WO2006058226A2

WO2006058226A2 - Modified dimeric streptavidins and uses thereof

Info

Publication number: WO2006058226A2
Application number: PCT/US2005/042766
Authority: WO
Inventors: Charles R. Cantor; Filiz M. Aslan
Original assignee: The Trustees Of Boston University
Priority date: 2004-11-24
Filing date: 2005-11-23
Publication date: 2006-06-01
Also published as: WO2006058226A3; WO2006058226A9

Abstract

The present invention provides novel modified streptavidin proteins, nucleic acids encoding such proteins, and uses of such proteins. Specifically, the invention provides single-chain dimeric streptavidin molecules. The invention also provides modified single-chain streptavidins with altered affinity to biotin and biotin derivatives, such as biotin-4-fluorescein. The modified single-chain dimeric streptavidin proteins can be used in analysis or composite separation either alone or in combination with ordinary streptavidin-biotin systems to visualize and/or separate composites and molecules. The single-chain streptavidin dimers of the invention also allow creating of more stable streptavidin tetramers. The single-chain streptavidin dimers of the invention can further be used as smaller sized functional streptavidin molecules in various in vivo methods, where the size of the conventional tetrameric streptavidin makes the use of such tetramers difficult or impossible. Such in vivo uses include diagnostic imaging and targeting of therapeutic molecules into individuals. The invention further provides methods of preparing and identifying novel modified single-chain streptavidin molecules.

Description

MODIFIED DIMERIC STREPTA VIDINS AND USES THEREOF

CROSS-REFERENCE TO RELATED APPLICATIONS

[001] This application claims the benefit under 35 U. S. C. 119(e) of U.S. provisional application Nos. 60/630,899, filed November 24, 2004, and 60/643,088, filed January 11, 2005, the contents of which are herein incorporated by reference in their entirety.

BACKGROUND Field of the Invention

[002] This invention relates to recombinant modified dimeric streptavidin proteins having two functional biotin binding sites, and to recombinant single-chain dimeric streptavidin proteins. These proteins can have an altered affinity for binding biotin, for example, an enhanced affinity to bind biotin-4-fluorescein. The invention further provides methods utilizing these proteins such as the recombinant single-chain dimeric streptavidin proteins, for example, for detection and isolation of targets. The invention also relates to nucleic acids encoding recombinant modified dimeric streptavidin proteins and to recombinant cells, phages, and protein chips which contain and/or express proteins encoded by these nucleic acids. Description of the Background

[003] Streptavidin, and its functional homolog avidin have been extensively used in biological and medical science due in large part to their ability to specifically bind biotin. Streptavidin-biotin binding has a very high affinity, and is one of the strongest known non-covalent interactions (N. M. Green, Methods Enzymol. 184:5-13,1990). This extraordinary affinity, coupled with the ability of biotin and its derivatives to be incorporated easily into various biological materials, endows streptavidin-biotin systems with great versatility.

[004] The advantages of streptavidin-biotin binding systems are numerous. The exceptionally high affinity and stability of the complex ensures complete reaction. Biotin's small size allows it to be conjugated to most molecules with no loss in molecular activity. Multiplicity of biotinylation sites combined with the tetrameric structure of streptavidin allows for amplification of the desired signal. The system is extremely versatile, as demonstrated by the large number of functional targets, binders and probes. The system is amenable to multiple labeling techniques, a wide variety of biotinylated agents and streptavidin-containing probes are commercially available. [005] Streptavidin-biotin complexes are used in a number of diagnostic and purification technologies. In general, a target molecule to be purified or detected or otherwise targeted is bound either directly to biotin or to a biotinylated intermediate. The binder may be almost any molecule or macromolecule that will complex with or conjugate to a target molecule. For example, if a particular antigen is the target, its binder would be an antibody. The biotinylated target is bound to streptavidin which may be bound to a probe for ease of detection. This basic technique is utilized in chromatography, cytochemistry, histochemistry, pathological probing, immunoassays, bioaffhity sensors and cross-linking agents, as well as more specific techniques such as targeting, drug delivery, flow cytometry and cytological probing.

[006] However, there is still room for improvement in the streptavidin-biotin systems. For example, the size of the streptavidin tetramer is still typically larger than ideal for in vivo applications. The large size can have disadvantageous pharmokinetics such as slow clearance from circulation and undesirable, nonspecific accumulation in organs like the kidney and liver.

[007] Moreover, the streptavidin tetramer is prone to dissociate which results in reduced binding in the in vitro assays.

[008] The mature streptavidin tetramer binds one molecule of biotin per subunit and the complex, once formed, is unaffected by most extremes of pH, organic solvents and denaturing conditions. Separation of wild type streptavidin from biotin requires harsh conditions, such as 8 M guanidine, pH 1.5, or autoclaving at 121⁰C for 10 minutes. [009] The origins of the unusually high binding affinity seen in streptavidin-biotin complexes has not been fully elucidated. X-ray crystallographic studies have shown that streptavidin's carboxyl and amino termini lie on the molecule's surface (P. C. Weber et al., J. Am. Chem. Soc. 114:3197-200, 1992). These termini have been modified by cleavage or conjugation with a minimal effect on biotin binding affinity. [0010] The streptavidin-biotin complex does not involve any covalent bonds, but does contain many hydrogen bonds, hydrophobic interactions and van der Waal interactions. These interactions are largely mediated by the aromatic side chains of tryptophan. Two tryptophan-lysine pairs are conserved between streptavidin and avidin. These pairs are found at positions 79-80 and 120-121 in streptavidin. Additional tryptophan residues in streptavidin are found at positions 21, 75, 92, and 108.

[0011] Trp-120 may play a role in maintaining local structures of streptavidin, particularly around the biotin-binding sites and the dimer—dimer interface. Strong hydrophobicity is observed around Trp-120 and three other tryptophan residues (Trp-79, 92 and 108) that make contact with biotin (P. C. Weber et al., Sci. 243:85-88,1989; C. E. Argara na et al., Nuc. Acids Res. 14:1871-82, 1986). In addition, hydrophobic interactions are the major force for the stable association of the two symmetric streptavidin dimers. Changes in local environment caused by the mutation of Trp-120 could prevent the molecule from folding correctly, resulting in diminished biotin-binding ability. In fact, the conversion of some amino acid residues located around the dimer- dimer interface to hydrophilic amino acids causes the formation of insoluble aggregates, probably due to random inter-molecular interactions.

[0012] Wild type streptavidin's strong affinity for biotin can also sometimes be a major drawback. The streptavidin-biotin binding system is essentially irreversible. The streptavidin-biotin bond is not affected by pH values between 2 to 13, nor by guanidine- HCl concentrations up to 8 M (neutral pH). The half-life for spontaneous dissociation of the streptavidin-biotin bond is about 2.5 years. The extremely strong binding of biotin to streptavidin means that biotinylated proteins can only be recovered from streptavidin supports under denaturing conditions. This sort of system is inappropriate for many procedures such as, one of its principal uses, the purification of delicate proteins. Native streptavidin-biotin cannot readily be used in sequential assays to detect specific types of biomolecules, macromolecular complexes, viruses or cells present in a single sample. The high affinity necessitates the use of harsh chemical reagents, complex procedures, and careful monitoring of the reactions. This also limits both yields and the ability to fully automate such reactions.

[0013] A number of methods have been developed in an attempt to create a releasable streptavidin-biotin or avidin-biotin conjugate. These methods include partly monomeric avidin beads, N-hydroxysuccinimide-iminobiotin and biotin or streptavidin cleavage. Monomeric avidin beads are formed by denaturing tetrameric avidin and coupling the denatured protein to chromatography beads. Thus, the so-called monomeric avidin is really a mixture of monomeric, dimeric and tetrameric proteins that have a binding affinity distributed between the wild type affinity of IO¹³ M^"' and the reduced affinity of 1O M^" . Thus, monomeric avidin beads produce low yields because some of the biotinylated products are irreversibly bound. Furthermore, the density and capacity of monomeric avidin beads is low.

[0014] N-hydroxysuccinimide-iminobiotin (NHS-iminobiotin) is a guanido analog of NHS-biotin with a pH sensitive binding affinity for streptavidin. The complete dissociation of NHS-iminobiotin from streptavidin occurs at low pH without the need for strong denaturants. A drawback to the NHS-iminobiotin system is that binding requires a pH of 9.5 or greater, while release requires a pH of less than 4. Thus, the use of NHS- iminobiotin is limited to those few molecules which are stable over a wide pH range. [0015] One method used to dissociate the streptavidin-biotin bond involves proteinase K digestion of streptavidin (M. Wilchek et al., Anal. Biochem. 171 :1-32, 1988). However, significant amounts of the streptavidin molecules remain attached even after proteinase K treatment. Proteinase K is useful only when the biotinylated product does not comprise proteins. Furthermore, this system precludes sequential assays or transfers of target.

[0016] Another method of release involves biotin cleavage of the binding partners, for example, of a cleavable biotin such as immunopure NHS-SS-biotin which is commercially available (Pierce Chemical Co.; Rockford, 111.). NHS-SS-biotin consists of a biotin molecule linked through a disulfide bond and an N-hydroxysuccinimide ester group that reacts selectively with primary amines. Using this group, NHS-SS-biotin is linked to a target molecule and the biotin portion removed by thiol cleavage. This complex approach is slow and of limited use since thiols normally disrupt native protein disulfide bonds. Furthermore, cleavage leaves a reactive sulfhydryl group that tends to react with other components of the mixture. Also, thiol-containing nucleic acids will no longer hybridize, severely limiting their usefulness.

[0017] A method of modifying streptavidin binding to biotin using mutant streptavidins that have alterations in the dimer interphase has been presented in PCT/USO 1/41027, and U.S. Patent No. 6,002,951.

SUMMARY OF THE INVENTION

[0018] We have now discovered novel modified streptavidin proteins, nucleic acids encoding such proteins, and uses of such proteins. Specifically, we have discovered single-chain dimeric streptavidin molecules. We also discovered that some of the novel single-chain dimeric streptavidins have a significantly increased affinity to fluorescently labeled biotin. Such modified streptavidin molecules can be used in analysis or composite separation either alone or in combination with ordinary streptavidin-biotin systems to visualize and/or separate composites and molecules. [0019] We have also discovered a method for making and using single-chain streptavidin dimers, that result in, for example, more stable tetramers. The single-chain streptavidin dimers of the invention can further be used as smaller sized functional streptavidin molecules in various in vivo methods, where the size of the conventional tetrameric streptavidin makes the use of such tetramers difficult or impossible. [0020] We have further discovered methods of preparing and identifying novel modified streptavidin molecules. Such methods can be used to screen for modified streptavidins with desired binding properties using a variety of techniques, such as display analysis, including but not limited to phage display, yeast display, and ribosome display analysis. Also, protein chip technology may be used for the screening purposes. The method allows screening for streptavidin molecules with additional mutations resulting in different binding affinity to, for example, biotin, and biotin derivatives, such as fluorescent biotins, such as biotin-4-fluorescein. The identified streptavidins can then be used in molecular separation and diagnostic assays alone or in combination with other streptavidins having different binding affinities to differently labeled biotin molecules. Such combinations allow creation of dual- or multi- screening/selection/purification methods in one single reaction.

[0021] Accordingly, in one embodiment, the invention provides a single-chain streptavidin dimer that combines two monomeric streptavidin subunits using circular permutation at or around the normal strong interface of two monomeric streptavidin subunits. One can use wild-type streptavidin subunits or core streptavidin subunits to make the single-chain dimeric streptavidins of the present invention. Core streptavidin typically consist of wild-type streptavidin (SEQ ID NO: 1) amino acids from about positions 13 to 139, 14 to 138, or 16 to 133 (see, e.g., U.S. Patent No. 6,022,951). [0022] The combined streptavidin subunits may also have mutations altering their affinity to biotin or biotin derivatives. For example, one or more tryptophan residues at positions 79, 120, 92, 108 may be mutated to alter the biotin binding affinity of the streptavidin subunits. For example, phenylalanine may be substituted for one or more of the tryptophan residues at positions 79, 120, 92, 108 (see, e.g., U.S. Patent No. 6,207,390). Amino acids which may be substituted for tryptophans or lysines include methionine, proline, isoleucine, leucine, valine, alanine, glycine, lysine (for tryptophan), phenylalanine, and derivatives and modifications of these amino acids (eg. beta-alanine, N-ethylglycine, 3-hydroxyproline, 4-hydroxyproline, allo-isoleucine, N-methylglycine, N-methylisoleucine, N-methylvaline, norleucine or norvaline). For example, the reduced- affinity single-chain dimeric streptavidin protein can comprise a phenylalanine, phenylalanine derivative (e.g. 4-amino-phenylalanine) or a phenylalanine modification (e.g. methylation) at position 79, 92, 108 or 120, and preferably at positions 79 and 120. In one embodiment, the mutation is Wl 2OK. Additionally, one may mutate lysines at positions 80, 121, 45, 94, and 1 1 1. Lysines at these positions have been suggested to participate in biotin binding (Avidin-Biotin Chemistry: A Handbook, M. D. Savage et al., ed., page 7, 1992). Additionally, one or more of the following mutations may be introduced to one or both of the streptavidin subunits of the single-chain dimeric streptavidin molecule: Asn23Ala, Ser27Glu, and Ser27Asp (see, e.g., U.S. Patent No. 6,368,813). Any combinations of these or other mutations known to affect biotin binding may be introduced to the streptavidin subunits present in the single-chain dimeric streptavidin molecule of the invention.

[0023] In one embodiment, the single-chain dimeric streptavidin molecule of the invention comprises mutations K24E, A28T, F50L, G89S, K148E, G181S, W120K, S26T, G32E, F50L, R171K, G247D, S251T, and S261N, or any combination thereof. In one preferred embodiment, the single-chain dimeric streptavidin molecule comprises mutations K24E, A28T, F50L, G89S, K148E, and Gl 8 IS. In another preferred embodiment, the single-chain dimeric streptavidin molecule comprises mutations S26T, G32E, F50L, R171K, G247D, S251T, and S261N.

[0024] For example, in one embodiment, the invention provides a single-chain streptavidin dimer of SEQ ID NO:2 as shown in Figure 7. Such bivalent constructs are useful, for example, in applications that involve bridging between two biotin-labeled molecules under conditions where native tetrameric streptavidin proves unsuitable. The single-chain streptavidin dimers also allow creation of molecules with two sites of differing biotin-binding affinity that can further broaden the range of streptavidin/biotin technology. For example, one can create a streptavidin that preferentially binds to fluorescently labeled biotin. Such a molecule can be used in combination with a traditional streptavidin with preferential binding to non-fluorescently labeled biotin, thus creating a double-binding assay that can be performed in one single reaction. [0025] In one embodiment, the invention provides single-chain streptavidin molecules that bind biotin-4-fluorescein with reversible and much higher affinity than to unlabeled biotin.

[0026] In another embodiment, the invention provides an isolated and purified single- chain streptavidin dimer mutant C2 comprising SEQ ID NO: 3 as shown in Figure 7. [0027] In yet another embodiment, the invention provides an isolated an purified single-chain streptavidin dimer mutant E2 comprising SEQ ID NO: 4 as shown in Figure 7.

[0028] In one embodiment, the modified streptavidins are engineered by destabilizing the dimer-dimer streptavidin interface together with a crossover-splicing procedure to convert the molecules to a single-chain dimeric (SCD) streptavidin molecules with two functional biotin-binding sites.

[0029] Accordingly, the invention also provides a method for separation and analysis of composites based on a fluorescence-tagged affinity system with tight, but still reversible interaction between the SCD streptavidin molecules and biotin-4-fluorescein. [0030] In one embodiment, the invention provides SCD streptavidins expressed in an expression system. In one preferred embodiment, the expression system is a phage- display system.

[0031] In another embodiment, the invention provides single-chain dimeric streptavidins attached on solid support, such as protein micro chips or micro beads. [0032] In yet another embodiment, the invention provides the use of expression libraries, such as display libraries, and use of protein-chips in screening for modified streptavidins that can bind to desired molecules, such as biotin derivatives, for example fluorescent biotins.

[0033] In one embodiment, the invention provides therapeutic and diagnostic methods using the single-chain dimeric streptavidins as described. In these methods, one preferably conjugates the single-chain dimeric streptavidin with a therapeutic or diagnostic molecule, such as a radioisotope, and delivers the molecule to an individual in a pharmaceutically acceptable carrier.

[0034] We have re-engineered streptavidin, which naturally forms a homotetramer with a high binding affinity to biotin. We destabilized the dimer-dimer interphase together with a crossover-splicing procedure or circular permutation to convert the streptavidin to a single-chain dimeric streptavidin (SCD) with two functional biotin- binding sites.

[0035] We introduced random mutations into this single-chain dimeric streptavidin. [0036] We also used phage display technology to express the single-chain dimeric streptavidins. For example, a pCANTAB 5 E vectors bearing the mutant streptavidin genes were transformed to E. coli TGl cells to form a bacterial library. The clones were expressed as fusion proteins on the 406-residue gene 3 protein (g3p) of Ml 3 bacteriophage. We panned the phages expressing the randomly mutated single-chain streptavidin dimers with biotinylated beads to optimize binding affinity and stability by phage display. The phage libraries can be panned with, for example, differently labeled biotins to identify mutants with optimal binding activity to such labeled biotins. Affinity- enriched phages were selected and sub-cloned into pET system with t7 RNA polymerase to produce soluble single-chain dimeric streptavidins.

DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 shows a table of the calculated free-energy-change contributions of residues that significantly influence the hydrophobicity of the dimer-dimer interface. [0038] Figure 2 shows a schematic presentation of the dimeric streptavidins showing on the lower left the mutant single chain dimeric streptavidin (SCD) and on the lower right the two-chain dimeric streptavidin (TCD).

[0039] Figures 3 A and 3B show the preferred mutation sites of single-chain dimeric streptavidin. Figure 3 A shows a schematic of the wild-type core streptavidin and Figure 3B shows a schematic of the mutant single chain streptavidin with circular permutation. Mutation sites of circularly permuted single-chain dimeric streptavidin. Two wild-type genes, diagrammed in (a), are cut, spliced and provided with tetrapeptide linkers as shown (b). L and R (left and right) designate the N-terminal and C-terminal halves of this construct. Residue numbering is according to wild-type streptavidin. [0040] Figure 4 shows the a cartoon representation of the structural details of the circularly permutated single-chain dimeric streptavidin of the present invention. GGGS

(SEQ ID NO: 14) linkers are depicted in ball & stick display. Residue numbering is according to wild-type streptavidin.

[0041] Figure 5 shows a schematic presentation of the pCANTAB 5 E phagemid cloning vector showing control regions and genealogy. This vector is designed such that the genes of interest can be cloned between the leader sequence and the main body of the fd gene 3 (the illustration depicts a single-chain antibody gene in that location). A sequence coding for a 13 -residue peptide tag ("E-tag" - AGAP VP YPDPLEP, SEQ ID

NO: 9) is present followed by an amber translational stop codon at the junction between the cloned gene and the sequence for g3p. Numbers give nucleotide residue positions.

(Amersham-Pharmacia manual.)

[0042] Figure 6 shows a table of the mutants selected from the 4th round panning of the rescued phagemid library constructed by the 3rd round error-prone PCR.

[0043] Figure 7 shows protein sequence alignment of NM-SCD (SEQ ID NO:2), C2 mutant (SEQ ID NO: 3), and E2 mutant (SEQ ID NO: 4) as compared to the consensus sequence (SEQ ID NO:5).

[0044] Figure 8 is a ribbon structure presentation of the C2 mutant dimer-dimer interface at the back side.

[0045] Figure 9 is a ribbon presentation of the E2 mutant dimer-dimer interface at the back side.

[0046] Figure 10 shows expression of the NM-SCD, C2 and E2 mutants. M refers to molecular standard marker.

[0047] Figure 11 shows the structure of biotin-4-fluorescein (B4F).

[0048] Figures 12A and 12B show how NM-SCD-tetramer binds B4F. Figure 12 A shows the B4F binding curve and Figure 12B shows results from the biotin/B4F competition analysis.

[0049] Figures 13A-13D show binding curves of mutants C2 and E2 in dimer (Figs.

13A and 13C) and tetramer (Figs. 13B and 13D) forms. Protein samples with the final monomer concentrations shown were mixed with 0.063 nM of biotin-4-fluorescein in a total volume of 150 μL. [0050] Figures 14A-14D show competition experiments for mutants C2 and E2 in dimer (14A and 14C) and tetramer (14B and 14D) forms.

[0051] Figures 15A-15D shows results of the reverse competition experiments with native streptavidin (Fig. 15A), NM-SCD (Fig. 15B), C2 dimer (Fig. 15C) and C2 tetramer

(Fig. 15D).

[0052] Figure 16 shows a table of binding constants of SCD proteins to B4F and biotin.

[0053] Figures 17A and 17B show examples of useful fluorescent molecules.

[0054] Figure 18 shows the nucleic acid sequence of a single chain dimeric streptavidin NM-SCD (SEQ ID No: 6) that is cloned between Ncol and Notl sites (bold and underlined).

[0055] Figure 19 shows the nucleic acid sequence of a single chain dimeric streptavidin of C2 mutant in cloned into a pCANTAB 5E vector (SEQ ID No: 7).

[0056] Figure 20 shows the nucleic acid sequence of the E2 mutant cloned in pET22b(+) vector at Ncol and Notl site (bold and underlined) (SEQ ID No: 8).

DESCRIPTION OF THE INVENTION

[0057] The present invention is directed to modified dimeric streptavidin molecules containing two streptavidin subunits in one single molecule, the modifications of such dimeric streptavidin molecules and methods of their use. The invention also provides nucleic acid sequences which encode these streptavidin subunits and molecules and recombinant cells which contain these sequences. Additionally, the invention provides methods for detecting and isolating small molecules, macromolecules and cells with streptavidin and to kits which contain streptavidin molecules and subunits of the invention. Methods of making modified dimeric streptavidin molecules, preferably single-chain dimeric streptavidin molecules are also provided. [0058] The single-chain dimeric streptavidin (SCD) molecule of the invention comprises two subunits of monomeric streptavidin covalently linked in one single polypeptide. The two streptavidin monomers are attached to each others using circular permutation to create a covalent cross-link between the two monomers and to create a new amino terminus and a new carboxyl terminus. The natural amino and carboxyl termini of the monomeric subunits are linked together using a linker peptide. For example, a linker or a bridge can be formed to link amino acids located in two different beta-strands of the monomeric streptavidin subunit. For example, amino acids 13 and 139 (corresponding to the wild-type sequence of streptavidin SEQ ID NO: 1), are located in two different beta-strands, and can be bridged to form a beta-turn. Such a beta-turn can be formed by addition of, for example, glycine residues, for example, 3, 4, 5, 6, or more glycine residues. It is possible to delete amino acid residues, for example, for the carboxyl terminal end and insert "spacer residues" such as glycine residues that will maintain conformation. Preferred linkers or bridges to form a beta-turn of the peptides of the invention are selected from GGGS (SEQ ID NO: 14) and SGGG (SEQ ID NO: 15). [0059] The circular permutation is preferably formed at or near the natural strong interface between two monomeric streptavidin subunits. In one preferred embodiment, the circular permutation cross-over is formed between amino acid residues corresponding to the amino acids 1 15 and 116, and 69 and 68 in the wild type streptavidin (SEQ ID NO: 1). For example, residue 115(GIy) and 69(Ser) can be covalently linked to form a cross¬ over and thus allow 116(GIu) to form a new amino terminus (N -terminus) and 68(GIy) to form a new carboxyl terminus (C-terminus). The covalently linked dimer can be regarded as a concatenation of two circularly permutated loops comprising structures of monomeric streptavidin, or modifications, and mutations thereof. Such modified or mutant streptavidin molecules may have altered biotin binding capacity as compared with the wild-type streptavidin. Examples of such modified streptavidins are described, infra. [0060] In creating the circularly permutated single-chain dimeric streptavidin from two streptavidin monomeric subunits, one preferably retains at least part, preferably the entire flexible loop near the biotin-binding site (Chu et al., Protein Science 7, 848-859). [0061] The monomeric streptavidin subunits that are concatenated using circular permutation can be wild-type or mutant streptavidin subunits. For example, one can form a single-chain dimeric streptavidin using at least one subunit, wherein W 120K mutation has been introduced. Other mutations can also be introduced to the single-chain dimeric streptavidin molecule of the invention. For example, one preferred mutant single-chain dimeric streptavidin comprises mutations K24E, A28T, F50L, G89S, K148E, and Kl 8 IS. Another preferred mutant single-chain dimeric streptavidin comprises mutations S26T, G32E, F50L,.R171K, G247D, S251T, and S261N. One can also combine streptavidin monomeric subunits with different biotin binding capacity to allow alteration in the biotin binding capacity of the single-chain dimeric streptavidin. Additionally, one can mix single chain dimeric streptavidins having different mutations. For example, one can mix C2 mutants (exemplified, infra), with E2 mutants (exemplified, infra), or a dimeric streptavidin molecule having the wild type monomeric subunits, or monomeric subunits carrying, for example one or two Wl 2OK mutations.

[0062] Based upon the present disclosure, one can modify the molecule's affinity to biotin. Such modifications can be performed, for example, by random mutagenesis a wildtype or mutant single-chain dimeric streptavidin as described, supra. For example, we have discovered modified dimeric streptavidin molecules having a significantly increased affinity to fluorescently labeled biotin relative to the native streptavidin. The binding of the biotin derivative with fluorescent molecule is tight but still reversible. [0063] Although native avidin and streptavidin both have a high affinity for biotin, they are different in many other respects. The two proteins have different molecular weights, electrophoretic mobilities and overall amino acid composition. Avidin is a glycoprotein found in egg whites and the tissues of birds, reptiles and amphibia. Like streptavidin, avidin has almost the same high affinity for biotin and exists as a tetramer with a molecular weight of between about 67,000 to about 68,000 daltons. Avidin also has a high isoelectric point of between about 10 to about 10.5 and contains carbohydrates which cause it to bind non-specifically to biological materials including cell nuclei, nucleic acids and lectins. These non-specific interactions make avidin less suitable than streptavidin for many applications.

[0064] Streptavidin is produced by the bacteria, Streptomyces avidini, and the wild type exists as a tetrameric protein having four identical subunits. That full length streptavidin monomer is 159 amino acids in length, some 30 residues longer than avidin. It contains no carbohydrate, and has an acidic isoelectric point of about 5.0 which accounts, in part, for the low non-specific binding level. Each subunit of streptavidin is initially synthesized as a precursor of 18,000 daltons which forms a tetramer of about 75,000 daltons. Secretion and post-secretory processing results in mature subunits having an apparent size of 14,000 daltons. Processing occurs at both the amino and carboxyl termini to produce a core protein of about 13,500 daltons, having about 125 to 127 amino acids. This core streptavidin forms tetramers and binds to biotin as efficiently as natural streptavidin. The amino acid sequence of the mature 160 amino acid streptavidin protein is as follows: XPSKDSKAQVSAAEAGITGTWYNQLGSTFIVTAGADGALTGTYESAVGNAESRY VLTGRYDSAPATDGSGTALGWTVAWKNNYRNAHSATTWSGQYVGGAEARINT QWLLTSGTTEANAWKSTLVGHDTFTKVKPSAASIDAAKKAGVNNGNPLDAVQQ (SEQ ID NO: 1)

[0065] Biotin, also known as vitamin H or cis-hexahydro-2-oxo-lH-thieno-(3,4)- imidazole-4-pentanoic acid, is an essential vitamin found in every living cell including bacteria and yeast. In mammals, the tissues having the highest amounts of biotin are the liver, kidney and pancreas. Biotin levels also tend to be raised in tumors and tumor cells. In addition to cells, biotin can be isolated from secretions such as milk which has a fairly high biotin content. Biotin has a molecular weight of about 244 daltons, much lower than its binding partners avidin and streptavidin. Biotin is also an enzyme cofactor of pyruvate carboxylase, trans-carboxylase, acetyl-CoA-carboxylase and beta-methylcrotonyl-CoA carboxylase which together carboxylate a wide variety of substrates. [0066] Only the intact bi cyclic ring of biotin is required for the strong binding to streptavidin. The carboxyl group of biotin' s pentanoic acid side chain has little to contribute to this interaction. Consequently, biotin derivatives, reactive to a variety of functional groups, can be prepared by modifying the pentanoic acid carboxyl group without significantly altering the target's physical characteristics or biological activity. This allows biotin to be conjugated to a number of target molecules. [0067] Streptavidin forms a highly stable complex with biotin. The hydrogen bonds between biotin and streptavidin are formed at least at Asn-23, Ser-27, Tyr-43, Ser-45, Asn-49, Ser-88, Thr-90, and Asp-128. The hydrophobic residues of the biotin-binding pocket in the streptavidin include at least Leu-25, Val-47, Trp-79, Trp-92, Trp-108, Leu- 110, and Trp-120. Particularly Trp-120 is involved in intersubunit contacts to biotin. Accordingly, any of these residues, alone or in combination, may be altered to modify the biotin binding of the single-chain dimeric streptavidins of the invention. For example, in one preferred embodiment, the invention provides a mutant single-chain dimeric streptavidin comprising mutations K24E, A28T, F50L, G89S, K148E, and Gl 81. In another embodiment, the invention provides a mutant single-chain dimeric streptavidin comprising mutations S26T, G32E, F50L, R171K, G247D, S251T, and S261N. [0068] For example, we calculated the free-energy-change contribution (ΔG) of residues that significantly influence the hydrophobicity of the dimer-dimer interphase as shown in Figure 1. The largest free-energy change can be attributed to the tryptophan at position 120 of the streptavidin.

[0069] Dimeric streptavidins can be formed by mutagenesis by destabilizing the dimer interface together with a crossover-splicing procedure.

[0070] To form a single-chain dimeric streptavidin (SCD), one combines the two monomer streptavidin subunits and performs a circular permutation to allow the dimer formation by a single chain. Circular permutation of polypeptides is a known technique in the art of protein engineering. For example, the amino acids present in the dimer interface of streptavidin preferably include at least amino acids at positions 115 and 1 16 and 68 and 69. For example, one can delete a portion of the molecule that is not important to the dimer interface and insert a linker of amino acids. Amino acid linkers are well known in the art and selected to maintain a specific conformation. For example, a string comprising glycines, such as SGGG (SEQ ID NO: 15), or GGGS (SEQ ID NO: 14), or other short sequence. For example, a cross-over can be formed between two streptavidin chains and it can be formed by inserting the glycine comprising amino acid string between the natural N and C termini of the monomeric streptavidin subunits, which allows the dimer formation between the combined subunits together with the circular permutation. For example, the string of glycines can be inserted between positions 13 and 139, or 13 and 138, or 16 and 116 of the wild-type core streptavidin, and by re- engineering to combine the subunits at the dimer interface (for example, see Figure 2). [0071] Unlike dimers that are formed of two separate polypeptide chains, the single chain proteins of this invention allow engineering of the proteins via various display libraries. There are a number of popular display technologies based on display of proteins or protein fragments on the surface of biological entities, for example, phage display (Phage Display - A Practical Approach, Ed. Tim Clackson and Henry B. Lowman, Oxfor University Press, 2004) and yeast display (Boder ET and Wittrup KD. (1997) Nat Biotechnol. 15:553-7. Yeast surface display for screening combinatorial polypeptide libraries; Feldhaus MJ et al., 2003, Nat Biotechnol. 21 :163-70. Flow- cytometric isolation of human antibodies from a nonimmune Saccharomyces cerevisiae surface display library), which rely on the replication of the organism to amplify the library and introduce errors. As an alternative to these in vivo systems, there one can also carry out the whole process in vitro using a so called, ribosome display analysis (Schaffitzel C et al. ,1999, J Immunol Methods 231 : 119-135. Ribosome display: an in vitro method for selection and evolution of antibodies from libraries; Weichhart T et al., 2003, Infect Immun. 71 :4633-41. Functional selection of vaccine candidate peptides from Staphylococcus aureus whole-genome expression libraries in vitro). In this case in vitro- generated transcripts are translated in cell extracts and RT-PCR is used to amplify the genetic information after the ligand-mediated isolation of mRNA-ribosome-protein complexes has taken place.

[0072] Use of such libraries allows screening of compounds that can bind to the dimer with varying affinities. The libraries also allow screening of a number of mutant dimers, for example randomly mutagenized single-chain dimers that can bind a desired compound. For example, one can screen for streptavidin mutants that have a desired binding affinity to biotins labeled with different fluorescent labels. Alternatively, one can screen for streptavidin molecules that have tight, yet reversible binding capacity to biotin or biotin derivatives.

[0073] In all of these display systems, one can use, magnetic sorting to allow an efficient and rapid isolation of interacting proteins (Siegel DL et al., 1997, J Immunol Methods 206:73-85. Isolation of cell surface-specific human monoclonal antibodies using phage display and magnetically-activated cell sorting: applications in immunohematology).

[0074] Accordingly, the present invention provides methods of using the single-chain modified streptavidin dimers of the present invention in display libraries. [0075] The single chain dimeric streptavidins can also be attached to solid support, such as protein chips, and thus provide convenient diagnostic applications and separation systems for, for example, biotin, or biotin derivative -labeled target molecules. [0076] Accordingly, in another embodiment, the invention provides methods of using the single-chain modified streptavidin dimers in the protein chips. Chips comprising the single-chain modified streptavidin molecules of the invention are also provided. [0077] In conventional streptavidin-biotin systems, there are many methods for detecting or purifying a given target. For example, the target may be directly biotinylated and complexed with the single-chain dimeric streptavidin or a mutant thereof. Alternatively a binder that complexes with the target may be the biotinylated component. There may, in fact, be more than one binder involved in a given system. The detectable probe may be bound to the streptavidin and the system may involve more than one detectable probe. Both the target and the support may be biotinylated, and the two can be complexed together with the single-chain dimeric streptavidin or a dimer thereof, which essentially has similar structure to that of a tetrameric wild-type streptavidin in that it has four biotin-binding sites. Many permutations are made possible by the variety of targets, binders and probes.

[0078] Another embodiment of the invention is directed to a method for detecting or purifying a target, in one embodiment, at least two targets simultaneously from a heterogeneous mixture which contains the target(s). The target is biotinylated using biotin or a biotin derivative or modification appropriate for the target. If one wishes to detect or purify two targets, one can label the first target with, for example, biotin, and the second target with, for example, biotin-4-fluorescein (B4F). One can also label targets with more than two, such as 3, 4, 5, 6, 7, or even more biotin derivatives, provided that all the biotin derivatives have different affinity, when compared to each other, to the modified streptavidins that are used to bind the labeled targets. Targets may be nearly any substance such as biological or inorganic substances. Biological substances include proteins and protein precursors, nucleic acids (DNA, RNA, PNA) and nucleic acid precursors (nucleosides and nucleotides), carbohydrates, lipids such as lipid vesicles, cells, biological samples and pharmaceuticals. Typical proteins which are detectable in conventional streptavidin/biotin systems, and useful herein, include cytokines, hormones, surface receptors, antigens, antibodies, enzymes, growth factors, recombinant proteins, toxins, and fragments and combinations thereof.

[0079] Subcellular components may also be purified by linking a ligand, with an affinity to the component, to the streptavidins of the invention. Proteins which can be purified include, but are not limited to cell adhesion molecules, antibody antigens, receptors ligands and antibodies. Specific affinity adsorbent moieties, such as wheat germ agglutinant, anti-idiotypic antibodies and dye ligands may be coupled to streptavidin to isolate glycosylated proteins such as SPl transcription factor, dye binding proteins such as pyruvate kinase and liver alcohol dehydrogenase, and other antibodies. Using this method, cellular and subcellular organelles may be rapidly purified using specific antibodies. [0080] The heterogenous mixture is contacted to the streptavidin or a mixture of different affinity streptavidins of the invention that may be fixed to a surface of a support or free in solution. Mixture is removed or the support removed from the mixture and the target purified. Alternatively, target(s) may be coupled to streptavidin(s) of the invention and biotin or biotin derivative attached to the support. In either situation, the result is the same. However, using single-chain dimeric streptavidin coupled to target, target may be isolated free of any biotin.

[0081] Streptavidins of the invention can be bound to a solid support or surface that is used to capture the labeled target(s). Solid surface can be, for example, plastic, glass, ceramics, silicone, metal, cellulose, and gels. Solid support can be, for example, beads, tubes, chips, resins, plates, wells, films, and sticks. The supports can be in the form of an array or in solution.

[0082] Accordingly, the invention provides a method for contacting a target, comprising biotinylating at least one target with biotin or biotin derivative under conditions such that a heterogeneous mixture is created, said heterogeneous mixture comprising target and biotinylated target; and contacting said heterogeneous mixture with a solid support, said solid support comprising at least one dimeric single-chain streptavidin molecule or a tetrameric streptavidin, comprising two dimeric single-chain streptavidin proteins of the invention.

[0083] Another embodiment of the invention is directed to a method for the detection of a disorder in a patient such as a human. Single-chain dimeric streptavidin, or a mutant derivative thereof, is naturally targeted to biotin. Biotinylation of a site within the body of the patient, such as, for example, using monoclonal or polyclonal antibodies coupled with biotin and specific for the site will target the coupled complex to the site. Single-chain dimeric streptavidin may be coupled with a pharmaceutical which can be used to treat the disorder. Treatable disorders include neoplasms, genetic diseases and infections (e.g. viral, parasitic, bacterial, fungal diseases).

[0084] As used herein, the terms "pharmaceutically acceptable" , "physiologically tolerable" and grammatical variations thereof, as they refer to compositions, carriers, diluents and reagents, are used interchangeably and represent that the materials are capable of administration to or upon a mammal without the production of undesirable physiological effects. The carrier must be pharmaceutically acceptable in the sense of being compatible with the other ingredients of the formulation and not deleterious to the recipient thereof. The preparation of a pharmacological composition that contains active ingredients dissolved or dispersed therein is well understood in the art and need not be limited based on formulation. Typically such compositions are prepared as injectable either as liquid solutions or suspensions, however, solid forms suitable for solution, or suspensions, in liquid prior to use can also be prepared. The preparation can also be emulsified or presented as a liposome composition. The active ingredient can be mixed with excipients which are pharmaceutically acceptable and compatible with the active ingredient and in amounts suitable for use in the therapeutic methods described herein. Suitable excipients are, for example, water, saline, dextrose, glycerol, ethanol or the like and combinations thereof. In addition, if desired, the composition can contain minor amounts of auxiliary substances such as wetting or emulsifying agents, pH buffering agents and the like which enhance the effectiveness of the active ingredient. [0085] We have shown that the single-chain streptavidin molecule of the invention is bound by biotin-4-fluorescein about 10⁵ times more strongly than by biotin. Accordingly, the invention provides systems and methods that, for example, can use single chain dimeric streptavidins that preferentially bind biotin-4-fluorescein in assays together with the single-chain dimeric streptavidins that preferentially bind biotin in the present invention. Accordingly, in embodiment, the invention provides non-mutant single-chain dimeric streptavisin and single-chain dimeric streptavidin molecules with mutations at various positions, such as mutants C2 and E2, described, infra.

[0086] Using the single-chain dimeric streptavidin molecules of the present invention one can develop and design a method wherein in one reaction, one can separate or identify different target molecules. For example, one can create a modified streptavidin that has different affinities to differently modified biotins. One can then mix these modified streptavidins to create a dual or triple or multiple step separation system in one reaction, wherein the targets labeled with biotins to which the streptavidins of the invention have different affinities can be identified or separated from a heterogenous mixture.

[0087] Without wishing to be bound by a theory, we suggest that there is a weak and a strong binding site in SCD. We have here shown how to engineer and express single- chain dimeric streptavidins. We have also shown how to make mutants of SCDs using, for example, error-prone PCR and how to select the mutants using a display system, such as phage-display.

[0088] We have used tight expression control to enhance protein production. One can also use several known systems for protein production to produce the single-chain strepatavidin molecules according to the invention. Useful protein production systems include well known prokaryotic, yeast, insect and mammalian systems. For example, one can use prokaryotic pBAD vector system or yeast Pichia based system to produce the single-chain dimeric streptavidin molecules. One may also use commercially available protein production and purification services such as those currently offered by, for example, Invitrogen Corporation, Carlsbad, CA; Orbigen Inc., San Diego, CA; GeneArt,

Toronto, Canada; Dyadic International Inc., Jupiter, FL; and ISC BioExpress, Kaysville,

UT.

[0089] We have shown that non-mutant single-chain dimeric streptavidins (NM-SCD) are mostly tetrameric, but that mutants, such as C2 or E2 exhibit dimer-tetramer equilibria. Our results also show that saturation with biotin and B4F favors tetramer formation.

[0090] Accordingly, we have developed single-chain dimeric streptavidins that can be used in a wide variety of biotechnological applications. For example, SCD and B4F can be used to create a fluorescence-tagged affinity system with tight, but still reversible interactions, that can be used alone, or sequentially or simultaneously with ordinary streptavidin-biotin system for composite separation and/or analysis.

[0091] The methods to screen for additional single-chain dimeric streptavidins with different binding properties allow development of not only dual, but also multiple analysis systems in one reaction, depending on the mixture of single-chain dimeric streptavidin mutant mixture used in the reactions. For example, 2, 3, 4, 5, 6, or even more mutant single-chain dimeric streptavidins recognizing differently labeled biotin molecules or having with different biotin binding kinetics may be mixed in one reaction to provide a separation or analysis system for 2, 3, 4, 5, 6, or even more different biotin-labeled molecules.

[0092] According to the teachings of the present invention one can prepare a multitude of different mutant single-chain dimeric streptavidins and screen them for desired properties. For example, the mutants can be screened using expression systems such as phage display. Phage display is an in vitro selection technique in which a peptide or protein is genetically fused to a coat protein of a bacteriophage, resulting in display of the fused protein on the exterior of the phage virion, while the DNA encoding the single- chain dimeric streptavidin resides within the virion. This physical linage between the displayed protein and the DNA encoding it allows screening of vast numbers of variants of the single-chain dimeric streptavidin, each linked to their corresponding DNA sequence, by a simple in vitro selection procedure called biopanning. [0093] For example, in its simplest form, biopanning is carried out by incubating the pool of phage-displayed variants of the single-chain dimeric streptavidin with a biotin- coated beads, washing away unbound phage, and eluting specifically bound phage by disrupting the binding interactions between the phage and the target. [0094] In this work, we have successfully constructed and expressed the single-chain dimer gene. Accordingly, the invention provides vectors and cells that are created to produce the single-chain dimeric strepatavidins of the present invention. [0095] Mutants C2 and E2 had 6 and 7 amino acid changes, respectively, compared to the non-mutant single-chain dimeric strepatavidin (NM-SCD). For convenience, in order to maintain the original numbering scheme for the amino acid residues, the interchain connection between residue 115 from the L domain ("chain A") and 69 from the R domain ("chain B") has been omitted. This allows the structure display program RASMOL© (a program for molecular graphics visualisation originally developed by Roger Sayle, Copyright (C) 1989, 1991 Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA) automatically to color the two domains separately and the user can select residues of interest by using the original streptavidin numbers followed by the separate chain labels. The linker residues were numbered 140- 143 in both domains.

[0096] Examination of the structure of the C2 mutant leads to several conclusions: (1) most of the changes have occurred on the original molecular surface of the tetramer, (2) no changes appear at the dimer-dimer interface (even though this is exposed to solvent), and (3) only one change (F29L) involves a residue close to the binding pocket (as defined by the 14 ligand-contacting residues in the native structure). Interestingly, there are two lysine -> glutamate substitutions and two glycine -> serine substitutions. The latter occur in surface loops, suggesting that serine may be a relatively benign substitution in such situations for β-barrel structures once folding has taken place. The absence of mutations at the monomer-monomer interface seems logical, though in the covalent dimer that interface should be less critical than for the two-chain dimer. More surprising is the lack of any changes at the dimer-dimer interface. Without wishing to be bound by a theory, we postulate that possibly the foπnation of folded (and thus detectable), expressed protein is favored by a tendency to form tetramers, hence dimer-dimer interface mutations would be selected against.

[0097] Mutant E2 presents a somewhat different picture. Four mutations (S52T, S62N, S136T and G142E) appear to be peripheral, neither close to the binding pocket nor at the dimer-dimer interface. The monomer-monomer interface again has suffered no significant change (though S62N may have a small effect on it), but the fringe of the dimer-dimer interface has been altered by the R103K mutation. Without wishing to be bound by a theory, while this represents a comparatively minor change in physicochemical terms (arginine and lysine both having large, positively charged side chains), it may account for the greater tendency of E2 to remain dimeric in the absence of biotin as compared with C2 and the non-mutant SCD. E2 also has "hits" in or near the biotin-binding pocket of the R domain (G48D and S52T), showing that it may have distinctly strong and weak binding sites. The L domain has only a single mutated residue in the vicinity of the binding pocket, and, curiously, the mutation is F29L - the single, binding pocket change found for C2.

[0098] Considering both mutants together, we can now show what types of mutation can be tolerated in such a single-chain dimer and still allow it to fold, have reasonable binding affinity, and be recoverable from cells expressing it. First, it is clear that the monomer-monomer interface should remain intact. This also correlates with experience with the two-chain dimer Stv-43 (Sano et al., Proc Natl Acad Sci U S A. 1997 Jun 10;94(12):6153-8). Secondly the dimer-dimer interface seems somewhat immune to changes, suggesting that it may play some role in determining which mutants are successfully expressed, folded and purified. Thirdly, changes do not occur frequently in or near the biotin-binding site.

[0099] The single-chain dimeric molecules have a fundamental asymmetry, which, without wishing to be bound by a theory, likely has a small effect on their binding sites. Furthermore, random mutations will exacerbate this asymmetry in virtually every instance. From the very outset, the L and R domains of the SCD are non-equivalent, and the selection of a single W120K mutation established the basic asymmetry for the NM- SCD and any of its derivatives. Since, the side chain on residue 120 projects outward away from the SCD, again, without wishing to be bound by a theory, we conclude that this is unlikely to affect the binding sites in the dimeric forms - though when tetramerization occurs W 120 should insert normally into a companion binding site across the dimer-dimer interface, while Kl 20 will behave abnormally with respect to its interaction partner (see the discussion of dimer-tetramer equilibrium below). Thus in any of the tetramers formed from the SCD constructs there will be two distinct types of biotin-binding site, each with a characteristic Kd value. Without wishing to be bound by theory, we assume that the sites with Wl 20 back in its normal (or near-normal) location will have tight ligand binding while the other sites will have much weaker binding. Some of the fluorescence polarization data appear to demonstrate the presence of strong and weak sites.

[00100] It is also likely that the number of mutations introduced in C2 and E2 is sufficient to cause local conformation changes that involve one or both binding sites. Without wishing to be bound by theory, such a conformation change propagates allosterically from one site to the other, though this seems unlikely given the lack of allosterism in native wild-type core streptavidin. Any conformation changes will differ between the two halves of the SCD (because of the asymmetry described above). This is an additional reason to expect different binding affinities at the two sites. The presence of an induced conformation change also shows a lower affinity for ligand binding to the SCDs as compared to binding to native streptavidin since some of the binding interaction energy is diverted to stabilizing the refolded conformation. Finally, again, without wishing to be bound by theory, if ligand binding stabilizes a more native-like conformation of the mutant SCDs, it favors tetramer formation - with concomitant creation of the near-native tight binding sites.

[00101] A variety of recent studies have examined the interconversion of streptavidin and avidin from tetramer to dimer to monomer. In addition to those cited previously, these include the rational design of active avidin monomers (Laitinen et al., J Biol Chem. 2003 Feb 7;278(6):4010-4. Epub 2002 Nov 27; Qureshi & Wong, Protein Expr Purif. 2002 Aug;25(3):409-15.) and the analysis of their aggregation states, including the effect of crystallization (Pazy et al, J Bacteriol. 2003 Jul;185(14):4050-6.) and ligand binding (Laitinen et al., J Biol Chem. 2001 Mar 16;276(11):8219-24. Epub 2000 Nov 13.). A general observation from these studies is that ligand binding tends to favor oligomerization, and that corresponds with our results also. Our FPLC results for the NM-SCD and mutants C2 and E2 show that the addition of unmodified biotin likely converts the dimers to tetramers. However, upon addition of biotin-4-fluorescein tetramer formation is prevented and mutants remain as dimers because biotin-4- fluorescein mimics the Wl 20 and blocks the binding site. Under such conditions, while having a single Wl 20 would somewhat favor tetramerization, biotin-4-fluorescein prevents Wl 20 from getting into its pocket on the other subunit hence preventing tetramerization.

[00102] The fluorescence polarization experiments had an additional variable: dimer- tetramer equilibrium. This accounts for the sigmoidal behavior of B4F binding curves. Any shift towards tetramer formation leads to longer correlation times, hence higher polarization values. We also show that during tetramerization the association of the remaining W 120 residue in each SCD with one of the binding pockets of the other SCD can convert that pocket into a much stronger binding site. [00103] Exactly how quenching will affect polarization values while somewhat uncertain it will not alter the conclusions from our experiments since these are largely based on simply empirical comparisons, not on any elaborate theory that depends upon precise quantitative measurements.

[00104] The present invention provides modified streptavidin molecules with remarkably tight binding of B4F to the SCD proteins (KD ~ 10^'10 M) which was a surprise, especially when compared to their rather modest affinity for biotin (KD ~ 10^" M). We tested both fluorescein and fluorescein glycinamide to see if they had detectable tight binding to the SCD proteins, but the results were negative. Therefore, without wishing to be bound by theory, there is likely no major contribution from some fortuitous binding site on the protein that has a high affinity for the fluorescein moiety. Saturating the solutions with biotin before testing with fluorescein did not alter this result. [00105] There is a possibility that the B4F binds to the biotin binding pocket in the SCDs in such a way that the fluorescein part of the molecule can orient so as to insert itself into the site normally occupied by tryptophan 120. The SCDs in their dimeric form have no way to configure a binding site with tryptophan present in this location (close inspection of the structure indicates that it is highly unlikely that the R domain can maneuver so as to insert the W 120 into its own pocket). However, the aromatic fluorescein group is linked to biotin by a flexible tether with at least three rotatable bonds. This allows it to loop around and place one of its aromatic rings into the opening where W120 binds in tetrameric streptavidin, i.e., B4F may be a ligand that brings part of its binding site with it. The additional stability gained from the interactions of the protein with the fluorescein moiety as well as with the biotin moiety could account for the increase by a factor of 104 in binding affinity compared with free biotin. The normal affinity of wild-type streptavidin for biotin is at least another factor of 104 greater. [00106] Here we show that B4F has significantly lower values for fluorescence polarization with native streptavidin than with the SCD proteins. Without wishing to be bound by a theory, that can be explained on the basis of less freedom of rotation of the fluorophore moiety of protein-bound B4F in the latter case - exactly what would be expected if it were binding in the tryptophan site. Consistent with this observation is the higher Kd for B4F binding to native streptavidin as compared to the SCD mutant proteins. In this case, the fluorescein moieties of B4F molecules have to compete with tryptophan at all four binding pockets.

[00107] To test the proposed orientation of the fluorescein moiety of B4F, we conducted a computational ligand-docking experiment using the program GOLD (Verdonk et al., Proteins. 2003 Sep l;52(4):609-23). The ligand file was prepared with MOE (Chemical Computing Group, Toronto, Canada) and 50 docking runs were made. From these the top hits were examined and one that positioned the fluorescein moiety in the postulated location was selected to illustrate the possible binding mode We show that all the top hits correctly placed the biotin moiety in its binding pocket with near- negligible RMSD values from the position of the ligand observed in the crystal structure. [00108] The docking result clearly shows a reasonable position for the fluorophore, and that it occupies nearly the same location as the tryptophan residue (W 120) that reaches across the dimer-dimer interface in wild-type streptavidin and that contributes so significantly to the biotin-binding affinity. [00109] Therefore, we show that computational analysis of the structures of these single chain dimeric molecules allows further characterization of the binding modes of

B4F, both to wild-type streptavidin and to the engineered mutants.

[00110] The present invention also provides cloned genes encoding the single-chain dimeric streptavidins with desired binding properties and methods for producing the encoded proteins.

[00111] Consequently, the invention provides the use of such molecules in new applications.

[00112] The constructs provided by the present invention are available and can be manipulated either by additional directed-evolution techniques, by site-directed mutagenesis techniques, or a combination of these.

[00113] The invention further provides that additional directed-evolution methods that can be used in high-throughput production to allow screening of a larger number of candidates.

[00114] The invention also provides uses of the single-chain dimeric streptavidins for applications that require prevention of tetramerization. The invention provides methods for preventing tetramerization by introducing mutations at the dimer-dimer interface, following in the results with the two-chain dimer.

[00115] Single-chain dimeric streptavidins may be bound to a solid support, for example, to facilitate detection and isolation procedures. Typical solid supports include the surfaces of plastic, glass, ceramics, silicone or metal. These components may be found in detection kits, biological sample analysis devices and environmental sampling aids. Particularly useful types of such components include beads, tubes, chips, resins, membranes, monolayers, plates, wells, films, sticks or combinations of the surfaces. Solid supports also include hydrogels which may be made of a variety of polymers such as acrylamide and hydroxyapatite, or biomolecules such as dextran, cellulose or agarose.

[00116] Binding of streptavidin to surfaces may be accomplished in several ways. A solid support may be derivatized with a moiety which can form a covalent bond with streptavidin, avidin or biotin. Alternatively many commercially available surfaces may be used to couple streptavidin, avidin or biotin. Example of such surfaces include agarose, cross linked agarose, acrylamide, agarose and acrylamide combinations, polyacrylic, cellulose, nitrocellulose membranes, nylon membranes, silicon and metal. These surfaces may be further modified to contain a carboxyl or other reactive group for crosslinking. Reagents suitable for crosslinking to solid surfaces include cyanogen bromide, carbonyldiimidazole, glutaraldehyde, hydroxysuccinimide and tosyl chloride. [00117] Single-chain dimeric streptavidins may also be coupled to a biological agent such as an antibody, an antigen, a hormone, a cytokine or a cell. Cells may be eukaryotic such as mammalian cells, prokaryotic such as bacterial cells, insect cells, parasitic cells, fungal cells or yeast cells. Coupling may be through electrostatic interaction or by covalent modification of one or both coupling partners. Covalent modifications are fairly stable when, for example, the coupled agent is subjected to the a biological environment such as occurs on administration to a host such as a mammal.

[00118] Given the advantages of streptavidin over avidin for biotechnological uses, the present SCD provides a promising point of departure for such an effort. [00119] The invention additionally provides a double W120K mutant SCD. [00120] The invention further provides binding assays and methods using the single- chain dimeric streptavidin and its mutants.

[00121] As discussed above, we have discovered that the non-mutant single-chain dimeric streptavidin (NM-SCD) has 10⁵ times as high an affinity for B4F than for biotin. Accordingly, the invention provides dual-binding systems where the biotin-labeled partners are readily dissociable at moderate biotin concentrations while B4F-labeled partners require more stringent conditions to be released. Of course, a B4F-labeled ligand (protein, nucleic acid, etc.) can also carry a built-in fluorescent tag. Numerous potential protocols could take advantage of such a system. For example, sequential use of SCD with B4F: in a solution with two targets, one B4F-tagged and the other biotin-tagged, one can use the single-chain dimer to pull out the B4F-tagged target and then use wild-type streptavidin to pull out the biotinylated target. This can also be reversed if desired. [00122] Suitable B4F derivatives useful in such methods are derivatives that allow the B4F derivative to be coupled to other molecules, particularly proteins and nucleic acids, without impairing its binding to the SCD. For example, one can use direct conjugation of lysine residues or amine-containing linkers to the fluorescein carboxyl group, which may have some steric hindrance, but should still be reactive with standard coupling agents like EDC or TSTU. One can also use synthesis of a B4F analog that has a suitable branch bearing a coupling site in the linker. B4F can be purchased from, for example, Biotium (Hayward, CA) who also provides custom synthesis.

Two-Chain Dimeric Streptavidins (TCD)

[00123] The sequences of the previously described two chain streptavidin dimers (TCDs) can be used in developing the single chain streptavidin molecules of the present invention. Following is a description of the useful TCDs.

[00124] For example, the two chain streptavidin consists of a dimer of streptavidin subunits, wherein each subunit contains a mutation of His 127 to prevent tetramer formation as well as mutations at the exposed subunit-subunit interface to reduce hydrophobicity. Preferred sites that are hydrophobic and should be mutated to remove hydrophobicity include Trpl20, Leul24, Vall25 as well as Hisl27. [00125] This hydrophobicity can be reduced by substituting one or more of the hydrophobic amino acid residues with more hydrophilic amino acids with care being taken that the side group substituted does not affect steric considerations because the resultant modified molecule should have a conformation approximating the wild-type structure. For example, one can substitute acidic amino acids such as Asp, GIu, uncharged amino acids such as Ser, Asn, GIn, and GIy, etc. Preferred substitutions are selected from the groups consisting of Asp, GIu, GIy, Ser, and Asn. More preferably Asp, Asn and Ser. In one preferred embodiment, Trpl20 is substituted with Asp, Leul24 is substituted with Asp or Asn, VaI 125 is substituted with Asp or Ser and His 127 is substituted with Asp.

[00126] In one preferred embodiment, the internal amino acids 1 through 15 can be deleted and the determinal amino acids from amino acid 130, preferably 133, still more preferably 139 can be deleted. For example, the streptavidin subunits can correspond to amino acids 1-159 with the above-mentioned substitution. In another embodiment the subunit can correspond to amino acids 13 to 139, 13 to 138, 13 to 133, 16 to 133, etc. [00127] The use of these subunits results in stable reduced size dimers that destabilize the dimer/dimer interfaces and therefore produce a dimeric protein instead of the wild type tetramer. The resultant molecule will have a reduced ability to bind biotin when changes are made at Trpl20 but it will still bind biotin. Preferably the molecule has an ability to bind biotin that is at least 50% and more preferably at least 60% and still more preferably at least 75% that of the wild-type streptavidin. In one embodiment, the molecule is made as a single chain molecule to enhance binding and stability. [00128] These dimeric streptavidin molecules can be used for any purpose the wild- type streptavidin can be used for. For example, it can be used in two-step immunotargeting. In the first step, the target is tagged with biotinylated antibody or other molecule containing a moiety that streptavidin will bind to. In this second step, the streptavidin, which is conjugated to a second molecule such as an imaging agent or radiotherapeutic agent is used to bind to, for example, the biotinylated antibody to tag the target. As a result of the smaller size of the streptavidin molecule, detrimental pharmacokinetics of the wild type molecule are reduced. For example, there is a reduction in problems resulting from slow clearance from the circulation and undesirable, nonspecific accumulation in organs such as the kidneys and liver, which both result from the large size of the tetramer.

[00129] The streptavidin-biotin binding system is an established fixture in biology due, at least in part, to the ability of streptavidin to non-covalently interact with biotin. This association is highly specific and quite strong with a binding constant of greater than 10¹⁵ M^"1. This tight binding can in some instances limit the usefulness of conventional streptavidin-biotin systems. Although molecules and cells can be isolated from complex mixtures, removal of one or the other of the binding partners is difficult. Dissociation is preferably accomplished under very harsh conditions such as 6-8 M guanidine-HCl, pH 1.5. Not surprisingly, such conditions also denature, and thereby inactivate or destroy most target biological substances.

[00130] Although streptavidin behaves better than antibodies as, for example, the radioisotope carrier for two step in vivo radioimmune imaging and therapy, its size is still much larger than ideal. The present invention constructs streptavidin molecules, are stable, reduced-size streptavidin mutants that destabilize the dimer-dimer interfaces so as to produce a dimeric protein instead of the wild-type tetramer. This dimeric streptavidin will speed up the delivery of radioactivity to the tumor targets. [00131] However, to be active in biotin-binding in a patient, a dimer must remain soluble in serum, and the individual monomers must remain joined, and not dissociate. This is accomplished by the present invention. The smaller size of the dimeric streptavidin allows this protein to better serve than the wild-type tetramer in a range of functions such as in an improved in vivo diagnostic or therapeutic agent by reducing the disadvantageous pharmacokinetics of natural tetrameric streptavidin, such as slow clearance from the circulation and undesirable, non-specific accumulation in organs like the kidney and liver.

[00132] By using molecular modeling, a dimeric streptavidin, STv-43, had previously been designed and produced in a functional form capable of binding biotin. Energetic calculations for the construction of Stv-43 which had an H127D mutation in addition to deletion of the hydrophobic loop (Gl 13-Wl 20), so that there would be no apparent contact with biotin in a dimeric molecule, indicated that this molecule could maintain a dimeric structure with sufficient solubility in aqueous media. However, experimental results showed that Stv-43 requires biotin to fold into a functional dimeric form, and it has limited stability without bound biotin. Purified Stv-43 without bound biotin gradually dissociates into monomers, which then aggregate. The dimeric molecule also has much lower affinity for biotin, Kc=IO^" , at least partly because of the lack of contacts made by Trp-120 of an adjacent subunit with biotin through the dimer-dimer interface. Although the loop removal increased the solvation free energy (ΔGs) to the target value of greater than -34 kcal/mol, it increased the electrostatic repulsion energy between the monomers, which caused the dimer to become unstable. The present invention uses a different strategy in which loop deletion is avoided.

[00133] Instead of loop deletion, we mutated, for example, L 124 and V 125 to more hydrophilic amino acids to increase the solubility of the two-chain dimeric protein. This design allowed us to obtain soluble two-chain dimeric streptavidins, which are folded into a functional dimeric form without using biotin.

[00134] Localization of tumor cells is a fundamental step in the treatment of cancer, and one way to do this is by attaching radioactive isotopes to tumor cells. Once they are attached to the tumor, its location in the body can be determined by detecting the emitted radiation. The most widely used method, known as radioimmunotargeting, attaches radioisotopes to cells via antibodies. In this method antibodies, can be produced such that they ideally bind to the surface of a certain cell type only and can be radiolabeled in vitro. These radiolabeled antibodies are injected into cancer patients. However, their large size is disadvantageous due to slow diffusion to tumor tissue from the bloodstream and retention in the kidney for long periods. During this time, the patient will be exposed to potentially harmful radiation levels. Furthermore, the radioactive emissions from antibodies trapped in the kidney produce a large amount of undesirable background noise in the tumor images.

[00135] The streptavidin tetramer has a molecular mass of only about 60,000 Daltons, while Mr for a typical antibody is about 150,000 Daltons. Pretargeting is a novel method of radioimmunotargeting. Pretargeting takes advantage of the strong noncovalent bond between biotin and streptavidin. This strategy has been developed to eliminate the disadvantages that are caused by using radiolabeled antibodies. Pretargeting involves coating the surface of tumor cells with antibodies that are covalently conjugated to biotin. The main difference from radioimmunotargeting is the absence of radioisotopes in the biotinylated antibody. After injection of the biotinylated antibody into the patient, the slow-moving biotinylated antibody can attach to the surface of the tumor. Since the antibody is not radioactive, there is no harmful radiation exposure to the patient. After the excess biotinylated antibodies have cleared the patient's kidneys and bloodstream, the tumor will remain coated with biotinylated antibodies. This concludes the pretargeting step. Next comes the targeting step in which radiolabeled streptavidin is injected into the patient. Radiolabeled streptavidin binds tightly to the biotin residues of the biotinylated antibody, which is attached to the tumor. Once the radiolabeled streptavidin binds the biotinylated antibodies, the tumor will be coated with radioisotopes, thereby achieving the diagnostic/therapeutic objective.

[00136] Dimeric streptavidin will not have as high binding as tetrameric streptavidin due to mutation of Wl 2OD. However, one can further increase the affinity of dimeric streptavidin by producing single-chain streptavidin which can be used with a phage display system to select the mutants having optimized binding. [00137] There are many other alterations that can be made in the subunits to effect other properties of the molecule. For example, many stable streptavidin proteins remain associated with biotin under conditions which would cause dissociation of biotin from wild-type streptavidin. For example, core streptavidin peptides form macromolecules of protein that have a higher affinity for biotin than wild-type streptavidin or even "natural core streptavidin" (containing the amino acid sequence of streptavidin from positions 13 to 139). Natural core streptavidin binds to about 0.94 to about 0.96 molecules of biotin per subunit of streptavidin and pH 7.4. At 6 M guanidine hydrochloride, pH 7.4, full- length and natural core streptavidin show about a 20% reduction in biotin binding (about 0.768 molecules biotin per subunit). At 4 M guanidine hydrochloride, pH 1.5, these same proteins show about a 15% reduction in biotin-binding affinity (about 0.826 molecules biotin per subunit). In contrast, Stv-13, a core streptavidin peptide (containing the amino acid sequence from 16 to 133) that forms tetrameric protein, shows no significant reduction in biotin-binding. Without wishing to be bound by theory, this stability may be due to the absence of two charged residues from the core (GIu 14 and Lys 134) as well as two polar residues (Ser 136 and Ser 139). Furthermore, at 6 M guanidine hydrochloride, pH 1.5, natural core retains only about 20% of its normal biotin-binding capability whereas Stv-13 retains over about 80%. Preferably, stable streptavidin proteins bind to at least about 0.80 molecules of biotin per subunit at 6 M guanidine, pH 7.4, and more preferably at least about 0.9 molecules of biotin per subunit at 6 M guanidine, pH 7.4. It is also preferable that stable proteins bind to 0.8 molecules of biotin per subunit at 6 M guanidine, pH 7.4, and 0.7 molecules of biotin per subunit at 6 M guanidine, pH 7.4. This enhanced binding to biotin may be due to the lack of steric hindrance caused by the presence of the amino and/or carboxyl terminal sequences. Neither of these sequences appear to be necessary for binding, but in fact may have in some way interfered with biotin binding. Accordingly, in one embodiment, one may mutate these residues to alter the biotin binding.

[00138] For example, hybrid tetrameric streptavidin proteins containing four core streptavidin peptides, two with aspartic acid and two with lysine at position 127, retain biotin more strongly than natural biotin under harsh conditions. Experiments show that at increased temperatures, both wild-type and natural core streptavidin loose biotin binding ability more quickly with increased temperatures. These streptavidin protein retains greater than about 90% of bound biotin at 6O⁰C, preferably greater than about 80% of bound biotin at 6O⁰C, and more preferably greater than about 50% of bound biotin at 80°C. Streptavidin protein with increased affinity for biotin may have a binding affinity of greater than about 10¹² M^'1 or about 10¹⁴ M^'1, preferably greater than about 10¹⁵ M^"1, and more preferably greater than about 10¹⁶ M^"1.

[00139] In another embodiment, the streptavidin molecule's attachment to biotin can be disrupted more easily than the wild-type streptavidin-biotin bond. The streptavidin-biotin bond involving streptavidin proteins of the invention may be disrupted through the addition of a fairly low concentration of biotin or biotin derivatives (biotin analogs) or modifications. The concentration of biotin which can be used to disrupt the streptavidin- biotin bond using streptavidin proteins of the invention is between about 0.1 mM to about 10 mM or, preferably, between about 0.3 mM to about 2 mM. In addition, elution may be performed in a high pH (e.g. 100 mM triethylamine, pH 11.5; 100 mM phosphate, pH 12.5), in a low pH (e.g. 100 mM glycine pH 4; 100 mM glycine pH 2.5; 100 mM glycine pH 1.8), in high salt (e.g. 5 M LiCl, 10 mM phosphate, pH 7.2; 3.5 M MgCl₂, 10 mM phosphate pH 7.2), or in the presence of ionic detergents (e.g. 1% SDS; 1% DOC), dissociating agents (e.g., 2 M urea; 8 M urea; 2 M guanidine HCl), chaotropic agents (e.g., 3 M thiocyanate), organic solvents (e.g., 10% dioxane; 50% ethylene glycol, pH 11.5; 50% ethylene glycol, pH 8), protease (protease K) or water. This type of versatility is a great advantage when utilizing conventional streptavidin-biotin detection or isolation procedures. Proteins are not destroyed and cells remain viable even after biotin has been removed.

[00140] In yet another embodiment, the streptavidin molecule is bound to a solid support, for example, to facilitate detection and isolation procedures. Typical solid supports include the surfaces of plastic, glass, ceramics, silicone or metal. These components may be found in detection kits, biological sample analysis devices and environmental sampling aids. Particularly useful types of such components include, but are not limited to beads (e.g., magnetic beads; Dynal), tubes, chips, resins, gels, membranes (e.g., porous membranes), monolayers, plates, wells, films, sticks or combinations of these surfaces. Solid supports also include hydrogels which may be made of a variety of polymers such as acrylamide and hydroxyapatite, or biomolecules such as dextran, cellulose or agarose.

[00141] Binding of streptavidin to surfaces may be accomplished in several ways. A solid support may be derivatized with a moiety which can form a covalent bond with streptavidin or biotin. Alternatively many commercially available surfaces may be used to couple streptavidin or biotin. Examples of such surfaces include agarose, cross linked agarose, acrylamide, agarose and acrylamide combinations, polyacrylic, cellulose, nitrocellulose membranes, nylon membranes, silicon and metal. These surfaces may be further modified to contain a carboxyl or other reactive group for crosslinking. Reagents suitable for crosslinking to solid surfaces include, for example, cyanogen bromide, carbonyldiimidazole, glutaraldehyde, hydroxysuccinimide and tosyl chloride, and others known to one skilled in the art.

[00142] One type of streptavidin, which may be used to facilitate coupling to a solid support is composed of a core streptavidin containing a plurality of cysteines at the protein's amino or carboxyl terminus, and preferably the carboxyl terminus. The cysteines facilitate binding to, for example, a thiolated support.

[00143] Preferably the streptavidin molecules of the invention are coupled to another molecule such as an imaging agent, a radiopharmaceutical, a biological agent such as an antibody, an antigen, a hormone, a cytokine, a cell or a pharmaceutical agent (for in vivo use). Such agents are well known in the art as is the method of coupling. See, for example, "Conjugate Vaccines", Contributions to Microbiology and Immunology, J. M. Crage and R.E. Lewis, Jr. (Eds), Carger Press, New York (1989) (incorporated herein by reference). Cells may be eukaryotic such as mammalian cells, prokaryotic such as bacterial cells, insect cells, parasitic cells, fungal cells or yeast cells. Coupling may be through electrostatic interaction or by covalent modification of one or both coupling partners. Covalent modifications are fairly stable when, for example, the coupled agent is subjected to the a biological environment such as occurs on administration to a host such as a mammal.

[00144] Another embodiment of the invention is directed to nucleic acids which encode a streptavidin subunit of the invention. Such nucleic acids may further comprise transcription or translational control regions to regulate transcription, translation or secretion of the recombinant protein. Control sequences can also be introduced to provide inducible expression. This is very useful as streptavidin is somewhat harmful to most cells. Recombinant nucleic acids may be introduced into bacterial cells, for example, by transformation, or into mammalian cells, for example, by transfection. [00145] Recombinant cells can be used to produce large quantities of recombinant molecule as needed or to provide a continuous source of recombinant streptavidin to a biological system. Recombinant cells, which can support the expression of streptavidin molecules or peptides (subunits) on the invention include eukaryotic cells such as mammalian and yeast cells and prokaryotic cells such as bacteria. [00146] In conventional streptavidin-biotin systems, there are many methods for detecting or purifying a given target. For example, the target may be directly biotinylated and complexed with the reduced substrate affinity streptavidin. Alternatively a binder that complexes with the target may be the biotinylated component. There may, in fact, be more than one binder involved in a given system. The detectable probe may be bound to the streptavidin and the system may involve more than one detectable probe. Both the target and the support may be biotinylated, and the two are complexed together with the reduced substrate affinity streptavidin. Many permutations are made possible by the variety of targets, binders and probes.

[00147] Another embodiment of the invention is directed to a method for detecting or purifying a target from a heterogeneous mixture which contains target. The target is biotinylated using biotin or a biotin derivative or modification appropriate for the target. Targets may be nearly any substance such as biological or inorganic substances. Biological substances include proteins and protein precursors, nucleic acids, such as single- or doublestranded DNA, RNA, siRNA, miRNA, PNA, aptamers, and nucleic acid precursors (nucleosides and nucleotides), carbohydrates, lipids such as lipid vesicles, cells, biological samples and pharmaceuticals. Typical proteins which are detectable in conventional streptavidin/biotin systems, and useful herein, include cytokines, hormones, surface receptors, antigens, antibodies, enzymes, growth factors, recombinant proteins, toxins, and fragments and combinations thereof.

[00148] Subcellular components may also be purified by linking a ligand, with an affinity to the component, to a streptavidin of the invention. Proteins, which can be purified include cell adhesion molecules, antibody antigens, receptors ligands and antibodies. Specific affmnity adsorbent moieties, such as wheat germ agglutinant, anti- idiotypic antibodies and dye ligands may be coupled to streptavidins of the invention to isolate glycosylated proteins such as SPl transcription factor, dye binding proteins such as pyruvate kinase and liver alcohol dehydrogenase, and other proteins, such as antibodies. Using this method, cellular and subcellular organelles may be rapidly purified using specific antibodies.

[00149] The heterogenous mixture is contacted to the reduced-affinity streptavidin, which may be fixed to a surface of a support of free in solution. Mixture is removed or the support removed from the mixture and the target purified. Alternatively, target may be coupled to streptavidin and biotin attached to the support. In either situation, the result is the same. However, using reduced-affinity streptavidin coupled to target, target may isolated free of any biotin.

[00150] Alternatively, when it is not as important to separate streptavidin or biotin from the purified target, a streptavidin with increased affinity for biotin may be used. This may be useful, for example, where the targeted substance is, for example, a malignant cell or a contaminant. The contaminant may be removed by the increased affinity streptavidin and disposed of. The increased affinity will ensure a more complete recovery than wild type streptavidin. Because wild type streptavidin is unstable under the extremes conditions of pH, salt, detergent, and disrupting agents, it is preferable to neutralize these agents before separation with streptavidin. Increased affinity streptavidin may reduce or eliminate the need for neutralization. This reduction or elimination will reduce processing time and complexity and contribute directly to cost reduction.

[00151] Using the methods disclosed herein, combination of detection and isolation procedures can also be utilized. For example, targets can be transferred from one support to another using a manual or automated apparatus. Sequential detection or purification techniques can also be used to purify targets to homogeneity. Such techniques were heretofore not possible when the streptavidin biotin bond could not be easily broken. In addition, nearly any conventional detection or isolation methodology that can be performed with conventional streptavidin-biotin procedures can be performed with the modified streptavidins of the present invention.

[00152] Another embodiment of the invention is directed to a method for the detection of a disorder in a patient such as a human. Reduced-affinity streptavidin is naturally targeted to biotin. Biotinylation of a site within the body of the patient, such as, for example, using monoclonal or polyclonal antibodies coupled with biotin and specific for the site, such as a tumor cell, will target the coupled complex to that site. Reduced- affinity streptavidin may be coupled with a pharmaceutical which can be used to treat the disorder. Treatable disorders include neoplasms, genetic diseases and infections (e.g., viral, parasitic, bacterial, and fungal infections).

[00153] Another embodiment of the invention is directed to a method for the isolation and culture of infectious agents from a patient. Body fluids, such as blood of a patient may be contacted with a support with antibodies specific for viral surface antigens. If the antibody was cross-linked to the solid support by a reduced-affinity streptavidin, bound infectious agents may be released without harm with a gentle elution technique. The isolated agents may be definitively identified by live culture. Infectious agents which can be isolated by this technique include, for example, slow viruses, malaria and infectious yeast.

[00154] Another embodiment of the invention is directed to a purification method for nucleic acids and nucleic acid-protein complexes. Nucleic acids can be immobilized to, for example, a column through a single-chain dimeric streptavidin complex (dimer thereof). The immobilized nucleic acid may be single or double stranded and it may comprised cloned sequence or random sequence. The column may be used to enrich for nucleic acid-binding proteins. The proteins bound to nucleic acids may be released without the use of nuclease or protease. The product may be studied, without the disruption of the protein nucleic acid bond by native gel electrophoresis (a gel mobility shift assay). This is an especially powerful tool for studying proteins with relatively low affinity for nucleic acids such as transcription factors.

[00155] The references cited herein and throughout the specification, including the examples, are herein incorporated by reference in their entirety.

[00156] The following experiments are offered to illustrate embodiments of the invention, and should not be viewed as limiting the scope of the invention.

EXAMPLES Example 1

[00157] In this and the following examples we demonstrate a methodology for creating derivatives of streptavidin that contain two biotin binding sites in a single polypeptide chain. Such bivalent constructs are useful in applications that involve bridging between two biotin-labeled molecules under conditions where native tetrameric streptavidin proves unsuitable. They also allow creation of molecules with two sites of differing biotin-binding affinity that can further broaden the range of streptavidin/biotin technology.

[00158] In addition to its diagnostic and therapeutic applications, streptavidin has been used for affinity purification of biotinylated macromolecules. However, its low dissociation constant does not allow the recovery of the targets back from the affinity matrix unless very harsh conditions are used. Therefore, use of directed-evolution strategies to engineer streptavidin mutants with reduced size and different biotin-binding affinities should provide an opportunity to exploit reversible interactions with biotinylated molecules in applications that require a wide spectrum of affinity. [00159] The main purpose of the work described has been to engineer a single-chain "dimeric" streptavidin. By a crossover-splicing procedure with circular permutation and the introduction of the W120K mutation (Laitinen et al., FEBS Lett. 1999 Nov 12;461(1- 2): 52-8), we constructed a single-chain dimeric streptavidin (SCD). Error-prone PCR was employed to generate random mutations in this single-chain dimeric streptavidin construct, following which we used phage display to optimize biotin binding affinity. Screening by ELISA detected stable proteins, but we did not separately seek to optimize stability.

Design and mutagenesis strategy

Destabilization of the dimer-dimer interface: Streptavidin, a homo-tetrameric protein, can be thought of as a dimer of stable subunit dimers. A pair of subunits associates very tightly to form the primary dimer, in which the subunit β-barrels have complementary curved surfaces that interact via numerous van der Waals contacts. Two primary dimers then combine to form the tetramer. Thus, each streptavidin molecule has two different subunit interfaces: (1) the strong interface between subunits in the stable primary dimer, and (2) the weaker interface between two stable dimers (dimer-dimer interface) in the tetramer (Reznik, et al. (1996) Nature Biotechnology 14, 1007-11 ; Sano, et al., (1996). Annals of the New York Academy of Sciences 799, 383-90; Sano et al., (1997) Proceedings of the National Academy of Sciences of the U. S. A. 94, 6153-6158). [00160] The extremely high stability of the streptavidin-biotin complex derives primarily from three main molecular recognition mechanisms: the first is the hydrogen bonding between the biotin and eight amino acid residues of streptavidin that are located at one end of each β-barrel near the dimer-dimer interface and define the biotin-binding site. As revealed by the X-ray crystallographic results, the ureido oxygen of biotin forms three hydrogen bonds arranged with tetrahedral geometry to stabilize an sp3 oxyanion (Weber, P. C, et al., 1989, Science 243, 85-8).

[00161] The second recognition mechanism depends on aromatic side chain contacts by four Tip residues, 79, 92, 108, and 120, that - together with Leu-25, Val-47, and Leu- 110 - form hydrophobic parts of the biotin-binding pocket and interact with the thiophene ring and the alkyl side chain of biotin through van der Waals forces. [00162] The third molecular recognition element consists of a flexible loop near the biotin-binding site (Chu, et al.,1998, Protein Science 7, 848-859). The integrity of the tetrameric structure, particularly the dimer-dimer interface, is essential to maintain the intersubunit contacts to biotin made by Trp-120 and Lys-121 (Reznik, et al.,1996, Nature Biotechnology 14, 1007-11 ; Salamone, F., 1994, The effect of mutations on the energetic stability of the streptavidin tetramer. Senior Project Final Report, Boston University; Sano, T., et al., 1996, Molecular engineering of streptavidin. Annals of the New York Academy of Sciences 799, 383-90). Mutation of Trp 120 to Lys (W120K) produces a stable two-chain dimeric streptavidin with reversible biotin-binding ability (Kd =2.8 xlO^"8 M) that remains dimeric in solution both in the absence and presence of biotin (Laitinen et al., 1999). The analog of W120 in sea urchin fibropellin (as yet the only known homolog of avidin and streptavidin) is Kl 041, and in streptavidin the single replacement of Wl 20 by lysine led to this successful expression of a two-chain dimeric protein. Based on that elegant result, in our SCD design we used the Wl 2OK mutation to prevent association across the dimer-dimer interface instead of using our set of four other mutations which also provided stable two-chain dimeric streptavidin mutants.

Fusion of two monomers with circular permutation

[00163] The first step to create a single-chain dimer construct requires performing a circular permutation, reordering of the residues within the peptide chain, on each monomer. Chu et al engineered a circularly permuted streptavidin with deletion of residues between 47-50 of the flexible loop (45-52) (Chu, et al., 1998, Protein Science 7, 848-859). To apply circular permutation to each monomer, we constructed a bridge between residues 13 and 139 of the core wild-type streptavidin to connect the initial N and C termini of the monomer with a GGGS linker. Residues 13 and 139 are located in two different β-strands, thus combining them forms a β-turn. Next, we determined a site to break on each peptide chain of the dimer so as to create four ends, two of which could be cross-connected to fuse the chains without major distortion of tertiary structure. Residues 115 (GIy) and 69 (Ser) served this purpose and were linked to combine the two monomers. The remaining two ends became the new N- and C- termini at 116 (GIu) and 68 (GIy), respectively. The covalently linked dimer can be regarded simply as the concatenation of the two circularly permuted loops.

[00164] The gene for commercially produced and completely truncated core streptavidin (Pahler, A., et al, 1987, Journal of Biological Chemistry 262, 13933-7) Alal3-Serl39 was used as the starting point since this protein does not aggregate and is very soluble, making it well-suited for biotechnological applications and a prime target for genetic engineering. In the work of Chu et al. (1998, Protein Science 7, 848-859) a flexible loop near the biotin-binding site was deleted and the new N- and C-termini was introduced near the biotin binding site. However, in our work, the flexible loop at that position was retained and new N- and C-termini were introduced distal to the binding site.

Optimization by error-prone PCR and phage display

[00165] Error-prone PCR: Generally, in error-prone PCR reactions standard PCR methods are modified so as to increase the natural error-rate of the polymerase (Cadwell, 1992, PCR Methods and Applications 2, 28-33; Cirino, P., et al., 2003, Methods in Molecular Biology 231, 3-10; Leung, D. W., Chen, E., and Goeddel, D.V., 1989, A method for random mutagenesis of a defined DNA segment using a polymerase chain reaction. Technique 1, 11-15.). Taq polymerase (Keohavong, et al., 1989, Proceedings of the National Academy of Sciences, U. S. A. 86, 9253-9257) is preferred due to its naturally high error-rate, with errors mostly biased toward AT to GC changes. However, there are other reported polymerases which introduce different types of mutations, for example more GC to AT changes (Cirino, P., et al., 2003, Methods in Molecular Biology 231, 3-10). Another way to increase the error rate is to substitute MnC12 in place of MgC12 (Lin-Goerke, J. L., et al., 1997, Biotechniques 23, 409-412.). In addition, higher concentrations of MgC12 (7 niM instead of 1.5 mM) can be used to stabilize non- complementary base pairs (Cadwell, et al., 1994, PCR Methods and Applications 3, S136-S140; Ling, L. L., et al., 1991, PCR Methods and Applications 1, 63-69). Thus, in order to create potentially useful streptavidin variants (derivatives) we subjected the gene encoding single-chain dimeric streptavidin (SCD) to random mutagenesis by three rounds of error-prone PCR. To date, there have been no reports of application of this technology to streptavidin. [00166] Phage Display: Phage display, a powerful technology which has many applications in different areas of biotechnology and medicine (such as protein engineering, study of ligand-receptor interactions, immunology, etc.), allows display on the surface of M 13 phages of large protein libraries that include many molecules with different properties (Burton, D. R. (1995). Phage Display. Immunotechnology 1, 87-94; de Bruin, R., et al., 1999, Nature Biotechnology 17, 397-399; Oneil, K. T. & Hoess, R. H. (1995). Phage Display - Protein Engineering by Directed Evolution. Current Opinion in Structural Biology 5, 443-449; Rapley, R. (1995). The Biotechnology and Applications of Antibody Engineering. Molecular Biotechnology 3, 139-154; Smith, G. P. & Petrenko, V. A. (1997). Phage display. Chemical Reviews 97, 391-410). This is followed by selection of the phage that display proteins with desired properties. Application of phage display is almost exclusively restricted to single-chain polypeptides (Cantor, C. R. & Smith, C. (1999). Genomics. 519; Dunn, I. S., 1996, Current Opinion in Biotechnology 7, 547-553), although preliminary efforts to develop a phage λ-based system to display multimeric proteins have been reported (Dunn, I. S. (1995). Journal of Molecular Biology 248, 497- 506; Hoess, R. H., 2002, Current Pharmaceutical Biotechnology 3, 23-8; Maruyama, I. N., et al., 1994, Proceedings of the National Academy of Sciences of the U. S. A. 91, 8273-8277; Sternberg, N. & Hoess, R. H., 1995, Proceedings of the National Academy of Sciences of the U. S. A. 92, 1609-1613). A study of phage with mutants in the major coat protein P8 (Sidhu, S. S., et al., 2000, Journal of Molecular Biology 296, 487-95) included successful display of wild-type streptavidin (16-133), possibly as a tetramer, though there was no evidence to support a multimeric structure (Clackson, T. & Wells, J. A., 1994, Trends in Biotechnology 12, 173-184; Winter, G., et al., 1994, Annual Review of Immunology 12, 433-55).

[00167] The display system we used for phage display is based on filamentous phage strain Ml 3 and exploits the interaction between a "phagemid" vector (Phagemids are highly modified Ml 3 particles carrying single-stranded DNA engineered to facilitate cloning, expression and selection of target proteins via phage display) and helper phage with an Ml 3 life cycle. Ml 3 belongs to the Ff class of the filamentous bacteriophages (genus Inovirus). Other common related phage are fl and fd (the three strains have 98% homology). These viruses all contain a circular, single-stranded DNA genome packaged in a long, cylindrical protein capsid. Other display systems based on bacteriophage T4 (Efimov, V. P., et al, 1995, Virus Genes 10, 173-177; Ren, Z. J., et al., 1996, Protein Science 5, 1833-1843) and λ (Sternberg, N. & Hoess, R. H., 1995, Proceedings of the National Academy of Sciences of the U. S. A. 92, 1609-1613) can also be used. [00168] Filamentous phages are flexible rods about 1 μm long and 6 nm in diameter (i.e., with an axial ratio of about 170:1). The mass of this particle is about 16.3 MDa, 87% of which is contributed by ca. 2700 copies of the 50 amino-acid-residue-long major coat protein g8p (gene 8 protein; also known as pVIII or P8). This forms a tube of helically arranged molecules that envelop the single-stranded DNA genome that contains 6407 bases encoding 10 different proteins. One of the special features of M 13 phage is the expression of several minor coat proteins. At one tip the Ml 3 phage particle has five copies each of the 406-residue gene 3 protein (g3p) and the 112-residue gene 6 protein (g6p). The other two minor coat proteins, the 33-residue gene 7 protein (g7p) and the 32- residue gene 9 protein (g9p), are located at the other tip. Minor coat proteins function as adsorption proteins on the tips of the phage (Smith, G. P. & Petrenko, V. A. (1997). Phage display. Chemical Reviews 97, 391-410).

[00169] To display the protein or peptide of interest, its gene sequence must be fused to that of one of the coat proteins so as to produce a chimeric protein. The phage protein part of such a chimera directs its incorporation into the phage coat where it anchors the non-phage-derived protein - on "display" for various assay and selection procedures. In this work we displayed SCD proteins fused to g3p which is the capsid protein most commonly used for this purpose. g3p has the disadvantage of displaying only about 5 molecules per phage particle, but large inserts can be packaged into phage reasonably well, since the foreign peptide is inserted at the very end of the packaged particle. This may create less steric hindrance when passing through the g6p exit pore, and/or be less disruptive of phage coat assembly than g8p fusions, which have been extensively used to display short peptides. However, large inserts tend to lower the phage infectivity or even make the phage noninfective limiting the ability to select a particular displayed protein (Smith, G. P., 1985, Science 228, 1315-1317). This problem is generally overcome by expressing the chimeric g3p from a phagemid together with a helper phage that provides the majority of the g3p in the cell, which is wild-type. Proteins fused to the carboxy- terminal portion of g3p missing domains Nl and N2 can be displayed on hybrid phage that retain their infectivity because of the presence of wild-type g3p (Barbas III, C. F., et al., 2001, Phage Display A Laboratory Manual. 2.7-2.8). Successful display of an active protein depends on (1) efficient translocation into the membrane; (2) proper folding; (3) avoidance of degradation in the periplasm, and (4) suitable packaging into phage. Foreign proteins also can be inserted between the Nl and N2 domains as well as between the N2 and CT domains of g3p (Krebber et al., 1997). These constructs, can retain (lower levels of) infectivity as long as the Nl and N2 domains can interact to generate a pilus- binding site. Recent work has focused on varying the positions of insertion in ways that select for protease-resistant proteins (Krebber, C, et al., 1997, Journal of Molecular Biology 268, 607-618) as well as using directed evolution of modified g3p domains to allow more efficient screening for the fused guest protein.

[00170] Classification of the phage-display systems can be done according to the arrangement of the coat protein genes. For fusions to g8p and g3p there are three kinds of display systems: types 3 and 8 phage, types 33 and 88 phage and types 3+3 and 8+8 phagemid systems. In a "type 3" vector there is a single phage genome with one gene 3 to which foreign DNA inserts are attached to encode a single type of chimeric g3p molecule. Theoretically, the virion displays the foreign peptide encoded by the insert on all five g3p molecules. However, proteolytic enzymes in the host bacterium may cleave the foreign peptide from some or even most copies of g3p, especially if the foreign peptide is large. Similarly, foreign peptides would be displayed on every copy of g8p and g6p in type 8 and hypothetical type 6 vectors (no type 6 vectors have been reported yet) (Smith, G. P. & Petrenko, V. A. (1997). Phage display. Chemical Reviews 97, 391-410). A type 88 vector contains two genes 8 which encode two different types of g8p molecules; one carries a foreign DNA insert and the other is wild-type. The resulting phage coat will assemble from both recombinant and wild-type g8p molecules. The latter usually predominate. This allows display of quite large proteins with hybrid g8p proteins on the virion surface, even though the phage assembly cannot be supported entirely by hybrid protein. Similarly, there are two genes in a type 33 vector: one is wild type and the other carries an additional piece of foreign (recombinant) DNA and both recombinant and wild-type g3p proteins are incorporated into the phage coat. Unlike the 33 system, in type 3+3 systems the two genes 3 are on separate genomes. A helper phage carries the wild-type gene while the recombinant gene resides on the phagemid genome. A phagemid, carrying a filamentous-phage replication origin, remains inactive until the cell is infected with the helper phage. After infection the phage replication proteins act on the phage origins of both the helper phage DNA and the phagemid DNA. Therefore, two types of progeny virions will be secreted: particles carrying the helper phage DNA and particles carrying phagemid DNA. Like the type 33 system, the coats of these virions are composed of a mixture of recombinant and wild-type g3p molecules. When cells are infected by a phagemid virion, the antibiotic resistance carried by phagemid is acquired by the cell. Cells infected only by a helper phage virion produce progeny helper phage in the normal way. However, the progeny virions will not be recombinant since the helper phage carries only a single wild-type gene 8, and is not accompanied by a phagemid encoding recombinant g8p. Type 3+3 systems are like type 8+8 systems, except that the phagemid carries the insert in gene 3 rather than gene 8. The recombinant g3p encoded by a type 3+3 phagemid usually lacks its N-terminal domain since cells expressing this domain resist superinfection by helper phage.

[00171 ] Filamentous phage can only infect strains of E. coli that contain the bacterial F pilus, encoded by the F plasmid, because these bacteriophages use the tip of the F conjugative pilus as a receptor. Ml 3 phage, as well as other Ff phages, do not kill the host E. coli during the productive infection. The infected E. coli host cells produce and secrete phage particles without undergoing lysis. Infection is initiated when the N- terminal domain of g3p (about 200 amino acids) attaches to the tip of the F pilus of a male E. coli, and the g3p-containing end of the particle enters the cell first. Upon entering the cell, the coat proteins of the phage dissolve into the surface envelope and the uncoated ssDNA is released into the cytoplasm. In the cytoplasm, the host DNA replication machinery converts the single-stranded phage DNA into a double-stranded replicative form (RF). This double- stranded genome then undergoes rolling-circle replication to produce ssDNA for packaging into new phage particles. The template for the transcription of phage mRNA is also the RF. Newly synthesized ssDNAs emerge through the cell envelope, in the process wrapping themselves with the coat proteins from the membrane to form intact virions. Progeny phage (several hundred per cell per division cycle) are secreted continuously without killing the host. Infected cells continue to divide, though more slowly than uninfected cells, and such cultures can yield more than 0.3 mg/mL of virus particles. Leader sequences direct transport of g3p and g8p proteins into the inner membrane of the bacterial cell. Immediately after the single-stranded progeny DNA molecules emerge from the inner bacterial membrane, they are assembled into mature phage. Thereafter the phage escape through the cell wall into the medium. Phage production continues until the cell eventually dies either from accumulated toxic Ml 3 phage components, such as g3p, or from cellular waste products. [00172] We expressed single-chain dimeric streptavidins as g3p fusion proteins and functionally displayed at the tip of the Ml 3 phage. We used commercially available phagemid vector pC ANTAB 5 E, which is designed such that the genes of interest can be cloned between the leader sequence and the main body of the Ml 3 gene 3 (Figure 5). A sequence encoding for a peptide epitope tag ("E-tag") is present, followed by an amber translational stop codon at the junction between the cloned gene and the sequence for the g3p (Figure 5). When TGl cells (supE strain of E.colϊ) are transformed with the recombinant vector, translation continues through the amber stop codon to produce the SCD-g3p fusion protein that will be displayed on the phage tip. In TGl cells, suppression of the amber stop codon is about 20% efficient. However, in non-suppressor HB2151 cells, the stop codon is recognized, and protein synthesis stops at the end of the SCD gene so that the g3p fusion protein is not made. At this point, the SCD protein is transported into the periplasmic space (because it contains the leader peptide - see Figure 3-2), but is not assembled into a phage particle since it lacks the gene 3 domain. The SCD protein, if soluble, may accumulate in the periplasm or become incorporated into inclusion bodies. [00173] The expression of the SCD-g3p gene is controlled by an inducible lac promoter present on pCANTAB 5 E. This promoter is in turn regulated by the lac repressor encoded by the lac Iq gene. The g3p fusion protein is not expressed when the lac repressor blocks transcription of the lac promoter by the E.coli DNA-dependent RNA polymerase. Because accumulation of g3p is toxic to the cell, the lac promoter must be tightly controlled prior to infection with M13KO7 helper phage to avoid g3p expression. Thus, strains of E.coli, such as TGl, contain the lac Iq gene. Moreover, because the lac promoter is relatively "leaky", expressions should be performed by growing the transformed cells more slowly at 30⁰C instead of 37⁰C and by adding at least 2% glucose into the medium. Glucose forces the transformed cells to shut down alternate metabolic pathways, further repressing the lac operon. If the transformed cells are grown without glucose at 37°C, g3p will be produced even in the presence of active lac repressor. During M13KO7 infection ("phage rescue"), glucose must be removed from the medium to allow expression of the SCD-g3p fusion protein. Under these conditions, the residual, low-level expression from the repressed lac promoter produces sub-lethal levels of g3p sufficient for phage assembly. Over-expression of either g3p or fusion proteins will kill the infected cells and no phage will be produced. Therefore, the standard method to induce the lac promoter with IPTG, which inactivates the lac repressor must be avoided. [00174] Soluble proteins can be produced both in TGl (suppressor strain) and HB2151 (non-suppressor strain) cells to varying degrees. In E.coli TGl cells, since the suppression of stop codon is only 20% effective, both phage-displayed and soluble recombinant proteins will be produced. The yield of soluble proteins is expected to be higher in HB2151 cells since no fusion proteins are produced in these cells. Thus, we expected that this strain would prove suitable for the production of soluble recombinant proteins. Once having found a desirable stv mutant clone by phage-display, we planned to produce soluble proteins in E.coli HB2151 and detect them using the anti-E tag antibody.

[00175] In this example, our strategy to produce SCD protein and useful mutants was as follows: (1) to clone the appropriate DNA sequence, (2) to generate mutants via error- prone PCR, (3) to clone these into pCANTAB 5 E, (4) to produce phage libraries, (5) to pan the phage libraries with biotinylated magnetic beads and capture the phages encoding proteins with novel properties such as higher binding ability to biotin and biotin-4- fluorescein, (6) to produce phages from singly infected bacterial colonies, and (7) to produce soluble proteins and characterize them.

[00176] Construction of phagemid vector for single-chain dimeric streptavidin: we used a hybrid "phagemid" vector system which combines the advantages of both phage and plasmid vectors (Vieira, J. & Messing, J., 1987, Methods in Enzymology 153, 3-11). A phagemid is a plasmid that includes a plasmid (double-stranded) origin of replication as well as an Ml 3 (phage-derived and single-stranded) origin of replication - also called the major intergenic-region. As with typical plasmids, phagemids carry an antibiotic- resistance marker, such as ampicillin resistance marker, to allow selection and propagation of the transformed cells. Phagemid systems have distinct advantages over direct cloning of the phage. High yields of double-stranded DNA can easily be obtained by simple plasmid preparation. Phagemid genomes can also maintain large DNA inserts more readily than phage genomes. Third, valency (i.e., the number of copies per phage particle) of the displayed fusion protein can be modulated by two-gene display systems (Type 3+3 and 8+8 phagemid and Type 33 and 88 phage systems) (Barbas III, C. F., Dennis R. Burton, Jamie K. Scott, Gregg J. Silverman. (2001). Phage Display A Laboratory Manual. 2.7-2.8; Smith, G. P. & Petrenko, V. A. (1997). Phage display. Chemical Reviews 97, 391-410).

[00177] M13KO7 (Amersham Pharmacia Biotech) and its derivative VCSMl 3 (Stratagene) are commonly used helper phage. The M13KO7 helper phage contains a defective origin of replication (IG region), thus the phagemid DNA gets replicated and packaged more efficiently than the helper phage genome. Therefore, most of the phage produced will contain phagemid DNA. Nevertheless, the M13KO7 origin has sufficient activity to replicate the helper phage genome in the absence of a competing phagemid. Cells infected with the helper phage can be selected specifically over non-infected cells due to the kanamycin resistance gene on the M13KO7 genome. After infection with the helper phage, the phagemid-containing E. coli culture is grown with both ampicillin and kanamycin. Ampicillin selects the cells containing phagemid and kanamycin selects for cells also infected with M13KO7.

[00178] Although both phagemid phage and helper phage bearing the same fusion protein will be affinity-selected during the panning step, only the phagemid phage can drive the production of the displayed fusion protein during the amplification step that follows panning. Therefore, the yields of phagemid phage obtained in selection experiments depend directly on the efficiency of phagemid-phage production (Barbas III, C. F., Dennis R. Burton, Jamie K. Scott, Gregg J. Silverman. (2001). Phage Display A Laboratory Manual. 2.7-2.8).

[00179] The first phagemid we attempted to use to clone the single-chain dimeric streptavidin was pHEN2, a 4.5 kb vector, which had an about 800-bp single-chain antibody (scFv) gene cloned between Sfil and Notl sites. Since Sfil digestion at a single site is never 100% (Nobbs, T. J. & Halford, S. E. (1995). DNA Cleavage at 2 Recognition Sites by the Sfii Restriction-Endonuclease - Salt Dependence of Cis and Trans Interactions between Distant DNA Sites. Journal of Molecular Biology 252, 399-411; Wentzell, L. M., et al, 1995, Journal of Molecular Biology 248, 581-595), removal of the scFv was not efficient enough to prevent obtaining a prohibitive number of background colonies despite trying many different methods to circumvent the problem. [00180] As an alternative choice we picked Amersham pCANTAB 5 E, a commercial phagemid to use for the cloning. This vector was engineered with a "stuffer" kanamycin resistance gene fragment (ca. 1200 bp), that the supplier removed by digestion with Sfil and Notl. Since this digestion can be easily monitored (a 1200 bp fragment can be seen on the gel and removed by gel filtration) the cut vector should be quite "clean." Nevertheless, digestion of pCANTAB 5 E is not absolutely 100% effective and there will be a small percentage of uncut vector in the Sfil/Notl-cut pCANTAB 5 E which may cause "background" colonies (information provided by Amersham Pharmacia technical support). We overcame this problem by increasing the yield for the ligation reaction of my starting fragments (P12(SfiI/BspeI) and P34(BspeI/NotI)) to obtain more Sfil/P1234/Notl product to ligate into pCANTAB 5 E. Our initial ligation reactions of Sfil/P1234/Notl into both pHEN2 or pCANTAB 5 E phagemids failed because the yields of Sfil/P1234/Notl from ordinary ligation reactions of P12(SfiI/BspeI) and P34(BspeI/NotI) were very low. Therefore, we first cloned Sfil/P1234/Notl into pET- 22b(+) (Novagen) whose cloning/expression region of the coding strand is transcribed by T7 RNA polymerase. This provided a reliable means to amplify the fragment. The plasmid pET-22b(+) vector does not carry an Sfil site, however, thus we first had to introduce one into it. After the successful production of Sfil/P1234/Notl in pET-22b(+) we inserted it into pCANTAB 5 E.

[00181] Introduction of Sfil and Notl sites into the two halves of the SCD gene: The gene for the circularly permuted SCD was initially ligated into the pET22 plasmid (3.5 kb) (Novagen-Madison, WI) between the Ndel and HindIII sites. This gene was assembled by combining two pieces each of which was initially cloned into pCR-Blunt vector (Invitrogen-Carlsbad, CA), a 3.5-kb cloning vector with T7 promoter that allows efficient in vitro transcription/translation and has a kanamycin resistance gene for selection of transformants. The first piece, called the "pCR-Blunt vector- 12" (P 12), has an Ndel site at the 5' end and a Bspel site at the 3 'end (3895 bp). The second fragment, "pCR-Blunt vector-34," (P34) has a Bspel site at the 5' end and a HindIII site at the 3' end (3915 bp). To subclone the genes Pl 2 and P34 into either pHEN2 or pCANTAB 5 E phagemids between Sfil and Notl sites, two primer sets (Table 3-1) were designed to replace the Ndel site at the 5' end of P12 with an Sfil site, and to replace the HindIII site at the 3 'end of P34 with a Notl site. PCR was performed as follows: denaturation at 94°C for 3 min, thirty cycles of PCR mutagenesis (94⁰C for 1 min; 58°C for 1 min; 68°C for 1 min) and 68°C for 10 min. Oligonucleotide primers (Table 3-1) were obtained from Integrated DNA Technologies (Coralville, IA). PCR products were confirmed on a 1.3% agarose gel: 395-bp PCR product of P12 [pCR-Blunt-P12(SfiI/BspeI)] obtained with primer sets 1 & 2] and 415-bp PCR product of P34 [pCR-Blunt-P34 (Bspel/NotI)] obtained with primer sets 3 and 4]. 395-bp pCR-Blunt-P12(SfiI/BspeI) and 415-bp pCR- Blunt-P34 (Bspel/NotI) were isolated using the Qiagen gel-extraction kit.

GCCGCCTCCGGAGGTGGATCTGCAGAGGCCGGCATCACC

GGCACCTGGTACAACCAGCTCGGCTCGACCTTCATCGTGA

CCGCGGGCGCCGACGGCGCCCTGACCGGAACCTACGAGT

CGGCCGTCGGCAACGCCGAGAGCCGCTACGTCTG

HindIII

ACCGGTCGTTACGACAGCGCCCCGGCCACCGACGGCTAGA

AGCTT

[00182] Table above shows the DNA sequence of P12(NdeI/BspeI) and P34 (Bspel/Hindlll). CAT is from the vector and CATATG codes for Ndel and AAGCTT codes for HindIII, C was deleted due to a primer design error and was re-inserted by PCR. C became T which changed the codon from alanine (GCC) to valine (GTC). [00183] Table below shows the oligonucleotides used to construct the Sfil and Notl sites. F and R refer to forward and reverse primer, respectively.

[00184] Joining SCD gene fragments in pCR-Blunt vector. pCR-Blunt- P12(SfiI/BspeI) and pCR-Blunt-P34 (Bspel/Notl) were digested with Bspel (TCCGGA) for 2 hours at 37⁰C, then each piece was isolated using the Qiagen gel-extraction kit after electrophoresis on a 1.3 % agarose gel, and the pieces were ligated at 16⁰C overnight with T4 DNA ligase. The 810-bp ligation product was confirmed on a 1.3% agarose gel and extracted. However, attemps to clone it into a pCR-Blunt vector with the Invitrogen Zero Blunt PCR cloning kit and T4 DNA ligase followed by transformation into One Shot TOPlO chemically competent cells (Kanr) were unsuccessful. The colonies from transformation were tested with Sfil and Notl digestions and none of them gave the 815- bp P 1234 and 3.5-kb blunt vector. Therefore, each PCR product was first cloned into the blunt vector prior to ligation in order to get more DNA and increase the ligation efficiency.

[00185] Zero Blunt PCR Cloning kit (Invitrogen) provides cloning of blunt PCR fragments or any blunt DNA fragment with a low background of non-recombinants. Direct selection of recombinants is possible by using the pCR-Blunt vector via disruption of a lethal gene. Since about ninty percent of the PCR products obtained using thermostable, proofreading polymerases will be blunt-ended, they can be ligated directly into pCR-Blunt vector without purification with cloning efficiencies varying from 80% to 95%. This vector contains both the kanamycin and Zeocin resistance genes for selection in E.coli. pCR-Blunt product also has the advantage of being analyzable by a single digestion with EcoRI enzyme since the gene is cloned between two EcoRI sites. Although other host strains may be used, E.coli TOPlO is recommended for general use with pCR-Blunt.

[00186] Cloning the SCD gene into pET22b(+): We subcloned the SCD gene into pET22b(+) to obtain larger amounts of the 810-bp gene. We first inserted an Sfil site into pET22b(+). The pHEN2 vector carries a 907-bp insert between HindIII and Notl sites that contains an Sfil site. We inserted that piece into pET22b(+) to provide the necessary Sfil site.

[00187] Subcloning of SfϊI/SCD/NotI from pET22b(+) into the phagemid vector: pET22b(+) carrying the 810-bp single-chain dimeric gene was digested first with Sfil (double-stranded oligodeoxynucleotides with Sfil sites were used during the digestion) and then with Notl, sequentially. The 810-bp single-chain dimeric gene was gel-purified and then desalted with G-50. We used pCANTAB 5 E as a cloning vector. [00188] Subcloning Sfil/SCD/Notl from pET22b(+) into pCANTAB 5 E: Amersham, pCANTAB 5 E is a 4.5-kb phagemid vector provided precut with Sfil and Notl, SAP- treated and gel-purified. This phagemid is part of the Recombinant Phage Antibody System (RPAS) expression-module kit. The components of this kit are: pCANTAB 5 E, M13KO7 helper phage T, E.coli TGl cells, E.coli HB2151 cells, 1OX One-Phor-All buffer PLUS (1OX OPA+buffer), control insert (AlOB, Sfil/Notl-digested), scFv marker (pUC18/A10B, Sfil/Notl-digested). I ligated about 150 ng of Sfil/Notl-digested, 810-bp SCD with about 250 ng of Sfil/Notl-digested pCANTAB 5 E (4.5 kb). [00189] Transformation efficiencies of both ABLE C and ABLE K cells (tested with pUC18) were between 10⁶-10⁷ cfu/mL. The ligation of the vector to itself gave 4 colonies with ABLE C cells and 10 colonies with ABLE K cells showing that the background level was really low. In addition, ligation of the control insert to pCANTAB 5 E gave about the same number of colonies as the SCD gene, showing that the ligation efficiency was very satisfactory. Ten colonies each were tested from ABLE C and ABLE K cells. [00190] The positive clones identified by Bspel digestion (giving 5.3-kb vector bands) were further confirmed by double digestion with Sfil and Notl (giving 4.5-kb vector and 810-bp insert bands) and triple digestion by SfII₅ Notl and Bspel (giving 4.5-kb vector and ca. 400-bp insert bands). All 20 colonies that were analyzed provided the right size insert and vector. The identities of the samples (two were chosen from each type of host cell) were further confirmed by DNA sequencing using the oligonucleotide primers shown in Table below.

[00191 ] The sequencing revealed that there was a C-deletion upstream of the Notl site in pCANTAB 5 E carrying SCD-stv caused by a faulty oligodeoxynucleotide used to introduce the Notl site. Oligos were re-ordered and C was introduced back into the SCD gene cloned into pCANTAB 5 E by PCR (Table below). pCANTAB 5 E/SCD was used as a template. PCR was performed as follows: denaturation at 94 oC for 3 min, 16 cycles of PCR mutagenesis (94°C for 30 sec; 65°C for 30 sec; 72°C for 6 min) and 72⁰C for 10 min. Final concentrations of primers, dNTPs and Pfu DNA polymerase (Stratagene) were 0.4 μM, 0.2 mM, and 1 μL of 2.5 units/μL, respectively. After confirming the 5.3-kb linearized PCR product on agarose gel, the reaction was treated with Dpnl resriction enzyme at 37°C for 3 hours. After Dpnl treatment, the linear plasmid DNAs carrying the desired mutation were recircularized by blunt-end ligation with T4 DNA Ligase (New England BioLabs) and transformed into E.coli by using ABLE C, ABLE K and TGl competent cells. The numbers of colonies that provided the right size insert (800 bp) and vector (4.5 kb) with ABLE C, ABLE K, and TGl cells were 5,1, and 3, respectively. Sequencing results showed that only the ABLE C strain provided the gene with the C properly re-inserted. TGl cells lacked it. Therefore, ABLE C/pC ANTAB 5 E/SCD was transformed into E.coli TGl cells before phage production. Table below shows the primers used to re-introduce C upstream of the Notl site in pCANTAB 5 E/SCD. Underlined C in forward primer replaced the missing base in the original PCR product. Sequence in bold is the Notl recognition site.

[00192] From this point forward all manipulations involving phage display of the SCD proteins were carried out using pCANTAB 5 E and protocols similar to those just described.

[00193] Phage production from pCANTAB 5 E/SCD: In these procedures plating on minimal medium is important to make sure that the host cells have the f pili that are essential for infection with M13KO7.

Table below shows the results of titer measurements. TGl /pCANTAB 5 E/SCD,

TGl/pHEN2/anti-human thyroglobulin, and TGl/pHEN2from 10³, 10⁵, and 10⁷ X dilutions.

[00194] Screening phage particles by ELISA: Phage-displayed proteins were detected and identified in an enzyme-linked immunosorbent assay (ELISA). Biotinylated BSA was used as a substrate to capture the TGl/pCANTAB 5 E/SCD. Human thyroglobulin (Tg) (0.324 mg/mL) (Sigma; product number BCR-457) was used as an antigen to capture TGl/pHEN2/anti-human thyroglobulin (positive control). Human thyrogobulin (0.324 mg/mL) reacted with TGl/pHEN2 lacking the insert served as a negative control. Binding of phage in ELISA was detected by HRP/anti-M13 monoclonal conjugate (Amersham, product code 279421-01).

[00195] The single-chain dimeric streptavidin cloned into the pCANTAB 5 E phagemid vector was expressed as a phage-displayed recombinant protein and detected and identified in an enzyme-linked immunosorbent assay (ELISA). Biotinylated BSA was used as a substrate to capture the SCD in this experiment. Human thyroglobulin (Tg), used as an antigen to capture TGl/pHEN2/anti-human thyroglobulin, served as a positive control. Human thyrogobulin was used to test TGl/pHEN2 with no insert as a negative control. The signal from the SCD was not as strong as that for the positive control, but it was not as weak as that for negative control. This result suggests that our single-chain dimeric streptavidin binds biotin, but does not bind it as strongly as the control antibody binds to its antigen. Therefore, we began introducing random mutations into SCD by error-prone PCR to form a library of mutant SCDs with different properties that could be displayed on the tip of the phage and panned with biotin-coated magnetic beads. [00196] Error-prone PCR with pCANTAB 5 E/SCD in E.coli TGl cells: pCANTAB 5 E carrying the SCD gene purified from TGl cells was used for three rounds of error-prone PCR to introduce random mutations. Buffer containing 0.5 mM MnCl₂ (Chen, K. Q. & Arnold, F. H., 1993, Proceedings of the National Academy of Sciences of the United States of America 90, 5618-5622) was used instead of MgC12 (1.0 - 2.5 mM) during error-prone PCR. When 2 mM MnCl₂ was used the reaction did not produce any PCR products at all.

[00197] The first set of primers used is shown in the Table showing the oligonucleotides used to construct the Sfil and Notl sites. Table below shows the components of the PCR reaction used with the conditions EPP-I . [00198] We designed a set of primers (Table below) in such a way that the first 10 bases of the forward primer and the first 6 bases of the reverse primer were from the pCANTAB 5 E vector.

[00199] Table below shows the set of primers used for the first round EP-PCR (AMITOF, Allston, MA). In this set of primers the number of the nucleotides from the vector part was 17 for the forward primer and 21 for the reverse primer

[00200] Primer concentrations of 0.1, 0.2, 0.3, 0.4 μM primers were used both with homemade Taq polymerase and commercial polymerase (Promega Taq DNA polymerase, 5 units/μL in storage buffer A; catalog # Ml 865, Promega) at annealing temperatures of 55 and 71.5°C and 10-minute elongation times for the first-round EP-PCR. [00201] The best results were obtained with 0.1 μM primer concentration and Promega Taq polymerase. At this concentration on the first round EP-PCR 100% of the product was 810-bp - there were no 300- or 1200-bp side products. The two annealing temperatures were equally effective. At 0.2 μM primer concentration and an annealing temperature of 55⁰C there was very little 300-bp product, however, there wasn't any at 71.5°C.

[00202] The product from the first-round EP-PCR obtained with 0.1 μM primers and Promega Taq was used for the second round, for which 0.1 μM primers were used with Promega Taq polymerase at elongation temperatures of 55 oC and 71.5°C. The second error-prone PCR provided the right-size product with 0.1 μM primer at both 55⁰C and 71.5°C annealing temperatures. The yield at 55°C was about 50% higher than that at 71.5°C. There were smaller amounts of some lower- and higher-molecular-weight side- products including about 10-20 ng of the 300-bp band. [00203] The 810-bp band obtained from the second round EP-PCR was used for the third round. We used 0.1 μM primers with elongation temperatures of 50, 55 and 71.5°C and Promega Taq polymerase. As a result of the third-round EP-PCR, we obtained about 200 ng of 810-bp PCR product at each annealing temperature. There was about 30 ng of 300-bp side-product as well as some 650-bp (ca. 20 ng) and ca. 20 ng of 1600-bp side- products. However, the yield of 810 bp desired product was sufficient to proceed. The 810-bp band from each annealing temperature (4X 200 ng = 800 ng ) was gel-extracted and purified with the Qiagen gel-extraction kit. This product was first digested with Sfil (50°C for 2 hrs) after which it was desalted with G-50 columns. Then, it was digested with Notl (37°C for 2 hrs) and analyzed on a 1.4% agarose gel from which the 810-bp band was gel-extracted and purified with the Qiagen gel-extraction kit, yielding about 150 ng of mutated SCD genes (SCD-EPP-P3) to be ligated into the pCANTAB 5 E phagemid. [00204] Displaying SCDs on the tip of the phage: Sfil- and NotI-digested, gel- purified DNA, containing single-chain dimeric genes from the third-round error-prone PCR was cloned into the pCANTAB 5 E phagemid vector.

[00205] Formation of the bacterial library: The ligation reaction of pCANTAB 5 E carrying the genes for single-chain dimeric streptavidin from the third-round EP-PCR was transformed into E.coli TGl electro-competent cells (Stratagene) by electroporation. It is common to produce very large libraries by electroporation since most phage-display vectors are designed to be introduced into E.coli in the form of naked DNA (Smith, G. P. & Petrenko, V. A. (1997). Phage display. Chemical Reviews 97, 391-410). [00206] The transformation efficiency was calculated as 9 x 10³ transformants/μg DNA. Titer of this bacterial library was 162 transformants/mL. 8 mL (1296 transformants) was used for phagemid library rescue.

[00207] Rescue of phagemid library: The phagemid libraries were rescued by plating the transformed cells.

[00208] Polyethylene glycol (PEG) precipitation: The amber stop codon (TAG) located between the cloned single-chain dimeric streptavidin and gene 3 sequences of pCANTAB 5 E is suppressed (read through) by an amber suppressor tRNA produced by E.coli TGl cells. However, suppression of the amber stop codon in TGl cells is only about 20% efficient, thus soluble proteins will be produced in addition to phage-displayed recombinant proteins. Soluble proteins, produced 80% of the time, may also compete for biotinylated beads during selection. Therefore, phage carrying displayed proteins were purified away from the soluble proteins by PEG precipitation.

[00209] The size of the phage library that was used for the first-round panning was 3 xlOl l transformants/mL according to the 10⁷ dilution, 2.1xl0¹⁰transformants/mL according to the 10⁵ dilution. The average of these two was taken as final the titer of the phage stock: 1.6 x 10^π transformants/mL. The final yield was 478 mL of PEG- precipitated phage library for the SCD.

[00210] Immobilization of biotinylated BSA on strep tavidin-coupled DYNABEADS®: To perform the panning of the phage libraries against biotin we wanted to use magnetic beads. However, commercially available biotin beads were unsatisfactory; therefore we biotinylated the streptavidin-coupled DYNABEADS®. [00211] Biotin-labeled bovine serum albumin (66 kDa), purchased from Sigma, is readily soluble in water at 1 mg/mL and has about 12 moles (8 -16 moles) of biotin covalently attached per mole of albumin. It was prepared by reacting the purified protein with the N-hydroxysuccinimide derivative of biotin under alkaline conditions in 1 mM sodium citrate, then purified by gel filtration and lyophilized (Sigma technical support). We prepared a 20 μg/mL solution of this material in phosphate buffered saline (IX PBS) at pH 7.4.

[00212] "DYNABEADS® M-280 Streptavidin" was purchased from DYNAL Biotech (Oslo, Norway). Streptavidin was covalently attached to the uniform polystyrene surface of the beads which are superparamagnetic. We suspended beads at a concentration of 10 mg/mL (6.7 x 108 beads/mL) in IX PBS, pH 7.4, containing 0.1% BSA and 0.02% NaN₃. [00213] In the following experiments we assumed a binding capacity of 5 mg biotinylated BSA per 1.0 mg of DYNABEADS®. This represents a conservative limit for binding capacity.

[00214] The streptavidin-binding ability of the biotinylated beads was tested and confirmed with ³⁵S-streptavidin (1.85 MBq, 0.1 μCi/μL, Amersham). Five samples were prepared by mixing biotin-beads (ca. 0.11 μM available biotin) with 0.01 μM, 0.02 μM, 0.03 μM, 0.045 μM and 0.09 μM of 35S-stv in 200 μL of 1 XPBS for 1 hour at 25⁰C. Biotin-beads were washed prior to mixing. Tubes were placed in the Dynal Magnetic Particle Concentrator (MPC) for 1-2 min to separate the beads from supernatant after which 50 μL of the supernatant was taken for the measurements with the liquid scintillation counter. The results imply that the biotinylated beads bind about 50% of the added streptavidin and reach a saturation level. The reason why the biotinylated beads cannot bind to all of the ³⁵S-StV even at a concentration where there is a 156-fold excess of bead-attached biotin over the calculated total number of ³⁵S-StV binding sites could be that a significant fraction of the ³⁵S-StV is denatured, accounting for the counts obtained in the supernatant. Since inactive streptavidin will not be able to bind to biotin, it will remain in the solution.

[00215] To verify that the ³⁵S-StV binding to my biotinylated-BS A-coupled DYNABEADS® M280 Streptavidin ("biotin beads") comes from tight, specific binding by native ³⁵S-StV molecules and not some form of weak, non-specific binding, we performed two control experiments. First, to samples pre-mixed with hot streptavidin and biotin beads, we added 6 times more cold streptavidin than the hot streptavidin present, mixed and incubated for 1-2 hours at 25°C. Beads were separated and ³⁵S-StV in the supernatant was counted. When I compared the average cpm values for free ³⁵S- streptavidin before and after adding cold streptavidin, there was no increase. This strongly suggests that binding by active ³⁵S-StV is not weak, but that our ³⁵S-StV was half dead.

[00216] In a second experiment, we added cold streptavidin incrementally up to one equivalent of hot streptavidin and then chased with ³⁵S-StV. The objective was to demonstrate the disappearance of sites capable of binding hot as a function of the amount of cold (undamaged) streptavidin. A final concentration of 0.9 pmoles/μL of biotin beads was mixed with varying amounts of cold streptavidin (0.005, 0.01, 0.015, 0.023, 0.035, 0.045 pmoles/μL, final concentration) and incubated with shaking at 1200 rpm at 25⁰C for an hour and a half in a total reaction volume of 200 μL. Then, 7 pmoles of hot streptavidin was added to a final concentration of 0.035 pmoles/μL and incubated for another hour in the same manner. The mixtures were placed in a Dynal-MPC for 2-3 minutes to separate the supernatants from the beads, the 50 μL of each supernatant was analyzed by liquid scintillation counting. As the amount of cold streptavidin increased, the amount of bound hot streptavidin decreased (counts in the supernatant - free ³⁵S-StV - increased). Thus the binding of cold streptavidin by biotin beads can block binding by hot streptavidin, implying that this occurs with intact, native ³⁵S-StV and does not simply reflect non-specific adsorption. [00217] Panning for high-affinity SCD mutants using biotinylated beads. We performed PEG precipitation and panning immediately upon rescue of the phagemid library because some phage-displayed recombinant protein preparations may be unstable (Amersham Pharmacia, Expression Module/Recombinant Phage Antibody System manual). PEG-precipitated phage libraries formed by the single-chain dimeric streptavidin exposed to three rounds of error-prone PCR and then cloned to pCANTAB 5 E were therefore used directly for four rounds of panning with biotin beads. [00218] First-round panning: 100 μL of the PEG-precipitated rescued phagemid library with a titer of l.61 x 1011 transformants/mL was used for the first-round panning. After adsorption to biotin beads the non-selected phage were rinsed off, then bound phage were eluted with triethylamine. This fraction, together with the remaining beads, was used to infect TGl cells in preparation for a second panning step. [00219] Second-round panning: Phage obtained from both the supernatant and the beads from the first-round panning were used to create new libraries for the second panning step. From this point on for phage rescue M13KO7 was used. [00220] Second-round panning of phagemid library derived from the supernatant fraction of the first-round panning. Panning followed the same protocol as in the first round. In the second round, 0.5 mL of PEG-precipitated phage resuspended in 2X YT (the whole library) was used with the same amount of biotin beads (1.5 mg which has ca. 660 pmoles of available biotin). However, during the washing steps more stringent conditions were used. Washing was done 20 times with 500 mL PBS+ 0.1% Tween-20 and 20 times with 500 mL of PBS to remove the detergent. Beads were treated with 100 mM triethylamine for 10 min as before to elute the phage. After elution, the supernatant fraction and the remaining beads were stored (in 100 mL of PBS + 0.1%BSA) at 4°C until the infection of E.coli cells.

[00221] Reinfection of E.coli with enriched phage clones from second-round panning of the library from the first-round supernatant was performed. Log-phase E.coli TGl cells were prepared by inoculating 200 mL of the overnight TGl culture into 20 mL 2X YT medium which was incubated at 37°C with shaking at 250 rpm until the culture reached OD600 of 0.3-0.6. 15 mL of an exponentially growing TGl culture was mixed with 1.75 mL of the eluted phage (supernatant) then incubated for 30 min at 37°C without shaking to allow for infection. 200 and 400 mL were plated with no dilution onto separate plates for titering. One mL of an exponentially growing culture of TGl was added to the beads that had been eluted with triethylamine and then incubated for 30 min at 37⁰C without shaking to allow for infection. 50 mL was plated without dilution on a plate to titer the beads.

[00222] For the preparation of the master plates the remaining cells were centrifuged at 330O g for 10 min. The pellet from the 10-mL supernatant-derived culture was resuspended in 1 mL 2X YT and plated onto two plates, 0.5 mL each. The pellet derived from the beads was resuspended in 0.5 mL plated all on one plate. The plates were incubated at 30°C overnight.

[00223] From the bead fraction 50 μL plated directly from the 1 mL culture gave 6 colonies, i.e. -120 transformants/mL of TGl cells. Direct plating of 200 and 400 μL from the 16.75-mL supernatant fraction gave 59 and 51 colonies, respectively. The average titer determined according to these plates was 2 x 10³ transformants/mL. The master plates from beads and the supernatant were used for rescue of the phagemid library for the third-round panning.

[00224] Second-round panning of phagemid library derived from the bead fraction of the first-round panning. One mg of biotin beads (ca. 440 pmoles of available biotin in 200 mL of PBS + 0.1% BSA) was mixed with 0.5 mL of PEG-precipitated phage from the bead fraction of first-round panning and incubated by shaking at room temperature for 2 hrs.

[00225] Reinfection ofE.coli with enriched phage clones from the beads of the second round panning. Log-phase E.coli TGl cells were prepared as before. 15 mL of an exponentially growing culture of TGl was mixed with 1.8 mL of the eluted phage and incubated for 30 min at 37⁰C without shaking to allow for infection. 200 and 400 mL were plated with no dilution onto separate plates for titering. After elution, beads (stored in 100 μL of PBS with 0.1% BSA) were mixed with 1 mL of an exponentially growing culture of TGl and incubated for 30 min at 37°C without shaking to allow for infection. 50 μL of this culture was plated without dilution to titer the bead library. For the preparation of the master plates the remaining cells were centrifuged at 3300 g for 10 min. The pellet from the 16.8-mL supernatant-derived culture was resuspended in one mL 2X YT and plated onto two plates, 0.5 mL each. The pellet derived from the beads was resuspended in 0.5 mL plated all on one plate. The plates were incubated at 30°C overnight.

[00226] Result of second-round panning from beads: Infection of 1 mL E.coli TGl cells with the beads gave 54 transformants/niL of TGl cells. From the supernatant directly plating of 200 and 400 μL from the 16.8 mL gave 1 and 4 colonies, respectively. Average titer determined according to these plates was 70 transformants/mL. The master plates from beads and the supernatant were used for rescue of the phagemid library for the third-round panning.

[00227] Production of phage clones from single infected bacterial colonies from second-round panning: From the bead-derived colonies obtained by second-round panning and rescue (starting with the bead fraction from first-round panning) we selected five to express clonally and test for desirable mutant SCD genes. We prepared frozen stocks of the PEG-precipitated phage (labeled Pl - P5) and determined their titer. Titers ranged from 4 x 10¹⁰ to 2 x 1011 cfu/mL indicating successful production of adequate amounts of mutant phage.

[00228] Screening phage clones P1-P5 by ELISA: Binding of phage to biotin in ELISA was detected by HRP/anti-M13 monoclonal conjugate. The results of the ELISA directed at biotin showed that the phages had binding at least as good as the non-mutant. We proceeded to grow cultures from single bacterial colonies, extracted the DNA and analyzed it by DNA sequencing. However, the sequences obtained did not relate to the sequence of streptavidin. These phages were produced from the third-round error-prone PCR and second-round panning of the beads, that led us to initially believe that too many mutations had been introduced. However, based on subsequent screening, the problem most likely stems from the choice of phage clones derived from the bead fraction after panning.

[00229] Analysis of mutants derived from various stages of error-prone PCR and panning: Rather than systematically explore the entire combinatorial tree of error-prone PCR and panning steps, we picked conditions that might prove favorable on the basis of all the information gathered up to this point. Four sets of conditions were used. [00230] Mutants from first- round error-prone PCR and first-round panning: I generated a new bacterial library from first round EP-PCR (710 cfu/mL) and carried out a single panning step on 2.8 x 10⁶ cfu of the phagemid. The solution fraction gave 13 colonies and the bead fraction 50-100 (non-uniform) colonies. DNA from ten of the bead-derived colonies, cut with Sfil and Notl failed to provide any clear bands on agarose gel electrophoresis. Gel ectrophoresis showed that four of the colonies from the solution had the right-size insert and the vector. These were analyzed and DNA sequencing confirmed that all (S3, S9, SlO, SI l) were related to the streptavidin sequence. Phage were produced from the four positive clones and PEG-purified stocks yielded titers in the range of lθ^lo-lθ' ' cfu/mL. The binding ability of these phage to biotin was tested with ELISA and there was blue signal both from the mutant phage and the positive control SCD-non-mutant. However, the signals from all four phage were weaker or similar to that of SCD-non-mutant.

[00231] We also performed ELISA assays on these mutants using goat anti- streptavidin antibody (Zymed) with a tetrameric streptavidin control. The goat anti- streptavidin assays gave weak binding for all four clones — indistinguishable from that of the NM-SCD (and, as shown by DNA sequencing, two of the phagemid clones recovered from panning were, in fact NM-SCD). Given that the bead fraction yielded no meaningful sequences even after only one round of EP-PCR, we concluded that this was not due to excessive mutation, but had some other cause. Thus for subsequent experiments we chose third-round EP-PCR products and used only the supernatant fractions from panning.

[00232] Mutants from third-round error-prone PCR and second-round panning: PEG-precipitated phage obtained from the supernatant of the second rescued phagemid library was exposed to second-round panning. The supernatant was used to infect E.coli cells. Eleven colonies from the master plate were analyzed: SAl, SA3, SA5, SA7, SA8, SA9, SAl 3, SAl 4, and SAl 6 (irregular numbers indicate omission of colonies that did not provide the right size insert (810 bp) as a result of Sfil and Notl digestion). These colonies were double-digested with Sfil and Notl and confirmed by DNA sequencing. Titers of these phages were between 10ⁿ-10¹² transformants/niL. Their binding ability to biotin was tested with ELISA directed at biotinylated-BSA and there was blue signal both from the phages and the positive control non-mutant. However, the signals from all of these nine phages were weaker or similar to that of non-mutant.

[00233] The ELISA directed at goat anti-streptavidin with tetrameric streptavidin as a positive control gave a strong blue color at A₄₅₀ of about one. However, the phages SAl, SA2, S A3, SA4, and SA5 from the second-round panning of the solution gave weaker signals with A₄₅₀ between 0.06 - 0.15.

[00234] Since our purpose was to select for high-affinity clones, we decided to try two additional rounds of panning to further enrich the clones.

[00235] Mutants from third-round error-prone PCR and third-round panning. The master plate was obtained from the solution part of second-round panning derived from solution part of the first-round panning and rescue of the third phagemid library by plating was performed as before.

[00236] Washing was done 30 times with 500 mL PBS+ 0.1 % Tween-20 and 10 times with 500 mL of PBS to remove the detergent. At step 8, incubation of beads with 100 mM triethylamine was performed for 20 minutes. The enriched phage clones were used for the infection of E. coli TGl cells. Titer of the supernatant phage solution was about 3 xlO⁴ transformants/niL. Ten colonies were analyzed by agarose gel electrophoresis and confirmed by DNA sequencing. The phage were produced from single infected bacterial colonies, then purified with PEG precipitation as before and checked for their titer.

[00237] DNA sequence of P 12(NdeI/BspeI) and P34 (Bspel/Hindlll). CAT is from the vector and CATATG codes for Ndel and AAGCTT codes for Hindlll. C was deleted due to a primer design error and was re-inserted by PCR. C became T which changed the codon from alanine (GCC) to valine (GTC)).ELISA directed to biotinylated BSA showed that the binding ability of these mutants to biotin is similar to or a little weaker than that of NM-SCD. DNA sequencing revealed that they fell into several sets within each of which all sequences are identical (Table 3-15).

[00238] Table 3-15. Sets of identical mutants from third-round panning.

[00239] ELISA directed to goat anti-streptavidin IgG-HRP showed A450 about 1 for tetrameric streptavidin, but gave background values for the mutants (A450 about 0.07 - 0.1). [00240] Mutants from third-round error-prone PCR and fourth-round panning:

The master plate was obtained from the solution part of third-round panning derived from the solution part of the second-round panning. The procedure followed was the same as in the previous section except that beads were washed only 10 times with PBS + Tween- 20. The final library titer was 6.7 xlO⁴ transformants/mL.

[00241] To obtain a wider selection of mutants, we switched to a 96- well plate to screen colonies from this (final) panning step.

[00242] This panning step gave more satisfactory results than the earlier ones. There were ~ 48 phages, which gave much higher signals than that of the non-mutant SCD. We chose eight mutants for further analysis and characterization (Table 3-17). These eight candidates were first analyzed by Sfil and Notl digestions and all of them gave the right- size vector (4.5 kb) and insert (810 bp) on gel electrophoresis and were confirmed by DNA sequencing. Differences from the amino acid sequence of the non-mutant SCD were determined from the DNA sequencing results.

[00243] Table 3-17. ELISA results from 96-well plate for the phages from fourth- round panning (third-round error-prone PCR) directed to biotinylated BSA. Note that the sequence position numbers here correspond to the SCD construct (and hence do not match native streptavidin sequence numbers). DNA sequences of non-mutant single- chain dimeric streptavidin (NM-SCD), C2 mutant, and E2 mutant were as follows: [00244] NM-SCD between Ncol and Notl sites:

[00245] CCATGGAGGCCAACGCCAAGAAGTCCACGCTGGTCGGCCACGACA CCTTCACCAAGGTGAAGCCGTCCGCCGCCTCCGGTGGTGGATCCGCCGAGGC CGGCATCACCGGCACCTGGTACAACCAGCTCGGCTCGACCTTCATCGTGACCG CGGGCGCCGACGGCGCCCTGACCGGAACCTACGAGTCGGCCGTCGGCAACGC CGAGAGCCGCTACGTCCTGACCGGTCGTTACGACAGCGCCCCGGCCACCGAC GGCAGCGGCACCGCCCTCGGTTGGACGGTGGCCTGGAAGAATAACTACCGCA ACGCCCACTCCGCGACCACGTGGAGCGGCCAGTACGTCGGCGGCGCCGAGGC GAGGATCAACACCCAGTGGCTGCTGACCTCCGGCACCGGCTCCGGAACCGCC CTCGGTTGGACGGTGGCCTGGAAGAATAACTACCGCAACGCCCACTCCGCGA CCACGTGGAGCGGCCAGTACGTCGGCGGCGCCGAGGCGAGGATCAACACCCA GTGGCTGCTGACCTCCGGCACCACCGAGGCCAACGCCTGGAAGTCCACGCTG GTCGGCCACGACACCTTCACCAAGGTGAAGCCGTCCGCCGCCTCCGGTGGTG GATCTGCAGAGGTCGGCATCACCGGCACCTGGTACAACCAGCTCGGCTCGAC

CTTCATCGTGACCGCGGGCGCCGACGGCGCCCTGACCGGAACCTACGAGTCG

GCCGTCGGCAACGCCGAGAGCCGCTACGTCCTGACCGGTCGTTACGACAGCG

CCCCGGCCACCGACGGCGCGGCCGCACTCGAGCACCACCACCACCACCAC

(SEQ ID NO: 10).

[00246] C2 in pCANTAB 5 E between Sfil and Notl sites:

[00247] GGCCCAGCCGGCCATGGAGGCCAACGCCAAGAAGTCCACGCTGGT

CGGCCACGACACCTTCACCAAGGTGGAGCCGTCCGCCACCTCCGGTGGTGGA

TCCGCCGAGGCCGGCATCACCGGCACCTGGTACAACCAGCTCGGCTCGACCC

TCATCGTGACCGCGGGCGCCGACGGCGCCCTGACCGGAACCTACGAGTCGGC

CGTCGGCAACGCCGAGAGCCGCTACGTCCTGACCGGTCGTTACGACAGCGCC

CCGGCCACCGACAGCAGCGGCACCGCCCTCGGTTGGACGGTGGCCTGGAAGA

ATAACTACCGCAACGCCCACTCCGCGACCACGTGGAGCGGCCAGTACGTCGG

CGGCGCCGAGGCGAGGATCAACACCCAGTGGCTGCTGACCTCCGGCACCGGC

TCCGGAACCGCCCTCGGTTGGACGGTGGCCTGGGAGAATAACTACCGCAACG

CCCACTCCGCGACCACGTGGAGCGGCCAGTACGTCGGCGGCGCCGAGGCGAG

GATCAACACCCAGTGGCTGCTGACCTCCAGCACCACCGAGGCCAACGCCTGG

AAGTCCACGCTGGTCGGCCACGACACCTTCACCAAGGTGAAGCCGTCCGCCG

CCTCCGGTGGTGGATCTGCAGAGGTCGGCATCACCGGCACCTGGTACAACCA

GCTCGGCTCGACCTTCATCGTGACCGCGGGCGCCGACGGCGCCCTGACCGGA

ACCTACGAGTCGGCCGTCGGCAACGCCGAGAGCCGCTACGTCCTGACCGGTC

GTTACGACAGCGCCCCGGCCACCGACGGCGCGGCCGCAGGTGCGCCGGTGCC

GTATCCGGATCCGCTGGAACCGCGTGCCGCATAG (SEQ ID NO: 11);

[00248] C2; Translation of 846 bp in pCANTAB 5 E without Sfil site and with E-tag:

281 aa : 28901.3 Da; pi: 5.20

[00249] MEANAKKSTLVGHDTFTKVEPSATSGGGSAEAGITGTWYNQLGSTLI

VTAGADGALTGTYESAVGNAESRYVLTGRYDSAPATDSSGTALGWTVAWKNN

YRNAHSATTWSGQYVGGAEARINTQWLLTSGTGSGTALGWTVAWENNYRNAH

SATTWSGQYVGGAEARINTQWLLTSSTTEANAWKSTLVGHDTFTKVKPSAASG

GGSAEVGITGTWYNQLGSTFIVTAGADGALTGTYESAVGNAESRYVLTGRYDSA

? ATOGAkAGAPVPYPDPLEPRAA (SEQ ID NO: 12) (E-tag is shown in bold and italics, 16 aa, and sequence derived from Notl (AA) in bold and underlined). [00250] E2 in pET22b(+)between Ncol and Notl sites with a His-tag:

[00251 ] CCATGGAGGCCAACGCCAAGAAGTCCACGCTGGTCGGCCACGACA

CCTTCACCAAGGTGAAGCCGACCGCCGCCTCCGGTGGTGAATCCGCCGAGGC

CGGCATCACCGGCACCTGGTACAACCAGCTCGGCTCGACCCTCATCGTGACC

GCGGGCGCCGACGGCGCCCTGACCGGAACCTACGAGTCGGCCGTCGGCAACG

CCGAGAGCCGCTACGTCCTGACCGGTCGTTACGACAGCGCCCCGGCCACCGA

CGGCAGCGGCACCGCCCTCGGTTGGACGGTGGCCTGGAAGAATAACTACCGC

AACGCCCACTCCGCGACCACGTGGAGCGGCCAGTACGTCGGCGGCGCCGAGG

CGAGGATCAACACCCAGTGGCTGCTGACCTCCGGCACCGGCTCCGGAACCGC

CCTCGGTTGGACGGTGGCCTGGAAGAATAACTACCGCAACGCCCACTCCGCG

ACCACGTGGAGCGGCCAGTACGTCGGCGGCGCCGAGGCGAAGATCAACACCC

AGTGGCTGCTGACCTCCGGCACCACCGAGGCCAACGCCTGGAAGTCCACGCT

GGTCGGCCACGACACCTTCACCAAGGTGAAGCCGTCCGCCGCCTCCGGTGGT

GGATCTGCAGAGGTCGGCATCACCGGCACCTGGTACAACCAGCTCGGCTCGA

CCTTCATCGTGACCGCGGGCGCCGACGGCGCCCTGACCGGAACCTACGAGTC

GGCCGTCGACAACGCCGAGACCCGCTACGTCCTGACCGGTCGTTACGACAAC

GCCCCGGCCACCGACGGCGCGGCCGCACTCGAGCACCACCACCACCACCAC

(SEQ ID NO: 13)

[00252] Conclusion. Single-chain dimeric streptavidin which is a covalently attached dimer, was used to form a library of mutants by error-prone PCR after which it was cloned into a phagemid vector to be used for phage display. Panned phage libraries provided some interesting mutants like C2 and E2 (Table 3-17) among others. These two mutants were chosen for further analysis and characterization. Example 2

[00253] Expression, Purification and Characterization of SCD-Streptavidin and SCD-Streptavidin Mutants. Phage display-selected SCD mutants that gave highest binding abilities according to ELISA results were used to produce phage from single, infected bacterial colonies. These phages were then used to infect E.coli HB2151 cells (a non-supressor strain in which the stop codon is recognized) to produce soluble proteins. However, due to a leaky lac promoter and the toxic features of streptavidin, expression in this strain of SCD proteins cloned in the phagemid was not successful. Therefore, genes for SCD proteins were subcloned into a pET vector which expresses them using T7 RNA polymerase under the tight regulation of the T7 promoter. This system allowed successful production of SCD proteins. Cells expressing each mutant were tested to find the cellular location of the engineered protein. Experiments showed that they all formed inclusion bodies which could be dissolved in Gu-HCl and folded. Structure and stability of the proteins were analyzed by FPLC chromatography and MALDI-TOF mass spectroscopy. Then their biotin-binding abilities were tested, first with ultrafree-MC centrifugal filtration units and ³H-biotin, and subsequently by fluorescence polarization, which allowed quantitative measurement of the dissociation constants. The results confirmed successful production of intact single-chain dimeric streptavidins and serendipitously revealed the potential utility of biotin-4-fluorescein as an affinity ligand for possible technological applications of these proteins. Small-scale expression of SCD proteins

[00254] Preparation of phage for mutants A8, C2, C6, D2, E2, E6, E9, F7: Production of phage from single, infected bacterial colonies was performed as before in section G4c of Chapter 3 and PEG precipitation as described before. Their titers are given in Table 4- 1. Average titer was l.lxlO¹ ' pfu (ampr)/mL, which was a very good result.

[00255] Infection of E.coli HB2151 cells with NM-SCD and mutants A8, C2, C6, D2, E2. Because sequencing results showed that mutants C6, E6 and E9 had the same change (at a single position), further work was confined to C6. Similarly, since C2 and F7 had the same sequences, only C2 was used. E.coli TGl produces a suppressor tRNA which allows suppression (read-through) of the amber stop codon, located between the SCD and gene 3 sequences of pCANTAB 5 E, though only with 20% efficiency. Thus, during the infection of E.coli TGl cells soluble proteins will be produced as well as phage-displayed recombinant proteins. Therefore, small-scale quantities of soluble proteins can be produced in TGl cells and screened as an alternative before immediately proceeding to use non-suppressor tRNA E.coli HB2151 cells. However, since the yield of the soluble proteins may be significantly higher in the HB2151 non-suppressor strain, we omitted the TGl -cell screening and proceeded directly to use recombinant phages produced from individual clones to infect E.coli HB2151 cells for larger-scale production of soluble recombinant proteins.

[00256] Because E.coli strain HB2151 infects quite poorly, we used selection of transductants on medium containing ampicillin, glucose, and nalidixic acid (SOBAG-N) to ensure that the resulting colonies were true nalr (nalidixic acid-resistant) HB2151 transductants and not carryover infected TGl cells.

[00257] Transfer of phagemids to HB2151 cells: HB2151 cells were streaked from their frozen glycerol stocks on fresh minimal medium plates and grow overnight at 37°C. Asingle colony from HB2151 cells was inoculated into 5 mL of 2X YT medium and incubate overnight at 37⁰C with shaking at 300 rpm.

[00258] Preparation of log-phase HB2151 cells: 1 : 100 (200 μL of overnight- grown HB2151 cells diluted with 20 mL of fresh 2X YT medium) was inoculated, and incubated at 37°C with shaking at 300 rpm until the culture reaches an OD600 of 0.3-0.5. 400 μL of the log-phase culture was transferred into a separate 10-mL Falcon tube for each clone. [00259] For each clone, 40 μL of recombinant phage was added (no PEG precipitation is required since HB2151 cells are a non-suppressor strain and no recombinant proteins will be produced), to 400 μL of log phase HB2151 cells, which were incubated with intermittent gentle shaking for 30 min at 37⁰C. 100 μL was plated on individual SOBAG- N plates from each clone and incubate overnight at 30⁰C. There were colonies on all plates, demonstrating successful transfer of the phagemids to HB2151. [00260] Localization of SCD proteins in E.coli HB2151 cells: The objective was to test the ability of HB2151 to serve as the host for production of the mutant SCD proteins. Since the subcellular location of the expressed protein product could not be predicted in advance, we examined all three principal compartments: extracellular solution, periplasmic space and inclusion bodies.

[00261] We tested NM-SCD and all five independent mutants A8, C2, C6, D2 and E2 for protein production in this system.

[00262] Soluble protein: Two colonies were inoculated from NM-SCD, A8, C2 (≡ F7), C6 (≡ E6 s E9), D2, E2 from SOBAG-N plates from the infection of HB2151 cells, in 10 mL of 2X YT medium supplemented with lOOμg/mL ampicillin and 2% glucose, and grown overnight at 3O⁰C with shaking at 250 rpm. 5 mL from the overnight cultures was added into 50 mL of freshly prepared 2X YT-AG medium, that was incubated for 1 hr at 30°C with shaking at 250 rpm. The culture was centrifuged at 1500 g for 20 minutes at 25⁰C and the supernatant was removed from the sedimented cells and discarded and the sedimented cells were resuspended in 50 mL of freshly prepared 2X YT-AI (NO GLUCOSE) and incubate for at least 5 hours at 30°C with shaking at 250 rpm. (final cone. Of IPTG = I mM).

[00263] The culture was divided into two separate centrifuge tubes and centrifuged at

1500 g for 20 minutes at 25⁰C and the supernatants were carefully removed (supernatants contain the extracellular soluble proteins) from both pellets and transferred to a single clean container. The supernatants were recentrifuged for 5 more minutes at 1500 g and then filter through a 0.45 μm filter and stored at 4°C. One pellet was reserved for preparation of the periplasmic extract and one pellet for preparation of the whole cell extract.

[00264] We next applied the following protocol to one of the pellet fractions of induced HB2151 cells.

[00265] Periplasmic extract: One of the two cell pellets obtained from above were resuspended in 0.5 mL of ice cold IX TES (100 mL IX TES buffer: 20 mL of 1 M Tris-

HCl (pH 8) (0.2 M Tris-HCl); 100 μl of 0.5 M EDTA (pH 8) (0.5 mM EDTA); 50 mL 1

M sucrose; Complete to 100 mL with deionized H2O; Filter with 0.22 μm pore-size filter and store at 2-8°C). 0.75 mL of ice-cold 0.2X TES was added and vortex to resuspend and incubated on ice for 30 min. The tubes were centrifuged at 13,000 rpm (MTX-150-

TOMY High Speed Micro Refrigerated Centrifuge, Induction Drive, Palo Alto, CA) for

10 minutes and the supernatant, which contains the soluble proteins from the periplasm, was carefully transferred to a clean tube, and stored at 4°C.

[00266] Two different methods, one based on boiling, the other on lysozyme digestion were used to prepare whole-cell extract from induced HB2151 cells.

[00267] Whole-cell extract (by boiling): One of the pellets obtained above was resuspend in 0.5 mL of PBS and boiled for 5 minutes, the cell debris was pelleted by centrifugation at 13,000 rpm (MTX- 150) for 10 minutes and the supernatant, which contains the intracellular soluble proteins, was transferred to a clean tube, and store at

4⁰C.

[00268] Whole-cell extract (with lysozyme): One of the pellets was resuspended in

0.5 mL of PBS. We added protease inhibitor cocktail for bacterial cells (Sigma no. P

8465, one mL of the cocktail solution is recommended for the inhibition of the protease activity found in 20 mL of cell lysate from 4 g (wet weight) of E. coli cells), 3.34 μL of

50 mg/niL lysozyme (final cone. 1 mg/mL), 2 μL of 1 M MgC12 (final cone. 12 mM), 0.16 μL of 10 mg/mL DNase I (final cone. 10 μg/mL) and mixed the mixture and incubated it at room temperature for 15-30 minutes, and then freezed and thawed it three times. The mixture was centrifuged at 13,000 rpm (MTX-150) for 20 minutes and passed carefully through a 0.45 μm filter (Millex-HA ultracleaning membrane filter, Millipore no. SLHA 025 OS) and then the filtrate, which contains the intracellular soluble proteins, was transferred to a clean tube and store at 4°C.

[00269] To have a comparison standard for the preparation of SCD-mutant proteins when expressed in HB2151 cells, we used above described protocols (including induction with IPTG) to produce the appropriate subcellular fractions from untransformed HB2151 cells.

[00270] ELISA analysis: ELISA analysis of sub-cellular fractions from IPTG- induced HB2151 cells directed at HRP/anti-E tag conjugates for the soluble fraction, periplasmic extract and whole-cell extract of the mutants and for the negative control HB2151 was performed in an attempt to localize the expressed proteins. [00271] SCD proteins contain a C-terminal 13-amino-acid peptide epitope tag (E-tag) which is recognized by an anti-E-tag monoclonal antibody. Therefore, HRP/anti-E-tag conjugate was used in ELISA assays to quickly identify E.coli colonies expressing soluble, antigen-positive E-tagged SCDs. Because ELISA assays can be quantitative, when used together with HRP/anti-E-tag conjugate, one can detect and assess the expression levels of SCD proteins in different cell fractions. [00272] ELISA assay: A 96-well plate was coated with 100 μL of the soluble fraction, periplasmic extract and whole-cell extract from cells expressing the mutant proteins and from negative control HB2151 cells using duplicate wells for each mutant and incubated overnight at 4⁰C. The overnight incubated plates were washed and the sites were blocked with BSA. The plates were coated with 100 μL of diluted HRP/anti-E tag conjugates (dilution ratio 1 :8000) prepared in blocking buffer (3% BSA-PBS) and incubated the plated at room temperature for an hour, and washed three times with PBS- 0.05% Tween 20, then 3 times with PBS.

[00273] Results of ELISA directed to HRP/anti-E tag conjugate with the soluble fractions of all the mutants did not yield any positive signal, meaning these soluble fractions do not contain detectable amounts of SCD mutant proteins. However, periplasmic extracts of all the mutants gave positive signals (A₄₅₀ ~ 0.13-0.6), indicating binding to the HRP/anti-E tag conjugate. The strongest signal was obtained with C2 (A₄₅o ~ 0.6) and E2 (A₄₅₀ ~ 0.5). Whole-cell extracts (boiled preparations) of all the mutants except E2 (A₄₅o ~ 0.08) gave positive signals (A₄₅₀ ~ 0.3-0.5; for negative control A₄₅₀- 0.07). The strongest signal was again obtained with C2 (A₄₅₀ ~ 0.5). Lysozyme preparations for all the mutants gave very strong signals. The one for E2 mutant was especially strong A₄₅₀ ~ 0,8; the others were around A₄₅₀ ~ 0.4-0.5. Negative control and background had A₄₅₀ ~ 0.05.

[00274] However, the anti-E-tag HRP-linked antibody should recognize both folded and unfolded proteins. For that reason, we also tested the ability of the mutants detected in the periplasmic extracts and the whole-cell extracts to bind to biotinylated BSA. Soluble fractions which gave negative signals with HRP/anti-E-tag conjugate were used as negative controls during this assay. Unfolded protein should show little or no affinity to the biotin-linked adsorbant. Plates were coated with 100 μL of 20 μg/mL biotinylated BSA solution prepared in IX PBS buffer, using duplicate wells for each mutant and negative controls (HB2151 periplasmic extract and whole-cell extract) and incubated overnight at 4°C. After the wells were washed with IX PBS and then with 3% BSA-PBS to block the non-specific binding sites, 100 μL of each of the cell fractions from the corresponding mutants was used to coat the wells. Then, binding was tested with HRP/anti-E-tag conjugate.

[00275] ELISA analysis directed to biotin and detected with HRP/anti-E tag showed that the solution fractions from all of the mutants tested had no activity (confirming the results from the direct HRP/anti-E-tag assays). There was no activity in the periplasmic extracts and in the whole-cell extracts prepared with boiling for mutants A8, C6 and D2. Mutants A8, C6 and D2 extracted from whole cells with lysozyme treatment showed very weak activities with ELISA directed to biotin (background A450 = 0.06; mutants = 0.1- 0.16). However, periplasmic extracts of mutants C2 and E2 showed some activity (A₄₅₀ ~ 0.2). While the C2 mutant prepared from whole-cell extracts with boiling showed some activity (A₄₅₀ ~ 0.25), mutant E2 had no activity (A₄₅₀ ~ 0.06) compared to background signals (negative control A₄₅₀ ~ 0.05).

[00276] In conclusion, samples prepared from whole cells by boiling showed activity only for the C2 mutant; the rest of the mutants gave background readings. Among the mutants prepared by lysozyme treatment, E2 had the highest binding activity (A450 ~ 0.75) for biotin while A8, C2, C6 and D2 gave similar signals (A₄₅₀ ~ 0.1 - 0.16). [00277] The non-mutant SCD is present in low amounts in the soluble fraction (A450 ~ 0.2), and whole-cell extracts prepared by boiling (A₄₅₀ - 0.18) as shown by weak signals with HRP/anti-E-tag. However, when it was prepared from periplasmic extracts (A₄₅₀ ~ 0.6) or by lysozyme treatment ( A₄₅₀- 0.78) it gave much stronger signals. ELISA analysis of NM-SCD directed to biotin and detected with HRP/anti-E-tag showed that soluble fractions, periplasmic extracts and whole-cell extracts prepared by boiling gave no signal (A₄₅₀ ~ 0.06). The whole-cell extracts prepared by using lysozyme gave much higher levels of binding activity (A₄₅₀ ~ 0.21). These results imply that the NM-SCD is present in the cytosol in the active form.

[00278] Scaling-up expression of SCDs in E. coli HB2151 cells. Small-scale pilot purification of the SCD-mutants C2 and E2 on anti-E-tag column. The matrix for the anti-E-tag column (Amersham Biosciences) contains a mouse monoclonal antibody specific for the 13-amino-acid E-tag located upstream of the stop codon in the SCD gene constructs. This matrix was produced by coupling the antibody to N-hydroxysuccinimide- activated, high-performance Sepharose and optimized so as to produce a stable covalent binding of the ligand, ensuring long column life. Non-specific cross reactions with E. coli proteins were minimized by careful selection of the mouse antibody. The column can be used at least 20 times if one follows the instructions for its care. [00279] Anti-E-tag column chromatography: The extract was dialyzed from the periplasmic and whole-cell extracts (from lysozyme digestion) containing the SCD proteins against one liter of binding buffer [0.2 M phosphate buffer, pH 7, with 0.05 % NaN3 and 1 mL of 100 mM PMSF (was added freshly) to a final concentration of 0.1 mM] in 3-15 mL Slide-A-Lyzer dialysis cassette (10 K MWCO membranes, Pierce) for 2-3 hrs to adjust the pH of the extract to 7. 100 mL of IX binding buffer was prepared (0.2 M phosphate buffer, pH 7; 0.05 % NaN3 from the stock 1OX binding buffer. The 1OX elution buffer (1.0 M glycine, pH 3) was diluted by adding 3 mL of elution buffer to 27 mL of distilled water. Collection tubes were prepared by adding 500 μL of neutralizing buffer (1 M Tris, 0.05 % NaN3, pH 8.2) for each 5-mL fraction to be collected from the anti-E tag column. The column was equilibrated with 15 mL of elution buffer and then immediately with 25 mL of binding buffer. The sample was applied using the syringe at a flow rate of ca. 5 mL/min. (To increase the binding the sample was re-loaded 2-3 times slowly.) The column was washed with 25 mL of binding buffer at ca. 5 mL/min to remove the excess unbound E.coli proteins. The bound SCD proteins were eluted from the anti-E-tag column with elution buffer. The first 4.5 mL of material eluted from the column was discareded, generally, it will not contain a significant amount of SCD mutant. The following 5 mL - which should contain the purified SCD proteins - was collected in one fraction. The absorbance of the anti-E-tag purified C2 and E2 was measured at 280 nm and the column was immediately re-equilibrated with 25 mL binding buffer.

[00280] To detect the purified proteins with HRP/anti-E tag conjugate, a 96-well ELISA plate was coated with 3X 100 μL from the 5-mL eluted fractions; 3X 100 μL of the first 4.5-mL eluted fractions; and 2X 100 μL of the loaded sample. The plates were incubated at 4⁰C overnight.

[00281] To assess the biotin-binding ability of the detected proteins, a 96-well ELISA plate was coated with 100 μL of 20 μg/mL biotinylated BSA solution prepared in IX PBS buffer, using three wells for each sample. After the wells were treated with IX PBS and 3% BSA-PBS as before, 100 μL of each of the fractions was used to coat the wells. The plates were incubated at 4°C overnight and test the binding with HRP/anti-E-tag conjugate.

[00282] ELISA results of C2 and E2 mutants, after purification on the anti-E-tag column, directed to HRP/anti-E tag conjugate showed positive binding (A₄₅₀ ca. 1 ; negative control = 0.06). ELISA results directed to biotin gave A₄₅₀ = 0.3 with C2 mutant and A₄₅₀= 1 with E2 mutant, indicating some activity.

[00283] Large-scale purification of mutants C2 and E2 on anti-E-tag column: SCD mutants were expressed with 1 mM IPTG following Protocol 4.5 for whole-cell extract with lysozyme (scaled-up for 1.0-L cultures). ELISA against HRP/anti-E-tag conjugate and biotin before purification and after purification with anti/E-tag-column gave positive signals with both mutants (A₄₅₀ Ca. 1-1.2; negative control ca. 0.06), Mutants were first purified with anti-E tag column and then run on FPLC. [00284] FPLC chromatography of crude C2 (obtained from the last 5-mL elution of the anti-E tag column) from expression in one liter gave 6 peaks. The major peak (#3) was at 12.91 mL. FPLC of E2 from one-liter expression gave the major peak at 13.75 mL and a minor peak at 16.71 mL, both in trace amounts . The retention volumes of protein standards on this column were 10.17 mL for tetrameric streptavidin (Mr = 60 kDa), 12.22 mL for chymotrypsinogen A (Mr = 25 kDa), and 13.35 mL for ribonuclease A (Mr = 13.7 kDa).

[00285] Peak obtained from FPLC of the C2-mutant in HB2151 (Figure 4-1) was analyzed by ELISA directed to biotin. A₄₅₀ readings of the peaks were as follows: #1 = 1; #2 = 1; #3, first half = 0.9; #3, second half = 0.9; #4 = 0.35. Peaks #5 and #6 as well as the negative control all gave A₄₅₀ = 0.06 (same as the blank).

[00286] SDS-PAGE of the first half of the third peak from FPLC showed some faint bands with sizes between 67 and 14.4 kDa. However, the second half of the third peak from FPLC gave a single clean peak at ~ 14.4 kDa which has almost the same migration distance as tetrameric streptavidin. The second peak-obtained from FPLC did not provide any major band. The first peak which is probably an aggregate obtained from FPLC showed many bands on SDS-PAGE sizes between 94 - 14.4 kDa. The fourth, fifth and sixth peaks from FPLC did not provide any bands on SDS-PAGE gel. [00287] Mutant E2 behaved similarly. SDS-PAGE of the fractions from the anti-E-tag column (before purification with FPLC) showed a band around 14.4 kDa (between 14.4 and 20 kDa), another one right above this one (ca. 20 kDa) and another one little above 30 kDa. After FPLC purification of this sample, the only band observed on SDS-PAGE was the one at ca. 14.4 kDa. The yield of E2 (major peak on FPLC) was estimated as 2 μg/L E.coli culture (according to SDS-PAGE). ELISA results directed to HRP/anti-E-tag conjugate and biotin with FPLC-purified E2 showed positive binding both to HRP/anti-E- tag and to biotin.

[00288] Binding ability of C2 (0.6 μg/μL-estimated by SDS-PAGE) and E2 (0.02 μg/μL-estimated by A280) mutants in HB2151 which showed single band at 14.4 kDa on SDS-PAGE after FPLC purification was also checked with ³H-biotin by using ultrafree- MC centrifugal filtration units (molecular mass cut-off 10 kDa from Millipore). 42, 210, and 420 pmoles (monomer) of C2-mutant was mixed with 10 pmoles of ³H-biotin and 1.5, 7.5, and 15 pmoles (monomer) of E2-mutant was mixed with 10 pmoles of ³H-biotin in total volumes of 100 μL. In both cases no binding was detected. [00289] The binding abilities of C2 and E2 mutants in HB2151 were also tested with PDlO columns packed with Sephadex G-25 (Amersham). Fifteen pmoles of C2 was mixed with 0.002 μCi/μL ³H-biotin and run on the column. In a parallel experiment 110 pmoles of mutant E2 was mixed with 0.002 μCi/μL ³H-biotin and chromatographed. C2 showed a small amount of binding, however, E2 did not.

[00290] The results of SDS-PAGE and FPLC showed that the mutants C2 and E2 as expressed in HB2151 exist as monomer-sized polypeptides. To verify that the cloned sequences of these mutants after transfer to HB2151 cells had not changed, the C2 and E2 phagemid inserts in HB2151 cells were confirmed by DNA sequencing. [00291] Expression of mutants in HB2151 in rich medium with biotin: Expression of SCD mutants in HB2151 did not produce an intact dimeric protein perhaps because we were using lethal genes with a leaky system that had incomplete repression caused by the lac promoter. The toxicity of these genes derives from the depletion of free biotin in the cell due to its tight binding to the mutant streptavidins. Depletion of free biotin inhibits biotin-dependent carboxylases, decarboxylases and transcarboxylases (Fall, R. R. (1979), Analysis of Microbial Biotin Proteins. Methods in Enzymology 62, 390-398). Inactivation of these enzymes blocks the first step of fatty acid biosynthesis and affects amino acid metabolism, gluconeogenesis, replenishment of the Krebs cycle, and uptake of substrates by some anaerobes (Szafranski, P., et al., 1997, Proceedings of the National Academy of Sciences of the U. S. A. 94, 1059-63). Under such sub-optimal physiological conditions cells probably synthesize proteins slowly making them more vulnerable to proteolysis before complete folding. They may also express higher levels of proteolytic enzymes. As a first step to tighten the regulation of their expression systems, we repeated the expressions under rich nutrient conditions - i.e., in 2X YT medium containing biotin to a final concentration of 50 μg/mL (0.2 mM) (Szafranski et al., Id.) since killing occurs by depletion of biotin.

[00292] Expression from HB2151 in rich medium: A single colony from a fresh E2/HB2151 plate was inoculated into 25 mL of 2X YT supplemented with 100 μg/mL ampicillin and 2% glucose (final concentrations) and grown by shaking 8 hrs to OD600 = 0.5-0.7. 10 mL of the above culture was transferred into each of two 50-mL Falcon tubes (one to use with biotin, the other without biotin) and centrifuged at 3500 rpm with a swinging bucket rotor for 10 minutes. The pellet was resuspended in a Falcon tube and transferred into 1-L flask with 200 mL of 2X YT supplemented with ampicillin to a final concentration of 100 μg/mL and in one flask biotin to final concentration of 50 μg/mL (0.2 mM). The OD600 of each suspension was measured. The cultures were shaken at 30°C and 250 rpm till they grew to an OD600 of ca. 0.5 before induction. Before induction 1 mL from each culture was saved and stored at 0-4°C as a zero-time control and another 1 mL from each culture was removed as an uninduced control to shake as long as the induced ones.

[00293] Treatment of the samples for SDS-PAGE: Samples were centrifuged down at 13,000 rpm in a MTX 150 rotor and the pellet was resuspended in 20 μL water. 20 μL of 2X SDS gel loading buffer was added. 10 μL per well was loaded on SDS-PAGE gels. 100 mM of IPTG was added to a final concentration of 1 mM. One mL from each culture was saved after the first hour, then every half hour for three more times, and were shaken until four hours after induction. After four hours' induction, the cultures were centrifuged at 6000 g (Sorvall RC-5B Plus, fixed angle GSA rotor) for 20 minutes. The supernatant was transferred into a clean flask and kept at 4⁰C. The pellets were weight, and 4 mL of IX TES and 250 μL of protease inhibitor cocktail were added per gram of pellet to each tube. Pellets were esuspend by vortexing. 25% TES was prepared in water and 4.8 mL of it was added to each tube, vortexed and incubated on ice for 40 minutes. Samples were centrifuged at 13,000 rpm for 10 minutes in an MTX-150 rotor and the supernatant and pellet were saved separately at 4⁰C. 4-20% (VWR) i-gel was run to monitor the expression.

[00294] The SDS-PAGE gels did not reveal any over-expressed bands as a result of using biotin. Hence this approach did not cure our problem with the expression levels of the single-chain dimeric streptavidin mutants.

[00295] Expression of NM-SCD, C2 and E2 in HB2151 with BL21 -Codon Plus RP cells. Production of heterologous proteins in E.coli depends on the availability of certain tRNAs which are abundant in the organisms from which the heterologous proteins are derived. If the protein that is to be expressed at high levels contain codons for which sufficient corresponding tRNAs do not exist in E.coli, the translation will be stalled since high-level expression of foreign proteins will deplete the pool of rare tRNAs. BL21- CodonPlus strains carry extra copies of genes encoding the tRNAs that most frequently limit translation of heterologous proteins in E.coli. Availability of tRNAs allow many heterologous recombinant genes to be expressed at high-levels in BL21-CodonPlus cells (which otherwise are poorly expressed in other BL21 strains). BL21-CodonPlus-RP cells contain extra copies of the argU and proL genes which encode tRNAs that recognize the arginine codons AGA and AGG and the proline codon CCC, respectively. The CodonPlus-RP strains carry the tRNAs that most frequently restrict translation of heterologous proteins of organisms that have GC-rich genomes such as S. avidinii. Moreover, these cells can be used for protein expression with vectors driven by T7 promoters (induced by the CE6 bacteriophage) as well as with vectors driven by non-T7 promoters.

[00296] Since SCD proteins have arginine codon bias, we repeated their expression with Stratagene, BL21 -Codon Plus RP competent cells in an attempt to overcome this bias. Different IPTG concentrations (0.03 mM, 0.5 and 1 mM) and induction times (2, 4, 6 hrs and overnight) were used both at 25°C and 30°C.

[00297] Expression from BL21 - Codon Plus RP cells: Transformed cells were plated on SOBAG (100 μg/mL ampicillin and 2% glucose) plates and incubated overnight at 30°C. Induction was performed with IPTG concentrations of 0.03, 0.06, 0.125, 0.25 and 0.5 mM both with BL21CodonPlus RP cells, and HB2151 cells. Expression after 2, 4, 6 hrs and overnight was monitored for BL21CodonPlus RP cells (at 30 oC) and 3.30, 6, 8 hrs and overnight for HB2151 cells (both at 25°C and 3O⁰C) with ELISA directed to anti-E-tag antibody. Plates were covered with 100 μL of cells and HB2151 cells as negative control and incubated overnight at 4°C. [00298] Expression of NM-SCD, C2 and E2 with BL21-CodonPlus-RP at 30°C or 25⁰C under different IPTG concentrations (0.5 and 1 mM) did not yield successful results. As induction time increased the OD450 decreased (after expressing 2 hrs OD450 = ca. 0.7 -0.8; after expressing 4 hrs OD450 = ca. 0.5 -0.6; after expressing 6 hrs OD450 = ca. 0.3 -0.5; after expressing overnight OD450 = ca. 0.1.).

[00299] Expressions in SCD/BL21 Codon Plus RP, ClI BL21 Codon Plus RP, and E2/ BL21 Codon Plus RP were also tried after reducing the concentration of glucose from 2% (w/v) to 0.2% (w/v) or adding 0.05% (v/v) glycerol during induction (Donovan, R. S., et al., 2000, Canadian Journal of Microbiology 46, 532-541; Su, Y. C, et al., 2003, Journal of Biochemistry and Molecular Biology 36, 493-498) with 0.05 mM IPTG to increase the protein expression level. Cultures were grown at 30⁰C to increase the production of correctly folded functional proteins (Glick, B. R. (1995). Metabolic Load and Heterologous Gene-Expression. Biotechnology Advances 13, 247-261; Su, Y. C, et al., 2003, Journal of Biochemistry and Molecular Biology 36, 493-498). OD600 were compared before induction, and after inducing for one to five hours. Before induction OD600-C2 = 0.35; OD600-E2 - 0.6; OD600-non-mutant = 0.6. After about an hour of induction the OD600 increased about 0.1-0.3 units. After 2 hrs, most OD600s stayed the same. After 2.5 hrs only non-mutant's OD600 increased to around 1 ; the others stayed the same. After 3 hrs, all remained the same. After total of 5.5 hrs all had OD600 of about 0.6.

[00300] SDS-PAGE gels of these expressions in BL21-CodonPlus-RP cells did not provide any over-expressed band either at monomeric size (ca.14.4 kDa) or at dimeric size (ca. 30 kDa).

[00301] Production of the SCD proteins in E.coli HB2151 cells: A single-chain dimeric streptavidin mutant E2, obtained from third-round error-prone PCR and from the rescue of the phage library (from TGl cells) by fourth-round panning, was used to infect E.coli HB2151 cells, induced with IPTG and expressed for 5 hrs. The protein product was purified by anti-E-tag column and FPLC. On the FPLC column this mutant acted as a monomer, however, it bound to biotin (as detected by ELISA with biotinylated BSA) and when the DNA from the HB2151 cells was sequenced, it still contained the gene for an intact dimer. Activity of the protein was checked by ELISA directed to biotinylated-BSA at the following steps and all gave positive results:in TGl cells; in HB2151 cells; after expression; after purification by anti-E-tag column; and after purification by FPLC. [00302] Expression of the mutants in HB2151 both at 3O⁰C and 25°C in 2X YT medium supplemented with 2% glucose was monitored from 1 to 10 hrs and C2 mutant from 1 to 8 hrs. with different IPTG concentrations. The SDS-PAGE of these expressions in HB2151 cells did not provide any overexpressed band either at 14.4 kDa (as a monomer) or at 30 kDa (as a dimer). Figure 4-4 shows the result of monitoring E2 expression. Mutant C2 gave a similar result. OD600 values of the cultures expressed for ca. 2-3 hrs gave ca. 0.6-0.8 after which it either decreased or did not change as the induction time was increased (OD600 ~ 0.4 -0.6 after 8 hrs)

[00303] MALDI analysis gave monomeric size. The sequences obtained from Edman sequencing was used for a BLAST search which gave the sequence of lysozyme. [00304] Subcloning of single-chain dimeric streptavidins into pET22b(+). Low-level expression can happen due to toxic or unstable proteins, or when the expression construct is not maintained in the cells during growth. In addition, such proteins may cause "leaky" expression before induction under conditions of poor culture growth or they may slow down the growth of the bacteria containing the correct plasmid (mutants with deletions may grow faster). In spite of the fact that the streptavidin gene is extremely lethal to the host cells, it can be expressed efficiently by using T7 RNA polymerase/T7 promoter expression systems (Sano, T. & Cantor, C. R. (1990). Proceedings of the National Academy of Sciences of the U. S. A. 87, 142-6; Studier, F. W. and Moffatt, Barbara A., 1986, Journal of Molecular Biology 189, 113-130; Studier, F. W., et al., 1990, Methods in Enzymology 185, 60-89.; Szafranski, P., et al., 1997, Proceedings of the National Academy of Sciences of the U. S. A. 94, 1059-63). Therefore, to express my single-chain dimeric streptavidin and its mutants we switched to the T7 expression system which in the past has allowed efficient production of various streptavidin mutants (Sano, et al., 1997, Methods in Molecular Biology 63, 119-28).

[00305] The cloning and expression region of the coding strand in pET22b(+) vector is transcribed by T7 RNA polymerase/T7 promoter expression systems to direct tightly regulated high-level expression. When this is used together with an E.coli expression strain like BL21-Gold(DE3)pLysS (Stratagene) which lacks the Lon and OmpT proteases that can degrade recombinant proteins, toxic genes can be expressed successfully as shown in our work.

[00306] Subcloning SCD genes into pET22b(+): pCANTAB 5 E carrying the single- chain dimeric streptavidin mutants (NM-SCD, C2 and E2) and pET22b(+) vector carrying the P1234 gene were simultaneously digested with Ncol and Notl at 37⁰C for 4 hrs. The 810-bp bands for NM-SCD, C2 and E2 and a 5.5 -kb band for the pET22b(+) were gel extracted and the pET22b(+) was treated with shrimp alkaline phosphatase (USB), at 37⁰C, and SAP was heat-inactivated at 65°C for 15 minutes. The inserts were ligated into pET22b(+) with Fast-link DNA ligation kit (Epicentre) for 10 minutes at room temperature. The ligase was heat-inactivated at 7O⁰C for 15 minutes. 2 μL of the ligation reactions was transformed with 50 μL of Transformax EClOO electrocompetent cells (Epicentre) and plated on LB/Ampicillin (100 μg/mL) and incubate at 3O⁰C overnight; The transformants were analyzed with Ncol and Notl digestions and monitore on 1% agarose gel. The constructs were confirmed by DNA sequencing. Each mutant was transformed with BL21-Gold(DE3)pLysS chemically competent cells (Stratagene). [00307] The transformants provided the 810-bp insert and the 5.5-kb vector from double digestion with Ncol and NotI. The transformants were further confirmed by DNA sequencing. Afterwards, each mutant was transformed into BL21-Gold(DE3)pLysS chemically competent cells (Stratagene).

[00308] Expression of single-chain dimeric streptavidin mutants sub-cloned into pET22b(+) with 1 mM IPTG and purification from inclusion bodies: Two transformants were analyzed from each mutant in BL21-Gold(DE3)pLysS chemically competent cells (Stratagene). First expressions were performed on an analytical scale by expressing 1 mL culture at 37°C and 1 mM IPTG and 3O⁰C with 0.5 mM IPTG. Both experiments provided the over-expressed single-chain dimeric gene and did not have noticeably different yields. In addition, NM-SCD was expressed for 3 hours, and no difference in expression level was observed between 2 and 3 hours. [00309] The result of the expression in 1 mL cultures monitored on SDS-PAGE showed that the single-chain dimeric streptavidin mutants were successfully expressed as intact dimers (~ 30 kDa protein) under the control of T7 promoter. Thus, we chose to perform subsequent expressions at 37°C for two hours with 1 mM IPTG. [00310] We next used 50-mL culture volumes to find out if the expressed proteins occur in solution, in the periplasmic extract or as inclusion bodies in the cytosol. Cells induced with 1 mM IPTG for 2 hrs were centrifuged at 5,000 g (Sorvall RC-5B Plus with fixed angle SS-34 rotor). Both supernatant and pellet fractions were kept for further analysis. The pellet was used first to check the periplasmic extract, then the inclusion bodies (cf. Protocols 4-3 & 4-5). The results showed that all the SCD proteins ended up in the inclusion bodies. Since the final yield was low, I used four liters of culture to express each mutant.

[00311] The procedure below is described for a culture volume of 500 mL and can be scaled up or down as needed.

[00312] Preparation of frozen lysate: BL21(DE3)Gold(pLysS) carrying pET22b(+) as an expression vector encoding single-chain dimeric streptavidin mutants were grown from a single colony overnight at 37°C at 250 rpm in 50 mL of LB supplemented with 100 μg/mL ampicillin and 50 μg/mL chloramphenicol. Approximately 25 mL of each fresh culture of BL21(DE3)Gold(pLysS) carrying pET22b(+) as an expression vector (1:20) was added into 500 mL fresh LB broth containing no selection antibiotics. The cultures were incubated with shaking at 250 rpm at 37°C for two hours. 1 mL of each culture was pipetted into a clean microcentrifuge tube and placed on ice before gel analysis. These served as the non- induced control samples. Another 1 mL of each culture was pipetted into a clean microcentrifuge tube and no IPTGwas added, but the sample was shaken alongside the IPTG-induced cultures for the same period of time. To the rest of the culture (498 mL) in each flask 5 mL of 100 mM IPTG (final cone. 1 raM) was added and incubate with shaking at 250 rpm at 37°C for 2 hrs. 1-mL sample after one hour of induction was taken and after the 2-hour induction period the cultures were placed on ice. 1 mL of each of the induced cultures was pipetted into a clean tube and first centrifuged then resuspended in 40 μL of water and then 40 μL of 2X SDS gel-loading buffer was added, and 40 μL of this solution was run on a Gradipore 4-20% iGel (Frenchs Forest NSW, Australia) Tris-glycine-SDS precast polyacrylamide mini-gel. The induced cells were centrifuged down at 6000 g (Sorvall RC-5B Plus, fixed angle GSA rotor) for 15-20 min. The cell pellet (inclusion bodies) was washed briefly with 100 mL of 100 mM NaCl, 10 mM Tris-HCl pH 8.0, 1 mM EDTA. The sample was centrifuged at 6000 g (Sorvall RC-5B Plus, fixed angle GSA rotor) for 15 minutes. The supernatant was discarded (Sano, T., et al., 1997, Methods in Molecular Biology 63, 119-28). The cell pellet was suspend in ca. 50 mL of 2 mM EDTA, 30 mM Tris-HCl pH 8.0, 0.1% Triton X-100 with vigorous shaking to lyse host cells. This cell lysate can generally be stored at - 70°C until used (Sano et al., Id.). If after adding resuspension buffer and vortexing, the pellet forms a lump and does not get resuspended and can be simply stored at -70°C. [00313] Preparation of inclusion bodies from lysate: The stored lysate was thawed, freezed and thawed three times with vortexing to insure complete lysis. The sample was sonicated for 30 seconds on ice (on power setting 4), with a pause of 30 sec, then sonicated again for about 30 sec. This was repeated three times. The sample was centrifuged at 13,000 rpm (Sorvall RC-5B Plus, fixed angle SS-34 rotor) for 10 minutes (the supernatant which has the cytosolic fraction was also kept) and resuspend in 10 mL of wash buffer (50 mM Tris-HCl, pH 8.0, 0.5% Triton X-100, 1 mM EDTA, 100 mM NaCl, 0.1% NaN₃) by sonicating as above, sample was again centrifuged, and the above- steps of sonication and centrifucation was repeated three times. After the last wash the supernatant was discarded and the pellet washed once with 10 mL of wash buffer without Triton X-100; the sonication and centrifucation steps were repeated once more and the resuspended pellet was centrifuged at 13,000 rpm (Sorvall RC-5B Plus, fixed angle SS-34 rotor) for 10 minutes. The pellet was resuspended in 100 - 500 μL of deionized water (i.e., 0.2-1 mL H₂O /6 L culture). Once the white paste was formed, the pellet was resuspended in 2.5 mL of 6 M Gu-HCl, 50 niM Tris-HCl, pH 7.5 (if the pH of folding buffer was 7.5) or 7 M Gu-HCl, pH 1.5 (30 mL of Gu-HCl / 6 L of culture). The inclusion bodies were incubated at 4⁰C overnight to make sure that resuspension is complete. After vortexing, to ensure that the paste dissolves completely the sample was centrifuged at 13,000 rpm (Sorvall RC-5B Plus, fixed angle SS-34 rotor) for ten minutes at 4⁰C to remove the insoluble particles. Next step was to proceed with folding. [00314] Folding of the single-chain tumeric proteins in pET22b(+) from inclusion bodies: This folding protocol was modified from (Chu, et al., 1998, Protein Science 7, 848-859). The first folding was done at pH 6, and provided proteins with good binding ability to ³H-biotin as determined by using ultrafree MC centrifugal filter devices (MW cut off = 10 IcDa - 20 μL from the bottom chamber for NM-SCD, C2, E2 and control reaction with no protein gave 49 cpm, 42 cpm, 42 cpm, and 413 cpm, respectively). However, the yield of the folded protein at pH 6 was low ~ 40 μg/L. We therefore used different pH's, taking into account that pi (C2) = 5.34 and pi (E2) = 5.79 to find the optimum pH for folding. 50-mL culture volumes were used to test the folding at different pH values. The following buffers were used for pH's 4.5, 6, and 7.5: pH 4.5: 50 mM NH₄OAc, 150 mM; NaCl, 5 mM EDTA, 0.1 mM PMSF; pH 6: 50 mM MES, 150 mM NaCl, 5 mM EDTA; 0.1 mM PMSF; pH 7.5: 50 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, 0.1 mM PMSF. Each 250 μL of inclusion bodies, obtained from 50 mL of culture and dissolved in Gu-HCl, were injected into three separate folding buffers (25 mL) by using 18 /4 G needle dropwise while stirring forcefully at 4⁰C. The stirring was continued overnight. One-tenth of the total volume of the concentrated protein was taken to check the ³H-biotin-binding ability of the samples folded at different pHs as before, using ultrafree MC centrifugal filter devices (MW cut off = 10 kDa). [00315] Proteins displayed binding to biotin at all three pH values. The highest yield obtained was at pH 7.5 as monitored on SDS-PAGE. Yield at pH 7.5 was about 80 μg/L; pH 6 = 40 μg/mL, at pH 4.5 = 20 μg/mL. The yield from the inclusion bodies was about 1 mg/(L culture). After folding it went down to about 80 μg/L (at pH 7.5) meaning that about 92% of the protein aggregated. [00316] After folding protein in pH 7.5 buffer, Gu-HCl concentration was reduced to two different concentrations: 80 mM and 80 μM. Injection of 250 μL of inclusion bodies in 7 M Gu-HCl into 50 mL folding buffer dilutes the Gu-HCl to 80 mM. This protein solution can be injected directly onto the FPLC column for purification without further desalting. We also diluted the solution containing 80 mM Gu-HCl with Centripreps and Centricons by washing with buffer several times so as to decrease the final Gu-HCl concentration to 80 μM. At both concentrations, the proteins preserved their biotin binding ability. Therefore, after folding, the protein can be concentrated with or without dilution and applied on the FPLC column.

[00317] Experiments done to optimize the folding conditions showed that the best result was obtained by dissolving the inclusion bodies either in 6 M Gu-HCl, 50 mM Tris- HCl, pH 7.5 or 7 M Gu-HCl, pH 1.5, followed by folding in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, with freshly added 0.1 mM (or 0.2 mM) PMSF, 0.02% Tween-20 (omitting Tween-20 apparently had no significant effect). [00318] Because the highest yield was obtained at pH 7.5, the rest of the folding experiments were done at this pH. These conditions were used to prepare protein from a 4-L culture volume by a suitable scale-up.

[00319] Inclusion bodies obtained from expression of SCD proteins tended to be difficult to filter through 0.45 μm filters. Therefore, prior to injection into the folding buffer they can be centrifuged at 13,000 rpm (MTX- 150 rotor) to try to eliminate as much as possible of the insoluble particles which otherwise would decrease the yield of folding. [00320] We folded inclusion bodies from 500-mL culture volumes in 250 mL folding buffer and concentrated the solutions to total volumes of 200 μL. A 1-μL sample (corresponding to ~ 10 mL culture volume) of each was mixed with 1 μL of 1200 cpm/μL ³H-biotin and 98 μL folding buffer. The control sample lacked protein. After 1 hr of incubation at 4°C with 300 rpm shaking, these were transferred to 10,000 nominal- molecular-weight-limit (NMWL) ultrafree-MC centrifugal filter devices (Millipore) and centrifuged at 5,000 rpm (MTX-150) for 5 minutes. A 20-μL aliquot from the bottom chamber was counted and compared with control samples. Since the controls have only ³H-biotin (no protein), they have higher counts in the bottom chamber. For example, the bottom chamber of the filter device with NM-SCD protein gave 35 cpm, and that of the control sample gave 272 cpm. NM-SCD protein and mutants C2 and E2 all gave very similar results: retention of about 90% of the added ^JH-biotin, meaning that they had significant binding activity.

[00321] We also analyzed the periplasmic fraction and showed that the proteins were mostly present in the inclusion bodies.

[00322] Gel-filtration purification and analysis of NM-SCD and mutants E2 and C2. We prepared the mutant proteins using FPLC as implemented on the Amersham Pharmacia LKB-Controller LCC - 501 Plus instrument. The column used was Tricorn high performance column Superdex 75 10/300 GL.

[00323] Eluates were collected in 0.5 mL fractions. AUF = 0.1-0.5 (0.2), sample loop was 200 μL, flow rate was 0.5 niL/min. Running buffer: 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA.

[00324] FPLC results of mutant E2. Gel-filtration column chromatography of crude E2/pET22b(+)/ BL21Gold(DE3)pLysS showed four peaks with retention volumes of: 7.78 mL, 10.90 mL, 13.25 mL, and 17 mL (Figure 4-6). Comparison of the retention volumes of these peaks with those of known standards revealed that the peak at retention volume 10.90 mL is likely a tetramer, and the one at 13.25 mL a dimer. The peak at 7.78 mL (void volume) is probably an aggregate. Identity of the 17-mL peak was not known (it appears consistently in every chromatogram and without wishing to be bound by a theory, we suggest that it is an artifact). The ligand binding ability of each fraction was tested with ³H-biotin in ultrafree-MC centrifugal devices. Tetramer and dimer fractions both showed appreciable binding. The aggregate was collected in two fractions, the first of which showed no biotin binding ability. The second aggregate fraction showed moderate binding, probably due to partially refolded protein occluded in the aggregate. The peak at 17 mL showed no biotin-binding. From this point on only tetramer and dimer peaks were used for further characterization.

[00325] Stabilities of the dimer and tetramer peaks were tested by re-running these peaks on the FPLC column both in the absence and presence of biotin. The column was pre-equilibrated with buffer containing 10 μM biotin and biotin was added to the dimer fraction (which had 9.2 μM monomeric binding sites) to a final concentration of 19.2 μM. The tetramer fraction, containing 7 μM monomeric sites, was mixed with biotin to a final concentration of 17 μM. Protein samples with added biotin were incubated at 4⁰C for 30- 60 min before chromatography. Both in the absence and presence of biotin the tetramer peak eluted at the same volume. The dimer peak, rechromatographed in the absence of biotin, provided two peaks: tetramer and dimer. However, upon adding biotin to the dimer peak and re-running it on a column pre-equilibrated with biotin, it gave only a tetramer peak. Aggregate and 17-mL peaks were re-run on the column in the absence of biotin and all eluted at the same elution volumes.

[00326] FPLC results of mutant C2. Gel-filtration column chromatography of crude C2/pET22b(+) BL21Gold(DE3)pLysS showed four peaks: at 8 mL, 11 niL, 12.5 mL, and 17 mL (Figure 4-7). Comparison of the retention volumes of these peaks with those of known standards revealed that the peak at retention volume 11 mL is likely a tetramer, and the one at 12.5 mL a dimer. The peak at 8 mL (void volume) is probably an aggregate. The identity of the 17-mL peak is not known. Stabilities of the dimer and tetramer peaks were tested by re-running them on the FPLC column both in the absence and presence of biotin. The column was pre-equilibrated with buffer containing 100 μM biotin and the dimer fraction (which had ~ 200 nM monomeric binding sites) was incubated with ~ 106 μM of biotin before loading on to the column. For the tetramer fraction, 4 μM monomer was incubated with 117 μM of biotin. Protein samples with added biotin were incubated at 4⁰C for 3-4 hrs. Both in the absence and presence of biotin the tetramer peak eluted at the same volume. The dimer peak, re-chromatographed in the absence of biotin, provided two peaks: tetramer and dimer. However, incubation of the dimer peak with biotin and re-running it on a column pre-equilibrated with biotin provided only the 17-mL peak, probably due to very low concentration of starting material.

[00327] FPLC results of NM-SCD. Gel filtration chromatography by FPLC of crude SCD-NM showed 7 peaks: at 8.53 mL, 9.45 mL, 11.33 mL, 17.31 mL, 19.05 mL, 20.29 mL, 21.80 mL (Figure 4.8). Comparison of the retention volumes of these peaks with those of known standards revealed that the peak at retention volume 11.33 mL is likely a tetramer. The peaks at 8.53 mL and 9.45 mL (void volume is ca. 8 mL) are probably aggregates. The rest is lokely be some low-molecular weight aggregates which we did not investigate. We suspected that one of the two shoulders on the right hand side of the 11.33 mL peak might be a dimer. Therefore, we collected those two fractions and ran them on the gel column again (Figure 4-9). This provided a peak at 11.3 mL (we speculated this to be the tetramer) and 12.5 mL (we speculated this to be the dimer). However, the MALDI result from the 12.5 niL peak did not provide expected dimer mass of 28 kDa, but well-defined m/e peaks near 8 and 10 kDA. Even with ziptip cleanup, no reasonable spectrum was obtained for the 11.3 mL peak obtained from rerunning the two- shoulder peak.

[00328] Stability of the NM-SCD-tetramer peak was tested by re-running this peak on the FPLC column both in the absence and presence of biotin (Figure 4-8). The column was pre-equilibrated with buffer containing 100 μM biotin and the tetramer fraction (which had 5 μM monomeric binding sites) was incubated with 1 11 μM biotin at 4°C for 2-3 hours. The tetramer peak eluted at the same volume both in the absence and presence of biotin.

[00329] How does the NM-SCD tetramer fraction behave in the absence of biotin? Since biotin-binding apparently stabilized the tetrameric form of the E2 mutant, the question arises as to whether or not the NM-SCD tetramer might be similarly stabilized. To test this we used streptavidin-coated DYNABEADS® to strip out any remaining biotin from this fraction. One mg of DYNABEADS® can bind at least 650 pinoles of biotin. A 6.5-mg batch of beads was first washed three times with 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA - each wash was done with 650 μL. Beads were resuspended in 600 μL of the same buffer used for washing. NM-SCD-tetramer fraction (40 μL, 1.05 x 105 nM containing 4210 pmoles of monomer, 1.5 μg/μL) was mixed with the streptavidin-coated magnetic beads and incubated overnight at 4°C. The overnight- incubated bead solution was placed in the Magnetic Particle Concentrator (MPC) and the supernatant was transferred into a clean tube. The second tube was also placed in the Dynal separator, after which the supernatant was pipetted out and transfered into a 2-mL Centricon and concentrated down to ~ 200 μL. This entire sample was run on a gel column with the FPLC system. The protein eluted at the same retention volume (11.19 mL) as a tetramer (panel d in Figure 4-8). This experiment confirmed that the tetramer fraction behaves as a tetramer in the absence of biotin both with and without treatment with DYNABEADS® (Rv = 11.19 mL and 11.29 mL, respectively, Figure 4-8b and d). [00330] Mass spectrometry (MALDI-TOF-MS) characterization of purified SCD proteins. Having had the pure proteins by FPLC, matrix-assisted laser desorption time- of- flight mass spectrometry (MALDI-TOF-MS) was used to determine the mass of NM- SCD, C2 and E2. Results were obtained on an Applied Biosystems Voyager-DE STR (Framingham, MA) in the Microchemistry Laboratory at Harvard University. [00331] Theoretical weights of NM-SCD, C2 and E2 are 28,376.8; 28,434.7; 28,499.9 Da, respectively as calculated from their sequences by using the ProtParam tool which was available online and which is now published at Gasteiger E., Hoogland C, Gattiker A., Duvaud S., Wilkins M. R., Appel R.D., Bairoch A.;Protein Identification and Analysis Tools on the ExPASy Server; (In) John M. Walker (ed): The Proteomics Protocols Handbook, Humana Press (2005) pp. 571-607, ISBN: 1-58829-343-2. These masses include the mass of the N-terminal methionine and the his tag (six histidines at the C- terminus).

[00332] The spectra were taken with an externally calibrated instrument. The range of calibration (or mass accuracy) depends on the external calibration conditions, measured mass, the location of the spot on the plate, the external calibrants used, the width of the peaks (adducts cause wider peaks), etc. The measured masses will not typically deviate from the true value more than 150 Da at the 28 kDa mass range (i.e., they will have ca. 0.5% error). The error can be much lower (around 50-70 Da). Mass accuracy can only be determined by repeated measurements. In addition, internal calibration can be used to bracket the mass being measured to further increase the accuracy. [00333] In the following text, masses of the mutant proteins are given in ascending order and their differences from the lowest mass are given in parentheses. The peaks with an excess mass of 224 Da may reflect the presence of sinapic acid (MW = 224 Da) which has a tendency to form adducts (Beavis, R. C. & Chait, B. T. (1996). Matrix-Assisted Laser Desorption Ionization Mass-Spectrometry of Proteins. Methods in Enzymology 270, 519). Multiple sinapic acid adducts might also explain some multiple peaks in the spectra. However, as noted below, the mass differences do not correspond well to multiples of 224 (or 206 - the dehydration form), perhaps due to the inability of the mass spectrometer to call out the right masses among closely spaced peaks. The peaks near 14 kDa (very close to half the size of the dimer) probably arise from the +2 ion of the parent peak at ca. 28 kDa.

[00334] MALDI-TOF-MS for NM-SCD-T and NM-SCD-D. NM-SCD (28,376.8 Da) cosists of two covalently attached ca. 14-kDa monomers. Accordingly, we expected to see ca. 28-kDa peaks in MALDI-MS. NM-SCD-tetramer (56,753.6 Da) contains two non-covalently associated "dimers." Accordingly, we also expected to see ca. 28-kDa peaks in its MALDI-MS. However, the spectrum for the tetramer fraction (NM-T) has peaks near a mass of 56.7 kDa. These peaks consist of two sets of triplets {56,797.89, 57,014.54 (+ 216.65), 57216.54 (+418.65) and 58973.89, 59,212.71 (+238.82), 59,453.01 (+479.12) Da}. Since the two dimers are associated non-covalently we were not expecting to see a 57 kDa peak. However, it is known that non-covalently associated protein subunits may fly together in MALDI-TOF-MS and that may explain the 57 kDa peak. The NM-SCD-T spectrum also has a set of triplet peaks near 28 kDa: 28,321.64; 28,409.19 (+87.55), 28,588.42 (+266.78). Moreover, there is a set of doublet peaks near 14 kDa: 14215.14, 14320.57 (+105.43).

[00335] The two shoulders on the right hand side of the tetramer peak (T) (Figure 4- 8a) were collected and run on the gel column again, providing a peak at 11.3 mL and 12.5 mL as seen in Figure 4-9. The "dimer" fraction (NM-D) at 12.5 mL had no significant MALDI peaks near the expected mass of 28 kDa but well defined peaks near 8 and 10 kDa were observed. Even with ziptip cleanup, no reasonable spectrum was obtained for the 11.3 mL peak obtained from the second run of two-shoulder peak. [00336] MALDI-TOF-MS for C2D and C2T. C2-"dimer" (28,434.7 kDa) contains two covalently attached ca. 14-kDa monomers, so I expected to see a 28 kDa peak in the MALDI-MS. C2-"tetramer" (56,869.4 kDa) consists of two-non-covalently associated "dimers." Accordingly, we also expected to see ca. 28 kDa peaks in the MALDI-MS. A peak near 57 kDa was observed both for C2D (56,892.43 Da) and C2T (56,911.38). Although these two dimers can associate non-covalently, we did not expect to see a 57- kDa peak. However, it is known that non-covalently associated protein subunits may fly together in MALDI-TOF-MS and that may explain the 57 kDa peak. The C2D has two sets of doublet peaks near 28 kDa: 28463.52; 28680.80 (+ 217.28); 28824.09 (+ 360.57); 28920.81 (+ 457.29). C2D also has a set of doublet peaks near 14 kDa: 14,194 Da, and 14,428.78 Da (+ 234.78), that are very close to half the size of the dimer (14,217.35 Da) and probably arise from the +2 ion of the parent peak at ca. 28 kDa. [00337] The C2T spectrum has a peak at 56,911.38 Da and two sets of doublet peaks ^' near 28 kDa: 28485.57; 28699.79 (+ 214.22); 28767.48 (+ 281.91); 28890.15 (+ 404.58) and one peak at 14,227.20 {very close to half the size of the dimer (14,217.35 Da) probably arises from the +2 ion of the parent peak at 28 kDa}. [00338] MALDI-TOF-MS for E2D and E2T. E2-"dimer" (28,499.9 kDa) contains two covalently attached ca. 14 kDa monomers. Accordingly, we expected to see a 28 kDa peak in the MALDI-MS. E2-"tetramer" (56,999.8 kDa) consists of two-non- covalently associated "dimers." Accordingly, we also expected to see ca. 28 kDa peaks in its MALDI-MS. As before in the case of NM-SCD and mutant C2, however, a peak near 57 kDa (57,141.74) was observed both for E2D and E2T. Although these two dimers can associate non-covalently, we did not expect to see a 57-kDa peak. However, as mentioned before, non-covalently associated protein subunits may fly together in MALDI-TOF-MS and that may explain the 57 kDa peak. Successful applications of MALDI-MS for characterization of non-covalent complexes have been reported (Borchers, C, et al., 1999, Biochemistry 38, 1 1734-11740; Bordini, E. & Hamdan, M. 1999, Rapid Communications in Mass Spectrometry 13, 1143-1151; Juhasz, P. & Biemann, K. 1994, Proceedings of the National Academy of Sciences of the United States of America 91, 4333-4337; Kiselar, J. G., et al., 1999, Biochemisty 38, 14185-14191; Rosinke, Bet al., 1995, Journal of Mass Spectrometry 30, 1462-1468; Tang, X., et al., 1995, Analytical Biochemistry 67, 4542-4548). For example, antibody-peptide complexes were shown to be preserved during mass spectrometric analysis under certain experimental conditions (Kiselar, J. G., et al., 1999, Biochemisty 38, 14185-14191). In addition, there is some controversy on determination of the stoichiometry of multiprotein complexes by using MALDI-MS (Sobott, F. & Robinson, C. V., 2002, Current Opinion in Structural Biology 12, 729-734).

[00339] E2T also had a set of doublet peaks: 28,562.98; 28790.25. The difference between these masses is 227.27 Da, which most probably arises from an adduct of sinapic acid (224 Da). The E2T spectrum also has a peak at 14,349.18 Da (very close to half the size of the dimer - 14,249.95 Da) and which probably comes from the +2 ion of the parent peak at 28 kDa. As a result, this spectrum shows the mass of the E2T as 28562.98 which is 63.08 Da (0.22 % error) more than that of theoretical mass. This is acceptable, given that the average mass of an amino acid residue is about 110 Da. [00340] The MALDI spectrum of E2D has a set of doublet peaks near 57 kDa: 57,033.21, and 57,194.42 (+161.21) and has two sets of doublet peaks near 28 kDa: 28547.84; 28771.17 (+223.33); 28620.05 (+72.21); 28959.01 (+411.17). The E2D spectrum also has a peak at 14,306.03 Da (very close to half the size of the dimer - 14,249.95 Da) which probably arises from the +2 ion of the parent peak at 28 kDa. [00341] The three engineered streptavidins displayed similar behavior in the mass spectrometer - consistent with the expected chain size (~ 28 kDa) and a tendency to form rather tight stable dimeric structures similar to wild-type tetrameric streptavidin. [00342] Characterization of biotin-binding behavior of SCD proteins by fluorescence polarization: Fluorescent molecules that are excited with plane-polarized light will emit light back into the fixed plane, meaning that the light remains polarized, but only if the molecules remain stationary during the excited-state lifetime of the fluorophore. However, dissolved molecules rotate and tumble and the orientations with which they emit light can differ greatly from the plane of initial excitation. [00343] The molecular rotational correlation (or relaxation) time τc (the time it takes to rotate through an angle of 68.5°) determines the polarization value of a molecule. Rotational correlation time (τc) is directly proportional to viscosity (η) times hydrated molecular volume (Vh) and inversely proportional to the gas constant (R) times the absolute temperature (T) (Cantor, C. & Schimmel, P. (1969). Techniques for the study of biological structure and function, 2. 3 vols, W.H. Freeman, San Francisco). [00344] τc = (η Vh) / RT

[00345] As we see from the above formula, polarization will correlate with molecular size or volume if viscosity and temperature are kept constant. Molecular volume may change due to binding or dissociation of two molecules, conformational changes, or degradation. Thus fluorescence polarization is ideal to study the binding of small- molecule fluorescent ligands to receptors. Because the size of the ligand is much smaller than that of receptor, binding causes a large increase in polarization. [00346] We used fluorescence polarization to study the interactions between a biotin - fluorescein conjugate and single-chain dimeric streptavidin mutants, so as to determine dissociation constants, and performed competition experiments to determine the binding of unmodified biotin as a competitor. All experiments were carried out in a sample chamber at 25 oC and in assay buffer {50 mM Tris-HCl (fluorescence-grade), pH 8, 150 mM NaCl, 5 mM EDTA, 0.1 mg/mL BSA (Sigma no. A-9647; minimum 96% electrophoresis grade)}. The pH of the buffer was adjusted to 7.5, then it was sterilized by filtration through a 0.22 μm filter (Millipore, Milex syringe-driven filter unit) in a total volume of 150 μL in borosilicate culture tubes. Samples were prepared in microcentrifuge tubes and the order of addition was: buffer, protein, ligand. [00347] Biotin-4-fluorescein (B4F, Mr = 644.7, Figure 11 ), developed as a small fluorescent ligand (absorption/emission = 494/523 nm) to accurately measure avidin or streptavidin concentrations (Gruber, H. J., et al., 1998, Biochimica Et Biophysica Acta- General Subjects 1381, 203-212; Kada, G., et al., 1999, Biochimica Et Biophysica Acta- General Subjects 1427, 33-43; Kada, G., et al., 1999, Biochimica Et Biophysica Acta- General Subjects 1427, 44-48), is a yellow solid, soluble in DMSO or buffers with pH > 7. Its solubility in water is low. (Strept)avidin concentrations as low as 200 pM can be accurately measured by titrating solutions of unknown concentration with this ligand. Assay buffers with 0.1 mg/mL BSA are recommended to prevent non-specific binding, especially at low protein concentrations (such as 200 pM). BSA does not interfere with the assay (Kada, G., et al., 1999, Biochimica Et Biophysica Acta-General Subjects 1427, 44-48). Wild-type streptavidin from Pierce was used during our experiments as a control (Pierce no. 21122, 1 mg/mL; 12 biotin-binding units per mg of protein). [00348] Other fluorescent ligands. As described below, we also tested fluorescein itself (as the sodium salt) and fluoresceinyl-glycinamide for binding to the SCD proteins. (See Figure 17 for their structures).

[00349] Binding curves for the calculation of Kd' of biotin-4-fluorescein. Protein samples with final monomer concentrations as indicated in the figure legends were mixed with 0.063 nM of B4F, so as to have [ligand]/Kd < 0.1 (i.e. assuming a Kd ~ 10^"9 M) in a total volume of 150 μL. All samples were prepared in duplicate and incubated at room temperature for 30 min before polarization measurements at 25°C. For example, figures 13a, 13b, 13c, and 13d show the binding curves for dimeric mutant C2, tetrameric C2, dimeric E2 and tetrameric E2, respectively.

[00350] The binding curve for the E2-tetramer was also determined by using ten times more protein and ligand. Comparison of the curves measured at the two concentrations showed that there was no significant difference between them, meaning that the mP values in this range are concentration-independent. In the first experiment, B4F concentration was 1 nM and the protein concentration range was 1 -20 nM. In the second experiment, B4F concentration was 10 nM and the protein concentration range was 10- 200 nM. [00351] Dissociation constants for B4F (Kd' ) were estimated assuming a simple one- step binding reaction. When the fluorescence intensity is half maximally quenched, the dye is half-bound: [S-B4F]₅₀% = [B4F]_free(5o%) = [B4F]o/2 Thus,

[00352] Kd¹ = ([S]_free x [B4F]_free(50%) / [S.B4F]₅₀% =[S]free = [S]total x [B4F]o/2, where [S-B4F]₅₀% molarity of bound B4F at half-saturation; [B4F]f_ree(₅G%) molarity of free B4F at half-saturation; [B4F]o=molarity of total B4F; [S]_fr_ee molarity of unoccupied biotin- binding sites; and [S]_totai ⁼ total protein monomer concentration. [00353] Since [S]o » [B4F]o, [S]₅₀ ~ [S]o (All protein concentrations are defined in monomeric units.)

[00354] Competition experiments to calculate Kd for unmodified biotin. Dissociation constants or binding affinities of non-fluorescent ligands can be measured indirectly by competition binding experiments. This can be done by either calculating the binding affinities from IC₅₀ curves, or by direct calculation once the K'd of the labeled ligand is known. The beauty of the competition experiments is their ability to determine the dissociation constant of a ligand without the need to label it. Competition experiments performed using fluorescence polarization require relatively high protein concentrations in order to bind significant amounts of the labeled ligand. Therefore, FP competition experiments should be designed such that the protein/Kd ratio is at least 1, so that the starting polarization value will represent at least 50% of the maximal shift. It has been reported that IC50 values determined by such methods can be greater than the actual dissociation constants of the unlabeled ligand (Panvera. (2002). Fluorescence Polarization. Technical Resource Guide, third edition Chapter 2, Receptor-Ligand Binding, 2-2).

[00355] In these competition experiments our purpose was to measure the dissociation constants of unmodified biotin for NM-SCD, C2 and E2 whose dissociation constants for biotin-4-fluorescein had already been determined. Therefore, single-chain dimeric streptavidin mutants (9 nM total monomer concentration) were first incubated with 0.16 nM of biotin-4-fluorescein for 30 minutes at room temperature and then chased with unmodified biotin. Assay volume was 150 μL. Assay buffer was 50 mM Tris-HCl, pH 8, 150 mM NaCl, 5 mM EDTA, 0.1 mg/mL BSA. BSA was included to prevent non¬ specific binding (Kada, G., et al., 1999, Biochimica Et Biophysica Acta-General Subjects 1427, 44-48). [00356] Preliminary experiment with E2-dimer. First, a preliminary experiment was run with E2-dimer to determine the concentration of unmodified biotin to add to displace the labeled biotin. Results showed that the binding of the fluorescent derivative - at least to our mutant - is so much stronger than the binding of biotin itself that 67 μM biotin needs to be added to observe an appreciable exchange between B4F and unmodified biotin.

[00357] Our preliminary experiment was designed such that [protein]/Kd >1, so that the starting polarization value will represent at least 50% of the maximal shift. The protein sample was mixed with B4F as before, and then a 50-fold excess of unmodified biotin (compared to the total monomeric binding sites) was added and the polarization monitored for half an hour. It stayed at the initial value. A few-hundred fold more excess biotin was added, but again no change in polarization value was observed. Finally, unmodified biotin was added so as to increase the final concentration from 450 nM to 67 μM. This time the polarization started to decrease slowly and continued doing so for more than an hour. These results were unusual since in the case of ³H-biotin binding a 50-fold excess of unmodified biotin would normally have been enough to measure exchange with wild-type streptavidin. However, our results for mutant E2 at high unmodified biotin concentration indeed do indicate exchange and the semi-log plot of these data is exactly what theory predicts. This suggests that the binding of the fluorescent derivative - at least to our mutant - is very much stronger than the binding of biotin itself. [00358] Competition experiments with NM-SCD, C2 and E2 to calculate Kd for unmodified biotin. Protein samples ~ 0.7%-saturated with biotin-4-fluorescein were chased with different concentrations of unmodified biotin. In the first experiment, BO, protein samples were mixed with 63 pM of labeled biotin and incubated at 25°C for 30 minutes, then their polarization was monitored. Polarization values started between 300- 350 mP and remained there. This means that the proteins bind to B4F tightly and do not dissociate over time. In the second and third experiments, Bl and B2, protein samples pre-incubated with B4F were chased with 182.μM and 246 μM of biotin, respectively. At both concentrations the proteins displayed some exchange, however, the polarization started to level off at around 240 mP meaning that the majority of the B4F remained bound to the protein. In the fourth experiment, B3, where 492 μM of biotin was added, the exchange reached an equilibrium between 150-250 mP which indicates that even at such a high biotin concentration a substantial amounts of B4F remained bound. In the final experiment, B4, biotin was added to a concentration of 725 μM at which point all of the mutants showed complete exchange except in the case of E2-D where the curve leveled off at around 70 mP.

[00359] Figures 14A-14D show examples competition experiments between NM- SCD-dimer and tetramer fractions pre-incubated with B4F then exposed to unmodified biotin. Although both proteins showed qualitative behavior similar to that just described for mutants C2 and E2 (displacement of B4F by high concentrations of unmodified biotin with first-order kinetics), their quantitative behavior had inconsistencies. Furthermore, MALDI analysis of the NM-SCD-dimer fraction from FPLC did not give a peak near the expected 28 kDa, suggesting that some degradation had occurred. [00360] Reverse-competition experiments: Protein incubated with biotin then mixed with biotin-4-fluorescein. A reverse-competition experiment was performed by adding different amounts of excess unmodified biotin (five-fold, fifty-fold, 611 -fold, and 6077- fold of total monomeric binding sites) to protein samples (9 nM monomer) and then chasing with an amount of biotin-4-fluorescein (0.16 nM) such that it can all bind, but only if it successfully displaces the unmodified biotin. The polarization then was followed for several hours. E2-dimer was tested with all of the biotin concentrations. At zero biotin concentration all B4F was bound. When the protein was first saturated with a five¬ fold excess of unmodified biotin, then chased with 0.16 nM of B4F, mP values were similar to those obtained in the absence of biotin as shown, for example in Table 4-3. This implies that all the B4F molecules managed to displace biotin and bind. Even with a 50-fold excess of biotin almost all the B4F was able to bind. However, a 611-fold excess of biotin caused the mP values to drop down to ca. 200 mP, which means that some of the fluorescently labeled biotin remained unbound, and with a 6077-fold excess of biotin, background mP values were obtained, implying that no B4F was bound (Table 4-3). Experiments done with all the other mutants pre-incubated with a five-fold excess of unmodified biotin (compared to the total monomeric biotin binding sites) showed similar results to those done with the E2-dimer, indicating that these mutants allow enough biotin to exchange so that all of the added B4F (0.16 nM) can bind. To explain the rise of the five-fold added biotin curve over the no-biotin curve I repeated the experiment with the C2-dimer. The results showed that the effect of pre-addition of unmodified biotin is so small that it is probably not significant. This experiment also demonstrated the reversibility and stability of B4F binding behavior in the presence or absence of a five¬ fold molar excess of biotin.

[00361] Conclusions from the competition experiments. Non-mutant single-chain dimer and mutants C2 and E2 had binding affinities for biotin much lower than that of wild-type tetrameric streptavidin. We showed that the initial polarization value of E2- dimer, preincubated with a 6077-fold excess of unmodified biotin, then chased with 0.16 nM of biotin-4-fluorescein, was 25.6. After 35 minutes the mP value increased to 55, meaning that E2-D still showed some binding to B4F even at such a high concentration of unmodified biotin. However, wild-type tetrameric streptavidin mixed with only a five¬ fold excess of unmodified biotin gave background (23 mP) as first reading and remained the same (- 17 mP) for 48 minutes. This means that biotin-4-fluorescein could not replace the biotin bound to wild-type tetrameric streptavidin, demonstrating that the binding ability of wild-type tetrameric streptavidin to bind biotin was very much higher than its ability to bind biotin-4-fluorescein, i.e., exactly the reverse of what we found for the SCDs. As already shown by the preliminary experiment the ability of E2-dimer to bind biotin-4-fluorescein (Kd ~ 10^"10 M) is much greater than its biotin-binding ability. After adding 6077-fold excess biotin we saw that B4F could not bind to the protein and polarization stays at background level. This showed that B4F dis not bind to some other (ectopic) site — in other words, the protein did not have a fluorescein-binding site distinct from the normal biotin-binding site.

[00362] C2-dimer had a slower binding to biotin-4-fluorescein than C2-tetramer: dissociation of B4F or exchange of B4F with unmodified biotin was smaller for C2D than that for C2T. E2-dimer had a faster binding to B4F than E2-tetramer. Dissociation of B4F or exchange of B4F with unmodified biotin was also faster for E2D than that for E2T.

[00363] IC₅o experiment for C2-Dimer and native tetrameric streptavidin. The inhibitory concentration 50% (IC₅₀) value determines the concentration of unlabeled ligand necessary to displace 50% of a labeled ligand (tracer) from its binding protein. IC₅₀ values vary with the concentration of the tracer and the protein and the experimental system, thus if these parameters are not constant between experiments they can not be easily compared. If these conditions can be kept constant, IC₅₀ values can be used to compare the relative affinities of a series of ligands. In a typical competition experiment, protein is incubated with labeled ligand and various concentrations of unlabeled competing ligand. As the concentration of unlabeled ligand increases, it competes with the labeled ligand. In an FP experiment, as the exchange occurs between labeled and unlabeled ligand, the polarization value decreases. We used the IC50 procedure to test the affinities of unlabeled biotin for native tetrameric streptavidin and for the C2-Dimer by its ability to displace biotin-4-fluorescein. C2-Dimer and native tetrameric streptavidin (9 nM monomer in both cases), preincubated with 0.16 nM of biotin-4- fluorescein, were mixed with 12 and 9 different unmodified biotin concentrations, respectively. The mixtures were incubated for 2-3 hours at room temperature then fluorescence polarization was measured. Points were from duplicate experiments, and each corresponded to the average of six measurements. [00364] Remeasurement of the polarization values for the same sets of samples after further incubation at 25°C for two days or overnight showed little change, showing that the initial values correspond to equilibrium binding.

[00365] As already noted, biotin competes only weakly with B4F for binding to the C2-Dimer protein. The IC₅₀ curve confirms this observation - background polarization values were only obtained at the highest concentration of unmodified biotin. However, although the data show some scatter, it strongly suggest a biphasic displacement of B4F by biotin, corresponding to one Kd in the micromolar range at about 3 μM, involving slightly less than half of the bound ligand, and a second Kd at 560 μM. Since all the SCD proteins have an inherent asymmetry due to one of the two Tip- 120 residues (in the C- terminal domain) remains unmutated, it would not be surprising to have two classes of binding site with different affinities for biotin. However, this consideration is further complicated by the dimer-tetramer equilibrium.

[00366] The IC₅₀ curve for native tetrameric streptavidin confirmed our previous conclusion as well as direct biotin/B4F competition experiments that ligands (either biotin or B4F) essentially bind irreversibly to the native protein.

[00367] Test of non-specific binding of fluorescein-Na+ salt and fluoresceinyl- glycinamide to NM-SCD and mutants C2 and E2. To understand why the association constants for biotin-4-fluorescein are at least 104 times stronger than for biotin, we tested the binding ability of the SCD and its mutants to fluorescein sodium salt and fluoresceinyl-glycinamide. First, the fluorescein compounds (0.063 nM) were incubated with protein samples (prepared in duplicate) at different concentrations (0.06-4 nM) for 30 minutes at 25°C in the absence of biotin, then their fluorescence polarization was monitored. The results were at the background level, meaning that SCD-NM, C2D, C2T and E2D, E2T did not bind to these fluorescein compounds at the concentrations used for FP experiments. Native tetrameric streptavidin, used as a control, also did not show any non-specific binding to fluorescein or fluoresceinyl-glycinamide. [00368] To study if pre-incubation with excess unmodified biotin will affect the results due to possible conformational changes, protein samples (9 nM monomers) were first saturated with excess unmodified biotin (725 μM), allowed to stand at 25⁰C for 5-10 minutes, and then incubated with 63 pM of fluorescein sodium salt and fluoresceinyl- glycinamide. The results were again at the background level, confirming that there are no sites on the biotin-bound protein molecules that can tightly bind these fluorescein derivatives.

[00369] In conclusion, there are neither contiguous (near the biotin-binding pocket) nor ectopic sites that these fluorescein compounds can bind to non-specifically with high affinity. Hence some other factor(s) must be responsible for the apparent tight binding of biotin-4-fluorescein to the SCD and its mutants, given that the competition experiments indicate a comparatively weak binding of biotin itself to these proteins. [00370] Thermal stability of NM-SCD and its mutants C2 and E2 at 70⁰C. NM- SCD and C2 and E2 were heated at 70°C for 3 minutes then cooled down at 25°C. Biotin-binding ability of these proteins was tested with 3H-biotin by ultrafree MC centrifugal filter devices. Additional protein samples, prepared the same way as the ones heated at 70⁰C, were incubated at room temperature and used as controls. All samples were prepared in duplicate.

[00371 ] Comparison of the counts of protein samples with those of control samples (Table 4-4) showed that after heating at 70°C all three proteins lost activity. Free biotin counts in the bottom chamber of the protein samples heated at 70⁰C were much higher than for those incubated at room temperature. Thus heating caused the proteins to lose their binding ability, and after cooling down they did not recover it. Unheated, control proteins were more active. As seen in Table 4-5 the counts from the negative control and those of the heated samples are very close to each other. Natural core streptavidin when heated to 70⁰C at a rate of 2°C/min and kept there for 10 minutes and then cooled down to 25°C again at the same rate loses 45% of its biotin-binding ability (Reznik, G. O., et al., 1996). Nature Biotechnology 14, 1007-11).

[00372] Stability of SCD and its mutants at 4⁰C. The biotin-binding abilities of the SCD and its mutants were re-measured by fluorescence polarization after storage at 4°C for different periods of time to find out whether they were still active or had undergone some degradation. Different concentrations of protein (0.18 - 2 nM) were mixed with 0.16 nM of biotin-4-fluorescein and incubated at 25°C for 30 minutes and their polarization values were measured. These experiments were also repeated by using ten times more ligand (0.63 nM) and protein (1.8 and 2.5 nM) and similar results were obtained. The results (Table 4-5) showed that the SCD proteins remained stable for 1-3 months though some of the proteins might have suffered some slight degradation because their mP readings were a little lower than when the binding experiment with fresh protein samples. Nevertheless, since fluorescence polarization measurements are quite sensitive to experimental conditions, the difference between the readings might be trivial. In addition, some mutants actually showed higher mP values, possibly due to increased tetramerization during storage which might have increased their stability. The lowest concentration of the protein samples routinely used for storage was 500 nM (dimer). Lower concentrations were prepared on the same day. All samples were prepared in folding buffer (50 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA at pH 7.5).

[00373] The mutants stored at 4⁰C for various periods of time were also tested for their ability to bind fluorescein-Na+ salt and fluoresceinyl-glycinamide as well as biotin-4- fluorescein. None of the proteins displayed any binding activity with these compounds. [00374] Quenching effect of protein on fluorescence emission from biotin-4- fluorescein, fluorescein-Na+ salt and fluoresceinyl-glycinamide. Normally one does not need to measure the fluorescence intensity when measuring ligand binding by FP, though we took measurements periodically to check instrument performance. However, as we increased the protein concentration of C2T, in experiments with biotin-4- fluorescein, we checked the overall intensity and observed that quenching occured. Total intensity of the 0.16 nM B4F with no protein was 425, and upon addition of 0.18 nM (monomeric unit) C2T the total intensity dropped to 260. When the C2T concentration was increased further to 0.25 nM (monomeric unit) total intensity became 208. We then systematically checked all the SCD mutants for quenching and obtained the results summarized in Table 4-6. Total intensity of the controls is from the average of four experiments, and that of the samples with protein is from the average of duplicate experiments. No significant quenching was observed with fluorescein-Na+ salt and fluoresceinyl-glycinamide in the experiments with these mutants confirming the absence of non-specific binding of these fluorophores to the SCD proteins and suggesting that the quenching effects we observed came from close contact between the fluorescein moiety of B4F and the protein.

[00375] Quenching was observed with tetrameric streptavidin as well. Upon addition of 1 iiM (monomer) tetrameric streptavidin total intensity was ca. 390. After the protein concentration was increased to 3 nM, 6 nM and 10 nM (all in monomeric units) the total intensities became 300, 150 and 35, respectively. The existence of quenching in my FP samples should not affect the interpretation of B4F binding results reported here, but it will demand close scrutiny in future studies of these and related proteins using fluorescence techniques. Example 3

[00376] Oligonucleotide Directed Mutagenesis of Streptavidin: Site directed mutagenesis without disturbing local environments around this residue can be accomplished by known means. For example, the codon encoding Tip- 120 was mutated to a codon encoding Asp.

[00377] pTSA-13, which carries the coding sequence for amino acids 16-133 of mature streptavidin was used as the starting material to make the modified streptavidin (FIG. 1). Basically, a phosphoxylated oligonucleotide of the desired sequence is used to mutate the codon TGG encoding for Trp on residue 120 to a condon encoding Asp, e.g., GAC. The reaction is initiated by hybridizing 10 pmoles of the phosphorylated oligonucleotide to the single stranded streptavidin DNA in a 10 μl reaction with 20 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 50 mM NaCl and 1 mM dithiothreitol (DTT). Elongation and mutation is initiated by the addition of 10 μl of 20 mM Tris-HCl, pH 7.5,10 mM MgCl₂, 10 mM DTT, 2 mM dATP, 2 mM dTTP, 2 mM dCTP, 2 mM dGTP, 10 mM ATP, 5 units bacteriophage T4 DNA ligase and 2.5 units of Klenow. This procedure is performed according to the in vitro mutagenesis kit supplied by Amersham. Subsequent procedures followed as recommended by Amersham. Resulting products created are used to transform competent E. coli cells. To select clones contained the desired mutations, the sequence was confirmed using a dideoxy chain termination procedure. The same technique is used to alter the other sites. Example 4

[00378] The mutated streptavidin of Example 1 was used to produce large quantities modified streptavidin protein. Because the expression of streptavidin in bacteria has a lethal effect to a cell, an inducible system was used. The DNA fragment comprising the sequence encoding the streptavidin mutant was excised from its vector with the restriction endonucleases Nde I and BamH I, and cloned into the same sites in the T7 expression vector pET-3a. Resultant plasmids were transformed in BL21(DE3) (pLysE) bacteria. [00379] To produce the modified streptavidin, BL21(DE3)(pLysE) cells carrying the expression plasmid was grown at 37°C in LB supplemented with 0.4% glucose, 150 μg/ml ampicillin and 25 μg/ml chloramphenicol until cultures reached a density of 0.6 at A. sub.600. Expression of the Phe-120 streptavidin was induced by the addition of a gratuitous inducer, IPTG, to a final concentration of 0.4 mM. Modified streptavidin was expressed for five hours at 37⁰C before the cells were harvested. Example 5

[00380] Purification of Expressed Streptavidin. Modified streptavidin protein produced by induced E. coli was purified. Cells expressing the mutant streptavidin were harvested by centrifugation at 1600.times.g for 10 minutes. Protein was purified from the insoluble fraction of cell extracts. Briefly, cells were pelleted, washed with an isotonic solution of 100 mM NaCl, 1 mM EDTA and 10 mM Tris, pH 8.0, and resuspended in a detergent solution of 2 mM EDTA, 30 mM Tris-HCl, pH 8.0, 0.1% Triton X-100. Lysis occurred under these conditions because the presence of T7 lysozyme in the cells. [00381] Nucleic acid in the extract was digested for 15 minutes at room temperature by the addition OfMgSO₄, DNase I and RNase A, to final concentrations of 12 mM, 10 μg/ml and 10 μg/ml, respectively. The insoluble fraction of the extract containing Phe- 120 streptavidin was isolated by centrifugation of the nuclease treated extract at 39,000 x g for 15 minutes. Pellets were washed with 2 mM EDTA, 30 mM Tris-HCl, pH 8.0, and 0.1% Triton X-100, and solubilized in 6 M guanidine hydrochloride, pH 1.5. [00382] Impurities were removed by dialysis against 6 M guanidine hydrochloride pH 1.5. Mutant streptavidin were renatured slowly by dialysis against 0.2 M ammonium acetate, pH 6. After renaturation, insoluble impurities were removed by centrifugation at 39,000xg. Supernatant containing the mutant streptavidin protein were removed and collected. Final purifications were performed by 2-iminobiotin affinity chromatography. [00383] Polyacrylamide gel electrophoresis analysis (PAGE-SDS) of the modified streptavidin was performed on protein expressed in E. coli carrying pTSA-38. Total cell protein of BL21(DE3)(pLysE), with or without pTSA-38, is analyzed using a 15% polyacrylamide gel.

[00384] Modified streptavidin can also be analyzed by SDS-PAGE. Briefly, approximately 3 μg of modified streptavidin was applied to a 15% polyacrylamide gel. The right lane contains molecular mass standard proteins. The molecular weight of the protein estimated to be approximately 13,000 daltons, which was consistent with the molecular mass obtained from the deduced amino acid sequence (12,600 daltons). Example 6

[00385] Determining the Biotin-Binding Affinity of the Modified Streptavidin. The biotin-binding affinities of wild type and modified streptavidin were determined by an equilibrium dialysis method using a micro dialyzer (Hoeffer Scientific). One hundred microliters each of D-[carbonyl-¹⁴C] biotin (2 nM-4 μM; 53 mCi/mmol; Amersham) and lOOμl of streptavidin (5.3 μg/ml, 0.42 μM subunits) were prepared separately in TBS (150 mM NaCl, 20 mM Tris-HCl, pH 7.4,0.02% NaN₃) solutions. Equilibrium dialysis analysis was begun by the placement of the two solutions into two opposing chambers of a micro dialyzer. Chambers were incubated at 3O⁰C with rotation for 48 hours and the concentration of labeled biotin in each chamber is measured by scintillation counting. Results were plotted on a Scatchard plot.

[00386] Comparison with the biotin-binding affinity of natural core streptavidin (4 x 10¹⁴ M^"1 at pH 7.0 at 25⁰C) indicated that the biotin-binding affinity is caused by the modifications. Example 7

[00387] Determination of Solubility Characteristics of Expressed Proteins. To determine the effect of the modifications on the solubility characteristics, the solubility of each core streptavidin molecule with and without biotin was investigated by varying the concentration of ammonium sulfate or ethanol in the solution.

[00388] Analysis of solubility in the absence of biotin was performed by adjusting the concentration of each core streptavidin to 5.7 nanomoles of subunits per mililiter in TBSl [Tris-buffered saline: 150 mM NaCl, 20 mM Tris-HCl, pH 7.4,0.02% NaN₃. This corresponds to 72 μ.g/ml for Stv-13 and 76 μg/ml for Stv-25 and natural core streptavidin. To 100 μ.l of this protein solution, 1.1 ml of an appropriate ammonium sulfate solution in TBS was added to adjust the final concentration of ammonium sulfate (final streptavidin concentration, 0.48 nanomoles of subunits per milliter). The mixture was allowed to stand at 30⁰C for 30 minutes and centrifuged at 2,200 x g for 20 minutes. The amount of soluble streptavidin in the supernatant fraction was determined by biotin- binding assays described below. The fraction of original streptavidin remaining in the supernatant was defined as the relative solubility.

[00389] Analysis of solubility in the presence of biotin was performed in a similar method to analysis in the absence of biotin. Biotin-binding sites of each streptavidin molecule are saturated by adding an equimolar amount of D-[carbonyl-¹⁴ Cjbiotin, prior to the addition of an ammonium sulfate solution. The amount of soluble streptavidin in the final supernatant was estimated from the radioactivity derived from bound biotin, determined by liquid scintillation counting.

[00390] Analysis of solubility using ethanol was similar to analysis made using ammonium sulfate with two modifications. After the addition of ethanol, the final volume was adjusted to 1.2 ml by the addition of an appropriate ethanol solution to make the final protein concentration constant for all samples. After incubation at 30°C for 30 minutes, centrifugation was performed at 13,000 x g for 20 minutes. Example 8

[00391] The following three sets of candidate changes were considered and tested by site-directed mutagenesis of streptavidin to create stable dimers with retention of high biotin binding affinity.

[00392] Because the intersubunit contact made by tryptophan- 120 to biotin bound by an adjacent subunit contributes significantly to biotin binding in tetrameric streptavidin, the biotin-binding affinity of dimeric streptavidin is still likely to be reduced by the subunit separation. (Note that tryptophan- 120 is the most hydrophobic side chain exposed to solvent in a dimeric molecule.)

[00393] The modified streptavidins prevent the dimer from dissociating into subunits since these are likely to have too many exposed hydrophobic residues to be stable as a folded soluble protein.

[00394] The three mutants were designed by molecular modeling using effective binding free energy calculations. In vitro site-directed mutagenesis was used to construct three different genes for two-chain dimeric streptavidin. Expression vectors encoding putative two-chain dimeric streptavidin were isolated and used to transform E. colt lysogens, BL21(DE3)(pLysS) and BL21(DE3)(ρLysE) by known techniques. These strains carry the T7 RNA polymerase gene under the lacUV5 promoter in the chromosome, and these are used for high-level expression of genes cloned into expression vectors containing bacteriophage T7 promoter. Each dimeric streptavidin mutant was expressed efficiently in E. coli using the T7 RNA polymerase/T7 promoter expression systems. The expressed streptavidins were insoluble in the cell and yielded inclusion bodies. These were solubilized in 7M guanidine hydrochloride (pH=1.5) and subsequent dialysis against neutral non-denaturing buffer gave soluble and renatured protein. Gel filtration chromatography was carried out to determine the size of these proteins. The results confirm that they are dimeric proteins with a molecular weight of 25kDa. Biotin- binding ability was determined by gel filtration using a Sephadex G-25 column (PD-IO) and ³H-Biotin (28.0 Ci/mmol) at room temperature. This dissociation constant of dimeric streptavidin is about 10^"7M.

[00395] Other embodiments and uses of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. All documents including U.S. patents and patent applications disclosed herein are specifically incorporated by reference in their entirety. The specification and examples should be considered exemplary only with the true scope and spirit of the invention indicated by the following claims.

Claims

What is claimed:

1. A single-chain dimeric streptavidin molecule comprising two streptavidin monomer subunits connected by a series of circular permutations of the two monomer streptavidin subunits leaving a single amino terminal and carboxyl terminal end.

2. The single-chain dimeric streptavidin molecule of claim 1 , wherein the series of circular permutations of the two streptavidin monomer subunits involves circularizing each streptavidin monomer subunit at their amino terminal and carboxyl terminal end and inserting a linker of amino acids to form a cross-connection between the two streptavidin monomer subunits at a normal interface between the subunits when they bind to form a dimer, thereby creating a new amino terminal end on one of the streptavidin monomer subunits, and a new carboxyl terminal end on the other monomer streptavidin subunit.

3. The single-chain dimeric streptavidin molecule of claim 2, wherein the linker is selected to form a beta turn between the two streptavidin monomer subunits.

4. The single-chain dimeric streptavidin molecule of claim 2 or, 3, wherein the cross-over linker is at amino acid residue 115 or 116 of one streptavidin monomer subunit, and 69 or 68 of the other streptavidin monomer subunit.

5. The single-chain dimeric streptavidin molecule of claim 2, 3, or 4, wherein the cross-over linker comprises a number of glycine amino acid residues.

6. The single-chain dimeric streptavidin molecule of claim 2, 3, or 4, wherein the cross-over linker comprises amino acids of SEQ Id NO: 14 or SEQ ID NO: 15.

7. The single-chain dimeric streptavidin molecule of claim 2, 3, or 4, wherein the linker between the amino terminal end and the carboxyl terminal end is inserted at positions corresponding to positions 13 and 139 of SEQ ID NO: 1.

8. The single-chain dimeric streptavidin molecule of claim 2, 3, 4, 5, 6, or 7, wherein the streptavidin monomer subunit is selected from the group of streptavidin peptides consisting of wild type streptavidin monomeric subunits and modified streptavidin monomeric subunits.

9. The single-chain dimeric streptavidin molecule of claim 8, wherein at least one of the wild-type or modified streptavidin monomeric subunits has at least one mutation selected from the group consisting of K24E, A28T, F50L, G89S, K148E, G181S, W120K, S26T, G32E, F50L, R171K, G247D, S251T, S261N, and combinations thereof.

10. The single-chain dimeric streptavidin molecule of claim 9, comprising mutations K24E, A28T, F50L, G89S, K148E, and G181S.

11. The single-chain dimeric streptavidin molecule of claim 9, comprising mutations S26T, G32E, F50L, R171K, G247D, S251T, and S261N.

12. An isolated nucleic acid comprising a sequence encoding the single-chain dimeric streptavidin of claims 1-11.

13. A cell comprising the isolated nucleic acid of claim 12.

14. A display library comprising at least one single-chain dimeric streptavidin of claims 1-11.

15. A solid support having attached thereon at least one single-chain streptavidin protein of claims 1-11.

16. A method of delivering to an individual a single-chain dimeric streptavidin of any of the preceding claims, in a pharmaceutically acceptable carrier.

17. The method of claim 16, wherein the single-chain dimeric streptavidin is attached to a therapeutic or diagnostic molecule.

18. A method for contacting at least a first target with at least a first biotin derivative, and at least a second target with at least a second biotin derivative, under conditions that a heterogeneous mixture is created, said heterogeneous mixture comprising at least a first biotinylated target and at least a second biotinylated target; contacting the heterogenous mixture with a solid support of claim 15, wherein the streptavidin molecule is a modified streptavidin with an affinity to the first biotin derivative wherein the affinity differs from the affinity of the second biotin derivative, to the second modified streptavidin, and the solid support comprises at least second streptavidin molecule wherein the streptavidin molecule is a second modified streptavidin with an affinity to the second biotin derivative wherein the affinity differs from that of the first modified streptavidin to the first biotin derivative.

19. The method of claim 18, wherein one of the modified streptavidin molecules binds to biotin-4-fluorescein with a greater affinity than to wild type biotin.