JP3267293B2 - Methods for enriching mutant proteins with altered binding - Google Patents

Description

Description: FIELD OF THE INVENTION The present invention relates to the production and systematic selection of novel binding proteins with altered binding to target molecules. More specifically, the present invention relates to a method for producing an exogenous polypeptide that mimics the binding activity of a natural binding partner. In a preferred embodiment, the present invention relates to the manufacture of therapeutic or diagnostic compounds that mimic proteins or non-peptidyl molecules, such as hormones, drugs and other small molecules, especially biologically active molecules such as growth hormone. About.

BACKGROUND OF THE INVENTION Binding partners are substances that specifically bind to each other, usually through non-covalent interactions. Examples of binding partners include ligand-receptor, antibody-antigen, drug-target, and enzyme-substrate interactions. Binding partners are extremely useful in both therapeutic and diagnostic fields.

In the past, binding partners have been identified by various methods, including collection from nature (eg, antibody-antigen, and ligand-receptor pairing), and by accidental identification (eg, using random screening of candidate molecules). Drug development). In some cases, these two methods are combined. For example, variants of proteins or polypeptides (eg, polypeptide fragments) have been prepared that include key functional residues involved in binding. These polypeptide fragments are then derivatized by methods similar to traditional drug development. Examples of such derivatization would include methods such as cyclization to conformationally constrain the polypeptide fragment to obtain new candidate binding partners.

The problem with conventional methods is that natural ligands may not have properties suitable for all therapeutic applications. Furthermore, polypeptide ligands may not be available for some target substances. Also,
Methods for producing non-natural synthetic binding partners are often expensive and difficult, and usually require complex synthetic methods to obtain each candidate. The inability to characterize the resulting candidate structure so that rational drug design methods can be applied to further optimize the candidate molecule further hinders these methods. I have.

Geys in an attempt to overcome these problems
en [Immun. Today 6: 364-369 (1985)] and Geysen et al. [Mol. The use of peptide synthesis was proposed. According to Geysen et al. (Ibid.), Short polypeptides, such as dipeptides, are first screened for the ability to bind to a target molecule. The most active dipeptide is then selected for the next test. The test consists of attaching another residue to the starting dipeptide (or by internally modifying a component of the first starting dipeptide), and then screening this set of candidates for the desired activity. This process is repeated until a binding partner with the desired properties is identified.

The method of Geysen et al. Is hampered by the inconvenience that the chemistry underlying the method, peptide synthesis, produces molecules with unclear or varied secondary and tertiary structures. As the number of selection iterations increases, random interactions accelerate between the various substituents of the polypeptide, and a truly random population of interacting molecules with reproducible conformation is not obtained. For example, interactions between amino acid side chains (which are widely separated in sequence but adjacent in space) occur arbitrarily. In addition, sequences that do not promote a conformationally stable secondary structure may provide complex peptide-side chain interactions, which may interfere with side chain interactions of certain amino acids with the target molecule. Such complex interactions are facilitated by the flexibility of the polyamide backbone of the polypeptide candidate. Also, candidates can exist in a number of conformations, making it difficult to identify the conformer that interacts with or binds to the target with maximum affinity or specificity, and provides rational drug design. Make it difficult.

A final problem when using Geysen's iterative polypeptide method is that there is currently no practical method that can produce, screen and analyze highly diverse alternative peptides. With 20 natural amino acids,
The total number of all combinations of hexapeptides that must be synthesized is 64,000,000. Even if such a highly diversified peptide is produced, a mixture of such highly diversified peptides can be rapidly screened to select a peptide having a high affinity for the target molecule. There is no way. at present,
Each "attached" peptide must be recovered in large enough to perform protein sequencing.

To overcome many of the problems inherent in Geysen's method,
Biological selection and screening were chosen as alternatives. Biological selection and screening are powerful tools for exploring protein function and isolating mutant proteins having desired properties [Shortle, Protei
n Engineering, Oxender and Fox, ARLiss, Inc., NY,
pp. 103-108 (1988); and Bowie et al., Science 247: 130.
6-1310 (1990)]. However, the resulting selection or screening is only applicable to one or a few related proteins.

More recently, Smith and his collaborators [Smith, Scien
ce 228: 1315-1317 (1985); and Parmley and Smit
h, Gene 73: 305-318 (1985)], inserts a short gene fragment into gene III of fd phage ("fusion phage") to convert a small protein fragment (10-50 amino acids) into a linear phage. It has been shown that the surface can be efficiently "displayed". The minor coat protein of gene III (approximately 5 copies at one end of the virion) is important for proper phage assembly and infection by adherence to E. coli fimbria [Rasched et al., Micr
obiol. Rev. 50: 401-427 (1986)]. Recently, "fusion phages" have been developed for exogenous proteins [Devlin et al., Science 249: 404.
-406 (1990)] or an antibody [Scott et al., Science 249: 386.
-390 (1990); and Cwirla et al., Proc. Natl. Acad. USA 8
7: 6378-6382 (1990)].

However, there are some important limitations when using such "fusion phage" to identify modified peptides or proteins with new or enhanced binding. First, less than 100, preferably less than 50, large inserts likely disrupt the function of gene III and thus disrupt phage assembly and infectivity.
Fusion phages have been shown to be useful only when displaying proteins with fewer amino acid residues [Parm
ley et al., Gene 73: 305-318 (1988)]. Second, conventional methods fail to select peptides from a library that have the greatest binding affinity for the target molecule. For example, after thoroughly screening a random peptide library with an anti-β endorphin monoclonal antibody,
Cwirla and co-workers have developed high-affinity peptides (Kd-0.4 μM) to medium-affinity peptides (Kd
〜10 μM) could not be separated. In addition, the parent β- with very high affinity (Kd-7 nM)
Endorphin peptide sequences were not selected from the epitope library.

Ladner [WO 90/02802] discloses a method for selecting novel binding proteins displayed on the outer surface of virus particles and cells (a heterologous protein has up to 164 amino acid residues. Is allowed).
This method contemplates isolating and amplifying the displayed protein to design a new group of binding proteins with the desired affinity for the target molecule. More specifically, Ladner displays a protein with an "initial protein binding domain" ranging from 46 residues (clambin) to 164 residues (T4 lysozyme) fused to the M13 gene III coat protein. "Fusion phage" are disclosed. Ladner states that it is relatively easy to arrange restriction sites in relatively small amino acid sequences,
It teaches the use of these proteins "less than necessary" and favors the use of a 58 amino acid residue bovine pancreatic trypsin inhibitor (BPTI). Small fusion proteins, such as BPTI, are preferred when the target is a protein or macromolecule, while relatively large fusion proteins, such as T4 lysozyme, are preferred for small target molecules, such as steroids, where such large proteins are small. This is because it has crevices and grooves into which molecules can be fitted. This preferred protein BPTI is described in Smith et al. Or de la Cruz et al. [J. Biol. Chem. 263: 4318-4322 (19
88)] or at one of its termini, together with a second synthetic copy of gene III, so that "partial" unmodified gene III protein is present. Proposed. Ladner does not address the problems of successfully selecting high affinity peptides from random peptide libraries, which hinders the prior art biological selection and screening methods.

Human growth hormone (hGH) is greatly involved in regulating normal human growth and development. This 22,000 dalton pituitary hormone, in particular, has linear growth (body formation),
Shows numerous biological effects, including lactation, macrophage activation, insulin-like and diabetic effects [Chawla,
RK, Ann. Rev, Med. 34: 519 (1983); Edwards, CK et al., Sci.
ence 239: 769 (1988); Thomer, MO et al., J. Clin. Invest. 8
1: 745 (1988)]. Growth hormone deficiency in children leads to dwarfism, which has been successfully treated by external administration of hGH for over a decade. hGH is a member of a group of homologous hormones including placental lactogen, prolactin, and other genetic and species variants or growth hormones [Nicoll, CS et al., Endocrine Reviews 7: 169.
(1986)]. hGH is unique among these, exhibits broad species specificity, and has cloned somatic receptors [Leung, DW
, Nature 330: 537 (1987)] or the prolactin receptor [Boutin, JM et al., Ce 53:69 (1988)]. The cloned hGH gene has been secretedly expressed in E. coli [Chang, CN et al., Gene 55: 189 (198
7)], and its DNA and amino acid sequences have been reported [Goeddel et al., Nature 281: 544 (1979); Gray et al., Gene 39:
247 (1985)]. The three-dimensional structure of hGH is not available. However, the three-dimensional folding pattern of porcine growth hormone (pGH) has been reported with normal resolution and precision [Abdel-Meguid, SS et al., Proc. Natl. Acad. Sci. USA 8
4: 6434 (1987)]. Human growth hormone receptor and antibody epitopes have been identified by homolog-scanning mutagenesis [Cunningham et al., Science 243: 1.
330 (1989)]. The structure of a novel amino-terminal methionyl bovine growth hormone containing a spliced sequence of human growth hormone including histidine 18 and histidine 21 has been shown [US Patent No. 4,880,910].

Human growth hormone (hGH) causes a variety of physiological and metabolic effects in various animal models, including linear skeletal growth, lactation, macrophage activation, insulin-like and diabetic effects [RKChawla et al., Annu. Rev. Me
d.34: 519 (1983); OGPIsaksson et al., Annu. Rev. Physio.
l. 47: 483 (1985); CKEdwards et al., Science 239: 769 (19
88); MOThorner and MLVance, J. Clin. Invest. 82: 7
45 (1988); JPHughes and HGFriesen, Ann. Rev. Phy
siol. 47: 469 (1985)]. These biological effects are related to hGH
And specific cell receptors.

Accordingly, it is an object of the present invention to provide a rapid and efficient method for the systematic production of candidate binding agents.

It is another object of the present invention to produce conformationally stable candidate binding agents displayed on the surface of phagemid particles.

Another object of the present invention is to produce a candidate binding substance consisting of a fusion protein of a phage coat protein and a heterologous polypeptide [wherein the polypeptide is 100 amino acids or more in length and 1 subunit And may be displayed on a phagemid particle (the polypeptide is encoded in the phagemid genome)].

Yet another object of the present invention is to provide or display all peptidyl moieties that may be involved in non-covalent binding interactions and to provide these moieties in a sterically restricted manner. It provides a method for producing and selecting binding materials with sufficient flexibility.

Another object of the present invention is the production of mutant growth hormones that exhibit a stronger affinity for growth hormone receptors and binding proteins.

It is another object of the present invention to provide a method for functionally linking a heterologous polypeptide and a phage coat protein such that a detectable fusion protein is produced in a suppressor host cell and only the heterologous polypeptide is produced in a non-suppressor host cell. The objective is to produce an expression vector phagemid containing a placed repressible stop codon.

Finally, it is an object of the present invention to provide a phagemid that rarely displays more than one copy of a candidate binding protein on the outer surface of the phagemid particle, thus achieving efficient selection of high affinity binding proteins. To produce particles.

These and other objects of the invention will be apparent from a consideration of the invention as a whole.

SUMMARY OF THE INVENTION The above object is to provide: (a) a replicable expression containing a first gene encoding a polypeptide and a second gene encoding at least a portion of a native or wild-type phage coat protein. A vector (where the first and second genes are heterologous, and the transcriptional regulatory element is operably linked to the first and second genes, so that a gene fusion encoding the fusion protein is obtained. (B) mutating at one or more selected positions in the first gene of the vector to generate a family of related plasmids; (c) generating appropriate host cells with the plasmids (D) infecting the transformed host cell with a helper phage having a gene encoding the phage coat protein; Under conditions suitable for the formation of recombinant phagemid particles containing at least a portion of the phagemid and capable of transforming the host, only a small amount of phagemid particles will display more than one copy of the fusion protein on the surface of the particles. Adjusting said conditions to culture said transformed and infected host cells; (f) contacting said phagemid particles with a target molecule to bind at least a portion of said phagemid particles to said target molecule; and (G) separating bound phagemid particles from unbound particles. A method for selecting a novel binding polypeptide comprising: Preferably, the method further comprises the step of transforming a suitable host cell with the recombinant phagemid particles that bind to the target molecule and repeating steps (d)-(g) one or more times.

Furthermore, a method for selecting a novel binding protein consisting of more than one subunit is achieved by selecting a novel binding peptide as follows: the selection method comprises:-one or more subunits A replicable expression vector containing a transcriptional regulatory element operably linked to DNA encoding a desired protein, wherein the DNA encoding at least one of the subunits is a phage coat protein Fused to the DNA encoding at least a portion of the DNA sequence; and mutating the DNA encoding the desired protein at one or more selected positions to produce a family of related vectors. Transforming a suitable host cell with the vector; encoding the transformed host cell with the phage coat protein Infecting a helper phage having the carrying gene; only a small amount of the phagemid particles, under conditions suitable for the formation of recombinant phagemid particles containing at least a portion of the plasmid and capable of transforming the host, Culturing the transformed and infected host cells by adjusting the conditions to display more than one copy of the phagemid particle on the surface of the particle; At least partially binding to the target molecule;
And separating the bound phagemid particles from the unbound particles.

In the method of the present invention, the plasmid is cultured under strict control of transcriptional regulatory elements such that the amount or number of phagemid particles displaying more than one copy of the fusion protein on the surface of the particles is less than about 1%. Preferably, the conditions are adjusted. Also, the amount of phagemid particles displaying more than one copy of the fusion protein is 10% of the amount of phagemid particles displaying a single copy of the fusion protein.
Preferably it is less than. Most preferably, this amount
Less than 20%.

Usually, in the method of the present invention, the expression vector is a DN encoding each subunit of the polypeptide.
Further containing a secretory signal sequence fused to A, the transcriptional regulatory element will be a promoter system. Preferred promoter systems are selected from the LacZ, λ _PL , TAC, T7 polymerase, tryptophan, and alkaline phosphatase promoters and combinations thereof.

Also, the first gene will usually encode a mammalian protein. Preferred proteins will be selected from: human growth hormone (hGH),
N-methionyl human growth hormone, bovine growth hormone, parathyroid hormone, thyroxine, insulin A chain,
Insulin B chain, proinsulin, relaxin A chain, relaxin B chain, prorelaxin, glycoprotein hormones such as follicle stimulating hormone (FSH), thyroid stimulating hormone (TSH) and leutinizing hormone (LH), glycoprotein hormone Receptor, calcitonin, glucagon, factor VIII, antibody, lung surfactant,
Urokinase, streptokinase, human tissue-type plasminogen activator (t-PA), bombesin, factor I
X, thrombin, hematopoietic growth hormone, tumor necrosis factors α and β, enkephalinase, human serum albumin, Mueller inhibitor, mouse gonadotropin-related peptide,
Microbial proteins such as β-lactamase, tissue factor protein, inhibin, activin, vascular endothelial growth factor, hormone or growth factor receptors; integrins;
Thrombopoietin, protein A or D, rheumatoid factor, nerve growth factor such as NGF-β, platelet growth factor, transforming growth factor (TGF) such as TGF
-Α and TGF-β, insulin-like growth factors I and I
I, insulin-like growth factor binding protein, CD-4, DN
Ase, latency-related peptide, erythropoietin, osteoinductive factors, interferons such as interferon α, β and γ, colony stimulating factor (CSF) such as M
-CSF, GM-CSF and G-CSF, interleukin (I
L), for example, IL-1, IL-2, IL-3, IL-4, superoxide dismutase; decay-accelerating factors, viral antigens, HIV envelope proteins, for example, GP120, GP14
0, Atrial natriuretic peptide A, B or C, immunoglobulin, and fragments of any of the above proteins.

Preferably, the first gene encodes a polypeptide of one or more subunits comprising about 100 or more amino acid residues, and comprises a plurality of rigid secondary proteins displaying a plurality of amino acids capable of interacting with a target. It will be folded to form a structure. Preferably, the first gene will be mutated at codons corresponding only to amino acids that can interact with the target, such that the integrity of the rigid secondary structure is preserved.

Usually, the method of the present invention comprises M13KO7, M13R408, M13-VCS
And a helper phage selected from PhiX174. A preferred helper phage is M13KO7, and a preferred coat protein is gene I of M13 phage.
II coat protein. A preferred host is E. coli, which is a protease-deficient strain of E. coli. Novel hGH variants selected by the method of the present invention were detected. A phagemid expression vector was constructed containing a suppressible stop codon functionally located between this polypeptide and the nucleic acid encoding the phage coat protein.

BRIEF DESCRIPTION OF THE FIGURES FIG. 1: Shows a method for displaying (exposing) large proteins on the surface of filamentous phage and a method for enriching for modified receptor binding. A plasmid phGH-M13gIII was constructed in which the entire coding sequence of hGH was fused to the carboxy-terminal domain of M13 gene III. Transcription of this fusion protein is under the control of the lac promoter / operator sequence and secretion is directed by the stII signal sequence. Phagemid particles are "helper" phages
Particles obtained by infection with M13KO7 and displaying pGH can be enriched by binding to an affinity matrix containing the hGH receptor. Wild-type gene III
(Derived from M13KO7 phage)
The fusion protein (derived from the phagemid phGH-M13g III) is illustrated by the multiple arrows of the copy, and by the hGH fold figure (replace the head of the arrow).

FIG. 2: Immunoblot of all phage particles shows that hGH co-migrate with phage. Phagemid particles purified with a cesium chloride gradient were applied to double wells and electrophoresed on a 1% agarose gel in 375 mM Tris, 40 mM glycine (pH 9.6) buffer. The gel was transferred to a transfer buffer [25] containing 2% SDS and 2% β-mercaptoethanol.
mM Tris (pH 8.3), 200 mM glycine, 20% methanol]
Immersion for 2 hours, then rinsed in transfer buffer for 6 hours. Next, the proteins in the gel were transferred to Immobilon membrane (Mill
(ipore). Membranes containing one set of samples were stained with Coomassie blue to indicate the location of phage proteins (A). Double membrane with polyclonal rabbit anti-
The membrane was immunostained for hGH by reacting with the hGH antibody and then with a horseradish peroxidase-conjugated goat anti-rabbit IgG antibody (B). Lane 1 contains the M13KO7 parent phage, which contains hG
Lack of H could only be observed in Coomassie blue stained membranes. Lanes 2 and 3 contained another preparation of hormone phagemid particles, which could be observed on both Coomassie and hGH immunostaining. Differences in migration distance between parental M13KO7 phage and hormone phagemid particles reflect differences in the size of the internally packaged genomes (
8.7kb and 5.1kb).

FIG. 3 is a diagram summarizing each step in a method for selecting an hGH-phage library in which codons 172, 174, 176 and 178 are randomized. Template molecule pH041 containing hGH (R178G, I179T) gene and unique Kpn I restriction site
5 was mutagenized as described herein and E. coli strain W
The first phagemid library (library 1) was obtained by electrotransfer into JM101. A portion (about 2%) of Library 1 was used directly in the first round of selection as described herein to obtain Library 1G. On the other hand, double-stranded
DNA (dsDNA) is prepared from Library 1 and the restriction enzyme Kp
nI digestion to eliminate template background, WJM10
Introduced into 1 to obtain Library 2. Subsequent rounds of selection (or Kpn I digestion; shaded boxes) and subsequent phagemid expansion were performed as indicated by the arrows, according to the methods described herein. Were sequenced by dideoxy sequencing the four independent clones from four independent clones and libraries 5G ⁶ from the library 4G ^4. All of these clones are hGH mutants (Glu 174 Ser, Phe 176 Tyr)
Had the same DNA sequence.

Figure 4: 2.8% of porcine growth hormone determined by crystallography
It shows the structural model of hGH derived from the fold diagram. hGH that strongly modulates binding to hGH-binding proteins
The positions of the middle residues are indicated in the shaded circles. Alanine substitutions that cause a 10-fold or greater decrease in binding affinity (●), a 4- to 10-fold decrease (•), an increase (○), or a 2- to 4-fold decrease (•) are shown. Spiral ring projections in the α-helical region revealed these polarities. Black, shaded, or unshaded residues indicate charge, polarity, or non-polarity, respectively. In helix 4, the most important residues for mutation are on the hydrophilic side.

Figure 5: Rounds 1 and 3 for the indicated route (hGH elution, glycine elution or glycine elution after preadsorption)
The amino acid substitutions at positions 172, 174, 176 and 178 of hGH found after sequencing a number of clones from the selection step of 178R). Non-functional sequences (ie, vector background, or other immature and / or frameshifted mutants) are indicated as "NF." Functional sequences containing non-silent pseudomutations (ie, other than the set of target residues listed above) are indicated by a "+". Appeared more than once but different among all sequencing clones
A protein sequence having a DNA sequence is indicated by “#”. Protein sequences which appear more than once in the sequencing clones and have the same DNA sequence are indicated by a "*". It should be noted that two different contaminating sequences were found after three rounds of selection (these clones did not match the cassette mutant, but did match the previously constructed hormone phage). The pS0643 contaminant is consistent with wild-type hGH-phage (hGH "KEFR"). Third
High pH for pool of round glycine-selected phages
The 0457 contaminant was the previously identified hGH mutant `` KSYR ''
Matches. The amplification of these contaminants underscores the ability of the hormone-phage selection method to select for rare, non-occurring mutants. Also, the convergence of these sequences was significant in all three pathways (ie, R or K occurred most frequently at positions 172 and 178;
Or F occurs most frequently at position 176, and S, T, A
And other residues occur at position 174).

FIG. 6: Sequence from phage selected on hPRLbp-beads in the presence of zinc. The display is the same as that described in FIG. Here the sequence convergence is unpredictable, but there appears to be a bias towards hydrophobic sequences under the most stringent (glycine) selection conditions (L, W and P residues are found more in this pool) .

FIG. 7: Sequence from phage selected on hPRLbp-beads in the absence of zinc. The display is the same as that described in FIG. In contrast to the sequence of FIG. 6, the sequence here appears to be relatively hydrophilic. After four rounds of selection by hGH elution, two clones (ANHQ and TLDT / 171V) are abundant in the pool.

FIG. 8: Sequence from phage selected on blank beads. The display is the same as that described in FIG. After three rounds of selection by glycine elution, no siblings were observed and background levels of non-functional sequences were retained.

FIG. 9: Construction of phagemid fl origin (ori) from pHO415. Cassette mutagenesis and hGH-
This vector for expression of the gene III fusion protein was constructed as follows. hGH-gene I under the control of the E. coli phoA promoter, containing the pBR322 and fl origins of replication.
II fusion protein (residues 1-191 of hGH followed by one Gly residue, which is fused to Pro-198 of gene III)
Plasmid pS0643 was constructed by oligonucleotide-directed mutagenesis of pS0132, which expresses The following oligonucleotides: Was used for mutagenesis. This oligonucleotide introduced an Xba I site (underlined) and an amber stop codon (TAG) following Phe-191 of hGH.

Figure 10: A depicts a plasmid pDH188 insert containing DNA encoding the light and heavy chains (variable and constant domain 1) of a Fab humanized antibody directed to the HER-2 receptor. is there. _VL and _VH are the variable regions of the light and heavy chains, respectively. C _k is the constant region of the human kappa light chain.
CH1 _G1 is the first constant region of the human γ1 chain. Both coding regions begin with the bacterial st II signal sequence. B
Illustrates the entire plasmid pBH188 containing the insert shown in 5A. After introducing this plasmid into E. coli SR101 cells and adding helper phage, the plasmid is packaged into phage particles. Some of these particles display Fab-pIII fusions (where pIII
Is a protein encoded by M13 gene III DNA). The segment in this plasmid diagram corresponds to the insert shown in 5A.

FIG. 11 (A to C in FIG. 11 are summarized): DNA encoding 4D5 Fab molecule expressed on the surface of phagemid
1 (SEQ ID NO: 25).
The amino acid sequence of the light chain (SEQ ID NO: 26) is shown as well as the amino acid sequence of the heavy chain pIII fusion (SEQ ID NO: 27).

FIG. 12: Enrichment of wild-type 4D5 Fab phagemid from mutant Fab phagemid. A mixture of wild type phagemid and mutant 4D5 Fab phagemid in a ratio of 1: 1,000 was selected on plates coated with the HER-2 receptor extracellular domain protein. After each round of selection, a portion of the eluted phagemid was infected into E. coli and plasmid DNA was prepared. Next, this plasmid DNA was transferred to EcoR V
And digested with PstI, separated on a 5% polyacrylamide gel and stained with ethidium bromide. Bands were visualized under UV light. Bands resulting from wild-type and mutant plasmids are indicated by arrows. The first round of selection eluted only under acid conditions. Subsequent rounds were eluted either by acid elution (left side of the figure) or by a humanized 4D5 antibody washing step using the method described in Example VIII followed by acid elution (right side of the figure). Three mutant 4D5 Fab molecules were prepared. That is, H91A (amino acid histidine at position 91 of the _VL chain was mutated to alanine; shown in the "A" lane in the figure), Y49A (amino acid tyrosine at position 49 of the _VL chain was mutated to alanine. ; Shown in the "B" lane in the figure), and Y92A (the amino acid tyrosine at position 92 of the _VL chain was mutated to alanine; shown in the "C" lane in the figure).
Amino acid position counting was performed according to Kabat et al. ["Sequences of Proteins in Immunological Inserts", 4th ed., USDept of Hea
lth and Human Services, Public Health Service, Nat '
l. Institute of Health, Bethesda, MD (1987)].

FIG. 13: Scatchard analysis of RIA affinity measurements described during the experimental protocol is shown. Bound labeled EC
The amount of D antigen is shown on the x-axis, and the value obtained by dividing the amount bound by the amount released is shown on the y-axis. The slope of the line indicates Ka,
The calculated Kd is 1 / Ka.

DETAILED DESCRIPTION OF THE INVENTION The following discussion may be best understood by referring to FIG. In its simplest form, the method of the present invention provides a novel binding polypeptide (such as a protein ligand) having a desired, usually high affinity for a target molecule from a library of structurally related binding polypeptides. Consist of ways to choose. Libraries of structurally related polypeptides fused to phage coat proteins are prepared by mutagenesis and preferably contain a single copy of each related polypeptide containing the DNA encoding the polypeptide. Displayed on the surface of phagemid particles. These phagemid particles are then contacted with a target molecule and the particles with the highest affinity for the target are separated from the low affinity particles. The high affinity binder is then amplified by infection of the bacterial host and the competitive binding step is repeated. This process is repeated until a polypeptide with the desired affinity is obtained.

The novel binding polypeptides or ligands produced by the methods of the present invention are themselves useful as diagnostic or therapeutic agents (eg, agonists or antagonists) used in the treatment of biological organisms. Also, rational drug design can be advanced using the structural analysis of the selected polypeptide.

As used herein, "binding polypeptide" refers to any polypeptide that binds to a target molecule with a selective affinity. Preferably, the polypeptide is a protein, most preferably a protein containing about 100 or more amino acid residues. Usually, the polypeptide will be a hormone or antibody or a fragment thereof.

“High affinity” as used herein means an affinity constant (Kd) of <10 ⁻⁵ M, preferably <10 ⁻⁷ M, under physiological conditions.

As used herein, a "target molecule" is any molecule for which it is desired to produce a ligand, and is not necessarily a protein. However, the target is preferably a protein, most preferably
This target is a receptor, such as a hormone receptor.

As used herein, “humanized antibody” refers to an antibody in which the complementarity determining regions (CDRs) of a mouse or other non-human antibody are bound to a human antibody framework. The framework of a human antibody refers to all human antibodies except CDRs.

I. Selection of Polypeptide for Display on Phage Surface The first step in the method of the present invention is to select a polypeptide having a rigid secondary structure exposed on the surface of the polypeptide for display on the phage surface. Is to choose.

As used herein, "polypeptide" refers to any molecule that can be expressed by a specific DNA sequence. A polypeptide of the present invention consists of more than one subunit, each subunit being a separate DNA
Coded by array.

"Rigid secondary structure" as used herein, for example, alpha-helix, 3 ₁₀ helix, [pi helices,
By parallel and anti-parallel β-sheets, as well as any polypeptide segment that exhibits a regular repeating structure found in the reverse turns and the like. Also, certain "disordered" structures that lack recognizable geometric order form domains or "patches" of amino acid residues that allow them to interact with the target, and the overall shape of the structure is Unless destroyed by amino acid substitutions, they are included in the definition of a rigid secondary structure. Certain unordered structures are thought to be combinations of reverse turns. The geometry of these rigid secondary structures is
It is well defined by the φ and ψ torsion angles around the α-carbon of the peptide “backbone”.

All that is required to expose the secondary structure to the polypeptide surface is a domain or “patch” of amino acid residues that can be exposed to and bind to the target molecule.
It is to provide. This is basically an amino acid residue that is replaced by mutagenesis, which results in a "library" of structurally related (mutant) binding polypeptides displayed on the surface of the phage. From which a novel polypeptide ligand is selected. Generally, mutagenesis or substitution of amino acid residues towards the interior of the polypeptide is avoided, thereby preserving the overall structure of the rigid secondary structure. Some substitutions of amino acids in the internal regions of the rigid secondary structure, especially with hydrophobic amino acid residues, may be tolerated since these conservative substitutions do not appear to distort the overall structure of the polypeptide. .

The higher affinity binding is selected by phagemid selection of a plurality of amino acid changes selected by the multiple selection cycles using the reversion cycle of the "polypeptide" selection. Following the first round of phagemid selection (which involves the selection of amino acids or first regions in the ligand polypeptide), additional rounds of phagemid selection are performed on other regions or amino acids of the ligand polypeptide. This phagemid selection cycle is repeated until the desired affinity of the ligand polypeptide is achieved. To illustrate this method, phagemid selection of hGH in Example VIII was performed in cycles. In the first cycle,
Mutations were made to amino acids 172, 174, 176 and 178 of hGH and phagemids were selected. In the second cycle, amino acids 167, 171, 175 and 179 of hGH were phagemid selected. In the third cycle, amino acid 1 of hGH
0, 14, 18 and 21 were phagemid selected. Optimal amino acid changes from the previous cycle can be introduced into the polypeptide before selection for the next cycle. For example, hG
H amino acid substitutions 174 (serine) and 176 (tyrosine)
Was introduced into hGH prior to phagemid selection of amino acids 167, 171, 175 and 179 of hGH.

From the foregoing, it will be appreciated that the amino acid residues forming the binding domain of the polypeptide may not be contiguously linked, but may be on different subunits of the polypeptide. That is, the binding domain follows a specific secondary structure at the binding site, not the primary structure. Therefore, mutations should be introduced into codons encoding amino acids within a particular secondary structure at sites from the interior to the exterior of the polypeptide so that they have the potential to interact with the target. Normal. For illustration, the positions of residues in hGH that are known to strongly modulate binding to hGH-binding proteins are shown in Figure 2 [Cunningham et al., Science 247: 1461-1465].
(1990)]. That is, representative sites suitable for mutagenesis include residues 172, 174, 176, and 178 of the helix, and residue 64 in the "disordered" secondary structure.

It is not necessary that a polypeptide selected as a ligand for a target binds normally to the target. That is, for example
Glycoprotein hormones such as TSH can be selected as ligands for the FSH receptor, and libraries of mutant STH molecules are used in the methods of the invention to obtain novel drug candidates.

That is, the present invention contemplates any polypeptide that binds to the target molecule and encompasses antibodies. Preferred polypeptides are those that have pharmaceutical use. More preferred polypeptides include: growth hormone (human growth hormone, des-).
Parathyroid hormone; thyroid stimulating hormone; thyroxine; insulin A chain; insulin B chain; proinsulin; follicle stimulating hormone; calcitonin; leutinizing hormone; glucagon; VIII; antibodies; lung surfactant; plasminogen activator [eg urokinase or human tissue-type plasminogen activator (t-PA)]; bombesin; factor IX; thrombin; hematopoietic growth factor; serum albumin (eg, human serum albumin); Muller inhibitor; relaxin A chain; relaxin B chain; prorelaxin; mouse gonadotropin-related peptide; microbial protein (eg, β-lactamase); ;I Hibin; activin;
Vascular endothelial growth factor; hormone or growth factor receptor;
Integrin; thrombopoietin; protein A or
D; rheumatoid factor; nerve growth factor (eg, NGF-
β); platelet-derived growth factor; fibroblast growth factor (eg, aFGF and bFGF); epidermal growth factor; transforming growth factor (TGF) (eg, TGF-α and TGF).
Insulin-like growth factors I and II; insulin-like growth factor binding protein; CD-4; DNase; latency-related peptide; erythropoietin; osteoinductive factor; interferons (eg, interferon α, β and γ); Stimulator (CSF) (eg, M-CSF, GM
Interleukin (IL) (eg, IL-1, IL-2, IL-3, IL-4); superoxide dismutase; decay-accelerating factor; atrial natriuretic peptide A, B or C; viral antigens (eg, HIV
Immunoglobulin; and fragments of any of the above polypeptides. In addition, one or more predetermined amino acid residues on the polypeptide may be substituted, inserted or deleted to obtain a product having, for example, improved biological properties. Also included are fragments of these polypeptides, particularly biologically active fragments. Further preferred polypeptides of the invention are human growth hormone,
Atrial natriuretic peptides A, B and C, endotoxin, subtilisin, trypsin, and other serine proteases.

A more preferred polypeptide hormone is any amino acid sequence produced in a first cell, specifically for a receptor on the same cell type (autocrine hormone) or a second cell type (non-autocrine). A polypeptide hormone that can be defined as an amino acid sequence that binds and causes a characteristic physiological response in receptor-bearing cells. Such polypeptide hormones include cytokines, lymphokines, neurotrophic hormones and pituitary gland polypeptide hormones such as growth hormone, prolactin, placental lactogen, luteinizing hormone, follicle stimulating hormone,
Thyrotropin, villous gonadotropin, corticotropin, α or β-melanocyte stimulating hormone, β-lipotropin, γ-lipotropin and endorphins;
Hypothalamic release inhibitory hormones such as corticotropin releasing factor, growth hormone releasing inhibitory hormone, growth hormone releasing factor; and atrial natriuretic peptide A;
Other polypeptide hormones such as B and C are included.

II. Obtaining the First Gene Encoding the Desired Polypeptide (Gene 1) The gene encoding the desired polypeptide (ie, a polypeptide having a rigid secondary structure) can be obtained by methods known in the art. [Generally, Sambrook
Et al., Molecular Biology: A laboratory Manual, Cold Spri
ng Harbor Press, Cold Spring Harbor, New York (198
9)]. If the sequence of this gene is known, the DNA encoding this gene may be synthesized chemically [Merrfield, J. Am. Chem. Soc. 85: 2149 (196
3)]. If the sequence of this gene is unknown, or if this gene has not been isolated before, cDNA
From a library (prepared from RNA from a suitable tissue in which the desired gene is expressed) or from a suitable genomic DNA
Can be cloned from a library. The gene is then isolated using a suitable probe. Suitable probes for a cDNA library include monoclonal or polyclonal antibodies (when a cDNA library is an expression library), oligonucleotides, and complementary or homologous cDNAs or fragments thereof. Probes that can be used to isolate the desired gene from a genomic DNA library include cDNAs encoding the same or similar genes.
Or its fragment, homologous genomic DNA or D
NA fragments, and oligonucleotides. Screening of a cDNA or genomic library with the selected probe is described in Sambrook et al.
Perform using the standard method described in ~ 12.

Another method for isolating the gene encoding the desired protein is described in Section 14 of Sambrook et al. (Supra).
Using the polymerase chain reactivity (PCR) described in US Pat. This method requires the use of oligonucleotides that hybridize to the desired gene.
That is, to obtain an oligonucleotide,
At least part of the DNA sequence must be known.

After isolating the gene, it can be inserted into a suitable vector for amplification, preferably a plasmid, as generally described by Sambrook et al. (Supra).

III. Construction of Replicable Expression Vectors Although several types of vectors are available and can be used in the practice of the present invention, plasmid vectors are the preferred vectors for use in the present invention. This is because these vectors can be constructed relatively easily and can be easily amplified. In general, plasmid vectors contain various components, including promoters, signal sequences, phenotypic selection genes, origins of replication, and other necessary components, as known to those skilled in the art.

The most commonly used promoters in prokaryotic vectors, lacZ promoter system, the alkaline phosphatase phoA promoter, the bacteriophage .lambda.P _L promoter (a temperature sensitive promoter), the hybrid trp regulated by tac promoter (lac repressor
-Lac promoter), tryptophan promoter,
And the bacteriophage T7 promoter.
See Section 17 of Sambrook et al. (Supra) for a general description of promoters. Although there are the most commonly used promoters, other suitable microbial promoters can be used as well.

Preferred promoters for the practice of the present invention are those that can be tightly regulated to control expression of the fusion gene. It is believed that a problem that remained unrecognized in the prior art was that the display of multiple copies of the fusion protein on the surface of the phagemid particles led to multiple point binding of the phagemid to the target. This effect, called "chelation", is to select the wrong "high affinity" polypeptide when multiple copies of the fusion protein are displayed on phagemid particles in close proximity to each other and the target is "chelated" It is thought to give the result. When multiple point binding occurs, the effective or apparent Kd will be as high as the product of the individual Kd for each copy of the indicated fusion protein. This effect may be the reason why Cwirla and coworkers (supra) were unable to separate medium-affinity peptides from relatively high-affinity peptides.

By tightly regulating the expression of the fusion protein such that only small amounts (ie, less than about 1%) of the phagemid particles contain multiple copies of the fusion protein,
It has been discovered that "chelation" has been overcome, allowing for the appropriate selection of high affinity polypeptides. That is, depending on the promoter, the culture conditions of the host are adjusted to maximize the number of phagemid particles containing a single copy of the fusion protein and minimize the number of phagemid particles containing multiple copies of the fusion protein.

Preferred promoters used in the practice of the present invention are
The lacZ promoter and the phoA promoter. this
The lacZ promoter is regulated by the lac repressor protein laci, and thus transcription of the fusion gene can be controlled by manipulating the amount of the lac repressor protein. By way of illustration, the lacZ promoter-containing phagemid is grown in a cell line containing a copy of the lac repressor gene, a repressor of the lacZ promoter. Examples of cell lines containing the laci gene include JM101 and XL1-Blue. Alternatively, host cells can be co-transfected with a plasmid containing both the repressor laci and lacZ promoters. Sometimes both of the above methods are used simultaneously. That is, phagemid particles containing the lacZ promoter are grown in a cell line containing the laci gene, and the cell line is co-transfected with a plasmid containing both the lacZ and laci genes. Usually, when it is desired to express the gene in the transfected host, an inducer such as isopropylthiogalactoside (IPTG) is added. However, in the present invention, this step is omitted and (a) expression of the gene III fusion protein is minimized to minimize copy number (ie, the number of gene III fusions relative to the number of phagemids);
(B) Prevent poor or improper packaging of phagemid caused by inducers such as IPTG even at low concentrations. Typically, when no inducer is added, the number of fusion proteins per phagemid particle is about 0.1 (number of bulk fusion proteins / number of phagemid particles). The most preferred promoter used in the practice of the present invention is phoA. This promoter is thought to be regulated by the amount of inorganic phosphate in the cell, which phosphate down regulates the activity of the promoter.
That is, the activity of the promoter can be increased by reducing the phosphate of the cell. The desired result is 2Y
Grow cells in a phosphate-rich medium such as T or LB,
This is achieved by controlling the expression of the gene III fusion.

Another useful component of the vectors used in the practice of the invention is a signal sequence. Usually, this sequence is located immediately 5 ′ to the gene encoding the fusion protein,
Thus, it is transcribed at the amino terminus of the fusion protein. However, in some cases, this signal sequence may be added to the other 5 'of the gene encoding the protein to be secreted.
It is shown that it exists at the position of. This sequence targets the protein to which it binds across the inner membrane of the bacterial cell. Encodes this signal sequence
DNA can be obtained as a restriction endonuclease fragment from any gene encoding a protein having a signal sequence. Suitable prokaryotic signal sequences include, for example, LamB or OmpF [Wong et al., Gene 68: 193.
(1983)], and can be obtained from genes encoding MalE, PhoA and other genes. Preferred prokaryotic signal sequences for the practice of the present invention are described in Chang et al. [Gene 55: 189.
(1987)], which is a thermostable enterotoxin II (ST II) signal sequence of Escherichia coli.

Another useful component of the vectors used in practicing the present invention is a phenotypic selection gene. A common phenotypic selection gene is a gene that encodes a protein that confers antibiotic resistance on a host cell. For explanation,
The ampicillin resistance gene (amp) and the tetracycline resistance gene (tet) are easy to use for this purpose.

Construction of suitable vectors containing the above components as well as the gene encoding the desired polypeptide (gene 1) is accomplished using standard recombinant DNA techniques described by Sambrook et al. (Supra). The isolated DNA fragments that are ligated to form the vector are cut, processed, and ligated together in a particular order and orientation to create the desired vector.

Cleavage the DNA with an appropriate restriction enzyme or enzymes in an appropriate buffer. Typically, about 0.2-1 μg of plasmid or DNA fragment is used with about 1-2 units of the appropriate restriction enzyme in about 20 μl of buffer solution. Appropriate buffers, DNA concentrations, and incubation times and temperatures are specified by the restriction enzyme manufacturer. Usually, an incubation time of about 1 or 2 hours at 37 ° C. is appropriate, but some enzymes require higher temperatures. After incubation, enzymes and other impurities are removed by extraction of the digestion solution with a mixture of phenol and chloroform, and DNA is recovered from the aqueous fraction by precipitation with ethanol.

To ligate the DNA fragments together to obtain a functional vector, the ends of these DNA fragments must be compatible with each other. In some cases, these ends are compatible after endonuclease digestion. However, in order to make them compatible for ligation, it may be necessary to first convert sticky ends, usually produced by endonuclease digestion, to blunt ends. To blunt the ends, the DNA is digested at least 15 ° C. with 15 units of Klenow fragment of DNA polymerase I (Klenow) in an appropriate buffer in the presence of four deoxynucleotide triphosphates. Process for a minute. Next, the DNA is purified by phenol-chloroform extraction and ethanol precipitation.

Using DNA gel electrophoresis, the cleaved DNA fragments can be size separated and selected. By either agarose or polyacrylamide matrix
DNA can be electrophoresed. The choice of matrix will depend on the size of the DNA fragments to be separated. Following electrophoresis, the DNA is matrixed by electroelution, as described in sections 6.30-6.33 of Sambrook et al. (Supra), or by melting agarose and extracting DNA therefrom when low melting agarose is used as the matrix. Extract from

DNA fragments to be ligated together (pre-digested with appropriate restriction enzymes so that the ends of each fragment to be ligated are compatible) are placed in solution in approximately equal amounts. This solution contains ATP, ligase buffer and ligase, for example, about 10 units of T4 DNA per 0.5 μg of DNA.
It will also contain ligase. When ligating a DNA fragment into a vector, the vector is first linearized by cutting it with the appropriate restriction endonuclease (s). The linearized vector is then treated with alkaline phosphatase or bovine intestinal phosphatase. This phosphatase treatment prevents self-ligation of the vector during the ligation step.

After ligation, the vector into which the foreign gene has been inserted is introduced into a suitable host cell. Prokaryotes are preferred host cells for the present invention. Suitable prokaryotic host cells include E. coli strains.
JM101, E. coli K12 strain 294 (ATCC No. 31,446), E. coli strain W
3110 (ACTT No.27,325), E. coli X1776 (ACTT No.31,53)
7), E. coli XL-1Blue (stratagene), and E. coli B
But many other strains of E. coli (eg, HB10
1, NM522, NM538, NM539), and many other prokaryotic genera and species can be used as well. In addition to the E. coli strains listed above, Bacillus (eg, Bacillus Sub.
tilis), other intestinal bacteria (eg, Salmonella typhimur
ium or Serratia marcesans), and various Pseud
All omonas species can also be used as hosts.

Transformation of prokaryotes is readily accomplished using the calcium chloride method described in Section 1.82 of Sambrook et al. (Supra). According to an alternative method, electroporation [Neumann et al., EMBO
J. 1: 841 (1982)]. Transformed cells are usually selected by growing on an antibiotic such as tetracycline (tet) or ampicillin (amp) (the transformed cells are isolated from these antibiotics by the presence of the tet and / or amp resistance genes on the vector). Resistant to substances).

After selection of the transformed cells, the cells are grown in culture and the plasmid DNA (or other vector into which the foreign gene has been inserted) is then isolated. Plasmid DNA can be isolated using methods known in the art. Two suitable methods are described in Section 1.25 of Sambrook et al. (Supra).
Small-scale and large-scale DNA preparations described in ~ 1.33. The isolated DNA can be purified by methods known in the art, such as those described in Section 1.40 of Sambrook et al. (Supra).
The purified plasmid DNA is then analyzed by restriction mapping and / or DNA sequencing. Usually, DNA sequencing is performed by Messing et al. [Nucleic Acids Res.
9: 309 (1981)] or Maxam et al. [Meth. Enzymol. 6
5: 499 (1980)].

IV. Gene Fusion The present invention contemplates fusing a gene encoding a desired polypeptide (gene 1) to a second gene (gene 2) such that a fusion protein is produced during transcription. ing. Usually, gene 2 is the phage coat protein gene, preferably the gene III coat protein of phage M13 or a fragment thereof.
Fusion of genes 1 and 2 can be achieved by inserting gene 2 at a particular site on a plasmid containing gene 1, or by inserting gene 1 at a particular site on a plasmid containing gene 2. it can.

Insertion of a gene into a plasmid requires that the plasmid be cut at the exact location where the gene is to be inserted. That is, a restriction endonuclease site must be present at this position (preferably the only site so that the plasmid is cleaved at only one site during restriction endonuclease digestion). The plasmid is digested, phosphatase treated and purified as described above. Then, by ligating the two DNAs together,
The gene is inserted into this linearized plasmid. Ligation can be achieved when the ends of the plasmid match the ends of the gene to be inserted. When cutting the plasmid with restriction enzymes and isolating the gene to be inserted (creating blunt ends or compatible cohesive ends), bacteriophage T4, as described in Section 1.68 of Sambrook et al. Using a ligase such as DNA ligase, incubating the mixture at 16 ° C. for 1-4 hours in the presence of ATP and ligase buffer,
DNA can be directly ligated together. If the ends are not compatible, they are first converted to Klenow fragment of DNA polymerase I or bacteriophage.
It must be blunted by the use of T4 DNA polymerase, which requires four deoxyribonucleotide triphosphates to fill the protruding single-stranded ends of the digested DNA. According to an alternative method, these ends can be replaced by nuclease S1 or mung bean nuclease (these are DNA
Which function by cutting back overhanging single strands of DNA). The DNA is then ligated using the ligase described above. In some cases, the reading frame of the coding region changes, so that the end of the gene to be inserted may not be smoothed.
Oligonucleotide linkers may be used to overcome this problem. This linker serves as a bridge connecting the plasmid and the gene to be inserted. These linkers
It can be prepared as double-stranded or single-stranded DNA by synthesis using a conventional method. These linkers have one end that matches the end of the gene to be inserted and are first linked to this gene using the ligation method described above. The other end of the linker is designed to be compatible with the plasmid for ligation. When designing the linker, care must be taken not to destroy the reading frame of the gene to be inserted or of the gene contained on the plasmid.
In some cases, it may be necessary to design the linker so that they code for some of the amino acids, or for one or more amino acids.

A DNA encoding a stop codon can be inserted between gene 1 and gene 2. Such stop codons are UAG (Amber), UAA (Ocher) and UGA (Opel) [Davis et al., Microbiology, Harper & Row, New
York, pp. 237, 245-47 and 274 (1980)]. The stop codon expressed in a wild-type host cell results in the synthesis of the gene 1 protein product without the gene 2 protein attached. However, growth in suppressor host cells results in the synthesis of a detectable amount of the fusion protein. Such suppressor host cells have been modified to insert amino acids at the stop codon position of the mRNA.
It contains a tRNA, thus giving the result of producing a detectable amount of the fusion protein. Such suppressor host cells are, for example, suppressor strains of E. coli [Bullock
Et al., Bio Techniques 5: 376-379 (1987)]. Such stop codons can be placed in the mRNA encoding the fusion polypeptide using any accepted method.

The repressible codon can be inserted between a first gene encoding a polypeptide and a second gene encoding at least a portion of a phage coat protein. Alternatively, this repressible stop codon can be inserted adjacent to the fusion site by substitution of the last amino acid triplet in the polypeptide or the first amino acid in the phage coat protein.
Propagation of the phagemid containing this repressible codon in a suppressor host cell results in the detectable production of a fusion polypeptide containing the polypeptide and coat protein. Growth of this phagemid in a non-suppressor host cell results in the termination of the polypeptide at the inserted repressible triplet encoding UAG, UAA, or UGA, resulting in a polypeptide that is not fused to the phage coat protein. Synthesized. In non-suppressor cells, the polypeptide is synthesized and secreted from the host cell due to the absence of the fused phage coat protein, which otherwise anchors the polypeptide to the host cell.

V. Modification (mutation) of Gene 1 at Selected Position Gene 1 encoding the desired polypeptide is
Alternatively, it can be modified at more selected codons. An alteration is defined as a substitution, deletion or insertion of one or more codons in the gene encoding the polypeptide. This modification results in a change in the amino acid sequence of the polypeptide as compared to the unmodified or native sequence of the same polypeptide. Preferably, the alteration is by replacing at least one amino acid with any other amino acid in one or more regions of the molecule. This modification can be made by various methods known in the art. These methods include, but are not limited to, oligonucleotide-mediated mutagenesis and cassette mutagenesis.

A. Oligonucleotide-mediated mutagenesis Oligonucleotide-mediated mutagenesis
This is a preferred method for preparing substitution, deletion and insertion mutants. This method is described in Zoller et al. [Nucleic Acids Re
s.10: 6487-6504 (1987)]. Briefly, Gene 1 is modified by hybridizing an oligonucleotide encoding the desired mutation to a single-stranded DNA template of a plasmid containing Gene 1 of the unmodified or native DNA sequence. . After hybridization, the DNA polymerase is used to synthesize the complete second complement of this template, into which the oligonucleotide primers have been introduced, encoding the selected modification in gene 1.

Usually, oligonucleotides of at least 25 nucleotides in length are used. The optimal oligonucleotide will have 12-15 nucleotides that are perfectly complementary to the template on each side of the nucleotide (s) encoding the mutation. As a result, the oligonucleotide becomes a single-stranded DNA
It ensures that it hybridizes properly to the template molecule. The oligonucleotide can be prepared by methods known in the art,
For example, Crea et al. [Proc. Natl. Acad. Sci. USA 75: 5765 (197
8)] can be easily synthesized using the method described in [1].

This DNA template can be derived from a bacteriophage M13 vector (commercially available M13mp18 and M13mp19 vectors are suitable), or Veira et al. [Met
h. Enzymol. 153: 3 (1987)] can be obtained only by a vector containing a single-stranded phage origin of replication. That is, the DNA to be mutated is
In order to create a single-stranded template, it must be inserted into any of these vectors. Preparation of single-stranded template is Sa
described in sections 4.21 to 4.41 of mbrook et al. (supra).

To modify the native DNA sequence, the oligonucleotide is hybridized to a single-stranded template under appropriate hybridization conditions. Next, a DNA polymerase, usually Klenow fragment of DNA polymerase I, is added, and the complementary strand of the template is synthesized using this oligonucleotide as a primer for synthesis. Thus, a heteroduplex molecule in which one strand of DNA encodes a mutated form of Gene 1 and the other strand (the first template) encodes the native, unmodified sequence of Gene 1. Is formed. The heteroduplex molecule is then introduced into a suitable host cell, usually a prokaryote such as E. coli JM101. After growing the cells, they are plated on agarose plates and screened using oligonucleotide primers radiolabeled with 32-phosphate to identify bacterial colonies containing the mutated DNA.

The method described immediately above can be modified such that a heteroduplex molecule is created in which both strands of the plasmid contain the mutation (s). The modifications are as follows:
The single-stranded oligonucleotide is annealed to the single-stranded template as described above. Two deoxyribonucleotides,
That is, deoxyriboadenosine (dATP), deoxyriboguanosine (dGTP) and deoxyribothymidine (dTTP)
To a modified thio-, termed dCTP- (aS).
Mix with deoxyribocytosine (available from Amersham). This mixture is added to the template-oligonucleotide complex. Addition of a DNA polymerase to this mixture creates a DNA strand identical to the template except for the mutated base. In addition, this new DNA strand
Contains dCTP- (aS) instead of dCTP, which serves to protect this chain from restriction endonuclease digestion. After nicking the heteroduplex template strand with an appropriate restriction enzyme, the template strand is mutated to the site (group) to be mutated.
Can be digested with Exo III nuclease or another suitable nuclease. The reaction is then stopped, leaving molecules that are only partially single-stranded. A complete DNA homoduplex is then formed using DNA polymerase in the presence of all four deoxyribonucleotide triphosphates, ATP and DNA ligase. This homoduplex was then transformed into E. coli JM10 as described above.
1, etc., into a suitable host cell.

Mutants having more than one amino acid to replace can be created by any of several methods. When these amino acids are in close proximity to each other in a polypeptide chain, they can be mutated simultaneously with one oligonucleotide encoding all of the desired amino acid substitutions. However, when amino acids are present at some distance from each other (separated by about 10 or more amino acids), creating a single oligonucleotide encoding all of the desired changes is a relative Difficult. Instead,
Either of two different methods can be used.

In the first method, an oligonucleotide is created independently for each of the amino acids to be substituted. Then, when these oligonucleotides are simultaneously annealed to the single-stranded template DNA, the second template synthesized from this template is synthesized.
The DNA strand will encode all of the desired amino acid substitutions. This first round is the same as described for the single mutation. That is, using wild-type DNA as a template, an oligonucleotide encoding the first desired amino acid substitution (s) is annealed to this template, and then a heteroduplex DNA molecule is created. The second round of mutagenesis uses the mutated DNA obtained from the first round of mutagenesis as a template. That is, the template already contains one or more mutations. Then
Annealing an oligonucleotide encoding another desired amino acid substitution (s) to this template results in
The DNA strand encodes mutations from both the first and second rounds of mutagenesis. This obtained DN
A can be used as a template, etc. in the third round of mutagenesis.

B. Cassette Mutagenesis This method is also a preferred method for preparing substitution, deletion and insertion mutants of Gene 1. This method is wells
[Gene 34: 315 (1985)]. The starting material is a plasmid (or other vector) containing gene 1, the gene to be mutated. The codon in gene 1 to be mutated has been determined. There must be only one restriction endonuclease site on each side of the determined mutation site (s). If such restriction sites are not present, they can be created using the above-described oligonucleotide-mediated mutagenesis method and introduced at the appropriate location in gene 1. After introducing restriction sites into the plasmid, the plasmid is cut at these sites to linearize it. Double stranded oligonucleotides encoding the DNA sequence between these restriction sites and containing the desired mutation are synthesized by conventional methods. The two chains are synthesized separately and then hybridized together in a conventional manner. This double-stranded oligonucleotide is called a cassette. This cassette is designed to have 3 'and 5' ends that match both ends of the linearized plasmid, so that it can be directly ligated to the plasmid. This plasmid will then contain the mutated DNA sequence of gene 1.

VI. Preparation of DNA Encoding a Desired Protein In another embodiment, the present invention contemplates the production of a variant of the desired protein containing one or more subunits. Usually, each subunit is encoded by another gene. Each gene encoding each subunit can be obtained by methods known in the art (see, eg, Section II). In some cases, it may be necessary to obtain genes encoding various subunits using independent methods selected from any of the methods described in Section II.

When constructing a replicable expression vector in which the desired protein contains more than one subunit, all subunits are replaced by the same promoter, usually located 5 'to the DNA encoding the subunit. Alternatively, each subunit may be regulated by an independent promoter appropriately oriented in the vector, wherein each promoter is operably linked to the DNA to be regulated.
The selection of the promoter is performed as described in section III above.

Reference should be made to FIG. 10 when constructing a replicable expression vector containing DNA encoding a desired protein having multiple subunits, where the vector is illustrated for illustration. And the DNA encoding each subunit of the antibody fragment is shown.
This figure usually shows one of the subunits of the desired protein.
Shows that one is fused to a phage coat protein such as M13 gene III. Usually, the gene fusion will contain its own signal sequence. It can be seen that another gene encodes another subunit or group of subunits, and each subunit usually has its own signal sequence. FIG. 10 also shows that a single promoter can regulate the expression of both subunits. Alternatively, each subunit may be independently regulated by a different promoter. The desired protein subunit-phage coat protein fusion construct can be prepared as described in Section IV above.

When constructing a group of variants of the desired multi-subunit protein, the DNA encoding each subunit in the vector may be mutated at one or more positions in each subunit. When multi-subunit antibody variants are constructed, preferred mutagenesis sites encode amino acid residues located in the complementarity determining regions (CDRs) of either the light chain, the heavy chain, or both chains. Matches the codon that is. Usually, this CDR is called a hypervariable region. The method of mutating the DNA encoding each subunit of the desired protein is performed substantially as described in Section V above.

VII. Preparation of Target Molecules and Binding to Phagemids Target proteins, such as receptors, can be isolated from natural sources or prepared by recombinant methods by methods known in the art. To illustrate, glycoprotein hormone receptors are described by McFarland et al. [Science 245: 494-499].
(1989)], and the non-glycosylated form expressed in E. coli is described by Fuh et al. [J. Biol. Chem. 265: 3111-3115 (1990)]. I have. Other receptors can be prepared by a conventional method.

Purify the target protein with a suitable matrix such as agarose beads, acrylamide beads, glass beads, cellulose, various acrylic copolymers, hydroxyalkyl methacrylate, polyacrylic and polymethacrylic acid copolymers, nylon, neutral and ionic carriers, etc. Can be combined. Binding of the target protein to the matrix is described in the literature [Mothods in Enzymol
ogy 44 (1976)] or by other means known in the art.

After binding the target protein to the matrix, the immobilized target is contacted with a library of phagemid particles under conditions suitable for binding at least a portion of the phagemid particles to the immobilized target. Usually, conditions including pH, ionic strength, temperature, etc. will be similar to physiological conditions.

Bound phagemid particles having a high affinity for the immobilized target ("binders") are separated from the low affinity particles (and thus those that do not bind to the target) by washing. The conjugate can be dissociated from the immobilized target by various methods. These methods include competitive dissociation using changes in wild-type ligand, pH and / or ionic strength, and methods known in the art.

Appropriate host cells are infected with the conjugate and helper phage, and the host cells are cultured under conditions suitable for the amplification of phagemid particles. The phagemid particles are then collected and the selection step is repeated one or more times until a conjugate having the desired affinity for the target molecule is selected.

Optionally, a library of phagemid particles can be
Can be continuously contacted with more than one immobilized target to improve selectivity for a particular target. For example, ligands such as hGH often have more than one natural receptor. In the case of this hGH, both the growth hormone receptor and the prolactin receptor bind to the hGH ligand. hGH
May be desirable for growth hormone receptors over prolactin receptors.
This involves first contacting the phagemid particle library with the immobilized prolactin receptor, eluting it with low affinity (ie, lower than wild-type hGH) for the prolactin receptor, and then binding this low affinity prolactin "binding" This can be accomplished by contacting the "body" or unconjugated with the immobilized growth hormone receptor and selecting a high affinity growth hormone receptor conjugate. In this case, a mutant of hGH that has low affinity for the prolactin receptor has therapeutic use even when the affinity for the growth hormone receptor is slightly lower than that of wild-type hGH. Will be. This same method can be used to improve the selectivity of a particular hormone or protein for its primary functional receptor over its clearance receptor.

In another aspect of the invention, improved substrate amino acid sequences can be obtained. These are used to create better "cleavage sites" for protein linkers.
Or it may be useful for better protease substrates / inhibitors. In this embodiment, an immobilizable molecule (eg, an hGH-receptor, biotin-avidin, or a molecule capable of covalently binding to a matrix) is fused to gene III by a linker. The linker is preferably 3 to 10 amino acids in length and acts as a substrate for a protease. Phagemids are constructed as described above (where a random mutation is made to the DNA encoding the linker region to prepare a random library of phagemid particles having different amino acid sequences at the binding sites). This library of phagemid particles is then immobilized on a matrix and exposed to the desired protease. Phagemid particles having a preferred or better substrate amino acid sequence in the liner region of the desired protease will be eluted, and an enriched pool of phagemid particles will be obtained initially, gorging the preferred linker. These phagemid particles are then repeated several more times to obtain an enriched pool of particles encoding the consensus sequence (s) (see Examples XIII and XIV).

VIII. Growth Hormone Variants and Methods of Use The cloned gene for hGH is expressed in a secreted form in Eschericha cola [Chang, CN et al., Gene 55: 189 (198
7)], and its DNA and amino acid sequences have been reported [Goeddel et al., Nature 281: 544 (1979)]; Gray et al., Gene 3
9: 247 (1985)]. The present invention describes novel hGH variants obtained using the phagemid selection method. 10, 1
4, 18, 21, 167, 171, 172, 174, 175, 176, 178 and
A human growth hormone variant containing a substitution at position 179 has been described. Mutants with relatively high binding affinity are listed in Table VI
I, XIII and XIV. Amino acid nomenclature for describing these variants is provided below. Growth hormone variants may be administered and formulated in the same way as regular growth hormone. The growth hormone variants of the present invention can be expressed in any recombinant system capable of expressing native or met hGH.

Formulations of hGH for therapeutic administration comprise hGH of the desired degree of purification.
Can optionally be a physiologically acceptable carrier, excipient or stabilizer [Remington's Pharmaceutical Sciences, 16
Edition, Osol, A. Ed. (1980)], and prepared for storage in the form of a lyophilized cake or aqueous solution. Acceptable carriers, excipients or stabilizers are non-toxic to recipients at the dosages and concentrations employed, and include buffers, for example, phosphate, citrate, and other organic acids. Buffer); antioxidants (including ascorbic acid);
Low molecular weight (less than about 10 residues) polypeptides; proteins (eg, serum albumin, gelatin, or immunoglobulins); hydrophilic polymers (eg, polyvinylpyrrolidone); amino acids (eg, glycine, glutamine,
Monosaccharides, disaccharides and other carbohydrates (including glucose, mannose or dextrin); chelating agents (eg, EDTA); divalent metal ions (eg, zinc, asparagine, arginine, or lysine);
A sugar alcohol (eg, mannitol or sorbitol); a counter ion that forms a salt (eg, sodium); and / or a nonionic surfactant [eg, Tween, Pluronic or polyethylene glycol (PEG)]. included. The formulation of the present invention may further contain a pharmaceutically acceptable buffer, amino acid, bulking agent, and / or nonionic surfactant. These include, for example, buffers, chelating agents, antioxidants, preservatives, cosolvents, and the like. Examples of these would include trimethylamine salt ("Tris buffer"), and disodium edetate. The phagemid of the present invention can be used to produce large amounts of hGH mutants without phage proteins. To express hGH variants that do not contain the gene III portion of the fusion, pS0643 and derivatives can simply be grown in non-suppressor strains such as 16C9. In this case, the amber codon (TAG) leads to termination of translation and free hormone is obtained without the need for independent DNA assembly. The hGH variant is secreted from the host and can be isolated from the culture medium.

8 hGH amino acids F10, M14, H18, H21, R167, D17
One or more of the amino acids 1, 175 and 1179 can be substituted with any amino acid other than the amino acid found at that position in the native hGH shown here. That is, amino acids F10, M14, H18, H2 shown here
1, 2, 3, 3 of R167, D171, T175 and 1179
4, 5, 6, 7, or all 8 can be substituted with any of the other 19 amino acids of the 20 amino acids listed below. In a preferred embodiment, the 8
All amino acids are replaced with another amino acid. The most preferred eight amino acids to be substituted are shown in Table XIV in Example XII.

Amino acid nomenclature Ala (A) Arg (R) Asn (N) Asp (D) Cys (C) Gln (Q) Glu (E) Gly (G) His (H) Ile (I) Ieu (L) Lys (K Met (M) Phe (F) Pro (P) Ser (S) Thr (T) Trp (W) Tyr (Y) Val (V) The one-letter nomenclature for the hGH variant is the hGH amino acid ( For example, glutamate 179) is listed first, followed by the inserted amino acid (eg, serine), resulting in (E1795S).

EXAMPLES Without further elaboration, it is believed that one skilled in the art can, using the preceding description and illustrative examples, fully implement and utilize the present invention. That is, the following specific examples illustrate preferred embodiments of the present invention and should not be considered as limiting the remainder of the disclosure in any way.

Example 1 Plasmid Construction of and hGH- preparation of phagemid particles plasmid phGH-M13g III (Fig. 1), M13KO7 ⁷ and hGH production plasmid pBO473 [Cunningham, BC, et al., Scien
ce 243: 1330-1336 (1989)]. Synthetic oligonucleotide: Was used to introduce a unique Apa I restriction site (underlined) after the last Phe191 codon of hGH in pBO473. Oligonucleotides: Using a unique Apa I restriction site (first underlined) and a Glu197-Amber stop codon (second underlined).
It was introduced into M13KO7 gene III. Oligonucleotides: Due to the unique Nhe I after the 3 'end of the gene III coding sequence
The site (underlined) is introduced. Double mutated M13K
The resulting 650 base pair (bp) Apa I from O7 gene III
The NheI fragment was converted to the large ApaI-NheI
It was cloned into the fragment to create plasmid pSO132. This allows the insertion of the glycine residue encoded from the Apa I site while inserting the carboxy terminus of hGH (Phe191).
Is fused to the Pro198 residue of the gene III protein, and this fusion protein is ligated to E. coli alkaline phosphatase (pho
A) Promoter and st II secretion signal sequence [Chang,
CN et al., Gene 55: 189-196 (1987)]. The phoA promoter was replaced with a lac promoter and operator for inducible expression of the fusion protein in rich media. 138 bp EcoR I- containing lac promoter, operator, and Cap binding site
The Xba I fragment was used to convert the oligonucleotide: [They border the desired lac sequence and introduce EcoR I and Xba I restriction sites (underlined)]. This lac fragment was gel purified and the large EcoR
The plasmid phGH-M13g was ligated to the I-XbaI fragment.
III was created. The sequences of all processed DNA junctions were confirmed by dideoxy sequencing [Sanger, F. et al.
Proc. Natl. Acad. Sci. USA 74: 5463-5467 (1977)]. R6
The 4A mutant hGH phagemid was constructed as follows. That is, Ala substitution of Arg64 (R64A) [Cunningham, BC, Wells,
JA, Science 244: 1081-1085 (1989)], and the Nsi I-Bg III mutant fragment of hGH [Cunningham et al. (Supra)] was transformed into a phGH-M13g III plasmid (FIG. 1).
Wild-type hGH cloned between the corresponding restriction sites in
The sequence was replaced. The R64A hGH phagemid particles were grown and titrated as described below for wild-type hGH-phagemid.

Plasmid was introduced into a male strain of E. coli (JM101),
Selected on carbenicillin plates. A single transformant was grown at 37 ° C. for 4 hours in 2YT medium (2 ml), and M1
The cells were infected with 3KO7 helper phage (50 μl). The infected culture was diluted in 2YT (30 ml), cultured overnight, and phagemid particles were collected by precipitation with polyethylene glycol [Vierra, J., Messing, J., Methods in Enzymolo.
gy 153: 3-11 (1987)]. Typically, phagemid particle titers were in the range of 2-5 × 10 ¹¹ cfu / ml. Cs
Cl density centrifugation [Day, LA, J. Mol. Biol. 39: 265-277 (196
9)] to remove any fusion proteins not bound to virions.

Example II Immunochemical analysis of hGH on fusion phage Rabbit polyclonal antibody against hGH was purified using protein A, and 50 mM sodium carbonate buffer (pH 10)
Microtiter plates (Nunc) at a concentration of 2 μg / ml at 4 ° C. for 16-20 hours. PB containing 0.05% Tween 20
After washing in S, hGH or hGH-phagemid particles were washed with buffer A [50 mM Tris (pH 7.5), 50 mM NaCl, 2 mM EDT
A, 5 mg / ml bovine serum albumin, and 0.05% Tween 2
0] to 2.0-0.002 nM. 2 at room temperature (rt)
After time, the plates were washed extensively and the indicated Mab [C
unningham et al. (above)] at rt for 2 hours in buffer A at 1 μg
/ ml. After washing, horseradish peroxidase-conjugated goat anti-mouse IgG antibody was allowed to bind for 1 hour at rt. After the last wash, peroxidase activity was assayed using the substrate o-phenylenediamine.

Example III Binding and Enrichment of hGH Binding Protein to Polyacrylamide Beads Oxiran Polyacrylamide Beads (Sigma)
Is an extra cysteine residue introduced by site-directed mutagenesis at position 237 (does not adversely affect hGH binding;
J. Wells: Unpublished). The purified extracellular domain of the hGH receptor (hGHbp) containing [Fuh, G. et al., J. Biol. Chem. 265: 3111-31].
15 (1990)]. This hGHbp is
Conjugated to a quantity of 1.7 pmol hGHbp / mg dry oxirane beads as measured by [ ¹²⁵ I] hGH binding to the resin, as recommended by the supplier. Then, all unreacted oxirane groups were blocked with BSA and Tris. As a control for non-specific binding of phagemid particles, BSA was similarly bound to the beads. Buffers for adsorption and washing are 10 mM Tris-HCl (pH 7.5),
It contained 1 mM EDTA, 50 mM NaCl, 1 mg / ml BSA and 0.02% Tween-20. The elution buffer contained 200 nM hGH or 0.2 M glycine (pH 2.1) in addition to the wash buffer. Parental phage M13KO7 was converted to hGH phagemid particles by approximately 300
Mix at a ratio of 0: 1 (initial mixture) and mix with either absorbent (5 μl each; 0.2 mg acrylamide beads) in a 50 μl volume at room temperature for 8-12 hours. The beads were pelleted by centrifugation and the supernatant was carefully removed. The beads are resuspended in 200 μl of wash buffer and
The mixture was mixed for one hour (washing solution 1). After the second wash (Wash solution 2), the beads are washed twice with 200 nM hGH each for 6-10 minutes.
Elution was carried out over time (eluent 1, eluent 2). The final elution was performed for 4 hours using glycine buffer (pH 2.1) to remove the remaining hGH phagemid particles (eluate 3). Each fraction was appropriately diluted in 2YT medium, mixed with fresh JM101, incubated at 37 ° C. for 5 minutes, and plated on LB or LB carbenicillin plates using 2YT soft agar (3 ml).

Example IV Construction of hGH-phagemid particles with a mixture of gene III products The gene III protein consists of 410 residues and is divided into two domains separated by a variable linker sequence [Armstrong, J. et al., FEBS Lett. 135 : 167-172 (198
1)]. The amino-terminal domain is required for attachment to E. coli fimbria, while the carboxy-terminal domain is integrated into the phage coat and required for proper phage assembly [Crissman, JW, Smith, GP , Virolog
y 132: 445-455 (1984)]. The signal sequence and amino-terminal domain of gene III were replaced by the stII signal and the entire hGH gene [Chang et al., Supra] by fusion to residue 198 in the carboxy-terminal domain of gene III (FIG. 1). This hGH-gene III fusion was converted to the β-
Plasmid containing the lactamase gene and the ColE1 origin of replication and the intergenic region of phage fl (pbGH-M13gI
II; placed under the control of the lac promoter / operator in FIG. 1). This vector can be easily maintained as a small plasmid vector by selection with carbenicillin, thereby providing a functional gene for propagation.
Dependence on the III fusion is avoided. This plasmid can also be efficiently packaged into virions (ie, phagemid particles) by infection with helper phage such as M13KO7 [Viera et al. (Supra)], thereby avoiding the problem of phage assembly. . The infectivity titer of phagemid based on transduction for carbenicillin resistance in this system varied from 2 to 5 × 10 ¹¹ colony forming units (cfu) / ml. The titer of M13KO7 helper phage in these phagemid stocks is 1010 ¹⁰
Plaque forming units (pfu) / ml.

This system was used to confirm a previous study [Parmley, Smith (supra)] that homogenous expression of large proteins fused to gene III was detrimental to phage production (data not shown). For example, phGH- by adding IPTG
Induction of the lac promoter in M13g III resulted in low phagemid titers. Furthermore, phagemid particles obtained by co-infection with M13KO7 containing an amber mutation in gene III gave very low phagemid titers (<10 ¹⁰ cfu / ml). We speculated that multiple copies of the gene III fusion bound to the phagemid surface could lead to multiple point binding ("chelation") of the fused phage to the immobilized target protein. Thus, to control the copy number of the fusion protein, E. coli JM101 containing constitutively high levels of the lac repressor protein
By culturing the plasmid in (lacl ^Q ), hGH
-Restricted transcription of the gene III fusion. phGH-M13g III
E. coli JM101 cultures containing E. coli were fully grown and infected with M13KO7 in the absence of the 1ac operon inducer (IPTG) (however, this system is flexible and has other genes
III can co-equilibrate with co-expression of the fusion protein). From the ratio of the amount of hGH per pillion (based on the hGH immunoreactive material in phagemid purified on a CsCl gradient), about 10% of the phagemid particles contained one copy of hGH.
It is estimated to contain the gene III fusion protein. Therefore, the titer of the fusion phage displaying the hGH gene III fusion is about 2-5 × ^{10 10} / ml. This number is
The titer of E. coli in the culture from which these were derived (~ 10 ⁸ -10 ⁹ / m
considerably larger than l). That is, on average, all E. coli cells produce 10-100 copies of phage decorated with the hGH gene III fusion protein.

Example V Structural Integrity of hGH-Gene III Fusions Immunoblot analysis of hGH-gene III phagemid (FIG. 2) shows that hGH cross-reactive material co-migrate with phagemid particles in an agarose gel. This indicates that hGH is tightly bound to the phagemid particles. HGH from this phagemid particle
-The gene III fusion protein migrates as a single immunostained band and hGH binds to gene III when hGH binds to gene III
Indicates that there is little decomposition of H. Wild-type gene III protein is clearly present, as about 25% of the phagemid particles are infectious. This is similar to the concentration assessed by UV absorption and specific infectivity assessment performed on similarly purified (by CsCl density gradient) wild-type M13 phage [Smith, GP (supra) and Parmley, Sm.
ith (above)]. That is, both the wild-type gene III and the hGH-gene III fusion protein are displayed in the phage pool.

It was important to ensure that the indicated tertiary structure of hGH was maintained, in order to obtain confidence that the results of the binding selection were converted to the native protein. The structural integrity of the indicated hGH gene III fusion protein was evaluated using a monoclonal antibody (Mab) to hGH (Table I).

The epitopes on hGH for these Mabs have been mapped [Cunningham et al., Supra], and binding to seven of the eight Mabs requires that hGH be properly folded. The IC ₅₀ values of all Mab were equivalent to wild-type hGH except for Mab 5 and 6. Both Mab 5 and 6 are known to have binding determinants near the carboxy terminus of hGH that are blocked in the gene III fusion protein. Mabs that react with both native and denatured hGH
1 relative an IC ₅₀ value, conformationally sensitive Mab 2～5,7
And 8 remain unchanged. That is, Mab 1
Serves as a good internal control of any errors in adapting the concentration of the hGH standard to the concentration of the hGH-gene III fusion.

Example VI Enrichment of Binding to Receptor Affinity Beads Previous investigators [Parmley, Smith (supra); Smith
(Supra); Cwirla et al. (Supra); and Devlin et al. (Supra)] fractionated phage by sorting on streptavidin-coated polystyrene dishes or microtiter plates. However, chromatographic systems allow for more efficient fractionation of phagemid particles displaying muteins with different binding affinities. We use non-porous oxirane beads (Sigm
a) was selected to avoid capture of phagemid particles in the chromatography resin. Furthermore, these beads have a small particle size (1 μm), maximizing the ratio of surface area to volume. Contains free cysteino residues
The extracellular domain of the hGH receptor (hGHbp) [Fuh et al. (supra)] binds these beads efficiently, and phagemid particles exhibit extremely low non-specific binding to beads bound only to bovine serum albumin. (Table II).

In a representative enrichment experiment (Table II), 1 part of hGH phagemid was mixed with> 3,000 parts of M13KO7 phage.
After one cycle of binding and elution, ¹⁰⁶ phage were recovered, the ratio M13KO7 phage phagemid was 2: 1. That is,> 5000-fold enrichment was obtained by one binding selection step. Additional elution by free hGH or acid treatment to remove residual phagemid resulted in higher enrichment. This enrichment was achieved using Smith elution from coated polystyrene plates using Smith.
And the collaborators gained [Smith,
GP (above) and Parmley, Smith (above)]. However, even smaller volumes are used for beads (200 μl vs 6 ml). There was almost no hGH phagemid enrichment over M13KO7 when using beads bound only to BSA. Slight enrichment observed for control beads (~ 10-fold over pH 2.1 elution; Table 2)
May be derived from trace impurities of bovine growth hormone binding protein present in BSA binding to the beads. Nevertheless, these data indicate that hGH phage enrichment is dependent on the presence of hGHbp on the beads, suggesting that binding occurs by a specific interaction between hGH and hGHbp.

We assessed the enrichment of wild-type hGH over relatively weak binding variants of hGH on fusion phagemids to further investigate the specificity of the enrichment and to reduce the binding affinity for purified hormone after fusion phagemid selection. Associated with the enrichment factor. HGH mutant in which Arg64 is replaced with Ala (R64A)
Was used to construct a fusion phagemid. This R64A mutant hormone has about a 20-fold decrease in receptor binding affinity compared to hGH [Kd values of 7.1 nM and 0.34 nM, respectively; Cunnin
gham, Wells (above)]. The titer of the R64A hGH-gene III fusion phagemid was comparable to that of wild-type hGH phagemid. After one round of binding and elution (Table III),
Wild-type hGH phagemid consists of two types of phagemid and M13
The mixture of KO7 was enriched 8-fold for phagemid R64A and 1010 ⁴ for M13KO7 helper phage.

Conclusions Wild-type gene III and gene I on phagemid particles
By displaying a mixture of II fusion proteins, virions displaying large, properly folded proteins as fusions to gene III can be assembled and propagated. Effective control of the copy number of the gene III fusion protein allows selection of large epitope libraries (> 10 ¹⁰ ), avoiding “chelation” still maintained at sufficiently high levels in the phagemid pool . We have shown that hGH (a 22 kD protein) can be displayed in its native, folded form. Binding selections eluted with free hGH and performed on receptor affinity beads efficiently enriched wild-type hGH phagemid over mutant hGH phagemid which was shown to have low receptor binding affinity . That is, phagemid particles having a reduced binding constant in the nanomolar range can be selected.

Protein-protein and antibody-antigen interactions are governed by discontinuous epitopes [Janin,
J. et al., J. Mol. Biol. 204: 155-164 (1988); Argos, P., Pro.
t.Eng. 2: 101-113 (1988); Barlow, DJ et al., Nature 322:
747-748 (1987); and Davies, DR et al., J. Biol. Chem.
263: 10541-10544 (1988)]. That is, residues directly involved in binding are close together in the tertiary structure, but are separated by residues not involved in binding. The screening system provided by the present invention further advantageously analyzes protein-receptor interactions and allows the isolation of discontinuous epitopes in proteins with novel and high affinity binding It is.

Example VII Selection of hGH Mutants from a Library Randomized at hGH Codons 172, 174, 176, 178 Construction of Templates A mutant of the hGH-gene III fusion protein was
Constructed using the method of el et al. [Meth. Enzymol. 154: 367-382 (1987)]. As template DNA, the plasmid pS0132 (containing the native hGH gene fused to the carboxy-terminal half of M13 gene III under the control of the alkaline phosphatase promoter) is grown in CJ236 cells with the addition of M13-K07 phage as a helper. Was prepared. Single-stranded uracil-containing DNA was prepared for mutagenesis to assay (1) mutations in hGH that greatly reduced binding to the hGH binding protein (hGHbp); and (2) assay for parental background phage. , A unique restriction site (Kpn I) that could be used to select against was introduced. T7 DNA polymerase and the following oligodeoxynucleotides: Was used to perform oligonucleotide-directed mutagenesis. This oligo was mutated in hGH (R178G, I179
Introduce a Kpn I site (shown above) with T). Clones from mutagenesis were screened by Kpn I digestion and confirmed by dideoxy DNA sequencing. This resulting construct, used as a template for random mutagenesis, was named pHO415.

Random mutagenesis in hGH helix 4 Again using Kunkel's method, codons 172, 174, 17
6,178 were targeted for random mutagenesis in hGH.
A single-stranded template was prepared from pHO415 as described above and the following oligos: hGH codons: Mutagenesis was performed using a pool of As shown above, this oligo pool has codon 179 as wild type (Ile).
To destroy the unique Kpn I site of pH0415 and random codons at positions 172, 174, 176 and 178 (NNS; where N is A, G, C or T, and S is G or C
) Is introduced. This codon choice cannot be used to create additional Kpn I sites for the above sequences. By selecting this NNS synonymous sequence, 3 sites
Two possible codons were obtained (one “stop” codon and at least one codon for each amino acid), for a total of (32) ⁴ = 1,048,576 possible nucleotide sequences, 12% of which were at least one stop codon ) Or (20) ⁴ = 160,000 possible polypeptide sequences and 34,
481 immature, terminated sequences (ie, sequences containing at least one stop codon) are obtained.

Propagation of first library Mutagenesis with phenol: chloroform (50: 5
Extracted twice at 0) and ethanol precipitated with excess carrier tRNA to avoid adding salts which would complicate the subsequent electroporation step. About 50 ng (15 fmol) of DNA is
In a total volume of 45 μl in the cuvette, WJM101 cells (2.8 ×
10 ¹⁰ cells / ml) were electroporated.

The cells were allowed to recover for 1 hour at 37 ° C. with shaking, then 2YT medium (25 ml), 100 μg / ml carbenicillin, and
M13-K07 (multiplicity of infection = 1000). By plating serial dilutions from this culture on a carbenicillin-containing medium, it was found that 8.2 × 10 ⁶ electrotransformants were obtained. After 10 minutes at 23 ° C., the culture was incubated overnight (15 hours) at 37 ° C. with shaking.

After overnight incubation, pellet the cells and add pLIB
Double-stranded DNA (dsDNA) designated as 1 was prepared by an alkaline lysis method. The supernatant was centrifuged once again to remove all remaining cells, and a phage named phage pool φ1 was PEG precipitated, and STE buffer [10 mM Tris (pH 7.6), 1
mM EDTA, 50 mM NaCl] (1 ml). Phage titers, measured as the hGH-g3p gene III fusion (hGH-g ³⁾ For recombinant phagemid containing plasmid colony forming units (CFU), and the plaque forming units (PFU) for M13-K07 helper phage did.

Binding selection using immobilized hGHbp 1. Binding: A portion of phage pool φ1 (6 × 10 ⁹ CFU, 6 × 10 ⁷ PFU) was buffered into buffer A [phosphate buffered saline, 0.5% BSA, 0.05%
Tween 20, 0.01% thimerosal] diluted 4.5 times,
In a 1.5 ml silane-treated polypropylene test tube, it was mixed with a suspension (5 μl) of oxirane-polyacrylamide beads conjugated to hGHbp (350 fmol) containing the Ser237Cys mutation. As a control, equal amounts of phage were mixed with beads coated with BSA alone in a separate tube. The phage were allowed to bind to the beads by incubating for 3 hours at room temperature (23 ° C.) with slow rotation (about 7 rpm). The following steps were performed at room temperature using a constant volume of 200 μl.

2. Washing: The beads were spun for 15 seconds and the supernatant was removed (Supernatant 1). The beads were washed twice by resuspending in buffer A to remove non-specifically bound phage / phagemid and then pelleted. A final wash was performed by rotating the beads in buffer A for 2 hours.

3. hGH elution: Phage / phagemid weakly bound to the beads was removed by stepwise elution with hGH. In the first step, the beads were spun with buffer A containing 2 nM hGH. After 17 hours, the beads were pelleted, resuspended in buffer A containing 20 nM hGH, spun for 3 hours, and then pelleted. In the last hGH cleaning,
The beads were suspended in buffer A containing 200 nM hGH,
Spin for hours and then pelletize.

4. Glycine elution: To remove the most tightly bound phagemids (ie, phagemids still bound after the hGH wash), the beads were washed with glycine buffer [1M glycine, HC
, pH 2.0], spun for 2 hours and pelletized. The supernatant (fraction “G”; 200 μl) was neutralized by the addition of 1 M Tris base (30 μl).

Fraction G eluted from hGHbp-beads (1 × 10 ⁶ CFU, 5 × 10 ⁴
PFU) was not substantially more phagemid rich than the K07 helper phage. We found that the K07 phage packaged during growth of the recombinant phagemid h
This is attributed to displaying the GH-g3p fusion.

However, when compared to fraction G eluted from BSA-coated control beads, the hGHbp-beads provided a 14-fold higher CFU. This is more than non-specifically bound phagemid,
It reflects the enrichment of phagemid displaying tightly bound hGH.

5. Proliferation: Part of fraction G eluted from hGHbp-beads (4.3x
10 ⁵ CFU) were used to infect log-growth WJM101 cells. For transduction, fraction G (100 μl) was transferred to WJM101 cells (1 μm).
l), incubate at 37 ° C for 20 minutes, then
07 (multiplicity of infection = 1000). The culture (25 ml of 2YT and carbenicillin) was grown as above and a second pool of phage (library 1
G, for first glycine elution) was prepared as described above.

Phage from library 1G (FIG. 3) were selected for binding to hGHbp beads as described above. Fraction G eluted from hGHbp beads contained CFU 30 times higher in this selection than Fraction G eluted from BSA beads. Once again, a portion of fraction G was propagated in WJM101 cells to yield library 1G ² (indicating that it has selected the library twice by glycine elution). Also, 2
Single-stranded DNA (pLIB 1G ² ) was prepared from this culture.

To reduce the amount of dsDNA of Kpn I test and restriction selection background (Kpn I ⁺⁾ template, a portion of pLIB 1G ² (about 0.5 [mu] g) was digested with Kpn I, WJ
Electroporation was introduced into M101 cells. These cells were grown in the presence of K07 (multiplicity of infection = 100) as described for the original library and a new phage pool, pLIB3, was prepared (FIG. 3).

Further, a part (about 0.5 μg) of the dsDNA derived from the first library (pLIB1) was digested with KpnI and electroporated directly into WJM101 cells. Transformants were recovered as described above, infected with M13-K07, and grown overnight to obtain a new phage library named phage library 2 (FIG. 3).

Sequential round selection (1) Excess (10 times CFU) of purified K07 phage (not displaying hGH) was added into the bead binding cocktail; and (2) replacing the hGH step elution with a short wash of buffer A alone Phagemids from both pLIB2 and pLIB3 were bound, eluted and propagated in a continuous round, except as described above, against phagemids derived from both pLIB2 and pLIB3. In some cases, XL1-Blue cells were used for phagemid propagation.

An additional digestion of the dsDNA with Kpn I was performed on pLIB 2G ³ and pLIB 3G ⁵ prior to the final round of bead binding selection (FIG. 3).

The isolated clones selected phagemid DNA sequencing LIB 4G ⁴ from four independently isolated clones and LIB 5G ⁶ four independently derived from,
Sequenced by dideoxy sequencing. All eight clones have the same DNA sequence: hGH codon: Had. That is, all of them encode the same mutant of hGH (E174S, F176Y). Residue 172 in these clones is Lys as in the wild type. Also, 17
The codon chosen for 2 is also identical to wild type hGH.
This is not surprising since AAG is the only lysine-codon possible from the set of synonymous "NNS" codons. In addition, residue 178-Arg is also identical to wild-type, except that the codon selected from the library was AAG and not the CGC found in wild-type hGH (the latter codon was also a set of "NNS" codons). Is possible).

Multiplicity of K07 infection The multiplicity of infection of K07 infection is an important parameter in the growth of recombinant phagemid. The multiplicity of infection of K07 must be high enough to ensure that substantially all cells transformed or transfected with phagemid can package new phagemid particles. In addition, the concentration of wild-type gene III in each cell is maintained high, reducing the likelihood of multiple hGH-gene III fusion molecules being displayed on each phagemid particle, thereby reducing chelation at binding. Should be reduced. However, when the multiplicity of infection of K07 is too high, packaging of K07 will compete with packaging of the recombinant phagemid. We have a background K0 of 1-10%
An acceptable phagemid yield of only 7 phages was found to be obtained when the multiplicity of infection of K07 was 100.

The phage pool was labeled as previously shown (FIG. 3). The multiplicity of infection (moi) is determined by the K
07 means multiplicity of infection (PFU / cell). C beyond PFU
FU enrichment is shown when purified K07 is added to the binding step. HGHbp over CFU eluted from BSA beads
The ratio of CFU eluted from the beads is shown. Fractionation of the KpnI-containing template (ie, pH0415) remaining in the pool was achieved by digesting the dsDNA with KpnI and EcoRI and reducing the product to 1%.
Run on agarose gel and stain negative for ethidium bromide
The DNA was measured by laser-scanning.

Receptor binding affinity of the hormone hGH (E174S, F176Y) The fact that a single clone was isolated from two different alternative pathways (FIG. 3) indicates that the double mutant (E174S, F176Y)
176Y) strongly binds to hGHbp. To determine the affinity of this hGH mutant for hGHbp,
We constructed this hGH mutant by site-directed mutagenesis using plasmid pB0720. This plasmid contains the wild-type hGH gene as a template, and the following oligonucleotides that change codons 174 and 176: hGH codons: It contains. The resulting construct, pH0458B, was introduced into E. coli strain 16C9 for expression of the mutant hormone. hGHb
Scatchard analysis of the competitive binding of hGH (E174S, F176Y) and ¹²⁵ I-hGH to p showed that this (E174S, F176Y) mutant had at least 5.0 binding affinity of wild-type hGH.
It was found that the binding affinity was twice as strong.

Example VIII Selection of hGH Variants from a Hormone-Phage Helix 4 Random Cassette Library Human growth hormone variants were prepared by the method of the invention using the phagemids described in FIG.

Construction of De-Fusionable Hormone-Phage Vectors We (1) improve the binding between hGH and the gene III part to make display of the hGH part more advantageous on phage; (2) substantially To limit the expression of the fusion protein to obtain a "monovalent representation"; (3) allow selection of restriction nucleases relative to the starting vector; (4) eliminate expression of the fusion protein from the starting vector; and (5) Cassette mutagenesis [Wells et al., Gene 34: with the aim of achieving easy expression of the corresponding free hormone from one hGH-gene III fusion mutant.
315-323 (1985)] and hGH-gene III fusion proteins.

The hGH-gene III fusion protein (hGH residues 1-19) containing the pBR322 and f1 origins of replication and under the control of the E. coli phoA promoter [Bass et al., Proteins 8: 309-314 (1990)].
1, followed by one Gly residue and the gene III Pro-198
Oligonucleotide-directed mutagenesis of pS0132 [Kunkel et al., Methods Enzymo
l.154: 367-382 (1987)] to construct plasmid pS0643 (Fig. 9). Oligonucleotides: [This oligonucleotide introduces an Xba I site (underlined) and an amber stop codon (TAG) following Phe-191 of hGH]. In the resulting construct pS0643, part of gene III has been deleted, resulting in two silent mutagenesis (underlined),
The following junction points are created between hGH and gene III: This sets the overall size of the fusion protein to 401 in pS0132.
It shortens from residue to 350 residues in pS0643. Experiments with monoclonal antibodies to hGH have shown that the hGH portion of the new fusion protein assembled into phage particles is more accessible than previous long fusions.

To grow phage that display hormones, pS06
Amber-suppressor strains of E. coli such as JM101 or XL1-Blue [Bullock et al., BioTechniques
5: 376-379 (1987)]. Shown above is a Glu substitution at the amber codon that occurs in the supE suppressor strain. Suppression by other amino acids is also possible in various well-known and commonly available E. coli strains.

To express hGH (or mutant) that does not contain the gene III portion of the fusion, pS0643 and derivatives can simply be grown in non-suppressor strains such as 16C9. In this case, the amber codon (TAG) leads to the termination of translation and provides free hormone without the need for independent DNA assembly.

To create a site for cassette mutagenesis,
pS0643 to oligonucleotide: [This destroys the unique Bg III site of pS0643]; [This inserts a unique BstE II site, one base frameshift, and a non-amber stop codon (TGA)]; and [This introduces a new BgIII site] to give the starting vector pH0509. Adding a frameshift with a TGA stop codon ensures that no Gene III fusion can be produced from the starting vector. The BstE II-Bg III segment was excised from pH0509 and replaced with a mutated DNA cassette at the desired codon. Other restriction sites for cassette mutagenesis at other positions in hGH were also introduced into this hormone-phage vector.

Cassette Mutagenesis within the hGH helix hGH codons 172, 174, 176 and 178 were targeted for random mutagenesis. The reason for this is that all these hG
H at or near H and significantly contributes to receptor binding [Cunningham and Wells, Science 244: 1081-10
85 (1989)]; all of which exist within a well-defined structure and occupy two “turns” on the same side of helix 4; and each of these represents at least one of the known evolutionary variants of hGH. This is because it has been substituted with one amino acid.

We use NNS (N = A / G / C / T; S =
G / C). This selection of the NNS synonymous sequence yields 32 possible codons (codons for at least one each amino acid) at four sites, for a total of (32) ⁴
= 1,048,576 possible nucleotide sequences, or (20)
⁴ = 160,000 possible polypeptide sequences are obtained. The only stop codon, amber (TAG), is possible through this codon selection, and this codon can be suppressed as Glu1 in the supE strain of E. coli.

Two synonymous (degenerate) oligonucleotides with NNS at codons 172, 174, 176 and 178 were synthesized, phosphorylated and annealed to construct a mutation cassette: The vector was prepared by digesting pH0509 with BstE II followed by Bg III. The product was run on a 1% agarose gel, the large fragment was excised, phenol extracted and ethanol precipitated. This fragment was treated with bovine intestinal phosphatase (Boehringer), then phenol: chloroform extracted, ethanol precipitated and resuspended for ligation with the mutation cassette.

Growth of the first library in XL1-Blue cells After ligation, the reaction product was digested once more with BstE II,
Then phenol: chloroform extraction, ethanol precipitation and resuspension in water [BstE II recognition site (GGTNACC)
Is created in a cassette containing a G at position 3 of codon 172 and an ACC (Thr) codon at position 174. However, in this process
Treatment with BstE II should not select for all possible mutation cassettes. The reason for this is that virtually all cassettes are heteroduplex and cannot be cleaved by this enzyme]. Approximately 150 ng (45 fmoles) of DNA was added to XL1-Blue cells (0.0
1.8x10 ⁹ cells in 45ml), single pulse voltage setting 2.49k
Electroporation was introduced at V (time constant = 4.7 ms).

The cells were shaken at 37 ° C in SOC medium for 1 hour.
Allowed to recover for time, then mixed with 2YT medium (25 ml), 100 mg / ml carbenicillin, and M13-K07 (moi = 100).
After 10 minutes at 23 ° C., the culture was incubated overnight (15 hours) at 37 ° C. with shaking. By plating serial dilutions from this culture on a medium containing carbenicillin, it was found that 3.9 × 10 ⁷ electrotransformants had been obtained.

After an overnight incubation, pellet the cells to pH05.
Double-stranded DNA (dsDN) named 29E (first library)
A) was prepared by the alkali dissolution method. The supernatant was centrifuged once more to remove all remaining cells, and a phage named phage pool φH0529E (the first phage library) was PEG precipitated, and STE buffer [10 mM Tris (pH
7.6), 1 mM EDTA, 50 mM NaCl] (1 ml).
Phage titers were measured as colony forming units (CFU) for recombinant phagemid containing hGH-g3p.
About 4.5 × 10 ¹³ CFU was obtained from the starting library.

Synonyms of the starting library (degenerate) 58 clones from the pool of electrotransformants were
Sequences were made in the region of the II-Bg III cassette. Of these, 17% correspond to the starting vector and 17%
Contained at least one frameshift and 7% contained non-silent (non-termination) mutations outside the four target codons. We found that 41% of clones had drawbacks due to any of the above criteria,
10 ⁷ initially concluded that what remains are a total functional pool of transformants. Nevertheless, this number almost exceeds the number of possible DNA sequences by a factor of 20. Therefore, we are convinced that we have all possible sequences shown in the starting library.

We evaluated the sequence of unselected phage to assess the degree of codon bias in mutagenesis (Table V). Although these results indicate that some codons (and amino acids) are more or less than random predictions, the library is very diverse and evidence of a large amount of "sibling" subordination Was found not to be present (Table VI).

Preparation of immobilized pGHbp and hPRLbp Wild-type hGHbp [Fuh et al., J. Biol. Chem. 265: 3111-3115 (1
990)] except that [Bass et al., Proteins
8: 309-314 (1990)].
("HGHbp-beads") was prepared. Competitive binding experiments with [ ¹²⁵ I] hGH showed that 58 fmol of functional hGHbp bound per 1 μl of bead suspension.

Prolactin receptor [Cunningham et al., Science 250: 170
9-1712 (1990)] and immobilized hPRLbp ("hPRLbp-beads") as described above using the 211 extracellular domain.
Was prepared. Competition binding experiments with [ ¹²⁵ I] hGH in the presence of 50 μM zinc resulted in 2.
1 fmol of functional hPRLbp was found to bind.

“Blank beads” were prepared by mixing oxirane-acrylamide beads with 0.6 M ethanolamine (pH 9.2) at 4 ° C.
Prepared by treating for 5 hours.

Binding selection using immobilized hGHbp and hPRLbp Hormone-phage binding to beads was performed in any of the following buffers: buffer A [PBS, 0.5% BS for selection using hGHbp and blank beads.
A, 0.05% Tween 20, 0.01% Methylosal]; Buffer B [50 mM Tris (pH 7.5), 10 mM MgCl ₂ , 0.5% BSA, for selection using hPRLbp in the presence of zinc (+ Zn ²⁺ )
Buffer C [PBS, 0.5% BSA, 0.05% Tween 20, 0.01% Thimerosal, 10 mM EDTA for selection using hPRLbp in the absence of zinc (+ EDTA); 0.05% Tween 20, 100 mM ZnCl ₂ ]; ]. Binding selection was performed according to each of the following routes: (1) binding to blank beads, (2) binding to hGHbp-beads, (3) hPRLb.
binding to p-beads (+ Zn ²⁺ ), (4) binding to hPRLbp-beads (EDTA), (5) two pre-adsorptions with hGHbp beads followed by binding of unadsorbed fraction to hPRLbp-beads (“ -HGHbp, + hPRLbp "selection), or (6) two pre-adsorptions with hPRLbp-beads and subsequent binding of unadsorbed fractions to hGHbp-beads (" -hPRLbp, + hPRLbp ").
GHbp ”). The latter two procedures are mutants that bind hPRLbp but not hGHbp, respectively, or hGH
It is expected to enrich for mutants that bind bp but not hPRLbp. Phage binding and elution in each cycle was performed as follows: 1. Binding: a portion of the hormone phage (usually 10 ⁹ -10 ¹⁰ CF
U) is mixed with an equal volume of non-hormonal phage (pCAT), diluted with a suitable buffer (A, B or C) and hGHbp, hPRL in a total volume of 200 ml in 1.5 ml polypropylene tubes.
Mix with bp or blank bead buffer (10 ml).
The phage was bound to the beads by incubating at room temperature (23 ° C.) for 1 hour with slow rotation (about 7 rpm). The following steps were performed at room temperature using a constant volume of 200 μl.

2. Washing: The beads were spun for 15 seconds and the supernatant was removed. To reduce the number of non-specifically bound phage, the beads were washed five times by brief resuspension in the appropriate buffer and then pelleted.

3. hGH elution: Phage weakly bound to the beads were removed by elution with hGH. The beads were spun with the appropriate buffer containing 400 nM hGH for 15-17 hours. This supernatant was used as “hGH eluate” and beads. The beads were washed by brief resuspension in buffer and pelleting.

4. Glycine elution: The most tightly bound phage (ie,
To remove any phage still bound after the hGH wash, suspend the beads in glycine buffer [add 0.2 M glycine to buffer A and bring to pH 2.0 with HCl], spin for 1 hour, Pelletized. The supernatant ("glycine eluate"; 200 μl) was neutralized by the addition of 1 M Tris base (30 ml) and stored at 4 ° C.

5. Proliferation: XL in logarithmic growth phase using a part of hGH eluate and glycine eluate from each set of beads under each set of conditions
Independent cultures of 1-Blue cells were infected. For transduction, the phage was mixed with XL1-Blue cells (1 ml) and incubated at 37 ° C.
This was performed by incubating for 20 minutes and then adding K07 (moi = 100). Cultures (25 ml of 2YT and carbenicillin) were grown as above and the next pool of phage was prepared as above.

Phage binding, elution and propagation were performed in consecutive rounds according to the cycle described above. For example, phage amplified from hGH eluate from hGHbp beads was selected once more with hGHbp beads, eluted with hGH, and then used to infect a fresh culture of XL1-Blue cells. Five rounds of selection and three rounds of expansion were performed for each of the above selection methods.

DNA sequencing of selected phagemids A portion of the phage from the hGH and glycine elution steps of each cycle was used for inoculation of XL1-Blue cells, which were plated on LB medium containing carbenicillin and tetracycline and derived from each phage pool. Independent clones were obtained. Single-stranded DNA was prepared from the isolated colonies, and the mutation cassette region was sequenced. this
The results of DNA sequencing are summarized in FIGS. 5, 6, 7 and 8 as deduced amino acid sequences.

Expression and Assay of hGH Mutants DNA from clones sequenced to determine the binding affinity of some of the selected hGH mutants to hGHbp
Was introduced into E. coli strain 16C9. As described above, this strain is a non-suppressor strain that terminates protein translation after the last Phe-191 residue of hGH. Although single-stranded DNA was used for these transformations, either double-stranded DNA or whole phage can be easily electroporated into non-suppressor strains for free hormone expression.

Mutants of hGH were prepared from cells that were osmotically shocked by ammonium sulfate precipitation as described for hGH [Olson et al., Nature 293: 408-411 (1).
981)]. Protein concentration was also measured by laser densitometry on a Coomassie stained SDS-polyacrylamide gel electrophoresis gel using hGH as a standard [Cunningham and Wells, Science 244: 1081-1085.
(1989)].

The binding affinity of each mutant was determined using an anti-receptor monoclonal antibody (Mab263) as described in the literature [Spencer et al., J. Biol.
hem. 263: 7862-7867 (1988); Fuh et al., J. Biol. Chem. 265: 3.
111-3115 (1990)] by substituting ¹²⁵ I hGH.

The results of the number of hGH mutants selected by the different routes (FIG. 6) are shown in Table VII. Many of these mutants have h
It has a stronger binding affinity for GHbp than wild-type hGH. The most improved mutant, KSYR, is wild-type
It has 5.6 times higher binding affinity than hGH. The weakest mutant selected among these tested mutants had only about 10-fold lower binding affinity than hGH.

Binding assays can also be performed on mutants selected for hPRLbp-binding.

Additive and non-additive effects on binding At some residues, substitution of a particular amino acid has substantially the same effect, independent of surrounding residues. For example,
Substitution of F176Y in the 172R / 174S background reduces binding affinity by 2.0-fold (RSFR vs. RSYR). Similarly, 1
In a background of 72K / 174A, the binding affinity of the F176Y mutant (KAYR) was determined by the corresponding 176F mutant [KA
FR; Cunningham and Wells (1989)].

On the other hand, the binding constants measured for some selected hGH mutants show the non-additive effects of some amino acid substitutions at residues 172, 174, 176 and 178. For example, in the 172K / 176Y background, substitution of E174S results in a mutant (KSYR) that binds hGHbp 3.7-fold more tightly than the corresponding mutant containing E174A (KAYR). However, in the 172R / 176Y background, the effects of these E174 substitutions are reversed. In this case, the E174A mutant (RAYR) is replaced with the E174S mutant (RS
It binds 1.5 times stronger than (YR).

Such non-additive effects on the binding of substitutions at adjacent residues indicate the use of protein-phage binding selection as a means to select optimal mutants from a randomized library at some positions. Things. In the absence of detailed structural information, and without such a selection method, a large number of substitution combinations would be tried before finding the optimal mutant.

Example IX Selection of hGH Mutants from Helix 1 Random Cassette Library of Hormone-Phage Using the method described in Example VIII, hGHbp and / or
Or another region of hGH involved in binding to hPRLbp,
That is, residues 10, 14, 18, and 21 of helix 1 were replaced with phGHam-g3p
Targeted for random mutagenesis in the vector (also known as pS0643; see Example VIII).

We chose to use the “amber” hGH-g3 construct (referred to as phGHam-g3p) because it appears to create a target protein hGH that is more accessible for binding. This is supported by data from a comparative ELISA assay for monoclonal antibody binding. p
S0132 [S. Bass, R. Greene, JA Wells, Proteins 8: 306 (1
990)] and phage generated from both phGHam-g3 and three antibodies (Medix2, 1B5.G2, and 5B7.C10) known to have binding determinants near the carboxy terminus of hGH [5]. BCCunningham, P.Jhurani, P.Ng, JAWel
ls, Science 243: 1330 (1989); BCCunningham and J.
A. Wells, Science 244: 1081 (1989); L. Jin and J. Well
s (unpublished result)], and helices 1 and 3
Antibody (Medix1) recognizing determinants in BCCunn
ingham, P.Jhurani, P.Ng, JAWells, Science 243: 1330
(1989); BCCunningham and JAWells, Science 24
4: 1081 (1989)]. Phagemid particles from pbGHam-g3 reacted more strongly with antibodies Medix2, 1B5.G2 and 5B7.C10 than phagemid particles from pS0132. In particular, the binding of pS0132 particles was reduced> 2000-fold for both Medix2 and 5B7.C10 and> 25-fold for 1B5.G2 compared to binding to Medix1. On the other hand, the binding of the phGHam-g3 phage was only about 1.5-fold, 1.2-fold and 2.3-fold weaker for the Medix2, 1B5.G2 and 5B7.C10 antibodies, respectively, compared to the binding to Medix1.

Construction of Helix 1 Library by Cassette Mutagenesis Alanine-Scanning Mutagenesis [BCCunningh
am, P. Jhurani, P. Ng. JAWells, Science 243: 1330 (198
9); BCCunningham and JAWells, Science 244: 1081
(1989), 15, 16] by mutating residues in helix 1 previously identified by hGH and / or binding of the extracellular domains of the hPRL receptor (designated hGHbp and hPRLbp, respectively). Was modulated. Cassette mutagenesis has been described essentially in the literature [JAWells, M. Vasser, DBP
owers, Gene 34: 315 (1985)].
This library contains oligonucleotides: 5'-GCC TTT GAC AGG TAC CAG GAG TTT G-3 'and 5'-CCA ACT ATA CCA CTC TCG AGG TCT ATT CGA TAA
Only Kpn I (h
PhGHam-g3 [TA] with GH codon 28) and Xho I (hGH codon 6) restriction sites (underlined) inserted.
Kunkel, JDRoberts, RAZakour, Methods Enzymol. 154:
367-382] using cassette mutagenesis to mutate all four residues at once (see Example VIII). The latter oligo also introduced a +1 frameshift (the last G underlined), which terminated translation from the starting vector and minimized wild-type background in the phagemid library.
This starting vector was named pH0508B. hGH residues 10, 1
The helix 1 library with mutated 4, 18, 21 comprises complementary oligonucleotides: The cassette prepared from XhoI-Kpn pH0508B
Created by ligation to the I fragment. This K
The pn I site was broken at the junction of the ligation product, allowing the use of restriction digestion for analysis of the non-mutated background.

This library contains at least 10 ⁷ separate transformants, and when the library is completely random (10 ⁶ different codon combinations), on average about 10 copies of each possible mutant hGH The gene is obtained. Helix 1 by restriction analysis using Kpn I
It was found that at least 80% of the library construct contained the inserted cassette.

Enrichment of hGH-phage binding from this library was performed using hGHbp immobilized on oxirane-polyacrylamide beads (Sigma Chemical Co.) as described previously (Example VIII). Four residues in helix 1 (F10, M14, H18 and H21) were similarly mutated, giving a non-wild type consensus after 4 and 6 cycles (Table VIII). Position 10 on the hydrophobic side of helix 1 tends to be hydrophobic, whereas positions 21 and 18 on the hydrophilic side are predominantly Asn dominant, with no clear consensus at position 14 (Table IX).

The binding constant of these hGH mutants to hGHbp is determined by expressing the free hormone mutant in non-suppressor E. coli strain 16C9, purifying the protein and competing displacement of labeled wt-hGH from hGHbp. (Example VIII buffered). As shown, some mutants bind more strongly to hGHbp than do wt-hGH.

Example X Selection of hGH Variants from Helix 4 Random Cassette Library Containing Mutations Found Previously by Hormone-Phage Enrichment Design of Mutant Proteins with Improved Binding by Repeated Selection Using Hormone-Phages From our experiments with new unbound homologues of hGH evolution variants, obtain cumulatively improved mutants of hGH for binding to specific receptors by combining multiple independent amino acid substitutions [BCCunningham, DJ Henner, JA Wells, Science 247:
1461 (1990); BCCunningham and JAWells, Proc. Na
tl.Acad.Sci.USA 88: 3407 (1991); HBLowman, BCCun
ningham, JA Wells, J. Biol. Chem. 266, printing (199
1)].

In an attempt to further improve a helix 4 double mutant (E174S / F176Y; see Example VIII) selected from a helix 4a library found to bind more tightly to the hGH receptor, the helix 4b library It was created. Mutations were made to residues around positions 174 and 176 on the hydrophilic surface of helix 4, using the E174S / F176Y hGH mutant as the background starting hormone.

Construction of helix 4b library by cassette mutagenesis [J. Wells, M. Vasser, DBPowers, Gene
34: 315 (1985)] by cassette mutagenesis. Cassette mutagenesis using a mutant of phGHam-g3 (Example VIII) with pre-inserted unique BstE II and Bg III restriction sites, mutating all four residues at once (see Example VIII) Using E174S /
Residues 167, 171, 175 and 17 in F176Y background
A 9 helix mutated helix 4b library was created. Complementary oligonucleotide: Cassette prepared from the vector BstE II-Bg III
Inserted into the site. This BstE II site was deleted in the ligated cassette. Fifteen unselected clones from this helix 4b library were sequenced. Of these, none lacked the cassette insert, 20% were frameshifted and 7% had non-silent mutations.

Results of hGHbp enrichment hGH-phage binding enrichment from this library was performed using hGHbp immobilized on oxirane-polyacrylamide beads (Sigma Chemical Co.) as described previously (Example VIII). Was. After 6 cycles of coupling,
A reasonable and clear consensus has emerged (Table XI). Interestingly, all positions are polar residues, especially Ser, Thr
And Asn (Table XII).

Assay of hGH mutants The binding constants of some of these hGH mutants to hGHbp were determined by expressing the free
Determined by assaying by competitive displacement of labeled wt-hGH from p (see Example VIII). As shown,
The binding affinity of some helix 4b mutants for hGHbp was stronger than that of wt-hGH (Table XII).
I).

hGH Variant Receptor Selectivity Finally, we began the analysis of the binding affinities of several more robust hGHbp binding mutants for their ability to bind hPRLbp. The E174S / F176Y mutant binds 200 times weaker to hPRLbp than hGH. Each of the E174T / F176Y / R178K and R167N / D171S / E174S / F176Y / I179T mutants binds> 500-fold weaker to hPRLbp than hGH. That is,
Novel receptor-selective mutants of hGH can be obtained by phage display.

Information content of specific residues identified by hormone-phagemid selection Three hGH-phagemid libraries (Example VII
Of the 12 residues mutated in I, IX, X), 4
Species showed strong (but not exclusive) conservation of wild-type residues (K172, T175, F176 and R178). Not surprisingly, there were residues that, when converted to Ala, caused the greatest disruption (4-60 fold) in binding affinity for hGHbp. Ala substitution has a weaker effect on binding (2-
A group of four other residues (F10, M14,
D171 and I179) were present, and these positions showed little wild-type consensus. Finally, another four residues (H18, H21, R167 and E174) that promoted binding to hPRLbp but not hGHbp were identified by the selection with hGHbp-beads against the wild-type hGH sequence. No consensus was shown. In fact, two residues (E174 and H21) increase binding affinity for hGHbp by 2- to 4-fold when Ala substituted [BCCunningham, PJ
hurani, P. Ng. JA Wells, Science 243: 1330 (1989);
C. Cunningham and JA Wells, Science 244: 1081 (198
9); BCCunningham, DJ Henner, JAWells, Science 24
7: 1461 (1990); BCCunningham and JAWells, Proc,
Natl. Acad. Sci. USA 88: 3407 (1991)]. That is, the alanine-scanning mutagenesis data is reasonably related to the variability of replacing each position. In fact, the reduced binding affinity caused by alanine substitution [BCCunningham, P. Jhurani, P. Ng, JA Wells, Scie
nce 243: 1330 (1989); BCCunningham and JAWell
s, Science 244: 1081 (1989); BCCunningham, DJHenn
er, JAWells, Science 247: 1461 (1990); BCCunningh
am and JAWells, Proc. Natl. Acad.sci. USA 88: 3407
(1991)] reasonably predicts the rate at which wild-type residues are found in the phagemid pool after 3-6 rounds of selection. Alanine-scanning information is useful for targeting side chains that modulate binding, and phage selection can be used to optimize side chains and the variability of each site (and / or combination of sites) for substitution. Suitable for defining Scanning mutation method [BCCunningh
am, P. Jhurani, P. Ng, JAWells, Science 243: 1330 (198
9); BCCunningham and JAWells, Science 244: 1081
(1989)] and phage display is a powerful tool for designing receptor-ligand interfaces and studying molecular evolution in vitro.

Mutations in repeated enrichment of hormone-phagemid libraries Known as hormone-phagemid enrichment to improve binding if the combination mutation in hGH has an additive effect on binding affinity for the receptor The resulting mutations can be combined simply by restriction fragment cleavage and ligation or mutagenesis to obtain a cumulatively optimized mutant of hGH.

On the other hand, mutations in one region of hGH that optimize receptor binding may be structurally or functionally incompatible with mutations in overlapping or different regions of the molecule. In such cases, hormone-phagemid enrichment can be achieved by any of several variations of the following iterative enrichment methods: (1) cassette or other mutagenesis of the molecule. A random DNA library can be created in each of the two (or possibly more) regions: a mixed library can then be created by ligation of restriction fragments from these two DNA libraries; ( 2) hGH variants that have been optimized for binding by mutations in one region of the molecule can be randomly mutated in a second region of the molecule, as in the example of a helix 4b library; (3) Two or more random libraries were identified for improved binding by hormone-phagemid enrichment. After this "coarse selection" of the optimized binding site, another library which is still partly remixed by ligation of the restriction fragments to obtain two or more regions of the molecule A single library can be obtained that is partially different in, and this library can then be further selected for optimized binding using hormone-phagemid enrichment.

Example XI Assembly of Fab Molecules on Phagemid Surface Plasmid Construction Plasmid pDH188 is a DNA encoding the Fab portion of a humanized IgG antibody called 4D5 that recognizes the HER-2 receptor.
It contains. This plasmid is contained in E. coli strain SR101 and has been deposited with the ATCC [Rockville, MD].

Briefly, this plasmid was prepared as follows. The starting plasmid is pS0132, which contains the alkaline phosphatase promoter described above. The DNA encoding human growth hormone was excised and subjected to a series of operations to make the ends of the plasmid compatible for ligation, and then the DNA encoding 4D5 was inserted. This 4D5
DNA contains two genes. The first gene is
It encodes the variable and constant regions of the light chain and contains DNA encoding the stII signal sequence at its 5 'end. The second gene contains four parts, the first of which is 5 '
DNA encoding the terminal st II signal sequence.
This is followed by the DNA encoding the variable domain of the heavy chain, followed by the DNA encoding the first domain of the heavy chain constant region, and further by the DNA encoding the M13 gene III.
Followed by The salient features of this construct are shown in FIG. The sequence of the DNA encoding 4D5 is shown in FIG.

Transformation of E. coli and generation of phage Plasmids were introduced into SR101 cells using both polyethylene glycol (PEG) and electroporation. PEG competent cells were obtained from Chung and Miler [Nucleic Acid
s Res. 16: 3580 (1988)]. Cells Competent for Electroporation, Zabarovs
ky and Winberg [Nucleic Acids Res. 18: 5912 (199
0)] and then transformed by electroporation. SOC medium [described by Sambrook et al. (Supra)] (1
ml), and grown at 37 ° C. for 1 hour with shaking. At this point, cell concentration was measured using OD ₆₀₀ light dispersion. The titrated KO7 phage stock was added to give a multiplicity of infection (MOI) of 100, and the phage was allowed to attach to the cells at room temperature for 20 minutes. The mixture was then diluted in 25 ml of 2YT broth [described by Sambrook et al. (Supra)] and incubated overnight at 37 ° C. with shaking. The next day, cells were pelleted by centrifugation at 5000 xg for 10 minutes, the supernatant was collected, and 0.5M NaCl and 4% PEG
Phage particles were precipitated using (final concentration) for 10 minutes at room temperature. Pellet the phage particles by centrifugation at 10,000 xg for 10 minutes, TEN [10 mM Tris (pH 7.6),
1 mM EDTA and 150 mM NaCl] (1 ml),
Stored at ° C.

Preparation of antigen-coated plates 0.5 mg / ml solution of BSA (control antigen) or 0.1 mg / ml solution of extracellular domain (EDC) of HER-2 antigen in 0.1 M sodium bicarbonate (pH 8.5) Using each ml, Fa
The wells of an lcon12 well tissue culture plate were coated. After applying the solution to the wells, the plate was incubated overnight at 4 ° C. on a rocking platform. The plates were then blocked by removing the first solution, applying blocking buffer [30 mg / ml BSA in 0.1 M sodium bicarbonate] (0.5 ml) and incubating at room temperature for 1 hour. Finally, the blocking buffer was removed and buffer A [PBS, 0.5% BSA, and 0.05% Tween 20] (1 ml)
Was added and the plates were stored at 4 ° C. for up to 10 days until used for phage selection.

Phage particles of phage selection methods about ¹⁰⁹ were mixed with 100-fold excess of KO7 helper phage and buffer A (1 ml). This mixture
0.5 ml was divided into two, one of which was applied to ECD coated wells and the other was applied to BSA coated wells. The plates were incubated for 1-3 hours at room temperature with shaking, and then washed three times for 30 minutes each with buffer A (1 ml each). Elution of phage from the plate was performed at room temperature in one of two ways: (1) 0.025 mg /
ml of purified Mu4D5 antibody (murine) for the first overnight, followed by acid elution buffer [0.2 M glycine (pH 2.
1), 0.5% BSA and 0.05% Tween 20] (0.5 ml) for 30 minutes, or (2) incubation with acid elution buffer alone. The eluate was then neutralized with 1 M Tris base and TEN (0.5 ml each) was added. These samples were then grown, titrated and stored at 4 ° C.

Adding a portion of the growth eluted phage of a phage in 2YT broth (0.4 ml), and mixed with mid-log Tokio E. coli strain SR101 about 10 ^8.
The phage is allowed to attach to the cells for 20 minutes at room temperature, then
Added to 5 ml of 2YT broth containing g / ml carbenicillin and 5 μg / ml tetracycline. The cells were grown for 4-8 hours at 37 ° C. until reaching mid-log phase.
Measuring the OD _600, the cells were co-infected with KO7 helper phage for generating phage. After obtaining the phage particles, they were titrated to determine the number of colony forming units (cfu). This was done by taking a portion of a serial dilution of a phage stock, infecting mid-log phase SR101, and plating on LB plates containing 50 vg / ml carbenicillin.

Measurement of RIA Affinity Affinity of Fab phage and h4D5 Fab fragment for ECD antigen was determined by competitive receptor binding RIA [Burt, DR, Recep
tor Binding in Drug Research, O'Brien, RA (ed.), p
p.3-29, Dekker, New York (1986)]. This ECD antigen was purified by the continuous chloramine-T method [De Larco, J. et al.
E. et al., J. Cell. Physiol. 109: 143-152 (1981)].
Labeled with ^125I . This gives 0.47 per mole of receptor
A radioactive tracer having a specific activity of 14 μCi / μg into which iodine was introduced was obtained. 0.5 ng (by ELISA) of Fab or Fab phage, 50 nCi of ¹²⁵ I-ECD tracer, and a range of unlabeled ECD (6.4 ng to 3277 ng)
A series of 0.2 ml solutions containing was prepared and incubated overnight at room temperature. Labeled ECD-Fab or ECD-Fab phage complex was ligated to anti-human IgG (Fitzgerald 40-GH
23) was separated from unbound labeled antigen by forming an assembly complex induced by the addition of 6% PEG8000. Centrifuge this complex (15,000xg
Pellet for 20 minutes) and the amount of labeled ECD (cp
m) was measured with a gamma counter. Dissociation constant (Kd)
Describes a Scatchard analysis [Scatchard, G., Ann. NYAcad.
Sci. 51: 660-672 (1949)]
[Munson, P. and Rothbard, D., Anal.
Biochem. 107: 220-239 (1980)]. This Kd value is shown in FIG.

Using test Rodogen method competitive cell binding, 40-5 murine 4D5 antibody by ¹²⁵ I
Labeled with a specific activity of 0 μCi / μg. A solution containing a fixed amount of labeled antibody and an increasing amount of unlabeled mutant Fab was prepared and added to almost the entire SK-BR-3 cell culture in a 96-well microtiter dish (final concentration of labeled antibody). Was 0.1 mM). 4
After overnight incubation at ℃, the supernatant was removed, the cells were washed, and the radioactivity bound to the cells was measured with a gamma counter. Improved version of the program LIGAND [Munso
n, P. and Rothbard, D. (supra)] were used to determine the Kd values by analyzing the data.

Deposit of plasmid pDH188 (ATCC No. 68563) has been made under the provisions and regulations of the Budapest Treaty on the International Recognition of Microbial Deposits for Patent Procedure (Budapest Treaty). This ensures the maintenance of viable microorganisms for 30 years from the date of deposit. This microorganism is available from the ATCC under the terms of the Budapest Treaty and with the consent of Genentech, Inc. and the ATCC, and has issued the relevant US patents or any US or foreign The earlier of the publication of the patent application guarantees that the progeny of this culture will be made permanently and unrestrictedly available to anyone, and
C The descendants will be available to the Commissioner of the United States Patent and Trademark Office, which has been entitled in accordance with §122, and to those determined by rules pursuant thereto, including 37 CFR §1.14 which specifically references 886 OG 638 That is guaranteed.

The assignee of the present application will notify and promptly replace a viable sample of the deposited microorganism when the deposited microorganism is dead, lost, or destroyed when cultured under appropriate conditions. I agree. The availability of a deposited culture should not be construed as a license to practice the present invention in violation of the rights granted under the authority of any government in accordance with national patent laws.

The foregoing specification is considered to be sufficient to enable one skilled in the art to practice the invention. The deposited embodiment is intended as an independent illustration of a particular embodiment of the present invention, and the present invention is not limited by the deposited microorganisms, as any functionally equivalent culture is within the scope of the present invention. Is not limited. The deposit of materials herein has acknowledged that the statements contained herein are unsuitable for enabling the practice of any aspect of the present invention, including the best mode. It is not intended to limit the scope of the claims to the particular examples. Indeed, various modifications of the invention in addition to those described and shown herein will become apparent to those skilled in the art from the foregoing description and fall within the scope of the appended claims. It is.

Although the invention has necessarily been described in connection with a preferred embodiment, those skilled in the art after reading the above description will appreciate that various modifications may be made to the subject matter described herein without departing from the spirit and scope of the invention. Changes, equivalent substitutions, and mutations could be made. That is, the present invention can be implemented by methods other than those specifically described in the present specification. Therefore,
The protection granted by this patent is limited only by the appended claims and equivalents thereof.

Example XII Selection of hGH Mutants from Helix 1 and Helix 4 Hormone-Phage Mutant Combinations Preparation of Additive Mutants of hGH Principle of Additivity [JA Wells, Biochemistry 29: 8509 (19
90)] shows that even when mutations in different parts of the protein do not interact, combining them results in an additive change in the free energy of binding to another molecule. (Changes are additive in terms of ΔΔG _binding or multiplicative in terms of Kd = exp [−ΔG / RT]). That is,
Mutations that give a two-fold increase in binding affinity are 3
When combined with a second mutation that causes a fold increase, one would expect to give a double mutation with a 6-fold increased affinity of the starting mutant.

hGH-Multiple mutations obtained by phage selection
To test whether bp (hGH-binding protein; extracellular domain of the hGH receptor) has a cumulatively good effect on binding, hG from the helix 1 library was tested.
The three most tightly bound variants of H (Example IX: F10A / M
14W / H18D / H21N, F10H / M14G / H18N / H21N, and F10F / M14
S / H18F / H21L), the mutation found in helix 4
b The three most tightly binding variants found in the library (Example X: R167N / D171S / T175 / I179T, R1
67E / D171S / T175 / I179, and R167N / D171N / T175 / I179
Combined with the mutation found in T).

HGH-phagemid double-stranded DNA (dsDNA) was isolated from each of the 1 helix mutants and digested with the restriction enzymes EcoRI and
Digested with BstXI. The larger fragment from each helix 4b mutant was then isolated and ligated with the smaller fragment from each helix 1 mutant and expressed as
A novel two-helix mutant shown in XIII was obtained. further,
All of these variants contained the E174S / F176Y mutation obtained in the previous hGH-phage binding selection (see Example X for details).

Construction of a Selective Combination Library of hGH The principle of additivity appears to apply to a large number of mutation combinations, but some combinations (eg, E17 with F176Y)
4S) is clearly non-additive (see Examples VIII and X). For example, in order to reliably identify the most tightly binding variant having four mutations in helix 1 and four mutations in helix 4, ideally all eight residues should be abruptly changed at once. Mutations would then select the pool of mutants that would most tightly bind overall. However, such a pool would consist of 1.1 × ^{10 12} DNA sequences (using NNS codon degeneracy) encoding 2.6 × ^{10 10} different polypeptides. Obtaining a random phagemid library large enough to ensure the display of all mutants (probably 10 ¹³ transformants) is not practical with current transformation methods.

We first address this challenge by using sequential rounds of mutagenesis to take the most tightly binding mutations from one library and then mutate the other residues to further improve binding. Worked (Example X). In a second method, we combine the best mutations from two independently selected libraries using the principle of additivity to create multiple mutants with improved binding (Above). Here we create 10, 14, 18, 21, 1 in hGH by creating a combinatorial library of random or partially random mutants.
Possible combinations of mutations at positions 67, 171, 175, and 179 were further explored. Pooled phagemids from the helix 1 library (independently 0, 2 or 4
Cycle sorting; Example IX) and pools from the helix 4b library (independently 0, 2 or 4 cycle sorting; Example X) were used to generate three different combinatorial libraries of hGH-phagemid, and this combination Mutant pools were screened for hGHbp binding. Since a certain amount of sequence diversity exists in each of these pools, the resulting combinatorial library can examine more sequence combinations than can be made by hand (e.g.
XIII).

Pool of one helix library (0, 2 or 4
2 rounds of hGH-phagemid from each
Isolate single-stranded DNA (dsDNA) and use restriction enzymes AccI and BstX
Digested with I. The larger fragment was then isolated from each helix 1 mutant pool and ligated with the smaller fragment from each helix 4b mutant pool to form a three combinatorial library pH0707A (Example IX).
And the unselected helix 1 and helix 4b pools described in X and X), pH0707B (twice selected helix 1 pool and twice selected helix 4b pool), and pH07
07C (helix 1 pool selected 4 times and helix 4b pool selected 4 times) were obtained. Plus, less DNA
Perform the same ligation using pH0707D, pH0707E and pH0707E.
No. 0707F (corresponding to 0, 2, and 4 rounds of starting library, respectively). In addition, all of these mutant pools contained the mutation E174S / F176Y obtained from the previous hGH-phage binding selection (see Example X for details).

Selection of combinatorial library of hGH-phage variants The ligation products pH0707A-F were processed and described (Example V
Electrolysis was performed on XL1-Blue cells as in III). Based on colony forming units (CFU), the number of transformants obtained from each pool was as follows: from pH0707A
2.4x10 ⁶ , pH0707B to 1.8x10 ⁶ , pH0707C to 1.6x10 ⁶ , p
8x10 ^5, 3x10 from pH0707E ⁵ from H0707D, and 4x10 from pH0707F ^5. hGH-phagemid particles were prepared and prepared in Example VIII
Selected for hGHbp-binding over 2-7 cycles as described.

Rapid Selection of hGH-Phagemid Libraries In addition to selecting phagemid libraries for tightly binding protein variants as measured by equilibrium binding affinity, the on-rate of binding to receptors or other molecules (k It is also important to screen for mutants that have altered either _on ) or off-rate (k _off ).
From thermodynamics, these rates are determined by the equilibrium dissociation constant Kd = ( _koff /
k _on ). Although we are part of the variants of a particular protein have Kd similar for binding, are assumed to have very different k _on and k _off. Conversely, the change in Kd for one mutant and another is
It may be due to an effect _{on kon,} an effect _{on koff} , or both. The pharmacological properties of the protein may depend _{on the} binding affinity or k _on or k _off (depending on the detailed mechanism of action). Here we
I tried to examine the effect of changes in k _on and identify hGH variants with speed - higher one. We assume that by increasing the binding time, and by increasing the washing time and / or enrichment with the cognate ligand (hGH), the selection can be selectively weighted towards k _off . .

Time course analysis of wild-type hGH-phagemid binding to immobilized hGHbp shows total hGH that can be eluted in the final pH2 wash.
Less than 10% of the phagemid particles (see Example VIII for overall binding and elution protocols)
Binds after a 1-minute incubation, but more than 90%
It appears to bind after a minute incubation.

For "rapid binding selection", phagemid particles from the pH0707B pool (independently selected twice for helices 1 and 4) were incubated with immobilized hGHbp for only 1 min, then with binding buffer (1 ml). Washed 6 times [hGH
The washing step was eliminated and the remaining hGH-phagemid particles were p
Eluted with H2 (0.2 M glycine in binding buffer) wash]. Enrichment of hGH-phagemid particles relative to non-displayed particles is due to the use of short-term binding and mutations in which hGH-phagemid binding selection binds tightly from the randomized pool, even when no cognate ligand (hGH) challenge is used. Showed to sort the body.

Assay for hGH mutants.The binding constants of some of these hGH mutants to hGHbp were determined by expressing the free hormone mutant in non-suppressor E. coli strain 16C9 or 34B8, purifying the protein,
And labeling w from hGHbp in the radioimmunoprecipitation assay
It was determined by assaying by competitive displacement of t-hGH (see Example VIII). In the following Table XIII-A, all variants in addition to phenylalanine ₁₇₆ substituted by glutamate ₁₇₄ and tyrosine ₁₇₆ substituted (E174S and F176Y) by serine _174, the hGH amino acids 10,14,18,21, It has the additional substitutions shown in the table at positions 167, 171, 175 and 179.

In Table XIV below, hGH variants were selected from the combinatorial library by the phagemid binding selection method. table
All hGH variants in XIV contain two background mutations (E174S / F176Y). Library pH0
707A (part A), pH0707B and pH0707E (part B)
Alternatively, the hGH-phagemid pool from pH0707C (Part C) was selected in 2-7 cycles for binding to hGHbp. The numerical value P indicates the fractional incidence of each variant type in a set of clones sequenced from each pool.

In Table XV below, hGH variants were selected from the combinatorial library by the phagemid binding selection method. table
All hGH variants in XV contain two background mutations (E174S / F176Y). The number P is the fractional incidence of a particular variant in all clones sequenced after 4 cycles of rapid binding selection.

In Table XVI below, hGHbp (1-238) and Mab5
Or ¹²⁵ I-labeled hormone using either Mab263
Binding constants were determined by competitive displacement of H0650BD or labeled hGH. Mutant H0650BD appears to bind 30-fold more tightly than wild-type hGH.

Example XIII Selective enrichment of hGH-phage and non-substrate phage containing protease substrate sequences. Contains the hGH gene fused to
Using this plasmid, hGH-phage particles in which the hGH-gene III fusion product is monovalently displayed on the phage surface can be obtained (Example IV). This fusion protein
Contains the entire hGH protein fused to the carboxy-terminal domain of gene III by a variable linker sequence.

To test the possibility of using phage display to select a preferred substrate sequence for a protein hydrolase, a genetically engineered mutant of subtilisin BPN 'was used [Carter, P. et al., Proteins: Structure. , function an
d genetics 6: 240-248 (1989)]. This mutant (hereinafter referred to as A64SAL subtilisin) has the following mutations: Ser24Cys, His64Ala, Glu156Ser, Gly169Ala and T
Contains yr217Leu. This enzyme has the essential catalytic residue His6
The lack of 4 greatly restricts its substrate specificity and preferentially hydrolyzes certain histidine-containing substrates [Carter et al., Science 237: 394-399 (198
7)].

Construction of hGH-Substrate-Phage Vector The sequence of the linker region in pS0132 is To create a substrate sequence for A64SAL subtilisin. This gives the protein sequence: Was introduced into the linker region between hGH and the carboxy-terminal domain of gene III (the first Phe residue in the above sequence is Phe191 of hGH).

Array: Is known to be a good substrate for A64SAL subtilisin [Carter et al. (1989); supra]. The resulting plasmid was named pS0640.

hGH-Substrate-Phage Selective Enrichment Phagemid particles derived from pS0132 and pS0640 were made as described in Example I. In the first experiment, each phage pool (10 μl) was added to 20 mM
Oxirane beads (prepared as described in Example II: 30 μl) were separately mixed in a buffer (100 μl) containing Tris-HCl (pH 8.6) and 2.5 M NaCl. The coupling and washing steps were performed as described in Example VII. The beads were then resuspended in the same buffer (400 μl) with or without 50 nM A64SAL subtilisin. After incubation for 10 minutes, the supernatant was collected and the phage titer (cfu)
Was measured. Table XVII shows that about 10-fold more substrate-containing phagemid particles (pS0640) were eluted in the presence of the enzyme than in the absence of the enzyme, and that of the non-substrate phagemid (pS0132) in the presence or absence of the enzyme. It shows that it elutes more than a case. Increasing enzyme, phagemid or bead concentrations did not change this ratio.

Improved Selection Enrichment Method In an attempt to reduce non-specific elution of immobilized phagemid, tightly binding hGH variants were introduced into pS0132 and pS0640 in place of the wild-type hGH gene. HG used
The H mutant is the mutant described in Example XI (pH 0650 bp), and the mutant Phe10Ala, Met14Trp, His18Asp, His21As
n, Arg167Asn, Asp171Ser, Glu174Ser, Phe176Try and Ile179Thr. This results in two new phagemids: pDM0390 (containing tightly binding hGH and no substrate sequence) and pDM0411 (tightly binding hGH and the substrate sequence Ala-Ala-His-Tyr-Thr-Arg
-Containing Glu). Also, the binding wash and elution protocol was changed as follows: (i) Binding: COSTAR 12-well tissue culture plate was
l / well 2 μg / well in sodium carbonate buffer (pH 10.0)
Coated with ml hGHbp for 16 hours. This plate is then
2 ml with 1 ml / well blocking buffer [phosphate buffered saline (PBS) containing 0.1% w / v bovine serum albumin]
Incubate for 10 hours, 10 mM Tris-HCl (pH 7.5), 1 mM
Washed with assay buffer containing M EDTA and 100 mM NaCl. The phagemid was again prepared as described in Example I, the phage pool was diluted 1: 4 with the assay buffer described above,
0.5 ml of phage per well was incubated for 2 hours.

(Ii) Washing: The plate was thoroughly washed with PBS + 0.05% Tween 20, and incubated with the washing buffer (1 ml) for 30 minutes. This washing step was repeated three times.

(Iii) Elution: The plate was washed with 20 mM Tris-HCl (pH 8.
6) Incubate for 10 minutes in an elution buffer consisting of +100 mM NaCl, then elute the phage with the above buffer (0.5 ml) with or without 500 nM A64SAL subtilisin.

Table XVII shows that the proportion of specifically eluted substrate-phagemid particles was dramatically increased compared to the method described above for pS0640 and pS0132. This may be due to the fact that the tightly binding hGH mutant has a significantly lower off-rate for binding to the hGH binding protein compared to wild-type hGH.

Example XIV Identification of Preferred Substrates for A64SAL Subtilisin Using Selective Enrichment of a Substrate Sequence Library Attempted to identify good substrate sequences from a library of random substrate sequences using the selective enrichment method described in Example XIII .

Construction of Vectors for Insertion of Randomized Substrate Cassettes Vectors suitable for introduction of randomized substrate cassettes and subsequent expression of a substrate sequence library were designed. The starting point was the vector pS0643 described in Example VIII.
Oligonucleotides: Site-directed mutagenesis using Apa I (GGGCCC) and Sal I between hGH and gene III.
(GTCGAC) A restriction site was introduced. This new construct
It was named pDM0253 (the actual sequence of pDM0253 is And the underlined base substitutions are due to spurious errors in the mutant oligonucleotide). Also, pDM041
By exchanging the fragment from 1 (Example XIII), the tightly binding hGH variant described in the examples was introduced. Transfer the obtained library vector to pDM0454
It was named.

Preparation of Library Cassette Vector and Insertion of Mutation Cassette
Digest with pa I, then Sal I, then 13% PEG8000 + 10 mM
precipitated with GCL _2, washed twice with 70% ethanol, and resuspended. This effectively precipitates the vector, but leaves a small Apa I-Sal I plug in solution [Paithank
ar, KR and Prasad, KSN, Nucleic Acids Research
19: 1346]. This product was run on a 1% agarose gel, the Apa I-Sal I digested vector was cut out, and
Purified using the prep kit (Pharmacia) and resuspended for ligation with the mutation cassette.

The inserting cassette contained a sequence similar to the DNA sequence in the linker region of pS0640 and pDM0411, but had codons for histidine and tyrosine residues in the substrate sequence replaced by randomized codons. Example VI
At each of the randomized positions as described in II
We chose to replace the NNS (N = G / A / T / C; S = G / C). The oligonucleotide used in the mutation cassette was Met. This cassette also destroys the Sal I site,
Thus, digestion with Sal I can be used to reduce vector background. These oligonucleotides did not phosphorylate prior to insertion into the Apa-Sal cassette site. The reason for this is that we fear that subsequent oligomerization of the cassette subpopulation may lead to spurious results with multiple cassette inserts. After annealing and ligation, the reaction product was phenol: chloroform extracted, ethanol precipitated, and resuspended in water.
Initially, to reduce the background vector
Digestion with Sal I was not performed. As described in Example VIII, about 200 ng was electroporated into XL1-Blue cells to prepare a phagemid library.

Selection of susceptible substrates from the substrate library The selection method used is the same as that described for pDM0411 and pDM0390 in Example XIII. After each round of selection, eluted phage were transduced into fresh XL1-Blue cell cultures and expanded with a new phagemid library as described for hGH-phage in Example VIII. Propagated. The progress of the selection procedure was monitored by measuring the titer of eluted phage and sequencing individual clones after each round of selection.

Table A shows serial phage titers for elution in the presence and absence of enzyme after 1, 2 and 3 rounds of selection. The ratio of specifically eluted phage: non-specifically eluted phage (ie, phage eluted with enzyme: phage eluted without enzyme) dramatically increases from round 1 to round 3. It is clear that this suggests that the good substrate population is increasing with each selection round.

Sequencing of the ten isolates from the starting library
All of these were shown to consist of the wild-type pDM0464 sequence. This means that after digestion with Apa I
It is very close to the ends of the DNA fragments and thus results in inefficient digestion. Nevertheless, there are only 400 possible sequences in the library, so this population should still be noted.

Tables B1 and B2 show the sequences of the isolates obtained after round 2 and round 3 selection. After two rounds of selection, there is clearly a high occurrence of histidine residues. This is exactly what was expected (example
As described in XIII, A64SAL subtilisin requires a histidine residue in the substrate because it uses a substrate-assisted catalytic mechanism). After three rounds of selection, each of the 10 clones sequenced has histidine in the randomized cassette. However, it should be noted that two of these sequences are of pDM0411 and were not present in the starting library and are therefore impurities.

Sequence Table (1) General Information (i) Patent Applicants: Genentech, Inc. Gailard, Liza Jay Henner, Dennis Jay Bath, Steven Green, Ronald Rowman, Henry Be Wells, James A. Matthews, David Jay (ii) Title of the invention: Method for enriching mutant proteins having altered binding properties (iii) Number of sequences: 27 (iv) Contact: (A) Addressee: Genentech, Inc. (B) Point San Bruno Boulevard No. 460 (C) City: South San Francisco (D) State: California (E) Country: United States (F) ZIP: 94080 (v) Computer reading ceremony: (A) Media type : 5,25 inch, 360Kb floppy disk (B) Computer: IBM PC compatible ( ) Operating system: PC-DOS / MS
-DOS (D) Software: patin (Genentech) (vi) Data of this application: (A) Application number: To be determined (B) Filing date: To date (C) Classification: To be determined (vii) Priority claim application No .: (A) Application No .: 07/743614 (B) Application date: August 9, 1991 (vii) Priority claim application data: (A) Application No .: 07/715300 (B) Application date: 1991 June 14, 1991 (vii) Priority application data: (A) Application No .: 07/683400 (B) Application date: April 10, 1991 (vii) Priority application data: (A) Application Number: 07/621667 (B) Application date: December 3, 1990 (viii) Patent attorney / agent information: (A) Name: Benson, Robert H. (B) Registration number: 30,446 (C) Reference / Organization Number: 645P4 (ix) Telephone contact information: (A) Phone number: 415 / 266-1489 (B) Fax number: 415 / 952-9881 (C) Telex number: 910 / 371-716 8 (2) Information of SEQ ID NO: 1 (i) Sequence characteristics: (A) Length: 36 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi ) Sequence: SEQ ID NO: 1: (2) Information of SEQ ID NO: 2 (i) Sequence characteristics: (A) Length: 36 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 2: (2) Information of SEQ ID NO: 3 (i) Sequence characteristics: (A) Length: 33 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 3: (2) Information of SEQ ID NO: 4 (i) Sequence characteristics: (A) Length: 30 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 4: (2) Information of SEQ ID NO: 5 (i) Sequence characteristics: (A) Length: 24 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 5: (2) Information of SEQ ID NO: 6 (i) Sequence characteristics: (A) Length: 23 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 6: (2) Information of SEQ ID NO: 7 (i) Sequence characteristics: (A) Length: 21 bases (B) Type: nucleic acid (C) Number of strands: single strand (D) Topology: linear (xi) Sequence: SEQ ID NO: 7: (2) Information of SEQ ID NO: 8 (i) Sequence characteristics: (A) Length: 63 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 8: (2) Information of SEQ ID NO: 9 (i) Sequence characteristics: (A) Length: 24 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 9: (2) Information of SEQ ID NO: 10 (i) Sequence characteristics: (A) Length: 33 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 10: (2) Information of SEQ ID NO: 11 (i) Sequence characteristics: (A) Length: 36 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 11: (2) Information of SEQ ID NO: 12 (i) Sequence characteristics: (A) Length: 36 bases (B) Type: nucleic acid (C) Number of strands: single strand (D) Topology: linear (xi) Sequence: SEQ ID NO: 12: (2) Information of SEQ ID NO: 13 (i) Sequence characteristics: (A) Length: 12 amino acids (B) Type: amino acid (D) Topology: linear (xi) Sequence: SEQ ID NO: 13: (2) Information of SEQ ID NO: 14 (i) Sequence characteristics: (A) Length: 27 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 14: (2) Information of SEQ ID NO: 15 (i) Sequence characteristics: (A) Length: 38 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 15: (2) Information of SEQ ID NO: 16 (i) Sequence characteristics: (A) Length: 30 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 16: (2) Information of SEQ ID NO: 17 (i) Sequence characteristics: (A) Length: 66 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 17: (2) Information of SEQ ID NO: 18 (i) Sequence characteristics: (A) Length: 64 bases (B) Type: nucleic acid (C) Number of strands: single strand (D) Topology: linear (xi) Sequence: SEQ ID NO: 18: (2) Information of SEQ ID NO: 19 (i) Sequence characteristics: (A) Length: 25 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 19: (2) Information of SEQ ID NO: 20 (i) Sequence characteristics: (A) Length: 33 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 20: (2) Information of SEQ ID NO: 21 (i) Sequence characteristics: (A) Length: 66 bases (B) Type: nucleic acid (C) Number of strands: single strand (D) Topology: linear (xi) Sequence: SEQ ID NO: 21: (2) Information of SEQ ID NO: 22 (i) Sequence characteristics: (A) Length: 58 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 22: (2) Information of SEQ ID NO: 23 (i) Sequence characteristics: (A) Length: 65 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 23: (2) Information of SEQ ID NO: 24 (i) Sequence characteristics: (A) Length: 64 bases (B) Type: nucleic acid (C) Number of chains: single-stranded (D) Topology: linear (xi) Sequence: SEQ ID NO: 24: (2) Information of SEQ ID NO: 25 (i) Sequence characteristics: (A) Length: 2178 bases (B) Type: nucleic acid (C) Number of strands: single strand (D) Topology: linear (xi) Sequence: SEQ ID NO: 25: (2) Information of SEQ ID NO: 26 (i) Sequence characteristics: (A) Length: 237 amino acids (B) Type: amino acid (D) Topology: linear (xi) Sequence: SEQ ID NO: 26: (2) Information of SEQ ID NO: 27 (i) Sequence characteristics: (A) Length: 461 amino acids (B) Type: amino acid (D) Topology: linear (xi) Sequence: SEQ ID NO: 27:

────────────────────────────────────────────────── ─── Continued on the front page (31) Priority claim number 715,300 (32) Priority date June 14, 1991 (1991. 6.14) (33) Priority claim country United States (US) (72) Inventor Bath, Steven, California 94061, Redwood City, Sheila Street, 1355 (72) Inventor Green, Ronald, USA North Carolina 27707, Durham, Norwood Ave. 1014 (72) Inventor Lohmann, Henry B. United States 94547, Hercules, Falcon Way 205 (72) Inventor Wells, James A. California, USA 94010, Burlingame, Columbus Avenue 1341 (72) Inventor Matthews, David J. USA 94122, California, San Francisco, Noriega Street No. 1527 (56) References Special Table 5-508076 (JP, A) International Publication 90/2809 (WO, A1) PROTEINS: Structur e, Function, and Genetics, 8 (1990) p. (58) Fields investigated (Int. Cl. ⁷ , DB name) C12N 15/09 C12N 7/00 BIOSIS (DIALOG) MEDLINE (STN)

Claims (10)

(57) [Claims] 1. A phagemid expression vector containing a transcriptional regulatory element operably linked to a gene fusion encoding a fusion protein, wherein the gene fusion encodes a first polypeptide. The gene and a second gene encoding at least a portion of the phage coat protein, and (a) the vector does not include the gene encoding the mature coat protein; (b) the complete phage genome Or (c) transforming host cells with the vector does not produce phage particles in the absence of helper phage, and (d) encodes an mRNA repressible stop codon.
A phagemid expression vector wherein an A triplet codon is inserted between the fusion termini of the first and second genes, or is substituted for the amino acid-encoded triplet codon adjacent to the gene fusion junction. 2. The phagemid vector according to claim 1, wherein the stop codon is UAG, UAA or UGA. 3. The phagemid expression vector according to claim 1, wherein the phage coat protein is encoded by filamentous phage gene III. 4. The phagemid expression vector according to claim 3, wherein the mammalian protein is a human protein. 5. The protein is a growth hormone, human growth hormone (hGH), des-N-methionyl human growth hormone, bovine growth hormone, parathyroid hormone, thyroxine, insulin A chain, insulin B chain, proinsulin, relaxin A Chain, relaxin B chain, prorelaxin, follicle stimulating hormone (FSH), thyroid stimulating hormone (T
SH), leutinizing hormone (LH), glycoprotein hormone receptor, calcitonin, glucagon, factor VI
II, antibody, lung surfactant, urokinase, streptokinase, human tissue-type plasminogen activator (t
-PA), bombesin, factor IX, thrombin, hematopoietic growth factor, tumor necrosis factors α and β, enkephalinase, human serum albumin, Muller inhibitor, mouse gonadotropin-related peptide, β-lactamase, tissue factor protein, inhibin, activin , Vascular endothelial growth factor,
Integrin receptor, thrombopoietin, protein A
And D, rheumatoid factor, nerve growth factor (NGF-
β), platelet growth factor, transforming growth factor (TGF); TGF-α and TGF-β, insulin-like growth factors I and II, insulin-like growth factor binding protein, CD
-4, DNase, latency-related peptide, erythropoietin, osteoinductive factor, interferon α, β and γ, colony stimulating factor (CSF), M-CSF, GM-CSF and GC
SF, interleukin (IL), IL-1, IL-2, IL-
3. IL-4, superoxide dismutase; disintegration promoter, viral antigen, HIV envelope protein GP120
The phagemid according to any one of claims 1 to 4, wherein the phagemid is selected from the group consisting of GP140, atrial natriuretic peptides A, B and C, proteins having more than one subunit of the above proteins, immunoglobulins, and fragments. Expression vector. 6. A host cell containing the phagemid expression vector according to claim 1 and a helper phage having a gene encoding a coat protein. 7. A phagemid particle comprising at least a part of the phagemid expression vector according to any one of claims 1 to 5, wherein the amount or number of the phagemid particle displaying two or more copies of the fusion protein on the particle surface is 20 or more. % Of the phagemid particles. 8. A method for selecting a novel binding polypeptide, comprising: (a) constructing a family of phagemid expression vectors according to any one of claims 1 to 5 (where the phagemid expression vector is (B) transforming a suitable host cell with the phagemid expression vector; and (c) transforming the transformed host cell with a recombinant phagemid. Infecting a helper phage having a gene encoding the phage coat protein sufficient to produce particles; (d) a host suitable and suitable for forming recombinant phagemid particles containing at least a portion of the expression vector The transformed and infected host cells are cultured under conditions that allow transformation of the fusion protein (where the fusion protein 2
The amount or number of phagemid particles displaying more than copies,
(E) contacting the recombinant phagemid particles with a target molecule to allow at least a portion of the phagemid particles to bind to the target molecule; and (f) not allowing the phagemid particles to bind to the target molecule. Separating from the particles of binding. 9. A group of phagemid particles produced by the method according to claim 8. 10. The phagemid expression vector of claim 8, wherein the first gene encodes a polypeptide linked to a linker amino acid sequence, and the phagemid expression vector comprises a mutant first gene encoding a mutant linker amino acid sequence. The method comprising the steps of: (f) contacting the bound phagemid particles with a protease capable of hydrolyzing at least a portion of the bound amino acid sequence of the bound phagemid particles; and (g) hydrolyzing phagemid particles. Isolating the particles; and, optionally, (h) further contacting a suitable host cell with the hydrolyzed phagemid particles, and steps (d)-(g).
9. The method of claim 8, wherein: is repeated.