EP0717848A1 - Biotin-analog conjugated antibodies for positive cell selection and release - Google Patents

Abstract

The invention provides a stable conjugate of a target cell binding protein, preferably an antibody, covalently linked to a biotin analog moiety for use in positive cell selection and release. Methods for positive cell selection and release using the stable conjugate are described.

Description

BIOTIN-ANALOG CONJUGATED ANTIBODIES FOR POSITIVE CELL SELECTION AND RELEASE

This application is a continuation-in-part of U.S. Serial No. 08/118,068, filed September 8, 1993, the contents of which are hereby incorporated by reference into the present disclosure.

Technical Field

The invention relates to biotin-analog/antibody conjugates and methods for the specific capture of target, cells from a heterogeneous cell suspension, followed by separation and release of the target cells. The general field is also known as positive cell selection.

Background

Several cell selection techniques have been considered as a part of adjunctive treatments for cancer, particularly when target cells are separated from peripheral blood, cord blood, or bone marrow. Cell selection techniques are divided into two broad categories, negative cell selection and positive cell selection.

Negative cell selection involves the selection and removal of potentially harmful cells from a patient's or a donor's blood or bone marrow. For instance, a treatment for metastatic cancer may involve removal of a sample of the patient's bone marrow prior to ablative chemotherapy or radiation, with the intent to replace the patient's bone marrow cells after the ablative therapy in order to replenish he atopoietic cells. To minimize the risk of returning metastatic tumor cells to the patient, negative cell selection or purging is applied to the patient's bone marrow sample prior to reinfusion. This type of negative cell selection is based on anti-tumor antibodies, linked to a solid phase such as magnetic beads, which specifically bind the tumor cells (Hardwick, A., et al., J Hematotherapy 1:379-386, 1992). The tumor cells bound to the solid phase are then removed from the bone marrow sample and discarded. In contrast to negative cell selection, the aim of positive cell selection is to bind desired cells and recover them in

viable condition. One of the greatest problems in positive cell selection is the retention of viability of the desired cells while effecting their release from the solid phase separation material. To retain cellular function, no organic solvent may be used, the pH must be maintained at around 7.0 - 7.4, and the temperature cannot be raised much higher than 37°C. Thus, conventional methods such as extremes of pH and temperature cannot be used for positive cell selection. Certain cell types may tolerate low levels of reducing agents such as dithiothreitol and/or chelating agents such as EDTA, while other target cells may not remain viable even under very mild reducing or chelating conditions.

The strong affinity of avidin for biotin has been employed in both negative and positive cell selection. In avidin/biotin based techniques, typically an antibody which is specific for the target cell is biotinylated according to one of several standard methods (Avidin-Biotin Chemistry; A Handbook. Eds. Savage, MD, et al.. Pierce Chemical Co, 1992). For negative selection, the target cell is bound by the biotinylated antibody, which in turn is bound to an avidin-coated solid phase, usually in column form. The non-bound cells are then recovered, and the negatively selected cells bound to avidin are discarded.

For positive cell selection, however, the very strong affinity of avidin for biotin is disadvantageous since the target cells are firmly held within the cell/antibody- biotin/avidin complex. Although the avidin/biotin interaction is non-covalent, the interaction is so strong (K_a=10¹⁵M^_1) that it resists wide extremes of pH, temperature, organic solvents, reducing agents, and most forms of enzymatic proteolysis. Conditions which are harsh enough to disrupt the biotin/avidin interaction also damage or kill the target cells.

Since the avidin/biotin interaction is so strong, the disruption of other bonds was proposed for the release of desired target antigens. Certain biotinylating agents have chemically cleavable covalent bonds within their spacer arms or form cleavable covalent bonds with target proteins (Sigler, G.F. US Patent Nos: 4,798,795 and 4,709,037; ilchek, M. , et al, German Pat. App. DE 3629194.A; Avidin- Biotin Chemistry: A Handbook,supra, p.41). The bonds are cleaved under reducing conditions employing dithiothreitol, mercaptoethanol, or sodium borohydride, but these conditions are generally too damaging to cells to be considered for selection of cells which must remain functional.

Attempts have also been made to substitute biotin with chemical moieties that bind to avidin with lower affinities (Basch et al., J. Immunoloσical Meth. 56:269-280 (1983)).

Several techniques for positive cell selection rely on mechanical means for disruption of antibody/epitope interactions for release of selected cells. Tissue culture flasks may be coated with a primary antibody which binds the target cells; after the unbound cells are washed away, the target cells are released by striking the sides of the flask (Lebkowski, JS, et al.. Transplantation 53:1101-1019, 1992). Another method for positive cell selection employs a "sandwich" technique which involves avidin bound to a biotinylated secondary antibody which binds a primary antibody, which in turn binds the target cell to form a complex. After separation of the complex from the heterogeneous cell suspension, the target cell is removed from the avidin by agitation to disrupt the interaction between the secondary and primary antibodies (Berenson, R.J., et al., US Patent Nos: 5,215,927 and 5,225,353). Mechanical release is disadvantageous for the obvious reason that cells may sustain damage during the release process, and it has been reported that low numbers of viable cells are recovered after mechanical release (Egeland, T., et al., Scand J Immunol 27: 439-444, 1988).

Another method for cell release involves proteolysis by enzymes such as papain and chymopapain. The target cells may be bound to magnetic beads via a primary antibody which is in turn bound to magnetic beads. After the cell/antibody/bead complex is removed from the heterogeneous cell suspension, the cells are released from the beads by proteolysis of the cell surface antigen or the antibody, or both (Hardwick, A., et al., J Hematotherapy 1:379-386, 1992; Civin, CE, et al.. In Bone Marrow purging and Processing Progress in Clinical And Biological Research. Vol. 333, Eds. S. Gross, et al., Alan R. Liss, Inc, New York, pp 387-402; Civin, CI, EP 0 395 355 Al; Hardwick, A., et al.. In Advances in Bone Marrow Purging and Processing- Progress in Clinical and Biological Research. Vol. 377, Eds. orthington-White, DA, et al., Wiley-Liss, Inc., New York, pp 583-589). Proteolysis by papain or chymopapain is advantageous over mechanical disruption because these enzymes are not generally harmful to cells.

There remains a need for a positive cell selection method which produces a high yield of functional target cells, and which relies on relatively inexpensive, benign reagents in a physiologically compatible solution.

Summary of the Invention

The invention provides a stable conjugate of a target cell binding protein, preferably an antibody, covalently linked to a biotin analog moiety for use in positive cell selection and release. The invention also provides a method for forming a stable conjugate of an antibody and a biotin analog moiety comprising bringing an N-hydroxysuccinimide (NHS) ester of a biotin analog into reactive contact with an antibody in a medium with a pH of about 8 and at a temperature of about 2°-8°C.

The invention also provides methods for the positive selection of target cells from a heterogeneous cell suspension.

One method involves the formation within the cell suspension of a complex formed by the target cell bound to a primary antibody linked to a biotin analog moiety, a secondary anti-biotin antibody bound to the biotin analog moiety and linked to a cell separation means. The complex is separated from the heterogeneous cell suspension, then incubated with a competitive compound, preferably authentic biotin, for which the secondary antibody has a higher affinity than it has for the biotin analog. The competitive compound displaces the biotin analog moiety from the binding site of the secondary antibody, thus releasing the target cells from the complex. The cell separation means may then be separated from the target cells.

Another method involves the formation within the heterogeneous cell suspension of a complex formed by the target cell bound to a primary antibody linked to a biotin analog moiety and a form of avidin, which is bound to the biotin analog moiety. The complex is separated from the heterogeneous cell suspension and then incubated with a competitive compound, preferably authentic biotin, for which avidin has a higher affinity than it has for the biotin analog. The competitive compound displaces the biotin analog moiety from the avidin binding site, thus releasing the target cells from the complex. The avidin may then be separated from the target cells.

The invention also provides compositions of matter which are intermediate complexes in the above methods. One intermediate complex comprises a cell separation means bound to an anti-biotin antibody, which in turn is bound to a biotin analog moiety covalently linked to a target cell binding protein. Another intermediate complex comprises avidin bound to a biotin analog moiety covalently linked to a target cell binding protein.

Brief Description of the Figure

Figure 1 depicts a preferred method of the invention whereby the target cell is bound to an antibody/biotin analog conjugate, which in turn is bound by an anti-biotin antibody linked to a magnetic bead. Release of the target cell is effected by elution with authentic biotin which displaces the biotin analog from the biotin binding site.

Figure 2 shows the mass spectrometry analysis of an antibody/biotin analog conjugate (2A) compared to the unconjugated antibody (2B).

Detailed Description of the Invention

This invention provides a stable conjugate of a target cell binding protein covalently linked to a biotin analog moiety, and methods using the conjugate for the positive selection and recovery of target cells from a heterogeneous cell suspension.

Herein, the term "target cell binding protein" refers to any protein which specifically binds to a cell surface marker on the target cell. Herein, the term "cell surface marker" is defined as any moiety present specifically on the target cell; such markers may include proteins, lipids, and carbohydrates. Examples of target cell binding proteins include lectins and antibodies, growth factors, hormones, extracellular matrix proteins and functional fragements thereof. Lectins typically bind to specific carbohydrate moieties on cells, while antibodies may be obtained which bind to specific protein epitopes, or to lipids or carbohydrates. The target cell binding protein is reacted with a biotin analog reagent under conditions which form a stable conjugate of the protein/biotin analog moiety; the conjugate retains the specific cell surface binding activity of the protein.

A preferred target cell binding protein is a primary antibody. Herein, the term "primary antibody" refers to an antibody which specifically binds to a cell surface antigen on the target cell. Herein, the term "secondary antibody" refers to an antibody which binds to the primary antibody, and the term "tertiary antibody" refers to an antibody which binds to the secondary antibody.

Figure 1 depicts a preferred embodiment of one method of the invention. For illustration, this method will first be described in a system for the positive selection of hematopoietic stem/progenitor cells bearing the CD34 cell surface antigen (Civin, CI, et al.. In Bone Marrow Purging and Processing - Progress in Clinical and Biological Research. Eds. S. Gross, et al., Alan R. Liss, Inc., New York, Vol. 333, pp 387-402). A conjugate is formed from a primary anti-CD34 antibody covalently linked to a biotin analog (1° Ab/BA) . A heterogeneous suspension of blood cells is incubated with the conjugate, thus forming a first complex comprising the target CD34+ cell bound to the conjugate. A second complex is formed by linking an anti- biotin antibody to a paramagnetic bead. The anti-biotin antibody of the second complex binds the biotin analog moiety of the first complex to form a third complex comprising the target cell, the 1° antibody/BA conjugate, the anti-biotin antibody, and the paramagnetic bead. The third complex is separated from the heterogeneous cell suspension by means of a magnet. The third complex is then incubated with authentic biotin, which competes for the binding site on the anti-biotin antibody. Since the antibiotin antibody has a higher affinity for authentic biotin than for the biotin analog, authentic biotin displaces the first complex and thereby releases the target cell from the bead.

A second method of the invention involves the use of avidin in place of the second complex in Figure 1. Thus, the first complex is formed as in Figure 1, but it is bound to avidin by means of the affinity of avidin for the biotin analog moiety, thus forming a second complex comprising the target cell, the l°Ab/BA, and avidin. After separation from the heterogeneous cell suspension, the second complex is incubated with authentic biotin, for which avidin has a higher affinity than it has for the biotin analog. The target cell is thus released from the avidin. Herein, the term "avidin" refers to authentic avidin or streptavidin. Preferably, the avidin is insolubilized by retention on a solid phase such as a column or paramagnetic beads, thus making the avidin separable from the target cells.

The successful practice of this invention requires a stable conjugate of a binding protein covalently linked to a biotin analog. Herein, the term "stable conjugate" refers to a compound which does not undergo substantial hydrolysis of the covalent bond linking the protein to the biotin analog moiety under normal laboratory and experimental conditions. Normal laboratory and experimental conditions are herein defined as temperatures between about 2°-37°C, a pH range of about 7 - 8.5, and exposure to usual laboratory room light for about 4 - 16 hours. Moreover, the stable conjugate of the invention remains functional while stored at 4°C for at least 16 hours, preferably for 2 weeks or much longer. The stable conjugate also remains functional when stored frozen at -20°C or lower for at least 2 weeks and preferably much longer.

In the present invention, the conjugate is preferably formed by incubating the binding protein with an N- hydroxysuccinimide (NHS) ester of a biotin analog in a medium at a pH of about 8, and at a temperature of about 2°-8°C, most preferably 4°C. After a reaction time of about 4 to about 16 hours, the conjugate may be stored at 4°C for longer than 16 hours without substantial loss of function. The conjugate is sufficiently stable to be used for positive cell selection under conditions which are practical for laboratory and clinical use and which are conducive to cell viability, i.e. normal light, pH, and temperatures as stated above. The stable conjugate of the invention is particularly suitable for clinical applications since it enables the practice of the invention method at room temperature and under normal lighting.

According to published methods for the covalent linkage of authentic biotin to antibodies, the molar ratio of biotin to protein should be limited to 2-3 moles biotin:mole antibody for retention of the antibody binding function (Avidin-Biotin Chemistry: A Handbook, supra, pp.86-87). Surprisingly, it was found that using the methods of the invention, a biotin analog could be incubated with an antibody at molar ratios of up to 30:1 without loss of the antibody's binding function. This high molar ratio overcomes the problem associated with low affinity binding by increasing the valency of the interaction. Therefore, the sum of the many low affinity interactions allows for efficient recovery of the target cells. However, these low affinity interactions can still be disrupted without using harsh conditions which reduce the viability of the recovered cells. Given the experimental examples herein, it will be apparent to one of skill in the present art that molar ratios employed in formation of the conjugate may be varied according to the intended use of the conjugate as well as the particular biotin analog employed. For example, a biotin analog which exhibits a relatively high binding affinity for the 2° antibody will require a smaller number of conjugated molecules to achieve the same efficiency compared to a lower affinity biotin analog. For instance, the ultimate yield of target cells may be optimized and made more cost-effective by increasing the molar ratio of biotin analog:antibody while decreasing the number of beads used.

Herein, the term "biotin analog" refers to a compound which shares sufficient physicochemical and structural homology with authentic biotin to be bound by an anti-biotin antibody or by avidin. Examples of biotin analogs include desthiobiotin and 2 '-iminobiotin (Biochem J 101:774, 1966), biotin sulfone (Methods in Enzymology, Vol XVIII, Vitamins and Coenzymes. Part A, 1970, p. 423), bisnorbiotin, tetranorbiotin, and oxybiotin. The binding affinities of several biotin analogs for avidin are known, but the binding affinities of anti-biotin antibodies for biotin analogs are expected to vary with each individual antibody. In general, the binding affinity of avidin for biotin analogs is lower than avidin's affinity for authentic biotin. For instance, the binding affinity of avidin for desthiobiotin is 5 X 10¹³M^_1, which is substantially lower than the affinity of avidin for authentic biotin (K_a=10¹⁵M^{~ x}) . Similarly, the affinity of avidin for 2 '-iminobiotin is 3.5 X lO¹^^-1, again much lower than avidin's affinity for authentic biotin. This means that authentic biotin has a great competitive advantage in displacing the biotin analog moieties on the conjugates of the invention from their binding sites on avidin.

The binding of biotin analogs by antibodies raised against authentic biotin is similar, although not identical, to binding by avidin. Anti-biotin antibodies may be either polyclonal or monoclonal. Polyclonal antibodies are generally raised in an animal such as a goat, sheep, or rabbit by injecting the animal with biotin conjugated to an antigenic adjuvant such as albumin. (Authentic biotin alone is not very antigenic since it is a small molecular weight molecule, and it is a vitamin naturally present in the blood. ) The resulting polyclonal serum is a unique combination of antibodies to various epitopes on biotin in conjunction with the adjuvant. Surprisingly, it was found that various affinity purified polyclonal ^" sera from different sources raised against biotin would bind biotin analogs with affinities appropriate for use in the present invention. That is, the affinity for the biotin analog was strong enough to bind the target cells to the beads, yet weak enough to be successfully competed off by authentic biotin. Monoclonal antibodies are generally raised by injecting a mouse or rat with the antigen, then fusing individual spleen cells with immortal myeloma cell lines to form hybridomas. Each hybridoma synthesizes a monoclonal antibody which is specific for an individual epitope on the antigen. Again, it was surprising to find that a mouse anti-biotin monoclonal antibody had appropriate affinities for a biotin analog and authentic biotin to make it useful in the present invention. Given the present disclosure, one of skill in the art of raising and testing antibodies will be able to obtain and identify other antibodies which will function in the present invention.

The methods of the invention require that a competitive agent displace the biotin analog moiety from its binding site on avidin or on the anti-biotin antibody. It is understood that the competitive agent must have a competitive binding advantage for the binding site occupied by the biotin analog. The preferred competitive agent is authentic biotin since its binding to both avidin and to anti-biotin antibodies is the strongest of the agents tested. Moreover, authentic biotin is a simple and well- known vitamin which should be obtainable in an infusible grade suitable for use in human therapy. It is also contemplated that other competitive agents may be useful in the practice of the invention. For instance, if the biotin analog moiety on the conjugate has a low affinity for avidin or the anti-biotin antibody as compared with desthiobiotin, then desthiobiotin may be used as the competitive agent. Desthiobiotin is also considered to be a relatively benign reagent since it can be biotransformed to biotin by biotin synthetase (Gene 80:39-48^', 1989; BBRC 88:132, 1979). The most preferred combination is desthiobiotin as the moiety on the conjugate, subsequently competed off with authentic biotin. Covalent linkages between a protein and a biotin analog reagent may be formed via amine groups, sulfhydryl groups, or carbohydrate groups on the protein. In forming the conjugates of the instant invention, biotin analog reactions may be conducted according to known reactions with authentic biotin (Avidin-Biotin Chemistry: A Handbook supra, pp.25-87). Biotin analog reagents may be chosen from the group consisting of N-hydroxysuccinimide (NHS) esters, p_-nitrophenyl (BNP) esters, hydrazides, pyridyldithio-, iodoacetyl-, and maleimidopropionyl- biotin analogs. Also, photoactivatable biotin analogs may be employed. These biotin analog reagents may be purchased or may be synthesized according to established methods.

Once the protein/biotin analog conjugate has been formed, it may be tested for retention of antigen binding capacity by established methods such as ELISA, Western blot, dot blot, and target cell capture. The conjugate may also be tested by the same methods for binding by an anti-biotin antibody or by avidin.

This invention provides methods for forming antibody/biotin analog conjugates which retain their primary antibody binding function. However, under certain circumstances it may be desirable to practice a variation of this invention using a type of "sandwich" technique which allows the primary target cell binding antibody to remain unconjugated. In a variation of this type, a secondary antibody specific for the primary antibody is conjugated to the biotin analog. For instance, if the primary antibody is a mouse monoclonal anti-CD34, the secondary antibody might be a goat-anti-mouse-Fc. Thus, a first complex is formed of the target cell, the primary antibody, and the secondary antibody/biotin analog conjugate. ^"The biotin analog is then bound to a second complex as in Figure 1, or to avidin, and subsequently eluted with native biotin as in Figure 1. An advantage of this variation is that it enables the production of a universal "kit" containing all the elements of the invention method except for the primary antibody, which could be selected by the user of the kit according to its intended application.

Use of this invention provides a sufficiently pure population of functional target cells in a physiologically compatible solution suitable for clinical applications.

The following experimental examples are offered by way of illustration and are not intended to limit the scope of the invention.

EXAMPLE 1 Covalent coupling of biotin analogs to anti-CD34 antibody. Monoclonal anti-CD34 antibodies (mouse) designated "9079" were produced by standard methods from hybridomas obtained from Becton-Dickinson (8G12 cell line).

N-hydroxy-succinimide-DL-Desthiobiotin (NHS-DL- Desthiobiotin) (Sigma, Cat. #H-2134) was dissolved in PBS without calcium or magnesium, pH 8.0, at a concentration of 1.5 mg/ml. Optionally, the NHS-DL-Desthiobiotin was first dissolved in a small amount of DMSO or DMF prior to addition to the aqueous solution. The solution was cooled in an ice-water bath to about 4°C. The monoclonal antibody was dissolved in the same buffer, pH 8.0, about 4°C, at a concentration of 1.0 mg/ml. Desthiobiotin was added to the antibody solution at molar ratios of 3.25, 7.5, 15, and 30:1 (desthiobiotin:antibody) . The solution was kept at 4°C for 4-16 hours, and then separated by a gel filtration column (PD-10, Pharmacia, #17-0851-01) to remove small molecular weight compounds such as free NHS and unbound desthiobiotin. The separated sample was kept at 4°C and tested for binding activity by Western blot and dot blot.

A second type of biotin analog, 2-iminobiotin, was also coupled to monoclonal antibody 9079 using the same method as above.

EXAMPLE 2

Determination of the Number of Conjugated Biotin Analogs Per Antibody Using MALDI-TOF Mass Spectrometry. This example shows the use Matrix Assisted Laser Desorption Ionization (MALDI) - Time of Flight (TOF) Mass Spectrometer (MS) for determining the quantity of biotin analog moieties attached to a target cell binding protein.

The antibody used for this analysis was 9079 which is specific for the CD34 cell surface antigen and has been previously described. Conjugation of the biotin analog desthiobiotin was performed as described in Example 1. Specifically, desthiobiotin was added to antibody at a molar ratio of 30:1 desthiobiotin:antibody. Following conjugation, the antibody/biotin analog sample and an unlabeled antibody sample were prepared for mass spectrometry analysis. Specifically, 1 mg each of the 9079 antibody and the desthiobiotin conjugated antibody were adjusted to 2.5 ml by addition of ammonium phosphate buffer. Buffer exchange was then performed using a PD10 column (Pharmacia) and elution with ammonium phosphate buffer. The eluent from each sample (3.5 mis) was concentrated by two separate centrifugations using a Centricon column (Centricon 30; A icon) . The buffer exchange procedure was again repeated with the ammonium phosphate buffer. The samples were sent for analysis to the Midwest Center for Mass Spectrometry (University of Nebraska-Lincoln) .

Molecular weights of both the unlabeled and desthiobiotin-labeled antibody were determined by MALDI-TOF mass spectrometry. The results of the analysis are shown in Figure 2 and are the average of more than 25 independent determinations. As can be seen in the figure, the molecular weight for the control antibody was 151,181 and 151,170 for the M⁺ and M** ions, respectively. The desthiobiotin-labeled antibody showed molecular weights of 154,055 and 154,030 for the M⁺ and MC ions, respectively.

Using the above molecular weights and given that the molecular weight of desthiobiotin is 214.3 with the elimination of one water molecule (MW=18) during the coupling procedure, it was calculated that each antibody molecule contained an average of 14.6 molecules of coupled desthiobiotin (SD ± 1.6). The accuracy was determined to be ± 0.1% and possibly as low as 0.016%. Conversely, results obtained with an HABA assay only revealed an average of 11.3 molecules of desthiobiotin per antibody molecule.

These results indicate that a molar ratio of greater than 10:1 can be achieved using biotin analog moieties conjugated to target cell binding proteins. Moreover, the results also indicate that MALDI-TOF mass spectrometry can be efficiently and accurately utilized to determine the number of moieties conjugated to large proteins such as antibodies (MW=150,000) . EXAMPLE 3 Coupling of anti-biotin antibody to paramagnetic beads. Polyclonal goat anti-biotin antibodies were obtained from three suppliers: Vector Labs, San Mateo, CA, Cat. #SP3300; CalBiochem, San Diego, CA, Cat. #203199; Pierce Chemical Co., Rockford, MD, Cat. # 31852. A monoclonal mouse anti- biotin antibody in ascites form was obtained from American Qualex, La Mirada, CA, Cat.#M2270, and purified by ProSep A column (Bioprocessing, Durham, UK, Cat. # 8422).

Tosyl activated paramagnetic polystyrene Dynal beads were obtained from Baxter Fenwal Division, Round Lake, IL. The tosyl activated paramagnetic beads in suspension were washed quickly with 50 mM borate buffer pH 9.5 (coupling buffer) . Anti-biotin antibodies were dissolved or reconstituted in coupling buffer at 1 mg/ml (polyclonal) or 0.5 mg/ml (monoclonal). Anti-biotin antibody in coupling buffer was added to the tosyl activated paramagnetic beads at 50 μg/mg beads (polyclonal) or 25 μg/mg beads (monoclonal) and incubation was carried out overnight at room temperature. Anti-biotin antibody-coupled paramagnetic beads were washed and stored at 4°C.

EXAMPLE 4 KGla capture and release using antibody/desthiobiotin conjugate. KGla is a human cell line (ATCC #CCL 246.1) that expresses CD34 antigen on its cell membrane and is used as a model system for initial testing or optimizing conditions for positive selection of CD34+ cells.

Anti-CD34 specific antibody-desthiobiotin conjugate was incubated with KGla cells at the concentrations indicated in Tables 1 - 4 for 30 minutes at room temperature (about 22°C) .

Other KGla cells were incubated with anti-CD34 antibody HPCA-1 (Becton-Dickinson) without biotin analog moieties. and subsequently bound to sheep-anti-mouse (SAM) Dynal beads as a positive control for depletion.

As a negative control, KGla cells were incubated with non¬ specific IgGl isotype and subsequently mixed with SAM beads.

Other KGla cells were incubated with antibody-biotin conjugates for use in control experiments to compare the release of cells by elution/competition with native biotin. The antibody-biotin conjugates were prepared ^"by standard chemical methods (Biotinylation Reagents. IN: Avidin- Biotin Chemistry: A Handbook. Eds. Savage, MD, et al. Pierce Chemical Company, 1992, pp. 25-86).

The experiments were carried out in low biotin DMEM containing 2% FBS. Generally, one million cells per tube were used. Cells were washed to remove excess (unbound) antibody conjugate. Cell-antibody-desthiobiotin complexes were then mixed with anti-biotin antibody coated paramagnetic beads at the indicated concentrations and incubation was carried out for 30 minutes at room temperature with continuous mixing. Bead-cell complexes were separated from unbound cells using a magnet. Solution containing unbound cells was saved for cell counting (depletion step). Bead-cell complexes were washed 3 times.

Cell-antibody-biotin complexes were treated in parallel to the above.

Biotin stock solutions were prepared by dissolving DL- biotin (Pierce Chemical Co., Rockford, IL, Cat. # 29129) in low biotin DMEM, 2% FBS. The captured cells were then released from the beads by incubating bead-cell complexes with biotin at 1 - 5 mM for 30 minutes at room temperature with continuous mixing. Biotin competed with the desthiobiotin-moiety for binding sites on the anti-biotin antibody, thus dislodging the cell/antibody-desthiobiotin complex from the bead/anti-biotin antibody complex.

The mixing was carried out by either shaking (Vortex Genie II) or by rocking (Nutator®) .

Paramagnetic beads were then separated from released cells using a magnet. Solution containing released cells was saved for cell counting (release step) . Cell counting was carried out using a microscope and a hemacytometer. Cell viability was determined by diluting the suspension in Trypan Blue stain.

The following calculations were carried out:

%Depletion = [1 - (# of cells in the unbound fraction/* of cells in the unbound fraction of IgGl*)] X 100 (*IgGl is an isotype control to determine non-specific binding of cells to beads.)

%Release = (# of cells in the released fraction/starting # of cells used, usually 10⁶) X 100

%Yield and Yield were calculated the same as %Release.

Results: As shown in Table 1, comparable depletion (i.e. cell capture on beads) was obtained using either desthiobiotin-conjugated anti-CD34 antibody (9079- Desthiobiotin) or the control (HPCA-1) at an antibody concentration of 0.2 μg/10⁶ cells/tube. Release of cells complexed with 9079-Desthiobiotin by incubation with native biotin was as high as about 78% (yield) . Non-specific release, as shown by vortexing of HPCA-1 complexes, represented about 18% in this experiment. These results indicate that specific release of cells using biotin competition was about 60% or better. Table 3 shows the results of a control experiment using anti-CD34 antibody 9079 coupled to biotin (9079B). As expected, the % depletion (i.e. capture of cells on beads) is high; however, the % release of cells from beads is very low, comprising only about 1% over control (HPCA-1). This experiment demonstrates that a biotin-conjugated antibody may be very suitable for negative selection, in cases where the cells bound to beads are to be discarded. However, for positive selection, in cases where the selected cells are meant to be released from the beads and retrieved, a biotin-based release system is not successful against a biotin-conjugated antibody.

One Million KGla was used/tube. ^* method used for the release

21

TABLES 2 & 3

Effect of Desthiobiotin:Antibody Ratio in Labeling and Cell to Bead Ratio on Cell Capture and Release

1 million KGla cells were used/tube. Antibody was Incubated with KGla cells and tben beads were added to cells. Blotlnvlatlon of 9079 Is through tbe carbobydrate residues. EXAMPLE 5 "Sandwich" method; KGla capture and release using secondary antibody/2-Iminobiotin conjugate.

A conjugate of 2 '-iminobiotin covalently linked to a goat- anti-mouse (GAM) polyclonal IgG (secondary antibody) was prepared according to the methods of Example 1. The primary antibody (mouse monoclonal anti-CD34 9079) was bound to KGla cells, then bound by the secondary antibody GAM/2 '-iminobiotin conjugate to form a complex. Magnetic beads linked to tertiary antibody, goat-anti biotin, were then bound to the complex via the affinity of the tertiary antibody for 2 '-iminobiotin. Cells were then released from the beads by incubation with DL-biotin as described.

Results: As shown in Table 4, % capture (% depletion) using the antibody/2-Iminobiotin conjugate was as high as 96%, which was comparable to that achieved with the positive control (HPCA-1). The % yield (% release) using the antibody/2-iminobiotin conjugate was as high as 35%. It is expected that the percent yield could be as high as 80% or greater under optimal conditions.

TABLE 4

obiotin

at t e amount Indicate In the Uble. Fiv. million bead, were «_*d

EXAMPLE 6 KGla spiking of a cell population; specific depletion and release.

KGla cells were fluorochrome labeled as follows: 100 mg of Bisensimide H33342 Fluorochrome (Calbiochem Cat. #382065) was dissolved in 50 ml of nanopure water (2 mg/ml) . To each flask of KGla cells was added 100 μl of the Fluorochrome dye solution; the cells were incubated in the dye for 30 minutes at 37°C. The labeled cells were then mixed at a ratio of 1:50 with Daudi cells. Daudi cells are cells that do not express the CD34 antigen on the cell surface. Capture and release were performed as above. The KGla cells were counted under fluorescence microscopy in the unbound fraction and the released fraction. The Daudi cells in the released fraction were also counted to determine the purity of the released fraction.

Percent purity was calculated as follows:

% Purity = (# of KGla cells in the released fraction X 100)/ # of total cells (Daudi + KGla) in the released fraction

Results: As shown in Table 5, using anti-CD34 antibody/desthiobiotin conjugate (9079-Desthiobiotin) a high percentage of depletion and specific yield were achieved. Optimal results were obtained using 0.2 ug of antibody conjugate/10⁶ cells. Moreover, the purity of the released cell population was in the range of 88% KGla cells. TABLE 5

Capture and Release of KGla Spiked in Daudi Cells

KGla cells were spiked in Daude cells (2%). One million KGla ceils were used KGla Cell to bead ratio is 1:10. Vortex was used in the exp. instead of Nutator

EXAMPLE 7 Capture and release of CD34+ cells from human mobilized peripheral blood and bone marrow.

Peripheral blood or bone marrow samples were washed 3 times to remove platelets by re-suspending the samples in Dulbecco's Modified Eagles Medium containing 1% human serum albumin (Baxter Hyland Division) (DMEM-HSA) followed by centrifugation for 10 minutes at 1000 RPM at room temperature. To remove cells which non-specifically bound to beads, washed blood samples were incubated with uncoated paramagnetic beads for 30 minutes at 37°C with manual end over end rotation every 10 minutes. The uncoated beads were separated from the blood samples using a magnet. The beads were washed once with DMEM/HSA, beads were separated, and the wash solution was re-pooled with the separated sample (pre-depletion step) . Optionally, the pre-depletion step was replaced by incubation of cells, after the sensitization step, with a pooled human Ig fraction (Gammagard, Baxter Hyland Division, Duarte, CA) . The sample was then incubated with anti-CD34 antibody- desthiobiotin for 30 minutes at 4°C with end over end rotation (this is known as the sensitization step) . Cells were washed to remove unbound antibody conjugate. The sensitized cells were then incubated with anti-biotin antibody coated paramagnetic beads for 30 minutes at 4°C with end over end rotation. Beads were separated from the sample using a magnet, leaving behind the negative fraction. Bead/cell complexes were washed three times with DMEM/HSA and the three washes were pooled with the negative fraction.

The bead/cell complexes were then incubated with biotin (5 mM) for 15 minutes at room temperature using end-over-end rotation.

As a positive control, cells were incubated with monoclonal antibody 9069 (anti-CD34, no biotin or biotin analogue) followed by binding to sheep-anti-mouse coated Dynal paramagnetic beads. Following magnetic separation from the peripheral blood cell suspension, the control bead/cell complexes were incubated for 15 minutes at 37°C with chymopapain, which enzymatically digested the antibodies binding the cells to the beads, thus releasing the cells.

Beads were separated using a magnet and the solution was saved (positive fraction) . Beads were washed once with DMEM/HSA and the wash was pooled with the positive fraction.

Cells were labeled for FACScan analysis as follows: HPCA- 2-FITC (Becton Dickinson Cat. #348053) was added to the 9069 chymopapain control cells to identify CD34+ cells. A mixture of mouse IgGl-FITC and IgG2-PE (Becton Dickinson, Cat. #340041) was added to cell samples as an isotype control, used to estimate the extent of non-specific antibody binding to cells. LeukoGATE (Becton Dickinson Cat. #340040) was added to cell samples as a means for identifying the total leukocyte population (LeukoGATE is a mixture of anti-CD45-FITC which labels leukocytes and anti- CD14-PE, which labels monocytes). Goat-anti-mouse IgG-FITC (Becton Dickinson Cat. #349031) was used as a label to determine the percentage of leukocytes bearing mouse monoclonal antibodies 9079 and 9069 bound to CD34.

For each label, 20 μl (1 μg) of test or isotype control antibody preparation was added to polypropylene tubes. Nucleated cells (IO⁶) were added per tube for the starting samples or negative fractions, and at least 2.5 X IO⁴ cells were added for the positive fractions. After 30 minutes incubation at 4°C, cells were washed twice using 2 ml of 2%FBS/0.02%NaN₃/PBS (washing buffer), and then resuspended in 300 μl of fixative. FACScan analysis was carried out on the different fractions (washed sample, predepleted fraction, negative fraction, and positive fraction) to determine % CD34+ cells in each fraction.

The flow cytometer was calibrated according to the manufacturer's recommendations. At least 10,000 events were collected for each sample, except that for HPCA staining of the starting material, pre-depleted sample, and negative fractions, at least 20,000 events were collected.

The leukocyte gate was determined by recalling the FLl(FITC) vs FL2 data from the file that contained the starting material (mPBSC) stained with LeucoGATE, and drawing a box (region Rl) around all the events positive in FL1 (CD45-FITC). Using the same file, forward scatter (FSC) vs side scatter (SSC) data was recalled for all those events located within Rl. A box was drawn around all leukocytes.

The % of CD34+ cells was determined by recalling those FL1 vs SSC events that fall within R2 (leukocyte analysis gate) from the file that contained starting material stained with simultest control, and comparing to the same dot-plot of the cells stained with positive stain (i.e. HPCA-2-FITC or GAM-FITC) . A box was drawn (region R3) that contained all the events that were positive for the positive stain and low SSC (%CD34+) .

Percent purity was determined by FACScan analysis (% CD34+ cells in the positive fraction) . Other values were calculated as follows:

% Capture = [1 - (%CD34+ in negative fraction from FACScan X # of negative cells/% CD34+ in pre-D sample X # cells from pre-D used) ] X 100

Final percent yield from pre-depleted cells: (%CD34+ in positive fraction X # positive cells) / (%CD34+ in pre-D sample X # cells used from pre-D)

Final percent from total MNC:

(%CD34+ in positive fraction X # cells by Coulter counting in positive fraction) /[ (%CD34+ in pre-D sample X

# cells used from pre-D) /(% yield from pre-D)] (Coulter counting counts mononuclear cells only, since the red cells are lysed by the counting solution.)

Results: The yield was 30 - 40% and the purity of the CD34+ cells in the positive fraction was in the range of 40-80%. It is expected that a yield of greater than 80% and purity of greater than 80% will be obtained once conditions are optimized.

EXAMPLE 8 Positive cell selection via antibody-desthiobiotin binding to avidin.

Anti-CD34 antibody was covalently linked to desthiobiotin or 2 '-iminobiotin according to the methods of Examples 1, 2, and/or 7 to form a conjugate. KGla cells were incubated with the antibody/biotin analog conjugate.

The sensitized cells were then incubated with avidin in the following forms: albumin(HSA) /avidin beads, avidin/glass beads, and streptavidin (Dynal) beads.

The biotin-analog moiety on the anti-CD34+ antibody bound to avidin with sufficient affinity to attach the cells to the avidin solid phase. Percent binding (%depletion) ranged as high as about 90%.

The KGla cells were then separated from the avidin by means of incubation with authentic biotin, which out-competed desthiobiotin for the binding sites on avidin due to the higher binding affinity of biotin for avidin. The percent recovery of KGla cells ranged as high as about 40%.

It is expected that this method could be successfully applied to the positive selection and recovery of CD34+ stem cells from human bone marrow or blood. Once the conditions are optimized, it is expected that the % yield will be as high as 50% or greater, with a purity of close to 80%.

EXAMPLE 9 Preparation of antibodies conjugated to the biotin analogs biotin sulfone, bisnorbiotin, tetranorbiotin. and oxybiotin.

There are many other biotin analogs, in addition to desthiobiotin and 2-iminobiotin, which could be linked to antibodies for use in this invention. Examples of useful biotin analogs include biotin sulfone, bisnorbiotin, tetranorbiotin, and oxybiotin. In addition to their antigenic resemblance to biotin, these analogs are expected to participate in reactions in a similar fashion as does biotin. Thus, NHS esters of these other biotin analogs are expected to react with the amine groups of antibodies as does NHS-desthiobiotin or NHS-2-iminobiotin (see Example 1 above) .

Moreover, other types of biotin analog reagents are expected to prove useful in forming conjugates with antibodies. Biotin analog-]D-nitrophenyl ester (BaNP) is another amine-reactive reagent which could be used to label antibodies.

Certain biotin analog hydrazides are expected to react with carbohydrate and carboxyl groups of antibodies to form useful conjugates. Biotin analog hexyl-pyridyldithio-propionamide (HPDP), iodoacetyl-LC-biotin analog, and 3 - ( N.- maleimidopropionyl)biotin analog are three reagents which are expected to react with sulfhydryl groups on antibodies.

Biotin analogs with photoactivatable reactive groups may also prove useful for the covalent linkage of biotin analogs to proteins.

Reaction protocols adapted from conventional biotin chemistry will be used to conjugate antibodies with various other biotin analogs (see "Biotinylation Reagents", supra) . Each new conjugate will be tested for its suitability for use in the present invention. Each conjugate will be tested for retention of antigen binding activity by conventional methods such as ELISA, Western blot, and dot blots. Then, each conjugate will be tested for binding by anti-biotin antibodies, or avidin, using conventional methods. Then the relative affinity of the anti-biotin antibodies, or avidin, for the new conjugate will be tested against biotin or other biotin analogs which will ultimately be used to compete for binding of the conjugate in order to release the cells. It is expected that several useful biotin analog/antibody conjugates can be made based on the principle that the conjugates bear enough resemblance to biotin for recognition by the anti-biotin agent (antibody or avidin), but the anti-biotin agent has a lower affinity for the conjugate than for native biotin, allowing competitive elution and thus release of the desired cells.

Claims

What is claimed is:

1. A composition of matter comprising a stable conjugate of at least one biotin analog moiety covalently linked to a cell binding protein.

2. The composition of claim 1 wherein said cell binding protein is an antibody.

3. The composition of claim 2 wherein said antibody is specific for CD34.

4. The composition of claim 1 wherein said biotin analog is selected from the group consisting of desthiobiotin, iminobiotin, biotin sulfone, bisnorbiotin, tetranorbiotin, and oxybiotin.

5. The composition of claim 1 wherein said stable conjugate further comprises at least 7.5 biotin analog moieties covalently linked to a cell binding protein.

6. A method for positive selection of target cells from a heterogeneous cell suspension, comprising

(a) providing within said cell suspension a first complex comprising a biotin analog covalently linked to a primary antibody bound to said target cells,

(b) providing within said cell suspension a second complex comprising a secondary antibody bound to a cell separation means, said secondary antibody binding to said biotin analog to form a third complex comprising the target cells, the primary antibody, biotin analog, secondary antibody, and cell separation means,

(c) separating said third complex from said cell suspension, and

(d) incubating the separated third complex with a competitive compound for which said secondary antibody has an affinity higher than for said biotin analog, thereby releasing said target cells from said third complex.

7. The method of claim 6 wherein said competitive compound is biotin.

8. The method of claim 6 further comprising, after step (d) , separating the cell separation means from the target cells.

9. The method of claim 6 wherein said cell separation means is a solid support selected from the group consisting of paramagnetic beads, columns, hollow fibers, glass beads, polysaccharide beads, and polystyrene tissue culture flasks.

10. The method of claim 6 wherein said primary antibody is specific for the cell surface antigen CD34.

11. The method of claim 6 wherein said biotin analog is selected from the group consisting of desthiobiotin, iminobiotin, biotin sulfone, bisnorbiotin, tetranorbiotin , and oxybiotin .

12. The method of claim 6, wherein said first complex further comprises at least 7.5 biotin analog moieties covalently linked to a primary antibody.

13. A method for positive selection of "target cells from a heterogeneous cell suspension, comprising

(a) providing within said cell suspension a first complex comprising a primary antibody bound to said target cells and a secondary antibody covalently linked to a biotin analog,

(b) providing within said cell suspension a second complex comprising a tertiary antibody bound to a cell separation means, said tertiary antibody binding to said biotin analog to form a third complex comprising the target cells, the primary antibody, secondary antibody linked to biotin analog, the tertiary antibody, and cell separation means,

(c) separating said third complex from said cell suspension, and

(d) incubating the separated third complex with a competitive compound for which said tertiary antibody has an affinity higher than for said biotin analog, thereby releasing said target cells from said third comple .

14. A method for positive selection of target cells from a cell suspension, comprising

(a) providing within said cell suspension a first complex comprising avidin bound to a biotin analog covalently linked to an antibody, said antibody specifically binding said target cells to form a second complex,

(b) separating said second complex from said cell suspension, and

(c) incubating the separated second- complex with a competitive compound for which said avidin has an affinity higher than for said biotin analog, thereby releasing said target cells from said complex.

15. The method of claim 14 further comprising, after step (c), separating the target cells from the avidin.

16. The method of claim 14 wherein said competitive compound is biotin.

17. The method of claim 14 wherein said primary antibody is specific for the cell surface antigen CD34.

18. The method of claim 14 wherein said biotin analog is selected from the group consisting of desthiobiotin, iminobiotin, biotin sulfone, bisnorbiotin, tetranorbiotin, and oxybiotin.

19. The method of claim 14, wherein said first complex further comprises at least 7.5 biotin analog moieties covalently linked to an antibody.

20. A method for positive selection of target cells from a heterogeneous cell suspension, comprising

(a) providing within said cell suspension a first complex comprising avidin bound to a biotin analog covalently linked to a secondary antibody, and a primary antibody bound to said secondary antibody, said primary antibody specifically binding said target cells to form a second complex,

(b) separating said second complex from said cell suspension, and

(c) incubating the second complex with a competitive compound for which said avidin has an affinity higher than for said biotin analog, thereby releasing said target cells from said second complex.

21. A composition of matter comprising a stable conjugate of at least one biotin analog moiety covalently linked to a secondary antibody.

22. The composition of claim 21, wherein said stable conjugate further comprises at least 7.5 biotin analog moieties covalently linked to a secondary antibody.

23. A composition of matter comprising a cell separation means bound to an anti-biotin antibody, and a biotin analog moiety covalently linked to a cell binding protein, said anti-biotin antibody binding to said biotin analog moiety to form a complex.

24. The composition of claim 23 wherein said target cell binding protein is an antibody.

25. The composition of claim 24 wherein said antibody is specific for CD34.

26. The composition of claim 23, further comprising at least 7.5 biotin analog moieties covalently linked to a cell binding protein.

27. A composition of matter comprising a cell separation means bound to an anti-biotin tertiary antibody, and a biotin analog moiety covalently linked to a secondary antibody, said anti-biotin tertiary antibody binding to said biotin analog moiety to form a complex.

28. A composition of matter comprising a target cell binding protein covalently linked to a biotin analog moiety, and avidin, said avidin binding to said biotin analog moiety to form a complex.

29. A composition of matter comprising a secondary antibody covalently linked to a biotin analog moiety, and avidin, said avidin binding to said biotin analog moiety to form a complex.

30. A method for forming a stable protein-biotin analog conjugate, comprising

bringing an N-hydroxysuccinimide (NHS) ester of a biotin analog into reactive contact with an antibody in a medium at a pH of about 8 and at a temperature of about 2°-8°C.