WO1991005260A1

WO1991005260A1 - A method for the identification and selection of antibody producing cells

Info

Publication number: WO1991005260A1
Application number: PCT/GB1990/001476
Authority: WO
Inventors: Cesar Milstein; Ermanno Gherardi; Martin Dreher; Richard Pannell
Original assignee: Medical Research Council
Priority date: 1989-09-28
Filing date: 1990-09-26
Publication date: 1991-04-18
Also published as: GB8921880D0; AU6510990A; EP0494237A1

Abstract

Antibody-producing organisms or cells (prokaryotic or eukaryotic) producing a particular antibody of interest are identified by locating individual organisms or cells on a surface in spaced-apart relationship; growing the organisms or cells under suitable conditions; applying to the organisms or cells after growth a filter layer having immobilized thereon a reagent which binds to the antibody of interest; and detecting antibodies bound to the antigen. Organisms or cells producing the antibody of interest may be subsequently selected and isolated, e.g. for further growth.

Description

A METHOD FOR THE IDENTIFICATION AND SELECTION OF

ANTIBODY PRODUCING CELLS

Field of the Invention

This invention concerns production of antibodies, and relates to a method of identifying antibody-producing organisms or cells which produce a particular antibody of interest.

Background to the Invention

The derivation of hybridoma clones secreting antibodies of desired specificity is typically achieved in three steps: (1) the growth of hybridoma cultures, (2) the

identification of the cultures secreting the antibody(ies) of interest and (3) the cloning of specific lines (Galfré and Milstein, 1981). The main limitation to this

procedure is the ratio of positive to negative clones.

This is because screening methods must be capable of detecting a single positive clone out of perhaps 10³ or 10⁴ or more negative ones. In addition, if such positive clones are mixed with negative ones, there is an increased probability of loss of valuable clones due to overgrowth by fast-growing negative competitors (Galfré et al.,

1980). Methods to avoid these shortcomings have been described in the literature, which rely on cloning before the screening step. Cloning is usually performed on solid supports, such as agar or agarose, but methyl cellulose has also been used (Dennis et al., 1982). Direct

detection of positive clones is performed using red cells or derivatised red cells in two ways: (1) Complement- dependent haemolytic plagues of red cells overlaid on developed clones (Kohler and Milstein, 1975) or mixed with hybridomas (Galfre and Milstein, 1981); (2) Binding of the red cells to a nitrocellulose replica containing adsorbed supernatant of the growing colonies (Sharon et al., 1979). Reference is also made to Kelsoe, 1987.

Summary of the Invention

According to one aspect of the invention there is provided a method of identifying antibody-producing organisms or cells which produce a particular antibody of interest, comprising locating individual organisms or cells on a surface in spaced-apart relationship; growing the

organisms or cells under suitable conditions; applying to the organisms or cells after growth a filter layer having immobilized thereon a reagent which binds to the antibody of interest; and detecting antibodies of interest bound to the filter layer.

The antibody-producing organisms or cells may be

prokaryolic organisms, such as bacterial cells, and are preferably cells which secrete antibody either into the cell periplasmic space or from the cell completely into surrounding medium. The organisms or cells may

alternatively be eukaryotic cells, eg mammalian cells such as mouse cells.

The antibody may be a complete antibody, including a bispecific antibody, or an antibody fragment such as an Fab or Fv fragment.

Suitable conditions for growth are selected depending on the nature of the organism or cell. Eukaryotic cells are preferably grown in a feeder layer, which in some

embodiments conveniently comprises a layer of soft agarose containing STO fibroblast feeder cells.

When dealing with eukaryotic cells, the feeder layer is preferably covered with a diffusion layer, eg of agar: this assists clustering of cells derived from a particular individual initial cell during cell growth, and so helps maintain separation of different cell colonies.

The filter layer conveniently comprises nitrocellulose.

The reagent immobilized on the filter layer may be antigen to the antibody. Antigen may be immobilized using known techniques, and may be bound to the layer directly or indirectly, eg via a secondary molecule such as another antibody, or using a biotin/avidin link. In an

alternative approach, the filter layer may be coated with anti-immunoglobulin or equivalent antibody capturing reagents.

Antibody of interest bound to immobilized reagent (either directly or indirectly) may be detected using known techniques, eg using radiolabelled or enzyme-labelled second antibody, as will be well known to those skilled in the art.

After identification of organisms or cells producing the antibody of interest, by detection of bound antibodies, the organisms or cells may be selected and isolated.

Isolation may be achieved either by removing the organisms or cells from the surface, or by physically isolating the organisms or cells on the surface, eg by providing a suitable barrier separating the organisms or cells of interest from the surroundings. Isolated organisms or cells may be grown further or otherwise treated.

The method of the invention enables rapid screening and identification of organisms or cells producing antibody of interest, using a colony replica assay. In preferred embodiments at least, the method used can improve

considerably both the growth of organisms or cells and the ease and speed of screening as compared with known

methods. Further advantages can be gained from the screening of multiple replicas prepared from a single dish. using this procedure, several thousand hybridoma clones can be screened in a few hours and selected

antigen-specific hybridoma lines can be derived less than two weeks from the date of fusion.

The method of the invention can be used to discriminate on the basis of antigen affinity of antibodies, by screening with different amounts of antigen immobilized on the filter layer, eg using different filter layers each with different concentrations of antigen immobilized thereon. The filter layers may be produced by linking antigen molecules to carrier molecules such as bovine serum albumin (BSA) or possibly another antibody in varying proportions (eg ranging from about 15 to about 1 molecules of antigen to each molecule of carrier) and coating the filter layer with a fixed constant amount of the antigen- carrier conjugate. This technique provides a convenient means for achieving defined filter coatings carrying desired and variable amounts of antigen. This approach will generally be used to detect higher affinity

antibodies, but also has application in detection of lower affinity antibodies. In a preferred aspect the present invention provides a method of selecting antibody-producing eukaryotic cells which produce a particular antibody of interest,

comprising locating individual cells on a surface in spaced-apart relationship; growing the cells under suitable conditions; applying to the cells after growth a filter layer having immobilized thereon a reagent which binds to the antibody of interest; detecting antibodies of interest bound to the filter layer; and selecting and isolating cells producing the antibody of interest.

Another preferred aspect of the invention provides a method of identifying antibody-producing prokaryotic organisms which produce a particular antibody of interest, comprising locating individual organisms on a surface in spaced-apart relationship; growing the organisms under suitable conditions; applying to the organisms after growth a filter layer having immobilized thereon a reagent which binds to the antibody of interest; and detecting antibodies of interest bound to the filter layer.

The invention will be further described, by way of

illustration, in the following Examples and by reference to the accompanying drawings, in which:

Figure 1 is a replica assay of anti-hapten hybridoma clones cultured on STO fibroblast feeders in soft

agarose;

Figure 2 is a replica assay of the anti-oxazolone

hybridoma clones NQ2/16.2 and NQ11/7.12;

Figure 3 is a graph of ¹²⁵I-Rb anti-mouse 1gG bound (cpm x 10³/cm²) versus Ox_{14. 4}-BSA (pmoles/cm²);

Figure 4 is a replica assay of NQ2/16.2 hybridoma clones using an indirect assay with an anti-immunoglobulin coated filter layer;

Figure 5 is a comparison of results for direct and indirect replica assays; and

Figure 6 illustrates schematically a bacterial colony assay for antibody fragments.

Example 1

Materials and Methods

Cell Cultures

The mouse myeloma NSO (Clark and Milstein, 1981) and the mouse hybridoma NQ2/16.2 Ag.8 (an azaguanine-resistant clone of the hybridoma N02/16.2; Kaartinen et al., 1983) were cultured in DMEM + 5% FCS. The mouse embryo

fibroblast line STO (Martin and Evans, 1975) was

maintained in DMEM + 10% FCS and passaged twice a week at a split ratio of 1:4-1:6. STO feeders were produced by incubating subconfluent cultures in DMEM + 10% FCS supplemented with mytomicin C (Sigma M 0503, 10 ug/ml in DMEM + 10% FCS) for 2 hours at 37°C, after which cells were washed twice with PBS, lifted with trypsin-EDTA and melting point agarose (BRL 5517UB) in DMEM + 20% FCS.

Dishes were kept on ice for 10 mins before being returned to the incubator and were used, typically, 2 days after plating. Derivation of Hybridomas

The fusion procedure previously detailed (Galfré and

Milstein, 1981) was followed. After fusion cells were resuspended at 1 x 10⁷ cells/ml in DMEM + 20% FCS and an equal volume of 0.5% agarose in DMEM + 20% FCS + HAT 2x was added. 2 mis of cell suspension in agarose (1 x 10⁷ spleen cells) were plated in each dish on STO feeders.

Dishes were left on ice for 5-10 mins and returned to the incubator where they were left, undisturbed, for a week.

After a week, plates were scored for growth using an inverted microscope and a 1 x 1 cm grid on a polyacetate sheet at the bottom of the plate. Clones with more than

8-16 cells were counted in at least 8 cm² across the plate.

Replica Assay of Antigen Specific Clones

10-12 days after fusion, dishes were overlaid overnight with 2 mis of 0.75% agar (Bacto Agar, Difco Laboratories) in DMEM equilibrated at 42°C.

Binding of hapten-BSA conjugates to 82mm diameter

nitrocellulose discs (Schleicher and Schuell, 40116) was performed overnight at 4ºC using 1.3 nmoles of BSA/filter in 2.5 mis of PBS (this corresponds to different

concentrations of hapten, depending on th substitution ratio of each hapten-carrier conjugate, see below). In the morning the filter was blocked with 2.5 mis of 5% FCS in PBS at 37ºC for 2 hours, blotted between two circles (15 cm diameter) of Whatman filter paper in 1 (cat. 1001 150) and applied on the surface of culture dishes for approximately 1 hour. Using a hypodermic needle dipeed in waterproof drawing ink (Rotring 591 0176), three

asymmetric alignment marks were then made by piercing the nitrocellulose filter and the culture. The filters were then peeled off gently and transferred (face up) in 10 cm dishes containing 5 mis of PBS + 5% FCS + rabbit

peroxidase-conjugated anti-mouse IgG(Dakopatts P260, diluted 1:500). Incubation with second antibody was for 1 hour at room temperature followed by 3 x 5 mins washings with Tween 20 (1g/1) in PBS (10-15 mis/dish) and a final washing with PBS. Filters were then incubated for about 5 mins with peroxide substrate (200 ug/ml diaminobenzidine, 0.075 ug/ml H₂O₂, 80 ug/ml NiCl₂ in PBS). Filters were washed with water and dried on Whatman paper. Dried filters (placed face up) were glued on to the paper sheet of A4 size, paper backed, polyacetate transparencies

(Kodak 358 4273). The alignment marks and the contour of the spots and the filter were drawn on the polyacetate sheet. This was then peeled off, turned upside down (so that the spots were in the same orientation of the

colonies in the Petri dish) and used for clone alignment. The polyacetate sheet was either cut into discs (and each disc aligned and taped at the bottom of the corresponding Petri dish) or kept intact and transferred on the stage of a dissecting microscope. In the latter case Petri dishes were aligned, one at a time, for colony picking. Clones were picked with a Pasteur pipette, drawn to a fine bore and transferred into 96 well plates containing 1,000 mitomycin-treated STO cells in 0.2 m/well of DMEM + 20% FCS + HAT. Further details of the procedure for clone alignment and picking are given in the Results section.

ELISA

This was performed in flexible polyvinylchloride 96 well assay plates coated overnight at 4°C with 10 p moles (carrier)/well of hapten-carrier conjugates. Plates were blocked with 5% FCS in PBS for 2 hours at 37°C, incubated with test antibodies for 2 hours at room temperature and washed three times with PBS. Incubation with peroxidase- conjugated second antibody (see above) was for 1 hour room temperature in 5% FCS in PBS followed by three washings with PBS and addition of ABTS (Sigma A-1888) substrate.

Hapten-Carrier Conjugates

2-phenyl-4-ethyoxmethylene-oxazolone (Ox) (BDH 44160 2W) was coupled to CSA (chicken serum albumin) and BSA, as described by Makela et al. (1978). Ox-BSA conjugates with 14.1, 3.1, 1.5 and 0.8 hapten/carrier molar ratios were prepared using different ratios of hapten/carrier (25, 10, 5 and 2.5 respectively) in the reaction.

2, 4-dinitrofluorozenene (DNP) (Sigma, D-1529) was

conjugated to keyhole limpet haemocyanin (KLH) and BSA in 0.1M borate buffer, pH 8.5.

Other Procedures

IgG monoclonal antibodies were purified from ascites or serum-free culture medium by affinity chromatography on Protein A-Sepharose 4B or Protein A-Superose columns

(Pharmacia-LKB) according to standard techniques.

Antibody purity was assessed by SDS-polyacrylamide gel electrophoresis (Laemmli, 1970) and antibody concentration was measured by A₂₈₀ assuming A₂₈₀=1.45 at 1 mg/ml. K_d values were measured (using phenyl-oxazolone gamma-amino butirate) by fluorescence quenching, essentially as described by Eisen (1984). Results

Clonal Growth and Replica Assay of Antigen-Specific

Hybridomas

Table 1 shows the results of a set of experiments in which mouse or rat spleen cells from animals which had received a primary immunisation with Ox-CSA or DNP-KLH were fused with the mouse myeloma NSO or the mouse hybridoma NQ2/16.2 Ag.8 and plated directly in soft agarose on STO feeders. The number of growing clones varied considerably, as expected, from one experiment to another, but was always greater than 1,000 clones/10⁸ spleen cells, and averaged 3554 clones/10⁸ spleen cells in the set of experiments reported here. Each clone represents a different fusion event, since the cells were plated before expansion. This ensures maximum representation of the initial "library".

Twelve days after fusion, clones secreting antigen- specific antibodies were identified using a replica assay on nitrocellulose discs coated with Ox-BSA or DNP-BSA. Clones were protected from direct contact with the paper by means of an intermediate agar layer, which was left overnight to allow diffusion of the antibodies. The antigen-coated nitrocellulose discs were then applied on the agar layer for 1 hour, and antibodies bound to nitrocellulose were detected with horseradish peroxidase- conjugated anti-mouse immunoglobulin. Figure 1 shows representative results of the procedure described, for fusions Q2R3 and NQS1. The figure shows the mirror image of antigen-specific clones from representative dishes.

Several features of the replica assay have become apparent during our studies and are discussed here, as they affect considerably the quality of the results.

(i) Age of the Culture

The assay was performed with cultures less than 8 to 10 days old (established lines) or 10 to 12 days old (newly established hybridomas). When older cultures were used, a dark and uniform antibody staining appeared on the filter and the localised signal due to individual clones was lost.

(ii) Clone Alignment

It was crucial to establish conditions which would allow unequivocal alignment of the replicas and their respective clones. Following the procedures described in Materials and Methods, this can usually be achieved without

difficulty. Each spot on the polyacetate sfyeet was numbered and, after the polyacetate sheet had been aligned with the dish, all the clones falling in the area marked were picked. Clones were classified as (A), to indicate a clone falling in the centre of the area marked, or (B), to indicate a peripheral clone. Sometimes two or more clones were too close to allow the identification of positive one(s). These were classified as (C), picked as a

mixture, re-cloned and re-screened with the replica assay. Re-cloning of all newly-established hy.bridomas is

eventually necessary to increase stability of the lines.

Selection of High-Affinity Anti-Hapten Antibodies with Multiple Replicas

Several methods have been proposed to rank the binding affinity of monoclonal antibodies based, typically, on RIA or ELISA results (van Heyningen et al., 1983; Griswold and Velson, 1984 and 1985; Beatty et al., 1987). We found that anti-oxazolone monoclonal antibodies previously isolated in our laboratory (Griffiths et al., 1984) bound differently to oxazolone-carrier conjugates in which the ratio of hapten to carrier was varied. Although both low- and high-affinity antibodies bound well to Ox-BSA

conjugates with a substitution ratio of greater than or equal to 14.4, the binding of low-affinity antibodies to Ox-BSA conjugates with substitution of less than or equal to 3.1 was modest. Thus, the ratio between the antibody binding to Ox_{14. 4}-BSA and to Ox_{1 .5}-BSA (or Ox_0.8-BSA) correlated inversely with the K_d values (Table 2). A similar finding has been reported with antibodies to other haptens (Haas, 1975; Haas and Layton, 1975; Herzenberg et al., 1980; Rothstein and Gefter, 1983; Lew, 1984).

Although the resolving power of the assay was maximal at low (1-10 nM) and identical concentrations of antibody (Table 2), the assay had sufficient resolution at the concentrations of antibody present in hybridoma

supernatants (100-1000nM).

We established a new set of hybridomas secreting

monoclonal antibodies to the hapten oxazolone. These were derived on STO feeders and identified using the replica assay (as described above) using nitrocellulose filters coated with Ox_14.4-BSA, thus allowing the isolation of both low- and high-affinity antibodies. Positive clones were transferred in wells and the supernatants from nine clones secreting IgG monoclonal antibodies were tested by ELISA for binding to Ox_14.4-BSA and Ox_{1. 5}-BSA (Table 3). These cultures were then expanded to allow purification of the antibodies and measurement of their binding affinities (Table 3).

Although the correlation between the K_d values and the ELISA results did not hold for all antibodies, it did nevertheless apply to the majority of the antibodies studied. For example, the best three antibodies in the ELISA assay (NQP2/10, NQS2/10 and NQS2/14) were 4th, 3rd and 1st in terms of binding affinity. Similarly, the last three (NQS2/3, NQS2/9 and NOP2/8), were the 5th, 8th and 9th out of 9 antibodies studied.

These results prompted us to study the possibility of identifying high-affinity anti-hapten antibodies directly from the replica assay. Figure 2 shows an example of this approach. Figure 2 is a replica assay of the anti- oxazolone hybridoma clones NQ2/16.2 and NQ11/7.12. Each cell line was subcloned in soft agarose (about 2500 cells/10 cm dish, without feeders) and assayed after 10 days using the colony replica assay. Nitrocellulose discs were coated with Ox_{14. 4}-BSA (first lift), Ox_{3 .1}-BSA

(second lift) and Ox_{0. 8}-BSA (third lift), and three sequential replicas of the anti-oxazolone antibodies secreted by the NQ2/16.2 and NQ11/7.12 lines (K_d=280 and less than or equal to 2nM respectively) were obtained from the same dish. The discrimination between the binding of the two antibodies was maximal at Ox_0.8-BSA (Figure 2) and at Ox_1.5-BSA (data now shown). In the experiment shown, the first replica was with Ox_{14. 4}-BSA followed by Ox_3.1- BSA and by Ox_0.8-BSA, but the results were the same where the order was reversed (data not shown). In other

experiments, we found that using the same assay

conditions, a clear discrimination could be obtained between NQ2/16.2 and NQ10/12.5 antibodies (K_d=280 and 10 nM respectively), whereas the discrimination between

NQ10/12.5 and NQ11/7.12 antibodies (K_d=10 and less than or equal to 2 nM respectively) was never complete.

Discussion

The desirability of achieving clonal growth of hybridomas immediately after fusion need not be emphasised. The loss of valuable hybridomas due to clonal competition has been observed several times in our laboratory (see, for

example, Galfre et al., 1980), as well as in other

laboratories. The most common method of initial

distribution of the hybrid population at a limiting dilution containing single viable cells per each

microculture is not only labour-intensive but also time- consuming. Clones must grow to a minimum cell density for supernatants to be assayed. During this period, the danger of negative variants arising with better growth characteristics is at its highest.

The procedure described here consistently achieves clonal growth of newly established hybridomas in soft agarose on a layer of STO fibroblasts. We do not use, routinely, feeder cells for cloning established hybridomas, but in some cases this proved essential for good yields of newly established hybrids. These are screened within 10-12 days, when clones contain at most 10 7-103 cells, and thus allows early manipulations to stabilise the lines.

Different sources of cells have been used, over the last twenty years, to sustain the clonal growth of myeloma and hybridomas. These include fibroblasts (Coffino and

Scharff, 1971; Sharon et al., 1979: Butcher et al., 1988), macrophages (Fazekas de St Groth and Scheidegger, 1980), thymocytes (Oi and Herzenberg, 1980; Davis et al., 1982) and splenocytes (Coding, 1986). Media conditioned by different cell lines and partially purified growth factors have also been employed successfully (Pintus et al., 1983; Nordan et al., Rathjen and Geczy, 1986; van Snick et al., 1986; Micklem et al., 1987; King and Sartorelli, 1987) and may well, sometime in the future, replace completely the use of feeder cells. We have chosen the STO fibroblasts for our studies because, in our experience, they proved to be more efficient than thymocytes, macrophages and several other fibroblast lines in supporting the growth of newly established hybridomas. As yet, we have not attempted to replace the use of these cells with partially purified or purified growth factors of fibroblastic, macrophagic or thymocytic origin.

The ability to derive large numbers of independent clones after fusion is only useful if coupled to a rapid colony assay for the identification of the clones of interest. In the replica assay we describe, the nitrocellulose is coated with the antigen of interest (Ox-BSA or DNP-BSA in our case), and blocked before being applied on the culture dishes. This is suitable for antigens (such as proteins and nucleic acids) which can be immobilised on

nitrocellulose. We have not yet attempted to immobilise cell membranes or intact cells to nitrocellulose with the view of extending the applications of the assay to these systems. Indirect methods are, however, more suitable for such cases and more generally applicable. This involves coating the nitrocellulose with anti-mouse (or rat) immunoglobulin and detection of antigen-specific clones by incubation with trace concentrations of labelled antigen. Direct adsorption of antibodies from the culture dish to uncoated nitrocellulose discs is very inefficient, probably due to competition with proteins in the culture medium (data not shown).

An important feature of this method is the possibility it affords in terms of multiple assays on a single or on separate replicas. We have used the assay in order to select for high-affinity anti-hapten antibodies and have exploited, for this purpose, the preferential ability of high-affinity antibodies to react with hapten-carrier conjugates with low substitution ratios. The results obtained (Figure 2) indicate that anti-oxazolone

antibodies with K_d in the range of 10-280 nM could be discriminated. It is possible that minor modifications to the assay could resolve antibodies with K_d less than 10 nM.

The replica assay has several other areas of application. For example, it would allow the rapid screening for bispecific antibodies, where the need for early cloning has been specifically emphasised (Suresh et al., 1986). Clones secreting bispecific antibodies could be detected by the binding of the antibodies to the first antigen (immobilised on nitrocellulose) followed by washing and incubation with a solution containing labelled second antigen. Another exciting application of the colony replica assay will be the screening of antigen-specific antibody fragments expressed in E. coli (Better et al., 1988; Skerra et al., 1988; Ward et al., 1989) or yeasts (Horwitz et al., 1988). Antigen-binding Fab fragments secreted in the medium could be detected essentially with the procedure described. The detection of Fv fragments will probably require several modifications of the basic protocol to ensure efficient detection of antigen-bound Fv. The results of the direct colony replica assays described above for detection of anti-hapten antibodies were optimal when concentrations of hapten-carrier conjugate of 25 pmoles (carrier) per cm² were employed. With the hapten- carrier conjugate used in most experiments (Ox_{14 .4}/BSA at a coupling ratio: 14.4/1) this corresponded to about 360 pmoles of hapten per cm² or about 20 nmoles of hapten per filter (8.2 cm diameter). If comparable concentrations of a protein antigen were required for optimal assay results, mg amounts of protein could be needed to screen entire fusions by the direct colony replica assay. This meant that the screening could not be used in the many instances in which the amount of antigen was limiting. We therefore set out to investigate if other membranes would provide a better assay at lower concentration of antigen.

Example 2

We have studied the amount of ¹²⁵I-Ox/BSA immobilized on several membranes when the concentration of antigen was increased up to 1 nmole (carrier) per cm². Under the conditions used, nitrocellulose showed the highest

capacity for protein binding (about 80% at 1 nmole/cm²). A nylon membrane and an iodoacetamide-activated membrane bound between 30% and 40% of the antigen at the highest concentrations; other chemically-activated membranes and a PVDF membrane bound comparatively less. The difference between nitrocellulose and the other membranes were equally marked at the lower coating conditions. Very similar results were obtained when 125I-rabbit IgG was used instead of ¹²⁵I-Ox/BSA.

Figure 3 shows how some of the membranes performed in an assay which mimicked the direct colony assay. Squares of membranes were coated with different concentrations of Ox/BSA from 0 to 10 pmoles/cm², blocked, incubated with the supernatant of an anti-hapten antibody (NQ2/16.2) and finally probed with ¹²⁵I- rabbit anti-mouse IgG, according to the following method.

Squares (0.7 x 0.7 cm) of each membrane were transferred to a 24 well tissue culture plate and incubated for 5 hours at room temperature with Ox/BSA (0-10 pmoles/cm²) in 0.1 ml of phosphate buffer saline (nitrocellulose, Hybond N, Immobilon P, and UAM-AE), or 0.5 M potassium carbonate, pH 9.3 (UAM-EA) or 0.1 M phosphate, pH 7.5 (UAM-IA). UAM membranes were blocked with 0.15 ml/well of 1M

ethanolamine, pH 8.5 and all membranes were incubated for 1 hour at 37 °C in 5% fetal calf serum in phosphate

buffered saline and washed twice with 0.25 ml/well of PBS. 0.1 ml of spent supernatant of NQ2/16.2 culture was added to each well and incubation continued for 1 hour.

Membranes were washed 3 x 5 mins with 0.25 ml/well of a solution of Tween 20 in phosphate buffered saline (1 g/1) and incubated for 1 hour with ¹²⁵I-rabbit anti-mouse IgG (50,000 pmole in 0.1 ml of 5% fetal calf serum in

phosphate buffered saline). Membranes were washed as carried out after the incubation with primary antibody, dried and counted.

The following different membranes were used in the assay: Immobilon P (Millipore), nitrocellulose 0.45 urn

(Schleicher and Schuell), Hybond N (Amersham), UAM-IA, UAM-EA and UAM-AE (Nygene).

The experiment clearly showed that nitrocellulose and PVDF had the higher signal/background ratio whereas the nylon and one of the chemically-activated membranes had much higher background and limited resolving power.

Example 3

Indirect Assay (Anti-Immunoglobulin-Coated Membranes)

The results with the direct assay and the search for a "universal" assay system reguiring limited amount of antigen for the screening of a large number of clones induced us to explore an indirect assay in which secreted antibodies are captured on the solid support and antigen specific clones are visualized by binding to radiolabelled (or alternatively labelled) antigen. We first explored this approach in model experiments which mimicked the assay with hybridoma clones. Squares of membranes were coated with increasing concentration of rabbit anti-mouse

IgG, blocked, incubated with NQ2/16.2 supernatant and finally probed with ¹²⁵I-Ox/BSA, according to the

following method.

NQ2/16.2 hybridoma cells were cloned in agarose in 10 cm dishes. Eight days after cloning, the dishes were

overlaid overnight with 2 ml of 0.75% agar in DMEM. Assay membranes were coated with 50 pmoles/cm² of rabbit anti- mouse IgG (Dako P259) for 5-6 hours at room temperature as described in Example 2 and blocked for 1 hour at 37ºC or 42°C (Hybond-N membrane) in 5% fetal calf serum in

phosphate buffered saline. Membranes were blotted between two circles of Whatman filter paper n.1 and applied for 1 hour to the surface of culture dishes at 37°C before incubation for 1 hour with 1 x 10⁶ cpm/ml of Ox/BSA.

Membranes were washed with lg/l Tween 20 in phosphate buffered saline (3 times x 5 mins), dried and exposed overnight at 70°C.

The following different membranes were used: 1: Immobilon P (Millipore), 2 and 5: nitrocellulose 0.45 um (Schleicher and Schuell), 3: Hybond N (Amersham), 4: UAM-IA (Nygene), 6: UAM-EA (Nygene).

The results of the experiment showed that in the model assay the chemically-activated and nitrocellulose

membranes produced comparable results and that the nylon membrane had, as in other experiments, very high

background. The PVDF membrane had the lowest background and the lowest signal but a good signal/background ratio.

The results of the experiment are shown in Figure 4 and clearly show that the PVDF membranes produced the best signal and the lowest background. In view of the

relatively low amount of capturing reagent bound to PVDF membranes (see above) it would appear that the results obtained with these membranes may be due to the

orientation of the anti-immunoglobulin reagent on the membrane or a higher retention of the capturing reagent during the course of the assay or both.

Example 4

Comparison of Direct and Indirect Procedures and the Use of Monoclonal Antibodies as Immobilizing Reagents in Colony Replica Assays

The results obtained with the indirect assays suggested a direct comparison with the direct assay. PVDF membranes were coated with either antigen or rabbit anti-mouse imunoglobulin reagents (at 50 pmoles/cm² ) and filter-bound anti-hapten (NQ2/16.2) antibodies were visualized with 1 x 10⁶ cpm/ml of ¹²⁵I-rabbit anti-mouse IgG or ¹²⁵I-Ox/BSA respectively.

The NQ2/16.2 hybridoma cells were cloned and used for colony replica assays as described in Gherardi et al.

(1990) and in Example 3. Immobilon P membranes were coated for 8 - 10 hours at room temperature with either anti-immunoglobulin reagents or antigen, as described below, blocked for 1 hour at 37°C in 5% fetal calf serum in phosphate buffered saline, blotted between 2 circles of Whatman n.1 paper and applied for 1 hour on the surface of culture dishes which had been overlaid overnight with 2ml of 0.75% agar. Filters were then incubated for 1 hour with radiolabelled antigen or anti-immunoglobulin, washed three times x 5 mins with Tween 20 in phosphate buffered saline, dried and exposed overnight at 70ºC.

The filters were coated and probed with the reagents listed in Table 4. The concentration of probe was 2.5 x 10⁶ cpm/ml in 2.5 ml/filter in each case.

The results are shown in Figure 5 and indicated that the indirect assay was at least five times more sensitive than the direct assay.

In the same experiment we compared the polyclonal rabbit anti-mouse IgG antibody with two monoclonal antibodies directed against the constant regions of the mouse heavy or kappa chains respectively for their ability to capture the secreted antibodies. The results (Figure 5) suggested that these monoclonal antibodies were superior to the polyclonal reagent and showed that both anti-heavy and anti-light chain reagents could be used in the capture assay.

Example 5

Colony Assay for Bacterial Cells

A. As illustrated in Figure 6, plasmid harbouring

bacteria are grown into colonies 10 on a nitrocellulose filter 12 placed on top of an agar plate 14 in a dish 16 with an appropriate antibiotic selection including 100 ug/ml Ampicillin in 1% glucose until a small colony has formed. The nitrocellulose filter with bacterial colonies is transferred from the agar plate to another which has been supplemented with IPTG. This arrangement facilitates the handling of the bacteria in several ways:

1) All bacterial colonies are transferred in one step from the agar plate with the growth medium to another agar plate 18 with the components necessary for induction (ie production of the antibody fragments), including 100 ug/ml Ampicillin and 1 mM IPTG.

2) If desired the filter can be used to produce an exact replica of the bacterial colonies. This allows duplicate screening to be performed.

3) Bacteria are partially immobilized on the filter matrix facilitating rescue of viable bacterial colonies in subsequent steps.

B. After the induction of the bacterial colonies the filter 12 with the induced colonies 18 is covered with another filter 20 in the form of a 0.22 u membrane of low protein binding properties (eg Millipore, CatNo: GVWP 090 50) and overlaid with an antigen coated nitrocellulose filter 22. The filters are placed on top of about 10 layers of 3MM Whatmann paper 24 soaked in osmotic shock buffer (0.2M NaBorate, 0.16M NaCl, pH8) and covered with about 20 layers of dry blotting paper 26. The osmotic shock buffer diffuses through the filter with the

bacteria, selectively releasing periplasmic bacterial proteins including the antibody fragments which penetrate the second inert filter by capillary action and bind to the antigen-coated top filter. Subsequently the

immobilized Fv-fragments on the antigen-coated filter can be visualized by, for instance, appropriate immunological reagents. The technique offers the following advantages:

1) The osmotic shock treatment releases the desired antigen binding protein and is carried by the flow of the blotting buffer to the antigen- coated filter. This process is completed within minutes.

2) In contrast to other protocols not all bacteria lyse. Enough of them survive the blotting experiment which is essential for later rescue of positive clones.

3) The low protein binding size exclusion filter between the bacterial and antigen filter prevents debris from the bacterial colonies from coming into direct contact with the antigen coated filter, which improves the signal to noise ratio and also holds the colonies (which may be very small) firmly in their position on the filter. C. We envisage that recombinant DNA techniques can be used to produce on any one filter several thousand bacterial colonies each with antibody fragments of

potentially different binding properties. The desired antibody fragment can be found by screening such a

"library" with the above method. For example, we have successfully applied the above method to a bacterial expression library of limited heterogeneity: We randomly mutated one amino-acid residue of a bacterially expressed anti-oxazolone Fv-fragment such that any amino-acid

(including the wild-type) would occur at this particular position. We determined the presence of functional antigen-binding Fv-fragment by colony screening: few positive ("antigen-binding") clones amongst many negatives were found. DNA sequence analysis of the plasmid carried by positive and negative bacterial clones showed that all the negative colonies have plasmids with an amino acid different from the wild-type and the few positives have the original wild-type sequence restored. Thus, we have demonstrated that only the wild-type can restore binding activity and that this particular amino-acid residue is essential for antigen-binding. Any other amino acid at its place destroys the binding site and no functional Fv- fragment can be assembled. We are in the process of extending these studies to more diverse libraries and also addressing the question of isolating higher-affinity Fv fragments using the above methods.

It is envisaged that the superior results obtained with the indirect colony assays will be paralleled in the bacterial system. Experiments are now in progress in which PVDF membranes coated with anti-tag or anti-light chain monoclonal antibodies are used to capture Fv or Fab fragments and these are detected by incubation with radiolabelled (or alternatively labelled) antigen.

Applications of Colony Assays

1. Improved Screening and Selection of Antibodies

Secreted by Hybridomas, Transfectants and EBV

Transformants

The ability to clone newly established hybridomas in agarose directly after fusion coupled with the

availability of colony assays extends considerably the yield and recovery of antigen-specific clones. Similarly the colony assays described here have been used

successfully to screen and detect antibodies expressed by transfectants. This is of interest because we often find that the concentration of antibody in the supernatant of transfectant cultures is in the order of 1% or less of the concentration present in the supernatant of many

hybridomas. A special application which is now being explored is the screening of clones of EBV transformants. The instability and low level of expression of EBV lines are well known and it is hoped that gene technology will be able to rescue poteitnal antibody specificities which could otherwise be lost. Colony assays may prove ideal to identify the clones with the relevant specificities and may avoid the need to clone the heavy and light chain genes of mixed populations of cells which then require much time being spent in expressing pairs of heavy and light chain to allow the identification of the antigen- specific one.

2. Screening of Complex Bacterial and Yeast Antibody Libraries It is likely that a number of clones in the range of 10⁴ to 10⁸ will need to be screened to rescue antigen positive pairs of heavy and light chain. There may currently be an absolute requirement of colony assays for the screening of such libraries.

3. Screening for High Affinity Antibody Variants

We have shown (see Example 1) that the direct colony assay could be successfully used to select for high affinity anti-hapten antibodies by using filters coated with hapten-carrier conjugates with decreasing substitution ratios. This screening and selection procedure is

applicable whenever high affinity antibodies to small molecules are to be derived or engineered.

We clearly envisage, however, that the indirect assay would provide a general assay system for selection of high affinity antibodies to both small molecules and large antigens. In this modification of the assay the

antibodies (or antibody fragments)- captured on the membrane are incubated with mixtures of radiolabelled and unlabelled antigen in which the concentration of the unlabelled species is varied and set at values which allow binding of the trace label only to the high (or higher) affinity antibodies.

4. Screening for Isotype Specific Antibody Variants

The choice of the capturing reagent in the indirect assay can be used to determine the isotype of the antibodies to be selected (for example after fusion or EBV

transformation). This requires the use of isotype specific monoclonal antibodies on the membrane (as shown in Example 4) and may be important when particular

isotypes have to be selected from the beginning in view of particular applications.

References

Beatty, J.D., Beatty, B.G. and Vlahos, W.G. (1987)

Measurement of monoclonal antibody affinity by non- competitive enzyme immunoassay. J. Immunol. Methods 100, 173.

Better, M., Chang, C.P., Robinson, R.R. and Horwitz, A.H. (1988) Escherichia coli secretion of an active chimeric antibody frament. Science 240, 1041.

Butcher, R.N., McCullough, K.C., Jarry, C. and Bryant, J. (1988) Mitomycin C-treated 3T3/ B (3T3/A31) cell feeder layers in hybridoma technology. J. Immunol. Methods 107, 245.

Clark, M.R. and Milstein, C. (1981) Expression of spleen cell immunoglobulin phenotype in hybrids with myeloma cell lines. Somatic Cell Genetics 7, 657.

Coffino, P. and Scharff, M.D. (1971) Rate of somatic mutations in immunoglobulin production by mouse myeloma cells. Proc. Natl. Acad. Sci. USA 68, 219.

Czerkinsky, C.C., Nilsson L.-A., Nygren, H. et al (1963) J. Immunol. Methods 65,109.

Davis, J.M., Pennington, J.E., Kubler, A.-M. and

Conscience, J.-F. (1982) A simple, single-step technique for selecting and cloning hybridomas for the production of monoclonal antibodies. J. Immunol. Methods 50, 161.

Eisen, H.N. (1964) Determination of antibody affinity for haptens and antigens by means of fluorescence quenching. In Methods in Medical Research, Vol. 10, H.N Eisen (Ed.), Year Book Medical Publishers, Inc., Chicago, p. 115.

Fazekas de St Groth, S. and Scheidegger, D. (1980)

Production of monoclonal antibodies: Strategy and tactics. J. Immunol. Methods 35, 1.

Galfré, G., Butcher, G.W., Howard, J.C., Wilde, CD. and Milstein, C. (1980) Clonal competition and stability of hybrid myelomas of mouse and rat origin. Transplantation Proc. XII, 371.

Galfré, G. and Milstein, C. (1981) Preparation of

monoclonal antibodies: Strategies and Procedures. Methods in Enzymol. 73, 3.

Gherardi, E., Pannell, R. & Milstein, C. (1990) J.

Immunol. Methods, 126, 61.

Griffiths, G.M., Berek, C., Kaartinen, M. and Milstein, C. (1984) Somatic mutation and the maturation of immune response to 2-phenyl oxazolone. Nature 312,271.

Griswold, W.R. and Nelson, D.P. (1984) A rapid method for the determination of antibody affinity. Immunology Letts. 7,229.

Griswold, W.R. and Nelson, D.P. (1985) Calculation of monoclonal antibody affinity constants directly from antibody dilution curves . Immunology Letts . 9 , 15 .

Haas, W. (1975) Separation of antigen-specific lymphocytes

II. Enrichment of hapten-specific antibody-forming cell precursors. J. Exp. Med. 141, 1015. Haas, W. and Layton, J.E. (1975) Separation of antigen- specific lymphocytes I. Enrichment of antigen-binding cells. J. Exp. Med. 141, 1004.

Herzenberg, L.A., Black, S.J., Tokuhisa, T. and

Herzenberg, L.A. (1980) Memory B cells at successive stages of differentiation. J. Exp. Med. 151, 1071.

Horwitz, A.H., Chang, C.P., Better, M., Hellstrom, K.E. and Robinson, R.R. (1988) Secretion of functional antibody and Fab fragment from yeast cells. Proc. Natl. Acad. Sci. USA 85, 8678.

Jerne, N.K., Nordin, A.A. & Henry, C. (1963) in "Cell- bound Antibodies" (Amos B. & Koprowski H., eds.), Wistar Inst. Press, Philadelphia, p.109.

Jerne, N.K. & Nordin, A.A. (1963) Science 140, 405.

Kaartinen, M., Griffiths, G.M., Markham, A.F. and

Milstein, C. mRNA sequences define an unusually restricted IgG response to 2-phenyloxazolone and its early

diversification. Nature 304,320.

Kelsoe, G., (1987) Cloning of mitogen-and antigen-reactive B lymphocytes on filter paper disks: Phenotypic and genotypic analysis of B cell colonies. Methods in

Enzymology, 150, 287-305.

King, I.C. and Sartorelli, A.C. (1989) Ewing sarcoma growth factor as a supplement for the cloning of

hybridomas. J. Immunol. Methods 118, 297. Köhler, G. and Milstein, C. (1975) Continuous cultures of fused cells secreting antibody of predefined specificity. Nature 256,495.

Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4.

Nature 227, 680.

Lew, A.M. (1984) The effect of epitope density and antibody affinity on the ELISA as analysed by monoclonal antibodies. J. Immunol. Methods 72, 171.

Makela, O., Kaartinen, M., Pelkonen, J.L.T. and

Karjalainen, K. (1978) Inheritance of antibody specificity V. Anti-2-phenyloxazolone in the mouse. J. Exp. Med. 148,1644.

Martin, G.R. and Evans, M.J. (1975) Differentiation of clonal lines of teratocarcinoma cells: Formation of embryoid bodies in vitro. Proc. Natl. Acad. Sci. USA 72, 1441.

Micklem, L.R., McCann, M.C. and James, K. The use of rat mixed-thymocyte culture-conditioned medium for hybridoma production, cloning and revival. J. Immunol. Methods

104,81.

Nordan, R.P. and Potter, M. (1986) A macrophage-derived factor required by plasmacytomas for survival and

proliferation in vitro. Science 233,556.

Pintus, C, Ransom, J.H. and Evans, C.H. (1983)

Endothelial cell growth supplement: A cell cloning factor that promotes the growth of monoclonal antibody producing hybridoma cells. J. Immunol. Methods 61, 195.

Rathjen, D.A. and Geczy, C.L. (1986) Conditioned medium from macrophage cell lines supports the single-cell growth of hybridomas. Hybridoma 5, 255.

Rothstein, T.L. and Gefter, M.L. (1983) Affinity analysis of idiotype-positive and idiotype-negative Ars-binding hybridoma proteins and Ars-immune sera. Mol. Immunol. 20, 161.

Sedgwick, J.D. & Holt, P.G. (1983) J. Immunol. Methods, 57,301.

Sharon, J., Morrison, S.L., and Kabat, E.A. (1979)

Detection of specific hybridoma clones by replica

immunoadsorption of their secreted antibodies. Proc.

Natl. Acad. Sci. USA 76, 1420.

Skerra, A. and Pluckthun, A. (1988) Assembly of a

functional immunoglobulin Fv fragment in Escherichia coli. Science 240, 1038.

Suresh, M.R., Cuello, A.C. and Milstein, C. (1986)

Bispecific monoclonal antibodies from hybrid hybridomas. Methods in Enzymol. 121,210. van Heyningen, V., Brock, D.J.H. and van Heyningen, V. (1983) A simple method for ranking the affinities of monoclonal antibodies. J. Immunol. Methods 62, 147. van Snick, J., Cayphas, S., Vink, A., Vyttenhove, C., Coulie, P.G., Rubira, M.R. and Simpson, R.J. (1986)

Purification and NH₂-terminal .amino acid sequence of a T cell derived lymphokine with growth factor activity for B- cell hybridomas. Proc. Natl. Acad. Sci. USA 83, 9679.

Ward, E.S., Gussow, D.H., Jones, P.T. and Winter, G.P. (1989) Expression and secretion of repertoires of VH domains in Escherichia coli: Isolation of antigen binding activities. In Progress in Immunology, Vol. 7, in press .

Ward, E.S., Gussow, D.H., Griffiths, A.D. et al (1989) Nature 341, 544.

Winter, G. & Milstein, C. (1990) Nature, in press.

Claims

1. A method of identifying antibody-producing organisms or cells which produce a particular antibody of interest, comprising locating individual organisms or cells on a surface in spaced-apart relationship; growing the

organisms or cells under suitable conditions; applying to the organisms or cells after growth a filter layer having immobilised thereon a reagent which binds to the antibody of interest; and detecting antibodies of interest bound to the filter layer.

2. A method according to claim 1, wherein the antibody- producing organisms or cells are eukaryotic cells, eg mammalian cells.

3. A method according to claim 2, wherein the eukaryolic cells are grown in a feeder layer.

4. A method according to claim 3, wherein the feeder layer comprises a layer of soft agarose containing STO fibroblast feeder cells.

5. A method according to claim 3 or claim 4, wherein the feeder layer is covered with a diffusion layer.

6. A method according to claim 5, wherein the diffusion layer comprises agar.

7. A method according to claim 1, wherein the antibody- producing organisms or cells are prokaryotic organisms, such as bacterial cells.

8. A method according to claim 7, wherein the cells secrete antibody.

9. A method according to any one of the preceding claims, wherein the antibody is a bispecific antibody.

10. A method according to any one of the preceding claims, wherein the filter layer comprises nitrocellulose.

11. A method according to any one of the preceding claims, wherein antigen to the antibody of interest is immobilized on the filter layer.

12. A method according to any one of claims 1 to 10, wherein the filter layer is coated with anti- immunoglobulin.

13. A method according to any one of the preceding claims, wherein antibody bound to immobilized reagent is detected using an enzyme-labelled second antibody.

14. A method according to any one of the preceding claims, further comprising selecting and isolating organisms or cells producing the antibody of interest.

15. A method according to claim 14, further comprising growing the isolated organisms or cells.

16. A method according to any one of the preceding

claims, wherein varying amounts of antigen are immobilized on the filter layer to enable discrimination on the basis of antigen affinity of antibodies.

17. A method of selecting antibody-producing eukaryotic cells which produce a particular antibody of interest, comprising locating individual cells on a surface in spaced-apart relationship; growing the cells under

suitable conditions; applying to the cells after growth a filter layer having immobilized thereon a reagent which binds to the antibody of interest; detecting antibodies of interest bound to the filter layer; and selecting and isolating cells producing the antibody of interest.

18. A method of identifying antibody-producing

prokaryotic organisms which produce a particular antibody of interest, comprising locating individual organisms on a surface in spaced-apart relationship; growing the

organisms under suitable conditions; applying to the organisms after growth a filter layer having immobilized thereon a reagent which binds to the antibody of interest; and detecting antibodies of interest bound to the filter layer.