Antibody arrays for high-throughput screening of antibody-antigen interactions.

We have developed a novel technique for high-throughput screening of recombinant antibodies, based on the creation of antibody arrays. Our method uses robotic picking and high-density gridding of bacteria containing antibody genes followed by filter-based enzyme-linked immunosorbent assay (ELISA) screening to identify clones that express binding antibody fragments. By eliminating the need for liquid handling, we can thereby screen up to 18,342 different antibody clones at a time and, because the clones are arrayed from master stocks, the same antibodies can be double spotted and screened simultaneously against 15 different antigens. We have used our technique in several different applications, including isolating antibodies against impure proteins and complex antigens, where several rounds of phage display often fail. Our results indicate that antibody arrays can be used to identify differentially expressed proteins.

Synthesis of Rh Fv phage-antibodies using VH and VL germline genes.

Antibodies to the D antigen of the Rh system use a restricted set of immunoglobulin V and J gene segments, especially VH DP50 and DP63, JH6, Vlambda DPL16 and Jlambda 2/3. These gene segments may confer a natural affinity on the antibodies for the D antigen and this hypothesis has been tested by constructing two single-chain Fv phage-antibody libraries based on the germline gene segments DP50 and DP63; structural variability was obtained by insertion of 11 amino acids in random sequence in the VHCDR3. 10 anti-D antibodies were selected from these libraries, each with a unique VHCDR3. In contrast, selections with the CcEe antigens produced antibodies reacting with the Rh polypeptide molecules but without strict blood group specificity. One of these latter DP50-based antibodies was converted into 12 different antibodies with specificity for E by replacing the original germline light chain with chains from a rearranged L chain library. The CDR1 and CDR2 sequences of the DP50-based antibodies were common to both anti-D and anti-E molecules; differentiation between D and E specificity was dependent on VHCDR3 sequences and their correct pairing with an appropriate L chain.

Prediction of the severity of haemolytic disease of the newborn. Quantitative IgG anti-D subclass determinations explain the correlation with functional assay results.

Sera containing anti-D, taken from 44 RhD-negative women with RhD-positive infants, were tested in antibody-dependent cellular cytotoxicity (ADCC) and monocyte monolayer assays (MMA) which used similar target and effector cell populations. In addition, the anti-D concentration was measured in the Auto Analyzer and the number of IgG1 and IgG3 anti-D molecules bound to the target red cells was measured by flow cytometry. The results of the functional assays and Auto Analyzer quantitation were examined for correlation with IgG subclass quantitation and all results were compared for their ability to predict the severity of haemolytic disease of the newborn (HDN). ADCC correctly predicted HDN in 39/44 (88.6%) cases, Auto Analyzer quantitation in 35/44 (79.5%) and the MMA in 32/44 (72.7%). For all three assays, the number of correct predictions was highest when the maternal serum contained both IgG1 and IgG3 anti-D. ADCC activity and HDN were correlated with the number of cell-bound IgG1 molecules (r > or = 0.58), but MMA activity was most closely correlated with the number of cell-bound IgG3 molecules (r = 0.68). Hence the superior predictive value of ADCC is due to its ability to reflect the IgG1 component of maternal anti-D, which has a better correlation than IgG3 anti-D with the severity of HDN.

Characterization of human blood group scFv antibodies derived from a V gene phage-display library.

We previously reported the initial characterization of five human single-chain Fv (scFv) antibody fragments specific for the blood group antigens B, D(Rh), E(Rh), Kpb and HI. The scFvs were isolated from a phage-antibody library constructed from the variable region genes of two non-immunized donors. In this paper we report the specificity, affinity and kinetics of antigen binding of these scFv fragments. All five scFvs agglutinated the appropriate red cell phenotype following the addition of a monoclonal antibody which recognizes a peptide tag incorporated into the scFv. The anti-B and anti-HI scFv molecules, which recognize high density carbohydrate antigens, spontaneously polymerized and agglutinated red cells directly. None of the antibody fragments showed cross-reactivity with other red cell antigens, with the exception of the anti-E which reacted weakly with E-negative cells. Specific scFv binding was confirmed by ELISA, flow cytometry and radioactive labelling. The anti-D scFv recognized 17,600 sites on cDE/cDE red cells with an association constant (Ka), of 5.2 x 10(7) M-1 and a rate constant for dissociation (koff) of 1.9 x 10(-2) s-1. The anti-E scFv recognized 29,800 and 39,800 sites on cDE/cDE red cells in two experiments with Kas of 8.4 x 10(6) and 4.4 x 10(7) M-1. The koff for this antibody was 2.7 x 10(-2) s-1. The results demonstrate that scFv antibody fragments specific for cell surface antigens and possessing affinities typical of the primary immune response can be obtained from a phage-display library.

Human antibody fragments specific for human blood group antigens from a phage display library.

We have isolated human antibody fragments with binding specificities against the blood group antigens of the ABO and I blood group systems (B and HI), of the Rh system (D and E) and of the Kell system (Kpb), by selecting a phage-antibody library on human red cells. The fragments, expressed in bacteria, were antigen-specific and effective in assays including agglutination and immunohistochemistry. Selection of phage-antibody libraries using intact cells seems to offer a promising alternative to hybridoma technology for the production of antibodies against cell surface molecules.

Human anti-self antibodies with high specificity from phage display libraries.

Recently we demonstrated that human antibody fragments with binding activities against foreign antigens can be isolated from repertoires of rearranged V-genes derived from the mRNA of peripheral blood lymphocytes (PBLs) from unimmunized humans. The heavy and light chain V-genes were shuffled at random and cloned for display as single-chain Fv (scFv) fragments on the surface of filamentous phage, and the fragments selected by binding of the phage to antigen. Here we show that from the same phage library we can make scFv fragments encoded by both unmutated and mutated V-genes, with high specificities of binding to human self-antigens. Several of the affinity purified scFv fragments were shown to be a mixture of monomers and dimers in solution by FPLC gel filtration and the binding kinetics of the dimers were determined using surface plasmon resonance (k(on) = 10(5)-10(6) M-1s-1, k(off) = 10(-2)s-1 and Ka = 10(7) M-1). The kinetics of association are typical of known Ab-protein interactions, but the kinetics of dissociation are relatively fast. For therapeutic application, the binding affinities of such antibodies could be improved in vitro by mutation and selection for slower dissociation kinetics.

Germline variable region gene segment derivation of human monoclonal anti-Rh(D) antibodies. Evidence for affinity maturation by somatic hypermutation and repertoire shift.

To date, there has been no systematic study of the process of affinity maturation of human antibodies. We therefore sequenced the variable region genes (V genes) of 14 human monoclonal antibodies specific for the erythrocyte Rh(D) alloantigen and determined the germline gene segments of origin and extent of somatic hypermutation. These data were correlated with determinations of antibody affinity. The four IgM antibodies (low affinity) appear to be derived from two germline heavy chain variable region gene segments and one or two germline light chain variable region gene segments and were not extensively mutated. The 10 IgG antibodies (higher affinity) appear to be derived from somatic hypermutation of these V gene segments and by use of new V gene segments or V gene segment combinations (repertoire shift). Affinity generally increased with increasing somatic hypermutation; on average, there were 8.9 point mutations in the V gene segments of the four IgM antibodies (Ka = 1-4 x 10(7)/M-1) compared with 19 point mutations in the V gene segments of the 10 IgG antibodies. The four highest affinity antibodies (Ka = 0.9-3 x 10(9)/M-1) averaged 25.5 point mutations. The use of repertoire shift and somatic hypermutation in affinity maturation of human alloantibodies is similar to data obtained in inbred mice immunized with haptens.

Differential exposure of epitopes on the human red cell antigen D (Rh).

The D polypeptide of the human Rh blood group system has a number of different epitopes on its surface. It is also known that there is considerable variation in the number of D antigen sites available to different human monoclonal anti-D antibodies. For instance, certain monoclonal antibodies recognise only a small number of sites on the red cell surface (about 9,000 sites/red cell on R1R2 cells) whereas other antibodies recognise a high number (about 20,000-30,000 sites/red cell). It has been found that cholesterol enrichment of the red cell membrane increases the number of sites available to those antibodies which recognize only a few sites but has no effect on those recognising many sites. The results are consistent with the view that access to some of the D epitopes is partially hindered by neighbouring molecules in the membrane and that alteration of the lipid content of the membrane changes it in such a way as to allow increased access to these obstructed epitopes.

Relative functional binding activity of IgG1 and IgG3 anti-D in IgG preparations.

The intention to replace polyclonal IgG anti-D with human monoclonal antibody in the prophylaxis of haemolytic disease of the newborn requires knowledge concerning the relative content of IgG1 and IgG3 anti-D in prophylactic IgG preparations that are in present use. This has been carried out using a functional assay in which the absolute amount of IgG1 and IgG3 anti-D present on red cells was determined after incubation with IgG preparations. The assay was carried out by flow cytometry on 17 samples; expressed as a percentage of the total, the average value for the amount of IgG3 anti-D on the cells was 8% (range 1-18%). Similar measurements were also made on the anti-D present in 18 samples of antisera; IgG3 anti-D formed a larger fraction of the total, the average value being 17% (range 0-60%) confirming previously reported estimates. It is suggested that some of the low values found for IgG3 in IgG preparations may be due do preferential loss during production.

Multiple epitopes recognised by human monoclonal IgM anti-D antibodies.

Bivalent 7S subunits were obtained from 7 purified IgM human monoclonal anti-D antibodies with 33-63% of the total protein-retaining functional binding activity. The number of sites per red cell recognized by these subunits ranged between 9,400 and 28,500. The values are taken to indicate that a number of different epitopes on the D polypeptide are being recognized. There was competition between IgG and IgM antibodies for binding to the red cells, indicating that both classes of antibody are recognizing epitopes on the same D polypeptide. The value of the functional affinity constants for the binding of the 7S subunits varied between 1.0 and 8.8 x 10(7) M-1 and there was a 3- to 16-fold increase on reduction of the ionic strength (0.05 M NaCl). Using agglutination in microwells, the end-point of the titres for the native pentamer IgM antibodies occurred at concentrations in the range 7-26 ng/ml, with one exception where the concentration required was 165 ng/ml.

Comparison of the reactions of the Rh-related murine monoclonal antibodies BS58 and R6A.

The murine monoclonal antibodies BS58 and R6A are known to recognize epitopes related to the human Rh system: neither antibody reacts with Rhnull cells and the BS58 antigen is not expressed by -D- or .D. cells. It is shown here that the numbers of BS58 and R6A antigen sites vary with Rh phenotype. Both epitopes are well represented on cells of the CDe/CDe, CDe/cDE and CDe/cde phenotypes; BS58 sites are markedly reduced on cde/cde and cDE/cde and are only just detectable on cDE/cDE cells when compared with R6A sites. The number of R6A sites per red cell ranged between 20,000 and 150,000. The evidence indicates that the BS58 epitope is not on the polypeptides carrying D or R6A, nor is it uniquely on one of the polypeptides carrying either C, c, E or e. It is suggested that the BS58 epitope is either common to all the CcED polypeptides or that it is present on a polypeptide which has not yet been identified biochemically.

Critical dependence on antibody for defence against mycoplasmas.

We have shown that mycoplasmas bind spontaneously to neutrophils, as well as directly activating the first component of complement with probable subsequent binding through neutrophil complement receptors. Thus, antibody is not required to opsonize mycoplasmas for neutrophil phagocytosis. Furthermore, mycoplasmas remain viable when phagocytosed in the absence of antibody and may be carried inside neutrophils to various parts of the body to cause infection (e.g. joints). We found that antibody alone inhibits the growth of mycoplasmas in vitro. These observations suggest that the role of antibodies is to control the growth of mycoplasmas on mucosal surfaces; neutrophils play no part in defence and may even aid dissemination of the infection. This helps to explain why patients with hypogammaglobulinaemia are prone to systemic mycoplasma infections.

Three epitopes on the human Rh antigen D recognized by 125I-labelled human monoclonal IgG antibodies.

Seven purified monoclonal antibodies specific for the D antigen of the human Rh blood group system were examined for the characteristics of their reactions with red cells (phenotype CcDEe). The average number of sites available for binding to the antibodies ranged from 8,900 to 26,000/cell. In mutual inhibition studies, all the antibodies inhibited each other but the extent of inhibition varied in that antibodies recognizing a lower number of sites only partially inhibited those recognizing a higher number of sites. It was concluded from the evidence that these six monoclonals recognize at least three different epitopes on the D peptide. In order to explain the variation in the number of epitopes on each polypeptide, it is suggested that there is heterogeneity in the placement of the molecule in the red cell membrane resulting in variation in access to the epitopes.

Radio-immunoassay of the functional activity of anti-D (Rh) preparations using a human monoclonal 125I-labelled anti-D.

The prophylactic effect of anti-D in the prevention of maternal immunization to the D antigen of the Rh system depends in the first instance on its ability to bind to fetal Rh-positive red cells. The functional activity of the injected anti-D in this respect is dependent on both the plasma concentration and functional affinity constant of the antibody, and both factors should be taken into account in assessing IgG anti-D preparations for clinical use. A radio-immunoassay utilizing 125-I-labelled human monoclonal anti-D has been developed which measures the amount of antibody bound to red cells after the addition of an aliquot of the test anti-D preparation. The assay is thus a functional test in that it measures the ability of the anti-D to react with the antigen. The functional activity of the test anti-D is compared to that of the international standard, and the concentration is then expressed in terms of microgram-equivalents, that is, its equivalence in functional terms to the stated amount of international standard. The radio-immunoassay was compared to the currently used agglutination assay, and the results were found to give good agreement in 18 out of the 21 samples assayed.

Production of human monoclonal IgG and IgM antibodies with anti-D (rhesus) specificity using heterohybridomas.

Heterohybridomas secreting human IgM and IgG anti-D antibodies of the rhesus blood group system have been established by fusion of EBV-transformed anti-D secreting cells with the mouse myeloma cells X63-Ag8.653. Both classes of antibody reacted with all Rh-positive cells, some Du cells but not with Rh-negative or DB cells. Concentrations of both antibodies reached between 25 micrograms/ml and 50 micrograms/ml in the culture supernatants. The cell lines have been maintained in culture for 14 months and have been shown to be suitable for large-scale production of antibody.

Antibody density on rat red cells determines the rate of activation of the complement component C1.

It is a common observation that there is variability in the rate of activation of C1, the first component of complement, when bound to immune complexes. The cause of this variation has been investigated with experiments designed to assess separately the effect of antibody, antigen and C1 density. Using 125I-labeled C1 and a rat monoclonal antibody specific for the class I antigen, it has been found that the rate of activation is primarily dependent on antibody density on the cell surface and not on antigen or C1 density. This finding supports the suggestion that direct contact between the C1r2C1s2 subcomponent of C1 and antibody may be required for potentiation of C1 activation.

IgG pair formation on one antigenic molecule is the main mechanism of synergy between antibodies in complement-mediated lysis.

Two rat IgG2b monoclonal antibodies bound to different epitopes on a offgle RT1Aa antigen molecule synergize in their induction of complement-mediated lysis of the red cells. The mechanism of this synergistic action has been investigated by determining the C1q:IgG relationship on red cells using the fluorescence-activated cell sorter and making use of the considerable variation in antigen density between individual red cells. With a synergistic pair of antibodies, the C1q:IgG ratio was approximately 1 and independent of antibody density per cell, indicating that the antibody molecules are present as closely associated pairs, each pair binding two C1q molecules and thus forming a cyclic tetramer. When one of the antibodies was used alone, the number of C1q molecules bound relative to the amount of antibody was higher than expected from theoretical considerations based on a presumed random distribution of antigen-antibody complex. This result can be explained if there is some mobility of the complex within the membrane. The interpretation of the results gives strong support to the hypothesis that C1q must bind bivalently to rat IgG2b complexes.

The mechanism of synergistic complement-mediated lysis of rat red cells by monoclonal IgG antibodies.

The mechanism of synergistic complement-mediated lysis of rat red cells was investigated using rat monoclonal antibodies against class I RT1Aa antigens. The increased lytic activity when using two antibodies simultaneously is due to the increase in the number of activated C1 molecules on the cell surface and this results from (a) an increase in the number of binding sites for C1q, (b) an increase in the functional affinity constant for C1q binding and (c) an increase in the rate of activation of C1. Complete lysis of red cells was only achieved if one member of the synergistic pair was of the gamma 2b isotype, and this isotype was the only one to which binding of 125I-labeled C1q could be detected. A partial synergistic effect was seen using an F(ab')2 fragment of antibody. Increased uptake and activation of C1 probably results both from the presence of two antibodies attached to each antigen molecule and from the formation of antigen-antibody catenars.