Characterization of HLA-reactive antibody in sensitized patients who require transplantation or platelet transfusion is paramount in the selection of acceptable HLA-mismatched donors. Conventional HLA antibody screening methods, such as complement-dependent cytotoxicity (CDC) (1) and flow cytometric analysis (2), use whole lymphocytes as targets for the antibody. Recently, ELISA-based antibody detection methods using isolated class I and II antigens from platelets (3, 4) and cell lines (5) have also been described. All of these methods, however, are constrained by the co-expression of different HLA classes and products of different loci, and this can be particularly confounding during dissection of component specificities in sera from highly sensitized patients. Linkage disequilibrium between the different HLA loci may also hinder the assignment of such specificity.
The construction of biotinylated recombinant HLA/peptide complexes offers a novel means of addressing this problem. These complexes (6) have been used in the form of tetramers for the detection of MHC/peptide-specific cytotoxic T cells (6, 7) by flow cytometric method.
Here, we demonstrate that HLA/peptide monomers can be effectively used in a novel ELISA assay, monoLISA, for the detection of HLA antibodies. In this assay, biotinylated HLA class I monomers are conjugated to streptavidin-coated microtiter plates in a “one-well, one-antigen” format, allowing absolute correlation of reactivity with specificity.
MATERIALS AND METHODS
Test sera were separated from the clotted blood of 100 transplantation and pretransplantation renal patients (test sera), and the negative control for both assays was from an untransfused blood group AB male. All patients were tested and were negative for human immunodeficiency virus (HIV), and sera from patients positive for hepatitis C virus (HCV) were excluded from this study. All sera were stored at −20°C.
Recombinant HLA molecules.
Biotinylated HLA-peptide monomers were synthesized as described previously (6) but were not tetramerized. Briefly, purified HLA heavy chain and β2 microglobulin were synthesized using a prokaryotic expression system (pET; Novagen, Milwaukee, WI). The heavy chain was modified by deletion of the transmembrane/cytosolic tail and COOH-terminal addition of a sequence containing the BirA enzymatic biotinylation site. Heavy chain, β2 microglobulin, and peptide were refolded by dilution. The 45-kD refolded product was isolated using fast protein liquid chromatography and biotinylated by BirA (Avidity, Denver, CO) in the presence of biotin (Sigma Chemical Co., St. Louis, MO), ATP (Sigma Chemical Co.), and Mg (2) (Sigma Chemical Co.). The biotinylated product was separated from the free biotin by gel filtration and ion exchange using fast protein liquid chromatography. The monomers used (6, 8) had been refolded with the peptides (9–14) listed in Table 1. The use of monomers refolded with HIV-and HCV-derived peptides avoided the possible confounding contribution of antibody specific for the peptide rather than the alloantigen.
All volumes were per well. We incubated 50 μl HLA monomer at approximately 0.5 ng·μl−1 in a streptavidin-coated, 96-well “Combiplate” (Labsystems, Finland) at 22°C for 30 min. The plate was washed 4 times with 200 μl wash buffer (PBS with 0.05% Tween 20). Serum, diluted 1:20 in dilution buffer (wash buffer with 5% skimmed milk powder, Tesco Ltd, UK), was incubated at 22°C for 30 min, then washed 3 times as before. Each well was incubated for 1 hr at 22°C with 100 μl of sheep anti-human IgG antiserum conjugated to horseradish peroxidase (Serotec, UK) diluted 1:10000 in dilution buffer and then washed 3 times as above. We added 100 μl 1 mg·ml−1 ortho-phenylenediamine dihydrochloride (Sigma, UK) solution in phosphate-citrate buffer with sodium perborate (Sigma, UK) and incubated for 15 min at 22°C in the dark. The reaction was stopped with 100 μl of 1N HCl and the absorbance (A) evaluated at 490 nm with a reference wavelength of 630 nm using a Dynatech MRX plate-reader. Results were corrected (δA) for nonspecific binding (background) by subtracting the absorbance of the blank well, which contained serum but no antigen, (A no antigen) from that of the test well (A test). Preliminary data (not shown) led to the assignment of an arbitrary cut-off value for δA of 0.5. To standardize the concentration of different monomers, the above method was performed identically, except that the mouse monoclonal, W6/32 (Serotec, UK) at a concentration of 1:1000 in dilution buffer, was substituted for the patient’s serum. Consequently, a rat anti-mouse IgG2a monoclonal conjugated to horseradish peroxidase (Serotec, UK), also at 1:1000 in dilution buffer, was used in place of the anti-human conjugate. The monomer was bound at a range of doubling dilutions, and the concentration giving a δA closest to a standard value (3 absorbance units) was chosen as the working concentration for that monomer.
Comparison of monoLISA with CDC.
To determine the sensitivity and specificity of the monoLISA assay with respect to the established standard, CDC, the new method was applied to 85 sera from renal dialysis or transplant patients. The sera were selected for the presence of immunoglobulin (Ig)G of the following specificities as determined by CDC: 36 anti-HLA-A2 sera, 28 anti-B8 sera and 25 sera showing either no alloreactivity (cytotoxic negativity), or alloreactivity to other class I molecules (irrelevant antibody). To ensure that the antibody did not bind to the peptide itself, all sera were drawn from HIV- and HCV-negative patients. The specificities are given in Table 2. Four of the sera with relevant IgG specificity contained antibodies with specificity for both HLA-A2 and -B8. Two different monomer/peptide combinations were used separately as target antigens with each of the sera; A2/gag and B8/HCV. The sequences of the peptides, which were derived from HIV and HCV proteins, respectively, are shown in Table 1.
Table 2 shows the CDC-defined specificity and δA for each of the 85 sera after testing against the A2/gag and B8/HCV monomers. These same data are presented graphically as categorical scatter plots in Figures 1a. and 1b. respectively. Briefly, 4/85 CDC-A2-negative sera were positive with the A2 monomer, whereas no CDC-A2-positive sera were negative with A2 monomer. The remaining 81 sera were concordant between the two methods, with 34 double positive sera and 47 double negatives. With the B8 monomer, there were 4 CDC-B8-ve/monoLISA+ve sera and, again, no CDC+ve/monoLISA-ve serum, with 27 double positive and 54 double negative sera. The monoLISA test using the A2 monomer exhibited 100% sensitivity and 92% specificity compared with CDC. Similarly, the B8 monoLISA attained 100% sensitivity and 93% specificity.
The influence of presented peptide on anti-HLA/monomer binding.
To investigate whether the presence of different peptides presented in the monomer would exert any effect on the strength of antibody binding, we used 4 recombinant HLA/peptide complexes comprising identical monomers but different peptides. This panel was available for HLA-A*1101and included 2 HIV-derived peptides (nef and pol) and 2 Epstein-Barr virus (EBV)-derived peptides (EBV1 and EBV2;Table 1). These monomer/peptide combinations were applied to sera containing antibodies against HLA-A11 as detected by CDC. To minimize the interference of any possible specific humoral response directed at the peptides, the peptides were derived from pathogens which were either absent from the patient population (HIV) or ubiquitous (EBV). Nine sera from sensitized patients were selected for the presence of anti-A11 antibodies. A negative control serum was also used for the absence of HLA-reactive Ig . These sera were reacted in duplicate with each of the 4 different A11 monomer/peptide combinations. Each of the monomer/peptides was also standardized with W6/32 in the same plates so that correction could be made for slight variations in actual monomer concentrations.
The results are shown in Figure 2. Each of the 10 sera gave grossly similar results with the different monomer/peptide combinations, and all sera had significantly higher δA values than the negative control (P <0.001, paired T). This demonstrates that bound peptide did not significantly interfere with reactivity in the monoLISA test.
The influence of glycosylation on anti-HLA/monomer binding.
Prokaryotic expressions systems such as Escherichia coli are potentially restricted by their inability to glycosylate proteins (15, 16). Because native HLA class I heavy chains have an N-linked carbohydrate moiety at asparagine 86 it is possible that antibodies specific for epitopes in proximity with this residue may have differential binding characteristics with recombinant monomers compared with native protein. The Bw4 and Bw6 motifs, present on all expressed HLA-B locus antigens, constitute a well-defined operationally dimorphic system (17). The residues responsible for these motifs are at positions 77–83 on the class I heavy chain. This region of the heavy chain sequence constitutes the most proximal known epitope to the glycosylation site. Anti-Bw6 reactivity was thus used to determine whether the absence of carbohydrate had a measurable effect on the binding of alloantibodies. Antisera of the Bw6 specificity were tested with monoLISA using the two monomers- B7/EBV and B8/HCV as Bw6 motif-bearing targets. Bw4 motif-bearing monomers were not available for testing. Antisera from 5 patients with CDC-defined anti-Bw6 reactivity, one AB serum and one Bw4-reactive serum were tested in duplicate against the two monomer/peptide combinations. Standardization between the two monomers was performed by measuring W6/32 reactivity as above. The sensitization details of the selected sera are given in Table 3.
The results are shown in Figure 3. All the sera reacted above the arbitrary cut-off δA of 0.5, and the negative controls both reacted at a lower level than the no-monomer wells. Variation in binding levels between the different sera was apparent, and with the exception of serum BC, the results obtained with the two monomers followed the same trend. These results show that the lack of a carbohydrate moiety did not reduce the binding of antisera to a very close epitope, i.e., the Bw6 motif. Indeed, the absorbance values obtained were mostly high in the range of values seen using sera against other epitopes.
The results of this study clearly demonstrate that HLA class I monomers can be used as a target for the detection of anti-HLA antibody in alloantisera. The potential benefits include objectivity and the ability to quantify antibody in a highly reproducible manner. In addition, this test uniquely ensures that the only target present is an HLA antigen, and that it is the only HLA antigen present. This one-well, one-antigen approach could prove extremely useful in the alloantibody screening of very highly sensitized patients. This is of particular importance in the identification of acceptable mismatches, which typically require serum screening against a panel of lymphocytes from several hundred donors. Using monoLISA should theoretically allow unequivocal definition of acceptable mismatches using a single microtitre plate assay. Furthermore, such plates could be prepared in advance, reducing assay time per serum from several weeks, in the case of CDC, to less than 3 hr.
This study demonstrates an excellent correlation between the techniques of CDC and monoLISA. With both the A2 and B8 monomers tested, there were positive reactions which had previously been negative by CDC. These reactions, which tended to be relatively weak, are to be expected in a system that is designed to detect IgG isotype-binding and not just the presence of cytotoxic antibodies. This phenomenon has previously has been reported using the ELISA method, PRA-STAT, which detected HLA-specific IgG antibodies relevant to transplant outcome that were not detected by CDC (18). More important, in our study, all the sera that showed positivity in the CDC test were also positive using monoLISA. This is a good demonstration that, at least using the sera tested, there was no abrogation of the antibody/antigen interaction using recombinant molecules in place of the natural ligand.
The use of monomers with identical HLA antigens, but different peptides within the groove offered an opportunity to analyze the importance of the peptide in alloantibody responses. Using a limited number of monomers of the same HLA specificity (A11), but with different peptides, we have demonstrated that the peptide did not have a large impact on alloantibody binding. This is an important observation, because it may suggest that single monomers with irrelevant peptides within the groove can be used for antibody screening in this assay. Clearly, more data are required concerning the influence of bound peptide on alloantibody binding because the combinations presented here constitute a very small sample of all those possible. The peptides used were those already refolded into the available monomers, and no attempt was made to displace them with other peptides that may have been more relevant to this assay. With further testing, it may become apparent that careful attention to peptide design is required on an individual basis for each monomer.
The reactivity exhibited by the two Bw6 bearing monomers suggests that the Bw6 motif can be detected using the monoLISA method. Because the putative Bw6 motif is situated close to the carbohydrate-bearing amino acid on HLA class I, it is suggested that lack of glycosylation of the monomers is not detrimental to binding, and thus detection, of antibody. The possibility, however, that the selected sera contained not only anti-Bw6 reactivity, but also antibodies reacting with B7- and B8-specific motifs that are spatially separate from the putative Bw6 region cannot be ruled out. However, this is unlikely, given that similar strengths of reactivity are demonstrated against both molecules with each individual serum. The exception is serum BC. With this serum, the monomers reacted differentially, the B8 eliciting approximately double the reactivity produced with B7. This disparity in reactivity can be explained, because the patient was sensitized with a graft bearing both B7 and B8. A later serum from the same patient showed a decrease in PRA that allowed the two specificities, B7 and B8, to emerge from the broader Bw6 specificity (data not shown). Presumably, the graft-presented B8 elicited a stronger response than did B7, to which the patient was exposed on the graft and in previous pregnancies.
The data presented here are based on the use of four class I monomers. Obviously, a large amount of work is required to synthesize all the remaining alleles, but we believe it possible, in the first instance, to significantly reduce the number needed for complete class I coverage. If one considers the minimum requirement for class I screening to be that each polymorphic residue is represented at least once on the panel, then a carefully selected set of 33 antigens fulfils this criterion (Table 4). This hypothesis is supported by data obtained using the Flowscreen technique (19), in which flow cytometric analysis of two pools of 12 cells is sufficient to detect most or all alleles. We predict that any HLA-reactive serum will bind to one or more of this smaller set of antigens and that reactivity with untested specificities could be inferred from the smaller set.
Ideally, the monoLISA assay would be applied to the screening of class II antibodies in addition to the class I method demonstrated here. At present, however, the synthesis of class II monomer has not been reported, although their construction is imminent. Given successful construction of these recombinant molecules, their efficacy in the assay would need to be tested as for class I.
The assay presented identifies the presence of HLA-reactive IgG. However, HLA antibodies of the IgM class have also been shown to be associated with poorer graft outcome (20). Although not tested in this study, the monoLISA assay could be adapted to include an IgM-specific conjugate, either as a separate reaction or together with the anti-human IgG to detect these antibodies.
This work demonstrates that recombinant monomeric HLA/peptide complexes may be useful for accurate definition of anti-HLA antibodies, particularly in individuals with broad panel reactivity. The method presented was tested on few alleles using a small number of sera, and the preliminary findings suggest a possible application to patient screening for solid organ transplantation or platelet transfusion. Before implementation however, synthesis of further monomers and subsequent testing against a larger number of sera will be required. The number of alleles that need to be synthesized would be considerably less than the allele count because there is redundant expression of most epitopes on more than one antigen.
MonoLISA could also potentially fulfil a role as an international standard source of anti-HLA screening targets. Once constructed, the limitless availability of monomer would prove particularly useful in the case of rarer antigens in the various populations. In addition, the monoLISA method could be used in an assay for direct allo-presentation to T cells as well as for allopeptide presentation. Certainly, the potential of the technique shown by this work is sufficient for a collaborative approach for monomer production to be undertaken.
The authors would like to thank Paul Klenerman, Pokrath Hansasuta, Sarah Rowland-Jones, and Gerry Gillespie for their generous donation of monomers, and Sara Marshall for help in the preparation of the manuscript.
1. Terasaki PI, McClelland JD. Microdroplet assay of human serum cytotoxins. Nature 1964; 204:998.
2. Harmer AW, Sutton M, Bayne A, Vaughan RW, Welsh KI. A highly sensitive, rapid screening method for the detection of antibodies directed against HLA class I and II antigens. Transpl Int 1993; 6:277.
3. Carroll K, Lannon B, O’Kennedy R. Optimal fixation of cells for use in solid-phase ELISA. J Immunol Methods 1990; 129:71.
4. Kao KJ, Scornik JC, Small SJ. Enzyme-linked immunoassay for anti-HLA antibodies: an alternative to panel studies by lymphocytotoxicity. Transplantation 1993; 55 (1):192.
5. Buelow R, Chiang TR, Monteiro F, et al. Soluble HLA antigens and ELISA: a new technology for panel reactive antibody and crossmatch testing. Transplantation 1995; 60 (12):1594.
6. Altman JD, Moss PAH, Goulder PJR, et al. Phenotypic analysis of antigen-specific T lymphocytes [published erratum appears in Science 1998 Jun 19; 280(5371): 1821]. Science 1996; 274 (5284):94.
7. Ogg GS, Dunbar RP, Romero P, Chen JL, Cerundolo V. High frequency of skin-homing melanocyte-specific cytotoxic T lymphocytes in autoimmune vitiligo. J Exp Med 1998; 188 (6):1203.
8. Reid SW, McAdam S, Smith KJ, et al. Antagonist HIV-1 Gag peptides induce structural changes in HLA B8. J Exp Med 1996; 184 (6):2279.
9. Parker KC, Bednarek MA, Hull LK, et al. Sequence motifs important for peptide binding to the human MHC class I molecule, HLA-A2. J Immunol 1992; 149 (11):3580.
10. Hill A, Worth A, Elliott T, et al. Characterization of two Epstein-Barr virus epitopes restricted by HLA-B7. Eur J Immunol 1995; 25 (1):18.
11. Koziel MJ, Dudley D, Afdhal N, et al. HLA class I-restricted cytotoxic T lymphocytes specific for hepatitis C virus: identification of multiple epitopes and characterization of patterns of cytokine release. J Clin Invest 1995; 96 (5):2311.
12. Couillin I, Culmann Penciolelli B, Gomard E, et al. Impaired cytotoxic T lymphocyte recognition due to genetic variations in the main immunogenic region of the human immunodeficiency virus 1 NEF protein [see comments]. J Exp Med 1994; 180 (3):1129.
13. Goulder P, Price D, Nowak M, Rowland Jones S, Phillips R, McMichael A. Co-evolution of human immunodeficiency virus and cytotoxic T-lymphocyte responses. Immunol Rev 1997; 159:17.
14. Gavioli R, Kurilla MG, de Campos Lima PO, et al. Multiple HLA A11-restricted cytotoxic T-lymphocyte epitopes of different immunogenicities in the Epstein-Barr virus-encoded nuclear antigen 4. J Virol 1993; 67 (3):1572.
15. Huth JR, Norton SE, Lockridge O, Shikone T, Hsueh AJ, Ruddon RW. Bacterial expression and in vitro folding of the beta-subunit of human chorionic gonadotropin (hCG beta) and functional assembly of recombinant hCG beta with hCG alpha. Endocrinology 1994; 135 (3):911.
16. Wang LF, Hum WT, Kalyan NK, Lee SG, Hung PP, Doi RH. Synthesis and refolding of human tissue-type plasminogen activator in Bacillus subtilis
. Gene 1989; 84 (1):127.
17. Parham P. Typing for class I HLA polymorphism: past, present and future. Eur J Immunogen 1992; 19 (5):347.
18. Worthington JE, Thomas AA, Dyer PA, Martin S. Detection of HLA-specific antibodies by PRA-STAT and their association with transplant outcome. Transplantation 1998; 65 (1):121.
19. Sutton PM, Harmer AH, Bayne AM, Welsh KI. The flow cytometric detection of alloantibodies in screening for renal transplantation. Transpl Int 1995; 8 (5):360.
20. Taylor CJ, Chapman JR, Ting A, Morris PJ. Characterization of lymphocytotoxic antibodies causing a positive crossmatch in renal transplantation: relationship to primary and regraft outcome. Transplantation 1989; 48 (6):953.