Since 1971, when the first panel-reactive antibody (PRA) tests were performed (1), we have been limited in identifying specificities in human leukocyte antigen (HLA) antisera by the fact that the sera had to be tested against whole cells containing multiple specificities of the A, B, C, and DR loci, or antigen extracts, which also contained multiple specificities that had been isolated from cells. In many cases, especially for high PRA sera, the specificities could not be defined by these tests. The ideal test would have a single antigen per reaction. In 1979, Parham purified HLA A2 and HLA B7 antigens by reacting cell extracts with monoclonal antibodies (mAbs) attached to affinity columns (2). We and other investigators (3–9) have used this method to produce antigens for antibody testing. However, the method is limited by the number of specific mAbs available and by the problem of reactions of different antigens with shared common epitopes. In 2000, 20 years later, Barnardo et al. described the production of HLA A2 and B8 using recombinant antigens produced in Escherichia coli (10). They showed that the A2 and B8 antigens produced would react with antisera against these specificities. They concluded that glycosylation was not necessary in producing antigens to be used for antibody characterization. However, this may not be generally true for all HLAs, because posttranslational modifications are essential for the function of many of the glycoproteins. In addition, the unfolded HLA produced in the E. coli expression system requires a refolding process that involves the association of exogenous beta-2 microglobulin to obtain correctly folded, functional recombinant HLA protein. As another approach, we developed a mammalian cell expression system of HLA-transfected cells. This system provides all of the crucial elements, including posttranslational modification, to produce mature glycosylated HLAs that are indistinguishable from native HLA class I antigen. By using this system, we describe the construction of a comprehensive panel of 110 single antigens and their application to a flow cytometry assay for detection of HLA antibodies. By examining high PRA sera reacting to more than 90% of the regular cell panel, we show that the single antigen panel facilitates the accurate assignment of HLA antibody specificities.
MATERIALS AND METHODS
Cloning and Obtaining Human Leukocyte Antigen Transfectants
Complementary DNAs (cDNAs) encoding specific HLA class I were generated by reverse transcription of mRNA isolated from Epstein-Barr virus (EBV) cell lines. The cDNA was amplified with polymerase chain reaction (PCR) with Taq polymerase (Invitrogen, Carlsbad, CA) using HLA class I specific primer pairs and cloned into an expression vector driven by the Rous sarcoma virus promoter (11). The resultant HLA cDNA clones were verified by sequence analysis and transfected by electroporation into a human cell line that lacks expression of HLA class I antigens (12). All cells were maintained in Roswell Park Memorial Institute 1640 medium (Irvine Scientific, Santa Ana, CA) supplemented with 15% fetal calf serum (Irvine Scientific). For transfection, 5×106 host cells and 20 μg of plasmid were electroporated in 500 μL medium. Transfected cells were incubated for 48 hr before addition of 800 μg/mL G418 (CalBioChem, San Diego, CA) for selection of positive transfectants. Functional single HLAs expressed by each of the HLA transfectants were detected by specific reaction with fluorescein isothiocyanate (FITC) conjugated or biotinylated HLA mAbs (One Lambda, Inc., Canoga Park, CA), and 105 cells were incubated with 10 μL conjugated mAb for 30 min at 4oC. The cells were washed three times with 1 mL phosphate-buffered saline (PBS) and pelleted by centrifugation at 350g for 5 min. If biotinylated mAb was used, 100 μL streptavidin-FITC was added to the cell pellet, and the cells were incubated for 30 min at 4oC. Cells were then washed three times with 1 mL PBS and pelleted by centrifugation at 350g for 5 min. The cell pellets were added with 0.5 mL fixing solution (PBS with 0.5% formaldehyde) and then analyzed on a flow cytometer by collecting 5000 green fluorescence events. Positive reactions showed FL1 channel shifts.
DNA Typing Confirmation of the Transfectants
Genomic DNA was extracted from approximately 2×106 HLA-transfectant cells using the QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). The final concentration of the isolated DNA was adjusted to 100 ng/μL. DNA typing of the HLA-transfectant was performed by PCR using the Micro SSP DNA Typing Trays (One Lambda, Inc.) with 0.25 units of recombinant Taq polymerase (Hoffman-LaRoche, Basel, Switzerland) per reaction well. The PCR reaction mixture was subjected to electrophoresis on a 2.5% agarose gel (FMC Seakem LE) prepared using the Micro SSP Gel System (One Lambda, Inc.). DNA fragments were detected by staining with ethidium bromide and photographed using an ultraviolet transilluminator (Alpha Innotech Corp., San Leandro, CA).
Detection of Single and Native Human Leukocyte Antigen on Antigen-Coated Beads with Monoclonal Antibodies
HLAs were purified by cell lysis and affinity chromatography (3) using EBV cell lines for native HLAs and transfectants for the single HLAs and then coated onto microbeads as previously described (5,6). Eight different color-coded microbeads, which were coated with different individual HLAs, were mixed together. The bead mixture (5 μL) was diluted to 90 μL with PBS and incubated with 10 μL biotinylated mAbs for 30 min at 4oC. The beads were washed three times with 1 mL PBS, pelleted by centrifugation at 9000g for 2 min, and then 100 μL streptavidin-FITC was added and incubated for 30 min at 4oC. The beads were washed three times with 1 mL PBS, pelleted by centrifugation at 9000g for 2 min, and then fixed with 0.5 mL fixing solution. The green and yellow fluorescence of 5000 events was then analyzed on a flow cytometer. Positive reactions showed FL1 channel shifts.
Screening Human Leukocyte Antigen Antibody in Human Sera by an Array of Microbeads
The assay was performed by using a set of the HLA bead groups. Each group (5 μL), which contained eight different antigen-coated beads, was incubated with 20 μL human serum separately for 30 min at 20oC to 25oC. The beads were washed three times with wash buffer, pelleted by centrifugation at 9000g for 2 min, and then incubated for 30 min with 100 μL FITC-conjugated F(ab)2 fragment of goat anti-human immunoglobulin G (Fc γ fragment specific, 1:100 dilution, Jackson Immuno Research Laboratories, West Grove, PA). The beads were washed three times with 1 mL wash buffer and fixed with 0.5 mL fixing solution. The green and yellow fluorescence of 5000 events were then analyzed on a flow cytometer.
Analyzing an Array of Human Leukocyte Antigen Microbeads on a Flow Cytometer
The microbeads were uniform fluorescent latex beads, 2 to 4 μm in size. The beads were excited at 488 nm, generating a maximum emission of 580 nm, which could be collected on the FL2 channel on a flow cytometer. The bead mixture contained eight different beads, each of which had a unique FL2 channel shift and could be separated from each other by the FL2 channel of a flow cytometer. The fluorescence of positive reactions of HLA antibodies to the beads was detected by the FL1 channel of the cytometer. The major population of beads was gated on the FSC versus SSC dot plot, and an FL2 versus FL1 dot plot was obtained on the gated beads. Gates were set for each bead population that had reacted with the negative control serum on the FL2 versus FL1 dot plot; the same gates were used to analyze all the testing sera on their FL2 versus FL1 dot plot. A positive reaction was indicated by a shift of the beads to the right of the gate. The positive-negative cutoff was 50% mean channel shift of the positive control after subtracting the negative control. The weak positive was defined as 50% to 75% of the shift, and the strong positive was defined as 75.1% to 100% of the shift.
Construction of Functional Human Leukocyte Antigen Class I Recombinant Single Antigens
A panel of HLA class I gene-transfected cell lines was constructed. All of these transfectants, including 34 A alleles, 57 B alleles, and 19 C alleles, were verified by DNA typing (Table 1). Transfectants reacted with anti-class I mAb and with the specific mAbs, whereas the host cells remained negative to all the class I mAbs.
The purified HLAs reacted specifically with HLA mAbs, detected by a flow cytometry-based assay using an array of microbeads with different fluorescent intensity. Each population of beads with the same fluorescent intensity was coated with a single HLA and could be distinguished from other bead populations by the unique fluorescent intensity on the FL2 mean channel shifts (Fig. 1). A specific interaction between the beads with the corresponding mAb was indicated by the FL1 mean channel shift of the beads when an FITC-conjugated tag was used for detection. As an example, a serologically characterized A3 mAb reacted specifically with the A3 antigen beads shown an FL1 channel shift on the FL2 versus FL1 dot plot (Fig. 1).
The reaction patterns of a panel of specific HLA mouse mAbs to the native and single antigen bead arrays are summarized in Table 2. The native antigens were purified from 32 EBV cell lines to form a panel of 32 antigen beads (6). The single antigens were purified from transfectants, including 21 A, 43 B, and 8 C alleles (Table 2). All the previously serologically defined specificities of these mAbs by cytotoxicity tests showed positive reactions with both the specific native and the specific single antigens measured by the FL1 mean channel shifts (Table 2). Cross-reactivities of these mAbs resulting from common epitopes shared by the antigens (13,14) were detected by both the single antigens and the native antigens using FlowPRA tests (6) (Table 2).
Single Antigen Provides a Sensitive Method for Antibody Detection
Compared with beads coated with multiple antigens extracted from whole cells as previously described (6), beads coated with single antigens had a higher sensitivity because of the higher concentration of the same antigens. This prediction has been proven by comparison tests of the two methods with the same antibody at series dilutions. An antibody at a certain dilution lost the reactivity to the antigen extracted from a whole cell, whereas it still remained reactive to the same antigen extracted from single transfectants.
Single Antigen Uncovered the Antibody Specificity Masked by the Broad Epitopes
When a serum has broad HLA antibody reactivity, with the use of limited cell panels, the reactions against many of the less frequent specificities are often undetected. The single antigen panel provides a tool to identify each single antigen reaction against the antibodies in the serum, regardless of the existence of broadly reactive antibodies. Extra antibody reactivity that was masked by broad specificities was revealed using the single antigen panel (Table 3). This is shown by using the same methodology with either a single antigen or native antigen panel to compare the detected anybody specificities. Although most reactivity was concordant, certain reactivity can be detected or eliminated only by the single antigen panel. As an example, antibodies to A3, A11, A24, and B7 from serum S1 were detected by a panel of either single antigen or native antigen purified from 88 cell lines. However, the following additional specificities were found only when the serum was analyzed on the single antigen panel: A31, B27, B42, B55, B56, B81, and B82 (Table 3). Serum S2 was assigned as having B5C (B5, 35, 37, 53, 15, 17, 18, 21, and 70) antibodies using the native antigen panel with cross-reactive group (CREG) analysis (13,14). However, when the serum was analyzed on the single antigen panel, not only were all the B5C specificities shown to be positive but additional reactions were found with B13, B46, B7801, B50, and B56 (Table 3). Two other sera (S3 and S4) with extra specificities detected by single antigen beads are shown. Aside from additional specificity assignments, a single antigen panel can also eliminate specificities that had been wrongly assigned by statistical analysis involving complex reactions produced by cells having multiple class I specificities. In serum S3, both B55 and B60 were excluded by the single antigen detection, although they belonged to the B7C CREG group assigned by using the native antigen panel (Table 3). These “additional specificities” detected by the single antigen panel are not the result of the nonconcordance of the reactions, because both the native and single antigens reacted to the serum. However, the native antigens cannot determine those positive antigens for those sera, because each reaction contains multiple antigens, and broadly reactive epitopes covered the reactivity of other antigens in the same reaction. Multiple native antigens cannot eliminate certain specificities in a reaction determined by the CREG analysis for the same reason.
Single Antigen Resolves the Specificities Present in High Panel-Reactive Antibody Sera
It is difficult to identify all the specificities present in high PRA sera that react against almost all cells in a panel. A group of high PRA sera reacting to more than 90% of the native antigen panel was tested against a single antigen panel using the same flow cytometric method with purified antigen-coated beads (Table 4). All of these sera showed specific reaction patterns to single antigens, and many clearly negative reactions were identified. Although the sera were more than 90% positive with cells and no specificity could be assigned, with single antigens the percent positive was less than 90%, and specificities were clearly defined.
A group of 10 sera from kidney transplant patients, who had rejected a graft, was tested against a single antigen panel with 15 A alleles and 14 B alleles. As shown in Table 5, in which a large circle indicates a strong positive reaction and a small circle indicates a weak positive reaction, antibodies were found to most of the specificities against which the patient had been immunized. Antibodies were found in 31 of 35 donor-specific reactions (89%). Most important, although these sera were reactive, they did not react with specificities present in the patient. This confirms the validity of the test reactions.
We describe here the production of an extensive panel of 110 HLA class I antigens by recombinant DNA technology from a mammalian cell expression system. The purified sin-gle antigens were then absorbed onto flow cytometry beads and demonstrated to be structurally intact by specific reactions with well-defined HLA-mAbs.
The single antigen test was shown to eliminate the interference from other HLAs and, therefore, showed improved test resolution. Masked specificities by a broad reactive antibody using a cell panel can be uncovered using the single antigen panel. In addition, the single antigen panel can also resolve the nonreactive antigens in the same CREG groups, which would be mistakenly assigned as positive antigens by CREG analysis using a cell panel.
The task of identifying antibodies present within highly reactive sera is not simply an academic one but a serious clinical problem. Patients waiting for an organ transplant and those who have life-threatening bleeding problems requiring platelets need to have an accurate way for donors to be selected who would not react with their antibodies. That is, in highly reactive sera, the most important question is which antigens are negative to the sera and, therefore, acceptable to the patient as organ or platelet donors. These single antigen beads finally free us from the need to guess at the antibodies present using various computer programs. Starting with programs developed by Mickey et al. in 1982 (15), many similar programs have been developed over the past 30 years to deal with the difficulty in identifying specificities within broadly reacting antisera (16). When dealing with sera that react to more than 80% of the panel, attempts have been made to use the negative reactions to identify the specificities. Three separate analyses using stochastic scores, genetic algorithms, and neural networks have been described (16). More recently, use of amino acid triplets to match also has been described to find compatible donors for highly sensitized patients (17,18). The HLAMatchmaker program was developed on the basis of such a computer algorithm to facilitate the finding of compatible donors (18). Single antigen bead testing would most likely outperform all of these statistical methods by directly defining the negative and positive antigens for highly sensitized transplant patients.
One potential problem, which could occur with single antigen beads, is that the particular allele used to represent the antigen could be negative, but the antibody might react with another allele. For example, if the cell with A*6801 type were used to make the recombinant antigen, but the antibody reacted with A*6802 allele, there might be a reaction that would be missed. We used the most common alleles to represent each antigen in this original list of 110 specificities. Other alleles can be added in the future using the same expression system. It is unlikely that all known alleles will be differentiated in a humoral immune response. Because most of the “antigens” are identified serologically, and most alleles within the serologic group react in the same way, it is more likely that the number of test antigens required will be closer to the already defined serologic groups, rather than the massive alleles now known. By further testing with the single antigen beads, all of the serologically important alleles should become apparent.
The authors thank Drs. R. A. Bray, M. Lopez-Cepero, and D. Cook for their helpful discussions; S. Rojo, T. Chen, L. Banh, D. Chau, H. Tran, and A. Caoyonan for assistance with single antigen production and quality control; and M. Lias and N. El-Awar for providing mAbs and sera information.
1. Terasaki PI, Kreisler M, Mickey RM. Presensitization and kidney transplant failures. Postgrad Med J 1971; 47: 89.
2. Parham P. Purification of immunologically active HLA-A and -B antigens by a series of monoclonal antibody columns. J Biol Chem 1979; 254: 8709.
3. Kao KJ. Plasma and platelet HLA in normal individuals: quantitation by competitive enzyme-linked immunoassay. Blood 1987; 70: 282.
4. Zaer F, Metz S, Scornik JC. Antibody screening by enzyme-linked immunosorbent assay using pooled soluble HLA in renal transplant candidates. Transplantation 1997; 63: 48.
5. Pei R, Wang G, Tarsitani C, et al. Simultaneous HLA class I and class II antibody screening with flow cytometry. Hum Immunol 1998; 59: 313.
6. Pei R, Lee JH, Chen T, et al. Flow cytometric detection of HLA antibodies using a spectrum of microbeads. Hum Immunol 1999; 60: 1293.
7. Tambur AR, Bray RA, Takemoto SK, et al. Flow cytometric detection of HLA-specific antibodies as a predictor of heart allograft rejection. Transplantation 2000; 70: 1055.
8. Gebel HM, Bray RA, Ruth JA, et al. Flow PRA to detect clinically relevant HLA antibodies. Transplant Proc 2001; 33: 477.
9. Worthington JE, Robson AJ, Sheldon S, et al. A comparison of enzyme-linked immunoabsorbent assays and flow cytometry techniques for the detection of HLA specific antibodies. Hum Immunol 2001; 62: 1178.
10. Barnardo MC, Harmer AW, Shaw OJ, et al. Detection of HLA-specific IgG antibodies using single recombinant HLA alleles. Transplantation 2000; 70: 531.
11. Gorman CM, Merlino GT, Willingham MC, et al. The Rous sarcoma virus long terminal repeat is a strong promoter when introduced into a variety of eukaryotic cells by DNA-mediated transfection. Proc Natl Acad Sci U S A 1982; 79: 6777.
12. Kavathas P, Bach FH, Demars R. Gamma ray-induced loss of expression of HLA and glyoxalase I alleles in lymphoblastoid cells. Proc Natl Acad Sci U S A 1980; 77: 4251.
13. Rodey GE, Neylan JF, Whelchel JD, et al. Epitope specificity of HLA class I alloantibodies. Hum Immunol 1994; 39: 272.
14. Fuller TC. Monitoring HLA alloimmunization. Clin Lab Med 1991; 11: 551.
15. Mickey MR, Ayoub G, Terasaki PI. Prediction of negative crossmatch: an aid for cost-effective kidney sharing. Transplant Proc 1982; 14: 279.
16. Clark BD, Leong S-W. Crossmatch prediction of highly sensitized patients. In: Terasaki PI, Cecka JM, eds. Clinical transplants 1992. Los Angeles, CA: UCLA Tissue Typing Laboratory, 1992, p. 435.
17. Takemoto S, Terasaki PI. HLA compatibility can be predicted by matching only three residues with outward oriented sidechains. Transplant Proc 1996; 28: 1264.
18. Duquesnoy RJ, Marrari M. HLAMatchmaker: a molecularly based algorithm for histocompatibility determination. II. Verification of the algorithm and determination of the relative immunogenicity of amino acid triplet-defined epitopes. Hum Immunol 2002; 63: 353.