Bioengineered soluble HLA-B7. Genesis, characterization, and occurrence of dimerization

Peptides derived from a soluble molecule of the human leukocyte antigen (HLA) class I cause apoptosis in gastric cancer cell lines

We have reported that the levels of the soluble molecule of the human leukocyte antigen class I (sHLA-I) in patients with advanced gastric cancer were significantly lower than those in patients with cancer in the early stages. However, the effect of sHLA-I on gastric cancer cells has not been elucidated. Using human gastric cancer cell lines, MKN28, MKN45, and MKN74, we evaluated the effects of sHLA-I on cell growth, DNA synthesis, and apoptosis induction. Three types of synthesized peptides derived from HLA-I were also examined for their capacity to induce apoptosis. sHLA-I and a synthesized peptide, nos. 220-232 of the alpha3 domain of HLA-B7, caused cell growth inhibition by inducing apoptosis in human gastric cancer cells. This peptide also inhibited the in vivo growth of cancer dissemination caused by an intraperitoneal injection of MKN45 into severe combined immunodeficient mice. In conclusion, sHLA-I and the peptides derived from HLA-I cause apoptosis in human gastric cancer cell lines.

Beta 2-microglobulin-free HLA class I heavy chain epitope mimicry by monoclonal antibody HC-10-specific peptide

mAb HC-10 loses its reactivity with HLA class I (HLA-I) H chain (HC) following its association with beta(2)-microglobulin (beta(2)m). Furthermore, the HC-10 defined epitope appears to be involved in the pathogenesis of spondyloarthropathies, because HC-10 reduced their incidence in HLA-B27(+)beta(2)m degrees /MHC class II knockout mice. This study has characterized the determinant recognized by HC-10. Panning of a phage display peptide library with HC-10 resulted in isolation of the motif PxxWDR, which could be aligned with P57, W60, D61, and R62 of the first domain of the HLA-I HC allospecificities reactive with HC-10. The (55)EGPEYWDR(N/E)T(64) (p-1) is the shortest motif-bearing peptide that reacts with HC-10 and inhibits its binding to soluble HLA-B7 HC, irrespective of whether N (p-1a) or E (p-1b) is present at position 63. By contrast, HC-10 did not react with six additional peptides, each bearing motif amino acid substitutions present in HC-10-not-reactive HLA-I allospecificities. The p-1-derived Qp-1, synthesized with the additional conserved Q54, which displays the highest in vitro reactivity with HC-10, was the only one to induce in mice IgG resembling HC-10 in their fine specificity. Mapping of the HC-10-defined determinant suggests that the lack of mAb reactivity with beta(2)m-associated HLA-I HC is caused by blocking by the peptide in the groove of beta(2)m-associated HLA-I HC, though a role of HC conformational changes following its association with beta(2)m cannot be excluded. This information contributes to our understanding of the molecular basis of the antigenic profiles of beta(2)m-free and beta(2)m-associated HLA-I HC and may serve to develop active specific immunotherapy of spondyloarthropathies.

Size variants of beta-2-microglobulin-free human leucocyte antigen class I heavy chain make different contributions to its serum increase in multiple myeloma

We previously showed that serum beta-2-microglobulin (beta2m)-free human leucocyte antigen (HLA) class I heavy chain (FHC) levels were increased in MM and correlate with disease activity. The present investigation, carried out in 124 multiple myeloma patients, studied the expression of the three size variants of FHC, namely the 42 kDa intact heavy chain (A variant, AV), released through a shedding process, and the truncated FHC (tFHC) 39 kDa (BV) and 36-35 kDa (CV) released by means of membrane-type metalloprotease activity. The increase in FHC correlated with a high expression percentage of BV (r = 0.32, P = 0.0002) and tFHC (r = 0.42, P < 0.0001), which could help to discriminate multiple myeloma from monoclonal gammopathy of undetermined significance (tFHC mean ratio = 3.2; Mann-Whitney U-test, P < 0.0001). tFHC levels highly correlated with other disease activity markers, namely haemoglobin (r = -0.35; Spearman's rank, P = 0.0001), percentage of bone marrow plasma cells (r = 0.4, P < 0.0001) and beta2m levels (r = 0.36, P < 0.0001), while only the last barely correlated (r = 0.2, P = 0.03) with AV. Finally, the 0.4, 0.57 and 0.71 mg/l BV, tFHC and (to a lesser extent) FHC cut-off values divided patients into two groups with different survival curves (P = 0.0005, P = 0.0025 and P = 0.04 respectively). These data are in favour of a correlation between disease aggressiveness and cleavage of these variants by membrane-type metalloprotease enzymes.

Beta-2 microglobulin-free HLA class I heavy chain (FHC) A3 and/or A30 soluble products contribute only minimally to serum FHC expression

No monoclonal antibodies (mAbs) are presently available to measure the total amount of beta2-microglobulin-free HLA class I heavy chain (FHC) in sera. The available ELISA-based double determinant immunoassay (DDIA), established to measure FHC, uses two mAbs (TP25.99 and HC-10) that recognize a monomorphic determinant expressed on all HLA-B/C FHC products and a determinant expressed only on some HLA-A FHC products. This restricted reactivity implies that, in addition to HLA-B/C, HLA-A FHC products are also detected in individuals bearing HLA A3 and/or A30 allotypes. The aim of this study was to establish whether such restriction results in the detection of low FHC levels in individuals lacking HLA A3 and/or A30 allospecificities. The FHC mean concentration (+/- SD) in 294 healthy blood/bone marrow donors (HBDs) was 0.24 (+/- 0.2) mg/l. The grouping of HBDs according to their HLA-A FHC product reactivity with one, both or no mAbs did not result in any statistically significant differences (Mann-Whitney test: P > 0.05) between their median FHC concentrations. Since the absence of differences in their FHC levels was not attributable to a difference in the percentage distribution of HLA allotypes associated with high or low HLA-B/C FHC expression, our results indicate that FHC HLA A3 and/or A30 products detected in DDIA by these two mAbs only minimally contribute to FHC serum expression and that the assay is not limited by the failure to detect HLA-A FHC products in A3- and/or A30- individuals.

Human liver allograft acceptance and the "tolerance assay". II. Donor HLA-A, -B but not DR antigens are able to trigger regulation of DTH

In our initial study of liver transplant recipients using the trans vivo delayed-type hypersensitivity (DTH) assay, we found that in donor derived B-LCL or sonicates of donor leukocytes triggered linked suppression of the response to recall antigens tetanus toxoid (TT) or Epstein-Barr virus (EBV). Since both donor antigen sources contain HLA class I and class II proteins, we wished to determine which donor HLA proteins were responsible for the linked suppression effect. PBMC from four liver transplant recipients with donor-specific DTH regulation were studied. Surprisingly, we found that single donor HLA-A or B antigens (4/4 patients) but not single HLA-DR (0/4) donor antigens triggered linked suppression of DTH. A dose response study of two patients revealed that donor-type HLA-DR antigens (0.5-500ng) were not capable of triggering any linked suppression; however, as little as 500pg of donor-type HLA-class I protein triggered linked suppression of DTH response to a recall antigen. These findings may have implications for the differential impacts of class I vs class II mismatching in organ transplantation. On a practical level, they indicate that soluble HLA-A and B antigens are the proper choice for detection of DTH regulation as part of a "tolerance assay" in human liver transplant recipients.

Clinical significance of soluble form of HLA class I molecule in Japanese patients with pancreatic cancer

In recent studies a soluble form of human leukocyte antigen class I (sHLA-I) has been found in blood, urine, ascitic fluid, and various other tissues. Research has been focused on the role of sHLA-I in the induction of immunotolerance in organ transplantation. To examine the role of sHLA-I in the immune system of patients with malignancy, we examined serum sHLA-I levels in patients with pancreatic, biliary, hepatic malignancy, and other diseases. We examined sHLA-I levels in the sera of patients with pancreatic cancer (n = 19), benign biliary disease and chronic pancreatitis (n = 20), hepatocellular carcinoma (n = 51), gallbladder cancer (n = 6), cholangiocellular carcinoma (n = 6), and in normal controls (n = 22), using enzyme-linked immunosorbent assay (ELISA). In patients with pancreatic cancer we also analyzed the relationship between sHLA-I and CA19-9, and the specificity and sensitivity of sHLA-I. When patients with acute or chronic hepatitis were excluded from analysis, the mean sHLA-I level in patients with pancreatic cancer was significantly higher than that of normal controls (p < 0.01) and patients with benign disease (p < 0.01), hepatocellular carcinoma (p < 0.01), gallbladder cancer (p < 0.05), and cholangiocarcinoma (p < 0.05). We determined a serum sHLA-I cutoff level for normal controls of 2000 ng/ml; serum levels of sHLA-I were higher than the cutoff in ten patients with pancreatic cancer, and serum levels of CA19-9 were lower than 37 IU/l in 9 of 14 patients; sensitivity and specificity were 88.2% and 85.5%, respectively. Serum levels of sHLA-I in pancreatic cancer patients were higher than in the other diseases, although we found that pancreatic cancer cell lines did not produce the sHLA-I. The evaluation of serum sHLA-I levels could have clinical significance in pancreatic cancer.

Soluble donor HLA class I and beta 2m-free heavy chain in serum of lung transplant recipients: steady-state levels and increases in patients with recurrent CMV infection, acute rejection episodes, and poor outcome

We determined the concentration of donor sHLA/beta(2)m and total beta(2)m-free heavy chain (HC) in the serum of lung transplant recipients with ELISA assays. While we were unable to detect specific donor beta(2)m-free HCs due to a lack of available antibodies, we could determine if events that led to an increase in the release of beta(2)m-free HC also led to an increase in the release of donor sHLA/beta(2)m, particularly the 36 kDa, proteolytically cleaved form. We found that lung transplants constituitively release donor sHLA/beta(2)m at ng/ml levels. The levels (both of donor sHLA/beta(2)m and total beta(2)m-free HC) were significantly increased in CMV-sero-negative recipients (but not in CMV-sero-positive recipients) at the onset of post-transplant CMV disease. Acute rejection episodes were also associated with an increased release of donor sHLA/beta(2)m, but not of beta(2)m-free HC. However, in patients with particularly poor outcome (i.e., graft loss within 1 year) there was a significant release of beta(2)m-free HC. Analysis of one such patient showed a predominance of 36 kDa forms of donor-sHLA/beta(2)m. Our data are consistent with the hypothesis that the metalloproteinase that cleaves beta(2)m-free HC is active during uncontrolled CMV infection and acute rejection. However, recall responses to CMV and controlled immune responses to donor may result in little or no activation of sHLA class I release.

Inhibition of Epstein-Barr virus (EBV)-specific CD8+ cytotoxic T lymphocyte (CTL) activity by soluble HLA class I in vitro

In the present study, the effects of soluble HLA (sHLA) class I molecules against EBV-specific CTL were examined. Two different sources of sHLA class I, either bioengineered spliced form of HLA-B7 (sB7) or natural production from EBV-transformed B cells (natural sHLA), were added during the induction of CTL or incubated with MHC-restricted CD8+ CTL, which were selected by immunobeads just before testing for their cytotoxic activity. Both sB7 and natural sHLA class I blocked the generation of CD8+ CTL and also inhibited the cytotoxic activity of established CTL in a dose-dependent manner. In both ways, natural sHLA class I was effective in 10-fold lower concentrations compared with sB7. The inhibitory effect did not require a sharing of the HLA allotypes between sHLA and the CTL. CTL, after being treated with sHLA, underwent apoptosis, which was considered here as the main mechanism.

Serum levels of beta-2-microglobulin-free heavy chain of HLA class I antigen in healthy individuals: relationship to their class I allotype

An ELISA-based double determinant immunoassay has been established to measure the soluble beta2-microglobulin (beta2m)-free heavy chain (FHC) of the HLA-B, -C (and HLA-A3, -A28 and -A30) class I molecular complex in sera from 212 HLA-typed healthy unrelated individuals. FHC was calculated by means of a standard curve constructed using serial concentrations of beta2m-associated HLA-class I heavy chain (HLA-I)/FHC purified from cultured human lymphoid cell C1R-sB7-supernatant. The mean FHC concentration (+/-SD) was 0.25 mg/l (+/-0.2). Its median concentration did not statistically differ between males and females, though the male/female ratio was greater in the high secretor (FHC >0.45 mg/l; mean + 1SD) than in the low secretor group (FHC < 0.05 mg/l; mean - 1SD). FHC < 0.05 mg/l was statistically (Fisher's exact test) associated with HLA-B17 (p = 0.003); FHC > 0.45 mg/l was statistically associated with HLA-B35 (p = 0.003) and -Cw4 (p = 0.002). None of these allele-positive groups showed a mean FHC concentration 1.5 times higher than that of the corresponding allele-negative ones. This allotype-dependent HLA-B and C FHC enhancement was less marked than that previously reported for HLA-I in individuals carrying HLA-A9 (and its splits). These results indicate that FHC could be a more valuable marker when its levels are compared among individuals carrying different allotypes. Moreover the lack of correlation between FHC and HLA-I levels measured in 52 HLA-A3, -A28 or -A30 positive individuals suggests that the two molecules may be regulated by different metabolic pathways and their serum expression may have a different biological significance.

Increased serum levels of beta2m-free HLA class I heavy chain in multiple myeloma

Serum levels of beta2-microglobulin (beta2m)-free HLA class I heavy chain (FHC) in 94 patients with multiple myeloma (MM) were higher than in 29 patients with monoclonal gammopathy of undetermined significance (MGUS) (P = 0.023) and in 97 sex- and age-matched healthy controls (P < 0.0001). Spearman correlation analysis indicated that in MM, FHC correlated with beta2m (r = 0.31, P = 0. 003) and the percentage of bone marrow plasma cells (BMPC%) (r = 0. 36, P = 0.002), whereas beta2m, in addition to BMPC% (r = 0.43, P = 0.0003), also correlated with creatinine levels (r = 0.63, P < 0.0001), haemoglobin levels (r = -0.35, P = 0.0007) and patient age (r = 0.34, P < 0.0011). Furthermore, MM patients with poor prognosis (beta2m >/= 6 mg/l) displayed higher FHC levels than those with a better prognosis (beta2m < 6mg/l) (P < 0.021). At variance from beta2m, these levels were not influenced by renal failure, as indicated by the lack of Spearman correlation of FHC with creatinine concentration and of statistical significance between the median FHC concentration of MM patients with creatinine < 176.6 micromol/l and those with creatinine >/= 176.6 micromol/l (P = 0.3). Stratification of patients according to disease activity and stage showed that FHC levels were only statistically different (P = 0.04) for disease activity, whereas beta2m and C-reactive protein were not. Taken together, our data indicate that serum FHC may be a useful disease marker in MM.

Specific sHLA in healthy donors and donor-specific sHLA in renal transplant patients

We studied cadaver kidney transplant recipients to determine if their serum levels of donor-specific class I sHLA correlated with graft outcome. Testing of sHLA was performed by an ELISA sandwich assay using allospecific monoclonal trapping antibodies and anti-beta2-mu detecting antibody. Sufficient sHLA sensitivity (<1 ng/ml) was achieved by using two synergistic trapping antibodies. Suitable antibodies were available for A2 and B7, and data were collected for these two antigens. Stability of these sHLA was determined in plasma and serum as were ranges of normal and background levels. Background levels varied substantially. Five A2- recipients of A2+ grafts and 5 B7- recipients of B7+ grafts were studied with appropriate sHLA levels measured pre-transplant and at intervals post-transplant. Graft outcome was assessed by serum creatinines, renal biopsies and/or therapy for rejection. In the 5 patients (3 A2- and 2 B7-) whose post-transplant donor-specific sHLA never exceeded immunological complications (e.g., post-operative ATN, ureteral obstruction) did not affect the correlation. In the 5 patients with post-transplant levels exceeding pre-transplant levels, subsequent evidence of rejection was observed. Periodic measurement of donor-specific sHLA should be a useful instrument for monitoring renal allograft rejection.

Beta2-micro-free HLA class I heavy chain levels in sera of healthy individuals. Lack of association with beta2-micro-associated HLA class I heavy chain levels and HLA phenotype

We have applied a double-determinant immune assay (DDIA) to measure soluble beta2-microglobulin (beta2-micro)-free HLA class I heavy chains in serum. The mean concentration of beta2-micro-free HLA class I heavy chains in serum from 120 healthy subjects was 0.21+/-0.24 microg/ml. The individual serum levels of beta2-micro-free HLA class I heavy chains had a wide distribution, did not seem to be related with HLA phenotype, were stable over time and did not change with age. The serum levels of soluble beta2-micro-free HLA class I heavy chains did not correlate with those of soluble beta2-micro-associated HLA class I heavy chains, suggesting that their release is independently regulated. Three forms of soluble beta2-micro-free HLA class I heavy chains, with apparent molecular masses of 44, 39 and 37-35 kD, respectively, circulate in human serum. These results provide a useful background to assess the serum level of soluble beta2-micro-free HLA class I heavy chains in pathological conditions and to evaluate their putative immunoregulatory function.

Purified truncated recombinant HLA-B7 molecules abrogate cell function in alloreactive cytotoxic T lymphocytes by apoptosis induction

BACKGROUND

Soluble MHC class I molecules are ubiquitous in human body fluids, including serum, urine, sweat, and cerebrospinal fluid. However, their biological function has remained unresolved. Membrane-derived human soluble MHC molecules (soluble human leukocyte antigen; sHLA) have been shown to induce apoptosis in alloreactive cytotoxic T lymphocytes (CTL). Here we report the efficacy of recombinant soluble HLA-B7 (rsHLA-B7) to modulate T-cell function.

METHODS

Primers of HLA-B7 were designed to allow amplification of a cDNA lacking the transmembrane and cytoplasmic domains yielding a truncated gene. rsHLA-B7 molecules were expressed in the human myeloma cell line 721.221 and purified by affinity chromatography using the BB7.7 mouse monoclonal antibody. CTL were generated from peripheral blood lymphocytes derived from healthy blood donors by stimulation with irradiated Epstein Barr virus-transformed HLA-B7-positive B cells. CTL were preincubated with rsHLA-B7, and cytotoxicity and apoptosis were tested according to standard procedure.

RESULTS

A total of 2 x 10(6) cells/ml secreted 10 microg/ml rsHLA-B7 as determined by a conformation-dependent ELISA, suggesting that rsHLA-B7 do not require the transmembrane and cytoplasmic regions for proper folding. After purification by affinity chromatography, rsHLA-B7 induced apoptosis in anti-HLA-B7 CTL, but not in anti-HLA-A2-specific, CTL. As a consequence, allorecognition of target cells by the CTL was significantly blocked.

CONCLUSION

Recombinant sHLA are sufficient binding cues for T cells, which efficiently induce apoptosis and block allorecognition of target cells by CTL. Thus, recombinant sHLA molecules may become a valuable new modality for specific immunological therapeutic intervention.

The metalloproteinase-mediated pathway is essential for generation of soluble HLA class I proteins by activated cells in vitro: proposed mechanism for soluble HLA release in transplant rejection

We and others have found donor-derived soluble beta2m-associated HLA class I proteins (sHLA/beta2m) in the serum of allograft recipients with acute and chronic rejection. Whether appearance of sHLA/beta2m and upregulated expression of donor cell-bound HLA/beta2m during allograft rejection are related events is unknown. Activation-induced upregulation of in vitro HLA/beta2m expression correlates with the surface expression of another form of HLA class I, namely beta2m-free HLA heavy chains (beta2m-free HC). We have shown that beta2m-free HC, but not beta2m-associated HC, are then cleaved by a specific membrane-bound metalloproteinase and released into supernatants as soluble 36 kDa proteins. We show now that activated peripheral blood lymphocytes produce predominantly the 36 kDa form of sHLA proteins which is present in supernatants as both beta2m-free HC and sHLA/beta2m. Importantly, the metalloprotease inhibitor BB-94 blocked not only the release of soluble beta2m-free HC, but also the appearance of sHLA/beta2m in cell supernatants. Low levels of 36 kDa beta2m-free HC were also present in human plasma of healthy donors. These data suggest an important role for the HLA class I-specific metalloproteinase in vivo in healthy individuals and during allograft rejection in the generation of soluble beta2m-free and beta2m-associated HLA proteins.

Serum concentrations of soluble HLA-class I and CD8 forms in patients with viral hepatic disorders

Soluble HLA-class I and CD8 molecules were determined by sandwich ELISA in patients with viral-induced hepatic disorders. As a whole, the patients with hepatic disorders (acute hepatitis: AH; chronic hepatitis: CH; liver cirrhosis: LC; hepatocellular carcinoma: HCC) showed higher sHLA-class I and sCD8 levels than normal controls (P < 0.001). AH patients had the highest sHLA-class I levels (mean, 3513 +/- 2112 ng/ml), followed by CH (2896 +/- 1290 ng/ml), LC (2293 +/- 1266 ng/ml), and HCC (2221 +/- 1212 ng/ml) sCD8 levels wer highest in AH, followed by HCC, LC, and CH, in that order. Among histologically defined C virus-positive patients, sHLA-I levels were higher in those with chronic active hepatitis (CAH) 2A (3802 +/- 1124 ng/ml) than in those with chronic persistent hepatitis (CPH; 2200 +/- 711 ng/ml; P < 0.01), the levels then decreased as the disease progressed (CAH2B, 3564 +/- 1783 ng/ml, LC, 2376 +/- 1265 ng/ml). In contrast, sCD8 values showed little difference among the disorders. sHLA-class I levels showed a positive correlation with sCD8 values both in whole patients and in patients with AH (P < 0.01), but no correlation was shown, in any patients, with biochemical parameters such as GPT and GOT. These findings, taken together, suggest that hepatic destruction is not the only cause of sHLA-class I production, but that sHLA-class I levels, together with sCD8 levels, may reflect immunological activity in hepatic disorders.

Soluble HLA class I in epithelial lining fluid of lung transplants: associations with graft outcome

We hypothesized that the small amounts of donor HLA-A and HLA-B proteins detected in the serum during organ allograft rejection are indicative of higher local releases within the graft itself. We determined the concentrations of total HLA class I (HLA-I) and, in selected cases, specific donor and host HLA-A and HLA-B proteins, in the epithelial lining fluid (ELF) sampled by bronchoalveolar lavage (BAL) of lung transplant recipients (n = 37) and of normal controls (n = 25). We found that 1) HLA-I proteins were enriched in the lung ELF relative to other proteins; 2) the concentration of HLA-I in the ELF of well-functioning transplants was similar to that in normal lungs; 3) HLA-I proteins and total proteins were elevated in the ELF of patients who developed chronic rejection or refractory acute rejection; 4) the concentration of HLA-I was correlated with the percentage of neutrophils but not with the percentage of lymphocytes in the ELF of transplanted lungs; and 5) only the percentage of lymphocytes was elevated in the ELF of transplant patients with active CMV infections. Total HLA-I from the ELF was found to contain a mixture of both donor- and recipient-type HLA-A and HLA-B proteins and the donor-type HLA-A2 was found to be highly enriched in the ELF relative to serum.

Natural killer clones recognize specific soluble HLA class I molecules

Enhancement of major histocompatibility complex (MHC) class I expression leads to protection from natural killer (NK) cell recognition in several systems. MHC class I gene products are released from the cell surface and can be found in sera as soluble forms. To investigate the possible immunoregulatory role of soluble HLA (sHLA) in NK cell-target recognition, several sHLA antigens were studied for their ability to induce NK cell cytotoxicity modulation. NK cell-target recognition was inhibited by the addition of sHLA during the cytotoxicity assay. Our results indicate that sHLA molecules can down-regulate NK killing at the effector level. Moreover, different NK clones are able to specifically recognize different sHLA antigens. Kp43 molecules seem to be involved in the NK recognition of sHLA-B7.

Detection of panel-reactive anti-HLA class I antibodies by enzyme-linked immunosorbent assay or lymphocytotoxicity. Results of a blinded, controlled multicenter study

A soluble HLA ELISA for the detection of anti-HLA class I IgG antibodies was developed and compared to complement-dependent microlymphocytotoxicity. ELISA plates were coated with a panel of sHLA class I antigens isolated from the culture supernatants of 46 different EBV-transformed phenotyped B-cell lines. After the incubation of the coated plates with test serum, bound antibodies were detected using a peroxidase-conjugated anti-human IgG antibody. Absorbance was read using an ELISA plate reader and assay results were analyzed by computer. Antibody specificities were determined by Fisher's exact test tail analysis. The reproducibility of ELISA assay results was evaluated in a blinded, controlled multicenter study. A total of 102 serum specimens from patients on waiting lists to receive kidney transplants were tested five times by ELISA in five different laboratories. The correlation coefficients (r) of %PRA values determined by ELISA ranged from 0.89 to 0.96, and the average agreement on qualitative assay results (antibody positive vs antibody negative) was 98%. Endpoint titration of several serum specimens demonstrated equivalent sensitivity of ELISA and microlymphocytotoxicity (using the anti-globulin antibody protocol). Most of the antibody specificities determined by ELISA were in agreement with specificities determined by microlymphocytotoxicity. To evaluate the correlation of ELISA and microlymphocytotoxicity (CDC) assay results the same 102 specimens were tested six times by CDC in five different laboratories. The interlaboratory correlation coefficient (r) of %PRA values determined by microlymphocytotoxicity ranged from 0.57 to 0.94, and the average agreement on qualitative assay results was 85%. A comparison of ELISA with microlymphocytotoxicity was performed using consensus microlymphocytotoxicity results. This showed a high correlation (r = 0.81) of %PRA values determined by ELISA and microlymphocytotoxicity. This demonstrates that the detection of anti-HLA class I antibodies by soluble HLA ELISA is a reliable alternative to microlymphocytotoxicity testing.