MHC class I molecules expressed on cell surfaces are composed of H chain, β2-microglobulin and any of a vast array of peptides. The role of peptide in the recognition of HLA class I by serum HLA Abs is unknown. In this study, the solid-phase assay of a series (n = 11) of HLA-A2-reactive, pregnancy-induced, human mAbs on a panel (n = 12) of recombinant monomeric HLA-A2 molecules, each containing a single peptide, revealed peptide selectivity of the mAbs. The flow cytometry membrane staining intensities on the HLA-A2-transduced cell line K562, caused by these mAbs, correlated with the number of monomer species detected by the mAbs. Flow cytometry staining on HLA-A2-bearing cell lines of a variety of lineages was indicative of tissue selectivity of these HLA-A2 mAbs. This tissue selectivity suggests that the deleterious effect on allografts is confined to alloantibodies recognizing only HLA class I loaded with peptides that are derived from tissue-specific and household proteins. Since Abs that are only reactive with HLA loaded with irrelevant peptides are expected to be harmless toward allografts, the practice of HLA Ab determination on lymphocyte-derived HLA deserves reconsideration.

Human histocompatibility leukocyte Ag class I molecules play a pivotal role in the defense toward intracellular pathogens. The activation of CD8+ T cells is initiated by recognition of pathogen-derived peptides bound in the HLA class I molecules of APCs, and this interaction requires at least partial MHC identity of T cells and APCs. In the case of MHC disparity, however, HLA molecules themselves are recognized as foreign, resulting in an adaptive alloimmune response to HLA class I. After blood transfusion, or transplantation of an incompatible graft, differences in the HLA type of donor and recipient can cause both formation of HLA-specific Abs (1) and T cell activation. These reactivities also occur during and after pregnancy (2) in a proportion of women. As a clinical consequence, preformed HLA Abs, detectable in a lymphocyte cross-match, are a contraindication to transplantation with an organ bearing the HLA Ags that are detected by these Abs.

Peptide embedded in alloantigen is a critical factor that determines the recognition of alloreactive, MHC class I-specific T cells (3). Crystallographic studies have shown that a role for peptide in this allorecognition is likely because subtle differences in conformational structure ensue when different peptides are incorporated in MHC class I (4, 5). This variability has also been detected serologically: the successful binding of some mouse monoclonal allo-MHC Abs to H2 class I Ags is governed by peptide, and, as these mAbs permit only a limited repertoire of peptides, they detect a limited number of H2 class I molecules on lymphocytic cell surfaces by flow cytometry (FCM) (6) and immunoprecipitation (7).

Whether serum HLA Abs are equally sensitive to peptide bound in the HLA class I groove has not been documented before. Monoclonal HLA Abs have been instrumental in determining epitopes that are responsible for alloantibody binding (8), but to what extent the shaping of the epitopes is peptide dependent is unknown. To determine whether peptide-induced changes of HLA class I conformation affect Ab binding in humans, we undertook the present study in which a series of recombinant HLA-A2 molecules, each containing a different peptide, served as ligands for human HLA mAbs. Differential binding characteristics indicated that the nature of the peptide bound in HLA class I is decisive for binding of HLA-A2 Abs and explained the limited binding patterns of some human mAbs to cell surface expressed HLA-A2.

HLA-A2 monomers (based on the HLA-A*0201 sequence) were synthesized as described previously (9), with minor modifications. Peptides used in HLA-A2 monomer synthesis were synthesized by solid-phase technology (10) and are listed in Table I.

Table I.

Peptides used in HLA-A2 monomer synthesis

PeptideGeneAmino Acid Sequence
HY SMCY FIDSYICQV 
PRA PRAME SLYSFPEPEA 
HA-1h KIAA0223 VLHDDLLEA 
HA-2 MYO1G YIGEVLVSV 
IB.54 Insulin HLVEALYLV 
HPV E7 12–20 HPV MLDLQPETT 
EII49 FASN FLFDGSPTYV 
EBV BMLF-1 GLCTLVAML 
Flue IMP (matrix 58–66) GILGFVFTL 
hCMV pp65 NLVPMVATV 
HIV Pol and RT2 ILKEPVHGV 
PR-1 Proteinase-3 VLQELNVTV 
PeptideGeneAmino Acid Sequence
HY SMCY FIDSYICQV 
PRA PRAME SLYSFPEPEA 
HA-1h KIAA0223 VLHDDLLEA 
HA-2 MYO1G YIGEVLVSV 
IB.54 Insulin HLVEALYLV 
HPV E7 12–20 HPV MLDLQPETT 
EII49 FASN FLFDGSPTYV 
EBV BMLF-1 GLCTLVAML 
Flue IMP (matrix 58–66) GILGFVFTL 
hCMV pp65 NLVPMVATV 
HIV Pol and RT2 ILKEPVHGV 
PR-1 Proteinase-3 VLQELNVTV 

Human hybridomas were established from B lymphocytes of seven HLA Ab-seropositive, multiparous women by EBV transformation, followed by electrofusion and hypoxanthine, aminopterin, thymidine, and ouabain selection of Ab-secreting EBV lines, and rigorously subcloning as described elsewhere (11). The Medical Ethics Committee of Leiden University Medical Center approved the use of human donors. Some individuals yielded multiple hybridomas. HLA specificities of human mAbs were determined by complement-dependent cytotoxicity (CDC)3 against large (n > 240) panels of HLA-typed PBLs (Table II). mAbs were used as hybridoma supernatants at Ig concentrations ranging from 0.6 to 2.5 μg/ml. Murine mAb TP25.99, which binds a monomorphic epitope in the α3 domain of HLA class I (12), was provided by Dr. S. Ferrone (Department of Immunology, Roswell Park Cancer Institute, Buffalo, NY). Murine mAb 20.4 (anti-human nerve growth factor receptor; NGFR) was obtained from American Type Culture Collection.

Table II.

Human monoclonal HLA Abs

mAbaHLA Specificity by CDCrbEpitopecIsotype
JOK3H4 A2 0.952 107W IgM,λ 
ROU2D3 A2/B17 0.944 62G IgM,λ 
SN230G6 A2/B17 0.904 62G IgG1,λ 
WIM1B3 A2/B17 0.815 62G IgM,λ 
HA5C2 A2/A28 0.711 149A 152V IgM,κ 
JOK2C7 A2/A28 0.533 149A 152V IgM,λ 
SN607D8 A2/A28 0.933 149A 152V IgG1,κ 
SN66E3 A2/A28 0.902 149A 152V IgM,κ 
WK4E3 A locus not A1/A24 0.826 166EW IgM,λ 
WK3D10 A2/A3/A23/A31/B7/B17/B13/B40/B21/B62 0.745 nd IgM,κ 
WIM8E5 A1/A10/A11/A9/A29/A30/A31/A33/A28d 0.578 nd IgG1,κ 
BVK1F9 B8 0.979 n/ae IgG1,κ 
mAbaHLA Specificity by CDCrbEpitopecIsotype
JOK3H4 A2 0.952 107W IgM,λ 
ROU2D3 A2/B17 0.944 62G IgM,λ 
SN230G6 A2/B17 0.904 62G IgG1,λ 
WIM1B3 A2/B17 0.815 62G IgM,λ 
HA5C2 A2/A28 0.711 149A 152V IgM,κ 
JOK2C7 A2/A28 0.533 149A 152V IgM,λ 
SN607D8 A2/A28 0.933 149A 152V IgG1,κ 
SN66E3 A2/A28 0.902 149A 152V IgM,κ 
WK4E3 A locus not A1/A24 0.826 166EW IgM,λ 
WK3D10 A2/A3/A23/A31/B7/B17/B13/B40/B21/B62 0.745 nd IgM,κ 
WIM8E5 A1/A10/A11/A9/A29/A30/A31/A33/A28d 0.578 nd IgG1,κ 
BVK1F9 B8 0.979 n/ae IgG1,κ 
a

mAbs with the same two- or three-letter prefix are from a single individual.

b

Correlation coefficient.

c

The epitopes involved in human mAb binding as determined by analysis of amino acid sequences of HLA class I molecules (5 ).

d

Plus HLA-A2 on cells homozygous for HLA-A2 only (15 of 16 individuals tested).

e

n/a, Not applicable.

Microtiter plates were coated with 1 μg of streptavidin (SA; Pierce) in 10 mM Tris (pH 9.0), blocked with 2% BSA in PBS, and washed with 0.05% Tween 20 in PBS. HLA-A2 monomers were bound at 16 ng/well for 2 h and incubated for 4 h with human mAbs in the presence of Complete (protease inhibitor mixture; Roche Applied Science), followed by HRP-labeled goat anti-human Ig (Southern Biotechnology Associates) for 90 min. Upon ABTS addition, color was developed for 1 h and read at 450 nm. The efficacy of binding of monomers to SA was determined with mAb TP25.99 followed by HRP-labeled goat anti-mouse IgG (Jackson ImmunoResearch Laboratories). All human mAb and TP25.99 incubations were done in duplicate. To correct for the various binding efficacies of monomers to SA (with the lowest binder, monomer-containing EBV peptide, and the best binder, monomer-containing HY peptide, yielding ODs of 0.65 and 1.42, respectively, in the TP25.99 assay), human mAb binding was reported as the ratio: (mean ODHuMAb − mean ODblank)/(mean ODTP25.99 − mean ODblank). None of the mAbs reacted with SA alone.

A single Ag-expressing line (SAL) was established as follows: a retroviral construct encoding HLA-A*0201 linked to the downstream internal ribosome entry sequence and the truncated form of NGFR (13) was transfected into φNX-A cells (provided by Dr. G. Nolan, Department of Microbiology and Immunology, School of Medicine, Stanford University, Stanford, CA). Transfectants were selected with puromycin (BD Clontech). Supernatant of transfected φNX-A was used to transduce the HLA class I and class II-negative erythroleukemia cell line K562. Transduced cells were sorted by FACS on the basis of NGFR positivity with mAb20.4 and are called SAL-A2. Renal cell carcinoma (RCC) lines MZ1774 and MZ1257 were made available by Dr. A. Knuth (University Hospital Zurich, Zurich, Switzerland), lymphoblastoid cell lines (LCL) were obtained from the 12th International Histocompatibility Workshop and a proximal tubular epithelial cell (PTEC) line VDS was established locally (14).

Human mAbs were incubated with 350,000 cells in 50 μl for 1 h on ice, washed, incubated with FITC-conjugated rabbit F(ab′)2 of anti-human IgG or IgM conjugates (DakoCytomation) on ice, washed, fixed with paraformaldehyde (1%), and analyzed on a FACSCalibur (BD Biosciences) equipped with CellQuest. For quantification of fluorescence intensity, the medians of total histograms of gated cells were recorded. Gated cells constituted 85% or more of the total cell population.

CDC on SAL-A2 was performed as described previously (15).

Aliquots of a precleared Nonidet P-40 lysate of 125I-labeled (by lactoperoxidase) LCL CALOGERO (HLA-A2,-, B61,-, Cw2,-) were sequentially incubated with suspensions of protein A-Sepharose CL-4B beads (Pharmacia) that had been preloaded with mAbs WIM8E5 or SN230G6. Upon washing, bound Ags were released from the beads in reducing sample buffer, electrophoresed on a 12% polyacrylamide-SDS gel, visualized by autoradiography, and quantified on a PhosphoImager (Molecular Dynamics).

Rank correlations were determined using Spearman’s rank test; Ab specificity correlations were determined using Pearson’s test with Yates correction (both from GraphPad InStat).

We produced a SAL (SAL-A2) by transduction of the HLA class I and II-negative cell line K562 with the HLA-A*0201 gene. FCM on this cell line, that expresses HLA-A2 with the K562-encoded repertoire of endogenously produced peptides, permits the relative quantification of HLA-A2 Ag recognized by various HLA-A2 mAbs. Eight human mAbs with HLA-A2 reactivity (Table II) showed reactivity by FCM with SAL-A2, but with a vast range of staining intensities (Table III). Differences in staining intensity were observed even among mAbs of identical specificity and isotype (compare mAbs ROU2D3 and WIM1B3). Titration experiments demonstrated that mAbs were used at saturating concentrations (data not shown). Six mAbs (ROUD2D3, SN230G6, SN607D8, SN66E3, WK4E3, and WK3D10) showed high staining intensity, exceeding a median channel value of 280, and 2 mAbs (WIM1B3 and WIM8E5) showed intermediate intensity. Three other mAbs (JOK3H4, Ha5C2, and JOK2C7) did not stain. Comparisons of staining intensities between isotype-matched mAbs revealed a ratio for WIM1B3:ROU2D3 (both IgM,λ) of 0.25 and likewise for WIM8E5:SN607D8 (both IgG1,κ) of 0.25. This suggests that mAbs WIM1B3 and WIM8E5 react with subsets of HLA-A2 molecules that contain a limited repertoire of bound peptides. In addition, we used CDC to test the lytic capacity of human mAbs on the SAL-A2 (Table III). The CDC reactivity correlated well with FCM staining intensity for all mAbs, except for two mAbs (HA5C2 and WIM1B3) that deviated considerably. Taken together, HLA-A2-reactive mAbs, even of identical isotype, react to varying degrees with HLA-A2 expressed on whole cells.

Table III.

Reactivity of human mAbs

mAbMedian fluorescenceaNo. of Monomers Detected by ELISAbCDCc
BVK1F9 
JOK2C7 11 
JOK3H4 11 
Ha5C2 12 17 
WIM1B3 81 99 
WIM8E5 114 50 
ROU2D3 286 89 
WK4E3 309 10 86 
WK3D10 339 100 
SN66E3 392 11 87 
SN607D8 419 12 74 
SN230G6 429 12 100 
mAbMedian fluorescenceaNo. of Monomers Detected by ELISAbCDCc
BVK1F9 
JOK2C7 11 
JOK3H4 11 
Ha5C2 12 17 
WIM1B3 81 99 
WIM8E5 114 50 
ROU2D3 286 89 
WK4E3 309 10 86 
WK3D10 339 100 
SN66E3 392 11 87 
SN607D8 419 12 74 
SN230G6 429 12 100 
a

Determined on SAL-A2; mAbs sorted by these data.

b

OD ratio (human mAb:TP25) of 0.100 was used as cutoff value for positivity.

c

Percentage of SAL-A2 cells lysed in CDC.

The availability of a panel of HLA-A2 monomers, each containing a different peptide, permitted analysis of the peptide selectivity of HLA-A2-specific human mAbs by ELISA. These 12 peptides (Table I), all containing HLA-A2 binding motifs, were based on viral, bacterial, and autoantigen-derived sequences. The monomer reactivity of human mAbs, expressed as an OD ratio is given in Fig. 1. We considered a monomeric HLA-A2/human mAb combination that showed an OD ratio of 0.100 or higher as positive. With this criterion, eight human mAbs showed reactivity with the monomers, but with vastly different reactivity patterns. These ranged from reactivity with the complete monomer panel, as exemplified in Fig. 1, A–C, by the three mAbs derived from donor SN (with mAb SN66E3 reacting with 11 of12 monomers), to reactivity with only one or three monomers as shown by the mAbs WIM1B3 and WIM8E5, respectively (Fig. 1, E and F). Intermediate degrees of selectivity were shown by 3 mAbs: the mAb ROU2D3 reactive with 7 of 12 monomers (Fig. 1,D), the mAb WK3D10 (8 monomers, Fig. 1,G), and WK4E3 (10 monomers, Fig. 1,H). The three remaining CDC defined, HLA-A2-reactive mAbs JOK2C7, JOK3H4, and Ha5C2 and the HLA-B8-reactive mAb BVK1F9 did not bind to any monomer (data not shown). The peptide selectivity shown by some mAbs is likely due to critical contributions of certain amino acids in the HLA-A2-bound peptides. This is illustrated for mAb WK4E3 that was intolerant for a unique combination of amino acids P and E in positions 6 and 7 in the two nonreactive HLA-A2-bound peptides (Table IV). The peptide selectivity was not confined to these positions for the other selective mAbs (data not shown).

FIGURE 1.

Reactivity of human mAbs with HLA-A monomers. Monomer-embedded peptides are given on the y-axis. Results are expressed as OD ratio (see Materials and Methods). Only mAbs causing an OD ratio >0.1 for one or more monomers are shown.

FIGURE 1.

Reactivity of human mAbs with HLA-A monomers. Monomer-embedded peptides are given on the y-axis. Results are expressed as OD ratio (see Materials and Methods). Only mAbs causing an OD ratio >0.1 for one or more monomers are shown.

Close modal
Table IV.

Analysis of aminoacids critical in binding of mAb WK4E3 to monomersa

PeptideWK4E3 OD RatioAmino Acid Position
12345678910
HY 0.567  
EBV 0.501  
EII49 0.484 
IB.54 0.412  
HA-2 0.355  
HA-1h 0.333  
PR-1 0.321  
HIV 0.299  
hCMV 0.280  
Flue 0.245  
HPV E7 0.056 P E  
PRA 0.040 P E 
PeptideWK4E3 OD RatioAmino Acid Position
12345678910
HY 0.567  
EBV 0.501  
EII49 0.484 
IB.54 0.412  
HA-2 0.355  
HA-1h 0.333  
PR-1 0.321  
HIV 0.299  
hCMV 0.280  
Flue 0.245  
HPV E7 0.056 P E  
PRA 0.040 P E 
a

Amino acids unique for preventing mAb binding are in bold. Monomers were sorted by mAb reactivity.

The number of monomers detected by each mAb in ELISA is included in Table III. When fluorescence intensity on SAL-A2 and the number of reactive monomers were compared for each HLA-A2 mAb, these values showed extremely good correlation (r = 0.800, p = 0.003 by Spearman’s rank test, Fig. 2). These data show that the amino acid sequence of the peptide bound in HLA-A2 dictates binding of the human mAbs to these molecules.

FIGURE 2.

Correlation of fluorescence intensity caused by mAbs on SAL-A2 and monomer panel binding of mAbs. (mAb BVK1F9 not included in this analysis).

FIGURE 2.

Correlation of fluorescence intensity caused by mAbs on SAL-A2 and monomer panel binding of mAbs. (mAb BVK1F9 not included in this analysis).

Close modal

To probe the relationship of HLA-A2 molecules that are recognizable by various HLA-A2 mAbs on cell surfaces, sequential immunoprecipitation was performed with two human mAbs of the IgG isotype on the HLA-A2 homozygous LCL CALOGERO. Exposing the CALOGERO lysate to three rounds of mAb SN230G6 resulted in precipitation of all available HLA-A2 molecules, as evidenced by bands visible at Mr 46,000 (H chain) and 13,000 (β2-microglobulin), but no bands visible upon three subsequent rounds of mAb WIM8E5 (Fig. 3,A). When the order of mAbs was reversed (Fig. 3 B), three rounds of WIM8E5 precipitated only a fraction of HLA-A2 visible at Mr 46,000 and 13,000 because three subsequent rounds of SN230G6 precipitated the preponderance of material at Mr 46,000 and 13,000. Quantification of bands in the latter sequential experiment showed that WIM8E5 precipitated 13% of the total H chain precipitated by WIM8E5 and SN230G6 together. Likewise, WIM8E5 precipitated 25% of the total β2-microglobulin. Thus, the WIM8E5 recognizable HLA-A2 molecules are a subset of the SN230G6 recognizable molecules.

FIGURE 3.

Immunoprecipitation on lysate of 125I-labeled CALOGERO, with two human mAbs analyzed by PAGE. A, Three rounds of immobilized SN230G6 (lanes 1–3) followed by three rounds of WIM8E5 (lanes 4–6). B, Three rounds of WIM8E5 (lanes 1–3) followed by three rounds of SN230G6 (lanes 4–6). Positions of H chain (HC, Mr 46,000) and β2-microglobulin (β2M; Mr 13,000) are interpolated from the positions of Mr markers in a separate lane.

FIGURE 3.

Immunoprecipitation on lysate of 125I-labeled CALOGERO, with two human mAbs analyzed by PAGE. A, Three rounds of immobilized SN230G6 (lanes 1–3) followed by three rounds of WIM8E5 (lanes 4–6). B, Three rounds of WIM8E5 (lanes 1–3) followed by three rounds of SN230G6 (lanes 4–6). Positions of H chain (HC, Mr 46,000) and β2-microglobulin (β2M; Mr 13,000) are interpolated from the positions of Mr markers in a separate lane.

Close modal

Cell surface-expressed HLA-A2 contains peptides derived from household proteins, virus-encoded proteins, and proteins uniquely expressed in cell lineages. HLA-A2 mAbs that show peptide selectivity also may harbor selectivity for certain tissues. To test this hypothesis, we quantified the expression of HLA-A2 by FCM on cell lines from various lineages. Ratios of expression of HLA-A2 calculated from peptide-selective mAbs are given in Table V. Although mAb WIM8E5 detects a small fraction of SN230G6-detectable HLA-A2 molecules on SAL-A2 (Fig. 4,A) and HLA-A2 homozygous LCL CALOGERO (Fig. 4,B), it detects a larger proportion of the HLA-A2 molecules expressed on the PTEC line VDS (Fig. 4,C). In contrast, WIM1B3 detects a larger fraction of ROU2D3-detectable HLA-A2, both on RCC lines (MZ1774 shown in Fig. 5,A) and on LCL (CALOGERO shown in Fig. 5,B), than on SAL-A2 (Fig. 5,C). The latter indicates that WIM1B3 and WIM8E5 exhibit diverse preferences for peptides. In contrast, the ratios of expression of HLA-A2 as detected by the two mAbs SN230G6 and SN607D8 were around 1.0 throughout this range of cell lineages (Table V), which is in line with the finding that these mAbs were not peptide selective on the monomer HLA-A2 panel.

Table V.

Differential binding of mAbs to cell lines of various lineages

DesignationCell TypeHLA TypeRatioa
mAb1→WIM8E5WIM1B3WIM1B3SN230G6
mAb2→SN230G6ROU2D3SN66E3SN607D8
MZ1774 RCC A2,A30 B13,B18  nib 0.73 0.60 0.87 
MZ1257 RCC A2,A3 B7,B44  0.26 0.42 0.29 1.13 
VDS PTEC A*02,A*03 B*0702,B*4402  0.77 0.15 0.10 0.84 
AWELLS LCL A2,- B44,-  0.26 0.69 0.58 0.97 
CALOGERO LCL A2,- B61,-  0.18 0.83 0.48 1.00 
WT24 LCL A2,- B27,-  0.26 0.92 0.57 1.08 
SAL-A2 Erythroid A*0201   0.25 0.26 0.19 1.03 
DesignationCell TypeHLA TypeRatioa
mAb1→WIM8E5WIM1B3WIM1B3SN230G6
mAb2→SN230G6ROU2D3SN66E3SN607D8
MZ1774 RCC A2,A30 B13,B18  nib 0.73 0.60 0.87 
MZ1257 RCC A2,A3 B7,B44  0.26 0.42 0.29 1.13 
VDS PTEC A*02,A*03 B*0702,B*4402  0.77 0.15 0.10 0.84 
AWELLS LCL A2,- B44,-  0.26 0.69 0.58 0.97 
CALOGERO LCL A2,- B61,-  0.18 0.83 0.48 1.00 
WT24 LCL A2,- B27,-  0.26 0.92 0.57 1.08 
SAL-A2 Erythroid A*0201   0.25 0.26 0.19 1.03 
a

Ratios of reactivity were determined by calculating (median mAb1 − median (negative control)/median mAb2 − median negative control) in which negative control was cells with conjugate only.

b

ni, Not informative.

FIGURE 4.

Flow cytometric characteristics of IgG isotype human mAbs. Solid line, SN230G6; gray area, WIM8E5; stippled, control. A, SAL-A2; B, CALOGERO; C, VDS.

FIGURE 4.

Flow cytometric characteristics of IgG isotype human mAbs. Solid line, SN230G6; gray area, WIM8E5; stippled, control. A, SAL-A2; B, CALOGERO; C, VDS.

Close modal
FIGURE 5.

Flow cytometric characteristics of IgM isotype human mAbs. Solid line, ROU2D3; gray area, WIM1B3; stippled, control. A, MZ1774; B, CALOGERO; C, SAL-A2.

FIGURE 5.

Flow cytometric characteristics of IgM isotype human mAbs. Solid line, ROU2D3; gray area, WIM1B3; stippled, control. A, MZ1774; B, CALOGERO; C, SAL-A2.

Close modal

Cell surface-expressed HLA class I encoded by a single allele is a heterogeneous set of molecules presenting an array of peptides generated from intracellularly processed proteins. HLA class I molecules of a single allele, but from different organs, contain different arrays of peptides and these differences are detectable by CTL (16). In this study, we present evidence that human HLA Abs have also have the capacity to distinguish peptide-induced alterations. Although the set of monomers used in the ELISA was limited, the human mAbs show a range of peptides selectivities that has been found in allo-CTL clones also (3). The combined use of human mAbs, HLA-A2 monomers, and the single Ag transductant was instrumental in demonstrating the decisive role of peptides in the recognition of MHC molecules by human alloantibodies. We observed differential binding of six HLA-A2-reactive mAbs with the HLA-A2 monomer panel. There was no relationship between the monomer reaction patterns and the HLA-A2 epitope specificity or the Ig isotype of the human mAbs. When these mAbs were confronted with HLA-A2 bearing a multitude of peptides, various subsets of HLA-A2 molecules reacted, yielding a range of staining intensities by FCM on the SAL-A2. Ratios of staining intensities were calculated for mAbs of identical isotype. Strikingly, biochemical analysis and FCM with two IgG mAbs, WIM8E5 and SN230G6, showed that in these divergent techniques, ratios of reactivity were in the same range. Furthermore, HLA-A2 molecules recognized by one mAb are nested within the larger set of HLA-A2 molecules seen by the other mAb. In this respect it is remarkable that the rank order of Ag densities reactive with mAbs as detected by FCM on SAL-A2, almost completely concurs with the rank order of the number of monomers recognized in the ELISA by the mAbs (Fig. 2). Three mAbs that react with HLA-A2+ lymphocytes by CDC failed to react with SAL-A2 by FCM and CDC, suggesting their preference for peptides that are not sufficiently provided in the erythroid precursor environment of K562. In contrast, the degree of susceptibility of SAL-A2 to complement-dependent lysis is only partially reflected in the number of monomers reactive with a given mAb. The notable exception here is mAb WIM1B3 which causes complete lysis of SAL-A2, yet binds only one monomer of the panel (Table III). Similar to CTL-induced lysis, complement-dependent lysis may require only a limited density of Ab-reactive Ags on the cell surface. That this threshold is low is underpinned by the fact that one mAb, Ha5C2, is able to cause a low degree of lysis by CDC on SAL-A2, despite lack of FCM staining.

The differentially reactive mAbs clearly see different subsets of cell surface-expressed HLA-A2. Several mechanisms may underlie differential mAb binding to HLA-A2 and these include peptide-induced conformational changes causing epitope disruption, steric hindrance by large side chains of peptides resulting in epitope blocking, and requirement for mAb to interact with Ag by two CDRs at distinct antigenic sites. Our exploratory study does not allow favoring any of these possibilities. The influence of peptide-induced conformational changes in MHC class I molecules on alloantibody binding has been shown with murine Kd-reactive mAbs (7). Another study with Kb-specific mAbs showed that by increasing the number of peptides, a larger number of Kb-reactive mAbs appear to be peptide dependent. Peptide pools eluted from Kb molecules that were immunoprecipitated with a peptide-selective mAb were successfully used to reconstitute empty Kb molecules on T2-Kb targets that were lysable by some allo-CTL clones (17). This suggests a relationship between alloreactive CTL and alloantibody recognition of MHC class I. Even deliberate alloimmunization and boosting with the SIINFEKL peptide-loaded tumor RMA-S yielded a peptide-MHC class I complex mAb (18) that nonetheless turned out to be tolerant to single amino acid substitutions (19). Peptide selectivity by mAbs has also been demonstrated in xenogeneic (mouse anti-HLA-B27) combinations (20, 21).

To our knowledge, the present study is the first demonstration of peptide selectivity of human HLA-alloreactive Abs and corroborates an earlier study (22) that showed peptide (des-Tyr1-γ-endorphin) blocking of HLA allosera. In this study, we took advantage of availability of recombinant monomeric HLA-A2 which provides a more convenient platform for peptide studies, compared with peptide reconstitution of TAP-deficient cell lines. Our results are at variance with data published by others. Using a similar ELISA method for screening sera for HLA Abs, Barnardo et al. (23) failed to observe a difference in serum Ab binding with HLA-A11 monomers (two peptides tested). Likewise, an anti HLA-A11 human mAb (24) detected no differences in cell surface-expressed, empty HLA-A11 molecules that were reconstituted with an array of peptides, while in the same study mouse mAbs of identical specificity were discriminatory. The findings of these two groups may have been fortuitous, as our present observation of five differentially HLA-A2-reactive human mAbs, which were derived from three different immune donors, indicates that peptide selectivity of HLA Abs is no isolated incident. This is further supported by preliminary data demonstrating that more than one-half of the HLA Abs present in supernatants of monoclonally seeded B cells of donor ROU, cultured after tetrameric HLA-A2-guided isolation (25), were peptide selective. (C. Eijsink and A. Mulder, unpublished observations). The human mAbs described here were derived from women who had been immunized by pregnancy. The fact that human mAbs with identical specificity (e.g., for the HLA-A2/B17 shared epitope) and isotype, but derived from different donors, show different monomer panel reactivity and different staining intensity by FCM suggests that in each donor a different set of peptides embedded in the allogeneic HLA-A2 molecule was responsible for the immunization that caused the clonal B cell expansion that we subsequently immortalized. Curiously, one individual, SN, yielded three HLA-A2 mAbs that were tolerant to peptide exchange, whereas the HLA-A2 mAbs generated from three other individuals were peptide selective. HLA Abs require one CDR for serological epitope recognition, with a critical contribution of a second CDR to stabilize Ag binding (26). An attractive hypothesis is that the SN-derived mAbs contain a second CDR that contacts a nonpolymorphic determinant exposed on the HLA backbone, but that the second CDR of the other mAbs interact with solvent-exposed amino acids of the embedded peptide. We expect Abs with even greater peptide-induced stringency to be present in the sera of women after HLA-A2 syngeneic pregnancies. The existence of such MHC class I-restricted humoral response was demonstrated in a serum Ab with HY specificity after bone marrow transplantation (27). More recently, phage display technology has enabled the isolation of recombinant Abs specific for the gp100 melanoma peptide-HLA-A2 complex from a nonimmune library (28), supporting the notion that Abs with TCR-like specificity are part of the normal humoral repertoire. The increased affinity of these TCR-like Abs (as compared with TCR themselves) makes them attractive tools for immunotherapy. The ELISA based on monomeric HLA class I loaded with peptides relevant to the purpose will be a useful tool for additional specificity testing of these novel Abs.

Besides advancing the knowledge of molecular interactions in humoral HLA recognition, a clinically relevant notion emerges from the present study. The peptide dependence of some human HLA-A2 mAbs implies that HLA Abs may harbor a certain degree of tissue specificity. Our FCM data on cells of various lineages confirm this concept. Having established that mAbs WIM1B3 and WIM8E5 recognize only fractions of total available HLA-A2 molecules on the erythroid lineage cell SAL-A2, we demonstrated that these two selective mAbs show increases in reactive fractions of HLA-A2 on cells of other lineages. However, among these two mAbs, the patterns of cell lineage preference were different from each other. The monomer panel reactivity patterns (Fig. 1, E and F) of these mAbs confirm this difference. Apparently, these two mAbs have preference for peptides that are expressed with greater abundance on renal epithelial cells from malignant (RCC) and primary origin (PTEC) than on the SAL-A2.

The expression of lineage-specific and/or virally encoded peptides in HLA-A2 is therefore the likely cause of this selectivity. Biochemical studies, with peptides that are derived from proteins uniquely expressed in relevant cell types and incorporated in HLA-A2 monomers should prove or negate this concept. If such tissue specificity of Abs is shown to exist, the implications for solid organ transplantation are far-reaching. Although Ab screening and cross-matching is performed on lymphocytes, the organ to be transplanted likely contains a set of peptides embedded in its HLA class I molecules that is partially different from the set of lymphocyte-expressed peptides. It is at least theoretically possible that patients who have been transplanted with a positive serological cross-match on lymphocytes, nonetheless have a well-functioning graft. By reanalyzing their pretransplant sera for the presence of Ab reactivity on relevant monomers, one would gain insight in the issue of tissue-specific Abs. In this light, it is important to note that tissue specificity has been observed in alloreactive CTLs (29, 30).

We are grateful to Dr. Alexander Knuth for making available RCC cell lines. We thank Drs. Dave Roelen and Jeroen van Bergen for critical reading of this manuscript.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

3

Abbreviations used in this paper: CDC, complement-dependent cytotoxicity; FCM, flow cytometry, LCL, lymphoblastoid cell line; NGFR, nerve growth factor receptor; PTEC, proximal tubular epithelial cell; RCC, renal cell carcinoma; SA, streptavidin; SAL, single Ag-expressing line.

1
van Rood, J. J., J. G. Eernisse, A. van Leeuwen.
1958
. Leukocyte antibodies in sera from pregnant women.
Nature
181
:
1735
.-1736.
2
Dankers, M. K., D. L. Roelen, N. Korfage, P. de Lange, M. Witvliet, L. Sandkuijl, I. I. Doxiadis, F. H. Claas.
2003
. Differential immunogenicity of paternal HLA class I antigens in pregnant women.
Hum. Immunol.
64
:
600
.-606.
3
Moris, A., V. Teichgraber, L. Gauthier, H. J. Buhring, H. G. Rammensee.
2001
. Characterization of allorestricted and peptide-selective alloreactive T cells using HLA-tetramer selection.
J. Immunol.
166
:
4818
.-4821.
4
Madden, D. R., D. N. Garboczi, D. C. Wiley.
1993
. The antigenic identity of peptide-MHC complexes: a comparison of the conformations of five viral peptides presented by HLA-A2.
Cell
19: 75
:
693
.-708.
5
Fremont, D. H., M. Matsumura, E. A. Stura, P. A. Peterson, I. A. Wilson.
1992
. Crystal structures of two viral peptides in complex with murine MHC class I H-2Kb.
Science
257
:
919
.-927.
6
Hogquist, K. A., A. G. Grandea, III, M. J. Bevan.
1993
. Peptide variants reveal how antibodies recognize major histocompatibility complex class I.
Eur. J. Immunol.
23
:
3028
.-3036.
7
Noun, G., M. Reboul, J. P. Abastado, C. Jaulin, P. Kourilsky, M. Pla.
1996
. Alloreactive monoclonal antibodies select Kd molecules with different peptide profiles.
J. Immunol.
157
:
2455
.-2461.
8
Tongio, M. M., I. I. Doxiadis, M. Laforet, G. M. T. Schreuder, N. Froehlich, G. Semana, L. Pratico, R. Fauchet.
1997
. 12th International Histocompatibility Workshop HLA class I monoclonal antibodies study. D. Charron, III, ed.
Genetic Diversity of HLA: Functional and Medical Implication
7
.-10. EDK Medical and Scientific Publisher, Paris.
9
Altman, J. D., P. A. H. Moss, P. J. R. Goulder, D. H. Barouch, M. G. McHeyzer-Williams, J. I. Bell, A. J. McMichael, M. M. Davis.
1996
. Phenotypic analysis of antigen-specific T lymphocytes.
Science
274
:
94
.-96.
10
Fields, G. B., R. L. Noble.
1990
. Solid phase peptide synthesis utilizing 9-fluorenylmethoxycarbonyl amino acids.
Int. J. Pept. Protein Res.
35
:
161
.-214.
11
Mulder, A., M. Kardol, J. Blom, W. B. Jolley, C. J. Melief, J. W. Bruning.
1993
. Characterization of two human monoclonal antibodies reactive with HLA-B12 and HLA-B60, respectively, raised by in vitro secondary immunization of peripheral blood lymphocytes.
Hum. Immunol.
36
:
186
.-192.
12
D’Urso, C. M., Z. G. Wang, Y. Cao, R. Tatake, R. A. Zeff, S. Ferrone.
1991
. Lack of HLA class I antigen expression by cultured melanoma cells FO-1 due to a defect in B2m gene expression.
J. Clin. Invest.
87
:
284
.-292.
13
Ruggieri, L., A. Aiuti, M. Salomoni, E. Zappone, G. Ferrari, C. Bordignon.
1997
. Cell-surface marking of CD34+-restricted phenotypes of human hematopoietic progenitor cells by retrovirus-mediated gene transfer.
Hum. Gene Ther.
8
:
1611
.-1623.
14
Woltman, A. M., S. de Haij, J. G. Boonstra, S. J. Gobin, M. R. Daha, C. van Kooten.
2000
. Interleukin-17 and CD40-ligand synergistically enhance cytokine and chemokine production by renal epithelial cells.
J. Am. Soc. Nephrol.
11
:
2044
.-2055.
15
Bruning, J. W., F. H. Claas, M. J. Kardol, Q. Lansbergen, A. M. Naipal, H. J. Tanke.
1982
. Automated reading of HLA-A,B,C typing and screening: the propidium iodide (PI) method.
Hum. Immunol.
5
:
225
.-231.
16
de Bueger, M., A. Bakker, J. J. van Rood, W. F. van der Woude, E. Goulmy.
1992
. Tissue distribution of human minor histocompatibility antigens: ubiquitous versus restricted tissue distribution indicates heterogeneity among human cytotoxic T lymphocyte-defined non-MHC antigens.
J. Immunol.
149
:
1788
.-1794.
17
Sherman, L. A., S. Chattopadhyay, J. A. Biggs, R. F. Dick, J. A. Bluestone.
1993
. Alloantibodies can discriminate class I major histocompatibility complex molecules associated with various endogenous peptides.
Proc. Natl. Acad. Sci. USA
90
:
6949
.-6951.
18
Porgador, A., J. W. Yewdell, Y. Deng, J. R. Bennink, R. N. Germain.
1997
. Localization, quantitation, and in situ detection of specific peptide-MHC class I complexes using a monoclonal antibody.
Immunity
6
:
715
.-726.
19
Messaoudi, I., J. LeMaoult, J. Nikolic-Zugic.
1999
. The mode of ligand recognition by two peptide:MHC class I-specific monoclonal antibodies.
J. Immunol.
163
:
3286
.-3294.
20
Wang, J., D. T. Yu, T. Fukazawa, H. Kellner, J. Wen, X. K. Cheng, G. Roth, K. M. Williams, R. B. Raybourne.
1994
. A monoclonal antibody that recognizes HLA-B27 in the context of peptides.
J. Immunol.
152
:
1197
.-1205.
21
Urban, R. G., R. M. Chicz, W. S. Lane, J. L. Strominger, A. Rehm, M. J. Kenter, F. G. UytdeHaag, H. Ploegh, B. Uchanska-Ziegler, A. Ziegler.
1994
. A subset of HLA-B27 molecules contains peptides much longer than nonamers.
Proc. Natl. Acad. Sci. USA
91
:
1534
.-1538.
22
Claas, F. H., J. J. van der Poel, R. Castelli-Visser, J. Pool, R. B. Chen, K. Y. Xu, J. J. van Rood.
1985
. Interaction between des-Tyr1-γ-endorphin and HLA class I molecules: serological detection of an HLA-A2 subtype.
Immunogenetics
22
:
309
.-314.
23
Barnardo, M. C., A. W. Harmer, O. J. Shaw, G. S. Ogg, M. Bunce, R. W. Vaughan, P. J. Morris, K. I. Welsh.
2000
. Detection of HLA-specific IgG antibodies using single recombinant HLA alleles: the MonoLISA assay.
Transplantation
70
:
531
.-536.
24
Zhang, Q. J., Y. Lindquist, V. Levitsky, M. G. Masucci.
1996
. Solvent exposed side chains of peptides bound to HLA A*1101 have similar effects on the reactivity of alloantibodies and specific TCR.
Int. Immunol.
8
:
927
.-938.
25
Mulder, A., C. Eijsink, M. J. Kardol, M. E. Franke-van Dijk, S. H. van der Burg, M. Kester, I. I. Doxiadis, F. H. Claas.
2003
. Identification, isolation, and culture of HLA-A2-specific B lymphocytes using MHC class I tetramers.
J. Immunol.
171
:
6599
.-6603.
26
Duquesnoy, R. J., A. Mulder, M. Askar, M. Fernandez-Vina, F. H. J. Claas.
2005
. HLAMatchmaker-based analysis of human monoclonal antibody reactivity demonstrates the importance of an additional contact site for specific recognition of triplet-defined epitopes.
Hum. Immunol.
66
:
749
.-761.
27
van Leeuwen, A., E. Goulmy, J. J. van Rood.
1979
. Major histocompatibility complex-restricted antibody reactivity mainly, but not exclusively, directed against cells from male donors.
J Exp. Med.
150
:
1075
.-1083.
28
Denkberg, G., C. J. Cohen, A. Lev, P. Chames, H. R. Hoogenboom, Y. Reiter.
2002
. Direct visualization of distinct T cell epitopes derived from a melanoma tumor-associated antigen by using human recombinant antibodies with MHC- restricted T cell receptor-like specificity.
Proc. Natl. Acad. Sci.
99
:
9421
.-9426.
29
Deckers, J. G., M. R. Daha, S. W. van der Kooij, F. J. van der Woude.
1998
. Epithelial- and endothelial-cell specificity of renal graft infiltrating T cells.
Clin. Transplant.
12
:
285
.-291.
30
Nguyen, T., B. Naziruddin, S. Dintzis, G. M. Doherty, T. Mohanakumar.
1999
. Recognition of breast cancer-associated peptides by tumor-reactive, HLA-class I restricted allogeneic cytotoxic T lymphocytes.
Int. J. Cancer
81
:
607
.-615.