HLA Epitopes: The Targets of Monoclonal and Alloantibodies Defined

Sensitization to human leukocyte antigens (HLA) in organ transplant patients causes graft rejection, according to the humoral theory of transplantation. Sensitization is almost ubiquitous as anti-HLA antibodies are found in almost all sera of transplant recipients. Advances in testing assays and amino acid sequencing of HLA along with computer software contributed further to the understanding of antibody-antigen reactivity. It is commonly understood that antibodies bind to HLA antigens. With current knowledge of epitopes, it is more accurate to describe that antibodies bind to their target epitopes on the surface of HLA molecular chains. Epitopes are present on a single HLA (private epitope) or shared by multiple antigens (public epitope). The phenomenon of cross-reactivity in HLA testing, often explained as cross-reactive groups (CREGs) of antigens with antibody, can be clearly explained now by public epitopes. Since 2006, we defined and reported 194 HLA class I unique epitopes, including 56 cryptic epitopes on dissociated HLA class I heavy chains, 83 HLA class II epitopes, 60 epitopes on HLA-DRB1, 15 epitopes on HLA-DQB1, 3 epitopes on HLA-DQA1, 5 epitopes on HLA-DPB1, and 7 MICA epitopes. In this paper, we provide a summary of our findings.


Introduction
Sensitization to HLA antigens in organ transplant patients causes graft rejection, according to the humoral theory of transplantation [1]. Sensitization is almost ubiquitous as it is evident in the detection of anti-HLA antibodies in the sera of recipients-in one study, almost all patients waiting for regraft of a kidney transplant have anti-HLA antibodies [2]. Determining specificity of the anti-HLA antibody has advanced in recent years using recombinant HLA single antigens (SA) coated on color-coded Luminex beads [3]. The reactivity of anti-HLA antibodies with HLA antigens and the phenomenon of cross-reactivity has been the subject of investigation for decades. Amino acid sequences of the HLA molecules which greatly contributed to our understanding of antibody and antigen reactivity has been introduced since 1963 [4][5][6][7][8][9][10][11][12]. Antibodies are commonly described as binding to HLA antigens; however, it is more accurate to describe the reactivity of the antibody as binding to specific epitopes on the surface of HLA antigens-epitopes are conformational amino acid arrangements and are the targets of antibodies. Some epitopes are private, found exclusively on one antigen; others are public epitopes shared by two or more antigens. The phenomenon of cross-reactivity in HLA testing, often explained as crossreactive groups (CREGs), of antigens with antibody can be clearly explained now by public epitopes-an antibody targeting a public epitope shows positive reaction with all antigens sharing the epitope.
Since 2006, we defined and reported on 194 HLA class I unique epitopes, including 138 epitopes on intact HLA class I (heavy chain + β2m + peptide), and 56 cryptic epitopes on dissociated HLA class I (heavy chain only) [13][14][15][16][17][18]. 110 epitopes on intact HLA class I were defined using murine monoclonal and human alloantibodies, and the remaining 28 epitopes were defined with naturally occurring anti-HLA antibodies. Naturally occurring (natural) HLA antibodies found in cord blood and healthy males were used to define the 56 cryptic epitopes on dissociated HLA class I. In addition, 83 HLA class II unique epitopes were defined and reported, including 60 epitopes on HLA-DRB1, 15 epitopes on HLA-DQB1, 3 epitopes on HLA-DQA1, and 5 epitopes on HLA-DPB1 [15,[19][20][21][22]. All HLA-DRB1 epitopes were defined using solely amino acid sequence data, in contrast to HLA-DQA1, HLA-DQB1, and HLA-DPB1 epitopes that were defined using human alloantibodies. Lastly, we defined and reported on 7 MICA epitopes using human alloantibodies [15,22]. In this paper, we provide a summary of our findings.

Materials and Methods
The principle we used to define HLA epitopes is summarized in (Figure 1). Briefly, if an antibody is determined to test positive with certain HLA antigens and negative with others, it is reasonable to assume that the antibody is targeting a specific epitope on the positive antigens. Epitopes are conformational arrangements of amino acids (aa) at sequence positions on the surface of antigens that must be within the binding span of the antibody. To define an epitope, a computer search, in published aa sequences of tested antigens, was performed to identify exclusively shared aa at one or more sequence positions among the positive antigens-these amino acids define the epitope.
Murine monoclonal antibodies or transplant recipient and healthy male HLA antibodies isolated from sera and cord blood by first adsorbing them onto appropriate recombinant HLA (rHLA) single antigen cells, then eluted with an acidic buffer (ImmunoPure IgG elution buffer, Pierce, Rockford, IL), and neutralized with 1 M TRIS-HCl pH 9.5 ( Figure 2) were all tested with the single antigen beads (One Lambda Inc., Canoga Park, CA) to determine the specificity of the antibodies [14]. HLA class I SA beads treated with a buffer that dissociates the peptide and the beta-2-microglobulin (β2m) from the heavy chain of the intact HLA antigens on the beads [17] were used to reveal the specificity of antibodies targeting epitopes on dissociated class I heavy chains. MFI values of 1000 or above were considered positive except when the overall reactions of an eluted antibody were weak, a cutoff of MFI 400 was used.
Computer software was utilized to search for exclusive amino acids in the structure of antigens showing positive reactions with an antibody. Searches were performed within sequences of HLA class I heavy chains, MICA antigens, DR beta chains, DQ beta and alpha chains, and DPB chains. All amino acid sequences were obtained from the HLA Informatics Group at the Anthony Nolan website [23]. One or more amino acids found exclusively at the same sequence positions in the chains of positive antigens, but not in the sequence positions of negative antigens, were designated as the defining amino acids for an epitope. The defining amino acid(s) had to be within the antibody binding span [24,25] estimated at 494 Å2-750 Å2 area ( Figure 3) and the aa(s) must be exposed at the surface of the antigen-exceptions are noted between parentheses (Table 1).
The efficacy of isolating HLA antibody from HLA sera with adsorption and elution assays, testing the eluted antibody with the SA beads to determine specificity and the definition of the epitope on the surface of positive antigens (corresponding to antibody specificity) are shown in (Figure 4). Alloserum with determined specificity A2, A68, A69, B57, and B58 was adsorbed separately with SA rHLA Main steps for epitope di nition Monoclonal antibody or alloantibody isolated from sera by adsorption-to then elution-from a single antigen recombinant cell line (Figure 2).
Test with 96 single antigen bead panels "Each single antigen bead has a unique HLA antigen (allele) attached" In this case, the results in the gure show that A2, 68, and 69 beads/antigen are positve while the rest (90 beads, not all shown) are negative-an indication that the antibody is targeting a unique epitope on the positive antigens.
(1) Antibody preparation and testing with single antigen beads aa molecular structure so ware to determine the following: (i) Amino acids at the exclusicely shared positions of the positive beads are exposed to the surface of the antigens for antibody to bind. epitope, the distance between any two aa's is within the binding span of  A6901 and B5801 cells. Eluted antibodies tested with the SA beads showed specificity A2, A68, and A69 and A2, B57, and B58, respectively. HLA antigens A2, A68, and A69 share an epitope defined by glycine (G) at position 62; therefore, 62G defines the epitope. Similarly, HLA antigens A2, B57, and B58 share an epitope defined by threonine (T) at position 142 or histidine (H) at position 145; therefore 142T or 145H define the epitope.

Results
3.1. Class I Epitopes on Intact Antigens. 138 unique epitopes were defined for one or a group of two or more intact HLA class I antigens. 110 unique epitopes were defined by using SA beads (Table 1, partial list; complete table in the supplemental information available online at https://doi.org/10.1155/2017/3406230) assays to test eluted alloantibodies that were adsorbed from human sera onto the surface of mammalian rHLA single antigen cells then eluted, and murine monoclonal antibodies to determine specificity of each antibody. Epitopes were defined by identifying exclusively unique amino acids among the positive antigens. Also defined were 28 unique epitopes targeted by naturally occurring anti-HLA antibodies found in sera of healthy males and in cord blood ( Table 2). All epitopes were defined by identifying exclusively unique amino acids among the positive antigens. Here, we present partial lists in tables and example figures of epitopes-complete tables and other figures can be found in the supplemental information document.
The number of epitopes defined for each antigen, using human alloantibodies, varied from 4 to 23 (Table 3). In general, there was no correlation between the number of epitopes and the frequency of antigen in the population. For example, for HLA A2, the most frequent antigen (f = 30.3% to 54%), we defined 16 epitopes while for A25,     indicates two or more positions/aa needed to define the epitope; amino acids not exposed at the surface of the HLA molecule are between parentheses; c epitope also shared by C-locus antigens (not shown here) is between square brackets.
with a frequency of f = 0.0% to 6.1%, we defined 19 epitopes. Class I epitopes were found to be shared by antigens of the same locus or by inter-locus antigens-BC or ABC. Epitopes are defined by 1, 2, 3, or 4 aa's. For example, 17 A-locus epitopes were defined by 1 aa, two B-locus epitopes by 4 aa's, or two ABC loci epitopes defined by 2 aa's. In addition, amino acids and positions on the HLA class I heavy chain epitopes were found at varying frequencies in epitope definitions.
The most frequent was position 163 located in the alpha 2 domain, and the aa threonine (T) was found to be the most frequent in our studies.
The following examples illustrate HLA class I epitopes for the A-locus, B-locus, and C-locus and AB-, BC-, and ABCloci antigens. Illustration shows SA beads specificity, antigens sharing the epitopes, and their position on the HLA class I heavy chain.
Single antigen bead speci city

62G
Adsorbed/eluted with B5801 Single antigen bead speci city  Plus sign "+" indicates two or more positions/aa needed to define the epitope; amino acids not exposed at the surface of the HLA molecule are between parentheses.

Cryptic Epitopes on Dissociated HLA Class I Antigens.
Naturally occurring anti HLA antibodies were detected in nonalloimmunized healthy males [26], and 96 of their target epitopes were defined [16]. 58 natural antibodies are only reactive with dissociated HLA class I antigens, heavy chain only (Table 4). 56 unique epitopes on dissociated HLA class I defined [16].
Epitope 5007 is shared by the HLA class I A-locus antigens A31 and A33 and defined by isoleucine (I) at the cryptic position 73. Antibody reactivity with the intact antigen is obstructed because position 73 is located under the peptide. It is slightly reactive with the intact HLA class I antigens. Reactivity increased by up to 10-fold with the dissociated antigens (heavy chain only)-when β2m and the peptide are dissociated from the heavy chain ( Figure 9). Epitope 5024 is shared by the HLA class I B-locus antigens B7, B42, B54, B55, B56, B67, B81, and B82 and defined by 66I + 70Q. Reactions strength of the antibody is stronger  Figure 9: Epitope 5007 shared by the HLA class I A-locus antigens A31 and A33 and defined by isolucine (I) at position 73. The epitope is accessible on the dissociated antigens and show stronger reactivity when the peptide has been dissociated from the heavy chain. Position 73 is not exposed in an intact HLA class I antigen. After acid buffer treatment and neutralization of the eluate, epitope 5007 becomes exposed and reacts with the antibody 10-fold. MICA * 027 213I/251R ND ND: not done; amino acids not exposed on the surface of the MICA antigen are shown between parentheses; a possible alternative epitope definitions are separated by "/"; epitopes.
with the unobstructed epitope after dissociation of the peptide from the heavy chain. Epitope 5037 is shared by the HLA C-locus antigens Cw4, Cw6, Cw17, and Cw18 and defined by 73A + 77N. Antibody reaction strength increases with the unobstructed epitope after removal of the peptide.
3.3. MICA Epitopes. MICA or MHC class I polypeptiderelated sequence A antigens have similar aa structure as the HLA class I ABC heavy chains. However, MICA antigens are not associated with a peptide and beta 2 microglobulin. Seven epitopes were defined for MICA antigens (Table 5).

Class II Epitopes
3.4.1. HLA-DRB1 Epitopes. Unlike class I epitopes, the 60 HLA class II B1 epitopes were defined based solely on amino acid sequence of the DR antigen beta chain where all epitopes can be defined by one single amino acid at one position ( Table 6). The number of epitopes for each DR antigen was from 8 to 21 epitopes.

3.4.2.
HLA-DQA1 and HLA-DQB1 Epitopes. Eighteen HLA class II DQB1 and DQA1 epitopes are defined using the adsorption and elution assays described in the Materials and Methods above. Fifteen of the epitopes are located on the beta chain of the DQ antigen and three on the alpha chain ( Table 7). The number of epitopes for DQB chains was 4-8 and only one for DQA chains (Table 8).
Sera from allosensitized patients can be expected to have anti-HLA antibodies to class I and II antigens. As illustrated in (Figure 12), this serum has antibodies directed against DR, DQ, and DP antigens. The serum was adsorbed with DQA1 * 02 01/DQB1 * 04 01 rHLA cells and the eluted antibody reacted with DQ4, DQ5, and DQ6 antigens which share epitope 2007 (Table 7).
One antigen mismatch can elicit an immune response to several epitopes on an HLA antigen. A serum from renal transplant patient with DQA1 * 02 01/DQB1 * 02 02 mismatch has two antibodies. One antibody targets epitope 2017 (defined by histidine (H) in position 52) on the DQA1 * 02 01 alpha chain and the other antibody targets   T  T  T  T  T  T  T  T  T  T  T  T  T  T (Table 7). As shown in the table, alternative epitope definitions are separated by "/." The efficacy of adsorption and elution assays is demonstrated where one serum with DQ specificity, including DQA1 * 02 01, underwent four separate adsorptions and elutions with rHLA DQ cells. Two of the cells have the relevant DQA1 * 02 01 chain, and the eluted antibodies show positive reactions with all heterodimers that contain the DQA1 * 02 01chain (red and green bars). However, eluents from adsorptions with irrelevant cells (no DQA1 * 02 01 chain) showed negative reactions (yellow and blue bars) ( Figure 13).
The following examples illustrate HLA-DQA1 and HLA-DQB1 epitopes. Illustration shows SA beads specificity, antigens sharing the epitopes and their position on the DQA1 and DQB1 chains.
HLA-DQA1 epitopes: epitope 2018 is shared by the alpha chains of the DQ4, DQ5, and DQ6 antigens and defined by glutamine (Q) at position 53.
HLA-DQB1 epitopes: epitope 2002 is shared exclusively by the beta chains of the DQ4 antigen and defined by leucine (L) in position 56.
Epitope 2022 is exclusive to DQB1 * 05 01 chain on the DQ5 antigen and defined by 125S + 126Q.
Epitope 2006 is shared by DQB1 * 03 01 (DQ7), DQB1 * 03 02 (DQ8), and DQB1 * 03 03 (DQ9) and  Epitope 2008 defined using mAb; b possible alternative epitopes are separated by "/"; epitopes that are defined by more than a single position/aa are separated by "+"; amino acids not exposed at the surface of the HLA molecule are between parentheses.
defined by proline (P) at position 55 on the beta chains of the DQ antigens.

Discussion
Cross-reactivity of antibodies with HLA antigens has been investigated for decades [4,5,[7][8][9]27]. Studies to identify HLA epitope, the target of antibodies, started more than 50 years ago [10], and numerous other studies followed since then [5,6,11,[28][29][30][31][32][33][34]. The amino acid structure of HLA antigen chains was reported for the HLA A2 in 1987 [12], and now, complete sequences of all known HLA antigen chains are readily available online [23]. Single antigens expressed on a mammalian cell line allowed us to simplify adsorption/elution assays and isolate one antibody from multispecific allosera, with multiple antibodies, and test the isolated antibody with the single antigen beads to more accurately determine antibody specificity. Isolated antibodies tested with large panels of HLA class I or class II single antigen beads were shown to be positive with certain antigens of the bead panels and negative with others. It is, therefore, reasonable to assume that the positive antigens share a public  Figure 12: Unabsorbed serum has antibodies with specificity to DR, DQ, and DP antigens (green bars). After adsorbing the serum with rHLA DQ cells, the eluted antibody shows specificity to DQ antigens only (red bars).
epitope which can easily be confirmed by looking at the amino acid sequences of these antigens. HLA single antigen bead assays are simplified assigning anti-HLA antibody specificities by simply listing all antigens that are positive with the serum or antibody. Because a positive antibody-antigen reaction indicates binding of antibody to the single antigen on the bead, we postulate that the antibody must be specific to the antigen. However, the single antigen beads assay often reveals more antibody specificities than other antibody detection assays. This is clearly seen when an immunological response to a mismatched antigen produces antibody specificity to nondonor antigens and in some cases unexpectedly to rare antigens. HLA antigens share public epitopes; therefore, the extra antibody specificity   Epitopes defined by more than a single position/aa are separated by "+"; amino acids not exposed at the surface of the HLA molecule are between parentheses.
of non-donor-specific and rare antigens can now be explained as antibody binding to public epitopes located on the positive antigens. Defining epitopes of HLA gives us better understanding of the breadth of non-donor-specific specificities found in sera. For example, specificity of antibodies to rare antigens like A80 and B76 were unexpectedly higher than the antigens' frequency (<0.5%) in the general population. The two antigens have 9 and 13 epitopes, respectively (Table 3). HLA epitopes were defined using computer software by searching, in published sequences of class I and class II antigens, for exclusive amino acids at the same position(s) that are shared by all positive-reacting antigens. Amino acid sequences and the 3D structures of available HLA antigens, used to ensure that aa's are exposed on the surface of the antigens, helped in defining close to 300 epitopes. Assay-positive antigens that share epitopes, defined by exclusively shared aa's, correspond to the antibody specificities. Although it is beyond the scope of assays used in our studies to determine the exact conformational arrangement of each epitope and all amino acids that constitute the epitope, the defining amino acids must be a focal part of the epitope. Public epitopes found exclusively on positive antigens and not on negative antigens are likely not coincidences. For several epitopes defined in our studies, the difference of one aa among alleles of the same antigen, at least one amino acid position can determine whether the allele is positive or negative with the antibody (Figure 5).
We have demonstrated that some antibodies target an epitope on one single antigen (private epitope) or an epitope on a group of two or more antigens (public epitopes). Furthermore, in anti-DQ antisera, immunological responses can produce antibodies to epitopes located on either or both polymorphic chains of the DQ antigens.
The usefulness of epitopes beyond determining correct antibody specificity in sera of transplant patients has been E E E   on HLA epitope-based matching for organ transplantation [35,36]. Wiebe reported on epitope matching to minimize de novo donor-specific antibodies to improve transplantation outcome [37] and Walton et al. reported on the usefulness of matching at the epitope level which protects against chronic lung allograft dysfunction [38].  Figure 15: HLA class II DP epitope 4003 shared by DPB chains DPB1 * 0201, DPB1 * 0402, DPB1 * 1001, and DPB1 * 1801 (red bars) and defined by 84D + 85E + 86A + 87V. Negative antigens that did not share epitope 4003 are shown in (gray bars).