Platelets from donors with consistently low HLA-B8, -B12, or -B35 expression do not undergo antibody-mediated internalization

Platelet (PLT) transfusions are essential for patients with chemotherapy-induced thrombocytopenia. Because of multiple causes, however, 10% to 27% patients can become refractory to PLT transfusion, which is defined as multiple subsequent ineffective transfusions.^1-5  This refractoriness is associated with increased risk for bleeding and reduced survival.⁴  Immunological refractoriness occurs in 4% to 8% of all patients receiving PLT transfusions. This is most frequently due to alloimmunization of the recipient against HLAs.^6,7  This alloimmunization results in the production of alloantibodies that target transfused PLTs and induce rapid clearance of the opsonized PLTs by uptake through splenic macrophages.^8-10 

There are currently 2 major strategies for management of patients with PLT refractoriness due to alloimmunization: (1) selecting HLA-identical donors from an HLA-typed donor file and (2) identifying acceptable HLA-mismatched donors by selecting PLTs negative for antigens that are targeted by anti-HLA antibodies from the recipient (ie, antibody specificity avoidance) or by selecting PLTs that are cross-match negative.^6,11,12  Although these methods result in acceptable posttransfusion increments in 70% to 90% of all cases,¹²  both have major limitations. Because the HLA system is highly polymorphic, a very large donor pool is needed, and for patients with multiple antibody specificities and/or rare HLA phenotypes, which occurs more frequently with increasing immigration, sufficiently compatible donors might not be available. An improved or alternative donor-selection strategy, as compared with fully HLA-matched PLTs products, would therefore be welcome to treat refractory patients.

A lower antigen expression may render PLTs less susceptible to antibody-mediated clearance by macrophages.^8,13-15  Therefore, strategies to minimize or reduce PLT HLA class I expression were explored as a means to avoid HLA matching to treat refractory patients. In this regard, using chloroquine phosphate isotonic fluid or citric acid HLA class I from the PLT surface was denaturated, causing a 70% to 80% loss of serologic activity.^16-19  Clinical application of these PLTs, however, resulted in varying outcomes, presumably due to manipulation-induced PLT function defects.²⁰  An alternative strategy for HLA class I elution may be to use donors that have naturally low expression of HLA class I on their PLTs, as has been described for certain HLA-B antigens. Indeed, a study that transfused HLA-B12–mismatched PLTs to refractory patients because of alloimmunization showed satisfactory increments in 69% of all transfusions.²¹  In this study, however, the density of HLA-B12 expression on the transfused PLTs was not determined, which might explain why some patients failed to show satisfactory increments. From all HLA-B antigens, the most frequent alleles in the Dutch donor population are HLA-B7 (27.3%), -B12 (25.3%), -B8 (22.7%), and -B35 (17.8%).²²  High variation of HLA-B8, -B12, and its split antigen B44 expression on PLTs has been reported, whereas this variation was not observed in HLA-B7.^23-26  Identifying donors with naturally low expression of frequent class I antigens on PLTs could significantly improve the availability of HLA-matched donors for PLT supportive care of immunized patients.

We investigated whether low HLA class I–expressing PLTs might overcome antibody-mediated clearance of transfused PLTs despite HLA mismatches and in this respect be an alternative strategy for fully HLA-matched transfusions. To do this, we first established the proportion of donors with consistently low expression of HLA-B8, -B12, and -B35 and then studied the effect of this low HLA expression on antibody-mediated internalization of PLTs by macrophages, which was shown to correlate with in vivo clearance.²⁷ 

To determine specific HLA class I expression and internalization of PLTs with high or low expression, citrated blood samples from HLA-typed donors were obtained after blood donation. Blood from these donors was collected multiple times, with a maximum of 10 samples per donor. HLA typing was performed in different HLA laboratories in The Netherlands as routine blood bank practice for PLT donors using low-, intermediate-, or high-resolution DNA or split-level serologic typing. Furthermore, to isolate monocytes or determine internalization of differently opsonized PLTs, blood samples from anonymous healthy individuals were obtained after they gave written informed consent. This study was approved by the Sanquin Ethical Advisory Board in accordance with the Declaration of Helsinki and according to Dutch regulations.

To specifically determine HLA-B8, -B12, and -B35 expression on PLTs, human monoclonal antibodies (mAbs) were selected from a locally produced mAb panel, produced in stirred-tank format and purified as previously described.^28,29  mAbs were labeled according to the manufacturer’s instructions using an Alexa Fluor 488 fluorescent labeling kit (Molecular Probes/Invitrogen, Breda, The Netherlands). Labeling and staining efficiency was standardized by flow cytometry analysis as previously described.²⁸  In short, after fluorescent labeling, a new lot of antibodies was compared with the previous lot using HLA-typed peripheral blood mononuclear cells. The dilution of the new lot was adjusted to achieve equal fluorescent signals for the new and the old batch. Staining was performed in saturating concentrations.

To determine specific HLA class I expression on PLTs, blood was diluted to 2.5 × 10⁶ PLTs/mL with PBS supplemented with 8 mM EDTA (Sigma, Zwijndrecht, The Netherlands). PLTs were stained 30 min at room temperature (RT) using 50 µg/mL anti-HLA-B8 (clone BVK1F9), anti-HLA-B12 (clone JOK3H5 combined with clone DK7C11) or anti-HLA-B35 (clone HDG8D9) Alexa Fluor 488–labeled antibodies and anti-CD61 labeled with PE-Cy7 (Beckman Coulter, Woerden, The Netherlands). Specific HLA class I expression on white blood cells (WBCs) from the same individuals was determined in whole blood supplemented with 8 mM EDTA and 50 µg/mL anti-HLA-B8, anti-HLA-B12, or anti-HLA-B35 Alexa Fluor 488–labeled mAbs and anti-CD45 labeled with PE-Cy5 (Beckman Coulter). After 30-min incubation at RT, red cells were lysed using IO Lysing Solution according to the manufacturer’s instructions (Beckman Coulter). Subsequently, samples were measured using flow cytometry (Cytomic FC500 MPL flow cytometry system with MXP software, Beckman Coulter) and analyzed with Kaluza Flow Cytometry Analysis Software (V1.3, Beckman Coulter). The gating strategy involved both selection of single PLTs or lymphocytes (based on forward and side scatter), and CD61-positivity for PLTs or CD45 positivity for WBCs (supplemental Figure 1, available on the Blood Web site). To determine background fluorescence levels, during each experiment, a negative control (ie, blood from a donor negative for the specified HLA type) was measured (n = 112, 90, and 76 for HLA-B8, -B12, and -B35). Specific HLA expression was presented relative to the averaged background median fluorescence as mean ± standard deviation (SD). Coefficients of variation were calculated by dividing the SD with the mean, multiplied by 100%.

Citrated blood from healthy individuals or donors with known low (ie, <2 SDs above background) or high (ie, above median) specific HLA class I expression was collected, and PLT-rich plasma was obtained by centrifugation at 200 radial centrifugal force for 10 min. PLT-rich plasma was washed and labeled for 20 min in PBS at RT with 3.75 µM PKH26 membrane dye (Sigma) with continuous gentle shaking. Subsequently, PLTs from healthy individuals were opsonized with anti-HPA1 amAb (humanized IgG1, clone B2G1; kindly provided by G. Vidarsson, Sanquin Research, Amsterdam, The Netherlands) or anti-pan HLA class I mAb (mouse IgG2a, clone W6-32; ITK Diagnostics, Uithoorn, The Netherlands), and PLTs from donors with the specific HLA antigens were opsonized with 50 µg/mL anti-HLA-B8, anti-HLA-B12, or anti-HLA-B35 human IgG1 mAbs. After 30-min opsonization at RT, PLTs were washed twice using sequestrine buffer (17.5 mM Na₂HPO₄, 8.9 mM Na₂EDTA, 154 mM NaCl, pH 6.9, containing 0.1% [wt/vol] bovine serum albumin [BSA]; all obtained from Merck Millipore, Amsterdam, The Netherlands).

Monocytes were isolated from fresh apheresis material using the Elutra Cell Separation System (Gambro, Lakewood, CO) and subsequently frozen until further use. The >90% purity of these monocytes was confirmed using flow cytometry. On day 0, monocytes were thawed and differentiated in a 6-well plate (Nunc/Sanbio BV, Uden, The Netherlands) to macrophages with 10 ng/mL granulocyte-macrophage colony-stimulating factor (Cellgenix, Freiburg, Germany) in Iscove modified Dulbecco medium (Lonza, Breda, The Netherlands) supplemented with 10% fetal calf serum (Bodinco, Alkmaar, The Netherlands), 100 U/mL penicillin, and 100 U/mL streptomycin (both from Gibco/Thermo Fischer Scientific, The Netherlands) at 37°C and 5% CO₂. On days 7 to 9, macrophages were incubated for 60 min at 37°C with opsonized PLTs. Subsequently, macrophages were washed with PBS + 0.5% BSA, harvested using 130 mM lidocaine (Sigma) with 10 mM EDTA (Merck Millipore), washed again in PBS, fixed with 3.7% paraformaldehyde (Sigma) in PBS, washed 2 more times in PBS + 0.5% BSA, and incubated with FITC-labeled anti-CD61 (Beckman Coulter) and APC or brilliant violet 650-labeled anti-HLA DR (both from BD biosciences, Breda, The Netherlands). After 30 min incubation at RT, cells were washed and analyzed using imaging flow cytometry (ImageStreamX Mark II Imaging Flow Cytometer, Merck Millipore). Samples were analyzed using IDEAS Application (IDEAS software V6.1.303.0, Merck Millipore). The gating strategy is presented in supplemental Figure 2 and involved selection of single macrophages (aspect ratio intensity vs area; for bright field and APC/BV650 channel) in focus (gradient root mean square; for bright field and APC/BV650 channel) and exclusion of false-positive cells (defined as intracellular fluorescence of the PLT-specific anti-CD61 membrane staining). PLT internalization by macrophages was determined by quantifying the amount of PKH fluorescence within macrophages (using fluorescence intensity within an intracellular mask based on HLA DR membrane staining). To correct for varying “baseline” internalization of PLTs (presumably caused by differences in quality and age [1 to 3 days] due to transportation), internalization of HLA-typed opsonized PLTs was presented relative to internalization of unopsonized PLTs from the same donor.

Statistical analysis and graphical presentation was performed using GraphPad Prism version 7.02 (GraphPad Software, La Jolla, CA). The percentage of false-negative test results (ie, <2 SDs above background) was determined using a sensitivity analysis in donors with high HLA class I expression on average, defined as >2 SDs above background, with at least 3 measurements. Correlations between HLA class I expression in WBCs and PLTs as well as PLT opsonization and internalization were calculated using a Pearson’s correlation test. To investigate the effect of zygosity on interindividual variation, expression data were divided based on zygosity and then normalized to the group average (ie, all donors, homozygous or heterozygous), and subsequently, an F-test was used to calculate statistical differences in variation between these groups. Statistical differences between interindividual and intraindividual variation in HLA class I expression were calculated using an F-test. Statistical differences in HLA expression were calculated using an unpaired Student t test. Statistical differences in internalization due to opsonization were calculated using a Wilcoxon matched-paired signed rank test, and differences in internalization between positive controls from PLTs with high or low expression were determined by Mann-Whitney U test. P < .05 was considered statistically significant.

PLTs and WBCs from HLA-B8– (n = 113), -B12– (n = 86), and -B35–positive donors (n = 66) were analyzed for the level of these specific HLA-B antigens. From donors positive for the broad HLA-B12 antigen, most donors were positive for the split antigen B44 (n = 85), and 6 donors were B45 positive. Because only 1 of these 6 B45-positive donors was B44 negative, potential differences between the HLA-B12 split antigens were not studied. All expression data are presented relative to the background values of donors negative for the specific antigen as relative median fluorescence intensities. PLTs from positive individuals showed a relative HLA-B8, -B12, and -B35 expression of 1.9 ± 0.9, 3.2 ± 2.6, and 3.1 ± 1.3 (mean ± SD), respectively (Figure 1), with a maximum variation between individuals of approximately fivefold, 12-fold, and eightfold, respectively. Even more interesting is that from all HLA-B8, -B12, and -B35 donors, 31.9%, 33.7%, and 9.1%, respectively, showed no or only minimal expression (defined as <2 SDs above background levels) of these HLA class I antigens on PLTs.

Next, we investigated the effect of zygosity on specific HLA class I expression. For this, donors were grouped as heterozygous or homozygous for a given HLA-B antigen, and the effect of this grouping on interindividual variation was investigated using an F-test. PLTs from homozygous HLA-B8 donors had significantly higher expression (2.7 ± 0.9, n = 48) than heterozygous B8 donors (1.3 ± 0.3, n = 65) (Figure 2A). This increased expression in homozygous individuals was also observed for HLA-B12 (5.0 ± 3.2 [n = 30] vs 2.8 ± 2.1 [n = 56]) (Figure 2B) and -B35 (4.2 ± 1.5 [n = 13] vs 2.8 ± 1.0 [n = 53]) donors (Figure 2C). In the group of homozygous HLA-B8, -B12, and -B35 donors, 12.5% (6 of 48 donors), 20.0% (6 of 30 donors), and 0.0% (0 of 13 donors), respectively, showed expression that was not or only minimally detectable. Furthermore, grouping HLA-B8 donors based on their zygosity significantly reduced interindividual variation in expression, wherein it decreased from homo- to heterozygous donors. However, in HLA-B12– and -B35–positive donors, although the density of expression was affected by zygosity, the interdonor variation was not. To investigate whether low antigen expression is a general donor characteristic, individuals who expressed two of the investigated antigens (ie, HLA-B8/B12, -B8/B35, or -B12/B35) were selected and the association of expression of these 2 antigens was investigated. Expression of these HLA-B antigens was not correlated, but the number of donors expressing 2 of the investigated antigens was limited (n = 6-8). However, donors with low expression of both B antigens were identified (supplemental Figure 3). To investigate whether the observed variations are exclusive to PLTs, specific HLA class I expression was also studied on WBCs from the same individual. HLA-B8, -B12, and -B35 expression on WBCs similarly varied between individuals and again only for HLA-B8 donors due to zygosity (Figure 3). The expression on WBCs clearly correlated with the observed variability of these antigens on PLTs, although expression on WBC always was strongly above the background level, showing that the observed variations are not PLT specific.

To investigate the consistency of expression on PLTs within an individual (ie, intraindividual variation), donors with at least 4 independent blood samplings at different time points were selected (Figure 4). The average follow-up time was 349, 413, and 371 days for HLA-B8, -B12, and -B35 donors, respectively. The variation in HLA-B8 (n = 22), -B12 (n = 22), and -B35 (n = 17) expression within each individual donor was significantly lower than the variation between individuals (Figure 4A-C), which was most pronounced in low-expressing donors. Because of these low intraindividual variations, individuals with consistently low expression can be used as possible HLA-mismatched PLT donors. However, some donors with a high expression on average may show a lower expression in another sample, which may affect the firm definition of a donor with consistently low expression if only 1 sample is tested. A sensitivity analysis (using a threshold of 2 SDs above background median fluorescence intensity defining low expression) showed that the possibility of a false-negative test result was 20.3%, 8.4%, and 7.6% for HLA-B8–, -B12–, and -B35–positive donors. As a result, 2 consecutive negative test results indicate with >95% certainty the identification of a donor with low HLA expression.

Next, the effect of reduced antigen expression on antibody-mediated internalization of PLTs by macrophages was studied. First, PLTs were incubated with increasing concentrations of mAbs (0.1, 1, and 10 µg/mL) that target either HPA1a or pan-HLA class I, resulting in increased opsonization (Figure 5A-B). The opsonization of PLTs led to antibody-mediated internalization by macrophages (Figure 5C-D), which strongly correlated with the level of antibody opsonization (r² = 0.89 for anti-HPA1a and r² = 0.94 for anti-pan HLA class I). To subsequently determine the relevance of the observed variations in HLA class I expression, PLTs from donors with high (ie, expression above median) or low (ie, <2 SD above background) HLA-B8, -B12, or -B35 expression were obtained and opsonized with the relevant human mAbs in saturating concentrations. Compared with unopsonized controls, internalization of PLTs obtained from donors with high specific HLA class I expression significantly increased due to opsonization, which was not observed for PLTs obtained from donors with low specific HLA class I expression (Figure 6A). PLT opsonization with an irrelevant antibody (ie, anti-HPA1a) did not result in differences in internalization of PLTs obtained from high- or low-expressing donors (Figure 6B). Antibody-mediated internalization of PLTs correlated with PLT antigen expression (Figure 6C).

HLA-matched PLT transfusions require the maintenance of a large HLA-typed donor pool. The efficiency of this HLA matching could be significantly improved if certain mismatches are allowed due to low or undetectable antigen expression. In this study, we explored the proportion of donors with naturally low expression of HLA-B8, -B12, or -B35 on PLTs. Second, the consistency of this expression was determined. Finally, the effect of this low expression on antibody-mediated internalization of PLTs by macrophages was investigated. Interindividual variation in HLA-B8, -B12, and -B35 expression on PLTs was high, with a maximum variation between individuals of approximately fivefold, 12-fold, and eightfold, respectively. The observed variations in the present study for HLA-B8 and -B12 were less than the previously reported eightfold and 35-fold variations.^23,24  This difference is probably because we used flow cytometry assay in our study, whereas a 2-stage ⁵¹Cr release assay was used previously.

Although the variation in specific HLA class I expression was most prominent on PLTs, expression also varied on WBCs. As PLT and WBC antigen expression appeared correlated, the observed interdonor variations are not PLT but donor specific and suggest involvement of (epi-)genetic factors. For HLA-B8, the inter-individual variations are indeed in part explained by zygosity. In addition, low expression of one HLA-B antigen was not associated with low expression of another B-antigen, possibly indicating that the density of antigen expression is dependent on the allele. However, although posttranslational modifications are known to affect surface expression of HLA-A, HLA-B, and HLA-C and thereby cause variations between HLA alleles,^30,31  this does not explain variations within one HLA allele. Furthermore, inflammation causes significant upregulation of mainly HLA-B on WBCs,³²  which is less likely to play a role in our studies, which only involved healthy donors. Clearly, more studies are warranted to further explore the underlying causes of these interindividual differences in HLA class I expression.

In contrast to the high interindividual variation, specific HLA class I expression appeared to be fairly constant within each individual, particularly in the low-expressing individuals. This allows identification of donors with consistently low HLA-B8, -B12 or -B35 expression. Because of this consistency, instead of testing each PLT concentrate for its level of HLA expression, donors can be marked in the HLA-typed donor file as “low expressor” for the specific HLA allele. This, in combination with a straightforward flow cytometry–based assay, allows simple clinical implementation. However, incidental low expression in donors with on average high expression requires retesting of each donor displaying low expression on the first measurement. Whatever the underlying reason for this, 2 consecutive measurements showing a low expression identifies a low-expressing donor with >95% certainty.

PLT concentrates with a naturally low level of specific HLA class I expression obviously have major advantages over the chemical treatment of PLTs to lower HLA expression. Although acid treatment results in a 70% to 80% loss of serological activity, the clinical effectiveness is disappointing. This has been attributed to manipulation-induced PLT dysfunction, thereby limiting clinical application.^16,18,20  The use of unmanipulated naturally low HLA class I expression on PLTs, which is present in 31.9%, 33.3%, and 9.1% of HLA-B8–, -B12–, and -B35–positive donors when using a cutoff <2 SDs above background, would not have this disadvantage. Homozygous donors with such low expression on PLTs, as observed for HLA-B8 and -B12, are of particular interest, because these only need to be matched for HLA-A, thereby greatly increasing the possibility of matching a patient with a donor.

For transfusion purposes, antigen expression on PLTs must be reduced to such an extent that anti-HLA class I antibody–mediated clearance of PLTs by macrophages is limited or completely abolished.^9,10  Because Lim and colleagues have previously shown that antibody-mediated internalization in vitro correlates with in vivo clearance,²⁷  our data showing an absence of antibody-mediated internalization of low-HLA-expressing PLTs by macrophages are of particular interest. Because this antibody-mediated internalization showed a linear correlation with the degree of antibody opsonization and PLT antigen expression, it is likely that PLT antigen expression will also correlate with clearance in vivo. However, the degree of antibody opsonization depends not only on antigen expression but also on antibody titer, avidity, and/or isotype. Therefore, the specific threshold below which HLA class I expression will not result in antibody-mediated clearance will be patient specific and can not be predicted at this point.

A limitation of this study is the lack of clinical data supporting the findings observed in our phagocytosis experiment, even though this experimental setup was proven to have clinical correlation.²⁷  If PLT transfusions with low expressing mismatched HLA-B alleles actually result in satisfactory increments in broadly immunized and refractory patients could be evaluated in a parallel-group semirandomized trial,³³  in which patients with immunological refractoriness should be randomized to receive either PLTs with high or low HLA-B8, -B12, or -B35 expression.

In summary, we show high interindividual variation for HLA-B8, -B12, and -B35 expression on PLTs, with 9% to 34% of donors showing reproducibly undetectable or low expression. PLTs with low HLA expression showed deficient antibody-mediated internalization by macrophages. Although such low HLA expression on PLTs is mostly found in heterozygous donors, even homozygous HLA-B8 and -B12 individuals may show expression at background levels. In theory, for such donors, only HLA-A needs to be taken into account as matching criteria, which would greatly facilitate donor selection for alloimmunized patients. Establishing a pool of donors with such consistently low expression of HLA-B antigens might even allow transfusions of highly immunized patients during acute situations. Future studies should explore clinical effectiveness of such a “low HLA expressed mismatch” strategy and determine which expression levels result in satisfactory increments in refractory patients due to alloimmunization.

The authors thank Susan Cuvalay (Sanquin Blood Bank) for her help with the transportation of study material, all individuals for giving informed consent, and G. Vidarsson for kindly providing the anti-HPA1a mAb.

This study was funded by Product and Process Development Cellular products (Sanquin PPOC) 09-026.

Contribution: A.S. designed research, performed experiments, analyzed and interpreted data, and wrote the manuscript; B.T. and A.B. designed research and critically reviewed manuscript; A.M. and F.H.C. developed mAbs and reviewed the manuscript; J.L. and J.S. performed experiments; S.M.v.H. interpreted data and reviewed the manuscript; and A.t.B., J.J.Z., and P.F.v.d.M designed research, interpreted data, and critically reviewed the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Pieter F. van der Meer, Plesmanlaan 125, 1066CX Amsterdam, The Netherlands; e-mail: p.vandermeer@sanquin.nl.