In utero exposure to alloantigens primes alloimmunization to platelet transfusion in mice

Abstract

Background:

Platelet transfusions remain a mainstay of treatment for many patients with thrombocytopenia, but can lead to alloantibodies to Human Leukocyte Antigens (anti-HLA) resulting in inadequate responses to subsequent platelet transfusions (refractoriness), as well as complicate transplantation. Despite substantial decreases in alloimmunization with the implementation of leukoreduction, a significant percentage of patients still become alloimmunized following platelet transfusions. It remains unclear why some patients make anti-HLA antibodies, but others do not make anti-HLA antibodies even with chronic transfusion. Antecedent pregnancy correlates with risk of alloimmunization due to platelet transfusion in humans - however, isolation of pregnancy as a single variable is not possible in human populations.

Study Design and Methods:

A tractable murine model of pregnancy and transfusion was engineered by breeding C57BL/6 (H-2^b) dames with BALB/c (H-2^d) sires. After pregnancy, female mice were transfused with leukoreduced platelets from F1 (H-2^b/d) donors that expressed the same paternal major histocompatibility complex (MHC) H-2^d alloantigens as the sires. Control groups allowed isolation of pregnancy or transfusion alone as independent variables. Alloimmunization was determined by testing serum for antibodies to H-2^d MHC alloantigens.

Results:

No alloantibodies were detected after pregnancy alone, or in response to transfusion of platelets alone; however, significant levels of alloantibodies were detected when pregnancy was followed by transfusion.

Conclusions:

These findings isolate antecedent pregnancy as a causal contribution to increased frequencies of alloimmunization by subsequent platelet transfusion in mice and provide a platform for ongoing mechanistic investigation.

1 |. INTRODUCTION

Transfusion of platelets (PLTs) remains an essential and life-saving therapy for patients with thrombocytopenia.^1,2 Despite advances in engineering platelet units with decreased immunogenicity (eg, leukoreduction), a significant number of patients still become alloimmunized to HLA after PLT transfusion,³ which can lead to inadequate responses to subsequent PLT transfusions (refractoriness) as well as complicating organ transplantation. Identifying HLA matched or cross-matched PLTs for refractory patients can be both technically and logistically challenging, resulting in delay or even absence of treatment, causing morbidity and/or mortality, as well as consuming substantial resources.⁴

It remains unclear why some patients become alloimmunized and others do not. Analysis of human data has indicated an association between prior pregnancy and risk of alloimmunization by PLT transfusion.^3,5 Due to the ethical and logistical limitations of human studies, the link between pregnancy and alloimmunization to PLT transfusion remains a correlation, and causality has not been directly assessed. Herein, we report a murine model designed to isolate pregnancy as an independent variable and to test if pregnancy alone is sufficient to increase alloimmunization to subsequent PLT transfusions.

2 |. MATERIALS AND METHODS

2.1 |. Mice

C57BL/6 (B6), UbiC-GFP (C57BL/6-Tg[UBC-GFP]30Scha/J), and BALB/c mice were purchased from Jackson Labs, Bar Harbor, ME. B6xBALB/c F1 and UbiC-GFPxBALB/c mice were bred in the BloodworksNW vivarium. All procedures were carried out in vivaria of either BloodworksNW or University of Virginia, and in accordance with approved IACUC protocols at BloodworksNW and/or University of Virginia.

2.2 |. Platelet transfusions and survival studies

Transfusion of Leukoreduced-PLTs (LR-PLTs) and post-transfusion PLT survival were performed as previously described.^6,7 Refractoriness was defined as 2 standard deviations (SD) < the mean 48 hour PLT recovery of naive controls. For survival studies, PLT donors were UbiC-GFPxBALB/c mice, which express both green florescent protein (GFP) and one copy of BALB/c alloantigens on PLTs.

2.3 |. Detection of alloantibodies

Serum was tested for alloantibodies by indirect immunofluorescence using splenocytes from BALB/c and B6 mice as targets, as previously described.⁶ Adjusted Mean Florescence Intensity (MFI) was determined by subtracting the MFI of negative control B6 targets from the signal obtained using BALB/c targets.

2.4 |. Statistical analysis

Results were analyzed using GraphPad PRISM 8.3. Differences between MFIs were evaluated using a nested One-Way ANOVA with Tukey post-hoc analysis. Differences in platelet refractoriness were evaluated by Fisher’s exact test.

3 |. RESULTS

The H-2 locus in mice is orthologous to HLA in humans, and contains three loci (K, D, and L) that express major histocompatibility complex (MHC) I genes (analogous to HLA-A, B, and C). BALB/c mice encode the H-2^d haplotype that expresses K^d, D^d, and L^d, whereas B6 mice encode the H-2^b haplotype that expresses K^b and D^b (B6 mice do not express a functional L gene). Exposure of mice to cells or tissues from a mouse of a different H-2 haplotype is a common model of HLA mismatch in humans.

The experimental design consisted of four separate groups (Figure 1A). B6 dames were bred with BALB/c sires and were then transfused with F1 H-2^d/b LR-PLTs (allogeneic LR-PLT) (group 1) to establish the combined effects of in utero exposure followed by transfusion of PLTs expressing the same alloantigens. Group 2 (bred with BALB/c sire followed by transfusion with B6 PLTs) and Group 3 (bred with B6 sire followed by allogeneic LR-PLTs) established isolated effects of exposure to alloantigen by either pregnancy or PLT transfusion alone, respectively. Group 4 (bred with B6 sire followed by transfusion with B6 LR-PLTs) established background from both pregnancy and transfusion but no exposure to alloantigen. All groups received four weekly transfusions of LR-PLTs. One week after the last transfusion, serum was isolated from all groups and tested for BALB/c specific alloantibodies. Subsequently, PLT survival studies were performed in each group by transfusing allogeneic GFP+ PLTs and assessing circulation 48 hours post-transfusion.

FIGURE 1.

A, Treatment schema. B6 (H-2^b) female mice were mated with either B6 (H-2^b) males or BALB/c (H-2^d) males. After weaning pups, mice were either transfused with 4 weeks of allogeneic (H-2^b/d) LR-PLT or 4 weeks of syngeneic B6 (H-2^b) LR-PLT. During week 5, all mice underwent phlebotomy for serum testing, followed by a transfusion of allogeneic H-2^d GFP PLT to test platelet survival. Since mating is not synchronized across mice, as a logistical necessity, post-partum females were accumulated overtime and then all transfused together - however, within a given experiment all dames had delivered within 3 weeks of each other and similar time ranges were present in all groups (between 3 and 6 weeks). B, The rate and magnitude of alloimmunization was significantly higher (P < .01) in the group that had in utero exposure to alloantigens (H-2^d) followed by four weekly allogeneic LR-PLT transfusions compared to the control group, which had in utero exposure to syngeneic antigens (H-2^b), followed by four weekly allogeneic LR-PLT transfusions. Tukey’s post-hoc analysis was used to compare each group with all other groups and group 1 was significantly different from groups 2, 3, and 4. No statistical difference between groups 2, 3, or 4 with each other. C, Separate mice that were dedicated to assessing antibody levels immediately after pregnancy were bled prior to breeding and immediately after delivery. No alloantibodies were detected from pregnancy alone. D, No significant difference was seen in PLT survival in any of the groups; although a trend was observed in which decreased PLT survival was observed only in the group exposed to alloantigen both in utero and then with subsequent transfusion (BALB/c sires and F1 PLTs) - this trend did not achieve statistical significance (P = .2241)

Antecedent pregnancy with a BALB/c sire and transfusion with allogenic LR-PLTs resulted in a significant (P < .01) increase in alloimmunization compared to all other groups (Figure 1B). While there was substantial variability in alloimmune responses within each group, as is common with mice despite using genetically identical animals,^6,8,9 no significant difference was observed when groups 2, 3, or 4 were compared with each other. Inflammation can affect alloimmunization to transfusion and phlebotomy induces inflammation; thus, the experimental design does not include phlebotomy until after four transfusions. This results in the first test for antibodies being 10–13 weeks after pregnancy. To test if pregnancy alone was sufficient to induce alloantibodies that faded over 10 weeks, additional studies were performed in which B6 dames were bred with BALB/c sires. No alloantibodies were detected in the post-pregnancy sera compared to the pre-pregnancy sera (Figure 1C). Positive control sera from previously alloimmunized mice gave a strong signal. Together, these data indicate that the combination of antecedent pregnancy with subsequent LR-PLT transfusion induces humoral alloimmunization. Alloantibodies are not induced by either pregnancy or transfusion alone in this model.

Combined exposure to alloantigen in utero and through subsequent transfusion (group 1) showed a decrease in PLT survival, but it did not achieve statistical significance (Figure 1C). However, categorical analysis of refractoriness (as defined as PLT survivals 2 SD below the mean of control mice [group 4]) revealed that 3 of 15 mice (20%) were refractory in group 1 whereas no mice were refractory in any other group; however, this result was not significant (P = .2241 by Fisher’s exact test).

4 |. DISCUSSION

In the current report, a reductionist animal model isolates pregnancy as a single variable and determines if it can cause increases susceptibility to alloimmunization by subsequent PLT transfusion. As with any mouse model it is necessary to consider how it aligns with known human biology. Anti-HLA can be detected in many women after pregnancy and increases in frequency with multiple pregnancies.^10,11 However, in the TRAP trial, there were women with a history of pregnancy who had no detectable anti-HLA antibodies upon enrollment in TRAP, but still had increased rates of alloimmunization and refractoriness after PLT transfusion compared to men and women who had never been pregnant.³ Several studies on anti-HLA and pregnancy have reported rapid evanescence and/or disappearance of alloantibodies after delivery.¹² Since alloantibodies can be transient in nature and are not routinely checked in the perinatal period, it is unknown if the TRAP participants in question had detectable alloantibodies after pregnancy that then evanesced, or if they never had alloantibodies at all. In the current study, mice only experienced a single pregnancy; multiple pregnancies may result in detectable alloantibodies without transfusion. However, the current data demonstrate that a single pregnancy can pre-dispose to subsequent alloimmunization by PLT transfusion without inducing detectable alloantibodies on its own.

In the current model, the use of F1 donors as the source of allogeneic LR-PLTs is required to see the described effect. The reason for this is that fully mismatched BABL/c PLTs cause strong alloantibody responses even in B6 dames that bred with B6 sires and have never previously encountered alloantigen (data not shown). This is consistent with previous reports using BALB/c PLT donors and B6 recipients.^6,8,9 The use of fully mismatched PLTs to study alloimmunization is also common in the field in models that study alloimmunization to PLTs transfusion.^13–15 One interpretation of this is simply a matter of antigen dose; with two copies of alloantigen, BALB/c LR-PLTs are a stronger immune stimulus than the one copy found on F1 PLTs. However, F1 LR-PLTs (but not BALB/C PLTs) also share an MHC locus with the recipient. These variables cannot be separated by the current studies, and it is also unknown how such factors may play a role in human alloimmunization as (in general), fine determination of degree of MHC mismatch and/or sharing is not known for studies of human PLT alloimmunization.

The mice that were immunized as a result of exposure to alloantigen in utero and through subsequent transfusion showed decreased post-transfusion PLT recoveries, although the decrease was not statistically significant. This is not because alloantibodides cannot clear PLTs in this system, as it has been previously reported that robust clearance of H-2^b/d F1 PLTs occurs in the same strain combination used in this report.^6,7 However, in the referenced papers, the immunogen was fully allogeneic cells. It is unclear why PLT clearance was not caused by alloantibodies in the current study, but could be due to either qualitative or quantitative effect. Indeed, the presence of alloantibodies increases the likelihood of refractoriness in humans; however, it has been well described that some people with even high titer alloantibodies are not refractory to incompatible PLTs, suggesting qualitative effects in addition to quantitative.¹⁶

In aggregate, this murine model isolates in utero exposure to alloantigen as a single variable, providing strong evidence of causality. As with all models systems, murine biology may not reflect human biology; however, in this case, the same correlation between pregnancy and transfusion induced alloantibodies and PLT refractoriness has been noted in humans.³ This murine model allows the elimination of other confounders not possible in human studies, and also allows future mechanistic studies into the underlying immunobiology.