Non-HLA Antibodies and Epitope Mismatches in Kidney Transplant Recipients With Histological Antibody-Mediated Rejection

Background Correlation between antibody-mediated rejection (ABMR) and circulating HLA donor-specific antibodies (HLA-DSA) is strong but imperfect in kidney transplant (KT) recipients, raising the possibility of undetected HLA-DSA or non-HLA antibodies contributing to ABMR. Detailed evaluation of the degree of HLA matching together with the identification of non-HLA antibodies in KT may help to decipher the antibody involved. Methods We retrospectively assessed patients with transplant biopsies scored following Banff’15 classification. Pre- and post-transplant serum samples were checked for HLA and non-HLA antibodies [MICA-Ab, angiotensin-II type-1-receptor (AT1R)-Ab, endothelin-1 type-A-receptor (ETAR)-Ab and crossmatches with primary aortic endothelial cells (EC-XM)]. We also analyzed HLA epitope mismatches (HLA-EM) between donors and recipients to explore their role in ABMR histology (ABMRh) with and without HLA-DSA. Results One-hundred eighteen patients with normal histology (n = 19), ABMRh (n = 52) or IFTA (n = 47) were studied. ABMRh patients were HLA-DSApos (n = 38, 73%) or HLA-DSAneg (n = 14, 27%). Pre-transplant HLA-DSA and AT1R-Ab were more frequent in ABMRh compared with IFTA and normal histology cases (p = 0.006 and 0.003), without differences in other non-HLA antibodies. Only three ABMRhDSAneg cases showed non-HLA antibodies. ABMRhDSAneg and ABMRhDSApos cases showed similar biopsy changes and graft-survival. Both total class II and DRB1 HLA-EM were associated with ABMRhDSApos but not with ABMRhDSAneg. Multivariate analysis showed that pre-transplant HLA-DSA (OR: 3.69 [1.31–10.37], p = 0.013) and AT1R-Ab (OR: 5.47 [1.78–16.76], p = 0.003) were independent predictors of ABMRhDSApos. Conclusions In conclusion, pre-transplant AT1R-Ab is frequently found in ABMRhDSApos patients. However, AT1R-Ab, MICA-Ab, ETAR-Ab or EC-XM+ are rarely found among ABMRhDSAneg patients. Pre-transplant AT1R-Ab may act synergistically with preformed or de novo HLA-DSA to produce ABMRhDSApos but not ABMRhDSAneg. HLA epitope mismatch associates with ABMRhDSApos compared with ABMRhDSAneg, suggesting factors other than HLA are responsible for the damage.


INTRODUCTION
Correlation between the detection of HLA donor-specific antibodies (HLA-DSA) and antibody-mediated rejection (ABMR) is strong but imperfect in kidney transplant (KT) recipients (1)(2)(3)(4)(5)(6)(7). Not all patients with pre-or post-transplant HLA-DSA have ABMR damage in their biopsies (8). Different groups have tried to identify characteristics of HLA-DSA that may predict ABMR (9)(10)(11)(12). There is also an active search for other invasive or non-invasive biomarkers for ABMR diagnosis (13)(14)(15). In the other hand, some patients have biopsies with histological findings suggestive of ABMR (ABMR h ) without circulating HLA-DSA (16), generating the concept of the existence of ABMR h DSA pos and ABMR h DSA neg cases. There is still limited literature describing the incidence of this type of ABMR without HLA-DSA, evaluating if these cases collectively show different clinical or histological characteristics or if non-HLA antibodies may explain the damage. Besides, controversial results in outcomes comparing ABMR h DSA pos and ABMR h DSA neg cases have been reported (17,18).
Based on the hypothesis that other antibodies may play a lead role in the case of ABMR histological damage with or without HLA-DSA, some groups have evaluated non-HLA antibodies in KT recipients (19,20). Although first reports connecting non-HLA antibodies and graft outcomes were published in 2005 (19,21), evidence is still weak and debated. Antibodies against specific alloantigens such as MICA (MICA-Ab) or MICB, or against autoantigens like angiotensin II type 1 receptor (AT 1 R-Ab), endothelin-1 type A receptor (ETAR-Ab), perlecan, agrin or vimentin, among others, have been reviewed recently (22). Some groups focused into the analysis of pathogenic antibodies directed against endothelial cellswhich express some of those but also other antigens-with endothelial cell crossmatches (23)(24)(25). The increased evidence that the prevalence of non-HLA antibodies in KT recipients is high (26), together with the heterogeneous post-KT clinical course of patients included in these studies (25) hamper the correct identification of deleterious non-HLA antibodies. On the other hand, HLA epitope mismatch (HLA-EM) assessment has gained interest as an added immune monitoring tool to provide a more precise evaluation of HLA matching (27)(28)(29). HLA-EM has been previously associated with the development of de novo HLA-DSA (30) and ABMR (31). The clinical relevance of HLA-EM analysis remains under discussion and its application is not generalized yet.
Here, we systematically explored pre-and post-KT serum samples for HLA and different types of non-HLA antibodies: MICA-Ab, AT 1 R-Ab and ETAR-Ab, and other non-HLA antibodies performing crossmatches with primary aortic endothelial cells (EC-XM). Additionally, we evaluated pre-KT HLA-EM load. We focused on KT patients with biopsies with Banff category 2 diagnosis and compared them with two other Banff diagnosis: category 1 or no abnormalities (normal), as a usual control group, and category 5 or interstitial fibrosis and tubular atrophy (IFTA), damage with not clear pathogenicity to evaluate the potential role of non-HLA antibodies in this case (32).

Study Population and Design
Prospective observational study performed in KT patients active at our transplant program in Hospital del Mar. A total of 234 consecutive clinical and surveillance renal biopsies were performed in ABO compatible KT after a negative CDC crossmatch (February 2011-June 2015). Ninety-two biopsies fulfilling Banff 2015 categories 3, 4 and 6 were excluded and 142 biopsies achieving categories 1, 2 or 5 were selected. From these 142 biopsies, we selected only one biopsy per patient according to these criteria: the first biopsy obtained after 3 months post-transplantation, unless a biopsy with category 2 diagnosis was available. Five biopsies were excluded due to unsuitable serum samples. Finally, 118 biopsies corresponding to 118 patients were included in the study (Supplementary Figure 1). Demographical and clinical data were collected as previously described (33), and follow-up was done until graftloss, death, 96 months post biopsy or July/2020. The study was approved by the Parc de Salut Mar Ethical Research Board (2010/ 3904/I) and all patients signed informed consents. All clinical and research activities reported are consistent with the Declarations of Istanbul and Helsinki.

Histological Scoring and Classification of the Biopsies
Biopsies were performed for indication or follow-up (including HLA-DSA detection without graft dysfunction). Processing was undertaken as previously described (33). All biopsies were scored by a pathologist following Banff 2015 classification and assigned to any of the six Banff categories (33). Category 2 included biopsies that met the first two Banff 2015-2019 criteria for ABMR histology, fulfilling the suspicious or full diagnosis of ABMR in Banff 2015 classification.

Sera Collection and Detection of HLA and Non-HLA Antibodies
One-hundred one available pre-KT and 118 post-KT serum samples collected contemporaneously to biopsies were retrospectively analyzed. HLA antibody testing (HLA-A, B, C, DRB and DQB) was performed as previously described (34) using Luminex HLA Single Antigen Bead assays (LABScreen, One Lambda, Canoga Park, CA). Antibodies against MICA antigens (*001, *002, *004, *007, *009, *012, *017, *018, *019, *027) were determined using LABScreen assay by Luminex Technology, according to the manufacturer's specifications (One Lambda, CA). MICA-Ab were considered positive if mean fluorescence intensity >1,000. MICA typing for donors and recipients was not available. AT 1 R-Ab and ETAR-Ab were measured using enzyme-linked immunosorbent-based assays (35) (One Lambda, CA), diluted 1:100, tested in duplicate and read on an Epoch Microplate Spectrophotometer (Bio-Tek, Winooski, VT). Samples with AT 1 R-Ab or ETAR ≥10 U/ml were considered positive based on previous studies and our receiver operating curve analysis.

Endothelial Cell Crossmatches
Primary human aortic endothelial cells (ECs) were isolated from aortic rings of explanted donor hearts (36). EC were cultured in M199 medium supplemented with 20% (vol/vol) FBS, penicillinstreptomycin (100 U/ml and 100 ug/ml; Invitrogen Life Technologies), sodium pyruvate (1 mM), heparin (90 ug/ml; Sigma-Aldrich) and EC growth supplement (20 mcg/ml; Fisher Scientific). ECs from passages 7-8 were frozen and used in the EC-XM. Two different ECs (phenotyped as follows, donor CAR: HLA A2, A68, B60, B65; and donor Y126: HLA A1, A11, B35, B37) were employed avoiding for each KT recipient any HLA class I match with the kidney graft which could yield a reaction towards donor-specific HLA antigens. A total of 2 ×10 5 ECs were incubated 30 min with 100 ul patient serum on ice. ECs were washed three times and incubated with 50 mc of 1:400 diluted FITC-AffiniPure F(ab') 2 Fragment Goat Anti-Human IgG Fc fragment (Jackson ImmunoResearch Laboratories) for 30 min on ice. After three washes, cell fluorescence was analyzed on a FACSCalibur flow cytometer using CellQuest software (BD Biosciences). Gates for forward and side scatter measurements were set on EC, and a minimum of 10,000 events was acquired. Positive EC-XM threshold was set at two standard deviations (50 Median Channel Shift) above the mean of negative control serum tests. EC-XM were only performed in 83 pre and 103 post-KT cases due to insufficient sample.

HLA Epitope Mismatch Characterization
HLAMatchmaker software according to the July 2020 update (ABC and DRDQDP Eplet Matching Program V3.1, http:// www.epitopes.net) was used to define potential HLA-EM between donors and recipients (37). High-resolution typing for all donors and recipients was performed or inferred using the HaploStats tool (www.haplostats.org) selecting the most likely high-resolution typing for HLA-A, B, C, DR and DQ according to three-five highest haplotype frequencies in the population of each one (Caucasian, African American, Asian or Hispanic).

Statistics
Data are presented as mean ( ± standard deviation), median, interquartile range, or number (percentage) based on data distribution. Comparisons between clinical variables were carried out using Student's T test for parametric continuous variables and U Mann-Whitney or Kruskal-Wallis test for nonparametric data. Chi-squared or Fisher's exact tests were used to test categorical variables. Survival analyses were performed using the Kaplan-Meier method using the log-rank test. Logistic regression analysis was used to estimate the odds ratio (OR) for ABMR h DSA pos development. All variables with a p-value <0.10 in the univariate analysis were included in the multivariate analysis. Statistical analysis was performed using SPSS v.27.0 (IBM Corp., Armonk, NY, USA) and p-values <0.05 were considered statistically significant.

Clinical Characteristics and Graft Survival
The selected 118 patients were grouped according to Banff diagnostic categories: category 1 or normal biopsy (n = 19), category 2 or ABMR histology (ABMR h , n = 52) and category 5 or IFTA (n = 47). Thirty patients (25.4%) lost their grafts and 13 died with a functioning graft (11%) during the study period. Death-censored graft survival 68 months after the biopsy [IQR  was worse in ABMR h cases than in those with IFTA or normal biopsies (Supplementary Figure 2). Baseline characteristics showed that normal histology patients were more frequently males, whereas ABMR h patients received grafts from younger donors and were more frequently retransplanted. ABMR h biopsies were less frequently surveillance biopsies and were performed later post-KT. At biopsy time, ABMR h patients had worse glomerular filtration rate (GFR) and higher proteinuria. Finally, IFTA patients were more frequently receiving calcineurin inhibitors and less on mTOR inhibitors ( Table 1).

Pre-Transplant Combination of HLA-DSA and Non-HLA Antibodies
Detection of pre-transplant HLA-DSA and/or AT 1 R-Ab were highly associated with ABMR h compared with IFTA and normal biopsies (66.7 vs. 25.6 vs. 21%, p <0.001, Figure 1C). Nine ABMR h cases presented with simultaneous HLA-DSA and AT 1 R-Ab  Figure 1D).

Patients With ABMR h With and
Without HLA-DSA From 52 patients with ABMR h 14 (26.9%) had no peri-biopsy HLA-DSA. ABMR h DSA pos cases were more frequently HLA sensitized, less well DR-matched with their donors and received more frequently a graft from a deceased donor than those ABMR h DSA neg . No differences were found in graft function or immunosuppression at biopsy ( Table 2). Patients showed similar microvascular inflammation, but diffuse C4d was more frequent in ABMR h DSA pos cases (27% vs 0%, p = 0.07, Table 2). Graft survival was similar between both groups ( Figure 2). We assessed pre-and post-transplant non-HLA antibodies in ABMR h DSA neg cases. Of 7 cases with pretransplant sample, two had EC-XM + but none showed MICA-Ab, AT 1 R-Ab or ETAR-Ab (Table 3A). After KT, one had coexistent MICA-Ab, AT 1 R-Ab and ETAR-Ab; one had   Table 1A). In 9/14 ABMR h DSA neg patients (64.3%) we could not identify any of the non-HLA antibodies studied.

HLA Epitope Mismatch Characterization
The median number of class I and class II HLA-EM in our cohort were 16 (0-36) and 18 (0-46) respectively. Among them, 10 class I and 7 class II HLA-EM were antibody-verified (HLA-EM ver ). We observed similar class I and class II HLA-EM ver in all three groups of study (data not shown). We compared the load of HLA-EM ver between ABMR h DSA pos and ABMR h DSA neg patients, finding similar class I but significantly higher class II and DRB HLA-EM ver in ABMR h DSA pos cases (8 vs 4.5, p = 0.046; 5 vs. 0.5, p = 0.044, Figure 3).

DISCUSSION
We report here that ABMR damage in KT recipients occurs in a significant proportion of cases without the detection of HLA-DSA at biopsy. We have evaluated the role of non-HLA antibodies, such as AT 1 R-Ab, ETAR-Ab, MICA-Ab or anti-EC antibodies detected with crossmatches and found they could not explain ABMR h DSA neg . Our results suggest a synergistic interaction between pre-transplant AT 1 R-Ab and HLA-DSA to    produce ABMR h DSA pos or facilitate de novo appearance of HLA-DSA, but not to induce ABMR h DSA neg . Interestingly, it appears more strongly associated with ABMR h than incompatibility evaluated through HLA-EM analysis. The relationship between ABMR h and HLA-DSA has been described in KT recipients for over 20 years (1,2). However, there is increased evidence that ABMR compatible histological lesions may be present in the graft without detectable circulating HLA-DSA (18,38). Up to 27% of our ABMR h patients did not show circulating HLA-DSA at the time of biopsy. This could be attributed to the inability of current techniques to detect these HLA antibodies or due to the participation of a different set of antibodies in graft damage. ABMR h DSA neg patients presented significantly lower class II and DRB HLA-EM compared with ABMR h DSA pos cases. This finding strengthens the hypothesis of the participation of other mechanisms of damage in these cases rather than non-detected HLA-DSA. However, neither AT 1 R-Ab, ETAR-Ab, MICA-Ab nor antibodies identified with EC-XM before or after KT were able to explain the ABMR h DSA neg cases in our study. We describe here that ABMR h patients without  HLA-DSA showed similar graft function, immunosuppressive treatment, histological features at biopsy and graft survival at the end of follow-up compared with ABMR h DSA pos cases. Like us, Sablik et al. (17) reported a similar histological phenotype in ABMR h DSA pos and ABMR h DSA neg patients, but a larger study by Senev et al. (18) found that ABMR h DSA pos biopsies were more frequently C4d positive compared with ABMR h DSA neg cases, as the unique histological difference between the groups. In our series, although C4d positivity was similar between both groups, C4d intensity was higher in the ABMR h DSA pos group. In our cohort, graft survival was similar between both groups, in agreement with results reported by Sablik et al. (17) but in contrast with the study from Senev et al. (18), which included mostly active ABMR cases without chronicity, unlike our cohort.
KT recipients may produce immune responses through indirect recognition against foreign proteins or even against own proteins expressed by the donor graft acting as autoantigens due to different factors that induce graft damage during the transplant process. These antibodies may then react against polymorphic alloantigens, like HLA related MICA or MICB, or against autoantigens like AT 1 R, ETAR, agrin, vimentin, perlecan, K-tubulin, etc. (39)(40)(41) which may be prevalent in KT recipients. Some of these autoantibodies and new ones recently validated (42) have not been evaluated in our cohort yet. They might explain some ABMR h DSA neg cases. Some groups have evaluated the relationship between antibodies directed against ECs-the barrier between donor and recipient -and graft survival (43), and exploratory studies have employed array techniques in limited series with antibodies against ECs validating potential target proteins with ELISA (44,45). Jackson and col. were able to identify four antigenic targets expressed on ECs in nine patients with ABMR h DSA neg (44). They found that antibodies against these proteins in pre-transplant sera predicted ABMR h DSA pos . In our cohort, of seven ABMR h DSA neg cases with pre-transplant samples, two had a positive EC-XM + , but none showed MICA-Ab, AT 1 R-Ab or ETAR-Ab. In line with our results, a recent report from Delville et al. (23) found that only 26% of patients with early acute ABMR h DSA neg had pretransplant AT 1 R-Ab using our same threshold of 10 UI/ml. Moreover, MICA-Ab were only detected in two of these ABMR h DSA neg cases. However, these cases had preformed IgG antibodies against constitutively expressed antigens of microvascular glomerular cells (23). Of note, our two cases with pre-transplant EC-XM + developed ABMR h within the first 12 months of KT, while the other twelve developed ABMR h later on. Unlike Lefaucheur et al. (46), despite employing the same threshold for AT 1 R antibodies, the presence of these antibodies in our ABMR h DSA neg cohort is negligible. Nevertheless, our overall prevalence of 25% in posttransplant AT 1 R-Ab is not different from theirs. Unfortunately, these authors do not analyze the relation between pre-transplant AT 1 R-Ab and ABMR.
We report here a strong and independent association between pre-transplant AT 1 R-Ab and ABMR h DSA pos development. AT 1 R can be found in several cell types such as vascular endothelial cells and binds to angiotensin II (39,47). First report linking AT 1 R-Ab and kidney allograft rejection suggested a potential relationship between AT 1 R agonistic antibodies and vascular injury (19,39). Subsequently, pre-or post-transplant AT 1 R-Ab detection have been linked to both rejection and allograft failure (19,48). Philogene et al. (24) described higher post-transplant AT 1 R-Ab levels in patients with ABMR compared with patients with cellular rejection or those without rejection, however, they provided no data regarding pretransplant AT 1 R-Ab. In another report (49), pre-and posttransplant AT 1 R-Ab were strongly associated with biopsy-proven rejection, not specifically ABMR. Some reports suggest that non-HLA and HLA-DSA antibodies may function in synergy (24,49). Taniguchi et al. (49) reported lower graft survival mainly in the presence of de novo AT 1 R-Ab and HLA-DSA at biopsy with lesions compared with those cases with HLA-DSA alone. Here we show a strong association of pre-transplant AT 1 R-Ab with post-transplant HLA-DSA, either persistent preformed or de novo, and with ABMR h DSA pos development. This association may be of utmost importance for KT outcomes. We previously reported the strong association among persistent preformed HLA-DSA and lower ABMR free survival, only surpassed by the development of de novo HLA-DSA (34). Moreover, here we show that all 16 ABMR h patients with pre-transplant AT 1 R-Ab had HLA-DSA at biopsy, nine of them maintained the preformed HLA-DSA and seven developed de novo HLA-DSA. We found no association between pre-transplant AT 1 R-Ab and graft survival, in line with other reports (26,49). In our multivariate analysis, pre-transplant HLA-DSA and AT 1 R-Ab were independent predictors for ABMR h . Our study may not be powered enough to assess the relationship between AT 1 R-Ab and graft loss. Given the strong and already known association between ABMR and increased risk of kidney allograft loss (34,(50)(51)(52), our data supports that pre-transplant AT 1 R-Ab assessment should be carefully considered in KT candidates.
In the last years, HLA-EM analysis has been proposed as a better strategy to prevent HLA-DSA development than antigen matching (29). Here we confirm that class II and DRB dnDSA development may be predicted with HLA-EM, as previously reported (30), however, only a weak association was observed with DQB dnDSA, probably due to the limited number of cases included. Interestingly, neither class II, DRB or DQB HLA-AM were able to predict dnDSA. As mentioned, the detection of lower number of class II and DRB HLA-EM in ABMR h DSA neg cases may contradict the idea of undetected HLA-DSA responsible for the damage. Class II and DRB HLA-EM associated with ABMR h DSA pos , although the existence of preformed HLA-DSA or AT 1 R-Ab are more potent predictors of ABMR h DSA pos in our experience. In our study, ABMR h DSA neg could not be explained by higher HLA-EM or by the non-HLA antibodies evaluated. Interestingly, an alternative mechanism to produce ABMR h termed "the missing-self hypothesis" has been proposed. According to it, the inability of graft EC to provide HLA I-mediated inhibitory signals to recipient circulating NK cells may trigger NK cell activation, resulting in endothelial damage and chronic vascular rejection (53).
The main limitation of our study is the restricted number of ABMR h DSA neg cases in the whole cohort. In order to further increase its number and the significance of the study, a multicenter trial is advisable. Besides, it is based on a mix of indication and surveillance biopsies which introduces heterogeneity in the timing and clinical picture of patients. Of note, EC-XM were performed with aortic cells which may not express the same proteins as a renal EC. Last, another limitation may be the use of inferred four-digit HLA typing for HLA-EM analysis. Despite careful estimation of second field HLA typing, we cannot rule out the possibility that some rare HLA genotypes are not correctly assigned, as recently suggested (54). However, ours is a large well characterized cohort of KT recipients, reflecting clinical practice, with thorough analysis of biopsies, including electron microscopy, crucial to detect some cases of ABMR h and with systematical study of HLA-DSA and a known set of non-HLA antibodies.
In summary, although the majority of patients with HLA-DSA at the time of biopsy show ABMR h , almost 30% of ABMR h patients did not show evidence of circulating HLA-DSA. These patients were more frequently HLA unsensitized pretransplant and less HLA matched but did not show other specific characteristics at transplantation or at biopsy. Neither AT 1 R-Ab, ETAR-Ab, MICA-Ab nor antibodies identified with EC-XM before or after KT were able to explain ABMR h DSA neg cases. Importantly AT 1 R-Ab with or without HLA-DSA before KT clearly increased the risk of ABMR h DSA pos , suggesting it should be included in the pre-transplant immune assessment together with HLA-DSA.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Parc de Salut Mar Ethical Research Board. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
MC designed the study, coordinated logistics, analyzed the results, and drafted the manuscript. LL-M analyzed the results and drafted the manuscript. DR-P analyzed the results and revised the manuscript. CBut coordinated lab procedures and revised the manuscript. JG contributed with the assessment of the graft biopsies. MP-S, CBur, AB, CA-C, and SS-U revised the manuscript. MF coordinated sample drawing and storage. NV supervised HLA and non-HLA antibody interpretation. ER and JP evaluated the design of the study and revised the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
This study was performed with funding from projects PI13/ 00598, PI16/00617, and PI20/00090 (Spanish Ministry of Health ISCIII FIS-FEDER); RD16/0009/0013 (ISCIII FEDER REDinREN) and 201822-10 (Fundacióla Maratóde TV3). MC received grants of the Sociedad Española de Nefrología and Parc de Salut Mar for a research stay at UCLA Immunogenetics Center (Los Angeles, USA). ER is supported by National Institute of Allergy and Infectious Diseases Grants RO1 AI135201. One Lambda provided reagents but had no role in the design of the study or the analyses and writing of the manuscript.