ABSTRACT. The wide application of high-throughput DNA sequencing methods in the typing of human and non-human MHC genes since the last decade has boosted the generation of big data in immunogenetics, where the identification of thousands of alleles defined at very high resolution can now be used to investigate detailed molecular variation at the individual level rather than global genetic profiles in populations. This poses new challenges in data analysis, as new biostatistical and computer tools need to be developed to visualize this substantial amount of molecular diversity and decipher it in terms of genetic evolution. Following the work carried out by the EFI Population Genetics Working Group and the Population Genetics, Anthropology and Evolution (PGAE) Component during the last (18th) IHI Workshop, we here present the results of extensive NGS data analyses at 11 HLA loci in a large set of more than 300,000 individuals from Europe, North Africa, Asia and America, as well as at 5 MHC loci in 63 chimpanzees and 43 bonobos, that we compared together by implementing and redesigning the hla-net.eu bioinformatics pipeline now available as a fully user-interactive interface. While the data of the 13th to 18th Workshops compared at the 2nd-field level do improve the global map of HLA population diversity, mostly correlated with geography, the NGS data analysed at the 3rd- and 4th-field levels significantly refine the characterization of HLA variation at the inter-individual level, emphasising remarkable differences between individuals within some isolated populations. This high-resolution dataset also allowed estimating HLA peptide-binding affinity thereby highlighting potential functional differences among individuals of different populations. Finally, we performed direct comparisons of HLA and homologous Patr and Papa genes among individuals of the different species, providing new insights into the evolutionary history of MHC in humans and their closest relatives.

ABSTRACT. The 5Mbp MHC genomic region is characterized by immense sequence and structural variation and is the most significant region of the genome affecting immune mediated disease. Despite this medical importance, the MHC remains inadequately represented in the current version of the human reference genome. We recently completed the sequences from the original MHC consortium cell lines, to include eight full-length haplotypes, including representatives from the four major HLA class II structures -DR1, -3, -4 and -5. Although this represents a substantial improvement from that previously available, these samples are of European descent and do not capture the full complexity of the global population. To expand the MHC reference set further, we first made technical improvements to our approach. To increase the sequencing efficiency, we developed new targeted nanopore long-read methods, resulting in greater depth of coverage through the MHC. We also modified our bioinformatics pipeline to utilize both long and short sequencing read data. Our combined methods provide the scaffolding and high depth of coverage to perform a fully comprehensive de novo assembly that can identify and integrate structural variations that were not previously characterized. We have now expanded the project to complete 100 reference haplotypes representing diverse ancestries with a goal of expanding to include another 500 MHC homozygous samples we have collected through multiple collaborations. These references will be integrated into the human reference genome. They will also be used to construct a population reference graph (PRG) that can in turn be used to study associations with disease through the entirety of ~180 genes and intergenic segments that comprise the MHC genomic region.

ABSTRACT. MICA and MICB are stress-induced ligands of the NKG2D receptor that regulate the activity of NK- and T cells. The MICA and MICB genes are located within the MHC complex and are in strong linkage disequilibrium with HLA-B. Since 2019, we have genotyped over two million samples for MICA and MICB and thereby identified 22,883 samples with a MICA gene duplication (1.0%) and 9,262 samples with a MICA gene deletion (0.4%). In certain populations, these frequencies can even exceed 5%. Since copy number variations might alter the level of MICA expression and consequently the immune responses against cancer or infectious diseases, we aimed for a detailed characterization of the underlying genomic organization. By using whole genome sequencing and targeted PCR on 53 samples with MICA duplications and 17 samples with MICA deletions, we mapped the recombination events to CT-repeat regions approximately 13.7 kb upstream of both MICA and MICB. These repeats lie within a 30 kB long segmental duplication that makes them susceptible to non-allelic homologous recombination. Even though all sequenced recombinations share a similar organization, the sequences of the individual haplotypes vary in detail and are indicative of independent genomic events. Taken together, our study highlights the high frequencies of MICA copy number variations and their organization in a particularly challenging genomic region. This knowledge will facilitate the development of MICA copy number-sensitive genotyping strategies that might be beneficial or even crucial for future applications.

ABSTRACT. MICA is a ligand for the NKG2D receptor and regulates the activity of NK and T cells. The MICA gene is located within the MHC complex in close proximity and linkage disequilibrium to HLA-B. Previously, a MICA gene deletion haplotype with strong linkage to HLA-B*48 has been described in Asian populations. In addition, several individual cases of non-HLA-B*48-linked MICA hemizygosity have been reported. To strengthen these initial observations, we analyzed MICA hemizygosity in over two million potential stem cell donors. Very high frequencies of MICA hemizygosity were identified in Mexicans (9.2%), Chileans (3.5%) and Black Americans (2.7%). While the previously described HLA-B*48-linked haplotype clearly dominates in Chileans, we identified other recurrent MICA gene deletion linkages: HLA-B*15:16:01G in Black Americans and HLA-B*39:06:02:02 in Mexicans. Interestingly, MICA hemizygosity is rare in European populations (Germans: 0.1%). This is inverse to the previously observed HLA-B*27:02-linked MICA gene duplication, which has its highest frequencies in the Eastern European populations (up to 2.9% in Croatians). Overall, in our cohort of two million predominantly European samples, we identified 9,262 MICA hemizygous samples. This observation clearly demonstrates that copy number variations of MICA are not rare and should be considered in MICA genotyping.

ABSTRACT. Telomeres protect the ends of normal chromosomes from degradation or abnormal fusions, with the consequent maintenance of genome stability. They consist of tandem repeats of nitrogenous bases and their number determines the maximum lifespan of a cell. The aging process is very complex and is determined by genetics, environmental, and lifestyle factors. Previous studies within the IHIWS component “Immunogenetics of Ageing” showed that longevity is associated with positive selection of HLA-DRB1*11 and DRB1*16 associated haplotypes, shown to be protective against diseases. Within the 18th IHIWS, we aimed to investigate the relevance of telomere length for successful ageing and its association with classical HLAs. In total 857 individuals: 316 unrelated elderly (>76 years) and 541 young controls (age 18–35 years),from Bulgaria, Turkey, Romania, and Poland were investigated. Telomere length was assessed using the Absolute Human Telomere Length Quantification qPCR Assay Kit. The comparison of telomere length in the analysed groups showed some significant differences between the studied populations (p<0.0001). Elderly from Romania had the shortest telomeres, while those from Turkey had the longest ones. Among young controls, Poles had the shortest while Romanians had the longest telomeres. Furthermore, an association between telomere length and the presence of HLA-DRB1 alleles was found. DRB1*11 was associated with longer telomere length in Bulgarians (p=0.004). This relationship was observed in both elderly (p=0.053) and young (p=0.019) Bulgarians. Furthermore, we found that Bulgarians carrying HLA-DRB1*11 and/or DRB1*16 had significantly longer telomeres than those with other HLA alleles (whole study group, p=0.001; elderly, p=0.007; young controls, p=0.001). These results highlight the inter-population differences in telomere length and suggest some associations with HLA. In addition, these data could help to identify immunogenetic profiles associated with longevity.

ABSTRACT. Immune-mediated diseases are among the strongest selective pressures driving human evolution. The major histocompatibility complex (MHC) locus that encodes human leukocyte antigen (HLA) proteins plays a vital role in our adaptive immune responses, and thus is crucial in understanding the human evolutionary process. Several non-competing natural selective regimes exist to explain how the MHC locus evolved, including balancing, and pathogen-driven positive and negative selection. However, owing to the complicated genomic structure of the MHC locus, it remains challenging to nominate specific genomic variations and haplotypes driving the observed selection signatures. In this work, we studied 1,832 African American (AA) individuals with majority African (74.4%) and European (25.6%) ancestry, and 594 Latino American (LAT) individuals with majority African (4.4%), European (49.7%) and Native-American (45.9%) genetic ancestry. By using whole-genome sequences in two admixed populations within the United States, we inferred 726 unique classical HLA alleles at two-field resolution including including three classical HLA class I (HLA-A, B, C) and five class II (HLA-DPA1, -DPB1, -DQA1, -DQB1, -DRB1) genes. In LAT, we observed 4.4% African ancestry genome-wide versus 7.7% in the extended MHC region, representing a 1.75-fold increase. In contrast, in AA, we did not observe a genome-wide significant deviation in local ancestry. Furthermore, we nominated specific classical HLA alleles (e.g. B*14:02, C*08:02 and DQB1:05:01) and long-range haplotypes (e.g., A*33:01~B*14:02~C*08:02 and DQA1*01:01~DQB1*05:01~DRB1*01:02) that are under recent selection and responsible for deviations in local ancestry in LAT. Our work provides clarification of the dynamics that shaped the MHC locus, and thus improves our understanding of how selection-driven alleles contributed to immune-mediated disease susceptibility today.

ABSTRACT. Human leukocyte antigens (HLA) are highly polymorphic proteins and essential for transplantation. With the exception of red blood cells, almost all of our cells include HLA-A, B, and C (class I), whereas immune cells primarily contain HLA-DR, HLA-DQ, and HLA-DP (class II). It is believed that heterozygous people at HLA loci will exhibit a wider variety of pathogen-derived peptides than homozygotes, increasing the likelihood of inducing a particular immune response. This heterozygote advantage hypothesis, when sequence level is taking into account, lead to the idea of a divergent allele advantage. Furthermore, HLA genes’ high polymorphism may also be explained by this mechanism. The Grantham distance between the two alleles of each HLA locus can be calculated using the HLA evolutionary divergence (HED) metric. Likewise, HED is a straightforward replacement for the size of the repertoire of peptides presented by the HLA molecules. An new application to calculate HED values for Class I and Class II loci is available from here: https://txor.shinyapps.io/ched/. Protein sequences corresponding to the peptide-binding domain (exons 2 and 3 for class I and exon 2 for class II) of each HLA molecule were obtained from the IMGT HLA database. This application was built in shiny, an R package for the development of interactive web apps. Genotyping at two–field (4 digits) resolution for all classical HLA loci A, B, C, DRB1 and DQB1 can be used as input for this HED calculator. The application allows calculations for individual loci where each pair of alleles are single picked and also using files with several HLA genotypes. From a clinical perspective, HED has been associated liver transplantation rejection, response to checkpoint medications in some malignancies, improved HIV management, and haematopoietic stem cell transplantation outcome. HED calculator is easy-to-use and do not require any programing skills from their users.

ABSTRACT. The major histocompatibility (MHC) locus houses the polymorphic human leukocyte antigen (HLA) genes and contains the most disease association signals in the genome. However, the complex sequence features and unique evolutionary history of the MHC make interpretation of these signals challenging. For instance, ~50% of non-coding sequence at the MHC is comprised of low complexity repeats and transposable elements, known to be reservoirs of genetic and structural variation leading to extensive population-level variation. Linkage disequilibrium adds additional challenges to interpretation. The complexity presented by these features is compounded by the difficulty of assembling highly variable genomic sequences. Traditional assembly approaches relying on reference genome alignment can leave short read datasets partially unaligned, or improperly aligned, and variation is missed. Long read sequencing is an option for overcoming such limitations in the future, though it is still cost-prohibitive and comes with a high error rate. Genome graphs have great potential but are still in developmental stages and are limited by the extent of sequenced MHC regions available as graph backbones. As an alternative to such challenges, we implement a hybrid approach: a reference genome-guided, de novo assembly method, previously established to scaffold genomes without a reference sequence to the genome of a related species. Aligned and unaligned reads are de novo assembled separately, after which, unique contigs are integrated into a single scaffold. We apply this to a human multiple sclerosis case-control cohort, comprised of ~3,000 individuals with full MHCs, sequenced at high coverage using Illumina NGS. In this way, we rapidly and sensitively characterize both coding and non-coding sequence variation associated with multiple sclerosis. Achieving high resolution MHC sequences for previously inaccessible disease cohorts could transform our understanding of the MHC in human disease.

ABSTRACT. The role of HLA-E in haematopoietic cell transplantation (HCT) has been investigated by a few studies but mostly for patients receiving well HLA matched transplants. Here we describe our analysis of HLA-E matching in 1,513 HCT UK patients and their unrelated donors. Genotypes were generated by full-length PacBio Single Molecule Real-Time DNA sequencing. We investigated HLA-E matching at both the protein and full allelic levels. A total of 943 (62.3%) patients were matched for HLA-E (HLA-Em) at the second field, whilst 821 (54.3%) were HLA-Em at full allelic resolution. In all 584 patients received a <10/10 HLA matched transplant, 290 (49.7%) of which were also mismatched for HLA-E (HLA-Emm). After adjusting for clinical factors that affected outcome prognoses, multivariate analysis showed patients receiving a <10/10 HLA matched donor, who were also HLA-Emm had a significantly reduced risk of relapse (HR 0.55; 95% CI 0.36 to 0.84, p=0.005) compared to their HLA-Em counterparts. Correspondingly, HLA-Emm was associated with a significantly reduced risk of Event-Free Survival (EFS) (HR 0.72; 95% CI 0.54 to 0.97, p=0.03). A beneficial but not significant effect on Overall Survival (OS) and Transplant-Related Mortality (TRM) was also observed for <10/10 HLA-Emm patients compared to HLA-Em cases (OS: HR 0.82, p=0.12; TRM: HR 0.83, p=0.34). We hypothesise that HLA-Emm in an already HLA mismatched setting is providing enough genetic disparity to improve the Graft-versus-Leukemia effect. In a complimentary study at our centre, 59.5% of patients with a choice of only <10/10 donors at verification typing were also HLA-Emm. Of those cases, 76.1% had a choice of more than one HLA-Emm donor. Together this suggests that in patients receiving a less well HLA matched HCT, who would otherwise have poorer post-transplant outcomes, HLA-Emm is beneficial for patient survival and reducing risk of relapse and is feasible to include in donor selection.

ABSTRACT. Background Many studies have reported the relevance of donor-derived cfDNA (dd-cfDNA) after lung transplantation (LT) to diagnose and monitor acute (AR) or chronic rejection or infection (INF), analysis of cfDNA fragment size is not sudied. The aim of this study was to determine the clinical relevance of dd-cfDNA and cfDNA size profiles in events (AR and INF) during the first month after LT. Methods This is a prospective, single-center study including 62 LT patients at the Marseille Nord Hospital (LARA project). Total cfDNA quantification was performed by fluorimetry and digital PCR, dd-cfDNA by NGS (AlloSeq cfDNA-CareDX®) and the size profile by BIABooster (Adelis®). After selection of small (< 160 bp) cfDNAs by BIABooster technology, chimerism analysis by ddPCR (HLA KMR marker, GenDX) was performed. A biopsy at D30 allowed establishing the following groups: stable, nonstable (AR, INF and AR+INF). Results Quantification of total cfDNA, for all methods, was not correlated with the patient's status at D30, percentage of dd-cfDNA was significantly higher for nonstable patients at D30 (p=0.0004). A threshold of 1.72% of dd-cfDNA determined the nonstable patients (NPV: 91.4%). The analysis of small sizes (80-120 bp) identified the INF with a threshold of 3.7% (PPV: 100%). An algorithm combining the two analyses allows to significantly differentiate the type of lesions due to allografts. Conclusions Our algorithm aims at guiding the performing of biopsies, which are invasive and risky for the patient. This combined noninvasive biomarker of allograft injury requires to be confirmed. For this purpose, a control cohort is being included and a multicenter study is planned.

ABSTRACT. The non-classical HLA class II molecules, HLA-DM and -DO, regulate peptide loading to other HLA class II molecules. They have been less well studied than classical HLA genes and are thought to be less polymorphic. Given their position in the major histocompatibility complex, these genes may be good candidates as haplotype markers. We have used PacBio Single Molecule Real-Time (SMRT) DNA sequencing to perform full gene sequencing of HLA-DMA, -DMB, -DOA, and -DOB from 49 IHIW cell lines. PCR amplicons encompassing the entire gene were pooled into barcoded adaptor libraries and SMRT DNA sequencing was performed on a PacBio Sequel. An in-house bioinformatics pipeline was used to analyse this data. A total of 17, 26, 17 and 18 unique sequences were generated for HLA-DMA, -DMB, -DOA, and -DOB, respectively. We observed 13 novel alleles, all with SNPs in intronic regions. Six reported typing results, previously observed to be homozygous, were shown to have two different intronic variants of the same coding allele. None of these results occurred in cell lines reported to be homozygous or consanguineous. Concordance with previously available typing for these genes was high, 96 out of 99. The differences between the previously reported typings and that derived here were all located with coding regions. A discrepant HLA-DOA typing was shown to differ from the previously reported sequence by a single SNP. The observation of multiple novel alleles within a small sample size suggests there is greater polymorphism within these genes than previously thought. However, it should be noted that IHIW cell lines cover a range of ethnicities and cannot be considered as a random sample of a population. The identification of zygosity changes within intronic regions when completing full gene sequencing highlights the importance of sequencing the entirety of these genes. These results therefore establish the foundations for larger-scale, more comprehensive future studies of HLA-DM and -DO.

ABSTRACT. Urine CXCL10 is a promising non-invasive biomarker for detection of renal allograft rejection. The aim of this study was to investigate the clinical utility of renal allograft monitoring by urine CXCL10 in a randomized trial. We stratified 241 patients, 120 into an intervention and 121 into a control arm. In both arms, urine CXCL10 levels were monitored at three specific time points (1, 3, and 6 months posttransplant). In the intervention arm, elevated values triggered performance of an allograft biopsy with therapeutic adaptations according to the result. In the control arm, urine CXCL10 was measured, but the results concealed. The primary outcome was a combined endpoint at one-year posttransplant (death-censored graft loss, or clinical rejection between month 1 and one-year, or acute rejection in one-year surveillance biopsy, or chronic active T cell-mediated rejection in one-year surveillance biopsy, or development of de novo donor-specific HLA-antibodies, or estimated GFR<25ml/min). The incidence of the primary outcome was not different between the intervention and the control arm (51% vs 49%; RR 1.04 [95% CI 0.81-1.34]; p=0.80). When including 175/241 (73%) patients in a per-protocol analysis, the incidence of the primary outcome was also not different (55% vs 49%; RR 1.11 [95% CI 0.84-1.47]; p=0.54). The incidence of the individual endpoints was not different as well. The CXCL10 burden, calculated as the mean of measurements at the three monitoring time points, predicted rejection in one-year surveillance biopsies defined by the Banff 2019 classification (p=0.01). CXC10 values often reflected inflammation processes in the allograft (rejection, polyomavirus-BK infection). In conclusion, this study could not demonstrate a beneficial effect of urine CXCL10 monitoring on one-year outcomes. However, CXCL10 can provide important information on the inflammatory status of the renal allograft (ClinicalTrials.gov_NCT03140514).

ABSTRACT. Chimpanzees are humans closest living relatives, and they share a common ancestor that lived about 5-7 million years ago. The human KIR gene family consists of 17 members, which may show polymorphisms. Each KIR haplotype contains a variable number of genes, including four framework genes; KIR3DL3, KIR3DL2, KIR2DL4, and KIR3DP1. Chimpanzees possess a comparable KIR gene system, with a total repertoire of 13 genes, which includes the above mentioned framework genes. Only three genes, KIR2DL4, KIR2DL5, and KIR2DS4, are considered orthologous based on their high nucleotide sequence similarity in humans and chimpanzees. The KIR repertoire in chimpanzees is further characterized by an expansion of the lineage III KIRs (both D1-D2 and D0-D1-D2 structures). In the past, three chimpanzee KIR haplotypes have been characterized by sequencing different BAC libraries. Twenty additional KIR haplotypes have been deduced by KIR genotyping of a panel of chimpanzees. Recently, we have characterized the KIR transcriptome of a West-African chimpanzee cohort, and deduced KIR haplotypes based on segregation analyses. However, some of the deduced haplotypes contain different combinations of KIR genes that are not covered with the BAC-library sequenced haplotypes, and therefore gene combinations and locations could only be inferred. In the present study, our aim was to resolve the chimpanzee KIR gene organizations by characterizing a variety of Patr-KIR haplotypes using Cas-9 enrichment and Oxford Nanopore Technologies sequencing approach. The data provides a detailed insight on the location of KIR genes and shows possible duplication of genes on a haplotype. Moreover, we discovered that some haplotypes can contain recombinant KIR genes, which may impact the binding and signal potential of the encoded receptors.

ABSTRACT. Kidney pair donation (KPD) is one of a few efficient strategies to improve the chance of receiving a compatible kidney for patients with incompatible living kidney donor. Czechia is involved in an international KPD together with Austria (since 2016) and Israel (since 2019), where the Prague transplant center is the coordinating center for these exchanges. In 2020 we developed a software called TXMatching to simplify the process of finding optimal matching between donor recipient pairs, especially for sensitized patients. There are two key features of the tool. First, it supports calculation of HLA matching and virtual crossmatching for both high and low resolution typing. This includes a novel algorithm that determines whether the reactivity of an antibody is against the alpha or beta chains of DQ and DP. Second, it allows broad customization to accommodate the needs of all the countries and pairs. This includes recipient’s MFI cutoff or acceptance criteria for donor (blood group, weight, age etc.). Moreover, when calculating the optimal solution, maximum sequence length, debt between centers, HLA scorers and other parameters are configurable. The software runs as a web application and is fully accessible from any PC or mobile device. The design ensures that the app is easy to use and at the same time gives each transplant center full control over their patients and donors. As proof that the software is useful, we present a case study of an immunologically complicated patient actively waiting for 5 years for his second kidney transplantation in both KPD and deceased donor programs that found compatible kidney on the first attempt with the new algorithm. This donor would have been rejected based on two antibodies that would have been considered donor-specific if the virtual crossmatch was not performed on a high resolution and no discrimination was made between DQA and DQB reactivity. The patient is now 18 months since transplant without any rejection incidents.

ABSTRACT. Genomic loss of mismatched HLA alleles is a possible driver of relapses after hematopoietic stem cell transplantation, especially in a haploidentical transplant. Early detection of HLA loss is needed for comprehensive clinical decision making. In case HLA loss did not occur and mismatched HLA alleles are still present, donor lymphocyte infusion or modulating immunosuppression is a treatment option. In the case HLA loss occurs, a second allo-transplant from a different donor is a treatment option. HLA loss can be measured by quantifying the levels of markers within the region where HLA genes are located. This can be measured by quantitative PCR, which is highly accurate. However, the number of markers currently available in commercial kits using qPCR are insufficient to cover the complete population and cannot detect all types of markers. We are investigating if this method can be developed using Next-Generation Sequencing (NGS) for increased informativity and combination with NGStrack chimerism measurement. HLA genes are clustered in the MHC region on the short arm of chromosome 6. Multiple short tandem repeat (STRs) can be found in between the HLA genes, with the number of repeat units being highly variable among individuals. These can be used as informative markers for detecting HLA loss using NGS. For this purpose, we have developed PCR assays for over 15 STRs interspersed between HLA-A (on one side of the MHC region) and HLA-DPB1 (on the other side of the MHC region). The amplicons are sequenced on an Illumina MiSeq and automatically analyzed in a new analysis tool. This tool reads the FastQ files, extracts the information on STR counts, and visualizes the STR lengths in graphs. Artificial chimeric samples were created and analyzed using this workflow, showing that STR markers within the MHC region can be applied to quantitatively monitor loss of specific HLA alleles within chimeric samples and can be an accurate indication of loss of a specific haplotype.

ABSTRACT. This study, part of a collaboration undertaken by a consortium of global blood services, industry and academia, assessed the performance of a universal blood donor genotyping platform capable of typing clinically relevant red cell (HEA), leukocyte (HLA) and platelet (HPA) antigens in a single test. DNA samples from consented blood donors were submitted by 7 blood services from Europe, Africa, North America and Australia with a genetic ancestry of 76% European, 14.3% African, 4.7% Hispanic, 3% South Asian, 1.4% other, 0.6% East Asian. Samples were collected centrally, plated and distributed for testing in laboratories in 3 different blood services. Genotyping of samples was performed in our laboratory using the GENETITAN-PROPEL workflow in batches of 94 on the Applied Biosystems™UK Biobank v2.2 Axiom™Array. HEA and HPA antigen inference was performed using the BloodTyper algorithm (Lane et al). HLA class I and II were imputed using Applied Biosystems™ HLA Analysis v2.12.0RC1 algorithm. Existing data on HEA, HLA and HPA types of the donors were used for concordance analysis. In phase one of the study, 4,510 samples were tested in our laboratory. Results showed that for HEA genotyping, a total of 58,856 comparisons were performed with concordant results in 99.88%. HLA and HPA typing records were available for a subset of donors. Initial analysis of 380 samples showed a 99.2% to 99.9% concordance rate for HLA- A, B, C, DRB1 and DQB1 but 98.6% for HLA-DPB1 at second field when compared typing previously performed by Next Generation Sequencing and PCR-SSOP. HPA typing results were available for HPA1,2,4,5,6,15. The probes for HPA-3 did not perform satisfactorily. For all other HPA antigens concordance was 100% except HPA-6 where a small number of discordances require further analysis. The study shows the genotyping array delivers highly concordant and reproducible results with a high level of agreement for samples tested at all three blood service laboratories.

ABSTRACT. The killer cell immunoglobulin-like receptor (KIR) cluster displays extensive chromosomal rearrangements that reshuffle gene content, reflected by a variable number of genes per haplotype. These recombination events not only diversify haplotype configurations, but might also generate novel hybrid entities, which consist of segments from two different genes. The common human KIR gene repertoire comprises 17 members, but is expanded by multiple reports of hybrid genes. For example, the IPD-KIR database documents 7 allelic variants for a hybrid KIR3DL1-KIR3DL2 gene, which are referred to as alleles of KIR3DL1. Most conventional typing techniques, however, might lack detection of these hybrid genes, suggesting that the pool of novel KIR entities may be larger. Other hybrid entities are discovered using haplotype sequencing and are often accompanied by gene deletions or introductions. Long read transcriptome and ultra-long read haplotype sequencing, utilizing PacBio and ONT platforms respectively, allowed definition and detection of hybrid KIR genes in macaques. We documented over 20 novel gene entities, which are represented by more than 60 allelic variants, expanding the previously defined gene repertoire of 22 macaque KIR genes. Chromosomal rearrangements mostly involved lineage II genes (KIR3D), but novel entities were also defined comprising segments of KIR3DL20 (lineage V), KIR1D (lineage III), KIR2DL04 (linage I), and pseudogene KIR2DP. Each hybrid KIR gene is in-frame and comprises all coding exons. The sophisticated mechanism that drives gene diversification is not completely understood yet. However, the extensive reshuffling of binding and signaling domains appears to generate receptors with differential biological functions. The relatively high rate of generating hybrid KIR genes might indicate an evolutionary strategy to protect against pathogen evasion and subversion.

ABSTRACT. The members of the killer cell immunoglobulin-like receptor (KIR) family display a variegated expression pattern on NK cells. This stochastic activation of KIR genes is orchestrated by multiple promoter regions that contain different sets of transcription factor binding sites. The proximal promoter, located directly upstream of each first exon, functions as a bidirectional switch that generates sense or antisense transcripts associated with gene activation or silencing, respectively. A distal promoter seems to modulate the activity of this switch via splicing of transcripts, whereas the intermediate promoter might be involved in tissue-specific expression. In humans, sequence variation in these promoter elements correspond with differential expression of KIR. For example, distinct promoter regions are defined for KIR2DL4, which displays non-stochastic expression, whereas for KIR3DL3, expression is lacking on circulating NK cells. Moreover, decreased expression of specific KIR allotypes might associate with SNPs in the distal promoter. In this study, we performed long read sequencing using an Oxford Nanopore Technologies platform to assemble complete KIR haplotypes in rhesus macaques, a species that is relevant for preclinical studies. These high-resolution haplotypes provide insights into the conservation and diversification of the KIR promoter regions. The bidirectional switch and the distal promoter seem to be largely conserved, indicating a strong selective pressure for these regulatory elements. In contrast, the intermediate promoter displays more sequence variation, especially for KIR3DL3 when compared to its counterpart in macaques, KIR3DL20. This latter gene is abundantly transcribed by circulating NK cells in macaques, indicating a diverged biological function. These insights enhance the currently limited understanding of KIR expression in macaques, and might help to translate the biological relevance of the differential KIR expression patterns.

ABSTRACT. Extreme polymorphism of human leukocyte antigens (HLA) and killer-cell immunoglobulin-like receptors (KIR) differentiates immune responses across individuals. Additional to T cell receptor interactions, subsets of HLA class I act as ligands for inhibitory and activating KIR, allowing natural killer (NK) cells to detect and kill infected cells. We investigated the impact of HLA and KIR polymorphism on the severity of COVID-19. High resolution HLA class I and II and KIR genotypes were determined from 403 non-hospitalized and 1,575 hospitalized SARS-CoV-2 infected patients from Italy collected in 2020. We observed that the activating KIR2DS4*001 allotype is associated with severe disease (OR=3.74, 95% CI 1.75-9.29, pc=0.003). KIR2DS4*001 in presence of its specific HLA ligands and inhibitory KIR3DL2*002 in absence its HLA ligand are also enriched in severe COVID-19 patients (OR=1.64, 95% CI 1.09-2.50, p=0.019), suggesting this combination acts in tandem to increase risk of developing severe COVID-19. We also observed the HLA class II allotype, HLA-DPB1*13:01 protects SARS-CoV-2 infected patients from hospitalization (OR=0.49, 95% CI 0.33-0.74, pc=0.019). These association analyses were replicated using logistic regression with sex and age as covariates. Autoantibodies against IFN-α associated with COVID-19 severity were detected in 26% of hospitalized patients. HLA-C*08:02 was more frequent in patients with IFN-α autoantibodies than those without, and KIR3DL1*01502 was only present in patients lacking IFN-α antibodies. We intend to expand this analysis in Greek and Spanish COVID-19 cohorts. These findings show that KIR and HLA polymorphism may play important roles in determining the clinical outcome following SARS-CoV-2 infection, by influencing the course both of innate and adaptive immunity.

ABSTRACT. NK cells can target cells lacking HLA class I molecules. This is regulated by inhibitory and activating signals from KIR and NKG2 receptors which can interact with highly polymorphic HLA molecules. HLA and KIR are inherited independently. To ensure self-tolerance, NK cells undergo a process termed education. Education makes the NK cell more receptive to activating stimuli. Uneducated NK cells are largely hyporesponsive. Human cytomegalovirus (HCMV) has been shown to impact the NK cell receptor repertoire and could therefore affect NK cell education. In reproduction, the role of education is still unknown. In early pregnancy, most of the leukocytes are NK cells and they regulate placentation and trophoblast invasion. We hypothesized that NK cell education is altered in women with recurrent pregnancy loss (RPL) compared to women with successful pregnancies (controls). Additionally, the effect of CMV infection on NK cell education was investigated. Protein expression of KIRs and NKG2A was assessed by flow cytometry. HLA typing was done using PCR-SSO (Luminex). HLA alleles were divided into KIR recognition epitopes. KIR expression was similar between RPL (n=41) and controls (n=16) with KIR2DL2/3 having the highest percentage positive cells followed by KIR2DL1 and KIR3DL1. No differences were found in the percentage of NKG2A educated NK cells between RPL (49.98%) and controls (55.50%) (p=0.33) or in the percentage of single KIR educated NK cells (RPL: 10.07%, C: 13.27, p=0.74). Only very few NK cells were double or triple KIR educated in both the RPL and the control group. Furthermore, CMV infection did not influence the percentage of educated NK cells in women with RPL. In conclusion, (1) The percentage of educated NK cells is not associated with RPL. Our results also suggest that (2) CMV infection does not impact the percentage educated NK cells in these women.

ABSTRACT. COVID-19 has aspects on its pathogenesis that still need elucidation, and an analysis of clinical and immunogenetic factors in each cohort of patients deems paramount to understand how genetic variability can explain the multiple clinical spectrum seen in patients infected with SARS-CoV-2. The aim of this study was to correlate the KIR polymorphism/HLA-I ligands’ interactions from patients and healthy subjects with either the susceptibility or severity to COVID-19 Genotyping of HLA-A, B, C and KIR genes were carried out from 459 symptomatic as well as 667 non-infected Spanish Caucasian individuals using Lifecodes HLA-SSO and KIR-SSO kits (Immucor™, USA) and analysed in the Luminex® in this unicentre case-control study performed at the University Hospital of Salamanca, Spain. Comparative KIR gene analysis showed that KIR2DS4 was significantly more representative in healthy versus infected individuals. When comparing subgroups of infected patients, KIR2DS3 had a higher frequency in those who progressed to a more severity disease and yet with higher mortality rate. Three functional combinations were significant on univariate analysis: KIR2DL2/C1, KIR2DS2/C1, and KIR2DS3/C1. However, in the multivariate analysis, only the KIR2DL2/C1 interaction remained significant [OR=15.2 (95% CI 1.5-147), p= 0.0189]. Compared to the solo-clinical characteristics predictive model, that included well-known comorbidity variables such as hypertension, age, sex, diabetes, C-reactive protein, dyslipidemia, smoking, ferritin, and fibrinogen, the clinical-and-KIR-based model showed a better ability to discriminate between severe and non-severe patients with higher sensitivity and specificity Our results support a fundamental role of KIR/ligand interaction in the clinical course of COVID-19. Since the KIR2DL2 gene has a high frequency in Spain (60%), the analysis of the KIR2DL2/C1 in symptomatic patients who require hospitalization could be helpful to better determine their prognosis

ABSTRACT. Killer-cell immunoglobulin-like receptor (KIR) genes regulate NK cell activity and thereby influence the immune response to infectious diseases and malignant cells. However, the underlying molecular interactions resemble a giant and largely unsolved puzzle. Variability in KIR gene repertoire is derived from KIR region haplotype diversity with varying KIR gene composition and high allelic diversity of each of the 17 KIR genes. Since we started allele-level KIR genotyping in 2018, we identified hundreds of novel KIR alleles and successfully characterized and submitted 478 novel alleles for KIR2DL1, -2DL4, -2DL5, -2DS1, -2DS2, -2DS3, -2DS5, -3DL3 and -3DP1 to the international database IPD-KIR in 2020. For these samples, we used our well-established dual redundant reference sequencing (DR2S) protocol of short-read Illumina- and long-read PacBio sequencing to obtain fully phased sequences in reference grade quality. However, recent advances in nanopore sequencing technology offered the opportunity to use an alternative platform for KIR gene characterization. As proof of concept, we first confirmed more than 30 PacBio/Illumina-characterized KIR2DL1 alleles using the novel ONT/Illumina-based approach with perfect concordance. Subsequently, we applied the ONT/Illumina process for the characterization of novel KIR2DL1, -2DS1, -3DS1 and -2DP1 alleles. By now, we finished reference grade sequence characterization of over 600 novel alleles. Among them are 52 novel KIR3DS1 alleles, which multiply the amount of full-length KIR3DS1 alleles in the newest IPD-KIR release (2.12, December 2022). These novel KIR sequences provide additional insight into KIR gene diversity and will support KIR genotyping in the future. Furthermore, since ONT requires far less capital investment compared with PacBio, these results lower the entry threshold for applying reference grade sequencing for the characterization of novel KIR alleles.

ABSTRACT. Breast cancer (BC) is the most common malignancy among women, with tumors varying in aggressiveness and histological and immunohistochemical subtypes. Natural killer cells (NK) are lymphocytes critical for recognizing and killing neoplastic cells and express the killer-cell immunoglobulin-like receptor (KIR) molecules on the cell surface, which modulate NK cell responses by recognizing human leukocyte antigen (HLA). Despite the well-established role of KIR-HLA in diseases, the variation of these genes in high resolution has never been analyzed in BC. Here, we analyzed the full extent of the KIR polymorphism in the context of their HLA ligands. We applied a State-of-the-Art next-generation sequencing method and bioinformatics pipeline to evaluate a cohort of 550 patients with BC and 747 controls. The significance threshold was set as 0.05 after Bonferroni corrections. The presence of three copies of KIR2DL5 was associated with protection (OR = 0.21, pcorr = 0.027), while the presence of three copies of KIR3DL1S1 was associated with increased risk to BC (OR=2.44, pcorr = 0.009). We observed that the duplication of KIR3DP1~KIR2DL4~KIR3DS1 significantly increases the risk of sporadic BC (OR = 2.42, 95%CI 1.30-4.61 p = 0.007). Moreover, the pair KIR2DL3*001+HLA-C1 ligand was significantly reduced in patients with the aggressive non-luminal HER2+ compared with all other subtypes (OR = 0.23, pcorr = 0.016), in addition to a strong association of KIR2DS1*002+HLA-C2 in patients with aggressive triple-negative tumors compared to other subtypes (OR = 3.34, pcorr < 1.4 x10-4). Finally, we observed a protective effect of the pair KIR2DL1*001+HLA-C2, significantly increased in patients with less severe BC grade I compared with the most severe grades II+III (OR = 0.25, pcorr = 0.038). In summary, we have provided solid evidence that the variation of KIR genes is implicated in BC and associated with its aggressive subtypes, pointing them as candidates for immunotherapies.

ABSTRACT. Multiple myeloma (MM) is a hematological malignancy caused by the clonal expansion of malignant plasma cells in the bone marrow. Myeloma cells are susceptible to killing by natural killer (NK) cells, but NK cells fail to control disease progression, suggesting immunosuppression in the MM setting. The strength of NK cell effector function is regulated by interaction between KIRs and self HLA class I molecules, during a process called ‘education’. The HLA and KIR repertoire is extremely diverse, thus this study aimed to characterize potential variances in genotypic composition of HLA Class I epitopes and KIRs between patients with MM and healthy controls. For this, 16 KIR genes and HLA group C1, C2, and Bw4 epitopes were analyzed in 172 MM patients and 196 healthy controls using SSO and Luminex-analysis. Compared to healthy controls, we did not observe specific KIR genes or genotypes, or HLA epitopes with higher prevalence among MM patients. The presence of all three HLA epitopes was not related to the occurrence of MM disease (30.4% in MM vs. 32.5% in controls; chi-square, p=0.672). However, MM patients were more likely to be C1-/C2+/Bw4+ (chi-square, p=0.05, OR 2.0). In line with this, the occurrence of both HLA-Bw4 and KIR3DL1 was higher in MM patients (chi-square, p=0.023, OR 1.6). Furthermore, MM patients were less likely to only have a single HLA-KIR pair (chi-square, p=0.02, OR 0.6). In contrast, MM patients were more likely to have two HLA-KIR pairs (chi-square, p=0.003, OR 2.0), specifically HLA-C2/KIR2DL1 and HLA-Bw4/KIR3DL1 (chi-square, p=0.008, OR 2.6). In conclusion, our results reveal an HLA epitope combination that is associated with the occurrence of MM. There were no specific KIR genotypes associated with MM. Based on HLA-epitope triple-positivity, haplo-SCT combined with KIR-ligand mismatching is feasible for MM patients. The role of having different HLA-epitopes, HLA-KIR pairs, or different amounts of HLA-KIR pairs should be further investigated.

ABSTRACT. Kidney transplantation remains the best therapeutic option against end-stage kidney disease. However, current immunosuppression poorly influences chronic rejection and the graft half-life is still around 10 years. Identification of biomarkers to anticipate risk of allograft injury may conduct to the design of innovative therapeutics adapted to the patient’s own risks (i.e. personalized medicine) to prevent graft failure and favor patients care. The KiT-GENIE genetic cohort is composed of 1,504 complete donor/recipient kidney transplant pairs from European ancestry from Nantes since 2000. We imputed the HLA alleles from SNPs genotyping using HIBAG to obtain high resolution typing, we used Easy-HLA (hla.univ-nantes.fr) to determine KIR ligand groups from HLA alleles. In parallel, we typed 247 recipients for KIR genes content. We performed an HLA genetic association study for time-to-kidney graft failure using Cox regression models. We added several covariates in the model such as recipient/donor age, recipient/donor sex, ancestry components from genomic analysis, epitopic mismatch score. We defined a Bonferroni multiple testing threshold of significance at P<8x10-4. We identified HLA-B*40:01 from the donor as significantly associated with time to graft failure (P=5.9x10-4, HR 1.8, frequency=4.3%, dominant genetic model). Moreover, we identified the KIRD2L2 recipient + HLA-C2 donor combination as significantly associated with time to graft failure (P=2.2x10-4, HR 3.0, frequency=19%, dominant genetic model). Overall, these results report important new findings on kidney transplantation graft failure: the first reported significant association with an HLA allele from the donor as well as a new KIR/HLA association. These very promising results need further investigation with a larger number of patients.

ABSTRACT. To evaluate possibility of predicting acute renal allograft rejection the present study evaluated lncRNA expression at various time points pre and post transplant. Study enrolment included ESRD patients who underwent renal transplantation and were subgrouped as Group 1-subjects with well functioning grafts and Group 2- subjects that experienced acute allograft rejection. RNA seq was performed on a selected cohort using samples collected at pre transplant, at the time of rejection, and post anti-rejection therapy time points. Differentially expressed (DE) RNAs and related biological pathways were identified from the RNA Seq data. Database for Annotation, Visualization and Integrated Discovery (DAVID) and Gene Ontology (GO) helped analyse the major function(s) of DE genes. Using the 2-side Fisher exact test, we determined the statistical significance of enrichment. In the present study, we compared lncRNAs expression among two data sets obtained from RNA obtained from i) set 1: Well functioning grafts vs those experiencing rejection and ii) set II: those experiencing rejection vs the same cases post anti-rejection therapy. Over 2000 RNAs were upregulated and >850 were downregulated in the set 1 analysis where as in set 2, lower number of differentially expressed RNA were identified (~360 upregulated and ~ 150 downregulated). These GO terms were associated with leukocyte activation, degranulation and regulation of cytokine secretion. Further, the involvement of pathways like cytokine-cytokine receptor interaction and TNF signaling was suggested on using Kyoto Encyclopedia of Genes and Genomes pathway (KEGG) analysis. qPCR was performed on samples obtained from an independent cohort of subjects experiencing acute rejection and this helped in validation of lncRNA identified from discovery cohort. The linkage of identified and validated lncRNAs with immune pathways indicates few of these might play key role in leading to allograft rejection.

ABSTRACT. Allograft rejection remains a major cause of kidney function loss after transplantation. The histological analysis of graft biopsy is the gold standard to diagnose rejection and guide immunosuppressive treatment. However, histology can yield inconclusive results, leading to a delay in optimal treatment. Moreover, once detected in histological analysis, injury is often irreversible. Therefore, we compared gene expression profiles of biopsies from patients without versus with early and late rejection to identify molecular changes associated with progression to rejection before histological changes occur. RNA was isolated from frozen biopsies, all taken at 3 months after transplantation, including samples from patients without rejection in the first year (n=6), with biopsy proven rejection at 3 months (n=4, early rejection group) and with no rejection at 3 months, but biopsy proven rejection at 12 months, (n=4, late rejection group). Gene expression analysis was performed with Nanostring’s Human Organ Transplant panel. Distinct gene expression profiles were observed between control and early rejection groups (138 genes differentially expressed). Importantly, gene expression, 103 genes were differently expressed at 3 months in the late rejection group as compared to control group. Interestingly, 83 of these genes were also differentially expressed between early rejection group and control and 3 month expression pattern of the late rejection group closely resembled that of the early rejection group. This pattern includes genes involved in oxidative stress, inflammasome and NLR signalling such as NOX4, FAS, JAK2, NFKB1, BCL2 and BMP7 (p-Adj < 1,641e-02). Our results demonstrate that a distinct gene expression profile precedes histological evidence of injury. This can potentially be used to predict rejection and adjust treatment to prevent it.

ABSTRACT. Liver transplantation is the main available treatment for acute and chronic hepatic failure. Since the transplant outcome is still limited and influenced by the ischemia-reperfusion injury grade of the graft, hepatic function is carefully monitored through the analysis of multiple hematological parameters. However, a single, specific and early marker of post-transplant liver damage has not been identified yet. Donor-derived cell-free DNA (dd-cfDNA), which is a potential biomarker of rejection and graft tissue damage in heart, lung and kidney transplanted patients, could represent this missing marker. The aim of this study was to recruit a cohort of adult liver transplant recipients to validate dd-cfDNA quantification as a powerful and non-invasive method for early identification of graft failure. We included 40 patients (male sex 80.0%; mean age 58,01±9,97 years). Blood samples were collected before transplantation, daily during the first 7 days post-transplant, then at day 28 and at months 3 and 6. Dd-cfDNA quantification was obtained from 418 samples using droplet digital PCR, targeting a list of 8 alleles of HLA-DRB1 gene. Results were correlated to hematological measurements, clinical exams and liver biopsies collected in case of suspicion of rejection. Dd-cfDNA levels showed a common trend in the entire cohort with a peak at the time of reperfusion of the organ and a progressive decrease of its levels over time, reaching percentages similar to those physiologically secreted by the liver after 7 days. The percentage of dd-cfDNA reflected the grade of ischemia-reperfusion injury and significantly correlated with hepatocytolysis indices used in clinical practice (p<0.0001). In the long term, it increased in patients with hepatic complications, such as acute rejection and increase of hepatocytolysis. Our results demonstrated the value of dd-cfDNA as a reliable biomarker of hepatic injury reflecting the ischemia-reperfusion damage in liver transplantation.

ABSTRACT. Rejection represents one of the main causes of graft failure and loss after heart transplantation. It is generally monitored by histopathological evaluations of endomyocardial biopsies (EMB), that are limited by interobserver variability, poor sensitivity, and invasiveness, particularly in pediatric patients. In addition, biopsies are often performed for clinical reasons and therefore when graft injury is possibly already present. Donor derived cell-free DNA (dd-cfDNA) is a potential biomarker of rejection and graft injury in solid organ transplantations. The study aims to assess the role of dd-cfDNA in rejection monitoring in a pediatric cohort of heart transplant recipients. We enrolled a pilot cohort of 21 pediatric heart recipients with an average age at transplant of 4.7±4.3 years and a transplant survival of 6.9±5.2 years. cfDNA was obtained from 38 plasma samples (1.8 samples/patient) and dd-cfDNA quantification was performed exploiting a ddPCR assay based on HLA-DRB1 polymorphisms that discriminate between donor and recipient DNA. Results were correlated with clinical and bioptic data collected contextually with blood samples. dd-cfDNA quantification was significantly higher in relation to graft injury events (acute or chronic) compared to stable patients (p=0.0080). We obtained a significant correlation with the EMB rejection grade (p=0.0146) and the pro-BNP value (p=0.0135). To assess the performances of the test, we performed a ROC analysis, obtaining an AUC of 0.74. With a positive cut-off of 0.13%, the test had 80.95% sensitivity and 64.71% specificity. Our preliminary results demonstrated that dd-cfDNA is a valid and reliable biomarker of both acute and chronic rejection in pediatric heart transplant recipients, and its quantification can help in the post-transplant management of patients with the aim to avoid unnecessary and invasive EMB when there is no suspicion of rejection.

ABSTRACT. Context: Donor-recipient (DR) mismatches in HLA genes have been associated with a poorer kidney graft survival. However, HLA mismatches alone do not explain long-term graft function decline. We ran a genome-wide survival study (GWSS) on a large monocentric cohort of kidney transplant donors in order to characterize non-HLA genetic factors associated with HLA and non-HLA antibody mediated rejection (ABMR), the main transplant complication causing graft loss. Methods: The KiT-GENIE French genetic cohort comprises 1647 donors of European ancestry for kidney transplants performed in Nantes since 2000. After genotyping and polymorphism (SNP) imputation with the TopMed reference panel, we performed a genetic association study on >8.2M SNPs for biopsy proven ABMR. Cox proportional hazards models notably adjusted for HLA mismatches were fit to assess the association between donors’ genotypes and biopsy proven ABMR. Statistical tests were computed using a saddle point approximation with the SPACox R package in order to limit the number of false positives. P-values below the genome-wide multiple testing Bonferroni correction threshold (5x10-8) were considered significant. Results: We identified a statistically significant association in an intergenic region of chromosome 6 (p= 3.5x10-8, HR=2.9) outside of the HLA complex, near a gene implied in vasculogenesis. Conclusion: Our genome wide analysis in a large homogeneous monocentric cohort revealed a non-HLA variant in donors associated with ABMR. Further validation in external cohorts and non-European individuals will be performed.

ABSTRACT. Kidney transplantation is the preferred treatment option for many patients with end-stage kidney disease. The outcome of such a treatment is most optimal when recipient and donor are human leukocyte antigen (HLA) matched. Through the activation of alloimmune T and B cells, HLA mismatches may lead to T-cell- or antibody-mediated rejection. During antibody-mediated rejection, the indirect pathway of allorecognition plays an important role. However, the role of the indirect pathway of allorecognition in acute T cell-mediated rejection (aTCMR) is not well defined. In this study, we aimed to evaluate the role of the indirect pathway of allorecognition in aTCMR using the Predicted Indirectly ReCognizable HLA Epitopes algorithm (PIRCHE-II), which is an algorithm that quantifies the amount of theoretical T-cell epitopes available for indirect allorecognition. As a measure for the potential indirect CD4+ T-cell alloreactivity, the PIRCHE-II score was calculated for 688 donor kidney-recipient combinations. A diagnosis of aTCMR was made in 182 cases; 121 cases of tubulo-interstitial rejection cases and 61 cases of vascular aTCMR. A strong association between the PIRCHE-II score and the incidence of first-time aTCMR was observed, in particular vascular rejection. This association was found mainly for the peptides derived from donor HLA-DR/DQ (PIRCHE-II DR/DQ): at one year after transplantation, the cumulative percentage of recipients with a vascular rejection was 12.7%, 8.6% and 2.1% within respectively the high, medium and low tertile of the PIRCHE-II DR/DQ score (p<0.001). In a multivariate regression analysis this association remained significant (p<0.001 for PIRCHE-II DR/DQ tertiles). In conclusion, indirect antigen presentation of donor HLA-peptides may significantly contribute to the risk for acute vascular rejection. This finding increases our current understanding of the pathogenesis of aTCMR and may contribute to risk stratification following kidney transplantation.

ABSTRACT. Organ transplant recipients show weaker immune responses to vaccines than immunocompetent individuals, which may be related to the repertoire of HLA-bound vaccine antigens presented to T lymphocytes. The HLA evolutionary divergence (HED) metric, which quantifies pairwise allele divergence at each HLA locus, provides a primary measure of the breadth of the immunopeptidome. We recently showed that high class I HED of the donor is a strong and independent driver of allograft rejection in a large cohort of liver transplant recipients. Here, in the same cohort, we explored the relation between HED, the size of the predicted immunopeptidome derived from vaccine antigens, and the quality of vaccine responses. We analyzed humoral response to the SARS-CoV-2 BNT162b2 vaccine (n= 310 patients; undetectable anti-spike IgG titers considered as no response, ≤250 BAU/mL as moderate and >250 BAU/mL as strong response) and Hepatitis B virus (HBV) vaccine (n= 424 patients; anti-HBs IgG <10 mIU/mL considered as no response, 10-100 mIU/mL as moderate and ≥100 mIU/mL as strong response). HED at HLA-A, -B, -C, -DRB1, -DQA1 and-DQB1 loci were measured using the Grantham distance. NetMHCIIpan-4.0 was used to predict the binding to HLA-DQ molecules of all possible 15mer peptides derived from the Spike and HBS sequences. For each vaccine, HED at the DQB1 locus, but not at the other loci, was significantly higher in responders than in non-responders (p= 0.0003), independent of response-associated covariates (age, time since transplant, immunosuppression). Moreover, for both vaccines, there was a strong relationship between DQB1 HED, the diversity of the immunopeptidome and the quality of the vaccine response. In conclusion, DQB1 HED is a critical determinant of humoral response to vaccines in liver transplant recipients. This metric could guide the design of future vaccines as it predicts the magnitude of the repertoire of vaccine-derived peptides presented to CD4 helper T cells.