Soluble HLA-DR serum levels are associated with smoking but not with acute coronary syndrome
J. Tolva (Helsinki, FI)
Elevated soluble HLA-DR (sHLA-DR) serum levels have been reported in HLA class II -associated inflammatory disorders. We have previously shown that the HLA-DRB1*01, HLA class II allele, may predispose an individual to acute coronary syndrome (ACS). sHLA-DR levels were measured in 477 Finnish ACS patients, 475 matched controls and 149 population cohort samples. Patients and controls were HLA-DRB1*01 genotyped by genomic real-time PCR and the population cohort was HLA-typed by SSP. sHLA-DR levels were measured by a sandwich enzyme-linked immunosorbent assay using commercial purified HLA-DRB1*01:01 to generate the standard curve. Logistic regression and ordinal logistic regression analyses were performed to evaluate the determinants of sHLA-DR levels. ACS patients had lower sHLA-DR serum levels than controls, but the difference became not significant after adjusting for smoking. However, current smoking was inversely associated with sHLA-DR levels both in patients (OR=0.59; 95% CI=0.95-0.99; p=0.016) and in controls (OR=0.36; 95% CI=0.56–0.86; p=0.000010). In addition, HLA-DRB1*01 positivity was associated with sHLA-DR levels in controls (OR=0.60; 95% CI=0.40-0.88; p=0.010). HLA-DRB1*04 and HLA-DRB1*03 were associated with sHLA-DR levels in the population cohort (OR=2.28; 95% CI=1.12-4.65, p=0.024 and OR=0.27; 95% CI=0.09-0.75, p= 0.012, respectively). The results indicate for the first time that lower sHLA-DR levels are associated with smoking, but not with ACS. This is an important finding because to our knowledge, the previous studies of sHLA-DR have not accounted for the possible confounding effects of smoking on sHLA-DR levels. Further studies are required to establish these novel results and explore the mechanisms behind the observed associations.
Recombination rates in the HLA region: Are mothers less stable than fathers?
W. Helmberg (Graz, AT)
Identification of crossover events can be by either sperm analysis or by genotyping of families with >= 2 children. We investigated high-resolution HLA genotyping data from the Type 1 Diabetes Genetics Consortium (T1DGC), which provided HLA genotypes of 2349 complete families consisting of mother, father and >= 2 children. Both HLA class I (A, B, and C) and class II (DRB1, DQA1, DQB1, DPA1, and DPB1) loci were genotyped to two-field resolution. Haplotypes were generated using identity-by-descent and assigned to the parents and children. Each locus within each haplotype was investigated based on the number of unambiguous assignments within a family.
If the haplotype assignment of one locus was based on only one child, or a parent was homzygote for a locus, this assignment was neglected, as it did not contribute information for a crossover event. Based on the children of these families, 12,844 HLA regions (maternal and paternal) were investigated for crossover events. Due to the incomplete haplotypes described before, 2520 had a missing assignment of HLA-A, 658 were missing HLA-A to HLA-C, 736 had no class I assignment, and 833 had no DQ-DP assignment. These missing assignments reduced the overall length of the MHC regions to be investigated for crossing over. HLA-A lies approximately 1.4 Mb from HLA-B and HLA-C, which lie about 1.2 Mb away from DRB1, which lies about 0.5 Mb from HLA-DPB1. 8097 chromosomes were informative for the entire distance from from A to DP. We detected 127 crossover events in the HLA region of maternal chromosomes and 74 in paternal chromosomes. Three individuals showed crossing over events on both chromosomes. The most frequent crossing over locations could be seen on the maternal MHC region in: DQB1-DPA1 (45), A-C(30), B-DRB1(28), on the paternal cases: A-C (23), DQB1-DPA1(21), B-DRB1(18). The observed crossing over rate for this cohort is, thus, at least ~3,7% for maternal and ~2,2% for paternal HLA region.
MHC Class I of olive baboons (Papio anubis): extraordinary diversity easily unraveled by long-read SMRT sequencing
G. Doxiadis (Rijswijk, NL)
The olive baboon (Papio anubis), an Old World monkey (OWM) species, is frequently used as a model animal in various fields of biomedical research. Most of the classical MHC class I genes of OWM exhibit copy number variation (CNV) next to allelic polymorphism, and may also display transcription level differences as well. Preliminary sequencing results on the MHC class I A and B of the olive baboon (Paan-A and -B) already suggested high CNV of class I genes in this species. These results were, however, hampered by insufficient read lengths. Therefore, full-length cDNA class I typing of 24 partially related olive baboons was performed by Pacific Biosciences' SMRT sequencing and subsequent data analysis with Geneious software. Using generic primers, we were able to determine 19 A and 80 B transcripts, half of which were newly defined. Based on segregation analysis of MHC alleles within related animals and together with known DRB typings, we were able to define 13 different Paan-A,-B, -DRB haplotypes and additionally three B haplotypes. A Paan class I haplotype appears to consist of one or two A and three to seven B transcripts. On one haplotype we could detect a transcribed pseudogene and a splice variant in addition to seven bona fide Paan-B alleles, a phenomenon which is rarely observed in macaques. Also in contrast to macaques, phylogenetic analyses of the A alleles appear not to allow their allocation to different loci which suggest homogenization. Furthermore, transcription level differences can rarely be observed for Paan-A alleles but only for Paan-B alleles. A further peculiarity seems to be the presence of B*02 variants on nearly all haplotypes. Thus, the genetic make-up of the olive baboons’ MHC seems even to exceed the complexity of other thoroughly-studied Old World monkey species as macaques.
The orthologues of HLA class II genes display more variability in macaques.
A. de Vos-Rouweler (Rijswijk, NL)
The human major histocompatibility complex (MHC) region encodes three types of class II molecules designated HLA-DR, -DQ and -DP. The HLA-DQ and -DP gene regions comprise both a duplicated tandem of A and B genes, whereas in macaques only one set of genes is present per region. In humans, the B genes of the DR region display copy number variation, which is even more prominent in macaques. The HLA-DRA gene, in contrast, is monomorphic. A substantial sequencing project on the DQA/DQB and DPA/DPB as well as DRA genes in various macaque populations resulted in the detection of more than 300 previously unreported full-length alleles. As shown by phylogenetic analyses, humans and macaques share trans-species lineages for the DQ genes, namely DQA1*01 and DQB1*06, respectively, whereas the DPA1 and DPB1 lineages in macaques appear to be species-specific. Amino acid variability plot analyses revealed that the DQ genes and DPB1 display more allelic variation in macaques than is encountered in humans. Additionally, DPA1 and DRA display high or moderate levels of polymorphism in macaques, which is in contrast to the nearly invariant HLA-DPA1 and -DRA genes. Moreover, the polymorphism of DPA1 is not restricted to exon 2, the part of the gene that encodes the peptide-binding region of the class II protein. Furthermore, the DRA alleles of macaques can be divided into two lineages, one of which, DRA*02, is virtually invariant as in humans and conserved in the macaques species, whereas the DRA*01 lineage displays all variations. Moreover, the numbers of different amino acids at certain positions in the encoded proteins are higher in macaques than in humans. This phenomenon is remarkably prominent at the contact positions of the peptide binding sites of the deduced macaque DPβ-chains. These differences in the MHC class II DP regions of macaques and humans suggest separate evolutionary mechanisms in the generation of diversity.
Large scale whole gene sequencing provides new insights into the evolution and function of HLA class I introns and exons
N. Cereb (Ossining, New York, US)
We have been sequencing whole gene HLA class I on a large scale at Histogenetics on PacBio RS II® platform since the beginning of July 2016. To date we have sequenced 150,000 samples for whole gene class I. We observed 1422, 1489 and 1443 unique sequences in HLA-A, -B and -C, respectively. We compared inter-allelic variation for each HLA class I gene (HLA-A, -B, -C) for 106,377 that were sequenced as of December 31st 2016. Here we present a summary of our findings for HLA-A alleles. Mean values of non-synonymous and synonymous divergence (dN and dS, respectively) for all codons for all pairwise comparisons among alleles were determined using SNPGenie, an implementation of the Nei-Gojobori method (Nei and Gojobori, 1986) for use with next-generation sequencing data (Nelson and Hughes, 2015; Nelson et al., 2015). Overall d (among unique alleles) for introns is 0.0236, as compared to overall dS for exons at 0.0392. Thus, neutral diversity among unique alleles is approximately 66% higher in exons than introns. This is consistent with the hypothesis that diversity at synonymous coding sites has been elevated as a result of linkage to non-synonymous sites, which experience over-dominant (positive balancing) selection. In contrast, intronic sites are less closely linked to non-synonymous coding sites than are synonymous coding sites, making them more likely to experience recombinational separation from coding sites and subsequent drift, which has acted to homogenize the sequences over evolutionary time. When we compared dN/dS ratios exons 2 and 3 are the only ones for which dN/dS substantially exceeds 1, consistent with over-dominant selection acting on peptide binding residues (PBR). However, We were surprised to see that, while exon 4 seems to experience strong purifying selection (dN/dS = 0.22), exon 5 is again higher (dN/dS=1.04; the only exon besides 2 and 3 with a ratio above 1). Since exon 5 does not encode the PBR, this ratio suggests either a relaxation of purifying selection or localized over-dominant selection. There have been reports that the transmembrane domains of HLA class I genes could be a determinant in Inhibition of a subset of Natural Killer Cells.
We will present similar analysis for HLA-B and -C and discuss evolutionary pathways and possible functionalities of various regions of class I genes.
Predicting HLA-DPB1 expression marker rs9277534 from exons 2 and 3: linkage analysis of over 37,000 samples
B. Schöne (Dresden, DE)
HLA-DPB1 allele mismatches between patients and unrelated donors increase the risk of acute graft-versus-host-disease (GvHD) after hematopoietic stem cell transplantation. If no DPB1 matched donors are available, the genotype defined by the Single Nucleotide Polymorphism (SNP) rs9277534 can be used to select donors that are well-tolerated despite DPB1 disparity. However, since rs9277534 resides within the 3’ untranslated region (UTR), it is usually not analyzed during routine genotyping. To investigate if the rs9277534 polymorphism can be reliably imputed by standard DPB1 typing, we analyzed 37,168 samples of mostly Caucasian origin for linkage between high-resolution DPB1 typing results and the rs9277534 alleles. DPB1 alleles were genotyped based on exons 2 and 3. The rs9277534 polymorphism was amplified by PCR and the SNP was called using the software FasType (DKMS Life Science Lab). Sequence data were generated on Illumina MiSeq and HiSeq instruments. Samples with rare linkage patterns were confirmed by whole-gene sequencing on PacBio RS II instruments. We confirmed all previously defined linkages between rs9277534 and 18 DPB1 alleles. In addition, we established the linkage for 41 DPB1 alleles at two-field resolution with so far unknown 3’ UTR sequences. Based on these linkages and the high resolution DPB1 genotyping results we were able to predict the rs9277534 genotype for 99.6% of the samples. The predictions conformed to the rs9277534 sequencing results for 99.99% of the 37,028 samples. (The 4 remaining discrepancies were all due to DPB1*34:01 variants.) In contrast, we demonstrate that linkage between DPB1 exon 2 sequences and rs9277534 is ambiguous for certain allele groups. We therefore conclude that DPB1 typing of exons 2 and 3 is sufficient to infer the DPB1 expression marker rs9277534 with very high accuracy in Caucasian samples. This information could be used to select donors with permissive HLA-DPB1 mismatches without screening for rs9277534 in addition.
The use of entropy to describe HLA molecular information
T. Goeury (Geneva 4, CH)
Defined in 1948 by C. Shannon, the term entropy refers to the amount of information encapsulated in a signal. We present here an application of the entropy concept to estimate the molecular information presented at different regions (i.e. exons, introns and UTRs) within and across HLA genes. To that aim, we used the DNA sequences of the IPD-IMGT/HLA Database to compute the entropy of each gene region at all classical loci HLA-A, -B, -C, -DRB1, -DQA1, -DQB1, -DPA1 and -DPB1. We first compared the information provided by full-genomic HLA sequences with those of exomic HLA sequences by computing, for each source of data, the ratio of the observed entropy to its maximum expected value. We show that except for a few lowly informative exons, the genomic data, despite a lower number of alleles currently defined, do not exhibit a recruitment bias compared to the exomic data, i.e., the entropy values are generally similar. We then analysed those patterns across all gene regions of all HLA genes. A striking result is a linear relationship observed between the entropy and the size of the gene regions. Furthermore, we found different behaviours of this relationship i) between coding and non-coding regions, the former appearing to accumulate molecular information per site at a higher rate than the latter, but also ii) between different HLA loci with similar rates of entropy accumulation per site but different entropy levels, DPA1 and DPB1 showing the lowest values and HLA-B the highest. Finally, based on the entropy of exons 2, exon 3 and exon 2 and 3 taken together, we computed the Mutual Information provided by these two gene regions, i.e. we estimated whether they provide redundant information to define the HLA alleles. We show that the Mutual Information not only differs between class I and class II genes, but also between different genes of each class. These results are crucial both for setting up HLA typing strategies and for better understanding the molecular evolution of the whole HLA genomic region.
HLA-G*01:04:01 and HLA-G*01:04:04 are differentially associated with asthma and splicing patterns
J. Di Cristofaro (Marseille cedex 05, FR)
HLA-G, expressed by a restricted set of tissues, including bronchial epithelial cells (HBEC), is a molecule with potent immunomodulatory activities. There are several HLA-G isoforms as a result of RNA alternative splicing. Currently the IPD-IMGT/HLA Database contains 20 2nd field alleles and 53 alleles of HLA-G at maximum resolution. HLA-G*01:04 was associated with lung transplantation impaired long term survival (Di Cristofaro et al. 2015), with recurrent miscarriages (Aldrich et al. 2001, Vargas et al. 2011) and conversely, with protection in acute renal rejection and end stage renal disease (Misra et al. 2014). The aims of the present study were to further gain insight into the predictive value of HLA-G*01:04 in a bronchial inflammatory disease, to analyze these regulatory region haplotypes and to predict in silico alternative splicing signals such as variations in HLA-G*01:04:XX alleles. HLA-G was sequenced by NGS from position -1983 to 3447 in 330 healthy individuals and 590 asthmatic patients, data were analyzed by Xegen Compagny. Variation in splicing sites was analyzed by the Human Splicing Finder program. HLA-G*01:04 was only represented by two alleles: HLA-G*01:04:01 and G*01:04:04, with identical regulatory regions and distinguishable by a synonymous SNP 1827G>A in exon 4. HLA-G*01:04:01 and G*01:04:04 were respectively higher and lower in healthy individuals than in asthmatic patients (p=0.044 and p=0.012). As compared to HLA-G*01:04:01, HLA-G*01:04:04 had 2 binding sites for SF2/ASF broken and 1 site for SRSF5 created. SF2/ASF has been shown to be involved in RNA stability and surveillance as well as translation initiation and enhancement. SRSF5 was shown to significantly increase exon skipping in alternative splicing process (Michlewski et al. 2008; Sebbag-Sznajder et al. 2012). These results support the distinction of the HLA-G*01:04 allele at 8 digits which could explain discrepancies in clinical studies as only HLA-G*01:04:04 seem to be associated with impaired inflammatory control. This may be due to differential alternative splicing affecting protein isoform expression (soluble or membrane-bound) and interaction with HLA-G receptors. These results support our previous study showing that HBEC from asthmatic patients displayed less functional transcriptional HLA-G isoforms than those from healthy donors (Carlini et al. 2017).