Comparative analysis in cynomolgus macaque identifies a novel human MHC locus controlling platelet blood counts independently of BAK1


Correspondence: Antoine Blancher, Laboratoire d'Immunogénétique Moléculaire, EA 3034, Faculté de Médecine Purpan, Université Paul Sabatier, Toulouse 3, IFR150 (INSERM), CHU de Toulouse, 1 avenue Jean Poulhes, TSA 50032, 31059 Toulouse Cedex 9, France.

Tel.: +33 5 61 32 34 32 and +33 05 61 32 34 34;

fax: +33 5 61 32 34 30.


The platelet blood count is a critical parameter in clinical assessment of hemostasis. It fluctuates little in a given healthy individual but there are important inter-individual variations, which partly depend on genetic factors [1-4]. One of these lies at the centromeric end of the major histocompatibility complex (MHC) and involves the BAK1 gene, which encodes a strongly pro-apoptotic protein known to control platelet lifespan [5].

The cynomolgus macaque (Macaca fascicularis) is a non-human primate phylogenetically close to humans that has received growing interest in experimental medicine as well as in comparative mapping of complex phenotypic traits [6]. We sought an influence of MHC polymorphism on platelet blood counts in 179 unrelated young cynomolgus monkeys (2.5–6 years old), born in captivity (F1) from breeders originated from Sicombrec in the Philippines. We genotyped 17 markers spanning the MHC, including 16 microsatellites and the DRB locus [6, 7]. As shown in Fig. 1(A), there was a significant association between two consecutive microsatellites, D6S2788 and D6S2782, and platelet counts (P = 1.1 × 10−4 and 1.9 × 10−5). A haplotype analysis using a sliding window of two markers across the MHC revealed five consecutive windows significantly associated with platelet blood counts in the same region (Fig. 1A). Allele recurrence in the successive windows suggested that only two haplotypes encompassing six markers from MICA to D6S2669, that map more than 2 Mbases upstream of BAK1, were involved. This was supported by a strong correlation between several of the associated alleles (data not shown). We identified the carriers of these two haplotypes, here referred to as H1 and H2 (frequencies 5.31% and 5.03%), associated with low and high platelet counts, respectively. There was no H1/H2 heterozygous subject. As shown in Fig. 1(B), effects of H1 and H2 were additive (P = 3 × 10−11). They explained 24% of the overall variance of the platelet counts.

Figure 1.

Influence of MHC on platelet blood counts. (A) Association of individual MHC markers (open circles) and of two consecutive marker haplotypes (dark squares) with the log of platelet counts of 179 macaques was tested with UNPHASED (v3.013). Markers were located on the MHC genetic macaque map. The horizontal dashed line indicates the Bonferroni-adjusted significance threshold (2.9 × 10−3). The most significantly associated alleles, which form the H1 and H2 haplotypes, are listed underneath the MHC map. (B) Additive effect of H1 and H2 haplotypes on platelet counts in macaques, tested by linear regression, adjusting for age and sex, using the lm function of R. X designates haplotypes other than H1 or H2. Number of subjects per group is given in brackets. (C–E) Association of human platelet counts with MHC between chromosomal positions 29.5 and 34 Mb (NCBI build 37). (C) Association of individual SNPs considering a linear model, assuming allele additive effects and adjusting for age, sex and cohort origin. The horizontal dashed line indicates the Bonferroni-adjusted threshold P (2.28 × 10−5) for 2189 SNPs tested. Dark symbols indicate significantly associated SNPs and shaded symbols non-associated SNPs. (D) Association after controlling for the top MHC SNP, rs5745568, in BAK1. (E) Association after controlling additionally for rs13437088 in MICA.

Examination of our Japanese cohort containing 14 967 subjects from the BioBank Japan Project revealed several SNPs potentially associated with platelet count in a region homologous to that we identified in the macaque model but also in the class II region (Fig. 1C), as also recently reported in Caucasians [3]. To rule out a long-range effect of BAK1 through linkage disequilibrium, we re-analyzed the data using a linear model after conditioning on rs5745568, the top associated SNP in BAK1 (Fig. 1D). The most significant association corresponded to a cluster of tightly associated SNPs in the upstream region of MICA, towards the centromeric end of the MHC class I region. The minor allele of the best associated SNP, rs13437088, had a frequency of 0.29 and was associated with a decrease of platelet counts (β = −3.75 ± 0.78, P = 7.53 × 10−6). Of note, association signals at HLA-B and in the class II region were removed, indicating their dependency on BAK1.

A further round of analyses, conditioned on both rs5745568 and rs13437088, identified an additional SNP, rs3793125, in the class III region, between the LY6G6F and LY6G6E genes (Fig. 1E). The minor allele of this SNP is rare (F = 0.011), specific to Asians, and has a strong effect on platelet counts (β = 11.4 ± 3.4, P = 0.0008). It is flanked by two other SNPs, rs2273612 and rs11967206, which are in perfect LD with it and show the same effect. Although the SNP does not reach the significance level adjusted for multiple testing, it is worth reporting because of the potential functional interest of the LY6G6 gene family (see below), in addition to the parallel association at D6S2782 in macaques.

Our study thus highlights an effect of genes in the proximal class I and class III regions on inter-individual variation of platelet blood counts, both in the macaque model and in humans. What genes precisely are involved and how they interrelate remain to be determined and will require additional investigations, linking genetic and functional information. As for the class I genes, platelets express a large amount of them at their membrane. It was proposed that these MHC class I molecules do not activate cytotoxic CD8 T cells and are involved in longer survival of allografts following transfusion [8]. Moreover, very interestingly, a recent study associated an allele of a microsatellite marker in the MICA gene with susceptibility to idiopathic thrombopenic purpura [9]. In the class III region, several members of the LY6G gene cluster are expressed in platelets. Among them, LY6G6B encodes a member of the immunoglobulin (Ig) superfamily that is specifically transcribed in megakaryocytes and platelets and is a negative regulator of platelet function [10, 11]. Moreover, Ly6g6b-KO mice present a severe impairment of megakaryocyte development and platelet function [12].

Altogether, our work demonstrates the role of MHC loci other than BAK1 in influencing platelets blood counts. It also underscores the value of non-human primates in investigating important aspects of blood platelet physiology.


This work was supported by The Ministère Français de la Recherche (EA 3034). The BioBank Japan Project was supported by the Ministry of Education, Culture, Sports, Science and Technology of the Japanese Government. We thank E. André (Bioprim, Baziège France) and M. Bigaud (Novartis, Bâle) for help in obtaining macaque blood samples, M. Blancher-Sardou for macaque blood counts, P. Tisseyre and M. Dutaur for DRB genotyping, and B. Atlan, A. Dauba, S. Despiau-Schiavinato and Hébrard for microsatellite genotyping.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.