No postcopulatory selection against MHC‐homozygous offspring: Evidence from a pedigreed captive rhesus macaque colony

The heterozygosity status of polymorphic elements of the immune system, such as the major histocompatibility complex (MHC), is known to increase the potential to cope with a wider variety of pathogens. Pre‐ and postcopulatory processes may regulate MHC heterozygosity. In a population where mating occurs among individuals that share identical MHC haplotypes, postcopulatory selection may disfavour homozygous offspring or ones with two MHC haplotypes identical to its mother. We tested these ideas by determining the incidence of MHC‐heterozygous and MHC‐homozygous individuals in a pedigreed, partially consanguineous captive rhesus monkey colony. Bayesian statistics showed that when parents share MHC haplotypes, the distribution of MHC‐heterozygous and MHC‐homozygous individuals significantly fitted the expected Mendelian distribution, both for the complete MHC haplotypes, and for MHC class I or II genes separately. Altogether, we found in this captive colony no evidence for postcopulatory selection against MHC‐homozygous individuals. However, the distribution of paternally and maternally inherited MHC haplotypes tended to differ significantly from expected. Individuals with two MHC haplotypes identical to their mother were underrepresented and offspring with MHC haplotypes identical to their father tended to be overrepresented. This suggests that postcopulatory processes affect MHC haplotype combination in offspring, but do not prevent low MHC heterozygosity.

encode for cell surface proteins with different functionalities, MHC I and II, both involved in the binding and presentation of peptides to specialized types of T cells belonging to the immune system (Falk, R€ otzschke, Stevanovi c, Jung, & Rammensee, 2006;Klein, 1986).
MHC I gene products are expressed on virtually all nucleated cells and are involved in generating immune responses to pathogens that cause intracellular infections, such as viruses and mycobacteria. In contrast, MHC II gene products are only expressed on professional antigen-presenting cells such as macrophages and dendritic cells and present peptides, mostly from extracellular origin. The MHC class I and II molecules may control susceptibility/resistance to many autoimmune and infectious diseases (McLaren & Carrington, 2015;Tiwari & Terasaki, 1981). The characteristics of the genes encoding for MHC I and II are their high degree of polymorphism (allelic variation) and their copy number variation (diversity) (Doxiadis, Otting, de Groot, Noort, & Bontrop, 2000;Otting et al., 2005;Robinson, Waller, Fail, & Marsh, 2006). This holds true not only for humans but also for nonhuman primates such as macaques (de Groot et al., 2012). MHC class I and II genes, which are inherited together on one chromosome, are called MHC haplotypes. MHC molecules are codominantly expressed and different MHC allotypes can bind and display a unique gradient of foreign peptides. As a consequence, MHC-heterozygous individuals can present more peptides selected from pathogens to T cells than MHC-homozygous ones and therefore have an improved fitness profile (heterozygous advantage hypothesis: Doherty & Zinkernagel, 1975;Carrington et al., 1999).
The influence of MHC-dependent sexual selection on the maintenance of MHC diversity in natural populations has been described (Jordan & Bruford, 1998;Kamiya et al., 2014;Penn & Potts, 1999).
Several processes may explain how the MHC influences mating behaviour and reproductive success such as precopulatory disassortative mating based on the MHC genotype (Piertney & Oliver, 2006) and postcopulatory MHC-related selective fertilization, implantation and/or selective abortion (Ziegler, Kentenich, & Uchanska-Ziegier, 2005). Precopulatory partner choice and postcopulatory processes may both prevent inbreeding and promote the best combination of polymorphic genes for the offspring. Inbreeding can lead to reduced viability in offspring. This may result from homozygosity of deleterious genes (Charpentier, Widdig, & Alberts, 2007;Leberg & Firmin, 2008;Pusey & Wolf, 1996) and from similarity of genes of the immune system, that is, the major histocompatibility complex (MHC).
Partner choice for unrelated partners is a precopulatory process that will maintain heterozygosity of the relevant MHC genes, as nonrelated individuals are likely to have different MHC haplotypes. Multiple precopulatory mechanisms make mating outbreeding more likely, including differential dispersal patterns or distance of the sexes, extra-pair copulations and partner choice for unrelated individuals (Pusey & Wolf, 1996). Such partner choice may be guided by the genetic characteristics of the partner concerning its MHC haplotypes. Typically, partners with different MHC haplotypes (mandril, Mandrillus sphinx: Setchell, Charpentier, Abbott, Wickings, & Knapp, 2010), a specific MHC or a more heterozygous MHC (fat-tailed dwarf lemur, Cheirogaleus medius: Schwensow, Fietz, Dausmann, & Sommer, 2008) are preferred (meta-analyses: Kamiya et al., 2014;Winternitz, Abbate, Huchard, Havl ıcek, & Garamszegi, 2017). However, when precopulatory mechanisms do not prevent copulating and breeding with related individuals, postcopulatory processes may reduce the proportion of inbred offspring (Tregenza & Wedell, 2000) and MHC-homozygous offspring.
Postcopulatory selection is typically investigated in a polyandrous setting, where a female mates with both related and unrelated males. In these polyandrous matings, females select against inbreeding, as unrelated males father relatively more offspring than related males (decorated field cricket, Gryllodes supplicans: Stockley, 1999; Gouldian finch, Erythrura gouldiae: Pryke, Rollins, & Griffith, 2010). In addition, polyandrous mating may also enhance embryo viability (Simmons, 2005). However, these studies do not address whether postcopulatory mechanisms also operate to reduce the proportion of MHC-homozygous offspring. Such postcopulatory mechanisms may concern selection of MHC-specific sperm by the ovum. There is some evidence for sperm selection in inbred laboratory mouse strains. Although blastocysts are on average more often MHC-homozygous than heterozygous, the proportion of heterozygous blastocysts seems to increase with external circumstances, such as infections (Olsson, Madsen, Ujvari, & Wapstra, 2004;Scofield, Schlumpberger, West, & Weissman, 1982;Wedekind, Chapuisat, Macas, & Rulicke, 1996).
Another possible postcopulatory selection process concerns the MHC disparity or similarity between maternal and foetal MHC that may result in selective implantation or selective abortion of embryos.
Data on the human MHC (human leucocyte antigen: HLA) matching and elevated foetal loss are reported of the Hutterites, U.S.A. (Ober, Hyslop, & Hauck, 1999), an inbred population that is characterized by a limited number of HLA haplotypes and a high natural fertility rate. In Hutterites, foetal loss is more likely to occur in couples sharing an entire HLA haplotype than not. In other human studies, HLA similarity of parents has also been linked to foetal loss (Ober & van der Ven, 1996). Similarly, pig-tailed macaque mothers that share MHC haplotypes with their mating partner have a lower fertility than females mating males with a different MHC haplotype (pig-tailed macaques, Macaca nemestrina: Knapp, Ha, & Sackett, 1996). Such reduced fecundity may (partly) be caused by a lower survival or implantation of MHC-homozygous embryos (review: Simmons, 2005). Surprisingly, in the Hutterite population, the number of surviving HLA homozygous individuals did not differ from expectations.
However, the observed number of heterozygous-compatible individuals, namely individuals that are HLA identical to the mother, was lower than expected (Ober, Hyslop, Elias, Weitkamp, & Hauck, 1998;Ober et al., 1999). Therefore, maternal-foetal incompatibility from the foetal perspective may be important in human pregnancy due to immune-related processes, which may need initiation/activation by HLA (MHC) incompatibility (humans: Ober et al., 1999;mice: Moldenhauer et al., 2009;Robertson et al., 2009).
The role of the MHC in pre-and postcopulatory selection may be due to the complete MHC haplotype or particular gene classes within the MHC haplotype. This research project investigates whether postcopulatory selection processes may affect the presence of particular combinations of MHC haplotypes in offspring from parents that share founder MHC haplotypes, that is, MHC haplotypes present in the founding rhesus macaques of the studied colony, relative to a Mendelian distribution of these haplotypes.
This was investigated retrospectively in a pedigreed rhesus macaque (M. mulatta) colony. We measured both the representation of MHC haplotypes and the constituting MHC class I (Mamu-A and Mamu-B) and MHC class II (Mamu-DRB) alleles (Doxiadis et al., 2013).
We analysed whether there is selection against offspring with particular MHC haplotypes. The representation for two types of offspring may be relatively low: (i) when the offspring is MHC homozygous for its complete MHC haplotype (Simmons, 2005) or for MHC class I and MHC class II genes or (ii) for heterozygous-compatible offspring, that is, when mother and offspring match for their MHC haplotypes (cf. Ober et al., 1998).

| Study animals and cell lines
The data set consists of rhesus monkeys that were born at the BPRC facilities. The colony was founded in the 1970s by about 140 animals. Twenty-nine males and 69 females were sufficiently successful in breeding and are thus the founders of the analysed cohort. In the past, the colony has been pedigreed for more than seven generations based on segregation of polymorphic MHC markers (Bontrop, Otting, de Groot, & Doxiadis, 1999;Bontrop, Otting, Slierendregt, & Lanchbury, 1995;de Groot, Doxiadis, Otting, de Vos-Rouweler, & Bontrop, 2014). Today, BPRC houses a selfsustaining breeding colony of about 650 rhesus macaques of Indian origin consisting of around 28 simultaneous breeding groups. The macaques are group-housed, mimicking their social organization in the wild (Thierry, 2007). Groups consist of multiple nonrelated mothers and their offspring and one unrelated non-natal male.
Females and their female offspring remain in the group, male offspring remains in the group until at approximately 4 years of age and can stay to an older age when aggression levels allow this.
Breeding is seasonal and typically occurs from November to February. In this colony, males and females can breed from an age of 3 years. When multiple males older than 3 years reside in the group, precopulatory mate preferences can be exerted . In addition, inbreeding can potentially occur between females and natal males (i.e., mother-son back-crossing and siblingcrossing) or young females and long-resident non-natal alpha males (father-daughter back-crossing). Parentage of all animals included in this study had been defined by parentage analysis based on 24 microsatellites localized on 11 different chromosomes (Andrade et al., 2004;Massen et al., 2012). This high number of microsatellites was intentionally chosen to be able to distinguish between related males with regard to having fathered the offspring.
The individuals get a health check on a yearly basis. Immatures receive their first health check between 1 and 2 years of age, when also their MHC haplotypes are determined. Immature mortality almost only occurs in the first month of life; almost all individuals that survived to 1 month also survived to 1 year (A. Louwerse, personal communication, April 7, 2017: N = 1749 births; 8.3% is born dead; 3.9% dies in first week; 1.5% between 1 week and 1 month; 1.6% between 1 month and 1 year; and 1.2% between 1 and 2 years). Lymphoblastoid B-cell lines and genomic DNA (gDNA) have been available from most of the animals in the colony for the last 20 years so that animals could also be typed retrospectively.

| Haplotype definition
A haplotype is defined as the combination of alleles of different genes which are located next to each other on the chromosome and are usually inherited together. MHC haplotypes of Indian rhesus macaques were initially defined by cosegregation of serologically defined MHC class I Mamu-A and Mamu-B and MCH class II DR antigens (Bontrop et al., 1995). Due to sequencing techniques, a high number of alleles and loci have been defined in this colony at the haplotype level (Doxiadis et al., 2013). Currently, a haplotype is The 176 MHC haplotypes known for this population concern combinations of 17 different Mamu-A, 18 Mamu-B and 22 Mamu-DRB types. Individuals that differ for their MHC haplotype may share the same Mamu-A and Mamu-B or Mamu-DRB.

| Categorization of relatedness
The MHC haplotypes of both parents and the four grandparents

| Statistics
We employed a relatively new method to determine whether the data fit a Mendelian distribution: the Bayesian binomial test to accept or reject the null hypothesis (Rouder, Speckman, Sun, Morey, & Iverson, 2009). This test compares a found distribution with an expected, null distribution. In contrast to most statistical tests, this test can not only be used to determine whether an observed distribution differs significantly from an expected distribution but also to determine how well an observed distribution significantly fits the expected, null distribution. In other words, how strong the evidence is for the null hypothesis relative to the alternative hypothesis. The so-called Bayes factor, that measures the odds of the null hypothesis over the alternative hypothesis, is interpreted as follows: Bayes factor > 3 is considered "some evidence"; Bayes factor > 10 is considered "strong evidence"; and Bayes factor > n 30 is considered "very strong evidence" for the null hypothesis over the alternative one (Rouder et al., 2009

| MHC-matching of mother and embryo
We investigated whether the offspring showed over-or underrep-

| DISCUSSION
We determined in a pedigreed rhesus macaque colony whether there was postcopulatory selection for or against particular MHC haplotype combinations when parents shared at least one founder MHC haplotype. Bayesian statistics showed that the distribution of MHC-heterozygous and MHC-homozygous individuals significantly fitted the expected Mendelian inheritance incidence, and therefore, MHC-homozygous individuals were not selected against. However, the overall distribution of the four possible inherited MHC haplotypes tended to differ significantly from expected, indicating selection on the combination of MHC haplotypes. As predicted, relatively few individuals share both MHC haplotypes with their mother, yet this was not significant. Individuals with a set of MHC haplotypes that were identical to their paternal MHC haplotypes tended to be overrepresented.
Postcopulatory processes may promote MHC-heterozygous offspring, resulting in a relatively low number of MHC-homozygous individuals among offspring from parents that share founder MHC haplotypes. However, Bayesian statistics that can confirm the null hypothesis (Rouder et al., 2009) show that the MHC haplotypes from 144 offspring of parents that share one inherited founder MHC haplotype significantly fit the expected 3:1 Mendelian distribution. Also the 10 offspring of parents that share more than one MHC haplotype have a distribution similar to the expected 1:1, but due to the low number of individuals, this cannot significantly fit the expected distribution (Rouder et al., 2009). When inbreeding per se, and not selection on MHC haplotype combinations, determined the effects, we would expect that they differed between on the one hand back-crossings and sibling-crosses, and on the other hand paring among distantly related individuals. We determined whether the type of paring affected the incidence of MHC-heterozygous offspring. However, we found some evidence that the proportion of MHC-homozygous offspring fitted the 3:1 distribution for all three types of parings, indicating that the relatedness among parents did not affect the proportion of MHC-homozygous offspring. These outcomes encompass several processes, including sperm selection during fertilization, implantation, embryo survival and early infant survival. We cannot differentiate between these diverse processes, while each may affect the distribution of MHC haplotypes. Altogether, we found no evidence for postcopulatory selection against MHC-homozygous offspring. As the MHC class I molecules, Mamu-A and Mamu-B, and the MHC class II molecules, DR, have distinct and different functions, it seems possible that one or two of these loci show a differential distribution of heterozygosity vs homozygosity. Moreover, many field studies use only one of these classes, mainly DRB, to determine whether individuals have the same or different MHC antigens (Grob, Knapp, Martin, & Anzenberger, 1998;Knapp et al., 1996;Pechouskova et al., 2015;Setchell & Huchard, 2010;Setchell, Abbott, Gonzalez, & Knapp, 2013;. Therefore, we calculated heterozygosity versus homozygosity frequencies for each locus independently. However, the distributions measured all showed very strong (3:1) or some (1:1) evidence that they fitted the expected 3:1 or 1:1 distributions, indicating that selection against homozygous MHC I A or B or MHC II DR genes was absent. These results suggest that none of the different functions of these genes promote MHC-heterozygous offspring through postcopulatory processes. In addition, the lack of postcopulatory selection based on specific MHC class I and/or MHC class II genes, as mostly defined in field studies, suggests that similarity in part of the MHC haplotype also does not incite postcopulatory selection against MHC-homozygous offspring.
As these results were based on larger subsets (Appendix S1-S4) than those available for individuals that are homozygous for their complete MHC haplotype and a broader application, we are convinced that these data have general biological significance for studies on MHC-matching between parents.
As the results indicate postcopulatory selection on the combina-  . As this tolerance is MHC-specific in mice (Moldenhauer et al., 2009;Robertson et al., 2009) and humans (Sharkey et al., 2012), seminal effects may be enhanced and increase implantation and survival of embryos when the seminal fluid matches not only one, but for two MHC haplotypes with the embryo. However, whether maternal-foetus MHC-matching reduces or paternal-foetus MHC-matching enhances their survival or that both processes are relevant remains to be conclusively established.
Sexual selection may favour the production of MHC-heterozygous offspring through pre-and postcopulatory processes (e.g., Setchell & Huchard, 2010;Tregenza & Wedell, 2000). The postcopulatory processes may concern two different phenomena: differentiation between sperm from kin and nonkin males in polyandrous matings, and within sire selection of specific MHC haplotypes resulting in MHC-heterozygous offspring. Our findings indicate that within sire selection based on offspring MHC haplotypes may take place among the heterozygous offspring, yet does not reduce the incidence of MHC-homozygous offspring. This suggests that when sexual selection processes increase the presence of MHC-heterozygous offspring, this may result from preferential fertilization by unrelated males, either through postcopulatory F I G U R E 1 Observed and expected distribution of major histocompatibility complex (MHC) haplotypes of offspring whose parents share one founder haplotype A (father: A f B; mother: A m C; A f and A m concern the same founder haplotype derived from father or mother, respectively). The overall differences tended to be significant (p = .053). Note that the offspring MHC haplotypes are labelled according to the parent from whom it originated: an offspring that has the same two MHC haplotypes as its father, AfB, obtained the B from its father and the Af from its mother STERCK ET AL.
| 3791 processes favouring fertilization by unrelated males in polyandrous matings; or through pre-copulatory processes that favour mating with unrelated males. Whether this is also present in rhesus macaques could not be determined in our study, as we did not systematically document the number of mating partners of females.
However, female rhesus macaques in our colony typically express mate choice and mate with multiple males when they have this option . Moreover, this is consistent with other studies, which showed that unrelated males are favoured over related males (Pusey & Wolf, 1996). Therefore, precopulatory processes favouring mating with unrelated males and postcopulatory processes in polyandrous matings favouring sperm from unrelated males may be the major reason for the observation that MHCheterozygous offspring are favoured, and not postcopulatory selection against offspring with MHC-homozygous haplotypes.
Altogether, postcopulatory selection based on rhesus macaque offspring's MHC haplotypes when parents share MHC haplotypes in a pedigreed colony takes place between fertilization and infant birth. This postcopulatory selection does not disfavour MHChomozygous offspring, yet may favour particular MHC-heterozygous offspring. However, the processes involved in this postcopulatory MHC selection remain to be established. Altogether, this study suggests that a high incidence of MHC heterozygosity in offspring will not result from postcopulatory selection against offspring with homozygous MHC haplotypes, but from the prevalence of paternity by unrelated males favoured by both pre-and postcopulatory processes.

ACKNOWLEDGEMENTS
We thank Rouder for expanding his program of Bayesian statistics, so we could calculate whether data significantly fit a 3:1 distribution.
We thank Han de Vries statistical advice, Annet Louwerse for providing data on the breeding colony and Jan Langermans for suggesting improvements of the manuscript.

DATA ACCESSIBI LITY
The data sets used for the analyses are provided in the supplementary data.