485umb_krawczak.qxd

Hum Genet (2001) 108 : 249–254DOI 10.1007/s004390100485 Ulrike Sauermann · Peter Nürnberg ·
Fred B. Bercovitch · John D. Berard · Andrea Trefilov ·
Anja Widdig · Matt Kessler · Jörg Schmidtke ·
Michael Krawczak
Increased reproductive success of MHC class II heterozygous males among free-ranging rhesus macaques Received: 27 November 2000 / Accepted: 11 January 2001 / Published online: 7 March 2001 Abstract Gene conversion and balancing selection have
sity (Parham and Ohta 1996). By virtue of indirect rea- been invoked to explain the ubiquitous diversity of the soning, over-dominant and frequency-dependent selection antigen-presenting proteins encoded in the vertebrate ma- have become the most widely accepted explanations for jor histocompatibility complex (MHC). In the present the maintenance of intra-species MHC polymorphism study, direct evidence for over-dominant selection pro- (Hedrick and Thomson 1983; Hughes and Nei 1988; moting MHC diversity in primates is provided by the ob- Takahata and Nei 1990; Hill et al. 1991) and a variety of servation that, in a large free-ranging population of rhesus parasites are thought to be critical for both models. In- macaques, males heterozygous at MHC class II locus deed, an influence of particular MHC alleles or of ho- Mamu-DQB1 sired significantly more offspring than ho- mozygosity for certain MHC allelic classes upon morbid- mozygotes (the male-specific selection coefficient s equals ity in pathogen-infected individuals has been observed in 0.34). This heterozygote advantage appeared to be inde- humans (Hill et al. 1991; Thursz et al. 1997, 1999; Car- pendent of the actual male Mamu-DQB1 genotype. No rington et al. 1999) and other primates (Evans et al. 1999; similar effect emerged for a captive group of monkeys of Sauermann et al. 2000). Choice of mating partners with similar genetic background but under veterinary care.
differing MHC genotypes (sexual selection), on the otherhand, can promote polymorphism without invoking hand-icaps and might be involved in kin recognition (Singh et al. 1987; Potts et al. 1991). Finally, a heterozygote excesspotentially explicable by maternal-fetal selection (Black The antigen-presenting proteins encoded by genes of the and Hedrick 1997) has been observed in isolated human major histocompatibility complex (MHC) play a key role populations. Prenatal wastage in MHC class-I-matched in many species in terms of their capability to initiate a couples has also been reported to occur in non-human pri- specific immune response. One of the most remarkable characteristics of these proteins is their persistent diver- Although none of the above theories are mutually ex- clusive, their relative importance has been subject to con-troversial debate. Previous studies that have examined the association between MHC genotype and reproductive fit- Deutsches Primatenzentrum, Arbeitsgruppe Primatengenetik, ness in higher vertebrates have generally adopted indirect measures of reproductive success, including mating pref- erences (Wedekind et al. 1995), male ornamentation (von Institut für Medizinische Genetik, Universitätsklinikum Charité, Schantz et al. 1996) and allele frequency distributions (Wenink et al. 1998). The scientific merit of these studies F. B. Bercovitch · J. D. Berard · Matt Kessler notwithstanding, however, the most rigorous test for the Caribbean Primate Research Center, University of Puerto Rico, strength of any MHC-associated selective pressure would PO Box 1053, Sabana Seca, Puerto Rico 00952, USA be to determine the actual number of progeny produced.
In an ongoing long-term study of the reproductive be- Institut für Humangenetik, Medizinische Hochschule, haviour of rhesus macaques (Macaca mulatta), we have been able to relate directly biological fitness (reproductive success) with genotypes at MHC class II locus Mamu- DQB1 in three social groups of monkeys from the Cayo University of Wales College of Medicine, Santiago colony, Puerto Rico. We have observed that sig- Heath Park, Cardiff CF14 4XN, UK e-mail: krawczak@cardiff.ac.uk nificantly more offspring are sired by heterozygous males compared with homozygous males, a result that provides 1987). Survival analysis of male rhesus macaques was performed direct evidence for over-dominant selection (heterozygote using the LIFETEST procedure as implemented in the SAS/STATsoftware package (Version 6, Fourth Edition, Chapter 26).
advantage) promoting MHC diversity in non-human pri- LIFETEST allows non-parametric survival time distributions to be mates. At the same time, mate choice has been found to be estimated from both failure (“dead”) and censoring (“alive”) data.
independent of Mamu-DQB1 genotype, so that sexual se- Differences in survival times were tested for significance by means lection could be excluded from playing an important role of a log-rank χ2 statistic. The expected degree of allele sharing be- tween reproductively successful mating partners was determinedvia simulation. To this end, the observed genotypes of mothers andactual sires were randomly redistributed over infants (10,000 repli-cations), and the mean allele sharing between parents recorded.
Goodness-of-fit between observed and expected allele sharing wasassessed by χ2 analysis. Finally, Fisher’s exact test was used to compare Mamu-DQB1 allele frequencies between male cate-gories, whereas mean ages of sires at first offspring birth were Three social groups of rhesus macaques were analysed in the pre- compared by Student’s t test.
sent study. Two of them (groups R and S) reside on Cayo Santi-ago, a 15-ha island located offshore Puerto Rico and populatedwith rhesus monkeys that have been subject to numerous behav- ioural studies (Kessler and Berard 1989). Cayo Santiago is a semi-natural enviroment where veterinary care is limited to tetanus in-occulations being given to all yearlings and 2-year-olds. During the annual trapping season, blood samples are taken for research purposes from all yearlings and other selected animals. Socialgroup M, in contrast, resides in an outdoor enclosure of 0.4 ha at Paternity analysis (Nürnberg et al. 1998) was performed Sabana Seca Field Station, Puerto Rico, where it has been kept for all animals residing in or born into groups S, R and M since 1984. Tetanus shots are given every year and Ivermectin, abroad-band antihelminth, is administered twice a year in group M.
of Cayo Santiago between 1989 and 1997 (n=1453). Data If animals become severely injured, lose weight and/or develop di- from the two free-ranging social groups R and S were arrhoea, they are removed from the group and appropriate treat- combined, since these troops live in close proximity and ment is initiated. Groups R and S are not subject to any veterinary regular male migration occurs between them. Of the 598 intervention. All three groups comprise decendants of the 409 mon-keys originally brought to Cayo Santiago in 1938.
sirehoods established in groups R/S, a total of 487(81.4%) were from a heterozygote male, whereas 111(18.6%) were from a homozygote (Table 1). Since all off- spring had descended from males that had been consid-ered as a potential sire for at least 200 infants, the same Paternity testing using 10 microsatellites (Nürnberg et al. 1998)was performed for all animals residing in or born into groups S, R threshold was also used to define which males had not and M between 1989 and 1997 (n=1453). Since the employed been reproductively successful (“vain potential sirehood”; markers yield a combined (theoretical) paternity exclusion rate 30 homozygotes, 80 heterozygotes). These criteria re- larger than 99.9%, no more than one potential sire was expected toresult for the majority of mother-infant pairs. For a given infant, allmales older than 40 months by, and alive at least 200 days prior to, Table 1 Male reproductive success and Mamu-DQB1 heterozy-
the infant’s date of birth were considered as probable sires. Candi- gosity (numbers in brackets refer to reproductively successful date sires for group R and S infants were allowed to come from both free-ranging groups. Sirehood was regarded as establishedwhen a male reached a log-likelihood ratio (LR) in favour of pa- ternity that was (1) larger than two (corresponding to a standard- ized paternity probability of 99%) and (2) at least one unit larger than the LR of any other male. Likelihood calculations (Krawczak 1999) were carried out based upon the overall microsatellite allelefrequencies observed in the colony and assuming a mutation rate of 10–3 per locus. By these means, 598 sirehoods could be estab- lished in groups S and R and 129 sirehoods could be established in Mamu-DQB typing was performed as previously described (Sauermann et al. 1996). Rhesus DQA and DQB loci are in strong linkage disequilibrium (Sauermann 1998), so that homozygosity for DQB1 could be confirmed by DQA1 analysis in order to mini- The mean number of offspring per potential sire was tested for sig- b2×11, 12, 13, 14, 19, 2×23 and 25 offspring nificant differences between paternal Mamu-DQB1 genotypes us- ing a randomization test with 10,000 replications (Edgington sulted in a mean offspring number (± standard error, SE) Table 2 Female reproductive success and Mamu-DQB1 het-
of 2.32±0.26 for heterozygous males and 1.52±0.25 for erozygosity (n number of females, µ mean number of offspring perfemale, SE standard error of mean female offspring number, P as homozygotes, suggesting a selection coefficient s of determined by a one-sided randomization test for heterozygote ad- (2.32-1.52)/2.32=0.34. The difference between the two genotypes was statistically significant (one-sided random-ization, P=0.03). The average heterozygosity per animal at the microsatellites used for paternity testing was 0.85 for Mamu-DQB1 homozygotes and 0.86 for heterozy-gotes. A potential genome-wide decrease in heterozygos- ity and any consequent inbreeding depression can there- fore be excluded from affecting Mamu-DQB1 homozy- In group M, which has been kept in an enclosure and is subject to medical care, 80 infants had been sired by 41heterozygous males (mean 1.95±0.74) as opposed to 69 reach statistical significance (one-sided randomization, offspring who were descendents of 19 homozygotes P=0.12). We may therefore conclude that the reproductive (mean 3.63±1.50). The criterion for a reproductively un- advantage associated with Mamu-DQB1 heterozygosity successful male was vain potential sirehood for at least 30 in free-ranging rhesus macaques on Cayo Santiago is animals. The mean difference between the two genotypes stronger in, or is confined to, males. No difference in off- was not statistically significant (two-sided randomization spring number per female was observed in group M.
test, P=0.29), mainly because most of the excess homozy-gote reproductivity could be attributed to four males only.
Nevertheless, the results imply that the Mamu-DQB1 lo- cus is subject to over-dominant selection among male rhe-sus macaques in the two free-ranging groups R and S but In order to assess whether the observed male heterozygote advantage in groups R and S was explicable in terms of agenotype-dependent bias in mating preference, the levelof Mamu-DQB1 allele-sharing between mothers and sires Over-dominant selection and male demography was determined for couples in which both partners hadbeen genotyped (Table 3). When the observed allele-shar- One possible explanation for a difference in male repro- ing was compared with its expectation based upon 10,000 ductive success associated with Mamu-DQB1 heterozy- simulations of random mating in the respective popula- gosity would be a shorter life expectancy of homozygotes.
tion, a trend towards increased allele-sharing only became However, the 358 heterozygous males in groups R and S apparent in group M (zero or one allele shared vs two al- that had been genotyped for Mamu-DQB1 had a mean life leles shared: χ2=3.27, 1 df, P=0.07). In groups R and S, span (± SE) of 155.1±5.8 months, whereas for the 114 the sharing of zero or one allele occurred slightly more homozygotes, the corresponding figure was 152.8±10.6 frequently than expected but the difference was not sig- months. These results indicate that the potentially fertile nificant (χ2=0.57, 1 df, P>0.45) . There was thus no evi- life span of heterozygotes is only marginally (i.e.
dence that mother-sire genotype similarity at Mamu- 115.1/112.8=1.02) longer than that of homozygotes and DQB1 conferred a reproductive advantage or disadvan- explains no more than approximately 4% of the excess tage in the two free-ranging groups. In addition, the fre- fertility of heterozygote males. Alternatively, heterozy- quency distribution of Mamu-DQB1 alleles among repro- gotes could have sired offspring earlier than homozy- ductively successful males was not significantly different gotes, thereby gaining an advantage over the latter at the from that observed among unsuccessful males (Table 4, beginning of their reproductive phase. The mean age of Fisher’s exact P=0.51 for groups R and S, P=0.90 for a sire at the birth of its first offspring was indeed lower for the 130 reproductively successful heterozygotes(100.2±2.8 months) compared with the 43 homozygotes Table 3 Mother-sire allele-sharing at Mamu-DQB1. Expected
(102.3±6.2 months) but the difference was only small and frequencies were determined by simulation involving the random not statistically significant (t=0.350, 171 df, P>0.3).
combination of mating partners in the respective population(10,000 replications) No Mamu-DQB1 heterozygote advantage among females Among the females in groups R and S with at least one offspring, Mamu-DQB1 heterozygotes had a slightly higher reproductive success than homozygotes (Table 2).
However, the difference in mean offspring number be- tween the two genotypes was only 13% and failed to Table 4 Mamu-DQB1 allele
tive pressure that is at least one order of magnitude higher Among the rhesus macaques residing on Cayo Santi- ago, MHC loci DQA, DQB and, to a lesser extent, DRB are in strong linkage disequilibrium with one another (Sauermann 1998; Khazand et al. 1999). This tight associ- ation need not necessarily be attributed to inbreeding since DQ-DRB haplotypes identical to those segregating on Cayo Santiago have also been found in rhesus macaques from other locations (Khazand et al. 1999).
Furthermore, our results imply that whole DQ-DRB hap- lotypes, rather than single alleles, are subject to over- dominant selection and theoretical considerations suggest that over-dominant selection would indeed be sufficient to explain at least the strong linkage disequilibrium observed between rhesus macaque DQA and DQB genes (Slatkin Inbreeding seems to be avoided in rhesus macaques by females choosing a new mating partner in each breeding group M). Thus, single alleles also did not appear to be season. A whole ensemble of traits is known to regulate particularly associated with any increased male reproduc- male mating success, including dominance status, age, troop tenure time, body weight, fat level and “novelty”(Berard et al. 1994; Bercovitch and Nürnberg 1996, 1997;Berard 1999). All of these traits vary during life. In the present report, we have highlighted Mamu-DQB1 as a ge-netic marker that is not subject to any environmental There is a growing body of literature regarding the influ- change but nevertheless has direct implications for male ence of MHC genotypes upon survival rates and repro- reproductive success. However, allele frequencies at ductive activity in various species. However, a subtle but Mamu-DQB1 are similar among reproducing and non-re- very important difference exists between these two pa- producing males in all three social groups. The existence rameters and actual reproductive success, the latter repre- of Mamu-DQB1 alleles specifically promoting male fe- senting the only critical issue for the evolutionary mainte- cundity can thus be excluded. Furthermore, since mothers nance of MHC diversity. The most rigorous test for any in free-ranging groups R and S show no preference for MHC-associated heterozygote advantage is therefore to sires with Mamu-DQB1 alleles differing from their own, perform a field study and to determine reproductive suc- no evidence has been found for an association between cess directly. Ideally, such a study would be conducted in Mamu-DQB1 and mate choice acting as a means of in- a natural environment and would carefully control for possible confounding factors. By the very nature of the The most plausible explanation for the increased repro- study design and purpose, however, such requirements are ductive success of free-ranging Mamu-DQB1 heterozy- gous males that is independent of the actual genotype is We have demonstrated here that, under the quasi-nat- increased resistance to the debilitating effects of injury ural conditions of Cayo Santiago, the reproductive suc- and parasite infection. Although most common parasites cess of male rhesus macaques is higher for heterozygotes including Stronglyoides, Balantidium coli and Trichuris at MHC class II gene locus Mamu-DQB1 than for ho- are unlikely to represent a major cause of death on Cayo mozygotes. It should be emphasized that this observation Santiago (Knezevich 1998), parasites such as helminths is unlikely to be attributable to false assignments of sire- may nevertheless cause lethargy and reduce the ability of hood since errors in paternity assessment are rare (Nürn- a male to obtain access to females. Variable propensity to berg et al. 1998) and Mamu-DQB1 genotyping is unlikely parasite load would thus explain why antihelminth treat- to be biased in any systematic way. Based upon theoreti- ment has served to augment the reproductive success of cal considerations and sequence data, selection maintain- Mamu-DQB1 homozygous males in captive group M. In- ing MHC diversity is generally thought to be weak in hu- terestingly, all three males removed from group M be- mans and selection coefficients (s) are expected to be of tween 1991 and 1995 and treated for parasite infestation the order of 0.01 (Satta et al. 1994). By contrast, our data because of diarrhoea/underweight status were Mamu- on rhesus macaques suggest that s equals 0.34, a figure al- most as high as empirical estimates obtained for HLA-A Since heterozygosity at Mamu-DQB1 does not appear and HLA-B loci in South Amerindian families, where s is to have any positive influence upon female reproductive 0.46 (Black and Hedrick 1997). Taken together, it emerges success, we may surmise that the selective pressure at this that, under certain demographic and ecological circum- locus is sex-specific. Such a discrepancy is not unex- stances, primates can be subject to MHC-associated selec- pected in a female-philopatric male-dispersed species such as macaques. Female rank, and therefore access to food, Evans DT, Knapp LA, Jing P, Piekarczyk MS, Hinshaw VS, which is necessary to raise infants, is strictly inherited Watkins DI (1999) Three different MHC class I molecules bindthe same CTL epitope of the influenza virus in a primate with culturally in rhesus macaques, thereby possibly rendering limited MHC class I diversity. J Immunol 162:3970–3977 genetic influences upon the reproductive success of fe- Fedigan LM, Zohar S (1997) Sex difference in mortality of Japan- males less relevant. In addition, males in free-ranging ese macaques: twenty-one years of data from the Arashiyama populations of non-human primates appear to be more West population. Am J Phys Anthropol 102:161–175 susceptible to parasites than females (Fedigan and Zohar Hedrick PW, Thomson G (1983) Evidence for balancing selection 1997; Knezevich 1998). This increased suceptibility is re- Hill AV, Allsopp CE, Kwiatkowski D, Anstey NM, Twumasi P, portedly stronger in juvenile males and may be related to Rowe PA, Bennett S, Brewster D, McMichael AJ, Greenwood stress caused by wounding. Male injuries, on the other BM (1991) Common West African HLA antigens are associ- hand, are especially frequent in monkeys upon group dis- ated with protection from severe malaria. Nature 352:595–600 Hughes AL, Nei M (1988) Pattern of nucleotide substitution at ma- persal and during the mating season (Wilson and Boelkins jor histocompatibility complex class I loci reveals overdomi- 1970), both of which represent critically important times for reproductive success. In agreement with this hypothe- Kessler MJ, Berard JD (1989) A brief description of the Cayo San- sis, no heterozygote advantage has been observed in tiago rhesus monkey colony. P R Health Sci J 8:55–59 Khazand M, Peiberg C, Nagy M, Sauermann U (1999) DQ-DRB group M where males are provided with antibiotic treat- haplotype analysis in rhesus macaques: evidence for a number of haplotypes displaying low allelic polymorphism. Tissue Preferential reproduction of Mamu-DQB1 heterozy- gous macaques in response to parasite load and/or stress Knapp LA, Ha JC, Sackett GP (1996) Parental MHC antigen shar- (wounding) would be a typical example of the way in ing and pregnancy wastage in captive pigtailed macaques. J Reprod Immunol 32:73–88 which the presence of an environmental challenge may Knezevich M (1998) Geophagy as a therapeutic mediator of en- ensure that the same threat is eventually overcome by the doparasitism in a free-ranging group of rhesus macaques species. Thus, non-specific selection, although potentially (Macaca mulatta). Am J Primatol 44:71–82 inefficient in the short term, may well be evolutionarily Krawczak M (1999) Statistical inference from DNA evidence. In: Epplen JT, Lubjuhn T (eds) A laboratory guide to DNA finger- favourable to a species when its environmental challenges printing/profiling. Birkhäuser, Basel, pp 229–244 tend to change rapidly. This also implies that, despite the Nürnberg P, Sauermann U, Kayser M, Lanfer C, Manns E, Widdig high selection coefficient obtained, selective pressure may A, Berard J, Bercovitch F, Kessler M, Schmidtke J, Krawczak not have been strong or constant enough in rhesus M (1998) Paternity assessment in rhesus macaques (Macacamulatta): multilocus DNA fingerprinting and PCR marker typ- macaques to evolve a means to recognize specific Mamu- Parham P, Otha T (1996) Population biology of antigen presenta- tion by MHC class I molecules. Science 272:67–74 Acknowledgements We thank Gerhard Hunsmann for his sup-
Potts WK, Manning CJ, Wakeland EK (1991) Mating patterns in port and encouragement, Eberhard Günther, Mauvis Gore and seminatural populations of mice influenced by MHC genotype.
Marion Nagy for critical comments on the manuscript, and Karsten Lohan, Heike Rössler, Ingrid Barth, Tatjana Riemenschneider, Satta Y, O’hUigin C, Takahata N, Klein J (1994) Intensity of nat- Christina Oberdieck, Christina Peiberg and Wolfgang Kühnau for ural selection at the major histocompatibility complex loci.
expert technical assistance. This work was supported by the Deutsche Forschungsgemeinschaft (Nu 50/3–2, Kr 1093/3–1, Kr Sauermann U (1998) DQ-haplotype analysis in rhesus macaques: 1093/5–2 and Sa 661/1–1) and the National Science Foundation implications for the evolution of these genes. Tissue Antigens (IBN 9209510; FBB). The Caribbean Primate Research Center is Sauermann U, Arents A, Hunsmann G (1996) PCR-RFLP-based Mamu-DQB1 typing of rhesus monkeys: characterization oftwo novel alleles. Tissue Antigens 47:319–328 Sauermann U, Stahl-Hennig C, Stolte N, Mühl T, Krawczak M, Spring M, Fuchs D, Kaup FJ, Hunsmann G, Sopper S (2000)Homozygosity for a conserved MHC class II DQ-DRB haplo- Berard J (1999) A four-year study of the association between dom- type is associated with rapid disease progression in SIV-in- inance rank, residency status and reproductive activity in male fected macaques: results from a prospective study. J Infect Dis rhesus macaques (Macaca mulatta). Primates 40:159–176 Berard J, Nürnberg P, Epplen JT, Schmidtke J (1994) Alternative Schantz T von, Wittzell H, Goransson G, Grahn M, Persson K reproductive tactics and reproductive success in male rhesus (1996) MHC genotype and male ornamentation: genetic evi- cence for the Hamilton-Zuk model. Proc R Soc Lond [Biol] Bercovitch FB, Nürnberg P (1996) Socioendocrine and morpho- logical correlates of paternity in rhesus macaques. J Reprod Singh PB, Brown RE, Roser B (1987) MHC antigens in urine as olfactory recognition cues. Nature 327:161–164 Bercovitch FB, Nürnberg P (1997) Genetic determination of pater- Slatkin M (2000) Balancing selection at closley linked, overdomi- nity and variation in male reproductive success in two popula- nant loci in a finite population. Genetics 154:1367–1378 tions of rhesus macacques. Electrophoresis 18:1701–1705 Takahata N, Nei M (1990) Allelic genealogy under overdominant Black FL, Hedrick PW (1997) Strong balancing selection at HLA and frequency-dependent selection and polymorphism of major loci: evidence from segregation in South Amerindian families.
histocompatibility loci. Genetics 124:967–978 Thursz MR, Thomas HC, Greenwood BM, Hill AV (1997) Het- Carrington M, Nelson GW, Martin MP, Kissner T, Vlahov D, erozygote advantage for HLA class-II type in hepatitis B virus Goedert JJ, Kaslow R, Buchbinder S, Hoots K, O’Brien SJ (1999) HLA and HIV-1: Heterozygote advantage and B*35-Cw*04 disadvantage. Science 283:1748–1752 Edgington E (1987) Randomization tests. Dekker, New York Thursz M, Yallop R, Goldin R, Trepo C, Thomas HC (1999) In- Wenink PW, Groen AF, Roelke-Parker ME, Prins HH (1998) fluence of MHC class II genotype on outcome of infection with African buffalo maintain high genetic diversity in the major hepatitis C virus. The HENCORE group. Hepatitis C European histocompatibility complex in spite of historically known pop- Network for Cooperative Research. Lancet 354:2119–2124 ulation bottlenecks. Mol Ecol 7:1315–1322 Wedekind C, Seebeck T, Bettens F, Paepke AJ (1995) MHC-de- Wilson AC, Boelkins RC (1970) Evidence for seasonal variation pendent mate preferences in humans. Proc R Soc Lond [Biol] in aggressive behaviour by Macaca mulatta. Anim Behav

Source: http://www.eva.mpg.de/pks/staff/widdig/pdf/Sauermann_et_al_2001_Hum_Genet.pdf