Male Homosexuality: Absence of Linkage to Microsatellite Markers at Xq28

See allHide authors and affiliations

Science  23 Apr 1999:
Vol. 284, Issue 5414, pp. 665-667
DOI: 10.1126/science.284.5414.665


Several lines of evidence have implicated genetic factors in homosexuality. The most compelling observation has been the report of genetic linkage of male homosexuality to microsatellite markers on the X chromosome. This observation warranted further study and confirmation. Sharing of alleles at position Xq28 was studied in 52 gay male sibling pairs from Canadian families. Four markers at Xq28 were analyzed (DXS1113, BGN, Factor 8, and DXS1108). Allele and haplotype sharing for these markers was not increased over expectation. These results do not support an X-linked gene underlying male homosexuality.

Previous studies have suggested that there is a genetic component in male sexual orientation. These include controlled family studies that have shown an increased frequency of homosexual brothers of homosexual index subjects as compared to heterosexual index subjects (1, 2) and twin studies, which have shown increased concordance for homosexual orientation in monozygotic as compared to dizygotic twins (3). On the other hand, the similar rates of male homosexuality in biological and adoptive siblings of male homosexual index subjects (3), coupled with methodological uncertainties in family and twin studies of homosexuality, suggest caution in accepting a genetic-epidemiological basis for homosexuality (4,5).

The strongest support for a genetic component in male sexual orientation came from the studies of Hamer et al.(6, 7), who posited the involvement of an X-linked gene at position Xq28, based on family recurrence patterns and molecular analysis of the X chromosome in sibships in which there were multiple brothers with homosexual orientation. Specifically, Hamer and colleagues obtained family history information from 76 gay male index subjects and 40 gay brother pairs about the sexual orientation of the first-, second-, and third-degree relatives, with follow-up interviews of a smaller proportion of relatives. They reported increased rates of homosexual orientation in the maternal uncles and male cousins through maternal aunts, which was suggestive of X-linked inheritance. Molecular analysis of the X chromosome revealed an excess of allele sharing in the region of Xq28 in 40 homosexual brother pairs (6) and, to a lesser extent, in a follow-up study of 33 additional pairs (7).

However, the evidence for X linkage has been questioned on theoretical and empirical grounds (8, 9). Most would agree that male homosexual orientation is not a simple Mendelian trait. There would be strong selective pressures against such a gene. Hamer's identification of a contribution from a gene near Xq28 to homosexuality in some families that were selected for X-linked transmission of that trait might be fraught with type 1 (false positive) error. This is important to consider, given the irreproducibility of many linkage reports for complex behavioral traits.

Given the political and social ramifications of gene linkage in homosexuality, we launched independent genetic studies of male sexual orientation in Canada. Specifically, we advertised in Canadian gay news magazines (Xtra and Fugue) for families in which there were at least two gay brothers. One hundred and eighty-two individuals responded to the advertisement. The respondents volunteered information about the sexual orientation of individuals in their families, including siblings, parents, uncles, aunts, and first cousins, although all members of the extended family were not directly interviewed. The 182 families included 614 brothers, 269 (44%) of whom were homosexual. There were 148 families with two gay sons, 34 families with three, and two families with four. The high rate of sibling concordance reflects the nature of the advertisement. The sample included 270 sisters, 49 (18%) of whom were said to be gay. This rate is high compared to the frequency of homosexual orientation in women as ascertained in most population-based studies, which suggest a sister concordance rate of 14% (10).

Our molecular analysis was based on 52 gay sibling pairs from 48 families who were willing to provide blood samples. Sexual orientation was confirmed for all subjects at the time of blood sampling by the direct questioning of a gay interviewer. The index subject read gay magazines and volunteered that he was gay, and this observation was corroborated by interviewing the gay brother. We believe that the rate of false positives, as in Hamer's study, was low (6). The sample included 46 families with two gay brothers. There were two families with three gay brothers, and these were considered as six pairs. Four markers were analyzed (DXS1113, BGN, Factor 8, and DXS1108), along a 12.5-centimorgan (cM) region of Xq28. The methods were as described (11). The allelic sharing is shown in Table 1.

Table 1

Allele sharing by state in 52 homosexual brother pairs.

View this table:

Genotyping was performed on DNA samples from the brothers themselves without genotyping of parents (alleles were identified by comparison with population-based controls, known as “identical by state,” rather than by confirmation of maternal transmission, known as “identical by descent”). Maternal DNA was difficult to obtain. As controls, we included an additional 33 sibling pairs who were concordant for multiple sclerosis. These were genotyped simultaneously with the gay sibships (Table 2). Allele scoring was performed independently by two evaluators who were blind to the status of the sibship.

Table 2

Allele sharing in 33 control brother pairs.

View this table:

A priori, a pair of brothers will share an X-linked maternal allele, identical by descent with probability = 1/2. Therefore, for a marker with heterozygosity H, under the null hypothesis of no excess sharing, a brother pair will share an allele identical by state with probability 1/2 + 1/2 (1 – H) = 1 – H/2. For an X-linked trait-influencing locus in the region, the sharing will be increased. For the distal three markers taken as a haplotype, the probability that brothers share the full haplotype is approximately (1 – H/2) (1 – θ)2, where θ is the recombination fraction in the entire interval (2.5% for all three markers), and H is the heterozygosity of the full haplotype (which is 1 minus the product of the homozygosities at the three loci, assuming linkage equilibrium).

Table 1 shows no excess sharing for any of the four markers tested nor for the haplotype of the distal three loci. These results are not consistent with an X-linked gene underlying sexual orientation in this region of the X chromosome.

We further analyzed these data with multipoint sib pair analysis by means of the computer program ASPEX (12). Multipoint lod (logarithm of the odds ratio for linkage) scores were calculated along the 12.5-cM region for two values of λs (2.0 and 1.5), where λs is the ratio for homosexual orientation in the brothers of a gay index subject, as compared to the population frequency, that is attributable to a gene in this region. Very strong exclusion is obtained for lod scores <–2.0, and strong exclusion is obtained for lod scores <−1.0 (13). As depicted in Fig. 1, λs values of two or greater can be very strongly excluded. Values of 1.5 or greater can also be strongly excluded. The lod scores were clearly negative for all values of λs .

Figure 1

Multipoint map for Xq28. Multipoint lod scores were calculated along the 12.5-cM region for two values of λs (2.0, solid line; 1.5, dashed line), where λs is the ratio for homosexual orientation in the brothers of a gay index subject, as compared to the population frequency, that is attributable to a gene in this region. Very strong exclusion is obtained for lod scores <–2.0, and strong exclusion is obtained for lod scores <−1.0.

Hamer and colleagues described linkage of male homosexuality to polymorphic markers at Xq28 in 40 brother pairs. The sharing was 33/40, deemed to be significant with a λs value of 2.86 (6). In a follow-up study of 33 gay male sibling pairs (32 informative), 22 shared all the Xq28 markers (7). Our sample comprised 46 sib pairs and 2 sib trios. The sharing of distal Xq28 markers in the 46 sib pairs was 20/46. For one of the sib trios, all three brothers shared the same X chromosome; for the other trio, two shared the same X chromosome and the other was different. Therefore, forming independent sib pairs by picking two pairs out of three for each sibship gives a total X-chromosome sharing of either 2 out of 4 or 3 out of 4. For comparison with the results of Hamer et al., we used the more favorable 3 out of 4, which gave a total of 23 out of 50 chromosomes shared for our sample. This result was highly different statistically from the first study of Hamer (6) (chi square value = 11.09,P <0.001) but not statistically different from the second study (7) (chi square = 3.21, P > 0.05). Combining the two replication studies gives a total sharing of 45 out of 82 (55%), which was not significantly different from 50% (chi square = 0.78, P > 0.30). Also, the sharing for the two replication studies combined is significantly different from the original study of Hamer et al. (chi square = 7.74,P < 0.01).

It is unclear from the original study to what extent families were excluded to produce the data set in which the positive linkage analysis was reported. Families were excluded if a father was gay or if there were any first-degree lesbian relatives. By these precise criteria, two sib pairs would have been excluded from our study (one with a gay father and one with two gay parents). For the remaining pairs, the linkage evidence was the same as for the entire group.

It is unclear why our results are so discrepant from Hamer's original study (6). Because our study was larger than that of Hamer et al., we certainly had adequate power to detect a genetic effect as large as was reported in that study. Nonetheless, our data do not support the presence of a gene of large effect influencing sexual orientation at position Xq28.

Although we found no evidence of linkage of sexual orientation to Xq28, these results do not preclude the possibility of detectable gene effects elsewhere in the genome.

  • * To whom correspondence should be addressed. E-mail grice{at}


View Abstract

Navigate This Article