Repetitive microsatellites mutate at relatively high rates and may contribute to the rapid evolution of species-typical traits. We show that individual alleles of a repetitive polymorphic microsatellite in the 5′ region of the prairie vole vasopressin 1a receptor (avpr1a) gene modify gene expression in vitro. In vivo, we observe that this regulatory polymorphism predicts both individual differences in receptor distribution patterns and socio-behavioral traits. These data suggest that individual differences in gene expression patterns may be conferred via polymorphic microsatellites in the cis-regulatory regions of genes and may contribute to normal variation in behavioral traits.
Social behavior evolves rapidly as evidenced by the diversity of species-typical social structures among closely related species; however, the underlying mechanisms of this rapid evolution are currently unknown. Evolution of species-typical behavioral traits requires behavioral diversity and a polymorphic genetic mechanism producing such diversity. The high levels of polymorphism in repetitive DNA sequences (microsatellites) make them useful as markers to distinguish among individuals. Expanded repeats in coding regions of genes are responsible for diseases such as Huntington's disease and the spinocerebellar ataxias (1). Recently, expansion and contraction in the coding regions of several developmental genes has been hypothesized to underlie the rapid and continuous evolution of snout morphology in domesticated dog breeds (2). Although there is not as much empirical evidence, variation in regulatory regions of genes likely plays a critical role in the generation of morphological variation by altering the timing and location of gene expression (3, 4). Microsatellites in the cis-regulatory regions of genes may significantly enhance the rate of evolution of gene expression patterns and selectable phenotypic traits (5–7).
Rodents of the genus Microtus (voles) show dramatic species differences in social structure (8). Prairie voles form lifelong attachments with a mate, are biparental, and show high levels of social interest (9). In contrast, the closely related montane vole does not pair bond, the males do not contribute to parental care, and they appear socially indifferent (10). Species differences in the pattern of vasopressin 1a receptor (V1aR) expression in the brains of these species contribute to the species differences in social structure (11–13). The species-specific patterns of V1aR expression appear to be regulated by differences in a microsatellite in the 5′ regulatory region of the gene encoding V1aR (avpr1a). This microsatellite is highly expanded in prosocial prairie and pine voles, consisting of several repeat blocks interspersed with nonrepetitive sequences (Fig. 1, A and B), compared with a very short version in the asocial montane and meadow voles (14). The species differences in this microsatellite have functional consequences in cell culture: In transcription reporter assays, changes in the microsatellite locus modify luciferase reporter activity in a cell-type–dependent manner (15).
Because microsatellites offer a mechanism of continuous phenotypic variation, it follows that there should also be a microsatellite-phenotype relation within a species. Accordingly, within the prairie vole species, there is variation in social behavior across different geographical populations and even within the same population in communal pens (16, 17). In addition, the length of the avpr1a microsatellite is variable [Fig. 1, B and C; (15, 18)] as are V1aR distribution patterns across individuals within the prairie vole species (19), although levels of intraspecific variation are lower than the levels of interspecific variation in both genotype and phenotype (19, 20). Whereas microsatellite length is associated with species differences in V1aR expression and social structure, we hypothesize that intraspecific variation in microsatellite length generates intraspecific variation in V1aR expression and, consequently, socio-behavioral traits.
To determine whether or not intraspecific variation in the microsatellite itself is sufficient to change gene expression, we created luciferase reporter constructs containing ∼3.5 kb of prairie vole avpr1a 5′ regulatory region, including the microsatellite locus. Holding the rest of the 5′ regulatory region constant, we interchanged two different prairie vole alleles of this microsatellite to test them against each other for their ability to drive the luciferase reporter in A7r5 cells. The two alleles only differ in length by 19 base pairs (bp), primarily attributable to an expanded GA repeat in the longer allele. In three independent cell culture experiments, the longer allele had significantly increased levels of luciferase activity compared with the shorter allele, demonstrating that intraspecific variation in the microsatellite itself modifies gene expression (Fig. 1D, two-tailed Student's t test with Welch's correction, P < 0.0001). Although the long allele showed higher levels of expression in this particular cell line that we tested, other evidence indicates that the effect may be cell-type dependent.
We next created 25 breeding pairs of prairie voles based on the length of their avpr1a microsatellite polymorphism to generate testable offspring that were homozygous for either longer or shorter than the average allele length. Two out of three litter cohorts were randomly cross-fostered within 6 to 8 hours after birth to reduce the potential confounding variable of rearing environment. There were no genotype differences in birth rates (Mann-Whitney, P > 0.05); however, short-allele breeder pairs had higher rates of pup mortality (12 out of 25, versus 3 out of 36, χ2 = 10.83, P < 0.001). This effect was only apparent with the first litter, as both genotype groups had high rates of survival after the pairs had become experienced parents. Because paternal care in prairie voles is influenced by vasopressin, as well as selective V1aR antagonists (21), we measured parental care during the first ten post-natal days. Twenty hours of observations revealed no genotype differences in the amount of time the pair were on the nest together or in the time each parent was alone on the nest (Fig. 2A, Mann-Whitney, P > 0.05). Additionally, there were no genotype differences in the amount of time the nest was left unoccupied by both parents simultaneously (Fig. 2A, Mann-Whitney, P > 0.05). There were also no genotype differences in nursing postures adopted by the females (Fig. 2B, Mann-Whitney, P > 0.05). In contrast, there were significant genotype differences in the frequency of pup licking and grooming (Fig. 2C, Mann-Whitney, P < 0.05). This genotype difference can be fully attributed to a higher grooming frequency of long-allele males compared with short-allele males (Mann-Whitney, P < 0.01), as there were no genotype differences in pup licking and grooming frequency by the females (Mann-Whitney, P > 0.05). These data coincide with the prevailing hypothesis in the literature that vasopressin systems contribute to male, not female, species-typical behaviors. The genotype differences in total licking and grooming (15 to 20%) are only slightly less than the 20 to 30% differences in grooming rates in high versus low licking and grooming rat dams, which have well-established consequences for the receiving offspring (22).
Vasopressin and the V1aR have been implicated in social recognition and investigation processes in a variety of species (23). We had preliminary correlative evidence (20) from a random sample of 20 animals from our regular laboratory breeding colony to suggest that the microsatellite would highly predict levels of V1aR in the olfactory bulb, and that V1aR levels in the olfactory bulb would be associated with social behavior. It can be difficult to control the experiment-wise variability in social behavior tasks due to variation contributed by the stimulus animal. Therefore, as an initial test of social behavior, we tested the F1 males for their behavior toward novel social odors. We placed each male in an empty arena for a 2-min acclimation period and then added a small cartridge filled with soiled bedding from the cage of an unrelated female of similar age. Long-allele males displayed a shorter latency to first approach, and higher frequency and longer duration of investigation compared with short-allele males (Fig. 3A, Mann-Whitney, P < 0.05). This effect appears to be specific to social odors, as there were no genotype differences in response to a banana-like odor, amyl-acetate (Fig. 3B, Mann-Whitney, P > 0.05).
To determine whether the behavioral differences observed in the social odor task would translate into differences in prosocial behavior toward another individual, a juvenile prairie vole (PND15 to 18) was placed in the test subject's home cage for 10 min, and the reaction of the test subject was recorded. As with the social odorant test just described, long-allele animals began investigating the juvenile with a reduced latency (Fig. 3C, Mann-Whitney, P < 0.05). There were no genotype differences in huddling or attack behavior, or in the frequency or duration of investigation measures as seen in the social odor task. Regardless, the genetic differences in approach latency have consequences for the probability of social contact in ethologically relevant contexts.
The partner preference test has been the hallmark assay for demonstrating a role for vasopressin and V1aR in pair bonding. We paired the test males with sexually receptive females for an overnight cohabitation. We used a truncated (18-hour) cohabitation period in an effort to increase the variability in the development of partner preferences. In the 3-hour partner preference test, long- and short-allele males spent equal amounts of time in total social contact (student's t test, P > 0.05). However, long-allele males spent more time with their partner compared with the stranger, whereas short allele-males did not (Fig. 3D; two-way ANOVA, preference and genotype as factors, planned t test for partner time versus stranger time P < 0.05 in long-allele males, P > 0.05 in short-allele males). Criteria for a partner preference are met when the test animal spends twice as much social contact time with the partner compared with the stranger. As a group, long-allele males displayed partner preferences (Fig. 3D, χ2 = 7.56, P < 0.05), whereas short-allele animals did not (χ2 = 2.4, P > 0.05), when compared with a random distribution of preferences. Because there were no genotype differences in total social contact time, the genotype differences in partner preference behavior could be due to differences in some aspect of social discrimination of the two females. Perhaps short-allele males need a longer cohabitation period to learn to distinguish partner from stranger.
To control for the possibility of a confounding effect of V1aR on anxiety, we tested the animals in the elevated plus maze and the open field test. Although there were subtle genotype differences in each of these tests (see supporting online material), neither test revealed genotype differences on the classic measures of trait anxiety: namely, time in the open arms of the elevated plus maze and time in the center of the open field test. Therefore, although there may be subtle genotype differences in some measure of emotionality, there were clearly no genotype differences in the classic measures of trait anxiety to sufficiently explain the observed differences in social behavior.
The genotype differences in social behavior are associated with robust genotype differences in V1aR binding in the brains of the F1 males (Fig. 4). Several brain regions showed up to 50% changes in the quantity of V1aR binding. In some brain regions, a long allele conferred a higher level of receptor binding (olfactory bulb, lateral septum), and in other regions a long allele conferred a lower level of receptor binding (amygdala, hypothalamus, posterior cingulate cortex). Microsatellites may generate phenotypic diversity by acting in a cell-type dependent manner, i.e., altering the pattern of gene expression across brain regions, rather than globally altering total levels of expression. This could generate heritable behavioral diversity by altering the neural circuits that are engaged in behavioral contexts that cause the central release of vasopressin.
Several of the brain regions with genotype differences in V1aR levels have previously been implicated in the behaviors described above. For example, V1aR activation in the lateral septum has been associated with both increased rates of pup licking and grooming, as well as increased partner preference formation (21, 24). Our long-allele males showed higher levels of V1aR in the lateral septum, higher rates of pup licking and grooming, and higher levels of partner preference formation. The long-allele males also showed higher levels of V1aR in the external plexiform layer of the olfactory bulb. Lesions of the olfactory bulb inhibit partner preference formation and social approach behaviors in prairie voles (25, 26). Also, pharmacological manipulation of vasopressin systems in both the olfactory bulb and lateral septum modulate social recognition in rats and mice (27–29). Therefore, genotype differences in V1aR binding in the lateral septum and olfactory bulb are likely candidates for serving as the neural substrates underlying the observed genotype differences in social behavior. Variation in the lateral septum and olfactory bulb probably influence the requisite social recognition component of pair bonding.
Previous comparative (11, 12), pharmacological (30), and viral vector gene-transfer (13, 31) experiments have demonstrated that species differences in V1aR levels in the ventral pallidum contribute to species differences in the development of partner preference formation in voles. Interestingly, V1aR levels in the ventral pallidum, which are associated with species differences in social structure, were not associated with genotype within the prairie vole species. Perhaps V1aR binding levels in the ventral pallidum are now genetically fixed by conservation of the 5′ end of the prairie-specific microsatellite, but the hypervariable 3′ end produces variation in other brain regions, such as the lateral septum, olfactory bulb, and hypothalamus, within this species.
Although we have established a clear functional role for the microsatellite in cell culture, the possibility still exists that some of the variation in V1aR binding patterns and behavior may be due to linkage disequilibrium with the microsatellite. However, the in vitro evidence in this manuscript combined with ample experimental evidence from other reports implicating V1aR in social behavior indicate that the microsatellite is likely a causal mechanism of behavioral trait variation through alterations in V1aR distribution.
We theorize that microsatellites in the regulatory regions of the avpr1a gene confer this locus with high levels of evolvability, which in itself may be a target of selection (32). Interestingly, four polymorphic microsatellites surround the human avpr1a gene (33). Two independent reports have indicated modest association of microsatellite alleles at the –3625 bp locus with autism (34, 35), which is a disease of profound social deficit. Considering that variation at this locus may have important implications for our own species-typical social behavior, we compared the publicly available avpr1a gene sequences of chimpanzees [Pan troglodytes (36)] and humans (37) and found that 360 bp in and around this microsatellite locus was deleted in chimpanzees, although the flanking regions were >96% conserved. In contrast, the same locus in the avpr1a gene in the bonobo (Pan paniscus), which is known for its socio-sexual reciprocity and bonding (38), has high homology with the human microsatellite (Fig. 5). Perhaps in primate species, as in vole species, both inter- and intraspecific variation in regulatory microsatellites of the avpr1a gene can give rise to behavioral variation via altered regulation of the distribution of this gene product across individuals.
Supporting Online Material
Materials and Methods