The serotonin_1A receptor gene as a genetic and prenatal maternal environmental factor in anxiety

WT Offspring of Heterozygote 5-HT_1AR-Deficient Mothers Exhibit Anxiety-Like Behavior.

We examined the effect of 5-HT_1AR inactivation, both as an offspring genotype and maternal environmental factor, on anxiety and stress-related behaviors in adult Swiss Webster (SW) mice. In the elevated plus maze (EPM) paradigm, mice freely explore the safe closed arms and the stressful open arms of the apparatus, and reduced time spent or distance traveled in the open arms is considered to indicate increased anxiety-related behavior (8). WT offspring of heterozygote knockout (H) mothers [WT(H)] exhibited increased anxiety-like behavior in the EPM as compared with WT offspring of WT mothers [WT(WT)] (Fig.1 A and C) and their behavior was similar to that of homozygote KO offspring of H mothers [KO(H)] and KO offspring of KO mothers [KO(KO)], indicating an unexpected role for maternal/parental 5-HT_1AR deficiency in offspring anxiety. Activity of WT(H) mice in the closed arm of the EPM was not significantly different from that of WT(WT) mice, indicating that the maternal/parental receptor deficit does not affect safe-area exploration (Fig. S1). Quantitative RT-PCR (qRT-PCR) experiments showed that WT(H) mice had normal 5-HT_1AR mRNA levels in the ventral granule cell layer (GCL) and CA1 region of the hippocampus (Fig. S2), a region relevant to anxiety (9), indicating that the increased anxiety-like behavior of WT(H) mice was not caused by paramutation.

To determine if the pre- and/or postnatal KO maternal environment is sufficient to produce anxiety in WT offspring, postnatal and embryonic cross-fostering were performed. Cross-fostering of WT pups to KO mothers at birth [WT(WT/KO)] resulted in only a partial anxiety phenotype, but WT mice implanted as 1-day old (E1) embryos into KO mice and then cross-fostered at birth by WT mothers [WT(KO/WT)] exhibited full anxiety (Fig. 1 B and D). In the embryonic cross-fostering experiments, fathers were not present during the pre- and postnatal period, resulting in an increase in the baseline level of EPM anxiety; nevertheless, the effect of the prenatal KO maternal environment was still apparent. In both EPM measures (time and distance), the extent of anxiety-like behavior in these WT(KO/WT) mice was comparable with that of WT(KO/KO) mice raised in the pre/postnatal KO environment, indicating that the prenatal 5-HT_1AR KO maternal environment is necessary and sufficient to elicit a full anxiety-like phenotype in this paradigm. Closed-arm activity was normal in cross-fostered mice except when both the mother and the offspring were receptor deficient [KO(KO/KO) mice displayed reduced safe-area activity], indicating an interaction between the offspring and maternal genotypes (Fig. S1B). This notion is supported by the reduced closed-arm exploration of KO(H) and KO(KO) but not WT(H) mice (Fig. S1A).

Next, we examined immobility time in the Porsolt forced swim test (FST). Although reduced immobility time typically is considered an indicator of an antidepressant-like response, in genetically modified mice, especially in those with increased anxiety-like behavior, it probably is more appropriate to interpret reduced immobility time as increased stress-reactivity or lack of appropriate coping (7, 10). Increased stress-reactivity of WT(H) animals in the FST indicated a maternal genotype effect (Fig. 1E), but only the combination of postnatal and prenatal KO maternal environment in WT(KO/KO) mice produced the phenotype (Fig. 1F). Therefore, this behavior is established by the KO maternal environment through the pre/postnatal period.

Open field (OF) is another test of anxiety-like behavior, and, as in the EPM, reduced time spent and distance traveled in the center area of the field is indicative of an anxiety-like phenotype (7). However the OF is considered less stressful than the EPM (11), and quantitative trait locus (QTL) studies identified both shared and OF-specific loci (12). Anxiety in the OF was completely offspring-genotype dependent, and neither the post- nor the prenatal maternal environment, and not even their combination, increased anxiety (Fig. 1 G and H).

Delayed Ventral Dentate Gyrus Development Correlates with the Maternally Transmitted Behaviors.

Next we investigated the possible cellular basis of EPM anxiety and increased escape-directed behavior in the FST transmitted nongenetically from 5-HT_1AR-deficient mothers to their offspring during pre/postnatal life. Because gestational stress, in addition to inducing permanent emotional deficits (13), also affects the development of the DG (14, 15), among other brain regions, we focused on the hippocampus, in particular on its ventral subdivision implicated in anxiety (9, 16, 17). Proliferation in the GCL of the DG was assessed by BrdU labeling of S-phase cells by unbiased stereology based on the optical fractionator principle; GCL volume was measured by the Cavalieri method. Because the level of proliferation is high in the developing DG, a single BrdU injection followed by a 3-h labeling period was sufficient to assess proliferation. Assessed at a total of eight anatomical levels through the dorsal and ventral DG, proliferation between postnatal days (P) 1 and 7 was found to be unaltered in offspring with H or KO mothers (Fig. S3). However, ventral but not dorsal GCL volume was significantly increased in WT(H), KO(H), and KO(KO) between P1 and P7 (Fig. 2A). Because cell density of DAPI-labeled cells was unchanged across the groups, the increased volume is related to an overall increase in cell number in WT(H), KO(H), and KO(KO) GCL. In the same P1–7 time frame, the GCL volume was relatively unchanged in WT(WT) mice, consistent with naturally occurring cell death (18). The relatively larger GCL volume in offspring with mutant mothers was normalized by P28.

Young neurons prominently express the Prospero homeobox protein, Prox1, which makes them selectively detectable at low antibody concentrations (19, 20). Young neurons showed a gradual increase in density from the “oldest” suprapyramidal two-fifths toward the intermediate two-fifths and the still-developing infrapyramidal one-fifth of P7 GCL (21) in WT(WT) mice, whereas the distribution of these young neurons in WT(H), KO(H), and KO(KO) mice was more consistent along the length of the GCL (Fig. 2 B–D). This distribution indicated that in the WT(WT) P7 GCL, the earliest-born cells already had passed the young-neuron stage, whereas in animals with mutant mothers, young neurons were still abundant. GCL volume positively and negatively correlated with Prox1 expression in the suprapyramidal (trend) and intermediate supra-/infrapyramidal GCL, respectively (Fig. 2E), suggesting that the immaturity and increased volume are related. Overall, the maternal 5-HT_1AR genotype minimally affects the proliferation of precursors and instead seems to target the differentiation/maturation of DG granule cells.

Genetic Inactivation of p16^ink4a Phenocopies Both the Cellular and Behavioral Phenotypes of the WT(H) Offspring.

The switch from proliferating to postmitotic cells and subsequent differentiation require the down-regulation of cyclin D and associated cyclin-dependent kinases (CDKs). The CDK inhibitor p16^ink4a, which is important in differentiation (22), was expressed in microdissected GCL at P3 and P7 (Fig. S4B). Incidentally, expression of p16^ink4a in neuronal precursors is suppressed by Bmi1 (23, 24), and therefore genetic deletion of p16^ink4a is not expected to result in increased proliferation in the developing DG. Indeed, the number of BrdU-labeled cells in P3 and P7 GCL of p16^ink4a KO mice was similar to that measured in WT mice (Fig. S5). However, GCL volume in p16^ink4a KO mice was increased at both P3 and P7 (Fig. 3A), similar to levels seen in mice exposed to the 5-HT_1AR-deficient maternal environment (Fig. 2A). Inactivation of p16^ink4a also phenocopied the prenatally determined EPM anxiety but only partly phenocopied the pre/postnatally specified stress response seen in WT(H) mice (Fig. 2 and Fig. 3 B and C). These data show that perturbations in early GCL development, both in p16^ink4a KO mice and in mice born to 5-HT_1AR-deficient mothers, correlate with increased adult anxiety-like behavior in the EPM. Because p16^ink4a inactivation was not restricted to the GCL or to the prenatal period, we cannot exclude the possibility that the p16^ink4a KO phenotype is caused by a nonhippocampal mechanism. Nevertheless, the finding that p16^ink4a KO mimics both the cellular and behavioral phenotypes of the receptor-deficient maternal genotype effect and specifically the prenatally determined EPM behavior suggests the involvement of early DG development in establishing the level of EPM anxiety.

Maternal 5-HT_1AR Genotype Effect Is Genetic-Background Dependent.

Although outbred strains such as the SW tend to respond to adverse or stimulating environments, inbred lines, in particular the C57BL6 (B6), are resilient to these effects (25, 26). No anxiety in the EPM and enhanced stress response in the FST developed in response to the 5-HT_1AR-deficient maternal environment when the strain background was switched from SW to B6 (>99.9% isogeneity). Instead, all anxiety manifestations were caused by the offspring's own receptor deficiency (Fig. 4), as reported previously by others with B6 and mixed B6 strains (27, 28). Taken together, the data indicate that maternal receptor deficit has no appreciable effect on offspring behavior in the B6 and B6-derived strains and are consistent with the reported resilience of B6 mice to environmental adversity (25, 26, 29).

Animals.

5-HT_1AR KO mice originally were generated on the 129SvEv background (7) and were backcrossed to the SW and B6 (Taconic) backgrounds >10 times. Both littermate (crossing heterozygote parents) and nonlittermate (breeding WT and KO mice separately) WT and KO offspring from at least three litters were generated. At least three litters per group were used.

Ink4a KO mice on the FVB/N background (congenic, FVB.129-Cdkn2a^tm2.1Rdp) were obtained from the International Mouse Strain Resource (National Cancer Institute). In the Ink4a experiments the SW and FVB/N strains were intercrossed. Genotyping was performed as previously described (7, 23).

For postnatal cross-fostering, pups were swapped between mothers of the same or different genotypes within 24 h of birth. For combined embryonic and postnatal cross-fostering, SW WT embryos (E0.5–E2.5) were implanted into pseudopregnant (7- to 9-week-old) SW 5HT_1AR WT(WT) or KO(KO) animals. Again, pups were cross-fostered within 24 h of birth to mothers of the same or different genotypes.

In all behavioral and immunohistochemical experiments, two separate cohorts of animals from a total of at least three different litters were used (except in the embryonic cross-fostering experiments with two or three litters).

Behavior.

Behavioral experiments were performed as described previously (7) and in SI Materials and Methods.

BrdU Immunohistochemistry and Stereological Analysis.

Mice received one s.c. 100-mg/kg injection of 10 mg/mL BrdU (Sigma). Perfusion, sectioning, BrdU immunohistochemistry, and stereological counting of positive cells were performed as described in a previously published protocol. Detailed procedures are described in SI Materials and Methods.

Statistical Analysis.

Two- and one-way ANOVAs with LSD post hoc tests were used for the behavioral experiments as described in the text and figure legends. Cell-count data were analyzed by two-way ANOVA with group and section as independent variables.