|Home | About | Journals | Submit | Contact Us | Français|
Genomic duplications spanning Xq28 are associated with a spectrum of phenotypes including anxiety and autism. The minimal region shared among affected individuals includes MECP2 and IRAK1, however, it is unclear which gene, when overexpressed, causes anxiety and social behavior deficits. We report that doubling MeCP2 levels causes heightened anxiety and autism-like features in mice, and alters the expression of genes that influence anxiety and social behavior, such as Crh and Oprm1. To test the hypothesis that alterations in these two genes contribute to the heightened anxiety and social behavior deficits, we analyzed MECP2 duplication mice (MECP2-TG1) with reduced Crh and Oprm1 levels. In MECP2-TG1 animals, reducing Crh, or its receptor, Crhr1, suppresses anxiety-like behavior; in contrast, reducing Oprm1 improves abnormal social behavior. These data demonstrate that increased MeCP2 levels impact molecular pathways underlying anxiety and social behavior, and provide novel insight into potential therapies for MECP2-related disorders.
The discovery that loss-of-function mutations in MECP2, the gene encoding Methyl-CpG-Binding Protein 2, cause the neurological disorder Rett syndrome (MIM 312750) led to the identification of other neuropsychiatric phenotypes caused by MECP2 mutations1–6. The most recently identified MECP2-related disorder is a genomic disorder that results from large non-recurrent duplications of chromosome Xq287–12 (MIM 300260). The shared region of overlap among affected individuals spans Interleukin-1 Receptor-Associated Kinase 1 and MECP212–14, suggesting that the overexpression of either one or both of these genes contributes to the features of the disorder. Autism is common in boys with duplications spanning MECP212, and anxiety is a co-morbid condition12. Individuals with triplications spanning MECP2 typically manifest a more severe phenotype7,11.
We previously demonstrated that mice overexpressing MeCP2 at twice the normal levels (MECP2-TG1) on an FVB/N background have motor defects, stereotypies and seizures15. However, whether MECP2-TG1 animals display heightened anxiety remained unanswered because mice on a pure FVB/N background develop premature retinal degeneration, a potential confounder for the interpretation of anxiety-like behavior16. In addition, it is unknown whether social behavior abnormalities are present in MECP2-TG1 mice. We therefore tested the hypothesis that the overexpression of MECP2 alone is sufficient to cause heightened anxiety and abnormal social behavior in mice.
Using F1 hybrid animals to overcome issues associated with pure inbred strains17,18, we found that F1 hybrid MECP2-TG1 animals expressing two-fold the normal levels of MeCP2 and F1 hybrid MECP2-TG3 animals expressing in excess of three-fold the normal levels of MeCP2 displayed anxiety-like behavior in the elevated plus maze and light dark box (Fig. 1a–d). The new data here unequivocally demonstrate that doubling MeCP2 levels indeed causes heightened anxiety-like behavior given the robust phenotype on two different F1 hybrid backgrounds. Thus, the absence of detectable anxiety-like behavior in previously studied MECP2-TG1 animals of a pure FVB/N background in the light-dark box task was likely due to impaired vision in that background15. To investigate social behavior abnormalities, we tested F1 hybrid MECP2-TG1 and -TG3 mice in the partition test for social interest and recognition19–21. We discovered that MECP2-TG1 and -TG3 animals displayed less interest in familiar and novel partner animals (Fig. 2a, b). Although it is possible that the social interaction deficit in MECP2-TG3 animals may be confounded by decreased activity, as evident in an open field (Fig. 1e, f), the findings in the MECP2-TG1 animals, which do not show any motor deficits at the ages tested, are most relevant to the genomic disorder. We therefore chose to further understand the social behavior abnormalities of MECP2-TG1 mice by subjecting these animals to the three chamber test for sociability22,23. MECP2-TG1 animals displayed a deficit in social approach behavior towards novel partner mice, without a deficit in interest towards a novel object (Fig. 2c, d), or a deficit in activity or preference for either chamber (Fig. 2e–h). It is noteworthy that MECP2-TG3 mice had a more severe phenotype in some tests (indicated with plus signs, Fig. 1, ,2),2), reminiscent of the human data in which more severe clinical phenotypes are observed in individuals with triplications spanning MECP28,13,14.
Because either the loss or gain of MeCP2 is known to impact gene expression levels24–26, we next sought to identify gene expression alterations that might contribute to the anxiety and social behavior phenotypes of MECP2-TG mice. We performed microarray experiments using RNA from the amygdala, an anatomical region important for anxiety and social behavior27, of MECP2-TG animals, animals that lack MeCP228 (Mecp2null/y), and the respective wild-type littermates. We focused on gene expression changes altered in opposite directions in the MECP2-TG animals compared with the Mecp2null/y animals, which are likely sensitive to MeCP2 dosage24,25. A total of 1,060 genes were altered in opposite directions in both MeCP2 mouse models compared with wild-type littermates (FDR-corrected p value (q) value < 0.05). Of these, 625 (~60%) genes were up-regulated and 435 (~40%) genes were down-regulated in the presence of excess MeCP2 (Fig. 3a, Supplementary Tables 1, 2). The gene ontology terms associated with these altered genes are listed in Supplementary Table 3. We then compared these 1,060 genes with phenotypic terms relevant to anxiety and social behavior in the Mouse Genome Informatics database and found a significant enrichment of genes whose mutations in mice caused anxiety-related behaviors and/or altered social behaviors (n=32, odds-ratio 1.88, p=0.0016). We selected these 32 genes and an additional 85 genes that have not been associated with anxiety- and/or social behavior-related defects for quantitative real-time reverse transcription-PCR (qRT-PCR) validation studies, and confirmed that 21 of the 32 anxiety- and/or social behavior-related genes (66% validation rate, p value < 0.05), and 58 of the 85 genes not implicated in anxiety and social behavior deficits (68% validation rate, p value < 0.05) are significantly altered in the MECP2-TG animals (Fig. 3b, c; Supplementary Tables 4, 5).
The identification of several anxiety and/or social behavior genes that are sensitive to MeCP2 levels suggests that altered dosage of some of these genes could modulate the behavioral phenotypes of MECP2-TG animals. Studies suggest that a 50% increase or decrease in gene dosage, as observed in humans with copy number variations, is sufficient to cause disease phenotypes29. Nine of the 21 anxiety- and/or social behavior-related genes that were significantly altered based on qRT-PCR analysis had at least a 50% fold change in the MECP2-TG animals (genes outlined in boxes in Supplementary Table 5). We therefore identified these genes as promising candidates to mediate either the heightened anxiety and/or social behavior deficits in this disease model.
To test the hypothesis that an expression change of 50% is sufficient to contribute to the anxiety and social behavior phenotypes of MECP2-TG mice, we focused on Corticotropin-releasing hormone (Crh), which encodes for the neuropeptide CRH, given evidence of increased anxiety-like behavior in mice that overexpress Crh30,31, and on the Opioid receptor, mu (Oprm1), which encodes the G-protein coupled mu-opioid receptor MOR, one subtype of opioid receptor that has been shown to play a role in aspects of emotional and social behavior32–34. Similar to previous studies35–38, we demonstrated that MeCP2 was bound to the promoters of these genes (Supplementary Fig. 1). Because Crh and Oprm1 are up-regulated in the amygdala in MECP2-TG animals, we bred female MECP2-TG1 animals to either male Crh+/− animals39 or male Oprm1+/− animals40 to genetically reduce the levels of these genes in MECP2-TG1 mice. We confirmed that Crh and Oprm1 expression levels were indeed reduced in MECP2-TG1; Crh+/− and MECP2-TG1; Oprm1+/− double mutant animals, respectively (Fig. 3d–h), then tested the behavioral consequences of these genetic reductions.
We found that anxiety-like behavior was subdued in MECP2-TG1 mice lacking one copy of Crh (Fig. 4a, b). In contrast, MECP2-TG1 mice lacking one copy of Oprm1 did not show any significant differences in anxiety-like behavior (Supplementary Fig. 2a, b). Both double mutant animals compared with their respective littermates showed normal exploratory activity in an open field (Fig. 4c, Supplementary Fig. 2c). Furthermore, we found that a 50% reduction in Crh decreased the stress-induced serum corticosterone levels in MECP2-TG1 animals (Fig. 4d). Because Avp levels can modulate anxiety41, we tested Avp expression and found that Avp levels were not significantly altered in the amygdalae of either Crh+/−, MECP2-TG1 or double mutant animals (Fig. 4e, f). Thus, these data suggest that the reduction in anxiety and stress-induced corticosterone levels is specific to the genetic reduction of Crh levels in the MECP2-TG1 animals. To ensure that the suppression of anxiety-like behavior was a direct consequence of modulating Crh levels, we explored this pathway further. CRH mediates its effects on anxiety-related behavior primarily through its predominant receptor in the brain, CRH receptor subtype 1, encoded by the Crhr1 gene42. Based on the anxiolytic effects of CRHR1 antagonists43,44, we reasoned that genetic reduction of Crhr1 or pharmacologic blockade with the CRHR1 antagonist, antalarmin, should also improve the anxiety-like behavior in MECP2-TG1 animals if the anxiety phenotype is mediated by the increase in Crh levels. We bred female MECP2-TG1 mice to male Crhr1+/− mice45 and found that the MECP2-TG1; Crhr1+/−double mutants were less anxious (Fig. 5a, b), and displayed normal exploratory activity in an open field (Fig. 5c). In addition, pharmacologic intervention with antalarmin, a CRH receptor antagonist, demonstrated that an acute dose of 60 mg/kg antalarmin prior to social defeat stress43 significantly reduced anxiety of wild-type animals in the elevated plus maze (Fig. 5d), and improved the anxiety-like behavior of MECP2-TG1 animals in the elevated plus maze and light-dark box (Fig. 5e, f).
Next, we evaluated social behavior in the double mutants and their littermates. Neither the genetic reduction of Crh, nor of Crhr1, affected the social deficit of MECP2-TG1 animals in the partition test (Supplementary Fig. 3a, b). However, MECP2-TG1; Oprm1+/− double mutant animals compared with MECP2-TG1 littermates spent significantly more time investigating familiar and novel partners (Fig. 6a). Furthermore, in the three chamber test, we discovered that MECP2-TG1; Oprm1+/− double mutant mice compared with MECP2-TG1 littermates spent more time investigating novel mice (Fig. 6b). Exploratory activity in the test apparatus, investigation of a novel object, and olfaction ability was comparable among the different test groups (Fig. 6c–e). Together, these data demonstrate that reducing the levels of Oprm1 ameliorates the social behavior phenotype caused by increased MeCP2 dosage.
Based on the smallest region of overlap in individuals with Xq28 duplications, we hypothesized that MECP2 duplication would be sufficient to cause anxiety- and autism-like phenotypes in mice. We discovered that mice of two F1 hybrid backgrounds with either twice or three times the levels of endogenous MeCP2 are anxious and socially impaired. These findings provide strong evidence that the overexpression of MeCP2 contributes to autism and anxiety in this particular genomic disorder. We also demonstrate that the MeCP2-dependent increase in expression of two anxiety- and social behavior-related genes contribute to disease phenotypes. In the MECP2-TG mice, we found that an increase in Crh levels contributed to anxiety-like behavior, whereas an increase in Oprm1 levels only impacted social behavior. Reducing the levels of either of these two genes specifically corrected the respective phenotypes. Thus, these data imply that MeCP2 regulates anxiety and social behavior through distinct pathways.
Of note, we previously showed that Crh is a bona fide MeCP2 target gene35, yet it is intriguing to find its expression up-regulated in both the Mecp2308/y and MECP2-TG mice. The Mecp2308 allele is originally described as a loss-of-function allele46; however, it may also have features of a hypermorphic allele given that it lacks key phospho-serine sites (S421 and S424) that regulate DNA binding47. This is supported by a recent study demonstrating that mice harboring Mecp2 mutations at these sites have neurophysiological, behavioral, and transcriptional changes similar to those observed in MECP2-TG1 mice48. Importantly, Crh levels which are increased in the MECP2-TG and Mecp2308 mice35 yet decreased in Mecp2-null mice24, correlate with heightened and reduced anxiety-like behavior49, respectively. In MECP2-TG animals, heightened anxiety-like behavior can be decreased by modulating the CRH signaling pathway.
Similar to other studies36–38, we show that MeCP2 binds to the promoter of Oprm1; however, we find that Oprm1 expression is up-regulated in the context of increased MeCP2 gene dosage. Furthermore, we demonstrate that increased Oprm1 levels likely underlie the social approach deficits in MECP2-TG1 mice as these deficits are improved by genetically reducing Oprm1 expression almost to wild-type levels. This result is interesting as pharmacological activation of MOR is associated with increased sociability34, underscoring a difference between the effects of direct MOR activation versus up-regulation of Oprm1 expression. Indeed, there are no reports regarding the effects of Oprm1 overexpression in rodents, thus it is challenging to compare our findings with existing studies related to MOR activity on social behavior. Our data highlight the complexity of the regulation of social behavior by Oprm1, and call for further exploration of the distinct effects of Oprm1 overexpression and MOR activation on such behavior.
In sum, our data highlight the importance of MeCP2 in governing genetic pathways related to normal anxiety and social behavior. It is conceivable that the phenotypes that arise due to an excess or loss of MeCP2 in mice and humans primarily result from misregulation of only a subset of genes that are altered by at least 50%, and that focusing on such genes could identify therapies that ameliorate the respective symptoms. These data provide a potential framework for investigating treatments of anxiety and social behavior in individuals carrying MECP2 duplications.
Mice were maintained on a 12 h light:12 h dark cycle with standard mouse chow and water ad libitum. For experiments related to the phenotypic analysis of anxiety and social behavior in the MECP2-TG lines, we generated F1 hybrid animals by mating female MECP2-TG1 and -TG3 mice of a pure FVB/N background to either wild-type male 129S6/SvEv (Taconic Farms, Inc., USA) or C57BL/6 mice (Jackson Laboratories, USA). MECP2-TG mice harbored a ~99 kb P1 artificial chromosome (PAC 671D9) containing only the human MECP2 genomic locus15. MECP2-TG1 and -TG3 adult male mice on both F1 hybrid genetic backgrounds, (FVB/N × 129S6/SvEv)F1 and (FVB/N × C57BL/6)F1, were viable through at least 20 weeks of age; therefore, it was feasible to test the behavior of these animals during adulthood. We tested only male mice, as our clinical study identified autism and heightened anxiety in boys with MECP2 duplication syndrome12. For studies related to the genetic interaction of MeCP2 with either Crh or Crhr1, we generated F1 hybrid mice by mating female MECP2-TG1 mice of a pure FVB/N background to either male Crh+/− or male Crhr1+/− animals39,45. Crh+/− animals, originally generated by targeted disruption of the pre-pro-Crh coding region39, were purchased from Jackson Laboratories on a pure C57BL/6 background. Crhr1+/− animals, originally generated by replacing exons 5 through 8 of Crhr1 locus with a PGK-neo cassette45, were also purchased from Jackson Laboratories on a pure C57BL/6 background. Both Crh and Crhr1 lines were backcrossed to 129S6/SvEv for seven generations prior to breeding with MECP2-TG1 females to obtain F1 hybrid progeny for the genetic interaction experiments. For studies related to the genetic interaction of MeCP2 with Oprm1, we generated F1 hybrid mice by mating female MECP2-TG1 mice of a pure FVB/N background to male Oprm1+/− animals that were maintained on a pure C57BL/6 background40. Oprm1+/− animals, originally generated by insertion of a PGK-neo cassette into exon 240, were purchased from Jackson Laboratories on a pure C57BL/6 background. For studies related to the pharmacological blockade of CRHR1 with antalarmin hydrochloride, we used (FVB/N × 129S6/SvEv)F1 animals that were generated by mating MECP2-TG1 female mice of a pure FVB/N background with wild-type 129S6/SvEv male mice (Taconic Farms Inc., USA). The subsequent wild-type male F1 progeny from this mating scheme were used in testing the effect of acute antalarmin treatment after social defeat stress43, whereas MECP2-TG1 F1 hybrid male progeny from this mating scheme were used in testing the effect of acute antalarmin treatment. In most cases, animals were housed 4 animals per cage with an equal balance of genotypes per cage. Animals used for the pharmacological studies were housed 4–5 animals per cage, and MECP2-TG1 and wild-type littermate F1 hybrid animals in these studies were housed in separate cages. All research and animal care procedures were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee.
Four (FVB/N × 129S6/SvEv)F1 MECP2-TG1 test colonies were generated for behavioral testing. Test colonies included both MECP2-TG1 animals and their wild-type littermates. The first MECP2-TG1 test colony was used to measure activity in an open field (8–9 weeks of life), to test anxiety-like behavior in the light-dark box (9–10 weeks of life), and to test social interaction in the partition test (12–13 weeks of life). The second MECP2-TG1 test colony was used to test anxiety-like behavior in the elevated plus maze (9–10 weeks of life). The third MECP2-TG1 test colony was used to test social interaction in the three chamber test (12 weeks of life). The fourth MECP2-TG1 test colony (16 – 20 weeks of life) was used for the pharmacological studies related to antalarmin. One (FVB/N × C57BL/6)F1 MECP2-TG1 test colony was generated to confirm the anxiety-like behavior and social behavior deficits; anxiety and social behavior tests were performed at equivalent timepoints listed for (FVB/N × 129S6/SvEv)F1 MECP2-TG1 test cohorts.
Two (FVB/N × 129S6/SvEv)F1 MECP2-TG3 test colonies were generated for behavioral testing. Test colonies included both MECP2-TG3 animals and their wild-type littermates. The first MECP2-TG3 test colony was used to measure activity in an open field (8 weeks of life), to test anxiety-like behavior in the light-dark box (8–9 weeks of life), and to test social interaction in the partition test (10–12 weeks of life). The second MECP2-TG3 test colony was used to test anxiety-like behavior in the elevated plus maze (8–9 weeks of life). One F1 hybrid FVB/N × C57BL/6 MECP2-TG3 test colony was generated to confirm the anxiety-like behavior and social behavior deficits; anxiety and social behavior tests were performed at equivalent timepoints listed for (FVB/N × 129S6/SvEv)F1 MECP2-TG3 test cohorts.
One test colony each was generated to test the effect of either Crh or Crhr1 haploinsufficiency in MECP2-TG1 animals in an open field, elevated plus maze, light dark box and partition test (~4–5 months of life). Test colonies included all possible animal genotypes; the parental genotypes (MECP2-TG1 and either Crh+/− or Crhr1+/−), wild-type littermates, and double mutant (MECP2-TG1; Crh+/− or MECP2-TG1; Crhr1+/−) animals.
One test colony was generated to test the effect of Oprm1 haploinsufficiency in MECP2-TG1 animals in an open field, elevated plus maze, light dark box, partition test, and three chamber test (~2–3 months of life). Test colonies included all possible animal genotypes; the parental genotypes (MECP2-TG1 and Oprm1+/−), wild-type littermates, and double mutant (MECP2-TG1; Oprm1+/−) animals.
Methods for behavioral tests and statistical analysis of behavioral data are presented in the Supplementary Note.
Total RNA was extracted from the amygdalae of Mecp2null/y animals28 and their wild-type littermates (6 weeks of life, n=4 per genotype), and from the amygdalae of male MECP2-TG3 and their wild-type littermates (6 weeks of life, n=5 per genotype) using Trizol (Invitrogen, USA). Microarray experiments were performed as previously described using the Affymetrix Mouse Exon 1.0 ST microarray24,25. Probe level data were normalized using the RMA method and the exonmap package. The resulting probe set expression summaries were then annotated to each gene and exon using the information from the Affymetrix using the na27 build of annotation for the MoEx 1 array (see MoEx-1_0-st-v1.na27.AFFX_README.NetAffx-CSV-Files.txt). Exon coordinate annotation for each gene were then obtained from the UCSC genome browser using the mm9 build of the mouse genome; probe sets were assigned to exons using the combined information from the Affymetrix annotation file and data from the UCSC browser; data on a total of 24,277 distinct gene symbols were considered. Once the normalized data were assigned to exons, a linear model was constructed extending our previous work24,25. This model considered RMA expression summaries at the level of probe sets with model terms for probe set, gene, genotype, genotype-exon interaction, and individual animal effects. The model was fit using R and the base method for analysis of variance. The error estimate and coefficient parameters from this model were then used to construct three linear contrasts: one contrast for the difference between WT and NULL mice using only the appropriate matched WT control; another contrast for the difference between the TG mice and their WT control; and a final contrast considered the sum of these two differences. The 2-sided p values for these linear contrasts were then converted to q-values using the p. adjust method in the base R installation. Values where the mean gene-level fold change difference exceeded 0.2 with an FDR q-value less than 0.05 were selected as differentially expressed, as well as genes where the total difference exceeded 0.4 with a q-value less than 0.05. Content analyses for the GeneOntology were performed as previously described24.
Methods for MGI database phenotype analysis are presented in the Supplementary Note.
Methods for ChIP-PCR are presented in the Supplementary Note.
For validation of expression profiling data, total RNA was extracted from the amygdalae of MECP2-TG3 animals and their wild-type littermates (n=4 of each genotype) using Trizol (Invitrogen, USA). For testing the effect of reducing either Crh or Oprm1 levels in MECP2-TG1 mice, and to check Avp expression in MECP2-TG1 × Crh+/− animals, total RNA was extracted from the hypothalami and amygdalae of the double mutant animals and their respective littermates (n=3–6 of each genotype). Three μg of RNA was used to synthesize cDNA according to the manufacturer’s protocol (Invitrogen, USA). QRT-PCR was performed as previously described20 using PerfeCTa qPCR FastMix (Quanta Biosciences, Inc., USA). Primers were designed to amplify a single gene product using an online primer design tool (Primer-BLAST, www.ncbi.nlm.nih.gov) or were obtained from an online public resource (PrimerBank50). Sequences are available upon request. Expression levels were normalized to S16 and data are represented as fold change relative to wild-type levels. Significant differences were determined using paired T tests.
Crh ISH probe was PCR-amplified from wild-type mouse brain cDNA, followed by digoxigenin labeling as previously described35. ISH was performed on brain tissue obtained from wild-type, Crh+/−, MECP2-TG1 and MECP2-TG1; Crh+/− animals (~4–5 months of life), and ISH signal intensity was quantified as previously described21,35,51.
Serum corticosterone levels (n=3–6 animals of each genotype for each measurement) were determined as previously described35. Briefly, basal corticosterone levels were obtained from animals that were undisturbed for at least 12 hours. Stress-induced corticosterone levels were obtained from animals restrained in 50 mL conical tubes for 30 minutes. For both test measurements, animals were rapidly decapitated after the indicated time period. Trunk blood was collected and placed in 1.5 mL conical tubes on ice for at least 30 minutes. Blood was centrifuged at max speed for 10 minutes. Serum was collected and analyzed using an enzyme-linked immunoassay (IDS Inc., USA).
RS, CMB and CMM performed experiments; RS and CMM analyzed the data. CS performed statistical analyses of microarray data. BM provided intellectual contribution to and initiated CRH genetic interaction studies. RS and HZ designed experiments, reviewed the data and wrote the manuscript. All authors reviewed the manuscript in its preparation. We thank Drs. Melissa Ramocki and Jeffrey Neul for critical reading of the manuscript, Drs. Corinne Spencer, Richard Paylor, and Paolo Moretti for advice on neurobehavioral tasks, the BCM Microarray core and the BCM Intellectual and Developmental Disabilities Research Center (IDDRC) In situ Hybridization and Neurobehavioral Cores for use of facilities. This work was funded by the U.S. National Institutes of Health Grants NS043124 (RS), NS073317 (CMM), NS057819 (HZ), and HD24064 (HZ, BCM IDDRC), Autism Speaks (Predoctoral Fellowship to RS), the Carl C. Anderson, Sr. and Marie Jo Anderson Charitable Foundation, the Simons Foundation, and the Rett Syndrome Research Trust (HZ). Huda Zoghbi is a Howard Hughes Medical Institute investigator.