MeCP2 regulates Gdf11, a dosage-sensitive gene critical for neurological function

Curation statements for this article:
  • Curated by eLife

    eLife logo

    eLife assessment

    Rett syndrome is one of the most frequently diagnosed neurodevelopmental conditions. The gene mutated in the condition, Mecp2, encodes for a transcriptional repressor, but genes functioning downstream of Mecp2 have remained difficult to clarify. Here the authors identify an important candidate gene, Growth Differentiation Factor 11 (GDF11) regulated by Mecp2 via epigenetic mechanisms. Further studies in mouse models demonstrate that genetic reduction of Gdf11 ameliorates behavioral deficits of Mecp2 duplication mice, and can function to produce neurobehavioral deficits in mice alone. These findings will be of interest to scientists working in mouse cognition, behavior, neurodevelopment, transcriptional and epigenetics.

This article has been Reviewed by the following groups

Read the full article See related articles


Loss- and gain-of-function of MeCP2 causes Rett syndrome (RTT) and MECP2 duplication syndrome (MDS), respectively. MeCP2 binds methyl-cytosines to finely tune gene expression in the brain, but identifying genes robustly regulated by MeCP2 has been difficult. By integrating multiple transcriptomics datasets, we revealed that MeCP2 finely regulates growth differentiation factor 11 ( Gdf11 ). Gdf11 is down-regulated in RTT mouse models and, conversely, up-regulated in MDS mouse models. Strikingly, genetically normalizing Gdf11 dosage levels improved several behavioral deficits in a mouse model of MDS. Next, we discovered that losing one copy of Gdf11 alone was sufficient to cause multiple neurobehavioral deficits in mice, most notably hyperactivity and decreased learning and memory. This decrease in learning and memory was not due to changes in proliferation or numbers of progenitor cells in the hippocampus. Lastly, loss of one copy of Gdf11 decreased survival in mice, corroborating its putative role in aging. Our data demonstrate that Gdf11 dosage is important for brain function.

Article activity feed

  1. eLife assessment

    Rett syndrome is one of the most frequently diagnosed neurodevelopmental conditions. The gene mutated in the condition, Mecp2, encodes for a transcriptional repressor, but genes functioning downstream of Mecp2 have remained difficult to clarify. Here the authors identify an important candidate gene, Growth Differentiation Factor 11 (GDF11) regulated by Mecp2 via epigenetic mechanisms. Further studies in mouse models demonstrate that genetic reduction of Gdf11 ameliorates behavioral deficits of Mecp2 duplication mice, and can function to produce neurobehavioral deficits in mice alone. These findings will be of interest to scientists working in mouse cognition, behavior, neurodevelopment, transcriptional and epigenetics.

  2. Reviewer #1 (Public Review):

    This study aimed at identifying genes that contribute to the neurological manifestations underlying Rett syndrome and MECP2 duplication syndrome, caused respectively by loss- and gain-of-function of the MECP2 gene. By interrogating murine and human transcriptomics datasets, the authors identified the growth differentiation factor 11 (Gdf11) as a gene whose expression is positively correlated with Mecp2. Through CUT&RUN approaches, the authors also provide initial evidence that Mecp2 regulates Gdf11 expression through epigenetic mechanisms.

    By crossing Mecp2 duplication mice (MECP2-TG1) with mice with monoallelic loss of Gdf11 (Gdf11tm2b/+), the authors succeeded to ameliorate part of the behavioral phenotypes of the MECP2-TG1 mice. The authors also provided compelling evidence that Gdf11 haploinsufficiency is deleterious per se, in keeping with the neurological manifestations documented in individuals with GDF11 loss-of-function variants. The authors also tried to tie the behavioral deficits resulting from Gdf11 haploinsufficiency to deficits in adult hippocampal neurogenesis but observed no differences in neural progenitor pools in the dentate gyrus of Gdf11tm2b/+ mice compared to controls.


    • The identification of Gdf11 as a downstream Mecp2 target derives from an unbiased approach combining multiple transcriptomic datasets. The authors started with the analyses of a dataset from a recent study rectifying Mecp2 expression with antisense oligonucleotide, and then extended to another 20 datasets from human postmortem studies or mouse models.
    • The correlation between Gdf11 and Mecp2 expression has been validated with rigorous mouse genetics approaches, using both Mecp2 null and Mecp2 duplication models.
    • The behavioral batteries used to characterize the neurological phenotypes of the Gdf11tm2b/+ and MECP2-TG1;Gdf11tm2b/+ lines are comprehensive and robust.
    • Sex is properly accounted for, as the tests have been conducted on both males and females and the data for animals of each sex are displayed.
    • The study advances the field in that it identified a potential disease modifier of MECP2-related disorders. Given that rectifying Gdf11 expression alleviates part of the behavioral anomalies in the Mecp2 duplication mouse, this study has implications for therapeutic developments in MECP2-related disorders, especially MECP2 duplication syndrome.
    • Beyond the repercussion for understanding the mechanisms of MECP2-related disorders, the study also provides face validity for the Gdf11tm2b/+ mouse as a model for GDF11 heterozygous loss-of-function variants associated with neurological phenotypes.


    • Gdf11 is critical for skeletal development, and this important information is not considered as a potential confounder or discussed in the manuscript. McPherron et al (1999) have shown that Gdf11-/- mice show skeletal abnormalities, in line with the skeletal phenotypes detected in individuals with monoallelic loss of GDF11. The observation of a truncated tail in Gdf11tm2b/tm2b neonates (Figure S3C) suggests that a skeletal phenotype might be also present in the Gdf11tm2b line. McPherron et al (1999) have also reported milder skeletal anomalies in Gdf11+/- mice, for example the presence of an additional thoracic segment with an associated pair of ribs. This information is missing in the manuscript. The authors did not investigate potential skeletal phenotypes in Gdf11tm2b/+ mice and how they might contribute to some of the behavioral outcomes, for example reduced latency to fall in rotarod.
    • One caveat not discussed in the frame of beneficial effects of Gdf11 reduction in MECP2-TG1 mice is the impact of Gdf11 loss on survival. The authors have shown that Gdf11tm2b/+ have reduced survival, and 30% MECP2-TG1 mice have shown to die between 20 weeks and 1 year of age (Collins et al., Human Molecular Genetics, 2004). Whether MECP2-TG1;Gdf11tm2b/+ mice have a further decrease in longevity compared to MECP2-TG1 mice has not been investigated or discussed. This is important to correctly interpret the health status of the MECP2-TG1;Gdf11tm2b/+ mice undergoing behavioral testing at 12 weeks of age (and the resulting behavioral outcomes). It also has ramifications related to therapeutic development.
    • The manuscript is missing a discussion about the potential cell-specific effects of the Mecp2-mediated regulation of Gdf11. Figure 1B shows that Mecp2 and Gdf11 expression is correlated in all datasets but in inhibitory neurons isolated from postmortem brains of individuals with Rett syndrome. Given the evidence of MECP2-related pathology in both excitatory and inhibitory neurons, this is an important area that remains unaddressed.
    • More caution should be taken when interpreting mouse behavior in relationship to complex human behavioral traits. Expressions like "anxious mice" should be avoided.
    • In open field test, MECP2-TG1 show no differences in distance in the center of the arena over the total distance traveled (Collins et al., Human Molecular Genetics, 2004). MECP2-TG1 mice in this study display reduced number of entries in the center of the arena, and this anomaly is rescued in MECP2-TG1;Gdf11tm2b/+ mice. The relationship between the two measures and how they relate to thigmotaxis is not explained.
    • The fear conditioning data should be interpreted with greater caution. First, during learning training, the percentage of time spent freezing in the second post-tone phase is expected to be higher compared to the time of administration of second tone or the first post-tone phase, unlike what observed in Figures S2B and S3I. Second, both MECP2-TG1 and Gdf11tm2b/+ mice have changes in freezing behavior during the learning phases (Figure S2B, S3I), which affect interpretation of changes in contextual and cue-dependent testing. This integration of data interpretation across the learning and testing phases is missing. Third, the cumulative plots showing the percentage of time spent freezing in testing phases (Figure 2C, 3E, S2B) are not informative with respect with the temporal dynamic of the behavior (over 5 min for the contextual testing and 6 minutes for the cued testing). Fourth, the general hypoactivity of MECP2-TG1 and general hyperactivity in Gdf11tm2b/+ are not considered as potential confounders of the freezing behaviors observed in the fear conditioning paradigms.
    • The statistical considerations are missing information on how data normality was assessed and outliers investigated and treated.

  3. Reviewer #2 (Public Review):

    Mecp2 is the causative gene for RTT and MDS, but the Mecp2 driven pathogenesis is not clearly defined. While Mecp2 is a regulator of gene expression, identifying downstream genes that are robustly regulated by Mecp2 have been a challenge. The authors utilized computational approach to identify Mecp2 regulated genes using previously published differentially expressed genes in hippocampi of MDS mice treated with Mecp2-specific ASO (Shao et. al., 2021). Through this analysis, the authors shortlisted Gdf11, which also validated in an additional 20 transcriptional profiles for Mecp2 perturbed rat, mouse, and human brain samples.

    The transcriptional regulation of Gdf11 by Mecp2 was confirmed using genetic murine models, including Mecp2 -knockout, Gdf11 mutant and Mecp2-tg1.
    Finally, the CUT and RUN analysis showed increased Mecp2 binding upstream of the Gdf11 TSS in Mecp2-tg1 hippocampi, which was lost in Mecp2 knockout hippocampus. Mecp2 loss increases H3K27me3, which suggested Mecp2 prevents transcriptional silencing of Gdf11. While these results provide mechanistic insight into the transcriptional control of Gdf11 by Mecp2, it remains unclear how Mecp2, which is generally a transcriptional suppressor increases Gdf11.

    The author elegantly demonstrates that normalization of Gdf11 levels in Mecp2-tg1 mice crossed with Gdf11 improves several behavioral deficits in the MDS mice model. In contrast, loss of one copy of gdf11 in mice caused neurobehavioral deficits using a battery of behavioral tests, such as elevated plus maze, rotarod, anxiety tests and shock-tone conditioning.

    Finally, the authors show that loss of one copy of gdf11 does not alter proliferation in the adult mouse SGZ or no gross changes in brain anatomy or volume of the dentate gyrus.

    Overall, the authors demonstrate that gdf11 is robustly regulated by mecp2, which coup provide new therapeutic options for Mecp2-related diseases, such as RTT and MDS. As discussed in the paper, additional studies are needed to test whether Gdf11 can rescue behavioral deficit in symptomatic RTT murine models.

  4. Reviewer #3 (Public Review):

    This manuscript provides evidence of the correlation of Gdf11 expression to MeCP2 protein levels, demonstration of phenotypic improvement of mice overexpressing MeCP2 by genetic reduction of Gdf11 levels, and characterization of the phenotypic effects of loss of one copy of Gdf11 on mouse behavior and survival. Significance of the work is driven by the understanding that both gain and loss of MeCP2 function, a transcriptional regulator, causes severe neurodevelopmental disease associated with widespread transcriptional changes. Furthermore, recent work has identified people with neurodevelopmental problems associated with heterozygous mutations in Gdf11. The results are potentially impactful in that the identification of a specific gene target of MeCP2 relevant to pathophysiology and the underlying molecular abnormalities associated could provide insight into future novel therapeutic interventions, as well as the initial characterization of an animal model of a different neurodevelopmental disorder. Furthermore, the work expands the understanding of aspects of the importance of gene dosage in neurodevelopmental disorders and outlines interesting approaches to dissect the underlying genetic network interaction.

    1. Careful bioinformatic evaluation of gene expression changes in MDS mice responsive to anti-sense oligonucleotide treatment that reduces MeCP2 RNA and protein levels to identify a set of genes whose expression was highly correlated with MeCP2 protein levels, restriction to genes of interest based on human predictive algorithms of loss-of function intolerance, followed by analysis of existing transcriptional profiles from multiple species (human, rat, mouse) to restrict focus to Gdf11
    2. Combinatorial use of reporter mouse lines and modern molecular genetic techniques to establish relationship between MeCP2 protein levels and Gdf11 locus binding and regional histone epigenic modifications to support model of direct transcriptional relationship between MeCP2 protein and Gdf11 transcription.
    3. Systematic phenotypic evaluation of the effect of reducing Gdf11 copy number in MDS mice to demonstrate amelioration of some phenotypes observed in MDS mice, as well as evaluation of the effect of Gdf11 copy number reduction on mouse phenotypes to demonstrate mouse phenotypic abnormalities that suggest that this mouse line can be a mouse model of the human disease caused by the heterozygous loss of function mutations in Gdf11

    1. There is a lack of detailed information on the exact composition of the various cohorts of animals used, the age and order of the specific behavioral assessments, and any accounting for the multiple behavioral test performed (to adjust for the multiple statistical tests).
    2. A number of the behaviors that showed improvement with genetic reduction of Gdf11 in MDS mice were behaviors in which the Gdf11 heterozygous mice showed the opposite behavioral abnormality as the MDS mice. For example, total distance in the open field in MDS mice was reduced compared to WT mice, whereas in Gdf11 het mice there is an increased amount of total distance traveled. Similar opposite directions are present in a number of the key phenotypic measures (elevated plus, conditioned fear). The presence of these opposing phenotypic abnormalities between MDS and Gdf11 het mice make interpretation of a partial amelioration of MDS phenotypes by genetic reduction of Gdf11 less clear, as the final "normalization" could reflect an additive effect of opposing phenotypes resulting in a pseudonormalization resulting from aberrant changes in completely independent underlying mechanisms, rather than directly associated with correcting underlying problems directly associated with MDS. Potentially most interesting, and worth commenting upon, are those opposite behavioral abnormalities (such as rotarod) that do not show improvement in the double mutant animals.
    3. The transparency and availability of the entirety of the data contributing to the manuscript (including behavioral data) could be improved by inclusion as supplemental tables or deposition into freely and readily available data repositories or websites (rather than indicating that it is available from corresponding author upon request).