1. Chromosomes and Gene Expression
  2. Neuroscience
Download icon

Losing Dnmt3a dependent methylation in inhibitory neurons impairs neural function by a mechanism impacting Rett syndrome

  1. Laura A Lavery
  2. Kerstin Ure
  3. Ying-Wooi Wan
  4. Chongyuan Luo
  5. Alexander J Trostle
  6. Wei Wang
  7. Haijing Jin
  8. Joanna Lopez
  9. Jacinta Lucero
  10. Mark A Durham
  11. Rosa Castanon
  12. Joseph R Nery
  13. Zhandong Liu
  14. Margaret Goodell
  15. Joseph R Ecker
  16. M Margarita Behrens
  17. Huda Y Zoghbi  Is a corresponding author
  1. Jan and Dan Duncan Neurological Research Institute at Texas Children’s Hospital, United States
  2. Department of Molecular and Human Genetics, Baylor College of Medicine, United States
  3. Genomic Analysis Laboratory, The Salk Institute for Biological Studies, United States
  4. Howard Hughes Medical Institute, The Salk Institute for Biological Studies, United States
  5. Department of Pediatrics, Baylor College of Medicine, United States
  6. Graduate Program in Quantitative and Computational Biosciences, Baylor College of Medicine, United States
  7. Computational Neurobiology Laboratory, The Salk Institute for Biological Studies, United States
  8. Program in Developmental Biology, Baylor College of Medicine, United States
  9. Medical Scientist Training Program, Baylor College of Medicine, United States
  10. Center for Cell and Gene Therapy, Baylor College of Medicine, United States
  11. Stem Cells and Regenerative Medicine Center, Baylor College of Medicine, United States
  12. Department Molecular and Cellular Biology, Baylor College of Medicine, United States
  13. Department of Psychiatry, University of California San Diego, United States
  14. Department of Neuroscience, Baylor College of Medicine, United States
  15. Howard Hughes Medical Institute, Baylor College of Medicine, United States
Research Article
Cite this article as: eLife 2020;9:e52981 doi: 10.7554/eLife.52981
4 figures, 2 tables, 3 data sets and 4 additional files

Figures

Figure 1 with 4 supplements
Dnmt3a cKO and Mecp2 cKO mice show overlapping as well as distinct neurological deficits.

(A) Mice that lack Dnmt3a or MeCP2 in inhibitory neurons present with hindlimb spasticity. (B) Obsessive grooming is increased in both cKO models. (C) Nest building, (D) grip strength, (E) open field, (F) fear conditioning tests revealed impairments in both cKO lines. (G) Self-injury in Dnmt3a cKO and Mecp2 cKO mice necessitated humane euthanasia. (H) Example traces of miniature inhibitory postsynaptic currents (mIPSCs) recorded in the dorsal striatum. Both cKO models show similar alterations in (I) amplitude. (J) Weekly body weight records for Dnmt3a cKO and Mecp2 cKO mice showed only Dnmt3a cKO mice (here separated by sex- see Materials and methods) were runted. (K) Dnmt3a cKO and Mecp2 cKO mice showed opposite alterations in acoustic startle response. (L) Only Mecp2 cKO mice displayed impairment on the parallel rod. M) Dnmt3a cKO mice had to undergo earlier euthanasia than Mecp2 cKO mice due to the severity of their self-lesioning. n = 11–52 (behavior), n = 5–9 mice per genotype with 24–50 neurons total (electrophysiology). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. See Supplementary file 1 for full statistics.

Figure 1—figure supplement 1
Protein levels of Dnmt3a and MeCP2 are reduced in inhibitory neurons of conditional knockout mice.

(A) Western blot of half brain hemisphere from Dnmt3a or Mecp2 cKO mice demonstrating loss of each protein. (B) Immunofluorescence (IF) images of WT, Dnmt3a cKO or Mecp2 cKO mice probing for Dnmt3a or MeCP2 (red), the Sun1-sfGFP-myc-PA fusion protein that marks the nuclear envelope and is dependent on Cre expression (green), and DAPI to mark genomic DNA (blue). n = 3 mice per genotype (western), n = 3 mice (IF) representative images for two brain regions shown.

Figure 1—figure supplement 2
Supplemental behavioral and physiological data for Dnmt3a cKO and Mecp2 cKO mice.

(A) Both mouse lines showed decreased rearing in the open field test. (B) Dnmt3a cKO mice showed impaired fear learning, whereas Mecp2 cKO mice did not differ from control mice. (C) Cue memory was normal in both cKO mice. (D–F) Tests for anxiety-like behaviors (open field, light dark or elevated plus maze, respectively). (G) Hot plate and (H) tail flick testing for nociceptive pain in both cKO mice. (I) Rotarod test for motor learning and coordination and (J) the partition test for social interaction. (K–N) Frequency, charge, rise and decay measures from mIPSCs from the striatum. (O) Only Mecp2 cKO had a trend for increased pre-pulse inhibition. n = 11–50 per genotype (behavior), n = 5–9 mice per genotype with 24–50 neurons total (electrophysiology), *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. See Supplementary file 1 for full numbers and statistics.

Figure 1—figure supplement 3
X-gal staining of LacZ+/-;Slc32a1-Cre+/- and control E14.5 embryos shows widespread expression of Cre transgene in the nervous system and select expression in peripheral tissues.

(A) X-gal staining of LacZ+/- control and (B) LacZ+/-;Slc32a1-Cre+/- reporter mice demonstrating specific staining of tissues with expression of Cre recombinase under the Slc32a1 promoter. (C–H) Images showing positive X-gal staining in nervous system tissues consistent across three biological replicates: (C) brain (D) spinal cord (E) dorsal root ganglia (F) inner ear (G) olfactory epithelium (H) and eye. (I–L) Images showing positive X-gal staining in non-neural tissues consistent across three biological replicates: (I) kidney (J) heart (K) skin (only in select regions: surrounding genital tubercle and tail and area on the face/nose) (L) ureter and bladder. Additional positive X-gal staining was noted in the (M) testis in the only embryo where the testis were visible, as well as rare, scattered cells of the intestine and pancreas (not shown). The observation of expression of Slc32a1-Cre in the testis is consistent with our noted germline recombination of floxed genes if F1, Dnmt3aflox/+;Slc32a1-Cre+/- males are used to generate F2 offspring; a breeding scheme we deliberately avoided to generate our experimental mice (see Materials and methods for more details).

Figure 1—video 1
Forepaw stereotypies are apparent in Dnmt3a cKO mice.

Video shows an example of forepaw stereotypies in a Dnmt3a cKO male mouse at 8 week of age.

Figure 2 with 1 supplement
Methylation in striatal inhibitory neurons remains stable without MeCP2, but without Dnmt3a there is global loss of mCH and some loss of mCG.

(A) Schematic of INTACT method used to isolate inhibitory neurons from the striata of WT, Dnmt3a cKO, or Mecp2 cKO mice that also conditionally express the INTACT allele. (B) Spearman correlation of methylation profiles from inhibitory neurons sorted from WT vs. Mecp2 cKO striatum, showing that methylation is stable in the absence of MeCP2. (C) Bar graph showing the global mCH level in each biological replicate (left). Example genes from DNA methylome sequencing tracks showing mCH signal in two biological replicates per genotype (right). (D) Bar graph showing the global mCG level in each biological replicate (left). Example genes from DNA methylome sequencing tracks showing mCG signal (right). Red bars indicate DMRs. (E) Genome wide correlation between mCH and mCG in wild-type mice showing poor correlation. (F) Genome wide correlation of Dnmt3a dependent mCG and mCH showing good correlation to indicate that mCH and mCG written by Dnmt3a are coupled. Correlation values for E and F are Pearson correlations designated as ‘rho’. n = 2 mice per genotype (see Materials and methods for specific genotype information). See Supplementary file 1 for replicate statistics.

Figure 2—figure supplement 1
Plot of the difference in genome wide mCH versus mCG methylation between the WT and MeCP2 cKO mice.

Person correlation designated as rho. Consistent with stable methylation in the absence of MeCP2, the difference centers at 0.

Figure 3 with 2 supplements
RNA-seq from sorted striatal inhibitory neurons of WT and cKO mice reveal MeCP2 is a restricted reader for Dnmt3a dependent methylation.

(A) RNA-seq data of sorted WT vs Dnmt3a cKO or Mecp2 cKO striatal inhibitory neurons that also express the INTACT allele. Red dots represent genes with altered expression in the knockout cells (padj <0.01). (B) Differentially expressed genes (DEGs) that overlap between knockout models. Inhibitory neurons that lack MeCP2 share ~40% of the same DEGs as the same neurons that lack Dnmt3a. Only ~12% of DEGs in inhibitory neurons that lack Dnmt3a are shared with the same neurons that lack MeCP2. (C) Plot of log2 fold-change for DEGs in Dnmt3a cKO and Mecp2 cKO models. DEGs that are only significantly misregulated in the Dnmt3a cKO model, only significantly misregulated in the Mecp2 cKO model, or common to both models are colored in blue, green, or orange, respectively. The plot shows that the DEGs common to both models have similar degree and direction of change.

Figure 3—figure supplement 1
The percentage of DEGs that overlap between cKO mouse models broken down as a function of (A) p-value, (B) direction of change (down- or up-regulated), (C) gene length.
Figure 3—figure supplement 2
The percentage of DEGs that overlap between parvalbumin (PV) and vasoactive intestinal polypeptide (VIP) neurons in the mouse cortex from Dnmt3a cKO (Nestin-Cre) and Mecp2 KO mouse models.

The data are a re-analysis of single-nuclear RNA sequencing (Stroud et al., 2017). The percentages are shown as all genes and genes broken down by direction of change (down- or up-regulated) for (A) PV and (B) VIP neurons. Consistent with our data there are few DEGs significantly misregulated in Dnmt3a cKO mice that are also significantly misregulated in the same neurons that lack MeCP2. Notably, our data show a significantly higher overlap of genes (40%) misregualted in Mecp2 cKO that are also misregulated in the Dnmt3a cKO.

Figure 4 with 2 supplements
Integrative gene expression and methylation analysis shows mCH and mCG loss contribute to Dnmt3a cKO DEGs, and reveals a strong mCH contribution to RTT.

(A) Gene-body mCH levels in different categories of DEGs in WT mice are plotted, highlighting that misregulated genes have higher mCH than genes that are unchanged and the significant differences between mCH levels on up- and down-regulated genes. (B) Gene-body mCG levels in different categories of DEGs in WT mice are plotted, demonstrating that misregulated genes (with the exception of MeCP2 down-regulated genes) have higher mCG than genes that are unchanged and the significant differences between mCG levels on up- and down-regulated genes. (C) Running average plot of log2fold change in gene expression for genes significantly misregulated only in the Dnmt3a cKO model versus the change in mCH methylation observed in the Dnmt3a cKO model (‘Dnmt3a dependent mCH methylation’) (left). Running average plot of log2fold change in gene expression for these same genes versus the change in mCG methylation observed in the Dnmt3a cKO (‘Dnmt3a dependent mCG methylation’) (right). (D) Running average plots of log2fold change in gene expression in the Dnmt3a cKO model for genes commonly misregulated in both cKO models versus Dnmt3a dependent mCH (right) and mCG (left). (E) Running average plots of log2fold change in gene expression in the MeCP2 cKO for genes commonly misregulated in both cKO models versus Dnmt3a dependent mCH (right) and mCG (left). (F) Running average plots of log2fold change in gene expression for genes that are only significantly misregulated in the MeCP2 cKO model. (G–J) R2 values from analysis in panels C-F shown as blue, orange or green dots, respectively, plotted over 1000 random repetitions of the analysis with each repetition containing the same number of non-DEGs (padj >0.01). The results of random repetitions are shown as gray dots. All plots were made with DEGs padj <0.01. n = 2 mice per genotype for methylation data. n = 4 mice per genotype (RNA-seq). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 4—figure supplement 1
Methylation levels on categories of DEG genes in WT, Dnmt3a cKO, and Mecp2 cKO mice with statistical comparisons.

(A) Averaged gene-body mCH levels (average of two replicates) in different categories of DEGs in WT mice, highlighting the direction of difference in mCH levels between up- and down-regulated DEGs. The box plots within each segment highlights the median of the distribution, while the red point represents the mean of the distribution. (B) Averaged gene-body mCG levels (average of two replicates) in different categories of DEGs in WT mice, highlighting the direction of difference in mCG levels between up- and down-regulated DEGs. The box plots within each segment highlights the median of the distribution, while the red point represents the mean of the distribution. (C) Gene-body mCH levels in different categories of DEGs in WT, Dnmt3a cKO and Mecp2 cKO mice. (D) Gene-body mCG levels in different categories of DEGs in WT, Dnmt3a cKO and Mecp2 cKO mice. (E) P-value matrix for comparison of mCH levels of different categories of genes. The boxes are colored such that more significant p-values are darker shades of red. (F) P-value matrix for comparison of mCG levels of different categories of genes. As in C, the boxes are colored such that more significant p-values are darker shades of red.

Figure 4—figure supplement 2
Integrative gene expression and methylation analysis excluding interneuron specific DEGs determined in PSEA analysis shows the same trends and statistical significance as in Figure 4.

(A–H) Running average and validation plots from Figure 4 re-plotted after removing DEGs that may represent changes in gene expression in the minor population of co-purified inhibitory interneurons. All plots were made with DEGs padj <0.01. n = 2 mice per genotype for methylation data. n = 4 mice per genotype (RNA-seq). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Tables

Table 1
Summary of behavioral and physiological test results in cKO models.
Symptom CategoryTest/observationResult
General HealthReduced body weightDnmt3a
Prematurely moribund (humane end due to lesioning)both
SpasticityHind limb claspingboth
Repetitive BehaviorForepaw stereotypiesDnmt3a
Perseverative grooming and self-injuryboth
Nociceptive painDelayed response to hot plateboth
Delayed response to tail flickDnmt3a
ApraxiaPoor nest buildingboth
Muscle StrengthDecreased grip strengthboth
MotorRotarodnone
Parallel rod footslip- more footslipsMecp2
Open field – decreased distance traveledboth
Open field – decreased vertical activityboth
AnxietyOpen field – time spent in centernone
Elevated plus maze- increased time in open armsMecp2
Light/dark boxnone
Learning and MemoryConditioned fear – decreased contextual fear responseboth
Conditioned fear – cuednone
Conditioned fear – decreased response to tone #2 during fear learning (training)Dnmt3a
Sensory processingIncreased acoustic startle responseDnmt3a
Paired pulse inhibitionnone
Social BehaviorPartitionnone
Inhibitory signalingAltered miniature inhibitory postsynaptic currentsboth
Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional
Information
AntibodyRabbit polyclonal anti-Dnmt3aSanta Cruz BiotechnologyCat. #: 20703; RRID:AB_2093990(1:1000)
AntibodyRabbit polyclonal anti-MeCP2Huda Zoghbi lab (in house)0535(1:10,000)
AntibodyRabbit polyclonal anti-histone H3AbcamCat. #: 1791; RRID:AB_302613(1:20,000)
Antibodygoat anti-rabbit-HRPBioRadCat. #: 170–5046;RRID:AB_11125757(1:10,000)
AntibodyMouse monoclonal anti-Dnmt3aNovusCat. #: NB120-13888; RRID:AB_789607(1:250)
AntibodyRabbit polyclonal anti-MycSigmaCat. #: C3956; RRID:AB_439680(1:200)
AntibodyRabbit monoclonal anti-MeCP2Cell Signaling
Technology
Cat. #: 3456; RRID:AB_2143849(1:500)
AntibodyChicken polyclonal anti-GFPAbcamCat. #: 13970; RRID:AB_300798(1:800)
AntibodyGoat anti-mouse Alexa Fluor 555InvitrogenCat. #: A21127; RRID:AB_141596(1:1000)
AntibodyGoat polyclonal anti-rabbit Dylight 488Bethyl LaboratoriesCat. #: A120-201D2; RRID:AB_10634085(1:750)
AntibodyGoat anti-chicken Alexa Fluor 488InvitrogenCat. #: A11039; RRID:AB_142924(1:750)
AntibodyDonkey anti-rabbit Alexa Fluor 568Thermo Fisher ScientificCat. #: A10042; RRID:AB_2534017(1:1000)
AntibodyMouse monoclonal anti-NeuNMilliporeCat. #: MAB377; RRID:AB_2298772(1:300)
(each batch needs to be empirically tested after labeling with Alexa Fluor 647)
AntibodyRabbit polyclonal anti-GFP 488Thermo Fisher ScientificCat. #: A-21311; RRID:AB_221477(1:1000)
OthercOmplete, EDTA-free Protease Inhibitor CocktailSigmaCat. #: 5056489001
OtherPierce Universal Nuclease for Cell LysisThermo Fisher ScientificCat. #: 88701
OtherOptimal Cutting
Temperature medium
VWRCat. #: 25608–930
OtherVECTASHIELD HardSet Antifade Mounting Medium without DAPIVector LaboratoriesCat. #: H-1400
OtherCryoseal XYLFisher ScientificCat. #: 22-050-262
OtherAqueous Glutaraldehyde
EM Grade, 10% 10 ML
Electron Microscopy SciencesCat. #: 16100Dilute fresh
OtherParaformaldehydeSigmaCat. #: P6148Make fresh
OtherX-GalGold BiotechnologiesCat. #: X4281C
OtherRNasin Ribonuclease InhibitorsPromegaCat. #: N261A
OtherUltraPure BSA (50 mg/mL)Thermo Fisher ScientificCat. # AM2618
OtherDPBS, no calcium, no magnesiumInvitrogenCat. #: 14190144
OtherUltraPure DEPC-Treated Water (1L)InvitrogenCat. #: 750023
Commercial assay, kitGE Healthcare Amersham ECL Prime Western Blotting Detection ReagentFisher ScientificCat. #: 45010090
Commercial assay, kitAPEX Alexa Fluor 647 Antibody Labeling KitThermo Fisher ScientificCat. #: A10475
Commercial assay, kitSingle Cell RNA Purification KitNorgen BiotekCat. #: 51800
Commercial assay, kitNuGEN Ovation RNA-Seq v2NuGenprotocol p/n 7102, kit p/n 7102–08
Commercial assay, kitRubicon ThruPlex
DNA-Seq
Rubicon Genomicsprotocol: QAM-108–002, kit p/n R400428
Commercial
assay, kit
EZ DNA Methylation-Direct KitZymoCat. #: D5021
OtherUnmethylated lambda DNA spike-inPromegaCat. #: D1521
Strain, Strain background (Mus musculus)Dnmt3aflox/flox miceDr. Margaret Goodell, Baylor College of Medicine (Can be purchased from Riken BRC)Cat. #: RBRC03731; RRID:IMSR_RBRC03731
Strain, Strain background (Mus musculus)Slc32a1-Cre+/+ miceJackson LaboratoryCat. #: 017535;
RRID:IMSR_JAX:017535
backcrossed to C57Bl/6J
Strain, Strain background (Mus musculus)Mecp2flox/flox and Mecp2flox/y miceMMRRCCat. #: 011918-UCD; RRID:MMRRC_011918-UCDbackcrossed to C57Bl/6J
Strain, Strain background (Mus musculus)R26-CAG-LSL-Sun1-sfGFP-Myc+/+ miceDr. M. Margarita Behrens, The Salk Institute for Biological Studies (can be purchased from Jackson Laboratory)Cat. #: 021039; RRID:IMSR_JAX:021039backcrossed to C57Bl/6J
Software, algorithmGraphPad Prism version 6GraphPad Softwarewww.graphpad.com
OtherpClamp10Molecular Deviceshttps://www.moleculardevices.com
Software, algorithmMinianalysis 6.0.3Synaptosoft Inchttp://www.synaptosoft.com/MiniAnalysis/
Software, algorithmSTAR aligner version 2.5.3aGitHubhttps://github.com/alexdobin/STAR
Software, algorithmFeatureCount v1.5.3Subreadhttp://subread.sourceforge.net
Software, algorithmDESeq2 v1.6.2Bioconductorhttps://bioconductor.org/packages/release/bioc/html/DESeq2.html

Data availability

DNA methylome data can be accessed through a web browser at http://neomorph.salk.edu/Striatum_Inhibitory_Neuron.php and at the Gene Expression Omnibus database (GEO) at accession number GSE124009. RNA-seq data can be accessed at GEO at accession number GSE123941.

The following data sets were generated
  1. 1
    NCBI Gene Expression Omnibus
    1. C Luo
    2. LA Lavery
    3. R Castanon
    4. JR Nery
    5. HY Zoghbi
    6. JR Ecker
    (2018)
    ID GSE124009. Loss of non-CpG methylation in inhibitory neurons impairs neural function through a mechanism that partially overlaps with Rett syndrome.
  2. 2
    NCBI Gene Expression Omnibus
    1. LA Lavery
    2. Y Wan
    3. HY Zoghbi
    (2018)
    ID GSE123941. Loss of non-CpG methylation in inhibitory neurons impairs neural function through a mechanism that partially overlaps with Rett syndrome.
The following previously published data sets were used
  1. 1
    NCBI Gene Expression Omnibus
    1. H Stroud
    2. SC Su
    3. S Hrvatin
    4. AW Greben
    5. W Renthal
    6. LD Boxer
    7. MA Nagy
    8. DR Hochbaum
    9. B Kinde
    10. HW Gabel
    11. ME Greenberg
    (2017)
    ID GSE103214. Early-life gene expression in neurons modulates lasting epigenetic states.

Additional files

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)