ABSTRACT
Individuals with monoallelic pathogenic variants in the histone lysine methyltransferase DOT1L display global developmental delay and varying congenital anomalies. However, the impact of monoallelic loss of DOT1L remains unclear. Here, we present a largely female cohort of 11 individuals with DOT1L variants with developmental delays and dysmorphic facial features. We found that DOT1L variants include missense variants clustered in the catalytic domain, frameshift, and stop-gain variants. We demonstrate that specific variants cause loss of methyltransferase activity and therefore sought to define the effects of decreased DOT1L function. Using RNA-sequencing of cultured neurons and single nucleus RNA-sequencing of mouse cortical tissue, we found that partial Dot1l depletion causes sex-specific transcriptional responses and disrupts transcription of synaptic genes. Further, Dot1l loss alters neuron branching and expression of synaptic proteins. Lastly using zebrafish and mouse models, we found behavioral disruptions that include sex-specific deficits in mice. Overall, we define how DOT1L loss leads to neurological dysfunction by demonstrating that partial Dot1l loss impacts transcription, neuron morphology, and behavior across multiple models and systems.
INTRODUCTION
Neurodevelopmental disorders (NDDs) are a diverse group of highly prevalent (0.3 – 18.5%) (1) conditions that manifest during development and impact central nervous system functions (2, 3). The spectrum of NDDs include intellectual disability, autism spectrum disorder (ASD), attention deficit/hyperactivity disorder (ADHD), communication disorders, specific learning disabilities, and motor disorders (4). The cause of NDDs is multifactorial, and includes both inherited and de novo, genetic variants with a notable overrepresentation of epigenetic regulators (5–9). One subset of epigenetic regulators, histone methyltransferases, are linked to numerous NDDs (10–12) and function by methylating histones to regulate transcription. Histone methyltransferases are critical for neurogenesis, neuronal migration, neuronal differentiation, synaptic plasticity and cognition (13) yet several disease-linked methyltransferases have not yet been studied in the context of neuronal function or animal behavior.
Prior exome sequencing studies identified variants in the histone methyltransferase DOT1L as a potential causative driver of NDDs (6, 12). More recent work identified two variants in DOT1L in individuals displaying ADHD (14) and nine monoallelic (presumed) de novo variants of DOT1L were identified in individuals with global developmental delay (15). Complete loss of DOT1L in mouse models is embryonic lethal (16) while in Drosophila, loss of grappa, the Drosophila DOT1L ortholog, leads to developmental delay and lethality (15). However, grappa is highly divergent from DOT1L and thus does not provide an ideal model to study an emerging human disorder. Further, while two previously identified variants were proposed to be gain-of-function based on human cell based-assays (15) most identified variants have unclear functional consequences. Lastly, while most prior work used full and transmitted knockout models, variants are typically monoallelic and de novo making it difficult to define the effect of partial DOT1L disruption from existing data. Thus, the underlying mechanisms linking DOT1L to NDDs remain unclear.
DOT1L is the sole methyltransferase responsible for depositing mono-, di-, and trimethyl methyl marks on the histone-fold domain on residue 79 of histone H3 (H3K79me) (17, 18). H3K79me is enriched in gene bodies peaking after the transcription start site (17), with higher methyl states linked to greater transcriptional output (19). DOT1L interacts with RNA polymerase II (20) and TFIID (21) and recruits effector proteins such as Menin (22) to regulate transcription. DOT1L functions in numerous cellular processes, including development (23) such as in neural progenitor proliferation and differentiation in the cortex, cerebellum, and spinal cord (24–27) and in maintaining the transcriptional state in differentiating neural progenitors(28–31). Further work demonstrated that stress modulates DOT1L expression and H3K79me in the nucleus accumbens and that monoallelic loss of Dot1l in the midbrain disrupts synaptic and mitochondrial genes (32). Cumulatively, this suggests that DOT1L is critical in neuronal development and neuronal function. Despite these advances, the majority of DOT1L research has focused on biallelic loss of DOT1L, which does not reflect the monoallelic nature of individual variants or has not examined effects on development and behavior. Thus, the consequences of monoallelic disruptions of DOT1L remain poorly understood.
Here, we identified 11 individuals with monoallelic variants in DOT1L displaying a spectrum of neurodevelopmental phenotypes and dysmorphic facial features. Using structural protein modeling, biochemical studies, and patient-derived cells, we found that several variants cause loss of DOT1L methyltransferase activity. Utilizing a dot1l knockdown system in zebrafish, we identified disruptions in motor responses to sensory stimuli. Harnessing both primary cultured cortical neurons bulk RNA-sequencing and in vivo cortical neuronal single-nucleus RNA-sequencing in mice, we show that partial loss of Dot1l affects transcription of critical neuronal genes linked to synaptic function and causes sex-specific transcriptional responses. Further, cortical neurons display disruptions in neuronal morphology upon partial Dot1l loss. Finally, we identified behavioral alterations upon both ubiquitous and neuron-specific monoallelic loss of Dot1l in mice. Together, our work demonstrates that partial loss of Dot1l causes transcriptional disruptions impacting cognitive function and provides insight into the neurodevelopmental disruptions found in individuals with DOT1L variants.
RESULTS
Identification of individuals with a spectrum of neurodevelopmental disorders and DOT1L variants
Given the recent discovery of DOT1L’s association with an emerging neurodevelopmental disorder, we searched specifically for individuals harboring variants in DOT1L. We collected a cohort of individuals through collaborating clinicians and GeneMatcher (33) with DOT1L variants identified through genome sequencing or exome sequencing. Criteria for inclusion consisted of individuals displaying developmental phenotypes with DOT1L variants not observed in multiple individuals from the general population and without additional known pathogenic variants.
Using these criteria, we compiled a cohort of 11 individuals with variants in DOT1L. All individuals have only a single monoallelic variant in DOT1L. Inheritance of these variants was de novo (6/11), maternal (1/11) or inconclusive due to one or both parents being unavailable to be sequenced (4/11). In the maternally inherited case, dysmorphic facial features were noted in the mother, but full phenotyping was not available. Variants include missense (9/11), frameshift (1/11) and stop-gain (1/11). Notably, all but one of the missense variants (8/9) are within the catalytic domain of DOT1L and affect amino acids that are rarely altered in humans (i.e. dn/ds score of <0.2), indicating intolerance to variation at these sites (Fig. 1A). According to gnomAD (v4.1.0) (34), DOT1L has a high probability of loss-of-function intolerance (pLI = 1, LOEUF = 0.32) and a high probability of deletion intolerance (pHaplo = 0.98). All variants were absent in gnomAD (v4.1.0) apart from one counted allele of p.L1067Dfs*66. The cohort displays a non-specific constellation of congenital anomalies, including craniofacial anomalies (10/11) such as midface hypoplasia (Table 1, Fig. 1B). Based on the information available at this time, there is no recognizable pattern of morphological differences that would suggest the diagnosis in the absence of molecular genetic testing. Additional individual phenotypes include intellectual disability (2/11), language delay (8/11), motor delay (7/11), and a diagnosis of ASD (2/11) (Table 1, Fig. 1B). Three additional individuals with variants in DOT1L were identified through the MSSNG (35) database with a diagnosis of ASD but are not included in the main cohort due to an inability to gather additional information (fig. S1A, S1B, table S2). Further, four individuals with variants in DOT1L also contained additional potential pathogenic variants or had a DOT1L variant found the general population and thus did not meet criteria for inclusion in the cohort. We include them here (fig. S1A, S1B, table S2) given that they shared some features with the main cohort and that we cannot rule out the possibility of incomplete penetrance of this disorder. Interestingly, while the cohort is not sufficiently powered to confidently determine sex enrichment and the prior smaller cohort (15) was split roughly equally by sex, 9 out of 11 the individuals in this cohort are female suggestive of possible sex bias.
(A) The landscape of missense tolerance of DOT1L from Metadome for individuals with ASD from MSSNG database, individuals with variants in the general population, or individuals harboring additional pathogenic variants. Schematic of DOT1L protein domains (purple = catalytic and yellow= nuclear localization signal) and locations of DOT1L variants (black = missense variant, red = stop/frameshift variant). (B) Clinical phenotypes of individuals with DOT1L variants from MSSNG, individuals with variants in the general population, or additional patho genic variants (Y = present, B = borderline, N = not present, -= not reported). (C) Methyltransferase activity of human DOT1Lor variants of DOT1L (R292C or E123K) on unmodified nucleosomes. Graph shows mean ± SE (n = 3/condi tion, unpaired two-tailed t-test). RLU = relative light units. (D) Sanger sequencing of human fibroblasts from individual 5 with variant DOT1L (D157N) and from age-and sex-matched control with wildtype DOT1L. (E) Alignment of catalytic domain variants to zebrafish Dot1I, mouse DOT1L, and human DOT1L using DIOPT. Individual variants are highlighted in bold. (F) Sanger sequence traces from injected animals with the three dot1I gRNAs used to generate crispants. (G) Schematic behavioral phenotyping paradigm in zebrafish dot1/ crispants. (H) Total distance traveled with lights on and off during the visual motor response assay in dot1/ crispant and control zebrafish (control: n = 107, dot1/ crispant: n = 89; 3 independent experiments, Kruskal-Wallis Test). All box plot bounds indicate the 25th and 75th percentiles, the black line shows the median, and whiskers extend to the minimum and maximum value that are no further than 1.5 * interquartile range. *p<0.05, **p<0.01, ****p<0.0001.
(A) The landscape of missense tolerance of DOT1L from Metadome. Schematic of DOT1L protein domains (purple = catalytic and yellow= nuclear localization signal) and locations of DOT1L variants (black = missense variant, red = stop/frameshift variant). (B) Images of dysmorphic facial features in individuals with DOT1L variants. (C) Structure of the DOT1L nucleosome complex (PDBID: 6NJ9) highlighting the position of variants (in cyan and blue spheres). Blue spheres represent variants tested by enzymatic assays in this study. DOT1L is in magenta, DNA in grey, histone H2A in pale yellow, histone H2B in red salmon, histone H3 in marine blue, and histone H4 in green. For clarity, ubiquitin is not shown. SAM is shown in green sticks. (D) Methyltransferase activity of human DOT1L or variant DOT1L (D157N) on unmodified nucleosomes. Graph shows mean ± SE (n = 3/condition, unpaired two-tailed t-test). RLU = relative light units. (E) H3K79 methyl levels in patient-derived fibroblasts from individual 5 (D157N variant) beside an age-and sex-matched control. (F) H3K79 methyl levels in Neuro-2A cells with variant Dot1/ (D157N) overexpressed, wildtype Dot1/, or empty vector control. (G) Schematic of Cas9 targeting of dot1/ and subsequent behavioral phenotyping paradigm. (H) Average distance traveled during dark flashes in dot1I crispant and control zebrafish (control: n = 113, dot1/ crispant: n = 98; 3 independent experiments, unpaired two-tailed t-test). (I) Area under the curve of response to increasing stimulus intensity in dot1I crispant and control zebrafish (AU = arbitrary units, control: n = 113 control, dot1/ crispant: n = 95; 3 independent experiments, Kruskal-Wallis Test). (J) Prepulse inhibition (PPI) to a medium intensity prepulse acoustic stimuli in dot1/ crispant and control zebrafish (control: n = 113 control, dot1/ crispant: n = 95; 3 independent experiments, Kruskal-Wallis Test). All box plot bounds indicate the 25th and 75th percentiles, the black line shows the median, and whiskers extend to the minimum and maximum value that are no further than 1.5 * interquartile range. **p<0.01, ****p<0.0001.
DOT1L variant individuals’ phenotypic spectrum and prevalences. ASD = autism spectrum disorder. ADHD = attention deficit hyperactivity disorder.
We next determined the location of missense variants in the catalytic domain of DOT1L based on a published structure of DOT1L (PDBID: 6NJ9) (36) (Fig. 1C). Variants are spread throughout the catalytic domain, including regions in close proximity to the binding pocket and nucleosome interface likely to affect DOT1L function. Given that most of the variants lie within the catalytic domain, we assessed methyltransferase activity via endpoint histone methyltransferase assays. We selected two previously published variants (15) (p.R292C and p.E123K) one of which was reported to have no effect (p.R292C) and the other of which was proposed to increase activity (p.E123K). We also assessed the p.D157N variant based on the identification of variants at residue 157 in two unrelated individuals. Methyltransferase assays demonstrated that p.R292C and p.D157N reduced methyltransferase activity (Fig. 1D, fig. S1C). In contrast, p.E123K increased activity, as previously reported (15). Further, human fibroblasts harboring the p.D157N had a decrease in all three H3K79me states compared to age-and sex-matched control fibroblasts further supportive of loss of catalytic activity in DOT1L (Fig. 1E, fig. S1D). Lastly, to determine the effect of p.D157N in an orthogonal system without the confound of different genetic backgrounds from primary human fibroblasts, we overexpressed wildtype Dot1l and variant Dot1l (p.D157N) in mouse Neuro-2A cells. Wildtype Dot1l increased H3K79me2 and, to a lesser extent H3K79me1/3 (Fig. 1F). However, variant Dot1l (p.D157N) had no detectable impact on H3K79me levels. Together, these data demonstrate that the p.D157N variant reduces catalytic activity. In addition to functional testing of DOT1L variants, we noted that two variants cause early stop codons that will either result in nonsense-mediated decay or a severely truncated protein. Further, these variants truncate DOT1L upstream of nuclear localization sequences (Fig. 1A), likely preventing DOT1L from performing established nuclear functions if translated into protein. Together, this suggests that both gain-and loss-of-function variants are found within DOT1L. Given our findings that both a previously published variant and the D157N variant have reduced catalytic activity and the discovery of two patients with stop-gain variants, we chose to examine the effects of partial loss-of-function DOT1L to more broadly model individuals with DOT1L variants and to better understand the role of DOT1L in the brain.
Loss of dot1l in zebrafish leads to exaggerated motor behavior in response to sensory stimuli
Given the de novo nature of the majority of individual variants, we aimed to characterize early behavioral disruptions utilizing a system that allows for allele disruptions in the offspring of zebrafish. Additionally, zebrafish provide a vertebrate model with high genetic similarity to humans (37), including DOT1L (Catalytic domain: 85% identity, Whole gene: 49% identity, 57% similarity, 21% gaps) (38) (fig. S1E). Zebrafish also develop robust stereotypical motor movements in response to sensory stimuli (visual or acoustic) detectable within the first six days of development. Prior work demonstrated that these behaviors are sensitive to mutations in genes associated with NDDs (39–41), suggesting their relevance to NDD pathophysiology.
To assess behavioral roles for DOT1L in early development, we first specifically disrupted the zebrafish dot1l gene using a CRISPR-Cas9-approach that generates biallelic null alleles (42) in over 90% of animals. Briefly, we injected three guide RNAs that target non-overlapping sites along the dot1l gene into fertilized embryos together with Cas9 protein, generating dot1l ‘crispants’. Control embryos were injected in parallel with three non-targeting gRNAs and Cas9. We first confirmed that each gRNAs targeted dot1l by sequencing (fig. S1F). Dot1l crispants, were viable to 6 days post-fertilization (dpf) and did not display obvious gross morphological defects (control injected n=164, dot1l crispant n=144, 4 independent experiments). Behavior of dot1l crispants was then assessed at 6dpf using a previously described pipeline that allows assessment of multiple sensorimotor behaviors including the visual motor response, responsiveness to flashes of light or darkness, and the acoustic startle response (39) (Fig. 1G, fig. S1G). Compared to controls, dot1l crispants displayed exaggerated motor responses to multiple sensory inputs. Specifically, dot1l crispants displayed increased movement in response to changes in illumination, as illustrated by increased distance travelled in the visual motor response assay (43) (fig. S1H) and increased movement in response to flashes of darkness (44) (Fig. 1H). In addition, dot1l crispants are hypersensitive to acoustic stimuli, displaying startle responses (45) following stimuli that do not elicit similar responses in controls (Fig. 1I). Further, dot1l crispants also show an increase in acoustic startle prepulse inhibition (46) compared to controls (Fig. 1J). Together, these results demonstrate that zebrafish dot1l controls responses to visual stimuli and is required for establishing the acoustic startle threshold and acoustic startle sensorimotor gating in zebrafish. Further, they demonstrate that DOT1L loss affects early developmental behaviors.
DOT1L regulates glutamatergic synaptic gene expression
Based on the broad neurodevelopmental phenotypes observed in individuals with DOT1L variants, and the robust effects of DOT1L loss in zebrafish behavior, we next tested the role of DOT1L in mouse models based on highly conserved DOT1L (Catalytic domain: 96% identity, Whole gene: 84% identity, 88% similarity, 1% gaps) (38) (fig. S1E). We began by defining the regulation of Dot1l and its target histone modification H3K79me in developing mouse neurons. We leveraged mouse primary cultured neurons derived from E16.5 cortices to generate a pure neuronal population and found that both Dot1l and H3K79me increase throughout neuronal maturation (fig. S2A, S2B). To model partial loss of Dot1l, we infected primary cortical neurons with short hairpin RNAs (shRNA) targeting Dot1l or a non-targeting control (n.t.). We confirmed Dot1l loss and H3K79me depletion upon infection of Dot1l shRNAs (fig. S2C, S2D) demonstrating that H3K79me is dynamically regulated and requires continued DOT1L function for H3K79me deposition in developing neurons.
(A) Dot1/ expression in primary cortical neurons from 5-20 days in vitro (DIV) Graph shows mean± SE. (n=6/DIV). (B) Representative western blot of H3K79me3 in primary cortical neurons from 5-20 DIV. (C) Dot1/ expression in primary cortical neurons infected with 3 different Dot1/ shRNAs or non-targeting control (n.t.). Graph shows mean± SE. (n.t.: n = 3, Dot1/ shRNA 1: n = 2, Dot1/ shRNA 2: n = 3, Dot1/ shRNA 3: n = 3, ANOVA with post-hoc pairwise t-test with Bonferroni correction). (D) Representative western blot of H3K79me2/3 in primary cortical neurons infected with Dot1/ shRNA or n.t. control. (E-G) Gene set enrichment analysis of genes involved in (E) GABAergic, (F) cholinergic, (G) or dopaminergic synaptic transmission. NES indicates normalized enrichment score. *p<0.05, **p<0.01.
Given the association between H3K79me and active gene expression, we next sought to determine the effect of partial loss of Dot1l on transcription in primary neurons. Following Dot1l depletion, we performed RNA-sequencing and found widespread changes in gene expression with 677 genes significantly up-regulated and 1050 genes significantly down-regulated (Fig. 2A). Gene ontology (GO) analysis indicated an enrichment of genes involved in synaptic transmission (such as GO:0099177 and GO:0050804) in down-regulated differentially expressed genes (DEGs) and no significant enrichment of GO terms in up-regulated DEGs (Fig. 2B). Given the dysregulation of synaptic-related genes, we further interrogated differentially expressed genes using SynGO (47) which demonstrated enrichment for pre-and post-synaptic compartment proteins, and synaptic cleft proteins suggesting widespread disruption of expression of synaptic genes (Fig. 2C). We next asked whether the observed changes in genes related to synaptic transmission were global or specific to a class of chemical synaptic transmission. To this end, we used gene set enrichment analysis to test for enrichment of genes related to glutamatergic, GABAergic, dopaminergic, and cholinergic synaptic transmission. Interestingly, glutamatergic transmission is enriched in down-regulated genes, while there is no significant enrichment of other classes of synaptic transmission (Fig. 2D, fig. S2E-G). In fact, down-regulated DEGs had significant overlap with the glutamatergic synaptic transmission gene set including genes such as Gria2 and Grin1, two glutamate receptor subunits that are critical for appropriate levels of glutamatergic transmission throughout the brain (Fig. 2E-G). Together these findings demonstrate that H3K79me is dynamically regulated in neurons by DOT1L and that partial Dot1l loss disrupts expression of critical synaptic genes.
(A) Volcano plot showing differen tially expressed genes (DEGs) from primary cortical neurons infected with Dot1I shRNA or non-targeting control shRNA (n = 3 biological replicates/condtion). Green dots represent genes with log2(FoldChange) >= 0.5 and FDR<= 0.05. (B) Biological process gene ontology analysis down-regulated genes. (C) SynGO analysis of cellular compartment in down-regulated DEGs. (D) Gene set enrichment analysis of genes involved in glutamatergic synaptic transmission. NES indicates normalized enrichment score. (E) Overlap of down-regulated DEGs and glutamatergic synaptic trans mission genes (hypergeometric test). (F) Heatmap of the 20 glutamatergic synaptic transmission genes that are down-regulated DEGs (R1-R3 denotes biological replicates). (G) RNA-seq genome browser tracks RNA-seq for Grin1 and Gria2. **p<0.01.
Dot1l loss impacts neuronal arborization and GluA2 levels
Given the disruption to expression of critical synaptic genes that we detected, we next tested the effect of partial Dot1l loss on neuronal morphology, synapses, and synaptic proteins. Neuronal branching and spine formation are critical components of neuronal maturation that allow for neuronal communication and downstream behaviors. To assess how Dot1l loss impacts neuronal architecture, we performed Sholl analysis on primary cortical neurons transfected with a Dot1l shRNA or control shRNA (Fig. 3A). Dot1l depleted neurons had a reduced number of intersections in comparison to controls indicative of reduced neuronal arborization (Fig. 3B,3C). In addition to neuronal branching, spine formation is critical for synapse development and essential for neuronal communication and memory consolidation. Interestingly, the spine density of Dot1l depleted neurons was increased (Fig. 3D). This could suggest aberrant spine development as seen in other developmental disorders (48) or be a compensatory mechanism to offset the loss of neuronal branching or decreased glutamatergic synapse function. Given the downregulation of genes involved in glutamatergic transmission, we also assessed whether the glutamatergic receptor subunit GluA2 protein is regulated by DOT1L in primary neurons. Using immunocytochemistry, we found that GluA2 is depleted upon Dot1l loss demonstrating that transcriptional disruptions functionally affect protein levels of critical synaptic genes (Fig. 3E, 3F). Conversely, overexpression of Dot1l in neurons did not result in significant changes to neuronal arborization or GluA2, suggesting gain-of-function variants may impact neurons through mechanisms that are distinct from loss-of-function variants (fig. S3A-F).
(A) Representative images of primary cortical neurons transfected with Dot1I or empty vector control. Scale bar= 20 µm. (B) Number of branch intersections per radius of transfected primary cortical neurons (control: n = 50 neurons from 8 biological replicates, Dot1/: n = 42 neurons from 8 biological repli cates). (C) Area under the curve (AUC) quantification of (B) (control: n = 50 neurons from 8 biological repli cates, Dot1/: n = 42 neurons from 8 biological replicates, mixed effect model). (D) Spine density of transfected primary cortical neurons (control: n = 50 neurons from 8 biological replicates, Dot1/: n = 42 neurons from 8 biological replicates, Kruskal-Wallis Test). (E) Representative images of GluA2 staining in shRNA-transfected primary cortical neurons. (F) Quantification of GluA2 staining in shRNA-transfected primary cortical neurons. (control: n = 13 neurons from 2 biological replicates, Dot1/ shRNA: n = 9 neurons from 2 biological replicates, Kruskal-Wallis Test). All box plot bounds indicate the 25th and 75th percentiles, the black line shows the median, and whiskers extend to the minimum and maximum value that are no further than 1.5 * interquartile range. n.s. = not significant.
(A) Representative images of primary cortical neurons transfected with Dot1f shRNA or non-targeting control. Scale bar = 20 µm. (B) Number of branch intersections per radius of shRNA transfected primary cortical neurons (control: n = 33 neu rons from 5 biological replicates, Dot1f shRNA: n = 24 neurons from 5 biological replicates). (C) Area under the curve (AUC) quantification of (B) (control: n = 33 neurons from 5 biological replicates, Dot1f shRNA: n = 24 neurons from 5 biological replicates, mixed effect model). (D) Spine density of shRNA-transfected primary cortical neurons (control: n = 33 neurons from 5 biological replicates, Dot1f shRNA: n = 23 neurons from 5 biological replicates, Kruskal-Wallis Test). (E) Representative images of GluA2 staining in shRNA-transfected primary cortical neurons. (F) Quantification of GluA2 staining in shRNA transfected primary cortical neurons. (control: n = 27 neurons from 3 biological replicates, Dot1f shRNA: n = 22 neurons from 3 biological replicates, Kruskal-Wallis Test). All box plot bounds indicate the 25th and 75th percentiles, the black line shows the median, and whiskers extend to the minimum and maximum value.*p<0.05, ****p<0.0001.
DOT1L regulates cortical gene expression in a sex-specific manner
Given the transcriptional disruptions in our in vitro primary cultured neurons and behavioral disruptions in our zebrafish model, we next sought to analyze the transcriptional effects of monoallelic loss of Dot1l in mice to model the monoallelic loss in individuals with DOT1L variants. We first examined Dot1l and H3K79me expression in mice during cortical development from E14 through postnatal day 28. We found that both Dot1l and H3K79me increase during this period, suggesting that DOT1L may play a role during this critical period of brain development (fig. S4A-D). Prior work thoroughly defined the effects of Dot1l loss on the transcriptome of stem cell populations and the effect of biallelic Dot1l loss on neurons early in development(24–31). However, to the best of our knowledge the effect of monoallelic loss has only been tested in the midbrain (32) with a focus on aging-related phenotypes and has not been tested beyond early development in brain regions relevant to the emerging disorder described here.
(A) Dot1/ expression in whole cortex from embryonic day 14 (E14) to postnatal day 0 (PO) in male and female mice. Graph shows mean± SE (E14[male: n = 5, female n= 3], E16[male: n = 4, female: n = 4], E18[male: n = 3, female: n = 6], P0[male: n = 4, female: n = 51). (B) Representative western of H3K79me3 levels in whole cortex from E14-P0 in male and female mice. (C) Dot1/ expression in cortical nuclei from P0-P28 in male and female mice (P0[female: n = 2], all remaining groups n = 3/timepoint/sex). (D) Representative western of H3K79me1 /2/3 in cortical nuclei from P0-P28 in male and female mice. (E) Dot1I expression in control and Dot1I HET whole cortex. Graph shows mean ± SE (control: n = 3 [1 male, 2 females]; Dot1/ HET: n = 4 [2 males, 2 females]). (F) Dot plot showing expression of cell type-specific marker genes for each cluster. (G-H) Number of up-and down-regulated differentially expressed genes (DEGs) within inhibitory clusters (G) and glia, endothelial, and claustrum (H) clusters. (I) Volcano plot showing DEGs from Dot1/ HET and control combined inhibitory clusters. (E) Biological process gene ontology analysis down-regulated DEGs from combined excitatory clusters. (J) Biological process gene ontology analysis down-regulated DEGs from combined inhibitory clusters. (K) Overlap of male and female up DEGs from combined excit atory clusters (hypergeometric test). (L) Overlap of down-regulated DEGs from primary neurons infected with Dot1I shRNA and down-regulated DEGs from combined excitatory clusters from Dot1I HET cortical nuclei (hypergeomet ric test). ** p < 0.01, **** p < 0.0001.
Given the notable increase and stabilization of Dot1l expression and H3K79me from P0-P28 and the lack of characterization of Dot1l after brain development, we assessed the transcriptional impact of monoallelic loss of Dot1l in 8-week-old cortical mouse tissue. We used a floxed Dot1l mouse model containing loxP sites flanking exon 2 of Dot1l that was crossed to a ubiquitously expressing Cre line under the human cytomegalovirus (CMV) minimal promoter that expresses during early embryogenesis (49) to generate Dot1lfloxed/+;CMV-Cre+/- (referred to as Dot1l HET) with littermate controls (Dot1l+/+;CMV-Cre+/-). Notably, parental lines that generated experimental cohorts included Dot1lfloxed/+ crossed to CMV-Cre+/+. This ensures that parents of experimental mice have wildtype DOT1L expression to avoid effects of parental partial loss of Dot1l which may affect the health of offspring and to better mimic the affected individuals in which most variants are de novo.
We first confirmed partial loss of Dot1l in Dot1l HET cortical tissue as expected (fig. S4E). Given that prior work established effects of complete Dot1l loss on neurogenesis and cortical layer development and our data demonstrating that partial loss of Dot1l robustly affects gene expression within neurons, we harnessed single nucleus RNA-sequencing to capture both changes in cell type identity and changes in gene expression in cortical tissue (Fig. 4A). Using 3 male and 3 female animals for both control and Dot1l HET, we identified 25 clusters that include 10 excitatory neurons clusters (Slc17a7+), 7 inhibitory neuron clusters (Gad2+), 2 microglia clusters (Ctss+ and Ptprc+), an astrocyte cluster (Gja1 + and Gnb4+), and an oligodendrocyte cluster (Mog+, Enpp6+, and Opalin +) (fig. S4F). Interestingly, we did not find altered proportions of neuronal cell types and only a modest increase in microglia cell types in Dot1l HET mice in comparison to control suggesting that partial Dot1l loss is not sufficient to alter cortical neuron identity as occurs following complete Dot1l deletion (24) (Fig. 4B). However, we found widespread disruption of gene expression across most excitatory and inhibitory neuron clusters and modest changes in non-neuronal clusters (Fig. 4C and fig. S4G, S4H). We detected the greatest effects in excitatory neuron clusters (Fig. 4C) and thus examined the effect of Dot1l loss on excitatory clusters as a whole. We found 880 significantly down-regulated genes and 310 genes significantly up-regulated upon Dot1l loss in vivo fitting with culture data demonstrating more genes are decreased in gene expression following partial Dot1l loss (Fig. 4D). Gene ontology analysis indicated an enrichment of genes involved in synaptic function (GO:0099072, GO:1903421, GO:0050804) in down-regulated DEGs and no significant enrichment of GO terms in up-regulated DEGs (Fig. 4E). Examining the excitatory cluster with the most DEGs (Ex_L2/3_1), showed similar effects with 602 significantly down-regulated genes and 221 genes significantly up-regulated (Fig. 4F). Gene ontology analysis of downregulated genes again indicated disruption of genes involved in synaptic function (GO:0050803, GO:0050804, GO:0099003, GO:0099536), a feature that was echoed in inhibitory neuron clusters as well (Fig. 4G and fig. S4I, S4J).
(A) UMAP (Uniform Manifold Approxi mation and Projection for Dimension Reduction) of single nucleus RNA sequencing from 8-week-old Dot1/ HET and control mouse cortex (n = 6 cortices per group[3 males, 3 females). (B) Proportion of nuclei in Dot1/ HET and control in each cluster. Red indicates fold difference > 1.8 and FDR < 0.05. (C) Number of up-and down-regulated differentially expressed genes (DEGs) in excitatory clusters. (D) Volcano plot showing DEGs from Dot1/ HET and control combined excitatory clusters. (E) Biological process gene ontology analysis down-regulated DEGs from combined excitatory clusters. (F) Volcano plot showing DEGs from Dot1I HET and control from Ex_L2/3_1. (G) Biological process gene ontol ogy analysis down-regulated DEGs from Ex_L2/3_1. (H) UMAP clustering with dots representing nuclei from Dot1/ HET or control separated by sex. (I) Volcano plots showing DEGs from male or female Dot1/ Het and control from combined excitatory clusters. (J) Overlap of male and female down DEGs from combined excitatory clusters (hypergeometric test). (K) Dot plots showing overall down DEGs (Syngr1, Camkv) and sex-specific down DEGs (Camk2n1, Syt4, Kcnj4, Unc5a) from combined excitatory clusters. ***p < 0.00001.
Given that the majority of individuals within the cohort were female, we interrogated whether there may be sex-specific transcriptional alterations upon monoallelic Dot1l loss. To parse sex-specific effects, we separated male and female cells and found the sexes were equally represented in each cluster (Fig. 4H). We again detected widespread gene expression changes in both male and female excitatory neuron clusters (Fig. 4I). However, we detected slightly more down-regulated genes in female neurons with 312 uniquely down-regulated in female and 222 genes uniquely down-regulated in males, with a similar effect in up-regulated genes (Fig. 4J, 4K and fig. S4K). Interestingly, we detected Dot1l-sensitive genes for which female neurons showed decreased expression compared to males even in control tissue suggesting baseline transcriptional differences in female neurons may contribute to different responses to Dot1l monoallelic loss. These findings demonstrate that there are both shared and sex-specific transcriptional programs down-regulated upon monoallelic Dot1l loss and that female neurons may be more sensitive to loss of Dot1l loss due to underlying differences in transcriptional states. Finally, we compared in vivo and in vitro RNA-sequencing gene sets. Genes unique to each system were identified as expected due to the difference in methods (whole cell analysis in the in vitro system versus nuclei-specific analysis in the in vivo system) and due to the differences in length and method of DOT1L depletion (5-day knockdown verses long-term genetic depletion). However, despite these differences, we identified 69 down-regulated genes shared between our in vitro Dot1l shRNA dataset and the in vivo Dot1l cKO dataset suggesting shared transcriptional disruptions between even highly distinct models of partial Dot1l loss (fig. S4L).
Monoallelic Dot1l loss alters early vocalization development and sociability
Previous studies using homozygous Dot1l mouse knockouts indicate that it is essential for hematopoiesis (50), cardio myocyte function (51), and neural progenitor proliferation and differentiation in the cortex, cerebellum, and spinal cord (24–27). However, defining the developmental and behavioral responses to monoallelic loss of Dot1l is critical to understand the implications of monoallelic variants in affected individuals. To the best of our knowledge such work has not been performed outside of one publication noting that heterozygous germline knockout mice were normal and fertile (16).
As previously reported (16), Dot1l HETs are viable and generated in approximately expected Mendelian ratios (fig. S5A, S5B). To assess impacts of monoallelic Dot1l loss on early development, we tracked developmental milestones including physical landmarks, and sensorimotor development in Dot1l HET and controls. Male Dot1l HET had no differences in weight but had delayed development of the visual placing response, a measurement of sensory development (Fig. 5A, fig. S5C). Female Dot1l HET pups weighed more than controls but had no delayed development (Fig. 5A, fig. S5C). Given the language delay seen in 8/11 individuals with DOT1L variants, and previous studies demonstrating ultrasonic vocalizations (USVs) changes in various NDD mouse models (52), we measured USVs in pups during 5 minutes of maternal separation at P6. Male Dot1l HET pups had significantly higher decibel calls, and a greater percentage of chevron type calls (Fig. 5B-E). Female Dot1l HET mice had decreased total USV calls with no differences in call characteristics (Fig. 5B-E). Finally, we found both male and female Dot1l HET mice were slower to complete a negative geotaxis assay where mice are placed face down on an angled platform to assess early motor and vestibular development (Fig. 5F).
(A) Pie chart of progeny from Dot11floxed/+ x CMV-Cre+/+ crosses (n = 168 mice from 24 litters). (B) Survival rate at weaning age (approximately 4 weeks) in Oot1/ HET and control mice (control: n = 37 mice from 9 litters, Dot1/ HET: n = 44 mice from 10 litters). (C) Developmental milestone achievements in male and female mice (male[control: n = 13, Dot1/ HET: n = 16]; female[control: n = 15, Dot1/ HET: n = 23], unpaired two-tailed t-test). (D) Total number of rears during the open field assay in male and female mice. (males[control: n = 13, Dot1I HET: n = 16]; female[ control: n = 15, Dot1/ HET: n = 23], unpaired two-tailed t-test). (E) Percent of time spent in the center of the open field arena in male and female mice (Kruskal-Wallis Test). (F) Total entries during the elevated zero maze in male and female mice (unpaired two-tailed t-test). (G) Total arm entries during the Y maze in male and female mice (unpaired two-tailed t-test). All box plot bounds indicate the 25th and 75th percentiles, the black line shows the median, and whis kers extend to the minimum and maximum value that are no further than 1.5 * interquartile range.
(A) Weight during first two weeks post-birth in male and female mice (male[control: n = 13, Dot1/ HET: n = 16]; female[control: n = 15, Dot1/ HET: n = 23], repeated measures ANOVA). (B) Total number of ultrasonic vocalizations (USVs) in P6 male and female pups (male[control: n = 15, Dot1/ HET: n = 22]; female[control: n = 22, Dot1/ HET: n = 22], Kruskal-Wallis Test). (C) Average volume in decibels (dB) of USVs in male and female pups (unpaired two-tailed t-test). (D) Average frequency in kilohertz (kHz) of calls in male and female pups. (E) Percent chevron calls out of total calls in male and female pups (Kruskal-Wallis Test). (F) Time to right self in negative geotaxis assay during development in male and female pups (male[conrol: n = 13, Dot1/ HET: n = 16]; female[control: n = 15, Dot1/ HET: n = 23], repeated measures ANOVA). (G) Total activity measured from beam breaks during 10-minute open field assay in 4 week old male and female mice (male[control: n = 13, Dot1/ HET: n = 16]; female[control: n = 15, Dot1/ HET: n = 23], Kruskal-Wallis Test). (H) Percent of time spent in open arms of the elevated zero maze in male and female mice (unpaired two-tailed t-test). (I) Percent spontaneous alternations out of the total number of triads possible in male and female mice.(unpaired two-tailed t-test). (J) Discrimination index between inter action time with the mouse or rock cylinder during the three-chamber social choice assay in male and female mice (Krus kal-Wallis Test). All box plot bounds indicate the 25th and 75th percentiles, the black line shows the median, and whiskers extend to the minimum and maximum value that are no further than 1.5 * interquartile range. *p<0.05, **p<0.01.
Next, we performed a battery of behavioral assays to assess motor and cognitive function in juvenile Dot1l HET and controls. We found no impairments in gross motor function in Dot1l HET mice in an open field assay (Fig. 5G, fig. S5D). Further, there was no evidence of anxiety-related behaviors measured using percent of time spent in open arms of the elevated zero maze and percent of time spent in the center of the open field assay (Fig. 5H, fig. S5E, S5F). We also detected no changes in working memory in Dot1l HET in comparison to controls measured using percent of spontaneous alternations completed in a Y maze (Fig. 5I, fig. S5G). To assess sociability, we performed the social choice assay where mice explore the 3-chamber arena with one chamber holding a rock, one chamber holding a mouse, and a neutral center chamber. Female Dot1l HET had a reduced time spent with the mouse measured using a discrimination index (time spent with mouse – time spent with rock / total interaction time) indicating sex-specific social behavior changes (Fig. 5J). Together, these data demonstrate that Dot1l HET mice have sex-dependent deficits in sensorimotor function, vocalization development, and sociability.
Neuronal Dot1l loss alters early vocalization development and sociability
Given the behavioral alterations seen in Dot1l HET, we next used a forebrain neuron specific Dot1l conditional knockout (cKO) mouse to assess whether behavioral alterations in the Dot1l HET mice can be attributed to monoallelic loss of Dot1l specifically in forebrain neurons. We confirmed partial loss of Dot1l and H3K79me in Dot1l cKO cortical tissue as expected (fig. S6A, S6B). We again recorded USVs in P6 pups and found both male and female Dot1l cKO mice had altered frequency of calls in comparison to controls and male Dot1l cKO mice had reduced down calls suggesting that DOT1L in forebrain neurons contributes to early vocalization deficits observed in ubiquitous Dot1l HET mice (Fig. 6A-D). We did not find any developmental delays or weight alterations suggesting effects observed in Dot1l HETs are independent of DOT1L function in neurons (fig. S6C-E). Similarly to Dot1l HET mice, Dot1l cKO had no motor activity or anxiety-related impairments (Fig. 6E, fig. S6F-I). In contrast, female Dot1l cKO had increased spontaneous alternations with no change observed in males (Fig. 6F, fig. S6J). Notably, we again found a sex-specific sociability deficit in the 3-chamber social test in female Dot1l cKO similar to Dot1l HET (Fig. 6G), indicating that DOT1L loss in neurons contributes to this effect. We also found long-term memory deficits in male Dot1l cKO mice in contextual fear conditioning (Fig. 6H). While no significant change was observed in females in contextual fear conditioning, freezing rates were low in control female mice, so it is possible that we lacked the dynamic range to detect differences. These data indicate that forebrain neuron-specific monoallelic depletion of Dot1l recapitulated sociability deficits in female Dot1l HETs and caused changes in vocalization behavior suggesting that expression of Dot1l in neurons contributes to specific behavioral alterations.
(A) Dot1I expression in whole cortex from embryonic day 14 (E14) to postnatal day 0 (PO) in male and female mice. Graph shows mean± SE (E14[male: n = 5, female n= 3], E16[male: n = 4, female: n = 4], E18[male: n = 3, female: n = 6], P0[male: n = 4, female: n = 5]). (B) Representative western of H3K79me3 levels in whole cortex from E14-P0 in male and female mice. (C) Developmental milestone achievements in male and female mice (male[control: n = 6, Dot1/ cKO: n = 11]; female[control: n = 4, Dot1/ cKO: n = 9], unpaired two-tailed t-test). (D) Weight during first two weeks post-birth in male and female mice (repeated measures ANOVA). (E) Time to right self in negative geotaxis assay during development in male and female pups (repeated measures ANOVA). (F) Total number of rears during the open field assay in male and female mice (male[control: n = 16, Dot1/ cKO: n = 16]; female[control: n = 14, Dot1I cKO: n = 14], unpaired two-tailed t-test). (G) Percent of time spent in the center of the open field arena in male and female mice (Kruskal-Wallis Test). (H) Percent of time spent in open arms of the elevated zero maze in male and female mice. (male[control: n = 9, Dot1/ cKO: n = 10; female[control: n = 14, Dot1/ cKO: n = 14], unpaired two-tailed t-test). (I) Total entries during the elevated zero maze in male and female mice (unpaired two-tailed t-test). (J) Total arm entries during the Y maze in male and female mice (unpaired two-tailed t-test). All box plot bounds indicate the 25th and 75th percentiles, the black line shows the median, and whiskers extend to the minimum and maximum value. * p < 0.05
(A) Total number of ultrasonic vocalizations (USVs) in P6 male and female pups (male[control: n = 15, Dot1! cKO: n = 14]; female[control: n = 21, Dot1I cKO: n = 19], Kruskal-Wallis Test). (B) Average volume in decibels (dB) of USVs in male and female pups (unpaired two-tailed t-test). (C) Average frequency in kilohertz (kHz) of calls in male and female pups (unpaired two-tailed t-test). (D) Percent of down calls out of total calls in male pups (Kruskal-Wallis Test). (E) Total activity measured from beam breaks during 10-minute open field assay in 4 week old male and female mice (male[control: n = 16, Dot1! cKO: n = 16]; female[control: n = 14, Dot1! cKO: n = 14], Kruskal-Wallis Test). (F) Percent of spontaneous alternations out of the total number of triads possible in male and female mice (unpaired two-tailed t-test). (G) Discrimination index between interaction time with the mouse or rock cylinder during the three-chamber social choice assay in male and female mice (unpaired two-tailed t-test). (H) Freezing activity after 2 weeks post-con textual fear conditioning in male and female mice (unpaired two-tailed t-test). All box plot bounds indicate the 25th and 75th percentiles, the black line shows the median, and whiskers extend to the minimum and maximum value that are no further than 1.5 * interquartile range. *p<0.05.
DISCUSSION
Here, we identified 11 individuals with DOT1L variants and NDDs, including developmental delays, ASD, and intellectual disability. We confirmed two missense variants disrupt methyltransferase activity of DOT1L which, along with two other variants that result in early stop codons, suggest that monoallelic loss-of-function of DOT1L can lead to observed phenotypes. We found dynamic regulation of H3K79me in post-mitotic cortical neurons and widespread transcriptional disruptions upon partial Dot1l loss in excitatory neurons both in vitro and in vivo. Further, loss of Dot1l alters neuron arborization, spine density, and expression of synaptic genes. In addition, we found that dot1l depletion in zebrafish increases activity in response to multiple sensory inputs. Finally, we show that both ubiquitous and neuron-specific monoallelic loss of Dot1l cause sex-specific vocalization disruptions and sociability deficits, with additional motor development deficits observed in full-body heterozygous mice. Together, this work demonstrates that partial Dot1l loss can lead to an emerging neurodevelopmental disorder and disrupt transcription, neuron morphology, and behavior.
We provide the first functional testing of two loss-of-function variants that indicate monoallelic loss of DOT1L causes neurological dysfunction. Recent work proposed that gain-of-function variants in DOT1L contribute to the neurodevelopmental disorders based on modeling in flies and human HEK293T cells (15). However, the fly ortholog of DOT1L, grappa, is highly divergent from human DOT1L (Catalytic domain: 65% identity, Whole gene: 24% identity, 35% similarity, 28% gaps) (38) and expression of wildtype human DOT1L did not rescue developmental defects caused by grappa loss. Thus, we sought to characterize additional DOT1L variants and model them in homologous systems. We confirmed one of the previously described variants does indeed function as a gain-of-function (p.E123K) supporting prior findings (15). However, we also found that other missense variants disrupt methyltransferase activity through multiple approaches. This, combined with modeling of partial Dot1l loss in multiple systems suggests that loss-of-function variants have profound functional consequences and are also likely causative in the identified neurodevelopmental disorder. Together, these findings place DOT1L in a growing group of epigenetic regulators for which either increases or decreases in function or expression can lead to neurodevelopment disorders (53, 54).
These findings point toward likely molecular changes and cell types responsible for the resulting phenotypes. Given the ample work illustrating the importance of DOT1L in early corticogenesis (24, 27), it is noteworthy that we did not detect major changes in cortical neuron identify following loss of just a single copy of Dot1l. Rather, we found robust changes in transcriptional programs, particularly in excitatory neurons, resulting in downregulation of genes related to synaptic function. In addition, we performed in-depth behavioral characterization of mice with monoallelic Dot1l loss in all tissue and in forebrain neurons. The behavioral overlap of altered early vocalization and sociability deficits suggests that DOT1L loss in neurons is at least partly responsible for behavioral deficits. Further, given that we detected sex-specific effects on both gene expression and behavior, our findings indicate that partial DOT1L loss has divergent effects based on sex. This is particularly intriguing given that our cohort includes more females than males, although whether this trend will be sustained as additional individuals are identified remains to be determined. Together, this work builds upon the previous modeling of Dot1l loss in the brain by indicating that partial loss of Dot1l is sufficient to cause changes in transcription, neuron maturation, and behavior and identifies transcriptional pathways and cell types that likely contribute to these deficits.
Several notable questions remain that will be critical to understanding the role of DOT1L in the brain and in neurodevelopmental disorders. Dynamic regulation of H3K79me is evident in primary cortical neuronal upon partial Dot1l loss fitting with prior work in the midbrain (32). However, it is unclear whether histone variant replacement or demethylase activity is responsible for H3K79me removal in the brain. Future work establishing where the deposition of this mark occurs in the neuronal genome and how this is affected by partial DOT1L loss will also be important for understanding its role in neuronal transcription. There is also conflicting evidence on whether methylation of H3K79 is required for DOT1L to fulfill its role in neuronal differentiation (29–31). Given that the majority of variants lie in the catalytic domain, our work suggests that H3K79me is important in neuron function. However, our findings also allow for a critical function for H3K79me to emerge after neuronal differentiation. Notably, several of the phenotypes that we detected in mouse models were evident in neuron-specific monoallelic loss of Dot1l suggesting this cell type is particularly sensitive to DOT1L dosage. Although we characterized the impact of partial Dot1l loss in the brain, whether transcriptional states of other cell types are also impacted in both mouse and human systems will be critical to understanding the role of DOT1L in contributing to developmental disorders. Lastly, predicted loss-of-function alleles for DOT1L are found within gnomAD that are not associated with notable phenotypes suggesting incomplete penetrance or the potential for attenuation of the disorder described through other unknown factors.
In summary, this work examines the impact of partial loss of Dot1l spanning from the transcriptional level to the behavioral level. This research provides insights into the effect of variants on DOT1L and the neuronal changes that may contribute to phenotypes observed in DOT1L loss-of-function variant individuals. Further, our findings expand on our understanding of DOT1L by demonstrating that disruption of a single copy of Dot1l is sufficient to disrupt neuronal function and contributes to an emerging neurodevelopmental disorder.
METHODS
Experimental design
The goal of this study was to examine the impact of monoallelic DOT1L variants in the brain. To accomplish this, we identified 11 individuals with variants in DOT1L with a spectrum of neurodevelopmental disorders. We assessed the methyltransferase activity of two previously published variants and one variant identified in our cohort to better understand how variants impact DOT1L function. In zebrafish, we defined early developmental behavioral disruptions upon dot1l loss. We then evaluated the impact of partial DOT1L loss on neuronal transcription and neuronal maturation in mouse primary cultured neurons. Further, we examined how monoallelic loss of Dot1l impacts transcription and behavior in mice using ubiquitous and forebrain specific depletion models.
Study Participants
Identification of DOT1L Variants
Variants in DOT1L were identified through connections made through collaborating clinicians, GeneMatcher (33), Deciphering Developmental Disorders Research Study, GeneDx, and the MSSNG (www.mss.ng) database. The first individual of interest (Individual E) was identified through a prior publication (6, 12). The remainder of participants were identified through GeneMatcher apart from the following: individual 2, individual 6, and individual F (identified through pre-existing collaboration), individual 3, 4, 8 (GeneDx), individuals A-C (MSSNG database). The initial GeneMatcher entry was made on May 23, 2023, and all matches were considered in this study submitted until April 2024 (table S3). Variants are reported according to Human Genome Variation Society (55) nomenclature in reference to the DOT1L transcript (NM_032482.3). Allele counts were gathered from gnomAD (v4.1.0), TOPMed Bravo, and RGC Million Exome Variant Browser (table S1, table S2). Pathogenicity of missense variants was predicted using an aggregation of the following databases: Metadome (56), Revel (57), and AlphaMissense (58) (table S1, table S2). Variants p.I85M (SCV004169212), E134K (SCV004169195), and Gln598* (SCV003804054) are available on ClinVar.
Ethical Statement
Human subject studies were approved consistent with the principles of research ethics and the legal requirements of the lead clinician authors’ jurisdiction(s) (The Hospital for Sick Children, Canada). Voluntary, informed consent was obtained from human participants, consistent with the institutional principles of research ethics and the legal requirements of each referring author’s jurisdiction. Ethical approvals were obtained for participation, phenotyping, sample collection and generation/derivation of affected individual and control fibroblasts (IRB#16-013278_AM118, Children’s Hospital of Philadelphia). The authors also confirm that human research participants provided written informed consent for publication of the images in Fig. 1.
Methyltransferase Activity
Expression and Purification of DOT1L and Mutants
DOT1L and mutants were expressed and purified as previously described (59). Briefly, the proteins were expressed in BL21 One Shot (DE3) (ThermoFisher) E. coli cells. They were grown at 37°C until reaching an OD600 equal to 0.6-0.8 and then were induced using 0.5mM IPTG for 3 hours at 37°C. The cells were harvested (Sorvall LYNX6000) and then lysed (AvestinEmulsiflexC3) (Lysis buffer: 500mM NaCl, 50mM Tris-HCl pH 8.0, 5% Glycerol, 5mM Imidazole, 2mM BME, 1x Protease Inhibitor). Lysate was incubated with Ni-NTA Beads (Qiagen). Protein was eluted (Elution buffer: 500mM NaCl, 50mM Tris-HCl pH 8.0, 5% Glycerol, 300mM Imidazole, 2mM BME) and cleaved by TEV protease overnight in dialysis (Dialysis buffer: 75mM NaCl, 20mM Tris pH 8.0, 5% Glycerol, 2mM BME). Sample was then purified over a HiTrap SP HP column (Cytvia) (Buffer A: 75mM NaCl, 25mM HEPES pH 7.5, 5% Glycerol, 2mM BME) and eluted with a linear salt gradient (75mM to 1000mM NaCl); and then further purified over HiLoad Superdex 200 16/600 size exclusion column (GE Healthcare) (150mM NaCl, 10mM HEPES 7.5, 2mM DTT). Protein was concentrated, then flash frozen in liquid nitrogen and stored in a -80°C freezer.
Purification of Widom 601 DNA
Widom 601 DNA was transformed into DH5a competent E. Coli cells (NEB) from a plasmid containing 8 copies of 147bp repeats flanked by EcoRV sites (60). The cells were grown overnight at 37°C, then harvested and lysed. The DNA was then further purified using established protocols (61).
Expression and Purification of Xenopus Histones
Xenopus laevis histones H2A, H2B, H3 and H4 were expressed and purified using previously published protocols (61). Briefly, the histone constructs were cloned in a pET-3 vector and grown in pLysS (DE3) cells (NEB) to an OD600 of 0.6-0.8 and induced at 0.5mM IPTG at 37°C for 3 hours. The protein was then extracted from inclusion bodies and purified over a size exclusion column Sephacryl S200 (Cytvia) followed by an SP anion exchange column (Tosoh). Proteins were then dialyzed in 1mM BME and then lyophilized using a Vertis Sentry lyophilizer.
Reconstitution of Nucleosomes
Unmodified nucleosome was assembled as previously described(60, 61). First equimolar ratios of unfolded histones H2A, H2B, H3, and H4 were mixed and dialyzed in refolding buffer. The assembled octamer was then purified over a size exclusion chromatography column Superdex 200 26/600(GE Healthcare) using refolding buffer. Nucleosomes were assembled by combining an equimolar quantity of octamer and Widom 601 DNA, followed by an overnight salt gradient dialysis using a peristaltic pump (Gilson).
Endpoint Methylation Assay
The endpoint methylation assay was performed as described (59). Assays on DOT1L and mutants with unmodified nucleosome were done in three replicates. Briefly, in methyltransferase buffer, 250nM and 125nM of DOT1L or mutants were combined with 1µM of nucleosome. In a volume of 20ml, the reaction was incubated at 30°C for 30min. The reaction was stopped using 5ml of 0.5% TFA. SAH production was determined using a MTase-Glo methyltransferase kit (Promega). The luminescence was measured using an EnSpire 2300 Multilabel plate reader (Perkin Elmer).
Western blotting
Protein lysates or histone samples were mixed with 5X Loading Buffer (5% SDS, 0.3M Tris pH 6.8, 1.1mM Bromophenol blue, 37.5% glycerol), boiled for 10 minutes, and cooled on ice. Protein was resolved by 4%–20% Tris-glycine or 16% Tris-glycine SDS-PAGE, followed by transfer to a 0.45-μm PVDF membrane for immunoblotting. Membranes were blocked for 1 hour at RT in 5% milk in 0.1% TBST and probed with primary antibody overnight at 4C. Membranes were incubated with secondary antibody for 1 hour at RT.
Cell culture
Human Fibroblasts
Individual 5 fibroblasts and age-and sex-matched control fibroblasts were donated from collaborating clinicians. Fibroblasts were cultured in DMEM (with 4.5 g L−1 glucose, L-glutamine and sodium pyruvate) supplemented with 15% FBS (Sigma-Aldrich, F2442-500ML) and 1% penicillin-streptomycin (Gibco, 15140122).
Neuro-2A cells
Neuro-2A cells were obtained from the American Type Culture Collection (ATCC), cultured in DMEM (with 4.5 g L−1 glucose, L-glutamine and sodium pyruvate) supplemented with 10% FBS (Sigma-Aldrich, F2442-500ML) and 1% penicillin-streptomycin (Gibco, 15140122) and maintained free of mycoplasma. N2A transfections were performed in DMEM using lipofectamine 2000 (Life Technologies, 11668027). Lipofectamine and DNA complexes were left on for overnight. Cells were harvested for analysis 2 days after transfection.
Primary neuronal culture
Cortices were dissected from E16.5 C57BL/6J embryos and cultured in neurobasal medium (Gibco 21103049) supplemented with B27 (Gibco 17504044), GlutaMAX (Gibco 35050061), penicillin-streptomycin (Gibco 15140122) in TC-treated twelve or six-well plates coated with 0.05 mg/mL Poly-D-lysine (Sigma-Aldrich A-003-E). At 3 DIV, neurons were treated with 0.5 µM AraC. Transfections were performed using lipofectamine 2000 (Life Technologies, 11668027). Neurons were put in a 1mM kynurenic acid solution during transfection to prevent excitotoxicity. Lipofectamine and DNA complexes were left on neurons for 15 minutes. Transfections were performed at 8 to 12 DIV for constructs expressing DOT1L and cells were fixed two to three days later. shRNA transfections were performed at 9 to 12 DIV and fixed three to four days later. Neuronal infections were transduced overnight with lentivirus containing the constructs described below. Virus was removed the following day, and neurons were cultured for 5-7 days.
Constructs
The GFP control plasmid was obtained from Addgene, pLenO-CMV-MCS-GFP-SV-puro (Addgene plasmid# 73582). The pET28-MHL-DOT1L (1–420) was received from the Armache lab originally purchased from Addgene (Addgene plasmid# 40736). The pDSV-DOT1l-HA-Flag-mRFP-nls and empty pDSV-mRFP-nls plasmids were received from the Vogel lab. The Sun1-GFP plasmid was a gift from Jeremy Nathans Lab, pCDNA3-CMV-Sun1-GFP-6xMyc. Dot1l shRNA and control Luciferase shRNAs were inserted into the pLKO.1 vector backbone (Addgene plasmid# 10878). Dot1l shRNA target sequences were as follows:
Dot1l shRNA 1: CCGGGTCCAGTTTGTACTGTCAATACTCGAGTATTGACAGTACAAACTGGACTTTTTG
Dot1l shRNA 2: CCGGCCTCGGTTTACACAGCTTCAACTCGAGTTGAAGCTGTGTAAACCGAGGTTTTTG
Dot1l shRNA 3: CCGGCGGCAGAATCGTATCCTCAAACTCGAGTTTGAGGATACGATTCTGCCGTTTTTG
DOT1L mutants were generated using sited directed mutagenesis using Pfu Turbo HotStart DNA polymerase (Agilent, 600322, for R292C and E123K) or NEB Q5 Polymerase (M0491S, for D157N), and primers were created using the DNA-based primer design feature of the online PrimerX tool or manually creating primers using ∼15 bp overlap strategy. Plasmid sequences were verified through Sanger sequencing and/or Plasmidsaurus long read sequencing.
Lentiviral production
HEK293T cells were cultured in high-glucose DMEM growth medium (with 4.5 g L−1 glucose, L-glutamine and sodium pyruvate), 10% FBS (Sigma-Aldrich F2442-500ML), and 1% penicillin-streptomycin (Gibco 15140122). Calcium phosphate transfection was performed with Pax2 and VSVG packaging plasmids. Viral media was removed 2 hours after transfection and collected at 48 and 72 hours later. Viral media was passed through a 0.45-μM filter and precipitated for 48 hours with PEG-it solution (40% PEG-8000 [Sigma-Aldrich P2139-1KG], 1.2 M NaCl [Fisher Chemical S271-1]). Viral particles were pelleted and resuspended in 200μL PBS.
RNA-sequencing
Library preparation & sequencing
RNA was isolated using Zymo Quick-RNA Miniprep Plus Kit (R1057). Libraries were generated using the Illumina TruSeq stranded mRNA library prep kit (Illumina 20020595). Prior to sequencing libraries were quantified by qPCR using a KAPA Library Quantification Kit (Roche 07960140001). Libraries were sequenced on an Illumina NextSeq 500/550; reads (75-bp read length, single end). Data can be accessed under the following GEO accession number: GSE279978.
Data processing and analysis
Reads were mapped to Mus musculus genome build mm10 with Star (v2.7.9a). The R packages DESeq2 (62) (v1.34.0) and limma (v3.50.3) via edgeR (v3.36.0) were used to perform differential gene expression analysis. We defined genes as differentially expressed where FDR < 0.05 and an absolute log2 fold change > 0.5. Volcano plots were generated using Enhanced Volcano. IGV tools (63) (v2.12.3) was used to generate genome browser views.
Gene ontology
PANTHER (64, 65) (v18.0) was used to perform an overrepresentation test against the biological process complete ontology using default parameters. SynGO (47) was used for synaptic gene ontologies and overrepresentation tests of differentially expressed genes. All expressed genes (defined as any gene that did not have an NA p adjusted value in the DeSeq2 output and did not have an NA gene name) was used as a background gene list.
Revigo
Revigo (66) was used to remove redundant terms and gather a concise list based on a published protocol(67). In brief, the Panther output of the Biological Process gene ontology terms and their associated FDR-corrected p-values were input into Revigo. Revigo input parameters used were: size of resulting list – small; remove obsolete GO terms – yes; species – Mus Musculus; semantic similarity measure – Resnik. Revigo output was then filtered using the following conditions: reference genes within a gene ontology term <= 3000, dispensability < 0.2, and fold enrichment > 1. The resulting top 10 gene ontology terms based on FDR-corrected p-values were displayed.
GSEA
The R package FGSEA (68) was used to perform pre-ranked gene set enrichment analysis (GSEA) based on log2 fold changes obtained from DESeq2 differential expression analysis. Genes without a defined adjusted p-value and genes with a base mean < 100 were removed prior to running GSEA. The GSEA (69, 70) database (https://www.gsea-msigdb.org/gsea/index.jsp) was used for synaptic transmission based gene sets.
Immunocytochemistry
GluA2 antibody (Synaptic Systems: 182103) was added to the media of live cells and incubated for 45 minutes. Cells were fixed in 4% PFA for 10 min and washed with PBS. Cells were blocked in blocking solution (PBS with 3% BSA and 2% serum) for at least 1 hour. Cell coverslips were then incubated with secondary antibody for 1 hour at room temperature. For detection of GluA2, Goat anti-Rabbit Alexa Fluor™ 647 (Thermo Fisher, A-21244; 1:500) was added to the secondary antibody solution. Nuclei were stained with DAPI (1:1,000 in PBS) for 10 min with washing in PBS. Coverslips were mounted onto microscope slides using ProLong Gold antifade reagent (Thermo Fisher).
Image acquisition
Cells were imaged on an upright Leica DM 6000, TCS SP8 laser scanning confocal microscope with 405-nm, 488-nm, 552-nm and 638-nm lasers. The microscope uses two HyD detectors and three PMT detectors. The objective used was a ×63 HC PL APO CS2 oil objective with an NA of 1.40. Type F immersion liquid (Leica) was used for oil objectives. Images were 175.91 × 171.91 µM2, 1,024 × 1,024 pixels and 16 bits per pixel. Coverslips were imaged with a z stack through the neuron.
Image analysis
Sholl, Spine, and GluA2 Analysis
Images were analyzed using ImageJ (v2.14.0/1.54f) software. A singular z stack image’s maximum projection of the GFP channel was generated. The image was traced in Simple Neurite Tracer (SNT) (71) and the Sholl analysis feature (72) was used to generate a data table with number of intersections per step size (radius step size = 10 μM). An R script (https://zenodo.org/records/1158612) was used generate graphs and summary statistics using a mixed effect model (73). Spine density was quantified from the maximum projection image. The three largest neurite branches were measured and projections from these branches were counted. Spines had to be > 0.4 μM and < 8 μM in order to be counted based off previous literature (74, 75). Spine density for each branch was calculated as number of spines/branch length and this value was averaged together for each neuron imaged. To quantify GluA2 levels, an in-house macro was created. In short, this macro creates an outline of the imaged neuron from each individual stack using the GFP channel and then measures the fluorescent intensity in that stack from the far-red channel used to stain GluA2. The fluorescent intensity of each image is normalized to the average intensity of the control transfected neurons.
Single nuclei RNA-sequencing (snRNAseq)
Nuclei Isolation
For each biological replicate, one cortical hemisphere of a mouse was dissected, and flash frozen in liquid nitrogen and stored at −80 °C. The nuclei isolation procedure used was modified from (76, 77). Tissue was homogenized in douncers using a loose pestle (∼10-15 strokes) in 1.2 mL of homogenization buffer supplemented with 1 mM DTT, 0.15 mM spermine, 0.5 mM spermidine, RNasin ® Plus Ribonuclease Inhibitor (Promega N2611), and EDTA-free protease inhibitor (Roche). A 5% IGEPAL-630 solution was added (107ul), and the homogenate was further homogenized with the tight pestle (∼10-15 strokes). The sample was then mixed with 1.3 mL of 50% iodixanol density medium (Sigma D1556) and added to a polypropylene thin wall tube (13.2 mL, Beckman and Coulter, 331372). The sample was then underlaid with a gradient of 30% and 40% iodixanol, and centrifuged at 10,000 x g for 18 minutes (no brake) in a swinging bucket centrifuge at 4 °C. Nuclei from control and Dot1l HET mice were individually counted (3 males, 3 females per genotype) and proportionally combined with all other biological replicates within each genotype. These samples were washed 3 times in DPBS and spun at 1000g for 5 min. Samples were resuspended in 1X Nuclei Buffer (10x Genomics PN-2000153/ 2000207) at a concentration of approximately 5,000 nuclei/ul for subsequent library preparation. All steps were performed on ice or at 4°C.
Library preparation & sequencing
For the generation of ATAC and Gene Expression libraries, the 10X Genomics Chromium Next GEM Single Cell Multiome ATAC + Gene Expression (CG000338 Rev F) protocol was followed. Briefly, 16,100 nuclei from each sample underwent a transposition reaction before being loaded on the 10X genomics Chromium controller to target 10,000 recovered nuclei per sample. The resulting barcoded transposed DNA and barcoded cDNA were then used to generate ATAC and gene expression libraries, respectively, following the manufacturer’s guidelines. Quality control was performed during library preparation using an Agilent Bioanalyzer and a Thermo Fisher Qubit. Prior to sequencing libraries were quantified by qPCR using a KAPA Library Quantification Kit (Roche 07960140001). Libraries were sequenced on an Illumina NextSeq 1000, using 28 cycles for Read 1, 10 cycles for the i7 index, 10 cycles for the i5 index, and 90 cycles for Read 2. Data can be accessed under the following GEO accession number: GSE279978.
Preprocessing of snRNAseq data
Paired end sequencing reads were processed using 10X Genomics Cellranger v5.0.1. Reads were aligned to the mm10 genome optimized for single cell sequencing through a hybrid intronic read recovery approach (78). In short, reads with valid barcodes were trimmed by TSO sequence, and aligned using STAR v2.7.1 with MAPQ adjustment. Intronic reads were removed, and high-confidence mapped reads were filtered for multimapping and UMI correction. Empty GEMs were also removed as part of the pipeline. Initial dimensionality reduction and clustering was performed prior to processing to enable batch correction and removal of cell free mRNA using SoupX (79). Raw expression matrices with counted, individual nuclei UMI and genes were used for subsequent steps and filtering by QC metrics.
Clustering and merging by genotype and comparison
Raw matrices for each individual genotype were converted to Seurat objects using Seurat 5.0.1 and filtered to remove UMIs with thresholds of > 200 minimum features, < 5% mitochondrial reads, and < 5% ribosomal reads. Each genotype (control and Dot1l HET, each containing 6 biological replicates) were merged to generate an object for the subsequent steps. Each dataset was normalized (NormalizeData) using the default scale factor of 10000, variable selection (FindVariableFeatures) was performed using 2000 features, then scaled and centered (ScaleData) using all features without regressing any variables. Dimensionality reduction with PCA (RunPCA) used the first 30 principal components and the nearest-neighbor graph construction (FindNeighbors) used the first 10 dimensions. Clustering (FindClusters) was next performed using a resolution of 0.5 before layers corresponding to each genotype were integrated (IntegrateLayers) using CCA Integration with a k weight of 60 and then rejoined (JoinLayers). The dataset per condition was then dimensionally reduced using the integrated CCA at with 30 dimensions (RunUMAP) and the same resolution of 0.5.
Marker gene identification
To identify marker genes for each cluster, differential expression analysis was performed using the Seurat function FindAllMarkers. Differentially expressed genes that were expressed at least in 25% cells within the cluster and with a fold change more than 0.5 (log scale) were considered marker genes. Cell identity was determined using well-established marker genes for major cortical cell types. Marker gene analysis led to the identification of 17 cortical neuron clusters (10 excitatory, 7 inhibitory), 1 subcortical neuron cluster, and 7 non-neuronal clusters. Neuronal clusters were annotated according to the cortical layer they occupy, or—if unidentifiable by cortical layer—according to the gene most differentially expressed in that cluster relative to all other excitatory or inhibitory neuronal clusters.
Differential gene expression analysis and parsing sex of nuclei
Differential gene expression analysis between control and Dot1l HET groups was performed using the Seurat function FindMarkers (min.pct =.001, logfc.threshold = 0.5) with a MAST test. Genes with an adjusted p-value < 0.05 and an absolute log2 fold change > 0.5 were considered differentially expressed between control and Dot1l HET. The sex of the nuclei was determined based on the following parameters: females were categorized as nuclei with Xist expression at or above the 70th percentile in comparison to total nuclei and with no expression of Eif2s3y or Ddx3y; males were categorized as nuclei with Xist expression below 70th percentile in comparison to total nuclei and non-zero expression of Eif2s3y or Ddx3y.
Gene ontology
PANTHER (64, 65) (v19.0) was used to perform an overrepresentation test against the biological process complete ontology using default parameters. All expressed genes (defined as any gene within the current Seurat object subset with a min.pct = .001) was used as a background gene list.
Revigo
Revigo (66) was used to remove redundant terms and gather a concise list based on a published protocol (67). In brief, the Panther output of the Biological Process gene ontology terms and their associated FDR-corrected p-values were input into Revigo. Revigo input parameters used were: size of resulting list – small; remove obsolete GO terms – yes; species – Mus Musculus; semantic similarity measure – Resnik. Revigo output was then filtered using the following conditions: reference genes within a gene ontology term <= 1000, dispensability < 0.1, and fold enrichment > 2. The resulting top 10 gene ontology terms based on FDR-corrected p-values were displayed.
Mice
A floxed Dot1l mouse line crossed with the Sun1-sfGFP line (JAX Strain #:030952) was received from Tanja Vogel. In brief, the floxed Dot1l mouse line is floxed at exon 2 causing a frameshift that results in an early stop codon and nonfunctional gene product (C57BL6/J background). The Dot1l mouse line was originally obtained from the Knockout Mouse Project (KOMP). Heterozygous floxed Dot1l mice (Dot1lfloxed/+;Sun1-sfGFP+/+) were crossed to the NEX-Cre line (80) for neuron-specific behavioral testing. Dot1l cKO were Dot1lfloxed/+;Sun1-sfGFP+/-;NEX-Cre+/- and controls were Dot1l+/+;Sun1-sfGFP+/-;NEX-Cre+/-. Heterozygous Dot1l mice that did not also harbor alleles from the Sun1-sfGFP line (Dot1lfloxed/+; Sun1-sfGFP-/-) were crossed to the CMV-Cre line (JAX Strain #:006054) for ubiquitous monoallelic Dot1l loss behavioral testing. Dot1l HET were Dot1lfloxed/+; CMV-Cre+/- and controls were Dot1l+/+;CMV-Cre+/-. All mice were housed in a 12-hour light-dark cycle and fed a standard diet. All experiments were conducted in accordance with and approval of the IACUC at the University of Pennsylvania.
Behavioral assays
Behavioral cohorts
Male and female controls (Dot1l+/+; Sun1-sfGFP+/-;NEX-Cre+/- or Dot1l+/+;CMV-Cre+/-), Dot1l HET (Dot1lfloxed/+;CMV-Cre+/-), and Dot1l cKO (Dot1lfloxed/+;Sun1-sfGFP+/-;NEX-Cre+/-) mice were tested in the behavioral tests described below: For control and Dot1l HET mice, two cohorts were generated a month apart from each other and used for developmental milestone testing from P1 – P18. Additionally, these mice were used at 4 weeks old at the onset of behavioral testing which included: elevated zero maze, open field, Y maze, social choice, and fear conditioning, in that order. The breakdown of these cohorts was as follows: Cohort 1[control: male = 4, female = 10, Dot1l HET: male = 2, female = 10], Cohort 2 control: male = 9, female = 5, Dot1l HET: male = 14, female = 13]. A third cohort of mice were used for ultrasonic vocalizations at P6 – P7. The breakdown was as follows: Dot1l HET cohort[litter = 13, control: male = 15, female = 22, Dot1l HET: male = 22, female = 22].
For control and Dot1l cKO mice, a cohort was generated for developmental milestone testing from P1 – P18 [control: male = 6, female = 4, Dot1l cKO: male = 11, female = 9]. A separate cohort [control: male = 16, female = 14, Dot1l cKO: male = 16, female = 14] of mice was used at 4 weeks old at the onset of behavioral testing which included: open field, Y maze, social choice, and fear conditioning, in that order. A third cohort of mice was used at four weeks old at the onset of behavioral testing for elevated zero maze [control: male = 9, female = 14, Dot1l cKO: male = 10, female = 14]. A fourth cohort of mice used for ultrasonic vocalizations at P6 – P7 [litter = 10, control: male = 15, female = 21, Dot1l cKO male = 14, female = 19]. For all behavioral testing, the experimenter was blinded to genotype of the mice.
Ultrasonic vocalizations
Multiple litters were used for both the Dot1l HET and Dot1l cKO cohorts. Pups at approximately P6 – P7 were individually placed into a soundproof chamber with fresh bedding. A Condenser ultrasound microphone (Avisoft-Bioacoustics CM16/CMPA, part #40011) microphone and UltraSoundGate 116H (Avisoft Bioacoustics, part # 41163, 41164) recording device was used with Avisoft-RECORDER USGH software. The recording sessions were 5 minutes in length and recorded with the following parameters: sampling rate = 375000 Hz, range = 15 – 180 kHz, and min whistle duration = 5 ms. USVs were analyzed using a MATLAB based software, VocalMat (81). Mice with fewer than 50 calls were excluded.
Elevated zero maze
The elevated zero apparatus consists of a circular shaped platform raised approximately 16 inches above the floor. Two opposing quadrants have raised walls (wall height = 4 inches, circle width = 2 inches) without a ceiling leaving these closed quadrants open to overhead light. The two remaining opposing quadrants were open (wall height = 0.25 inches). Mice were placed into a closed quadrant and allowed to freely explore for 5 minutes. The entire testing session was recorded, and videos were analyzed using ANY-maze software.
Open field
Mice were placed into an empty arena (15 inches x 15 inches) and allowed to freely explore for 10 minutes. Activity was measured using beam breaks recorded using Photobeam Activity System Open Field software (San Diego Instruments) and percent of center activity was quantified as number of beam breaks in the center / total beam breaks *100.
3-chamber social choice assay
The social choice test was carried out in a three-chambered apparatus, consisting of a center chamber and two outer chambers. Before the start of the test and in a counter-balanced manner, one end chamber was designated the social chamber, into which a stimulus mouse would be introduced, and the other end chamber was designed the nonsocial chamber. Two identical, clear Plexiglas cylinders with multiple holes to allow for air exchange were placed in each end chamber. In the habituation phase of the test, the experimental mouse freely explores the three chambers with empty cue cylinders in place for 10 min. Immediately following habituation, an age-and sex-matched stimulus mouse was placed in the cylinder in the social chamber while a rock was simultaneously placed into the other cylinder in the nonsocial chamber. The experimental mouse was tracked during the 10 min habituation and 10 min social choice phases. All testing was recorded, and videos were analyzed manually.
Y maze
The Y maze test was performed on a Y shaped apparatus composed of 3 enclosed arms equidistant apart (3 in wide x 5 in wall x 15 in long). Mice were handled for 2 minutes each on 3 consecutive days immediately prior to the onset of testing. For Y maze testing mice were placed at the distal end of the arm closest to the experimenter of the Y maze apparatus. Mice were allowed to freely explore for 8 minutes. Entries into each arm were defined as all four paws of the mouse entering. A spontaneous alternation was defined as a consecutive entry into each of the 3 arms without returning to the arm that the mouse had been in immediately prior. Spontaneous alternation triads over the total number of possible triads based on the total entries were calculated as spontaneous alternations/ (total entries -2). All testing was recorded, and videos were analyzed manually.
Contextual and cued fear conditioning
Mice were handled for 2 minutes each the day immediately prior to the onset of testing. On training day, mice were placed in individual chambers for 2 minutes followed by a loud tone lasting 30 second that co-terminated with a 2 second, 1.25-mA foot shock. One minute later mice received another tone-shock pairing and were then left undisturbed for an additional 1 minute in the chamber before being returned to their home cage. Freezing behavior, defined as no movement except for respiration, was determined before and after the tone-shock pairings and scored by MedAssociates VideoFreeze software. To test for context-dependent learning, we placed mice back into the same testing boxes 24 hours later for a total of 5 minutes without any tone or shock, and again measured the total time spent freezing. Following an additional 24 hours, we tested for cue-dependent fear memory by placing the mice into a novel chamber consisting of altered flooring, wall-panel inserts, and vanilla scent. After 2 minutes in the chamber, the cue tone was played for a total of 3 minutes, and the total time spent freezing during the presentation of this cue tone was recorded. Long-term contextual and cued fear memory were again tested with the same protocol at 14 days (contextual) or 15 days (cued) post-training.
Zebrafish experiments
Experiments were conducted on 6 dpf larval zebrafish (Danio rerio, TLF strain) raised in E3 medium at 29 °C on a 14:10 h light cycle. At this developmental stage the sex of the organism is not yet determined. Breeding adult zebrafish were maintained at 28 °C on a 14:10 h light cycle. Crispants were generated as described(42) by Kroll et al. Three gRNAs targeting three different regions across the dot1l locus were designed using ChopChop v3 (https://chopchop.cbu.uib.no/). Custom Alt-R CRISPR-Cas9 crRNAs (IDT) were annealed with tracrRNA (IDT, #1072533) to form gRNAs which were subsequently complexed with Cas9 protein (IDT, #1081061) to make the final ribonucleoprotein (RNP) complex. Three non-targeting crRNAs (IDT, #1072544, 1072545, 1072546) were used to make the RNP for controls. Single-cell wildtype (TLF) zebrafish embryos were then microinjected within 15minutes of fertilization with 1nl of RNP mix containing 357pg (10.1fmol) of each gRNA and 5029pg (30.5fmol) of Cas9. Embryos displaying acute toxicity or damage from microinjection were removed from analysis. The remaining embryos were raised to 6dpf at which point they were arrayed on a 100-well plate and multiple sensorimotor behaviors including the visual motor response, responsiveness to flashes of light or darkness, and the acoustic startle response were assessed as described previously(39). To confirm that each of the three gRNA-Cas9 RNP complexes was able to target the predicted dot1l locus and cause mutations, genomic DNA was also extracted from dot1l crispants at 6dpf. The predicted target sites were amplified by PCR using primers that flank the region, and the PCR product was then sent for Sanger sequencing. Each of the three RNPs caused mutations at the predicted target site that were not present in control injected embryos.
Sequences
Statistical analysis
All statistical analyses were performed using readily available code in R. Number of replicates and details of statistical tests are reported in figure legends. Shapiro-Wilk’s method was used to test for normality of a given dataset. Detailed information on statistical tests as well as all relevant test statistics can be found in table S4.
Data Availability
The exome/genome sequencing will be made available upon request provided that privacy and consent criteria are preserved. RNA-sequencing and single nucleus RNA-sequencing data generated in this study can be accessed under the following GEO accession number GSE279978. Variants p.I85M (SCV004169212), E134K (SCV004169195), and Gln598* (SCV003804054) are available on ClinVar. All data are available in the main text or the supplementary materials. Any additional data will be made available within two weeks upon request to corresponding author.
https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE279978
Funding
NIH NINDS grant 1F31NS129242 (MM). NIH/NICHD grant P50 HD105354, Intellectual and Developmental Disabilities Research Center (research support). NIH NINDS grant 1R01NS134755 (EK). NIH NIMH grant 1DP2MH129985 (MB, EK). NIH NIMH grant R00MH111836 (MB, EK). Klingenstein-Simons Fellowship from the Esther A. & Joseph Klingenstein Fund (MB, EK). Simons Foundation (MB, EK). Alfred P. Sloan Foundation Research Fellowship FG-2020-13529 (MB, EK). Brain and Behavior Research Foundation NARSAD Young Investigator Award (MB, EK). SickKids Research Institute (ARD). Azrieli Precision Child Health Platform (ARD). NIHR Manchester Biomedical Research Centre NIHR203308 (SB). MRC Epigenomics of Rare Diseases Node MR/Y008170/1 (SB). Miguel Servet program from Instituto de Salud Carlos III, Spain CP22/00141 (DNB). Netherlands Organisation for Scientific Research ZonMw Vidi, grant 09150172110002 (TSB). EpilepsieNL (TSB). CURE Epilepsy (TSB). NIH NINDS grant K08NS135125 (PDC). University of Pennsylvania Autism Spectrum Program of Excellence (ASPE) (JM, PDC). ANID-Chile Fondecyt grant #1211411 (GMR, VF). "Joan Oró" of the Secretary of Universities and Research of the Department of Research and Universities of the Government of Catalonia with code 2024 FI-1 00075 (BEA). European Union (BEA).
Author contributions
MM designed, performed, and analyzed most experiments. MB performed in vitro RNA-sequencing and immunocytochemistry. KL supported mouse work and analyzed behavioral tests and immunocytochemistry data. ARD gathered clinical information. PC and JM performed zebrafish experiments. RL performed methyltransferase activity experiments. AC performed p.D157N overexpression experiment. AP supported single nucleus RNA-sequencing work. VF, GMR, CM, ALS, CP, GMSM, RS, TSB, CMR, JL, IA, DNB, CO, BEA, FL, KC, AG, JL, XL, AV, AMI, XY, SB, KV, MJ, MK, PS, CIGM, SB, and JLM provided variant clinical information. MG led zebrafish experiments. KA led methyltransferase activity experiments. GC led clinical information compilation. EK led the project.
Competing interests
KV has received honoraria as an advisory board member, travel expenses and speaker fees from Biogen, Santhera, Orchard, ITF and Novartis, outside the submitted work. JLM is an employee of and may own stock in GeneDx, LLC. All other authors declare they have no competing interests.
Data and materials availability
The exome/genome sequencing will be made available upon request provided that privacy and consent criteria are preserved. RNA-sequencing and single nucleus RNA-sequencing data generated in this study can be accessed under the following GEO accession number GSE279978. Variants p.I85M (SCV004169212), E134K (SCV004169195), and Gln598* (SCV003804054) are available on ClinVar. All data are available in the main text or the supplementary materials. Any additional data will be made available within two weeks upon request to corresponding author.
Acknowledgments
We thank the patients and their families for sharing data and samples. The authors wish to acknowledge the resources of MSSNG (www.mss.ng), Autism Speaks and The Centre for Applied Genomics at The Hospital for Sick Children, Toronto, Canada. We also thank the participating families and clinicians for their time and contributions to this database, as well as the generosity of the donors who supported this program. We thank Dr. Tanja Vogel for sharing DOT1L;Sun1-sfGFP mouse line and Dr. Andrea Stout for microscopy support. Behavioral procedures were performed at the Neurobehavior Testing Core at the University of Pennsylvania. Single nucleus RNA-sequencing library preparation was performed by the Single Cell Core at Children’s Hospital of Pennsylvania. We thank the Bhoj lab for providing fibroblasts.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.
- 8.
- 9.↵
- 10.↵
- 11.
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.
- 26.
- 27.↵
- 28.↵
- 29.↵
- 30.
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
