Mutations of the histone linker H1–4 in neurodevelopmental disorders and functional characterization of neurons expressing C-terminus frameshift mutant H1.4

Abstract

Rahman syndrome (RMNS) is a rare genetic disorder characterized by mild to severe intellectual disability, hypotonia, anxiety, autism spectrum disorder, vision problems, bone abnormalities and dysmorphic facies. RMNS is caused by de novo heterozygous mutations in the histone linker gene H1–4; however, mechanisms underlying impaired neurodevelopment in RMNS are not understood. All reported mutations associated with RMNS in H1–4 are small insertions or deletions that create a shared frameshift, resulting in a H1.4 protein that is both truncated and possessing an abnormal C-terminus frameshifted tail (H1.4 CFT). To expand understanding of mutations and phenotypes associated with mutant H1–4, we identified new variants at both the C- and N-terminus of H1.4. The clinical features of mutations identified at the C-terminus are consistent with other reports and strengthen the support of pathogenicity of H1.4 CFT. To understand how H1.4 CFT may disrupt brain function, we exogenously expressed wild-type or H1.4 CFT protein in rat hippocampal neurons and assessed neuronal structure and function. Genome-wide transcriptome analysis revealed ~ 400 genes altered in the presence of H1.4 CFT. Neuronal genes downregulated by H1.4 CFT were enriched for functional categories involved in synaptic communication and neuropeptide signaling. Neurons expressing H1.4 CFT also showed reduced neuronal activity on multielectrode arrays. These data are the first to characterize the transcriptional and functional consequence of H1.4 CFT in neurons. Our data provide insight into causes of neurodevelopmental impairments associated with frameshift mutations in the C-terminus of H1.4 and highlight the need for future studies on the function of histone H1.4 in neurons.

Introduction

Rahman syndrome (RMNS, also referred to as HIST1H1E syndrome, MIM #617537) is a rare neurodevelopmental disorder recently reported. Symptoms of RMNS can vary but include mild to severe intellectual disability (ID), hypotonia in newborns and infants, anxiety, autism spectrum disorder (ASD), vision problems, brittle bones and teeth and dysmorphic facial features including high hairline, wide-set eyes and low-set ears (1–8). A total of about 40 individuals have been reported worldwide since the disorder was originally characterized (1). Given this small number of patients, the full spectrum of RMNS clinical presentations remains to be fully delineated.

RMNS is caused by heterozygous de novo pathogenic variants in the single exon H1–4 gene (previous gene symbol HIST1H1E, OMIM:142220, NM_005321), which encodes the linker histone H1.4. Most pathogenic variants identified are small insertions or deletions (indels) that shift the codon reading frame. Because this reading frame does not have an immediate stop codon after the indels, these mutations result in truncated proteins that not only lack the normal C-terminal domain of the 219-amino acid protein but also generate a shared, novel C-terminal Frameshifted Tail (H1.4 CFT) of approximately 40 amino acids (1,3,5). Data from other groups raise the possibility that cellular pathophysiology in RMNS could arise from gain-of-function or dominant negative effects of the mutant protein (5).

Histones function in the nucleus to organize genomic DNA, with DNA wrapping around octamers of the core histones (H2A, H2B H3 and H4) in a repeating structure called the nucleosome. The H1 histone linker proteins flank nucleosomes at the DNA entry and exit sites, where they are thought to play a role in chromatin compaction (9). The positive charge of the histone H1 linkers allows these proteins to neutralize and bind to negatively charged DNA (10). The shared H1.4 CFT has a net negative charge, potentially disrupting this interaction (3). However, the functional importance of this charge shielding for chromatin structure or gene regulation is not well understood.

H1–4 is just one member of the histone H1 family, which is composed of 11 members in humans. The histone H1 family includes both germ cell-specific and somatic subtypes, which are expressed in nearly all cells, including neurons (11). The somatic histone H1s are divided into replication-dependent (H1–1 through −5) and replication-independent (H1–0, H1-x) subtypes based on their preferential expression and presumed incorporation into chromatin during specific phases of the cell cycle (12). Histone H1.4 is a 219-amino acid protein that shares > 80% amino acid identity with each of the four other replication-dependent histone H1s (10). The high homology among the H1s suggests common functions for H1 family members, which may explain why knockout mice lacking single H1s, including knockout of the mouse H1f4 gene (the homolog of human H1-4), fail to show obvious developmental phenotypes (13). At the cellular level, members of the histone H1 family have been implicated in several gene regulatory processes including DNA methylation, chromatin compaction and regulation of repressive histone modifications (12,14). Consistent with possible disruption of chromatin regulation in RMNS, one group reported a distinct methylation profile in peripheral blood samples of a cohort of affected individuals compared with healthy controls (15). However, whether H1.4 CFT results in neurodevelopmental phenotypes including ASD and ID via effects on chromatin, and more importantly how H1.4 CFT acts in neurons to disrupt brain function, remain entirely unknown.

To better understand the pathophysiology underlying H1.4 CFT, we gathered a cohort of 7 individuals with C-terminus mutations and examined brain phenotypes. We also determined the transcriptional and cellular consequences of exogenous wild-type (WT) and H1.4 CFT expression in primary rat neurons, a context that models the predicted gain-of-function effects of the endogenous mutation. Our data reveal that the H1.4 CFT disrupts synaptic gene expression and physiology in postmitotic neurons, raising the possibility that aberrant function of H1.4 CFT within neurons of the brain may contribute to neurological deficits in patients with RMNS.

Results

Genetics and phenotypes of subjects with RMNS

To expand our understanding of the spectrum of H1–4 sequence variants and clinical phenotypes associated with RMNS, we identified seven individuals with heterozygous sequence variants in the H1–4 gene from clinical exome sequence (ES) (Fig. 1) (see Supplementary Material, Fig. S1 for histone H1 family gene nomenclature). Consistent with prior reports, most variants in the H1–4 gene were clustered in a 100 bp region between c.360 and c.450 in the C-terminal domain of the transcript and result in frameshift mutations predicted to produce mutant proteins that have a strongly negative net charge compared to WT H1.4 protein. Six of the variants in our cohort are recurrent and have been reported by other groups, but one variant (c.392dupC in subject #4) is newly reported here.

Figure 1

Summary of variants in H1–4. The H1–4 gene encodes a 219-amino acid protein composed of three domains: a short N-terminal tail, a globular domain and a long C-terminal tail. Amino acid positions delineating the protein domains are shown as blue arrows. Above the gene schematic (black boxes) are mutations detected in the patients recruited for this study. Solid black boxes indicate that the individual was previously reported by another group. Black box with red outline indicates patients that have not been reported by other groups. Mutations that have previously been published are drawn under the peptide schematic, with color of the box corresponding with the publication reporting the mutation. (Purple (6); Green (1); Pink (4); Auburn (38); Red (3); Grey (2); Blue (5); Yellow (7)).

The clinical features we collected from a review of medical records and parent-completed questionnaires are summarized in Figure 2A and Supplementary Material, File S2. RMNS subjects were originally characterized as having a syndrome of ‘Intellectual disability and overgrowth’ (1). However, only two subjects have macrocephaly in our cohort (Fig. 2B). These data are consistent with observations reported in Takenouchi et al. (4), which also reported a subject with a H1.4 CFT lacking overgrowth, and suggest that overgrowth is not an obligate finding for RMNS. Also, consistent with previous reports (2,5), the major features of subjects carrying C-terminal frameshift mutations in our cohort included dysmorphic facies (7/7, 100%), ID (7/7, 100%); ASD (3/4, 75%); motor delay (7/7, 100%) and vision issues (6/6, 100%). Despite the preponderance of neurodevelopmental features in this cohort, our review of their health records failed to reveal a consistent brain architectural defect among the six individuals in this study who had magnetic resonance imaging (MRI) (Supplementary Material, File S2). A prior study of 30 subjects with H1.4 CFT reached a similar conclusion (2) and reported highly varied findings among the 14 subjects with MRI data. The common findings of brain MRI include hypoplastic corpus callosum, delayed myelination and ventriculomegaly (Supplementary Material, File S2) (2,5).

Figure 2

Clinical presentation of subjects in this study. For subjects in this study (Subjects 1–7), medical and health history was collected, and guardians were asked to self-report in a survey covering additional health and development metrics. (A) In the table, gray boxes indicate that the phenotypes reported by parents do differ from that of a typically developed child. White box indicates that respective questions were reported by parents as unremarkable or not able to assess (NR). The prevalence of previously published phenotypes is also included. (B). Comparison of growth parameters of our subjects against CDC clinical growth charts (percentile). Numbers correspond to growth parameters for each subject. Thick gray bar indicates average growth percentile, thin grey bar indicates SD.

To determine whether H1–4 mutations might associate with neurodevelopmental and other phenotypes independent of an RMNS diagnosis, we conducted a systematic review of the Baylor Genetics database (one of the largest and non-public clinical ES databases) of ~17 000 individuals undergoing clinical ES. This revealed 15 variants in the UTRs, 11 in-frame amino acid deletions, 403 missense variants and 94 silent variants in addition to individuals described in Figure 1. Among individuals undergoing clinical chromosomal microarray analysis, we also identified three individuals with phenotypes of developmental delay or ASD who had large copy number variant (CNV) gains of 565-602 kb in regions encompassing the H1–4 gene. Whether these variants are causative for ASD/ID in the subjects and/or suggestive of a dosage effect of H1.4 is unknown but may be relevant for further study.

The clinical consequences of haploinsufficiency resulting from heterozygous H1–4 variants in humans have not been described. Therefore, we sought to examine the phenotypes of individuals with possible haploinsufficiency of the H1–4 gene. We were unable to identify individuals with H1–4 CNV losses in the Baylor clinical genetics ES database. We did, however, identify and gather clinical information for three individuals with predicted loss-of-function variants in the N-terminus of H1–4 (Supplementary Material, Fig. S2). It is noted that one of these variants is a de novo c.1A > T substitution that disrupts the ATG start codon. Two other variants are a maternally inherited c.100insT (p.K34Ifs^*13) and a de novo c.265delA (p.S89Afs^*140). These three variants are predicted to be loss of function. But it is of note that the variant of c.265delA (p.S89Afs140) is predicted to result in a long frameshifted tail that is in a different reading frame from the C-terminus of H1.4 CFT. It is also not clear whether the c.265delA (p.S89Afs140) may result in unstable mRNA or H1.4 protein. From the clinical information we obtained, these individuals are phenotypically distinct from those diagnosed with RMNS as they lack the common features of infantile hypotonia, and delayed motor skills associated with H1–4 CFT mutations. Because these are single cases for each loss of function variant, the pathogenicity of these variants remains to be determined. However, these initial clinical findings provide additional support for the hypothesis that H1.4 CFT carries a gain-of-function, rather than a loss-of-function, of the WT protein.

Conserved frameshifts and net charge changes in H1 family

The cluster of RMNS associated C-terminal frameshift mutations in H1–4 raises the question of why similar disease-associated mutations are not found in other histone H1 family members. To investigate the consequence of indels in H1–1 through −5, we computationally introduced 1 bp indels throughout the length of the H1–1, H1–2, H1–3, H1–4 and H1–5 gene sequences and determined the consequence of these frameshifts on the both the length of the protein translated and the change in charge of the resulting peptide compared with the WT protein (Fig. 3). We found that all histone H1 family members could produce out-of-frame proteins of various sizes if indels were introduced at any one of several positions (Supplementary Material, Fig. S3). However, the frameshift mutations in the H1–4 RMNS C-terminal hotspot were unique for their ability to result in peptides with a large negative change (Fig. 3). We also observed that insertions in a similar region of H1–5 could have a lesser but potentially significant effect on protein charge. However, when we queried the clinical ES database at Baylor College of Medicine for frameshift variants in H1–5 that were associated with neurodevelopmental phenotypes, we found none. There were nine individuals with frameshifts in H1–1 (three of these mutations were inherited, five were not tested for origin) and two individuals with frameshift variants in H1–2 (Supplementary Material, File S3); however, these mutations had minimal effects on net protein charge. Taken together, these data are consistent with the hypothesis first proposed by Tatton-Brown et al. that pathogenic mutations in the C-terminus of H1–4 are those that disrupt the positively charged C-terminal tail, impairing the ability of this H1 histone to interact with negatively charged DNA (12,14).

Figure 3

Frameshift modeling of the H1 family of genes. R was utilized to systematically create 1 bp duplications (left) or 1 bp deletions (right) throughout the length of the H1–1 (red), H1–5 (yellow), H1–2 (green), H1–3 (blue) and H1–4 (purple) genes, translate the sequences until a stop codon was reached and calculate net charge change from a WT protein of corresponding length. Frameshift mutations found in the clinical ES database are points outlined in black. Black box around H1–4 points shows previously published H1.4 CFT frameshift mutations (2,5).

Altered distribution of H1.4 mutant protein in nuclei of rat hippocampal neurons

A core phenotype of individuals with H1.4 CFT is neurodevelopmental impairment; however, the consequences of H1.4 CFT expression on the structure or function of neurons is completely unknown. To elucidate the possible impact of H1.4 CFT on neurons, we compared the consequences of exogenously expressing either WT or a frameshift mutant H1.4 protein (c.430dupG), the most common H1–4 variant found in RMNS in cultured primary embryonic rat neurons (Fig. 4A). Immunostaining of cells infected with lentivirus containing the Myc-tag of the dual FLAG-myc-H1.4 constructs showed that most cells expressed the exogenous human proteins (Fig. 4B). Since these neurons also express the endogenous WT protein, our model mimics the heterozygous expression of H1–4 CFT in human RMNS. Consistent with previous experiments in other cell types (5), we found that both WT H1.4 and H1.4 CFT were detected exclusively in the nucleus (Fig. 4B). However, the percent overlap between H1.4 and DAPI signal, which shows the distribution of genomic DNA, was lower in neurons expressing H1.4 CFT (Fig. 4C). These data suggest there is a difference either in nuclear morphology or in the distribution of mutant histone H1.4 compared with WT within the nucleus. Indeed, the nuclear size of neurons expressing H1.4 CFT is significantly larger than that of WT (Fig. 4D). Yet even controlling for size, we found that H1.4 CFT had a greater localization towards the periphery of the nucleus compared with the distribution of the WT protein (Fig. 4E). These results indicate a potential large-scale change in chromatin or nuclear organization associated with the expression of H1.4 CFT.

Figure 4

Exogenous expression of FLAG-myc-Tagged H1.4 WT and H1.4 CFT in rat primary neurons. (A) Western blotting of whole cell extract of primary neurons with anti-FLAG antibody. (B) Representative images of DIV7 rat hippocampal neurons exogenously expressing FLAG-myc-WT H1.4 or frameshift mutant FLAG-myc-H1.4 CFT (c.430dupG mutation in H1.4) (stained with myc in red) is expressed in the nucleus of neurons (labeled with Map2 in green). (C) CellProfiler Quantification of percent overlap between Myc and the DAPI signal of neurons (P < =  0.000001) and (D) of nuclear size from DAPI signal (P = 0.0015). N = 3 experiments, n = 234 WT H1.4 nuclei, n = 511 mutant H1.4 nuclei. Unpaired two-tailed t-test, ^** < = 0.005, ^**** < = 0.0001. (E) Representative images of pixel intensity of myc signal (H1.4) across the diameter of neurons exogenously expressing WT H1.4 or H1.4 CFT. Error bars represent SD.

RMNS frameshift mutant H1.4 disturbs transcriptional programs in hippocampal neurons

To determine if the distinct nuclear distribution of H1.4 CFT resulted in corresponding changes in gene expression, we performed bulk RNA-seq from biological replicates of DIV7 rat hippocampal neuron cultures expressing WT H1.4, H1.4 CFT or GFP as a control (n = 4 WT or H1.4 CFT, n = 3 GFP). All samples had similar quality control metrics (Supplementary Material, Fig. SA–C). By principal component analysis analysis, the GFP and WT samples clustered together, whereas H1.4 CFT expressing neurons clustered on their own, indicating a specific effect of the mutant H1.4 protein on the overall transcriptional program (Supplementary Material, Fig. S4D). We also observed similar overall gene rank comparisons for H1.4 CFT verses either GFP or WT H1.4, again suggesting a selective effect of H1.4 CFT on overall gene expression programs (Supplementary Material, Fig. S4E).

When gene expression levels between the treatment conditions were compared pairwise using DESeq (16), only nine genes were significantly differentially expressed comparing neurons expressing GFP with those expressing the exogenous WT H1.4 protein (Supplementary Material, File S3). These data indicate that differences in the level of H1.4 protein expression do not impact gene expression. By reverse transcription and quantitative PCR (qRT-PCR) we quantified expression of the endogenous rat H1f4 and compared it with the exogenous human H1–4. We found that both that rat and the human were expressed at similar levels relative to the housekeeping gene Gapdh (Fig. 5A). The H1.4 CFT transcript was expressed at higher levels than that of WT H1.4. However, when we probed against the FLAG tag of the dual FLAG-myc-H1.4 constructs and quantified western blots of rat neurons expressing WT or CFT H1.4, we found no significant difference in the amount of protein (Fig. 5B), suggesting that transcripts for CFT H1.4 may be more stable than those of WT H1.4. Prior studies have shown that knockout of single H1 family members leads to upregulation of other H1s, potentially compensating for the loss (13). The replication-dependent H1s are not polyadenylated and therefore are poorly represented in our sequencing data. Thus, to determine whether rat H1 family members changed expression upon exogenous expression of human H1.4, we used qRT-PCR to quantify the expression of the rat homologs of human H1–1, −2, −4 and −5. Our data show that exogenous expression of human H1–4 was associated with reduced expression of rat H1f2 and H1f5 compared with neurons expressing only GFP (Fig. 5C). Taken together with the knockout data (13), these findings suggest that the replication-dependent histone H1 subfamily have the ability to compensate for changes in H1 expression levels. These data are important because they validate the model of exogenous expression as a specific means to find the aberrant functions of H1.4 CFT independent of any general consequences of H1.4 overexpression.

Figure 5

RNA-seq results from rat hippocampal neurons. RNA from DIV 7 rat hippocampal neurons exogenously expressing WT H1.4, H1.4 CFT (with mutation c.430dupG), or a GFP infection controls were sequenced. (A) Relative mRNA expression of exogenous human H1–4 WT or CFT or endogenous rat H1f4 mRNA in rat hippocampal neurons (p = 0.4375). Paired t-test. (B) Quantification of exogenous WT and CFT H1.4 protein expression in rat hippocampal neurons as in Figure 4A. FLAG signal from the dual FLAG-myg tagged H1.4 constructs is normalized to actin in the same lane (P = 0.2971); n = 7 (C) Expression of H1f1, −f2, −f4 and −f5 in rat neurons. Rat H1–3 (H1f6) is testes specific and not somatic. One-way analysis of variance (ANOVA) [H1f1 p = 0.054, H1f2 P = 0.000008 (RM: GFP versus WT P = 0.497, WT versus CFT P = 0.0023), H1f4 P = 0.24, H1f5 P = 0.022 (WT versus GFP P = 0.033)]. One-way ANOVA with multiple corrections, ^* < =  0.05, ^** < =  0.005, ^*** < =  0.0005, ^**** < =  0.0001. (D) Volcano plot of transcripts identified by RNA-seq. Statistical cutoffs used were Log2FoldChange > = 1 or −1, and P < = 0.5. Transcripts differentially expressed are shown in red, non-significant genes are shown in gray. Transcripts that met the Log2Fold change but not P-value cutoffs are shown in green, transcripts that met P-value cutoffs but log2FoldChange are shown in blue. (E) Significant GO terms of differentially expressed genes, graphed by −Log (10) of P-value for each term. (F) Expression of synaptic genes (left) and receptors (right) differentially expressed between DIV 7 rat hippocampal neurons exogenously expressing WT H1.4 or H1.4 CFT. (Grin3a P = 0.00029, Nrn1 P = 0.0051, Lrrtm2 P = 0.0041, Slittk2 P = 0.00086, Cadps2 P = 0.00066, Rasgtf2 P = 0.0035, Tacr3 P = 0.000003, Oxtr P = 0.0014, Hctr2 P = 0.049, Mrap2 P = 0.0012, Nbpwr1 P = 0.00042, Cckbr P = 0.0002). Unpaired two-tailed t-test, ^* < = 0.05, ^** > = 0.005, ^*** > = 0.0005, ^**** > = 0.0001.

By contrast to the seeming tolerance of rat neurons to exogenous WT H1.4 expression, we found a total of 398 differentially expressed genes comparing WT with H1.4 CFT expressing neurons. In total, 249 genes were significantly elevated in H1.4 CFT versus WT, and 149 genes were significantly lower in H1.4 CFT compared with WT (Fig. 5D, Supplementary Material, File S4). A nearly identical set of differentially expressed genes was observed in comparing control GFP expression to H1.4 CFT expression (Supplementary Material, Fig. S5A, Supplementary Material, File S4), again suggesting the transcriptional consequences are selective to expression of H1.4 CFT. These gene expression differences did not arise from distinctions in the cellular composition of the cultures, because we found no significant difference in typical gene markers of neurons or astrocytes, the two major cell types in dissociated cultures, when comparing neurons expressing GFP, WT H1.4 or H1.4 CFT (Supplementary Material, Figs. S5B and C).

Gene ontology (GO) analysis revealed enrichment among the differentially expressed genes for the Biological Processes of synaptic signaling, biological adhesion, behavior, locomotion and regulation of ion transport (Supplementary Material, Fig. S5D). We were especially interested to see that H1.4 CFT expressing neurons showed reduced expression of a set of gene products that function in synaptic transmission and downstream signal transduction, as well as a set of neuropeptide receptor encoding genes that are implicated in behavioral modulation (Fig. 5F). Taken together these data suggest that expression of H1.4 CFT in postmitotic neurons is sufficient to disrupt expression of specific set of genes that contribute to interneuronal communication and intraneuronal signaling.

H1.4 CFT disturbs action potential frequency and synchrony in hippocampal neurons

Given the disruption of neuronal gene expression programs we observed by RNA-seq, to determine if the expression of H1.4 CFT results in functional changes in neurons, we first assessed dendritic morphology, which determines key aspects of synapse formation between neurons. We did not find any difference in length and number of primary or secondary neurites at DIV7 between neurons expressing either exogenous WT or H1.4 CFT (Fig. 6A). In addition, Sholl analysis performed on these data did not note a difference in number of crossings of cell neurites for the neurons exogenously expressing H1.4 CFT compared with neurons expressing WT protein (Fig. 6B).

Figure 6

Effects of frameshift mutant H1.4 CFT (with c.430dupG) on firing rate and synchrony in rat hippocampal neurons. (A) Neurite outgrowth tracing shows no difference between number of primary dendrites (P = 0.248), length of primary dendrites (P = 0.511), number of secondary neurites (P = 0.926) or length of secondary neurites (0.835). N = 3 experiments, n = 74 WT H1.4 neurons, n = 75 CFT H1.4 neurons. Unpaired two-tailed t-test. (B) Sholl analysis of these traces is not significant (P = 0.2344), repeated measures two-way ANOVA, Tukey correction. N = 3 experiments, n = 76 WT H1.4 neurons, n = 75 CFT H1.4 neurons. Error bars represent SD. (C) Representative images showing neurons at DIV14. Red = Myc (H1.4) or GFP, Green = MAP2, Blue = DAPI. (D) Representative traces of action potential (AP) waveforms from single electrodes of a 16 MEA recording. The average waveform is shown in black. (E) Raster plots showing spontaneous firing of APs in WT H1.4, H1.4 CFT and GFP expressing neurons over a 100-s time period. Each row represents an electrode, each dash on the plot indicates a spike. Synchronous spikes, in which multiple electrodes have activity simultaneously, are colored in blue. Non-synchronous spikes are colored in black. (F) Quantification of firing rate and synchrony throughout development in H1.4 CFT expressing neurons compared with WT H1.4 and GFP expressing neurons. N = 3 experimental plates, n = 4–6 replicas of each condition per plate. Mixed effects analysis with multiple comparisons, for multiple comparisons: WT versus H1.4 CFT ^* < = 0.05, ^** < = 0.01, ^*** < = 0.001, GFP versus H1.4 CFT # < = 0.05, ## < = 0.01, ### < = 0.001, brackets denote P-value summary of overexpression (firing rate P = 0.0006, Synchrony P = 0.0017).

To determine if there is a functional consequence of H1.4 CFT for neural network activity, we measured action potential spike frequency and synchrony in cultured hippocampal neurons plated on 16 electrode multielectrode arrays (Fig. 6C). We detected spiking with similar waveforms regardless of whether neurons were infected with the WT H1.4, H1.4 CFT or GFP (Fig. 6D and E). For neurons expressing either WT H1.4 or GFP as control, both firing rate detected at a single electrode and the synchrony of firing between electrodes increased from DIV 8–13, which are consistent with increased connectivity and synaptic maturation over this period (Fig. 6F, left). However, neurons expressing H1.4 CFT showed significantly lower levels of firing with a delayed rise throughout days in culture and significantly reduced synchrony when compared to WT H1.4 and GFP groups (Figure 6F, right). Thus, these data provide the first evidence that H1.4 CFT disrupts neuronal physiology.

Discussion

In this study we report H1–4 mutations and clinical phenotypes for several new subjects, furthering our understanding of the genetics of RMNS and the scope of neurodevelopment related comorbidities to RMNS. Importantly, we show that expression of mutant H1.4 CFT in neurons disrupts neuronal firing and synaptic gene expression programs, providing a novel insight into the underlying causes of neurodevelopmental impairments in RMNS. Taken together these data significantly expand our understanding of how mutations in H1–4 lead to RMNS.

One of the most significant challenges in understanding RMNS is unraveling the neurodevelopmental pathophysiology observed in individuals with H1.4 CFT. However, this also presents an important opportunity for elucidating the functions of chromatin regulation in the developing brain. Like RMNS, many non-syndromic cases of ID and ASD have been shown to arise from de novo pathologic variants in chromatin regulators (3,17). The other major class of ID/ASD-associated genes are synaptic proteins; however whether there is a privileged relationship between ID/ASD-associated chromatin regulators and synaptic development or function is poorly understood (18). There is limited evidence of brain structural abnormalities in the small number of RMNS patients that have been studied by MRI (2), although the numbers of subjects evaluated remains small. Identification of additional subjects with pathogenic variants in H1–4 will provide a larger pool for future evaluation of brain development and expand the spectrum of clinical phenotypes.

Our systematic modeling of frameshift mutations across the closely related H1–1 through −5 genes confirmed that C-terminus frameshift mutations in H1–4 are unique in their impact on the charge of the resulting truncated proteins. This may also explain why similar frameshift variants in other histone H1 genes have not been implicated as causative in other human genetic disorders, despite the evidence from mouse knockout studies that suggest functions of this subfamily of histone H1 proteins are substantially overlapping (19). The positive charge of the histone H1 proteins is thought to be important for shielding the negative charge of the linker DNA at the entry and exit points of nucleosomes (12). Our data further support the hypothesis that changes in charge of the C-terminus contributes to the pathophysiology of H1–4 frameshift mutations, although whether DNA charge shielding, alteration of protein–protein interactions, changes in protein structure or some other molecular change is at the root of the downstream cellular consequences of the frameshift requires further study.

The histone H1 proteins organize chromatin structure in genomic DNA and are presumed to play a role in coordinating transcription (20). Surprisingly little is known about the transcriptional functions of histone H1.4 in large part because knockout of this single H1 family member is functionally compensated by other H1s (13). To fill this gap in knowledge, we took advantage of the genetic evidence that H1.4 CFT may act in a gain of function manner to profile gene expression changes induced by expressing either exogenous WT or H1.4 CFT in dissociated cultures of embryonic rat hippocampal neurons. Our data show selective impairment of neuronal gene expression programs including those that are related to synaptic function and neuropeptide signaling, which are strong candidates to underlie the neurological features of associated with H1–4 frameshift mutations. For example, GRIN3A is a component of synaptic NMDA-type glutamate receptors that modulates activity-regulated gene transcription in neurons and functions to control critical period timing (21,22). Experience dependent translation of the secreted protein neurogranin, NRGN, has been implicated in memory formation (23). LRRTM2 is a postsynaptic binding receptor for the presynaptic ASD/ID associated adhesion protein encoded by Nrxn1 (24). Rare variations in the secretory protein CADPS2 gene are associated with ASD (25). Finally, the oxytocin receptor OXTR stands out among the neuropeptide receptors for the clinical association of oxytocin with neuropsychiatric disorders (26). While we were able to identify these changes, the molecular mechanism underlying how these transcriptional changes are created remains to be elucidated. Future studies should make use of protein and chromatin immunoprecipitation methods to determine protein–protein interactions and genome wide distribution changes in neurons expressing H1.4 CFT.

We further show that exogenous expression of H1.4 CFT in primary embryonic rat hippocampal neurons in culture reduces action potential firing, demonstrating that the coordinate disruption of these genes induced by H1.4 CFT is sufficient to alter key aspects of neuronal physiology. These data represent the first evidence that H1.4 changes gene expression and function in postmitotic neurons. Importantly, the fine tuning of the synaptic wiring diagram has been implicated in neurodevelopmental disorders of ID and ASD including those associated with mutations of chromatin regulators (27,28), so understanding how a single-gene mutation in H1–4 can lead to synaptic deficits will have relevance for the broader field. Future studies examining H1.4 CFT in transgenic mice have potential to facilitate broader understanding of disease pathophysiology in RMNS.

Materials and Methods

Subjects

This study was approved by the Duke University Institutional Review Board and the Yale University School of Medicine. Written consent was obtained from all subjects. Clinical ES reports and medical records (including neuropsychological reports) were collected from individuals. Families also completed a brief questionnaire regarding clinical features and medical issues associated with subjects. For each subject, mutations of H1–4 reported by clinical genetics diagnostic laboratories were reviewed and confirmed by a clinical geneticist.

In Silico frameshift predictions in H1 genes

Human mRNA sequences for each H1 family member (H1–1 through −5) were obtained from Gencode GRCh38/hg38 Assembly. R programming software was used to model single base pair duplications and deletions into the H1–1 through −5 coding regions. Please see https://github.com/WestLabDuke/H1.4-repository for source code and more information.

Exogenous lentiviral expression of WT and H1.4 CFT

We generated a dual FLAG-myc-tagged WT and c.430dupG frameshift mutant H1–4 cloned into a modified FUGW vector (Addgene plasmid #14883 (29)). The FLAG-myc tag was placed at the N-terminus in frame with the ATG, consistent with constructs used in prior studies (30–32) as not to disrupt the C-terminus. A GFP expressing plasmid was used as a control (Addgene cat#53188). Lentivirus was produced in HEK293T cells using lentiviral packing plasmids p-CMV-VSV-G (Addgene plasmid #8454) and pCMV-dR8.2dvpr (Addgene plasmid #8455 (33)). Briefly, lentiviral packing and overexpression plasmids were transfected into 293 T cells using calcium phosphate transfection. Supernatant was collected 3 days post transfection and virus was concentrated by ultracentrifugation. For neuron experiments, viruses were functionally titered in primary rat hippocampal neuron cultures. Rat neurons were infected on DIV1 at a multiplicity of infection of 1 in Basal Medium Eagle (BME) (Sigma B1522) with 0.4 μg/ml added polybrene.

Primary rat hippocampal neuron culture

Hippocampi of CD IGS E18.5 rat embryos (Charles River Laboratories) were dissected as previously described (34). Both male and female embryos were used for all experiments, and all experiments were conducted in accordance with an animal protocol approved by the Duke University Institutional Animal Care and Use Committee. For cultures, briefly, hippocampi were removed and immediately placed in cold Ca²⁺ and Mg²⁺-free HEPES-buffered solution (ThermoFisher Cat:14175–095). Tissue was then trypsinized in 3 ml TrypLE Express (ThermoFisher Cat:12604013) for 15 min at 37°C. Trypsin-containing buffer was then replaced with BME (Sigma B1522) supplemented with Glutamax (ThermoFisher Cat:35050061), 10% fetal bovine serum and penicillin/streptomycin and triturated using flame-narrowed Pasteur pipettes. Cells were plated at a density of 120 k/well on coverslips coated with poly-D-lysine (Sigma P7280) and laminin (Sigma L2020). On day in vitro (DIV) 1, the media was exchanged with Neurobasal medium (ThermoFisher Cat:21103049) supplemented with B-27 (ThermoFisher Cat:17504044), Glutamax and penicillin/streptomycin.

Western blot

Cells were lysed in sodium dodecyl sulfate plus beta-mercaptoethanol, boiled for 10 min to denature and run on a 12% polyacrylamide gel. Blots were transferred for 1.5 h at 300 mA constant on to a nitrocellulose membrane. Membranes were blocked for 1 h in 5% milk/TBS + 1% Tween, probed with primary antibody (mouse anti-Flag (M1), Sigma F3165, 1:1000 or mouse anti-actin, Millipore MAB1501, 1:10000) overnight in blocking solution, rinsed and incubated with infrared conjugated secondary antibodies (Biotium) for 2 h at room temperature (RT). Blots were imaged using the Li-Cor Odyssey (9120) system.

Immunocytochemistry

Primary rat hippocampal neurons were cultured on coverslips coated with poly-d-lysine and laminin. At DIV 7 or DIV14, cells were fixed in 4% paraformaldehyde for 10 min at RT and rinsed with phosphate buffered saline. Neurons were permeabilized in PBS + 0.2% Triton-X and blocked in 10% Goat serum prior to incubation with primary antibody (chicken anti-MAP2 1:3000 Millipore AB5543; mouse anti-Myc 1:200 Sigma M4439) overnight at 4°C. Cells were then washed and incubated with fluorescent secondary antibody (Invitrogen) for 1 h at RT. Coverslips were then counterstained with DAPI and mounted on slides. Images were captured on a confocal microscope (Leica SP8).

Nuclear morphology analysis and Sholl analysis

Microscopy images were blinded before all analysis. CellProfiler (35) was used for nuclear morphology analysis to determine overlap of myc and Hoechst signal in each nuclei and cell size. Representative images of nuclear distribution were produced using FIJI ImageJ (36) by quantifying intensity across the diameter of each nucleus. For Sholl analysis, neuron projections were traced using FIJI ImageJ and analyzed using the Sholl plugin.

RNA sequencing

DIV7 hippocampal neurons exogenously expressing WT or mutant H1.4 CFT were lysed directly in Trizol Reagent (Invitrogen cat# 15596026). RNA was purified using Direct-zol mini-prep kit (Zymo Research cat#R2050) with on column DNAse treatment. RNA was sent to Novogene (CA, USA) for quality control and sequencing. 150 bp paired-end sequencing was performed on the NovoSeq 6000 instrument with a minimum of 20 M reads per sample. Sequence read quality was ensured and reads were aligned to the Rattus norvegicus (mRatBN7.2) reference genome. Differentially expressed genes were identified using DESeq2 (16) with padj < = 0.5 and log2FoldChange > = 1. Differentially expressed genes in the RMNS mutant was determined by determining DE genes between WT and mutant H1.4, then removing DE genes which appeared in WT versus GFP infection control (n = 9 genes). Source code can be found at https://github.com/WestLabDuke/H1.4-repository and FASTQ files have been deposited at GEO accession number GSE180609. Gene set enrichment analysis was performed using MSigDB and GO gene sets (37).

Reverse transcription and quantitative PCR

RNA DIV 7 rat hippocampal neurons (see RNA-seq sample preparation) were isolated using Trizol Reagent. RNA was treated with DNAse (NEB cat# M0303) prior to cDNA synthesis (Invitrogen SuperScript III cat#18080051) with random hexamers (H1 transcripts are predicted to not be polyadenylated) per manufacturers protocols. qRT-PCR was performed on a QuantStudio3 machine (ThermoFisher Scientific) using SYBR green master mix (ABI cat#4667659). Primers for qRT-PCR are listed in Supplemental Material, File S1. Primers for rat H1f1 (human H1–1), rat predicted transcript H1f2 (human H1–2), rat H1f5 (human H1–5), exogenous human H1–4 (spanning from the FLAG tag to the protein coding region) and endogenous rat H1f4. To compare the relative expression of endogenous rat H1f4 and exogenous human H1–4 in the hippocampal expression assays, we first corrected each primer pair for the relative efficiency (e) of amplification as calculated from the slope of a standard curve (e = 10^-1/slope). We then used this efficiency to impute the quantities of each transcript [adj Ct = log2(e^CT)]. Values were normalized for Gapdh in the same sample as a control for sample handling. Presence of a single amplicon was confirmed by running qRT-PCR products on agarose gel.

Multi-electrode array recordings

E18.5 CD IGS rat hippocampal neurons were plated in a 48-well multi-electrode array (MEA) plate (Lumos 48, Axion Biosystems) at a density of 60 000 cells/well. Wells were coated with poly-d-lysine and laminin prior to plating. Cells were infected on DIV 1 and half-media changes were performed at DIV 10 and 13 with Neurobasal culture medium supplemented with B-27, Glutamax and penicillin/streptomycin. Extracellular recordings were performed at 37°C with 5% CO₂ using the Maestro MEA system with AxIS software (Axion Biosystems). Data were acquired at 12.5 KHz and was filtered with a Butterworth bandpass filter of 200 Hz–3 kHz. Spike detection was performed to detect action potentials with a 6× standard deviation (SD). Electrodes were considered active if they displayed more than 5 spikes/min. Recordings were collected for 5 min every day from DIV 8 to 14 and data were analyzed for spike frequency and synchrony index using Axion Biosystem’s Neural Metrics Tool.

List of Abbreviations
 
• RMNS

  Rahman syndrome.

 
• H1.4 CFT

  H1.4 protein with mutant C-terminal Frameshift Tail.

 
• EBV

  Epstein–Barr Virus.

 
• ESC

  Embryonic stem cells.

 
• ASD

  Autism spectrum disorder.

 
• ID

  Intellectual disability.

 
• WT

  Wild type.

 
• qRT-PCR

  Real-time quantitative reverse transcription PCR (polymerase chain reaction).

Data and Statistical Analysis

All data are expressed as mean ± SD. Analyses were performed with Prism 8 statistical package (Graphpad, San Diego, CA). Figure legends contain experiment numbers (N), group sizes (n) and statistical test performed for each experiment. The experimenters were blind to sample genotype and treatment prior to data collection and analysis.

Acknowledgements

The authors would like to thank the subjects described in this study and their families. We also thank: Melyssa Minto for computational assistance, Fowzan S. Alkuraya and Hanan E. Shamseldin for confirming the de novo variant in subject #10, Janet Wooton for assistance in gathering clinical data from patients and their families and Scott Soderling for sharing MEA equipment. Thanks David Allis and Alexey A. Soshnev for valuable discussion.

Conflict of Interest statement. The Department of Molecular and Human Genetics at Baylor College of Medicine receives revenue from clinical genetic testing conducted at Baylor Genetics.

Funding

National Institute of Health (1R01NS098804 to A.E.W., MH098114 to YH.J., MH104316 to YH.J., HD087795 to YH.J); the HIST1H1E Genetic Syndrome Research Foundation (to YH.J).

Authors’ contributions

M.W.T. and YH.J. designed the human clinical data portion of study. YH.J., J.A.R., H.S., W.C., N.B. and S.A.T. collected patient genetic and phenotypic data. M.W.T., M.V.G. and A.E.W. designed the neuron experiment portion of the study. M.W.T., M.V.G., W.D.T., B.M.G. and A.I.A. performed all experiments. M.W.T., M.V.G., A.E.W. and Y.H.J. wrote the manuscript. All authors revised the manuscript and approved the final version for publication. Please direct queries regarding clinical data and phenotypes to [email protected]; wet lab experimental questions to [email protected]