MSL2 ensures biallelic gene expression in mammals

Abstract

In diploid organisms, biallelic gene expression enables the production of adequate levels of mRNA^1,2. This is essential for haploinsufficient genes, which require biallelic expression for optimal function to prevent the onset of developmental disorders^1,3. Whether and how a biallelic or monoallelic state is determined in a cell-type-specific manner at individual loci remains unclear. MSL2 is known for dosage compensation of the male X chromosome in flies. Here we identify a role of MSL2 in regulating allelic expression in mammals. Allele-specific bulk and single-cell analyses in mouse neural progenitor cells revealed that, in addition to the targets showing biallelic downregulation, a class of genes transitions from biallelic to monoallelic expression after MSL2 loss. Many of these genes are haploinsufficient. In the absence of MSL2, one allele remains active, retaining active histone modifications and transcription factor binding, whereas the other allele is silenced, exhibiting loss of promoter–enhancer contacts and the acquisition of DNA methylation. Msl2-knockout mice show perinatal lethality and heterogeneous phenotypes during embryonic development, supporting a role for MSL2 in regulating gene dosage. The role of MSL2 in preserving biallelic expression of specific dosage-sensitive genes sets the stage for further investigation of other factors that are involved in allelic dosage compensation in mammalian cells, with considerable implications for human disease.

Main

Sexually reproducing organisms inherit one copy of each chromosome from each parent, resulting in a diploid state in the somatic cells of the offspring. The majority of genes exhibit balanced expression from both paternal and maternal alleles^{4,5,6,7,8,9,10,11,12}.

Haploinsufficient genes exhibit obligately biallelic expression because two transcribing copies of the gene are necessary to produce a functional amount of protein^13,14,15. Loss of expression from one of the two alleles is sufficient to result in diseases^13,14,15.

In flies and mammals, males are the heterogametic sex exhibiting hemizygosity of X-linked genes. Dosage compensation is required to adjust allelic expression of X-linked genes to compensate for differences in gene dosage between the sexes. In mammals, one X chromosome is inactivated in females^6,16, whereas, in flies, the MSL histone acetyltransferase complex upregulates transcription of the single male X chromosome to match the expression levels of the two X chromosomes in females^17,18,19.

MSL2, a component of the MSL complex, interacts with X-linked long non-coding RNAs to determine specificity for the single male X chromosome in flies¹⁷. It has been proposed that the conserved function of MSL2 across dipterans and mammals involves dosage regulation of developmental genes²⁰. To date, conventional gene expression analysis has been insufficient to comprehend the full range of MSL2 function.

Hybrid mouse cell line models

To examine the role of MSL2 in gene dosage regulation in mammals, we used hybrid mouse embryonic stem (ES) cells. Male cell lines were derived from reciprocal CAST/EiJ mother × C57BL/6 father (CaBl) or C57BL/6 mother × CAST/EiJ father (BlCa)²¹ and female cell lines were derived from CAST/EiJ mother × C57BL/6 father (CaBl) or 129S1/SvImJ mother × CAST/EiJ father (9sCa) crosses⁵ (Fig. 1a). Given the high genetic similarity between the C57BL/6 and 129S1/SvImJ mouse strains (Methods), we refer to these crosses as reciprocal. Wild-type (WT) hybrid ES cells were differentiated into neuronal progenitor cells (NPCs) (Fig. 1a) and the single-cell-derived Msl2 knockout (KO) was generated independently in ES cells and NPCs using CRISPR–Cas9 (Fig. 1a and Supplementary Table 1). The Msl2 KO was validated at both the RNA and protein levels (Extended Data Fig. 1). The catalytic core of the MSL complex, MOF, is also a component of the KANSL complex²². The levels of KANSL-complex members and pluripotency markers remained unchanged in Msl2 KO ES cells (Extended Data Fig. 1a,b,d,e,g). In agreement with previous findings^19,20, histone modifications, such as H4K16ac, which is added by the MSL complex^18,23, exhibited minimal changes after the loss of MSL2 in ES cells and NPCs (Extended Data Fig. 1a,c,d,f,g–i).

An MSL2 mutant (H64Y), which was reported to abolish MSL2’s ubiquitin ligase activity²⁴, showed disrupted binding to MOF and MSL1 (Supplementary Fig. 2a–c). Expression of known MSL2 targets^20,25 was comparably reduced in both Msl2-KO and Msl2^H64Y mutant cells (Supplementary Fig. 2d), supporting the idea that the observed effects were specific to the loss of MSL2 function.

Allele-specific gene expression analysis

After establishing WT and Msl2-KO hybrid cell lines, we performed RNA-sequencing (RNA-seq) analysis. Quality control of the cells was performed using karyotyping analysis (Methods and Supplementary Fig. 2f,g). We performed three types of analyses: (1) standard (non-allele-separated) differential expression analysis; (2) allelic differential expression analysis of WT versus Msl2 KO to determine the individual gene expression changes for allele 1 (C57BL/6 or 129S1) and allele 2 (CAST); (3) allele-specific differential expression analysis to identify genes with differential expression specific to a single allele. For each gene, we calculated the allele-specific log₂-transformed fold change (log₂[FC]) by dividing the allele-2 fold change by the allele-1 fold change obtained by allelic differential expression analysis (Methods).

Standard differential expression analysis identified more than 1,100 differentially expressed genes in ES cells (Extended Data Fig. 2a,b) and more than 2,500 differentially expressed genes in NPCs (Fig. 1b and Extended Data Fig. 2b). In general, there was a notable similarity among all NPCs (Extended Data Fig. 2c,d). Given that the MSL complex activates transcription^18,23, we focused on downregulated genes. In 60–80% of cases, both alleles showed similar downregulation (Extended Data Fig. 2e). Gene ontology (GO) analysis suggested a role for MSL2 in regulating essential neuronal differentiation and brain development genes exclusively in NPCs and not in ESCs (Extended Data Fig. 2f,g). Notably, the frequency of genes showing allelic bias was higher in NPCs (>800 genes) than in ES cells (around 350 genes) (Fig. 1b and Extended Data Fig. 2a,e,h). Many of these were missed by the standard differential expression analysis (Fig. 1b (red dots)).

We classified NPC genes with standard or allele-specific downregulation into five distinct categories; 300 random genes showing no gene expression changes after Msl2 KO were used as the control category (Fig. 1c,d, Methods, Extended Data Fig. 3a,b and Supplementary Fig. 3). The majority of downregulated genes was classified as bi-to-bi-down genes, exhibiting biallelic expression in WT and biallelic downregulation in Msl2-KO cells. These genes mostly showed log₂[FC] values of >−1 allelic downregulation (Extended Data Fig. 3c). Another class of genes, monoallelic allele 1 or allele 2 (A1/A2) to none (mono^A1/A2-to-none), comprised genes that were initially expressed monoallelically and were silenced after Msl2 deletion. Most genes in this class exhibited log₂[FC] values of ≤−2, indicating complete loss of expression (Extended Data Fig. 3c). Notably, a class of genes that we named bi-to-mono^A1/A2 genes was initially biallelically expressed and became monoallelic after Msl2 deletion. Bi-to-mono genes are particularly interesting because most of them were borderline affected in the standard analysis, failing to reach the significance threshold (Fig. 1b and Extended Data Fig. 3a). However, separating the alleles revealed a substantial change in gene expression with log₂[FC] values of ≤−2 on one allele, indicating near-total loss of monoallelic expression, whereas the other allele remained unaffected (Extended Data Fig. 3c). These findings highlight the advantage of using a hybrid system, as these categories of genes would have been overlooked by conventional differential expression analysis. Our subsequent analyses focused on bi-to-mono genes with log₂[FC] values of less than −2. To verify that the bi-to-mono changes were not due to NPC subcloning, we performed RNA-seq analysis of three additional WT NPC clones (Supplementary Fig. 4). In conclusion, allele-specific differential expression analysis revealed a new class of MSL2-regulated genes.

Deciphering the bi-to-mono switch in Msl2 KO

Clustering revealed that approximately 80% of bi-to-mono genes showed consistent expression changes in at least two NPCs, indicating a high degree of reproducibility across cell lines (Fig. 2a and Supplementary Table 2). Given the importance of biallelic expression for haploinsufficient genes, we compared curated lists of haploinsufficient genes in humans (Supplementary Table 3) to bi-to-mono genes in NPCs (Fig. 2a (pink genes)). While only 9% of genes in the mouse genome were haploinsufficient, 21–22% of bi-to-mono genes exhibited haploinsufficiency in each NPC line (Extended Data Fig. 4a). The majority of MSL2-regulated haploinsufficient genes displayed high haploinsufficiency scores (Methods) and a notable proportion was associated with human neurological disorders (Fig. 2b and Extended Data Fig. 4b). Most MSL2-regulated haploinsufficient genes were not annotated as triplosensitive (Fig. 2b and Extended Data Fig. 4c), but were intolerant to loss-of-function mutations (Extended Data Fig. 4d), suggesting that they are more responsive to reduced, rather than increased, dosage…