MSL2 ensures biallelic gene expression in mammals

In diploid organisms, biallelic gene expression enables the production of adequate levels of mRNA1,2. This is essential for haploinsufficient genes, which require biallelic expression for optimal function to prevent the onset of developmental disorders1,3. 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.


Supplementary Fig. 2 MSL2 RING domain mutant in male CaBl ESCs and karyotyping analysis of all ESCs and NPCs. (a)
Western blot analysis of MSL2, MOF and NANOG levels in parental WT, 2 Msl2 -KO and 3 MSL2 H64Y mutant male CaBl ESC clones.RNA POL II serves as a loading control.(b) Western blot analysis of indicated histone modifications in parental WT, 2 Msl2 -KO and 3 MSL2 H64Y mutant male CaBl ESC clones.H3 serves as a loading control.(c) Immunoprecipitations (IP) of MOF, MSL2 and IgG control using nuclear lysates from parental WT, 2 Msl2 -KO and 2 MSL2 H64Y mutant male CaBl ESC clones.3% Inputs (lanes 1, 5, 9, 13 and 17), IgG IPs (lanes 2, 6, 10, 14 and 18), MOF IPs (lanes 3,7,11,15,and 19) and MSL2 IPs (lanes 4,8,12,16 and 20)  (f ) Comparison of the CPM (Counts per million) of Bisulfite-Seq data for allele 2 (A2) to allele 1 (A1) on each chromosome of male CaBl (left), male BlCa (middle) and female 9sCa (right) WT (top) and Msl2 -KO (bottom) NPCs.Theoretically, if the copy number of each allele is the same, the CPM value from allele 1 and allele 2 should be equal.If that is not the case and one allele showed more coverage than the other allele, it indicates that a chromosome copy number difference occurred.Chromosomes with abnormal copy numbers were defined if the fold change of allele-2 CPM / allele-1 CPM was beyond 0.8 to 1.2.As an example, for the male CaBl NPC Msl2 -KO clone, the fold change (allele-2 CPM / allele-1 CPM) was between 0.8-1.2 for most autosomes except for Chr1.Thus, the entire Chr1 was omitted from the male CaBl NPCs RNA-seq DE analysis.
(g ) For cell lines without BS-seq, differential expression (DE) analysis comparing allele 2 to allele 1 with RNAseq are performed to identify the allele biased genes on each chromosome of female CaBl NPCs (first), male CaBl ESCs (second), male BlCa ESCs (third) and female 9sCa ESCs (forth) WT (top) and Msl2 KO (bottom).The number of allele-1-biased genes and allele-2-biased genes should be equal if the copy number of each allele is the same.Otherwise, one allele will show much more biased genes than the other allele.Chromosomes with abnormal copy numbers were defined if the log2[FC] of allele-2 DEgene number / allele-1 DEgene number was beyond -1 to 1.As an example, for the male BlCa ESC Msl2 -KO clone, the log2[FC] (allele-2 CPM / allele-1 CPM) was between -1 to 1 for most autosomes except for Chr19.Thus, the entire Chr19 was omitted from the male BlCa ESCs RNA-seq DE analysis.Supplementary Fig. 3 Comparison of expression between WT and Msl2 KO of differentially expressed genes in male CaBl/BlCa and female 9sCa NPCs.(a-c) Scatterplots of the normalized counts (log2) of WT vs Msl2 KO for allele 1 (A1) (left) and allele 2 (A2) (right) for each gene within the categories bi-to-mono , bi-to-bi-down and random (n=300) genes for female 9sCa (a) and male BlCa (b) and CaBl (c) NPCs.(c) Heatmap showing allelic gene expression of three additional WT clones for bi-to-mono A2 (top) and bi-to-mono A1 genes (bottom) for male CaBl (left) and male BlCa (middle) and female 9sCa NPCs (right).Percentages of biallelically expressed genes compared to all expressed genes in each bi-to-mono category are indicated.

Supplementary Fig. 4 Comparison of additional WT clones and
Monoallelic expression can be random in different NPC clones derived from ESCs 4,5,28 .Consequently, the biallelic expression state of certain genes can vary across different clones.Nevertheless, monoallelic expression remains stable during cell division 4,5,28 .Hence, our strategy for generating Msl2 KO involved selecting a single clone from a mixed population of differentiated NPCs and performing CRISPR/Cas9 editing on the pure population of that clone (Supplementary Fig. 4a).This approach ensures that any observed changes from biallelic to monoallelic expression upon Msl2 KO are not influenced by NPC subcloning.To validate this, we picked 3 more WT NPC clones for male CaBl and BlCa and female 9sCa NPCs and performed RNA-seq.Our analysis revealed that, for each cell line, 50%-70% of the bi-to-mono genes exhibited biallelic expression in each additional wild-type (WT) sample (Supplementary Fig. 4b,c).This analysis confirms that the observed bi-to-mono changes are due to MSL2 loss rather than NPC subcloning.

Supplementary Fig. 5 scMultiomics data and escapee analysis of female 9sCa and
CaBl WT and Msl2 -KO NPCs.(a) Heatmap illustrating allelic gene expression of X chromosome genes in 10 WT NPCs.RNA-seq data from a total of ten female WT NPC clones, including three clones from a previously published study, four single-cell clones differentiated from 9sCa ESCs, and three single-cell clones differentiated from CaBl ESCs generated in our lab were used.Escapees in each clone were identified using the following criteria: inactivated X allele normalized counts > 10 and 0.1 < (WT inactivated X allele)/(allele-1 + allele-2) < 0.9.All facultative escapees were further categorized into common in over 5 NPC clones and in less than 5 NPC clones.(b) Heatmap showing allelic gene expression of all XCI escape genes in female WT and Msl2 -KO 9sCa (left) and CaBl (right) NPCs.In female 9sCa and CaBl NPC clones, 131 and 106 of the expressed X-chromosomal genes were identified to escape XCI, respectively.By applying a stringent cutoff for gene expression change (log2[FC] <-2), we identified 19 and 3 genes (labeled in black box) out of these two gene lists as high confidence MSL2-regulated escapees in female 9sCa and CaBl NPC clones, respectively.(c) UMAPs of female 9sCa WT and Msl2 -KO NPC single cell data based on batch effect corrected RNA (left) and chromatin accessibility analysis (right) respectively (see Methods for batch effect correction).The total number of cells analyzed per condition is indicated in the figure.Cells are colored by their sample names.WT2 in female 9sCa NPCs is the same as WT2 in Supplementary Fig. 4 b,c .(d) Pearson correlation of gene counts between bulk RNA-seq and scRNA-seq in female 9sCa WT and Msl2 -KO NPCs.scRNA-seq gene counts were calculated by merging the total counts of all cells per gene.(e) Normalized counts of RNA expression and chromatin accessibility of genes from each category for individual alleles of female 9sCa WT and Msl2 -KO NPCs.(f,g) RNA expression (left) and chromatin accessibility (right) on ATAC UMAPs for escape genes in female 9sCa WT and Msl2 -KO NPCs.Allele frequency is indicated above each plot (See Methods for allele frequency analysis).
To understand the heterogeneity of escape genes, we thoroughly characterized the escapees of several independent WT NPC clones.Escape genes were identified with the following criteria: inactivated X allele normalized counts > 10 and 0.1< (WT inactivated X allele)/(allele-1 + allele-2) < 0.9.For this analysis, we used RNA-seq data from a total of ten female WT NPC lines each generated from a single cell clone.This list included three published cell lines (1 CAST Xi and 2 CAST Xa) and seven cell lines generated in our laboratory including four lines (1 CAST Xi and 3 CAST Xa) differentiated from 9sCa ESCs and three lines (1 CAST Xi and 2 CAST Xa) differentiated from CaBl ESCs (Supplementary Fig. 5a).In total, this analysis contains 3 NPC lines with CAST Xi and 7 NPC lines with CAST Xa.We found constitutive escapees that were biallelically expressed in all ten cell lines (Supplementary Fig. 5a).All facultative escapees were further categorized into commonly expressed in over 5 NPC lines and in less than 5 NPCs.Remarkably, 70% of facultative escapees were common in less than 5 NPCs, indicating the high degree of variability of escapees among different WT NPCs (Supplementary Fig. 5a).A complete summary of all escape genes identified in female 9sCa and CaBl NPCs are shown in Supplementary Fig. 5b.We found that the vast majority of MSL2-regulated escapees were specific to one cell line, including 16 escapees in 9sCa (excluding Mecp2 , Rab9 and Rbm41 ) and 3 escapees in CaBl NPC clones (Supplementary Fig. 5b, Fig. 2i).We believe that this relatively small number of MSL2-regulated genes likely reflects the heterogeneity of the escape process in NPCs differentiated in vitro illustrated in Supplementary Fig. 5a.
Msl2 -KO clones in male CaBl/BlCa and female 9sCa NPCs.(a) Schematic for the generation of isogenic single cell WT and Msl2 -KO NPCs.To obtain Msl2 -KO NPCs, we first differentiated ESCs into NPCs.From this population, a single clone was selected and proliferated to yield a pure population of cells.Subsequently, we used CRISPR/Cas9 to knock out MSL2 in the pure population, resulting in the generation of several Msl2 -KO clones.(b) In each of the three NPC lines male CaBl (top) and BlCa (middle) and female 9sCa (bottom), expression levels of indicated gene subsets for additional WT clones (WT1-3, dark grey), the original WT (light grey) and Msl2 KO (pink) for individual alleles (allele 1: left; allele 2: right) obtained from allele-specific DE analysis are depicted.Significance was scored by nonparametric Wilcoxon rank-sum test (two-sided), *p<0.05,**p<0.01,***p<0.001,NS: p>0.05.
are shown.IPs were analyzed by Western blot analysis of MSL complex members MSL2, MSL1 and MOF.DHX9 and RNA POL II serve as negative controls.(d,e) RT-qPCR analyses of Msl2 exon1, Msl1 , Msl3 , and Nanog mRNA levels (d) and MSL2 target genes ( Tsix , Bex2 , Zfp185 , and Bscl2) (e) in parental WT, 2 Msl2 -KO and 3 MSL2 H64Y mutant male CaBl ESC clones.mRNA levels were normalized to Rplp0 .Results are represented as relative values normalized to WT.Data are presented as mean values +/-SEM, n=6 independent experiments.