Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Feb 24.
Published in final edited form as: Dev Cell. 2014 Feb 24;28(4):351–365. doi: 10.1016/j.devcel.2014.01.017

Random Monoallelic Gene Expression Increases upon Embryonic Stem Cell Differentiation

Mélanie A Eckersley-Maslin 1,2, David Thybert 3, Jan H Bergmann 1, John C Marioni 3, Paul Flicek 3, David L Spector 1,2,*
PMCID: PMC3955261  NIHMSID: NIHMS560517  PMID: 24576421

Summary

Random autosomal monoallelic gene expression refers to the transcription of a gene from one of two homologous alleles. We assessed the dynamics of monoallelic expression during development through an allele-specific RNA sequencing screen in clonal populations of hybrid mouse embryonic stem cells (ESCs) and neural progenitor cells (NPCs). We identified 67 and 376 inheritable autosomal random monoallelically expressed genes in ESCs and NPCs respectively, a 5.6-fold increase upon differentiation. While DNA methylation and nuclear positioning did not distinguish the active and inactive alleles, specific histone modifications were differentially enriched between the two alleles. Interestingly, expression levels of 8% of the monoallelically expressed genes remained similar between monoallelic and biallelic clones. These results support a model in which random monoallelic expression occurs stochastically during differentiation, and for some genes is compensated for by the cell to maintain the required transcriptional output of these genes.

Introduction

The majority of gene expression in diploid cells is carried out through expression of both alleles of each gene. However, several interesting cases of monoallelic expression, in which there is transcription from only one allele, have been documented. Well characterized and extensively studied examples include X-chromosome inactivation (reviewed in Guidi et al., 2004; Schulz and Heard, 2013), and genomic imprinting (reviewed in Bartolomei and Ferguson-Smith, 2011; McAnally and Yampolsky, 2010). Interestingly, random monoallelic expression can also occur on autosomes independently of the parental origin and genotype (reviewed in Chess, 2012a; Guo and Birchler, 1994). For example, the immune system utilizes monoallelic expression to ensure each B-cell expresses a single uniquely rearranged immunoglobulin receptor (Pernis et al., 1965). Additionally, neurons express olfactory receptors (ORs) in a monogenic and monoallelic manner to provide cell-identity and aid in neural connectivity (reviewed in Chess et al., 1994). However, random autosomal monoallelic expression is not limited to specialized gene families, as it has been reported to occur at individual gene loci throughout the genome of a few cell types examined (Gimelbrant et al., 2007; Jeffries et al., 2012; Li et al., 2012; Zwemer et al., 2012). Yet, despite the identification of such genes, detailed molecular characterization and potential biological consequences of random monoallelic expression remains unknown.

The extent of random monoallelic expression varies from 2% in neural stem cells (Jeffries et al., 2012; Li et al., 2012) to 10% in lymphoblasts (Gimelbrant et al., 2007; Zwemer et al., 2012). Interestingly, only a small number of genes have been identified in common across these studies, suggesting that monoallelic expression may be established during development in a lineage or cell-type specific manner. However, random monoallelic expression has not been studied in the context of a developmental paradigm.

Exclusive expression from one allele renders the cell susceptible to loss-of-heterozygosity effects that could result in deleterious disease-related phenotypes. Monoallelic expression has been hypothesized to contribute to cellular diversity and identity, as is the case for the ORs and immunoglobulins (reviewed in Chess, 2012b), or may be a mechanism for regulating the transcriptional output of genes, although this has not been vigorously analyzed. Alternatively, rather than being an active process, the switch to monoallelic expression may instead reflect the stochastic nature of gene regulation occurring independently at the two alleles.

We performed an allele-specific RNA-sequencing screen for random autosomal monoallelic expression during differentiation of mouse embryonic stem cells (ESCs) to neural progenitor cells (NPCs). Interestingly, we observed a 5.6 fold increase in monoallelic expression during differentiation, from just 67 genes (<0.5%) in ESCs to 376 genes (3.0%) in NPCs, indicating that the establishment of monoallelic expression occurs during early development. Detailed genomic and molecular characterization of these genes revealed that DNA methylation was not sufficient for the mitotic inheritance of monoallelic expression, nor was there evidence for differential nuclear positioning of the active versus inactive alleles. However, specific histone modifications were sufficient to distinguish the active and inactive alleles, and likely contribute towards maintaining monoallelic expression across cell divisions. Interestingly, in a subset of monoallelically expressed genes, transcriptional compensation through up-regulation of the single-active allele preserved the biallelic levels of the respective mRNA in the cell. These results support a model where stochastic gene regulation during differentiation results in monoallelic expression, and for some genes, the cell is able to compensate transcriptionally to maintain the required transcriptional output of these genes. Therefore random monoallelic expression exemplifies the stochastic and plastic nature of gene expression in single cells.

Results

Identification of monoallelically expressed genes upon differentiation of mouse embryonic stem cells to neural progenitor cells

To identify random autosomal monoallelically expressed genes in mouse ESCs and NPCs, we used male cells derived from a F1 hybrid between C57Bl/6J and CAST/EiJ mice in which the high density of single nucleotide polymorphisms (SNPs) allowed us to quantify allele-specific expression for 82.8% of transcripts. We expanded 6 single-cell derived clones from both ESCs and induced NPCs (Fig 1A, S1). Assuming inheritance of monoallelic expression across cell divisions, all cells within each single-cell derived clone are expected to express the same alleles. However, different clones should show random selection of alleles, allowing the identification of mitotically inheritable random monoallelically expressed genes.

Figure 1. Identification of monoallelically expressed genes in ESCs and NPCs.

Figure 1

Open in a new tab

(A) Schematic of the allele-specific RNA-sequencing screen used to identify monoallelically expressed genes. For both ESCs and NPCs, 6 single cell derived clones were generated and transcripts categorized as either C57Bl/6J biased (orange), CAST/EiJ biased (blue), biallelic (orange + blue) or not expressed/assessable (grey). Transcripts were further grouped into three classes based on their expression across clones, with Class A representing the high confidence random monoallelically expressed genes, Class B which following additional filtering represent additional monoallelically expressed genes and Class C representing the non-random monoallelically expressed genes. (B) Graphical representation of the p-value and d-score thresholds used to categorize transcripts (circles) in a given clone as either monoallelic (|d-score|>0.4), allele-bias (0.18<|d-score|<0.4) or biallelic (|d-score|<0.18). (C) Examples of Class A, B and C transcripts showing the behavior of individual clones (filled circles) with respect to the d-score and p-value. (D) Summary table of number of genes from Class A, B, B-filtered, and C for both ESCs and NPCs, and their percentage of all assessable transcripts. See also Fig S1, Table S1–S2.

For each clone, ~5×107 reads were mapped to both the C57Bl/6J and CAST/EiJ transcriptomes using BWA (Li and Durbin, 2010) (Fig S1E). To control for possible loss of heterozygosity, mouse diversity SNP arrays were run on genomic DNA, and transcripts within aneuploid regions were excluded from further analysis (see supplementary materials and methods). For all assessable transcripts, the number of reads corresponding to each allele at each SNP position (minimum of 5 read coverage) was used to determine whether there was evidence of allele-specific expression based on two metrics: a d-score representing the ratio of allelic expression; and a p-value (see supplementary materials and methods). Assessable transcripts were then classified as monoallelic (|d-score|≥0.4, p-value<10−8), allele biased (0.18≤|d-score|<0.4, p-value<10−8) or biallelic (|d-score|<0.18) (Fig 1B). Based on the patterns of expression bias observed across clones, transcripts were subsequently grouped into one of three classes of monoallelically expressed genes (Fig 1C): Class A transcripts had at least one clone classified as monoallelic for the C57Bl/6J allele and at least one for the CAST/EiJ allele, and represent high confidence random monoallelically expressed genes as they clearly show a random choice of allele. Class B transcripts had at least one clone classified as monoallelic for either the C57Bl/6J or CAST/EiJ alleles, but not both. These transcripts were further filtered to select those in which there was one high-confidence biased clone (p-value<10−10 and |d-score|>0.35) and one high-confidence biallelic clone (|d-score|<0.1). With a larger number of clones, these filtered class B transcripts would likely be assigned to class A. Finally class C transcripts represent non-random monoallelically expressed genes. In this class, all clones showed bias towards the same allele with no evidence that the second allele is transcribed. These transcripts include imprinted genes, in addition to genes in which one allele was inactive due to a cis mutation, and as such were not included in further analysis.

In ESCs, of the 13,699 assessable genes, only 1 was classified as class A, with another 66 class B filtered, giving a total of 67 monoallelically expressed genes or 74 transcripts, representing only 0.49% of assessable genes (Fig 1D, Table S1, S2). Interestingly, this low number of genes increased 5.6 fold during differentiation to 376 genes (86 class A and 302 class B filtered) in the NPCs, corresponding to 602 transcripts or 3.0% of assessable genes (Fig 1D, Table S1, S2). This set included Thrsp and several members of the Protocadherin family which have been previously reported to be monoallelically expressed (Esumi et al., 2005; Kaneko et al., 2006; Wang et al., 2007). The increase in monoallelic expression during differentiation suggests that the establishment of monoallelic expression occurs upon cell-fate specification early in development.

Validation of monoallelically expressed genes

Validation of the screen was first performed by Sanger sequencing of PCR products containing informative exonic SNPs for 20 different genes in both ESCs and NPCs. Clones were classified as monoallelic or biallelic (Fig 2A), then subsequently compared to the RNA-sequencing screen results. 76 of 82 (93%) PCR products were in agreement with the RNA-sequencing screen (Fig S2A, B), demonstrating the robustness of our approach.

Figure 2. Validation of monoallelic gene expression.

Figure 2

Open in a new tab

(A) Representative traces from Sanger sequencing of PCR products containing informative exonic SNPs (arrow) from cDNA from biallelic (first column) or monoallelic (second and third columns) clones for four separate monoallelically expressed genes. (B) Chromatin immunoprecipitation (ChIP) for RNA polymerase II large subunit (blue) or control IgG (black) for three separate gene promoter regions between monoallelic (m, dark blue), allele-bias (ab, medium blue) or biallelic (b, light blue) clones. Error bars represent SEM of at least 3 biological replicates. (C) Representative traces of Sanger sequencing of ChIP products containing informative SNP (arrow) revealing associated alleles for monoallelic and biallelic clones. (D) RNA-DNA FISH validation of six separate monoallelically expressed class A genes in NPCs. Representative 3D projections of RNA-FISH (top row, green) and DNA-FISH (second row, red) image stacks. Third row shows a merge of RNA- and DNA-FISH with DAPI to visualize total DNA (blue). Arrows denote actively transcribing alleles, arrowheads inactive alleles. Scale bar represents 5μm. Bottom row shows quantification of independent clones for each of the 6 genes for NPC clones that were either biallelic, allele-biased or monoallelic for the respective gene. Percentage of expressing cells having either 1 (dark grey) or 2 (light grey) RNA-FISH signals representing monoallelic and biallelic cells respectively. 100 cells were analyzed per sample. See also Fig S2.

Next, monoallelic expression was confirmed at the level of transcription by chromatin immunoprecipitation (ChIP) for all forms of RNA polymerase II. Levels of pull-down were similar between monoallelic and biallelic clones within the body of four randomly selected genes (Fig 2B). Importantly, Sanger sequencing of SNPs within the amplicons used confirmed that RNA polymerase II was specifically associated with only the active allele in monoallelic clones compared to both alleles in biallelic clones (Fig 2C), confirming that monoallelic expression is due to the exclusive transcription of only one of the two alleles in the cell.

We further validated monoallelic expression at single-cell resolution by RNA fluorescence in situ hybridization (RNA-FISH). By using fluorescently labeled probes targeting both exonic and intronic sequences of the target gene, nascent RNA at the sites of transcription can be visualized as fluorescent spot(s) within the nucleus (Fig 2D top row). These RNA-FISH spots co-localize with the gene locus visualized by subsequent DNA-FISH in the same cells (Fig 2D second row), confirming that they are indeed sites of transcription. We calculated the percentage of expressing cells exhibiting monoallelic or biallelic expression and successfully validated 6 out of 6 class A monoallelically expressed genes (Fig 2D). For example, a single Acyp2 signal was detected by RNA-FISH in 93.5% of expressing cells in a monoallelic clone. In contrast 57.5% of cells in a biallelic clone showed 2 active alleles of Acyp2. Likewise, expression from 1 allele was confirmed for 84.3%, 96.5%, 93.3% and 89.6% of cells in monoallelic clones for Ror2, Pdzrn4, Gas6 and Acot1 respectively. In this way, the RNA-FISH analysis confirmed at single cell resolution the results of RNA-sequencing analysis.

Importantly, RNA-FISH confirmed monoallelic expression for 3 class B genes in NPCs of a pure genetic background (Fig S2C). A single transcribing allele was observed in 54.8%, 67.7% and 73.9% of expressing cells for Atp1a2, Arap1 and Mavs respectively, confirming that monoallelic gene expression is independent of the genetic background and not due to differences between the two parental strains in the hybrid cell lines.

Dynamics of monoallelic expression during differentiation

During differentiation there is 5.6 fold increase in monoallelic expression, coinciding with the loss of pluripotency and gain of lineage commitment (Fig 1D). Interestingly, we observed very few (<2%) monoallelically expressed genes in common between ESC and NPC (Fig 3A). Instead, the majority of monoallelically expressed genes were biallelically expressed in the other cell type (Fig 3B–C, S3A–B). Thus, monoallelic expression, while maintained across cell divisions, is not maintained during the transition from ESC to NPC.

Figure 3. Dynamic changes in monoallelic expression during differentiation.

Figure 3

Open in a new tab

(A) There is little overlap between ESC and NPC monoallelic transcripts. (B) Status of the NPC monoallelic transcripts in ESCs. (C) Status of the ESC monoallelic transcripts in NPCs. (D) Quantification of ESC and NPC monoallelic transcripts as monoallelic (light grey), allele-biased (dark grey) or not expressed/assessable (black). Bars represent standard deviation amongst 6 clones. (E) Expression level (NRPK) distribution of all assessable and monoallelically expressed genes in both ESCs and NPCs. (F) Expression levels decrease, increase or remain unchanged during ESC to NPC differentiation. (G) Expression level changes for ESC monoallelically expressed genes. (H) Expression level changes for NPC monoallelically expressed genes. See also Fig S3, S4.

Interestingly, for 98.9% of monoallelically expressed genes, at least one clone was either biallelic and/or did not express the respective gene (Table S2). Within a single clone, ~60% of the monoallelically expressed genes show biallelic expression (Fig 3D), suggesting that monoallelic expression may reflect variation in gene expression regulation between two homologous alleles. This contrasts with imprinted genes and X-chromosome inactivation, where all cells exhibit strict monoallelic expression, and implies that, rather than being tightly regulated, random monoallelic expression is not an active decision required for cell survival or differentiation.

Importantly, the distribution of expression levels of the monoallelically expressed genes was not dramatically different from all assessable transcripts (Fig 3E, S3C–D). The small, yet statistically significant difference in the expression level for NPCs is unlikely to be of biological significance. Furthermore, reducing the stringent expression level thresholds used for the screen did not result in a large increase in the number of monoallelically expressed genes (Table S3), confirming monoallelically expressed genes have a similar expression profile to all expressed genes.

We next determined whether any genomic features of monoallelically expressed genes distinguished them from other expressed genes. Unlike for imprinting and ORs, the random monoallelically expressed genes were distributed throughout the genome and did not fall into any genomic clusters (Fig S4A). Monoallelically expressed genes showed similar GC density at their promoters to all assessable genes (Fig S4B), in contrast to the reduced GC density previously reported for ORs (Clowney et al., 2011). Analysis of 174 mammalian and 530 vertebrate transcription factor motifs revealed that although 10 motifs were differentially enriched at the promoters, they were not sufficient to distinguish monoallelically expressed genes from all assessable genes (Fig S4C). Furthermore, while there was a small decrease in evolutionary conservation of monoallelically expressed genes, this was not as dramatic as what is observed for ORs (Fig S4D). Finally, gene ontology analysis using DAVID revealed a slight enrichment in glycoproteins involved in signaling (Fig S4E). Thus, random monoallelically expressed genes are not distinguished from other genes by these genomic features.

Finally, we compared the changes in expression levels of monoallelically expressed genes during differentiation (Fig 3F–H). Expression of the majority of ESC monoallelically expressed genes either decreased (50%) or did not change (32.4%) during differentiation (Fig 3G), despite the majority being biallelically expressed in NPCs. Furthermore, only 13.1% of the NPC monoallelically expressed genes are expressed at lower levels in the NPCs compared to ESCs (Fig 3H), despite the fact that 55.2% switch from biallelic to monoallelic expression during differentiation (Fig 3B), again suggesting that monoallelic expression is not a mechanism for reducing transcript levels.

DNA methylation does not regulate monoallelically expressed genes

One intriguing aspect of monoallelic expression is that the transcriptional imbalance between the active and inactive alleles is maintained across cell generations. DNA methylation is the most widely accepted mechanism through which the transcriptional state of a gene can be inherited and maintained in daughter cells (Smith and Meissner, 2013), and distinguishes active and inactive alleles of both imprinted (Kelsey and Feil, 2013) and X-linked genes (Schulz and Heard, 2013). To assess a potential role for DNA methylation, we performed bisulfite analysis of both CpG high and CpG low promoters of 10 monoallelically expressed genes, and compared methylation levels between biallelic, allele-biased, and monoallelic NPC clones (Fig 4A–B, S5B). Globally, we did not observe a correlation between the extent of DNA methylation with the overall expression level (Fig S5A). If DNA methylation differentially marked the active and inactive alleles, monoallelic clones should contain a mix of methylated and unmethylated molecules. For seven out of ten genes tested, we did not see any evidence for allele-specific DNA methylation in the monoallelic clones (Fig 4A, S5B, data not shown). However, three genes (Npl, Cbr3 and Fkbp7) contained a monoallelic clone in which there was clear separation between methylated and unmethylated reads (Fig 4A, B, data not shown). In the case of Cbr3, the amplicon used contained an informative SNP that allowed us to assign the bisulfite treated reads to either the C57Bl/6J or CAST/EiJ allele, confirming the unmethylated and methylated reads were in fact derived from the active and inactive alleles respectively (Fig. 4B). The bisulfite analysis reflected predominantly levels of 5-methylcytosine, as 5-methylcytosine (5meC) and 5-hydroxymethylcytosine (5hmeC) DNA immunoprecipitation revealed 5meC but little to no 5hmeC present at the promoters of 6 monoallelically expressed genes analyzed (Fig S5C–F).

Figure 4. DNA methylation does not regulate monoallelic expression.

Figure 4

Open in a new tab

(A) Bisulfite traces of five separate class A genes promoters for NPC clones that were biallelic (top) or allele-biased/monoallelic (middle and bottom). Filled and open circles represent methylated and unmethylated CpGs respectively. Each row within a group represents a single bisulfite treated molecule. For regions containing allele information (Gas6), the alleles are separated by a line and labeled accordingly. (B) Bisulfite traces for Cbr3 for a biallelic (left) clone (left), untreated monoallelic (middle) and 5-Azacytidine treated monoallelic (right) clone. Line separates reads from CAST/EiJ (top) and C57Bl/6J alleles (bottom) (C) Sanger sequencing results of cDNA from a biallelic (left), untreated monoallelic (middle) and 5-Azacytidine treated monoallelic (right) clone of Cbr3. Both treated and untreated monoallelic samples show only one allele expressed (arrow). See also Fig S5.

To test whether this distinguishing differential DNA methylation was involved in maintaining monoallelic expression through the cell cycle, we treated the cells with 5-Azacytidine, which inhibits DNA methyltransferases leading to global DNA demethylation. Following 5 days of treatment, which was sufficient to allow several cell divisions to occur, the inactive allele lost all methylation marks (Fig 4B). However, when we analyzed expression from the two alleles by PCR amplification including an informative SNP within the cDNA, we failed to see reactivation of the inactive allele for Cbr3 (Fig 4C). This also held true for an additional 6 genes tested, including Npl and Fkbp7 (data not shown). Thus, DNA methylation alone does not regulate the expression status of random monoallelically expressed genes.

Active and inactive alleles are differentially marked by H3K4 and H3K9 methylation

As we did not find a general role for DNA methylation, we next investigated whether histone modifications may distinguish the active and inactive alleles. We screened promoter regions of monoallelically expressed genes with a panel of 9 well-characterized histone modifications implicated in gene transcription or gene silencing (Black et al., 2012), by chromatin immunoprecipitation (ChIP) (Table S4). Methylation of histone H3 at lysine 4 (H3K4) is associated with actively transcribed regions of the genome (Black et al., 2012). For all gene promoters tested, there was an increase in the levels of both associated H3K4me2 (Fig 5A) and H3K4me3 (Fig 5C) between monoallelic and biallelic clones, consistent with biallelic clones having twice the number of active alleles. Importantly, SNP analysis by Sanger sequencing revealed that only the active allele in monoallelic clones was associated with methylated H3K4 (Fig 5B, D), compared to both alleles in biallelic clones, for all genes tested.

Figure 5. Alleles are marked by differential histone modifications.

Figure 5

Open in a new tab

Chromatin immunoprecipitation (ChIP) analysis for H3K4me2 (A, B), H3K4me3 (C, D) and H3K9me3 (E, F). Analysis of regions within 200bp of the transcription start site for 2 class A (Tubb2a and Cbr3) and 1 class B (Serpinh1) genes. Pull down quantification as percentage of input for H3K4me2 (A), H3K4me3 (C) and H3K9me3 (E) for individual clones which are monoallelic (m, dark blue/red), allele-bias (ab, medium blue/red) or biallelic (b, light blue/red) for the respective clone. IgG (black) shows non-specific pull down. Error bars represent SEM of 3–4 biological replicates. Sanger sequencing traces of ChIP-qPCR products for H3K4me2 (B), H3K4me3 (D) and H3K9me3 (F) for a monoallelic and biallelic clone for each of the three genes tested showing allele(s) associated with the respective histone modification. See also Fig S6.

After identifying modifications specifically marking the active allele, we investigated whether there were any modifications associated with the inactive allele. Tri-methylation of histone H3 lysine 9 (H3K9me3) and lysine 27 (H3K27me3) are two well-characterized marks of transcriptionally silent genes (Black et al., 2012). ChIP analysis revealed a decrease in the levels of H3K9me3 associated with the promoters of monoallelic versus biallelic clones (Fig 5E). Moreover, Sanger sequencing revealed that H3K9me3 was specifically associated with the inactive allele (Fig 5F). In all cases examined we did not observe any specific association of H3K27me3 with the inactive allele (Fig S6A) and analysis of published H3K27me3 ChIP-seq data sets in NPCs (Mikkelsen et al., 2007), revealed that only 2% of monoallelically expressed genes have detectable H3K27me3 at their promoters (Fig S6B). We also did not observe any preferential association with the inactive allele for other marks of inactive chromatin (Table S4), including mono-methylation of histone H4 lysine 20 (H4K20me1). Additionally, tri-methylation of histone H4 lysine 20 (H4K20me3), which has been implicated in marking OR gene choice (Magklara et al., 2011), was not observed at these genes (Table S4).

Nuclear organization of active versus inactive alleles

Since nuclear positioning of genes has been correlated with transcriptional activity (Hübner et al., 2013), we were interested in assessing whether differences in nuclear position may distinguish the active and inactive alleles. We performed RNA-DNA-FISH analysis in NPCs and analyzed the position of the active and inactive alleles in three-dimensions. However, we did not find any evidence for preferential positioning of the inactive allele towards heterochromatic foci (Fig 6A), nor the nuclear periphery (Fig 6B), as for most genes examined, both alleles had similar interaction frequencies with these domains despite their difference in transcriptional state. We also did not see evidence for allelic pairing of the active and inactive alleles (data not shown). Furthermore, global analysis of the monoallelically expressed genes did not show any preferential association within or at the borders of lamin-associated domains (LADs) (Fig 6C–D) (as defined in Peric-Hupkes et al., 2010). Therefore, the nuclear positioning of these genes does not play a determining role in distinguishing monoallelic expression.

Figure 6. Active and inactive alleles do not show preferential nuclear positioning.

Figure 6

Open in a new tab

Bar graph showing the proportion of active (light grey) and inactive (dark grey) alleles associated with either heterochromatic foci (A) or the nuclear periphery (B) for 6 separate class A monoallelically expressed genes in NPCs. Measurements were performed in 3D. (C) Proportion of monoallelically expressed genes which are located within (black) or outside (white) Lamin Associated Domains (LAD). (D) Minimal distance of genes from the nearest LAD in Mb.

A sub-set of monoallelically expressed genes exhibit transcriptional compensation

Next, we examined the impact of monoallelic expression on the transcriptional output in the cell. We performed linear regression analysis to compare expression levels of individual monoallelically expressed genes across the independent NPC clones to determine if there was a correlation between the extent of allelic imbalance, reflected in the d-score, and total expression level. If the levels of expression remained constant across clones, the slope of the linear regression line, a, would be equal or close to 0 (Fig 7A upper). Alternatively, if monoallelic clones had half the transcript level of biallelic clones, the slope a would be equal to the y-intercept b (Fig 7A lower). Thus we classified genes as either following the dosage of active alleles ( 0.75<-ab<1.25) (Fig 7D,E), or showing evidence of transcriptional compensation ( ∣-ab∣<0.35) (Fig 7B,C). Using these criteria, we identified 30 monoallelically expressed genes (8%) with evidence for transcriptional compensation (Table S5), and 54 genes (15.4%) that followed the dosage of active alleles. The remaining genes either showed intermediate responses or were highly variable and so not able to be confidently classified based on data from 6 clones. We validated the linear regression analysis by quantitative RT-PCR confirming transcriptional compensation for 7 out of 9 genes (78%) and dosage sensitivity for 7 out of 11 (64%) genes tested (Fig 7B–E, data not shown). Interestingly, the genes that exhibited transcriptional compensation were enriched for DNA binding proteins and transcription factor activity (Table S5), although the confidence of enrichment was low (p-value 0.037). The transcriptional compensation is intriguing as it suggests that for these genes the exact level of transcript is more critical than for others. Furthermore, it supports a model in which the biological consequences of monoallelic expression is not to reduce transcript levels in the cell, but rather may be a reflection of the stochastic nature of gene regulation at independent alleles.

Figure 7. A subset of monoallelically expressed genes upregulate the single active allele.

Figure 7

Open in a new tab

(A) A hypothetical gene that undergoes compensation (top) or no compensation (bottom). For the trend line y=ax+b, the non-compensated genes would be expected to have a=−b or a/−b=1. Compensated genes would have a=0 or a/−b<1. Example of genes classified as compensated (a/−b<035) (B–C) or non-compensated (a/−b>0.75) (D–E). Red crosses represent individual clone normalized RPK from RNA-seq, blue circles represent mean normalized expression of three biological replicates by Q-PCR, errors bars represent standard deviation from the mean. Lines represent linear regression lines of best fit. See also Table S5. (F) Monoallelic gene expression increases during differentiation of ESCs to NPCs, coinciding with the loss of pluripotency and gain of lineage commitment. Stochastic regulation of homologous alleles results in random monoallelic gene expression. Upon differentiation of ESCs to NPCs, there is independent low probability regulation of the two alleles, resulting in a mixed population of NPCs containing 0, 1 or 2 active alleles. The active and inactive alleles are distinguished through H3K4me2/3 (green circles) and H3K9me3 (red squares) respectively, which may contribute to the clonal inheritance of allelic imbalance. Monoallelic expression can either result in dosage sensitivity, where the cell has half the levels of mRNA (yellow line) as biallelic cells; or transcriptional compensation in which the cell upregulates the single active allele.

Discussion

We performed an allele-specific RNA-sequencing screen and identified a 5.6 fold increase from just 67 to 376 genes in random autosomal monoallelic expression during differentiation of mouse ESCs to NPCs, indicating that monoallelic expression is acquired upon lineage commitment. Importantly, this study provides a detailed and extensive molecular characterization of random monoallelic expression, revealing that histone modifications, not DNA methylation or nuclear organization, distinguish the active and inactive alleles. Interestingly, monoallelic expression is not required by the cell, since some clones exhibit biallelic expression, supporting a model in which stochastic gene regulation occurring independently at the two alleles results in monoallelic gene expression, and for some genes is compensated for transcriptionally to maintain the required level of expression of these genes.

We propose that random monoallelic expression exemplifies a stochastic aspect of gene regulation that takes place upon the initiation of specific differentiation programs resulting in global changes in chromatin and gene expression (Fig 7F). If the probability of gene activation or repression is less than 1, this would result in a mixed population of cells containing 0, 1 or 2 active alleles, which, once established and not detrimental to the cell, could be subsequently maintained across cell generations and propagated clonally. Probabilistic models of stochastic gene regulation have been previously proposed for specific examples of monoallelic expression, including Albumin in hepatocytes (Michaelson, 1993), Ly49 receptors in natural killer cells (Held and Kunz, 1998), and interleukins in T-lymphocytes (Guo et al., 2005). In all cases, the two alleles are independently regulated with a low activation probability, possibly due to limiting accessibility of key activating factors. One outcome of this independent regulation is that it results in both monoallelic and biallelic cells in a mixed population. Indeed, at least one biallelic clone is observed for almost all monoallelically expressed genes, consistent with an independent stochastic regulation model. The outcome of monoallelic expression for some genes may be unfavorable if the cell requires a specific level of transcript that cannot be accommodated for by the single active allele, thus resulting in cell death. However, for those genes for which either the exact level of transcript is not critical, or for those that are able to compensate transcriptionally, monoallelic expression represents a viable outcome for the cell.

One key finding of our study is that there is significantly more monoallelic expression in NPCs compared to pluripotent ESCs supporting the establishment of lineage or cell-type specific random monoallelic expression during early development (Fig 7F). Consistent with this, screens performed in neuronal cell types (Jeffries et al., 2012; Li et al., 2012; Wang et al., 2007) have a higher degree of overlap with our study than those in more distant cell types, such as lymphoblasts and fibroblasts (Gimelbrant et al., 2007; Zwemer et al., 2012). Several factors could contribute to the lack of extensive monoallelic expression in ESCs. Not only are ESCs unique in their pluripotent potential and dynamic open chromatin (reviewed in Fisher and Fisher, 2011), but ESC populations are highly heterogeneous both in terms of transcriptional profiles and developmental potency (Huang, 2011; Martinez Arias and Brickman, 2011). Within a colony, ESCs cycle between different states of developmental potential continuously adjusting their transcriptional program (Canham et al., 2010). Thus while the initial frequency of monoallelic expression may be similar to that of differentiated cell types, these allelic imbalances may not be maintained as efficiently and thus not clonally propagated.

This study represents a detailed and extensive molecular characterization of the differences between active and inactive alleles of random autosomal monoallelically expressed genes. Intriguingly, DNA methylation, important for other examples of monoallelic expression including genomic imprinting (reviewed in Kelsey and Feil, 2013), was not sufficient to distinguish nor maintain monoallelic gene expression of the genes analyzed. While allele-specific DNA methylation has been previously reported and used to identify random monoallelically expressed genes (Wang et al., 2007), a direct role for DNA methylation driving monoallelic expression has not been shown. Indeed, DNA methylation does not maintain active and inactive alleles of the monoallelically expressed Cubilin gene in kidney and intestinal cell lines (Aseem et al., 2013). Additionally, allele-specific DNA methylation does not drive monoallelic expression in human cells in the absence of DNA sequence variation effects (Gutierrez-Arcelus et al., 2013).

Importantly, we did, however, observe that the active and inactive alleles were associated with H3K4me2/3 and H3K9me3 respectively. Histone modifications have been shown to mark other examples of monoallelically expressed genes, including X-inactivated genes by promoter-restricted H3K4me2 (Rougeulle, 2003), and ORs by H3K9me3 and H4K20me3 (Magklara et al., 2011), consistent with our results. Interestingly, we did not see evidence for the Polycomb-associated H3K27me3 repressive mark, whereas H3K9me3 was present. It remains to be determined whether these histone modifications are actively involved in the inheritance of the transcriptional state or if simply reflect the transcriptional status of the respective alleles.

The organization of genes within the nucleus has been linked to transcriptional output (reviewed in Hübner et al., 2013). While nuclear positioning has been implicated in monoallelic expression of ORs (Clowney et al., 2012) immunoglobulins (Skok et al., 2001), and Gfap in astrocytes (Takizawa et al., 2008), we did not observe any differences in the position of active versus inactive alleles for the genes examined in this study.

There may be yet additional undetected characteristics that distinguish the active and inactive alleles of monoallelically expressed genes that may also play a role in maintaining the difference in transcriptional state across cell divisions. Asynchronous DNA replication timing, in which the active allele replicates earlier in S-phase than the inactive allele (Hiratani and Gilbert, 2009) has been observed at monoallelically expressed genes (Donley et al., 2013; Dutta et al., 2009). However, as the monoallelically expressed genes are interspersed amongst biallelic genes within the same DNA replication timing domains (Alabert and Groth, 2012), it is unlikely that asynchronous DNA replication timing contributes to the monoallelic state.

Surprisingly, transcript levels for some monoallelically expressed genes did not follow the active allele dosage. Transcriptional compensation has been reported for heterozygous knockout mice that show comparable mRNA and/or protein levels to their wild-type counterparts, including Mks1 (Wheway et al., 2013) and Bag3 (Homma et al., 2006), both identified as monoallelically expressed in NPCs. However, there are examples of genes in which the heterozygous mice have reduced transcript levels, including the monoallelically expressed genes Cth (Kaasik et al., 2007) and Cstb (Ishii et al., 2010), suggesting that transcriptional upregulation is not only gene specific but also cell type specific. Transcriptional compensation has also been observed in non-mammalian systems, including Drosophila (McAnally and Yampolsky, 2010) and Maize (Guo and Birchler, 1994) in which mRNA levels do not strictly follow the dosage of the gene. For those genes that exhibit compensation, it will be of interest to determine the mechanisms by which transcriptional compensation maintains the total level of mRNA in the cell, potentially through levels and accessibility of specific transcription factors, feedback loops sensing the levels of mRNA and/or protein in the cell (Guidi et al., 2004), or autoregulation (Trieu et al., 2003). This ability to tune the transcriptional output of an allele in response to either genetic or epigenetic inactivation of the second allele has important biological consequences, especially in the interpretation of copy number variants, as these may not necessarily result in a change of transcript and protein product. In this way, random autosomal monoallelic gene expression illustrates the remarkable plasticity and stochasticity of gene regulation in mammalian cells.

Experimental Procedures

Cell culture

ESCs were cultured using standard procedures in medium containing 1,000U/ml leukemia inhibitory factor (Millipore) with irradiated MEF feeders (Global Stem) on gelatin-coated plates. ESCs were removed from feeder cells by soaking twice on gelatin-coated plates for 1 hour each prior to sample collection or cell differentiation. Differentiation was performed using a protocol adapted from (Conti et al., 2005) by culturing ESCs in the absence of MEF feeders in 50:50 DMEM/F12:Neurobasal medium (Gibco) supplemented with 1x N2 (Gibco), 1x B27 (Gibco), 40mg/L insulin (Sigma), 25μg/ml BSA Fraction V (Gibco) at 0.5×106–2.0×106 cells/10cm plate for 6 days. Cells were then resuspended in N2 expansion medium (DMEM/F12, 50μg/ml BSA fraction V, 10ng/ml EGF (Preprotech), 10ng/ml FGF (Preprotech), 1μg/ml laminin (Invitrogen), 1x N2) and plated onto uncoated T75 flasks to allow for neurosphere outgrowth. Following 4 days, neurospheres were collected by mild centrifugation and plated onto gelatin-coated plates in N2 expansion medium. Following 2–3 passages, cells represented a homogeneous population of NPCs. Single ESCs and NPCs were seeded through limiting dilutions in 96 wells and expanded to obtain clonal populations.

Allele-specific RNA sequencing screen

RNA from 6 independent single-cell clones was isolated using Trizol reagent (Ambion) and polyA+ RNA isolated (Oligotex kit, Qiagen). Stranded libraries were prepared using a protocol adapted from (Parkhomchuk et al., 2009) for paired-end sequencing on the Illumina GA IIx platform. Reads were mapped with BWA to both C57Bl/6J and CAST/EiJ transcriptomes. Affymetrix Mouse Diversity SNP arrays (The Jackson Laboratory, Bar Harbor, Maine, USA) were performed for DNA from NPC clones to control for locus heterozygosity. For detailed statistical analysis see extended materials and methods. For each assessable transcript (≥5 reads/SNP) in each clone, a significance test and d-score representing the weighted difference from 50:50 expected ratio between the two alleles calculated. Transcripts were then classified as monoallelic (|d-score|≥0.40, p-value ≤10−8), allele-biased (0.18≤|d-score|<0.40, p-value ≤10−8), biallelic (|d-score|<0.18 and/or p-value>10−8), not expressed (expression is lower than 5.8 NRPK), non-assessable, or other, and assigned into three classes of monoallelically expressed transcripts based on patterns across clones, or a fourth class containing all other transcripts: Class A transcripts have at least one C57Bl/6J and one CAST/EiJ biased clone; Class B transcripts have at least 1 biased clone and one biallelic clone and were further filtered to obtain additional monoallelic transcripts; and Class C transcripts showed bias in all clones but only towards one allele.

Validation of monoallelically expressed genes

RNA was isolated using Trizol reagent (Ambion) and converted to cDNA (Applied Biosciences RT reagents). PCR amplification of exonic SNPs was performed (Phusion High-Fidelity Polymerase, NEB), products were gel purified and subjected to Sanger sequencing. Sequencing traces were analyzed using 4Peaks 1.7.2 (Mekentosj) and scored independently of screen results. Non-clonal NPC cDNA or genomic DNA was used to confirm the presence of the SNP. See Supplemental Table 3 for primer sequences and SNP information.

RNA and DNA fluorescence in situ hybridization (FISH)

Probes for both RNA and DNA FISH were generated from BAC or fosmid DNA (Acot1 WI1-2795P12; Acyp2 RP23-405O2; Arap1 WI1-0101I18; Atp1a2 WI1-1389O12; Cap2 RP23-105K14; Gas6 WI1-0153N19; Mavs WI1-1832A20; Pdzrn4 RP23-322L18; Ror2 RP23-280O5) by Nick Translation (Abbott Molecular Inc.) with red or green fluorescently conjugated d-UTP nucleotides (Enzo Life Sciences) for 10 hours at 15 degrees. Probe size was verified by agarose gel electrophoresis to be 50–400nt. Probes were mixed with competitor DNA, lyophilized and resuspended in 50% deionized formamide, 2x SSC and 10% dextran sulfate.

RNA-DNA FISH was performed sequentially with separate images taken for both RNA-FISH and DNA-FISH. Cells were grown on coverslips, fixed in freshly prepared 4% formaldehyde for 15 minutes and permeabilized in 0.1% Triton for 5 minutes on ice in the presence of 5mM vanadyl ribonucleoside compex (NEB). RNA FISH was performed by hybridizing prepared denatured probes on coverslips overnight at 40 degrees centigrade. DNA FISH required prior treatment with 0.1mg/ml RNase A (Invitrogen) for 1 hour at 37 degrees centigrade, followed by heat denaturation in 70% formamide, 2x SSC for 5 minutes at 80 degrees centigrade, and hybridization with denatured probe overnight at 37 degrees centigrade. Following hybridization, cells were washed in 2xSSC/50% formamide, 2xSSX then 1xSSX at 37 degrees, counterstained with DAPI and mounted in antifade containing 10% glycerol and 1mg/ml p-Phenylenediamine (Sigma).

Samples were imaged using an Applied Precision DeltaVision Core wide-field fluorescence microscope system (GE Healthcare, Issaquah, WA) equipped with a PlanApo 60x 1.40 numerical aperture objective lens (Olympus America). Image stacks were taken at 0.2nm intervals throughout the entire cell and deconvolved using Applied Precision SoftWoRx software version 4.2.1. with default parameters. Separate projections of RNA-FISH/DAPI and DNA-FISH/DAPI images were overlayed in Photoshop using heterochromatin foci as a guide. Image analysis was performed manually using Applied Precision SoftWoRx software.

Quantitative RT-PCR

cDNA was prepared as above and used for quantitative RT-PCR using SYBR green reagents (Applied Biosciences). Q-PCR was performed using the following forward and reverse primer sequences: 18S GGGCCCGAAGCGTTTACTTT, CGCCGGTCCAAGAATTTCAC; Acot1 CATCACCTTTGGAGGGGAGC, TGTACCTTTCCCCAACCTCC; Capn5 ACACGTCAGAGGAATGGCAG, GGATGCTCAGGTAGGACGTG; Cbr3 GTCCCTCTGACATGTCGTCC, CGTTAAGTCCCCCGTACTCC; CycloB GACAGACAGCCGGGACAAGC, GGGGATTGACAGGACCCACA; Gapdh1 GGTGGTGAAGCAGGCATCTG, CGGCATCGAAGGTGGAAGAG; Rhoj GGCCACTCTCTTACCCCAAC, GAGGCATGCAGTCCTTCAGT. Three biological replicates for each sample were used for each experiment, and values normalized to the geometric mean of at least three separate housekeeping genes.

Bisulfite analysis

500ng of purified DNA (Trizol reagent, Ambion) was converted with bisulfite (EZ DNA methylation-Gold kit, Zymo) according to the manufacturer’s instructions. Primers amplifying A-T SNPs within promoter CpG islands were designed using MethPrimer (Li and Dahiya, 2002). PCR was performed using OneTaq Hot Start DNA Polymerase (NEB). See Supplemental Table 5 for primer. Products were gel purified and cloned (Topo-TA, Invitrogen). Clones were sent for Sanger sequencing and analyzed using BiQ Analyzer (Bock et al., 2005).

Chromatin immunoprecipitation

Chromatin was prepared from cells, and immunoprecipitations performed as described in extended experimental procedures. Quantitative RT-PCR was performed on immunoprecipitated DNA using primers amplifying promoter or genic regions of monoallelically expressed genes (see Supplemental Table 4) and normalized to input DNA. Products containing informative SNPs were subsequently analyzed by Sanger sequencing. At least three biological replicates were analyzed per sample.

Supplementary Material

01
02
03
04

Highlights.

  • Monoallelic expression increases 5.6 fold during ESC to NPC differentiation

  • Independent stochastic regulation of the alleles results in monoallelic expression

  • Histone modifications, not DNA methylation, distinguish active and inactive alleles

  • In 8% of cases the single active allele transcriptionally compensates

Acknowledgments

We thank C. Vakoc (CSHL) for kindly providing the C57Bl/6J x CAST/EiJ ESC cell line, A. Mills (CSHL) for AB2.2 ESC line, C. Davis, J. Drenkow and T. Gingeras (CSHL) for assistance in RNA-Sequencing library preparation, E. Hodges and G. Hannon (CSHL) for assistance in DNA methylation studies, S. Hearn for assistance in microscopy, MJ. Delás for assistance with DNA-FISH, members of the Spector Laboratory for discussion and comments. M.A.E-M. and D.L.S. conceived the study, designed the experiments, and wrote the manuscript; M.A.E-M. and J.H.B. performed experiments; D.T., J.M. and P.F. designed and performed all bioinformatic analysis. M.A.E.-M. is supported by a Genentech Foundation Fellowship and George A. and Marjorie H. Anderson Fellowship. D.T, J.C.M and P.F are supported by the European Molecular Biology Laboratory. D.T. and P.F. are supported by the Wellcome Trust (WT095908). J.H.B. is supported by a DAAD Post-Doctoral Fellowship. Research in the Spector lab is supported by NIGMS 42694 and NCI 2P30CA45508.

Footnotes

Accession numbers

RNA-seq and microarray data are available in the Array Express database under accession numbers E-MTAB-1822 and E-MTAB-1823 respectively.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Alabert C, Groth A. Chromatin replication and epigenome maintenance. Nat Rev Mol Cell Biol. 2012;13:153–167. doi: 10.1038/nrm3288. [DOI] [PubMed] [Google Scholar]
  2. Aseem O, Barth JL, Klatt SC, Smith BT, Argraves WS. Cubilin expression is monoallelic and epigenetically augmented via PPARs. BMC Genomics. 2013;14:405–423. doi: 10.1186/1471-2164-14-405. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bartolomei MS, Ferguson-Smith AC. Mammalian genomic imprinting. Cold Spring Harbor Perspectives in Biology. 2011;3:a002592. doi: 10.1101/cshperspect.a002592. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Black JC, Van Rechem C, Whetstine JR. Histone lysine methylation dynamics: establishment, regulation, and biological impact. Molecular Cell. 2012;48:491–507. doi: 10.1016/j.molcel.2012.11.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bock C, Reither S, Mikeska T, Paulsen M, Walter J, Lengauer T. BiQ Analyzer: visualization and quality control for DNA methylation data from bisulfite sequencing. Bioinformatics. 2005;21:4067–4068. doi: 10.1093/bioinformatics/bti652. [DOI] [PubMed] [Google Scholar]
  6. Canham MA, Sharov AA, Ko MSH, Brickman JM. Functional heterogeneity of embryonic stem cells revealed through translational amplification of an early endodermal transcript. PLoS Biol. 2010;8:e1000379. doi: 10.1371/journal.pbio.1000379. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chess A. Mechanisms and consequences of widespread random monoallelic expression. Nature Reviews Genetics. 2012a;13:421–428. doi: 10.1038/nrg3239. [DOI] [PubMed] [Google Scholar]
  8. Chess A. Random and Non-Random Monoallelic Expression. Neuropsychopharmacology. 2012b;38:55–61. doi: 10.1038/npp.2012.85. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chess A, Simon I, Cedar H, Axel R. Allelic inactivation regulates olfactory receptor gene expression. Cell. 1994;78:823–834. doi: 10.1016/s0092-8674(94)90562-2. [DOI] [PubMed] [Google Scholar]
  10. Clowney EJ, Legros MA, Mosley CP, Clowney FG, Markenskoff-Papadimitriou EC, Myllys M, Barnea G, Larabell CA, Lomvardas S. Nuclear aggregation of olfactory receptor genes governs their monogenic expression. Cell. 2012;151:724–737. doi: 10.1016/j.cell.2012.09.043. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Clowney EJ, Magklara A, Colquitt BM, Pathak N, Lane RP, Lomvardas S. High-throughput mapping of the promoters of the mouse olfactory receptor genes reveals a new type of mammalian promoter and provides insight into olfactory receptor gene regulation. Genome Research. 2011;21:1249–1259. doi: 10.1101/gr.120162.110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Conti L, Pollard SM, Gorba T, Reitano E, Toselli M, Biella G, Sun Y, Sanzone S, Ying QL, Cattaneo E, et al. Niche-Independent Symmetrical Self-Renewal of a Mammalian Tissue Stem Cell. PLoS Biol. 2005;3:e283. doi: 10.1371/journal.pbio.0030283. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Donley N, Stoffregen EP, Smith L, Montagna C, Thayer MJ. Asynchronous Replication, Mono-Allelic Expression, and Long Range Cis-Effects of ASAR6. PLoS Genetics. 2013;9:e1003423. doi: 10.1371/journal.pgen.1003423. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Dutta D, Ensminger AW, Zucker JP, Chess A. Asynchronous Replication and Autosome-Pair Non-Equivalence in Human Embryonic Stem Cells. PLoS ONE. 2009;4:e4970. doi: 10.1371/journal.pone.0004970. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Esumi S, Kakazu N, Taguchi Y, Hirayama T, Sasaki A, Hirabayashi T, Koide T, Kitsukawa T, Hamada S, Yagi T. Monoallelic yet combinatorial expression of variable exons of the protocadherin-α gene cluster in single neurons. Nature Genetics. 2005;37:171–176. doi: 10.1038/ng1500. [DOI] [PubMed] [Google Scholar]
  16. Fisher CL, Fisher AG. Chromatin states in pluripotent, differentiated, and reprogrammed cells. Current Opinion in Genetics & Development. 2011;21:140–146. doi: 10.1016/j.gde.2011.01.015. [DOI] [PubMed] [Google Scholar]
  17. Gimelbrant A, Hutchinson JN, Thompson BR, Chess A. Widespread Monoallelic Expression on Human Autosomes. Science. 2007;318:1136–1140. doi: 10.1126/science.1148910. [DOI] [PubMed] [Google Scholar]
  18. Guidi CJ, Veal TM, Jones SN, Imbalzano AN. Transcriptional compensation for loss of an allele of the Ini1 tumor suppressor. J Biol Chem. 2004;279:4180–4185. doi: 10.1074/jbc.M312043200. [DOI] [PubMed] [Google Scholar]
  19. Guo L, Hu-Li J, Paul WE. Probabilistic Regulation in TH2 Cells Accounts for Monoallelic Expression of IL-4 and IL-13. Immunity. 2005;23:89–99. doi: 10.1016/j.immuni.2005.05.008. [DOI] [PubMed] [Google Scholar]
  20. Guo M, Birchler JA. Trans-acting dosage effects on the expression of model gene systems in maize aneuploids. Science. 1994;266:1999–2002. doi: 10.1126/science.266.5193.1999. [DOI] [PubMed] [Google Scholar]
  21. Gutierrez-Arcelus M, Lappalainen T, Montgomery SB, Buil A, Ongen H, Yurovsky A, Bryois J, Giger T, Romano L, Planchon A, et al. Passive and active DNA methylation and the interplay with genetic variation in gene regulation. Elife. 2013;2:e00523. doi: 10.7554/eLife.00523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Held W, Kunz B. An allele-specific, stochastic gene expression process controls the expression of multiple Ly49 family genes and generates a diverse, MHC-specific NK cell receptor repertoire. Eur J Immunol. 1998;28:2407–2416. doi: 10.1002/(SICI)1521-4141(199808)28:08<2407::AID-IMMU2407>3.0.CO;2-D. [DOI] [PubMed] [Google Scholar]
  23. Hiratani I, Gilbert DM. Replication timing as an epigenetic mark. Epigenetics. 2009;4:93–97. doi: 10.4161/epi.4.2.7772. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Homma S, Iwasaki M, Shelton GD, Engvall E, Reed JC, Takayama S. BAG3 deficiency results in fulminant myopathy and early lethality. Am J Pathol. 2006;169:761–773. doi: 10.2353/ajpath.2006.060250. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Huang S. Systems biology of stem cells: three useful perspectives to help overcome the paradigm of linear pathways. Philos Trans R Soc Lond, B, Biol Sci. 2011;366:2247–2259. doi: 10.1098/rstb.2011.0008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Hübner MR, Eckersley-Maslin M, Spector DL. Chromatin organization and transcriptional regulation. Current Opinion in Genetics & Development. 2013;23:89–95. doi: 10.1016/j.gde.2012.11.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Ishii I, Akahoshi N, Yamada H, Nakano S, Izumi T, Suematsu M. Cystathionine gamma-Lyase-deficient mice require dietary cysteine to protect against acute lethal myopathy and oxidative injury. Journal of Biological Chemistry. 2010;285:26358–26368. doi: 10.1074/jbc.M110.147439. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Jeffries AR, Perfect LW, Ledderose J, Schalkwyk LC, Bray NJ, Mill J, Price J. Stochastic Choice of Allelic Expression in Human Neural Stem Cells. Stem Cells. 2012;30:1938–1947. doi: 10.1002/stem.1155. [DOI] [PubMed] [Google Scholar]
  29. Kaasik A, Kuum M, Aonurm A, Kalda A, Vaarmann A, Zharkovsky A. Seizures, ataxia, and neuronal loss in cystatin B heterozygous mice. Epilepsia. 2007;48:752–757. doi: 10.1111/j.1528-1167.2007.00985.x. [DOI] [PubMed] [Google Scholar]
  30. Kaneko R, Kato H, Kawamura Y, Esumi S, Hirayama T, Hirabayashi T, Yagi T. Allelic gene regulation of Pcdh-alpha and Pcdh-gamma clusters involving both monoallelic and biallelic expression in single Purkinje cells. J Biol Chem. 2006;281:30551–30560. doi: 10.1074/jbc.M605677200. [DOI] [PubMed] [Google Scholar]
  31. Kelsey G, Feil R. New insights into establishment and maintenance of DNA methylation imprints in mammals. Philos Trans R Soc Lond, B, Biol Sci. 2013;368:20110336. doi: 10.1098/rstb.2011.0336. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Li H, Durbin R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics. 2010;26:589–595. doi: 10.1093/bioinformatics/btp698. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Li LC, Dahiya R. MethPrimer: designing primers for methylation PCRs. Bioinformatics. 2002;18:1427–1431. doi: 10.1093/bioinformatics/18.11.1427. [DOI] [PubMed] [Google Scholar]
  34. Li SM, Valo Z, Wang J, Gao H, Bowers CW, Singer-Sam J. Transcriptome-Wide Survey of Mouse CNS-Derived Cells Reveals Monoallelic Expression within Novel Gene Families. PLoS ONE. 2012;7:e31751. doi: 10.1371/journal.pone.0031751. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Magklara A, Yen A, Colquitt BM, Clowney EJ, Allen W, Markenscoff-Papadimitriou E, Evans ZA, Kheradpour P, Mountoufaris G, Carey C, et al. An Epigenetic Signature for Monoallelic Olfactory Receptor Expression. Cell. 2011;145:555–570. doi: 10.1016/j.cell.2011.03.040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Martinez Arias A, Brickman JM. Gene expression heterogeneities in embryonic stem cell populations: origin and function. Current Opinion in Cell Biology. 2011;23:650–656. doi: 10.1016/j.ceb.2011.09.007. [DOI] [PubMed] [Google Scholar]
  37. Mattout A, Meshorer E. Chromatin plasticity and genome organization in pluripotent embryonic stem cells. Current Opinion in Cell Biology. 2010;22:334–341. doi: 10.1016/j.ceb.2010.02.001. [DOI] [PubMed] [Google Scholar]
  38. McAnally AA, Yampolsky LY. Widespread transcriptional autosomal dosage compensation in Drosophila correlates with gene expression level. Genome Biol Evol. 2010;2:44–52. doi: 10.1093/gbe/evp054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Michaelson J. Cellular selection in the genesis of multicellular organization. Lab Invest. 1993;69:136–151. [PubMed] [Google Scholar]
  40. Mikkelsen TS, Ku M, Jaffe DB, Issac B, Lieberman E, Giannoukos G, Alvarez P, Brockman W, Kim TK, Koche RP, et al. Genome-wide maps of chromatin state in pluripotent and lineage-committed cells. Nature. 2007;448:553–560. doi: 10.1038/nature06008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Parkhomchuk D, Borodina T, Amstislavskiy V, Banaru M, Hallen L, Krobitsch S, Lehrach H, Soldatov A. Transcriptome analysis by strand-specific sequencing of complementary DNA. Nucleic Acids Research. 2009;37:e123. doi: 10.1093/nar/gkp596. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Peric-Hupkes D, Meuleman W, Pagie L, Bruggeman SWM, Solovei I, Brugman W, Gräf S, Flicek P, Kerkhoven RM, van Lohuizen M, et al. Molecular Maps of the Reorganization of Genome-Nuclear Lamina Interactions during Differentiation. Molecular Cell. 2010;38:603–613. doi: 10.1016/j.molcel.2010.03.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Pernis B, Chiappino G, Kelus AS, Gell PG. Cellular localization of immunoglobulins with different allotypic specificities in rabbit lymphoid tissues. J Exp Med. 1965;122:853–876. doi: 10.1084/jem.122.5.853. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Rougeulle C. Promoter-restricted H3 Lys 4 di-methylation is an epigenetic mark for monoallelic expression. Human Molecular Genetics. 2003;12:3343–3348. doi: 10.1093/hmg/ddg351. [DOI] [PubMed] [Google Scholar]
  45. Schulz EG, Heard E. Role and control of X chromosome dosage in mammalian development. Current Opinion in Genetics & Development. 2013;23:109–115. doi: 10.1016/j.gde.2013.01.008. [DOI] [PubMed] [Google Scholar]
  46. Skok JA, Brown KE, Azuara V, Caparros ML, Baxter J, Takacs K, Dillon N, Gray D, Perry RP, Merkenschlager M, et al. Nonequivalent nuclear location of immunoglobulin alleles in B lymphocytes. Nat Immunol. 2001;2:848–854. doi: 10.1038/ni0901-848. [DOI] [PubMed] [Google Scholar]
  47. Smith ZD, Meissner A. DNA methylation: roles in mammalian development. Nature Reviews Genetics. 2013;14:204–220. doi: 10.1038/nrg3354. [DOI] [PubMed] [Google Scholar]
  48. Takizawa T, Gudla PR, Guo L, Lockett S, Misteli T. Allele-specific nuclear positioning of the monoallelically expressed astrocyte marker GFAP. Genes & Development. 2008;22:489–498. doi: 10.1101/gad.1634608. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Trieu M, Ma A, Eng SR, Fedtsova N, Turner EE. Direct autoregulation and gene dosage compensation by POU-domain transcription factor Brn3a. Development. 2003;130:111–121. doi: 10.1242/dev.00194. [DOI] [PubMed] [Google Scholar]
  50. Wang J, Valo Z, Smith D, Singer-Sam J. Monoallelic expression of multiple genes in the CNS. PLoS ONE. 2007;2:e1293. doi: 10.1371/journal.pone.0001293. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Wheway G, Abdelhamed Z, Natarajan S, Toomes C, Inglehearn C, Johnson CA. Aberrant Wnt signalling and cellular over-proliferation in a novel mouse model of Meckel-Gruber syndrome. Dev Biol. 2013;377:55–66. doi: 10.1016/j.ydbio.2013.02.015. [DOI] [PubMed] [Google Scholar]
  52. Zwemer LM, Zak A, Thompson B, Kirby A, Daly MJ, Chess A, Gimelbrant A. Autosomal monoallelic expression in the mouse. Genome Biology. 2012;13:R10. doi: 10.1186/gb-2012-13-2-r10. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

01
NIHMS560517-supplement-01.pdf (17.6MB, pdf)
02
NIHMS560517-supplement-02.xlsx (66.9KB, xlsx)
03
NIHMS560517-supplement-03.xlsx (27.8MB, xlsx)
04
NIHMS560517-supplement-04.xlsx (48.1KB, xlsx)

RESOURCES