RNA sequencing-based screen for reactivation of silenced alleles of autosomal genes

Abstract In mammalian cells, maternal and paternal alleles usually have similar transcriptional activity. Epigenetic mechanisms such as X-chromosome inactivation (XCI) and imprinting were historically viewed as rare exceptions to this rule. Discovery of autosomal monoallelic autosomal expression (MAE) a decade ago revealed an additional allele-specific mode regulating thousands of mammalian genes. Despite MAE prevalence, its mechanistic basis remains unknown. Using an RNA sequencing-based screen for reactivation of silenced alleles, we identified DNA methylation as key mechanism of MAE mitotic maintenance. In contrast with the all-or-nothing allelic choice in XCI, allele-specific expression in MAE loci is tunable, with exact allelic imbalance dependent on the extent of DNA methylation. In a subset of MAE genes, allelic imbalance was insensitive to DNA demethylation, implicating additional mechanisms in MAE maintenance in these loci. Our findings identify a key mechanism of MAE maintenance and provide basis for understanding the biological role of MAE.


Fig.S2
Concordance between UMI vs non-UMI based AI measurement from Screen-seq. 3

Fig.S5
Dose response curve for Adnp2 and cell viability of Abl.1 clone in 'biological replicate experiment'. 7

Fig.S9
Change in Xist allele imbalance in Abl.1 cells is very subtle on exposure to 5-aza-dC treatment. 11

Fig.S10
Changes of allele-specific expression of autosomal genes appear independent of CpG islands. 12 . Genomic DNA (gDNA) mixes ranging from 0 to 0.1 and 0.9 to 1 in 0.025 increments, as well as 0.25, 0.5, and 0.75, were made from parental liver tissue and shown on X-axis (Expected maternal AI). For AI measured from Screen-seq (Measured AI, Y-axis), data points are an average of 3 biological replicates and error bars show standard deviation. Red line denotes linear fit to smoothen all data points. Expected and measured AI were highly concordant (R 2 ≥ 0.99) at >1000 reads/SNP for readout genes.

Fig.S2. Concordance between UMI vs non-UMI based AI measurement from Screen-seq.
We compared AI sensitivity for UMI and non-UMI assays, by designing both types of assays for a subset of genes where the position of SNPs allowed that. For this, we used mixes prepared from total RNA from the spleens of the mice of the parental mouse strains. Both types of assays were designed for a subset of genes (Adamtsl4, Adnp2, Dnajc12, Smtnl2) where the position of SNPs allowed that. AI measurements were highly concordant between the UMI and non-UMI assays (R 2 ≥ 0.97) at >1000 reads/SNP.  Effect of 5-aza-cytidine (5-aza-C), a closely related analog of 5-aza-dC, on AI. AI was measured using ddPCR after treating cells for 2 days (black circles) and 5 days (red circles) and shown on the Y-axis in Abl.1 cells for left -Col6a5 and for right -Dnajc12 readout genes. (A) 5-aza-dC exposure/recovery experiment in Abl.1 cells. Cells were exposed to 0.2, 0.5 or 1 µM 5-aza-dC in growth medium for 2 days. Control cells were not treated (NT) with drug and only grown in growth medium. After 2 days, cells exposed to the drug were washed and grown only in growth medium (without drug, called recovery phase) for the rest of the days. Cells were collected on days 2, 5, 7, 9, and 12. For cell numbers on these days measured using automated trypan blue assay, see Suppl. Table S7. (B) AI measurements for Dnajc12 on day 9 after recovery. Scatterplots for 20,000 droplets targeting the readout gene, Col6a5. 5-aza-dC concentration used during first 2 days is shown in the plots. Black circles: empty droplets; blue: droplets with the Cast paternal allele amplified (labeled by FAM fluorophore); red: droplets with the 129 maternal allele amplified (labeled by HEX fluorophore). Ratio of red:blue droplets (AI) is given in the top left of each scatterplot. AI value written in red is the maternal AI. AI of Xist was measured using ddPCR. Scatterplots for 20,000 droplets targeting Xist in control untreated samples (upper panel) and 5-aza-dC treated Abl.1 cells (lower panel). cDNA samples from day 7 of screening were assessed using ddPCR with allele-specific fluorescent probes. Grey: empty droplets; blue: droplets with Cast allele amplified; red: droplets with 129 allele amplified. Ratio of red:blue droplets (AI) is written on the upper right corner of each plot. Next to red and blue droplets are the copies/µl measured for 129 and Cast Xist allele.

Supplementary Figure S7
Supplementary Figure S10 Comparing the CpG islands in the 200kb genomic region window for readout genes (shown in red) Col6a5, Dnajc12 (genes responsive to DNA demethylation) and Fam217b (gene not responsive to DNA demethylation). The CpG island tracks (shown in green) and genomic coordinates were obtained from UCSC browser (mm10).