Figure 1.
(A) F0 and F1 mice were generated via the depicted crosses. The schematic diagram illustrates example expression patterns for a cis effect, trans effect, and parent-of-origin effect. For a cis effect, in the F1 hybrids, the Cast/EiJ allelic expression relative to the C57BL/6J allelic expression recapitulates the ratio of gene expression levels in the F0 homozygotes. For a trans effect, the F1 hybrids express the Cast/EiJ and C57BL/6J alleles equally. For a parent-of-origin effect, there is preferential expression of the maternal allele (as depicted) or the paternal allele, as seen by comparison of the reciprocal F1 hybrids. (B) An overview of the workflow is shown.
Table 1.
Agreement between F0 biological replicates.
Table 2.
Agreement between F1 biological replicates.
Table 3.
Accuracy of X chromosomal read mapping in F1 samples.
Figure 2.
Characterization of parent-of-origin effects in the retina.
Autosomal genes polymorphic between C57BL/6J and Cast/EiJ were analyzed in retinas from reciprocal F1 hybrids. Higher Bayes factors indicate greater likelihood of imprinting. (A) Non-imprinted genes with Bayes factor <0.1 (gray) and <0.001 (orange) are depicted. (B) Parent-of-origin effects with preferential expression of the paternal (blue) or maternal (red) allele with Bayes factor ≥10 (light) and ≥30 (dark) are depicted. (C) Top-ranked (low rank number) genes are enriched for known imprinted genes. (D) Genes with strong evidence of imprinting in the retina (Bayes factor ≥30) that exhibit preferential expression of the paternal (blue) or maternal (red) allele are depicted. Green, special cases—see text for discussion of Rtl1 and Grb10. Filled squares, genes previously reported as imprinted in other tissues. Empty triangles, not previously reported as imprinted. A230006K03Rik appears twice because it is associated with two Ensembl ID's, a protein-coding gene and a lincRNA.
Figure 3.
Comparison of differentially expressed and cis-effect genes associated with photoreceptor CREs.
Genes were classified as being associated with CRX ChIP-seq peaks (CBR-associated) or not. (A) Differentially expressed (DE) autosomal genes were identified using DESeq at 5% FDR. The proportions of genes with higher expression in F0 Cast/EiJ than F0 C57BL/6J at various fold changes are shown. (B) Cis-effect autosomal genes were identified using MMDIFF. Proportions of genes with higher expression in F1 Cast/EiJ allele than F1 C57BL/6J allele at various fold changes are shown. P-values were calculated with two-tailed Fisher's exact test. N.S. = not significant, *<0.05, **<0.01, ***<0.001, **** <0.0001.
Figure 4.
Classification of genes by mechanism of gene regulatory divergence.
(A) Genes were classified as conserved (yellow; largely obscured), cis (green), trans (red), or cis and trans (purple). (B) Cis- and trans-effect genes were further subcategorized as to whether the cis and trans effects acted in the same (“+” sign; pink and brown) or opposite (“−” sign; orange and blue) directions, and whether the cis (CAPS; orange and pink) or trans (CAPS; blue and brown) effect was stronger. Insets, magnified views.
Figure 5.
Analysis of variant density in photoreceptor CREs.
(A) The number of Cast/EiJ (top) or Spret/EiJ (bottom) SNPs and indels relative to C57BL/6J was determined in 50 bp windows (sliding 25 bp at a time) across the 2 kb region centered on CBRs. Phylogenetic conservation for CBRs is based on PhastCons scores as found in [6]. The highlighted area corresponds to the central 300 bp region. (B) Histogram showing frequency of variants (SNPs+indels) in the 1 kb region centered on all CBRs (black), CBRs associated with cis-effect genes (green), and CBRs associated with trans-effect genes (red). Total bar height was normalized to 1 for each category.
Figure 6.
Cis-effect genes associated with retinal disease and photoreceptor CREs.
(A) Cis-effect genes associated with CRX ChIP-seq peaks were matched against the RetNet database of retinal disease genes. The yellow circle highlights Sag. (B) Sag locus, mm9. Top: Screenshot from UCSC Genome Browser [77]. DNaseI hypersensitivity site (DHS) signals are from ENCODE data [8]. Bottom: Enlargement of boxed region. The 0.7 kb promoter region is depicted here. Locations of known Cast/EiJ variants [19] are depicted as green tic marks (SNPs) or blue tic marks (indels). (C) Retinal explant electroporation was used to assay the activity of the 0.7 kb Sag promoter region of B6 and Cast alleles. Representative images are shown here for the B6 (top) and Cast (bottom) 0.7 kb Sag promoter constructs driving DsRed expression. Rho-CBR3-eGFP served as the loading control. (D) Quantification of the cis-regulatory activity measured by the explant electroporation assay. Error bar represents SEM. P-value was calculated with one-tailed Wilcoxon rank-sum test.
Figure 7.
Comparison of cis effects and trans effects between liver and retina.
(A) Using the same analytic pipeline as for retina, genes in the liver were classified as conserved (yellow; largely obscured), cis (green), trans (red), or cis and trans (purple). (B) Cis- and trans- regulated genes were further subcategorized as to whether the cis and trans effects acted in the same (“+” sign; pink and brown) or opposite (“−” sign; orange and blue) directions, and whether the cis (CAPS; orange and pink) or trans (CAPS; blue and brown) effect was stronger. (C) Number of genes classified as cis in liver and conserved in retina, cis in both tissues, or cis in retina and conserved in liver. (D) Number of genes classified as trans in liver and conserved in retina, trans in both tissues, or trans in retina and conserved in liver. (E) Correlation between genes classified as cis in both tissues. Pearson r values for F0 samples (left) and F1 samples (right) are shown. (F) Correlation between genes classified as trans in both tissues. Pearson r values for F0 samples (left) and F1 samples (right) are shown. Insets, magnified view.