A novel Cbx1, PurB, and Sp3 complex mediates long-term silencing of tissue- and lineage-specific genes

Abstract

miRNA-based cellular fate reprogramming offers an opportunity to investigate the mechanisms of long-term gene silencing. To further understand how genes are silenced in a tissue-specific manner, we leveraged our miRNA-based method of reprogramming fibroblasts into cardiomyocytes. Through screening approaches, we identified three proteins that were downregulated during reprogramming of fibroblasts into cardiomyocytes: heterochromatin protein Cbx1, transcriptional activator protein PurB, and transcription factor Sp3. We show that knockdown of Cbx1, PurB, and Sp3 was sufficient to induce cardiomyocyte gene expression in fibroblasts. Similarly, gene editing to ablate Cbx1, PurB, and Sp3 expression induced fibroblasts to convert into cardiomyocytes in vivo. Furthermore, high-throughput DNA sequencing and coimmunoprecipitation experiments indicated that Cbx1, PurB, and Sp3 also bound together as a complex and were necessary to localize nucleosomes to cardiomyocyte genes on the chromosome. Finally, we found that the expression of these genes led to nucleosome modification via H3K27me3 (trimethylated histone-H3 lysine-27) deposition through an interaction with the polycomb repressive PRC2 complex. In summary, we conclude that Cbx1, PurB, and Sp3 control cell fate by actively repressing lineage-specific genes.

Keywords: cardiac development; cardiac muscle; cardiomyocyte; fibroblast; regeneration; reprogramming; transcription repressor.

Figure 1

Cbx1, PurB, and Sp3 are repressors of cardiomyocyte-specific genes.A, RNA-Seq read counts from the study by Hodgkinson et al. (18) were analyzed for the mRNA levels of the indicated genes in freshly isolated cardiomyocytes and noncardiomyocytes. Individual data points (open circles) and mean (horizontal bar) are shown. One-way ANOVA with Bonferroni post hoc tests was used to determine significance; ∗∗∗p < 0.001. N = 3. B, cardiac fibroblasts were transfected with the direct cardiac reprogramming cocktail miR combo. A nontargeting miRNA (negmiR) was used as a control. Four days after transfection, expression of the indicated transcription factors was determined by quantitative PCR (qPCR). Expression values were normalized to negmiR. N = 5. Individual data points (open circles) and mean (horizontal bar) are shown. One-way ANOVA with Bonferroni post hoc tests was used to determine significance; ∗∗∗p < 0.001. C, cardiac fibroblasts were transfected with siRNA targeting an individual putative repressor. A nontargeting siRNA was used as a control. After 4 days, expression was determined by qPCR. Expression values were normalized to the control siRNA. N = 3. Individual data points (open circles) and mean (horizontal bar) are shown. One-way ANOVA with Bonferroni post hoc tests was used to determine significance; ∗∗∗p < 0.001. D, cardiac fibroblasts were transfected with siRNA targeting an individual putative repressor. A nontargeting siRNA was used as a control. After 14 days, expression of the indicated cardiomyocyte-specific genes was determined by qPCR. Expression values were normalized to the control siRNA. The heatmap summarizes the results of ten cardiomyocyte-specific genes. Increased expression of greater than twofold and with a significance <0.05 is shown in red. N = 3 to 5. One-way ANOVA with Bonferroni post hoc tests was used to determine significance; ∗p < 0.05. E, cardiac fibroblasts were transfected with siRNA targeting an individual putative repressor. A nontargeting siRNA was used as a control. After 14 days, the cells were incubated with antibodies to the cardiomyocyte-specific protein Actn2. Representative images are shown. N = 4. The scale bar represents 50 microns. F, higher resolution images of the cells shown in E are shown to display sarcomeres. The scale bar represents 50 microns. N = 4. Quantification of the percentage of Actn2+ cells displaying sarcomeres. Individual data points (open circles) and mean (horizontal bar) are shown. One-way ANOVA with Bonferroni post hoc tests was used to determine significance; ∗∗p < 0.01.

Figure 2

Loss of Cbx1, PurB, and Sp3 expression is associated with cardiomyocyte differentiation.A, cardiac fibroblasts were transfected with either a control nontargeting siRNA or a pool of siRNAs targeting Cbx1, PurB, and Sp3. After 14 days, expression of the indicated cardiomyocyte, fibroblast, endothelial, and neuronal-specific genes was determined by quantitative PCR (qPCR). For completeness, a number of general genes were included. Expression values were normalized to the control siRNA. N = 3. B, RNA was extracted from iPS cells and iPS-derived cardiomyocytes. Expression of the indicated genes was determined by qPCR, and expression values in iPS-derived cardiomyocytes were normalized to iPS cells. N = 4. Individual data points (open circles) and mean (horizontal bar) are shown. One-way ANOVA with Bonferroni post hoc tests was used to determine significance; ∗∗∗p < 0.01. C, RNA-Seq of iPS differentiation into cardiomyocytes from the study by Churko et al. (20). Expression values of the indicated genes in iPS-derived cardiomyocytes were normalized to iPS cells. N = 1. D, RNA-Seq from the study by Pavlovic et al. (21). Cardiac tissue was isolated from 12 individuals and used to generate iPS cells. These iPS cells were then differentiated into cardiomyocytes. RNA-Seq was performed with the source cardiac tissue; undifferentiated iPS cells; as well as iPS-derived cardiomyocytes. N = 12. Read number is shown. Individual data points (open circles) and mean (horizontal bar) are shown. One-way ANOVA with Bonferroni post hoc tests was used to determine significance; ∗∗p < 0.01 and ∗∗∗p < 0.001. iPS, inducible pluripotent stem.

Figure 3

Ablation of Cbx1, PurB, and Sp3 expression reprograms fibroblasts into cardiomyocytes.A, cardiac tissue was isolated from 8-week-old Fsp1-Cre:tdTomato mice. In these mice, Fsp1 fibroblasts are marked permanently with tdTomato. Tissue slices were incubated with tdTomato (red) and repressor (green) antibodies. Nuclei (blue) were stained with 4′,6-diamidino-2-phenylindole (DAPI). Representative images from three individual mice. The scale bar represents 50 microns. B, cardiomyocytes and fibroblasts were isolated from 1-day-old C57BL6 mice. RNA was analyzed for the expression of the cardiomyocyte marker Scn5a, the fibroblast marker Postn, as well as the expression of the three repressors. Expression values are shown as a fold enrichment in fibroblasts when compared with cardiomyocytes. N = 3. Individual data points (open circles) and mean (horizontal bar) are shown. One-way ANOVA with Bonferroni post hoc tests was used to determine significance; ∗∗∗p < 0.001. C, guide RNAs (gRNAs) for Cbx1, PurB, and Sp3 were cloned into a plasmid containing CRISPR-associated protein 9 (Cas9), and the resulting construct was transfected into cultured cardiac fibroblasts. After 7 days, protein extracts were probed for the presence of Cbx1, PurB, or Sp3. N = 3. Representative blots are shown with the loading control Gapdh. D, the Cbx1, PurB, and Sp3 gRNAs were cloned into a lentivirus-generating plasmid containing Cas9. Control nontargeting gRNA was cloned into the same plasmid as a control. Lentiviral particles were isolated and injected into the heart of an 8-week-old C57BL6 mouse. One week after cardiac injection, tissue slices were analyzed for repressor expression (green). Nuclei (blue) were visualized via DAPI. The scale bar represents 50 microns. Representative images from three individual mice. E, the Cbx1, PurB, and Sp3 gRNAs were cloned into a lentivirus-generating plasmid containing Cas9. Lentiviral particles were injected into the hearts of fibroblast lineage-tracing mouse Fsp1-Cre:tdTomato. In this model, fibroblasts and their progeny are permanently labeled with the fluorescent protein tdTomato. Two months after injection, heart sections within 500 microns of the injection site were incubated with tdTomato and cardiac troponin-T (cardiomyocyte-specific marker) antibodies. Representative images are shown. The scale bar represents 50 microns. N = 3 per group. The number of cardiomyocytes derived from the reprogramming of fibroblasts (tdTomato+ cardiac troponin-T+) is expressed as a percentage of the total cardiomyocyte (cardiac troponin-T+) population. A two-tailed t test was used to determine significance between the two groups; ∗∗∗p < 0.001.

Figure 4

Cbx1, PurB, and Sp3 bind to cardiomyocyte-specific genes in fibroblasts.A, chromatin derived from cardiac fibroblasts was incubated with antibodies for Cbx1, PurB, or Sp3. An isotype antibody was used as a control. Immunoprecipitated DNA was analyzed by high-throughput sequencing. Bioinformatic approaches were used to determine Cbx1-, PurB-, and Sp3-binding sites. The Venn diagram details the number of genes with Cbx1-, PurB-, and Sp3-binding sites. B, Cbx1-, PurB-, and Sp3-binding sites in the cardiomyocyte-specific genes Ttn, Ryr2, and Kcnj6. C, Cbx1-, PurB-, and Sp3-binding peaks in the Nebl gene. D, Gene Ontology analysis of the genes to which Cbx1, PurB, and Sp3 were bound.

Figure 5

Cbx1, PurB, and Sp3 regulate chromatin architecture. Cardiac fibroblasts were transfected with siRNAs targeting Cbx1, PurB, or Sp3. A nontargeting siRNA was used as a control. After 7 days, chromatin was isolated and digested with micrococcal nuclease (MNase). Following MNase digestion, the resulting undigested DNA was submitted for high-throughput sequencing (MNase-Seq) and mapped to the mouse genome. A, MNase digestion was optimized to give rise to one nucleosome. Read lengths were analyzed after sequencing and summed. As expected, the majority of read lengths were 1 nucleosome is size (∼150 bp). B, MNase accessibility signals around transcription start sites (TSSs). The y-axis represents the read number for each 10 bp bin normalized to the effective genome size for the mouse. C, nucleosomes (black bars) were plotted on the cardiomyocyte-specific genes Ryr2 and Actn2. D, nucleosomes in noncardiomyocyte genes.

Figure 6

Cbx1, PurB, and Sp3 regulate the PRC2 complex.A, endogenous Cbx1 and PurB was immunoprecipitated from cardiac fibroblast cell lysates. An isotype control antibody was used as a control. Immunoprecipitates were immunoblotted with a Cbx1, PurB, and a Sp3 antibody. The first lane contains cell extract (1/10th immunoprecipitation input). Representative immunoblots are shown from three independent experiments. B, cardiac fibroblasts were transfected with siRNAs targeting Cbx1, PurB, and Sp3. A nontargeting siRNA was used as a control. After 4 days, cell lysates were immunoblotted with H3K27me3 and H3 antibodies. Immunoblotting for Gapdh was used as a loading control. Representative immunoblots are shown from four independent experiments. C, endogenous Cbx1, PurB, and Sp3 was immunoprecipitated from cardiac fibroblast cell lysates. An isotype control antibody was used as a control. Immunoprecipitates were immunoblotted with an Eed antibody. The first lane contains cell extract (1/10th immunoprecipitation input). Representative immunoblots are shown from three independent experiments. D, cardiac fibroblasts were transfected with either a nontargeting control siRNA or a PurB targeting siRNA. After 3 days, endogenous Cbx1 was immunoprecipitated from cell lysates. An isotype control antibody was used as a control. Immunoprecipitates were immunoblotted with a Cbx1 and a Sp3 antibody. The first lane contains cell extract (1/10th immunoprecipitation input). Representative immunoblots are shown from three independent experiments. H3K27me3, trimethylated histone-H3 lysine-27.