Role of the SAF-A/HNRNPU SAP domain in X chromosome inactivation, nuclear dynamics, transcription, splicing, and cell proliferation

Abstract

SAF-A/HNRNPU is conserved throughout vertebrates and has emerged as an important factor regulating a multitude of nuclear functions, including lncRNA localization, gene expression, and splicing. Here we show the SAF-A protein is highly dynamic and interacts with nascent transcripts as part of this dynamic movement. This finding revises current models of SAF-A: rather than being part of a static nuclear scaffold/matrix structure that acts as a stable tether between RNA and chromatin, SAF-A executes nuclear functions as a dynamic protein, suggesting contacts between SAF-A, RNA, and chromatin are more high turnover interactions than previously appreciated. SAF-A has several functional domains, including an N-terminal SAP domain that binds directly to DNA and RNA. Phosphorylation of SAP domain serines S14 and S26 is important for SAF-A localization and function during mitosis, however, whether these serines are involved in interphase functions of SAF-A is not known. In this study we tested for the role of the SAP domain, and SAP domain serines S14 and S26 in X chromosome inactivation, protein dynamics, gene expression, splicing, and cell proliferation. Here we show that the SAP domain, and SAP domain serines S14 and S26 are required to maintain XIST RNA localization and XIST-dependent histone modifications on the inactive X chromosome, to execute normal protein dynamics, and to maintain normal cell proliferation. In addition, we present evidence that a Xi localization signal resides in the SAP domain, enabling SAF-A to engage with the Xi compartment in a manner distinct from other nuclear territories. We found that the SAP domain is not required to maintain gene expression and plays only a minor role in mRNA splicing. We propose a model whereby dynamic phosphorylation of SAF-A serines S14 and S26 mediates rapid turnover of SAF-A interactions with nuclear structures during interphase. Our data suggest that different nuclear compartments may have distinct requirements for the SAF-A SAP domain to execute nuclear functions, a level of control that was not previously known.

Fig 1. The SAP domain and phosphorylatable SAP domain serines S14 and 26 are required for cell proliferation.

A. Schematic depicting full-length SAF-A (isoform a, 825 amino acids) with domains drawn to scale. B. Design of the SAF-A-AID-mCherry degron to replace the endogenous SAF-A genes in RPE-1 cells. C. SAF-A transgenes with alanine (phosphomutant) or aspartic acid (phosphomimetic) mutations at positions S14 and S26, or expressing a SAP domain deletion. D. Western blot analysis of SAF-A cell lines to monitor expression of SAF-A-AID-mCherry (α-RFP panel) and SAF-A-GFP (α-GFP panel) in cell lines. Tubulin was used as a loading control. Lane 1: untagged RPE-1 parental cell line. Lane 2: SAF-A degron cell line, either without (lane 2), or with (lane 3) doxycycline and IAA. Lanes 4-7: cells expressing SAF-A-GFP transgenes, treated with doxycycline and IAA. Molecular weight markers are indicated to the right of the panel. E. Immunofluorescence of cells expressing tagged SAF-A, as indicated by the panel insets. Treatment with doxycycline and IAA as indicated led to complete depletion of SAF-A-AID-mCherry in the vast majority of cells. Integrated SAF-A-GFP transgenes showed uniform expression in the cell population after induction. Bar, 40 mm. F-G. PCNA and Ki-67 immunofluorescence to monitor cell proliferation. Cells were treated with doxycycline and IAA to induce endogenous SAF-A depletion and SAF-A-GFP transgene expression, and were compared to control cell populations. Cells were treated for 1, 3, 6, or 10 days as indicated. SAF-A expression is the same as the panel insets shown in E. All images in panels E-G are rendered as a maximum projection of a 3D stack. Bar, 10 mm. H. Quantitation of cell populations with PCNA- and Ki-67-positive cells expressed as percent of the total population. 300 cells were scored for PCNA and Ki-67 immunofluorescence in two biological replicates; the average and standard deviation is shown.

Fig 2. The SAP domain and SAP domain serines S14 and 26 impact XIST RNA localization and PRC1/PRC2-dependent histone modifications on the Xi.

A. XIST RNA FISH and DAPI staining in RPE-1 cells, and cells expressing SAF-A-GFP transgenes, 24 hours after treatment with doxycycline and IAA. Images are rendered as a maximum projection of a 3D stack. Bar, 10 μm. B. Quantitative measurement of XIST RNA foci per cell. The term foci refers to the number of resolvable objects rather than individual molecules. Image stacks were acquired for at least 100 nuclei per genotype for two biological replicates and analyzed in Fiji software to count XIST RNA foci. Measurements are depicted as violin superplots. The average of each replicate is depicted as an open circle, whereas the average of both replicates is depicted as a horizontal line. The standard deviation of the two averages is shown. Cell n for quantitation of XIST RNA particles is: RPE-1 n = 193 and n = 211, SAF-A^wt-GFP n = 137 and n = 236, SAF-A depleted 1 n = 122 and n = 220, SAF-A^AA-GFP n =170 and n = 206, SAF-A^DD-GFP n = 133 and n = 198, SAF-A^ΔSAP-GFP n = 160 and n = 158. Statistical comparison of number of XIST RNA particles was performed using a one-way ANOVA followed by Tukey’s tests with a Bonferroni correction. The p-value comparisons between SAF-A^wt and all other genotypes are as follows: SAF-A^wt versus SAF-A^wt-GFP, ns. SAF-A^wt versus SAF-A depleted, p = 0.001. SAF-A^wt versus SAF-A^AA-GFP, ns. SAF-A^wt versus SAF-A^DD-GFP, p = 0.004. SAF-A^wt versus SAF-A^ΔSAP-GFP, p = 0.001. C. Immunofluorescence of histone H3K27me3 and H2AK119ub and DAPI staining, 24 hours after treatment with doxycycline and IAA. SAF-A-GFP allele expression is indicated to the left of the panel. Images are rendered as a maximum projection of a 3D stack. Bar, 10 μm. D. Quantitation of cell populations with H3K27me3 and H2AK119ub enrichment on the Xi. 100 cells were scored for H3K27me3 and H2AK119ub enrichment on the Xi in two biological replicates; the average and standard deviation is shown. Statistical comparison was performed using a t-test to determine p-values for H2AK119ub enrichment. SAF-A^wt versus SAF-A^wt-GFP, p = 0.86, ns. All other comparisons were statistically significant: SAF-A^wt versus SAF-A depleted, p = 0.0065. SAF-A^wt versus SAF-A^AA-GFP, p = 0.0174. SAF-A^wt versus SAF-A^DD-GFP, p = 0.0084. SAF-A^wt versus SAF-A^ΔSAP-GFP, p = 0.0033. The same analysis was performed to determine p-values for H3K27me3 enrichment. SAF-A^wt versus SAF-A^wt-GFP, p = 0.8698, ns. All other comparisons were statistically significant: SAF-A^wt versus SAF-A depleted, p = 0.0019. SAF-A^wt versus SAF-A^AA-GFP, p = 0.0021. SAF-A^wt versus SAF-A^DD-GFP, p = 0.0069. SAF-A^wt versus SAF-A^ΔSAP-GFP, p = 0.0005.

Fig 3. The SAF-A SAP domain is important for nuclear dynamics and Xi localization.

A. Live cell analysis of SAF-A^wt-GFP and SAP domain mutations as indicated. Cells were analyzed 24 hours after doxycycline and IAA treatment. Each image represents a single 0.2 mm slice. Bar, 10 mm. The proportion of cells showing the depicted localization patterns for each allele are as follows: SAF-A^wt-GFP, 91% of cells; SAF-A^AA-GFP, 96% of cells; SAF-A^DD-GFP 100% of cells; SAF-A^ΔSAP-GFP 100% of cells (n ≥ 50). B. Images from a typical FRAP experiment using SAF-A^wt-GFP. C. FRAP recovery curves for SAF-A^wt-GFP, SAF-A^AA-GFP, SAF-A^DDGFP, and SAF-A^ΔSAP-GFP, with and without transcriptional inhibition (LDC). The standard deviation of recovery time is indicated in light colored error bars. D. Table of the t^1/2 recovery time and the immobile fraction for all SAF-A-GFP proteins, with and without transcriptional inhibition. E–F. Coimmunoprecipitation analysis of SAF-A-GFP alleles and bulk chromatin. E. Western blot analysis of SAF-A-GFP and histone H3 in soluble extracts (top panels) and α-GFP immunoprecipitations (bottom panels). F. Western Blot quantitation of histone H3 in eluates, expressed as a ratio to the amount of SAF-A-GFP in immunoprecipitations. SAF-A^wt-GFP was normalized to 1.0. The graph depicts the mean and SEM of n = 5 experiments. Statistical pairwise comparisons for each allele were performed using a Wilcoxon signed rank test to determine the p-value, which was ns.

Fig 4. SAF-A depletion does not reactivate gene expression on the inactive X chromosome.

A. Allele-specific gene expression was calculated from RNA-seq libraries using a combination of PAC and edgeR in RPE-1 cells. ‘a to ‘b ratios are plotted by gene for each chromosome. B. Average a:b ratio for all genes on the X chromosome plotted for RPE-1 and SAF-A depleted cells. C. Allele-specific ATAC-seq was calculated using PAC and is plotted by gene for each chromosome. D. comparison of allele-specific ATAC-seq reads by gene plotted for RPE-1 cells and SAF-A depleted cells. E. Allele-specific Cut-and-Run for H3K4Me3 was calculated using PAC. ‘a/b’ ratio is plotted by gene for each chromosome. F. Allele-specific Cut-and-Run is plotted for RPE-1 cells and SAF-A depleted cells for X chromosome genes.

Fig 5. SAF-A depletion leads to changes in gene expression that accumulate over multiple cell divisions.

A-C. SAF-A was depleted for 24, 48, or 72 hours through addition of auxin. Gene expression was evaluated by RNA-seq and EdgeR. MD plots depict differentially expressed genes at each time point. Genes with an FDR < 0.01 are colored. D-E Upset plots depicting the intersection of up and downregulated genes observed at 24, 48, and 72 hours after SAF-Adepletion. F. Gene Ontology analysis of the genes downregulated following acute SAF-A depletion.

Fig 6. SAF-A depletion leads to widespread changes in mRNA splicing.

A. mRNA splicing was evaluated in SAF-A depleted cells (24 hours) using rMATS. Density plot showing changes in exon inclusion. B. CDF plot illustrating the magnitude of changes in exon inclusion (of all exons significantly changed following SAF-A depletion) in all domain mutants. C. MISO plot showing an example altered exon. D. GO analysis of genes with altered splicing. E. CDF plot showing enrichment of all mRNAs, mRNAs with increased exon inclusion, and mRNAs with decreased exon inclusion in SAF-A RIP-seq libraries from interphase cells.