A PRMT5-ZNF326 axis mediates innate immune activation upon replication stress

Abstract

DNA replication stress (RS) is a widespread phenomenon in carcinogenesis, causing genomic instability and extensive chromatin alterations. DNA damage leads to activation of innate immune signaling, but little is known about transcriptional regulators mediating such signaling upon RS. Using a chemical screen, we identified protein arginine methyltransferase 5 (PRMT5) as a key mediator of RS-dependent induction of interferon-stimulated genes (ISGs). This response is also associated with reactivation of endogenous retroviruses (ERVs). Using quantitative mass spectrometry, we identify proteins with PRMT5-dependent symmetric dimethylarginine (SDMA) modification induced upon RS. Among these, we show that PRMT5 targets and modulates the activity of ZNF326, a zinc finger protein essential for ISG response. Our data demonstrate a role for PRMT5-mediated SDMA in the context of RS-induced transcriptional induction, affecting physiological homeostasis and cancer therapy.

Fig. 1.. HU induces global changes to the transcriptome.

(A) Volcano plot displaying differentially expressed genes and ERVs in MCF10A cells upon HU treatment (1 mM) for 24 hours. Up-regulated genes are defined by requiring false discovery rate (FDR) < 0.05 and log₂FoldChange > 1 (up-regulated following HU treatment) or log₂FoldChange < −1 (down-regulated following HU treatment). (B) Left: Volcano plot displaying gene set enrichment analysis (GSEA) of 50 hallmark gene sets for comparison of MCF10A ± HU (1 mM) for 24 hours. Right: Volcano plot displaying IFN response gene expression distribution in HU (1 mM) versus no treatment control for 24 hours. (C) Left: Enrichment plot from GSEA for comparison of MCF10A ± HU (1 mM) 24 hours. Right: Heatmap of normalized read counts from RNA-seq of MCF10A ± HU (1 mM) for 24 hours. The heatmap was generated using the ClustVis webtool (71). (D) Volcano plot displaying ERVs expression distribution in HU (1 mM) versus no treatment control for 24 hours. (E) Enrichment plot of ERV signature from GSEA for comparison of MCF10A treated with HU (1 mM) versus no treatment control for 24 hours.

Fig. 2.. A chemical genetic screen identifies key epigenetic modulators of HU-induced transcriptional changes.

(A) qPCR of CXCL10 expression from MCF10A untreated or treated with HU (1 mM). Statistical significance was analyzed by unpaired Student’s t test of values obtained from two independent experiments. (B) Schematic workflow for generating HiBiT tag knock-in to CXCL10 locus in MCF10A and HiBiT-CXCL10 detection by luminescence. The illustration was created with BioRender. (C) Luminescence readout from HiBiT-CXCL10 knock-in cell clone ± HU for 24 hours. (D) Workflow for epigenetic drug screen to determine modulators of HU-induced CXCL10 expression. (E) Scatterplot for epigenetic drugs down-/up-regulated HU-induced CXCL10 expression. (F and G) Validated hits from (E) with cotreated HU and epigenetic drugs for 24 hours in MCF10A WT cell by qPCR (F) and Western blot (G).

Fig. 3.. RS induces a PRMT5-dependent ISG expression pattern.

(A) CXCL10-HiBiT expression from CXCL10-HiBiT knock-in MCF10A cells treated with DMSO, HU (1 mM), HU (1 mM) + GSK591 (1 μM), and HU (1 mM) + EPZ015666 (1 μM) for 48 hours. Statistical significance was analyzed by unpaired Student’s t test of values obtained from three independent experiments. (B) Western blot of MCF10A treated with DMSO, HU, GSK591, and HU + GSK591 and blot for IRF7, ISG15, STAT1, and GAPDH (glyceraldehyde-3-phosphate dehydrogenase). (C) qPCR of MCF10A cells treated with DMSO, AZD6738 (ATRi) (5 μM), and ATRi (5 μM) + GSK591 (1 μM) for 6 days. Statistical significance was analyzed by unpaired Student’s t test of values from three independent experiments (*P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001). Red asterisk, statistical difference between DMSO versus ATRi; black asterisk, statistical difference between ATRi versus ATRi + GSK591. (D) Western blot of MCF10A treated as in (C). (E) Multiplex ELISA for IFNs and cytokines from supernatants of MCF10A treated as in (B). Red asterisk, statistical difference between DMSO versus HU; black asterisk, statistical difference between HU versus HU + GSK591. P value was analyzed by unpaired Student’s t test of values from two independent experiments (*P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ns, not significant). (F) Volcano plot displaying GSEA of 50 hallmark gene sets comparing MCF10A treated with HU versus HU + GSK591 for 24 hours. (G) Enrichment plot of GSEA IFNA hallmark comparing MCF10A treated with HU versus HU + GSK591 for 24 hours. (H) Heatmap of normalized read counts from RNA-seq of MCF10A treated with HU versus HU + GSK591 for 24 hours. The heatmap was generated using the ClustVis webtool (71). (I) Volcano plot displaying ERV expression distribution in MCF10A treated with HU versus HU + GSK591 for 24 hours. (J) Enrichment plot of ERV signature from GSEA comparing MCF10A treated with HU versus HU + GSK591 for 24 hours.

Fig. 4.. HU leads to an increase in nuclear SDMA.

(A) MCF10A cells treated with DMSO, HU (1 mM), GSK591(1 μM), and HU (1 mM) + GSK591 (1 μM) for 24 hours and stained for PRMT5. Images were captured and analyzed using the Operetta system. (B) A violin plot displaying median value with interquartile range for the ratio for nuclear PRMT5 staining intensity over cytoplasmic PRMT5 staining intensity from at least 1000 cells from (A). P value was analyzed by unpaired Student’s t test of median values obtained from three independent experiments. (C) MCF10A cells treated with DMSO, HU, GSK591, and HU + GSK591 and stained for SDMA. Images were captured and analyzed using the Operetta system. Representative images in one of three independent experiment are shown. (D) A violin plot displaying median value with interquartile range for the nuclear SDMA staining intensity from at least 1000 cells from (C). P value was analyzed by unpaired Student’s t test of median values obtained from three independent experiments. (E) Western blot for MCF10A cells treated as in (C).

Fig. 5.. HU treatment leads to increased symmetric dimethylation of ZNF326.

(A) Schematic workflow to identify symmetric dimethylated (SDMA) proteins and peptides. The illustration was created with BioRender.com. (B) Volcano plots displaying SDMA-enriched proteins in MCF10A cells treated with 1 mM HU versus 1 μM SGC2096a, a negative control for PRMT5 inhibitor GSK591, for 6 days (analysis I); SDMA-enriched proteins in MCF10A cells treated with 1 mM HU versus 1 mM HU + 1 μM GSK591 for 6 days (analysis II); SDMA-enriched peptides in MCF10A cells treated with 1 mM HU versus 1 μM SGC2096a for 6 days (analysis III); and SDMA-enriched peptides in MCF10A cells treated with 1 mM HU versus 1 mM HU + 1 μM GSK591 for 6 days (analysis IV). All volcano plots were created with VolcaNoseR app (72). (C) Venn diagram showing common hits in analysis I, II, III, and IV. The Venn diagram was created with InteractiVenn web-tool (73). (D) Western blot of MCF10A cells treated with DMSO, HU (1 mM), GSK591(1 μM), or HU (1 mM) + GSK591 (1 μM) for 6 days and co-IP performed with SDMA antibody.

Fig. 6.. ZNF326 regulates expression of IRF7 and ISG15 induced by HU.

(A) Control siRNA–, ZNF326 siRNA–, or PRMT5 siRNA–transfected MCF10A cells ± HU (1 mM, 48 hours) subjected to proximity ligation assay (PLA) for PRMT5 and ZNF326. Images were captured and analyzed using the Operetta system. (B) A violin plot displaying the median number of spots with interquartile range in nuclear and cytoplasm compartment from at least 1000 cells from the PLA assay as described in (A). Statistical significance for difference in number of spots between treatments was analyzed by unpaired Student’s t test of values obtained from a representative of two independent experiments. (C) qPCR of MCF10A cells treated with control siRNA– or ZNF326 siRNA–transfected MCF10A cells ± HU (1 mM, 48 hours). Statistical significance was analyzed by unpaired Student’s t test of values obtained from three independent experiments. Comparisons were made between ZNF326 siRNAs + HU versus control siRNA + HU. *P ≤ 0.05; **P ≤ 0.01. (D) Western blot of MCF10A treated as in (C). Western blot for ZNF326 showed one band, which is different from ZNF326 Western blot with two bands in Fig. 5D. The difference may originate from the usage of IP lysis buffer in Fig. 5D and RIPA in (D). (E) qPCR of MCF10A cells with empty vector, WT ZNF326, or ZNF326 with mutation of arginine residues 173 and 175 to lysine constructs transfected with control siRNA or ZNF326 siRNAs ± HU (1 mM). Statistical significance was analyzed by unpaired Student’s t test of values obtained from three independent experiments. Comparisons were made between WT ZNF326 transfected with ZNF326 siRNA + HU versus ZNF326 with mutation of arginine residues 173 and 175 to lysine transfected with ZNF326 siRNA + HU. *P ≤ 0.05; **P ≤ 0.01.