Background
Serine/arginine-rich splicing factor 3 (SRSF3), previously named as SRp20 and SFRS3, is the smallest member of serine/arginine-rich (SR) protein family, well known for its regulatory roles in RNA metabolism and functions, such as pre-mRNA splicing [
1‐
4], mRNA 3′ end processing [
5,
6], mRNA export from nucleus [
7‐
9] and cap-independent translation [
10,
11]. SRSF3 was also implicated in the regulation of chromatin structure and function because of its association with interphase chromatin but not with hyperphosphorylated mitotic chromosomes [
12].
Physiologically, SRSF3 is essential for embryo development since
Srsf3-null mouse embryos failed to form blastocysts and died at the morula stage [
13]. Mice with hepatocyte-specific knockout of
Srsf3 exhibited altered hepatic architecture, prolonged expression of fetal liver markers, impaired glucose homeostasis and reduced cholesterol synthesis, suggesting that
Srsf3 is indispensable for hepatocyte maturation and metabolic function in mice [
14].
Pathologically, there is increasing evidence indicating that SRSF3 plays an important role in tumorigenesis. In a mouse model of mammary tumorigenesis, it was observed that SRSF3 was remarkably increased during the development of mammary cancer [
15]. In human ovarian tumors, we found that SRSF3 was overexpressed in invasive ovarian cancer at all stages and its overexpression was critical for tumor cell growth and maintenance of transformation properties [
16,
17]. Knockdown of SRSF3 expression causes growth inhibition or apoptosis of ovarian cancer cells, depending on the extent of SRSF3 knockdown [
16]. SRSF3 was also found upregulated in a variety of other tumors, such as cervical cancer and rhabdomyosarcoma [
18]. It was showed that ectopically expressed SRSF3 promoted cell growth and transformation of human and mouse fibroblasts [
18]. In addition, knockdown of SRSF3 resulted in G1 arrest and downregulation of several G1/S transition-related genes in colon cancer cells [
19] and led to p53-dependent cellular senescence in fibroblasts [
20]. Besides the tumor promoting role, a recent study found that SRSF3 might function as a suppressor of hepatic carcinogenesis, because mice with hepatocyte-specific knockout of
Srsf3 invariably developed hepatocellular carcinoma at late ages [
21].
Our previous studies mentioned above raise questions why SRSF3 is required for ovarian cancer cell growth and how it contributes to the neoplastic transformation. In the present study, we show that knockdown of SRSF3 suppresses expression of breast cancer 1, early onset (BRCA1), BRCA1 interacting protein C-terminal helicase 1 (BRIP1), and RAD51 recombinase (RAD51). These genes all play important roles in the homologous recombination (HR)-mediated DNA damage repair pathway [
22,
23]. Correspondingly, we observed impaired HR-mediated DNA damage repair (HRR) activity and accumulation of DNA double-strand breaks (DSBs) after SRSF3 knockdown. We also provide evidence suggesting that SRSF3 possibly regulates the expression of above genes through an epigenetic pathway.
Discussion
In this report, we present data showing that knockdown of SRSF3 results in downregulation of BRCA1, BRIP1 and RAD51 expression and causes impaired HRR activity. These results suggest a novel role for SRSF3 in the regulation of HRR pathway.
HRR is a major mechanism to repair DSBs, which are the most deleterious form of DNA damage and can be generated by exogenous insults as well as endogenous factors [
37]. In dividing cells like cancer cells, DSBs are mainly caused by endogenous factors (endogenous DSBs, EDSBs), such as reactive oxygen species (ROS) and replication stress [
37], and can be induced by activated oncogenes [
38‐
41]. It was estimated that EDSBs were produced at the rate of ~50 per cell per cell cycle in the normal human cells [
42]. In cancer cells, this rate could be higher because of the effects of increased oncogene activity. DSBs are repaired primarily by two mechanisms: non-homologous end-joining (NHEJ) and HRR [
43,
23]. NHEJ repairs DSBs by promoting direct ligation of DNA ends, which frequently introduces insertions, deletions, substitutions and even chromosome rearrangements. In contrast, HRR repairs DSBs faithfully by using homologous sister chromatids as the template to guide the repairing process and thus playing a pivotal role in the maintenance of genomic stability [
43,
23]. HRR involves following steps: DSB recognition, damage signal transduction and break repair by HR [
23]. The six downregulated genes shown in Fig.
1c all have a role or roles in this repair pathway [
22,
23,
44]. For example, BRCA1 helps to direct the cell to choose HRR over NHEJ to repair DSBs during S and G2 phase [
44]. BRCA1 is also required for the recruitment of RAD51to the damage sites [
45], which is necessary for homology search and subsequent strand exchange with intact sister chromatid duplex DNA [
23].
If DSBs are left unrepaired or aberrantly repaired, the outcome would be cell death or genomic instability. Although genomic instability is a characteristic of most cancers and is believed to facilitate the development of permanent oncogenic changes in the genome [
46], there is no evidence suggesting that cancer cells could tolerate continuous DNA damage generation after generation. On the contrary, a relatively stable genome is essential for any cell, normal or tumor, to grow and survive [
47], and it is cancer cell’s reliance on a stable genome that makes DNA-damaging agents to be effective in cancer treatment.
Given the more frequent occurrence of spontaneous DSBs in cancer cells and the importance of a relatively stable genome for cell growth and survival, it is logical that cancer cells need upregulated HRR activity to keep their genomes from continuous alterations. Otherwise, accumulated DSBs or genomic alterations would eventually lead to cell death. The new role of SRSF3 in the regulation of HRR pathway provides a mechanism for cancer cells to meet this need. Therefore, it is no wonder that almost all invasive ovarian tumors that we examined overexpressed SRSF3 and knockdown of SRSF3 induced growth inhibition and cell death [
16]. Analysis of the serous ovarian cancer microarray dataset from The Cancer Genome Atlas project shows that SRSF3, BRCA1, RAD51, XRCC2 and BLM are upregulated in tumors compared to normal ovaries, as shown in Additional file
3: Figure S7, supporting the notion that tumor cells need enhanced HRR activity.
The new role of SRSF3 discovered in this study also suggests a new paradigm to understand the tumorigenic process. It is widely accepted that activated oncogenes are a driving force of tumorigenesis [
48,
49]. However, they alone cannot cause cancer. Instead, activated oncogenes induce senescence or cell death in normal and partially transformed cells due to their induction of DNA damage and DNA damage response (DDR) [
49,
40]. According to current tumorigenic model, after oncogene activation, further genetic or epigenetic changes in tumor suppressor genes are needed to overcome replicative stress and make tumorigenesis proceed (Fig.
8, left panel) [
48,
49]. Our observation suggests that there exist another mechanism to promote tumorigenesis. That is, during neoplastic transformation, which could be initiated by oncogene activation, SRSF3 is upregulated by presently unknown factor(s) and confers cells enhanced capability to carry out HRR and thus allows cells to bypass replicative stress and complete transformation process (Fig.
8, right panel). This new mechanism may explain not only the development of tumors that lack mutations or alterations in tumor suppressors involved in DNA damage repair and response but also the overexpression of RAD51 found in a wide variety of human tumors, including BRCA1-deficient ones [
50,
51]. Overexpression of RAD51 can rescue the defects caused by depletion of BRCA1 and thus may contribute to the genesis of BRCA1-deficient tumors [
51].
Finally, the results shown in Fig.
7 provide a clue to understand the molecular mechanisms behind the new role of SRSF3. Based on those results, we hypothesize that SRSF3 regulates the expression of HRR-related genes indirectly through an epigenetic pathway. That is, SRSF3 controls alternative splicing of KMT2C, whose splice variants determine the methylation status of H3K4, by which the transcriptional activities of HRR-related genes are set. To test the hypothesis, more work will be needed to establish causal relationships between the changed alternative splicing of KMT2C and reduced methylated H3K4 and between reduced H3K4me3 and suppressed expression of HRR-related genes.
Methods
Cell cultures
Ovarian cancer cell line A2780 sublines, A2780/SRSF3si1, A2780/SRSF3si2 and A2780/LUCsi, were established in our previous study [
16]. These sublines were grown in DMEM supplemented with 10 % FBS and 2 mM L-glutamine at 37 °C, 5 % CO
2. 293 T cells were purchased from the American Type Culture Collection (ATCC) and grown in the same media as A2780 sublines.
Microarray analysis
Total RNAs were extracted from A2780/SRSF3si2 cells grown in the presence or absence of Doxy (0.1μg/ml) for 3 days using TRIzol reagent (Life Technologies, Grand Island, NY) and treated with TURBO DNA-free kit (Life Technologies). The prepared total RNA samples were submitted to Asuragen (Austin, TX) for expression profiling by Affymetrix Human Exon 1.0 ST Array (Affymetrix, Santa Clara, CA). The microarray data were analyzed using Partek Genomics Suite Version 6.6 (Partek, St. Louis, MO) to determine the differentially expressed or spliced genes. Gene ontology analysis was performed also using Partek Genomics Suite Version 6.6.
RT-PCR and qPCR
Total RNAs were extracted with TRIzol reagent from cultured cells and treated with TURBO DNA-free kit. cDNAs were synthesized from 2 μg of total RNAs with High Capacity cDNA Reverse Transcription Kit (Life Technologies). Non-quantitative RT-PCR reactions were set up with Phusion Green Hot Start II High-Fidelity DNA Polymerase (Thermo Fisher Scientific, Waltham, MA). qPCRs were set up with Fast SYBR Green Master Mix (Life Technologies) and run in StepOne Plus Real-Time PCR System (Life Technologies). The primer pairs for RT-PCR and qPCR were the same for each gene and they are BRCA1 prime pair 5′-ACTCTGAGGACAAAGCAGCG-3′ and 5′- CATCCCTGGTTCCTTGAGGG-3′, BRIP1 primer pair 5′- CGCTTTAGGAATAACCCAAGT-3′ and 5′- CTCATTGTCCTGTATATTGGTT-3′, RAD51 primer pair 5′- TTTGGCCCACAACCCATT TC-3′ and 5′- TTAGCTCCTTCTTTGGCGCA-3′, SRSF3 primer pair 5'-AATTGGAACGGGCTTTTGGC-3' and 5'-CCATCTAGCTCTCGGACTGC-3', and GAPDH primer pair 5′-GGGGCTGGCATTGCCCTCAA-3′ and 5′-GGCTGGTGGTCCAGGGGTCT-3′. The expression level of each gene was determined by the comparative CT (ΔΔCT) method [
52] with GAPDH as the endogenous control and the subline A2780/LUCsi cells grown in the absence of Doxy as the reference. The primer pair for amplification of KMT2C cDNA between exon 44 and exon 46 was 5′-AGCACTGACACGTTTACCCA-3′ and 5′- AAGCCGGAGTGTTAGTGAGC-3′.
Western blotting
Whole cell lysates were prepared with 1x sample buffer (50 mM Tris pH 6.8, 2 % SDS, 10 % glycerol, 5 % β-mecaptoethanol and 0.002 % bromphenol blue) and sonicated with Sonifier Cell Disrupters (Branson Ultrasonics, Buffalo Grove, IL). Western blotting was performed as described previously [
53]. The antibodies for BRCA1, RAD51, SRSF3 and γ-H2AX were purchased from Santa Cruz Biotechnology (Dallas, TX, cat# sc-642, sc-8349, sc-13510 and sc-101696, respectively) and the antibodies for BRIP1 and UPF1 were from Cell Signaling Technology (Danvers, MA; cat# 4578S, 12040S, respectively). Quantitation of western blotting results was performed with Volume Tools program contained in Image Lab software (Bio-Rad Laboratories, Hercules, CA).
Apoptosis assay
Cells were fixed in 4 % paraformaldehyde for 10 min and then stained in a solution of Hoechst 33342 (Life Technologies) for 15 min. Apoptotic cells and non-apoptotic cells were counted under fluorescent microscope manually with computer assistance.
ChIP
Chromatin DNAs were isolated from A2780/SRSF3si2 cells treated with or without Doxy for 3 days and immunoprecipitated with ChIP-IT Express Enzymatic kit (Active Motif, Carlsbad, CA) and RNA polymerase II antibody (mAb) (Active Motif, Cat # 39097) or Negative control mouse IgG (Santa Cruz Biotechnology, cat# sc-2762) by following the manufacturer’s instruction. The primer pairs for non-quantitative PCR and qPCR were the same for each gene and they are following: BRCA1 primer pair, 5′-GGACGTTGTCATTAGTTCTTTGGT-3′ and 5′-TCTTCAACGCGAAGAGCAGA-3′; BRIP1 primer pair, 5′-GGGCTCCGCTTTATTTGCTC-3′ and 5′-CAGTTGAGATCCCCGAGACC-3′; RAD51 primer pair, 5′-GCTGGGGCGAAAACACAAG-3′ and 5′-GACTTCTCGCTCGAACCCAT-3′; and GAPDH primer pair, 5′- TACTAGCGGTTTTACGGGCG-3′ and 5′- AGGCTGCGGGCTCAATTTAT-3′. Non-quantitative PCRs and qPCRs were set up as described in 2.2. The immunoprecipitated DNAs were quantitated by standard curve method. The standard curve was generated with input chromatin DNA samples at concentrations of 50 ng, 5 ng, 0.5 ng and 0.05 ng per ul.
Immunofluorescent staining
A2780/SRSF3si2 cells were grown on poly-L-lysine-coated glass coverslip in the presence or absence of Doxy for 3 days before subjected for staining. The cells were fixed in ice-cold methanol for 10 min followed by air-dry. Afterwards, the cells were blocked in 5 % normal donkey serum (Jackson ImmunoResearch, West Grove, PA) for 1 h before they were incubated with γ-H2AX antibody (Cell signaling Technology, Cat # 9718S, 1:400 dilution) for 1 h and then with Dylight 488-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch, Cat # 711-485-152, 1:200 dilution) for 45 min. The cells were rinsed in 1xPBS for three times after each incubation step. Finally, the coverslips were mounted on glass slides with VECTASHIELD Mounting Medium containing 4′, 6-Diamidino-2-phenylindole dihydrochloride (DAPI) (Vector Laboratories, Burlingame, CA).
HR assay
The direct repeat green fluorescent protein (DR-GFP) reporter was used to measure HR activity in 293 T cells with or without SRSF3 knockdown. Briefly, 293 T cells grown in 12-well plate were infected at multiplicity of infection 5 with lentiviruses expressing SRSF3si1, SRSF3si2 or LUCsi for 12 h. Two days after infection, these cells were co-transfected with plasmids pDRGFP, pCBASceI (Addgene, Cambridge, MA) and pmCherry-N1 (Clontech Laboratories, Mountain View, CA) by calcium phosphate precipitation method [
54]. The transfected cells were subjected to flow cytometric analysis for GFP-positive and mCherry-positive cells two days after transfection. The percentages of GFP-positive cells were normalized to the percentages of mCherry-positive cells before comparison.
Statistical analysis
Unless otherwise stated, Student’s t-test was used in comparisons between samples. All tests were two-sided and p-values < 0.05 were considered significant.
Accession numbers
The microarray data reported in this paper were deposited in Gene Expression Omnibus (GEO) database. The accession number is GSE71745.
Competing interests
The authors declare that they have no competing interest.
Authors’ contributions
XH conceived, designed and performed the experiments, analyzed the data and wrote the manuscript. PZ performed the experiments and analyzed the data. Both authors read and approved the final manuscript.