Results

Chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-seq) analysis reveals a role of SMYD3 in DNA repair.

In previous work (15), we have identified that cyclin A1 is a downstream target of SMYD3 and H2A.Z.1. In order to characterize at a genomic scale the unexplored aspects of the relationship among SMYD3, H2A.Z.1, and their cellular targets, we performed SMYD3 and H2A.Z.1K101me2 ChIP-seq experiments. We performed three times of paired-end sequencing for each sample to get a total of 30 million reads. The output results were pooled together for further analysis. The signals of SMYD3 and H2A.Z.1K101me2 were normalized to input signals (Fig. 1a). We compared the distribution map of about 20 thousand genes with reported H2A.Z and H2A.Zac (acetyl K4 + K7 + K11) ChIP-seq database (GEO accession number GSM1059388) (44).

Results showed that SMYD3, H2A.Z, and H2A.Z PTM proteins did not share similar tendency of distribution, in which SMYD3 was recruited more at the promoter region (-5000 bp to -1000 bp) than at coding sequences, whereas H2A.Z and its PTM proteins were more centered around the transcription start site (TSS). Compared to H2A.Z and H2A.Zac, which showed equal distribution around TSS, H2A.ZK101me2 signal levels were higher at the promoter region (-1250 bp to TSS) than at the downstream of the TSS. Next, we examined unique genes identified from the ChIP-seq data. Using cutoff parameters of P < 0.001 and distance to TSS (-5000 bp to +200 bp), SMYD3 and H2A.Z.1K101me2 were associated with 2406 and 3029 genes, respectively (Fig. 1a and 1b). Gene ontology analysis showed that SMYD3 regulated genes involved in regulation of transcription factor activity, DNA binding, enzyme activity and

nucleocytoplasmic transport. H2A.ZK101me2 regulated genes that were involved in protein transport and cell cycle. Furthermore, we investigated the relationship between the enrichment of SMYD3 and H2A.Z.1K101me2 on chromatin and gene expression levels. We compared their occupancy and the gene expression data from our previous report (15). We then separated genes into four classes based on their increase or decrease in expression, namely they were downregulated or upregulated in shLuc versus shSMYD3 cells (Table 1 and 2), and in H2A.Z.1^WT versus H2A.Z.1^K101Q in endogenous H2A.Z knockdown MCF7 cells (Table 3 and 4). Since SMYD3 is a transcriptional co-activator (45), we focused on genes that were downregulated in the shSMYD3 cell (Fig. 2a and Table 1). Gene ontology (GO) analysis showed that 5 genes were involved in DNA repair, including EXO1, FANCI, MDC1, POLQ, and RAD54B, and 4 genes were related to M phase, including ANLN, EXO1, OIP5, and RAD54B (Table 5). To verify the accuracy of the microarray data, qRT-PCR was performed for 5 DNA repair genes, and all genes exhibited similar fold as originally identified in the microarray analysis (Fig.

2b). We then went back to review the microarray data and found that GO analysis suggested a role of SMYD3 in response to DNA damage stimulus (Fig. 3a). Therefore, we aimed to study whether SMYD3 is involved in DNA repair.

Microarray data analysis identifies SMYD3-regulated expression of DNA repair machinery.

Increased expression of SMYD3 can promote cancer proliferation (7) and metastasis (16). To explore additional and novel roles of SMYD3 in biological processes, we analyzed our previously conducted whole-genome microarray data of RNAs isolated from shLuc vs. shSMYD3 MCF7 cells (GEO accession number GSE58048), in which a lentivirus shRNA infection system was used for stable

knockdown of SMYD3 (15). 449 genes were downregulated upon SMYD3 knockdown.

The gene ontology (GO) analysis indicated that these genes were mainly involved in cell cycle, DNA metabolic process, response to DNA damage stimulus, cell proliferation and macromolecular complex subunit organization (Fig. 3a). Previous reports provided evidence that SMYD3-dependent histone methylations are essential for cell cycle and cell proliferation. Intriguingly, SMYD3 is associated with DNA damage response (DDR) in the top three categories of GO analysis. Since SMYD3 has not been linked to DDR or DNA repair, the mechanism was further analyzed.

We first investigated whether SMYD3 knockdown cells are more vulnerable to DNA damage stress such as IR. To examine the repair rate following IR, the formation of γH2A.X foci were used as a marker for DNA damage. Following exposure to 1.67 Gy of IR, SMYD3 knockdown cells significantly delayed the removal of γH2A.X foci at 48 and 72 hr compared to the shLuc controls (Fig. 3b). Conversely, exogenous expression of SMYD3, but not the catalytic dead mutant SMYD3^Y239F proteins, in the SMYD3 knockdown cells significantly restored the defects at 24 and 72 hr compared to that in the shSMYD3 with the expression of the vector control (Fig. 3c). The clonogenic survival following exposure to IR was further examined, and knockdown of SMYD3 led to impeded formation of colonies (Fig. 4a). Similar to that in MCF7 cells, shSMYD3 MDA-MB-231 and AU565 cells were more vulnerable to IR stress (Fig. 4b and 4c).

We next analyzed the effect of IR on SMYD3’s cellular location. The majority of SMYD3 is located in the cytoplasm (10), and we wondered whether it would translocate into the nucleus upon DNA damage insults. Cells were exposed to increasing dosages of IR and assayed for the translocation of SMYD3 protein to the nucleus at 1 and 3 hours.

The distribution of SMYD3 did not show any noticeable difference after IR treatment (Appendix Fig. 2a). Furthermore, the gene and protein expression of SMYD3 were not augmented after IR treatment (Appendix Fig. 2b). These results suggest that DNA damage does not modulate SMYD3 expression and location.

We further examined whether SMYD3 affects genome integrity in cells. To determine if the loss of SMYD3 is associated with increased DNA damage, we performed a single-cell gel electrophoresis (comet) assay. The comet assay revealed increases of damage rate and tail moment in shSMYD3 compared to those in shLuc cells at 72 h (Fig. 5). We also investigated micronuclei formation, a well-established indicator of genome instability (46,47), which occurs through the aberrant segregation of chromosomes or acentric chromosomal fragments. Compared to the shLuc control, knockdown of SMYD3 caused an increase in IR-induced micronuclei at 72h (Appendix Fig. 3). Taken together, these data suggest a role of SMYD3 in DNA repair mechanism.

SMYD3 mediates the HR pathway.

Since mammalian DSB repair was achieved mainly by two mechanisms, HR, and NHEJ, we wondered whether SMYD3 is involved in these pathways. We used cells with well-characterized GFP-based chromosomal reporters to detect the efficiency of HR.

The reporter contains an I-SceI recognition sequence, which would be cleaved upon I-SceI expression to generate a DSB. DSB repair by HR using the direct repeat within the reporter cassette as a template results in an intact GFP gene. The repair efficiency was then quantified by flow cytometry. For the plasmid-based end-joining assay, a linearized plasmid harboring a luciferase reporter gene was used. Repair efficiency was measured by the luciferase activities of linearized reporter constructs compared with that of the intact plasmid. Results demonstrated that SMYD3 knockdown significantly

hampered HR repair by 55-70% compared to the control cells (Fig. 6a). In contrast, SMYD3 knockdown did not change the NHEJ activity compared to the control cells (Fig. 6b). As the controls for the HR and NHEJ assays, knockdown of EXO1 reduced the HR activity and knockdown of Ku70 impaired the NHEJ activity by 56-73% and 47-50%, respectively (Fig. 6a and 6b). We also examined whether SMYD3 was required by MMEJ using a plasmid-based MMEJ assay and found that SMYD3 did not display any effect on the efficiency of MMEJ, while knockdown of the control, POLQ, reduced MMEJ activity by 49-58% (Fig. 6c). The knockdown effectiveness of each cell lines used was confirmed by qRT-PCR (Fig. 7). Moreover, the exogenous expression of SMYD3, but not the SMYD3^Y239F proteins, restored the HR activity of the shSMYD3 cells (Fig. 8). These results identify a role of SMYD3 in HR repair.

SMYD3 knockdown downregulates HR gene expressions.

To understand the exact role of SMYD3 in HR repair, we analyzed microarray-identified genes that were related to DDR and found that 32 of them are involved in DNA repair, including genes that are in response to oxidative stress, base excision repair, cell cycle checkpoint, interstrand cross-links, HR, NHEJ and MMEJ (Table 6). Among these 32 genes, 13 genes are implicated in the HR pathway (Table 7).

These genes range from the early step of DNA damage mediators, kinase transducer to downstream effectors that execute error-free repair process (48). We performed qRT-PCR analysis to confirm their expressions and found that all genes exhibited similar fold differences in mRNA expression as initially identified in the microarray analysis (Fig. 9).

Among these genes, MDC1 plays the earliest role in HR repair (49) and participates in the initial recruitment of BRCA1 (50-54) to promote DNA end resection

for HR (55). Moreover, EXO1 is the major exonuclease for efficient end resection (26), and RAD54B is a DNA-dependent ATPase required for efficient chromatin remodeling during strand invasion (29,30). We checked the influence of SMYD3 depletion on MDC1, EXO1, and RAD54B. Consistently, the protein levels of MDC1 and EXO1 were significantly reduced after inhibition of SMYD3. Besides, a marginal reduction of the RAD54B protein was observed (Appendix Fig. 4).

To gain further insight into how SMYD3 regulates HR activity, we investigated the significance of SMYD3 on the formation of MDC1 foci. After IR treatment, MDC1 foci were diminished in SMYD3-depleted cells at 1 to 4 hrs. And these MDC1 foci were disappeared at 24 to 72 hrs, even in the shLuc cells (Fig. 10a). Moreover, after IR treatment, the formation of BRCA1 foci in SMYD3 knockdown cells was impaired as well (Fig. 10b). In contrast, lack of SMYD3 did not affect the assembly of 53BP1 foci (Appendix Fig. 5). These data indicated that SMYD3 deficiency weakens HR partly through downregulation of MDC1, and thereby compromising the recruitment of BRCA1 at DSBs.

SMYD3 knockdown impaired the formation of BRCA1 foci. We further investigate whether it was resulted from the change of mRNA and protein expression of BRCA1 per se or affected by the reduction of MDC1. We observed that the mRNA and protein

level of BRCA1 diminished in SMYD3-depleted cells (Appendix Fig. 6A and 6B).

Intriguingly, our microarray data showed that the expression of BARD1 (BRCA1-Associated RING Domain-1) decreased in SMYD3-depleted cells (Appendix Fig. 6A). BARD1 is vital in the rapid relocation of BRCA1 to DNA damage sites, and the interaction with BARD1 increases the E3 ligase activity and the stability of BRCA1 (56). We observed that MDC1-depleted cells showed decreased BRCA1 foci/protein

expression as well (Appendix Fig. 6C and 6D). Therefore, the reduction of BRCA1 foci/protein in SMYD3-depleted cells might due to the synergistic effects caused by BARD1 and MDC1.

SMYD3 controls the expression of HR genes through methylating histone H3K4.

SMYD3 was initially reported to methylate histone H3K4 to modulate the accessibility of chromatin architecture and to form a complex with RNA polymerase II (RNAPII) and RNA helicase HELZ to drive its target genes (7). To explore the epigenetic regulation of SMYD3 on these HR genes, we examined the recruitments of SMYD3, H3K4me3, and RNAPII pSer5, which is a required RNAPII phosphorylation for the transcriptional initiation (57), to the MDC1, EXO1, and RAD54B promoter regions. The putative TATA box (region TA) was retrieved from GPMiner (58). ChIP experiments showed a direct binding of SMYD3 to the MDC1 promoter (region S3) at

~500 bp upstream of the transcription start site (TSS) (Fig. 11a). SMYD3 preferred to enrich at region S3 (blue bar) than region TA (red bar) (Fig. 11b). In contrast, RNAPII pSer5 (Fig. 11c) and histone H3K4me3 (Fig. 11d) exhibited greater binding at region TA than region S3. Furthermore, knockdown of SMYD3 led to significant decreases of SMYD3, H3K4me3, H3 and RNAPII pSer5 at various degrees at both regions S3 and TA (Fig. 11b-e). Also, the enrichment of H3K4me3-modified histones (H3K4me3/H3) declined significantly in SMYD3 knockdown cells (Fig. 11f). A similar tendency was observed in the promoter regions of EXO1 and RAD54B (Fig. 12). These results suggest that SMYD3 may trigger HR gene expression by directly binding to the promoters to create an active histone mark for transcription.