Phosphorylation of proliferating cell nuclear antigen promotes cancer progression by activating the ATM/Akt/GSK3β/Snail signaling pathway

Proliferating cell nuclear antigen (PCNA) and its posttranslational modifications regulate DNA metabolic reactions, including DNA replication and repair, at replication forks. PCNA phosphorylation at Tyr-211 (PCNA-Y211p) inhibits DNA mismatch repair and induces misincorporation during DNA synthesis. Here, we describe an unexpected role of PCNA-Y211p in cancer promotion and development. Cells expressing phosphorylation-mimicking PCNA, PCNA-Y211D, show elevated hallmarks specific to the epithelial-mesenchymal transition (EMT), including the up-regulation of the EMT-promoting factor Snail and the down-regulation of EMT-inhibitory factors E-cadherin and GSK3β. The PCNA-Y211D–expressing cells also exhibited active cell migration and underwent G2/M arrest. Interestingly, all of these EMT-associated activities required the activation of ATM and Akt kinases, as inactivating these protein kinases by gene knockdown or inhibitors blocked EMT-associated signaling and cell migration. We concluded that PCNA phosphorylation promotes cancer progression via the ATM/Akt/GSK3β/Snail signaling pathway. In conclusion, this study identifies a novel PCNA function and reveals the molecular basis of phosphorylated PCNA-mediated cancer development and progression.

Proliferating cell nuclear antigen (PCNA) 2 is a ring-shaped homotrimeric DNA clamp that is essential for many cellular processes, including DNA replication and DNA repair (1)(2)(3). The roles of PCNA in these processes are largely regulated through its posttranslational modifications. One of PCNA's modifications is its phosphorylation at tyrosine 211 (Tyr-211) by epidermal growth factor receptor (EGFR), whose overexpression and/or activation is associated with a variety of cancers and their progression (4 -8). Tyr-211-phosphorylated PCNA has been shown to increase the half-life of PCNA and facilitate cell proliferation (9).
PCNA is an essential component of the initiation and resynthesis reactions of DNA mismatch repair (10 -12), a critical cellular mechanism that ensures genome stability primarily by correcting mispairs generated during DNA replication (13). However, we recently demonstrated that Tyr-211-phosphorylated PCNA not only inhibits DNA mismatch repair, but also induces nucleotide misincorporation during DNA synthesis, leading to a hypermutator phenotype (14). This hypermutability may give tumor cells more flexibility to adapt to new environments during progression. EGFR overexpression and activation promote tumor cell motility and invasion (15,16), and both EGFR and PCNA have been used as diagnostic and prognostic markers for many tumors (17)(18)(19)(20)(21)(22), indicating that they both promote tumor progression. We therefore hypothesized that EGFR's tumor promotion/progression activity is executed, at least in part, through activities triggered by PCNA-Y211 phosphorylation.
To test this hypothesis, we used an inducible Tet-On system to express phosphorylation-mimicking or nonphosphorylation-mimicking PCNA in HeLa cells, and we analyzed the resulting cells for cancer progression. We demonstrate here that cells expressing phosphorylation-mimicking PCNA, a Tyr-211 to Asp-211 substitution (PCNA-Y211D or PCNA-YD), exhibit characteristics of the epithelial-mesenchymal transition (EMT), including the up-regulation of EMT-promoting factor Snail and the down-regulation of EMT-inhibitory factors E-cadherin and GSK3␤. Accordingly, these cells exhibit active cell migration and undergo G 2 /M arrest. Strikingly, all of these EMT-associated events require the activation of ATM and Akt. This study, therefore, has identified a novel function for Tyr-211-phosphorylated PCNA in promoting cancer metastasis by activating the ATM/Akt/GSK3␤/Snail signaling pathway.
To explore the mechanism by which phosphorylated PCNA promotes tumor progression, we conducted RNA-Seq in cells expressing PCNA-WT and PCNA-YD, then performed a genome-wide analysis of differentially expressed genes (DEGs). These analyses identified 4088 DEGs (Log 2 ͉-fold change͉ Ͼ0, Padj Ͻ0.05). The gene set enrichment analysis (GSEA), a highthroughput method that yields a functional profile of the gene or protein set involved in a biological process (30,31), revealed a gene set related to EMT hallmarks (NES ϭ 2.15, FDRq ϭ 0.002) as a top candidate (Fig. 1D). We, therefore, hypothesized that phosphorylated PCNA promotes EMT. We performed a transwell migration assay (32)(33)(34). The results indeed showed that cells expressing PCNA-YD exhibit a strong migrating activity, as significantly more HeLa-PCNA YD cells migrated from the upper chamber to the lower chamber than any of the control cell lines (HeLa, HeLa-PCNA WT , or PCNA YF ) (Fig. 1, E  and F).

Up-regulation of Snail is associated with phosphorylated PCNA-mediated EMT
Transcription factor Snail is an important driving factor of EMT, as it represses the expression of E-cadherin, a transmembrane glycoprotein that connects epithelial cells together at adherens junctions to prevent EMT (35)(36)(37)(38). To determine if Snail and E-cadherin are involved in PCNA-YD-promoted EMT, we performed Western blotting experiments and showed that the protein level of Snail in PCNA YD cells is at least 41% more than in nonphosphorylated control cells ( Fig. 2A). Also, little E-cadherin was detected in PCNA YD cells, but the protein was relatively abundant in all three control cell lines ( Fig. 2A). To determine whether the up-regulated Snail is responsible for the observed EMT in PCNA YD cells, we knocked down Snail expression using an shRNA. As expected, partial depletion of Snail was associated with increased expression of E-cadherin (Fig. 2B). The transwell migration assay revealed that Snail knockdown dramatically reduced the cell migration capability of PCNA YD cells (Fig. 2, C and D). We, therefore, conclude that phosphorylated PCNA promotes EMT via up-regulating transcription factor Snail.

Up-regulation of Snail in PCNA YD cells is mediated through the PI3K/Akt pathway
To determine the molecular basis by which Snail is up-regulated in HeLa-PCNA YD cells, we conducted Ingenuity Pathway

Phosphorylated PCNA stimulates cancer progression
Analysis (IPA) of the DEG profile, as described (39,40). Among the top seven canonical pathways identified that are unique to PCNA YD cells, as opposed to PCNA WT cells, we found that the phosphatidylinositol 3-kinase (PI3K)/Akt pathway is the most active one, with a positive Z-score of 1.81 (Fig. 3A), suggesting that the PI3K/Akt pathway is associated with PCNA-YDmediated EMT. We also compared the IPA data in PCNA YD cells with those in HeLa and HeLa-PCNA YF cells and plotted the enrichment scores as a heat map in Fig. 3B. The results showed that activation of the PI3K/Akt signaling pathway was the most enriched.
ATM (ataxia telangiectasia, mutated), a PI3K-related kinase, has been shown to promote cancer metastasis via stabilizing Snail (41). It is also known that the PI3K/Akt pathway downregulates glycogen synthase kinase 3 ␤ (GSK3␤) and up-regulates Snail (42)(43)(44)(45)(46). We hypothesized that these molecules (ATM, Akt, GSK-3␤, and Snail) lie in the phosphorylated PCNA-mediated EMT signaling pathway. Upon inducing PCNA molecules by doxycycline (DOX), we analyzed cell lysates from various PCNA-expressing cells for activation of ATM and Akt, as well as up-and down-regulation of Snail and GSK3␤, respectively. As shown in Fig. 3C, PCNA YD cells, but not the three control cell lines (HeLa, PCNA WT , and PCNA YF ), expressed a high level of Ser-1981-phosphorylated ATM (active form). Increased phosphorylation of both Akt (Ser-473, active form) and GSK3␤ (Ser-9, inactive form) was also observed in cells expressing PCNA-YD, but not in the three control cell lines (Fig. 3C). Accordingly, high levels of Snail were associated with these phenomena (Fig. 3C). To verify the involvement of ATM signaling in PCNA-YD-promoted EMT, we treated cells with wortmannin, a PI3K-Akt pathway inhibitor (47)(48)(49). We found that wortmannin treatment dramatically reduced the phosphorylation levels of ATM, Akt, and GSK3␤ (Fig. 3D). We also observed a dramatic reduction in Snail expression (Fig. 3D). In contrast, the treatment led to an increase in E-cadherin expression (Fig.  3D). Consistent with these anti-EMT phenomena, the transwell migration assay showed that wortmannin treatment blocked PCNA YD cells' ability to migrate (Fig. 3, E and F). These results strongly suggest that the EMT in PCNA YD cells is processed through the PI3K/Akt/GSK3␤/Snail signaling axis.
We examined this idea further with experiments that directly inactivated ATM. First, we knocked down the expression of ATM in HeLa-PCNA YD cells using shRNAs against ATM and measured their EMT-related activities. The results showed a 45% reduction in ATM in PCNA YD cells (Fig. 4A). This partial reduction in ATM essentially blocked Akt activation. As expected, the knockdown was also associated with increased expression of GSK3␤ and decreased expression of Snail (Fig.  4A). More importantly, the partial depletion of ATM substantially reduced PCNA YD cells' migration capacity (Fig. 4, B and C). Second, we treated HeLa-PCNA YD cells with KU-55933, an ATM-specific inhibitor. As expected, KU-55933 treatment inactivated ATM and blocked the downstream EMT process, much like what we observed in ATM knockdown cells (Fig. 4, D-F). We, therefore, conclude that the ATM/Akt/GSK3␤/ Snail signaling axis executes phosphorylated PCNA-mediated EMT.

Phosphorylated PCNA stimulates cancer progression Cells expressing PCNA-YD arrest at G 2 /M
It is well-documented that EMT is associated with G 2 /M cell cycle arrest (50), although the cause and effect relationship is unclear. We therefore analyzed cell cycle distribution in cells expressing individual forms of PCNA. The results showed that all three control cell lines (HeLa, HeLa-PCNA WT , and HeLa-PCNA YF ) demonstrated essentially the same pattern of cell cycle phases, but the pattern in PCNA YD cells is visibly different from that of the control cells (Fig. 5, A and B), as PCNA YD cells show more than twice (35% versus 17%) as many G 2 /M cells as each of the control cell lines (Fig. 5C), suggesting a G 2 /M arrest in PCNA YD cells. Morphologically, many PCNA YD cells exhibit a much larger nucleus (see red arrows) than control cells in Hoechst stain (Fig. 5D), indicating abnormal mitotic division in PCNA YD cells. Consistent with this assumption, PCNA YD cells contained a higher percentage of cells with a DNA content greater than 4N, compared with control cells. The production of these abnormal nuclei appears to depend on phosphorylated PCNA, because the percentage of cells with an abnormal nucleus is proportional to the length of DOX treatment (Fig.  5E). We then analyzed Cdc2 and Cdc25C, both of which are hallmarks of G 2 /M arrest (51-53). We observed high levels of phosphorylated Cdc2 (Tyr-15) and phosphorylated Cdc25C (Ser-216) in PCNA YD cells (Fig. 5F). Correspondingly, we detected phosphorylation of Chk1 and Chk2, downstream substrates of activated ATR and ATM, respectively, and upstream kinases of Cdc25C, in PCNA YD cells. These results indicate that PCNA phosphorylation-induced EMT is associated with G 2 /M arrest, which is probably triggered through the ATM/ATR signaling pathway (51)(52)(53). However, given the fact many PCNA YD cells underwent nuclear degradation when treated with an ATR inhibitor (Fig. S1), we believe that ATR activation in HeLa-PCNA YD cells is unrelated to EMT, but it is essential for cell survival and other cellular functions (54,55).

Discussion
PCNA is a critical cell proliferation factor that orchestrates essentially all metabolic reactions at the replication fork, including DNA replication and DNA repair (1-3). PCNA can be phosphorylated at Tyr-211 by tumor-promoting factor EGFR (9). We have shown previously that Tyr-211-phosphorylated PCNA inhibits DNA mismatch repair and induces nucleotide misincorporation during DNA synthesis, thereby induc-
We revealed the involvement of phosphorylated PCNA in EMT through a gene set enrichment analysis of RNA-Seq data derived from cells expressing individual isoforms of PCNA examined in this study, which identified molecular hallmarks specific to EMT in PCNA YD cells (Fig. 1D). Consistent with this prediction, we found that PCNA YD cells indeed display EMT characteristics. First, these cells exhibit a migration activity that is much more active than that of control cells (Fig. 1, E and F); second, EMT-promoting factor Snail and EMT-inhibitory factor E-cadherin are up-and down-regulated, respectively ( Fig.  2A), in PCNA YD cells; and third, Snail knockdown prevents PCNA YD cells from migrating (Fig. 2, C and D). We show that Snail-mediated EMT appears to be activated by the PI3K/Akt signaling pathway. Evidence supporting this notion initially came from the IPA of the DEG's profile, where the PI3K/Akt signaling pathway shows the most active one in PCNA YD cells. Correspondingly, we found that PCNA YD cells, but not control cells, expressed high levels of activated ATM and Akt (Fig. 3C).
The activation of ATM and Akt is coupled with the downregulation of GSK3␤ and the up-regulation of Snail (Fig. 3C). Inhibition of the ATM/Akt signaling pathway by wortmannin, KU-55933, or ATM knockdown blocks the up-regulation and down-regulation of Snail and GSK3␤ (Figs. 3D and 4, A and D), respectively, as well as the migration activity of PCNA YD cells (Figs. 3E and 4, B and E).
On the basis of previously published data and the results presented here, we propose a signaling cascade for PCNA phosphorylation-induced EMT (Fig. 6). It has been shown that phosphorylated PCNA inhibits DNA mismatch repair and induces misincorporation during DNA synthesis (14); cells with phosphorylated PCNA make numerous errors during replication. Even though the molecular basis underlying phosphorylated PCNA-induced misincorporation is unclear, it has been postulated that the modified PCNA recruits an error-prone translesion polymerase to carry out DNA synthesis (14). The nonprocessive nature of translesion polymerases can lead to a severe delay in DNA replication. Although the delay in replication induces G 2 /M arrest, both the replication errors and the delays cause replication stress to activate the ATM-mediated DNA damage response. The activated ATM, which also facilitates G 2 /M arrest (51), then triggers a cascade of signaling reactions,
However, there are many uncertainties. Although activation of the ATM/ATR DNA damage response pathway can trigger G 2 /M cell cycle arrest (51), recent studies also suggest that the process of EMT induces G 2 /M arrest (49). It is well-established that genomic instability can lead to polyploidy/aneuploidy, which can result from G 2 /M arrest and abnormal cell division, leading to cancer development and progression (56). Thus, whether the G 2 /M arrest observed in HeLa-PCNA YD cells is induced by replication stress-activated ATM signaling or is a result of EMT is unknown. It is possible that these processes mutually promote each other. Another big question concerns the exact abnormal event or signal generated by the phosphorylated PCNA-mediated reaction at the replication fork that triggers the ATM/Akt signaling pathway to promote EMT. Future studies are required to address these important questions.

Cell culture and materials
HeLa and HeLa knock-in cell lines were cultured in DMEM with 10% FBS at 37°C in a humidified atmosphere with 5% (v/v) CO 2 . To induce the expression of the knock-in PCNA, cells were treated with 1.0 g/ml of DOX for 4 days. When present, the concentration of KU55933 (Selleck, S1092) or wortmannin (Selleck, S2758) used was 10 M or 2 M, respectively.

Cell cycle analysis by flow cytometry
For cell cycle determination, cells were cultured in media with 10 g/ml Hoechst for 10 min before harvesting. Cells were then fixed with 75% ethanol before cytometry analysis.

Phosphorylated PCNA stimulates cancer progression In vitro transwell cell migration assay
An in vitro migration assay was performed using a 24-well Boyden chamber (32). Approximately 5 ϫ 10 4 cells (in 100 l media) treated with DOX for 4 days were added to the upper chamber in serum-free medium containing DOX. The lower compartment was filled with 650 l medium containing 10% FBS and DOX. All cells were seeded in the upper part of the Boyden chamber and incubated for 12 h. Nonmigrated cells were scraped from the upper surface of the membrane with a cotton swab. Migrated cells remaining on the bottom surface were fixed with 4% neutral buffered formaldehyde and then stained with 0.5% crystal violet for 20 min. The migratory phenotypes were determined by counting the cells that migrated to the lower side of the filter using microscopy at 10ϫ magnification. Five fields were counted for each filter, and each sample was assayed in triplicate.

shRNA experiment
shRNAs for ATM and Snail were synthesized at the Center for Biomedical Analysis at Tsinghua University and cloned into the pLKO.1 vector. A scrambled shRNA was used as a control, whose sequence is CCGGCAACAAGATGAAGAGCACCAACTCG-AGTTGGTGCTCTTCATCTTGTTGTTTTT. Five different shRNAs were used for each shRNA knockdown. Western blot analysis was used to determine protein knockdown. We then selected the two most effective knockdown clones. For shSnail, effective knockdown was from sequences CCGGCCAGGCT-CGAAAGGCCTTCAACTCGAGTTGAAGGCCTTTCGAGC-CTGGTTTTTG and CCGGGCAGGACTCTAATCCAGAGT-TCTCGAGAACTCTGGATTAGAGTCCTGCTTTTTG. For shATM, these sequences were CCGGCAAACGAAATCTCAG-TGATATCTCGAGATATCACTGAGATTTCGTTTGTTT-TTTG and CCGGGTATTACCTTTCGTGGTATAACTCGA-GTTATACCACGAAAGGTAATACTTTTTTG.

RNA-Seq and analysis
RNAs were isolated from cells treated with DOX for 4 days using TRIzol Reagent (Invitrogen), and sequencing libraries were generated using NEBNext ® Ultra TM RNA Library Prep Kit for Illumina ® (New England Biolabs). The resulting libraries were sequenced on an Illumina platform and 150-bp pairlength reads were collected. Differentially expressed genes of two groups (HeLa-PCNA YD versus HeLa, HeLa-PCNA WT , or HeLa-PCNA YF ) were calculated using software DEGSeq (1.12.0). The resulting p values were adjusted using the Benjamini-Hochberg procedure. Genes with an adjusted p value Ͻ0.05 were assigned as differentially expressed. The complete unedited RNA-Seq datasets are available at Gene Expression Omnibus (GEO) database (accession number GSE127276).
The differentially expressed genes were used for GSEA analysis (30,31). The gene set collection used was the h.all.v6.2.symbols.gmt [Hallmarks] gene sets database. GSEA analyzed 1000 permutations, and enrichment statistic was weighted. For IPA, the differentially expressed coding genes identified by the RNA-Seq at a p Ͻ0.05, -fold change Ͼ2 were uploaded to the IPA software for canonical pathway analysis. The analysis was conducted based on prior knowledge of pathway network stored in the Ingenuity Knowledge Base. The p value represents the significance of a given pathway and the Z-score predicts the pathway activation or inactivation.

Western blot analysis
Whole cell lysates were obtained in the radioimmunoprecipitation assay lysis buffer containing protease and phosphatase inhibitors. Protein concentrations were determined using BCA Protein Assay Reagents (Thermo Fisher). Proteins were separated by SDS-PAGE on polyacrylamide gels, transferred onto PVDF membranes, and detected by Western blot analysis using antibodies against specific proteins.

Statistical analysis
All statistical analyses were performed with one-way analysis of variance (ANOVA) test using GraphPad Software. Data were expressed as mean Ϯ S.D. and were considered statistically significant if p values were less than 0.05 or 0.01, as indicated.