Uev1A Promotes Breast Cancer Cell Migration and EMT by Up-Regulating CT45A Expression via the AKT Pathway

Introduction UEV1A encodes a ubiquitin-E2 variant closely associated with tumorigenesis and metastasis, but its underlying mechanism in promoting tumorigenesis remain to be investigated. Methods In this study, we experimentally manipulated UEV1A and CT45A gene expression and monitored their effects on cancer-related gene expression, cell migration and the signal transduction cascade. It was found that UEV1A overexpression induces CT45A family gene expression in breast cancer cells. Indeed, ectopic expression of UEV1A is sucient to induce CT45A and its downstream genes involved in tumorigenesis, epithelial-mesenchymal transition (EMT), stemness and metastasis, and promote cell migration and EMT. Consistently, depletion of CT45A abolishes the above effects, indicating that CT45A is a critical downstream effector of Uev1A. The Uev1A-induced cell migration and EMT is dependent on AKT signaling but independent of NF-κB signaling, indicating that CT45A acts downstream of the AKT pathway. Based on previous reports and observations in this study, we propose that the Ubc13-Uev1A complex activates AKT through K63-linked polyubiquitination, which leads to enhanced CT45A expression, stimulated cell migration and EMT. Since similar effects were also observed in a colorectal cancer cell line, we propose that the Ubc13/Uev1A-AKT-CT45A axis promotes tumorigenesis and metastasis in a broad range of tissues.

Despite the fact that Uev1A and Mms2 are two major Uevs in mammalian cells and share a similar biochemical activity, they appear to function differently: Ubc13-Mms2 is required for DNA-damage response, whereas Ubc13-Uev1A is involved in NF-κB activation [18] and AKT activation [6]. Previous studies demonstrated that Uev1A-Ubc13 represses stress-induced apoptosis in HepG2 cells [20] and promotes breast and colon cancer metastasis through the NF-κB signaling pathway [19,21]. Meanwhile, Uev1A-Ubc13 promotes breast cancer cell survival and chemoresistance through the AKT pathway [6].
Consistently, chemical inhibition of the Uev1A-Ubc13 interaction suppresses cells survival and proliferation of diffuse large B-cell lymphoma cells [22]. These results collectively indicate that Uev1A is involved in tumorigenesis and metastasis.
The PI3K/AKT signaling pathway is an essential node in mammalian cells and is closely associated with various biological functions including cell growth, survival, proliferation, migration, resistance to apopotosis, differentiation, metabolism and angiogenesis [23][24][25][26]. In addition, this pathway is frequently found to be abnormally activated and altered in many human malignancies, which induces chemoresistance and malignant transformation [27][28][29][30].
In this study we found that overexpression of UEV1A induced CT45A expression in breast cancer cells in an Ubc13-dependent manner, while depletion of Uev1 inhibited CT45A expression. This study revealed that ectopic expression of CT45A could upregulate expression of its downstream genes related to tumorigenic, EMT, stemness and metastasis, and promote breast cancer EMT and cell migration. A series of experimental results support a notion that CT45A is a critical downstream gene of the AKT signaling pathway but not the NF-κB signaling pathway. Since similar effects were also observed in a colorectal cancer cell line, the Uev1A-/Ubc13-AKT-CT45A axis in tumorigenesis may be a general phenomenon.
Hence, this study suggests a potential therapeutic target in the treatment of breast and colorectal cancers.

Plasmids and cell transfection
Human UEV1A and CT45A open reading frames (ORFs) were ampli ed as KpnI-XhoI fragments and cloned into plasmid vector pcDNA4.0/TO/HA (+) as previously described [19]. The mutated Ubc13binding site (F38E) in UEV1A was designed based on a previous study with Mms2-F13E [17]. The CT45A small interfering RNA (siRNA) was purchased from GenePharma (Shanghai, China). The sequence for CT45A siRNA is 5'-GGAGAGAAAAGGAUCAGAUUU-3'. The modi ed sequence for UEV1A small hairpin RNA (shRNA, sc-38606-v) and negative control shRNA (sc-108080) delivered by lentiviral particles were obtained from Santa Cruz Biotechnology, Inc. The lentiviral particle infection of MDA-MB-231 and MCF7 breast cancer cells was performed following instructions of the supplier.
RNA preparation and quantitative real-time RT-PCR (qRT-PCR) Total RNAs were extracted from cultured MDA-MB-231, MCF7 breast cancer and HCT116 colorectal cancer cells using Trizol (Invitrogen, 15596018). First-strand cDNA was synthesized with 1 μg of total RNAs with TransScript® All-in-One First-Strand cDNA Synthesis SuperMix (TransGen, AT341-01) according to manufacturer's instructions. qRT-PCR analysis based on SYBR® Premix Ex Taq™ (Takara, RR420A) was performed on the BioRad real-time PCR machine. The speci c primers are listed in Supplementary Table 1. The data analysis was performed using the 2 -ΔCT comparative cycle threshold method [31] from three independent experiments.

Protein extraction and western blotting
Cells were grown to log phase and lysed with whole-cell extraction buffer (150 mM NaCl; 1% NP-40; 10% glycerol; 1 mM EDTA; 50 mM Tris; 1 mM PMSF) and protease inhibitor cocktail for mammalian cells (Roche). Proteins in cell extracts were separated by 8-12% SDS-PAGE gels and transferred to PVDF membrane. The membrane was blocked with 5% milk/BSA and incubated with speci c primary antibodies followed by secondary antibodies. The following antibodies were used: anti-AKT (#4691, Cell

Cell migration assay
In vitro cell migration ability was measured by a Transwell assay without Matrigel coating, using 8-μmpore-size polycarbonate membrane lters in 24-well culture plates. After starving in FBS-free DMEM medium for 24 hours, 2×10 5 HCT116, 5×10 4 MDA-MB-231 and 2×10 5 MCF7 cells were seeded in the upper chamber, while the lower surface of the lter was coated with 10% FBS-DMEM as chemoattractants. The cells were allowed to migrate for 24 hours and the cells migrated to the lower surface of the lter were counted in ve random elds under a light-microscope at high magni cation. These experiments were done at least in triplicate.

Statistical analysis
The statistical signi cance of differential ndings between the control and experimental groups was determined by student's t-test as implemented by Microsoft Excel 2016 (*, P <0.05; ** P <0.01; and ***, P <0.001).

Results
Uev1A upregulates CT45A expression in a Ubc13-dependent manner We performed a microarray analysis by comparing UEV1A-overexpressed and vector control MDA-MB-231 breast cancer cells, which revealed 47 genes upregulated by more than vefold in UEV1A-overexpressed MDA-MB-231 cells (Supplementary Table S2). Interestingly, 16 out of 47 belong to cancer/testis antigens (CTAs), among which CT45A family members are most highly elevated in UEV1A-overexpressed MDA-MB-231 cells ( Figure S1A). CTAs are tumor-associated and testis-derived speci c immunogenic antigens closely associated with spontaneous immune responses in cancer patients [32,33]. They are not expressed in nearly all normal tissues except testis after birth, but are highly expressed in various types of cancers [34][35][36][37][38][39]. Within CTAs, the CT45A gene family comprises 10 genes designated as CT45A1 to CT45A10, which are distinct but highly conserved, as their amino-acid sequences exhibit more than 98% identity [40] ( Figure S1B). To independently examine the role of Uev1A in upregulating CT45A expression and its biological implications, UEV1A was cloned into a pcDNA4.0/TO/HA(+) vector and then trasfected into MDA-MB-231-TR and MCF7 cells to construct stable cell lines as previously reported [19]. The level of UEV1A ectopic expression was monitored by western blot against the HA-tag after 10 μg/ml doxycyline (Dox) treatment ( Figure S2A,B). Then CT45A expression was measured by qRT-PCR and found to be signifcantly upregulated in UEV1A-overexpressed MDA-MB-231-TR ( Figure 1A) and MCF7 ( Figure 1B) cells. It has been reported that UEV1A is upregulated in MDA-MB-231 and MCF7 cells by 2.8-and 4-fold, respectively [19]. To ask whether this moderate overexpression of UEV1A contributes to CT45A upregulation in breast cancer cells, we suppressed the endogenous UEV1A expression in MDA-MB-231 and MCF7 cells using shRNAs delivered by lentiviral particles as previously reported [19]. It was found that two independent shUEV1A constructs, shUEV1A-1 and shUEV1A-2, reduced UEV1A expression in MDA-MB-231 cells by 43% and 60% ( Figure S3A), and in MCF7 cells by 71% and 85% ( Figure S3B), respectively, compared to contral shRNA-treated cells. Meanwhile, the CT45A transcript levels were also reduced ( Figure 1C, D). To further ask whether Uev1A upregulates CT45A expression in a Ubc13dependent manner, we constructed stable MDA-MB-231-TR and MCF7 cell lines expressing Dox-inducible Uev1A-F38E mutant protein ( Figure S2A,B), which is known to abolish physical interaction between Mms2/Uev1A and Ubc13 [7,17,19]. As expected, UEV1A-F38E failed to upregulate CT45A mRNA levels in both MDA-MB-231-TR ( Figure 1E) and MCF7 ( Figure 1F) cells. These observations collectively indicate that Uev1A upregulates CT45A expression in a Ubc13-dependent manner in breast cancer cells.
Uev1A positively regulates CT45A downstream gene expression in breast cancer cells CT45A has been reported to act as a proto-oncogene through upregulating tumorigenic and metastatic genes [41]. We rst measured the transcript level of several CT45A previously-reported [41] downstream genes thought to be involved in tumoregenesis, EMT, stemness and metastasis after CT45A ectopic expression. The expression of some tumoregenesis-associated genes, including those encoding RAS exchange factor (RASGEF1A), melanoma antigen family member (MAGED4B), homeobox B6 (HOXB6 and HOXD13) was indeed signi cantly higher in CT45A-overexpressed MDA-MB-231 ( Figure 2A) and MCF7 ( Figure 2B) cells than their respective control cells. Expression of several EMT, stemness and metastasis related genes, including TWIST1, KIT, aldehyde dehydrogenase 1 family, member A1 (ALDH1A1), CXCR4 and/or SULF2 were also upregulated in CT45A-overexpressed MDA-MB-231 ( Figure 2C) and MCF7 ( Figure  2D To ask whether Uev1A is a critical regulator for CT45A-induced migration, we successfully depleted CT45A by approximately 50% using siRNA in MDA-MB-231 ( Figure S3C) and MCF7 ( Figure S3D) cells. Similar effects were also observed in UEV1A-overexpressed MDA-MB-231 ( Figure 3D) and MCF7 ( Figure  4D) cells. The above treatment does not affect the expression of UEV1A (Figures 3E and 4E), but the moderate CT45A depletion in UEV1A-overexpressed cells markedly reduced cell migration as determined by a transwell assay without matrigel ( Figures 3F,G and 4F,G). The above ndings allow us to conclude that CT45A is a critical regulator for Uev1A-induced migration in breast cancer cells, as partial depletion of CT45A can reverse cell migration in UEV1A-overexpressed breast cancer cells.

CT45A promotes metastasis in other type of cancer cells
To ask whether UEV1A overexpression also increases CT45A expression in other cancer cells, we created stable UEV1A-transfected HCT116 colorectal cancer cell lines as previously described ( Figure 5A) [21], in which CT45A was moderately upregulated upon UEV1A ectopic expression, and this upregulation was not observed in the UEV1A-F38E-expressed HCT116 cells ( Figure 5B). To ask whether this moderate overexpression of UEV1A contributes to CT45A upregulation in colorectal cancer cells, we suppressed the endogenous UEV1A expression in HCT116 cells by using shRNAs delivered by lentiviral particles as previously reported [21]. It was found that two independent shUEV1A constructs, shUEV1A-1 and shUEV1A-2, reduced UEV1A expression in HCT116 cells by 55% and 65%, respectively ( Figure S4A), compared to control shRNA-treated cells. Meanwhile, the CT45A transcript levels were also reduced ( Figure 5C). To ask whether ectopic expression of CT45A could promote metastasis in other types of cancer cells, HCT116 cells were transiently transfected with pcDNA4.0/TO/HA-CT45A and the CT45A expression was monitored by western blot analysis against HA-tagged CT45A ( Figure 5D). The CT45A ectopic expression resulted in concommitant increase in HCT116 cell migration by sevenfold ( Figure  5E,F), indicating that CT45A could also promote tumorigenesis and metastasis in other types of cancer cells. To further ask whether CT45A is a critical regulator for UEV1A-induced migration, we depleted CT45A by using siRNA in UEV1A-overexpressed HCT116 cells. As shown in Figure S4B, CT45A was depleted by 44%. The above treatment does not affect the expression of UEV1A ( Figure S4C), but the moderate CT45A depletion in UEV1A-overexpressed HCT116 cells markedly reduced cell migration as determined by a transwell assay without matrigel ( Figure 5G,H). The above ndings indicate that Uev1A induces colorectal cancer cell migration through upregulating CT45A genes.
Depletion of CT45A can reverse EMT in UEV1A-overexpressed breast cancer cells Epithelial-mesenchymal transition (EMT) is closely associated with cancer progression, cancer cell metastasis and drug resistance [42,43]. Cells undergoing EMT display increased expression of mesenchymal genes including N-cadherin, bronectin and vimentin, and decreased expression of epithelial genes including E-cadherin, occulin and ZO-1 [44]. It was reported that overexpression of CT45A could induce breast cancer EMT, and thus foster cancer metastasis by upregulating EMT master gene TWIST1 [41]. To further investigate the potential molecular mechanisms by which CT45A regulates breast cancer cell migration, we monitored alterations of EMT markers, including N-cadherin and vimentin, two well-characterized mesenchymal markers, and E-cadherin, a well-known epithelial marker [44,45]. Consistent with breast cancer cell migration, increased mRNA levels of N-cadherin and vimentin and decreased E-cadherin were found upon CT45A overexpression in MDA-MB-231 ( Figure 6A) and MCF7 ( Figure 6B) cells. We also assessed the effect of CT45A on cellular N-cadherin and E-cadherin at protein levels. Firstly, we monitored cellular N-cadherin and E-cadherin levels in MDA-MB-231 and MCF7 cells and found that MDA-MB-231 and MCF7 cells produced detectable N-cadherin and E-cadherin, respectively ( Figure 6C). Interestingly, ectopic expression of CT45A increased N-cadherin in MDA-MB-231 cells and decreased E-cadherin in MCF7 cells ( Figure 6C,E,F), suggesting that cell migration stimulated by ectopic CT45A expression is likely due to the enhanced EMT in breast cancer cells. To address whether Uev1A is a critical upstream regulator of CT45A-induced EMT, we depleted CT45A by using siRNA in UEV1Aoverexpressed MDA-MB-231 and MCF7 breast cancer cells (Figures 3D and 4D), which signi cantly increased E-cadherin protein levels in UEV1A-overexpressed MCF7 cells and decreased N-cadherin protein levels in UEV1A-overexpressed MDA-MB-231 cells ( Figure 6D,G,H). Collectively, these results support a notion that Uev1A can serve as an important regulator for CT45A-induced EMT in breast cancer cells.
Uev1A regulates CT45A expression through the AKT signaling pathway Since Uev1A has been reported to be associated with NF-κB activation [19][20][21] and AKT activation [6], we wish to investigate molecular mechanisms by which Uev1A regulates CT45A expression. To ask whether Uev1A regulates CT45A expression through the NF-κB pathway, MDA-MB-231 and HCT116 cells transiently overexpressing UEV1A were treated with the NF-κB pathway inhibitor Bay11-7082 [46] and its e cacy was measured by the nuclear P65 level (Figure S5A,C). The CT45A transcript level was not signi cantly reduced in UEV1A-ovexpressed MDA-MB-231 ( Figure S5B) and HCT116 cells ( Figure S5D) by treatment with Bay11-7082, indicating that Uev1A upregulation of CT45A expression is independent of the NF-κB pathway. To ask whether Uev1A regulates CT45A expression through the AKT signaling pathway in breast cancer cells, phosphorylation levels of both AKT-Thr308 and AKT-Ser473 in MDA-MB-231 and MCF7 cells transiently overexpressing UEV1A were rst monitored by western blot and found to be increased ( Figure 7A). In contrast, overexpression of UEV1A-F38E failed to induce AKT phosphorylation at both residues ( Figure 7A), indicating that the effects of Uev1A on AKT is dependent on its interaction with Ubc13. These observations allow us to conclude that excessive Uev1A promotes the Uev1A-Ubc13 complex formation, which activates the AKT signaling pathway. To further address whether Uev1A promotes CT45A expression through the AKT signaling pathway, we examined effects of PI3K/AKT pathway inhibitor LY294002 [47] on MDA-MB-231 and MCF7 cells with ectopic UEV1A expression. As seen in Figure 7B, the AKT-Ser473 phosphorylation level was markedly decreased after LY294002 treatment in UEV1A-overexpressed MDA-MB-231 and MCF7 cells compared to those without the inhibitor treatment. We then examined CT45A expression and found that, compared to cells without LY294002 treatment, the CT45A transcript level was signi cantly reduced in UEV1A-overexpressed MDA-MB-231 ( Figure 7C) and MCF7 ( Figure 7D) cells after 10 μM LY294002 treatment. After 20 μM LY294002 treatment, the CT45A transcript further decreased to levels below the vector control cells without the inhibitor treatment ( Figure 7C,D). It was previously reported that insulin-like growth factor (IGF-1) is an important activator of the PI3K/AKT signaling pathway [48,49]. To further investigate whether CT45A is indeed a direct downstream gene of the AKT signaling pathway, we treated MDA-MB-231 ( Figure 7E) and MCF7 ( Figure 7F) cells with IGF-1, and found that the AKT-Ser473 phosphorylation level was dramatically increased after IGF-1 treatment compared to untreated cells. Under the above experimental conditions, the CT45A mRNA levels were signi cantly increased in MDA-MB-231 ( Figure 7G) and MCF7 ( Figure 7H) cells after IGF-1 treatment. Collectively, we conclude that Uev1A-Ubc13 regulates CT45A expression through the AKT signaling pathway in breast cancer cells.

Discussion
Previous reports have identi ed CT45A as a chemosensitivity mediator and immunotherapy target in ovarian cancer [40,50]. In addion, CT45A has no detectable expression in normal tissues after birth, except for the testis, but it is closely associated with the progression and development of various cancers [41,[51][52][53][54]. In particular, it is highly expressed in cancer stem cells (CSCs), but not in differentiated cells [55], indicating that it is a promising biomarker for diagnosis and treatment of cancer patients. However, exactly how the CT45A family genes function in these processes remain unclear.
The CT45A family genes were brought to our attention based on our preliminary microarray data from which CT45A family genes were among the highest induced genes following UEV1A overexpression in MDA-MB-231 breast cancer cells. This obervation was independently con rmed in two breast cancer cell lines, although the levels of CT45A induction after UEV1A overexpression vary. In this study, we rst investigated the correlation between CT45A and tumorigenesis using breast cancer cell models. At the beginning of our investigation, the CT45A gene family was thought to comprise six members (CT45A1-CT45A6) and their amino-acid sequences share more than 98% identity; hence we cloned one of them (CT45A1) to represent all members. Consistantly, siRNAs used in this study were designed to target all six CT45A family genes. Recently, the CT45A family has been updated to 10 genes in NCBI, and their aminoacid sequences still share more than 98% identity [40], making our initial experimental designs still valid.
We overexpressed CT45A in MDA-MB-231 and MCF7 breast cancer cells and found that CT45A could promote cell migration, EMT and its downstream tumorigenic, EMT, stemness and metastatis related genes expression, indicating that CT45A plays an important role in breast cancer tumorigenesis.
A previous study showed that CT45A protein has a DEAD/H box which has RNA helicase activity and putative nucleic acid binding function [41]. RNA helicases of DEAD box family are required for gene expression and transcrition by interacting with RNA polymerase II (Pol II ) [56], whether CT45A interacts with RNA Pol II or other transcription factors to promote tumorigenesis and metastasis remains to be further elucidated.
This study investigated the correlation between Uev1A and CT45A in breast cancer cell migartion and EMT. It was found that Uev1A upregulates CT45A expression in a Ubc13-dependent manner in one colorectal cancer and two breast cancer cell lines. In a reverse expreriment, depletion of UEV1A in the above three cancer cell lines signi cantly inhibited the upregulation of CT45A, indicating that Uev1A plays a critical role in the upregulation of CT45A. Siminarly, Uev1A positively regulates the expression of CT45A downstream tumorigenic, EMT, stemness and metastatis related genes in breast cancer cells. Moreover, under our experimental conditions, we found that N-cadherin was readily detectable in MDA-MB-231 but not MCF7 cells, while E-cadherin was detected in MCF7 but not MDA-MB-231 cells. Furthermore, ectopic expression of CT45A could increase N-cadherin in MDA-MB-231 cells and decrease E-cadherin in MCF7 cells, both of which are expected to promote tumorigenesis. Indeed, CT45A depletion in UEV1A-overexpressed cells reduced EMT and cell migration to a level comparable to that of controltransfected cells. These ndings together indicate that Uev1A is a critical regulator of CT45A-induced cell migration and EMT in breast cancer.
In order to determine through which signaling pathway(s) Uev1A upregulates CT45A expression, we treated UEV1A ectopic expression cells with NF-κB and PI3K/AKT pathway inhibitors and found that inhibition of AKT markedly decreased CT45A expression, while inhibition of NF-κB activity had no observable effects. To further con rm that CT45A is a direct downstream gene of the AKT pathway, we treated breast cancer cells with the AKT pathway activator IGF-1 and found that the IGF-1 treatment leads to CT45A induction. The AKT signaling pathway is closely associated with many biological processes such as cell proliferation, migration and differentiation [24]. It has been reported that AKT undergoes the TRAF6-triggered K63-linked polyubiquitination, which is critical for AKT membrane localization, phosphorylation and subsequent activation [57,58]. Since Uev1A-Ubc13 is the only known E2 complex to regulate K63-linked polyubiquitination leading to the AKT pathway activation in breast cancer [6], this study reveals a novel Uev1A/Ubc13-AKT-CT45A axis to promote breast cancer cell migration and EMT ( Figure 8). Given limited but consistent observations in a colorectal cancer cell line, the above signaling cascade may be expanded to other types of cancers.

Conclusions
Overexpression of UEV1A is su cient to activate the AKT pathway in breast cancer cell lines, which in turn upregulates CT45A expression to promote breast cancer cell migration and EMT. These observations provide a potential therapeutic target in the treatment of breast cancer. TN and ZW participated in the project design and carried out all experiments. TN wrote the initial draft manuscript. WX conceived the study, participated in the project design, manuscript preparation and submission. All aothors read and approved the nal manuscript.