HeLa cells were obtained from the American Type Culture Collection (CCL-2) and Ca Ski cells (87020501) were purchased from the European Collection of Authenticated Cell Cultures. HeLa cells were cultured in DMEM, supplemented with 10% FBS (Sigma-Aldrich; Merck KGaA), 1,000 mg/l glucose (Sigma-Aldrich; Merck KGaA), 2 mM L-Glutamine (Sigma-Aldrich; Merck KGaA) and 1 mM sodium pyruvate (Sigma-Aldrich; Merck KGaA). Ca Ski cells were cultured in RPMI-1640 medium (Sigma-Aldrich; Merck KGaA), supplemented with 10% FBS and 2 mM L-Glutamine. All cells were maintained in a humidified atmosphere of 5% CO₂ and 37˚C.

To mimic hypoxic conditions, 1x10⁵ cells were treated with 500 µM cobalt chloride (CoCl₂) for 0-60 min or 8, 16 and 24 h. For EGF treatment, 5x10⁴ HeLa cells were seeded into 24-well plates containing 12-mm round glass cover slips or on 100-mm dishes. Following 24 h of incubation at 37˚C, the cells were serum-starved overnight and subsequently treated with 100 ng/ml EGF for various times.

Different clones of the HECTD1 KD in HeLa or Ca Ski cells were established through the stable transfection of individual of 4 short hairpin (sh)RNAs (cat. no. TF304134; OriGene Technologies, Inc., which consists of 4 shRNAs against HECTD1, cat. nos. FI316529, FI316530, FI316531 and F1316532). The control cells (Ctrl, NC) were stably trans-fected with an empty shRNA RFP Cloning Vector (scramble, pRFP-C-RS vector; cat. no. TR30014; OriGene Technologies, Inc.). The RFP expression cassettes in the shRNA constructs were excised with XhoI/BglII and blunt-end ligation.

miRTarBase was used to identify hypothetical target sequences for HECTD1 (http://mirtarbase.mbc.nctu.edu.tw/php/search.php). microRNA (miR)-21 and miR-210 antagomirs were purchased from Ambion; Thermo Fisher Scientific, Inc. and the scrambled negative control (anti-miR-scr) was obtained from Shanghai GenePharma Co. Ltd. These antagomirs have been previously described and verified (13-15). Cells were transfected with 50 nM miR-21/-210 antagomir or 25 nM anti-miR-scr using RNAifectin™ (ABM, Inc.), according to the manufacturer's protocol.

hsa-miR-21-5p and hsa-miR-210-3 RT-qPCR kits (Applied Biosystems; Thermo Fisher Scientific, Inc.) were used for RT-qPCR. RT-qPCR was performed using a TaqMan PCR kit (Applied Biosystems; Thermo Fisher Scientific, Inc.), according to the manufacturer's protocol. The following primers were used for qPCR: U6 forward, 5'-CTC GCT TCG GCA GCA CA-3' and reverse, 5'-AAC GCT TCA CGA ATT TGC GT-3'; and HECTD1 forward, 5'-AAT GAA CCA GGG TCA ACT GC-3' and reverse, 5'-TGT GTT TGT CCA CTG GCA TT-3'. The cycling conditions were as follows: 95̊C for 10 min, followed by 30 cycles of (95̊C for 15 sec, 60̊C for 30 sec, 72̊C for 40 sec), then 95̊C for 60 sec. Expression levels were calculated using the 2^-ΔΔCq method (16) and normalized to the loading control U6. All experiments were performed >3 times and expressed as a log₂ scale or fold change of treatment/control. Expression changes with a log₂ ratio of treatment/control >1 or <-1 and P<0.05 were considered significant.

Anti-SNAIL (cat. no. 3879), a nt i- SLUG (c at. no. 9585), a nt i-E - c a d he r i n (cat. no. 3195), anti-N-cadherin (cat. no. 4061), anti-ERK1/2 (cat. no. 4695), anti-phospho-ERK1/2 (Thr202/Tyr204; cat. no. 4374) and anti-GAPDH (2118L) primary antibodies were purchased from Cell Signaling Technology, Inc.; the anti-HECTD1 primary antibody (cat. no. CSB-PA010273GA01HU) was obtained from Cusabio Technology LLC; the anti-ubiquitin primary antibody (cat. no. ab7780) was obtained from Abcam; and the anti-HECTD1 (M03), clone 1E10 primary antibody (cat. no. H00025831-M03) was purchased from Abnova Corp. The FITC-conjugated and Alexa Fluor 546-conjugated anti-rabbit (A-11035) or anti-mouse (A-11003) secondary antibodies were purchased from Invitrogen; Thermo Fisher Scientific, Inc. and the horseradish peroxidase (HRP)-conjugated anti-mouse (cat. no. 1706516) or anti-rabbit (cat. no. 1706515) secondary antibodies were obtained from Bio-Rad Laboratories, Inc. Anti-α-tubulin antibody was obtained from Abcam (ab52866, 1:250). Human recombinant EGF and EGF-Alexa Fluor 488 were purchased from Invitrogen; Thermo Fisher Scientific, Inc. The GFP-SNAIL plasmid (cat. no. 16225) was obtained from Addgene, Inc. and the HECTD1-HaloTag^® human ORF in pN21A clone (cat. no. FHC05410) was purchased from Promega Corp. Leptomycin B (LMB; cat. no. L2913) and ivermectin (IVE; cat. no. I8898) were purchased from Sigma-Aldrich; Merck KGaA.

Briefly, 2.5x10⁶ shRNA-transfected HeLa cells were plated on 100-mm dishes for 16 h. Following the treatment with 100 µg/ml CHX for the indicated time periods, cell pellets were harvested and lysed on ice in the supplemented RIPA buffer. Protein expression was detected by western blot analysis and quantified by densitometric analysis with ImageJ software. Three independent experiments were performed. For LMB or IVE treatments, HeLa cells were treated with or without 50 nM LMB or IVE (1 µM) for 4 h and sequentially treated with 100 ng/ml EGF.

Cells were seeded into 24-well plates containing 12-mm round cover glass slips. Prior to treatment, the cells were serum-starved for 16 h. Following treatment, the cells were washed twice with PBS and fixed with 4% formaldehyde for 15 min at room temperature. The cells were subsequently permeabilized with 0.15% Triton-X100/PBS for 15 min at room temperature and washed twice with PBS. The cells were then blocked with 5% BSA/PBS for 1 h at room temperature and incubated at 4˚C overnight with anti-HECTD1 and anti-SNAIL primary antibodies diluted 1:200 in blocking buffer. Following primary antibody incubation, the slides were washed with PBS 3 times for 15 min and were subsequently incubated with Alexa Fluor-conjugated secondary antibodies (1:500) for 1 h at room temperature. The slides were then washed 3 times with PBS and the nuclei were stained with DAPI for 3 min at room temperature. The cells were subsequently mounted with prolonged gold anti-fade mountant (Invitrogen; Thermo Fisher Scientific, Inc.) and the stained cells were visualized using the Nikon A1R confocal microscope (Nikon Corp.).

The fluorescent intensities of HECTD1 and SNAIL were quantified using NIS-Elements software (Nikon Corp.). The nuclear regions were selected with the DAPI channel and the measured intensities were subtracted from the background intensity (non-cell region) to obtain the fluorescent intensity of each protein. At least 3 independent experimental repeats were performed.

The analysis of the subcellular localization of HECTD1 was performed using two different prediction softwares: RSLpred (17) and PSORT II (18).

The cells were washed twice with pre-chilled PBS at 4˚C and total protein was extracted on ice for 10 min using 200 µl RIPA lysis buffer, containing 10% proteinase inhibitor. The cells were collected into a 1.5 ml Eppendorf tube, vortexed and lysed on ice for 30 min. Subsequently, the cells were centrifuged for 15 min at 6,500 x g at 4˚C and the supernatant was collected. Total protein was quantified using a protein assay (Bio-Rad Laboratories, Inc.) and 20 µg protein/lane were separated on 12% SDS-PAGE to detect ERK1/2 and phospho-ERK1/2 expression and 4% SDS-PAGE to detect HECTD1 expression. The separated proteins were subsequently transferred onto PVDF membranes and blocked with 1% milk powder. The membranes were incubated with the following primary antibodies at 4˚C overnight: Anti-ERK1/2 (1:200), anti-phospho-ERK1/2 (1:1,000), anti-HECTD1 (1:200) and anti-GAPDH (1:100,000). Following primary antibody incubation, the membranes were incubated with HRP-conjugated secondary antibodies (1:5,000) for 1 h at room temperature. Protein bands were visualized using the Pierce™ ECL Plus Substrate (Pierce; Thermo Fisher Scientific, Inc.). Protein expression was quantified using ImageJ software (National Institutes of Health) and normalized to the loading control, GAPDH.

To analyze cell proliferation, the CellTiter96^® AQueous One Solution Cell Proliferation assay (MTS; cat. no. G3582; Promega Corp.) was used according to the manufacturer's protocol. Following serum starvation overnight, a total of 5x10³ cells/well were seeded into 96-well plates with or without 100 ng/ml EGF treatment. Following incubation for 30, 48 and 72 h, 20 µl MTS solution were added to each well and the cells were incubated for 3 h at 37˚C. The absorbance was measured at 490 nm using a Biotek 96-well plate reader (BioTek Instruments, Inc.). The absorbance was normalized to the absorbance at 0 h. A single experiment was performed.

The wound healing assay for HeLa cells was performed as previously described (12). Three independent experimental repeats were performed.

A total of 4x10⁵ Ca Ski or HeLa cells/ml were resuspended in serum-free DMEM and 350 µl cell suspension was plated in the upper chambers of 24-well Transwell plates (8.0 µm; cat. no. MCEP24H48; Merck KGaA). A total of 1 ml DMEM or RPMI-1640 medium supplemented with 20% FBS was plated in the lower chambers. Following incubation for 20 h at 37̊C, the non-migratory cells remaining in the upper chamber were removed using cotton swabs. The migratory cells were fixed with 70% ethanol for 15 min and subsequently stained with 0.1% crystal violet for 15 min at room temperature. The stained cells were counted in four randomly selected fields using a microscope (ECLIPSE Ti2, Nikon). Three independent experimental repeats were performed.

Cell pellets were lysed with IP lysis buffer (20 mM Tris-HCl, pH 8.0, 137 mM NaCl, 1% NP40 and 2 mM EDTA supplemented with 1% protease inhibitor cocktail) on ice for 20 min. The lysate was centrifuged at 9,000 x g for 5 min at 4̊C and the supernatant was collected and transferred to pre-cooled fresh tubes. The protein concentration was equilibrated with the IP lysis buffer. Subsequently, 2 µl anti-GFP GF28R antibody (Thermo Fisher Scientific, Inc.) were added per 500 µg protein sample prior to being incubated overnight at 4˚C. Normal mouse IgG (1:1,000, ab188776, Abcam) was used as a negative control. The lysates were subsequently incubated with pre-washed protein G agarose beads (20 µl/500 µg protein) for 1 h at 4˚C with gentle agitation. The beads were washed 3 times with IP lysis buffer and then centrifuged at 1,600 x g for 3 min at 4̊C. Subsequently, the beads were heated for 5 min at 95˚C in 2X Laemmli sample buffer and target proteins were detected by western blot analysis using specific antibodies as described above. Two independent experimental repeats were performed.

Following 24 h of transfection with GFP-SNAIL or GFP, HeLa cells were washed twice with PBS and incubated in serum-free medium supplemented with 1 nM MG132 or DMSO overnight at 37˚C. For the endogenous ubiquitination assay, starved cells were harvested as pellets and resuspended in serum-free medium. The pellets were centrifuged 9,000 x g for 5 min at 4̊C and lysed with ubiquiti-nation lysis buffer (50 mM Tris, pH 7.5, 1 mM EDTA, 150 mM NaCl, 0.1% Triton X-100, complete protease inhibitor cocktail, 100 µM MG132 and 100 µM N-ethylmaleimide) on ice for 15 min, followed by centrifugation at 12,000 x g for 5 min at 4˚C. The supernatant was collected and used to determine the total protein concentration. Equal concentrations of total protein were immunoprecipitated with GFP antibody and protein G agarose beads, and ubiquitination was subsequently detected using western blotting. Whole cell lysate, which was 5% of total protein, was used as the control. Two independent experimental repeats were performed.

The expression levels of HECTD1 were examined in normal and cancer tissues obtained from the Human Protein Atlas (HPA; http://www.proteinatlas.org) using immunohistochemistry (IHC). Briefly, IHC was performed using anti-HECTD1 (1:50, cat. no. HPA002929; Sigma-Aldrich; Merck KGaA) and anti-SNAIL (1:500, cat no. HPA069985; Sigma-Aldrich; Merck KGaA) primary antibodies, and the source of these tumors and the pathological analysis can be found in the online supplementary materials. Kaplan-Meier plots were generated based on the association analysis between mRNA expression levels and patient survival, and patients were subsequently divided into 2 groups based on the level of expression as the 'Low' or 'High' expression groups. Log-rank P-values are displayed. 329 cervical cancer samples from The Catalogue Of Somatic Mutations In Cancer (COSMIC) and 309 samples from The Cancer Genome Atlas (TGCA) were analyzed using whole-exome sequencing.

Statistical analysis was performed using unpaired one-way ANOVA (Tukey's multiple comparisons test was used as a post-hoc test) or subsequent Student's t-test using GraphPad Prism 7 software (GraphPad Software, Inc.) and data are presented as the means ± SD or SEM, as indicated in the figure legends. P<0.05 was considered to indicate a statistically significant difference.

It has previously been demonstrated that the degradation of SNAIL occurs through a proteasome-mediated mechanism involving several E3 ubiquitin ligases (19-21). SNAIL has been suggested to interact with HECTD1, a HECT domain E3 ubiquitin ligase involved in regulating the dynamic nature of adhesive structures in cells (11). Therefore, it was hypothesized that HECTD1 may be involved in the protein degradation of SNAIL. In this study, Co-IP assays were used to investigate whether HECTD1 physically interacted with SNAIL protein; HECTD1 was present in the immunoprecipitates of cell lysates transfected with an expression vector for GFP-SNAIL, but not in those collected following transfection with the control vector alone (Fig. 1A), although endogenous SNAIL could not be detected in a pulldown in HaloTag-HECTD1 transfected cells. These results suggest that the proteins interact with each other in vivo. To elucidate the potential role of HECTD1 in cancer, several cell lines were established in which HECTD1 was knocked down (HECTD1-KD cells) using shRNA. One of these HECTD1-KD cells established from TF304134, demonstrated markedly reduced expression levels compared with the scrambled shRNA-transfected cells (Ctrl); >90% of HECTD1 expression was knocked down in HeLa cells (Fig. S1A). Since HECTD1 is an E3 ubiquitin ligase, the HECTD1-mediated ubiq-uitination of SNAIL was investigated. The GFP-SNAIL vector was overexpressed in HeLa cells and ubiquitinated proteins were purified using anti-GFP antibody beads, and SNAIL expression was subsequently analyzed by western blot analysis. The ubiquitination of SNAIL was markedly increased following treatment with the proteasome inhibitor, MG132, which blocks ubiquitin-proteasome degradation (Fig. 1B). Furthermore, the HECTD1-KD cells were observed to have reduced ubiquitination levels of SNAIL (Fig. 1B), whereas cells transfected with expression pN21A plasmids containing HECTD1 (HaloTag^®-HECTD1) and the GFP-SNAIL plasmid demonstrated that HECTD1 expression promoted the ubiquitination of SNAIL (Fig. 1C). Overall, these data suggest that SNAIL proteins are the target of HECTD1-mediated ubiquitination.

To identify whether HECTD1 is involved in the degradation of SNAIL, the CHX chase assay was used. Compared with the control cells (Ctrl), cells transfected with HECTD1-KD exhibited markedly decreased degradation levels of SNAIL proteins (Figs. 2A and S2), suggesting that HECTD1 may be one of the E3 ubiquitin ligases that mediates the stability of SNAIL proteins.

It was subsequently hypothesized that the localization of HECTD1 within cells may affect HECTD1-mediated SNAIL degradation. Thus, to examine this hypothesis, the localization of SNAIL in HeLa Ctrl and HECTD1-KD cells was analyzed. Consistent with the results obtained by western blot analysis (Fig. 2A), it was observed that the immunoreactivity of nuclear SNAIL proteins was increased in the HECTD1-KD cells compared with the Ctrl cells (Fig. 2B and C). In contrast to the localization of HECTD1 (Fig. 3A), SNAIL expression was found to be constitutive and dependent on serum/growth factors in HeLa cells (Fig. 2B). Compared with serum starvation (-EGF), the immunoreactivity of nuclear SNAIL protein was increased following the addition of EGF (Fig. 2B and D); however, SNAIL was predominantly expressed in the nucleus of the HECTD1-KD cells, regardless of the treatment condition. These results strongly suggest that HECTD1 may play an important role in controlling the abundance of SNAIL present in the nucleus.

HECTD1 contains 8 putative nuclear localization sequences (NLS) and 4 nuclear export signals (NES; Fig. S3A and B). Through using two localization prediction software, it was discovered that HECTD1 expression was localized within the nucleus (Fig. S3C). Consistent with this finding, endogenous HECTD1 was also found to be localized in the nucleus following serum deprivation, despite its otherwise cytoplasmic localization (Fig. 3A and B), which is similar to FBXL5 (19), but unlike many other types of SNAIL1 E3 ubiquitin ligases, which are localized in the cytosol (21). Notably, EGF treatment stimulated the export of HECTD1 to the cytosol (Fig. 3C and D), indicating that HECTD1 may shuttle back and forth between nucleus and cytoplasm. No changes were observed in the intensity of HECTD1 immunoreactivity during the 4-h period (Fig. 3A and B).

It has been demonstrated that XPO1/CRM1 mediates the nuclear export of numerous types of protein (22). LMB, a well-known natural inhibitor of XPO1/CRM1, has been observed to reduce nuclear export (23). The treatment of HeLa cells with LMB resulted in the accumulation of HECTD1 in the nucleus (Fig. 3C and D), suggesting that HECTD1 may be actively exported from the nucleus to the cytosol.

To investigate the location of HECTD1-mediated SNAIL degradation by ubiquitination, the nuclear export or import of HECTD1 was blocked by LMB or IVE (which is a specific inhibitor of importin α/β-mediated nuclear import), respectively, and the nuclear signals and expression levels of SNAIL were determined by fluorescence intensity. LMB not only increased the nuclear signal of HECTD1, but also that of SNAIL (Fig. 3E and F); however, it was suggested that the nuclear localization of SNAIL was independent of XPO1/CRM1 in HeLa cells. The total expression levels of SNAIL were unaltered (Fig. 3G), indicating that LMB may block the nuclear export of both proteins, but the degradation of SNAIL does not occur in the nucleus. IVE treatment blocked the nuclear import of both proteins (Fig. 3E and F) and decreased the total SNAIL expression levels by 30% (Fig. 3G). Overall, these results suggest that HECTD1-mediated SNAIL degradation occurs in the cytoplasm, but not in the nucleus.

Alterations in cellular adhesion, migration, invasion and morphology are essential for EMT. The knockdown of HECTD1 induced a mesenchymal appearance in the HeLa cells (Fig. 4A). The Ctrl HeLa cells displayed a 'cobblestone (cubed)', epithelial-like phenotype (24), whereas the HECTD1-KD cells acquired a spindle-like, elongated phenotype, which is typical of mesen-chymal cells, suggesting that the knockdown of HECTD1 may induce changes in cell morphology similar to EMT processes.

MEF cells obtained from Hectd1-homozygous mutant (Hectd1^R/R) embryos exhibited an accelerated cell spreading/migratory phenotype (11). Similarly, the wound healing assays demonstrated that closure of the wound occurred more rapidly in the HECTD1-KD compared with the Ctrl cells (Fig. 4B); furthermore, in the Transwell assay, the number of migrated cells was higher in KD cells as compared to the Ctrl cells (Fig. 4C). However, the knockdown of HECTD1 in the HeLa cells did not significantly modify the proliferation rate of the cells (Fig. S1B). Thus, the downregulation of HECTD1 expression levels may be strongly associated with abnormal cell migration and invasion, but not with proliferation.

In addition, the alteration in the expression levels of specific transcription factors, such as SNAIL and SLUG, are also typical features of EMT (20). It has been reported that the EGF-receptor (EGFR) is overexpressed in cervical cancer (25), and SLUG/SNAIL are downstream mediators of EGFR-stimulated re-epithelization (26). Thus, the study investigated whether the protein expression levels of several important factors in EGF-mediated EMT were associated with the knockdown of HECTD1 expression. The Ctrl HeLa cells expressed low levels of SNAIL, whereas the HECTD1-KD cells demonstrated high expression levels of SNAIL (Figs. 4D and S4A). Similar results were found for SLUG. The adhesion molecule, E-cadherin, is considered to be an important signaling marker for EMT (27); reduced E-cadherin expression levels were observed in the HECTD1-KD cells, which indicated that these cells had a lower cell adhesive ability. The expression level of vimentin remained unaltered, while that of N-cadherin was found to be slightly decreased. Overall, these data suggested that the knockdown of HECTD1 may potentiate EMT.

In cervical cancer cells, EGF has been reported to induce EMT through the upregulation of the expression of SNAIL (28) and via the activation of the AKT and ERK signaling pathways (29). Furthermore, hypoxia activates the EGFR signaling pathway to induce EMT (30). In this study, to determine the changes in the EGFR-signaling pathway following stimulation of cells with EGF, total ERK1/2 and phospho-ERK1/2 protein expression levels were investigated by western blot analysis at different time points (Fig. 5). Compared with the Ctrl cells, the expression levels of total ERK1/2 were not markedly altered in the HECTD1-KD cells (Figs. 5A and S4B). However, the phosphorylation of ERK1/2 progressively decreased following a 30-min incubation period with EGF in the Ctrl cells, whereas the phospho-ERK1/2 expression levels remained high in the HECTD1-KD cells at 60 min following EGF stimulation (Fig. 5A and B). In addition, the knockdown of HECTD1 increased cell motility (Fig. 4B and C) and EGFR signaling was observed to be associated with an increased phospho-ERK1/2 expression (Fig. 5A), suggesting that the knockdown of HECTD1 and the EGF signaling pathway may synergize to promote EMT in these cells. It was also particularly intriguing to observe the reduction of total ERK2 protein levels in the HECTD1-KD cells.

Upon screening for factors that regulated the expression levels of Hectd1 in mouse embryonic fibroblasts cells, Hectd1 expression was observed to be downregulated during hypoxic and heat treatment (11). HECTD1 is a putative BH3-only protein under hypoxic conditions (12); it has been reported that hypoxia is involved in the EMT processes (31), and that both miR-21 and miR-210 are also involved in this process (32,33). In this study, miRTarBase was used to identify hypothetical target sequences for HECTD1. Six targeting sequences of mir-21 and mir-210 were found in the 3'UTR of HECTD1. Thus, HECTD1 may be a target of miR-21 and miR-210 (34). By inducing hypoxic conditions in cells using CoCl₂, a factor involved in inducing EMT in cancer, the association between the expression levels of HECTD1 and those of miR-21 and miR-210 were investigated. The expression levels of HECTD1 progressively decreased under hypoxic conditions in HeLa cervical cancer cells (Fig. 6A); however, the expression levels of miR-21 and miR-210 were increased in the HeLa cells following CoCl₂ treatment (Fig. 6A).

The involvement of miR-21/-210 in the inhibition of HECTD1 expression was further investigated by trans-fecting the cells with their specific antagomirs (α-mir) (35). The presence of the mir-210 antagomir compared with the scramble (α-scr), followed by CoCl₂ treatment, resulted in significantly increased mRNA expression levels of HECTD1 (Fig. 6B; P<0.001). Similar results were obtained with miR-21 antagomirs, although to a lesser extent.

To determine whether the genetic abnormalities of HECTD1 expression may affect its function in both cervical cancer cells and in other types of cancer cells, 329 cervical cancer samples from The Catalogue Of Somatic Mutations In Cancer (COSMIC) (36) and 309 samples from The Cancer Genome Atlas (TGCA) were analyzed using whole-exome sequencing (37). A total of 24 mutations were identified throughout the HECTD1 gene, including one frameshift, twenty missense and three silent mutations (Fig. 7A, and Tables SI and SII). Although all these variants were spread over the entire gene, and they occurred most commonly in those areas which encoded the evolutionally conserved ANK-repeats, the heavily phosphorylated area and the HECT-domain (Fig. 7A). Compared with other types of cancer tissue, such as ovarian cancer, these genetic abnormalities were exclusively observed in small cell carcinoma, squamous cell carcinoma and endo-cervical adenocarcinoma (Fig. 7A, and Tables SI and SIII).

Using tissue microarrays, the association between the expression of HECTD1 and SNAIL in cervical cancer samples was investigated. A significant inverse association was observed between the nuclear expression of HECTD1 and SNAIL; all 4 samples displayed a low HECTD1 expression in the presence of a high SNAIL expression (Figs. 7B and S5).

Moreover, Kaplan-Meier analysis of squamous cell carcinoma using the HPA (38) revealed that markedly higher HECTD1 expression levels were associated with a longer relapse-free survival (P=4.86x10^-2; Fig. 7C). By contrast, higher SNAIL expression levels were associated with a shorter relapse-free survival (P=2.25x10^-2). The association between HECTD1 expression in cancer tissue samples and the survival rate in a number of large public clinical databases was also investigated (39). In breast, gastric, lung and kidney cancer, lower expression levels of HECTD1 were significantly associated with shorter survival times for patients with cancer (Fig. S6). Furthermore, consistent with the role of miR-210 in regulating HECTD1 expression (Fig. 6), higher expression levels of miR-210 were associated with lower expression levels of HECTD1 and a shorter patient survival (Fig. S7), which is similar to results obtained in cervical cancer, where it was reported that the upregulation of miR-210 was associated with a poorer prognosis (40).