UBE2J2 promotes hepatocellular carcinoma cell epithelial-mesenchymal transition and invasion in vitro

Ubiquitin-conjugating enzyme E2 J2 (UBE2J2) is an ubiquitin proteasome component that responds to proteotoxic stress. We found that UBE2J2 was highly expressed in cellular protrusions of HCCLM3 metastatic hepatocellular carcinoma (HC) cells. Immunohistochemical analyses showed that UBE2J2 was expressed at higher levels in HC patient tissues than in corresponding non-tumor tissues. Because cellular protrusions are important for cell invasion, we hypothesized that UBE2J2 promotes HC cell invasion. We used chip-based surface plasmon resonance (SPR) to assess possible mechanisms of UBE2J2-regulated HCCLM3 cell invasion. We found that p-EGFR interacted with UBE2J2, and this finding was confirmed by co-immunoprecipitation analysis. UBE2J2 overexpression activated endothelial-mesenchymal transition in the non-invasive SMMC7721 HC cell line, and promoted invasion. UBE2J2 silencing reduced HCCLM3 cell invasion and endocytosis, and downregulated p-EGFR expression. p-EGFR inhibition by lapatinib reduced UBE2J2-promoted cell invasion, suggesting p-EGFR is important for UBE2J2-mediated HCCLM3 cell invasion. These findings demonstrate that endocytosis by HC cells is closely related to invasion, and may provide new anti-HC therapeutic targets. UBE2J2 may also be a novel biomarker for clinical HC diagnosis.


INTRODUCTION
Hepatocellular carcinoma (HC) is one of the most common cancers worldwide [1,2]. Metastasis is the primary cause of cancer patient death [3], and HC diagnostic and therapeutic outcomes remain unsatisfactory [2]. Additional studies of the molecular mechanisms underlying HC cell metastasis, and new anti-HC therapeutic targets are urgently needed [2].
Cell invasion is closely related to tumor metastasis [4]. Before invasion, most malignant tumor cells undergo epithelial-mesenchymal transition (EMT) to detach from surrounding tissue [5,6]. Biomarkers, including E-cadherin, snail, slug, vimentin, CLDN-1, and N-cadherin, can be used to detect EMT or MET (reverse process of EMT). Cellular protrusions can drive cancer cells to dissociate from their surroundings and penetrate into the vasculature [7]. Highly dynamic actins assembled by the Arp2/3 complex and formins, along with dynamin and cortactin, both membrane trafficking machinery components, reportedly accumulate in finger-like protrusions [8][9][10]. Direct RNA sequencing (DRS) analysis of hepatocellular carcinoma cell protrusions will enhance our understanding of liver tumor cell metastasis and invasion.
This study found that UBE2J2 was highly expressed in HCCLM3 cell protrusions. We investigated mechanisms of UBE2J2-regulated HCCLM3 cell invasion and the relationship between endocytosis and cell invasion, providing new insights into HC metastasis and invasion regulation.

UBE2J2 is highly expressed in HCCLM3 cell protrusions
We isolated cellular protrusions and cell bodies from HCCLM3 (a highly metastatic HC cell line) and SMMC7721 cells (a low metastatic HC cell line) via Boyden chamber isolation assay as described previously [19] (Figure 1). More than 7500 genes were sequenced using DRS and listed according their expression ratios in HPs (HCCLM3 protrusions) versus HBs (HCCLM3 cell bodies). Table 1 lists the 50 genes with the highest HP/ HB expression ratios. PYCARD is related to inflammatory and apoptotic signaling pathways [20], and STXBP2, ATP6V0D1, and UBE2J2 are responsible for endocytosis [21][22][23][24]. Our DRS results suggest that endocytosis is related to HC cell metastasis and invasion.
Protein lysates were analyzed using western blotting (Figure 2A-2B). UBE2J2 was more highly expressed in HCCLM3 cell protrusions than in SMMC7721 cells and cell bodies. RT-qPCR results revealed that UBE2J2 transcription rates were greater in HCCLM3 protrusions than in cell bodies and SMMC7721 cells ( Figure 2C).

UBE2J2 knockdown decreases HCCLM3 cell invasion
UBE2J2 was silenced using specific siRNAs in HCCLM3 cells, and cell invasion was measured using the Boyden chamber assay. The number of migrated cells was lower in UBE2J2-silenced cells than in controls (Figure 4), indicating that UBE2J2 regulates HCCLM3 cell invasion.

UBE2J2 binds p-EGFR
Chip-based surface plasmon resonance (SPR) was used to screen for potential UBE2J2-associating proteins. SPR biosensors monitor refractive index (measured in resonance units, RU) alterations caused by binding of analytes to ligands immobilized on sensor chips [26]. Anti-UBE2J2 antibody was immobilized on CM5 sensor chips. RU values increased from 2300 to 3800 when cell-extracted proteins bound the immobilized antibody ( Figure 7A). An RU increase from 3800 to 5100 indicated that p-EGFR bound UBE2J2 ( Figure 7A-7B). Co-IP analysis results further confirmed that p-EGFR binds UBE2J2 ( Figure 7C-7E). Abbreviations: HP: HCCLM3 protrusions; HB: HCCLM3 cell bodies; SP: SMMC7721 protrusions; SB: SMMC7721 cell bodies; "/": gene could not be detected; "-": little report of the given gene with respect to cell invasion; "√": reports of the given gene with respect to cell invasion.

Figure 1: Cellular protrusion and cell body isolation. HCCLM3 and SMMC7721 cells were seeded on invasion inserts with 1μm
(diameter) pore membranes, allowing only protrusions to pass through. Twelve h after cells reached confluence, cellular protrusions and cell bodies were isolated using cell scrapers, and total RNA and proteins were collected.

UBE2J2 and p-EGFR were enriched in HC tissues
Serial sections of metastasized HC tissues were stained using anti-UBE2J2 and anti-p-EGFR antibodies ( Figure 8A). UBE2J2-positive and p-EGFR-positive staining was observed in 86.7% (13/15) and 73.3% (11/15) of samples, respectively ( Figure 8B, P<0.01), indicating that p-EGFR and UBE2J2 were highly expressed in HC tissues. The highest-scoring UBE2J2 and p-EGFR positive staining locations are indicated by white arrows ( Figure 8A).

Lapatinib attenuates UBE2J2-mediated cell invasion
Lapatinib (10 nM), a p-EGFR inhibitor [27], was used to explore the role of p-EGFR in UBE2J2-regulated cell invasion. Lapatinib-mediated p-EGFR suppression attenuated the effects of UBE2J2 on cell invasion ( Figure 10A-10B). Western blotting confirmed that Lapatinib suppressed p-EGFR, resulting in E-cadherin downregulation ( Figure 9C). p-EGFR inhibition also reduced the suppression on E-cadherin by UBE2J2 ( Figure 10C). These results together suggest that UBE2J2regulated HCCLM3 cell invasion depends on p-EGFR.

DISCUSSION
Cellular protrusions are highly dynamic structures involved in cell invasion [9]. To investigate how cellular protrusions promote invasion, mRNA from HCCLM3 cell protrusions and cell bodies was analyzed by direct RNA sequencing (DRS). Of the first four genes in these results not yet associated with cell invasion, three (ATP6VOD1, STXBP2, and UBE2J2) are important for endocytosis [21][22][23][24].
Western blotting and RT-qPCR analyses showed that UBE2J2 was highly expressed in HCCLM3 cell protrusions. STXBP2 and ATP6V0D1 were also highly expressed (data not shown here). IHC analyses showed UBE2J2-positive staining in most HC tissues compared with corresponding non-tumor tissues (Figure 3), indicating that UBE2J2 might be a useful biomarker for HC diagnosis. UBE2J2 was silenced in HCCLM3 cells using specific siRNA, and cell invasion was measured via Boyden chamber assay. Invasion was decreased in silenced cells, indicating that UBE2J2 regulates HCCLM3 cell invasion.
The EMT-MET switch is fundamental to tumor metastasis [37]. EMT allows cancer cells in the primary tumor site to break through the basement membrane and enter the bloodstream through intravasation [38]. Invasive tumor cells that survive this process usually undergo MET within the new environment. Because UBE2J2 promotes HCCLM3 cell invasion, we assessed whether or not UBE2J2 knockdown might induce MET in these cells. We found that numerous MET biomarkers, including β-catenin, CLDN-1,     N-cadherin, slug, snail, vimentin, ZO-1, MMP-9, were downregulated following UBE2J2 knockdown, and E-cadherin was upregulated. UBE2J2 overexpression in non-invasive SMMC7721 cells appeared to induce EMT and cell invasion, indicating that UBE2J2 regulates the EMT-MET switch.
To explore potential mechanisms of UBE2J2regulated HCCLM3 cell invasion, we screened for UBE2J2-interacting proteins using chip-based SPR. Numerous cell invasion-related protein antibodies were tested, but only p-EGFR bound UBE2J2. p-EGFR controls cell invasion via AKT and MMPs [39,40]. We assessed HC cell invasion and protein levels following UBE2J2 silencing or p-EGFR inhibition. While p-EGFR levels appeared dependent on UBE2J2 expression, p-EGFR inhibition reduced UBE2J2promoted HCCLM3 cell invasion. Our findings indicate that UBE2J2 binds p-EGFR to promote HCCLM3 cell invasion.
Transferrin is an endocytosis indicator [41]. We found that UBE2J2 knockdown suppressed transferrin endocytosis. Because UBE2J2 also promotes HC cell invasion, we hypothesize that endocytosis is closely related to invasion. Protein ubiquitin (UB) modification is an important aspect of endocytosis [42]. Proteins secreted by other cells are captured by cell membrane receptors, internalized via endocytosis, sorted, and degraded by proteases in the lysosome [12,42]. Our study confirmed that UBE2J2 positively regulates HC cell endocytosis. p-EGFR reportedly stabilizes snail and slug to trigger EMT and tumor metastasis [43]. We showed that UBE2J2 binds p-EGFR to promote HC cell invasion and EMT. The UBE2J2 hydrophobic carboxyl terminus anchors to the ER membrane and associates with ubiquitin ligases to degrade cargo proteins in the lysosome [44]. Based on information from the microenvironment, cells make adjustments, such as triggering EMT or MET, reorganizing actin and tubulin, reconstructing cytoskeletons, and migrating ( Figure 11).
In conclusion, the p-EGFR-UBE2J2 complex appears to promote HCCLM3 cell invasion and endocytosis. Our findings demonstrate that endocytosis in HC cells is closely related to invasion, and may provide new anti-HC therapeutic targets. UBE2J2 may also be a novel biomarker for clinical HC diagnosis.

Cell culture
Cell culture supplies were purchase from Life Technologies (Carlsbad, USA) and Corning (New York, USA). Human liver cancer cell lines, HCCLM3 and SMMC7721, were bought from the cell bank at the Chinese Academy of Sciences. Cells were grown in DMEM supplemented with 10% FBS, in an incubator with 5% CO 2 at 37°C.

Protrusion isolation and direct RNA sequencing (DRS)
Cellular protrusion and cell body mRNA was extracted as described [19]. Cells were seeded on invasion inserts with 1μm pore membranes (from BD Company). Twelve h after cells reached confluence, cellular protrusions were cut using a cell scraper, and mRNA was extracted using TRIzol. DRS was conducted via BGI RNA-Seq (Quantification) Analysis (BGI Tech, Shenzhen, China). The experiment was repeated three times and genes not detected at any time were excluded. More than 7,500 genes were analyzed and listed according to the ratio of expression in protrusions (HP) versus cell bodies (HB).

Cell invasion assay
Cell invasion assays were performed as described [45,46]. Invasion inserts with 8μm pore membranes (Corning, New York, USA) were coated with fibronectin (Sigma-Aldrich, Missouri, USA) as described [47]. Cells were pretreated with a specific or control siRNA for 6 h. Pretreated cells were seeded onto inserts to reach confluence in 12 h, and then cultured for another 24 h. After fixation with 4% formaldehyde, non-invading cells on the upper sides of membranes were removed by cotton swab. Invading cells were stained with 0.08% trypan blue for 15 min as described [46] and photographed using a bright-field light microscope. Cells were counted in five random fields using Image-Pro Plus 6.0 from Media Cybernetics (MD, USA).

Western blotting
Cells were lysed in RIPA buffer to extract whole-cell proteins [49]. Protein concentrations were measured using the DC Protein Assay Kit I (Bio-Rad, California, USA). Equal amounts of protein were subjected to 10% SDS polyacrylamide gel electrophoresis, and then transferred to PVDF membranes. The following primary antibodies were used: UBE2J2 (Millipore, Massachusetts, USA), EMT antibody kit, EGFR, and p-EGFR (Tyr1068) (Cell Signaling, Massachusetts, USA). Horseradish peroxidaseconjugated secondary antibody signals were detected and measured using the Luminescent Image Analyzer Tanon 5200 (Shanghai, China). Band densities were measured via ImageQuant software (Molecular Dynamics, Sunnyvale, CA, USA), and expressed as percentage of the β-actin band density.

UBE2J2 cloning and overexpression
UBE2J2 plasmid was constructed by cloning the UBE2J2 PCR amplicon with specific primers into the XbaI site in the pEGFP-N3 plasmid (Clontech). UBE2J2 forward cloning primer: 5'-GCTCTAGAGCATGACCC CTTATGAAGGTGG-3', and reverse: 5'-GCTCTAGA GCTCACTCCTGCGCGATGCT-3'. After ligation, the amplicon was transfected into DH5α competent cells followed by plasmid extraction and transient transfection of UBE2J2 into SMMC7721 cells using Lipofectamine 2000 as described [20]. After 24 h transfection, cells were subjected to Boyden chamber assay to assess cell invasion. Cell lysates were analyzed via western blotting with EMT biomarker antibodies to examine the effects of UBE2J2 on EMT.

Immunofluorescence staining
HCCLM3 cells were cultured in glass bottom dishes and transfected with UBE2J2 siRNA for 24 h using Lipofectamine 2000 as described [19]. After 3 h of serumfree starvation, cells were treated with transferrin (20ug/ ml) for 30 min. Cells were fixed with 4% formaldehyde for 15 min and permeabilized with 0.25% Triton X-100 [52]. Cells were incubated with p-EGFR primary and Dylight 488 Donkey Anti-Rabbit secondary antibodies as described [53,54]. Cells were examined using an Olympus confocal laser-scanning microscope and images were analyzed using IPP6.0 (Image-Pro Plus 6.0).

Statistical analysis
Data are expressed as means ± standard deviation (SD) of three independent experiments, and were analyzed with SPSS software using Student's t test. P<0.05 was considered statistically significant.