Gain of UBE2D1 facilitates hepatocellular carcinoma progression and is associated with DNA damage caused by continuous IL-6

LO2, QSG -7701, SMMC -7721, LM3, Hep3B and Huh7 cells were achieved from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China) and cultured in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum at 37 °C, 5% CO2. Where indicated, cells were treated with IL-6 (PeproTech, Rocky Hill, NJ) for the indicated time. Without indications, cells were treated with IL-6 at 1 ng/ml for 48 h. IL-6-LT cells were treated with 1 ng/ml IL-6 for more than two months with IL-6 replenishment every 2 days.

Total cell and tissues protein extracts were separated by SDS-PAGE, transferred to nitrocellulose membrane and incubated with antibodies specific for UBE2D1(Abcam, Hong Kong, China), p53(Sigma-Aldrich, Saint Louis, MO), CHK1(Cell Signaling Technology, Boston, USA), phosphorylated CHK1(Cell Signaling Technology), RPA(Sigma-Aldrich), γ-H2Ax(Cell Signaling Technology), STAT3 and phosphorylated STAT3(Cell Signaling Technology), RAD51B(Abcam), cleaved PARP, Caspase 9 and Caspase 3 (Cell Signaling Technology) and β-actin (Sigma-Aldrich). IRdye800-conjugated goat anti-rabbit IgG and IRdye700-conjugated goat anti-mouse IgG were then incubated with the blots. Antibody binding was detected on Odyssey infrared scanner (Li-Cor Biosciences Inc.).

Immunohistochemistry of UBE2D1 was performed on 5 μm paraffin sections of tissue samples. UBE2D1 antibody (Abcam) was applied at 1:100 dilution. Vectastain Elite ABC kits (Vector, Burlingame, CA, USA) and ImmPACT DAB Substrate (Vector) were used to detect UBE2D1 expression. The staining sections were visualized in Olympus microscopy (IX71; Olympus, Japan).The score was evaluated as four levels:0 (−), 1(+), 2(++), 3(+++). The quantitative analyses of immunohistochemistry images were performed using Image Software Pro Plus 6.0.

To obtain the overexpression of UBE2D1, STAT3 and RAD51B, the indicated cells were transfected with recombinant lentiviruses. Recombinant lentiviruses containing UBE2D1, STAT3 and RAD51B with negative control were purchased from Obio Technology (Shanghai, China). For transfection of UBE2D1, STAT3 and RAD51B into indicated cells, 1 × 10⁶ transducing units were added to the medium of cultured cells, and supernatant was replaced with complete culture media after 24 h. The overexpression of genes was confirmed by real-time PCR and Western blot with specific primers or antibodies 96 h after infection. For construction of stably overexpressed UBE2D1, STAT3 and RAD51B, the cells were transfected with 1 × 10⁶ transducing units’ lentivirus, and the transfected cells were selected by using puromycin at the concentration of 0.5 μg/ml to 1.0 μg/ml for indicated cells. The overexpression of target genes was confirmed by real-time PCR and Western blot.

UBE2D1 and STAT3 interference was mediated by siRNA. RAD51B interference experiments were performed using shRNA vector (genechem, shanghai, China). For transfections, 75 nM siRNA or 4 μg shRNA and negative controls were transfected into indicated cells with Lipofectamine 3000 (Invitrogen). The oligonucleotide sequences of siRNAs were presented in Additional file 1: Table S3.

Cells were incubated in normal culture medium with 10 μl CCK-8 (Dojindo, Kumamoto, Japan) in 96-wellplate for 2 h. Then the cell numbers were estimated by value of optical density (OD) at 450 nm. The cell proliferation rates were determined at 0, 12, 24, 48, 60 and 72 h after treatment.

Indicated cells were cultured in 6-well plate with 500 cells per well. Then the plates were incubated at 37 °C with 5% CO₂ for 14 days with medium renewal every 4 days. After being fixed with 3.7% formaldehyde for 20 min, the cells were stained with 0.5% Crystal violet dye.

Cells for immunofluorescence analysis were prepared on 12 × 12 mm glass slides in 12-well plate. The cells were firstly incubated with specific antibodies of γ-H2Ax(Cell Signaling Technology), p53(Abcam), UBE2D1(Abcam) and then the goat anti -rabbit IgG (Alexa Fluor 594, Invitrogen; cy3,sangon) was used as the secondary antibody. The nuclei were stained by DAPI, and images were captured under the Olympus microscopy (IX71; Olympus, Japan).

For the detection of apoptosis, TUNEL staining was performed according to the manufacturer’s instruction (Biotool, Houston,TX,USA). Cells were cultured in slides in 12-well plate, and fixed with 4% paraformaldehyde solution. After permeabilizing slides in 0.2% Triton X-100 solution, cells were incubated with reaction mixture for 1 h. Nuclei were stained by DAPI, and images were captured under the Olympus microscopy (IX71; Olympus, Japan).

HCC tissues and paired adjacent noncancerous tissues(at least 3 cm away from tumor boarder) in Cohort, normal hepatic tissues(distal normal liver tissue of liver hemangioma patients), hepatitis liver tissues, precancerous lesions and paired tissues without lesions, serum samples were collected from Eastern Hepatobiliary Surgery Hospital, Shanghai, China (Additional file 1: Table S1 and Table S2). For precancerous lesions, we refers to high grade dysplastic nodules (HGDN), which is defined as nodules with high malignant potential and high grade atypia. The non-lesions tissues were collected from tissues at least 1 cm away from HGDN edge. The HGDN was diagnosed by two pathologists and met the criteria for HGDN reported previously [23]. All patients provided written informed consent, and the experiments were approved by ethics committee of Second Military Medical University.

All animal experiments were approved by Institutional Animal CareandUse Committee(IACUC) of Second Military Medical University, Shanghai, China. Animals were maintained under standard conditions, in accordance with institutional animal care guidelines. HCC cells with UBE2D1 overexpression and control or continuous IL-6 stimulation were injected subcutaneously into the bilateral armpit of nude mice, and tumors were collected after 3–4 weeks to evaluate the tumor growth. As cells with UBE2D1 overexpression or continuous IL-6 stimulation and control cells were injected subcutaneously into the left and right armpits respectively of the same nude mice, the Wilcoxon signed-rank test for statistical testing was adopted.

To elucidate the roles of UBE2D1 in HCC progression, we detected its expression in HCC tissues. We found that UBE2D1 was significantly upregulated in HCC tissues compared with corresponding noncancerous hepatic tissues in a set of HCC Cohort (n = 108) and Cohort from online database(GSE14520, n = 209) (Fig. 1a and Additional file 1: Figure S1A). Then we examined the UBE2D1 expression in the precancerous lesions and paired adjacent tissues without lesions to find that precancerous lesions had a higher UBE2D1 level than adjacent normal tissues (Fig. 1b), indicating the pro-tumor role of UBE2D1 at the early stage of HCC. To further confirm the increase of UBE2D1 in hepatocellular carcinogenesis, we also measured its transcript level in normal hepatic tissues and hepatitis liver tissues without HCC to find that the upregulation of UBE2D1 was only in cancer cells but not in hepatitis cells (Fig. 1c), indicating UBE2D1 as a potential early HCC biomarker. Next we confirmed the upregulation of UBE2D1 on the protein level by western blotting and immunohistochemistry (IHC) in HCC (Fig. 1d and Additional file 1: Figure S1B).

To further explore whether UBE2D1 expression level was associated with the clinical outcome of HCC patients, we performed Kaplan-Meier analysis to reveal that the high expression level of UBE2D1 in tumor tissues was correlated with reduced overall survival of the HCC patients in GSE14520 (Fig. 1e). These results demonstrated that UBE2D1 was significantly overexpressed in HCC and premalignant tissues and was a prognostic predictor for HCC patients.

We next evaluated the functional roles of UBE2D1 in HCC growth in vitro and in vivo. Firstly, we detected the UBE2D1 abundance in HCC cell lines and normal hepatic cell lines to find that hepatic cells had a lower UBE2D1 level than most of HCC cells (Additional file 1: Figure S2A). Blotting analysis of proteins from nuclear and cytoplasmic fractions revealed that UBE2D1 mainly located in the cytoplasm (Additional file 1: Figure S2B). To determine the role of UBE2D1 in HCC pathogenesis, we conducted the lentivirus-based overexpression and small interfering RNAs (siRNAs)-mediated knockdown (Fig. 2a). We conducted the clone formation assays in HCC cells line LM3 and hepatic cells line QSG-7701 to find that overexpression of UBE2D1 enhanced the clone formation potential, while UBE2D1 knockdown reduced the clone numbers (Fig. 2b). Cell proliferation evaluated by CCK-8 assays was decreased in UBE2D1-silent LM3 and QSG-7701 cells and increased in UBE2D1 overexpressed cells (Fig. 2c and Additional file 1: Figure S2C). Additionally, knockdown of UBE2D1 increased the markers of apoptosis, including cleaved PARP, caspase 3 and caspase 9, while upregulation of UBE2D1 suppressed the apoptosis (Fig. 2d). And the Annexin V-FITC staining confirmed the apoptosis-suppressing role of UBE2D1(Additional file 1: Figure S2D). Moreover, UBE2D1 significantly increased the expression of marker of stem cell-like phenotype and decreased the expression of hepatic specific genes(Additional file 1: Figure S2E). Finally, we explored the role of UBE2D1 in HCC carcinogenesis in vivo by transplanting the UBE2D1 stably-overexpressed cells into the bilateral armpit of nude mice. We found that UBE2D1 could promote the tumor growth in vivo (Fig. 2e).

Since UBE2D1, as an ubiquitin-conjugating enzyme, was reported to be associated with tumor-suppressor protein p53, a key regulator in HCC carcinogenesis, we confirmed the regulation of UBE2D1 on p53 in QSG-7701, LM3 and SMMC-7721 cells. Firstly, we measured the basal expression level of p53 in these three cell lines(Additional file 1: Figure S3A). Next, We found that p53 accumulated following UBE2D1 knockdown, while overexpression of UBE2D1 reduced the p53 protein level (Fig. 3a, Additional file 1: Figure S3B and S3C), Immunofluorescent staining of p53 confirmed the negative regulation of UBE2D1 on p53(Additional file 1: Figure S3D). Meanwhile, p53 transcript level was not affected by UBE2D1 (data not shown). We then treated the UBE2D1 downexpressed and control cells with the proteasome inhibitor MG-132 to block the protein degradation. As a result, MG132 abolished the upregulation of p53 caused by UBE2D1 knockdown in western blot and immunofluorescent staining (Fig. 3b, Additional file 1: Figure S3E and S3F). To further confirm the ubiquitination-dependent degradation of p53 by UBE2D1, we performed the immune-precipitated assays with MDM2 specific antibody in UBE2D1-silent QSG-7701 cells. Results revealed that UBE2D1 knockdown decreased p53-MDM2 interaction(Fig. 3c).

Then we further investigated whether UBE2D1 promoted cell growth through p53 inhibition. As hepatic cell line QSG-7701 and cancer cell line LM3 are p53 wild-type, we examined the cellular functions and mechanisms in other p53 wild-type hepatic cell line L02 and cancer cell line SMMC-7721, p53 mutant-type cancer cell line Huh7 and p53-deficient HCC cell line Hep3B [24]. Results showed that UBE2D1 could promote cell proliferation and repress cell apoptosis in p53 wild-type cell lines, but not in p53 mutated and deficiency cells(Fig. 3d-f and Additional file 1: Figure S3G). Altogether, these results showed that UBE2D1 triggered the degradation of p53 to promote HCC growth.

As copy number variations is the crucial event early in hepatocarcinogenesis, and controls the gene expression to initiate cancers [25], we investigated whether the upregulation of UBE2D1 in HCC was attributed to copy number variations. By detecting the genomic levels of UBE2D1 in the same set of HCC Cohort, we found that the genome content of UBE2D1 was significantly amplified in HCC and was correlated with its transcriptional level (Fig. 4a, b). Additionally, the HCC tissues had a higher genomic level of UBE2D1 than normal liver tissues or hepatitis liver tissues (Fig. 4c), corresponding to the transcript level. Correlation analysis of UBE2D1 genome content with patients’ information showed that genomic level of UBE2D1 was associated with the gender(Additional file 1: Table S4). Compared with female patients, males had a higher UBE2D1 genomic level both in HCC tissues and adjacent nontumor tissues (Fig. 4d). Gender disparity is a Characteristic for HCC [2], and males are about three to five times more likely to develop HCC than females [26]. The results indicated that the copy number variations of UBE2D1 was associated with gender in HCC, and may be a reason for the gender disparity. Interestingly, we found that the genomic level of UBE2D1 was also associated with the serum IL-6 level of HCC patients, and patients with high serum IL-6 concentration had a higher UBE2D1 genomic level (Fig. 4e). Previous studies have confirmed the crucial roles of IL-6 in the gender disparity of HCC [27, 28]. So we hypothesized that the gender difference of UBE2D1 genome was attributed to the IL-6. To verify this hypothesis, we treated the SMMC-7721 cells with continuous IL-6 for long time to evaluate the genomic alterations of UBE2D1. We found that the genomic content was amplified after continuous IL-6 stimulation for more than 4 months (Fig. 4f). Taken together, the results showed that copy number gain contributed to UBE2D1 overexpression and was associated with the high IL-6 level.

Next we explored the mechanisms of genomic alterations induced by continuous IL-6 stimulation. As activated DNA damage response and genomic instability induced by exogenous stimulation was the primary reason for genomic alterations in cancer, we detected the effects of IL-6 on DNA damage and genomic instability. IL-6 activated the DNA damage response marked by phosphorylated CHK1 and RPA [29] (Fig. 5a and Additional file 1: Figure. S4A). And continuous IL-6 maintained the activated DNA damage marker and significantly enhanced the genomic instability marked by γ-H2Ax (Fig. 5b). Moreover, continuous IL-6 enhanced the UBE2D1 protein levels and inhibited the p53 expression, but not short-term IL-6 stimulation (Fig. 5a and Additional file 1: Figure S4B). Immunofluorescence of γ-H2Ax foci confirmed the increasing of genomic instability under IL-6 (Fig. 5c). As STAT3 activation is the main event in IL-6 signaling pathway [30], we next performed the lentivirus-mediated STAT3 overexpression to find that STAT3 overexpressed enhanced the DNA damage and genomic instability as continuous IL-6 (Fig. 5d), and interfering STAT3 in continuous IL-6 treated cells abolished the increase of DNA damage and genomic instability caused by IL-6(Fig. 5e). Interestingly, in continuous exogenous IL-6 treated cells, the IL-6 expression was significantly increased, indicating an autocrine IL-6 loop involved in the process (Additional file 1: Figure S4C).

To further explore the underlying DNA damage checkpoint protein regulated by IL-6 and STAT3, we overlapped the STAT3 ChIP-seq data [31, 32] with genes associated with DNA damage repair summarized from previous study. Next we detected these candidate genes in continuous IL-6 treated cells to find that RAD51B was significantly repressed in both QSG-7701 and SMMC-7721 cells with continuous IL-6 stimulation (Fig. 5f and Additional file 1: Figure S4D). And continuous IL-6 could provoke a gradual decline of RAD51B (Additional file 1: Figure S4E). The immunoblotting confirmed the suppression of RAD51B by IL-6 and STAT3 on protein level (Fig. 5g). We then confirmed the genomic binding of STAT3 in the promoter of RAD51B using ChIP with phosphorylated STAT3 specific antibody (Fig. 5h). And the suppressing effect of STAT3 on Rad51b promoter was further confirmed by Rad51b reporter assay (Additional file 1: Figure S4F). Moreover, interfering STAT3 in continuous IL-6 treated cells can abolish the decrease of RAD51B caused by IL-6 (Fig. 5i and Additional file 1: Figure S4G). Next, we detected the functions of RAD51B on DNA damage and genomic instability to find that overexpression of RAD51B could inhibit DNA damage and genomic instability, while RAD51B knockdown enhanced the response (Fig. 5j and Additional file 1: Figure S4H). Restoring the RAD51B expression could rescue the DNA damage response and genomic instability(Fig. 5k), indicating that continuous IL-6 activated the DNA damage and genomic instability through RAD51B. Next, we found that Rad51b stably underexpressed cells had increased UBE2D1 protein levels and decreased p53 expression (Additional file 1: Figure S4I). And long-term Rad51b silent could increase the UBE2D1 genomic content (Additional file 1: Figure S4J).

Next, we explored the relationship between continuous IL-6 and UBE2D1 copy number gain in HCC carcinogenesis. As continuous IL-6 played a different role from short-term stimulation in cancer [33], we first explored the effects of continuous IL-6 in HCC. We found that continuous IL-6 significantly facilitated the cell proliferation (Fig. 6a and Additional file 1: Figure S5A), clone formation potential (Fig. 6b and Additional file 1: Figure S5B) and tumor growth in vivo(Additional file 1: Fig. S5C). Additionally, interfering the STAT3 or restoring the RAD51B expression in continuous IL-6 treated cells would rescue the cell proliferation rate and clone formation potential (Fig. 6c, d).

As continuous IL-6 could increase the genomic content of UBE2D1 in part of HCC cells, we next explored the significance of UBE2D1 amplification under IL-6. Results showed that cells harboring UBE2D1 amplification under IL-6 enhanced the UBE2D1 protein level, while decreased p53 protein level, and exhibited reduced cell apoptosis and senescence phenotype than cells without UBE2D1 genomic alterations under IL-6 (Fig. 6e, f and Additional file 1: Figure S5D-F), indicating that UBE2D1 amplification enhanced the HCC promoting effects of continuous IL-6 stimulation.

Finally, we investigated the clinical significance of IL-6-RAD51B-UBE2D1 axis in HCC. By detecting the expression of RAD51B in the HCC cohort and precancerous lesions, we found that RAD51B was downregulated in both tumor tissues and precancerous lesions (Fig. 6g, h). Moreover, the transcript level of RAD51B was also associated with the serum IL-6 level of HCC patients, and patients with high serum IL-6 concentration had a lower RAD51B level (Fig. 6i). Low RAD51B expression in adjacent nontumor tissues suggested a poor recurrence-free survival of HCC patients (Fig. 6j). And STAT3 transcript level was negatively correlated with RAD51B in GSE14520(Additional file 1: Figure S5G).