ID1 promotes hepatocellular carcinoma proliferation and confers chemoresistance to oxaliplatin by activating pentose phosphate pathway

MHCC97H (97H), a high-metastatic human HCC cell line, was established in Liver Cancer Institute of Zhongshan Hospital [15]. MHCC97H cells from the 12th to the 15th passage were used in our experiments. Hep3B, a low metastatic potential HCC cell line was purchased from the America Type Culture Collection(ATCC, HB 8064™). The oxaliplatin-resistant HCC cell lines MHCC97H–OXA and Hep3B–OXA selected at a 25umol/L concentration of oxaliplatin were successfully established from MHCC97H and Hep3B, by exposing cells to gradually increasing oxaliplatin (Sigma, St. Louis, MO, USA) from 2 umol/L to 25umol/L in our laboratory [14]. The IC50 value of surviving HCC cells treated with oxaliplatin was about 10-fold as high as that of their parental cells (Additional file 1: Figure S1).MHCC97H and MHCC97H–OXA were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, Gibco Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS, Life Technologies/Gibco). Hep3B and Hep3B–OXA cells were cultured in minimum essential medium (MEM) supplemented with 10% fetal bovine serum. Cells(1 × 10⁶) were seeded into 25 cm culture flask for 72 h per passage.

For gene knockdown, short hairpin RNA (shRNA) clones targeting ID1 were delivered by lentiviral infection.shRNA targeting sequences for ID1 genes are 5′-gatccAACTCGGAATCCGAAGTTGGATTCAAGAGATCCAACTTCGGATTCCGAGTTTTTTTTg-3′.5 × 10⁴ cells were seeded in 6-well plates 24 h before addition of lentivirus and 8 μg/mL polybrene to the growth medium, and then cultured for additional 72 h. Three days After transfection, ID1 knockdown was confirmed by Western blot analysis and qPCR (Fig. 1b).For rescue experiments, 2 μg G6PD plasmid was transfected into ID1-knockdown 97H–OXA and 3B–OXA cells for 48 h using Lipofectamine 2000 (Life Technologies, Carlsbad, CA, USA).

Human Cancer Pathway Finder™ PCR Array that profiles the expression of 84 genes representative of 9 different biological pathways (listed in Table 1) was purchased from SABiosciences (Qiagen Company, Milano, Italy). Total RNA isolated from 97H–Ctrol and 97H–shID1 cells was used for screening by real-time PCR according to the manufacturer’s instruction. Target genes whose expressions were differentially regulated (with at least 2-fold difference) were selected and validated using real-time PCR and Western blot.

Total RNA was isolated using Trizol reagent (Invitrogen, Carlsbad, CA) and reverse-transcribed to cDNA using PrimeScript RT reagent Kit (DRR037A, Takara, Japan). For real-time PCR (RT-PCR), SYBR Premix Ex Taq (DRR081, Takara, Japan) was used according to the manufacturer’s instructions. The qPCR primers used are provided in Additional file 2: Table S1. Western blots were performed as described previously [16]. Images were captured through Gel Doc XR system (Bio-Rad, Philadelphia, PA), and analyzed using Image LabTM software (version 2.0). Antibodies against ID1 and c-MYC were brought from Santa Cruz Biotechnology (Santa Cruz Biotechnology, CA, USA). Antibodies against G6PD, β-catenin, TCF-4 and Lamin B were brought from Abcam (AbCam, Cambridge, UK). Antibodies against Caspase-3 and Caspase-9, and Ki67, ERK, and pERK antibodies were obtained from Cell Signaling Technology (Cell Signaling Technology, Danvers, MA). Antibody against GAPDH was from Kangcheng Technology (Shanghai, China).

CCK-8 assay was carried out as previously described [17]. Briefly, 100 μL cells (2 × 10³/ml) were seeded in triplicated wells of a 96-well microplate. After 5-day culture, 10 μL CCK-8 solution (Beyotime, Haimen, China) was added to each well and incubated at 37 °C for 2 h. Absorbance values were expressed as percentages relative to the controls.

For colony formation ability assay, cells were plated in 60 mm² culture dishes at 2 × 10³ cells/well for 7–14 days. Colonies were stained with Giemsa (Beyotime, Haimen, China), and surviving colonies (a colony was defined as >50 cells) were counted.

Cell apoptosis was determined by flow cytometry using FITC Annexin V Apoptosis Detection Kit (BD Pharmingen, San Diego, CA, USA). Briefly, 1 × 10⁶ cells were washed with ice-cold phosphate-buffered saline (PBS) twice and resuspended in 200 μl 1 × binding buffer containing 2.5 μl Annexin V-PE for 15 min at room temperature in the dark. After incubation, the population of Annexin V-positive cells was evaluated using a FACS Aria cytometer (Becton cytomics FC500, Fremont, CA, USA).

The enzyme activity of G6PD was assessed using a G6PD assay kit (Sigma, St. Louis, MO, USA), according to the manufacturer’s instructions. The combined activity of G6PD and 6-phosphogluconate dehydrogenase (6PGD), the second enzyme of the PPP that also produces NADPH, was determined by the rate of conversion of NADP+ to NADPH in the presence of glucose-6-phosphate (G6P). G6PD activity was calculated by subtracting 6PGD from the combined activity. The reaction buffer contained 50 mM Tris (pH 8.1) and 1 mM MgCl₂, and the substrate concentration was G6P(200 uM), 6-phosphogluconate(6PG,200 μM) and NADP+ (100 μM). Enzyme activities were normalized based on the protein concentration, which was determined by a Bio-Rad protein assay kit (Bio-Rad, Richmond, CA, USA).

NADPH levels and NADP+/NADPH ratios were determined using the NADP+/NADPH Quantification kit (BioVision, Mountain View, CA, USA). Intracellular reactive oxygen species (ROS) generation was measured by using Cell ROX Deep Red reagent (Life Technologies Carlsbad, CA, USA) according to manufacturer’s protocol. Analysis and data acquisition were performed in Beckman cytomics FC500 by using CellQuest Pro (BD) analysis software (FACS analysis).

Coimunoprecipitation was performed according to the manufacture’s protocol (Cell Signaling, Frankfurt, Germany). Briefly, 5 × 10⁵ cells were lysed in lysis buffer (100 mM TRIS/HCl pH 7.5, 150 mM NaCl, 1 mM MgCl₂, 0.25% NP-40, 5 mM NaF, 1 mM Na₃VO₄, 10 μg/mL Pepstatin, 100 μM PMSF and 3 μg/mL Aprotinin) and incubated with Rabbit serum (DAKO, Glostrup, Denmark)) for 1 h on ice and Agarose A/G-protein bead (Santa Cruz Biotechnology, CA, USA) for 30 min. The supernatant was immunoprecipitated with ID1 antibody (1:500) overnight at 4 °C. Immune complexes were precipitated with Agarose A/G-protein bead for 2 h at 4 °C min and analyzed by immunoblotting. Comparable results were obtained in at least two-independent experiments.

Cells growing to 70% confluence were transfected in triplicate with pGL3-G6PD-luc, pcDNA c-MYC or pGL3-Basic (Promega, Madison, WI, USA), along with pRL-TK. After 48-h transfection, cells were collected and the luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI) according to the manufacturer’s protocol. The luciferase activity of pRL-TK served as internal control.

Male athymic BALB/c nude mice (Shanghai Institute of Material Medicine of the Chinese Academy of Sciences) were raised in specific pathogen-free conditions. Animal care and experimental protocols were conducted in accordance with guidelines established by the Shanghai Medical Experimental Animal Care Commission. Cells (5 × 10⁶ cells per mouse) were injected subcutaneously into the upper left flank region of each mouse to produce tumors. Seven days later, mice were treated with 0.1 ml oxaliplatin (10 mg/kg/week) via tail vein injection for 4 weeks. The excised tumors were fixed with 4% formaldehyde and embedded in paraffin. The expression of Ki-67, Caspase-3 and Caspase-9 was detected using immunohistochemistry, and apoptotic cells in tumors were detected by TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) according to the manufacturer’s instructions (Boyetime, Shanghai, China). The percentage of TUNEL-positive cells to the total cell number was recorded as an apoptotic index.

Jaspar (http://​jaspardev.​genereg.​net), an open-access database of the transcription factors binding preferences in multiple species, was used to predict potential transcription factor binding sites. Survival analysis data were obtained from TCGA database using the UCSC Cancer Genomics Browser (https://​genome-cancer.​ucsc.​edu). A total of 371 HCC samples were collected in the dataset. The relationship between ID1/G6PD and the prognosis of HCC patients was explored.

It was found in our previous study that oxaliplaitin-resistant HCC cells acquired stem cell-like traits, and gene profile analysis revealed 267 up-regulated and 65 down-regulated genes in oxaliplatin-treated HCC tumors, compared with GS-treated HCC tumors [14]. These differentially expressed genes were related to chemokines, chemokine receptors, signal transduction, inflammation, proliferation, development, and metabolism. Among them, ID1 was highly expressed in oxaliplatin-treated HCC tumors. Therefore, we hypothesized that ID1 was probably activated in chemoresistant HCC cells. To confirm this finding, the expression of ID1 in oxaliplaitin-resistant HCC cells was examined. As expected, the expression level of ID1 in 97H–OXA and Hep3B–OXA cells was significantly higher than that in their parental cells (Fig. 1a). We further established 97H–OXA-shID1 and 3B–OXA-shID1 cells transfected with ID1shRNA to down-regulate endogenous ID1 expression (Fig. 1b). Colony formation ability assay (Fig. 1c) and CCK-8 assay analysis (Fig. 1d) revealed that ID1 knockdown inhibited the proliferation of HCC cells markedly in the absence of oxaliplaitin (P < 0.05). In addition, silencing ID1 expression induced the apoptosis MHCC97H–OXA and 3B–OXA cells (Fig. 1e). We then investigated whether ID1 was also involved in HCC chemoresistance. After 48-h treatment with a series of diluted concentrations of oxaliplatin, the half-maximal inhibitory concentration (IC50) of oxaliplaitin in ID1 knockdown 97H–OXA-shID1 and 3B–OXA-shID1 cells was significantly lower than that in control cells (Fig. 1f). These data demonstrated that ID1 was involved in the regulation of proliferation, apoptosis and chemoresistance in oxaliplaitin-resistant HCC cells.

Based on the vitro findings described above, we examined the impact of ID1 on HCC malignant proliferation, apoptosis and chemoresistance in vivo. Xenografts in nude mice were established by subcutaneous injection of 97H–OXA-Ctrol cells and 97 L–OXA-shID1 cells into nude mice as described in the Methods section. Seven days after implantation, nude mice were treated with 0.1 ml oxaliplatin (10 mg/kg/week) via tail vein injection for 4 weeks. In all cases, the mean size of tumors formed by 97H–OXA-Ctrol group was significantly larger than that in 97H–OXA-shID1 group (P < 0.05) (Fig. 2a and b). In addition, the final mean tumor weight in 97H–OXA-Ctrol group was significantly higher than that in 97H–OXA-shID1 group (P = 0.002) (Fig. 2c), indicating that ID1 knockdown inhibited tumor proliferation of 97H–OXA cells in vivo. We then examined in vivo apoptosis of the xenograft, using TUNEL staining. Conversely with the tumor size, tumors derived from 97H–OXA-shID1 cells displayed a higher percentage of TUNEL-stained cells compared with the tumor cells (Fig. 2d and e). Furthermore, we detected the expression levels of Caspase-3, Caspase-9 and Ki-67 by immunohistochemistry and found that the expressions of Caspase-3 and Caspase-9 were significantly increased in 97H–OXA-shID1 tumors compared with 97H–OXA-Ctrol tumors (Fig. 2f). In addition, the expression of Ki-67 in 97H–OXA-shID1 group was significantly decreased compared with that in 97H–OXA-Ctrol group. These results indicate that knockdown of ID1 expression inhibited cell proliferation, promoted cell apoptosis, and chemosensitized oxaliplaitin-resistant HCC cells to oxaliplaitin in vivo.

To quantitatively examine the mechanism of ID1 on cell proliferation and oxaliplaitin resistance, 84 genes related to cell proliferation, apoptosis, cell cycle, angiogenesis, invasion, and metastasis were evaluated via Human Cancer Pathway Finder™ PCR Array in 97H–Ctrol cell line and 97H–shID1 cell line. After knockdown of ID1, 12 genes exhibited significant changes in mRNA expression relative to control (expression ratio showing greater than 2.0-fold or less than 0.5-fold difference compared with the control group). Among the differentially expressed genes, 7 genes were up-regulated and 5 genes were down-regulated (Table 1 and Fig. 3a). Further real-time PCR and Western blot verified that ID1 knockdown significantly decreased the expression of G6PD, a key rate-limiting enzyme in pentose phosphate pathway (Fig. 3b and c).

Knowing that the key enzyme of pentose phosphate pathway G6PD may be a downstream target of ID1, we wondered whether ID1 induced-oxaliplatin resistance was mediated by the activation of pentose phosphate pathway. Thus, the enzyme activity of G6PD, the metabolic product of pentose phosphate pathway NADPH and the levels of intracellular reactive oxygen species (ROS) were further assayed in 97H–OXA-Ctrol cells and 97H–OXA-shID1cells. Compared with 97H–OXA-shID1 cells, the 97H–OXA-Ctrol cells exhibited a higher activity of G6PD (P = 0.003) (Fig. 3d) with a significant increase in NADPH levels (P = 0.008) (Fig. 3e) and a significant decrease in intracellular ROS level (P = 0.005) (Fig. 3f). The same results were also obtained in 3B–OXA-Ctrol cell line and 3B–OXA-shID1 cell line (Fig. 3d-f). Furthermore, after transfecting G6PD plasmid into ID1-knockdown 97H–OXA-shID1 or 3B–OXA-shID1 cells, the enzyme activity of G6PD was recovered (Fig. 3d) with an elevated NADPH level (Fig. 3e) and a decreased intracellular ROS level (Fig. 3f). More importantly, the recovery of G6PD expression seemed to induce oxaliplatin chemoresistance (P = 0.005 and P = 0.001) (Fig. 4a) in 97H–OXA-shID1 or 3B–OXA-shID1 cells via promoting cell proliferation (Fig. 4b) and inhibit cell apoptosis (Fig. 4c). These results indicate that ID1 promoted malignant proliferation and conferred oxaliplatin chemoresistance to HCC cells by activating pentose phosphate pathway through its downstream target G6PD.

Knowing that G6PD mRNA expression in ID1-knockdown cells was reduced significantly as compared with that in control cells (Fig. 3c), suggesting that ID1 regulated G6PD at the transcriptional level. Next, by using dual luciferase assay, we investigated whether ID1 could regulate G6PD transcriptional activity. As shown in Fig. 5a, knockdown of ID1 expressions in 97H–OXA and 3B–OXA cell lines inhibited the ID1 G6PD promoter activity (P < 0.05). Due to the lack of the basic DNA binding region, ID1 protein usually interacts with the basic-helix-loop-helix (bHLH) transcription factor by forming heterodimers in regulating its downstream gene expression. To decipher the molecular mechanism by which ID1 regulates G6PD, we screened for bHLH transcription factors (Additional file 3: Table S2) that could possibly bind G6PD promoter by using bioinformatic analysis (http://​jaspardev.​genereg.​net/​). The result showed that the promoter sequence of G6PD harbored 7 potential c-MYC binding sites (Fig. 5b), indicating that c-MYC may be involved in ID1-induced G6PD promoter activation. Further luciferase reporter assay confirmed that after transfecting c-MYC plasmid, G6PD promoter activity was restored devoid of ID1 in 97H–OXA-shID1 and 3B–OXA-shID1 cells (Fig. 5a). This result indicates that the activating effect of ID1 on the G6PD promoter was dependent on the endogenous expression of c-MYC. To experimentally confirm whether c-MYC directly targets G6PD promoter, we transfected different lengths of G6PD promoter reporter vectors with or without c-MYC plasmids in 293 T cells. The constructs used for reporter assays are illustrated in Fig. 5C. Luciferase reporter assay showed that G6PD promoter activity was low without transfecting c-MYC plasmid (Fig. 5d). After co-transfecting c-MYC plasmid, a remarkable increase in luciferase activity was observed from full-length G6PD promoter reporter constructs, but not from the reporter vector containing −1500 bps to +1 bps promoter fragment or −1000 bps to +1 bps promoter fragment, suggesting that −2000 to −1500 region on G6PD promoter was responsible for c-MYC-mediated transcriptional activation (Fig. 5d). This region contains one c-MYC potential binding position (GGATATAAAC) on the G6PD promoter. Mutation of this position markedly reduced the G6PD promoter activity induced by c-MYC (Fig. 5e). These results strongly imply that the regulatory effect of ID1 on G6PD promoter activity was mediated by c-MYC expression.

Next, we asked whether c-MYC expression was regulated by ID1. As seen in Fig. 6a and b, c-MYC mRNA and protein expressions were reduced when ID1 expression was suppressed in 97H–OXA-shID1 cells or 3B–OXA-shID1 cells. ID proteins belong to class V of HLH factors that lack the basic domain; they dimerise with E proteins and prevent their DNA interaction, thus, acting as dominant negative of E proteins [18, 19]. Since c-MYC is a bHLH protein, it is entirely possible that ID1 directly interacts with c-MYC, controlling and affecting G6PD transcriptional activity. By coimmunoprecipitation, we detected a measurable amount of ID1/c-MYC complexes (Fig. 6c). However, strikingly, ID1 binding to c-MYC did not appear to have negative effects on the transcriptional activity of G6PD. In luciferase reporter analyses, G6PD transactivation was gradually enhanced when the pGL3-ID1 plasmid concentration was increased from 0.5 to 2 μg in the same transfection reaction (Fig. 6d), due to the argument of c-MYC expression (Fig. 6e). These results indicate that although ID1 could directly bind to c-MYC, it did not suppress its downstream gene G6PD transcriptional activity.

Knowing that ID1 regulated self-renewal of both normal and cancer stem cells via Wnt/β-catenin dependent c-MYC transcription activation [20], and ID1 promoted c-MYC expression through MAPK/ERK signal pathway in HCC [21], we next explored whether ID1 enhanced c-MYC activities through these two pathways. As is illustrated in Fig. 6f, Wnt/β-catenin pathway, rather than MAPK/ERK pathway, was activated in ID1 over-expression 97H–OXA and 3B–OXA cell lines. Down-regulation of ID1 expression inhibited Wnt/β-catenin pathway activation in 97H–OXA-shID1 cells and 3B–OXA-shID1 cells. Treatment with FH535 (Sigma-Aldrich, St. Louis, MO, USA), a small-molecule inhibitor of Wnt/β-catenin, repressed ID1-mediated c-MYC/G6PD expression in 97H–OXA and 3B–OXA cell lines (Fig. 6g and h). Together, these results indicate that ID1 induced c-MYC activation through Wnt/β-catenin pathway, thus promoting G6PD transcription, which in turn activated pentose phosphate pathway and conferred chemoresistance to oxaliplatin in HCC (Fig. 7).

Knowing that ID1/G6PD signaling promoted tumor cell proliferation and chemoresistance in HCC, we wondered whether ID1/G6PD signaling could predict unfavourable prognosis in HCC patients. To prove this postulation, we downloaded a gene expression dataset from TCGA database using the UCSC Cancer Genomics Browser (https://​genome-cancer.​ucsc.​edu). HCC patients were stratified as ID1/G6PD high or ID1/G6PD low based on the median values of ID1 and G6PD expression. Survival analysis indicated that the expressions of ID1 and G6PD were associated with a poor survival outcome in HCC patients (Log rank P = 0.001 and P < 0.001) (Additional file 4: Figure S2A and 2B). Moreover, the most unfavourable prognosis was observed in patients with coexpression of ID1 and G6PD among the entire cohort (Log rank P < 0.001) (Additional file 4: Figure S2C).These results demonstrate that ID1/G6PD signaling could predict a poor clinical outcome in HCC patients.