PRDM8 exhibits antitumor activities toward hepatocellular carcinoma by targeting NAP1L1

PATIENTS AND SPECIMENS

Two independent cohorts of HCC patients were enrolled in this study. For the first cohort (cohort 1), 60 pairs of HCC tissues and corresponding adjacent nontumorous tissue samples and related clinical information were obtained from patients who received hepatectomy at Department of Hepatobiliary Surgery, the First Affiliated Hospital of Nanjing Medical University between January 2015 and December 2016 with the patients' consent, and 12 paired HCC samples in cohort 1 were randomly selected for western blotting analysis. For the second cohort (cohort 2), tissue microarrays (TMAs) containing 398 pairs of HCC tissues with complete clinicopathological and follow-up data were purchased from Shanghai Zhuoli Biotech (Shanghai, China).

The study protocol which conformed to the ethical guidelines of the 1975 Declaration of Helsinki was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University.

STATISTICAL ANALYSIS

All data are shown as means and standard errors of the mean. Prism 6 (GraphPad Software, La Jolla, CA, USA) and SPSS 21.0 (IBM SPSS software, NY, USA) were used for the Wilcoxon matched-pairs test, Student t test, chi-square test, Fisher exact test, or Pearson correlation test when appropriate. Kaplan-Meier analysis and Cox proportional hazards regression model were used for survival analysis. Values of P < 0.05 were considered statistically significant.

Detailed methodology can be found in supplementary materials and methods.

PRDM8 EXPRESSION WAS DOWN-REGULATED IN HCC AND CORRELATED WITH A POOR OUTCOME

We examined the mRNA expression level of PRDM8 in 60 human HCC tissue samples and their matched nontumorous liver specimens by quantitative reverse transcription polymerase chain reaction (RT-qPCR). The results revealed that PRDM8 mRNA level was markedly down-regulated in HCC tissues (Fig. 1A,B). Decreased protein expression level of PRDM8 in HCC tissue was further confirmed in 12 paired samples randomly obtained from cohort 1 using western blotting (Fig. 1C). Immunohistochemistry (IHC) also confirmed that the protein expression level of PRDM8 was down-regulated in HCC tissues in cohort 1 (Fig. 1D). The clinicopathological analysis revealed that PRDM8 expression was closely correlated with tumor multiplicity, Edmonson stage, tumor size, tumor encapsulation, vascular invasion, and tumor-node-metastasis (TNM) stage (Supporting Table S1). In addition, TMA results confirmed that the expression level of PRDM8 protein was significantly lower in HCC than that in peritumoral tissues (Supporting Fig. S1A,B). Further survival analysis on cohort 2 showed that low PRDM8 group had obvious higher recurrence probability than those in high PRDM8 group, whereas no significant effect on overall survival was observed (Fig. 1E). Multivariate Cox regression analysis demonstrated that low PRDM8 expression was an independent predictive indicator for recurrence-free survival (hazard ratio [HR] = 1.742, 95% confidence interval [CI], 1.056-2.874; P = 0.030) for HCC patients (Table S2).

PRDM8 INHIBITED CELL PROLIFERATION, MIGRATION AND INVASION IN VITRO

We performed RT-qPCR and western blotting to investigate the expression of PRDM8 in HCC cell lines and noncancerous liver cell lines. We found the mRNA and protein expression levels of PRDM8 were reduced in HCC cell lines compared with noncancerous cell lines (Supporting Fig. S2A,B). According to the expression level of PRDM8 in HCC cell lines, we stably overexpressed PRDM8 in low PRDM8-expressing YY8103 and SMMC7721 cells by lentivirus, named YY8103-PRDM8 and SMMC7721-PRDM8, respectively (Supporting Fig. S2C). High PRDM8-expressing Focus and HepG2 cells were selected to knock down the expression level of PRDM8 using lentivirus carrying small hairpin RNA (shRNA). The knockdown efficiency of three shRNAs for PRDM8 was confirmed through comparison with negative control at both mRNA and protein levels (Supporting Fig. S2D,E). Among the three shRNAs tested, shRNA#1 generated the most consistent knockdown results across Focus and HepG2 cell lines, and was thus chosen for subsequent studies.

To determine whether PRDM8 is functionally involved in HCC cells, we performed a variety of in vitro assays to evaluate the effect of PRDM8 overexpression on cell functions including proliferation, migration, and invasion. The results of Cell Counting Kit-8 (CCK-8) assays showed that up-regulation of PRDM8 significantly suppressed cell proliferation of YY8103 and SMMC7721 (Fig. 2A). Colony formation assays confirmed the inhibitory effect of PRDM8 overexpression on cell proliferation (Fig. 2B). In wound healing migration assay, PRDM8-overexpressed cells exhibited delayed wound healing (Fig. 2C). Transwell assays demonstrated that cell migration and invasion were strongly impaired by overexpression of PRDM8 in YY8103 and SMMC7721 cells (Fig. 2D).

Then we assessed the impact of PRDM8 knockdown on HCC cell functions. The CCK-8 and colony formation assays revealed that down-regulation of PRDM8 significantly promoted cell proliferation of Focus and HepG2 (Supporting Fig. S3A,B). As shown in Supporting Fig. S3C, the migration ability of Focus and HepG2 cells with PRDM8 knockdown was increased compared with the control groups. Results from Transwell assays further showed that PRDM8 deficiency enhanced the abilities of HCC cells to migrate and to invade (Supporting Fig. S3D). These results were consistent with the PRDM8 overexpression data, and they indicated that PRDM8 could be a key player in the regulation of cell functions of HCC.

PRDM8 INDUCED G1/S CELL CYCLE ARREST AND ENHANCED APOPTOSIS IN VITRO

We quantified cell cycle distribution using flow cytometry and found that the number of cells was increased at G1 phase and decreased at S phase in PRDM8-overexpressed YY8103 and SMMC7721 cells (Fig. 3A). In contrast, the percentage of G1 phase cells in HepG2-shPRDM8 and Focus-shPRDM8 cells was significantly lower than that in control groups (Fig. 3B). Flow cytometry analysis using Annexin V-APC/7-AAD staining demonstrated that cell apoptosis was induced by up-regulating PRDM8 in HCC cells (Fig. 3C), whereas knockdown of PRDM8 in HepG2 and Focus inhibited cell apoptosis (Fig. 3D).

PRDM8 SUPPRESSED TUMOR GROWTH AND METASTASIS IN VIVO

The HCC cells with stable overexpression or knockdown of PRDM8 were implanted into nude mice to examine the expansion and migration of tumor cells in vivo. Each mouse was monitored once every three days, and the mice were euthanized after four weeks. Tumors derived from SMMC7721-PRDM8 cells grew slower, and the average tumor weight was significantly lower in mice inoculated with PRDM8-overexpressed SMMC7721 cells at the fourth week (Fig. 4A). In mice inoculated with HepG2-shPRDM8 cells, average tumor volume was markedly higher at the fourth week than in control mice, and tumors were also heavier in mice inoculated with HepG2-shPRDM8 (Fig. 4B). Hematoxylin and eosin, IHC, Ki-67, and terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end-labeling (TUNEL) staining of the xenograft tumors are shown in Fig. 4C, and the data further confirmed the antitumor activities of PRDM8 toward HCC.

Detection of metastasis in a nude mouse model transplanted with HCC cells indicated that overexpression of PRDM8 was significantly related to reduced metastasis of SMMC7721 according to photon flux detection (Fig. 5A). In mice injected with HepG2-shPRDM8 cells, we observed increased metastasis of tumor cells when compared to controls (Fig. 5B). Histological analysis of lung tissue showed that incidence of lung metastases in SMMC7721-PRDM8 group mice was decreased, while the incidence was higher in HepG2-shPRDM8 group mice compared with the controls (Fig. 5C–E). In addition, the SMMC7721-PRDM8 group had a longer overall survival time than the SMMC7721 control group (Fig. 5F). Consistently, HepG2-shPRDM8 group had worse survival outcome compared to control group (Fig. 5G).

PRDM8 INHIBITED HCC PROGRESSION VIA PI3K/AKT/MTOR SIGNALING

Considering the pivotal role of PI3K/AKT/mTOR signaling in tumor progression, we investigated the effect of PRDM8 on the specific signaling cascade. We found that PRDM8 overexpression reduced phosphorylation of AKT and mTOR in HCC cells (Fig. 6A), indicative of inhibition of PI3K/AKT/mTOR signaling. Conversely, PRDM8 knockdown enhanced the activity of PI3K/AKT/mTOR signaling (Fig. 6B).

In addition, PRDM8 overexpression altered the expression pattern of the key regulators that are indispensable in cell cycle regulation, tumor metastasis, and cell apoptosis. Western blotting results indicated that the expression levels of Cyclin D1, Cyclin E1, CDK2, CDK4, CDK6, MMP2, MMP9, VEGF, and Bcl-2 were reduced, while p27, Bax, and cleaved caspase-3 levels were up-regulated in PRDM8-overexpressed cells (Fig. 6C). In Focus-shPRDM8 and HepG2-shPRDM8 cells, the levels of Cyclin D1, Cyclin E1, CDK2, CDK4, CDK6, MMP2, MMP9, VEGF, and Bcl-2 expression were increased, and down-regulated expression levels of p27, Bax, and cleaved caspase-3 were also observed (Fig. 6D). According to the results of previous study, the expression of Prdm8 is complementary to that of Hes5 in mouse neurogenesis.18 Thus, we attempted to examine the effect of PRDM8 overexpression or knockdown on HES5 mRNA expression in HCC. However, no significant results were found (data not shown). In addition, previous works reported that reduced expression of PRDM8 was caused by hypermethylation in patients with dyskeratosis congenital, aplastic anemia, and Down syndrome.19, 20 Based on our results, down-regulated PRDM8 expression in HCC cell lines was not significantly restored by treatment with 5-aza-2′-deoxycytidine (data not shown). Prediction of upstream microRNAs was performed using TargetScan (http://www.targetscan.org) and miRanda (http://www.microrna.org). Several microRNAs including miR-302, miR-372, miR-373, miR-425-5p, and miR-520 were suggested as potential upstream regulators of PRDM8.

We promoted the activity of AKT signaling by IGF-1 treatment, and found that AKT induction restored the decreased proliferation, migration, and invasion in PRDM8 overexpressed SMMC7721 cells (Fig. 7). Consistently, pharmacological inhibition of AKT (MK-2206) abrogated the increased proliferation, migration, and invasion results from PRDM8 knockdown in HepG2-shPRDM8 cells (Supporting Fig. S4). Taken together, these data indicated that PRDM8 suppressed HCC progression through blocking PI3K/AKT/mTOR signaling.

PRDM8 SUPPRESSED PI3K/AKT/MTOR SIGNALING AND HCC PROGRESSION VIA NAP1L1

The immunoprecipitation (IP) assay was performed to explore the potential binding protein of PRDM8. As presented in Fig. 8A, further mass spectrometry identification indicated that NAP1L1 was the captured protein by PRDM8. Co-IP verified the specificity of this interaction, suggesting that PRDM8 could bind with NAP1L1 (Fig. 8B,C). Next, we analyzed the mRNA expression of NAP1L1 in the tissue samples of patients in cohort 1. An increased level of NAP1L1 was found in HCC tissues compared with the corresponding peritumor tissues (Fig. 8D). Person correlation analysis also presented a negative correlation between the expression of PRDM8 and NAP1L1 (Fig. 8E). Immunofluorescence results showed that PRDM8 co-localized with NAP1L1 in cytoplasm and nucleus (Fig. 8F). PRDM8 ectopic expression led to reduced NAP1L1 protein level in YY8103 and SMMC7721 cells (Fig. 8G). Conversely, PRDM8 knockdown caused increased NAP1L1 protein (Fig. 8H). To determine whether altered NAP1L1 expression played a role in PRDM8 suppression of HCC progression, we up-regulated the expression of NAP1L1 in SMMC7721-PRDM8 cells using lentivirus (Supporting Fig. S5A). The results revealed that NAP1L1 overexpression rescued the attenuated PI3K/AKT/mTOR signaling activity (Supporting Fig. S5B), and increased proliferation, migration, and invasion of PRDM8-overexpressed SMMC7721 cells (Supporting Fig. S5C–H). In addition, we knocked down NAP1L1 in HepG2-shPRDM8 cells (Supporting Fig. S6A). NAP1L1 knockdown blunted the ability of PRDM8 shRNA to promote phosphorylation of AKT and mTOR (Supporting Fig. S6B), and restored the increased HCC cell proliferation, migration, and invasion (Supporting Fig. S6C–H). Moreover, based on the results of CCK-8 and transwell assays, NAP1L1 alone could exert an antiHCC effect, and overexpressing PRDM8 when NAP1L1 was silenced had an obvious synergistic effect (Supporting Fig. S7A,B).

According to a recent study, depletion of NAP1L1 in hepatocytes leads to the down-regulation of 358 genes, and Ingenuity Pathway Analysis of the downregulated genes indicates that PI3K/AKT signaling is one of the top canonical pathways. GAB1, an important upstream regulator of PI3K/AKT/mTOR, is among those downregulated genes.21 We therefore speculated that NAP1L1 might activate PI3K/AKT/mTOR signaling through GAB1. As shown in Supporting Fig. S7C, western blotting results validated that HCC cells overexpressing NAP1L1 exhibited a higher GAB1 expression level, while HCC cells silencing NAP1L1 presented a decreased GAB1 expression. GAB1 might be the key mediator that activated PI3K/AKT/mTOR in the downstream of NAP1L1. Collectively, these observations indicated that PRDM8 negatively regulated NAP1L1 expression, which was important for PRDM8 to suppress the growth, migration, and invasiveness of HCC cells.