NOD-like receptor X1 functions as a tumor suppressor by inhibiting epithelial-mesenchymal transition and inducing aging in hepatocellular carcinoma cells

From January to December 2008, 635 patients with HCC were recruited. Enrollment criteria were as follows: (a) definitive HCC diagnosis, (b) no prior cancer treatment, (c) complete resection of all tumor nodules with margins confirmed free of cancer by histologic examination, and (d) availability of complete clinicopathologic and follow-up data [5]. HCC diagnosis was based on histopathology. The Barcelona Clinic Liver Cancer (BCLC) staging system was used to assess tumor stage [2]. Tumor differentiation was determined according to the Edmondson grading system. Liver function was assessed with the Child–Pugh scoring system [2]. Approval for use of human subjects was obtained from the research ethics committee of Zhongshan Hospital. Informed consent was obtained from each subject.

Patients were prospectively monitored by serum α-fetoprotein (AFP) testing, abdomen ultrasonography, and chest X-ray every 1–6 months as previously described [5]. Follow-up ended in August 2017. Time to recurrence (TTR) was defined as the interval between surgery and the diagnosis of any type of recurrence including intra- or extrahepatic recurrence as identified by magnetic resonance imaging or computed tomography. OS was defined as the interval between treatment and death of any cause or the last observation date.

Huh7, HepG2, and 7721 cell lines were purchased from the Cell Bank at the Institute of Biochemistry and Cell Biology, China Academy of Science (Shanghai, China). MHCC97L, MHCC97H, and HCCLM3 cell lines were previously generated in our institute. All cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS) supplemented with 100 IU/mL penicillin and 100 μg/mL streptomycin and incubated at 37 °C in a humidified atmosphere with 5% CO₂. All cell culture reagents were obtained from Life Technologies (Thermo Fisher Scientific, Waltham, MA, USA).

Cell apoptosis was analyzed by flow cytometry using annexin V-fluorescein isothiocyanate (FITC). Apoptosis detection kits (BD Biosciences, San Jose, CA, USA) were used according to the manufacturer’s protocol. Briefly, cells exposed to different treatments were harvested and suspended in binding buffer. An aliquot of 100 μL was incubated with 5 μL annexin V-FITC and 5 μL propidium iodide for 15 min in the dark, and 400 μL binding buffer (1×) was added to each sample. The stained cells were analyzed by flow cytometry within 1 h.

HCCLM3 cells were fixed in 4% paraformaldehyde and blocked with 5% bovine serum albumin. Afterwards, 0.1% Triton was used for permeabilization followed by blocking with 5% bovine serum albumin. Then, FITC-conjugated mouse anti-human NLRX1 antibodies (1:30; BioLegend, San Diego, CA, USA) were added and incubated overnight at 4 °C. HCCLM3 cells were also counterstained with DAPI (Sigma-Aldrich, St. Louis, MO, USA). Images were captured using an IX-71 fluorescent microscope (Olympus, Tokyo, Japan).

Total RNA extraction was conducted by RNeasy mini kit (Qiagen, Hilden, Germany), and cDNA was synthesized via Quantitect Reverse Transcription Kit (Qiagen) according to the manufacturer’s instructions. Target genes were quantified using FastStart Universal SYBR Green Master (Roche Diagnostics, Basel, Switzerland), and DNA amplification was carried out using a LightCycler 480 (Roche Diagnostics). The relative quantities of target gene mRNAs compared to an internal control were determined using the ΔCq method. Reverse transcription-polymerase chain reaction (RT-PCR) conditions were as follows: 5 min at 95 °C, followed by 40 cycles of 95 °C for 10 s and 60 °C for 60 s. GAPDH was used as an internal control. Primers and probes are listed as follows: NLRX1, F: 5′-CGACCAGATGATCGTATCC-3′ R: 5′-TGCGTCACTGAGGTGTTTCCTGCC-3′; E-cadherin, F: 5′-TTGCTACTGGAACAGGGACAC-3′ R: 5′-CCCGTGTGTTAGTTCTGCTGT-3′; N-cadherin, F: 5′-TTATCCTTGTGCTGATGTTTGTG-3′ R: 5′-TCTTCTTCTCCTCCACCTTCTTC-3′; Vimentin, F: 5′-CCTTGACATTGAGATTGCCACCTA-3′ R: 5′- TCATCGTGATGCTGAGAAGTTTCG-3′; Snail1, F: 5′- TCCAGAGTTTACCTTCCAGCA -3′ R: 5’-CTTTCCCACTGTCCTACTCTG -3′; Twist1, F: 5’-GTCCGCAGTCTTACGAGGAG-3′ R:5′-GTCTGAATCTTGCTCAGCTTGTC-3′; beta-actin, F: 5′-CTGAGGACAAGCCACAAGATTA-3′ R: 5′-ATCCACCAGAGTGAAAAGAACG-3′.

Protein from HCC cells were lysed in complete radioimmunoprecipitation assay buffer for WB analyses. All protein lysates were quantified using a quantitative bicinchoninic acid protein assay. A total of 30 μg protein was mixed in sodium dodecyl sulfate (SDS) loading dye containing 20 mg/mL dithiothreitol reducing agent, then boiled for 5 min. Proteins were separated by SDS-polyacrylamide gel electrophoresis using 4–12% Bis-Tris gels and wet transferred to nitrocellulose membrane (Bio-Rad, Hercules, CA, USA). Nitrocellulose membranes were blocked for 1 h with 10% non-fat milk and incubated overnight with primary antibodies, washed five times with Tris-buffered saline containing Tween, and incubated for 2 h at room temperature with the appropriate secondary antibodies. Then, protein expression was determined by chemiluminescent reagents (Thermo Fisher Scientific).

For the proliferation assay, control or NLRX1-modulated HCC cells were aliquoted into a 96-well plate at 1000/100 μL per well. At the indicated time points, 20 μL of cell counting kit 8 (CCK-8) solution (Dojindo, Kamimashiki-gun, Kumamoto, Japan) was added to determine the number of viable cells in each well.

Cell invasion ability was evaluated by Transwell (Corning, Corning, NY, USA) assays as previously described. Briefly, HCC cells subjected to different treatments were collected and washed with phosphate-buffered saline (PBS). Next, 10⁵ cells were seeded in the upper chamber with a MatriGel-coated membrane (dilution 1:6), and the lower chambers were supplied with DMEM containing 10% FBS to act as chemo-attractant. After 24 or 48 h of incubation at 37 °C, cells that had already migrated or invaded to the lower surface of membrane were fixed with 4% methanol, stained with crystal violet, and counted in 10 random × 200 microscopic fields per sample.

Cell apoptosis was analyzed by flow cytometry using annexin V-FITC. Apoptosis Detection Kits (BD Biosciences) were used according to the manufacturer’s protocol. Briefly, cells were harvested and suspended in binding buffer (1×). An aliquot of 100 μL was incubated with 5 μL annexin V-FITC and 5 μL propidium iodide for 15 min in the dark, and 400 μL binding buffer (1×) was added to each sample. The stained cells were analyzed by flow cytometry within 1 h.

The resected specimens were embedded in paraffin and stored at 4 °C. The construction of the TMA and immunohistochemistry protocol were described previously [9]. Briefly, immunohistochemical staining was performed using the avidin-biotin-peroxidase complex method. After rehydration and microwave antigen retrieval, primary anti-human-NLRX1 antibodies were applied to slides for overnight incubation at 4 °C. Then, secondary antibody incubation was conducted at 37 °C for 30 min. Staining was performed with 3′3-diaminobenzidine tetra hydrochloride, and counterstaining was performed with Mayer’s hematoxylin. We included negative control slides with the primary antibodies omitted in all assays. Immunohistochemical staining was independently assessed by two pathologists.

Cignal Finder RTK signaling 10-Pathway Reporter array was used to uncover the potential down-stream signaling pathway controlled by NLRX1. Control, NLRX1-KD Huh7 cells and control, NLRX1-OE HCCLM3 cells were subsequently transfected with a mixture of a transcription factor-responsive firefly luciferase reporter and a constitutively expressing Renilla construct. The relative activity of each pathway was decided by luciferase/Renilla and normalized by untreated controls. Experiments were performed in triplicates.

For the mouse xenograft model, 6-week-old male nude mice were purchased from the Chinese Science Academy (Shanghai, China). Mice were subcutaneously implanted with Huh7 cells infected with lentivirus (Huh7, Huh7-NLRX1^KD; 3 × 10⁶). Tumor volume was measured twice a week, and tumor growth was calculated as the follow equation: larger diameter × (small diameter)²/2. Five weeks after HCC cell injection, the mice were sacrificed, and tumor tissues were resected for hematoxylin and eosin staining.

The Flag-Nlrx1, Flag-Nlrx1R1, and Flag-Nlrx1R2 truncations were subcloned into pcDNA3.1 vectors. The shNlrx1 was subcloned into a PLKO.1 vector. To transiently express a given protein, cells were transfected with Lipofectamine (Life Technologies, Carlsbad, CA, USA). For 6-cm plates, a total of 6 μg DNA was used for transfection, whereas 12 μg DNA was used to obtain saturated effects. A retroviral vector pMSCV was used to stably knock-down (KD) NLRX1 or negative control (NC). The retrovirus was produced by 293 T packaging cells. HCC cells were infected with these viruses or control virus and then selected against puromycin for 3 days before use in assays.

HCC cellular senescence was evaluated by detecting the activity of senescence-associated β-galactosidase (SA-β-gal) with Senescence β-Galactosidase Staining Kits (Beyotime, China) according to the manufacturer’s instruction. Briefly, cells were seeded into six-well plates, cultured with the kit reagents for 48 h, fixed for 15 min at room temperature with 1 mL fixative solution, and then washed three times with PBS. Next, the cells were incubated overnight at 37 °C with a staining solution mixture containing X-gal. After cells were rinsed with PBS, they were observed for the development of the blue coloration under a light microscope (× 400).

Expression profiling was conducted using Affymetrix 3’ IVT Expression microarray. Briefly, total RNA was extracted using RNeasy Mini Kit (Qiagen), then first-strand cDNA was synthesized by reverse transcription. Afterwards, labeled aRNA was synthesized and purified for further hybridization. Raw data were obtained for further analysis.

Statistical analyses were performed using SPSS 20.0 software (IBM, Armonk, NY, USA). Experimental values for continuous variables are expressed as the mean ± standard error of the mean. Chi-squared tests, Fisher’s exact probability tests, and Student’s t tests were used when appropriate to evaluate the significance of differences between groups. If variances within groups were not homogeneous, a nonparametric Mann–Whitney test or Wilcoxon signed-rank test was used. The relationships between NLRX1 expression and TTR or OS were analyzed using Kaplan–Meier survival curves and log-rank tests, respectively. P < 0.05 was considered statistically significant.

First, we investigated the expression patterns of NLRX1 between HCC tumor tissues and paired adjacent normal liver tissues. RT-PCR showed that 44.23% (23/52) of tumors exhibited significantly lower expression of NLRX1 compared with paired normal liver tissues, while only 9.60% (5/52) of tumors showed NLRX1 upregulation (Fig. 1a). WBs confirmed downregulation of NLRX1 within tumor tissue (Fig. 1b).

To explore whether low NLRX1 expression was associated with disease prognosis, a TMA containing 575 patients who underwent curative resection was immunostained. NLRX1 expression status was stratified according to Fig. 1c and Table 1. For the 575 patients with HCC enrolled, the median TTR and OS were 20.00 months (range 0.60–59.00) and 45.00 months (range 2.50–59.00), respectively. Kaplan–Meier analysis revealed significantly shorter median OS in patients with low NLRX1 was compared with those whose status was high (12.80 months vs. not reached, P < 0.001; Fig. 1d). In addition, patients with low NLRX1 had significantly shorter TTR compared with patients with high expression (median 32.00 months vs. not reached, P < 0.001; Fig. 1e). Univariate analysis showed that TTR correlated with NLRX1 expression status, AFP level, tumor number, tumor size, satellite lesions, vascular invasion, differentiation, and BCLC stage (all P < 0.05, Table 2), while OS correlated with NLRX1 expression status, Child–Pugh score, AFP level, tumor size, satellite lesions, vascular invasion, differentiation, and BCLC stage (all P < 0.05, Table 2). Multivariate analysis revealed that NLRX1 expression status was an independent indicator for both TTR (hazard ratio [HR] 1.87, 95% confidence interval [CI] 1.26–2.76, P = 0.002; Table 3) and OS (HR 2.26, 95% CI 1.44–3.56, P = 0.001; Table 3). Collectively, these results indicate that NLRX1 might act as a tumor suppressor and be a useful prognostic biomarker in HCC.

Next, we investigated the effect of NLRX1 on HCC cells. First, the expression status of NLRX1 in six HCC cell lines were evaluated with RT-PCR and WB. Interestingly, NLRX1 levels were low in cell lines with high metastasis potential, such as MHCC97H and HCCLM3, while it was high in cell lines with low metastasis potential, such as HepG2 and Huh7 (Fig. 2a). Based on these results, Huh7 and HCCLM3 cells were selected for further KD and over-expression (OE) experiments, respectively. The efficiencies of NLRX1 modification were validated by RT-PCR and WB (Fig. 2b). Intracellular localization of over-expressed NLRX1, which is mainly in the cytoplasm, was validated by immunofluorescent staining (Fig. 2c).

The effect of NLRX1 on tumor cell apoptosis was also investigated using flow cytometry. We found that NLRX1-OE significantly increased HCCLM3 cell apoptosis (P < 0.05). Conversely, NLRX1-KD significantly decreased the apoptosis rate of Huh7 cells (P < 0.05, Fig. 2d). Transwell assays were then conducted to explore the effect of NLRX1 on HCC cell invasiveness. The invasive potential of HCCLM3 cells was significantly decreased following NLRX1 OE (P < 0.05, Fig. 2e), while NLRX1-KD significantly promoted Huh7 cell invasive potential (P < 0.05, Fig. 2e).

To investigate the effect of NLRX1 on tumor growth, we first conducted CCK-8 assays. We found that NLRX1 overexpression could hinder cell proliferation in HCCLM3 cells, while knocking down NLRX1 in Huh7 resulted in significant higher proliferation rates (Fig. 2f). Next, 3 × 10⁶ Huh7-Vector (control) and KPNA3-KD Huh7 cells were subcutaneously implanted into nude mice. Mice injected with KPNA3-KD Huh7 developed significantly larger tumor volumes than those injected with Huh7-Vector cells after 5 weeks (Fig. 2g). Collectively, these findings indicate that NLRX1 effectively restrained HCC proliferation.

EMT is a key event that can directly induce tumor invasion and is defined by the loss of epithelial characteristics and acquisition of a mesenchymal phenotype. Hallmarks of this procedure are the loss of E-cadherin expression and upregulation of mesenchymal-related markers such as N-cadherin and vimentin [23, 24]. Therefore, we first observed the cell morphology changes after NLRX1 modulation, and we found that HCCLM3 cells exhibited epithelial-like morphology after NLRX1-OE, while Huh7 cells displayed an elongated spindle shape, distinctive than the rounded appearance in their parental cells after NLRX1 knock-down (Fig. 3a). Next, we performed RT-PCR to evaluate the expression of several EMT-related markers. NLRX1 OE significantly increased mRNA E-cadherin mRNA levels and downregulated N-cadherin, vimentin, Snail1, and Twist1 (Fig. 3b). Meanwhile, NLRX1 KD in Huh7 cells induced an opposite expression pattern (Fig. 3b). WB showed that expression patterns of E-cadherin, N-cadherin, and Snail1 after NLRX1 modification were similar to those observed with RT-PCR. However, Twist1 and vimentin protein levels were not different according to WB (Fig. 3c). These results demonstrated that NLRX1 can successfully suppress EMT in HCC by slightly decreasing Snail1 expression.

To further clarify the mechanism underlying how NLRX1 suppress EMT process, a Cignal Finder RTK signaling 10-Pathway Reporter array was used. Results showed that PI3K-AKT signaling pathway was the most altered pathway due to NLRX1 modulation (Fig. 3d). We further performed a QPCR array including critical genes involved in PI3K-AKT signaling pathway. We found that key genes regulating cell proliferation and migration including P21 (CDNK1A), CCNE2, EIF4E, MDM2, BCL2, and FN1 were significantly upregulated due to NLRX1-KD, while several tumor suppressor genes including TSC2, RPTOR, and TP53 were downregulated due to NLRX-KD in Huh7 cells (Additional file 1: Table S1). Activation of PI3K-AKT signaling is a major event in HCC progression [25‐27], and recent studies demonstrated that the PI3K-AKT pathway induces EMT in a Snail-dependent manner [28]; thus, we inferred that PI3K-AKT pathway might be the key downstream pathway of NLRX1 which contributed to the suppression function of NLRX1. Therefore, we further investigated whether NLRX1 could inhibit PI3K-AKT pathway activation. We found that NLRX1-OE HCCLM3 cells exhibited significantly lower levels of phosphorylated AKT compared with control cells (Fig. 3e). Moreover, as a key downstream target of AKT activation, we also examined the activation status of mammalian target or rapamycin (mTOR). NLRX1-OE markedly inhibited the phosphorylation of mTOR (Fig. 3e), indicating that NLRX1 suppresses PI3K-AKT signaling. To validate the critical role of PI3K-AKT signaling pathway, we conducted rescue experiments via using an AKT activator, SC79. We found that re-activating AKT could effectively abolish the inhibition effect of NLRX1-KD on snail1 expression (Fig. 3f). In addition, re-activation of AKT could restore the mesenchymal-like phenotype of HCCLM3 cells regardless of NLRX1 overexpression (Fig. 3g). Taken together, our results suggest that NLRX1 effectively suppresses the EMT process by downregulating Snail1 expression via inhibition of the PI3K-AKT pathway.

Cell senescence is considered as a key step in apoptosis, and this pathway is mainly controlled by P21 [29], which is a key downstream target of AKT [30]. Given that NLRX1 can induce HCC cell apoptosis and suppress PI3K-AKT signaling, we further investigated whether NLRX1 promotes cell apoptosis by hindering P21 expression. We found that NLRX1-OE could effectively upregulate P21 expression (Fig. 4a), and expressions of down-stream targets of P21 including CDK1 and CDK2 were significantly repressed due to high level of P21 expression caused by NLRX1 overexpression, while knocking down NLRX1 in Huh7 resulted in increased expression of CDK1 and CDK2 (Fig. 4b). To further validate the role of P21 in NLRX1 induced senescence, we knocked down P21 in NLRX1-OE cells, which was validated by WB assays (Fig. 4c), and performed SA-β-gal activity evaluation. Results showed that NLRX1-OE significantly increased SA-β-gal activity, indicating a higher proportion of aging in NLRX1-OE cells compared to cells transfected with vector, and P21 depletion attenuated the effect of NLRX1 OE (Fig. 4d). Collectively, these data indicate that NLRX1 promotes cell senescence via the PI3K-AKT-P21 axis, thus contributing to HCC cell apoptosis.

We attempted to identify the critical fragment of NLRX1 that contributes to its suppressive function. We generated three types of plasmids expressing the following peptides: fragment 75–556 (R1), fragment 556–974 (R2), and fragment 1–974 (full length), and transfected them into HCCLM3 cells (Fig. 5a). We found that transfection of the R2 plasmid, but not R1, could induce HCC cell aging similar to expression of the full-length fragment (Fig. 5b). Moreover, Transwell assays showed that R2 plasmid transfection significantly inhibited cell invasion like the full-length fragment; however, the R1 plasmid failed to mimic the effect of full-length NLRX1 (Fig. 5c). Based on these results, the 556–974 fragment of NLRX1 was identified as a critical contributor to its suppressive function.