Background
Radiofrequency ablation (RFA) has been established as the standard care for unresectable early hepatocellular carcinoma (HCC) with complete response rates exceeding 90% [
1,
2]. Due to its superiorities such as effectiveness, minimal invasiveness, safety and repeatability, RFA has been advocated to treat the patients with medium-large HCC [
3,
4]. However, RFA efficacy diminishes and local tumor recurrence increases (10–52.4%, depending on tumor size) [
4‐
6], which greatly impairs survival of patients. When RFA is used to treat HCC beyond 3 cm, local tumor recurrence is a common phenomenon because it is difficult for RFA to secure a sufficient safety margin in three dimensions to ablate microsatellite or microvascular invasion around the tumor [
7]. So, local tumor recurrence arises from those minimal or occult residual tumors. Even more, rapid and aggressive tumor progression after incomplete RFA has been increasingly reported [
8‐
13], albeit its mechanisms are not fully understood.
Previous studies have shown that the change of HCC cells in response to heat stress, such as sarcomatous appearance, heat-resistant subline, enrichment of cancer stem cells or occurrence of epithelial–mesenchymal transition (EMT) [
14‐
19], is associated with accelerated tumor progression after suboptimal heat treatment. However, these studies neglect the potent signals from tumor microenvironment that may favor tumor progression.
Hepatocellular carcinoma occurs within a fibrotic tumor microenvironment rich in activated hepatic stellate cells (HSCs). Evidences have shown that mutual interactions between HCC and activated HSCs promote the tumorigenicity, growth, migration, invasion, angiogenesis and metastasis of HCC [
20‐
24]. More importantly, Rozenblum et al. observed the mass accumulation of the activated HSCs at the perimeter of the ablation zone [
25]. This prompted us to hypothesize that the cross-talks between activated HSCs and residual HCC cells would enhance the tumor progression after incomplete thermal ablation and they could be potential therapeutic targets.
Here, we provided the evidence how activated HSCs enhanced tumor progression of heat-treated residual HCC as the following: (i) activated HSCs-derived condition medium (HSC-CM) aggravated malignant phenotypes of heat-exposed residual HCC cells (ii) periostin (POSTN) mediated the tumor-promoting effects of HSC-CM (iii) vitamin D analog calcipotriol blocked POSTN secretion from activated HSCs (iv) calcipotriol plus cisplatin suppressed the in vivo activated HSCs-promoted tumor progression of the residual HCC cells via inhibition of POSTN expression and an increase of apoptosis.
Methods
Cell culture and conditioned medium (CM) collection
HCC cell lines MHCC97H and MHCC97H with integrin-β1 knockdown (Liver Cancer Institute of Fudan University, Shanghai, China), Hep3B and HepG2 (ATCC, USA), Huh7 (Japanese Cancer Research Bank) were grown in DMEM with 10% FBS (Gibco) and 1% penicillin/streptomycin. Cell lines were authenticated by short tandem repeat validation analysis during the study period. Primary human hepatic stellate cells (pHSCs) (Sciencell, USA) were maintained in the provided medium and LX2 cells (a gift from S. Friedman) were cultured in DMEM with 2% FBS. All cell cultures were carried out in a 37 °C incubator with a humidified atmosphere in presence of 5% CO2.
As in our previous description [
26], conditioned medium was collected from activated HSCs (HSC-CM) and anti-human POSTN antibody (2.5 μg/mL) (Abcam, Cambridge, UK) was added into HSC-CM to neutralize the activity of POSTN.
To obtain conditioned medium from calcipotriol-treated HSCs, pHSCs or LX2 cells were pre-stimulated using 10 ng/mL TGF-β1 and then incubated with 100 nM calcipotriol (Sigma-Aldrich) for 12 h, replenished with fresh medium for another 24 h and the medium was collected for the subsequent experiments.
In vitro heat treatment
As previously described [
26], HCC cells were heated at the pre-set temperatures, seeded into 96-well plates and cell viability was measured at 48 h after heat treatment. IT50 was calculated to indicate the temperature of inducing a 50% reduction in cell viability compared with the 37 °C control. The IT50 data was used to simulate in vivo sublethal heat condition.
Cell proliferation
Cell proliferation was measured using the cell viability assay WST-1 or BrdU ELISA kit (Roche, Manheim, Germany) according to the manufacturer’s instructions. Briefly, HCC cells were subjected to sublethal heat treatment and plated in 96-well plates (8 × 103 cells/well). Equivalent amount of HSC-CM or control medium was added into each well and replaced daily to remove cell debris. The absorbance was measured at 450 nm on a multiskan spectrum reader (Thermo Scientific).
Cell co-culture were performed using a transwell system containing the 0.4 μm pore filter insert (Millipore, Switzerland) allowing medium components freely diffusion but restricting cells migration. Briefly, heat-treated HCC cells were plated in the bottom alone or together with HSCs cells in the transwell insert (4:1 ratio). After 4 days, the proliferation of HCC cells in the bottom wells were measured using WST-1 reagent.
Cell migration and invasion assay
As previously described [
27], migration and invasion assays were carried out. Briefly, in the migration assay, heat-treated residual HCC cells (1 × 10
5) were seeded into the upper chamber while 600 μL HSC-CM or control medium was added into the lower chamber, and then incubated for 36 h. The invasion assay was done in a similar manner except transwell insert pre-coated with Matrigel (BD Biosciences) and culture for 72 h. At the end of experiments, the cells in the upper chamber were removed with cotton swabs and the cells on the lower surface of the inserts were fixed, stained and photographed from five random fields (100× magnification) under phase contrast microscope (Leica, Germany).
After sublethal heat treatment, HCC cells (2 × 103 cells) were seeded into 6-well plates and equal volume of HSC-CM or control medium was then added into each well for 8 days. Cell colony with a diameter larger than 50 μm was counted using ImageJ software (National Institutes of Health, Bethesda, MD, USA) after fixed with methanol, stained with crystal violet and photographed.
Flow cytometry analysis
Cell apoptosis was detected using annexin V staining kit (Invitrogen, Carlsbad, California, USA) according to the manufacturer’s instruction. In brief, heat-treated HCC cells were incubated with HSC-CM or control medium for 3 days. Then, cells were harvested and resuspended in annexin-binding buffer (1 × 106 cells/mL). Subsequently, appropriate amount of Alexa Fluor 488 annexin V and PI working solution were added. Early and late apoptosis rates were analyzed by FACS Calibur flow cytometer (BD Biosciences, USA) and FlowJo software (Tree Star Inc, Ashland, Ore).
Quantitative reverse transcription-PCR (qRT-PCR)
Total RNA was extracted using TRIzol reagent, transcribed and amplified with the use of RevertAid First Strand cDNA Synthesis kit and the Maxima SYBR Green qPCR Master Mix kit (Thermo Fisher Scientific, Waltham, MA, USA). The relative gene expression was calculated using the equation 2
−ΔΔCt where the Ct value of GAPDH or β-actin was used as the normalization. The PCR primers are shown in Additional file
1: Table S1.
Western blot analysis
Western blot was performed as the previously described [
26]. Briefly, total protein was extracted using RIPA buffer added with PMSF and phosphatase inhibitors. Protein concentration was determined using BCA protein assay (Millipore, Switzerland). Then, 20 μg of protein was separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes (PVDF, Millipore, USA). After blocked with 5% non-fat milk, the membrane was incubated with primary antibody against PCNA (1:2000), Snail (1:1000), Vimentin (1:1000), E-cadherin (1:1000), N-cadherin (1:1000), integrin β1 (1:1000), Shc (1:1000), phosopho-Shc (1:2000), ERK1/2 (1:1000) or phosopho-ERK1/2 (Thr202/Tyr204) (1:2000) (Cell Signaling Technology, USA), periostin (1:1000), collagen I (1:5000), α-SMA (1:300) or anti-vitamin D receptor antibody (1:1000) (Abcam, USA), GAPDH (1:1000), tubulin (1:1000) or β-actin (1:1000) (Beyotime, China) at 4 °C overnight and the corresponding HRP-conjugated secondary antibody for 1 h at room temperature on the next day. The membrane was developed with enhanced chemiluminescence (ECL; New Cell & Molecular Biotech Co., China).
Immunohistochemical analysis
As previously described [
28], immunohistochemistry was conducted using the streptavidin–peroxidase two-step method. Briefly, tissue section was deparaffinised, rehydrated, heated with antigen retrieval, blocked with 3% hydrogen peroxide and then incubated with primary antibody against POSTN (1:200), α-SMA (1:100), PCNA (1:400), cleaved-Caspase-3 (1:100) or E-cadherin (1:100) (Cell Signaling Technology, USA) at 4 °C overnight. The immunoreactivity and nuclear counterstain was performed using EnVision two-step visualization system (GeneTech, Shanghai, China) and hematoxylin. Photographs of three representative fields (200× magnification) were captured by microscope (Leica, Germany).
Dual luciferase assay
The 1284-base pair cDNA containing the coding region of vitamin D receptor (VDR), named GV141-VDR, was purchased from Genechem (Shanghai, China). A luciferase reporter construct plasmid GV238 that contained the 5ʹ-flanking region of the POSTN gene from − 2000 to − 1 relative to the transcription start site (corresponds to − 2000 relative to the first AGA of 5′ UTR, amplified with the primers 5′-GATAGGTACCGCAAAGAACGACTAGGTTAAAATTG-3′ and 5′-AGATCTCGAGGAACTCTTTCCAGGAAGCATCGG-3′), as well as a series of deletion constructs (deleted at intervals of 500 base pairs from the 5ʹ end), were generated by GeneChem (Shanghai, China). Based on the conserved motif of VDR, the JASPAR database (
http://jaspar.genereg.net/) was used to predict the binding sites of VDR on POSTN promoter sequence. Two putative sites in the promoter sequence of POSTN were identified (− 1829 to − 1815 and − 1415 to − 1401). To determine functionally relevant VDR binding sites, the putative binding sites for the seed region of VDR were mutated using site-specific sequence mutagenesis (CCCCACGGTTTCCAT). All promoter constructs were sequenced. HEK 293T cells were seeded into 24-well plates at 50–60% confluence per well. Cells were transfected with 1 μg POSTN promoter plasmids containing the firefly luciferase reporter, 1 μg VDR expression vector GV141-VDR and 20 ng Renilla luciferase expressing construct (as an internal control) into HEK 293T cells using Lipofectamine 2000 reagent (Invitrogen, CA, USA) and according to the manufacturer’s protocol, and cells were dealt with 100 nM calcipotriol for 12 h after 12 h of transfection. Twenty-four hours after transfection, the cells were detected for luciferase activity using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA) and multi-plate reader. Firefly luciferase activity is normalized to the Renilla luciferase activity. Each experiment was independently performed at least two times in triplicate.
Gene expression profile
Total RNA was extracted and purified from heat-exposed residual MHCC97H cells treated with or without 100 ng/mL POSTN using miRNeasy Mini kit (QIAGEN, Germany). Genome-wide expression profiling was conducted using Affymetrix GeneChip Human Genome U133 Plus 2.0 Array according to the manufacturer’s instructions. Briefly, total RNA samples were amplified, labelled and purified to obtain biotin-labeled cRNA using GeneChip 3′IVT PLUS reagent Kit (Affymetrix, CA, US). The arrays were hybridized, washed and stained using the GeneChip Hybridization Wash and Stain Kit, and then scanned with a GeneChip Scanner 3000 (Affymetrix, CA, US) for subsequent generation of raw data. The gene expression data were processed and further analyzed using MAS 5.0 algorithm, Affy packages in R. Genes significantly differentially expressed were selected based on fold change > 2 or < 0.5, and T-test with P < 0.05. Gene annotation was based on Gene Ontology analysis. KEGG pathway enrichment analysis was conducted to examine the functional association between differentially expressed genes and to generate the significant gene networks. Differentially expressed genes were mapped to STRING database (
http://string-db.org/) and then imported into Cytoscape software (
http://www.cytoscape.org/) to acquire protein–protein interaction (PPI) network. The microarray data have been deposited into Gene Expression Omnibus (GEO) database (GSE108853).
Gene expression profiles of HCC cohorts (374 tumoral samples and 50 normal control samples) were downloaded from Genomic Data Commons Data Portal (GDC,
https://gdc.cancer.gov/). Data processing and analysis were performed by using R Statistical Software. The expression enrichment of a gene signature was assessed significance for differences between the two groups using unpaired two-tailed Student’s t-test. The strength of associations between two parameters was assessed using Spearman’s correlation test. Survival plots were generated using Kaplan–Meier analyses and splitting the tumor samples into high- and low-expression groups. The median of tumor tissues was token as the threshold and significance for differences of survival between both groups was assessed by a log-rank test.
Animal experiments
All animal experiments were conducted according to the guidelines of the Shanghai Medical Experimental Animal Care Commission and approved by the Ethical Committee on Animal Experiments of Fudan University, Shanghai, China.
Tumorigenicity assay was performed as described [
29]. In brief, 2 × 10
4 heat-treated residual MHCC 97H cells resuspended in 50% growth-factor-reduced Matrigel (BD Biosciences) were subcutaneously injected into the flanks of NOD/SCID mice alone or with 100 ng/mL POSTN. Mice were sacrificed 2 months post injection, at which time tumors were harvested.
As we previously described [
26], 4-week old BALB/c nude mice were inoculated subcutaneously with heat-treated residual MHCC 97H cells (2 × 10
7, n = 6) with pHSCs (ratio, 4:1) into the flanks of mice. Eighteen tumor-bearing mice were randomly assigned into three treatment groups: saline, calcipotriol (60 μg/kg, intraperitoneal, daily), or calcipotriol (60 μg/kg, intraperitoneal, daily) + cisplatin (4 mg/kg, twice a week). After 2 weeks, mice were sacrificed and tumor samples were harvested.
To examine the presence of activated HSCs in the tumors, CFSE labelled pHSCs (5 × 106) with heat-treated residual MHCC 97H cells were inoculated subcutaneously into nude mice. At 2 weeks post injection, mice were euthanized and tumors were cryosectioned to detect the HSCs using fluorescence microscope (Leica, Germany).
Statistical analysis
All statistical calculations were performed using Prism 6.0 (GraphPad Software, Inc., La Jolla, CA). Data were expressed as either mean ± standard deviation (SD) or percentage (%) when appropriate. Continuous data were analyzed by unpaired Student’s t-test or ANOVA between two groups or multiple groups whereas categorical data were compared using Chi-square test or Fisher’s exact test. Differences were considered statistically significant when P < 0.05 (two-side).
Discussion
Local ablation, preferably RFA, is currently considered as the standard and first-line treatment option for patients with unresectable early-stage HCC [
1,
2]. However, when it is expanded to the treatment of medium and large lesions, high local recurrence accounts for the worse prognosis due to incomplete ablated tumors [
4,
5]. Recently, accelerated tumor progression after incomplete RFA has been increasingly reported [
8‐
10,
13,
36]. In this study, we firstly demonstrate that activated HSCs promote the tumor progression of residual HCC after sublethal heat treatment through the release of POSTN. Second, we show that this detrimental effect could be reversed by vitamin D analogue calcipotriol, and cisplatin has an addition to the inhibitory effect of calcipotriol on heat-treated residual HCC, suggesting that calcipotriol plus cisplatin could be applied to thwart the accelerated progression of residual HCC after suboptimal heat treatment. Our findings have clinical implications in improving the survival outcome of RFA.
In contrast to previous studies focusing on the changes of HCC cells response to heat treatment [
14,
18,
19,
37‐
39], we provide a new evidence that activated HSCs enriched in tumor-promoting microenvironment could promote the accelerated progression of heat-treated residual HCC through POSTN secretion. POSTN is one of pivotal proteins from activated HSCs and becomes a better anti-fibrotic therapeutic target [
40]. It is also involved in initiating the process of proliferation and EMT in cancer cells [
30]. In addition, stromal expression of POSTN is associated with high aggressiveness of HCC [
41]. To our knowledge, there is no study of the interaction between activated HSCs and post-treatment residual HCC via POSTN.
Similar to the function of fibroblasts in the process of wound healing [
42], activated HSCs are recruited into the peri-ablative zone after RFA [
25]. Therefore, it is plausible to explore the cross-talks between activated HSCs and heat-treated residual HCC cells as a tumor promoting mechanism. In the present study, we showed that POSTN from activated HSCs promoted the proliferation, motility, invasion prominent activation of EMT as well as decreased apoptosis in heat-exposed residual HCC cells. The result is consistent to the previous reports of POSTN accelerating the progression in other tumors [
43,
44]. Moreover, using gene microarrays, we identified Shc as a molecule with biological importance in POSTN-mediated signaling networks. Further analysis of data from TCGA revealed POSTN expression was strongly associated with markers of HSC activation, proliferation, EMT and Shc-mediated signaling, and the co-expression of POSTN and ERK2 conferred poor-survival prognosis in the TCGA-HCC cohort. ERK1/2 are known direct downstream mediator of Shc. ERK activation has been involved in the regulation of proliferation and EMT of HCC cells [
14,
37,
38]. Recently, it has been reported that increased Shc3 expression results in activation of MEK/ERK in HCC independently of c-Raf [
45]. Shc is signaling partner for integrins [
46‐
48]. In this study, we demonstrated that POSTN promoted the malignant behaviors of heat-exposed residual HCC cells via integrin β1 and p52Shc-ERK1/2 activation, indicating that integrin β1-Shc-ERK axis as major responsible pathway for delivering signals from POSTN to heat-exposed residual HCC cells.
Considering the important roles of activated HSCs on the accelerated growth of heat-exposed residual HCC cells, we sought to revert activated HSCs to quiescence or block the POSTN secretion, which might help prevent the progression of heat-treated residual HCC cells. Vitamin and its analogues have been shown to attenuate fibrosis and delay tumor progression through suppressing the activities of fibroblasts, stellate cells [
33‐
35]. In scleroderma, the vitamin D analogue inhibits TGF β-induced POSTN expression to reduce fibrosis [
49]. Another study has reported that calcipotriol is capable of inhibiting POSTN expression in pancreatic ductal adenocarcinoma [
33]. In line with the above studies, calcipotriol significantly reduced POSTN secretion from activated HSCs and inhibited the subsequent pro-tumorigenic effects of activated HSC on heat-exposed residual HCC cells. The two VDR transcription factor binding sites in the POSTN promoter was identified as crucial negative regulators of POSTN expression in HSCs, suggesting that POSTN expression in HSCs is regulated by calcipotriol via the VDR axis. Because VDR was highly expressed on the HSCs relative to HCC cells, the observed therapeutic effect of calcipotriol inhibiting tumor growth likely results from suppressing activated HSCs. In this study, calcipotriol plus cisplatin showed the additive and therapeutic effects, suggesting combination strategies could be used to suppress the accelerated progression of residual HCC after insufficient heat treatment.
This study has several limitations. First, although we demonstrate the role of POSTN in tumor progression of heat-treated residual HCC, we could not exclude the other factors implicated in post-inflammation reaction after RFA that will promote tumor progression of heat-treated residual HCC, such as a Th1 cytokine pattern, cellular infiltration at the periablational zone, heat shock proteins. Second, it is better to have a group of tumor-bearing mice treated with cisplatin alone in experimental therapy. However, we showed that conventional anticancer agent cisplatin had an addition to the inhibitory effect of calcipotriol on heat-treated residual HCC. Cisplatin is a common therapeutic agent used for chemotherapy in HCC and has been shown to have tumor inhibition in the HCC xenografts [
50]. Whether there exists a synergistic relationship between calcipotriol and cisplatin needs to be verified in additional animal studies. Third, we employed a subcutaneous tumor model of implanting heat-treated residual HCC cells and activated HSCs in mice to assess the response to experimental therapy. Better animal models (an orthotopic model of HCC, rabbit VX2 hepatoma) are needed to verify our findings. Fourth, calcipotriol treatment carries risks of local and systemic side effects because of its toxicity. Topical application of calcipotriol or the use of cholecalciferol as an alternative to calcipotriol may be a safe and effective therapy.
Authors’ contributions
RXC and JFC conceived and designed the study, RZ, XHL, MM, JC, JC, DMG performed the experiments, RXC and RZ analyzed the data and wrote the manuscript. All authors read and approved the final version of the article and the authorship list. All authors read and approved the final manuscript.