Stress Conditions Induced by Locoregional Therapies Stimulate Enrichment and Proliferation of Liver Cancer Stem Cells

 

Yuchen Huo, Wendy S. Chen, Jin Lee, PhD, Gen-Sheng Feng, and Isabel G. Newton

 

 

The risk of hepatocellular carcinoma (HCC) recurrence after percutaneous thermal ablation (PTA), transarterial chemo- embolization, and transarterial embolization (TAE) could be related to their effects on residual cancer cells. Sublethal heat exposure, hypoxia, and chemotherapy have been shown to enrich the proportion of HCC cells bearing cancer stem cell (CSC) markers (1–4). CSCs are rare, self-renewing, pluripotent cells bearing cell surface markers such as CD44, CD90, CD13, CD133, EpCAM, OV6, CD24, and keratin 19 (2,5–7). Pathways involved in hepatocellular carcinoma cancer stem cell (hCSC) self-renewal, including the canonical Wnt/b-catenin and PI3K-AKT-mTOR path- ways (8–11), represent potential targets for inhibition. If, as some studies suggest, hCSCs are responsible for recurrence, then combining PTA, transarterial chemoembolization, or TAE with hCSC inhibition could reduce recurrence at its could be repurposed to limit HCC recurrence after locore- gional treatment. The present study builds on prior studies demonstrating a heat- and hypoxia-induced increase in hCSC markers (2,3) by testing the hypothesis that condi- tions simulating PTA, TAE, or transarterial chemo- embolization stimulate hCSC enrichment and proliferation and that this can be curbed with niclosamide and other hCSC inhibitors source (3,12–16). One study found that partial PTA of orthotopic N1S1 rat HCC tumors resulted in enrichment of CD44þ/CD90þ cells, especially along the tumor ablation margin; however, in vitro pretreatment of these cells with a dual PI3K-mTOR inhibitor prevented CD44þ cell enrich- ment (2).

No current anti-HCC therapy targets hCSCs. One chal- lenge is the fact that the signaling pathways important for hCSC renewal and maintenance can be common to normal stem cells (9). For patients with hepatic insufficiency, stem cell inhibition could impair regeneration and lead to liver failure. Ideal inhibitors would target hCSCs while sparing normal stem cells. One such candidate is the US Food and Drug Administration (FDA)-approved antihelminthic drug niclosamide and its more bioavailable salt, niclosamide ethanolamine. Niclosamide promotes apoptosis and auto- phagic cell death and inhibits proliferation, migration, and metastasis in HCC, acute myeloid leukemia, glioblastoma and cancers of the lung, breast, ovary, and adrenal gland (17–28). It inhibits CSCs and pathways important for CSC renewal, including the Wnt/b-catenin, AKT, and EGFR/Ras/ Raf pathways (17,18,26,28–32). Niclosamide’s effects are largely specific to cancer cells, rendering it safe and well tolerated (21,32). It also acts as a mild mitochondrial un- coupler, stimulating lipid metabolism and reducing hepatic steatosis (29), a risk factor for HCC. Studies in vitro and in patient-derived HCC xenografts also showed that niclosa- mide binds to cdc37, inhibiting its interaction with heat shock protein 90, thereby interrupting signaling pathways critical for hepatocarcinogenesis (21). Thus, niclosamide

 

 

MATERIALS AND METHODS

Hepatocellular Carcinoma Cultures

HCC cultures were prepared from human the HCC cell lines HepG2 (product HB8065; American Type Culture Collec- tion, Manassas, Virginia) and PLC/PRF/5 (PLC; product CRL 8024; ATTC). Cells were cultured in high-glucose Dulbecco’s modified Eagle medium (DMEM) containing glutamine   (product   11965;   Thermo   Fisher   Scientific, Waltham, Massachusetts), 1% penicillin/streptomycin, poly- L-lysine (product P4707; Millipore Sigma, Burlington, Massachusetts), and nonessential amino acid (product M7145; Millipore Sigma), with or without 10% fetal bovine serum (FBS) (product F8192; Millipore Sigma) supple- mentation. Cells were maintained in a humidified 37oC cell culture chamber with 5% CO2.

 

Percutaneous Thermal Ablation Simulation

Cultured cells were trypsinized, suspended, and placed in 1.5-mL tubes in a water bath set at 37oC (control) or 46.5oC for 10 minutes, the median lethal dose (LD50) temperature. The cells were subsequently replated on cul- ture plates, and the medium was changed twice daily to remove dead cells and debris. On day 3, the cell density was adjusted to equalize the density between the simulated percutaneous thermal ablation (sPTA) and control condi- tions. Cells were analyzed on days 4 or 5, when confluency reached 70%. Transarterial Embolization Simulation Cultured cells were switched to FBS-free medium and placed in a hypoxia chamber at 1% O2 for 72 hours, the time point determined in preliminary experiments at which there are detectable changes in the remaining cells.

 

Transarterial Chemoembolization Simulation

Cultured cells were treated as for simulated transarterial embolization (sTAE), with 2 nM of doxorubicin added to the medium during the first 48 hours. The half maximal inhibitory concentration (IC50) for doxorubicin was determined by MTT assay (described below, where MTT is 3-(4,5-dimethylthiazol-2-Yl)-2,5-diphenyltetrazolium bromide).

 

Metabolic Stress Through Glutamate Deprivation

Cultured cells were incubated in DMEM lacking glutamine (product 11960015; Thermo Fisher Scientific) for at least 7 days prior to analysis. This time was sufficient to observe detectable changes in the remaining cells, as determined in preliminary experiments.

 

Identification and Characterization of hCSC

Antibodies used for flow cytometry, Western blotting, and immunofluorescence included the following: anti-human CD44   (IM7)-APC   (1:200   dilution;   code   17-0441-82; Thermo Fisher Scientific), anti-human CD133 (TMP4)-PE antibody (1:200 dilution; code 12-1338-42; Thermo Fisher Scientific), anti-human EpCAM antibody (1:200 dilution; code 324223, Biolegend, San Diego, California), anti-HIF1a monoclonal antibody (1:1000 dilution; immunofluorescence dilution 1:100; code ERP16897; Abcam, Cambridge, Mas- sachusetts), anti-b-catenin antibody (H-102) (Western blot- ting dilution 1:1000; immunofluorescence dilution 1:100; code SC7299, Santa Cruz Biotechnology, Dallas, Texas), and anti-GAPDH monoclonal antibody (Western blotting dilution 1:5000; code 2118S; Cell Signaling Technology, Danvers, Massachusetts). For flow cytometry, HCC cells were labeled with the following anti-human antibodies: CD44-APC, CD133-PE, or EpCAM-BV605. Labeled cells were analyzed using a Fortessa or Canto analyzer (BD Biosciences, Franklin Lakes, New Jersey).

Stemness was assessed by colony-forming assay (in HepG2 cells) or sphere-forming assay (in PLC cells), depending on the growth tendencies of each cell line. For the soft agar colony-forming assay, HepG2 cells were suspended in medium containing 0.6% Noble agar and plated at 5000 cells per well in 6-well plates. After 3 weeks, the time required in preliminary experiments to permit colony growth, the colonies were counted. For the sphere- forming assay, PLC cells were transferred to ultra-low attachment plates and cultured in the high-glucose DMEM-F12 medium (product 11320033; Thermo Fisher Scientific) supplemented with 20 ng/mL epidermal growth factor (product E5036; Sigma-Aldrich, St. Louis, Mis- souri), 20 ng/mL b-FGF (product F0291; Sigma-Aldrich), and 1% B27 supplement (product 17504044; Thermo Fisher Scientific). After 10–12 days of growth, spheres were counted.

For immunofluorescence, HepG2 and PLC cells were seeded on poly-L-lysine-coated coverslips, placed in a 24- well plate, and cultured overnight. The cells were fixed on coverslips using paraformaldehyde and permeabilized with 0.15% TritonX-100. After cells were incubated overnight with primary antibodies, they were stained with secondary antibodies conjugated either to green fluorescent protein or red    fluorescent    protein.    40,6-Diamidino-2-phenylindole (DAPI) was used as a nuclear counterstain at the mounting step. (Microscopy are shown at an original magnification of 10   or 20 ).

Proliferation was measured by using an MTT prolifera- tion assay (product M6494; Thermo Fisher Scientific). After cells were exposed to the 37oC or 46.5oC bath, they were allowed to recover for 5 days, and HCC cells were seeded on a 96-well plate at a density of 10,000 cells/well. After 48 hours of cell growth, 5 mg/mL MTT reagent was added, and the absorbance at 550 nm was measured.

For Western blotting, cells were lysed in radio- immunoprecipitation assay buffer, and protein concen- trations were standardized by using the Bradford assay (Bio-Rad Laboratories, Philadelphia, Pennsylvania). Equal amounts of proteins were run in an 8%–12% sodium dodecyl sulfate gel. The proteins were trans- ferred to nitrocellulose membranes, which were incu- bated with primary antibodies overnight at 4oC. After the incubation with horseradish peroxidase-conjugated secondary antibodies, membranes were imaged with a Bio-Rad imager.

 

Hepatocellular Carcinoma Cancer Stem Cell Inhibitors

The b-catenin inhibitors FH535 (product F5682; Millipore Sigma) and XAV939 (product S1180; SelleckChem, Hous- ton, Texas) were used. The PI3K inhibitor LY294002 (product L9908; Millipore Sigma) was used. Niclosamide (NEN) (product N3510; Millipore Sigma), an antihelminthic drug and multi-kinase inhibitor, was used. Dimethyl sulf- oxide (DMSO) (product 276855; Millipore Sigma) served as a control. The IC50 of each drug was determined by MTT assay. If the tested IC50 was greater than 10 μM, then 10 μM of the drug was used to avoid the nonspecific inhibition that occurs at higher doses. Inhibitors were administered in cells in fresh medium on the third day after sPTA because this was when the cells demonstrated elevated cyclin-D1 expression on Western blots. Cells were incubated in the inhibitor-containing medium for 48 hours.

 

Data Analysis and Statistics

Flow cytometry data were analyzed using FlowJo (Tree Star, Ashland, Oregon) and Excel (Microsoft, Redmond, Washington) software. A Student t-test or one-way analysis of variance (ANOVA) followed by post hoc pairwise compari- son using the unpaired t-test was performed using FlowJo, Caliper (Caliper Corp., Newton, Massachusetts), and Excel software. A P value of less than 0.05 was considered statis- tically significant. Institutional Review Board approval was unnecessary, as human subjects were not used in this study.

 

 

RESULTS

sPTA Stimulates Enrichment of HCC Cells with a CSC Phenotype

For both cell lines, sPTA stimulated an enrichment of cells bearing hCSC markers. Five days after sPTA, HepG2 cells demonstrated an enrichment of CD133þ/EpCAMþ and CD44þ cells (Fig 1), and PLC cells demonstrated an enrichment of CD44þ/CD133þ and CD44-/CD133þ cells (Fig E1; available online on the article’s Supplemental Material page at www.jvir.org). The HepG2 cells exposed to sPTA also demonstrated a higher proliferation rate, an effect that did not reach significance in PLC cells (Fig 2a). After sPTA, HepG2 cells exhibited significantly higher colony-forming capacity, and PLC cells exhibited higher sphere-forming capacity, both of which are stemness assays (Fig 2b, c). Next, the effects of sPTA on proteins involved in proliferation, metastasis, and self-renewal were assessed by   Western   blotting   and   immunofluorescence   assays.

Cyclin-D1 was interrogated as a marker of proliferation and Wnt7a as a marker of senescence. Metastatic potential was assayed using markers of epithelial-mesenchymal transition, which is characterized by a decrease in E-cadherin and an increase in N-cadherin. Both cell lines responded to sPTA by expressing lower Wnt7a and higher cyclin-D1, compat- ible with an escape from cell cycle arrest and a shift toward a proliferative state (Fig 3a). PLC cells expressed higher b- catenin, and HepG2 cells expressed higher p-Erk1/2 at 5 days after sPTA. PLC cells also demonstrated higher cyclin-D1 and N-cadherin and lower Wnt7a after sPTA by immunofluorescence (Fig 3b).

 

sTAE and Transarterial Chemoembolization  Simulation Stimulate Enrichment of HCC Cells with a CSC Phenotype

For both cell lines, transarterial embolization simulation (sTAE) stimulated enrichment of HCC cells with a CSC phenotype. After sTAE, HepG2 cells demonstrated enrich- ment of CD133þ/EpCAMþ cells (Fig 4a), and PLC cells demonstrated an increase in the CD44þ/CD133þ and CD44-/CD133þ populations (Fig E2; available online on the article’s Supplemental Material page at www.jvir.org). HCC cells also exhibited increased expression of CSC pathway proteins by Western blotting at 24 and 72 hours CD44, a reported CSC marker, was not detected in HepG2 cells. Graphic representation of the flow cytometry data: mean CD133þ/EpCAMþ expression level was 28.94% after normal conditions and 46.96% after sTAE treatment. (b) Western blotting was performed using HepG2 cells (left) and PLC lysates (right) under normal conditions (in duplicate) or for 24 hours (in duplicate) or 72 hours (in triplicate) after sTAE conditions. GAPDH served as the loading control. sTAE stimulated expression of phosphorylated Akt (p-Akt) and phosphorylated Erk1/2 (p-Erk1/2) in HepG2 cells at 72 hours and expression of p-Erk and hypoxia-inducible factor-1a (HIF-1a) in PLC at 24 hours after sTAE. However, cyclin-D1 levels were suppressed by sTAE in a manner that was partially recovered after 72 hours in PLC cells. FBS ¼ fetal bovine serum after sTAE (Fig 4b). HepG2 and PLC expressed increased p-Erk1/2 at 72 (HepG2) and 24 (PLC) hours after sTAE. HepG2 demonstrated an increase in p-AKT at 72 hours after sTAE. HepG2 cells and, to a lesser extent PLC cells, exhibited lower cyclin-D1 expression at 24 and 72 hours after sTAE, suggesting suppression of proliferation. PLC cells expressed HIF-1a at 24 hours but not 72 hours after sTAE, which was not observed in HepG2 cells. Transarterial chemoembolization simulation elicited effects similar to those of sTAE. After transarterial chemoembolization simulation, HepG2 cells demonstrated an enrichment of CD133þ/EpCAMþ and CD44þ populations, and PLC cells exhibited an increase in CD44þ/CD133þ cells (Fig E3; available online on the article’s Supplemental Material page at www.jvir.org). Given that stress from conditions simulating PTA, TAE, and transarterial chemoembolization induced hCSCs, PLC cells were exposed to metabolic stress by glutamate deprivation to determine whether it would also stimulate hCSCs. Long-term glutamate deprivation resulted in enrichment of the CD44þ and CD44þ/CD133þ PLC cell populations (Fig E4; available online on the article’s Supplemental Material page at www.jvir.org). Conversely, the CD44-/CD133þ population decreased.

NEN and CSC   Pathway   Inhibitors Suppress   the   sPTA-Induced   Stimulation of hCSC Enrichment and Proliferation Conditions simulating locoregional therapies for HCC resulted in enrichment and stimulation of hCSCs, yet no clinically available treatment specifically targets this likely source of recurrence. To test whether targeted inhibitors suppress the stress-induced stimulation of hCSCs, HCC cells were exposed to sPTA and then treated with the b- catenin inhibitor FH535 or XAV939, the PI3K inhibitor LY294002, or the multi-kinase/mixed inhibitor NEN at various concentrations. Treatment with each of the 4 in- hibitors suppressed sPTA-induced enrichment of CD133þ/ EpCAMþ HepG2 cells (Fig 5), whereas only FH535 or NEN suppressed enrichment of CD44-/CD133þ PLC cells (Fig 6). FH535 treatment of control PLC cells resulted in an increase in the CD44—/CD133þ population. A significant decrease in proliferation was observed in controls and sPTA-exposed cells treated with all 4 inhibitors for the Hep3B cells and in the presence of LY294002, XAV939, and NEN (2 μM) for the PLC cells (Fig E5; available online on the article’s Supplemental Material page at www.jvir.org).

 

 

DISCUSSION

Inhibiting the source of HCC recurrence after PTA, TAE, and transarterial chemoembolization would result in more effective and durable treatment responses. The current study tested the hypothesis that the stress conditions induced by PTA, TAE, and transarterial chemoembolization stimulate hCSCs, which are inherently resistant to standard anticancer treatments. Stimulated hCSCs represent a likely source of recurrence, which could be mitigated through targeted hCSC inhibition. These data supported the hypothesis by showing that sPTA, sTAE, and transarterial chemo- embolization simulation provoked an enrichment of cells bearing an hCSC phenotype. Moreover, the enrichment of hCSCs after transarterial chemoembolization simulation was similar to that after sTAE, confirming that doxorubicin is ineffective against hCSCs. Cells that survived sPTA also exhibited greater proliferation, indicating that heat stress has a stimulatory effect on the surviving cells. Finally, these data show that NEN and inhibitors of b-catenin and PI3K sup- pressed the sPTA-induced hCSC enrichment and prolifera- tion. These data support and expand upon those of prior studies, demonstrating that conditions related to locoregional therapies stimulate cells with features sugges- tive of CSCs (2,13,14).

The tested stress conditions induced enrichment of cells bearing hCSC markers and exhibiting the CSC property of self-renewal using colony-forming and sphere-forming as- says. Populations bearing different hCSC markers were identified in the HepG2 and PLC cells, reflecting the hetero- geneity of the human HCCs from which they were derived.

This observation is concordant with published data demon- strating different CSC markers across different cell lines and in different patients (33,34). It is not surprising that these data also revealed cell line-specific differences in the expression of CSC signaling pathway components and vulnerability to in- hibitors. The heterogeneity observed among these 2 cell lines predicts variations in future studies in vivo.

Whereas both HepG2 and PLC cells demonstrated elevation of the proliferation marker cyclin-D1 after sPTA, only HepG2 cells exhibited significantly increased prolif- eration in MTT assays. The nature of this discrepancy will be explored in future experiments. Also, proliferation as measured by cyclin-D1 was increased after sPTA but was suppressed at 24 hours after TAE. This difference may reflect the 72-hour duration of sTAE stress exposure versus 10 minutes for sPTA, resulting in a longer recovery period for the cells exposed to sTAE. Alternatively, the nature of the stress could exert differential effects on the proliferative capacity of the surviving cells. This and the effect of sTAE on proliferation after 72 hours and the effects of hCSC in- hibitors on sTAE-induced hCSC enrichment will be the focus of future studies.

The current study provides insight into the pathways activated in HCC cells after exposure to conditions simu- lating locoregional therapies. Proteins associated with the Wnt/b-catenin, PI3K-AKT-mTOR, and MAPK/ERK signal transduction pathways involved in CSC renewal increased after sPTA and sTAE, as observed in prior studies (8–11).

Cells that survived sPTA demonstrated a change in protein expression compatible with an escape from senescence, increased proliferation, and increased epithelial-to- mesenchymal transition, suggesting greater malignant and metastatic potential. HIF-1a, known to stimulate CSCs (6), was induced at 24 hours but not at 72 hours after sTAE in PLC cells, suggesting a potential mechanism for sTAE- induced hCSC enrichment. Future studies will examine the effects of sPTA on HIF-1a, as HIF-1a upregulation has also been observed after PTA with radiofrequency ablation in rats (9). It is unclear why HepG2 cells did not exhibit the sTAE-related increase in HIF-1a observed at 24 hours in PLC cells. Whether this phenomenon is cell line-specific or it occurs at times not assessed in the current study remains to be determined. The observation that NEN suppressed sPTA- induced hCSC enrichment and proliferation in both cell lines is likely due to the multiple mechanisms by which NEN inhibits hCSCs. The specific b-catenin and PI3K in- hibitors suppressed both enrichment and proliferation of the HepG2 cells, but the pattern was more mixed for the PLC cells, suggesting cell line-specific differences in vulnera- bilities to the specific inhibitors. Further studies are neces- sary to understand the mechanisms whereby hCSC inhibitors suppress the stress-induced stimulation of hCSC enrichment and proliferation.

This research has limitations related to the in vitro models used and the simplified conditions simulating locoregional therapies. Although in vitro studies are necessary precursors to experiments in animals, they cannot fully reflect the complexity of the pathophysiology of HCC or its in- teractions with the host immune response. Although these cell lines were derived from human HCCs and thus reflect the heterogeneity observed in patients, their immortalization results in a much larger proportion of hCSCs than those in vivo. Expansion of the hCSC population facilitates the study of changes in response to stress conditions and in- hibitors but could impact these observations, which require in vivo confirmation. The in vitro design also cannot predict the effect of inhibitors on normal tissues and potential sys- temic toxicities, which could render them unsafe in patients, especially for those with borderline liver function. Although the stress-induced hCSC enrichment and stimulation of proliferation suggest a mechanism for recurrence, the in vitro design does not permit direct testing of recurrence or the effect of the hCSC inhibitors on recurrence. Finally, the conditions simulating PTA, TAE, and transarterial chemo- embolization are necessarily simplified but do not represent the complete response elicited by these treatments in vivo. The implications of this study are that combining PTA, TAE, or transarterial chemoembolization with adjuvant hCSC inhibitors could potentially reduce HCC recurrence at its source. The ideal adjuvant would have a broad anti- hCSC effect, low systemic side effects, and a durable response. NEN represents a promising candidate, as it is well tolerated and inhibits HCC through various mecha- nisms. NEN has the additional benefit of inhibiting hepatic steatosis, which is a risk factor for HCC. Approval of NEN by the FDA makes clinical translation more straightfor- ward. If NEN proves efficacious in vivo, future clinical trials could include locoregional therapy plus NEN administered systemically (orally or intravenously) or directly (percutaneously or intra-arterially) at the time of locoregional therapy or immediately before or after treatment, whichever proves most effective.

In conclusion, these data show that stress conditions associated with locoregional therapies stimulate hCSC enrichment and proliferation, which can be suppressed by XAV-939 and inhibitors of pathways important for hCSC renewal. Future studies in vivo will assess whether combining locoregional therapies with adjuvant hCSC in- hibitors reduces recurrence.