A microRNA-7/growth arrest specific 6/TYRO3 axis regulates the growth and invasiveness of sorafenib-resistant cells in human hepatocellular carcinoma

CELL CULTURE

The Huh-7 and Hep3B cell lines used in this study were kindly provided by Associate Professor Nicholas Shackel, Centenary Institute, Sydney, Australia. Two sorafenib-resistant sublines were established, namely Huh-7/SR1 and Huh-7/SR2, by culturing the cells in increasing doses of sorafenib. The origin and culture conditions of these cells are detailed in the Supporting Materials and Methods.

REAGENTS

A full list of reagents is included in the Supporting Materials and Methods.

CELL VIABILITY, PROLIFERATION, AND COLONY FORMATION ASSAY

To determine the sensitivity of HCC cell lines to sorafenib, miR-7, and TYRO3, cells were treated with increasing concentrations of drug, miRNA, and siRNA, or after transfection with a single concentration of miRNA/siRNA, and the effect on cell viability was measured over time, as described.18 Two-dimensional and three-dimensional colony formation assays were performed, as described.15 Further details are outlined in the Supporting Materials and Methods.

MIGRATION AND INVASION ASSAYS

Migration and invasion were assessed using transwell assays with polycarbonate membrane inserts with 8.0 μm pore size (Costar) coated with or without Matrigel (Corning), as described in Supporting Materials and Methods.

RNA EXTRACTION, QUANTITATIVE REVERSE-TRANSCRIPTION POLYMERASE CHAIN REACTION, AND miRNA TARGET VALIDATION (LUCIFERASE REPORTER ASSAYS)

Full methods are described in Supporting Materials and Methods.

WESTERN BLOT ANALYSIS

Immunoblotting was performed using antibodies detailed in the Supporting Materials and Methods. Densitometry was performed using ImageJ version 1.51h software (http://rsb.info.nih.gov/ij/).

PHOSPHO-RECEPTOR TYROSINE KINASE ARRAY

Phospho-receptor tyrosine kinase (Phospho-RTK) array (R&D Systems) was performed according to the manufacturer's instructions. Lysates were prepared from Huh-7 cells transfected with 30 nM miRNA precursor molecules for 48 hours, and 500 μg protein was used for each array. Densitometry was performed using ImageJ version 1.48 software (http://rsb.info.nih.gov/ij/).

ORTHOTOPIC LIVER CANCER MODEL

For details on methodology, please see Supporting Materials and Methods.

PUBLIC ARRAY DATA

To investigate the clinical significance of TYRO3 expression in human HCC, a Kaplan-Meier survival curve was plotted using data sets from The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/, TCGA-LIHC, RNASeqV2 (data type), Illumina HiSeq 2000 (platform), downloaded and accessed October 2015) in the R survival package. For information about patient samples and expression analysis, refer to Supporting Materials and Methods.

STATISTICAL ANALYSIS

All results are represented as mean ± SEM except for quantitative reverse-transcription polymerase chain reaction (qRT-PCR), which is presented as mean ± SEM of ΔΔCt. Graph Pad Prism (version 7.0, GraphPad Software) was used to calculate statistical significance between test (e.g., miR-7-5p/siTYRO3) and control (e.g., miR-negative control [NC]/NC siRNA) treated groups using the Student t test (one-tailed, unpaired), one-way analysis of variance (ANOVA), two-way ANOVA, ANOVA with repeated measures, and Mann-Whitney U test as appropriate. P < 0.05 was considered statistically significant wherein *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

miR-7 IS A POTENT SUPPRESSOR OF HCC GROWTH IN VITRO AND IN VIVO

The constitutive activation of EGFR in HCC leads to persistent downstream signaling through the PI3-Kinase and AKT, p21-activated kinases, and Raf/MEK/ERK pathways, transducing growth signals to tumor cells and promoting tumor progression and dissemination.19 As miR-7 is down-regulated in HCC,13, 20 its replacement should reduce EGFR signaling together with other pathways containing miR-7 targets. We initially evaluated the effects of miR-7 overexpression in vitro using a human HCC-derived cell line Huh-7 and found that miR-7 produced a significant antiproliferative (P = 0.0435), antimigratory (P < 0.0001), and anti-invasive (P = 0.0024) phenotype in Huh-7 cells and reduced their clonogenic capacity (P < 0.01) (Fig. 1A, panel i-iv; Supporting Fig. S1A). miR-7 was cytotoxic at very low concentrations to Huh-7 (50% effective concentration [EC₅₀] = 1.514 nM), in a well-differentiated human hepatocellular carcinoma derived cell line; HEPG2 (EC₅₀ = 2.115 nM) and human hepatoma derived cell line; and PLC/PRF/5 (Alexander cell) (EC₅₀ = 0.4812 nM) cells (Fig. 1B; Supporting Fig. S1B, panel i and ii). The growth inhibitory effect of miR-7 on HCC cell viability was unrelated to the level of endogenous miR-7, which varies across the three cell lines (Supporting Fig. S1B, panel iii). miR-7 induced a G0/G1 cell-cycle arrest associated with apoptosis, as evidenced by an increase in caspase3/7 activity and cleaved poly(adenosine diphosphate ribose) polymerase protein in a western blot (Supporting Fig. S1C).

We next explored the effects of miR-7 in an orthotopic model of human HCC, using a NOV340-miR-7 mimic, as described in the Materials and Methods. The NOV340-miR-7 mimic had a similar effect on Huh-7 cell viability as the Ambion miR-7 mimic used for the in vitro work (Supporting Fig. S1D). The initial in vivo investigation found that miR-7 expression was enriched in both murine liver and human HCC cell line-derived tumor tissues following tail vein injection of the NOV340-miR-7 formulation (Supporting Fig. S1E). We then conducted a short miR-7 replacement therapy trial in which nonobese diabetic SCID IL2rγ null [NOD.Cg-PrKdc^SCIDil2rg^tm1wjl/SzJ] or NSG immunodeficient mice received Huh-7 cells orthotopically; once the cells reached a critical mass (equivalent to AFP secretion of 1,000 ng/mL), they commenced a 2-week alternate day intravenous treatment regimen of miR-7 or control (Fig. 1C). Analysis of the mice at 14 days showed that the tumor burden in the mice receiving miR-7 was dramatically reduced, accompanied by markedly lower AFP levels (almost 100-fold lower) (Fig. 1D). This was associated with a major reduction in tumor size and weight (Fig. 1F,G; Supporting Fig. S1F, panel i). Remarkably, in the miR-7 recipient group (n = 8), no visible tumors were observed macroscopically in six of the mice, explaining why there are only two tumor weights plotted for miR-7 in Fig. 1F (Supporting Fig. S1E,F panel ii). Taken together, these data indicate that miR-7 is a potent inhibitor of human HCC cell proliferation, migration, and invasion in vitro and tumor growth in vivo.

THE TAM FAMILY MEMBER TYRO3 IS ELEVATED IN HCC AND IS A NOVEL miR-7 TARGET GENE

Given the magnitude of effect induced by miR-7 in HCC, we set out to identify new targets of the miRNA within the RTK family, using an unbiased phospho-RTK antibody array. Densitometry analysis of the 42-member RTK array showed that miR-7 reduced the amount of phosphorylated EGFR and insulin-like growth factor 1 receptor, consistent with each being a validated target of miR-7 (Fig. 2A, #2 and #5, respectively). In addition, miR-7 reduced the levels of phosphorylated fibroblast growth factor receptor 4 (FGFR4) and TYRO3 (Fig. 2A, #3 and #6, respectively), and Target Scan analysis revealed FGFR4 and TYRO3 contain putative miR-7 seed sequences in their 3' untranslated region (UTR) (Supporting Fig. S2A) that are well conserved across primates. FGFR4 has been implicated in HCC and is a potential therapeutic target.21 Small molecule inhibitors have been developed against FGFR4 (e.g., BLU9931), and much is known about its expression in HCC.22 In contrast, interest in TYRO3 in HCC is just emerging, and given its association with resistance to both chemotherapy and targeted therapies in chronic myeloid leukemia and ovarian cancer,23, 24 we speculated that aberrant TYRO3 expression might be an important determinant of SR in HCC. To date, there are no specific small molecule TYRO3 inhibitors that have been developed as they either target all of the TAM family members or Axl alone. Thus, miR-7, by directly targeting TYRO3, could represent an alternative approach to specifically inhibit TYRO3 expression in HCC.

When miR-7 was overexpressed in Huh-7 or HEP3B cells, we observed TYRO3 mRNA and protein expression were significantly reduced (Fig. 2B; Supporting Fig. S2B). Furthermore, we identified TYRO3 as a direct functional gene target of miR-7 by using a luciferase reporter assay with constructs containing either the wild-type or mutant TYRO3 3'-UTR (Fig. 2C). Transient overexpression of miR-7 significantly reduced luciferase activity of the wild-type TYRO3 but not the mutant compared to the miR-NC (Fig. 2C, panel ii). Additional clinical validation of the importance of TYRO3 expression in HCC was achieved within a cohort of 368 patients from the TCGA; this showed that TYRO3 expression is significantly increased in tumor compared to surrounding normal tissue (Fig. 2D). Further, patients with HCC with higher expression of TYRO3 had a significantly worse prognosis and reduced survival (Fig. 2E).

TYRO3 REGULATES PROLIFERATION, MIGRATION, AND INVASION OF HCC CELLS THROUGH THE PI3-KINASE/AKT PATHWAY

To evaluate the effects of regulating TYRO3 expression in HCC (Huh-7 and HEP3B cells), we used two commercially available TYRO3 siRNAs (named #5 and #6), each targeting different regions at the 3′-end of TYRO3 mRNA. This region is distinctive among all three transcript variants ensuing from alternative 5′-end splicing, which can regulate subsequent biological function.25 We found a functional difference between the two siRNAs in migration assays whereby TYRO3 #6 siRNA did not alter migration in HEP3B cells (Supporting Fig. S3). However, using both the xCELLigence and transwell migration assay, TYRO3 siRNA #6 was found to be more potent than TYRO3 siRNA #5 in Huh-7 cells as it decreased both migration (P < 0.0001) and invasion (P = 0.0043) (Fig. 3B,C; Supporting Fig. S3). TYRO3 function has been previously examined in HEP3B but not Huh-7 cells.6 Thus, we elected to investigate TYRO3 function in Huh-7 cells for all the functional assays in this study. Transient knockdown of TYRO3 in Huh-7 cells significantly reduced their rate of proliferation (P < 0.05) (Fig. 3A). To identify TYRO3-induced signaling pathways involved in oncogenic processes in HCC, the siRNA-mediated TYRO3-depleted Huh-7 cells were analyzed for various downstream signaling molecules. For the PI3-Kinase/AKT pathway, the amount of phosphorylated AKT (serine 473) is considerably reduced in TYRO3-deficient Huh-7 cells, whereas phosphatase and tensin homolog (a negative regulator of the pathway19) remained intact (Fig. 3D). In contrast, for the nuclear factor kappa B and RAF/MEK/ERK pathways, phosphorylated transcription factor p65 (RelA; serine 536) and phosphorylated ERK1/2 (threonine 202/204) are increased on TYRO3 depletion (Fig. 3D, ERK1/2 increased with siTYRO3 #6 only). These results suggest that the canonical PI3-Kinase/AKT pathway transduces the protumorigenic signaling of TYRO3 in HCC. Furthermore, in cells in which TYRO3 expression is reduced, alternative pathways may become active, such as RelA/nuclear factor kappa B signaling. Of note, we have previously reported that RelA is a direct target of miR-7.15

ENDOGENOUS miR-7 REGULATES THE TYRO3-MEDIATED METASTATIC PHENOTYPE IN HCC

To comprehend the biological significance underlying the miR-7/TYRO3 axis in Huh-7 cells, “rescue of function” experiments were conducted whereby the endogenous miR-7 levels were depleted using anti-miR-7 for 48 hours followed by sequential depletion of TYRO3 for a further 24 hours (Supporting Fig. S4A). We hypothesized that if TYRO3 is a physiological direct target of miR-7, elimination of endogenous miR-7 would induce a phenotype characteristic of TYRO3 overexpression. Subsequent introduction of TYRO3 siRNAs should antagonize this gain of function. We found by using anti-miR-7, depletion of endogenous miR-7 did not affect Huh-7 cell proliferation (Fig. 4A; Supporting Fig. S4B) but significantly stimulated their migration and invasion (Fig. 4C,D). When the cells lacking endogenous miR-7 were exposed to TYRO3 siRNA, we observed anti-miR-7 rescued the level of TYRO3 protein, consistent with anti-miR-7 nullifying the effect of any endogenous miR-7 (Supporting Fig. S4C). Subsequent depletion of TYRO3 in these cells did not alter Huh-7 cell viability (Fig. 4B) but significantly antagonized the extent of migration and invasion provoked by anti-miR-7 (Fig. 4C,D). We further observed that the TYRO3-dependent prometastatic effect in HCC is partly mediated by altered expression of E-cadherin (Supporting Fig. S5; Fig. 3D). Taken together, these data provide additional evidence for an important inverse regulation relationship between miR-7 and TYRO3 in HCC.

miR-7 REVERSES THE REBOUND SURGE OF PROINFLAMMATORY CHEMOKINES UPON TYRO3 DEPLETION IN Huh-7 CELLS

Loss of TAM receptors in HCC has been associated with altered expression of some key proinflammatory cytokines and chemokines,26 which could contribute to increased growth. Thus, we investigated the regulation of expression of chemokine (C-X-C motif) ligand (CXCL)10 and interleukin (IL)-8/CXCL8, two well-characterized HCC-associated chemokines with putative miR-7 binding sites. We found with qRT-PCR that TYRO3 depletion significantly increased expression of CXCL10 (rho = 0.943; P = 0.017) and IL-8 (rho = 0.886; P = 0.033), suggestive of a rebound surge of proinflammatory chemokines (Fig. 4E). Furthermore, overexpression of miR-7 significantly decreased the expression of IL-8 (P = 0.025) and CXCL10 (P = 0.0034) in Huh-7 cells (Fig. 4F). Based on these findings, we concluded that miR-7 not only fine tunes TYRO3 expression in Huh-7 cells but also down-regulates key proinflammatory chemokines. This effect of miR-7, combined with the other data above, serves to illustrate how broad its effects are and also exemplifies its potency as an inhibitor of HCC cell viability.

SORAFENIB RESISTANCE IN HCC IS OVERCOME BY miR-7 SUPPRESSION OF TYRO3 FUNCTION

To understand the mechanisms driving acquired resistance to sorafenib in HCC, two sorafenib-resistant cell lines Huh-7/SR1 (EC₅₀ = 42.92 μM) and Huh-7/SR2 (EC₅₀ = 75.94 μM) were established (Fig. 5A). miR-7 expression was significantly reduced in both sorafenib-resistant cell lines (Fig. 5B, panel i), whereas TYRO3 expression was markedly increased in the more resistant Huh-7/SR2 cell line (Fig. 5B, panel ii). We found that the EGFR pathway becomes less active during acquiring SR (Fig. 5C). The expression of TYRO3 protein is elevated in the Huh-7/SR2 cell line consistent with its mRNA profile and is accompanied by higher levels of phospho-RelA(Fig. 5C). Phospho-AKT and phospho-ERK1/2 were differentially expressed between the two sorafenib-resistant lines (Fig. 5C). In addition, we found increased TYRO3 expression in HEP3B cells, and these cells were less sensitive to sorafenib than the parental Huh-7 cells, with an EC₅₀ of ∼14 μM (Supporting Fig. S6). Further characterization showed the two resistant cell lines were functionally very distinct from their precursors. Compared to the parental cells, the Huh-7/SR1 cells were more proliferative whereas the Huh-7/SR2 cell lines were less proliferative in nature (Fig. 5D, panel i). Both sorafenib-resistant lines were more migratory and invasive than the parental cell line (Fig. 5D, panel ii and iii). Aberrant activation of TYRO3 is reported to stimulate epithelial–mesenchymal transition in colorectal cancer and regulate tumor dissemination.27 We observed that Huh-7/SR2 cells lost their epithelial polarity by expressing more mesenchymal cell markers (α-smooth muscle actin and vimentin) and less epithelial E-cadherin compared to their parental counterparts (Fig. 5E, panel i and ii). The sorafenib-resistant cells also transcribed more TWIST1 (Fig. 5E, panel iii). Furthermore, depleting TYRO3 levels in both the parental and Huh-7/SR2 cells reduced their viability in a dose-dependent manner, and the EC₅₀ of siTYRO3 in the Huh-7/SR2 cells was much less than that of the parental cells (0.009 nM versus 2.516 nM) (Fig. 5F, panel i and ii). Finally, by using concentrations near the EC₅₀ of siTYRO3, we found that combination therapy of siTYRO3 with sorafenib (5.5 μM, ∼highest achievable dose in plasma) significantly increased the cytotoxic effect of sorafenib in the parental Huh-7 cell line (Fig. 5F, panel iii), although this effect was not synergistic (using the Bliss Additivity model28).

We next investigated the effects of miR-7 on the sorafenib-resistant cells. We observed that similar to siTYRO3, miR-7 significantly reduced viability of both Huh-7/SR1 and Huh-7/SR2 cells in a dose-dependent fashion and markedly attenuated their migration and invasion (Fig. 6A,B). miR-7 reinforced E-cadherin expression in the Huh-7/SR2 cell line and induced a global reduction of key survival molecules (TYRO3, phospho-AKT, panAKT, phospho-ERK, and total RelA), previously shown to be up-regulated in these cells (Fig. 6C, and compare to Fig. 5C). We next stimulated the Huh-7/SR2 cells with the ligand to TYRO3, namely growth arrest specific 6 (GAS6), in the presence and absence of miR-7. We observed that in the absence of miR-7, GAS6 significantly stimulated the migration and invasion of the Huh-7/SR2 cell lines, and this effect was dramatically reduced in the presence of miR-7 (Fig. 6D). Further, in the presence of miR-7, the GAS6-induced increase in migration and invasion was abolished, consistent with miR-7 effectively inhibiting the GAS6-TYRO3 pathway.29 Finally, when we treated sorafenib-resistant Huh-7 cells with a constant dose of miR-7 and increasing doses of sorafenib, the EC₅₀ curve shifted significantly downwards (Fig. 6E,F). The Bliss Additivity model28 demonstrated synergy between miR-7 (at 5 nM) and sorafenib in both sorafenib-resistant cell lines (Supporting Table S1 and S2; 0.1-75 μM and 25 μM sorafenib in SR2 and SR1, respectively).