Membrane-to-Nucleus Signals and Epigenetic Mechanisms for Myofibroblastic Activation and Desmoplastic Stroma: Potential Therapeutic Targets for Liver Metastasis?

Introduction

Cancer Associated Fibroblasts from Liver Metastases Originate Predominantly from Hepatic Stellate Cells

HSTCs are liver specific pericytes that reside in the space of Disse between sinusoidal endothelial cells and hepatocytes, where they are quiescent and store vitamin A. They control ECM turnover and release paracrine, autocrine, juxtacrine, and chemoattractant factors to regulate blood flow of hepatic vasculature and maintain homeostasis of the liver microenvironment (12). HSTCs are a key contributor to liver fibrosis in response to injury by activating into myofibroblasts responsible for the excessive ECM deposits, growth factors and cytokines (13) (14). Quiescent HSTCs express neuronal cell markers, including glial fibrillary acidic proteins (GFAP), and desmin (15). When cancer cells are colonized in sinusoidal area of liver lobules, they extravasate into the space of Disse to make a direct contact with desmin positive HSTCs to induce their transdifferentiation into CAFs (10) (16), which are phenotypically distinct from the lipid-droplet rich quiescent HSTCs. During HSTC activation, cells shed their lipid droplets, develop α-SMA positive stress fibers, and express additional mesenchymal markers including vimentin, PDGF receptor alpha (PDGFR-α) and beta (PDGFR-β). Activated HSTCs are proliferative, migratory, contractile and fibrogenic, contributing to the formation and remodeling of the desmoplastic stroma of liver metastases.

Cell lineage tracing in genetic modified mouse models demonstrate that HSTCs give rise to 82-96% of hepatic myofibroblasts under various liver injuries (14), and that mesothelial cells, which constitute a single layer of flat epithelial cells covering the liver, are another progenitor of hepatic myofibroblasts via mesothelial-mesenchymal transition (17). Portal fibroblasts may be another contributor to hepatic myofibroblasts (18). Although epithelial-to-mesenchymal transition (EMT) has been proposed as a mechanism for myofibroblastic activation in the liver, genetic labeling of hepatocytes and cholangiocytes in mice and tracing of the labeled cells in liver injury animal models failed to support this notion (19) (20). Bone marrow derived fibrocytes and mesenchymal stem cells can be activated into hepatic myofibroblasts (15). However, contribution of these cells to CAFs found in liver metastases remains to be determined and, therefore, this topic offers unique opportunities for further experimentations.

Bench and Bedside Evidence Supporting a Tumor Promoter Role of Cancer Associated Fibroblasts Activated from Hepatic Stellate Cells

Clinical studies demonstrate that gene expression signatures obtained by expression profiling of CAFs predict poor clinical outcome of patients with lung cancer, colon cancer, invasive ductal breast carcinoma, and esophageal squamous cell carcinoma (21) (22) (23) (24). In liver metastases of patients affected by colon, gastric or pancreas adenocarcinomas cancers, CAF/activated HSTCs are easily detected by α-SMA staining, which correlate with the degree of fibrous stroma (desmoplasia) (25). Studies, primarily performed during the last decade have found that, in patients with hepatocellular carcinoma (HCC) or intrahepatic cholangiocarcinoma (ICC), the presence of desmoplasia or, at least of intratumoral CAF/activated HSTCs, correlates with the occurrence of larger tumors sizes and poor patient survival (26) (27). Consequently, these discoveries have fuelled a large amount of studies, which seek to identify CAF associated molecules as either mechanistic regulators or biomarkers for these processes. For instance, Utispan K et al. compared gene expression profile of CAF/activated HSTCs of cholangiocarcinoma (CCA) patients with that of normal liver fibroblasts and they found that periostin was overexpressed in CAF/activated HSTCs of CCA patients. Furthermore, they found that patients with high levels of periostin in their cancer had significantly shorter survival time than those with low levels, indicating CAF/activated HSTCs derived ECM components such as periostin as prognostic markers for CCA patients (28). With a similar goal, Badioda et al., have implanted tumor cells into the liver of control and discoidin domain receptor 2 (DDR2) knockout mice via intrasplenic injection and they found that DDR2 mice developed three folds more liver metastases than control mice, which contained higher densities of CAF/activated HSTCs (29). Similarly, mice deficient for IQ motif containing GTPase activating protein 1 (IQGAP1) developed significantly more liver metastases than control mice, which associated with more CAFs/activated HSTCs (10). In vitro, conditioned medium of activated HSTCs stimulated tumor cell proliferation, migration and invasion (27) (30) and induced EMT phenotype of tumor cells (31). In a tumor/HSTC coimplantation model, HSTCs promoted tumor initiation and progress in mice (10) (27) (32), and in a rat ICC model, depletion of CAF/activated HSTCs by a cytotoxic drug navitoclax resulted in reduction of tumor size and metastasis secondarily (33). Thus, the desmoplastic reaction, which forms the tumor microenvironment of liver metastases, is not just a response or a barrier to invasion of tumor cells, but rather a niche for cancer cells to implant and progress (34). Consequently, the primary cell constituent of this tissue, CAFs, provide a large amount of currently known cellular mechanisms and molecular mediators that are essential for modulating cancer growth.

Cancer Associated Fibroblasts Derived from Activated Hepatic Stellate Cells Promote Liver Metastasis by Paracrine actions

Activated HSTCs promote liver metastases by multiple mechanisms (Fig. 1): (a) they are sources of PDGF, HGF, SDF-1/CXCL12, TGF-β and so on, which are potent mitogens and chemoattractants of cancer cells (10) (16) (35). Colorectal cancer cells express CXCR4, receptor of SDF-1/CXCL12, and this SDF-1/CXCL12-CXCR4 pathway has been shown to mediate liver specific metastasis of these tumors (35). Additionally, TGF-β promotes migration and invasion of tumor cells by inducing their EMT transition. (b) MMPs, released from CAFs/activated HSTCs, facilitate tumor cell migration and invasion by breaking down ECM and cleaving E-cadherin and disassembling adherens junctions of tumor cells (36). In addition, these proteases are involved in the processing of signaling molecules. For instance, TGF-β is usually anchored in the ECM by a covalently link to a large glycoprotein called latency associated protein (LAP) and it is released by MMPs through proteolytic cleavage of LAP from the ECM (12) (37). (c) TGF-β suppresses anti-tumor responses of a variety of inflammatory cells, including natural killer (NK) cells, dendritic cells, macrophages, neutrophils, CD8+ and CD4+ T cells and regulatory T cells. It has been shown that activated HSTCs inhibit T-cell responsiveness, T-cell mediated cytotoxicity and accelerate T-cell apoptosis by releasing B7-H1 (38) (39). (d) Activated HSTCs regulate tumor angiogenesis by releasing angiogenic factors such as VEGF, angiopoietin 1 or 2, and MMP9 (29) (30) (40). Additionally, ECM components, such as fibronectin, collagen and laminin, facilitate tumor angiogenesis by activating integrin mediated signaling in endothelial cells (41). (e) The desmoplastic and hypovascularized stroma formed by activated HSTCs and ECM acts as a physical barrier for drug delivery and confers chemo- and radio- resistance to cancer tissue. For example, activated HSTCs confer tumor resistance to cisplatin by releasing HGF which activates Met on HCCs to promote cancer cell survival (31). Activated HSTCs contribute to chemo resistance of cholangiocarcinoma as well by simulating production of IL-1α, C5α, IL-1β and G-CSF cytokines in a HSTC/tumor cell coculture (27). Furthermore, ECM and inflammatory cytokines in the tumor stroma protect tumor cells from radiation therapy-induced death in part by activating integrin mediated intracellular signaling of tumor cells (42). Thus, CAFs/activated HSTCs are emerging as a bonafide target of therapies that seek to improve the efficacy of both chemotherapy and radiation therapy (5) (34).

Key Intracellular Signaling Pathways for Hepatic Stellate Cells Activation

Myofibroblastic activation of HSTCs requires paracrine signals from cancer cells and other cell populations within the hepatic microenvironment to initiate intracellular signaling of HSTCs (Fig. 2). Below we list key signaling cascades governing HSTC activation. We need to point out that HSTC activation under tumor invasion is a relatively new and understudied topic as compared to HSTC activation under fibrotic stimuli. Some mechanisms presented here are therefore adapted from those in liver fibrosis studies, which we think are applicable to both liver diseases. Furthermore, each receptor mediated intracellular signaling does not act alone. In fact, it often cross-talks, regulates or converges with others to form complexes which function as an interconnected signaling network for HSTC activation (Fig. 2).

Extracellular Matrix initiated Mechanotransduction for HSTC Activation

Progression of cancer and fibrosis caused by overproduction, deposition and cross-linking of ECM proteins leads to a stiff microenvironment. In this regard, transglutaminases, lysyl oxidases and prolyl hydroxylases are three examples of enzymes that mediate ECM cross-linking (56). Cells respond to physical signals of the ECM by forming integrin- and actomyosin-linked focal adhesions that sense mechanical force and stiffness of the ECM and initiate mechanotransduction of the cell. Thus, in addition to soluble biochemical factors, mechanotransduction induced by changes of the stiffness of the surrounding ECM has been proposed as another key factor for myofibroblastic activation of cells. Indeed, Olsen et al. have shown that when cultured in a stiff environment, freshly isolated rat HSTCs differentiated into myofibroblasts without TGF-β stimulation (67). In fact, TGF-β stimulation potentiated HSTC activation in response to a stiff environment (67). In the original study, in which this phenomenon was described, HSTCs were cultured on insert polyacrylamide supports of variable but precisely defined shear modulus (stiffness) ranging from 0.4 to 12 kPa, and cell morphology and α-SMA IF were used to assess HSTC activation. HSTCs cultured on a support of 0.4-1 kPa, representing the stiffness of normal liver (0.3-0.6 kPa), failed to differentiate into myofibroblasts. In contrast, cells on a support of 8-12 kPa showed myofibroblastic morphology and α-SMA positive stress fibers. Interestingly, when plated on Teflon, which is 4-5 orders of magnitude stiffer than cirrhotic liver, HSTCs failed to differentiate into myofibroblasts because they were not able to generate an effective mechanical tension on the polymer Teflon (67). Together, these data support the notion that HSTC activation requires mechanical tension generated within the cell in response to a stiff environment.

It appears that increased matrix stiffness leads to increased cell-generated tension, which ultimately leads to α-SMA expression and ECM deposition of activated HSTCs. Thus, it is critical to define the signaling pathways that are activated by mechanical signals and, which in turn, transduce these signals into defined gene expression patterns within the nucleus. In this regard, recent studies have begun to identify several signaling pathways as mediators of mechanotransduction of the cell. For example, as noted earlier, TGF-β is secreted as a latent form and anchored into ECM by LAP which directly binds to certain integrins. Single molecule analysis demonstrates that release of the active form of TGF-β is mechanosensitive; if cells are attached to a soft ECM, LAP-TGF-β complexes remain intact. On the other hand, if cells are attached to a stiff ECM, the resistance of ECM against cell-generated tension leads to LAP-TGF-β breakdown and the consequent release of active TGF-β (57). Of the numerous proteins within the focal adhesions, FAK is a key molecule mediating mechanotransduction. Besides the capability of FAK to crosstalk with both the PI3K and MAPK pathways, its activation also couples to RhoA/ROCK signaling (68), a step which is essential for generating actomyosin-dependent tension and reestablishment of force equilibrium within a cell. Samuel et al. demonstrated that overexpression of ROCK2 in skin of mice activated RhoA/ROCK signaling and led to tissue stiffening, intracellular tension, β-catenin nuclear translocation and β-catenin mediated hyperproliferation of epidermal cells in mice, all effects which were abrogated by inhibitors of actomyosin contractility (69). Although these findings were originally reported in epidermal cells, we believe that it is likely that FAK serves as an important mechanotransducer for HSTC activation. Complementing this observation, Yes-associated protein (YAP) and TAZ (WW domain-containing transcription regulator protein 1, also called WWTR1) have been recently identified as novel mechanosensors and mechanomediators in response to cues from the cellular microenvironment (70). YAP/TAZ become oncogenic through their ability to interact with the transcription factors TEAD/TEF in the nucleus to regulate cancer-associated gene expression networks (71). YAP/TAZ are downstream effectors of Hippo signaling, a pathway that controls cell proliferation and organ size. In the Hippo signaling cascade, cell-cell contacts activate LATS1/2 serine/threonine kinases, which induces phosphorylation and sequestration of YAP/TAZ in the cytoplasm and their subsequent ubiquitin-mediated degradation (72). Dupont et al. found that ECM stiffness and cell spreading increased nuclear accumulation and function of YAP/TAZ, which was abrogated by Rho inhibitor C3 or actin depolymerization agent latrunculin A (70). Additionally, Calvo et al. found that YAP of CAFs isolated from mouse mammary carcinomas was nuclear as compared to cytoplasmic YAP found in normal fibroblasts used as control (73). Furthermore, YAP in the nucleus of CAFs promoted the expression of ANLN and DIAPH3 which stabilized actomyosin fibers and led to further ECM stiffening required for maintenance of CAF phenotypes (73). Thus, YAP/TAZ, which are novel mechano sensors and transducers of the cell, may play important roles in the process of myofibroblastic activation of HSTCs.

Epigenetic Mechanisms for HSTC Activation

The receptor mediated intracellular signaling cascades and ECM medicated mechanotransduction, as mentioned above, eventually converge in the nucleus to regulate gene transcription. Indeed, the downstream effectors of these signaling pathways are often transcription factors and chromatin modification proteins that together constitute epigenetic mechanisms to induce the expression of gene networks responsible for the activated HSTC phenotype (Fig. 2). Thus, the epigenetic mechanisms are an obliged requisite for myofibroblastic activation of HSTCs.

Conclusions and Perspectives on Therapeutic Targeting of CAF/activated HSTCs

Liver metastasis is dependent on bidirectional interactions between cancer cells and the microenvironment of the liver. HSTCs are one of important components of prometastatic liver microenvironment for their transdifferentiation into CAFs critical for tumor implantation, propagation and invasion in the liver. CAFs/activated HSTCs also confer chemo and radio resistance of cancer. Activation of HSTCs into CAFs is regulated by a complex interconnected intracellular signaling network and downstream orchestrated gene transcriptional events to induce genes responsible for activated HSTC phenotype and repress genes responsible for quiescent HSTC phenotype. CAF/activated HSTCs thus present a therapeutic target for preventing or reducing liver metastasis, improving drug delivery, and reducing tumor resistance to anti-cancer therapy.

Therapeutic targeting of cancer cells specifically has progressed slowly for decades because each patient's cancer cells harbor individual genetic mutations and display unique biological behaviors, making them difficult to target. CAF/activated HSTCs are targetable because they are significantly more sensitive than tumor cells to cytotoxic drugs such as navitoclax (ABT-263), a BH3 mimetic (33). In rats orthotopically transplanted with ICC tumor cells, navitoclax treatment increased apoptosis of CAF/activated HSTCs, which resulted in reduced tumor size and metastasis, providing a “proof-of-principle” for depletion of CAF/activated HSTCs as an anticancer strategy for tumor growth in the liver (33) (34). In a mouse model of pancreatic ductal adenocarcinoma, combined therapy with gemcitabine to target tumor cells and Hh inhibitor IPI-926 to target desmoplastic stroma, improved delivery and efficacy of gemcitabine, and extended median survival of mice as compared with either drug alone (4). Congruently, in phase I/II and phase III clinical studies, combination of nab-paclitaxel (Abraxane) to cause depletion of tumor stroma and Gemcitabine significantly extended survival of patients with advanced/metastatic pancreatic cancer as compared to Gemcitabine alone (86) (87). Based upon these findings, nab-paclitaxel plus gemcitabine has received FDA approval in 2013 as a first-line treatment for patients with metastatic pancreatic adenocarcinoma. Interestingly, however, using the same experimental paradigm (4), Rhim at al. recently demonstrated that IPI-926 alone or combined with gemcitabine, both are known to target CAFs, shortened survival of mice (88). Additionally, CAF depletion in a genetic modified pancreatic cancer mouse model shortened mouse survival but enhanced the efficacy of anti-CTLA4 (Ipilimumab) immunotherapy (7). Together, these data reveal that, though it is still poorly understood, targeting CAFs is beneficial for ameliorating pancreatic cancer in experimental animal models. Thus, more in depth investigations are needed to more solidly establish the beneficial outcomes of combinational therapies that target both CAFs and tumor cells for the treatment of liver metastasis.

PDGF and TGF-β signaling are two major intracellular signaling pathways that control HSTC activation. We recently found that activated HSTCs express a high level of PDGFRα in addition to PDGFRβ and that PDGFRα is required for TGF-β signaling in HSTCs, highlighting PDGFRα as a target for inhibiting both PDGF and TGF-β signaling of HSTCs (51). Crenolanib besylate, a drug currently under investigation (89), and approved anti-cancer drug Imatinib (Gleeve, ST1-571) and Sunitinib, are designed to inhibit both PDGFR receptors. It is worth investigating if these drugs inhibit both PDGF and TGF-β signaling of HSTCs and suppress both tumor cells and the tumor microenvironment. Targeting TGF-β signaling of the tumor microenvironment is a promising direction in cancer therapy for tumors that lost TGF-β mediated tumor suppression pathways (90). Drugs targeting TGF-β signaling are still under investigation with some of them showing encouraging results in clinical trials. Besides IPI-926, GDC-0449 is another promising Hh inhibitor currently being tested in clinical trials. GDC-0449 has been shown to inhibit accumulation of liver fibroblasts and induce HCC regression in aged mice deficient for multidrug resistant gene 2 (Mdr2−/− mice) (59). Furthermore, continuously elucidating epigenetic regulation of gene expression for HSTC activation may help lead to new directions for drug development, for instance, delivering miRNAs or screening for novel compounds that target histone methylation or acetylation to inhibit myofibroblastic activation of HSTCs and the tumor microenvironment.