Activation of Hepatic Stellate Cells During Liver Carcinogenesis Requires Fibrinogen/Integrin αvβ5 in Zebrafish.

Hepatocellular carcinoma (HCC) is one of the most common cancers and it usually develops from a background of liver fibrosis or inflammation. The crosstalk between tumor cells and stromal cells plays an important and stimulating role during tumor progression. Previously we found in a krasV12-induced zebrafish HCC model that oncogenic hepatocytes activate hepatic stellate cells (HSCs) by up-regulation of serotonin and activate neutrophils and macrophages by up-regulation of cortisol. In the present study, we found a novel signaling transduction mechanism between oncogenic hepatocytes and HSCs. After krasV12 induction, fibrinogen was up-regulated in oncogenic hepatocytes. We reasoned that fibrinogen may bind to integrin αvβ5 on HSCs to activate HSCs. Consistent with this notion, pharmaceutical treatment using an antagonist of integrin αvβ5, cilengitide, significantly blocked HSC activation and function, accompanied by attenuated proliferation of oncogenic hepatocytes and progression of liver fibrosis. On the contrary, adenosine 5'-diphosphate, an agonist of αvβ5, activated HSCs significantly that further stimulated the tumor progression and liver fibrosis. Interestingly, in human liver disease samples, we detected an increased level of fibrinogen during tumor progression which indicated the potential role of fibrinogen signaling in HCC progression. Thus, we concluded a novel interaction between oncogenic hepatocytes and HSCs through the fibrinogen related pathway in both the zebrafish HCC model and human liver disease samples.

Introduction

Hepatic stellate cell (HSCs) are liver specific mesenchymal cells. While quiescent HSCs are dormant, activated HSCs are involved in liver disease progression [1]. In a normal liver, HSCs as a vitamin A storing cells are in the quiescent and non-proliferative stage [2]. Upon liver injury, various cytokines, such as TGFB, IL1B and TNF, are produced to instigate the transformation of HSCs into disease-promoting state [3]. Activated HSCs are an important component in the liver tumor microenvironment (TME) and a key regulator of hepatic fibrosis and cirrhosis [4]. In human HCC patients, liver fibrosis shows a close relationship with HCC incidence and recurrence [5]. As we reported previously, serotonin activates HSCs through activation of serotonin receptor 2B and stimulates liver fibrosis, as characterized by deposition of collagen and laminin to the TME. Moreover, accelerated progression of liver fibrosis promotes tumor progression in our zebrafish HCC model [6].

Recently, it has been found that integrin, a surface receptor on HSCs could receive signal from TME and is critical to liver fibrosis [7]. Integrin signaling has been shown to be crucial for HSC survival and function. Disruption of integrin on HSCs inhibits HSC proliferation and induces HSC apoptosis through increase of Caspase 3 activity [8]. Pharmacological blockage of integrin attenuated liver fibrosis in carbon tetrachloride-induced hepatic fibrosis in mice [9].

We have previously generated several inducible HCC models in zebrafish by overexpression of an oncogene specifically in hepatocytes [9], [10], [11], [12], [13]. In our kras-expressing zebrafish HCC model, we have observed a significant liver enlargement in 6-dpf (day post fertilization) larvae after 3 days of induction of kras expression [14]. In the current study, we found that after 4 days of doxycycline exposure, both HSC density and ratio of activation were increased significantly. A blood glycoprotein, fibrinogen, was up-regulated after kras induction in oncogenic hepatocytes (OHs). Fibrinogen has a high affinity to integrin and could activate it [15]. Pharmacological manipulation of integrin by its agonist, stimulated HSC proliferation and secretion of alpha-smooth muscle actin (a-Sma) and Tgfb1; meanwhile, we observed an accelerated tumor progression and a higher fibrosis level. In contrast, the antagonist treatment suppressed the secretion of a-Sma, Tgfb1 and attenuated both tumor and fibrosis progression. Furthermore, we confirmed that there was a progressively increased fibrinogen level in human patient samples with liver diseases, suggesting that fibrinogen signal is also involved in HCC progression in human patients.

Materials and Methods

Zebrafish Husbandry

All zebrafish were maintained according to the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and the protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of the National University of Singapore (Protocol Number: R17-0868). Tg(fabp10:rtTA2s-M2; TRE2:EGFP-kras) in a Tet-on system for induction of oncogenic kras expression in hepatocytes [10], Tg(fabp10:DsRed; ela3l:EGFP) for isolation of hepatocytes [16], and Tg(hand2:EGFP) for visualization of EGFP-labeled HSCs were used in this study and they are referred to as kras+, fabp10+, and hand2+ respectively.

Induction of Oncogene Expression and Chemical Treatments

Both kras induction and chemical treatments were conducted in larvae from 3 dpf to 7 dpf. The chemicals used included doxycycline (dox) (20 μg/ml; D9891; Sigma), cilengitide (10 μm; SML1594; Sigma) and ADP (Adenosine 5′-diphosphate) (10 μm; A2754; Sigma). The dosages were selected based on the highest all-survival concentrations as determined in preliminary experiments [17], [18].

Photography and Image Analyses

At the end of the chemical treatments, more than 20 larvae from each treatment group were selected for imaging analyses. All the larvae were anesthetized in tricaine (0.08%; E10521; Sigma) and immobilized in methylcellulose (3%; M0521; Sigma) before imaging. Each larva was photographed individually from the left lateral side using an Olympus microscope. 2D liver size were measured using ImageJ as previous described [19].

Histological and Immunocytological Analyses

All of the fish larvae were fixed in 4% paraformaldehyde (P6148; Sigma) in phosphate buffered saline, embedded in 1.5% Bacto-agar (A5306; Sigma), immersed in 30% sucrose (S7903; Sigma) in phosphate buffered saline and cryo-sectioned at 8 μm thickness using a Leica Cryostats CM1950, followed by immunofluorescence (IF) staining. The primary antibodies used included anti-proliferating cell nuclear antigen (PCNA) (FL-261; Santa Cruz Biotechnology, Dallas, TX), anti–Caspase 3 (C92-065; BD Biosciences, Singapore), anti–collagen I (ab23730; Abcam), anti-laminin (L9393; Sigma), anti-glial fibrillary acidic protein (Gfap) (154474; Abcam), anti-a-Sma (ab15734; Abcam, Singapore), anti-fibrinogen (sc69775; Santa Cruze) and anti-Tgfb1 (04-953; EMD Millipore, Billerica, MA). Anti-rabbit or anti-mouse secondary antibodies were purchased from Thermo Fisher Scientific. At least 10 fish larvae from each group were examined and at least one image was taken for each larva.

Fluorescence-activated Cell Sorting (FACS), RNA Extraction and RT-qPCR

7-dpf larvae (after removal of the tail from mid-intestine) were used for FACS using a cell sorter (BD Aria). Hepatocytes and HSCs were isolated from fabo10+ (or kras+) and hand2+ larvae respectively based on their dsRed or GFP expression. Total RNA was extracted using RNeasy mini kit (Qiagen, Hilden, Germany). QuantiTect Whole Transcriptome Kit (Qiagen) was used to synthesize and amplify cDNA, as we described previously [20]. The sequences of PCR primers used are β-actin (forward: CTCTGGGTCACCGCTTCTTT; reverse: CAGATGCTCACGAAACCCCT), fgα (forward: GGGCAGAGAACGATGGTCAA; reverse: AGTCGAAGTTGGCGGTCAAT), fgβ (forward: CTCCATCCACGGCTATGTCC; reverse: CCAACGCCTGCCAAAATCAA); fgγ (forward: ACGACTTTGGTGACGATCCC; reverse: GACCATCGGGACTTGCGTAT); itgav (forward: GTACGAGGCTGACCTGATCG; reverse: AAGATCACACACGACCACCC; itgb5 (forward: AACCCACTCAACAGGACAGC; reverse: ACATTCACAGAACGGACCCC). RT-qPCR was performed with cDNA as the template using the LightCycler ® 480 SYBR Green I Master system (Roche). Reactions were conducted in triplicate for each sample. Gene expression levels in each control or transgenic liver samples were normalized with the expression level of β-actin mRNA as the internal control. The log2 fold changes in expression in the transgenic samples as compared with the control samples were then calculated using the CT method [21] according to the formula: log2 fold changes = -∆∆CT = -[(CT gene of interest–CT β-actin) transgenic sample–(CT gene of interest–CT β-actin) control sample].

Human Patient Samples

Paraffin-embedded human liver disease progression tissue microarray slides were purchased from Biomax, Inc (Derwood, MD) (LV8011a). Samples were classified into 4 groups: normal, inflammation, cirrhosis, and HCC. Information of histopathology of all patients was provided by Biomax. Patient sample slides were performed with H&E staining and IHC staining for fibrinogen.

Statistical Analyses

For statistical significance between two groups, 2-tailed unpaired Student t test was performed using GraphPad Prism version 7.00 for Windows. Statistical data are presented as means ± SEM.

Results

KrasV12-expression Induced Liver Proliferation and Fibrosis in Zebrafish Larvae

In previous study, we observed a robust tumor progression with rapid increases of hepatocyte proliferation and fibrosis after kras induction in zebrafish adult [6], [20]. To examine kras-induced tumorigenesis in zebrafish larva, heterozygous kras+ transgenic zebrafish were crossed with WT zebrafish to get kras+ and WT larvae. They were exposed to dox for 4 days from 3 dpf to 7 dpf. Gross morphology of the whole larvae was examined after the treatment and we observed a significant liver enlargement after 4 days of induction (Figure 1A). To investigate the mechanism for hepatocarcinogenesis, cell proliferation, apoptosis and hepatic fibrosis were examined by IF staining. As expected, kras-expressing hepatocytes had a higher rate of cell proliferation and apoptosis than their WT siblings (Figure 1, B and C). Kras-expression increased both proliferation and apoptosis of hepatocytes, but the percentage of proliferating cells (26.5%) was much higher than the percentage of apoptotic cells (9.5%), thus causing a liver enlargement. Hepatic fibrosis is a marker of early HCC, characterized by collagen and laminin secretion [22]. By IF staining, we observed significant increases of collagen I and laminin deposition in the kras-expressing liver (Figure 1, D and E). It has been reported that both total HSC density and ratio of activated HSCs are increased in HCC patients [23]. Here we also found significant increases of the density of total HSCs as marked by Gfap staining and the ratio of activated HSCs as marked by a-SMA staining in kras-expressing liver in the larvae (Figure 1F); this is consistent with our previous observation in adult kras-expressing livers [24].

Figure 1

Characterization of kras-induced hepatocarcinogenesis in zebrafish larvae. 3-dpi kras+ or WT larvae were treated with 20 μg/ml for 4 days. More than 20 larvae were analyzed in each group. (A) Gross morphology of kras+ or WT larvae after 4 days induction (left lateral view). kras expressing liver is marked by GFP expression and WT liver is outlined. Quantification of 2D liver size was shown in the right panel. (B-E) IF staining of PCNA (B), Caspase 3 (C), Collagen I (D) or Laminin (E) on liver sections of kras+ and WT larvae as indicated. Livers are marked by dash lines. Int indicates intestine. Quantifications of staining signals based on percentages of liver area are shown in the right panels. (F) IF co-staining of GFAP (red) and a-Sma (blue) on liver sections of kras+ and WT larvae as indicated. Quantifications of total HSC density based on Gfap+ cells and ratio of activated HSCs based on a-SMA staining in liver sections are presented on the right. *P < .05. Scale bars: 20 μm.

Fibrinogen Up-Regulation in Hepatocytes and Integrin αvβ5 Expression in HSCs

In human HCC patients, fibrinogen level is closely associated with advanced tumors and poor survival [25]. To examine whether fibrinogen was up-regulated in kras-expressing liver, IF staining of fibrinogen was conducted on liver sections of kras+ and WT zebrafish (Figure 2A). Following induction of kras expression, there were dramatic increases of Fibrinogen+ cells in hepatocyte (Figure 2B). To further confirm the up-regulation of fibrinogen gene expression following kras induction, hepatocytes were isolated by FACS for RT-PCR analyses of fibrinogen gene expression. As fibrinogen protein is composed of Fgα, Fgβ, and Fgγ subunits [26], the expression of all the three genes (fga, fgb and fgg) for the three subunits was measured by RT-qPCR. As shown in Figure 2C, indeed there was an up-regulation of fibrinogen mRNAs in OHs.

Figure 2

Fibrinogen production and integrin αvβ5 expression in the liver. (A) IF staining of Fibrinogen on liver sections of kras+ and WT larvae. Livers in the WT picture is marked by a dash line. (B) Quantification of Fibrinogen-positive hepatocytes. (C) Fold change of expression of fibrinogen genes (fga, fgb an fgg) kras-expressing hepatocytes compared to WT hepatocytes. Integrin receptor, (D-E)expression of itgav (D) and itgb5 (E) mRNAs in hepatocytes and HSCs. Normal hepatocytes and HSCs were sorted by FACS based on DsRed and GFP expression from fabp10+ and hand2+ transgenic zebrafish respectively. Relative expression levels are shown with the value from hepatocytes arbitrarily set as 1. *P < .05. Scale bar: 20 μm.

It has been reported that fibrinogen can activate HSCs through integrin αvβ5, which consists of the αv subunit and β5 subunit [27]. The αv subunit is the key regulator in the fibrinogen signal transduction and it is largely expressed by neoplastic cells to promote liver disease progression and tumor metastasis in human patients [28]. To investigate the expression of Integrin αvβ5 in the zebrafish liver, hepatocytes and HSCs were isolated from fabp10+ and hand2+ fish by FACS. Expression of both subunit genes, itgav and itgb5, were measured by RT-qPCR. Both genes showed a significantly higher expression in HSCs than that in hepatocytes (Figure 2, D and E). These observations indicated potentially active interaction of fibrinogen and integrin pathway in HSCs.

Manipulation of Integrin αvβ5 Activity Affect HSC Density, Activation and Function

To examine the effects of activation and inhibition of integrin αvβ5 on HSCs during kras-induced hepatocarcinogenesis, kras+ larvae and WT siblings were treated with dox together with either an antagonist (cilengitide) [29] or an agonist (ADP) [30] of integrin. To elucidate the effects on HSC density and activation, IF co-staining of Gfap and a-SMA was performed in the liver section of the larvae (Figure 3A). As quantified in Figure 3, B and C, agonist treatment significantly increased the HSC density and the ratio of activated HSCs, while the antagonist treatment only decreased the activation ratio.

Figure 3

Effects of manipulation of integrin αvβ5 activity on HSC. 3-dpi kras+ and WT larvae were treated cilengitide or ADP together with dox for 4 days. Liver sections were stained for various molecular markers and more than 20 fish were analyzed in each group. (A) IF co-staining of Gfap (red) and a-SMA (blue). (B) Quantification of total HSC density. (C) Quantification of ratio of activated HSCs. (D) IF co-staining of Gfap (red) and Tgfb1 (blue). (E) Quantification of percentage of Tgfb1-expressing HSCs. (F) IF co-staining of Gfap (red) and Caspase 3 (blue). (G) Quantification of percentage of apoptotic HSCs. Livers are marked by dash lines. Int indicates intestine. *P < .05. Scale bar: 20 μm.

It has been reported that activated HSCs secret Tgfb1, which in turn plays a prominent role in stimulating liver fibrogenesis [31]. To examine Tgfb1 expression in HSCs after kras induction in the presence of an antagonist or an agonist, IF co-staining of Gfap and Tgfb1 was conducted in the liver sections (Figure 3D). The agonist increased the percentage of Tgfb1-expressing HSCs, while the antagonist showed the opposite effect in the kras+ larvae. In contrast, neither the agonist nor antagonist caused any significant change of Tgfb1 expression in wild type (WT) siblings (Figure 3E). IF co-staining of Gfap with Caspase 3 was carried out to determine apoptotic HSCs (Figure 3F). The antagonist treatment showed an increase, but statistically not significant, of percentage of apoptotic HSCs in kras+ larvae, while the agonist treatment had no detectable effect on HSC apoptosis (Figure 3G). Hence, we concluded that the agonist treatment significantly promoted the HSC function and proliferation, while the antagonist inhibited HSC activation and induced HSC apoptosis.

Effects of Manipulation of Integrin αvβ5 Activity on Liver Proliferation, Apoptosis and Fibrosis

Following the demonstration of increased Integrin αvβ5 activity to activate HSCs, we further examined its effect on kras-induced liver tumorigenesis. As shown in Figure 4, A and B, inhibition of Integrin αvβ5 by cilengitide attenuated the kras-induced liver enlargement while stimulation of Integrin αvβ5 by ADP further enlarged oncogenic liver. To elucidate the underlying mechanism for the change in the liver size, IF stainings for hepatocyte proliferation (PCNA) and apoptosis (Caspase 3) were carried out (Figure 4, C and E). Antagonist treatment significantly inhibited the proliferation of oncogenic hepatocytes and slightly increased the hepatocyte apoptosis while the agonist treatment enhanced the hepatocyte proliferation, but had no effects on the hepatocyte apoptosis. In contrast, both antagonist and agonist treatments had no effects on cell proliferation and apoptosis in WT livers (Figure 4, D and F).

Figure 4

Effects of manipulation of integrin αvβ5 activity on tumorigenesis. 3-dpi kras+ and WT larvae were treated cilengitide or ADP together with dox for 4 days. Liver size were measured and liver sections were stained for cell proliferation and apoptosis. More than 20 fish were analyzed in each group. (A) Gross morphology of larvae after 4 days of treatment (left lateral view). The WT livers were outlined. (B) Quantification of liver size. (C) IF staining of PCNA on liver sections. (D) Quantification of PCNA-positive cells in the liver. (E) IF staining of Caspase 3 on liver sections. (F) Quantification of Caspase 3-positive cells in the liver. Livers are marked by dash lines. Int indicates intestine. *P < .05. Scale bar: 20 μm.

Furthermore, whether or not the antagonist and agonist treatments affected the progression of liver fibrosis was investigated. Collagen I and laminin has been used as the liver fibrosis markers [32]. To measure the progression of liver fibrosis, collagen I and laminin were detected by IF staining (Figure 5, A and C). Antagonist of integrin αvβ5 significantly inhibited both collagen I and laminin expression, while the agonist showed the opposite effects (Figure 5, B and D). There were no significant differences in WT livers among different treatments.

Figure 5

Effects of manipulation of integrin αvβ5 activity on progression of liver fibrosis. 3-dpi kras+ and WT larvae were treated cilengitide or ADP together with dox for 4 days. Liver sections were stained for fibrosis markers, Collagen I and Laminin. More than 20 fish were analyzed in each group. (A) IF staining of Collagen I on liver sections. (B) Quantification of percentage of Collagen I positive area in the liver. (C) IF staining of Laminin on liver sections. (D) Quantification of percentage of Laminin positive area in the liver. Livers are marked by dash lines. Int indicates intestine. *P < .05. Scale bar: 20 μm.

Increased Fibrinogen Expression in Human Liver Disease Patients

To determine whether the increase of fibrinogen that we observed in the zebrafish HCC model would also be observed in human liver disease conditions, we also analyzed a tissue array containing multiple human liver disease samples. Patients were classified into 4 groups according to the liver disease progression, normal (n = 5), inflammation (n = 7), cirrhosis (n = 16) and HCC (n = 30), respectively. H&E staining were performed to confirm their histology (Figure 6A). Fibrinogen was stained (Figure 6B) and we observed a progressive increase of fibrinogen level during liver disease progression (Figure 6C). Fibrinogen level showed a significant increase in HCC patients, compared to the normal liver. It suggested the role of fibrinogen signaling on liver disease progression.

Figure 6

Fibrinogen level in human liver disease samples. A panel of liver disease samples from human patients was examined for histology by H&E staining and for fibrinogen levels by antibody staining. These samples were categorized into normal, inflammation, cirrhosis, and HCC. (A) H&E staining of human liver disease samples. (B) IHC staining of antibody against Fibrinogen. (C) Quantification of the percentages of fibrinogen-positive cells in human liver disease samples. *P < .05. Scale bar: 20 μm.

Discussion

HCC usually develops from a background of chronic inflammation and/or fibrosis. During progression of carcinogenesis, there would be a substantial crosstalk between tumor cells and stromal cells. The tumor derived factors activate stromal cells and stimulate various growth factors, chemokines and cytokines expression to in turn promote carcinogenesis [33]. In human HCC patients, stromal cells and genes specifically expressed by stromal cells have been used as clinical markers in prognosis of liver disease. HSC signature expression is associated to overall survival [34] and provide various molecular markers for diagnosis [35]. Essentially, the interaction between oncogenic hepatocytes and stromal cells in the TME provide a better environment for tumor growth. In recent years, increased studies show that interaction not only causes the tumor initiation, but also accelerates tumor progression [36]. In our previous studies on this kras-induced zebrafish HCC model, increased serotonin production by oncogenic hepatocytes resulted in activation of HSC and production of Tgfb1 to stimulate hepatocarcinogenesis [6]. Similarly, tumor associated neutrophils and macrophages were greatly activated at least partly owing to cortisol secreted by oncogenic hepatocytes. The activated tumor-associated neutrophils and macrophages also secreted Tgfb1 to promote the tumorigenesis in the kras-induced zebrafish HCC model [20].

Here in this study, we identified a new signaling pathway between oncogenic hepatocytes and stromal cells. After kras-induction, we detected significant up-regulation of both fibrinogen mRNAs and proteins. Integrin αvβ5 as a fibrinogen receptor was highly expressed in HSCs compared to that in hepatocytes. By using an antagonist and an agonist of integrin αvβ5 to treat the zebrafish, we found that the antagonist treatment inhibited HSC proliferation and deferred carcinogenesis. On the contrary, the agonist treatment promoted HSC proliferation and Tgf1b expression to promote carcinogenesis, indicating the importance of fibrinogen signaling on HSC survival and the significance of HSCs on progression of liver fibrosis and carcinogenesis. Consistent with our finding, it has been previously reported that a fibrinogen-like protein activated HSCs and subsequently promoted fibrosis and tumorigenesis in the cirrhotic patients with HCC [37]. Moreover, pharmacological inhibition of fibrinogen and heterozygous fibrinogen mutation prevented the kidney fibrosis in mice, where Tgfb signaling pathway has also shown to be involved in fibrinogen induced fibrosis [38]. Thus, our results together with these previous studies indicated the important role of fibrinogen signal on fibrogenesis.

In human patients, fibrinogen expression has been reported to be increased progressively with clinical tumor stage and higher plasma fibrinogen level has been associated with a poor prognosis [39]. Fibrinogen could also serve as a useful predictor of clinical HCC progression and recurrence [40], [41]. In our study, we detected a progressive increase of fibrinogen level during the liver disease progression using liver tissue array slides with human liver disease. HCC patients showed a significant increase of fibrinogen level compared to humans with normal liver or liver disease before HCC stage.

In sum, the combination of studies on zebrafish HCC model and human liver disease samples show that the interaction between oncogenic hepatocytes and stromal cells promote the progression of carcinogenesis and liver fibrosis through fibrinogen and integrin αvβ5 related signaling pathway. Importantly, we show that the activation of fibrinogen signaling is conserved from human HCC patients to our zebrafish HCC model that proved again our finding on zebrafish model could translate to human patients which help gain more knowledge on potential mechanisms of human carcinogenesis. Pharmacological targeting of fibrinogen signaling could be a potential therapy for human HCC patients to attenuate the progression of liver fibrosis and carcinogenesis. However, in a thioacetamide induced cirrhosis rat, cilengitide treatment reduced integrin αvβ5 but increased liver fibrosis [42]; thus, caution is needed in use of integrin antagonists in human liver disease. It has been reported that integrin-targeted cancer immunotherapy helps improve the tumor targeting specificity and reduces the harm on healthy tissue [43], [44]; thus combination of integrin antagonists with cancer targeted immunotherapy may represent a promising development of future therapeutic approaches to cancer treatment.