Extracellular matrix transcriptome dynamics in hepatocellular carcinoma

Matrix Biology 32 (2013) 393–398

Contents lists available at ScienceDirect

Matrix Biology journal homepage: www.elsevier.com/locate/matbio

Brief report

Extracellular matrix transcriptome dynamics in hepatocellular carcinoma Michael B. Duncan ⁎ Section of Gastroenterology/Hepatology, Department of Medicine, Medical College of Georgia, Georgia Regents University, Augusta, GA 30912, United States Department of Biochemistry and Molecular Biology, Medical College of Georgia, Georgia Regents University, Augusta, GA 30912, United States Georgia Regents University Cancer Center, Georgia Regents University, Augusta, GA 30912, United States

a r t i c l e

i n f o

Article history: Received 3 April 2013 Received in revised form 17 May 2013 Accepted 21 May 2013 Keywords: Cancer RNA-Seq Collagen Proteoglycan Glycoprotein

a b s t r a c t The extracellular matrix undergoes extensive remodeling during hepatocellular carcinoma and functions as a critical component of the tumor microenvironment by providing a substratum for cell adhesion and serving as a reservoir for a variety of cytokines and growth factors. Despite the clinical correlation between ECM deposition and hepatocellular carcinoma progression, it remains unclear how global extracellular matrix gene expression is altered in hepatocellular carcinoma and the molecular pathways that govern this change. Herein, a comprehensive analysis of the extracellular matrix transcriptome using an RNA-sequencing dataset provided by The Cancer Genome Atlas consortium was conducted and indicates substantial differential gene expression of key extracellular matrix collagens, glycoproteins, and proteoglycans in hepatocellular carcinoma. This analysis also reveals alternative expression of extracellular matrix gene transcript variants that could impact biological activity and serves as a framework for exploring the dynamic nature of the extracellular matrix transcriptome in cancer and identifying candidate genes for future exploration. © 2013 Elsevier B.V. All rights reserved.

1. Introduction It is estimated that 80–90% of all cases of hepatocellular carcinoma are preceded by cirrhosis (Seitz and Stickel, 2006). In cirrhosis, reduced liver function is due in part to the excessive deposition of extracellular matrix (ECM) and activation of various stromal cell types. These changes in the ECM also provide a permissive environment for tumor development through enhanced integrin signaling, paracrine cross-talk between tumor and stromal cells, increased tissue stiffness, and altered growth factor sequestration by ECM (Zhang and Friedman, 2012). This suggests that HCC progression and clinical outcomes are tightly linked to the tissue remodeling and enhanced matrix deposition within the stromal tissue environment that typifies cirrhotic liver pathology. From a clinical perspective, curative therapy for HCC when moderate to extensive cirrhosis has occurred is often not effective (Ercolani et al., 2003). These observations provide strong evidence that the extracellular matrix in the hepatic microenvironment plays a critical role in HCC progression and can hamper therapeutic outcomes. The ECM is a fundamental component of the tumor stroma, promoting cancer hallmark capabilities to tumor cells via enhanced cell adhesion and serving as a reservoir for growth factors and cytokines

⁎ Medical College of Georgia, Georgia Regents University, Augusta, GA 30912, United States. Tel.: +1 706 721 5484. E-mail address: [email protected]. 0945-053X/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.matbio.2013.05.003

(Hanahan and Weinberg, 2011; Lu et al., 2012). Due to its distinct molecular composition, the liver ECM is structurally unique from most tissue ECMs and lacks the typical electron dense nature of most basement membrane structures. Quantitative and qualitative studies convincingly show that as liver tissue changes during cirrhosis, so does the composition of the ECM (Gressner and Weiskirchen, 2006; Zeisberg et al., 2006). However questions remain regarding the origins of these changes and their impact on cancer progression. In particular how the ECM is altered in HCC and the molecular pathways that govern this change are largely unknown. Proteomic approaches have been developed to begin determining the types of ECM and ECM related molecules expressed in the tumor microenvironment of various cancers (Hynes and Naba, 2012; Naba et al., 2012). The genetic programs and resulting expression that give rise to these unique ECM profiles in these and other cancer types remain to be fully elucidated. Next generation sequencing represents a significant advance in our ability to study genomic alterations and global gene expression profiles in cancer (Shendure and Ji, 2008; Han, 2012). In particular, RNA sequencing technology provides sensitive detection of gene expression, analysis of non-coding RNA, and the discovery of novel transcripts in cancer (Mardis and Wilson, 2009; Wang et al., 2009). Recently, The Cancer Genome Atlas (TCGA, http://cancergenome.nih. gov/) has initiated an effort to provide publicly accessible data on several cancers including HCC. In this study the TCGA dataset was used to develop a preliminary assessment of changes in the extracellular matrix transcriptome in hepatocellular carcinoma. These

394

M.B. Duncan / Matrix Biology 32 (2013) 393–398

results provide new insight into the ECM gene expression pattern in the liver and the substantial changes it undergoes during HCC.

2. Results 2.1. ECM expression in normal versus HCC The TCGA provides gene expression datasets (RNA sequencing) for a variety of cancers including HCC (LIHC dataset). For this study, normalized RNA counts derived from the RSEM (RNA-Seq expression estimation by Expectation–Maximization) algorithm were acquired from the TCGA data portal and analyzed. The RSEM algorithm is a statistical model that estimates RNA expression levels from RNA sequencing counts (Li and Dewey, 2011). Recently a comprehensive list of ECM and ECM related molecules has been reported (Hynes and Naba, 2012; Naba et al., 2012). These studies have classified ECM genes into “core” and “associated” ECM subtypes, in which the “core” sub-type refers to proteins that are directly incorporated into the ECM structure, while the “associated” sub-type refers to proteins that interact with or alter the ECM. This list of “core” ECM genes was utilized to conduct an analysis of ECM gene expression in human HCC. ECM gene expression was assessed based on the RSEM counts. A broad distribution of ECM gene expression was observed as indicated by RSEM normalized counts in both normal liver (n = 20) and HCC (n = 34) samples (Fig. 1A). More ECM genes were expressed at greater than 50 counts in HCC as compared to normal, suggesting a general increase in ECM gene expression. A scatter plot comparing gene expression between normal and HCC reveals a broad spectrum of expression among collagens, glycoproteins, and proteoglycans of the ECM (Fig. 1B). Based on the scatter plot nearly a third of the analyzed ECM genes were potentially alternatively expressed in HCC.

A RSEM counts:

< 50

>50-100

>100-500 >500-1500

>1500

120

# Genes

100 80

In order to establish the role of the ECM in HCC, it is essential to determine what ECM genes are differentially expressed in the tumor setting. To determine specific genes that were significantly differentially expressed, a volcano plot of the statistical difference between normal and HCC samples (p-value of Student's t-test) versus fold change in gene expression of normal versus HCC was analyzed (Fig. 2A). Based on the volcano plot (see Experimental procedures for detail), a diverse set of ECM genes was differentially expressed (both over and under expressed) in HCC. Fig. 2B depicts the number of collagens, glycoproteins and proteoglycans that are up or down regulated or did not change significantly during HCC. Criteria of a two-fold difference in gene expression and a p ≥ 0.05 were set in order to establish differential expression. An assessment of the three major ECM gene families indicate that 40.91%, 18.97%, and 25.00% of collagens, glycoproteins, and proteoglycans are up-regulated respectively and 6.82%, 15.38%, and 5.56% of collagens, glycoproteins, and proteoglycans were down regulated in HCC. 2.3. Highly expressed ECM genes in HCC In order to establish potential roles of ECM genes that are differentially expressed in HCC, it is important to determine if they are expressed at significantly high levels. ECM genes were plotted based on their RSEM normalized counts. A threshold of greater than 1500 counts was established in order to focus on those genes that were expressed to an appreciable extent in the liver. It is important to note that a number of differentially expressed ECM genes fell below this established threshold and a complete list of all ECM genes analyzed in the dataset is provided in Supplemental Table 1. We observed that several collagens including COL1A1, COL1A2, COL4A1, COL4A2, COL5A1, COL5A2, and COL6A3 met this threshold and were up regulated in HCC (Fig. 3A). No collagens meeting the expression threshold were down regulated. The proteoglycans AGRN, HSPG2, and VCAN were up regulated in HCC while DCN and PRG4 proteoglycans met the expression threshold but were down regulated in HCC (Fig. 3B). ECM glycoproteins that met the expression threshold including CYR6, ECM1, FGA, FGB, FGG, FGL1, IGFALS, IGFBP1, IGFBP3, IGFBP4, LRG1, MFAP3L, and THBS1 were down regulated, while SMOC1, SPARCL1, SPARC and VWF were up regulated in HCC (Fig. 3C).

60 40

2.4. Isoform variants of differentially and highly expressed ECM genes

20 0

Collagen Glycoprotein Proteoglycan

N T

N T

N

T

N T

N

T

B Hepatocellular Carcinoma (RSEM normalized counts)

2.2. Differential expression of ECM in HCC

6

10

5

10

104 10 3 10 2 101 10 0

Isoform variants of ECM genes influence their localization and function. RNA sequencing allows for determining the relative contribution of these isoforms. We examined how isoform levels change during differential expression in HCC. Of the ECM genes that were down regulated in HCC, FGL1 and MFAP3L demonstrated switches in the relative abundance of their predominant isoforms (Fig. 4A). Of the up-regulated genes that met the expression criteria VCAN, HSPG2, COL6A3, and SPARCL1 showed a significant change in isoform abundance in HCC (Fig. 4B). 3. Discussion

10-1 10-2 -2 0 -1 10 10 10

10

1

10

2

10

3

10

4

10

5

10

6

Normal Liver (RSEM normalized counts) Fig. 1. ECM gene expression in normal liver (N) versus hepatocellular carcinoma (T) tissue. (A) RSEM normalized count distribution in normal liver (n = 20) and hepatocellular carcinoma (n = 34). (B) Scatter plot of RSEM normalized gene expression in normal liver and hepatocellular carcinoma tissues. Normalized values were averaged and plotted on a log scale. The gray shaded area indicates less than two fold difference in RSEM normalized value between normal and HCC tissue.

3.1. Extracellular matrix and hepatocellular carcinoma It is well appreciated that the ECM plays critical roles in developmental, homeostatic, and tumorigenic processes in liver biology (Bedossa and Paradis, 2003; Kalluri, 2003; Hynes, 2009; Yurchenco, 2011). However, a systematic understanding of the composition of the HCC ECM and how it is regulated has not yet been fully elaborated. Understanding the molecular constituents of the ECM as well as the gene expression programs that regulate their expression will

M.B. Duncan / Matrix Biology 32 (2013) 393–398

395

Fig. 2. Differential ECM gene expression in hepatocellular carcinoma. (A) Volcano plot representing fold change versus p-value from Student's t-test. Red dashed line indicates a cutoff of statistical significance at p = 0.05. (B) Pie charts representing the total number of collagen, glycoprotein, and proteoglycan genes that were up-regulated (green), down-regulated (red), or had no change (gray) in expression.

provide a significant advance in determining how tumor ECM develops and modulates the HCC tumor microenvironment. 3.2. Collagens Several members of the collagen family of ECM molecules were up regulated in the LIHC dataset. Type I collagen is the predominant fibrillar collagen associated with cirrhotic scar tissue in the liver

A

5000

0

10000 15000 20000

COL1A1 COL1A2 COL3A1 COL4A1 COL4A2 COL5A1 COL5A2 COL6A1 COL6A2 COL6A3 COL18A1

B

RSEM normalized counts 0

2000

4000

6000

8000

AGRN BGN DCN HSPG2 PRG4 SRGN NO CHANGE

VCAN

RSEM normalized counts

C

RSEM normalized counts 0

(Bedossa and Paradis, 2003). The deposition of this fibrillar collagen promotes tissue stiffness, which is a poor prognostic for cirrhosis and HCC development (Zhao et al., 2010; Motosugi et al., 2013). Type IV collagen is a major ECM component of the vasculature and is highly associated with established and angiogenic vessels in HCC (Grigioni et al., 1987). The alpha 1 and alpha 2 chains are commonly expressed in the liver and were significantly up regulated in the dataset. Type V collagen is considered a minor fibrillar collagen,

DOWN-REGULATED UP-REGULATED

20000 40000

150000

250000

AEBP1 CRIM1 CTGF CYR6 ECM1 FGA FGB FGG FGL1 FN1 GAS6 IGFALS IGFBP1 IGFBP2 IGFBP3 IGFBP4 IGFBP5 IGFBP7 LAMA5 LAMB1 LAMB2 LAMC1 LRG1 LTBP3 LTBP4 MFAP3L NID1 OIT3 PCOLCE RELN SMOC1 SPARCL1 SPARC SPON2 SPP1 TGFBI THBS1 TSKU VTN VWA1 VWF

Fig. 3. Relative abundance of differentially expressed ECM genes in hepatocellular carcinoma. (A) Collagen, (B) proteoglycan, and (C) glycoprotein genes that averaged greater than 1500 normalized counts were considered highly expressed. Up-regulated genes are indicated in green, down-regulated genes are indicated in red and unchanged genes are gray. Error bars (non-capped black line) in each bar graph represent standard error of the mean.

396

M.B. Duncan / Matrix Biology 32 (2013) 393–398

A

Up Regulated Normal

B

Down Regulated

HCC

Normal

VCAN

FGL1

HSPG2

MFAP3L

HCC

3.4. Insulin-like Growth Factor Binding Proteins The insulin growth factor (IGF) signaling axis plays a critical role in proliferation, survival, and the metastatic capacity of HCC cells (Wu and Zhu, 2011). Insulin-like Growth Factor Binding Protein (IGFBP) family members regulate IGF mitogenic activity by modulating their bioavailability and circulating half-life (Kelley et al., 1996; Kuemmerle, 2012). An overall decrease in IGFBP levels, as was revealed in differential expression analysis, would likely enhance IGF signaling during HCC progression. 3.5. Hepatocyte-specific ECM gene expression

COL6A3

SPARCL1 Fig. 4. Changes in isoform expression in differentially expressed ECM genes. The pie chart represents the relative contribution of ECM transcript isoforms in normal liver and HCC that are (A) up-regulated or (B) down-regulated.

however it plays an essential role in collagen fibrillogenesis (Wenstrup et al., 2004a). Heterozygous animals are viable, but develop abnormal collagen fibrils. This genetic evidence coupled with biochemical studies suggests that COL5A1 plays a critical role in initiation of fibril assembly (Wenstrup et al., 2004b). Type VI collagen is up regulated in various cancers and pre-clinical models suggest that it plays a key role in promoting tumorigenesis and resistance to chemotherapy (Arafat et al., 2011; Sherman-Baust et al., 2003).

3.3. Proteoglycans Agrin (AGRN), perlecan (HSPG2) and versican (VCAN) were the only proteoglycans up regulated in the HCC dataset. Agrin is a heparan sulfate proteoglycan known to accumulate in the liver during cirrhosis and HCC (Tatrai et al., 2006). Perlecan is an important vascular basement membrane component and endorepellin is an angiostatic matricryptin derived from perlecan (Mongiat et al., 2003; Whitelock et al., 2008). Versican is a large chondroitin sulfate proteoglycan that is considered to have pro-tumorigenic properties (Sheng et al., 2005; Kim et al., 2009; Ricciardelli et al., 2009). A recent report also suggests that the 3′-untranslated region of versican mRNA can regulate microRNA activity and promotes cell proliferation and escape form apoptosis (Fang et al., 2012). Versican has been shown to be over-expressed at the protein level in a chemically induced rodent model of HCC (Jia et al., 2012). Several proteoglycans in the LIHC dataset were down regulated. Proteoglycan 4 (PRG4) is expressed throughout the gastrointestinal tract,1 however it is best known as a chondrocyte proteoglycan that plays a role in joint lubrication (Coles et al., 2010). A role for this molecule in cancer has not yet been established. Decorin (DCN) is a member of the small leucine rich proteoglycan family and binds to type I collagen fibrils (Brown and Vogel, 1989; Hedbom and Heinegard, 1989; Svensson et al., 1995). Studies suggest that this molecule limits oncogenic transformation in vivo and cancer cell growth by modulating receptor tyrosine kinase signaling including Met receptor and EGF receptor complexes, and angiogenesis by controlling VEGF and thrombospondin gene expression levels (Grant et al., 2002; Neill et al., 2012). These findings suggest that the down regulation observed in HCC may facilitate tumor progression.

1 Based on high transcript count in the TCGA LIHC dataset and data from the human protein atlas (www.proteinatlas.org).

Hepatocyte specific ECM genes including the fibrinogen alpha (FGA), fibrinogen beta (FGB), fibrinogen gamma (FGG) and fibrinogen like-1 (FGL1) were down regulated in the LIHC dataset. The decrease in gene expression of these liver specific ECM genes likely reflects dedifferentiation of hepatocytes during tumor progression (Ishiyama et al., 2003). 3.6. Other ECM molecules Thrombospondin 1 (THBS1) was decreased in the HCC dataset. This multifunctional glycoprotein plays a role in various processes related to cancer including angiogenesis, cell adhesion, inflammation, and hemostasis (Bonnefoy et al., 2008; Kazerounian et al., 2008; Lopez-Dee et al., 2011). Thrombospondin 1 was recently shown to be a negative regulator of liver regeneration in mice due to its ability to activate TGF beta (Hayashi et al., 2012). Its down-regulation at the gene expression level may reflect a mechanism to regulate TGF beta driven cell proliferation, apoptosis, and angiogenesis in the tumor microenvironment. Von Willebrand factor is a common marker for endothelial cells and plays a critical role in hemostasis (Wagner, 1990; Franchini and Mannucci, 2008; Franchini et al., 2013). Increased gene expression of this ECM associated glycoprotein likely reflects the angiogenic response associated with HCC tumorigenesis (Xiong et al., 2009). Sparc, Smoc1, and Sparcl1 are all members of the Sparc family of glycoproteins that are commonly expressed in normal and malignant tissues. While previous studies suggest that Sparc is overexpressed in HCC, its precise role in tumorigenesis remains unclear. Genetic evidence in various cancers suggests that Sparc can suppress tumorigenesis, however it is also thought to play a role in various processes including proliferation, matrix deposition, and angiogenesis (Chlenski et al., 2006; Arnold et al., 2010; Said et al., 2013). 3.7. Transcript variants Most ECM proteins are modular in nature having distinct structural domains that promote self-assembly as well as binding sites for various ligands, and cell attachment (Hynes, 2012). Alternative splicing of ECM genes is common and can result in multiple structurally unique proteins arising from a single gene. A particular advantage of RNA sequencing lies in its ability to provide quantitative analysis of transcript variants that arise from alternative splicing events. In the LIHC dataset a number of genes displayed changes in variant abundance. HSPG2, SPARCL1, FGL1, and MFAP3L all demonstrated differential expression of their transcript variants, however these variants to date have not been well characterized in the context of cancer. Also differentially expressed in our dataset, various VCAN variants are known to elicit unique cellular responses in vitro and tumor specific variants of COL6A3 are expressed in pancreatic cancer (Sheng et al., 2005; Arafat et al., 2011; Said et al., 2012; Yang and Yee, 2013). Analysis of these genes and others in the LIHC dataset suggests that differential expression of transcript variants is a common feature of HCC and has the potential to impact tumor phenotype.

M.B. Duncan / Matrix Biology 32 (2013) 393–398

3.8. Conclusions Developing systematic approaches to understanding how unique ECM compositions arise in a given tissue and are altered during cancer progression represents a unique opportunity to determine the role that ECM plays in caner progression. The analysis presented herein suggests that there are key groups of ECM molecules that are differentially expressed in HCC. This study provides a framework for how next generation sequencing may be utilized for exploring the mechanisms involved in ECM gene expression and identifies novel ECM molecules that may impact HCC progression. 4. Experimental procedures 4.1. The Cancer Genome Atlas dataset Data was downloaded from the TCGA website (http://cancergenome. nih.gov/). The Liver Hepatocellular Carcinoma (LIHC) dataset was accessed from the TCGA data portal (https://tcga-data.nci.nih.gov/tcga/). The dataset used in this analysis was generated using 2 × 48 base pair paired end reads acquired on an Illumina HiSeq 2000 sequencing platform at The University of North Carolina at Chapel Hill. Normalized RNA-Seq expression estimation by Expectation–Maximization (RSEM) gene expression and isoform expression were utilized for analysis. RSEM is a statistical method for analyzing RNA sequencing data. This is a robust method for mapping reads and estimating gene and isoform abundance, while taking read uncertainty (RNA-Seq reads that map to multiple genes or transcript isoforms) into consideration (Li and Dewey, 2011). ECM genes were identified based on the curated list provided by Naba et al. (2012). The complete list of ECM genes, their symbol, ID, and alternative names are provided in Supplemental Table 2. Transcript variants were identified based on the UCSC identifier and compiled with the respective gene identifier. 4.2. Differential expression Volcano plot is a scatter plot commonly used to graphically display differentially expressed data points (i.e. gene transcripts) contained in a large dataset (i.e. sequencing dataset). Fold change is plotted against p-value such that transcripts undergoing statistically significant change in expression approach upper left and right portions of the graph. 4.3. Statistical analysis Data are expressed as the mean ± standard error of the mean. Statistical significance was determined by an unpaired sample Student's t test using a two-tailed distribution. Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.matbio.2013.05.003. Acknowledgments This work was supported by the Department of Medicine at the Medical College of Georgia, Georgia Regents University. I would also like to thank The Cancer Genome Atlas for providing access to the Liver Hepatocellular Carcinoma (LIHC) dataset. References Arafat, H., Lazar, M., Salem, K., Chipitsyna, G., Gong, Q., Pan, T.C., Zhang, R.Z., Yeo, C.J., Chu, M.L., 2011. Tumor-specific expression and alternative splicing of the COL6A3 gene in pancreatic cancer. Surgery 150, 306–315. Arnold, S.A., Rivera, L.B., Miller, A.F., Carbon, J.G., Dineen, S.P., Xie, Y., Castrillon, D.H., Sage, E.H., Puolakkainen, P., Bradshaw, A.D., Brekken, R.A., 2010. Lack of host

397

SPARC enhances vascular function and tumor spread in an orthotopic murine model of pancreatic carcinoma. Dis. Model. Mech. 3, 57–72. Bedossa, P., Paradis, V., 2003. Liver extracellular matrix in health and disease. J. Pathol. 200, 504–515. Bonnefoy, A., Moura, R., Hoylaerts, M.F., 2008. The evolving role of thrombospondin-1 in hemostasis and vascular biology. Cell. Mol. Life Sci. 65, 713–727. Brown, D.C., Vogel, K.G., 1989. Characteristics of the in vitro interaction of a small proteoglycan (PG II) of bovine tendon with type I collagen. Matrix 9, 468–478. Chlenski, A., Liu, S., Guerrero, L.J., Yang, Q., Tian, Y., Salwen, H.R., Zage, P., Cohn, S.L., 2006. SPARC expression is associated with impaired tumor growth, inhibited angiogenesis and changes in the extracellular matrix. Int. J. Cancer 118, 310–316. Coles, J.M., Zhang, L., Blum, J.J., Warman, M.L., Jay, G.D., Guilak, F., Zauscher, S., 2010. Loss of cartilage structure, stiffness, and frictional properties in mice lacking PRG4. Arthritis Rheum. 62, 1666–1674. Ercolani, G., Grazi, G.L., Ravaioli, M., Del Gaudio, M., Gardini, A., Cescon, M., Varotti, G., Cetta, F., Cavallari, A., 2003. Liver resection for hepatocellular carcinoma on cirrhosis: univariate and multivariate analysis of risk factors for intrahepatic recurrence. Ann. Surg. 237, 536–543. Fang, L., Du, W.W., Yang, X., Chen, K., Ghanekar, A., Levy, G., Yang, W., Yee, A.J., Lu, W.Y., Xuan, J.W., Gao, Z., Xie, F., He, C., Deng, Z., Yang, B.B., 2012. Versican 3′-untranslated region (3′-UTR) functions as a ceRNA in inducing the development of hepatocellular carcinoma by regulating miRNA activity. FASEB J. 27, 907–919. Franchini, M., Mannucci, P.M., 2008. Von Willebrand factor: another janus-faced hemostasis protein. Semin. Thromb. Hemost. 34, 663–669. Franchini, M., Frattini, F., Crestani, S., Bonfanti, C., Lippi, G., 2013. von Willebrand factor and cancer: a renewed interest. Thromb. Res. 131, 290–292. Grant, D.S., Yenisey, C., Rose, R.W., Tootell, M., Santra, M., Iozzo, R.V., 2002. Decorin suppresses tumor cell-mediated angiogenesis. Oncogene 21, 4765–4777. Gressner, A.M., Weiskirchen, R., 2006. Modern pathogenetic concepts of liver fibrosis suggest stellate cells and TGF-beta as major players and therapeutic targets. J. Cell. Mol. Med. 10, 76–99. Grigioni, W.F., D'Errico, A., Mancini, A.M., Biagini, G., Gozzetti, G., Mazziotti, A., Garbisa, S., 1987. Hepatocellular carcinoma: expression of basement membrane glycoproteins. An immunohistochemical approach. J. Pathol. 152, 325–332. Han, Z.G., 2012. Functional genomic studies: insights into the pathogenesis of liver cancer. Annu. Rev. Genomics Hum. Genet. 13, 171–205. Hanahan, D., Weinberg, R.A., 2011. Hallmarks of cancer: the next generation. Cell 144, 646–674. Hayashi, H., Sakai, K., Baba, H., Sakai, T., 2012. Thrombospondin-1 is a novel negative regulator of liver regeneration after partial hepatectomy through transforming growth factor-beta1 activation in mice. Hepatology 55, 1562–1573. Hedbom, E., Heinegard, D., 1989. Interaction of a 59-kDa connective tissue matrix protein with collagen I and collagen II. J. Biol. Chem. 264, 6898–6905. Hynes, R.O., 2009. The extracellular matrix: not just pretty fibrils. Science 326, 1216–1219. Hynes, R.O., 2012. The evolution of metazoan extracellular matrix. J. Cell Biol. 196, 671–679. Hynes, R.O., Naba, A., 2012. Overview of the matrisome — an inventory of extracellular matrix constituents and functions. Cold Spring Harb. Perspect. Biol. 4, a004903. Ishiyama, T., Kano, J., Minami, Y., Iijima, T., Morishita, Y., Noguchi, M., 2003. Expression of HNFs and C/EBP alpha is correlated with immunocytochemical differentiation of cell lines derived from human hepatocellular carcinomas, hepatoblastomas and immortalized hepatocytes. Cancer Sci. 94, 757–763. Jia, X.L., Li, S.Y., Dang, S.S., Cheng, Y.A., Zhang, X., Wang, W.J., Hughes, C.E., Caterson, B., 2012. Increased expression of chondroitin sulphate proteoglycans in rat hepatocellular carcinoma tissues. World J. Gastroenterol. 18, 3962–3976. Kalluri, R., 2003. Basement membranes: structure, assembly and role in tumour angiogenesis. Nat. Rev. Cancer 3, 422–433. Kazerounian, S., Yee, K.O., Lawler, J., 2008. Thrombospondins in cancer. Cell. Mol. Life Sci. 65, 700–712. Kelley, K.M., Oh, Y., Gargosky, S.E., Gucev, Z., Matsumoto, T., Hwa, V., Ng, L., Simpson, D.M., Rosenfeld, R.G., 1996. Insulin-like growth factor-binding proteins (IGFBPs) and their regulatory dynamics. Int. J. Biochem. Cell Biol. 28, 619–637. Kim, S., Takahashi, H., Lin, W.W., Descargues, P., Grivennikov, S., Kim, Y., Luo, J.L., Karin, M., 2009. Carcinoma-produced factors activate myeloid cells through TLR2 to stimulate metastasis. Nature 457, 102–106. Kuemmerle, J.F., 2012. Insulin-like growth factors in the gastrointestinal tract and liver. Endocrinol. Metab. Clin. North Am. 41, 409–423 (vii). Li, B., Dewey, C.N., 2011. RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinforma. 12, 323. Lopez-Dee, Z., Pidcock, K., Gutierrez, L.S., 2011. Thrombospondin-1: multiple paths to inflammation. Mediators Inflamm. 2011, 296069. Lu, P., Weaver, V.M., Werb, Z., 2012. The extracellular matrix: a dynamic niche in cancer progression. J. Cell Biol. 196, 395–406. Mardis, E.R., Wilson, R.K., 2009. Cancer genome sequencing: a review. Hum. Mol. Genet. 18, R163–R168. Mongiat, M., Sweeney, S.M., San Antonio, J.D., Fu, J., Iozzo, R.V., 2003. Endorepellin, a novel inhibitor of angiogenesis derived from the C terminus of perlecan. J. Biol. Chem. 278, 4238–4249. Motosugi, U., Ichikawa, T., Koshiishi, T., Sano, K., Morisaka, H., Ichikawa, S., Enomoto, N., Matsuda, M., Fujii, H., Araki, T., 2013. Liver stiffness measured by magnetic resonance elastography as a risk factor for hepatocellular carcinoma: a preliminary case–control study. Eur. Radiol. 23, 156–162. Naba, A., Clauser, K.R., Hoersch, S., Liu, H., Carr, S.A., Hynes, R.O., 2012. The matrisome: in silico definition and in vivo characterization by proteomics of normal and tumor extracellular matrices. Mol. Cell. Proteomics 11 (M111 014647).

398

M.B. Duncan / Matrix Biology 32 (2013) 393–398

Neill, T., Schaefer, L., Iozzo, R.V., 2012. Decorin: a guardian from the matrix. Am. J. Pathol. 181, 380–387. Ricciardelli, C., Sakko, A.J., Ween, M.P., Russell, D.L., Horsfall, D.J., 2009. The biological role and regulation of versican levels in cancer. Cancer Metastasis Rev. 28, 233–245. Said, N., Sanchez-Carbayo, M., Smith, S.C., Theodorescu, D., 2012. RhoGDI2 suppresses lung metastasis in mice by reducing tumor versican expression and macrophage infiltration. J. Clin. Invest. 122, 1503–1518. Said, N., Frierson, H.F., Sanchez-Carbayo, M., Brekken, R.A., Theodorescu, D., 2013. Loss of SPARC in bladder cancer enhances carcinogenesis and progression. J. Clin. Invest. 123, 751–766. Seitz, H.K., Stickel, F., 2006. Risk factors and mechanisms of hepatocarcinogenesis with special emphasis on alcohol and oxidative stress. Biol. Chem. 387, 349–360. Shendure, J., Ji, H., 2008. Next-generation DNA sequencing. Nat. Biotechnol. 26, 1135–1145. Sheng, W., Wang, G., Wang, Y., Liang, J., Wen, J., Zheng, P.S., Wu, Y., Lee, V., Slingerland, J., Dumont, D., Yang, B.B., 2005. The roles of versican V1 and V2 isoforms in cell proliferation and apoptosis. Mol. Biol. Cell 16, 1330–1340. Sherman-Baust, C.A., Weeraratna, A.T., Rangel, L.B., Pizer, E.S., Cho, K.R., Schwartz, D.R., Shock, T., Morin, P.J., 2003. Remodeling of the extracellular matrix through overexpression of collagen VI contributes to cisplatin resistance in ovarian cancer cells. Cancer Cell 3, 377–386. Svensson, L., Heinegard, D., Oldberg, A., 1995. Decorin-binding sites for collagen type I are mainly located in leucine-rich repeats 4–5. J. Biol. Chem. 270, 20712–20716. Tatrai, P., Dudas, J., Batmunkh, E., Mathe, M., Zalatnai, A., Schaff, Z., Ramadori, G., Kovalszky, I., 2006. Agrin, a novel basement membrane component in human and rat liver, accumulates in cirrhosis and hepatocellular carcinoma. Lab. Invest. 86, 1149–1160. Wagner, D.D., 1990. Cell biology of von Willebrand factor. Annu. Rev. Cell Biol. 6, 217–246.

Wang, Z., Gerstein, M., Snyder, M., 2009. RNA-Seq: a revolutionary tool for transcriptomics. Nat. Rev. Genet. 10, 57–63. Wenstrup, R.J., Florer, J.B., Brunskill, E.W., Bell, S.M., Chervoneva, I., Birk, D.E., 2004a. Type V collagen controls the initiation of collagen fibril assembly. J. Biol. Chem. 279, 53331–53337. Wenstrup, R.J., Florer, J.B., Cole, W.G., Willing, M.C., Birk, D.E., 2004b. Reduced type I collagen utilization: a pathogenic mechanism in COL5A1 haplo-insufficient Ehlers–Danlos syndrome. J. Cell. Biochem. 92, 113–124. Whitelock, J.M., Melrose, J., Iozzo, R.V., 2008. Diverse cell signaling events modulated by perlecan. Biochemistry 47, 11174–11183. Wu, J., Zhu, A.X., 2011. Targeting insulin-like growth factor axis in hepatocellular carcinoma. J. Hematol. Oncol. 4, 30. Xiong, Y.Q., Sun, H.C., Zhang, W., Zhu, X.D., Zhuang, P.Y., Zhang, J.B., Wang, L., Wu, W.Z., Qin, L.X., Tang, Z.Y., 2009. Human hepatocellular carcinoma tumor-derived endothelial cells manifest increased angiogenesis capability and drug resistance compared with normal endothelial cells. Clin. Cancer Res. 15, 4838–4846. Yang, W., Yee, A.J., 2013. Versican V2 isoform enhances angiogenesis by regulating endothelial cell activities and fibronectin expression. FEBS Lett. 587, 185–192. Yurchenco, P.D., 2011. Basement membranes: cell scaffoldings and signaling platforms. Cold Spring Harb. Perspect. Biol. 3. Zeisberg, M., Kramer, K., Sindhi, N., Sarkar, P., Upton, M., Kalluri, R., 2006. De-differentiation of primary human hepatocytes depends on the composition of specialized liver basement membrane. Mol. Cell. Biochem. 283, 181–189. Zhang, D.Y., Friedman, S.L., 2012. Fibrosis-dependent mechanisms of hepatocarcinogenesis. Hepatology 56, 769–775. Zhao, G., Cui, J., Qin, Q., Zhang, J., Liu, L., Deng, S., Wu, C., Yang, M., Li, S., Wang, C., 2010. Mechanical stiffness of liver tissues in relation to integrin beta1 expression may influence the development of hepatic cirrhosis and hepatocellular carcinoma. J. Surg. Oncol. 102, 482–489.