Platelet-derived growth factor D: A new player in the complex cross-talk between cholangiocarcinoma cells and cancer-associated fibroblasts


  • Carmen Berasain Ph.D.,

    1. Division of Hepatology and Gene Therapy, Centro de Investigación Médica Aplicada, Universidad de Navarra, Pamplona, Spain
    2. Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas, Clínica Universidad de Navarra, Pamplona, Spain
    Search for more papers by this author
  • Matias A. Ávila Ph.D.

    Corresponding author
    1. Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas, Clínica Universidad de Navarra, Pamplona, Spain
    • Division of Hepatology and Gene Therapy, Centro de Investigación Médica Aplicada, Universidad de Navarra, Pamplona, Spain
    Search for more papers by this author

  • Potential conflicto of interest: Nothing to report.

  • See Article on Page 1042.

Address reprint requests to: Matias Ávila, Ph.D., Division of Hepatology and Gene Therapy, Centro de Investigación Médica Aplicada, Universidad de Navarra, Avenida Pío XII, N55, 31008 Pamplona, Spain. E-mail:; fax: +34-948-194717.


cancer-associated fibroblasts




extracellular matrix


enhanced green fluorescent protein


epithelial-mesenchymal transition


hepatocellular carcinoma


hepatocyte growth factor


hepatic stellate cells


c-Jun N-terminal kinase


platelet-derived growth factor


stromal cell-derived factor 1


alpha smooth muscle actin.

Cholangiocarcinoma (CCA) is a primary liver malignancy and a devastating disease with a very poor prognosis and increasing worldwide incidence.[1, 2] Besides liver fluke infection and primary sclerosing cholangitis, risk factors for CCA development are not completely known. However, conditions associated with chronic hepatic inflammation, such as viral infection, alcohol consumption, diabetes, and obesity, are increasingly being recognized as major risk factors for this malignancy that may be of relevance for a larger population.[3] The link between chronic inflammation and CCA development, its genetic heterogeneity, as well as the limited efficacy of standard and molecularly targeted therapies tested so far are common features shared with hepatocellular carcinoma (HCC).[3-9] Importantly, at variance with HCC, tumors originating from cells lining the biliary tree are frequently accompanied by a dense, reactive desmoplastic stroma surrounding the malignant ducts.[10] One characteristic and abundant cellular component of this desmoplastic stroma are alpha smooth muscle actin (α-SMA)-positive myofibroblasts, also known as cancer-associated fibroblasts (CAFs).[11] These mesenchymal cells appear not to be innocent bystanders in CCA progression. Accumulating evidence demonstrates that α-SMA-expressing CAFs indeed play an active role in tumor progression, their abundance correlating with decreased patient survival.[11, 12] The origin of CAFs in CCA is not completely clear. It is possible that these cells come from different origins, most likely hepatic stellate cells (HSCs) and/or portal or periductal fibroblasts, but also circulating bone marrow–derived precursor cells.[11, 12] In addition, the possibility of CAFs originating from tumor cells undergoing an epithelial-mesenchymal transition (EMT) has also been proposed.[11] Activated CAFs are known to produce potent paracrine signals that increase apoptosis resistance, growth, invasiveness, and metastasis of CCA cells. These effects are mediated by a variety of CAF-secreted factors, including matricellular proteins, such as periostin, tenascin-C, and thrombospondin-1, extracellular matrix (ECM) proteases, chemokines, such as stromal cell-derived factor 1 (SDF-1), and growth factors, such as hepatocyte growth factor (HGF), or, as more recently recognized, platelet-derived growth factor (PDGF).[11-14] Complex interactions between these ECM components and growth factors trigger convergent intracellular signaling pathways, promoting increased CCA cell invasion, metastasis, and survival.[12]

The important influence of the desmoplastic stroma on CCA progression suggests that pharmacological targeting of pathways involved in this cross-talk, or even a more selective targeting of CAFs,[15] may provide novel therapeutic opportunities to treat this deadly tumor. To this end, in addition to better understanding the influence of the stromal component on CCA cells, a detailed knowledge of the cellular origin and the key mechanisms in the formation of tumor reactive stroma is of critical importance. A study published in this issue of Hepatology sheds new light on central aspects of CAF biology in CCA (Fig. 1).[16] In the first place, Cadamuro et al.[16] approach the issue of the cellular source of CAFs in biliary malignancies, in particular, their potential derivation from tumoral cells through an EMT process.[11] In a collection of intrahepatic and extrahepatic human CCA tissues, the researchers certainly found positive staining for a panel of phenotypic EMT markers, including Snail1 and Twist. However, coexpression of cytokeratin 7 and α-SMA was never observed in CCA bile ducts, leading the researchers to conclude that a complete transition toward a mesenchymal phenotype did not occur. Interestingly, this phenotypic study was complemented with an elegant, more functional in vivo approach. Human CCA cells (the EGI-1 cell line), engineered to express a fluorescent marker (enhanced green fluorescent protein; EGFP), were xenotransplanted by intraportal injection in immunodeficient mice. After engrafting, xenotransplanted cells undergoing a complete EMT and CAF conversion would be expected to express both the EGFP marker as well as α-SMA. As anticipated, intrahepatic tumors with an abundant stroma formed around EGFP-expressing CCA cells, which were also positive for a human chromosome Y-probe. However, coincident labeling between EGFP and α-SMA, or the human Y-probe and α-SMA, was never observed, whereas all CAFs stained positive for α-SMA and mouse Y-probe. These observations constitute compelling evidence indicating that tumor-infiltrating CAFs would not be generated through an EMT process of CCA cells, at least under these experimental conditions.

Figure 1.

Formation of tumor reactive stroma in CCA. Role of the PDGF-D/PDGFRβ signaling system in CAF recruitment. The origin of CAFs in the CCA-associated stroma is not completely known. Different potential cellular sources have been proposed, including portal or periductal fibroblasts, HSCs, circulating bone marrow-derived precursor cells and also CCA cells through an EMT process. This latter possibility was ruled out in the present study, because no α-SMA-positive mesenchymal cells derived from malignant CCA cells were observed. A novel paracrine cross-talk between CCA cells and CAFs was demonstrated. PDGF-D expression was not detected in normal cholangiocytes, but it was significantly up-regulated and released to the extracellular medium in CCA cells. PDGF-D-mediated activation of PDGFRβ present in CAFs elicits a potent migratory response, resulting in CAF recruitment. This migratory effect triggered by PDGF-D/PDGFRβ on CAFs involves the activation of the small Rho GTPases (such as Rac1 and Cdc42) and the JNK pathway.

In view of the unlikely epithelial origin of CAFs in CCA, the focus of this study shifted to the elucidation of potential alternative mechanisms involved in the recruitment of the reactive tumor stroma. In particular, the role of the PDGF-signaling system was addressed. The PDGF family includes five dimeric ligand isoforms (PDGF-AA, PDGF-BB, PDGF-AB, PDGF-CC, and PDGF-DD) as well as two tyrosine kinase receptors (PDGFRα and PDGFRβ). PDGF-mediated cross-talk between activated myofibroblasts and cholangiocytes has been reported on in models of chronic biliary tract inflammation and fibrogenesis and is attracting increasing attention in CCA biology.[12-14] Systematic immunohistochemical analysis of human CAA tissues revealed that tumor cells were strongly positive for PDGF-A and PDGF-D, weakly expressed PDGF-B, and were negative for PDGF-C and PDGFRβ.[16] On the other hand, α-SMA-expressing CAFs were extensively positive for PDGFRβ. Intriguingly, the negative expression of PDGFRβ in human CCA cells found in this study seems to be at variance with other recent reports.[13, 14] Nevertheless, the prominent expression of PDGFRβ in CAFs and its cognate ligand, PDGF-D, in CCA cells, together with the emerging role of this growth factor in tumor development,[17] prompted the researchers to examine the function of PDGF-D in CAF recruitment. In a series of in vitro experiments, it was cogently demonstrated that CCA cells, in contrast with normal cholangiocytes, secreted high amounts of PDGF-D, and that the presence of this growth factor in conditioned media of tumoral cells elicited a potent migratory response on CAFs. The involvement of PDGF-D in this response was supported by its attenuation in the presence of the PDGFRβ inhibitor, imatinib, or upon small interfering RNA-mediated knockdown of PDGF-D expression in CCA cells. In the context of the CCA tumoral microenvironment, these findings complement the recently reported antiapoptotic effects of CAF-produced PDGF-BB on CCA cells, which are also mediated through the PDGFRβ receptor.[13] Furthermore, Cadamuro et al. now demonstrate that downstream of PDGFRβ, CAF recruitment is mediated through the activation of small Rho GTPases as well as the c-Jun N-terminal kinase (JNK) pathway. Previous studies identified PDGF-D as a prominent factor up-regulated in experimental models of biliary injury and liver fibrosis, with strong activating effects on myofibrobalsts.[18] The current findings of Cadamuro et al. extend the role of PDGF-D in the progression of hepatic disease beyond liver fibrogenesis, revealing this growth factor as a candidate effector in the formation of tumor reactive stroma in CCA. In view of these remarkable observations, it would be interesting to directly examine the contribution of PDGF-D to CAF recruitment in an in vivo model of CCA.[19] From a translational point of view, this study provides further support for the evaluation of selective PDGFRβ inhibitors in large clinical trials of human CCA, following up on preliminary, but encouraging, recent clinical observations.[20]

  • Carmen Berasain, Ph.D.1,2

  • Matias A. Ávila, Ph.D.1,2

  • 1Division of Hepatology and Gene Therapy Centro de Investigación Médica Aplicada Universidad de Navarra Pamplona, Spain

  • 2Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas Clínica Universidad de Navarra Pamplona, Spain