Conjunction junction, what’s the function? CCN proteins as targets in fibrosis and cancers

Abstract

Cellular communication network (CCN) proteins are matricellular proteins that coordinate signaling among extracellular matrix, secreted proteins, and cell surface receptors. Their specific in vivo function is context-dependent, but they play profound roles in pathological conditions, such as fibrosis and cancers. Anti-CCN therapies are in clinical consideration. Only recently, however, has the function of these complex molecules begun to emerge. This review summarizes and interprets our current knowledge regarding these fascinating molecules and provides experimental evidence for their utility as therapeutic targets.

INTRODUCTION

Matricellular proteins are dynamically expressed and are secreted into the extracellular microenvironment, notably the extracellular matrix (ECM). They do not play a stable, structural role in the ECM but regulate cell function by interacting with ECM structural proteins and cell-surface receptors, proteases, hormones, and other bioeffector molecules, and may sequester and modulate activities of specific growth factors (13).

Members of the cellular communication network (CCN) family of matricellular proteins, which include CCN1–6, were first identified in the late 1980s and early-to-mid 1990s (66, 92, 95, 96). Initially, each protein was named based on what type of screen was used to initially identify each protein. In some cases, the name of the protein was based on its structure, such as cysteine-rich 61 (cyr61), or what induced the expression of each protein, such as wnt-inducible secreted protein 1–3 (WISP1–3) or where is was expressed, such as hypertrophic chondrocyte-specific gene product (Hcs24) or nephroblastoma-overexpressed (nov) or by its purported activity, such as connective tissue growth factor (CTGF) or ecogenin (15, 96, 112). In some cases, most notably for connective tissue growth factor, the original names were subsequently found to be misleading, generating much confusion and misunderstanding in the literature and with the lay public. Of course, at the time of their original discovery, it was not recognized that each protein was a member of a related family. This fact was first determined by Bork (10). All CCN proteins share a common four-module domain structure, except CCN5, which lacks a COOH-terminal heparin-binding domain.

On the basis of efforts by many laboratories and, notably, by the International CCN Society, as of 2018, HUGO (a committee of the Human Genome Organisation) now recognizes officially that, in order, Cyr61, CTGF, nov, and WISP1–3 should be renamed CCN1–CCN6 (97). In all cases, people who were the discoverers or pioneers in the early study of each protein had to give up emotional attachments to largely misleading, and consequently nonheuristically useful, names. The story of how we came to rename these proteins has been superbly outlined elsewhere (96, 97) and is beyond the scope of this review.

Since their initial discovery, CCN proteins have been recognized as being of potential developmental and pathological importance due to the observation that their expression is highly spatiotemporally regulated (66). They possess signal peptides and are secreted through the Golgi into the cellular microenvironment during development and under conditions of tissue remodeling and repair. They are also potently dysregulated in pathological conditions in which fibroblasts are activated, such as in fibrotic disease and cancers. That CCN2 (also known as CTGF) was potently induced when transforming growth factor (TGFβ), but not a myriad of other cytokines, was applied to fibroblasts, which led to an initial hypothesis that CCN2 could be a mediator of the profibrotic effects of TGFβ (30). We now know that this initial hypothesis was considerably oversimplistic. However, this hypothesis led to the formation in 1994 of a company, FibroGen, that, initially, was focused on developing anti-CTGF strategies to treat fibrosis. Notably and incredibly, no thanks in part to a reluctance among some to recognize CTGF not as a growth factor but as a matricellular protein, as of the summer of 2019, that is, 25 years later, the anti-CCN2 antibody (FG-3019, pamrevlumab) is now in Phase III trials for both idiopathic pulmonary fibrosis (IPF) and pancreatic cancer (https://www.fibrogen.com/pamrevlumab/). Intriguingly, the actual mechanism underlying the pathological roles and mechanisms of CCN2, and, indeed, the remaining CCN proteins, has remained largely elusive.

In this article, I focus specifically on what we know to be the mechanistic roles of CCN proteins in fibrosis and cancers, and why, mechanistically, these proteins may be appropriate therapeutic targets.

CCN GENE REGULATION

The mechanisms and factors that regulate CCN gene expression have been expertly covered elsewhere, and the reader is referred to these reviews (18, 66). In summary, CCN proteins are regulated by precisely the same pathways that one would expect, should CCN proteins be regulated by and promote a fibrotic and oncogenic microenvironment. As discussed above, since the mid-1990s, CCN2 has been known to be highly induced in fibroblasts by the potent profibrotic cytokine TGFβ (31, 36, 42) and to be highly overexpressed in all fibrotic conditions, including scleroderma, idiopathic pulmonary fibrosis, diabetic nephropathy, atherosclerosis, macular degeneration, gingival hyperplasia, and liver fibrosis (46, 66, 80). Moreover, CCN proteins are generally overexpressed in tumor cells and in stroma, notably in pancreatic and breast cancers and melanoma (4, 66, 125).

In fibroblasts, TGFβ-induced CCN2 gene expression does not require de novo protein synthesis and involves not only the canonical activin-like kinase (ALK5)/Smad pathway, but also the noncanonical mechanotransduction/adhesion pathway, mediated by focal adhesion kinase (FAK)/TGFβ-activated kinase (TAK1)/extracellular signal-regulated kinases (ERK)/yes-activated protein (YAP)/TEA domain family member (TEAD)/Ets1 (30, 32, 36, 57, 67, 68, 91, 119). These results are particularly interesting, as cell attachment to extracellular matrix is obligatory for the appearance of myofibroblasts and the initiation and perpetuation of fibrosis (60, 113). Similarly, TGFβ induces CCN1 by ALK5/FAK/TAK1/ERK/YAP (91, 114). Indeed, the literature considers CCN1 and CCN2 to be prototypical YAP/transcriptional coactivator with PDZ-binding motif (TAZ) targets, and to be particularly responsive to both mechanotransduction and by the ability of actin networks (18). As might be expected, the induction of CCN proteins by mechanical strain is linked to the activation of latent TGFβ (32). Crucially, CCN1 and CCN2 are also induced by hypoxia, and specifically by hypoxia-inducible factor 1-α (HIF1-α; 35, 58, 119, 126).

The noncanonical proadhesive signaling pathway, but not ALK5/Smads, is important for the constitutive upregulation of CCN2 in pathological conditions, including scleroderma (22, 36, 37, 108, 124). That Smad3 was not involved in the hyperactivity of the CCN2 promoter in SSc fibroblasts led to the initial conclusion in 2001 that autocrine TGFβ signaling was insufficient to perpetuate the fibrotic phenotype of SSc dermal fibroblasts (36). Indeed, an activated adhesive pathway is a fundamental characteristic of SSc fibroblasts, and may be linked to a decreased expression of the adhesive signaling repressor phosphatase and tensin homolog (PTEN) (88, 89).

Although the regulation of the other CCN family members has been less studied, it is known that CCN3 is reciprocally regulated to both CCN1 and CCN2 in a yin/yang fashion (103, 104). For example, in fibroblasts, TGFβ suppresses CCN3 mRNA in a fashion that involves ALK5/FAK/ERK, but not YAP/TAK1 (91). That CCN3 and CCN2/CCN1 are reciprocally regulated may have therapeutic implications (see below).

The important consideration to remember is that CCN protein expression does not reflect a particular pathology per se, but of whether or not the individual system examined is being subjected to increased loading/mechanical strain or to hypoxia. This feature is important to consider when interpreting published observations (see below).

CCNs AS MATRICELLULAR PROTEINS

Probably the most helpful progress in the last 20 yr has been the recognition that CCNs are matricellular proteins (17). Many years have been expended uncovering specific CCN receptors, under the assumption that, at least, CTGF was a growth factor. We now know that this assumption is in error. An excellent review focusing on historical attempts to uncover CCN family receptors has been published elsewhere, and the reader is referred to that article (62). Indeed, perhaps the strongest evidence that CCN proteins are properly considered as matricellular proteins has arisen due to the attempts to find a receptor. That CCNs are secreted into the ECM, together with the discovery that CCN proteins bind directly to integrins, the receptors mediating ECM signaling, initially led to the proposal that CCNs are similar to and may be considered as matricellular proteins (63). The integrins through which CCNs signal vary widely depending on the situation, and include: α_vβ₃, α_vβ₅, α₅β₁, α₆β₁, αIIbβ₃, αMβ₂, and αDβ₂. CCN2 can also interact with heparan sulfate proteoglycans (HSPGs) such as syndecan 4. Similarly, CCNs can bind ECM components such as fibronectin. As such, in vitro, CCNs act as proadhesive molecules; in fact, adhesion assays are probably the only robust, universally agreed-upon in vitro assays for assessing CCN activity, at least of full-length CCN proteins.

FAK plays a critical role downstream of integrin engagement, and mediates CCN activities (6, 20, 53, 115). Similarly, CCN1, and likely other CCN proteins such as CCN2, activates YAP via integrin αvβ3 (90). As discussed above, Ccn1/2 expression is also dependent on FAK activation and YAP (28, 108, 109), consistent with a hypothesis that an autocrine proadhesive signaling loop operating via FAK/YAP promotes Ccn1/2 expression and function. In this regard, it is important to remember the point, described above, that it is now understood among experts within the fibrotic field that an autocrine proadhesive signaling loop operating via integrin-mediated FAK activation is both necessary and sufficient to perpetuate the fibrotic phenotype (60, 71, 113).

Integrin receptor binding domains decorate CCN molecules. Thus, it is important to remember that CCN activities mediated by one integrin binding domain are likely to be distinct from those mediated by a different integrin binding domain; for example, an amino acid mutation (D125A) at the CCN1 binding site for α_vβ₃/α_vβ₅ that abolishes CCN1 binding to integrins α_vβ₃ and α_vβ₅ and impairs α_vβ₃/α_vβ₅-dependent functions does not affect other CCN1-dependent functions mediated by α₆β₁ integrin (21). Evaluating the biological significance of these individual interactions requires assessment in vivo, and, testing each individual interaction, although informative, may not necessarily give an accurate picture of what the in vivo role of the entire CCN protein may be at a particular juncture [see section CCN1 (cyr61)].

CCNs also interact with receptors other that integrins; for example, CCN2 signals through TrkA (26, 122), and binds both LRP1 (51, 107) and the mannose 6 phosphate receptor (7). The in vivo role of these receptors in mediating CCN activity is unclear.

These observations are further complicated by recent findings, supported by earlier functional data, that, to be active, CCN proteins may have to be cleaved into COOH-terminal heparin-binding fragments (49). Indeed, truncated and full-length CCN2 and CCN3 isoforms have been observed (16, 92). These issues might be particularly important in the case of cancer and fibrosis, which are characterized by the overproduction of many different kinds of proteases. The dysregulation of proteolysis would increase the relative ratio of truncated over full-length isoforms, therefore causing abnormal signaling. Of course, the receptors responsible for signaling by fragments might differ than those that have been detected using full-length material. Indeed, it is not a novel idea that CCN fragments may have different activities than full-length protein (16, 92–94).

The important consideration is that these proteins are not growth factors, making it understandable how it has been essentially impossible to establish robust cell-based assays (other than those examining cell adhesion). To date, the effects of CCN proteins have been only effectively studied in vivo. Nonetheless, it is probable that, in the future, the generation of 3D culture models could prove invaluable.

CCN2 (CTGF)

I will start with the most studied member of the CCN family, CCN2. Initial studies largely focused on the correlation of high CCN2 expression with clinical outcome, that is, as a potential surrogate marker of disease (66, 104). The initial observation that CCN2 was induced by TGFβ was extremely misleading, as CCN2 is not exactly a downstream mediator of TGFβ’s effects, but rather, when initially present, acts as cofactor to amplify TGFβ action, through adhesive/ERK pathways. This was first shown, elegantly, by Takehara and subsequently by other groups (52, 82, 83, 109). This notion is supported by the fact that fibroblast-specific expression of CCN2 is utterly dispensable for normal, cutaneous tissue repair in mice, including the formation of myofibroblasts in response to injury (76). Thus, it is somewhat frustrating that people still have the erroneous impression that CCN2 is a downstream mediator of TGFβ, as no expert in the CCN field has made that claim, nor is it based on any experimental observations.

Because of the misconceptions outlined above, it has awaited the development of key reagents, notably genetic knockout animals and antagonistic antibodies (FG-3019) to demonstrate conclusively in vivo activities of CCN2 and, hence, to validate CCN2 as a bona fide therapeutic target. The anti-CCN2 antibody FG-3019, which may act as a clearance antibody, is a high-affinity humanized monoclonal antibody (14). In models of pancreatic cancer, this antibody has been shown since 2006 to be effective at blocking metastasis in models of pancreatic cancer and also in enhancing the response to chemotherapy, without affecting the efficacy of drug delivery (25, 85). This antibody also blocks melanoma metastasis, impedes ovarian cancer progression, and prolongs survival in response to chemotherapy in acute lymphoblastic leukemia (64).

FG-3019 also successfully protects against animal models of idiopathic pulmonary fibrosis (IPF), including radiation-induced fibrosis (106). Although FG-3019 was tested in human diabetic patients (1), it appears that Phase I/II trials have not been initiated or reported on, possibly due to lack of efficacy because of possible issues with lack of antibody penetration into target tissues in diabetic patients. However, FG-3019 has had a successful Phase II trial for IPF (65, 102). In terms of pancreatic cancer, the results of a Phase I/II trial were published in an OMICS journal that is unfortunately no longer accessible online. Nonetheless, Phase III trials have been initiated for IPF (Zephyrus, March 2023) and pancreatic cancer (Lapis, December 2023), and a Phase II trial is ongoing for muscular dystrophy, with an expected completion date of April 2021.

Genetic analyses using animal models have shown that CCN2 expression by fibroblasts is required for skin fibrosis. and the skin and lung fibrosis caused by loss of PTEN (i.e., by the activation of adhesive signaling/Akt), and liver fibrosis (74, 75, 89, 98). CCN2 is required for the formation of myofibroblasts in fibrotic, but, as mentioned above, not in tissue repair, models (74, 76). In fibrosis, myofibroblasts are derived from “your favorite” progenitor cells, including resident fibroblasts that acquire a progenitor cell-like phenotype, that stably differentiate into myofibroblasts and are, therefore, positive for progenitor cell markers such as NG2, platelet derived growth factor receptor (PDGFRβ), or Sox2 (sex-determining region Y-box 2) (23, 56, 73, 100, 116). This population does not appear to be a significant contributor to tissue repair (50, 76, 117). Although synthetic (collagen-expressing) fibroblasts clearly play an important role in both cutaneous tissue repair and fibrosis, only in fibrotic/cancer models do synthetic fibroblasts (33, 71, 72, 74) generate stable myofibroblasts, likely through a progenitor cell-like intermediate (117). CCN2 is not required for the formation of the preponderance of myofibroblasts during tissue repair but is required for the differentiation of the progenitor cell intermediate to the myofibroblasts that occur in fibrotic/cancer models (39, 74, 76, 116). This latter activity appears to occur in response to activated mechanotransduction pathways.

In cancer models, CCN2 expression by CAFs is also required for metastasis, including the formation of new blood vessels via vasculogenic mimicry, apparently because CCN2 is required for the expression of both periostin and integrin α11 (ITGA11) (40, 41). In human melanoma patients, CCN2 expression correlates with a highly fibrotic subset of ITGA11/COL1A1/FAP-expressing CAFs that is correlated with poor clinical outcome (117). CCN2 expression is independent of BRAF mutational status, suggesting that CCN2 may be a target for patients resistant to BRAF inhibitors (40). As might be anticipated, in human melanoma patients, CCN2 expression correlated with the development of a fibrotic tumor stroma (40).

Collectively, these data suggest that the role of CCN2 in vivo is to promote the differentiation of progenitor cells to a myofibroblast phenotype, and that the presence of these CCN2-dependent myofibroblasts correlates with a poor clinical prognosis. Thus, CCN2 would appear to be a bona fide therapeutic target.

CCN1 (cyr61)

CCN1 is the second most studied CCN family member. The available clinical data also suggest that CCN1 is also a target for both fibrosis and for cancer (4, 5, 59, 99), although the data are somewhat confusing. It should be pointed out that perceived divergences in the published data may be explained because, sometimes, the effects of endogenous CCN1 are examined, and, at other times, the effects of exogenously added or overexpressed CCN1 are evaluated. Overexpression of CCN1, and, indeed, of all CCN proteins, causes an unfolded protein response, a stress-generated response in the endoplasmic reticulum, and cellular death (11). Also, observed effects may differ based on the molecular basis of why CCN proteins are being expressed. As discussed above, generally when endogenous CCN proteins are induced, it is due to mechanical loading or hypoxia; if triggers other than mechanical loading or hypoxia are examined, it is entirely conceivable that different results may be obtained.

I will start with data consistent with the notion that targeting CCN1 may be a useful therapeutic strategy in both cancers and fibrosis. As discussed above, CCN1, like CCN2, is induced by both TGFβ and hypoxia. CCN1 promotes angiogenesis and tumor growth in vivo (2) and promotes models of breast cancer by promoting tumor cell extravasation and protecting from anoikis (38). Similarly, CCN1 promotes progression of pancreatic tumors and metastasis by promoting neoangiogenesis, via hedgehog signaling, and also resistance to chemotherapy (4, 5, 34, 77, 78). The data involving the role of CCN1 in melanoma are fairly limited; however, one study reported, as expected, that hypoxia and ECM stiffness promoted CCN1 expression in endothelial cells, and that knockout of Ccn1 in endothelial cells inhibited the binding of melanoma cancer cells to blood vessels (99). This phenomenon, a critical step in the transit of cancer cells through the vasculature during metastasis (99), is consistent with the notion that CCN1 promotes metastasis in melanoma.

The data regarding the role of CCN1 in tissue repair and fibrosis are somewhat contradictory. CCN1 expression by fibroblasts contributes to bleomycin-induced skin fibrosis; loss of CCN1 expression by fibroblasts results in a disorganized, nonlinear collagen fiber network, suggesting that CCN1 may act as a molecular chaperone to promote collagen organization (99). Moreover, in humans, CCN1 is overexpressed in acute lung injury; in mice, adenoviral-mediated overexpression of CCN1 in lung promoted inflammation, causing fibrosis and the expression of profibrotic proteins (29, 59). Adenoviral-based overexpression of CCN1 by in portal myofibroblasts in liver, however, causes cellular senescence that suppressed fibrogenesis; as pointed out above, this appears to be because this causes an unfolded protein stress response (11, 12).

One thing to remember is that CCN proteins, including CCN1 and CCN2, readily generate reactive oxygen species (ROS) (19, 47). Of course, low levels of ROS act to promote signaling, whereas high levels of ROS act to promote stress (81). For example, ROS, generated by NOX4, works as a signaling molecule downstream of integrin β1 to promote profibrotic gene expression in healthy fibroblasts and also mediates the fibrotic phenotype of scleroderma fibroblasts and animal models of scleroderma (70, 84, 110, 123). Conversely, ROS is also a key component of a stress response, for example, that generated in the endoplasmic reticulum during an unfolded protein response. This latter response can be cytotoxic. Thus, low levels of CCN1 may promote signaling via ROS but much higher amounts of CCN1, which may or may not be physiologically relevant, may cause cell mortality and, hence, indirectly due to causing (myo)fibroblast death, limit excessive collagen deposition in fibrotic models (11, 12, 48).

Of course, the physiological role of CCN1 entirely depends on which integrin is engaged; CCN1 promotes cell proliferation, survival, and angiogenesis via integrin α_vβ₃, but it promotes apoptosis and senescence through integrin α₆β₁ and heparan sulfate proteoglycans, as well as a sustained level of ROS (61). Thus, mutation of a single integrin α₆β₁ binding domain in CCN1 results in a senescence-defective CCN1 protein and enhances fibrotic responses to wounding (45). Conversely, CCN1 induces a proinflammatory M1-like genetic program in macrophages and promotes adhesion of activated macrophages via integrin α_Mβ₂ and HSPGs (3). Thus, the observed results may differ depending on which integrin is being engaged by CCN1 and the relative affinity of CCN1 to each integrin.

In summation, CCN1 and CCN2 may generate distinct but overlapping activities and depending on the context, both appear to contribute to fibrosis and cancers. These observations suggest that careful modulation of the pathological effects of both proteins might be useful therapeutically to treat these conditions.

CCN3 (nov)

CCN3 is somewhat less studied that CCN1, but is equally confusing. CCN3, or nov, was initially discovered as an integration site for the MAV retrovirus in avian nephroblastomas (44). This gene was designated nov because it was highly expressed in all nephroblastomas, compared with adult kidney where its expression was low, yet detectable. In this initial report, CCN3 was observed to inhibit growth of chicken embryonic fibroblasts, and thus provided the first evidence that the CCN family contains negative regulators (44, 92). The COOH-terminal module of CCN3 was sufficient to induce a strong (80%) cellular growth inhibition, due to the ability of CCN3 to act as a brake on the cell cycle, resulting in an accumulation of CCN3-treated cells in the S-phase of the cell cycle (9). Thus emerged the notion of coregulation by CCN proteins where CCN3 may act as a counter-regulator interfering with a pathogenic process caused by another CCN protein (44, 91, 92, 103, 104).

The role of CCN3 in cancers appears to be stage-specific and complex. For a detailed essay, the reader is referred to Ref. 125. To explain what CCN3 may be doing in cancers, I will, for simplicity, focus on melanoma. CCN3 expression is initially high. Generally, the primary tumors and skin metastases appear to lose CCN3 expression; this loss appears to permit growth and invasion (27, 118). Conversely, CCN3 expression appears to be significantly higher in distant, visceral metastases, and to permit the establishment of tumor cells in their new location (118). Consistent with this notion, CCN3, when exogenously added to 1205Lu melanoma cells, decreased cell invasion and the transcription and activation of matrix metalloproteinase-2 and -9 (27). Conversely, the high levels of CCN3 at distal visceral sites are correlated with the ability of CCN3 to promote melanoma cells’ adhesive ability and integrin expression (118), which would permit cells to embed in their new location. Collectively, these data are consistent with the idea that loss of CCN3 permits the development of a fibrotic microenvironment, surrounding the initial tumor site, which is required for cancer cells to metastasize.

Indeed, the available data indicated that loss of CCN3 expression is permissive for fibrosis. In anti-Thy1-induced glomerulonephritis, CCN1 and CCN2 expression was increased, whereas CCN3 expression was decreased. Glomerular cell proliferation was positively correlated with CCN1 and CCN2 expression, but negatively correlated with CCN3 expression (121). In the mesangial cell culture model of diabetic renal fibrosis, CCN3 and CCN2 work in a yin/yang-like manner (105). When applied to mesangial cells, TGFβ induced CCN2, but suppressed CCN3 expression. Conversely, this ability was blocked in the presence of recombinant human CCN3 (rhCCN3) (105). Similarly, CCN3 blocks TGFβ-induced CCN1 and CCN2 mRNA expression in human dermal fibroblasts (91). Finally, exogenously added CCN3 blocked the fibrosis seen in the BTBR ob/ob mouse, and, more specifically, protected against, and even reversed, the podocyte loss and the establishment of glomerular hypertrophy in this model (103).

Generally speaking, CCN3 expression is high, and CCN1 and CCN2 expression low, during tissue homeostasis. During pathologies or in disease models, CCN1 and CCN2 expression increases, whereas CCN3 expression decreases. These data have led to the hypothesis that adding exogenous CCN3, by restoring the CCN1/2:CCN3 ratio to that found in healthy tissue, could be used as an antifibrotic therapy (103–105). Precisely why CCN3 behaves antagonistically to CCN1/CCN2 is unclear; it may act as a dominant negative by blocking CCN1/2 binding to the cell surface by binding to the same integrins engaged by CCN1/2, thus preventing CCN1/2 from bringing proteins binding CCN1/2 to the cell surface.

CCN4–6 (WISP1–3)

There is relatively little known about these three proteins. For a review on the role of CCN4 in cancers, the reader is referred elsewhere (86). Briefly, CCN4 appears to stimulate aggressive behaviors of various cancers, except in lung cancer and prostate cancer cells, where CCN4 does the opposite. In terms of fibrosis, CCN4 inhibition by a CCN4 monoclonal antibody in vivo significantly attenuated CCl₄-induced liver injury and the progression of liver fibrosis (69). Similar results are also seen in a mouse model of pulmonary fibrosis; neutralizing monoclonal antibodies specific for CCN4 reduced the expression of genes characteristic of fibrosis and markedly attenuated lung fibrosis, by decreasing collagen deposition and improving lung function and survival (55). Recently, it was shown that CCN4 is a novel maker of obesity in diabetic patients (54).

CCN5, a ~29–35 kDa protein that lacks the heparin-binding carboxy-terminal domain, acts as a dominant negative molecule. CCN5 may be a novel therapeutic agent to treat cardiac hypertrophy and heart failure. CCN5 can reverse established cardiac fibrosis by inhibiting the generation of and enhancing apoptosis of myofibroblasts in the myocardium (43). CCN5 is also a tumor suppressor gene in pancreatic cancer. CCN5 protein is overexpressed in normal and preneoplastic cells, but it is mostly undetected or minimally detected in various pancreatic and breast cancer cell lines and tissue samples (4). Functional studies demonstrate that the exposure of pancreatic cancer cells to CCN5 recombinant protein reverses epithelial/mesenchymal transition (24). Whether CCN5 acts by soaking up proteins binding CCN1/CCN2 or whether CCN5 binds cell surface receptors without engaging additional receptors, such as heparan sulfate-containing proteoglycans, which bind the carboxy-terminal domain of CCN proteins is unclear.

The role of CCN6 is largely unknown; but Ccn6^fl/fl;MMTV-Cre mice developed invasive mammary carcinomas with histopathological features and gene signatures resembling human spindle cell metaplastic breast carcinoma. Ccn6^fl/fl;MMTV-Cre mammary tumors were highly aggressive and metastasized in 46% of the cases. CCN6 loss, due to mutations or other mechanisms, results in deregulated BMP4 and IGF signaling, with increased p-38/TAK1 activation, and activation AKT1 (79).

CONCLUSION

The available evidence strongly supports the idea that coordinated, collective action of CCN proteins and their partners controls fibrosis and cancer progression. The different actions that CCN proteins mediate vary widely, depending on the proteins and receptors they interact with, and the resultant downstream signaling pathways they stimulate or suppress. The outcome, therefore, also depends on the bioavailablity of the various partners, on the affinity of each protein for the particular partner, and the domains of each CCN molecule that are physically present in the microenvironment (Fig. 1). They truly form a cellular communication network that integrates signals that are occurring at a particular point in time. Indeed, stage-specific therapeutic intervention targeting CCN proteins is warranted.

Fig. 1.

Combinatorial interactions dictate cellular communication network (CCN) function. Each domain of CCN proteins interacts with different ligands. What CCN domains are present in the microenvironment depends on which CCN molecules are being synthesized [for example, CCN5 lacks the heparin-binding carboxy-terminal (CT) domain], protease activity, and differential splicing. The affinity of CCN molecules for each ligand is likely to vary. Thus, the overall biological effect of CCN proteins can vary widely depending on bioavailability of CCN proteins, their modules/fragments, and their interacting partners. CCN proteins act as central mediators of mechanotransduction. BMP, bone morphogenetic protein; HSPG, heparan sulfate proteoglycans; LRP, lipoprotein receptor-related protein; TSP, thrombospondin-1; VWC, von Willebrand factor C.

GRANTS

This work was supported in part by Arthritis Society Grant SOG-17-0039, Canadian Institutes of Health Research Grant MOP-77603, and Natural Sciences and Engineering Research Council of Canada Grant RGPIN-2016-04756.

DISCLOSURES

AL is a shareholder (less than 5% or $500,000) in FibroGen.

AUTHOR CONTRIBUTIONS

A.L. drafted manuscript; edited and revised manuscript; and approved final version of manuscript.