Defining the versican interactome in lung health and disease

Abstract

The extracellular matrix (ECM) imparts critical mechanical and biochemical information to cells in the lungs. Proteoglycans are essential constituents of the ECM and play a crucial role in controlling numerous biological processes, including regulating cellular phenotype and function. Versican, a chondroitin sulfate proteoglycan required for embryonic development, is almost absent from mature, healthy lungs and is reexpressed and accumulates in acute and chronic lung disease. Studies using genetically engineered mice show that the versican-enriched matrix can be pro- or anti-inflammatory depending on the cellular source or disease process studied. The mechanisms whereby versican develops a contextual ECM remain largely unknown. The primary goal of this review is to provide an overview of the interaction of versican with its many binding partners, the “versican interactome,” and how through these interactions, versican is an integrator of complex extracellular information. Hopefully, the information provided in this review will be used to develop future studies to determine how versican and its binding partners can develop contextual ECMs that control select biological processes. Although this review focuses on versican and the lungs, what is described can be extended to other proteoglycans, tissues, and organs.

INTRODUCTION

The epithelial surfaces of the lungs are the largest mucosal surface of the body that has direct contact with the outside environment, thereby exposing the lungs to a broad array of particulates, toxins, and pathogens. These constant insults result in repetitive inflammation and injury, requiring pulmonary host defenses to maintain the normal lung function necessary for proper gas exchange and oxygen delivery to the systemic circulation. To protect against the various external insults, the early innate host defenses of the lungs consist of physical and cellular defenses (Fig. 1). Physical host defenses include anatomical barriers such as the airway architecture and the epithelial mucosal barrier, mucociliary clearance, and cough reflexes (2). Cellular defenses are mediated by airway epithelial cells, resident immune cells such as alveolar macrophages, and leukocytes recruited in response to an insult (Fig. 1) (2–4). It is now recognized that the extracellular matrix (ECM) also plays a critical role in lung health and disease (5). However, there is a significant knowledge gap in our understanding of how the ECM regulates innate immunity in the lungs.

Figure 1.

Pulmonary host defenses include anatomical barriers such as the epithelial mucosal barrier, mucociliary clearance, and the cough reflexes and cellular responses of the innate and acquired immune systems. Central to the early host response in the lungs is the innate immune system that recognizes microbial pathogens through pattern recognition receptors (PRR). An example of a PRR is Toll-like receptor-4 (TLR4) on macrophages (MΦ) and other cells in the lungs. TLR4 recognizes pathogen-associated molecular patterns, such as lipopolysaccharide (LPS), a component of the cell wall of Gram-negative bacteria; it also recognizes DAMPs originating from the ECM, such as hyaluronan, biglycan, and versican. Upon activation of TLR4, cytokines, chemokines, and growth factors are synthesized. These mediators are a critical component of the innate immune response, which leads to the pulmonary recruitment of neutrophils (PMN) and other immune cells into the lungs. A significant gap in our knowledge is a lack of understanding of how the spatial-temporal changes in versican and the versican-interactome control the innate immune response in the lungs. DAMPs, damage/danger-associated molecular pattern; ECM, extracellular matrix. [Adapted with permission from Gill et al. 2010 (1)].

The ECM is a complex scaffold surrounding cells that imparts mechanical and biochemical properties, which control cell differentiation, proliferation, survival, and migration. Genomic and proteomic analysis of the ECM provides a more complete picture with the known list of constituents being called the “matrisome” (6). The matrisome includes the “core matrisome” that is made up of over 300 proteins that include ECM glycoproteins, collagens, and proteoglycans (PGs) (6, 7). In addition, the matrisome contains ECM-modifying enzymes, ECM-binding growth factors, and other ECM-associated proteins (6, 8, 9). The ECM is dynamic with the remodeling of its components, leading to the development of select microenvironments that control biological processes in the lungs including development, homeostasis, and disease (10–15). Proteoglycans are essential constituents of the core matrisome that play a critical role in determining the mechanical and cellular properties in normal and diseased lungs.

Proteoglycans are a family of molecules with complex macromolecular structures (16, 17). The basic structure of PGs includes a core protein and one or more covalently attached glycosaminoglycan (GAG) side chain(s). Glycosaminoglycans are long linear negatively charged polymers of repeating disaccharides classified into four groups, heparan sulfate/heparin, chondroitin sulfate/dermatan sulfate, hyaluronan (HA), and keratan sulfate. Most GAGs are covalently attached to a core protein; HA is the exception as it is synthesized at the plasma membrane and is not associated with a core protein (18, 19). The PG family encompasses 43 distinct genes with numerous alternatively spliced variants divided into four major classes: intracellular, cell-surface, pericellular, and extracellular PGs (16, 18). The extracellular chondroitin sulfate PGs (CSPG) consist of small leucine-rich PGs, including biglycan, decorin, and lumican, and large aggregating hyalectans versican, aggrecan, neurocan, and brevican (16, 18). This review will focus on versican, a CSPG, and its binding partners—“the versican interactome”—which are essential in lung health and disease (17, 20–22).

Versican is an integral component of the ECM during embryonic development but is absent from mature, healthy lungs (23). In contrast to healthy lungs, versican is consistently reexpressed and accumulates in human patients with various lung diseases and mouse models of lung inflammation and injury (20, 21, 23–34). What is not well understood is what role versican plays in lung inflammation and injury. To address this limitation, we constructed the Vcan^tm1.1Cwf mouse strain (i.e., Vcan^fl/fl mouse), in which loxP sites were placed in introns 3 and 4, to allow Cre recombinase-mediated deletion of Vcan exon 4 and subsequent frameshift mutation resulting in a premature STOP codon near the beginning of exon 5. The development of the Vcan^fl/fl mouse strain provides a mouse model in which versican can be deleted in a time- or cell-specific manner. Three collaborating laboratories have subsequently developed mice with versican deficiency in various cells in the lungs. These novel strains of mice include the R26/Vcan^−/− mouse strain, which has a global deletion of versican when treated with tamoxifen; the LysM/Vcan^−/− mouse strain, which lacks versican in myeloid cells and type II epithelial cells; and the SPC/Vcan^−/− mouse strain, which lacks versican in airway epithelial cells including club cells and type II epithelial cells in the lungs (24, 26, 27). When R26/Vcan^−/− mice were treated with polyinosinic-polycytidylic acid [poly(I:C)], the recovery of recruited cells in bronchoalveolar lavage (BAL) fluid was significantly decreased, suggesting that under these conditions versican is proinflammatory (26). In contrast, when LysM/Vcan^−/− mice were treated with poly(I:C), the total number of cells in the BAL fluid was significantly increased, suggesting an anti-inflammatory phenotype (24). Studies performed in the SPC/Vcan^−/− mice exposed to the respiratory syncytial virus (RSV) observed a significant increase in neutrophils and monocytes in the BAL fluid of mice, suggesting that under the conditions studied, versican is anti-inflammatory (27).

In summary, the different results from R26/Vcan^−/−, LysM/Vcan^−/−, and SPC/Vcan^−/− mice show that versican can have both pro- and anti-inflammatory roles. These findings suggest that versican has immunomodulatory properties that provide contextual extracellular control of the immune response in the lungs. What is not known are the mechanisms responsible for the contextual nature of the versican-enriched ECM in the lungs.

Based on the findings from these studies in three strains of versican-deficient mice, our initial interpretation was that differences in the cellular source of versican could in part explain the contextual nature of versican. In this scenario, myeloid and epithelial cells would synthesize an anti-inflammatory form of versican. In contrast, stromal cells or endothelial cells would produce a proinflammatory form of versican. However, this interpretation likely does not adequately explain the contextual nature of versican. Other considerations include the agonist or signaling pathway responsible for the increased expression of versican. Two pathways, the β-catenin/T-cell factor pathway in stromal cells (35, 36) and the type I interferon (IFN) signaling pathway in macrophages and fibroblasts are known to regulate Vcan expression (24, 37). Whether select agonists such as Toll-like receptor (TLR) agonists, bacteria, or viruses or one of the two known signaling pathways regulating versican expression promote a versican-enriched ECM that is pro- or anti-inflammatory is not yet known.

Two additional mechanisms that could influence the immunomodulatory properties of a versican-enriched ECM are changes in the structure of versican or differences in the synthesis of select versican binding partners. Structural changes to versican could include differences in the isoform of versican synthesized, proteolytic modifications of the core protein, or alterations in the sulfation pattern of the CS side chains of versican. One would also expect that the agonist or signaling pathway will impact the expression of components within the versican interactome, as well as versican itself. Unfortunately, we know very little about how spatial-temporal alterations in versican expression and accumulation, versican structure, or the versican-interactome will alter the lungs’ immune response. What is apparent is that through its interactions with its many binding partners, versican can be considered an integrator of complex information in the ECM.

Studies using two distinct agonists, influenza virus and cockroach antigen (CRA) provide some insight into spatial-temporal alterations in versican expression in the lungs (30, 37). In mice treated with mouse-adapted influenza A virus (IAV, A/Puerto Rico/8/34, H1N1), versican expression peaked between 6 and 9 days postinfection (dpi) and remained significantly elevated at 12 dpi. An increase in versican expression was seen in both myeloid cells expressing CD68 and stromal cells expressing the platelet-derived growth factor receptor β. In both cell types, the increased versican expression was regulated by both type I IFN-dependent and -independent signaling pathways (37). The type I IFN-independent signaling is most likely the result of signaling through the β-catenin/T-cell factor pathway (35, 37). In addition to differences in the signaling pathway involved with increasing versican expression, the changes in the accumulation of versican from being predominantly epithelial in PBS controls to developing strong immunopositivity in perivascular, peribronchiolar, and alveolar septal regions throughout the clinical course of influenza reflect the dynamic and fluid nature of the versican-enriched ECM.

Evaluation of lungs of mice exposed to CRA shows positive immunostaining for versican beginning at 6–12 h and remaining increased for up to 8 wk after exposure (30). Versican staining in response to CRA occurred in similar anatomical regions as was observed for influenza virus—peribronchiolar, perivascular, and alveolar septa. In addition, the authors provide evidence suggesting that epithelial cells, fibroblasts, and macrophages were the cellular source of versican in mice exposed to CRA. Taken together, these two studies similarly showed that the positive staining for versican correlated with the influx and spatial location of inflammatory cells, suggesting that the versican-enriched matrix was proinflammatory. What is not known is whether differences in the cellular source or signaling pathway responsible for the synthesis of versican or members of the versican interactome exist in specific microenvironments (e.g., perivascular vs. peribronchiolar) in the lungs in mice exposed to influenza virus versus CRA (Fig. 2).

Figure 2.

Versican accumulation and the development of select microenvironments in the lungs of a mouse infected with influenza A virus for 6 days. We know very little about the changes in the versican-interactome in these tissue microenvironments. Blue arrows highlight positive staining for versican in the peribronchiolar space directly beneath the airway epithelium. A gray arrow highlights versican accumulation in the perivascular space of a vessel adjacent to the bronchiole. The black arrow points to an expanded perivascular space (presumptive edema) adjacent to a bronchiole that has positive staining for versican. The purple arrow highlights positive staining for versican in the alveolar septa. The tissue was counterstained with hematoxylin. *Blood vessel lumen. AV, alveolar space; BL, bronchiolar lumen.

Although the studies described above provide a glimpse at the contextual nature of the versican-enriched matrix and the spatial-temporal heterogeneity in versican accumulation, they provide limited information regarding changes to the many molecules known to bind to versican, the versican-interactome. This review provides detailed information about the interactions of versican with its binding partners and, to the extent possible, what is known regarding how these interactions modify the innate immune response and pulmonary inflammation. The primary goal is to better understand how versican and select components of the versican-interactome provide contextual extracellular control of the host response to external insults to the lungs.

VERSICAN STRUCTURE

The hyalectans are a distinct family of PG named for their dual ability to bind HA and lectins (16, 38). When expressed as a whole molecule (designated V0), versican is the largest member of the hyalectan family. In this review, we focus our discussion on the most pertinent features of the versican-interactome in lung health and disease. The versican gene (VCAN) has 15 exons, in which the N-terminal G1 domain (exons 2–6) and the C-terminal G3 domain (exons 9–15) encompass two CS-binding domains, the α-GAG (exon 7) and β-GAG (exon 8) domains (Fig. 3). The G1 domain of versican has A, B, and B’ subdomains. The A subdomain is an immunoglobular (Ig)-like motif, followed by the B and B’ subdomains that have a “link module structure” that binds to HA and hyaluronan proteoglycan link protein 1 (HAPLN1) (39, 40). The G3 domain has two epidermal growth factor (EGF)-like repeats, a carbohydrate recognition domain (also called the lectin-like domain) and a complement binding protein-like motif. The G3 domain binds to tenascins, fibronectin, fibrillins, fibulin-1, fibulin-2, β-integrin, and P-selectin glycoprotein-1 (PSGL-1) (reviewed in Refs. 41 and 42). The CS sidechains of the αGAG and βGAG domains interact with cell surface receptors, cytokines, chemokines, lipoproteins, and proteases (reviewed in Refs. 17, 21, 41, 42).

Figure 3.

Schematic showing the four versican isoforms of versican binding to hyaluronan through the link modules in the G1 domain. The G1 domain of versican has A, B, and B’ subdomains. The A subdomain is an immunoglobulin (Ig)-like motif, followed by the B and B’ subdomains with a “link module structure that binds to HA.” The C-terminal G3 domain has two epidermal growth factor (EGF)-like repeats, a carbohydrate recognition domain (CRD, also called the lectin-like domain) and a complement binding protein (CBP)-like motif. The two CS binding domains, the α-GAG and β-GAG domains, are interposed between the G1 and G3 domains. The V3 isoform of versican lacks both GAG-binding domains and therefore lacks CS side chains. The V4 isoform of versican contains the N-terminal portion of the β-GAG domain (i.e., the green region in V1) interposed between G1 and G3. The CS side chains and the multiple domains of the versican-core protein provide many binding sites for versican’s binding partners. CS, chondroitin sulfate; GAG, glycosaminoglycan.

Versican mRNA can undergo alternative splicing at exons 7 and 8, resulting in five isoforms with significant differences in CS content and overall size (Fig. 3). V0 contains both the αGAG and βGAG domains (∼370 kDa), V1 contains only the βGAG domain (∼263 kDa), V2 contains only the αGAG domain (∼180 kDa), and V3 lacks both GAG-binding domains (∼74 kDa) (39, 43–45). The fifth isoform of versican, V4, contains the G1 and G3 domains and an N-terminal portion of the βGAG domain (46). Identification of the consensus sequence for chondroitin sulfate attachment sites in the αGAG and βGAG domains of versican suggest that the potential number of GAG chains on human versican is 17–23 for V0, 12–15 for V1, 5–8 for V2, and 0 for V3 (47). The core protein of versican can also be modified by proteases, including A disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS) and matrix metalloproteinases (MMPs) (29, 48–52). Degradation of versican by ADAMTS and MMPs results in the development of bioactive molecules, such as versikine, that play an important role in health and disease (reviewed in Refs. 21, 39, 53–55).

VERSICAN INTERACTOME

Other excellent reviews have described the diverse binding partners of versican (41, 42). In this review, we broaden the definition of the versican interactome based on the classification developed for the matrisome by Hynes and Naba (6, 9). The new definition of the versican interactome includes proteins that are categorized as versican-associated factors, versican-core matrisome, and versican modifying enzymes (Table 1). Versican-associated factors include HA and several versican-associated proteins that are incorporated into the versican-HA macromolecular complex developed under conditions of tissue inflammation (22, 56). These Vcan-HA-affiliated proteins include tumor necrosis factor-stimulated gene-6 (TSG6), heavy chains (HC), and HAPLN1. Versican-associated factors also include adhesion molecules such as CD44, L-selectin, PSGL-1, and β-integrin; secreted factors such as cytokines and chemokines that bind to the CS side chains of versican and develop transient interactions (Table 2); and cell surface receptors such as Toll-like receptor-2 (TLR2), TLR4, and epidermal growth factor receptor (EGFR) that bind to versican and activate specific signaling pathways (84–86). The versican-core matrisome proteins include thrombospondin-1, tenascins, fibulin-1, fibulin-2, fibrillins, type I collagen, and fibronectin, which bind predominantly to the G3 domain of versican and develop stable interactions. And finally, the versican-modifying enzymes include select ADAMTS family members and MMPs. These enzymes bind versican at different locations resulting in the degradation and removal of versican from tissues. Degradation of versican also results in the development of versican fragments such as versikine that are biologically active (39, 53). To fully understand the contextual nature of the versican-enriched matrix will require the development of studies that evaluate multiple components of the versican-interactome together with changes to versican.

Table 1.

Versican-interactome

Versican Interactome	Binding Partners
Versican-Associated: HA-Affiliated Proteins	TSG-6, heavy chains (HC), and HAPLN1
Versican-Associated: Adhesion Molecules	CD44, L-selectin, PSGL-1, and b-integrin
Versican-Associated: Secreted Factors	Chemokines, cytokines, and growth factors
Versican-Associated: Cell Surface Receptors	TLR2, TLR4, and EGFR
Versican-Core Matrisome	Thrombospondin-1, Tenascins, Fibulin-1, Fibulin-2, Fibrillins, Type I collagen, and Fibronectin
Versican-modifying enzymes	ADAMTS-1, -4, -5, -9, and -20, MMP-1, -2, -3, -7, -9, and -12

Hynes et al. and Naba et al. (6, 9). EGFR, epidermal growth factor receptor; HA, hyaluronan; PSGL-1, P-selectin glycoprotein-1; TSG-6, tumor necrosis factor-stimulated gene-6; TLR, Toll-like receptor.

Table 2.

Versican-associated secreted factors

Inflammatory Mediator	GAGs and PGs that Interact	References
CXC-Chemokines
CXCL1, CXCL2, CXCL3, CXCL4, CXCL8, CXCL10, CXCL12	HS, CS, HSPGs, and CSPGs	(57–61)
CC-Chemokines
CCL2, CCL3, CCL4, CCL5, CCL11	HS, collagen XVIII, and versican	(57–59, 62–65)
Proinflammatory cytokines
IL-1α IL-1β, IL-2, IL-5, IL-6, IL-7, IL-12, TNFα, IFNγ	HS, heparin, perlecan, CD44v3, CSPGs	(66–75)
Anti-inflammatory cytokines
IL-4, IL-10, IFNβ	HS	(76, 77)
Growth factors
FGFs, VEGF, GM-CSF, TGFβ	HS and heparin	(78–83)

Adapted with permission from Kang et al. (17). CS, chondroitin sulfate; CSPG, chondroitin sulfate proteoglycan; GAG, glycosaminoglycan; PG, proteoglycan.

Versican-Associated Factors

Versican-HA-affiliated factors.

Hyaluronan.

Hyaluronan (HA), a critical component of the ECM, is broadly expressed in tissues, including the lungs throughout development and maturation (56, 87). The HA molecule comprises alternating β-1,4-d-glucuronic acid and β-1,3-N-acetylglucosamine subunits and contributes to the viscoelasticity of tissues in both health and disease (88). HA is synthesized by a family of cell surface HA synthases (HAS1, HAS2, and HAS3) that extrude HA into the extracellular environment as a nonsulfated GAG polymer (89). The HAS isoforms display different HA synthesis kinetics and can produce various lengths of HA polymers contributing to the heterogeneity of HA molecular weight in the microenvironment. HAS isoforms are differentially expressed during tissue injury and repair and are also subject to posttranslational modifications that can further regulate their activity (90, 91). Enzymatic turnover of HA is facilitated by the activity of hyaluronidases, which also display tissue-specific expression and differential regulation during injury and repair (92).

HA is primarily secreted as high molecular weight HA (HMW-HA, ≥1,000 kDa); therefore, the conformation of the HA molecule can be highly variable depending on its interaction with water and the presence of other molecules that bind HA in the microenvironment. For example, HA exists in the form of a loose pericellular glycocalyx that facilitates locomotion and cell division (93). In contrast, it is also found as highly condensed extracellular cable-like structures that avidly bind leukocytes in other contexts (94). Thus, the interactions of HA with HA-binding molecules (also termed hyaladherins) are highly diverse, context-specific, and dependent on the size of the HA molecule (22). HMW-HA can be degraded into lower molecular weight fragments by hyaluronidases and reactive oxygen species, facilitating the generation of low molecular weight HA (LMW-HA) (95). Prior studies show that different molecular sizes are in part responsible for the diverse biological activities of HA (96–98). For example, HMW-HA suppresses inflammatory cell activation, inhibits scar formation, and suppresses inflammation, whereas LMW-HA promotes these effects (96, 99–102). However, understanding the dichotomous effects of HMW-HA and LMW-HA in inflammation is likely an oversimplification, making it a critical area of future investigation.

In addition to serving as an extracellular scaffold, HA has been shown to participate in cell signaling pathways through the engagement of cell surface receptors, including primary HA receptors such as CD44, RHAMM, LYVE-1, and pathogen-associated molecular pattern receptors such as TLRs, TLR2 and TLR4. HA binding to these receptors regulates a range of activities, including cell proliferation, migration, and inflammatory responses (89). Interestingly, HA molecules of different sizes can elicit diverse activities from engagement of the same receptor (103). One hypothesized mechanism for these differences in signal transduction is that HMW-HA can bind multiple receptors at once, clustering the activation signaling at the cell membrane. In contrast, LMW-HA is unable to aggregate the receptors in the same fashion, thereby activating different signaling pathways. In addition, LMW-HA is able to engage cell surface receptors from which HMW-HA is sterically excluded (56) Finally, the binding of HA to cell surface receptors can be potentiated through the interaction of HA with hyaladherins, which may enhance the affinity of the interaction and promote HA-cell signaling (104). Examples of these hyaladherins and their contributions to HA and versican-mediated signaling are discussed throughout this review.

The investigation of HA and its modifications in the lung during health and injury is a growing field, with significant advancements in recent years. Studies in both human and animal models have demonstrated that HA is present in limited amounts in healthy lung tissue and is confined primarily to perivascular and peribronchiolar spaces with minimal amounts present in the alveoli (105, 106). During lung injury and repair, HA increases in these areas and is associated with the influx of inflammatory cells and interactions with HA may further influence the inflammatory process (107–109). Elevated HA levels have been reported in patients with asthma (110, 111), chronic obstructive pulmonary disease (112), interstitial pulmonary fibrosis (113), and several occupational lung diseases (reviewed in Ref. 107). These findings are paralleled in animal models of lung injury, including bleomycin-induced lung fibrosis (114), antigen-induced asthma (106, 115), ozone-induced lung injury (116), and models of lung infection (24, 25, 117). A common feature of these studies is that, in general, HA levels correlate with the severity of lung disease and the degree of inflammation observed in the lungs, suggesting an essential role for HA in disease pathogenesis and progression. HA also plays a critical role during lung development and regeneration (103, 118–120). Indeed, remodeling of the ECM through deposition of a HA-enriched provisional matrix is an essential step in wound repair (121). For a more detailed discussion of HA’s role in lung health and disease, we recommend several excellent recent reviews on this topic (22, 96, 107, 109, 121–123). Our focus for the present review will center on the interactions of HA and versican in the lung.

A discussion of the biological effects of versican without considering its interactions with HA would be difficult because factors that increase versican expression generally increase HA expression leading to the increased accumulation of both HA and versican in tissues (21). Versican is known to bind HA through a site in its G1 domain (Fig. 3) (124, 125) and both molecules are often colocalized to regions of lung inflammation or tissue repair. Seminal studies by >Bensadoun and colleagues (32) reported that HA and versican were colocalized in three distinct types of fibrotic disease in humans, including acute respiratory distress syndrome (ARDS), bronchiolitis obliterans organizing pneumonia (BOOP), and idiopathic pulmonary fibrosis (IPF). In normal lungs, both HA and versican were observed in the subepithelial connective tissue and the media of blood vessels with minimal staining present in the alveoli. In each disease state, versican and HA staining were increased in the fibrotic foci and colocalized with staining for myofibroblasts. Interestingly, these versican-rich areas stained positively for procollagen I with minimal staining for mature collagen leading the investigators to speculate that these versican-enriched areas may influence the early repair processes in the lung during fibrotic disease (32). In later studies, the same authors reported similar findings of versican and HA-enriched regions in the connective tissue surrounding granulomas in a variety of granulomatous lung diseases leading them to the conclusion that a versican-rich provisional matrix is essential for lung remodeling regardless of the underlying inflammatory process (33).

Asthma is another chronic lung disease where versican and HA play a critical role in airway inflammation and responsiveness; increased airway deposition of versican and HA has been reported in subjects with fatal asthma and remodeled airways (126, 127). The postulated mechanism of the contribution of the subepithelial deposition of the versican-HA-enrich ECM was that these molecules could affect the tissue fluid balance through increased osmotic activity, thereby causing airway edema and compression, leading to airway obstruction. Subsequent studies demonstrated increased proteoglycan production from isolated lung fibroblasts that correlated with asthma severity (128). Using a cockroach antigen sensitization model of asthma in mice, members of our group demonstrated increased subepithelial accumulation of versican and HA and increased HAS expression that correlated with inflammatory changes observed in the airways of sensitized mice (30). The immunostaining of tissues from studies using the chronic cockroach antigen exposure protocol (115) demonstrated that increases in versican paralleled the time course and distribution of HA as well as inflammatory changes in the lungs of challenged mice (30). Furthermore, parallel studies performed in differentiated bronchial epithelial cells from children with asthma also displayed increased versican and HAS expression raising the possibility that the epithelium may contribute to the subepithelial deposition of these molecules as well (30).

Versican binds to HA through its G1 domain as described in the next section on hyaluronan and proteoglycan link protein 1. Studies using human lung fibroblasts (HLFs) treated with poly(I:C) demonstrated the increased formation of HA cable structures decorated with versican that displayed enhanced binding of monocytes (129). In separate studies, poly(I:C) treatment of HLFs led to the increased accumulation of versican, which correlated with increased accumulation of HA within the cell layer and enhanced binding of monocytes (130). Pretreatment of the HLFs with antibodies against versican inhibited the HA-dependent binding of monocytes suggesting a cooperative effect of versican and HA, leading to an increase in inflammation following treatment with poly(I:C). Subsequent studies in mice revealed an essential role for versican in lung inflammation in response to treatment with poly(I:C), further suggesting a proinflammatory role for versican in the lungs (26). In these studies, poly(I:C) administration significantly increased the accumulation of versican and HA, with the greatest increases occurring in the perivascular and peribronchial regions where positive staining for HA was colocalized with inflammatory cells. Similarly, treatment of versican-deficient (R26/Vcan^−/−) mice with poly(I:C) demonstrated that the global lack of versican protected the mice from the increased inflammatory response seen in wild-type mice. Interestingly, isolated fibroblasts from the R26/Vcan^−/− showed a significant decrease in leukocyte adhesion to HA cables and a decrease in the expression of select proinflammatory cytokines (26).

In summary, the increased expression and synthesis of versican and HA is observed in multiple types of pulmonary inflammation and injury in human subjects and mouse models. The evidence from these studies suggests that versican-HA interactions are essential. Whether and how versican-HA interactions play a role in pulmonary health and disease is not yet known. Further studies are required to delineate the nuances of versican-HA interactions and the potential for this interaction to play a vital role in developing select immunomodulatory microenvironments in the lungs.

Hyaluronan and proteoglycan link protein 1.

Hyaluronan and proteoglycan link protein 1 (HAPLN1), also called cartilage link protein 1 (CRTL1), is an HA-binding protein (hyaladherin) whose functional role is to stabilize the interactions between HA and the hyalectans such as versican and aggrecan (124, 131–133). The G1 domains of versican and aggrecan include two link module units (Fig. 3), which share 95% homology with HAPLN1 (40). The link module has been extensively reviewed (134–137). Link modules also are found in the HA-binding proteins, CD44, TSG6, LYVE1, stablin-1, and CAB61358 (135). Other members of the HAPLN family include HAPLN-2, -3 and -4, with HAPLN-2 and HAPLN4 expression restricted to the brain and nervous system (134).

Studies using HAPLN1-null mice found that most mice died soon after birth due to respiratory failure (138, 139). In addition to perinatal lethality, defects in cartilage development (chondrodysplasia), delayed bone formation, attenuated perineuronal nets, and cardiac malformations were observed (138–141). All these abnormal phenotypes are linked to an altered ECM in the HAPLN1-null mice, reinforcing the importance of HAPLN1 in modifying the ECM. Regarding HAPLN1 and the lungs, HAPLN1 gene expression was increased in mouse lungs at embryonic day 20.5 and in 2-day-old mice. In contrast, the amount of HAPLN1 mRNA in healthy adult mice was minimal (138). Perinatal death in mice lacking HAPLN1 was due to respiratory failure caused by the collapse of the cartilage-containing trachea and extrapulmonary bronchi resulting in collapsed alveoli (138, 139, 142). The work of Evanko et al. (40) shows that HAPLN1 stabilizes the binding of versican to HA synthesized by HLF, which may facilitate stabilizing cell adhesion. This work also showed that HAPLN1 increased ECM compaction and promoted myofibroblast formation suggesting a potential role for HAPLN1 in the fibrotic remodeling of the lungs.

Heavy chains and tumor necrosis factor-stimulated gene-6.

Heavy chains (HCs) and tumor necrosis factor-stimulated gene-6 (TSG-6) are two other HA-binding proteins that can modify the HA matrix. Inter-alpha-trypsin inhibitor (IαI) is a composite protein composed of one or two HCs covalently attached to the single CS side chain of the bikunin core protein. This protein-GAG-protein structure of IαI makes it a unique molecule. Only free bikunin is detected in urine, whereas, in blood and other tissues, bikunin is present either alone or with one or two HCs linked (143). Functions of IαI include hyaluronidase inhibitory activity, attenuation of complement activation, and binding invading pathogens (144–146). For detailed reviews on IαI, please see Refs. 22, 147, and 148. However, the most well-known function of IαI is as an HC donor to HA for the formation of the HC-HA matrix through a TSG-6-dependent process (149–153).

The individual components of IαI have different biological functions when they are present as separate entities versus the IαI complex. Bikunin is involved in various biological processes, including cell stress, apoptosis, serine protease inhibition, aging, and inhibition of cytokine release. HCs can inhibit complement and bind with other ECM components (154, 155). There are currently six HCs with differential expression in tissues during development and disease (147, 148). HC1-3 and -5 are synthesized as prepropeptides; the C-terminal is cleaved before they are covalently attached to the bikunin CS chain (148). HC4 does not contain the cleavage site found in HC1-3 and -5, suggesting it exists as a free HC form in circulation (148, 156). HC6 has only been identified at the gene level (147). HC1 can self-associate, providing a mechanism for HC-HA cross linking and matrix stabilization. HC1 can also interact with arginine-glycine-aspartate (RGD)-containing ligands such as fibronectin and vitronectin (155). The latter interactions of HCs suggest a more diverse role in the ECM beyond HA, where HCs could act as linkers between HA and RGD-containing proteins.

TSG-6 is an inflammation-associated secreted protein that has been implicated in critical and diverse tissue-protective and anti-inflammatory properties, e.g., mediating many of the immunomodulatory and beneficial activities of mesenchymal stem/stromal cells (137). These properties of TSG-6 are either direct, via binding of its link domain to HA and PGs, or indirect, via enzymatic modification of HA (157, 158). TSG-6 enzymatic activity catalyzes HC’s transfer and covalent binding from the IαI family of proteins to HA, forming HC-HA (137). Through this highly conserved activity, TSG-6 cross-links HA and creates a compacted HC-HA ECM (159).

The formation of the HC-HA matrix is implicated in many biological processes, including female fertility, leukocyte adhesion, and macrophage polarization (147, 149, 160–162). This structural change enhances the binding of HA to its receptors and enhances the formation of “HA cable” structures known to bind leukocytes with greater affinity (56, 104, 163–165). In addition, TSG-6 synergizes with the viral mimetic poly(I:C) and extends its catalytic function in the formation of HC-HA to the actual induction of the synthesis of HA matrices. TSG-6 increases the accumulation of HA in the cell-associated matrix, partially by preventing its dissolution from the cell-associated matrix into the conditioned medium, but primarily by inducing HA synthesis (166).

Published data suggest that IαI and TSG-6 play essential roles in lung inflammation and injury. Like versican, IαI has a contextual role in lung inflammation in that it can be anti- or proinflammatory. Studies using bikunin-deficient mice have demonstrated that the lack of bikunin/IαI results in increased airway hyperresponsiveness and pulmonary inflammation in murine asthma models, suggesting a protective role of IαI in lungs (161). IαI also binds vitronectin to promote bronchial epithelial repair after injury and can inhibit complement activation, reducing complement-dependent lung injury (145, 167). IαI is also able to decrease airway epithelial sodium channel activation, which contributes to the pathophysiology of cystic fibrosis (CF) (168). In many tissues, TSG-6 plays a vital role in inhibiting inflammation and promoting wound repair (137). However, evidence shows that TSG-6 plays a different role during lung disease and is associated with acute and chronic airway inflammation (107). TSG-6 expression in the lung has been explored in the setting of cigarette smoke using primary human bronchial epithelial cells, where TSG-6 expression was found to be increased in current smokers and induced by TNFα and IL-1β (169). In that study, TSG-6 expressed by bronchial epithelial cells was also found to remove HCs from bikunin, promoting bikunin’s anti-protease properties and suggesting an essential function in balancing pro- and anti-inflammatory signals in the airway.

In follow-up to their prior study, Lauer and colleagues (166) showed that adding exogenous recombinant TSG-6 to the culture medium of airway smooth muscle cells enhanced the addition of HCs from IαI to HA and increased leukocyte binding to HA cables in response to poly(I:C). The enhanced generation of the HC-HA matrix in response to added TSG-6 also resulted in thicker HA cables, increased accumulation of HA in the cell-associated matrix, and decreased HA shedding. Most studies of TSG-6 and HC-HA in the lung evaluate chronic airway inflammation such as asthma, with few studies evaluating their role in acute airway inflammation. For example, a study by Swaidani and colleagues using a mouse model of ovalbumin-induced allergic airway inflammation and mice lacking the ability to form HC-HA matrix (TSG-6^−/− mice) demonstrated decreased HA deposition, reduced eosinophilic airway inflammation, attenuated airway hyperresponsiveness (AHR), and the absence of an HC-HA ECM in the subepithelial space (170). Interestingly, no changes in TH2 cytokine profiles were found between wild-type and TSG-6^−/− mice suggesting a direct role of the ECM in the inflammatory changes reported. Studies in humans with asthma also demonstrate that HC-HA is present in the airway basement membrane and correlates with asthma severity (110). Leukocytes were embedded in the HC-HA-enriched ECM of individuals with asthma, suggesting that HC-HA is directly linked to airway inflammation (110).

Similarly, the lungs from patients with CF showed formation of an HC-HA matrix, which may also contribute to the chronic inflammation observed in CF lung disease (171). More recently, Stober et al. (172) demonstrated that AHR induced by ozone or LMW-HA depended on TSG-6 in mice. Furthermore, HC-HA and HC-HA derived from LMW-HA were necessary to mediate the proinflammatory effects in their model. The addition of exogenous TSG-6 led the authors to conclude that the rapid formation of LMW HC-HA was required to promote the downstream signaling leading to airway inflammation and enhanced AHR. Although studies have linked TSG-6 activity to airway inflammation in asthma and airway hyperreactivity models, the role of TSG-6 in acute respiratory infections has not been widely explored (110, 172). TSG-6 was upregulated in BAL fluid and whole lung homogenates of mice infected with the influenza A virus. Increased TSG-6 corresponded to the presence of HC-HA and correlated with the degree of inflammation and impaired lung function in influenza-infected mice (173). Treatment with intranasal hyaluronidase decreased HC-HA accumulation, reduced the overall inflammatory response, and improved lung function in influenza A-infected mice (173).

The complex interplay of IαI, TSG-6, and HC-HA in lung inflammation is not well understood. HC-HA generated either in vivo or in vitro in cell cultures enhances leukocyte adhesion (104, 166, 174). However, when a matrix is generated using isolated HA, TSG-6, and IαI, the formation of HC-HA did not show augmented adhesion of CD44⁺ cells compared with HA alone (175). These data suggest that other associated molecular components in the HC-HA-enriched matrix may contribute to leukocyte adhesion. Interactions with these other matrix molecules may ultimately determine a pro- or anti-inflammatory environment. Currently, we know very little about the composition of the HC-HA matrix, such as which HCs are present and what other molecules are present in this matrix. Studies by Abbadi et al. (176) reveal that tracheal epithelial cells form an HC3-HA matrix adhesive for leukocytes under normal conditions. Beyond this, the cellular sources of the increased IαI present during lung disease are not known, but studies show that the matrix generated by this complex plays a role in determining the phenotype of immune cells (162, 177). Future studies will need to connect the biological function of the HC-HA matrix in the lungs to its composition during development, tissue homeostasis, and lung disease.

Although TSG-6 and versican are both differentially regulated during inflammation, few studies have directly evaluated the interaction of these two important molecules. In a study by Kuznetsova et al. (178), versican G1 bound to immobilized link module of TSG-6, suggesting that versican and TSG-6 may interact directly through their link modules. Aside from this, most studies have focused on TSG-6 and versican interactions with HA as part of the larger microenvironment during inflammation (21). Indeed, both molecules have been shown to be important components of HA cables and blocking of either molecule impacts the ability of the HA matrix to interact with leukocytes in a variety of cellular models (26, 94, 176, 179–181).

Only one study has reported a direct interaction between versican and IαI. In this study, a complex consisting of the versican G1 domain, HCs, and HA was detected in the edematous granulation tissues of human pressure ulcers and in inflamed tissues in a mouse model of moist would healing (182). Furthermore, the versican G1 fragment-HC-HA complex was associated with macrophage accumulation, suggesting an important role in modulating inflammatory changes in this model. With many studies demonstrating the interactions of TSG-6-HA, versican-HA, and HC-HA in inflammatory conditions, it would be hard not to speculate that TSG-6, versican, HCs, and HA form a macromolecular complex under certain inflammatory conditions. Immunohistochemistry staining of diabetic pancreas tissues showed increased staining for HA, IαI, and versican compared with control and all three molecules and TSG-6 were observed in the inflamed regions of the pancreas (183). This observation suggests the formation of an HC-HA-TSG-6-versican-enriched matrix promotes tissue inflammation.

Bell et al. (173) reported increased production of IαI in BAL fluid and formation of HC-HA complex in lungs in influenza-infected mice. In a study by our group, Brune et al. (37) reported increased accumulation of versican in lungs after influenza infection in C57BL/6 mice. Interestingly, these two manuscripts showed similar increases in IαI and versican in lungs at 8 and 9 dpi. The similar temporal expression of versican and IαI could suggest that a versican-HC-HA ECM may be formed at 8–9 dpi with influenza virus. It is interesting that TSG-6, versican, and IαI are all reported to have both pro- and anti-inflammatory roles in lungs and can all interact with HA to modify HA matrix in disease. Future studies determining if TSG-6, HCs, and versican operate in a cooperative or interdependent fashion to alter the biological activity of the ECM could offer further insight into how the versican interactome provides fine control over the development of microenvironments during the clinical course of a disease.

Versican-associated adhesion molecules.

CD44.

CD44 is a type I transmembrane cell surface glycoprotein consisting of a single polypeptide chain expressed on embryonic stem cells and other types of cells, including connective tissues and bone marrow. CD44 expression is also upregulated in subpopulations of cancer cells and is recognized as a molecular marker for cancer stem cells. CD44 has at least 10 variable exons encoding a segment of the extracellular domain, termed exons v1–v10 and has cell-specific glycosylation (184). CD44 is a major cell surface receptor for GAGs and HA, and most of CD44’s function can be attributed to its ability to bind and internalize HA (185, 186). Functions of CD44 include mediating cell-cell and cell-ECM interactions (185), lymphocyte homing (187), and tumor metastasis. See reviews on CD44 structure, functions, and its involvement in cancer (188–191).

During development, the amount of HA in the lungs decreases as the lungs mature. A study showed that this reduction is due to the internalization of HA by CD44-positive macrophages in lungs, and that blocking antibodies against CD44 increased the concentration of HA while decreasing the number of HA-containing macrophages in the lungs (192). CD44 has been shown to play a role in resolving lung inflammations caused by bleomycin. CD44-deficient mice showed impaired clearance of apoptotic neutrophils, persistent accumulation of HA fragments, and impaired activation of transforming growth factor-β1 (TGFβ1) (193). Restoring CD44 on hematopoietic cells can reverse these effects. Loss of CD44 also disrupts alveolar macrophage lipid homeostasis and exacerbates oxidized lipid-induced lung inflammation (194). Treating mice with blocking antibody to CD44 reduced lung fibrosis, suggesting CD44, together with HAS2, regulate matrix-invading fibroblast phenotypes (195). Studies have implicated CD44 in lung cancer tumorigenesis (196–198) making it a therapeutic target for cancer treatment (199). See detailed reviews on CD44 in cancer listed above.

CD44 and versican have been reported to interact in different biological systems. The interaction is through the carbohydrate-binding domains of CD44 and the CS chains on versican and this interaction is independent of the sulfation pattern (200, 201). However, a study using melanoma cells overexpressing the V3 isoform of versican showed that V3 could also interact with CD44, providing evidence that CD44 interacts independently of the CS chains on versican (202). In development, the loss of the versican-HA matrix can lead to cellular senescence via CD44 (203).

Versican and CD44 interaction is also implicated in cancer development (204, 205). Downregulation of versican resulted in a decrease in CD44 expression and cleaved form of CD44 in lymphoma (206), and upregulation of versican resulted in increased expression of CD44 in cancer-associated fibroblasts (207). HA is implicated in most studies on versican and CD44 interactions as the third player, forming a versican-HA-CD44 complex. Overexpression of versican by either stromal or tumor cells forms a pericellular matrix enriched in versican, HA, and CD44, which increases tumor cell proliferation (203). Inhibition of CD44 inhibited the formation of this pericellular matrix and motility and invasion of ovarian cancer cells (208). The versican-HA-CD44 complex has become a target for new anticancer therapies based on these findings.

L-selectin.

L-selectin (CD62L) is a type-I transmembrane glycoprotein and cell adhesion molecule constitutively on most circulating leukocytes. L-selectin mediates the initial steps of the adhesion cascade, the capture and rolling of leukocytes on endothelial cells. This event enables leukocytes to migrate out of the vasculature into surrounding tissues during inflammation and immune surveillance. For detailed reviews, please see Refs. 209–212. The major enzyme responsible for the shedding of the L-selectin ectodomain is a disintegrin and metalloproteinase (ADAM)17 (213, 214). The soluble circulating L-selectin from ectodomain shedding can be used as a plasma biomarker for leukocyte activity during inflammation (215–217).

L-selectin-deficient mice are susceptible to pulmonary infection compared with wild-type controls, indicating that L-selectin plays a role in pulmonary host defenses (218, 219). In several animal and lung disease models, L-selectin was demonstrated to effect multiple aspects of neutrophil trafficking in lungs, particularly with regard to neutrophil margination, sequestration, and emigration (219–223). Furthermore, L-selectin cleavage from the neutrophil surface by ADAM17 is involved in neutrophil recruitment and enhanced neutrophil effector functions, resulting in enhanced bacterial clearance from lungs (218). An interesting observation using L-selectin-deficient mice in an asthmatic model was that the airway hyperresponsiveness was abrogated in L-selectin-deficient mice even though there was no difference in the acute airway lung inflammation between the deficient and the wild-type controls, indicating a role for L-selectin in the development of airway hyperresponsiveness but not allergic inflammation in this asthma model (224).

Binding between L-selectin and versican derived from the renal adenocarcinoma cell line ACHN was mediated by the carbohydrate-binding domain of L-selectin and CS chains for versican (200, 201, 225). This interaction is sulfation dependent, with the over-sulfated CS/DS chains recognized by L-selectin, indicating that these chains are important in L-selectin-mediated cellular responses (200). Although studies reporting interactions between L-selectin and versican have only been done in the kidney, with increased versican expression in the lungs under inflammatory conditions, we can speculate that L-selectin can bind to versican in the lungs to promote leukocyte infiltration during inflammation.

P-selectin glycoprotein ligand-1.

P-selectin glycoprotein ligand-1 (PSGL-1) is a glycoprotein expressed on the cell surface of all leukocytes, including platelets, neutrophils, monocytes, and lymphocytes. It is constitutively expressed on some leukocytes (e.g., neutrophils), while in others (e.g., T cells), it is only expressed after activation. As such, PSGL-1 is a vital regulator of homeostasis and innate and adaptive immunity. PSGL-1 was first identified for its role as an adhesion molecule that mediates the tethering of leukocytes to endothelial selectins, thereby mediating migration and adhesion of leukocytes both constitutively and in response to inflammatory stimuli (226). PSGL-1 also impacts leukocyte function via nonselectin interactions; as a cell signaling transmembrane receptor, ligation of PSGL-1 influences essential cellular functions, including rolling of neutrophils and cytokine secretion by macrophages and dendritic cells (227). More recently, PSGL-1 has been shown to regulate multiple aspects of T-cell function (including T-cell proliferation, exhaustion, and survival) and is recognized as a critical immune checkpoint regulator of T-cell function (227, 228).

In the pulmonary setting, the interactions between PSGL-1 on lung type 2 innate lymphoid cells (ILCs) and P-selectin on platelets have consequences for ILC-mediated secretion of T_H2 cytokines in allergic inflammation and asthma (229). In chronic obstructive pulmonary disease (COPD), elevated expression of PSGL-1 is associated with the dysregulated recruitment of circulating leukocytes into the lungs, which is a characteristic of the disease (230). PSGL-1 also has been shown to mediate clearance of Streptococcus pneumoniae by neutrophils, suggesting a protective role in this disease (231).

Versican is the only ECM molecule known to bind PSGL-1 (232). Although this interaction has not been extensively studied, the findings are intriguing. The interaction occurs between a motif on PSGL-1, which is distinct from that involved in selectin-binding, and the C-terminal G3 domain of versican, which has a selectin-like structure. This interaction results in multimerization of the versican G3 domain and consequent aggregation of the leukocytes expressing PSGL-1, suggesting that versican might have essential roles in modulating leukocyte function. Multiple possible outcomes of PSGL-1 binding to versican are hypothesized depending on the tissue site of interaction, including retention of leukocytes in the bloodstream limiting extravasation into tissues, retention of leukocytes at sites of inflammation, or acceleration of thrombus formation. An additional consideration is that PSGL-1-expressing leukocytes produce MMPs, potentially amplifying the consequences of versican degradation, which is discussed in depth in the Versican-Modifiying Enzymes.

The interactions and consequences of versican binding to this critical regulator of immune cell function warrant further study. To date, nothing is known regarding the role of versican-PSGL-1 interactions in the lungs during development, homeostasis, or inflammation.

Beta-1 integrins.

Beta-1 (β1) integrins are cell-surface receptors that mediate cell-cell and cell-matrix interactions. In humans, integrin β1 has 4 variants of subunits (β1A–D) that differ in their cytoplasmic domains and can dimerize with 10 different α subunits (α1–α9, αv) (233). Integrin β1 plays a vital role in many biological processes including development, cell differentiation, and proliferation. Homozygous β1 integrin null embryos are embryonic lethal (234). Integrin β1 acts as transmembrane linkers, sending and receiving signals between the ECM and the actin cytoskeleton. In ECM, β1 integrin binds various ECM components, including collagens, laminins, fibronectin, and vitronectin (233). The cytoplasmic domain of β1 integrin can interact with not only intracellular anchor proteins such as talin, filamin, and α-actinin but also signaling molecules such as focal adhesion kinase and integrin-linked kinase (233). Intracellular signals can either enhance or inhibit the ability of integrin binding to their extracellular ligands, therefore allowing bidirectional transfer of information through the cell membrane. For detailed reviews, see Refs. 233 and 235.

β1 integrin plays a critical role in regulating lung development and homeostasis. Studies using mice that can selectively delete β1 integrin in lung epithelium have showed that epithelial β1 integrin during both early and late lung development affect airway branching morphogenesis, epithelial cell differentiation, alveolar septation, and regulation of alveolar homeostasis (236). Deletion of β1 integrin in adult mouse lung after completion of development in type 2 alveolar epithelial cells showed that the deficient mice exhibited COPD-like pathology characterized by emphysema, lymphoid aggregates, and increased macrophage infiltration (237). These studies showed that β1 integrin suppresses inflammatory signaling to maintain homeostasis in lung epithelial cells. Furthermore, β1 integrin has also showed to mediate eosinophil migration and activation in allergic inflammation (238). β1 integrin is also involved in tumor progression in lungs, specifically in assisting cancer cell adhesion to ECM (239). Other functions include signal transduction, tumorigenicity, and growth regulation, for detailed reviews see Refs. 239 and 240.

The C-terminal domain of versican, G3 interacts with β1-integrin. In studies using G3 mutant constructs, the region of interaction was shown to encompass the carbohydrate recognition and complement-binding domains, but not the EGF-like domains (241). In astrocytoma cells, this interaction was found to activate focal adhesion kinase, enhance β1-integrin and fibronectin expression, promote cell adhesion, and protect cells from free radical-induced cell apoptosis (241, 242). These studies raise several interesting questions which have not yet been addressed. The authors speculate that the G3 mutant construct increased the binding activity of integrins to fibronectin by forming complexes containing these three molecules. They also propose that the G1 and G3 domains of versican have different and counteracting roles in cell adhesion, with G1-hyaluronan interactions being antiadhesive but G3-integrin interactions being proadhesive so that shedding of different domains of versican might modulate cell adhesion and survival (241). In addition, the β1 integrin binding region of versican’s C-terminal domain does not contain the RGD consensus-binding motif for integrins, indicating a new and, as yet undefined mechanism of integrin-binding (241). It is also notable that there is currently no information regarding the potential role of versican interactions with β1-integrin in the context of lung development or inflammation.

Versican-associated secreted factors.

The glycosaminoglycan interactome.

Glycosaminoglycans (GAGs) have an important role in the binding of a select set of molecules to versican as the CS side chains interact with a wide array of secreted factors, including growth factors, morphogens, chemokines, ECM proteins, and their bioactive fragments, receptors, lipoproteins, and pathogens (243). This extensive collection of binding partners mediates diverse GAG functions from ECM assembly to immunomodulation in various contexts, including embryonic development, homeostasis, angiogenesis, cancer, neurodegenerative diseases, and infection (243). The GAG interactome detailing GAG-protein interaction and interaction networks has recently been reviewed elsewhere (243). Our discussion will focus on the interactions of secreted factors known to bind with the chondroitin sulfate side chains of versican.

The binding of secreted factors to versican is multifaceted due to the alternative splicing of GAG domains across versican isoforms. Additional versatility is conferred to GAG interactions with secreted factors through various processes that alter GAG side-chain structure. These include hyperelongation of GAG chains, epimerization of glucuronic acid to iduronic acid resulting in the conversion of chondroitin sulfate to dermatan sulfate, and changes in the sulfation site patterns that yields five possible chondroitin sulfate subtypes (42, 244). GAG side-chain structure determines their abilities to impact the availability of secreted factors for receptors, control the development of secreted factor gradients, and sequester factors from proteolytic degradation (62, 245–248).

Chemokines.

The work of Hirose et al. (57) shows that versican binds select chemokines and regulates their chemokine function. This work shows that versican binds to CCL2/MCP-1, CCL8/MCP-2, CXCL10/IP-10, CXCL4/PF4, CXCL12/SDF-1β, CCL5/RANTES, CCL20/LARC, and CCL21/SLC but not to CXCL5/ENA-78, CXCL1/GROα, CXCL8/IL-8, CCL7/MCP-3, CCL3/MIP-1α, CCL4/MIP1β, CCL18/PARC, CCL17/TARC, or CCL19/ELC (57). Furthermore, they examined the ability of versican, CS, or heparan sulfate (HS) to induce integrin activation by secondary lymphoid tissue chemokine (CCL21/SLC) and calcium mobilization in lymphoid cells expressing a receptor for CCL21/SLC, CC chemokine receptor 7 (57). This study showed that HS supported integrin-dependent binding and calcium mobilization by CCL21/SLC, whereas versican or CS B inhibited these cellular responses. This suggests that CCL21 will be proinflammatory in microenvironments enriched with HS, but anti-inflammatory in microenvironments enriched in versican or CS-B (57).

GAGs play an essential role in immobilizing chemokine depots for interaction with leukocytes. The model by which chemokines are presented to leukocytes by GAGs has recently been refined by evidence that they are presented in solution while sequestered within the glycocalyx, instead of being exclusively GAG-bound (245). Work by Tanino et al. (246) showed that dermatan sulfate (i.e., chondroitin sulfate-B) inhibited the binding of mouse CXCL1/KC to immobilized heparin, suggesting that in mice, this key neutrophil chemotactic factor could bind to the versican side chain under certain conditions. In addition, they investigated how the kinetics of chemokine-GAG interactions contribute to chemokine functions and found that in the lungs, rapid association and disassociation of KC from heparin contributed to higher plasma KC levels and more effective neutrophil recruitment (246). GAG chain sulfation pattern has also been shown to impact chemokine binding to versican with over-sulfated chains demonstrating inhibited binding of CCL21/SLC, CXCL-10/IP-10, CXCL4/PF-4, and CXCL12/SCF-1B compared with control versican (200). The potential inhibition of KC binding to heparin by CS-B and the inhibition of chemokine binding due to altered sulfation are possible mechanisms by which versican may impact the kinetics of chemokine-GAG interactions and therefore chemokine activity in the lungs.

Cytokines.

Interactions between cytokines and heparin and heparin sulfate GAGs have been extensively described, while relatively fewer interactions between cytokines and chondroitin sulfate GAGs or versican have been elucidated (243). One relatively well-characterized interaction is between chondroitin sulfate and IFN gamma (IFN-γ), an immunoregulatory cytokine produced by activated T lymphocytes and NK cells (66, 249). The work of Camejo et al. (249) shows that IFN-γ binds to CSPGs secreted by human arterial smooth muscle cells. They demonstrated that ECM-bound IFN-γ was functionally active, associated closely with CD44, and induced higher expression of MHC II antigen in hASMCs compared with soluble IFN-γ suggesting that CSPGs such as versican could play a critical role in immobilizing IFN-γ for induction of immunomodulatory functions (67, 249). In addition, soluble GAGs likely compete with IFN-γ receptors for IFN-γ and have been shown to inhibit IFN-γ binding to membrane-bound receptors (66). These findings suggest that the assembly and turnover of a CSPG-enriched ECM and/or versican-enriched ECM may regulate IFN-γ availability and functions.

Versican-associated cell surface receptors.

Innate immune receptors.

The innate immune system is the first line of defense against microbial pathogens in the lungs. Genetically encoded pattern recognition receptors (PRRs), present in both immune and nonimmune cells, recognize a set of microorganism-derived pathogen-associated molecular patterns (PAMPs), such as lipopolysaccharide (LPS). The activation of PRRs through their interaction with PAMPs initiates an inflammatory cascade consisting of the synthesis of inflammatory mediators, recruitment of leukocytes into the site of infection, and the transition to adaptive immunity (3, 4, 250). In the case of tissue injury and cancer, PRRs recognize host-derived endogenous factors, the damage/danger-associated molecular patterns (DAMPs), which initiate the innate immune response and the development of sterile inflammation (203, 251). Proteoglycans, including versican, are DAMPs that activate the innate immune system and have been extensively reviewed (203, 252).

The activation of PRRs by PAMPs often involves the engagement of various coreceptors, such as the recognition of LPS by the TLR4/MD2/CD14 complex, which promotes specific intracellular signaling cascades and secretion of proinflammatory mediators (3, 253–255). In contrast, the recognition of the DAMP, HA, occurs through the activation of TLR4/MD2/CD44, resulting in sterile inflammation (256). The activation of TLR4/MD2/CD14 and TLR4/MD2/CD44 by LPS and HA, respectively, results in the induction of entirely different sets of genes (256). The importance of PRRs and the innate immune response in the lungs has been extensively reviewed (2–4, 22).

The first study showing the interaction of versican with a PRR is by Kim et al. (34), showing that Lewis lung carcinoma cells secreted versican that activated macrophages through the coreceptor complex of TLR2/TLR6/CD14. This work also showed that the activation of TLR2/TLR6 complexes increased TNFα synthesis, which enhanced metastatic growth of Lewis lung carcinoma cells (34). Work by Tang et al. (86) shows that versican binds TLR2 and differentiates dendritic cells to an anti-inflammatory/immunosuppressive phenotype. The activation of TLR2 by versican and polarization of dendritic cells to an immunosuppressive phenotype is a mechanism by which tumor cells evade the immune system (86, 257). Secretion of versican by gliomas, the most common tumor of the central nervous system, activates TLR2 signaling in microglia/brain macrophages, promoting tumor expansion (258). Studies performed in vitro using ovarian tumor cells and macrophages provided similar results; tumor-derived versican was found to activate TLR2-dependent signaling in macrophages, leading to secretion of the tumor growth factor, hCAP18/LL-37, and subsequent promotion of ovarian tumor growth (259).

The proteolytic degradation of versican results in the generation of fragments that also activate PRRs. Versikine, the 70 kDa N-terminal fragment of versican, activates macrophages to increase type I IFN-stimulated gene signatures and production of IL-6, IL-1β, IL-12p40, and CCL2 (85). The findings of a proinflammatory phenotype stimulated by versikine contrast with the anti-inflammatory/immunosuppressive phenotype developed in response to intact versican (257, 260). Versikine not only stimulated the synthesis of select cytokines but also induced the expression of versican, suggesting a positive autoregulatory loop (85). The observation that versikine induces a type I IFN signature in myeloma cells is interesting because this signaling pathway increases versican expression in macrophages and stromal cells; these findings suggest that versikine could induce versican expression in myeloma cells through a type I IFN-dependent mechanism (24, 37).

Epidermal growth factor receptor.

The epidermal growth factor receptor (EGFR) is a member of the family of ERBB receptors that includes EGFR/ERBB1, ERBB2, ERBB3, and ERBB4 (261). All ERBB receptor family members have an extracellular ligand-binding region, a single membrane-spanning region, and a cytoplasmic tyrosine-kinase-containing domain and are found in tissues of epithelial, mesenchymal, and neuronal origin (261). EGFR plays a role in the development of numerous human cancers, including lung cancer, and is, therefore, a target for EGFR tyrosine kinase inhibitor therapies (261, 262).

EGFR plays a role in development, tissue homeostasis, and chronic lung disease in the lungs. A study in neonatal mice shows that the lack of EGFR in the developing lungs results in changes consistent with neonatal respiratory distress syndrome (263). Lesions in the EGFR-deficient mice included decreased staining for surfactant proteins, increased thickening, and cellularity in the alveolar septa and collapsed alveoli. In mouse models of asthma using chronic exposure of adult mice to house dust mite allergen, activation of EGFR in the airway epithelium resulted in the development of airway hypersensitivity and remodeling (264).

The mechanistic studies of Zhangyuan and colleagues (84) show that the EGF-like motif of versican V1 (Fig. 3) is required to promote proliferation, invasion, and metastasis in hepatocellular carcinoma by the EGFR-PI3K-AKT axis. This work provides strong evidence that the binding of the G3 domain of versican to the EGFR activates this receptor’s signaling pathway providing a novel role for versican in the progression of hepatocellular carcinoma. In contrast, the work by Hernández et al. shows that the overexpression of the versican isoform, V3, in MeWo human melanoma cells decreases in vitro and in vivo tumor growth. Reduced tumor growth in MeWo cells overexpressing the V3 isoform of versican probably occurred through the inhibition of epithelial growth factor binding to CD44 and the EGFR, altering signaling pathways that regulate cell proliferation and migration (202). In a separate study, the correlation between versican accumulation and its association with human EGFR1, 2, and 3 (EGFR, HER-2, and HER-3) and CD44 was evaluated using immunohistochemistry and tissues obtained from various stages of canine mammary gland tumors (205). This work suggested but did not prove a relationship between versican and tumor invasiveness. Although there is strong evidence suggesting that versican-EGFR interactions play a role in cancer progression, it is unknown whether this interaction plays a role in lung development or other types of lung disease such as asthma. Therefore, future work will need to determine whether the binding of the EGF-like motif in the G3 domain of versican to EGFR plays a role in lung health and disease.

Versican Core Matrisome

Versican binds to several ECM proteins that are part of the core matrisome. The seven ECM proteins that bind to versican include tenascin, thrombospondin-1(TSP-1), fibulin-1, fibulin-2, fibrillin-1, type I collagen, and fibronectin. Four of these proteins, tenascin, TSP-1, fibulin-1, and fibulin-2, are matricellular proteins (MCP). The other three proteins, fibrillin-1, type I collagen, and fibronectin, are considered structural proteins of the ECM.

Matricellular proteins.

Matricellular proteins have many of the same characteristics as versican. For example, MCPs are expressed and accumulate in tissues during embryonic development; MCPs decrease postnatally in healthy tissues and increase in response to tissue inflammation and injury; and MCPs have no structural role in tissues, but they do bind to structural ECM proteins such as collagen (265–274). Finally, in his commentary on the function of matricellular proteins, Paul Bornstein emphasizes that the sometimes diverse and contextual functions of matricellular proteins could be explained by their numerous molecular interactions (265). Based on these various interactions, he postulated that “matricellular proteins function as integrators of complex extracellular information imparted by extracellular protein motifs and cell surface receptors (265, 275).” What was not described in the commentary by Paul Bornstein in 1995 was the potential for the binding of versican to matricellular proteins to increase their ability to act together as integrators of extracellular information in lung health and disease. The matricellular proteins TNC, TSP-1, fibulin-1, and fibulin-2 play an essential role in controlling lung function in health and disease (276).

Tenascin-R and tenascin-C.

Tenascin-R (TN-R) and Tenascin (TN-C) are glycoproteins with complex modular structures that enable diverse molecular interactions with cell surface adhesion molecules (e.g., contactins, integrins), other ECM molecules (e.g., fibronectin, versican, aggrecan, brevican, and neurocan), growth factors (e.g., PDGF, FGF, and TGF-β), and receptors (e.g., EGF-R). Through these interactions, the tenascins are often associated with inhibition of cell spreading and formation of focal adhesions, and an increase in motility, invasive behavior, and proliferation (277, 278). These roles are reflected in the tissue expression patterns of TN-C and TN-R. TN-C is widely expressed in the developing embryo but is localized to sites of high cell turnover, high tensile stress, plasticity, and tissue remodeling in the adult. TN-R expression is restricted to the central and peripheral nervous system; its most important function is in the normal development of perineuronal nets, which are necessary for the development of visual cortical plasticity (279, 280).

TN-C is highly expressed in the lungs during development and is important for branching morphogenesis and alveolarization (281). There is little TN-C in the healthy adult lung, but it is highly reexpressed in many chronic lung diseases (e.g., COPD, IPF, RDS, and asthma). Although TN-C deficiency during development is not fatal, adult mice lacking TN-C show persistent lung function impairment under basal conditions associated with decreased smooth muscle cell numbers and smooth muscle actin content (282). On the upside, mice lacking TN-C have reduced lung disease phenotype during diseases characterized by smooth muscle hypertrophy, including acute lung injury, selective epithelial lung injury, or ovalbumin-induced asthma (282).

Both TN-R and TN-C interact with the C-terminal lectin domain of versican (and other lecticans), enabling the formation of macromolecular complexes with HA (283, 284). This interaction between TN-R, versican, and HA is essential for forming perineuronal nets. This organization is thought to impact diverse processes, including neural cell adhesion and migration, axon pathfinding, synaptogenesis and plasticity; growth factor and cytokine action, neuronal survival, structural organization of the ECM, and restricting access to the highly specialized microenvironment of perineuronal nets. Mice deficient in TN-R have abnormal aggregation of perineuronal CSPGs and exhibit severe behavioral disorders (42). Expression of TN-C and versican appear to be similarly upregulated in several diseases, including fibrotic lung diseases (285, 286). However, little is known about their mechanisms of coregulation or how their interaction contributes to lung health or pathology.

Thrombospondin-1.

Thrombospondin-1 (TSP-1) is a homotrimeric glycoprotein with multiple functional domains that enable intracellular and extracellular interactions with various protein and nonprotein ligands, including other matricellular proteins, structural components of the ECM, cell receptors, growth factors, and proteases (275). It is known that TSP-1 is capable of binary interactions with these ligands and multimolecular interactions that depend on the local composition of TSP-1 network components and the specific biological context. Thus, TSP-1 has critical and complex roles in embryogenesis, hemostasis, angiogenesis, inflammation, and tissue remodeling.

Studies of TSP-1 in mouse models of lung injury exemplify some of this complexity. TSP-1 was found to regulate the production of IL-10 by macrophages, and TSP-1-deficient mice were shown to be prone to LPS-induced lung injury due to defective macrophage IL-10 production and consequent persistent neutrophilic lung inflammation (287). TSP-1 also was found to inhibit the protease LasB, one of the virulence factors secreted by Pseudomonas aeruginosa; and TSP-1-deficient mice showed reduced survival, impaired host defense, and increased lung permeability with exaggerated neutrophil activation in response to acute Pseudomonas aeruginosa infection (288). However, TSP-1 is also a potent neutrophil elastase inhibitor; TSP-1-deficient mice showed increased neutrophil elastase activity, enhanced bacterial clearance from the lungs, and increased survival in response to Klebsiella pneumoniae infection (289). Thus, TSP-1 can be anti-inflammatory or proinflammatory, depending on the context.

TSP-1 binds to the link module motif in TSG-6, aggrecan, and versican (178, 290). The relationship between TSP-1 and versican has been examined in a few studies. Both were coordinately upregulated in the ECM of vascular smooth muscle cells during a TLR3-induced inflammatory response and organized into fibrillary structures containing elastin. The molecular events responsible for matrix assembly in this scenario are unknown, but versican may act as a bridging molecule that mediates TSP-1 binding to other ECM components, such as fibrillin-1, present in microfibrils (178). Both were coordinately downregulated in the left ventricles of mice in a model of cardiac aging (291). Versican and TSP-1 were among 18 genes related to the ECM, adhesion, and stem cell signaling that showed increased expression in the lungs of mice subjected to spaceflight stress (292). Much remains to be understood regarding the interactions between versican and TSP-1 in inflammatory disorders of the lungs.

Fibulin-1 and fibulin-2.

Fibulin-1 (Fbln-1) and -2 (Fbln-2) are among the larger members (∼90 kDa) of the fibulin family of glycoproteins, have structural similarities to fibrillin-1, and are secreted by various cells. Through their interactions with fibrillin-1 and TGF-β, fibulin-1 and -2 have dual roles as structural proteins and as matricellular proteins (273, 293). As structural proteins, these fibulins link FBN-1 with elastin and several other molecules, including laminin, fibronectin, fibrinogen, and PG; in this way, fibulin-1 and -2 are essential stabilizing components of the complex ECM. As matricellular proteins, these fibulins modulate TGF-β signaling and are, in turn, modulated by TGF-β signaling. Fibulin-1 is an essential binding partner and regulator of ADAMTS-1-mediated proteolysis, indicating an important role in ECM turnover (294). In these ways, fibulin-1 and -2 influence cellular processes critical to development, homeostasis, and disease.

In the pulmonary context in humans, fibulin-1 levels are elevated in circulation and in BAL fluid of patients with asthma or IPF compared with healthy individuals (295, 296). In cultured lung fibroblasts, fibulin-1 was found to stimulate cell viability and proliferation, associated with decreased secretion of TGF-β; fibulin-1 also was found to stimulate deposition of fibronectin and perlecan, which are thought to serve as growth factor depots, into the ECM of lung fibroblasts from patients with chronic obstructive pulmonary disease (COPD) or IPF (297). In the cigarette smoke mouse model of the chronic inflammation associated with COPD, mice lacking fibulin-1 have decreased airway inflammation, lung remodeling, and improved lung function compared with control mice (298). For these reasons, it is believed that fibulin-1 contributes to lung fibrosis and pulmonary tissue remodeling in chronic diseases. Whether fibulin-2 has a potential role in lung inflammation has not been extensively studied. However, fibulin-2 gene expression is elevated in the lungs of mice exposed to hyperoxic conditions (299).

The interactions of versican with fibulin-1 and -2 involve the lectin repeat structure of the C-terminal G3 domain on versican with the core protein of the fibulins. Fibulin-1 is strongly expressed in tissues where versican is expressed in developing and normal adult tissues (284). Both fibulin-1 and -2 are prominent components of endocardial cushion tissue in the developing heart and are associated with versican in HA-rich matrices (300). In addition to intact versican, ADAMTS-mediated proteolysis of versican is integral to the formation and differentiation of endocardial cushion mesenchyme (301), and fibulin-1 is essential for this proteolytic process (302). Fibulin-1 deficiency in mice causes a significant reduction in ventricular levels of the ADAMTS-mediated versican cleavage product, leading to suppression of cardiomyocyte proliferation, which is necessary for normal development (302). Fibulin-2 expression is coregulated, and the protein is colocalized with versican in the HA-rich matrix of mammary glands during puberty and early pregnancy. These findings are associated with the increased adhesiveness of mouse mammary epithelial cells to local matrix components, suggesting that fibulin-2 and versican alter the cell-matrix interaction to allow normal mammary ductal outgrowth and development (303). Fibulin-2 and versican also are coexpressed in the HA-rich matrix of atherosclerotic lesions; this interaction is thought to facilitate smooth muscle cell migration in response to the injury phase of vessel wall repair (304).

These findings suggest that the interactions between fibulin-1 and -2 with versican and their binding partners, including ADAMTS-1 and HA, are important in tissue remodeling. Whether these interactions are altered or have a role in lung inflammation has not yet been described.

Structural proteins of the core matrisome.

Fibrillin-1.

Fibrillin-1 (FBN-1) is a large extracellular glycoprotein (∼350 kDa) secreted mainly by fibroblasts, smooth muscle cells, and endothelial cells. It is the predominant component of the microfibrils found in the complex ECM of connective tissues. Its major direct binding partners include growth factors, tropoelastin, and integrins. Thus, in addition to contributing to tissue integrity and architecture, FBN-1 regulates growth factor signaling, cell behavior, and immune responses (305). The growth factor, TGF-β is a key regulator of microfibril assembly (306); in turn, microfibrils and the complex ECM have a major role in regulating the bioavailability and activity of TGF-β (307). The interactions between microfibrils and TGF-β are critical, as evidenced by a variety of connective tissue disorders, most notably Marfan syndrome, caused by mutations in the FBN-1 gene, which disrupt its ability to bind TGF-β. As a building block of the complex ECM, FBN-1 also has an important role in elastogenesis, as microfibrils provide a scaffold for elastic fiber assembly; thus, the microfibril and elastin networks are tightly linked (308).

Normal lung function depends on the proper interactions and assembly of microfibrils and elastin as these networks define structures throughout the lungs, including the pulmonary vascular system, the bronchi, and respiratory units, and the pleural connective tissue (309). Perinatal inflammation can cause downregulation of FBN-1 and other elastic fiber assembly components (e.g., fibulins 4 and 5), disrupting elastic fiber organization and lung development in preterm infants (310). Alternatively, overexpression of FBN-1 in the tight-skin mouse results in an abnormal pulmonary ECM and profound airspace enlargement that is characteristic of bronchopulmonary dysplasia and cigarette smoke-induced emphysema in humans (311). Thus, perturbations of FBN-1 expression can have severe pulmonary consequences.

The microfibril network also involves direct and indirect interactions with other ECM molecules, including versican and many others relevant to the versican interactome, such as HA, fibronectin, fibulin, and laminin (312, 313). The interaction between versican and FBN-1 is direct and involves the C-terminal G3 domain of versican with the region of FBN-1, which is essential to its interactions with TGF-β and whose disruption is causative to the development of Marfan syndrome (307). Versican and FBN-1 microfibrils have been identified in healthy and diseased tissues (314). Macromolecular complexes of versican, HA, and FBN-1 have been found in the ciliary and vitreous bodies of the healthy eye; imbalances in the composition of these complexes and proteolytic degradation of components of these complexes are suggested to contribute to several ocular disorders (315, 316). Versican, HA, and FBN-1 also have been colocalized in the stroma of breast cancer tissues; alterations in these components are suggested to influence the biological phenotypes of tumor cells and the progression of the disease (317). The versican G1 domain cleavage product is also found in healthy skin; this fragment is shown to self-aggregate and bind to intact versican and enhances recruitment of HA to the microfibril network. It is hypothesized that the microfibril network forms a depot for this versican fragment, which can be called upon to capture more HA at sites of inflammation as needed (318). In these ways, the interactions between versican and FBN-1 connect the microfibril and elastin networks to the HA-rich matrix in defining the architecture of the complex ECM.

To date, nothing is known regarding the interactions of versican with FBN-1 and microfibrils in the lungs during development, homeostasis, or inflammation.

Collagen type I.

Fibrillar collagens, including collagen type I, are the largest component of connective tissues (319, 320). All are modular proteins consisting of three polypeptide chains with helical structures; collagen type I is heterotypic with two α1 chains and one α2 chain. The fibrillar collagens are synthesized and secreted into the ECM as soluble precursors, which undergo significant posttranslational processing to produce the mature collagen molecules that self-assemble into fibrils. This processing involves enzymatic cleavage of the N- and C-termini propeptides, hydroxylation of lysine and proline residues, glycosylation of hydroxylated lysines, and cross-linking collagen chains; these steps are necessary for proper assembly of the collagen fibrils and integration into the complex ECM of healthy tissues. Further proteolytic processing of mature collagen is a feature of many chronic diseases. Collagen type I is susceptible to cleavage by MMP-8, MMP-9, and prolyl endopeptidase (PE), resulting in the generation of the proline-glycine-proline (PGP) tripeptide, which is chemotactic for neutrophils. The chemotactic activity of PGP is thought to be due to its sequence similarity to the CXC chemokine, IL-8. As neutrophils contain active MMP-8, MMP-9, and PE, this indicates a self-perpetuating cycle of neutrophilic inflammation (321, 322).

In healthy lungs, the triple helical structure of collagen type I (and also type III) forms a tight fibrous network in the interstitial matrix throughout the large conducting airways, bronchi, and bronchioles, providing the strength, stability, and structure required for their proper function (323, 324). Several chronic pulmonary disorders are characterized by aberrant deposition or turnover of collagen type I (323, 325). IPF is characterized by excessive collagen type I deposition in the alveolar parenchyma, causing expansion of fibrotic foci that are associated with advanced disease. TGF-β mediates these changes by inducing differentiation of myofibroblasts, enhancing collagen gene transcription, and altering the balance between MMPs and TIMPs, which favors collagen deposition. In asthma, chronic allergic airway inflammation is characterized by deposition of collagen type I in the ECM of both the basement membrane and interstitial compartments, mediated by TGF-β and reorganization of the complex ECM, in part mediated by inflammatory immune cells (including eosinophils, neutrophils, mast cells, and Th2 cells). Chronic obstructive pulmonary disease (COPD) is characterized by excessive breakdown of collagen type leading to lung tissue destruction and emphysema. In cigarette smoke-induced COPD, this is due to activation and recruitment of innate immune cells (especially macrophages and neutrophils) into the lungs; secretion of MMPs and PE by these immune cells promotes cleavage of collagen type I and generation of chemotactic PGP leading to propagation of the inflammatory response (321, 322).

It has long been known that the interactions between versican and collagen type I have an important role in organizing the complex interstitial matrix surrounding fibroblasts (326) and that collagen type I synthesis occurs in versican-rich areas containing myofibroblasts in both granulomatous (sarcoidosis, extrinsic allergic alveolitis, and tuberculosis) and nongranulomatous fibrotic lung diseases (ARDS, BOOP, and IPF) (32, 33). Collagen type I and versican also have been colocalized in more recent studies of human diseases, including distal airways of lung tissue from patients with ARDS (327), large airways of patients with COPD (328), and the myofibroblast core of the fibroblastic focus in IPF lung specimens (329). In vitro studies show that TGF-β can mediate the increased expression of collagen type I and versican in bronchial epithelial cells (330). Although changes in collagen type I and versican expression are thought to be significant in the progression of these diseases, the nature of their interaction and how they contribute to lung pathology are not well-understood. It is likely that other versican-binding partners in the complex ECM, including HA and fibronectin, are involved in this dynamic interaction.

Fibronectin.

Fibronectin (FN) is found in the circulation (plasma FN) and tissues (tissue FN) during development and wound healing. Plasma FN is a component of the provisional matrix in the early stage of wound healing, where it is essential for platelet function and clot formation (331); tissue FN is an important component of the later stages of wound healing, tissue remodeling, and regeneration. In tissues, FN is one of the major glycoproteins of both interstitial and basement membrane ECMs, where it interacts with numerous binding partners; it is capable of self-association to form multimeric FN fibrils and also interacts with integrins, cell surface receptors, and matrix-degrading enzymes, as well as several other components of the complex ECM. Through these multiple interactions, FN influences matrix assembly and cellular processes throughout tissue development and wound healing (332–335). Both FN synthesis and subsequent turnover are essential to the healing process, and disruption to either process contributes to many chronic conditions. FN synthesis is regulated by TGF-β, and turnover is thought to be regulated by ADAMTS9 and MMP14 (335).

In developing lungs, FN is thought to have roles in airway formation, alveolar epithelial cell differentiation, lung growth, and maturation (336). Dysregulation of FN synthesis and turnover in chronic pulmonary diseases has long been described. These disorders are characterized by the transition of quiescent tissue fibroblasts into activated myofibroblasts which secrete several profibrotic molecules, including TGF-β and FN. More recently, it has been shown that the EDA isoform of FN (FN-EDA) is highly upregulated by TGF-β during activation of myofibroblasts (337). The importance of FN-EDA in lung pathology was shown in studies of lungs from bleomycin-treated FN-EDA-null mice, in which TGF-β activation, accumulation of myofibroblasts, ECM, and fibrosis were all diminished compared with control mice; these processes are essential to the resolution of inflammation is indicated by the higher mortality of FN-EDA-null mice in response to bleomycin (338).

Versican and FN can interact via both the G3 domain and the chondroitin sulfate chains of versican with multiple domains of FN. That versican expression is altered along with FN and other matrix molecules as part of the extensive remodeling of the ECM that occurs in fibrotic diseases is well-described (128, 329, 339). However, little is known about how FN-versican interactions contribute to the fibrotic process (326). It is known, though, that fibrosis is dependent on HA, which also binds to FN, and whose interactions with CD44 and TLR4 both promote and diminish maintenance of the myofibroblast phenotype via complex stimulation and feedback regulation of TGF-β (340). Thus, it is not unlikely that FN-versican interactions have important consequences in pulmonary fibrosis.

Versican-modifying enzymes.

The majority of the enzymes that are known to cleave the protein core of versican belong to the matrix metalloproteinase (MMP) and the A disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS) families of proteases (21). Both of these types of proteases require a zinc ion for catalytic activity, are synthesized as latent enzymes requiring activation, are active at neutral pH, and are known to catalyze the turnover of the ECM during development, homeostasis, and various disease conditions (341–345). MMPs and ADAMTS proteases do not recognize single consensus cleavage sites and have multiple substrates beyond versican (346). These proteases may catalyze, activate, or deactivate ECM proteins and ECM-associated secreted factors, including collagens, fibronectin, laminins, PG, cytokines, and chemokines, thereby adding complexity to their activity and role within the versican-enriched ECM (343, 345–348). Four endogenous tissue inhibitors of metalloproteinases (TIMPs) have been identified that inhibit the activity of MMPs and ADAMTS proteases (343, 349, 350).

MMPs and ADAMTSs are increasingly being recognized for their contextual regulatory and immunomodulatory roles in various disease and inflammatory settings (344, 348, 351). In vivo, the role and activity of these proteases are likely heavily influenced by their colocalization with either activating or inhibiting cofactors and their substrates (344, 346, 348, 352). We will focus our discussion of versican-modifying enzymes on proteases belonging to these families and on what is known about how they interact within the versican-enriched ECM. However, it is important to note that plasmin has been identified as a versican-modifying enzyme in a primate model of flow-induced intimal atrophy and may be an important component of the versican interactome in context where MMP activity toward versican does not predominate (350).

MMP and ADAMTS protease activity has become intimately intertwined with the contextual nature of the versican-enriched ECM as these proteases can interact with versican in several different ways (1, 21, 346). In disease and inflammatory contexts, imbalance between protease activation and inhibition of proteases by TIMPs contributes to significant degradation of the ECM (343, 353, 354). Cleavage of the versican core protein by MMP and ADAMTS proteases leads to degradation and turnover of the versican-enriched ECM, which has important implications for the function of both intact and cleaved versican products (53, 347, 355, 356). The cleavage of versican by proteases is thought to be sequential and yields bioactive fragments, known as matrikines, with distinct functions in a step-wise manner (55, 357). In addition, versican’s GAG side chains can bind directly with MMPs, ADAMTSs, and TIMPs, allowing their activity to be directly regulated by versican (1, 52, 352, 358). This regulatory effect is in part through the colocalization of protease with activators or inhibitors, some of which are members of the versican interactome, such as fibronectin or fibulin-1 (53, 346, 351). GAG side chains also indirectly regulate the function of proteases toward their chemokine and cytokine substrates which can be bound and sequestered by versican’s GAG side chains, concealing them from proteolysis (57, 345). Subsequent cleavage of the core versican protein into bioactive fragments containing GAG side chains may contribute to chemotactic gradients for these secreted factors and play an important role in inflammatory contexts.

Versican has multiple cleavage sites in its core protein that are known to be cleaved by ADAMTS proteases, including ADAMTS-1, -4, -5, -9, -15, and -20 (49, 359–363). ADAMTS-8 is also predicted to cleave versican as it has been shown to cleave aggrecan and shares homology with the ADAMTS proteoglycanases that have activity against versican (55). In the V0 isoform, ADAMTS cleavage sites are located at Glu⁴⁰⁵-Gln⁴⁰⁶ in the αGAG domain and at Glu¹⁴²⁸-Ala¹⁴²⁹ in the βGAG domain (V0 human sequence enumeration) (49, 363). These cleavage sites correspond to single sites in the V1 and V2 isoforms at locations Glu⁴⁴¹-Ala⁴⁴² and Glu⁴⁰⁵-Gln⁴⁰⁶, respectively (49, 363). Both cleavage sites have been validated in vivo using antibodies against the neoepitopes generated by ADAMTS cleavage with anti-DPEAAE⁴⁴¹ (V1 human enumeration) corresponding to a βGAG domain cleavage site and anti-NIVSFE⁴⁰⁵ (V0/V2 human enumeration) corresponding to the αGAG domain cleavage site (49, 363).

The versican fragments generated from cleavage at these locations have been studied in various contexts (52, 55, 85, 364–367). The cleavage product of V1 at Glu⁴⁴¹-Ala⁴⁴² generates a versican fragment with the neoepitope DPEAAE, which has been named versikine and is conserved between human and mouse (49, 55). The cleavage product of V0/V2 at Glu⁴⁰⁵-Gln⁴⁰⁶ that generates a versican fragment with the neoepitope NIVSFE has been named glial hyaluronate-binding protein (GHAP). This neoepitope has only been identified in humans and is found predominately in brain tissue (363, 368). In mice, the neoepitope NIVNSE, which corresponds to the Glu⁴⁰⁵-Gln⁴⁰⁶ ADAMTS cleavage site within the mouse sequence was recently validated and will enable the researcher to better understand the bioactivity of GHAP in future studies (367). Several other ADAMTS versican cleavage sites have been predicted using in vitro and modeling techniques but still await in vivo validation (51, 369).

Versican degradation by ADAMTS proteinases plays an important role in development, inflammation, and disease states. In development, the activity of ADAMTS-5, -9, and -20 was elegantly shown to work in combination to degrade versican in the regulation of interdigital web regression (366). In this context, fibulin-1 was demonstrated to be an important cofactor for ADAMTS activity, and versikine, the V1 degradation product, demonstrated bioactivity by inducing apoptosis in interdigital webs (366). The bioactivity of versikine has been implicated in disease contexts as well and has been shown to promote epithelial-mesenchymal transition resulting in cell migration and proliferation and acts as a danger-associated molecular pattern in the context of multiple myeloma.

In the context of lung disease, two critical studies have investigated the role of ADAMTS proteases in the remodeling and degrading of the versican-enriched ECM in pulmonary organs during influenza infection. Boyd et al. (355) investigated the role of ADAMTS4 in the lungs. They found that mice lacking ADAMTS4 had higher levels of versican accumulation combined with fewer CD8 lymphocytes, less alveolar inflammation, and less protein accumulation in the airways compared with wild-type controls at 9 dpi with influenza. In ADAMTS4-null mice, there was no change in the amount of gene expression of other known ADAMTS versicanases, thus identifying ADAMTS4 as the primary ADAMTS protease in the lungs during influenza infection (355). In addition, it was shown that ADAMTS4-competent fibroblasts promoted T-cell migration across a versican barrier compared with ADAMTS4-deficient fibroblasts in vitro (355). In contrast, McMahon et al. (356) found that ADAMTS5 was the primary ADAMTS protease responsible for the degradation of the versican-enriched ECM in the mediastinal lymph nodes during influenza infection. Similar to the effect of ADAMTS4 deficiency in the lungs, T-cell migration from the lymph nodes was impaired in influenza-infected ADAMTS5-deficient mice (356). In addition, versikine was decreased in the mediastinal lymph nodes of ADAMTS5-deficient mice, suggesting that versican proteolysis by ADAMTS and/or the bioactivity of versikine may be necessary for T-cell migration in the context of influenza infection in pulmonary organs (356). Although Boyd and colleagues (37) did not assess versikine accumulation in the lungs, versikine accumulation has been shown to peak from baseline at 6 dpi with influenza in wild-type mice.

Versican’s core protein is known to be cleaved by MMPs, including MMP-1, -2, -3, -7, -8, -9, and -12, but unfortunately, specific versican fragments and neoepitopes have not been consistently identified in these studies (29, 48, 50, 368). Versican derived from rabbit lungs with hydraulic edema was shown to be sensitive to both MMP-2 and MMP-9 degradation, with the activity of MMP-2 being greater than that of MMP-9 (29). In atherosclerotic lesions, MMP-7 expression and positive immunostaining for versican were localized to the same area. Additionally, MMP-7 degraded versican in vitro more efficiently than MMP-1, -2, -3, and -9 (48). A cleavage site is known for the activity of MMP-1, -2, -3, and -9, which have been demonstrated to cleave versican extracted from the bovine brain at Glu⁴⁰⁵-Gln⁴⁰⁶, a site shared with ADAMTS proteases, generating GHAP.

In addition, MMP-8 and MMP-12 have been demonstrated to generate a bioactive versican degradation product named VCANM, which is cleaved at position Tyr³³⁰⁵ in the G3 domain with the neoepitope KTRFGKMKPRY (50). VCANM is a potential biomarker for atherosclerotic disease with high plasma levels correlating to patients with acute coronary syndrome, however, low levels of VCANM were associated with acute exacerbation of chronic obstructive pulmonary disease and idiopathic interstitial pneumonia (50, 353, 354). Very little is known about the bioactivity of VCANM in different disease contexts. Therefore, more research is needed to better understand the role of VCANM in lung health and disease.

In the context of influenza infection, a study by Bradley et al. identified neutrophils as the predominant pulmonary cellular source of MMP-9. It demonstrated that the MMP-9-deficient mice had impaired neutrophil migration into airways, resulting in improved weight loss from severe influenza infection (347). Although the activity of MMP-9 toward versican was not considered in this study, MMP-9 secretion by neutrophils was shown to be induced by TNFα, an important proinflammatory versican-associated secreted factor in the versican interactome (347). To investigate a possible relationship between neutrophil migration and versican during influenza infection, the predominant MMP-9 cleavage site on the versican core would need to be elucidated and any resulting versican fragments would need to be defined.

One of the prominent challenges to gaining a better understanding of the contextual nature of the versican interactome, especially regarding how versican-modifying enzymes impact the immune response during infection and disease, is that versican-modifying enzymes act on multiple substrates beyond versican and may cleave versican at multiple sites generating multiple fragments with bioactivity. This complicates mechanistic interpretation of phenotypes in mouse models deficient in a particular MMP or ADAMTS protease as the phenotype may be due to loss of substrate function, substrate accumulation, or generation of bioactive substrate fragments. Recently, two mouse models have been generated that address this challenge by abolishing the ADAMTS cleavage site in the versican βGAG domain (365, 367). Although these mice have been used primarily to understand the versican interactome during development and wound healing, their application in studies of inflammation and disease states will lead to an improved understanding of the contextual nature of the versican interactome.

SUMMARY

Contextual Nature of the Versican-Enriched Matrix

The development of genetically engineered mice that are conditionally deficient in versican provides evidence of the contextual nature of the versican-enriched matrix (24, 26). The mechanisms whereby versican develops a pro- or anti-inflammatory ECM remain a significant knowledge gap. The contextual nature of the ECM is best illustrated by the interactions of IαI and TSG-6, which result in an HC-HA ECM that can be either pro- or anti-inflammatory. Studies by Stober et al. (172) show that tracheal epithelial cells form an HC3-HA matrix adhesive for leukocytes, supporting a proinflammatory role for the HC-HA ECM. In contrast, the HC-HA complex derived from the amniotic membrane promotes an M2 phenotype in macrophages and reduces inflammation in vitro and in vivo (162, 177, 370, 371). The HC-HA complex of amniotic fluid contains HC1 and pentraxin but not HC2, HC3, bikunin, or TSG-6, suggesting that the HC1-HA-pentraxin matrix is anti-inflammatory (162, 370). Based on these studies, it is apparent that the composition of the HC-HA matrix defines its biological function as pro- or anti-inflammatory.

Whether the addition of versican to the HC-HA matrices will change their function is not known. In fact, little is known about the composition or inflammatory nature of the versican-HA-enriched matrix formed in the lungs during the development of lung disease. The works of Kang et al. (26) and Potter-Pergio et al. (130) provide evidence that versican binding to HA is required for the adhesion of both monocytes and T cells to the HA matrix in vitro. We believe that to understand the biological function of the versican-enriched matrix in lungs, it is essential that future studies examine spatial and temporal changes not only in the expression of versican and HA but must include other members of the versican-interactome.

Six Important Questions to Determine Contextual Nature of the Versican-Enriched Matrix

Investigating the following six questions will lead to a better understanding of the mechanisms whereby the binding of versican to select molecules of the versican-interactome determines the contextual nature of the versican-enriched ECM in the lungs.

Who.

What members of the versican-interactome are synthesized near regions of versican accumulation in the lungs under various conditions?

What.

What is the specific agonist (e.g., TLR agonist or disease process) or signaling pathway (β-catenin/T-cell factor or type I IFN) that stimulates the development of the versican-enriched matrix?

When.

What are the temporal dynamics of the expression, synthesis, accumulation, and degradation of versican and the versican-interactome in response to a specific agonist?

Where.

What are the spatial dynamics of the expression, accumulation, and degradation of versican (Fig. 2) and the versican-interactome in response to a specific agonist? The spatial dynamics include the cellular source of versican synthesis (e.g., myeloid vs. mesenchymal), the anatomical location (e.g., perivascular vs. peribronchiolar), or the compartment where versican accumulates (e.g., vasculature, lung tissue, or airways).

How.

What are the mechanisms whereby versican—the isoform, core protein domains, CS side-chains, or degradation products (Fig. 3)—binds to and alters the function of its binding partners in the ECM?

Why.

What is the outcome of versican binding to select components of the versican-interactome in defining cellular phenotype, leukocyte migration, lung function, and the immune response in the lungs (Fig. 1)?

CONCLUSION

Information provided in this review shows the complexity of the ECM, its ability to provide fine control of cellular processes in the lungs, and the contextual nature of the versican-enriched ECM. Although this review has focused on versican and the versican interactome in the lungs, the interactions between versican and its binding partners can control cellular phenotype and function in other tissues and organ systems in health and disease. In addition, the interactions between other proteoglycans and components of the matrisome will most likely play a role in the contextual nature of the ECM. Therefore, information provided in this review is relevant to other organs systems in health and disease.

GRANTS

This work was supported in part by the following: National Institute of Allergy and Infectious Diseases Grants 1R01AI136468-01 and 1R01AI130280 to W.A.A. and C.W.F.; University of Washington RDP (SINGH19R0) to C.W.F.; National Institute of Allergy and Infectious Diseases R21AI147536; National Heart, Lung, and Blood Institute R01HL122895 and National Institute of Allergy and Infectious Diseases R01AI136468 (to W.A.A.); and National Heart, Lung, and Blood Institute K08HL135266 (to S.R.R.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

C.W.F. prepared figures; F.T., J.E.B., M.Y.C., S.R.R., W.A.A., and C.W.F. drafted manuscript; F.T., J.E.B., M.Y.C., S.R.R., W.A.A., and C.W.F. edited and revised manuscript; F.T., J.E.B., M.Y.C., S.R.R., W.A.A., C.W.F. approved final version of manuscript.