Matrix proteoglycans in tumor inflammation and immunity

Abstract

Cancer immunoediting progresses through elimination, equilibrium, and escape. Each of these phases is characterized by breaching, remodeling, and rebuilding tissue planes and structural barriers that engage extracellular matrix (ECM) components, in particular matrix proteoglycans. Some of the signals emanating from matrix proteoglycan remodeling are readily co-opted by the growing tumor to sustain an environment of tumor-promoting and immune-suppressive inflammation. Yet other matrix-derived cues can be viewed as part of a homeostatic response by the host, aiming to eliminate the tumor and restore tissue integrity. These latter signals may be harnessed for therapeutic purposes to tip the polarity of the tumor immune milieu toward anticancer immunity. In this review, we attempt to showcase the importance and complexity of matrix proteoglycan signaling in both cancer-restraining and cancer-promoting inflammation. We propose that the era of matrix diagnostics and therapeutics for cancer is fast approaching the clinic.

INTRODUCTION

Extracellular matrix (ECM) remodeling is a crucial regulator of tumor progression (1, 2). The ECM is a diverse, noncellular, three-dimensional network composed of proteoglycans (PGs) and glycosaminoglycans (GAGs) as well as fibrous proteins, including collagens, fibronectins, elastins, and laminins (3–5). Its composition is dynamic and changes substantially from organ to organ. The ECM is an integral structural component of the tumor microenvironment (TME), the latter also incorporating an intricate milieu of endothelial, mesenchymal, and immune cells that collectively host and support the tumor. The importance of ECM remodeling to local invasion, angiogenesis, and metastasis is well established (6, 7). However, the contribution of matrix composition in regulating the cancer-immunity cycle and shaping the tumor immune microenvironment is only beginning to be systematically explored (8).

The dynamic interfacing between the developing tumor and the host immune response has been termed “immunoediting” and consists of three distinct stages: elimination, equilibrium, and escape (9) (Fig. 1). During the first stage, elimination, the host immune system is successfully able to detect and destroy cancer cells before the tumor reaches a detectable size. During the equilibrium stage, however, surviving tumor cells undergo genetic and/or epigenetic changes (“edits”) that compromise the host immune system’s ability to eliminate the tumor cells, leading to a “dormant” tumor with stagnant growth and a dynamic, albeit precarious, equilibrium between the tumor and host immune response. The final stage, escape, is characterized by tumor outgrowth, made possible by diminished or lost tumor antigen repertoire (which may also happen during the equilibrium stage), secretion of cytokines that promote cell growth, division, angiogenesis, and immune suppression as well as the recruitment of regulatory, protumor, immune cells (17). Each immunoediting stage is accompanied by extensive ECM breaching, deposition, and reshaping, resulting in the release or alteration of immunomodulatory ECM components, such as small leucine-rich proteoglycans (SLRPGs), large matrix proteoglycans (e.g., versican), hyaluronan (HA), and heparan sulfate fragments. True to their derivation from stroma and tissue plane restructuring, these mediators act as “danger signals” [termed damage-associated molecular patterns (DAMPs) or “alarmins”], to activate immune system sensor mechanisms (17–19). In this review, we shall specifically focus on the role of matrix proteoglycans and their products in shaping tumor inflammation and immunity.

Figure 1.

Matrix proteoglycans in cancer evolution and immunoediting. Cancers are “wounds that do not heal.” Immunoediting refers to the distinct stages of interfacing between the host immune system and the developing cancer. There are three stages: elimination, equilibrium, and escape (see text for details). Extracellular matrix remodeling regulates key signaling pathways at each step. The case is illustrated by versican (VCAN), a versatile large matrix proteoglycan that possesses diverse context-specific immunoregulatory or immunostimulatory roles, depending on its abundance, isoform structure, and proteolysis status (10). We propose the following model based on work by our group and others (11–13): during elimination, robust VCAN proteolysis releases the immunostimulatory fragment (matrikine), versikine, that regulates stimulatory dendritic cells (conventional type 1 dendritic cells, cDC1) promoting immune destruction of the tumor through enhanced neoantigen recognition. In equilibrium, a precarious balance is established between immunostimulatory proteolyzed VCAN products and parental, nonproteolyzed immunoregulatory VCAN that allows the propagation of tumors with active, but progressively ineffectual, immune surveillance. In escape, progressive fibrosis and immunosuppression [e.g., through TGF-β (14)] attenuates VCAN proteolysis and results in the unbalanced immunoregulatory activity of nonproteolyzed VCAN. In these latter stages, immunosuppressive signals from receptors recognizing collagen may further promote immune exclusion (15, 16). Created with BioRender.com.

PROTEOGLYCANS IN TUMOR IMMUNITY: AN OVERVIEW

PGs comprise a heterogeneous group of complex macromolecules that form an important structural and functional component of the ECM (20). Structurally, PGs consist of a core protein onto which one or more glycosaminoglycan (GAG) chains are covalently attached. Known PGs can be classified into four families based on criteria that include cellular and subcellular localization, overall gene/protein homology, and the utilization of specific protein modules: intracellular, cell surface, pericellular, and extracellular (21) (Table 1). GAGs, also known as mucopolysaccharides, are negatively charged polysaccharides composed of repeating disaccharide units. GAGs, based on the type of monosaccharides and presence or absence of modification by sulfation, are categorized into the following: chondroitin sulfate (CS)/dermatan sulfate (DS), keratan sulfate (KS), heparan sulfate (HS), and hyaluronan (HA). The sulfate-enriched negatively charged GAG chains in PGs bind water and positively charged soluble ligands such as cytokines, chemokines, growth factors, matrix metalloproteinases (MMPs), and cell surface receptors, facilitating the function of PGs as important mediators of cell proliferation, differentiation, adhesion, and dissemination within the ECM, as well as inflammation and angiogenesis (154). The bioactivity of PGs is further modulated by enzymes present within the tumor microenvironment, such as sheddases, proteases, and heparanases, that cleave these biomolecules and modify the structure and function of associated PG and their bound ligands.

Table 1.

Key proteoglycans organized per class, with cellular source(s), and selected functional roles indicated

Proteoglycan	Expression	Source(s)	Functional Role(s)	References
Serglycin	Intracellular	Immune cells, endothelial cells, fibroblasts, chondrocytes, smooth muscle cells, fetal liver, nasopharyngeal, and hematopoietic cancer cells	Complexes with and modulates the activities of mast cell proteases, e.g., chymases, tryptases, and carboxypeptidase A; acts as a vehicle for delivery of granzyme B in cytotoxic cells; triggers CD44 signaling in tumor cells; regulates IL-8 secretion in cancers; via tumor-derived exosomes, it protects human multiple myeloma from complement attack by binding C1q and MBL in the TME	21, 22–24, 25–42
Syndecan-1	Cell surface	Normal and malignant plasma cells; normal and malignant epithelial cells	Sequesters T-cell specific CC-chemokines, and in soluble form, with VEGFR2 and VLA-4; On the cell surface, engages CXCR4, VLA-4, and VEGFR2; upregulates IL-23 and Notch ligand DLL4, which polarizes CD4+ T cells toward the Treg, Th17, and Th1 phenotypes	21, 43–56, 57–64
Syndecan-2	Cell surface	Mesenchymal cells, tumor-associated stromal cells	Modulates TFG-β signaling to control tumorigenesis via TFG-β-induced genes PD-L1 and CXCR4 in TASC	21, 43–56, 65
Syndecan-3	Cell surface	Neuronal tissue; tumor-associated macrophages; tumor endothelial cells; bladder, prostate, breast, and pancreatic cancer cell lines	Accentuates the IFN-γ-regulome, including upregulation of CD274 and CD8; regulates hypoxia-resistance in a HIF1α-dependent manner	21, 43–56, 66,67
Syndecan-4	Cell surface	Ubiquitous expression	Ligand for dendritic cell-associated heparan sulfate proteoglycan-integrin ligand (DC-HIL) to attenuate T-cell activation; sequesters TFG-β on the cell surface	21, 43–56, 68–72
Chondroitin sulfate proteoglycan 4	Cell surface	Monocytes/macrophages	Activates monocytes to secrete IL-1β and induce monocyte-dependent B cell proliferation in vitro	21, 73–75
Betaglycan	Cell surface	Ubiquitous expression	Accentuates TGF-β signaling or interferes with TGF-β receptor type I and TGF- β receptor type II complexing; in soluble form, can sequester TGF-β; upregulates Cdc42 via β-arrestin-2 in vivo	76–81
Glypican-1	Cell surface	Ubiquitous expression	FGF signaling; PTEN attenuation; Akt and β-catenin signaling; recognition by NK cells via NKp46 and NKp30; interacts with VEGF via HS chains	21, 82, 83, 84
Glypican-2	Cell surface	Embryonic central nervous system tissue; neuroblastoma cells	Upregulates expression of N-Myc, which in turn promotes transcription of the glypican-2 in a positive feedback loop	21, 82, 83
Glypican-3	Cell surface	Ovary, breast, lung, and kidney epithelial cells	Sequesters VEGF in TME; inhibits Zeb1 transcription factor; influences PI3K/Akt, p38, ERK1/2, and Wnt signaling pathways (context dependent); induces cytotoxic T-cell and macrophage recruitment in HCC	21, 82, 83, 85–87
Glypican-4	Cell surface	Ubiquitous expression	Accentuates Wnt signaling and pluripotency; reduces sensitivity of pancreatic cancer cells to 5-fluorouracil	21, 82, 83
Glypican-5	Cell surface	Central nervous system	Acts through FGF, HGF, and Wnt1α to promote cell division; tumor suppressor in certain contexts, e.g., glioma	21, 82, 83
Glypican-6	Cell surface	Gallbladder, urinary bladder, liver, appendix, and kidney epithelial cells	CD8A mRNA transcription; Wnt signaling modulation; tumor infiltration of cytotoxic T-cells	21, 82, 83, 88
Perlecan	Pericellular	Stromal cells	Sequesters bioactive molecules (via GAG chains)	21, 89–93
Agrin	Pericellular	Stromal cells	Activates VEGF-VEGFR2 pathway; stimulates angiogenesis	21, 94
Biglycan	Extracellular	Stromal cells; macrophages; immune cells	Binds TLR2/4; forms complexes with CD14-TLR4 and P2X4/P2X7 respectively; promotes tumor progression	95, 96–104, 105
Decorin	Extracellular	Stromal cells	Binds TLR2/4; sequesters TGF-β; upregulates PDCD4; inhibits tumor growth	95, 105–107
Lumican	Extracellular	Stromal cells	Enhances LPS-dependent TLR4 signaling; tumor-dependent impact on EMT, immune infiltration and inflammation	21, 97, 108–110
Versican	Extracellular	Stromal cells; myeloid cells	Binds hyaluronan, P and L selectins, CD44, and TLR2; promotes accumulation of tumor-associated macrophages and decrease intraepithelial tumor infiltrating lymphocytes; positively correlated with Th2 and Treg transcriptional signatures, negatively correlated with cytotoxic T-cell signatures	22, 10–13, 111–148
Aggrecan	Extracellular	Cartilage and perineural net stromal cells	Forms aggregates for load-bearing structural support and mechanical properties to joints; protects neurons via cation sequestration and mechanically stabilization of synaptic connections	21
Brevican	Extracellular	Neural tissue stromal cells	Interacts with tenascin-R and fibulin-2; neural tissue injury and repair, Alzheimer’s disease, glioma tumorigenesis	21
Neurocan	Extracellular	Neural tissue stromal cells	Alleviates mechanical and ischemic damage in central nervous system; inhibits neurite outgrowth in vivo; assists in neural guidance during development	21
Hyaluronan and proteoglycan link protein 1	Extracellular	Stromal cells	Binds to aggrecan complexes in cartilage to offer structural support; NF-κB pathway activation; myeloma drug resistance	149–153

Tumor evolution results in significant alterations in PG expression and composition (21, 95, 155). Altered expression of GAG-synthesizing enzymes during malignant transformation affects the type and fine structure of GAG chains attached to PGs. Whereas these modifications have been classically thought to facilitate local invasion, angiogenesis, and metastasis, there is increasing evidence that PG remodeling—such as proteolysis of the protein core and alterations to the GAG sidechains—has further implications in shaping tumor-immune cross talk (22–24). As will be discussed in greater detail, these implications include the diversification of cell-surface PG functionalities through regulated shedding, the synthesis and degradation of cytokine- and chemokine-sequestering GAG chains, and the proteolytic release of immunomodulatory proteolytic fragments (matrikines). Our understanding of role that matrix PG remodeling plays in shaping tumor-immune cross talk will continue to increase as this growing field matures.

In recent years, the application of high throughput multiomics technologies has allowed the comprehensive cataloging and characterization of PG and other matrix components in diverse cancer types (5, 156, 157). For example, Naba et al. (156) developed a bioinformatic approach to predict the in silico “matrisome” consisting of ECM proteins and associated factors. They reported a detailed composition of human melanoma extracellular matrices (156). Both tumor and stromal cells were found to contribute, although differentially, to the secretion of the proteins making up the tumor ECM (156). Similar strategies could reveal novel molecular players, whose functional contributions can then be further studied, in addition to holding significant promise for biomarker identification in various cancer settings.

INTRACELLULAR PG—SERGLYCIN

Serglycin (Fig. 2) forms a class on its own, as it is the unique intracellular PG (21). Initially, it was identified as the dominant PG expressed in normal immune cells such as monocytes, macrophages as well as their malignant counterparts in hematopoietic cancers. However, serglycin is also expressed in nonhematopoietic normal and malignant cell types, such as endothelial cells, fibroblasts, chondrocytes, smooth muscle cells, fetal liver, murine embryonic stem cells, and nasopharyngeal cancer cells (22–25). The type and sulfation pattern of GAG chains attached to the serglycin core protein vary between cell types and different species (26). Serglycin-knockout mice have helped unravel its functional roles in both normal and pathological immunity (26, 27). Serglycin is localized in secretory granules and vesicles for proper intracellular storage and secretion of bioactive molecules. For example, serglycin forms complexes with, and modulates, the activities of mast cell proteases such as chymases, tryptases, and carboxypeptidase A among others (28, 29). Serglycin-deficient mast cells showed reduced sensitivity to apoptosis as a result of diminished storage and release of mast cell proteases into the cytosol and impaired caspase-3 activation (30). In cytotoxic T lymphocytes (CTLs), serglycin binds to granzyme B and acts as a vehicle for its delivery into the target cells during cytotoxic cell granule-mediated cell death (31). Lack of serglycin expression may affect CTL and NK cells’ ability to kill target cells (32, 33).

Figure 2.

Immunomodulatory roles of intracellular PG. Serglycin (SRGN) can modulate the activation status of the complement system. SRGN-CD44 interaction activates intracellular signaling pathways, such as NF-κB, promoting tumor aggressiveness. SRGN regulates IL-8 and TGF-β secretion, thus controlling crucial inflammatory networks in the TME. PG, proteoglycan. Created with BioRender.com.

Serglycin expression has been demonstrated in multiple cancer types including breast cancer, lung cancer, myeloma, lymphoma, mastocytoma, and thymoma. Secreted serglycin triggers CD44 signaling on tumor cells (34, 35). CD44 regulates various cellular processes such as migration, proliferation, and apoptosis (36). CD44 is also a primary receptor for HA, which will be discussed in further detail later. Guo et al. (35) showed that serglycin secreted by nonsmall cell lung cancers (NSCLCs) and/or cancer-associated fibroblasts (CAFs) in the TME can bind CD44 on tumor cells to promote malignant phenotypes by inducing stemness through the transcription factor Nanog, accompanied with chemoresistance and anoikis resistance. Serglycin, via the CD44/NF-κB/claudin-1 (CLDN1) axis, was shown to upregulate vimentin expression and hence promote epithelial-to-mesenchymal transition (EMT). Chu et al. (37) reported that in nasopharyngeal carcinoma (NPC), serglycin-CD44 interaction activates the MAPK/β-catenin axis to induce CD44 receptor expression in a positive feedback loop, resulting in NPC tumor cell stemness and chemoresistance. Serglycin is highly expressed in triple-negative breast cancer (TNBC) and it induces the expression and secretion of TGF-β2, which subsequently promotes EMT (38). Binding of serglycin to CD44 was shown to induce CREB phosphorylation, which in turn activated TGF-β2 transcription. Conversely, TGF-β2/TGFR1 signaling was shown to positively regulate serglycin expression by activating downstream SMAD2/3 signaling. Bouris et al. (39) showed that serglycin regulates IL-8 secretion in breast carcinoma cells and induces IL-8/CXCR2 signaling axis to activate PI3K, Src, and Rac GTPase activation. Furthermore, serglycin overexpression was shown to promote an aggressive phenotype, EMT, and drug resistance in estrogen receptor-α (ER-α) positive MCF-7 breast cancer cells. Skliris et al. (40) showed that serglycin secreted by human multiple myeloma cells may protect cancer cells from complement system attack by specifically inhibiting the classical and the lectin pathways through direct binding to C1q and MBL, without hampering alternative pathway activity. The data demonstrate a role for serglycin as a modulator of tumor immunity in the TME, where it can defend cancer cells from immunotherapy-induced complement attack. Recently, a novel function of serglycin was reported as a critical component of tumor-derived exosomes. Tumor-derived exosomes act as potent regulators of cell-to-cell interactions in the tumor microenvironment due to their ability to carry nucleic acids, proteins, and lipids and to transfer their cargo into the target cells (27, 41). They promote tumor growth through alteration of phenotypic and functional attributes of the recipient cells in the TME, resulting in drug resistance, inhibition of host antitumor responses, and attenuation of immunotherapy efficacy (41). Serglycin was detected in exosomes derived from the cell culture supernatants of human myeloma cell lines and from the serum of patients with myeloma (42). Exosomes derived from serglycin-knockdown cells, but not from control cells, lacked several proteins that are required for mediating crucial biological processes, such as translational initiation, cell adhesion, response to drug, and intracellular protein transport. Moreover, exosomes from serglycin-knockdown cells failed to induce an invasive phenotype in myeloma cells or promote migration of macrophages (42). Altogether these findings suggest that serglycin plays a key role in the encapsulation of proteins in exosomes as well as their transfer into target cells residing in the TME.

CELL SURFACE PGS—SYNDECANS, CSPG-4/NG2, BETAGLYCAN, AND GLYPICANS

Syndecans (Fig. 3) are a family of single-pass transmembrane PGs consisting of four members: syndecan-1 (Sdc-1, CD138), syndecan-2 (Sdc-2, fibroglycan), syndecan-3 (Sdc-3, N-syndecan), and syndecan-4 (Sdc-4, amphiglycan) (21). They are expressed in a developmentally and cell type-specific manner. Sdc-1 is expressed mainly by plasma cells and epithelial cells, whereas Sdc-2 and Sdc-3 are expressed by mesenchymal cells and neurons, respectively. Sdc-4 shows a ubiquitous expression pattern in vertebrates. Structurally, they consist of an ectodomain, a transmembrane domain, and an intracellular domain. The predominant GAG covalently attached to the ectodomain protein core is HS and occasionally CS, which makes syndecans “hybrid PGs.” Syndecans are involved in a wide variety of biological functions, including cell-cell and cell-ECM interactions, cell survival, adhesion, migration, and cell signaling through binding to numerous soluble molecules via their HS chains (43, 44). Many syndecans, if not all, can also function as soluble heparan sulfate PGs (HSPGs) via proteolysis-mediated release of their ectodomain. This shedding is an important posttranslational modification that regulates the amount of HSPG present on the cell surface versus the pericellular microenvironment. The role of syndecans in tumor initiation and progression has been extensively studied (45–47). Both cell-surface and soluble syndecans regulate multiple aspects of tumor cell adhesion and cell migration (44, 47). Cell-surface syndecans in co-operation with integrins control cell adhesion to extracellular fibronectin (48) and promote cell spreading through actin and fascin bundling (49). Cell-surface syndecans may enhance tumor endothelial cell proliferation and motility by binding growth factors (such as FGF-2 and VEGF), presenting them to their high-affinity receptors, and protecting them from inactivation (50, 51). Syndecans can also bind to and induce conformational changes to growth factor receptors, activating downstream signaling pathways that control cell proliferation and motility (46, 52). The ectodomain of soluble syndecans may also bind to certain chemoattractants, such as IL-8, effectively creating chemotactic gradients (53). Finally, ectodomains of soluble syndecans may increase endothelial cell membrane protrusion, migration, capillary tube formation, and cell-cell interactions by competitively inhibiting cell-surface syndecans (54).

Figure 3.

Immunomodulatory roles of transmembrane PG. Syndecan-1 (Sdc-1) regulates NF-κB and STAT-dependent inflammatory networks in tumor cells as well as TME immune cell polarization/activity through Notch. CSC, cancer stem cell; IBC, inflammatory breast cancer; PG, proteoglycan. Created with BioRender.com.

All four syndecans have been reported to be expressed in a variety of cancer types (46). They have been detected in varied components of the TME such as inflammatory cells, fibroblasts, blood vessels, and the ECM. Syndecan accumulation in the tumor stroma could be a consequence of either induction or shedding (55, 56). Many epithelial tumors exhibit stromal expression of syndecans; for example, Sdc-1 has been detected in the stromal compartment of breast cancer, colorectal cancer, endometrium cancer, and gastric cancer (57). In nonepithelial cancers such as multiple myeloma, the stroma stains intensely for Sdc-1, likely derived from tumor shedding (58). Notably, in myeloma, malignant plasma cells express high levels of Sdc-1 (CD138), which is often used as a lineage marker for this cancer.

Syndecans display diverse roles in cancer, acting as either tumor promoters or suppressors, depending on the type and stage of the tumor. Using a murine model of colitis-related colon carcinoma, Gallimidi et al. (59) demonstrated that Sdc-1-deficient mice showed higher susceptibility to malignant transformation due to increased local production of IL-6, activation of STAT3 and downstream effector genes with key roles in colonic tumorigenesis. In contrast, Sdc-1 silencing in inflammatory breast cancer (IBC) SUM-149 cells caused reduced colony formation and 3-D spheroid formation. Sdc-1 knockdown was shown to downregulate NF-κB and STAT3 signaling and reduce IL-6, IL-8, CCL20, gp130, and EGFR mRNA in both SUM-149 and non-IBC breast cancer SKBR3 cells (60, 61).

Sdc-1 shapes the cellular composition of the inflammatory environment by modulating soluble mediators. For example, Sdc-1 can sequester T-cell-specific CC-chemokines through its HS-chains and thereby inhibit T-cell-driven inflammation (62). Saleh et al. (63) focused on the immunomodulatory role of Sdc-1 in the polarization of CD4⁺ T cells within breast cancer TME. Conditioned media from Sdc-1-knockout SUM-149 breast cancer cells significantly enhanced ex vivo polarization of CD4⁺ T cells toward T_reg (Foxp3⁺CD4⁺), Th₁₇ (IL-17⁺CD4⁺), and Th₁ (IFN-γ⁺CD4⁺) via upregulation of IL-23 and Notch ligand DLL4. The Rapraeger group (64) reported a novel mechanism by which Sdc-1 engages CXCR4, VLA-4 (very late antigen-4), and VEGFR2 (vascular endothelial growth factor receptor-2) to modulate cell migration across a variety of cell types. They showed that VEGFR2 and VLA-4 docks in the extracellular domain of sSdc-1 (shed Sdc-1) and VEGFR2 causes PKA-mediated phosphorylation of VLA-4 on S988, which promotes metastasis and immunosuppression.

Loftus et al. (65) showed that Sdc-2 expressed on the cell surface of tumor-associated stromal cells (TASCs), controls tumorigenesis via TGF-β signaling. Decreased Sdc-2 activity in TASCs inhibited TGF-β-induced immunosuppressive genes such as PD-L1 and CXCR4. This effect correlated with reduced ability of TASCs to suppress T-cell proliferation in vitro. Consistently, they observed T-cell activation and tumor inhibition in an immune-competent syngeneic breast cancer model.

Although Sdc-3 is known mainly to be expressed in neuronal tissues, some studies demonstrate it to be expressed in the TME as well as in cancer cell lines such as bladder cancer, prostate cancer, breast cancer, and pancreatic cancer (66). In the TME, Sdc-3 is expressed on tumor cells, TAMs, and endothelial cells. Sdc-3 expression was shown to be induced under hypoxic conditions in CT26 colon carcinoma cells in a HIF-1α-dependent manner (66). In addition, Sdc-3 expression was strongly upregulated in murine macrophage cell line RAW-264.7 treated with proinflammatory cytokine IFN-γ but not anti-inflammatory IL-4. This finding is in line with an earlier study by Takeda et al., which showed that IFN-γ stabilizes HIF1α in macrophages (67). Furthermore, Sdc-3 expression was shown to positively correlate with an IFN-γ-regulome, including CD274 and CD8, across the majority of TCGA tumor types and a better overall survival in melanoma tumors expressing a hypoxia transcriptional signature (66).

Sdc-4 is the major HSPG expressed on the dendritic cell (DC) surface. In the TME, DCs play an important role in T-cell priming and tumor rejection; however, tumor-associated DCs are often polarized into a tolerogenic state that allows cancer cells to escape immune surveillance. In Sdc-4-deficient mice, Lewis lung carcinoma (LLC) tumors were growth-impaired and were characterized by increased infiltration of mature DCs, NK cells, and NKT cells (68). Sdc-4 is a ligand for dendritic cell-associated heparan sulfate proteoglycan-integrin ligand (DC-HIL). DC-HIL is a highly glycosylated type I transmembrane protein of 125 and 95 kDa containing an extracellular Ig-like domain, constitutively expressed by APCs and acts as a negative regulator of T-cell activation (69). DC-HIL has been shown to bind HS chains on Sdc-4 expressed on activated T cells, and this binding attenuates T-cell activation (69–71). Chung et al. (72) showed that malignant T cells from patients with Sézary syndrome (SS) and cutaneous T-cell lymphoma (CTCL) cell lines constitutively expressed Sdc-4 featuring distinct HS moieties at high levels compared with those expressed by normal T cells activated in vitro. Using these HS moieties, Sdc-4 was shown to inhibit activation of T cells through two independent mechanisms: first, by binding to DC-HIL and second, by sequestering immunosuppressive cytokine TGF-β on the cell surface. These findings highlight the importance of HSPGs, like Sdc-4, in regulating tumor immunity.

Chondroitin sulfate proteoglycan 4 (CSPG-4) is a highly glycosylated, single-pass transmembrane PG that carries one CS chain and a large ectodomain (21). It is also referred to as melanoma-associated chondroitin sulfate proteoglycan (MCSP), high-molecular-weight melanoma-associated antigen (HMW-MAA), or neuron-glial antigen 2 (NG2). It was first reported in malignant melanoma as a highly immunogenic tumor antigen and subsequently implicated in the pathobiology of multiple solid tumors as well as hematological cancers. Due to its restricted or low expression in normal tissues, overexpression in certain tumor types, as well as functionally important role in supporting tumor growth and metastasis, it has been investigated as a potential immunotherapy target (73). Erfurt et al. (74) reported the presence of natural CSPG-4 specific CD4⁺ T cells that may confer immunosurveillance against melanoma. In some patients, attenuation of this response is associated with tumor progression. Activated human monocytes/macrophages also produce CS-containing PGs (75) that may subsequently act in a paracrine manner. CSPG-4 activates monocytes to secrete IL-1β and induce monocyte-dependent B cell proliferation in vitro, which can be inhibited by TGF-β and CD44 monoclonal antibodies (75).

Free GAGs can also have diverse immunomodulatory actions. Aoyama et al. (158) showed that chondroitin sulfate B (an older term for dermatan sulfate) could stimulate the proliferation of murine B cells in vitro via a mechanism involving translocation of protein kinase C (PKC) β from cytosol to membrane and increased AKT phosphorylation, but not extracellular signal-regulated kinase (ERK) activation. Yang et al. (159) showed that hyaluronic acid (HA) or chondroitin sulfate A (CSA) could promote the differentiation of immature human monocyte-derived dendritic cells in vitro, suggesting that HA or CSA have potential to modulate immune responses. A significant body of literature has established a clear role for distinct molecular weight HA polymers in regulating immune responses, acting through receptors CD44 and RHAMM (reviewed in Ref. 160).

Betaglycan, also known as TGF-β type III receptor (TGFBR3), is a single-pass transmembrane PG belonging to the TGF-β coreceptor superfamily. TGF-β is an immunosuppressive cytokine that represses dendritic cell antigen-presentation and inhibits T-cell function (76). Altered TGFBR3 expression has been reported in multiple tumor types. Reduced TGFBR3 has been shown in advanced neuroblastoma, ovarian carcinoma, endometrial carcinoma, nonsmall cell lung cancer, prostate cancer, breast cancer and pancreatic cancer (77). However, in high-grade non-Hodgkin’s lymphoma and B-CLL, betaglycan was found to be upregulated, suggesting a paradoxical tumor-promoting role. Eickelberg et al. (78) showed that betaglycan can prevent the complexing of TGF-β type I and type II receptors, whose sequential activation is necessary for TGF-β signaling, in the renal epithelial cell line LLC-PK1. They determined that the GAG chains of exogenously expressed betaglycan in LLC-PK1 cells had a higher molecular weight, relative to those of the L6 cell line, whose betaglycan promotes TGF-β signaling (78). Shedding of TGFBR3 from the tumor cell surface generates soluble sTGFBR3, which sequesters TGF-β, thus antagonizing downstream TGF-β signaling. Therefore, betaglycan may exert its function as a dual modulator of TGF-β signaling: in its membrane-bound form, it enhances TGF-β signaling, whereas in its soluble form, it acts as a decoy inhibitor (79). Hanks et al. (80) showed that loss of tumor-expressed TGFBR3/sTGFBR3 activated TGF-β-dependent expression and activity of immunomodulatory enzyme indoleamine 2, 3-dioxygenase (IDO) in plasmacytoid DCs (pDCs) and CCL22 chemokine expression in myeloid DCs (mDCs). This alteration within the locoregional DC population induced infiltration of immunosuppressive FOXP3⁺ T_reg cells, correlated with a reduction in CD8⁺ T cells in the TME and suppressed tumor-associated antigen-specific T-cell responses, such as T-cell proliferation and IFNγ secretion. Finally, betaglycan may also have important roles in cell migration, irrespective of TGF-β signaling. Mythreye et al. (81) found that, in the absence of TGF-β, betaglycan was able to act through the scaffold protein β-arrestin-2 to upregulate Cdc42 and impair cell migration in the ovarian and breast cancer cell lines Ovca429 and MDA-MB231, as well as in the healthy ovarian epithelial cell line NOSE007. This effect was not related to sTGFBR3, and both the intracellular domain and GAG chains bound to extracellular moieties of betaglycan were important for this activity to occur (81).

Glypicans are HSPGs bound to the plasma membrane via a glycosylphosphatidylinositol (GPI) anchor (21). Glypicans are evolutionarily conserved and the human genome includes six distinct glypican genes (GPC-1 to GPC-6) that encode the six glypican proteins. Based on their amino acid sequence, GPC proteins can be further categorized into two families consisting of GPC-1, GPC-4 and GPC-6, and GPC-3 and GPC-5, respectively. Glypicans can regulate and interact with a wide gamut of soluble and insoluble factors such as cytokines, chemokines, growth factors, morphogens, enzymes, and receptors (82, 83). They are versatile PGs that play a dual role in either activating or inhibiting cellular signals, which may be attributed to their property to interact through their HS-GAG chains as well as their protein cores. They may act as either tumor suppressors or tumor promoters depending on the type and stage of tumor (83).

Glypican-3 is one of the most studied glypicans in human cancer and it has been proposed as a promising immunotherapy target. Glypican-3 antigenic peptide (amino acids 114–152) generates specific peptide-reactive cytotoxic T cells causing tumor regression without inducing autoimmunity (85). Wang et al. (86) showed that human dendritic cells genetically engineered to express GPC-3 were able to induce a cytotoxic T-cell-mediated antitumor response in vivo against GPC-3-expressing hepatocellular carcinoma cells (HCC). Furthermore, GPC-3 expression induces the recruitment of macrophages into human HCC tumors (87). Expression of GPC-6 positively correlated with mRNA levels of CD8A, a marker of T-cell infiltration, and improved overall survival in early stage ovarian cancer (88). Suppression of GPC-1 expression in pancreatic cancer cells rendered them resistant to recognition and lysis by NK cells through cytotoxicity receptors NKp46 and NKp30 (84).

PERICELLULAR AND BASEMENT MEMBRANE ZONE PGS—PERLECAN AND AGRIN

Perlecan or heparan sulfate proteoglycan 2 (HSPG2) is a complex modular HSPG consisting of ∼500 KDa protein core divided into five domains with homology to SEA (sea urchin sperm protein), N-CAM (neural cell adhesion molecule), IgG, LDL receptor, and laminin (21). It is a fully secreted PG that resides in the pericellular region and is a major component of the basement membrane. The majority of perlecan is produced by the cells of the reactive stroma in the TME, although it can also be produced by cancer cells (89, 90). It has been shown to be present in the desmoplastic stromal region surrounding lung, renal, prostate, and breast cancer lesions (91). Cytokines such as transforming growth factor-β (TGF-β) and tumor necrosis factor-α (TNF-α) regulate transcription levels of perlecan in normal stromal cells, tumor cells, and a subset of bone marrow stromal cells (91). In breast cancer, TGF-β was shown to positively regulate perlecan synthesis, whereas IFN-γ had a negative effect (92). Perlecan not only acts as a physical barrier but its HS-containing GAG chains sequester numerous bioactive molecules such as chemokines, cytokines, and enzymes. For example, perlecan forms a complex with FGF2 (fibroblast growth factor-2) and sequesters it in the basement membrane and stroma (93). Matrix metalloproteinases (MMPs), sulfatases, and heparanases present in the TME have been identified as the enzymes that modulate the molecular state of perlecan. For example, MMPs act on the perlecan core to release bioactive fragments, which influence cell adhesion, invasion, and angiogenesis. Sulfatases and heparanases act on the HS chains, thus releasing bound growth factors and cytokines to act on their target cells. Perlecan modifier enzymes promote degradation of the perlecan-rich stroma that may transform the TME from a “hostile” state to a “permissive” one that allows tumor growth, angiogenesis, and metastasis (reviewed in Ref. 89).

Agrin is another pericellular or basement membrane HSPG (21). It has a multimodal structure that is homologous to that of perlecan, with several splice variants. Agrin has been shown to stimulate angiogenesis in the TME by activating the VEGF-VEGFR2 pathway (94).

SMALL LEUCINE-RICH PGS—BIGLYCAN, DECORIN, AND LUMICAN

Small leucine-rich PGs (Fig. 4), encompassing 18 distinct gene products and multiple splice variants, form one of the largest families of extracellular PGs (21). They are characterized by a small protein core (36–42 kDa) and a central region composed of leucine-rich repeats (LRRs) (21, 95, 161). SLRPGs are grouped into five classes (I–V) based on several criteria such as evolutionary conservation, protein and gene homology, and chromosomal organization (21). SLRPGs bind to collagens, receptor tyrosine kinases (RTKs), TGF-β, TNF-α, and bone morphogenetic protein (BMP), thus modulating innate immune receptors and signaling pathways (161, 162). In this section, we will discuss well-studied major SLRPGs involved in regulating immune responses in the TME, such as biglycan, decorin, and lumican.

Figure 4.

Immunomodulatory roles of small leucine-rich PG (SLRPGs). Soluble biglycan and decorin bind TLR-2 and -4, resulting in production of various cytokines and chemokines, such as proinflammatory TNF-α and IL-1β, as well as anti-inflammatory IL-10. Decorin sequesters TGF-β and reduces the availability of miR-21, which results in decrease in IL-10 secretion. Biglycan can cluster TLR2/4 and purinergic receptors, P2X_4/P2X₇. APC, antigen-presenting cell; PG, proteoglycan. Created with BioRender.com.

Biglycan, a class I SLRPG bearing chondroitin sulfate GAG chains, is well known for its structural and proinflammatory functions in the ECM (96). It is overexpressed in multiple cancer types such as gastric cancer, ovarian cancer, pancreatic cancer, colon cancer, and melanoma (97, 98). Biglycan can either be secreted de novo by activated macrophages and other resident immune cells or be proteolytially freed from the ECM on tissue injury or stress. Soluble biglycan can act as a danger signal through binding Toll-like receptors (TLR)-2 and -4, resulting in production of TNF-α, IL-1β, CCL2, CXCL1, CXCL-2, CXCL-13, and CCL5 (95). Using a transient transgenic mouse model where biglycan was overexpressed by hepatocytes and released into the bloodstream, Zeng-Brouwers et al. (99) demonstrated biglycan accumulation in the kidneys caused leukocyte infiltration in the renal parenchyma concurrent with abnormal renal levels of chemoattractants CXCL1, CXCL2, CCL2, and CCL5. Furthermore, using mice deficient in either TLR adaptor proteins MyD88 or TRIF, they discovered that biglycan utilizes TLR-2/4 signaling that engages the adaptor molecule MyD88 (myeloid differentiation primary response 88), whereas it modulates T-cell infiltration exclusively using TLR-4 and the adaptor molecule TRIF (Toll/interleukin 1 R domain-containing adapter inducing interferon-β) (99, 100). In macrophages, soluble biglycan has been reported to cluster TLR2/4 and purinergic receptors, P2X_4/P2X₇, which in turn triggers the activation of caspase-1 through the NLRP3 (NOD-like receptor protein 3) inflammasome, followed by the release of IL-1β and IL-18 (101). Another study showed that binding of biglycan to CD14 and TLR2 in macrophages induces TNF-α and HSP70 release, whereas biglycan-CD14-TLR4 complex causes CCL5 (RANTES) secretion (102). During tumorigenesis, interaction of biglycan and TLR receptors stabilizes NADPH oxidases (NOX-1, -2, and 4) and modulates reactive oxygen species (ROS) production, which in turn regulates IL-1β secretion (103). Biglycan-TLR2/4-mediated ROS generation activates macrophages and dendritic cells to release CXCL13, a major chemoattractant for B cells and B1 cells (104). Thus, biglycan acts at different stages of the cancer immune cycle where it can bridge innate and adaptive immune responses.

Decorin is another class-I SLRPG with dermatan sulfate side chains that possesses well-established antitumorigenic functions (95, 106). Like biglycan, it can bind TLR2/4 in macrophages to initiate a downstream signaling and secretion of TNF-α, IL-12p70, and CCL2 (95, 107). Decorin modulates the cytokine milieu toward proinflammation by regulating TGF-β/miR-21/PDCD4 (programmed cell death protein 4) axis, to control anti-inflammatory and immunosuppressive IL-10 secretion by macrophages. Decorin reduces the availability of TGF-β-regulated oncogenic microRNA miR-21 by sequestering TGF-β. The tumor suppressor PDCD4 is translationally inhibited by miR-21 and is itself a translational repressor of IL-10. Reduced oncogenic miR-21 causes an increased PDCD4 production, which results in decreased IL-10 secretion. This pathway, through modulation of the inflammatory cytokine milieu in the TME, inhibits tumor growth.

The actions of biglycan and decorin can often be characterized as antagonistic. For example, although TGF-β induces biglycan production by stromal fibroblasts, it inhibits decorin transcription. In addition, biglycan increases the affinity of TGF-β to its receptor and this biglycan-driven TGF-β signaling creates a permissive protumor TME, unlike decorin, which promotes an antitumor TME (105).

Lumican is a member of class-II SLRPGs bearing keratan sulfate GAGs. It is characterized by the presence of 10 leucine-rich repeats in its central region (21). Although several studies ascribe an antitumor role to lumican, its actual role in a particular tumor context is determined by the specific interplay among its abundance, distribution, and tumor stage/type (108). High lumican expression correlates with poor prognosis in colorectal cancer, pancreatic cancer, and breast cancer. Lumican has been shown to participate in inflammatory signaling and EMT (97). Interestingly, lumican affects peripheral monocyte extravasation and Fas-FasL signaling, thereby modulating tumor-associated inflammation (109). It also enhances LPS-dependent TLR4 signaling (108). Zhang et al. (110) studied the role of lumican in the colon cancer microenvironment: using in silico approaches, they showed that lumican expression is positively correlated with infiltration of immune cells such as B cells, monocytes, macrophages, TAMs, DCs, CD8⁺ T cells, CD4⁺ T cells, and T_reg cells.

LARGE EXTRACELLULAR PGS—VERSICAN AND OTHER HYALECTANS

Versican (VCAN; Fig. 5) is a large matrix proteoglycan, whose actions have been extensively documented in embryonic morphogenesis, inflammation, as well as directing biochemical signaling and cell fate decisions (111). Unlike other Hyalectan (HA-binding) family members, such as aggrecan, brevican, and neurocan (21), VCAN is expressed in a wide variety of tissues throughout the body and has crucial nonredundant roles in organ development and disease (112, 113). The VCAN core protein consists of an N-terminal G1 domain, a C-terminal G3 domain, and CS chain-binding regions (113, 114). The G1 domain is composed of an immunoglobulin (Ig)-like module, followed by two hyaluronan (HA)-binding domains (link modules). VCAN’s G3 domain consists of two epidermal growth factor (EGF)-like repeats, followed by carbohydrate recognition domain (lectin-like, CRD) and complement binding protein (CBP)-like subunits (113). Depending on the number and species of GAG regions present, full-length (V0) and 3 common splice variants of VCAN have been described: V1, V2, and V3. An additional variant, V4, has been reported in breast cancer (115).

Figure 5.

Immunomodulatory roles of large extracellular PG. Tumor- or stromal-derived versican (VCAN) leads to DC dysfunction through TLR-2 activation. Stromal nonproteolyzed VCAN also promotes exclusion of CD8⁺ T cells from TME. Versikine, an N-terminal bioactive fragment (matrikine), promotes CD8⁺ influx in the TME through regulation of cDC1 abundance and activation. The balance between nonproteolyzed VCAN and versikine appears to critically influence tumor immune infiltration density. DC, dendritic cell; PG, proteoglycan. Created with BioRender.com.

Versican is of central relevance to several hallmarks of cancer (116) and plays important roles in both malignant transformation and tumor progression (117). There are various sources of VCAN production in the tumor microenvironment: tumor cells, stromal cells, tumor-associated myeloid cells, and possibly lymphoid cells in some contexts (10). Increased versican expression has been observed in a wide range of malignant tumors and has been associated with both cancer relapse and poor patient outcomes (10). Versican is a central partner in extracellular matrix (ECM) assembly and tissue homeostasis through key protein-protein or protein-carbohydrate interactions (118). VCAN engages binding partners both in the extracellular matrix, such as hyaluronan and link protein (HAPLN1, see below) through the N-terminal G1 domain (119) and on the surface of cells in the tumor microenvironment including P and L selectins (120), lymphoid tissue chemokines (121, 122), CD44 (115) and TLR2 (123, 124). VCAN, thus orchestrates a network of crucial intrinsic signals that dictate immune and inflammatory cell phenotype in the tumor microenvironment (125).

VCAN is not only produced by myeloid cells but also regulates the myeloid cell composition and regulatory networks in the tumor (98, 124). Tumor-derived VCAN leads to DC dysfunction through TLR2 activation. TLR2 ligation promotes secretion of autocrine IL-10 and IL-6 as well as sustained upregulation of the cell-surface receptors for these cytokines, which decreases the threshold for STAT3 activation. This positive feedback loop renders DCs dysfunctional and impedes downstream Th1 and cytotoxic lymphocyte (CTL) differentiation (11, 12).

VCAN plays a central role in the regulation of tumor-associated macrophages (TAMs). VCAN may function as a danger-associated molecular pattern (DAMP) molecule that interacts with Toll-like receptors (TLRs), such as TLR2 on macrophages and monocytes, to promote the production of inflammatory cytokines, including tumor necrosis factor-α (TNFα) and other proinflammatory cytokines in various tumor environments (123, 124, 126–129). In the setting of mesothelioma, tumor-derived versican promotes tumor progression by shaping an immunosuppressive milieu, mainly by impairing macrophage antitumor activities. Mice harboring versican-deficient tumors presented fewer tumor/pleural macrophages and neutrophils, and fewer pleural T-regulatory cells, compared with the control animals (130). In the 4T1 breast carcinoma model, versican expression, identified in late stages of progression, was associated with a high number of peritumoral infiltrating TAMs (131). The impact of VCAN in modulating the functional phenotypes of intratumoral macrophages was described in hematological malignancies as well. VCAN was shown to promote myeloma-associated monocytes/macrophages through TLR2/6 signaling in multiple myeloma (MM), thus triggering trophic IL-1β and IL-6 upregulation (132). In addition, the role of VCAN in loss of immune surveillance in human MM was further supported by the demonstration that myeloid-derived versican transcription was strongly associated with altered myeloid polarization toward an immunosuppressive phenotype, contraction of protective T-cell stem-like (Tcf1⁺) memory, expansion of dysfunctional/exhausted T effectors and consequently, MM progression (133). Moreover, macrophages with a unique immunosuppressive signature (expressing versican, ENTPD1, and STAB1) were associated with persistence of minimal residual disease postautologous stem cell transplant for myeloma and increased risk for relapse (134, 135).

There is a well-established inverse correlation between stromal VCAN accumulation and cytotoxic lymphocyte infiltration and/or activity in multiple human cancers. Tumor-infiltrating lymphocyte (TIL) density has been shown to predict outcomes and correlate with antitumor therapeutic responses, particularly to checkpoint inhibition and other novel immunotherapies (136, 137). VCAN accumulation was inversely correlated with intraepithelial TIL abundance in a large cohort of samples obtained from patients with colorectal cancer (138). Similar conclusions were drawn from analysis of large cohorts of breast, pancreatic, esophageal, and neuroendocrine tumors (139). Earlier work in cervical cancer was also concordant with these observations (140). A meta-analysis of RNA-Seq data by Pearce et al. (141) demonstrated that VCAN, extrapolated from a global matrisome signature, positively correlated with Th2 and T regulatory responses and negatively correlated with cytotoxic T-cell markers (142). VCAN⁺ tumor-associated macrophage (TAM) accumulation conferred a survival disadvantage in patients with cutaneous skin melanoma; these TAMs were less abundant in ICB therapy responders (143). Preliminary data from a Phase II trial showed significantly lower recurrence rates of metastatic colorectal cancers expressing low to moderate levels of VCAN or high levels of VCAN proteolysis, following neoadjuvant stereotactic body radiation therapy (SBRT) and anti-PD-1 therapy (144). Altogether, these findings suggest that VCAN accumulation impedes antitumor CD8⁺ infiltration and/or antitumor responses.

Contrary to stromal nonproteolyzed VCAN, which promotes T-cell exclusion, VCAN proteolytic fragments (matrikines) appear to possess opposing, immunostimulatory properties (13, 145). Versikine, an N-terminal VCAN-matrikine, is generated through cleavage of a Glu⁴⁴¹-Ala⁴⁴² bond (V1 enumeration) in the GAG-β domain of V1 isoform, generating a neoepitope, DPEAAE (146). This bioactive fragment promotes conventional dendritic cell type 1 (cDC1) recruitment, survival, and activation across tumor types and genetic backgrounds (13). Therefore, the balance between nonproteolyzed VCAN and versikine appears to critically influence the levels of immune infiltration. VCAN proteolysis at Glu⁴⁴¹-Ala⁴⁴² is associated with robust CD8⁺ infiltration in multiple myeloma bone marrow (147, 148) as well as solid tumors (138).

Hyaluronan and proteoglycan link protein 1 (HAPLN1) is an intriguing VCAN binding partner and structural homolog (149, 150). It has a molecular weight of 45–52 kDa and consists of a signal peptide (SP), one immunoglobulin-like (Ig), and two HA-binding link modules (proteoglycan tandem repeats, PTR: PTR1, and PTR2) (151, 152). HAPLN1, produced by bone marrow stromal cells from patients with MM, activates an atypical NF-κB pathway in MM cells that may contribute to proteasome-inhibitor drug resistance (153).

CONCLUDING REMARKS

Protective immune responses by the organism against the existential threat posed by a nascent tumor can be understood in the discrete steps of the “cancer-immunity cycle.” This cycle has been classically studied from a cell- and soluble mediator-centric vantage point but the contribution of tumor matrix signaling appears essential. Disruption of the host tissue by the developing tumor is sensed as “danger” by a sentinel network featuring specialized antigen-presenting cells that subsequently migrate from the site of damage to the regional lymph nodes to prime effector lymphocytes. Seminar work by Schaefer and others demonstrated SLRPG as essential “danger” signals activating antigen-presenting cells. The next hurdle to be overcome concerns the trafficking of effectors back to the tumor site to fight the tumor and restore homeostasis. Later work highlighted large PG-derived matrikines, such as versikine, as regulators of chemokine networks controlling immune effector cell influx. Thus, tissue barriers are breached at both the early (“danger”) and latter (“effector”) stages of the cancer-immunity cycle, and cues emanating from extracellular matrix (ECM) remodeling regulate the activity of the cellular actors at each step (163). Successful tumors that overcome these defenses potently co-opt inflammatory networks to promote their own survival and propagation. In this review, we highlighted some of the essential roles of matrix proteoglycans in regulating cancer-immune cross talk as well as opportunities for therapeutic intervention.

GRANTS

The work in the authors’ laboratory is supported by the National Institutes of Health (NIH)/National Cancer Institute (R01CA252937), the American Cancer Society (127508-RSG-15-045-01-LIB), the Leukemia and Lymphoma Society (6551-18), the UW Trillium Myeloma Fund, and the Robert J. Shillman Foundation.

DISCLOSURES

F.A. is listed as inventor on US patent US20170258898A1: “Versikine for inducing or potentiating an immune response.” None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

G.D., A.C., A.P., and F.A. conceived and designed research; G.D., A.C., A.P., and F.A. performed experiments; G.D., A.C., A.P., and F.A. analyzed data; G.D., A.C., A.P., and F.A. interpreted results of experiments; G.D., A.C., A.P., and F.A. prepared figures; G.D., A.C., A.P., and F.A. drafted manuscript; G.D., A.C., A.P., and F.A. edited and revised manuscript; G.D., A.C., A.P., and F.A. approved final version of manuscript.