Proteoglycans of basement membranes: Crucial controllers of angiogenesis, neurogenesis, and autophagy

Abstract

Anti-angiogenic therapy is an established method for the treatment of several cancers and vascular-related diseases. Most of the agents employed target the vascular endothelial growth factor A, the major cytokine stimulating angiogenesis. However, the efficacy of these treatments is limited by the onset of drug resistance. Therefore, it is of fundamental importance to better understand the mechanisms that regulate angiogenesis and the microenvironmental cues that play significant role and influence patient treatment and outcome. In this context, here we review the importance of the three basement membrane heparan sulfate proteoglycans (HSPGs), namely perlecan, agrin and collagen XVIII. These HSPGs are abundantly expressed in the vasculature and, due to their complex molecular architecture, they interact with multiple endothelial cell receptors, deeply affecting their function. Under normal conditions, these proteoglycans exert pro-angiogenic functions. However, in pathological conditions such as cancer and inflammation, extracellular matrix remodeling leads to the degradation of these large precursor molecules and the liberation of bioactive processed fragments displaying potent angiostatic activity. These unexpected functions have been demonstrated for the C-terminal fragments of perlecan and collagen XVIII, endorepellin and endostatin. These bioactive fragments can also induce autophagy in vascular endothelial cells which contributes to angiostasis. Overall, basement membrane proteoglycans deeply affect angiogenesis counterbalancing pro-angiogenic signals during tumor progression, and represent possible means to develop new prognostic biomarkers and novel therapeutic approaches for the treatment of solid tumors.

INTRODUCTION

The basement membrane (BM) is a meshwork of extracellular matrix (ECM) proteins at the periphery of the cells and it is broadly expressed in the tissues of Metazoans¹. Since its first discovery in the skeletal muscle, advances have been made in clarifying its composition, structures, interactome, genetics and function. Indeed, BMs are structural constituents of the pericellular matrix and the theater of cell/matrix interactions, especially those established between proteoglycans and proteinases or growth factors, which often play a major role in pathology^2–4.The major structural components of the basement membranes are collagen IV, laminins, nidogen, and heparan sulfate (HS) proteoglycans (HSPGs). These extracellular components are first secreted as soluble molecules and are subsequently assembled into an insoluble scaffold adhering to the cells^5;6. Embedded in the meshwork of structural proteins are matricellular proteins that provide signals to the adjacent cells influencing their behavior^7–9. Among these SPARC and nephronectin play a prominent role in tumor growth^10;11. This complex milieu also includes growth factors and cytokines which are often bound to the HS chains of proteoglycans, the only type of proteoglycans of the BM. Other important components of the BMs include collagen XV, which is primarily a chondroitin sulfate proteoglycan^12;13, and collagen XVIII present in the stromal interface and proteinases and their inhibitors, which dynamically modulate ECM remodeling. In this review we will exclusively focus on the three BM HSPGs known to directly affect developmental and cancer associated angiogenesis as well as autophagy and various genetic disorders where BMs are central to their pathophysiology.

Angiogenesis is the formation of new blood vessels from pre-existing vasculature and is a well-established hallmark of cancer¹⁴. Cancer cells in fact can trigger angiogenesis to support their growth with oxygen and nutrients. Anti-angiogenic therapy can interfere with this process and has been approved for the treatment of some tumor types in combination with conventional therapy or, more recently, immunotherapy^15;16. Angiogenesis is a tightly regulated process¹⁷ and is the result of a plethora of interconnected events promoting or inhibiting molecular pathways, including several types of proteoglycans and glycosaminoglycans such hyaluronan and HS^18–34. Notably the enzymatic complex involved in HS bisysnthesis is quite complex³⁵ and lead to the formation of substle changes in sulfation pattern that have biological effects^36;37. In addition HS and heparin are involved in the regulation of intracellular functions such as protease storage in mast cell granules ^3;38–40 and regulation of immune checkpoints^41–43. These charged glycosaminoglycans are subjected to further processing by heparanases³¹ which are directly involved in presenting growth factors to their cognate receptors. The VEGFA/vascular endothelial growth factor receptor 2 (VEGFR2) signaling axis is the most studied and the chief target for the development of anti-angiogenic drugs^44–47. Although they have been used clinically to treat a number of tumor types, their impact fell short of clinical expectations, with the survival rates increasing only few months with considerable side effects⁴⁸. Recent research has shown that increased HS synthesis in dormant cancer cells predicts poor patient prognosis and promotes recurrence by enhancing the survival of dormant residual tumor cells⁴⁹. Thus, a better understanding of the molecular mechanisms regulating this process and the discovery of novel biomarkers predicting the efficacy of anti-angiogenic therapy are clinically relevant requisites.

In this context, we will critically assess the role of perlecan, agrin and collagen XVIII in regulating angiogenesis. Foremost, these molecules are in tight contact with part of the matrix and thus can exert a deep impact on its biology in the early phases of the angiogenic program. Second, these molecules not only directly affect angiogenesis, but they also indirectly affect the bioactivity of angiogenic growth factors and cytokines as well as autophagy of vascular endothelial cells. Third, their cleavage profoundly alters the angiogenic potential of these HSPGs through the generation of active fragments suitable for the development of new prognostic markers.

The sophisticated angiogenic properties of these three BM HSPGs may substantially change depending on their remodeling^50;51. In particular, perlecan is a multimodular proteoglycan characterized by a very complex protein core which grants the possibility to modulate cellular functions also via the interaction with several other ECM molecules, growth factors and surface receptors^52–54. Together with agrin, perlecan establishes collateral interactions among the BM proteins and the receptors expressed at the cell surface of adjacent cells, associating nidogen and laminins to specific integrins, dystroglycan and sulfated glycolipids¹. Agrin is a prominent constituent of the BM expressed in various isoforms in different tissues and known to be crucial for neuromuscular junctions ⁵⁵ and cancer progression⁵⁶. Collagen XVIII, a collagen/proteoglycan hybrid harboring HS chains in the N-terminal domain and between the interrupted triple-helical collagenous sequences, is also involved in angiogenesis⁵⁷. In analogy to other HSPGs such as syndecan 2 and 4⁵⁸, the HS chains may also regulate the conformational dynamics of HSPG protein cores especially around their attachment sites. In the following sections we will discuss the specific role of each HSPG in the context of angiogenesis and tumor progression.

PERLECAN: A MULTIMODULAR AND MULTIFUNCTIONAL PROTEOGLYCAN

Perlecan is one of the giants of the ECM molecules displaying three HS chains attached to the N-terminal region of a ~470 kDa protein core^59;60 (Figure 1A), and it is encoded by a colossal gene encompassing 95–97 exons⁶¹. Like many ECM molecules, its expression is positively regulated by transforming growth factor β (TGFβ)^62;63, whereas Interferon-γ exerts an inhibitory effect⁶⁴. The “perlecan” eponym derives from the original observation using rotary shadowing electron microscopy in which, thanks to its globular domains, perlecan appears like a “string of pearls”^65;66. Originally isolated from Engelbreth-Holm-Swarm murine tumors, perlecan is widely expressed in the BM of most vascular and avascular tissues of different species^67–73 and it is also associated with the cell surface⁷⁴. Its biology and its effect on the surrounding cells is complicated by the fact that the molecule can vary to a high extent due to posttranslational modifications including attachment of three HS chains at the N-terminus, several O-linked oligosaccharides dispersed along the protein core and the possible substitution with chondroitin sulfate⁷⁵. Perlecan plays a critical role in cell invasion and adhesion, cardiovascular and skeletal muscle development, neural crest development, lipid metabolism, apoptosis, synaptogenesis, cartilage homeostasis, smooth muscle cell proliferation, epidermal and osteophyte formation, stem cell niche, and mechanobiology^76–86. A further level of complexity in the angiogenic activity of perlecan derives from the proteolytic processing of its protein core, a key post-translational modification regulating its bioactivity^87;88. Endorepellin is the C-terminal fragment of perlecan encompassing three laminin-like globular domains separated by two doublets of EGF-like repeats (Figure 1B) (see below). Genetic ablation of the Hspg2 gene, which encodes perlecan, causes embryonic lethality due to myocardial BM failure under mechanical stress which eventually leads to intrapericardial hemorrhage⁸⁹. The few surviving pups die soon after birth due to severe cephalic and cartilage defects⁹⁰. Interestingly Hspg2^−/− mice manifest abnormal position of aorta, pulmonary and coronary arteries^91;92, suggesting that perlecan plays a pivotal role in cardiovascular development. These original findings were subsequently confirmed in zebrafish where Hspg2 knockdown leads to profound cardiovascular malformations⁹³. Perlecan binds to dystroglycan for proper acetylcholinesterase clustering at the neuromuscular junction⁹⁴, and for preserving axonal and synaptic stability during Drosophila development⁹⁵. Moreover, perlecan is involved in maintaining epithelial polarization⁹⁶, heart stability⁹⁷, and for providing structural and contractile development of human cadiomyocytes⁹⁸ via the activity of multiple receptors (Figure 1C). Notably, perlecan directly affects developmental angiogenesis by interacting with the α2β1 integrin⁹⁹ and by regulating the VEGF/VEGFR2 signaling axis¹⁰⁰. Hspg2^−/− mice rescued by a construct harboring Hspg2 cDNA driven by the Col2a1 promoter, exhibit an overall increased propensity to develop aortic dissection²⁸. Genetic suppression of perlecan in keratinocytes has shown that perlecan is essential for epidermal formation regulating both the survival and terminal differentiation steps of keratinocytes¹⁰¹. Furthermore, Hspg2^−/− mice exhibit impaired bone calcium signaling and reduced anabolic response to loading¹⁰², and manifest mild muscle hypertrophy, decreased myostatin expression, and changes in muscle fiber composition¹⁰³.

FIGURE 1.

Structural domains of perlecan and functional receptors. (A) The five domains are marked in Roman numerals from the N-terminus to the C-terminus. SEA module (Sperm protein, Enterokinase and Agrin homology sequence). (B) Schematic representation of the C-terminal domain endorepellin with different proteases and their associated cleavage sites (C) Schematic representation of the various validated surface receptors involved in endorepellin angiostatic, autophagic, and mitophagic properties. For additional details see the text.

Perlecan domain III binds the extracellular α2δ1 subunits of the voltage-sensitive calcium channels, the binding site of the common painkiller and anticonvulsive drug gabapentin¹⁰⁴. Gabapentin disrupts the perlecan/α2δ1 interaction and reveals a novel mechanosensosry complex within osteocytes required for bone force transmission and sensitive to gabapentin¹⁰⁴. Another interesting observation is based on studies using a murine model of Schwartz-Jampel syndrome, where perlecan deficiency disrupts pericellular matrix formation, bulk matrix content and glycosaminoglycan deposition at the articular surface of cartilage¹⁰⁵. Overall, these data indicate that perlecan plays key functions in bone, cartilage and skeletal muscle homeostasis and mechanosensing.

Notably, anti-sense Hspg2 knockdown in endothelial cells causes abnormalities in cell surface anticoagulant ability. Specifically, when porcine endothelial cells are seeded onto 3D polymeric matrices and implanted adjacent to carotid arteries subjected to deep injury, the untransduced parent endothelial cells prevent occlusive thrombosis, in contrast to HSPG2-deficient cells which are completely ineffective¹⁰⁶. These results indicate that cell surface perlecan is required for inhibiting thrombosis after deep vascular injury, and contributes to endothelial cell-mediated inhibition of intimal hyperplasia¹⁰⁶. Not only does perlecan affect physiological vascular formation¹⁰⁷, but also cancer-associated angiogenesis. The first indication that perlecan could play a role in tumor development came from the observations that metastatic melanomas display increased deposition of the molecule in the pericellular matrix of tumor cells¹⁰⁸. Subsequently, it was found that perlecan supports high-affinity binding of fibroblast growth factor 2 (FGF2)¹⁰⁹, a well-established pro-angiogenic cytokine¹¹⁰, both to cells with poor HS content, and to soluble FGF receptor forms, thereby enhancing angiogenesis¹⁰⁹. In particular, the affinity of the binding of FGF2 to perlecan isolated from different endothelial cell types is relatively high and represents a substrate for the adhesion to vascular cells¹¹¹. In addition to a direct interaction with FGF2, perlecan promotes both the autocrine and paracrine functions of the growth factor, suggesting that perlecan may serve as a key co-receptor for FGF2¹¹². A further indication of a prominent role of perlecan in angiogenesis is supported by the observation that vessels developed in prostate carcinoma xenografts, display tumor cell-derived perlecan along the BM, suggesting that the deposition of perlecan could function as a scaffold for vessel sprouting¹¹³. Subsequently, the angiogenic properties of perlecan have been confirmed by several in vitro and in vivo studies¹¹⁴. The down-regulation of perlecan leads to decreased angiogenesis and tumor growth in colon carcinoma and melanoma preclinical models¹¹⁵, and its expression is required for strain-mediated endothelial cell growth¹¹⁶. Notably, suppression of perlecan expression in melanomas halts malignant progression by directly impacting tumor cell aggressiveness¹¹⁷ or by impairing the response to FGF2 and VEGFA¹¹⁸. However, this effect seems to be tumor type-dependent, as fibrosarcoma cells form more aggressive tumors upon perlecan knockdown¹¹⁹. Owing to its complex molecular structure, perlecan can interact with several growth factors and cytokines involved in angiogenesis, including platelet derived growth factor AA and BB^75;120, FGF1⁷⁵, FGF7 ¹²¹, FGF9⁷⁵, FGF18^122;123, hepatocyte growth factor⁷⁵, and FGF-binding protein¹²⁴. Furthermore, perlecan binds to progranulin^125–127 and extracellular matrix protein 1 involved in bone formation and angiogenesis 1^128;129, and it also regulates prostate cancer cell growth via the Sonic Hedgehog pathway^130;131. Finally, engineered short forms of perlecan carrying HS chains enhance angiogenesis by potentiating growth factor signaling¹³², and circulating levels of perlecan correlate with estrogen receptor posive breast cancer patients¹³³. Thus, perlecan and its fragments can deeply impact and finely regulate their angiogenic potential in the tumor microenvironment.

Studies using zebrafish have demonstrated that perlecan expression is required for the proper expression and localization of VEGFA, and endothelial cells exposed to perlecan display increased VEGFR2 activation⁹³. Perlecan is an intrinsic component of the VEGFA/VEGFR2 signaling axis and its downregulation in zebrafish leads to similar phenotypes observed upon the knockdown of VEGFA¹³⁴, VEGFR2¹³⁵, and phospholipase C_γ1¹³⁶ a downstream effector of the receptor. As a further indication of the interdependence between these molecules and of a positive feedback loop, microvascular endothelial cells challenged with VEGFA increase perlecan expression in a VEGFR2-dependent manner¹³⁷. The use of soluble domains of VEGFR1 and VEGFR2 to sequester VEGFA leads to vascular pruning followed by vascular rebound. Importantly, during this last phase both VEGFA and perlecan are extensively deposited along the tumor vasculature, and this correlates with increased VEGFR2 activation¹³⁸. In line with these results, the linkage of VEGFA to the ECM milieu leads to a protracted activation of VEGFR2, further suggesting that this signaling pathway is profoundly modulated by the microenvironment¹³⁹. Accordingly, HSPGs act in trans, through the trapping of the active VEGFR2 signaling complex causing sustained activation of the receptor and increased angiogenesis¹⁴⁰. Collectively, these results suggest that perlecan functions as a double-edged sword, regulating both the availability and distribution of the HS-binding cytokines, as well as their interaction and consequent activation of the cognate receptors.

It is well established that ischemic stroke causes blood-brain barrier (BBB) breakdown due to significant damage to the integrity of BBB components and perlecan is directly involved in this process. An elegant study using a transient middle cerebral artery occlusion model, found larger infarct volumes and more BBB leakage in conditional Hspg2^−/−--Transgenic mice vis-à-vis control mice¹⁴¹. In addition, Hspg2^−/−--Transgenic mice showed a decrease in pericytes, suggesting that perlecan in vivo regulates pericyte recruitment to support BBB maintenance and repair following ischemic stroke¹⁴¹. Perlecan is also involved in the maintenance of corneal epithelium as Hspg2^−/− -Transgenic mice show microphthalmos and thinner corneal epithelium vis-à-vis wild type mice¹⁴².

Transgenic mice lacking the portion of domain I where the chains are covalently attached to the protein core¹⁴³ have been utilized to investigate the potential consequences of HS-deficiency¹⁴⁴. Interestingly tumor growth is significantly impaired in these mutant mice, as well as FGF2-induced angiogenesis¹⁴⁵. Thus, the pro-angiogenic function of perlecan is exerted both through the interaction with receptors such as VEGFR2 and with the ligands likely liberated by the heparanase activity¹⁴⁶, a complex enzyme with multiple enzymatic and non-enzymatic activities^31;147–151. During morphogenesis of the salivary gland cleavage of HS chains from perlecan induces the release of FGF10¹⁵², among likely other pro-angiogenic cytokines, thus promoting vascular sprouting. Moreover, perlecan growth factor-binding domain I when conjugated with laminin-E8 fragments promotes differentiation of pluripotent stem cells ¹⁵³ and enhances maturation of grafted dopaminergic progenitor cells¹⁵⁴. Overall, these features provide a role for perlecan and its HS chains in neuronal development with properties that partially overlap with those of agrin, the other major BM HSPG (see below).

The discovery of the 85-kDa perlecan fragment known as endorepellin was fortuitous. Perlecan domain V was used as a bait in a yeast two-hybrid screen in search for new molecular partners of this HSPG. One of the strongest identified interacting proteins was the C-terminal fragment of collagen XVIII which included the anti-angiogenic molecule endostatin¹⁵⁵. Surprisingly, and contrary to the function of the parent molecule, endorepellin itself exerted potent angiostatic effects¹⁵⁵. Recombinant endorepellin inhibits endothelial cell migration through a collagen I network as efficiently as endostatin. Giving the anti-endothelial effects the C-terminal fragment of perlecan we coined the term “endorepellin”, and its activity was detected at nanomolar concentrations in angiogenic assays both in vitro and in vivo^155;156. The module involved in the interaction with endostatin is LG2, the second laminin-like globular domain of endorepellin, and the interaction seems to weaken the activity of the two molecules, although this has never been tested in vivo. Nonetheless, this interaction occurs in vivo, as demonstrated by the quantitative double labeling with immunogold performed on murine renal tissues, highlighting a consistent co-localization of perlecan and endostatin¹⁵⁷.

An intriguing observation is that endorepellin seems to play a protective role during renal development ¹⁵⁸, although this line of research has not been continued and, thus, we do not know if this of physiological significance. In summary, while the C-terminal/C-terminal interaction between perlecan and collagen XVIII may play a critical role in the assembly and maintenance of BMs, it is likely that it could also regulate the angiogenic potential of endothelial cells during sprouting angiogenesis.

Treatment of tumor bearing mice with recombinant endorepellin results in its accumulation in the perivascular zone of the tumors and leads to decreased angiogenesis, higher hypoxic levels and reduced tumor growth¹⁵⁹. In human, porcine and murine endothelial cells, endorepellin causes endothelial cell disassembly of the actin cytoskeleton and focal adhesions through the α2β1 integrin¹⁶⁰.The angiostatic function of the molecule is at least partially exerted through a dual receptor antagonism: endorepellin simultaneously engages VEGFR2¹⁶¹ and integrin α2β1, both required for proper endorepellin activity¹⁶². While LG2 is required for the interaction with endostatin, most of the endorepellin angiostatic activity is exerted via the LG3 domain which is released through the action of the bone morphogenetic protein-1 (BMP-1)/Tolloid-like family of metalloproteases (Figure 1B) and its activity requires amino acid residues involved in Ca²⁺ coordination^156;163. Notably, endorepellin enhances collagen-platelet responses via the α₂β₁ integrin¹⁶⁴. Additionally, whole perlecan and, to a lesser extent, endorepellin enhance glycoprotein VI-dependent responses and thrombus formation¹⁶⁵. It has also been reported using an ex vivo model of angiogenesis in a 3D collagen type I matrix¹⁶⁶, that endorepellin evokes an angiostatic stress signaling axis in endothelial cells¹⁶⁷. Activation of caspase-3 in endothelial cells triggers extracellular release of cathepsin L which in turn cleaves endorepellin liberating the terminal LG3¹⁶⁸. Recently, LG3 fragments have been found in apoptotic exosome-like vesicles, designated ApoExos, that are released by apoptotic endothelial cells¹⁶⁹ and generated by autolysosomes and caspase-3, and containing immunogenic bodies^170;171.

It has been reported that an intact LG3 module is generated by a specific cleavage occurring between Asn⁴¹⁹⁶ and Asp⁴¹⁹⁷ also detectable in live cells¹⁶³. Accordingly, the substitution of Asn⁴¹⁹⁶ with an Ile residue prevents the release of the active fragment which potentially occurs in different organisms as the cleavage sequence is highly conserved across species¹⁶³. Not only endothelial cells, but also mast cells secrete and process perlecan generating C-terminal fragments of the protein core including endorepellin⁴⁰. The crystal structure of the LG modules from different ECM molecules has been resolved and indicates that these units fold autonomously^172–174. Unlike endostatin, the LG3 domain of perlecan is positively charged with a predicted isoelectric point of ~5.1¹⁶³ as originally modeled on the crystal structure of the LG5 domain of the α2 chain of laminin¹⁷². These data were subsequently confirmed when the crystal structure of endorepellin LG3 was solved¹⁷⁵. The LG3 has a Ca²⁺ binding site that appears to be preformed, suggesting that the bound Ca²⁺ ion, rather than structural rearrangements, contributes to its angiostatic activity ¹⁷⁵. Given the electrostatic surface potential, LG3 has no affinity for heparin or HS, nonetheless once liberated it halts endothelial cells migration and inhibits capillary morphogenesis through the disruption of actin stress fibers, as observed with the parent endorepellin molecule¹⁵⁶.

The LG3 domain is required for the interaction with integrin α₂β₁ whereas LG1 and LG2 engage VEGFR2 thus evoking dual receptor antagonism^162;176, repressing hypoxia-inducible factor 1 (HIF-1α) and VEGFA and simultaneously inhibiting nuclear factor of activated T cells (NFAT1)¹⁶². Importantly, the release of these fragments is not restricted to cancer as it also occurs in other pathological conditions. For example, LG3 fragments have been detected in the urines of patients with renal disease¹⁷⁷, and have been implicated as novel regulators of obliterative remodeling associated with allograft vascular rejection¹⁷⁸. Release of LG3 secondary to caspase-3 activation of vascular apoptosis generates anti-LG3 autoantibodies that have an effect on both native and transplanted kidneys¹⁷⁹.

LG3 modules can be a potential biomarker of severity in IgA nephropathy¹⁸⁰, a marker of physical activity¹⁸¹, and can interact with the spike protein of SARS-CoV-2¹⁸². Moreover, LG3 can be neuroprotective in severe experimental ischemic stroke¹⁸³ and the parent perlecan itself improves blood spinal cord repair following spinal cord injury¹⁸⁴. Finally, LG3 levels can be used as serological biomarkers for breast cancer as the plasma levels of the endorepellin LG3 are significantly lower in breast cancer patients vis-à-vis healthy donors¹⁸⁵. All these studies point to the biological complexity of endorepellin and LG3 module in physiology and pathology.

AGRIN: A PROTEOGLYCAN WITH NEUROMUSCULAR AND SYNAPTIC FUNCTIONS

Agrin is an HSPG present in basement membranes of several cell types and was identified several decades ago as one of BM components associated with acetylcholine receptor (AChR) clustering¹⁸⁶ and synaptic differentiation¹⁸⁷. Agrin binds to the nerve muscle BM via laminins¹⁸⁸ as visualized by electron microscopy¹⁸⁹. Like perlecan, agrin is characterized by a multidomain structure with several alternatively spliced isoforms mostly dependent on the tissue context (Figure 2A). Indeed, the N-terminus of the molecule can be differentially spliced to generate a transmembrane agrin isoform, predominantly expressed in the brain, or a BM secreted molecule, broadly present throughout various organs⁵⁵. Agrin can be further spliced at additional sites present in the C-terminus domain, where the laminin-globular domains LG1–3 are located. The splice variants exert a profound impact on the agrin-mediated functions, such as the aggregation of AChR¹⁹⁰. Apart from the regulation of the mRNA levels, and consequently the expression levels of the molecule in the various tissues, agrin is extensively modified by post-translational modifications, such as by glycosylation and addition of the HS chains at the N-terminus and in the backbone, respectively¹⁹¹ (Figure 2A). Binding of agrin to synaptic basal lamina is mediated by the N-terminal agrin (NtA) domain which is followed by a tandem of nine Follistatin-like (FS) repeats forming a rod-like spacer to the laminin G-like domains of the molecule. Alternative splicing of the AGRN gene produces a variety of transmembrane (SN) or secreted (LN) agrin variants (Figure 2B). Neurotrypsin cleaves agrin locally at the synapse generating two C-terminal fragments of ~90 and ~22 kDa¹⁹². As agrin plays an important role in the formation and maintenance of excitatory synapses, its local synaptic processing suggests that the neurotrypsin/agrin system could regulate adaptive reorganizations of the synaptic circuitry in the context of cognitive functions, such as learning and memory.

FIGURE 2.

Structural domains of agrin and their functional receptors. (A) Schematic representation of agrin molecule. NtA, N-terminal agrin domain; TM, transmembrane domain; F, Follistatin-like domains; S/T, Serine/Threonine-rich region; LE, laminin-epidermal growth factor (EGF)-like domains; SEA, Sperm protein, Enterokinase and Agrin homology sequence; EG, EGF-like domains; LG, laminin-globular domains. (B) Schematic representation of the cleavage sites generating the agrin C-terminal functional module. (C) Cell surface receptors known to interact with agrin, its functional module, or cleaved LG3 unit. For additional details see the text.

A key role of agrin was shown by the generation of global and nerve-specific Agrn^−/− mice. These mutant animals die perinatally due to respiratory distress caused by the aberrant stimulation of the diaphragm where neuromuscular synapses fail to develop normally⁵⁵. Additionally, agrin is relevant in the formation of immunological synapses, orchestrating the organization of lipid rafts in T cells¹⁹³. Agrin plays a role in age-related processes as deficiency of skeletal muscle agrin contributes to the pathogenesis of age-associated sarcopenia¹⁹⁴. Agrin favors the interaction between stromal cells and hematopoietic stem cells, being relevant for the maturation and activation of monocytes¹⁹⁵. Thus, the critical role of agrin in the regulation of both nerve and immunological synapses suggests that this HSPG provides pivotal microenvironmental cues in the modulation of cell-cell interactions. Neuronal agrin promotes proliferation of human myoblasts in a age-dependent manner¹⁹⁶, suggesting that agrin might also be involved in the aging process as previously proposed for perlecan¹⁹⁷. Interestingly, agrin, apart from the nervous system and immune cells, is expressed by endothelial cells, being deposited along blood vessels in highly vascularized tissues, like brain, retina, lungs and kidneys¹⁹⁸. Depending on the tissue context, agrin can exert contrasting activities on the vasculature. Agrin is an essential component of the BM of the BBB, where, along with laminins, it regulates the proper connection among endothelial cells, pericytes and astrocytes¹⁹⁹. Interestingly, as discussed above, it has been reported that perlecan also supports BBB maintenance and repair following ischemic stroke¹⁴¹.

Agrin interacts directly and indirectly with multiple receptors (Figure 2C).Through its interaction with α-dystroglycan (Figure 2C), agrin regulates clustering of the aquaporin 4 channels, thus enhancing astrocytes polarization in the BBB²⁰⁰. On the other side, agrin strengthens endothelial cell junctions by enhancing proper localization of VE-cadherin and ZO-1, further reinforcing the BBB²⁰¹. Indeed, agrin LG3 module is crucial for the formation and stabilization of this complex, as LG3 binding to LRP4 β−1 domain induces oligomerization and activation of MuSK²⁰². In support of this LG3-mediated bioactivity, administration of a strong myostatin inhibitor (ActR-Fc-nLG3) a protein complex that harbors soluble activin receptor linked to the C-terminal agrin LG3 domain, increases neuromuscular junction stability and endurance in aging mice²⁰³. It has been reported that a signaling pathway involving agrin, Lrp4 and Ror2 (receptor tyrosine kinase orphan receptor 2) regulates adult hippocampal neurogenesis²⁰⁴. Antibodies against LRP4 and agrin are pathogenic in Myasthenia Gravis further supporting a central role for this complex in neuromuscular activity and homeostasis²⁰⁵

Agrin is enriched in the early wound microenvironment and acts as a mechanosensosry molecule to coordinate a competent skin wound healing²⁰⁶. Finally, agrin promotes chondrocyte homeostasis and requires both LRP4 and α-dystroglycan to enhance cartilage formation in vitro and in vivo ²⁰⁷. In line with these results, agrin-containing collagen gels cause long-term regeneration of bone and cartilage, suggesting a potential therapeutic effect of soluble agrin for joint surface regeneration²⁰⁸. In glioblastoma patients, the tumor-associated vessels, particularly in the inner core of the tumor, lose agrin due to the proteolytic activity of endothelial cell-derived MMP3^209;210. This alteration impacts on the transport of fluids through the BBB, increasing the risk of edema. On the contrary, in other pathological conditions, the expression of agrin increases significantly. Indeed, in a model of diabetic retinopathy, upon inflammatory stimuli, retinal endothelial cells display a higher expression of both agrin and collagen IV. This process associates with overexpression of the major endothelial adhesion molecules E-selectin, VCAM1 and ICAM1, which in turn enhances immune cell recruitment²¹¹. A bioinformatics-based pan-cancer cell analysis has shown that agrin expression correlates with immune cell function and response to Type I interferon²¹². Together, these findings reinforce the concept that angiogenesis and the immune response are two tightly interconnected processes²¹³.

In lung carcinomas, agrin drives cancer progression by interacting with the transmembrane receptor NOTCH1 to release its intracellular structural domain, thus activating the Notch signaling axis²¹⁴. Indeed, lung carcinoma patients with high AGRN expression are more susceptible to lymph node metastases and have a worse prognosis²¹⁴. In cholangiocarcinoma and hepatocellular carcinoma (HCC) agrin is highly expressed in tumor vessels²¹⁵. Importantly, while agrin is present at very low levels along the blood vessels of healthy liver, under cirrhosis and also in HCC this HSPG is prominently deposited along the newly-formed tumor vasculature²¹⁶, and high agrin expression correlates with HCC progression and poor prognosis²¹⁷. Further, agrin plays a role in the progression of oral squamous cell carcinomas as silencing of Agrn in orthotopic models of squamous cell carcinomas reduces tumor aggressiveness²¹⁸.

Mechanistically, through the binding to the lipoprotein receptor 4 (LRP4), an agrin enriched ECM sustains the activation of FAK, which in turn favors the acquisition of aggressive phenotypes by HCC cells, such as the expression of EMT markers²¹⁹. The agrin-mediated alterations in HCC occur through the coordination of key angiogenic pathways influencing endothelial cell behavior²²⁰. At an early tumor stage agrin supports endothelial cell proliferation and migration, thus promoting the formation of new blood vessels. This potentially occurs through the engagement of the integrin β1-LRP4-MuSK complex, which in turn boosts FAK activation²²⁰. The engagement of this axis enhances the stabilization of VEGFR2 during tumor angiogenesis by promoting the downstream activation of ERK1/2, AKT and eNOS, further sustaining tumor angiogenesis²²⁰. In addition, the alterations induced by the knockdown of agrin in endothelial cells also affect the homing of HCC cancer cell, likely affecting tumor progression and invasion²²⁰. In agreement with these findings, a single dose of agrin is capable of reducing ischemia-reperfusion injury and ameliorating cardiac function²²¹ suggesting that agrin could be a novel additional therapeutic agent to improve cardiac failure following infarct. However, we should point out that a recent report utilizing an endothelial cell-specific agrin knockout has shown that agrin is not required for developmental vasculogenesis as the mutant mice develop normal vascular barrier integrity and vasoreactivity in the adult stage²²². Moreover, the growth of both primary and metastatic tumor cells was not affected by endothelial cell-specific ablation of the Agrn gene²²².

Overall, the role of agrin in endothelial cells seems to be multifaceted and context dependent. On one side, in glioblastoma agrin preserves the integrity of the BBB, thus preventing the dangerous accumulation of interstitial fluids. On the contrary, in HCC agrin facilitates a more permissive tumor microenvironment, enhancing tumor angiogenesis and promoting cancer cell aggressiveness. Importantly, agrin can be produced by several types of cells in the tumor stroma as well as by the tumor cells²²³, and its levels and impact on tumor progression vary depending on the mutual activation of the various actors. Although there is no scientific data concerning the C-terminal activity of agrin, we envision that, given the structural homology to perlecan LGs, agrin LGs may play a similar angiostatic role.

COLLAGEN XVIII: A HYBRID HERPARAN SULFATE PROTEOGLYCAN WITH ANGIOSTATIC ACTIVITY

Collagen XVIII belongs to a subfamily of evolutionarily conserved and widely expressed non-fibrillar BM-associated collagens known as multiplexins (multiple triple-helix domains with interruptions) (Figure 3A)²²⁴. The high expression of collagen XVIII makes it one of the most abundant components of the epithelial and vascular BMs²²⁴. The collagen XVIII gene is localized to chromosome 21 and harbors 43 exons and 2 promoters²²⁵. Collagen XVIII plays a critical role in the maintenance of retinal structure and in neural tube closure²²⁶. Furthermore, mutations in the COL18A1 gene cause Knobloch syndrome, a disease characterized by severe ocular and brain malformations²²⁷. The ubiquitous collagen XVIII is expressed as three isoforms, named short, medium and long, differing in their N-terminal non-collagenous (NC) domain and in tissue distribution^228;229. Collagen XVIII has a trimeric structure displaying a central area of homologous α1 chains and contains ten collagenous regions interspersed with eleven non-collagenous domains, and at least three heparan sulfate chains²³⁰ (Figure 3A). Collagen XVIII is a true member of the HSPG family, unlike its homologue collagen XV, which harbors chondroitin sulfate chains¹³. Of note, collagen XV is involved in various disease mechanisms²³¹. For example, Col15a1^−/− mice exhibit a complex cardiac phenotype and a predisposition to stress-induced cardiac responses²³², whereas the same mutant mice show neuroprotective features after ischemic stroke²³³. Collagen XV is also expressed in various BM zones^234;235 where it plays key roles in cancer microenvironment^236;237, cell adhesion²³⁸, osteogenesis²³⁹,leukocyte influx²⁴⁰, developing motor axons²⁴¹, and in slow muscle precursors which deposit a collagen XV-enriched matrix that guides motor axon navigation²⁴². Of note, recent data in Drosophila suggest that collagen XV, via an interaction with the discoidin domain receptor, regulates ensheathment in larval nerves²⁴³.

FIGURE 3.

Structural domains of collagen XVIII and their functional receptors. (A) Schematic representation of the structural domains of collagen XVIII. (B) Schematic representation of the different structures that constitute the C-terminal fragment of collagen XVIII, with indication of the cleavage sites used by several proteases. (C) Cell surface receptors known to facilitate endostatin anti-angiogenic and pro-autophagic activities. For additional details see the text.

Collagen XVIII harbors a globular TSP-1 domain which is homologous to the N-terminal domain of Thrombospondin-1, whose function still remains elusive²³⁵. In this context, overexpression of Tsp-1 domain affects eye growth and cataract formation²⁴⁴. Due to alternative splicing, collagen XVIII can also contain a cysteine-rich domain homologous to the Drosophila frizzled domain and/or an acidic segment A. One of the first indications that collagen XVIII was a negative regulator of angiogenesis came from ex vivo experiments employing aortic explants from Col18a1^−/− mice, which showed that loss of collagen XVIII expression associates with increased vessel sprouts²⁴⁵. On the same line, Col18a1 ablation increases both vascularization and vessel permeability in atherosclerosis²⁴⁶ and causes early cerebral vessel disease²⁴⁷. As vascular permeability deeply affects the delivery and thus the efficacy of various drugs to the tumors as well as the intra- and extra-vasation of tumor and immune cells, altered expression of collagen XVIII could markedly impact patient outcome. Indeed, high levels of collagen XVIII correlate with poor clinical response in breast cancer patients²⁴⁸, and can represent a potential prognostic indicator in non-small cell lung carcinomas²⁴⁹. An opposite effect was found in the retinas of Col18a1^−/− mice where ocular vessels display impaired vascularization²²⁶. It is possible that the effect may depend on the tissue-specific microenvironment or differences between physiological or pathological conditions. Nonetheless, these results support a prominent role of the parent collagen XVIII in angiogenesis.

More recently additional biological roles for collagen XVIII have been discovered. Mice lacking either all three collagen XVIII isoforms, or specifically, the medium and long isoforms develop insulin resistance and glucose intolerance²⁵⁰. Moreover, Col18a1^−/− mice subjected to high fat diet show a significant increase in serum triglyceride values and an increased hepatic steatosis vis-à-vis with controls²⁵⁰. These data indicate that collagen XVIII may play a role in the regulation of glucose tolerance, insulin sensitivity and lipid homeostasis, primarily through its ability to regulate adipose tissue expansion. Another novel role for collagen XVIII has been reported to occur during renal development. Specifically, this proteoglycan is involved in regulating the renal epithelial tree patterning via its N-terminal domain, common to all the isoforms ²⁵¹.

Like perlecan, the processing of collagen XVIII gives rise to C-terminal domains with angiostatic activity. The breakthrough discovery of endostatin as a potent angiogenesis inhibitor (Figure 3B)²⁵² has opened a new path towards the discovery novel endogenous angiogenesis inhibitors of potential clinical use. The basic idea is that these natural angiostatic molecules would be “safe” and less likely to evoke detrimental side effects. Specifically, endostatin is a 22-kDa protein fragment generated by the cleavage of collagen XVIII in the NC1 domain which is target of many matrix metalloproteinases as well as cathepsin and elastases^253–256 (Figure 3B). Again, there is a common theme in the proteolytic processing as both perlecan and collagen XVIII can generate endorepellin/LG3 or endostatin via cathepsin, respectively. Unlike endorepellin, endostatin is a heparin-binding molecule²⁵⁷. Of note, an endostatin-like fragment derived from collagen XV has been found in circulating blood ²⁵⁸ and was named restin²⁵⁹ for its ability to inhibit angiogenesis^259;260. However, very little has been published after their initial identification and there is published evidence that restin does not affect in vivo tumorigenicity of cervical carcinoma²⁶¹.

Crystallographic analyses of endostatin show a cluster of 11 Arg residues creating a large positively charged patch that may serve as a binding site for heparin and HS²⁵⁷, and mutations in this region abrogate the angiostatic function of the molecule²⁶². However, endostatin does not require the binding to heparin to inhibit the migration of endothelial cells²⁶³, and thus may halt angiogenesis in a HS -independent manner. Besides integrin α₅β₁, endostatin also engages integrin α_vβ₃^264-267 (Figure 3C) with both integrins playing a major role in angiogenesis.

Endostatin inhibits the migration, proliferation, and viability exclusively of vascular endothelial cells leaving non-vascular cells unharmed^252;263;268. In analogy to some of the biological functions of agrin and perlecan, the NC1/endostatin domain of collagen XVII can regulate affect axonal guidance and neuromuscular junctions in C. Elegans^269;270. Thus, collagen XVIII is another example of BM HSPGs regulating synapse organization. In support of this role in the nervous system, it has been reported that collagen XVIII is associated with vascular amyloid depositions and senile plaques in Alzheimer’s disease brains²⁷¹. Indeed, recombinant endostatin can generate amyloid fibrils that bind and are cytotoxic to neuroblastoma cells²⁷², and can also induce endothelial cell detachment²⁷³.

Tumor progression is associated with a significant decrease in the expression of the endostatin in human hepatocellular carcinomas ²⁷⁴, although circulating levels of endostatin are increased in pancreatic cancer patients and then normalized after surgical removal of the primary pancreatic cancer²⁷⁵. There are several proposed mechanisms of action reflecting the complexity of the repertoire of the endostatin interactome. For example, a possible mechanism through which endostatin impairs angiogenesis is through the inactivation of metalloproteinases²⁷⁶, whose activation is required to allow the migration and proper sprouting of endothelial cells²⁷⁷. Also, it has been proposed that endostatin may block VEGF-mediated signaling via direct interaction with VEGFR2²⁷⁸. A very interesting and novel finding is the discovery that endostatin can be produced by a large extracellular proteasome 20S present in various murine and human tumor cells but not in untransformed cells²⁷⁹, a process that can be enhanced by CD147-evoked effects on MMP-9 and secreted proteasome 20S²⁸⁰. As circulating proteasome 20S is emerging as a chief indicator of tumor progression, the possibility of proteasomes controlling the production of angiostatic endostatin and perhaps endorepellin should be contemplated.

Treatment with recombinant endostatin induces tumor cell quiescence without overt signs of drug resistance even when the injections are protracted for long periods of time²⁸¹. Moreover, treatment of tumor-bearing mice with near infrared-labeled endostatin has shown that the molecule targets the tumor and tumor-associated vasculature, suggesting that the antiangiogenic effect is exerted via a local rather than a systemic mechanism²⁸². Interestingly, near infrared-labeled decorin, another proteoglycan with angiostatic properties^283–294, also specifically localizes to the tumor and its vasculature when systemically delivered to tumor-bearing mice²⁹⁵.

The engagement of endostatin with α5β1 integrin leads to downregulation of the FAK/Ras/p38MAPK/ERK signaling pathway and the consequent inhibition of the expression of HIF-α/VEGFA²⁹⁶. In addition, endostatin by simultaneously engaging α5β1 and caveolin-1 alters the endothelial cell cytoskeleton via the activation of Src²⁹⁷ and promotes actin stress fiber disassembly through the decrease of RhoA²⁹⁸. Furthermore, endostatin can also inhibit Wnt signaling leading to β-catenin degradation, thus impairing cell cycle progression via cyclin D1 inhibition²⁹⁹. A modified recombinant form of endostatin, named Endostar, can also block the Wnt/β-catenin pathway by suppressing the expression of T cell factor transcriptional activity and angiogenesis³⁰⁰. The inhibitory effects on endothelial cell function are also exerted through the downregulation of the VEGFR2/VEGFA signaling axis which occurs via a direct binding to VEGFR2 and through the decrease of VEGFA expression^278;301. Finally, it has been proposed that cell surface glypicans (Figure 3C) are low-affinity endostatin receptors³⁰². Overall, the mechanism of action of endostatin is likely more complex as a genome-wide expression profiling indicates that endostatin affects the expression of a plethora of genes involved in angiogenesis³⁰³. Clinical trials evaluating the efficacy of endostatin or its mutated form Endostar are still ongoing and benefits have been observed in the treatment of different tumors utilizing new delivery strategies³⁰⁴ or combinatorial treatments^305–310. Endostar can synergize with radiotherapy to inhibit the growth of cervical and non-small cell carcinomas³¹¹, and can also block Lewis lung carcinomas when combined with antibodies targeting the programmed cell death protein 1 (PD-1)³¹². As a functional link between angiogenesis and inflammation has been well established³¹³, the use of endostatin in combination with immunotherapy may represent a new promising approach for cancer treatment. An important conceptual advance that emerged from these studies is the idea that a molecule embedded in the basement membrane can give rise, via partial proteolysis, to soluble fragments that function in a paracrine/endocrine manner, deeply impinging on the biology of adjacent cells. In line with this concept, soluble NC1 fragments containing endostatin can be detected in a number of human and murine organs²⁵⁵ and may serve as endogenous regulators of angiogenesis. The variable concentration of these circulating soluble factors, as well as their degradation, which may vary depending on the expression of proteases in the different tissues, likely impact not only on the regulation of angiogenesis but also on the efficacy of anti-angiogenic therapies.

AUTOPHAGY: NEW ROLES FOR THE C-TERMINAL HSPG PROCESSED MODULES

A new area of research in the past few years is the realization that proteoglycans and their processed forms can activate intracellular catabolic pathways such as canonical autophagy and mitophagy^78;314–318. The first observation that a proteoglycan could evoke autophagy was discovered serendipitously in endothelial cells by following the fate of paternally expressed gene 3 (Peg3), an imprinted tumor suppressor gene³¹⁹. Specifically, decorin, a small leucine-rich proteoglycan ^53;320 involved in controlling angiogenesis and RTK activities^{285;295;321–326} evoked the synthesis Peg3 and its relocation into Beclin 1⁺ and LC3⁺ autophagosomes in both microvascular and macrovascular endothelial cells leading to suppression of angiogenesis³¹⁹. Decorin activates AMPK³²⁷ and evokes transcription of Beclin 1 and LC3 via Peg3, thereby leading to a protracted autophagic program^{317;328–330} and also mitophagy^78;331–333. Soon thereafter, it was discovered that endorepellin also causes autophagy in vascular endothelial cells^334;335. We should point out another common theme: the parent perlecan functions in an anti-autophagic manner³³⁶. For instance, perlecan helps in maintaining muscle mass by suppressing autophagy^336–338. As mentioned above, endorepellin functions as a dual receptor antagonist to the α2β1 integrin receptor and VEGFR2^{16;77;336;337}. This dual receptor antagonism results in the down regulation of the VEGFR2 axis as well as the suppression of the α2β1-integrin signaling pathway, ultimately causing the inhibition of VEGFA transcription as well as the disassembly of actin stress fibers and focal adhesion, respectively¹⁶. We discovered that endorepellin binds to the ectodomain of VEGFR2 via the two proximal LG modules LG1 and LG2 (Figure 4A)^16;78;334. This binding site only partially overlaps with the major binding site of VEGFR2 natural ligand VEGFA^16;77. Although VEGFA and endorepellin interact with VEGFR2 at separate binding pockets (Ig_2–3 and Ig_3–5 respectively), endorepellin competes with VEGFA and acts as an allosteric inhibitor to suppress endothelial cell migration and angiogenesis^34;78 by preventing downstream signaling initiated by phosphorylation of Tyr₁₁₇₅ in VEGFR2³³⁴. Moreover, endorepellin activates protein tyrosine phosphatase SHP-1³³⁹, an enzyme capable of dephosphorylating VEGFR2 at Tyr₁₁₇₅. This dephosphorylation induces the physical dissociation and translocation of the Src homology 2 (SH2) domain that contains the adaptor protein (Shb) to the endothelial nucleus and the release of PLCγ from VEGFR2. This, in turn, impairs the signaling along the PI3k/Akt/mTOR and PKC/JNK/Ap-1 pathways^16;34;77. The loss of PLCγ and Shb binding prevents the activation of three VEGFA pro-angiogenic pathways: the PLCγ/calcineurin/RACK1/NFAT1 pathway, the PLCγ/PI3K/PDK1/AKT/mTOR pathway and the Shb/PKC/KNK/AP1 pathway⁷⁸. The blocking of these three pathways ultimately results the downregulation of NFAT1, HIF1A and AP1 leading to angiogenic inhibition⁷⁸.

FIGURE 4.

A comprehensive view of angiostatic, pro-autophagic, and pro-mitophagic signaling pathways exploited by endorepellin or endostatin in vascular endothelial cells. (A) The endorepellin LG1 and LG2 domains interact with VEGFR2 leading to mitophagy and autophagy, whereas the LG3 domain interacts with α2β1 integrin leading to actin disruption and angiostasis. (B) Endostatin-evoked autophagy and angiostasis in endothelial cells by interacting with the α5β1 integrin and by preventing binding of VEGFA to VEGFR2 respectively. For additional details see the text.

Overall, the binding of endorepellin to the VEGFR2 ectodomain results in the attenuation of the two main VEGFR2 signaling axes. This eliminates the suppression of autophagy that is induced by the mammalian target of rapamycin (mTOR), a potent autophagic repressor (Figure 4A)^16;34. Endorepellin-evoked signaling via VEGFR2 induces the dephosphorylation of the receptor via activation of SHP-1 causing the suppression of downstream proangiogenic effectors⁷⁷.

Notably, LG1 and LG2 are capable of evoking autophagy via VEGFR2, independently of LG3 engagement with the α2β1 integrin. Furthermore, only the LG1/LG2 domains can evoke autophagy via VEGFR2^{16;77;334;337}. In endothelial cells, endorepellin is a pro-autophagic agent that, through transient non-canonical phosphorylation of VEGFR2 at Tyr₁₁₇₅, evokes adenosine monophosphate kinase (AMPKα)-dependent induction of the microtubule-associated light chain protein 3 (LC3), Beclin 1 and p62^16;78;336. This causes inhibition of the mTOR pathway and the induction of paternally expressed gene 3 (PEG3) and the formation of large autophagosomes (Figure 4A)^{78;317;336;337}. PEG3 is a master regulator of autophagy that is associated with increased levels of LC3-positive autophagosomes^78;334. Notably, PEG3 interacts with Beclin 1 and LC3 to assist in autophagosome formation^78;336. In the absence of PEG3 endorepellin loses its ability to transcriptionally induce autophagy and fails to increase Beclin 1 and LC3 levels^77;78. Under normal physiological conditions PLCγ and the adapter protein Shb are bound to VEGFR2 at Tyr₁₁₇₅ and, once phosphorylated, they activate the PI3K/Akt/mTOR and PKC/JNK/Ap-1 pathways⁷⁷. However, endorepellin forces a redistribution of class III PI3K and Vps34 from the plasma membrane to large intracytoplasmic Beclin1-positive autophagosomes^16;77. Normally, Vps34 can be found at the plasma membrane but when autophagy is initiated Vps34 is mobilized into phagophores, the initial formation stage of autophagosomes^77;78. Overall, endorepellin causes the co-localization of PEG3 with both key autophagic markers LC3-II and Beclin 1⁷⁷. However, this process can be blocked by small molecule inhibitors of VEGFR2 tyrosine kinase¹⁶. In addition, endorepellin induces mitophagy (Figure 4A)³³⁶. This endorepellin-evoked pro-mitophagic transcriptome leads to an enhanced expression of mitostatin and Parkin inducing mitochondrial depolarization³³⁶, and ultimately resulting in reduced vessel growth.

Located near the N-terminus of LG3 is the potential binding site for the α2β1 integrin⁷⁷. This interactions induces the disassembly of the actin cytoskeleton and focal adhesions, thereby suppressing endothelial cell migration (Figure 4A)^16;77;334, and inhibiting autophagic gene expression ^16;336. In addition, engagement of α₂β₁ integrin by LG3 induces the recruitment of SHP-1 tyrosine phosphatase on its cytoplasmic tail (Figure 4A)⁷⁸. Notably, there are opposing functions within a single proteoglycan structure to regulate catabolic levels needed to maintain normal cell homeostasis³³⁶. The LG1 and LG2 domains induce the expression of BCN1, PEG3 and MAPLC3A genes. In contrast, the LG3 domain suppresses the levels of all these genes⁷⁷. However, LG1 and LG2 signaling carry a greater signaling potential and can override the signaling pathway and effects of LG3 and α2β1⁷⁷. Within 5–10 minutes of binding to both VEGFR2 and α2β1, endorepellin downregulates and allosterically inhibits VEGFA-induced angiogenesis^34;77;78. However, cells that lack α2β1 but express VEGFR2 are incapable of interacting with or responding to endorepellin³⁴⁰. Finally, endorepellin has been shown to evoke autophagic degradation of hyaluronan synthase 2 (HAS2) in endothelial cells³⁴¹. As HAS2 is the main enzyme capable of producing large amounts of pro-angiogenic and pro-tumorigenic HA^342;343 and central to tumor progression in breast cancer³⁴⁴, this biological activity offers a novel mechanism to regulate in vivo angiogenesis.

The role of HSPGs in controlling autophagy, lipid metabolism and neurodegenerative pathologies is another emerging field of research³⁴⁵. During the development of Drosophila neuromuscular junctions, muscle-specific knockdown of HS biosynthetic machinery evokes autophagy³⁴⁶. In addition, HS production is required for maintaining proper autophagic processes in the fat body, the Drosophila main energy storage and nutritional sensing organ. In support of these data, HS affects autophagy, lifespan, and responses to oxidative stress in Drosophila Parkin³⁴⁷. Finally, there is evidence that HS-degrading enzymes such as heparanase can affect autophagy³¹. Indeed, the non-enzymatic form of heparanase enhances gastric tumor proliferation via TFEB-dependent autophagy³⁴⁸. Together, these studies establish that HS and heparanase are critical regulator of autophagy, a biological process necessary for proper assembly of postsynaptic membrane specialization³⁴⁶.

Like endorepellin, endostatin can also evokes autophagy in endothelial cells^349;350 and can also induce autophagic cell death³⁵¹. Endostatin-evoked autophagy can occur through binding to the α5β1 integrin and via the downregulation of the Wnt/β catenin pathway^16;78. When β-catenin levels are decreased, endostatin increases autophagic flux rate, resulting in higher Beclin1 levels and autophagic vacuoles in endothelial cells¹⁶. This results in the dissociation of the Beclin1/Bcl-2 complex and in a pro-autophagic stimulus¹⁶. Ultimately, endostatin induces autophagy via integrin α5β1, while simultaneously controlling the levels of Beclin 1 and β-catenin to induce the downregulation of the anti-apoptotic protein Bcl-2, an inhibitor of autophagy and apoptosis, and MAP kinase activities^16;336. However, if β-catenin levels are increased there will be a downregulation of Beclin 1 in the presence of endostatin^16;336. Overall, there are three major functions of endostatin which include: the inhibition of the p38-MAPK/ERK signaling cascade through α5β1 binding, the suppression of HIF-1α/VEGFA and Wnt signaling-dependent downregulation of β-catenin to suppress oncogenes like cyclin D1 and Myc, and the disassembly of actin via Src-dependent- p190RhoGAP activation (Figure 4B)¹⁶.

Uninhibited by endostatin, the VEGFA/VEGFR2 complex is categorized by phosphorylation of Tyr₁₁₇₅, and ultimately, cell proliferation and migration pathways by which VEGFR2 induces angiogenesis. VEGFA binding to VEGFR2 induces activation of phospholipase C-γ (PLCγ) via phosphorylation of Tyr₁₁₇₅. Activated PLCγ hydrolyzes phosphatidylinositol (4,5)-bisphosphate (PIP2) to produce inositol 1,4,5-triphosphate (IP3) and diacylglycerol (DAG), a physiological activator of PKC. This allows VEGFR2 to capitalize upon ERK in the PKC/Raf/MEK/ERK signaling axis, sending ERK to the nucleus, binding transcription factors, inducing gene expression, and ultimately contributing to cell proliferation ³⁵². Cell migration is another crucial aspect of the VEGFA/VEGFR2 signaling pathway inhibited by endostatin. Cell migration is normally supported by VEGFR2 through phosphorylation of Tyr₁₁₇₅ which regulates the SHB/FAK/paxillin and the SHB/PI3K/Rac pathways, or through phosphorylation of Tyr₁₂₁₄, which regulates the NCK/p38/MAPKAPK2/3 pathway (Figure 4B) ³⁵².

In addition, endostatin can be chemically modified with low molecular weight heparin (LMWH) to exhibit higher activity and have a better heat tolerance in comparison to native endostatin¹⁶. Treatment with endostatin containing an RGD (Asn-Gly-Arg) tripeptide at its N-terminus leads to impaired tumor growth and associates with pronounced angiostatic activity¹⁶. Also, there are several recombinant forms of endostatin that possess enhanced angiostatic activity and improved stability¹⁶. Studies have shown that the application of endostar can inhibit the progression and the pathological process of breast cancer while concurrently reducing the expression of pro-angiogenesis molecules in tumor-bearing mice³⁵³. Furthermore, P125A is a mutated form of endostatin that exhibits a higher binding affinity to endothelial cells with enhanced autophagic and antiangiogenic effects^16;78;354. P125A evokes autophagy in a Vps34/Beclin 1/LC3 dependent manner³⁵⁴. In fact, the systematic administration of endostatin and its modified forms can suppress several types of malignant tumors such as fibrosarcomas, melanomas, and hemangioendotheliomas¹⁶.

PERSPECTIVES AND UNUNSWERED QUESTIONS

We have provided a global assessment of the respective roles of three BM HSPGs and their C-terminal modules and have contextualized their roles in angiogenesis and autophagy. Besides their role in protecting the brain from potential injuries, perlecan, agrin and collagen XVIII also actively regulate neurogenesis and axonal migration. This is not surprising considering that neurons extend their axons toward their synaptic targets and shape functional neural circuits often using the same molecular pathways exploited by endothelial cells during sprouting. An important question is why only HSPGs are intrinsic BM constituents. We do not have a clear answer for this, but it is quite surprising that all known BMs contain at least one of the HSPGs covered here. A potential explanation is that their HS chains act as a reservoir of growth factors highly needed during development, tissue regeneration and angiogenesis. A common feature of BM HSPGs is their large size and the position of the HS chains, often located at the N-termini. Moreover, the C-terminal domains contain bioactive protein modules that carry biological activity and bind various integrins and surface receptors.

The BM HSPGs play a pivotal role in the regulation of angiogenesis, finely tuning and aiding the process of tissue vascularization. However, during cancer progression BM HSPGs contribute to extensive ECM remodeling and the discharge of bioactive fragments with opposite effects to their parent molecules. Each HSPG and its respective C-terminal module have unique relationships with one another through which they exert control over pathological development of blood supply in the tumor microenvironment, with these relationships being closely modulated by ECM remodeling within the tumor microenvironment.

Overall, proteoglycans have the unique ability to either induce or inhibit angiogenesis and autophagy in the tumor parenchyma and surrounding stromal cells. Most importantly, endorepellin and endostatin exercise dual receptor antagonism to promote angiostatic or pro-autophagic activity. Although autophagy is upregulated upon nutrient deprivation and unfavorable conditions, autophagic regulation by matrix factors like proteoglycans is a nutrient-independent mechanism. HS chains themselves can also regulate autophagy. However, the precise molecular mechanism through which HS blocks autophagy needs to be fully elucidated. Moreover, we do not know which HSPG is responsible for this activity, nor do we know which specific receptor is responsible for modulating this intracellular catabolic event.

While perlecan, endorepellin, collagen XVIII, and endostatin have been well studied, the latter of which has already undergone promising clinical trials, many questions remain unanswered as the field strives towards maximizing the therapeutic potential of these molecules. Much remains to be determined about agrin, particularly where glioblastoma is concerned, as a relationship has been identified between this HSPG and the disease, a relationship that remains to be capitalized upon. The observation that loss of agrin in the tumor microenvironment is associated with glioblastoma complications suggests that agrin should be further studied to determine its potential as a therapeutic target in the context of this lethal disease. Overall, agrin diverse and context-dependent roles render further exploration necessary, particularly focusing on agrin laminin globular units, which may provide insight into the differential organ system-dependent role of agrin.

The clinical trials performed with Endostar have shown exciting potential with limited adverse side effects following its administration. However, further investigations are required to improve Endostar as an option for cancer treatment. Two distinct drawbacks have been identified with Endostar, the first being its relative instability, making it difficult to store and produce on a large scale, and the other being the high concentration of the drug needed to produce beneficial effects for the patients. Deeper exploration of collagen XVIII and the endostatin molecule are needed to resolve these issues. Additionally, the vast majority of trials conducted with Endostar have been restricted to a relatively small patient population in China, with all participants being categorized as having late-staged disease. An expansion of these trials both geographically as well as to participants with cancer in earlier developmental stages are needed to validate the initial findings of these trials³⁵⁵.

Endorepellin, having similar angiostatic effects, may also serve as a promising therapeutic candidate, but several questions must be answered before this can be confirmed. First, greater in vivo exploration of endorepellin effects must be achieved to determine possible adverse effects of its use as a treatment option. While it has been shown that endostatin effect is confined to the vasculature, similar determinations regarding the interaction between endorepellin and the microenvironment must be made. Practical questions concerning endorepellin stability and manufacturability also remain. It would be interesting to generate clinical studies using a combination of endorepellin and endostatin potentially targeting two distinct integrins, α2β1 and α5β1, respectively.

Provided that these endogenous fragments do not induce undesirable side effects, much effort should be put in the near future to exploit this reservoir of anti-angiogenic molecules and create new approaches to improve cancer treatment. This could be achieved, in part by modifying these bioactive fragments to ameliorate their stability and activity, but also by attempting combinatorial treatments in association with conventional chemotherapy and immunotherapy. In addition, these fragments may also be exploited for the development of much needed predictive biomarkers for patient-tailored treatments.

In conclusion, we have critically assessed the essential functions of three BM HSPGs and their C-terminal fragments as they relate to angiogenesis and autophagy, as well as providing early reports of these molecules’ functions in other physiological processes relating to cancer growth, development and neurogenesis. These molecules reveal themselves to be clear controllers of the tumor microenvironment and the need for further investment into the exploration of how we might exploit them to develop novel therapies is proven by the promising results that already exist with the endostar project. These complex and highly conserved proteoglycans provide a clear lens through which we may make novel discoveries within the field of matrix biology in new body systems, and ultimately provide newer, safer, and more reliable therapeutic options to the global patient population.

DATA AVAILABILITY STATEMENT

Data sharing is not applicable to this article as no new data were created or analyzed in this study.