NG2 Proteoglycan Enhances Brain Tumor Progression by Promoting Beta-1 Integrin Activation in Both Cis and Trans Orientations

cancers

Review NG2 Proteoglycan Enhances Brain Tumor Progression by Promoting Beta-1 Integrin Activation in both Cis and Trans Orientations

William B. Stallcup

Sanford Burnham Prebys Medical Discovery Institute, Cancer Center, Tumor Microenvironment and Cancer Immunology Program, 10901 North Torrey Pines Road, La Jolla, CA 92037, USA; [email protected]; Tel.: +1-858-646-3100

Academic Editor: Helen M. Sheldrake Received: 28 February 2017; Accepted: 29 March 2017; Published: 31 March 2017

Abstract: By physically interacting with beta-1 integrins, the NG2 proteoglycan enhances activation of the integrin heterodimers. In glioma cells, co-localization of NG2 and α3β1 integrin in individual cells (cis interaction) can be demonstrated by immunolabeling, and the NG2-integrin interaction can be conﬁrmed by co-immunoprecipitation. NG2-dependent integrin activation is detected via use of conformationally sensitive monoclonal antibodies that reveal the activated state of the beta-1 subunit in NG2-positive versus NG2-negative cells. NG2-dependent activation of beta-1 integrins triggers downstream activation of FAK and PI3K/Akt signaling, resulting in increased glioma cell proliferation, motility, and survival. Similar NG2-dependent cis activation of beta-1 integrins occurs in microvascular pericytes, leading to enhanced proliferation and motility of these vascular cells. Surprisingly, pericyte NG2 is also able to promote beta-1 integrin activation in closely apposed endothelial cells (trans interaction). Enhanced beta-1 signaling in endothelial cells promotes endothelial maturation by inducing the formation of endothelial junctions, resulting in increased barrier function of the endothelium and increased basal lamina assembly. NG2-dependent beta-1 integrin signaling is therefore important for tumor progression by virtue of its affects not only on the tumor cells themselves, but also on the maturation and function of tumor blood vessels.

Keywords: brain tumor progression; brain tumor vascularization; NG2 proteoglycan; beta-1 integrin; pericyte; endothelial cell; macrophage; tumor vessel function; Cre-Lox gene ablation

1. Introduction The membrane-spanning NG2 proteoglycan (Figure1), also known as chondroitin sulfate proteoglycan-4 (CSPG4), has only limited signal transducing capability of its own [1,2]. However, the ability of NG2 to promote signaling through β1 integrins [3–9] and through receptor tyrosine kinases [10–15] makes the proteoglycan a signiﬁcant accessory contributor to transmembrane signaling. Studies with a number of different immature or progenitor cell types indicate that NG2-enhanced signaling processes promote cell proliferation, motility, and survival, all of which are valuable properties for progenitor cells [16,17]. These characteristics would also be expected to beneﬁt tumor progression, and indeed, expression of NG2 is correlated with tumor malignancy and poor patient outcome in several types of cancer [3,18–26]. NG2 knockdown impairs tumor progression [3], and in the one instance where it has been tested, forced expression of NG2 in melanoma cells has been shown to enhance tumor growth [27]. In this respect, most research has been focused on NG2 expression in the case of gliomas and melanomas, since NG2 is often found on the tumor cells themselves in these neoplasms. However, because NG2 is also expressed by several elements of the tumor stroma, including macrophages and microvascular pericytes [28], NG2 expression is also relevant to many

Cancers 2017, 9, 31; doi:10.3390/cancers9040031 www.mdpi.com/journal/cancers Cancers 2017, 9, 31 2 of 15 other cancers in which the tumor cells are NG2-negative. Although the properties of tumor cells are Cancers 2017, 9, 31 2 of 15 obviously important in dictating malignancy, the impact of host stromal elements is proving to be surprisinglyAlthough powerful the properties in determining of tumor cells the are aggressiveness obviously important of tumorin dictating growth malignancy, and spread. the impact Therefore, in orderof to host appreciate stromal elements the full is impactproving to of be NG2 surprisingly on tumor powerful progression, in determining it is the important aggressivene toss understand of tumor growth and spread. Therefore, in order to appreciate the full impact of NG2 on tumor mechanisms by which the proteoglycan exerts it effects not only on tumor cells, but also on various progression, it is important to understand mechanisms by which the proteoglycan exerts it effects types of hostnot only cells. on tumor In this cells, review, but also we on will various focus types on theof host NG2-dependent cells. In this review, activation we will offocusβ1 on integrins the as a mechanismNG2 that-dependent impacts activation the contributions of 1 integrins of both as a mechanismtumor and that stromal impacts cells the to contributions tumor progression. of both A key aspect oftumor this researchand stromal has cells been to tumor the ﬁndingprogression. that A NG2key aspect is capable of this research of activating has beenβ the1 integrins finding that when the two moleculesNG2 is capable are expressed of activating in the 1 integrins same cell when (cis the interaction), two molecules as are well expressed as when in the the same two cell molecules (cis are interaction), as well as when the two molecules are expressed by two different cells that are closely expressed by two different cells that are closely apposed (trans interaction). apposed (trans interaction).

Figure 1. Domain structure of NG2. NG2 is a type-1 transmembrane protein with several distinct Figure 1.structuralDomain dom structureains. Domain of NG2.1. Bold magenta NG2 is bars a type-1 = laminin transmembrane G domains; S-S = proteindisulfide withbonds. several Domain distinct structural2. domains.Bold yellowDomain bar = collagen 1. Bold binding magenta domain; bars Irregular = laminin red Gline domains; = chondroitin S-S =sulfate disulﬁde chain bonds.attachedDomain 2. Bold yellowat S- bar999. =Domain collagen 3. Blue binding Y shapes domain; = N-linked Irregular oligosaccharides; red line = chondroitin Orange arrowheads sulfate chain = sites attached of at S-999. Domainproteolytic 3. Blue cleavage. Y shapes Transmembrane = N-linked domain oligosaccharides;. Double green lines Orange = plasma arrowheads membrane. = sitesCytoplasmic of proteolytic cleavage.domainTransmembrane. Blue circles = domain sites of. threonine Double phosphorylationgreen lines = plasma at T-2256 membrane. and T-2314; GreenCytoplasmic circle = domain. C-terminal PDZ-binding motif; Gray grid lines = proline-rich segment. Figure taken from [17]. Blue circles = sites of threonine phosphorylation at T-2256 and T-2314; Green circle = C-terminal PDZ-binding2. Cis Interaction motif; Grayof NG2 grid with lines 1 =Integrins proline-rich segment. Figure taken from [17]. When expressed in the same cell, NG2 and 1 integrins co-localize and interact physically, 2. Cis Interactionpromoting ofan NG2 activated with conformationβ1 Integrins for the integrin that has important consequences for cell behavior. These cis interactions occur in both tumor and stromal cells. When expressed in the same cell, NG2 and β1 integrins co-localize and interact physically, promoting2.1. an NG2 activated in Glioma conformation Cells for the integrin that has important consequences for cell behavior. These cis interactionsUsing detergent occur extractsin both oftumor the NG2and- stromalexpressing cells. U87 human glioma cell line, we have demonstrated a physical interaction between NG2 and 31 integrin by showing that the integrin 2.1. NG2co in-immunoprecipitates Glioma Cells with NG2. Similar results were obtained with the NG2-negative human UsingU251 detergent glioma cell line extracts following of expression the NG2-expressing of full-length rat U87 NG2 human[3]. Although glioma glycosaminoglycan cell line, we have chains are often important for proteoglycan function [29,30], we have not detected a role for the α β demonstratedsingle NGa physical2 chondroitin interaction sulfate chainbetween in NG2 NG2 interaction and 3 with1 integrin integrins by, showing suggesting that that the the integrin co-immunoprecipitatesinteraction is protein with-protein NG2. in Similarnature. We results also used were U251 obtained cells to investigate with the changes NG2-negative in cell biology human U251 glioma cellthat line are followingcaused by forced expression expression of full-length of several key rat NG2 NG2 variants [3]. Although [7,8]. These glycosaminoglycan single amino acid chains are often important for proteoglycan function [29,30], we have not detected a role for the single NG2 chondroitin sulfate chain in NG2 interaction with integrins, suggesting that the interaction is protein-protein in nature. We also used U251 cells to investigate changes in cell biology that are caused by forced expression of several key NG2 variants [7,8]. These single amino acid variants are relevant to the phosphorylation status of the NG2 cytoplasmic domain, which can be modiﬁed Cancers 2017, 9, 31 3 of 15

Cancers 2017, 9, 31 3 of 15 at Thr-2256 by protein kinase C-alpha (PKCα) and at Thr-2314 by the proline-dependent kinase ERK. Thevariants T2256E are relevant NG2 variant to the phosphorylation mimics the PKC statusα-phosphorylated of the NG2 cytoplasmic state ofdomain, NG2, which while can the be T2314E  variantmodified mimics at the Thr ERK-phosphorylated-2256 by protein kinase state. C-alpha Conversely, (PKC ) and the atT2256V Thr-2314 and by the T2314V proline variants-dependent represent kinase ERK. The T2256E NG2 variant mimics the PKC-phosphorylated state of NG2, while the non-phosphorylatable forms of NG2. When expressed in U251 cells, all of these forms of NG2 T2314E variant mimics the ERK-phosphorylated state. Conversely, the T2256V and T2314V variants co-localizerepresent with nonα3-phosphorylatableβ1 integrin, consistent forms of withNG2. ourWhen evidence expressed that in U251 the cells, interaction all of these between forms of the two moleculesNG2 isco- mediatedlocalize with by 3 the1 integrin, NG2 ectodomain consistent with rather our evidence than the that cytoplasmic the interaction domain. between However,the there aretwo two molecules distinct is mediated sites of NG2- by theα 3NG2β1 co-localization,ectodomain rather which than the are cytoplasmic dictated by domain. the phosphorylation However, state ofthere the are proteoglycan two distinct (Figuresites of NG22). When-31 co phosphorylated-localization, which at are Thr-2314, dictated NG2by the and phosphorylation its α3β1 binding partnersta localizete of the toproteoglycan arrays of microprotrusions(Figure 2). When phosphorylated on the apical at cell Thr surface-2314, NG2 of U251and its cells 31 (Figure binding2 A–D). partner localize to arrays of microprotrusions on the apical cell surface of U251 cells (Figure 2A–D). The phosphomimetic T2314E NG2 variant is also highly concentrated in these microprotrusions, The phosphomimetic T2314E NG2 variant is also highly concentrated in these microprotrusions, while thewhile non-phosphorylatable the non-phosphorylatable T2314V T2314V variant variant failsfails to to occupy occupy these these arrays. arrays. Electron Electron microscopy microscopy of of immune-goldimmune- labeledgold labeled cells cells reveals reveals these these NG2-rich NG2-rich structures to to resemble resemble microvilli microvilli averaging averaging about about 2 µm in2 lengthm in [ length7]. Very [7] similar. Very similar NG2-rich NG2 microprotrusions-rich microprotrusions are seen are onseen the on apical the apical surfaces surfaces of melanoma of cells [31melanoma]. In contrast, cells [31] when. In phosphorylated contrast, when phosphorylated at Thr-2256, NG2 at Thr and-2256, its NG2 associated and its α associated3β1 partner 3 localize1 to broadpartner lamellipodia localize to(Figure broad2 E–H). lamellipodia The phosphomimetic (Figure 2E–H). The T2256E phosphomime variant alsotic T2256E spontaneously variant also localizes to lamellipodia,spontaneously while localizes the T2256V to lamellipodia, variant fails while to do the so [ T2256V8]. In wound variant healing fails to (scratch) do so [8] assays. In wound performed healing (scratch) assays performed in vitro, it is apparent that these NG2-rich lamellipodia represent in vitro, it is apparent that these NG2-rich lamellipodia represent the leading edges of motile cells [7]. the leading edges of motile cells [7].

Figure 2. NG2 and 31 integrin co-localization on U251 glioma cells. NG2-transfected U251 cells Figurewere 2. NG2 double and immunolabeledα3β1 integrin with co-localization rabbit anti-NG2 on U251 (A,C,E glioma,G) and cells. mouse NG2-transfected monoclonal antibody U251 cells were doubleagainstimmunolabeled either 1 (B,F) or with3 (D rabbit,H) integrin. anti-NG2 (A–D (A) U251,C,E, Gcells) and expressing mouse monoclonalthe T2314E NG2 antibody variant. against either βNG21 (B and,F) or3α31 (Dco,-Hlocalize) integrin. to microprotrusions (A–D) U251 cells from expressing the apical the cell T2314E surface. NG2(E–H variant.) U251 cells NG2 and α3β1 co-localizeexpressing the to microprotrusions T2256E NG2 variant. from NG2 the and apical31 co cell-localize surface. to prominent (E–H) U251 lamellipodia cells expressing on the the T2256Eleading NG2 variant.cell edge. NG2Bar inand (D) =α 53 βm.1 co-localize to prominent lamellipodia on the leading cell edge. Bar in (D) = 5 µm. We hypothesize that the site of NG2 phosphorylation determines the ability of its cytoplasmic domain to interact with known intracellular scaffolding partners such as ezrin, MUPP1, GRIP1, and Wesyntenin hypothesize-1 [7,8,32 that–34] the. In site turn, of NG2these phosphorylation cytoplasmic interactions determines dictate the the ability localization of its cytoplasmicof the domainNG2 to- interact31 complex with to knownspecific membrane intracellular microdomains scaffolding in apical partners microprotrusions such as ezrin, or lamellipodia. MUPP1, GRIP1, and syntenin-1The functional [7,8 ,importance32–34]. In of turn, these theselocalization cytoplasmic patterns interactionsbecomes apparent dictate when the we localization compare the of the cell biology of the NG2 variants. While motility cannot be stimulated in T2256V variants, the T2256E NG2-α3β1 complex to speciﬁc membrane microdomains in apical microprotrusions or lamellipodia. species is spontaneously motile, with NG2-positive lamellipodia forming the leading edges of The functionalmigrating importance cells. Increased of thesemotility localization is also seen patterns in wild type becomes NG2 transfectantsapparent when in which we compare PKC the cell biology of the NG2 variants. While motility cannot be stimulated in T2256V variants, the

T2256E species is spontaneously motile, with NG2-positive lamellipodia forming the leading edges of migrating cells. Increased motility is also seen in wild type NG2 transfectants in which PKCα activity is stimulated by treatment with phorbol, 12-myristate, 13-acetate (PMA). In contrast, T2314E variants Cancers 2017, 9, 31 4 of 15 exhibitCancers spontaneously 2017, 9, 31 increased proliferation, a property also seen in wild type NG2 transfectants4 of 15 in whichactivity ERK is isstimulated directly by stimulated treatment wit byh phorbol, the upstream 12-myristate, presence 13-acetate of constitutively (PMA). In contrast, active T2314E MEK-DD. Predictably,variants proliferation exhibit spontaneously cannot be increased stimulated proliferation, in T2314V a property variants. also The seen importance in wild type of NG2α3 β1 in mediatingtransfectants either motility in which or ERK proliferation is directly stimulated is indicated by bythe theupstream ability presence of anti- βof1 constitutively antibody to active inhibit both processes.MEK In-DD. addition, Predictably, immunolabeling proliferation cannot with be the stimulated conformationally-sensitive in T2314V variants. The HUTS-21 importance monoclonal of antibody3 [135 in] revealsmediating that either the motilityβ1 integrin or proliferation subunit is is highly indicated activated by the ability in association of anti-1 with antibody NG2 to in both T2256Einhibit lamellipodia both processes. and in In TT2314E addition, microprotrusionsimmunolabeling with [8 ].the Interestingly, conformationally motility-sensitive and HUTS proliferation-21 monoclonal antibody [35] reveals that the 1 integrin subunit is highly activated in association with appear to be mutually exclusive. Spontaneously motile T2256E variants are very poorly proliferative, NG2 in both T2256E lamellipodia and in TT2314E microprotrusions [8]. Interestingly, motility and while spontaneouslyproliferation appear proliferative to be mutually T2314E exclusive. variants Spontaneously exhibit verymotile low T2256E motility. variants We are proposevery poorly that the phosphorylationproliferative, statewhile of spontaneously NG2, determined proliferative by the T2314E balance variants of PKC exhibitα and very ERK low activitiesmotility. We that are controlledpropose by the that cellular the phosphorylation environment, state is ofan NG2, important determined aspect by of the the balance cell’s of choice PKC to and proliferate ERK or migrate.activities As a that result are ofcontrolled its phosphorylation, by the cellular environment, the anchorage is an important pattern ofaspect NG2 of the is ablecell’s tochoice direct to α3β1 localizationproliferate to speciﬁc or migrate. membrane As a result microdomains of its phosphorylation, where it canthe anchorage interact with pattern machinery of NG2 is that able promotes to   either motilitydirect 3 (lamellipodia)1 localization to orspecific proliferation membrane (microprotrusions). microdomains where A it representation can interact with of machinery these localized that promotes either motility (lamellipodia) or proliferation (microprotrusions). A representation of signaling mechanisms is shown in Figure3. these localized signaling mechanisms is shown in Figure 3.

Figure 3. Spatial localization of functional NG2 interactions. On the apical cell surface, the T-2314 Figure 3. Spatial localization of functional NG2 interactions. On the apical cell surface, the T-2314 phosphorylated NG2 proteoglycan (green oval) activates 31 integrin (yellow dimer) signaling to phosphorylatedpromote enhanced NG2 proteoglycan proliferation. (green NG2 also oval) promotes activates cell α proliferation3β1 integrin via (yellow potentiation dimer) of growth signaling to promotefactor/growth enhanced factor proliferation. receptor signaling NG2 also (indicated promotes here cell by proliferation red FGF and viablue potentiation FGFR dimer). of On growth factor/growthleading edge factor lamellipodia, receptor signaling T-2256 phosphorylated (indicated here NG2 by activates red FGF 3 and1 integrin blue FGFRsignaling dimer). to promote On leading edge lamellipodia,enhanced motility. T-2256 NG2 phosphorylated-dependent integrin NG2 activates signalingα 3 canβ1 alsointegrin enhance signaling cell survival to promote via enhanced the motility.PI3K/AKT NG2-dependent pathway. integrin In addition, signaling NG2 can provides also enhance a linkage cell between survival the via cell the surface PI3K/AKT and the pathway. In addition,extracellular NG2 matrixprovides via a its linkage interaction between with type the cell VI collagen surface (triple and the turquoise extracellular rods). Signaling matrix via its pathways are shown in abbreviated format due to space consideration. Figure taken from [17]. interaction with type VI collagen (triple turquoise rods). Signaling pathways are shown in abbreviated formatAn dueother to spaceimportant consideration. consequence Figure of takenNG2-dependent from [17]. 1 integrin activation is an increase in PI3K/AKT survival signaling [3]. Use of the conformationally-sensitive HUTS-21 antibody reveals Anotherthat 1 integrin important activation consequence occurs in of NG2 NG2-dependent-expressing U87 gliomaβ1 integrin cells in activation similar fashion is an to increase that in PI3K/AKTobserved survival in NG2 signaling-transfected [3]. U251 Use ofcells. the In conformationally-sensitive both glioma cell lines, 1 activation HUTS-21 is accompanied antibody reveals by that increased phosphorylation of AKT. This NG2-dependent survival signaling via the PI3/AKT axis β1 integrin activation occurs in NG2-expressing U87 glioma cells in similar fashion to that observed results in increased glioma cell resistance to treatment with chemotherapeutic drugs, including β in NG2-transfectedcarboplatin, cisplatin, U251 cells. doxorubicin, In both and glioma TNF cell. This lines, drug1 resistance activation is is absent accompanied in NG2-negative by increased phosphorylationparental U251 of cells AKT. and This in NG2 NG2-dependent-knockdown U87 survival cells. The signaling involvement via of the PI3K/AKT PI3/AKT signaling axis results is in increasedindicated glioma by cell increased resistance sensitivity to treatment to drug with treatment chemotherapeutic in the presence drugs, of the including PI3K inhibitor carboplatin, cisplatin, doxorubicin, and TNFα. This drug resistance is absent in NG2-negative parental U251 cells and in NG2-knockdown U87 cells. The involvement of PI3K/AKT signaling is indicated by increased sensitivity to drug treatment in the presence of the PI3K inhibitor wortmannin. The importance of upstream β1 integrin activation in drug resistance is demonstrated by the ability of anti-β1 antibodies Cancers 2017, 9, 31 5 of 15 Cancers 2017, 9, 31 5 of 15 wortmannin. The importance of upstream 1 integrin activation in drug resistance is demonstrated by the ability of anti-1 antibodies to increase drug sensitivity. The relevance of these findings to gliomato increase growth drug in sensitivity. vivo is emphasized The relevance by of the these reduc ﬁndingsed growth to glioma of NG2 growth-knockdownin vivo is U87 emphasized tumors comparedby the reduced to control growth NG2 of-positive NG2-knockdown U87 tumors, U87 likely tumors due compared to diminished to control cell proliferation NG2-positive in U87 the knockdowntumors, likely tumors. due to In diminished addition, the cell effect proliferation of NG2 expression in the knockdown on drug tumors.resistance In is addition, demonstrated the effect by anof NG2even expressionfurther reductio on drugn in resistance growth of is the demonstrated NG2-knockdown by an eventumors further as a result reduction of administration in growth of the of α TNFNG2-knockdown [3]. In contrast, tumors growth as a result of control,of administration NG2-positive of TNF U87 [3 tumors]. In contrast, is unaffected growth ofby control, TNF α administrationNG2-positive U87 (Figure tumors 4). is unaffected by TNF administration (Figure4).

Figure 4. Effect of NG2 knockdown on growth of U87MG gliomas. Cohorts of 5 NOD-SCID mice Figure 4. Effect of NG2 knockdown on growth of U87MG gliomas. Cohorts of 5 NOD-SCID mice were innoculated subcutaneously with 5 × 106 U87MG glioma cells that had been transduced with were innoculated subcutaneously with 5 × 106 U87MG glioma cells that had been transduced with lentivirus (LV) for expression of either control shRNA or NG2 shRNA. NG2 expression was reduced lentivirus (LV) for expression of either control shRNA or NG2 shRNA. NG2 expression was reduced by 75% in the NG2 shRNA group compared to the control shRNA group. One cohort of mice from by 75% in the NG2 shRNA group compared to the control shRNA group. One cohort of mice from each tumor group also received IP injections of TNF (150 g/kg) once daily on days 3–10 following each tumor group also received IP injections of TNFα (150 µg/kg) once daily on days 3–10 following tumor cell injections. A. Final tumor sizes in the 4 cohorts of mice. B. Kinetics of tumor growth in the tumor cell injections. A. Final tumor sizes in the 4 cohorts of mice. B. Kinetics of tumor growth in the 4 cohorts of mice. NG2 NG2 knockdown knockdown significantly signiﬁcantly slows slows U87MG U87MG tumor tumor growth, growth, likely likely due due to to loss loss of of NG2NG2-dependent-dependent cell cell proliferation. The The NG2 NG2 knockdown knockdown tumors tumors exhibit exhibit an an additional sensitivity to  treatmenttreatment with with TNF α (lacking in in the control tumors) due to loss of NG2NG2 pro-survivalpro-survival effects.effects. Data takentaken from [3]. [3].

An intriguing, but untested, possibility related to glioma cell survival is suggested by the An intriguing, but untested, possibility related to glioma cell survival is suggested by the finding finding that expression of 61 integrin by glioma stem cells is important for interaction of these that expression of α6β1 integrin by glioma stem cells is important for interaction of these cells with cells with laminin in the perivascular niche [36]. This niche enhances survival and helps maintain laminin in the perivascular niche [36]. This niche enhances survival and helps maintain stemness in stemness in this population. A cis interaction of NG2 with 61 on the glioma stem cell surface this population. A cis interaction of NG2 with α6β1 on the glioma stem cell surface might be important might be important for promoting activation of the integrin to enhance its binding to laminin in the for promoting activation of the integrin to enhance its binding to laminin in the perivascular niche. perivascular niche. 2.2. NG2 in Pericytes 2.2. NG2 in Pericytes Vascularization is a key aspect of tumor progression, with the vascular switch serving as the point Vascularization is a key aspect of tumor progression, with the vascular switch serving as the at which tumors have recruited sufficient vasculature to expand beyond a very small volume [37,38]. point at which tumors have recruited sufficient vasculature to expand beyond a very small volume The interaction between endothelial cells and smooth muscle-like pericytes is critical for microvessel [37,38]. The interaction between endothelial cells and smooth muscle-like pericytes is critical for formation, maturation, maintenance, and remodeling [39,40]. Our ability to identify and study microvessel formation, maturation, maintenance, and remodeling [39,40]. Our ability to identify and pericytes at an early point in their participation in microvessel development has been much improved study pericytes at an early point in their participation in microvessel development has been much by using NG2 as a marker for immature pericytes [41–43]. In addition, our work with germline NG2 improved by using NG2 as a marker for immature pericytes [41–43]. In addition, our work with null mice [12] has indicated that NG2-deficient pericytes are restricted in their ability to support germline NG2 null mice [12] has indicated that NG2-deficient pericytes are restricted in their ability neovascularization. This is emphasized in retinal and corneal models of neovascularization, which to support neovascularization. This is emphasized in retinal and corneal models of reveal reduced recruitment and participation of NG2 null pericytes in the formation of new blood neovascularization, which reveal reduced recruitment and participation of NG2 null pericytes in the vessels [10,44]. formation of new blood vessels [10,44]. The importance of NG2-dependent β1 integrin activation in pericyte recruitment has been The importance of NG2-dependent 1 integrin activation in pericyte recruitment has been demonstrated more clearly by in vitro studies using human brain microvascular pericytes [9]. demonstrated more clearly by in vitro studies using human brain microvascular pericytes [9]. siRNA-mediated knockdown of NG2 in pericytes leads to a 60% decrease in β1 integrin activation siRNA-mediated knockdown of NG2 in pericytes leads to a 60% decrease in 1 integrin activation

Cancers 2017, 9, 31 6 of 15

Cancers 2017, 9, 31 6 of 15 (judged by immunolabeling with the conformationally-sensitive HUTS-21 antibody), further resulting(judged in by a 40%immunolabeling reduction in with phosphorylation the conformationally of the downstream-sensitive HUTS effector-21 antibody),FAK. This decreasefurther inresulting NG2-dependent in a 40%β reduction1/FAK signaling in phosphorylation has an important of the impact downstream on pericyte effector behavior. FAK. This NG2 decrease knockdown in pericytesNG2-dependent exhibit a1/FAK 4-fold signaling decrease has in proliferationan important andimpact a 3-fold on pericyte decrease behavior. in PDGF-BB NG2 knockdown stimulated migration.pericytes These exhibit changes a 4-fold in decrease pericyte in biology proliferation are likely and to a underlie 3-fold decrease the deficits in PDGF in recruitment-BB stimulated of NG2 nullmigration. pericytes These seen changes in retinal in and pericyte corneal biology angiogenesis are likely modelsto underlie [10, 44the]. deficits in recruitment of NG2 null pericytes seen in retinal and corneal angiogenesis models [10,44]. 2.3. NG2 in Macrophages 2.3. NG2 in Macrophages Preliminary data suggest that cis interactions between NG2 and integrins may also be important for macrophagePreliminary recruitment data suggest to tumors. that Wecis generated interactions myeloid-specific between NG2 NG2 and null integrins (Mac-NG2ko) may also mice be by crossingimportant NG2 for floxed macrophage mice [45] with recruitment LysM-Cre to mice tumors. [46, 47 We].Growth generated of intracranial myeloid-specific B16F10 NG2 melanomas null is(Mac greatly-NG2ko) retarded mice in by Mac-NG2ko crossing NG2 mice floxed compared mice to[45] tumor with growth LysM-Cre in control mice [46,47] mice (Figure. Growth5A,B), of a phenomenonintracranial B16F10 that correlates melanomas with is greatly significantly retarded reduced in Mac accumulation-NG2ko mice ofcompared macrophages to tumor in tumorsgrowth in thein NG2 control null mice mice (Figure [28,48 5].A,B), In the a phenomenon absence of large that numberscorrelates of with tumor significantly macrophages, reduced vascularization accumulation of of macrophages in tumors in the NG2 null mice [28,48]. In the absence of large numbers of tumor tumors in Mac-NG2ko mice is greatly impaired, possibly due to loss of macrophage-derived factors macrophages, vascularization of tumors in Mac-NG2ko mice is greatly impaired, possibly due to that stimulate tumor vascularization [49,50]. In vitro studies with human THP1 macrophages reveal loss of macrophage-derived factors that stimulate tumor vascularization [49,50]. In vitro studies with that NG2 knockdown results in loss of activation of both β1 and β2 integrins in these cells [28], as human THP1 macrophages reveal that NG2 knockdown results in loss of activation of both 1 and judged by immunolabeling with the conformationally sensitive HUTS-21 and mAb-24 [51] antibodies, 2 integrins in these cells [28], as judged by immunolabeling with the conformationally sensitive respectively. We hypothesize that NG2-dependent activation of macrophage α4β1 and αMβ2 may be HUTS-21 and mAb-24 [51] antibodies, respectively. We hypothesize that NG2-dependent activation important for binding of these integrins to VCAM1 and ICAM1 expressed by endothelial cells [52–55]. of macrophage 41 and M2 may be important for binding of these integrins to VCAM1 and TheseICAM1 interactions expressed are by key endothelial aspects of cells myeloid [52–55] cell. These arrest interactions and subsequent are key transmigration aspects of myeloid across cell the vasculararrest and endothelium. subsequent Ablation transmigration of NG2 across might the weaken vascular these endothelium. interactions Ablation to an extent of NG2 that impairsmight macrophageweaken these extravasation interactions from to an the extent circulation that impairs into tumors. macrophage Confirmation extravasation of these from possibilities the circulation requires additionalinto tumors. experimentation. Confirmation of these possibilities requires additional experimentation.

Figure 5. Decreased brain tumor progression in myeloid and pericyte-specific NG2 null mice. Figure 5. Decreased brain tumor progression in myeloid and pericyte-speciﬁc NG2 null mice. Although Although primary melanomas arise mostly in the skin, the brain is a common site for melanoma primary melanomas arise mostly in the skin, the brain is a common site for melanoma metastasis [56]. metastasis [56]. We used intracranial B16F10 melanomas as a model of melanoma growth in the We used intracranial B16F10 melanomas as a model of melanoma growth in the brain. B16F10 tumors brain. B16F10 tumors in control versus Mac-NG2ko mice (A) and in control versus PC-NG2ko mice in control versus Mac-NG2ko mice (A) and in control versus PC-NG2ko mice (C) were evaluated at (C) were evaluated at 10 days after tumor initiation via intracranial microinjection (2 × 104 cells per 10 days after tumor initiation via intracranial microinjection (2 × 104 cells per mouse). Graphs quantify mouse). Graphs quantify tumor volumes in control versus Mac-NG2ko mice (B) and in control tumor volumes in control versus Mac-NG2ko mice (B) and in control versus PC-NG2ko mice (D). Data versus PC-NG2ko mice (D). Data in B represent 21 control and 20 Mac-NG2ko mice. Data in D in B represent 21 control and 20 Mac-NG2ko mice. Data in D represent 22 control and 18 PC-NG2ko represent 22 control and 18 PC-NG2ko mice. *p < 0.01 compared to controls. Data taken from [48]. mice. *p < 0.01 compared to controls. Data taken from [48].

Cancers 2017, 9, 31 7 of 15

3. Trans Interactions of NG2 with β1 Integrins In addition to its ability to activate β1 integrins in a cis orientation, NG2 also appears able to activate β1 in a trans orientation when the two molecules are expressed on two different closely apposed cells. While this was initially surprising to us, this phenomenon has a precedent in the case of homotypic interactions between cadherin molecules, which can occur in both cis and trans orientations [57,58]. Interactions between Eph receptors and ephrins are also observed to occur in both cis and trans orientations [59]. The intimate interaction between pericytes and endothelial cells in microvessels provides a good example of trans interaction between NG2 and β1 integrin.

3.1. Effects of Soluble NG2 on Endothelial Cells An understudied aspect of NG2 biology is the proteolytic shedding of the proteoglycan’s ectodomain from cell surfaces [60–63] (see also Figure1). While neither the precise mechanism underlying NG2 shedding [64] nor the functional significance of shedding [65,66] are well-understood, it is significant that the shed ectodomain can substitute for membrane-bound NG2 in supporting certain signaling functions, including potentiation of growth factor signaling [10]. Extending this line of investigation, we found that soluble, recombinant NG2 ectodomain [67] is effective in promoting angiogenesis in a corneal vascularization assay [4]. This action of NG2 is based on its ability to stimulate endothelial cell motility, morphogenesis, and tube formation, as demonstrated using in vitro assays. We subsequently identified α3β1 integrin and galectin-3 as NG2 binding partners on the endothelial cell surface, and demonstrated the ability of soluble NG2 to activate β1 integrin in both mouse and human endothelial cells [4] via use of the conformationally sensitive β1 antibodies 9EG7 (mouse) [68] and HUTS-21 (human). These studies suggest the possibility that NG2 shed from pericytes might be able to act at a distance to aid in recruiting endothelial cells to sites of angiogenesis.

3.2. Effects of Pericyte NG2 on Endothelial Cells Our studies on the effects of soluble NG2 on endothelial cells further suggested the possibility that membrane-bound NG2 on pericyte surfaces could be involved in pericyte interaction with endothelial cells. This was tested in vitro using human pericyte and endothelial cells grown on opposite sides of transwell membranes with 0.4 µm pores [9]. These pores are too small to allow the passage of cells [69], but are large enough to allow contact between pericyte and endothelial cell processes, generating an “in-contact” model of pericyte-endothelial cell interaction (Figure6A). A “non-contact” model is generated by growing endothelial cells on the membrane and pericytes on the bottom of the chamber Figure6B). Compared to non-contact co-cultures, the presence of control, NG2-positive pericytes in the in-contact model improves the barrier function of the endothelial monolayer, as judged by decreased leakage of FITC-dextran from the upper to lower chamber Figure6C,D). This decreased permeability is due to the formation of ZO-1 positive endothelial junctions in the endothelial monolayer under the inﬂuence of interactions with pericytes (Figure6E–H). Increased endothelial junction formation and improved barrier function are not seen in the non-contact model or when NG2 knockdown pericytes are used in the assay, showing that cell-cell contact and NG2 expression are required for the effect of pericytes on endothelial cells. The effect of NG2 is based on NG2-dependent activation of β1 integrin signaling in the endothelial monolayer, as shown by immunolabeling with the conformationally-sensitive HUTS-21 antibody [9]. This ability to activate endothelial β1 integrin is lacking in NG2 knockdown pericytes (Figure6I,J). Cancers 2017, 9, 31 8 of 15 Cancers 2017, 9, 31 8 of 15

FigureFigure 6. 6. NG2NG2 knockdown knockdown in pericytes pericytes reduces integrin activation activation and and formation formation of of endothelial endothelial junctions.junctions. Pericyte-endothelial Pericyte-endothelial cell cell interactions interactions were were investigated investigated using in-contact using in (-Acontact) and non-contact (A) and (Bnon) Transwell-contact ( co-cultures.B) Transwell HUVEC: co-cultures. human HUVEC: umbilical human vein umbilical endothelial vein cells. endothelial HBVP: cells human. HBVP: brain vascularhuman pericytes. brain vascular NG2 knockdown pericytes. NG2 in pericytes knockdown reveals in that pericytes NG2 expression reveals that is required NG2 expression for enhanced is barrierrequired function for enhanced (reduced barrier leakage function of FITC-dextran (reduced from leakage upper of FITCto lower-dextran chamber) from by upper the endothelial to lower monolayerchamber) by in the the endothelial in-contact modelmonolayer (C) butin the not in- incontact the non-contact model (C) but model not in (D the). GAPDH non-contact siRNA model and negative(D). GAPDH (scrambled siRNA and NG2) negative siRNA (scrambled serve as controls NG2) siRNA for NG2 serve siRNA as controls in these for NG2 experiments. siRNA in these In the in-contactexperiments. model In the (E ),in- NG2contact expression model (E), by NG2 pericytes expression is needed by pericytes for formation is needed offor the formation ZO-1 positive of the endothelialZO-1 positive junctions endothelial that underliejunctions barrier that underlie function. barrier Expression function. of Expression NG2 by pericytes of NG2 does by pericytes not affect endothelialdoes not affect junction endothelial formation junction in the formation non-contact in the model non-contact (F). Junctions model ( areF). Junctions quantiﬁed are as quantified percentage ofas CD31-positive percentage of CD31 cells with-positive ZO-1 cells positive with ZO junctions.-1 positive Robust junctions. formation Robust of formation ZO-1 positive of ZO- endothelial1 positive junctionsendothelial (green) junctions is observed (green) when is observed NG2-positive when NG2 pericytes-positive are usedpericytes in the are in-contact used in themodel in-contact (G: blue =model DAPI). (G Endothelial: blue = DAPI). junction Endothelial formation junction is impaired formation in the is impaired in-contact in model the in when-contact NG2 model is knocked when downNG2 inis knocked pericytes down (H). In in the pericytes in-contact (H). model, In the iβn1-contact integrin model, activation 1 integrin in endothelial activation cells in isendothelial reduced by knockdowncells is reduced of NG2 by in knockdown pericytes ( I of). In NG2 the in non-contact pericytes model,(I). In the endothelial non-contactβ1 integrin model, endothelialactivation is  not1 affectedintegrin by activation NG2 knockdown is not affected in pericytes by NG2 (J ).knockdown The HUTS-21 in pericytes antibody ( wasJ). The used HUTS to quantify-21 antibodyβ1 integrin was activationused to quantify in HUVECs. 1 integr Activationin activation is quantiﬁed in HUVECs. as the percentage Activation of is CD31-positive quantified as thecells percentage with HUTS-21 of positiveCD31-positive junctions. cells Data with taken HUTS from-21 positive [9]. junctions. Data taken from [9].

Cancers 2017,, 9,, 3131 9 of 1515

These in vitro findings provide insight into the deficits in vascularization of intracranial B16F10 tumorsThese in pericytein vitro-findingsspecific NG2 provide null insight mice (PC into-NG2ko). the deficits These in vascularization mice are generated of intracranial by crossing B16F10 NG2 tumorsfloxed mice in pericyte-specific [45] with pdgfrb NG2-Cre nullmice mice [70,71] (PC-NG2ko).. Greater than These 95% mice of pericytes are generated in PC- byNG2ko crossing mice NG2 are floxednegative mice for [45 NG2] with [9,28,48] pdgfrb-Cre. Progression mice [70 of,71 intracranial]. Greater than melanomas 95% of pericytes is significantly in PC-NG2ko impaired mice in arePC-NG2ko negative mice, for NG2 compared [9,28,48 to]. control Progression mice of (Figure intracranial 5C,D). melanomas As initially is suggested significantly by work impaired in the in PC-NG2kogermline NG2 mice, null compared mouse [72,73] to control, this reduced mice (Figure tumor5C,D). growth As initiallystems from suggested deficits byin the work structure in the germlineand function NG2 of null tumor mouse blood [72,73 vessels], this reduced[9,48]. U tumornderlying growth these stems deficits from indeficits vascularization in the structure is a anddiminished function ability of tumor of NG2 blood null vessels pericytes [9, 48 to]. effectively Underlying ensheath these deficitsendothelial in vascularization cells, as judged is by a diminisheddouble immunolabeling ability of NG2 for null the pericytes endothelial to effectively marker CD31 ensheath and the endothelial pericyte cells,marker as judgedPDGFR by (Figure double β immunolabeling7A,B). Pericyte coverage for the endothelialof the endothelial marker tube CD31 is andreduced the pericyteby almost marker 40% in PDGFR tumors (Figurein PC-NG2ko7A,B). Pericytemice. The coverage decrease ofin thethis endothelial key interaction tube between is reduced pericytes by almost and endothelial 40% in tumors cells in is PC-NG2koaccompanied mice. by Thea 30% decrease decrease in in this basal key interactionlamina assembly between (as pericytesjudged by and collagen endothelial IV immunolabeling) cells is accompanied and by by a a 35% 30% decrease in basalendothelial lamina junction assembly formation (as judged (as byrevealed collagen by IV ZO immunolabeling)-1 immunolabeling), and byreminiscent a 35% decrease of the in endothelialvitro results junction obtained formation with NG2 (as knockdown revealed by pericytes ZO-1 immunolabeling), in transwell co- reminiscentcultures with of endothelial the in vitro resultscells. As obtained a result with of these NG2 knockdown structural deficits pericytes in intumor transwell blood co-cultures vessels, vessel with endothelial patency is decreasedcells. As a resultalmost of 2- thesefold, while structural vessel deficits leakiness in tumor is increased blood vessels,almost 3 vessel-fold. A patency 5-fold isincrease decreased in tumor almost hypoxia 2-fold, whilereflects vessel the decreased leakiness blood is increased flow to the almost tumors 3-fold. caused A 5-foldby these increase structural in tumorand functional hypoxia deficiencies reflects the decreased[9,28]. blood flow to the tumors caused by these structural and functional deficiencies [9,28].

Figure 7. Pericyte ensheathment of endothelial cells in tumors in NG2 null mice. Vessels in Figure 7. Pericyte ensheathment of endothelial cells in tumors in NG2 null mice. Vessels in intracranial intracranial B16F10 tumors were compared at 10 days in control (A) versus PC-NG2ko (B) mice, and B16F10 tumors were compared at 10 days in control (A) versus PC-NG2ko (B) mice, and in control in control (C) versus MacNG2ko (D) mice by immunolabeling sections for CD31 (endothelial cells) (C) versus MacNG2ko (D) mice by immunolabeling sections for CD31 (endothelial cells) and PDGFRβ  (pericytes).and PDGFR Pericyte (pericytes). (green) Pericyte ensheathment (green) ensheathment of endothelial of cellsendothelial (blue) iscells incomplete (blue) is inincomplete PC-NG2ko in vesselsPC-NG2ko (B), vessels as judged (B), by as gaps judged in coverageby gaps in (arrows). coverage These (arrows). gaps These are not gaps present are not in control present vessels in control (A). Invessels Mac-NG2ko (A). In Mac vessels-NG2ko (D), manyvessels pericytes (D), many (red, pericytes arrows) (red, fail toarrows associate) fail withto associate endothelial with cells endothelial (green). Thesecells (green). detached These pericytes detached are pericytes not seen inare control not seen vessels in control (C). These vessels images (C). These also highlightimages also the highlight reduced diameterthe reduced of vesselsdiameter in of Mac-NG2ko vessels in Mac mice,-NG2ko a phenotype mice, a that phenotype is not observed that is not in PC-NG2koobserved in mice. PC-NG2ko Bar in Dmice. = 20 Barµm. in DataD = 2 taken0 m. fromData [taken9,48]. from [9,48].

This change in pericyte-endothelial cell interaction in tumors in PC-NG2ko mice can be This change in pericyte-endothelial cell interaction in tumors in PC-NG2ko mice can be contrasted contrasted with the changes observed in tumors in Mac-NG2ko mice. Even though pericytes still with the changes observed in tumors in Mac-NG2ko mice. Even though pericytes still express NG2 express NG2 in the Mac-NG2ko mice, pericytes largely fail to interact with endothelial cells in in the Mac-NG2ko mice, pericytes largely fail to interact with endothelial cells in tumors in these tumors in these mice (Figure 7C,D). Structural and functional properties of tumor vessels in mice (Figure7C,D). Structural and functional properties of tumor vessels in Mac-NG2ko mice are Mac-NG2ko mice are correspondingly diminished compared to PC-NG2ko mice [48], likely correspondingly diminished compared to PC-NG2ko mice [48], likely accounting for the slower tumor accounting for the slower tumor growth seen in myeloid-specific NG2 null mice (Figure 5A,B). These growth seen in myeloid-speciﬁc NG2 null mice (Figure5A,B). These observations emphasize the fact observations emphasize the fact that NG2 is only one of many functional players involved in determining the ability of pericytes to interact effectively with endothelial cells.

Cancers 2017, 9, 31 10 of 15 that NG2 is only one of many functional players involved in determining the ability of pericytes to interact effectively with endothelial cells.

4. Future Prospects Our work clearly establishes the ability of the NG2 proteoglycan to activate β1 integrin signaling in the context of several processes that impact brain tumor progression. These include not only the malignant properties of the tumor cells, but also the properties of tumor blood vessels that are essential for tumor growth. Nevertheless, there are several aspects of the work in which additional mechanistic understanding is needed in order to fully appreciate the nature and importance of the NG2-integrin interaction.

1. What domains of NG2 and the integrin heterodimer are responsible for their interaction? Our results indicate that the NG2-integrin interaction is mediated by the extracellular domains of the two proteins, rather than by the cytoplasmic domains. However, we do not know what part of the NG2 ectodomain is involved in the interaction, or whether NG2 binds to the α or β integrin subunit (or to both). This also leaves open the questions of whether NG2 preferentially interacts with specific β1 heterodimers and how the spatial requirements for the interaction can be satisfied in both the cis and trans orientations. 2. Does NG2 influence crosstalk between integrins and receptor tyrosine kinases? Crosstalk between integrins and receptor tyrosine kinases is proposed to be an important aspect of amplifying and/or modulating transmembrane signaling [74–77]. Since NG2 has the ability to activate both of these signaling entities, it is of interest to know how the proteoglycan may affect crosstalk that occurs between the two and what implications this may have for the properties (e.g., proliferation, motility, survival) of both tumor and stromal cells. 3. How does NG2 interact with the cytoskeleton and how is this interaction affected by phosphorylation of the NG2 cytoplasmic domain? We propose that the phosphorylation state of NG2 determines its interaction with cytoplasmic scaffolding proteins and thus the localization of both NG2 and its β1 binding partner to specific membrane microdomains (lamellipodia versus apical microprotrusions). It remains for us to identify these cytoplasmic scaffolding proteins and to demonstrate how they anchor the NG2-integrin complex at sites that specifically influence proliferation versus motility. 4. How does NG2-dependent cis activation of integrins influence the biology of macrophages? Although we can demonstrate enhanced activation of both β1 and β2 integrins in NG2-positive THP1 macrophages, we have not yet established whether this influences macrophage interaction with the vascular endothelium (e.g. via binding to VCAM1 or ICAM1) or the ability of macrophages to extravasate from the circulation into tumors and other sites of inflammation. This will be an important addition to the spectrum of cellular events that are impacted by NG2.

Acknowledgments: This work was supported by NIH grant R01 CA095287 and by Sanford Burnham Prebys Lab Funding Initiative to WBS. I thank Regina Kapono for assistance with manuscript preparation and Karolina Kucharova for assistance with figure preparation. Conflicts of Interest: The authors declare no conflict of interest. Funding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

References

1. Fang, X.; Burg, M.A.; Barritt, D.; Dahlin-Huppe, K.; Nishiyama, A.; Stallcup, W.B. Cytoskeletal reorganization induced by engagement of the NG2 proteoglycan leads to cell spreading and migration. Mol. Biol. Cell 1999, 10, 3373–3387. [CrossRef][PubMed] 2. Majumdar, M.; Vuori, K.; Stallcup, W.B. Engagement of the NG2 proteoglycan triggers cell spreading via rac and p130cas. Cell. Signal. 2003, 15, 79–84. [CrossRef] Cancers 2017, 9, 31 11 of 15

3. Chekenya, M.; Krakstad, C.; Svendsen, A.; Netland, I.A.; Staalesen, V.; Tysnes, B.B.; Selheim, F.; Wang, J.; Sakariassen, P.O.; Sandal, T.; et al. The progenitor cell marker NG2/MPG promotes chemoresistance by activation of integrin-dependent PI3K/Akt signaling. Oncogene 2008, 27, 5182–5194. [CrossRef][PubMed] 4. Fukushi, J.; Makagiansar, I.T.; Stallcup, W.B. NG2 proteoglycan promotes endothelial cell motility and angiogenesis via engagement of galectin-3 and alpha3beta1 integrin. Mol. Biol. Cell 2004, 15, 3580–3590. [CrossRef][PubMed] 5. Iida, J.; Meijne, A.M.; Spiro, R.C.; Roos, E.; Furcht, L.T.; McCarthy, J.B. Spreading and focal contact formation of human melanoma cells in response to the stimulation of both melanoma-associated proteoglycan (NG2) and alpha 4 beta 1 integrin. Cancer Res. 1995, 55, 2177–2185. [PubMed] 6. Knutson, J.R.; Iida, J.; Fields, G.B.; McCarthy, J.B. CD44/chondroitin sulfate proteoglycan and alpha 2 beta 1 integrin mediate human melanoma cell migration on type IV collagen and invasion of basement membranes. Mol. Biol. Cell 1996, 7, 383–396. [CrossRef][PubMed] 7. Makagiansar, I.T.; Williams, S.; Dahlin-Huppe, K.; Fukushi, J.; Mustelin, T.; Stallcup, W.B. Phosphorylation of NG2 proteoglycan by protein kinase C-alpha regulates polarized membrane distribution and cell motility. J. Biol. Chem. 2004, 279, 55262–55270. [CrossRef][PubMed] 8. Makagiansar, I.T.; Williams, S.; Mustelin, T.; Stallcup, W.B. Differential phosphorylation of NG2 proteoglycan by ERK and PKCalpha helps balance cell proliferation and migration. J. Cell Biol. 2007, 178, 155–165. [CrossRef][PubMed] 9. You, W.K.; Yotsumoto, F.; Sakimura, K.; Adams, R.H.; Stallcup, W.B. NG2 proteoglycan promotes tumor vascularization via integrin-dependent effects on pericyte function. Angiogenesis 2014, 17, 61–76. [CrossRef] [PubMed] 10. Cattaruzza, S.; Ozerdem, U.; Denzel, M.; Ranscht, B.; Bulian, P.; Cavallaro, U.; Zanocco, D.; Colombatti, A.; Stallcup, W.B.; Perris, R. Multivalent proteoglycan modulation of FGF mitogenic responses in perivascular cells. Angiogenesis 2013, 16, 309–327. [CrossRef][PubMed] 11. Goretzki, L.; Burg, M.A.; Grako, K.A.; Stallcup, W.B. High-afﬁnity binding of basic ﬁbroblast growth factor and platelet-derived growth factor-AA to the core protein of the NG2 proteoglycan. J. Biol. Chem. 1999, 274, 16831–16837. [CrossRef][PubMed] 12. Grako, K.A.; Ochiya, T.; Barritt, D.; Nishiyama, A.; Stallcup, W.B. PDGF (alpha)-receptor is unresponsive to PDGF-AA in aortic smooth muscle cells from the NG2 knockout mouse. J. Cell Sci. 1999, 112, 905–915. [PubMed] 13. Grako, K.A.; Stallcup, W.B. Participation of the NG2 proteoglycan in rat aortic smooth muscle cell responses to platelet-derived growth factor. Exp. Cell Res. 1995, 221, 231–240. [CrossRef][PubMed] 14. Sugiarto, S.; Persson, A.I.; Munoz, E.G.; Waldhuber, M.; Lamagna, C.; Andor, N.; Hanecker, P.; Ayers-Ringler, J.; Phillips, J.; Siu, J.; et al. Asymmetry-defective oligodendrocyte progenitors are glioma precursors. Cancer Cell 2011, 20, 328–340. [CrossRef][PubMed] 15. Nishiyama, A.; Lin, X.H.; Giese, N.; Heldin, C.H.; Stallcup, W.B. Interaction between NG2 proteoglycan and PDGF alpha-receptor on O2A progenitor cells is required for optimal response to PDGF. J. Neurosci. Res. 1996, 43, 315–330. [CrossRef] 16. Stallcup, W.B. The NG2 proteoglycan: past insights and future prospects. J. Neurocytol. 2002, 31, 423–435. [CrossRef][PubMed] 17. Stallcup, W.B.; Huang, F.J. A role for the NG2 proteoglycan in glioma progression. Cell Adh. Migr. 2008, 2, 192–201. [CrossRef][PubMed] 18. Cattaruzza, S.; Nicolosi, P.A.; Braghetta, P.; Pazzaglia, L.; Benassi, M.S.; Picci, P.; Lacrima, K.; Zanocco, D.; Rizzo, E.; Stallcup, W.B.; et al. NG2/CSPG4-collagen type VI interplays putatively involved in the microenvironmental control of tumour engraftment and local expansion. J. Mol. Cell Biol. 2013, 5, 176–193. [CrossRef][PubMed] 19. Smith, F.O.; Rauch, C.; Williams, D.E.; March, C.J.; Arthur, D.; Hilden, J.; Lampkin, B.C.; Buckley, J.D.; Buckley, C.V.; Woods, W.G.; et al. The human homologue of rat NG2, a chondroitin sulfate proteoglycan, is not expressed on the cell surface of normal hematopoietic cells but is expressed by acute myeloid leukemia blasts from poor-prognosis patients with abnormalities of chromosome band 11q23. Blood 1996, 87, 1123–1133. [PubMed] Cancers 2017, 9, 31 12 of 15

20. Behm, F.G.; Smith, F.O.; Raimondi, S.C.; Pui, C.H.; Bernstein, I.D. Human homologue of the rat chondroitin sulfate proteoglycan, NG2, detected by monoclonal antibody 7.1, identifies childhood acute lymphoblastic leukemias with t(4;11)(q21;q23) or t(11;19)(q23;p13) and MLL gene rearrangements. Blood 1996, 87, 1134–1139. [PubMed] 21. Shoshan, Y.; Nishiyama, A.; Chang, A.; Mork, S.; Barnett, G.H.; Cowell, J.K.; Trapp, B.D.; Staugaitis, S.M. Expression of oligodendrocyte progenitor cell antigens by gliomas: implications for the histogenesis of brain tumors. Proc. Natl. Acad. Sci. USA 1999, 96, 10361–10366. [CrossRef][PubMed] 22. Al-Mayhani, M.T.; Grenfell, R.; Narita, M.; Piccirillo, S.; Kenney-Herbert, E.; Fawcett, J.W.; Collins, V.P.; Ichimura, K.; Watts, C. NG2 expression in glioblastoma identifies an actively proliferating population with an aggressive molecular signature. NeuroOncology 2011, 13, 830–845. [CrossRef][PubMed] 23. Persson, A.I.; Petritsch, C.; Swartling, F.J.; Itsara, M.; Sim, F.J.; Auvergne, R.; Goldenberg, D.D.; Vandenberg, S.R.; Nguyen, K.N.; Yakovenko, S.; et al. Non-stem cell origin for oligodendroglioma. Cancer Cell 2010, 18, 669–682. [CrossRef][PubMed] 24. Price, M.A.; Colvin Wanshura, L.E.; Yang, J.; Carlson, J.; Xiang, B.; Li, G.; Ferrone, S.; Dudek, A.Z.; Turley, E.A.; McCarthy, J.B. CSPG4, a potential therapeutic target, facilitates malignant progression of melanoma. Pigment Cell Melanoma. Res. 2011, 24, 1148–1157. [CrossRef][PubMed] 25. Wang, J.; Svendsen, A.; Kmiecik, J.; Immervoll, H.; Skaftnesmo, K.O.; Planaguma, J.; Reed, R.K.; Bjerkvig, R.; Miletic, H.; Enger, P.O.; et al. Targeting the NG2/CSPG4 proteoglycan retards tumour growth and angiogenesis in preclinical models of GBM and melanoma. PLoS ONE 2011, 6, e23062. [CrossRef][PubMed] 26. Yadavilli, S.; Hwang, E.I.; Packer, R.J.; Nazarian, J. The Role of NG2 Proteoglycan in Glioma. Transl. Oncol. 2016, 9, 57–63. [CrossRef][PubMed] 27. Burg, M.A.; Grako, K.A.; Stallcup, W.B. Expression of the NG2 proteoglycan enhances the growth and metastatic properties of melanoma cells. J. Cell Physiol. 1998, 177, 299–312. [CrossRef] 28. Stallcup, W.B.; You, W.K.; Kucharova, K.; Cejudo-Martin, P.; Yotsumoto, F. NG2 Proteoglycan-Dependent Contributions of Pericytes and Macrophages to Brain Tumor Vascularization and Progression. Microcirculation 2016, 23, 122–133. [CrossRef][PubMed] 29. Clausen, T.M.; Pereira, M.A.; Al Nakouzi, N.; Oo, H.Z.; Agerbaek, M.O.; Lee, S.; Orum-Madsen, M.S.; Kristensen, A.R.; El-Naggar, A.; Grandgenett, P.M.; et al. Oncofetal Chondroitin Sulfate Glycosaminoglycans Are Key Players in Integrin Signaling and Tumor Cell Motility. Mol. Cancer Res. 2016, 14, 1288–1299. [CrossRef][PubMed] 30. Viola, M.; Bruggemann, K.; Karousou, E.; Caon, I.; Carava, E.; Vigetti, D.; Greve, B.; Stock, C.; De Luca, G.; Passi, A.; et al. MDA-MB-231 breast cancer cell viability, motility and matrix adhesion are regulated by a complex interplay of heparan sulfate, chondroitin-/dermatan sulfate and hyaluronan biosynthesis. Glycoconj. J. 2016.[CrossRef][PubMed] 31. Garrigues, H.J.; Lark, M.W.; Lara, S.; Hellstrom, I.; Hellstrom, K.E.; Wight, T.N. The melanoma proteoglycan: restricted expression on microspikes, a specific microdomain of the cell surface. J. Cell Biol. 1986, 103, 1699–1710. [CrossRef][PubMed] 32. Barritt, D.S.; Pearn, M.T.; Zisch, A.H.; Lee, S.S.; Javier, R.T.; Pasquale, E.B.; Stallcup, W.B. The multi-PDZ domain protein MUPP1 is a cytoplasmic ligand for the membrane-spanning proteoglycan NG2. J. Cell Biochem. 2000, 79, 213–224. [CrossRef] 33. Stegmuller, J.; Werner, H.; Nave, K.A.; Trotter, J. The proteoglycan NG2 is complexed with alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors by the PDZ glutamate receptor interaction protein (GRIP) in glial progenitor cells. Implications for glial-neuronal signaling. J. Biol. Chem. 2003, 278, 3590–3598. [CrossRef][PubMed] 34. Chatterjee, N.; Stegmuller, J.; Schatzle, P.; Karram, K.; Koroll, M.; Werner, H.B.; Nave, K.A.; Trotter, J. Interaction of syntenin-1 and the NG2 proteoglycan in migratory oligodendrocyte precursor cells. J. Biol. Chem. 2008, 283, 8310–8317. [CrossRef][PubMed] 35. Luque, A.; Gomez, M.; Puzon, W.; Takada, Y.; Sanchez-Madrid, F.; Cabanas, C. Activated conformations of very late activation integrins detected by a group of antibodies (HUTS) specific for a novel regulatory region (355-425) of the common beta 1 chain. J. Biol. Chem. 1996, 271, 11067–11075. [PubMed] 36. Lathia, J.D.; Gallagher, J.; Heddleston, J.M.; Wang, J.; Eyler, C.E.; Macswords, J.; Wu, Q.; Vasanji, A.; McLendon, R.E.; Hjelmeland, A.B.; et al. Integrin alpha 6 regulates glioblastoma stem cells. Cell Stem Cell 2010, 6, 421–432. [CrossRef][PubMed] Cancers 2017, 9, 31 13 of 15

37. Hanahan, D.; Folkman, J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 1996, 86, 353–364. [CrossRef] 38. Bergers, G.; Benjamin, L.E. Tumorigenesis and the angiogenic switch. Nat. Rev. Cancer 2003, 3, 410. [CrossRef] [PubMed] 39. Armulik, A.; Abramsson, A.; Betsholtz, C. Endothelial/pericyte interactions. Circ. Res. 2005, 97, 512–523. [CrossRef][PubMed] 40. Bergers, G.; Song, S. The role of pericytes in blood-vessel formation and maintenance. NeuroOncology 2005, 7, 452–464. [CrossRef][PubMed] 41. Ozerdem, U.; Grako, K.A.; Dahlin-Huppe, K.; Monosov, E.; Stallcup, W.B. NG2 proteoglycan is expressed exclusively by mural cells during vascular morphogenesis. Dev. Dyn. 2001, 222, 218–227. [CrossRef] [PubMed] 42. Ozerdem, U.; Monosov, E.; Stallcup, W.B. NG2 proteoglycan expression by pericytes in pathological microvasculature. Microvasc. Res. 2002, 63, 129–134. [CrossRef][PubMed] 43. Ozerdem, U.; Stallcup, W.B. Early contribution of pericytes to angiogenic sprouting and tube formation. Angiogenesis 2003, 6, 241–249. [CrossRef][PubMed] 44. Ozerdem, U.; Stallcup, W.B. Pathological angiogenesis is reduced by targeting pericytes via the NG2 proteoglycan. Angiogenesis 2004, 7, 269–276. [CrossRef][PubMed] 45. Chang, Y.; She, Z.G.; Sakimura, K.; Roberts, A.; Kucharova, K.; Rowitch, D.H.; Stallcup, W.B. Ablation of NG2 proteoglycan leads to deficits in brown fat function and to adult onset obesity. PLoS ONE 2012, 7, e30637. [CrossRef][PubMed] 46. Clausen, B.E.; Burkhardt, C.; Reith, W.; Renkawitz, R.; Forster, I. Conditional gene targeting in macrophages and granulocytes using LysMcre mice. Transgenic. Res. 1999, 8, 265–277. [CrossRef][PubMed] 47. Stockmann, C.; Doedens, A.; Weidemann, A.; Zhang, N.; Takeda, N.; Greenberg, J.I.; Cheresh, D.A.; Johnson, R.S. Deletion of vascular endothelial growth factor in myeloid cells accelerates tumorigenesis. Nature 2008, 456, 814–818. [CrossRef][PubMed] 48. Yotsumoto, F.; You, W.K.; Cejudo-Martin, P.; Kucharova, K.; Sakimura, K.; Stallcup, W.B. NG2 proteoglycan-dependent recruitment of tumor macrophages promotes pericyte-endothelial cell interactions required for brain tumor vascularization. Oncoimmunology 2015, 4, e1001204. [CrossRef][PubMed] 49. Coffelt, S.B.; Lewis, C.E.; Naldini, L.; Brown, J.M.; Ferrara, N.; De Palma, M. Elusive identities and overlapping phenotypes of proangiogenic myeloid cells in tumors. Am. J. Pathol. 2010, 176, 1564–1576. [CrossRef][PubMed] 50. De Palma, M.; Venneri, M.A.; Galli, R.; Sergi Sergi, L.; Politi, L.S.; Sampaolesi, M.; Naldini, L. Tie2 identifies a hematopoietic lineage of proangiogenic monocytes required for tumor vessel formation and a mesenchymal population of pericyte progenitors. Cancer Cell 2005, 8, 211–226. [CrossRef][PubMed] 51. Dransfield, I.; Cabanas, C.; Barrett, J.; Hogg, N. Interaction of leukocyte integrins with ligand is necessary but not sufficient for function. J. Cell Biol. 1992, 116, 1527–1535. [CrossRef][PubMed] 52. Forlow, S.B.; White, E.J.; Barlow, S.C.; Feldman, S.H.; Lu, H.; Bagby, G.J.; Beaudet, A.L.; Bullard, D.C.; Ley, K. Severe inflammatory defect and reduced viability in CD18 and E-selectin double-mutant mice. J. Clin. Investig. 2000, 106, 1457–1466. [CrossRef][PubMed] 53. Jin, H.; Su, J.; Garmy-Susini, B.; Kleeman, J.; Varner, J. Integrin alpha4beta1 promotes monocyte trafficking and angiogenesis in tumors. Cancer Res. 2006, 66, 2146–2152. [CrossRef][PubMed] 54. Stark, K.; Eckart, A.; Haidari, S.; Tirniceriu, A.; Lorenz, M.; von Bruhl, M.L.; Gartner, F.; Khandoga, A.G.; Legate, K.R.; Pless, R.; et al. Capillary and arteriolar pericytes attract innate leukocytes exiting through venules and ‘instruct’ them with pattern-recognition and motility programs. Nat. Immunol. 2013, 14, 41–51. [CrossRef][PubMed] 55. Ulyanova, T.; Priestley, G.V.; Banerjee, E.R.; Papayannopoulou, T. Unique and redundant roles of alpha4 and beta2 integrins in kinetics of recruitment of lymphoid vs. myeloid cell subsets to the inflamed peritoneum revealed by studies of genetically deficient mice. Exp. Hematol. 2007, 35, 1256–1265. [CrossRef][PubMed] 56. Cohen, J.V.; Tawbi, H.; Margolin, K.A.; Amravadi, R.; Bosenberg, M.; Brastianos, P.K.; Chiang, V.L.; de Groot, J.; Glitza, I.C.; Herlyn, M.; et al. Melanoma central nervous system metastases: current approaches, challenges, and opportunities. Pigment Cell Melanoma. Res. 2016, 29, 627–642. [CrossRef][PubMed] Cancers 2017, 9, 31 14 of 15

57. Bunse, S.; Garg, S.; Junek, S.; Vogel, D.; Ansari, N.; Stelzer, E.H.; Schuman, E. Role of N-cadherin cis and trans interfaces in the dynamics of adherens junctions in living cells. PLoS ONE 2013, 8, e81517. [CrossRef] [PubMed] 58. Wu, Y.; Jin, X.; Harrison, O.; Shapiro, L.; Honig, B.H.; Ben-Shaul, A. Cooperativity between trans and cis interactions in cadherin-mediated junction formation. Proc. Natl. Acad. Sci. USA 2010, 107, 17592–17597. [CrossRef][PubMed] 59. Arvanitis, D.; Davy, A. Eph/ephrin signaling: networks. Genes Dev. 2008, 22, 416–429. [CrossRef][PubMed] 60. Nishiyama, A.; Lin, X.H.; Stallcup, W.B. Generation of truncated forms of the NG2 proteoglycan by cell surface proteolysis. Mol. Biol. Cell 1995, 6, 1819–1832. [CrossRef][PubMed] 61. Martin, S.; Levine, A.K.; Chen, Z.J.; Ughrin, Y.; Levine, J.M. Deposition of the NG2 proteoglycan at nodes of Ranvier in the peripheral nervous system. J. Neurosci. 2001, 21, 8119–8128. [PubMed] 62. Jones, L.L.; Yamaguchi, Y.; Stallcup, W.B.; Tuszynski, M.H. NG2 is a major chondroitin sulfate proteoglycan produced after spinal cord injury and is expressed by macrophages and oligodendrocyte progenitors. J. Neurosci. 2002, 22, 2792–2803. [PubMed] 63. Larsen, P.H.; Wells, J.E.; Stallcup, W.B.; Opdenakker, G.; Yong, V.W. Matrix metalloproteinase-9 facilitates remyelination in part by processing the inhibitory NG2 proteoglycan. J. Neurosci. 2003, 23, 11127–11135. [PubMed] 64. Asher, R.A.; Morgenstern, D.A.; Properzi, F.; Nishiyama, A.; Levine, J.M.; Fawcett, J.W. Two separate metalloproteinase activities are responsible for the shedding and processing of the NG2 proteoglycan in vitro. Mol. Cell Neurosci. 2005, 29, 82–96. [CrossRef][PubMed] 65. Sakry, D.; Neitz, A.; Singh, J.; Frischknecht, R.; Marongiu, D.; Biname, F.; Perera, S.S.; Endres, K.; Lutz, B.; Radyushkin, K.; et al. Oligodendrocyte precursor cells modulate the neuronal network by activity-dependent ectodomain cleavage of glial NG2. PLoS Biol. 2014, 12, e1001993. [CrossRef][PubMed] 66. Joo, N.E.; Miao, D.; Bermudez, M.; Stallcup, W.B.; Kapila, Y.L. Shedding of NG2 by MMP-13 attenuates anoikis. DNA Cell Biol. 2014, 33, 854–862. [CrossRef][PubMed] 67. Tillet, E.; Ruggiero, F.; Nishiyama, A.; Stallcup, W.B. The membrane-spanning proteoglycan NG2 binds to collagens V and VI through the central nonglobular domain of its core protein. J. Biol. Chem. 1997, 272, 10769–10776. [PubMed] 68. Lenter, M.; Uhlig, H.; Hamann, A.; Jeno, P.; Imhof, B.; Vestweber, D. A monoclonal antibody against an activation epitope on mouse integrin chain beta 1 blocks adhesion of lymphocytes to the endothelial integrin alpha 6 beta 1. Proc. Natl. Acad. Sci. USA 1993, 90, 9051–9055. [CrossRef][PubMed] 69. Kaya, M.; Ahishali, B. Assessment of permeability in barrier type of endothelium in brain using tracers: Evans blue, sodium fluorescein, and horseradish peroxidase. Methods Mol. Biol. 2011, 763, 369–382. [CrossRef] [PubMed] 70. Foo, S.S.; Turner, C.J.; Adams, S.; Compagni, A.; Aubyn, D.; Kogata, N.; Lindblom, P.; Shani, M.; Zicha, D.; Adams, R.H. Ephrin-B2 controls cell motility and adhesion during blood-vessel-wall assembly. Cell 2006, 124, 161–173. [CrossRef][PubMed] 71. Stenzel, D.; Nye, E.; Nisancioglu, M.; Adams, R.H.; Yamaguchi, Y.; Gerhardt, H. Peripheral mural cell recruitment requires cell-autonomous heparan sulfate. Blood 2009, 114, 915–924. [CrossRef][PubMed] 72. Gibby, K.; You, W.K.; Kadoya, K.; Helgadottir, H.; Young, L.J.; Ellies, L.G.; Chang, Y.; Cardiff, R.D.; Stallcup, W.B. Early vascular deficits are correlated with delayed mammary tumorigenesis in the MMTV-PyMT transgenic mouse following genetic ablation of the NG2 proteoglycan. Breast Cancer Res. 2012, 14, R67. [CrossRef][PubMed] 73. Huang, F.J.; You, W.K.; Bonaldo, P.; Seyfried, T.N.; Pasquale, E.B.; Stallcup, W.B. Pericyte deficiencies lead to aberrant tumor vascularizaton in the brain of the NG2 null mouse. Dev. Biol. 2010, 344, 1035–1046. [CrossRef] [PubMed] 74. Borges, E.; Jan, Y.; Ruoslahti, E. Platelet-derived growth factor receptor beta and vascular endothelial growth factor receptor 2 bind to the beta 3 integrin through its extracellular domain. J. Biol. Chem. 2000, 275, 39867–39873. [CrossRef][PubMed] 75. West, X.Z.; Meller, N.; Malinin, N.L.; Deshmukh, L.; Meller, J.; Mahabeleshwar, G.H.; Weber, M.E.; Kerr, B.A.; Vinogradova, O.; Byzova, T.V. Integrin beta3 crosstalk with VEGFR accommodating tyrosine phosphorylation as a regulatory switch. PLoS ONE 2012, 7, e31071. [CrossRef][PubMed] Cancers 2017, 9, 31 15 of 15

76. Dan, L.; Jian, D.; Na, L.; Xiaozhong, W. Crosstalk between EGFR and integrin affects invasion and proliferation of gastric cancer cell line, SGC7901. Onco. Targets Ther. 2012, 5, 271–277. [CrossRef][PubMed] 77. Takada, Y.; Takada, Y.K.; Fujita, M. Crosstalk between insulin-like growth factor (IGF) receptor and integrins through direct integrin binding to IGF1. Cytokine Growth Factor Rev. 2017.[CrossRef][PubMed]

© 2017 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).