The role of caveolin-1 in tumors of the brain - functional and clinical implications

Abstract

Background

Caveolin-1 (cav-1) is the major structural protein of caveolae, the flask-shaped invaginations of the plasma membrane mainly involved in cell signaling. Today, cav-1 is believed to play a role in a variety of disease processes including cancer, owing to the variations of its expression in association with tumor progression, invasive behavior, metastasis and therapy resistance. Since first detected in the brain, a number of studies has particularly focused on the role of cav-1 in the various steps of brain tumorigenesis. In this review, we discuss the different roles of cav-1 and its contributions to the molecular mechanisms underlying the pathobiology and natural behavior of brain tumors including glial, non-glial and metastatic subtypes. These contributions could be attributed to its co-localization with important players in tumorigenesis within the lipid-enriched domains of the plasma membrane. In that regard, the ability of cav-1 to interact with various cell signaling molecules as well as the impact of caveolae depletion on important pathways acting in brain tumor pathogenesis are noteworthy. We also discuss conversant causes hampering the treatment of malignant glial tumors such as limited transport of chemotherapeutics across the blood tumor barrier and resistance to chemoradiotherapy, by focusing on the molecular fundamentals involving cav-1 participation.

Conclusions

Cav-1 has the potential to pivot the molecular basis underlying the pathobiology of brain tumors, particularly the malignant glial subtype. In addition, the regulatory effect of cav-1-dependent and caveola-mediated transcellular transport on the permeability of the blood tumor barrier could be of benefit to overcome the restricted transport across brain barriers when applying chemotherapeutics. The association of cav-1 with tumors of the brain other than malignant gliomas deserves to be underlined, as well given the evidence suggesting its potential in predicting tumor grade and recurrence rates together with determining patient prognosis in oligodendrogliomas, ependymomas, meningiomas, vestibular schwannomas and brain metastases.

Introduction

Caveolae are flask-shaped, cholesterol-rich invaginations of the plasma membrane, basically involved in signal transduction, vesicular transport, cholesterol hemostasis and substrate metabolism [1]. Today, they are believed to play a role in a variety of disease processes including cancer. Among the 3 structural proteins of caveolae, caveolin-1 (cav-1) has been the most widely studied and is, thus, most well-known so far. Cav-1 is a 178 amino acid and 22 kD integral membrane protein ubiquitously expressed in various cell types with the highest levels in adipocytes, endothelial cells, fibroblasts, smooth muscle cells and epithelial cells. Being the main structural protein component, cav-1 is essential for both the formation and the functioning of caveolae. Cav-1 also governs the majority of caveolae functions and has been implicated in the regulation of cell cycle progression, cell proliferation and cell death [2, 3]. For these purposes, cav-1 directly interacts with a vast number of proteins involved in cell signaling through its homologous domain, i.e., the caveolin-scaffolding domain, which binds to the caveolin-binding motifs in the proteins. Other constitutional proteins of caveolae, cav-2 and -3, are structurally similar to cav-1, but although cav-2 is largely co-expressed with cav-1, cav-3 expression is specific to skeletal muscle [4]. More recently, another family of proteins, known as the ‘cavin family’, has been defined as a key regulator of caveolae dynamics [5]. However, these proteins and caveolins other than cav-1 are beyond the scope of this review.

From when its potential role in oncogenic cell transformation [6] and carcinogenesis [7] was first suggested onwards, cav-1 has been subject of investigation in various cancer types. As a consequence, it has been found that the expression of cav-1 is either upregulated or downregulated in different human cancers including breast cancer [8], prostate cancer [9], leukemia [10], renal cancer [11], pancreas cancer [12] and thyroid cancer [13]. As such, the tumor suppressor or oncogenic effects of cav-1 seem to vary depending on the type of tissue and the tumor itself. Recent work has revealed variations in cav-1 levels even during tumor development, with low levels at earlier stages and high levels at advanced stages [14]. Importantly, cav-1 has been shown to contribute to cell survival, angiogenesis, metastasis and chemo- and radiotherapy resistance in several cancer types, thereby playing a significant role in cancer prognosis as well [15–17].

Since first detected in the resident cells of the brain [18–21], evidence has emphasized a role of cav-1 in several molecular pathways involved in the pathobiology and natural behavior of brain tumors, especially glioblastomas, leading to tumor growth, progression, invasion, aggressiveness, metastasis and resistance to chemoradiotherapy. In this review, we comprehensively discuss the role of cav-1 in glial, non-glial and metastatic brain tumors along with its contributions to treatment challenges such as limited transport across the blood tumor barrier (BTB) and resistance to therapy, which are noteworthy.

Primary brain tumors of glial origin

Primary tumors of the central nervous system (CNS) are substantial causes of cancer-related deaths in adults [22] and they are the second leading cause of pediatric cancer-related deaths, after leukemia [23]. Among all types, tumors originating from the supportive glial cell lineages including astrocytes, oligodendrocytes and ependymal cells, i.e., gliomas, are the most common ones in both adults and children. In this section, we discuss the role of cav-1 in gliomas, focusing on astrocytic, oligodendroglial and ependymal tumors separately given their individual biologic, pathologic, genetic and behavioral characteristics.

Astrocytic tumors

Compared to others, WHO grade IV astrocytoma, also known as glioblastoma, is dreadful, given its aggressive anaplastic, infiltrative and angiogenic properties [24–26]. Maximum surgical resection followed by chemoradiotherapy is the current standard of care for glioblastomas. Despite ample efforts to improve its treatment, the median survival of newly diagnosed glioblastoma patients still does not go beyond 15 months [27, 28]. Owing to the refractory behavior of glioblastomas to current therapeutic strategies, improvement of these strategies as well as the development of novel targeted strategies are mandatory. In fact, glioblastomas have already outpaced other CNS tumors and have become the focus of many preclinical and clinical studies over the years, including those exploring the cav-1-astrocytoma relation. Initial reports revealed non-significant changes in tumoral cav-1 levels compared to that in normal brain [29], but increased expression was noted later on [30]. Moreover, a grade-specific variability in expression pattern and intensity with a significant association between high cav-1 levels and high histologic grades were observed [31]. This positive correlation was affirmed by the more prominent caveolin mRNA levels observed in human glioblastoma tissues compared to those in normal brain tissues [30]. In addition to the tumor tissue itself, cav-1 levels were recently also evaluated in exosomes isolated from plasma of glioblastoma patients [32] and in myeloid cells obtained from human glioblastoma tissues [33] (Table 1). Below, we describe the molecular mechanisms linked to cav-1 in the pathobiology of astrocytic brain tumors, thereby focusing on glioblastoma.

Table 1.

Studies with human tissues demonstrating cav-1 expression in primary brain tumors of glial origin

Tissue Type	Type and Distribution of the Tumors (n)	Control Group (n)	Cav-1 Detection Method	Cav-1 Status (n)	Significance	Associated Findings (n)	Reference
Tumor	Astrocytoma (24) • Grade I (4) • Grade II (4) • Grade III (7) • Grade IV (9)	Normal brain tissue (5)	Western blotting for immunodetection of protein expression	Positive in all, except one grade III astrocytoma (23)	No significance between the tumors and control group	Significantly lower expression of RhoA and RhoB in tumors compared to control Inverse correlation between the tumor grade and expression of RhoA and RhoB No significance between Rac1 expression in the tumors compared to control	[29]
Tumor	Astrocytoma (6) • Grade IV (6)	Normal brain tissue (2)	qRT-PCR for m-RNA expression	Positive in all	21- to 321-fold increase in tumors compared to control group	Colocalization of EGFR, but not EGFRvIII, with cav-1 within the lipid rafts Highest cav-1 expression in the only tissue that expressed EGFRvIII	[30]
Tumor	Glioma (64) • Grade II Astrocytoma (7) • Grade III Astrocytoma (6) • Grade IV Astrocytoma (16) • Gliosarcoma (3) • Grade II Oligodendroglioma (12) • Grade III Oligodendroglioma (10) • Grade II Oligoastrocytoma (4) • Grade III Oligoastrocytoma (6)	Normal brain tissue (10)	H&E staining for staining pattern and intensity, immunohistochemistry for protein expression	Positive in all astrocytomas Negative in all grade II (12) and 70% of grade III oligodendrogliomas (7) Positive in 50% of grade II (2) and 83% of grade III (5) oligoastrocytomas	Grade-specific pattern (progressively acquired membrane pattern with increasing tumor grade) in astroglial tumors and when positive, grade-specific intensity in all gliomas (higher intensity with higher grades)	1p/19q deletion in 92% of grade II (11) and 70% of grade III (7) oligodendrogliomas Cav-1 negativity in all 1p/19q deleted tumors Significant correlation between 1p/19q deletion and lack of cav-1 staining	[34]
Tumor	Glioma (73) • Grade II Astrocytoma (7) • Grade III Astrocytoma (17) • Grade IV Astrocytoma (20) • Grade II Oligodendroglioma (11) • Grade III Oligodendroglioma (5) • Grade II Oligoastrocytoma (8) • Grade III Oligoastrocytoma (5)	Normal brain tissue (8)	Immunohistochemistry for protein expression	Positive in 14% of grade II (5), 63% (12) of grade III and 90% (18) of grade IV astrocytomas Positive in 45% (5) of grade II and 60% (3) of grade III oligodendrogliomas Positive in 25% (2) of grade II and 60% (3) of grade III oligoastrocytomas	No significance according to the histological subtype Significance between grade II and grade IV gliomas with high cav-1 in high grade tumors	Significant correlation of high cav-1 immunoexpression with p53 overexpression but not with Ki-67 or EGFR overexpression in astroglial-derived tumors No significant correlation between cav-1 expression and histologic grade, Ki-67, EGFR, and p53 overexpression in any of the tumors with oligodendroglial component 1p/19q deletion in 91% of grade II (10) and 100% of grade III (5) oligodendrogliomas, and 12.5% (1) grade II and 20% (1) grade III oligoastrocytomas. No significant correlation between 1p/19q deletion and absence of cav-1 staining	[31]
Tumor	Oligodendroglioma (87) • Grade II Oligodendroglioma (33) • Grade III Oligodendroglioma (21) • Grade II Oligoastrocytoma (10) • Grade III Oligoastrocytoma (16) • Grade IV (Glioblastoma with oligodendroglial component) (7)	Blood vessels within the specimens for positive control	H&E staining and immunohistochemistry for cav-1 expression	Positive in 9% of grade II (3), 10% (2) of grade III oligodendrogliomas Positive in 20% of grade II (2) and 44% (7) of grade III oligoastrocytomas Positive in 71% of glioblastomas (5)	Significant association between cav-1 expression and tumor grade/type with higher levels in high-grade gliomas with an astroglial component Significant association between cav-1 expression and shorter survival	1p/19q deletion in 55% of grade II (18) and 62% of grade III (13) oligodendrogliomas, and 70% (7) of grade II and 6% (1) of grade III oligoastrocytomas, and 14% (1) of glioblastomas Significant correlation between 1p/19q deletion and lack of cav-1 immunoreactivity	[35]
Tumor	Ependymoma (22) • Grade II (14) • Grade III (8)	–	H&E staining for staining pattern and intensity, immunohistochemistry for protein expression	Expression over the cut-off value (30%) in 50% (11) and below the cut-off value in 50% (11)	Significant correlation of only cav-1 with poor overall survival in grade II tumors	Significant correlation of histological grade, extent of surgery, Ki-67, p53, cav-1, and EGFR expression with overall survival Significant correlation between cav-1 expression and Ki-67, EGFR, and p53 immunoreactivity Colocalization of cav-1 and EGFR immunoreactivity in 60% of the tumor cells Significant correlation between cav-1 expression and poor overall survival	[36]
Tumor	Glioma (subtypes not-specified) (108) • Grade II (21) • Grade III (38) • Grade IV (35)	Normal brain tissue (14)	qRT-PCR for m-RNA expression	Low expression in 83% (90) and high expression in 17% (18) of the tumors	When present, positive correlation between increased cav-1 expression and tumor aggressiveness	Accompanied increase in ITGA5 and TGFβ levels in correlation with tumor aggressiveness Negative correlation between cav-1 expression and ITGA5 and TGFβ expression in 29% (31) of the tumors Low levels of all 3 genes in 71% (77) of the tumors	[37]
Tumor and plasma exosomes	Astrocytoma • Grade IV Astrocytoma (343) • Exosomes isolated from plasmas of grade IV astrocytoma patients (8)	Control plasma (5)	Western blotting for immunodetection of protein expression qRT-PCR for m-RNA expression	Enriched cav-1 in exosomes obtained from tumor patients	–	Upregulated mRNAs of hypoxia-induced exosome components in hypoxic regions of tumor tissues	[32]
Myeloid cells	Astrocytoma • TAMs isolated from grade IV astrocytoma patients (10)	Normal brain tissue (3)	Cell sorting to obtain purified TAMs qRT-PCR for m-RNA expression of exvivo sorted TAMs samples	Positive in all	Significantly upregulated (6.32 fold) in TAMs from tumor patients compared to control group	GBM-mediated suppression of TAMs was mediated by cav-1 Restored myeloid cell function after suppression of cav-1	[33]
Tumor	Ependymoma (176) • Grade II (107) • Grade III (69)	–	H&E staining and immunohistochemistry for protein expression	Positive (>30%) in 48% (83) of the tumors	–	Worse prognosis with positive staining only in the infratentorial group	[38]

Cav-1: Caveolin-1, EGFR: Epidermal growth factor receptor, H&E: Hematoxylin and eosin, qRT-PCR: Quantitative reverse transcriptase polymerase chain reaction, TAMs: Tumor associated myeloid cells, TGFβ: Tumor growth factor β

Caveolin-1 and Rho GTPases

The Rho family of small GTPases are involved in actin cytoskeletal dynamics, axonogenesis, synaptic plasticity, neuronal development and migration [39–41]. In particular, RhoA, RhoB and Rac1 are key actors in cell migration [42]. The expression of Rho proteins has been found to be increased in tumors with metastasizing capacity such as lung, breast and colon tumors [43]. When localized in the plasma membrane, Rho proteins are situated in caveolae with a direct physical interaction with cav-1 [44]. Initial work examining the possible role of a crosstalk between cav-1 and Rho GTPases in human astrocytomas, that are known to metastasize rarely [45], revealed reduced Rho family protein levels together with unvaried cav-1 expression levels in the tumor tissues compared to normal brain tissues [29]. Specifically, the RhoA and RhoB protein levels were found to be decreased with inverse correlations to tumor grade, whereas a concomitantly reduced expression of Rac1 failed to expose such a correlation. As a result, Rho proteins and cav-1 were concluded to play a synergistic role in the failure of gliomas to metastasize. In contrast, increased expression of cav-1 and a more complex crosstalk between Rho GTPases and cav-1 involved in various glioma-associated features, including the regulation of glioma cell migration, BTB permeability and resistance to radiotherapy, which will be discussed in the following sections, have been disclosed. To be noted, current data regarding Rho GTPases in glioma aggressiveness and tumor progression support deregulation of these proteins in response to their upstream activators. Similarly, Rac1 is supposed to positively correlate with astrocytoma grade, as it has been shown to inhibit RhoA, thereby promoting invasive glioma cell behavior [46, 47].

Caveolin-1 and the extracellular matrix

Hyaluronic acid constitutes a substantial portion of the extracellular matrix in the brain [48]. The ubiquitous primary receptor of hyaluronic acid, CD44, has been implicated in the migration and invasion of tumor cells [49, 50]. In parallel, gliomas have been found to highly express CD44 in correlation with their adhesive, migratory and invasive behavior [51, 52]. Recently, CD44 has also been considered as a marker of cancer stem cells, a group of highly malignant cells identified in a number of cancers including glioblastoma, with the ability of self-renewal and proliferation [53–55]. In human glioblastoma-derived cell lines, inhibition of matrix metalloproteinase (MMP) has been found to be associated with suppression of CD44-dependent cell migration, suggesting a requirement of CD44 cell surface processing by MMP for migration [56]. In fact, on the cell surface, CD44 has been found to be localized together with the membrane type 1-matrix metalloproteinase (MT1-MMP) at the leading lamellipodia edge of motile cells [57, 58] and within the lipid enriched membrane domains containing cav-1 [59]. For this reason, the caveolar location of MT1-MMP and CD44 has been proposed to play a regulatory role in hyaluronic acid binding to the cell surface, as MT1-MMP-mediated cleavage of CD44 was found to be critical in promoting tumor cell migration and as this binding was found to be upregulated by the depletion of caveolae [60]. A common localization of MT1-MMP, CD44 and RhoA within caveolae [44, 59] is thought to play a role in the infiltrative glioma phenotype as well. As expected, it has been found that overexpression of RhoA GTPase upregulates MT1-MMP expression and triggers CD44 shedding from the glioma cell surface, thereby decreasing their adhesion to hyaluronic acid [61].

Caveolin-1 and the epidermal growth factor receptor

Epidermal growth factor receptor (EGFR) gene amplification is found in more than 40% of human glioblastomas, with a type III mutation (EGFRvIII) being the most common rearrangement [62, 63]. EGFR mutations increase the proliferative capacity of gliomas and reduce their apoptosis. As a result, tumorigenicity is enhanced, particularly resulting from constitutive tyrosine kinase activity and aberrant EGFR signaling [64–66]. In glioma cells, EGFR has been found to co-localize and interact with cav-1 on the cell membrane [30]. The interaction between EGFR and cav-1, which downregulates its tyrosine kinase activity, has been found to be mediated by the caveolin-binding motif of the receptor [67]. In response to EGF, EGFR rapidly leaves the caveolae domain. This means that phosphorylation-induced dissociation of EGFR from caveolae is related to a more invasive phenotype given the involvement of EGFR-dependent cellular transformation and EGFR trafficking in and out of caveolae [30]. Consistently, it has been found that when the tyrosine kinase activity of EGFR was inhibited in glioma cells, localization of EGFR within caveolae was increased while the transformed cellular phenotype was decreased. On the other hand, in the case of a mutation such as EGFRvIII, receptor internalization and EGFR trafficking from caveolae was found to be disrupted, leading to an abnormal cell behavior due to a prolonged residence of mutant EGFR in the caveolae [68].

Evidence from several preclinical and clinical studies has revealed substantial contributions of PI3K/Akt/mTOR signaling and STAT3 activation to glioma cell survival [69–71]. When activated constitutively, STAT3 was found to be co-expressed with EGFR and to target STAT3/JAK2 sensitized glioma cells to anti-EGFR alkylating agents in vitro [72]. EGFR may also upregulate the gp130/JAK/STAT3 pathway in glioma cells through pro-inflammatory cytokine IL-6 secretion, which is known to be associated with increased glioma cell survival [73]. Moreover, selective inhibition of EGFR phosphorylation has been found to alter the distribution of EGFR towards cav-1-rich lipid domains and, ultimately, reduced Akt/mTOR and gp130/JAK/STAT3 signaling, leading to an arrest of the glioma cell cycle, while inducing apoptosis [74].

Based on these results, cav-1 appears to be a critical player in the suppression of EGFR signaling [75]. Therefore, strategies aimed at inducing EGFR binding to caveolae could be of benefit for the treatment of glioblastomas. More so, therapeutic agents modulating cellular cholesterol levels may disrupt caveolae and exhibit potent anti-tumor effects while decreasing EGFR phosphorylation and inhibiting receptor signaling [76] (Fig. 1).

Fig. 1.

Effect of caveolar localization of EGFR and cav-1-EGFR crosstalk on glioma cell survival. In response to EGF, EGFR leaves its caveolar localization. This phosphorylation-induced dissociation of EGFR from caveolae is associated with its activated tyrosine kinase function, which in turn activates Ras/Raf/MEK/MAPK, PI3K/Akt/mTor and gp130/JAK/STAT3 signaling pathways involved in glioma cell survival. Since interaction between EGFR and cav-1 downregulates the tyrosine kinase activity of the receptor, altered distribution of EGFR towards caveolae potentially reduces EGFR-mediated glioma survival while inducing apoptosis. Cav-1: Caveolin-1, EGF: Epidermal growth factor, EGFR: Epidermal growth factor receptor, IL6: Interleukin-6, IL6R: Interleukin-6 receptor, PDK1: 3-phosphoinositide-dependent protein kinase-1, PI3K: Phosphatidylinositol-4,5-bisphosphate 3-kinase

Caveolin-1 and calcium channels

Evidence indicates that ion channels play an important role in glioma invasion by regulating cell size or by participating in Ca²⁺ signaling in migratory cells [77–79]. Accordingly, migrating glioma cells exhibit oscillatory changes in intracellular Ca²⁺ levels that correlate with the velocity of migration [80, 81]. Hence, ion channels allowing Ca²⁺ influx such as non-voltage-gated calcium channels, i.e., those belonging to the canonical transient receptor potential (TRPC) channel family, have been considered essential for cell signaling during chemotaxis and migration [82]. The expression of TRPC1 channels has been found to be enhanced both in glioma-derived cell lines and in primary patient-derived malignant glioma tissues [83, 84], thereby promoting the migration of tumor cells along an EGF gradient, which was lost after pharmacological inhibition of the channels [85]. These channels were found to be localized in the leading edge of the migrating glioma cells, where they co-localized with markers of caveolae lipid rafts [86]. Accordingly, disruption of the lipid rafts by cholesterol depletion was found to impair the TPRC1-mediated Ca²⁺ influx and, subsequently, chemotaxis toward EGF in the glioma cells. Besides, stimulation with EGF, whose receptor (EGFR) is also localized in caveolae, increased glioma cell chemotaxis, TRPC currents and TRPC1 channel localization to the leading edge of the migrating cells [85]. Based on these findings, the integrity of lipid rafts and activated TRPC channels have been highlighted as unavoidable components of glioma cell chemotaxis [87, 88].

Caveolin-1 and integrins

Integrins, a class of heterodimeric cell adhesion receptors, have been considered as therapeutic targets in various cancers, including glioblastoma [89–92]. Cav-1 is crucial in the signaling and dynamics of integrins. In addition to its regulatory effect on integrin trafficking and internalization [93], cav-1 acts as an adaptor, linking various integrins to growth, adhesion and migration signaling pathways [94, 95]. PCR-based array analysis of glioma cells revealed an additional regulatory role of cav-1 in integrin expression at the transcriptional level. Indeed, cav-1 has been found to act as an integrin suppressor and its expression has been found to be inversely correlated with α5β1 integrin levels in glioma cells [96]. This observation could, at least partly, be explained by a possible escape of p44/42-MAPK signaling from inhibition in case of cav-1 loss, thereby inducing the expression of α5β1 integrins. Concordantly, it was previously reported that in cav-1 knockout animals [97] loss of cav-1 resulted in the hyperphosphorylation of p44/42-MAPK, which in turn shifted glioma cells towards a more proliferative and invasive phenotype [96]. In this manner, α5β1 integrins may play a mediatory role in the phenotypic alterations of the cav-1 manipulated cells, leading to modifications in their growth pattern and clonogenicity. Evidence from a later study has pointed to the TGFβ/TGFβRI/Smad2 pathway in cav-1-mediated expression of α5β1 integrins [37], which is in line with previous reports emphasizing TGFβ as an oncogenic actor leading to increased invasiveness of gliomas and a concomitant poor prognosis [98, 99]. In addition to α5β1 integrins, cav-1 expression was found to be inversely correlated with TGFβ signaling, such that its overexpression antagonized TGFβ and the subsequent phosphorylation of Smad2, which subsequently resulted in a reduced expression of the α5β1 integrins, duly derogating the clonogenic capacity of glioblastoma cells. Depletion of cav-1 has been found to prevent the phosphorylation of ERK1/2 as well, which may serve as an extra and direct route independent of TGFβ signaling in the cav-1-mediated expression of α5β1 integrins through the phosphorylation of Smad2 (Fig. 2). It should be noted that several preclinical and clinical studies targeting the TGFβ pathway have revealed promising results [100–104]. However, as observed in vitro, the co-existence of both cav-1 positive and negative cells within a tumor may lead to both negative and positive effects on α5β1 integrins and TGFβ, respectively [96]. Although the crosstalk between cav-1, TGFβ and integrins in glioblastomas requires further research, based on the current findings, this crosstalk may be one of the reasons for the unsatisfactory response of gliomas to current therapies.

Fig. 2.

Proposed mechanisms of cav-1-mediated expression of α5β1 integrins. Cav-1 keeps p44/42-MAPK signaling under inhibition by prohibiting its phosphorylation-induced activation. Cav-1 also antagonizes TGFβ and TGFβRI activation, precluding phosphorylation of Smad2/3 and subsequent increased expression of α5β1 integrins. Besides TGFβ/TGFβRI, Smad2/3 can be phosphorylated by ERK1/2, which is also downregulated by cav-1. cav-1: Caveolin-1, TGFβ: Tumor growth factorβ, TGFβRI-II: Tumor growth factorβ receptor I-II

Caveolin-1 and connexins

Connexins are structural subunits of gap junctions that mediate direct exchanges of ions and small molecules, thereby contributing to intercellular communication between adjacent cells and various cellular compartments, which has been termed gap junction intercellular communication (GJIC) [105]. A possible protective role of connexin43, the major protein of gap junctions in astrocytes, has been considered as it was suggested to be downregulated in gliomas with a negative correlation between its protein level and the proliferative activity of the tumor [106, 107]. Conversely, others argued for increased migratory capacity, invasive behavior and resistance to chemotherapeutics due to increased connexin43 expression in gliomas, leading to a negative association between its expression and patient survival [108–111]. Interestingly, dynamic changes in connexin43 during tumor progression and its contribution to angiogenesis and other fundamental characteristics of glioblastomas have been reported. Specifically, rapid proliferation of the tumor cells during earlier stages of tumor development and angiogenesis have been associated with downregulation of connexin43, whereas enhanced tumor invasion in later stages was found to be associated with its upregulation [112]. In human glioma cells, a direct interaction between connexin43 and cav-1 has been reported [112]. Connexin43 was found to be linked to cav-1 through its carboxy-terminal, which has previously been shown to be important in the control over glioma invasiveness [113]. In addition, it has been found that disruption of the cav-1-connexin43 crosstalk may result in decreased GJIC [112]. Consequently, inhibition of connexin43-mediated invasive glioma behavior depends on the integrity of lipid rafts and the establishment of proper GJIC between cancer cells and healthy parenchymal cells. This conclusion is actually not surprising as a possible role of abnormal GJIC in carcinogenesis has been reported before [114, 115].

Caveolin-1 and immune cells

Immune cells play an important role in the tumor microenvironment (TME) and the generation of local tumor immunosuppression, neo-angiogenesis, invasion and recurrence [116, 117]. Myeloid cells, particularly macrophages, constitute the dominant immune cell type infiltrating gliomas [118–120]. Glioblastomas have the ability to inhibit the myeloid cell response leading to tumor-induced immunosuppression. Therefore, despite promising results of immunotherapy targeting immune cell infiltration in gliomas [121, 122], strategies addressing the efficacy of tumor-induced immunosuppression are mandatory. Cav-1 is expressed in immune cells [123] and exhibits a suppressive effect on inflammation by reducing proinflammatory cytokine production and endothelial nitric oxide synthase (eNOS) activity while inducing anti-inflammatory cytokines [124]. A search for a single, dominant molecule modulating glioblastoma-mediated myeloid cell suppression revealed cav-1 as candidate, featuring upregulation in myeloid cells [33]. In fact, cav-1 expression was found to be upregulated in these cells in the presence of glioblastoma cells, and impaired myeloid cell function was found to be restored upon inhibition of cav-1 expression. Thus, cav-1 seems to have the potential to regulate glioblastoma-mediated suppression of myeloid cell function. Although current evidence for such a modulation of myeloid cells through cav-1 is still preliminary, pharmacological inhibitors of cav-1 may be of benefit in augmenting the impact of immunotherapy strategies against glioblastomas that contain large numbers of functionally impaired myeloid cells.

Caveolin-1 and tumor necrosis factor receptor CD40

The glioma microenvironment harbors several cytokines, chemokines and growth factors. Among many others, tumor necrosis factor-alpha (TNF-α), a master proinflammatory cytokine of inflammation as well as the TME, plays an important role in glioblastoma progression by enhancing its proliferation, invasiveness and angiogenesis [125]. Several years ago, activated nuclear factor-κB (NF-κB) signaling has been found to be responsible for the resistance of glioma cells to TNF-α-induced apoptosis, as reversal of this activation sensitized the cells to apoptosis [126–128]. CD40 is a member of the TNF receptor superfamily that is expressed primarily by antigen presenting cells and a broad range of tumors including gliomas [129–131]. CD40 mediates the activation of NF-κB involved in resistance of tumor cells to apoptosis [132]. However, although TNF-α has been found to promote its expression in glioblastoma, CD40 failed to activate NF-κB signaling in TNF-α treated tumor cells [133]. This observation may be explained by a decreased anchoring of CD40 in cav-1 rich lipid rafts upon TNF-α exposure, which was critical in formerly reported CD40-mediated NF-κB activation of lymphoma cells [132]. The biological significance of CD40 expression in cancer cells and the exact molecular mechanisms of CD40-mediated apoptosis are yet to be elucidated. Nevertheless, the relatively higher expression of CD40 and its ligand CD40L in lower grade gliomas compared to glioblastomas has been shown to be associated with better clinical outcomes and prolonged survival rates [131]. Based on these data, targeting caveolae and CD40 clustering in them may serve as a potential therapeutic strategy for glioblastomas, as CD40 activity necessarily requires functional caveolae [134].

Caveolin-1 and exosomes

Substantial evidence indicates that hypoxia can promote the survival and dissemination of tumor cells, thereby leading to increased therapy resistance and poor patient outcomes [135–137]. As such, hypoxic regions within solid tumors are considered major determinants of tumor aggressiveness. Similar to other malignant solid tumors, glioblastomas characteristically harbor large areas of hypoxia and necrosis [138]. Hence, hypoxia and angiogenic proteins such as cytokines, growth factors and proteases mediating angiogenesis and remodeling of the extracellular matrix have been suggested as therapeutic targets in various tumor types, including malignant gliomas [139–141]. Exosomes, nanosized extracellular micro-vesicles containing intracellular contents such as mRNAs, microRNAs and proteins, can be transferred to other cells within the tumor mass [142, 143]. In this respect, exosomes can mediate crosstalk between different cell types in the TME [144]. A potential role of exosomes in the hypoxic response of tumor cells was suspected by the higher number of exosomes produced by hypoxic tumor cells compared to normoxic tumor cells [145]. In a mouse glioblastoma xenograft model, it was found that tumors grown in the presence of exosomes exhibited enhanced vascularization compared to control tumors, and that hypoxic exosomes more potently stimulated tumor growth than normoxic exosomes [32] (Table 1). In addition, it was found that exosomes from glioblastoma-bearing mice exhibited increased levels of IL-8, which is a well-known hypoxia-regulated cytokine [146] that plays a role in glioma aggressiveness [147, 148]. It has also been reported that cav-1 can be regulated by hypoxia [149] and it has been encountered in plasma exosomes of cancer patients [150]. Exosomes containing certain microRNAs were found to be internalized by cav-1 and lipid raft-dependent endocytosis, and to be involved in the proliferation, migration and invasion of malignant cells [151]. Concordantly, in glioblastoma patients the number of cav-1 enriched exosomes was found to be more prominent in hypoxic regions. Based on these findings, besides other factors such as IL-8, platelet derived growth factors (PDGFs) and MMPs, cav-1 is considered to be a noninvasive biomarker that reflects the hypoxic signaling and oxygenation status and, thus, the aggressiveness of malignant gliomas [32].

Oligodendroglial tumors

Oligodendrogliomas arise from cells that generate axonal myelin sheaths of the CNS, and account for less than 10% of all diffuse gliomas [152]. Isolated chromosome 1p deletion or co-deletion of chromosome 1p and 19q is being used to define a subgroup of oligodendroglial tumors with an increased response to chemotherapy, regardless tumor grade and, thus, with a longer disease-free and overall survival [152–156]. In addition to this prognostic importance, discrimination between oligodendrogliomas and astrocytomas solely based on the presence of 1p/19q co-deletion in conjunction with an IDH mutation has been accepted lately [157]. Since there is less published literature on oligodendroglioma models than on glioblastoma models, it is not surprising that cav-1 has so far not been subject of such oligodendroglioma research. Furthermore, evidence regarding a role of cav-1 in oligodendrogliomas acquired from exvivo human oligodendroglioma studies is limited and, thus, arguable (Table 1). In a series of 64 gliomas, for example, all the low grade and the majority of the high grade pure oligodendrogliomas were found to be negative for cav-1 expression and this negativity correlated with 1p/19q deletion regardless the grade of the tumor, while all astrocytomas were positive [34]. Besides that, in mixed oligoastrocytomas, wherein cav-1 expression was accompanied by the presence of an astrocytic component, the protein level was found to be directly proportional to the tumor grade. As a consequence, cav-1 was initially believed to be a potential marker of astrocytoma-oligodendroglioma differentiation. In later years, cav-1 expression has been revealed in pure oligodendrogliomas, albeit with a more frequent expression in tumors harboring an oligodendroglial component, such as pure oligodendrogliomas and oligoastrocytomas, compared to pure astrocytomas, without a correlation with the 1p/19q status of the tumor [31]. In another study on oligodendroglial and mixed oligoastrocytic tumors of any grades, although cav-1 was mainly negative in most of the cases, its expression was found to correlate with the grade and type of tumor, when present [35]. The authors concluded that cav-1 expression was associated with a shorter survival and thus a poor prognosis, as the expression levels were inversely correlated with the 1p/19q status. This association was even found to be stronger in higher grade oligodendrogliomas, in which cav-1 expression was the only independent prognostic indicator. Although current data regarding the role of cav-1 expression in oligodendroglial tumors are heterogeneous, yet a possible correlation between cav-1 expression, tumor grade and patient survival is notable. Such a correlation should be confirmed in larger series of primary human tumor samples and in experimental studies evaluating the potential contribution of cav-1 expression to the development and progression of oligodendrogliomas.

Ependymal tumors

Ependymomas are rare tumors of the CNS arising from the ependymal lining of the cerebral ventricles and central canal of the spinal cord, constituting only 2% of all adult intracranial tumors, with a worse prognosis when located at the spinal cord and a shorter progression-free survival when located supratentorially [158–160]. Initially, variations in cav-1 expression according to the grade of the tumor and the EGFR status were evaluated in ependymomas [36], based on evidence indicating upregulated cav-1 expression and its co-localization with EGFR in glioblastomas [30]. Indeed, co-expression of cav-1 with EGFR has been related to poor patient outcomes in a series of 22 adult intracranial ependymomas, in line with earlier reports underlining EGFR as an independent prognosticator of this type of brain tumor [161]. Aside from histological grade, extent of resection, and Ki-67, p53 and EGFR protein levels, which were directly associated with overall patient survival, only cav-1 expression was found to correlate with a poor survival, particularly in low grade ependymomas. Based on this notion, potential molecular pathways linking cav-1 overexpression to unfavorable patient outcomes were explored, but no specific mutations in the relevant genes were encountered. This result is reminiscent of former work divulging a lack of cav-1 gene mutations in human cancers of organs other than the brain [162, 163], although the role of cav-1 mutations in the migration, invasiveness and aggressiveness of the transformed cells has been mentioned in previous in vitro studies [8, 164]. More recently, EGFR and cav-1 were found to be of limited prognostic value in a larger series of adult and pediatric ependymomas. In this study, encompassing 176 ependymomas with intracranial and spinal localizations, less than half of the tumors exhibited EGFR and cav-1 expression and, even when present, their co-expression did not correlate with unfavorable patient outcomes [38]. So, based on current literature, the biological and clinical significance of cav-1 expression in ependymomas remains questionable. The available data so far should, therefore, be considered as preliminary evidence indicating a possible cav-1 involvement in ependymal tumor development with a potential impact on patient prognosis (Table 1).

Caveolin-1 and the challenges of malignant glioma treatment

With respect to chemoradiotherapy, insufficient drug delivery as well as drug resistance are the major challenges faced in the management of malignant gliomas [165, 166]. In this section, we discuss conversant causes of difficulties in the treatment of malignant glial tumors, thereby focusing on the molecular mechanisms involving cav-1.

Limited transport across the blood tumor barrier

The presence of a biological barrier between the CNS and the blood circulation precludes its exposure to chemotherapeutic agents, which is one of the major determinants of drug efficacy [167]. This physical barrier formed by tight junctions together with very low rates of transcytosis are two unique structural features of endothelial cells within the CNS, bringing the blood brain barrier (BBB) forward as a highly specific seal [168–170]. Under pathologic conditions such as the presence of neoplastic tissue, vascular endothelial barrier functions may be altered. Its alterations in permeability may be mediated either through a transcellular or through a paracellular pathway involving caveolae and tight junctions, respectively. Although the BTB is structurally diverse from the BBB in some aspects, limited transportation of anti-tumor drugs is their commonality [171]. The restricted delivery of drugs across the BTB is, therefore, a major obstacle in the treatment of CNS tumors [172]. As such, targeted opening of the barriers allowing sufficient entry of anti-tumor agents may be a promising strategy in the treatment of these tumors. To date, increased caveola-mediated transcytosis has been implicated in an enhanced permeability of the BBB [169, 173–175]. Accordingly, a number of studies, including experimental glioma and BTB models aimed at investigating feasible approaches that would allow selective opening of the BTB without damage to the normal brain tissue, have resulted in targeted caveolae-mediated transport through cav-1 (Fig. 3). It has been found, for example, that low frequency ultrasound (LFU) can effectively open the BTB in experimental models by increasing Ca⁺² influx into endothelial cells leading to activation of eNOS [176, 177]. In addition to increased nitric-oxide levels, it was found that LFU irradiation activated tyrosine kinase Src and subsequently increased the phosphorylation of cav-1, which in turn enhanced tyrosine kinase-dependent caveolae-mediated transcytosis [178, 179]. A well-known vasoactive peptide, bradykinin, was also found to exhibit potential to enhance the delivery of chemotherapeutics across the BTB, either by increasing endothelial transport [180, 181] or by rearranging the actin cytoskeleton, thereby downregulating the expression of tight junctions [182]. Excitingly, it was found that combination of LFU with bradykinin further increased the barrier permeability in vivo, while prolonging the opening time compared to their separate use [183]. It has also been found that Minoxidil sulfate, a selective adenosine 5′-triphosphate-sensitive potassium channel (KATP channel) activator, can selectively open the BTB by increasing cav-1 expression in both the tumor capillary endothelium and the tumor cells [184, 185]. Similar to ATP-sensitive potassium channels, induced activation of calcium-activated potassium channels (Kca channels), which is mediated through FoxO1 upregulation, was found to increase the barrier permeability. FoxO1 is a member of the Fox transcription factor subfamily with a direct regulatory effect on the promoter of the cav-1 encoding gene [186]. In an in vitro BTB model, it was found that induced activation of Kca channels increased the expression of FoxO1, the activated form acting in the nucleus, and led to increased expression of cav-1 and caveolae-mediated transcytosis in a time-dependent manner. Specifically, after an initial and transient positive effect on BTB permeability, sustained activation of the Kca channels activated the ROS/PI3K/protein kinase B signaling pathway, which then led to increased phosphorylated FoxO1 levels in immortalized human cerebral microvascular endothelial cells. Eventually, phosphorylated and deactivated FoxO1 moved to the cytosol causing a decrease in the levels of dephosphorylated FoxO1 and cav-1 [187].

Fig. 3.

Summary of the proposed molecular mechanisms regarding substrates that have the potential to augment the permeability of the BTB through cav-1 and caveolae-mediated transcytosis. LFU increases nitric oxide levels as well as Src tyrosine kinase-mediated phosphorylation of cav-1, which in turn enhances caveolae-mediated transcytosis. Bradykinin enhances potential delivery across the BTB either by increasing endothelial transport or by rearranging the actin cytoskeleton. When used in combination, LFU and bradykinin exhibit an increased effect on barrier permeability compared to their separate use. Potassium channels in the plasma membrane also contribute to the regulation of the permeability of the BTB. Enhanced permeability due to induced activation of Kca channels is mediated through an increased expression of FoxO1, leading to increased cav-1 levels. However, following the initial positive affect, sustained activation of these channels induces phosphorylation of FoxO1 through activation of the ROS/PI3K/PKB signaling pathway. Next, the inactivated form, phosphorylated FoxO1, moves to the cytosol leading to decreased expression of cav-1. VEGF increases the permeability of the BTB through upregulated expression of cav-1 and through the opening of tight junctions. Co-administration of VEGF and papaverin, further enhances the permeability compared to their individual use. EMAP-II activates tyrosine kinase/RhoA/Rho kinase signaling and increases the expression as well as the phosphorylation of cav-1. The T-domain of the diphtheria toxin interacts with cav-1, activates Src kinase and leads to increased cav-1 phosphorylation which, subsequently, phosphorylates and inactivates the repressor nuclear protein, Egr-1. miRNA-132 increases the permeability through the inhibition of PTEN, which has an inhibitory effect on tyrosine kinase Src. Consequently, increased Src phosphorylation augments phosphorylated cav-1 levels and leads to increased transcytosis. In addition to the transcellular pathway, cav-1 contributes to the paracellular pathway by means of increased expression of tight junction proteins. Cav-1: Caveolin-1, DT: Diphtheria toxin, Egr-1: Early growth responsive-1, EMAP-II: Endothelial monocyte-activating polypeptide-II, VEGF: Vascular endothelial growth factor, Kca channels: Calcium-activated potassium channels, LFU: Low frequency ultrasound, miRNA: MicroRNA, NO: Nitric oxide, eNOS: Endothelial nitric oxide synthase, PI3K: Phosphatidylinositol-4,5-bisphosphate 3-kinase, PKB: Protein kinase B, PTEN: Phosphatase and tensin homolog deleted on chromosome 10, TJ: Tight junctions

Vascular permeability factor, also known as vascular endothelial growth factor (VEGF), is a potent growth factor displaying multiple functions in the regulation of angiogenesis and vascular permeability. The highly vascular nature of glioblastomas has long since been attributed to the high VEGF expression levels in these tumors [188]. Concordantly, therapies targeting VEGF are currently considered encouraging in the treatment of malignant gliomas. However, despite promising results from preclinical studies, randomized phase III clinical trials have failed to achieve improved overall survival rates using anti-angiogenic agents alone, or in combination with chemoradiotherapy in newly diagnosed glioblastoma patients [189, 190]. Nevertheless, experimental studies have revealed a contribution of VEGF to the mediation of BTB permeability. Specifically, it has been reported that VEGF may increase the permeability of the BTB in gliomas through an upregulated expression of cav-1, whereupon the number of pinocytotic vesicles increased, as also through the opening of tight junctions [191, 192]. Recent evidence from a rat glioma model revealed that co-administration of VEGF and papaverine, an opium-like alkaloid effective in vasodilatation, further enhanced the BTB permeability compared to their individual use. Similar to VEGF, papaverine was found to affect the BTB permeability through both paracellular and transcellular pathways. Through the paracellular route, VEGF reduced the expression of claudin-5, whereas papaverine caused a redistribution of occluding. Both molecules augmented caveola-mediated transcytosis through the transcellular route [192].

Another candidate for targeted opening of the BTB is endothelial monocyte-activating polypeptide-II (EMAP- II), a pro-inflammatory cytokine exhibiting potential anti-tumor activity by inducing apoptosis and autophagy of the tumor cells, and inhibiting neo-angiogenesis [193]. In regulating BTB permeability through the transcellular route, EMAP-II has been found to increase the expression and phosphorylation of cav-1 along with cav-2 via the tyrosine kinase/RhoA/Rho kinase signaling pathway [194]. In addition, it was found that EMAP-II may contribute to the paracellular pathway, both in vivo and in vitro, by decreasing the expression of tight junction proteins [195, 196]. Also, diphtheria toxin has been found to increase the permeability of the BTB in experimental models by upregulating the expression of cav-1. The T domain of this toxin, which is synthetized by Corynebacterium diphtheriae, interacts with cav-1 through its caveolin-binding motif. As a result, cav-1 is phosphorylated by the activated tyrosine kinase Src. Subsequently, phosphorylated cav-1 was found to cause phosphorylation and inactivation of repressor Egr-1 (early growth responsive-1), a nuclear protein with a transcriptional regulatory function, promoting cav-1 expression [197].

MicroRNAs (miRNA) are short, noncoding RNAs that can regulate gene expression. In recent years, there is a rapidly increasing consideration on the role of miRNAs in pathologic brain conditions including glioblastoma, in particular in tumor progression and resistance to chemoradiotherapy [198–200]. Among many others, increased expression of miRNA-132 in glioma endothelial cells has been found to downregulate PTEN expression (phosphatase and tensin homolog, deleted on chromosome 10), leading to an increase in phosphorylated cav-1 [201]. As PTEN was previously shown to dephosphorylate and inhibit tyrosine kinase Src activity, reduced PTEN levels in glioma cells have been found to be associated with increased phosphorylation of cav-1, increased caveolae-mediated transcytosis and, finally, increased delivery across the BTB.

Although a regulatory role of cav-1 in the transcellular pathway is widely accepted, its contribution to the paracellular route is less well defined. Currently, there is lack of a consensus whether an increase or a decrease in cav-1 expression enhances barrier permeability in the brain through the regulation of tight junction proteins. So far, cav-1-tight junction crosstalk in the BTB and the impact of altered cav-1 expression levels has been studied in brain tumors only once. Importantly, this study, using rat glioma-derived microvascular endothelial cells, indicated a positive regulatory effect of cav-1 on tight junctions. It was found that the permeability of the BTB was increased after cav-1 gene silencing due to reduced expression and opening of tight junctions, while exogenous overexpression of cav-1 resulted in the opposite effect [202]. This finding is in contrast to the permeability enhancer function of cav-1 via the transcellular pathway, in which it directly influences the quantity of endocytic vesicles involved in caveolae-mediated transcytosis.

Collectively, these data support the hypothesis that enhancing caveolae-mediated transcytosis in microvascular endothelial cells of the brain and the tumor cells through the regulation of cav-1 expression has the potential to facilitate the efficacy of chemotherapeutic drugs. In fact, targeting transcytotic transport may represent a promising approach given the lack of physicochemical restrictions, which would hinder the transport of a drug via this way [173, 203]. On the other hand, the possible entrance of unwelcome substances besides the target drug into the brain via the transcellular route should be emphasized. It should also be noted that, in light of evidence of distinct consequences of its up- or downregulation in different routes, a more complex role of cav-1 in striking the balance between the paracellular and transcellular pathways may be highly feasible.

Resistance to chemotherapy

Among all gliomas, glioblastomas are considered incurable as they exhibit a high degree of resistance to contemporary anti-tumor drugs. Unfortunately, effective solutions are yet to be found, although the molecular mechanisms of chemo-resistance have been widely studied. As yet, only a limited number of studies has been aimed at investigating the role of cav-1 in this process, and attributed its possible regulatory role in glioma resistance to chemotherapy.

Currently, temozolomide, an oral alkylating agent used alone or in combination with radiotherapy, is the standard of care in glioblastoma. However, a considerably high proportion of glioblastomas resists this therapy as a result of DNA repair and post-transcriptional regulation of gene expression, impeding the cytotoxic effect of the drug on the tumor cells [204, 205]. Temozolomide has been shown to modify cav-1 expression both in vivo and in vitro, and this effect was, at least partly, found to be mediated through activation of tyrosine kinase Src [206]. Additionally, overexpression of cav-1 was found to be associated with an increased sensitivity of glioblastoma cells to temozolomide [207].

Tamoxifen, a potent estrogen receptor (ER) antagonist that is widely used in breast cancer patients, has been suggested as an emerging option in glioblastomas [208]. Specifically, it has been found that tamoxifen may cause cell cycle arrest and induce apoptosis in tumor cells [209], leading to its primary beneficial effect on chemo-resistant glioblastomas [210]. Results from a recent in vitro glioma study have, for example, revealed ER-α36 expression, which is a novel variant of ER-α involved in tamoxifen resistance in breast cancer [211, 212]. Notably, it has been found that ER-α36 anchors to the glioma cell membrane via cav-1. Based on this finding, it has been concluded that cav-1 may play a role in regulating tamoxifen sensitivity of glioblastoma cells through ER- α36 [213].

P-glycoprotein, also termed multidrug resistance protein-1, is an ATP-dependent efflux pump responsible for the extrusion of toxins, xenobiotics and drug molecules out of cells [214]. High p-glycoprotein expression in astrocytes and endothelial cells at the BBB as well as glioma cells [215, 216] has been proposed to play a significant role in establishing the chemo-resistant phenotype of brain tumors [217, 218]. In view of this, increased caveolae levels in multi-drug resistant cells and a caveolar localization of p-glycoprotein with a direct physical interaction with cav-1 [219] support a possible collaboration of these two proteins in drug transport processes in the brain [219, 220].

Resistance to radiotherapy

Radiotherapy is well established as part of standard malignant glioma care. Similar to chemotherapy, however, resistance may hamper its response and may, ultimately, lead to a dismal clinical outcome [221]. Therefore, studies focusing on the molecular mechanisms underlying radio-resistance in glioblastomas have gradually gained momentum over the years. As such, increased expression of cav-1 following ionizing radiation leading to resistance to DNA damage and a more radioresistant phenotype has been found to occur in a variety of human cancers [222, 223]. The specific role of cav-1 in radio-resistant brain tumors has so far, however, only been examined in a few studies. Firstly, cav-1 expression has been reported to increase following ionizing radiation together with RhoA and survivin, a member of the inhibitor of apoptosis protein family. In a study using human glioblastoma xenografts [224], the authors found, consistent with earlier reports, that increased survivin expression was associated with radio-resistance, whereas overexpression of RhoA led to radio-sensitivity and a decreased proliferative capacity of the tumor cells [225, 226]. In this system, the role of cav-1 overexpression has remained unclear as cav-1 per se failed to affect glioma cell proliferation. In contrast, a more influential role of cav-1 in response to radiation due to interaction with Tie2, a receptor tyrosine kinase widely expressed in endothelial cells and involved in the pathogenesis of several cancers including glioblastomas [227–230], was recently disclosed. Enhanced co-localization of Tie2 with clathrin heavy chains in the cell membrane upon radiation exposure [231] and its trafficking between the cell membrane and the nucleus have been found to be essential in conferring radio-resistance to cancer cells [232]. In glioma cells, Tie2 was found to be actively involved in the regulation of its own trafficking by phosphorylating cav-1, associating with caveolae and, resultantly, mediating the generation of Tie2-cav-1 complexes which then translocated to the nucleus. Eventually, it was found that after interacting with cav-1, Tie2 binding to chromatin and DNA repair complexes in the nucleus resulted in a radio-resistant phenotype of the tumor cells. As anticipated, inhibition of both Tie2 and cav-1 increased the radio-sensitization of the glioma cells, indicating that tightly regulated membrane trafficking of Tie and Tie-cav-1 crosstalk may serve as a practical target in the management of radio-resistance in gliomas [233].

Primary brain tumors of non-glial origin

Until today, among various non-glial primary tumors of the brain, cav-1 has been implicated only in meningiomas and vestibular schwannomas. Below, we discuss the role of cav-1 in these tumor types, particularly in terms of expression patterns and clinical consequences (Table 2).

Table 2.

Studies with human tissues demonstrating cav-1 expression in primary brain tumors of nonglial origin and brain metastasis

Sample Type	Type and Distribution of the Tumors (n)	Control Group (n)	Cav-1 Detection Method	Cav-1 Status (n)	Significance (n)	Associated Findings (n)	Reference
Tumor	Meningioma (62) • Grade I (34) • Grade II (28)	Normal leptomeningeal tissue (10)	Immunostaining for staining intensity and distribution	Positive in 95% of the tumors with variable staining (moderate to strong in 43, evident reaction in more than 50% of the tumor cells in 38). Staining in less than 10% of the cells in the control group	Higher staining intensity and distribution in atypical subtype and parasagittal localization Higher staining intensity and distribution in grade II tumors compared to grade I	Significantly higher cav-1 intensity and distribution in cases with Ki-67 levels equal or greater than 4% Significant association of higher cav-1 intensity and distribution with worse outcome Significantly higher cav-1 intensity and distribution in recurrent meningiomas	[234]
Tumor	Meningioma (62) • Grade I (34) • Grade II (28)	Normal brain tissue (2)	Immunostaining for staining intensity and distribution	Positive in all	Low expression in 55% (34), high expression in 45% (28) of the tumors	Significantly higher histological grade and Ki-67 expression with higher cav-1 expression Significantly higher microvascular density in tumors with higher cav-1 staining intensity and distribution	[235]
Tumor	Meningioma (20) • Grade I (14) • Grade II (5) • Grade III (1)	Normal brain tissue (6)	iTRAQ-based quantitative proteomics analysis, ESI-quadrupole-TOF, Q-Exactive MS, ELISA for protein analysis	–	Increased expression in low and high grade meningiomas	In addition to cav-1, increased expression of several more tissue proteins such as, complement factor B, Y box protein, vinculin, SHC-binding protein, and guanine nucleotide-binding protein G[i] subunit alpha in low and high grade meningiomas	[236]
Tumor	Sporadic vestibular schwannoma (25)	Tibial nerve (3)	qRT-PCR for m-RNA expression SAM to explore differential gene expression Immunostaining for cav-1 positivity	Negative in all	Differential expression of cav-1 Uniform downregulation of both cav-1 mRNA and protein levels (12.9 fold compared to control group)	Differential expression of several more cancer-related genes like TGFB3, VCAM1, EDNRA, GLI1, GLI2, PRKAR2B, and FZD1 ERK pathway in the central core linking these differentially expressed genes	[237]
Tumor	Vestibular Schwannoma (31) • Sporadic (28) NF2-associated (3)	Auricular nerve (2), cervical nerves (2), facial nerve (1), vestibular nerve (1), a nerve from the VIII cranial pair (1), commercial normal human adult Schwann cell (1)	qRT-PCR for m-RNA expression	Downregulated	One of the top 30 downregulated genes (4.9 fold change compared to control groups)	Possible activation of the MET signaling pathway by SPP1, ITGA4/B6, PLEXNB3/SEMA5A and cav-1 in the development and maintenance of vestibular schwannomas	[238]
Tumor	Non-small cell lung carcinoma metastasis (34)	Vascular endothelium for positive control	Immunostaining for cav-1 positivity	Positive in 53% (18) of the metastatic tumors Positive in 21% (7) of the corresponding primary lung carcinomas of the metastases	Significant increase in brain (but not in adrenal) metastases compared to the corresponding primary lung carcinomas	Correlation between cav-1 expression and aggressiveness / progression of the lung cancer	[239]
Tumor	Breast cancer metastasis (50)	Tumor tissue of breast DCIS (50) Tumor tissue of breast IDC (50)	Immunostaining for staining intensity and cav-1 positivity	Positive in 20% (10) Downregulated	Significant decrease in cav-1 expression in brain metastases compared to control groups	pSTAT3 positivity in 68% (34) of the metastatic tumors Significantly increased pSTAT3 levels in the brain metastases compared to control groups Significant negative correlation between pSTAT3 and cav-1 in brain metastases	[240]
Tumor	Non-small cell lung carcinoma metastasis (69)	–	H&E staining and immunohistochemistry for protein expression qRT-PCR analyses for cav-1 gene mutation in 54 of the tumors Cav-1 FISH	Mild to strong positive in 41% (28)	–	Significant association between cav-1 expression of brain metastasis and histologic subtype for adenocarcinomas Significant association between cav-1 expression in the brain metastasis and increased risk of death Cav-1 was an independent predictor of poor survival Significant association of RT with reduced risk of death in patients with cav-1-negative brain metastasis and with increased risk of death in those with cav-1- positive brain metastasis	[241]

Cav-1: Caveolin-1, DCIS: Ductal carcinoma in-situ, ELISA: Enzyme-linked immunosorbent assay, ESI-quadrupole-TOF: Electrospray ionization quadrupole time-of-flight, FISH: Fluorescent in situ hybridization, H&E: Hematoxylin and eosin, IDC: Invasive ductal carcinoma, iTRAQ:Isobaric tags for relative and absolute quantification, MS: Mass spectrometer, NF2: Neurofibromatosis type 2, qRT-PCR: Quantitative reverse transcriptase polymerase chain reaction, RT: Radiotherapy, SAM: Significance analysis of microarrays

Meningiomas

Meningiomas, the most common primary extra-axial tumors of the brain [242], were initially reported to exhibit moderate to strong cav-1 expression in human tissues with significant differences according to the grade and histopathological subtype of the tumor. In particular, recurrent meningiomas and meningiomas with high Ki-67 levels showed high levels of cav-1 expression, which were in turn associated with a relatively poor survival. As a result, increased cav-1 expression was concluded to be a feasible indicator of meningioma aggressiveness leading to worse clinical outcomes and recurrence [234]. In later years, the possible regulatory role of cav-1 in meningioma progression was questioned given its potential to stimulate neo-angiogenesis in other tumors [243, 244]. As expected, a significant positive correlation between cav-1 expression and the micro-vessel density of meningiomas was observed [235]. Even though the pathophysiologic mechanisms underlying the involvement of cav-1 in the neo-angiogenesis of meningiomas are yet to be defined, VEGF-dependent mechanisms have been suspected [235, 245]. This suspicion was based on a considerably reduced VEGF-mediated pathological angiogenesis in cav-1 knockout mice [246] and the ability of VEGF to induce cav-1 expression through Src-mediated phosphorylation [247]. A recent study aimed at investigating cav-1 alterations in the tissue proteome identified increased expression of cav-1 along with many other proteins, such as complement factor B, Y box protein and vinculin, in human meningiomas of different grades, thereby recommending cav-1 as one of the new candidate marker proteins for low and high grade meningiomas [236]. Consequently, knowledge obtained from the abovementioned work indicates that increments in cav-1 expression may serve as a biomarker predicting the grade and recurrence of meningiomas, as well as its prognosis. Nevertheless, the exact role of cav-1 expression in the different steps of meningioma pathogenesis, including growth, invasion, aggressiveness and neo-angiogenesis, remains to be established.

Vestibular schwannomas

Vestibular schwannomas are usually benign and slow-growing tumors of the CNS arising from the myelin-forming Schwann cell sheath and they constitute ~6% to 8% of all intracranial tumors [248]. So far, the identification of mutations in the NF2 gene has been the major culprit in understanding the pathobiology of vestibular schwannomas [249]. Loss of function mutation of the ‘merlin’ tumor suppressor protein encoded by NF2 has been found to be essential for vestibular schwannoma pathogenesis. More than a decade ago, studies aimed at identifying genes and pathways contributing to vestibular schwannoma pathogenesis other than NF2 resulted in an altered expression of several genes including SPARC, RBM5, AKT1, PLAT, PLAU, FGF1 and TP53 [250, 251]. Likewise, a recent microarray-based study of 25 cases unveiled uniform down-regulation of the cav-1 gene in sporadic vestibular schwannomas in favor of a possible tumor suppressor effect [237]. Notably, these findings were confirmed by a later study reporting downregulation of the cav-1 gene and a concomitant deregulation of stress-related proteins, and upregulation of the HEPACAM gene [238]. No significant changes between different tumor characteristics such as the presence of cystic components, which are known to be associated with more unpredictable behavior compared to pure solid vestibular schwannomas, were noted [252, 253]. Nonetheless, the biological significance of alterations in the expression levels of any of these genes, including cav-1, in the pathogenesis of vestibular schwannomas remains to be established.

Metastatic brain tumors

Based on emerging evidence supporting its potential to interact with cell adhesion molecules and to induce loss of polarity during migration [254], variations of cav-1 expression have also been investigated in brain metastases. Interestingly, metastatic tumors of lung origin were found to express cav-1 in a higher percentage of cases compared to their non-metastatic counterparts. In addition, a significant increase in cav-1 expression in brain metastases, but not in other organ metastases, compared to the primary tumors, was reported [239]. Notably, cav-1 expression was found to serve as an indicator of a poor survival and an independent predictor of poor radiotherapy responses in resected brain metastases of non-small cell lung cancer patients [241]. However, the molecular mechanisms underlying brain metastasis involving cav-1 are far from being resolved, and the only available data to date point at activation of STAT3 signaling and subsequent inhibition of cav-1 expression in breast cancer cells. More specifically, it was found that activation of STAT3 inhibited cav-1 transcription by directly binding to its promoter sequence, while cav-1 negatively regulated STAT3 activation, thereby precluding the invasion of brain-metastatic cancer cells [240] (Table 2). Although limited, the above findings indicate a possible contribution of increased cav-1 expression to the pathophysiology and prognosis of patients with brain metastases. Based on these notions, cav-1 may serve as a novel therapeutic target for the future development of individualized therapies given its potential site-specific and cell type-specific expression in metastatic tumors.

Conclusions and perspectives

After it was first suggested to have a role in oncogenesis, cav-1 has gained remarkable attention in cancer research over the years. Today, accumulating evidence has confirmed a substantial involvement of cav-1 in human tumor development, including variations in expression and its association with tumor progression, invasion, metastasis and therapy resistance. In line with in vivo data, a significant correlation between tumor aggressiveness and increased expression of cav-1 has been noted in several studies using ex vivo human glioblastoma tissues. In vitro studies revealed a more complex role of cav-1 and indicated that increased cav-1 expression was associated with tumor hypoxia and glioma-mediated myeloid cell dysfunction, leading to a more aggressive and invasive tumor phenotype. In addition, cav-1 was found to be a critical player in the suppression of EGFR signaling as well as the inhibition of integrin expression. Similarly, it was found that disruption of caveolae lipid rafts impaired TPRC1-mediated Ca²⁺ influx and subsequent chemotaxis, while the inhibition of connexin43-mediated invasive glioma behavior was found to be dependent on the integrity of the lipid rafts. Cav-1 was also found to act at various stages and in multiple pathways affected by brain tumor treatment. In vitro studies revealed promising results indicating that upregulated expression of cav-1 enhanced caveolae-mediated transcytosis in microvascular endothelial cells of the brain and the tumor cells, leading to increased permeability of the BTB and facilitating the efficacy of chemotherapeutic drugs. On the contrary, although overexpression of cav-1 resulted in increased sensitivity to temozolomide, it also resulted in increased radio-resistance in glioblastoma cells. Despite varying data regarding the involvement of cav-1 in oligodendroglial as well as ependymal tumors, more consistent evidence was obtained from studies evaluating the cav-1-meningioma association, which pointed at cav-1 as a potential indicator of meningioma aggressiveness giving rise to worse clinical outcomes and high recurrence rates. While cav-1 has been suggested to act as a potential tumor suppressor in vestibular schwannomas, increased expression of cav-1 was found to be strongly associated with a poor radiotherapy response and a poor survival in patients with metastatic brain tumors.

Collectively, the data discussed in this review highlight that cav-1 has the potential to be pivot in the molecular mechanisms underlying the pathobiology and behavior of brain tumors. This critical contribution may, at least partly, be attributed to its co-localization with illustrious players in tumorigenesis within the lipid-rich domains of the plasma membrane. For this reason, the ability of cav-1 to interact with molecules involved in cell signaling, as well as the impact of caveolae depletion on important signaling pathways in brain tumor pathogenesis, is noteworthy. Moreover, the regulatory effect of cav-1-governed transcellular vesicular transport mechanism on BTB permeability is exciting and could be of benefit to overcome the restricted transport of therapeutic drugs across brain barriers. In addition, despite limited data, evidence regarding associations between cav-1 expression and brain tumors other than gliomas deserves to be underlined.

Authors’ contributions

PEO analyzed the literature, wrote the review and draw the figs. UO structured the sections, designed and substantially contributed to the development of the review. JT and JHZ supervised and critically revised the paper. All authors have read and approved the submitted manuscript.

Compliance with ethical standards

Conflict of interest

The authors declare no conflict of interest.