Intersectin scaffold proteins and their role in cell signaling and endocytosis

Abstract

Intersectins (ITSNs) are a family of multi-domain proteins involved in regulation of diverse cellular pathways. These scaffold proteins are well known for regulating endocytosis but also play important roles in cell signaling pathways including kinase regulation and Ras activation. ITSNs participate in several human cancers, such as neuroblastomas and glioblastomas, while its downregulation is associated with lung injury. Alterations in ITSN expression have been found in neurodegenerative diseases such as Down Syndrome and Alzheimer’s disease. Binding proteins for ITSNs include endocytic regulatory factors, cytoskeleton related proteins (i.e. actin or dynamin), signaling proteins as well as herpes virus proteins. This review will summarize recent studies on ITSNs, highlighting the importance of these scaffold proteins in the aforementioned processes.

1. Introduction

ITSN was first isolated from Xenopus laevis oocytes during an expression cloning screen with a Src homology 3 (SH3) class II peptide ligand [1]. The 1270 amino acid protein contained two Eps15 homology (EH) domains, a coiled coil (CC) domain and five SH3 domains. The name intersectin derived from the hypothesis that this scaffold protein could potentially bind numerous targets through each of its modular domains creating a macromolecule [1]. ITSN orthologs are found in all metazoans. Mammals have two ITSN genes, ITSN1 and ITSN2, while lower organisms usually encode only a single isoform [2]. A special case is the fish Oryzias latipes, also known as the Japanese rice fish, which shows a genome duplication resulting in three ITSN genes. ITSN genes frequently encode two major splice forms referred to as ITSN long (ITSN-L) and ITSN short (ITSN-S). The longer isoforms are co-linear with ITSN-S but possess an extended C-terminus encoding three additional domains: a Dbl homology (DH) domain, a Pleckstrin homology (PH) domain and a C2 domain. The DH and PH domains function in concert as guanine nucleotide exchange factors (GEFs) for Rho family GTPases revealing that ITSNs participate in GTPase regulation [3]. These additional domains are frequently absent in lower eukaryotes (reviewed in [2]) suggesting that the primordial functions of ITSNs are retained within the domains of the short isoform. In addition to these two major splice variants, there are numerous minor splice variants of ITSNs, a number of which alter the binding specificities of various domains within the protein (reviewed in [2]). Both ITSN genes and their isoforms play important roles in a variety of biological pathways as described in detail in this review.

2. Structure and subcellular localization

The modular nature of ITSNs suggests that these scaffolds associate with multiple protein ligands (Figure 1). Indeed, the EH domains bind proteins targets containing Asp-Pro-Phe (NPF) motifs including endocytic components such as Epsin, Dab2, Numb, and Stonin2, among others [4–6]. The CC domains promote homo- and heterodimerization of ITSNs [7] as well as interaction with proteins from the endocytic (e.g. Eps15) [5] and exocytic machinery (e.g. SNAP-25) [8]. The SH3 domains bind a wide variety of proteins involved in diverse processes including endocytosis (e.g. dynamin and synaptojanin) [1, 9, 10], actin rearrangement (e.g. N-WASP) [11], cell survival (e.g., PI3KC2β) [12], and cell signaling (e.g., Sos, Cbl, and RALT) [13–16] (reviewed in [2, 7, 17]). Some ITSN orthologs in lower eukaryotes have fewer SH3 domains. For example, D. melanogaster and A. gambie ITSNs only possess four SH3 domains. The DH-PH domains in ITSN-L specifically activate Cdc42 but not other Rho family members [3, 18]. The C2 domain of ITSN-L is also implicated in GTPase regulation as well as membrane trafficking [19]. The ITSN ortholog from the pathogenic fungus C. neoformans, Cin1, is a significantly more divergent member of the ITSN family [20]. In contrast to ITSN members from many lower eukaryotes, Cin1 is expressed as both short and long isoforms similar to mammalian ITSNs, with the long isoform present at roughly twice the levels of the shorter isoform [20]. The Cin1 long isoform lacks the C2 domain and only possesses two SH3 modules. Finally, Cin1 is unique in possessing a WH2 (WASP homology 2) domain which functions as an actin monomer-binding motif (Figure 1) (reviewed in [21]).

Figure 1. ITSN short and long isoforms.

Human ITSN is comprised by 2 EH domains, 1 CC domain and 5 SH3 domains. Additionally, the long isoform contains a DH, a PH and a C2 domain. Both short and long isoforms are also found in mice. Invertebrates usually lack the long isoform, as shown in the worm C. elegans. Besides, some organisms as Drosophila only exhibit 4 SH3 domains. The pathogenic fungus C. neoformans is an exception, since the DH and PH domains are present in its structure. Furthermore, it contains a single EH domain and 2 SH3 domains, with an additional WH2 domain. Abbr: H. sapiens: Homo sapiens; M. musculus: Mus musculus; C. elegans: Caenorhabditis elegans; D. melanogaster: Drosophila melanogaster; C. neoformans: Cryptococcus neoformans.

The expression and subcellular localization of ITSN isoforms varies. ITSN1-S is expressed in many tissues localizing throughout the cytoplasm and perinuclear regions [22]. Both ITSN1-S and ITSN2-S localize to clathrin-coated vesicles [22, 23] and ITSN1-S is also associated with small invaginations of the plasma membrane named caveolae [24]. ITSN1-L is neuron-specific where it regulates the dendritic spine development [11, 25, 26] while ITSN2-L is more widely expressed and found in the cytosol of endothelial cells and associated with vesicular structures [27].

3. Function

3.1 Endocytosis and exocytosis

One of the best known functions of ITSNs is their role in endocytosis. Multiple studies have described interactions between different domains of ITSNs and components of the endocytic machinery including dynamin, AP2, FCHo1/2, Eps15 and Epsin family proteins [6, 7, 28–30]. ITSNs are involved in membrane trafficking by clathrin- and caveolin-mediated endocytosis [31]. Loss-of-function mutations in the C. elegans ITSN ortholog lead to reduced synaptic vesicles and a reduction in their recycling. In Drosophila, mutation of dap160 is lethal. Interestingly, silencing ITSNs in mammals leads to defects in endocytosis although total loss of murine ITSN genes does not lead to lethality [32, 33].

The large GTPase dynamin, which is involved in plasma membrane fission during clathrin-mediated endocytosis, co-localizes with ITSNs on clathrin-coated vesicles [34, 35] and binds ITSNs SH3 domains [1, 8, 36, 37]. Clathrin accumulation is coordinated by several endocytic adaptors, such as adaptor protein 2 (AP2) and Dab2. AP2 allows concurrent binding to clathrin and the lipid PIP₂, while Dab2 recruits EH domain scaffold proteins including ITSNs to regulate integrin β1 endocytosis [6]. ITSN1 together with Eps15 forms a complex with FCHo1/2, resulting in the initiation of clathrin-coated pit (CCP) formation [28].

ITSN2 is also involved in endocytosis. As with ITSN1, ITSN2 localizes to endocytic vesicles and overexpression [23] or silencing of ITSN2 alters endocytosis [28]. ITSN2 also interacts with dynamin and FCHo1/2 proteins to define the sites of CCP formation [28]. Crystallization of the C2 domain of human ITSN2-L suggests a role in the assembly of endocytic vesicles [19]. The structure of the C2 domain of ITSN2 is highly similar to the C2 domains in synaptotagmin I and rabphilin. Specifically, the high degree of structural identity between the core β-sandwiches of these three C2 domains suggests that this β-sandwich represents a scaffold for vesicular transport [19]. Additional evidence for the role of ITSN2-L in endocytosis is provided by its interaction and co-localization with RABEP1 (Rab GTPase-binding effector protein 1, also named Rabaptin 5) [7], a protein involved in endocytic membrane fusion and membrane trafficking of recycling endosomes. RABEP1 regulates early endosomal transport through interaction with small Ras-related GTPases, Rab4 and Rab5. Overexpression of ITSN2-L represses RABEP1-enhanced endosome aggregation and induces its degradation [38]. ITSN1 also interacts with RABEP1/Rabaptin 5 [7] although its role regulating RABEP1/Rabaptin5 degradation has not been examined.

ITSNs also regulate exocytosis; however, their precise role in this process remains unclear. Okamoto and colleagues first described the direct interaction between the CC domain of ITSN1 and the SNAREs SNAP23 (ubiquitous) and SNAP25 (specific to brain), which are essential for regulated exocytosis in a variety of cell lines [8]. After reporting that Cdc42 facilitated exocytosis in neuroendocrine cells, Malcombe et al. demonstrated that ITSN1-L regulates exocytosis through activation of Cdc42 [39]. The same group showed that ITSN1 was present at exocytic sites in PC12 and chromaffin cells and found that ITSN1 is an essential component of the exocytic machinery in these cells [40]. In support of these findings, ITSN1-knockout mice exhibit defective exocytosis in chromatin cells [32]. Together, these findings suggest that ITSN1-L may provide a link for the exo-endocytotic membrane trafficking in secretory cells.

3.2 Actin cytoskeleton rearrangement

Rho GTPases are key regulators of the actin network [41]. ITSN-L isoforms act as GEFs for Cdc42, which induces the formation of actin-rich surface protrusions called filopodia to regulate cell polarity and actin cytoskeleton. One of the effectors of Cdc42 that allows filopodia formation is the Wiskott-Aldrich syndrome protein (WASP) and its variant N-WASP which are expressed in hematopoietic cells and ubiquitously, respectively. Both proteins are ITSN binding partners [3, 7], that also interact with WIP (WASP interacting protein), a widely expressed actin-binding protein that regulates stability, location and function of WASP and N-WASP (Figure 2). WIP participates in cellular signaling, endocytosis and actin cytoskeleton remodeling (reviewed in [42]) and is present in actin-rich membrane protrusions that regulate matrix degradation in cells [43]. Association between ITSN and WIP was first suggested after identifying WIP as a binding partner of ITSN2 in a yeast two-hybrid screen [7]. WIP interacts with ITSN1 and ITSN2 in vitro and in vivo [44]. Furthermore, interaction between ITSN and N-WASP activates N-WASP, which binds the Arp2/3 (actin-related proteins 2 and 3) complex to stimulate actin filament nucleation and initiate actin polymerization (Figure 2) [44].

Figure 2. Actin polymerization pathway.

ITSN-L interacts with the Rho GTPase Cdc42, its effector N-WASP and the WASP interacting protein WIP. N-WASP and WIP also bind each other. Interaction between ITSN-L and N-WASP activates N-WASP, which binds the Arp2/3 complex, resulting in the stimulation of the actin filament nucleation and initiation of actin polymerization.

The role of ITSN in actin regulation has also been reported in other organisms. ITSN2-L regulates Cdc42 activity and actin cytoskeleton dynamics during early Xenopus development [45]. In Drosophila, Dap160 is enriched in plasma membrane fractions, and localized to areas adjacent to the active zone [34]. Dap160 interacts with Nervous wreck (Nwk) and together with Cdc42 promotes Wsp (WASP ortholog in Drosophila)-mediated actin polymerization [46]. Both Dap160 and Nwk bind the actin cytoskeleton regulator Wsp and mutations in these genes exhibit similar phenotypes, characterized by synaptic bouton overgrowth in neuromuscular junction (NMJ) synapses [47]. Interestingly, Nwk expression levels are severely reduced in dap160 mutant synapses, suggesting that this abnormal morphology may be due in part to deregulation of Nwk [34]. Overall, these observations demonstrate that ITSN plays an important role in actin cytoskeleton rearrangement.

3.3 Signaling pathways

ITSNs are also key participants in a variety of cellular signaling pathways as outlined below:

A) Ras Family GTPases

ITSN1 regulates Ras family GTPases, including Cdc42, Rac, and Ras (reviewed in [17, 48]). Ras proteins cycle between an active GTP-bound form and an inactive GDP-bound form. GEFs bind to Ras-GDP resulting in the release of GDP and formation of nucleotide-free Ras (nfRas). The resulting nfRas is loaded with GTP due to the high concentration of GTP vs GDP in the cytoplasm resulting in the formation of active, RasGTP. Inactivation of Ras occurs by hydrolysis of GTP to GDP through the intrinsic GTPase activity of Ras. However, due to the relatively low enzymatic activity of Ras, this function is enhanced through the action of GTPase activating proteins (GAPs), which stimulate Ras-mediated GTP hydrolysis returning Ras to the inactive GDP-bound state (Figure 3).

Figure 3. Ras-PI3KC2β-ITSN1 model.

Ras GTPases cycle between their inactive Ras-GDP and their active Ras-GTP states. Binding of GEFs to Ras-GDP causes the release of the GDP, leading to a nucleotide-free Ras (nfRas). This nfRas gets loaded with nucleotide due to the high concentration of GTP in the cytoplasm, generating active Ras-GTP. The presence of GAPs stimulates the hydrolysis of GTP by the intrinsic GTPase activity of Ras, returning Ras to its inactive Ras-GDP state. Our group showed that nfHRas interacts with PI3KC2β, stabilizing HRas in such state and blocking PI3KC2β’s activity. However, binding of ITSN1 to PI3KC2β causes the dissociation of the nfHRas:PI3KC2β complex, which activates the kinase and at the same time, allows for the transition of nfRas to the Ras-GTP bound form. GEF: Guanine exchange factor. GAP: GTPase activating protein.

As noted earlier, ITSN-L isoforms serve as GEFs for Cdc42. However these scaffolds regulate additional Ras family GTPases independent of the DH-PH region. ITSN1 overexpression induces Ras activation on intracellular vesicles resulting in stimulation of signal transduction pathways [49]. This activation of Ras by ITSN1 may be through recruitment of the Ras GEF Sos1 (Son of Sevenless homolog 1), through the SH3 domains of ITSN1-S binding the Pro-rich region of Sos1 [14]. Consistent with this model, overexpression of the SH3 domains of ITSN1 blocks epidermal growth factor-mediated Ras activation and activation of the ERK-MAPK pathway [50]. However, an alternative mechanism has recently been described [12]. ITSN1 interacts with and activates a novel phosphatidylinositol 3-kinase (PI3K), PI3K class II beta (PI3KC2β) [51]. PI3KC2β has a Ras binding domain (RBD) and co-localizes with Ras and ITSN1 on vesicles. Interestingly, PI3KC2β forms a complex with H-Ras when the latter is in its nucleotide-free state, unlike class I PI3Ks that bind Ras in its active, GTP-bound state [12]. This nucleotide-free H-Ras complex inhibits the lipid kinase activity of PI3KC2β while simultaneously preventing H-Ras from loading with nucleotides. The interaction of ITSN1 with PI3KC2β has been proposed to stimulate the dissociation of nfHRas:PI3KC2β complex, allowing Ras to load with GTP resulting in its activation on intracellular vesicles (Figure 3). In addition, disruption of the nfHRas:PI3KC2β leads to loss of PI3KC2β repression by Ras, thereby leading to its activation. Even though this model has only been examined for H-Ras, a similar mechanism may apply to other Ras isoforms (K-Ras and N-Ras), although experimental evidence is currently lacking. This model may provide a potential explanation for the observed tumor suppressor activity of WT RAS alleles toward mutationally activated RAS [52]. Ras activating mutations are found in about 30% of human cancers. In such cases, Ras is trapped in the GTP-bound state due to its reduced ability to hydrolyze GTP to GDP, which results in increased activation of MAPK pro-proliferation and AKT anti-apoptotic pathways. However, WT Ras is predominantly found in its GDP state and when overexpressed it can suppress the transformation induced by the oncogenic form [53, 54]. According to the above model, PI3KC2β would compete for nucleotide-free WT Ras, after the dissociation of GDP from WT Ras [12]. Capturing WT Ras in its nf state would prevent GTP loading and thereby reduce the activation of the abovementioned downstream pathways.

B) TGFβ receptor I kinase/Alk5 pathway

As described in section 3.1, ITSN1 knockdown results in defects in both caveolae- and clathrin-mediated endocytosis. The impairment of these pathways leads to the upregulation of alternative pathways that include the presence of enlarged endocytic structures. These endocytic structures internalize the transforming growth factor β (TGFβ) receptor I, also named Alk5, resulting in its degradation [55, 56]. The effects of such degradation are important, since the TGFβ pathway is involved in cell proliferation, differentiation and apoptosis. This pathway begins with the capture of TGFβ by TGFβ receptor II dimers, which immediately recruit and catalyze the phosphorylation of a type I receptor (Alk5) dimer. Both dimers form a heterotetrameric complex that can activate either the Smad2/3 proteins or Smad-independent pathways [57] (Figure 4). The canonical TGFβ signaling pathway includes the recruitment of R-Smad by the Smad anchor for receptor activation (SARA), followed by R-Smad dissociation and subsequent activation of Smad2/3. Alternatively, the Alk5/TGFβRII heteromeric complex recruits the adaptor ShcA [58] followed by the formation of ShcA/Grb2/Sos complex and subsequent activation of the Ras-Raf-MAPK cascade (Figure 4). Since ITSN and Grb2 compete for Sos [14], reduced levels of ITSN in cells may increase Sos availability for Grb2, favoring the Alk5-Sos1-Grb2 pathway over the Alk5-Smad-SARA route. Studies in knockdown ITSN1-S mouse lungs demonstrate a preference for the Alk5-Sos1-Grb2 pathway leading to persistent activation of ERK1/2 and the proliferation of pulmonary endothelial cells [55]. These findings suggest that ITSN attenuates Ras-ERK activation by the TGFβ pathway in contrast to previous work suggesting that ITSN promotes Ras activation [50, 59].

Figure 4. TGFβ-RI/Alk5 pathways.

TGFβ is first captured by TGFβ receptor II (in brown) and TGFβ receptor I or Alk5 (in green) then to form a complex that may activate Smad proteins (left branch) or the Smad independent pathway (right branch). Canonically, SARA (Smad anchor for receptor activation) recruits R-Smad, allowing for its phosphorylation, which causes a conformational change that leads to R-Smad dissociation. Phosphorylated R-Smad forms a complex with Smad4, which consequently activates Smad2/3 proteins. Alternatively, the TGFβ-RII/TGFβ-RI tetrameric complex binds ShcA, which recruits Grb, that in turn, binds Sos. This complex activates Ras, leading to ERK activation. Abnormal levels of ITSN1 may shift the balance between SARA/Smad and ShcA/Grb2/Sos pathways. Since ITSN and Grb compete for Sos, reduced levels of ITSN1 result in an increased Sos availability for Grb and may favor the ShcA/Grb2/Sos pathway.

C) Receptor tyrosine kinases (RTKs)

RTK signaling pathways are critical for normal cell growth, differentiation and development, and alterations in RTK regulation contribute to many diseases including cancer. Endocytosis is essential for the regulation of RTK function, and a number of studies reveal that ITSN acts as a nexus between the endocytic process and RTK signal transduction. ITSN1 promotes the ubiquitylation and degradation of the epidermal growth factor receptor (EGFR) tyrosine kinase through the activation of Cbl, an E3 ubiquitin ligase [13]. ITSN activates Cbl through a complex network of protein:protein interactions. In addition to binding Cbl, ITSN binds Spry2, a Cbl inhibitor, and the tyrosine phosphatase Shp2, a Cbl activator [16, 60]. These interactions lead to decreased tyrosine phosphorylation of Spry2, which disrupts binding to Cbl’s N-terminal SH2-like domain, thereby resulting in Cbl activation. In addition to Spry2, ITSN1 interacts with additional Cbl regulatory proteins, including ALIX and CIN85 [7] suggesting a complex network of ITSN-mediated interactions in the regulation of Cbl, and therefore RTK ubiquitylation.

ITSN1 and ITSN2 interact with RALT (receptor-associated late transducer), also known as MIG6 (Mitogen-inducible gene 6 protein). RALT is a negative regulator of EGFR which inhibits the catalytic activity of the receptor by binding the kinase domain [61]. Interestingly, RALT is capable of driving internalization of non-phosphorylated, non-ubiquitylated EGFR, leading to EGFR degradation. The endocytic domain of RALT responsible for this internalization binds the SH3 regions of ITSN1 and ITSN2. Silencing of ITSNs reduces RALT-mediated endocytosis of EGFR, with ITSN2 knockdown exhibiting the higher effect [15]. Thus, RALT couples the EGFR to activation-independent, clathrin-mediated endocytosis by interaction with ITSNs.

Ephrin type-B receptor 2, EphB2, is another RTK that co-localizes with ITSN1. EphB2 mediates important developmental processes in the nervous system and stimulates ITSN1-L’s GEF activity together with Numb to promote dendritic spine formation [11, 25]. The interaction between ITSN1 and EphB2 is mediated by the kinase domain-containing fragment of EphB2 and the N-terminal region of ITSN1 [11]. EphB2 activation induces the clustering of Numb with the receptor along dendrites suggesting that Numb is recruited to EphB2 receptor complexes in such conditions [25] (further discussed in section 3.5).

D) Kinases and phosphatases

An important family of protein kinases affected by ITSN is the mitogen activate protein kinases (MAPKs) group. Although ITSN1 stimulates Ras activation, overexpression of ITSN1-S does not activate the ERK pathway [49]. Instead, ITSN stimulates EGFR internalization, which indirectly affects ERK activation [13]. ITSN1 stimulates gene expression through activation of a JNK-dependent but ERK-independent pathway [59]. However, ITSN1 and EGFR cooperate to enhance transcriptional activation and this synergy requires ERK-MAPK activity [59]. As discussed above, ITSN1 attenuates Ras-ERK activation by TGFb in lung cell [55]. Thus, ITSN1 activates JNK and indirectly regulates ERK activation by receptors.

ITSN also interacts with WNK (with-no-lysine) kinases, characterized by an abnormal location of the catalytic lysine. Mutations on WNK1 and WNK4 result in hypertension and hyperkalemia. Both kinases stimulate clathrin-dependent endocytosis of ROMK1 (renal outer medullar potassium 1) and binding of ITSN1 to WNK1 or WNK4 is essential for such stimulation [62]. As discussed in 3.3A, ITSN1 regulates PI3Ks. These lipid kinases phosphorylate the 3′ hydroxyl group of phosphatidylinositol. ITSN1 interacts with and activates PI3KC2β [12, 51] and reduced expression of ITSN1 inhibits the PI3KC2β-AKT pathway [51, 63].

While phosphorylation of molecules represents a critical event in cell signaling, dephosphorylation plays an equally important role in regulating this process. ITSNs are also implicated in regulating several phosphatases. ITSN1 binds the SH2 domain-containing inositol 5-phosphatase 2 (SHIP2). In response to epidermal growth factor, SHIP2 recruits ITSN1 to the plasma membrane [64]. ITSN1 also binds synaptojanin, a lipid phosphatase with an important regulatory role in vesicle uncoating in neurons [34, 35]. Both SHIP2 and synaptojanin belong to the 5′ phosphoinositide phosphatase family, which dephosphorylate PIP₂ and PIP₃. As discussed in Section 3.3C, ITSN1 binds the tyrosine phosphatase Shp2 to regulate the tyrosine phosphorylation state of Spry2 and Cbl activity [16].

3.4 Cancer and lung injury

ITSNs are also involved in human tumorigenesis. Overexpression of either ITSN1-L or ITSN1-S promotes oncogenic transformation of fibroblasts [49, 65]. ITSN1 contributes to the tumorigenic properties of several human cancers including neuroblastoma and glioblastoma [66, 67]. Furthermore, examination of the ONCOMINE™ database reveals that ITSN1 is over-expressed in several human cancers including pancreatic carcinomas, liposarcomas and Wilm’s tumor, suggesting a broader role for this scaffold in human cancers.

Neuroblastoma (NBL) is the most common extracranial solid tumor in children, usually affecting children under the age of five. These tumors are biologically and clinically highly heterogeneous [68]. Characterization of more than 1000 tumors reveals a lack of common genetic alterations in these tumors [69]. Although amplification or overexpression of the MYCN protooncogene is present in roughly 20% of NBLs, the molecular drivers of the remaining tumors remain unclear [69]. Recently, ITSN1 was found to be expressed in both MYCN amplified and non-amplified human NBL tumors and cell lines [67]. More importantly, silencing ITSN1 expression decreased anchorage-independent growth in vitro and tumor growth in athymic nude mice [67]. The ITSN1 target PI3KC2β, also contributes to NBL tumorigenesis, although its role was not as significant as ITSN1 [63], suggesting that additional ITSN1-regulated pathways are involved in NBL tumorigenesis.

ITSN1 has also emerged as a potential target in Glioblastoma multiforme (GBM). GBM represents the largest and most malignant subgroup of gliomas, which in turn, represent 80% of malignant brain tumors. GBMs are highly invasive and their prognosis remains poor. Treatment regimes include surgery when possible, radiotherapy and frequently chemotherapy. However the median survival time of patients typically does not exceed 15 months [70]. Thus, defining the critical signaling pathways involved in GBM tumorigenesis represents an important step in developing therapies for treatment of GBMs. Recent studies indicate that ITSN1-S is involved in migration and invasion of human glioma cells [71] and is essential for malignant glioma proliferation [66].

More recent work suggests that ITSN1 may contribute to lung dysfunction. ITSN1-S is present in lung endothelial cells (ECs), and is involved in regulating lung vascular permeability and ECs survival [72]. ITSN1 knockdown results in poor recruitment of dynamin 2 to endocytic sites, interfering with the detachment of caveolae from plasma membrane. The limited formation of free vesicular carriers results in deficient endocytosis leading to increased endothelial barrier dysfunction and pulmonary edema [72].

ITSN2 has also been implicated in cancer; however, in contrast to ITSN1, elevated ITSN2 levels in breast cancer patients is associated with longer disease-free survival of patients [73], suggesting that ITSN2 may play a protective or tumor suppressive role in tumorigenesis. Future work will be necessary to define the precise mechanisms contributing to the contrasting roles of ITSN1 and ITSN2 in oncogenesis.

3.5 ITSN in neurons and neurodegenerative diseases

ITSN1 is known to be a major binding partner for the neuronal GTPase dynamin1 and high expression levels of ITSN1 are found in neurons. In C. elegans ITSN is found in the nervous system and enriched in presynaptic regions [74] while Dap160 is present both in central and peripheral neurons in D. melanogaster [35, 75]. In the adult rat brain, ITSN1 is found in layer III of the neocortex, hippocampus, globus pallidus, subthalamic nucleus, and substantia nigra [76]. Although both ITSN1 isoforms are expressed in the brain, ITSN1-L is specific to neurons whereas ITSN1-S is restricted to glia [32]. Synaptic transmission occurs upon stimulation of synaptic vesicle fusion with the plasma membrane releasing neurotransmitters into the synaptic cleft. For fast neurotransmission, a pool of release-ready synaptic vesicles termed the readily releasable pool is required. In neurons from the auditory brainstem, ITSN1 is crucial for the replenishment of this vesicle pool independent of endocytic membrane retrieval [77]. While mutations in ITSN in both Drosophila and C. elegans lead to defects in synaptic vesicle recycling [34, 35, 74, 78], ITSN1 silencing in rat hippocampal neurons did not affect this process, although defects in transferrin internalization and dendritic spine development were detected [26]. Among invertebrates, ITSN1 is found in pre-synaptic regions; however, in mammals localization is mostly post-synaptic, particularly in dendritic spines. Although the specific mechanism(s) that accounts for these differences is unknown, the additional domains within ITSN-L may contribute to these observations.

In Drosophila, loss of Dap160 ultimately results in lethality [34, 35]. In contrast, ITSN1 null-mutants in C. elegans are viable [74, 78], although each of these genetic models exhibit enlarged, irregular vesicles [34, 35, 74, 78]. Similarly, knockouts of one or both murine ITSNs are also viable but other defects are noticeable [32, 33]. At weaning (3 weeks of age), ITSN1 KO mice are smaller than their wild type mice and about 10% of them fail to thrive. Despite that, ITSN1 null mutants’ weight is recovered by 3 months of age, although brain mass relative to body is smaller than in wild type mice. ITSN2 KO brains are indistinguishable from WT mouse brains and the average weight of double mutant mice at weaning is no different from ITSN1 KO mutant, indicating that loss of ITSN1 affects early weight gain and brain size in mice. Hippocampal basal synaptic transmission and plasticity at Schaffer CA1 synapses is not affected by loss of ITSN1 and/or ITSN2. However, brain structure abnormalities are detected in ITSN1 mutant mice, such as the absence of midline corpus callosum. Loss of ITSN2 does not have such effect. Additionally, ITSN1 KO results in severe deficits in learning and memory, not detected in ITSN2 KO mice [33]. Deletion of ITSN1 also affects chromaffin cells exocytosis, which is impaired even in the absence of ITSN1-L alone [32]. In endothelial cells, silencing ITSN1-S reduces caveolae-dependent endocytosis [24, 79]. However, in multiple cell types, overexpression of ITSNs inhibits endocytosis (reviewed in [17, 48]). The fact that both silencing and overexpression of ITSN result in similar endocytic defects indicates the importance of tightly regulating ITSN expression levels for controlling these biochemical processes.

An important post-synaptic function of ITSN1-L is the regulation of the dendritic spine development. ITSN1 interacts with various proteins involved in neuron formation and development. As noted earlier, an important protein in this process is Numb, which interacts with both ITSN1 and ITSN2 isoforms. Numb localizes to dendritic spines and regulates spine development in hippocampal neurons [25]. Suppression of Numb results in the reduction of protrusion density and length. Numb forms a complex with EphB2, NMDA-type glutamate receptors together with ITSN1 in the post-synaptic membrane. Stimulation by Ephrin-B promotes the clustering of Numb with EphB2 in dendrites [25]. Furthermore, binding of Numb to ITSN1-L stimulates its GEF activity toward Cdc42 to promote spine morphogenesis [11, 25]. These results are consistent with the essential role of Cdc42 in stimulating filopodia and neurite outgrowth during brain development [80, 81].

Another binding partner of ITSN is endophilin A. Besides its implication in endocytosis, endophilin A participates in synaptic vesicle (SV) recycling. ITSN regulates endophilin function and aids in recruitment of endophilin A to sites of clathrin-mediated SV recycling, allowing vesicle uncoating. Indeed, ITSN1 knockout mice accumulate clathrin-coated vesicles in synapses [82]. ITSN1 may also have a role in synaptic plasticity due to its interaction with the neuron specific isoform of the stable tubule only polypeptide (STOP), a microtubule stabilizing protein [83]. The SH3 domains of ITSN1 bind the Pro-rich motifs of STOP, and both ITSN1 and STOP co-localize in dendrites of rat hippocampal neurons. Knockout of STOP in mice results in a dramatic loss of microtubule cold stability and causes reduction in synaptic vesicle density and defects in both short-term and long-term plasticity [84]. These results open the possibility for ITSN1 to play a role in synaptic plasticity although further investigation is required.

ITSN1 is associated with several neurodegenerative diseases. Human ITSN1 is localized to chromosome 21q22.2, a region closely linked with the Down Syndrome phenotype. Indeed, ITSN1 is overexpressed in patients with Down Syndrome [85]. In addition, ITSN1 is one the most highly induced genes in Alzheimer Disease brains [86–88]. A common characteristic of both diseases is enlargement of the early endosomal compartment [89]. This phenotype resembles the effects cause by deregulated ITSN1 overexpression suggesting that both diseases share a common problem in endocytic trafficking. ITSN1 has also been associated with Huntington disease, where elevated ITSN1-S levels result in increased aggregation of mutant hungtintin (htt) protein leading to enhanced neuron dysfunction [90].

3.6 Herpes virus biology

Kaposi’s sarcoma-associated herpes virus (KSHV), named human herpes virus 8, is responsible for Kaposi’s sarcoma (KS) and several human lymphomas. KSHV mainly infects endothelial cells and B cells. In KSHV genome, K15 is an essential gene for KSHV pathogenesis. The K15 protein consists of a 12 transmembrane sequence and intracellular N-terminal and C-terminal tails. The C-terminal region of K15 comprises short peptide motifs, including SH3 domain-binding sites, such as the conserved Pro-rich sequence PPLP. Lim et al. reported that PPLP selectively binds proteins involved in the regulation of endocytosis, including ITSN2, which directly binds these Pro-rich sites of K15 through its SH3 domains, specifically the SH3-C [91]. This interaction was demonstrated in vitro and in cells. Subcellular fractionation and immunofluorescence experiments showed that ITSN2 and K15 co-localized to endosomal compartments, specifically to discrete vesicular structures in the cytoplasm of B cells. This co-localization was dependent on the ITSN2-K15 interaction, suggesting a possible role for K15 in regulating endocytic trafficking in KSHV-infected B cells. Additionally, the B cell receptor (BCR) was found in the same subcellular fraction as K15 suggesting K15 may affect BCR function. Lim and colleagues studied the effect of ITSN2 and K15 in BCR internalization and found that overexpression of ITSN2 in B cells inhibited BCR internalization [91], consistent with the inhibition of transferring uptake by ITSN2 [23] and the inhibition of endocytosis by overexpression of SH3 domains of ITSN [92]. These findings indicate that the interaction between K15 and ITSN2 links the viral protein to regulation of surface receptor expression, such as BCR. Furthermore, K15-ITSN2 interaction connects K15 to endocytic trafficking suggesting that viral proteins may have the ability to influence the machinery in infected cells.

Epstein-Barr virus (EBV) provides a similar example to KSHV K15. EBV, known as human herpes virus 4, is one of the most common viruses in humans. Most infections occur before the age of five and present no symptoms; however, EBV represents the etiological agent of infectious mononucleosis in adolescents. EBV primarily infects cells of the oropharyngeal epithelium and B cells but it has also been associated with diverse lymphomas and nasopharyngeal carcinoma, resulting in its classification as an oncogenic virus. EBV’s life cycle observes different latency states and Latent Membrane Protein 2A (LMP2A) is an important protein to maintain the latent infection. LMP2A is a structural homologue of K15, and its tail also contains Pro-rich SH3 binding motifs that mediate interaction with ITSN1. Furthermore, LMP2A regulates phosphorylation of ITSN1 by tyrosine kinases [93]. Both examples demonstrate the binding of ITSN to key proteins of human herpes virus. However, the pathological importance of these interactions in herpes virus biology requires further investigation.

4. Concluding remarks and future directions

ITSNs are evolutionary conserved scaffold proteins implicated in a wide variety of biochemical processes. First described as endocytic regulators, ITSNs are now known to play important roles in several signaling pathways (e.g. MAPK), as well as in human diseases (e.g. cancer). Regulation of ITSN levels is likely important for its biological effects since both elevated and reduced ITSN expression result in specific biochemical abnormalities. These results are likely due to the fact that as a scaffold, ITSNs coordinate the assembly of multiprotein complexes. Thus, too much or too little of ITSN may result in defective complex formation and thereby, impairment in the respective process [94].

Another important aspect of ITSNs is the selection of binding partners that allow for the varied functions described above. Splicing of ITSNs provides one possible explanation for the diverse functions of this scaffold. For example, the extended C-terminus of ITSN-L allows for activation of Cdc42. Furthermore, splice variants in the various ITSN domains lead to differential interaction with various targets (reviewed in [2, 17]). Given the ability of ITSNs to homo- and heterodimerize [7] and interact with the related EH-containing family Eps15 [37], there likely exist a large number of unique tetrameric complexes in the cell, each with distinct ligand preferences. Post-translational modification of ITSNs (phosphorylation, ubiquitylation, etc.) may also regulate the interaction with various targets adding an additional layer of complexity to the ITSN network. Although ITSN1 and ITSN2 share many common interacting proteins and may in part have overlapping functions, these proteins likely have distinct roles as well. Indeed, loss of individual ITSN proteins results in distinct phenotypes despite the presence of the other protein.

In conclusion, ITSNs are multi-domain scaffold proteins that have critical roles in a variety of biochemical processes, not limited to endocytosis. Future experiments will help determine the mechanisms by which these multi-protein complexes operate, and allow for a more comprehensive understanding of the function of ITSN in each pathway.