Unique Intracellular Trafficking Processes Associated With Neural Cell Adhesion Molecule and Its Intracellular Signaling

Abstract

Homophilic binding of the neural cell adhesion molecule (NCAM) results in intracellular signaling, which also involves heterophilic engagement of coreceptors such as the fibroblast growth factor receptor (FGFR) and receptor protein tyrosine phosphatase-α (RPTPα). NCAM's own cellular dynamic itinerary includes endocytosis and recycling to the plasma membrane. Recent works suggest that NCAM could influence the trafficking of other receptor molecules that it associates with, particularly the FGFR. Furthermore, it was demonstrated that NCAM could undergo proteolytic processing upon activation. A processed fragment of NCAM, together with an N-terminal fragment of focal adhesion kinase (FAK), is translocated into the nucleus. Here, the authors discuss these rather unique (though not without precedence and analogues) receptor trafficking activities that are associated with NCAM and NCAM signaling.

Keywords::

• fibroblast growth factor receptor (FGFR)
• neural cell adhesion molecule (NCAM)
• protein trafficking
• signaling

INTRODUCTION

The neural cell adhesion molecule (NCAM) (or CD56) is the prototypic neural system–enriched cell adhesion molecule of the immunoglobulin (Ig) superfamily. It has three major splice isoforms (NCAM120, NCAM140, and NCAM180), with different but overlapping distributions in the central nervous system and other tissues. All three of these shares the same extracellular domain consisting of five Ig-like domains and two fibronectin type III (FNIII) repeats. NCAM120 terminates with a glycophosphatidyl (GPI) anchor, whereas NCAM140 and NCAM180 NCAM are transmembrane proteins with intracellular domains of different sizes (Maness and Schachner 2007). NCAM carries a unique carbohydrate modification in the form of α-2,8-linked polysialic acid, and the degree of polysialylation of the molecule is developmentally regulated and affects its function (Hildebrandt et al. 2010).

NCAM plays a myriad of roles in neural development, neuritogenesis, neuronal migration, synaptogenesis, and neuronal survival. Homophilic NCAM binding in trans engages multiple intracellular signaling mechanisms, and a complex signaling network is associated with NCAM's effect in neuritogenesis and neuronal survival (Maness and Schachner 2007). NCAM activates cyclic adenosine monophosphate (cAMP)-dependent protein kinase A (PKA) and phosphatidylinositol 3-kinase (PI3K) (Ditlevsen et al. 2003). Its signaling also involves heterophilic interactions with coreceptors such as the fibroblast growth factor receptors 1 and 2 (FGFR1/2) (Cavallaro et al. 2001) and receptor protein phosphatase-α (RPTPα) (Bodrikov et al. 2005, 2008). NCAM-FGFR interaction leads to activation of phospholipase Cγ (PLCγ) and diacylglycerol lipase (DGL), resulting in arachidonic acid (AA) production and elevation of intracellular calcium (Williams et al. 1994). The latter activates Ca^{2 +} -calmodulin–dependent kinase II (CaMKIIα), and enhances NCAM interaction with RPTPα as well as protein kinase C (PKC), both of which are strengthened by cobinding of spectrin. Acting within the context of lipid rafts, clustering of NCAM also induces RPTPα-mediated activation of Fyn, engagement of the focal adhesion kinase (FAK), and the activation of the Ras-Raf-extracellular signal-regulated kinase 1 and 2 (Erk1/2) cascade (Beggs et al. 1997; Kolkova et al. 2000; Bodrikov et al. 2005). NCAM signaling from within and outside lipid rafts thus modulates cytoskeletal dynamics and transcriptional processes that are critical for neurite outgrowth (Maness and Schachner 2007).

The C3 peptide, which has been identified by recombinant screens and binds to the first Ig module of NCAM, elicits neuritogenesis in a FGFR-dependent manner (Rønn et al. 2000; Kiryushko et al. 2003). Interestingly, synthetic NCAM-binding peptides that interact with its FNIII domain induced neurite outgrowth through a pathway that involves heterotrimeric G proteins, and that is independent of FGFR and Fyn (Hansen et al. 2007). The above illustrates the complex diversity in NCAM-associated signaling. A careful dissection of the pathways involved also revealed that activation of PLCγ and CaMKIIα are only necessary for neurite outgrowth induction, but largely dispensable for the neuroprotective effect of NCAM (Ditlevsen et al. 2007). The various structural, functional, signaling, developmental, and pathological aspects of NCAM have been extensively and authoritatively reviewed (Maness and Schachner 2007; Ditlevsen et al. 2008; Schmid and Maness 2008). The reader can also find updated and detailed knowledge of these in an extensive collection of reviews in Advances in Experimental Medicine and Biology (Berezin, 2009).

The cellular trafficking and dynamics of NCAM is not as well defined compared to, for example, the structurally and functionally related neuronal Ig family member L1/neuron-glia cell adhesion molecule (NgCAM) (Yap et al. 2010). Like L1/NgCAM, surface NCAM is endocytosed, and subsequently recycled back to the plasma membrane, with a minor fraction degraded through the late endosome-lysosomal pathway. The endocytosis of NCAM appears to be developmentally regulated, and is dependent upon its monoubiquitination, which serves as an endocytosis signal (Diestel et al. 2007). The subcellular trafficking dynamics and fate of NCAM upon homophilic interaction and signaling activity are not particularly well documented. Recent findings have, however, indicated that NCAM and receptors associated with it undergo unique and interesting membrane trafficking activities. Our purpose here is to highlight and discuss some of these interesting receptor trafficking processes elicited by NCAM and NCAM-associated signaling.

NCAM BINDING INFLUENCES FGFR TRAFFICKING

Cell contact–dependent neurite outgrowth stimulated by cell adhesion molecules appears to require a receptor tyrosine kinase–induced activation of second messenger pathways that causes calcium influx into neurons. It was shown earlier in 1994 by Doherty and colleagues that activation of FGFR underlies neurite outgrowth stimulated by NCAM, L1/NgCAM, and N-cadherin (Williams et al. 1994). The functional association between NCAM and FGFR is not limited to neural cells, and has been documented in cell types as diverse as cancerous pancreatic β cells (Cavallaro et al. 2001) and T lymphocytes (Kos and Chin 2002). All three major isoforms of NCAM could interact with FGFR though their FNIII modules (Kiselyov et al. 2003). Unlike the latter's canonical ligand FGFs, NCAM-FGFR interaction is absolutely dependent upon an acid box motif at the linker region between FGFR's Ig domains D1 and D2 (Sanchez-Heras et al. 2006). Although NCAM overexpression appeared to antagonize FGF signaling (Francavilla et al. 2007), certain NCAM-derived peptides (known as encamins) could also bind FGFR in an agonistic manner to induce neurite outgrowth (Hansen et al. 2008).

The FGF-FGFR signaling pathway has been extensively studied. Upon FGF binding, FGFR undergoes dimerization and auto tyrosine phosphorylation. Phosphotyrosines serve as docking sites for multiple adaptor and effector proteins such as Shc, Src, phospholipase Cγ, STAT1, Gab1, and FRS2α. These activate several downstream signal pathways, including the Ras-Raf-Erk1/2 pathway (usually resulting in cell proliferation) and PI3K-Akt pathway (usually associated with cell survival) (Eswarakumar et al. 2005). Like other receptor tyrosine kinases, ligand binding induces FGFR internalization and Cbl-mediated ubiquitination (Wong et al. 2002), followed by transport towards the late endosome-lysosomal pathway for degradation (Dikic and Giordano 2003). However, as reported recently by Cavallaro and colleagues (Francavilla et al. 2009), binding of soluble NCAM to FGFR appears to affect its subsequent trafficking differently from FGF, leading instead to an increase in receptor recycling to the cell surface.

The authors started out with a comparative biochemical analysis of FGFR signaling in NIH3T3 and HeLa cells, by FGF-2, a soluble form of NCAM ectodomain fused to Fc and the NCAM-derived FGL peptide. NCAM-Fc and FGL could recapitulate FGFR-dependent function of membrane-bound NCAM, but clearly stimulates different FGFR downstream pathways compared to FGF-2. In HeLa cells in particular, NCAM-Fc induced phosphorylation of Src (whereas FGF does not), and elicited a much longer sustained phosphorylation of the adaptor FRS-2α and the downstream kinase Akt. Interestingly, these ligands also activated the Erk1/2 pathways differently. FGF activation of Erk1/2 occurs through Ras, whereas that by NCAM-Fc acts through Src.

To understand the basis of prolonged downstream effector activation by NCAM over that by FGF, the authors performed analysis of FGFR dynamics with ectopically expressed FGFR. The rate of FGFR endocytosis upon FGF and NCAM treatments were not significantly different from each other, as determine by immunofluorescence (IF) imaging and biochemical assays. However, whereas FGFR internalized upon FGF treatment were largely and quickly degraded, those endocytosed upon NCAM treatment remain stable for prolonged periods. A possible explanation for this arises with the finding that unlike FGF, both NCAM and NCAM-derived peptide did not induce significant FGFR ubiquitination, or an association between Cbl and FRS-2α.

What happened to the internalized FGFR induced by NCAM? Both IF and biochemical analyses indicated that FGFR in cells stimulated by NCAM is channeled to a recycling pathway back to the cells surface. After 60 min of treatment, FGFR in FGF-treated cells was found in lysosomes, whereas those by NCAM were in the recycling endosome compartment and exhibited extensive colocalization with Rab11, the small GTPase that regulates cell surface recycling. Src activation by NCAM underlies the difference in the trafficking activities observed, as FGFR recycling is abolished by Src inhibitors. In the presence of Src inhibitors, Cbl formed a complex with FRS-2α even with NCAM stimulation. The authors also noted that unlike FGF, which primarily enhanced proliferation, NCAM instead stimulated cell migration. This activity of NCAM is dependent on its ability to promote FGFR recycling, and is attenuated by Rab11-silencing or expression of a Rab11 dominant-negative mutant (as well as other methods that inhibit endosomal recycling). In fact, if cell surface recycling of FGFR is promoted by expressing a dominant negative version of Cbl, FGF itself could stimulate migration as well as Src phosphorylation.

The findings of Cavallaro and colleagues add to the emerging mechanism of signaling regulation of receptors via modulation of endosomal trafficking itinerary and receptor fate. One could draw immediate comparisons with recent findings pertaining to another receptor tyrosine kinase, namely the epidermal growth factor receptor (EGFR). Upon binding to EGF, EGFR is also quickly ubiquitinated by c-Cbl, internalized, and targeted for degradation in the lysosome. However, depending on ligand concentration and cellular context, there may be a certain degree of surface recycling. Other than EGF, several different molecules could also bind EGFR and elicit varying degree of mitogenic responses. A recent report compared the effect of six different EGFR ligands on receptor trafficking resulting from ligand binding (Roepstorff et al. 2009). Interestingly, although all ligands promoted internalization, different ligands appear to induce greatly contrasting itineraries upon internalization. EGF, as reported previously, induced lysosomal degradation of the majority (but not all) EGFRs. Two other ligands, the heparin-binding EGF-like growth factor and betacellulin, caused a more complete EGFR lysosomal degradation. In contrast, however, transforming growth factor α (TGFα) and epiregulin binding resulted in complete receptor recycling. Amphiregulin likewise promoted EGFR recycling with different kinetics. Furthermore, in line with that observed for FGFR, another report has shown that regulated Src activation in Madin-Darby Canine Kidney (MDCK) cells could trigger EGFR internalization. However, the internalized EGFR is redistributed to recycling endosomes and not targeted for lysosomal degradation (de Diesbach et al. 2010).

The surface recycling of both FGFR and EGFR has another additional layer of analogous connection, in the context of cancer. NCAM has been shown to promote tumor cell–matrix adhesion through the formation of an FGFR signaling complex (Cavallaro et al. 2001), presumably through enhancement of β1-integrin– mediated cell-matrix adhesion by activating FGFR signaling. EGFR's recycling to the cell surface, particular in conjunction with α5β1 integrin, as coordinated by the Rab coupling protein (RCP), promotes tumor cell migration and invasion in three-dimensional (3D) matrices (Caswell et al. 2008). It was also recently shown by Vousden's laboratory that mutant forms of p53 have a gain-of-function property in promoting invasion and metastasis. These activities of mutant p53 is reflective of enhanced integrin and EGFR cell surface recycling by RCP, which results in constitutive activation of EGFR/integrin signaling (Muller et al. 2009). These recent findings suggest that modulation of receptor tyrosine kinase signaling by enhancement of surface recycling underlies features of cancer pathology that pertains to cell adhesion and migration. Cell adhesion molecules, such as integrin and NCAM, interact and cross-talk extensively with receptor tyrosine kinases in this regard. As noted earlier, a major portion of internalized NCAM is itself recycled back to the cell surface after internalization. However, NCAM's trafficking itinerary upon signaling could take on a more unconventional path, as discussed below.

NCAM SIGNALING ELICITS NUCLEAR TRANSLOCATION OF ITS PROTEOLYTIC FRAGMENT

NCAM-induced neurite outgrowth requires its association with lipid rafts, mediated via its palmitoylation (Little et al. 1998). This raft association facilitates activation of the two nonreceptor tyrosine kinases, Fyn and the focal adhesion kinase (FAK) (Maness and Schachner 2007; Schmid and Maness 2008). Schachner's group has recently reported an intriguing phenomenon associated with NCAM's signaling engagement of FAK, in that both could be proteolytically processed, with their processed products translocated to the nucleus (Kleene et al. 2010). The group has shown earlier that the transmembrane forms of NCAM could be proteolytically processed by the tumor necrosis factor α–converting enzyme (TACE) (or a disintegrin and a metalloprotease 17, ADAM17) to generate a soluble 110-kDa fragment, and this processing is apparently required for NCAM's neuritogenic activity (Kalus et al. 2006).

In the more recent work (Kleene et al. 2010), the authors showed that neuronal NCAM binds to calmodulin through a binding motif at its cytoplasmic domain, and that this interaction is required for the neurite outgrowth promoting activity of NCAM. Loss-of-function mutation of this motif did not alter NCAM's cell surface expression, palmitoylation, or lipid raft localization. However, FAK activation (but not Fyn activation) by NCAM homophilic interaction (or by an interaction-mimicking antibody) was abolished. Interestingly, calmodulin-dependent FAK activation by NCAM stimulation was accompanied by its proteolysis into a phosphorylated N-terminal 50-kDa fragment and a nonphosphorylated C-terminal 55-kDa fragment. That an N-terminal fragment of FAK could be nuclear-localized has been previous documented (Stewart et al. 2002), and the current authors showed that this indeed occur in both Chinese hamster ovarian (CHO) cells and neurons.

Intriguingly, the authors also noted a membrane-spanning 50-kDa NCAM proteolytic fragment generated by calmodulin-dependent NCAM signaling in NCAM-transfected CHO cells and neurons, which could apparently also become nuclear-localized. This proteolytic fragment, unlike the 110-kDa soluble (or extracellular) NCAM fragment generated by TACE/ADAM17, was formed as a result of NCAM cleavage by an extracellular serine protease with plasmin-like activity. As it is recognizable by antibodies against either the intracellular domain or the extracellular domain of NCAM, this fragment likely consists of the intracellular domain, the transmembrane domain, and a truncated extracellular domain. The correspondingly released extracellular domain fragment of 55 kDa could also be detected in the cell culture supernatants by the authors. A prominent point of note here is that the 50-kDa NCAM fragment and the N-terminal fragment of FAK generated by NCAM signaling could be found to colocalize in the neuronal nuclei.

How does the 50-kDa NCAM fragment enter the nucleus? Surface biotinylated NCAM could be found in both the endoplasmic reticulum (ER) and the cytoplasm. The authors showed elegantly, using ER membrane and cytosol derived from stimulated and biotinylated cells coincubated with cytosol and nuclei of untreated cells, that truncated NCAM's likely path to the nucleus is through the ER and cytoplasm. Biotinylated NCAM fragment could be extruded out from the ER membrane into the cytosol, and transported from the cytosol into the nucleus, strictly in a Ca²⁺-calmodulin–dependent manner. The process of ER extrusion does not appear to involve the Sec61-based ER retrotranslocation, as it is not halted by reagents targeting Sec61.

The work by Schachner's group places NCAM amongst a growing number of cell surface or secretory proteins that share the same virtue of some degree of documented nuclear translocation (Planque 2006). Of course, the best documented examples of functional nuclear translocation of proteolytic fragments of surface proteins pertain to γ-secretase–mediated regulated intracellular proteolysis (RIP) of molecules that include Notch and the amyloid precursor protein (APP) (Medina and Dotti 2003; Xia and Wolfe 2003). The generation of the 110-kDa of NCAM by TACE/ADAM17 (Kalus et al. 2006) is a classical case of ectodomain shedding mediated usually by the ADAM family of metalloproteases (Arribas and Borroto 2002; Huovila et al. 2005). On the other hand, the 50-kDa membrane-spanning fragment of NCAM was clearly not derived from either of these processes, as its generation was not inhibited by inhibitors of γ-secretase or metalloprotease (Kleene et al. 2010). It is unclear what the extracellularly released 55-kDa soluble fragment actually does. The proteolytic product of interest here (the 50-kDa fragment) is also not a soluble cytosolic polypeptide, but rather a membrane-spanning protein. How could such a protein be translocated to the nucleus?

Several receptor tyrosine kinases, including EGFR and the homologous ErbB-2, are transported to the nucleus upon endocytosis as polypeptides with an intact membrane-spanning or transmembrane domain (Lin et al. 2001; Wang et al. 2010). Endocytosed EGFR could follow retrograde membrane traffic to the ER (Johannes and Popoff 2008) where it associates with Sec61β, and retrotranslocated from the ER to the cytoplasm (Lo et al. 2006). Importin-β–engaging putative nuclear localization signals (NLS) have been identified for the EGFR (Lin et al. 2001; Lo et al. 2006). The 50-kDa NCAM fragment, however, escapes from the ER membrane into the cytosol in a Sec61-independent manner. Its subsequent nuclear translocation mechanism (and whether it involves of an importin-dependent carrier) is also unclear at the moment.

What is the truncated NCAM fragment doing in the nucleus? As the process of polypeptide extraction from the ER membrane into the cytosol would most probably involve protein denaturation, whether the denatured polypeptide could refold into a functional form remains unclear. Assuming that functional refolding occurs, perhaps aided by nuclear chaperone proteins, the authors speculated that it might interact with transcription factor complexes or components of the chromatin to modulate transcription, perhaps of genes that are downstream of the NCAM signaling pathway. Although mutation of the calmodulin binding motif and blocking calmodulin-dependent molecular interactions abolished measurable NCAM functions and its proteolysis/nuclear transport of processed fragment, whether the two events are linked by causality is not shown. A demonstration that NCAM fragment nuclear translocation is directly responsible for NCAM-mediated functions awaits gain-of-function experiments involving the ectopic expression of the truncated fragment alone. This could be complemented by loss-of-function mutations (such as any putative NLS located at the intracellular domain of NCAM) that could abolish the nuclear translocation of the fragment. Another possibility is that NCAM nuclear translocation is coupled to, or in some way connected with, the nuclear import of FGFR (Maher 1996). Importin-b translocated nuclear FGFR1 induces c-Jun expression, which may be involved in the regulation of cell proliferation (Reilly and Maher 2001). The NCAM fragment may have hitched a ride with FGFR, and may speculatively act in conjunction with FGFR in influencing nuclear transcription. These are exciting prospects that warrant future investigations.

EPILOGUE

In this brief update, we have focused on trafficking processes of an NCAM-associated membrane-bound signaling receptor, as well as NCAM itself. These are of course not the only interesting protein trafficking events associated with NCAM. For example, it has also been recently shown that NCAM promotes dopamine D2 receptor (D2R) internalization and subsequent degradation via direct interaction with a short peptide located within the latter's third intracellular loop (Xiao et al. 2009), and in this context functions as a modulator of the dopaminergic system. In fact, the expression of polysialylated NCAM in the medial prefrontal cortex of rodents is conversely modulated by dopamine acting through D2R (Castillo-Gómez et al. 2008).

The unique subcellular trafficking events associated with NCAM discuss above were highlighted because they illustrate emerging paradigms in NCAM function and activity in relation with other signaling molecules/pathways. The influence of NCAM on FGFR recycling highlights the role of endosomal recycling of a myriad of molecules, mediated by the Rab11 family of small GTPases and their effectors, in regulating cell migration and invasion (Mosesson et al. 2008; Tang and Ng 2009), as well as neurite outgrowth (Eva et al. 2010). On the other hand, demonstration of nuclear translocation of NCAM (and its possible functionality within) may have uncovered yet another direct molecular connection between signaling at the plasma membrane and modulation of gene activity in the nucleus. The signaling mechanisms discussed above are perhaps more widespread than previously appreciated, and we could perhaps expect more examples of such to be revealed in the near future.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.