Faculty of Medical and Health Sciences, Centre for Brain Research, The University of Auckland, Auckland, New Zealand
Department of Anatomy with Radiology, Faculty of Medical and Health Sciences, The University of Auckland, Auckland, New Zealand
Address correspondence and reprint requests to Maurice A. Curtis, Centre for Brain Research, Faculty of Medical and Health Sciences, The University of Auckland, Private Bag 92019, Auckland, New Zealand. E-mail: firstname.lastname@example.org
Cellular interactions mediated by the neural cell adhesion molecule (NCAM) are critical in cell migration, differentiation and plasticity. Switching of the NCAM-interaction mode, from adhesion to signalling, is determined by NCAM carrying a particular post-translational modification, polysialic acid (PSA). Regulation of cell-surface PSA-NCAM is traditionally viewed as a direct consequence of polysialyltransferase activity. Taking advantage of the polysialyltransferase Ca2+-dependent activity, we demonstrate in TE671 cells that downregulation of PSA-NCAM synthesis constitutes a necessary but not sufficient condition to reduce cell-surface PSA-NCAM; instead, PSA-NCAM turnover required internalization of the molecule into the cytosol. PSA-NCAM internalization was specifically triggered by collagen in the extracellular matrix (ECM) and prevented by insulin-like growth factor (IGF1) and insulin. Our results pose a novel role for IGF1 and insulin in controlling cell migration through modulation of PSA-NCAM turnover at the cell surface.
Neural cell adhesion molecules (NCAMs) are critically involved in cell differentiation and migration. Polysialylation (PSA)/desialylation of NCAMs switches their functional interaction mode and, in turn, migration and differentiation. We have found that the desialylation process of PSA–NCAM occurs via endocytosis, induced by collagen-IV and blocked by insulin-like growth factor (IGF1) and insulin, suggesting a novel association between PSA–NCAM, IGF1/insulin and brain/tumour plasticity.
Cell-to-cell and cell-to-extracellular matrix (ECM) adhesions are essential for the development and function of multicellular organisms. The neural cell adhesion molecule (NCAM), abundantly expressed in the nervous system and in tumour cells, modulates events such as migration, differentiation and synaptic plasticity (Bonfanti 2006) by operating as a signalling receptor-like structure for homophilic and heterophilic inputs (Hinsby et al. 2004). NCAM is subject to an unusual post-translational modification, polysialic acid (PSA). PSA plays a modulatory role in NCAM-mediated adhesions by promoting NCAM binding with ECM-components (Hinsby et al. 2004). In contrast, non-polysialylated-NCAMs display predilection for homophilic -cis-trans NCAM-NCAM adhesions (Soroka et al. 2003). Accordingly, PSA acts as a switch between cell-to-cell and cell-to-ECM interactions and in doing so adjusts NCAM function from adhesion to signalling (Gascon et al. 2007; Senkov et al. 2012).
Despite understanding the importance of PSA–NCAM, regulation of cell surface PSA–NCAM remains a poorly understood process with modulation of the expression and/or activity of the two Golgi-associated polysialyltransferases, ST8siaIV (PST) and ST8siaII (STX) (Eckhardt et al. 1995) considered the main regulatory mechanism. The developmental regulation of PST and STX is correlated with that of PSA–NCAM at the cell surface (Bonfanti 2006). The activity of PST/STX is strongly dependent on intracellular stores of Ca2+ and, thus, regulated by perturbations in the concentrations of this ion (Bruses and Rutishauser 1998). Consequently, developmental up-regulation of NMDAR, and subsequent activation of Ca2+ signalling pathways, is associated with a reduction in PSA levels in adult brainstem synapses (Bouzioukh et al. 2001a). Based on the premise that PSA at the cell surface is the resulting net balance between PSA–NCAM synthesis and turnover we demonstrate that PST/STX enzymatic activity per se does not modulate cell-surface PSA–NCAM. Instead, our results demonstrate that PSA–NCAM synthesis and turnover are controlled by different mechanisms which, being independently regulated, allow for fine-tuning of cell-surface PSA–NCAM.
The ECM constitutes a 3D physical-scaffold which plays critical roles in tissue morphogenesis and homeostasis (Frantz et al. 2010). Constant cell–ECM interactions account for the heterogeneous and highly dynamic remodelling of the extracellular milieu and for modulation of key cellular events. The dynamic role of the ECM on the biology of the cells is particularly illustrated by the interaction of integrin-family members with specific components of the extracellular milieu. This in turn activates endocytic/exocytic receptor trafficking and intracellular signalling pathways aimed at remodelling the extracellular microenvironment and induce cytoskeleton rearrangements required for cell migration (Caswell et al. 2009; Huttenlocher and Horwitz 2011). Cell–ECM interactions are influenced by a number of mechanisms, including growth factor signalling (Rozario and DeSimone 2010), and implicated, among other cellular events, in differentiation and migration in neurons, invasion and metastasis in tumour (Guvakova 2007; Sachdev and Yee 2007).
In this study, turnover of PSA–NCAM was investigated in human rhabdomyosarcoma TE671 cells (Daniel et al. 2001). We provide evidence that PSA–NCAM turnover is principally achieved by endocytosis, induced upon interactions with ECM-components such as collagen type IV and modulated by insulin-like growth factor 1 (IGF1) and insulin. Induction of PSA–NCAM turnover promoted accumulation of non-polysialylated-NCAMs and, thereby, a switch in the NCAM-mediated adhesion mode. The results of this study reveal a novel role for IGF1 and insulin in modulation of cell-surface PSA–NCAM which has functional implications for PSA in cell migration (Rutishauser 2008; Brennaman and Maness 2010) and tumour cell metastasis (Amoureux et al. 2010).
TE671 cells (CRL-10636) were obtained from ATCC-Cryosite and routinely cultured with Dulbecco's Modified Eagle Medium: Nutrient Mixture F-12 (DMEM/F12) containing 10% foetal bovine serum (FBS), 100 units/mL of penicillin, 0.1 mg/mL of streptomycin and 2 mM glutamine (PSG) and 1 mM sodium pyruvate (NaPyr) (Invitrogen, Auckland, NZ, USA). For experimentation, 3.75 × 104 per cm2 cells were grown in 24-well plates overnight, washed 2x with DMEM/F12 medium and overlaid with 200 μL of diluted ECM, left to gelify for 1 h (this time referred as experimental t = 0 h) and cultured for the specified times in 1 mL of DMEM/F12 containing 1 mM NaPyr and glutamax (this medium referred in text as normal medium) and the specified supplementations as follows: 1 μM CI-A23187 (calcium ionophore; in EtOh); 500 mM sucrose; 2 μM PAO [phenylarsine oxide; in dimethyl sulfoxide (DMSO)]; 2 μM NNA (l-Nitro-N-Arginine; in DMSO; Sigma, Auckland, NZ, USA); 1.5 ng/mL basic fibroblast growth factor (bFGF); 2.3 ng/mL transforming growth factor beta (TGF-beta); 12 pg/mL PDGF; 1.5 ng/mL EGF (human; Peprotech, Rehovot, Israel); 15 ng/mL IGF1 (human; R&D, Minneapolis, MN, USA); 100 mM insulin (bovine; Sigma); Protease inhibitor (Roche, Auckland, NZ, USA); 2 mM EDTA (Sigma). For EndoN (AbCys, Les Ulis, France) experiments, cells were treated with 1.5 units of the enzyme in normal medium for 3 h [this time referred as t = 0 h except for the study of NCAM-mediated interaction modes (Fig. 4), where t = 0 h refers to non-EndoN-treated cells], washed three times with normal medium and cultured as specified. ECM preparation: MAT and MATGFR (BD Biosciences, Franklin Lakes, NJ, USA) were used at 1 : 4 dilution (approx 2 mg/mL), collagen type I (rat tail; Invitrogen) was used at 0.750 mg/mL and collagen type IV (human placenta; Sigma) at 0.2 mg/mL, in normal medium containing the specified medium-supplementation. Cell viability was assessed with Alamar blue (Invitrogen) according to manufacturer's instructions (Figure S1).
For protein extraction, cells were collected, pelleted by centrifugation and lysed in 30 μL of lysis buffer using a bullet blender (NextAdvance, Averill Park, NY, USA) according to manufacturer's instructions. Electrophoresis and immunoblot were performed as described (Monzo et al. 2012) with mouse IgM α-PSA–NCAM 1 : 3000 (Millipore, Darmstadt, Germany), mouse-IgG α-GAPDH 1 : 1500 (Abcam, Cambridge, UK), rabbit IgG α-NCAM 1 : 2000 (Santa Cruz, Dallas, TX, USA) and goat α-mouse-IgM-HRP (horseradish peroxidase) 1 : 3000 (Serotec, Oxford, UK), sheep α-mouse-IgG-HRP 1 : 1500 and donkey α-rabbit 1 : 2000-HRP (GE Healthcare, Buckinghamshire, UK). Correlation between PSA–NCAM amount and GADPH loading control, as well as specificity and correct protein size, are presented with uncropped images for each antibody used in this study in Figure S1.
Developed films were scanned with Hp Scanjet-5470c with greyscale settings and analysed with Image J 1.46 (National Institutes of Health, Bethesda, MD, USA) by obtaining the integrated densities for the PSA–NCAM and corresponding GAPDH bands. PSA–NCAM integrated densities were normalized with the corresponding GAPDH. Results are presented as mean% average relative to control set as value = 100 ± SD (n = 3 independent experiments; each independent experiment was performed with the corresponding controls, and image analysis was performed per individual gels). Figures presented in this study correspond to individual-gel images adjusted for brightness/contrast (adjustments affected in all cases the entire image including controls). Sample order in figures was rearranged (indicated by a vertical-dotted bar) to maintain data-description coherence.
Cells intended for immunocytochemistry were cultured in 96-well plates. Primary antibodies: PSA–NCAM 1 : 3000; Goat α-mouse-IgM Alexa 488 and Hoechst (1 : 500; Invitrogen). Images were acquired with equal exposure times and settings between samples and processed with Leica Application Suite V3.0 (Wetzlar, Germany). Staining controls with omission of either primary or secondary antibodies were applied and revealed absence of signal. For confocal microscopy cells were cultured in 24-well plates on glass coverslips and immunostained as described (Dieriks et al. 2011) with rabbit anti-Rab5 (1 : 500; Abcam), PSA–NCAM (1 : 600), goat α-mouse-IgM-Alexa 488 and goat α-rabbit Alexa 594 (1 : 250). Confocal recordings were done with FV1000 confocal microscope (Olympus, Tokyo, Japan) and 40x oil immersion lens (NA 1.00).
Cells intended for RT-qPCR were cultured in six-well plates and sampled as indicated above. RNA extraction (bullet blender), reverse transcription and qPCR were performed as described (Monzo et al. 2012). qPCR data were analysed with gene-specific standard curves and mRNA amounts subsequently normalized to GAPDH and calibrated to control/t = 0 h/no-ECM (as specified in figures). Results are presented as mean fold change at t = 12 h ± SEM in target-gene mRNA relative to control/t = 0h/no-ECM (black bar; set as value = 1) except in Figure S6, where values represent mRNA at t = 0 h relative to no-ECM control (n =3 independent experiments). Primer used (5′-3′; efficiency%; NCBI-reference): CATGAGAAGTATGACAACAGCCT/GTCCTTCCACGATACCAAAGT for GAPDH (87; NM_002046.4); TCA GTGGTGTGGAATGATGATTC/CCGGCGTCGTCGATGT for NCAM (89; NM_000615.6); GAGCAGTTTTAAGCCTGGTG ATG/TAGGAGGCTATGTAGATCATGAGAAATG for PST (84; NM_005668); GCTGACCAACAAAGTCCACATC/GAAACGTG TGGCCAGGGTAT for STX (108; NM_006011); GGAGATGATGTCAACATTGAAAAGG/GGTGAGGCGGAACTTGTTTCT for n-NOS (106; NM_000620); GGTGGAAGCGGTAACAAAGG/TGCTTGGTGGCGAAGATGA for i-NOS (n. d; NM_000625.4); AGATCGGCACGAGGAACCT/TCCATGCAGACAGCCACATC for e-NOS (108; NM_000603.4); ACAGACGGCTCGCTCCCCTC/TAGGCAAAGCTGCCCGCCAC for IGF1R (107; NM_000875.3), CTGCCGCTATGAAATTGAGTGG/CGCCGCTCAGAGAACAAGTT for IGF2R (115; NM_000876.2); CATCCGGGGATCACGACTG/AAGATGACCAGCGCGTAGTTA for INSR (102%; NM_ 000208.2).
Statistical significance was assessed using two-sample t-test with unequal variance, and significant differences were attributed to p ≤0.05 (specific p-values are indicated in the text).
All procedures conformed to the University of Auckland Code of Conduct for Research and Biosafety Regulations.
Pharmacological perturbation of intracellular calcium pools inhibits NCAM polysialylation
Perturbation of the intracellular Ca2+ was reported to prevent NCAM polysialylation (Bruses and Rutishauser 1998). Following a working hypothesis for the amount of PSA–NCAM at the cell surface as the net result between synthesis and turnover, we used a calcium ionophore (CI) to investigate the mechanisms governing PSA–NCAM turnover. Endoneuraminidase N (EndoN; a glycoside hydrolase enzyme highly specific for cleaving α-2-8-linked PSA from NCAM)-treated TE671 cells were cultured with media supplemented with CI or EtOh. CI-cultures were unable to resume NCAM polysialylation, in contrast to EtOh controls, demonstrating that CI inhibited PSA–NCAM synthesis in these cells (Fig. 1). No differences in PSA–NCAM were observed after 6 h of culture in normal medium (control) in comparison to t = 0 h.
Turnover of PSA–NCAM is induced by the ECM
As per our working hypothesis, we assumed that in non-differentiating cells, such as TE671, PSA–NCAM synthesis and turnover mechanisms occur at similar rates resulting in a constant amount of cell surface PSA–NCAM despite its continuous recycling. Accordingly, inhibition of PSA–NCAM synthesis would allow us a window to solely study the turnover of the molecule from the cell surface and, conversely, inhibition of PSA–NCAM turnover would allow for the study of its synthesis/transport to the cell surface. The working hypothesis was tested with time-course cultured TE671 cells in normal- and CI-medium for 12 h. Cells did not show changes in PSA–NCAM in normal medium; however, changes in PSA–NCAM were also not observed when using CI-medium (Fig. 2). Cells were cultured in the absence of ECM, which is in contrast with the in vivo scenario where cells are three-dimensionally (3D) embedded in the ECM, which provides fundamental physicochemical stimuli. The role of the ECM in cell development together with that of NCAM and its polysialylated form in modulating interactions between cells and the extracellular milieu (Bonfanti 2006) prompted us to investigate the influence of the ECM in PSA–NCAM turnover. As ECM, we used the less biochemically complex ‘growth factor reduced’ version (MATGFR) of a basement membrane matrix preparation (Matrigel; MAT). Cells time-cultured in MATGFR with normal medium showed no differences in PSA–NCAM; however, a remarkably progressive-reduction in cell-surface PSA–NCAM was observed in MATGFR cultures with CI-medium (43.1 ± 3.9% amount of PSA–NCAM at t = 12 h, relative to t = 0 h; 100 ± 14.4%; ♣p = 0.002; Fig. 2 and Figure S2). Immunocytochemistry corroborated these results by showing a reduction in fluorescence signal for PSA–NCAM in cells after 12 h of culture in MATGFR with CI-medium (Fig. 2d). EndoN-treated cells cultured with CI-medium were unable to resume NCAM polysialylation during the entire length of the culture time in both ECM conditions (Fig. 2e and f). Thus, despite the efficient inhibition of NCAM polysialylation in no-ECM and MATGFR, PSA–NCAM turnover was only achieved in the presence of ECM.
PSA–NCAM turnover is mediated by endocytosis
To address the PSA–NCAM turnover mechanism, we first cultured protein extracts from TE671 cells in MATGFR with normal and CI/EtOh-supplemented media for 12 h and analysed for PSA–NCAM abundance. The results revealed no difference in PSA–NCAM between t = 12 h and t = 0 h in any cell culture condition (Figure S3a) and, disregarding components of the ECM acting as enzymatic catalysts, pointed towards a cellular-based processing of the molecule. Cell surface proteins, including NCAMs, are targets of matrix metalloproteases (MMP)-mediated pericellular proteolysis as part of an orchestrated cell signalling mechanism (Werb 1997) and MMP-mediated NCAM-ectodomain shedding was suggested as the likely mechanism for NCAM desialylation at the cell surface (Hinkle et al. 2006). In our studies we could not detect PSA–NCAM in the supernatant and MATGFR samples of cultures with CI-medium for 12 h (Figure S3b). Additional tests were performed in MATGFR cell cultures with CI-medium supplemented with protease inhibitors (PI; Figure S3c) and EDTA (Figure S3d) showing no differences in PSA–NCAM between CI-cultures and CI-cultures supplemented with PI/EDTA. Taken together, our data strongly suggested that PSA–NCAM desialylation in TE671 cells was mainly an intracellular process. We then tested sucrose which, at hypertonic concentrations, interferes with the endocytic pathway (Hansen et al. 1993). MATGFR cultures with CI-medium supplemented with sucrose blocked PSA–NCAM turnover (91.4 ± 15.1% vs. 42.4 ± 14.5%, respective amount of PSA–NCAM for CI-sucrose and CI-cultures at t = 12 h relative to t = 0 h; 100 ± 14.1%; *p = 0.015; Fig. 3a). Subsequently, we tested phenylarsine oxide [PAO; inhibitor of receptor internalization; (Sorensen et al. 1998)]. PAO in CI-MATGFR cultures completely prevented PSA–NCAM turnover (119 ± 27.6% vs. 46.5 ± 5.9% and 44.9 ± 21.6%, respective amount of PSA–NCAM for CI-PAO, CI and CI-DMSO cultures at t = 12 h relative to t = 0 h; 100 ± 11.5%; *p = 0.001; Fig. 3b and Figure S3). Immunocytochemistry corroborated the above data by showing a strong fluorescence signal for PSA–NCAM in TE671 cells after 12 h of culture in MATGFR with CI-PAO-medium in comparison to that for CI-medium (Fig. 3c). In agreement with previous studies (Minana et al. 2001; Diestel et al. 2007), small numbers of PSA–NCAM molecules were recurrently found co-localizing with Rab5 (early endocytic marker) as soon as 1 h following culture in MATGFR with CI and during subsequent culture time-points (Fig. 3d). The small amount of positive co-localization events between PSA–NCAM and Rab5 per time-point analysed was in agreement with PSA–NCAM turnover following a slow but progressive endocytosis process (Fig. 2c). Our results were in agreement with previous studies in the rodent vagal complex, cultured astrocytes and cortical neurons (Bouzioukh et al. 2001a; Minana et al. 2001; Diestel et al. 2007) and implicated the endocytic pathway in the turnover process of PSA–NCAM. MATGFR cultures in PAO-medium (to prevent PSA–NCAM internalization) displayed a significantly increased (~ fourfold) amount of PSA–NCAM (436.0 ± 144.6%) after 12 h relative to control at t = 0 h (100 ± 4.6%; *p =0.01; Fig. 3fi and fii), thus corroborating the working hypothesis. PSA–NCAM turnover in MATGFR cultures with CI was also prevented when cell culture medium was additionally supplemented with NG-nitro-l-Arginine (non-specific NO-synthase inhibitor; (Griffith and Kilbourn 1996); Figure S3fi and fii), and gene expression pattern analysis at the mRNA level showed no changes for the neural NO-synthase isozyme (n-NOS) between the culture conditions used, which differentially induced PSA–NCAM turnover (Fig. 2) or during the culture time (Figures S4 and S6). NO induces endocytosis via activation of guanylyl cyclase and subsequent production of PIP2 (Liu et al. 2009), and its activity is associated with synaptic plasticity in cultured rat hippocampal neurons (Micheva et al. 2003). Taken together, our data demonstrated a regulatory role for NO in the MATGFR-induced PSA–NCAM internalization process in our cell system similar to that occurring in vivo.
Turnover of PSA–NCAM modulates the NCAM-mediated interaction mode
NCAM molecules can interact homophillically, on opposing (-trans) and same (-cis) surfaces, as well as engage in heterophilic ligations (Gascon et al. 2007). The functional in vivo role for PSA-switch of the NCAM-interaction modes (Senkov et al. 2012) prompted us to assess the effects of PSA–NCAM turnover in our cellular model. First, a scenario devoid of PSA was generated by culturing TE671 cells with EndoN. No-ECM cultures showed initially (t = 0 h) two NCAM bands corresponding to molecular weights (MW) of approximately 220- and 140-KDa; however, subsequent EndoN-treatment resulted in a progressively reduced MW shift for the upper 220-KDa NCAM band to approximately 140-KDa (Fig. 4a). TE671 cells most commonly express the 140-KDa isoform of NCAM (Daniel et al. 2001); thus, a shift in its MW reflected the NCAM post-translational configuration – polysialylated (220-KDa) or desialylated (140-KDa) (Daniel et al. 2001). MATGFR cultures also displayed initially (t = 0 h) two NCAM bands of approximately 140- and 220-KDa; however, culture in EndoN resulted in no progressive MW shift for the 220-KDa NCAM band (Fig. 4b). EndoN completely removed PSA from NCAM by t = 2 h in both no-ECM and MATGFR cultures and promoted NCAMs to be non-polysialylated for the entire experimental time (Fig. 4a and b). Thus, as detailed above, the presence of a large 220-KDa NCAM band in a PSA-devoid scenario was indicative of native 140-KDa NCAMs interacting (i.e. with a post-translational modification) with ligands other than PSA. We then addressed the NCAM-mediated interaction mode upon endogenous (i.e. non-EndoN mediated) turnover of PSA–NCAM. No-ECM in normal- or CI-medium showed no major changes over time for the 220-KDa NCAM band (Fig. 4ci and cii); this in agreement with PSA–NCAM turnover not being induced in these culture conditions (shown in Fig. 2). The time-course for MATGFR in normal- and CI-medium (Fig. 4d) displayed a similar pattern for NCAM as that observed for no-ECM (Fig 4c) with an intense 220-KDa NCAM band during the entire experimental time (Fig. 4di and dii). However, as for EndoN-treated cultures (Fig. 4b), the lack of progressive shift for the 220-KDa NCAM band in MATGFR culture conditions with CI correlated with a remarkable reduction in PSA–NCAM at the cell surface (shown in Fig. 2). Thus, the presence of a large 220-KDa NCAM band in MATGFR with CI-medium was indicative of non-polysialylated-NCAMs interacting with ligands other than PSA (Fig. 4dii) and, therefore, of a switch in the NCAM-mediated interaction mode. These data suggested a functional correlation between the PSA–NCAM turnover in TE671 cells and that reported in vivo (Senkov et al. 2012). Culture conditions with CI-medium also showed a progressive increase in the intensity for the 140-KDa NCAM as a result of the production and subsequent accumulation of non-polysialylated-NCAMs. Interestingly, this was matched with a moderate but nevertheless significant down-regulated expression for NCAM gene (Figures S4 and S6) and suggested the activation of a negative-feedback mechanism to prevent excessive accumulation of non-polysialylated-NCAMs.
The data obtained clearly indicated that PSA–NCAM turnover were modulated by components of the ECM. To further address the role of ECM-components in PSA–NCAM turnover on TE671 cells we used MAT (complete version of Matrigel). First, cells treated with EndoN and cultured in MAT with CI-medium could not engage in NCAM-polysialylation during the 12 h experimental time (Fig. 5a). We then assessed MAT for a role in promoting PSA–NCAM internalization. MAT cultures in normal medium (control) displayed a larger amount of PSA–NCAM (100 ± 9.6%) than that for MATGFR cultures in normal medium (71.3 ± 8.3%, amount of PSA–NCAM relative to MAT-control after 12 h of culture; Fig. 5b); however, whereas MATGFR cultures with CI-medium showed a ≈ 59% reduction (40.9 ± 7.8% amount of PSA–NCAM in MATGFR cultures with CI relative to MATGFR in normal medium), the reduction of PSA–NCAM for MAT cultures with CI relative to MAT cultures in normal medium reached only ≈ 34% (66.2 ± 6.2% amount of PSA–NCAM in MAT cultures with CI relative to MAT-control vs. 100 ± 9.6%; Fig. 5b). Cells in MATGFR with CI reduced the amount of PSA–NCAM to 29.5 ± 8.4% relative to that of MAT cultures in normal medium (Fig. 5b). The significantly different (**p = 0.01) capacity for promoting PSA–NCAM internalization between MAT and MATGFR was indicative that the biochemical composition of the ECM was playing a role in modulation of PSA–NCAM turnover in TE671 cells. The expression patterns for NCAM, PST, STX, n-NOS, i-NOS and e-NOS genes were not significantly different between MAT and no-ECM or MATGFR cultures (Figures S4 and S6), further supporting the notion that PSA–NCAM turnover was principally directed by ECM-components.
The data obtained with MAT and MATGFR indicated that whereas common components in both ECMs were inducing PSA–NCAM internalization, missing components in the latter, such as GFs, could be acting as inhibitors of the process. MATGFR cultures with CI-medium were additionally supplemented with individual GFs (at their original concentration in MAT) and tested for PSA–NCAM turnover after 12 h. MATGFR cultures in CI-medium showed a reduced amount of PSA–NCAM which was not significantly different from that in CI-medium supplemented with bFGF, TGF-β, PDGF and EGF (41.1 ± 4.2 vs. 39.6 ± 1.1%, 48.2 ± 2.5%, 57.8 ± 15.1% and 48.8 ± 11.9%, respective amount of PSA–NCAM relative to normal medium control, 100 ± 15.5%). In contrast, MATGFR cultures with CI-medium additionally supplemented with IGF1 and its related hormone insulin displayed a significantly larger amount of PSA–NCAM than that in unsupplemented CI-medium (78.1 ± 6.4% and 81.6 ± 6.9% vs. 41.1 ± 4.2%, respective amounts of PSA–NCAM relative to normal medium control; *p = 0.003; **p = 0.00003; Fig. 5c). The expression pattern for IGF1R and INSR (insulin receptor) genes was not significantly different between the cell culture conditions used (Figures S5 and S6).These results revealed a prominent inhibitory role for IGF1/insulin in the PSA–NCAM turnover of TE671 cells.
MAT and MATGFR share a basic composition of laminin/entactin and collagen type IV. Therefore, a common component between MAT and MATGFR could be inducing PSA–NCAM turnover in TE671 cells. Accordingly, we tested collagen, one of the most abundant protein components of the ECM. Because we were unable to generate collagen type IV 3D-gels, we first analysed collagen type I (not reported as component in MAT or MATGFR) 3D-gels. Cultures in collagen I showed no significant differences in PSA–NCAM after 12 h in any culture media used (108.5 ± 18.1% vs. 92.3 ± 20.1% and 101.7 ± 8.2%, respective PSA–NCAM amount in CI, CI + IGF1 and CI + insulin media, relative to normal medium control, 100 ± 16.7%; Fig. 5d). In contrast, cultures in collagen I + IV with CI-medium showed a significant induction of PSA–NCAM turnover after 12 h, which was prevented when CI-medium was supplemented with IGF1/insulin (59.1 ± 7.3% vs. 87.0 ± 9.5% and 100.1 ± 7.5%, respective PSA–NCAM amount in CI, CI + IGF1 and CI + insulin media, relative to normal medium control, 100 ± 3.8%; *p =0.007; Fig. 5e). This data revealed the involvement of collagen type IV as an inducer of PSA–NCAM turnover in TE671 cells.
The balance between PSA–NCAM and NCAM at the cell surface is critical for cell migration and synaptic plasticity (Rutishauser 2008). Cell surface PSA–NCAM was proposed to be a direct function of its synthesis rate (Soares et al. 2000; Angata and Fukuda 2003); however, in TE671 cells, inhibition of PSA–NCAM synthesis is a necessary but not sufficient condition to reduce cell-surface PSA–NCAM. Turnover of PSA–NCAM was achieved when its synthesis was inhibited (Bruses and Rutishauser 1998) but only upon activation of PSA–NCAM internalization by the ECM. Thus, cell-surface PSA–NCAM/NCAM ratios in TE671 cells are the resulting net balance between two major mechanisms, synthesis and turnover, independently regulated but acting in tight coordination (Bonfanti 2006).
Cell lysates cultured in ECM did not show changes in PSA–NCAM or was PSA detected in culture media or ECM samples, thus posing PSA–NCAM desialylation as an intracellular mechanism involving internalization of the molecule in TE671 cells. Furthermore, the effective prevention of PSA–NCAM turnover by endocytosis inhibitors and the co-localization of PSA–NCAM with Rab5 corroborated the involvement of the endocytic pathway in PSA–NCAM turnover (Bouzioukh et al. 2001b; Minana et al. 2001). Shed NCAM-ectodomains were previously reported devoid of PSA (Hinkle et al. 2006; Brennaman and Maness 2008) and our data concurs that MMP-mediated proteolysis of NCAM occurs after desialylation of PSA–NCAM.
Recycling of adhesion receptors such as integrins via the endosomal pathway promotes the dynamics required for cell migration (Bretscher 2008; Huttenlocher and Horwitz 2011). In concurrence with previous studies (Bouzioukh et al. 2001a; Minana et al. 2001; Diestel et al. 2007), our data pose modulation of PSA–NCAM/NCAM at the cell surface as a recycling process, whereby NCAMs are polysialylated at the Golgi and transported to the cell surface. Subsequently, cell surface PSA–NCAMs are internalized into acidic endosomes where PSA is degraded, because of the labile glycosidic linkages of sialic acids in low pH environments (Manzi et al. 1994), and NCAMs sorted for re-polysialylation at the Golgi and transported to the cell surface. In this scenario, inhibition of NCAM-polysialylation results in de novo synthesized and/or sorted non-polysialylated-NCAMs being transported to the cell surface, where they are now susceptible to MMP-mediated ectodomain shedding (Hinkle et al. 2006), while cell-surface PSA–NCAM is continuously being internalized into the cytosol. The continuous supply of non-polysialylated-NCAMs, in combination with the continuous internalization of PSA–NCAM, leads to increased non-sialylated-NCAM at the cell surface and consequently to a progressive switch in the NCAM-mediated interaction status of the cell.
PSA exerts steric impediment between cells and modulates NCAM-mediated interactions (Rutishauser 2008). Our results showed a PSA-dependent pivotal shift in the MW of NCAM suggesting the establishment of NCAM-interactions with ligands other than PSA and, thereby, a switch in the overall NCAM-interaction mode. In the absence of PSA, NCAM displays stronger affinity for homophilic interactions (Hinsby et al. 2004), thus our results might reflect the formation of large zipper-like NCAM-NCAM complexes in cis/trans (Soroka et al. 2003). Alternatively, a number of partners have been described for heterophilic NCAM-interactions such as L1, FGF and glutamate receptors, HSPG and collagens (Storms and Rutishauser 1998; Gascon et al. 2007; Senkov et al. 2012). This poses the possibility for non-polysialylated-NCAMs in our cellular model interacting with ECM- or cellular components, the latter requiring ECM-components such as HSPGs for modulation of their strength (Nielsen et al. 2010), and forming large structural complexes.
Dysregulated NCAM-mediated adhesions are remarkably associated with non-polysialylated-NCAMs and correlate with gross brain-wiring abnormalities in mouse models (Brennaman and Maness 2008; Hildebrandt et al. 2009). Accordingly, our results showing a significant down-regulation for NCAM gene expression (Figure S4) suggest the activation of a negative-feedback regulatory mechanism for NCAM in conditions inducing the production of non-polysialylated-NCAMs [i.e. polysialylation of NCAM is inhibited: CI-medium or up-regulation of NMDAR in vivo (Bouzioukh et al. 2001b)]. Such mechanisms would avoid excessive accumulation of non-polysialylated-NCAM at the cell surface preventing unregulated NCAM-mediated interactions with potentially severe consequences (Weinhold et al., 2005).
Our results point toward ECM-components acting as a trigger-switch in a ligand-receptor endocytosis-mediated PSA–NCAM turnover. Using ECMs of different composition, we identified a novel function for collagen type IV and IGF1/insulin as respective inducer and inhibitors of PSA–NCAM internalization in TE671 cells. IGFs in the ECM play functional roles in events such as turnover of cell-adhesions (Guvakova 2007). IGF1R forms dynamic complexes with αv-integrin which are disrupted upon IGF1 interaction causing integrin relocalization and a subsequent shift from cell-to-cell to focal contacts that promotes cell migration (Canonici et al. 2008). We did not detect such structural interaction between PSA–NCAM and IGF1R (data not shown); however, the formation of PSA–NCAM–IGF1R complexes may be weak or transient, or mediated through an unknown linker protein in TE671 cells. Further research is required to unravel the functional cross-talk between IGF1R and the turnover of PSA–NCAM; however, we speculate that likely pivotal-candidates for IGF1R-PSA–NCAM cross-talk are integrin-family members. NCAM is known to interact with β1-integrin (Petridis et al. 2011), involved in modulation of integrin-dependent cell migration (Diestel et al. 2005) and, similar to L1, proposed to act as co-endocytosis carrier for integrin-family members and in recycling of focal adhesions and integrin-dependent migration processes (Diestel et al. 2007; Schmid and Maness 2008). Interaction between components of the ECM, such as collagen, and integrin-family members leads to endosomal trafficking of several molecules and to the establishment of polarized/asymmetric adhesions promoting cell migration (Hynes 2002; Caswell et al. 2009; Huttenlocher and Horwitz 2011). The implication of collagen type IV in PSA–NCAM turnover in TE671 cells also suggests the involvement of integrin-family members in the PSA–NCAM internalization process.
Modulation of cellular responses by the ECM is largely established as a direct function of its dynamic topology and biochemical composition (Frantz et al. 2010). Moreover, our results suggest a functional association between the in vivo dynamics of the ECM biochemical composition and the balance of PSA–NCAM/NCAM affecting key cellular events such as development and migration (Rutishauser 2008). Increasing evidence from clinical and pre-clinical studies pose IGF1/insulin as key players in CNS neuroplasticity (Reagan 2010). Our data suggest an interesting functional correlation between IGF1/insulin, PSA–NCAM turnover, synaptic dysfunctions and neurodegenerative diseases associated with dysregulated PSA–NCAM interactions such as Alzheimer's and schizophrenia (Hildebrandt et al. 2009; Brennaman and Maness 2010).
The authors would like to thank the Fisher Family Trust, the Auckland Medical Research Foundation, the Marsden Fund, the Health Research Council of NZ and the Herring grant from the Manchester Trust for funding this study. HJM was funded by the Gus Fisher Postdoctoral Fellowship. Authors thank Natacha Coppieters for technical support and Sheryl Feng, Marika Eszes and Anne-Marie Alborn for research assistance. The authors have no conflict of interest.
HJM: Experimental design and experimental work, data analysis and manuscript writing.
TIHP: Experimental work and manuscript writing.
BVD: Experimental work and manuscript writing.
DJ: Experimental work and manuscript writing.
RLMF: Data analysis and critical analysis of the manuscript.
MD: Data analysis and critical analysis of the manuscript.
MAC: Experimental design, data analysis and manuscript writing.