Research ArticleDEVELOPMENTAL NEUROSCIENCE

Temporal proteomics of NGF-TrkA signaling identifies an inhibitory role for the E3 ligase Cbl-b in neuroblastoma cell differentiation

See allHide authors and affiliations

Science Signaling  28 Apr 2015:
Vol. 8, Issue 374, pp. ra40
DOI: 10.1126/scisignal.2005769

Abstract

SH-SY5Y neuroblastoma cells respond to nerve growth factor (NGF)–mediated activation of the tropomyosin-related kinase A (TrkA) with neurite outgrowth, thereby providing a model to study neuronal differentiation. We performed a time-resolved analysis of NGF-TrkA signaling in neuroblastoma cells using mass spectrometry–based quantitative proteomics. The combination of interactome, phosphoproteome, and proteome data provided temporal insights into the molecular events downstream of NGF binding to TrkA. We showed that upon NGF stimulation, TrkA recruits the E3 ubiquitin ligase Cbl-b, which then becomes phosphorylated and ubiquitylated and decreases in abundance. We also found that recruitment of Cbl-b promotes TrkA ubiquitylation and degradation. Furthermore, the amount of phosphorylation of the kinase ERK and neurite outgrowth increased upon Cbl-b depletion in several neuroblastoma cell lines. Our findings suggest that Cbl-b limits NGF-TrkA signaling to control the length of neurites.

INTRODUCTION

Neuroblastoma arises from cells of the developing sympathetic nervous system and is the most common extracranial tumor in childhood. It is a remarkable heterogeneous disease with a clinical representation ranging from highly malignant tumors to tumors that either spontaneously regress or differentiate into a benign state (1). The survival rate is less than 50% among high-risk patients receiving extensive therapy, whereas it can reach 90% for low-risk patients requiring observation only and no therapy (1). This striking dichotomy in clinical outcomes emphasizes the need for reliable prognostic stratification. Furthermore, driving neuroblastoma tumor cells toward a benign differentiated state has emerged as a very useful clinical strategy. For instance, retinoic acid, which induces morphological differentiation of neuroblastoma cells in culture, is now used as “differentiation therapy” in the standard treatment regimen of high-risk neuroblastoma (2, 3). Thus, a better understanding of the molecular mechanisms underlying neuroblastoma differentiation holds promise for improving the treatment of high-risk neuroblastoma patients.

Dissecting the nerve growth factor (NGF) signaling network in the context of human neuroblastoma is highly relevant for several reasons. First, NGF and its high-affinity receptor tropomyosin-related kinase A (TrkA; encoded by NTRK1) play a major role in the developing nervous system, especially in sympathetic neural development (4). The NGF-TrkA pair is an interesting candidate in neuroblastoma differentiation because one hypothesis for the origin of neuroblastoma focuses on defects in differentiation (5), and because NGF-TrkA signaling is considered important for driving sympathoadrenal progenitor cells toward more committed cell fates upon differentiation into sympathetic neurons (6). Second, TrkA protein abundance is often high in tumors from low-risk patients, and these tumors hold the unique propensity to spontaneously differentiate (7). Finally, reexpression of exogenous TrkA in cell culture model systems of neuroblastoma renders cells sensitive to NGF-induced differentiation (811).

NGF-TrkA signaling has been extensively studied in the rat pheochromocytoma cell line (PC12), a model used to study neuronal signal transduction (1214). Upon NGF binding, the receptor tyrosine kinase (RTK) TrkA undergoes homodimerization, autophosphorylation, and activation. Consequently, downstream effectors including Src homology 2 (SH2) domain–containing transforming protein (Shc), fibroblast growth factor receptor substrate 2 (FRS2), and phospholipase Cγ (PLCγ) are recruited directly to the activated receptor by binding SH2 domains (15). These adaptors ultimately bridge to the major canonical TrkA downstream signaling pathways, including the phosphatidylinositol 3-kinase (PI3K)–AKT and RAS–mitogen-activated protein kinase (MAPK) pathways, leading to cell survival and differentiation (16). In particular, sustained activation of extracellular signal–regulated kinase (ERK) in response to NGF, but not epidermal growth factor (EGF), leads to differentiation in PC12 cells (17). As demonstrated for other RTKs, TrkA ubiquitylation, internalization, and turnover can be promoted by E3 ubiquitin ligases such as NEDD4-2, tumor necrosis factor receptor–associated factor 6 (TRAF6), and Cbl (casitas B-lineage lymphoma) (1820). Various cellular responses can be triggered by the activation of TrkA depending on the specific cellular context; neurons differentiate, whereas breast cancer cells proliferate and medulloblastoma cells undergo apoptosis (21, 22). This highlights the importance of studying TrkA signaling in a cell type–specific manner. Reported studies aimed at deciphering the NGF signaling network that mediates neuronal differentiation in human cells have largely focused on changes in protein abundance (23, 24). However, tight regulation of intracellular signaling networks and cellular outcome decision-making are strictly dependent on the specific timing of well-coordinated molecular events (17, 25, 26). The temporal coordination of the phosphorylation of, for example, RTKs and their adaptor proteins can direct and control the flow of information (2729). Thus, to effectively exploit the directionality of activated RTKs toward a certain cellular response, a more complete understanding of dynamic signaling regulation is needed. With regard to NGF-TrkA signaling, a greater overview of the early signaling events triggered by the ligand-activated receptor and of the link to late signaling events, such as proteome changes, is required. Here, we addressed these critical aspects by exploring NGF-TrkA signaling associated with neurite outgrowth in neuroblastoma cells.

RESULTS

Temporal proteomics of NGF-TrkA signaling reveals dynamic regulation of downstream events

To study early and late signaling events associated with NGF-induced neurite outgrowth in neuroblastoma cells, we developed a tetracycline (Tet)–inducible system for TrkA expression in the human neuroblastoma cell line SH-SY5Y (herein called SH-SY5Y-TR-TrkA). Tet induction and NGF stimulation of these cells resulted in the phosphorylation of TrkA and the downstream signaling proteins AKT and ERK (Fig. 1A and fig. S1, A and B). Furthermore, neurite outgrowth (a morphological characteristic of neuronal differentiation) was detected in response to long-term stimulation with NGF (at 24 and 48 hours) in SH-SY5Y-TR-TrkA cells (Fig. 1B and fig. S1C). Because these findings were consistent with previous studies (11), we used this cell line to model TrkA signaling that induces differentiation in neuroblastoma cells.

Fig. 1 Temporal regulation of NGF-TrkA signaling defined by the interactome, phosphoproteome, and proteome.

(A) Western blotting showing dynamic activation of TrkA, AKT, and ERK in NGF-stimulated SH-SY5Y-TR-TrkA cells. p, phosphorylated. (B) Quantification of NGF-induced neurite outgrowth in SH-SY5Y and SH-SY5Y-TR-TrkA cells stimulated for 24 and 48 hours. Data are means ± SD of three experiments. **P < 0.01 in comparison to unstimulated control cells by two-sample Student’s t test with Bonferroni correction. (C) Graphical overview of the time scale for proteomics analysis (interactome, phosphoproteome, and proteome) of NGF-TrkA signaling. h, hours; ′, min. (D) List of NGF-induced interactors of TrkA that were robustly quantified by at least five peptides in the MS analysis and with a minimum twofold ratio change (table S1). (E) Overview of phosphoproteomic data presenting number of identified phosphorylation sites and phosphoproteins after NGF stimulation of SH-SY5Y-TR-TrkA cells. (F) Phosphorylation sites and distribution by amino acid from the phosphoproteomic data. (G) Overview of proteome data showing the number of identified proteins and those that significantly changed in abundance, tested by significance B with a Benjamini-Hochberg false discovery rate (FDR) <5%. (H) Overlap between time points for regulated proteins in the proteome.

To measure dynamic changes in NGF-TrkA signaling, we used a quantitative proteomics approach (30), which combined stable isotope labeling by amino acids in cell culture (SILAC) for quantification (31), peptide fractionation and enrichment, and high-resolution liquid chromatography–tandem mass spectrometry (LC-MS/MS) (fig. S1D). We analyzed cell lysates obtained after NGF stimulation for 5 to 120 min and 24 to 48 hours, which we define as early and late signaling, respectively (Fig. 1C and fig. S1, D and E). Overall, this setup provided us with three different layers of information on NGF-TrkA signaling: the TrkA interactome, phosphoproteome, and proteome. Among the early signaling events, we measured the dynamic recruitment of adaptor proteins to the activated receptor (meaning, the interactome) as well as the phosphorylation of proteins downstream of the activated receptor (the phosphoproteome). In the late signaling events, we measured changes in protein abundance associated with NGF-induced neurite outgrowth (the proteome) (Fig. 1C).

For the interactome analysis, we performed immunoprecipitation of TrkA using lysates from triple-SILAC–labeled, TrkA-expressing SH-SY5Y cells stimulated for 0, 5, or 10 min with NGF, and analyzed the mixed samples by LC-MS/MS (fig. S1D). We identified and quantified 1923 proteins and obtained good reproducibility between biological replicates (table S1 and fig. S2A). We identified 45 proteins that were recruited to TrkA upon NGF stimulation with significant ratios (table S1), 30 of which are presented in Fig. 1D. In addition to many of the known signaling interactors of TrkA, including Shc, growth factor receptor–bound protein 2 (Grb2), PLCγ, and the PI3K complex (15), we identified several proteins not previously known to interact with TrkA, including Cbl-b (Fig. 1D).

To analyze the phosphoproteome, we stimulated SH-SY5Y-TR-TrkA cells with NGF for 10, 45, or 120 min, which encompassed the sustained AKT and ERK signaling activity downstream of TrkA (Fig. 1A). At the later time point, activation of newly synthesized TrkA protein by NGF could affect phosphorylation of proteins (Fig. 1A). However, we were only interested in measuring phosphorylation of proteins derived from the activation of TrkA initially present at the cell surface at the time of NGF stimulation. We thus used cycloheximide (CHX) to efficiently block the appearance of the immature, 110-kD TrkA precursor after 60 min of NGF and the increase in the mature, 140-kD form of TrkA (fig. S1E) (32). Furthermore, the kinetics of TrkA down-regulation observed at 45 and 60 min was consistent with previous studies (33). In the phosphoproteomics setup, we therefore also compared cells that were exposed to NGF for 2 hours to those in which protein synthesis was blocked with CHX while cells were stimulated with NGF (fig. S1D). The phosphoproteomics setup included two parallel triple-SILAC experiments composed of two time-course experiments as described in the Supplementary Materials (fig. S1D). We identified and quantified 12,739 phosphorylated sites, 10,256 of which were confidently localized to serine (89.1% of the total), threonine (10.4%), or tyrosine (0.5%) residue in the peptide sequence (class I) within 2649 proteins with high reproducibility between replicates (Fig. 1, E and F, and fig. S2, B to D, and table S2). We deemed phosphorylation sites regulated if their ratios were higher or lower than the 2.5% most up- or down-regulated nonphosphorylated peptides, respectively; thus, cutoffs were individually determined for each time point. The condition for 120-min NGF + CHX was solely used to adjust the cutoffs for the 120-min time point. This was based on the assumption that phosphorylation sites regulated upon NGF stimulation downstream of the receptor initially present at the cell surface displayed a comparable behavior under the two conditions (with or without CHX). Most of the regulated phosphorylated sites showed an increased ratio (1851 sites, 18%), whereas 892 sites had decreased ratios (8.7%) (Fig. 1F). Phosphorylated tyrosine residues were almost twofold enriched among the regulated sites (from 0.5 to 1%) (Fig. 1F), supporting a more pronounced involvement of tyrosine phosphorylation in the early signaling downstream of TrkA.

To analyze the proteome, we stimulated SH-SY5Y-TR-TrkA cells with NGF for 0, 24, or 48 hours in a triple-SILAC setup to measure changes in protein abundance associated with neurite outgrowth. Total protein extracts from whole-cell lysates separated by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and in situ digested with trypsin were analyzed by LC-MS/MS (fig. S1D). We quantified in total 6076 proteins, of which 5096 were quantified in both biological replicates, which showed a good overlap (table S3 and fig. S3, A and B). We found 193 proteins with significant changes in protein abundance (Fig. 1, G to H, fig. S3C, and table S3).

This comprehensive proteomics analysis, covering early as well as late signaling events, represents a resource of NGF-mediated changes in recruitment of adaptors to the activated receptor, phosphorylation of downstream proteins, and protein abundance associated with neurite outgrowth in this neuroblastoma cell line.

The dynamic TrkA interactome identifies regulators of receptor ubiquitylation and stability

Because cellular outcome decisions are affected by coordinated protein-protein interactions (27, 28, 34), we reasoned that TrkA-mediated neurite outgrowth could depend on adaptor proteins bridging the receptor with downstream effectors. Gene Ontology (GO) term enrichment analysis of the 45 identified dynamic interactors of TrkA revealed that proteins with roles in migration, axon guidance, and endocytosis were specifically enriched (Fig. 2A), suggesting that the early recruitment of specific adaptors to TrkA is involved in the differentiation process. A functional network of the dynamic TrkA interactors was grouped into two modules: one related to signaling and the other related to ubiquitylation and endocytosis (Figs. 1D and 2B). Because protein function can be inferred from the presence of certain functional domains, we added information on the presence of the following domains: SH2, SH3, pleckstrin homology (PH), phosphotyrosine binding (PTB), tyrosine kinase binding (TKB), protein kinase, really interesting new gene (RING), and homologous to the E6AP carboxyl terminus (HECT) domains. The first five domains are known protein-protein interaction domains, whereas the protein kinase, RING, and HECT domains are catalytic domains that are characteristic of protein kinases and E3 ubiquitin ligases, respectively.

Fig. 2 Overview of the NGF-dependent dynamic TrkA interactome.

(A) Significantly overrepresented GO terms for biological process among TrkA interactors. (B) Dynamic protein interaction network centered on NTRK1 (TrkA in yellow) including the 45 TrkA interactors dynamically recruited upon NGF stimulation for 5 and 10 min. All interactors are represented by their gene names. Color and bar coding are described to the right of the network image. (C and D) Immunoblots of immunoprecipitated TrkA showing dynamic regulation of receptor phosphorylation (C) and ubiquitylation (D) and of receptor interaction with PLCγ, GAREM2, Shc, and Grb2 (C), or Cbl-b and Cbl (D) upon NGF stimulation for the indicated time points. A negative control (NC) without antibody for immunoprecipitation (IP) was included. Data are representative of three experiments.

Among the known signaling interactors of TrkA, we validated the interaction of Shc, Grb2, and PLCγ (Fig. 2C). In addition, we identified several proteins not previously known to interact with TrkA (Figs. 1D and 2B). For instance, we found a novel NGF-dependent interaction between TrkA and the Grb2-associated adaptors GAREM and GAREM2, which were previously associated with EGF and insulin-like growth factor 1 (IGF-1) signaling (35, 36). We validated the interaction with GAREM2 by TrkA coimmunoprecipitation and Western blot analysis (Fig. 2C). GAREM2 is functionally important for IGF-1–induced neurite outgrowth in neuroblastoma cells (35), suggesting that this adaptor is a general regulator of signal transduction underlying neurite outgrowth.

Because little is known about the events that regulate TrkA turnover, we focused on the module related to ubiquitylation and endocytosis (Fig. 2B). In our immunoprecipitation assays, we identified two ubiquitylation sites on TrkA, Lys447 and Lys775 (herein, K447 and K775), of which K447 was recently reported (37). We determined that these sites had increasing occupancy in response to NGF stimulation, with a kinetic profile similar to the TrkA-specific phosphorylation sites Tyr496 or Tyr676 and Tyr680: The kinetic profile for K447 resembled that of Tyr496, whereas that for K775 resembled that of the kinase activation loop sites Tyr676 and Tyr680 (fig. S4, A and B). From the gel slices in which TrkA was identified, we also identified diglycine remnant–modified, ubiquitin-derived peptides with K11, K48, and K63 as putative lysine linkage sites to other ubiquitin molecules in chain structure. All of these sites displayed an increase in SILAC ratio upon NGF stimulation (fig. S4B), suggesting that the increased ubiquitylation of TrkA in response to NGF (Fig. 2D and fig. S4, A and B) was caused by the recruitment of an E3 ubiquitin ligase. Among the E3 ubiquitin ligases, we found that Cbl-b, mind bomb 1 (Mib1), HECT, UBA, and WWE domain–containing protein 1 (HUWE1) and autocrine motility factor receptor (AMFR) were dynamically recruited to activated TrkA.

For further analyses, we focused our attention on Cbl-b for several reasons. Most importantly, Cbl-b exhibited the highest SILAC ratio at both time points compared to the other E3 ubiquitin ligases (table S1). Second, Cbl-b contains a TKB domain, making it relevant to study in relation to phosphotyrosine signaling (Fig. 2B). Last, because its family member Cbl mediates TrkA ubiquitylation to mark it for degradation (20). We found that Cbl-b displayed a dynamic interaction with TrkA upon NGF stimulation, whereas Cbl did not (Fig. 1D and table S1). Intrigued by these findings, which were validated by TrkA coimmunoprecipitation and Western blot analysis (Fig. 2D), we examined in more detail the behavior of the two Cbl family members. We studied the functional relevance of the dynamic Cbl-b versus the constitutive Cbl interaction with the receptor by analyzing TrkA ubiquitylation after Cbl-b or Cbl depletion by small interfering RNA (siRNA). After verifying the specificity and efficiency of Cbl-b and Cbl knockdown (fig. S4, C and D), we observed that the ubiquitylation, but not the phosphorylation, of TrkA decreased in Cbl-b–depleted as well as in Cbl-depleted cells stimulated with NGF for 10 min (Fig. 3, A and B, and fig. S4, C and D). However, the effect of Cbl-b depletion on TrkA ubiquitylation was much more pronounced compared to Cbl depletion. We also noticed that the size of the TrkA receptor was reduced, evident by the gel shift, most likely due to a decrease in basal receptor modification.

Fig. 3 The dynamic TrkA interactome identifies Cbl-b as a negative regulator of TrkA stability.

(A and B) Western blotting and quantification of TrkA ubiquitylation in lysates from SH-SY5Y-TR-TrkA cells transfected with a control (siGFP) or siRNA against CBLB (siCBLB) (A) or CBL (siCBL) (B) and stimulated with NGF for 10 min. An NC without antibody for immunoprecipitation was included. (C) Immunoblot and quantification of TrkA abundance after Cbl-b or Cbl depletion in SH-SY5Y-TR-TrkA cells pretreated with CHX and then stimulated with NGF for the indicated time points. Vinculin, loading control. (D) SH-SY5Y-TR-TrkA cells were treated with NGF for the indicated time, and lysates were immunoprecipitated for phosphotyrosine and immunoblotted for Cbl-b and Cbl. (E) Immunoblot and quantitation of Cbl-b and Cbl abundance in SH-SY5Y-TR-TrkA cells pretreated with CHX and then stimulated with NGF. All data are means ± SD of three experiments. *P < 0.05 and **P < 0.01 compared to controls by two-sample Student’s t test (A and B) with Bonferroni correction (C and E). Blots are representative of three experiments.

Furthermore, TrkA stability upon CHX treatment and subsequent NGF stimulation was significantly increased in the absence of Cbl-b, but not upon Cbl depletion (Fig. 3C). These findings point to Cbl-b as the prominent E3 ubiquitin ligase involved in TrkA ubiquitylation and turnover in neuroblastoma cells in response to NGF, whereas Cbl only plays a role in receptor ubiquitylation but not turnover. Phosphorylation of tyrosine residues within the N terminus is important for the ubiquitin ligase activity of the Cbl family members (3840). In response to NGF treatment for 2 to 10 min, tyrosine phosphorylation of Cbl-b, but not Cbl, markedly increased (Fig. 3D), suggesting that Cbl-b recruitment and activity are stimulated by phosphorylation in response to NGF. Furthermore, we could detect a rapid 75% reduction in Cbl-b protein abundance 45 min after NGF stimulation in CHX-treated cells, whereas the abundance of Cbl protein remained constant (Fig. 3E). Together, these data demonstrated that Cbl-b has several unique functions compared to Cbl in modulating TrkA in neuroblastoma cells. Despite a shared ability to affect TrkA ubiquitylation upon NGF stimulation, only Cbl-b was dynamically recruited to TrkA, tyrosine-phosphorylated, and rapidly degraded and had the ability to affect TrkA turnover in response to NGF.

Phosphoproteomics links sustained signaling activation with neurite outgrowth in neuroblastoma

Cbl-b and Cbl have established roles as negative regulators of RTK signaling, and here, we described Cbl-b as a negative regulator of TrkA. Alongside processes controlling signaling attenuation, a wealth of signaling events is activated upon NGF binding to TrkA. We next analyzed these changes with a specific focus on sustained signaling activation and its influence on neurite outgrowth. To define TrkA signaling dynamics, we examined the regulated phosphoproteome by fuzzy c-means clustering, including the 2048 regulated phosphorylated sites with a ratio for all the three time points upon stimulation with NGF. We defined proteins phosphorylated in response to NGF stimulation as “responders.” The analysis revealed seven distinct temporal profiles corresponding to sustained (clusters 1, 2, and 3), early or late decreasing (clusters 4 and 5), cycling (cluster 6), and transient (cluster 7) responders (Fig. 4A and table S2). GO enrichment analysis showed a significant overrepresentation of members of MAPK signaling and neuronal differentiation in the sustained responders (clusters 1 to 3) (Fig. 4B). Our analysis also revealed that sites on the same protein can have different temporal profiles as exemplified by, for example, Shc1 and Dock7 (fig. S5A). The phosphoproteins with more strongly increased ratios per site were protein kinases, adaptors, and GTPase (guanosine triphosphatase) regulators involved in MAPK signaling and differentiation (fig. S5A); this reinforces the interconnection between these two processes in neuroblastoma cells.

Fig. 4 Sustained signaling activation underlies neurite outgrowth.

(A) Clustering based on fuzzy c-means of the 2048 phosphorylation sites quantified at 0 to 120 min. (B) Hierarchical clustering of GO terms for biological process enriched within at least one of the signaling clusters shown in (A). Red boxes highlight terms specifically enriched within a cluster and whose members are shown in fig. S5A. (C) Kinase groups significantly represented in clusters 1 to 3 by NetworKIN analysis (P < 0.05, one-sided Kolmogorov-Smirnov test with FDR correction). (D) Percentage of neurite-bearing cells upon treatment with the indicated kinase inhibitors and NGF for 24 hours. Data are means ± SD of three experiments. *P < 0.05 and **P < 0.01 in comparison to DMSO-treated cells (two-sample Student’s t test with Bonferroni correction).

To identify the protein kinases mediating the neurite outgrowth response, we performed a cluster-dependent sequence motif analysis of the phosphorylation sites. All clusters except cluster 1 displayed a strong preference for proline residues in the +1 position to the phosphorylated residue, and phosphorylation sites within the sustained clusters 1 to 3 exhibited a preference for the R-X-X-pS/T motif (fig. S5B). In addition, we used the integrative computational approach NetworKIN (41) on the sustained clusters to predict kinase-substrate relations based on consensus sequence motifs and protein-protein interactions. We identified 23 kinase groups to be significantly overrepresented (Fig. 4C), including the classical basophilic kinases protein kinase A (PKA), PKB (AKT), and PKC, which explains the preference for arginine residues in the −3 position for the phosphorylated sites belonging to clusters 1 to 3. Likewise, the preference for proline residues in +1 is explained by proline-directed kinases, such as ERK1 and p38, which are activated downstream of the MAPK2 kinase group (Fig. 4C and fig. S5B).

To further dissect the functional relevance of these kinases for neurite outgrowth, we used a range of pharmacological kinase inhibitors to selectively target mitogen-activated or extracellular signal–regulated protein kinase kinase 1/2 (MEK1/2), PI3K-AKT, ERK5, p38, c-Jun N-terminal kinase (JNK), p90 ribosomal S6 kinase (RSK), glycogen synthase kinase 3 (GSK3), and PKC. All used inhibitors were specific for their targets (fig. S6A) and, with the exception of the JNK inhibitor (JNK-IN-8), significantly reduced the number of neurite-bearing cells in response to NGF (Fig. 4D). For cells treated with the JNK inhibitor, we observed a trend toward longer and less ramified neurites compared to dimethyl sulfoxide (DMSO)–treated control, suggesting that JNK signaling impedes neurite outgrowth under normal conditions (fig. S6B). Together, these data underline the importance of performing temporal phosphoproteomics to interconnect kinase networks, TrkA sustained signaling activation, and neurite outgrowth in neuroblastoma cells.

Cbl family of proteins modulates ERK activation and neurite outgrowth in neuroblastoma cells

Because sustained NGF-TrkA signaling is important for neurite outgrowth (Fig. 4D), we also wanted to test the impact of TrkA proximal regulatory events. Therefore, we examined the contribution of the Cbl family to neurite outgrowth in our neuroblastoma model cell line. We first examined neurite outgrowth in Cbl-b–depleted and control SH-SY5Y-TR-TrkA cells in the absence of NGF or upon 24 hours of NGF stimulation. Neurite outgrowth was detected upon NGF stimulation in control cells and also in Cbl-b–depleted cells independent of NGF stimulation (Fig. 5, A and B). The already high number of neurite-bearing cells in Cbl-b–depleted cells did not further increase upon NGF stimulation (Fig. 5, A and B). Moreover, we observed an increased abundance of basal ERK phosphorylation in control cells depleted of Cbl-b (Fig. 5C), which is consistent with an essential role of ERK activity in mediating neurite outgrowth in neuroblastoma cells (Fig. 4D). To test if the effect observed upon Cbl-b depletion was independent of TrkA expression, we depleted Cbl-b in several neuroblastoma cell lines that have a low abundance of TrkA relative to SH-SY5Y-TR-TrkA cells (fig. S7, A and B). In the parental SH-SY5Y cell line, Cbl-b depletion resulted in a phenotype similar to that of TrkA-induced cells (Fig. 5, D to G). We confirmed these findings in two additional neuroblastoma cell lines, CLBGA and NB1 (Fig. 5H and fig. S7C). Furthermore, our findings were validated in NB1 cells using single sequence siRNA-mediated depletion of Cbl-b (fig. S7, D and E). Together, these results showed that Cbl-b negatively regulated ERK activation and neurite outgrowth independent of TrkA expression in neuroblastoma cells.

Fig. 5 Cbl-b inhibits ERK activation and neurite outgrowth.

(A to C) Representative images of neurite outgrowth (red arrows) (A) in SH-SY5Y-TR-TrkA cells treated with the indicated siRNA and stimulated with NGF for 24 hours. Neurites are quantified in (B), and lysates from the cells in each condition were immunoblotted for phosphorylated ERK in (C). Vinculin, loading control. Scale bar, 50 μm. (D to G) Images (D) and quantification (E) of neurite outgrowth (red arrows) and immunoblotting (F) and quantification (G) of phosphorylated ERK in SH-SY5Y-TR-TrkA and parental SH-SY5Y cells treated with control or CBLB siRNA. Scale bar, 50 μm. (H) Quantification of neurite outgrowth in CLBGA and NB1 cells treated as indicated. Representative images are shown in fig. S7C. All data are means ± SD of three experiments. *P < 0.05 and **P < 0.01 in comparison to siGFP control cells (two-sample Student’s t test).

We also examined the effect of Cbl depletion on neurite outgrowth and ERK activation. These experiments were performed with single siRNA sequences targeting CBL (Fig. 6A). Similar to Cbl-b depletion described above, we found that Cbl depletion also elicited the outgrowth of neurites from SH-SY5Y-TR-TrkA cells (Fig. 6, B and C) and increased ERK phosphorylation independent of NGF stimulation (Fig. 6, D and E). Furthermore, the increase in ERK activation upon depletion of Cbl-b and Cbl was not impaired by inhibition of TrkA activity using the inhibitor GW441756 (24) (Fig. 6F), suggesting that the observed effects rely on mechanisms downstream of TrkA. The phenotype observed upon the individual depletion of Cbl-b and Cbl had a striking resemblance to NGF-induced neurite outgrowth. In addition, NGF stimulation reduced Cbl-b and Cbl protein abundance in SH-SY5Y-TR-TrkA cells, but not in parental SH-SY5Y cells (Fig. 6G), indicating that low abundance of Cbl family members is required for the NGF-TrkA–elicited cellular response. Investigating the underlying mechanism, we detected only a minor, not statistically significant change in Cbl-b and Cbl mRNA amounts after stimulation with NGF for 24 hours (Fig. 6H). Thus, a reduction in protein abundance of Cbl family members plays a major role in neurite outgrowth, which is in line with both proteins showing increased ubiquitylation in response to NGF (Fig. 6I). However, only the Cbl-b protein abundance was efficiently reduced at early time points of NGF stimulation (Fig. 3E). A possible explanation for this finding could be differences in the relative abundances between the two proteins, with NGF having a more pronounced effect on the less abundant protein (42). These data suggest that regulated degradation of the Cbl family of proteins plays a major role in NGF-TrkA–mediated responses in neuroblastoma cells.

Fig. 6 The Cbl family of proteins inhibits ERK-dependent neurite outgrowth and is regulated in response to NGF.

(A to E) Immunoblot of Cbl abundance (A), images and quantification of neurite outgrowth (orange arrows) (B and C), and immunoblotting and quantification of phosphorylated ERK (D and E) in SH-SY5Y-TR-TrkA cells transfected with control siRNA (siGFP) or one of two siRNAs against CBL (siCBL #1 or #3). Blot is representative and data are means ± SD of three experiments. *P < 0.05 and **P < 0.01 compared to siGFP cells (two-sample Student’s t test with Bonferroni correction). (F) Immunoblotting of lysates from SH-SY5Y-TR-TrkA cells transfected with siGFP, siCBLB, or siCBL and treated with 2 μM GW441756 (TrkA inhibitor). Blots are representative of two experiments. (G) Immunoblotting for Cbl-b and Cbl abundance in SH-SY5Y-TR-TrkA and parental SH-SY5Y cells treated as indicated. Blots are representative of two experiments. (H) Cbl-b and Cbl mRNA amounts in SH-SY5Y-TR-TrkA cells after 24 hours of NGF stimulation. Data are means ± SD of three experiments (one-sample Student’s t test). (I) Immunoprecipitation and immunoblotting for Cbl-b or Cbl ubiquitylation in SH-SY5Y-TR-TrkA cells treated with NGF for up to 10 min. An NC without antibody for immunoprecipitation was included. Blots are representative of three experiments.

TrkA signaling has been extensively studied in rat PC12 cells (13, 14). Unlike SH-SY5Y-TR-TrkA cells, the abundance of Cbl-b and Cbl protein was unaffected in PC12 cells stimulated with NGF for up to 48 hours (fig. S8A). Because Cbl was previously associated with TrkA in PC12 cells, although with contradicting functions (18, 20), we tested the dynamic interaction between Cbl-b and TrkA in PC12 cells. Upon immunoprecipitation of Cbl-b and Western blot analysis for TrkA, we found no dynamic increase in association between the two proteins (fig. S8B). Depletion of Cbl-b in PC12 cells (fig. S8C) showed no difference in ERK activity between control and depleted cells upon NGF treatment (fig. S8D). Cbl-b depletion was also not sufficient to induce neurite outgrowth in the absence of NGF (fig. S8E). These data suggest a clear species-specific and cell context–dependent difference in TrkA signaling between the transfected human neuroblastoma cell line SH-SY5Y-TR-TrkA and the rat cell line PC12. To explain the differential phenotype between these two cell lines, we looked at the relative protein abundance of TrkA, Cbl, and Cbl-b, because it has been previously suggested that EGF-induced down-regulation of Cbl-b may depend on the relative protein abundance of Cbl-b and the EGF receptor (42). Although TrkA, Cbl-b, and Cbl were expressed at relatively lower amounts in the PC12 cells compared to SH-SY5Y-TR-TrkA cells (fig. S8F), Western blot analysis did not reveal any interprotein relation within the same cell line. Therefore, we performed a proteome analysis of PC12 cells including a triple-SILAC experiment with stimulation of cells with NGF for 0, 24, and 48 hours (fig. S8G and table S4). Using the intensity-based absolute quantification (iBAQ) method (43), we measured the relative abundance of TrkA, Cbl-b, and Cbl in PC12 and SH-SY5Y-TR-TrkA cells (tables S3 and S4). We used iBAQ values from the unstimulated condition (“light”-labeled cells) derived from both proteome analyses and presented the data in absolute values and relative to Cbl-b (fig. S8H). The comparison of iBAQ values revealed that the relative abundance between the proteins differed within the same cell line. We found the TrkA abundance to be in excess of Cbl-b (>100-fold) but similar to Cbl in SH-SY5Y-TR-TrkA cells, which, in agreement with Ettenberg et al. (42), explains the preferential rapid degradation of Cbl-b compared to Cbl in response to NGF observed in these cells (Fig. 3E). In contrast, neither Cbl-b nor Cbl was expressed at lower amounts compared to TrkA in PC12 cells, which explains why we did not see any regulation of Cbl family protein abundance in response to NGF in PC12 cells (fig. S8A). In line with this, the lack of regulation of Cbl-b and Cbl in the parental SH-SY5Y cells (Fig. 6G) may also be explained by the undetectable amount of TrkA in these cells (fig. S7B).

Thus far, our findings support Cbl-b to be a previously uncharacterized E3 ubiquitin ligase for TrkA with the ability to promote receptor ubiquitylation and decrease receptor stability and whose abundance is rapidly reduced in response to NGF stimulation in SH-SY5Y-TR-TrkA cells. Low abundance of the Cbl-b and Cbl proteins is necessary for neurite outgrowth as supported by concomitant ERK activation, thereby establishing Cbl-b and Cbl as modulators of neurite outgrowth in neuroblastoma cells.

Changes in the proteome are associated with ERK-regulated responses

Because Cbl-b and Cbl protein abundances were reduced by long-term NGF stimulation in SH-SY5Y-TR-TrkA cells (Fig. 6G), we analyzed other changes in protein abundance associated with neurite outgrowth. With a focus on the proteins identified in the proteome analysis (Fig. 1G), we performed fuzzy c-means clustering of their temporal abundance profiles, and four distinct patterns of regulation emerged. Clusters 1 and 2 comprised the proteins with steadily increasing abundance and abundance reaching a plateau after 24 hours of NGF, respectively. The proteins with decreased abundance fell into clusters 3 and 4, showing steadily decreasing abundance and proteins with a more transient abundance profile, respectively (Fig. 7A). We identified the classical neuronal markers such as NGF-responsive neuropeptide VGF and growth-associated protein 43 (GAP43) (belonging to cluster 1), which confirmed the induction of a differentiated phenotype upon long-term NGF stimulation. We also found several protein kinases and transcriptional regulators with a change in protein abundance. Of these, we validated by Western blot analysis glucose-regulated protein 78 (GRP78, also known as HSPA5), which has been associated with the more differentiated phenotype of neuroblastoma tumors and suggested as an independent marker of good prognosis (44) (Fig. 7B). Several regulators of ERK activity, such as dual specificity phosphatase 4 (DUSP4), sprouty 4 (Spry4), and lysine-deficient protein kinase 2 (WNK2), were also regulated (Fig. 7B) (4547), suggesting a complex interplay between these proteins in the regulation of neurite outgrowth. It is known that DUSP4 can dephosphorylate ERK and JNK (48); however, despite an increasing DUSP4 abundance, we found that ERK phosphorylation was sustained (Fig. 5C). This suggests DUSP4 selectivity toward JNK, which is in agreement with our observation that JNK inhibition facilitates neurite outgrowth (Fig. 4D). The increase in GRP78 and DUSP4 in response to NGF was blocked by the use of CHX (Fig. 7C), which supports that new protein synthesis is the underlying cause of their increase in protein abundance. Several protein kinases decreased in abundance, including the RTK anaplastic lymphoma kinase (ALK) with an established role in poor-prognosis neuroblastoma (1) (Fig. 7B). Apart from VGF, none of the above-mentioned proteome changes for SH-SY5Y-TR-TrkA cells were confirmed in the PC12 proteome (table S4), thus underscoring species-specific and cell context–dependent effects.

Fig. 7 Proteome changes are associated with ERK-regulated responses.

(A) Dynamic clustering of all proteins for which abundance was significantly regulated by NGF over 48 hours. (B) Immunoblots of selected proteins in SH-SY5Y-TR-TrkA cells stimulated with NGF for 24 or 48 hours. Blots are representative of three experiments. (C) Immunoblotting for GRP78, DUSP4, and vinculin in SH-SY5Y-TR-TrkA cells pretreated with CHX and stimulated with NGF stimulation for the indicated time points and lysates. Blots are representative of two experiments. (D) Significantly overrepresented GO terms for biological process of NGF-regulated proteins. (E) Significantly enriched transcription factor binding sites for genes of NGF-regulated proteins. (F) Model of NGF-induced signaling in the SH-SY5Y-TR-TrkA neuroblastoma cell line. Findings supported by our study are indicated with orange dotted arrows.

Enrichment analysis of the regulated proteins revealed only borderline significant GO terms (Fig. 7D); however, we found that the proteins were preferentially encoded by genes with binding sites for certain transcription factors (Fig. 7E). The two most significant of these, cAMP (adenosine 3′,5′-monophosphate) response element–binding (CREB) and Myc and Max, are well-known mediators of ERK signaling (49, 50), which we identified in the proteome analysis (table S3). Therefore, they may mediate ERK-induced transcriptional responses supporting the proteome changes associated with neurite outgrowth.

In conclusion, our extensive temporal proteomic analysis of TrkA signaling serves as a functional resource to identify candidate biomarkers and regulators of neurite outgrowth related to neuroblastoma differentiation. This is supported by the identification of Cbl-b as an E3 ubiquitin ligase for TrkA and by the finding that low Cbl-b protein abundance enhanced ERK activation and consequently promoted neurite outgrowth in neuroblastoma cells (Fig. 7F).

DISCUSSION

This study addressed NGF-TrkA signaling underlying neurite outgrowth in a neuroblastoma cell line model using comprehensive interactome, phosphoproteome, and proteome analysis combined on a temporal scale. Besides representing a resource of NGF-TrkA signaling events in neuroblastoma cells, it validated the unique combination of MS-based quantitative proteomics and functional assays as crucial means to elucidate molecular mechanisms related to cellular outcome decisions (27). We identified an inhibitory role for Cbl-b in NGF-TrkA signaling and confirmed a regulation of Cbl-b analogous to T cell receptor and epidermal growth factor receptor (EGFR) signaling (42, 5153). However, several regulatory and mechanistic aspects remain to be elucidated. For example, how Cbl-b becomes phosphorylated and whether the interaction with activated TrkA is direct or mediated by other proteins have not yet been addressed. Cbl-b contains a TKB domain, which, like SH2 domains, can directly bind phosphorylated tyrosines on an activated RTK. Another likely mechanism of TrkA–Cbl-b interaction involves the SH2B adapter protein 2 (SH2B2, also known as APS), which was identified in our interactome data to be among the proteins displaying the highest interaction ratios. SH2B2 associates with TrkA requiring the phosphorylation of only the catalytic activation loop tyrosines of TrkA (54), and one of its major functions is recruitment of Cbl family members to facilitate their tyrosine phosphorylation (5558). Studies of TrkA mutants suggest additional pathways for supporting ERK activation independent of the involvement of the classical adaptors Shc and PLCγ. In the presence of a TrkA double mutant defective in recruiting Shc and PLCγ, NGF is still able to support ERK activation and neurite outgrowth in SH-SY5Y but not in PC12 cells (11, 59). Because this double mutant preserves the ability to trans-phosphorylate the catalytic activation loop tyrosines required for SH2B2 binding, we speculate that the rise in ERK activation may be elicited via dynamic recruitment of SH2B2 and Cbl-b to TrkA in SH-SY5Y-TR-TrkA cells. However, further studies are required to address this. Another unclear aspect of Cbl-b biology is how the protein is rapidly degraded in response to growth factor stimulation. Our data suggested regulation of Cbl-b protein stability as a key regulatory mechanism, but whether the proteasomal, lysosomal, or both degradation pathways are involved remains elusive (42). However, on the basis of a high similarity between the NGF-dependent kinetics of Cbl-b and TrkA down-regulation, we speculate that Cbl-b is degraded together with internalized TrkA in our cellular system. In agreement with EGFR-mediated down-regulation of Cbl-b (42), excess of TrkA compared to Cbl-b, evident for SH-SY5Y-TR-TrkA cells and not PC12 cells, was associated with Cbl-b down-regulation. These data supported the hypothesis that the down-regulation of Cbl-b by RTKs depends on a defined stoichiometric relationship at the protein level. Although we focused on the Cbl family in this study, the other E3 ubiquitin ligases AMFR, HUWE1, and Mib1 may also be important for NGF-dependent TrkA ubiquitylation, internalization, and degradation, suggesting a more complex interplay between E3 ubiquitin ligases regulating these processes such as recently proposed for EGFR (60).

We showed in several ways that low abundance of Cbl-b and Cbl protein was linked to neurite outgrowth. An intriguing theory to explain these findings is the potential influence of turnover and/or activation of other RTKs, which, in turn, would promote sustained ERK phosphorylation and neurite outgrowth. Indeed, the Cbl family of E3 ubiquitin ligases has many known RTK substrates regulating neurite outgrowth in a neuroblastoma setting, such as insulin-like growth factor 1 receptor (IGF-1R) and fibroblast growth factor receptor 1 (FGFR1) (35, 6163). Our findings argue for development of strategies to interfere with the function of Cbl family of proteins as a method for modulating neurite outgrowth and potentially differentiation in neuroblastomas.

The neuroblastoma research field has gained valuable information from the extensive use of mRNA expression profiling, and our protein level data suggested that complementation by MS-based proteomics could advance the field further. Proteomics analysis of neuroblastoma tissue samples could, for instance, determine whether low-abundance Cbl-b and Cbl correlate with neuroblastoma differentiation. Ultimately, this comprises valuable protein level information in the quest to identify stronger prognostic protein biomarkers as well as novel therapeutic targets. In addition, our phosphoproteomics approach refined the view of NGF-TrkA downstream effector protein kinases categorized as “sustained responders.” These were proven important for neurite outgrowth using a small-molecule inhibitor–based assay. Despite potential inhibitor-related off-target effects, our data support the idea that the sustained signaling code may confer functional specificity for neurite outgrowth. Thus, future treatment of neuroblastoma may rely on therapeutics used as precision tools to specifically block or redirect certain cellular signaling responses toward desired dynamic profiles.

In conclusion, the data presented here provide insight into the regulation of TrkA ubiquitylation and turnover by Cbl-b and revealed an intriguing role for the Cbl family of proteins as modulators of neurite outgrowth in neuroblastoma cells. Therefore, our large-scale analysis of temporal NGF-induced TrkA signaling extends our understanding of NGF-TrkA–mediated responses and offers a wealth of candidates for further investigation.

MATERIALS and METHODS

Reagents

The following commercial reagents were used: Tet hydrochloride and CHX (Sigma-Aldrich); Murine 2.5S Nerve Growth Factor (Promega); the PI3K inhibitor LY294002 (Cell Signaling Technology); the MEK inhibitor U0126 (Cayman Chemical Company); the RSK inhibitor BID1870, the JNK inhibitor JNK-IN-8, and the PKC inhibitor Gö6979 (Calbiochem); the MEK5/ERK5 inhibitor BIX02189, the p38 inhibitor BIRB796, and the GSK3 inhibitor CHIR99021 (Selleckchem). Stock solutions for all inhibitors were prepared in DMSO and stored at −20°C.

Antibodies against phosphorylated TrkA (raised in rabbit against Tyr674/675, corresponding to Tyr680/681 for TrkA isoform II), TrkA, phosphorylated or total PLCγ (raised in rabbit), phosphorylated or total AKT (raised in rabbit), phosphorylated or total ERK1/2 (raised in mouse or rabbit, respectively), phosphorylated or total ERK5 (raised in rabbit), phosphorylated or total ATF-2 (raised in rabbit), phosphorylated or total GSK3 (raised in rabbit), phosphorylated or total HSP27 (raised in rabbit or mouse, respectively), ALK, DUSP4, BIP (GRP78), and lamin A/C (all raised in rabbit) were purchased from Cell Signaling Technology. Antibody against vinculin (raised in mouse) was from Sigma-Aldrich. Antibodies against Cbl-b and ubiquitin (P4D1) (both raised in mouse), Spry4, Lyn, and Cbl (C-15) (each raised in rabbit) used in immunoprecipitation assays were from Santa Cruz Biotechnology. Antibodies to GAP43, TrkA, and WNK2 (each raised in rabbit) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and Shc (both raised in mouse) used in immunoprecipitations were from Abcam. Antibodies to Grb2 and c-Cbl (both mouse) were from BD Biosciences, and that to phosphotyrosine (4G10, raised in mouse) used in immunoprecipitations was from EMD Millipore. The antibody to GAREM2 (raised in rabbit) was provided by H. Konishi (University of Hiroshima, Hiroshima, Japan) and was previously described (35). Secondary horseradish peroxidase–conjugated antibodies against rabbit or mouse immunoglobulin G (IgG) and raised in goat were from Jackson ImmunoResearch Laboratories.

Cell culture

The human neuroblastoma cell lines SH-SY5Y, CLBGA, and NB1 were cultured in RPMI 1640 (Gibco) supplemented with 10% fetal bovine serum (Gibco), 2 mM l-glutamine (Gibco), and penicillin (100 U/ml) and streptomycin (100 μg/ml) (Gibco). The CLBGA and NB1 cell lines were a gift from F. Speleman (University of Ghent, Ghent, Belgium). The rat pheochromocytoma PC12 cell line was cultured in RPMI 1640 supplemented with 10% horse serum (Gibco), 5% fetal bovine serum (Gibco), 2 mM l-glutamine (Gibco), 1 mM sodium pyruvate (Gibco), and penicillin (100 U/ml) and streptomycin (100 μg/ml) (Gibco). PC12 cells were cultivated on BioCoat Collagen I Cellware (Corning Incorporated).

Generation and maintenance of stable cell lines

The number of cell lines with high endogenous TrkA expression is limited, and constitutive ectopic expression of TrkA in a neuroblastoma setting is often associated with receptor autophosphorylation and loss of expression upon long-term culture (64). We therefore developed a Tet-inducible system for TrkA expression. SH-SY5Y cells were stably transfected with the plasmid for expression of the Tet repressor (TR) pcDNA6/TR (Invitrogen), and single cell clones were obtained after selection in the presence of blasticidin (7.5 μg/ml) (Invitrogen). A clone with high expression of TR resulted in the cell line SH-SY5Y-TR and was then stably transfected with the TrkA expression plasmid pT-REx-DEST30-TrkA. The sequence for TrkA corresponded to that of TrkA isoform II, which is specifically expressed in neuronal cells (65). The sequence of codon-optimized TrkA complementary DNA (cDNA), which did not alter the wild-type TrkA amino acids, was cloned into a Tet-inducible vector pT-REx-DEST30 (Invitrogen) as previously described (http://tinyurl.com/llp9k8u). Single cell clones were generated by antibiotic selection using blasticidin (7.5 μg/ml) and G418 (500 μg/ml) (Sigma-Aldrich). A clone with the ability to express a high amount of TrkA upon Tet induction resulted in the cell line SH-SY5Y-TR-TrkA. Culturing conditions for SH-SY5Y-TR and SH-SY5Y-TR-TrkA were similar to parental SH-SY5Y with the addition of their respective selection antibiotics. All cell lines were maintained at 37°C in a humidified atmosphere at 5% CO2.

Cell stimulation, lysis, and Western blotting

Cells were cultured in six-well plates in complete medium and serum-starved for 6 hours in serum-free medium before short-term (5 to 120 min) stimulation with NGF (100 ng/ml). Whenever the SH-SY5Y-TR-TrkA cell line was used, TrkA expression was always induced using Tet (1 μg/ml) 48 hours before NGF stimulation unless otherwise stated. To block for new protein synthesis, the translational inhibitor CHX (Sigma-Aldrich) was used (20 μg/ml), and cells were preincubated for 1 hour before NGF stimulation.

Whole-cell extracts were generated by lysing the cells in modified radioimmunoprecipitation assay (RIPA) buffer [50 mM tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.1% sodium deoxycholate with the addition of 5 mM β-glycerophosphate, 5 mM sodium fluoride, 1 mM sodium orthovanadate, and 1 Complete Inhibitor Cocktail tablet per 10 ml (Roche)]. Protein concentration was determined using Bio-Rad Bradford reagent. Proteins (20 to 80 μg per lane) were resolved by SDS-PAGE using 4 to 12% bis-tris gradient gels (NuPAGE, Invitrogen) and MOPS buffer and transferred onto Whatman Protran nitrocellulose membranes (Sigma-Aldrich). Membranes were stained for total protein amounts with Ponceau S (Sigma-Aldrich) and blocked with phosphate-buffered saline and 0.1% Tween 20 containing either 5% skim milk powder (Sigma-Aldrich) or 5% albumin from bovine serum (Sigma-Aldrich) before incubating with the primary antibodies as indicated. Incubation with primary antibodies was performed overnight at 4°C and followed by 1 hour of incubation with species-specific peroxidase-conjugated secondary antibodies. The Novex ECL Chemiluminescent Substrate Reagent Kit (Invitrogen) was used to visualize the protein bands, and bands were detected by exposure to Hyperfilm (Amersham, GE Healthcare). To detect multiple proteins, the antibodies were removed from the membrane in stripping buffer [500 mM glycine (pH 2.5)], washed, and then reprobed with antibodies against the corresponding proteins. Western blot analyses were performed on at least three independent sets of lysates with similar results. Densitometric analysis was performed with the ImageJ software.

Immunoprecipitation

Cell extracts for immunoprecipitation were prepared using immunoprecipitation lysis buffer containing 50 mM tris-HCl (pH 7.5), 150 mM NaCl, 1 mM calcium chloride, 1% Triton X-100 with the addition of 5 mM β-glycerophosphate, 5 mM sodium fluoride, 1 mM sodium orthovanadate, and one Complete Protease Inhibitor Cocktail tablet (Roche) per 10-ml solution. Cell lysates (1 to 2 mg) were precleared with protein G–Sepharose (Invitrogen) and rabbit or mouse IgG from serum (Sigma-Aldrich) depending on the species of antibody used for immunoprecipitation. Lysates were then incubated overnight at 4°C with TrkA antibody (Abcam), phosphotyrosine antibody (4G10; EMD Millipore), and Cbl-b or Cbl antibody (Santa Cruz Biotechnology) with subsequent binding to protein G–Sepharose for 1 hour. After five washes with lysis buffer, the bound proteins were eluted by boiling in SDS sample buffer, resolved by SDS-PAGE, and analyzed by Western blotting.

Neurite outgrowth assay

For experiments including chemical inhibitors, cells were seeded in six-well plates in complete medium and serum-starved for 6 hours in serum-free medium before NGF stimulation (100 ng/ml) for 24 hours. Before NGF stimulation, cells were preincubated for 30 min with chemical inhibitors at the following concentrations: 20 μM LY294002, 20 μM U0126, 2 μM BID1870, 5 μM JNK-IN-8, 10 nM Gö6979, 5 μM BIX02189, 10 μM BIRB796, and 3 μM CHIR99021; each was determined to be the lowest concentration required to inhibit the primary target kinase or a downstream target. Control cells were preincubated with DMSO (0.001%). Neurite outgrowth was evaluated 24 hours after stimulation. When cells were treated with NGF alone or transfected with siRNA, cells were seeded in six-well plates in complete medium (unless otherwise indicated), and images were obtained after treatment with NGF at the indicated time points or 3 days after transfection, respectively. Each condition was performed in triplicate. Images were obtained using a phase-contrast microscope (DM1000, Leica) equipped with a DFC420C camera (Leica). Neurite-bearing cells were evaluated on at least five images for each condition in triplicate, and cells with neurites longer than the length of two cell bodies were manually counted.

RNA interference and transfection

SH-SY5Y, SH-SY5Y-TR-TrkA, CLBGA, NB1, and PC12 cells were transfected using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer’s instructions, and all the assays were performed 48 to 72 hours after transfection. Double-stranded Stealth siRNA oligonucleotides targeting human CBLB (sequence #1: 5′-UCAUCCCACCCUGUUUCCCUGAAUU-3′; sequence #2: 5′-GGUCCAUCUUCAGAGAAGAAAUCAA-3′; sequence #3: 5′-CAUGGGAGAGGGUUAUGCCUUUGAA-3′), human CBL (sequence #1: 5′-CCAGCAGAUUGAUAGCUGUACGUAU-3′; sequence #2: 5′-GCGGAGAAUCAACUCUGAACGGAAA-3′; sequence #3: 5′-CCUACCAGGACAUCCAGAAAGCUUU-3′), and rat CBLB (sequence #1: 5′-GGAAUCUCACAAAGCUGUCCCUUAU-3; sequence #2: 5′-GGGACGGCAAUAUCCUACAGACCAU-3′; sequence #3: 5′-CCGAAGUCUUCUCCAUGCAUGGUUA-3′) were purchased from Invitrogen. Cells were transfected with a mixture of all three CBLB- or CBL-targeting siRNA duplexes for a final concentration of 100 nM unless otherwise stated. As negative control, Stealth RNAi siRNA GFP (green fluorescent protein) Reporter Control duplex (Invitrogen) was used at a final concentration of 100 nM. Silencing of gene expression was monitored by Western blotting of cell lysates with an antibody against Cbl-b, recognizing Cbl-b of both human and rat origin or an antibody against Cbl.

Quantitative real-time reverse transcription polymerase chain reaction

Total RNA was isolated using the Qiagen RNeasy kit according to the manufacturer’s instructions. Total RNA was quantified using a NanoDrop spectrophotometer (Thermo), and cDNA was synthesized from 1 μg of total RNA using the QuantiTect Reverse Transcription Kit (Qiagen). Reverse transcription reactions were diluted 1:5 with H2O and stored at −20°C. Two microliters of each sample was mixed with Brilliant II SYBR Green QPCR Master Mix (Agilent Technologies) according to the manufacturer’s instructions. The following primers (250 nM final concentration) were used for qPCR (quantitative real-time reverse transcription polymerase chain reaction): GAPDH: 5′-CAGCGACACCCACTCCTCCA-3′ (forward) and 5′-GCTGGTGGTCCAGGGGTCTT-3′ (reverse) (Eurofins MWG Operon); CBLB: 5′-TGCCGATGCTAGACTTGGACGA-3′ (forward) and 5′-TGATGTGACTGGTGAGTTCTGCC-3′ (reverse); CBL: 5′-GCACGTTCAGTCTGGATACCTC-3′ (forward) and 5′-GCAGTTTTGGCACAGGAAGAGG-3′ (reverse) (OriGene Technologies). qPCR was performed using a Stratagene Mx3005P instrument with the following conditions: 10 min at 95°C followed by 40 cycles of 30 s at 95°C, 1 min at 55°C, and 1 min at 72°C. Experimental data were analyzed with MxPro software. Each sample was measured in technical triplicates, and each condition in biological triplicates. Quantification is presented as the ratio between the amount of target gene and the amount of GAPDH mRNA in each sample and relative to the untreated control using the ΔΔCt method (66).

SILAC labeling

For quantitative MS-based proteomics, SH-SY5Y-TR-TrkA and PC12 cells were labeled in SILAC RPMI (PAA Laboratories GmbH) supplemented with 10 or 15% dialyzed fetal bovine serum (Sigma), respectively, 2 mM l-glutamine (Gibco), penicillin (100 U/ml), and streptomycin (100 μg/ml) for at least 2 weeks to ensure complete incorporation of amino acids. Three cell populations were obtained: one labeled with natural variants of the amino acids (light label; Lys0, Arg0) (Sigma), the second labeled with medium variants of amino acids {L-[2H4]Lys (+4) and L-[13C6]Arg (+6)} (Lys4, Arg6), and the third labeled with heavy variants of the amino acids {L-[13C6,15N2]Lys (+8) and L-[13C6,15N4]Arg (+10)} (Lys8, Arg10). Medium and heavy variants of amino acids were purchased from Cambridge Isotope Laboratories.

Sample preparation for MS analysis

For strong cation exchange (SCX) chromatography and TiO2-based phosphopeptide enrichment, cells from light, medium, and heavy SILAC conditions were lysed separately at 4°C in ice-cold modified RIPA buffer [50 mM tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.1% sodium deoxycholate with the addition of 5 mM β-glycerophosphate, 5 mM sodium fluoride, 1 mM sodium orthovanadate, and one Complete Protease Inhibitor Cocktail tablet (Roche) per 50 ml of solution]. Proteins were precipitated overnight at −20°C in fourfold excess of ice-cold acetone. The acetone-precipitated proteins were solubilized in denaturation buffer [10 mM Hepes (pH 8.0), 6 M urea, 2 M thiourea], and 8 mg of protein from each SILAC condition was mixed 1:1:1. Proteins were reduced with 1 mM dithiothreitol for 60 min, alkylated with 5.5 mM chloroacetamide for 60 min, and digested initially with endoproteinase Lys-C for 3 hours (Wako), diluted fourfold with Milli-Q, and then digested with trypsin (modified sequencing grade, Sigma) overnight. All the steps were performed at room temperature. Enzyme activity was quenched by acidification of the samples with trifluoroacetic acid (TFA). The peptide mixture was desalted and concentrated on a C18 Sep-Pak cartridge (Waters) and eluted with 50% acetonitrile, 0.1% TFA. Peptide fractionation by SCX chromatography (67) was performed in a 1-ml Resource S column (GE Healthcare) on an ÄKTA FPLC system (GE Healthcare). The eluted peptide mixture was loaded directly onto a 10-ml injection loop and separated by a linear gradient from 100% SCX buffer A [5 mM KH2PO4 (pH 2.7), 30% ACN] to 30% SCX buffer B [5 mM KH2PO4 (pH 2.7), 350 mM KCl, 30% ACN] for 30 min, followed by isocratic (100%) buffer B for 6 min at a flow rate of 1.0 ml/min. Fractions of 2 ml were collected, of which some were pooled. Phosphopeptides were enriched using titansphere chromatography essentially as described previously (67). Briefly, titanium dioxide (TiO2) beads (10 μm, Titansphere, GL Sciences) were incubated with a solution of 2,5-dihydroxybenzoic acid (DHB) (20 mg/ml) (Sigma-Aldrich) in 80% ACN, 0.1% TFA for 30 min at room temperature. About 1 mg of TiO2-DHB beads was added to each SCX fraction or fraction pool and incubated under rotation for 30 min at room temperature. Early SCX fractions, mostly enriched in phosphopeptides, were incubated twice with TiO2-DHB beads for better coverage. The beads were washed once with SCX buffer B and once with 40% ACN, 0.5% TFA, and transferred in 80% ACN, 0.5% acetic acid on top of a C8 stage tip. The bound phosphopeptides were eluted directly into a 96-well plate by 5% NH4OH followed by 10% NH4OH, 25% ACN. The eluate was immediately concentrated in a SpeedVac at 60°C and acidified with 5% ACN, 1% TFA. Each sample was then desalted and concentrated on a C18 stage tip.

For Coomassie staining and in-gel digestion, cells from light, medium, and heavy SILAC conditions were lysed separately at 4°C in immunoprecipitation lysis buffer and modified RIPA for interactome and proteome samples, respectively. For the interactome analysis, TrkA was immunoprecipitated from 7 mg of protein in parallel for each SILAC condition before immunoprecipitation eluates were combined, and for proteome analysis, 50 μg of protein amounts from each SILAC condition was mixed before SDS-PAGE. The SILAC mixes were fractionated on 4 to 12% bis-tris gradient gels (Invitrogen) and stained with the Colloidal Blue Kit (Invitrogen) according to the manufacturer’s protocol. Gel lanes were divided into 5 (interactome) or 10 (proteome) different slices, and each gel slice was cut into 1-mm3 cubes. Gel slices were destained with 50% ethanol in 25 mM ammonium bicarbonate solution and dehydrated with absolute ethanol. Proteins were digested with trypsin (modified sequencing grade, Sigma) overnight. Trypsin activity was quenched by acidification with TFA, and peptides were extracted from the gel pieces with increasing concentrations of acetonitrile in 0.5% acetic acid. Organic solvent was evaporated in a SpeedVac vacuum centrifuge. Each sample was then desalted and concentrated on a C18 stage tip (68).

Liquid chromatography–tandem mass spectrometry

Peptides from all samples were eluted from C18 stage tips using 40% ACN, 0.5% acetic acid before online nanoflow LC-MS/MS analysis. Peptide mixtures were analyzed using an EASY-nLC system (Proxeon) connected to a Q Exactive mass spectrometer (Thermo Fisher Scientific) through a nanoelectrospray ion source. Peptides were separated in a 15-cm analytical column (75-μm inner diameter) in-house packed with 3-μm reversed-phase C18 beads (ReproSil-Pur AQ, Dr. Maisch) with a 180-min gradient from 6 to 60% acetonitrile in 0.5% acetic acid at a flow rate of 250 nl/min. Standard mass spectrometric parameters were as follows: spray voltage, 2 kV; no sheath and auxiliary gas flow, heated capillary temperature, 275°C; S-lens radio frequency level of 50%. The Q Exactive was operated in data-dependent acquisition mode using the “sensitive scanning method,” essentially as described (69) with a few modifications. Full-scan MS spectra [mass/charge ratio (m/z), 300 to 1750; resolution, 70,000 at m/z 200] were detected in the Orbitrap analyzer after accumulation of ions at 1e6 target value based on predictive AGC from the previous scan. For every full scan, the 10 most intense ions were isolated and fragmented (collision energy: 25%) by higher-energy collisional dissociation (HCD) with a fixed injection/fill time of 120 ms and 35,000 resolution. Finally, the dynamic exclusion was set to 30 s.

Mass spectrometry data analysis

Raw MS files were analyzed using MaxQuant software version 1.2.6.13 or 1.4.0.6 using the Andromeda search engine (70, 71) by which the precursor MS signal intensities were determined and SILAC triplets were automatically quantified. Proteins were identified by searching the HCD-MS/MS peak lists against a target/decoy version of the complete human UniProt database supplemented with commonly observed contaminants such as porcine trypsin and bovine serum proteins. Tandem mass spectra were matched with an initial mass tolerance of 6 ppm on precursor masses and 20 ppm for fragment ions. Cysteine carbamidomethylation was searched as a fixed modification. Protein N-acetylation, N-pyroglutamine, oxidized methionine, and phosphorylation of serine, threonine, and tyrosine were searched as variable modifications for the phosphoproteomics experiment. Protein N-acetylation, oxidized methionine, and deamidation of asparagine and glutamine were searched as variable modifications for the interactome and proteome experiments. Site localization probabilities were determined by MaxQuant using the PTM (post-translational modification) scoring algorithm (70, 72). The data sets were filtered by posterior error probability to achieve an FDR below 1% for peptides, proteins, and modification sites. Only peptides with Andromeda score >40 (unmodified and modified) were included in the total peptide list. Minimal peptide length was seven amino acids.

Bioinformatic analysis

For the phosphoproteomics data, only peptides with a phosphorylation site localization probability of at least 0.75 (class I, shown in table S2) (72) were included in the bioinformatic analyses. To identify phosphorylation sites with significantly regulated ratios, we compared the ratio distributions of all quantified phosphopeptides with all nonphosphorylated peptides that we expect not to change and therefore specify our technical variance. To determine the level of regulation, each analyzed time point was considered an independent experiment and cutoffs for up- and down-regulation were set to allow for an estimated 5% false-positive rate based on the distribution of ratios of identified and quantified nonphosphorylated peptides. Thus, cutoffs were individually determined for each time point, and the condition for 120-min NGF + CHX was solely used to adjust the cutoffs for the 120-min time point assuming that phosphorylation sites regulated upon NGF stimulation downstream of the receptor initially present at the cell surface displayed a similar behavior under the two conditions (±CHX) (table S2). For interactome data, ratios of identified and quantified dynamic interactors were normalized to the ratio of TrkA to account for uneven efficiency during individual pull-downs performed in parallel. Significantly changing dynamic interactors and proteome changes were identified for proteins with a ratio at all analyzed time points, and regulation was determined by significance B testing (P < 0.05) using Perseus version 1.3.9.10 (tables S1 and S3) (70).

Temporal profiling was performed for phosphorylation site changes and proteome expression changes by fuzzy c-means clustering using GProX 1.1.12 by requesting seven or four clusters, respectively, with a fuzzification parameter of 2 and 100 algorithm iterations (73). Clustering was performed on phosphorylation sites and proteins with a ratio at all time points and significantly regulated to at least one time point.

GO enrichment analysis was performed using InnateDB (74), and significantly overrepresented GO terms for biological process were determined using Fisher’s exact test and P values were corrected using the Benjamini-Hochberg FDR test to account for multiple testing. The analysis of cluster-assigned phosphoproteins was performed on each individual cluster, and significantly enriched GO terms within at least one signaling cluster were visualized by hierarchical clustering. Significantly overrepresented GO terms within the interactome and proteome data were represented in bar plots.

The overview of phosphoproteins represented in fig. S5A was done manually based on their function as protein kinase, adaptor, GTPase regulator, or other. In addition, to visualize the dynamics of the sites, heatmaps of the log2 ratios were created using the software R including only sites assigned to sustained signaling (clusters 1 to 3). Proteins included were grouped into “MAPK signaling,” which included the following associated significantly regulated GO terms: “MAPK cascade,” “nerve growth factor receptor signaling,” “activation of MAPKK activity” and “activation of MAPK activity,” and “Differentiation,” which included “axonogenesis,” “positive regulation of neuron differentiation,” “axon guidance,” and “neuron projection development.”

To asses for sequence bias around the regulated phosphorylation sites, sequence motif logo plots (±6 amino acids adjacent to the identified phosphorylated sites) were generated and visualized using the IceLogo software (75) with default parameters (P < 0.01). The analysis was performed independently on each group of cluster-assigned phosphorylation sites and compared with nonregulated site sequences, which was used as a common background. The nonregulated phosphorylation sites were defined as sites with ratios within less than 1 SD away from the mean of the distribution of identified nonphosphorylated peptides.

Kinase-substrate relationships were predicted using the NetworKIN algorithm (41), which combines the catalog of kinase consensus motifs from the NetPhorest atlas (41) with information for protein-protein associations from the STRING database (23203871). To calculate the differential kinase activity, predictions for all phosphorylation sites assigned to one of the sustained signaling clusters 1 to 3 were compared to the nonregulated sites. The analysis was restrained to those kinases that were experimentally identified within the neuroblastoma cell line used based on the MS analyses, resulting in the inclusion of 228 different kinases. We calculated the differential activity in each of the three different sets on the basis of a one-sided Kolmogorov-Smirnov test for increased activity. For each kinase, we compared the score distribution of its assigned sites to the distribution of the sites assigned to the kinase from the background set. The reported P values are adjusted for the FDR (76).

The protein network based on TrkA, interactome data were obtained by using the STRING database (version 9.1) (77). To ensure high confidence, only interactions assigned as experimentally verified were used and a confidence score over 0.7 was required. Further network analysis and visualization were performed by using the Cytoscape platform (version 3.1) (78). Enrichment analysis of transcription factor binding sites was performed using the DAVID bioinformatics resource with an FDR cutoff of 0.1% (79).

Estimation of site occupancy

We estimated the site occupancy in terms of modifications (phosphorylation or ubiquitylation) based on regulated phosphopeptide and diGly-containing peptide SILAC ratios, respectively, as previously described (80). The calculation is based on a set of equations that makes use of the information encoded in the SILAC ratio of the modified peptide (x), its corresponding nonmodified version (y), and the overall ratio of the protein based on the median SILAC ratio of all other unmodified peptides (z). The analysis is performed pairwise between the untreated control sample (“light” SILAC cells, termed “a”) and the NGF-stimulated sample (“medium” or “heavy” SILAC cells, termed “b”). The proportion between a modified peptide and its corresponding unmodified peptide counterpart in untreated control sample is calculated as a = (zy)/(xz), and in NGF-stimulated samples as b = [x*(zy))/(y*(xz)]. On the basis of this, the fractional modification (phosphorylation or ubiquitylation) site stoichiometry in control and NGF-stimulated samples can be calculated as a/(1 + a) and b/(1 + b) for each time point, respectively. This method applies equally well in case of missed cleavages induced by phosphorylation or ubiquitylation and assumes that (i) all three SILAC ratios are measured accurately (like in the Orbitrap mass analyzer); (ii) there are two states for the same peptide (modified and nonmodified); and (iii) the SILAC ratio changes between states.

Statistical analysis

Statistical tests were applied for experiments performed in three biological replicates. Results shown are the mean of three measurements ± SD. For neurite outgrowth experiments, P values were calculated by a two-sample Student’s t test [Figs. 1B, 4D, 5 (B, E, and H), and 6C and fig. S7D]. The test was applied on measured percentages of neurite-bearing cells, and normality of data was assumed because underlying counts of cells either positive or negative for bearing neurites were n > 10. A one-sample Student’s t test was used for calculating P values for the quantification experiments [Figs. 3 (A, B, and E), 5G, and 6 (E and H) and fig. S7E]. Here, ratios of control versus treatment were log2-transformed, and the mean of three experiments was tested to be different from 0. In Fig. 3C, P values were calculated using a two-sample Student’s t test on slopes calculated by the linear least squares regression model. Bonferroni correction was used to correct for multiple t test comparisons. Significance was concluded when P < 0.05 or P < 0.01 as indicated by * or **, respectively, in the figures.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/8/374/ra40/DC1

Fig. S1. Characterization of the Tet-inducible system for TrkA expression and experimental workflow.

Fig. S2. Quality control of interactome and phosphoproteome data.

Fig. S3. Quality control of proteome data.

Fig. S4. TrkA modifications and depletion of Cbl family of proteins.

Fig. S5. Sustained signaling and sequence motif analysis.

Fig. S6. Effects of inhibition of sustained signaling on neurite outgrowth.

Fig. S7. Increased neurite outgrowth in neuroblastoma cell lines upon Cbl-b depletion.

Fig. S8. Cbl-b functions are specific for neuroblastoma cells.

Table S1. Summary of SH-SY5Y-TR-TrkA interactome data.

Table S2. Summary of data from two biological replicates of TrkA-expressing SH-SY5Y cells stimulated with NGF for 10, 45, and 120 min.

Table S3. Summary of SH-SY5Y-TR-TrkA proteome data.

Table S4. Summary of PC12 proteome data.

REFERENCES AND NOTES

Acknowledgments: We thank members of the Proteomics Program at Novo Nordisk Foundation (NNF) Center for Protein Research (CPR) for valuable comments. The GAREM2 antibody was a gift from H. Konishi. Funding: The work carried out in this study was supported by the European Union’s 7th Framework Programme (contract no. 259348-ASSET). C.F. was supported by a long-term EMBO fellowship J.V.O. was supported by a Sapere Aude Research Leader grant from the Danish Council for Independent Research. Work at the NNF CPR is funded in part by a donation from the NNF (grant no. NNF14CC0001). Author contributions: S.L. and J.H.S. developed the Tet-inducible cellular system. A.E. and J.H.S. supervised S.L. K.P.T. generated the graphical representation of Fig. 2B and fig. S5A. H.H. performed the NetworKIN analysis shown in Fig. 4C. L.J.J. supervised K.P.T. and H.H., and edited the manuscript and did the analysis for Fig. 7E. A.-K.P. performed experiments shown in Figs. 3D and 6I and fig. S7B. D.B.B.-J. edited the figures and helped with the MS of the TrkA interactome samples. D.B.B.-J. and K.B.E. did the data analysis for Fig. 1D. K.B.E. generated and analyzed the data shown in the remaining figures under supervision of C.F. K.B.E. and J.V.O. performed all downstream MS data analysis. K.B.E., C.F., and J.V.O. designed the experiments, critically evaluated the results, and wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The raw MS data and associated tables have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository with the data set identifiers PXD001115 and PXD001478.
View Abstract

Stay Connected to Science Signaling

Navigate This Article