Phosphorylation of the Unique C-Terminal Tail of the Alpha Isoform of the Scaffold Protein SH2B1 Controls the Ability of SH2B1α To Enhance Nerve Growth Factor Function

ABSTRACT

The scaffold protein SH2B1, a major regulator of body weight, is recruited to the receptors of multiple cytokines and growth factors, including nerve growth factor (NGF). The β isoform but not the α isoform of SH2B1 greatly enhances NGF-dependent neurite outgrowth of PC12 cells. Here, we asked how the unique C-terminal tails of the α and β isoforms modulate SH2B1 function. We compared the actions of SH2B1α and SH2B1β to those of the N-terminal 631 amino acids shared by both isoforms. In contrast to the β tail, the α tail inhibited the ability of SH2B1 to both cycle through the nucleus and enhance NGF-mediated neurite outgrowth, gene expression, phosphorylation of Akt and phospholipase C-gamma (PLC-γ), and autophosphorylation of the NGF receptor TrkA. These functions were restored when Tyr753 in the α tail was mutated to phenylalanine. We provide evidence that TrkA phosphorylates Tyr753 in SH2B1α, as well as tyrosines 439 and 55 in both SH2B1α and SH2B1β. Finally, coexpression of SH2B1α but not SH2B1α with a mutation of Y to F at position 753 (Y753F) inhibited the ability of SH2B1β to enhance neurite outgrowth. These results suggest that the C-terminal tails of SH2B1 isoforms are key determinants of the cellular role of SH2B1. Furthermore, the function of SH2B1α is regulated by phosphorylation of the α tail.

INTRODUCTION

Alternative splicing provides avenues for cells to optimize their responses. Membrane receptors for growth factors and hormones are often alternatively spliced to optimize cellular signaling events for a single ligand, which enables distinct tissue- and cell-specific changes necessary for cellular homeostasis (1, 2). Ligand signaling can also be modulated by the presence of different isoforms of scaffold proteins and other signaling proteins. The scaffold protein SH2B1 (SH2-B, PSM) is recruited to multiple receptor tyrosine kinases (3–9). It exists in 4 distinct isoforms (α, β, γ, and δ) that share their first 631 amino acids (10). These shared amino acids contain dimerization, pleckstrin homology (PH), and SH2 domains, a nuclear localization and export sequence, and several proline-rich regions (Fig. 1A) (11). Each isoform contains a unique C-terminal tail of 39 to 125 amino acids. The isoforms arise from alternative splicing involving exon skipping and/or alternative 5′ splice donor sites (10, 12). The different isoforms differ in their tissue distribution. SH2B1β and SH2B1γ are ubiquitously expressed, whereas SH2B1α and SH2B1δ are expressed primarily in the brain (13). In vitro studies indicate that the different isoforms differ in their levels of efficacy in promoting a variety of functions, including mitogenesis in response to platelet-derived growth factor (PDGF), insulin, and insulin-like growth factor 1 in NIH 3T3 and 293T cells (14) and insulin-stimulated glucose and amino acid transport, glycogenesis, and lipogenesis in 3T3-L1 adipocytes (15).

FIG 1.

The C-terminal tail of SH2B1α regulates SH2B1's ability to enhance NGF-mediated neurite outgrowth and translocation to the nucleus. (A) Schematic of SH2B1α, SH2B1β, and SH2B1 1–631. DD, dimerization domain; NLS, nuclear localization sequence; NES, nuclear export sequence; PH, pleckstrin homology domain; SH2 domain; P, proline-rich domains; Y, tyrosine. The unique C-terminal tails are noted in green and red. Numbers indicate amino acids in rat and mouse sequences. (B) PC12 cells transiently expressing GFP or GFP-tagged SH2B1β, SH2B1α, or 1–631 were incubated with 25 ng/ml NGF. Percentages of GFP-expressing PC12 cells with neurite outgrowths at least twice the length of the cell body were determined on the indicated days. Results shown are mean values ± standard errors of the means (SEM) (n = 3). (C) Areas under the curve (AUCs) were determined from the data in panel B. *, P ≤ 0.05 compared to the value for cells expressing GFP alone (−). (D) 293T cells transiently expressing the indicated GFP-SH2B1 variant were treated with or without 20 nM leptomycin B (LMB) for 6 h. Live cells were imaged by confocal microscopy. Scale bar = 20 μm. (E, F) Fluorescence ratios of GFP-SH2B1 variants in the nucleus versus the cytoplasm (+LMB cells) (E) and in the plasma membrane versus the cytoplasm (−LMB cells) (F) from the experiments for which representative images are shown in panel D. The fluorescence ratios were determined from line scans using MetaVue. The locations of the line scans used for SH2B1β are noted by red lines. Results shown are mean values ± SEM (n = 47 to 80 cells from 3 or 4 independent experiments). *, P ≤ 0.05 compared to the results for GFP-SH2B1β.

Sh2b1^−/− mice are obese, hyperphagic, leptin resistant, and insulin resistant (16, 17). They also exhibit a decreased response to aortic banding-induced cardiac hypertrophy (18). In humans, genome-wide association studies (GWAS) have identified SH2B1 as a gene associated with body mass index (19, 20). Individuals with gene deletions within a region of chromosome 16p11.2 that includes the SH2B1 gene exhibit early-onset obesity and greater than expected insulin resistance (21, 22). More recently, nonsynonymous mutations in the SH2B1 gene have been identified by screening a cohort of individuals from the Genetics of Obesity Study (GOOS) who exhibited severe early-onset childhood obesity and greater than expected insulin resistance (13, 23).

Restoration of the SH2B1β isoform to Sh2b1^−/− mice using the neuron-targeted enolase promoter largely restores the lean phenotype (24), indicating that neuronal SH2B1 plays a key role in controlling energy balance. SH2B1 isoforms are important regulators of cellular signaling for receptors for multiple growth factors and cytokines that are critical for the differentiation and/or function of various subsets of nerves. These include receptors for nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), glial cell-derived neurotrophic factor (GDNF), fibroblast growth factor, leptin, and insulin (7, 8, 12, 25–29). Much of what we know about the role of SH2B1β in neurotrophic factor signaling has been determined in the context of signaling by the endogenous NGF receptor TrkA in PC12 cells. PC12 cells, derived from a rat pheochromocytoma, can be induced to differentiate into sympathetic neuron-like cells by NGF. NGF causes PC12 cells to produce neurite outgrowths and express neuron-specific genes (30–32). In response to NGF, SH2B1β is recruited, via its SH2 domain, to the activated NGF receptor TrkA, which in turn appears to phosphorylate SH2B1β (7, 9). Knockdown studies showed that SH2B1 isoforms are required for maximal NGF-stimulated neurite outgrowth and the expression of a subset of the NGF-sensitive genes (31, 33). Live-cell confocal microscopy of green fluorescent protein (GFP)-tagged SH2B1β (GFP-SH2B1β) revealed that SH2B1β localizes primarily to the plasma membrane and in the cytoplasm of PC12 and 293T cells (34). However, all SH2B1 isoforms have both a nuclear localization sequence (NLS) and nuclear export signal (NES). Mutating the NES or adding an inhibitor of nuclear export, leptomycin B (LMB), causes GFP-tagged SH2B1β to accumulate in the nucleus (33, 35); LMB also causes endogenous SH2B1 to accumulate in the nucleus. These results provide strong evidence that SH2B1β cycles through the nucleus. Blocking the ability of SH2B1β to cycle through the nucleus blocks the ability of SH2B1β to stimulate NGF-induced neurite outgrowth and gene expression, suggesting that cycling through the nucleus may be required for at least a subset of SH2B1 functions important for NGF responses.

The brain-specific expression of the α and δ isoforms of SH2B1 suggests that these two isoforms may play particularly important roles in neurons. Curiously, however, while we found that the β isoform of SH2B1 potently enhances neurite outgrowth of PC12 cells, the α isoform has little to no effect, nor does the α isoform appear to cycle through the nucleus (23). These findings suggest that the α and β isoforms have different functions and that the unique C-terminal tails of SH2B1α and/or SH2B1β may be potent regulators of SH2B1 function. For the current study, we hypothesized that relatively small differences in proteins due to differential splicing can dramatically regulate the function of those proteins. We first asked whether the unique α tail inhibits or the unique β tail promotes the actions of SH2B1. We then examined how the α and β tails regulate the function of the 631 N-terminal amino acids shared by all SH2B1 isoforms and which steps in NGF signaling are affected. Our results provide strong evidence that the unique C-terminal tails of the SH2B1 isoforms are key determinants of the cellular roles of the isoforms and that the function of SH2B1α is regulated by phosphorylation/dephosphorylation of a tyrosine in the C-terminal tail of SH2B1α.

RESULTS

The unique C-terminal tail of the α isoform of SH2B1 inhibits the ability of SH2B1 to enhance NGF-mediated neurite outgrowth and affects its subcellular localization.

Previously, we showed that, unlike overexpression of SH2B1β, overexpression of SH2B1α does not enhance the ability of NGF to promote neurite outgrowth in PC12 cells (23). To determine whether the unique β tail enables SH2B1β to enhance neurite outgrowth or if the unique α tail prevents SH2B1α from enhancing neurite outgrowth, we compared the function of SH2B1α and SH2B1β to that of the N-terminal 631 amino acids of SH2B1 (SH2B1 1–631) shared by both isoforms (Fig. 1A). We transiently expressed GFP or GFP-tagged SH2B1α, SH2B1β, or SH2B1 1–631 in PC12 cells and determined the percentages of transfected cells exhibiting neurites after 1, 2, and 3 days of NGF treatment (Fig. 1B and C). In contrast to SH2B1α, which had no effect on neurite outgrowth, SH2B1 1–631 enhanced NGF-induced neurite outgrowth to an extent similar to that seen with SH2B1β, suggesting that the unique C-terminal tail of SH2B1α prevents the N-terminal 631 amino acids of SH2B1 from enhancing neurite outgrowth.

Mutating the NLS in SH2B1β abrogates the ability of SH2B1β to both cycle through the nucleus and enhance NGF-mediated neurite outgrowth (33). These results suggest that SH2B1 might need to enter the nucleus to promote neurite outgrowth. Because the α tail prevents SH2B1α from enhancing NGF-induced neurite outgrowth, we next examined whether the α tail also prevents SH2B1α from cycling through the nucleus. Using live-cell confocal microscopy, we visualized and quantified the subcellular localization of transiently expressed GFP-tagged SH2B1α, SH2B1β, and SH2B1 1–631 in 293T cells before and after treatment with the nuclear export inhibitor LMB (Fig. 1D). In the steady state (absence of LMB), SH2B1 1–631, like SH2B1α and SH2B1β, localizes primarily to the plasma membrane and in the cytoplasm (Fig. 1D). In the presence of LMB, SH2B1 1–631, like SH2B1β, accumulates in the nucleus to a much greater extent than SH2B1α (Fig. 1D and E), consistent with the C-terminal tail of SH2B1α preventing SH2B1α from entering the nucleus. Confocal microscopy also revealed that SH2B1 1–631, like SH2B1β, localizes to the plasma membrane to a greater extent than SH2B1α (Fig. 1D and F). These results suggest that the unique C-terminal tail of SH2B1α interferes with the ability of the region of SH2B1 shared by all SH2B1 isoforms to localize to the plasma membrane.

Mutating Tyr753 in the unique C-terminal tail of SH2B1α enables SH2B1α to enhance NGF-mediated neurite outgrowth and cycle through the nucleus.

The results shown in Fig. 1 raise the possibility that some function of the C-terminal tail (e.g., protein-protein interaction) prevents SH2B1α from enhancing neurite outgrowth and cycling through the nucleus. The C-terminal end of the α tail (S₇₅₄FV) contains a consensus recognition motif for PDZ domains (10, 36, 37) that is not present in the other SH2B1 isoforms. To gain insight into whether binding of the α tail to a PDZ domain-containing protein might decrease the ability of SH2B1α to enhance neurite outgrowth, enter the nucleus, and/or localize to the plasma membrane, we mutated key residues in this potential PDZ binding domain. SH2B1α with Ser754 mutated to Ala (S754A) or Val756 to Asp (V756D) did not enhance NGF-mediated neurite outgrowth to a statistically significant extent (Fig. 2) or accumulate in the nuclei of LMB-treated cells (Fig. 3A and B). However, SH2B1α with Tyr753 mutated to Phe (Y753F) enhanced neurite outgrowth to an extent that approached that of SH2B1β (Fig. 2) and accumulated in the nucleus to a significantly greater extent than SH2B1α, although to a lesser extent than SH2B1β (Fig. 3A and B). In contrast, SH2B1α with Tyr753 mutated to the phosphomimetic glutamic acid (Y753E) did not enhance NGF-mediated neurite outgrowth or accumulate in the nucleus. None of the mutations altered the ability of SH2B1α to localize to the plasma membrane (Fig. 3A and C). These data suggest that the C-terminal tail of SH2B1α does not alter the activity of SH2B1α by binding directly to a PDZ domain-containing protein. Rather, phosphorylation of Tyr753 may prevent SH2B1α from enhancing NGF-induced neurite outgrowth and entering the nucleus.

FIG 2.

Mutating Tyr753 to phenylalanine in its C-terminal tail enables SH2B1α to enhance NGF-mediated neurite outgrowth. (A) Live PC12 cells transiently expressing GFP or the indicated GFP-SH2B1 variant were treated with 25 ng/ml NGF. The percentages of GFP-expressing cells with neurite outgrowths at least twice the length of the cell body were determined on the indicated days. Results shown are mean values ± SEM (n = 3). (B) AUCs were determined from the data shown in panel A. *, P ≤ 0.05 compared to the results for control cells expressing GFP alone (−).

FIG 3.

Tyr753 in SH2B1α regulates the ability of SH2B1α to translocate to the nucleus. (A) 293T cells expressing GFP-tagged SH2B1β, SH2B1α, or the indicated SH2B1α mutant were treated with or without 20 nM leptomycin B (LMB) for 6 h. Live cells were imaged by confocal microscopy. Scale bar = 20 μm. (B, C) Fluorescence ratios of GFP-SH2B1 variants in the nuclear region versus the cytoplasm (+LMB) (B) and in the plasma membrane versus the cytoplasm (−LMB) (C) from the experiments for which representative images are shown in panel A. The fluorescence ratios were determined from line scans using MetaVue. Results shown are mean values ± SEM (n = 16 to 55 cells from 2 to 4 independent experiments). #, P ≤ 0.05 by Student t test comparing the results for GFP-SH2B1α variants to the results for GFP-SH2B1β; *, P ≤ 0.05 by one-way ANOVA with Dunnett's posttest comparing the results for GFP-SH2B1α variants to the results for GFP-SH2B1α.

The C-terminal tail of SH2B1α prevents SH2B1α from enhancing the expression of NGF-responsive genes.

Previous studies showed that SH2B1β enhances the expression of a subset of NGF-responsive genes, including Plaur, Mmp3, Mmp10, and FosL1 (31). Plaur is an early response gene that encodes urokinase plasminogen activator receptor (UPAR). UPAR is located on the plasma membrane, where it binds to and activates plasminogen activator (uPA). UPAR has been shown to be necessary for NGF-mediated neurite outgrowth in PC12 cells (38, 39). It is also thought to be an essential player for neuronal differentiation in vivo (40). Matrix metalloproteases, such as MMP3 and MMP10, are thought to be critical for the degradation of the extracellular matrix that is required for neuronal axon and dendrite extension (41). The Fos-like antigen 1 (FosL1, Fra1) gene is a member of the Fos family of immediate early-response genes (42). As seen with the neurite outgrowth assays, overexpression of either SH2B1β or SH2B1 1–631 significantly enhanced NGF-mediated expression of Plaur, Mmp10, and FosL1, whereas SH2B1α had no effect (Fig. 4). Mutating Tyr753 to Phe enabled SH2B1α to enhance the expression of Plaur, Mmp10, and FosL1. A similar trend was seen with Mmp3, although the differences did not achieve statistical significance. These data provide further evidence that the α tail negatively regulates the region of SH2B1α common to all SH2B1 isoforms by a mechanism requiring phosphorylation of Tyr753, thereby suppressing the ability of SH2B1α to enhance NGF-mediated functions.

FIG 4.

Deletion of the isoform-specific C-terminal tail of SH2B1α or mutation of Tyr753 in the α tail enables SH2B1α to enhance NGF-mediated gene expression. PC12 cells stably overexpressing GFP alone (−) or the indicated GFP-tagged SH2B1 variant were treated with 100 ng/ml NGF for 6 h. mRNA was collected, and relative levels of Plaur (A), Mmp10 (B), FosL1 (C), and Mmp3 (D) mRNA were determined by qPCR. Relative mRNA abundance was normalized to the geometric mean of the levels of Hprt and Cyclophilin A mRNA. Data were normalized to values obtained for GFP-SH2B1β-expressing cells treated with NGF. mRNA levels for untreated samples are plotted but were usually too low to be visible on the graph. Results shown are mean values ± SEM for 3 independent experiments. *, P ≤ 0.05 compared to the results for GFP cells treated with NGF.

The C-terminal tail of SH2B1α inhibits the ability of the region of SH2B1 shared by all isoforms to enhance NGF signaling pathways; mutation of Tyr753 reverses that inhibition.

To gain insight into the mechanism by which the α tail inhibits the ability of SH2B1α to enhance NGF-induced neurite outgrowth and gene expression, we next examined the effect of the α tail on the ability of the shared region of SH2B1 to affect NGF signaling pathways. NGF-mediated neurite outgrowth in PC12 cells is initiated by NGF binding to and activating the receptor tyrosine kinase TrkA (43). The resulting autophosphorylated Tyr (pTyr) in TrkA recruits signaling proteins via their SH2 and/or PTB domains. Multiple signaling pathways are activated, including those leading to rapid and prolonged activation of extracellular signal-regulated kinase 1 (Erk1) and Erk2, Akt, and phospholipase C-gamma (PLC-γ) (43). Our laboratory and/or others have previously shown that overexpression of SH2B1β in PC12 cells enhances and prolongs NGF-induced phosphorylation of activating amino acids in Akt (Ser473) (44, 45) and PLC-γ (Tyr783) (45). Modest enhancement of the phosphorylation of activating amino acids (Thr202 and Tyr204) in Erk1 and -2 has been observed in some studies (7, 33, 45). Consistent with these previous results, overexpression of SH2B1β in PC12 cells enhanced and prolonged the phosphorylation of Tyr783 in PLC-γ (Fig. 5A, top panel, and B) and Ser473 in Akt (Fig. 5A, third panel, and C) but did not have a statistically significant effect on the phosphorylation of Thr202/Tyr204 in Erk1 and Erk2 (Fig. 5A, fifth panel, and D). In contrast to SH2B1β, SH2B1α had no effect on PLC-γ phosphorylation (Fig. 5A, top panel, and B) and actually modestly inhibited NGF's stimulation of phosphorylation of Akt, Erk1, and Erk2 (Fig. 5A, third and fifth panels, and C and D). Overexpression of SH2B1 1–631 showed enhanced and/or prolonged phosphorylation of PLC-γ, Akt, Erk1, and Erk2 (Fig. 5), indicating that the α tail and, in the case of Erk1 and Erk2, the β tail inhibit the ability of the region of SH2B1 shared by all isoforms to enhance NGF signaling via these proteins. Mutating Tyr753 to Phe enables SH2B1α, like SH2B1β and SH2B1 1–631, to enhance NGF-induced phosphorylation of PLC-γ and Akt. The results for the Y753F mutant suggest that phosphorylation of Tyr753 prevents SH2B1α from enhancing NGF signaling pathways. The resulting inability of SH2B1α to enhance NGF signaling pathways could contribute to the inability of SH2B1α to enhance NGF-induced neurite outgrowth and gene expression.

FIG 5.

The isoform-specific C-terminal tails of SH2B1 modulate NGF-dependent phosphorylation of NGF signaling proteins. (A) Serum-starved (15 h) PC12 cells stably overexpressing GFP alone (−) or a GFP-tagged SH2B1 variant as indicated below the gels were treated with 25 ng/ml NGF for 10 or 60 min. Proteins in cell lysates were immunoblotted with the indicated antibodies. Cell lysates from all cell lines were run on the same gel. (B to D) The relative signal intensities of the indicated phosphorylated proteins were normalized to the signal intensity for that protein. AUCs were determined from the normalized signals and further normalized to the GFP signal at 10 min. Results shown are mean values ± SEM (n = 3 independent experiments). *, P ≤ 0.05 compared to the results for GFP cells.

When proteins were separated using gradient gels designed to visualize different states of phosphorylation, NGF caused an upward shift and substantial broadening in the migration of GFP-tagged SH2B1β, SH2B1α, and SH2B1α Y753F (Fig. 5A, bottom panel). The upward shift in SH2B1β has been shown to be due primarily to serine/threonine phosphorylation (46). Phosphorylation of some of these serines in SH2B1β (Ser161 and Ser165) appears to be important for SH2B1β to leave the plasma membrane, enter the nucleus, and enhance both NGF-induced neurite outgrowth and gene expression (34). Surprisingly, SH2B1 1–631 showed a much more modest change in migration in response to NGF than did SH2B1β, SH2B1α, or SH2B1α Y753F. These data are consistent with the C-terminal tails of both SH2B1α and SH2B1β increasing the serine/threonine phosphorylation of SH2B1 in response to NGF.

The C-terminal tail of SH2B1α inhibits the ability of the region of SH2B1 shared by all isoforms to enhance NGF-induced autophosphorylation of TrkA; mutation of Tyr753 reverses that inhibition.

Some studies have suggested that SH2B1 isoforms, recruited to TrkA via their shared SH2 domain (7), activate TrkA as monitored by NGF-induced phosphorylation of Tyr490 in TrkA (47). Differences in the ability of the different isoforms of SH2B1 to enhance TrkA activity and/or the phosphorylation of specific tyrosines in TrkA would be expected to alter their ability to enhance NGF signaling pathways. We used phosphospecific antibodies and our stably transfected PC12 cell lines to examine the impact of SH2B1β, SH2B1α, SH2B1α Y753F, and SH2B1 1–631 on the ability of NGF to stimulate phosphorylation of Tyr674/Tyr675 and Tyr490 (Fig. 6). Tyrosines 674 and 675 are the activating tyrosines of TrkA; their phosphorylation correlates with TrkA activity (48). Phospho-Tyr490 recruits both Shc and FRS2, both of which are thought to lead to activation of Erk1 and Erk2 (49, 50). SH2B1β modestly enhanced the phosphorylation of both Tyr674/Tyr675 and Tyr490 (Fig. 6). SH2B1 1–631 was more effective than SH2B1β (Fig. 6A to C), suggesting that the β tail has a modest inhibitory effect on the ability of the region of SH2B1 shared by all isoforms (amino acids 1 to 631) to enhance TrkA activity. In contrast, overexpression of SH2B1α had no effect on tyrosyl phosphorylation of either Tyr490 or Tyr674/Tyr675 in TrkA unless Tyr753 in SH2B1α was mutated to Phe (Fig. 6D to F), in which case phosphorylation of TrkA was stimulated to a similar extent as in the presence of SH2B1β. These data are consistent with the α tail suppressing the ability of SH2B1α to enhance NGF-stimulated TrkA activity and subsequent phosphorylation of Tyr490, unless Tyr753 in the α tail is mutated to Phe.

FIG 6.

The isoform-specific C-terminal tails of SH2B1 modulate NGF-mediated TrkA phosphorylation. (A, D) Serum-starved (15 h) PC12 cells stably overexpressing GFP alone (−) or a GFP-tagged SH2B1 variant as indicated below the gels were treated with 100 ng/ml NGF for 10 and 60 min. Proteins in cell lysates were immunoblotted with the indicated antibodies. GFP and GFP-SH2B1β lanes for panels A and D contain aliquots of the same cell lysates. Immunoblots are representative of 3 independent experiments. n.s, nonspecific. (B and C, E and F) Relative signal intensities of the indicated phosphorylated proteins in panels A and D were normalized to the signal intensity for that protein. AUCs were determined from the normalized signals and further normalized to the GFP signal at 10 min. Results shown are mean values ± SEM (n = 3 independent experiments). *, P ≤ 0.05 compared to the results for GFP cells.

The inability of SH2B1α to enhance TrkA responses could result from an inability to be recruited to activated TrkA. Using a 293T overexpression system, we found that SH2B1α, SH2B1β, SH2B1α Y753F, and SH2B1 1–631 coprecipitate with constitutively active TrkA (Fig. 7A). Consistent with SH2B1α and -β both being recruited via their SH2 domains to phosphorylated tyrosines in TrkA, neither coprecipitated with kinase-dead, unphosphorylated TrkA (Fig. 7A). These results suggest that the decreased ability of SH2B1α to enhance TrkA responses is not due to a decreased ability of SH2B1α to bind to TrkA. Analysis of the protein levels of SH2B1β, SH2B1α, and SH2B1α Y753F following NGF treatment for 0 to 4 h (Fig. 7B) suggests that the inability of GFP-SH2B1α to enhance TrkA responses is not due to decreased stability of GFP-SH2B1α relative to the stability of SH2B1β or SH2B1α Y753F.

FIG 7.

SH2B1β, SH2B1α, SH2B1α Y753F, and SH2B1 1–631 bind equally to TrkA and exhibit similar levels of protein expression over a 4-h time course. (A) 293T cells transiently coexpressing a Myc-tagged SH2B1 variant and either mCherry alone (−), wild-type mCherry-TrkA (WT), or kinase-dead mCherry-TrkA (KD) as indicated below the gels were lysed, and mCherry-TrkA was immunoprecipitated with anti-mCherry antibody. Cell lysates and anti-mCherry antibody immunoprecipitates (IP) were immunoblotted (IB) with anti-Myc antibody to detect coprecipitating Myc-SH2B1 variants. The immunoblots are representative of 3 independent experiments. (B) Serum-starved (15 h) PC12 cells stably overexpressing GFP alone or a GFP-tagged SH2B1 variant as indicated below the gels were treated with 25 ng/ml NGF for the indicated times. Proteins in cell lysates were immunoblotted with the indicated antibodies. The immunoblots are representative of 3 independent experiments.

TrkA phosphorylates multiple tyrosines in SH2B1α and SH2B1β, including Tyr753 in SH2B1α.

Our results thus far suggest that phosphorylation of Tyr753 in the α tail of SH2B1α likely inhibits the function of SH2B1α. We have shown previously that SH2B1β is phosphorylated on tyrosines in response to NGF (7), presumably by the NGF receptor TrkA. We therefore hypothesized that Tyr753 is a substrate of TrkA. To gain insight into which tyrosines in SH2B1α and SH2B1β are phosphorylated by TrkA, we mutated each individual tyrosine in Myc-tagged SH2B1α or SH2B1β to a phenylalanine. The Myc tag was used because it does not contain any tyrosines. We then transiently coexpressed each SH2B1 with TrkA in 293T cells, immunoprecipitated each Myc-SH2B1 using anti-Myc antibody, and immunoblotted with antibodies specific to phosphorylated tyrosines. The results shown in Fig. 8A reveal that both SH2B1α and SH2B1β are highly phosphorylated on tyrosines when coexpressed with TrkA. No phosphorylation of SH2B1α or SH2B1β was detected in the absence of TrkA. Mutating either Tyr55 or Tyr439 in SH2B1β, but not any other tyrosine, significantly reduced tyrosyl phosphorylation of SH2B1β (Fig. 8B, top panel). Mutating both Tyr55 and Tyr439 abrogated tyrosyl phosphorylation of SH2B1β (Fig. 8C, top panel). Immunoblotting cell lysates with antibody highly specific to pTyr439 in SH2B1 (51) confirmed that Tyr439 in SH2B1β is phosphorylated when TrkA and SH2B1β are coexpressed (Fig. 8C, third panel). Phosphorylation of Tyr439 was not appreciably altered by mutating Tyr55. These results suggest that Tyr55 and Tyr439 are the two main substrates of TrkA in SH2B1β. In the case of SH2B1α, mutating Tyr753 dramatically reduced the tyrosyl phosphorylation of SH2B1α (Fig. 8A). Mutation of the other individual tyrosines did not result in a reproducible detectable reduction in the overall tyrosyl phosphorylation of SH2B1α, suggesting that Tyr753 is the primary substrate of TrkA. However, SH2B1 Y753F retained some tyrosyl phosphorylation, indicating that TrkA phosphorylates additional tyrosines in SH2B1α. Since TrkA appears to phosphorylate Tyr55 and Tyr439 in SH2B1β (Fig. 8B and C), we tested whether mutating Tyr753, Tyr55, and Tyr439 in different combinations would further reduce TrkA phosphorylation of SH2B1α (Fig. 8D, top panel). When all three were mutated, we saw an almost total absence of tyrosyl phosphorylation of SH2B1α, suggesting that SH2B1α is phosphorylated on Tyr55 and Tyr439 in addition to Tyr753. Immunoblotting cell lysates with anti-phospho-Tyr439 SH2B1 antibody confirmed that coexpression of SH2B1α with TrkA results in phosphorylation of Tyr439 (Fig. 8D, third panel). The levels of phosphorylation of Tyr439 were not significantly affected by mutating Tyr753 and Tyr55, alone or together. Collectively, these data are consistent with Tyr753 in the C-terminal tail of SH2B1α being a major substrate for TrkA. Additionally, Tyr439 and, most likely, Tyr55, both in the N-terminal region shared by the 4 isoforms of SH2B1, appear to be substrates of TrkA in both SH2B1α and SH2B1β.

FIG 8.

TrkA phosphorylates Tyr753 in SH2B1α, as well as Tyr439 and, potentially, Tyr55 in SH2B1β and SH2B1α. 293T cells transiently coexpressing a Myc-tagged SH2B1 variant and either TrkA (B) or mCherry-TrkA (A, C, and D) as indicated below the gels were lysed, and proteins were immunoprecipitated with anti-Myc antibody. Cell lysates and anti-Myc antibody immunoprecipitates (IP) were immunoblotted (IB) with the indicated antibodies. The blots shown are representative of 2 (A, D) or 3 (B, C) independent experiments.

SH2B1α inhibits SH2B1β-induced neurite outgrowth of PC12 cells.

All SH2B1 isoforms contain a common SH2 domain (10), and thus, all isoforms would be expected to compete for binding to TrkA in response to NGF. Furthermore, SH2B1α and SH2B1β are both expressed in all areas of the brain tested (13). These observations, combined with our finding that, in contrast to SH2B1β and SH2B1α Y753F, SH2B1α is unable to promote NGF-induced tyrosyl phosphorylation of TrkA and NGF signaling pathways, led us to ask whether coexpressing SH2B1α in the same cells as SH2B1β or SH2B1α Y753F would inhibit the ability of SH2B1β or SH2B1α Y753F, respectively, to promote neurite outgrowth. PC12 cells stably expressing GFP or GFP-tagged SH2B1β, SH2B1α, or SH2B1α Y753F were transiently transfected with the red fluorescent protein (RFP) tdTomato (tdT) or tdT-tagged SH2B1β, SH2B1α, or SH2B1α Y753F. Neurites were assessed in cells expressing both the GFP and tdT tags after 2 days of NGF treatment (Fig. 9A). In “control” cells stably expressing GFP, transient overexpression of the various tdT-tagged forms of SH2B1 resulted in a profile of neurite outgrowth similar to that obtained when the equivalent GFP-tagged SH2B1 was transiently expressed in PC12 cells (Fig. 1B and 2A). In cells stably overexpressing GFP-tagged SH2B1β or SH2B1α Y753F, transient expression of tdT-SH2B1α inhibited neurite outgrowth by about 50%, whereas transient expression of tdT, tdT-SH2B1β, and tdT-SH2B1α Y753F had minimal effect. In cells stably expressing GFP-SH2B1α, overexpression of tdT-SH2B1β and tdT-SH2B1α Y753F still enhanced neurite outgrowth but only to about half the level observed in the cells stably expressing GFP. These results could not be attributed to a reduction in the transient expression of tdT-SH2B1α versus that of SH2B1β or tdT-SH2B1α Y753F (Fig. 9B). The results shown in Fig. 9 are consistent with SH2B1α competing with SH2B1β (and SH2B1α Y753F) to bind to TrkA via their shared SH2 domains and thereby inhibiting the ability of SH2B1β (and SH2B1α Y753F) to enhance neurite outgrowth.

FIG 9.

SH2B1α inhibits the ability of SH2B1β to enhance NGF-mediated neurite outgrowth. (A) Live PC12 cells stably expressing GFP alone (−) or the indicated GFP-SH2B1 variant were transiently transfected with tdTomato (TdT) alone (−) or the indicated tdT-tagged SH2B1 variant and incubated with 25 ng/ml NGF for 2 days. The percentages of cells expressing both TdTomato and GFP with neurite outgrowths at least twice the length of the cell body length were determined on day 2. Mean values and ranges from 2 independent experiments are shown. *, P ≤ 0.05 compared to the results for cells stably expressing GFP and transiently expressing TdT-SH2B1β. (B) Cell lysates from a parallel experiment to the one whose results are shown in panel A were immunoblotted with the indicated antibodies.

Because SH2B1α and SH2B1β can dimerize (23) and multiple lines of evidence suggest that SH2B1β must cycle through the nucleus to enhance NGF-induced neurite outgrowth and gene expression (33, 35), we also looked to see whether coexpressing SH2B1α impairs SH2B1β cycling through the nucleus. In preliminary experiments, we found that coexpressing RFP mCherry-tagged SH2B1α (mCherry-SH2B1α) and GFP-SH2B1β in PC12 cells subsequently treated with NGF did not impact the ability of SH2B1β to accumulate in the nucleus in the presence of LMB (data not shown). This finding suggests that SH2B1α does not impair the ability of SH2B1β to enhance neurite outgrowth as a consequence of preventing it from cycling through the nucleus.

DISCUSSION

It is becoming increasingly evident that alternative splicing plays an important role in increasing the diversity of cellular responses both between and within different tissues. In this study, we show that two different isoforms of the scaffold protein SH2B1, SH2B1α and SH2B1β, which differ only in their C-terminal amino acids, representing 15 and 6% of the total protein sequence, respectively, have profoundly different effects on NGF signaling. SH2B1β enhances NGF-dependent phosphorylation of TrkA, Akt, and PLC-γ, gene expression, and neurite outgrowth of PC12 cells, whereas SH2B1α does not. We further provide strong evidence that the α tail inhibits the ability of the N-terminal 631 amino acids shared by all isoforms of SH2B1 to enhance these NGF functions, rather than the β tail enabling them. In fact, the β tail may also inhibit, to a lesser extent, certain actions (e.g., phosphorylation of Erk1 and Erk2 and TrkA activity) of the N-terminal region of SH2B1 shared by all isoforms. The latter possibility is supported by findings that SH2B1 1–631 is as effective, if not more so, than SH2B1β in all of these activities. We identify potential TrkA substrate sites within SH2B1α and β and provide strong evidence that phosphorylation of a single amino acid (Tyr753) in the C-terminal tail of SH2B1α is responsible for the inhibition of the functions of the shared region of SH2B1. Finally, we provide evidence that when coexpressed, SH2B1α inhibits the ability of SH2B1β to enhance neurite outgrowth.

Several lines of evidence support our hypothesis that phosphorylation of a single amino acid (Tyr753) in the C-terminal tail of SH2B1α is critical for the α tail's inhibition of the functions of the shared region of SH2B1. First, mutating Tyr753 to phenylalanine enables SH2B1α to act like SH2B1β and enhance multiple functions of NGF. In contrast, mutating Tyr753 to a phosphomimetic amino acid (glutamic acid) yields a phenotype similar to that of SH2B1α, suggesting that a negative charge in position 753 (e.g., from phosphorylation or from glutamic acid), rather than the presence of a tyrosine per se, inhibits the ability of SH2B1α to enhance NGF functions. Second, TrkA-dependent tyrosyl phosphorylation of SH2B1α is greatly reduced when Tyr753 is mutated to Phe, providing strong evidence that TrkA is capable of phosphorylating Tyr753. Thus, it seems likely that upon recruitment of SH2B1α to activated TrkA, Tyr753 becomes phosphorylated, which would essentially turn SH2B1α off until Tyr753 is dephosphorylated. We speculate that other tyrosine kinases will also be found to phosphorylate Tyr753. Such cross talk would enable other ligands that activate tyrosine kinases to negatively modulate NGF functions by reversibly turning off SH2B1α. One can also envision phosphorylation of Tyr753 being a switch that turns off some activities and turns on other activities or enables SH2B1α to enhance the actions of one tyrosine kinase and inhibit the actions of another. Indeed, a turned-on SH2B1α has been detected in other cell lines with other ligands. In metabolic studies of the insulin response in 3T3-L1 adipocytes, SH2B1α enhanced insulin-mediated glucose and amino acid transport and lipogenesis, and in contrast to the results we detected with NGF in PC12 cells, it enhanced Akt activity to a level similar to that seen with SH2B1β (15). In NIH 3T3 and 293T cells, SH2B1α and SH2B1β both enhanced the ability of insulin, insulin-like growth factor 1, and PDGF to promote mitogenesis (14). Consistent with Tyr753 playing an important role in regulating the function of SH2B1α, the C-terminal tail of SH2B1α, including the Tyr753-containing motif RAINNQY₇₅₃SFV, is highly conserved within mammals. In addition, comparable tyrosines are found in the C termini of SH2B1 family members SH2B2/APS (Tyr618 in rodents and humans) and SH2B3/Lnk (Tyr536 in rodents and Tyr572 in humans), as well as in a Drosophila melanogaster SH2B ortholog (GenBank accession number AAF56523). Tyr618 in SH2B2/APS has been shown to be phosphorylated by the insulin receptor (52, 53); mass spectrometry has identified phospho-Tyr618 in SH2B2 and phospho-Tyr536/Tyr572 in SH2B3 in a variety of tissues (54).

So how might phosphorylation of Tyr753 inhibit the actions of SH2B1α? Tyr753 resides in a potential PDZ binding motif (55). This raises the possibility that phosphorylation of Tyr753 promotes the binding of the C-terminal tail of SH2B1α to a PDZ domain-containing protein that inhibits the actions of SH2B1α or inhibits the binding of a PDZ domain-containing protein that enhances the actions of SH2B1α. Arguing against these hypotheses is our finding that mutating critical residues expected to disrupt the putative PDZ binding domain does not alter the ability of SH2B1α to enhance NGF-induced neurite outgrowth. In SH2B2/APS, the Tyr analogous to Tyr753 binds to c-Cbl (56, 57). The binding of a member of the Cbl family of E3 ligases to phospho-Tyr753 in SH2B1α could promote degradation of SH2B1α. Arguing against this, we did not observe an NGF-dependent decrease in the levels of SH2B1α in our stable PC12 cell line. Additionally, mutating Tyr753 to glutamic acid, a mutation that would be expected to prevent recruitment of any SH2 domain-containing protein, did not enable SH2B1α to promote neurite outgrowth, suggesting that phospho-Tyr753 may not be inhibiting the actions of SH2B1α by recruiting an SH2 domain-containing protein like Cbl. The analogous tyrosine in Lnk (Tyr536) has been proposed, when phosphorylated, to recruit PLC-γ1, Grb-2, and phosphatidylinositol 3-kinase (PI3K) (58). However, it is hard to imagine how recruiting signaling proteins that are thought to be important for NGF-induced neurite outgrowth would prevent, rather than enhance, the ability of SH2B1α to augment NGF-dependent neurite outgrowth.

Phosphorylation of Tyr753 could, through a conformational change, cause or stabilize a conformational change in the α tail that blocks the NLS. Because the NLS is required for both binding of SH2B1β to the plasma membrane and entry into the nucleus (34), such interference could help explain why SH2B1α both appears to bind to the plasma membrane to a lesser extent than SH2B1β and shows a greatly reduced ability to cycle through the nucleus. Decreased localization to the plasma membrane could impair the recruitment of SH2B1α to TrkA, and decreased availability of the NLS to bind to importins could decrease the cycling of SH2B1α through the nucleus. Both of these are thought to be important for the ability of SH2B1β to enhance neurite outgrowth and gene expression (7, 31, 33). In further support of this, mutating Tyr753 to phenylalanine modestly increased the ability of SH2B1α to cycle through the nucleus, although it did not affect its ability to localize to the plasma membrane. Further experiments are required to determine whether phosphorylation or dephosphorylation of Tyr753 regulates the binding of the C-terminal tail of SH2B1α to specific proteins that interfere with SH2B1α function and/or causes a conformational change that interferes with SH2B1α function. While we think it less likely, we also cannot rule out the possibility that mutating Tyr753 causes a conformational change independent of phosphorylation that relieves the inhibitory action of the α tail on the functions of the shared region of SH2B1.

Previously, we showed that SH2B1β is phosphorylated on Tyr439 and Tyr494 by JAK2, both in the context of a 293T overexpression system (59) and in response to growth hormone (GH) stimulation of 3T3-F442A cells (51). Here, we provide evidence that TrkA, like JAK2, is able to phosphorylate SH2B1β on Tyr439. We also provide evidence that in the presence of TrkA, Tyr55 in SH2B1β is phosphorylated, rather than Tyr494. Tyr55 and Tyr439 in SH2B1α also appear to be TrkA substrates, although they appear to be phosphorylated to a lesser extent than Tyr753. Both Tyr55 and Tyr439 were implicated as substrates of SH2B1α and SH2B1β because of reduced anti-pTyr antibody signals when they were individually mutated and a further reduction when both were mutated together. Tyr439 was confirmed as a substrate using an anti-phospho-Tyr439 SH2B1 antibody. Tyr55 was mentioned as a substrate of TrkA without documentation by Qian et al. (9). Thus, TrkA and JAK2 share at least one substrate site (Tyr439) in SH2B1α and SH2B1β but appear to each have one kinase-specific site (Tyr55 for TrkA versus Tyr494 for JAK2). The shared Tyr439 site would allow for some SH2B1 functions to be shared by ligands that activate TrkA and JAK2, while the kinase-specific phosphorylation sites would allow for some functions to be specific to ligands that activate JAK2 versus TrkA. The shared Tyr439 site, along with the JAK2-specific Tyr494 site, has been implicated in SH2B1β enhancement of GH-stimulated membrane ruffling in 3T3-F442A fibroblasts (59) and GH-stimulated motility of RAW 264.7 macrophages (51). Thus, it is tempting to speculate that in response to NGF, phospho-Tyr439 in SH2B1β and SH2B1α recruits a protein important for neuronal migration or the cytoskeletal rearrangements that occur during neuronal differentiation. The function of phospho-Tyr55 remains to be determined.

Because of the critical role that phosphoserines play in regulating the function of SH2B1, we were interested to see that NGF stimulates both an upward shift in migration and a broadening of the SH2B1α band in immunoblots, similar to that seen previously with SH2B1β (46). This NGF-dependent change in SH2B1 migration was greatly reduced for SH2B1 1–631. It will be intriguing to find out whether the C-terminal tails of the different isoforms affect which Ser/Thr are phosphorylated, how phosphorylation of the different Ser/Thr affects the function of the different isoforms, and what kinases and phosphatases phosphorylate and dephosphorylate, respectively, the various Ser/Thr.

Despite significant effort, it is still not totally clear how SH2B1β enhances NGF-induced neurite outgrowth. Previous work (7, 9) and/or work presented here indicate that SH2B1 isoforms are recruited via their shared SH2 domains to phosphorylated Tyr674/Tyr675 in TrkA and are phosphorylated by TrkA on multiple tyrosines. Our work using phosphospecific antibodies to various tyrosines in TrkA indicates that the binding of SH2B1β to TrkA modestly increases the activity of TrkA, as monitored by increased phosphorylation of the activating Tyr674/Tyr675 in TrkA. Thus, the modest activation of TrkA by SH2B1β may result, at least in part, from stabilizing the activated state of TrkA, similar to how SH2B1β has been hypothesized to stabilize the active state of JAK2 (60). A more active TrkA would be expected to increase phosphorylation of other tyrosines in TrkA, allowing TrkA to recruit additional signaling molecules for longer periods of time. Consistent with this, SH2B1β increased the phosphorylation of Tyr490. Phospho-Tyr490 has been shown to recruit Shc and FRS-2 proteins, both of which are implicated in activating Erk1 and Erk2 (61–64). The finding that SH2B1β is more potent at stimulating and prolonging the phosphorylation of PLC-γ and Akt than of Erk1 and Erk2 suggests that SH2B1β has a greater effect on the phosphorylation of the TrkA tyrosines that lead to activation of PLC-γ and Akt than it has on the phosphorylation of Tyr490. Alternatively, neither SH2B1 nor pTyr490 is rate limiting for the pathways by which NGF activates Erk1 and Erk2 under our experimental conditions. In contrast to SH2B1β and SH2B1 1–631, SH2B1α did not stimulate the tyrosyl phosphorylation of any of the tested Tyr in TrkA, nor did it stimulate the phosphorylation of Akt, PLC-γ, Erk1, or Erk2. However, mutation of Tyr753 to Phe enabled SH2B1α to enhance phosphorylation of TrkA, as well as PLC-γ and Akt. In contrast to the relatively small changes in phosphorylation associated with proteins in the NGF signaling pathways tested, SH2B1β, SH2B1 1–631, and SH2B1α Y753F robustly stimulated NGF-dependent neurite outgrowth and expression of Plaur, Mmp10, and FosL1, whereas SH2B1α had no effect. The discrepancy between the robust SH2B1β- and SH2B1α Y753F-induced enhancement of NGF-dependent neurite outgrowth and gene expression and the more modest increases in phosphorylation of TrkA and downstream signaling pathways suggest that (i) there are some yet-to-be-determined signaling pathways that are major contributors to the ability of SH2B1 to regulate neurite outgrowth and gene expression, (ii) nucleus-localized isoforms of SH2B1 directly promote neurite outgrowth and gene expression at the level of the nucleus, or (iii) the robust neurite outgrowth and gene expression responses are the consequence of multiple small changes that combine to cause a more substantial effect.

In summary, these data together suggest that for many of the functions of SH2B1, the region common to the four isoforms of SH2B1 (amino acids 1 to 631) has an intrinsic level of activity that is modulated by the unique C-terminal tails of the various isoforms. The effect of the β tail varies, but in general, SH2B1β maintains the ability to enhance the various functions of NGF that we have studied. In contrast, the α tail blocks much of the intrinsic activity present in the region of SH2B1 common to all the isoforms. Intriguingly, the ability of the α tail to block the intrinsic function of the common region of SH2B1 appears to be regulated by phosphorylation at Tyr753 in the α tail. This suggests that as-yet-unknown inputs could regulate the function of SH2B1α. Our data suggest that at least part of the ability of SH2B1β and the inability of SH2B1α to enhance neurite outgrowth and gene transcription starts at the level of the TrkA receptor. Since SH2B1β and SH2B1α have the same SH2 domain, we would expect them to compete for binding to a phosphorylated tyrosine in TrkA. Our data also suggest that when SH2B1α binds, TrkA phosphorylates Tyr753, which by some unknown mechanism inactivates SH2B1α, thereby preventing enhancement of TrkA activity. Because all SH2B1 isoforms share the same SH2 domain and SH2B1 isoforms are reported to bind to multiple neurotrophic factor receptors, we predict that SH2B1α would similarly compete with the other isoforms of SH2B1 for binding to TrkB and other neurotrophic factor receptors. One can envision that by so doing, SH2B1α helps to fine tune the response to neurotrophic factors, which regulate not only the expression of neuronal proteins and the growth of neuronal projections but also the pruning of neuronal projections, including projections in the brain that regulate energy expenditure.

MATERIALS AND METHODS

Antibodies.

Rabbit polyclonal antibodies to Myc (A14) (immunoblotting [IB], 1:1,000; immunoprecipitation [IP], 1:80) (Fig. 8B and C), and mouse monoclonal antibody to Myc (9E10) (IB, 1:2,000) (Fig. 8A and D) or Trk (C-14, sc-11) (1:500) were from Santa Cruz. Mouse monoclonal antibodies to Erk1 and Erk2 (4696S) (1:2,000) and Akt (2920S) (1:2,000), rabbit monoclonal antibodies to phospho-Akt (Ser473) (4058L) (1:1,000), phospho-TrkA (Y490) (4619S) (1:500), and phospho-TrkA (Y674/Tyr675) (4621P) (1:500), and rabbit polyclonal antibodies to doubly phosphorylated Erk1 and Erk2 (Thr202/Tyr204) (9101S) (1:1,000), PLC-γ (2822S) (1:1,000), and phospho-PLC-γ (Tyr783) (2821S) (1:500) were from Cell Signaling. Rabbit polyclonal antibody to Myc (A190-105A) (IP, 1:450) (Fig. 8B and C) was from Bethyl Laboratories. Mouse monoclonal antibody to phosphotyrosine (4G10) (1:1,000) and rabbit polyclonal antibody to TrkA (06-574) (1:1,000) were from Millipore. Mouse monoclonal antibody to mCherry (orb66657) (IP, 1:500) (Fig. 7A) was from Biorbyt. Rabbit polyclonal antibody to GFP (632592) (1:2,000) (Fig. 6) was from Clontech. Rabbit polyclonal antibody to phospho-SH2B1 (Tyr439) (1960) (1:1,000) was made in collaboration with Millipore and has been described previously (51). Goat anti-GFP antibody conjugated to IRDye800 (600-132-215) (IB, 1:20,000) (Fig. 5) and rabbit polyclonal antibody to RFP (600-401-379) (1:1,000) (Fig. 9B) used to detect levels of tdTomato (tdT) were from Rockland Immunochemicals. Goat anti-rabbit IgG conjugated to IRDye 700 (827-11081) (1:20,000) and goat anti-mouse IgG conjugated to IRDye 800 (926-32210) (1:15,000) were from Li-Cor Biosciences.

Plasmids.

cDNAs encoding mouse SH2B1α (GenBank accession number AF421138) (10) and GFP- and Myc-tagged rat SH2B1β (GenBank accession number NM_001048180) (7) have been described previously. cDNAs encoding rat mCherry-TrkA and untagged TrkA were gifts from Chengbiao Wu (University of California at San Diego) (Fig. 7A and 8A, C, and D) (65) and David Ginty (Harvard University) (Fig. 8B) (47), respectively. The sequence encoding SH2B1α was subcloned into pEGFPC1 (Clontech) and Prk5 vector containing an N-terminal Myc tag. Mutations were introduced into SH2B1β and SH2B1α using the QuikChange II site-directed mutagenesis kit (Stratagene) according to the manufacturer's protocol. Kinase-dead (KD) TrkA was prepared by mutating Lys547 to an Ala (K547A). The primers used are listed in Table 1. To generate vectors encoding SH2B1β and SH2B1α N-terminally tagged with TdT, the enhanced GFP (eGFP) in GFP-SH2B1α and GFP-SH2B1β in the pEGFPC1 vector was replaced with TdT from the ptdTomato-N1 vector (632532; Clontech) using the AgeI (5′) and BsrGI (3′) restriction sites. New mutations and constructs were confirmed by sequencing (University of Michigan DNA Sequencing Core).

TABLE 1.

Primers for site-directed mutagenesis

Species	Target		Direction	Primer sequence (5′ → 3′)
	Protein	Mutation
M. musculus	SH2B1α S752	S752A	Forward	CGAGCCATTAATAATCAGTACGCATTTGTGTGAGATACC
			Reverse	GGTATCTCACACAAATGCGTACTGATTATTAATGGCTCG
	SH2B1α V754	V754D	Forward	CAGTACTCATTTGACTGAGATACCTGCCCACCCTC
			Reverse	GAGGGTGGGCAGGTATCTCAGTCAAATGAGTACTG
	SH2B1α Y753	Y753E	Forward	CGAGCCATTAATAATCAGGAGTCATTTGTGTGAGATACC
			Reverse	GGTATCTCACACAAATGACTCCTGATTATTAATGGCTCG
	SH2B1α Y753	Y753F	Forward	CGAGCCATTAATAATCAGTTCTCATTTGTGTGAG
			Reverse	CTCACACAAATGAGAACTGATTATTAATGGCTCG
R. norvegicus	SH2B1β G632	Stop codon (1–631)	Forward	CAGCGGCAGCAGTGACGGGAGCAGGCTG
			Reverse	CAGCCTGCTCCCGTCACTGCTGCCGCTG
M. musculus	SH2B1α Y48	Y48F	Forward	CGACGTTTTCGCCTCTTCCTGGCCTC
			Reverse	GAGGCCAGGAAGAGGCGAAAACGTCG
	SH2B1α Y55	Y55F	Forward	CCACAGTTTGCAGAGCCGGGAGCAGAGG
			Reverse	CCTCTGCTCCCGGCTCTGCAAACTGTGG
	SH2B1α Y354	Y354F	Forward	CCTTCAGAGTTCATCCTGGAGACAAGTGATGCG
			Reverse	CGCATCACTTGTCTCCAGGATGAACTCTGAAGG
	SH2B1α Y439	Y439F	Forward	GTCGCAGGGAGCTTTTGGGGGCCTCTCAGACC
			Reverse	GGTCTGAGAGGCCCCCAAAAGCTCCCTGCGAC
	SH2B1α Y494	Y494F	Forward	CCCCTCTCTACCCCCTTCCCTCCCCTGGATAC
			Reverse	GTATCCAGGGGAGGGAAGGGGGTAGAGAGGGG
	SH2B1α Y525	Y525F	Forward	CAGCCCCTCTCAGGCTTTCCTTGGTTCCAC
			Reverse	GTGGAACCAAGGAAAGCCTGAGAGGGGCTG
	SH2B1α Y564	Y564F	Forward	GACAAGGCGTGGTGAATTTGTCCTCACTTTCAACTTCC
			Reverse	GGAAGTTGAAAGTGAGGACAAATTCACCACGCCTTGTC
	SH2B1α Y624	Y624F	Forward	CAGTGATGTTGTCCTTGTCAGCTTTGTGC
			Reverse	GCACAAAGCTGACAAGGACAACATCACTG
R. norvegicus	SH2B1β Y55	Y55F	Forward	CCACAATTTGCAGAGCCGGGAGCAGAGGCTG
			Reverse	CAGCCTCTGCTCCCGGCTCTGCAAATTGTGG
	SH2B1β Y439	Y439F	Forward	GTCGCAGGGAGCTTTTGGAGGCCTCTCAGACC
			Reverse	GGTCTGAGAGGCCTCCAAAAGCTCCCTGCGAC
	TrkA K547	K547A	Forward	GTGGCTGTCGCGGCACTGAAGGAGACATCTG
			Reverse	CAGATGTCTCCTTCAGTGCCGCGACAGCCAC

Cell culture and transfection.

PC12 cells (CRL-1721; ATCC) were plated on dishes coated with rat tail type I collagen (BD Biosciences) and maintained at 37°C with humidified air at 5% CO₂ in normal growth medium (RPMI 1640 medium, A10491-01; Life Technologies) containing 10% horse serum (HS; Atlanta Biologicals) and 5% fetal bovine serum (FBS; Life Technologies). PC12 cells were transiently transfected using Lipofectamine LTX (Life Technologies) for 24 to 72 h according to the manufacturer's instructions. Pooled PC12 cell lines stably expressing GFP or GFP-tagged SH2B1β, SH2B1α, SH2B1α Y753F, or SH2B1 1–631 were generated by transfecting cells with the plasmid encoding that protin using Lipofectamine LTX for 72 h. Cells were grown in G418 medium (normal growth medium containing 0.5 mg/ml G418 [Cellco; Corning]) for 5 days, in G418-free medium for 7 days, and then in G418 medium for an additional 30 days. The top 60% of GFP-positive cells were gated and sorted from non-GFP-expressing cells by fluorescence-activated cell sorting (FACS; Beckman Coulter MoFlo Astrios, University of Michigan Flow Cytometry Core). Sorted cells were maintained in G418 medium. Prior to experimental use, cells were incubated overnight in deprivation medium (RPMI 1640 containing 1% bovine serum albumin [BSA; Proliant Biologicals]).

293T cells were maintained at 37°C with humidified air at 5% CO₂ in Dulbecco's modified Eagle's medium (DMEM) (11965-092; Life Technologies) supplemented with 1 mM l-glutamine, 0.25 μg/ml amphotericin, 100 U/ml penicillin, 100 μg/ml streptomycin (DMEM culture medium), and 8% calf serum (CS; Life Technologies). 293T cells were transiently transfected using calcium phosphate precipitation (66) or polyethyleneimine (PEI) (23966; Polysciences). Briefly, PEI (1 μg/μl) was mixed with DNA at a 4:1 (wt/vol) ratio in Opti-MEM (Gibco, 31985-070; Life Technologies) and incubated for 10 min at room temperature. Cells were used after 24 or 48 h.

Neurite outgrowth.

PC12 cells or stably transfected PC12 cells were plated in 6-well dishes and transiently transfected with the indicated construct. Cells were treated with 25 ng/ml mouse NGF 2.5S (356004 [BD Biosciences] [Fig. 1 and 2] or BT.5025 [Harlan Bioproducts] [Fig. 4 to 6, 7B, and 9]) in RPMI 1640 medium containing 2% HS and 1% FBS. GFP-positive cells (Fig. 1 and 2) or GFP- and tdT-positive cells (Fig. 9) were visualized by fluorescence microscopy (20× or 40× objective, Nikon Eclipse TE200) after 1, 2, or 3 days as indicated. One hundred GFP- or GFP- and tdT-positive cells in three different areas of each plate were scored for the presence of neurites ≥2 times the length of the cell body (total of 300 cells per condition per experiment). Each experiment was carried out 2 (Fig. 9) or 3 times (Fig. 1 and 2) (total of 600 or 900 cells per condition, respectively). The percentage of cells with neurites was determined by dividing the number of cells with neurites by the total number of GFP-positive cells (Fig. 1 and 2) or GFP-/tdT-positive cells (Fig. 9) counted.

Immunoblotting and immunoprecipitation.

Stably transfected PC12 cells were seeded at 10 × 10⁶ cells per 10-cm collagen-coated dish in normal growth medium. After 24 h, cells were incubated overnight in deprivation medium. Cells were incubated at 37°C with NGF (25 or 100 ng/ml) as indicated, placed on ice, and washed two times with phosphate-buffered saline (PBS; 10 mM NaPO₄, 140 mM NaCl, pH 7.4) containing 1 mM Na₃VO₄, pH 7.3 (PBSV). Cells were lysed with ice-cold L-RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 2 mM EGTA, 0.1% Triton X-100, pH 7.2) (Fig. 5A) for 10 min or with modified L-RIPA buffer (20 mM Tris, 150 mM NaCl, 1% Triton X-100, 1 mM EGTA, 1 mM EDTA, pH 7.4) (Fig. 6, 7B, and 9B) for 30 min on ice. Both lysis buffers were supplemented with 20 mM NaF, 1 mM Na₃VO₄, and protease inhibitors (1 mM phenylmethylsulfonyl fluoride [PMSF], 10 μg/ml aprotinin, 10 μg/ml leupeptin). Cell lysates were centrifuged at ∼15,000 × g for 10 min at 4°C, and protein concentrations of the supernatant were determined using the DC protein assay (Bio-Rad). Equal amounts of proteins were resolved using 5-to-12% gradient (Fig. 5), 10% (Fig. 6 and 7B), or 9% (Fig. 9B) SDS-PAGE gels and transferred to low-fluorescence PVDF (Bio-Rad) membranes.

Transfected 293T cells were maintained in normal growth medium, incubated at 37°C, placed on ice, and washed 2× with PBSV. Cells were then lysed on ice for 5 to 10 min with ice-cold L-RIPA buffer containing 20 mM NaF, 1 mM Na₃VO₄, 1 mM PMSF, 10 μg/ml aprotinin, 10 μg/ml leupeptin (Fig. 7A and 8). Cell lysates were centrifuged at ∼15,000 × g for 10 min at 4°C. For immunoprecipitation, antibodies to Myc (A14 [1:80] [Fig. 8A and D] or c-Myc [1:450] [Fig. 8B and C]) or mCherry (1:500) (Fig. 7A) were preconjugated to protein A-agarose beads (Repligen) overnight at 4°C, and then the beads were added to cell lysates containing equal amounts of protein for 2 to 3 h at 4°C under constant rotation. Bound proteins were eluted and resolved on 9% (Fig. 8) or 7.5% (Fig. 7A) SDS-PAGE gels. Proteins were transferred onto nitrocellulose (Bio-Rad) (Fig. 8B and C) or low-fluorescence PVDF (Bio-Rad) (Fig. 7A and 8A and D) membranes.

For both PC12 and 293T cells, blots were incubated overnight at 4°C with antibodies in 10 mM Tris, 150 mM NaCl, pH 7.4, 0.1% Tween 20, 3 to 5% BSA (1% chicken egg albumin for blots probed with anti-pTyr antibody), followed by secondary antibody for 1 h at room temperature. Bands were visualized using the Odyssey infrared imaging system (Li-Cor Biosciences) and quantified using Image Studio Lite 4.0 (Fig. 5) or 5.2 (Fig. 6).

Live-cell imaging.

GFP, GFP-tagged SH2B1β, SH2B1 1–631, or SH2B1α, or the indicated GFP-tagged SH2B1α mutant was transiently expressed in 293T cells seeded on 10-mm glass-bottom dishes (MatTek) in normal growth medium. Cells were imaged in Ringer's buffer (10 mM HEPES, 155 mM NaCl, 2 mM CaCl₂, 5 mM KCl, 1 mM MgCl₂, 10 mM NaH₂PO₄, 10 mM glucose, pH 7.2) using a 60× water immersion objective on an Olympus FV500 laser-scanning confocal microscope and FluoView version 5.0 software (Morphology and Image Analysis Core of the Michigan Diabetes Research Center). A multiline argon blue laser was used to excite GFP fluorescence at 488 nm. All cells were monitored for 4 to 6 h after the addition of 20 nM leptomycin B (LMB; Sigma) and imaged at the time when the nuclear-to-cytoplasmic ratio of GFP-SH2B1β in GFP-SH2B1β-expressing cells became ≥1. The relative levels of GFP-SH2B1 in the plasma membrane, cytoplasm, and nucleus were determined from 16-bit images using the line scan function in MetaVue 6.0 (Universal Imaging, Sunnyvale, CA). Image planes were chosen to allow maximal visualization of the plasma membrane and nucleus. MetaVue was used to draw a 15-pixel-wide line scan perpendicular to the plasma membrane or nuclear membrane. For each line scan, the peak pixel intensity at the plasma membrane and average pixel intensity within the nucleus or cytoplasm were used to generate plasma membrane/cytoplasm or nucleus/cytoplasm ratios (34, 51). To avoid bias, line scans were generated by an individual blinded to the experimental conditions. Two or more individual experiments were conducted with each form of GFP-SH2B1. For each condition, the signals from 16 to 80 cells were quantified by line scan. Dying or dead cells and cells with a particularly high or low expression level were excluded (usually <10% of the cells). A too-high level of expression resulted in a plasma membrane/cytoplasm signal that was abnormally low, thought to be due to saturation of a limited number of binding sites in the plasma membrane. A too-low level of expression of SH2B1 prevented accurate determination of the plasma membrane/cytoplasm and nuclear/cytoplasm ratios due to an inability to accurately determine SH2B1 levels in the cytoplasm.

qPCR.

PC12 cells stably expressing GFP or the indicated GFP-SH2B1 variant were incubated at 37°C in deprivation medium overnight. Cells were treated with or without 100 ng/ml NGF (Harlan) for 6 h. Total RNA was extracted using TRIzol (Ambion; Life Technologies) according to the manufacturer's instructions, and RNA quality was confirmed by optical density using a Nanodrop spectrophotometer. cDNA was generated from the RNA using the TaqMan reverse transcription reagent kit (N808-0234; Applied Biosystems) according to the manufacturer's instructions. The transcript levels of Plaur, Mmp10, Mmp3, FosL1, Hprt (67), and Cyclophilin A (68) were determined in triplicate per gene by quantitative PCR (qPCR) using a TaqMan kit (Invitrogen), SYBR green I (Life Technologies), and an Eppendorf Realplex2 instrument using Mastercycler software. The primers used are listed in Table 2. Cycle threshold values were normalized to the geometric mean values of the cycle thresholds (69) for Hprt and Cyclophilin A, whose expression did not differ between the different cell lines or with NGF treatment (data not shown). The results were then normalized to the value obtained for GFP-SH2B1β cells treated with NGF.

TABLE 2.

Primers for qPCR

R. norvegicus gene target	Direction	Primer sequence (5′ → 3′)
Mmp3	Forward	TGAAGATGACAGGGAAGCTGG
	Reverse	GGCTTGTGCATCAGCTCCAT
Mmp10	Forward	GAAATGGTCACTGGGACCCTC
	Reverse	TGCGCAGCAACCAGGAATA
FosL1	Forward	GCAAGCGCAGACACAGACAG
	Reverse	CTTGGCACAAGGTGGAACTTC
UPAR	Forward	CAAAGCACAGAACGGAGCG
	Reverse	GCCACAGCCTTTGGTGTAGG
Cyclophilin A	Forward	AACTTTCGTGCTCTGAGC
	Reverse	ATGGCGTGTGAAGTCACC
Hprt	Forward	CTCATGGACTGATTATGGACAGGAC
	Reverse	GCAGGTCAGCAAAGAACTTATAGCC

Statistics.

Statistics were performed using GraphPad Prism 7.01. For neurite outgrowth and Western blot quantification, time course curves were converted to areas under the curve (AUCs). For qPCR experiments, relative transcript expression levels were normalized to those obtained with NGF-treated GFP-SH2B1β cells. When the means of the data for some conditions are near zero, the equal variance assumption for analysis of variance (ANOVA) is not met. In these cases, to equalize the variance, data were transformed by taking the log of the relative expression (Fig. 4). Similarly, to equalize the variance of the AUC data shown in Fig. 5 and 6, data were transformed by taking the log of the relative expression. Statistical significance was determined using one-way ANOVA with repeated measures and multiple comparison analysis with Dunnett's posttest or the Student t test (Fig. 3B and C) as indicated. Data show the mean values ± standard errors of the means. P values of ≤0.05 were considered significant.