Regulation of insulin and type 1 insulin-like growth factor signaling and action by the Grb10/14 and SH2B1/B2 adaptor proteins


  • Bernard Desbuquois,

    Corresponding author
    • Institut Cochin, Départment d'Endocrinologie, Métabolisme et Cancer, Université Paris-Descartes, Institut National de la Santé et de la Recherche Médicale, Unité 1016, et Centre National de la Recherche Scientifique, Unité Mixte de Recherche 8104, Paris, France
    Search for more papers by this author
  • Nadège Carré,

    1. Institut Cochin, Départment d'Endocrinologie, Métabolisme et Cancer, Université Paris-Descartes, Institut National de la Santé et de la Recherche Médicale, Unité 1016, et Centre National de la Recherche Scientifique, Unité Mixte de Recherche 8104, Paris, France
    Current affiliation:
    1. Institut National de la Santé et de la Recherche Médicale, Unité 693 and Faculté de Médecine Paris Sud, Université Paris Sud, Le Kremlin Bicêtre, France
    Search for more papers by this author
  • Anne-Françoise Burnol

    1. Institut Cochin, Départment d'Endocrinologie, Métabolisme et Cancer, Université Paris-Descartes, Institut National de la Santé et de la Recherche Médicale, Unité 1016, et Centre National de la Recherche Scientifique, Unité Mixte de Recherche 8104, Paris, France
    Search for more papers by this author


B. Desbuquois, Département d'Endocrinologie, Métabolisme et Cancer, Faculté de Médecine, 24 rue du Faubourg Saint Jacques, 75014 Paris, France

Fax: +33 1 47 53 27 03

Tel: +33 1 47 53 27 08



The effects of insulin and type 1 insulin-like growth factor (IGF–1) on metabolism, growth and survival are mediated by their association with specific receptor tyrosine kinases, which results in both receptor and substrate phosphorylation. Phosphotyrosine residues on receptors and substrates provide docking sites for signaling proteins containing SH2 (Src homology 2) domains, including molecular adaptors. This review focuses on the regulation of insulin/IGF–1 signaling and action by two adaptor families with a similar domain organization: the growth factor receptor-bound proteins Grb7/10/14 and the SH2B proteins. Both Grb10/14 and SH2B1/B2 associate with the activation loop of insulin/IGF–1 receptors through their SH2 domains, but association of Grb10/14 also involves their unique BPS domain. Consistent with Grb14 binding as a pseudosubstrate to the kinase active site, insulin/IGF-induced activation of receptors and downstream signaling pathways in cultured cells is inhibited by Grb10/14 adaptors, but is potentiated by SH2B1/B2 adaptors. Accordingly, Grb10 and Grb14 knockout mice show improved insulin/IGF sensitivity in vivo, and, for Grb10, overgrowth and increased skeketal muscle and pancreatic β–cell mass. Conversely, SH2B1-depleted mice display insulin and IGF–1 resistance, with peripheral depletion leading to reduced adiposity and neuronal depletion leading to obesity through associated leptin resistance. Grb10/14 and SH2B1 adaptors also modulate insulin/IGF–1 action by interacting with signaling components downstream of receptors and exert several tissue-specific effects. The identification of Grb10/14 and SH2B1 as physiological regulators of insulin signaling and action, together with observations that variants at their gene loci are associated with obesity and/or insulin resistance, highlight them as potential therapeutic targets for these conditions.


Akt serine/threonine protein kinase/protein kinase B


adaptor protein with PH and SH2 domains


adipocyte lipid-binding protein


Bcl2 antagonist of cell death


brain-derived neurotrophic factor


Bcl2 interacting mediator of cell death


between the PH and SH2 domains


proto-oncogene c-Abl


Casitas B-lineage lymphoma proto-oncogene


CCAAT/enhancer binding protein


c–Cbl-associated protein


proto-oncogene c-Crk


downstream of tyrosine kinase


epidermal growth factor


receptor tyrosine-protein kinase erbB2


extracellular signal-regulated kinase


focal adhesion kinase


fibroblast growth factor receptor


tyrosine-protein kinase Fyn


Grb2-associated binder

GIGYF protein

Grb10-interacting GYF protein


growth factor receptor bound protein


glycogen synthase kinase


glutathione S–transferase


hepatocyte nuclear factor 4


insulin-like growth factor


type 1 insulin-like growth factor


IGF–1 receptor


insulin receptor


insulin receptor kinase


insulin receptor substrate


Janus kinase


lymphocyte-specific adapter protein


mitogen-activated protein kinase


mouse embryonic fibroblasts


transforming protein N-Ras/GTPase N-Ras


neuronal precursor cell-expressed developmentally down-regulated


nerve growth factor


platelet-derived growth factor receptor


3′–phosphoinositide-dependent protein kinase


phosphoprotein enriched in diabetes


pleckstrin homology


phosphoinositide 3–kinase


phosphorylated insulin receptor interacting region


protein kinase C


peroxisome proliferator-activated receptor


SH2 domain-containing signaling mediator


phosphatase and tensin homolog


phosphotyrosine phosphatase




proto-oncogene tyrosine-protein kinase receptor Ret


receptor tyrosine kinase


Src homology 2


Src homologous and collagen protein


SH2 domain containing inositol phosphatase 2


small hairpin RNA


small interfering RNA


suppressor of cytokine signaling


sterol regulatory element binding protein


tyrosine-protein kinase Tec


tyrosine-protein kinase Tie2


mammalian target of rapamycin


mTOR multiprotein complex 1


T-cell protein tyrosine phosphatase


protein kinase Cmath formula interacting protein


Insulin and insulin-like growth factors (IGFs) control metabolism, growth and survival in many mammalian tissues but play distinct physiological roles. The major action of insulin is to regulate glucose and lipid metabolism in muscle, fat and liver, whereas IGFs affect mainly cell growth and differentiation, with type 2 insulin-like growth factor (IGF–2) acting in the fetus and with type 1 insulin-like growth factor (IGF–1) acting postnatally. In addition, both insulin and IGF–1 are involved in the regulation of lifespan and in the development of neoplasia.

The first step in insulin/IGF–1 action is their association with structurally related receptor tyrosine kinases (RTKs) at the cell surface. This results in receptor phosphorylation and the recruitment and phosphorylation of two main families of substrates: the insulin receptor substrate (IRS) and Src homologous and collagen (Shc) proteins. In thurn these substrates recruit, through their phosphotyrosine residues, effector proteins that contain Src homology 2 (SH2) domains, leading to activation of two major signaling pathways: the phosphoinositide 3–kinase (PI3K)/Akt serine/threonine kinase and Ras/extracellular signal-regulated kinase (ERK) pathways (Fig. 1). The former mediates acute metabolic actions of insulin via serine phosphorylation of multiple targets, whereas the latter is involved in the regulation of cell proliferation and differentiation. Both pathways also regulate gene expression at the level of transcription, and, for the PI3K/Akt pathway, translation [1-4].

Figure 1.

Canonical insulin/IGF–1 signaling pathways. The main components of the PI3K/Akt and Ras/ERK pathways, including receptors, substrates (in orange), adaptors and transducers (in yellow), serine/threonine kinases (in green), downstream components (in blue) and some negative regulators (in purple) are shown. Adapted from [4].

In addition to IRS and Shc proteins, various substrates [Grb2-associated binder (Gab), Dok, Crk and SH2B) and adaptors/scaffolds (growth factor receptor-bound proteins Grb7/10/14, Rack1, β–arrestin and cytohesin) that are recruited and/or phosphorylated by the insulin receptor (IR) or the IGF–1 receptor (IGFR) modulate the PI3K/Akt and Ras/ERK pathways and/or elicit specific biological responses. Other pathways have been implicated in insulin signaling, including activation of non-RTKs such as Janus kinases (JAKs), activation of class II and III PI3K and PIKfyve, and inhibition of protein tyrosine phosphatases (i.e. PTP1B) and lipid phosphatases [i.e. phosphatase and tensin homolog (PTEN) and SH2 domain-containing inositol phosphatase–2 (SHIP2)] by generation of reactive oxygen species [4].

Insulin/IGF–1 signaling is also subjected to negative regulation through feedback mechanisms and cross-talk from other pathways. Major counter-regulatory mechanisms include dephosphorylation of signaling components by protein and lipid phosphatases, modification of serine/threonine residues of IR and IRS proteins by phosphorylation and O–GlcNacylation, and recruitment to IR and/or IRS of negative regulators such as suppressor of cytokine signaling (SOCS) and Grb7/10/14 adaptor proteins [4].

This review focuses on the regulation of insulin/IGF–1 signaling and action by the Grb10/14 and SH2B1/B2 members of the Grb7/10/14 and SH2B protein families. First identified as partners of the activated epidermal growth factor (EGF) receptor and regulators of immune cell activation, respectively, these two families display a similar domain organization and associate with a variety of receptor and non-receptor tyrosine kinases, as well as many other partners. Within the Grb7/10/14 family, Grb10 and Grb14 have emerged as major negative regulators of insulin/IGF–1 action on metabolism and growth [5-8], whereas Grb7 regulates focal adhesion kinase (FAK)-mediated cell migration [9]. Within the SH2B family, the role of SH2B2 in insulin action remains unclear, but SH2B1 has been been identified as a positive regulator of insulin, IGF–1 and leptin action [10]. In addition, as described below, variants at the Grb10/14 and SH2B1 gene loci have been shown to be associated with obesity, altered body fat distribution and/or insulin resistance in humans.

Grb7/10/14 and SH2B proteins

Structure and oligomerization

The Grb7/10/14 and SH2B adaptor protein families each have three members. The SH2B family members are SH2B1, SH2B2 and SH2B3, originally named as SH2B [also known as SH2 domain-containing signaling mediator (PSM), adaptor protein with PH and SH2 domains (APS) and Lnk, respectively. Both families possess a SH2 domain near the C–terminus, a central pleckstrin homology (PH) domain and several proline-rich sequences in the N–terminal region (Fig. 2). In the Grb7/10/14 family, a Ras-associating (RA) domain precedes the PH domain, and the PH and SH2 domains are separated by a conserved region referred to as the BPS domain (between the PH and SH2 domains) or the phosphorylated insulin receoptor-interacting region (PIR), which is unique to this family. In the SH2B family, an N–terminal dimerization domain and multiple consensus sites for tyrosine and serine/threonine phosphorylation are present. Grb7/10/14 and SH2B proteins are products of single genes, but alternative splicing and/or use of alternative translation initiation sites generate several isoforms. Thus, Grb10 occurs as six isoforms (α, β, γ, δ, ε and ζ), that differ at their N–termini, SH2B1 as four isoforms (α, β, γ and δ), which are identical at their N–termini but differ at their C–termini shortly after the SH2 domain, and SH2B2 as two isoforms (α and β), which possess (α) or lack (β) a SH2 domain. Sequence identity among members of the Grb7/10/14 and SH2B families is highest in the SH2 domain (approximately 60–70% and 70–80%, respectively) but identity between the SH2 domains of the two families is only 25–30%.

Figure 2.

Schematic representation of the Grb7/10/14 and SH2B family members. Both families possess a SH2 domain at or near the C–terminus, a central PH domain and a proline-rich region near the N–terminus. In addition, Grb7/10/14 family members possess an RA (Ras-associating) domain and a PIR (phosphorylated insulin receptor-interacting) or BPS (between PH and SH2) domain, and SH2B family members possess an N–terminal dimerization domain. Grb7/10 and SH2B1/B2 occur as several isoforms generated by alternative splicing and/or use of alternative translation initiation sites. Within the Grb7/10/14 family, human Grb10β, as well as human Grb10ε and ζ (not shown), differ from human Grb10γ by a common extended N–terminal region and additional unique segments at the extreme N–terminus, and Grb10β differs from other Grb10 isoforms by a truncated PH domain. Mouse Grb10α and Grb10δ are very similar to human Grb10γ but possess additional mouse-specific regions between the proline-rich and RA domains. Within the SH2B family, SH2B1α, β, γ and δ differ at their C–terminus shortly after the SH2 domain, and SH2B2β differs from SH2B2α by the absence of a SH2 domain. For each isoform, the number of amino acids is indicated on the right. Adapted from [8] and [10]. Permission for reproduction granted by Elsevier Limited, License number 3046401296649.

Both Grb10/14 and SH2B12/B2 proteins form oligomers in solution and in cells. Grb10γ has been identified by gel chromatography as an elongated dimer [11], and Grb14 has been shown by bioluminescence resonance energy transfer to undergo insulin-induced dimerization in cells [12]. Based on crystal structures, Grb10/14 homodimerize through their RA/PH and SH2 domains [13, 14], whereas SH2B1 and SH2B2 homo- and heterodimerize through their N–terminal domains [15, 16].

Grb7/10/14 and SH2B SH2 domains

The SH2 domain mediates association with phosphotyrosine residues of partner RTKs either entirely (SH2B) or in part (Grb7/10/14), and, unlike for other signaling proteins, binds to phosphotyrosines in the kinase activation loop. Typically, SH2 domains consist of a core anti-parallel β–sheet formed by seven β–strands (βA–βG) and two α–helices (αA and αB); a conserved Arg residue at βB5 in the sequence FLVRES and a pair of Lys residues at βD1/βD3 are involved in phosphotyrosine recognition. In the Grb7/10/14 SH2 domain, the EF loop possesses a 4–5 amino acid extension, non-glycyl residues are present at BG3 (Val/Ile), EF1 (Asp) and BC5 (Lys/Gln), and the binding pocket for the p + 3 residue is masked. In the SH2B1/B2 SH2 domain, the five residues after the βD strand are disordered, and the α–helix that follows is extended. These features favor binding of the turn-containing phosphotyrosine sequences found in the activation loop of IR/IGFR [8].

Although SH2 domains are typically monomeric, the Grb10/14 SH2 domain in solution is dimeric; the dimer interface is made up of conserved residues at the SH2 C–terminal α–helix [11, 13]. The SH2 domain of SH2B2, unlike that of SH2B1, is also dimeric; Trp475 at βF2 is a key determinant of dimerization [17]. The different oligomeric status of the SH2B1/B2 SH2 domains has been suggested to be the reason for the preferential association of SH2B1 with JAK2 and of SH2B2 with IR [18].

Grb7/10/14 non-SH2 domains

The PIR/BPS domain interacts specifically with IR/IGFR and mediates inhibition of receptor kinase activity. Although unfolded in solution [19], this domain acquires structure when bound to the phosphorylated IR [13].

The PH domain mediates in vitro and in vivo association with membrane phosphoinositides, with highest affinity for D3 and D5 phosphoinositides [14, 20, 21]. Studies using a fluorescent polarization assay have shown that the Grb10 PH domain binds phosphoinositides with a higher affinity (Kd 4–10 μm) than the Grb14 PH domain (Kd 28–100 μm) [14].

The RA domain targets Grb7/10/14 proteins to insulin-activated GTPases at the plasma membrane [14, 22]. Crystallographic studies have shown that the RA and PH domains of Grb10, together with the intervening 40 amino acid linker, physically associate to form an integrated structural unit that is dimerized via a C–terminal helical extension of the PH domain [14]. The two domains share an extensive interface but their GTPase and phosphoinositide binding sites remain fully accessible. Some hydrophobic residues at the RA/PH interface are conserved, indicating that the model also applies to Grb14. The function of this integrated RA/PH unit is to favorably position the RA domain for binding to membrane-anchored GTPases. Mutagenesis studies suggest that the RA and PH domains are both involved in the interaction of Grb14 with N–Ras [14].

The N–terminal region contains a conserved proline-rich sequence (PS/AIPNPFPEL) and other PXXP motifs that are involved in the binding of SH3 domains. These sequences have been shown to mediate the binding of Grb10 to the SH3 domain of c–Abl and the GYF motifs of Grb10-interacting GYF (GIGYF) proteins [8], and have been implicated in the association of Grb14 with the SH3 domain of phospholipase CCγ [23].

SH2B N–terminal dimerization domain

This conserved domain of approximately 60 amino acids, referred to as a phenylalanine zipper, mediates the homo- and heterodimerization of SH2B1/B2 adaptors [15, 16]. It consists of a compact U–shaped four-helix bundle with a stack of inter-digitated phenylalanine side chains. Dimerization of SH2B2 has been shown to induce the dimerization and activation of IR and IGFR β–subunit mnomeric constructs [15, 16].

Binding partners of Grb7/10/14 and SH2B adaptors

Grb7/10/14 and SH2B adaptors associate with a variety of proteins of diverse function, but with some preferences depending on the interacting partners (Table S1). In addition to insulin and IGF–1 receptors, the major partners identified for both families include several RTKs [platelet-derived growth factor receptor (PDGFR), fibroblast growth factor receptor (FGFR) and Ret] and docking proteins (IRS1, Shc and p85 PI3K); partners for the Grb710/14 family include the non-receptor tyrosine kinase FAK, several serine/threonine kinases, the tyrosine phosphatase PTP1B and the ubiquitin ligase neuronal precursor cell-expressed developmentally down-regulated 4 (Nedd4); partners for the SH2B family include the RTKs of brain-derived neurotrophic factor (BDNF) and nerve growth factor (NGF), the JAK receptor-associated tyrosine kinases, the ubiquitin ligase c–Cbl, and several actin-regulating proteins.

Unlike docking proteins recruited by activated RTKs, such as IRS and Shc, Grb7/10/14 proteins are generally not direct RTK substrates. However, Grb7 is tyrosine-phosphorylated by Tie2, Ret and HER2/erbB2 tyrosine kinases [24, 25] and FAK [26], Grb10 is tyrosine-phosphorylated by Tec, Src and Fyn kinases [27-29], and Grb14 is tyrosine-phosphorylated by Tie2 [29, 30] and light-activated Src kinase [31]. In contrast, SH2B adaptors are substrates of both RTKs and JAKs, and their phosphotyrosine residues provide docking sites for other binding partners [10]. Both families of adaptors undergo basal and growth factor/cytokine-stimulated serine/threonine phosphorylation.

In addition to their role in the regulation of insulin/IGF–1 signaling, both adaptor families regulate, negatively or positively, signaling mediated by other RTKs. For instance, Grb7 and Grb14 inhibit FGFR-mediated signaling [23, 32], whereas Grb10 and Grb14 enhance signaling mediated by PDGFR [33] and Ret [34], respectively. In addition, SH2B adaptors regulate signaling and action mediated by JAKs upon JAK activation in response to ligand occupancy of associated receptors [10]. Thus, SH2B1 enhances the effects of growth hormone on cell motility and of leptin on energy balance and body weight, SH2B2 inhibits erythropoietin action, and SH2B3 inhibits cytokine-mediated production of B–cell precursors and hematopoetic progenitor cells [10].

Expression of Grb10/14 and SH2B1/B2 adaptors

Tissue distribution and subcellular localization

Grb10 is an imprinted gene, being expressed predominantly from the maternally inherited allele in most human and mouse tissues except brain [8]. Consistent with their role in insulin signaling, Grb10/14 mRNA and protein are expressed in skeletal muscle and white adipose tissue, two major insulin target tissues, as well as heart and kidney [35-38]. A high level of expression is also observed for Grb10 in pancreatic islets [39-42] and for Grb14 in liver [37, 38] and retinal rod photoreceptor cells [43]. Although also expressed in insulin target tissues, SH2B1/B2 adaptors are more broadly distributed than Grb10/14 adaptors are. SH2B1 mRNA [44-48] and protein [49-51] are expressed in skeletal muscle, kidney, spleen, pancreas, lung, brain, heart, ovary, testis, adipose tissue and liver, with a preferential expression of α/δ or β/γ isoforms depending on the tissue. SH2B2 expression is more restricted, affecting mainly skeletal muscle, adipose tissue and heart [52-54].

As for many insulin signaling proteins, Grb10/14 are associated with both cytoplasm and membranes. In cultured cell lines, Grb10 has been found to be localized in mitochondria in association with Raf1 kinase [55], and Grb14 has been found to be localized in a low-density microsomal fraction in association with tankyrase 2 [56]. In rat liver, Grb14 is predominantly localized in microsomal and cytosolic fractions [57], and it is partly associated with the nuclear fraction in rat retina, consistent with the presence of a functional nuclear localization signal at the Grb14 N–terminus [21]. Although residing mainly at the plasma membrane and in the cytoplasm, SH2B1 has been shown to undergo constitutive nucleo-cytoplasmic shuttling in neuronal cells, mediated by nuclear localization and export sequences [10].

Expression in human and animal disease: correlations with impaired growth, insulin resistance, obesity and cancer

Despite the identification of Grb10 as a growth-inhibiting protein, expression of Grb10 in mouse tissues declines during postnatal growth deceleration, together with that of other imprinted genes, both growth-inhibiting and growth-promoting [58]. However, two other observations indicate a negative correlation between expression of the Grb10 gene maternal allele and growth. First, duplication of proximal chromosome 11, which harbors the murine Grb10 gene, is associated with prenatal growth retardation. Second, disruption of the E3 ubiquitin ligase Nedd4 gene in mice causes severe growth retardation and perinatal lethality, associated with over-expression of Grb10 protein in embryonic fibroblasts [59]. This phenotype is rescued by disruption of the maternal allele of the Grb10 gene, suggesting that Grb10 over-expression mediates the growth defect observed in Nedd4-deficient mice.

Genome-wide association studies in humans have revealed an association of single nucleotide polymorphisms at the Grb10 [60, 61] and Grb14 [62] gene loci with type 2 diabetes, and, when accounting for body mass index or glycemic traits, with increased fasting blood glucose and plasma insulin [63, 64]. The Grb14 locus was identified as one of 16 loci that modulate body fat distribution as determined by waist/hip ratio, serum triglycerides, plasma insulin and insulin sensitivity [65]. A recent meta-analysis of genetic variants using the Metabochip (a custom array designed to study variants associated with metabolic diseases) also identified Grb14 as a susceptibility locus for type 2 diabetes [66]. In addition, single nucleotide polymorphisms at the SH2B1 gene were shown to be associated with obesity [67-71] and visceral fat accumulation [72], and a chromosomal deletion that eliminates the SH2B1 gene was associated with early-onset severe obesity and insulin resistance [73, 74].

Grb10/14 expression in insulin target cells is negatively correlated with insulin sensitivity. The Grb14 mRNA and protein levels in adipose tissue are increased in human type 2 diabetes and rodent models of insulin resistance [75]. The Grb14 mRNA level is also increased in skeletal muscle of obese human subjects, with partial reversal upon weight loss induced by gastric bypass [76]. However, despite opposing effects of Grb10/14 and SH2B1 on insulin signaling, the SH2B1 mRNA level in adipose tissue is increased in obese db/db mice and normal mice fed a high-fat diet, possibly reflecting counter-regulation [49].

Changes in expression of the Grb10/14 genes, albeit variable, have been described in cancer. In some studies, Grb10 [77] and Grb14 [34, 78, 79] were over-expressed in human tumors and tumor cell lines, and immunity to Grb10 was associated with inhibition of breast tumor growth in mice [80]. However, in a recent meta-analysis of microarray data, Grb10 expression was decreased in many human tumors, consistent with Grb10 acting as a tumor suppressor [81]. These opposing changes are not necessarily conflicting as, depending on the interacting receptor, Grb10/14 may inhibit or stimulate cell proliferation. Thus, the over-expression of Grb14 in human thyroid cancer is consistent with the ability of Grb14 to enhance cell proliferation through activation of Ret signaling [34].

In a microarray analysis of cardiomyocyte-specific transgenic mice with increased or decreased PI3K activity subjected to myocardial infarction, Grb14 expression in the heart was correlated with PI3K activity and cardiac function [82]. Together with the cardiac dysfunction observed in Grb14-deficient mice, these findings suggest that Grb14 is a PI3K-regulated gene that confers a cardioprotective phenotype. Grb14 has also been identified as a susceptibility gene for elevated blood pressure [83].

Regulation of expression in cultured cells

Consistent with the over-expression of Grb14 in tissues of hyperinsulinemic humans and rodents, Grb14 mRNA and protein levels are up-regulated by insulin in cultured adipocytes [75] and mammary cells [79]. Although it is unknown whether Grb10/14 genes are regulated by insulin at the transcriptional level, a recent study has demonstrated epigenetic silencing of the human Grb14 gene by the transcription factor HNF–4α, together with that of the PED gene, which is over-expressed in type 2 diabetes [84]. Within the SH2B family, the expression of SH2B1 in 3T3L1 adipocytes [49, 85] and of SH2B2 in L6 myoblasts [86] increases upon insulin-induced differentiation. In differentiated myoblasts, SH2B2 expression is down-regulated by resistin and insulin, as is IRS1 and IRS2 expression [86].

Involvement of Grb10/14 in insulin and IGF–1 action

Association of Grb10/14 with activated insulin and IGF–1 receptors

First detected by yeast two-hybrid and glutathione S–transferase (GST) pull-down assays, interactions of Grb7 [87], Grb10 [27, 36, 88-93] and Grb14 [12, 38, 94] with IR and of Grb10 with IGFR [36, 91, 93, 95, 96] have been demonstrated by co-immunoprecipitation and bioluminescence resonance energy transfer of over-expressed proteins in CHO, COS and HEK293 cells. Endogenous Grb14 also associates with ligand-activated IR in 3T3L1 adipocytes and rat liver [38, 57] and with light-activated IR in rat retina [21]. In cultured cells, the association of Grb14 with the IR is a rapid process and persists after removal of insulin from the medium [97].

Recruitment of Grb10 [27, 88] and Grb14 [14, 98] to activated IR in cells results in their translocation from the cytosol to the plasma membrane. Following insulin injection of rats, liver endogenous Grb14 associates with the activated IR first at the plasma membrane and then in endosomal fractions, suggesting that it may be in part co-internalized as a complex with the IR [57]. In rat retina, light-induced activation of the IR is accompanied by a major redistribution of Grb14 from inner to outer segments of rod photoreceptor cells; this process requires photoactivation of rhodopsin but no direct Grb14–rhodopsin interaction [99]. However, Grb14 redistribution in the retina is not IR-dependent as it is unaffected by ablation of photoreceptor-specific IR. In mouse embryonic fibroblasts (MEFs) over-expressing the IGFR, Grb10 has been shown by confocal microscopy to undergo co-internalization with ligand-activated IGFR and the ubiquitin ligase Nedd4 through both clathrin-dependent and -independent pathways [100].

Molecular determinants of Grb7/10/14 association with insulin and IGF–1 receptors

Based in part on the crystal structure of the Grb14 BPS domain in complex with IRK (Fig. 3), a model for the interaction of the Grb14 BPS–SH2 module with the IR has been proposed [13] (Fig. 4). In this model, the three phosphotyrosines of the insulin receptor kinase (IRK) activation loop are engaged by Grb14. pTyr1158 is bound to the canonical binding pocket of the SH2 domain and coordinated by Arg466 at βB5; pTyr1162 is coordinated by Lys484/486 at βD1/βD3 in the SH2 domain; pTyr1163 interacts with Trp406 in the BPS region. The N–terminal conserved sequence of the BPS region LVAMDF(376–381), in which the non-phosphorylatable Leu376 replaces substrate tyrosine, binds as a pseudo-substrate to the substrate peptide-binding groove of IRK. In addition to the interaction of BPS Trp406 with IRK pTyr1163, Arg387 in the BPS region interacts with Lys1168 in the IRK activation loop, and Gly403, Leu404 and Trp406 in the BPS α1 helix interacts with Arg1039, Leu1038 and Ile1042 in the IRK αC helix, respectively (Fig. 3). The BPS Leu404–IRK Leu1038 interaction probably explains the selectivity of IR–Grb14 complex formation, as these two residues are not conserved in the related proteins IGFR and Grb10, respectively [101].

Figure 3.

Crystal structure of the Grb14 BPS–IRK complex. (A) Diagram of the complex showing the N- and C–terminal lobes (dark and light grey), catalytic loop (orange) and activation loop (green) of phosphorylated IRK, and the BPS region of Grb14 (purple). The Grb14 BPS region adopts secondary structure when bound to IRK, in the form of two short β–strands (β1 and β2) and a 16-residue α–helix (α1), with its N–terminal portion positioned in the substrate peptide binding groove of IRK. (B) View of the pseudo-substrate inhibitory region of the Grb14 BPS region with superimposed IRS1-derived substrate peptides bound to IRK (in black) and IGFR kinase (in blue). In the BPS pseudo-substrate region, Leu376 replaces substrate Tyr but, in common with substrate peptides, hydrophobic residues are present at positions +1 (Val377), +3 (Met379) and +5 (Phe381) C-terminally to Leu376. (C) Stereo view of the interactions within the Grb14 BPS region and between the Grb14 BPS region and IRK, showing side chains and hydrogen bonds/salt bridges (dashed black lines). The conformation of the Grb14 BPS region is stabilized by Arg387, which is salt-bridged to Asp380 (+4) and Glu394, by Val388, which makes contact with Val377 (+1) and Met379 (+3), and by Asn391, which is hydrogen-bonded to Glu394. Val388 in the Grb14 BPS pseudo-substrate region makes contact with Glu1216 and Leu1219 in the IRK αG helix, Arg387 and Glu394 the Grb14 BPS region make contact with Lys1168 in the IRK activation loop, and Gly403, Leu404 and Trp406 in the Grb14 BPS α1 helix make contact with Arg1039, Leu1038 and Ile1042 in the IRK αC helix, respectively. An unusual interaction occurs between the BPS-conserved residue Trp406 and IRK pTyr1163. Reproduced from [13] with permission. Permission granted by Elsevier Limited. License number 3046411246926.

Figure 4.

A simplified representation of the interactions of Grb10/14 and SH2B1/B2 adaptors with the IR. Both Grb10/14 and SH2B1/B2 adaptors associate as dimers with phosphorylated IR β subunits; in addition to their SH2 domains, dimerization involves their RA/PH (Grb10/14) and N–terminal (SH2B1/B2) domains. Grb10/14 associate with IR through both their SH2 domain, which engages pTyr1158 and 1162 in the IR activation loop (green stars), and their BPS domain, which engages pTyr1163 (red star) and binds as a pseudo-substrate inhibitor to the IRK active site (in yellow). The RA and PH domains, by interacting with GTPases and phosphoinositides at the plasma membrane, contribute to the Grb10/14 association with IR. Consequently, substrate access to IR and IR catalytic activity are inhibited by Grb10/14 despite the protection of phosphotyrosines in the IR activation loop from tyrosine phosphatases such as PTP1B. In contrast, although targeted to the plasma membrane by their PH domain, SH2B1/B2 associate with the IR activation loop solely through their SH2 domain, leaving the IRK active site fully accessible. The enhanced IR phosphorylation and catalytic activity induced by SH2B1/B2 may result from protection of phosphotyrosine residues in the IR activation loop from PTP1B action.

Consistent with structural studies, mutations of Lys1030 at the ATP-binding site of IRK and of the three tyrosines in the kinase activation loop, particularly Tyr1162 and Tyr1163, abrogate or reduce association of Grb10/14 with IR [38, 87, 88, 91, 93] (Table S2). In contrast, phosphorylation of tyrosines residues in the IR juxtamembrane and C–terminal regions is not required for interaction, although, based on phosphopeptide competition studies, Tyr972 in the juxtamembrane region is involved in interaction with Grb10 [27]. Mutations of Leu1038 in the IR α–helix and Lys1168 in the IR activation loop markedly reduce association of IR with Grb14 [101].

Mutagenesis studies have identified several residues in the BPS domain of human [13, 14] and rat [101] Grb14 that are critical for interaction with IR (Table S2). In human Grb14, these include Ala378 in the pseudo-substrate region and Leu404 and Trp406 in the α–helix. In rat Grb14, these include Arg385 (equivalent of human Grb14 Arg387), which stabilizes the conformation of the BPS region. Also important is the phosphorylation status of rat Grb14 Ser370 [101]. In addition to the BPS domain, the SH2, RA and PH domains are required for high-affinity interaction of human Grb14 with the IR, as shown by the inhibitory effects of mutations of Arg466, Lys484/486 and Phe519 (SH2 domain), Lys140 (RA domain) and Lys252 (PH domain) (Table S2). However, Gln348 and Asn349, which are involved in human Grb14 RA–PH dimerization, are not required for interaction [14].

Effects of Grb7/10/14 on insulin and IGF–1 receptor kinase activity

Both full-length and BPS-containing fragments of Grb7/10/14 adaptors inhibit in vitro IR/IGFR catalytic activity towards peptide substrates [97, 102] (Fig. 5). This involves a non-competitive mechanism with an order of potency Grb14 > Grb10 > Grb7, and IR activity more potently inhibited than IGFR activity [97]. Consistent with these in vitro effects, over-expression of Grb10 [81, 92, 103-105] and Grb14 [38, 94, 97] in cells inhibits insulin- and/or IGF–1-stimulated tyrosine phosphorylation of endogenous substrates (IRS1, IRS2, Shc and p62Dok). Conversely, depletion of endogenous Grb10 by small interfering RNA (siRNA) interference potentiates insulin/IGF-induced substrate phosphorylation [81, 103, 106, 107]. Endogenous Grb14 appears to exert a negative feedback control on IRK in liver [57].

Figure 5.

Schematic representation of the sites at which Grb14 modulates insulin signaling in cells. Association of Grb14 with the activated IR inhibits substrate access to the IRK active site and IR catalytic activity (1); PTP1B-induced dephosphorylation of Tyr972 in the IR juxtamembrane domain contributes to inhibition of IRS1 association with the IR (2). As a result, activation of downstream PI3K/Akt and Ras/ERK signaling pathways (3), and serine phosphorylation of Akt substrates such as GSK3 and Foxo1 (4) are inhibited. The inhibitory effect of Grb14 on insulin signaling is enhanced by its direct interaction with the Akt-activating serine kinase PDK1 (5) and its indirect interaction with the serine kinase PKCζ through the molecular adaptor ZIP; the latter interaction mediates Grb14 serine phosphorylation (6). Although it inhibits phosphorylation of Akt substrates, Grb14 stimulates insulin SREBP–1c activation (7) and SREBP–1c-dependent expression of glycolytic and lipogenic genes, but the interacting partner involved has not yet been identified.

Although it inhibits substrate phosphorylation, Grb10/14 hardly affects in vitro IR phosphorylation [97], and, when over-expressed in cells, Grb10/14 do not affect [105] or even increase [97] IR phosphorylation. Conversely, depletion of Grb10 in cells causes a decrease in ligand-stimulated IGFR phosphorylation [106], and disruption of Grb10/14 genes in mice causes a decrease in in vivo insulin-stimulated IR phosphorylation [37, 39]. As Grb14 inhibits PTP1B association with the IR and PTP1B-induced IR dephosphorylation [97], these effects reflect protection of phosphotyrosines in the IR activation loop from dephosphorylation. Studies using site-specific antibodies have indeed shown that, in cells expressing IR and PTP1B, Grb14 co-expression maintains phosphorylation of tyrosines in the IR activation loop, while favoring dephosphorylation of Tyr972 in the juxtamembrane domain [12, 108]. As Tyr972 is the main docking site for IRS1, this may contribute to the ability of Grb14 to inhibit the association of IRS1 with the IR, although such association is disrupted by Grb10 in the absence of Tyr972 dephosphorylation [105]. In retina photoreceptor cells, Grb14 also protects light-activated IR against PTP1B-induced dephosphorylation, but, as described below, it does so through direct interaction with PTP1B and inhibition of PTP1B activity.

Modulation of insulin and IGF–1 signaling and action by Grb10/14 in cultured cells

As anticipated from their inhibitory effects on IR/IGFR kinase, when over-expressed in cells, Grb10/14 inhibit the activation of downstream PI3K/Akt and ERK1/2 signaling pathways [8, 38, 92, 94, 97, 103, 105], as well as distal effects of insulin such as stimulation of DNA and glycogen synthesis [8, 38, 79, 109] (Fig. 5). GST fusion proteins containing Grb10/14 BPS and SH2 domains also inhibit insulin/IGF-induced mitogenesis in fibroblasts [91, 93] and insulin-induced maturation of Xenopus laevis oocytes [32, 101, 110]. The inhibitory effects of Grb14 mutants on insulin signaling generally correlate with their ability to interact with the IR (Table S2). Interestingly, mutation of Grb10 at two mitogen-activated protein kinase (MAPK) phosphorylation sites (Ser150 and 476) reduces the ability of Grb10 to inhibit insulin-stimulated IRS1 phosphorylation in cells, indicating that phosphorylation of these sites is required for full inhibition [111].

Consistent with over-expression studies, depletion of endogenous Grb10 by siRNA interference increases insulin-dependent phosphorylation of Shc, MAPK and Akt in HeLa cells [81, 103, 107], as well as IGF-dependent Akt and ERK1/2 phosphorylation and DNA synthesis in NIH3T3 cells [81, 106, 107]. Likewise, Grb14 depletion in primary hepatocytes enhances insulin-induced activation of Akt and ERK1/2, phosphorylation of glycogen synthase kinase 3 (GSK3) and FoxO1 and expression of gluconeogenic enzymes, as well as insulin-induced inhibition of glucose production [112]. Grb14-depleted MEFs also display an increased proliferation rate, which is reversed by Grb14 expression [101].

Paradoxically, stimulatory effects of Grb10 over-expression on mitogenic effects of insulin and IGF–1 in NIH3T3 fibroblasts [33, 113] and on metabolic effects of insulin in L6 myoblasts and 3T3L1 adipocytes [114] have also been reported. Whether, as previously proposed [8], these opposite effects reflect the use of different Grb10 isoforms and domains, cell lines that are not representative of insulin target cells, and/or non-physiological levels of Grb10 expression remains unclear.

Positive effects of Grb14 in cultured hepatocytes have also been described [112]. Despite the improved activation of the PI3K/Akt pathway, Grb14 depletion in these cells inhibits insulin-induced maturation of the transcription factor sterol regulatory element binding protein 1c (SREBP–1c), and consequently stimulation of glycogen synthesis and expression of glycolytic and lipogenic enzymes (Fig. 5). These findings suggest that Grb14 acts as a positive regulator downstream of the PI3K/Akt pathway, possibly by interacting with a partner involved in SREBP–1c maturation. The increased inhibition of glucose production and decreased stimulation of lipogenesis induced by insulin in Grb14-depleted hepatocytes are expected to both oppose the insulin resistance associated with obesity and liver steatosis.

Functional consequences of Grb10/14 interactions with signaling proteins downstream insulin/IGF–1 receptors

Based in part on mutagenesis studies, some effects of Grb10/14 in cultured cells (summarized in Table S3 and Fig. 5) have been linked to their ligand-dependent or -independent interaction with signaling components downstream of IR/IGFR. Some of these interactions favor positive rather than negative effects on PI3K/Akt and Ras/Erk signaling pathways, and therefore their physiological relevance remains to be established.

The association of Grb10 through its SH2 domain with the IR substrate Gab1 has been implicated in positive regulation of cell proliferation [113]. Over-expression in fibroblasts of a Gab1 mutant lacking the Grb10-binding region reduces activation of p44/42 MAPK by insulin/IGF–1, and disruption of the Gab1 gene severely reduces the ability of Grb10 to potentiate insulin/IGF–1-induced cell proliferation.

The constitutive interaction of Grb10 with mitochondrial Raf1 serine kinase has been shown to mediate positive regulation of cell survival through PI3K/Akt- and Ras/ERK-dependent phosphorylation and inactivation of BAD, a pro-apoptotic protein [115]. Cells over-expressing SH2-mutated forms of Grb10, which do not bind to Raf1, undergo apoptosis [55, 116]. In addition, Grb10- and Raf1-depleted MEFs display increased sensitivity to apoptosis in response to Bad expression; this sensitivity is rescued by expression of Grb10 and Raf1, respectively [115]. Mutagenesis studies have shown that the SH2, proline-rich and PH domains of Grb10, and its Akt phosphorylation site and 14.3.3 binding site, are necessary for rescue, as are the RA domain and kinase activity of Raf1 [115].

The interaction of Grb10 with BIM L, a pro-apoptotic member of the Bcl2 protein family, has been found to be involved in positive regulation of cell survival through BIM L inactivation [41, 117]. Over-expression of Grb10 and BIM L in HEK293 cells inhibits apoptosis in a BIM L phosphorylation-dependent manner [117]. Conversely, depletion of Grb10 in mouse pancreas by a lentivirus shRNA strategy induces apoptosis of both exocrine and endocrine cells, predominantly α cells, together with increased expression of BIM L in pancreatic islets [41]. Consistent with apoptosis of α cells, the plasma glucagon level is decreased and glucose tolerance is increased in these mice.

In a large-scale analysis of the phosphoproteome regulated by the serine/threonine kinase mTOR, a major controller of cell growth, Grb10 has been identified as a direct substrate and binding partner of the mTORC1 multi-protein complex [81, 117]. Serine phosphorylation of Grb10 by mTORC1, which stabilizes the Grb10 protein, is required for inhibition by Grb10 of ligand-induced IR/IGFR phosphorylation, IRS phosphorylation, and activation of downstream PI3K/Akt and Ras/ERK signaling pathways. These results underline the importance of the mTORC1 pathway in Grb10-dependent regulation of insulin/growth factor signaling.

The indirect association of Grb14 with atypical protein kinase Cζ (PKCζ) mediated by the molecular adaptor ZIP (PKCζ-interacting protein) has been implicated in Grb14 serine phosphorylation in vitro and insulin-induced Grb14 phosphorylation in CHO–IR cells [110]. However, whether Grb14 simultaneously associates with ZIP/PKCζ and the activated IR is unknown. Serine phosphorylation of Grb14 by PKCζ increases its inhibitory effect on insulin-induced activation of IRK and maturation of Xenopus oocytes [110].

The direct interaction of Grb14 with the Akt-activating serine kinase 3′-phosphoinositide-dependent protein kinase (PDK1) has been implicated in both stimulatory and inhibitory effects of Grb14 on insulin signaling. Based on effects of Grb14 mutated at the PDK1-binding motif YAKYEF (human Grb14 sequence 191–196), this interaction has been shown to facilitate insulin-dependent recruitment of PDK1 to the cell membrane and activation of Akt in HEK293 cells [98]. However, the Grb14–PDK1 interaction is also required for full inhibition by Grb14 of insulin action in Xenopus oocytes and serum-induced proliferation of Grb14-depleted MEFs [101].

The light-dependent association of Grb14 with PTP1B in the retina results in inhibition of PTP1B activity [31]. Phosphorylation of Tyr347 in the Grb14 BPS domain, mediated by light-induced activation of retinal Src kinase, is required for these effects. Grb14-induced inhibition of retinal PTP1B activity has been implicated in the ability of Grb14 to protect light-activated retinal IR from dephosphorylation.

Modulation of in vivo insulin and IGF–1 sensitivity by Grb10/14 adaptors

Transgenic mice over-expressing the Grb10 maternal allele show postnatal growth retardation, hyperinsulinemia, glucose intolerance and insulin resistance [118, 119]. At 3 months of age, some transgenic lines with a high level of Grb10 expression in the pancreas display severe hyperglycemia, decreased fasting plasma insulin, severe atrophy of pancreatic acinar cells, β–cell enlargment and islet dysmorphism.

Disruption of the maternal allele of the Grb10 gene results in an opposite phenotype comprising over-growth and increased insulin sensitivity [39, 40, 120, 121]. Over-growth affects embryos, placenta and neonates [120, 121], and is maintained throughout life, despite attenuating with age. Adult Grb10-deficient mice display an increased total body mass, resulting from an increase in lean and muscle mass, and reduced adiposity. Muscle enlargement is due to an increase in myofiber number, with no change in fiber size and type, indicating that Grb10 is involved in muscle development during embryogenesis [122]. This is accompanied, in neonate mice, by an up-regulation of genes involved in cancer/proliferation and myogenic signaling pathways [122]. Adult Grb10-deficient mice also show improved in vivo glucose tolerance and insulin sensitivity, enhanced insulin-induced IRS1, Akt and MAPK phosphorylation in skeletal muscle and adipose tissue, and enhanced IGF-induced IRS1 phosphorylation in muscle [39, 40]. The latter observation suggests that, although embryonic over-growth of Grb10-deficient mice is IGF–2/IGFR-independent [120], improved IGF–1 signaling may mediate postnatal over-growth, at least in part. Pancreas-specific disruption of the Grb10 gene enhances insulin/IGF–1 signaling in islets, β–cell mass and insulin content, and insulin secretion [42]. In addition, it improves glucose tolerance in mice fed a high-fat diet and protects mice from streptozotocin-induced β–cell apoptosis [42]. Ablation of the Grb10 paternal allele, which is selectively expressed in brain, results in behavioral changes but does not affect growth and insulin sensitivity [123].

Disruption of the Grb14 gene results in a slight decrease in body mass and liver mass, an increase in heart mass, and, after 16 weeks of age, an improved in vivo glucose tolerance and insulin sensitivity [37]. It also enhances insulin-induced stimulation of glucose transport in skeletal muscle, as well as glycogen synthesis in liver and muscle. These metabolic effects are accompanied, despite reduced phosphorylation of the liver IR, by an increased ability of insulin to stimulate IRS1 tyrosine phosphorylation and to activate the PI3K/Akt pathway in liver and skeletal muscle, but not adipose tissue. Despite a similar improvement of this pathway in liver of Grb14-deficient mice [37] and Grb14-depleted cultured hepatocytes [112], insulin-induced glycogen synthesis is enhanced in the former but abrogated in the latter. However, studies on Grb14-deficient mice were performed a long time after disruption of the Grb14 gene, and this may have allowed some adaptations to occur.

Interestingly, as for Grb14 deficiency, PTP1B deficiency in mice enhances glucose tolerance, insulin sensitivity and insulin-induced activation of the IRS/PI3K/Akt pathway [124]. However, unlike in Grb14-deficient mice, the activated IR in PTP1B-deficient mice is protected from dephosphorylation. This suggests that PTP1B and Grb14 both act as inhibitors of IRK activity, and that the relative expression of these two regulators controls the level of IR phosphorylation.

Despite differences in tissue and developmental expression, the Grb10 and Grb14 adaptors exhibit partially overlapping functions. To further delineate the physiological role of these two adaptors, double knockout mice were generated [125]. These mice exhibit an increase in lean mass comparable to that of Grb10-deficient mice, indicating that lean mass regulation is a Grb10-specific function. In addition, although disruption of either Grb10 or Grb14 genes enhances insulin-induced IRS1 phosphorylation and Akt activation in muscle, this is not increased further by dual disruption, and, under certain conditions, may be attenuated by increased IR hypophosphorylation and/or decreased IRS1 expression. Unlike single knockout mice, double knockout mice are protected from the glucose intolerance induced by high-fat feeding, and high fat-fed double knockout mice show reduced expression of IRS1 in liver and muscle. These findings suggest that Grb10 and Grb14 exert non-redundant effects on whole-body glucose homeostasis, which may involve distal steps of the insulin signaling pathways.

Gene disruption studies in mice have revealed heart- and retina-specific effects of Grb14 that differ from those observed in classical insulin target tissues. Thus, Grb14-deficient mice develop a cardiomyopathy comprising an increase in heart and atrial weight, a decrease in cardiac function and a decrease in phosphorylated Akt in the heart [82]. Ablation of Grb14 in mice also results in a loss of light-dependent activation of IRK in rod photoreceptor cells, which results from increased PTP1B-induced IR dephosphorylation [31].

The phenotypes of Grb10 and Grb14 knockout mice are summarized in Fig. 6 and Table S4.

Figure 6.

Major consequences of Grb10, Grb14 and SH2B1 deficiency in mouse tissues and cells. Effects induced by systemic or targeted disruption of the Grb10 [39, 40, 42, 122], Grb14 [37, 99] and SH2B1 [46, 49, 85] genes are shown in blue, and effects induced by Grb10 shRNA [41] and Grb14 siRNA [112] strategies are shown in yellow. In some SH2B1-deficient mice, SH2B1 expression in the central nervous system was restored [85].

Role of SH2B1/B2 in insulin and IGF–1 action

Association of SH2B1/B2 adaptors with insulin and IGF–1 receptors

Interactions of SH2B1 with activated insulin [44, 47, 51, 126] and IGF–1 receptors and of SH2B2 with activated IR [53, 127] have been demonstrated by yeast two-hybrid, GST pull-down and co-immunoprecipitation assays. Unlike Grb10/14, SH2B2 and, to a lesser extent, SH2B1 undergo insulin-dependent tyrosine phosphorylation in CHO–IR and 3T3L1 cells [51, 53, 128]. In SH2B2, Tyr618 at the C–terminus is the major site phosphorylated; expression of the TyrSH2B2Phe mutant in cells prevents both phosphorylation of wild-type SH2B2 and its association with the IR [53]. SH2B2 is also phosphorylated at Ser588 in insulin-treated cells [129].

Analysis of the crystal structure of the SH2 domain of SH2B2 in complex with phosphorylated IRK has shown that, as for the Grb14 SH2 dimer, the SH2B2 SH2 dimer associates with two IRK molecules and engages both pTyr1158 and pTyr1162 of the kinase activation loop [17] (Fig. 7). pTyr1158 is bound to the canonical pY binding pocket of the SH2 domain and is salt-bridged to Arg437 at βB5, and pTyr1162 is coordinated by Lys455/457 at βD1/βD3. Unlike Grb10/14 adaptors, association of the SH2 domain of SH2B2 with IRK leaves the substrate-binding site fully accessible (Fig. 4).

Figure 7.

Crystal structure of the SH2B2 SH2 domain–IRK complex. Interactions between SH2B2(SH2) (in green) and the IRK activation loop (in yellow) are shown as black dashed lines. pTyr1158 in the IRK activation loop is salt-bridged to invariant Arg437 (βD5) in the canonical phosphate-binding pocket and also interacts with the side chains of Arg436 (αA2), Ser439 (βB7) and Thr441 (BC2), and with the backbone nitrogen of Glu440 (BC1). pTyr1162 is salt-bridged to Lys455/457 (βD1/βD3) and is positioned by Arg1164 for recognition by this lysine pair. Reproduced from [17] with permission. Permission granted by Elsevier Limited. License number 3046420085853.

As with Grb10/14, the association of SH2B1 [17, 51] and SH2B2 [53, 127, 136] with the activated IR is critically dependent on phosphorylation of the three tyrosines of the IR activation loop. In addition, the ability of SH2B1/B2 to associate with the IR and to potentiate insulin-induced IR and IRS1/IRS2 tyrosine phosphorylation requires the integrity of their SH2 domain, as shown by adverse effects of mutations of amino acids that make contact with the IR activation loop (Arg555/560 in SH2B1; Arg437/442, Lys455/457 and Arg460 in SH2B2) and are involved in SH2 domain dimerization (Trp475 in SH2B2) [17, 18, 50, 53, 130] (Table S5). In contrast, deletion of the N–terminal sequence 1–504 does not affect SH2B1 activity [50, 130], although it is required for SH2B1 action in neurones [131].

Effects of SH2B1 on insulin/IGF–1 signaling and action in vitro and in cultured cells

In a cell-free system, purified GST–SH2B1 dose-dependently stimulates IRK activity as determined by tyrosine phosphorylation of GST–IRS1 [131]. In CHO, NIH3T3 and HEK293 cells stably expressing the IR, SH2B1 over-expression increases basal and insulin-stimulated IR and IRS1 phosphorylation, as well as Akt and Erk activation [50, 128, 132] (Fig. 8). In 3T3L1 adipocytes, over-expressed SH2B1 [49] and membrane-permeant SH2B1 peptides [133] also enhance insulin/IGF-induced activation of the IRS1/PI3K/Akt pathway. Conversely, depletion of endogenous SH2B1 in cultured hepatocytes results in a decrease in IR/IRS1 phosphorylation and PI3K/Akt activation [130], as do silencing of the SH2B1 gene and expression of dominant-negative SH2B1 mutants in NIH3T3 fibroblasts [133].

Figure 8.

Schematic representation of sites at which SH2B1 modulates insulin signaling in cells. Association of SH2B1 with the activated IR enhances IR catalytic activity (1) and IRS1 association with IR (2). This in turn potentiates the activation of downstream PI3K/Akt and Ras/ERK signaling pathways (3) and distal insulin metabolic and mitogenic actions. In addition, the direct association of SH2B1 with IRS proteins (4), by blocking PTP1B-induced IRS dephosphorylation, contributes to activation of the PI3K/Akt pathway.

Distal effects of insulin and IGF–1, including DNA synthesis in fibroblasts [48, 134] and glucose transport, amino acid uptake, glycogen synthesis and lipogenesis in 3T3L1 preadipocytes [133] are potentiated by over-expressed SH2B1. The effectiveness of SH2B1 isoforms decreases in the order α > β > δ > γ for metabolic effects [133] and γ > δ > α > β for mitogenic effects [48]. Conversely, insulin/IGF–1 effects are inhibited by putative dominant-negative SH2B1 domain-specific peptides [133, 135]. Insulin-induced differentiation of 3T3L1 adipocytes is also increased by SH2B1β over-expression, together with induction of the adipogenic genes peroxisome proliferator-activated receptor γ (PPARγ), C/EBPα and aP2.

By protecting phosphorylated IRS proteins from PTP1B-induced dephosphorylation, the insulin-dependent association of SH2B1 with IRS1/IRS2 contributes to the ability of SH2B1 to potentiate insulin signaling [130] (Fig. 8). In cells expressing the SH2B1-insensitive IR mutant Y1158F, SH2B1 increases insulin-induced IRS1 phosphorylation and PI3K/Akt activation [130]. Whether SH2B1 participates in formation of a trimeric IR/SH2B1/IRS1 complex in response to insulin, comparable to the leptin-induced trimeric JAK2/SH2B1/IRS1 complex [136], is unknown.

Effects of SH2B2 on insulin signaling and action in cultured cells

The results of several studies are consistent with a positive regulatory role for SH2B2α in early steps of insulin signaling. When over-expressed in CHO–IR cells, SH2B2α increases and prolongs insulin-induced IR autophosphorylation and IRS1 phosphorylation, as well as ERK and/or Akt activation [128, 137]. SH2B2β, which lacks the SH2 C–terminal domain of SH2B2α, inhibits the ability of SH2B2α and SH2B1 to increase insulin-induced IRS1 phosphorylation through dimerization with these isoforms [54].

By mediating the recruitment of the ubiquitin ligase c–Cbl and the c–Cbl-associated protein CAP to the activated IR, SH2B2 has been shown to facilitate the phosphorylation of c–Cbl and membrane translocation of the glucose transporter Glut4 in 3T3L1 adipocytes [128, 138, 139]. SH2B2 has been proposed to initiate a PI3K-independent signaling pathway involving CAP, c–Cbl, the adaptor Crk, the GTPase exchange factor C3G and the Rho GTPase TC10 as downstream components [140, 141]. However, this pathway does not appear to be functional in skeletal muscle, and knocking down its components in adipocytes does not disrupt insulin-stimulated glucose transport [4]. In addition, the ability of over-expressed SH2B2 to enhance insulin-induced Glut4 translocation and glucose uptake in 3T3L1 cells was not confirmed in a later study [52]. Thus, at the present time, the role of the SH2B2/CAP/Cbl pathway in insulin signaling remains questionable.

The interaction of SH2B2 with Enigma, a PDZ and LIM domain-containing protein, has been implicated in the regulation of insulin-induced glucose transport by affecting actin cytoskeleton organization [142, 143]. In NIH3T3 cells co-expressing SH2B2 and Enigma, insulin increases their association and co-localization in large F–actin-containing ruffles. In 3T3L1 adipocytes, over-expression of Enigma inhibits insulin-induced Glut4 translocation and actin cytoskeleton remodeling, and the SH2B2 binding domain of Enigma is required for these effects. As Enigma is over-expressed in adipose tissue of obese diabetic patients, it has been suggested that, by interacting with SH2B2, Enigma acts as an inhibitor of insulin-induced actin remodeling and Glut4 translocation [142, 143].

Effects of SH2B1 and SH2B2 on insulin and IGF–1 action in vivo

Gene disruption studies in mice are consistent with a positive effect of SH2B1 on insulin/IGF–1 signaling and action, but two distinct phenotypes have been described, probably due to different genetic backgrounds (Fig. 6 and Table S6). In some studies, SH2B1-deficient mice developed age-dependent glucose intolerance, insulin resistance and impaired activation of the PI3K/Akt and Ras/Erk pathways in liver, muscle and adipose tissue [50, 144]. These mice also developed, by 21 weeks of age, a phenotype of leptin resistance associated with hyperphagia, obesity and energy imbalance [145]. As both phenotypes were reversed by restoration of neuronal SH2B1 expression, peripheral insulin resistance was initially attributed to obesity induced by neuronal leptin resistance [85], although neuronal insulin resistance cannot be excluded. However, when restricted to peripheral insulin target tissues, SH2B1 deficiency exacerbated glucose intolerance induced by a high-fat diet and led to impaired insulin signaling in tissues of high fat-fed mice, supporting a role for peripheral SH2B1 in the regulation of insulin action [130]. Peripheral SH2B1 depletion also led to a decrease in adipose tissue mass and an impaired ability of MEFs to differentiate into adipocytes, indicating that adipose SH2B1 controls adipogenesis [85].

In another series of studies, disruption of the SH2B1 gene led to growth retardation 2–6 weeks after birth, impaired fertility, developmental defects in gonadal organs, a reduced response of oocytes to IGF–1, and reduced adiposity and adipogenic gene expression [46, 49]. In addition, MEFs of SH2B1-deficient mice showed a reduced ability to differentiate into adipocytes, together with a decrease in IRS1 phosphorylation, Akt activation and PPARγ induction in response to insulin and IGF–1 [49]. These changes were reversed by over-expression of SH2B1, suggesting that this adaptor, by potentiating the PI3K/Akt signaling pathway, is a key regulator of PPARγ-mediated adipogenesis.

Unlike for SH2B1, the phenotypes of SH2B2-deficient mice described in two independent studies do not support a positive regulatory role for SH2B2 on insulin signaling. (Table S6). In one study, disruption of the SH2B2 gene led to an increase in whole-body insulin sensitivity and insulin-simulated glucose transport in isolated adipocytes, suggesting a negative role for SH2B2 on insulin signaling [52]. In another study, combined disruption of the SH2B1 and SH2B2 genes caused a lesser increase in plasma insulin level than disruption of the SH2B1 gene alone, also supporting negative regulation of insulin sensitivity by SH2B2 [144].

Regulation of insulin-like signaling and action by the SH2B adaptor in Drosophila

The insulin/IGF signaling pathway is structurally and functionally conserved, as shown by its ability to regulate growth, metabolism, stress resistance, reproduction and lifespan in Caenorhabitis elegans and Drosophila [146, 147]. Two independent groups have shown that the single SH2B protein expressed in Drosophilia (dSH2B) acts as a positive regulator of insulin-like signaling [148-150]. Disruption of the dSH2B gene results in reduced body size and weight, growth retardation, reduced female fertility, increased longevity, increased resistance to starvation and oxidative stress, increased lipid and triglyceride content in fat bodies, and increased hemolymph carbohydrate level. These changes are accompanied by an impaired PI3K/Akt insulin signaling pathway as determined by a decrease in membrane-associated PtdIns3P and Akt activation in tissue extracts. Consistent with this phenotype, siRNA-induced depletion of SH2B in isolated insect cells inhibits insulin-induced tyrosine phosphorylation of Chico (the Drosophilia IRS homolog), Akt and Erk activation, as well as Drosophilia FoxO (dFoxO) phosphorylation and nucleo-cytoplasmic translocation. Systemic over-expression of dSH2B in Drosophilia results in a phenotype opposite to that of dSH2B depletion. However, whereas targeted expression of SH2B in fat bodies decreases lipid and glucose levels, neuron-specific expression of SH2B decreases life span and resistance to oxidative stress, indicating different physiological functions of dSH2B in these two tissues. Of note, the phenotype of SH2B-deficient Drosophilia differs from that of SH2B1-deficient mice, as life span and resistance to oxidative stress are increased in the former but decreased in the latter [150]. Thus, whereas the ability of dSH2B to improve insulin-like signaling, growth and metabolism is maintained in mammalian SH2B1, its ability to decrease resistance to oxidative stress and longevity is not.

Role of Grb10/14 and SH2B2 adaptors in insulin and IGF–1 receptor protein degradation

By mediating recruitment of the ubiquitin ligase Nedd4 to the IGFR, Grb10 has been implicated in IGFR ubiquitination and degradation [151, 152]. Interestingly, crystallographic studies showed that the Nedd4 binding site in the Grb10 SH2 domain is distant from the phosphotyrosine-binding pocket, thus enabling formation of a trimeric Nedd4–Grb10–IGFR complex [153]. In MEFS over-expressing Grb10 and Nedd4, the IGFR undergoes a ligand-dependent multi-ubiquitination and internalization through clathrin- and caveolin-dependent pathways [100]. Grb10 and Nedd4 are co-internalized with the IGFR, but, once in endosomes, the receptor is targeted to lysosomes for degradation [100]. Consistent with these results, over-expression of Grb10 induced by Nedd4 depletion leads to a decrease in IR/IGFR abundance at the cell surface and co-localization of Grb10 and the IGFR β-subunit in intracellular compartments [59].

Based on the ability of Grb10 over-expression to down-regulate the IR level in CHO–IR cells, and of Grb10 depletion to up-regulate it, Grb10 has also been implicated in IR ubiquitination and degradation, but the ubiquitin ligase responsible has not been identified [154]. The IR level in quadriceps muscle and adipose tissue is increased [39] or unchanged [40] in Grb10-deficient mice,, and the IR level in muscle and liver is unchanged in Grb14-deficient mice [37]. However, in Grb10/14 double knockout mice, the IR level in muscle and adipose tissue is increased twofold, suggesting a redundant role for these adaptors in negative regulation of IR expression [125]. So far, however, there is no evidence that IR down-regulation in insulin target tissues contributes to the ability of Grb10/14 to inhibit insulin signaling.

SH2B2 has been implicated in ligand-dependent IR ubiquitination and degradation by mediating recruitment to the IR of c–Cbl [155] and Asb6, a SOCS box protein that binds the elongin B–C ubiquitin ligase complex [156]. The involvement of c–Cbl in IR degradation was initially suggested by the ability of over-expressed SH2B2 to enhance insulin-dependent IR ubiquitination and c–Cbl phosphorylation in CHO–IR cells [155]. However, these findings were not confirmed in a more recent study, and SH2B2 over-expression did not affect internalization of cell-surface biotinylated IR [157]. Furthermore, in neither 3T3L1 adipocytes over-expressing SH2B2 nor major insulin-sensitive tissues of SH2B2-deficient mice was the IR level increased [52], arguing against a physiological role for SH2B2 in the regulation of IR protein expression.

Concluding remarks

Within the past years, IR and IGFR have been identified as important partners of Grb10/14 and SH2B1/B2 adaptors, and the mechanisms whereby these adaptors associate with receptors and differentially modulate receptor catalytic activity have been clarified. Some questions that remain to be answered are the spatial and temporal aspects of these regulations, the relative roles of Grb14 and PTP1B as negative regulators of IR/IGFR activity upon receptor internalization, and the ability of other tyrosine phosphatases, such as PTPα, PTPε and TC–PTP [124], to dephosphorylate the receptors.

Based on siRNA-based depletion studies in cultured cells, the predominant action of Grb10/14 on insulin/IGF–1 signaling is clearly inhibitory. The phenotypes of Grb10- and Grb14-deficient mice are also consistent with overlapping but non-redundant inhibitory effects of Grb10/14 on insulin sensitivity and glucose homeostasis, and with an inhibitory action of Grb10 on growth and skeletal muscle development. However, the heart and retina phenotypes of Grb14-deficient mice suggest that the effects of Grb14 are tissue-specific and not merely restricted to inhibition of insulin action. Conditional knockout models are required to better characterize Grb10/14 tissue-specific effects.

The phenotypes of SH2B1-deficient mice described in two independent series of studies are consistent with a positive regulation by SH2B1 of insulin/IGF–1 signaling, and, in one series, of leptin signaling. Insulin signaling and action in peripheral tissues appears to be regulated both indirectly by neuronal SH2B1 through the control of leptin-dependent, and possibly also insulin-dependent, adiposity, and directly by peripheral SH2B1; a major effect of the latter is to enhance insulin/IGF-dependent adipocyte differentiation.

Due to their co-expression in many tissues, Grb10 and Grb14 are expected to compete with each other, and with SH2B1 and/or SH2B2, for binding to IR/IGFR. Likewise, IR and IGFR compete with each other, and with other RTKs, for recruitment of Grb/SH2B adaptors. Selectivity, and therefore functional consequences, will depend on the relative expression levels and binding affinities of interacting partners. Due to the presence of a specific IR/IGFR-interacting domain in Grb10/14, the latter adaptors are expected to display a higher affinity for IR/IGFR than for other RTKs, and to be recruited preferentially by IR/IGFR compared to SH2B adaptors. Another potential determinant of Grb10/14 affinity for IR/IGFR is their state of serine/threonine phosphorylation, as suggested by the enhanced inhibitory effect of the phosphorylated forms on signaling.

Some effects of Grb10/14 and SH2B1/B2 on insulin/IGF signaling in cultured cells appear to involve their association with effectors downstream of the receptors. Additional studies are required to assess whether these interactions occur physiologically and to identify interacting partners involved in effects that are yet not well explained, such as the ability of Grb14 to inhibit SREBP–1c activation in hepatocytes.

The association of single nucleotide polymorphisms at the Grb10, Grb14 and SH2B1 gene loci with obesity and insulin resistance in humans warrants the development of therapeutic strategies focused on these adaptors, such as the design of compounds that are able to specifically inhibit the interaction of Grb10/14 with the IR. Given the recent identification of Grb10 as a negative regulator of muscular development and β–cell mass and function, this adaptor may be a potential therapeutic target for human wasting pathologies and diabetes.

Overall, the role of Grb10/14 and SH2B1 as regulators of insulin/IGF1 signaling and action is now well established. However, much remains to be done to identify new partners and functions of these adaptors, to assess the importance of IR/IGFR as physiological targets of SH2B1 relative to other RTKs, to identify the factors that regulate Grb10/14 and SH2B1 gene expression, and to assess the implications of altered adaptor function in human disease.