Asb6, an Adipocyte-specific Ankyrin and SOCS Box Protein, Interacts with APS to Enable Recruitment of Elongins B and C to the Insulin Receptor Signaling Complex* (cid:1)

The APS adapter protein plays a pivotal role in cou-pling the insulin receptor to CAP and c-Cbl in the phos-phatidylinositol 3-kinase-independent pathway of insu-lin-stimulated glucose transport. Yeast two-hybrid screening of a 3T3-L1 adipocyte library using APS as a bait identified a 418-amino acid ankyrin and SOCS (sup-pressor of cytokine signaling) box protein Asb6 as an interactor. Asb6 is an orphan member of a larger family of Asb proteins that are ubiquitously expressed. How-ever, Asb6 expression appears to be restricted to adipose tissue. Asb6 was specifically expressed in 3T3-L1 adipocytes as a 50-kDa protein but not in fibroblasts. In Chinese hamster ovary-insulin receptor (CHO-IR) cells Myc epitope-tagged APS interacted constitutively with FLAG-tagged Asb6 in the presence or absence of insulin stimulation and insulin stimulation did not alter the interaction. In 3T3-L1 adipocytes, insulin receptor activation was accompanied by the APS-dependent recruitment of Asb6. Asb6 did not appear to undergo tyrosine phosphorylation. Immunofluorescence and confocal microscopy studies revealed that Asb6 colocalized with APS in CHO cells and in 3T3-L1 adipocytes. In immunoprecipitation

Insulin promotes glucose transport into target tissues by causing the exocytosis of the glucose transporter GLUT4 from intracellular vesicles to the plasma membrane (1). There are thought to be two pathways required for insulin-stimulated glucose transport: an insulin receptor substrate-PI 1 3-kinase pathway and a PI 3-kinase-independent pathway involving APS, CAP, and c-Cbl. The PI 3-kinase-independent pathway is initiated by the binding of APS to the activation loop of the insulin receptor (2,3). APS is constitutively bound to CAP and undergoes tyrosine phosphorylation on Tyr 618 allowing it to bind to the variant SH2 domain of c-Cbl (3,4). This is followed by the tyrosine phosphorylation of c-Cbl on tyrosines 700 and 774 resulting in the phosphorylated c-Cbl binding to the SH2 domain of Crk (3,5). Following the insulin-stimulated phosphorylation of c-Cbl, the CAP/Cbl complex migrates to the caveolin-rich lipid rafts, a movement facilitated by the interaction of the CAP SOHO domain with flotillin, a protein in lipid rafts (6). This allows the Crk/C3G complex to be recruited to this microdomain, where C3G activates the small G protein TC10 (7)(8)(9)(10). The activation of TC10 occurs independently of PI 3-kinase and is crucial for insulin-stimulated Glut4 translocation (1).
The domain structure of APS includes proline-rich regions and a number of candidate tyrosine, serine, and threonine phosphorylation sites (11,12), implying that APS has the potential to generate a repertoire of novel protein-protein interactions involved in insulin-stimulated signal transduction pathways. We initiated a search for such proteins using the yeast two hybrid system and identified a member of the Asb family that is specifically expressed in adipocytes.
Yeast Two-hybrid Screening-The Matchmaker GAL4 Two-Hybrid System 3 (Clontech) was used to identify APS interacting proteins. Full-length APS was fused to the C-terminal of the GAL4 DNA binding domain in the pGBKT7 vector to generate the bait plasmid pGBK(APS). Standard yeast two-hybrid procedures were used to transform Saccharomyces cerevisiae strain AH109 initially with the pGBK(APS) bait, followed by transformation with a 3T3-L1 adipocyte library in * This work was supported by project grants from Diabetes UK and Novo Nordisk. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  pGAD-GH vector (kindly provided by Alan Saltiel, Life Sciences Institute). The resulting transformants were plated onto selective medium lacking tryptophan, leucine, and histidine and containing 2 mM 3-AT (3-amino-1,2,4-triazole) and incubated at 30°C until colonies formed. His ϩ colonies were picked after 10 and 15 days and streaked onto selective media lacking tryptophan, leucine, histidine, and adenine. An 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside filter lift assay was performed on the His ϩ Ade ϩ clones to identify expression of the ␤galactosidase reporter gene. Library plasmids were rescued from the His ϩ Ade ϩ LacZ ϩ clones and transformed into Escherichia coli XL10Gold (Stratagene) under ampicillin selection to select for the library plasmids. The library plasmids were sequenced, and nucleotide and amino acid sequences were compared with submitted sequences in the GenBank TM data base with the BLAST program.
The interacting library plasmids were isolated from the yeast transformed into XL10Gold E. coli and plated onto LB agar plates containing 100 g/ml ampicillin. A total of 1.05 ϫ 10 6 clones were screened. A total of eight specific interacting clones were identified, and all eight of the specific, positive interacting clones were subjected to automated dideoxy-DNA sequencing using GADF and GADR (forward and reverse primers for pGAD vectors). One of the clones, clone 10.16, showed complete identity with Asb6 and the clone was in frame with the GAL activation domain. The sequence of 10.16 includes the entire translated sequence, including the "ATG" start codon at residue 2153 of the Asb6 sequence.
Stable Expression of Asb6 -The cDNA encoding Asb6 was tagged with the FLAG epitope and cloned into the vector pIRES puro (Clontech) and the plasmid was transfected into CHO.T-APS cells using FuGENE. The cells were placed under puromycin selection 48 h later. Two weeks later, surviving colonies were isolated by limiting dilution and screened by anti FLAG blotting.
Generation of Asb6 Antibody-Purified GST-Asb6 protein was prepared as described above, dissolved in 1ϫ Laemmli sample buffer, electrophoresed, and a slice excised from the gel at 70 kDa. The gel slice was dried and used to generate antisera in two rabbits by Eurogentec (Herstal, Belgium). The resulting antisera was tested on 3T3-L1 adipocyte lysate separated on a SDS-PAGE with non-immune sera as a negative control.
Immunofluorescent Staining-Immunofluorescence and confocal imaging of Asb6 and APS in CHO cells or 3T3-L1 adipocytes were performed as described previously (2) using goat anti-APS (Santa Cruz Biotechnology) or rabbit anti-Asb6 and Alexa-Fluor 594 donkey-antigoat and Alexa-Fluor 488 anti-rabbit (heavy and light chains).

RESULTS
Yeast Two-hybrid Screening Identifies Asb6 as an Interactor-One of the clones identified in the yeast two-hybrid screen contained the complete coding sequence of an ankyrin and suppressor of cytokine signaling (SOCS) box protein Asb6, in frame with the GAL4 DNA binding domain (Fig. 1). The predicted amino acid sequence of Asb6 shows that it is a 418amino acid-long peptide consisting of a novel N-terminal region, followed by a series of six ankyrin repeats (aa 65-286) and a C-terminal SOCS box (aa 358 -413). The SOCS box contains the BC box consensus sequence. The predicted size of Asb6 is ϳ46.2 kDa, but on SDS gels, the protein displays a molecular mass of 50 kDa possibly as a result of phosphorylation. There are a number of consensus serine/threonine phosphorylation sites predicted by NetPhos (13).
Asb6 Is Specifically Expressed in Adipocytes-In a previous study by Kile et al. (14), the expression of a number of different Asb family members was examined by Northern blotting. Interestingly, there was no detectable Asb6 expression in all the tissues examined in the Northern blot suggesting that it has a highly restricted expression. Consistent with this, we could not FIG. 2. Expression of Asb6 in adipocytes. a, whole cell lysates from CHO cells co-expressing insulin receptor, APS and Asb6, and 3T3-L1 adipocytes were subjected to electrophoresis and immunoblotting with a rabbit polyclonal anti-Asb6 antibody and visualized using chemiluminescence. b, human adipose tissue lysates obtained from needle biopsies were analyzed by electrophoresis and immunoblotting with the same rabbit polyclonal anti-Asb6 antibody.
detect it in CHO cells or 3T3-L1 fibroblasts, but it was readily detectable using a specific anti-Asb6 antibody in transfected cells, 3T3-L1 adipocytes, and human adipose tissue biopsies (Fig. 2).
APS and Asb6 Interact in Transfected Cells-We analyzed the interaction of APS and Asb6 in intact cells using ectopic expression in CHO.T cells (Fig. 3a). Asb6 was tagged with the FLAG epitope and stably transfected into CHO cells stably expressing insulin receptor and APS (CHO.T-APS cells) (2, 4). Asb6 was then immunoprecipitated using anti-FLAG-agarose, and the immunoprecipitates were blotted with anti-APS anti-body. The results reveal that these proteins were constitutively bound in transfected cells, and insulin stimulation did not alter the interaction. Although a number of tyrosine-phosphorylated bands were observed in the immunoprecipitates (Fig. 3a, third  panel), these did not change with insulin, and Asb6 did not appear to undergo tyrosine phosphorylation. The SOCS box is known to function to bind Elongins B and C. We tested for the presence of Elongin C in the anti-FLAG immunoprecipitates and found that Elongin C was readily detectable in the immunoprecipitates of transfected CHO cells expressing Asb6 (Fig. 3a). FIG. 3. APS co-precipitates with Asb6 and Elongins in CHO cells or 3T3-L1 adipocytes. a, CHO.T cells co-expressing APS and Asb6 were serum-starved and then stimulated with insulin. Following cell lysis, FLAG-tagged Asb6 was immunoprecipitated using anti-FLAG-agarose, and the immunoprecipitates were analyzed by electrophoresis and immunoblotting with anti-FLAG antibody, anti-APS antibody, anti-phosphotyrosine RC20-horseradish peroxidase, and anti-Elongin C. Mouse IgG-agarose was used as a nonspecific control. Bound antibodies were visualized using luminol-based chemiluminescence. b, specificity and recognition of Asb6 by a rabbit antibody raised against Asb6. 3T3-L1 adipocyte lysates were subjected to electrophoresis and immunoblotting with two diffent anti-Asb6 antisera (at varying dilutions using a multichannel Decaprobe (Hoefer) apparatus (upper panel) and then anti-Asb6#2 was used for immunoprecipitation and immunoblotting in CHO.T-APS-Asb6 cells (lower panel). The cells were serum-starved, stimulated with insulin, and then lysed. Asb6 was then immunoprecipitated (lower panel) using either anti-FLAG or anti-Asb6 polyclonal. Non-immune serum was used as a nonspecific control. The whole cell lysates and immunoprecipitates were subjected to electrophoresis and immunoblotting with polyclonal anti-Asb6 antiserum. Bound antibody was visualized by chemiluminescence. c, activation of the insulin receptor in 3T3-L1 adipocytes leads to recruitment of APS, Asb6, and Elongins B/C. The insulin receptor was activated with dephostatin and then precipitated using insulin-agarose. The precipitates were divided and analyzed by electrophoresis and immunoblotting for insulin receptor (Anti-IR), phosphotyrosine (RC20HRP), APS (Anti-APS), Asb6 (Anti-ASB6) and Elongins B and C. Elongins B/C were probed for simultaneously. WCL, whole cell lysates; Nim, non-immune (preimmune) serum.

ASB6 and APS Interact in Adipocytes and Elongins B and C
Are recruited-In 3T3-L1 adipocytes, we used a rabbit anti-Asb6 antibody to study the interaction, since the endogenous protein was not tagged. The specificity of the antibody was tested both for immunoblotting and immunoprecipitation in adipocytes (Fig. 3b, upper panel) and CHO.T-APS-Asb6 cells (Fig. 3b, lower panel). Next, the insulin receptor was activated using dephostatin (15), and the insulin receptor was isolated using insulin-agarose to enhance the efficiency of precipitation, since many insulin receptor monoclonals do not bind to mouse receptors with high affinity. Furthermore, this was done to minimize the potential of IgG cross-reaction, since Asb6 comigrates with the immunoglobulin heavy chain at 50 kDa. As shown in Fig. 3c, activation of the insulin receptor, demonstrated by antiphosphotyrosine immunoblotting, was accompanied by recruitment of APS and Asb6. In addition, Elongins B and C could be detected in the immunoprecipitates after insulin receptor activation.
APS and Asb6 Co-localize in Cells-Using immunofluorescence and confocal imaging, we analyzed the cellular distribution of APS and Asb6 in transfected CHO cells and 3T3-L1 adipocytes. In CHO cells, APS and Asb6 were labeled using antiFLAG and antiMyc antibodies. Fig. 4). In 3T3-L1 adipocytes, endogenous APS and Asb6 were visualized using goat anti-APS and rabbit anti Asb6 (Fig. 5). In both cell types, co-localization of APS and Asb6 was observed when the immunofluorescent images were merged.
Mapping the Regions of APS and Asb6 That Interact-The regions of APS and Asb6 that interact were mapped using pull-downs with wild-type and deletion mutants of GST fusion proteins. To establish the region of Asb6 that mediates the binding to APS a series of deletions of the GST-Asb6 constructs were made and purified. The series of immobilized GST-Asb6 deletions were used for GST "pull-downs" of APS derived from CHO.T-APS cells (Fig. 6), with GST alone used as a negative control. The panel shows that APS is precipitated by the fulllength Asb6, Asb6(aa 1-275) and Asb6(aa 123-418), and to a lesser extent to Asb6(aa 69 -418) and Asb6(aa 269 -418), with no APS binding to Asb6(aa 1-82) and Asb6(aa 1-133) detected. Similar levels of GST-Asb6 fusion protein deletions and negative GST control were used for the pull-down. These results indicated that the region of Asb6 primarily responsible for binding to APS lies between amino acids 123 and 275.
Similarly, regions of APS were expressed as GST fusions, and the purified fusion proteins were incubated with cell lysates derived from CHO.T-Asb6 cells in a "pull-down assay" (Fig. 7). The precipitates were then probed by immunoblotting with anti-FLAG antibody. The region of APS that interacts with Asb6 is contained in amino acids 117-466.
Asb6 Undergoes Insulin-stimulated Degradation along with APS-Since Asb6 contains a SOCS box that binds to Elongins, it is likely that the recruitment of Asb6 facilitates degradation of associated proteins. To test this hypothesis, we examined the effects of prolonged insulin stimulation on the degradation of Asb6 and APS in CHO cells (Fig. 8). Serum-starved CHO cells were stimulated with varying concentrations of insulin overnight and the lysates were analyzed for the levels of Asb6 and APS. Insulin caused a dose-dependent decrease in the levels of both Asb6 and APS.

DISCUSSION
In a search for novel interactors for the APS adapter protein expressed in adipose tissue, we screened a 3T3-L1 adipocyte library using the yeast two-hybrid system. One of the interactors we identified is the Asb6 protein (16). Asb6 is specifically expressed in 3T3 L1 adipocytes and human fat tissue but not in fibroblasts. In an analysis of Asb1 knock-out mice, Asb6 was not expressed in any of the tissues examined by Kile et al. (14), suggesting that it is highly selectively expressed. The interaction of Asb6 and APS was confirmed using several techniques including GST fusion protein pull-downs and co-precipitation in both transfected and native cells expressing endogenous levels of protein. Finally, co-localization studies with immunofluorescence and confocal microscopy in both overexpressing and CHO cells and 3T3-L1 adipocytes confirmed the cellular interaction of APS and Asb6. Since neither APS nor Asb6 are expressed in commonly used cultured cells, it was necessary to overexpress them in a heterologous expression system to analyze the interaction in detail.
Ankyrin repeats are found in more than 400 different proteins that contain from 2 to more than 20 tandemly arrayed repeat units. In each case where ankyrin repeat structures have been examined, they have been shown to comprise helixturn-helix motifs linked together by loops. There is evidence that the loops are sites for protein-protein interactions for which ankyrin repeats provide a stable platform (17). SOCS boxes are ϳ40 amino acids long and are found in five distinct families of proteins (16). The SOCS boxes bind to Elongins B and C, potentially acting as an adaptor to couple proteins to the ubiquitination or proteosomal compartments (18,19). There have been studies into Asb1 knock-out mice that showed an increase in testicular anomalies; mice that overexpressed fulllength and truncated forms of Asb1 all showed normal devel-opment, indicating some sort of redundancy between protein family members (14) The amino acid structure of Asb6 predicts a 418-residue peptide, made up of a novel N-terminal region, followed by a series of six ankyrin (Ank) repeats (aa 31-256) and a C-terminal SOCS box (aa 358 -413) domain. By using deletions of GST-Asb6 fusion proteins the region of Asb6 involved in the interaction with APS lies within the C-terminal region of the Ank repeat region (aa 123-275). This is consistent with the proposed role of Ank repeats being involved as a stable platform for protein-protein interactions (17). The SOCS box was first identified in the protein SOCS-1 (20), and it was demonstrated that overexpression of SOCS-1 inhibited both interleukin-6-induced receptor phosphorylation and STAT activation. The SOCS family of proteins consist of variable N-terminal regions, followed by a central SH2 domain and a C-terminal SOCS box. The other members of the family that possess a SOCS box differ from the SOCS family proteins in the domains upstream of this motif and have been characterized accordingly. In place of the SH2 domains the Asb proteins contain ankyrin repeats, the WD-40 SOCS boxes contain WD-40 repeats, the SPRY SOCS boxes contain SPRY domains, and the Ras-like proteins contain GTPase domains. It has been demonstrated that the SOCS box associates with Elongins B and C (18,19). The region within the SOCS box responsible for binding Elongins B and C has been termed the BC box consisting of a consensus amino acid sequence ((A,P,S,T)LXXXCXXX-(A,I,L,V)), previously identified in Elongin A and von Hippel-Lindau (VHL) tumor suppressor protein (21)(22)(23). The BC box consensus sequence is present in Asb6 (aa 372-381) (PLKHL-CRVSI), and this can be predicted to be involved in the binding of Elongins B and C. There are established precedents for the role of SOCs proteins in regulating components of the insulin signaling pathway (24 -27), and this is another example.
Initially, the Elongin BC complex was identified as a positive regulator of RNA polymerase II elongation factor Elongin A (28). The Elongin BC complex has also been identified as a component of the multiprotein VHL tumor suppressor protein complex (22,23). The VHL tumor suppressor protein complex is an E3 ubiquitin ligase (29) that targets the ␣-subunits of the hypoxia-inducible transcription factors HIF1 and HIF2 for ubiquitination (30). The C-terminal domain of VHL protein, homologous to the SOCS box, interacts with Elongin C. VHL and Elongin C are found in complex with Elongin B, the Cullin protein family member Cul-2, and RING finger protein called Roc1 or Bbx1. This complex has E3 ubiquitin ligase activity (17). For SOCS-1 there is now direct evidence that it can act as the substrate recognition component of an E3 ubiquitin ligase complex to regulate the half-life of Vav (31)and JAKs (32,33). From these results, it can be postulated that Asb6 may play an analogous role in the ubiquitination of APS or additional proteins involved in the APS-Asb6 complex, possibly including the IR. The Ank repeat region appears to provide a specific binding site for APS, which was not affected by insulin-stimulated tyrosine phosphorylation of APS, and the SOCS box of Asb6 can be predicted to recruit the Elongin BC ubiquitin-ligase complex.
It is interesting to note that the APS knock-out mice (34) show a gain of function, and it is therefore tempting to specu-late that this is due to loss of the negative regulatory function of Asb6 associated with APS.
In summary (Fig. 9), we have identified the ankyrin and SOCs box protein Asb6, along with the Elongin BC complex, as a component of the insulin receptor signaling complex recruited by the APS adapter protein in 3T3-L1 adipocytes and ectopically transfected CHO cells. We have satisfied the criteria for a bona fide interaction using co-precipitation and co-localization in both transfected and untransfected native cells. We have also demonstrated a functional consequence of the interaction that is likely to be of physiological relevance. Furthermore, we have identified a function for an orphan member of the Asb family, which has hitherto not been suspected. Previously we and others (2,3,35) have shown that APS plays a positive role in insulin signaling. The identification of a SOCs box protein recruited by APS would suggest that APS may also play a negative role in regulating signal transduction by enabling the recruitment of Asb6 and the Elongins.