Tankyrase modulates insulin sensitivity in skeletal muscle cells by regulating the stability of GLUT4 vesicle proteins

Tankyrase 1 and 2, members of the poly(ADP-ribose) polymerase family, have previously been shown to play a role in insulin-mediated glucose uptake in adipocytes. However, their precise mechanism of action, and their role in insulin action in other cell types, such as myocytes, remains elusive. Treatment of differentiated L6 myotubes with the small molecule tankyrase inhibitor XAV939 resulted in insulin resistance as determined by impaired insulin-stimulated glucose uptake. Proteomic analysis of XAV939-treated myotubes identified down-regulation of several glucose transporter GLUT4 storage vesicle (GSV) proteins including RAB10, VAMP8, SORT1, and GLUT4. A similar effect was observed following knockdown of tankyrase 1 in L6 myotubes. Inhibition of the proteasome using MG132 rescued GSV protein levels as well as insulin-stimulated glucose uptake in XAV939-treated L6 myotubes. These studies reveal an important role for tankyrase in maintaining the stability of key GLUT4 regulatory proteins that in turn plays a role in regulating cellular insulin sensitivity.

Insulin regulates glucose uptake into both muscle and fat cells by stimulating the movement of the facilitative glucose transporter GLUT4 from intracellular storage vesicles called GLUT4 storage vesicle (GSV) 3 to the plasma membrane (PM) (1). Insulin triggers GLUT4 translocation via the canonical phosphatidylinositol 3-kinase/Akt signal transduction pathway, including the phosphorylation of the Akt substrate AS160, a Rab GTPase-activating protein (RabGAP) that is localized to GSVs (1)(2)(3)(4)(5). Phosphorylation of AS160 is thought to regulate the GTP loading of the cognate Rab, RAB10, to initiate translocation of GSVs to and fusion with the PM (1, 6 -8).
Detailed molecular analyses of GLUT4 translocation have focused on the molecular composition of the GSVs in which GLUT4 is stored under basal conditions. This has led to the identification of the aminopeptidase IRAP, the sorting protein sortilin that plays a key role in the biogenesis of GSVs (9 -15), the v-SNAREs VAMP2, VAMP3, and VAMP8 that play a role in docking and fusion of GSVs with the PM (16 -18) and several Rab GTPases including Rab10 (3).
Another protein found to be associated with GSVs via its interaction with IRAP is tankyrase (19). Tankyrase 1/2 are members of the poly(ADP-ribose)polymerases (PARPs) that catalyze poly(ADP)-ribosylation or PARsylation, a protein post-translational modification involving addition of a large number of linear and/or branched ADP-riboses onto target proteins (20). PARsylated proteins can be targeted for ubiquitination and proteasomal degradation by the E3 ubiquitin ligase RNF146 that interacts with and is allosterically activated by the PAR chain (21,22).
Tankyrase 1/2 have been implicated in insulin-stimulated glucose uptake in adipocytes (23). Insulin enhances tankyrase activity possibly via mitogen-activated protein kinase-mediated phosphorylation (19). Tankyrase 1 knockdown or tankyrase inhibition using the broad-range PARP inhibitor PJ34 resulted in impaired insulin-stimulated GLUT4 translocation to the PM or insulin resistance in adipocytes (23). The mechanism by which tankyrase affects insulin-stimulated GLUT4 translocation has not been established (23). Furthermore, the role of tankyrase in skeletal muscle, the tissue that plays a central role in whole body insulin-mediated glucose uptake, is not known.
The development of more selective tankyrase inhibitors, such as XAV939 (24,25), has been the focus of anti-cancer therapeutics in an effort to inhibit aberrant Wnt/␤-catenin signaling (26). Tankyrase PARsylates AXIN, which is part of the ␤-catenin destruction complex, and targets it for degradation, thereby leading to the stabilization and activation of the ␤-catenin pathway (25).
In this study, we utilized XAV939 (24,25) to inhibit tankyrase 1/2-mediated PARsylation in L6 myotubes and showed that this impaired insulin-stimulated glucose uptake. We applied label-free quantitative proteomics (27) to determine XAV939mediated changes in the proteome and discovered that tankyrase inhibition resulted in a reduction in protein levels of a number of GSV proteins. A similar reduction in GSV proteins was observed when tankyrase 1 was knocked down using a tankyrase 1-specific siRNA. Proteasome inhibition reversed the XAV939-mediated effect on glucose uptake and GSV protein levels in L6 myotubes. These data show an important role for tankyrase in glucose metabolism in skeletal muscle cells and provide new insights into the mechanism of insulin resistance mediated by tankyrase dysfunction.

XAV939 inhibited tankyrase activity in L6 skeletal muscle cells
To determine the role of tankyrase in glucose uptake in muscle, the small molecule tankyrase inhibitor XAV939 (24,25) was used to inhibit tankyrase 1/2 in L6 myotubes. To confirm its efficacy in these cells, we incubated differentiated L6 myotubes with different doses of XAV939 for various times and assessed tankyrase inhibition (Fig. 1A). XAV939 treatment increased tankyrase 1/2 protein levels in a dose-and time-dependent manner, as previously shown (25), confirming that XAV939 inhibited tankyrase auto-PARsylation and subsequent degradation. Consistent with efficient tankyrase inhibition with XAV939 treatment, a dose-and time-dependent decrease in ␤-catenin protein levels was observed (Fig. 1A). However, PTEN abundance was not affected by XAV939 treatment in skeletal muscle cells (Fig. 1A), which was inconsistent with a previous study performed in cancer cell lines showing that upon tankyrase 1/2 knockdown, protein levels of the tankyrase substrate PTEN were increased (28). This suggests that regulation of PTEN protein levels may be cell type-specific. Global protein PARsylation in response to XAV939 treatment was also assessed (Fig. 1B). L6 myotubes were treated with or without XAV939, followed by enrichment of PARsylated proteins and immunoblotting with anti-poly(ADP-ribose) (PAR) antibodies and a decrease in global PARsylated proteins was observed with XAV939 treatment (Fig. 1B), confirming that XAV939 inhibits tankyrase 1/2-mediated PARsylation in skeletal muscle cells.

XAV939 treatment impaired glucose uptake
We next examined the effects of XAV939 on glucose uptake in differentiated L6 myotubes. We observed a significant dosedependent defect in insulin-stimulated 2-deoxyglucose (2-DG) uptake after 24 and 48 h of XAV939 treatment but not after 6 h (Fig. 1C), indicating that tankyrase inhibition in L6 myotubes results in insulin resistance, consistent with previous reports in adipocytes (23). XAV939 treatment for 24 -48 h also significantly affected basal glucose uptake in these cells. We next examined whether this defect in glucose uptake upon tankyrase inhibition was due to a defect in insulin signaling. L6 myotubes were incubated with or without XAV939 (1 M, 24 h) and a range of insulin doses and insulin signaling was assessed by immunoblotting (Fig. 1D). XAV939 treatment had no effect on insulin-stimulated phosphorylation of Akt (pThr-308, pSer-473) or its substrate AS160 (pThr-642), indicating that the impairment in 2-DG uptake upon tankyrase inhibition was not due to defective insulin signaling. Taken together, these data indicate that XAV939-mediated tankyrase inhibition resulted in insulin resistance in L6 myotubes.

Proteomic analysis of tankyrase inhibition in skeletal muscle cells
To ascertain which proteins were involved in this negative regulation of glucose metabolism, we next used MS-based quantitative proteomics to study the total proteome in this cellular model. We incubated L6 myotubes with or without XAV939 (1 M, 24 h) and subjected the samples to label-free quantitative MS analysis. This led to the identification of 4,387 proteins and quantification of 4,374 proteins ( Fig. 2A, Table S1) using MaxQuant as described under "Experimental procedures" (27). Of these, 4,096 proteins were quantified in at least 2 replicates of both control and XAV939-treated samples ( Fig.  2A, Table S2). The samples presented very consistent protein intensity distribution and were highly correlated (Ͼ0.98).
Data analysis revealed 103 down-regulated proteins and 104 up-regulated proteins in response to XAV939 treatment using a 1.5-fold change threshold or exclusively quantified in one condition (Tables S3 and S4). Consistent with our immunoblotting results (Fig. 1A), ␤-catenin was down-regulated (0.66-fold, p value ϭ 0.01) in response to XAV939 treatment, and tankyrase 2 was up-regulated (only quantified in all 3 XAV939-treated samples), thereby validating the approach. We next performed gene ontology overrepresentation analysis on the 207 differentially expressed proteins. A number of biological processes were significantly enriched upon tankyrase inhibition, including "vesicle-mediated transport," which is implicated in glucose transport and GLUT4 exocytosis, the process which is defective upon tankyrase inhibition in L6 myotubes (Fig. 1C). Interestingly, 3 of the XAV939-regulated proteins that belong to the process "vesicle-mediated transport" are associated with GSVs (VTI1B, VAMP8, and RAB10) (3). Of particular interest were VAMP8 and RAB10, as both proteins have been shown to play an integral role in insulin-stimulated GLUT4 translocation (7, Figure 1. XAV939 treatment impaired glucose uptake in muscle. L6 myotubes were incubated with DMSO (Ctrl) or the indicated doses and times of XAV939 (XAV), followed by serum starvation and incubation with or without 100 nM insulin (C) or as indicated (D). A, DMSO or XAV939-treated myotubes were lysed and immunoblotted with the indicated antibodies (14-3-3 was used as a loading control). B, DMSO or XAV939-treated (1 M, 24 h) myotube lysates were enriched for PARsylated proteins and subjected to immunoblotting with indicated antibodies (14-3-3 was used as a loading control in whole cell lysate (WCL)). C, DMSO or XAV939-treated cells were subjected to [ 3 H]2-DG uptake experiments. Data are mean Ϯ S.E., n ϭ 6 independent experiments; *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001 compared with corresponding DMSO control. D, DMSO or XAV939-treated myotubes were lysed and and immunoblotted with the indicated antibodies (14-3-3 was used as a loading control). 18,29) and both were down-regulated upon XAV939 treatment (VAMP8, 0.63-fold and RAB10, 0.56-fold).

GSV protein analysis
Insulin triggers glucose transport in fat and muscle cells by initiating the movement of intracellular GLUT4-containing storage vesicles (GSVs) to the PM and fusion with the PM. Tankyrase inhibition resulted in impaired insulin-stimulated GLUT4 translocation in L6 myotubes (Fig. 1C) and it was of particular interest to observe that the protein levels of two crucial GSV proteins, VAMP8 and RAB10, were down-regulated in response to XAV939. We therefore focused our analysis on GSV proteins in the proteomics dataset. We filtered our data for previously reported GSV proteins (1,3,12,30,31) and identified 79 GSV proteins in our dataset (Fig. 2B, Table S5). Among these, RAB10, VAMP8, and RRAS were down-regulated and VTI1B was up-regulated in response to XAV939 treatment (Fig. 2B). To verify these findings and to examine additional GSV proteins not quantified by MS, we subjected control and XAV939-treated L6 myotube lysates to immunoblotting using specific antibodies for various GSV proteins. A number of crucial GSV proteins were reduced upon XAV939 treatment, including RAB10, VAMP8, SORT1, and GLUT4 (Fig. 2, C and D). Intriguingly, other bona fide GSV proteins, including VAMP2 and IRAP, were not modulated following inhibition of tankyrase indicating that this phenomenon was not simply en bloc degradation of GSVs.

Validation of the role of tankyrase on GSV protein levels in skeletal muscle cells
To validate the above findings using XAV939 we next utilized another tankyrase inhibitor, the broad-spectrum PARP inhibitor PJ34 (23). Consistent with XAV939, PJ34 (40 M) caused an acute and significant reduction in 2-DG uptake after just 1 h of treatment and this suppression was more pronounced after 2 h (Fig. 3A). Insulin-stimulated Akt phosphorylation was reduced by PJ34 treatment at both Akt phosphorylation sites at a dose of 40 M PJ34 (Fig. 3B). However, there was no reduction in Akt activity as assessed by phosphorylation of the Akt substrate AS160, which was not reduced by PJ34 treatment. Although it is not clear why Akt phosphorylation was affected by PJ34 treatment, these data suggest that Akt signaling was normal despite the reduction in its phosphorylation and it is likely that the residual Akt activity was sufficient as very little Akt activity is required for maximal downstream signaling (32). Notably, PJ34 treatment affected the protein levels of the GSV proteins RAB10 and GLUT4 similar to XAV939 treatment (Fig. 3, C and D).

Tankyrase regulation of insulin sensitivity
therefore tested whether the observed phenotypes in response to drug treatment were due to a specific effect on tankyrase by knocking down tankyrase 1. L6 myotubes were transfected with pooled TNKS1 siRNAs or scrambled siRNA for 24 or 48 h, resulting in efficient knockdown of TNKS1 (ϳ90%) but not TNKS2 (Fig. 4, A and C). TNKS1 knockdown resulted in decreased ␤-catenin protein levels confirming that tankyrase activity was abolished and also that TNKS1 is likely the major tankyrase in L6 myotubes (Fig. 4, C and D). Basal and insulinstimulated 2-DG uptake was significantly impaired after 48 h of TNKS1 knockdown (Fig. 4B), confirming that TNKS1 knockdown reproduced the results obtained with tankyrase inhibition using 2 independent drugs in L6 myotubes. Notably, the protein levels of several GSV proteins, including RAB10, GLUT4, and SORT1, were significantly reduced upon TNKS1 knockdown (Fig. 4, C and D). However, TNKS1 knockdown affected insulin signaling, resulting in reduced phosphorylation of Akt and the Akt substrate AS160 (Fig. 4E). Hence it is not clear whether the impairment in 2-DG uptake was due to reduced GSV proteins or defective Akt signaling or a combination of both. Nevertheless, these data confirmed an important role for tankyrase in insulin-stimulated glucose uptake and GSV protein stability in L6 myotubes. Furthermore, these data demonstrated that TNKS1, rather than TNKS2, was the key regulator for these processes.

Proteasome inhibition reversed XAV939-mediated defects in GSV protein levels and glucose uptake in skeletal muscle cells
Our data show that tankyrase inhibition affected GSV protein stability and glucose uptake in L6 myotubes. In light of the known role of tankyrase activity in proteasomal protein degradation (21,22) we next tested the role of the proteasome in the loss of GSV proteins following inhibition of tankyrase activity. To achieve this, we utilized the proteasome inhibitor MG132 (33). MG132 treatment reversed XAV939-induced down-regulation of ␤-catenin as well as that of GSV proteins, including RAB10, GLUT4, SORT1, and VAMP8, indicating that proteasomal degradation was responsible for the XAV939-mediated reduction of these proteins (Fig. 5, A and B). Furthermore, 6 h of MG132 treatment also reversed the XAV939-mediated impairment in 2-DG uptake (Fig. 5C). Notably, treatment with MG132 alone for 6 h significantly increased GLUT4 protein levels as well as both basal and insulin-responsive glucose uptake to a similar extent to that observed with the XAV939 reversal. As expected, MG132 treatment did not affect insulindependent phosphorylation of Akt or AS160 (Fig. 5D). Furthermore, MG132 had no effect on tankyrase protein levels upon siRNA-mediated knockdown as MG132 acts post-translationally. In summary, these data suggest that glucose uptake is highly correlated to GSV protein levels and that tankyrase inhibition significantly accelerates GSV protein turnover.

Tankyrase protein levels were reduced in skeletal muscle from insulin-resistant mice
Given that tankyrase inhibition resulted in insulin resistance in L6 skeletal muscle cells, we next determined whether tankyrase levels or its activity were affected in skeletal muscle from insulin-resistant mice. C57Bl/6J mice fed a high-fat/highsucrose diet (HFHSD) for 6 weeks were insulin-resistant as displayed by impaired glucose tolerance (Fig. 6A). Tankyrase 1 protein levels were significantly reduced in skeletal muscle from HFHSD-fed mice compared with chow-fed mice (Fig. 6, B and C). However, protein levels of ␤-catenin or GSV proteins in skeletal muscle were unaffected by the HFHSD, suggesting that the reduction in tankyrase, although significant, was not sufficient for downstream effects.

Discussion
Over the years, tankyrase has been implicated in many biological processes including carbohydrate metabolism. However, it remains unclear exactly how tankyrase mediates these effects on glucose homeostasis or if it does so in skeletal muscle.

Tankyrase regulation of insulin sensitivity
In the present study, we used two different pharmacologic inhibitors of tankyrase (XAV939, PJ34) as well as siRNA knockdown of TNKS1 to show that in L6 myotubes, tankyrase has a profound effect on regulating insulin sensitivity. Although these 3 ways of modulating tankyrase differentially affected insulin signaling, they all had in common the targeted degradation of GSV proteins that include GLUT4 itself and RAB10, and in some cases also included SORT1 and VAMP8. The proteasome appears to play a role in this effect as inhibition of the proteasome rescued the impairments in glucose uptake and GSV protein levels that were triggered upon tankyrase inhibition.
In adipocytes, PARP inhibition (PJ34) or TNKS1 knockdown also resulted in reduced insulin-stimulated GLUT4 translocation to the PM (23), indicating a similar role for tankyrase in both insulin-sensitive cell types. However, whereas in muscle cells the primary defect appeared to be on reduced levels of GSV proteins, in adipocytes there was a redistribution of GLUT4 within the cell possibly into a non-insulin-responsive compartment. One possibility that we favor is that these two mechanisms may be related. One model that explains both observations is that tankyrase activity is required to maintain the integrity or stability of GSVs by, for example, blocking their fusion with other intracellular compartments such as endosomes or lysosomes. Inhibition of tankyrase may overcome this step leading initially to redistribution of GLUT4, the principle phenotype observed in adipocytes (23), commensurate with subsequent degradation of GSV proteins, the main effect seen in muscle cells. The difference in each cell type may simply reflect differences in the kinetics of transit through these different compartments in each cell type. It was intriguing that not all GSV proteins appeared to behave in the same way in muscle cells in that IRAP and VAMP2 did not undergo tankyrase-dependent degradation indicating that certain cargo may escape sorting into the degradative pathway.
The 4 GSV proteins that showed reduced protein levels upon tankyrase inhibition in muscle all play crucial roles in GLUT4 trafficking, aside from being localized to GSVs. GLUT4 itself is obviously the most important GSV protein and responsible for insulin-stimulated glucose uptake in myotubes. The vesicle-associated membrane protein (VAMP) VAMP8 is a v-SNARE that regulates fusion of vesicles with the PM and it has previously been implicated in insulin-regulated GLUT4 trafficking in adipocytes (18). Most notably, it has been observed that simultaneous disruption in the expression of VAMP2, VAMP3, and VAMP8 is required to completely block insulin-stimulated GLUT4 translocation to the PM in adipocytes and this defect could be rescued by re-expression of any one of these three VAMPs (18), indicating an important role for VAMP8 in this process. Furthermore, whole body deletion of VAMP8 affected

Tankyrase regulation of insulin sensitivity
glucose metabolism and glucose uptake in skeletal muscle in mice (34). RAB10 is a GTPase localized on GSVs (3) that is a substrate for the RabGAP AS160 (4) and its knockdown in adipocytes has a profound effect on insulin sensitivity (7,8,35). Furthermore, mice with adipose-specific deletion of RAB10 showed impaired whole body glucose homeostasis (29). Although other Rabs have been implicated in GLUT4 trafficking in L6 cells, including RAB8A, RAB13, and RAB14 (36,37), these studies were conducted in undifferentiated myoblasts and it is possible that RAB10 only has a crucial role in differentiated cells as is the case in adipocytes (35). SORT1 has been implicated in sorting of GLUT4 into GSVs in adipocytes (11,13) and myotubes (38) as well as GLUT4 retrograde traffic from endosomes to the trans-Golgi network thereby preventing GLUT4 traffic to lysosomes and subsequent degradation (39). So, the concerted loss of each of these proteins is likely to have a profound effect on insulin sensitivity.
It has been reported that the Wnt/␤-catenin signaling pathway plays a role during myogenesis (40). Because ␤-catenin was markedly reduced in XAV939-treated myotubes, one possibility was that this affected myotube differentiation thereby causing the reduction in GSV protein expression. We therefore assessed whether any myogenesis or skeletal muscle proteins were changed in L6 myotubes in response to XAV939 treatment. Although some myogenesis markers were not identified in our dataset (Myf5, MyoD, and myogenin), a number of skeletal muscle-specific proteins were quantified in our dataset and were not significantly changed in response to XAV939 treatment, including skeletal muscle contractile apparatus proteins (Acta1, 1.11-fold change XAV939/control; Mylpf, 1.01; Tnni1, 1.10; Tnni2, 1.09; Tnnt3, 0.96) (41) as well as skeletal muscle glucose metabolism proteins (Gys1, 1.11; Pfkm, 1.05; Phka1, 0.90; Pygm, 0.87) (42). These data indicate that XAV939 treatment had no effect on myotube differentiation and the reduction in GSV proteins is due to direct or indirect effects of tankyrase inhibition.
One question that arises from these studies is how these GSV proteins are stabilized upon proteasome inhibition? It seems unlikely that the transmembrane domain containing GSV proteins such as GLUT4, SORT1, or VAMP8 undergo proteasomal degradation as at least GLUT4 is principally degraded via lysosomes (43,44). One possibility is that tankyrase inhibition induces the proteasomal degradation of just one of these proteins, like RAB10, which results in the mis-targeting of all GSV proteins and their lysosomal degradation. Notably, GLUT4 protein levels are reduced upon knockdown or deletion of several cytoplasmic proteins, including AS160, its close homologue TBC1D1 and the retromer complex (39,45,46). In this instance, blockade of proteasomal activity would prevent degradation of these regulatory factor(s), thus overcoming this deleterious mechanism. It is unlikely that this regulatory factor is targeted for proteasomal degradation via tankyrase-mediated PARsylation because upon tankyrase inhibition the protein levels of tankyrase substrates would be increased rather than decreased. Altogether, these data suggest that the reduction in GSV protein levels upon tankyrase inhibition is likely due to an indirect effect of tankyrase inhibition involving proteasome activity. The slower reversal of insulin-regulated glucose uptake (6 h) compared with the GSV protein levels (2 h) likely reflects the slower reassembly of all GSV proteins into functional insulin-responsive vesicles, a process that may well require several hours under these conditions.
Given the prominent role of tankyrase in targeting its substrates for degradation, it was somewhat surprising that of the Ͼ200 regulated proteins identified upon XAV939 treatment, about 50% were down-regulated indicating indirect effects of tankyrase inhibition as is the case with ␤-catenin where upon tankyrase inhibition, its negative regulator Axin is stabilized, thereby leading to ␤-catenin degradation (25). Gene ontology enrichment of the XAV939-regulated proteins identified "cell proliferation" as the most significantly down-regulated process, which is consistent with the role of tankyrase in Wnt/␤-catenin signaling and cell cycle, thus justifying the efforts spent on development of more specific tankyrase inhibitors for cancer treatment. Other proteins that were changed upon XAV939 treatment included proteins involved in various kinds of posttranslational modifications, e.g. MINK1, DDRGK1, AKAP11, MNAT1, and thioredoxin2 (TXN2). Changes in their protein expression may affect the downstream signal transduction cascades via phosphorylation or redox events, implying potential cross-talk between PARsylation and other post-translational modifications. This is interesting in light of the effects of tankyrase 1 knockdown on Akt signaling, showing reduced insulin-stimulated Akt and AS160 phosphorylation. This is in contrast to 3T3-L1 adipocytes, where tankyrase 1 knockdown resulted in impaired insulin-stimulated glucose uptake in the absence of a signaling defect (23). One reason for this difference might be that L6 myotubes are less insulin sensitive than 3T3-L1 adipocytes (47,48). Tankyrase knockdown was the only Figure 6. Tankyrase 1 protein is reduced in skeletal muscle tissue from insulin-resistant mice. C57Bl/6J mice were fed a chow diet (Ch) or high-fat/ high-sucrose diet (HFHSD or HF) for 6 weeks. A, an intraperitoneal GTT at a glucose dose of 2 g/kg of lean mass was performed on the mice after 5 weeks on the diet and blood glucose was measured during the GTT. B, mice were euthanized after 6 weeks and skeletal muscle (quadriceps) was isolated, lysed, and immunoblotted with the indicated antibodies. 14-3-3 was used as a loading control. C, quantification of immunoblots in B of the indicated proteins is shown; data are mean Ϯ S.E.; n ϭ 4; *, p Ͻ 0.05 compared with chow.

Tankyrase regulation of insulin sensitivity
tankyrase modulating approach that showed impaired Akt activity as neither of the inhibitors showed a defect in phosphorylation of the Akt substrate AS160. The reduction in Akt phosphorylation observed with PJ34 had no effect on Akt activity consistent with the fact that little Akt phosphorylation is required for maximal phosphorylation of Akt substrates (32). The reason for the discrepancy in insulin signaling between tankyrase inhibition and knockdown might be that in the case of the inhibitors, the tankyrase protein, whereas inactive, is still present or rather increased, suggesting that tankyrase may have activity-independent effects on insulin signaling at least in L6 myotubes. Even though the different approaches of tankyrase modulation (knockdown and 2 inhibitors) showed different effects on Akt signaling, all 3 approaches converged on insulin resistance as shown by reduced 2-DG uptake as well as a reduction in GSV protein levels, the latter being the likely cause of the observed insulin resistance upon tankyrase inhibition in these cells.
Although GLUT4 levels are reduced in adipose tissue of insulin-resistant mice as well as of humans with T2D (49,50) this is not the case in skeletal muscle (51). Consistent with this GLUT4 protein levels were normal in insulin-resistant muscle from HFHSD-fed mice. Interestingly, tankyrase 1 protein levels were significantly reduced in insulin-resistant muscle, albeit with no effect on protein levels of GSV proteins or ␤-catenin, suggesting that the reduction in tankyrase 1 was not sufficient for downstream effects. Nevertheless, these data show that tankyrase 1 protein levels were impaired in insulin-resistant muscle tissue. Notably, ␤-catenin has previously been implicated in insulin resistance. It was part of a human skeletal muscle gene expression signature that was diagnostic for insulin resistance and inhibition of ␤-catenin inhibited insulin-stimulated glucose uptake in L6 myotubes (52) and in 3T3-L1 adipocytes (53). Consistent with a role for ␤-catenin in insulin sensitivity, drug-induced ␤-catenin stabilization enhances insulin-mediated glucose uptake in adipocytes (53). As ␤-catenin stabilization also occurs as a consequence of tankyrase activity (25), it is conceivable that tankyrase activation, which reportedly occurs in response to insulin (19), may in fact enhance insulin sensitivity. It is also possible that tankyrase activity is involved in GSV protein stabilization. Hence, it will be of interest to determine whether dysregulated ␤-catenin activity contributes to the altered stability of GSV proteins or if this represents a separate regulatory pathway that contributes to insulin resistance in parallel.
Taken together, this study for the first time unveils the total proteome regulation of tankyrase inhibition in skeletal muscle cells, which paves the way for expanding our knowledge of tankyrase dysfunction-induced insulin resistance. This study will also likely have an impact on the use of tankyrase inhibitors in humans as this treatment might result in insulin resistance.

PARsylated protein enrichment
PARsylated proteins were enriched as previously described (59). Briefly, cells were lysed in SDS lysis buffer (1% SDS, 10 mM HEPES, pH 8.5). Lysates were mixed with m-aminophenylboronic acid-agarose. After 1 h incubation at room temperature, beads were washed with SDS wash buffer (1% SDS, 100 mM HEPES, 150 mM NaCl, pH 8.5) and subsequently with wash buffer containing 100 mM HEPES and 150 mM NaCl (pH 8.5). Beads were mixed with 3 M ammonium acetate (pH 5.0) for 1.5 h and washed once with SDS lysis buffer. Beads were then incubated with 4ϫ SDS-PAGE loading buffer at 95°C for 10 min and the eluate was subjected to immunoblotting analysis.

SDS-PAGE and Western blotting
Protein concentrations were determined via BCA assay and 10 g of proteins were resolved by SDS-PAGE. The gel was transferred to PVDF membranes. Membranes were blocked in 5% skim milk powder in TBST (0.1% Tween 20 in TBS) for 2 h followed by an overnight incubation at 4°C with the indicated primary antibody solutions. Membranes were incubated with an appropriate secondary antibody at room temperature for 1 h before signals were detected using a Chemidoc. Immunoblots were quantified by ImageJ software.

[ 3 H]2-DG uptake assay
This assay was performed as previously described (30). Briefly, L6 myotubes grown in 24-well plates were serum Tankyrase regulation of insulin sensitivity starved for 2 h in basal ␣-MEM and washed with 37°C Krebs-Ringer buffer with 0.2% BSA. Cells were then stimulated with 100 nM insulin for 20 min in Krebs-Ringer buffer with 0.2% BSA. To determine nonspecific glucose uptake, 25 M cytochalasin B was added to control wells prior to addition of [ 3 H]2-DG (0.125 Ci/well, 50 M unlabeled 2-DG) during the final 5 min of insulin stimulation. Following rapid washes with ice-cold PBS, cells were solubilized in 1% Triton X-100 in PBS on a shaker for 1 h and assessed for radioactivity by scintillation counting using a ␤-scintillation counter. All data were normalized to protein concentration.

Mouse experiments
All experiments were approved by the University of Sydney Animal Ethics Committee. Male C57Bl/6J mice were grouphoused on a 12-h light/dark cycle with free access to chow diet (13% calories from fat, 65% carbohydrate, 22% protein) or HFHSD (47% fat (7:1 lard to safflower oil ratio), 32% carbohydrate, 21% protein), and water. Mice were subjected to an intraperitoneal glucose tolerance test (GTT, 2 g/kg lean mass) after 5 weeks on either chow or HFHSD. Mice were euthanized after 6 weeks on the diet, quadriceps muscle was removed, freezeclamped, frozen and powdered, followed by immunoblotting.

Sample preparation for MS analysis
L6 myotubes were washed with ice-cold PBS and harvested in lysis buffer (6 M urea, 2 M thiourea, 25 mM TEAB, 0.1% SDS). After sonication, cell lysates were centrifuged at 15,000 ϫ g for 30 min at room temperature. The supernatant was precipitated using pre-chilled acetone overnight. Protein pellets were resuspended in buffer containing 6 M urea, 2 M thiourea, 25 mM TEAB. After reduction with 10 mM DTT at room temperature for 60 min and alkylation with 25 mM iodoacetamide at room temperature for 30 min in the dark, the protein mixture was digested using LysC (enzyme:substrate, 1:100) at 30°C for 2 h. The protein solution was then diluted 1:5 using 25 mM TEAB and further digested with trypsin (enzyme:substrate 1:50) at 37°C overnight. The peptide mixture was desalted using stagetips and dried by vacuum centrifugation for total proteome analysis.

Mass spectrometry analysis and data processing
MS-based total proteome analysis was performed on a Dionex UltiMate 3000 coupled to a Q-Exactive Plus in positive polarity mode. Peptides were separated using an in-house packed 75 m ϫ 40-cm column (1.9 m particle size, C18AQ) with a gradient of 10 -35% acetonitrile containing 0.1% formic acid over 360 min at 250 nl/min at 55°C. The MS1 scan was acquired from 350 to 1550 m/z (70,000 resolution, 3e6 automatic gain control, 100 ms injection time) followed by MS/MS data-dependent acquisition of the top 20 ions with higher-energy collisional dissociation (17,500 resolution, 1e5 automatic gain control, 60 ms injection time, 27 NCE, 1.2 m/z isolation width).
Raw data were processed using MaxQuant (version 1.5.3.24) (60) against a Uniprot rat database (May 2016, 37,402 entries) with default settings and minor changes. Oxidation of methionine and acetylation of protein N terminus were set as variable modifications and carbamidomethylation of cysteine was set as a fixed modification, "re-quantify," "second peptides searching," and "match between runs" were enabled. Both peptide spectral match and protein false discovery rate were set to 1%. "Label-free quantification method" (27) was used for protein quantification and normalization. The data were filtered to retain proteins quantified in at least 2 samples in each of the control and XAV939-treated conditions. Bioinformatics analysis was mainly performed using the LIMMA package (61) in the R programming environment to identify proteins differentially regulated with XAV939 treatment. Functional annotation was done by David version 6.8␤ (released in May 2016, https:// david-d.ncifcrf.gov/) 4 (62). The MS proteomics data have been deposited at the ProteomeXchange Consortium with identifier PXD007182.

Statistical analysis
Data are presented as mean Ϯ S.E., unless otherwise indicated. Statistical analyses were performed using t test or analysis of variance using GraphPad Prism software. Significance was set at p Ͻ 0.05 and p values are indicated.