Loss of CHIP Expression Perturbs Glucose Homeostasis and Leads to Type II Diabetes through Defects in Microtubule Polymerization and Glucose Transporter Localization

Recent evidence has implicated CHIP (carboxyl terminus of Hsc/Hsp70-interacting protein), a co-chaperone and ubiquitin ligase, in the functional support of several metabolism-related proteins, including AMPK and SirT6. In addition to previously reported aging and stress intolerance phenotypes, we find that CHIP -/- mice also demonstrate a Type II diabetes-like phenotype, including poor glucose tolerance, decreased sensitivity to insulin, and decreased insulin-stimulated glucose uptake in isolated skeletal muscle, characteristic of insulin resistance. In CHIP-deficient cells, glucose stimulation fails to induce translocation of Glut4 to the plasma membrane. This impairment in Glut4 translocation in CHIP-deficient cells is accompanied by decreased tubulin polymerization associated with decreased phosphorylation of stathmin, a microtubule-associated protein required for polymerization-dependent protein trafficking within the cell. Together, these data describe a novel role for CHIP in regulating microtubule polymerization that assists in glucose transporter translocation, promoting whole-body glucose homeostasis and sensitivity to insulin.

paraformaldehyde in PBS (w/v). GFP was visualized directly to detect myc-Glut4-GFP. Indirect 128 immunofluorescence of endogenous Glut4 was carried out by incubating the Glut4 antibody (1:200) and 129 visualizing with an AlexaFluor488 secondary antibody (Life Technologies). In some instances, cells were 130 stained with TexasRed-X Phalloidin (Life Technologies) for 45 min. For indirect immunofluorescence 131 using ex vivo muscle fibers, the fixed fibers were transferred to blocking buffer containing 50 mM 132 glycine, 0.25% BSA, 0.04% TritonX-100, and 0.05% sodium azide in PBS, and permeabilized in buffer 133 containing 1% BSA and 0.5% TritonX-100 in PBS for 30 min, after which time fibers were incubated 134 overnight with the α-tubulin antibody diluted 1:1000 in blocking buffer. After 3 PBS washes of 30 min 135 each, the myofibers were incubated for 2 h with an AlexaFluor 488-conjugated goat anti-mouse antibody 136 (Life Technologies) and washed 3 times in PBS for 5 min each. All cells and muscle fiber preparations 137 were mounted with glass cover slips using Vectashield mounting media with DAPI (Fisher Scientific, absorbance correlates with the amount of exposed myc epitope and thus can be used to calculate the 154 amount of membrane-inserted myc-Glut4-GFP. 155 156 Two-dimensional differential in gel electrophoresis (2D-DIGE), matrix-assisted laser desorption 157 ionization time of flight (MALDI-TOF) mass spectrometry, and functional clustering. Differential protein 158 expression using 2D-DIGE comparing wild-type and CHIP-deficient mouse embryonic fibroblast protein 159 extracts was carried out as previously described (44). 34 spots identified via 2D-DIGE (unpaired t-test, p 160 < 0.05, FDR < 25%) were identified via MALDI-TOF mass spectrometry and analyzed using the DAVID 161 functional annotation clustering tool (14,21). The mean fold-enrichment (log2) of each functional cluster 162 (x-axis) and corresponding proteins (y-axis) were analyzed using a Pearson's centered, complete linkage 163 clustering analysis as previously described (8).

165
Single muscle fiber preparation. Wild-type and CHIP-deficient mice were fasted overnight and split into 166 two groups. The first group of mice maintained the fast, whereas the second received an IP injection of 167 20% D-glucose (2 µg glucose/g body mass). After 40 min, mice were sacrificed by cervical dislocation, 168 and the gastrocnemius was immediately removed and placed in 2% paraformaldehyde (w/v) for 1 h, then 169 rinsed several times in PBS. Small bundles of 1-3 fibers were then teased away with fine forceps.

171
Statistics. Statistical tests were performed as indicated in the methods, figure legends, and table legends.

174
CHIP is necessary for optimal glucose homeostasis. CHIP-deficient mice demonstrate an accelerated 175 aging phenotype as well as other maladies, including muscle wasting, that suggest metabolic imbalances 176 (35). Given the high level of CHIP expression detected in human skeletal muscle (4), the association 177 between muscle atrophy and altered glucose homeostasis (47), and the established link between the loss may have altered glucose homeostasis stemming from skeletal muscle dysfunction. Consistent with our 180 hypothesis, mice lacking CHIP exhibited a mild hyperglycemia that was more pronounced under fasted 181 conditions (Fig. 1A). The change in glucose homeostasis was accompanied by a trend towards an increase 182 in insulin levels that was unmasked in the fed state (Fig. 1A), suggesting that decreased insulin 183 responsiveness in the peripheral tissues of CHIP-deficient mice may result in compensatory 184 hyperinsulinemia typical of Type II diabetes (18). To test this, a glucose tolerance test was performed on 185 wild-type and CHIP-deficient 129SvEv mice to determine the acute ability of tissues in these mice to 186 remove glucose from the blood. CHIP-deficient mice failed to clear glucose from the blood efficiently 187 compared to wild-type mice (48 ± 9% increase in the AUC comparing CHIP -/to wild-type mice, Fig.   188 1B). To measure whether the defect in glucose tolerance was due to a reduction in insulin release or 189 insulin receptor sensitivity, an insulin tolerance test was performed. We found that intraperitoneal 190 administration of insulin had a smaller effect on lowering blood glucose levels in CHIP-deficient mice at 191 all time points tested (49 ± 6% increase in the AUC comparing CHIP -/to wild-type mice, Fig. 1C), 192 suggesting that insulin sensitivity is impaired in CHIP -/mice. To directly assess insulin sensitivity, we 193 performed a hyperinsulinemic-euglycemic clamp assay. During insulin infusion, CHIP-deficient mice 194 required half the rate of glucose infusion compared to wild-type mice to maintain blood glucose levels 195 (Table 1), indicating whole-body insulin resistance. This finding was supported by the fact that CHIP-196 deficient mice were unable to suppress hepatic glucose production to the same extent as wild-type mice, 197 and there was a dramatic reduction in the ability of CHIP-deficient gastrocnemius to take up 2-198 deoxyglucose during hyperinsulimia (45 ± 9%; Table 1). This poor glucose tolerance phenotype and 199 impaired glucose uptake in skeletal muscle was recapitulated in CHIP -/mice bred on a mixed C57BL/6 × 200 129 background ( Fig. 1D and Table 1). Taken together, these data demonstrate that the absence of CHIP 201 leads to a Type II diabetes-like phenotype under physiological conditions due to impaired glucose uptake 202 and insulin resistance in skeletal muscle.
CHIP is necessary for optimal Glut4 translocation to the membrane following insulin stimulation.

205
We hypothesized that the severe defect in glucose uptake associated with CHIP deficiency ( Table 1) may be linked to the inability of Glut4, the major glucose transporter found in skeletal muscle 207 and adipose tissue, to translocate from the cell interior to the plasma membrane in response to insulin 208 stimulation, thereby hindering the cellular uptake of glucose (22). Following insulin stimulation, a series 209 of intracellular signaling pathways promote Glut4 movement from intracellular vesicles to the cell 210 membrane (9). Once Glut4 reaches the cell membrane, it fuses and undergoes exocytosis to facilitate the 211 diffusion of glucose through the cell membrane. Glut4 translocation is essential for successful insulin 212 signaling and glucose uptake by skeletal muscle cells (23,26,39,45,51). We employed lentiviral 213 transduction of control or CHIP shRNA in mouse skeletal muscle C2C12 cells (shCONT and shCHIP cell 214 lines) to investigate if a reduction in CHIP expression influences insulin-dependent Glut4 translocation.

215
To measure Glut4 translocation, we transfected shCONT and shCHIP C2C12 cells with a myc-Glut4-216 GFP construct (myc N-terminus, GFP C-terminus; (52)) and examined the location of Glut4-GFP protein 217 following 4 h serum starvation before and after a challenge with 25 mM glucose and 100 nM insulin.

218
After serum starvation and prior to glucose/insulin stimulation (time 0), Glut4-GFP was largely localized 219 to the perinuclear region in both the shCONT and shCHIP C2C12 cells ( Fig. 2A). However, following 30 220 min of insulin stimulation, Glut4-GFP translocated to the membrane in shCONT cells, but remained in a 221 perinuclear region in the shCHIP cells ( Fig. 2A). To verify this lack of Glut4 movement in response to 222 insulin in shCHIP cells, we took advantage of the myc tag on the Glut4-GFP construct and performed a 223 colorimetric assay of surface myc-Glut4-GFP using O-phenylenediamine dihydrochloride (OPD). In this 224 method, insertion of myc-Glut4-GFP protein into the cell membrane results in the exofacial positioning of 225 the myc tag. This tag is subsequently labeled with a secondary antibody conjugated to peroxidase and 226 then is treated with OPD reagent, which results in a colorimetric reaction that can be monitored by 227 measuring light absorbance at 492 nm (50). Using this assay, we confirmed that, whereas the membrane This lack of Glut4 translocation to the membrane in the shCHIP cells suggests that CHIP is involved in 231 insulin-dependent trafficking of Glut4 to the cell membrane.  (Fig. 5). In contrast, this 295 lattice appearance was strikingly absent in CHIP-deficient myofibers, revealing a lack of microtubule 296 polymerization at the myofiber surface (Fig. 5). Taken together, we observe hyperglycemia and insulin 297 resistance in CHIP-deficient mice, and we demonstrate that shCHIP cells exhibit decreased microtubule 298 polymerization and Glut4 translocation. We propose that CHIP plays a critical role in promoting insulin-299 stimulated microtubule polymerization and the movement of Glut4 to the membrane, facilitating the 300 import of glucose into the cell and maintaining glucose homeostasis.

303
Glucose transporter translocation is an important cellular mechanism in glucose homeostasis, especially 304 in insulin-sensitive tissues such as skeletal muscle. Previously, we reported that mice lacking CHIP 305 expression exhibit phenotypes such as accelerated aging and decreased tolerance to cardiac stressors such 306 as ischemia reperfusion and pressure overload (35). Here we add to the characterization of the CHIP-deficient phenotype by demonstrating that CHIP-deficient mice also develop a Type II diabetic 308 phenotype, including glucose intolerance (Fig. 1B, 1D), insulin resistance (Fig. 1C, Table 1), and 309 decreased glucose uptake in skeletal muscle (Table 1). The apparent cause for this aberrant 310 glucose/insulin response is at least in part due to the disruption of microtubule reorganization in response 311 to insulin stimulation (Fig. 3A, 3B, 3C, and 5) that is necessary for the glucose transporter Glut4 to 312 translocate to the cell membrane and facilitate the entry of glucose into the cell (Fig 2A, 2B, and 3A).

313
These data support the hypothesis that CHIP expression is necessary for maintaining optimal 314 glucose/insulin signaling and provides further evidence of an intersection between protein quality control 315 and metabolic pathways.

317
The idea that CHIP, a co-chaperone/ubiquitin ligase involved in protein quality control pathways, could 318 play a role in metabolic homeostasis evolved over the last several years. The discovery that CHIP forms a 319 complex with the stress-induced kinase SGK1 (serum-and glucocorticoid-regulated kinase-1; (6)