Discoidal HDL and apoA-I-derived peptides improve glucose uptake in skeletal muscle

Lipid-free apoA-I and mature spherical HDL have been shown to induce glucose uptake in skeletal muscle. To exploit apoA-I and HDL states for diabetes therapy, further understanding of interaction between muscle and apoA-I is required. This study has examined whether nascent discoidal HDL, in which apoA-I attains a different conformation from mature HDL and lipid-free states, could induce muscle glucose uptake and whether a specific domain of apoA-I can mediate this effect. Using L6 myotubes stimulated with synthetic reconstituted discoidal HDL (rHDL), we show a glucose uptake effect comparable to insulin. Increased plasma membrane GLUT4 levels in ex vivo rHDL-stimulated myofibers from HA-GLUT4-GFP transgenic mice support this observation. rHDL increased phosphorylation of AMP kinase (AMPK) and acetyl-coA carboxylase (ACC) but not Akt. A survey of domain-specific peptides of apoA-I showed that the lipid-free C-terminal 190–243 fragment increases plasma membrane GLUT4, promotes glucose uptake, and activates AMPK signaling but not Akt. This may be explained by changes in α-helical content of 190–243 fragment versus full-length lipid-free apoA-I as assessed by circular dichroism spectroscopy. Discoidal HDL and the 190–243 peptide of apoA-I are potent agonists of glucose uptake in skeletal muscle, and the C-terminal α-helical content of apoA-I may be an important determinant of this effect.

protein sample by using desalting columns (GE Healthcare) equilibrated with PBS, pH 7.4. Protein purity was analyzed by SDS-PAGE, and concentration was determined by the BCA method (Pierce) or using a nanodrop 2000c spectrophotometer (Thermo Scientifi c).

Production of reconstituted HDL
1-palmitoyl-2-oleoyl-sn -glycero-3-phosphocholine (POPC) or 1,2-dimyristoyl-sn -glycero-3-phosphocholine (DMPC) (Avanti Polar Lipids) was dissolved in chloroform:methanol (3:1), which was evaporated under a stream of nitrogen gas, and the resulting lipid fi lm was resuspended in PBS. For POPC rHDL, deoxycholate was added to the POPC emulsion at a 2:1 molar ratio (deoxycholate:POPC) and incubated with apoA-I at a 156:1 molar ratio (phospholipid:protein) for 1 h at 22°C ( 16 ). Deoxycholate was removed from the POPC rHDL preparation by extensive dialysis against PBS. DMPC rHDL was prepared according to Ref. 17 . Briefl y, the DMPC emulsion was passed through a polycarbonate membrane with 100 nm pore size using the Lipo-soFast system (Avestin) a minimum of 20 times. The resulting vesicles were incubated with apoA-I at a 156:1 molar ratio (phospholipid:protein) for 4 days at 22°C. ApoA-I dimers, indicative of rHDL formation, was confi rmed by blue native PAGE (Invitrogen). POPC vesicles were prepared by fi rst passing the POPC emulsion through a polycarbonate membrane with 400 nm pore size followed by passage through a 100 nm pore size membrane using the LiposoFast system (Avestin) a minimum of 20 times. Treatments were performed with POPC rHDL unless otherwise indicated.

Circular dichroism spectroscopy
Circular dichroism (CD) measurements were performed on a Jasco J-810 spectropolarimeter equipped with a Jasco CDF-426S Peltier set to 25°C. ApoA-I (full-length and 190-243 fragment) was diluted to 0.1 mg/ml in PBS (fi nal concentration was 25 mmol/l phosphate, 25 mmol/l NaCl, pH 7.4), placed in a 0.1 mm quartz cuvette, and after extensive purging with nitrogen, scanned in the region 200-260 nm (scan speed was 20 nm/min). Averages of fi ve scans were baseline-subtracted (PBS buffer; 25 mmol/l phosphate, 25 mmol/l NaCl), and the ␣ -helical content was calculated from the molar ellipticity at 222 nm as previously described ( 18 ).

Western blotting
Prior to stimulation, cultured cells were serum starved for 4 h in serum-free ␣ -MEM, and all subsequent treatments, including insulin or phenformin as positive controls (Sigma), were performed in serum-free ␣ -MEM. After treatments, cells were washed with ice-cold PBS and lysed on ice using a nondenaturing lysis buffer (1% Triton X-100, 50 mmol/l Tris, 150 mmol/l NaCl, pH 8.0) containing protease and phosphatase inhibitors (Roche). Lysates were centrifuged at 16,000 g , 20 min at 4°C, and then BCA protein assay (Pierce) was performed on supernatants. Equal protein amounts were separated by SDS-PAGE and transferred to nitrocellulose membranes. pAMPK, AMPK, pACC, pAkt, Akt (Cell Signaling), and tubulin (Sigma) were used for immunodetection with IRDye 800CW and 680RD secondary antibodies (LI-COR). Blots were imaged using the Odyssey Fc system and quantifi ed using Image studio v2.0 software.

Glucose uptake measurements
Prior to stimulation cells were starved for 2 h in serum-free ␣ -MEM, and all subsequent treatments, which were performed in triplicate and included cytochalasin B (Sigma) as a measure of cell-associated nonspecifi c radioactivity, were performed in Despite these fi ndings, it is not clear whether discoidal HDL is also capable of specifi cally regulating muscle glucose uptake and whether this occurs via AMPK. Given that HDL subspecies interact differently with cellular receptors at the vascular wall for cholesterol effl ux and that discoidal HDL is a potent structure for this interaction ( 14 ), we hypothesized that discoidal HDL would be highly effective in the stimulation of glucose uptake in muscle.
Herein, we investigated the effects of synthetic discoidal HDL (rHDL) and apoA-I-derived peptides on glucose uptake, intracellular signaling, and GLUT4 translocation to the plasma membrane using L6 myotubes and fl exor digitorum brevis (FDB) fi bers. We show that rHDL produces insulin-like effects in these models, and we identify a novel peptide candidate that induces responses comparable to those of rHDL.

Expression and purifi cation of recombinant human apoA-I and apoA-I variants
Human apoA-I variants (full-length and truncated variants produced by site-directed mutagenesis, corresponding to amino acids 1-243 and to amino acids 1-189, 44-189, 44-243, and 190-243, respectively) were expressed in Escherichia coli strain BL21 Star (DE3)pLysS cells (Invitrogen) from the human apoA-I gene containing a hexa-His affi nity tag at the N-terminus ( 15 ). Briefl y, the gene (full-length or truncated variants of the gene) was cloned into the pEXP-5 plasmid (Novagen Inc.), transferred into the bacteria, and cultivated at 37°C in LB medium with 50 µg/ml of ampicillin and 34 µg/ml of chloramphenicol. Protein expression was induced for 3-4 h following the addition of 0.5 mmol/l isopropyl-␤ -thiogalactopyranoside (Sigma). Following cell disruption, apoA-I was purifi ed from the soluble fraction of the cells using a His-Trap-Nickel-chelating column (GE Healthcare) and a mobile phase of phosphate-buffered saline (PBS), pH 7.4, with 3 mol/l guanidine. The protein was then washed in PBS (pH 7.4) containing 100 mmol/l imidazole, and then eluted with PBS containing 500 mmol/l imidazole. Imidazole was removed from the  ( 36 ), discoidal HDL (middle) ( 16 ), and spherical HDL (right) ( 37 ) states, indicating signifi cant structural rearrangements in HDL maturation. The individual apoA-I molecules are shown in orange, blue, and gray, respectively. Yellow indicates the core of phospholipids, cholesterol, and cholesteryl esters . glucose uptake ( Fig. 2A ). DMPC as the lipid constituent in protein-free lipid vesicles and as the phospholipid constituent of rHLD was also tested. Whereas rHDL particles synthesized from apoA-I (30 µg/ml) and DMPC (0.16 mmol/l) induced glucose uptake to a level similar to insulinstimulated cells, DMPC vesicles alone did not stimulate glucose uptake (results not shown).
Blue native PAGE was performed on all rHDL preparations to confi rm the formation of 10 nm diameter discoidal apoA-I dimers. Fig. 2B is a representative Coomassie-stained gel that shows monomeric lipid-free apoA-I ( ‫ف‬ 28 kDa) and the size of POPC rHDL ( ‫ف‬ 10 nm diameter). These data clearly show that rHDL exerts a potent effect on glucose uptake in muscle and that the rHDL-mediated glucose uptake is apoA-I protein dependent.
To support the glucose uptake observations, the ability of rHDL to increase the amount of GLUT4 in the plasma membrane was assessed by immunofl uorescence microscopy of intact FDB muscle fi bers, isolated from a transgenic mouse model with muscle-specifi c HA-GLUT4-GFP expression ( 20 ). The HA epitope present on the fi rst exofacial loop of the HA-GLUT4-GFP construct allows detection of GLUT4 inserted into the plasma membrane. Intact FDB fi bers were incubated ex vivo with rHDL, followed by fi xation and HA antibody labeling of nonpermeabilized uptake buffer (140 mmol/l NaCl, 20 mmol/l HEPES, 5 mmol/l KCl, 2.5 mmol/l MgSO 4 , 1 mmol/l CaCl 2 , pH 7.4). After stimulation, treatments were replaced with 10 µmol/l 2-deoxy-D-glucose (Sigma) and 1 µCi/ml 2-[ 3 H]deoxy-D-glucose (Perkin Elmer) in uptake buffer for 15 min at room temperature. Cells were then washed twice with ice-cold PBS and lysed with 1 mol/l NaOH on ice. Lysates were collected and radioactivity was quantifi ed by scintillation counting.

Confocal microscopy
Fixed cells were imaged using a confocal LSM 510 microscope (Zeiss) using a 40× objective, NA 1.3, using BP 505-530 and LP 650. Images were collected with the LSM software.

Statistical analysis
All data are displayed as mean ± SEM unless indicated otherwise. Where appropriate, analysis was performed by two-tailed Student t -test or one-way ANOVA with Bonferroni's post hoc test using Microsoft Excel and Graph Pad Prism software. P р 0.05 was considered signifi cant.

rHDL induces glucose uptake and GLUT4 translocation in skeletal muscle cells
To investigate the effect of discodial HDL (rHDL) on GLUT4 translocation and glucose uptake, we produced recombinant human apoA-I and reconstituted HDL needed for cell incubations.
L6 myotubes were incubated with 2 µmol/l (60 µg/ml) discoidal rHDL (expressed as total protein concentration of apo A-I; given two apoA-I molecules per particle, this corresponds to 1 µmol/l discodial rHDL) for 1 h. The rHDL treatment induced a glucose uptake that was 2.3 ± 0.39-fold ( P р 0.05) over basal, which was similar to insulin stimulation (2.4 ± 1.0-fold) ( Fig. 2A ). To test for the contribution of the constituent phospholipid to rHDL-induced glucose uptake, rHDL made with POPC was compared with 100 nm POPC vesicles containing no apoA-I protein. Incubation with empty POPC vesicles (0.10 mmol/l) did not induce immunofl uorescence imaging of FDB fi bers labeled with HA-antibody was performed after ex vivo incubation with the 190-243 peptide. Relative to basal conditions, these images show greater membrane levels of GLUT4 protein in response to the 190-243 peptide ( Fig. 4C ), thus supporting the fi ndings in Fig. 4B . We next used western blotting to examine the signaling pathway activated by the 190-243 peptide. Incubation of L6 myotubes with increasing concentrations (2, 10, and 20 µmol/l) of 190-243 peptide resulted in phosphorylation of AMPK ( Fig. 4D ). In contrast, no Akt phosphorylation was observed. Phenformin at 1 mmol/l and insulin at 100 nmol/l were used as positive controls for phosphorylation of AMPK and Akt, respectively. Finally, our initial analyses on lipid-free apoA-I-induced signaling in L6 myotubes treated with liver X receptor (LXR) agonist to induce overexpression of ABCA1 suggest a non-ABCA1-dependent signaling pathway (data not shown).

Structural conformation of lipid-free 190-243 fragment resembles the conformation of lipid-bound apoA-I
We hypothesized that binding of cellular lipids to the 190-243 fragment upon incubation with L6 myotubes and FDB fi bers may be necessary for its glucose uptake-inducing effect. It is known that the 190-243 fragment can promote cholesterol effl ux from cultured cells and form discoidal particles by solubilization of lipid in solution ( 21,22 ), which can be visualized as oligomers on native PAGE. To assess oligomer formation indicative of lipid binding, purifi ed 190-243 peptide and conditioned media from cells treated with the 190-243 fragment for 1 h were run on a blue native PAGE and Coomassie stained ( Fig. 5A ). Under both conditions, 190-243 appeared as a single band at approximately 40 kDa corresponding to a tetramer. Lipidfree apoA-I and rHDL were included on the gel as a fulllength protein monomer and dimer reference. Although the presence of minute amounts of lipids in the 190-243 tetramers cannot be excluded, the migration distance is clearly different from the rHDL particles formed by the 190-243 peptide in interaction with cultured baby hamster kidney (BHK) cells expressing human ABCA1 ( ‫ف‬ 10 nm rHDL formed; approximately corresponding to the 242 kDa marker protein) and from interaction with phospholipid multilamellar vesicles ( ‫ف‬ 17 nm rHDL) ( 21 ). Moreover, our fi ndings on the oligomeric state of the cells. Both insulin and rHDL treatment induced translocation of GLUT4 into the sarcolemma plasma membrane as detected by HA signal ( Fig. 2C , upper panel). The lower panel in Fig. 2C displays total GLUT4 detected by GFP signal merged with the HA signal in nonstimulated and stimulated muscle fi bers. Due to steric hindrance, labeling of the transverse tubules was limited, and therefore, the plasma membrane GLUT4 translocation was assessed only at the sarcolemma.

C-terminal 190-243 peptide of lipid-free apoA-I increases glucose uptake in skeletal muscle
The relative effect of specifi c regions of apoA-I to increase glucose uptake was investigated using full-length and truncated protein fragments corresponding to resi-  membrane and thereby induce glucose uptake. From our work, it is clear that discoidal HDL promotes glucose uptake in cultured skeletal muscle, eliciting an effect comparable to insulin. Furthermore, we have made the discovery that the 190-243 peptide, corresponding to the C-terminal domain of apoA-I, is itself an effi cient agonist for glucose uptake.
Currently only two studies have shown that both lipidfree apoA-I and mature spherical HDL can increase glucose uptake in skeletal muscle via the AMPK signaling pathway ( 6,7 ). However, what had not been clearly addressed was the effi cacy of discoidal HDL, an important consideration given the marked alterations in structure that apoA-I undergoes during HDL maturation (depicted in Fig. 1 ). While the study by Drew et al. ( 7 ) found increased ACC phosphorylation in muscle biopsies after 4 h rHDL infusion, indicative of AMPK activation, it is diffi cult to specifi cally attribute this observation to discoidal HDL. As the authors acknowledge, rHDL is likely to become rapidly remodeled to mature HDL in the plasma. By using both synthetic discoidal HDL containing different phospholipid species and phospholipid vesicles lacking an apoprotein component, we were able to show that the glucose uptake response to rHDL is apoA-I dependent. Whether this involves lipid exchange of rHDL with the L6 myotubes, a capability that the phospholipid vesicles essentially lack, is not clear. Although rHDL does not refl ect the physiological heterogeneity of 190-243 peptide is in agreement with those on the 198-243 peptide that self-associates as tetramers in solution ( 23 ).
As depicted in Fig. 1 , the structure of apoA-I in the apo-state is signifi cantly different from the structural organization of the same protein in rHDL particles. This structural transition of the lipid-binding process involves a major increase in ␣ -helical secondary structure (from about 44-55% in the apo-state to 78% ␣ -helical secondary structure in discoidal HDL) ( 17,24 ). As rHDL is potent in stimulation of glucose uptake in myotubes, we speculated that lipid-free 190-243 fragment may adopt an amphiphatic ␣ -helix in solution. To investigate this, CD spectroscopy spectra were obtained at a protein concentration of 0.1 mg/ml for the 190-243 fragment and full-length protein for comparison ( Fig. 5 ). The helical content was estimated from their molar ellipticities at 222 nm to be 17% (or 21% at 0.2 mg/ml; data not shown) and 56% for 190-243 fragment and full-length apoA-I, respectively, suggesting an increase in helical structure of the fragment (see Discussion).

DISCUSSION
This study analyzed the capability of discoidal HDL and the potency of subdomains of apoA-I to promote translocation of the GLUT4 glucose transporter to the plasma the apoA-I protein. It was evident that peptides containing the C-terminal domain had an improved capacity to induce glucose uptake in relation to peptides lacking this region and that the 190-243 fragment alone carried the most potent effect. It is interesting to note that structural studies reveal that this region carries the highest lipid-binding capacity in relation to N-terminal and central domains ( 23,27 ) and that a 198-243 peptide self-associates into a tetrameric organization ( 23 ). Indeed, lipid binding is impaired in an apoA-I mutant lacking this region ( 21,28 ), whereas this fragment alone can initiate cholesterol effl ux from J774 macrophages that have been stimulated to produce high ABCA1 levels and can also lead to formation of peptide/ lipid particles with apparent diameters of 10 or 17 nm depending on whether they have been incubated with multilamellar vesicles or BHK cells ( 21 ). ApoA-I is known to interact with ABCA1, with cholesterol effl ux resulting ( 14 ). While we cannot completely exclude the possibility that minor amounts of cholesterol and phospholipids are taken up by the tetrameric peptide, the native gel analysis clearly show that discoidal-like HDL particles are not formed. In conclusion, we observe no evidence of lipid binding to the 190-243 peptide after incubation with L6 myotubes, allowing us to speculate that the glucose uptake effect of this peptide is independent of lipid effl ux.
This picture is in partial agreement with Drew et al ( 7 ) who, despite showing lipid effl ux to HDL, found this to be independent of ABCA1-dependent induced glucose uptake. This is perfectly feasible on the basis of Vedhachalam et al. ( 28 ) who proposed a two-step process of lipid effl ux to apoA-I, which involves ABCA1 binding followed by membrane lipid site association. It is possible that in skeletal muscle, the 190-243 fragment binds ABCA1 but does not continue to the second step of binding to the membrane domain, perhaps due to internalization, an event known to be necessary for the AMPK signaling and glucose uptake effects of lipid-free apoA-I ( 6 ).
It is notable that in our experiments apoA-I shows a clear trend toward increasing glucose uptake, but in contrast to other studies ( 6, 7 ), it is not signifi cant, whereas rHDL and the 190-243 peptide produce signifi cant insulin-like effects. We speculated that the 190-243 fragment better resembles the effect of full-length apoA-I in rHDL than in the apo-state due to structural rearrangements of the peptide. To analyze this, we used CD spectroscopy to compare secondary structure contents of intact apoA-I protein and the peptide fragment.
In the apo-state, this region of the intact protein contains no helical structure ( 29 ), and the obtained 17-21% average ␣ -helical content of the 190-243 fragment thus means a net increase in helical structure of this region. Under certain conditions, this may be even higher as reported in previous analyses ( ␣ -helical content is 33 ± 3% in Ref. 22 ). The results suggest that the 190-243 fragment in the apo-state adopts an rHDL-resembling structure in solution ( Fig. 6 ) that, similar to the previously described 198-243 peptide, adopts a tetrameric organization ( 23 ); we therefore propose that this structural organization is the preferred structure to induce glucose uptake in skeletal muscle. nascent discoidal HDL, rHDL offers the advantage of allowing defi ned oligomeric states to be studied and is free from plasma contaminants and modifi cations, such as oxidation and glycation, which have been shown to infl uence HDL function ( 25,26 ). Despite the use of a supraphysiological concentration of apoA-I, our signaling data in rHDL-treated L6 myotubes also concur with previous fi ndings that showed activation of AMPK and its downstream target ACC but found no increase in Akt phosphorylation. This is an important feature of apoA-I/HDL action in muscle that allows glucose uptake to be maintained independently of the insulin signaling pathway.
Evidence for the direct effect of apoA-I in its lipid-free and various lipid-bound states on skeletal muscle glucose uptake is now strong, having been verifi ed in murine C2C12 myotubes ( 6 ), primary human cultures from diabetic patients ( 7 ), and now in rat L6 myotubes. Using myofi bers from the HA-GLUT4-GFP transgenic mouse, this study has added a further, highly sensitive model that confi rms the direct infl uence of discoidal HDL on GLUT4 plasma membrane levels.
To exploit the properties of apoA-I for therapeutic means, we initiated the search for unique functional peptides by analyzing the glucose uptake effect of defi ned regions of required to elicit this effect. From the example of apoA-I mimetics currently under development in the fi eld of cardiovascular disease ( 34 ), the production and testing of smaller therapeutic peptides derived from knowledge of the 190-243 fragment may provide promise, with the apoA-I mimetic L-4F preventing insulin resistance in vivo via effects on insulin signaling and lipid metabolism ( 35 ). Future work is now needed on small peptide regions of the 190-243 fragment, which may prove to be viable treatment agents for insulin resistance and type 2 diabetes.
Although a role for the ABCA1 receptor has been shown for HDL and lipid-free apoA-I-induced glucose uptake ( 7 ), this does not exclude the involvement of alternative receptors. A potential candidate is the ecto-F1-ATPase, which has been shown to act as an apoA-I receptor in hepatocytes ( 30 ), adipocytes ( 31 ), and endothelial cells ( 32,33 ), but this interaction is yet to be characterized in skeletal muscle. The possibility of novel yet unidentifi ed receptors for apoA-I also remains. Our discovery of the novel action of the 190-243 peptide provides a valuable tool for future investigations aimed at identifying novel receptor targets. Such work will not only help to refi ne mechanisms of apoA-I/HDL action, but will themselves represent exciting pharmacological targets.
In conclusion, this study has fi lled the gap in knowledge regarding the effect of discoidal HDL on skeletal muscle glucose uptake, showing that the activity of apoA-I in this protein arrangement is as potent as it is in the lipid-free and mature spherical HDL states. We have also identifi ed the 190-243 peptide as harboring the relevant characteristics