The RabGAPs TBC1D1 and TBC1D4 Control Uptake of Long-Chain Fatty Acids Into Skeletal Muscle via Fatty Acid Transporter SLC27A4/FATP4

As direct downstream effectors of both, AKT (protein kinase B) and AMP-dependent kinase AMPK, the two RabGTPase-activating proteins (RabGAPs) TBC1D1 and TBC1D4 have been shown in previous studies to act as critical regulators of skeletal muscle glucose metabolism (1–5). RabGAPs accelerate the intrinsic GTP hydrolysis activity of Rab proteins, a large family of small GTPases that control a multiplicity of cellular vesicle trafficking and organelle translocation processes (6–8). In the past, several Rab proteins were described to be direct targets of the two RabGAPs and, in addition, to be associated to GLUT4 vesicles, namely Rab8a, Rab10, Rab14, and Rab28 (3,4,9–11). Rab8b, in addition, has been described as a direct downstream effector of TBC1D1, but its role in skeletal muscle substrate metabolism is not yet clear (4,12). Interestingly, both TBC1D1 and TBC1D4 have been implicated to play an important role in lipid metabolism as well. Deficiency in either Tbc1d1 or Tbc1d4 as well as the combined knockout of both RabGAPs in mice leads to enhanced in vivo lipid utilization, indicated by a lowered respiratory quotient (5,13). In fact, TBC1D-family GAPs reciprocally regulate glucose uptake and fatty acid oxidation (FAO) in skeletal muscle through independent signaling pathways and, at least in mice, this control is dependent on the muscle fiber composition (5). We have previously demonstrated that insulin- and AICAR-stimulated glucose uptake is normal in glycolytic extensor digitorum longus (EDL) muscle from Tbc1d4-deficient (D4KO) mice, whereas EDL muscles from Tbc1d1-deficient (D1KO) mice show impaired response toward these stimuli (5). In accordance, EDL muscle from D1KO but not D4KO mice shows lower abundance of GLUT4, presumably resulting from missorting of the protein, and may, in part, explain the reduced glucose transport. At that point, it seems to be clear that insulin- and AMPK-signaling pathways converge at the level of the two RabGAPs and that TBC1D1 and TBC1D4 are required for sorting of GLUT4 in skeletal muscle with TBC1D1 representing the major signal transducer for glucose transport in glycolytic skeletal muscle and TBC1D4 in oxidative skeletal muscle and adipose tissue (1,5,14,15). Most importantly, basal FAO is equally elevated in EDL muscle from D1KO and D4KO mice, but only the former shows this increase also in the oxidative soleus (SOL) muscle (5). These data indicate that despite the fact that both RabGAPs contribute to the effect on FAO, this influence is not directly related to skeletal muscle glucose uptake, as D4KO mice demonstrate normal glucose uptake but increased FAO in EDL muscle. Importantly, these effects are strictly dependent on the functional GAP domain since no changes in lipid utilization were monitored in cultivated muscle cells overexpressing a nonfunctional R941K mutant of Tbc1d1 (16).

Skeletal muscle FA uptake has been shown to be dependent on several distinct mechanisms including diffusion through the plasma membrane, as postulated for short-chain FAs (SCFAs), and complex transport systems specific for long-chain FAs (LCFAs) involving a variety of different membrane transporter and binding proteins (17,18). At least three different FA transporters have been described to control skeletal muscle LCFA uptake (19). The most extensively studied FA transporter is FAT/CD36, which exclusively binds LCFAs and has been implicated in lipid storage of different tissues and also pathophysiological impairments such as obesity and type 2 diabetes (20). Besides FAT/CD36, at least two more FA transporters, both members of the SLC27 family of FA transport proteins (FATP), have been reported to be relevant in skeletal muscle lipid metabolism, FATP1 (SLC27A1) and FATP4 (SLC27A4) (21). In L6 rat muscle cells, Tbc1d4 deficiency led to enhanced influx of FAs associated with an increase in FA translocase FAT/CD36 and FABPpm expression (22), and in mouse cardiomyocytes, knockdown (Kd) of Tbc1d4 led to redistribution of FAT/CD36 to the cell surface (23). Of note, previous studies have provided evidence for the involvement of a regulated translocation mechanism for FA uptake facilitators (e.g., FAT/CD36, FATP1, and FATP4) (19,24–26) in analogy to the mechanism described for the uptake of glucose through GLUT4 in insulin-sensitive cells (27,28). However, there are no data available so far on the specificity of RabGAP-regulated lipid metabolism with regard to the FA species (17–19).

The aim of the current study was to characterize the molecular mechanisms that are responsible for the increased FA use upon inactivation of Tbc1d1 or Tbc1d4 in skeletal muscle cells and to determine the responsible downstream targets of the two RabGAPs.

Chemicals and buffer ingredients are listed in Supplementary Table 1. Buffers and cell culture media are listed in Supplementary Table 2.

Mice were kept in accordance with the National Institutes of Health guidelines for the care and use of laboratory animals, and all experiments were approved by the ethics committee of the State Ministry of Agriculture, Nutrition and Forestry (Reference numbers 84–02.04.2013.A352 and 84–02.04.2017.A345, State of North Rhine-Westphalia, Germany). Unless indicated otherwise, three to six mice per cage were housed at 22°C and a 12-h light–dark cycle with ad libitum access to food and water. After weaning, animals received a standard chow diet (Ssniff, Soest, Germany). Male mice were used at the age of 12–25 weeks. Generation of D1KO and D4KO mice was described previously (5,16,29). Heterozygous Cd36-knockout (CD36KO) mice (30) were bred with either D1KO or D4KO mice and the F1 generation was intercrossed to generate homozygous double-deficient D1/CD36KO and D4/CD36KO mice. Isolation of genomic DNA from mouse tail tips was performed with the InViSorb Genomic DNA Kit II (Invitek, Berlin, Germany). Genotyping of mice was performed by PCR with primers for the Tbc1d1 and Tbc1d4 knockout as described (5). Genotyping of the Cd36 knockout was conducted as described (30). Sequences of all genotyping primers are listed in Supplementary Table 3.

The in vivo electroporation (IVE) protocol was adapted from previous reports (31–33). Briefly, mice were anesthetized using isoflurane, and their hind limbs were shaved. Then 30 µL saline containing 15 units hyaluronidase (Sigma-Aldrich, Steinheim, Germany) were injected in the proximal and distal part of the front and the back of the shank to target EDL and SOL muscle. Following a 1-h recovery in their cages, the mice were anesthetized again with isoflurane and 4 µg siRNA oligonucleotides specific for Rab8a, Rab10, or Slc27a4/Fatp4, respectively, in 30 mL sterile saline were injected into the same leg regions as before. Slc27a4/Fatp4 siRNA oligonucleotides were mixed with Invivofectamine 2.0 (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions to a final concentration of 0.7 mg/mL. In each animal, one muscle was used as a control and therefore injected with non-target (NT) siRNA oligonucleotides. The siRNA injection was followed by the application of a pair of tweezer electrodes across the distal limb connected to an ECM-830 electroporator device (Square Wave Electroporation System ECM 830; BTX). Eight pulses (80 V, 20 ms, 1 Hz) were applied to each leg following recovery of the animals. Three (Slc27a4/Fatp4) or seven (Rab-GTPases) days after IVE intervention, mice were subjected to ex vivo experiments. Only muscle samples with a minimum Rab protein reduction of 20% were considered for further statistical analysis.

C2C12 myoblasts were grown to 80% confluence in DMEM (PAA Laboratories) supplemented with 10% FBS and 1% penicillin/streptomycin at 37°C and 5% CO₂. Myotube differentiation was induced by switching to DMEM containing 3% horse serum. Cells were used for assays when fully differentiated (usually 7 days). For Kd of Tbc1d1, Tbc1d4, the different RabGTPases or FA transporters, C2C12 cells were transfected with 50 nmol/L siRNA duplexes using DharmaFECT1 on day 3 of differentiation. Experiments were conducted at day 6 of differentiation. Sequences from siRNA oligonucleotides used are listed in Supplementary Table 4.

C2C12 myotubes (6 days after differentiation) were serum-starved for 2 h (DMEM [high glucose, 4.5 g/L] supplemented with 1% FA-free BSA) before adding Krebs-Ringer-HEPES buffer (see Supplementary Table 2), supplemented with 40 μmol/L FA-free BSA. To start the uptake assay, a Krebs-Ringer-HEPES-based buffer (radioactive “HOT” buffer) containing either 8.5 nmol/L ³H-palmitate, ³H-oleate, or ³H-butyrate (Supplementary Table 5), 2.5 μmol/L FA-free BSA and 5 μmol/L unlabeled palmitate, bound in a molar ratio of 2:1 to BSA, was added to each well for 5 min at 37°C/5% CO₂. Afterward, cells were placed on ice, HOT buffer was immediately aspirated, and each well was intensively washed three times with ice-cold washing buffer. Residual washing buffer was completely aspirated, and cells were lysed in protein lysis buffer (with protease and phosphatase inhibitor cocktail) and cleared protein lysates were prepared. A total of 50 µL of the lysates were used for liquid scintillation counting, the rest was used for protein measurement. Radioactivity was measured by scintillation counting and normalized to protein concentrations. [1-¹⁴C]-palmitic acid oxidation assays were performed essentially as described (29,34). Briefly, C2C12 myotubes in 48-well plates were incubated with HOT oxidation buffer (11.8 μmol/L ¹⁴C-palmitate, 6.24 μmol/L FA-free BSA and 1 μmol/L L-carnitine in C2C12 differentiation medium) for 4 h at 37°C/5% CO₂. Then, supernatants were transferred to fresh 48-well plates, acidified with 1 mol/L HCl and incubated overnight with NaOH-soaked filter papers. The remaining cells were washed with PBS and analyzed for protein content. Trapped radioactivity was determined by scintillation counting and normalized to the amount of cellular protein.

Assays were done essentially as described (16). Briefly, EDL and SOL muscles were incubated in Krebs-Henseleit buffer containing 15 mmol/L mannitol, 5 mmol/L glucose, 3.5% FA-free BSA, 4 mCi/mL ³H-palmitate, and 300 mmol/L unlabeled palmitate at 30°C for 2 h. After absorption of FAs to activated charcoal, FAO was determined by measuring tritiated water using a scintillation counter.

Protein lysates (10–30 μg) were separated by 8–12% SDS-PAGE and transferred onto nitrocellulose membranes by tank blotting (Amersham Protran 0.45 µm). Membranes were blocked for 1 h with 5% fat-free powdered milk in Tris-buffered saline with Tween (Supplementary Table 2), incubated with primary antibodies and secondary horseradish peroxidase–conjugated antibodies as described in Supplementary Table 6 and developed with enhanced chemiluminescence reagent (Perkin Elmer). For analysis of SLC27A4/FATP4 membrane localization, Kd of Tbc1d1 was conducted in C2C12 myotubes as described above and subjected to a fractionation protocol modified from (35). Briefly, cells were dissolved in homogenization buffer A and centrifuged at 500g for 5 min at 4°C to remove cell debris. A fraction of the supernatant was stored for later determination of total SLC27A4/FATP4, the rest was subjected to an ultracentrifugation step (100,000g, 4°C, 30 min) to spin down membranes. Supernatants were discarded and crude membrane pellets were resuspended in homogenization buffer B. Buffer compositions are listed in Supplementary Table 2. Fractions A and B were analyzed via Western blot analysis as described above.

RNA was isolated using RNeasy Mini Kit and cDNA was synthesized using GoScript Reverse Transcriptase Kit (Promega) with 1 µg RNA and random hexanucleotide primers (Roche). For quantitative real-time PCR (qPCR), specific primers were used with the GoTaq quantitative PCR Master Mix on a StepOne Plus device (Applied Biosystems). Primer sequences are shown in Supplementary Table 7. Analysis was performed using the 2^−∆∆Ct-method (36) with Tbp (TATA-box binding protein) as reference gene. Quantification of Tbc1d1 and Tbc1d4 copy number was performed as described previously (29).

Skeletal muscle total FA (TFA) content and specific fractional compositions of FAs were determined by gas chromatography (37). FA data were further used to calculate the Δ9-desaturase index (C16:1/C16:0 or C18:1/C18:0) and the Δ5-desaturase index (C18:2/C20:4) as well as the sums of TFA, nonsaturated FAs, monounsaturated FAs (MUFAs), saturated FAs (SFAs), essential FAs (C18:2 + C18:3) or nonessential FAs (C16:0 + C16:1 + C18:0 + C18:1) (38,39). Nomenclature used indicates Cx:y (x, number of carbons in the FA; y, number of double bonds in the FAs).

All experiments were performed with at least n = 3 independent samples and are shown as mean values ± SEM. Statistical significance (P < 0.05) was calculated with appropriate tests (unpaired or paired two-tailed Student t test and one- or two-way ANOVA) using GraphPad Prism 8 software as detailed in the respective figure legends.

The data and critical resources supporting their reported findings, methods, and conclusions are available from the corresponding author upon reasonable request.

To investigate the mechanism of RabGAP-regulated FA utilization, we performed siRNA-mediated Kd of Tbc1d1 and Tbc1d4 in differentiated C2C12 myotubes and, subsequently, determined the uptake of the LCFAs ³H-palmitate and ³H-oleate or the SCFA ³H-butyrate, respectively. The rationale behind this experiment was to determine the specificity for the uptake of distinct lipid species such as saturated versus unsaturated (palmitate, oleate) and LCFA versus SCFA in RabGAP-deficient muscle cells. To validate successful Kd of the two target genes, expression of Tbc1d1 and Tbc1d4 mRNA was determined via qPCR. Treatment of the cells with target-specific siRNA oligonucleotides reduced the mRNA expression of Tbc1d1 or Tbc1d4 by 70–85% (Fig. 1A and B). ³H-palmitate uptake was increased in C2C12 myotubes after Kd of Tbc1d1 and Tbc1d4 compared with cells transfected with unspecific NT siRNA duplexes (Fig. 1A and B). Moreover, both Tbc1d1 and Tbc1d4 deficiency led to a significantly increased oxidation of ¹⁴C-palmitate (Fig. 1C). Interestingly, Kd of Tbc1d1 and Tbc1d4 resulted in increased uptake of ³H-oleate into the muscle cells as well (Fig. 1D). In contrast, uptake of ³H-butyrate was unchanged in C2C12 myotubes following either Tbc1d1 or Tbc1d4 depletion (Fig. 1E). To assess the relative contribution of each of the two RabGAPs in skeletal muscle FA uptake and oxidation, we determined expression levels of Tbc1d1 and Tbc1d4 transcripts by measuring mRNA copy numbers via qPCR using a standard calibration curve as described (29). Tbc1d1 gene expression increased during the course of differentiation, whereas Tbc1d4 transcript levels were unchanged in differentiated C2C12 myotubes compared with the undifferentiated myoblasts (Fig. 1F).

We conducted Kd experiments of the previously reported RabGAP-target RabGTPases Rab8a, Rab8b, Rab10, Rab14, and Rab28 in C2C12 myotubes and analyzed ³H-palmitate uptake in these cells. Gene expression levels of the different Rab genes were reduced between 40 and 60% 4 days after transfection of target-specific siRNA oligonucleotides (Fig. 2A–E). Although ³H-palmitate uptake was significantly reduced in C2C12 myotubes after Kd of Rab8a, Rab8b, Rab10, Rab14, depletion of Rab28 did not result in alterations of FA influx (Fig. 2A–E). Next, we conducted Kd of Rab8a and Rab10 in skeletal muscle from C57BL/6J mice via IVE as described in the research design and methods section. The Kd efficiency was ∼35% in the EDL and ∼41% in the SOL muscle (Supplementary Fig. 1A). Seven days after transfection of siRNA, ³H-palmitate oxidation was measured ex vivo in intact skeletal muscles. Interestingly, FAO was decreased by ∼24% in the oxidative SOL muscle but not in the glycolytic EDL muscle following Rab8a silencing (Fig. 2F). Rab10 Kd, however, did not lead to changes in ³H-palmitate oxidation (Supplementary Fig. 1B). Gene expression levels of Rab8a, Rab10, and Rab14 were not affected by either Tbc1d1 or Tbc1d4 Kd in C2C12 myotubes (Supplementary Fig. 1C and D).

To further elucidate the function of the RabGAPs as a signaling hub between lipid and glucose metabolism in skeletal muscle, we determined gene expression as well as protein abundance of key enzymes of energy substrate flux in RabGAP-deficient skeletal muscle and cultured muscle cells (Supplementary Fig. 2). Pdk4 mRNA and protein were increased in Tbc1d1 Kd myotubes and tibialis anterior (TA) skeletal muscle from D1KO mice (Supplementary Fig. 2A) but not in EDL and SOL muscle from D1KO and D4KO animals (Fig. 3A–D). Also, proteins for oxidative phosphorylation (OXPHOS) in EDL and SOL muscle were not altered compared with wildtype (WT) controls (Supplementary Fig. 2D–G). The FA transporter FAT/CD36 has been shown to contribute to FA uptake in skeletal muscle (20). However, protein abundance of FAT/CD36 was not altered in either EDL or SOL muscle from D1KO and D4KO mice, respectively (Fig. 3A–D), or in Tbc1d1 Kd myotubes and TA muscle from D1KO animals (Supplementary Figs. 2B and 3A). In contrast, protein abundance of the FA transporter SLC27A4/FATP4 was significantly increased in both EDL and SOL muscles from D1KO and D4KO mice compared with the WT littermate controls (Fig. 3A, B, G, and H). However, this difference did not reach significance level in TA muscle from D1KO animals or C2C12 myotubes following Tbc1d1 Kd (Supplementary Figs. 2C and 3A).

To study the relation of Cd36, Tbc1d1, and Tbc1d4, we generated global Cd36-RabGAP double-deficient mice by crossbreeding CD36KO animals with either D1KO or D4KO animals, respectively, as described in the research design and methods section. Intact isolated skeletal muscles from 12- to 16-week-old male knockout mice and respective WT littermate controls were used to analyze FA uptake ex vivo. As illustrated in Fig. 4A, uptake of ³H-palmitate was moderately reduced in EDL muscles from CD36KO animals, but substantially increased in D1KO or D4KO muscles, respectively. Importantly, the elevated FA uptake in muscles from D1KO mice was maintained or even further increased in muscles double-deficient in Cd36 and Tbc1d1 (D1/CD36KO). Similarly, Tbc1d4 knockout increased FA uptake in Cd36-deficient muscles (D4/CD36KO). Interestingly, D1/CD36KO muscles showed higher FA uptake than D4/CD36KO muscles (Fig. 4A).

To investigate the contribution of SLC27A4/FATP4 in the elevated FA uptake in RabGAP-deficient muscle cells, we conducted Kd experiments in C2C12 myotubes and measured ³H-palmitate uptake following depletion of Slc27a4/Fatp4 and Cd36, respectively, as well as combined Kd of Slc27a4/Fatp4 plus Tbc1d1 or Tbc1d4 (Kd efficiency shown in Supplementary Fig. 3B). In C2C12 myotubes, Kd of Slc27a4/Fatp4 resulted in a moderate reduction in ³H-palmitate uptake that did not reach statistical significance (Fig. 4B), whereas Cd36 depletion had no effect on FA uptake (Supplementary Fig. 3C). Conversely, Kd of Tbc1d1 or Tbc1d4 resulted in a substantial increase in FA uptake compared with control cells. Combined Kd of Slc27a4/Fatp4 and each of the RabGAPs completely abrogated the increase in FA transport observed in either Tbc1d1 or Tbc1d4 Kd cells (Fig. 4B). Kd of Slc27a4/Fatp4 did not lead to changes in Cd36 gene expression and vice versa (Supplementary Fig. 3D).

To further investigate the potential mechanism of RabGAP-regulated FA transport, we conducted siRNA-mediated Kd experiments of Tbc1d1 in C2C12 myotubes and measured plasma membrane and total SLC27A4/FATP4 abundance via Western blot analysis. Interestingly, membrane SLC27A4/FATP4 localization was increased upon Tbc1d1-deficiency compared with NT control cells (Fig. 4C). In accordance to our previous results (5), ex vivo FAO was enhanced in EDL muscle from D1KO and D4KO and SOL muscle from D1KO mice, respectively, following IVE of NT siRNA oligonucleotides. Most notable, Kd of Slc27a4/Fatp4 via IVE technology abolished the elevated rate of FAO in EDL and SOL muscle of D1KO and D4KO, respectively (Fig. 4D–F).

Next, we investigated distribution of different lipid species in skeletal muscle from D1KO mice. Despite the fact that the overall gene expression profile of cultured muscle cells or skeletal muscle after TBC1D1 depletion did not show marked changes (Supplementary Fig. 4A and C), we speculated that the proportions of cellular lipid species might be altered due to the enhanced FA uptake and oxidation in D1KO skeletal muscle. Total amount of FAs showed a trend to increase in gastrocnemius muscle of D1KO (Fig. 5A) with changes in the composition of different FA species. In skeletal muscle from D1KO mice, the amount of SFA was reduced (Fig. 5B and E), MUFA content was increased (Fig. 5C and E), and no change in polyunsaturated FA levels were observed (Fig. 5D and E). A more detailed analysis of the individual percent of distinct FA species within the muscle revealed that the amount of four distinct FA species was different between the two genotypes. Skeletal muscle from D1KO mice showed lower amounts of saturated palmitic acid (C16:0) and polyunsaturated arachidonic acid (C20:4). In contrast, levels of monounsaturated palmitoleic acid (C16:1) and oleic acid (C18:1) were significantly increased in gastrocnemius muscle samples from D1KO animals (Fig. 5F). Deduced from the measured values for each FA, enzyme activity of stearoyl-CoA desaturase-1, an enzyme that catalyzes the conversion of C16:0 and C18:0 SFAs into the MUFAs palmitoleic acid (C16:1) (Fig. 5G) and oleic acid (C18:1) (Fig. 5H) was increased in D1KO skeletal muscle. In contrast, activity of Δ5-desaturase, calculated as the ratio of arachidonic acid (C20:4) and α-Linoleic acid (C18:3) was significantly decreased in gastrocnemius muscle from D1KO mice (Fig. 5I).

In the current study, we investigated the contribution of the RabGAPs TBC1D1 and TBC1D4 to FA uptake and oxidation in skeletal muscle. Our results demonstrate that both RabGAPs regulate entry of LCFAs through a RabGTPase-dependent pathway that involves the LCFA transport protein 4 (SLC27A4/FATP4).

Previous studies showed that ablation of either Tbc1d1, Tbc1d4, or both RabGAPs in skeletal muscle results in greatly reduced insulin-stimulated glucose uptake due to compromised trafficking of the GLUT4 transporter (1,5,13–16,40,41). Moreover, we and others reported that RabGAP-deficiency also results in increased uptake and oxidation of palmitic acid in skeletal muscle and cultured muscle cells (16,42). Because overexpression of intact Tbc1d1, but not a GAP-inactive R941K mutant, decreased palmitate uptake and oxidation in both isolated skeletal muscle and cultured muscle cells (16,43), we speculated that the two RabGAPs may regulate uptake of glucose and FAs through distinct Rab-dependent pathways.

Neither the abundance of proteins involved in mitochondrial OXPHOS, nor mitochondrial copy number or citrate synthase activity was altered in Tbc1d1-deficient skeletal muscle (40). Electroporation-mediated overexpression of Tbc1d1 was reported to reduce FAO and β-hydroxyacyl-CoA dehydrogenase activity, a key enzyme of mitochondrial β-oxidation (43) whereas β-hydroxyacyl-CoA dehydrogenase activity was unaltered in skeletal muscle from Tbc1d1 knockout rats that displayed increased in vivo and ex vivo skeletal muscle fat oxidation (42). In this study, mitochondrial OXPHOS proteins were not altered in EDL and SOL muscle of D1KO and D4KO mice, and consistent differences in key enzymes for energy metabolism, such as PDK4 were not observed in the muscles, presumably reflecting the complexity of energy metabolism in different muscle types and fibers (44,45). An increased influx of LCFAs may therefore not be driven by increased mitochondrial activity but rather result from elevated activity or abundance of FA transporters or other nonmitochondrial metabolizing enzymes (46). In accordance with previous studies in skeletal muscle (16,42), we show here that Kd of both Tbc1d1 or Tbc1d4 specifically increases uptake of both saturated and unsaturated LCFAs, whereas uptake of SCFAs into skeletal muscle cells remains unaltered. This indicates that RabGAP-deficiency increases FA transport via specialized LCFA transporter enzymes rather than activating passive diffusion through the plasma membrane, which has been proposed to be the predominant mechanism for SCFA uptake (19,20,47).

Several RabGTPases have been implicated to play roles in the translocation of GLUT4 including Rab4, Rab8a, Rab8b, Rab10, Rab11, Rab13, and Rab14 (48,49). Here, we show that Kd of Rab8a, Rab8b, Rab10, and Rab14 decreases palmitate uptake in both basal and insulin-stimulated muscle cells (Supplementary Fig. 4B), whereas depletion of Rab28, which is also a substrate for TBC1D1 and TBC1D4 (3), has no effect (Fig. 2E). These data suggest that FA uptake utilizes a specific subset of RabGTPases downstream of TBC1D1 and TBC1D4, which partially overlaps with the substrate specificity of the RabGAPs in vitro (3). More specifically, Rab8a may represent the major RabGAP substrate mediating FAO as Kd of Rab8a but not Rab10 in intact isolated skeletal muscle via IVE technology led to a decrease of ³H-palmitate oxidation (Fig. 2F and Supplementary Fig. 1B). A reduction in FAO was detected following Rab8a Kd in intact SOL but not EDL muscle, indicating a more complex relationship between RabGAP-RabGTPase interaction and different types of skeletal muscle fibers. Divergent mechanisms of lipid accumulation and mitochondrial function have already been described for oxidative/slow-twitch and glycolytic/fast-twitch muscle fibers (50,51). Further studies are required to investigate the specific contribution of individual RabGTPases to FA uptake and metabolism in muscle cells.

Our findings indicate a possible involvement of LCFA-specific metabolizing enzymes and/or transporters expressed in skeletal muscle including FA translocase FAT/CD36 and FA transporters of the FATP protein family (25,52). FAT/CD36 has been implicated in FA uptake in skeletal muscle and the heart (30,53). In analogy to the insulin-regulated GLUT4, FAT/CD36 was found to undergo recruitment from intracellular pools to the plasma membrane in response to insulin and contraction (24,54,55). Consistent with a possible role of RabGAPs in the regulation of FAT/CD36 traffic and subcellular localization, Kd of Tbc1d4 as well as overexpression of constitutively active Rab8a, Rab10, and Rab14 led to a redistribution of FAT/CD36-associated immunofluorescence to the plasma membrane in cardiomyocytes (23). In previous studies, RabGAPs were also related to the expression levels of Cd36. Mikłosz et al. (22,56) found in rat L6 myotubes that Tbc1d4 silencing increased the expression levels of Cd36, whereas transient overexpression of Tbc1d1 in mouse skeletal muscle led to reduced palmitate oxidation without altering FAT/CD36 protein abundance (43). Tbc1d1-deficient TA muscles from our knockout mice did not display increased mRNA and protein levels of FAT/CD36, although there was a trend toward increased levels of expression. For further in-depth analysis of a possible function of FAT/CD36 in RabGAP deficiency, we followed a genetic approach by generating Cd36/RabGAP double-deficient mice (D1/CD36KO; D4/CD36KO) and analyzing FA uptake into intact isolated skeletal muscle. As expected, skeletal muscle from CD36KO mice displayed reduced uptake of palmitate compared with WT littermates as reported previously (30). However, muscles from D1/CD36KO and D4/CD36KO double knockout animals showed similarly elevated FA uptake as compared with the single RabGAP knockout muscles with an intact Cd36 gene. These findings rule out a major contribution of FAT/CD36 in the observed elevated FA uptake in response to RabGAP-deficiency in mouse skeletal muscle. In addition to FAT/CD36, the FA transport protein SLC27A4/FATP4 has been reported to catalyze uptake of LCFAs into skeletal muscle where its intrinsic acyl-CoA activity may contribute to its transport activity (57). Similar to FAT/CD36, plasma membrane SLC27A4/FATP4 was found to be increased in response to stimuli such as insulin and contraction in skeletal muscle (54). EDL and SOL skeletal muscle from both D1KO and D4KO mice displayed an increase in SLC27A4/FATP4 protein, indicating that this transporter might contribute at least in part to the elevated FAO in RabGAP-deficient skeletal muscle. However, a contribution of this FA transporter in the elevated fat utilization of the animals may not be strictly dependent on protein amount alone but involve intracellular trafficking and translocation processes. As FATP4-null mice display neonatal lethality (58,59), crossbreeding and generation of FATP4/RabGAP double-deficient mice was not possible. Therefore, we first analyzed palmitate uptake in cultured muscle cells after Kd of the FATP4 gene Slc27a4 and each of the two RabGAPs, revealing that the increase in palmitate uptake in RabGAP-deficiency is strictly dependent on the presence of SLC27A4/FATP4 in these cells. Likewise, Kd of Slc27a4/FATP4 completely abrogated the elevated FAO in intact isolated EDL and SOL muscle from D1KO and D4KO mice. Interestingly, Kd of Tbc1d1 or Tbc1d4 was associated with an enrichment of Slc27a4/FATP4 protein in the plasma membrane, indicating that RabGAPs may regulate both abundance and subcellular localization of the protein. However, further studies are required to elucidate the precise molecular mechanism of RabGAP-dependent regulation of FA transport.

Global Tbc1d1 deficiency in mice is associated with a moderate increase in MUFA and a concomitant decrease of SFAs in skeletal muscle. As MUFA content of muscle lipids is increased with insulin resistance and obesity and has been shown to correlate with Δ9-desaturase activity in human muscle cells, it appears that RabGAPs may contribute to systemic insulin sensitivity through alterations in lipid composition (60). A recent study in L6 myotubes showed that Kd of Tbc1d4 also led to alterations in the ratio of SFA and MUFA (56). Interestingly, indigenous people of Greenland frequently carry a loss-of-function mutation in the related TBC1D4 gene, which may reflect a genetic adaptation to the traditional hypoglycemic, fat-rich diet of the Inuits (61,62). However, the alterations in the FA profile of gastrocnemius muscle from D1KO mice were rather minor, which might be due to the diet and age of the mice, as well as the skeletal muscle type analyzed. Further studies are required to investigate alteration in metabolic fluxes of RabGAP-deficient muscle tissue.

Collectively, our data demonstrate that both TBC1D1 and TBC1D4 specifically control entry of LCFAs into skeletal muscle through a mechanism that requires a subset of RabGTPases involved also in GLUT4 translocation. FA transporter SLC27A4/FATP4 is a likely candidate mediating RabGAP-dependent lipid uptake, resulting in an altered lipid composition. A possible regulatory pathway of RabGAP-dependent SLC27A4/FATP4 trafficking within the muscle cell can be hypothesized, the exact mechanisms, however, remain to be determined. Unraveling the mechanisms underlying RabGAP-dependent lipid flux in skeletal muscle is an important step toward the understanding of skeletal muscle adaptations during the pathophysiology of insulin resistance and type 2 diabetes.

Acknowledgments. The authors thank Angelika Horrighs, Anette Kurowski, Heidrun Podini, Carina Heitmann, Dagmar Grittner, Antonia Osmers, Annette Schober, Lothar Bohne, Martina Schiller, Jennifer Schwettmann, and Denise Schauer (all from German Diabetes Center, Leibniz Center for Diabetes Research at Heinrich Heine University Duesseldorf) for expert technical assistance. The authors also thank Dr. Maria Febbraio (Cleveland Clinic, Cleveland, OH) for generously providing the Cd36 knockout mice.

Funding. This work was supported by the German Center for Diabetes Research (DZD) (82DZD00202 to H.A.-H.) of the Federal Ministry for Education and Research and the Ministry of Science and Research of the State North Rhine-Westphalia and was funded, in part, by grants from the Deutsche Forschungsgemeinschaft (CH1659/1-1 to A.C.), and the European Foundation for the Study of Diabetes (EFSD)/Novo Nordisk program (to H.A.-H.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. T.B., A.C., and H.A.-H. wrote the manuscript and analyzed and interpreted the data. T.B., L.E., S.E., I.Z., I.S., F.M., C.S., H.B., S.C., Z.Z., and J.K. performed the experiments and analyzed data. A.C. and H.A.-H. were involved in the study design and contributed to data interpretation. A.C. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.