Rab (Ras-related proteins in brain) GTPases are key proteins responsible for a multiplicity of cellular trafficking processes. Belonging to the family of monomeric GTPases, they are regulated by cycling between their active GTP-bound and inactive GDP-bound conformations. Despite possessing a slow intrinsic GTP hydrolysis activity, Rab proteins rely on RabGAPs (Rab GTPase-activating proteins) that catalyze GTP hydrolysis and consequently inactivate the respective Rab GTPases. Two related RabGAPs, TBC1D1 and TBC1D4 (=AS160) have been described to be associated with obesity-related traits and type 2 diabetes in both mice and humans. Inactivating mutations of TBC1D1 and TBC1D4 lead to substantial changes in trafficking and subcellular distribution of the insulin-responsive glucose transporter GLUT4, and to subsequent alterations in energy substrate metabolism. The activity of the RabGAPs is controlled through complex phosphorylation events mediated by protein kinases including AKT and AMPK, and by putative regulatory interaction partners. However, the dynamics and downstream events following phosphorylation are not well understood. This review focuses on the specific role and regulation of TBC1D1 and TBC1D4 in insulin action.
Rab (Ras-related proteins in brain) GTPases comprise the largest family within the monomeric 27 kDa Ras-like G protein superfamily [1–5]. Rab proteins cycle between active GTP-bound ‘on' and inactive GDP-bound ‘off' conformational states [5,6]. There are two enzymes controlling the Rab cycle: (a) guanine nucleotide exchange factors activate Rab proteins via promoting GDP release and subsequently GTP binding of the Rab proteins. Thereafter, (b) Rab GTPase-activating proteins (RabGAPs) inactivate Rabs via catalyzing the conversion of GTP into GDP by stimulating GTP hydrolysis [7,8]. The molecular switch between the active and the inactive form of Rabs controls membrane trafficking pathways such as vesicle formation, vesicle transport, vesicle tethering/docking and vesicle fusion with the target membrane [5,9]. RabGAPs are often, but not always multi-domain proteins possessing a highly conserved TBC (or GAP) domain. This GAP domain contains ∼200–300 amino acids and is responsible for catalyzing GTP hydrolysis of Rab proteins via a ‘dual arginine glutamine finger' mechanism . Based on this mechanism, all known RabGAPs except Rab3GAP possess this catalytic motif, indicating that they act as functional GAP proteins. So far, more than 40 different RabGAPs have been identified based on sequence homology [11,12], and at least 30 were studied in more detail with respect of their putative biological functions. This review will focus on (a) the specific function of the two RabGAPs, TBC1D1 and TBC1D4, as downstream AKT targets in insulin-stimulated GLUT4 traffic, and (b) the molecular mechanism of how phosphorylation may regulate the activity and function of TBC1D1 and TBC1D4.
GLUT4 mediates insulin-regulated glucose uptake
In mammals, a total of 13 facilitative hexose transporter genes have been identified so far [13,14]. GLUT4 (SLC2A4 gene) constitutes the main insulin-responsive glucose transporter and is prominently expressed in a limited number of insulin-responsive cell types including skeletal muscle, white and brown adipose cells, and cardiomyocytes . In non-stimulated cells, GLUT4 resides in intracellular storage vesicles (GSVs) and is constantly but slowly recycling between this compartment and the plasma membrane (PM). Insulin (and skeletal muscle contraction) strongly accelerates exocytosis of the GLUT4-containing vesicles, leading to a rapid and reversible redistribution of GLUT4 from GSVs to the PM, and, subsequently, to increased influx of glucose into the cells [16,17]. While the molecular mechanism of GLUT4 translocation is still unclear, there is compelling evidence that a phosphorylation cascade downstream from the insulin receptor (IR) triggers directional movement of GLUT4 vesicles and fusion with the PM . Currently, the most distal targets of the IR signaling cascade projecting to the GLUT4 translocation pathway are represented by TBC1D1 and TBC1D4, both being identified initially as AKT substrates in 3T3-L1 adipocytes and mouse skeletal muscle [19–22].
TBC1D1 and TBC1D4 are associated with metabolic traits and diseases
Mutations in TBC1D1 have been linked with obesity-related traits in humans [23–25] and mice [26–28]. Intriguingly, genetic variants of TBC1D1 were also associated with traits related to muscle growth and fat content in domesticated pigs and chicken [29–31], suggesting a possible evolutionarily conserved role of TBC1D1 in regulating body composition in vertebrates. Moreover, mutations in TBC1D4 have been associated with insulin resistance in humans [32,33]. Importantly, a common loss-of-function mutation in TBC1D4 (p.Arg684Ter) has been discovered in the Greenlandic Inuit population where the homozygous carriers of the mutant allele show severely impaired postprandial disposal of glucose and a more than 10-fold increased risk for developing type 2 diabetes (T2D) . In fact, TBC1D4 (p.Arg684Ter) is the major genetic cause for T2D in both the Greenlandic and Canadian Inuit population . There is now compelling evidence that both TBC1D1 and TBC1D4 are critical regulators in the insulin-stimulated glucose uptake [28,36–39].
AKT and AMPK phosphorylation sites in TBC1D1 and TBC1D4 show partial overlap
Insulin-stimulated activation of AKT has been shown to result in phosphorylation of TBC1D4 at several Ser/Thr residues (Figure 1). Overexpression of TBC1D4 mutants where the AKT phosphorylation sites were replaced by alanine impaired insulin-stimulated GLUT4 translocation in 3T3-L1 adipocytes, whereas overexpression of a GAP-inactive mutant had no effect . Similar studies were conducted with TBC1D1  and, as a result, it was concluded that this phosphorylation shuts down the GAP function, resulting in a de-inhibition of Rabs and triggering GLUT4 exocytosis (Figure 2, model 1). In particular, AKT phosphorylates TBC1D4 at least at six phosphorylation consensus motifs (RSRCSS318V, RRRHAS341A, RSLTSS570L, RGRLGS588M, RRRAHT642F and RKRTSS751T) [40,41]. Among these motifs, only those corresponding to Ser570 and Thr642 in TBC1D4 are also found in the homologous TBC1D1. AKT phosphorylates TBC1D1 at Ser507 and Thr596 in response to insulin (Figure 1) . Overexpression of inactivating mutations of the AKT phosphorylation sites in TBC1D4 impaired GLUT4 translocation, showing that phosphorylation by AKT is essential for insulin-stimulated GLUT4 translocation to the PM . In addition to the insulin-stimulated AKT phosphorylation, phosphorylation by AMPK in response to exercise has been suggested as an independent mechanism for regulating TBC1D1 and subsequently affecting GLUT4 trafficking and cellular glucose transport. AMPK has been shown to phosphorylate TBC1D1 at Ser 237, Thr489, Ser660 and Ser700, while TBC1D4 was shown to be phosphorylated at Ser341, Ser588 and Ser751, in vitro [43–46]. However, phosphorylation of distinct sites may occur after activation of both AMPK and AKT, including Ser596 in TBC1D1 and Ser341, Ser642 and Ser704 in TBC1D4 (Figure 1) [36,46,47]. Even though all known phosphorylation sites do not map even closely to the GAP domains of both TBC1D1 and TBC1D4 (Figure 1), the study of the Saltiel group provided evidence that phosphorylation of TBC1D4 altered the activation state of Rab10, in vivo .
Possible regulation of RabGAPs in GLUT4 translocation.
Knockout of TBC1D1 and TBC1D4 alters GLUT4 targeting
Early overexpression studies of phosphosite mutations in Tbc1d1 and Tbc1d4 in cultivated adipocytes indicated a possible mechanism how GLUT4 translocation might be regulated at the RabGAP level [40,42,49]. In the basal state, retention of GLUT4-containing vesicles is achieved by constitutively active RabGAPs that inactivate (presumably vesicle-associated) Rab GTPases, thus preventing GLUT4 from translocating to the PM. Phosphorylation of TBC1D1 and TBC1D4 is thought to inhibit their intrinsic GAP activity, leading to increased levels of active, GTP-bound Rab GTPases, which triggers GLUT4 translocation (Figure 2, model 1). Unexpectedly, RabGAP knockout mice showed substantially reduced abundance of GLUT4 protein in tissues normally expressing the respective RabGAP protein [28,37–39,50]. More specifically, GLUT4 protein levels are reduced in oxidative Soleus muscle and adipocytes from Tbc1d4- but not from Tbc1d1-deficient mice, whereas the latter present decreased GLUT4 abundance in glycolytic muscles like the EDL (Extensor digitorum longus) muscle. Interestingly, Slc2a4 (GLUT4 gene) levels are lower in Tbc1d4 knockout adipocytes but not in Soleus muscle from the same mice [28,37,38,50]. Because mRNA levels of the Slc2a4 gene are unchanged in skeletal muscles from RabGAP knockout mice, the reduced amount of GLUT4 probably reflects missorting and posttranslational degradation of the protein in this tissue . Interestingly, quantitation of GLUT4 protein using the membrane-impermeable glucose photoprobe ATB-BMPA demonstrated that, in isolated adipocytes from RabGAP knockout mice, the proportion of basal GLUT4 in the PM is increased, as would be expected from interference with a retention mechanism. The same observation has been made for Gastrocnemius and Quadriceps skeletal muscles, whereas GLUT4 content in the PM is not altered at the basal state in the oxidative Soleus muscle, indicating differences in GLUT4 protein turnover between the tissues [28,37,38,50]. While TBC1D1 and TBC1D4 are probably involved in regulating the intracellular retention of GLUT4 in the absence of a stimulus, other sorting steps, including but not limited to endosomal/lysosomal trafficking, may also be controlled by the RabGAPs. In concordance with the tissue-specific expression pattern of the two RabGAPs, numerous in vitro and in vivo studies demonstrated the specific impact of each of the two RabGAPs on glucose uptake and/or GLUT4 translocation in different types of skeletal muscle and in adipose cells. All studies published so far show an impairment of insulin-stimulated glucose uptake in response to the lack of Tbc1d4 in adipose cells [37–39]. Two studies also reported a moderately increased basal glucose transport in isolated primary adipocytes [28,38,50], whereas a study from our laboratory did not observe significant changes in basal glucose uptake in fat cells from Tbc1d4-deficient mice . These conflicting results may result from differences in genetic background, diet, age and other environmental factors.
In skeletal muscle, the function of the two RabGAPs is highly related to the expression pattern of the respective proteins. Deletion of Tbc1d4, which is predominantly expressed in the oxidative Soleus muscle, results in reduced insulin-, AICAR- and contraction-stimulated glucose transport in exactly this muscle without affecting glycolytic skeletal muscles. Conversely, deletion of Tbc1d1, the predominant RabGAP in glycolytic EDL muscle reduces insulin-, AICAR- and contraction-stimulated glucose transport in those fibers without affecting adipose cells and oxidative muscle. Because adipose cells mainly express Tbc1d4, it explains that knockout of Tbc1d4 but not Tbc1d1 reduces insulin-stimulated glucose transport in these cells. In oxidative Soleus muscle and glycolytic EDL muscle, basal glucose uptake was not increased in response to the lack of either of the two RabGAPs as observed in adipose cells [28,37,38]. However, in Gastrocnemius and Quadriceps muscle, both of which have a more heterogeneous fiber-type composition, PM GLUT4 was increased in the basal state in RabGAP knockout mice, resulting in a decreased fold change in the insulin response [28,38]. However, in general, deficiency in both RabGAPs leads to a combined phenotype, showing reduced insulin-stimulated glucose uptake into isolated adipocytes and all isolated skeletal muscle types examined, ex vivo [28,37–39,51].
While the results for total GLUT4 protein abundance in adipocytes and skeletal muscle are consistent throughout the different studies, quantification of GLUT4 membrane localization in both basal and insulin-stimulated states has produced somewhat contradictory findings as different fold changes for the insulin-stimulated GLUT4 membrane translocation/cell surface expression were reported for the different RabGAP knockout models [28,37–39]. These differences might be explained by the use of different technologies (e.g. fractionation, photolabeling, tracer experiments) for measuring the abundance of GLUT4 in the PM in different cell and muscle types. In summary, RabGAP deficiency leads to alterations in GLUT4 trafficking, GLUT4 protein abundance and insulin-stimulated glucose transport. The causal relation between the observed phenotypes remains to be clarified and more in-depth mechanistic analysis is needed to resolve current discrepancies.
A growing list of upstream regulators and downstream effectors of TBC1D1 and TBC1D4
Perhaps due to the large numbers of Rabs and associated regulatory proteins, only a limited number of Rab substrates and possible RabGAP binding/interaction partners are known.
The Lienhard laboratory has conducted a series of RabGAP assays in vitro using recombinant GST-GAP-domain fusion proteins expressed in Escherichia coli to identify possible Rab substrates of TBC1D4 and TBC1D1 [42,49]. Rab2a, 8a, 8b, 10 and 14 were shown to be substrates for both, TBC1D4 and TBC1D1 [42,49,52], consistent with the high sequence homology of the two GAP domains in TBC1D1 and TBC1D4 at the amino acid level. In particular, Rab10 has been suggested to be required for insulin-stimulated GLUT4 translocation in adipocytes [53–55], while Rab8a has been implicated in regulating GLUT4 translocation in skeletal muscle [56,57]. Recent studies indicated that additional Rabs such as Rab13 and Rab28 are downstream substrates for TBC1D1 and TBC1D4 in insulin target cells, and may play roles in regulating GLUT4 traffic as well [58–60]. Moreover, we recently showed that Rab32, 33B and Rab34 are also substrates for the recombinant GAP domains of both TBC1D1 and TBC1D4 in vitro . However, the possible involvement of these Rabs in GLUT4 trafficking remains to be investigated further. Importantly, the question remains open to what degree the recombinant GAP domains reflect the native full-length RabGAP proteins in terms of substrate specificity, efficacy and regulation. Because attempts to purify and characterize full-length RabGAPs have failed [42,49], validation of Rab substrates for full-length TBC1D1 and TBC1D4 has not been conducted.
In 2006, Ramm et al. identified ubiquitously expressed 14-3-3β as the interaction partner possibly involved in TBC1D4-regulated GLUT4 translocation , and this finding was extended to TBC1D1 as well . 14-3-3 interaction with the RabGAPs depends on phosphorylation of specific Ser/Thr residues . Phosphorylation of TBC1D1 at Ser237 and Thr596 and TBC1D4 at Ser341 and Thr642 were shown to be essential for 14-3-3 binding [44,62]. Interestingly, 14-3-3 proteins have different affinity to these phosphorylation sites in response to different stimuli . Since the second PTB domain is located between these two phosphorylation sites, binding 14-3-3 might change the function of the second PTB domain [45,62]. It has been demonstrated that phosphorylation increases the capacity of 14-3-3 binding to TBC1D4 [63,64], and this phosphorylation has been shown to be essential for GLUT4 translocation [40,42,44] and whole-body glucose homeostasis . Since the mutant in which the phosphorylation sites were substituted to alanine inhibited GLUT4 translocation [40,45,66], it has been suggested that phosphorylation and 14-3-3 binding may suppress the GAP activity of TBC1D1 and TBC1D4 [40,66] (Figure 2, model 1).
It has been previously shown that insulin-regulated aminopeptidase (IRAP) is associated with the cargo of GSVs . IRAP (=vp165) is a type II membrane protein with a single transmembrane segment, a large extracellular/intravesicular catalytic domain and an N-terminal 109 amino acid cytosolic domain which is involved in its localization during insulin-regulated trafficking [68,69]. The cytosolic domain of IRAP interacts with the second PTB domain of TBC1D4 under basal conditions and thus may keep TBC1D4 in proximity to the GSVs. Upon insulin stimulation TBC1D4 is reported to dissociate from IRAP and, consequently, is released into the cytoplasm  (Figure 2, model 2). In fact, localization of the RabGAPs might be regulated through phosphorylation-dependent interactions with GLUT4 vesicle-recruiting proteins such as IRAP. In this scenario, the RabGAP dissociates from the GLUT4 vesicle and is no longer available to act on the respective Rab proteins. Consequently, Rab GTPases in their GTP-bound form then trigger the redistribution of GLUT4 from GSVs to the PM [40,49] (Figure 2, model 2). Consistent with this hypothesis, Jordens et al. showed that knockdown of IRAP in 3T3-L1 adipocytes resulted in increased basal GLUT4 . However, they found that recruitment of TBC1D4 to GSVs occurred even in the absence of IRAP, indicating a more complex requirement of RabGAP targeting of the GLUT4 vesicles . Further research is required in order to understand how phosphorylation may affect the GAP activity and/or alter subcellular distribution of TBC1D1 and TBC1D4.
APPL1 and APPL2 are adiponectin receptor-binding proteins which were found to interact with AKT and TBC1D1 during GLUT4 translocation signaling . APPL1 and APPL2 share over 50% identity and a similar domain structure. Both are negative regulators of insulin signaling via two distinct mechanisms . Overexpression of APPL2 inhibits insulin-simulated glucose uptake, in vitro and in vivo. This inhibitory effect starts by interaction of TBC1D1 with APPL2 and blocks AKT phosphorylation of Ser235, consequently decreasing Thr596 phosphorylation of TBC1D1 . This may prevent inactivation of TBC1D1 and impair GLUT4 translocation.
It has been established that TBC1D1 and TBC1D4 play important roles in GLUT4 traffic. However, the molecular mechanism of this process and the specific function of each of the two RabGAPs in glucose metabolism are still not fully understood. Interestingly, TBC1D1 and TBC1D4 are similar in structure and function, but the phenotypes in knockout mouse models and individuals carrying mutations are different. This may reflect different roles of the RabGAPs in different tissues and/or different downstream effectors required for glucose and energy homeostasis, respectively. Information about the impact of phosphorylation of TBC1D1 and TBC1D4 on the catalytic GAP activity and/or protein interactions is at present speculative, since our current understanding of the molecular mechanisms of RabGAP regulation is based mainly on in vitro studies that are limited to overexpression of mutants, and biochemical characterization of truncated GAP domains and candidate Rabs expressed in E. coli.
This work was supported by the Ministry of Innovation, Science and Research of the State of North Rhine-Westphalia (MIWF NRW) and the German Federal Ministry of Health (BMG), and was funded in part by grants from the Deutsche Forschungsgemeinschaft [SFB1116] and the German Academic Exchange Service (DAAD).
The Authors declare that there are no competing interests associated with the manuscript.