Insulin increases glucose uptake by increasing the rate of exocytosis of the facilitative glucose transporter isoform 4 (Glut4) relative to its endocytosis. Insulin also releases Glut4 from highly insulin-regulated secretory compartments (GSVs or Glut4 storage vesicles) into constitutively cycling endosomes. Previously it was shown that both overexpression and knockdown of the small GTP-binding protein Rab14 decreased Glut4 translocation to the plasma membrane (PM). To determine the mechanism of this perturbation, we measured the effects of Rab14 knockdown on the trafficking kinetics of Glut4 relative to two proteins that partially co-localize with Glut4, the transferrin (Tf) receptor and low-density-lipoprotein-receptor-related protein 1 (LRP1). Our data support the hypothesis that Rab14 limits sorting of proteins from sorting (or ‘early’) endosomes into the specialized GSV pathway, possibly through regulation of endosomal maturation. This hypothesis is consistent with known Rab14 effectors. Interestingly, the insulin-sensitive Rab GTPase-activating protein Akt substrate of 160 kDa (AS160) affects both sorting into and exocytosis from GSVs. It has previously been shown that exocytosis of GSVs is rate-limited by Rab10, and both Rab10 and Rab14 are in vitro substrates of AS160. Regulation of both entry into and exit from GSVs by AS160 through sequential Rab substrates would provide a mechanism for the finely tuned ‘quantal’ increases in cycling Glut4 observed in response to increasing concentrations of insulin.
- glucose transporter isoform 4 (Glut4)
Insulin regulates blood glucose homoeostasis in part by increasing glucose uptake into muscle and adipose tissue [1–4]. Glucose uptake in adipocytes and muscle is rate-limited by the number of facilitative glucose transporters present in the plasma membrane (PM). Insulin increases glucose uptake by redistributing glucose transporter isoform 4 (Glut4) from intracellular compartments to the PM, a process known as Glut4 translocation. Glut4 is actively cycling between the PM and intracellular pools under both basal and insulin-stimulated conditions. The proportion of Glut4 at the cell surface is dependent on the relative rates of insertion of Glut4 into the PM (exocytosis) and clearance from the PM (endocytosis). It is also dependent on the amount of Glut4 that is cycling. Insulin regulates Glut4 translocation by increasing its rate of exocytosis relative to endocytosis. Insulin also regulates the distribution of Glut4 between a constitutive actively cycling endosomal pool and a non-cycling/very slowly cycling sequestered pool. The total amount of Glut4 expressed in cells is also regulated, through both transcription and turnover/degradation.
Glut4 is internalized from the PM into sorting (or ‘early’) endosomes (SE) through a slow endocytic pathway [5–9]. Glut4 is then sorted from endosomes into a slow constitutive recycling pathway through endosomal recycling intermediate compartments (ERCs). The transferrin (Tf) receptor is delivered to the same SE as Glut4. However, it is internalized and recycled through fast pathways that are kinetically distinct from those followed by Glut4. The α2-macroglobulin (α2M) receptor low-density-lipoprotein-receptor-related protein 1 (LRP1) cycles through both the slow (Glut4) and the fast (Tf receptor) constitutive pathways. In adipocytes, Glut4 and LRP1 also traffic through a specialized highly regulated insulin-sensitive sequestration pathway through Glut4 storage vesicles (GSVs). The Tf receptor is largely excluded from the GSV pathway. Thus, the trafficking of these three proteins identifies three kinetically distinct endosomal cycling pathways in adipocytes. Analysis of both kinetics and subcellular distribution data for these proteins indicates that insulin regulates three steps in the sequestration pathway: sorting from endosomes into the GSVs, release/priming of the sequestered GSVs, and tethering/docking/fusion of primed GSVs . Adipocytes also express an efficient mechanism for retrieval of Glut4 and LRP1 from lysosomal degradation . Identification of the protein machineries involved in each of these steps is essential to elucidate the mechanism of regulation of these processes by insulin.
In adipocytes, the release/priming of GSVs is regulated by the insulin-sensitive Rab GTPase-activating protein Akt substrate of 160 kDa (AS160) . This regulation occurs through the inhibition of the GTP-loading and activation of a Rab protein(s) bound to the sequestered GSVs [10,11]. Knockdown of AS160 increases the basal rate of Glut4 exocytosis, increasing cell surface Glut4 [7,12]. Expression of the constitutively active phosphorylation mutant 4P-AS160 inhibits Glut4 translocation [10,11]. Insulin also regulates sorting from endosomes into the GSVs, through an unknown mechanism . AS160 knockdown inhibits this sorting step, presumably through dysregulation of an additional Rab(s) [7,8]. Effects of AS160 have been observed in endosomes as well as at the PM, consistent with the idea that AS160 regulates multiple steps in Glut4 trafficking [7,8,13–15]. Tethering/docking/fusion is regulated by Akt and phosphoinositide 3-kinase (PI3K) through an AS160-independent mechanism , possibly through direct regulation of Rab effector proteins [14,16].
Glut4 traffics through multiple pathways, therefore many Rabs are found on Glut4-containing compartments, including Rabs 1–8, 10, 11, 14, 18 and 35 [17–19]. Of these, Rabs 2, 8, 10 and 14 are AS160 substrates in vitro [18,19] and knockdown of Rabs 8, 10 and 14 affect cell-surface Glut4. Interestingly, Rabs 10 and 14 have the largest roles in Glut4 trafficking in adipocytes [20–23], whereas Rabs 8, 13 and 14 have more dominant roles in muscle cells [24–26]. Consistent with a role downstream of AS160, knockdown of Rab10 reverses the elevation of cell-surface Glut4 caused by AS160 knockdown in adipocytes . Likewise, overexpression of either Rab8 or Rab14 in muscle cells is sufficient to overcome the decrease in cell-surface Glut4 caused by expression of constitutively active 4P-AS160 .
Knockdown of Rab10 decreases cell-surface Glut4 and can partially rescue the basal increase observed in AS160-knockdown adipocytes [20–23]. Overexpression of either wild-type Rab10 or a GTP hydrolysis-resistant mutant (Rab10 QL) increases basal cell-surface Glut4 [20,23]. Rab10 is found on Glut4 vesicles that fuse to the PM in insulin-stimulated, but not in basal cells . These vesicles do not contain the Tf receptor. Multiple functional connections between activated Rab10 and the exocyst complex have been described [22,23,27,28]. Together, these data support the hypothesis that Rab10-GTP primes GSVs for fusion by allowing interaction with effectors of vesicle tethering. Consistent with this hypothesis, the effects of Rab10 knockdown on the kinetics of trafficking of both Glut4 and α2M–LRP1 are exactly the phenotype expected for the Rab that limits GSV exocytosis .
Rab14 is also required for Glut4 translocation; knockdown of Rab14 decreases insulin-stimulated cell-surface Glut4 [22,23]. Surprisingly, overexpression of either wild-type Rab14 or GTP hydrolysis-deficient (constitutively activated) Rab14 QL also decreases cell-surface Glut4 in adipocytes . Expression of these constructs leads to enlarged SE, which accumulate Glut4. The Tf receptor is also found in these compartments. A role for Rab14 in endosomal trafficking is also suggested by total internal reflection fluorescence (TIRF) microscopy, which noted the presence of Rab14 on Glut4-containing vesicles derived from the constitutive endosomal pathway that were fusing with the PM in both basal and insulin-stimulated adipocytes . The microscopy data suggest that Rab14 limits fusion of constitutive ERC-derived vesicles, whereas Rab10 limits exocytosis of the insulin-regulated GSVs. However, the effects of Rab14 knockdown on the trafficking of sorting sequence mutants of Glut4 strongly supports a role for Rab14 in the trafficking of Glut4 from endosomes into the specialized Rab10-dependent GSV pathway rather than in vesicle fusion .
Many variables affect cell-surface Glut4, including the level of Glut4 expression, the proportion of Glut4 in cycling compared with non-cycling compartments (e.g. degradative and biosynthetic compartments, or inactive GSVs), the relative distribution of Glut4 in the constitutive compared with regulated pathways, as well as the relative rate constants of exocytosis (kex) and endocytosis (ken) of Glut4. Furthermore, kex can be affected through alterations in either the constitutive ERC or regulated GSV exocytic pathways, due to changes in either entry into or exit from these pathways. Therefore, to determine how Rab14 affects cell-surface Glut4, the effects of depletion of Rab14 on the trafficking kinetics, expression and subcellular distribution of Glut4 were measured in both adipocytes and fibroblasts. The effects on the trafficking of α2M–LRP1 and the Tf receptor were also examined. Our data and mathematical modelling support the proposition that the AS160 substrates Rab14 and Rab10 regulate sequential steps in Glut4 trafficking–Rab14 limits sorting/trafficking from endosomes into the highly regulated GSV sequestration pathway, whereas Rab10 limits exocytosis from the GSVs. However, Rab14 also limits trafficking through the constitutive Rab10-independent ERC pathway, suggesting that it functions at a step shared in the ERC and GSV pathways.
The experimental data described in the present paper were generated in high-throughput format experiments and were collected at the same time as the Rab10 data previously described ; the control data are identical, allowing direct comparison between these studies. Data were also simultaneously collected for Rab8a and Rab8b (results not shown).
3T3-L1 cells were obtained from the A.T.C.C. Cells were passaged as fibroblasts, plated into 96-well plates and either analysed after reaching confluence or differentiated into adipocytes .
3T3-L1 fibroblasts were infected with lentivirus encoding haemagglutinin (HA)–Glut4/GFP as described in . After recovery from lentiviral infection, cells were infected with retroviruses expressing shRNAs targeting Rabs, or a control sequence not targeting any mouse gene as described in . The retroviruses were a gift from Dr Gustav Lienhard (Dartmouth Medical School, Hanover, NH, USA) [20,21]. The effect of shRNA expression on the level of expression of Rab14 in our cells has been previously reported [22,29]. Expression of Rab14, but not Rab10, reversed the inhibition of Glut4 translocation observed in our Rab14-knockdown cells . Expression of constitutively active Rab14 or overexpression of wild-type Rab14 had the same effect as knockdown of Rab14 on Glut4 translocation .
Antibodies and reagents
HA.11 monoclonal antibody (α-HA; Covance) was purchased as ascites and purified and labelled with Alexa Fluor 647 (AF647) as described in . α2M was purchased from AssayPro, labelled with AF647, then activated using methylamine (Sigma), as described in .
Glut4 kinetics assays
Kinetic experiments were performed as described in . For all experiments, cells were serum-starved for 2 h at 37°C in low-serum medium [LSM: 0.5% FBS in Dulbecco's modified Eagle's medium (DMEM)]. Insulin (100 nM) was added to the medium, and cells were further incubated as indicated.
To check the validity of the rate constants measured in adipocytes, the fraction of total Glut4 expected at the PM (PMcalc) was estimated using kex, Ymax and ken determined from the kinetics experiments (Figures 2 and 3) using the following equations:
This was compared with the actual fraction of Glut4 at the cell surface (PMobs) measured in a third independent set of experiments (Figure 1):
There is excellent agreement between the measured and calculated values for cell surface Glut4 in control and Rab14 knockdown cells (Table 1) as previously described.
α2-Macroglobulin uptake and surface labelling
Experiments were performed as described in . Cells were serum starved for 2 h in serum-free DMEM and then incubated with or without 100 nM insulin for 30 min prior to incubation with AF647–α2M (with or without insulin) for increasing times, then analysed by flow cytometry. The rate constants of α2M–LRP1 exocytosis were calculated from uptake and cell-surface binding data using eqns (1)–(4) as described in .
Transferrin efflux and Tf receptor surface labelling
Experiments were performed as described in . Cells were pre-incubated for 2 h in LSM (0.5% FBS in DMEM), then incubated with AF647-conjugated iron-loaded (holo) Tf for 30 min, then with excess unlabelled holo Tf for increasing times. Cell-surface Tf receptor was measured in cells treated with or without insulin by labelling with biotinylated anti-Tf receptor antibody (anti-mouse CD71-Biotin, Leinco Technologies) and AF647–streptavidin (Invitrogen/Molecular Probes) at 4°C.
Flow cytometry, gating and analysis were performed as described in . Briefly, labelled cells were placed on ice, washed, then resuspended using collagenase (Type III; 1 mg/ml in 2% BSA/PBS; Worthington Biochemicals) in PBS (adipocytes) or PBS/0.5 mM EDTA (fibroblasts). Cells were gently filtered and analysed by flow cytometry (BD Accuri C6). Detection thresholds were set on FSC-H (forward scatter–height) and SSC-H (side scatter–height) for adipocytes or on FSC-H for fibroblasts. Logarithmic intensities of scattered light and three fluorescence channels (FL1, 488 nm excitation/533 nm emission; FL3, 488 nm excitation/>670 nm emission; FL4, 640 nm excitation/675 nm emission) were collected for each cell. Selective gating using light scatter and FL3 (autofluorescence) was used to distinguish adipocytes from fibroblasts and necrotic cells present in the co-cultures after differentiation . ‘Adipocytes’ contain lipid droplets, express endogenous Glut4 and are highly sensitive and responsive to insulin . Selective gating using light scatter and FL1 (GFP) was used to distinguish cells infected with HA–Glut4/GFP from uninfected cells . The geometric mean fluorescence of the gated populations were determined using FCS Express (De Novo Software). The mean fluorescence values of uninfected cells were subtracted from the mean values of FL1 and FL4 (AF647) measured for the infected cell population within the same sample to correct for autofluorescence (FL1) and non-specific labelling (FL4).
Data analysis, modelling and simulations
Data analysis was done essentially as described in [8,9]. Statistical significance was assessed by Student's t test or two-way ANOVA (Figures 1A and 1B, 4B–4D, 5A, 5B, 5D and 5E, 6B, 6D and 6E) or by comparing fits of the data sets using either single exponents (Figures 1C, 2, 3, 4E and 4F, 6A and 4C) or lines (Figures 5C and 5F) and calculating the P value for the null hypothesis that both sets of data were best fit by the same function (***P≤0.0001, **P≤0.001, *P≤0.015).
To fit the data with more complicated models, simultaneous fits of multiple data sets were done [8,9]. To estimate the values of the parameters, sets of simultaneous least-squares optimizations of each model were performed. The data sets considered simultaneously were the fibroblast data (Figures 4A and 4E) or the adipocyte data (Figures 1C, 2 and 3) for both control cells and Rab14-knockdown cells. The data from Rab10-knockdown cells were also included . This allowed a comparison of multiple independent optimized models for each Rab14-knockdown hypothesis (16 for each hypothesis tested in fibroblasts, 32 for each hypothesis tested in adipocytes; Supplementary Tables S1 and S2). Inclusion of the Rab10-knockdown data stabilized and improved the Rab14 fits.
The data were consistent across three separate retroviral infections, with multiple independent experiments done on different days using each batch of cells (biological replicates), and multiple replicate samples in each experiment (technical replicates). n=biological replicates unless otherwise specified.
Rab14 knockdown decreased cell-surface Glut4 in insulin-stimulated adipocytes but increased cell-surface Glut4 in basal cells
Knockdown of Rab14 in differentiated adipocytes decreased cell-surface Glut4 by 61% relative to control adipocytes after insulin stimulation as measured by AF647–α-HA binding (Figure 1A) . Cell-surface Glut4 levels are proportional to the total amount of Glut4 expressed in cells. Thus, it was possible that Rab14 affected cell-surface Glut4 through effects on Glut4 expression. Knockdown of Rab14 did decrease total cellular HA–Glut4/GFP by 26% relative to control cells (Figure 1B). To correct for this difference, mean fluorescence ratios (MFR; mean fluorescence AF647–α-HA/mean fluorescence GFP) were calculated for each sample. Using the corrected values, the proportion of total Glut4 at the cell surface in insulin-stimulated cells was decreased by 47% in Rab14-knockdown cells relative to control adipocytes and was increased by 23% in basal Rab14-knockdown adipocytes relative to control cells (Figure 1C; Table 1).
A decrease in cell-surface Glut4 might be detected if the levels were measured before the new insulin-stimulated steady state was reached. However, cell-surface Glut4 rapidly rose to new steady-state levels after insulin addition in both control and Rab14-knockdown cells, and remained elevated for at least 90 min in both (Figure 1C). In fact, the rate of transition from the basal to insulin-stimulated state was 25% faster in Rab14-knockdown cells than in control cells (kobs = 0.15 min−1, t1/2= 4.5 min control; kobs=0.20 min−1, t1/2= 3.5 min Rab14 knockdown). Thus, Rab14 knockdown did not inhibit the rate of response to insulin. Therefore, Rab14 knockdown must be affecting Glut4 trafficking (exocytosis and/or endocytosis).
Rab14 knockdown inhibited Glut4 exocytosis in insulin-stimulated adipocytes but accelerated exocytosis in basal cells
A decrease in the proportion of total Glut4 at the PM could be due to a change in the fraction of Glut4 trapped in non-cycling compartments (e.g. degradative or biosynthetic compartments) or to a change in the kinetics of cycling of Glut4. To measure the amount of cycling Glut4 and the intrinsic rate constant of exocytosis (kex), α-HA uptake assays were performed (Figure 2). In these experiments, AF647-labelled α-HA is added to the cell culture medium, and the cells were incubated at 37°C for increasing times. As HA–Glut4/GFP is inserted into the PM, it is labelled. It then continues to cycle, carrying the antibody with it. Single exponential fits of the labelling (antibody uptake) yield approximations of the overall kex (kex=kobs, the observed relaxation rate constant) and the size of the cycling pool (Ymax). In Rab14-knockdown cells, the cycling pool size in insulin-stimulated cells was not significantly different from control cells (Table 1, Ymax). Therefore, Rab14 knockdown must be affecting Glut4 trafficking kinetics in these cells. In fact, knockdown of Rab14 decreased kex by 32% relative to the control cells after insulin stimulation (kex= 0.034 min−1 control, 0.023 min−1 Rab14 knockdown; Figures 2A and 2B and Table 1).
In contrast, knockdown of Rab14 accelerated Glut4 exocytosis in basal cells (Figures 2C and 2D; P<0.0004 that Rab14 knockdown and control data are best fit by the same exponential function). In basal adipocytes, there are two pools of cycling Glut4, an actively cycling ERC pool, and a sequestered very slowly cycling/non-cycling GSV pool . Therefore, an increase in Glut4 exocytosis could be due either to a change in the kinetics of trafficking through one of these pathways or to a change in the relative distribution of Glut4 between these two pathways. This can be modelled as either a cycling ERC pool with a non-cycling GSV (fitted with a single exponent; Table 1) or as two cycling pools with two rate constants (results not shown; ). Using either model showed that Rab14 knockdown increased basal α-HA uptake relative to control cells by increasing Glut4 in the actively cycling ERC pool (Ymax was increased by 34%; Table 1) and inhibiting sequestration in GSVs, with little effect on the rate of constant of exocytosis from the ERCs (Figures 2B and 2D).
Rab14 knockdown accelerated Glut4 endocytosis in adipocytes
A change in the proportion of Glut4 at the cell surface might also be caused by a change in the intrinsic rate constant of endocytosis (ken). To measure ken, transition kinetics assays were performed using the PI3K inhibitor LY294002 (LYi; Figure 3; ). LYi rapidly inhibits exocytosis of Glut4 (kex LYi=0.004 min−1), with no effect on the rate constant of endocytosis. Therefore, the observed relaxation rate constant for the transition to the new steady state after addition of inhibitor (kobs) is a good approximation of ken (kobs =ken+kex; ken ≈ kobs–0.004 min−1). We see no effect of insulin on the rate constant of endocytosis of Glut4 in our cells, therefore the rate constant of endocytosis was measured only in insulin-stimulated adipocytes [6,8]. Rab14 knockdown increased ken 40% relative to the control cells (ken=0.13 min−1 control, 0.18 min−1 Rab14 knockdown; Figure 3 and Table 1). This increase in ken is not an artefact of shRNA expression or viral infection, as it was not observed in cells expressing shRNA specific for a number of other proteins (AS160, insulin-responsive aminopeptidase, sortilin or CDP138;  and results not shown). This difference in ken accounts for the increase in the observed relaxation rate constant for transition from the basal to insulin-stimulated state in the Rab14-knockdown cells (Figure 1C; kobs=ken + kex; 0.13+0.03=0.16 min−1, t1/2= 4.2 min control; 0.18+0.02=0.20 min−1, t1/2 = 3.4 min Rab14 knockdown). The increase in ken together with the decrease in kex can fully account for the effects of knockdown of Rab14 on cell-surface Glut4 in adipocytes (Table 1). The excellent agreement between the observed and calculated values for cell-surface Glut4 provides internal validation of the methods used to estimate ken and kex. We believe that the increase in ken in Rab14-knockdown cells is due to redistribution of Glut4 between a ‘slow’ endocytic pathway, which is the dominant pathway in adipocytes (ken=0.053 min−1) and a ‘fast’ endocytic pathway that Glut4 shares with the Tf receptor (ken=0.6 min−1) (see Figure 7). Differentiation of fibroblasts into adipocytes also causes a shift between these pathways .
Rab14 knockdown inhibited Glut4 exocytosis in fibroblasts with no effect on endocytosis
It has been suggested that Rab14 limits exocytosis of Glut4 from constitutive recycling pathways, whereas Rab10 limits exocytosis of the sequestered GSVs . The ERC pathway is found in both fibroblasts and adipocytes, whereas the GSV pathway is adipocyte-specific. The ERC pathway in fibroblasts is kinetically identical with the slow constitutive pathway in adipocytes , and the morphology of the compartments are very similar [2,29], indicating that this pathway is shared in the two cell types. Therefore, to assess effects on the constitutive pathway, the effects of Rab14 knockdown on the trafficking kinetics of Glut4 were measured in fibroblasts (Figure 4). Knockdown of Rab14 decreased cell-surface Glut4 in basal cells and at all time points after insulin stimulation relative to control cells by 25–30% (Figure 4A). As observed in adipocytes, Rab14 knockdown decreased total Glut4 levels 13% relative to control fibroblasts (Figure 4B).
Rab14 knockdown did not change the fraction of Glut4 actively cycling in fibroblast cells (Ymax was unaffected; results not shown). The decrease in cell-surface Glut4 in fibroblasts was due to a 30–40% decrease in Glut4 exocytosis under both basal and insulin-stimulated conditions (kex calc= 0.02 and 0.039 min−1 control, 0.014 and 0.022 min−1 Rab14 knockdown; Figure 4D). In contrast with adipocytes, Rab14 knockdown did not significantly affect Glut4 endocytosis in fibroblasts (ken=0.22 min−1 control, 0.18 min−1 Rab14 knockdown; Figures 4E and 4F). Glut4 endocytosis is significantly faster in fibroblasts than in differentiated adipocytes (ken=0.22 min−1 in fibroblasts compared with 0.13 min−1 in adipocytes) . Endocytosis of Glut4 is even slower in primary adipocytes than in cultured cells (ken=0.053 min−1; ). We hypothesize that differentiation decreases the rate constant of endocytosis of Glut4 due to a redistribution of Glut4 from a fast pathway that Glut4 shares with the Tf receptor, to a slow ‘adipocyte-specific’ pathway, and not through changes in the intrinsic rate constants of internalization through these pathways [8,9]. Rab14 knockdown may inhibit this redistribution, perhaps by changing the recycling pathway followed by the Glut4 (see Figure 7). Glut4 does not traffic through the Rab10-dependent GSV pathway in fibroblasts . Therefore, the inhibition of Glut4 exocytosis caused by Rab14 knockdown in fibroblasts shows that Glut4 is sorted from a Rab14-dependent Rab10-independent constitutive recycling pathway into a highly regulated Rab10-dependent pathway after differentiation.
Rab14 knockdown accelerates LRP1 exocytosis in adipocytes but not in fibroblasts
LRP1 cycles with Glut4 through both the constitutive ERCs and regulated GSV pathways [6,8]. Thus, we expected that Rab14 knockdown would affect LRP1 trafficking in the same way that it affected Glut4 trafficking: increasing LRP1 exocytosis in basal adipocytes, but inhibiting it in insulin stimulated cells. To study the trafficking of LRP1, we examined the binding and uptake of its ligand α2M (Figure 5). α2M uptake is a function of both the rate of internalization of the α2M–LRP1 complex and the amount of unoccupied LRP1 at the cell surface (slope=kenLRP1surface). Unexpectedly, Rab14 knockdown increased cell-surface LRP1 by 50% and accelerated α2M uptake 2-fold in both basal and insulin-stimulated adipocytes (Figures 5A and 5C). The observed increase in uptake was due in part to a 40% increase in ken for α2M–LRP1 (Table 2), as observed for Glut4. Like Glut4, LRP1 is internalized through both the ‘slow’ Glut4 endocytosis pathway and the ‘fast’ Tf receptor pathway in adipocytes and fibroblasts [6,8], and this increase in ken is probably due to redistribution between these pathways. However, an increase in the rate of internalization with no change in exocytosis would decrease cell-surface LRP1, decreasing uptake. The increase in cell-surface LRP1 and α2M uptake in Rab14-knockdown cells was due to a 2-fold increase in the rate constant of exocytosis of LRP1 under both basal and insulin-stimulated conditions (kex =0.021 min−1 basal, 0.044 min−1 insulin control; 0.041 min−1 basal, 0.084 min−1 Rab14 knockdown; Figure 5B, Table 2). Interestingly, AS160 knockdown also had differential effects on Glut4 and LRP1 trafficking . AS160 knockdown increased cell-surface LRP1, accelerated AF647–α2M uptake and increased LRP1 kex 2-fold, whereas it decreased cell-surface Glut4 by 20% and decreased Glut4 kex by 27% in insulin-stimulated adipocytes. As observed for Rab14, AS160 knockdown accelerated exocytosis of both Glut4 and LRP1 in basal adipocytes.
If Rab14 knockdown increases α2M–LRP1 trafficking in adipocytes due to effects on the constitutive ERC pathway, then Rab14 knockdown in fibroblasts should also increase exocytosis of LRP1. However, Rab14 knockdown decreased cell-surface LRP1 in fibroblasts by 40% (Figure 5D). The decrease in cell-surface LRP1 was due to both a decrease in kex and an increase in ken (Figure 5E and Table 2). The opposing effects on kex and ken resulted in no net change in α2M uptake in fibroblasts (Figure 5F). Thus, Rab14 knockdown increased LRP1 exocytosis in adipocytes but not fibroblasts. This indicates that Rab14 knockdown accelerates LRP1 exocytosis via effects on the specialized regulated GSV pathway found in adipocytes, not the constitutive ERC cycling pathway.
Rab14 knockdown does not affect Tf receptor exocytosis in adipocytes or fibroblasts
Most (90–95%) of the Tf that is internalized and delivered to endosomes is recycled with its receptor through a very fast exocytic pathway . LRP1 can also recycle through this fast pathway, whereas Glut4 is largely excluded. To verify that there was no effect of Rab14 knockdown on the kinetics of the fast Tf receptor recycling pathway, we measured the rates of efflux of labelled Tf and cell-surface levels of the Tf receptor in adipocytes and fibroblasts (Figure 6). In these experiments, cells were preloaded with AF647-conjugated iron-loaded Tf, then incubated with excess unlabelled Tf at 37°C. When the iron-depleted Tf is recycled back to the cell surface, it dissociates from its receptor. Therefore, the rate of loss of the labelled Tf is a measure of the rate of receptor recycling (kobs =krec). There was no significant difference in the rates of exocytosis of Tf or in the cell-surface levels of the Tf receptor between control and Rab14-knockdown cells in either adipocytes or fibroblasts. The rate constants of recycling of the Tf receptor (krec=0.1–0.15 min−1; Figure 6E) were much higher than the rate constants of exocytosis of Glut4 and LRP1 (kex = 0.005–0.04 min−1; Tables 1 and 2) in both fibroblasts and adipocytes, as previously reported [8,9].
Modelling and simulations: Rab14 limits sorting into the specialized regulated secretory compartments and into the constitutive ERC pathway
We have developed a model that describes the unique trafficking of Glut4 relative to the Tf receptor in adipocytes (Figure 7; ). Steps 1–3 represent the slow constitutive recycling pathway shared by fibroblasts and adipocytes (black dotted lines). Steps 4 and 5 represent the highly regulated specialized GSV pathway found only in adipocytes (black bold lines). Most (90–95%) of the Tf that is internalized and delivered to endosomes is recycled with its receptor through an additional very fast exocytic pathway (grey lines). The rate constants of both endocytosis and exocytosis are much higher through this pathway than through the Glut4 pathways (ken=0.5 min−1; kex=0.1–0.2 min−1). LRP1 can recycle through the fast Tf receptor pathway in both adipocytes and fibroblasts, whereas Glut4 is largely excluded from this pathway.
This model can be described mathematically as a series of ordinary differential equations that describe the transfer of Glut4 between four compartments: PM, SE, ERCs and GSVs, with a single rate constant for each of five steps (ken, ksort, kfuseE, kseq and kfuseG). The SE is defined functionally as the compartment where Glut4, LRP1 and the Tf receptor are sorted into three distinct recycling pathways. The ERC is defined functionally as a rate-limiting kinetic intermediate in the slow constitutive recycling pathway. There is an additional rate constant for the recycling of LRP1 from SE to the PM via the Tf receptor pathway (krec). All of the trafficking assays for both Glut4 and α2M–LRP1 in both fibroblasts and adipocytes can be accurately simulated using these differential equations [8,9].
To test the hypothesis that Rab14 knockdown affects one or more of these rate constants, different models were created in which the rate constants were either unique for each cell type or shared between the control and Rab14-knockdown cells. Hypotheses testing for all possible combinations of rate constants were examined (Supplementary Table S3). The values for the rate constants in each model were optimized by simultaneously fitting the data from both control and Rab14-knockdown cells for fibroblasts (Figure 4; four data sets) or adipocytes (Figures 1–3; six data sets).
To examine the effects of Rab14 knockdown on the constitutive endocytic recycling pathway, models of the fibroblast data testing three hypotheses were compared: Rab14 knockdown affects (1) ksort, (2) kfuseE or (3) ksort and kfuseE (Figure 4 and Supplementary Table S3). All three hypotheses were able to accurately represent the effects of Rab14 knockdown on the basal to insulin and insulin+LYi transition experiments (Figure 4A, grey lines and results not shown). In models testing hypotheses 1 or 2, ksort and kfuseE were significantly different in the Rab14-knockdown and control cells (Supplementary Table S3). Thus, effects on either vesicle fusion (kfuseE) or endosomal sorting (ksort) can account for the inhibition of Glut4 trafficking by Rab14 knockdown in fibroblasts. Allowing both to vary did not improve the optimized models.
To examine the effects of Rab14 knockdown in adipocytes, models of the adipocyte data testing eight hypotheses were compared (Figure 8 and Supplementary Table S3). To test the hypothesis that Rab14 knockdown affects only Glut4 trafficking through the constitutive recycling pathway, we examined optimized models where (1) ken and ksort, (2) ken and kfuseE, or (3) ken, ksort and kfuseE were allowed to vary between control and Rab14-knockdown adipocytes. In the models testing hypotheses 1 or 2, ksort and kfuseE were significantly different in the Rab14-knockdown and control adipocytes. Allowing both ksort and kfuseE to vary did not improve the adipocyte models.
All of the data for Glut4 could be well simulated as effects on ken, kfuseE and/or ksort (Figure 8A, purple lines, and results not shown). Thus, these simulations demonstrate that effects on either sorting into or exocytosis from the constitutive ERC recycling pathway would be sufficient to account for the decrease in cell-surface Glut4 and inhibition of Glut4 trafficking caused by Rab14 knockdown in both fibroblasts and adipocytes. However, LRP1 exocytosis and α2M uptake were unaffected, not accelerated, in these models (Figure 8B, purple lines). Furthermore, cell-surface LRP1 was decreased, not increased, in these models (results not shown). Thus, the hypothesis that Rab14 knockdown affects trafficking only through the constitutive ERC pathway is insufficient .
Alterations to ken and kfuseG (hypothesis 4) were also unable to accurately simulate the differential effects of Rab14 knockdown on Glut4 and α2M–LRP1 trafficking in adipocytes. Although effects on ken and kfuseG can accurately simulate the effects of Rab14 knockdown on Glut4 exocytosis (Figure 8A, red lines, and results not shown), a decrease in kfuseG inhibits, not increases, LRP1 exocytosis, decreasing α2M uptake and cell-surface LRP1 in insulin-stimulated cells (Figure 8B, red lines, and results not shown). Thus, the hypothesis that Rab14 affects GSV exocytosis is also not correct.
The effects of Rab14 knockdown on Glut4 as well as LRP1 in adipocytes can be accurately recapitulated in models where Rab14 knockdown affects sorting from SEs into the GSVs (hypothesis 5) Rab14 knockdown affects–ken and kseq (Figure 8, blue unbroken lines). The decrease in kseq inhibits sequestration in basal adipocytes, increasing the amount of Glut4 and LRP1 recycling through the ERC pathway. After insulin stimulation, sorting into the ERC and GSV pathways becomes rate-limiting in the Rab14-knockdown cells relative to the control cells, causing these proteins to accumulate in the SE. LRP1 can recycle with the Tf receptor from this compartment via the fast pathway, whereas Glut4 does not. Thus, inhibition of kseq inhibits Glut4 exocytosis, but accelerates LRP1 exocytosis, after insulin stimulation. Therefore, decreasing kseq and increasing ken is sufficient to simulate all of the effects of Rab14 knockdown on trafficking in adipocytes. However, they are insufficient to account for the observed phenotypes in fibroblasts, since the sequestration pathway is not active in fibroblasts. Therefore, the hypothesis that Rab14 knockdown affects only trafficking into the GSV sequestration pathway is also insufficient . Rab14 knockdown must affect both sorting into the GSVs and cycling through the constitutive ERC pathway.
The model testing hypothesis 6, i.e. Rab14 knockdown affects ken, kseq and ksort, accurately simulates all of the data from Rab14-knockdown adipocytes and fibroblasts (Figure 8, blue dashed lines, Figure 4, grey dashed line, and results not shown). In this model, Rab14 knockdown inhibits ksort to a similar degree in both adipocytes and fibroblasts (30% in adipocytes, 40% in fibroblasts; Supplementary Table S3). Allowing kfuseE to vary–hypothesis 7, i.e. Rab14 knockdown affects ken, kseq and kfuseE, or 8, i.e. Rab14 knockdown affects ken, kseq, ksort and kfuseE did not improve the simulations (Figure 8; blue dotted lines, and results not shown). These results strongly support the hypothesis that Rab14 limits transfer of Glut4 and LRP1 from the SE into both the regulated GSV pathway and the constitutive ERC pathway , but does not affect recycling of the Tf receptor from this compartment (Figure 6).
Differentiation of fibroblasts into adipocytes leads to the expression of insulin-sensitive protein machineries that inhibit exocytosis of Glut4 20–30-fold under basal conditions. The RabGAP AS160 is one of the inhibitory proteins induced during adipocyte differentiation that limits Glut4 exocytosis [11,31]. Under basal conditions, AS160 is present on GSVs where it inhibits the GTP loading of a Rab or Rab(s) required to make the vesicles competent to fuse to the PM [7,8,10]. After insulin stimulation, AS160 is phosphorylated, inactivated, binds to 14-3-3 proteins and dissociates from the vesicles allowing activation of the Rabs [18,32,33]. To identify which Rab is the AS160 substrate that limits GSV exocytosis, we examined the effects of knockdown of two Rabs that co-localize with Glut4–Rabs 10 and 14–on the trafficking of Glut4 and LRP1 in fibroblasts and adipocytes. These Rabs are AS160 substrates in vitro and are found on immuno-isolated Glut4-containing vesicles in adipocytes [18,19]. Previously, we showed that knockdown of Rab10 has the precise phenotype expected for the Rab that regulates GSV exocytosis . Rab10 knockdown inhibited the exocytosis of both Glut4 and LRP1 in adipocytes under both basal and insulin-stimulated conditions. Rab10 knockdown had no effect on trafficking of these proteins in fibroblasts, where there is no GSV pathway.
In contrast, the phenotype of Rab14 knockdown is very different. Rab14 knockdown inhibited Glut4 exocytosis in adipocytes after insulin stimulation, decreasing cell-surface Glut4. However, Rab14 knockdown accelerated exocytosis of LRP1 after insulin stimulation, and accelerated exocytosis of both Glut4 and LRP1 in basal adipocytes. Rab14 knockdown inhibited exocytosis of both Glut4 and LRP1 in fibroblasts, where there are no GSVs. These data show that Rab14 affects Glut4 trafficking through a very different mechanism than Rab10.
TIRF microscopy showed that Glut4 and insulin-regulated aminopeptidase (IRAP) traffic to the PM in vesicles decorated with either Rab14 or Rab10 . In contrast, very few of the fusing vesicles contained other Rabs that co-localize with Glut4, including Rabs 2, 4a, 4b and 11. Most of the fusing Rab14-containing vesicles also contained the Tf receptor, whereas those decorated with Rab10 did not. Furthermore, Rab14 decorated vesicles were observed in both basal and insulin-stimulated cells, whereas Rab10 was found on fusing vesicles only after insulin stimulation. These observations lead to the hypothesis that Rab14 mediates Glut4 delivery to the PM via vesicles from the ERCs, whereas Rab10 mediates GSV fusion. Consistent with this hypothesis, it is possible to simulate the effects of Rab14 knockdown on Glut4 trafficking in both fibroblasts and adipocytes through effects on the rate of fusion of ERC-derived recycling compartments (kfuseE; Figure 4A, grey dotted line, and Figure 8A, purple dotted line). However, an inhibition of endosomal fusion would inhibit, not accelerate basal Glut4 exocytosis. Furthermore, it would have little effect on α2M uptake and would decrease cell-surface LRP1 levels in adipocytes (Figure 8B, purple dotted line, and results not shown). This is not what was observed. Thus, effects on exocytosis through the constitutive pathway are insufficient by themselves to account for the phenotypes observed in Rab14-knockdown adipocytes.
An alternative hypothesis is that Rab14 limits sorting from endosomes into the regulated GSV pathway. This hypothesis is strongly supported by the observation that Rab14 knockdown and mutation of the TELEY sorting motif in Glut4 appear to affect the same step in Glut4 trafficking . The TELEY motif directs Glut4 into the GSV sequestration pathway. Consistent with this hypothesis, inhibition of endosomal sorting and packaging into GSVs (kseq) can account for the differential effects of Rab14 knockdown on Glut4 and LRP1 exocytosis (Figure 8; blue unbroken lines). In fact, these differential effects on Glut4 and LRP1 trafficking are a defining feature expected for perturbations of sorting into the sequestration pathway, and are not observed in simulations of perturbations in any other step. Thus, our data support the hypothesis that Rab14 and Rab10 limit trafficking through sequential steps in the Glut4 trafficking itinerary . Consistent with published data , our simulations predict only a small (12%) additive effect on the decrease in cell-surface Glut4 in insulin-stimulated cells if both Rab14 and Rab10 are knocked down compared with knockdown of Rab10 alone (results not shown).
Although an effect on kseq alone is sufficient to account for all of the phenotypes observed in adipocytes, this hypothesis is insufficient to account for the observed effects of Rab14 knockdown on Glut4 and LRP1 in fibroblasts. The model that best fits our data is that Rab14 limits sorting from SE into both the constitutive ERC and the highly regulated GSV recycling pathways followed by Glut4 and LRP1, but has no effect on protein sorting into the very fast Tf recycling pathway (Figures 6 and 7). Consistent with this hypothesis, differentiation (and expression of AS160) decreases the basal rate of sorting into the ERC (ksort) 4-fold (Supplementary Table S3 and ).
The hypothesis that Rab14 functions at endosomal sorting is consistent with previous studies of the effects of Rab14 knockdown on Glut4 trafficking in adipocytes [23,29]. Interestingly, expression of either a constitutively active GTP hydrolysis-resistant mutant or overexpression of wild-type Rab14 inhibited Glut4 translocation and resulted in the accumulation of enlarged SE structures that were rapidly filled by Glut4 and Tf internalized from the PM . The rate of transfer of Glut4 from these SE to later compartments (both the ERC and GSVs) was significantly inhibited in cells with elevated Rab14 activity (as measured by quantitative electron microscopy). Rab14 knockdown also caused Glut4 to accumulate in SE compartments and inhibited the rate of transfer of Glut4 to later compartments . However, this redistribution was not detected by fluorescence microscopy (there was no swelling of the endosomes in Rab14-knockdown cells) [23,29]. This model can explain the observation that knockdown of both Rab10 and Rab14 has an additive effect on decreasing the total Glut4 accumulated near the PM (as determined by TIRF microscopy; ), but not in the amount of Glut4 inserted into the PM . Our simulations predict that there would be approximately the same amount of Glut4 in the PM in the double-knockdown cells as in Rab10-knockdown cells (0.056 compared with 0.061), as is observed. However, there would be four times as much Glut4 accumulated in the SE in the double-knockdown cells than in the Rab10-knockdown cells, with proportionately less Glut4 in small transport vesicles. The SE lie outside the TIRF zone. Thus, the double knockdown might have less Glut4 in small vesicles near the cell surface (within the TIRF field), with no change in the amount inserted into the PM.
Both Rab14 and Rab10 are substrates for AS160 in vitro. Simultaneous regulation of both Rab14 and Rab10 via AS160 would be an ideal mechanism to ensure co-ordinate regulation of both sorting into and exocytosis from GSVs (Figure 7). Kinetics and subcellular distribution data strongly support the idea that insulin regulates both of these steps [2,8]. Although most of the Glut4 in basal adipocytes accumulates in the very slowly cycling GSVs, a small fraction (10–15%) of the Glut4 cycles through the constitutive ‘fibroblast’ pathway through the ERC [5,6,8,34,35]. The distribution of Glut4 in both the GSV and endosomal pathways in basal cells was originally observed by quantitative electron microscopy . In order to maintain the endosomal pool, the rate constant for transfer of Glut4 from the SE into the GSVs (kseq) must be very low in basal cells . If it were not, all of the Glut4 would accumulate in the GSVs due to the very slow rate-limiting step in this exocytic pathway (kfuseG basal < 0.0007 min−1). Consistent with this, kseq is very slow in the optimized fits of the control adipocyte data (kseq basal = 0.003–0.004 min−1, Supplementary Table S3). However, insulin must increase kseq to allow Glut4 to repopulate the GSVs after stimulation of the initial burst of exocytosis–two-thirds of the intracellular Glut4 is found in the GSVs in insulin-stimulated cells at steady state [2,8]. An increase in kseq after insulin stimulation is also required to account for the effects of Rab10 knockdown on the steady-state distribution of Glut4 . Consistent with these observations, kseq increases significantly in the optimized models after insulin stimulation (kseq insulin = 0.3–0.4 min−1; Supplementary Table S3).
Our data are consistent with a previous study that concluded that Rab14 functions in the sorting of Glut4 and LRP1 into the GSV pathway . However, despite strong supporting evidence, regulation of this sorting/sequestration step remains controversial , and is a major difference between our models. Previously it was concluded that although both Rab10 and Rab14 are in vitro substrates of AS160, Rab10 is a substrate in cells, but Rab14 is not. This was based on the observation that Rab10 knockdown reversed the increase in basal cell-surface Glut4 caused by AS160 knockdown in adipocytes, whereas Rab14 knockdown did not. However, Rab14 knockdown accelerates basal Glut4 trafficking in adipocytes through inhibition of sequestration (Figure 2 and Table 1). Thus, our studies predict that Rab14 knockdown would enhance (10%), not reverse, the increase in cell-surface Glut4 in basal adipocytes caused by AS160 knockdown. Both AS160 and Rab14 knockdown inhibit sequestration after insulin stimulation, and they are not expected to be additive under these conditions, as was observed. Thus, the published data are consistent with AS160 regulating Rab14 to limit transport through the sorting/sequestration step.
The effects of AS160 knockdown on the trafficking of Glut4 and α2M–LRP1 in adipocytes are complex [7,8]. AS160 knockdown increases the rate of exocytosis of both Glut4 and LRP1 in basal cells, increasing cell-surface levels 4-fold and 1.7-fold respectively. Insulin increases cell-surface Glut4 and LRP1 an additional 5-fold and 1.5-fold in knockdown cells. However, AS160 knockdown decreased cell-surface Glut4 by 20% and inhibited the maximal rate of exocytosis of Glut4 by 27% after insulin stimulation relative to control cells. Interestingly, AS160 knockdown accelerated exocytosis of LRP1 2-fold relative to control cells. The complex phenotypes observed in AS160-knockdown cells can be fully explained through effects on Rabs 10 and 14 (Figure 7). Loss of AS160 would lead to constitutive activation of Rab10 on the GSVs, increasing the rate constant of exocytosis for Glut4 and LRP1. Glut4 accumulates behind a second AS160-independent Akt- and PI3K-dependent step that regulates GSV fusion , although the protein(s) that regulate this fusion step are currently unknown. Stimulation of the fusion step by insulin leads to further translocation of Glut4 and LRP1 in the AS160-knockdown cells. Loss of AS160 would also lead to the constitutive activation of Rab14. Over-activation of Rab14 inhibits kseq, leading to redistribution of LRP1 and Glut4 from the very slowly cycling GSVs into the actively cycling endosomal recycling pathway. Thus, inhibition of kseq would increase exocytosis of both Glut4 and LRP1 in basal cells. It would also cause inhibition of Glut4 exocytosis and acceleration of LRP1 exocytosis after insulin stimulation, when trafficking out of the endosome becomes rate-limiting for Glut4. LRP1 recycles with the Tf receptor from these compartments. Interestingly, co-ordinate activation of kseq and kfuseG through incremental inhibition of AS160 and activation of Rabs 14 and 10 can simulate the stepwise ‘quantal’ release of Glut4 observed with increasing concentrations of insulin ([5,34,35]; results not shown).
A unifying hypothesis for all of the observed phenotypes is that Rab14 is required for maturation of endosomes from an early form that allows fusion of newly endocytosed vesicles to a later form that no longer supports delivery of newly internalized content. The Tf receptor and LRP1 are rapidly recycled from the early fusion-competent vesicles. After maturation, Glut4 and LRP1 (and approximately 5% of the Tf receptor) are sorted into the constitutive ERC and the regulated GSV pathways. The content that remains after this sorting is transferred to the lysosomes. The known effectors of Rab14 strongly support a role in endosomal maturation. Rab14 shares interactions with the Rab4-binding protein RUFY1/RabIP4  and with the Rab11-binding protein Rab11FIP1c [37,38]; these interactions occur through different contact sites and can occur simultaneously. When co-localization of a pulse of Tf with endosomal Rabs is monitored, Tf is found first in compartments marked by the SE marker Rab5, then Rab4, then Rab14 and finally Rab11 . These shared interactions and the kinetics of co-localization suggest that Rab14 may function as an intermediate in the Rab4 and Rab11 sorting and recycling pathways, and that it may participate in a Rab cascade with Rab4 and Rab11. Both Rab4 and Rab11 are utilized in Glut4 trafficking [40–42]. Interestingly, overexpression of both Rab4 and RUFY1 results in enlarged endosomes similar to the phenotype produced by Rab14 overexpression in adipocytes . Based on these biochemical interactions, it is likely that Rab14 functions in sorting from endosomes into the Rab11-dependent ERC pathway. However, it is unclear whether the Rab14 found on fusing Glut4- and Tf-positive vesicles is also required for tethering and fusion , or whether it is simply being carried to the cell surface on the membranes after endosomal sorting. Consistent with the latter, RUFY1 has its functional role on endosomes, but it increases in the PM in response to insulin . The functional significance of interactions between Rab14, RUFY1 and Rab11FIP1c remain unclear. In contrast with what is known in endosomal trafficking, the Rab14 effectors required for sorting Glut4, LRP1 and cargo proteins such as IRAP from endosomes into the Rab10-dependent GSV pathway, as well as the interaction of Rab14 with protein machineries such as GGA, sortilin and TUG that drive and regulate this sorting remain to be elucidated [45,46]. However, we show that Rab14 knockdown exhibits the defining phenotype expected for perturbation of a protein that limits sorting into the highly regulated specialized GSV sequestration pathway. Simultaneous regulation of both sorting into GSVs (via Rab14) and exocytosis out of GSVs (via Rab10) by AS160 would provide a mechanism for the dose-dependent ‘quantal’ release of Glut4 observed in response to insulin in adipocytes.
Cynthia Mastick conceived and co-ordinated the study, analysed the data and prepared the manuscript. Cynthia Mastick and Adelle Coster designed, performed and analysed the mathematical modelling and simulations. Paul Brewer, Estifanos Habtemichael and Irina Romenskaia designed, performed and analysed the experiments and preparation of all figures. All authors contributed to the writing of the manuscript, preparation of figures, reviewed the results and approved the final version of the manuscript.
This work was supported by the American Diabetes Association [grant number 1-12-BS-132]; and the National Institute of General Medical Sciences from the National Institutes of Health, USA [grant number P20GM103440].
We are grateful to Gustav Lienhard (Dartmouth Medical School, Hanover, NH, USA) for the retroviral Rab14 shRNA construct and Robert Hoeben (Leiden University Medical Centre Leiden, Netherlands) for the lentiviral HA–Glut4/GFP construct. This research includes computations using the Linux computational cluster Katana, supported by the Faculty of Science, UNSW Australia.
Abbreviations: AF647, Alexa Fluor 647; AS160, Akt substrate of 160 kDa; DMEM, Dulbecco's modified Eagle's medium; ERC, endosomal recycling intermediate compartment; Glut4, glucose transporter isoform 4; GSV, Glut4 storage vesicle; HA, haemagglutinin; IRAP, insulin-regulated aminopeptidase; LRP1, low-density-lipoprotein-receptor-related protein 1; LSM, low-serum medium; LYi, LY294002; MFR, mean fluorescence ratio; PI3K, phosphoinositide 3-kinase; PM, plasma membrane; SE, sorting or ‘early’ endosome(s); Tf, transferrin; TIRF, total internal reflection fluorescence; α-HA, anti-HA antibody; α2M, α2-macroglobulin
- © 2016 The Author(s). published by Portland Press Limited on behalf of the Biochemical Society