A Rictor-Myo1c Complex Participates in Dynamic Cortical Actin Events in 3T3-L1 Adipocytes

Abstract

Insulin signaling through phosphatidylinositol 3-kinase (PI 3-kinase) activates the protein kinase Akt through phosphorylation of its threonine 308 and serine 473 residues by the PDK1 protein kinase and the Rictor-mammalian target of rapamycin complex (mTORC2), respectively. Remarkably, we show here that the Rictor protein is also present in cultured adipocytes in complexes containing Myo1c, a molecular motor that promotes cortical actin remodeling. Interestingly, the Rictor-Myo1c complex is biochemically distinct from the previously reported mTORC2 and can be immunoprecipitated independently of mTORC2. Furthermore, while RNA interference-directed silencing of Rictor results in the expected attenuation of Akt phosphorylation at serine 473, depletion of Myo1c is without effect. In contrast, loss of either Rictor or Myo1c inhibits phosphorylation of the actin filament regulatory protein paxillin at tyrosine 118. Furthermore, Myo1c-induced membrane ruffling of 3T3-L1 adipocytes is also compromised following Rictor knockdown. Interestingly, neither the mTORC2 inhibitor rapamycin nor the PI 3-kinase inhibitor wortmannin affects paxillin tyrosine 118 phosphorylation. Taken together, our findings suggest that the Rictor-Myo1c complex is distinct from mTORC2 and that Myo1c, in conjunction with Rictor, participates in cortical actin remodeling events.

The mammalian target of rapamycin (mTOR) is an atypical serine/threonine kinase that integrates intracellular and extracellular signals from numerous pathways to regulate key cellular processes, including metabolism and cell growth. The actions of mTOR are mediated through at least two distinct mTOR complexes that mediate signaling pathways of receptors such as the insulin receptor (22, 35). In one of these complexes, mTOR acts with the regulation-associated protein of mTOR (Raptor) to regulate cell growth through phosphorylation of the eukaryotic initiation factor 4E-binding protein and the translational signaling protein p70 S6 kinase. A second mTOR complex was more recently found to mediate insulin receptor signaling to serine/threonine protein kinase B (PKB)/Akt (reviewed in reference 20). Akt activation precedes and mediates many of the metabolic actions of insulin, and its full activation is associated with its insulin-dependent phosphorylation at threonine 308 (Thr308) and serine 473 (Ser473) (reviewed in reference 29). Whereas the protein 3-phosphoinositide-dependent protein kinase 1 (PDK1) had been reported to phosphorylate Akt at Thr308 in response to insulin, the protein kinase that phosphorylates Akt at Ser473 was unknown for a long time. Surprisingly, this protein kinase was identified as mTOR in a complex with rapamycin-insensitive companion of mTOR (Rictor) (11, 24). Thus, mTOR functions within independent complexes to modulate distinct downstream targets.

The existence of two TOR complexes was first reported for Saccharomyces cerevisiae by Loewith et al. (19). According to them, in S. cerevisiae TOR complex 1 (TORC1) consists of TOR1 or TOR2, KOG1, and LST8, whereas TORC2 contains TOR2, AVO1, AVO2, AVO3, and LST8. Other laboratories have also reported the existence of two distinct mTOR complexes. In mammalian cells, mTOR complex 1 (mTORC1) consists of mTOR, GBL (or mLST8), and Raptor and is involved in the regulation of cell size (17). mTORC2, however, consists of mTOR, GBL, SIN1, and Rictor (or mAvo3) and is involved in the full activation of Akt (7, 11, 13, 24, 34). Interestingly, Akt phosphorylates and negatively regulates the Rheb GTPase GAP protein TSC2, causing the activation of the mTOR-Raptor complex and modulation of its downstream bioeffects (reviewed in references 1 and 20).

Interestingly, another function attributed to mTORC2 is the regulation of actin organization (23), particularly under serum-restimulated conditions (12), but the mechanisms behind this have not been defined. Sarbassov et al. showed that both Rictor and mTOR knockdown in HeLa cells resulted in a disrupted actin cytoskeleton arrangement, with actin fibers present in the whole cell and with less prominent cortical actin than that in control cells (23). Consistently, Jacinto et al. also showed that knocking down components of the mTORC2 in NIH 3T3 cells prevented actin polymerization and cell spreading, particularly in serum-restimulated cells (12). In this study, we report results that provide a novel connection between the functions of a component of the mTORC2 and regulation of the actin cytoskeleton. We report here that Rictor coimmunoprecipitates with the actin-based molecular motor myosin 1c (Myo1c) in 3T3-L1 adipocytes. Remarkably, our studies suggest that the Rictor-Myo1c complex is biochemically and functionally distinct from the Rictor-mTOR complex. Furthermore, we report here that the Rictor-Myo1c complex participates in dynamic cortical actin events in cultured adipocytes.

MATERIALS AND METHODS

Materials and chemicals.

Mouse anti-Myc (clone 9E10) monoclonal antibody was purchased from Neomarkers Inc. Rabbit anti-Rictor polyclonal antibody (BL2181) and rabbit anti-Raptor polyclonal antibody (BL888) were purchased from Bethyl Laboratories. Rhodamine-phalloidin was purchased from Molecular Probes. Unless otherwise mentioned, all other antibodies were purchased from Cell Signaling Technologies. Latrunculin B was purchased from BioMol. Wortmannin was purchased from Sigma-Aldrich. Rapamycin was purchased from Calbiochem.

DNA constructs.

The construction of 3×HA-Myo1c (IQ domain plus tail domain) has been described previously (3). The myc-Rictor construct (Addgene plasmid 11367) was obtained from Addgene and has been described previously (23). The 3×HA-Myo1c(full) (motor domain plus IQ domain plus tail domain), 3×HA-Myo1c(motor+IQ) (motor domain plus IQ domain), and 3×HA-Myo1c(tail) (tail domain) plasmids were constructed by subcloning a Myo1c coding sequence encompassing residues 1 to 1028 (full), 1 to 759 (motor+IQ), or 760 to 1028 (tail), amplified by PCR, into the HindIII and BamHI restriction sites of pCMV5-triple HA (30) in frame with three N-terminal hemagglutinin (HA) sequences. The green fluorescent protein (GFP)-Myo1c plasmid was constructed by subcloning a full-length Myo1c coding sequence into the HindIII and BamHI restriction sites of pEGFP-C1 plasmid (Clontech). The constructs were sequenced, and expression was verified in HEK 293T cells prior to experiments.

Cell culture, cell treatments, and transfection of 3T3-L1 adipocytes and HEK 293T cells.

3T3-L1 murine fibroblasts were cultured and differentiated as described previously (8). Differentiated 3T3-L1 adipocytes were transfected on day 4 postdifferentiation by electroporation with small interfering RNAs (siRNAs) as described previously (15). For adipocyte treatments, cells were serum starved for 3 h in Dulbecco modified Eagle medium (DMEM) plus 1% penicillin-streptomycin and then treated with the reagents described in the figure legends. For latrunculin B treatment, 3T3-L1 adipocytes were incubated without (dimethyl sulfoxide [DMSO]) or with 2 μM latrunculin B (in DMSO) in DMEM supplemented with 10% fetal bovine serum (Gibco) plus 1% penicillin-streptomycin (Gibco) for 3 h before coimmunoprecipitation (IP) experiments. HEK 293T cells were cultured in DMEM supplemented with 10% fetal bovine serum (Gibco) plus 1% penicillin-streptomycin (Gibco). For HEK 293T transfections, 3 × 10⁶ cells per 150-mm plate were plated the day before transfection. The following day, the 293T cells were transfected with 10 μg of myc-Rictor and 10 μg of either 3×HA-Myo1c(full), 3×HA-Myo1c(motor+IQ), 3×HA-Myo1c(IQ+tail), or 3×HA-Myo1c(tail) construct DNA, using Lipofectamine 2000 (Invitrogen). At 6 h posttransfection, the medium was changed, and at 48 h posttransfection, IP and immunoblot assays were performed.

siRNA oligonucleotides.

siRNA duplexes were synthesized and purified by Dharmacon Research, Inc. (Lafayette, CO). The following siRNA oligonucleotides were used in this study: Myo1c siRNA, 5′-GATCATCTGTGACCTGGTATT-3′; Rictor siRNA, 5′-GCGAGCTGATGTAGAATTGTT-3′; and scrambled siRNA, 5′-CAGTCGCGTTTGCGACTGG-3′.

MEF IP assays.

Rictor wild-type (WT) and knockout mouse embryonic fibroblasts (MEFs) were scraped into lysis buffer (20 mM Tris base, pH 7.9, 20 mM NaCl, 1 mM EDTA, 5 mM EGTA, 20 mM beta-glycerophosphate, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 25 nM calyculin A, 1 tablet/50 ml protease inhibitor [Roche Molecular Biochemicals]) containing 0.25% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS). Lysates were centrifuged at 13,500 rpm for 10 min. Aliquots of the supernatants containing equal amounts of protein, as measured by Bradford assay (Bio-Rad), were added to 15 μl of settled protein A DynaBeads (Invitrogen) preincubated with normal rabbit immunoglobulin G (IgG) or Rictor polyclonal antibody and incubated at 4°C for 2 h. The Rictor antibody was raised against a prokaryotic recombinant glutathione S-transferase (GST)-Rictor (1507-1708) polypeptide and affinity purified by binding to a biotinylated Rictor (1507-1708) polypeptide expressed in the pinpoint vector. Beads were washed three times with 1 ml of lysis buffer, twice with 1 ml of lysis buffer containing 0.5 M NaCl, and again with 1 ml of lysis buffer. The beads with adsorbed proteins were added to lysis buffer and 1× sodium dodecyl sulfate (SDS) sample buffer and heated at 95°C for 10 min. Samples were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). The gel was silver stained, and bands were excised and submitted for tryptic digestion and mass spectrum analysis.

3T3-L1 adipocyte and 293T IP assays.

3T3-L1 murine fibroblasts were grown to confluence in 150-mm dishes and differentiated as described previously (8). On day 7 postdifferentiation, the adipocytes were serum starved for 3 h in DMEM plus 1% penicillin-streptomycin and either left serum starved or stimulated with 100 nM insulin for 15 min. HEK 293T fibroblasts were also grown and transfected in 150-mm dishes. Following treatments or transfections, cells were rinsed twice with cold phosphate-buffered saline (PBS) and lysed in 1 ml of ice-cold S-50 lysis buffer (50 mM HEPES [pH 7.4], 5 mM sodium pyrophosphate, 5 mM β-glycerophosphate, 10 mM sodium fluoride, 2 mM EDTA, 2 mM sodium orthovanadate [Na₃VO₄], 1 mM phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin, 10 μg/ml leupeptin) containing either 0.3% CHAPS or 1% Triton X-100. Lysates were centrifuged at 13,000 rpm for 10 min. Total protein concentration was measured by Bradford assay (Bio-Rad). Ten micrograms of each indicated antibody or control IgG was added to 2 mg of total protein and incubated overnight at 4°C, followed by incubation with swelled protein A- or protein G-Sepharose beads (Sigma) for 2 h at 4°C. The beads with captured proteins were washed four times with lysis buffer before being boiled for 5 min in Laemmli buffer. Samples were resolved by SDS-PAGE and proteins transferred to polyvinylidene fluoride membranes for immunoblotting.

Immunoblot assays.

For immunoblots, cells were lysed with the indicated detergent, and total protein concentration was measured as described previously (14). Unless otherwise stated, 50 μg of total protein was resolved by SDS-PAGE. Proteins were transferred to polyvinylidene fluoride membranes, and membranes were probed with specific monoclonal or polyclonal antibodies overnight at 4°C. Bound primary antibody was visualized using horseradish peroxidase-conjugated secondary antibody (Vector Laboratories) and enhanced chemiluminescence reagents (Pierce).

Live-cell imaging and fixed-cell rhodamine-phalloidin staining.

For live-cell imaging experiments, day 4 3T3-L1 adipocytes were cotransfected with 50 μg plasmid DNA and 10 nM siRNA, as described previously (15), and seeded into 35-mm-diameter plastic tissue culture dishes with glass coverslip bottoms (Mat-Tek) for live-cell imaging, six-well culture plates for fixed-cell assay, and 100-mm culture plates for immunoblot analysis. At 72 h posttransfection, the adipocytes were serum starved for 5 h. For live-cell imaging, adipocytes were washed three times and incubated in Krebs-Ringer-HEPES buffer containing 0.5% bovine serum albumin and 1 mM sodium pyruvate (KBP buffer). Images of fluorescently labeled live adipocytes were obtained using a Yokogawa spinning-disk confocal scan head (model CSU10; Solamere Technology Group, Salt Lake City, UT) illuminated with an argon-ion laser (Reliant 150 M; Laser Physics, West Jordan, UT) set at 488 nm. Images of live cells were collected at a frequency of 1 image every 5 seconds for 10 min, with an exposure time of 100 ms. Metamorph Image acquisition and analysis software (Universal Imaging) was used to control the hardware, acquire data, and process final images. For fixed-cell analysis of membrane ruffles, 3T3-L1 adipocytes were washed three times with PBS and fixed in 4% paraformaldehyde in PBS for 20 min at room temperature. Cells were then permeabilized with PBS containing 0.1% Triton X-100 for 20 min and stained with rhodamine-phalloidin (Invitrogen) for 30 min. Samples were viewed by fluorescence microscopy using a Zeiss Axiophot2 fluorescence microscope (Zeiss, New York, NY) and Axiovision software.

Cell quantification.

Following rhodamine-phalloidin staining of adipocytes, slides were analyzed using a Zeiss Axiophot fluorescence microscope. Adipocytes were identified by the presence of lipid droplets. Fluorescein isothiocyanate-positive (positive for GFP empty vector [control] or GFP-Myo1c) 3T3-L1 adipocytes were evaluated for the presence of membrane ruffles. For determination of the percentage of cells showing membrane ruffles, adipocytes that were positive for GFP-Myo1c and displayed dense actin staining at the cell periphery were scored as positive for membrane ruffles. This number was divided by the total number of GFP-Myo1c-positive cells analyzed. Results were reported as percentages of cells showing membrane ruffles. For membrane ruffle size analysis, the ruffle size per cell size on GFP-Myo1c-positive adipocytes from both scrambled siRNA- and Rictor siRNA-transfected cells was measured using Axiovision software and normalized to GFP intensity for each cell analyzed. At least 60 cells were analyzed for each experiment, and data shown are the averages for three independent experiments. All results are given as means ± standard errors, unless stated otherwise.

Statistical analysis.

Statistical differences were determined by analysis of variance. P values of <0.05 were considered significant.

RESULTS

Myo1c is identified as a potential interacting partner with Rictor in WT and Rictor-deficient MEFs.

To better understand how Rictor functions to promote full Akt activation and cortical actin remodeling, we attempted to identify proteins that coimmunoprecipitate with Rictor. In order to assert the specificity of polypeptides in IP protocols, we took advantage of Rictor-deficient MEFs derived from Rictor knockout mice as well as MEFs from WT mice (27). Although the Rictor knockout MEFs grew poorly, displaying growth arrest after six or seven divisions, we were able to successfully immunoprecipitate Rictor from both WT and Rictor-deficient MEFs. In our IP experiments, the majority of the bands recovered from the WT and Rictor-depleted MEFs were essentially identical (Fig. 1A). However, there were three regions on the gel where the polypeptide bands observed for WT immunoprecipitates were missing for Rictor-deficient MEF lysates (Fig. 1A and B [enhanced in Fig. 1B]). Bands in these regions were excised from WT MEF lysate lanes and analyzed by tryptic digestion and mass spectrometry. The bands in the region near a molecular weight (MW) of 212,000 were identified to be mTOR and Rictor, thus strongly supporting the validity of our approach. The third band, near an MW of 116,000 (Fig. 1B), was identified as the unconventional myosin Myo1c (Fig. 1C). The three bands corresponding to mTOR, Rictor, and Myo1c that appeared in the immunoprecipitates from WT MEFs were absent from the corresponding gel segment for the IPs performed with Rictor-deficient MEFs (Fig. 1A and B [enhanced in Fig. 1B]), suggesting that in addition to mTOR, Rictor coimmunoprecipitates with the unconventional myosin Myo1c.

FIG. 1.

Rictor coimmunoprecipitates with the unconventional myosin Myo1c. (A) Identification of Myo1c as an immunoprecipitating partner of Rictor. Lysates from WT and Rictor-deficient MEFs (from Rictor knockout mice) were subjected to control rabbit IgG or Rictor polyclonal antibody IP. After electrophoresis of samples in SDS-polyacrylamide gels, the gels were silver stained. The bands in the regions shown with solid triangles were excised and identified by mass spectrometry analysis. The proteins that immunoprecipitated with Rictor from WT MEFs but were absent in samples treated with control rabbit IgG and in samples derived from Rictor knockout MEFs were analyzed further. As shown in the top panel, the upper two bands were mTOR (above MW of 212,000) and Rictor (below MW of 212,000). MWs in figures are thousands. (B) The boxed region of the gel in panel A was enhanced for better visualization of a band in the region around an MW of 116,000. This band was identified as Myo1c. (C) Quantification of results. Whereas five Myo1c peptides were found following IP of Rictor from WT MEFs, no Myo1c peptides were found following IP of Rictor from Rictor-deficient MEFs. Other proteins were identified, and the numbers of peptides found following Rictor IP from WT and Rictor knockout (KO) MEFs are shown. (D) IP between endogenous Myo1c and Rictor in 3T3-L1 adipocytes subjected to 0.3% CHAPS lysis conditions. Differentiated 3T3-L1 adipocytes were serum starved or stimulated with 100 nM insulin for 15 min and then lysed with 0.3% CHAPS lysis buffer. IP studies were then performed using the 0.3% CHAPS 3T3-L1 adipocyte lysates and antibodies to mTOR, Rictor, Myo1c, and nonimmune rabbit IgG (pull down). Membranes were then blotted for mTOR, Rictor, Myo1c, and Raptor. (E) IP of endogenous Myo1c and Rictor in 3T3-L1 adipocytes under 1% Triton X-100 lysis conditions. Differentiated 3T3-L1 adipocytes were serum starved or stimulated with 100 nM insulin for 15 min and then lysed with 1% Triton X-100 lysis buffer. IP studies were then performed using the 1% Triton X-100 3T3-L1 adipocyte lysates and antibodies to mTOR, Rictor, Myo1c, and nonimmune rabbit IgG (pull down). Membranes were then blotted for mTOR, Rictor, Myo1c, and Raptor. The results shown are representative of three independent experiments. In all 3T3-L1 adipocyte IP experiments 50 μg of total protein was used for lysate blots and 2 mg of total protein was used for IP experiments. −, basal conditions; +, 100 nM insulin stimulation.

Rictor and Myo1c coimmunoprecipitate under basal and insulin-stimulated conditions in 3T3-L1 adipocytes.

Myo1c had previously been identified in our laboratory as an important factor in potentiating insulin-stimulated GLUT4 translocation in 3T3-L1 adipocytes (3, 4). To confirm IP between endogenous Rictor and Myo1c in 3T3-L1 adipocytes, we performed further experiments. We observed that Rictor coimmunoprecipitated with Myo1c in 3T3-L1 adipocytes under both basal and insulin-stimulated (100 nM) conditions (Fig. 1D). No significant changes in amounts of Myo1c that precipitated with Rictor occurred following insulin stimulation. Consistently, by immunoprecipitating Myo1c with anti-Myo1c, we were also able to precipitate Rictor, confirming the existence of an endogenous Rictor-Myo1c complex in 3T3-L1 adipocytes. Under lysing conditions containing 0.3% CHAPS, Rictor but not Myo1c was also observed to coimmunoprecipitate with mTOR, but neither Rictor nor Myo1c was observed to coimmunoprecipitate with Raptor (Fig. 1D).

The Rictor-mTOR complex has previously been reported to be sensitive to detergent and disrupted by 1% Triton X-100 (23). Therefore, we determined whether the Rictor-Myo1c complex was also sensitive to detergent by immunoprecipitating Rictor or Myo1c from 3T3-L1 adipocyte lysates prepared with 1% Triton X-100. Consistent with the previously published results (23), mTOR did not coprecipitate with Rictor from lysates prepared in 1% Triton X-100 (Fig. 1E). In contrast, the Rictor-Myo1c complex was maintained under these conditions, and no apparent difference was observed when cells were treated with or without insulin (Fig. 1E). Furthermore, under these lysing conditions, neither Rictor nor Myo1c coimmunoprecipitated with mTOR or Raptor (Fig. 1E). Altogether, these results confirm that there are endogenous complexes containing Rictor and significant amounts of Myo1c in 3T3-L1 adipocytes under both basal and insulin-stimulated conditions.

The Rictor-Myo1c complex, unlike the Rictor-mTOR complex, is stable through chronic rapamycin treatment of cultured adipocytes.

The data presented in Fig. 1 suggested that the Rictor-Myo1c complex might be biochemically distinct from the Rictor-mTOR complex. To test this idea further, we determined whether the Rictor-Myo1c complex was sensitive to rapamycin, since prolonged rapamycin treatment of 3T3-L1 adipocytes and other cells is known to result in physical and functional disruption of the Rictor-mTOR complex (25). Consistent with published results, treatment of 3T3-L1 adipocytes in cell culture medium with 100 nM rapamycin for periods of up to 72 h significantly decreased phosphorylation of Akt at Ser473 (Fig. 2A and B). This correlated with a reduction in Rictor-mTOR interaction, initially visible 48 h following rapamycin treatment (Fig. 2C and D). These data suggest a disassembly of mTORC2, which mediates Akt activation at Ser473. Interestingly, protracted treatment of 3T3-L1 adipocytes with 100 nM rapamycin did not result in a significant disruption of the Rictor-Myo1c complex (Fig. 2C and D), suggesting that the Rictor-Myo1c complex differs in its sensitivity to rapamycin compared to the Rictor-mTOR complex.

FIG. 2.

Rictor-Myo1c complex is insensitive to chronic rapamycin treatment. (A) Effect of prolonged rapamycin treatment on Akt phosphorylation (serine 473) in 3T3-L1 adipocytes. Differentiated 3T3-L1 adipocytes were treated with 100 nM rapamycin for 0, 1, 24, 48, or 72 h. Adipocytes from each time point were lysed with 0.3% CHAPS lysis buffer, and lysates were immunoblotted for Akt phosphorylation at serine 473 as well as for total Akt. The images shown are representative blots from three independent experiments. (B) Quantitative assessment of the results in panel A, based on densitometry analysis of three independent experiments. (C) Effect of prolonged rapamycin treatment on Rictor-mTOR and Rictor-Myo1c complex stability. IP studies were performed by immunoprecipitating Rictor and immunoblotting for mTOR or Myo1c, using the 0.3% CHAPS 3T3-L1 adipocyte lysates from panel A. The images shown are representative blots from three independent experiments. (D) Quantitative assessment of the results in panel C, based on densitometry analysis of three independent experiments. In all 3T3-L1 adipocyte IP experiments, 50 μg of total protein was used for lysate blots and 2 mg of total protein was used for IP experiments. R, rapamycin treatment; pAkt, phospho-Akt (serine 473); IB, immunoblot. *, P < 0.05.

The head domain of Myo1c is responsible for Myo1c's interaction with Rictor.

We next sought to determine the domain of Myo1c responsible for the interaction with Rictor. Myo1c is a class I myosin, consisting of three major domains, namely, a head domain, a neck domain, and a tail domain (Fig. 3A). The head domain has conserved ATP- and actin-binding sites, and this domain has been suggested to be important in actin polymerization in other class I myosins (18). The neck domain contains three isoleucine-glutamine (IQ) binding sites that bind calmodulin, and it has been implicated in Ca²⁺-sensitive membrane binding in hair cells (9). The Myo1c tail domain is rich in basic amino acid residues and has been reported to anchor the protein to inositol 1,4,5-trisphosphate- and phosphatidylinositol 4,5-bisphosphate-rich regions of the plasma membrane (10). To determine which domain of Myo1c is responsible for the association with Rictor, we prepared four different triple influenza virus HA (3×HA)-tagged Myo1c constructs, namely, 3×HA-Myo1c(motor+IQ+tail) (full length), 3×HA-Myo1c(motor+IQ), 3×HA-Myo1c(IQ+tail), and 3×HA-Myo1c(tail) (Fig. 3A). Each construct was cotransfected with a myc-Rictor (full-length) construct into HEK 293T cells, and IP studies were performed by immunoprecipitating myc-Rictor and immunoblotting for HA. In these IP studies, we observed that only the full-length [3×HA-Myo1c(motor+IQ+tail)] and 3×HA-Myo1c(motor+IQ) Myo1c constructs coimmunoprecipitated with myc-Rictor, suggesting that Rictor's association with Myo1c occurred through its motor domain (Fig. 3B). Furthermore, DMSO treatment alone of 3T3-L1 adipocytes had no effect on cortical actin (Fig. 3C, top panels), but latrunculin B treatment of 3T3-L1 adipocytes disrupted the integrity of cortical actin (Fig. 3C, bottom panels). However, we observed that actin microfilament disruption had no effect on Rictor-Myo1c IP (Fig. 3D), suggesting that the Rictor-Myo1c complex is not mediated by intact actin filaments.

FIG. 3.

Rictor coimmunoprecipitates with the motor domain of Myo1c. (A) Triple-HA-tagged Myo1c constructs are shown. In addition to the full-length Myo1c(motor+IQ+tail) construct, Myo1c(motor+IQ), Myo1c(IQ+tail), and Myo1c(tail) constructs were prepared for cotransfection and IP experiments with a myc-tagged Rictor construct. (B) IP studies of myc-Rictor and 3×HA-Myo1c constructs. HEK 293T cells were cotransfected with myc-Rictor and either 3×HA-Myo1c(full), 3×HA-Myo1c(motor+IQ), 3×HA-Myo1c(IQ+tail), or 3×HA-Myo1c (tail). At 48 h posttransfection, cells were lysed with 1% Triton X-100 lysis buffer. Lysates were immunoprecipitated with the myc tag and immunoblotted for the HA or myc tag. Nonimmune rabbit IgG was used as a control for IP. The image shown is a representative blot from three independent experiments. (C) Assessment of latrunculin B effect on cortical actin integrity. 3T3-L1 adipocytes were incubated with or without 2 μM latrunculin B. At 3 h postincubation, cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and stained with rhodamine-phalloidin for cortical actin analysis. The images shown are representative frames of cells for each condition. Bars, 10 μm. (D) IP studies of endogenous Rictor and Myo1c following latrunculin B treatment. 3T3-L1 adipocytes were incubated without (−) or with (+) 2 μM latrunculin B. At 3 h postincubation, cells were lysed with 0.1% Triton X-100 lysis buffer. Lysates were immunoprecipitated with Rictor and immunoblotted for Myo1c. Nonimmune rabbit IgG was used as a control for IP. The images shown are representative blots from three independent experiments. MWs in the figure are thousands.

Myo1c is not required for mTORC2-dependent Akt activation in 3T3-L1 adipocytes.

Since the Rictor-mTOR complex was previously implicated in the phosphorylation of Akt at serine 473 and since Rictor also coimmunoprecipitates with Myo1c (Fig. 1), we tested whether Myo1c is necessary for mTORC2-mediated Akt phosphorylation. With increasing doses of insulin, Rictor knockdown cells showed a reduction in the level of Akt phosphorylation at serine 473 compared to control cells, whereas Myo1c-depleted cells did not (Fig. 4A and B). These results suggest that Myo1c is not required for mTORC2-dependent Akt activation.

FIG. 4.

Rictor depletion but not Myo1c depletion results in decreased Akt (serine 473) phosphorylation. (A) siRNA-mediated Rictor and Myo1c silencing. Differentiated 3T3-L1 adipocytes were transfected with either scrambled siRNAs (S), siRNAs to Rictor (R), or siRNAs to Myo1c (M). At 72 h posttransfection, the adipocytes were serum starved for 3 h and then left serum starved (0.0 nM insulin) or stimulated with 0.1 nM, 1 nM, 10 nM, or 100 nM insulin for 15 min. Lysates from each condition were analyzed by immunoblotting to determine the levels of Rictor depletion, Myo1c depletion, Akt phosphorylation (serine 473), and total Akt. The images shown are representative immunoblots from three independent experiments. (B) Quantitative assessment of the results in panel A, based on densitometry analysis of three independent experiments. Significance levels for comparison of Myo1c siRNA-transfected cells to Rictor siRNA-transfected cells are as follows: *, P < 0.05; **, P < 0.01. Significance levels for comparison of scrambled siRNA-transfected cells to Rictor siRNA-transfected cells are as follows: #, P < 0.05; ##, P < 0.01.

Myo1c or Rictor depletion compromises paxillin phosphorylation at tyrosine 118.

Previous work has shown that Rictor and the mTORC2 can also regulate actin organization (12, 23). Interestingly, data from our laboratory suggest that Myo1c overexpression in 3T3-L1 adipocytes induces insulin-independent cortical actin remodeling (4). We therefore hypothesized that the Rictor-Myo1c complex may function to regulate dynamic cortical actin events. Tyrosine phosphorylation of paxillin has been associated with regulation of the dynamics of the actin network (28), and experiments were designed to test the effect of Rictor or Myo1c silencing on paxillin phosphorylation at tyrosine 118. We observed a considerable reduction in paxillin phosphorylation at tyrosine 118 following either Rictor or Myo1c depletion (Fig. 5A), suggesting that both Rictor and Myo1c are required for optimal regulation of dynamic actin polymerization.

FIG. 5.

Rictor and Myo1c both regulate paxillin phosphorylation at tyrosine 118, in an mTOR- and PI3K-independent manner. (A) Rictor and Myo1c silencing compromises paxillin phosphorylation at tyrosine 118 (Y118). Differentiated 3T3-L1 adipocytes were transfected with either scrambled siRNAs, siRNAs to Rictor, or siRNAs to Myo1c. At 72 h posttransfection, the adipocytes were lysed in 1% SDS buffer and immunoblots were performed to assess Rictor, Myo1c, phospho-paxillin (Y118), and total paxillin levels. The images shown are representative blots from three independent experiments. (B) Effect of prolonged rapamycin treatment on phospho-paxillin level. Differentiated 3T3-L1 adipocytes were either left unstimulated and untreated (−/−), left unstimulated but treated with 100 nM rapamycin for 72 h (−/+), stimulated with 100 nM insulin for 15 min but left untreated (+/−), or stimulated with 100 nM insulin for 15 min and treated with 100 nM rapamycin for 72 h (+/+). In all cases where it was present, insulin was added for the final 15 min of the experiment. Following stimulation and treatment, the adipocytes were lysed in 1% SDS and immunoblots were performed to assess phospho-paxillin (Y118), total paxillin, phospho-Akt (S473), and total Akt levels. The images shown are representative blots from three independent experiments. (C) Effect of wortmannin treatment on phospho-paxillin level. Differentiated 3T3-L1 adipocytes were either left unstimulated and untreated (−/−), treated with 100 nM wortmannin for 30 min but left unstimulated (−/+), left untreated but stimulated with 100 nM insulin for 15 min (+/−), or treated with 100 nM wortmannin for 30 min and stimulated with 100 nM insulin for 15 min (+/+). Following treatment and stimulation, the adipocytes were lysed and immunoblots were performed to assess phospho-paxillin (Y118), total paxillin, phospho-Akt (S473), and total Akt levels. The images shown are representative blots from three independent experiments. Ins, 100 nM insulin; Rap, 100 nM rapamycin; Wort, 100 nM wortmannin.

Paxillin phosphorylation at tyrosine 118 is not dependent on mTOR or phosphatidylinositol 3-kinase (PI3K).

Whereas earlier studies with both yeast and mammalian cells suggested that the functions of the mTORC2 complex are not compromised by the mTOR inhibitor rapamycin (12), other results (25) (Fig. 2) imply that protracted rapamycin treatment of cells destabilizes mTORC2 and compromises its ability to phosphorylate Akt. Apart from phosphorylating Akt, the mTORC2 complex is believed to influence actin organization, although the effect of prolonged rapamycin treatment on actin organization has not been studied. We observed that Akt phosphorylation at serine 473 was less pronounced in insulin-plus-rapamycin-treated adipocytes than in adipocytes treated with insulin alone, indicating that the rapamycin treatment was effective, as expected (Fig. 5B). In contrast, paxillin phosphorylation at tyrosine 118 was unaffected by prolonged treatment of 3T3-L1 adipocytes with 100 nM rapamycin (Fig. 5B), suggesting that mTOR activity was not necessary for paxillin phosphorylation and actin reorganization.

Our previous work suggests that Myo1c regulates cortical actin remodeling in a PI3K-independent manner (4), and the results presented here (Fig. 5) are consistent with a model in which Myo1c in complex with Rictor participates in dynamic cortical actin polymerization. Taken together, these observations lead to the prediction that paxillin phosphorylation at tyrosine 118 should not be affected by inhibition of the PI3K pathway. As expected, Akt phosphorylation at serine 473 was significantly inhibited in adipocytes treated with insulin plus wortmannin relative to that after treatment with insulin alone (Fig. 5C). As with rapamycin-treated cells, paxillin phosphorylation at tyrosine 118 was not affected by treatment of 3T3-L1 adipocytes with 100 nM wortmannin (Fig. 5C), suggesting that paxillin phosphorylation at tyrosine 118 and dynamic cortical actin events were regulated in a PI3K-independent manner. These results show that neither chronic rapamycin treatment nor wortmannin has an effect on paxillin phosphorylation at tyrosine 118 and thus suggest that the actions of Myo1c and Rictor on the regulation of actin dynamics are likely independent of PI3K and mTOR.

Rictor depletion abrogates Myo1c-induced membrane ruffling in 3T3-L1 adipocytes.

The data presented in Fig. 1 to 5 are consistent with a model in which Myo1c and Rictor both act in a pathway that results in cortical actin remodeling. We previously observed that overexpression of Myo1c in 3T3-L1 adipocytes results in dramatic insulin-independent membrane ruffling (4). To determine whether Rictor is required for Myo1c-induced membrane ruffling, Rictor was depleted in adipocytes expressing GFP-Myo1c, and these adipocytes were analyzed. Immunoblot analysis revealed that Rictor levels were significantly depleted in Rictor siRNA-transfected adipocytes (Fig. 6A). In live-cell imaging analysis, GFP-Myo1c- and scrambled siRNA-transfected adipocytes showed dramatic insulin-independent membrane ruffling, with Myo1c concentrated in the regions of membrane ruffling (Fig. 6B; see Movie S3 in the supplemental material), as we had previously observed (4). Strikingly, membrane ruffling was significantly subdued in GFP-Myo1c- and Rictor siRNA-transfected adipocytes (Fig. 6B; see Movie S4 in the supplemental material). Adipocytes cotransfected with GFP empty vector and scrambled or Rictor siRNA did not show any significant membrane ruffling (Fig. 6B; see Movies S1 and S2 in the supplemental material).

FIG. 6.

Rictor depletion attenuates Myo1c-induced, insulin-independent membrane ruffling in 3T3-L1 adipocytes. (A) Rictor knockdown in 3T3-L1 adipocytes cotransfected with GFP-Myo1c and scrambled siRNA or Rictor siRNA. Differentiated 3T3-L1 adipocytes were cotransfected with GFP empty vector (GFP-E/V) or GFP-Myo1c and scrambled siRNA or Rictor siRNA. At 72 h posttransfection, a subset of these cells were lysed in 1% SDS and immunoblots were performed to assess Rictor and Akt (loading control) levels. (B) Rictor depletion abrogates Myo1c-induced, insulin-independent membrane ruffling in 3T3-L1 adipocytes (live-cell analysis). Differentiated 3T3-L1 adipocytes were cotransfected with GFP empty vector or GFP-Myo1c and scrambled siRNA or Rictor siRNA. At 72 h posttransfection, a subset of these cells were serum starved for 5 h and then incubated in Krebs-Ringer-HEPES buffer containing 0.5% bovine serum albumin and 1 mM sodium pyruvate. Membrane ruffling was observed by live monitoring of the cells for 10 min, with imaging done at 5-s intervals. The images shown are bright-field images and images at three time frames, each 5 min apart, for a representative GFP empty vector- and scrambled siRNA-transfected adipocyte, a GFP-Myo1c- and scrambled siRNA-transfected adipocyte, a GFP empty vector- and Rictor siRNA-transfected adipocyte, and a GFP-Myo1c- and Rictor siRNA-transfected adipocyte. Scr, scrambled siRNA; Ric, Rictor siRNA. Bars, 10 μm.

In order to quantitatively assess the effect of Rictor depletion on Myo1c-induced membrane ruffling, transfected adipocytes were serum starved prior to fixation and then permeabilized and probed with rhodamine-phalloidin. Whereas GFP-Myo1c- and scrambled siRNA-transfected adipocytes showed distinct and extensive membrane ruffles (Fig. 7A and B), the incidence of membrane ruffles (shown by discontinuous cortical rhodamine-phalloidin staining) in GFP-Myo1c- and Rictor siRNA-transfected adipocytes was significantly reduced (Fig. 7A and B). Furthermore, by analyzing membrane ruffles of both GFP-Myo1c- and scrambled siRNA-transfected and GFP-Myo1c and Rictor siRNA-transfected adipocytes, membrane ruffle size appeared to be significantly larger for GFP-Myo1c- and scrambled siRNA-transfected adipocytes than for GFP-Myo1c- and Rictor siRNA-transfected adipocytes (Fig. 7C). GFP empty vector-transfected adipocytes showed no significant membrane ruffles (Fig. 7A). Together, these data suggest that Rictor is required for Myo1c-induced, insulin-independent membrane ruffling in 3T3-L1 adipocytes.

FIG. 7.

Rictor depletion abrogates Myo1c-induced, insulin-independent membrane ruffling in 3T3-L1 adipocytes (fixed-cell analysis). (A) Differentiated 3T3-L1 adipocytes were cotransfected with GFP empty vector or GFP-Myo1c and scrambled siRNA or Rictor siRNA and then seeded on glass coverslips. At 72 h posttransfection, these cells were serum starved for 5 h, fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and stained with rhodamine-phalloidin for membrane ruffle analysis. The images show representative adipocytes for each condition. Bars, 10 μm. (B) For each condition, the percentage of ruffled cells was determined by scoring the number of GFP-Myo1c-positive cells that had membrane ruffles against the total number of cells counted. At least 60 cells were evaluated in each experiment. The averaged data from three independent experiments are shown, with standard deviations. **, P < 0.01. (C) Effect of Rictor knockdown on Myo1c-induced, insulin-independent membrane ruffle size. 3T3-L1 adipocytes were transfected and prepared as described above. For each condition, membrane ruffle size was determined by measuring the size of discernible membrane ruffles for GFP-Myo1c-positive cells under each condition and dividing it by the total size of the cell, as determined using Axiovision software. For each cell evaluated, the GFP intensity was determined and the average ruffle size/total cell size ratio from each condition was normalized to the average GFP intensity of all evaluated cells under each condition. At least 60 cells were assessed in each experiment. The box-and-whisker plots show (from the top down) maximum, third quartile, median, first quartile, and minimum values. **, P < 0.01.

DISCUSSION

The key novel finding reported here is the unexpected IP of the unconventional myosin Myo1c with Rictor, a previously identified protein in the mTORC2 that regulates Akt activation (Fig. 1 to 3). Remarkably, the data presented indicate that there are at least two biochemically and functionally distinct Rictor complexes in 3T3-L1 adipocytes, namely, the Rictor-mTOR complex (mTORC2) and the Rictor-Myo1c complex (Fig. 8). The Rictor-mTOR complex is sensitive to Triton X-100 and long-term treatment of cells with rapamycin, whereas the Rictor-Myo1c complex is not (Fig. 2). Functionally, the Rictor-mTOR complex regulates Akt phosphorylation at Ser473 (Fig. 4), while the Rictor-Myo1c complex seems to participate in cortical actin remodeling and membrane ruffling (Fig. 6 and 7).

FIG. 8.

Working model suggesting that Rictor may function in two pathways. Our data suggest that Myo1c and Rictor form a complex in 3T3-L1 adipocytes that is biochemically and functionally distinct from the Rictor-mTOR complex. While the Rictor-mTOR complex (I) that phosphorylates and activates Akt is rapamycin sensitive, our data suggest that the Rictor-Myo1c complex (II) acts in an insulin-, rapamycin-, and wortmannin-insensitive manner to participate in dynamic cortical actin events.

In a previous study designed to understand the role of Myo1c in GLUT4 trafficking and membrane dynamics, we observed that Myo1c overexpression induced dramatic cortical actin remodeling (membrane ruffling) in 3T3-L1 adipocytes, in a serum- and insulin-independent manner (4). This observation suggested that Myo1c might mediate the effect of insulin on membrane ruffling in 3T3-L1 adipocytes, which is supported by our observation that Myo1c depletion in 3T3-L1 adipocytes does indeed attenuate insulin-induced membrane ruffling (data not shown). Interestingly, expression of Myo1c in cultured adipocytes induces membrane ruffling even in the presence of wortmannin, suggesting that Myo1c may function downstream or independent of PI3K in activating membrane ruffling. Our studies here suggest that formation and maintenance of the Rictor-Myo1c complex are not dependent on insulin, rapamycin, or wortmannin (Fig. 1, 2, and 4). Therefore, to determine the functional relevance of Rictor's association with Myo1c, we performed experiments that focused on Myo1c-induced, not insulin-, rapamycin-, or wortmannin-dependent, membrane ruffling. By using insulin-independent, Myo1c-induced membrane ruffling as the readout for a specific Myo1c-dependent event, we could test whether or not Rictor participated in this event. Furthermore, our previous results (Fig. 1 to 5) suggested that the Rictor-mTOR complex is independent of the Rictor-Myo1c complex, and thus we eliminated possible contributions of the Rictor-mTOR complex (whose activity is dependent on serum and insulin) and observed the role of Rictor directly on Myo1c function. The data presented in Fig. 6 and 7 clearly show that Rictor is required for Myo1c-mediated membrane ruffling. Taken together, our results suggest that Myo1c overexpression induces insulin-independent cortical actin remodeling and that Rictor depletion attenuates this Myo1c-induced event.

Tyrosine 118 phosphorylation of the scaffolding protein paxillin is thought to be an important regulator of cytoskeleton-dependent changes in a cell (reviewed in reference 28). Insulin is known to induce dramatic 3T3-L1 adipocyte membrane ruffling (reviewed in reference 16), but we found that insulin stimulation does not result in a significant increase in the phosphorylation state of paxillin at tyrosine 118 (Fig. 5). This observation suggests that insulin-induced membrane ruffling in 3T3-L1 adipocytes has no effect on the phosphorylation state of paxillin at tyrosine 118, consistent with previous findings (31). However, we observed that Rictor or Myo1c depletion resulted in a significant reduction in paxillin phosphorylation at tyrosine 118 (Fig. 5). These data indicate that the effects of insulin versus the effects of expression of Myo1c are mediated, at least in part, by independent mechanisms leading to the induction of membrane ruffling in 3T3-L1 adipocytes. Whereas our result is novel, it is consistent with previously published work from our lab (21) and others (31) suggesting that there may be mechanisms that regulate actin dynamics in adipocytes that are distinct from the insulin-dependent pathways. Furthermore, the idea that the Rictor-Myo1c complex participates in cortical actin remodeling events that are distinct from the events regulated by insulin is supported by our observation that neither PI3K inhibition by wortmannin nor mTOR inhibition by rapamycin affects paxillin phosphorylation at tyrosine 118 in 3T3-L1 adipocytes (Fig. 5). Further work is required to uncover the underlying mechanisms that regulate actin dynamics in response to insulin and Myo1c.

The participation of class I myosins in actin dynamics has been reported for yeast class I myosins, such as Myo3p and Myo5p, which are activated through a phosphorylation event at a serine residue in the head domain by members of the Ste20p family (32, 33). Even though there is no direct evidence of this regulation for mammalian myosin I orthologs, it is possible that mammalian class I myosins could be regulated by factors that interact with their head domains. Our finding that the Myo1c head domain is important for Myo1c association with Rictor supports such a hypothesis, but further work is required to test this.

Finally, a potential mechanism by which Rictor may regulate cortical actin remodeling is through activation of small GTPase activity. AVO3, the budding yeast ortholog of Rictor, is reported to contain a RasGEFN domain that exists N-terminal to the catalytic GDP/GTP exchange domain in some guanine nucleotide exchange factors for Ras-like small GTPases (6). This raises the possibility that Rictor may activate GTPase signaling (19). Rho family GTPases play an important role in the regulation of cortical actin dynamic events in numerous cell types (2), including adipocytes (16), and in NIH 3T3 cells mTORC2 has been shown to regulate actin organization through Rho family GTPases (12). It is thus conceivable that Myo1c in complex with Rictor could be part of a machinery that also leads to the activation of Rho family GTPases in 3T3-L1 adipocytes. Future work will be needed to test these hypotheses and to determine the mechanism through which Rictor-Myo1c participates in cortical actin dynamics.