Serine 474 phosphorylation is essential for maximal Akt2 kinase activity in adipocytes

Abstract

The Ser/Thr protein kinase Akt regulates essential biological processes such as cell survival, growth, and metabolism. Upon growth factor stimulation, Akt is phosphorylated at Ser⁴⁷⁴; however, how this phosphorylation contributes to Akt activation remains controversial. Previous studies, which induced loss of Ser⁴⁷⁴ phosphorylation by ablating its upstream kinase mTORC2, have implicated Ser⁴⁷⁴ phosphorylation as a driver of Akt substrate specificity. Here we directly studied the role of Akt2 Ser⁴⁷⁴ phosphorylation in 3T3-L1 adipocytes by preventing Ser⁴⁷⁴ phosphorylation without perturbing mTORC2 activity. This was achieved by utilizing a chemical genetics approach, where ectopically expressed S474A Akt2 was engineered with a W80A mutation to confer resistance to the Akt inhibitor MK2206, and thus allow its activation independent of endogenous Akt. We found that insulin-stimulated phosphorylation of four bona fide Akt substrates (TSC2, PRAS40, FOXO1/3a, and AS160) was reduced by ∼50% in the absence of Ser⁴⁷⁴ phosphorylation. Accordingly, insulin-stimulated mTORC1 activation, protein synthesis, FOXO nuclear exclusion, GLUT4 translocation, and glucose uptake were attenuated upon loss of Ser⁴⁷⁴ phosphorylation. We propose a model where Ser⁴⁷⁴ phosphorylation is required for maximal Akt2 kinase activity in adipocytes.

Introduction

The serine/threonine kinase Akt (protein kinase B) is a key regulatory node in a range of cell signaling networks that control essential biological processes such as cell survival, proliferation, and metabolism (1). Upon growth factor stimulation, there is an increase in plasma membrane phosphatidylinositol (3,4,5) triphosphate (PIP₃)⁷ (2, 3). Akt then binds PIP3 via its pleckstrin homology (PH) domain and accumulates at the plasma membrane (4, 5). This interaction causes a conformational change, allowing PDK1 to phosphorylate Akt within its kinase domain at Thr³⁰⁹ (in Akt2; Thr³⁰⁸ in Akt1 and Thr³⁰⁵ in Akt3) and mTORC2 to phosphorylate Akt within its hydrophobic motif (HM) at Ser⁴⁷⁴ (in Akt2; Ser⁴⁷³ in Akt1 and Ser⁴⁷² in Akt3) (6, 7). These phosphorylation events enhance Akt kinase activity, allowing Akt to phosphorylate a number of substrates and facilitate several biological processes. For example, phosphorylation of TSC2 and PRAS40 activates mTORC1 and protein synthesis (8–10), phosphorylation of AS160/TBC1D4 facilitates GLUT4 translocation and glucose uptake (11, 12), and phosphorylation of FOXO proteins regulates their transcriptional activity via nuclear exclusion (13, 14).

Although there is consensus that Akt Thr³⁰⁹ phosphorylation is required for its activation (6, 15–17), the role of Ser⁴⁷⁴ phosphorylation remains controversial. Initial studies that assessed Akt activity in vitro reported that loss of Ser⁴⁷⁴ phosphorylation severely diminished Akt kinase activity (6, 17). Accordingly, Akt crystal structures show that an active kinase conformation is only possible by mimicking Ser⁴⁷⁴ phosphorylation with a S474D substitution (18, 19). From this, a model arose whereby Ser⁴⁷⁴ phosphorylation was integral to a high level of Akt kinase activity.

However, the accuracy of this model has been disputed in more recent studies in cells, where Ser⁴⁷⁴ phosphorylation was required for Akt activity only toward specific substrates. These studies abolished Ser⁴⁷⁴ phosphorylation through genetic or pharmacological ablation of its upstream kinase, mTORC2. For example, in mouse embryonic fibroblasts lacking SIN1, one of the subunits of mTORC2, there is loss of FOXO1/3a phosphorylation, but other Akt substrates (TSC2 and GSK3) and mTORC1 effectors (S6K and 4EBP1) are unaffected (20). Similarly, phosphorylation of FOXO3a and AS160 is impaired, but GSK3β phosphorylation is unaltered, in adipocytes from mice lacking the mTORC2 subunit RICTOR (21). Further, rat adipocytes treated with small-molecule inhibitors of mTORC2 exhibit decreased phosphorylation of PRAS40 but not GSK3β (22). These studies have implicated Ser⁴⁷⁴ phosphorylation as a driver of Akt substrate specificity. Alternatively, a recent study using RICTOR knockout mice indicates that Ser⁴⁷⁴ phosphorylation is dispensable for maximal phosphorylation of AS160, PRAS40, FOXO1, and GSK3β in adipose tissue (23). Nevertheless, use of mTORC2 deletion or inhibition to study the role of Akt Ser⁴⁷⁴ phosphorylation may be confounded by several factors, including the induction of compensatory mechanisms because of a chronic absence of mTORC2 (24), lack of specificity of mTORC2 inhibitors (25), and interference from other downstream actions of mTORC2 that are independent of Akt Ser⁴⁷⁴ phosphorylation, such as other phosphorylation sites on Akt (26–28).

A chemical genetics approach originally designed by Green et al. (29) overcomes these disadvantages by facilitating acute prevention of Ser⁴⁷⁴ phosphorylation in cells without perturbing mTORC2 activity. An Akt phosphomutant can be engineered to contain a W80A mutation in its PH domain, which confers resistance to the allosteric pan-Akt inhibitor MK2206 without compromising its activation. Treatment of cells ectopically expressing this mutant with MK2206 acutely eliminates endogenous Akt activity, allowing specific analysis of the W80A Akt mutant (29–31). We utilized this system to address how loss of Ser⁴⁷⁴ phosphorylation influences Akt2 activity toward a number of substrates and insulin-regulated processes in 3T3-L1 adipocytes. We show that insulin-stimulated phosphorylation of four Akt substrates was reduced by ∼50% in the absence of Ser⁴⁷⁴ phosphorylation. Similarly, several insulin-regulated processes were partially impaired following loss of Ser⁴⁷⁴ phosphorylation. These findings support a model where Ser⁴⁷⁴ phosphorylation is essential for maximal Akt2 kinase activity in adipocytes.

Results

The Akt2 W80A mutation confers MK2206 resistance and is a powerful tool to study Akt2 mutants

Studying the activity of ectopically expressed mutant proteins in cells can be confounded by the presence of their endogenous counterparts. To circumvent this and examine the role of Akt2 Ser⁴⁷⁴ phosphorylation in 3T3-L1 adipocytes (hereafter referred to as adipocytes), we employed a chemical genetics approach originally designed by Green et al. (29). MK2206 is an Akt inhibitor that works by preventing Akt recruitment to the plasma membrane (PM), which is essential for its activation (5, 32, 33). However, Akt with the W80A mutation is resistant to MK2206 inhibition. Treatment of cells ectopically expressing W80A Akt with MK2206 prevents activation of endogenous Akt, allowing specific assessment of the W80A Akt mutant upon insulin stimulation (Fig. 1A) (29–31).

Figure 1.

The Akt2 W80A mutation confers MK2206 resistance and has negligible impact on Akt2 activation at the plasma membrane in 3T3-L1 adipocytes. A, the Akt W80A mutation confers resistance to the Akt inhibitor MK2206. Treatment of cells ectopically expressing W80A Akt with MK2206 prevents activation of endogenous Akt, allowing specific activation of the W80A Akt mutant upon insulin stimulation. IR, insulin receptor; P, phosphorylation site. B, 3T3-L1 adipocytes were electroporated with TagRFP-T-WT or W80A Akt2, and plasma membrane recruitment was assessed using live-cell TIRF microscopy. Adipocytes were exposed to 1 nm insulin. Representative images for 3–7 independent experiments are presented. C, quantification of B (WT, 41 cells from three independent experiments; W80A, 60 cells from seven independent experiments; mean ± S.E.). D, 3T3-L1 adipocytes were electroporated with TagRFP-T WT or W80A Akt2, and plasma membrane recruitment was assessed using live-cell TIRF microscopy. Adipocytes were exposed to 10 μm MK2206 for 5 min, followed by 1 nm insulin. Representative images for two independent experiments are presented. E, quantification of D (WT, 15 cells from two independent experiments; W80A, 59 cells from two independent experiments; mean ± S.E.).

To validate the W80A system, we assessed insulin-stimulated PM recruitment of TagRFP-T-WT and W80A Akt2 in adipocytes using live-cell total internal reflection fluorescence (TIRF) microscopy. There was no significant difference in the kinetics of insulin-stimulated WT and W80A Akt2 PM recruitment (Fig. 1, B and C). Aligning with previous studies (34, 35), both WT and W80A Akt2 exhibited an overshoot in PM localization in response to insulin. However, pretreatment with 10 μm MK2206 abolished PM accumulation of WT Akt2, whereas robust accumulation of W80A Akt2 was observed under identical conditions (Fig. 1, D and E). These data are consistent with the Akt2 W80A mutation conferring MK2206 resistance but having negligible impact on Akt2 recruitment to the PM.

We next generated a series of N-terminally FLAG-tagged Akt2 constructs that were engineered to express Akt2 with Thr³⁰⁹ or Ser⁴⁷⁴ mutated to alanine to prevent phosphorylation; WT Akt2 (WT), Akt2 with the W80A mutation (W80A), Akt2 with W80A and T309A mutations (W80A-T309A), and Akt2 with W80A and S474A mutations (W80A-S474A) (Fig. 2A). These were stably expressed in adipocytes to similar degrees at ∼75% of endogenous Akt2 (Fig. 2, B–D). Ectopic FLAG-Akt2 could be separated from endogenous Akt2 (Fig. S1). As expected, treatment of adipocytes expressing WT Akt2 with 10 μm MK2206 blocked insulin-stimulated phosphorylation of Akt at Thr³⁰⁹ and Ser⁴⁷⁴ and phosphorylation of its substrates TSC2, PRAS40, FOXO1/3a, and AS160 (Fig. 2, E and F). MK2206 did not completely block insulin-stimulated phosphorylation of the Akt substrate GSK3β at Ser⁹ (Fig. S2), suggesting GSK3β is also phosphorylated by another kinase at this site. Therefore, we did not use GSK3β phosphorylation as a readout of Akt activity in this study because we could not confidently attribute changes in GSK3β phosphorylation to Akt. In contrast, adipocytes expressing W80A Akt2 displayed robust insulin-stimulated phosphorylation of Akt and its substrates in the presence of 10 μm MK2206, again implicating W80A Akt2 as being MK2206-resistant (Fig. 2, E and F). Consistent with previous reports (6, 15–17), Thr³⁰⁹ phosphorylation was required for Akt2 activity because loss of phosphorylation at this site abolished insulin-stimulated phosphorylation of Akt substrates (Fig. 2, E and F). Intriguingly, loss of Thr³⁰⁹ phosphorylation resulted in hyperphosphorylation of Akt at Ser⁴⁷⁴ upon insulin stimulation, indicating interdependence between the two phosphorylation sites. These data highlight that the Akt2 W80A mutation confers resistance to MK2206 and that this can be used to analyze the activity of Akt2 mutants in the absence of endogenous Akt activity in adipocytes.

Figure 2.

W80A Akt2 is resistant to MK2206 and can be used to study Akt2 mutants in 3T3-L1 adipocytes independent of endogenous Akt. A, FLAG-tagged Akt2 constructs were generated: WT, W80A, W80A-T309A, and W80A-S474A Akt2. Each consisted of mutations in the PH domain, kinase domain (KD), and HM. B, lysates from 3T3-L1 adipocytes stably expressing empty vector (EV), WT, W80A, W80A-T309A, or W80A-S474A Akt2 were immunoblotted using antibodies as specified, with 14-3-3 as a loading control. A representative blot for three independent experiments is shown. C, quantification of FLAG expression in B (n = 3, mean ± S.E.). AU, arbitrary unit. D, quantification of Akt2 expression in B relative to the empty vector cell line, which represents endogenous Akt2 (n = 3, mean ± S.E.). E, 3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were treated with 10 μm MK2206 for 5 min followed by 1 nm insulin for 10 min. Lysates were immunoblotted with antibodies as specified, with 14-3-3 as a loading control. A representative blot for three independent experiments is presented. Band 1 and Band 2 indicate shifts in Akt2 gel migration. p, phosphorylated. F, quantification of E (n = 3, mean ± S.E.).

Akt2 Ser⁴⁷⁴ phosphorylation is required for maximal phosphorylation of TSC2, PRAS40, FOXO1/3a, and AS160

How Ser⁴⁷⁴ phosphorylation regulates Akt kinase activity remains controversial. MK2206-resistant W80A Akt was a valuable tool to assess the role of Ser⁴⁷⁴ phosphorylation in the activation of Akt2 in adipocytes through direct mutation of this phosphorylation site. We first assessed the role of Ser⁴⁷⁴ phosphorylation in Akt2 activity toward its canonical substrates TSC2, PRAS40, FOXO1/3a, and AS160. We did so by measuring the temporal dynamics by which these substrates were phosphorylated following insulin stimulation in adipocytes stably expressing W80A Akt2 or W80A-S474A Akt2. This was conducted in the presence of MK2206 to specifically activate the ectopic mutant. We have reported previously that a small degree of Akt phosphorylation is sufficient to achieve maximal substrate phosphorylation (36, 37); thus, we utilized a submaximal insulin dose (1 nm) to ensure that changes in Akt kinase activity were detectable.

In adipocytes expressing W80A Akt2, there was a time-dependent increase in phosphorylation of all substrates following insulin stimulation (Fig. 3, A and B). Distinct phosphorylation kinetics were observed between each substrate, aligning with previous reports of a lack of concordance in the behavior of Akt substrates (36, 38). Nevertheless, in adipocytes expressing W80A-S474A Akt2, substrate phosphorylation plateaued at ∼50% of that observed in adipocytes expressing W80A Akt2 (Fig. 3, A and B). These data suggest that Ser⁴⁷⁴ phosphorylation plays an important role in maximizing Akt2 kinase activity toward TSC2, PRAS40, FOXO1/3a, and AS160 in adipocytes, with no evidence for substrate specificity.

Figure 3.

Akt2 Ser⁴⁷⁴ phosphorylation is required for maximal insulin-stimulated phosphorylation of PRAS40, AS160, FOXO1/3a, and TSC2 in 3T3-L1 adipocytes. 3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were treated with 10 μm MK2206 for 5 min, followed by 1 nm insulin for the times specified. A, lysates were immunoblotted with antibodies as specified, with 14-3-3 as a loading control. A representative blot for three to four independent experiments is presented. Band 1 and Band 2 indicate shifts in Akt2 gel migration. p, phosphorylated. B, quantification of A (n = 3–4, mean ± S.E., two-way analysis of variance; *, p < 0.05; ns, not significant). AU, arbitrary unit; p, phosphorylated.

Throughout the insulin time course, total phosphorylation of Akt at Thr³⁰⁹ was not significantly different between adipocytes expressing W80A and W80A-S474A Akt2 (Fig. 3, A and B). This finding is corroborated by previous studies where loss of Ser⁴⁷⁴ phosphorylation did not affect Thr³⁰⁹ phosphorylation (6, 20), supporting the notion that defective Akt2 substrate phosphorylation upon loss of Ser⁴⁷⁴ phosphorylation is not due to decreased Thr³⁰⁹ phosphorylation.

Akt2 Ser⁴⁷⁴ phosphorylation is required for maximal insulin-stimulated mTORC1 activation and protein synthesis

We next assessed whether loss of Akt2 Ser⁴⁷⁴ phosphorylation alters the activation of the insulin signaling network beyond the immediate phosphorylation of Akt substrates. In response to insulin, Akt activates mTORC1 via phosphorylation of its substrates TSC2 and PRAS40 (8, 9). mTORC1 then phosphorylates its substrates, which facilitate a variety of responses, including an increase in the rate of protein synthesis (Fig. 4A) (10, 39, 40). Thus, we first assessed whether loss of Ser⁴⁷⁴ phosphorylation affects insulin-stimulated activation of mTORC1 and protein synthesis in adipocytes. We assessed mTORC1 activation by the phosphorylation status of its substrate, P70 S6K. In adipocytes expressing WT or kinase-dead W80A-T309A Akt2, MK2206 blocked the ability of Akt2 to activate mTORC1 in response to insulin (Fig. 4, B and C). Similarly, following MK2206 treatment, adipocytes expressing WT Akt2 could not facilitate insulin-stimulated protein synthesis (Fig. 4D), aligning with these events being activated downstream of Akt. In the presence of MK2206, adipocytes expressing W80A-S474A Akt2 exhibited partially impaired P70 S6K phosphorylation, S6 phosphorylation and protein synthesis compared with adipocytes expressing W80A Akt2 (Fig. 4, B–D). These data, together with diminished TSC2 and PRAS40 phosphorylation (Fig. 3, A and B), suggest that Akt2 Ser⁴⁷⁴ phosphorylation is essential for maximal insulin-stimulated mTORC1 activation and protein synthesis.

Figure 4.

Akt2 Ser⁴⁷⁴ phosphorylation is required for maximal insulin-stimulated mTORC1 activation, protein synthesis, FOXO nuclear exclusion, GLUT4 translocation, and glucose uptake in 3T3-L1 adipocytes. A, Akt facilitates insulin-stimulated GLUT4 translocation, glucose uptake, mTORC1 activation, protein synthesis, and FOXO nuclear exclusion. B, 3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were treated with 10 μm MK2206 for 5 min, followed by 1 nm insulin for 10 min. Lysates were immunoblotted with antibodies as specified, with 14-3-3 as a loading control. A representative blot for three independent experiments is presented. Band 1 and Band 2 indicate shifts in Akt2 gel migration. p, phosphorylated. C, quantification of B (n = 3, mean ± S.E., two-tailed paired t test; *, p < 0.05). AU, arbitrary unit. D, 3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were treated with 10 μm MK2206 for 5 min followed by 1 nm insulin for 1 h and assessed for [³H]leucine incorporation for protein synthesis (n = 4, mean ± S.E., two-tailed paired t test; *, p < 0.05). E, 3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were electroporated with FOXO1-mNeonGreen. Cells were treated with 10 μm MK2206 for 5 min, followed by 1 nm insulin and nuclear exclusion assessed using live-cell spinning-disk confocal microscopy. Representative images for three independent experiments are presented. F, quantification of E (WT, 45 cells from three independent experiments; W80A, 63 cells from three independent experiments; W80A-S474A, 63 cells from three independent experiments; mean ± S.E.). G, 3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were electroporated with pHluorin-GLUT4-mRuby3. Cells were treated with 10 μm MK2206 for 5 min followed by 1 nm insulin and GLUT4 translocation to the plasma membrane and assessed using live-cell TIRF microscopy. Representative images for three independent experiments are presented. H, quantification of G (WT, 78 cells from three independent experiments; W80A, 138 cells from three independent experiments; W80A-S474A, 135 cells from three independent experiments, mean ± S.E.). I, 3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were treated with 10 μm MK2206 for 5 min followed by 1 nm insulin for 20 min and assessed for [³H]2-deoxyglucose uptake (n = 4, mean ± S.E., two-tailed paired t test; *, p < 0.05).

Akt2 Ser⁴⁷⁴ phosphorylation is required for maximal insulin-stimulated FOXO nuclear exclusion

Akt phosphorylates FOXO proteins to regulate their transcriptional activity via nuclear exclusion (Fig. 4A) (13, 14). Therefore, we next assessed the functional consequence of Akt-mediated FOXO phosphorylation by examining whether Akt2 Ser⁴⁷⁴ phosphorylation is required for insulin-stimulated FOXO1 nuclear exclusion. In adipocytes expressing WT Akt2, MK2206 inhibited insulin-stimulated FOXO1 nuclear exclusion (Fig. 4, E and F), consistent with it being Akt-dependent. In adipocytes expressing W80A-S474A Akt2, the rate of FOXO1 nuclear exclusion was reduced compared to adipocytes expressing W80A Akt2 (Fig. 4, E and F). These observations, together with impaired FOXO1/3a phosphorylation (Fig. 3, A and B), suggest that Akt2 Ser⁴⁷⁴ phosphorylation is required for maximal insulin-stimulated FOXO nuclear exclusion.

Akt2 Ser⁴⁷⁴ phosphorylation is required for maximal insulin-stimulated GLUT4 translocation and glucose uptake

Finally, we assessed whether Akt2 Ser⁴⁷⁴ phosphorylation is required for insulin-stimulated GLUT4 translocation and glucose uptake in adipocytes. Akt activation is both necessary and sufficient for insulin-stimulated glucose uptake (41). This is predominantly achieved through the phosphorylation of AS160, which is required for translocation of GLUT4 to the PM (Fig. 4A) (30, 42). Consistent with this, MK2206 blocked insulin-stimulated GLUT4 translocation and glucose uptake in adipocytes expressing WT Akt2 (Fig. 4, G–I). In the presence of MK2206, both GLUT4 translocation and glucose uptake were activated by insulin in adipocytes expressing W80A Akt2; however, only partial activation was achieved in adipocytes expressing W80A-S474A Akt2 (Fig. 4, G–I). These observations, together with impaired AS160 phosphorylation (Fig. 3, A and B), suggest that Akt2 Ser474 phosphorylation is essential for maximal insulin-stimulated GLUT4 translocation and glucose uptake.

Discussion

Akt is a major regulatory node for numerous biological processes, but the mechanism by which it is activated remains controversial. It is widely accepted that Thr³⁰⁹ phosphorylation is essential for Akt activation (6, 15–17); however, the role of Ser⁴⁷⁴ phosphorylation is unclear. Here we examined how loss of Ser⁴⁷⁴ phosphorylation influences Akt2 activity toward a number of substrates and insulin-regulated processes in 3T3-L1 adipocytes. Insulin-stimulated phosphorylation of four canonical Akt substrates was reduced by ∼50% in the absence of Ser⁴⁷⁴ phosphorylation. Accordingly, we measured reduced efficiency of several insulin-regulated processes in the absence of Ser⁴⁷⁴ phosphorylation. Overall, our findings suggest that Ser⁴⁷⁴ phosphorylation regulates the full activation of Akt2 toward a cadre of substrates and biological processes in adipocytes.

Previous studies have proposed an alternate model where Ser⁴⁷⁴ phosphorylation is required for Akt activity only toward select substrates. Evidence for this model is largely derived from studies where Ser⁴⁷⁴ phosphorylation was prevented via ablation of its upstream kinase, mTORC2 (20–22). However, by interfering with the upstream kinase of Ser⁴⁷⁴, these studies have several disadvantages, such as possible induction of long-term compensatory mechanisms and interference with other downstream actions of mTORC2 independent of Akt Ser⁴⁷⁴ phosphorylation. MK2206-resistant W80A Akt offered a valuable tool to overcome these disadvantages and interrogate the role of Akt2 Ser⁴⁷⁴ phosphorylation in cells through direct mutation of Ser⁴⁷⁴. Despite examining a cadre of Akt substrates and downstream biological processes, many of which were identical to those examined in prior studies, our data do not illustrate selective signaling in the absence of Ser⁴⁷⁴ phosphorylation but, rather, a global decrease in Akt2 activity. We cannot exclude that these differences between our data and previous reports are due to cell-specific effects, as each study was undertaken in a different cell line, or that the role of Ser⁴⁷⁴ phosphorylation differs for each Akt isoform. However, we speculate that these disparities are more likely reflective of other events downstream of mTORC2 conferring Akt substrate specificity rather than Ser⁴⁷⁴ phosphorylation itself. For example, mTORC2 has been shown to phosphorylate other sites on Akt, such as Thr⁴⁵⁰ on its turn motif (27, 28), and Ser⁴⁷⁷/Thr⁴⁷⁹ on its hydrophobic motif (26). This and other reported mechanisms of Akt substrate specificity (43) deserve further investigation largely because of their applicability to cancer treatments. Akt inhibitors have been tested in clinical trials; however, although they can successfully inhibit cancer growth, hyperglycemia often accompanies treatment (44, 45). This is not surprising considering Akt controls glucose metabolism. To design cancer therapeutics without these metabolic toxicities, it is of interest to therapeutically target specific actions of Akt rather than using pan-Akt inhibitors.

Akt2 undergoes differential SDS-PAGE migration upon insulin stimulation as a result of changes in its phosphorylation status (Fig. S3). FLAG-W80A, FLAG-W80A-T309A, and FLAG-W80A-S474A Akt2 did not differ in their gel migration in the basal state (Fig. 2B; indicated as Band 2 in Figs. 2E, 3A, and 4B). However, upon insulin stimulation, there was decreased migration of FLAG-W80A and FLAG-W80A-T309A Akt2 (indicated as Band 1 in Figs. 2E, 3A, and 4B) but not FLAG-W80A-S474A Akt2. These data suggest that Ser⁴⁷⁴ phosphorylation, but not Thr³⁰⁹ phosphorylation, is required for the insulin-stimulated band shift in Akt2. It is well established that some but not all phosphorylation events can result in a band shift (46). This can be attributed to the charge of the residues surrounding the phosphosite; only in the appropriate local environment does addition of a phosphate group decrease SDS binding and, consequently, migration (47). Nevertheless, this, coupled with the requirement for Ser⁴⁷⁴ phosphorylation for the band shift, leads us to propose that the insulin-stimulated band shift in Akt2 is caused by either Ser⁴⁷⁴ phosphorylation itself or other Ser⁴⁷⁴ phosphorylation–dependent phosphorylation sites. For example, we speculate that phosphorylation of Akt2 at Ser⁴⁷⁸ may be dependent on Ser⁴⁷⁴ phosphorylation as they are closely localized, increase with similar kinetics upon insulin stimulation (38), and interdependence between the corresponding sites on Akt1 has been reported (26). Elucidating whether this is the case may help dissect why Ser⁴⁷⁴ phosphorylation is required for maximal Akt2 kinase activity, particularly as phosphorylation events on Akt other than that at Thr³⁰⁹ and Ser⁴⁷⁴ are increasingly being shown to have substantial impact on Akt kinase activity (26, 48).

A major advantage of the W80A system used in this study is that it allowed investigation of Akt mutants in cells rather than in vitro. Our work in cells suggests that Akt can achieve half of its maximal kinase activity without Ser⁴⁷⁴ phosphorylation, which is in striking contrast to in vitro studies showing no or very little kinase activity in the absence of Ser⁴⁷⁴ phosphorylation (6, 17, 49). This difference is likely reflective of in vitro systems not reproducing complex cellular control points, comprising signal amplification, macromolecular crowding, interacting proteins, protein localization, and positive/negative feedback (36, 49, 50). As substantial Akt activation occurs without Ser⁴⁷⁴ phosphorylation in the cell, our data align with studies reporting that Ser⁴⁷⁴ phosphorylation poorly correlated with Akt activity (51); however, Ser⁴⁷⁴ phosphorylation is used in countless studies as a proxy of Akt activity. Rather, we suggest using phosphorylation of Akt at Thr³⁰⁹ or its substrates as more accurate indices of Akt activation (52).

Although we show a global defect in insulin-stimulated Akt2 activity upon loss of Ser⁴⁷⁴ phosphorylation, a recent study by Beg et al. (30), using the same W80A system in 3T3-L1 adipocytes, reported that insulin-stimulated GLUT4 translocation and mTORC1 activation were unaffected by loss of Akt2 Ser⁴⁷⁴ phosphorylation. There are differences between our studies that may explain these disparities. Most importantly, Beg et al. (30) used a dose of MK2206 (1 μm) they and others have shown to be insufficient to completely impair substrate phosphorylation by endogenous Akt (53). Having residual endogenous Akt activity will likely alter the behavior of the ectopic W80A Akt mutant, as Akt activity is extremely sensitive to temporal changes in its activation (54–56). Even a minute amount of residual endogenous Akt activity may skew signaling outcomes, as only a small percentage of the total Akt pool is required to achieve a maximal response for many downstream processes (36, 37). In contrast, we utilized 10 μm MK2206, as it was sufficient to block endogenous Akt activation, allowing specific analysis of ectopic W80A Akt2 (Figs. 1, D and E, and 2, E and F). Furthermore, to control for endogenous Akt inhibition, Beg et al. (30) used untransfected 3T3-L1 adipocytes. However, based on our studies, it is important to use adipocytes expressing FLAG-WT Akt2 to ensure effective comparison with adipocytes expressing FLAG-W80A Akt2. Nevertheless, Beg et al. (30) also reported that insulin-stimulated GLUT1 translocation requires Akt2 Ser⁴⁷⁴ phosphorylation, which we support (Fig. S4). These data suggest that further inquiry may reveal Akt substrates that control GLUT1 translocation and require Ser⁴⁷⁴-phosphorylated Akt for activation. However, an alternate explanation is that maximal Akt activity may be required to trigger insulin-stimulated GLUT1 translocation; this result may not reflect Akt substrate specificity per se but, rather, a difference in the dose–response relationship between Akt activation and various downstream processes.

The mechanism by which substantial Akt activity is possible in the cell without Ser⁴⁷⁴ phosphorylation remains elusive. Akt crystal structures suggest that HM binding to the N-lobe of the kinase domain is required for an active conformation and that Ser⁴⁷⁴ phosphorylation is essential for this interaction (18, 19). We speculate that this interaction can occur in the absence of Ser⁴⁷⁴ phosphorylation in cells; this weak interaction could be more favorable in cells than in vitro because of macromolecular crowding or stabilization by interacting proteins (50). Intriguingly, other AGC kinases, such as protein kinase A, lack a phosphosite within their HM, yet it retains the ability to interact with its kinase domain and induce activity (18, 19, 57). Thus, Ser⁴⁷⁴ phosphorylation may increase the affinity of the HM for the kinase domain but not be a prerequisite for the interaction or Akt activation. Alternatively, it has recently been found that the HM interacts with the PH–kinase domain linker to relieve PH domain–mediated autoinhibition and increase Akt activity (58). Whether this interaction can occur without Ser⁴⁷⁴ phosphorylation remains to be elucidated. Uncovering the mechanism by which substantial Akt kinase activity is possible without Ser⁴⁷⁴ phosphorylation is essential, as there is interest in developing therapeutic agents that inhibit Akt by targeting its hydrophobic motif (59).

Experimental procedures

Cloning

TagRFP-T-Akt2 has been described previously (34). TagRFP-T-Akt2 with the W80A mutation was generated using site-directed mutagenesis (60). For FLAG-Akt2 constructs, Akt2 was amplified by PCR from human cDNA with addition of attB sites at either end to be compatible with gateway cloning. Akt2 cDNA was inserted into pDONR221 (Thermo Fisher Scientific) using Gateway^TM BP Clonase^TM II Enzyme Mix (Invitrogen) and then into the retroviral vector pMIG-t-sapphire-puromycin (pMIG-tsap-puro) using Gateway^TM LR Clonase^TM II Enzyme Mix (Invitrogen). A FLAG tag was then added to the N terminus of Akt2 using Gibson assembly cloning (61). Site-directed mutagenesis (60) was used to create FLAG-tagged W80A, W80A-T309A, and W80A-S474A Akt2 plasmids. For FOXO1-mNeonGreen, the N-terminal region of FOXO1 (1–380) containing the nuclear localization and nuclear export sequences was codon-optimized and synthesized as a gene block (IDT Technologies). An mNeonGreen tag was placed at its C terminus, and this was cloned into the mammalian expression vector pcDNA3.1. The mutations S209A, H212A, and S215A were introduced to impair DNA binding and phosphorylation by STK4 as described previously (62). pHluorin-GLUT4-mRuby3 was generated from pHluorin-GLUT4-tdTomato (63) by replacing tdTomato with mammalian expression-optimized mRuby3 synthesized as a gene block (IDT Technologies). Primer and construct sequences are available upon request. All constructs were confirmed by Sanger sequencing.

Cell culture

3T3-L1 fibroblasts obtained from the Howard Green Laboratory (Harvard Medical School) were cultured in high glucose DMEM (Gibco by Life Technologies) supplemented with 10% (v/v) FBS (Gibco by Life Technologies), and 1× GlutaMAX (Gibco by Life Technologies) at 37 °C and 10% CO₂. All cell lines were mycoplasma-free. For the generation of 3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants, fibroblasts were transduced in the presence of 8 μg/ml hexadimethrine bromide (Polybrene, Sigma-Aldrich) with pMIG-tsap-puro (empty vector control) or pMIG-tsap-puro-FLAG-Akt2 retrovirus generated from Plat-E cells (Cell Biolabs, San Diego, CA) as described previously (64). Transduced cells were then selected using 2 μg/ml puromycin. Cells were differentiated into adipocytes as described previously (65), and used for experiments 7–12 days after initiation of differentiation. To minimize differences between cells expressing each Akt2 mutant, retroviral transduction was performed simultaneously using the same population of 3T3-L1 fibroblasts. Cells were infected with high virus titers so that there was minimal cell loss upon antibiotic selection. Then, to minimize differences in genetic drift, each cell population was subject to identical culturing conditions, used for experiments at the same passage, and not passaged more than 10 times. Macroscopically, we observed no morphological differences between cell populations at any stage throughout their differentiation into adipocytes. Furthermore, we performed the central experiments comprising Figs. 3, A and B, and 4, B–D, with three independently derived cell populations. Each achieved identical results; Ser⁴⁷⁴ phosphorylation was required for maximal Akt2 kinase activity.

Western blotting

3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were serum-starved with DMEM containing 1× GlutaMAX and 0.2% BSA (w/v) for 2 h. Cells were then exposed to 10 μm MK2206 (Selleckchem) or DMSO (Sigma-Aldrich) vehicle control for 5 min, followed by 1 nm insulin. We utilized 10 μm MK2206 to inhibit endogenous Akt, as we have shown previously that this dose is required to completely inhibit Akt activation (53). Cells were then placed on ice, washed with cold PBS, lysed with 1% (w/v) SDS in PBS containing protease inhibitors (Roche Applied Science) and phosphatase inhibitors (2 mm Na₃VO₄, 1 mm Na₄O₇P₂, and 10 mm NaF), and tip probe–sonicated. Lysates were centrifuged at 13,000 × g for 15 min at 4 °C. The lipid layer was removed, and protein content was quantified using the Pierce^TM BCA Protein Assay Kit (Thermo Scientific). 10 μg of lysate was then resolved by SDS-PAGE, transferred to PVDF membranes, and immunoblotted as described previously (39). All primary antibodies were from Cell Signaling Technology, but 14-3-3 was from Santa Cruz Biotechnology. Densitometry analysis was performed using ImageStudioLite version 5.2.5 (LI-COR). Band intensities were normalized to the loading control (14-3-3). Statistical tests were performed using GraphPad Prism version 7.0.

Live-cell TIRF microscopy

Matrigel-coated coverslips were prepared as described previously (34). To assess insulin-stimulated GLUT4 translocation to the plasma membrane, 3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were electroporated 7 days post-differentiation with 6 μg of pHluorin-GLUT4-mRuby3. To assess insulin-stimulated Akt2 recruitment to the plasma membrane, 3T3-L1 adipocytes were electroporated 6–8 days post-differentiation with 6 μg of WT or W80A TagRFP-T-Akt2. Then cells were placed onto the Matrigel-coated coverslips as described previously (34). 24–48 h after electroporation, adipocytes were serum-starved for 2 h and then incubated at 37 °C with Krebs–Ringer–phosphate–HEPES (KRPH) buffer (0.6 mm Na₂HPO₄, 0.4 mm NaH₂PO₄, 120 mm NaCl, 6 mm KCl, 1 mm CaCl₂, 1.2 mm MgSO₄, and 12.5 mm HEPES (pH 7.4)) supplemented with 10 mm glucose, 1× minimum Eagle's medium amino acids (Gibco by Life Technologies), 1× GlutaMAX, and 0.2% (w/v) BSA. The cells were then treated with 10 μm MK2206 and/or 1 nm insulin using a custom-made perfusion system. Images were acquired using the Nikon Ti-Lapps H-TIRF module angled to image ∼90 nm into cells. Temperature and humidity were maintained using an Okolab cage incubator and temperature control. The change in pixel intensity over time was determined for each individual cell using Fiji (66). The average pixel intensity over time for each cell was normalized to its average intensity over the basal period.

[³H]2-deoxyglucose uptake assay

3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were serum-starved for 2 h, washed with PBS, and incubated at 37 °C with KRPH buffer. Cells were stimulated with 10 μm MK2206 for 5 min or DMSO vehicle control, followed by 1 nm insulin. To determine background radiation, 25 μm cytochalasin B (Sigma-Aldrich) as a glucose transport inhibitor was added to a well of cells following 10-min insulin stimulation. After 15 min of insulin stimulation, 50 μm unlabeled 2-deoxyglucose and 0.25 μCi/ml [³H]2-deoxyglucose (Perkin-Elmer Life Sciences) was added to all cells for 5 min. Glucose uptake was terminated by placing the plate on ice, washing with cold PBS, and lysing the cells with 1% (w/v) Triton X-100 in PBS. Lysates were added to a scintillation vial with 3 ml of Ultima Gold^TM XR liquid scintillation mixture (Perkin-Elmer Life Sciences). The radioactivity of each sample in disintegrations per minute was determined using a Tri-Carb liquid scintillation counter (Perkin-Elmer Life Sciences). The protein content of each sample was assessed using the Pierce BCA Protein Assay Kit (Thermo Scientific). Assays were performed in duplicate, and the average of the duplicate was considered one biological replicate. The disintegrations per minute for each condition were converted to nanomoles of 2-deoxyglucose, and data were normalized to protein levels. The value for cytochalasin B–treated cells was subtracted from all other values as a background control. Statistical tests were performed using GraphPad Prism version 7.0.

[³H]leucine incorporation assay for protein synthesis

3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were serum-starved in leucine-free DMEM supplemented with 0.2% BSA and 20 mm HEPES (pH 7.4) for 2 h. Cells were then treated with 10 μm MK2206 for 5 min, followed by 5 μCi/ml [³H]leucine (Perkin-Elmer Life Sciences) and 1 nm insulin for 1 h. To determine background radiation, 10 μm cycloheximide, a translation inhibitor, was added 30 min before [³H]leucine to a single well. Leucine incorporation was terminated by placing the plate on ice, washing with cold PBS, and then incubating cells with cold 10% (v/v) TCA for 10 min to lyse cells and precipitate protein. Pellets were washed three times in cold 10% (v/v) TCA to remove free [³H]leucine. Pellets were resuspended in 50 mm NaOH with 1% (w/v) Triton X-100 in PBS at 65 °C for 30 min and then added to a scintillation vial with 3 ml of Ultima Gold^TM XR liquid scintillation mixture (Perkin-Elmer Life Sciences). The radioactivity of each sample in disintegrations per minute was determined using a Tri-Carb liquid scintillation counter (Perkin-Elmer Life Sciences). The protein content of each sample was assessed using the Pierce BCA Protein Assay Kit (Thermo Scientific). Assays were performed in triplicate, and the average of the triplicate was considered one biological replicate. The disintegrations per minute for each condition were converted to nanomoles leucine, and data were normalized to protein levels. The value for cycloheximide-treated cells was subtracted from all other values to account for background radiation. Statistical tests were performed using GraphPad Prism version 7.0.

Live-cell spinning-disk confocal microscopy

Matrigel-coated coverslips were prepared as described previously (34). To assess insulin-stimulated FOXO nuclear exclusion, 3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were electroporated 6 days post-differentiation with 6 μg of FOXO1(1–380; S209A, H212A, S215A)-mNeonGreen. Then cells were placed onto the Matrigel-coated coverslips as described previously (34). 24 h after electroporation, adipocytes were serum-starved for 2 h and then stained with Hoechst 33342 for 10 min. Adipocytes were then incubated at 37 °C with KRPH buffer supplemented with 10 mm glucose, 1× minimum Eagle's medium amino acids, 1× GlutaMAX, and 0.2% (w/v) BSA. The cells were then treated with 10 μm MK2206 and 1 nm insulin using a custom-made perfusion system. Images were acquired on a Nikon Ti inverted confocal microscope with an Andor Diskovery spinning-disk system and an Andor Ixon 888 EmCCD camera (Andor). Temperature and humidity were maintained using an Okolab cage incubator and temperature control. For each individual cell, the median pixel intensity of the cytosol was divided by the median pixel intensity of the nucleus using Fiji (66).

λ protein phosphatase assay

3T3-L1 adipocytes were serum-starved for 2 h and then treated with 100 nm insulin for 10 min. Cells were placed on ice, washed with cold PBS, and harvested in 1% (w/v) Triton X-100 in PBS. Cells were homogenized by passing through a 22-gauge needle six times and a 27-gauge needle six times prior to centrifugation at 12,000 × g for 15 min. 100 μg of the supernatant was treated with 1 mm MnCl₂ (New England Biolabs), 1× NEBuffer for protein metallophosphatases (New England Biolabs), and either 800 units of λ protein phosphatase (New England Biolabs) or phosphatase inhibitors (10 mm Na₃VO₄, 5 mm Na₄O₇P₂, or 50 mm NaF) at 30 °C for 15 min. Reactions were stopped with 2% (w/v) SDS and then resolved by SDS-PAGE.

Subcellular fractionation

3T3-L1 adipocytes stably expressing FLAG-Akt2 mutants were serum-starved for 2 h and then exposed to 10 μm MK2206 for 5 min, followed by 0.5 nm insulin for 20 min. Cells were placed on ice, washed with cold PBS, and harvested in cold HES buffer (20 mm HEPES, 1 mm EDTA, and 250 mm sucrose (pH 7.4)) containing phosphatase and protease inhibitors. All subsequent steps were carried out at 4 °C as described previously (67). Briefly, cells were homogenized using a Dounce tissue grinder prior to centrifugation at 500 × g for 10 min. The supernatant was centrifuged at 13,550 × g for 12 min to pellet the plasma membrane and mitochondria/nuclei. This pellet was resuspended in HES buffer and again centrifuged at 13,550 × g for 12 min. The pellet was then resuspended in HES buffer, layered over high-sucrose HES buffer (20 mm HEPES, 1 mm EDTA, and 1.12 m sucrose (pH 7.4)), and centrifuged at 111,160 × g for 60 min in a swing-out rotor. The PM fraction was collected at the interface between the sucrose layers and pelleted by centrifugation at 235,000 × g for 75 min. The PM pellet was resuspended in HES buffer containing phosphatase and protease inhibitors. Protein concentrations for each fraction were determined using the Pierce BCA Protein Assay Kit (Thermo Scientific).

Author contributions

A. L. K. data curation; A. L. K. formal analysis; A. L. K. validation; A. L. K., K. C. C., D. M. N., and A. Z. investigation; A. L. K. visualization; A. L. K., K. C. C., and D. M. N. methodology; A. L. K. writing-original draft; A. L. K., J. R. K., D. J. F., J. G. B., and D. E. J. writing-review and editing; K. C. C., J. R. K., D. J. F., J. G. B., and D. E. J. supervision; K. C. C. project administration; J. R. K., D. J. F., J. G. B., and D. E. J. funding acquisition; D. J. F., J. G. B., and D. E. J. conceptualization; D. E. J. resources.