1. Biochemistry and Chemical Biology
Download icon

The structural determinants of PH domain-mediated regulation of Akt revealed by segmental labeling

  1. Nam Chu
  2. Thibault Viennet
  3. Hwan Bae
  4. Antonieta Salguero
  5. Andras Boeszoermenyi
  6. Haribabu Arthanari  Is a corresponding author
  7. Philip A Cole  Is a corresponding author
  1. Division of Genetics, Department of Medicine, Brigham and Women's Hospital, United States
  2. Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, United States
  3. Department of Pharmacology and Molecular Sciences, Johns Hopkins School of Medicine, United States
  4. Department of Cancer Biology, Dana-Farber Cancer Institute, United States
Research Article
  • Cited 0
  • Views 794
  • Annotations
Cite this article as: eLife 2020;9:e59151 doi: 10.7554/eLife.59151

Abstract

Akt is a critical protein kinase that governs cancer cell growth and metabolism. Akt appears to be autoinhibited by an intramolecular interaction between its N-terminal pleckstrin homology (PH) domain and kinase domain, which is relieved by C-tail phosphorylation, but the precise molecular mechanisms remain elusive. Here, we use a combination of protein semisynthesis, NMR, and enzymological analysis to characterize structural features of the PH domain in its autoinhibited and activated states. We find that Akt autoinhibition depends on the length/flexibility of the PH-kinase linker. We identify a role for a dynamic short segment in the PH domain that appears to regulate autoinhibition and PDK1-catalyzed phosphorylation of Thr308 in the activation loop. We determine that Akt allosteric inhibitor MK2206 drives distinct PH domain structural changes compared to baseline autoinhibited Akt. These results highlight how the conformational plasticity of Akt governs the delicate control of its catalytic properties.

Introduction

Akt1 (termed Akt in the present work) is a Ser/Thr kinase that is a critical node in cell signaling and connects growth factor receptor activation to cell growth and metabolic regulation (Manning and Toker, 2017; Liao and Hung, 2010; Fruman et al., 2017). The Akt subfamily of kinases includes closely related paralogs Akt1-3 and are members of the larger AGC kinase family, comprised of about 60 members of the kinome (Leroux et al., 2018; Pearce et al., 2010). Akt is a 480 amino acid protein that contains an N-terminal PH (pleckstrin homology) domain followed by a catalytic (kinase) domain and culminates in a regulatory C-terminal disordered tail (C-tail) (Figure 1A). Our current understanding of Akt activation is that, in response to growth factor stimulation, phosphatidyl inositol diphosphate (PIP2) is converted to PIP3 by PI3-kinase and PIP3 recruits Akt to the plasma membrane via its PH domain (Manning and Toker, 2017; Haeusler et al., 2018). Upon recruitment, Akt is phosphorylated by both mTORC2 kinase at Ser473 and PDK1 kinase at Thr308 and this dual phosphorylation leads to activation of Akt (Sarbassov, 2005; Alessi et al., 1996). Additionally, Akt can also be phosphorylated at Ser473 by DNA-PK upon activation of DNA damage response in the nucleus (Bozulic et al., 2008). In an alternative pathway of Akt activation, Akt may be phosphorylated on Ser477 and Thr479 by Cdk2/cyclinA instead of Ser473 and this may happen in the cell nucleus (Liu et al., 2014). Excluding Thr308, the three other phosphorylations (Ser473, Ser477 and Thr479) are located in the disordered C-terminal tail.

Akt domain architecture, two distinct activation mechanisms mediated by C-tail phosphorylations, and the current model for drug-induced autoinhibited form.

(A) Schematic representation of Akt domains with phosphorylation sites of interest highlighted as red balls. (B) Cartoon model depicted for two distinct Akt activation mechanisms induced by phospho-Ser473 and dual phospho-Ser477/Thr479 (Chu et al., 2018). Without C-tail phosphorylations, Akt remains in an inactive autoinhibited state where the PH and kinase domains interact intramolecularly. The mTORC2-mediated Ser473 phosphorylation activates Akt by inducing an interaction between the C-tail and the PH-kinase domain linker, dislodging the PH domain from the kinase domain. Alternatively, the dual Cdk2/cyclinA-mediated pSer477/pThr479 is presumed to activate Akt by interaction with the activation loop and PH domain. Note that, although C-terminal phosphorylation is shown preceding activation loop Thr308 phosphorylation by PDK1, it is uncertain in normal cell signaling whether C-terminal phosphorylation or Thr308 phosphorylation comes first in Akt activation or if it is in random order. (C) Crystal structure of Akt (aa 1–443) bound to allosteric inhibitor compound VIII (left, PDB: 3O96, [Wu et al., 2010]) and cartoon model illustrating the current model of the allosteric drug-induced autoinhibited form of Akt with the non-phosphorylated C-tail.

Over the past two decades, several aspects of the regulatory mechanisms of Akt have been described. An autoinhibitory mechanism involving an intramolecular interaction between its N-terminal PH domain and C-terminal kinase domain is generally accepted (Bellacosa et al., 1998; Calleja et al., 2009; Calleja et al., 2007; Cole et al., 2019; Figure 1B). The structural basis for this autoinhibitory conformation has been proposed based on high resolution X-ray crystal structures of near full-length Akt in complex with small molecule catalytic inhibitors such as compound VIII (Wu et al., 2010; Lapierre et al., 2016; Ashwell et al., 2012; Figure 1C). A mechanistically related allosteric inhibitor of compound VIII, MK2206, has been widely used in cancer therapeutic clinical trials aimed to block Akt action (Hirai et al., 2010; Sangai et al., 2012; Yap et al., 2011; Nitulescu et al., 2016).

While phosphorylation at Thr308 is important for kinase activity in the context of the isolated Akt kinase domain, C-terminal phosphorylation shows significant impact only in the context of full-length Akt (Chu et al., 2018; Cole et al., 2019; Alessi et al., 1996; Calleja et al., 2009; Yang et al., 2002a; Yang et al., 2002b). Recent structural studies suggest that phosphorylation at Ser473 concurrently interacts with a PH-kinase domain linker basic patch, especially the sidechain of Arg144, and the kinase N-lobe via Gln218 (Chu et al., 2018; Figure 1B). A model based on these structural interactions postulates that these interactions with phosphorylated Ser473 (pSer473) help to displace the PH domain from the kinase domain thereby stimulating catalytic activity. Phosphorylation of Ser477/Thr479 is suggested to relieve autoinhibition of Akt in a distinct fashion, potentially by interaction with the activation loop and/or the PH domain (Chu et al., 2018; Figure 1B).

However, the roles and conformational status of the PH domain in the active and inactive states of Akt have been controversial. It has been suggested that PIP3 activates Akt not simply by membrane recruitment but by dislodging the PH domain from the kinase domain independent of promoting Akt phosphorylation by upstream kinases (Ebner et al., 2017; Lučić et al., 2018). In contrast to these studies, this allosteric mechanism of Akt activation by PIP3 was not confirmed by our group where neither soluble nor vesicle-embedded PIP3 stimulated Akt kinase activity (Chu et al., 2018). Moreover, the binding affinities of soluble PIP3 with pSer473 Akt vs. non-phospho C-tail Akt were essentially identical (Chu et al., 2018). In comparison, soluble PIP3 binding to pSer477/pThr479 was attenuated by about fourfold compared to pSer473 and non-C-terminally phosphorylated Akt (Chu et al., 2018). Remarkably, crystal structures of near full-length Akt in complex with compound VIII and related allosteric inhibitors reveal that the PIP3 binding pocket on the PH domain is occluded by its interactions with the kinase domain (Wu et al., 2010; Figure 1C).

X-ray crystal structures of near full-length Akt in the absence of allosteric inhibitors, as well as of active Akt including the PH domain have not yet been reported. Thus, our understanding of the structure and conformation of the PH domain in the context of the full-length Akt protein is incomplete. In this study, we have used a combination of enzymatic analysis of linker-altered Akts and solution NMR analysis of segmentally isotopically labeled semisynthetic Akt forms to characterize the structural features of the PH domain within the full-length protein in differentially phosphorylated states. These structural measurements combined with biochemical data highlight regions in the PH domain that appear to show distinct changes in the context of C-terminal phosphorylation and allosteric drug inhibition.

Results

Role of linker length and flexibility in autoinhibition and activation of Akt

Intramolecular interactions between the Akt PH and kinase domains are presumed to enforce autoinhibition of Akt's enzymatic activity. Interaction of pSer473 with the linker basic patch, particularly Arg144, appears to help relieve such autoinhibition. We have attempted to determine the role of linker length and flexibility that may contribute to how the pSer473-Arg144 interaction influences the relief of autoinhibition. A prior cellular study meant to investigate this possibility by inserting a flexible hexa-Gly segment in the middle of the PH-kinase linker showed reduced Akt signaling in response to growth factors. The results of these experiments were complicated to interpret, however, because hexa-Gly insertion led to reduced pSer473 levels (Chu et al., 2018).

To characterize the role of linker flexibility in more depth, we prepared the semisynthetic wt and hexa-Gly insertion Akt forms containing C-terminal phosphorylation using expressed protein ligation (Muir et al., 1998; Figure 2—figure supplement 1). In this procedure, aa1-459 of wt Akt and the hexa-Gly insertion segments fused to an intein were expressed in insect cells along with PDK1 and phosphatase inhibitor okadaic acid to ensure complete Thr308 phosphorylation. The C-terminal thioester fragments were ligated to N-Cys C-tail peptides (aa460-480) containing pSer473, pSer477/pThr479, or no C-tail phosphorylation and the semisynthetic proteins further purified to greater than 80%. Near stoichiometric Thr308 phosphorylation of these Akt forms was confirmed by western blot. It should be noted that Thr450 phosphorylation which is necessary for Akt stability occurs in insect cells concurrently with Akt production (Chu et al., 2018).

We then analyzed the catalytic properties of the wild type (wt) and hexaGly mutant proteins (Figure 2A). It should be noted that the apparent wt catalytic efficiencies (kcat/Km values) were relatively similar to those reported previously for wt Akt in various states of phosphorylation generated in this fashion (within ~3-fold, [Chu et al., 2018]). Some variation in kinase activities between different preparations can be explained by possible post-translational modification differences in the recombinant protein portion of the semisynthetic Akts which may occur in insect cells. To minimize the effects of catalytic differences among insect cell preparations, where possible we compared Akt forms with distinct C-terminal phosphorylation status prepared from the same batch of recombinant Akt thioester. Moreover, the small contamination (~10%) of slightly truncated Akt in these preps (Figure 2—figure supplement 1) is not expected to alter significantly the kcat/Km values measured (see Materials and methods).

Figure 2 with 1 supplement see all
The PH-kinase domain linker length/flexibility affects Akt activation.

(A) Schematic illustration for radiometric kinase assay using a biotinylated GSK3 peptide as Akt substrate and steady-state kinetic plots for v/[E] versus [ATP] with 20 µM GSK3 peptide for semisynthetic pThr308 Akt proteins WT (blue) versus the linker 6xGly insertion (red) with non-p C-tail (left), pSer473 (middle) and dual-pSer477/pThr479 (right), n = 2. (B) Enzyme catalytic efficiencies (apparent kcat/Km) obtained from (A) for each semisynthetic Akt protein, WT (left) and 6xGly insertion (right), two independent repeats were performed for each assay, S.D. shown.

There was a 3.4-fold reduction in kcat/Km of hexaGly pSer473 Akt relative to that of wt pSer473 Akt (Figure 2B). In contrast, the non-phosphorylated C-tail hexaGly Akt was ~8 fold more active compared with the corresponding wt Akt. These results support a role for linker length or flexibility as a contributing factor to activation of Akt orchestrated by pSer473-basic linker patch interaction. This also indicates that autoinhibition is alleviated in the non-C-tail phosphorylated state if this linker is too flexible.

In contrast to the results with pSer473, the activity of the dual C-terminal phosphorylated (pSer477/pThr479) Akt showed only a minor difference (<2 fold) when comparing the hexaGly linker insertion vs. wt Akt (Figure 2B). These findings suggest that linker length or flexibility is not an important factor in pSer477/pThr479-mediated Akt activation, consistent with an alternative mode of regulation for this dual phosphorylated C-tail form.

Segmental isotopic labeling of Akt by protein semisynthesis

We embarked on an effort to further characterize the domain structural interactions between the PH domain and the kinase domain within Akt using solution NMR. However, given the size of full-length Akt (56 kDa), the poor expression yields and low activity of E. coli-expressed kinase domain and the difficulty of achieving site-specific phosphorylation, we chose to develop a new three-piece expressed protein ligation-based segmental labeling protocol (Liu et al., 2009; Xu et al., 1999; Figure 3A and see Materials and methods).

Figure 3 with 1 supplement see all
Semisynthesis strategy to generate segmentally 15N, 13C, 2H isotopically labeled Akt containing distinct C-tail phospho forms.

(A) Schematic representation of the semisynthesis strategy. (B) Alignment of Akt1 (aa 118–127) and Akt2 (aa 119–129) linkers with Akt1 Ser122 (ligation site, mutated to Cys in our study) and Akt2 Cys124 highlighted (red). (C–D) SDS-PAGE analyses of EPL reaction between S122C, Δ121, pThr308, pSer473 Akt fragment and isotopically labeled PH thioester fragment (C); and segmentally isotopically labeled pThr308, pSer473 full-length Akt purified from (C) using size exclusion chromatography (D), lanes 2–4: purified full-length Akt diluted 10-fold from stock, loaded volumes are 2.5, 5 and 10 µl, respectively, lanes 5–9: BSA standards 0.25, 0.5, 1, 2, 4 µg, dashed line: deletion of unrelated samples. M: protein markers (kDa).

The PH domain (aa 1–121) is expressed in E. coli and isotopically labeled with (13C), 15N and 2H to ensure optimal relaxation properties (Figure 3—figure supplement 1A). The linker-kinase domain segment (aa 122–459) was expressed in Sf9 insect cells using the baculovirus system and phosphorylated at Thr308 in vitro using recombinant GST-PDK1, achieving site-selective phosphorylation and baseline activity. The kinase domain was not isotopically labeled, thus greatly reducing spectral crowding in context of the full-length Akt. The C-terminal tails (aa 460–480) were prepared by solid phase peptide synthesis in order to efficiently generate the desired phospho-forms of Akts: pSer473, pSer477/pThr479 or the non-phosphorylated C-tail variants (Figure 2—figure supplement 1A–C).

To obtain segmental labeling of full-length Akt, both the PH and kinase domains were fused to GyrA inteins, allowing C-terminal thioester formation. The C-terminal thioester form of the Akt kinase domain was then reacted with the requisite C-terminal peptides possessing an N-terminal Cys residue. Then, a N-Cys on the kinase domain was exposed after TEV protease cleavage, and reacted with the PH domain C-terminal thioester. Note that the natural Ser122 of Akt1 was replaced by Cys at this position – as needed for chemoselective ligation during the assembly of the semisynthetic protein – but the corresponding native residue in Akt2 is a Cys, indicating that this substitution should be well-tolerated (Figure 3b). This sequential three-piece ligation strategy facilitated the production of milligram quantities of highly pure (monomeric), full-length Akt, which was triply labeled in the PH domain, with phosphorylation at Thr308 and specific and stoichiometric C-terminal phosphorylation states (Figure 3C and D, and Figure 3—figure supplement 1B and C). Kinase assays with segmentally labeled pSer473 Akt revealed that its enzymatic activity was within twofold of that of the unlabeled semisynthetic enzyme prepared in a two-piece ligation of aa1-459 with aa460-480 as described previously (Chu et al., 2018; Figure 3—figure supplement 1D).

NMR analysis of segmentally labeled Akt forms

Given the limited solubility and stability of these full-length Akt forms under suitable NMR conditions, assignments were obtained for the isolated PH domain (see Materials and methods) and were subsequently transferred to the corresponding peaks in spectra of the segmentally labeled Akt proteins. About 60% of backbone hydrogen and nitrogen chemical shifts of residues in the segmentally labeled Akt forms could be unambiguously assigned using this procedure.

The interaction of the PH domain with the various phospho variants of the C-tail, in the context of the segmentally labeled full-length Akt, was initially assessed by 15N-1H-HQSC spectra using combined chemical shift perturbations (CSP) and peak broadening (Figure 4 and Figure 5). It is clear from a comparison of spectra of the isolated PH domain with that of the PH domain the context of full-length Akt (Figure 5—figure supplement 1) that the PH domain undergoes extensive interactions with the kinase domain and/or the C-tail, and potential conformational rearrangements. The CSP and line broadening are spread throughout the PH domain, including regions that interact with the PIP3 analog inositol hexaphosphate (IP6) (Figure 5—figure supplement 2B and E). Note that the CSPs observed on the PH domain are on average three-fold higher in the context of full-length Akt than those induced on the free PH domain by IP6 or the addition of the kinase domain in trans (Figure 5—figure supplement 1 and Figure 5—figure supplement 2A and D). These findings highlight the importance of the intramolecular PH-kinase domain interactions in intact Akt that can be captured only in the context of the full-length protein.

NMR reveals differences in the PH domain dependent on C-tail phospho states.

(A) Overlay of 15N-1H HSQC spectra of the PH domain in the context of full-length semisynthetic Akt with non-phosphorylated C-tail (blue) and pSer473 (red). Select residue-specific assignments are shown. (B) Overlay of 15N-1H HSQC spectra of the PH domain in the context of full-length semisynthetic Akt with non-phosphorylated C-tail (blue) and pSer477/pThr479 (green). Select residue-specific assignments are shown. (C) Expanded spectra around the peaks assigned to Ile6, Gln43, Asp44, Thr65 and Thr101. Overlay of all three spectra, same color coding. Arrows indicate chemical shift perturbations. Note that contour levels have been plotted lower for the non-p-C Gln43 and pSer477/pThr479 Asp44 peaks.

Figure 5 with 3 supplements see all
The PH domain of Akt interacts with the kinase domain differently depending on C-tail phosphorylations.

(A) Cartoon representation of secondary structure elements (rectangle for β-strands, zigzag for α-helices) in Akt PH domain. Color coding represents regions with distinct binding modes to the kinase domain. Star indicates that the α-helix is present only when Akt is bound to IP4. (B–C) Combined chemical shift perturbations derived from spectra in Figure 4 and plotted along the PH domain primary sequence for pSer473 (B) and pSer477/pThr479 (C) referenced to non-p-C. Dashed red line corresponds to the standard deviation to the mean, excluding outliers (higher than 3xStDev). Grey bars indicate peaks that disappeared from the spectrum, also indicating strong interaction. Red area highlights the short hinge primarily studied here. Negative bars (−0.05) indicate non-assigned residues, negative bars (−0.025) indicate residues which assignment could not be easily transferred or recovered in the context of full-length Akt. (D) Structure of allosteric drug inhibited Akt (PDB: 3O96, [Wu et al., 2010]) with PH domain as ribbon in the front and kinase domain as surface in the back (N-lobe in light blue, C-lobe in dark blue). Color coding of secondary structure elements in the PH domain corresponds to (A). Allosteric inhibitor MK2206 is displayed in white. Main Akt domains are labeled. (E–F) Same structure representation as in (D) with the most significantly affected residues (chemical shift perturbations higher than the standard deviation) colored in red in the PH domain in case of pSer473 (E) and pSer477/pThr479 (F). Non-affected residues are shown in light orange and non-assigned residues in grey. Representations rotated by 180° are shown. (G) Statistical bar and whisker plots of saturation transfer efficiencies from CST data (Figure 5—figure supplement 3) for each C-tail phospho-state, categorized and color coded according to secondary structure elements as in (A) and (D). A ratio of 0 indicates maximum saturation transfer efficiency (very tight interaction) whereas a ratio of 1 indicate no saturation transfer (no interaction).

Next we focused on comparing the spectral differences in the PH domains among the three C-tail phospho-states: pSer473, pSer477/pThr479, and non-phosphorylated C-tail (non-p-C) (Figure 4 and Figure 5). While the N-terminus (aa 1–41, β-strands 1 to 3) seems to show relatively minor differences in response to variations in the C-tail phospho-states, the region encompassing β-strands 4 and 5 (aa 72–90) show comparatively larger perturbations depending on which C-tail phospho-states are present. The effects of CSP and line broadening can be induced by a direct interaction or an allosteric change, both of which result in changes in the electronic environment of the observed proton. To distinguish between direct and indirect binding interactions between the kinase and PH domains among the different Akt forms, we employed cross saturation transfer (CST) NMR measurements, which exclusively provides information on the direct interaction interface (Wüthrich, 2000; Takahashi et al., 2000). Perdeuterated, 15N-labeled, PH domain was expressed and covalently linked to the protonated kinase domain (and the various phospho-variants) as discussed above. In a CST experiment, we selectively magnetize the protonated protein in the aliphatic region, in this case the kinase domain, and monitor the transfer of magnetization to the perdeuterated partner, the PH domain. Since CST experiments are performed in 70% 2H2O (30% H2O) solvent to minimize non-specific effects due to spin diffusion, the inherent sensitivity of the experiment, which is concentration limited, is further reduced. Hence, we considered the CST effect in a contiguous stretch of amino acids as a true positive. Upon saturation of the kinase domain, peak intensities of select resonances in the PH domain were effectively reduced by as much as 90% (Figure 5G and Figure 5—figure supplement 3), underscoring the direct intramolecular interaction between the PH and kinase domains among the various Akt phospho-forms. The results from CST the experiments along with those derived from the CSP analysis are discussed below.

Aligning with the CSP data, no striking differences, in a contiguous stretch of amino acids were observed in the regions encompassing β-strands 1–3 and 4–7 (Figure 5G and Figure 5—figure supplement 3). This indicates that the modes of activation of Akt may depend on subtle changes in the binding interface rather than on major differences in binding affinity. However, a small but distinct difference in the CST was observed for β-strands 4–7 across the three tested forms of Akt (Figure 5G). Specifically, the CST for β-strands 4–7 of pSer473 is lower than that of the non-phosphorylated and pSer477/pThr479 forms. This difference could be the result of a marginally weaker interaction of β-strands 4–7 of the PH domain with the kinase domain in the pSer473 form. This is consistent with the biochemical data.

Focusing on the strong CST effects, the C-terminal alpha-helix (aa 100–109) of the PH domain showed a significant CSP between the pSer473 and pSer477/pThr479 forms (Figure 4 and Figure 5B–E). There were no significant changes in the chemical shifts of the C-terminal alpha-helix PH domain between the non-p-C and pSer473. On the contrary, the spectrum of pS477/pThr479 shows marked CSPs compared to the non-p-C form on the residues corresponding to the C-terminal helix. Thus, we surmise that this PH domain C-terminal alpha-helix plays a specific role in the mode of activation of Akt by the dual pSer477/pThr479 modifications. Notably, results from the CST experiments reinforced this possibility since the peaks of residues corresponding to the C-terminal helix exhibit marked reduction in intensity (near-complete disappearance) in the pSer477/pThr479 form (Figure 5G and Figure 5—figure supplement 3).

Furthermore, residues Gln43 and Asp44 of the PH domain which are in an extended loop between β-strands 3 and 4 show either relatively large CSPs or marked peak broadening beyond detection in the phospho-C-tail forms relative to the non-phospho-C-tail form (Figure 4 and Figure 5B and C). Similarly, the CST behavior for Gln43 and Asp44 was closely linked the phospho-state of the C-tail (Figure 5G and Figure 5—figure supplement 3) with an apparent loosening of the interaction for pSer473 but an apparent tightening for pSer477/pThr479. We therefore posit that this region plays a role in governing Akt regulation and investigated this possibility below.

Role of a short hinge, aa 44–46, in Akt regulation

Prior X-ray crystal structures of the Akt PH domain have shown that aa 44–46 (DVD) located between β-strands 3 and 4 undergo a helix to loop transition between the inositol tetraphosphate-bound PH domain and the apo form (Thomas et al., 2002; Milburn et al., 2003; Meuillet, 2011; Carpten et al., 2007; Figure 5—figure supplement 2F and G). Given the fact that peaks assigned to Gln43 and Asp44 show either the highest CSP or extensive peak broadening as a function of C-tail phosphorylation status described above (Figure 4), we hypothesized that this conformational change contributes to Akt regulation in the intact protein. To test the hypothesis that this short hinge affects the capacity of Akt to bind PIP3, we recombinantly expressed a triple mutant – D44G/V45P/D46G – of the isolated PH domain aimed at disrupting the potential of this segment to form a stable helix with Gly and Pro residues that favor a random coil secondary structure (Pace and Scholtz, 1998). We then measured the affinity of wt and triple mutant PH domains to bind PIP3, using a fluorescence anisotropy assay; however, we did not detect any significant difference in the binding affinity (Figure 6—figure supplement 1A).

To characterize the effect of the short hinge further, we produced wt and triple mutant semisynthetic Akt variants using a standard expressed protein ligation strategy in which recombinant aa 1–459 thioester was generated in Sf9 insect cells along with co-overexpression with PDK1 to afford Thr308 phosphorylation. Using western blotting, we noted that the degree of phosphorylation at Thr308 was reduced compared to wt Akt prepared under the same conditions (Figure 6—figure supplement 1B and C). We thus treated partially Thr308 phosphorylated triple mutant (D44G/V45P/D46G) Akt1 aa 1–459 with PDK1 in an in vitro kinase reaction which appeared to complete the phosphorylation of Thr308 to a level matching that of wt Akt1 obtained simply from PDK1 co-expression (Figure 6—figure supplement 1E). Achieving similar pThr308 levels for wt and D44G/V45P/D46G Akt is necessary to determine how the PH hinge mutation affects catalysis in the same context.

We then ligated Akt aa 1–459 recombinant thioesters with synthetic peptides containing aa 460–480 monophosphorylated at Ser473, diphosphorylated at Ser477 and Thr479, or non-phosphorylated (Figure 6—figure supplement 1D). Kinase assays performed with the triple mutant pSer473 full-length Akt semisynthetic protein showed that its catalytic efficiency, kcat/Km, is reduced ~10 fold relative to wt pSer473 under these conditions (Figure 6A and B). In contrast, the pSer477/pThr479 activation was essentially unaffected by the PH domain triple mutation (DVD - > GPG). The differences in sensitivity of the DVD triple mutation to the two C-tail phosphorylated forms are consistent with the distinct CST behaviors noted above (Figure 5G and Figure 5—figure supplement 3).

Figure 6 with 1 supplement see all
A short hinge 44–46 aa in the PH domain governs Akt activation using a two-pronged approach.

(A) Steady-state kinetic plots v/[E] versus [ATP] with 20 µM GSK3 peptide for semisynthetic pThr308 Akt proteins WT (blue) versus D44G/V45P/D46G (GPG) mutant with non-P C-tail (left), pSer473 (middle) and pSer477/pThr479 (right), n = 2. (B) Catalytic efficiencies (apparent kcat/Km values) for each semisynthetic Akt phospho form of WT (left) and GPG mutant (right) obtained from kinase assays in (A), two independent repeats were performed for each assay, S.D. shown. (C) Cellular analysis of the effect of GPG mutant on Akt phosphorylation. Left, Akt and GAPDH antibodies western blot of cell lysate at different time points after stimulation, as indicated, Neg.: non-transfected cells stimulated with insulin/IGF-1; (-): transfected with DNA plasmid but not stimulated with insulin/IGF-1 (left). Right, quantification of Akt Thr308 phosphorylation level (blue for WT and green for GPG mutant) using ImageJ2 (n = 5, SEM shown, 0.01 < p < 0.05). (D) Time course kinase assay for PDK1-catalyzed Akt Thr308 phosphorylation (schematic illustration). Left, western blot of semisynthetic pSer473 Akt (WT or GPG) after incubation with PDK1 for different times, as indicated. Right, corresponding calculated half-time of completion obtained from two independent repeats and S.D. shown.

To examine the cellular significance of the D44G/V45P/D46G triple mutation, we transfected the full-length Akt triple mutant into an Akt1/Akt2 knockout HCT116 human colon cancer cell line. Cells were treated with insulin and IGF1 to stimulate the PI3K/Akt pathway that classically leads to phosphorylation of Thr308 as part of the activation mechanism. Interestingly, Thr308 phosphorylation was significantly impaired in the D44G/V45P/D46G triple mutant relative to transfected wt Akt (Figure 6C). These results are consistent with the insect cell expression experiments described above and suggest that the intramolecular PH domain-kinase domain interactions are altered in a way that interferes with PDK1-catalyzed phosphorylation of the activation loop.

To examine this in more depth, we performed an in vitro analysis of the kinetics of PDK1-mediated Thr308 phosphorylation of wt and GPG triple mutant pSer473 Akt. This revealed a ~ 2.5-fold reduction in Thr308 phosphorylation of the mutant protein, suggesting the reduced accessibility of Thr308 to PDK1 (Figure 6D).

Structural changes in Akt induced by allosteric inhibitors

Crystal structures of near full-length Akt1 in complex with various allosteric inhibitors have been widely accepted to represent the inactive conformation of Akt, because the PH and kinase domains are glued together (Wu et al., 2010; Calleja et al., 2009; Parikh et al., 2012; Lapierre et al., 2016; Ashwell et al., 2012). In these inhibitor-bound structures, the PIP3-binding pocket of the PH domain is obstructed which would be expected to reduce PIP3-binding affinity. Our recent work has revealed that PIP3-binding affinity is similar for the pSer473 and non-C-tail phosphorylated Akt forms. In this light, it has been difficult to reconcile the structural studies. To investigate this further, we measured PIP3 affinity to Akt in the context of either MK2206 or compound VIII allosteric Akt inhibitors (Figure 7, Figure 7—figure supplement 1A). These experiments have shown that the allosteric inhibitors dramatically weaken the binding of PIP3 to Akt, by more than 20-fold (Figure 7).

Figure 7 with 1 supplement see all
The allosteric drug-induced Akt inhibited form is distinct from its native autoinhibited form.

(A) Binding assays of phospholipid PIP3 with full-length Akt in presence of 20 µM MK2206. Chemical structure of MK2206 is shown on the right. The fluorescence anisotropy measurements (n = 2) were carried out and fit to quadratic binding isotherms, and Kd values shown ± S.D. (B) Cartoon model illustrated the distinct conformational structure of PH domain of allosteric drug-bound Akt when compared to that of natural autoinhibited form.

We then used NMR to assess the impact of MK2206 on the conformation of the PH domain with non-C-tail phosphorylated Akt, where the PH domain was segmentally labeled. Based on the 15N-1H HSQC analysis, there were marked chemical shift changes throughout the PH domain (Figure 8A,D). The amplitudes of these chemical shift perturbations were generally much larger than those observed when comparing different C-terminal phospho-forms of Akt. Notably, some of the most significant chemical shift perturbations were detected in the PIP3-binding interface (e.g. aa Trp11, Leu12) (Figure 8D). We also performed a CST experiment in the same manner as described above, in the presence of MK2206 (Figure 8B,E and F) and observed two principal effects mediated by the compound. The short hinge of aa 44–46 seems to separate further from the kinase domain and/or C-terminal tail while the C-terminal alpha-helix, in contrast, appears to interact more tightly with the kinase domain or C-tail. These inhibitor-induced changes likely result from a combination of altered structural and dynamic features relative to the baseline autoinhibited form of Akt. These could include a reorientation of the PH domain with respect to the kinase domain or C-tail (placing the C-terminal helix closer) as well as conformational and dynamic rearrangements within the PH domain itself. In a complementary approach, we performed a fluorescence trans binding assay of split Akt (separate PH and kinase domains) that MK2206 can enhance the binding affinity between the two domains in an intermolecular fashion (Figure 7—figure supplement 1B and C). This is consistent with a related study that showed that allosteric inhibitors indeed can stimulate kinase-PH domain interaction (Parikh et al., 2012; Weisner et al., 2015).

The allosteric drug-induced Akt inhibited form is distinct from its native autoinhibited form.

(A) Overlay of 15N-1H HSQC spectra of the PH domain in the context of full-length semisynthetic Akt with non-phosphorylated C-tail in the presence of the allosteric inhibitor MK2206 (red) or in its absence (blue). Select residue-specific assignments are shown. (B) Expanded 15N-1H HSQC cross saturation transfer (CST) spectra around the peaks assigned to Phe27, Thr65 and Ala102. Reference (unsaturated) spectrum is shown in blue and its saturated counterpart in red. (C) Cartoon representation of secondary structure elements (rectangle for β-strands, zigzag for α-helices) in Akt PH domain. Color coding represents regions with distinct binding modes to the kinase domain. Star indicates that the α-helix is present only when Akt is bound to IP4. (D) Combined chemical shift perturbations corresponding to spectra in (A) plotted along the primary sequence for the PH domain in the context of semisynthetic full-length Akt with non-p C-tail with MK2206, referenced to the control without drug. Dashed red line corresponds to the standard deviation to the mean, excluding outliers (higher than 3xStDev). Grey bars indicate peaks that disappeared from the spectrum, also indicating strong interaction. Red area highlights the short hinge primarily studied here. Negative bars (−0.05) indicate non-assigned residues, negative bars (−0.025) indicate residues which assignment could not be easily transferred or recovered in the context of full-length Akt. (E) Saturation transfer efficiency derived from spectra in (B) plotted as the ratio of peak intensities of saturated over unsaturated spectra against the PH domain primary sequence. A ratio of 0 indicates maximum saturation transfer efficiency (very tight interaction), whereas a ratio of 1 indicate no saturation transfer (no interaction). An indicative dashed red line has been drawn at 0.5. Red area highlights the short hinge primarily studied here. Negative ratios (−0.05) indicate non-assigned residues, negative ratios (−0.025) indicate residues that were not present in 70% 2H2O spectra. (F) Statistical bar and whisker plots of saturation transfer efficiencies from CST data (B,E), categorized and color coded according to secondary structure elements as in (C). (G) Structure of allosteric drug inhibited Akt (PDB: 3O96, [Wu et al., 2010]) with PH domain as ribbon in the front and kinase domain as surface in the back (N-lobe in light blue, C-lobe in dark blue). Significantly affected residues (chemical shift perturbations higher than the standard deviation) are colored in red. Non-affected residues are shown in light orange and non-assigned residues in grey. A representation rotated by 180° is shown.

Discussion

Our understanding of the molecular basis of Akt regulation by its PH domain have primarily derived from Akt crystal structures that either lack the PH domain or contain an allosteric inhibitor, although HD-exchange studies have also been performed on S473D Akt (Lučić et al., 2018). We note that prior studies have indicated that Asp473 is a poor mimic of pSer473 in kinase activity studies (Chu et al., 2018) suggesting that Asp473 may not relieve autoinhibition efficiently. Here, we look directly at the Akt PH domain in the context of full-length protein by solution NMR using a segmental labeling approach. These experiments show that specific regions in the PH domain show chemical shift perturbations in the presence of C-terminal phosphorylation that can stimulate Akt activity. Moreover, there are distinct changes in Akt that is phosphorylated on Ser473 versus Ser477/Thr479, consistent with differential interactions of the phospho-sites and the kinase domain and linker dependence on the phosphorylation state (Chu et al., 2018).

These studies highlight the value of segmental labeling using semisynthesis to make larger proteins dependent on insect cell expression amenable to NMR. The simplified spectra of the ~120 aa PH domain, prepared in E. coli, embedded in the 480 aa Akt protein, mostly produced in Sf9 cells, was straightforward to assign by correlation with the isolated PH domain. We suggest that such segmental labeling strategies be considered for solution phase structural studies of big proteins requiring eukaryotic expression that are now commonly considered out of reach by NMR.

We focused on one short segment in the PH domain based on the NMR studies which is the short hinge consisting of the three amino acid motif aa 44–46, DVD, henceforth referred to as the DVD motif. This hinge region was sharply affected by C-terminal phosphorylation. Prior crystal structures of the isolated PH domain have shown that this DVD motif transitions between alpha-helix and loop depending on whether the domains were complexed with the PIP3 mimic, IP4 (Thomas et al., 2002; Milburn et al., 2003; Askham et al., 2010; Meuillet, 2011). By mutating DVD to GPG to destabilize possible helix formation, we showed that C-terminal phosphorylation of Ser473 was poorly able to stimulate Akt kinase activity. This suggests that the normal dynamics (loop to helix transition) of this DVD motif is necessary for allowing the PH domain to dislodge from the kinase domain in response to pSer473. In contrast, activation of Akt by pSer477/pThr479 was insensitive to this mutation, despite changes in chemical shift in the hinge region (aa 43–44). These findings can be interpreted as consistent with a distinct mode of activation by these non-classical phosphorylation events (Chu et al., 2018; Liu et al., 2014). The GPG mutation in the isolated PH domain did not influence its ability to bind PIP3 indicating that the dynamics of this DVD segment is not intimately connected to the thermodynamics of PIP3 binding. Interestingly, our results show that the DVD motif can also impact PDK1's ability to phosphorylate Thr308 in Akt's activation loop. We surmise that the helical state of aa 44–46 in Akt may facilitate a catalytically competent interaction between PDK1 and Akt.

Our results also show how flexibility/length in the linker between the kinase domain and PH domain can modulate the ability of C-terminal phosphorylation to activate Akt. Insertion of a hexaGly linker led to reduced activation in the case of pSer473, but only minimally in the case of pSer477/pThr479. The results with the pSer473 form of Akt suggests that engagement of Arg144 by pSer473 may tug on the linker to weaken the PH domain-kinase domain interaction and that this effect is attenuated by increasing linker flexibility which could reduce linker tension (Chu et al., 2018). Prior cellular studies revealed that growth factor induced pSer473 phosphorylation is decreased in the hexaGly mutant, presumably because the C-tail phosphorylation is not as shielded from cellular phosphatases when the linker is less rigid. Notably, the baseline activity of non-C-tail-phosphorylated Akt is elevated in the hexaGly insertion mutant relative to the wt Akt enzyme. This can be understood to result from a floppier linker, enlarging the volume sampled by the PH domain, lowering its effective concentration relative to the kinase domain.

Our NMR analysis helps explain the paradox that crystal structures of allosteric inhibitors in complex with Akt show that the PH domain PIP3 pocket is obstructed but the non-phospho-C-tail form of Akt binds PIP3 potently. Indeed, we confirm here that in the presence of allosteric inhibitor, PIP3 affinity of Akt is markedly reduced. These results indicate that the conformation of the baseline non-phosphorylated-C-tail form of Akt is not identical to the allosteric-bound conformation. NMR data show dramatic chemical shift changes in the PH domain between the non-phosphorylated-C-tail and the inhibitor-bound form of Akt that likely reflect these differences. It is possible that the non-phosphorylated-C-tail Akt state is comprised of an ensemble of conformations, and one of the minor populations binds to allosteric inhibitors via conformational selection. Although NMR dynamics experiments, due to their very long measurement times are not feasible with our current experimental system, future alternative approaches may allow this to be investigated.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional information
Gene (human)Akt1Addgene and
DOI:10.1093/nar/gkh238
Addgene # 9021
Strain, strain background (E. coli)Rosetta 2(DE3) pLysSNovagenCat. No.: 71400Competent cells
Strain, strain background (E. coli)DH10BacInvitrogenCat. No.: 18297010Competent cells
Cell line (insect cell)Sf21Invitrogen11497–013
Cell line (insect cell)f9Invitrogen11496–015
Cell line (human)Akt1/2-/- HCT116DOI:10.1073/pnas.0914018107Colon cancer cell line; The Akt1-/-and Akt2-/-HCT116 colon cancer cell line was a gift from Dr. Bert Vogelstein (Johns Hopkins University) (Ericson et al., 2010). These cells were authenticated by western blot showing the absence of Akt and by the lack of signaling response to growth factors. They were also shown to be mycoplasma-free by PCR testing.
Antibodypan-Akt (11E7)
(Rabbit monoclonal)
Cell Signaling TechnologyCat. No.: 4685S, RRID:AB_10698888WB (1:1000 for cell-based assays; 1:20000 for activation assays)
AntibodyAkt phospho-Thr308 (D25E6)
(Rabbit monoclonal)
Cell Signaling TechnologyCat. No.: 13038S,
RRID:AB_2629447
WB (1:1000, for cell-based assays; 1:10000 for activation assays)
AntibodyAkt phospho-Ser473 [EP2109Y]
(Rabbit monoclonal)
AbCamCat. No.: ab81283,
RRID:AB_2224551
WB (1:1000)
AntibodyGAPDH (14C10) (Rabbit monoclonal)Cell Signaling TechnologyCat. No.: 2118S,
RRID:AB_561053
WB (1:5000)
AntibodyHRP conjugated, anti-Rabbit IgG (Goat monoclonal)Cell Signaling TechnologyCat. No.: 7074S, RRID:AB_2099233WB (1:5000)
Chemical compound, drugDeuterium oxide (D, 99.8%)Cambridge Isotope LaboratoriesDLM-4–99.8-1000
Chemical compound, drugAmmonium chloride (15N, 99%)Cambridge Isotope LaboratoriesNLM-467–1
Chemical compound, drugD-Glucose (U-13C6, 99%; 1,2,3,4,5,6,6-D7, 97–98%)Cambridge Isotope LaboratoriesCDLM-3813–2
Chemical compound, drugCeltone base powder (13C, 98%+; D, 97%+; 15N, 98%+)Cambridge Isotope LaboratoriesCGM-1030P-CDN-1
Chemical compound, drugD-Glucose (U-13C6, 99%)Cambridge Isotope LaboratoriesCLM-1396–2
Chemical compound, drug20 Fmoc-amino acids, Fmoc-Ser(HPO3Bzl)-OH, and Fmoc-Thr(HPO3Bzl)-OHP3Bio systems
Chemical compound, drugSulfo-Cy5-NHS esterLumiprobeCat. No.: 43320
Chemical compound, drug32P-ATPPerkin ElmerCat. No.: NEG002Z2-50UC
Chemical compound, drugPierce avidinThermo ScientificCat. No.: 21128
Chemical compound, drugRecombinant human insulinThermo ScientificCat. No.: 12585014
Chemical compound, drugHuman insulin-like growth factor 1 (hIGF-1)Cell Signaling TechnologyCat. No.: 8917SC
Software, algorithmImageJ2DOI:10.1186/s12859-017-1934-z
Software, algorithmGraphPad Prism version 8.2.1GraphPad
Software, algorithmNmrPipeDOI:10.1007/BF00197809
Software, algorithmCCPNmr Analysis version 2.4DOI:10.1002/prot.20449
Software, algorithmhmsISTDOI:10.1007/s10858-012-9611-z
Software, algorithmTopspin version 3.6Bruker

Peptide synthesis

Request a detailed protocol

Peptides corresponding to residues 460–480 of Akt1 (CVDSERRPHFPQFSYSASGTA) and kinase substrate peptide N-ε-biotin-lysine GSK3 (RSGRARTSSFAEPGGK) were synthesized according to an Fmoc-based solid phase strategy previously described (Chu et al., 2018). The peptides were purified using reverse-phase C18 HPLC (VYDAC) using a gradient of 30% to 50% (v/v) of acetonitrile mixed with water containing 0.05% trifluoroacetic acid at a flow rate of 10 mL/min for 40 min. Pure fractions (>95%) were identified by MALDI mass spectrometry, combined and concentrated on a rotovap and then lyophilized to dryness. Peptide concentrations were determined by amino acid analysis.

Expression of Akt proteins and GST-PDK1

Request a detailed protocol

The baculovirus-insect cell system was employed to express Akt-Mxeintein-CBD (CBD: chitin binding domain) constructs (Akt1 aa 2–459 and 122–459) and GST-PDK1 according to the procedures previously reported (Chu et al., 2018). In particular, the in vivo Akt Thr308 phosphorylation was obtained by co-expression of the Akt(2-459)-Mxeintein-CBD fusion construct with GST-PDK1 in Sf9 insect cells with M.O.I.s (multiplicity of infections) of 5.0 and 1.0 for the baculovirus containing Akt and GST-PDK1, respectively. After growing infected Sf9 cells for ~36 hr at 27°C, 25 nM of the phosphatase inhibitor okadaic acid (Cell signaling technology-CST) was added, and the cells permitted to grow for an additional 16 hr, and then harvested.

To obtain triply labeled 15N, 13C, 2H Akt PH domain, the pTXB1 plasmid containing Akt (aa 1–121)-MxeIntein-CBD was expressed in E. coli Rosetta (DE3)/pLysS (Invitrogen) following the established protocol (Gronenborn et al., 1991; Coote et al., 2018). Briefly, the E. coli cells were grown in 1 L of M9 minimal medium (6 g/L Na2HPO4 (Sigma if not stated otherwise), 3 g/L KH2PO4, 0.5 g/L NaCl, 0.25 g/L MgSO4, 11 mg/L CaCl2, 2 g/L deuterated-13C-glucose (Cambridge Isotopes), 1 g/L 15NH4Cl (Cambridge Isotopes), 100 mg/L ampicillin and 20 mg/L chloramphenicol) in D2O, and was further supplemented with trace elements (50 mg/L EDTA, 8 mg/L FeCl3, 0.1 mg/L CuCl2, 0.1 mg/L CoCl2, 0.1 mg/L H3BO3, and 0.02 mg/L MnCl2) and the vitamins biotin (0.5 mg/L) and thiamin (0.5 mg/L) in shaker flasks at 37°C until OD600 = 0.5, then 1 mL of 0.5 M IPTG was added to induce the expression and the cultures were further incubated for 24 hr at 16 °C. Cells were pelleted and stored in −80 °C freezer for the next steps.

Semisynthesis of segmentally isotopically labeled Akt

Request a detailed protocol

To produce full-length Akt containing segmentally triply labeled 15N, 13C, 2H PH domain and the C-tail site-specific phosphorylations at either Ser473, Ser477/Thr479 or no phosphorylations on these residues, a sequential expressed protein ligation (EPL) strategy involving three peptide/protein pieces was developed. After resuspending the E. coli cells expressing isotopically labeled PH domain-MxeIntein-CBD in lysis buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 1 mM EDTA, 10% Glycerol, 0.1% Triton X-100, one protease inhibitor tablet (Roche)), the cells were lysed by french press and the mixture was clarified by centrifugation at 17,500 g for 40 min at 4°C. The unlabeled insect cells expressing Akt (aa122-459-MxeIntein-CBD) were suspended in lysis buffer and lysed in a 40 ml Dounce homogenizer on ice, and the mixture was clarified as described above for the PH domain. The insect cell expressed protein was also passed through fibrous cellulose to remove chitinase as described previously (Bolduc et al., 2013). Next, both N-Tags-TEV-S122C-Akt kinase domain (aa 122–459)-MxeIntein-CBD (N-tags: N-terminal Flag-HA-6xHis) and triply labeled Akt PH domain (aa 1–121)-MxeIntein-CBD proteins were purified by affinity chromatography from the cell lysates using chitin beads. After loading onto chitin beads, elution of the protein C-terminal thioester forms of both the Akt kinase and PH domains via intein cleavage using MESNA (sodium mercaptoethylsulfonate) according to established protocols (Chu et al., 2018). The obtained N-Tags-TEV-S122C-Akt kinase domain thioester was phosphorylated at Thr308 in vitro using recombinant GST-PDK1 (Chu et al., 2018), and then ligated with the synthetic N-Cys containing C-terminal Akt peptides (aa 460–480) containing variable phosphorylations in the first ligation buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 1 mM TCEP, 100 mM MESNA, 10 mM EDTA, 10% glycerol, 1 mM PMSF) for 5 hr at room temperature and then maintained overnight at 4°C. The ligation product N-Tags-TEV-S122C-Akt aa 122–480 fragment was purified by size exclusion chromatography (SEC) on a Superdex 75 10/300 GL column (GE Healthcare) with the second ligation buffer (100 mM HEPES pH 7.8, 500 mM NaCl, 100 mM MESNA, 0.5 mM IP6 (phytic acid sodium salt, Sigma), 1 mM TCEP, 1 mM PMSF). The purified fractions (>90%) were combined, concentrated and mixed with triply labeled Akt PH domain thioester at the concentration of 5 mg/mL, 300 µg of TEV protease was added to initialize the second EPL at 4 °C for 65 hr. The ligation yields were assessed by coomassie-stained SDS-PAGE and typically shown to be greater than 90%. The resulting full-length Akt (aa 1–480) was separated from the excess of Akt PH domain thioester by using SEC purification on a Superdex 200 10/300 GL column (GE Healthcare) and buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 2 mM beta-mercaptoethanol, 0.5 mM PMSF, 10% glycerol), concentrated, fast frozen and stored at −80°C.

Semisynthetic Akt proteins (wt, hexaGly, DVD mutant) used for kinase assays were generated by the previously described two-piece EPL strategy (co-overexpression with GST-PDK1 and okadaic acid treatment during insect cell expression) with the intein-mediated Akt (aa 1–459)-thioester fragment was reacted with the synthetic N-Cys containing C-terminal Akt peptides (aa 460–480) in the first ligation buffer as described above to afford semisynthetic full-length Akt proteins. The semisynthetic Akt proteins were purified by SEC with the Akt purification and storage buffer. The small contamination (~10%) of slightly truncated Akt proteins in these likely results from incomplete ligation or slight proteolytic cleavage of the ca. 40 aa N-terminal tag. As the C-terminal tail is critical in supporting Akt catalytic activity and the absence or presence of the N-terminal tag does not appear to affect catalytic activity (Chu et al., 2018), we do not think this contamination will have a big impact on the measured catalytic efficiencies.

NMR experiments

Request a detailed protocol

All NMR experiments were conducted on a Bruker spectrometer operating at 750 MHz, equipped with a TCI cryoprobe and z-shielded gradients. All data were processed using Topspin (Bruker) or NmrPipe (Delaglio et al., 1995) and analyzed with CCPNmr (Vranken et al., 2005). Samples of approximately 600 μM 2H, 13C, 15N-labeled Akt PH domain (aa 1–121) in 50 mM HEPES pH 6.5, 500 mM NaCl, 2 mM IP6, 0.5 mM TCEP and 5% v/v 2H2O were prepared for backbone resonance assignment (Auguin et al., 2004). Standard triple resonance backbone experiments: HSQC, HNCA, HNCOCA, HNCO, HNCACO, HNCACB were used. For all the backbone triple resonance experiments, 10% Nyquist grid, in the indirect dimension, was sampled using Poisson-Gap sampling. The resulting non-uniformly sample spectrum was reconstructed using the hmsIST (Hyberts et al., 2012). The NMR experiments were performed at 20°C. An additional HNCA spectrum was measured at 13°C, as well as a temperature titration (from 13°C to 20°C), allowed us to leverage the previously published assignments from Auguin et al., 2004. 66% of non-proline backbone resonances could be assigned.

HSQC spectra of 100–120 μM segmentally isotopically labeled full-length Akt were measured in 50 mM HEPES pH 6.5, 50 mM NaCl, 0.5 mM TCEP (NMR buffer) and 5% v/v 2H2O using TROSY-HSQC (Pervushin et al., 1997). Recycle delays of 1 s were used and 768 transients (and 256 increments in the indirect dimension) were accumulated, leading to measurement times of 54 hr. Chemical shifts were calibrated using internal water calibration in NmrPipe. Combined chemical shift perturbations and associated standard deviations were calculated according to the protocol from Schumann et al., 2007.

Cross saturation transfer experiments (Shimada, 2005) were measured with on (0 ppm) and off-resonance saturation (−30 ppm) in an interleaved fashion using 100–120 μM samples in NMR buffer. The samples were prepared in 70% v/v 2H2O to minimize spin-diffusion through solvent. Saturation time was set to 3 s and recycle delays to 1 s. Composite WURST pulses (25 milliseconds in duration) were used for saturation. Number of transients was approximately 400 and number of increments to 232 leading to measurement time of about 100 hr.

Kinase activity assays

Request a detailed protocol

The steady-state kinetic parameters of semisynthetic Akt proteins were determined by phosphorylating N-ε-biotin-lysine GSK3 peptide (RSGRARTSSFAEPGGK) in radiometric reactions as described elsewhere (Qiu et al., 2009; Chu et al., 2018), with minor modification. The biotinylated GSK3 peptide substrate concentration was kept at 20 µM (5–10-fold above the peptide substrate Km values for Akt protein phosphorylated on Thr308 [Chu et al., 2018]) in all the assays. The buffer used contains 50 mM HEPES pH 7.5, 10 mM MgCl2, 1 mM EGTA, 2 mM DTT, 1 mM sodium orthovanadate, 0.5 mg/mL BSA (kinase reaction buffer),~0.42 µCi γ-32P-ATP, and varying amounts of ATP (0–2 mM). All the Akt proteins used in this assay contain N-terminal affinity tags Flag, HA and 6xHis unless otherwise noted. The kinase reactions were performed at 30°C for 10 min and quenched by adding 50 mM EDTA, and then 100 µg of Avidin (Pierce) was added to each sample and incubated for 20 min at room temperature. Samples were transferred to centrifugal 10 kDa MWCO filtration units (Nanosep 10K, PALL) and washed five times with 120 μl of washing buffer (0.5 M sodium phosphate, 0.5 M NaCl, pH 8.5). The filtration units were placed in 5 mL scintillation fluid and counted by Beckman liquid scintillation counter (Beckman LS6500). The kcat/Km (ATP) values were calculated using the standard Michaelis-Menten equation using a non-linear fit with Prism (TM) and can be considered ‘apparent’ values performed at a fixed, although most likely near-saturating peptide substrate concentration (Chu et al., 2018).

Fluorescence anisotropy measurements

Request a detailed protocol

The binding affinity of Akt PH domain with phospholipid PIP3 was determined using fluorescence anisotropy. Varying amounts of Akt PH domain were mixed with 50 nM fluorescein-labeled soluble PIP3 (Cayman Chemical) in binding buffer (50 mM HEPES pH 7.5, 2 mM DTT, 0.05 mg/ml ovalbumin) and incubated at room temperature for 30 min. Fluorescent anisotropy spectra were recorded by Multi-Mode Microplate Reader (Biotek Instruments) at 23°C with three different replicates. The Kd values were obtained by fitting the data to quadratic binding equation as described before (Seamon et al., 2015; Weiser et al., 2017; Chu et al., 2018).

Microscale thermophoresis (MST) analysis

Request a detailed protocol

For use in MST analysis, N-terminally Cy5 labeled, PH-deleted S122C Akt (aa 122–480) was prepared by pretreating Sulfo-Cy5-NHS ester (Lumiprobe) with MESNA to efficiently convert NHS ester into thioester that can selectively react with N-terminal Cysteine 122 as described in Dempsey et al., 2018. The binding affinity of Cy5 labeled, S122C Akt (aa 122–480) and Akt PH domain (aa 1–121) in a binding buffer containing 50 mM HEPES pH 7.5, 150 mM NaCl, 5% (v/v) glycerol, 0.05% (v/v) Triton X-100, 0.1 mg/mL ovalbumin, 5 mM DTT was carried out by MST using MONOLITH NT.115 (NanoTemper). The Cy5-labeled Akt (aa 122–480) (20 nM) was mixed 1:1 with different amounts of Akt PH domain in a two-fold dilution series from 19.8 nM to 650,000 nM for the measurement at 23°C. For the binding assays with the presence of Akt allosteric inhibitor MK2206, 20 nM Cy5-labeled Akt (aa 122–480) was pre-incubated with 40 µM MK2206 on ice for 30 min and then mixed 1:1 with Akt PH domain as described above. Each binding assay was repeated twice.

PDK1 phosphorylation of Akt assays

Request a detailed protocol

The PDK1 phosphorylation of Akt assays were carried out following the protocol previously described (Chu et al., 2018). Briefly, 1 μM Akt was mixed with 10 nM GST-PDk1 in the activation buffer (50 mM HEPES, pH 7.5, 2 mM DTT, 10 mM MgCl2) and incubated at 30°C, the reaction was triggered by adding 1 mM ATP. 10 μl of the reaction was collected at the indicated time points and quenched by the addition of 4xSDS-loading buffer. The SDS samples were loaded on SDS-PAGE gel for western blot analysis with anti-pT308 or pan-Akt primary antibodies.

Western blots

Request a detailed protocol

After transferring protein from SDS-PAGE gels to nitrocellulose membranes using an iBlot (Thermo Fisher) system, the membranes were blocked with 5% (w/v) BSA in TBS-T buffer at room temperature for 30 min. Membranes were incubated with anti-Akt (pan or phospho) primary antibodies (CST) at a 1:10,000 dilution with 5% (w/v) BSA in TBS-T buffer overnight at 4°C, washed three times of 10 min with TBS-T buffer, and incubated with secondary HRP-linked antibody (CST) for 1 hr at room temperature and following with three washing times. Membranes were developed with Amersham ECL Western blotting detection reagents (GE Healthcare) and imaged by a GeneSys (G:BOX, SynGene) imaging system.

Mammalian cell signaling assays

Request a detailed protocol

The Akt1-/- and Akt2-/- HCT116 colon cancer cell line was a gift from Dr. Bert Vogelstein (Johns Hopkins University) (Ericson et al., 2010). These cells were authenticated by western blot showing the absence of Akt and by the lack of signaling response to growth factors. They were also shown to be mycoplasma-free by PCR testing. Cells were cultured in McCoy’s 5A (Invitrogen) supplemented with 10% (v/v) FBS (Sigma) and 1% (v/v) penicillin/streptomycin (Gibco) at 37°C and 5% CO2. When the cell reached ~70% confluence in six-well plates, the cells were transfected with 1 μg of pcDNA3.1-Flag-HA-Akt plasmids complexed with 2 μL Lipofectamine 3000 (Invitrogen) and 2 μL P3000 reagent (Invitrogen) in Opti-MEM medium (Gibco) for 24 hr at 37°C and 5% CO2. When indicated, the cells were rinsed twice by PBS and serum-starved for 18 hr in McCoy’s 5A with 0.5% FBS and 1% penicillin/streptomycin, and stimulated with 100 ng/mL of insulin (Thermo Fisher Scientific) and 60 ng/mL human IGF-1 (CST) for variable times (5, 10, 20, and 40 min) at 37°C and 5% CO2. The cells were lysed by adding 180 μL RIPA buffer (CST) containing 1x complete protease inhibitor tablet and 1 mM PMSF, and gently shook for 30 min at 4 °C. 50 μg of total protein (BCA assay) was loaded on SDS-PAGE gels. Membrane transfer and Western blotting was carried out as described above with 1:1000 dilution for primary antibodies: Akt1, phospho Akt1.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
    Physiological regulation of akt activity and stability
    1. Y Liao
    2. MC Hung
    (2010)
    American Journal of Translational Research 2:19–42.
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54

Decision letter

  1. John Kuriyan
    Senior and Reviewing Editor; University of California, Berkeley, United States
  2. Kevin N Dalby
    Reviewer; The University of Texas at Austin, United States
  3. Neel Shah
    Reviewer

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Thank you for submitting your article "The Structural Determinants of PH Domain-Mediated Regulation of Akt Revealed by Segmental Labeling" for consideration by eLife. Your article has been evaluated favorably by three peer reviewers, and John Kuriyan has overseen the evaluation as the Reviewing and Senior Editor. The following individuals involved in the review of your submission have agreed to reveal their identity: Kevin N. Dalby (Reviewer #1); Neel Shah (Reviewer #3).

The reviewers have discussed the reviews with one another, and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation.

The manuscript by Cole, Arthanari, and co-workers aims to enhance our understanding of activation and autoinhibition of the protein kinase Akt1 (AKT). The authors use a combination of protein semisynthesis, enzymological assays, and NMR spectroscopy to identify structural changes that distinguish variously autoinhibited and activated states of AKT.

Three particularly notable conclusions stand out from this work. First, the authors lay out an elegant route to making segmentally-isotopically labeled and site-specifically phosphorylated samples of a multi-domain protein that has mostly eluded NMR studies, thus far. Second, the NMR experiments with different phospho-forms of AKT, which could only have been obtained through their synthetic methods, substantiate the hypothesis that distinct phosphorylation events activate AKT via different mechanisms. Finally, their NMR data show that the auto-inhibited state of AKT observed in crystal structures is almost certainly a drug-induced conformation, and is unlikely to be the predominant structure in solution in the absence of an allosteric inhibitor. The study provides good evidence that the conformational dynamics of the linker (or "tension", but note that reviewers object to this word) between the PH domain of AKT and the kinase domain plays a role in mediating the activation of AKT by phosphorylation of Ser473. Moreover, it suggests this is not the case for the phosphorylation of Ser477/Thr479. The authors identify a small region within the PH domain that adopts a helical state and mediates the phosphorylation of Thr308 by PDK1 in intact cells.

The work presented here provides new methods to study kinase structure/function, and the results are likely to guide future biological investigations, and drug discovery efforts focused on AKT.

The review process did perceive weaknesses in the manuscript. Some aspects of the work may be considered incremental, as it focuses on linker length (which was partly examined in previous work from this group) and a single loop region in the PH domain. Given the tools at this group's disposal, it seems the data in this paper are underutilized, and in some cases, the experimental design seems ill-conceived (see comment #5 below). There is a concern that, as presented, the authors provide interesting observations, but the data are not interpreted in a manner that yields a definitive advance in our mechanistic understanding of AKT regulation.

Despite these weaknesses, the reviewers agreed that the impressive demonstration of segmental labeling to probe the effects of phosphorylation and to investigate PH domain conformation are important advances. The finding that the drug-induced conformation of Akt may not represent the physiological conformation is also important. The paper should become acceptable for publication in eLife once it is revised to take into account the comments below. Note that under the present circumstances we are only asking for textual revisions, but if additional data that support the claims are available, they could be included.

Comments to address:

1) Work by Lucic et al. https://doi.org/10.1073/pnas.1716109115 contains important contributions to understanding AKT regulation and yet is not described or referenced in the submitted manuscript. Please include it in the revised manuscript.

2) The authors suggest that the putative regulatory interaction between pSer473 and R144 "exerts tension on the PH-kinase domain linker, thereby pulling on the PH domain and severing its contacts…" The use of the word "tension" throughout is misleading as the authors have not measured force in any of their experiments. Instead, it seems more appropriate to think about the pSer473/R144 interaction as promoting a shift in the conformational ensemble of the full-length protein that leads to destabilization of the PH/kinase interface.

3) The authors claim the results of increasing linker length (addition of the hexaGly insert) "support a role for linker tension," but really linker length has been probed. A simple explanation may be that the increased linker length and flexibility in the region adjacent to R144 might create an increased entropic cost in forming the pSer473/R144 contact. There seems to be an assumption in these experiments that the pSer473/R144 interaction is unchanged upon the insertion of hexaGly.

4) The NMR data, as presented in Figure 4 are difficult to see. Perhaps the spectra could be expanded to show key spectral changes. More importantly, the NMR changes are never mapped on the PH domain structure. Figure 4F simply shows regions of secondary structure, but the data itself (Figure 4D and E) could be significantly more informative if mapped onto the PH domain structure. This is true for all of the HSQC data throughout the paper.

5) What are the phosphorylation states at T308 for all of the protein samples generated via semisynthesis? It appears that Thr308 is phosphorylated in the protein samples prepared for the NMR work. This is puzzling as the authors are probing the transition from autoinhibited to activated AKT via C-tail phosphorylation. So it seems that as long as the activation loop is phosphorylated they won't capture the fully autoinhibited form. Moreover, the scheme in Figure 1B indicates that C-tail phosphorylation occurs before activation loop phosphorylation providing another reason to study C-tail phosphorylation effects in the absence of activation loop phosphorylation. Please clarify and comment on this issue. Can the authors provide MS data confirming the identities (and purity) of all semisynthetic products generated in this study? If additional data supporting the analysis are available, it might be worth including in the manuscript.

6) The structural distinction between the natural auto-inhibited state and the drug-induced state is interesting but poorly defined here. A potentially important finding is that the AKT crystal structures containing small molecules that “staple” the PH domain to the kinase domain may not be revealing the native autoinhibitory interface. The NMR mapping in this paper could provide insight into the PH/kinase interface in the absence of drug, but the authors stop short of using their data in this way. Is there some reason that the data have not been pushed to provide a more definitive understanding of the PH-kinase interaction? It would be illustrative and helpful to see the residues with significant chemical shift perturbations in Figure 6 highlighted on the structure of the drug-bound autoinhibited state. A discussion of how the extended interface upon drug binding differs from the native autoinhibited state is warranted. What can be inferred from the cross saturation transfer experiments? Was such an experiment done in the presence of the allosteric inhibitor? This experiment isn't necessary, but a general discussion of structural models that can be inferred from the data in this manuscript would strengthen the manuscript.

7) Cross-saturation transfer NMR data are also underutilized. Examining the quantitative differences between non-p-C and pS475 should provide more precise insight into how serine phosphorylation might be affecting the conformational preference of the PH domain. It's difficult to see from Figure 4—figure supplement 3 how the cross-saturation transfer effect is stronger for non-phosphorylated versus tail-phosphorylated samples, as stated in the text. Could this be quantified in some way – perhaps showing the mean peak intensity ratios for each sample or a distribution of these ratios across all samples?

8) The heavy focus on the loop following b3 (residues Gln43 and Asp44) seems driven (according to the text) by more dramatic chemical shift perturbation and more extensive line broadening in the phosphorylated forms compared to non-p-C. This seems a stretch as there are certainly other regions that show equally or nearly the same extent of spectral change (it's possible that the HSQC data itself makes the differences more evident, but expanded spectral data showing individual HSQC peaks are not provided).

9) The remainder of the paper focuses on a mutation of the native DVD sequence (residues 44-46) to GPG in an effort to "favor a random coil secondary structure in this region." The authors do not characterize this mutant to determine whether the mutation successfully shifted secondary structure preference of the PH domain but instead put the mutation directly into their full-length semi-synthetic AKT. Despite the lack of characterization of the mutant, this is the one place in this paper where the authors see a dramatic result; Figure 5A shows that the mutation of the DVD motif causes a large decrease in the steady-state kinetics of AKT in its pSer473 form compared to wild type. The other phosphorylated form of AKT (pSer477/pThr497) is unaffected by the mutation in the PH domain. Clearly understanding the mechanistic basis for this difference would be an advance in this field but the manuscript as presented does not sufficiently explain the observations.

10) In the Discussion, the authors interpret the failure of pSer473 to stimulate AKT activity as being due to changes in the "normal dynamics" of the DVD motif. Dynamics can be directly measured by NMR to support this conclusion, but this is surprisingly not included in the manuscript. The observation that activation loop phosphorylation (at Thr308) is also diminished upon mutation of the DVD motif to GPG might suggest that the mutation in the PH domain enhances the autoinhibitory contact between PH and kinase domains and/or alters the accessibility of the activation loop to PDK-1. Unfortunately, as presented, the manuscript provides little to advance our understanding of the precise role of this region of the PH domain in AKT regulation.

11) In the final part of this work the authors explore the effects of drug binding on PIP3 binding as well as PH domain NMR chemical shifts. The data are not interpreted in a satisfactory manner; the authors state that "Notably, some of the most significant chemical shift perturbations were detected in the PIP3-binding interface…" but there are significant changes elsewhere in the PH domain as well. There is also no attempt to take into account the likely significant spectral changes that will arise from the drug itself.

12) Finally, the last sentence of the paper suggests conformational heterogeneity might be present in the non-phosphorylated C-tail form of AKT. This is very logical and raises the question of whether the authors see any evidence for the existence of multiple states in solution for any of the segmentally labeled forms they have studied in this submitted paper or the previous 2018 publication?

13) Some of the kinetic parameters for the WT semi-synthetic AKT proteins shown in Figure 2 seem to differ a little from the parameters reported earlier in Chu et al., 2018. For example, the kcat/Km for A2 (full-length AKT phosphorylated on T308) in Figure 2 of Chu et al. is 28-fold smaller than the value reported in the current manuscript. In the present manuscript, the concentration range of ATP used is 0-2000 µM, whereas the Km values reported for A2, A1, and A6 (Figure 2 of Chu et al.) are 200, 5100, and 3700 µM, respectively. Perhaps the authors could clarify the apparent discrepancy.

14) Perhaps mention that the reported kcat/Km's are apparent values as the second substrate (the peptide in this case) is not present at a saturating concentration.

15) It would be helpful in Figure 1 if the various states were labeled as "active" or "inactive." In Figure 1B, the ribbon diagram of the PH-kinase unit is drawn upside. In this and other figures (e.g., 4F), it would be helpful if structural components and parts were clearly labeled on the figures.

16) Can the authors comment on the purity of the samples used in Figure 2? Based on the figure supplement SDS-PAGE analysis, there appears to be a ~10-20% impurity. Can this be identified by MS analysis of the full-length protein to confirm that it isn't a species that will interfere with the analysis of enzyme kinetics?

https://doi.org/10.7554/eLife.59151.sa1

Author response

The work presented here provides new methods to study kinase structure/function, and the results are likely to guide future biological investigations, and drug discovery efforts focused on AKT.

The review process did perceive weaknesses in the manuscript. Some aspects of the work may be considered incremental, as it focuses on linker length (which was partly examined in previous work from this group) and a single loop region in the PH domain. Given the tools at this group's disposal, it seems the data in this paper are underutilized, and in some cases, the experimental design seems ill-conceived (see comment #5 below). There is a concern that, as presented, the authors provide interesting observations, but the data are not interpreted in a manner that yields a definitive advance in our mechanistic understanding of AKT regulation.

Despite these weaknesses, the reviewers agreed that the impressive demonstration of segmental labeling to probe the effects of phosphorylation and to investigate PH domain conformation are important advances. The finding that the drug-induced conformation of Akt may not represent the physiological conformation is also important. The paper should become acceptable for publication in eLife once it is revised to take into account the comments below. Note that under the present circumstances we are only asking for textual revisions, but if additional data that support the claims are available, they could be included.

We agree that the strengths of our manuscript are the combination of the technical novelty as well as the mechanistic insights in Akt regulation and inhibition. We readily concede that our studies have not fully illuminated the complete set of conformational changes and altered intramolecular interactions in the various Akt states. A fundamental limitation of our work is the absence of NMR analysis of the kinase domain and C-terminus which are not isotopically labelled. This partly stems from requirement that the kinase domain should be expressed in insect cells to obtain a functional and well-behaved protein. For NMR studies of the kinase domain demands deuteration, which is challenging in eukaryotic expression systems. Another challenge that we faced in this investigation is the complexity and size of the Akt forms as well as their limited stabilities and solubilities for NMR analysis. Despite these impediments, we believe that we have gained important new information about how Akt is regulated by its PH domain in various phosphorylation states and in the presence of allosteric inhibitor. In addition, we posit that these studies will propel additional efforts by our lab and others in future studies to generate an even deeper understanding of this fascinating enzyme.

Comments to address:

1) Work by Lucic et al. https://doi.org/10.1073/pnas.1716109115 contains important contributions to understanding AKT regulation and yet is not described or referenced in the submitted manuscript. Please include it in the revised manuscript.

We thank the reviewers for pointing this out. While this work was related to the Ebner et al., 2017 paper, we agree that it is important to cite this study and briefly discuss it. We have now done so in the Introduction and the Discussion of the revised manuscript, noting its use of HD exchange analysis, but pointing out its reliance on the phosphomimic S473D rather than pSer473 that we believe limits its impact on mechanistic interpretations.

2) The authors suggest that the putative regulatory interaction between pSer473 and R144 "exerts tension on the PH-kinase domain linker, thereby pulling on the PH domain and severing its contacts…" The use of the word "tension" throughout is misleading as the authors have not measured force in any of their experiments. Instead, it seems more appropriate to think about the pSer473/R144 interaction as promoting a shift in the conformational ensemble of the full-length protein that leads to destabilization of the PH/kinase interface.

We think that this is a fair criticism and have dialed back the description of tension as the likely basis for the hexaGly insertion. We now mention alterations in tension as a possible basis for the catalytic changes along with changes in length and flexibility conferred by hexaGly insertion in the revised manuscript.

3) The authors claim the results of increasing linker length (addition of the hexaGly insert) "support a role for linker tension," but really linker length has been probed. A simple explanation may be that the increased linker length and flexibility in the region adjacent to R144 might create an increased entropic cost in forming the pSer473/R144 contact. There seems to be an assumption in these experiments that the pSer473/R144 interaction is unchanged upon the insertion of hexaGly.

We fully agree with these astute comments and as noted above, have revised the manuscript accordingly to emphasize linker length and flexibility rather than tension.

4) The NMR data, as presented in Figure 4 are difficult to see. Perhaps the spectra could be expanded to show key spectral changes. More importantly, the NMR changes are never mapped on the PH domain structure. Figure 4F simply shows regions of secondary structure, but the data itself (Figure 4D and E) could be significantly more informative if mapped onto the PH domain structure. This is true for all of the HSQC data throughout the paper.

We agree with the reviewers that it is difficult to appreciate the changes in the HSQC spectra of the original manuscript. To remedy this, we have created two new figures in the revised manuscript. Figure 4 shows an enlarged overlay of the HSQCs that we believe are easier to evaluate. Furthermore, Figure 4C shows a few selected peaks in a “zoomed in” manner. Figure 5 now includes representations of the most significant chemical shift perturbations highlighted on the structure of the MK2206-inhibited Akt relative to the signals of the pSer473 and pSer477/pThr479 Akts (Figure 5D and E). Likewise, Figure 8 now focuses on the NMR data for the effect of MK2206 and includes expanded peaks (Figure 8B) and a representation of the chemical shift perturbations projected on the three-dimensional structure (Figure 8G).

5) What are the phosphorylation states at T308 for all of the protein samples generated via semisynthesis? It appears that Thr308 is phosphorylated in the protein samples prepared for the NMR work. This is puzzling as the authors are probing the transition from autoinhibited to activated AKT via C-tail phosphorylation. So it seems that as long as the activation loop is phosphorylated they won't capture the fully autoinhibited form. Moreover, the scheme in Figure 1B indicates that C-tail phosphorylation occurs before activation loop phosphorylation providing another reason to study C-tail phosphorylation effects in the absence of activation loop phosphorylation. Please clarify and comment on this issue. Can the authors provide MS data confirming the identities (and purity) of all semisynthetic products generated in this study? If additional data supporting the analysis are available, it might be worth including in the manuscript.

In the production of purified Akts analyzed by NMR or enzymatically, the Thr308 has been phosphorylated at near stoichiometric levels by PDK1 either by co-expression or in vitro kinase treatment in all cases but one. This exception was for the study of the DVD to GPG triple mutant that was prepared specifically for kinetic analysis of PDK1 treatment of the full-length semisynthetic Akt in the DVD/GPG context. We performed this experiment since the HCT116 transfection experiments with DVD/GPG Akt unexpectedly revealed diminished Thr308 phosphorylation in response to growth factor. We have clarified this status in the revised manuscript.

In all of the other Akts including the DVD/GPG Akt that was prepared for determining the kcat/Km values of its own kinase activity, Thr308 was phosphorylated in this study. The reason that we elected to do this is based on previous work from our team (Chu et al., 2018) which showed that, in the absence of the PH domain, C-terminal phosphorylation is inconsequential for achieving full kinase activity. In contrast, Thr308 phosphorylation still matters with kinase domain only. In this prior work, both Thr308 and Ser473 phosphorylation in the context of full-length Akt provide about equally important and synergistic impacts on Akt activity. If either one is missing, the kinase activity is similar to having no phosphorylation at both Thr308 and Ser473, although Thr308 phosphorylation confers a lower Km of peptide substrate whereas Ser473 phosphorylation does not.

With respect to the cartoon in Figure 1B showing C-terminal phosphorylation preceding Thr308 phosphorylation, this is by no means established and different labs have concluded an ordered or random process about whether Thr308 or pSer473 comes first. We have added a disclaimer to the figure legend to clarify this point.

It is an interesting question raised by the reviewers regarding how Thr308 phosphorylation would impact the conformation of Akt and agree that this could also be investigated by NMR. Based on our previous measurements that a non-phosphorylated Thr308 form of Akt bound PIP3 about 2.5-fold more tightly compared with pThr308 Akt, there are likely at least subtle differences in the PH domain of the autoinhibited states +/-pThr308. We hope to analyze this issue further using NMR in future studies. We note, though, that the enhanced of affinity of Akt for PIP3 with non-phosphorylated Thr308 Akt is the opposite of what is observed with Akt in complex with allosteric inhibitor and this underscores the distinct Akt conformations that occur with MK2206 vs. MK2206-free Akt.

Regarding protein purity, we have relied on coomassie-stained SDSPAGE rather than mass spectrometry since the early stages of this project. In some unpublished mass spec analysis from our lab on Akt from several years ago when we first started working with the enzyme, we saw evidence for multiple PTMs including additional phosphorylations in the insect expressed component of the semisynthetic protein that was produced. We do not know the stoichiometries of these additional phosphorylations and have not mapped the sites. We do not have a simple way to remove the extra phosphorylations since treatment with alkaline phosphatase would hydrolyze the critical pThr450 that stabilizes Akt. So we have lived with this PTM heterogeneity, and we concede that this can complicate the interpretation of the structural and enzymatic studies. In the revised manuscript, we now allude to this limitation in the Results section, but also point out that within a set of experiments to assess different C-tail modification effects, we use the same batches of recombinant core kinase or PH-kinase to ensure apples to apples comparisons.

6) The structural distinction between the natural auto-inhibited state and the drug-induced state is interesting but poorly defined here. A potentially important finding is that the AKT crystal structures containing small molecules that “staple” the PH domain to the kinase domain may not be revealing the native autoinhibitory interface. The NMR mapping in this paper could provide insight into the PH/kinase interface in the absence of drug, but the authors stop short of using their data in this way. Is there some reason that the data have not been pushed to provide a more definitive understanding of the PH-kinase interaction? It would be illustrative and helpful to see the residues with significant chemical shift perturbations in Figure 6 highlighted on the structure of the drug-bound autoinhibited state. A discussion of how the extended interface upon drug binding differs from the native autoinhibited state is warranted. What can be inferred from the cross saturation transfer experiments? Was such an experiment done in the presence of the allosteric inhibitor? This experiment isn't necessary, but a general discussion of structural models that can be inferred from the data in this manuscript would strengthen the manuscript.

As explained above, Figure 8 now focuses on the NMR data for the effect of MK2206 and includes expanded peaks (Figure 8B) and a representation of the chemical shift perturbations in the context of the Akt structure (Figure 8G). We also performed CST experiment for the non-phosphorylated full-length Akt in the presence of MK2206 (Figure 8B, expanded peaks from CST spectra). One of the challenges we faced with the CST experiments is the sensitivity of the measurement. In addition to concentration limited, the CST experiments were conducted in 70% D2O to minimize, water mediated spin diffusion, which further reduced the intensity of the weaker peaks. In response to the reviewer’s comment, we mitigated this problem by considering CST data from groups of neighboring amino acids and we present the CST data in a corroborating manner (Figure 8E and F, CST data as bar plot and statistical bar and whisker plot, respectively).

Interestingly, none of the most affected residues in this context directly contact MK2206 in the crystal structure (this may be partly due to lack of assignment for some residues). From the chemical shift perturbation and CST data, we can infer two main phenomena: (i) the short hinge associated with aa 44-46 is “released” (CSPs are smaller and CST weaker in the presence of MK2206) and (ii) the C-terminal helix is interacts more tightly with the kinase domain and/or C-terminal tail. Additional experiments would be needed to characterize these findings in greater detail, but we speculate that the results suggest one or a mix of two possibilities when MK2206 binds to autoinhibited Akt: i) a reorientation of the PH domain with respect to the kinase domain or C-tail; ii) conformational/dynamic rearrangements within the PH domain itself.

We now include the following lines in the revised manuscript based on the CST data:

“We also performed a CST experiment in the same manner as described above, in the presence of MK2206 (Figure 8B,E and F) and observed two principal effects mediated by the compound. The short hinge of aa 44-46 seems to separate further from the kinase domain and/or C-terminal tail while the C-terminal α-helix, in contrast, appears to interact more tightly with the kinase domain or C-tail.”

We modified the manuscript to clarify the possible interpretation of NMR data:

“These inhibitor-induced changes likely result from a combination of altered structural and dynamic features relative to the baseline autoinhibited form of Akt. These could include a reorientation of the PH domain with respect to the kinase domain or C-tail (placing the C-terminal helix closer) as well as conformational and dynamic rearrangements within the PH domain itself.”

7) Cross-saturation transfer NMR data are also underutilized. Examining the quantitative differences between non-p-C and pS475 should provide more precise insight into how serine phosphorylation might be affecting the conformational preference of the PH domain. It's difficult to see from Figure 4—figure supplement 3 how the cross-saturation transfer effect is stronger for non-phosphorylated versus tail-phosphorylated samples, as stated in the text. Could this be quantified in some way – perhaps showing the mean peak intensity ratios for each sample or a distribution of these ratios across all samples?

We thank the reviewers for this insightful comment. Although we think that interpreting the average of the CST effect over the entire PH domain might not be so edifying, we agree that the CST data can be presented in a more quantitative way (and we have explained the challenges in interpreting the CST data above). We have added a section in Figure 5 of the revised manuscript in which the CST effects over the four parts of the PH domain that we defined are presented for each phospho-state (Figure 5G). Statistical analysis of these PH domain parts revealed that the only significant differences are found for the short hinge (in which pSer473 makes it looser while pSer477/pThr479 makes it tighter) and for the C-terminal helix which shows a tighter interaction only in the pSer477/pThr479 case. We have highlighted these correlations in the revised manuscript.

8) The heavy focus on the loop following b3 (residues Gln43 and Asp44) seems driven (according to the text) by more dramatic chemical shift perturbation and more extensive line broadening in the phosphorylated forms compared to non-p-C. This seems a stretch as there are certainly other regions that show equally or nearly the same extent of spectral change (it's possible that the HSQC data itself makes the differences more evident, but expanded spectral data showing individual HSQC peaks are not provided).

We modified Figure 4 accordingly to add a few selected peaks, including Gln43 and Asp44, in an expanded spectrum (Figure 4C).

We included the following lines in the manuscript based on the CST data:

“Similarly, the CST behavior for Gln43 and Asp44 was closely linked on the phospho-state of the C-tail (Figure 5G and Figure 5—figure supplement 3) with an apparent loosening of the interaction for pSer473 but an apparent tightening for pSer477/pThr479.”

9) The remainder of the paper focuses on a mutation of the native DVD sequence (residues 44-46) to GPG in an effort to "favor a random coil secondary structure in this region." The authors do not characterize this mutant to determine whether the mutation successfully shifted secondary structure preference of the PH domain but instead put the mutation directly into their full-length semi-synthetic AKT. Despite the lack of characterization of the mutant, this is the one place in this paper where the authors see a dramatic result; Figure 5A shows that the mutation of the DVD motif causes a large decrease in the steady-state kinetics of AKT in its pSer473 form compared to wild type. The other phosphorylated form of AKT (pSer477/pThr497) is unaffected by the mutation in the PH domain. Clearly understanding the mechanistic basis for this difference would be an advance in this field but the manuscript as presented does not sufficiently explain the observations.

We agree that a detailed understanding of the DVD hinge in Akt regulation is incomplete. We think that the CST effects that the reviewers encouraged us to more quantitatively analyze, nicely correlate with the divergent catalytic impacts of DVD mutation between the two C-tail phospho-states. We believe that this unexpected structure-activity relationship in this part of the PH domain between the pSer473 and pSer477/pThr479 forms adds to our overall understanding of how Akt regulation is achieved by distinct patterns of C-tail phosphorylation.

10) In the Discussion, the authors interpret the failure of pSer473 to stimulate AKT activity as being due to changes in the "normal dynamics" of the DVD motif. Dynamics can be directly measured by NMR to support this conclusion, but this is surprisingly not included in the manuscript. The observation that activation loop phosphorylation (at Thr308) is also diminished upon mutation of the DVD motif to GPG might suggest that the mutation in the PH domain enhances the autoinhibitory contact between PH and kinase domains and/or alters the accessibility of the activation loop to PDK-1. Unfortunately, as presented, the manuscript provides little to advance our understanding of the precise role of this region of the PH domain in AKT regulation.

We agree with the reviewers that NMR dynamics experiments would be the best way to validate this hypothesis. However, typically a set of relaxation, CEST or relaxation-dispersion spectra takes about 40-60 times longer than a single 2D (due to their pseudo-3D nature and the longer recycling delays they require). In our case we are limited by concentration and sample stability. In our case a single 2D 15N-TROSY-HSQC took about 50 hrs and at the end of the experiment, we saw heavy precipitation, which makes such dynamics experiments infeasible. In the revised manuscript, we modified the text to make clear that our hypothesis about the dynamics is inferred from the previously published crystal structures rather than from NMR data:

“This suggests that the normal dynamics (loop to helix transition) of this DVD motif is necessary for allowing the PH domain to dislodge from the kinase domain in response to pSer473.”

11) In the final part of this work the authors explore the effects of drug binding on PIP3 binding as well as PH domain NMR chemical shifts. The data are not interpreted in a satisfactory manner; the authors state that "Notably, some of the most significant chemical shift perturbations were detected in the PIP3-binding interface…" but there are significant changes elsewhere in the PH domain as well. There is also no attempt to take into account the likely significant spectral changes that will arise from the drug itself.

We now recognize this shortcoming. As discussed above, we have discussed the MK2206 NMR changes in greater detail in the revised manuscript.

12) Finally, the last sentence of the paper suggests conformational heterogeneity might be present in the non-phosphorylated C-tail form of AKT. This is very logical and raises the question of whether the authors see any evidence for the existence of multiple states in solution for any of the segmentally labeled forms they have studied in this submitted paper or the previous 2018 publication?

We did not detect any proof of conformational heterogeneity in our spectra (e.g. peak doubling in the HSQCs). However, this can be due to a number of limitations including insufficient signal-to-noise ratio (due to low stability and long measurement times required) and incompatible time regimes (fast to intermediate exchange). We would have loved to perform relaxation dispersion or CEST experiments, which can shed light on the presence of minor states. As explained above, NMR dynamic experiments that theoretically are able to characterize such phenomenon are inaccessible in this experimental system. Therefore, we cannot address such heterogeneity by NMR at this stage. We have added the following sentence to the revised manuscript:

"Although NMR dynamics experiments due to their very long time scales are not feasible with our current experimental system, future alternative approaches may allow this to be investigated."

13) Some of the kinetic parameters for the WT semi-synthetic AKT proteins shown in Figure 2 seem to differ a little from the parameters reported earlier in Chu et al., 2018. For example, the kcat/Km for A2 (full-length AKT phosphorylated on T308) in Figure 2 of Chu et al. is 28-fold smaller than the value reported in the current manuscript. In the present manuscript, the concentration range of ATP used is 0-2000 µM, whereas the Km values reported for A2, A1, and A6 (Figure 2 of Chu et al.) are 200, 5100, and 3700 µM, respectively. Perhaps the authors could clarify the apparent discrepancy.

The reviewers are correct that the catalytic parameters of the wt phospho- and non-phospho-Akt enzymes analyzed here do not precisely match the data in Chu et al., 2018, they are within 3-fold for each state (and 2.8-fold but not 28-fold different for A2). We speculate that this may be due to some heterogeneity of PTMs in the recombinant portion of the enzymes made in insect cells and have added this potential explanation to the revised manuscript.

14) Perhaps mention that the reported kcat/Km's are apparent values as the second substrate (the peptide in this case) is not present at a saturating concentration.

We have added the qualifier "apparent" as suggested by the reviewers. However, we previously showed that the Km of the GSK3 peptide for all forms of Akt analyzed that contain pThr308 are ca. 3 μm meaning that we are likely working with saturating peptide substrate concentration in the cases analyzed here.

15) It would be helpful in Figure 1 if the various states were labeled as "active" or "inactive." In Figure 1B, the ribbon diagram of the PH-kinase unit is drawn upside. In this and other figures (e.g., 4F), it would be helpful if structural components and parts were clearly labeled on the figures.

In the revised manuscript, we modified all of the figures displaying the Akt structure to show the canonical kinase domain structure with the N-lobe on top (e.g. Figure 1B, Figure 5D,E,F, Figure 8G). We also labeled the different domains of Akt as well as the allosteric inhibitor MK2206 in Figure 5F (previous Figure 4F).

16) Can the authors comment on the purity of the samples used in Figure 2? Based on the figure supplement SDS-PAGE analysis, there appears to be a ~10-20% impurity. Can this be identified by MS analysis of the full-length protein to confirm that it isn't a species that will interfere with the analysis of enzyme kinetics?

The semisynthetic protein samples displayed in the SDSPAGE Figure 2—figure supplement 1 show about a 10% impurity. This impurity likely results from incomplete ligation of the C-tail peptide and/or trace proteolysis at the N- or C-terminus. Since prior work from our lab and others indicate that C-tail deletion of Akt is severely catalytically impaired. Moreover, these Akt forms have a 40 aa epitope tag at the N-terminus and we have shown previously (Chu et al., 2018) that the this tag (and its removal) has little or no effect on catalytic activity. Thus, we do not believe this impurity band should impact the interpretation of mutations/phosphorylation and their influence on kcat/Km values of the Akt forms in this study.

https://doi.org/10.7554/eLife.59151.sa2

Article and author information

Author details

  1. Nam Chu

    1. Division of Genetics, Department of Medicine, Brigham and Women's Hospital, Boston, United States
    2. Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, United States
    3. Department of Pharmacology and Molecular Sciences, Johns Hopkins School of Medicine, Baltimore, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Writing - original draft, Writing - review and editing
    Contributed equally with
    Thibault Viennet
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9717-5007
  2. Thibault Viennet

    1. Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, United States
    2. Department of Cancer Biology, Dana-Farber Cancer Institute, Boston, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Writing - original draft, Writing - review and editing
    Contributed equally with
    Nam Chu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5349-0179
  3. Hwan Bae

    1. Division of Genetics, Department of Medicine, Brigham and Women's Hospital, Boston, United States
    2. Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, United States
    Contribution
    Data curation, Formal analysis, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5252-252X
  4. Antonieta Salguero

    1. Division of Genetics, Department of Medicine, Brigham and Women's Hospital, Boston, United States
    2. Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, United States
    3. Department of Pharmacology and Molecular Sciences, Johns Hopkins School of Medicine, Baltimore, United States
    Contribution
    Data curation, Formal analysis, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Andras Boeszoermenyi

    1. Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, United States
    2. Department of Cancer Biology, Dana-Farber Cancer Institute, Boston, United States
    Contribution
    Formal analysis, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Haribabu Arthanari

    1. Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, United States
    2. Department of Cancer Biology, Dana-Farber Cancer Institute, Boston, United States
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Investigation, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    hari_arthanari@hms.harvard.edu
    Competing interests
    No competing interests declared
  7. Philip A Cole

    1. Division of Genetics, Department of Medicine, Brigham and Women's Hospital, Boston, United States
    2. Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, United States
    3. Department of Pharmacology and Molecular Sciences, Johns Hopkins School of Medicine, Baltimore, United States
    Contribution
    Conceptualization, Supervision, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    pacole@bwh.harvard.edu
    Competing interests
    Senior editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6873-7824

Funding

National Cancer Institute (CA74305)

  • Philip A Cole

Claudia Adams Barr Program in Innovative Cancer Research

  • Haribabu Arthanari

Austrian Science Fund (Schroedinger Fellowship)

  • Andras Boeszoermenyi

American Heart Association (19POST34380800)

  • Andras Boeszoermenyi

National Institutes of Health (EB002026)

  • Haribabu Arthanari

Kwanjeong Educational Foundation (Pre-doctoral fellowship)

  • Hwan Bae

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank TK Harris for helpful advice and encouragement. HA acknowledges funding from the Claudia Adams Barr Program for Innovative Cancer Research. AB was supported by an Austrian Science Fund’s Schrödinger Fellowship (J3872-B21) and an American Heart Association’s fellowship (19POST34380800). Maintenance of the NMR instruments used for this research was supported by NIH grant no. EB002026. PAC and AS are grateful for funding from NIH grant CA74305. HB thanks the Kwanjeong Educational Foundation for support.

Senior and Reviewing Editor

  1. John Kuriyan, University of California, Berkeley, United States

Reviewers

  1. Kevin N Dalby, The University of Texas at Austin, United States
  2. Neel Shah

Publication history

  1. Received: May 20, 2020
  2. Accepted: August 3, 2020
  3. Accepted Manuscript published: August 3, 2020 (version 1)
  4. Version of Record published: August 19, 2020 (version 2)

Copyright

© 2020, Chu et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 794
    Page views
  • 162
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Biochemistry and Chemical Biology
    Erica N Thomas et al.
    Research Article

    Similar to DNA replication, translation of the genetic code by the ribosome is hypothesized to be exceptionally sensitive to small chemical changes to its template mRNA. Here we show that addition of common alkylating agents to growing cultures of E. coli leads to accumulation of several adducts within RNA, including N(1)-methyladenosine (m1A). As expected, the introduction of m1A to model mRNAs was found to reduce the rate of peptide-bond formation by three orders of magnitude in a well-defined in vitro system. These observations suggest that alkylative stress is likely to stall translation in vivo and necessitates activation of ribosome-rescue pathways. Indeed, the addition of alkylation agents was found to robustly activate the transfer-messenger RNA system, even when transcription was inhibited. Our findings suggest that bacteria carefully monitor the chemical integrity of their mRNA and they evolved rescue pathways to cope with its effect on translation.

    1. Biochemistry and Chemical Biology
    2. Cell Biology
    Mikel Garcia-Marcos et al.
    Tools and Resources

    Heterotrimeric G-proteins are signal transducers involved in mediating the action of many natural extracellular stimuli as well as of many therapeutic agents. Non-invasive approaches to manipulate the activity of G-proteins with high precision are crucial to understand their regulation in space and time. Here, we developed LOV2GIVe, an engineered modular protein that allows the activation of heterotrimeric G-proteins with blue light. This optogenetic construct relies on a versatile design that differs from tools previously developed for similar purposes, i.e. metazoan opsins, which are light-activated GPCRs. Instead, LOV2GIVe consists of the fusion of a G-protein activating peptide derived from a non-GPCR regulator of G-proteins to a small plant protein domain, such that light uncages the G-protein activating module. Targeting LOV2GIVe to cell membranes allowed for light-dependent activation of Gi proteins in different experimental systems. In summary, LOV2GIVe expands the armamentarium and versatility of tools available to manipulate heterotrimeric G-protein activity.