In vitro reconstitution of Sgk3 activation by phosphatidylinositol 3-phosphate

Abstract

Serum- and glucocorticoid-regulated kinase 3 (Sgk3) is a serine/threonine protein kinase activated by the phospholipid phosphatidylinositol 3-phosphate (PI3P) downstream of growth factor signaling via class I phosphatidylinositol 3-kinase (PI3K) signaling and by class III PI3K/Vps34-mediated PI3P production on endosomes. Upregulation of Sgk3 activity has recently been linked to a number of human cancers; however, the precise mechanism of activation of Sgk3 is unknown. Here, we use a wide range of cell biological, biochemical, and biophysical techniques, including hydrogen-deuterium exchange mass spectrometry, to investigate the mechanism of activation of Sgk3 by PI3P. We show that Sgk3 is regulated by a combination of phosphorylation and allosteric activation. We demonstrate that binding of Sgk3 to PI3P via its regulatory phox homology (PX) domain induces large conformational changes in Sgk3 associated with its activation and that the PI3P-binding pocket of the PX domain of Sgk3 is sequestered in its inactive conformation. Finally, we reconstitute Sgk3 activation via Vps34-mediated PI3P synthesis on phosphatidylinositol liposomes in vitro. In addition to identifying the mechanism of Sgk3 activation by PI3P, our findings open up potential therapeutic avenues in allosteric inhibitor development to target Sgk3 in cancer.

Keywords: CISK; Sgk3; allosteric regulation; endosome; hydrogen–deuterium exchange mass spectrometry (HDX-MS); phosphatidylinositide 3-kinase (PI 3-kinase); phospholipid; serine/threonine protein kinase.

Figure 1

Sgk3 localizes specifically to PI3P-containing endosomes. All images are representative of three independent transfections and multiple fields of view. A, HeLa cells were transfected with mCherry-Sgk3^WT and serum-starved overnight. B, starved cells were treated with 100 nM wortmannin for 20 min. C–E, cells were cotransfected with mCherry-Sgk3^WT and eGFP-Sgk3^R90A mutant. F, fluorescence intensity profile of mCherry Sgk3^WT and eGFP-Sgk3^R90A along the white line shown in C–E. G–I, cells were cotransfected with mCherry-Sgk3^WT and eGFP-EEA1^FYVE. J, fluorescence intensity profile of mCherry Sgk3^WT and eGFP-EEA1^FYVE along the white line shown in G–I.

Figure 2

Sgk3 is specifically and allosterically activated by PI3P. A, Sgk3^WT kinase activity was assayed in the presence (closed black square) or absence (closed red square) of 100 μM Crosstide substrate with 0–4 mol % PI3P liposomes or with 4 mol % PI(4,5)P₂ liposomes (open black square) as a negative control. In parallel, binding of Sgk3 to 0–4 % PI3P liposomes (full blue circles) or 4% PI(4,5)P₂ (open blue circle) as a negative control was measured. Dotted black line indicates basal Sgk3^WT kinase activity of unbound protein, dotted red line indicates no-substrate control. Kinase activity was read out radiometrically by liquid scintillation and liposome binding by Coomassie densitometry. Error bars are the standard deviation of three independent experiments. B, same as (A) with Sgk3^D286A kinase-dead mutant. C, schematic of Vps34-Sgk3 coupled kinase assay. Phosphorylation of Crosstide was assayed by phosphorimaging. PI3P synthesis was measured by liquid scintillation. D, Vps34-Sgk3 coupled kinase assay. Crosstide phosphorylation was measured in the presence (full blue line) or absence (dashed blue line) of 50 nM Vps34. At the same time, PI3P synthesis was measured in the presence (full green line) or absence (dashed green line) of 50 nM Vps34. The binding of Sgk3 to the liposomes was determined by a liposome pelleting assay at the indicated time points following termination of the reaction with 2.5 mM EDTA (black curve). Error bars are the standard deviation of three independent experiments.

Figure 3

Sgk3 is autoinhibited by its PX domain. A, comparison of deuterium incorporation in Sgk3 PX domain with 0% versus 5% PI3P liposomes. Plot: difference in number of deuterons incorporated as a function of peptide centroid (defined as the amino acid number corresponding to the center of each peptide) over the PX domain. Negative numbers indicate protection (blue). Peptides showing significant deuterium exchange differences (>7%, >0.5 Da, and p < 0.01) were mapped on the structure of the human Sgk3 PX domain (PDB:6EDX). The binding site for PI3P is indicated by superimposing the structure of the PX domain of p40phox in complex with PI3P (PDB: 1H6H) and rendering the PI3P in spheres. B, comparison of deuterium incorporation in Sgk3^WTversus Sgk3 PX domain in the presence of 0% PI3P liposomes (unbound). Plot: difference in number of deuterons incorporated as a function of peptide centroid over the PX domain. Positive numbers indicate domain exposure (red). C, comparison of deuterium incorporation in Sgk3^WT unbound (0% PI3P) versus bound (5% PI3P) for the PX domain (left) and the kinase domain (right). A homology model of Sgk3 kinase domain (Rosetta) (75) was used to map the changes. Color coding expressing the magnitude of changes in deuterium incorporation is on bottom right. All measurements were done in triplicates, error bars are the standard deviations.

Figure 4

Autoinhibition of Sgk3 impairs binding to PI3P. A, binding of Sgk3^WT full-length (blue squares) and FITC-labeled Sgk3 PX domain (black squares) to 0–4 mol % PI3P liposomes was monitored in single reaction using fluorescence (FITX-PX) or Coomassie densitometry (full-length Sgk3) in 15% SDS-PAGE gels. Curves were fitted with a one binding site model. Error bars are the standard deviations of three independent experiments. B–D, distribution of eGFP-PX domain (B) or mCherry-Sgk3^WT (full-length) (C) in serum-starved HeLa cells. All images are representative of three independent transfections and multiple fields of view. E, quantification of fluorescence intensity of both constructs along the white line shown in images B–D.

Figure 5

Model for the allosteric activation of Sgk3 by PI3P. Cytosolic Sgk3 is allosterically inhibited by its PX domain, which is relieved by PI3P binding (1). This is followed by phosphorylation of the Sgk3 activation loop by PDK1 as a consequence of its association with endosomal PI(3,4)P₂ (2). Phosphorylation activates Sgk3 (3), thereby permitting downstream substrate phosphorylation (4).