PHLPP2 is a pseudophosphatase that lost activity in the metazoan ancestor

Abstract

The phosphoinositide 3-kinase (PI3K) pathway is a major regulator of cell and organismal growth. Consequently, hyperactivation of PI3K and its downstream effector kinase, Akt, is observed in many human cancers. Pleckstrin homology domain leucine-rich repeat-containing protein phosphatases (PHLPP), two paralogous members of the metal-dependent protein phosphatase family, have been reported as negative regulators of Akt signaling and, therefore, tumor suppressors. However, the stoichiometry and identity of the bound metal ion(s), mechanism of action, and enzymatic specificity of these proteins are not known. Seeking to fill these gaps in our understanding of PHLPP biology, we unexpectedly found that PHLPP2 has no catalytic activity in vitro. Instead, we found that PHLPP2 is a pseudophosphatase with a single zinc ion bound in its catalytic center. Furthermore, we found that cancer genomics data do not support the proposed role of PHLPP1 or PHLPP2 as tumor suppressors. Phylogenetic analyses revealed an ancestral phosphatase that arose more than 1,000 Mya, but that lost activity at the base of the metazoan lineage. Surface conservation indicates that while PHLPP2 has lost catalytic activity, it may have retained substrate binding. Finally, using phylogenomics, we identify coevolving genes consistent with a scaffolding role for PHLPP2 on membranes. In summary, our results provide a molecular explanation for the inconclusive results that have hampered research on PHLPP and argue for a focus on the noncatalytic roles of PHLPP1 and PHLPP2.

Keywords: Akt; PHLPP; cancer; phosphatase; signaling.

Fig. 1.

PHLPP2 exhibits no detectable activity against Akt. (A) Structure of the Akt1 kinase domain depicting activation loop (T308, blue) and hydrophobic motif (S473, cyan) phosphopeptides used in phosphatase assays. (B) Protein phosphatase activity of purified PHLPP2^D1024N (red) compared to PHLPP2^WT (blue) against the activation loop peptide of Akt. (C) Dephosphorylation of the activation loop peptide by three independent preparations of recombinant PHLPP2 purified from Sf9 insect cells (blue) ±12.5 nM okadaic acid (red). (D) Dephosphorylation of the hydrophobic motif peptide by three independent preparations of recombinant PHLPP2 purified from Sf9 insect cells (cyan) ±12.5 nM okadaic acid (red). (E) Dephosphorylation of the activation loop peptide by purified PP2A (blue) ±12.5 nM okadaic acid (red). (F) Dephosphorylation of the hydrophobic motif peptide by purified PP2A (cyan) ±12.5 nM okadaic acid (red). (G) Peptide fingerprinting mass spectrometry to identify proteins in recombinant PHLPP2 purified from insect cells. Red: molecular weight of Sf9 PP2A heterotrimers. (H) Schematic for immunoprecipitation of PHLPP2 from HEK293 cells and subsequent phosphatase assay. (I) Phosphatase activity of immunoprecipitated PHLPP2 determined after successive wash steps ±12.5 nM okadaic acid. B = beads, SN = supernatant.

Fig. 2.

PHLPP2 is a pseudophosphatase. (A) Structure of the PP2A catalytic subunit bound to the covalent inhibitor microcystin-LR. (B) Structure of the PP2A catalytic subunit bound to okadaic acid. (C) Schematic for covalently coupling microcystin-LR to NHS-activated sepharose beads (26). (D) Tandem mass spectrometry analysis of proteins bound to microcystin-LR-conjugated beads after incubation with cell lysates of Sf9 cells heterologously overexpressing PHLPP2. (E) Purification scheme for affinity-based removal of PP2A family phosphatases from purified PHLPP2. (F) Dephosphorylation of activation loop (blue) and hydrophobic motif (cyan) peptides by PHLPP2 purified according to the scheme in E ± 12.5 nM okadaic acid (red). (G) Dephosphorylation of the generic phosphatase substrate para-nitrophenyl phosphate (pNPP, blue) by PHLPP2 purified according to the scheme in E ± 12.5 nM okadaic acid (red) or lambda phosphatase (green). Assay buffer includes 2 mM MnCl₂.

Fig. 3.

PHLPP2 is a zinc-binding protein. (A) Catalytic site of PPM1A phosphatase (gray) superimposed on the AF2 model of the PHLPP2 PP2C domain (lilac). M1, M2, and M3 metal ions shown in spheres; metal coordinating amino acids shown in stick representation. PPM1A residue numbers shown in parentheses. (B) Multiple sequence alignment of the metal ion-coordinating motifs of PHLPP1, PHLPP2, and PHLPP-like proteins across a wide evolutionary time. PPM1A residue numbers shown above the alignment in parentheses. (C) ICP-MS chromatograms for recombinant human PHLPP2. (D) Calibration of PHLPP2:Zn²⁺ stoichiometry for three biological replicates. (E) Differential alkylation of PHLPP2 under native and denaturing conditions, coupled to mass spectrometry. (F) Differential alkylation of five different peptides of PHLPP2 (two representing buried cysteines, 2 representing surface exposed cysteines, and one encompassing C799). (G) Quantum mechanical simulation of zinc ion coordination sphere using the ONIOM method (32). Results of two quantum mechanical frameworks (B3LYP and M062X) are shown. (H) Geometry of the zinc-binding site from the simulations showing bond angles and lengths.

Fig. 4.

Cancer genomics does not support a role of PHLPP1 or PHLPP2 as tumor suppressors. (A) Distribution of nonsynonymous mutations within the coding sequence of Akt1 (Tier 1 Cancer Gene Census cancer gene). Red: known hotspot mutations. (B) Number of synonymous and nonsynonymous substitutions per 1,000 amino acids reported in COSMIC for: PHLPP1 and PHLPP2; the tumor suppressor genes PTEN and TP53; the oncogenes KRAS and BRAF; the olfactory receptor genes OR2A4 and OR51E2. (C) Distribution of nonsynonymous mutations within the coding sequence of the genes represented in (A). Red: known hotspot mutations in CGC genes. (D) CNV for the genes represented in (B). Green: CNG; red: CNL.

Fig. 5.

PHLPP exhibits a conserved arrangement of its regulatory domains. (A) Domain architecture of human PHLPP2 [RA domain, magenta; PH domain, orange; LRR (gray); PP2C, blue; pseudo-catalytic site of PP2C domain, red]. (B) Rotary shadowing electron microscopy of recombinant PHLPP2. (C) Top-ranked AlphaFold2 model of human PHLPP2 [putative active site of PP2C domain, red]. (D) Predicted alignment error (PAE) plot for AF2 prediction in (C). PHLPP2 domain boundaries are indicated by colored boxes. (E) CryoEM maps of human PHLPP2 (conformation 1) at 6 Å resolution with the AlphaFold2-predicted structure of PHLPP2 fit to the map. (F) CryoEM maps of human PHLPP2 (conformation 2) at 6 Å resolution with the AlphaFold2-predicted structure of PHLPP2 fit to the map.

Fig. 6.

Evolution and diversification of PHLPP genes. (A) Maximum-likelihood phylogeny of the LRR-PP2C protein family. Species and clade names have been added and the number of sequences within collapsed nodes are noted in parentheses. The gain and loss of genes, domains, and the PP2C active site (PA) have been mapped across the phylogeny based on Dollo parsimony. A species tree based on the taxa present in the phylogeny and current eukaryotic taxonomy (50) is shown for reference. The full phylogeny, domain annotations, and active site residues can be viewed at https://itol.embl.de/shared/9pitR7NesER8. (B) Representative domain architectures of proteins from across the LRR-PP2C phylogeny. Taxonomic groups are noted with cartoons obtained from Phylopic.org and protein lengths are shown in amino acids (a. a.). S. c., Saccharomyces cerevisiae; R. i., Rhizophagus irregularis; C. o., Capsaspora owczarzaki; C. l., Corallochytrium limacosporum; H. s. 1, Homo sapiens PHLPP1; H. s. 2, Homo sapiens PHLPP2; C. e., Caenorhabditis elegans; E. m., Ephydatia muelleri; D. d., Dictyostelium discoideum; S. s., Stenamoeba stenapodia; T. f., Tritrichomonas foetus; A. a., Anaeramoeba ignava. RA, Ras binding domain; AC, adenylate cyclase; PH, PH-domain; DD, dimerization domain; LRR, leucine-rich repeat; AGC, AGC kinase domain; PP2C, PP2C domain. (C) Sequence logos showing PP2C active site conservation. PPM1A residues are noted for reference. (D) AlphaFold3 prediction of S. stenapodia LRR-PP2C. Inset: predicted coordination of M1 and M2 metal ions in the catalytic site. (E) Phosphatase activity of S. stenapodia LRR-PP2C against pNPP. (F) Domain architectures, AlphaFold2-predicted structures, and their associated confidence metrics (PAE plots). The rmsd of the LRR-PP2C domains (over C_α atoms) of each structure from the ancestral LRR-PP2C phosphatase of Entamoeba histolytica is shown for comparison. Red squares: pseudophosphatase (inactive); green circles: phosphatase (active).

Fig. 7.

PHLPP may have an alternative scaffolding role on membranes. (A) Surface conservation maps of vertebrate PHLPP. Low conservation (blue) to high conservation (red). Dashed white line: surface of the RA domain used to bind RAS in Cyr1. (B) Electrostatic surface potential map of the PH domain of H. sapiens PHLPP2. Phosphoinositide headgroup ligands superimposed from experimentally determined structures of homologous ArhGAP9 and PLEKHA7 PH domains identified by FoldSeek (65). (C) Lipid-protein overlay assay of PHLPP2 with phospholipids. Quantified signal is the mean of two independent replicates. (D) Conservation of the FLAP subdomain of the PHLPP PP2C domain. Secondary structure topology of the PP2C domain, indicating the FLAP subdomain in red. Circle Inset: superposition of AlphaFold-derived structures of FLAP-subdomain of S. stenapodia LRR-PP2C and a homologous PP2C phosphatase from the bacterium Desulfobacterota bacterium. (E) Phylogenomic profiling of gnathostomes to identify correlated gene presence and absence patterns for PHLPP2. Six of the most highly correlated genes are shown, alongside the presence/absence patterns of previously reported PHLPP2 substrates, Akt and PKC.