Crystal structure and mechanistic studies of the PPM1D serine/threonine phosphatase catalytic domain

Abstract

Protein phosphatase 1D (PPM1D, Wip1) is induced by the tumor suppressor p53 during DNA damage response signaling and acts as an oncoprotein in several human cancers. Although PPM1D is a potential therapeutic target, insights into its atomic structure were challenging due to flexible regions unique to this family member. Here, we report the first crystal structure of the PPM1D catalytic domain to 1.8 Å resolution. The structure reveals the active site with two Mg²⁺ ions bound, similar to other structures. The flap subdomain and B-loop, which are crucial for substrate recognition and catalysis, were also resolved, with the flap forming two short helices and three short β-strands that are followed by an irregular loop. Unexpectedly, a nitrogen-oxygen-sulfur bridge was identified in the catalytic domain. Molecular dynamics simulations and kinetic studies provided further mechanistic insights into the regulation of PPM1D catalytic activity. In particular, the kinetic experiments demonstrated a magnesium concentration-dependent lag in PPM1D attaining steady-state velocity, a feature of hysteretic enzymes that show slow transitions compared with catalytic turnover. All combined, these results advance the understanding of PPM1D function and will support the development of PPM1D-targeted therapeutics.

Keywords: B-loop; Mg(2+) ions; NOS bridge; PPM1D; catalytic domain; crystal structure; enzyme kinetics; flap subdomain; metal-dependent protein phosphatases; serine/threonine protein phosphatases.

Figure 1

Domain organization, construct design, and biophysical characterization of PPM1D variants.A, domain organization of human PPM1D. PPM1D comprises the catalytic domain (residues 1–420) and the regulatory domain (residues 421–605). The catalytic domain has two unique loop regions, the P-loop and the B-loop. A set of crystallographic constructs (#1–7) were designed to screen for soluble protein with improved crystallization propensity. The protein constructs comprised an N-terminal His₆-tagged, SUMOstar fusion protein, and SUMO protease cleavage site. B, phosphopeptide concentration dependence of phosphatase activity by PPM1D variants. The activity was measured with 0 to 400 μM ATM (1981pS) phosphopeptide. Data represent the means ± SD of two titrations. C, thermal stability of PPM1D constructs #5 and #6 was measured with the fluorescence-based nanoDSF method. Data are presented as mean values ± SD of two independent sample runs. D, microscale thermophoresis (MST) binding assay of MgCl₂ to PPM1D constructs #6 and #7. Graphs are represented as F_norm against ligand concentration (MgCl₂). Normalized fluorescence data were exported from MO. Affinity Analysis software version 2.3 (NanoTemper Technologies). Dissociation constants were determined using the one site-total binding model with nonlinear curve fitting in GraphPad Prism 9. Error bars represent the mean values ± SD of each data point calculated from two independent thermophoresis measurements.

Figure 2

Overall structural features of PPM1D.A, cartoon and surface representation of the PPM1D crystal structure. The ⍺-helices, β-sheets, and loop regions are shown as a cartoon colored with cyan, magenta, and light gray, respectively. The N- and C-termini of the protein are labeled. The two bound metal ions MG 1 (M1) and MG 2 (M2) are shown as green spheres. The PEG ligand in the active site is shown as sticks colored with tan (carbon atoms) and red (oxygen atoms). The flap subdomain (green) and the B-loop (blue) are close to the active site. Locations of the five-point mutations (G8R, P33E, R250A, R258A, and R259A) within crystallization construct #6 are labeled. The metal binding residues shown in the key are highlighted in coral color. B, close-in view of the active-site residue side chains is shown in stick representation. Metal-coordinating water molecules are represented as red balls and hydrogen bonds (distance < 3.4 Å) are indicated in purple. M1 and M2 jointly coordinate the bridging water molecule/hydroxide ion (blue). M2 is also coordinated by Asp105, Gly106, and three water molecules. M1 is also coordinated by Asp105, Asp314, Asp366, and two water molecules. Side chains are labeled with their respective residues. C, metal-binding residues in the PPM1D active site. Hydrogen bonds (distance < 3.4 Å) are indicated as purple dashed lines. D, the surface electrostatic potential features of the PPM1D active site (electronegative, red and electropositive, blue, as indicated by the heat scale). Figures were generated using UCSF ChimeraX (83). E, PEG binding in the active site with coordinating water molecules. The H-bonds (green dotted lines) with the distances indicated in Å, hydrophobic contacts (arcs with spokes), and interacting water molecules (cyan) and magnesium (pink) are shown. The two-dimensional representation was generated using the Ligplot+ diagram (84).

Figure 3

Overall three-dimensional structural comparisons of the flap subdomain among various PPM protein phosphatases.A, cartoon representation of PPM1D highlighting three features of interest: the active site (surface), the flap subdomain (green), and the B-loop (blue). B, three-dimensional structural comparisons are shown among different PPM structures, including human PPM1D (blue, PDB ID 8T2J), cyanobacterial PPM tPphA (purple, PDB ID 5ITI), Mycobacterium tuberculosis PstP (green, PDB ID 1TXO), and Streptococcus agalactiae SaSTP (plum, PDB ID 2PK0). C, the structural superimposition of human PPM1D (blue, PDB ID 8T2J) with human PPM1A (salmon, PDB ID 1A6Q, pink, PDB ID 6B67), Arabidopsis thaliana PPM Hab1 (gray, PDB ID 4LA7), and Arabidopsis thaliana PPM Abi2 (magenta, PDB ID 3UJK), human PPM1K (purple, PDB ID 2IQ1) is presented. D, comparisons of the flap domain of PPM1D (blue, PDB ID 8T2J) with PPM1H (orange, PDB ID 7L4J, chain A). Compared to PPM1D, the PPM1H flap is extended. E, a cartoon view of the PPM1D structure, colored according to temperature factor values from blue (low) to white (high). The stick representation shows the B-loop and part of the flap subdomain. Figures were generated using UCSF Chimera (85).

Figure 4

Comparisons between the PPM1D crystal structure and the structure predicted by AlphaFold.A, superposition of the PPM1D crystal structure (gray) and the AlphaFold-predicted structure (blue, https://alphafold.ebi.ac.uk/entry/O15297) (43, 44). B, close-in view of the superposition of the flap region with the conformation of the B-loop in the PPM1D crystal structure depicted with green highlight. C, packing of the B-loop in the crystal structure. PPM1D is shown as a ribbon cartoon. The B-loop (residues 246–267) is highlighted in magenta. The two Mg²⁺ atoms in the active site are shown as green spheres. Other proteins in the crystal lattice that contact the B-loop are represented as translucent volumes in differing tints. Two sites of major contact are indicated by yellow ellipsoids. Figures were generated using UCSF Chimera (85).

Figure 5

Time course studies of PPM1D (1–420) phosphatase activity toward the ATM phosphopeptide substrate implicate a slow enzyme conformational transition.A, dependence of PPM1D phosphatase activity on substrate and magnesium ion concentrations at 300 s. Lines connect data points with constant [MgCl₂]. B, progress curves for PPM1D phosphatase activity display a lag in attaining steady state velocities. Reactions contained 104 μM ATM substrate and indicated [MgCl₂]. Nonlinear curve fitting was used to estimate the hysteretic enzyme parameters v_ss and k. The annotated straight lines indicate the steady state rates. C, PPM1D steady state velocities exhibit hyperbolic dependence on substrate concentrations with magnesium ion concentration-dependent values of the apparent Michaelis–Menten parameters, appV_max and appK_M. D, secondary analysis of the apparent steady state parameters, appV_max and appK_M. The apparent V_max, appV_max is hyperbolically dependent on [MgCl₂] with parameters V_max^SA = 17.0 s^-1 and K_M^SA = 21.8 mM (left plot). The best-fit curve is indicated by the dashed blue line. The apparent K_M for substrate, appK_M (right plot) is weakly dependent on [MgCl₂]. The high value of appK_M at 6.25 mM MgCl₂ is not reproduced by the model curve (dotted red line). E, hysteretic enzyme relaxation times (1/k) for the transient phase exhibit complex dependencies on magnesium ion and substrate concentrations. Dashed lines connect data points with constant [MgCl₂]. F, the presence of a lag phase is dependent on the order of addition. Panels A – E were based on the Rho-PBP assay and were initiated by the addition of 8.15 nM PPM1D (1–420) (n = 4). Results depicted in panel F were based on phosphate detection by Biomol green and contained 5 nM PPM1D (1–420), 50 μM ATM substrate, and 25 mM MgCl₂ (n = 3).

Figure 6

Functional and structural analysis of PPM1D in the oxidized and reduced state.A, the overall structure of PPM1D is shown in cartoon representation with the Lys336-Cys346 cross bridge shown in stick representation. The corresponding 2mFo-DFc electron density maps are depicted in blue, with a contour level of 1σ. The NOS bridge between Lys336 (located on Helix 6, colored cyan) and Cys346 (located on Helix 7, magenta) has 100% occupancy. B, a close-up view of the NOS bridge between Lys336 and Cys346. C, PPM1D crystal structure (PDB ID 8T2J), showing the positions of residues Lys336, Cys329, Cys346, and Cys374, as well as the active site. D, steady-state kinetic analysis of PPM1D in the reduced (black) and oxidized (red) state. Data are presented as mean values ± SEM of two independent sample runs. E, multiple sequence alignment of 16 vertebrate Protein phosphatase 1D (PPM1D) homologs for the 327 to 350 segment. Human PPM1D is listed at the top with secondary structure and Lys336 and Cys346 positions numbered. The corresponding aligned lysine and cysteine residues are indicated with bold blue or red font. Conservation of both residues occurs in mammals (pink shading) and birds (blue shading).

Figure 7

Variations in the mechanical linkage between the active site and NOS bridge of PPM1D revealed by corresponding nodes of the Motion Tree analyses. The four protein systems are as follows: (A) three Mg²⁺ ions without an NOS bridge, (B) two Mg²⁺ ions without an NOS bridge, (C) three Mg²⁺ ions with an NOS bridge, and (D) two Mg²⁺ ions with an NOS bridge. The blue and red segments indicate the two semi-rigid domains. Figures were generated using the PyMOL Molecular Graphics System, Version 2.5.5 Schrödinger, LLC (https://www.pymol.org/).