Allosteric inhibition of PPM1D serine/threonine phosphatase via an altered conformational state

Abstract

PPM1D encodes a serine/threonine phosphatase that regulates numerous pathways including the DNA damage response and p53. Activating mutations and amplification of PPM1D are found across numerous cancer types. GSK2830371 is a potent and selective allosteric inhibitor of PPM1D, but its mechanism of binding and inhibition of catalytic activity are unknown. Here we use computational, biochemical and functional genetic studies to elucidate the molecular basis of GSK2830371 activity. These data confirm that GSK2830371 binds an allosteric site of PPM1D with high affinity. By further incorporating data from hydrogen deuterium exchange mass spectrometry and sedimentation velocity analytical ultracentrifugation, we demonstrate that PPM1D exists in an equilibrium between two conformations that are defined by the movement of the flap domain, which is required for substrate recognition. A hinge region was identified that is critical for switching between the two conformations and was directly implicated in the high-affinity binding of GSK2830371 to PPM1D. We propose that the two conformations represent active and inactive forms of the protein reflected by the position of the flap, and that binding of GSK2830371 shifts the equilibrium to the inactive form. Finally, we found that C-terminal truncating mutations proximal to residue 400 result in destabilization of the protein via loss of a stabilizing N- and C-terminal interaction, consistent with the observation from human genetic data that nearly all PPM1D mutations in cancer are truncating and occur distal to residue 400. Taken together, our findings elucidate the mechanism by which binding of a small molecule to an allosteric site of PPM1D inhibits its activity and provides insights into the biology of PPM1D.

Fig. 1. Generation of a homology model and use of computational genetics to characterize PPM1D.

A PPM1D homology model of residues 1–377 highlighting key structural regions of interest. The N-terminus (purple), loop (brown), hinge (cyan), and flap (light purple) are highlighted along with predicted alpha-helices (red) and beta-sheets (yellow), with the remainder of the protein in green. The loop and flap region are shown with an overlaying surface representation to emphasize the lack of structure predicted by the model. The metal-binding residues and metal cations are highlighted as spheres in black and white, respectively. A 2D representation is shown below the homology model. B PPM1D paralog values superimposed on the homology model. For each residue, the degree of identity across the human PPM1D paralogs was assessed and scored from 0 to 10 (purple represents full conservation, white represent no conservation). C Representation of genetic constraint in the human germline at each residue as calculate by the missense tolerance ratio. A ratio of less than 0.68 and 0.46 are in the bottom 25% and 5% of scores across the protein, respectively.

Fig. 2. Deep mutational scanning identifies residues important for PPM1D function.

A Schematic of engineered K562 cells with wild-type TP53 and GFP knocked into the C-terminus of the CDKN1A (p21) locus. In the presence of a DNA damaging agent, the DNA damage response (DDR) pathway induces p53 activation which transactivates the p21 locus, leading to expression of p21-GFP. After treatment with a DNA damaging agent such as daunorubicin, cells expressing a truncated PPM1D with a mutation that does not impair the function of the protein (red) suppresses the DDR and GFP expression whereas cells expressing truncated PPM1D that does impair protein function (blue) does not suppress the DDR or GFP expression. B Effect of mutations on PPM1D-mediated suppression of the DDR. Cells expressing truncated PPM1D (tPPM1D) or a mutation that does not affect protein function (tPPM1D-K238A) have lower p21-GFP levels at baseline and in the presence of daunorubicin. In contrast, cells expressing truncated PPM1D with mutations that impair catalytic function (tPPM1D-D314A, tPPM1D-DGH(105-107)→AAA, or tPPM1D-MED(21–23)→AAA) have higher levels of p21-GFP at baseline and in the presence of daunorubicin. N = 3 biologically independent samples. Data are presented as mean values ± SEM. Source data are provided as a Source Data file. C Representation of deep mutagenesis screen. Each column represents the amino acid position and each row represents that amino acid change in the variant including stop codons, silent mutations, or the average of all the missense mutations. For each variant the log₂(GFP_Hi/GFP_Lo) was calculated and scaled with red representing a greater ratio (impaired activity), and blue representing a lesser ratio (retained activity). See Supplemental Table 4 for source data. D Overlap of heatmap missense mutations on theoretical model. The average effect of the missense variants at each amino acid position are overlaid onto the theoretical model of truncated PPM1D.

Fig. 3. GSK2830371 functions as a non-competitive inhibitor and binds with high affinity to PPM1D at an allosteric site.

A Activity of GSK2830371 in the FDP (red) and p38 MAPK (black) enzymatic assays. FDP is cleaved by PPM1D in a substrate-independent fashion while p38 MAPK threonine 180 is dephosphorylated in a substrate-specific manner. N = 3 independent samples runs. Data are presented as mean values ± SD. B Mechanism of inhibition studies using non-linear fit of Michaelis–Menten enzyme kinetics where hydrolysis of FDP is measured in relative fluorescence units (RFU). Data suggest noncompetitive mode of action with α ≥ 1. N = 2 independent samples runs. Data are presented as mean values ± SD. C Differential scanning calorimetry (DSC) and fluorimetry (DSF) showing the melting temperature of PPM1D_1–420 in the absence (black) and presence (red) of GSK2830371. N = 3 independent sample runs. Data are presented as mean values. D Surface plasmon resonance analysis of the interaction of GSK2830371 with PPM1D_1–420. The K_d is calculated from the k_on and k_off measurements. E Structure of GSK2830371 and the GSK2830371 analog 1d used in the isothermal calorimetry experiments. F The thermodynamic profile of the interaction of the GSK2830371 analog (1d) with PPM1D_1–420 as measured by isothermal calorimetry. N = 3 independent sample runs. Source data are provided in the Source Data file.

Fig. 4. PPM1D exists in multiple conformational states reflected by movement of the flap domain.

A HDX-MS measurements of deuterium uptake for PPM1D_1–420 over time. The peptides analyzed run along the X-axis from the N-terminus (left) to C-terminus (right) and vertical tick marks between peptides are shown to indicate regions of missing coverage. Deuterium incorporation is represented as percent fractional uptake on the Y-axis, calculated as described in the “Materials and Methods” section. The 10 s exposure time point is shown in red to highlight the uptake values mapped in panel B. Reference plots are reported in Suppl. Data 1, the raw values of the plots are reported in Suppl. Data 2B Fractional uptake at 10 s exposure for PPM1D_1–420 mapped onto the homology model. C Spectral analysis showing the change in EX1 kinetics for peptide ₂₇₀ AVARALGDL ₂₇₈ located in the flap region in PPM1D_1–420 and PPM1D_Δhinge. D Relative differences in deuterium uptake between PPM1D_Δflap compared to PPM1D_1–420. E Relative differences in deuterium uptake between PPM1D_Δhinge compared to PPM1D_1–420. F Differences shown in panels (D, E) between PPM1D_Δhinge (left) or PPM1D_Δflap (right) and PPM1D_1–420 mapped onto homology model. Negative differences are represented in blue indicating protection from exchange, while positive differences are shown in red, indicating exposure to exchange as a result of the deletion. Darker colors indicate a stronger relative difference. For panels (A, D, E) the bars below the X-axis represent the N-terminus (purple), loop (brown), hinge (cyan), flap (light purple), and C-terminus (black) of the protein. Source data are provided as a Source Data file.

Fig. 5. GSK2830371 binding alters the conformational sampling of the flap domain of PPM1D.

A Relative differences in deuterium uptake between PPM1D_1–420 in the presence of GSK2830371 compared to PPM1D_1–420 without GSK2830371 as assessed by HDX-MS. Vertical tick marks between peptides on the X-axis are shown to indicate regions of missing coverage. The bars below the X-axis represent the N-terminus (purple), loop (brown), hinge (cyan), flap (light purple), and C-terminus (black) of the protein. Reference plots are reported in Suppl. Data 1, the raw values of the plots are reported in Suppl. Data 2B EX1 analysis showing the change in deuterium uptake for peptide ₂₇₀ AVARALGDL ₂₇₈ located in the flap region in PPM1D_1–420 in the absence (left) and presence (right) of GSK2830371. C Sedimentation populations of PPM1D_1–420 in the absence (black) and presence (red) of GSK2830371 as measured by SV-AUC. D Sedimentation populations of PPM1D_Δflap in the absence (black) and presence (red) of GSK2830371 as measured by SV-AUC.

Fig. 6. The Hinge, not the flap, mediates GSK2830371 binding.

A Surface plasmon resonance of PPM1D_Δflap and GSK2830371 (red represents raw data and black represents fitted kinetic parameters). B Relative differences in deuterium uptake between PPM1D_Δflap in the presence of GSK2830371 compared to PPM1D_Δflap in the absence of GSK2830371 as assessed by HDX-MS. Vertical tick marks between peptides on the X-axis are shown to indicate regions of missing coverage. The bars below the X-axis represent the N-terminus (purple), loop (brown), hinge (cyan), flap (light purple), and C-terminus (black) of the protein. C Activity of GSK2830371 in the FDP (top) and p38 MAPK (bottom) enzymatic assays with PPM1D_1–420 (black) or PPM1D_Δhinge (red). N = 2 independent samples runs. Data are presented as mean values ± SD. D EX1 analysis showing the change in deuterium uptake for peptide ₂₇₀ AVARALGDL ₂₇₈ located in the flap region in PPM1D_1–420 and PPM1D_Δhinge in the absence (left) and presence (right) of GSK2830731. Source data are provided in the Source Data file.

Fig. 7. Truncations proximal to amino acid 400 destabilize PPM1D.

A Distribution of PPM1D truncating mutations in solid tumor sequencing data from GENIE (v5.0) obtained from cBioportal (www.cbioportal.org). B Effects of different truncations on the thermal stability of PPM1D by differential scanning calorimetry. N = 3 independent sample runs. Data are presented as mean values ± SD. Source data are provided as a Source Data file. C Circular dichroism analysis of PPM1D_1–420 (black), PPM1D_1–400 (pink) and PPM1D_1–377 (brown). D Relative differences in deuterium uptake between PPM1D_1–420 and PPM1D_1–400 (top) and between PPM1D_1–420 and PPM1D_1–377 (bottom) as assessed by HDX-MS. Vertical tick marks between peptides on the X-axis are shown to indicate regions of missing coverage. The bars below the X-axis represent the N-terminus (purple), loop (brown), hinge (cyan), flap (light purple), and C-terminus (black) of the protein.