Target-Based Discovery of an Inhibitor of the Regulatory Phosphatase PPP1R15B

Abstract

Protein phosphorylation is a prevalent and ubiquitous mechanism of regulation. Kinases are popular drug targets, but identifying selective phosphatase inhibitors has been challenging. Here, we used surface plasmon resonance to design a method to enable target-based discovery of selective serine/threonine phosphatase inhibitors. The method targeted a regulatory subunit of protein phosphatase 1, PPP1R15B (R15B), a negative regulator of proteostasis. This yielded Raphin1, a selective inhibitor of R15B. In cells, Raphin1 caused a rapid and transient accumulation of its phosphorylated substrate, resulting in a transient attenuation of protein synthesis. In vitro, Raphin1 inhibits the recombinant R15B-PP1c holoenzyme, but not the closely related R15A-PP1c, by interfering with substrate recruitment. Raphin1 was orally bioavailable, crossed the blood-brain barrier, and demonstrated efficacy in a mouse model of Huntington's disease. This identifies R15B as a druggable target and provides a platform for target-based discovery of inhibitors of serine/threonine phosphatases.

Keywords: Huntington’s disease; PPP1R15B; drug discovery; eukaryotic initiation factor-2; neurodegenerative diseases; protein misfolding; protein phosphatase 1; protein quality control; proteostasis; stress response.

Graphical abstract

Figure S1

Sequence Alignment of Human R15A and R15B, Related to Figure 1 Sequences of human R15A and R15B using CLUSTAL O (version 1.2.1) shows 23.8% sequence identity (using a Percent Identity Matrix) (Sievers et al., 2011).

Figure 1

An SPR-Based Assay with Reconstituted R15 Holophosphatases Measures Affinities of R15A Inhibitors (A) Coomassie-stained gel showing recombinant proteins used in (B): MBP-R15A^325-636-His and MBP-R15B^340-698-His. (B) Normalized steady-state binding curves from SPR showing binding of R15A (○ cyan) and R15B (▲ magenta) to bio-GBZ immobilized on the streptavidin sensor chip surface. Bio-GBZ (biotinylated GBZ) is an R15A inhibitor as potent as GBZ (Tsaytler et al., 2011). (C) Coomassie-stained gel showing recombinant biotinylated PP1c (bio-PP1c; partially purified) and purified recombinant R15s (Input). Bio-PP1c, captured on neutravidin beads, bound R15A and R15B (Bound). Lower panel: immunoblot showing bio-PP1c. (D) Cartoon depicting the reconstitution of R15 holophosphatases (R15A^325-636-PP1c and R15B^340-698-PP1c; see STAR Methods) on a streptavidin (SA) SPR chip. (E and F) Normalized SPR steady-state binding curves showing binding of GBZ (E) or Sephin1 (F) to R15A-PP1c (○ cyan) and R15B-PP1c (▲ magenta) reconstituted on the streptavidin sensor chip surface. Representative results of three independent experiments are shown. See also Figures S1 and S2.

Figure S2

Selectivity of Compounds, Related to Figures 1 and 2 (A) Normalized steady-state binding curves from SPR showing binding of MBP (maltose binding protein, a control) (◇ orange) to bio-GBZ immobilized on the streptavidin sensor chip surface. K_D values were calculated with a steady-state affinity model by the Biacore T200 analysis software (Biaevaluation Version 1.0) by plotting equilibrium response units against protein concentration. For comparing the results the steady-state binding curves were normalized against R_max (maximum binding capacity of the surface based on the respective steady-state curve). (B, C, D, and F) Response units plotted against compound concentration showed no binding of GBZ (B), Sephin1 (C), Raphin1 (D) or compound C3 (F) to PP1c (■) immobilized on the streptavidin sensor chip surface. (E) Raphin1 does not inhibit PP1c. Dephosphorylation of difluoro-4-methylumbelliferyl phosphate by PP1c is inhibited by Calyculin A but not by GBZ, Sephin1 or Raphin1. Representative results of three independent experiments are shown in each panel.

Figure 2

SPR-Based Screening with Reconstituted Holophosphatases Identifies Raphin1 (A) Overview of the screening strategy. Numbers in dark gray squares refer to the number of compounds screened and selected for the following step; numbers in dashed squares show the number of excluded compounds. The arrow represents the experimental assay, and diamonds represent analysis steps. During the data analysis, compounds were rejected based on sensorgram quality or no binding to either holophosphatase. (B) Results of the SPR screening showing binding affinities of 54 compounds to R15A-PP1c versus R15B-PP1c, plotted as pK_D values (negative logarithm of the dissociation constant). The 15 compounds that did not bind R15A-PP1c nor R15B-PP1c were not represented on the graph. To depict compounds only binding to R15A-PP1c on the graph, we arbitrarily set the dissociation constant for R15B-PP1c to 25 μM (the highest concentration used in the screen). (C) Normalized SPR steady-state binding curves showing binding of Raphin1 to R15A-PP1c (○ cyan) and R15B-PP1c (▲ magenta) reconstituted on the streptavidin sensor chip surface. (D) Response units plotted against compound concentration showed no binding of compound C3 to R15A-PP1c (○) or R15B-PP1c (▲) reconstituted on the streptavidin sensor chip surface. Representative results of three independent experiments are shown. See also Figure S2.

Figure S3

Guanabenz Has No Measurable Effects on eIF2α Phosphorylation and on the Rates of Protein Synthesis in Unstressed Cells, Related to Figure 3 (A) Immunoblots of the indicated proteins in HeLa cells lysates treated with GBZ at 10 μM for the indicated time. (B) Upper panel: Autoradiogram of newly synthesized proteins radiolabeled with ³⁵S-methionine from HeLa cells lysates treated with GBZ at 10 μM for the indicated time. Lower panel: Coomassie staining of the gel. Representative results of three independent experiments are shown.

Figure 3

Raphin1 Inhibits R15B in Cells, Inducing a Transient Increase of eIF2α Phosphorylation and Attenuation of Protein Synthesis (A, C, E, and G) Top: immunoblots of the indicated proteins in HeLa (A and C), R15a −/− (E), or R15b −/− (G) cells lysates treated with the indicated compounds at 10 μM for the indicated time. Bottom: quantifications of eIF2α phosphorylation in immunoblots as shown above. Data are means ± SEM; n = 3. ^∗p < 0.05; ^∗∗p < 0.01 by an unpaired two-tailed Student’s t test in comparison to 0 hr time point. ns, not significant. (B, D, F, and H) Upper panel: autoradiogram of newly synthesized proteins radiolabeled with ³⁵S-methionine in HeLa (B and D), R15a −/− (F), or R15b −/− (H) cells treated with the indicated compounds at 10 μM for the indicated time. Lower panel: Coomassie-stained gel. Representative results of three independent experiments are shown. (I) Cartoon illustrating the activity of Raphin1. See also Figures S3 and S4.

Figure S4

Effects of Raphin1 at 10 or 20 μM, Related to Figure 3 (A) Measurement of Raphin1 stability in cell culture media over time at 37°C. Data are means ± SEM, n = 2. (B and C) Immunoblots (top) of the indicated proteins in HeLa cells lysates treated with Raphin1 at 10 (B) or 20 μM (C) for the indicated time. Representative results of four independent experiments are shown. Quantifications (bottom) of eIF2α phosphorylation in immunoblots such as shown above. Data are means ± SEM, n = 4. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001 by unpaired two-tailed Student t test in comparison to 0 hr time point. ns, not significant. (D and E) Upper panel: Autoradiogram of newly synthesized proteins radiolabeled with ³⁵S-methionine in HeLa cells treated with Raphin1 at 10 (D) or 20 μM (E) for the indicated time. Lower panel: Coomassie-stained gel. Representative results of three independent experiments are shown. (F) HeLa cells were plated in a 96-well plate and treated with indicated concentrations of Raphin1 in the presence of CellTox Green Dye (Promega). Cell confluency and green fluorescence (representing dead or dying cells) was measured as a function of time using the IncuCyte ZOOM system (Essen BioScience). Data is expressed as % of dead cells (described in the STAR Methods). Representative results of three independent experiments are shown. Each data point represents means ± SEM.

Figure 4

Raphin1 Inhibits Recombinant R15B (A–C) Coomassie-stained gels showing limited trypsin (5 nM) digestion of R15A and R15B in the presence of Guanabenz (A), Sephin1 (B), Raphin1 (C), or DMSO (vehicle) carried out for 5 min at 22°C. (D) Immunoblots of P-eIF2α and eIF2α following a dephosphorylation reaction by R15-PP1c holoenzymes in the presence of Raphin1. (E) Immunoblots of P-eIF2α and eIF2α following a dephosphorylation reaction by R15 chimera-PP1c holoenzymes in the presence of Raphin1. (F) Immunoblots of P-eIF2α binding to MBP-tagged R15B^340–698 or R15A^325–636 in the presence or absence of Raphin1. Representative results of three independent experiments are shown.

Figure 5

Raphin1 Promotes Proteasome- and p97-Dependent Degradation of R15B (A, B, and D) Immunoblots of the indicated proteins in HeLa cells lysates treated with Raphin1 at 10 μM, in the absence (A) or presence of the proteasome inhibitor MG-132 at 10 μM (B) or the p97 inhibitors NSM-873 or CB-5083 at 1 μM (D) for the indicated time. Representative results of three independent experiments are shown. (C) Quantifications of immunoblots corresponding to experiments shown in (B). Data are means ± SEM, ^∗∗∗∗p < 0.0001 by two-way ANOVA. ns, not significant. (E) Quantifications of immunoblots corresponding to experiments shown in (D). Data are means ± SEM, n = 3. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001 by two-way ANOVA. ns, not significant. (F) A proposed model depicting Raphin1 mechanism of action: Raphin1 binds to R15B and induces a conformational change that in cells results in its degradation in a proteasome- and p97-dependent manner.

Figure S5

Raphin1 Has No Measurable Adrenergic Activity in Cells, Related to Figure 6 α-2 adrenergic activity measurement in cells expressing human recombinant receptor following GBZ or Raphin1 treatment. Data are means ± SEM, n = 3 per group.

Figure 6

No Adverse Effects of Raphin1 Treatment on Body Weight Gain, Pancreatic and Liver Function, or Memory in Mice (A) Concentration of Raphin1 in the brain and plasma at the indicated time after a single oral administration of Raphin1 at 2 mg/kg. Data are means ± SEM; n = 3. (B) Total body weight gain of wild-type mice treated with Raphin1 at 2 mg/kg or vehicle for 10 weeks. Data are means ± SEM; n = 10 per group. (C) Glucose tolerance test on wild-type mice after treatment with Raphin1 at 2 mg/kg or vehicle for 8 weeks. Data are means ± SEM; n = 8 (vehicle) and n = 9 (Raphin1). (D) Oil Red O staining of liver in wild-type mice treated with Raphin1 or CHX at 40 mg/kg or vehicle. Scale bar, 10 μm. (E) Quadrant occupancy after training and removal of the platform in the Morris water maze of wild-type mice treated with Raphin1 at 2 mg/kg or vehicle for 2 weeks. Data are means ± SEM, n = 9 (vehicle) and n = 10 (Raphin1). (F and G) Response to fear conditioning—freezing to context (F) or freezing to cue (G)—of wild-type mice treated with Raphin1 at 2 mg/kg or vehicle for 3 weeks. Data are means ± SEM; n = 10 per group. There were no significant differences between Raphin1- and vehicle-treated mice as revealed by an unpaired two-tailed Student’s t test (B, C, E, F, and G). See also Figures S5 and S6 and Videos S1, S2, and S3.

Figure S6

Raphin1 Has No Adverse Effect on Body Weight Gain in Mice, Does Not Cause Liver Steatosis, or Affects Memory, Related to Figure 6 (A) Body weight gain of wild-type mice treated with Raphin1 at the indicated concentration or vehicle for 15 days. Data are means ± SEM, n = 3 per group. (B) Oil Red O staining of liver in wild-type mice treated with vehicle or Raphin1 at 2 mg/kg for 10 weeks. Scale bar, 10 μm. (C and D) Performance in the learning phase of Morris Water Maze of mice treated with Raphin1 at 2 mg/kg or vehicle for 2 weeks. Data are means ± SEM, n = 9 (vehicle) or n = 10 (Raphin1). Parameters measured: Distance (C) and latency (D) to locate a hidden platform in training sessions for 5 consecutive days. (E) Performance in the conditioning phase of fear conditioning in mice treated with Raphin1 at 2 mg/kg or vehicle for 3 weeks. Data are means ± SEM, n = 10 per group. Parameter measured: Freezing response during the conditioning session, where a light/tone [conditioned stimulus – CS] and a foot shock [aversive unconditioned stimulus – US] were applied. There were no significant differences between Raphin1 and vehicle treated mice as revealed by the unpaired two-tailed Student t test (A, C, D, E).

Figure 7

Raphin1 Normalizes Weight and Reduces Accumulation of SDS-Resistant Huntingtin Assemblies and Inclusions in Mutant Huntingtin Transgenic Mice (A) R15a and Chop mRNA levels (qPCR) in affected tissues from symptomatic CMT-1B or symptomatic HD^82Q mice relative to their wild-type littermates. Data are means ± SEM; n = 5 (CMT-1B) and n = 6 (HD^82Q). ^∗p < 0.05 by unpaired two-tailed Student’s t test. ns, not significant. (B) Body weight gain of wild-type and HD^82Q mice treated with Raphin1 at 2 mg/kg or vehicle once a day. Data are means ± SEM, n = 22, 24, 23, and 21 for WT vehicle, WT Raphin1, HD^82Q vehicle, and HD^82Q Raphin1, respectively. ^∗∗p < 0.01; ^∗∗∗∗p < 0.0001 by two-way ANOVA with Tukey’s multiple comparisons test. ns, not significant. (C) Immunoblots of cortex lysates of 2-month-old wild-type and HD^82Q mice treated with Raphin1 at 2 mg/kg or vehicle from 4 weeks of age daily. 2B4 antibody revealed SDS-resistant huntingtin assemblies, and 1C2 detected SDS-soluble huntingtin. (D) Quantifications of huntingtin SDS-resistant assemblies from immunoblots such as (C) from 2-month-old mice following treatment with Raphin1 or vehicle from 4 weeks of age. Data are means ± SEM; n = 6. ^∗p < 0.05 by an unpaired two-tailed Student t’s test. (E) Representative images of nuclear huntingtin inclusions revealed with 2B4 antibody (green) in the cortex of 2-month-old HD^82Q mice treated with Raphin1 at 2 mg/kg or vehicle from 4 weeks of age daily. Nuclei, Hoechst 33342 (blue). Scale bar, 10 μm. (F) Quantification of nuclear huntingtin inclusions (see STAR Methods) revealed with 2B4 antibody in the cortex of 2-month-old mice following treatment with Raphin1 or vehicle from 4 weeks of age. Data are means ± SEM; n = 6. ^∗p < 0.05 by an unpaired two-tailed Student’s t test. (G and H) Upper panel: autoradiogram of newly synthesized proteins radiolabelled with 75% ³⁵S-methionine + 25% ³⁵S-cysteine in brains of wild-type (G) or R15a −/− (H) mice treated with Raphin1 (40 mg/kg) for the indicated time. Lower panel: Coomassie-stained gel. (I) Cartoon depicting the platform of assays to enable the target-based discovery of holophosphatase inhibitors. A platform of biophysical, biochemical, and cell-based assays was used to identify selective holophosphatase inhibitors. An SPR screen identifies compounds (hits) binding to a target holophosphatase and an SPR counter screen performed with a different holophosphotase filters for selective binders. For selectivity, a filter of dissociation constant of at least 5-fold for one holophosphatase over the other is set. Hits from the primary binding screen are validated in cellular assays measuring target and pathway engagement. On-target inhibitors (lead) lose activity in cells knocked out for the target. Biochemical assays (optional) examine the mechanism of action and further confirm selectivity of the compounds. In vivo efficacy studies were carried out with validated compounds. See also Figure S7.

Figure S7

Raphin1 Is Beneficial in HD^82Q Mice, Related to Figure 7 Note that the results shown here were obtained with a separate cohort than the ones shown in Figures 7B, 7D, and 7F. (A) Total body weight gain of wild-type and HD^82Q mice treated orally with Raphin1 at 2 mg/kg or vehicle once a day for four weeks daily. Data are means ± SEM, n = 27, 26, 19, 21 for WT Vehicle, WT Raphin1, HD^82Q Vehicle and HD^82Q Raphin1, respectively. ^∗∗p < 0.01 by two-way ANOVA with Tukey’s multiple comparisons test. (B) Quantifications of huntingtin assemblies from immunoblots such as (Figure 7C) performed on cortex lysates from 2.5-month-old mice following treatment with Raphin1 or vehicle from 4 weeks of age. Data are means ± SEM, n = 3. ^∗p < 0.05 by unpaired two-tailed Student t test. (C) Quantifications of nuclear huntingtin inclusions (see STAR Methods) revealed with 2B4 antibody in the cortex of 2.5-month-old mice following treatment with Raphin1 or vehicle from 4 weeks of age. Data are means ± SEM, n = 7 (vehicle) and n = 10 (Raphin1). ^∗p < 0.05 by unpaired two-tailed Student t test.