2-Sulfonylpyrimidines: Mild alkylating agents with anticancer activity toward p53-compromised cells

Significance

Cancers with mutant p53 often show increased metastasis, genomic instability, and higher chemoresistance. The development of drugs targeting tumors with mutant p53 background is a current strategy for anticancer therapy. We found that certain activated electrophilic 2-sulfonylpyrimidines are a new class of thiol-reactive anticancer agents. These agents are especially effective in killing cancer cells with mutant or inactivated p53 or impaired reactive oxygen species detoxification and have relatively low cytotoxicity toward normal cells; they are mild electrophiles, some of which will, for example, stabilize mutant p53 by selective targeting of its thiol groups and have little general alkylating reactivity.

Abstract

The tumor suppressor p53 has the most frequently mutated gene in human cancers. Many of p53’s oncogenic mutants are just destabilized and rapidly aggregate, and are targets for stabilization by drugs. We found certain 2-sulfonylpyrimidines, including one named PK11007, to be mild thiol alkylators with anticancer activity in several cell lines, especially those with mutationally compromised p53. PK11007 acted by two routes: p53 dependent and p53 independent. PK11007 stabilized p53 in vitro via selective alkylation of two surface-exposed cysteines without compromising its DNA binding activity. Unstable p53 was reactivated by PK11007 in some cancer cell lines, leading to up-regulation of p53 target genes such as p21 and PUMA. More generally, there was cell death that was independent of p53 but dependent on glutathione depletion and associated with highly elevated levels of reactive oxygen species and induction of endoplasmic reticulum (ER) stress, as also found for the anticancer agent PRIMA-1^MET(APR-246). PK11007 may be a lead for anticancer drugs that target cells with nonfunctional p53 or impaired reactive oxygen species (ROS) detoxification in a wide variety of mutant p53 cells.

The tumor suppressor p53 plays a key role in regulating cell cycle arrest, DNA repair, apoptosis, and cell senescence and is inactivated either by mutation or up-regulation of proteins such as MDM2 or MDMX in virtually all human cancers (1–4). The function of p53 may be impaired by mutation of residues involved directly in DNA binding (contact mutants) or mutations elsewhere in the DNA-binding domain (DBD) that destabilize it (structural mutants) (5). Reactivation of oncogenic mutants of p53 is a target in cancer therapy (3, 6). The destabilized p53 cancer mutant Y220C, for example, can be reactivated both in vitro and in cancer cells by small molecules that bind to a mutationally induced crevice on the surface of the protein and stabilize it (7–9). More generally, the compounds PRIMA-1 and APR-246, also known as PRIMA-1^MET, preferentially induce growth suppression and apoptosis in cancer cells harboring either contact or destabilized mutants of p53 (10, 11). Both compounds decompose, however, into the bioactive methylene quinuclidinone (MQ), which contains a Michael acceptor group that reacts with nucleophilic thiols through a 1,4-addition (12). The nucleophilic –SH of cysteine can be selectively modified by various electrophiles, such as maleimides, iodoacetamides, α-halocarbonyls, Michael acceptors, activated thiols, and methane/phenylthiosulfonates (13). For a number of other thiol reactive compounds, such as the maleimide MIRA-3, or STIMA-1, similar biological effects were observed (11, 14). Despite their common reactivity toward thiols, the molecular basis for the observed biological effects of these compounds in cells is not completely clear and may be multifactorial (6).

We have discovered here a small 2-sulfonylpyrimidine molecule, PK11000, which stabilizes the DBD of both WT and mutant p53 proteins by covalent cysteine modification, without compromising DNA binding. These compounds were both mild and selective alkylating agents. We analyzed the thiol reactivity and mechanism of p53 stabilization of 2-sulfonylpyrimidines and evaluated their effects in different cancer cell lines, and found that they generally exert strong anticancer activity toward p53-compromised cells, involving up-regulation of p53 target genes and a strong increase in cellular reactive oxygen species (ROS) levels.

Results

Indications of Covalent Modification.

We screened an in-house fragment library targeted to bind in the mutationally induced cavity in the p53 cancer mutant Y220C using differential scanning fluorimetry (DSF) to identify ligands that increase its thermal stability (15, 16). The best hit, 5-chloro-2-methanesulfonylpyrimidine-4-carboxylic acid (PK11000), raised the protein melting temperature (T_m) of the stabilized p53-Y220C DBD (T-p53C-Y220C) (9) by ΔT_m > 1.2 K (Fig. 1A). We suspected from the kinetics of inhibition of aggregation measured by light scattering (Fig. 1B), however, that PK1100 was binding to the protein covalently, and not noncovalently as designed. ¹⁵N-¹H heteronuclear single quantum coherence (HSQC) NMR spectra of ¹⁵N-labeled T-p53C-Y220C (94–312) with PK11000 concentrations ranging from 218 to 1,000 µM confirmed that PK11000 binds to the DNA-binding domain (Fig. S1). The observed peak shifts did not change in a concentration-dependent manner, indicating either slow or intermediate exchange-like behavior or covalent modification. Interestingly, PK11000 mainly induced peak shifts of residues in direct proximity of Cys124, Cys182, and Cys277 (Fig. 1C). The chemical shifts of several residues within the helix 1 and loop 1 region (including Cys124 and Cys277) were particularly large, indicating a strong effect of PK11000 binding on the chemical environment of this region (Fig. S1). A less-pronounced effect of PK11000 was observed for the region around Cys182. PK11000 raised the T_m of full-length p53 that did not contain the Y220C mutation by 1.5 K at 1 mM, confirming that the stabilization was not by binding to the mutant cavity.

Fig. 1.

PK11000 bound to and stabilized p53 DBD. (A) Melting curve of the stabilized p53-Y220C DBD (T-p53C-Y220C) recorded via differential scanning fluorimetry in absence (blue) or in presence of 1 mM PK11000 (magenta). (B) Inhibition of T-p53C-Y220C aggregation by PK11000. Aggregation kinetics were recorded by monitoring light scattering at 500 nm with 3 μM protein at 37 °C in standard assay buffer [25 mM KPi pH 7.2, 150 mM NaCl, 1 mM TCEP, and 5% (vol/vol) DMSO]. Unlike for noncovalent binders that inhibit aggregation where the amplitude of the reaction remains unchanged and just the rate constants being lower, the amplitude of light scattering was substantially decreased at 250, 500, and 1,000 µM of PK11000, suggesting the presence of two species: unmodified p53-Y220C and a covalent adduct of PK11000 and p53. (C) Mapping of PK11000 induced peak shifts in the ¹⁵N-¹H HSQC NMR spectrum onto the structure of the p53-Y220C DBD. Large peak shifts are highlighted in red; medium shifts in orange; and small shifts in yellow.

Fig. S1.

DSF ΔT_m values of different p53 mutants (8 μM) after incubation with diverse 2-sulfonylpyrimidine compounds (250 μM) for 30 min.

PK11000 Alkylates Two Cysteine Residues of p53.

We identified covalent modification of cysteines in p53 using electrospray ionization (ESI) mass spectrometry experiments. This covalent modification was unexpected because electrophilic reactivity of this type of compound under mild aqueous conditions has not been noted previously, although amines have been reported to react with 2-sulfonylpyrimidines at high concentrations in dimethyl sulfoxide (17). A nucleophilic aromatic substitution (SNAr) reaction between PK11000 and a cysteine should lead to elimination of methyl sulfinic acid and an increase in the protein mass by 156.5 Da (Fig. 2A). Incubation of 50 µM T-p53C-Y220C with various PK11000 concentrations for 4 h at 20 °C yielded specific mass increases of either 157 Da or 2 × 157 Da (Fig. 2B), confirming the proposed SNAr reaction mechanism. At low PK11000 concentrations (250 µM) with a 5:1 compound:protein ratio, we observed a mixture of mono and dialkylated proteins. The maximum number of modified cysteines never exceeded two, even at a PK11000 concentration as high as 5 mM. Incubation of stabilized full-length p53 (T-p53) with 2 mM PK11000 also yielded exactly two modifications. T-p53C-Y220C contains six at least partly solvent-accessible cysteines (plus three zinc coordinating and two buried cysteines) (18). In contrast, the active form of APR-246, MQ, alkylates up to five cysteines of the R175H mutant at 2 mM concentration, and up to nine cysteines with 5 mM MQ (12). This selectivity of PK11000 at neutral pH conditions for specific p53 cysteines suggests that 2-sulfonylpyrimidines react under biologically ambient conditions only with SNAr-accessible and highly nucleophilic cysteines and are, therefore, useful tools for selective chemical protein modification.

Fig. 2.

PK11000 alkylated cysteines of p53 DBD. (A) SNAr reaction mechanism for PK11000 cysteine alkylation. (B) ESI (ES+) mass spectra of 50 µM T-p53C Y220C incubated for 4 h at 20 °C with no compound (black) or 250, 500, 1,000, and 5,000 µM PK11000 (red).

Specific Modification of Cys182 and Cys277 by PK11000 Increases Protein Stability Without Compromising DNA-Binding Affinity.

HSQC NMR data suggested three potential candidates for alkylation by PK11000: Cys124, Cys182, and Cys277 (Fig. S1). To determine which two of these three residues undergo alkylation, we monitored the effect of PK11000 on Cys-to-Ser mutants C124S/C182S, C124S/C277S, C182S/C277S, and C124S/C182S/C277S by ESI mass spectrometry. Treatment with PK11000 yielded no modification for the triple-mutant C124S/C182S/C277S and the double-mutant C182S/C277S, but exactly one modification for the mutants C124S/C182S and C124S/C277S (Fig. 3A), thus confirming specific alkylation of Cys182 and Cys277 by PK11000. Accordingly, Cys182 and Cys277 are the most reactive nucleophiles on the p53 DBD for SNAr reactions with 2-sulfonylpyrimidines.

Fig. 3.

ESI (ES+) mass spectra of different cysteine to p53 core DBD serine mutants after incubation without (black) and with PK11000 (red). (A) T-p53C C124S/C182S/C277S and (D) T-p53C C182S/C277S mutants showed no cysteine modification by PK11000, whereas (B) T-p53C C124S/C182S and (C) T-p53C C124S/C277S showed one cysteine modification by PK11000.

The mass spectrometry data also agreed well with DSF measurements of the effects of PK11000 on p53 protein stability (Fig. S2). Though 30 min incubation of the p53 DBD with 250 μM PK11000 increased the melting temperature by ∼3 °C, the compound had no effect on the stability of the C182S/C277S mutant. Reaction of the C124S/C182S and C124S/C277S mutants with PK11000 showed alkylation of Cys277 (ΔT_m = 3.6 K) had a stronger stabilizing effect than that of Cys182 (ΔT_m = 1.2 K).

Fig. S2.

¹⁵N-¹H HSQC NMR spectrum of the p53 Y220C core domain (red) with 1,000 µM (blue), 436 µM (yellow), and 218 µM (green) PK11000 at 293 K.

Cys277 forms weak polar interactions with bases in the major groove of bound DNA (19, 20). However, incubation of T-p53 with 1 mM PK11000 or PK11007 and PK11010, two structural analogs with larger ring substituents (see Fig. S2 for chemical formulas), for 2 h had little effect on the binding of p53–GADD45a, with K_d ranging from 26 to 36 nM vs. 31 nM for the untreated protein (Fig. S3). Alkylation of p53 by 2-sulfonylpyrimidine derivatives did not alter its DNA-binding capability.

Fig. S3.

Alkylation of full-length p53 with 2-sulfonylpyrimdines does not compromise its DNA binding capability. Stabilized full-length p53 was incubated with 1 mM PK11000, PK11007, PK11010, and 5% DMSO for 2 h at 4 °C and then titrated into a cuvette containing 20 nM carboxyfluoresceine-labeled Gadd45a DNA. Fluorescence polarization data were recorded and fitted to the Hill equation including a linear drift term.

To check the more general reactivity of 2-sulfonylpyrimidines, we incubated 50 µM of bacterial N-acetylneuraminate lyase, with 2 mM PK11000. The tetramer contains four cysteines, one of which is located on the surface of the tetramer but not fully solvent exposed (21). We observed no chemical protein modification, showing that this compound has low general reactivity and modifies only particularly reactive cysteine residues.

Structural Effects of Cys182 Modification.

We soaked crystals of the Y220C DBD with PK11000 and determined the structure of the complex at 1.42 Å resolution. Cys182 and Cys277 are freely accessible in the crystal lattice and, after initial refinement of the model, the difference density map showed positive density at these cysteines. For Cys277 the density was not clear enough to model unambiguously the modification, whereas for Cys182 there was conclusive electron density for a chloropyrimidine moiety covalently linked to the sulfur (Fig. 4). The carboxylate group was not well-resolved, suggesting ring flipping. The modification forms only few interactions with the rest of the protein, primarily via His178. Interestingly, however, modification of Cys182 affected the conformational equilibrium of this region. In the unmodified form, Cys182 adopts two alternative conformations. Upon modification, however, only one side-chain conformation was observed, and there was a change in the backbone conformation of residues 182–184.

Fig. 4.

Structural effect of Cys182 modification by PK11000. Superposition of the structure of the p53 cancer mutant Y220C with (gray) and without (green) PK11000 shows that Cys182 on the surface of the L2 loop is modified by the alkylating agent, with the covalent modification pointing toward the solvent. Chain A of the asymmetric unit is shown. Alkylation fixes Cys182 in a defined conformation, but there is little interaction between the modification and the rest of the protein. The figure was generated using PyMOL (www.pymol.org).

Reactivity and p53 Stabilizing Effects of 2-Sulfonylpyrimidine Derivatives.

We measured the rate of reaction of PK11000 with glutathione (GSH) by ¹H-NMR spectroscopy at 20 °C at equimolar concentrations of reagents. The second-order rate constant was 1.37 L⋅mol⁻¹⋅s⁻¹ (Fig. 5), 1,000, 100, and 10 times lower, respectively, than for the reported Michael acceptors, 1-penten-3-one, methyl propiolate, and methyl acrylate with GSH (22). The GSH reactivity of the acrylamide moiety, which is present in the FDA-approved anticancer drugs ibrutinib and canertinib, is seven times less reactive than PK11000 (23). Considering the clear correlation of compound–GSH reactivity with toxicity (22, 24, 25), the reactivity of the PK11000 is close to the level of therapeutically applied thiol reactive agents.

Fig. 5.

¹H-NMR kinetic measurements of PK11000-glutathione adduct formation. Adduct concentrations were calculated by integration of an aromatic product peak at 8.52 ppm. Data were fitted to a second-order reaction equation for equimolar educt concentrations (65).

We then examined the effects of analogs of PK11000 (Fig. 6A) on the thermal stability of p53 after 15-, 30-, 60-, and 120-min incubation. There was increasing stabilization of T-p53C-Y220C with PK11000 over time, rising to a maximum ΔT_m of ∼2.5 K (Fig. 6B). Substituting the 4-chloro group of PK11000 with bromine (PK11002) yielded slightly higher p53 T_m shifts of ∼3 K after 60-min incubation. The largest T_m shifts (>3 K) were observed for PK11007 and PK11010, which also appeared to react faster than other 2-sulfonylpyrimidines, because they reached their maximum effect after only 15 min. These two compounds share an electron withdrawing 4-N-(5-methyl/ethyl-1,3,4-thiadiazol-2-yl)carboxamide substituent, which may increase the reaction rate. Interestingly, large, substituted acetanilide or 5-benzyl-thiadiazol-2-yl-carboxamide substituents on the pyrimidine scaffold of PK11013–15 resulted in a significant destabilization of p53 (Fig. 6C). Compounds with electron donating groups, such as 3,5 dimethyl (PK11017) or 4-hydroxyl (PK11011) substituents, had no effect on p53 protein stability. Interestingly, 2-chloro-pyrimidines, which are reported to exhibit SNAr reactivity (26), had no significant effect on the p53 melting temperature. Overall, these findings clearly show the potential for fine-tuning the reactivity of the 2-sulfonylpyrimidine core scaffold through appropriate substitutions.

Fig. 6.

Effects of diverse 2-sulfonylpyrimidines on p53 stabilization. (A) Library of diverse 2-sulfonylpyrimidines and structurally related compounds. (B and C) Time-dependent stabilization (DSF ΔT_m) of T-p53C-Y220C after 15-, 30-, 60-, and 120-min incubation at room temperature with stabilizing or destabilizing/nonreactive 2-sulfonylpyrimidine compounds, respectively.

Mutant p53 Cancer Cells Were Extra Sensitive to PK11007 Treatment.

Given the well-studied anticancer effects of thiol reactive agents, such as PRIMA-1, APR-246, and MIRA-3 (10–12), we examined the effects of 2-sulfonylpyrimdines on cancer cells. It is unlikely that biological effects in cancer cells are exclusively caused by modification of only one specific protein such as p53. A multitude of other thiol-containing molecules, such as the abundant intracellular antioxidant GSH and proteins that possess sufficiently nucleophilic and surface-accessible cysteines are likely to undergo modification, which may lead to a complex perturbation of various cellular processes and pathways.

We measured cell viability of four p53 wild-type cell lines (WI-38, HUH-6, NUGC-4, SJSA-1) and four p53 mutant cell lines (HUH-7, NUGC-3, SW480, MKN1) after 24 h incubation with PK11007 (see Table S1 for a detailed description of the tested cell lines). There was a large viability reduction in mutant p53 cell lines MKN1 (V143A), HUH-7 (Y220C), NUGC-3 (Y220C), and SW480 (R273H/P309S) at concentrations ranging from 15 to 30 µM (Fig. 7A). The p53 WT cancer cell lines HUH-6 and NUGC-4, as well as the fibroblast cell line WI-38, were less sensitive to PK11007, with reduced cell viability only at high concentrations of compound (60 and 120 µM). Interestingly, the p53 WT cell line SJSA-1, which is known for its high intracellular MDM2 levels, was as sensitive to PK11007 as the tested mutant p53-containing cancer cells. Compound-mediated cell death of the mutant p53-containing cells HUH-7 and MKN1 was already observed after 3 or 6 h, respectively, whereas viability decrease of the p53 WT cell line HUH-6 happened only after 10 h (Fig. S4). There was a stronger selectivity for the p53 mutant cell lines also for other 2-sulfonylpyrimidines, such as PK11010 and PK11029, albeit only at higher compound concentrations (Fig. S5).

Table S1.

Description of cell lines

Cell line	p53 status	Organism	Tissue type	Disease	ATCC/JCRB code
WI-38	WT	Human	Lung	—	CCL-75
HUH-6	WT	Human	Liver	Hepatoblastoma	JCRB0401
NUGC-4	WT	Human	Stomach	Gastric adenocarcinoma	JCRB0834
SJSA-1	WT	Human	Fibroblast (lung)	Osteosarcoma	CRL-2098
HCT-116	WT	Human	Colon	Colorectal carcinoma	CCL-247
SW480	R273H-P309S	Human	Colon	Dukes’ type B, colorectal adenocarcinoma	CCL-228
HUH-7	Y220C	Human	Liver	Hepatocellular carcinoma	JCRB0403
NUGC-3	Y220C	Human	Stomach	Gastric adenocarcinoma	JCRB0822
MKN1	V143A	Human	Stomach	Adenosquamous carcinoma	JCRB0252
H1299	Null	Human	Lung	Non–small cell lung cancer	CRL-5803

ATCC, American Type Culture Collection; JCRB, Japanese Collection of Research Bioresources.

Fig. 7.

Biological effects of PK11007 on diverse cancer cell lines and one human fibroblast cell line. (A) Concentration-dependent viability reduction of WI-38 (fibroblast), NUGC-4 (p53 WT), HUH-6 (p53 WT), SJSA-1 (p53 WT), SW480 (p53 R273H/P309S), NUGC-3 (p53 Y220C), HUH-7 (p53 Y220C), and MKN1 (p53 V143A) cells after treatment with PK11007 for 24 h. The mutant p53 cells HUH-7, MKN1, and NUGC-3 were more sensitive to PK11007 treatment as indicated by strong viability reduction at low compound concentrations. Cell viability was measured in quadruplicate and normalized with the values of blank (viability = 1) and no cell (viability = 0) controls. Data are shown as mean ± SEM. (B) Incubation of the isogenic H1299 (p53^−/−), H1299 (p53 H175), HCT116, and HCT116 p53^−/− cancer cells with PK11007 yielded a comparable viability reduction after 24 h. (C) Cell viability of HUH-6, HUH-7, and MKN1 after p53 knockdown via siRNA. Cells were treated with PK11007 for 24 h (48 h for 15 µM PK11007 MKN1 sample). In HUH-6 and HUH-7, cell death was independent of p53, whereas it was partially dependent on p53 in MKN1 cells. (D) Western blots of NUGC-4, NUGC-3, MKN1, HUH-6, and HUH-7 cancer cells after 3 h (6 h for MKN1) treatment with PK11007. p53 target genes p21, PUMA, and MDM2 were up-regulated not only in MKN1 (p53-V143A), NUGC-3, and HUH-7 (both p53-Y220C), but also in HUH-6 and NUGC-4 (both p53 WT) cells. With increasing PK11007 concentration, the molecular weight of p53 gradually increased to ∼3 kDa for the mutant p53 cell lines (HUH-7, MKN1, and NUGC-3), suggesting hyperalkylation of unfolded/aggregated p53. The protein level of the antioxidative gene GSTP1 was higher in NUGC-4 (p53 WT) than in NUGC-3. The asterisks at the p53 target gene levels of HUH-6 and HUH-7 cells highlight the genes for which a different β-actin control was used. (E) Quantification of relative mRNA levels of p53 target genes via real-time PCR. Cells were treated with DMSO or 15 µM (MKN1, HUH-7) or 20 µM (NUGC-3, HUH-6) PK11007 for 6 h (4.5 h for HUH-7). p21 and PUMA mRNA levels were especially up-regulated in mutant p53 cells, whereas in p53 WT cells (HUH-6) no increased transcription was observed. Significance levels were calculated with the Student t test (***P < 0.001; **P < 0.01; *P < 0.05). (F) Western blots of protein levels of UPR key markers in MKN1, HUH-6, and HUH-7 cells. PK11007 treatment for 3 h (6 h for MKN1) increased levels of spliced XBP-1 and CHOP (especially in HUH-7). (G) Inhibition of cellular glutathione synthesis by buthionine sulfoximine strongly potentiated cell viability reduction by PK11007 in HUH-7, NUGC-3, and MKN1 mutant p53 cancer cells. (H) Determination of relative intracellular ROS levels via CellROX Deep Red fluorescence after incubating four cancer cell lines with 30 or 60 µM PK11007 for 2 h. PK11007 caused increase of ROS in all tested cell lines. However, at high doses, the increase of relative ROS levels was higher in HUH-7, NUGC-3, and MKN1 cells. Median fluorescence levels were determined in triplicate with error bars depicting the SE. Significance levels were calculated using a one-way ANOVA with the Bonferroni post hoc test for mean comparison (***P < 0.001; **P < 0.01; *P < 0.05).

Fig. S4.

Cell viability time course for (A) HUH-6 (p53 WT), HUH-7 (p53 Y220C), and (B) MKN1 cancer cells at 30 or 60 µM PK11007 (10 and 30 µM for MKN1). Cell viability was measured in quadruplicates and normalized with the values of blank (viability = 1) and no cell (viability = 0) controls. Data are shown as mean ± SEM.

Fig. S5.

Biological effects of 2-sulfonylpyrimidines on diverse cell lines. Concentration-dependent viability reduction of WI-38 (p53 WT, noncancer), NUGC-4 (p53 WT), HUH-6 (p53 WT), SJSA-1 (p53 WT), SW480 (p53 R273H/P309S), NUGC-3 (p53 Y220C), HUH-7 (p53 Y220C), and MKN1 (p53 V143A) cells after compound treatment for 24 h. The strongest effects on cell viability were observed for PK11007 and PK11012. Mutant p53 cell lines were mostly significantly more sensitive for these compounds than p53 WT and noncancer cell lines. This behavior was also observed for PK11010 and PK11029, however only at higher compound concentrations. PK11003 led to a strong viability decrease; however, it did not distinguish clearly between mutant and WT p53 cell lines. Compounds that contain a negatively charged carboxyl group, such as PK11000 and PK11009, did not lead to a significant cell viability decrease in the tested cell lines at concentrations below 120 µM. Mutant p53 cell lines were also extra sensitive to compounds that destabilized the p53 core domain in vitro, such as PK11012 and PK11015. Cell viability was measured in quadruplicate and normalized with the values of blank (viability = 1) and no cell (viability = 0) controls. Data are shown as mean ± SEM.

Additionally, we tested PK11007 with the p53^−/− cell line H1299 and with it containing p53 R175H (Fig. 7B). PK11007 strongly reduced viability in both cell lines, comparable to the sensitivity of HUH-7 and NUGC-3 cancer cells. Compared with p53-WT–containing HCT116, the isogenic HCT116 p53^−/− cell line was less sensitive (Fig. 7B). Down-regulation of p53 protein levels via siRNA did not change PK11007-mediated viability reduction in HUH-6 and HUH-7 cells (Fig. 7C), which reinforces that PK11007 can induce cell death independently of p53. In MKN1 cells (p53-V143A), however, there was a significant viability difference between control and p53 knockdown samples at 15 and 20 µM PK11007, indicating that cell death is partly dependent on mutant p53 in this cell line. APR-246 was not as effective as PK11007 in inducing cell death for the cancer cell lines tested (Fig. S6). Cell viability was reduced only at high concentrations of APR-246 (≥60 µM). Our WT and mutant p53 cell lines were similarly sensitive after 24-h treatment, unlike previous different sets of cells after longer treatment (10, 12).

Fig. S6.

Viability reduction of APR-246 in NUGC-3 (p53 WT), HUH-6 (p53 WT), NUGC-4 (p53 Y220C), and HUH-7 (p53 Y220C) cancer cells after treatment for 24 h. Cell viability was measured in quadruplicates and normalized with the values of blank (viability = 1) and no cell (viability = 0) controls. Data are shown as mean ± SEM.

PK11007 Increased Levels of p53 Target Genes and Activated the Unfolded Protein Response Pathway.

PK11007 up-regulated protein levels of the p53 target genes p21, MDM2, and PUMA in a mostly concentration-dependent manner in NUGC-3 (p53-Y220C), HUH-7 (p53-Y220C) and MKN1 (p53-V143A) cells, suggesting partial restoration of transcriptional activity to destabilized p53 mutants (Fig. 7D). The molecular weight of blotted p53 gradually increased by ∼3 kDa at 30 and 60 µM PK11007 in all mutant p53 cell lines, probably from alkylation of 11 residues in its denatured state (11 × 255 ∼ 2.8 kDa). In contrast, neither the molecular weight nor level of p53 changed in the WT p53 cell lines HUH-6 and NUGC-4, indicating a smaller amount of alkylated, denatured p53 and possibly a higher tolerance against the thiol reactivity of PK11007.

Levels of GST P (GSTP1), an important enzyme for the detoxification of hydrophobic electrophiles via conjugation with glutathione, were slightly higher in the p53 WT NUGC-4 cells than in the mutant p53 NUGC-3 cells and may be one factor for the higher resistance against thiol-reactive compounds in p53 WT cell lines. PK11007 also increased p53 activity in HUH-6 and NUGC-4 cells, as indicated by the increase of MDM2, PUMA, and p21 protein levels. At high concentrations, we observed in some cases a decrease in p53 target gene levels (e.g., p21 levels for NUGC-3 and MKN1), despite up-regulation of the same protein at lower concentrations.

PK11007 increased transcription of p53 target genes in three mutant p53 cell lines after 6-h treatment (Fig. 7E). PUMA and p21 mRNA levels were up-regulated by a factor of 2 upon treatment of NUGC-3 (20 µM PK11007), MKN1 (15 µM), and HUH-7 (15 µM) cells, as well as NOXA for the latter two. p53 mRNA levels did not change, except for a slight increase in HUH-7 cells. MDM2 levels were halved in MKN1 and NUGC-3 cells. There was no significant change in p53 target gene mRNA levels for the p53 WT cell line HUH-6 (20 µM PK11007).

Additionally, we observed activation of the unfolded protein response (UPR) pathway, which is triggered by endoplasmic reticulum (ER) stress. PK11007 up-regulated protein levels of spliced XBP-1, a key marker for UPR activation, in a concentration-dependent manner in HUH-7 cells and to a lesser degree also in HUH-6 cells. At 20 µM PK11007, spliced XBP-1 was also up-regulated in MKN1 cells (Fig. 7F). A similar observation was made for APR-246, which induced transcription of XBP-1 and splicing of its mRNA in Saos-2 and Saos-2-His273 cancer cells (27). Expression of CHOP, another UPR marker, was also increased by PK11007, especially in HUH-7 cells.

PK11007 Induced Mainly Caspase-Independent Cell Death.

We measured caspase activity and loss of membrane integrity after 6-h incubation with PK11007 in cancer cells. There was no significant activation of caspases 3 and 7 in HUH-6 and HUH-7 cells after treatment with 30 or 60 µM PK11007, suggesting that the reduction in viability of these cell lines did not result from caspase-mediated apoptosis (Fig. S7A). APR-246 treatment also did not increase that caspase activity. PK11007 treatment resulted in a loss of membrane integrity in mutant p53 HUH-7 cells but not in HUH-6 cells (Fig. S7B). A combination of 15 µM PK11007 and 100 µM buthionine sulfoximine (BSO), an inhibitor of glutathione synthesis, had a synergistic effect on cytotoxicity. APR-246 treatment did not increase membrane permeability in HUH-6 or HUH-7 cells, consistent with the relatively low viability reduction of APR-246 in these cell lines. PK11007 slightly increased caspase 3/7 activity without inducing membrane permeabilization in SW480 (15% increase at 15 µM), SJSA-1 (40% increase at 30 µM), and HCT116 (10% increase at 60 µM) cells (Fig. S7C), suggesting that it may kill cancer cells in a similar way as does APR-246, PRIMA-1, or MIRA-3, which are known to activate caspases in various cancer cell lines (10, 11, 14, 28). Although PK11007 caused an increase in caspase 3 and caspase 7 activities in several cases, reduction of viability of very sensitive cancer cell lines such as HUH-7 or MKN1 appeared to result mainly through caspase-independent pathways.

Fig. S7.

PK11007 induces cell death via caspase-independent pathways. HUH-6 and HUH-7 cells were treated for 6 h with PK11007, APR-246, and Nutlin-3. (A) PK11007 and APR-246 treatment did not significantly increase caspase 3 and caspase 7 activities, only Nutlin-3 yielded a significant caspase induction in HUH-6 cells. (B) Membrane permeability was significantly increased for high PK11007 concentrations or in combination with BSO in HUH-7 cells, whereas HUH-6 cells were not affected. APR-246 and Nutlin-3 did not significantly induce cytotoxicity. (C) PK11007 slightly increases caspase 3/7 activity in SW480 (at 15 µM), SJSA-1 (at 30 µM), and HCT116 (at 60 µM) cancer cells. All samples were measured in quadruplicate, with error bars depicting the SEM. Significance levels were calculated using a one-way ANOVA with the Bonferroni post hoc test for mean comparison (***P < 0.001; **P < 0.01; *P < 0.05).

PK11007 Viability Reduction Was Potentiated by Glutathione Depletion.

GSH is the major redox buffer in cells and is crucial for many enzymatic and nonenzymatic antioxidant reactions that decrease oxidative stress (e.g., ROS) and maintain the redox state of the cell (29). Because of its high abundance in the cell in the millimolar range and its freely accessible thiol group (30), GSH is a prime target for modification by selective thiol alkylators. APR-246–mediated growth suppression is potentiated by inhibition of GSH synthesis via BSO, an inhibitor of glutamate cysteine ligase (12). To assess whether the observed cell viability reduction for PK11007 is also enhanced by BSO, we incubated HUH-7, HUH-6, NUGC-3, NUGC-4, and MKN1 cell lines with 15 µM PK11007, 100 µM BSO, or a combination of both (Fig. 7G). BSO treatment alone did not affect viability in any cell line. Combination of PK11007 and BSO resulted in a substantially stronger viability reduction in the mutant p53 cell lines MKN1, HUH-7, and NUGC-3 than PK11007 alone. This strong synergistic effect was not observed in WT p53 HUH-6 and NUGC-4 cells.

PK11007 Induces ROS Especially in Mutant p53 Cells.

The biological effects of APR-246 may not be directly dependent on the p53 status but rather on the perturbation of the intracellular GSH/ROS balance and the respective cellular context (31–33). To test whether PK11007 also increases ROS levels, we incubated NUGC-3, NUGC-4, HUH-6, HUH-7, and MKN1 cells with PK11007 for 2 h and stained the cells with CellROX Deep Red dye, which exhibits a strong fluorescent signal upon oxidation by ROS. There were elevated ROS levels in all cell lines after 2 h. In the mutant p53 cells MKN1, HUH-7, and NUGC-3, however, the ROS increase was higher at 60 µM PK11007 than in NUGC-4 and HUH-6 cells, suggesting that the higher PK11007 sensitivity (and faster viability reduction in HUH-7) of the mutant p53 cell lines is mediated by a stronger ROS induction (Fig. 7H). Basal and PK11007-induced ROS levels in MKN1 cells were at least twofold higher than in other cell lines, which may have contributed to the strong induction of cell death at low PK11007 concentrations (Fig. 7A). Treatment of HUH-6 and HUH-7 cells with N-acetylcysteine (NAC) reduced ROS levels of untreated cells (Fig. S8), which is in line with its well-known antioxidative effect (34). NAC also prevented PK11007-mediated ROS increase in both HUH-6 and HUH-7 cells. This effect was most likely not only caused by the antioxidant effect of NAC, but also the result of adduct formation with the alkylating agent, as shown for APR-246 (12).

Fig. S8.

NAC prevents PK11007-mediated ROS formation. (A) Determination of intracellular ROS levels via CellROX Deep Red fluorescence after incubating HUH-6 and HUH-7 cancer cells with 5 mM NAC, 400 µM tert-Butyl hydroperoxide (TBHP), or a combination of 30 µM PK11007 and 5 mM NAC for 1.25 h. NAC does not only decrease basal ROS levels, but also completely prevents PK11007 from increasing ROS levels. (B) Relative intracellular ROS levels via CellROX Deep Red fluorescence after incubating four cancer cell lines with 30 or 60 µM PK11007 for 2 h. PK11007 led to an increase of ROS in all tested cell lines; however, at high doses the relative ROS level was significantly higher in HUH-7 and NUGC-3 cells, whereas the ROS levels in MKN1 cells remained on the level of the p53 WT cell lines. Median fluorescence levels were determined in triplicates. Error bars depict SEM values.

Discussion

We have identified 2-sulfonylpyrimidines as a biologically active class of selective thiol alkylators; they react rapidly and selectively with free thiol groups in neutral aqueous buffers at room temperature, allowing the introduction of specific molecular probes at solvent-exposed cysteines. Reactivity of the scaffold can be further fine-tuned by diverse substitutions that modify the electron density of the aromatic ring. Alkylation of the surface-exposed cysteines 182 and 277 by PK11007 and similar 2-sulfonylpyrimdines stabilized the p53 DBD in vitro and did not compromise its DNA-binding affinity.

PK11007 effectively killed cancer cells, especially with null or mutant p53 background, and showed only low cytotoxicity toward normal cells. Other 2-sulfonylpyrimidines were either less effective (e.g., PK11000, PK11010, and PK11029) or less selective (e.g., PK11003, PK11012, and PK11015) in inducing cell death (Fig. S5). Only C182 and C277 were readily alkylated in native protein incubated at 20 °C. Denaturation of the protein at 37 °C in the cell, which happens to a much larger extent in cells containing highly overexpressed, unstable mutant p53 (e.g., NUGC-3 MKN1, HUH-7), exposes further cysteine side chains, and up to 11 were alkylated. Immunofluorescence experiments showed that there was also a significant decrease in unfolded p53 (using Pab 240) after PK11007 treatment and some increase in folded p53 (using Pab1620) on the NUGC-3 cell line, showing an increase in the fraction of native protein (Fig. S9).

Fig. S9.

PK11007 changes the folding state of mutant p53 (Y220C) in NUGC-3 cells. PK11007 treatment led to a significant reduction of unfolded p53 (Pab 240) and a slight increase in folded p53 (Pab 1620). Hoechst 33342 dye was used to stain the nucleus. To facilitate comparison of Pab 1620 and Pab 240 fluorescence levels, we increased exposure of all raw pictures by 2 eV.

PK11007 increased protein and mRNA levels of the p53 target genes p21 and PUMA, especially in mutant p53 cell lines, suggesting partial reactivation of mutant p53 and its transcriptional activity. PUMA and p21 protein levels were up-regulated also in p53 WT cell lines NUGC-4 and HUH-6; however, at low compound concentrations, PUMA and p21 mRNA levels did not increase in HUH-6 cells, which may explain their higher resistance to PK11007. Ultimately, these results suggest that up-regulation of PUMA, p21, and NOXA may not be exclusively mediated by mutant p53 reactivation.

Although PK11007 up-regulated diverse p53 target genes and was clearly more effective in mutant p53 or null p53 cancer cells, cell death was in some cases independent of p53, as shown by p53 knockdown in HUH-6 and HUH-7 cells and the comparable viability reduction in isogenic H1299 and H1299-R175H cells. However, PK11007 induced cell death in a partially mutant p53-dependent manner in MKN1 cells. The difference in sensitivity between p53 knockdown and control was comparable to that reported after treatment of the mutant p53 cancer cell lines DMS273 and DMS53 with APR-246 (35). APR-246, which also kills mutant p53 cancer cells more effectively, either induces cancer cell death in a mutant p53-dependent manner (31, 35) or works exclusively via p53-independent mechanisms (33). These findings suggest that thiol and p53 alkylating compounds like PK11007 and APR-246 exert their anticancer effects via diverse mechanisms, depending on the respective genetic background and cellular context of cancers.

There was only a small difference in reduction of viability between the isogenic cell lines HCT116 and HCT116 p53^−/−, which lacks expression of full-length p53 (36), suggesting that cellular effects of PK11007 are mainly independent of p53 in this cell line. According to the catalog of somatic mutations in cancer (COSMIC) in the canSAR database (37, 38), HCT116 contains several missense or frameshift mutations in important antioxidative genes such as thioredoxin reductase 1 (TXNRD1), which is also inhibited by APR-246 and is an important gene for its cellular efficacy (39), and other thioredoxin containing proteins, GST alpha 2 (GSTA2), and GSTP1 (see Table S2 for a detailed description of gene mutations). Potential defects in the cellular ROS detoxification system may explain the high sensitivity for PK11007.

Table S2.

Somatic mutations in antioxidative genes of tested cell lines

Protein family	Abbreviation	NUGC-3	NUGC-4	HUH-7	HUH-6	SJSA-1	MKN1	SW480	H1299	HCT116
Glutathione peroxidases	GPX	—	—	—	—	—	—	—	—	GPX4 (c.325-2A > G)*
Glutathione S-transferases	GST	—	—	—	GSTK1 (p.M100I)	—	GSTM2 (p.N59S)	—	—	GSTA2 (c.273-3C > T; c.249G > T);GSTP1 (G208V)*
Thioredoxin-containing proteins	TXN	—	—	—	TXNDC11 (p.N440N; p.N467N)	—	—	—	—	TXN2 (p.K152N); TXNDC5 (p.F368S); TXNDC12 (p.L7L); TXNRD1 (c.1_2insT; c.451_452insT)*
Peroxiredoxins	PRDX		—	—	—	—	—	—	—	—
Superoxide dismutases	SOD	—	—	—	—	—	—	—	—	—
Catalase	CAT	—	—	—	—	—	—	—	—	—
Glutaredoxins	GLRX	—	—	—	—	—	—	—	—	—
Glutamate-cysteine ligase	GCL	—	—	—	—	—	—	—	—	—
Glutaminase	GLS2	—	—	—	—	—	—	—	—	—
Glutathione synthetase	GSS	—	—	—	—	—	—	—	—	—
Gene mutation data		2,237†	889†	932†	1,036†	785†	774†	16‡	145‡	5,100†

^*Mutation “syntax” follows the Human Genome Variation Society nomenclature recommendations (2).

^†Data taken from the COSMIC Cell Lines Project.

^‡Data taken from the COSMIC database.

Changes in cellular ROS levels can alter expression of many genes, activate cell-signaling cascades, and induce apoptotic or necrotic cell death at high intracellular levels (29, 40). PK11007 induced ROS more effectively in mutant p53 cancer cell lines, which may be the reason for stronger viability reduction in NUGC-3, HUH-7, and MKN1 cells. Depending on the cellular context, PK11007 treatment led also to induction of membrane permeabilization (HUH-7) or increased caspase activity (SJSA-1, SW480). The strong synergism of PK11007 and BSO further suggests that the observed ROS increase resulted from depletion of intracellular glutathione levels via GSH-PK11007 adduct formation. There are similar effects of piperlongumine, a Michael acceptor-containing natural product that selectively kills cancer cells (41).

Although p53 knockdown in HUH-6 did not lead to an increased sensitivity for PK11007, in some cases p53 WT cancer cells may be more resistant to ROS stress than p53-deficient cancer cells. While promoting prooxidant and apoptotic pathways at high stress levels, p53 exerts prosurvival and antioxidant responses at modest or transiently elevated oxidative stress levels (29). ROS directly increase p53 activity (29, 42), which leads to up-regulation of several genes with antioxidative effects, including GSTP1 (43), glutathione peroxidase 1 (GPX1), mitochondrial superoxide dismutase 2 (SOD2) (44), or phosphate-activated mitochondrial glutaminase (GLS2), which is an important enzyme for providing l-glutamate as substrate for GSH synthesis (45). Increased ROS levels and oxidation of DNA bases were observed upon knockdown of p53, in cancer cells with mutant p53, or cancer cells with elevated MDM2 levels (44). Further, siRNA knockdown of p53 led to a significantly higher ROS level in RKO cells upon treatment with H₂O₂ (44). p53 activity is modulated through phosphorylation by ROS-activated kinases (42) but is also regulated via direct redox modification. Redox-sensitive cysteines of the p53 DBD can be directly oxidized to various redox states, including formation of p53-GSH disulfides, silphenic acids, sulphinic acids, and sulphonic acids. Modification of Cys-124, -141, and -182 by GSH significantly reduces p53 DNA-binding affinity (46, 47), whereas oxidation of Cys277 leads to a decreased affinity for the GADD45a response element (48). Although thiol-reactive compounds like PK11007 or APR-246 increase intracellular ROS levels, they may promote transcriptional activity of unstable p53 mutants not just by stabilization but also via alkylation of redox-sensitive cysteines (e.g., C182, C277), preventing their oxidation or conjugation with GSH, and reduction of intracellular GSH levels (49). The usually very high ROS and GSH levels in mutant p53 cell lines may be one factor contributing to their increased sensitivity for PK11007.

PK11007 up-regulated two key markers of the UPR: spliced XBP-1 and CHOP, indicating induction of ER stress that may contribute to the observed cell death, especially in MKN1 and HUH-7 cancer cells. Similarly, APR-246 induced transcriptional activation of the XBP-1 gene and cleavage of its mRNA in SW480, Saos-2 and Saos-2-His273 cells (27). ER stress originates from an impaired protein folding capacity in the ER and results in accumulation of unfolded proteins. Excessive ER stress can lead to apoptotic cell death via induction of oxidative stress or also induce autophagy (50, 51). The p53 target gene PUMA is a known mediator of ER stress-induced apoptosis, which is in many cases p53 independent (52). ER stress-induced transcription of PUMA can be mediated by the transcription factors E2F1, which also activates p21 transcription, TRB3, and CHOP (53–56). PK11007-induced up-regulation of CHOP may therefore contribute to increased PUMA mRNA levels. Induction of ER stress is currently exploited as a strategy for anticancer treatment (51) and may contribute to the induction of ROS and p53-independent cell death by PK11007.

Taken together, our findings suggest that PK11007 and other thiol-modifying compounds, such as PRIMA, MIRA, and STIMA, despite alkylating and stabilizing the p53 protein, exert their antitumor function not only via reactivating p53 but also via other cellular mechanisms, such as increase of cellular ROS to toxic levels and activation of the UPR (33). This finding would also explain the seemingly contradictory reports that PRIMA-1, MIRA-1, and STIMA-1 induce cell death also in cell lines with DNA contact mutants that cannot be rescued by simple protein stabilization (12, 11, 14). Irrespective of mechanism, all these reagents have a greater activity toward cancer cells whose p53 status is compromised by mutation or other factors than to normal cells. PK11007 and other 2-sulfonylpyrimidines may have potential as leads for anticancer drugs.

Materials and Methods

Materials.

The in-house fragment library used for the DSF screening assay was purchased from Enamine in 96-well plate format at 20-mM compound concentrations in DMSO. Plates were stored at −20 °C. Derivatives of PK11000 were purchased from Enamine, Vitas-M Laboratory, and Key Organics. APR-246 (PRIMA-1^MET) and buthionine sulfoximine were purchased from Santa Cruz Biotechnology. DMEM High Glucose GlutaMAX and RPMI medium 1640 GlutaMAX were obtained from Life Technologies Ltd.

Protein Expression and Purification.

Plasmids for expression of the cysteine mutants C124/277S, C124/182S, and C182/277S of the p53 DBD were generated with the QuikChange II Site-Directed Mutagenesis Kit (Agilent Technologies). A pET24a vector with the stabilized p53 DBD was used as template (16). The stabilized DBD of the p53 mutant Y220C (T-p53C-Y220C) and T-p53C cysteine mutants were expressed and purified as described (7). Escherichia coli N-acetylneuraminate lyase was produced as described (21).

Differential Scanning Fluorimetry.

Melting temperatures of p53 variants were determined by DSF measurements using SYPRO Orange (Invitrogen) as the fluorescent probe as described (7). A final protein concentration of 8 μM in standard phosphate buffer [25 mM potassium phosphate pH 7.2, 150 mM NaCl, 1 mM TCEP, 5% (vol/vol) DMSO] was used for the DSF measurements. ΔT_m DSF values were calculated by subtracting the average T_m of the control samples from the average T_m of the respective compound samples. All samples were measured in triplicate.

HSQC-NMR.

¹H-¹⁵N HSQC spectra of uniformly ¹⁵N-labeled T-p53C-Y220C (75 μM) and compounds were recorded and analyzed as described (7). Briefly, the spectra were acquired at 293K on a Bruker Avance-800 spectrometer using a 5-mm inverse cryogenic probe. Compound samples were mixed with protein immediately before the NMR measurement. Spectra analysis was performed using Sparky 3.11430 and Bruker Topspin 2.0 software. A previously described assignment map of the p53-Y220C DBD was used to label residues (57).

Aggregation Kinetics.

Aggregation kinetics of the p53 Y220C DBD (94–312) was measured as described (7). Briefly, light scattering was recorded at 37 °C at 500 nm as excitation and emission wavelengths using a Horiba FluoroMax-3 spectrophotometer. Experiments were performed in standard phosphate buffer (as described above) with 3 μM protein.

X-Ray Crystallography.

Crystals of T-p53C-Y220C were grown as described (58); they were soaked for 4 h in a solution of 30 mM PK11000 in cryo buffer [19% (vol/vol) polyethylene glycol 4000, 20% (vol/vol) glycerol, 10 mM sodium phosphate, pH 7.2, 100 mM Hepes, pH 7.2, 150 mM NaCl] and flash-frozen in liquid nitrogen. An X-ray data set was collected at 100 K at beamline I03 of the Diamond Light Source. The data set was integrated using XDS (59) and scaled using SCALA (60) within the CCP4 suite of programs (61). The structure was determined by rigid-body refinement in PHENIX (62) using PDB ID code 2J1X as a starting model, and subsequently refined with iterative cycles of manual model-building in COOT (63) and refinement with REFMAC5 (64). Data collection and refinement statistics are given in Table S3.

Table S3.

X-ray data collection and refinement statistics

Variable	Value
Data collection
Space group	P212121
a, b, c, Å	65.11, 71.08, 105.25
Molecules per asymmetric unit	2
Resolution (Å)*	29.5–1.42 (1.50–1.42)
Unique reflections	92,560
Completeness (%)*	99.8 (99.5)
Multiplicity*	5.5 (5.3)
Rmerge (%)*,†	6.4 (57.2)
<I/σI>*	17.4 (3.9)
Wilson B value, Å2	9.5
Refinement
No. of protein atoms‡	3,136
No. of water atoms	541
No. of zinc atoms	2
No. of ligand atoms	20
Overall B value, Å2	15.2
Rcryst, %§	17.8
Rfree, %§	19.3
RMSD bonds, Å	0.007
RMSD angles, °	1.3
PDB ID code	5LAP

^*Values in parentheses are for the highest-resolution shell.

^†R_merge = ∑(I_h,i − <I_h>)/∑I_h,i.

^‡Number includes alternative conformations.

^§R_cryst and R_free = ∑||F_obs|− |F_calc||/∑|F_obs|, where R_free was calculated over 5% of the amplitudes chosen at random and not used in the refinement.

Mass Spectrometry.

To check for alkylation of T-p53C-Y220C (94-312) and T-p53, we added DMSO stocks of compounds to a reaction buffer containing 50 μM protein, 25 mM potassium phosphate pH 7.2, 150 mM sodium chloride, and 1 mM Tris(2-carboxyethyl)phosphine hydrochloride (TCEP), yielding a final concentration of 5% (vol/vol) DMSO. The samples were incubated for 4 h at 20 °C on a rotating shaker. The protein sample was then diluted to 5 µM with 100 mM ammonium acetate buffer and desalted using Millipore C4 ZipTips. The mass of the proteins was determined by electrospray mass spectrometry with a Waters (Micromass) LCT TOF mass spectrometer in ESI (ES+) mode.

Reaction Kinetics.

¹H-NMR spectra were recorded at 298 K on a Bruker AVANCE III 600 spectrometer. The NMR sample contained 1 mM PK11000 and 1 mM glutathione in 25 mM phosphate pH 7.2, 150 mM NaCl, 1 mM TCEP, and 5% (vol/vol) DMSO-d6 buffer. Aromatic proton peaks at 8.51 ppm (adduct) and 8.93 ppm (PK11000) were integrated to give concentrations of PK11000 and its GSH adduct over time. The data were then fitted with a second-order kinetics equation for equimolar educt concentrations using KaleidaGraph (65)

$$c\left(pro\right)=c_0-\frac{1}{k_{2}t+c_0}.$$

Fluorescence Polarization DNA-Binding Assay.

A total of 5 μM stabilized full-length p53 was incubated with 1 mM compound for 2 h at 4 °C. The fluorescence polarization DNA binding assay was then performed, as described (57), in a 25-mM sodium phosphate pH 7.2, 1 mM TCEP, 150 mM sodium chloride, 5% (vol/vol) DMSO, 0.05% Tween 20, and 0.2 mg/mL BSA assay buffer with 20 nM of a 5′ fluorescein-labeled GADD45 DNA response element (66). Data were fitted to the Hill equation including a linear drift term using KaleidaGraph.

Cell Culture.

H1299 cells were a kind gift from Carol Prives, Columbia University, New York; H1299 cells with constitutively expressed p53 R175H were a kind gift of Fiona M. Townsley, MRC Laboratory of Molecular Biology, Cambridge, UK; and HCT116 p53^−/− cell lines were a kind gift of Bert Vogelstein, Johns Hopkins Medicine, Baltimore. HCT116 (WT p53), SW480 (p53-R273H/P309S) and SJSA-1 (WT p53) were purchased from ATCC; HUH-7 (p53-Y220C^+/+), HUH-6 (WT p53^+/+). NUGC-3 (p53-Y220C^+/+), NUGC-4 (WT p53^+/+), and MKN1 (p53-V143A^+/+) cells were obtained from the Japan Health Science Research Resources Bank and cultured as described (8). Briefly, cells were maintained in DMEM (HUH-6, HUH-7, HCT116, SW480) or in RPMI 1640 (NUGC-3, NUGC-4, H1299, H1299 R175H, MKN1, SJSA-1) medium with 10% (vol/vol) FCS and 1% antibiotic stock mix (10,000 U/mL penicillin, 10,000 mg/mL streptomycin) and incubated in a humidified incubator at 37 °C with 5% (vol/vol) CO₂. The RPMI medium for culturing H1299 R175H was supplemented with 600 µg/mL G418.

Cell Viability Assay.

Cell viability was measured using the CellTiter-Fluor cell viability assay kit (Promega). Cells were seeded in 96-well plates at 7,500–15,000 cells per well and incubated overnight. Samples were prepared in medium with a twice-as-high compound, and DMSO concentration then added to an equivalent volume of growth medium, yielding a final DMSO concentration of 0.5%. After incubating of cells for 23 h or the respective time period, CellTiter-Fluor reagent was added to each well and incubated again for 45 min. Fluorescence was then recorded on a PHERAstar plate reader (BMG Labtech) using 400/500-nm excitation/emission filters. The MKN1 time course and siRNA p53 knockdown experiments were performed with the CellTiter-Glo 2.0 assay kit (Promega). Luminescence was recorded with a Centro XS³ LB 960 microplate luminometer (Berthold Technologies). Experiments were performed in quadruplicate.

p53 Knockdown.

p53 protein levels were down-regulated via transfection of human-specific p53 siRNA (Qiagen) using the INTERFERin siRNA Transfection Reagent (Polyplus). Negative control siRNA (Qiagen) was used as negative control for the siRNA transfection. The knockdown was confirmed by Western blots (Fig. S10).

Fig. S10.

Protein levels of p53 after treatment with nontarget or p53 siRNA in MKN1, HUH-6, and HUH-7 cancer cells.

Western Blots.

Cells were seeded in six-well plates at 0.5–0.8 million cells per well and incubated overnight at 37 °C and 5% (vol/vol) CO₂. Cells were harvested after treatment with PK11007 at 0.5% DMSO per well for 3 h and lysed in RIPA buffer (Sigma-Aldrich) containing one cOmplete EDTA-Free Protease Inhibitor mixture tablet (Roche Diagnostics) per 50 mL RIPA buffer. Protein levels were determined using Coomassie Plus protein assay reagent (Thermo Scientific). SDS gel electrophoresis was conducted using NuPAGE 4–12% (vol/vol) Bis-Tris gels (Life Technologies) loading 20 µg protein per lane. Proteins were electroblotted onto a Millipore Immobilon-P PVDF membrane (Millipore). Membranes were then blocked with PBS containing 5% (mass/vol) dried skimmed milk for 1 h at RT, incubated with primary antibodies for 1 h (or overnight at 4 °C), and then with secondary antibodies coupled to horseradish peroxidase for 1 h. The blots were treated with GE ECL or ECL PRIME chemiluminescent detection reagent (Little Chalfont) and exposed to Fuji Super RX-N medical X-ray film for detection. The following antibodies were used: p53 (DO-1), p21 (187), GSTP1 (3F2C2), CHOP (sc-793), XBP-1 (sc-7160; Santa Cruz Biotechnology), PUMA (ab9643), β-actin (AC-15), MDM2 (2A10; Abcam). Anti–mouse-HRP (sc-2005) and anti–rabbit-HRP (sc-2004) antibodies were both obtained from Santa Cruz Biotechnology.

Real-Time PCR.

HUH-6, NUGC-3, HUH-7, and MKN1 cells were treated with PK11007 or DMSO control for 6 h (4.5 h for HUH-7). Total RNA was extracted and purified using RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. Synthesis of cDNA was performed with iScript cDNA Synthesis Kit (Bio-Rad). Real-time PCR was performed using the Rotor-Gene SYBR Green Kit (Qiagen) on a Rotor-Gene 6000 (Corbett Life Science) PCR cycler. The relative standard curve method was used to quantify relative mRNA levels. Each sample was measured in triplicate.

Caspase-3/7 and Cytotoxicity Assay.

Caspase-3/7 activity and cytotoxicity were measured using the ApoTox-Glo Triplex assay (Promega), performed as described (67). Briefly, cells were incubated with compound or DMSO control for 6 h. After 1 h incubation with Caspase-Glo 3/7 reagent, we recorded luminescence using a Centro XS³ LB 960 microplate luminometer (Berthold Technologies). Experiments were performed in quadruplicate.

Detection of Intracellular Reactive Oxygen Species.

Compounds and controls were added to yield a final sample DMSO concentration of 0.25%. After incubation for 1.5 h at 37 °C and 5% (vol/vol) CO₂, and another 30 min after adding 0.5 µM CellROX Deep Red reagent (Life Technologies) and 1 µM SYTOX Blue Reagent (Life Technologies), we measured in triplicate the median CellROX Deep Red fluorescence of live cells using an Eclipse Flow Cytometry Analyzer (Sony Biotechnology Inc.).

SI Materials and Methods

Immunofluorescence experiments were performed as described (8). Briefly, NUGC-3 cells were treated with 25 μM PK11007 or DMSO control for 2 or 4 h, fixed with 4% (vol/vol) paraformaldehyde, and permeabilized with 0.5% (wt/vol) Triton X-100. The following antibodies were used: anti-p53 antibody Pab 1620 (Abcam) and anti-p53 antibody Pab 240 (Abcam). Hoechst 33342 (Cell Signaling) stain was used to stain the nucleus of cells. Imaging was performed using a Leica TCS SP8 confocal microscope.