Discovery of Novel Aldo-Keto Reductase 1C3 Inhibitors as Chemotherapeutic Potentiators for Cancer Drug Resistance

Abstract

As a crucial target which is overexpressed in a variety of cancers, aldo-keto reductase 1C3 (AKR1C3) confers chemotherapeutic resistance to many clinical agents. However, a limited number of AKR1C3-selective inhibitors are applied clinically, which indicates the importance of identifying active compounds. Herein, we report the discovery, synthesis, and evaluation of novel and potent AKR1C3 inhibitors with structural diversity. Molecular dynamics simulations of these active compounds provide reasonable clarification of the interpreted biological data. Moreover, we demonstrate that AKR1C3 inhibitors have the potential to be superior therapeutic agents for re-sensitizing drug-resistant cell lines to chemotherapy, especially the pan-AKR1C inhibitor S07-2010. Our study identifies new structural classes of AKR1C3 inhibitors and enriches the structural diversity, which facilitates the future rational design of inhibitors and structural optimization. Moreover, these compounds may serve as promising therapeutic adjuvants toward new therapeutics for countering drug resistance.

In the human body, aldo-keto reductase family 1 member C (AKR1C) enzymes have been identified with differential tissue expression and proved to possess overlapping but differing pleiotropic activities.¹ Among these isoforms, AKR1C3, known as 17β-hydroxysteroid dehydrogenase type 5 (17β-HSD5), is involved in the biosynthetic pathway of potent androgen (testosterone and 5α-dihydrotestosterone) and estrogen (17β-estradiol) available for the androgen receptor (AR) and estrogen receptor (ER) in target tissues, respectively.² Like prostaglandin F synthase, AKR1C3 also catalyzes the conversion of prostaglandin D₂ (PGD₂) into 9α,11β-PGF_2α, which acts to prevent myeloid differentiation and facilitates proliferation of tumor cells by preventing peroxisome proliferator-activated receptor γ (PPARγ) activation.³ Thus, AKR1C3 plays a pivotal role in the regulation of cell proliferation and differentiation in both hormone-dependent and hormone-independent manners.

Dysregulated AKR1C3 expression is associated with the development and poor prognosis of a variety of cancers; overexpression of AKR1C3 is also related to the resistance to radiotherapy and chemotherapy in many tumor cells.^4,5 AKR1C3 mediates the development of cancer cells’ resistance to anticancer drugs such as anthracyclines in the treatment of breast cancer and acute myeloid leukemia (AML),⁶ and enzalutamide⁷ and abiraterone acetate⁸ in the treatment of castration-resistant prostate cancer (CRPC). In addition, AKR1C3 is also involved in the development of colon cancer cells’ and breast cancer cells’ resistance to cisplatin⁹ and chronic myeloid leukemia cells’ resistance to imatinib.¹⁰ Thus, AKR1C3 has been recognized as not only a potential diagnostic or prognostic marker but also a novel therapeutic target for overcoming chemoresistance in cancer therapy. The effectiveness of AKR1C3-selective inhibitors combined with chemotherapeutic drugs has been confirmed in colon cancer cells,¹¹ liver cancer cells,¹² AML cell lines,¹³ and prostate cancer cells.^14,15

Considerable efforts have been invested in the discovery of AKR1C3 inhibitors.^5,15 As per our knowledge to date, no AKR1C3-specific inhibitor has been successfully marketed. ASP9521¹⁶ and BAY-1128688¹⁷ were evaluated in phase I/II clinical trials that were terminated due to toxicity or lack of efficacy. The non-steroidal anti-inflammatory drug (NSAID) indomethacin is still in phase I/II clinical trials to surmount drug resistance to enzalutamide (ClinicalTrials.Gov NCT02935205).⁷ Significant efforts are still needed to discover superior AKR1C3 inhibitors for clinical research.

The aim of the present study is to identify potent AKR1C3 inhibitors with new scaffolds. These hit compounds may provide a good starting point for further optimization and are important tools to better understand AKR1C3’s role in the development of cancer. In the work described here, we screened four commercial compound libraries by using in silico approaches and enzymatic assay. Five inhibitors were identified. Finally, we explored a new combination therapy using AKR1C3 inhibitors with chemotherapeutic agents in drug-resistant cancer cells.

Coupling of pharmacophore-based virtual screening with docking-based virtual screening can efficiently improve the hit rate. By performing qualitative HipHop pharmacophore-based virtual screening¹⁸ and cascade molecular docking, we obtained 15 candidates with novel skeletons in our study (Figure 1 and Table S2). The inhibitory capacity of the candidates toward AKR1C3 at 10 μM was first evaluated. Five compounds exhibited more than 50% inhibitory capacity (Figure S6). Subsequently, those five hits were proposed for a dose-dependent study of their inhibitory activity on AKR1C3 and their inhibitory capacity against three other isoforms of AKR1C. The inhibitory potency was determined by measuring the inhibition of the NADP⁺-dependent oxidation of the pan-AKR1C substrate S-tetralol.¹⁹ All the hit compounds showed IC₅₀ values toward AKR1C3 ranging from 0.13 to 2.08 μM (Table 1). S07-2008 showed high selectivity toward AKR1Cs. Although the inhibitory activity of S07-2001 toward AKR1C3 was approximately 13-fold less than that of S07-2008, it still exhibited selectivity toward other AKR1C isoforms. Intriguingly, S07-2010 showed sub-micromolar potency toward all the AKR1Cs studied and was thus regarded as a pan-AKR1C inhibitor.

Figure 1.

Schematic representation of the virtual screening protocol. The four commercial databases used were Lifechemicals, Chemdiv, Enamine, and Interbioscreen.

Table 1. AKR1C Isoform Inhibitory Activity and Selectivity of Hit Compounds.

	inhibitory activity, IC50 (μM)a				IC50 ratio
compd	1C1	1C2	1C3	1C4	1C1:1C3	1C2:1C3	1C4:1C3
S07-2001	>100	>100	2.08 ± 1.26	>100	>48.1	>48.1	>48.1
S07-2005	2.88 ± 1.10	50.03 ± 20.67	0.13 ± 0.03	0.75 ± 0.32	22.2	384.8	5.8
S07-2008	>100	>100	0.16 ± 0.09	>100	>625.0	>625.0	>625.0
S07-2009	11.18 ± 4.39	5.71 ± 2.78	0.20 ± 0.12	39.58 ± 12.77	55.9	28.55	197.9
S07-2010	0.47 ± 0.16	0.73 ± 0.35	0.19 ± 0.08	0.36 ± 0.15	2.47	3.84	1.89

^aThe data are expressed as the mean ± SEM of three independent experiments.

Because the synthesis routes of the five commercial compounds have not been revealed, we developed efficient routes to obtain the expected target molecules, which was beneficial for further structure–activity relationship studies. Due to the presence of the chiral center in S07-2005, the compound was synthesized as racemates without separation of the enantiomer. Subsequent biological evaluation still used commercial compounds.

As shown in Scheme 1, (4-cyanophenyl)methanesulfonyl chloride (2) was synthesized from 4-(chloromethyl)benzonitrile (1) in two steps, converting the chlorine to sulfonyl chloride under acidic conditions. Subsequent addition of ammonia gave 3. The intermediate 3 was reduced to intermediate 4 by lithium aluminum hydride. Methyl 4-acetyl-1H-pyrrole-2-carboxylate (6) was synthesized from methyl 1H-pyrrole-2-carboxylate (5) through Friedel–Crafts acylation, followed by hydrolysis to yield 7. The product 4-acetyl-N-(4-(sulfamoylmethyl)benzyl)-1H-pyrrole-2-carboxamide was synthesized from the intermediates 4 and 7 via a condensation reaction.

Scheme 1. Synthesis of Compound S07-2001.

Reagents and conditions: (a) (i) SCNH₂, EtOH, reflux 1 h; (ii) 6 M H₂SO₄, NaOCl, ethyl ether, 0–15 °C, 30 min. (b) NH₄OH(aq), DCM, rt, 3 h. (c) LAH, THF, rt, 3 h. (d) Acetyl chloride, AlCl₃, CS₂, 0–60 °C, 4 h. (e) LiOH, MeOH/THF, overnight. (f) HATU, DIPEA, DMF, rt, 12 h.

In Scheme 2, in the presence of paraformaldehyde and MgCl₂, triethylamine was used as a base to afford the intermediate 9, which was then cyclized to get compound 10. The cyano group on the intermediate 10 was hydrolyzed to give a carboxylic acid derivative, 11. Finally, the intermediate 6-ethoxychromane-3-carboxylic acid (12) was obtained via catalytic hydrogenation. Another crucial intermediate, 15, was synthesized in two steps, involving the Vilsmeier–Haack reaction and Borch reduction. 5-((6-Ethoxy-N-methylchromane-3-carboxamido)methyl)-2-methylfuran-3-carboxylic acid was synthesized using 12 and 15, followed by hydrolysis with 1 N LiOH in overall good yields.

Scheme 2. Synthesis of Racemic Compound S07-2005.

Reagents and conditions: (a) (CH₂O)_n, MgCl₂, Et₃N, 70 °C, 12 h. (b) Acrylonitrile, Dabco, 78 °C, 12 h. (c) (i) 1 N NaOH, EtOH, 80 °C, 10 h; (ii) 3 N HCl, 25 °C, 10 min. (d) H₂, Pd/C, MeOH, 25 °C, 48 h. (e) Ar, POCl₃, DMF, 25 °C, 24 h. (f) CH₃NH₂, 25 °C, CH₃BNNa, MeOH. (g) HATU, DIEA, 25 °C, 12 h. (h) THF, LiOH, H₂O, 25 °C, 48 h.

The synthesis of target compound S07-2008 is described in Scheme 3. The intermediate 19 was obtained from 4-(chloromethyl)-3,5-dimethylisoxazole (17) and methyl salicylate (18) and then hydrolyzed using NaOH as base. Finally, the intermediate 20 was treated with 4-(2-aminoethyl)benzenesulfonamide to obtain the target compound.

Scheme 3. Synthesis of Compound S07-2008.

Reagents and conditions: (a) K₂CO₃, KI, MeCN, 85 °C, 24 h. (b) NaOH, EtOH, H₂O, 25 °C, 6 h. (c) 4-(2-Aminoethyl)benzenesulfonamide, HATU, DIPEA, DMF, 25 °C, 12 h.

The synthesis of 1,2,4-oxadiazole derivative 23 from arylamidoximes 22 and succinic anhydride in high yields is described in Scheme 4. The 1,2,4-oxadiazole derivative 23 and the intermediate 15 from Scheme 2 then underwent a condensation reaction to give the compound 24 upon hydrolysis under the same reaction conditions in Scheme 4.

Scheme 4. Synthesis of Compound S07-2009.

Reagents and conditions: (a) NH₂OH (50 wt%), EtOH, 80 °C. (b) Succinic anhydride, 130 °C, 40 min. (c) 15, HATU, DIPEA, DMF, rt. (d) LiOH, MeOH/THF, rt, overnight.

The target compound 7-(2-((6-oxo-4-propyl-1,6-dihydropyrimidin-2-yl)thio)acetyl)-1,3,4,5-tetrahydro-2H-benzo[b]azepin-2-one was synthesized in four steps (Scheme 5). First, the initial compound 25 was treated with hydroxylamine hydrochloride using NaOH as base. Polyphosphoric acid was used as solvent, and the direct conversion of 3,4-dihydronaphthalen-1(2H)-one oxime (26) to amides via Beckmann rearrangement proceeded in good to excellent yields. Then, compound 27 was converted to intermediate 28 through Friedel–Crafts acylation using CS₂ as solvent. Finally, S07-2010 was obtained via substitution reaction using cesium carbonate as base at room temperature.

Scheme 5. Synthesis of Compound S07-2010.

Reagents and conditions: (a) hydroxylamine hydrochloride, NaOH. (b) PPA, 125 °C, 30 min. (c) Bromoacetyl bromide, AlCl₃, CS₂, reflux, 4 h. (d) 6-Propyl-2-thiouracil, Cs₂CO₃, DMF, rt, 12 h.

To better confirm the capacity of lead compounds targeting AKR1C3 and illustrate how the potency is achieved, molecular dynamics (MD) simulations were performed based on preliminary docking poses achieved from virtual screening. The inhibitor–AKR1C3 complexes stayed in an equilibrated and converged state and remained stable for 100 ns of simulation, indicating the stability of these complexes (Figure S7). The total binding free energies of the five complexes were below −110 kcal/mol, where the van der Waals energy and electrostatic energy were the major favorable contributors for the binding, whereas the polar solvation energies with positive values opposed binding (Table 2).

Table 2. Binding Free Energy (MM-PBSA) Results for Hit Molecules with AKR1C3 (kcal/mol).

compd	EvdWa	Eelb	Egbc	Esurfd	ΔGgase	ΔGsolvf	Δ totalg
S07-2001	–132.2844	–85.1602	102.1911	–15.7357	–217.4446	86.4554	–130.9892
S07-2005	–120.5874	–27.8248	51.9686	–14.4545	–148.4122	37.5141	–110.8980
S07-2008	–123.9503	–106.6039	118.5988	–15.6784	–230.5543	102.9204	–127.6339
S07-2009	–119.2172	–33.3193	55.3454	–14.3933	–152.5365	40.9521	–111.5844
S07-2010	–127.8439	–78.4384	106.9256	–15.5571	–206.2823	91.3685	–114.9138

^avan der Waals energy.

^bElectrostatic energy.

^cPolar solvation energy.

^dNon-polar solvation energy.

^eTotal gas-phase free energy.

^fTotal solvation free energy.

^gTotal binding free energy.

The ligand binding site of AKR1C3 is hydrophobic, consisting of an oxyanion site (OS), a steroid channel (SC), and three sub-pockets (SP). (Figure 2A). Tyr55 and His117 in the OS are highly conserved in AKR1Cs, playing an important role in the proton transfer. Notably, among AKR1Cs, AKR1C3 has the largest SP1 pocket due to the conformational changes of Trp227, Phe306, and Phe311, enabling it to accommodate specific ligands.²⁰ Therefore, inhibitors anchored to the OS and SP1 might have enhanced activity and selectivity. For these compounds, the total energies of key residues in AKR1C3 binding site were below −1 kcal/mol, indicating that these residues contribute to stabilizing the binding conformation during the dynamic process (Figure S7). The predicted binding patterns of the five complexes in MD simulations are presented in Figure 2 and Figure S8. It was observed that S07-2001 bound to OS, SP1, and SP3, forming H-bonds to catalytic His117 in a linear conformation, while the arylamino ring penetrated into the SP1 pocket by interacting with Phe306 and Phe311 and the sulfonamide group extended into SP3 pocket. S07-2008 embedded into three sub-pockets, selectivity in AKR1Cs can be gained by taking advantage of the structural differences in Phe306. In the binding pattern, the benzenesulfonamide tail in S07-2008 formed a π–π stacked interaction with Ph306 and a H-bond with Gln222. Due to the structural flexibility, the “V” shape conformation of S07-2008 and the “C” shape of S07-2009 were assumed to be the optimal conformations, enabling AKR1C3 to accommodate them. For S07-2005 and S07-2010 with conformational constraints, the bulky substitutions could provide strong hydrophobic interactions in the SP1 and SP2 pockets. Phe311 in SP1 and Trp86 and Trp227 in SP2 formed π–π stacked interactions with S07-2005 and S07-2010. All interactions stabilized the conformation and combination of the ligands lying in the OS, SP1, and SP2 pockets, which may explain the activity toward AKR1C3.

Figure 2.

MD simulation results. (A) Overall structure of AKR1C3 (PDB code: 3R94). The residues at the OS (Tyr55, His117) are shown in yellow stick mode; the residues in SP1 (Ser118, Asn167, Phe306, Phe311, Tyr319) are shown in green stick mode; the residues in SP2 (Trp86, Ser129, Trp227, Phe311) are shown in red stick mode; the residues in SP3 (Tyr24, Glu192, Ser221, Gln222, Tyr305) are shown in light blue stick mode; Leu54 in the SC is shown in white stick mode. Predicted binding patterns of S07-2001 (B), S07-2005 (C), S07-2008 (D), S07-2009 (E), and S07-2010 (F) in the last frame of MD simulations. Inhibitors are shown in light blue stick mode; NADP⁺ is shown in blue stick mode; key residues are shown in wheat stick mode. H-bonds are represented with green dotted lines; π–π stacking is depicted with purple dotted lines.

AKR1C3 has been proved to be closely related to anthracycline and cisplatin (DDP), which are involved in multi-drug resistance.⁴ We successfully established doxorubicin (DOX)-resistant breast cancer cells MCF-7/DOX and DDP-resistant lung cancer cells A549/DDP to confirm the combined effects of AKR1C3 inhibitors (Figures S9 and S10). Compared with the parental cells, the morphology of the drug-resistanct cell lines completely changed with varying levels of AKR1C3. S07-2010 exhibited obvious cytotoxicity on MCF-7/DOX and A549/DDP, with IC₅₀ = 127.5 and 5.51 μM, respectively, while other compounds had no cytotoxicity at high concentrations (Figure S11).

Subsequently, combination treatment of drug-resistant cancer cells with a range of concentrations of chemotherapeutic agents and AKR1C3 inhibitor was performed to determine the effects on cell viability. Treatment with AKR1C3 inhibitors alone showed weak cytotoxicity against cancer cells. In previous experiments, we observed that these inhibitors had high onset doses in these two drug-resistant cancers, with the exception of S07-2010. In A549/DDP cells, S07-2010 could obviously reverse the resistance of the cancer cells to DDP and inhibited tumor proliferation at low concentrations, while other compounds with better selectivity toward AKR1C isoforms exhibited significant differences only at high concentrations (Figure 3). In MCF-7/DOX cells, significant adjuvant effects were observed when these AKR1C3 inhibitors were incubated with 25 μM DOX. This concentration of DOX provided a 5% reduction of cell viability alone, which was potentiated by the action of S07-2001 or S07-2008 at 50 μM to provide a 27% reduction of cell viability. When a comparison was made at 25 μM DOX with the pan-inhibitor S07-2010 at 10 μM, a reduction of cell viability of 29% was observed. As for S07-2005 and S07-2009, combining them with DOX showed no significant difference at multiple concentrations (Figure S12). These data illustrate that S07-2010 showed the most potent adjuvant effect on both MCF-7/DOX and A549/DDP cells.

Figure 3.

Synergistic activity of AKR1C3 inhibitors. (A–F) The combination of AKR1C3 inhibitors and chemotherapeutic agents synergistically inhibited the proliferation of cancer cells. Columns show mean ± SD (n = 3). *p < 0.1, **p < 0.01, ***p < 0.001, and ****p < 0.0001 for cells treated with specific chemotherapeutic agents vs combination treated cells; ns = not significant. (G) Cytotoxic effect (Fa)–combination index (CI) plot showing the CI value of the combination at fixed concentrations. CI > 1.10 indicates antagonism, CI = 0.9–1.10 indicates an additive interaction, CI < 0.9 indicates synergism, and CI < 0.3 indicates strong synergism.

Based on these observations, the combination index (CI) values were calculated by the Chou–Talalay method.²¹ In MCF-7/DOX cells, synergistic effects of candidates were observed at all dosages after a 48-h incubation (CI < 0.9). Strong synergistic drug effects could be seen upon treatment with S07-2001 or S07-2010 and DOX, with relative low CI values in MCF-7/DOX. The CI values of S07-2008, S07-2005, and S07-2009 also indicated that these AKR1C3-selective inhibitors were sufficient to sensitize MCF-7/DOX cells to DOX at all concentrations (Figure S12). In A549/DDP cells, CI values of S07-2010 ranged from 0.2 to 0.6, indicating a strong degree of synergistic drug effects in the co-treatment experiments. However, the CI values of S07-2001 and S07-2008 indicated the combination at all concentrations was not as synergistic as that in MCF-7/DOX. Combination of S07-2005 and S07-2009 with DPP did not show significant synergistic action; additive interaction and antagonism even occurred at multiple doses (Figure S13).

Compared to the parental cells, MCF-7/DOX cells have a higher fold-induction of AKR1C3,²² indicating the important role of AKR1C3 in decreasing the sensitivity of MCF-7 to DOX, while in A549/DDP cells, AKR overexpression (not only AKR1C3) is part of the cancer chemotherapeutic resistance phenotype to platin-based cancer chemotherapeutics.⁴ These results may explain why the pan-AKR1C inhibitor S07-2010 provided a better effect than the more selective compounds. To confirm our hypothesis, we chose representative AKR1C3 inhibitor indomethacin as adjuvant agent in these two drug-resistant cancer cell lines. With the exception of compound S07-2001, the hit compounds showed better inhibitory activity against AKR1C3 than indomethacin (Figure S14) in our assay. Compared with the five hit compounds, synergistic action was also observed in MCF-7/DOX cells when indomethacin was combined with DOX, but the AKR1C3-selective inhibitor also failed to exert S07-2010-like activities at high concentrations in A549/DDP cells.

The acquired resistance of tumors may involve a series of AKR enzymes. If only one of the enzymes is inhibited, then the other enzymes still mediate tumor resistance. Both AKR isoform-selective inhibitors and pan-AKR1C inhibitors can restore chemosensitivity in drug-resistant cell lines. For the development of AKR1C3-selective inhibitors or AKR1C pan-inhibitors, they need to be placed in specific application environments. For CRPC, it has been confirmed that AKR1C3 is more closely related, so AKR1C3-selective inhibitors are more promising. For diseases in which the AKR family is highly expressed, pan-inhibitors may improve clinical outcomes.⁴ For example, in AML, pan-inhibition can promote apoptosis and/or differentiation of leukemia cells, while AKR1C3-selective inhibitors are insufficient to exert anti-tumor effects.²³ In our experiment, S07-2010 showed obvious synergistic effects on both drug-resistant cell lines. Studies have shown that AKR overexpression is associated with DDP resistance in many different cancer cells. Therefore, S07-2010 has the potential to be developed as an AKR1C pan-inhibitor for reversing the resistance of lung cancer cells to DDP. As for MCF-7/DOX, the expression levels of some AKR-related proteins need to be determined. For the molecular mechanism of AKR in mediating tumor resistance, it is still necessary to develop highly active and highly AKR1C3-selective inhibitors for comparison. Subsequent structural optimization is extremely necessary to acquire inhibitors with greater potency and increased cell activity. Also, the effects of the inhibitors on signals related to survival and proliferation need to be assessed.

Subsequently, Hoechst 33342 staining was used to observe the cells’ nuclear morphology of the four groups after the cells were treated with drugs for 48 h in MCF-7/DOX. The nuclear morphology of cells treated with DOX alone did not undergo marked changes. Compared with the blank control group, i.e., the DOX group, the number of the cells treated with S07-2010 combined with DOX was greatly reduced and increasing chromatin condensation and nuclear fragmentation could be observed (Figure 4A). Meanwhile, cell apoptosis in A549/DDP was detected by flow cytometry after DDP and/or S07-2010 treatment for 48 h. Compared with the blank control group, the percentage of cells that showed apoptosis was significantly increased in the drug treatment groups. The percentage of early apoptosis cells was 16.67% in S07-2010 combined with DDP. These results indicated that S07-2010 strengthened the cytotoxicity of chemotherapeutic agents in drug-resistant cells. A colony formation test indicated that combinations of chemotherapeutic agents and S07-2010 significantly inhibited the proliferation of drug-resistant cells compared with the blank control group.

Figure 4.

(A) S07-2010 combined with chemotherapeutic agents promoted apoptosis of cancer cells. Hoechst 33342 staining showed the nuclear morphology of MCF-7/DOX cells after DOX and/or S07-2010 treatment for 48 h. At least three captured fields were randomly selected. Annexin V-FITC/PI assay was performed to quantitative analysis of apoptotic cells of A549/DDP cells after DDP and/or S07-2010 treatment for 48 h. (B) Combination of S07-2010 with DOX or DDP inhibited proliferation and reversed the resistance in cancer cells. MCF-7/DOX cells were treated with 0.1% DMSO (control), DOX (25 μM), S07-2010 (25 μM), or DOX plus S07-2010. A549/DDP cells were treated with 0.1% DMSO (control), DDP (5 μg/mL), S07-2010 (1 μM), or DDP plus S07-2010.

In the work described in this manuscript, we applied an effective hierarchical in silico virtual screening method to identify potential AKR1C3-selective inhibitors and a pan-AKR1C inhibitor, which enriches the structural diversity and will no doubt provide a very good starting point for further optimization. Our results suggest that AKR1C3-selective inhibitors may serve as adjuvant agents for the synergistic treatment of doxorubicin-resistant breast cancer. However, we serendipitously find that, at least in cisplatin-resistant A549 cancer cells, selective inhibition of AKR1C3 is insufficient to elicit an anticancer effect, and multiple AKR1C inhibition may be required to increase chemotherapeutic sensitivity.

Glossary

Abbreviations

AKR1C

aldo-keto reductase family 1 member C

AML

acute myeloid leukemia

AR

androgen receptor

CI

combination index

CRPC

castration-resistant prostate cancer

DDP

cisplatin

DOX

doxorubicin

ER

estrogen receptor

MD

molecular dynamics

NSAID

non-steroidal anti-inflammatory drug

OS

oxyanion site

PGD₂

prostaglandin D2

PPARγ

peroxisome proliferator-activated receptor γ

SC

steroid channel

17β-HSD5

17β-hydroxysteroid dehydrogenase type 5

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.2c00175.

• Additional information and results about in silico experiments and biological assays; experimental details; and characterization for all final compounds, including Figures S1–S14 and Tables S1–S3 (PDF)

The authors declare no competing financial interest.