Discovery of (R)-2-(6-Methoxynaphthalen-2-yl)butanoic Acid as a Potent and Selective Aldo-keto Reductase 1C3 Inhibitor

Abstract

Type 5 17β-hydroxysteroid dehydrogenase, aldo-keto reductase 1C3 (AKR1C3) converts Δ⁴-androstene-3,17-dione and 5α-androstane-3,17-dione to testosterone (T) and 5α-dihydrotestosterone, respectively, in castration resistant prostate cancer (CRPC). In CRPC, AKR1C3 is implicated in drug resistance, and enzalutamide drug resistance can be surmounted by indomethacin a potent inhibitor of AKR1C3. We examined a series of naproxen analogues and find that (R)-2-(6-methoxynaphthalen-2-yl)butanoic acid (in which the methyl group of R-naproxen was replaced by an ethyl group) acts as a potent AKR1C3 inhibitor that displays selectivity for AKR1C3 over other AKR1C enzymes. This compound was devoid of inhibitory activity on COX isozymes and blocked AKR1C3 mediated production of T and induction of PSA in LNCaP-AKR1C3 cells as a model of a CRPC cell line. R-Profens are substrate selective COX-2 inhibitors and block the oxygenation of endocannabinoids and in the context of advanced prostate cancer R-profens could inhibit intratumoral androgen synthesis and act as analgesics for metastatic disease.

Graphical Abstract

INTRODUCTION

The development of castrate resistant prostate cancer (CRPC) in patients that have undergone androgen deprivation therapy (ADT) is driven by reactivation of androgen receptor (AR) signaling within the tumor in the presence of castrate levels of circulating androgens.^1,2 AR reactivation results from adaptive intratumoral androgen biosynthesis^3–5 and from changes in the AR itself, including gene amplification,⁶ AR mutations that make the receptor ligand permissive,^7,8 and the appearance of AR splice variants^9,10 which make the receptor constitutively active.

The importance of the conversion of adrenal androgens into testosterone (T) and 5α-dihydrotestosterone (5α-DHT) or de novo androgen biosynthesis within the tumor is firmly established by the therapeutic efficacy of abiraterone acetate (AA), an inhibitor of P450c17 (17α-hydroxylase/17,20 lyase) in CRPC patients.^11–15 However, concerns over the need for coadministration of prednisone with AA to prevent adrenal insufficiency and the rapid appearance of drug resistance indicate a pressing need for new therapeutic agents.^16,17

Aldo-keto reductase 1C3 (AKR1C3), also known as type 5 17β-hydroxysteroid dehydrogenase (HSD17B5), is a 17-ketoreductase that catalyzes the NADPH dependent conversion of androgen precursors, 4-androstene-3,17-dione (Δ⁴-AD) and 5α-androstan-3,17-dione (5-Adione), to yield the potent androgens, T and 5α-DHT, respectively (Scheme 1).^18,19 AKR1C3 also catalyzes the conversion of androsterone to 5α-androstane-3α,17β-diol, the precursor of 5α-DHT in the backdoor pathway.^20,21 AKR1C3 is one of the most highly overexpressed steroidogenic enzymes in CRPC compared to normal prostate tissue and prostate cancer.^22–26 Moreover, it is dramatically upregulated by ADT.^23,27,28 Upon ADT, AKR1C3 is induced by the TMPRSS2-ERG fusion protein, whereby the ERG transcription factor can override the repressive effects of the AR binding to the AKR1C3 promoter.²⁹ AKR1C3 also plays a role in resistance to P450c17 inhibition by AA observed in prostate cancer cell lines and xenografts.^16,17 AKR1C3 may also act as an AR selective coactivator that promotes tumor growth, and this coactivator function could be blocked with small molecule enzyme competitive inhibitors.³⁰

Scheme 1. Central Role of AKR1C3 in Androgen Biosynthesis in Prostate Cancera.

^aDHEA = dehydroepiandrosterone, Δ⁴-AD = Δ⁴-androstene-3,17-dione, Adione = 5α-androstane-3,17-dione, Δ⁵-Adiol = Δ⁵-androstene-3β,17β-diol, 5α-DHT = 5α-dihydrotestosterone. Enzymes are referred to by their gene names and are italicized: HSD3B1 = 3β-hydroxysteroid dehydrogenase type 1, SRD5A 5α-reductase type 1 or type 2, AKR1C2 = type 3 3α-hydroxysteroid dehydrogenase, and HSD17B6 = 17β-hydroxysteroid dehydrogenase type 6.

Enzalutamide (ENZ) is an AR superantagonist which is also used to treat CRPC patients,^31–33 and drug resistance also occurs with this agent. ENZ drug resistance in cell lines and in xenografts could be surmounted with indomethacin, a non-steroidal antiinflammatory drug (NSAID) and competitive inhibitor of AKR1C3 identified in this laboratory.^34,35 We are repurposing NSAIDs to be devoid of COX-1 or COX-2 inhibition while maintaining their inhibitory potency for AKR1C3. We have previously reported on indomethacin and fenamate analogues that have the requisite nanomolar inhibitory potency and selectivity for AKR1C3.^36,37 These agents do not inhibit COX isozymes, nor do they inhibit AKR1C1 and AKR1C2, which are required for the inactivation of 5α-DHT within the prostate.^38–40

Naproxen, (S)-2-(6-methoxynaphthalen-2-yl)propanoic acid is a NSAID that is used clinically to block cyclooxygenase (COX) mediated inflammation. It is also a potent AKR1C3 inhibitor that inhibits the AKR1C3 catalyzed reduction of the bioreductive drug PR-104 in multiple human cancer cells lines and a lung cancer xenograft model.⁴¹ However, naproxen also inhibits AKR1C2,³⁴ which limits its therapeutic potential in CRPC.

We set out to develop naproxen-based AKR1C3 inhibitors that would be useful in CRPC but are devoid of inhibitory activity on COX and other AKR1C enzymes. We report the discovery of (R)-2-(6-methoxynaphthalen-2-yl)butanoic acid as a competitive, potent, and selective AKR1C3 inhibitor which identifies R-profens as a new class of repurposed NSAIDs for CRPC patients. R-Profens are also substrate selective inhibitors of COX-2 oxygenation of endocannabinoids, raising the prospect that in the context of prostate cancer they may prevent intratumoral androgen biosynthesis and act as analgesics in metastatic disease.^42,43

RESULTS

Chemistry

S-Naproxen 1 was the precursor of compounds 5, 6, and 9–11 (Schemes 2 and 3). A common intermediate was S-methyl-2-(6-trifluoromethylsulfonyloxynaphthalene propanoate, 4. This compound was synthesized from S-naproxen, which was converted to O-demethyl naproxen 2 under acidic conditions followed by esterification to yield 3.⁴² Subsequent addition of trifluoromethane sulfonic anhydride in base gave 4. Compound 4 was converted to compound 5 via the 6-vinylogous intermediate and deesterification. The protected acid 3 was converted to the ethoxy intermediate using iodoethane and desesterification and gave compound 6 (Scheme 2). Coupling of 4 with sodium triisopropylsilanethiolate followed by deprotection with tetrabutylammonium fluoride gave (S)-2-(6-(methylthio)naphthalen-2-yl)propanoate 8, which upon base hydrolysis yielded the racemic acid, 9. The racemic acid was further oxidized with m-chloroperoxybenzoic acid to yield racemic 2-(6-methylsulfinyl)naphthalene-2-yl)-propanoic acid 10, Scheme 3. The steps to compound 11 are identical except the final oxidation of the methyl-thio derivative to yield the methylsufonyl derivative used potassium peroxymonosulfate (Oxone), Scheme 3. Compound 14 was synthesized in two steps from 2-bromo-6-methoxy-naphthalene 12.⁴² Compound 15 was synthesized from racemic 14 via 1,1′-carbonyldiimidazole coupling of methanesulfonamide, Scheme 3.

Scheme 2. Synthesis of Racemic Naproxen Analogues 2–6a.

^aReagents and conditions: (i) 48% HBr, AcOH, reflux, 3 h; (ii) TMSCl, CH₃OH, 25 °C, 2 h; (iii) (CF₃SO₂)₂O, Et₃N, DCM, 25 °C, 1 h; (iv) CH₂ = CHBF₃K, Cs₂CO₃, Pd(PPh₃)₄, Et₃N, EtOH, 50 °C, 16 h; (v) (OAc)₂Pd, t-But₃P, HCO₂H, 25 °C, 12 h; (vi) 3 M KOH/CH₃OH, reflux, 3 h; (vii) KOH/C₂H₅I, 25 °C, 30 min; (viii) 3 M KOH/CH₃OH, reflux, 2 h.

Scheme 3. Synthesis of Racemic Naproxen Analogues 7–11 and 13–15a.

^aReagents and conditions: (i) Pd(PPh₃)₄, [(CH₃)₂CH]₃SiSH, C₆H₆, reflux, 4 h; (ii) TBAF, CH₃I, 25 °C, 2 h; (iii) 3 M KOH/CH₃OH, reflux, 2 h; (iv) mCPBA, DCM 0 °C, 1 h; (v) KHSO₅·0.5KHSO₄· 0.5K₂SO₄, (CH₃)₂C = O/H₂O 25 °C, 2 h; (vi) Mg, I₂, THF, reflux, 1 h; (vii) CH₃CH₂CHBrCO₂CH₃, THF, reflux, 2 h; (viii) 3 M KOH/CH₃OH, reflux, 2 h, (ix) CDI, CH₃SO₂NH₂, DBU, DCM, 25 °C, 4 h.

AKR1C3 and AKR1C2 Inhibitor Screening

Naproxen analogues were synthesized to explore the effect of modifications of its structure on AKR1C3 activity and selectivity. On the basis of the observation that other NSAIDs require the carboxylic acid group for optimal inhibition of the AKR1C enzymes,⁴⁴ most of the synthesized compounds retained this functional group. Modifications were made primarily on the chiral α-carbon or the 6-position of the naphthalene ring. Because of the presence of the chiral center in the molecule, most of compounds were initially assayed as racemates and screened for AKR1C2 and AKR1C3 inhibition without separation of the enantiomers.

The ability of the compounds to inhibit the NADP⁺ dependent oxidation of S-tetralol catalyzed by AKR1C3 and AKR1C2 was determined and IC₅₀ values obtained, Figure 1 and Table 1. Because the S-tetralol assays were performed at K_m, the IC₅₀ values for the two enzymes were directly comparable. Naproxen inhibited AKR1C3 with an IC₅₀ value of 180 nM and was not selective for AKR1C3 over AKR1C2, consistent with previous studies.⁴⁵ As naproxen, 1 is an S-enantiomer, its R-enantiomer, 1a, was evaluated for AKR1C3 inhibition and selectivity. R-Naproxen inhibited AKR1C3 and AKR1C2 with IC₅₀ values of 50 nM and 2.75 µM, respectively, which makes it more potent and more selective for AKR1C3 than naproxen. Replacement of the 6-OMe group of naproxen with an –Et to give 5 or an –OEt to give 6 did not change the potency and selectivity for AKR1C3. On the other hand, an isosteric replacement of the 6-OMe with the thiomethyl group, –SMe, to give compound 9, led to a 3-fold increase in the inhibitory potency for AKR1C3 (IC₅₀ = 60 nM) over that seen with naproxen while the AKR1C2 inhibitory potency remained unchanged. This translated to 25-fold selectivity for AKR1C3 over AKR1C2 by compound 9. Because the –SMe group can be metabolized to the S(=O)Me and S(=O)₂Me groups, compounds 10 and 11 containing these functional groups at the 6-position were synthesized and screened for AKR1C3 activity and selectivity. Compared to 9, both compounds 10 and 11 displayed a greater than 15-fold loss of inhibitory potency on AKR1C3.

Figure 1.

Structure of naproxen analogues.

Table 1.

Structure and AKR1C3/AKR1C2 Inhibitory Potency of Naproxen Analogues

compd	R1	R2	R3	AKR1C3 IC50 (µM) ± SE (n = 4)	AKR1C2 IC50 (µM) ± SE (n = 4)	ratio IC50 value (AKR1C2:AKR1C3)
S-naproxen (1)	–OH	–Me	–OMe	0.18 ± 0.04	1.26 ± 0.15	7
R-naproxen (1a)	–OH	–Me	–OMe	0.05 ± 0.004	2.75 ± 0.35	56
(±)-5	–OH	–Me	–Et	0.12 ± 0.0075	1.9 ± 0.11	15
(±)-6	–OH	–Me	–OEt	0.10 ± 0.0076	2.4 ± 0.24	24
(±)-9	–OH	–Me	–SMe	0.060 ± 0.0054	1.5 ± 0.14	25
(R)-9a	–OH	–Me	–SMe	0.05 ± 0.0015	4.35 ± 0.61	87
(S)-9b	–OH	–Me	–SMe	0.07 ± 0.004	1.35 ± 0.09	19
(±)-10	–OH	–Me	–S(=O)Me	1.05 ± 0.13	6.3 ± 0.35	6
(±)-11	–OH	–Me	–S(=O)2Me	0.82 ± 0.1	3.0 ± 0.14	4
(±)-14	–OH	–Et	–OMe	0.12 ± 0.014	7.6 ± 0.95	58
(R)-14a	–OH	–Et	–OMe	0.11 ± 0.01	48.1 ± 5.0	437
(S)-14b	–OH	–Et	–OMe	0.12 ± 0.01	1.72 ± 0.37	14
(±)-15	–NHSO2Me	–Et	–OMe	6.0 ± 1.0	34 ± 1.8	6
16a	–OH	–H	–OMe	0.65 ± 0.11	19.04 ± 3.7	29

^aWindsor et al.⁶⁵

The N-(methylsulfonyl)acetamide analogue, 15, displayed a significant loss of inhibitory activity for AKR1C3 and AKR1C2, with IC₅₀ values of 6.0 and 34 µM, respectively, underscoring the need for a free carboxylic acid group for optimal inhibition of the AKR1C enzymes. Next, the replacement of the α-Me group of naproxen with an –Et gave 14, which was the most AKR1C3 selective racemate evaluated. With an IC₅₀ value of 120 nM against AKR1C3, compound 14 was more potent as an AKR1C3 inhibitor than naproxen and was 58-fold selective for AKR1C3 over AKR1C2. Because of its favorable properties, the racemic mixture of 14 was separated to give the R- (14a) and the S- (14b) enantiomers. The configuration of the enantiomers was inferred by comparing the elution order of the enantiomers with that of naproxen and R-naproxen when the same chiral column and mobile phase was used (see Supporting Information). Surprisingly, while the AKR1C3 inhibitory potency of the enantiomers was similar and not significantly different from the inhibitory potency of the racemate, the AKR1C2 inhibitory potency of the enantiomers was markedly different. The S-enantiomer inhibited AKR1C2 with an IC₅₀ of 1.72 µM, while the R-enantiomer displayed an IC₅₀ value of 46.4 µM against AKR1C2. This translated to 14- and 437-fold selectivity for AKR1C3 inhibition over AKR1C2, respectively. Eliminating the chiral center of naproxen by the removal of the α-Me group to give 16 led to a loss of inhibitory potency and selectivity for AKR1C3, indicating the importance in retaining the R-configuration. The R- and S- enantiomers of compound 9 were also examined based on the selectivity achieved with the R-enantiomer 14a. Compound 9a showed 87-fold selectivity for AKR1C3, which was 5-fold less than the selectivity observed with compound 14a, demonstrating that the –OMe was preferred over the more bulky –SMe.

Compound 14a was the most selective AKR1C3 inhibitor identified from the primary screen. When tested for inhibition of AKR1C1, it displayed selectivity for AKR1C3 over AKR1C1 inhibiting the latter with an IC₅₀ value of 50 µM, which translates to a 500-fold selectivity for AKR1C3 (Figure 2.) Similarly, selectivity was observed when the compound was screened against AKR1C4.

Figure 2.

Inhibitory effect of compound 14a on AKR1C1–4.

Effect of 14a on AKR1C2 Catalyzed Reduction of 5α-DHT

Compounds 14a and 14b were next evaluated for the ability to inhibit AKR1C2 catalyzed NADPH dependent reduction of 5α-DHT. This was necessary to establish the lack of inhibitory activity on AKR1C2 using its physiologically relevant ketosteroid substrate. In this reaction, compound 14b inhibited AKR1C2 in a dose-dependent manner over the concentration range of 3–100 µM, achieving 70% inhibition of enzyme activity at 100 µM (Figure 3). In contrast, 14a did not display any significant inhibition of AKR1C2 at <100 µM concentrations. Flufenamic acid (FLU), a nonselective inhibitor of the AKR1C enzymes, displayed about 70% inhibition of AKR1C2 at 10 µM concentration.

Figure 3.

Effect of compounds 14a (green) and 14b (blue) on AKR1C2 catalyzed reduction of 5α-DHT.

Mode of AKR1C3 Inhibition by 14a

The pattern of AKR1C3 inhibition by 14a was evaluated in by measuring the NADP⁺ dependent oxidation of S-tetralol and by measuring the NADPH dependent reduction of Δ⁴-androsten-3,17-dione. Compound 14a competitively inhibited the AKR1C3 catalyzed oxidation of S-tetralol, with a K_i value of 31 nM (Figure 4a). The same mode of inhibition was also observed when the reduction of Δ⁴-AD was monitored, albeit with a much higher K_i value of 750 nM (Figure 4b).

Figure 4.

(a) Competitive Inhibition of AKR1C3 catalyzed oxidation of S-tetralol by 14a. (b) Competitive Inhibition of AKR1C3 catalyzed Reduction of Δ⁴-AD (Δ⁴-androstene-3,17-dione) by 14a.

Inhibition of COX-1

Naproxen inhibited COX-1 with an IC₅₀ value of 61 nM. Relative to naproxen, 14b displayed a 30-fold loss of inhibitory potency on COX-1, with an IC₅₀ value of 1.93 µM (Figure 5). There was a profound loss of inhibitory activity on COX-1 activity by the respective R-enantiomers, compounds 1a and 14a. Both compounds displayed less than 20% inhibition of COX-1 activity at the highest inhibitor concentration (100 µM) tested. This is consistent with reported structure–activity relationship studies on the COX-1 inhibitory effects of naproxen and its analogues.

Figure 5.

Inhibition of COX-1 by naproxen analogues.

A comparison of inhibitory potency of the compounds against AKR1C3 and COX-1 (Table 2) shows that 14a was almost a thousandfold more selective for AKR1C3 over COX-1. Compound 14, which is a racemate of 14a and 14b, had no inhibitory activity on COX-2 at the maximum concentration of arachidonic acid tested as substrate (10 µM).

Table 2.

Inhibitory Potency of Compounds on AKR1C3 and COX-1

compd	AKR1C3 IC50 (µM)	COX-1 IC50 (µM)	COX-2 IC50 (µM)	ratio (COX-1 IC50:AKR1C3 IC50)a	ratio (COX-2 IC50:AKR1C3 IC50)
naproxen 1	0.18	0.061	0.90	0.34	15
R-naproxen 1a	0.05	>100	>25	>2000	>500
14a	0.11	>100	no inhibitiona	>910	∞
14b	0.12	1.93	no inhibitiona	16	∞

^aTaken from Duggan et al.⁴²

Effect on AR Reporter Gene Assay

Compound 14a was next evaluated for a direct effect on AR signaling. Compound 14a did not effect the trans-activation of the AR mediated by 5α-DHT. There was a concentration-dependent increase in luciferase activity when HeLa13 cells containing a stably transfected AR and an androgen response element driven luciferase gene were treated with increasing concentrations of 5α-DHT. The luciferase activity peaked at 0.3 nM 5α-DHT, however, there were no significant change in the 5α-DHT induced luciferase expression when cells were treated with increasing concentrations of 5α-DHT in the presence of 10 µM compound 14a, Figure 6.

Figure 6.

Effect of compound 14a on DHT induced AR gene expression. DHT alone (purple) and DHT plus 10 µM compound 14a.

Inhibition of AKR1C3-Mediated Production of Testosterone

LNCaP-AKR1C3 cells were used to probe the metabolism of Δ⁴-AD in the presence and absence of compound 14a. Radio-labeled [³H]-Δ⁴-AD was incubated with carrier Δ⁴-AD (100 nM) with LNCaP-AKR1C3 cells placed in medium supplemented with charcoal dextran stripped fetal bovine serum (CD-FBS) which is devoid of androgens. Androgens were extracted after 48 h and the aqueous fraction subject to β-glucuronidase treatment to liberate androgen conjugates that we have previously shown to exist due to the robust uridine-5′-diphospho-glucuronosyltransferse (UGT) activity in these cells. Metabolites were separated by radiochromatography using TLC. LNCaP-AKR1C3 cells were able to metabolize [³H]-Δ⁴-AD to [³H]-androsterone glucuronide and [³H]-testosterone glucuronide. However, in cells that were treated with 30 µM of compound 14a, the production of [³H]-testosterone glucuronide was significantly inhibited (Figure 7).

Figure 7.

Inhibition of testosterone formation in LNCaP-1C3 cells with compound 14a. (A) Conversion of 100 nM Δ⁴-AD to testosterone in LNCaP-AKR1C3 cells following digestion with β-glucurondiase. (B) Same experiment performed in the presence of 30 µM compound 14a. (C) Statistical analysis of n = 3 versus indomethacin as a positive control *p value <0.001; p value = 0.001.

Inhibition of AKR1C3-Mediated AR Gene Expression

Compound 14a was also tested for its ability to block the Δ⁴-AD mediated expression of PSA in LNCaP-AKR1C3 cells by Western blot analysis Figure 8. Treatment of these cells with 100 nM Δ⁴-AD led to a robust increase in PSA expression. This increase in PSA was reduced when the cells were treated with 100 nM Δ⁴-AD in the presence of 30 µM of compound 14a.

Figure 8.

Inhibition of Δ⁴-AD induced PSA expression in LNCaP-AKR1C3 cells by compound 14a. (A) Immunoblot. (B) Densitometric traces of immunoblots with normalization of PSA to β-tubulin for biological replicates (n = 3).

DISCUSSION

CRPC is currently treated with either AA or ENZ, however, patients rapidly develop drug resistance leading to an increase in median survival time of only 3–4 months. One mechanism of drug resistance is overexpression of AKR1C3. AKR1C3 represents a rational target due to its vital role as “gatekeeper” for the production of potent androgens regardless of the pathway used and its ability to function as a coactivator for the AR. This underscores the intense attention the search for AKR1C3 inhibitors has generated.^{30,36,37,46–53}

NSAIDs are known to be pan inhibitors of the AKR1C enzymes. This is thought to be as a result of the interaction of the carboxylate groups of NSAIDs with the catalytic site of the AKR1C enzymes.⁵³ The inhibition of AKR1C3 by NSAIDs is attained at therapeutic concentrations that are required for COX inhibition. This makes using these compounds as leads a promising approach because the resulting analogues may have similar pharmacokinetic profiles and therefore may be well tolerated. We have conducted structure–activity relationship studies on the NSAID naproxen and identified a R-enantiomer that differs from R-naproxen by the simple substitution of an ethyl group for a methyl group with therapeutic potential for CRPC.

The presence of small lipophilic groups at the 6-position of naproxen is optimal for COX inhibitory activity.^54,55 To determine the contribution of the 6-OMe group to the inhibitory potency of naproxen for AKR1C3, analogues with small substituents were synthesized and evaluated. With the exception of the 6-SMe, all the other substituents at the 6-position did not significantly improve AKR1C3 potency or selectivity.

All the compounds evaluated were superior inhibitors of AKR1C3 than AKR1C2. Because the AKR1C enzymes differ primarily in the enzyme subpockets, it is likely that the larger and more flexible subpockets of AKR1C3 allow for better interaction with the enzyme.^44,56

Naproxen, with a chiral center at the α-carbon, is used as the S-enantiomer for anti-inflammatory activity. This is due to the R-enantiomer being significantly less active as a COX inhibitor, Figure 5.^42,54 Naproxen is likewise a potent AKR1C3 inhibitor. However, R-naproxen, compound 1a, has never been evaluated for AKR1C3 inhibitory activity. Relative to naproxen, compound 1a was found to be more potent as an AKR1C3 inhibitor and less potent as an AKR1C2 inhibitor, which translated into an increased selectivity for AKR1C3 relative to AKR1C2. The higher affinity of 1a over naproxen for AKR1C3 is supported by crystallographic studies conducted by Flanagan et al.⁴⁸ These authors reported that while both S- and R-naproxen bind to AKR1C3 when the respective enantiomers were soaked separately into preformed crystals of the AKR1C3· NADP⁺ complex, only the R- enantiomer was bound in the these crystals when cosoaked with both enantiomers. This observation was attributed to the number of intermolecular interactions made by the enantiomers with the enzyme despite having relatively similar binding poses with AKR1C3. Relative to naproxen, R-naproxen made more contacts with side chains in the active site and cofactor, leading to a greater complementarity.⁴⁸ This would suggest the existence of stereochemical specificity in the interaction of R-profens with AKR1C3 and loss of specificity with AKR1C2. To further explore this observation, compound 14 with an α-Et group and its respective enantiomers (R)-14a and (S)-14b were subsequently evaluated. Surprisingly, the expected stereochemical specificity was not apparent in the interaction of the enantiomers 14a and 14b with AKR1C3 but was observed with AKR1C2 as the two enantiomers displayed markedly different inhibitory potency toward AKR1C2. The different inhibitory potency of the enantiomers on AKR1C2 was subsequently confirmed using a steroidal substrate for AKR1C2. Modeling studies of enantiomers 14a and 14b indicate that they adopt remarkably distinct binding poses with AKR1C2 (see Figures 9 and 10).

Figure 9.

Alignment of 14b and 14a in the AKR1C3 active site. AKR1C3 residues (green), 14b (yellow), 14a (purple), Dotted line: possible hydrogen bond. OX: oxyanion site (residues highlighted in pink). Ligand alignments were performed using LigAlign plugin in Pymol. The template crystal structures of the AKR1C3· NADP⁺ complexes were chosen from the RCSB Protein Data Bank (PDB: 3UFY and 3R58) (also see Supporting Information).

Figure 10.

Comparison of 14b and 14a binding to AKR1C2 and AKR1C3. 14b (yellow) and 14a (purple) binding modes in AKR1C2 (cyan) and AKR1C3 (green). Template crystal structure of the AKR1C2·NADP⁺ complex was taken from the RCSB Protein Data Bank (PDB: 4JQ1) (also see Supporting Information).

Ligand alignments of enantiomers 14a and 14b to the existing crystal structures (3UFY and 3R58) indicate that they adopt similar binding poses with AKR1C3 (Figure 9). Both analogues 14a and 14b show a similar binding mode to that of R-naproxen with the ethyl group occupying the subpocket 3 (SP3) formed by residues Tyr-24, Glu-192, Ser-217, Ser-221, and Gln-222.⁴⁷ The remainder of the molecule extends into the subpocket 1 (SP1) pocket composed of Ser-118, Asn-167, Phe-306, Phe-311, and Tyr-319. The ligand alignments revealed closer proximity of carboxylic group of 14a compared to 14b at the oxyanion site (Tyr-55, His-117, and the NADP⁺ cofactor). A slight increase in inhibitory potency and selectivity of 14a may be due to the favorable H-bond interaction with the oxyanion site.

Compounds 14a and 14b display remarkably different binding poses to each other with AKR1C2 (Figure 10). While analogue 14a binds in the active site of AKR1C2 similarly to AKR1C3, compound 14b further extends into the SP3 pocket with no interaction with the SP1 pocket. When the structures of AKR1C2 and AKR1C3 with either 14a or 14b bound were overlaid, additional reasons for the AKR1C3 selectivity of compound 14a were found. The Leu 308 side chain at the AKR1C2 binding site sterically clashes with 14a, which may significantly decrease its preference toward binding in the SP1 pocket of AKR1C2, resulting in a loss of inhibitory potency for AKR1C2.

Compound 14a competes with S-tetralol and the physiologically relevant steroidal substrate, Δ⁴-AD, for binding to AKR1C3, albeit with different binding constants. The difference in binding constant is related to the formation of two different inhibitor complexes.³⁴ During the competitive inhibition of S-tetralol, oxidation the E·NADP⁺·I complex forms (where I = inhibitor) to yield a K_i = 31 nM for 14a. By contrast, during the competitive inhibition of Δ⁴-AD reduction, the E·NADPH·I complex forms to yield a K_i = 750 nM for 14a. Thus, compound 14a displays a 20-fold preference for the E·NADP⁺ complex, Figure 11. Which complex is inhibited in cells is unclear.

Figure 11.

Kinetic mechanism of the reactions catalyzed by AKR1C3. (A) Oxidation reaction where the substrate S is S-tetralol and the product P is tetralone, and (B) reduction reaction where the substrate is Δ⁴-androstene-3,17-dione and the product is testosterone. S = substrate, P = product, and E = enzyme.

Indomethacin, another NSAID, has been shown to have greater affinity for the E·NADP⁺ complex, presumably due to the interaction between the negative charge on the compound and the positive charge of NADP⁺.⁴⁴ Because most AKR1C3 inhibitors that have a carboxylic acid group are anchored to the enzyme’s catalytic site via this functional group,^{37,47,48,56,57} compound 14a with a negatively charged carboxylic acid group at pH 7 will likewise be expected to have similar interaction. This would bring the negative charge of 14a close to the cofactor where it will form a stronger interaction with NADP⁺ relative to NADPH. The stronger interaction between 14a and NADP⁺ would account for the lower K_i value obtained for 14a during the oxidation of S-tetralol by AKR1C3 when compared to the value obtained during the reduction of Δ⁴-AD by AKR1C3.

Compound 14a, had no effect on the transactivation of the AR mediated by 5α-DHT in HeLa cells. These cells were also shown to express AKR1C3 endogenously (Adeniji and Penning, unpublished). Thus, if AKR1C3 acts as an AR coactivator in these cells, then compound 14a differs from GTx560 in that it does not block coactivator function.³⁰ This would suggest that not all competitive inhibitors of AKR1C3 are able to inhibit the coactivator function of this protein. However, coactivator function and its inhibition can be cell context dependent.

Consistent with the stereoselective inhibition of COX observed with naproxen, compound 14a was likewise devoid of inhibitory activity on COX, which is essential because chronic COX inhibition is not desirable in the context of CRPC management. We anticipate that due to the close structural similarity of naproxen with compound 14a, where the methyl group of R-naproxen was simply replaced by an ethyl group in compound 14a, that compound 14b would have favorable absorption, distribution, metabolism, elimination, and toxicological (ADMET) properties that would enable its rapid transit into man. However, it is recognized that the addition of the extra methylene group may effect the compounds’ lipohilicity and hence affect these properties.

We also note with interest that (R)-profens are substrate-selective inhibitors of endocannabinoid oxygenation by COX-2 and suggest that they will have analgesic benefit by this mechanism.^42,43 As a consequence, (R)-profens in CRPC may not only inhibit intratumoral androgen synthesis but may alleviate pain associated with the metastatic disease.

CONCLUSION

We have identified (R)-2-(6-methoxynaphthalen-2-yl)butanoic acid as a potent AKR1C3 inhibitor. This compound competitively inhibits AKR1C3 and displays selectivity for AKR1C3 over other AKR1C enzymes and COX. It was also efficacious at blocking AKR1C3 in LNCaP-AKR1C3 cells, which is a model for CRPC cells. This compound represents a potential therapeutic agent that can be used in the management of CRPC either alone or in combination with AA or ENZ, where it may improve efficacy and reduce the incidence of resistance to these drugs.

EXPERIMENTAL SECTION

Chemistry

General: All commercially available reagents and anhydrous solvents were ACS grade or better and were used as received.

Analytical thin-layer chromatography was carried out using glassbacked plates coated with fluorescent silica gel 60 F254 from Whatman (Partisil LK6D). Spots were visualized under natural light and UV illumination at λ = 254 and 365 nm. Flash chromatography was conducted on a Biotage SP1 automated flash chromatography system equipped with a fixed wavelength UV detector (λ = 254 nm). Samples were preabsorbed onto ready-made silica gel samplets and then applied on to normal-phase flash chromatography cartridges (Biotage KP-SIL, size according to requirements) and elution with a 0–100% EtOAc/hexane (0.5% acetic acid) gradient. ¹H and ¹³C NMR spectra were recorded at 400 and 100 MHz, respectively, using a Bruker AV-400 with sample changer (BACS 60). A purity of ≥95% (unless otherwise indicated) for the final compounds in this study was confirmed by analytical HPLC on a Waters HPLC system with PDA detector (set at λ = 254 nm) equipped with a Supelco Supelcosil LC-18 reverse-phase column (15 cm Å ~ 3 mm, 5 µm). Compounds screened in the current study also were characterized with respect to their mp and HRMS.

Preparation of Compounds

(S)-Methyl 2-(6-(trifluoromethylsulfonyloxy)naphthalen-2-yl)propanoate, 4

To (S)-naproxen (1, 3 g, 13 mmol) in acetic acid (20 mL) was added 48% HBr (11.2 g, 7.5 mL, 138 mmol) at 0 °C.⁵⁷ Following 3 h of reflux, water (40 mL) was added to precipitate out the product 2, which was isolated by filtration. To the crude product 2 was added methanol (35 mL) and TMSCl (1.7 g, 2.0 mL, 15.7 mmol), and the mixture was stirred at room temperature for 2 h.^58,59 The solvent was removed in vacuo, resulting in a tan solid 3, which was dissolved in CH₂Cl₂ (20 mL). Following the addition of triethylamine (2.6 g, 3.6 mL, 26.1 mmol) at 0 °C, trifluoromethanesulfonic anhydride (2.8 g, 2.7 mL, 15.7 mmol) was added dropwise and the mixture was warmed to room temperature and allowed to stir for 1 h. The mixture was then diluted with diethyl ether, quenched with 1 M HCl, and washed with saturated sodium bicarbonate and brine. The organic layer was dried over MgSO₄ and then concentrated, resulting in the desired product 4 (4.5 g, 95%). ¹H NMR (400 MHz, CDCl₃) δ 1.57 (d, 3H), 3.65 (s, 3H), 3.89 (q, 1H), 7.34 (dd, J = 4, 8 Hz, 1H), 7.52 (dd, J = 4, 8 Hz, 1H), 7.70 (d, J = 4 Hz, 1H), 7.77 (s, 1H), 7.81 (d, J = 8 Hz, 1H), 7.86 (d, J = 8 Hz, 1H). MS m/z: 363 (M + H)⁺.

2-(6-Ethylnaphthalen-2-yl)propanoic Acid, 5

(S)-Methyl 2-(6-(trifluoromethylsulfonyloxy)naphthalen-2-yl)propyl ester 4 (4.2 g, 11.6 mmol) and 1 M of cesium carbonate (20 mL) were added to a solution of potassium vinyltrifluoroborate (2.4 g, 17.5 mmol) in EtOH (100 mL). Then tetrakis(triphenylphosphine) palladium (672 mg, 0.59 mmol) and triethylamine (2.4 g, 3.2 mL, 23.8 mmol) were added to the mixture and stirred 50 °C for 16 h. The reaction mixture was cooled to room temperature, water (100 mL) was added, and the resulting mixture was stirred at room temperature for 30 min. The reaction mixture was filtered, resulting in (S)-methyl 2-(6-vinylnaphthalen-2-yl)propanoate as a brown solid (2 g), which was used directly to the next step without further purification. To a round-bottom flask purged with argon was added THF (35 mL), palladium(II) acetate (52 mg, 0.24 mmol), and tri-tert-butyl phosphine (94 mg, 113 mL, 0.46 mmol). This was brought to reflux and allowed to stir for 30 min. The reaction was then cooled, and (S)-methyl 2-(6-vinyl naphthalen-2-yl)propanoate (2 g, 8.9 mmol) and formic acid (2.7 g, 2.3 mL, 59 mmol) were added to the reaction mixture. The mixture was again brought to reflux for 30 min, then cooled to room temperature and allowed to stir for 12 h. The reaction mixture was filtered through a bed of Celite and then concentrated in vacuo to give the crude (S)-methyl 2-(6-ethylnaphthalen-2-yl)propanoate (700 mg), which was added to 14 mL of a 3 M KOH solution in MeOH. The reaction was held at reflux for 3 h, then cooled and quenched with water. The reaction mixture was extracted with ethyl ether (3×). The aqueous layer was then acidified with 1 M HCl and then extracted into ethyl ether (3×), washed with brine, dried with MgSO₄, and concentrated under reduced pressure to give a crude residue, which was purified by column chromatography (n-Hex:EtOAc 4:1) to afford the pure racemic product 5 (450 mg, yield 94% last step, purity 99.4%, HPLC retention time 11.93 min, melting point 120–122 °C). ¹H NMR (600 MHz, DMSO-d₆) 1.18 (t, J = 5.5 Hz, 3H), 1.38 (d, J = 6 Hz, 3H), 2.68 (q, J = 4.5 Hz, 2H), 3.75 (q, J = 4.5 Hz, 1H), 7.30 (dd, J = 7.5, 2.5 Hz, 1H), 7.34 (dd, J = 7.5, 2.5 Hz, 1H), 7.58 (s, 1H), 7.66 (s, 1H), 7.71–7.73 (m, 2H). ¹³C NMR (150.9 MHz, DMSO-d₆) 15.98, 18.90, 28.75, 45.19, 125.47, 125.94, 126.43, 127.74, 127.92, 128.03, 131.98, 132.70, 138.41, 141.66, 175.86. HRMS: m/z calcd for C₁₅H₁₆O₂ (M – H)⁻ 227.1078, found 227.1079.

2-(6-Ethoxynaphthalen-2-yl)propanoic Acid, 6

Potassium hydroxide (828 mg, 14.8 mmol) in 4 mL of methanol was added to a round-bottom flask. (S)-Methyl 2-(6-hydroxynaphthalen-2-yl)propyl methyl ester 3 (2.3 g, 9.8 mmol) in DMF (25 mL) was then added to the flask. The reaction was then allowed to stir at room temperature for 30 min. Iodoethane (3.0 g, 1.5 mL, 19.6 mmol) was added, which was allowed to stir at room temperature for 3 h. The reaction mixture was quenched with water and extracted with CH₂Cl₂ (3×), and the organic layer was then dried over MgSO₄, filtered, and finally concentrated in vacuo the crude product as a yellow solid (450 g), to which was added 8 mL of a 3 M KOH solution in MeOH. The reaction was held at reflux for 2 h, then cooled and quenched with water. The reaction mixture was extracted with ethyl ether (3×). The aqueous layer was then acidified with 1 M HCl and then extracted into ethyl ether (3×), washed with brine, dried with MgSO₄, and concentrated under reduced pressure to give pure racemic product as an off-white solid 6 (360 mg, yield 85% last step, purity 98%, HPLC retention time 11.28 min, melting point 148–150 °C). ¹H NMR (600 MHz, DMSO-d₆) 1.44 (t, J = 6 Hz, 3H), 1.49 (d, J = 6.5 Hz, 3H), 3.85 (q, J = 4.5 Hz, 1H), 4.19 (q, J = 4.5 Hz, 1H), 7.20 (dd, J = 2, 8.5 Hz, 1H), 7.32 (d, J = 2 Hz, 1H), 7.44 (dd, J = 2, 8.5 Hz, 1H), 7.76 (s, 1H), 7.78 (d, J = 8.5, 1H), 7.85 (d, J = 8.5, 1H). ¹³C NMR (150.9 MHz, DMSO-d₆) 15.10, 18.92, 45.06, 63.54, 106.83, 119.37, 126.00, 126.83, 127.28, 128.80, 129.56, 133.73, 136.74, 156.79, 175.95. HRMS: m/z calcd for C₁₅H₁₆O₃ (M – H)⁻ 243.1027, found 243.1026.

(S)-Methyl 2-(6-(Triisopropylsilylthio)naphthalen-2-yl)-propanoate, 7

To a solution of (S)-methyl 2-(6-(trifluoromethylsulfonyloxy)naphthalen-2-yl)propyl methyl ester 4 (3 g, 8.3 mmol) in benzene (20 mL) was added Pd(PPh₃)₄ (0.95 g, 0.8 mmol) and sodium triisopropylsilanethiolate⁶⁰ (2.1 g, 9.9 mmol) dissolved in THF (10 mL). The solution was refluxed for 4 h, then quenched with water and extracted with ethyl acetate. The organic layer was dried over MgSO₄ and concentrated. The crude product was purified by flash chromatography using 10:1 hexane:ethyl acetate to give pure product (2.9 g, 87%). ¹H NMR (400 MHz, CDCl₃) δ 1.10 (d, 18 H), 1.25 (m, 3H), 3.65 (s, 3H), 3.86 (q, 1H), 7.31 (m, 2H), 7.41 (dd, J = 4, 8 Hz, 1H), 7.54 (dd, J = 4, 8 Hz, 1H), 7.67 (m, 2H). MS m/z: 403 (M + H)⁺.

(S)-Methyl 2-(6-(Methylthio)naphthalen-2-yl)propanoate, 8

To 15 mL of THF was added (S)-methyl 2-(6-(triisopropylsilylthio)-naphthalen-2-yl)propanoate 7 (2.5 g, 6.2 mmol) followed by tetrabutylammonium fluoride (3.3 g, 3.6 mL, 12.4 mmol). The mixture was allowed to stir at room temperature for 2 h. Methyl iodide (2.6 g, 1.2 mL, 18.6 mmol) was then added, and the resulting mixture was stirred for an additional 2 h at room temperature. The reaction mixture was extracted with ethyl ether, dried over MgSO₄, and concentrated in vacuo. The crude product 8 was purified by flash chromatography using 7:1 hexane:ethyl acetate to give pure product (1.2 g, 75%). ¹H NMR (400 MHz, CDCl₃) δ 1.58 (d, 3H), 2.57 (s, 3H), 3.67 (s, 3H), 3.87 (q, 1H), 7.37 (dd, J = 4, 8 Hz, 1H), 7.42 (dd, J = 4, 8 Hz, 1H), 7.58 (s, 1H), 7.67 (s, 2H), 7.70 (d, J = 8 Hz, 1H). MS m/z: 261 (M + H)⁺

2-(6-(Methylthio)naphthalen-2-yl)propanoic Acid, 9

To (S)-methyl 2-(6-(methylthio)naphthalen-2-yl)propanoate 8 (975 mg, 3.7 mmol) was added 20 mL of a 3 M KOH solution in MeOH. The reaction was refluxed for 2 h, cooled, and then quenched with water. Following extraction with ethyl ether, the aqueous layer was acidified with 1 M HCl and then extracted into ethyl ether, washed with brine, dried with MgSO₄, and concentrated under reduced pressure to give pure product 9 (876 mg, yield 95%, purity 99.1%, HPLC retention time 11.39 min, melting point 172–174 °C). ¹H NMR (600 MHz, DMSO-d₆) δ 1.45 (d, J = 6.5 Hz, 3H), 2.57 (s, 3H), 3.82 (q, J = 5 Hz, 1H), 7.38 (dd, J = 2, 8.5 Hz, 1H), 7.45 (dd, J = 2, 8.5 Hz, 1H), 7.68 (s, 1H), 7.74 (s, 1H), 7.79–7.82 (m, 2H). ¹³C NMR (150.9 MHz, DMSO-d₆) 15.12, 18.85, 45.18, 63.54, 122.67, 125.84, 127.19, 127.32, 128.47, 131.14, 133.02, 136.13, 138.55, 175.79. HRMS: m/z calcd for C₁₄H₁₄O₂S (M – H)⁻ 245.0642, found 245.0637.

2-(6-(Methylsulfinyl)naphthalen-2-yl)propanoic Acid, 10

2-(6-(Methylthio)naphthalen-2-yl)propanoic acid 9 (65 mg, 0.26 mmol) was dissolved in 8 mL of dichloromethane. To this solution was added m-chloroperoxybenzoic acid (64 mg, 0.29 mmol), which was allowed to stir at 0 °C for 1 h. The reaction mixture was extracted with ethyl ether, dried over MgSO₄, and concentrated in vacuo. The crude product was purified by flash chromatography using 10:1 dichloromethane: methanol to give pure product 10 (46 mg, yield 62%, purity 98.6%, HPLC retention time 7.16 min, melting point 130–132 °C). ¹H NMR (600 MHz, DMSO-d₆) δ; 1.40 (d, J = 6.5 Hz, 3H), 2.74 (s, 3H), 3.83 (q, J = 5.5 Hz, 1H), 7.50 (dd, J = 2.5, 8.5 Hz, 1H), 7.63 (dd, J = 2, 8.5 Hz, 1H), 7.83 (s, 2H), 7.98 (dd, J = 2.5, 8.5 Hz, 1H), 8.16 (s, 1H). ₁₃C NMR (150.9 MHz, DMSO-d₆) 18.81, 43.50, 45.28, 120.83, 123.79, 126.45, 127.86, 127.86, 129.29, 131.10, 133.11, 141.20, 143.96, 175.59. HRMS: m/z calcd for C₁₄H₁₄O₃S (M + H)⁺ 263.0736, found 263.0741.

2-(6-(Methylsulfonyl)naphthalen-2-yl)propanoic Acid, 11

2-(6-(Methylthio)naphthalen-2-yl)propanoic acid 9 (82.5 mg, 0.33 mmol) was dissolved in 10 mL of acetone:water (2:1). To this solution was added Oxone (445.5 mg, 0.73 mmol), which was allowed to stir at room temperature for 2 h. The reaction mixture was extracted with ethyl ether (3×), dried over MgSO₄, and concentrated in vacuo to give pure product 11 (56 mg, yield 60%, purity 97.8%, HPLC retention time 8.78 min, melting point 180–182 °C). ¹H NMR (600 MHz, DMSO-d₆) δ 1.45 (d, J = 6 Hz, 1H), 3.24 (s, 3H), 3.92 (q, J = 5.5 Hz, 1H), 7.61 (dd, J = 2.5, 8.5 Hz, 1H), 7.91 (dd, J = 2.5, 8.5 Hz, 1H), 7.94 (s, 1H), 8.13–8.14 (m, 2H), 8.51 (s, 1H). ¹³C NMR (150.9 MHz, DMSO-d₆) 18.75, 44.05, 45.37, 122.93, 126.40, 128.34, 128.41, 129.78, 130.04, 131.16, 135.26, 138.16, 142.89, 175.44. HRMS: m/z calcd for C₁₄H₁₄O₄S (M + H)⁺ 279.0686, found 279.0926.

2-(6-Methoxynaphthalen-2-yl)butanoic Acid, 14

To a flask charged with 18 mL of THF and purged with argon was added magnesium ribbon (435.6 mg, 18 mmol) and a crystal of iodine. To this solution was added 2-bromo-6-methoxynaphthalene 12 (3.6 g, 15 mmol) in 18 mL of THF. An off-brown color appeared, and the reaction was allowed to reflux for 1 h. The reaction was then allowed to cool, and methyl 2- bromobutyrate (5.4 g, 3.4 mL, 30.2 mmol) was added. Following 2 h of reflux, the reaction was quenched with 1 M HCl, then extracted with ethyl ether (3×), washed with brine, dried with MgSO₄, and concentrated under reduced pressure. The crude mixture was suspended in 18 mL of a 3 M KOH solution in MeOH, held at reflux for 2 h, cooled, and then quenched with water. The reaction mixture was extracted with ethyl ether (3×). The aqueous layer was then acidified with 1 M HCl, extracted into ethyl ether (3×), washed with brine, dried with MgSO₄, and concentrated under reduced pressure to give crude product 14. The product was purified using HPLC on a C18 column (2 g, yield 55%, purity 99.8%, HPLC retention time 15.8 min, melting point 115–117 °C) to give pure racemic product. ¹H NMR (600 MHz, DMSO-d₆) δ 0.84 (t, J = 7 Hz, 3H), 1.74–1.77 (m, 1H), 2.02–2.06 (m, 1H), 3.52 (t, J = 5.5 Hz, 1H), 3.86 (s, 3H), 7.11 (dd, J = 2.5, 8.7 Hz, 1H), 7.28 (d, J = 2.5 Hz, 1H), 7.40 (dd, J = 2.5, 8.7 Hz, 1H), 7.71 (s, 1H), 7.76–7.80 (m, 2H). ¹³C NMR (150.9 MHz, DMSO-d₆) 12.56, 26.50, 53.01, 55.63, 106.17, 119.16, 126.69, 126.99, 127.31, 128.86, 129.57, 133.78, 135.24, 157.56, 175.40. HRMS: m/z calcd for C₁₅H₁₆O₃ (M – H)⁻ 243.1027, found 243.1024.

2-(6-Methoxynaphthalen-2-yl)-N-(methylsulfonyl)butanamide, 15

To an ice-cold mixture (0–5 °C) of 2-(6-methoxynaphthalen-2-yl)butanoic acid 14.⁴² (87 mg, 0.39 mmol) in dry CH₂Cl₂ (3 mL) under argon was added 1,1′-carbonyldiimidazole (63 mg, 0.39 mmol). After the reaction mixture was stirred for 2 h at 0–5 °C, methanesulfonamide (33 mg, 0.39 mmol) and diazabicyclo[5.4.0]-undec-7-ene (59 mg, 0.39 mmol) (1 equiv) were added. The mixture was left stirring for another 4 h at ambient temperature before it was quenched by the addition of glacial acetic acid (52 uL) and diluted with additional CH₂Cl₂ (1 mL). The organic layer was separated and washed with 10% NaH₂PO₄ buffer (pH 4) (2 × 3 mL) and water (3 × 3 mL), dried over Na₂SO₄, filtered, and concentrated under reduced pressure to give a crude residue. The raw product was purified by flash chromatography (SiO₂, ethyl acetate/hexane gradient) to afford the product (52 mg, yield 45%, purity 98.5%, HPLC retention time 11.20 min, melting point 144–146 °C) of the pure title compound. ¹H NMR (600 MHz, DMSO-d₆) δ 0.82 (t, J = 7.46 Hz, 3H), 1.73–1.75 (m, 1H), 2.02–2.06 (m, 1H), 3.18 (s, 3H), 3.61 (t, J = 7.04 Hz, 1H), 3.85 (s, 3H), 7.14 (dd, J = 9.5, 2.1 Hz, 1H), 7.27 (d, J = 2.2 Hz, 1H), 7.39 (dd, J = 9.2, 2.0 Hz, 1H), 7.69 (s, 1H), 7.77 (d, J = 9.7 Hz, 1H), 7.79 (d, J = 9.8 Hz, 1H). ¹³C NMR (150.9 MHz, DMSO-d₆) 12.37, 26.37, 41.35, 53.35, 55.61, 106.14, 119.22, 126.78, 126.84, 127.38, 128.75, 129.62, 133.88, 134.24, 157.67, 173.72. HRMS: m/z calcd for C₁₆H₁₉NO₄S (M – H)⁻ 320.0962, found 320.0966.

Determination of the Configuration of Naproxen Enantiomers 9a and 9b and 14a and 14b

The configuration of compounds was inferred by comparing the order of elution of the two enantiomers of compound 9 (9a and 9b) and the two enantiomers of compound 14 (14a and 14b) from chiral columns with the order of elution of naproxen (S-isomer) and R-naproxen from the same column. A mixture of R- and S-naproxen, R-naproxen, S-naproxen, compound 9, 9a, and 9b were resolved on a Chiral AD-RH column (150 mm × 4. 6 mm) using an isocratic elution method with a mobile phase of 60% HPLC grade water:40% acetonitrile containing 0.1% formic acid. By contrast, a mixture of R- and S-naproxen, R-naproxen, S-naproxen, compound 14, 14a, and 14b were resolved on a ChiralPak 1C (150 mm × 4.6 mm) column using an isocratic elution method. The mobile phase was 0.05%TFA in hexane:0.05% TFA in EtOH (98.5:1.5%) with a flow rate of 0.5 mL/min at room temperature (see Supporting Information). Absorbance was monitored at 230 nm except for compounds 9a and 9b where absorbance was monitored at 254 nm.

Ligand Alignment and Molecular Docking

The structural models of the ligand–enzyme interactions used in the ligand alignment and docking experiments are based on the existing crystal structures of AKR1C3·NADP⁺ complexed with R-naproxen, AKR1C3·NADP⁺ complexed with S-naproxen, and AKR1C2·NADP⁺ complexed with S-naproxen. In the first stage, R-naproxen was docked to the AKR1C2·NADP⁺ structure to generate AKR1C2·NADP⁺ complexed with R-naproxen because no crystal structure of this complex has been published. The second round of molecular modeling aligned 14a and 14b to their template ligand–enzyme structures. The results were overlaid to compare differences in the binding interactions of naproxen analogues in AKR1C2 and AKR1C3. For complete details, see Supporting Information

Enzyme Purification

Homogenous recombinant enzymes AKR1C1–4 were prepared and purified as previously described.⁶¹ Enzymes were purified to constant specific activity and their purity established by SDS-PAGE, see Supporting Information. All AKR enzymes were stored at −80 °C in 20 mM potassium phosphate buffer pH 7.0 containing 30% glycerol, 1 mM EDTA, and 1 mM β-mercaptoethanol. Under standard assay conditions, the specific activity of AKR1C1 for the NAD⁺ dependent oxidation of 1-acenaphthenol (Sigma) was 2.0 µmol min⁻¹ mg⁻¹. The specific activities of AKR1C2 and AKR1C3 for the NAD⁺ dependent oxidation of S-tetralol (Sigma) were 1.5 and 2.0 µmol min⁻¹ mg⁻¹, respectively, and the specific activity of AKR1C4 for the NAD⁺ dependent oxidation of androsterone (Steraloids) was 0.3 µmol min⁻¹ mg⁻¹ and are in agreement with previously published values. The specific activities of AKR1C enzymes were determined by measuring the formation of NADH at 340 nm using a Beckman DU640 spectrophotometer. A typical assay solution contained 100 mM potassium phosphate pH 7.0, 2.3 mM NAD⁺, 200 µM 1-acenaphthenol for AKR1C1, 3.0 mM S-tetralol for AKR1C2 and 1C3, and 75 µM androsterone for AKR1C4, 4% acetonitrile (v/v). The mixtures were incubated at 25 °C (AKR1C1 and 1C4) or 37 °C (AKR1C2 and 1C3) for 3 min, followed by adding a serial dilution of enzyme solution to a final volume of 1 mL to initiate the reaction. After continuously monitoring for 5 min, the increase in UV absorption using different concentrations of enzyme were recorded to calculate the initial velocity of substrate oxidation and determine enzyme-specific activity.

COX-1 was purified to homogeneity from ram seminal vesicles.⁶² The purity of the enzyme was assessed by SDS-PAGE (see Supporting Information). The specific activity of COX-1 for the conversion of arachidonic acid to prostaglandin H₂ (PGH₂) was coupled to the oxidation of N,N,N′,N′-tetramethyl-1,4-phenylenediamine (TMPD) and was found to be. 1.0 µmol min⁻¹ mg⁻¹. The specific activity of COX-1 enzyme was determined by measuring the formation of oxidized TMPD at 610 nm using Synergy 2 plate reader (BioTek). A typical assay solution (200 µL) contained 100 mM Tris-HCl (pH 8.0), 2 µM Hemin (Sigma), 5% DMSO, a serial dilution of COX-1 enzyme solution, 80 µM TMPD (Sigma), and 20 µM arachidonic acid (Sigma). Reagents were mixed and incubated at 25 °C for 5 min followed by adding a mixture of TMPD and arachidonic acid to initiate the reaction. Specific activity was determined based on the initial velocity of the reduction of PGG₂ formed in the reaction.

Enzyme Assays

S-Tetralol Oxidation Assay

The inhibitory potency of an individual compound against the AKR1C isoforms was determined by monitoring the NADP⁺ dependent oxidation of S-tetralol catalyzed by the AKR1C enzymes using a Synergy 2 plate reader (BioTek). Reaction systems (200 µL) contained 100 mM potassium phosphate buffer (pH 7.0), 4% DMSO, 200 µM NADP⁺, a serial dilution of compounds, S-tetralol, and AKR1C enzymes. The concentration of S-tetralol used in the inhibition assays using AKR1C1, 1C2, 1C3, and 1C4 was 5, 22.5, 165, and 25 µM, respectively, which was equal to their K_m values in order to make a direct comparison of IC₅₀ values. The concentration of AKR1C1, 1C2, 1C3, and 1C4 was 111, 86, 95, and 552 nM, respectively. Reagents were mixed and incubated at 37 °C for 10 min followed by adding AKR1C enzymes to initiate the reaction. A continuous fluorometric assay (Ex, 340 nm; Em, 460 mM) to measure NADPH formation was conducted at 37 °C for 5 min, and the IC₅₀ value of each compound was calculated as previously described. To determine the pattern of inhibition, five fixed concentrations of S-tetralol were used and four different concentrations of inhibitor were used. The equations for COMP, NONCOMP, and UNCOMP were globally fit to the initial velocity data using Grafit 5.0.6.

Reduction of Ketosteroids by Radiochemistry

Compounds were tested for their ability to inhibit the AKR1C3 catalyzed, reduction of [³H]-Δ⁴-AD and the AKR1C2 catalyzed reduction of [³H]-5α-DHT. Compounds were incubated with purified recombinant enzyme and radiolabeled steroid in 100 mM phosphate buffer, pH 7.0, and 4% DMSO at 37 °C. The reaction was initiated by the addition of 1 mM NADPH, and aliquots were removed over time and subsequently quenched by the addition of ice-cold ethyl acetate. The ethyl acetate fraction was extracted and dried in vacuo. Steroid reference standards and extracts were dissolved in 50 µL of ethyl acetate and applied to LK6D silica TLC plates (Whatman Inc., Clifton, NJ). TLC plates were developed using a dichloromethane/ethyl acetate (80:20 v/v) solution and were scanned with a Bioscan System 200 plate reader (Washington, DC). The percentage of radioactivity in the product peak was calculated assuming that the combined radioactivity in the substrate and product peaks was 100%. Initial velocities were computed by converting the percentage of radioactivity in the product peak by the specific radioactivity of the starting material to generate pmol/min. The pattern of AKR1C3 inhibition was determined using five fixed concentrations of [³H]-Δ⁴-AD using 5 concentrations of inhibitor. Equations for COMP, NONCOMP, and UNCOMP were fitted to the initial velocity data as described above.

COX-1 Assay

The effect of the compounds on COX-1 activity were determined by monitoring the oxidation of TMPD when coupled to COX catalyzed formation of prostaglandin H₂ (PGH₂) from PGG₂ using arachidonic acid as substrate. In brief, 200 µL of reaction solution was composed of 100 mM Tris-HCl (pH 8.0), 2 µM Hemin (Sigma), 5% DMSO, a serial dilution of compounds, COX-1 enzyme (175 nM), 80 µM TMPD (Sigma), and 20 µM arachidonic acid (Sigma). Reagents were mixed and incubated at 25 °C for 5 min followed by adding a mixture of TMPD and arachidonic acid to initiate the reaction. A continuous colorimetric assay to measuring TMPD oxidation at 610 nm was conducted using a Synergy 2 plate reader at 25 °C for 5 min, and IC₅₀ value of each compound was calculated as previously described.³⁷

Cell-Based Assays

AR-Luciferase Assay

HeLa13⁶³ cells stably expressing the AR and a luciferase reporter gene construct were used for this assay as previously described.^37,47 Briefly, cells were cultured in phenol red free media supplemented with 5% CDFBS (CSS media) for 48 h. The cells were then harvested and plated in a 96-well plate for 6–7 h, after which fresh CSS media (see below) containing 1 nM 5α-DHT in the presence of inhibitor was added. After 20 h incubation, the media was removed and the luciferase activity in the cells was measured using the Bright Glo kit (Promega) according to the manufacturer’s instructions. Fold induction of luciferase was evaluated relative to untreated cells.

Western Blots

LNCaP cells stably expressing AKR1C3 (LNCaP-AKR1C3 cells)⁶⁴ seeded at a density of 1.5 × 10⁶ cells were plated in 6 cm dishes containing phenol red free RPMI-1640 media supplemented with 5% CDFBS, 1% Pen/Strep, and 2 mM l-glutamine (CSS media). The cells were incubated for 24 h, after which media was aspirated and fresh CSS media containing 100 nM Δ⁴-AD plus and minus inhibitors. The cells were incubated for 24 h, after which they were harvested in RIPA lysis buffer supplemented with protease inhibitors at 4 °C. Lysate protein concentration was determined by the Bradford assay using the BIORAD protein dye (Bio-Rad). The samples were subjected to electrophoresis on a 12% SDS-PAGE gel that was subsequently transferred to a nitrocellulose membrane. The membrane was probed with the appropriate antibodies as follows: anti-β-tubulin (Millipore, no. 05-661) and anti-PSA (Meridian Life Science, Inc., no. K92110R). Blots were imaged using the ECL reagent (Pierce) and GelDoc XR+ system and Image Lab software (Bio-Rad).

Radiometric Assay of Androgen Metabolism

LNCaP-AKR1C3 cells seeded at a density of 1.5 × 10⁶ cells were plated in 6-well plates in CSS media. The cells were allowed to incubate for 24 h, after which the media was aspirated and fresh CSS media with 7.5 nM (1.26 µCi) of [³H] Δ⁴-AD and 92.5 nM of cold Δ⁴-AD (to obtain a final concentration of 100 nM) was added in the presence and absence of 30 µM of compound 14a in each well. The cells were incubated for 48 h, after which time the media was collected for analysis as previously described.⁶³ The media was extracted twice with cold ethyl acetate. To determine radioactivity of each phase, a portion of each fraction was added to Ultima Gold (PerkinElmer Life Sciences) scintillation fluid and analyzed on a TriCarb 2100 (Packard Instruments, PerkinElmer Life Sciences) and all CPM counts were blank-adjusted and reported as corrected cpm. The aqueous phases were acidified to pH 6.5 with acetic acid and subjected to treatment with 400 U of Echerichia coli β-glucuronidase at 37 °C for 24 h. The deconjugated androgens were re-extracted as described above, dried in vacuo, and redissolved in 100 µL of ethyl acetate for separation on the multichannel LK6D Silica TLC plates. Plates were developed using dichloromethane/ethyl acetate (80:20 v/v) and scanned on the BioScan 200 plate reader.

ABBREVIATIONS USED

Δ⁴-AD

Δ⁴-androstene-3,17-dione

AA

abiraterone acetate

Adione

5α-androstane-3,17-dione

ADT

androgen deprivation therapy

AKR1C3

aldo-keto reductase 1C3

AR

androgen receptor

COX

cylcooxygenases (PGH₂-synthase I and II)

CRPC

castration resistant prostate cancer

5α-DHT

5α-dihydrotestosterone

ENZ

enzalutamide

NSAID

nonsteroidal anit-inflammatory drug

PSA

prostate specific antigen

TLC

thin layer chromatography