Synthesis and structure-activity relationships for 1-(4-(piperidin-1- ylsulfonyl)phenyl)pyrrolidin-2-ones as novel non-carboxylate inhibitors of the aldo-keto reductase enzyme AKR1C3

Article abstract of DOI:10.1016/j.ejmech.2013.01.047

High expression of the aldo-keto reductase enzyme AKR1C3 in the human prostate and breast has implicated it in the development and progression of leukemias and of prostate and breast cancers. Inhibitors are thus of interest as potential drugs. Most inhibitors of AKR1C3 are carboxylic acids, whose transport into cells is likely dominated by carrier-mediated processes. We describe here a series of (piperidinosulfonamidophenyl)pyrrolidin-2-ones as potent (<100 nM) and isoform-selective non-carboxylate inhibitors of AKR1C3. Structure-activity relationships identified the sulfonamide was critical, and a crystal structure showed the 2-pyrrolidinone does not interact directly with residues in the oxyanion hole. Variations in the position, co-planarity or electronic nature of the pyrrolidinone ring severely diminished activity, as did altering the size or polarity of the piperidino ring. There was a broad correlation between the enzyme potencies of the compounds and their effectiveness at inhibiting AKR1C3 activity in cells.

Full text of DOI:10.1016/j.ejmech.2013.01.047

European Journal of Medicinal Chemistry 62 (2013) 738e744
Contents lists available at SciVerse ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Synthesis and structureeactivity relationships for 1-(4-(piperidin-1-ylsulfonyl)
phenyl)pyrrolidin-2-ones as novel non-carboxylate inhibitors of the aldo-keto
reductase enzyme AKR1C3
,
,
Daniel M. Heinrich a, Jack U. Flanagan a b, Stephen M.F. Jamieson a b, Shevan Silva a, Laurent J.M. Rigoreau c,
Elisabeth Trivier c, Tony Raynham c, Andrew P. Turnbull d, William A. Denny a
,b,
*
a Auckland Cancer Society Research Centre, The University of Auckland, Private Bag 92019, Auckland 1142, New Zealand
b Maurice Wilkins Centre for Molecular Biodiscovery, The University of Auckland, Private Bag 92019, Auckland 1142, New Zealand
c Cancer Research Technology Ltd, Wolfson Institute for Biomedical Research, The Cruciform Building, Gower St., London, WC1E 6BT, UK
d Cancer Research Technology Ltd, Birkbeck College, University of London, London, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
High expression of the aldo-keto reductase enzyme AKR1C3 in the human prostate and breast has
implicated it in the development and progression of leukemias and of prostate and breast cancers. In-
hibitors are thus of interest as potential drugs. Most inhibitors of AKR1C3 are carboxylic acids, whose
transport into cells is likely dominated by carrier-mediated processes. We describe here a series of
(piperidinosulfonamidophenyl)pyrrolidin-2-ones as potent (<100 nM) and isoform-selective non-car-
boxylate inhibitors of AKR1C3. Structureeactivity relationships identified the sulfonamide was critical,
and a crystal structure showed the 2-pyrrolidinone does not interact directly with residues in the
oxyanion hole. Variations in the position, co-planarity or electronic nature of the pyrrolidinone ring
severely diminished activity, as did altering the size or polarity of the piperidino ring. There was a broad
correlation between the enzyme potencies of the compounds and their effectiveness at inhibiting
AKR1C3 activity in cells.
Received 16 October 2012
Received in revised form
24 January 2013
Accepted 27 January 2013
Available online 9 February 2013
Keywords:
Aldo-keto reductase
Inhibitor
Crystal structure
Computer aided drug design
Ó 2013 Elsevier Masson SAS. All rights reserved.
1. Introduction
prostaglandin F synthase activity [3]. This has made them inter-
esting targets for the development of inhibitory small-molecule
Enzymes belonging to the aldo-keto reductase (AKR) 1C sub-
family are important modulators of both the steroid hormone and
the leukotriene cascades. Of the AKR1C enzymes, the C3 isoform
(also known as type 5 17-b-hydroxysteroid dehydrogenase) is of
particular interest because of its high expression in the human
prostate and breast, where it is responsible for producing (and
over-producing) testosterone, 17-b-estradiol and 20-a-hydrox-
yprogesterone [1]. As such, AKR1C3 is implicated in the develop-
ment and progression of leukemias and of prostate and breast
drugs that could be of potential use in controlling these diseases,
and a number of such compounds have been reported [4,5].
Important issues in such drug development are obtaining
selectivity between the four isoforms AKR1C1-4, which play dif-
ferent biological roles [6] and between these and the cyclo-
oxygenase (COX) enzymes. The latter is highlighted by the fact
that many inhibitors of the AKR1C subfamily are structurally rela-
ted to the NSAID class of COX inhibitors, such as the anthranilates
flufenamic [7] (1), meclofenamic [8] (2) and mefenamic [9] (3) acids
(Table 1). We have determined the crystal structures for a range of
NSAIDs bound in the AKR1C3 active site [10,11] that show, along
with flufenamic acid [7], that the drug carboxylate group occupies
the oxyanion hole, making H-bonds to the Y55 and H117 residues.
cancers [2], and to
a lesser extent leukemias through its
Abbreviations: AKR, aldo-keto reductase; COX, cyclo-oxygenase; DCM,
dichloromethane; DMSO, dimethyl sulfoxide; NADPH, nicotinamide adenine
dinucleotide phosphate; NSAID, non-steroidal anti-inflammatory drugs; PDB, Pro-
tein Data Bank; ROCS, rapid overlay of chemical structures; TEA, triethylamine; TFA,
trifluoroacetic acid.
* Corresponding author. Auckland Cancer Society Research Centre, The University
of Auckland, Private Bag 92019, Auckland 1142, New Zealand. Tel.: þ649 923 6144.
E-mail address: b.denny@auckland.ac.nz (W.A. Denny).
We have also recently shown [4] that
a series of (dihy-
droisoquinolyl)sulfonamidobenzoic acids (e.g., 4) are potent and
very selective inhibitors of AKR1C3 (Table 1), with the carboxylate
binding similarly in the oxyanion hole of the enzyme.
The majority of known AKR1C inhibitors are carboxylic acids,
whose transport into cells is likely dominated by carrier-mediated
0223-5234/$ e see front matter Ó 2013 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ejmech.2013.01.047

D.M. Heinrich et al. / European Journal of Medicinal Chemistry 62 (2013) 738e744
739
Table 1
Potency and selectivity of carboxylate-based AKR1C inhibitors.
Table 2
Variation of the lipophilic binding moiety.
No
Fm
X
AKR (IC50
1C1
, m
M)a
1C2
1C3
1C4
No
IC50 (m
M)a
9
A
A
A
>30
>30
>30
0.052
>30
>30
AKR1C1
AKR1C2
AKR1C3
AKR1C4
COX1b
COX2b
1
2
3
4
2.64
3.16
3.91
20.3
3.14
8.74
6.97
>30
0.41
0.54
0.56
0.013
>100
>100
>100
>30
3.0
0.22
25
9.3
0.7
2.9
>10
10
11
>30
>30
0.14
>10
>30
a
IC50 values are the average of 2 or more determinations.
COX data for compounds 1e3 from Ref. [18]; for compound 4 from Ref. [4].
b
12
A
>30
0.094
>30
processes rather than passive diffusion [12], and thus difficult to
predict. Therefore, we were interested that one of the hits from
a high-throughput screen seeking novel and selective inhibitors of
AKR1C3 was the pyrrolidine 12, which was a very potent (IC50
13
14
A
A
11.88
0.088
0.094 mM) and selective inhibitor [4]. Furthermore, in a cell-based
assay evaluating analogues for their ability to block AKR1C3 from
metabolising a known dinitrobenzamide substrate [4], 12 was more
potent relative to its enzyme inhibitory activity than a carboxylic
acid analogue (ratio IC50(enz)/IC50(cell) is 0.48 for 4 versus 8.5 for
12), suggesting a pharmacological disadvantage for the acids in
cells. We therefore undertook a study of the structureeactivity
relationships around the pyrrolidine lead 12, and an investigation
of its binding to the enzyme.
>30
>30
>30
15
16
A
A
>30
>30
>30
>30
0.084
0.20
>30
>30
2. Chemistry
Compounds of Table 2, where the amide-containing ring was
varied, were prepared from the halogenated sulfonamide in-
termediates 38aek, which were in turn prepared from known
sulfonyl chlorides 36a,b and substituted piperidines 37aek
(Scheme 1). Reaction of 38aek with pyrrolidinone in the pres-
ence of CuI, N,N’-dimethylethylenediamine and K2CO3 gave com-
pounds 10, 12e21 in moderate yields. Compound 11 was prepared
similarly from sulfonamide 39. Compounds 22 and 23, where the
sulfonamide was replaced by amide and amine respectively, were
prepared by coupling the benzoyl chloride 40 and benzyl bromide
41 with 2-methylpyrrolidine, and reacting the resultant in-
termediates 42, 43 with 2-pyrrolidinone as above (Scheme 1).
The compounds of Table 3 fix the 2-methylpiperidine ring of 12
and vary the key pyrrolidone unit. Coupling of 2-methylpiperidine
37a with substituted benzenesulfonic acids 36a,b,iep gave the
corresponding sulfonamide intermediates 38a,b,iep, which were
then reacted with 2-pyrrolidinine as before to give compounds 24e
30 (Scheme 2). Reduction of the 4-nitro intermediate 38q gave the
amine 38r, which was reacted with succinic or phthalic anhydride,
pivaloyl or isobutyryl chloride or 3-chlorosulfonyl chloride to give
respectively compounds 31e35 of Table 3.
17
A
>30
>30
0.056
>30
18
19
20
A
A
A
>30
>30
>30
>30
8.98
1.55
>30
>30
>30
21
A
21.2
22
23
B
B
C]O
CH2
>30
>30
a
IC50 values <10 mM are the mean for 2 or more determinations.
inhibitory activity of the compounds against the AKR1 isoforms
C1eC4 was performed with a competitive fluorescence assay,
where a non-fluorescent ketone probe (probe 5) [12] selective for
the AKR1C enzyme isoforms is reduced to a fluorescent alcohol in
the presence of AKR1C enzyme and NADPH. Compounds with
biochemical IC50 values <100 nM were tested for activity at
3. Enzyme biochemistry and cell biology
The structures of the new compounds and their activities in the
enzyme and cellular assays are shown in Tables 2 and 3. The

740
D.M. Heinrich et al. / European Journal of Medicinal Chemistry 62 (2013) 738e744
Table 3
Variation of the pyrrolidinone binding unit.
No
X
AKR (IC50
1C1
, m
M)a
1C2
1C3
1C4
24
25
>30
>30
>30
5.61
>30
>30
3.77
>30
26
27
>30
>30
28
29
30
31
10.2
>30
2.77
>30
Scheme 1. Synthesis of the compounds of Table 1. Reagents and conditions: (i) DCM,
TEA, piperidine, 20 ꢀC; (ii) pyrrolidin-2-one, CuI, N,N’-dimethylethylenediamine,
K2CO3, 100 ꢀC, 18e24 h; (iii) pyrrolidine, TEA, DCM, 20 ꢀC, 1 h; (iv) 2-methylpiperidine,
TEA, DCM, 20 ꢀC, 1 h (40)/K2CO3, tol, reflux, 3 h (41).
preventing the AKR1C3-dependent aerobic reduction of the dini-
trobenzamide PR-104A to its hydroxylamine metabolite in an
HCT116 cell line that overexpresses AKR1C3. Compound IC50 values
were calculated by fitting the inhibition data to a four-parameter
>30
>30
>30
32
>30
33
34
4-NHCOtBu
4-NHCOiPr
>30
>30
>30
>30
5.94
12.5
>30
>30
35
5.22
a
As for Table 2.
logistic sigmoidal doseeresponse curve using Prism 5.02 (Graph-
Pad, La Jolla, CA, USA).
4. Crystallography
The structure of 12 bound in the AKR1C3 active site was
determined by X-ray crystallography (PDB code 4H7C), as detailed
previously [4]. This showed that unlike the carboxylate containing
inhibitors with known binding modes, the pyrrolidinone unit made
no direct contact to the oxyanion hole (Fig. 1), while the sulfo-
nylpiperidine was located in the same pocket as the tetrahy-
droquinoline of 4 [4].
Scheme 2. Synthesis of the compounds of Table 2. Reagents and conditions: (i) DCM,
TEA, piperidine, 20 ꢀC; (ii) CuI, N,N’-dimethylethylenediamine, K2CO3, dioxane, 100 ꢀC,
18e24 h; (iii) EtOAc, 10% Pd/C, H2 (5 atm), 20 ꢀC, 8 h; (iv) DCM, TEA, 20 ꢀC.

D.M. Heinrich et al. / European Journal of Medicinal Chemistry 62 (2013) 738e744
741
the crystal structure. Compounds 15 and 16 probe the positioning
of the methyl group; the 3-methyl analogue 15 is as potent as 12,
but the 4-methyl analogue 16 loses some potency.
With the crystal structure in hand, we next sought to identify
sites for incorporating additional methyl and hydroxyl units around
the 2R-piperidine ring. A series of analogues that investigated these
substitutions at the 5- and 6-positions were modelled into the
active site by superimposition onto 12 using ROCS (Rapid Overlay of
Chemical Structures, Openeye Scientific Software), and energy
minimised within the active site (the structures considered are
illustrated in Table S2). Substitution patterns within the methyl or
hydroxyl series with the lowest ligand-protein energy guided fur-
ther analogue synthesis. Data for the analogues synthesised is
presented in Table 2. The 2S,6S-dimethyl analogue 17 is as potent as
the lead 12, and has the lowest interaction energy within the
dimethyl substituted series. The activity data for compounds 18 and
19 indicate that a hydroxyl group in this region of the active site is
not favoured, although there is an enantiomeric preference, with
stereoisomer 19 preferred over 18. This preference is consistent
with the energy values calculated for similar compounds with
equivalent hydroxyl substitutions (Table S2).
Compounds 20e21 explore polarity in the ring, which is clearly
deleterious, with the morpholine 20 and piperazine 21 losing all
activity. Finally, the role of the sulfonamide linker was examined
with the corresponding amide and amine 22 and 23. As expected,
these compounds were completely inactive, consistent with the
sulfonyl providing a rigid directionality in the molecule that pre-
arranges it for the AKR1C3 active site.
Fig. 1. Crystal structure of 12 complexed with AKR1C3. Protein Data Bank (PDB) code
4H7C. See Supplementary Material Table S1 for X-ray data collection and refinement
information. Compound 12 is shown in magenta sticks, while. While key amino acids
that define the pyrolidine-pheny-sulfone moiety binding site are labelled and shown
in yellow sticks. Water molecules are represented as red spheres. Electron density of
the ligand, NADP co-factor, water molecules and amino acids is also shown as a mesh.
Interactions between the pyrrolindyl group and a water molecule, as well as the li-
gands sulponyl moiety and Ser118. (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)
The compounds of Table
3 examine the role of the 2-
pyrrolidinone. The crystal structure of 12 bound to the enzyme
(Fig. 1) shows that while this does not make direct H-bond contacts
with the Y55 and H117 residues that constitute the “oxyanion hole”
as does the carboxylate of 4, the pyrrolidinone does occupy the SP3
pocket. The nature and positioning of the pyrrolidinone is therefore
expected to be important, and this was confirmed. The ring size of
the cyclic amide is critical, with the azetidinone 24 and piperidone
25 having IC50s for AKR1C3 respectively 60-fold and 40-fold less
than that of 12. Moving the pyrrolidinone from the 4-position to
either the 3-position (26) or the 2-position (27) lead to complete
5. Results and discussion
The crystal structure of 12 bound in the AKR1C3 active site
revealed that to accommodate the pyrrolidinone group at the 4
position of the phenylsulfone, compared to a carboxylate at the 3
position, the (methylsulfonyl)piperidine-phenyl was displaced
further into the SP1 pocket [1] compared to 4 where it can interact
with the side chain of Ser118. The compounds of Table 2 explore the
role of the sulfonamide substituent and probe its fit to the enzyme
hydrophobic pocket bounded by residues Met120, Asn167, Tyr216,
Phe306, Phe311, Tyr317, Pro318 and Tyr319 (Fig.1). In the acid series
the dihydroisoquinoline appeared to be the preferred unit, as
removal of the benzene ring caused a substantial (30-fold) loss of
potency against AKR1C3 [3]. This is not the case in the 2-
pyrrolidinone series, where there is only a 2.7-fold loss in po-
tency going from the dihydroisoquinoline 9 to the piperidine 10,
and only 1.8 between 9 and the (racemic) 2-methylpiperidine 12.
This is of importance since the series is relatively insoluble com-
pared to the previous acid series, and going from 9 to 10 markedly
lowers the lipophilicity (Clog P from 3.35 to 2.49 respectively).
However, there is a very marked dependence on ring size, with the
loss of activity (IC50s > 30 mM). Keeping the pyrrolidinone at the 4-
position but altering its conformation by steric crowding from
a flanking 2-methyl group on the central benzene ring (compound
28) diminished activity very severely. The corresponding 2,5-
dimethyl analogue 29, where the sulfonamide conformation may
also be affected, was completely inactive. Interestingly, while
altering the electronics of the central ring by making it a pyridine
(30) also reduced activity, it was less severe. However, altering the
electronics (rather than the conformation or position) of the pyr-
rolidine abolishes activity (pyrrolidine-2,5-dione 31, indoline-1,3-
dione 32). Linear amides 33 and 34 retain some, but greatly
diminished, activity, as does the isothiazolidine dioxide 35. Overall,
this SAR supports the crystal structure, suggesting that positioning
of the pyrrolidine oxygen is constrained by the SP3 pocket and
required for the compound to make the necessary H-bond to the
structured water molecule.
pyrrolidine analogue 11 being completely inactive (IC50 > 30 mM).
The dihydroisoquinoline unit of 9 may have an improved inter-
action with the SP1 pocket compared to the 2-methylpiperidine of
12 and contribute to the 1.8 fold increase in potency of 9 compared
to 12, although both compounds are 4 and 7 fold less potent than 4
respectively. This lower activity may result from the loss of an
oxyanion hole interaction, along with the displacement of the
butterfly shaped sulfonamide core common to the compounds.
Exploration of the chirality of the methyl substituent on 11 showed
The selectivity for AKR1C3 over other human AKR1C isoforms
was determined for all compounds with IC50 values <10
mM, but
none were active at inhibiting AKR1C1, 2 or 4 (IC50 values > 30
m
M)
(Tables 2 and 3). Representative compounds 10 and 12 were also
evaluated for inhibition of the cyclooxygenase enzymes COX-1 and
COX-2, and were found to be inactive at 10 mM.
All compounds with AKR1C3 IC50 values <100 nM were eval-
uated for their ability to block enzyme activity in human HCT-116
colon cancer cells engineered to over-express AKR1C3. Inhibitory
activity was assessed by measuring the inhibition of AKR1C3
that all the activity resides in the R-enantiomer 14 (IC50 0.088 mM),
which is 135-fold more potent than the corresponding S-enan-
tiomer 13, this is consistent with the R-enantiomer determined in

742
D.M. Heinrich et al. / European Journal of Medicinal Chemistry 62 (2013) 738e744
catalysed metabolism of an exogenous dinitrobenzamide substrate,
PR-104A, exclusively metabolised by AKR1C3 under aerobic con-
ditions to its cytotoxic hydroxylamine (PR-104H) and amine (PR-
104M) metabolites [14]. As shown previously for a series of
carboxylate-based inhibitors (e.g., 4) [4], the effectiveness of com-
pounds at inhibiting AKR1C3 activity in these cells broadly corre-
lated with their enzymic activity, but in the present case the
compounds were 2e8 fold more potent in the cellular assay
(Table 4). This contrasted with results for the carboxylate-based
series [3], which were almost universally less potent in the cell-
based assay. We speculate that this may be due to poorer passive
uptake of the carboxylate series.
reaction was stirred at 20 ꢀC until completion (TLC). The solvent
was removed under reduced pressure and the crude product was
purified by column flash-chromatography on silica to give the
sulfonamide.
B: Aryl halide (1 eq.), CuI (1 eq.), N,N’-dimethylethylenediamine
(2 eq.), amide (12 eq.) and K2CO3 (3 eq.) were dissolved in dry
dioxane (0.2 M) and heated at 100 ꢀC for 18e24 h. The mixture was
cooled to 20 ꢀC and solvent was removed under reduced pressure.
The crude product was purified as for procedure A to give the
amide [15].
C: Pd(OAc)2, (0.01 eq.), Xantphos (0.015 eq.), amide (1.2 eq.), aryl
halogen (1.0 eq.) and 1,4 dioxane (1 M) were heated at 100 ꢀC for
24 h. The mixture was cooled to 20 ꢀC, filtered, the solvent
evaporated under reduced pressure and the crude product was
purified as for procedure A to give the amide [16].
6. Conclusions
D: The aryl amine was dissolved in DMSO (0.8 M) and anhy-
dride/acid chloride added to the solution and the mixture was
heated for 12e18 h at 120 ꢀC. The reaction was allowed to cool to
20 ꢀC, diluted with EtOAc and washed with water, and the crude
product was purified as for procedure A to give the amide [17].
E: The compound was dissolved in EtOAc (0.2 M) and 10% Pd/C
(10 wt%) was added. The resulting mixture was stirred under H2
(5 atm) for 8 h at 20 ꢀC. The mixture was filtered through celite, the
solvent evaporated and the crude was purified for procedure A to
give the amine.
SAR studies of potent and selective non-carboxylate AKR1C3
inhibitor 12 showed that while the sulfonamide was still critical (cf
12 with 31,32), as in the carboxylate series represented by 4 [4],
there was much more tolerance for the sulfonamide substituent,
with a range of monocyclic six-membered ring analogues retaining
activity and AKR1C selectivity (Table 2). Crystal structure studies
show that the 2-pyrrolidinone was located in the SP3 pocket but
did not bind to the oxy-anion hole, and variations in the position,
co-planarity or electronic nature of the pyrrolidinone ring abol-
ished or severely diminished activity. The effectiveness of com-
pounds at inhibiting AKR1C3 activity in cells broadly correlated
with their enzyme inhibitory activity (Table 4).
7.3. Compounds of Table 2
7.3.1. 1-(4-(Piperidin-1-ylsulfonyl)phenyl)pyrrolidin-2-one (10) by
procedures A and B (Scheme 1A)
7. Experimental protocols
4-Bromobenzenesulfonyl chloride (36a) (1 g, 3.90 mmol, 1 eq.)
was dissolved in a mixture of dry DCM (50 ml, 0.2 M) and TEA
7.1. General conditions
(591 ml, 5.85 mmol, 1.5 eq.). piperidine (52a) (460 ml, 4.68 mmol,
Final products were analysed by reverse-phase HPLC (Alltima
1.2 eq.) was added and the reaction was stirred at 20 ꢀC until
completion (TLC). The solvent was removed under reduced pres-
sure and the crude product was purified by column flash-
chromatography on silica to give 1-(4-bromophenylsulfonyl)
piperidine (38a) (1.19 g, 3.9 mmol, quantitative yield). 1H NMR
C18 5
mm column, 150 ꢁ 3.2 mm; Alltech Associated, Inc., Deerfield,
IL) using an Agilent HP1100 equipped with a diode-array detector.
Mobile phases were gradients of 80% CH3CN/20% H2O (v/v) in
45 mM NH4O2CH at pH 3.5 and 0.5 ml/min. Purity was determined
by monitoring at 330 ꢂ 50 nm and was >95%. Final product purity
was also assessed by combustion analysis carried out in the
Campbell Microanalytical Laboratory, University of Otago, Dunedin,
New Zealand. Melting points were determined on an Electro-
thermal 2300 Melting Point Apparatus. NMR spectra were obtained
on a Bruker Avance 400 spectrometer at 400 MHz for 1H and
100 MHz for 13C spectra.
(CDCl3)
d
7.67 (dt, J ¼ 8.72, 2.16 Hz, 2H), 7.61 (dt, J ¼ 8.72, 2.16 Hz,
2H), 2.99 (t, J ¼ 5.48 Hz, 4H), 1.89 (mc, 4H), 1.44 (mc, 2H). This was
used directly.
A mixture of 38a (92 mg, 303
1 eq.), N,N’-dimethylethylenediamine (606
2-one (350 mg, 3.63 mmol, 12 eq.) and K2CO3 (90 mg, 606
m
mol, 1 eq.), CuI (58 mg, 303
mol, 2 eq.), pyrrolidin-
mol,
mmol,
m
m
3 eq.) were dissolved in dry dioxane (5 ml, 0.2 M) and heated at
100 ꢀC for 18e24 h. The mixture was cooled to 20 ꢀC and solvent
was removed under reduced pressure. The crude product was pu-
rified as for procedure A to give 10 (48% yield). mp 180e182 ꢀC. 1H
7.2. General procedures
A: Sulfonyl chloride (1 eq.) was dissolved in a mixture of dry
DCM (0.2 M) and TEA (1.5 eq.). Amine (1.2 eq.) was added and the
NMR (CDCl3)
d
7.81 (td, J ¼ 9.05, 2.28 Hz, 2H), 7.75 (td, J ¼ 9.08,
2.16 Hz, 2H), 3.91 (t, J ¼ 7.00 Hz, 2H), 2.98 (t, J ¼ 5.44 Hz, 4H), 2.66 (t,
J ¼ 7.92 Hz, 2H), 2.21 (mc, 2H), 1.64 (mc, 4H), 1.41 (mc, 2H). HRMS
(ESI): m/z ¼ [M þ H]þ calculated for C15H21N2O3S: 309.1267; found:
309.1276; [M þ Na]þ calculated for C15H20N2O3SNa: 331.1087;
found: 331.1093; HPLC purity: 99.6%. Anal calcd for C15H20N2O3S: C,
58.42; H, 6.54: N, 9.08. Found: C, 58.40; H, 6.54: N, 8.85.
See Supplementary Material for details of the syntheses of
related compounds 11e23 from 36a and substituted piperidines
37bek.
Table 4
IC50 values for compounds 9,12, 14,15,17 against recombinant AKR1C3 and in HCT-
116/AKR1C3 cells.
No
IC50 (nM)
AKR1C3a
HCT-116b
4
9
12
14
15
17
13
52
94
88
84
56
27
24
11
19
15
22
7.4. Compounds of Table 3
7.4.1. 1-(4-(2-Methylpiperidin-1-ylsulfonyl)phenyl)azetidin-2-one
(24)
Coupling of 1-(4-iodophenylsulfonyl)-2-methylpiperidine (36b)
and azetidin-2-one by procedure B gave 24 (yield 92%); mp 149e
a
As for Table 2.
b
Concentration of drug to inhibit the conversion of nitroaromatic prodrug PR-
104A to the hydroxylamine PR-104H in HCT-116 cells engineered to over-
express AKR1C3, as determined by LC/MS/MS; see Experimental section.

D.M. Heinrich et al. / European Journal of Medicinal Chemistry 62 (2013) 738e744
743
151 ꢀC. 1H NMR (CDCl3)
d
7.79 (mc, 4H), 4.23 (mc, 1H), 3.90 (t,
(td, J ¼ 13.36, 3.81 Hz, 1H), 3.01 (dt, J ¼ 12.81, 2.52 Hz, 1H), 2.92 (sbr,
4H), 1.68e1.32 (m, 5H), 1.11 (d, J ¼ 6.93 Hz, 3\H). HRMS (ESI): m/
z ¼ [M þ H]þ calculated for C16H21N2O4S: 337.1217; found:
337.1208; [M þ Na]þ calculated for C16H20N2O4SNa: 359.1036;
found: 359.1029. HPLC purity: 99.1%. Anal calcd for C16H20N2O4S: C,
57.12; H, 5.99; N, 8.33. Found: C, 55.60; H, 6.02; N, 7.88.
J ¼ 7.08 Hz, 2H), 3.70 (td, J ¼ 15.40, 1.60 Hz, 1H), 2.98 (dt, J ¼ 12.80,
2.48 Hz, 1H), 2.65 (t, J ¼ 7.93 Hz, 2H), 2.20 (mc, 2 H), 1.65e1.33 (m,
5H),1.08 (d, J ¼ 6.88 Hz, 3H). HRMS (ESI): m/z ¼ [M þ H]þ calculated
for C15H21N2O3S: 309.1267; found: 309.1268; [M þ Na]þ calculated
for C15H20N2O3SNa: 331.1087; found: 331.1082; HPLC purity: 99.8%.
Anal calcd for C15H20N2O3S$0.25H2O: C, 57.58; H, 6.60; N, 8.95.
Found: C, 57.70; H, 6.27; N, 8.70.
See Supplementary Material for details of the syntheses of
related compounds 32e35 from benzenesulfonamide 38r.
7.4.2. 1-(4-(2-Methylpiperidin-1-ylsulfonyl)phenyl)piperidin-2-one
(25)
7.5. COX assays
Coupling of 36b and piperidinone by procedure C gave 25 (4%
yield); mp 124e126 ꢀC. 1H NMR (CDCl3)
d
7.82 (td, J ¼ 8.77, 2.48 Hz,
COX assays were carried out by GVK Biosciences Ltd., Biology,
28A, IDA Nacharam, Hyderabad 500 076 India: www.gvkbio.com
2H), 742 (td, J ¼ 8.85, 2.44 Hz, 2H), 4.25 (mc, 1H), 3.74e3.62 (m, 3H),
3.00 (dt, J ¼ 12.68, 2.48 Hz, 1H), 2.58 (t, J ¼ 6.52 Hz, 2H), 2.20e1.90
(mc, 4H), 1.70e1.32 (m, 5H), 1.11 (d, J ¼ 6.89 Hz, 3H). HRMS (ESI): m/
z ¼ [M þ H]þ calculated for C17H25N2O3S: 337.1580; found:
337.1576; [M þ Na]þ calculated for C17H24N2O3SNa: 359.1400;
found: 359.1393. HPLC purity: 99.3%.
7.6. Production and purification of recombinant AKR1C protein
Recombinant AKR1C3 protein was purified from a plasmid
vector incorporating the amplified gene in Escherichia coli BL21
(DE3) cells as described elsewhere [4]. AKR1C1, AKR1C2 and
AKR1C4 protein was purified in the same method from expression
vectors supplied as a gift by Dr Chris Bunce, University of
Birmingham.
7.4.3. 1-(3-(2-Methylpiperidin-1-ylsulfonyl)phenyl)pyrrolidin-2-
one (26)
Reaction of 3-bromobenzenesulfonyl chloride (36l) and 2-
methylpiperidine by procedure A gave 1-(3-bromophenylsulfonyl)-
2-methylpiperidine (38l) (93% yield). 1H NMR (CDCl3)
d 7.97 (t,
J ¼ 1.76 Hz,1H), 7.75 (d, J ¼ 7.80 Hz,1H), 7.66 (td, J ¼ 8.01, 0.80 Hz,1H),
7.36 (t, J ¼ 7.92 Hz,1H), 4.24 (mc,1H), 3.72 (td, J ¼ 13.49, 2.72 Hz,1H),
3.01 (dt, J ¼ 12.93, 2.52 Hz,1H),1.67e1.23 (m, 5H),1.09 (d, J ¼ 6.92 Hz,
3H). This was used directly.
7.7. Inhibition of AKR1C enzyme activity
The AKR1C enzyme inhibitory activity of the compounds was
determined by measuring their ability to prevent AKR1C-
dependent reduction of a non-fluorescent ketone probe, Probe 5,
to a fluorescent alcohol in the presence of NADPH as previously
described [4]. The compounds and known AKR1C3 inhibitors (flu-
fenamic acid, meclofenamic acid and mefenamic acid; Sigmae
Aldrich, Auckland, New Zealand) were tested at multiple concen-
Coupling of 38l and 2-pyrrolidone by procedure B gave 26 (61%
yield); mp 116e118 ꢀC. 1H NMR (CDCl3)
d
¼ 7.99e7.95 (m, 1H), 7.94
(dd, J ¼ 2.24, 1.00 Hz, 1H), 7.59 (ddd, J ¼ 7.76, 1.52, 1.08 Hz, 1H), 7.49
(dd, J ¼ 15.89, 7.73 Hz, 1H), 4.25 (mc, 1H), 3.90 (t, J ¼ 7.04 Hz, 2H),
3.73 (td, J ¼ 13.33, 3.80 Hz, 1H), 3.02 (dt, J ¼ 12.65, 2.56 Hz, 1H), 2.63
(t, J ¼ 7.89 Hz, 2H), 2.20 (mc, 2H), 1.70e133 (m, 5H), 1.10 (d,
J ¼ 6.92 Hz, 3H) HRMS (ESI): m/z ¼ [M þ H]þ calculated for
trations in duplicate between 0.1 nM and 100 mM in 2% DMSO to
generate AKR1C enzyme inhibition data. Compound IC50 values
were calculated by fitting the inhibition data to a four-parameter
logistic sigmoidal doseeresponse curve using Prism 5.02 (Graph-
Pad, La Jolla, CA, USA). All IC50 values <10 mM are reported as the
mean of 2 or more separate determinations.
C
C
17H23N2O3S: 323.1424; found: 323.1426; [M þ Na]þ calculated for
17H21N2O3SNa: 345.1243; found: 345.1238; HPLC purity: 100%.
Anal calcd for C16H22N2O3S: C, 59.60; H, 6.88; N, 8.69. Found: C,
59.60; H, 6.76; N, 8.65.
See Supplementary Information for details of the syntheses of
related compounds 27e30 from benzenesulfonyl chlorides 36a,b,
lep.
7.8. Inhibition of cellular AKR1C3 activity
7.4.4. 1-(4-(2-Methylpiperidin-1-ylsulfonyl)phenyl)pyrrolidine-2,5-
dione (31)
Reaction of 4-nitrobenzenesulfonyl chloride and 2-methyl
piperidine by procedure A gave 2-methyl-1-(4-nitrophenylsulfonyl)
The cellular AKR1C3 inhibitory activity of the compounds was
determined in HCT-116 cells engineered to over-express AKR1C3 by
measuring the inhibition of AKR1C3-dependent aerobic metabo-
lism of an exogenous substrate (PR-104A) to its hydroxylamine
metabolite (PR-104H). The quantitation of PR-104H by LC-MS/MS
and transfection of the HCT-116/AKR1C3 cell line have been
described previously [3,13]. Compound IC50 values were calculated
from four-parameter logistic sigmoidal doseeresponse curves that
were fitted to the inhibition data using Prism 5.02. See
Supplementary Material for the structures of PR-10A and PR-104H.
piperidine (38q) (91% yield). 1H NMR (CDCl3)
d
8.34 (td, J ¼ 8.93,
2.36 Hz, 2H), 8.00 (td, J¼ 8.96, 2.32 Hz, 2H), 4.334.23 (m,1H), 3.76 (ddd,
J ¼ 13.12, 5.40, 2.16 Hz, 1H), 3.04 (dt, J ¼ 12.88, 2.40 Hz, 1H), 1.67e1.30
(m, 5H),1.11 (d, J¼ 6.92 Hz, 3H). HRMS (ESI): m/z¼ [M þ H]þ calculated
for C12H17N2O4S: 285.0904; found: 285.0903; [M þ Na]þ calculated for
C12H16N2O4SNa: 307.0723; found: 307.0720, HPLC purity: 99.7%.
Hydrogenation of 38q by procedure
methylpiperidin-1-ylsulfonyl)aniline (38r) (97% yield). 1H NMR
(CDCl3)
E
gave 4-(2-
d
7.59 (d, J ¼ 8.60 Hz, 2H), 6.66 (td, J ¼ 8.69, 2.68 Hz, 2H),
Acknowledgements
4.19 (mc, 1H), 4.05 (sbr, 2H), 3.64 (td, J ¼ 13.21, 3.48 Hz, 1H), 2.95 (dt,
J ¼ 12.61, 2.32 Hz, 1H), 1.70e1.30 (m, 5H), 1.08 (d, J ¼ 6.88 Hz, 3H).
HRMS (ESI): m/z ¼ [M þ H]þ calculated for C12H19N2O2S: 255.1162;
This work was partially funded by the Deutsche For-
schungsgemeinschaft under grant HE6009/1e1. (to D.H.). We thank
Ms Emma Hamilton for help with the enzyme studies, Dr Chris
Guise and Dr Adam Patterson for the HCT-116/AKR1C3 cell line, and
Dr Chris Bunce for the AKR1C1, AKR1C2 and AKR1C4 expression
vectors, and the New Zealand National eScience Infrastructure for
access to computational support.
found: 255.1157; [M
þ
Na]þ calculated for C12H18N2O2SNa:
277.0981; found: 277.0973. HPLC purity: 99.9%.
Coupling of 38r with succinic anhydride by procedure D gave 31
(yield 44%); mp 203e206 ꢀC. 1H NMR (CDCl3)
2.32 Hz, 2H), 7.50 (td, J ¼ 8.85, 2.28 Hz, 2H), 4.31e4.21 (m, 1H), 3.71
d
¼ 7.92 (td, J ¼ 8.80,

744
D.M. Heinrich et al. / European Journal of Medicinal Chemistry 62 (2013) 738e744
[7] A.L. Lovering, J.P. Ride, C.M. Bunce, J.C. Desmond, S.M. Cummings, S.A. White,
Appendix A. Supplementary data
Cancer Res. 64 (2004) 1802e1810.
[8] M.C. Byrns, T.M. Penning, Chem. Biol. Interact. 178 (2009) 221e227.
[9] D.R. Bauman, S.L. Rudnick, L.M. Szewczuk, Y. Jin, S. Gopishetty, T.M. Penning,
Mol. Pharmacol. 67 (2005) 60e68.
[10] V.J. Jackson, Y. Yosaamadja, J.U. Flanagan, C.J. Squire, Acta Crystallogr. F68
(2012) 409e413.
[11] J.U. Flanagan, Y. Yossatmadja, R.M. Teague, M.Z.L. Chai, A.P. Turnbull, C.J.
Squire, PLoS One 7 (8) e43965, doi:10.1371/journal.pone.0043965.
[12] T. Ogihara, I. Tamai, A. Tsuji, A. J. Pharm. Sci. 88 (1999) 1217e1221.
[13] D.J. Yee, V. Balsanek, D. Sames, J. Am. Chem. Soc. 126 (2004) 2282e2283.
[14] C.P. Guise, M.R. Abbattista, R.S. Singleton, S.D. Holford, J. Connolly, G.U. Dachs,
S.B. Fox, R. Pollock, J. Harvey, P. Guilford, F. Donate, W.R. Wilson,
A.V. Patterson, Cancer Res. 70 (2010) 1573e1584.
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.ejmech.2013.01.047.
References
[1] M.C. Byrns, Y. Jin, T.M. Penning, J. Steroid. Biochem. Mol. Biol. 125 (2011) 95e104.
[2] J.C. Desmond, J.C. Mountford, M.T. Drayson, E.A. Walker, M. Hewison, J.P. Ride,
Q.T. Luong, R.E. Hayden, E.F. Vanin, C.M. Bunce, Cancer Res. 63 (2003) 505e512.
[3] T.M. Penning, M.E. Burczynski, L.M. Jez, H.-K. Lin, H. Ma, M. Moore, K. Ratnam,
N. Palackal, Mol. Cell. Endocrinol. 171 (2001) 137e149.
[15] A. Klapars, X. Huang, S.L. Buchwald, A general and efficient copper cat-
alyst for the amidation of aryl halides, J. Am. Chem. Soc. 124 (2004)
7421e7428.
[16] J. Yin, S.L. Buchwald, Org. Lett. 2 (2000) 1101e1104.
[17] J.J. Cui. WO 2004/76412 A2, published 10th Sept 2004.
[18] T.D. Warner, F. Giuliano, L. Vojnovic, A. Bukasa, J.A. Mitchell, J.R. Vane, Proc.
Natl. Acad. Sci. U. S. A. 96 (1999) 7563e7568.
[4] S.M.F. Jamieson, D.G. Brooke, D. Heinrich, G.J. Atwell, S. Silva, E.J. Hamilton,
A.P. Turnbull, L.J.M. Rigoreau, E. Trivier, C. Soudy, S.S. Samlal, P.J. Owen, E. Schroeder,
T. Raynham, J.U. Flanagan, W.A. Denny, J. Med. Chem. 55 (2012) 7746e7758.
[5] A.O. Adeniji, B.M. Twenter, M.C. Byrns, Y. Jin, M. Chen, J.D. Winkler,
T.M. Penning, J. Med. Chem. 55 (2012) 2311e2323.
[6] J.M. Jez, M.J. Bennett, B.P. Schlegel, M. Lewis, T.M. Penning, Biochem. J. 326
(1997) 625e636.