Structure–Activity Relationship for the First-in-Class Clinical Steroid Sulfatase Inhibitor Irosustat (STX64, BN83495)

Structure–activity relationship studies were conducted on Irosustat (STX64, BN83495), the first steroid sulfatase (STS) inhibitor to enter diverse clinical trials for patients with advanced hormone-dependent cancer. The size of its aliphatic ring was expanded; its sulfamate group was N,N-dimethylated, relocated to another position and flanked by an adjacent methoxy group; and series of quinolin-2(1H)-one and quinoline derivatives of Irosustat were explored. The STS inhibitory activities of the synthesised compounds were assessed in a preparation of JEG-3 cells. Stepwise enlargement of the aliphatic ring from 7 to 11 members increases potency, although a further increase in ring size is detrimental. The best STS inhibitors in vitro had IC50 values between 0.015 and 0.025 nm. Other modifications made to Irosustat were found to either abolish or significantly weaken its activity. An azomethine adduct of Irosustat with N,N-dimethylformamide (DMF) was isolated, and crystal structures of Irosustat and this adduct were determined. Docking studies were conducted to explore the potential interactions between compounds and the active site of STS, and suggest a sulfamoyl group transfer to formylglycine 75 during the inactivation mechanism.


Introduction
The inhibition of steroid sulfatase (STS) as a new target for endocrine therapy has attracted considerable attention over the past two decades after recognition that the STS pathway could also be a significant source of oestrogens alongside those originating from aromatase, the enzyme that aromatises androgens to oestrogens. Evidence to support this hypothesis includes: 1) a millionfold higher STS activity than aromatase activity in liver as well as normal and malignant breast tissues, [1] 2) the origin of oestrone (E1) from oestrone sulfate (E1S) in breast cancer tissue is~10-fold greater than that from androstenedione, [2] and 3) STS expression is an important prognostic factor in human breast carcinoma. [3,4] Most oestrogens that originate from the aromatase pathway are converted into and stored in the body as sulfate conjugates that per se are biologically inactive. However, this reservoir of oestrogen sulfates could significantly contribute to overall oestrogenic stimulation of the growth and development of hormone-dependent tumours when STS catalyses the hydrolysis of substrates such as E1S to E1, and dehydroepiandrosterone sulfate (DHEA-S) to DHEA. The formation of DHEA via the STS pathway accounts for the production of 90 % of androstenediol (Adiol). Although structurally an androgen, Adiol possesses oestrogenic properties. It is~100-fold weaker than oestradiol [5][6][7][8] and has a lower affinity for the oestrogen receptor. [9] However, the 100-fold higher concentrations of Adiol in the circulation have led some to speculate that it may have oestrogenic properties equipotent to oestradiol. [10] Thus, STS is an attractive and novel target for rendering potentially more effective oestrogen deprivation through therapeutic intervention in hormone-dependent cancers such as those of the breast, endometrium, and prostate.
Considerable progress has been made since the early 1990s in the development of STS inhibitors. Many structurally (steroidal and nonsteroidal) and mechanistically (principally reversible and irreversible) diverse inhibitors have been developed. However, compounds that contain the pharmacophore for irreversible inhibition of STS, i.e., an aryl sulfamate ester, have consistently shown distinctive and potent in vitro and in vivo inhibitory activities. [11][12][13] One compound, the nonsteroidal inhibitor 1 (Irosustat, STX64, BN83495, Figure 1), is the first STS inhibitor to enter clinical trials for postmenopausal patients with ad-Structure-activity relationship studies were conducted on Irosustat (STX64, BN83495), the first steroid sulfatase (STS) inhibitor to enter diverse clinical trials for patients with advanced hormone-dependent cancer. The size of its aliphatic ring was expanded; its sulfamate group was N,N-dimethylated, relocated to another position and flanked by an adjacent methoxy group; and series of quinolin-2(1H)-one and quinoline derivatives of Irosustat were explored. The STS inhibitory activities of the synthesised compounds were assessed in a preparation of JEG-3 cells. Stepwise enlargement of the aliphatic ring from 7 to 11 members increases potency, although a further increase in ring size is detrimental. The best STS inhibitors in vitro had IC 50 values between 0.015 and 0.025 nm. Other modifications made to Irosustat were found to either abolish or significantly weaken its activity. An azomethine adduct of Irosustat with N,N-dimethylformamide (DMF) was isolated, and crystal structures of Irosustat and this adduct were determined. Docking studies were conducted to explore the potential interactions between compounds and the active site of STS, and suggest a sulfamoyl group transfer to formylglycine 75 during the inactivation mechanism.
vanced hormone-dependent breast cancer and has shown encouraging results. [14,15] Progress has been made since the completion of this first trial. [16] Currently, 1 is undergoing phase I trials for advanced prostate cancer and phase II trials for endometrial and advanced breast cancer.
On the discovery of 1 as a potent STS inhibitor, a basic study was carried out to provide a preliminary structure-activity relationship (SAR). [17] The main focus of that work was on ring contraction (from 7-down to 6-and 5-membered rings: compounds 2 and 3, Figure 1) and expansion (from 7-to 8membered rings: 4, Figure 1) of the aliphatic ring of 1. In addition, a tricyclic oxepin derivative of 3 (compound 5, Figure 1) was synthesised and evaluated. Herein we report a more extensive SAR study for 1, further expansion of the aliphatic ring size, N,N-dimethylation of the sulfamate group, relocation of the sulfamate group to another position, introduction of a substituent adjacent to the sulfamate group, and exploration of a series of quinolin-2(1H)-one and quinoline derivatives of 1. The biological activities of the synthesised compounds were assessed in a preparation of JEG-3 cells. In addition, an azomethine adduct of 1 and N,N-dimethylformamide (DMF) is reported. The crystal structures of 1 and its azomethine adduct were determined. Docking studies were conducted to explore the potential interactions between the compounds and the active site of STS.

Results and Discussion
Chemistry With the exception of ethyl 2-oxocyclotridecanecarboxylate, which is available commercially, the starting cyclic b-keto esters required for the synthesis of tricyclic coumarins 6 b-9 b and 11 b were prepared by treating the corresponding cycloalkyl ketone with diethyl carbonate in the presence of two equivalents of sodium hydride at room temperature. [18] The parent tricyclic coumarins were formed under Pechmann conditions by cyclising resorcinol and the corresponding ethyl 2oxocycloalkylcarboxylates in the presence of an equimolar mixture of trifluoroacetic acid and concentrated sulfuric acid as the condensing agent (Scheme 1). The yields of the tricyclic coumarins ranged from 14 to 33 %, presumably due to severe ring strain experienced by cycloalkenyl rings, in particular cy-clononene and cycloundecene, during the cyclisation of the cyclic b-keto esters with resorcinol.
An earlier method was used for the sulfamoylation of parent hydroxycoumarins (Scheme 1). This involved treating a solution of the phenol in anhydrous N,N-dimethylformamide (DMF) with sodium hydride followed by the addition of a freshly concentrated solution of sulfamoyl chloride in toluene, which was prepared according to the method of Woo et al. [19] The synthesis of 12 was initially attempted by deprotonation of 1 in N,N-dimethylacetamide (DMA) with sodium hydride at 0 8C followed by N,N-dimethylation with methyl iodide (Scheme 2). However, compound 12 obtained by this route was persistently contaminated by a trace amount of 3methoxy-8,9,10,11-tetrahydrocyclohepta[c]chromen-6(7H)-one, which is most likely the product of desulfamoylation of 1 followed by methylation of the phenol released (compound 1 a) under the reaction conditions employed. This ethereal contaminant was particularly difficult to remove, and hence a different synthetic approach was sought. Compound 12 was subsequently prepared with high purity by heating 1 a in N,N-dimethylcyclohexylamine with N,N-dimethylsulfonyl chloride (Scheme 2). Scheme 1. Synthesis of tricyclic coumarin sulfamates (6)(7)(8)(9)(10)(11). Reagents and conditions: a) 2 NaH, N 2 , 15 h, RT; b) concd H 2 SO 4 /CF 3 COOH, 3 h, 0 8C!RT; c) anhydrous DMF, NaH, N 2 , H 2 NSO 2 Cl, 0 8C!RT.
Scheme 2. Synthesis of 12, the N,N-dimethyl derivative of 1. Reagents and conditions: a) NaH, CH 3 I, 0 8C (12 obtained in this manner was contaminated by a trace amount of the 3-methoxy derivative of 1 a); b) N,N-dimethylcyclohexylamine, Me 2 NSO 2 Cl, 90-95 8C, 1 h. Similar to 1, the synthesis of 13 b was achieved by a Pechmann route, although resorcinol was replaced by 4-methoxybenzene-1,3-diol (13 a) as starting material, which was prepared according to the method of Godfrey et al. (Scheme 3). [20] Sulfamoylation of a solution of 13 b in DMA gave the methoxylated tricyclic coumarin sulfamate 13.
The synthesis of 2-hydroxy-8,9,10,11-tetrahydrocyclohepta[c]chromen-6(7H)-one (14 a) was carried out by allowing hydroquinone to react with methyl 2-oxo-1-cycloheptanecarboxylate under Pechmann conditions (Scheme 4). As anticipated, the isolated yield of 14 a was extremely low (3 %) due to the 2-position of hydroquinone not being electronrich and hence activated for ring closure by a Pechmann mechanism. Nonetheless, a sufficient quantity of 14 a was isolated for further sulfamoylation to give the 2-sulfamate 14.
Compound 15 is a low-yielding azomethine adduct of 1 with DMF. Only a very small amount of 15 was isolated during a very large-scale synthesis of 1 that was performed for determination of its crystal structure. With an earlier method for conducting sulfamoylation, which involves the use of sodium hydride in excess for deprotonating the phenolic parent compound 1 a in DMF prior to the addition of sulfamoyl chloride, the formation of 15 is anticipated, as we reported earlier a similar azomethine adduct between 2-nitrophenyl sulfamate and DMF. [21] It is reasoned that the presence of excess sodium hydride in the reaction mixture deprotonates the sulfamate group of 1 after its formation, and the resulting anion under-goes a nucleophilic attack on the formyl group of DMF to give compound 15 upon subsequent dehydration, as illustrated in Scheme 5.
The key intermediate for synthesising the rest of the quinoline and quinolinone derivatives reported herein is compound 17, which was prepared by O-benzyl protection of 16 a (Scheme 6). After deprotonation of 17 with sodium hydride and heating the resulting anion with methyl iodide, the Nmethyl derivative 18 a was obtained in high yield. Debenzylation by hydrogenation gave the phenolic quinolinone 18 b, which was sulfamoylated to give the 5-methyl quinolinone sulfamate 18.
The 3-O-benzyl-protected quinolinone 17 was converted into the 6-chloroquinoline 19 a with phosphorus oxychloride. Holding 19 a at reflux in anhydrous DMF with freshly prepared sodium methoxide gave the 6-methoxyquinoline 19 b. The 6methylquinolinyl sulfamate 19 was obtained by first debenzylating 19 b followed by sulfamoylating the phenolic derivative 19 c.
Quinolinones 20 and 22 and quinolines 21 and 23 were prepared by a different route from their corresponding lower members 18 and 19. Holding the anion of 17 at reflux in DMF with either 1-bromopentane or 1-bromo-3-phenylpropane rendered a mixture of both the N-(20 a and 22 a) and O-alkylated (21 a and 23 a) derivatives. Interestingly, the isolated yields of quinolinones 20 a (62 %) and 22 a (55 %) were both found to be higher than their quinoline counterparts 21 a (41 %) and 23 a (42 %), suggesting that N-alkylation is slightly more favourable under the reaction conditions. In addition, both quinolinones were retained longer by silica in flash chromatography than quinolines, suggesting that 20 a and 22 a are more polar than 21 a and 23 a. Debenzylation by hydrogenation of 20 a-23 a in the usual manner gave the phenolic derivatives 20 b-23 b, which upon sulfamoylation gave the corresponding sulfamates 20-23.

Crystal structures
A crystal of 1 with approximate dimensions of 0.25 0.10 0.08 mm was used for data collection. As shown in Figure 2 b, molecules of 1 interact via a network of intermolecular hydrogen bonds. In particular, one proton of the sulfamate NH 2 group (H1B) interacts with the carbonyl oxygen atom (O5) of the coumarin ring in a proximate molecule, whereas the other NH proton (H1A) interacts with an oxygen atom (O2) of the SO 2 group of a neighbouring sulfamate group. Additionally, there are possible intermolecular p-p interactions present (centroid C9-C10-C15-C16 to centroid C1-C2-C3-C4-C5-C6 distance = 3.52 ). As predicted in previous work by molecular modelling, the 7-membered aliphatic ring of 1 is in the chair form (Figure 2 a,b), which is similar to that of cycloheptene with the C=C moiety taking the place of one of the ring carbon atoms in the cyclohexane chair. [17] A crystal of 15 with approximate dimensions of 0.25 0.13 0.10 mm was used for data collection. As shown in Figure 2 c, the tricyclic coumarin scaffold of 15 has a similar conformation to that observed for 1. The stereochemistry is unambiguously E at the double bond of its (dimethylamino)methylene sulfamoyl group, suggesting that steric effects might be a contributing factor in the more favourable formation of the trans geometric isomer via the route in Scheme 5, with the bulky dimethylamino and arylsulfamoyl motifs placed diametrically opposite before the antiperiplanar elimination of water. As for 1, the aliphatic ring of 15 is clearly in the chair form. Crystal structures of two other tricyclic coumarin sulfamates 6 and 7 with larger ring sizes were also obtained and have been reported elsewhere. [22] Structure-activity relationship and molecular modelling Altogether, ten tricyclic coumarin sulfamates are compared in this work, out of which the syntheses of six final compounds are reported for the first time. These compounds contain a core bicyclic coumarin ring system, but differ in the size of the third (aliphatic) ring. The lowest member of the series studied is 2, because having an aliphatic ring smaller than the 5-membered cyclopentenyl would be synthetically challenging due to the significant ring strain of a cyclobutene or cyclopropene. The increase in size of the third ring was carried out in a stepwise fashion from 5 to 15 members, although the 14-membered derivative was omitted, primarily due to the lack of commercial availability of cyclotetradecanone as starting material. We evaluated the STS inhibitory activities of the tricyclic coumarin sulfamates 1-4 and 6-11 in a placental microsome preparation, and the results were reported in a previous publication. [23] For reference and comparison, these results are listed in Table 1. In this assay, 7 (10-membered third ring) proved to be the most potent STS inhibitor of the series in vitro, with an IC 50 value of 1 nm, although 1 (7-membered third ring), 6 (9membered third ring), and 8 (11-membered third ring) were also potent, with IC 50 values ranging from 8 to 13 nm. The least potent congeners of the series were 2 (5-membered third ring) and 11 (15-membered third ring), the IC 50 values for which were found to be 200 nm or higher. While it is not clear why the IC 50 value for 4 (8-membered third ring) is not of the same order of magnitude as its immediate lower (1) and higher (6) congeners, but is instead significantly higher at 30 nm, it is apparent that the size of the third ring in this series of compounds has a marked effect on the potency of compounds against STS. Interestingly, it was found that 7 is only marginally more potent than 1 in vivo despite its IC 50 value in placental microsomes at 1 nm being eightfold lower than that of 1. [23] Despite its relatively weak activity in vitro (IC 50 = 370 nm, placental microsomes), 11 was found to be the most potent tricyclic coumarin sulfamate in vivo, inhibiting rat liver STS activity by 23 and 94 % when assayed 24 h after administration at respective doses of 0.1 and 1 mg kg À1 , [23] which may be explained, among other things, by a depot effect relating to its high log P value.
We recently replaced the placental microsome preparation with a JEG-3 cell preparation as the standard assay for screening the in vitro STS inhibitory activities of compounds. The advantage of using intact growing JEG-3 cells is that they allow testing of the compounds under conditions that closely resemble the tissue/physiological situation in which the drug must first cross the plasma membrane before it can reach the target (STS) enzyme. These human choriocarcinoma cells have abundant STS enzyme activity, are easy to grow, and are less expensive to use than purified enzyme or placental microsomes. We therefore re-tested the STS inhibitory activities of the tricyclic coumarin sulfamates in JEG-3 cells, and their IC 50 values are listed in Table 1. As expected for a cell-based assay, the IC 50 values against STS obtained for the series of compounds are much lower than those obtained from the cell-free placental microsome assay. However, the overall in vitro inhibitory profile observed is similar, with potency increasing as the size of the third aliphatic ring increases from 5 to 11 members, but then decreasing as the ring size increases further. The most potent compounds observed are 6-8, the IC 50 values of which are between 0.015 and 0.025 nm, whereas 11 is the weakest STS inhibitor in vitro. These results suggest that the ability of compounds to cross the cell membrane and then to interact with the active site of STS is optimal with compounds 6-8, when the aliphatic ring contains 8-10 carbon atoms. Unexpectedly, there is a dramatic decrease in potency observed when the size of the third ring increases from 11 to 12 members. There is a five orders of magnitude difference between the IC 50 values of 8 and 9.
To examine the possible interactions of tricyclic coumarin derivatives with amino acid residues within the active site of STS, these molecules were docked into the crystal structure of STS (PDB ID: 1P49). [24] Importantly, the poses discussed are assumed to be those that form immediately prior to the irreversible inactivation of the enzyme by sulfamoyl transfer. Although it is currently not known what residue is involved, these docking results would be predictive of inactivation of the gem-diol form of the formylglycine residue 75 (FG75) by sulfamoyl transfer. The docking results for 1, 7, and 9 are shown in Figure 3 a and those for 7 and 11 in Figure 3 b. In common with 1 and 7, as shown in Figure 3 a, the rest of the compounds in the series, apart from compound 11, bind with the sulfamate down by the catalytically crucial FG75 residue and the calcium ion. This leaves the third aliphatic ring residing in a predominantly hydrophobic pocket formed by R98, T99, L103, V177, F178, T180, G181, T484, H485, V486, F488, and F553. As the size of the third ring increases from 5 to 11 members (compounds 1-4 and 6-8), it gives a more favourable contact with these residues, with the first and second rings (the coumarin moiety) and the sulfamate occupying nearly identical positions. This may partly explain the increase in potency of these compounds in general as the third aliphatic ring increases in size. As shown in Figure 3 a, and exemplified by compounds 1 and 7, the carbonyl groups of these compounds are within hydrogen bonding distance from the backbone NH group of G100 (~3 ). This additional interaction may be a contributing factor that further assists the binding of these molecules to the enzyme active site. The docking pose of compound 9 (12membered third ring) is different from that of its lower congeners. Presumably due to steric hindrance rendered by the bulk of its third ring, 9 binds with the coumarin ring rotated in the binding site (Figure 3 a). As a result, its carbonyl group is no longer positioned to form a hydrogen bond to G100. The same observations can be made for compound 10 (13-membered third ring), as it shows a docking pose similar to that of compound 9 (not shown). With compound 11, the 15-membered third ring is too large to fit in the binding site in the same orientation as it does for compounds 1-4 and 6-10. In contrast to its congeners, 11 binds upside down in the binding site (Figure 3 b) which is a much poorer binding pose. The GOLD docking scores for compounds 1-4 and 6-10 are all in the range of 52-57 which are not sufficiently different to allow any correlation to be made between their docking poses and IC 50 values. However, 11 has a significantly lower GOLD docking score of 38 which may reflect the much poorer IC 50 observed for this compound.
The N,N-dimethylation of 1 to give compound 12 renders the compound inactive in vitro as an STS inhibitor (Table 1). This supports previous findings that a free sulfamate group is a prerequisite for potent irreversible inhibition of STS in vitro. Hence, N-(piperidino), [25] N,N-(dibenzyl)sulfamate, [25] and N,N-dimethyl derivatives of oestrone 3-O-sulfamate (EMATE) [26] were found to be weak reversible or inactive inhibitors of STS in placental microsomes. Only N-acetylated EMATE, but not the benzoyl derivative, inhibits STS irreversibly, albeit much less potently than EMATE. [25] However, compound 12 was found to behave differently in vivo. When administered orally to nude mice, 12 inhibits liver STS activity potently at doses of 1 and 10 mg kg À1 . [27] Moreover, if 12 is applied topically at 1 and 10 mg kg À1 , it also inhibits skin as well as liver STS effectively. [27] This shows that 12 is able to be absorbed via the percutaneous route and could then inhibit STS in the liver and possibly in other tissues throughout the body. We reason that demethylation of 12 occurs enzymatically in vivo, releasing 1 which is then the agent that inhibits STS.
Keeping a free sulfamate group at the 3-position of 1 but introducing a methoxy group at the 2-position renders the resulting compound 13 a weaker STS inhibitor in JEG-3 cells (IC 50 = 78 nm for 13 versus 1.5 nm for 1, Table 1). A similar pattern was observed with 2-methoxyestrone 3-O-sulfamate (IC 50 = 30 nm), which was found to be a weaker STS inhibitor than EMATE (IC 50 = 4 nm) in a preparation of placental microsomes. [28] Having a bulkier aliphatic substituent positioned next to an aryl sulfamate has also been found to confer weaker inhibition of STS, presumably due to steric hindrance. [28] The relocation of the sulfamate group in 1 from the 3-to the 2-position renders a significant decrease in STS inhibitory activity of the resulting compound 14 (Table 1). It is reasoned that the high inhibitory activity observed for 1 is due to its sulfamate group being in a position conjugated to the a,b-unsaturated lactone moiety of the coumarin ring. As a result, the parent phenol 1 a has a lower pK a value and is hence a better leaving group than unsubstituted phenol. We postulate that this effect would more effectively facilitate the transfer of the sulfamoyl group of 1 to an essential amino acid residue in the Table 1. Inhibition of STS activity in placental microsomes (PM) and JEG-3 cells by tricyclic coumarin sulfamates 1-4 and 6-11, the N,N-dimethyl derivative of 1 (compound 12), the 2-methoxy derivative of 1 (13), 6-oxo-6,7,8,9,10,11-hexahydrocyclohepta[c]chromen-2-yl sulfamate (14), and the azomethine adduct of 1 and DMF (15 STS active site and inactivate the enzyme as a result. Relocation of the sulfamate group from the 3-to the 2-position to give 14 would essentially disrupt this process, as the pK a of the parent phenol 14 a is expected to be close to that of unsubstituted phenol. It is also possible that a sulfamate group placed at the 2-position might not be presented properly and effectively to essential amino acid residue(s) in the enzyme catalytic site responsible for its subsequent activation, resulting in less effective inactivation of the enzyme.
The coumarin moiety has been the core bicyclic template for the development of nonsteroidal STS inhibitors by our research group. Other phenols of bicyclic nonsteroidal moieties such as tetrahydronaphthalene; [26] flavones, isoflavones, flava-nones; [29,30] and chromenone and thiochromenone [31] have also been sulfamoylated and explored by us and other research groups for designing STS inhibitors with varying degrees of success. In this work, we studied the effects of replacing the coumarin ring system of 1 with either a quinolin-2(1H)-one or a quinoline moiety. Their respective N-alkylated and alkoxyl derivatives were also investigated for STS inhibitory activity. As shown in Table 2, all compounds inhibit STS weakly in JEG-3 cells. The best STS inhibitor is the unsubstituted quinolinone derivative 16 (IC 50 = 240 nm or 98 % inhibition at 10 mm), although it is 160-fold less potent than 1 (IC 50 = 1.5 nm, Table 1). This is closely followed by the quinoline derivative 19, which inhibits STS by 68 % at 10 mm, although the inhibition remains weak. These results further confirm that the coumarin ring is essential for the potent STS inhibitory activity observed for 1. This is attributed to several factors. With 16, 18, and 19 docked into the STS active site in a fashion similar to that of 7 ( Figure 4), we postulate that electronic factors such as the pK a values of parent phenols could play a significant role for the results observed. To explore this possible causative factor further, the pK a values of 7-hydroxy-2H-chromen-2-one (25, represents 1 a, the parent phenol of 1), 7-hydroxyquinolin-2(1H)-one (26, represents 16 a, the parent phenol of 16), 7-hydroxy-1,4-dimethylquinolin-2(1H)-one (27, represents 18 b, the parent phenol of 18), and 7-methoxynaphthalen-2-ol (28, represents 19 c, the parent phenol of 19) as calculated by ACD/Labs software version 11.01 were compared ( Figure 5). As shown, the pK a value of 1 a is expected to be between 1 and 2 log units lower than those of 16 a, 18 b, and 19 c. This factor suggests that 1 a is a much better leaving group than 16 a, 18 b, and 19 c, rendering the sulfamate group of 1 a a much stronger sulfamoylating species for the inactivation of the enzyme, and hence 1 is a more potent STS inhibitor than the quinolinone and quinoline derivatives.  N-Methylation of 16 to give 18 (IC 50 = 2400 nm, Table 2) is detrimental to activity, as this substitution produces a 100-fold decrease in the IC 50 value observed for 18 against STS. For both quinolinone and quinoline series, further enlargement of the substituent from a methyl group to either an n-pentyl or a phenethyl group significantly abolishes the STS inhibitory activities of the resulting compounds. It is possible that these substituted molecules no longer bind effectively to the active site of STS due to steric hindrance caused by the bulk of the substituent.
Finally, replacement of the bridging oxygen atom of the sulfamate group in 16 with an NH moiety to give a sulfamido group abolishes the activity of the resulting compound 24 as an STS inhibitor. A similar finding was observed with oestrone 3-sulfamide. [19] We postulate that, unlike the sulfamate group of 16, an enzyme-catalysed breaking of the SÀN bond of the sulfamido group of 24 is unlikely to take place because, among other things, the parent amine 24 a is a very poor leav-ing group. As a result, it is not anticipated that 24 would be able to inactivate STS to any degree by sulfamoylating the active site, but such an approach could provide leads for reversible STS inhibitors.

Conclusions
The nonsteroidal inhibitor Irosustat, STX64 (1) is the first agent to enter clinical trials for postmenopausal patients with advanced hormone-dependent breast cancer, and has shown encouraging results. In this work, we conducted a range of SAR studies on this drug. Expansion of the size of the aliphatic ring of 1 generally provides more potent derivatives against STS in JEG-3 cells, with best activities observed if the ring is between 9 and 11 members. However, further increasing the ring size is unfavourable, as inhibitory activities were observed to drop significantly. Molecular docking studies suggest that the aliphatic ring of 1 and its derivatives sit in a hydrophobic pocket within the enzyme active site with better contacts made with the enclosing amino acid residues as the ring size increases up to 11 members. Larger derivatives 9 and 10, and in particular 11, dock less well into the active site. Positioning of the sulfamate moiety close to the catalytic FG75 may be predictive of sulfamoyl transfer to this residue in the inactivation process. N,N-Dimethylation of the sulfamate group of 1 is detrimental to in vitro activity, as compound 12 is inactive. This supports previous findings which showed that a free sulfamate group (H 2 NSO 2 O À ) is a prerequisite for potent and irreversible STS inhibition. Introducing a methoxy group at the 2-position of 1 significantly decreases the activity of the resulting 13, probably as a result of steric factors. A detrimental effect to activity is also observed with relocation of the sulfamate group of 1 from the 3-to the 2-position of the molecule. We postulate that the decrease in activity of compound 14 is due to its sulfamate group not being in a conjugated position to the a,bunsaturated lactone moiety of the coumarin ring, which affects the ability of 14 to sulfamoylate and inactivate the enzyme. An azomethine adduct between 1 and the solvent DMF used in the sulfamoylation of 1 a was isolated. Its crystal structure shows that the stereochemistry is E at the double bond of its (dimethylamino)methylene sulfamoyl group. Replacing the coumarin ring system of 1 to give a series of quinolin-2(1 H)one and quinoline derivatives produces essentially weak inhibitors of STS. Only the lowest members of the series inhibit STS. This confirms the unique property of the coumarin system in the design of nonsteroidal STS inhibitors that are structurally related to 1.
In summary, most of the modifications made to the clinical drug 1 decrease potency in vitro. Only a moderate enlargement of its aliphatic ring results in derivatives that are more potent STS inhibitors in vitro. However, it remains to be explored whether such compounds would show significant advantages over 1 if put through pre-clinical trial development.

Experimental Section
In vitro sulfatase assay: Biological assays were performed essentially as described previously. [32] The extent of in vitro inhibition of STS activities was assessed by using intact monolayers of JEG-3 human choriocarcinoma cells. STS activity was measured with [6, H]E1S (50 Ci mmol À1 , PerkinElmer Life Sciences) over a 1 h period.
Molecular modelling: All ligands were built and minimised using Schrçdinger software running under Maestro version 9.0. The crystal structure of human placental oestrone/DHEA sulfatase (PDB ID: 1P49) [24] was used for building the gem-diol form of STS. This involved a point mutation of the ALS75 residue in the crystal structure to the gem-diol form of the structure using editing tools within the Schrçdinger software. The resulting structure was then minimised with the backbone atoms fixed to allow the gem-diol and surrounding side chain atoms to adopt low-energy confirmations. GOLD was used to dock the ligands 25 times each into the rigid protein, with the binding site being defined as a 10 sphere around the ALS75 sulfate. The docked poses were scored using the GOLDScore fitness function.
General methods for synthesis: All chemicals were purchased from either Aldrich Chemical Co. (Gillingham, UK) or Alfa Aesar (Heysham, UK). All organic solvents of analytical reagent grade were supplied by Fisher Scientific (Loughborough, UK). Anhydrous N,N-dimethylformamide (DMF), N,N-dimethylacetamide (DMA), and tetrahydrofuran (THF) were purchased from Aldrich. Sulfamoyl chloride was prepared by an adaptation of the method of Appel and Berger [33] and was stored as a solution under N 2 in toluene as described by Woo et al. [19] Thin-layer chromatography (TLC) was performed on pre-coated plates (Merck TLC aluminium sheets silica gel 60 F 254 , Art. No. 5554). Product(s) and starting material were detected by viewing under UV light and/or treating with a methanolic solution of phosphomolybdic acid followed by heating. Flash column chromatography was performed using gradient elution (solvents indicated in the text) on wet-packed silica gel (Sorbsil C 60 ). IR spectra were determined with a PerkinElmer 782 infrared spectrophotometer, and peak positions are expressed in cm À1 . 1 H and 13 C NMR spectra were recorded with either a Jeol Delta 270 MHz or a Varian Mercury VX 400 MHz spectrometer. Chemical shifts (d) are reported in parts per million (ppm) relative to tetramethylsilane (TMS) as an internal standard. Coupling constants (J) are recorded to the nearest 0.1 Hz. Mass spectra were recorded at the Mass Spectrometry Service Centre, University of Bath. FAB mass spectra were measured using m-nitrobenzyl alcohol as the matrix. Elemental analyses were performed by the Microanalysis Service, University of Bath. Melting points were determined using a Reichert-Jung Thermo Galen Kofler block and are uncorrected. HPLC was undertaken using a Waters 717 instrument equipped with an autosampler and PDA detector. The column used, conditions of elution, and purity of sample are as indicated for each compound analysed.