Aromatase and Dual Aromatase-Steroid Sulfatase Inhibitors from the Letrozole and Vorozole Templates

Concurrent inhibition of aromatase and steroid sulfatase (STS) may provide a more effective treatment for hormone-dependent breast cancer than monotherapy against individual enzymes, and several dual aromatase–sulfatase inhibitors (DASIs) have been reported. Three aromatase inhibitors with sub-nanomolar potency, better than the benchmark agent letrozole, were designed. To further explore the DASI concept, a new series of letrozole-derived sulfamates and a vorozole-based sulfamate were designed and biologically evaluated in JEG-3 cells to reveal structure–activity relationships. Amongst achiral and racemic compounds, 2-bromo-4-(2-(4-cyanophenyl)-2-(1H-1,2,4-triazol-1-yl)ethyl)phenyl sulfamate is the most potent DASI (aromatase: IC50=0.87 nm; STS: IC50=593 nm). The enantiomers of the phenolic precursor to this compound were separated by chiral HPLC and their absolute configuration determined by X-ray crystallography. Following conversion to their corresponding sulfamates, the S-(+)-enantiomer was found to inhibit aromatase and sulfatase most potently (aromatase: IC50=0.52 nm; STS: IC50=280 nm). The docking of each enantiomer and other ligands into the aromatase and sulfatase active sites was also investigated.


Introduction
The growth and development of the most common form of breast malignancies, hormone-dependent breast cancer (HDBC), is promoted by the presence of oestrogenic steroids. Currently, the most widely used therapies for the treatment of this disease focus on blocking the action of these steroids, either by the use of selective, oestrogen receptor modulators, such as tamoxifen, or by inhibiting their biosynthesis through inhibition of the aromatase enzyme complex. Third-generation aromatase inhibitors (AIs), currently finding widespread appli-cation in the clinic, comprise the nonsteroidal compounds anastrozole and letrozole, and the steroidal exemestane. [1][2][3] Although these compounds were initially used in patients for whom tamoxifen therapy had failed, data from a number of clinical trials suggest that AIs provide a more effective first-line therapy against HDBC as a result of their superior efficacy and toxicology profile. [4][5][6] Of these AIs, there is evidence to suggest that letrozole is superior to anastrozole in suppressing oestrogen levels in breast tissue and plasma in patients with postmenopausal breast cancer. [7] It is unclear how this difference will translate to the clinic, however, the Femara versus Anastrozole Clinical Evaluation (FACE) trial should help to determine whether any differences in efficacy exist between these two AIs. [8] A promising new therapy for the treatment of HDBC has arisen from the development of inhibitors of steroid sulfatase (STS). [9] This enzyme is believed to be virtually ubiquitous Concurrent inhibition of aromatase and steroid sulfatase (STS) may provide a more effective treatment for hormone-dependent breast cancer than monotherapy against individual enzymes, and several dual aromatase-sulfatase inhibitors (DASIs) have been reported. Three aromatase inhibitors with subnanomolar potency, better than the benchmark agent letrozole, were designed. To further explore the DASI concept, a new series of letrozole-derived sulfamates and a vorozolebased sulfamate were designed and biologically evaluated in JEG-3 cells to reveal structure-activity relationships. Amongst achiral and racemic compounds, 2-bromo-4-(2-(4-cyanophen-yl)-2-(1H-1,2,4-triazol-1-yl)ethyl)phenyl sulfamate is the most potent DASI (aromatase: IC 50 = 0.87 nm; STS: IC 50 = 593 nm). The enantiomers of the phenolic precursor to this compound were separated by chiral HPLC and their absolute configuration determined by X-ray crystallography. Following conversion to their corresponding sulfamates, the S-(+)-enantiomer was found to inhibit aromatase and sulfatase most potently (aromatase: IC 50 = 0.52 nm; STS: IC 50 = 280 nm). The docking of each enantiomer and other ligands into the aromatase and sulfatase active sites was also investigated. throughout the body and is responsible for the conversion of alkyl and aryl steroid sulfates to their unconjugated and biologically active forms. Primarily, STS catalyses the conversion of oestrone sulfate, a biologically inactive steroid found at high levels in the plasma of postmenopausal women, to oestrone. In breast cancer tissue, it has been shown that ten times more oestrone originates from oestrone sulfate than from androstenedione. [10] In addition, STS controls the formation of dehydroepiandrosterone (DHEA) from DHEA-sulfate (DHEA-S). DHEA can be subsequently converted to androst-5-ene-3b,17b-diol, an androgen with oestrogenic properties capable of stimulating the growth of breast cancer cells in vitro [11] and inducing mammary tumours in vivo. [12] The pharmacophore for irreversible STS inhibition has been identified as a substituted phenol sulfamate ester, and a number of steroidal (e.g., oestrone-3-Osulfamate, also known as EMATE) and nonsteroidal inhibitors (e.g., Irosustat, also known as STX64, BN83495) have been developed. [9,[13][14] Irosustat, discovered by our group, has been evaluated in a phase 1 clinical trial for the treatment of postmenopausal patients with metastatic breast cancer and has shown promising results. [15] The advantages of a single chemical agent with the ability to interact with multiple biological targets have been recently highlighted. [16][17][18] A possible application of this concept for the treatment of HDBC would be the combination of the pharmacophores for both aromatase and STS inhibition into a single molecular entity. One approach to achieve this would be insertion of the pharmacophore for STS inhibition into an established AI, whilst maintaining the features necessary for aromatase inhibition. We previously reported three series of dual aromatase-sulfatase inhibitors (DASIs) based on different AIs: examples include, compounds 1 and 2 based on letrozole, [19][20] compound 3 based on YM511 (4), [21][22][23][24] and compound 5 based on anastrozole. [25] In a complementary approach, we also reported a series of DASIs obtained following introduction of the pharmacophore for aromatase inhibition into a biphenyl template primarily designed for STS inhibition (e.g, 6). [26] In preliminary work on the design of a prototype letrozolebased DASI, it was hoped that dual aromatase and sulfatase inhibition could be achieved by replacing both para-cyano groups present in letrozole with sulfamate groups, whilst retaining both the triazole and the diphenylmethane moieties necessary for potent aromatase inhibition. [19] A lead compound, bis-sulfamate 1 exhibited IC 50 values of 3044 nm for aromatase and > 10 mm for STS when evaluated in JEG-3 cells. Further iterations improved inhibition of both aromatase and sulfatase, [20] and the most potent AI identified was (AE )-2 in which only one of the para-cyano groups is replaced with a sulfamate group. Compound 2 inhibited aromatase and STS with IC 50 values of 3 nm and 2600 nm, respectively. The enantiomers of 41, the phenolic precursor of 2, were separated by chiral HPLC and converted into their corresponding sulfamates; [27] the R-configuration provided the most potent aromatase inhibitor (R: 3.2 nm; S: 14.3 nm), whilst the S-configuration proved to be the best STS inhibitor (S: 553 nm; R: 4633 nm). [20] Here, we report the further investigation of the structure-activity relationships (SAR) of letrozole-derived DASIs by evaluating the effect on inhibitory activity of increasing linker length between the triazole and the STS pharmacophore, and replacing the para-cyano-substituted ring with a para-chloro-substituted ring. The enantiomers of one compound were separated and their absolute configuration was determined by X-ray crystallography. We also report the synthesis and in vitro inhibitory activities of the first dual inhibitor derived from the third generation AI, vorozole. according to the conditions described by Okada et al. [31] For the synthesis of sulfamate 14, it was envisaged that alkylation of n-butyllithium-deprotonated [32] 4-((1H-1,2,4-triazol-1-yl)methyl)benzonitrile [19] with (4-(chloromethyl)phenoxy)triisopropylsilane would provide a route to 13; this reaction failed to provide the desired product. However, when the alkylating agent was switched to the more reactive 12, the desired product could be obtained. Deprotection of the phenol was achieved using tetra-n-butylammonium fluoride and the product could be used without further purification. Phenol 13 was subsequently converted to sulfamate 14 using the conditions described above. Starting from aldehyde 7, reduction with sodium borohydride and conversion of the resulting benzyl alcohol to the chloride gave compound 16. This is a more reactive alkylating agent than its nonbrominated counterpart, and it was successfully used as the alkylating agent for the synthesis of sulfamate 18 according to the route described above.
The synthesis of sulfamates 22, 29, 35 and 40 is detailed in Scheme 2. Compound 22 was obtained in three steps from 19, which was prepared according to Avery et al. [33] Reaction of 19 with 4-((1H-1,2,4-triazol-1-yl)methyl)benzonitrile as described above was followed by deprotection and sulfamoylation to furnish 22. Sulfamate 29 was prepared in a similar manner using bromide 26, which was itself prepared from methyl 2-(3bromo-4-hydroxyphenyl)acetate 23. [23] Following triisopropylsilyl (TIPS) protection of the phenolic hydroxy group in 23, ester 24 was reduced with lithium borohydride and the resulting benzyl alcohol was converted to benzyl bromide 26, and the synthesis of 29 was completed using the steps described above. The bottom part of Scheme 2 describes the route for the synthesis of sulfamates 35 and 40. Sulfamate 35 was synthesised from 30, which was prepared as described by Avery et al. [33] From alcohol 30, Dess-Martin oxidation gave aldehyde 31, which was reacted with 4-chlorophenylmagnesium bromide to give 32. The alcohol was converted to chloride 33 with thionyl chloride, and this was reacted with 1,2,4-triazole in acetone with concomitant loss of the TIPS protecting group to give 34. Finally, sulfamoylation as described above furnished 35. Sulfamate 40 was prepared analogously from 23 following TIPS protection of the phenol and lithium borohydride reduction of ester 24. N,N-Dimethysulfamate 42 was successfully prepared by heating a mixture of 1-[(4-cyanophenyl)(3-bromo-4-hydroxyphenyl)methyl]-1H- [1,2,4]triazole [20] and N,N-dimethylsulfamoyl chloride in N,N-diisopropylethylamine (DIPEA) (Scheme 3).
Vorozole-derived sulfamate 51 was prepared from benzoic acid 43, which was synthesised as described by Dener et al. [34] from 3-methoxy-4-nitrobenzoic acid (Scheme 4). Formation of the benzotriazole ring was achieved by treatment of 43 with a mixture of sodium nitrite and hydrochloric acid in water. Subsequent formation of methyl ester 45 and lithium borohydride reduction gave compound 46. This approach to the synthesis of the benzotriazole ring ensures that the methyl group is placed on the correct nitrogen atom in the triazole ring. A more concise route to 45 was explored via the alkylation of methyl-1H-benzotriazole-5-carboxylate with methyl iodide in the presence of potassium carbonate but this gave a mixture of three regioisomers from which it was difficult to separate the individual benzotriazol-1-yl isomers. The oxidation of alcohol 46 with potassium permanganate in dichloromethane [35] gave aldehyde 47 in moderate yields. However, excellent yields of the aldehyde could be obtained by oxidation with the trichloroisocyanuric acid/catalytic 2,2,6,6-tetramethylpiperidinooxy (TEMPO) system reported by Giacomelli et al. [36] Aldehyde 47 was subsequently reacted with the Grignard reagent generated from 4-benzyloxybromobenzene to give alcohol 48. This was converted to the corresponding chloride with thionyl chloride and quickly reacted with 1,2,4-triazole in the presence of potassium carbonate to give 49. Deprotection of the phenol was achieved by catalytic hydrogenation with palladium on carbon to give 50, and the formation of the corresponding sulfamate was achieved using the conditions described above to furnish 51.
Inhibition of aromatase and steroid sulfatase activity by sulfamoylated compounds The in vitro inhibition of aromatase and STS activity by each sulfamate was measured in a preparation of an intact monolayer of JEG-3 cells. The results are reported as either IC 50 values or as a percentage of inhibition at 10 mm, and are compared to the reference AI letrozole [19] and the reference STS inhibitor STX64 [21] ( Table 1). The biological activities of both 2 and 52 have been reported previously. [20] All of the sulfamates tested in this series are potent inhibitors of aromatase with IC 50 values 39 nm and, in addition, some compounds also exhibit moderate-to-potent STS inhibition. In this assay, two compounds, 22 and 29 (IC Arom 50 = 0.22 and 0.12 nm, respectively) are more potent than the reference AI letrozole (IC Arom 50 = 0.89 nm). Several compounds in this study contain a para-chloro-substituted phenyl ring rather than the para-cyano-substituted ring found in letrozole and letrozole-based DASI 2. This substitution is present in compounds capable of potent aromatase  [a] Data taken from Wood et al. [19] [b] Data taken from Woo et al. [21] [c] Data taken from Wood et al. [20] [d] Percent inhibition at 10 mm. Mean IC 50 values AE SD were determined from incubations carried out in triplicate in a minimum of two separate experiments. inhibition. For instance, this moiety is present in the potent AI (AE )-vorozole (IC Arom 50 = 2.59 nm), [37] and furthermore, a derivative of letrozole with both para-cyano groups replaced by parachloro groups has been reported to inhibit aromatase activity with an EC 50 value of 8.8 nm in a rat ovarian microsome assay. [38] For this series, comparison of the activities of the para-cyano-substituted compounds with their para-chloro-substituted counterparts reveals that, for the two sets of compounds 2/11 and 18/40, dual inhibitory activity is retained following the switch in substitution (e.g., 2: IC Arom The importance of the positioning of a hydrogen-bond acceptor (e.g., CN, NO 2 ) in the molecule relative to the triazole/imidazole ring for potent aromatase inhibition has been extensively discussed in the literature. [39][40] Interestingly, in this series, replacement of the cyano group with the weaker hydrogen-bond accepting chloro substituent maintains good aromatase inhibitory activity, possibly due to a complex interaction between the hydrogen-bond donor in the active site, the halide and the p system of the connecting aromatic ring. [41] The effect played by the linker between the aromatase and STS pharmacophores on dual inhibitory activities is illustrated by three series of compounds: 1) 52, 14 and 22; 2) 2, 18 and 29; 3) 11 and 40. In each series, lengthening the linker (n) results in an increase in aromatase inhibition; this is illustrated by comparing compounds 2 (n = 0; IC Arom = 0.12 nm). This correlates with the small increase in aromatase inhibition observed in the aw-diarylalkyltriazole series of compounds with inhibitory activities for 4-(3-(4-fluorophenyl)-1-(1H-1,2,4-triazol-1-yl)propyl)benzonitrile and linker extended 4-(4-(4-fluorophenyl)-1-(1H-1,2,4-triazol-1-yl)butyl)benzonitrile reported as 0.19 mm and 0.12 mm, respectively. [32] These finding are also in agreement with those in a YM511-derived DASI series, with activities for 4-(((4-cyanophenyl)(4H-1,2,4-triazol-4-yl)amino)methyl)phenyl sulfamate and 4-(2-((4-cyanophenyl)(4H-1,2,4-triazol-4-yl)amino)ethyl)phenyl sulfamate being 100 nm and 2.1 nm, respectively. [23] A small increase in linker length is beneficial for enhanced STS inhibition (2: IC STS 50 = 2600 nm, vs 18: IC STS 50 = 593 nm), but further extension of the linker has a detrimental effect on inhibitory activity (29: IC STS 50 > 10 000 nm). Extending the linker length was shown to be detrimental to STS activity in the YM511-derived DASI series, with a decrease in activity from 227 nm for 4-(((4cyanophenyl)(4H-1,2,4-triazol-4-yl)amino)methyl)phenyl sulfamate to > 10 000 nm for 4-(2-((4-cyanophenyl)(4H-1,2,4-triazol-4-yl)amino)ethyl)phenyl sulfamate, which has one extra methylene unit in the linker. This decrease in STS inhibition could be due to an increase in flexibility in the molecule as the linker is extended, resulting in less favourable binding of the compound in the active site.
As in our previous investigation into letrozole-and YM511derived DASIs, derivatives containing a halogen positioned ortho to the sulfamate are better AIs than their nonhalogenat-ed counterparts. This trend is exhibited in both the para-

nm)
. This and previous results suggest that the higher aromatase inhibition can be attributed to the increased lipophilicity conferred by the halogen. [20] Similarly, we previously discovered that the presence of a halogen group ortho to the sulfamate increases STS inhibitory activity, and this trend holds true in this series for both pairs of compounds 14/18 (14: IC STS 50 = 3517 nm vs 18: IC STS 50 = 593 nm) and 36/40. The increase in STS inhibitory activity is reasoned to be caused by a lowering of the pK a of the phenol, enhancing its leaving group ability. Presumably, the deleterious effect on activity caused by an increase in linker length for compounds 22 and 29 is too large for any halogen-induced increase in inhibition to be observed.
The N,N-dimethylsulfamate-containing compound 42 is a weaker AI than its corresponding demethylated counterpart 2, despite the increase in lipophilicity conferred by dimethylation, which normally benefits aromatase inhibition. The weak STS inhibition exhibited by 42 in vitro is anticipated based on our previous work on N,N-dimethylated sulfamates. [13,42] For example, despite the poor inhibitory activity against STS [43] exhibited by the N,N-dimethylated derivative of STX64 in vitro, this compound has been shown to almost completely inhibit mouse liver and skin STS activities 24 h after oral administration. [44] This suggests that N,N-demethylation may occur in vivo to provide a compound capable of STS inhibition, and it might also be the case that, although 42 is inactive in vitro, it may act as a prodrug of 2 in vivo. Further work is required to explore the potential in vivo conversion of compound 42 to 2.
Compound 51 is the first reported example of a sulfamatecontaining vorozole derivative. Vorozole is a third generation, aromatase-selective AI, that entered phase 3 clinical trials, but further development was discontinued when no improvement in median survival was obtained compared to megestrol acetate. [45] Nonetheless, we explored the feasibility of designing a DASI that is structurally related to vorozole. Incorporation of the STS inhibitory pharmacophore into the molecule was achieved by replacement of the para-chloro substituent attached to the phenyl ring in vorozole with a sulfamate group. Compound 51 exhibits good inhibitory activity against aromatase, although it is a weaker AI than the corresponding letrozole derivative 52 and, like 52, also exhibits poor STS inhibition. However, based on previous observations, the introduction of appropriate substitutions onto the sulfamate-bearing ring in 51 would be expected to improve inhibitory activity against both enzymes, indicating the feasibility of a DASI based on the vorozole template.

Inhibition of aromatase activity by parent phenols
The aromatase inhibition for the phenols described in this paper is tabulated in Table 2. The loss of the sulfamate group following irreversible inactivation of STS by a sulfamate-based DASI will result in the formation of the corresponding phenol. The quantity of phenol produced by this mechanism is limited in principle once all the STS activity has been inactivated. [46] However, degradation of the sulfamate following prolonged circulation in plasma might provide an additional route for the formation of the phenol. As these phenols still contain the pharmacophore for aromatase inhibition they have the potential to act as AIs in their own right.
The most potent phenol in this series against aromatase is 28 (IC Arom 50 = 0.02 nm) and three compounds, 17, 21 and 28 (IC Arom 50 = 0.21, 0.16, 0.02 nm, respectively) are more potent than the reference AI letrozole (IC Arom 50 = 0.89 nm) in this assay. With the exception of 34, the phenols are either equipotent or slightly better inhibitors of aromatase compared to their corresponding sulfamates.
In common with the trends observed for their sulfamoylated counterparts, positioning a halogen ortho to the phenol results in an increase in aromatase inhibitory activity, as seen for example in compounds 13 and 17 (IC Arom 50 = 2.9 nm vs 0.21 nm, respectively), and lengthening the linker is also beneficial for aromatase inhibition, as seen for example in compounds 13 and 21 (IC Arom 50 = 2.9 nm vs 0.16 nm, respectively).

Chiral HPLC and absolute structure determination
In order to enrich the SAR for letrozole-derived DASIs with their target proteins and to allow comparison with the inhibitory activities of the enantiomers of 2, the activities of each enantiomer of 18, one of the most promising DASIs in this current series, were determined. To avoid any complications arising from decomposition of the sulfamate during separation, resolution by chiral HPLC was performed with 17, the parent phenol of the sulfamate, an approach previously used in the preparation of the enantiomers of 2. [20] The literature contains a number of reports on the resolution of AIs by chiral HPLC with a particular focus on imidazolecontaining compounds: for example, fadrozole hydrochloride, which was separated with a Chiralcel OD column. [47] Using con-ditions similar to those we reported previously for the separation of phenol 43, the enantiomers of phenol 17 were separated on a Chiralpak AD-H analytical column with methanol as the mobile phase (see Experimental Section for further details). The first enantiomer eluted from the column with a retention time of 3.80 min (17 a), whereas the second enantiomer eluted with a retention time of 8.2 min (17 b) giving greater peak separation than that previously obtained for 43. This separation was subsequently scaled-up and successfully performed on a Chiralpak AD-H semi-prep column to separate 700 mg of the racemate with injections of 1.5-2.0 mL of a 20 mg mL À1 methanol solution of 17. Conversion of 17 a and 17 b into their corresponding sulfamates was achieved with excess sulfamoyl chloride in DMA. We previously reported that the sulfamoylation step proceeds without loss of enantiomeric purity in the preparation of the enantiomers of 2, 2 a and 2 b. [20] The optical rotation for each enantiomer of the phenol and corresponding sulfamate was measured (data given in the Experimental Section).
Previously, in the absence of suitable crystals of 2 a,b and 41 a,b for X-ray analysis, the absolute configuration of each enantiomer had to be established using vibrational and electronic circular dichroism in conjunction with time-dependent density functional theory calculations of their predicted properties. Fortuitously, crystals suitable for X-ray analysis could be obtained from ethyl acetate solutions of both 17 a and 17 b, and the absolute configuration of each enantiomer was determined from the X-ray crystal structure of 17 a. [48] The crystal structure obtained for 17 a is shown in Figure 1, allowing the unambiguous elucidation of the absolute configuration of 17 a as R-(À).

Inhibitory activities of chiral sulfamates and their parent phenols
The difference in aromatase and STS inhibition exhibited by each enantiomer of 18 was evaluated following separation of the enantiomers of phenolic precursor 17 by chiral HPLC and conversion to their corresponding sulfamates. For comparison, the aromatase and STS inhibitory activities of each enantiomer of 18 and the aromatase inhibitory activities of the enantiomers of 17 are shown in Table 3 along with those previously obtained for the enantiomers of 2 and 41. Previous studies have suggested that there is often a large difference in aromatase inhibition observed between the enantiomers of chiral AIs. For vorozole, [37] there is a 32-fold difference in activity, with the S-configuration being the most active (S-(+): IC 50  There is a larger 210-fold difference in aromatase inhibitory activity between the enantiomers of fadrozole hydrochloride [47] with the S-enantiomer  50 = 280 nm). This increase in the difference of aromatase inhibition exhibited by each enantiomer could be a result of the increase in asymmetry of the molecule following extension of the linker length. Comparison of the aromatase inhibitory activity between the enantiomers of 2 and 18 reveals that in each case the most potent AI is the dextrorotatory enantiomer and that despite possessing different absolute configuration, the same three-dimensional relationship between the triazole ring and the para-cyanophenyl ring is present in the two most potent enantiomers, R-(+)-2 b and S-(+)-18 b. The spatial disposition of the heterocyle and para-cyanosubstituted ring in 2 b and 18 b resembles that present in the most potent enantiomers of the chromenone-based AI series, [49] suggesting that this is the most favourable orientation of these groups for potent aromatase inhibition. For STS inhibition, there is a switch in the most potent enantiomer from the levorotatory for 2 a to the dextrorotatory for 18 b, and the reason for this is currently unclear.
For the parent phenols, both 17 a and 17 b are more potent AIs than their corresponding sulfamates, which is in accordance with the trend previously described. There is a 36-fold difference in aromatase inhibition for the two enantiomers, with the best AI being S-(+)-17 b. Significantly, the best aromatase inhibitory activity is obtained with the S-(+)-enantiomer of both the phenol and sulfamate providing further confirmation that this is the optimal three-dimensional relationship between the triazole ring and the para-cyanophenyl ring for potent inhibition.

Molecular modelling
In order to examine the possible interaction of 18 a and 18 b with amino acid residues within the active site, these molecules were docked into the human aromatase crystal structure (PDB: 3EQM) [50] along with the natural substrate of the enzyme, androstenedione. For the first time, we also report the result of the docking of letrozole into  [a] Data taken from Wood et al. [19] [b] Data taken from Woo et al. [21] [c] Data taken from Wood et al. [21] Mean IC 50 values AE SD were determined from incubations carried out in triplicate in a minimum of two separate experiments. the active site of aromatase. We previously verified [24] the suitability of the crystal structure for use in docking studies by removing the cocrystallised androstenedione from the substrate binding site and using the docking program GOLD [51] to dock the steroid back in. The results of this experiment indicated that the best pose of the docked androstenedione overlays the crystal structure very well.
The results of the docking of letrozole and androstenedione into the aromatase active site are shown in Figure 2. The 17keto oxygen atom in androstenedione is able to form a hydrogen bond with the backbone amide of Met 374 (2.75 ), and this interaction is mimicked by one of the benzonitrile groups present in letrozole (bond distance = 3.11 ). Additionally, for the AI YM511, we recently reported [24] that the predicted docking conformation of this compound places the benzonitrile group in a similar position in the active site, and this is able to form a hydrogen-bond interaction with Met 374. Hydrogenbond interactions with either a benzonitrile moiety or another group placed at a suitable position relative to the triazole/imidazole ring are known to be important for potent aromatase inhibitory activity in nonsteroidal AIs. [39][40] The other benzonitrile group present in letrozole is able to interact with Ser 478 (2.31 ).
The docking of 18 a and 18 b into the aromatase active site is shown in Figure 2. Both 18 a and 18 b overlay androstenedione with their sulfamates positioned close to the 17-keto group in androstenedione. In a manner similar to letrozole, the benzonitrile groups of both 18 a and 18 b are able to form hydrogen-bond interactions with Ser 478 (3.54 and 2.27 , respectively), and for 18 b, there is an additional interaction with His 480. There is no obvious structural explanation for the difference in aromatase inhibitory activity observed for the two enantiomers. To date, no human aromatase crystal structure complexed with a nonsteroidal AI has been reported. It might be more relevant and informative to dock 18 a and 18 b into such a crystal structure when it becomes available in the future.
The docking of STX64, 18 a and 18 b (two poses) into the crystal structure of STS (PDB: 1P49 [52] ) is shown in Figure 3.

Conclusions
A range of DASIs structurally similar to the potent clinical AI letrozole and one compound similar to vorozole were synthesised and evaluated for aromatase and sulfatase inhibitory activity in JEG-3 cells. In order to realise molecules capable of dual inhibition, the known pharmacophore for STS inhibition (a phenol sulfamate ester) and the pharmacophore for aromatase inhibition (an N-containing heterocyclic ring) were incorporated into a single molecule.
For racemic compounds, the most potent AI identified is 29 (IC Arom 50 = 0.12 nm), while the most potent inhibitor of STS is 40 (IC STS 50 = 180 nm). Consideration of the developing SAR for these derivatives reveals that extending the linker between the aromatase and STS pharmacophores is beneficial for aromatase inhibition, but this is balanced by the detrimental effects on STS inhibition resulting from extension of the linker beyond two carbon atoms. As anticipated, the addition of a halogen in the ortho position to the sulfamate group results in an increase in both aromatase and STS inhibitory activity. Compounds capa- ble of potent aromatase and STS inhibition can be obtained following exchanging the para-cyano group for a para-chloro substituent, suggesting that this group is able to replicate interactions within the enzyme active site. Compound 51, the first sulfamate-containing vorozole derivative, exhibits good inhibitory activity against aromatase meriting further investigation. Even with an IC 50 value in the micromolar range for STS inhibition, based on established precedent, it is likely that this compound will be effective in vivo on both enzymes.
The enantiomers of 17, the phenolic precursor of 18, one of the most potent dual inhibitors in the current study, were separated by chiral HPLC and the absolute configuration of one enantiomer was unambiguously established using X-ray crystallography. Following conversion to the corresponding sulfamate and biological evaluation, it was established that there is a 60-fold difference in aromatase inhibition, with S-(+)-18 b being the most potent. This enantiomer has the same spatial disposition of the heterocycle and para-cyano-substituted ring as that found in R-(+)-2 b, the most potent enantiomer discovered in our previous study. For STS inhibition, there is a fourfold difference in inhibition with the most potent enantiomer also being S-(+)-18 b.
Molecular modelling studies indicate that both YM511 and letrozole dock into the human aromatase crystal structure with a benzonitrile group occupying a similar area of space to the 17-keto oxygen atom of androstenedione. For 18 a and 18 b, the sulfamate group is predicted to occupy this same area of space with their benzonitrile group able to form hydrogen bonds with Ser 478. The molecular modelling study suggests no obvious structural explanation for the difference in aromatase inhibitory activity of 18 a and 18 b. For STS, both 18 a and 18 b dock in a similar orientation to that of STX64.
These results further demonstrate the feasibility of designing a DASI based on the letrozole or vorozole templates and provide a basis for continuing pre-clinical development of such compounds for the treatment of HDBC using a multitargeted strategy.

Experimental Section
In vitro aromatase and sulfatase assays: Biological assays were performed essentially as described previously. [22] The extent of in vitro inhibition of aromatase and sulfatase activities was assessed using intact monolayers of JEG-3 human choriocarcinoma cells, which were chosen because these cells constitutively express both enzymes maximally. Aromatase activity was measured using [1b-  Molecular modelling: Models of androstenedione, letrozole, STX64, 18 a and 18 b were built and minimised using the Schroedinger software running under Maestro version 9.0. The GOLD docking program (version 5.0) [51] was used to dock the models into the aromatase crystal structure (PDB: 3EQM). [50] The binding site was defined as a 10 sphere around the androstenedione that is present in the crystal structure. A distance constraint of 2.30 was applied between the ligating triazole nitrogen atom of the ligand to the haeme iron atom. The ligands were then docked to the rigid enzyme a total of 25 times each and scored using the GOLD-Score fitness function. To remove strain from the docked poses, the systems were put through an energy minimisation procedure using the Impact module of the Schroedinger software.
The crystal structure of human placental oestrone/DHEA sulfatase (PDB: 1P49 [52] ) was used for building the gem-diol form of steroid sulfatase (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 Schroedinger 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. The docked poses were scored using the GOLDScore fitness function.
Crystallographic data: CCDC 806541 (17 a) contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via http://www.ccdc.cam.ac.uk.
General methods for synthesis: All chemicals were purchased from either Aldrich Chemical Co. (Gillingham, UK) or Alfa Aesar (Heysham, UK). All organic solvents of AR 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 [53] and was stored under N 2 as a solution in toluene as described by Woo et al. [54] Thin layer chromatography (TLC) was performed on pre-coated aluminium plates (Merck, silica gel 60 F 254 ). Product spots were visualised either by UV irradiation at 254 nm or by staining with either alkaline KMnO 4 solution or 5 % w/v dodecamolybdophosphoric acid in EtOH, followed by heating. Flash column chromatography was performed using gradient elution on either pre-packed columns (Isolute) on a Flashmaster II system (Biotage) or on a Teledyne ISCO CombiFlash R f automated flash chromatography system with RediSep R f disposable flash columns. 1 H and 13 C NMR spectra were recorded on 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 (UK). Fast atom bombardment (FAB) mass spectra were measured using m-nitrobenzyl alcohol as the matrix. Elemental analyses were performed by the Microanalysis Service, University of Bath (UK). Melting points (mp) were determined using either a Stuart Scientific SMP3 or a Stanford Research Systems Optimelt MPA100 and are uncorrected. Optical rotations were measured with a machine supplied by Optical Activity Ltd using 5 cm cells.
LC/MS was performed using a Waters 2790 machine with a ZQ Mi-croMass spectrometer and photodiode array (PDA) detector. The ionisation technique used was either atmospheric pressure chemi-cal ionisation (APCI) or electrospray ionisation (ESI). A Waters "Symmetry" C18 column (packing: 3.5 mm, 4.6 100 mm) and gradient elution were used (MeCN/H 2 O, 5:95 at 0.5 mL min À1 !95:5 at 1 mL min À1 over 10 min). HPLC was undertaken using a Waters 717 machine with an autosampler and PDA detector. The column used was either a Waters "Symmetry" C18 (packing: 3.5 mm, 4.6 150 mm) or a Waters "Sunfire" C18 (packing: 3.5 mm, 4.6 150 mm) with an isocratic mobile phase consisting of CH 3 CN/H 2 O (as indicated) with a flow rate of 1 mL min À1 . Analytical chiral HPLC was performed on a Chiralpak AD-H column (250 4.6 mm, 5 mm) with MeOH as the mobile phase, a flow rate of 1.2 mL min À1 and a PDA detector. Semi-preparative HPLC was performed with a Waters 2525 binary gradient module and a Chiralpak AD-H (250 20 mm) semi-prep column with MeOH as the mobile phase at a flow rate of 10 mL min À1 , injecting 1.5-2.0 mL of a 20 mg mL À1 solution and a run time of 25 min.
Method A: Condensation of carbinols with triazole: Substrate, 1,2,4-triazole and para-toluenesulfonic acid (p-TsOH), dissolved/suspended in toluene were heated at reflux with a Dean-Stark separator for 24 h. The reaction mixture was allowed to cool, and the solvent was removed in vacuo.
Method B: Hydrogenation: Pd/C (10 %) was added to a solution of substrate in THF/MeOH (1:1). The solution was stirred overnight under an H 2 atmosphere (maintained using a balloon). Excess H 2 was removed, and the reaction mixture was filtered through Celite, washing with THF and MeOH, and then the solvent was removed in vacuo.
Method C: Sulfamoylation: A solution of sulfamoyl chloride (H 2 NSO 2 Cl) in toluene was concentrated in vacuo at 30 8C to furnish a yellow oil, which solidified upon cooling in an ice bath. DMA and substrate were subsequently added and the mixture was allowed to warm to RT and stirred overnight. The reaction mixture was poured into H 2 O and extracted with EtOAc (2 ). The organic layers were combined, washed with H 2 O (4 ) and brine, dried (MgSO 4 ), filtered, and the solvent was removed in vacuo.