sign in sign up
<<
Home , Deoxycholic acid, Tetracycline

Molecular and Cellular Endocrinology 428 (2016) 133e141

Contents lists available at ScienceDirect

Molecular and Cellular Endocrinology

journal homepage: www.elsevier.com/locate/mce

Identification of novel inhibitors of the sulfate carrier ‘sodium- dependent organic anion transporter’ SOAT (SLC10A6)by pharmacophore modelling*

Gary Grosser a, Karl-Heinz Baringhaus b, Barbara Doring€ a, Werner Kramer b, * Ernst Petzinger a, Joachim Geyer a, a Institute of Pharmacology and Toxicology, Justus Liebig University of Giessen, 35392 Giessen, Germany b Sanofi-Aventis Deutschland GmbH, R&D, LGCR, Structure, Design and Informatics, Building G 878, 65926 Frankfurt am Main, Germany article info abstract

Article history: The sodium-dependent organic anion transporter SOAT specifically transports sulfated steroid Received 27 November 2015 and is supposed to play a role in testicular steroid regulation and male fertility. The present study aimed Received in revised form to identify novel specific SOAT inhibitors for further in vitro and in vivo studies on SOAT function. More 25 February 2016 than 100 compounds of different molecular structures were screened for inhibition of the SOAT- Accepted 21 March 2016 mediated transport of sulfate in stably transfected SOAT-HEK293 cells. Available online 23 March 2016 Twenty-five of these with IC50 values covering four orders of magnitude were selected as training set for 3D pharmacophore modelling. The SOAT pharmacophore features were calculated by CATALYST and Keywords: consist of three hydrophobic sites and two hydrogen bond acceptors. By substrate database screening, SOAT m DHEAS transport compound T 0511-1698 was predicted as a novel SOAT inhibitor with an IC50 of 15 M. This value was ASBT confirmed by cell-based transport assays. Therefore, the developed SOAT pharmacophore model Pharmacophore model demonstrated its suitability in predicting novel SOAT inhibitors. Inhibitor © 2016 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC- SLC10 ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction (Mueller et al., 2015). However, sulfo-conjugated steroid hormones can actively be imported into specific target cells via uptake carriers Sulfated steroid hormones, such as dehydroepiandrosterone and, after hydrolysis by the steroid sulfatase (StS) (so-called sulfa- sulfate (DHEAS) or estrone-3-sulfate, are usually considered to be tase pathway), contribute to the overall steroid regulation (Reed biologically inactive metabolites as they cannot activate classical et al., 2005; Pasqualini and Chetrite, 2012; Labrie, 2015). In 2004 steroid receptors (Strott, 2002). They are present in the blood cir- we identified a novel uptake carrier, named sodium-dependent culation at quite high concentrations, but generally exhibit low organic anion transporter SOAT, which specifically transports membrane permeation due to their physicochemical properties sulfated steroid hormones, such as DHEAS, estrone-3-sulfate, b- -3-sulfate, sulfate, and -3- sulfate in a sodium-dependent manner (Geyer et al., 2004). In rat Abbreviations: ASBT, apical sodium-dependent acid transporter; BSP, bro- and mouse, Soat showed a broader tissue expression, including mosulfophthalein; DHEAS, dehydroepiandrosterone sulfate; FRT, Flp recombinase , testis and lung (Geyer et al., 2004; Grosser et al., 2013), while target site; HEK293, Human Embryonic Kidney 293 cells; IC50, inhibitory concen- þ in humans SOAT is predominantly expressed in the testis with tration 50; NTCP, Na /taurocholate co-transporting polypeptide; PBS, phosphate- buffered saline; QSAR, quantitative structure-activity relationship; SLC10, lower expression in the placenta and breast tissue (Geyer et al., solute carrier family 10; SOAT, sodium-dependent organic anion transporter; TC, 2007). In the testis, SOAT/Soat was localised to spermatocytes and ; TLCS, taurolithocholic acid 3-sulfate. round spermatids and, therefore, this carrier is supposed to play a * This manuscript is dedicated to Prof. Dr. Ernst Petzinger who suddenly died role in testicular steroid regulation and male fertility (Fietz et al., fi during nalizing of this manuscript, in remembrance of his person and his aca- 2013; Grosser et al., 2013). More recently, SOAT expression was demic achievements. * Corresponding author. Justus Liebig University of Giessen, Institute of Phar- also demonstrated in breast tissue (unpublished data). Here, macology and Toxicology, Biomedical Research Center Seltersberg (BFS), Schu- SOAT may contribute to the -dependent proliferation due bertstr. 81, 35392 Giessen, Germany. to the import of estrone-3-sulfate and even DHEAS, making SOAT a E-mail address: [email protected] (J. Geyer). http://dx.doi.org/10.1016/j.mce.2016.03.028 0303-7207/© 2016 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc- nd/4.0/). 134 G. Grosser et al. / Molecular and Cellular Endocrinology 428 (2016) 133e141 potential drug target for anti-proliferative therapy. number NM_000452. SOAT ( symbol SLC10A6) phylogenetically belongs to the solute carrier family SLC10, which overall consists of seven mem- bers (SLC10A1eSLC10A7). The founding members of this carrier þ 2.3. Inhibition studies in SOAT-HEK293 and ASBT-HEK293 cells family, the hepatic Na /taurocholate co-transporting polypeptide NTCP (SLC10A1) and the intestinal apical sodium-dependent bile For inhibition studies, 24-well plates were coated with poly-D- acid transporter ASBT (SLC10A2) are important factors for the for better attachment of the cells. Cells (1.25 105 per well) maintenance of the enterohepatic circulation of bile acids between were plated and grown in standard medium for 72 h. SOAT and the liver and the gut (Geyer et al., 2006; Doring€ et al., 2012). ASBT expression was induced by pre-incubation with tetracycline Furthermore, ASBT is of particular interest in several pharmaco- (1 mg/ml). SOAT and ASBT non-expressing control cells were logical aspects. Regarding molecular drug design, ASBT-mediated not pre-treated with tetracycline. Before starting the transport uptake from the gut is used to improve the oral of experiments, cells were washed three times with phosphate- drugs by coupling them with bile acids (Kramer et al., 1994; Kramer, buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 1.5 mM KH PO , 2011). Furthermore, ASBT is a drug target for cholesterol-lowering 2 4 7.3 mM Na HPO , pH 7.4, 37 C) and equilibrated with transport therapy, because inhibition of reuptake in the gut stimu- 2 4 buffer (142.9 mM NaCl, 4.7 mM KCl,1.2 mM MgSO , 1.2 mM KH PO , lates de novo bile acid synthesis from cholesterol in the liver 4 2 4 1.8 mM CaCl , 10 mM glucose and 20 mM HEPES, adjusted to pH (Kramer and Glombik, 2006). 2 7.4). Then, cells were pre-incubated with transport buffer con- In the present study, we searched for novel SOAT inhibitors for taining increasing concentrations of the tested compound for further in vitro and in vivo studies in order to clarify the role of SOAT 5 min. For control, transport buffer without the test compound was in testicular steroid regulation and for proliferation of hormone- used. Transport measurements were started by adding 200 nM dependent breast cancer cells. Therefore, we analysed the SOAT [3H]DHEAS or 200 nM [3H]TC, while keeping the concentrations of inhibitory pattern in detail with a set of more than 100 different the test compounds constant. All uptake experiments were per- compounds and established a 3D pharmacophore model for the formed over 5 min at 37 C. Although SOAT uptake of DHEAS was identification of further inhibitors of SOAT. Because of the very high only linear over 1 min (Geyer et al., 2007), this longer uptake phase amino acid sequence identity of SOAT and ASBT, we also compared was required to ensure adequate uptake ratios for subsequent IC the SOAT pharmacophore model with the already established 50 determinations. However, comparative uptake experiments with pharmacophore model of rabbit Asbt (Baringhaus et al., 1999)in 1 min or 5 min uptake phase revealed identical IC values for order to determine the potential cross-inhibition between both 50 selected compounds (data not shown). The transport phase was carriers. Such a cross-inhibition after oral application of an SOAT terminated by removing the transport buffer and washing the cells inhibitor would be of pharmacological relevance as ASBT inhibition five times with ice-cold PBS. Cell-associated radioactivity and would hamper the enterohepatic circulation of bile acids. protein contents were determined as described previously (Geyer et al., 2007). For calculation of the IC values, the negative con- 2. Material and methods 50 trol (uptake in not carrier expressing HEK293 cells) was set to 0% and the respective positive control (uptake in carrier expressing 2.1. Chemicals and radiochemicals HEK293 cells without inhibitor) was set to 100%. All chemicals, unless otherwise stated, were from Sigma- eAldrich. Zeocin and hygromycin were purchased from Invitrogen. Materials used for cultivation of HEK293 cells were purchased from 2.4. Generation of the 3D QSAR model Gibco and SigmaeAldrich. [3H]Dehydroepiandrosterone sulfate ([3H]DHEAS, 94.5 Ci/mmol) and [3H]taurocholic acid ([3H]TC, 10 Ci/ The 3D quantitative structure-activity relationship (QSAR) mmol) were purchased from PerkinElmer Life Sciences. SAL-II-68, model was generated using the CATALYST software (version 2.11, SAL-II-156 and EMe I 4 were kindly provided by Dr. Alakurtti and Accelrys). All molecules of the training set were built in CATALYST Prof. Yli-Kauhaluoma. RR Scymnol sulfate was kindly provided by and transformed into 3D structures followed by a local mini- Prof. Fricker. All compounds used for the validation of the SOAT misation using the CHARMM-like force. Subsequent conforma- pharmacophore model, such as T 0511-1698, were purchased from tional analysis within a 20 kcal/mol energy window (“best Enamine. searching procedure”) yielded a diverse and representative set of up to 250 conformations per molecule (Smellie et al., 1995). 2.2. Cloning and establishment of cell lines CATALYST uses molecular structures as templates consisting of chemical functions positioned in space. It is assumed that the most Cloning of human SOAT and establishment of a stably trans- relevant biological features bind effectively with complementary fected SOAT-HEK293 cell line was reported previously (Geyer et al., functions on the respective binding protein. The model generation 2007). Briefly, the full-length SOAT cDNA was cloned into the Flp-In in CATALYST (CatHypo module) starts with the selection of the pcDNA5/FRT/TO expression carrying a Flp recombinase most relevant biological features from the most active compounds target site (FRT) and a hygromycin resistance gene (Invitrogen). In of the training set (Sprague and Hoffmann, 1997). CATALYST then this vector, SOAT expression is under the control of the cytomeg- evaluates all generated models against all compounds of the alovirus promoter and a tetracycline operator sequence. The SOAT- training set covering a broad range of activity. Only models that HEK293 cell line was established using the Flp-In expression sys- best explain the entire structure-activity relationship in the tem and the commercially available Flp-In T-Rex 293 host cell line training set are finally reported. This statistical relevance takes into (Invitrogen) with stable expression of the tetracycline repressor. In account the cost of each model relative to the null hypothesis as the absence of tetracycline, the tetracycline repressor effectively well as the correlation coefficients. CATALYST pharmacophores are binds to the tetracycline operator sequence and blocks SOAT tran- described by a set of hydrophobic, hydrogen bond donor and scription from the cytomegalovirus promoter. In the same way, a acceptor functions, as well as positively and negatively ionisable stably transfected ASBT-HEK293 cell line was established with the features distributed within 3D space. The hydrogen bonding fea- human ASBT cDNA sequence according to GenBank accession tures are vectors, while all other functions are points. G. Grosser et al. / Molecular and Cellular Endocrinology 428 (2016) 133e141 135

Table 1 Bile acids are inhibitors of the SOAT-mediated DHEAS transport.

Bile acid R1 R2 R3 R4 R5 IC50 (mM) OH H OH OH OH 177.6 OH H OH H OH 11.2 OH H b-OH H OH 384.8 Deoxycholic acid OH H H OH OH 100.1 OH H H H OH 10.4 7-Ketolithocholic acid OH H O H OH 164.2

Lithocholic acid 3-sulfate OSO3H H H H OH 4.2 Hyocholic acid OH OH OH H OH 971.8 OH OH H H OH 172.6 - Glycocholic acid OH H OH OH NHCH2CO2 284.8 - Glycodeoxycholic acid OH H H OH NHCH2CO2 46.7 - Glycochenodeoxycholic acid OH H OH H NHCH2CO2 26.6 - Glycoursodeoxycholic acid OH H b-OH H NHCH2CO2 100.8 - Taurocholic acid OH H OH OH NHCH2CH2SO3 65.3 - Taurodeoxycholic acid OH H H OH NHCH2CH2SO3 76.3 - Taurochenodeoxycholic acid OH H OH H NHCH2CH2SO3 38.1 - Tauroursodeoxycholic acid OH H b-OH H NHCH2CH2SO3 49.6 - Taurolithocholic acid OH H H H NHCH2CH2SO3 3.0 - Taurolithocholic acid 3-sulfate OSO3H H H H NHCH2CH2SO3 0.5

3. Results

3.1. Initial screening Table 2 Inhibitory potency of compounds with steroid-like structure on Although SOAT has no general transport activity for bile acids, SOAT-mediated DHEAS transport. several bile acids were quite good SOAT inhibitors in previous ex- IC50 (mM) periments (Geyer et al., 2007). For more systematic analysis of the Estrone-3-sulfate 22.1 inhibitory pattern of SOAT, a series of physiologically occurring bile 3 Pregnenolone-3-sulfate 9.1 acids was used to obtain IC50 values by inhibition of [ H]DHEAS b-estradiol-3-sulfate 145.9 transport of stably transfected SOAT-HEK293 cells. Table 1 shows b-estradiol-3,17-disulfate 133.2 the structures of the bile acids used with the corresponding IC50 -21-sulfate 323.8 > values, which cover about four orders of magnitude Sodium cholesteryl sulfate 1000 m e m Cortisone 29.6 (0.5 M 972 M). Direct comparison of lithocholic acid and >1000 taurolithocholic acid with their corresponding sulfated forms Corticosterone >1000 revealed 2- to 6-fold higher inhibitory potency with the sulfate Aldosterone >1000 Androstendione >1000 group at position 3 (R1). The secondary bile acids, hyocholic acid Androsterone >1000 and hyodeoxycholic acid, are characterised by additional hydroxyl >1000 groups at position 6 (R2), which decreased the inhibitory potency Prednisone >1000 compared to chenodeoxycholic acid and lithocholic acid by 17- and >1000 87-fold, respectively. Among all bile acids tested, hyocholic acid Testosterone >1000 > with its four hydroxyl groups was identified as the least potent Dihydrotestosterone 1000 > ¼ m Estrone 1000 inhibitor of SOAT (IC50 971.8 M). At position 7 (R3), the following 17b-estradiol >1000 inhibitory pattern was found, starting with the most potent com- Estriol ~100 pounds: H (lithocholic acid) z 7a-OH (chenodeoxycholic ~900 acid) > ¼O (7-ketolithocholic acid) > 7b-OH (ursodeoxycholic acid). Ouabain >1000 > Furthermore, the presence of a hydroxyl group at position 12 (R ) 1000 4 Digitonin 4.1 fi signi cantly reduced the inhibitory potency of the bile acid (e.g. Digitoxigenin >1000 chenodeoxycholic acid vs. cholic acid). In contrast, substitutions of Finasteride >1000 > the bile acids at R5 with glycine showed no general trend for better Wortmannin 1000 136 G. Grosser et al. / Molecular and Cellular Endocrinology 428 (2016) 133e141 or lower inhibition, while substitution with taurine at that position similar compounds with different IC50 values, as well as structurally revealed a distinct improvement of inhibition, in particular, for divergent compounds with similar IC50 values. Overall, the IC50 tauroursodeoxycholic acid and taurolithocholic acid 3-sulfate, but values covered four orders of magnitude (IC50 ¼ 0.15 mMe1000 mM) not for taurochenodeoxycholic acid. In summary, 3a-monohydroxy (Table 3). Fig. 1 illustrates some of the training set compounds. bile acids and bile acids with a sulfate group at position 3 are the Additional structural information for the used bile acids is given in most potent inhibitors of SOAT from the group of bile acids. How- Table 1, and compounds previously used as the training set for the ever, it has to be noted that taurolithocholic acid 3-sulfate is not Asbt pharmacophore (Baringhaus et al., 1999) are given in Fig. 2. only a potent inhibitor, but also a week substrate of SOAT (Geyer The best hypothesis for the calculated SOAT pharmacophore model, et al., 2007). proposed by CATALYST, is illustrated in Fig. 3 and consists of three Bile acids normally have a typical cis-trans-trans conformation hydrophobic sites (blue spheres) and two hydrogen bond acceptors of the A, B and C rings of the steroid nucleus. In contrast, sulfo- (two green spheres, with the smaller sphere giving the origin and conjugated steroid hormones, which are substrates of SOAT, are the larger sphere giving the direction of the hydrogen bond). Fig. 4 characterised by a trans-trans-trans conformation of the rings. depicts the correlation between the experimentally measured IC50 Consequently, we additionally determined the IC50 values of a set of values versus the predicted activities of the training set com- trans-trans-trans steroid hormones (Table 2). Most of these com- pounds, whereby the measured IC50 values of >1000 were set to pounds did not inhibit SOAT when they were not sulfo-conjugated. 10.000. The correlation coefficient was at r ¼ 0.89 ± 0.32, demon- In contrast, the sulfo-conjugated forms e estrone-3-sulfate and strating good predictive ability of this SOAT pharmacophore model. pregnenolone-3-sulfate e showed strong inhibition of the DHEAS For example, the best inhibitory bile acid TLCS (IC50 ¼ 0.5 mM) maps transport by SOAT. Both compounds have previously been identi- all five features of the pharmacophore model as follows (Fig. 3A): fied as substrates of SOAT (Geyer et al., 2007). Therefore, the the hydrophobic features cover the methyl groups at positions 18 inhibitory effect of these compounds can be regarded as a and 21, as well as the A-ring of the steroid nucleus, and the sulfate competition with another transported substrate (DHEAS). In the group at position 3, as well as the sulfonyl group of the taurine case of b-estradiol-3,17-disulfate and corticosterone-21-sulfate it is residue, act as hydrogen bond acceptors. In contrast, the less potent currently unknown if these compounds are also transported by inhibitors bromosulfophthalein (BSP) (IC50 ¼ 3.6 mM) and S 1647 SOAT or not. Interestingly, whereas cortisol with a hydroxyl group (IC50 ¼ 1.1 mM) only matched four of the five pharmacophore fea- at position 11 did not show any inhibition of SOAT, cortisone with a tures (Fig. 3B and C, respectively). While BSP does not cover one of ketone group at the same position revealed pronounced inhibitory the hydrogen bond acceptor groups (Fig. 3B), S 1647 lacks mapping potency, indicating that the oxidation status at the C11 position is of of a hydrophobic site (Fig. 3C). significant relevance for SOAT inhibition. Finally, digitonin, a steroid In order to identify novel SOAT inhibitors, the Sanofi-Aventis glycoside from digitalis purpurea with a cis-trans-cis conformation database of commercially available molecules (~12 million mole- of the steroid rings, was a powerful inhibitor of SOAT. cules) was screened for compounds matching the pharmacophore and revealed 67,360 hits. After applying a >50% fitting value, 3437 virtual hits remained. This number was decreased by a shape query 3.2. 3D QSAR pharmacophore model of SOAT of TLCS to 180 hits. A shape restriction to S 1647 limited the number of hits to 75 compounds and an intersection of both (TLCS and Apart from bile acids and steroid hormones, many further S 1647) shape queries limited the number of hits to four com- compounds were tested as potential inhibitors of SOAT. For devel- pounds. For the shape query of BSP, no compound was found. From opment of the pharmacophore model, 25 of these compounds were these hits, compounds with particularly low estimated IC50 values selected and used as a training set. These included structurally were selected for validation of the SOAT pharmacophore model and were used for experimental determination of the IC50 values. Whereas, T 5239532 (IC50 ¼ 137 mM), T 5573915 (IC50 ¼ 57 mM) and Table 3 ¼ m Compound training set for calculation of the SOAT pharmacophore model. T 5854015 (IC50 9 M) showed higher than predicted IC50 values (being 0.83 mM, 1.43 mM, and 1.28 mM, respectively), the predicted Training set compounds IC50 (mM) measured IC50 (mM) predicted and measured IC50 values were identical for compound T 0511- S 0960 0.15 0.077 1698, being 15 mM(Table 4, Fig. 5). Therefore, the 3D pharmaco- Taurolithocholic acid 3-sulfate 0.5 0.64 phore model was suitable in predicting novel SOAT inhibitors of S 1647 1.1 2.2 completely different chemical structures compared to the com- BSP 3.6 17 SAL-II-68 3.6 11 pounds of the training set. Closer analysis revealed that only four of EMe I 4 5.4 16 five features of the SOAT pharmacophore are mapped by compound Lithocholic acid 10.4 21 T 0511-1698, and one of the hydrogen bond acceptors is not covered S 9202 18.7 36 (Fig. 5B), which might explain the lower inhibitory power of T 0511- S 9086 23.1 23 fi RR Scymnol sulfate 23.3 50 1698 compared with TLCS, which mapped all ve pharmacophore Cortisone 29.6 280 features. Furthermore, all compounds were additionally tested for L-Thyroxine 49.5 250 potential cross-inhibition of ASBT. Only compound T 0511-1698 Tauroursodeoxycholic acid 49.6 32 showed weak interaction with ASBT with IC50 of 350 mM(Table 4). S 9087 50 38 Taurocholic acid 65.3 79 SAL-II-156 66.9 16 3.3. Comparison of the SOAT and ASBT pharmacophore models Deoxycholic acid 100.1 83 7-Ketolithocholic acid 164.2 110 Because of the high sequence identity between SOAT and ASBT, Cholic acid 177.6 76 we were interested to determine whether the 3D pharmacophore 4-Methylumbelliferyl sulfate 255.7 1400 Ursodeoxycholic acid 384.8 130 models of these carriers might overlap. Therefore, we tested 14 Hyocholic acid 971.8 200 compounds that were used as the training set of the previously Sodium cholesteryl sulfate >1000 430 published Asbt pharmacophore (Baringhaus et al., 1999)(Fig. 2) for Cortisol >1000 1000 SOAT inhibition and compared the measured IC50 values (Table 5, Wortmannin >1000 4700 Fig. 6). Direct comparison of these data indicated no significant G. Grosser et al. / Molecular and Cellular Endocrinology 428 (2016) 133e141 137

Fig. 1. Chemical structures of the molecules that were included in the training set (Table 3) for generation of the SOAT pharmacophore model.

Fig. 2. Chemical structures of the molecules that were used as the training set for generation of the Asbt pharmacophore model by Baringhaus et al. (1999) and that were also experimentally tested for SOAT inhibition in the present study (Table 5).

correlation of the IC50 values between Asbt and SOAT (r ¼ 0.54 and whereas the thioketo group of the same ring maps the hydrogen p ¼ 0.04; Fig. 6). However, certain compounds were good inhibitors bond acceptor of the Asbt pharmacophore. The other hydrogen for both carriers, such as S 3720 (IC50 ¼ 8 mM for Asbt and 1.1 mM for bond acceptor of the SOAT pharmacophore, as well as the hydrogen SOAT) and S 1647 (IC50 ¼ 4 mM for Asbt and 1 mM for SOAT), which bond donor of the Asbt pharmacophore, remains unfilled. From this allows us to suggest that both pharmacophore models might share direct comparison of the pharmacophore mapping of S 3720, it can some common domains. Both pharmacophore models have three be concluded that both pharmacophore models overlap in their hydrophobic features, but differ in their hydrogen bond donor/ hydrophobic features, but significantly differ in the orientation of acceptor features. Interestingly, fitting of compound S 3720 into their hydrogen bond donor/acceptor groups. both pharmacophore models showed mapping for all three hy- drophobic features by the identical structural domains (chloro 4. Discussion group, benzyl group, and phenyl group) of S 3720 (Fig. 7). However, one hydrogen bond acceptor of the SOAT pharmacophore fits to the In 2004, SOAT was identified as a novel member of the SLC10 ketone group of the 2-thioxo-pyrimidin-4,6-dion ring of S 3720, carrier family, which includes the bile acid transporters NTCP and 138 G. Grosser et al. / Molecular and Cellular Endocrinology 428 (2016) 133e141

Fig. 4. Correlation diagram between the predicted IC50 values (mM) of the training set compounds (Table 3) by using the statistically most significant hypothesis of the SOAT

pharmacophore model and the experimentally measured IC50 values. Correlation co- efficient reported by CATALYST: r ¼ 0.89 ± 0.32.

(Geyer et al., 2007; Fietz et al., 2013; Galuska et al., 2013; Schweigmann et al., 2014). Sulfated steroid hormones are present in the plasma at much higher concentrations than the corre- sponding free steroid forms, and are considered as a reservoir for the synthesis of free steroid hormones, which have potent regula- tory functions (Labrie, 2015). While lipophilic free steroid hor- mones are able to pass the cell membrane by diffusion, the negatively charged steroid sulfates can access the intracellular compartment only by carrier-mediated import via SOAT or other uptake carriers (Geyer et al., 2007; Mueller et al., 2015). After cleavage of the sulfate group by the steroid sulfatase these im- ported steroid sulfates can participate in the steroid regulation at nuclear and androgen receptors (Reed et al., 2005; Pasqualini and Chetrite, 2012). SOAT is highly expressed in germ cells of the human testis and is supposed to play a role in testicular steroid regulation and male fertility by importing sulfated steroid hormones (Fietz et al., 2013; Grosser et al., 2013). Furthermore, Fig. 3. Pharmacophore model of SOAT with matching of the compounds TLCS (A), BSP SOAT was localised in hormone-dependent breast cancer tissue (B) and S 1647 (C). The three hydrophobic areas of the pharmacophore model are (unpublished data) and here might be of pathophysiological rele- illustrated by blue spherical regions, each hydrogen bond acceptor is displayed in vance due to steroid sulfate import. In the present study, we aimed green by two orbital regions, representing the position of an accepted hydrogen (larger to develop a common feature-based 3D pharmacophore model for sphere) and the electronegative charge of the hydrogen bond acceptor position fi (smaller sphere). (A) TLCS maps all five features of the molecules: two proximate SOAT in order to identify novel and speci c SOAT inhibitors for hydrophobic sites at the methyl groups of position 18 and 21, as well as the A-ring of further in vivo and in vitro studies in both fields. the steroid nucleus ( blue); the electronegative sulfate group at position 3 as well Despite the high phylogenetic relationship to ASBT, SOAT is as the sulfonyl group of the taurine residue act as hydrogen bond acceptors (green). (B) transport-negative for the bile acids taurocholic acid, cholic acid, The three hydrophobic sites (light blue) of BSP fit to the model, as well as one hydrogen bond acceptor (green), but the second hydrogen bond acceptor remains free. (C) Both chenodeoxycholic acid and deoxycholic acid (Geyer et al., 2007). hydrogen bond acceptors (green) and two hydrophobic sites (light blue) fit to com- However, several bile acids are quite good inhibitors of SOAT, even pound S 1647, but one hydrophobic site is not covered. The spheres used for shape in the low micro molar range. Bile acids have certain structural restriction of BSP and S 1647 are additionally indicated in grey. similarities to steroid hormones, but with one major difference: whereas the A/B rings of the steroid nucleus are trans orientated in the steroid hormones, they have a cis orientation in the bile acid molecules, and this might discriminate between steroid-like SOAT ASBT (Geyer et al., 2004, 2006). SOAT specifically transports substrates and inhibitors (Geyer et al., 2007; Doring€ et al., 2012). sulfated steroid hormones, including DHEAS, 16a-hydroxy-DHEAS, Interestingly, further hydroxyl groups at positions 7 and/or 12 of androstenediol-3-sulfate, estrone-3-sulfate, b-estradiol-3-sulfate, the bile acid molecule, which are not present in steroid hormones, and pregnenolone sulfate in a strictly sodium-dependent manner significantly decreased the inhibitory power of the respective bile G. Grosser et al. / Molecular and Cellular Endocrinology 428 (2016) 133e141 139

Table 4 Validation of the SOAT pharmacophore model.

Predicted SOAT inhibitors Shape query IC50 (mM) predicted SOAT IC50 (mM) measured SOAT IC50 (mM) measured ASBT T 5239532 TLCS 0.83 137 >1000 T 5573915 S 1647 1.43 57 >1000 T 5854015 S 1647 1.28 9 >1000 T 0511-1698 S 1647 and TLCS 15 15 350

Fig. 6. Correlation of the experimentally measured IC50 values (mM) for all compounds depicted in Fig. 2 and listed in Table 5 between Asbt and SOAT. The values for Asbt were reported before by Baringhaus et al. (1999) and the values for SOAT were experimentally determined in the present study.

structure for the development of novel specific SOAT inhibitors. Fig. 5. Chemical structure (A) and SOAT pharmacophore mapping (B) of compound Apart from bile acids, further non-steroidal SOAT inhibitors have T 0511-1698, which was used for validation of the SOAT pharmacophore model (see fi u also Table 3). been identi ed, including the toxins 1-( -sulfooxyethyl)pyrene and 2-sulfooxymethylpyrene and the hormone L-thyroxine, as well as the hepatodiagnostic dye and NTCP substrate BSP. Furthermore, acid, indicating that a non-substituted steroid nucleus might be one the propanolamine compounds S9087, S8214, S9202 and S9203, as of the possible core structures that are required for binding to SOAT. well as the barbiturate compound S3740, showed significant inhi- However, as a steroid compound generally has the potential for bition of the SOAT transport with IC50 < 50 mM. The latter com- cross-inhibition of ASBT, it is not regarded as a promising core pounds were previously used as training-set compounds for

Table 5 Comparison of IC50 values of the Asbt training set compounds between Asbt and SOAT.

Training set compounds IC50 values for Asbt (reported by Baringhaus et al., 1999)IC50 values for SOAT (measured in the present study) S 9087 0.3 50 S 9203 1 37.7 S 0382 3 75.2 S 1647 4 1.1 S 3068 4 1 S 8214 4 14.9 S 3740 8 1.1 S 0925 16 ~1000 S 1690 20 0.7 S 0960 30 0.15 S 9202 90 18.7 S 0381 700 ~1000 S 9086 1500 23.1 S 8005 3000 ~1000 140 G. Grosser et al. / Molecular and Cellular Endocrinology 428 (2016) 133e141

NTCP are able to transport sulfated steroid hormones (Petzinger and Geyer, 2006), and could also show cross-inhibition with SOAT inhibitors. These carriers are largely expressed in the liver and kidney. So, inhibition of these carriers could reduce the overall hepatobiliary and renal of steroid sulfates. As these car- riers play also an important role for the excretion of many drugs, relevant drug-drug-interactions cannot be excluded for novel SOAT inhibitors. In conclusion, we were able to develop a 3D pharmacophore model for SOAT with three hydrophobic sites and two hydrogen bond acceptors. The bile acid with the highest inhibitory power, TLCS, mapped all five features. SOAT inhibition by compound T 0511-1698 was predicted by the pharmacophore model with the correct inhibitory power, demonstrating validity and reliability of the model for certain compounds. Furthermore, compound T 5854015 showed an even lower IC50 of 9 mM and absent cross- inhibition with ASBT and so is the most promising candidate for further studies. Although SOAT and ASBT show high sequence ho- mology, their pharmacophore models significantly differ in their hydrogen bond donor/acceptor groups, which might explain the lack of correlation of the IC50 values between both carriers for most of the compounds tested. On the other hand, some compounds showed cross-inhibition of SOAT and ASBT, probably because of the overlapped mapping to the three hydrophobic features, which are equally orientated in both pharmacophore models. Therefore, novel SOAT inhibitors should always be tested for cross-inhibition of ASBT and vice versa.

Acknowledgements

The authors thank Anita Neubauer and Klaus Schuh for their excellent technical help. We also thank Dr. Sami Alakurtti (VTT Technical Research Centre of Finland) and Prof. Jari Yli-Kauhaluoma (Faculty of Pharmacy, University of Helsinki) for providing the compounds SAL-II-68, SAL-II-156 and EMe I 4, as well as Prof. Fig. 7. Mapping of compound S 3740 to the pharmacophore models of (A) SOAT and (B) Asbt (published at Baringhaus et al., 1999). Fricker (Institute of Pharmacy and Molecular Biotechnology, Uni- versity of Heidelberg) for providing RR scymnol sulfate. generation of the Asbt pharmacophore model (Baringhaus et al., References 1999) and all showed significant inhibition of the Asbt, which Baringhaus, K.H., Matter, H., Stengelin, S., Kramer, W., 1999. Substrate specificity of þ might be explained by partial overlap of the two pharmacophore the ileal and the hepatic Na /bile acid cotransporters of the rabbit. II. A reliable þ models for Asbt and SOAT. Therefore, a specific and non-toxic SOAT 3D QSAR pharmacophore model for the ileal Na /bile acid cotransporter. J. Lipid e inhibitor could not be established so far. However, based on the Res. 40 (12), 2158 2168. Doring,€ B., Lütteke, T., Geyer, J., Petzinger, E., 2012. The SLC10 carrier family: SOAT pharmacophore model developed in the present study, transport functions and molecular structure. Curr. Top. Membr. 70, 105e168. several new molecules of different chemical structures (Fig. 5A) Fietz, D., Bakhaus, K., Wapelhorst, B., Grosser, G., Günther, S., Alber, J., Doring,€ B., were identified that all showed low or even no cross-inhibition Kliesch, S., Weidner, W., Galuska, C.E., Hartmann, M.F., Wudy, S.A., Bergmann, M., Geyer, J., 2013. Membrane transporters for sulfated steroids in with ASBT. Based on these structures, more potent SOAT in- the human testis-cellular localization, expression pattern and functional anal- hibitors can now be developed. ysis. PLoS One 8 (5), e62638. However, it has to be stated that the SOAT pharmacophore Galuska, C.E., Hartmann, M.F., Sanchez-Guijo, A., Bakhaus, K., Geyer, J., Schuler, G., Zimmer, K.P., Wudy, S.A., 2013. Profiling intact steroid sulfates and unconju- model also has some limitations, as the inhibitory power prediction gated steroids in biological fluids by liquid chromatography-tandem mass did not fit with the experimentally measured IC50 values for all spectrometry (LC-MS-MS). Analyst 138, 3792e3801. compounds analysed. Furthermore, there are some discrepancies Geyer, J., Godoy, J.R., Petzinger, E., 2004. Identification of a sodium-dependent fi organic anion transporter from rat adrenal gland. Biochem. Biophys. Res. between the tting status into the pharmacophore model and the Commun. 316 (2), 300e306. inhibitory power of certain compounds. For example, BSP as well as Geyer, J., Wilke, T., Petzinger, E., 2006. The solute carrier family SLC10: more than a compound T 0511-1698 both map only four of the five features of family of bile acid transporters regarding function and phylogenetic relation- e the SOAT pharmacophore, whereas the real IC value was correctly ships. Schmiedeb. Arch. Pharmacol. 372, 413 431. 50 Geyer, J., Doring,€ B., Meerkamp, K., Ugele, B., Bakhiya, N., Fernandes, C.F., Godoy, J.R., predicted for T 0511-1698, but not for BSP. This might indicate that Glatt, H., Petzinger, E., 2007. Cloning and functional characterization of human further features could be relevant for SOAT inhibition which are not sodium-dependent organic anion transporter (SLC10A6). J. Biol. Chem. 282 (27), e covered by the SOAT pharmacophore model. 19728 19741. Grosser, G., Fietz, D., Günther, S., Bakhaus, K., Schweigmann, H., Ugele, B., Brehm, R., The present study only analysed cross-inhibition of the SOAT Petzinger, E., Bergmann, M., Geyer, J., 2013. Cloning and functional character- inhibitors with ASBT, as the phylogenetically most related carrier to ization of the mouse sodium-dependent organic anion transporter Soat SOAT. But it has to be noted that apart from SOAT further carriers (Slc10a6). J. Steroid Biochem. Mol. Biol. 138, 90e99. Kramer, W., Wess, G., Neckermann, G., Schubert, G., Fink, J., Girbig, F., Gutjahr, U., from the Organic Anion Transporter (OAT) family and from the Kowalewski, S., Baringhaus, K.H., Boger, G., Enhsen, A., Falk, E., Friedrich, M., Organic Anion Transporting Polypeptide (OATP) family as well as Glombik, H., Hoffmann, A., Pittius, C., Urman, M., 1994. Intestinal absorption of G. Grosser et al. / Molecular and Cellular Endocrinology 428 (2016) 133e141 141

peptides by coupling to bile acids. J. Biol. Chem. 269 (14), 10621e10627. Reed, M.J., Purohit, A., Woo, L.W., Newman, S.P., Potter, B.V., 2005. Steroid sulfatase: Kramer, W., Glombik, H., 2006. Bile acid reabsorption inhibitors (BARI): novel molecular biology, regulation, and inhibition. Endocr. Rev. 26 (2), 171e202. hypolipidemic drugs. Curr. Med. Chem. 13 (9), 997e1016. Schweigmann, H., Sanchez-Guijo, A., Ugele, B., Hartmann, K., Hartmann, M.F., Kramer, W., 2011. Transporters, Trojan horses and therapeutics: suitability of bile Bergmann, M., Pfarrer, C., Doring,€ B., Wudy, S.A., Petzinger, E., Geyer, J., acid and peptide transporters for drug delivery. Biol. Chem. 392 (1e2), 77e94. Grosser, G., 2014. Transport of the placental estriol precursor 16alpha-hydroxy- Labrie, F., 2015. All sex steroids are made intracellularly in peripheral tissues by the dehydroepiandrosterone sulfate (16alpha-OH-DHEAS) by stably transfected mechanisms of intracrinology after menopause. J. Ster Biochem. Mol. Biol. 145, OAT4-, SOAT-, and NTCP-HEK293 cells. J. Steroid Biochem. Mol. Biol. 143, 133e138. 259e265. Mueller, J.W., Gilligan, L.C., Idkowiak, J., Arlt, W., Foster, P.A., 2015. The regulation of Smellie, A., Teig, S.L., Towbin, P., 1995. Poling - promoting conformational variation. steroid action by sulfation and desulfation. Endocr. Rev. 36 (5), 526e563. J. Comp. Chem. 16, 171e187. Pasqualini, J.R., Chetrite, G.S., 2012. Hormonal enzymatic systems in normal and Sprague, P.W., Hoffmann, R., 1997. CATALYST pharmacophore models and their cancerous human breast: control, prognostic factors, and clinical applications. utility as queries for searching 3D databases. In: Van de Waterbeemd, H., Horm. Mol. Biol. Clin. Invest 9 (1), 25e63. Testa, B., Folkers, G. (Eds.), Computer-assisted Lead Finding and Optimization, Petzinger, E., Geyer, J., 2006. Drug transporters in . Schmiedeb. pp. p.225e240. Zürich: Helv Chim Acta. Arch. Pharmacol. 372 (6), 465e475. Strott, C.A., 2002. Sulfonation and molecular action. Endocr. Rev. 23 (5), 703e732.