The FASEB Journal express article 10.1096/fj.02-0026fje. Published online September 5, 2002.
A concerted, rational design of type 1 17β-hydroxysteroid dehydrogenase inhibitors: estradiol-adenosine hybrids with high affinity Wei Qiu, Robert L. Campbell,† Anne Gangloff, Philippe Dupuis, Roch P. Boivin, Martin R. Tremblay, Donald Poirier, and Sheng-Xiang Lin Oncology and Molecular Endocrinology Research Center, Laval University Medical Center (CHUL) and Laval University, Quebec, G1V 4G2, Canada †
Present address: Department of Biophysics and Biophysical Chemistry School of Medicine, Johns Hopkins University, Baltimore, MD 21205, USA Corresponding authors: Donald Poirier and Sheng-Xiang Lin, Oncology and Molecular Endocrinology Research Center, CHUL, 2705 boul. Laurier, Quebec, QC, G1V 4G2, Canada. Email:
[email protected] W. Qiu and R. L. Campbell contributed equally to the work. ABSTRACT Human estrogenic 17β-hydroxysteroid dehydrogenase (17β-HSD type 1) catalyzes the final step in the synthesis of active estrogens that stimulate the proliferation of breast cancer cells. Based on the initial premise to make use of the binding energies of both the substrate and cofactor sites, and molecular modeling starting from the enzyme structure, several estradiol-adenosine hybrids were designed and synthesized. Among these hybrids, EM-1745 with a linker of 8-CH2 groups is proved to be the best competitive inhibitor with a Ki of 3.0 ± 0.8 nM. The crystal structure of the EM-1745 enzyme complex at 1.6 Å provides evidence at atomic resolution of strong interactions between both the steroid and cofactor moieties and the enzyme molecule, as illustrated by a σAweighted 2Fo-Fc electron density map contoured at 3.0 σ. The substrate entry loop is further stabilized in this complex compared with previous complexes of the enzyme. These results confirm our initial strategy of combining studies of structural biology and enzyme mechanism in the inhibitor design, which may be applied to other steroidogenic enzymes involved in human diseases. Key words: hybrid inhibitor • inhibition kinetics • protein crystallography • molecular modeling • drug design
B
reast cancer is the number one cancer in women in Canada, with 19,500 estimated new cases in 2001. This cancer alone will account for ∼30% of all new cancer cases in women, according to the National Cancer Institute of Canada (1). It is also the most prevalent cancer in women in the United States, with 203,500 projected new cases in 2002, accounting for ∼31% of all new cancer cases in women according to the American Cancer Society (2).
Humans are the only species whose adrenal gland secretes a large quantity of steroid precursors that are converted to active estrogens and androgens in peripheral tissues (3, 4). Members of the 17β-hydroxysteroid dehydrogenase/17-ketosteroid reductase family (17β-HSD/KSR, simplified to 17β-HSD, below) catalyze the last step in the biosynthesis of all active estrogens and androgens. These sex hormones control the proliferation and function of sex steroid-sensitive normal and malignant tissues (5, 6). The estrogenic type 1 17β-HSD (17β-HSD1) catalyzes the formation of the most potent estrogen, 17β-estradiol (E2) and a less active estrogen, 5-androstene-3β, 17β-diol (5-diol). These estrogens stimulate the proliferation of breast cancer cells (7–10). The critical importance of 17β-HSD1 in breast cancer has recently been demonstrated more directly by its higher mRNA level in postmenopausal than premenopausal breast cancers (11). Due to the high affinity of the steroid for its cognate receptor, the blockage of estrogen binding to the estrogen receptor for the treatment of related diseases has been more successful than the finding of an inhibitor for estrogen synthesis (12). Nevertheless, the inhibition of the estrogen receptor alone by an antihormone will lead to the accumulation of active steroids that will decrease the efficiency of the treatment to block estrogen function. The search for an enzyme inhibitor for estrogen biosynthesis has thus become very important. The design of 17β-HSD1 inhibitors has been attempted during the past few decades, but unfortunately without significant progress to justify their use in estrogen-dependent breast cancer treatment (13). The studies have mainly focused on the modification of the steroid part. Affinity ligands were first synthesized to study the active site of the enzyme (14–20). Enzyme-generated inhibitors, such as 16-methylene E2 and acetylenic 16-secoestradiol, were developed thereafter (21–23). A series of 16α-(halogenoalkyl) E2 was also developed to inactivate 17β-HSD1 (24, 25). All the aforementioned inhibitors contain a reactive functional group, which inactivates the enzyme by forming a stable covalent bond with the enzyme (irreversible inhibitors). However, the following categories of compounds were described as reversible inhibitors of 17β−HSD1: 1) estrone (E1) derivatives with pyrazole or isoxazole fused to the 16,17 position on the D-ring (26), 2) E2 derivatives bearing a long 7α-undecanamide side chain (27, 28), and 3) an E2 derivative bearing a 6β-thiaheptanamide side chain (29). 16-Oxoestrone represents a special inhibitor because it was reversible at neutral pH 7.2 and irreversible at basic pH 8.5 (30). Several flavonoids were also reported to inhibit 17β-HSD1 (31–33). In the light of some previous experiments described hereafter, the affinity of a conjunction of two ligands could be much higher than that of each ligand with respect to the enzyme. Indeed, it was found that the affinity of the reaction intermediate Phe-AMP to yeast phenylalanyl-tRNA synthetase was much higher than that of either Phe or AMP (34). Moreover, the dissociation constant KPhe·AMP (4×10–9 M) was much closer to the product of KPhe and KAMP than those of either KPhe (30×10–6 M) or KAMP (1x10–3 M) alone. This is in agreement with theoretical approximations (34, 35). It was then pointed out that if the formation of the covalent bond between Phe and AMP is not followed by essential conformational changes in the enzyme as opposed to the two ligands bound separately (a likely case, due to the small size of the substrates), one should expect the following:
∆GPhe.AMP ≅ ∆GPhe + ∆GAMP and thus: KPhe.AMP ≅ KPhe.KAMP Following a similar hypothesis, synthesizing new types of hybrid inhibitors for 17β-HSD1 that would combine moieties from NAD(P)H and analogs of E2 was proposed. A recent report used the substrate-cofactor synthesized in situ during the enzyme reaction (36). In this case, the enzyme-catalyzed formation of NADP-dihydrofinasteride provided a potent bisubstrate analog inhibitor for human 5α-reductase, contributing to the treatment of benign prostate hyperplasia. Taking advantage of the potential affinity and selectivity from bisubstrate inhibitor, other successful attempts have also been reported recently in insulin receptor tyrosine kinase (37), serotonin N-acetyltransferase(38). Our preliminary binding study revealed that ADP and NADP have similar affinities to 17βHSD1, indicating that the ADP part of the cofactor plays the major role in the binding with the enzyme (Lin et al. unpublished). This observation is consistent with the fact that the nicotinamide ring has a weaker electron density than the rest of the cofactor, apparently due to a lack of direct interaction with the active site (39). Therefore, the NADP moiety in the above compounds may be substituted by adenosine. We thus attempted our modeling starting from an E2 core and an adenosine core from NADP. Modeling and chemical synthesis studies were carried out closely to rapidly approach a good lead compound. MATERIALS AND METHODS Modeling of hybrid inhibitors The E2 structure in the 17β-HSD1-E2 complex at 2.3 Å (40) and the adenosyl portion of the NADH modeled in the structure of the apo-enzyme (41) were originally used to create the first hybrid inhibitor. The nicotinamide ribose diphosphate portion of NADH was deleted and replaced by an alkylcarbonyl linker attached in the C16β-position of E2 and connected to the 5'-carbon of the adenosyl ribose (Fig. 1). The initial linker contained 10 methylene groups in the modeling. The torsion angles of the linker were adjusted to place the adenosine moiety in a position similar to that seen in the NADH and NADPH models. Keeping the positions of the enzyme, the E2 moiety and the adenine moiety fixed, the model was submitted to energy minimization. The linker length was modified by adding or deleting methylene groups so as to give linkers of length 5, 6, 7, 8, 9, and 12 methylene groups. These were then energy-minimized, again keeping the E2 and adenine rings fixed. The resulting structures are referred to as belonging to the “fixed” group. The fixed structures were then minimized with only the E2 moiety fixed. The resulting structures are referred to as the “E2-fixed” group. The E2-fixed structures were further minimized with no constraints on the inhibitors. The latter resulting structures are referred to as the “free” group. Finally, the E2-fixed structures were minimized in the complete absence of the enzyme for comparison purposes.
All minimizations and model building steps were carried out with Insight II (Accelrys). The protocol began with a minimization by steepest descents followed by a conjugate gradient minimization to a maximum derivative of 0.001 kcal/mol. In all cases, the enzyme structure was constrained to the initial structure. Molecular surfaces of the final models were calculated using the algorithm of Connolly (42), with a probe radius of 1.4 Å and a dot density of 24 dots/Å2. The minimal buried surface areas were calculated by determining the portion of the buried surface that was
