Selectivity of Osilodrostat as an Inhibitor of Human Steroidogenic Cytochromes P450

Abstract

Osilodrostat (LCI699) is a potent inhibitor of the human steroidogenic cytochromes P450 11β-hydroxylase (CYP11B1) and aldosterone synthase (CYP11B2). LCI699 is FDA-approved for the treatment of Cushing disease, which is characterized by chronic overproduction of cortisol. While phase II and III clinical studies have proven the clinical efficacy and tolerability of LCI699 for treating Cushing disease, few studies have attempted to fully assess the effects of LCI699 on adrenal steroidogenesis. To this end, we first comprehensively analyzed LCI699-mediated inhibition of steroid synthesis in the NCI-H295R human adrenocortical cancer cell line. We then studied LCI699 inhibition using HEK-293 or V79 cells stably expressing individual human steroidogenic P450 enzymes. Our studies using intact cells confirm the potent inhibition of CYP11B1 and CYP11B2 with negligible inhibition of 17-hydroxylase/17,20-lyase (CYP17A1) and 21-hydroxylase (CYP21A2). Furthermore, partial inhibition of the cholesterol side-chain cleavage enzyme (CYP11A1) was observed. To calculate the dissociation constant (K_d) of LCI699 with the adrenal mitochondrial P450 enzymes, we successfully incorporated P450s into lipid nanodiscs and carried out spectrophotometric equilibrium and competition binding assays. Our binding experiments confirm the high affinity of LCI699 to CYP11B1 and CYP11B2 (K_d ≈ 1 nM or less) and much weaker binding for CYP11A1 (K_d = 18.8 μM). Our results confirm the selectivity of LCI699 for CYP11B1 and CYP11B2 and demonstrate partial inhibition of CYP11A1 but not CYP17A1 and CYP21A2.

1. Introduction

Osilodrostat (LCI699) is the first FDA-approved cortisol synthesis inhibitor for treating Cushing disease (CD). Most commonly occurring in women and with an incidence of 0.2-5.0 per million people per year, overt Cushing syndrome (CS) is a life-threatening illness characterized by chronic overproduction of cortisol [1]. The most prevalent cause of CS is pituitary adenomas (CD), which autonomously secrete excessive adrenocorticotropic hormone (ACTH) [1]. In such cases, the first line of treatment involves the surgical removal of the tumors. While effective in skilled centers, patients may require pharmacologic treatment post-surgery, as the risk of recurrence is high (15-66%) [2]. Hypercortisolemia can cause hypertension, skin atrophy, glucose intolerance, depression, and other co-morbid conditions [3]; consequently, adequate control of cortisol production is essential to prevent morbidity and mortality in patients with CS. LCI699 inhibits steroid 11β-hydroxylase (CYP11B1), thereby preventing the conversion of 11-deoxycortisol to cortisol. In two phase III trials, LCI699 normalized 24-hour urine-free cortisol in significantly more CD patients than placebo and maintained disease control over many months, with improvement in disease-related co-morbidities [4–6].

Initially, however, LCI699 was developed by Novartis as an aldosterone synthase (CYP11B2) inhibitor for the treatment of hypertension, including primary aldosteronism, the most common form of secondary hypertension. CYP11B2 catalyzes the final three steps of aldosterone synthesis from 11-deoxycorticosterone (DOC). While in-vitro studies established some selectivity of LCI699 for CYP11B2 over CYP11B1, phase II clinical studies revealed that LCI699 led to the accumulation of 11-deoxycortisol [7] and a slight reduction in ACTH-stimulated serum cortisol [8]. Given these findings, LCI699 was then successfully repurposed as a CYP11B1 inhibitor.

While in-vitro studies have shown that LC699 is a potent CYP11B1 and CYP11B2 inhibitor, with reported IC₅₀ values of 2.5 nM and 0.7 nM, respectively, few studies have established the global effects of LCI699 on adrenal steroidogenesis [9, 10]. A recent study compared the steroid profiles of patients with ACTH-dependent CS treated with either LCI699 or metyrapone, a drug used off-label in the United States (and approved by the European Medicines Agency) for treating CS [11]. Despite the markedly greater potency of LCI699 over metyrapone in-vitro as a CYP11B1 inhibitor, this study found that LCI699 caused less accumulation of 11-deoxycortisol and androgens than metyrapone, despite similar normalization of urine-free cortisol. Because CYP11B1 inhibition allows accumulation of cortisol intermediates, particularly 17-hydroxyprogesterone, which can be converted to androgens, the authors suggested that LCI699 also inhibits CYP17A1 (17-hydroxylase/17,20-lyase) and CYP21A2 (21-hydroxylase), but the data could not exclude simultaneous inhibition of CYP11A1 (cholesterol side-chain cleavage enzyme). Given these findings, we comprehensively characterized the effects of LCI699 on human adrenal steroidogenesis and on the adrenal cytochromes P450, CYP11A1, CYP11B1, CYP11B2, CYP17A1, and CYP21A2 (Fig. 1). We examined the effects of LCI699 on steroid synthesis in the NCI-H295R human adrenocortical cancer cell line, in V79 or HEK-293 cells stably expressing individual steroidogenic P450 enzymes, and with purified recombinant steroidogenic P450 enzymes.

Fig. 1. Overview of Studied Adrenal Steroidogenic Pathways.

Different colors are used to highlight the involvement of individual enzymes in multiple reactions. Abbreviations: B, corticosterone; 18OHB, 18-hydroxycorticosterone; DHEA, dehydroepiandrosterone; 17Preg, 17-hydroxypregnenolone; 17OHP, 17-hydroxyprogesterone; A4, androstenedione; 3βHSD2, 3β-hydroxysteroid dehydrogenase type 2; other enzyme abbreviations as in the text.

2. Materials and methods

2.1. Reagents and Chemicals

LCI699 was obtained as gifts from Novartis Pharmaceuticals and from Prof. Emily Scott, University of Michigan (obtained from Selleckchem). The sources of biochemicals [12–14] and reagents for mass spectrometry, including steroid calibrators and internal standards, [15, 16] are as described previously or as specified in subsections 2.2–2.6.

2.2. Cellular Inhibition Assay

After treatment with 10 μM forskolin in medium with 1% serum for 72 h, NCI-H295R cells were seeded in 12 well plates at 80% confluency and experimentally treated in serum-free medium with 0-1000 nM LCI699, 1000 nM ketoconazole, or 1000 nM metyrapone (final ethanol content <1% v/v). The medium was removed at 24 h, and steroids were assayed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) as described [16]. The Δ⁵-steroids pregnenolone, 17-hydroxypregnenolone, and dehydroepiandrosterone we measured in parallel [14]. These steroids combined amounted to only <5% the amount of 11-deoxycortisol + cortisol; consequently, no further analysis was performed with these data.

HEK-293 cell lines stably expressing CYP17A1 or CYP21A2 and V79 cells stably expressing CYP11A1, CYP11B1, or CYP11B2 were generated as described for AKR1C9 [17]. Briefly, cells were transfected in 10 cm dishes with 1.5 μg of pcDNA3 plasmid driving expression of each enzyme using FuGENE6. Cells were split 1:25 or thinner 48 h after transfection and then treated with medium containing G418 at 1.6 mg/mL to select for clones with stable plasmid integration. Individual clones were picked with a sterile pipette, expanded in medium with 0.5 mg/mL G418, and tested for enzyme activity. Clones with high activity were subsequently used for inhibition experiments.

V79 or HEK-293 cell lines were seeded in 12 well plates at 80% confluency and incubated in a medium containing 1 μM substrate and 0-1000 nM LCI699 (final ethanol content <1% v/v). Aliquots of medium were removed after 5 hrs, and steroids were extracted and analyzed by LC-MS/MS. The substrates used were 22R-hydroxycholesterol for CYP11A1, progesterone for CYP17A1, 17-hydroxyprogesterone for CYP21A2, 11-deoxycortisol for CYP11B1, and corticosterone rather than DOC for CYP11B2, to limit the number of intermediate steroids and facilitate data analysis.

The half-maximal inhibitory concentrations (IC₅₀) were calculated using GraphPad Prism software version 9.0.0 for Mac OS X. For the NCI-H295R cells, the measured products were plotted as a function of inhibitor concentration. In the case of the stable cell lines, the percent inhibition was plotted as a function of inhibitor concentration. Data were fit to the dose-response inhibitor (three-parameter) equation.

2.3. Protein Expression in Escherichia coli and Purification

CYP11B2 and CYP11B1 were expressed as previously reported [18]. After expression, the pelleted cells were resuspended in cold lysis buffer (50 mM potassium phosphate buffer (pH 8.0), 20% (v/v) glycerol, 0.5 M sodium chloride, protease inhibitor cocktail (Bimake), 0.1 mM DTT) and passed once through a French press system at 16,000 PSI. Proteins were solubilized through the slow addition of 1.5% (w/v) sodium cholate and 1.5% (v/v) Tween 20, and the supernatant was collected after 20 minutes of centrifugation at 35,000 RPM at 4 °C. The lysate was incubated overnight with Ni-NTA resin (Qiagen) previously equilibrated with 3 column volumes (CV) of equilibration buffer (50 mM potassium phosphate buffer (pH 8.0), 20% (v/v) glycerol, 0.5 M sodium chloride, 1% (w/v) sodium cholate, 1% (v/v) Tween 20, 0.1 mM DTT). The resin was loaded into a column and washed with 5 CV of washing buffer 1 (50 mM potassium phosphate buffer (pH 8.0), 20% (v/v) glycerol, 0.5 M sodium chloride, 1% (w/v) sodium cholate, 1% (v/v) Tween 20, 10 mM imidazole, 0.1 mM DTT) and then with 5 CV of washing buffer 2 (50 mM potassium phosphate buffer (pH 8.0), 20% (v/v) glycerol, 1% (w/v) sodium cholate, 1% (v/v) Tween 20, 20 mM imidazole, 0.1 mM DTT, 0.1 mM ATP). CYP11B proteins were eluted with 5 CV of elution buffer (50 mM potassium phosphate buffer (pH 7.4), 20% (v/v) glycerol, 0.5% (w/v) sodium cholate, 0.05% (v/v) Tween 20, 250 mM imidazole, 0.1 mM TCEP). Imidazole was removed by buffer exchange using PD-10 columns (GE Healthcare, 8.3 mL bed volume). The proteins were further purified with His-Select resin (Millipore) by using the same buffers. CYP11A1, adrenodoxin, adrenodoxin reductase, and membrane-scaffold protein 1D1 (MSP1D1) were expressed and purified as previously reported [19–22].

2.4. P450-nanodiscs Preparation

The procedure was adapted from a previously described method [21]. P450-nanodiscs were prepared by mixing the MSP1D1, cholate-solubilized lipids, and P450 enzymes in a 1:64:0.05 ratio in disc buffer (100 mM potassium phosphate (pH 7.4), 50 mM NaCl, 20 mM sodium cholate buffer), using 16:0-18:1 PC (POPC) and 16:0-18:1 PS (POPS) lipids at a 4:1 (POPC:POPS) ratio. The resulting reaction mixture was incubated on a shaker at 4°C. After 1 h of incubation, 0.25 g/mL of Bio-Beads^™ (Bio-Rad) was added to slowly remove sodium cholate and to initiate nanodisc assembly. After another 1 h, an additional 0.50 g/mL of Bio-Beads^™ was added, for a total of 0.75 g/mL. The sample was incubated in a tube rotator overnight at 10 RPM. The sample was aspirated from the Bio-Beads and centrifuged at 13,000 x g for 5 minutes before purification through size-exclusion chromatography on a Superdex 200 10/300 GL column using an Akta model UPC-900 FPLC and 50 mM potassium phosphate buffer (pH 7.4). P450 and MSP1D1 incorporation into the nanodiscs was assessed through SDS gel electrophoresis and SEC analysis. Purified nanodiscs were stored at 4°C and used within 2-3 days.

2.5. Reconstituted Enzyme Assays

Assays contained 0.16 μM CYP11B2, 9.6 μM adrenodoxin, and 0.16 μM adrenodoxin reductase in 0.2 mL 50 mM potassium phosphate buffer (pH 7.4). The reaction mixture was pre-incubated at 37°C for 5 min, followed by the addition of either 10 μM LCI699 or 0.5 μM DOC dissolved in ethanol (final ethanol content <1% v/v) and pre-incubated at 37°C for 8 min. In some experiments, either inhibitor or substrate was next added to have both compounds present but added in a different order. NADPH (1 mM) was added, and the final incubation was continued at 37°C for 10 min to allow the synthesis of quantifiable 18-hydroxylase and 18-oxidase products. The reaction was stopped by the addition of 20 μL of 0.1 M HCl, followed by vortexing. Steroids were extracted and analyzed using LC-MS/MS [16].

2.6. Ligand Binding Assays.

Substrate and inhibitor binding affinities to CYP11A1, CYP11B1, and CYP11B2 enzymes were determined by monitoring UV-visible (350-700 nm) spectral changes. P450-nanodiscs were diluted to 0.16 μM with 50 mM potassium phosphate (pH 7.4) in a 1-cm quartz cuvette. After a 5 min incubation at 25°C, increasing amounts of ligand stocks in ethanol were added (final ethanol content ≤1% v/v). Difference spectra were obtained by subtracting the absorbance signal of individual titration concentrations from the spectra of P450-nanodiscs in the absence of titrant. The absorbance differences (at the wavelength maximum and minimum) were then plotted as a function of the added ligand concentrations. Dissociation constants (K_d) were calculated using GraphPad Prism software version 9.0.0 for Mac OS X. For substrate binding analysis, data were fitted to the one-site-binding equation. For LCI699 binding to CYP11B1 and CYP11B2, the data obtained required fitting to the tight-binding Morrison equation (1) [23].

$$\Delta A=\Delta A_{max}\frac{\left(\left[E\right]+\left[L\right]+K_d\right)-\sqrt{{([E]+[L]+K_d)}^2-4[E][L]}}{2[E]}$$ (1)

The binding affinities of LCI699 for CYP11B1 and CYP11B2 were also calculated using a competition binding assay with DOC. P450-nanodiscs were pre-incubated with DOC for 8 minutes, after which increasing amounts of LCI699 were added (final ethanol content < 4% v/v). The absorbance differences (at the wavelength maximum and minimum) were then plotted against the log of the added LCI699 concentrations. The resulting curves were analyzed using a dose-response inhibitor (four-parameter) equation in GraphPad Prism. The dissociation constant for LCI699 (K_d) was calculated from the IC₅₀ obtained using the Cheng-Prusoff equation (2), where K_D is the binding constant for DOC and “IC₅₀” refers to the concentration of LCI699 that achieves 50% DOC displacement. In this case, the equation is used with binding parameters rather than enzymatic activity/inhibition constants, using the same theoretical basis of active-site competition.

$$K_d=\frac{{\text{IC}}_{50}}{1+\frac{[\text{DOC}]}{K_D}}$$ (2)

3. Results

3.1. Effect of LCI699 on Steroidogenesis in NCI-H295R cells

To assess the global effects of LCI699 on adrenal steroidogenesis, NCI-H295R cells were utilized [24]. To stimulate steroid production, the cells were pre-treated with 10 μM forskolin prior to incubation with fresh serum-free medium containing various concentrations of LCI699, 1000 nM ketoconazole, or 1000 nM metyrapone. Conditioned medium aliquots were collected 24 h post-treatment and analyzed via LC-MS/MS.

LCI699 inhibited cortisol and aldosterone production in a dose-dependent manner to 13 ± 4% and 2 ± 3% of baseline with IC₅₀ values of 8.4 ± 1.5 nM and 3.2 ± 0.9 nM, respectively (Fig. 2). Furthermore, corticosterone fell to 1 ± 2% of the baseline while androstenedione remained unchanged (Fig. 2). These data are consistent with potent inhibition of CYP11B1 and CYP11B2 with little “overflow” of 21-carbon precursors into the adrenal 19-carbon steroidogenesis pathway via CYP17A1. In contrast, 11-deoxycortisol but not DOC appeared to increase at higher concentrations of LCI699, but these changes did not reach statistical significance (Fig. 2). Unlike the human adrenal zona fasciculata, however, NCI-H295R cells are relatively deficient in CYP11B1 expression similar to most adrenocortical carcinomas [25], such that the cortisol content of the conditioned medium is roughly half that of 11-deoxycortisol at baseline. The sum of 11-deoxycortisol + cortisol remained relatively constant across all LCI699 concentrations, yet the abundance of 11-deoxycortisol relative to cortisol limited our ability to detect rises in 11-deoxycortisol. Similarly, aldosterone concentrations were an order of magnitude lower than DOC, which minimized DOC accumulation from CYP11B2 inhibition; furthermore, forskolin treatment does not induce maximal aldosterone production [26].

Fig. 2. Inhibition of steroid production in NCI-H295R cells.

After forskolin stimulation, H295R cells were treated with 0–1000 nM LCI699, 1000 nM ketoconazole (Keto.), 1000 nM metyrapone (Mety.), or vehicle in a fresh serum-free medium. Steroid levels measured by LC-MS/MS are given as the mean ± standard error of the mean (SEM) of three individual experiments.

3.2. V79 and HEK-293 Cell Lines Stably Expressing Individual Enzymes

To corroborate the results obtained from NCI-H295R cells, HEK-293 and V79 Chinese hamster cells stably expressing individual steroidogenic P450 enzymes were generated. V79 cells have been employed to study mitochondrial P450 enzymes, as they lack endogenous P450 activity and do not require co-transfection with redox partners (Adx and its reductase) [15,16]. HEK-293 cells, on the other hand, have been widely employed to study microsomal cytochrome P450-mediated drug and steroid metabolism [27]. In V79 cells expressing CYP11B1, LCI699 inhibited the conversion of 11-deoxycortisol to cortisol with an IC₅₀ of 9.5 ± 0.5 nM (Fig. 3). In comparison, LCI699 inhibited the conversion of corticosterone to aldosterone with an IC₅₀ of 0.28 ± 0.06 nM in V79 cells expressing CYP11B2. In V79 cells expressing CYP11A1, LCI699 showed partial (<25%) inhibition of pregnenolone formation at 1000 nM. In HEK-293 cells stably expressing CYP17A1 or CYP21A2, LCI699 showed negligible activity inhibition (<1%) at concentrations up to 1000 nM. These data confirm that LCI699 potently inhibits CYP11B1 and CYP11B2 with negligible inhibition of CYP17A1 and CYP21A2 yet partial inhibition of CYP11A1 at high concentrations.

Fig. 3. Inhibition of CYP11A1, CYP11B1, and CYP11B2 activity in V79 cells.

Cells were plated and incubated in a medium containing 1 μM 22-R-hydroxycholesterol, 11-deoxycortisol, and corticosterone for CYP11A1(♦), CYP11B1(•), and CYP11B2(■) expressing cells, respectively, and 0–1000 nM LCI699. Data were fit to the dose-response inhibitor (three-parameter) equation to determine the half-maximal inhibitory concentration (IC₅₀) of 9.5 ± 0.5 nM and 0.28 ± 0.06 nM for CYP11B1 and CYP11B2, respectively. Data represent the mean ± standard error of the mean (SEM) of three individual experiments.

3.3. Ligand Binding Titrations in Nanodiscs

To directly determine the equilibrium binding constants (K_d) of LCI699 with adrenal mitochondrial P450 enzymes, we performed spectroscopic studies with P450s incorporated into lipid nanodiscs. LCI699 binds tightly to purified CYP11B2 and CYP11B1 in nanodiscs and affords type II spectral changes (Fig. 4). Data from the optical titration were fit to the tight-binding model in equation (1) and yielded K_d values of 0.6 ± 2.0 nM and 0.3 ± 0.9 nM for CYP11B1 and CYP11B2, respectively. The large relative errors in these measurements are due to the sharpness of the inflection point under conditions where K_d << [enzyme], and difference spectra using lower enzyme concentrations would be too weak to measure. While these values should be interpreted with some caution, the data are consistent with our cell-culture results. In contrast, CYP11A1 showed much weaker binding to LCI699 with a K_d of 18.8 ± 4.5 μM derived from a robust titration curve (Fig. 4).

Fig. 4. Equilibrium binding titration of LCI699 to P450-incorporated nanodiscs.

(A) CYP11A1(♦), (B) CYP11B1(●), and (C) CYP11B2(■) nanodiscs at 0.16 μM were equilibrated with increasing concentrations of LCI699 in 50 mM potassium phosphate buffer (pH 7.4). (D) Calculated equilibrium dissociation constants (K_d). The inserted image in each panel shows the difference spectra between samples with and without LCI699. Data represent the mean ± standard deviation of three individual experiments.

3.4. Competitive Binding Experiments

Given the limitations of the direct binding experiments, we also determined the K_d of LCI699 for CYP11B1 and CYP11B2 using competition experiments with our P450-nanodisc system. In this strategy, we relied on the distinct spectral shifts induced upon binding of substrate (DOC) and LCI699 inhibitor to the P450s, which allowed us to monitor displacement of substrate at various LCI699 concentrations. To this end, both CYP11B1 and CYP11B2 were incubated with the common substrate DOC (K_D determined as 420 and 270 nM with CYP11B1 and CYP11B2, respectively) prior to the addition of LCI699. During titration with inhibitor, the spectrum gradually shifted from Type I to Type II (Fig. 5A). LCI699 completely outcompeted DOC (25 μM) bound to CYP11B1 in nanodiscs and yielded an IC₅₀ of 90 ± 3 nM (Fig. 5B). Using the Cheng-Prussoff equation (2), these data afforded a K_d of 1.5 ± 0.1 nM, with a similar value yet much lower relative error than the direct binding experiments. Thus, the two binding studies consistently yield K_d values for LCI699 binding to CYP11B1 of close to 1 nM. In contrast, concentrations up to 10 μM of LCI669 could not fully shift the spectrum of CYP11B2 in nanodiscs from Type I to Type II (Fig. 6A, 6B), suggesting that LCI699 cannot fully outcompete the previously bound DOC (0.5 μM). The addition of Adx further limited DOC displacement from CYP11B2 by LCI699 (Fig. 6B).

Fig. 5. CYP11B1-nanodiscs competitive binding assay.

(A) UV-visible spectral changes of 0.16μM CYP11B1-nanodiscs in 50 mM potassium phosphate buffer (pH 7.4). The addition of 25 μM DOC causes the absorption maximum of the Soret band to shift from 417 nm (dark purple) to 390 nm (blue). Subsequently, with increasing concentrations of LCI699, the maximum of the Soret band shifts to 419 nm (orange). (B) Dose-response curve of CYP11B1-nanodisc. Increasing concentrations of LCI699 can outcompete the bound DOC. Data were fit to the dose-response inhibitor (four-parameter) equation to determine the half-maximal inhibitory concentration (IC₅₀) of 90 ± 3 nM. Data represent the mean ± standard deviation of three individual experiments.

Fig. 6. CYP11B2 competition binding assay.

(A) UV-visible spectral changes of 0.16 μM CYP11B2-nanodiscs in 50 mM potassium phosphate buffer (pH 7.4). The addition of 0.5 μM DOC causes the absorption maximum of the Soret band to shift from 417 nm (dark purple) to 390 nm (blue). Subsequently, increasing the concentration of LCI699 up to 10 μM shifts the maximum of the Soret band to 414 nm (cyan). (B) Dose-response curve of CYP11B1-nanodisc with (▲) and without adrenodoxin (■). LCI699 cannot outcompete the bound DOC in either condition. Data represent the mean ± standard deviation of three individual experiments.

3.5. CYP11B2 Inhibition studies

To address the unexpected results obtained for LCI699 displacement of DOC bound to CYP11B2, we assess LCI699 inhibition of CYP11B2-catalyzed DOC turnover. Pre-incubation of CYP11B2 nanodiscs with LCI699 (10 μM) completely abrogated aldosterone production, with only traces of the initial product corticosterone formed from 0.5 μM DOC. Conversely, pre-incubation with the same concentrations of DOC followed by sequential addition of LCI699 and then NADPH yielded similar amounts of aldosterone as in the absence of LCI699, yet much fewer total products, when including corticosterone and 18-hydroxycorticosterone intermediates (Fig. 7). When CYP11B2 was pre-incubated with 0.5 μM corticosterone, addition of 10 μM LCI699 did not completely inhibit 18-hydroxycorticosterone and aldosterone formation (1% the yield without LCI699, not shown). These data are consistent with a population of CYP11B2•DOC complexes that does not dissociate during the timeframe of the experiment and affords aldosterone despite subsequent exposure to LCI699.

Fig. 7. LCI699-mediated inhibition of CY11B2-nanodiscs.

Corticosterone (A), 18-hydroxycorticosterone (B), and aldosterone (C) were quantified with LC-MS/MS. CYP11B2-nanodiscs (0.16 μM) were pre-incubated with either DOC (0.5 μM) or LCI699 (10 μM) in 50 mM potassium phosphate buffer (pH 7.4) before the addition of either NADPH (first bar), LCI699 or DOC in the order specified, then NADPH (second and third bars), or no NADPH control (fourth bar). Data represent the mean ± standard deviation of three individual experiments.

4. Discussion

Using three different model systems, we confirmed that LCI699 is a potent inhibitor of CYP11B1 and CYP11B2, with IC₅₀ fo 0.1-10 nM using V79 or NCI-H295R cells. Using purified recombinant enzymes, spectral binding studies required quadratic fitting and substrate competition experiments to calculate K_d values (~1 nM) due to such tight binding. In contrast, we found no inhibition of CYP17A1 or CYP21A2 at LCI699 concentrations up to 1000 nM. We also detected partial inhibition of CYP11A1 at 1000 nM LCI699 and spectroscopic evidence of binding to the heme iron of CYP11A1 with a K_d of 18.8 μM. In all systems, LCI699 was about a 3-fold more potent inhibitor of CYP11B2 than of CYP11B1. The internally consistent data from several different systems affirm the selectivity of LCI699 for the adrenal mitochondrial P450s in the order CYP11B2>CYP11B1>>CYP11A1.

While ligand binding and activity of microsomal (type 2) P450s have been extensively studied using nanodiscs [28, 29], we report the first data obtained with mitochondrial (type 1) P450 enzymes in nanodiscs. With microsomal P450 systems, the redox partner POR must be also incorporated in the nanodisc on the same side as the P450, which might account for the low specific activity often seen relative to liposomal reconstituted systems. In contrast, mitochondrial P450s use adrenodoxin, a soluble iron-sulfur protein, as the direct electron transfer protein. Hence, every mitochondrial P450 in a population of nanodiscs will have full access to its redox partner adrenodoxin from the solution, which allows maximal reconstitution of activity. We have found that both substrate binding and turnover are as efficient or better using CYP11B1 and CYP11B2 in nanodiscs as in liposomal reconstituted assays (data not shown).

An unexpected result was that the addition of LCI699 could neither completely displace CYP11B2-bound DOC nor block its metabolism to aldosterone (Fig. 6, 7), despite extremely tight LCI699 binding when added alone (Fig. 4). A possible explanation for this behaviour, which is consistent with the data in Fig. 7, is two populations of CYP11B2 molecules, one of which has a very slow dissociation rate of DOC. This finding is consistent with the high processivity of CYP11B2 with DOC as the substrate, which maximizes aldosterone production [30]. These data might explain the incomplete reduction of aldosterone observed in clinical trials of LCI699 [5, 31], despite its high potency as a CYP11B2 inhibitor. Our data suggest that LCI699 exposure would have to remain constantly high relative to intra-adrenal DOC to fully inhibit aldosterone production during clinical use. In contrast, LCI699 completely displaced DOC from CYP11B1, which is consistent with the efficacy observed in the treatment of CD, in which it rapidly reduced mean 24-hour urine-free cortisol. On the other hand, aldosterone is often low in CS due to the mineralocorticoid activity of excess cortisol, so the clinical significance of these findings remains ambiguous.

Furthermore, we observed consistent inhibition of and binding of LCI699 to CYP11A1, albeit incomplete and weak relative to its effects on CYP11B1 and CYP11B2. Nevertheless, this finding might explain the results of a retrospective study, which found less cortisol precursor accumulation in patients with ACTH-dependent CS during LCI699 treatment than for those treated with metyrapone [11]. This study also found higher substrate/product ratios in patients treated with LCI699 than for metyrapone, suggesting partial CYP21A2 (17OHP/11-deoxycortisol) and CYP17A1 inhibition (17OHP/androstenedione); however, the authors recognized that reduced substrate influx could also explain their data, and 21-deoxycortisol was not measured. Our data are consistent with the latter explanation that LCI699 partially inhibits CYP11A1 but not CYP21A2 or CYP17A1, and our data are consistent with the reduction of total steroid production in HAC15 cells treated with LCI699 [10]. The limited precursor accumulation in experiments with LCI699 treatment of NCI-H295R cell cultures should be interpreted cautiously (Fig. 2), because these cells are relatively deficient in CYP11B1 activity compared to the normal adrenal cortex and produce much less aldosterone than cortisol.

The limitations of our studies are similar to those for any pharmacologic experiments using model systems. The calculated inhibition and binding constants observed, while consistent in rank order for the three mitochondrial P450 enzymes, varied among the systems used. Differences in accessible substrate concentrations, enzyme expression, and cellular penetration/efflux of the drug might explain some of this variance. In addition, we did not study the influence of adrenodoxin in detail with purified enzymes, and we neither determined nor varied the abundance of adrenodoxin in the V79 or NCI-H295R cells. The data in Fig. 7 suggest that adrenodoxin enhances substrate binding for CYP11B2, as previously shown [30], and another study demonstrated that adrenodoxin enhances, in a concentration-dependent manner, the catalytic efficiency, substrate affinity, and potency of LCI699 as an inhibitor of purified CYP11B1 [32]. More detailed biochemical studies of LCI699 in-vitro and analyses of steroid changes in treated patients will enhance our understanding of the drug’s pharmacology and guide its use in patients with CS.

Abbreviations

Adx

adrenodoxin

CYP or P450

Cytochrome P450

CYP11A1

cholesterol side-chain cleavage enzyme

CYP11B1

11β-hydroxylase

CYP11B2

aldosterone synthase

CYP17A1

17-hydroxylase/17,20-lyase

CYP21A2

21-hydroxylase

DOC

11-deoxycorticosterone

FPLC

fast protein liquid chromatography

HPLC

high-performance liquid chromatography

IPTG

isopropyl-β-D-thiogalactopyranoside

LC-MS/MS

liquid chromatography-tandem mass spectrometry

NADPH

nicotinamide adenine dinucleotide phosphate (reduced form)

16:0-18:1 PC (POPC)

1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine

16:0-18:1 PS (POPS)

1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine

UV

ultraviolet light