Test compounds: 1. Pumpkin seed oil of Uromedic® pumpkin seeds (Material 3000000293, Omega Pharma batch: 511062, vendor/batch: 603827/283484), one of the active substances in “GRANUFINK® Prosta” (Omega Pharma Deutschland GmbH). 2. Δ7-sterols isolated from Uromedic® pumpkin seed oil; 3. Pumpkin seed soft extract (DER 15-25:1) from Uromedic® pumpkin seeds; extraction solvent ethanol 92% (m/m) (Material 3000000299, Omega Pharma batch: 531057, vendor/batch: 600516/15000533); active substance of GRANUFINK® Prosta forte 500 mg (Omega Pharma Deutschland GmbH), supplying approximately 30 mg/day Δ7-sterols [14] (Supplementary Table S1).

The test compounds were dissolved and stored in DMSO (dimethyl sulfoxide) at 4˚C.

Commercial hexane extracted saw palmetto fruit extract (Serenoa repens W. Bartram; DER 7-11:1) served as positive control (5α-reductase―inhibition). Finasteride (C₂₃H₃₆N₂O₂, 5α-reductase―inhibitor) was used to quantify 5α-reductase inhibition, and DHT to quantify AR binding.

AR (recombinant rat protein, ligand binding domain, code: A15675) was from ThermoFisher Scientific (Braunschweig, Germany). Dihydrotestosterone (DHT) (code: A83809), testosterone (code: T1500) and finasteride (code: F1293) were from Sigma (Taufkirchen, Germany). [³H]-Dihydrotestosterone ([³H]-DHT, code: NET 453), [1,2,6,7-³H(N)]-testosterone, code: NET370) and Ysi Copper His-Tag SPA Beads (code: RPNQ0096) were from PerkinElmer (Rodgau, Germany). Other chemicals were from Sigma or VWR.

Sprague Dawley rats (Elevage Janvier, Le Genest, Saint Isle, France) were sacrificed by decapitation. Liver was dissected on ice, frozen in liquid nitrogen and stored at −80˚C. For the experiments at pH 7.0 (5α-R1), rat liver microsomes were prepared as described [20] . For the experiments at pH 5.0 (5α-R2), a crude rat liver enzyme preparation was used as described [21] .

The enzyme reactions were mainly carried out as described [21] . In a total volume of 90 μl radioactive (³H-testosterone; 0.5 μCi) and non-radioactive (9.5 μM) testosterone, 0.5 mM beta-NADPH and liver microsomes/crude liver enzyme preparation were incubated in assay buffer (0.1 M Tris-citrate, pH 5.0 or 7.0) in a shaking water bath at 37˚C for 20 or 60 min. The assay was terminated by the addition of 10 μl 2.8 M NaOH. Educt (³H-testosterone) and products (³H-DHT, ³H-alpha-adiol and other metabolites) were separated by HPLC from each other using an adapted [22] and validated method, and using a Beckman System Gold HPLC (Beckman Instruments, San Ramon, CA, USA) fitted with an Onyx monolitic C18 (Phenomenex Ltd. Deutschland, Aschaffenburg, Germany) reversed phase column (100 × 3 mm). Chromatography was performed isocratically for 15 min at a flow rate of 0.5 ml/min with 0.2% NH₄OH in 65% methanol as mobile phase, separating ³H-testosterone and ³H-DHT. Radioactivity in collected fractions was determined in a scintillation counter (WinSpektral, Wallac, Finland). Data typically represent results gained by two experiments performed in duplicate (see results section).

Data evaluation: The inhibition by 10 µM finasteride (3.7254 µg/ml) was considered 100% inhibition, the inhibitory activity of the vehicle/solvent 0% inhibition. The inhibition of the test compounds was calculated relative to that of finasteride. IC₅₀ values were determined by non-linear regression (algorithm “sigmoidal dose-response”; GraphPadPrism, San Diego, USA).

The competition of each test compound (or DHT) with [³H]-DHT for binding to the rat recombinant AR was measured by detecting [³H]-DHT. The basic assay conditions were taken from Freyberger and Ahr [23] and adapted to the scintillation proximity assay (SPA) format, in which beads emit light when a radio-labeled molecule is in proximity. Incubation was in assay buffer (50 mM Tris-HCl pH 7.5, 800 mM NaCl₂, 10 mg/ml human γ-globulin, 0.05% bovine serum albumine) at room temperature for 90 min. Then Ysi Copper His-Tag SPA beads were added and the incubation continued for another 60 min. The radioactivity-induced light was determined by a microplate reader (Microbeta, Wallac, Finland). For each test compound, two measurement series were done, each in duplicate.

Data evaluation: In order to exclude binding to non-specific sites, incubations using the radioactive ligand [³H]-DHT and unlabeled DHT as the natural ligand of the AR were run simultaneously at different concentrations. Excess unlabeled DHT occupied the high-affinity specific sites and blocked the binding of [³H]-DHT to the AR. As a result, the radio-labeled ligand only bound to the non-specific sites. Such radioactive binding in the presence of excess unlabeled DHT was considered non-specific binding.

The amount of [³H]-DHT bound in the absence of unlabeled DHT was referred to as total binding. Specific binding was calculated as:

Total (radioactive) binding − Non-specific (radioactive) binding = Specific binding

Thus, the data of the reference compound DHT are presented as total bound radioactivity (in counts per minute, cpm). Test compound data are presented as specific radioligand binding to the receptor. The specific inhibition in % was calculated as (% specific radioligand binding mean ? 100) × (−1). IC₅₀ values were determined by non-linear regression (algorithm “sigmoidal dose-response”; Graph Pad Prism, San Diego, USA).

Rat liver microsomes were shown to obtain higher levels of 5-alpha-reductase compared to rat prostate microsomes [20] . Therefore, rat liver microsomes were used as source of 5α-R1. For 5α-R2, a crude liver preparation was found to be more active. 5α-R1 exhibits a broad pH optimum (between pH 6.0 - 8.5), whereas 5α-R2 shows a narrow acidic pH optimum (pH 5 - 5.5) [4] . To distinguish 5α-R1 from 5α-R2, the assays for 5α-R1 were performed at pH 7.0, and those for 5α-R2 at pH 5.0 [21] . Experiments at pH 5.0 led to a very low, almost negligible signal from the rat liver microsomal preparation (source of 5α-R1; data not shown).

At pH 7.0 the reference compound finasteride inhibited the activity of 5α-R1 derived from rat liver microsomes with an IC₅₀ of 48 nM (Figure 1). This compares to a published IC₅₀ of 21 nM [20] , demonstrating test system suitability. A commercial hexanic extract of Serenoa repens [5] was able to inhibit 5α-R1 with an IC₅₀ = 4.9 mg/ml and 5α-R2 with an IC₅₀ = 1.8 mg/ml.

The inhibitory potential of the test compounds on 5α-R1 was tested using our validated assay at pH 7, favoring 5α-R1 activity. Table 1 shows that pumpkin seed oil (5 mg/ml), pumpkin seed soft extract (5 mg/ml) and Δ7-sterols (0.5 mg/ml) had similar inhibitory capacities, between 70% - 95% compared to 10 µM finasteride. The results for Δ7-sterols showed a dose dependency. The calculated IC₅₀ values are indicated.

The inhibitory potential of the test compounds on 5α-R2 was tested using our validated assay at pH 5, favouring 5α-R2 activity. Table 2 shows that pumpkin seed oil (5 mg/ml), pumpkin seed soft extract (5 mg/ml) and Δ7-sterols (0.5 mg/ml) had similar inhibitory capacities, between 25% - 50% compared to 10 µM finasteride. The results for Δ7-sterols showed a dose dependency. The calculated IC₅₀ values are indicated.

a-c: see Table 2.

a: Inhibition relative to 10 µM (3.7254 µg/ml) finasteride (100% inhibition); activity with vehicle DMSO was considered 0% inhibition. b: Both experiments in duplicate; means of duplicate measurements shown; c: IC₅₀ values: concentration of test compound causing 50% inhibition. d: This value was judged as outlier and not taken into consideration.

DHT is the natural ligand of the AR. DHT competed with [³H]-DHT for binding to the AR with an IC₅₀ of 8.5 nM (Figure 2). This is in agreement with an IC₅₀ of 4.2 nM described in literature [23] , demonstrating test system suitability.

Overlaid mean specific binding data of pumpkin seed soft extract, pumpkin seed oil and Δ7-sterols are depicted graphically (Figure 3(a) and Figure 3(b)); numeric data is shown in Supplementary Tables S2-S4. All test compounds show typical concentration dependent binding. Strongest specific AR binding was observed for Δ7-sterols (mean IC₅₀ = 0.2 mg/ml, corresponding to 0.48 mmol/l) and weaker binding for pumpkin seed oil (mean IC₅₀ = 0.4 mg/ml) and pumpkin seed soft extract (mean IC₅₀ = 1.1 mg/ml) (Table 3).

a: Calculated using the molecular mass of the main Δ7-sterol within pumpkin seeds: Δ^7.25-stigmastadienol − 412 g/mol [14] .

As previously published, rat liver tissue expresses 5-alpha-reductases. Microsomal liver preparations were used as source for 5α-R1 (isoform I). However, the same was found unsuitable for measurement of isoform II at pH 5.0, which was instead investigated in crude liver preparations. Assay validation was done for both preparations using known enzyme inhibitors. Both pumpkin seed soft extract and oil showed similar inhibitory capacities (both: IC₅₀ < 5 mg/ml for 5α-R1 and IC₅₀ ≈ 6 mg/ml for 5α-R2). The isolated Δ7-sterols (IC₅₀ = 0.3 mg/ml for 5α-R1, IC₅₀ = 1.0 for 5α-R2) were more potent than the pumpkin seed soft extract or oil, most likely since the Δ7-sterols are the main active principle within pumpkin seeds regarding anti-androgenic activity.

The test compounds inhibited the [³H]-DHT binding to AR in a concentration-dependent manner (Table 3), with IC₅₀ (Δ7-sterols) = 0.2 mg/ml), IC₅₀ (pumpkin seed oil) = 0.4 mg/ml and IC₅₀ (pumpkin seed soft extract) = 1.1 mg/ml. This dose-dependency indicates the capacity of all three test compounds to bind to the AR.

Among the tested substances, the Δ7-sterols had the lowest IC₅₀ and thus bind most potently to the AR.

Chang and Liao classified the antagonistic effects of different cyclic hydrocarbons on the AR [24] . Accordingly, the binding of Δ7-sterols (mean IC₅₀ of 0.48 mmol/l) corresponds to moderately active substances (IC₅₀ range 0.1 - 2 mmol/l). In a published experiment, human cells were pre-incubated with Δ7-sterols and then treated with DHT. In a concentration-dependent manner, a reduced binding of DHT to its binding sites could be demonstrated [18] . This goes along with our findings.

The inhibition of 5-alpha-reductase by Δ7-sterols and their inhibitory binding to AR are plausible due to the similar structure of these compounds compared to DHT [18] [25] . Several studies support that our in vitro observations also take effect in vivo:

Trevisan et al. 2012 found that orally administered Δ7 sterol (alpha-spinasterol) was well absorbed by mice [26] . Pumpkin seed or pumpkin seed oil in the diet of BPH animal models inhibited prostate growth [13] [27] [28] [29] . Effects were observed with only 40 mg pumpkin seed oil/kg body weight [28] . Furthermore, patients taking 90 mg of a Δ7 sterol mixture isolated from Uromedic® pumpkin seeds 3 - 4 days before prostatectomy had a lower DHT content in the operated prostate tissue than control patients [17] . Thus anti-androgen effects seen in our studies presumably take effect also in vivo, while, the safety of pumpkin seed soft extract and pumpkin seed oil has been established by human data [11] [30] [31] [32] . As a next step, the isolated Δ7 sterols (versus seed soft extract and seed oil) could be tested in BPH animal models, and finally in BPH patients.

In order to get more reliable IC₅₀ values, more experiments using at least 3 different concentrations of the test compounds (especially in the case of the 5-alpha-reductases), and using more independent measurements should be done. Since all our experiments were done using rat enzymes, potential differences in binding and inhibition to human orthologues cannot be excluded.