*4.2. Surface Plasmon Resonance*

SPR analyses were carried out at 25 ◦C using the optical biosensors Biacore T200 and Biacore 8K (GE Healthcare, Chicago, IL, USA) and sensor chips of CM5 series S type (Cytiva, Marlborough, MA, USA). HBS-N (10 mM HEPES, 150 mM NaCl, pH 7.4) (Cytiva) was used as a running buffer for CYP51A1 immobilization. Carboxyl groups of biosensor chip dextran were activated for 5 min by injection of the 1:1 mixture of 0.2 M 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) and 0.05 M N-hydroxysuccinimide (NHS) at a flow rate of 5 μL/min, followed by 1 min wash with HBS-N buffer at the same flow rate. Next, CYP51A1 (25 μg/mL) in 10 mM sodium acetate (pH 5.0) was injected into the working channel of the biosensor for 5 min at a flow rate of 5 μL/min. The final level of immobilization was 13,500 RU (13.5 ng of protein). Reference channel without immobilized CYP51A1 was used to correct the effects of the non-specific binding of analytes to the chip surface.

Baicalein and luteolin were prepared as 10 mM stock solutions in 100% dimethyl sulfoxide (DMSO). Experimental samples of baicalein and luteolin were prepared in an HBS-N buffer at the concentration range 10–100 μM and 1% DMSO. The same amount of solvent was added to the HBS-N running buffer to minimize bulk-effects introduced by the difference between the refractive indexes of the running buffer and the experimental samples. Refractive indexes of running buffer and experimental samples were matched with a precision refractometer RX-5000 (Atago, Saitama, Japan). If needed, the concentration of solvent in the running buffer was corrected according to the equation:

$$\mathcal{C}(DMSO)\_{running\ buffer} = \mathcal{C}(DMSO)\_{sample} \times \frac{\eta\_1 - \eta\_2}{\eta\_3 - \eta\_2}.$$

where *C(DMSO)running buffer*—DMSO final concentration in running buffer, *C(DMSO)sample* —DMSO concentration in experimental sample, *η*1—analyzed sample refractive index, *η*2—HBS-N buffer refractive index, *η*3—HBS-N buffer containing the DMSO of the same concentration as experimental sample refractive index.

Luteolin 7,3- -disulfate 10 mM stock solution and experimental samples at final concentrations of 10–100 μM were prepared with HBS-N buffer without organic solvent. The same buffer was used as a running buffer with luteolin 7,3- -disulfate to minimize the bulk-effects on the obtained experimental data. A total of 10 mM stock solution of lanosterol was prepared in ethanol. Lanosterol experimental samples at the final concentrations of 10–100 μM, as well as the running buffer, were prepared by the same protocol as for baicalein and luteolin but in ethanol instead of DMSO.

Low molecular weight compounds were injected through biosensor channels (working and reference) at a flow rate of 10 μL/min (luteolin 7,3- -disulfate) and 50 μL/min (baicalein and luteolin) for 6 min. Dissociation of the formed CYP51A1/compound complexes were registered at the same flow rate for no less than 6 min after the sample injection. After each biosensor cycle, a bound analyte was removed with two-times injection of regenerating solution (2 M NaCl, 1% CHAPS) at a flow rate of 30 μL/min for 30 s.

SPR sensorgrams were processed in Biacore T200 Evaluation Software v.1.0 (GE Healthcare) and BIAevaluation Software v 4.1.1 (GE Healthcare) using "1:1 (Langmuir) binding" and "Two-state (conformational change) binding" data processing models. The 1:1 (Langmuir) binding model is a model for the 1:1 interaction between compound (C) with immobilized protein (P), and is equivalent to the Langmuir isotherm for adsorption to a surface: C + P ↔ CP. Two-state (conformational change) binding model describes a 1:1 binding of compound (C) to immobilized protein (P) followed by a conformational change in the complex (CP ↔ CP\*). It is assumed that the conformationally changed complex can dissociate only through the reverse of the conformational change: C + P ↔ CP ↔ CP\*. The final kinetic parameters were obtained from the models with best fit of the experimental curves according to the minimum of the obtained chi2 value. The equations describing used models are as follows:


#### *4.3. Spectral Titration Analysis*

Spectrophotometric titration was used to determine the apparent dissociation constants (*Kdapp*) for the enzyme–ligand complexes. The spectral measurements were performed on Cary Series UV-Vis-NIR (Agilent Technologies, Santa Clara, CA, USA) spectrophotometer using tandem quartz cuvettes (1 cm optical path) to exclude the absorption of ligands. Natural substrate, lanosterol, at final concentration 5 μM was added before the titration to the CYP51A1 (final concentration 4 μM) in 50 mM potassium phosphate buffer, pH 7.4. For titration, the ligand solution was added to the experimental cuvette (baicalein and luteolin 7,3- -disulfate were added up to a final concentration of 30 μM, luteolin was added up to a final concentration of 15 μM) and an equal volume of solvent

was added to the control cuvette. The difference spectra were recorded after each addition of ligand at room temperature in the range of 350–500 nm. The apparent dissociation constants were determined by plotting the absorbance changes in the difference spectra versus the concentration of free ligand and evaluated by using the Hill equation (OriginPro 8.1 statistical data analysis package):

$$A\_{\rm obs} = A\_{\rm max} \times \left(\frac{\mathbb{S} \times n}{K\_{dapp} \times n + \mathbb{S} \times n}\right)^{\prime}$$

where *Aobs*—the observed change in the absorption, *Amax*—the absorbance change at ligand saturation, *Kdapp*—the apparent dissociation constant for the ligand–enzyme complex, *S*—the ligand concentration, *n*—a Hill coefficient.

## *4.4. Enzyme Assay*

Lanosterol 14-alpha-demethylase activity of human CYP51A1 was determined at 37 ◦C in 50 мM KPB, 4 mM MgCl2, 0.1 mM DTT in presence of lipids (0.15 мg/mL mixture 1:1:1 of L-α-dilauroyl-sn-glycero-3-phosphocholine, L-α-dioleoyl-sn-glycero-3-phosphocholine and L-α-phosphatidyl-L-serine). The final concentrations of CYP51A1 and CPR were 0.5 and 2.0 μM, respectively. Aliquots of concentrated recombinant proteins were mixed and preincubated for 5 min at room temperature. Lanosterol (10 mM stock solution in ethanol) was added to the reaction mixture at a final concentration of 50 μM. Tested compounds were added to the reaction mixture at a final concentration of 25 μM. To estimate the apparent IC50, the following concentrations of luteolin 7,3- -disulfate were used: 5, 10, 25, 50 and 100 μM. Ketoconazole at a concentration of 5 μM was used as a positive control. After 10 min of preincubation at 37 ◦C, the reaction was started by adding NADPH at a final concentration of 0.25 mM. Aliquots (0.5 mL) were taken from the incubation mixture at chosen time intervals. Steroids were extracted with 5 mL of ethyl acetate. The mixture was vigorously mixed, the water and organic phases were separated by centrifugation at 3000 rpm for 10 min. The organic layer was carefully removed and dried under argon flow. A total of 50 μL of methanol was added to the pellet and steroids were analyzed on a computerized Agilent 1200 series HPLC instrument (Agilent Technologies, USA) equipped with Agilent Triple Quad 6410 mass spectrometer (Agilent Technologies). Samples were analyzed by gradient elution on Zorbax Eclipse XDB C18 column (4.6 × 150 mm; 5 μm) (Agilent Technologies). A total of 0.1% (*v/v*) FA in water was used as the mobile phase A and 0.1% (*v/v*) FA in methanol:1-propanol mix (75:25, *v/v*) as mobile phase B. The gradient was 75–100% B in 0–5 min. The flow rate was 500 μL per min. The column temperature was maintained at 40 ± 1 ◦C. Mass spectrometry experiments were performed with atmospheric pressure chemical ionization source (APCI) at positive ion mode. The following APCI settings were used: gas temperature 200 ◦C, vaporizer 250 ◦C, gas flow 7 L/min, nebulizer pressure 40 psig, Vcap 4000 V, corona 4 μA, fragmentor 100 V. The data acquisition mode was MS2Scan from 200 to 550 Da.

#### *4.5. Molecular Docking*

Crystal structure of CYP51A1 PDB ID 3LD6 was used for docking. 3D structures of luteolin (CID 5280445) and luteolin 7,3- -disulfate (CID 44258153) were obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/, accessed on 7 November 2020). Removal of water and ligand molecules from the original protein PDB files and molecular docking over the entire surface of CYP51A1 were performed automatically in the Flare software package (Cresset, Litlington, UK) with default settings [59]. Docking hypotheses were arranged according to score functions values: Lead Finder (LF) Rank Score, LF dG, LF VSscore. The lower is the LF Rank Score, the higher is the likelihood that the docked pose reproduces the crystallographic pose. LF dG has been designed to perform accurate estimation of the free energy of protein–ligand binding for a given protein–ligand complex. LF VSscore has been designed to produce maximum efficiency in

virtual screening experiments, i.e., to assign higher scores to active ligands (true binders) and lower scores to inactive ligands. Molecular graphics visualization tool Maestro version 12.5.139 (Schrödinger, New York, NY, USA) was used to analyze the selected docking hypotheses.

#### **5. Conclusions**

In this work, we identified a new ligand of human CYP51A1 among natural flavonoid luteolin 7,3- -disulfate—that inhibits 14α-demethylase activity. Potential inhibitory mechanisms include blocking of either a substrate access channel or the interaction with a redox partner. Obtained results suggest further exploration of polyphenols for the cholesterol lowering ability and anti-cancer potential via CYP51A1.

**Author Contributions:** Conceptualization, L.K., P.E. and E.Y.; methodology, L.K., P.E. and E.Y.; formal analysis, L.K., P.E., T.S., P.S., S.F., I.G. and A.G.; investigation, L.K., P.E., E.Y., Y.M. and O.G.; resources, T.S., S.U., A.P., A.A., O.S. and N.S.; data curation, L.K., P.E. and E.Y.; writing—original draft preparation, P.E.; writing—review and editing, L.K., P.E., E.Y., Y.M., O.G., A.G. and N.S.; visualization, L.K., P.E. and E.Y.; supervision, N.S.; project administration, A.I. All authors have read and agreed to the published version of the manuscript.

**Funding:** Surface plasmon resonance analysis was funded by the Russian Foundation for Basic Research (grant № 20-04-00014) and was performed using the equipment of "Human Proteome" Core Facility of the Institute of Biomedical Chemistry supported by MINOBRNAUKI, Agreement № 075-15-2019-1502 from 5 September 2019. Catalytic activity analysis was funded by SPSR of Belarus "Chemical processes, reagents and technologies, bioregulators and bioorganic chemistry", (project 2.3.1.1). Spectral titration analysis and molecular docking simulations were done in the framework of the Russian Federation fundamental research program for the long-term period for 2021–2030.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Datasets are available from the authors.

**Conflicts of Interest:** The authors declare no conflict of interest.

**Sample Availability:** Samples of the compounds are available from the authors.

#### **References**

