Androgenetic alopecia (AGA) is a well-known condition characterized by hair loss, affecting both men and women (Inui and Itami, 2011; Ntshingila et al., 2023). Traditionally, the mechanism of AGA involves the conversion of testosterone to dihydrotestosterone (DHT) by 5-α-reductase inside cells. DHT then affects the hair follicles, shortening the hair cycle by regulating the duration of anagen and lengthening the catagen and telogen phases, ultimately leading to hair loss (Ryan et al., 2011). The most crucial events during the anagen-to-catagen transition induced by testosterone are the degeneration of dermal papilla cells (DP) and the apoptosis of the outer root sheath (ORS). The degeneration of DP cells reduces the nutrient supply to hair cells, contributing to hair loss. Additionally, the ORS, which surrounds the hair, is a vital hair tissue, and its cell death leads to a decrease in the structural support of the hair, resulting in hair loss (Lolli et al., 2017). Therefore, the degeneration of DP cells and the cell death of ORS cells contribute to the initiation of male pattern baldness by inducing a shortened hair cycle.

Finasteride and dutasteride, as 5-α-reductase inhibitors, and minoxidil, as an ATP-sensitive K⁺ channel regulator, have been developed for conventional treatments of androgenetic alopecia. However, these treatments have reported controversial efficacy and various side effects. The side effects of minoxidil include scalp dryness and skin irritation, while those of finasteride and dutasteride include depression and erectile dysfunction. Based on this premise, several natural compounds derived from Urtica dioica, Humulus lupulus, Serenoa repens, Vitis vinifera, Pygeum africanum, Crocus sativus, Medicago sativa, Linum usitatissimum, Brassica oleracea var. italica, and Cucurbita pepo have been proposed for the treatment of alopecia (Cho and Kim, 2020).

Several lines of evidence suggest that testosterone induces various physiological phenomena through both conventional genomic and non-genomic pathways (Pi et al., 2005; Pi and Quarles, 2012). Upon binding to GPRC6A, testosterone activates the downstream signaling protein Gq, leading to the activation of PLCβ-PIP2-IP3 signaling cascade (Ko et al., 2014). Subsequently, an increase in intracellular calcium (Ca²⁺) levels activates Duox1, a Nox isozyme predominantly present in keratinocytes involved in skin barrier formation. The activation of Duox1 generates reactive oxygen species (ROS), ultimately leading to apoptosis of hair follicle cells.

Particularly, testosterone is known to promote the anagen-to-catagen transition, contributing to hair loss. In pathological conditions like hair loss, intracellular Nox activation results in increased intracellular reactive oxygen species (ROS). Therefore, it can be hypothesized that excessive production of toxic ROS induces the degeneration of DP cells and the cell death of ORS. The process of male pattern baldness through testosterone-ROS mediated cell signaling represents a novel molecular pathological mechanism. Based on this concept, we suggest potential natural candidates for treating AGA through inhibition of Nox isozyme in hair follicle cells.

Materials

Testosterone (17β-Hydroxy-3-oxo-4-androstene; T1500), N,N′-Dimethyl-9,9′-biacridinium dinitrate (lucigenin; M8010) and β-nicotinamide adenine dinucleotide phosphate hydrate (NADPH; N5755) were purchased from Sigma Aldrich (St. Louis, MO, USA). 2’,7’–dichlorofluorescin diacetate (DCF-DA; D-399) was purchased from Molecular Probes (Waltham, MA, USA).

Cell culture

Ker-CT (hTERT/CDK4 immortalized human keratinocytes, CRL-4048) cells were purchased from ATCC^® and cultured in EpiLife^TM medium (Gibco, Waltham, MA, USA, MEPI500CA) supplemented with EDGS (EpiLife^TM Defined Growth Supplement; Gibco, S0125) and 1% Penicillin-Streptomycin solutions (Welgene, Gyeongsan, Korea, LS202-02).

Preparation of primary mouse keratinocytes

Primary keratinocytes were isolated from skin of newborn C57BL/6J mice. Newborn mice were sacrificed by asphyxiation using CO₂ and washed with 70% ethanol and phosphate-buffered saline. The trunk skin was stripped and floated on the 0.25% trypsin solution (Cellgro, Manassas, VA, USA, 25-050) in 6 well culture plate. After overnight incubation at 4°C, the epidermis was separated from dermis and mechanically chopped in KBM medium without serum. The tissue was filtered through a 100-micron cell strainer (Falcon, Corning, NY, USA, 352360) and centrifuged. After removed the supernatant, cells were plated onto collagen (bovine collagen coating solution, Cell Applications, San Diego, CA, USA, 125-100)-coated 6 well plate at a density of 10⁶ per well.

Measurement of intracellular ROS by DCF-DA

After cells were starved overnight with serum free medium, they were stimulated by testosterone with or without Tanshinone and its isotypes. Cells were washed with Hanks’ balanced salt solution (HBSS) and incubated with HBSS containing 10 µM DCF-DA for 10 min. The fluorescence was measured using a LSM510 confocal microscope (Carl Zeiss, Oberkochen, Germany) at an excitation wavelength of 488 nm and an emission wavelength of 515-540 nm. Five groups of each cells were randomly selected, and the mean relative fluorescence intensity was measured with a Carl Zeiss vision system (LSM510, version 2.3). All experiments were repeated at least three times.

TdT-UDP nick end labeling (TUNEL) assay

Apoptotic cells were detected by the TUNEL technique using an In Situ Cell Death Detection Kit, Fluorescein (Roche, Basel, Switzerland, 11684795910). Cells were incubated with testosterone with or without Tanshinone I and Tanshinone IIA for 24 h. They were fixed with 3.5% paraformaldehyde for 1 h at room temperature and permeabilized with pre-chilled 0.5% triton X-100 in PBS for 10 min at room temperature. These cells were then stained with 4′,6-diamidino-2-phenylindole (DAPI) for 10 min and mounted with mounting solution (Sigma, St. Louis, MO, USA, M1289). Fluorescence was measured by confocal microscopy. Over four points, samples were randomly detected and the percentage of apoptotic cells was determined by counting numbers of positively stained cells using Image J software.

Nox inhibitor screening

Nox1, 2, 4, Duox1, and Duox2 inhibitory activities were determined by previous report (Joo et al., 2016). To specifically inhibit Nox isozymes, transgenic Drosophila Duox knockdown lines over-expressing human Nox1, 2, or 4 were established. Genotypes used were human Nox1 (UAS-hNOX1/UAS-DUOX-RNAi; Da-GAL4/+), Nox2 (UAS-hNOX2/UAS-DUOX-RNAi; Da-GAL4/+) and Nox4 (UAS-hNOX4/UAS-DUOX-RNAi; Da-GAL4/+), human Duox1 (UAS-hDUOX1/UAS-DUOX-RNAi; Da-GAL4/+), human Duox2 (UAS-hDUOC2/UAS-DUOX-RNAi; Da-GAL4/+) (Ha et al., 2005). Transgenic flies were homogenized with ice-cold PBS containing protease inhibitors. Membrane-enriched human Nox1, Nox2, Nox4, Duox1, or Duox2 was then harvested. Membranes containing human Nox1, Nox2, Nox4, Duox1, or Duox2 served to monitor ROS production by lucigenin chemiluminescence in the absence or presence of a chemical compound. The reaction medium consisted of HEPES-buffered salt solution (pH 7.4), 400 µM lucigenin (10,10-dimethyl-bis-9,9-bisacridinium nitrate) and 500 µM NADPH.

In silico complex modeling

Using the crystal structure of the dehydrogenase domain of Nox5 originating from Cylindrospermum stagnale (csNOX5-DH) as a template (PDB code: 5O0X), the dehydrogenase domain of hDuox2 (hDuox2-DH) was modeled using the Swiss-Model server (Bordoli et al., 2009). The hDuox2-DH/FAD complex was generated by substituting csNox5 with hDuox2 in the crystal structure of the csNox5-DH/FAD complex. Structures of Tanshinone derivatives were prepared with the ChemDraw 3D (Perkin Elmer informatics, MA, USA). Molecular docking simulations of Tanshinone derivatives into the hDuox2-DH were carried out using AutoDock Vina (Trott and Olson, 2010).

Tanshinone I and Tanshinone IIA efficacy test in testosterone-treated dorsal skin of C57BL/6J male mice

To estimate the effect of Tanshinone I and Tanshinone IIA on androgenic alopecia, C57BL/6J male mice (8-week-old) were anesthetized with isoflurane, and their back skin hair was shaved. Hairs were completely removed with appropriate amount of hair removal cream. One day after of hair removal, the mice were treated with testosterone (200 µg/100 µL) for 3 days. Subsequently, 50 µM and 100 µM concentrations of Tanshinone I and Tanshinone IIA were diluted with 70% ethanol and topically administrated to the back-skin of each testosterone-treated mice group. The treatment was repeated once daily for 12 days. After the treatment period, mice were sacrificed, and their back skins were collected. Tissues were stored in frozen section media (Leica, Wetzlar, Germany, 3801480) at –80°C until further processing. Sections were then prepared at a thickness of 10-12 µm.

Hematoxylin & Eosin (H&E) staining for the measurement of hair length

Mouse back-skin tissues were dried for 10 min and fixed with 10% neutral buffered formalin (Sigma, HT501320) for 5 min at room temperature. Tissues were stained with hematoxylin (VECTOR Laboratories, Newark, NJ, USA, H-3404) for 1-2 min and followed by rinsing with tap water for 3 min. Then tissues were stained with Eosin Y (Sigma, HT110132). Tissues were washed three times with deionized water for 5 min in between every process. Tissues were dehydrated by 95% ethanol twice for 1 min and 100% ethanol twice for 3 min. Finally, tissue slides were mounted using mounting solution (Thermo Fisher Scientific, Waltham, MA, USA, 6769007). Images were obtained with light microscope (Nikon Eclipse E200, Tokyo, Japan) and the length of hair follicles were measured using Image J software. At least 50 follicles were counted to evaluate average hair length of one organism.

Statistical analyses

All values are presented as mean ± SD or SEM. Statistical significance between groups was determined using two-tailed Student’s t-test. P-value less than 0.05 was considered statistically significant.

Identification of Tanshinone compounds as Nox inhibitors

Our previous reports indicated that Duox1 is involved in testosterone-dependent hair loss (Ko et al., 2014). To identify chemical compounds having Nox inhibitory activity for the therapeutic development for AGA, purified Drosophila membranes expressing human Duox1 were used to monitor ROS production by determining the oxidation of lucigenin, evident as chemiluminescence, in the absence or presence of chemical compound. The natural product library (Catalog Number L1400, Selleckchem, Houston, TX, USA) was screened to identify Nox inhibitors. We identified Tanshinone I, Tanshinone IIA, Tanshinone IIB, and Cryptotanshinone (Fig. 1). Tanshinone compounds are classified as abientane diterpenes isolated from Salvia miltiorrhiza, commonly known as Danshen in Chinese. Danshen is a well-known herb in traditional Chinese medicine as a therapeutic remedy for coronary heart diseases, vascular diseases, stroke, hyperlipidemia, endangiitis, arthritis and hepatitis (Zhou et al., 2005; Dong et al., 2011; Zhang et al., 2012a). Tanshinone I exhibited high inhibitory activity with IC₅₀ values of 2.6 μM for hDuox2 and 3.9 μM for hNox2, Tanshinone IIA showed IC₅₀ values of 2.2 μM for hNox1 and 2.8 μM for hDuox1, Tanshinone IIB displayed IC₅₀ values of 5.2 μM for hNox5 and 8.9 μM for hNox1 and hDuox2, while Cryptotanshinone demonstrated IC₅₀ values of 2.1 μM for hDuox2 and 3.7 μM for hDuox1, indicating a broad inhibitory effect across Nox isozymes (Fig. 1, Table 1). All tanshinone compounds exhibited comparable efficacy to the well-known pan-Nox inhibitor APX-115 (Joo et al., 2016). Salvianolic acid A and B were well-known abundant compounds in Danshen and exhibit good antioxidant activity. However, these two Salvianolic acid compounds did not show any effect on Nox inhibition (Supplementary Fig. 1).

Table 1 The IC₅₀ value of Tanshinone compounds on hNox1, hNox2, hNox4, hNox5, hDuox1, and hDuox2

	IC50 (μM)
	hNox1	hNox2	hNox4	hNox5	hDuox1	hDuox2
Tanshinone I	12.9	14.1	4.9	3.9	6.5	2.6
Tanshinone IIA	2.2	3.4	7.2	1.9	2.8	3.6
Tanshinone IIB	8.9	10.4	17.8	5.2	11.9	8.9
Cryptotanshinone	4.6	4.4	7.9	4.3	3.7	2.1
APX-115	2.6	0.9	2.3	2.2	2.5	0.7

Figure 1. Chemical structure of Tanshinone compounds and Nox inhibition profiles of Tanshinone compounds. (A) Chemical structure of Tanshinone compounds from Salvia miltiorrhiza. (B) Drosophila membranes overexpressing hNox1, hNox2, hNox4, hDuox1, and hDuox2 were subjected to ROS measurement with Lucigenin.

To explore the specificity of Tanshinone compounds from Salvia miltiorrhiza in inhibiting ROS production mediated by Nox, we conducted off-target experiments. Initially, we performed a scavenging assay to assess ROS scavenging capabilities of Tanshinone compounds. We used lucigenin, which reacts with exogenously added H₂O₂ and emits luminescence. The well-known ROS scavenger N-acetyl-L-cysteine (NAC) or the Tanshinone compounds were incubated with lucigenin in the presence of H₂O₂. We observed that NAC reduced the luminescence intensity from oxidized lucigenin. However, the Tanshinone compounds did not inhibit the luminescence generated from lucigenin, indicating that the compounds lack scavenging activity for ROS (Supplementary Fig. 2A). Subsequently, we investigated whether the compounds inhibited enzymes other than Nox that generate H₂O₂, such as glucose oxidase (GO) and xanthine oxidase (XO). NAC effectively eliminated ROS generation dependent on GO and XO, whereas the Tanshinone compounds did not exhibit such inhibition (Supplementary Fig. 2B). These findings suggest that all Tanshinone compounds are specific inhibitors of Nox and do not have ROS scavenging activity.

Molecular docking simulation for binding modes of Tanshinone compounds in Duox2

To elucidate how Salvia miltiorrhiza diterpenoid compounds suppress the activity of Duox2, an isozyme of Nox, the in silico docking simulations of these compounds were performed. Nox is a membrane protein consisting of a catalytic core with a transmembrane helical domain (TM) and a C-terminal cytosolic dehydrogenase domain (DH). The DH domain contains binding sites for flavin adenine dinucleotide (FAD) and NADPH, while the TM domain has two heme binding sites. Nox isozymes, such as Nox5, Duox1, and Duox2, are known to be regulated by the EF-hand domain which is attached to the N-terminal region (Wu et al., 2021). When a Nox isozyme is activated by a structural change in the EF-hand domain due to increased intracellular calcium, electron transfer occurs sequentially from NADPH to FAD, inner heme, outer heme, and O₂. During this NADPH oxidation process, π-π interactions are formed between the nicotinamide ring of NADPH and the isoalloxazine ring of FAD, facilitating the movement of electrons from NADPH to FAD. According to the docking simulations, all Salvia miltiorrhiza diterpenoid compounds were docked at the NADPH-binding site, forming π-π stacking interactions with the isoalloxazine moiety of FAD (Fig. 2). The binding affinities (ΔG) of Tanshinone I, Tanshinone IIA, Tanshinone IIB_R, and Tanshinone IIB_S were estimated as –8.5, –9.3, –8.8, and –9.3 kcal/mol, respectively. Both Cryptotanshinone_R and S forms displayed affinities of –9.5 kcal/mol. Considering the consistent binding patterns of all Tanshinones and the high sequence similarity among the Nox isozymes, it can be inferred that Salvia miltiorrhiza diterpenoid compounds would similarly inhibit Nox isozymes including Duox2 by interfering with NADPH-binding which is required for activating the enzymes during the cascading signal transduction in response to testosterone.

Figure 2. The binding modes of Tanshinone derivatives. The active-site cavities are color-coded: red, blue, and white indicate negative, positive, and neutral electrostatic potentials, respectively. Tanshinone derivatives are depicted as sticks, with green sticks symbolizing FAD molecules.

Tanshinone I and Tanshinone IIA suppress testosterone-mediated ROS generation

The hair follicle is composed of epithelial cells, known as root sheaths and the dermal papilla. Root sheath cells encircle the hair shaft, and the dermal papilla plays a crucial part in the cycles of hair growth (Yang and Cotsarelis, 2010). Nox4 and Duox1 are predominant isozymes found in DP and ORS cells, respectively (Supplementary Fig. 3). We evaluated the effects of four Tanshinone compounds on testosterone-induced ROS inhibition and analyzed which of the compounds exhibited the most superior inhibitory effect on testosterone-mediated ROS generation. Stimulation of Ker-CT (immortalized human foreskin keratinocyte cell) cells and primary mouse keratinocytes with Tanshinone I, Tanshinone IIA, Tanshinone IIB, or Cryptotanshinone resulted in the suppression of testosterone-induced ROS generation (Fig. 3A, 3B). Among them, Tanshinone I and Tanshinone IIA exhibited better inhibitory activities on testosterone-induced ROS generation, compared to Tanshinone IIB and Cryptotanshinone. To further analyze the inhibitory potency of Tanshinone I and Tanshinone IIA, we evaluated their effects at different concentrations in Ker-CT cells. The IC₅₀ values were determined to be 47.73 nM for Tanshinone I and 151 nM for Tanshinone IIA (Fig. 3C-3E). Taken together, Tanshinone I and Tanshinone IIA effectively inhibited the testosterone-induced intracellular H₂O₂ generation by suppressing Duox1, which is the predominant isozyme in Ker-CT cells.

Figure 3. Tanshinone compounds inhibit testosterone-dependent ROS generation. (A) Inhibitory activities of Tanshinone compounds in Ker-CT cells (A) and primary mouse keratinocytes (B). Ker-CT cells and primary mouse keratinocytes were pretreated with 10 μM Tanshinone I, Tanshinone IIA, Tanshinone IIB, and Cryptotanshinone. After pretreated each Tanshinone compounds, cells were stimulated with testosterone (200 ng/mL) for 10 min. Testosterone-induced ROS generation was assessed by confocal microscopic analysis of DCF fluorescence. Data are representative of three repeated experiments and are presented as mean ± SD (n=3). Concentration-dependent Nox inhibition of Tanshinone I (Tan I) (C) and Tanshinone IIA (Tan IIA) (D). Ker-CT cells were pretreated with indicated concentration of Tanshinone I and Tanshinone IIA for 1 h. Following pre-treatment with Tanshinone I and Tanshinone IIA, Ker-CT cells were stimulated with testosterone (200 ng/mL) for 10 min. Testosterone-induced ROS generation was monitored by confocal microscopic analysis of DCF fluorescence. Data represent three repeated experiments and are shown as mean ± SD (n=3). (E) IC50 of Tanshinone I and Tanshinone IIA.

Tanshinone I and Tanshinone IIA regulate testosterone-mediated cell death

Previously, we reported that testosterone-dependent H₂O₂ generation is involved in the cell death of keratinocytes. To investigate whether Tanshinone I and Tanshinone IIA could regulate testosterone-induced cell death, we performed a terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay, a hallmarker of cell death, in Ker-CT cells. Based on the IC50 values of Tanshinone I and Tanshinone IIA, we applied 100 nM and 500 nM of each compound to Ker-CT cells in the absence or presence of testosterone. It was observed that testosterone-induced cell death was significantly reduced at both 100 nM and 500 nM of Tanshinone I and Tanshinone IIA (Fig. 4). The inhibitory activity of Tanshinone I on testosterone-induced cell death was better than that of Tanshinone IIA, consistent with the results of intracellular ROS inhibition activity. Moreover, we performed cell viability assay with Tanshinone I and Tanshinone IIA in the presence of testosterone in Ker-CT cells. Treatment of Ker-CT cells with Tanshinone I and Tanshinone IIA resulted in suppressed testosterone-dependent cell death (Supplementary Fig. 4). These results indicate that Tanshinone I and Tanshinone IIA effectively inhibit testosterone-induced Nox activity in Ker-CT cells, thereby suppressing intracellular H₂O₂ production and consequent cell death.

Figure 4. Tanshinone I and Tanshinone IIA suppress testosterone-dependent cell death. (A) Primary mouse keratinocytes were isolated from C57BL/6 mice and treated with or without 200 nM testosterone along with Tanshinone I (Tan I) (at concentrations of 100 nM and 500 nM) or Tanshinone IIA (Tan IIA) (at concentrations of 100 nM and 500 nM) for 24 h. Following incubation, the cells were fixed and subjected to TdT-UDP nick end labeling for 1 h at 37°C. TUNEL-positive cells were visualized by fluorescence microscopy. The percentage of apoptotic cells was determined by examining eight random high-power fields was determined. (B) Data are shown as mean ± SEM.

Tanshinone I and Tanshinone IIA inhibit testosterone-induced hair loss in C57BL/6J mice

To determine the efficacy of Tanshinone I and Tanshinone IIA in a testosterone-induced AGA mice model, both compounds were applied to the back skin of C57BL/6J mice. The backs of 8-week-old C57BL/6J mice were shaved, and mice with pink skin, indicating the resting (telogen) phase in the hair cycle, were selected. Testosterone was applied to the shaved area for 3 days to inhibit hair growth, followed by co-application of Tanshinone I and Tanshinone IIA with testosterone for 12 days. Tanshinone I and Tanshinone IIA were applied at concentrations of 50 and 100 μM, each in 100 μL, and photographs were taken every 2 days to assess hair loss status based on skin color. By the 6th day after application, grayish hair started to grow from the skin, and by the 12th day, black hair was observed in some areas of the back skin (Fig. 5A, 5B). The application of Tanshinone I and Tanshinone IIA significantly enhanced hair growth in back skin of mice, compared with mice treated testosterone alone (Fig. 5B).

Figure 5. Effect of Tanshinone I and Tanshinone IIA on hair growth in an androgenetic alopecia mouse model. (A) Schematic of hair growth measurement from an androgenetic alopecia mouse model. The dorsal skins of 8-week-old C57BL/6 mice were topical treated daily with testosterone (200 μg) for 3 days, following by co-treated with Tanshinone I (at concentrations of 50 μM or 100 μM) and Tanshinone IIA (at concentrations of 50 μM or 100 μM) for 12 days. (B) Images of dorsal skin of mice treated with Tanshinone I and Tanshinone IIA as indicated time.

On the 12th day after application, mice were sacrificed, and the back skin was obtained. The obtained tissue was stained with H&E, and hair follicle length was measured using ImageJ software, from the bottom of the hair bulb to the epidermal pore along the hair shaft. Hair follicle lengths were measured for more than 50 hair follicles per sample, and the average values were calculated (Fig. 6). It was confirmed that hair follicle length significantly increased in the groups co-treated with testosterone and 50 μM and 100 μM of Tanshinone I or Tanshinone IIA, compared to the control group treated with testosterone alone. Furthermore, the application of 100 μM of Tanshinone I or Tanshinone IIA exhibited better efficacy in promoting hair follicle growth compared to 50 μM of each compound (Fig. 6). We next conducted a comparative experiment of Tanshinone I and Tanshinone IIA against positive control therapy. We used 5% minoxidil in a topical form as a positive control because Tanshinone I and Tanshinone IIA are formulated for topical application on the scalp. A 20 μM dose of either Tanshinone I or Tanshinone IIA was applied to the back-skin of C57BL6 mice, and the subsequent skin color changes to black were measured as a marker of hair growth. The hair growth enhancement activities of Tanshinone I and Tanshinone IIA were found to be comparable to that of 5% minoxidil (Supplementary Fig. 5). Thus, it can be concluded that Tanshinone I and Tanshinone IIA, which effectively inhibit testosterone-activated ROS generation, exhibit hair growth efficacy in a male pattern baldness model.

Figure 6. Effect of Tanshinone I and Tanshinone IIA on hair follicle length in an androgenetic alopecia mouse model. (A) Dorsal skin tissues of androgenic alopecia mouse treated with Tanshinone I (Tan I) (at concentrations of 50 μM or 100 μM) and Tanshinone IIA (Tan IIA) (at concentrations of 50 μM or 100 μM) for 12 days were stained with H&E. Scale bar=200 μm. Hair follicle lengths of Tanshinone I (at concentrations of 50 μM or 100 μM) and Tanshinone IIA (at concentrations of 50 μM or 100 μM)-treated groups or control groups for 12 days were measured. (B) Data shown are mean (± SEM) of results from 4 to 5 independent sample preparations.

Testosterone-induced hair loss is a well-established phenomenon. Testosterone is converted into dihydrotestosterone (DHT) by the action of 5-α-reductase enzymes in the cytoplasm (Inui and Itami, 2011; Lolli et al., 2017; Ntshingila et al., 2023). The interaction between DHT and androgen receptors (AR) stimulates the expression of numerous genes, leading to various physiological effects, including the miniaturization of hair follicles, the transition from the anagen to the catagen phase, and ultimately male pattern baldness. In response to this understanding, commercial drugs like Dutasteride and Finasteride have been developed to inhibit 5-α-reductase activity (Adil and Godwin, 2017). However, these inhibitors, originally designed for prostate cancer treatment, come with several side effects including impotence, decreased libido, erectile dysfunction, testicular pain, and ejaculation disorders. Consequently, there is growing anticipation for the development of alternative hair loss treatments.

In addition to the well-known concept of testosterone-induced hair loss, we have also elucidated a non-genomic pathway involving the activation of GRPC6A-Calcium-Duox1. This pathway works in tandem with the genomic pathway triggered by testosterone. Within hair follicle cells, Duox1, the predominant isozyme, is directly activated by testosterone-induced mobilization of intracellular calcium. This activation leads to significant conformational changes in the regulatory PHLD and calcium-binding domains (EF1 and EF2) within the cytosolic layer of Duox1 (Wu et al., 2021). In the absence of calcium, the EF module adopts a more contracted shape in conjunction with the dehydrogenase (DH) domain. However, in the presence of intracellular calcium, EF2 shifts away from the DH domain, while the PHLD rotates away from the transmembrane domain (TMD) and DH domains. Consequently, the DH domain forms a stable interaction with the TMD, serving as a site for FAD and NADPH binding, thereby stimulating the generation of reactive oxygen species (ROS). This non-genomic pathway, by activating Duox1 in hair follicle cells, contributes to the production of ROS. Furthermore, our research sheds light on the molecular mechanism of testosterone-dependent apoptosis, where testosterone triggers the generation of H2O2 through the activation of Duox1 (Fig. 7). Ultimately, the oxidative stress induced by testosterone results in cell death in hair follicle cells, leading to hair loss (Trüeb, 2015, 2021). In the context of the hair loss market, there is a growing demand for new drugs to mitigate the side effects of existing 5-α-reductase inhibitors. As a response, the development of Nox inhibitors has emerged as a promising solution to address these challenges in the alopecia market.

Figure 7. Proposed model for the function of Tanshinone in androgenetic alopecia. The binding of testosterone to GPRC6A triggers downstream signaling pathway involving intracellular calcium mobilization and Nox activation. Nox-mediated ROS generation induces apoptosis of keratinocytes. Tanshinone suppresses Nox activity, ultimately inhibiting hair loss.

Tanshinone compounds were initially identified from the root of Salvia miltiorrhiza, commonly known as Danshen in traditional Chinese medicine (Peixin et al., 2012; Li et al., 2020). These compounds have been extensively researched for their diverse pharmacological properties, including anti-inflammatory and anticancer effects. They are notably recognized for their potential therapeutic applications in cardiovascular diseases, neurodegenerative disorders, and dermatological conditions. Among the Tanshinones, Tanshinone IIA stands out as the primary fat-soluble component of Salvia miltiorrhiza and plays a role in the androgen receptor pathway (Zhang et al., 2012a). Due to its structural resemblance to testosterone, tanshinones function as antagonists of the androgen receptor (AR) pathway. In studies conducted on prostate cancer cells, Tanshinones exhibited a concentration-dependent suppression of cell growth and AR-dependent transcription. Tanshinone IIA also regulated the expression of the androgen receptor downstream target, prostate-specific antigen (PSA) (Zhang et al., 2012b). Interestingly, the mechanisms through which Tanshinones inhibit AR transcriptional activity differ from those of traditional AR antagonists such as flutamide and bicalutamide (Ju et al., 2005). Moreover, these conventional AR antagonists, like flutamide and cyproterone, do not effectively inhibit testosterone-induced reactive oxygen species (ROS) generation, implying that their binding sites may vary from those of GPRC6A. Our findings demonstrate that Tanshinone effectively suppresses testosterone-induced ROS generation in a dose-dependent manner. Nonetheless, further research is needed to elucidate the precise binding site of Tanshinone with GPRC6A.

It has been reported that Tanshinone compounds have therapeutic effects, such as anti-cancer, anti-inflammatory, anti-fibrosis activities, on various diseases (Peixin et al., 2012; Li et al., 2020). Tanshinone-dependent MAPK, NFkB, and caspase inhibitions in cancer cells is involved in promoting apoptosis and suppressing migration. Tanshinone compounds also have a significant therapeutic effect on cardiovascular diseases. Tanshinone-mediated inhibition of MAPK and NFkB regulates the suppression of pro-inflammatory cytokines such as IL1β and TNFα. Most of these therapeutic activities were derived from the regulations of protein phosphorylation. Several lines of evidence indicated that ROS regulates protein phosphorylation. In general, cytosolic protein phosphorylation was mediated by the balance between protein tyrosine kinases (PTKs) and protein tyrosine phosphatases (PTPs). PTPs contain a thiolate anion in a cysteine residue at their active center. Nox-mediated ROS generation inhibits PTPs activity by oxidizing the thiolate anion to sulfenic acid (-SOH) in the active center. This oxidation of PTPs disrupts the balance between PTKs and PTPs. As a result, PTK activity is relatively enhanced, leading to increased tyrosine phosphorylation of cell signaling-related proteins like MAPK and NFkB. It is known that the application of Tanshinone results in the suppression of Nox-mediated ROS generation, leading to significant therapeutic effects in various diseases.

Tanshinone I and Tanshinone IIA exhibit potent Nox inhibitory activity regarding testosterone-mediated ROS generation (Fig. 3). Interestingly, the Nox inhibitory activity of Tanshinone I is greater than that of Tanshinone IIA. However, the hair growth-enhancing activity of Tanshinone I is similar to that of Tanshinone IIA in a testosterone-mediated hair growth suppression animal model (Fig. 5, 6). Although Tanshinone IIA shows lower in vitro Nox inhibition activity compared to Tanshinone I, its in vivo activity is comparable to that of Tanshinone I. This suggests that the in vitro Nox inhibitory activities of Tanshinone I and Tanshinone IIA do not correlate with their hair growth activities in the androgenetic alopecia mouse model. Since Tanshinone IIA contains one more methyl group than Tanshinone I, it may possess slightly more hydrophobic properties, which could allow it to penetrate the skin barrier more easily. Therefore, it is likely that the pharmacokinetics (PK) of Tanshinone IIA in skin tissues are more favorable than those of Tanshinone I. However, detailed PK studies of both Tanshinone I and Tanshinone IIA are needed to provide a mechanistic explanation.

In summary, we have identified diterpenoid compounds, namely Tanshinone I, Tanshinone IIA, Tanshinone IIB, and Cryptotanshinone, as potent Nox inhibitors. These Tanshinone compounds have demonstrated effective inhibition of testosterone-mediated intracellular H₂O₂ levels, thereby preventing subsequent cell apoptosis in keratinocytes. Furthermore, both Tanshinone I and Tanshinone IIA have shown the ability to suppress testosterone-dependent alopecia in the back skin of C57BL/6 mice. Based on our findings, we propose that these Tanshinone compounds hold significant promise as therapeutic agents for the treatment of male AGA.

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) [RS-2024-00398295 to Y.S.B, NRF-2022R1A2C3006924 to S.-S.C., and NRF-2021R1A2C1091259 to I.H.L.] and by the Starting growth Technological R&D Program (20165925) funded by the Ministry of SMEs and Startups (MSS, Korea).

Y.S.B. has filed a Korean patent (KR 10-2017-0012315) covering its applications. The rights to the patent were transferred to Celros Biotech (Seoul, Korea).