PAEs were prepared by superfine crushing of pilose antler and then extracted with acetic acid. Brifly, fresh pilose antlers were ultra-finely grinded, crushed, and then washed till no blood remained. After that, it was submerged in water and the pH was adjusted to 3.5 using glacial acetic acid for an entire night. Then collect the supernatant and put in a dialysis bag and desalted to reach a neutral pH, and the liquid was concentrated with a rotary evaporator. The precipitates were removed by centrifugation, and the supernatant was then collected and preserved using a lyophilizer for freeze-drying. We then diluted the PAEs lyophilized powder to different drug concentrations using a gel prepared with CMC-Na, glycerol, and pure water.

The pilose antler tissue samples were ground and crushed by adding liquid nitrogen, and then the lysis solution was added to make the cells fully lysed, and the RNA was extracted and purified. The quality of total RNA was checked using an Agilent 2100 Bioanalyzer. Then the mRNA with polyA tail was enriched by magnetic beads with OligodT, and the obtained RNA was fragmented by interrupting the buffer, and then the cDNA was synthesized by reverse transcription with the addition of random N6 primers and modified. Then PCR amplification was carried out by the added specific primers, and after the PCR products were heat denatured into single stranded, they were looped with a piece of bridge primer to obtain a single-stranded circular DNA library, and finally up-sequencing was carried out.

Reads with low quality, splice contamination, and excessive unknown base N content are then filtered out of the raw data from sequencing, and the filtered data are called clean reads. The clean reads were de novo assembled using Trinity, and then clustered and de-redundantly assembled transcripts were clustered and de-redundantly assembled using Tgicl to obtain the Unigene. Bowtie2 (http://bowtie-bio.sourceforge.net/) was applied to align the clean reads to the Unigene, and RSEM (http://deweylab.biostat.wisc.edu/) was used to calculate the gene expression levels in the pilose antler samples. Finally, we integrated the network pharmacology approach and de novo transcriptomics data of pilose antler to speculate on the potential mechanism of PAEs for AGA treatment.

Using Androgenetic alopecia as the keyword, we collected AGA disease-related gene information in DisGeNET database (https://www.disgenet.org/) and GeneCards database (https://www.genecards.org/). Then the jvenn (http://www.bioinformatics.com.cn/static/others/jvenn/example.html) was applied to take the intersection analysis between pilose antler transcriptomics data and AGA disease targets to obtain the key targets of PAEs for AGA treatment.

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses of key targets of PAEs for AGA treatment were performed using the Bioinformatics online tool (https://bioinformatics.com.cn). The GO enrichment analyses included biological processes (BP), cellular component (CC), and molecular function (MF), while the KEGG enrichment mainly showed pathway-related content.

The 6-week-age C57BL/6 male mice were housed in the environment of a 12-h cycle of light and dark, with an ambient temperature of 20°C–24°C and 45%–65% humidity. Electric hair scissor (Codos, Shenzhen, China) and commercial hair removal gel (veet, Reckitt Benckiser) were used to gently depilate the dorsal hair.

Seven groups of six-week-old mice were randomly assigned: Control group (CON), TES-induced AGA model group (TES-group), PAEs gel group (Concentration:1, 2, 4, 8 mg/mL), and positive control group (MIN).

Existing studies have shown that topical application of TES can mimic the pathological characteristics of AGA (Zhang et al., 2016; Lin et al., 2019; Yuan et al., 2021; Ding et al., 2024). Compared to intraperitoneal and subcutaneous injection methods, topical application of TES reduces stress and pain in mice while creating a localized high concentration of hormone environment on the dorsal skin. The AGA mouse model was established and presented with a few minor adjustments from Zhang in this work (Zhang et al., 2016). In brief, the TES (Nature standard, ST053801, Shanghai, China) solution (1 mg/mL) was prepared in 75% ethanol solution. From the second day, mice in each group were applied topically to the back with 150 μL TES solution for 26 straight days except the CON group. Mice in the CON group were received 75% ethanol solution. Subsequently, 100 μL of PAEs and MIN were applied topically after 2 h of TES treatment. The PAEs group received a mixture of PAEs gel at a concentration of 1, 2, 4, and 8 mg/mL, while the MIN group received 5% MIN.

We observed the darkness of the mice’s skin color daily which indicated the hair cycle conversion. The occurrences of skin darkening and hair growth were recorded on days 1, 4, 6, 8, 12, 18, 22, and 26 post-topical application. Digital photographs were taken at these time points, and the level of hair growth and pigmentation was quantified using ImageJ software (Son et al., 2018) in a specific region of the back skin (3.0 × 4.0 cm). Meanwhile, the weight of the mice was recorded at each time point.

Micro-characteristic changes in AGA mice were observed in a representative area of dorsal skin under ×60 objective using medical electronic dermoscopy (MED) (BN-WG-1001, Nanjing). Furthermore, the morphology of hair subtype and hair shaft length were recorded through 180 × and 160 × objective respectively.

Mice’s dorsal skin and HFSCs were collected for immunoblotting analysis, and proteins were separated using a polyacrylamide gel (GenScript, M00654), and then they were moved onto a polyvinylidene fluoride membrane (Millipore, ISEQ00010). The appropriate primary monoclonal antibody was incubated on the membrane for a whole night at 4 °C after blocking with blocking solution (NCM Biotech, P30500), then exposed to a secondary antibody that had been coupled to horseradish peroxidase. Using enhanced chemiluminescence (Millipore, WBLUR0500) substrates and an imaging system (BIO-RAD, ChemiDoc), protein bands were identified, and quantified by ImageJ software. Every example blot’s optical density was calibrated using internal control proteins (Gapdh, Proteintech, 10494, 1:5000). The antibodies dilutions were as followings: AKT (rabbit, Abclonal, AP1208, 1:1000), p-AKT-Ser473 (rabbit, Abclonal, AP0637 1:1000), p-PI3K (rabbit, Abcam, ab182651, 1:2000), PI3K (rabbit, Abcam, ab227204 1:2000), p-GSK3β-Ser9 (rabbit, Abclonal, AP1088, 1:2000), GSK3β (rabbit, Abclonal, A11731, 1:2000), β-catenin (rabbit, Abclonal, A19657, 1:2000), WNT3A (rabbit, Abclonal, A0642, 1:2000), WNT10B (rabbit, Abclonal, A16717, 1:2000).

Total RNA was obtained using the extract kit. Subsequently, the Master Mix (TOYOBO, FSQ-201) was used to make the mRNA become complementary DNA (cDNA), and the resultant cDNA was used as the template. The 40 cycles in the thermal cycling program were set following the instructions. Using the 2^−ΔΔCt technique, relative mRNA expression levels in each sample were measured by normalizing to Gapdh. The sequences of primer are displayed in Table 1.

Mice’s dorsal skin was harvested on days 1, 4, 6, 8, 12, 18, 22, and 26. The samples were preserved in 4% paraformaldehyde for a full day, then dehydrated and paraffin-embedded to produce 4 μm slices. Following deparaffinization using xylene and rehydration through varying ethanol concentrations, then stained using the kit (Solarbio, G1120). Microscopic observation and image acquisition were performed utilizing a microscope (Olympus, CKX41). The Image-Pro Plus software was used to measure the skin and fat layer’s thickness, HFs’ length, and hair bulb’s diameter.

For immunofluorescence analysis, back skin tissues from mice at day 12 were embedded in OCT compound (Tissue Tek, 4583, American), and 10 μm frozen sections were made using a freezing microtome (CryoStar NX70, Epredia, China). These sections were permeabilized for half an hour, then treated for 1 h with immunofluorescence blocking solution (5% BSA in TBS), and with primary antibodies and secondary fluorescence-conjugated antibodies: anti-chicken (Bioss, bs0310G, 1:500), anti-rabbit (Abcam, ab150077, 1:500) at RT before DAPI staining (Yuanye, Shanghai, S19119). Images were captured by confocal microscope (Olympus, FV1000MPE). The dilutions of the antibodies were as follows: KI67 (rabbit, Abcam, ab15580, 1:500), Cytokeratin 15 (K15) (rabbit, Abcam, ab52816, 1:200), Cytokeratin 14 (K14) (chicken, Biolgend, 906004, 1:100), β-Catenin (mouse, Abclonal, A19657, 1:200), p-AKT-Thr308 (rabbit, CST, 13038S), CD31 (R&D Systems, AF3628). We conducted a statistical analysis of CD31-labeled blood vessels around DP following Deng et al. (2022).

Rat primary immortalized HFSCs were purchased from Guangzhou Chuyuanyun Medical Technology Co. The cell culture medium was consisted of 2% FBS, 1% penicillin-streptomycin-amphotericin B solution, and 1% culture additives, and the cells were cultured in a humid environment at 37°C with 5% CO₂.

The effect of PAEs on HFSC proliferation at different drug concentrations was tested by Cell Counting Kit 8 (CCK8; Biosharp, China). Cells were inoculated in 96-well plates at a density of 5 × 10³ cells/well. The next day, the culture was continued for 24h after adding PAEs or PAEs plus inhibitors (IWP-2 and API-2 block the Wnt pathway and AKT activity, respectively), and then the CCK8 solution was added to detect the absorbance.

Cell clone formation assay was used to evaluate the cell proliferation and clone-forming ability of HFSC after PAEs treatment with or without AKT or Wnt pathway inhibitors. Cells were inoculated in 6-well plates at a density of 500 cells/well, and inhibitors (IWP-2 and API-2) and PAEs were treated alone or jointly the next day and cultured continuously for 14 days. Then the cells were fixed with 4% paraformaldehyde and then stained with crystal violet solution. We observed the cells under a microscope and took photographs, defining cluster of more than 50 cells as a cell colony. We ensure that each counted cluster contains over 50 cells and consistently apply this method across all colonies to maintain uniformity and accuracy in the experiment. Using ImageJ software, we counted the colonies and calculated the colony formation rate.

Graph Prism 9 was used to determine the statistical differences in One-way ANOVA analysis. The data were expressed as mean ± standard deviation (SD). The caption for every figure provides information on the number of animals and p values. p < 0.05 was considered significant.

Beginning in the seventh week of postnatal life, the mice enter the second telogen phase of hair growth on the back skin, which lasts around 6 weeks (Müller-Röver et al., 2001). To confirm PAEs have an impact on hair development, we carried out a pre-experiment. PAEs gel was applied at concentrations of 1, 2, 4, and 8 mg/mL to the back of AGA mice at the seventh week of postnatal life. We found PAEs promoted hair growth in AGA mice, with the greatest benefit occurring at the concentration of 8 mg/mL, which was equivalent to the effect of the positive control MIN group (Supplementary Figures S1A, B). These observations indicated that PAEs have a hair growth-promoting effect on AGA mice. 8 mg/mL PAEs was used in subsequent experiments to reveal the underlying mechanism of hair regrowth ability in AGA mice.

We then observed a complete hair cycle of hair growth in AGA mice. On postnatal day 49, 8 mg/mL PAEs gel was applied for 26 days in a row to test the change of PAEs in AGA mice’s hair cycle (Figure 1A). Skin color could be a biomarker to determine the stage of the HF cycle because the activation of the new anagen phase was indicated by a considerable darkening due to pigmentation in HFs during the anagen phase (Plikus and Chuong, 2008). As illustrated in Figure 1B, the skin color in PAEs group changed from pink to gray at day 8, suggesting the mice began to enter the anagen. In contrast, the TES group remained pink. On day 12, the PAEs-treated group’s skin color was noticeably darker than that of the TES group, and nearly identical to the MIN group. On day 18, the PAEs group’s mice skin tone darkened to a greater extent than that of the AGA mice. From day 18 to day 22, the mice’s skin color started to change from dark to gray and pink to different extents, suggesting the mice began to enter the catagen, while the variability of PAEs group was slighter than the other three groups. In addition, the same trend also applies to hair density (Figures 1C, D). On day 26, almost all mice are restored to telogen. Meanwhile, each group showed the same trend of body weight changes under different time points (Figure 1E), indicating topical administration is harmless to mice. These findings indicated that PAEs could stimulate hair regrowth by accelerating the anagen and delaying catagen in AGA mice.

To investigate the histological changes of AGA mice after PAEs treatment, dorsal skin was collected on days 1, 4, 6, 8, 12, 18, 22, and 26, and examined histologically (Figure 2A). Five indicators were assessed (Figures 2B–F): the skin and fat layer’s thickness, the HFs’ length, the hair bulb’s diameter, and the percentage of HFs in the specified hair cycle phases. Based on the previously described classification, the hair cycle progression can be analyzed, on days 4, 6, 8, and 12, the HF in TES + PAEs and CON groups was more than those in the TES group. Additionally, AGA mice had earlier regression from day 18 to day 22 in comparison to the other groups. The histological pathology data also supported the notion that PAEs regulated the hair cycle by activating the anagen and delaying the entry of catagen in AGA mice. Furthermore, when compared to the CON and PAEs-treated groups, the AGA mice at these periods exhibited considerably thinner skin and fat layers as well as smaller hair bulbs. Taken together, these findings indicate that PAEs regulates the hair cycle by alleviating TES-induced thinner skin and fat layer, hair bulb diminishes, and HF shortened in AGA mice.

We also estimated the effect of HF counts after PAEs treatment in AGA mice, The changes in the HF number of each group in transverse sections were assessed on days 1, 4, 6, 8, 12, 18, 22, and 26. It is apparent that PAEs raised the HF numbers in comparison to those in AGA mice (Supplementary Figures S2A, B). Furthermore, it has been reported that the hair morphogenesis in AGA mice can be affected, leading to a thinning of the hair shafts (Fu et al., 2021). We also found that the ratio of the hair shaft diameter of guard hair to the inner root sheath (IRS) diameter increased in the TES + PAEs group than the TES group (Supplementary Figures S1C, D), indicating PAEs alleviated the hair thinner in AGA mice.

To confirm our findings and validate our theory, the MED was used to observe the micro-characteristic changes of dorsal hair, each hair subtype’s morphology, and hair shaft length in each group. From the images captured by MED, it’s clearly found that the PAEs could alleviate the reduction in number and thickness in AGA mice, especially on days 12, 18, and 22 (Figure 3A). Furthermore, skin color observed from the MED images reflects the variation of the hair cycle, which also supported our above findings to some extent.

Additionally, the mouse hair coat is made up of four HF subtypes (guard, awl, auchene, and zigzag). To explore how PAEs affect hair substyles in AGA mice, the dorsal hair was collected on day 26. The shorter hair shafts of AGA mice are evident, particularly in the guard and auchene hair (Figures 3B, C). Moreover, in four groups presented with rows of medulla cells, the thickness of all subtypes was markedly thinner morphologically in AGA mice (Figures 3D, E). These findings suggest that PAEs regulates the process of hair regeneration, hence mitigating the variations of four subtypes’ hair in AGA mice.

The process of hair formation and morphogenesis (such as thickness and length) requires the participation of multiple HF types of cells (Saxena et al., 2019). HFSC plays an important role in HF regeneration and hair cycle regulation, its interaction with hair bulb cells and ORS cells is also crucial for the formation of HF structure (Fuchs, 2007). In this study, we used K15 (Figure 4A) as the HFSC marker in the bulge of HF, Ki67 (Figure 4B) as the cell proliferation marker in the hair bulb, and K14 (Figure 4C) as the cell marker in the ORS (Coulombe et al., 1989; Inoue et al., 2009). The immunofluorescence staining results of mouse skin at day 12 indicated that the count of Ki67⁺, K15⁺, and K14⁺ cells was increased in the PAEs group compared to AGA mice (Figures 4D–F). Taken together, these findings reveal that PAEs reverses the decrease in HFSCs, ORS cells, and hair bulb cells in AGA mice.

Network pharmacology integrates information from public databases to analyze the overall mechanisms of drug action and predict its effects on diseases. Since PAEs have the effect of promoting hair growth in AGA mice, we conducted de novo transcriptomic analysis of pilose antler, and combined with AGA disease targets from public databases, predicted the target genes of pilose antler for treating AGA.

The analysis results indicated that a total of 13065 component targets were collected from PAEs transcriptomics data, and then 504 AGA-related targets were collected from the GeneCards database and DisGeNET database. Utilizing the jvenn method, we identified the intersection of disease-related targets and PAEs targets, resulting in 130 potential targets for PAEs in treating AGA (Figure 5A). Subsequently, these 130 core targets underwent GO and KEGG enrichment analysis.

From the KEGG analysis, we can see some growth factors such as transforming growth factor-β1/2 (TGF-β), IGF1, VEGF, FGF2/7, and HGF, suggesting that the therapeutic effect of PAEs on AGA may be related to the function of the above growth factors (Supplementary Figures S3).

From the cellular component (CC) section of the GO analysis entries, it can be seen that the β-Catenin-TCF complex plays a significant role in biological regulation (Figure 5B). Meanwhile, The KEGG pathway analysis revealed the top 16 enriched signaling pathways, prominently including MAPK, Ras, PI3K-Akt, and Wnt signaling pathways (Figure 5C). Among the four signaling pathways, it has been shown that MAPK (Akilli Ozturk et al., 2015), PI3K-AKT (Liu et al., 2021) and Wnt signaling pathways (Veltri et al., 2022) play important roles in the regulation of hair growth and growth cycle. The original KEGG pathway maps for Wnt-Catenin and the PI3K-AKT pathways were in Supplementary Figures S4A, B. Considering that the PI3K-AKT, and Wnt signaling pathways are interconnected by GSK3β, we then tested whether these two signaling pathways were altered when investigating the mechanism of PAEs promoting hair growth in AGA mice, aiming to attempt to verify this potential mechanism.

The Wnt/β-catenin pathway plays a critical role in HF development (Deschene et al., 2014; Jin et al., 2021). To determine if PAEs might affect this pathway, the mRNA level of up-and downstream genes of Wnt/β-catenin was examined, including Axin1, Axin2, Lef1, Tcf7, c-Jun, Lgr6, and Survivin (Figures 6A–G), and protein level of β-catenin (Figures 6H–J, M) and its two ligands (WNT3A and WNT10B) (Figures 6J–L) in back skin of each group. It was found that the above-mentioned genes and the proteins of β-catenin, WNT3A, and WNT10B were dramatically upregulated by PAEs in AGA mice, suggesting that this pathway could be triggered by PAEs.

GSK-3β could adversely regulate β-catenin, and the inhibitory phosphorylation site GSK-3β at Ser9 could be stimulated by AKT and then induce a cascade reaction to produce β-catenin (Zhang et al., 2019). Meanwhile, the stimulation of the PI3K/AKT pathway is necessary for new HF generation which is mediated by skin-derived precursors and epidermal stem cells (Chen et al., 2020). Therefore, the p-PI3K, p-AKT, and p-GKS-3β-Ser9 protein levels were tested in AGA mice after PAEs treatment. As illustrated in Figures 7A–D, PAEs reversed the suppression of these protein levels significantly. This suggests that PAEs may have a promoting effect on the accumulation of β-Catenin by activating the PI3K-AKT pathway. Furthermore, the PI3K/AKT pathway plays an essential role in angiogenesis (Karar and Maity, 2011). The new blood vessels surrounding DP provide nutrients to support hair growth, Studies have reported that under the influence of androgen receptors, the blood vessels surrounding the DP degenerate, resulting in a decrease in their number (Deng et al., 2022). We found the PAEs perhaps increased the number of blood vessels around the DP (Figures 7E, F) in AGA mice.

The detection of protein and gene changes related to both the PI3K-AKT and Wnt-β-Catenin pathways in AGA mice validated our speculation, suggesting that the treatment of PAEs in AGA mice is closely related to the AKT and Wnt pathways. Then We also examined the link between the activation and proliferation of HFSC in vitro and the AKT and wnt pathways after PAEs treatment. HFSC is closely related to the regulation of the HF cycle and the promotion of the formation of HF structures such as the hair shaft and the IRS (Lin et al., 2020). Meanwhile, the AKT and Wnt pathways have an important role in promoting the proliferation and differentiation of epidermal stem cells and hair regrowth (Wang et al., 2017). We found that PAEs at 250–1000 μg/mL had a significant pro-proliferative effect on HFSC, with a proliferation rate of 150.4% at 1000 μg/mL (Figure 8A). In addition, the protein level of p-AKT, p-GSK3β-Ser9, and β-Catenin was obviously increased in HFSC after PAEs treatment (Figures 8B–E).

To verify whether the activation of PAEs on HFSC is mediated through the AKT and Wnt pathways. We first detected the proliferative effect of HFSC after blocking the AKT activity and Wnt pathway. API-2 is a specific AKT inhibitor, mainly by suppressing the AKT phosphorylation at activation sites (Thr308 and Ser473). IWP-2 is an inhibitor of Wnt pathway. From the results of immunofluorescence of KI67 (Figures 9A, B) and clone formation experiments (Figures 9C, D), it can be seen that the proliferation and clone formation ability of HFSC was significantly enhanced after PAEs treatment. However, this effect was significantly reversed after blocking AKT activity and the Wnt pathway in HFSC. And the proliferative activity of HFSC in the PAEs + IWP-2/API-2 group could not be restored. It was suggested that the promotion of HFSC proliferation by PAEs might be closely related to the activation of the AKT and Wnt pathways.

To further validate our conjecture, we intervened with PAEs in HFSC that blocked AKT activity and Wnt pathway, and detected the level of relevant proteins and genes in the AKT and Wnt pathways. As can be seen from the results of immunofluorescence (Figures 9E, G) and Western blot (Figures 9F, H, I), the protein levels of p-AKT at both sites (Thr308 and Ser473) and p-GSK3β-Ser9 showed an obvious trend to increase after PAEs treatment in HFSC. However, the protein levels of p-AKT and p-GSK3β-Ser9 decreased significantly after blocking the AKT activity and did not recover after PAEs intervention. We also tested the mRNA levels of up-and downstream genes in the Wnt pathway, such as Survivin, LEF1, Axin1, Axin2, Lgr6, Tcf7, and Ccnd1 (Supplementary Figures S5A) and the levels of β-Catenin, WNT3A and WNT10B (Supplementary Figures S5B–G), then found these gene expressions increased in HFSC after PAEs treatment. However, the expressions of these genes were obviously suppressed after blocking the Wnt pathway, and the expressions did not recover under the PAE treatment. These results showed that the proliferation of HFSC induced by PAEs may be closely related to the activation of the AKT and Wnt pathways.