High-throughput screening of small molecule Wnt/β-catenin activators for hair and nail growth

Abstract

Hair and nails, though structurally different, both rely on Wnt/β-catenin signaling for regeneration. To identify small molecules that enhance this pathway, we screened 5,170 FDA-approved and natural compounds using HEK293 cells with a TCF/LEF luciferase reporter. Nine compounds significantly activated Wnt signaling, including antivirals (imidocarb, proflavine, aminoacridine), anticancer agents (entinostat, enzastaurin, abemaciclib), and GSK-3β inhibitors (BIO, CP21R7). Aminoacridine and proflavine markedly upregulated RSPO3 and RSPO4 while suppressing WIF1. In mice, aminoacridine—especially combined with minoxidil—showed the strongest hair regrowth, while proflavine and imidocarb enhanced nail elongation. These results reveal new therapeutic candidates for hair and nail regeneration.

High-throughput screening identifies FDA approved and natural compounds that activate Wnt/β-Catenin signaling, promoting hair and nail regrowth through cell-based and animal models, with potential for repurposing in regenerative therapy.

Introduction

One study on hair follicle (HF) and nail unit has revealed intriguing similarities and differences from both phenotypic and genotypic perspectives, reflecting the complexity and specificity of human integumentary structures¹. Phenotypically, both hair and nails primarily serve a protective function, with hair shielding the scalp and nails safeguarding fingertips. This protective role is largely facilitated by the production of hard keratin, a commonality hinting at a shared evolutionary pathway^2,3. However, hair and nails show significant differences in structure, with hair characterized by long, twisted fibers offering flexibility and nails distinguished by their flat layers of keratin conferring rigidity⁴.

In the case of the second genotype, similarities and differences in gene expression related to hair and nails offer a fascinating area of study. Both hair follicles and nails are primarily regulated by the Wnt/β-catenin signaling pathway³. However, expression levels of specific genes within this pathway show considerable variations, reflecting their distinct morphological characteristics⁵. For instance, the R-spondin family of proteins, particularly RSPO3 and RSPO4, play a significant role in these structures. RSPO3 is more associated with hair follicle development, while RSPO4 shows a stronger link to nail growth^6,7.

Such differential expression suggests a fine-tuning of the Wnt pathway, where certain genes are upregulated in one structure but not the other, leading to unique phenotypic traits of hair and nails. Previous studies have shown that single-cell analysis of hair follicle-related genes and spatial transcriptome analysis of tissues surrounding hair follicles can help discover targets for drug development and studying mechanisms⁷. For example, genes such as LGR5 and LGR6 have been implicated in hair follicle stem cell maintenance and activation, whereas their role in nail physiology is less clear. Similarly, specific keratin genes exhibit distinct expression patterns in hair and nails, contributing to their different structural properties⁸.

Understanding these genetic nuances provides critical insights into the molecular basis of skin appendage development and offers potential therapeutic targets for conditions affecting hair and nails. Recently, advances in hair restoration research have led to the development of techniques focused on the collection, culture, and proliferation of follicular dermal papilla (DP) cells. These DP cells play a critical role in hair follicle formation and growth. Cultured DP cells can be used to create hair germ, an early differentiated form of hair follicles, which can then be transplanted to promote hair growth. This approach represents a promising avenue in treating hair loss and baldness^9–11.

However, challenges remain in this process. Notably, it has been observed that expression levels of activation genes in dermal papilla cells are significantly decreased when these cells are grown in cell culture conditions¹². Furthermore, as these cells undergo successive passages in culture, the expression of these critical activation genes shows a progressive decline^13,14. This reduction in gene expression correlates with a decreased efficacy of establishing hair follicles after transplantation. In other words, when DP cells are cultured longer and passaged more, they are less effective in generating functional hair follicles post-transplantation¹⁵.

This phenomenon highlights a key hurdle in hair follicle bioengineering: maintaining intrinsic hair-inducing properties of dermal papilla cells throughout an in vitro culture process. Overcoming this challenge is essential for improving the success rate of hair follicle transplantation and offers a significant area for further research and development in the field of regenerative medicine for hair loss.

Building on the foundation of previous genetic analysis of hair follicle and nail unit bioinformatics, our research team has embarked on a dual-faceted screening process using High-Throughput Screening (HTS) technology. This approach is designed to identify potential Wnt agonists and antagonists, focusing on compounds from a comprehensive library of 5170 natural products and FDA-approved drugs. Our primary objective is to find molecules capable of activating the canonical Wnt/β-catenin signaling pathway, which is central to the development and growth of hair follicles and nails, as shown in Fig. 1.

Fig. 1. Schematic diagram of the research process from high-throughput screening to the discovery of a leading compound.

a Wnt activity and cytotoxicity of drugs assessed through TCF/LEF reporter assay for a library of 5170 drugs. b Measurement of Wnt-related RNA expression level and increase in hair embryo length as a secondary efficacy test for candidate drugs. c Selection of lead compound by evaluating drug efficacy in the hair and nail growth observation model.

On the one hand, we focused on identifying Wnt agonists that could simultaneously activate canonical Wnt signals and the R-Spondins/LGR pathway. This could potentially lead to more effective treatments for hair and nail growth by harnessing the synergistic effects of these pathways. On the other hand, our research also aimed to systematically discover Wnt antagonists that could modulate the Wnt/β-catenin signaling pathway, thus influencing gene expression relevant to hair and nail growth.

Further evaluations of the efficacy of these compounds were conducted through both in vitro and in vivo tests. In vitro tests allowed us to analyze molecular and cellular responses to these compounds in a controlled environment. Subsequently, we plan to conduct animal experiments to observe direct effects of these Wnt modulators (both agonists and antagonists) on hair and nail development.

These in vivo studies are crucial for understanding the physiological impact of Wnt modulation in complex biological systems. By integrating HTS, in vitro, and in vivo methodologies, we aimed to unravel the complex genetic and molecular underpinnings of hair and nail biology, potentially leading to innovative treatments for various conditions related to these integumentary structures.

Materials and methods

Primary high-throughput screening

For primary high-throughput screening, HEK293 cells stably integrated with the firefly luciferase gene under the control of TCF/LEF response elements (Cat. #60501, BPS Bioscience, San Diego, CA, USA) were utilized. These cells served as a model to evaluate the activation of Wnt signaling. The screening targeted two major libraries: an FDA-approved drug library containing 2570 compounds (Cat. #L1300, Selleckchem, Houston, TX, USA) and a natural product drug library comprising 2600 compounds (Cat. #L1400, Selleckchem, Houston, TX, USA).

HEK293 cells were cultured in MEM medium (Cat. #SH30024.01, Hyclone) supplemented with 10% Fetal Bovine Serum (FBS) (Cat. #S1480, Biowest), 1% non-essential amino acids (Cat. #SH30238.01, Hyclone), 1 mM sodium pyruvate (Cat. #SH30239.01, Hyclone), and 1% Penicillin/Streptomycin (Cat. #SV30010.01, Hyclone). These cells were maintained at 37 °C in a humidified atmosphere containing 5% CO₂.

For the screening, 3×10³ cells per well were seeded into a 384-well plate. After an overnight incubation to allow cell attachment, each compound from the libraries was diluted with the medium to achieve a final concentration of 5 µM in the well. Dimethyl sulfoxide (DMSO) was used as the solvent for each compound before dilution with the medium. Each drug/compound was used for treatment in duplicate for 24 hours. Lithium chloride (LiCl) at 10 mM concentration was employed as a positive control.

To assess both cell viability and luciferase activity post-drug treatment, the ONE-Glo™ + Tox Luciferase Reporter and Cell Viability Assay (Cat. # E7120, Promega) was used following the manufacturer’s instructions. A Mithras multi-reader (Cat. #LB940, Berthold Technologies, Bad Wildbad, Germany) was utilized for measurements. Drugs were selected based on the criterion that the average of two luciferase measurement values per drug was more than five times higher than that of the control group treated with the vehicle (0.1% DMSO). From this screening, seven drugs that showed activity consistently on three or more separate occasions were chosen for further analysis.

Secondary biological in vitro assay

RT-PCR

Human hair follicle dermal papilla cells (Promocell, Heidelberg, Germany, used at passages 2–5) were seeded into 60 mm dishes at a density of 3 × 10⁴ cells/well with DMEM high glucose serum medium supplemented with 10% FBS and 1% antibiotic-antimycotic solution. Cells were treated with seven candidate drugs at 5–10 µM, which was a high concentration of β-catenin in the nucleus, and cultured for 24 hours.

Total RNA was extracted using the RNeasy mini kit (Qiagen, Valencia, CA, USA). Template cDNAs were obtained by reverse transcription of total RNAs using an oligo (dT) primer and a PrimeScript™ RT reagent Kit (TAKARA, Japan). Amplification was carried out using SYBR® Green Realtime PCR Master Mix (Toyobo, Japan). Sequences of real-time reverse transcription-polymerase chain reaction (RT-PCR) primers used in this study are listed in Supplementary Table 1. They were designed following a previous study. Expression levels of target genes were normalized against the expression of β-actin, a housekeeping gene.

Hair and nail germ formation and elongation assay

For the formation and elongation assay of hair and nail germs, primary skin dermal fibroblasts, skin epidermal keratinocytes, hair follicle dermal papilla cells, and nail fibroblasts (onychofibroblasts) were obtained from patients under ethical considerations. Cells from only 1–5 passages were utilized in all experiments. This part of the study was performed after obtaining approval from the Institutional Review Board (IRB) of Samsung Medical Center, ensuring compliance with ethical research standards (IRB number:2021-12-134). Informed consent was obtained from all participating patients following ethical research guidelines.

In the experimental setup, spheroids were created. For each spheroid, precise counts of 2000 fibroblasts and 2000 keratinocytes were used. These cells were mixed with 60 µl of DMEM supplemented with 10% FBS. The cell mixture was then placed in U-Shaped-Bottom Microplates (Thermo Fisher, Waltham, MA, USA) designed for spheroid formation. Drug compounds used for the assay were sourced from Selleckchem (Houston, TX, USA). Each compound was used at a final concentration of 5 µM.

Spheroids were initially cultured and maintained for a period of 2 days to establish their form. Following this incubation period, drug compounds were administered to these spheroids, which were then further cultured for an additional 5–7 days. This duration was chosen to allow sufficient time for drugs to exert their effects on spheroids.

To quantify the elongation of each spheroid, images were obtained using an inverted microscope (CKX53, Olympus, Shinjuku, Tokyo, Japan). Obtained images were measured using ImageJ software (National Institutes of Health, Bethesda, MD, USA). For statistical robustness, more than 10 spheroids per experimental group were measured. To ensure the reliability and reproducibility of results, the entire experiment was repeated twice. This methodical approach allowed for a comprehensive assessment of the impact of drug compounds on hair and nail germ formation and elongation.

To measure germ elongation, we used the maximum Feret diameter from 2D images obtained via microscopy. ImageJ software (NIH, Bethesda, MD, USA) was used for analysis. This approach was chosen because the boundaries between fibroblasts and keratinocytes were not consistently distinguishable across all spheroids. The Feret diameter served as a standardized and reproducible measure of length.

Animal model and surgical procedures

Anagen induction assay with C57BL/6 mice

For the anagen induction assay, 56 male C57BL/6 mice aged 7 weeks weighing between 20 g and 22 g were procured from Orient BIO (Seoul, Korea). Prior to the commencement of experiments, these mice underwent an acclimatization period of one week under laboratory conditions. The ethical aspect of this study was rigorously adhered to, with all mouse experiments receiving approval from the Samsung Medical Center Institutional Animal Care and Use Committee (Approval code: 20210617001). Procedures were conducted in accordance with the AAALAC International Guidelines and the United States Animal Welfare Act to ensure humane and ethical treatment of animals.

Mice were housed under controlled conditions, including a 12-hour light/dark cycle, a regulated temperature, and a humidity range of 35–60%. They were provided with standard mouse food and water. After the one-week adaptation period, 8-week-old mice were ready for experimental procedures.

For the assay, all mice were anesthetized using isoflurane. Their dorsal areas measuring approximately 2 cm in width and 4 cm in length were shaved using hair clippers to prepare the skin for treatment application. These mice were then randomly divided into eight groups (seven mice per group): a vehicle group, an MXD (MXD) group, an imidocarb (1%) group, a profavine (1%) group, an abemaciclib (1%) group, an aminoacridine (1.0%) group, an aminoacridine (0.3%) + MXD (0.3%) group, and an aminoacridine (1.0%) + MXD (0.3%) group. The vehicle group received a phosphate buffered saline (PBS) solution without any active treatment.

For drug preparation, aminoacridine was dissolved in a mixture of propylene glycol (10%, v/v) and ethanol (60%, v/v) to achieve a concentration of 1% (w/v). MXD was similarly dissolved in a mixture of propylene glycol and ethanol to have a final concentration of 0.3% (w/v). Each group was treated with 200 µL of the respective solution, which was applied topically to shaved areas of mice once daily for 28 days^15,16.

At the conclusion of the 28-day treatment period, the efficacy of each treatment was assessed and mice were humanely euthanized using CO₂ gas. This protocol allowed for a comprehensive evaluation of the potential of each compound to induce anagen phase in hair follicles, providing valuable insights into their effectiveness for hair growth promotion.

Nail growth assay with C57BL/6 mice

In the nail growth assay, six male C57BL/6 mice per group were used to evaluate the efficacy of various drug treatments on toenail growth. For each mouse, three nails from the middle digit of the forepaw were measured. Due to occasional limitations in CT imaging angles, not all nails could be reliably assessed; however, a minimum of 11 nails per group were included in the final analysis.

To minimize the risk of bone damage or digit tip regeneration, toenails were trimmed one day prior to drug application and the initial CT scan. After trimming and the baseline scan, mice received treatment by immersing their forepaws in a solution containing 0.3% (w/v) of the test compound in 30% ethanol for 10 seconds, five times per week for 14 days.

CT scans were performed on day 0 and day 14 to monitor changes in toenail length. For accurate and reproducible quantification in the CT images, specific Hounsfield Unit (HU) parameters were applied. The tip of the toenail was identified by adjusting the Window Level (WL) to 287 and the Window Width (WW) to 332. To locate the first toe joint, the settings were adjusted to a WL of 693 and WW of 436. These calibrated settings enabled precise and consistent measurements across all samples.

Image analysis

To assess the efficacy of treatments in promoting hair growth, systematic image analysis was conducted for hair loss sites of mice. Photographs of these sites were captured at specific intervals: 1, 2, 3, and 4 weeks following the commencement of each treatment. These time points were chosen to provide a comprehensive view of the hair regrowth process over the course of treatment.

For the analysis of hair regrowth, a scoring system was implemented. This scoring was based on the percentage of hair regrowth observed in the treated area, allowing for a quantifiable and standardized assessment of the treatment’s effectiveness. Scoring criteria were as follows: score 0, no hair regrowth observed; score 1, 0–20% hair regrowth; score 2, 20–40% hair regrowth; score 3, 40–60% hair regrowth; score 4, 60–80% hair regrowth; and score 5, 80–100% hair regrowth.

In vivo fluorescence imaging analysis

For in vivo fluorescence imaging analysis, we utilized an IVIS spectrum system (Caliper Life Sciences), a sophisticated tool for capturing fluorescence images of live subjects. This method was applied to evaluate the distribution and penetration of the medication applied topically to mice. In each case, a dose of 100 μl of the medication was applied to the skin on the back of a mouse. To monitor dynamics of the medication’s absorption and distribution over time, fluorescence transmittance was assessed at 5, 60, 120, and 360 minutes post-application. These intervals were strategically chosen to provide a comprehensive view of the medication’s behavior within the first few hours after application.

To ensure consistency and accuracy in the imaging process, all images were acquired under identical field conditions. This uniformity in imaging conditions was crucial for reliable comparison and analysis of the data. Additionally, visibility of the region of interest (ROI) in the fluorescence image was automatically normalized using the Living Image software. This software is integral to the IVIS system, allowing for precise quantification and analysis of fluorescence signals.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Results

High-throughput screening for Wnt agonist discovery and assessment of cytotoxicity

Our HTS process focused on evaluating Wnt-stimulating effects and cytotoxicities of 5170 FDA-approved drugs and natural products using the TCF/LEF promoter assay. Identified Wnt agonists that were effective in three or more repeated experiments were primarily classified into three categories: antibacterial agents (as shown in Fig. 2a–c), anticancer agents (Fig. 2d–g), and GSK3beta inhibitors (Fig. 2h, i). To enable direct comparison, Fig. 2j presents the maximum luciferase expression values for each drug using a standardized y-axis scale.

Fig. 2. Evaluation of Wnt activity level by concentration of Wnt signaling agonists selected through high-throughput screening.

After treatment with (a) Imidocarb, (b) Proflavine, (c) Aminoacridine, (d) Entinostat, (e) Tucidinostat, (f) Enzastaurin, (g) Abemaciclib, (h) CP21R7, or (i) BIO at concentrations of 10 nm, 100 nM, 1 μM, 5 μM, 10 μM, and 100 μM for 24 hours in HEK293 cells, the expression level of luciferase was observed. j Comparison of maximal luciferase expression of (a~i).

Most drugs demonstrated maximum luciferase expression at concentrations between 1 µM and 10 µM. Notably, enzastaurin showed the most significant effect at a higher concentration of 100 µM (Fig. 2f). Regarding the maximum effect of each drug, five drugs (aminoacridine, entinostat, tucidinostat, enzastaurin, and CP21R7) exhibited a more potent effect than 10 mM LiCl, a positive control. In particular, CP21R7 displayed an activity level that was 10 times higher than that of 10 mM LiCl (Fig. 2j).

Among the nine effective drugs, CP21R7 and BIO, both GSK3 beta inhibitors, are already known for their Wnt signal activation effects^17,18. Entinostat, an HDAC inhibitor, has also been recently reported to be able to activate the Wnt signal¹⁹. This HTS led to the discovery of six new drugs as potential Wnt agonists.

In terms of cytotoxicity, there was a noticeable variation depending on the drug class. Antibacterial drugs (Fig. 3a–c) showed minimal toxicities, even at high concentrations (100 µM or more). However, anticancer drugs (Fig. 3f, g) and GSK3beta inhibitors (Fig. 3h, i) exhibited strong toxicities at a concentration of 100 µM or higher. Among anticancer drugs, HDAC inhibitors (Fig. 3d, e) were observed to be relatively less toxic.

Fig. 3. Assessment of cytotoxicity by concentration of selected drugs through a high-throughput screening technique.

a Imidocarb, (b) Proflavine, (c) Aminoacridine, (d) Entinostat, (e) Tucidinostat, (f) Enzastaurin, (g) Abemaciclib, (h) CP21R7, and (i) BIO(6-Bromoindirubin-3’-oxime) of Wnt agonists were used to treat HEK293 cells at concentrations of 10 nm, 100 nM, 1 μM, 5 μM, 10 μM, and 100 μM, respectively. Toxicity was observed after 24 hours of treatment.

Effects of candidate drugs on expression levels of Wnt-related RNAs in human dermal papilla cells

To evaluate the in vitro efficacy of the nine selected drugs, RT-PCR analysis was conducted to observe their effects on the expression levels of Wnt pathway-related genes in hair follicle cells.

Among the R-spondin genes, proflavine and abemaciclib demonstrated strong overall effects on RSPO3 (Fig. 4a) and RSPO4 (Fig. 4b) expression. Additionally, aminoacridine specifically increased the expression of RSPO3, while enzastaurin uniquely boosted RSPO4 expression.

Fig. 4. Effects of candidate drugs on expression levels of Wnt-related RNAs in human dermal papilla cells.

Each drug was used to treat dermal papilla cells at the highest concentration of luciferase expression in the HTS assay. After 24 hours, RNA level changes of (a) RSPO3, (b) RSPO4, (c) WIF1, (d) DKK1, (e) TCF7, (f) LEF1, (d) RGS2, (e) TCF7, (f) SERP2,(g) RUNX1, (h) RUNX2, (i) LEF1, (j) RGS2, (k) ALPL, (l) NOG, (m) LGR4, (n) LGR5, (o) LGR6, (p) BMP2, (q) BMP4, (r) BMP5, and s BMP7 genes were measured using quantitative RT-PCR. The effect of each drug was quantified by correcting the value of β-actin to 1.

With respect to the Wnt inhibitors WIF1 and DKK1, proflavine markedly increased WIF1 expression by 7.7-fold, the highest among all tested drugs. In contrast, DKK1 expression was significantly decreased by proflavine to approximately 3.7-fold, the lowest observed (Fig. 4c, d).

For the Wnt promoters TCF7 and LEF1, proflavine, enzastaurin, and CP21R7 substantially increased the expression of TCF7 (Fig. 4e) and LEF1 (Fig. 4i), with a more prominent effect noted for LEF1. These drugs also significantly elevated the expression of downstream targets of the TCF/LEF pathway, including SERP2 (Fig. 4f), RUNX1 (Fig. 4g), and RUNX2 (Fig. 4h). Moreover, the expression of hair germ activation genes—LEF1 (Fig. 4i), RGS2 (Fig. 4j), ALPL (Fig. 4k), and NOG (Fig. 4l)—remained relatively high following treatment with these compounds.

In the case of LGR receptors for R-spondins, LGR4 (Fig. 4m) showed a modest increase after Wnt agonist treatment, while LGR6 (Fig. 4o) exhibited a 4.7-fold increase following proflavine administration. Notably, LGR5 (Fig. 4p) was exceptionally upregulated—by approximately 150-fold—after proflavine treatment, indicating a strong induction of LGR5 expression among the LGR family.

The expression of bone morphogenetic proteins (BMPs) was analyzed. Contrary to expectations, BMP genes usually showing opposing effects to Wnt signaling exhibited increases after treatment with Wnt agonist drugs. Notably, BMP2 (Fig. 4p) and BMP4 (Fig. 4q) were increased approximately 100 times or more after treatment with CP21R7 and over 30 times after treatment with proflavine. Additionally, proflavine increased expression levels of BMP5 (Fig. 4r) and BMP7 (Fig. 4s) by 5.4 times and 14.6 times, respectively. These results demonstrate varied and complex effects of candidate drugs on the expression of genes involved in the Wnt signaling pathway and related processes in hair follicle cells, providing valuable insights into their potential as therapeutic agents for hair growth.

Effects of various drugs on hair and nail germ length

In our study, a comparative analysis was conducted to evaluate effects of various drugs on hair and nail germ length. Notably, treatment with imidocarb showed the most significant impact, resulting in the highest average hair germ length change of approximately 334.7 ± 35.64 μm. This suggests a strong potential for imidocarb to promote hair germ growth. Other drugs including enzastaurin, MXD, and abemaciclib also considerably increased hair germ length, with average changes recorded at approximately 308.89 ± 25.32 μm, 308.12 ± 33.14 μm, and 300.54 ± 34.85μm, respectively, as shown in Fig. 5. Similarly, in the evaluation of nail germ length, imidocarb treatment again resulted in the most significant growth, with an average increase of approximately 324.99 ± 45.20 μm. Enzastaurin, MXD, and abemaciclib also notably increased nail germ lengths with average changes of approximately 306.36 ± 63.54 μm, 308.13 ± 45.60 μm, and 277.30 ± 15.85 μm, respectively, as illustrated in Fig. 5d. These findings highlight the potential of these drugs, especially imidocarb, in enhancing the growth of both hair and nail germs. Their mechanisms of action and therapeutic implications warrant further investigations.

Fig. 5. Effects of different drugs on hair and nail germ length.

a Spheroids formed from diverse origins (skin, hair, and nail) exhibit varying shapes at different time points. b Hair germ and nail germ length changes over a 7-day period. c–e A figure illustrating effects of different drugs on hair and nail germ length, presenting average measurements under each treatment condition. *The experiment was repeated and the average length (μm) was calculated for each treatment group. Data were normalized by the length of germ on day 7 and plotted as mean ± SD (n = 4). For all symbols conferring statistical significance: one-way ANOVA was used. Data were compared to the vehicle control group. *p < 0.05; **p < 0.01.

Animal model and surgical procedures

Animal model for nail growth

In our study, we concentrated on assessing the impact of various drugs on nail growth by measuring changes in toenail length over a 14-day period. Results from our comparative analysis revealed that MXD treatment was the most effective one, leading to the highest average increase in toenail length of approximately 0.66 mm. This significant growth indicates MXD’s potential as an effective promoter of toenail growth. Other drugs such as imidocarb and proflavine also showed promising results. Imidocarb, in particular, demonstrated an impressive average increase in toenail length of about 0.74 mm, while proflavine exhibited an increase of approximately 0.59 mm. These findings are detailed in Fig. 6e, highlighting the potential of both imidocarb and proflavine as promising agents for enhancing toenail growth. This study thus sheds light on possible therapeutic applications of these drugs for conditions related to nail growth and health, offering valuable insights for further research and development in this area.

Fig. 6. Effects of various drugs on toenail length using CT imaging.

a Pre- and post-nail cutting, (b) precise nail length measurements, (c, d) imaging at time 0 and after a 14-day period, (e) effects of different drugs on nail length after a 2-week duration. Data were normalized to the length of each nail that grew over 14 days and plotted as mean ± SD (n > 11). One-way ANOVA was used to compare with the vehicle group. *p < 0.05; **p < 0.01 for all symbols conferring statistical significance.

Anagen induction assay of Wnt agonists in C57BL/6 mice

To investigate the effects of aminoacridine drugs on mouse hair regrowth, an anagen induction assay was performed with C57BL/6 N mice. MXD was used as a positive control. The dorsal skin of each 8-week-old C57BL/6 N male mouse was shaved for this purpose. For treatments including vehicle, aminoacridine, aminoacridine with MXD, imidocarb, and proflavine, each drug was topically applied to the mouse’s dorsal skin daily for 28 days. Mice were then observed for morphological changes at weekly intervals for 4 weeks, as depicted in Fig. 7.

Fig. 7. Anagen induction assay of Wnt agonists using C57BL/6 mice.

After hair removal, anagen induction was performed on the dorsal skin of each mouse. a Representative images showing hair regrowth progression over four weeks in different treatment groups, including vehicle, imidocarb (1%), proflavine (1%), abemaciclib (1%), aminoacridine (1.0%), MXD (0.3%), aminoacridine (0.3%) + MXD (0.3%), and aminoacridine (1.0%) + MXD (0.3%). b Hair regrowth scores were evaluated weekly (0, no hair regrowth; 1, 0–20% regrowth; 2, 20–40% regrowth; 3, 40–60% regrowth; 4, 60–80% regrowth; 5, 80–100% regrowth). Data were photographed weekly to quantify hair growth scores and plotted as mean ± SD (n = 7). One-way ANOVA was used to compare with the vehicle group. *p < 0.05; **p < 0.01 for all symbols conferring statistical significance.

Observations revealed distinct patterns of hair regrowth across different treatment groups. After one week, the aminoacridine with MXD group uniquely exhibited a darkening of mouse back skin (Fig. 7a), indicating early signs of hair regrowth. By week 2, both aminoacridine and aminoacridine with MXD groups showed early hair growth. Specifically, the aminoacridine with MXD group had a hair regrowth area score of 2.78 ± 0.53, translating to a 29.8% average regrowth area percentage. The MXD group serving as the positive control also developed darkened skin by this time (Fig. 7b). By week 3, the MXD group’s hair regrowth area score was 2.71 ± 0.56 (26% average regrowth area percentage), whereas the aminoacridine with MXD group exhibited a significantly higher score of 3.14 ± 0.50, corresponding to a 49% average regrowth area percentage. The aminoacridine group had a regrowth area score of 2.40 ± 0.40, which was about 22.5% average regrowth area percentage. By the fourth week, both MXD and aminoacridine with MXD groups showed significant hair regrowth, with scores of 3.34 ± 1.86 (54% average regrowth area percentage) and 4.14 ± 0.89 (78% average regrowth area percentage), respectively. Notably, by the end of the fourth week, all experimental groups, including those treated with aminoacridine, imidocarb, and proflavine, demonstrated more pronounced hair growth effects than vehicle control groups.

In vivo fluorescence imaging analysis with C57BL/6 mice

In vivo fluorescence imaging analysis was performed with an IVIS spectrum system to evaluate skin permeability of aminoacridine in mice. As shown in Supplementary Fig. 1 (Fig. S1), aminoacridine was consistently absorbed into the skin irrespective of alcohol concentration in the solution (ranging from 30% to 60%). Notably, around 70% of the drug was absorbed within the first 60 minutes post-application. This absorption rate increased over time, reaching over 90% after 360 minutes.

Statistical analysis and reproducibility

All experiments were repeated at least three times with independent biological replicates. Results from independent experiments were not combined to create a single set of data. All results are expressed as mean ± standard deviation. They were plotted and analyzed using GraphPad Prism version 9 and subjected to one-way analysis of variance (ANOVA) with statistical significance set at p < 0.05.

Discussion

Our study delves into phenotypic and genotypic intricacies of hair follicles and nail units, shedding light on their similarities and differences which are pivotal for understanding human integumentary structures. The challenge in treating alopecia has long been the inability of drugs and hair transplantation techniques to regenerate hair follicles. This has made hair follicle neogenesis a pivotal strategy in alopecia treatment, as highlighted in previous studies^20,21. Despite the recognized role of Wnt signaling in hair follicle biology, few Wnt agonists have made the transition to clinical use for hair loss treatment^22,23. To address this gap, our research focused on identifying new drugs, with a particular emphasis on safe, side effect-free Wnt agonists suitable for broad human application, leveraging both natural products and FDA-approved drugs.

In this study, a high-throughput screening strategy involved treating HEK293 cells with TCF/LEF engineered luciferase using 5170 drugs. It measured the binding degree of introduced β-catenin to the promoter within the nucleus, indicating Wnt pathway signaling activation. This comprehensive screening revealed nine effective drugs: three antiviral drugs, four anticancer drugs, and two GSK-3 inhibitors^16,24,25. Among these nine drugs, CP21R7 and BIO, which are GSK3 beta inhibitors, have been reported to have Wnt signal activation effects. Additionally, entinostat, an HDAC inhibitor, has been recently reported to be able to activate the Wnt signal^26,27.

As a result, 9 types of effective drugs were identified, including 3 antiviral drugs, 4 anticancer drugs, and 2 GSK-3 inhibitors^28–30. Notably, CP21R7 and BIO are GSK3 beta inhibitors known for their Wnt activation effects. Entinostat, an HDAC inhibitor, has also been reported to be able to activate the Wnt signal^15,31. Our HTS process uniquely identified six new drugs acting as Wnt agonists, marking the first extensive investigation in the world with a focus on identifying treatments for hair loss and nail issues that could stimulate the Wnt/β-catenin signaling pathway.

RT-PCR analysis was conducted to assess effects of these nine drugs on expression levels of Wnt pathway-related genes in hair follicle cells. Aminoacridine, abemaciclib, and proflavine notably increased expression levels of RSPO3 and RSPO4 genes, with aminoacridine notably enhancing RSPO3 levels by 19-fold, the highest among tested drugs. The importance of RSPOs in enhancing canonical Wnt/β-catenin signaling and their roles in stem cell biology have been well established^32,33. Moreover, the human genome harbors four genes that encode related RSPO proteins^34–36. Based on their physical interactions, RSPOs were first described as Wnt signaling agonists³⁷. RSPOs collaborate with Wnt proteins to enhance the activity of the canonical Wnt/β-catenin signaling pathway. Notably, LGR5 identified as a stem cell marker alongside two closely related proteins, LGR4 and LGR6, are specific receptors for RSPOs. Extensive research has significantly advanced our understanding of the functional roles of RSPO/LGR and Wnt signaling in stem cell biology^38,39.

We have previously shown that Wnt /β-catenin activation in the onychodermis contributes to the proliferation of the nail matrix epithelium and hair matrix via RSPO4-mediated signaling⁴⁰. Our previous findings also highlighted high expression levels of RSPO4 and LGR6 in the DP, suggesting their roles in the proliferation of mesenchymal and epithelial progenitors during hair follicle regeneration⁴¹.

When the Wnt-stimulating effects of these drugs were compared, CP21R7 exhibited the highest luciferase activity, surpassing lithium chloride, the positive control, by tenfold. On the other hand, proflavine displayed the lowest effect. It showed only 20% in the positive control’s Wnt stimulation. Interestingly, proflavine, potentially a lead compound, did not initially qualify as a lead compound based solely on signal activation from a promoter study. However, leveraging a gene bank derived from bioinformatics studies provided deeper insights into genes influencing target cells. This approach involved secondary screening of hits identified in the promoter assay. It propelled us closer to developing disease-specific drugs by scrutinizing target cell responses and gene expression.

Furthermore, the 3D spheroid structure formed by combining fibroblasts and keratinocytes initially presented as a mixed configuration akin to the arrangement of skin cells in the first row of Fig. 4a. Subsequently, these two cell types segregated into distinct regions while coexisting. Notably, only dermal papilla cells exhibited a germ-like structure, whereas spherical fibroblasts and keratinocytes elongated in a unidirectional manner. This germ-like structure serves as a widely used method to evaluate drug-induced hair growth effects by observing length changes⁴².

In this study, spheroids were created using identical skin keratinocytes and fibroblasts of various origins (skin, hair, and nail). The shape of the spheroid was influenced by the source of fibroblasts as depicted in Fig. 4a.

Interestingly, nail cells demonstrated the creation of a similar germ-like structure resembling hair cells. MXD, a well-known hair growth drug, increased lengths of both nail and hair germs. Both nail and hair germs exhibited analogous responses to drug treatments. Among the nine drugs tested, imidocarb and proflavine significantly increased the expression of all four types of fresh DP markers and hair germ markers identified by Ohyama⁴³. However, only imidocarb was effective in altering lengths of both hair and nail germs. This finding highlights imidocarb’s unique capability. Contrary to proflavine, which solely upregulated gene markers, imidocarb also manifested these genetic changes into tangible morphological alterations in hair and nail growth. This dual functionality of imidocarb highlights its potential as a valuable agent in treatments for hair and nail regeneration, suggesting a broader impact than merely genetic upregulation.

The experiment was conducted during the first telogen phase, which differs significantly between sexes. In male C57BL/6 mice, this phase lasts from postnatal day (P) 42 to P53, whereas in females it extends from P42 to P72⁴⁴. Therefore, female mice would need to be 10–12 weeks old to match the experimental window. However, the longer and more variable telogen phase in females may introduce greater individual differences, potentially increasing data variability. Furthermore, androgenetic alopecia is primarily influenced by male hormones, and hair growth is significantly affected by hormonal fluctuations. To minimize confounding factors related to the estrous cycle and ensure experimental consistency, we used only male mice in this study.

In our animal study, we intentionally excluded drugs with known adverse effects, specifically four chemotherapeutic agents (entinostat, tucidinostat, enzastaurin, and abemaciclib) known for their toxicities. Additionally, we excluded non-tonic formulations, specifically drugs that could not be dissolved in a tonic solvent composed of water and ethanol, such as CP21R7 and BIO. This approach allowed us to focus on the efficacy of new compounds. Aminoacridine, imidocarb, and proflavine were selected due to their potential benefits in promoting hair and nail growth. These compounds were assessed through rigorous animal experiments, applying the same criteria for exclusion to ensure consistency across our study focused on both hair and nail growth. Notably, a combination of aminoacridine and MXD exhibited higher hair growth effects than MXD alone, which served as a positive control.

A striking observation in the aminoacridine with MXD group was blackening of the skin on mice’s backs within a week, a change associated with the onset of the anagen phase of hair growth. This effect was not observed in the MXD-only group, suggesting a unique action of aminoacridine. In vitro tests revealed that aminoacridine significantly increased RSPO3 RNA levels, indicating its potential role in activating this gene. This finding aligned with findings of Niida et al.⁴³ showing that RSPO3 could induce a proliferative effect through standard Wnt signaling, suggesting a possible mechanism for aminoacridine’s effects on hair growth. Although aminoacridine resulted in the highest level of RSPO3 expression, there was a lack of consistency in results between animal experiments and hair germ experiments. This discrepancy, despite aminoacridine’s apparent effectiveness, indicates a complex interaction at play. It necessitates further investigation to fully understand the drug’s mechanisms and effects in different biological contexts.

Additionally, visible skin discoloration was observed in images following treatment with aminoacridine and proflavine. However, we confirmed that these discolorations could be easily removed using a tonic solution composed of 30% ethanol and 70% water, indicating that the observed color changes were not due to intrinsic skin alterations but rather the presence of residual compounds.

When the effects of various drugs, including a positive control, on toenail length were assessed, we observed that the collected drugs generally stimulated fingernail and toenail growth. However, imidocarb stood out, showing statistically significant effects across the board. This finding is consistent with observed changes in nail germ length. Imidocarb’s distinct efficacy in promoting nail growth at a statistically significant level highlights its potential as a targeted treatment for nail growth enhancement. The consistency of its effects on both toenail growth and nail germ length underscores the reliability and potential applicability of imidocarb in clinical settings for nail-related conditions.

Furthermore, our in vivo fluorescence permeability imaging tests showed that aminoacridine had good skin permeability, which was not adversely affected by increased alcohol concentrations. This characteristic is crucial for its effectiveness as a topical treatment.

In summary, our extensive screening of 5170 drugs using HTS techniques led to the discovery of six new drugs, among which aminoacridine, abemaciclib, and the combination of aminoacridine with MXD showed promising results in animal models. These drugs demonstrate potential as effective therapeutic agents for hair regeneration by activating Wnt/β-catenin signalling pathways.

However, our study has some limitations. First, we only tested three drugs in animal models. Further research involving more drugs is necessary to determine optimal concentrations and assess their safety and effectiveness comprehensively. Additionally, large-scale clinical trials are essential to confirm the safety and efficacy of these potential Wnt/β-catenin activators. Such trials could pave the way to develop effective treatments for hair loss and nail health issues.

Author contributions

G.A., H.-S.J., and D.L.: data analysis and writing of the manuscript; J.S.: idea formation; M.K.: in vitro testing.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Dr. Martina Rauner and Dr. Ophelia Bu. A peer review file is available.

Data availability

The numerical source data underlying the graphs and charts presented in this study are provided in the Source Data file. All other data supporting the findings of this study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Financial disclosure

The authors have the following to disclose: support provided by the Ministry of Health and Welfare (HI21C0694, Development of treatment for intractable skin diseases using low-molecular compounds with enhanced Wnt/beta-catenin), Republic of Korea.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09656-1.