Morphometric prediction of mesenchymal stromal cell-like immunosuppressive capacity of human hair follicle dermal sheath cup cells: an implication for regenerative medicine in hair loss diseases

Abstract

Background

Human hair follicle dermal sheath cup cells (DSCCs) hold promise as a cell source of regenerative medicine treatment for hair loss owing to their ability to secrete growth factors and/or signal pathway activators. The therapeutic effect of autologous DSCCs transplantation for male/female pattern hair loss (PHL) was demonstrated in a phase III equivalent clinical study. Intralesional inflammation has been implicated in the pathophysiology of various hair loss diseases, including PHL. As DSCCs possess mesenchymal stem/stromal cell (MSC)-like properties and MSCs are immunosuppressive, we investigated whether they exhibit immunoregulatory capabilities comparable to MSCs and developed an in vitro morphometric assay to predict this capability.

Methods

DSCCs were isolated via microdissection and propagated in vitro. Their conformity to MSC criteria was assessed based on cell surface antigen expression and differentiation potential. Furthermore, immunoregulatory capabilities were assessed by co-culturing DSCCs with anti-CD3/28 antibody-stimulated peripheral blood mononuclear cells (PBMCs) and examining the suppression mechanisms through pharmacological intervention. Multiple lots of DSCCs derived from various donors and manufacturing conditions were cultured and analyzed by phase-contrast microscopy to obtain their morphometric profiles. Parameters correlating with the expression levels of immunomodulatory factors were used to create a predictive model. Additional DSCC lots were manufactured to validate the predictive model.

Results

Similar to MSCs, cell differentiation assays revealed that DSCCs exhibited multipotency, and they did not express co-stimulatory molecules in response to immunogenic stimuli, suggesting low immunogenicity. Moreover, co-culture experiments with allogeneic PBMCs revealed that DSCCs reduced T cell proliferation (from 78 to 5%) and increased the proportion of FOXP3-positive regulatory T cells (from 3 to 10%) relative to the baseline. This capability was indoleamine 2,3-dioxygenase 1 (IDO)-dependent and affected by IDO expression levels. Image analysis of DSCCs revealed morphometric parameters that were highly correlated with IDO expression, and a model established using these parameters was able to predict IDO expression levels in the validation lots.

Conclusions

Results suggest that human DSCCs can help regenerative medicine approaches, not only for supporting hair follicle regeneration but also for suppressing microinflammation that potentially contributes to hair loss, and this can be readily evaluated using a newly developed morphometric analysis.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13287-025-04764-x.

Background

Development and regeneration of hair follicles (HFs) depend on the well-orchestrated interactions between hair-inductive mesenchymal and receptive epithelial components [1]. The dermal sheath (DS) is a connective tissue structure that envelops the main body of the hair follicle. It consists of multiple layers of cylindrical epithelial structures, and studies have suggested that it contains cell populations with hair-inducing activity [2]. The cells located at the proximal cup-shaped end of the HF are known as DS cup (DSC) cells (DSCCs), which can be obtained by microdissection and subsequent in vitro propagation [3]. DSCCs have garnered attention for their potential contribution to HF regeneration. When transplanted into mouse ears, murine DSCCs were incorporated into the dermal papilla (DP), a specialized mesenchymal component that regulates hair follicle development and regeneration, and promoted hair growth [4]. Implanted allogeneic DSC aggregates yielded new HFs containing DSC-derived cells in humans [5]. Moreover, injected human DSCCs were shown to migrate into human HF structures reconstituted in immunodeficient mice [6]. In recent Phase IIb and III equivalent studies, autologous DSCC injection improved hair density and cumulative hair diameter in male/female pattern hair loss (PHL) [3, 7]. These results indicate the HF regenerative potential of this cell population.

Mesenchymal stromal/stem cells (MSCs) have attracted interest in the field of regenerative medicine because of their low immunogenicity and intrinsic immunosuppressive capacity in allogeneic transplantation [8, 9]. MSCs have been used for the treatment of graft-versus-host disease and intractable fistulas associated with Crohn’s disease [10, 11], and for the treatment of alopecia areata, an autoimmune hair loss disease [12].

A major limitation to developing effective MSC-based immunosuppressive therapies is that they exhibit notably different immunosuppressive capacities depending on the donors and culture conditions. Several approaches have been useful for the assessment of immunosuppressive capacity of MSCs in vitro [13, 14]; however, their reliability and versatility remain elusive. Cell morphology has been widely used for monitoring cell status. Using advanced image processing techniques, the morphological features of cultured MSCs have been successfully linked to their quality attributes [15–17]. Furthermore, the combination of automated microscopy and machine learning models can predict cell quality attributes, without damaging cells by the biochemical stress that accompanies the measurement procedures [18]. Results suggest that noninvasive, morphology-based quality prediction approaches are advantageous for solving the bottlenecks in cell manufacturing.

HF-derived cells exhibit MSC-like properties [19–21]. In mice, DSCC transfusion prolonged the engraftment of pancreatic islet transplantation [22]. Human DS cells derived from the lower portion of HF fulfilled the 2006 International Society for Cellular Therapy (ISCT) criteria for MSCs, including multilineage differentiation capacity and cell surface antigen expression [23, 24]. Furthermore, these cells suppressed peripheral blood mononuclear cell (PBMC) proliferation, induced M2-type macrophages, and promoted the generation of regulatory T cells. Human DSCCs express PD-L1 and inhibit IFN-γ production from PBMCs [25]. The contribution of microinflammation to the pathophysiology of PHL has been suggested [26, 27]. Therefore, detailed investigation of MSC-like properties, particularly immunosuppressive capacity, in DSCCs used in the aforementioned clinical studies may be clinically significant [7].

In this study, we examined the MSC-like properties of DSCCs, with a focus on their immunogenicity and immunoregulatory capacity represented by allogeneic T cell suppression. Furthermore, we determined the usefulness of adopting morphometric parameters, originally developed for the prediction of MSC immunosuppressive activity, for evaluating the clinical potential of DSCCs.

Materials and methods

DSC dissection and DSCCs culture

A tissue biopsy of occipital scalp skin was collected from the Skin-Clinic. HFs were isolated from each sample, and DSCs were dissected under a stereomicroscope as described previously [3]. The isolated DSCs were transferred into culture plates and incubated at 37 °C and 5% CO₂ in culture medium for expansion. DSCCs at passage 2 were suspended in cryopreservation liquid and stored at − 80 °C. The DSCCs were re-cultured, and passage 3 cells were used for immunosuppression experiments. All patients involved in this study provided written informed consent. The research was conducted with the approval of the human research ethics committees at the Shiseido Global Innovation Center (#18000030).

MSC expansion

Bone marrow-derived MSCs (BM-MSCs) and adipose-derived MSCs (AD-MSCs) were purchased from PromoCell (Heidelberg, Germany) and cultured in the manufacturer’s recommended medium (Mesenchymal Stem Cell Growth Medium 2, PromoCell). The cells were cultured at 37 °C in a 5% CO₂ atmosphere. Passage 3 cells were used for the subsequent experiments.

Analysis of cell surface antigens

To phenotype the DSCCs, passage 3 cells were incubated and stained with CD34(581), CD45(H130), CD73(AD2), CD90(5E10), and CD105(43A3) (BioLegend, San Diego, CA). DSCCs, BM-MSCs, and AD-MSCs were assessed for expression of human leukocyte antigens [HLA-ABC (W6/32), HLA-DR(L243)], co-stimulatory molecules [CD80(2D10), CD86(BU63)], and immunosuppressive molecules [PD-L1(MIH2), FASL(NOK-1), HLA-G(87G), CD200(OX-104)] before and after 10 ng/mL IFN-γ treatment for 3 days (All reagents were purchased from BioLegend). Stained cells were examined using a Guava easyCyte (Luminex, Austin, TX) or SH800S (Sony, Tokyo, Japan) flow cytometer. Data were analyzed using FlowJo software (BD Biosciences, Franklin Lakes, NJ).

Differentiation capacity of DSCCs

For adipogenic differentiation, the DSCCs were seeded at a density of 4 × 10⁴ cells per well in 12-well plates, cultured for 3 days in culture medium, then moved to a differentiation medium for 2 weeks. For osteogenic differentiation, the DSCCs were seeded at a density of 2 × 10⁴ cells per well in 12-well plates, cultured for 3 days in culture medium, then shifted to a differentiation medium for 3 weeks. For chondrogenic differentiation, 8 × 10⁴ cells per well in 12-well plates were cultured in 96-well plates with differentiation medium for 3 weeks. The various differentiation media were from the StemPro Differentiation Kits purchased from Thermo Fisher Scientific (Waltham, MA). After differentiation induction, the cells were fixed with 4% paraformaldehyde (Wako, Osaka, Japan) and stained with Oil Red O (ScienCell, Carlsbad, CA) for adipogenic differentiation, Alizarin Red S (ScienCell) for osteogenic differentiation, and Alcian Blue (ScienCell) for chondrogenic differentiation.

Detection of IDO expression

Whole cell protein extracts were prepared from DSCCs, BM-MSCs, and AD-MSCs, which were either nonactivated or activated with 10 ng/mL IFN-γ treatment for 3 days. Indoleamine 2,3-dioxygenase 1 (IDO) protein expression was measured using the Human IDO ELISA kit (Abcam, Cambridge, UK), based on the manufacturer’s instructions. Optical densities were measured at 450 nm using the VICTOR3Multilabel Plate Reader (PerkinElmer, Waltham, MA, USA).

PBMC preparation and T cell activation

PBMCs from healthy donors were purchased from Precision for Medicine (Bethesda, MD), and labeled using the CellTrace CFSE Cell Proliferation Kit or the CellTrace Violet Cell Proliferation Kit following the manufacturer’s instructions (Thermo Fisher Scientific). Labeled PBMCs were resuspended in RPMI-1640 containing 10% fetal bovine serum (Thermo Fisher Scientific) and seeded at 2 × 10⁵ cells per well in 96-well plates. For the T cell stimulation group, the labeled PBMCs were stimulated with Dynabeads Human T-Activator CD3/CD28 (αCD3/28) (Thermo Fisher Scientific) for T cell activation.

Immunosuppression assays

Labeled PBMCs stimulated by αCD3/28 were co-cultured with DSCCs or MSCs. After 5 days, the cells were collected and labeled with anti-CD4(OKT4) and anti-CD8(SK1) antibodies (BioLegend) to assess the dose-dependent immunosuppressive effects of DSCCs at various PBMC/DSCC ratios (1:1, 1:0.5, 1:0.1 cell ratios). To determine the potential inhibition mechanism of activated T cell proliferation, labeled PBMCs stimulated with αCD3/28 were co-cultured with 1 × 10⁵ DSCCs, followed by 1 mM L-1-methyltryptophan (L-1MT) (Merck, Darmstadt, Germany) or 100 μg neutralized anti-PD-L1 antibody (29E2A3, BioLegend), respectively. After 5 days, the cells were collected and labeled with anti-CD4 and anti-CD8 antibodies. Moreover, T cell proliferation was determined by flow cytometry analysis of CFSE or Violet intensity.

Analysis of regulatory T cell inducibility

PBMCs stimulated by αCD3/28 were co-cultured with 1 × 10⁵ DSCCs. After 5 days, the cells were fixed and labeled with anti-CD4, anti-CD25 (M-A251, BioLegend), and anti-FOXP3 (PCH101, Thermo Fisher Scientific) antibodies. The formation of regulatory T cells (Tregs) was determined by the percentage of CD4⁺FOXP3⁺ or CD4⁺CD25⁺FOXP3⁺ cells among the total CD4⁺ T cells using flow cytometry. To investigate the putative mechanism, 1 mM L-1MT or 100 µg neutralized anti-PD-L1 antibody was added to the co-culture.

Image acquisition

DSCCs were seeded into Cell BIND^® 6-well plates (Corning Incorporated, Corning, NY) at a density of 18,000 cells per well (three wells for each lot). Phase-contrast images of DSCCs were acquired using a Nikon BioStation CT (Nikon Corporation, Tokyo, Japan) at 4 × magnification to create 8 × 8 tiled images covering 15.3 mm² per well. Image acquisition was conducted every 6 h from 18 to 96 h post-seeding. The culture medium was changed 48 h after seeding. Images for each DSCC lot were captured at the passage number indicated in Supplementary Table S1 and cryopreserved until further use (co-culture with PBMCs and IDO expression level measurements).

Image processing and morphological profiling

Image processing and morphological profiling were performed using the method described by Imai et al. [18]. The acquired images were analyzed using CL-Quant software (Nikon Corporation), which revealed individual regions occupied by each cell and measured 13 morphological descriptors that reflected 8 shape-associated and 5 intensity-associated features for each region (Supplementary Table S2). For the area and cell count plot, the mean of the area and the total number of identified cell regions (cell count) were calculated for all cells within the three tiled images. Thus, each profile for a single DSCC lot consisted of 14 values (14 time points). For principal component analysis (PCA) and the prediction models, the mean of each morphological descriptor was calculated for all cells within a single tiled image as morphological parameters. As a result, the morphological profile for each well containing a single DSCC lot consisted of 182 values (14 time points × 13 morphological parameters). PCA and visualization were performed using R software (version 4.2.2), by selectively using the 182 values.

Construction of prediction models

The parameters which significantly correlated with IDO expression were identified among the 182 parameters. Of these, the two parameters with the highest correlation coefficient were selected from each category (shape and intensity features) at the same time point. A linear regression model using these two parameters as explanatory parameters and the IDO expression levels as objective parameters was constructed using multiple regression analysis with R software. The constructed linear regression model was applied to the data obtained from separate DSCC lots cultured for validation, and its predictive accuracy was assessed.

Statistical analysis

Statistical analyses were performed using R software (version 4.2.2). The data were presented as the mean ± SD. Comparisons among more than three experimental groups were conducted using a one-way ANOVA with Tukey’s test, whereas comparisons between two experimental groups were done using a Student’s t-test. Statistical significance was defined as P of < 0.05.

Results

DSCCs satisfy the criteria for MSCs

The DSCCs used in the present study were aseptically cultured using the same manufacturing protocols described in previous clinical studies [3, 7]. Briefly, DSCs were mechanically isolated from the cup-like portion of the DS from the HFs isolated from the occipital region of the human scalp, which involved the inversion of the cup and separation of the DP (Fig. 1a–c). DSCCs were obtained by explant culture of isolated DSCs (Fig. 1d). They adhered to plastic, expressed phenotypic markers (CD105 +, CD90 +, CD73 +, ≥ 95%; CD45 +, CD34 +, ≤ 2%), and exhibited trilineage differentiation potency (Supplementary Fig. S1a, b), satisfying the ISCT minimal criteria for MSCs [23].

Fig. 1.

Isolation of DSC and phenotype of IFN-γ-activated DSCCs. A–D DSC isolation procedures. (A) A scalp tissue obtained by punch biopsy, (B) an intact hair follicle, (C) dissected and inverted DSC and DP before separation, and (D) morphological features of DSCCs at day 7 of primary culture. Dashed line: dissected location; bar: 500 μm. E, F DSCCs from three donors and two types of MSCs were activated with IFN-γ for 48 h and subjected to flow cytometric analysis using antibodies against anti-human leukocyte antigens and co-stimulatory molecules, HLA-ABC, HLA-DR, CD80, and CD86 (E), and immunosuppressive molecules, PD-L1, CD200, FASL, and HLA-G (F). G The mean fluorescence intensity of surface markers on DSCCs derived from three donors in the presence and absence of IFN-γ (**P < 0.01, ***P < 0.001, Student’s t-test). H IDO protein levels of IFN-γ-activated MSCs and DSCCs, and nonactivated MSCs and DSCCs extracts as measured by ELISA. Mean ± SD (*P < 0.05, ***P < 0.001, Student’s t-test)

DSCCs are comparable to MSCs in immunogenicity and immunosuppressive capacity

The position statement by the MSC Committee of ISCT in 2013 suggested that the evaluation of the immunomodulatory properties of MSCs may be done by IFN-γ activation [14]. Therefore, IFN-γ-activated DSCCs from three different manufacturing lots were analyzed by flow cytometry to assess changes in antigen-presenting and immunoregulatory molecules. Notably, representative immunosuppressive molecules, including PD-L1, CD200, FASL, and HLA-G, which are important for maintaining the immune privilege of the HF environment, were of particular interest [25, 28–30]. Two different sources of MSCs, namely BM-MSCs and AD-MSCs, were simultaneously examined as controls. Inactivated DSCCs, nor BM-MSCs or AD-MSCs, minimally expressed HLA class II (HLA-DR, ≤ 2%) or co-stimulatory molecules (CD80 and CD86, ≤ 2%) (Fig. 1e). Although DSCCs, BM-MSCs, and AD-MSCs moderately expressed PD-L1, they lacked expression of the immunosuppressive molecules, CD200, FASL, and HLA-G (Fig. 1f).

Following IFN-γ stimulation, activated DSCCs, BM-MSCs, and AD-MSCs showed increased expression of HLA-ABC, HLA-DR, and, notably, PD-L1 (Fig. 1f). In DSCCs, the expression of HLA-ABC, HLA-DR, and PD-L1 was significantly increased following IFN-γ stimulation, whereas no significant increase was observed for CD80 and CD86 (Fig. 1g). Moreover, CD80/86 was not upregulated by other proinflammatory factors, such as TNF-α and LPS (Supplementary Table S3). A persistent lack of CD80 and CD86 expression, irrespective of HLA class II upregulation, suggested a lack of ability to present antigens on DSCCs, BM-MSCs, and AD-MSCs. These results suggest that DSCCs, BM-MSCs, and AD-MSCs are discernible in terms of low immunogenicity and immunosuppressive capacity. The absence of crucial antigen-presenting molecules, together with the persistent expression of PD-L1, further supports an immunomodulatory role for DSCCs.

DSCCs upregulate IDO following IFN-γ stimulation

IDO contributes to the maintenance of the immune privilege of a tissue [31]. It is also an important factor that determines the immunoregulatory capacity of MSCs [32–34]. When DSCCs, BM-MSCs, and AD-MSCs were stimulated with IFN-γ, IDO expression was increased (BM-MSCs: P < 0.05, AD-MSCs: P < 0.001, DSCCs: P < 0.001, Fig. 1h). The extent of upregulation was greater in all DSCC lines compared with that in BM-MSCs and AD-MSCs.

DSCCs exert a suppressive effect on allogenic T cell proliferation comparable to MSCs

To examine the immunosuppressive properties, DSCCs, BM-MSCs, and AD-MSCs were co-cultured with CFSE-labeled PBMCs stimulated with anti-CD3 and CD28 antibody (αCD3/28). DSCCs significantly reduced T cell proliferation from 77.63% ± 3.35% to 5.14% ± 1.85% (P < 0.001) (Fig. 2a, b). The extent of inhibition was comparable to that of AD-MSCs (from 77.63% ± 3.35% to 5.48% ± 3.36%, P < 0.001) and was greater than that of BM-MSCs (from 77.63% ± 3.35% to 20.69% ± 9.21%, P < 0.001). Moreover, DSCCs suppressed the proliferation of CD8 + T cells with the magnitude of suppression analogous to that observed in CD4 + T cells (from 87.07% ± 4.43% to DSCCs: 7.54% ± 5.80%, BM-MSCs: 18.96% ± 16.90%, AD-MSCs: 6.49% ± 3.71%) (Fig. 2a, b). These results suggest potent T cell suppressive capacity of DSCCs, which was comparable to that of MSCs.

Fig. 2.

Effects of DSCCs and MSCs on T cell proliferation. A Representative flow cytometry histograms showing T cell proliferation after 5 days of co-culture with/without DSCCs or MSCs. Experimental conditions: PBMCs alone (red), with αCD3/28 (orange), with DSCCs + αCD3/28 (yellow), with BM-MSCs + αCD3/28 (green), and with AD-MSCs + αCD3/28 (blue). B Quantification of T cell proliferation showing the mean ± SD of the proliferation rates from three independent experiments in (A). C Quantification of T cell proliferation showing the mean ± SD of the proliferation rates after 5 days of co-culture in the presence/absence of DSCCs at the indicated ratio (PBMCs/DSCCs). DSCCs were obtained from three donors. (***P < 0.001, Tukey HSD test)

The inhibitory effect of DSCCs on T cell proliferation was further assessed by altering the ratio between DSCCs and PBMCs in co-culture experiments. The suppressive effect of DSCCs decreased as the DSCC dilution increased. Even at a DSCC/PBMC ratio of 1:10, a statistically significant inhibition of CD4 + and CD8 + T cell proliferation was observed (Fig. 2c, P < 0.001). This dose-dependency further supported the immunosuppressive function of DSCCs. In allogeneic PBMC culture without stimulation and DSCCs; T cell proliferation was not induced. The T cell proliferation ratio in the PBMC-alone group was 0.29% ± 0.23% for CD4 + T cells and 0.21% ± 0.07% for CD8 + T cells, and 0.35% ± 0.17% for CD4 + T cells and 0.39% ± 0.19% for CD8 + T cells in the unstimulated PBMCs and DSCCs co-culture group (Fig. 2c). Thus, DSCCs alone failed to elicit allogeneic T cell activation, which further supports the low immunogenicity of DSCCs.

IDO inhibitor, and not the PD-L1 neutralizing antibody, eliminates the ability of DSCCs to suppress T cell proliferation

To determine the function of IDO and PD-L1 in the immunosuppressive mechanisms of DSCCs, an IDO-specific antagonist, L-1MT, and anti-PD-L1 neutralizing antibody were added to a co-culture of DSCCs and allogenic T cells [35]. Inhibition of IDO with L-1MT abolished the ability of DSCCs to suppress CD4 + (78.92% ± 11.24% vs. 6.26% ± 3.89%) and CD8 + (76.79% ± 13.11% vs. 9.41% ± 5.18%) T cell proliferation compared with that of the control (Fig. 3a, b, P < 0.001), whereas anti-PD-L1 neutralizing antibodies did not increase CD4 + (5.35% ± 3.34%) and CD8 + (9.50% ± 5.27%) T cell proliferation (Fig. 3a, b). IDO expression was not observed in DSCCs co-cultured with unstimulated PBMCs. Its expression was detectable only when DSCCs were co-cultured with αCD3/28-stimulated PBMCs (Supplementary Fig. S2). Thus, DSCCs express IDO in response to T cells following activation. These results suggest that IDO, and not PD-L1, is a primary factor in DSCC-induced T cell suppression.

Fig. 3.

DSCCs inhibit the proliferation of activated T cells via IDO. A Representative flow cytometry histograms showing T cell proliferation after 5 days of co-culture with DSCCs in the presence of IDO inhibitor (L-1MT) or anti-PD-L1 neutralizing antibodies. Experimental conditions: PBMCs alone (red), with αCD3/28 (orange), with DSCCs + αCD3/28 (yellow), with DSCCs + αCD3/28 + L-1MT (green), with DSCCs + αCD3/28 + anti-PD-L1 antibody (blue), with DSCCs + αCD3/28 + vehicle + mIgG2b (violet). B Quantification of T cell proliferation showing the mean ± SD of the proliferation rates from three independent experiments in (A) (***P < 0.001, ns: not significant, Tukey HSD test). DSCCs of all experiments were from three donors

Induction of regulatory T cells in DSCC and T cell co-culture

The induction of Tregs is important for immune regulation. The ability of DSCCs to induce Tregs was examined by flow cytometric analysis of PBMCs cultured with or without αCD3/28 stimulation and DSCCs to measure the proportion of forkhead box P3 (FOXP3) + and CD25 + FOXP3 + Tregs in CD4 + T cells. CD4 + FOXP3 + (9.56% ± 0.78% vs. 2.56% ± 0.22%, P < 0.001) and CD4 + CD25 + FOXP3 + (8.58% ± 0.84% vs. 2.30% ± 0.21%, P < 0.001) Tregs were significantly increased in co-cultured DSCCs and stimulated PBMCs compared with controls, which were stimulated with PBMCs alone (Fig. 4a). Notably, a significant difference was observed in the percentage of Tregs between co-cultures of DSCCs with unstimulated and those with stimulated PBMCs (CD4 + FOXP3 + 3.46% ± 0.18%, CD4 + CD25 + FOXP3 + 1.61% ± 0.07%, P < 0.001) (Fig. 4a). Thus, DSCCs promote Treg development under these conditions.

Fig. 4.

DSCCs induce regulatory T cells in an IDO-dependent manner. A Representative flow cytometry histograms showing regulatory T cell formation. PBMCs were cultured for 5 days alone, with αCD3/28, DSCCs, or DSCCs + αCD3/28. Harvested PBMCs were labeled with anti-CD4, FOXP3, and CD25 antibodies to measure the percentage of regulatory T cells. Representative data of three independent experiments are shown (mean ± SD) (***P < 0.001, Tukey HSD test). B, C Representative flow cytometry contour plots and quantification data of T cell proliferation showing the induction level of regulatory T cells after 5 days of co-culture with DSCCs alone, in the presence of IDO inhibitor (L-1MT) or anti-PD-L1 neutralizing antibody. Representative data of three independent experiments are shown (mean ± SD) (***P < 0.001, Tukey HSD test). DSCCs were from three donors

IDO inhibitor impairs regulatory T cell induction in DSCC and T cell co-culture

To determine the role of IDO in Treg induction by DSCCs, the IDO inhibitor L-1MT was added to DSCC and T cell co-culture. Interestingly, IDO inhibition significantly disrupted the ability of DSCCs to induce Tregs (Fig. 4b, c). Moreover, the CD4 + FOXP3 + and CD4 + CD25 + FOXP3 + T cell fraction in the IDO inhibition group was significantly decreased compared with those in the control group (4.11% ± 0.53% vs. 12.50% ± 1.22%, P < 0.001; 2.28% ± 0.58% vs. 6.91% ± 0.81%, P < 0.001). This indicates a role of IDO in DSCC-mediated Treg induction. As previous reports suggested a possible contribution of PD-L1 to induce Tregs by MSCs [36], anti-PD-L1 neutralizing antibody was added to the DSCC and T cell co-culture; however, the suppression of Treg induction was not observed (CD4 + FOXP3 + 13.00% ± 1.11%, CD4 + CD25 + FOXP3 + 7.73% ± 0.89%). This suggests the presence of a distinct mechanism for Treg induction between DSCCs and MSCs.

Different DSCC lots exhibit distinct immunomodulatory capacity and positively correlate with IDO expression levels

MSCs may have a reduced ability to inhibit T cell proliferation through replicative senescence, and this potency may be associated with IDO expression [37]. To determine whether DSCCs, like MSCs, have an impaired immunosuppressive capacity in association with replicative senescence, a morphological study was done to identify parameters associated with cell conditions and IDO expression levels using 10 DSCC lots with different passage numbers that were originally prepared as training data. Early-passage and late-passage DSCC lots were used for this study (Supplementary Table S1). Late-passage DSCCs (Lot 2, 4, 6, 8, 10) exhibited flattened cell bodies and cytoplasmic hypertrophy compared with early-passage DSCCs (Lot 1, 3, 5, 7, 9), suggesting that they were in replicative senescence (Fig. 5a). Additionally, late-passage DSCCs (Lot 2, 4, 6, 8, 10) exhibited decreased IDO expression levels compared with early-passage DSCCs (Fig. 5b, c). The greatest difference in IDO expression was observed between Lots 1 and 2, with a 35-fold decrease, whereas the smallest difference was observed between Lots 3 and 4, with a 2.5-fold decrease.

Fig. 5.

Late-passage DSCCs show reduced IDO production with a decreased capacity to suppress T cell proliferation. A Morphological images of each DSCC lot after 48 h of cultivation. Bar: 200 μm. B Protein levels of IDO after IFN-γ stimulation in each DSCC lot as determined by ELISA. The results are shown as the mean ± SD. C The cumulative IDO expression levels in early-passage and late-passage lots of DSCC. (**P < 0.01, Mann–Whitney U test. D Correlation between IDO production in 10 DSCC lots and their immunoregulatory capacity as evaluated by DSCC and CD3/28-stimulated CFSE-labeled PBMCs co-culture. Left: CFSE dilution of CD4 + cells shown as MFI. Middle: CFSE dilution of CD8 + cells shown as MFI. Right: Induction of regulatory T cells shown as the percentage of FOXP3-positive cells among CD4 + cells (**P < 0.01, Pearson’s product-moment correlation)

Based on a CFSE analysis, the mean fluorescence intensity (MFI) of the CD4 + and CD8 + T cell populations exhibited a positive correlation with IDO expression levels (Fig. 5d, CD4 + T cell: R = 0.84, P < 0.01, CD8 + T cell: R = 0.82, P < 0.01). Furthermore, a positive correlation between IDO expression and the percentage of Tregs was also evident (high Treg induction ability) (Fig. 5d, %FOXP3 + in CD4 + T cell, R = 0.85, P < 0.01). These results suggest that DSCCs in replicative senescence were more prone to downregulate IDO, which potentially disrupts their capacity to regulate T cell-mediated immunity.

Morphometric analysis enables a distinction among respective DSCC subgroups

Based on the observation above that reduced IDO expression is accompanied by replicative senescence, we hypothesized that the immunomodulatory capacity of DSCCs may be predicted by their morphological characteristics.

Using an automated image acquisition device in combination with morphometric analysis software, the mean values of 13 parameters from the morphological profile (Supplementary Table S2) over time (14 time points from 18 to 96 h post-seeding) were compared among the DSCC lots. The cell counts varied among the DSCC lots, and the distinction between early-passage (odd-numbered lots) and late-passage DSCC lots (even-numbered lots) was difficult to establish. However, the cell area was larger in the late-passage DSCC lots compared with those of the early-passage DSCC lots at all time points (Fig. 6a, Supplementary Fig. S3a).

Fig. 6.

Morphological features of DSCCs predict the degree of IDO expression. A Time course analysis of the “Area” in 10 DSCC lots. B Principal component analysis mapping of 10 DSCC lots based on morphological profile. Morphological parameters at four time points (24, 48, 72, and 96 h) following cell seeding were used. Left: 8 parameters related to cell shape are displayed. Right: 5 parameters related to cellular intensity are shown. The same color dots represent each lot (3 wells per lot). Dotted circles indicate clusters of late-passage lots. C The correlation between morphological features of DSCCs and IDO expression levels was assessed using the Area and SD of intensity (see Supplementary Tables S3) at 48 h following cell seeding. D IDO protein levels following IFN-γ stimulation in each DSCC lot (for test data) were analyzed by ELISA. The results are plotted as the mean ± SD. E Comparison between the experimentally detected and predicted IDO expression levels obtained from a linear regression model using morphological features of “Area” and “SD of intensity” at 48 h following seeding. The predicted (IDOpred) and experimental (IDOexp) IDO expression levels from the test dataset (10 DSCC lots), the linear fit (solid line), and the reference IDOpred = IDOexp line are shown for comparison (**P < 0.01, ***P < 0.001, Pearson’s product-moment correlation)

Subsequently, PCA was conducted using all morphological profiles consisting of 182 values (14 time points × 13 morphological parameters) to determine if early- and late-passage lots may be classified. The late-passage lots clustered in the high PC1 region enabled them to be distinguished from the early-passage lots (Supplementary Fig. S3b). Furthermore, the morphological profiles at four time points (24, 48, 72, and 96 h) with respect to either shape or intensity features was visualized by PCA (Fig. 6b). The clusters formed exclusively by the late-passage lots were observed, suggesting that the classification is possible with morphological profiles consisting of fewer time points and morphological parameters.

Prediction of IDO expression in DSCCs using a newly invented morphometric analysis

Because IDO expression reflects the immunomodulatory capacity of DSCCs (Fig. 5d), the association between morphological profiles and IDO expression in DSCCs was further examined. The correlation coefficients for all measured morphometric parameters at four time points (24, 48, 72, and 96 h) are shown in Supplementary Fig. S3c. Among the shape-related parameters, “Area” showed the highest correlation with IDO expression, and among the intensity-related parameters, “SD of Intensity” showed the highest correlation. These two parameters were selected for model construction to avoid multicollinearity by including only one parameter from each feature category. Among the values for “Area” and “SD of Intensity”, the 48 h time point showed the second highest correlation coefficient with IDO expression (R = − 0.85 and R = 0.78 respectively; P < 0.01) (Fig. 6c) and 96 h showed the highest (Supplementary Fig. S3d). Although the “Area” and “SD of Intensity” parameters at 96 h showed slightly higher correlation coefficients than those at 48 h, we adopted the 48 h parameters for model construction to enable earlier prediction of IDO expression, thereby greater advantages for quality control during the manufacturing process.

Compared with the early-passage lots, the most of late-passage lots tended to exhibit larger “Area” values and lower “SD of Intensity” values, which are consistent with the general morphological characteristics of senescent cells. As described above, DSCC lots with low IDO expression may exhibit the morphological characteristics of senescent cells. A linear regression model constructed using the two parameters at the 48 h culture time point (“Area” and “SD of intensity”) successfully predicted IDO expression levels with high correlation in the training dataset (Supplementary Fig. S3e, f, R = 0.88, P < 0.01).

Using ten newly manufactured DSCC lots as the test samples (Supplementary Table S1), the aforementioned model, adopting the two parameters was assessed for its capacity to predict IDO expression levels. Bona fide IDO expression of test DSCC lots was experimentally measured (Fig. 6d). Importantly, predicted IDO expression levels were highly correlated with those experimentally detected in individual lots (Fig. 6e, R = 0.88 P < 0.001). These results suggest that DSCC immunosuppression can be predicted non-invasively through a morphometric analysis.

Discussion

In this study, we demonstrated that DSCCs—as cell products with robust quality control indicators—may be useful for controlling inflammatory diseases and allogeneic transplantation through their immunomodulatory effects. Rahmani et al. found that DSCCs can differentiate into hair-related mesenchymal cells, such as DP cells and sheath cells [38]. Therefore, the potential of DSCCs to differentiate into hair follicle-related mesenchymal cells, which are reduced in PHL [4, 39], may contribute to the improvement observed in the aforementioned clinical trials [3, 7]. The current study demonstrates the additional advantage of using DSCCs as therapy for PHL.

The DSCCs examined in the present study met the minimal criteria for MSCs [23]. The promise of MSCs for treating inflammatory diseases is well known [40]. Their use for treating graft-versus-host disease is particularly notable due to their various immunomodulatory effects [10]. For diseases in which autoimmunity results in a loss of hair follicle structure, DSCCs may not only suppress autoimmune responses to treat hair loss disorders caused by inflammation but also have a potential to differentiate into hair follicle related mesenchymal cells to replenish the lost cells, thereby restoring their function. DSCC treatment offers a more targeted approach compared with conventional MSC therapies.

Some studies have also categorized dermal sheath cells prepared from HFs as MSCs based on characteristics that meet the aforementioned criteria and proposed their therapeutic use for immunomodulation and wound healing [24, 41]. However, the DSCCs used in this study originated from microdissected dermal sheath cups (DSCs) as the starting material for cultivation. Therefore, the cells used in previous studies which were directly derived from the outgrowth of explant HFs culture may not be equivalent to the DSCCs. This distinction is significant because DSCs contain cells capable of inducing hair follicle formation and differentiating into DP and dermal sheath cells [4, 38]. The biological properties of dermal sheath cells differ considerably depending on their anatomical origin, as DSCCs, but not cells derived from the non-bulbar dermal sheath, can induce hair follicle growth [4]. This distinction is similar with that of MSCs. MSCs from different origins, such as BM-MSCs and AD-MSCs, exhibit distinct functions [42, 43]. As the DSCCs used in this study were derived solely from isolated DSCs and did not include other components, further functional dissection as well as evaluation of their hair-inductive capacity and MSC properties, is necessary to fully elucidate the mechanisms underlying the efficacy observed in PHL clinical studies [3, 7].

Consistent with the previous observation that DSCCs induce T cell apoptosis through PD-L1 [25], the DSCCs used in this study expressed PD-L1 in a steady state and upregulated this immunosuppressive molecule when exposed to IFN-γ. Unexpectedly, PD-L1 did not appear to play a role in suppressing T cells. This may be attributed to the difference in T cell activation methods. In contrast to other studies in which T cells and DSCCs were co-cultured, αCD3/28 was added to stimulate T cell proliferation in our co-culture system. Previous studies focused on the elucidation of the mechanism maintaining the immune privilege of HFs [25], whereas our study emphasized the dissection of the immunosuppressive machinery. These differences may account for the discrepancy of the roles of IDO and PD-L1 described in this study versus other previous studies. PD-L1 enhances allograft survival following allogeneic transplantation [44]. DSCCs expressing IDO and PD-L1 efficiently evade the host immune response, as DSCCs have been transplanted previously into allogeneic hosts and engrafted successfully [5]. Thus, DSCCs hold promise as a preferential cell source for cell-based regenerative medicine because of their low immunogenicity.

Our previous studies demonstrated that autologous DSCC transplantation into PHL patient’s results in hair thickening [3, 7]. Sacchidanand et al. suggested that the immune response may be involved in the progression of PHL [45]. Improvement of hair thinning following DSCC transplantation may be partially attributed to the anti-inflammatory properties of DSCCs. The pathogenesis of female PHL is heterogenous compared with that of the male PHL which is androgen-dependent [46, 47]. Interestingly, we found that female PHL patients showed better treatment outcomes compared with male patients [7], which suggests that the contribution of microinflammation to female PHL patients may be greater [26]. The extent to which the immunosuppressive function of DSCCs plays a role in the improvement observed in clinical trials needs to be dissected, not only for efficacy evaluation, but also to elucidate female PHL pathophysiology.

In the present study, we constructed a prediction model using morphological profiles that were noninvasively obtained as explanatory parameters and IDO expression levels as objective parameters. Previous studies have described various methods for predicting the immunomodulatory effects of MSCs. For example, the mixed lymphocyte reaction assay using noninvasively acquired morphological profiles of MSCs was established to predict the overall immunosuppressive capacity [15, 18]. Although these evaluations, which directly use lymphocytes (or PBMCs), may more accurately reflect the in vivo immune response, the results may greatly vary due to multiple factors, including HLA differences between donors and responders, and the intrinsic variability of PBMCs [13]. The results of the present study suggest that IDO expression may be used to predict the effect of DSCCs on T cell suppression and, therefore, provide an objective parameter for constructing a predictive model. Our findings indicate that, in DSCCs, IDO is a quantitative marker that correlates strongly with T cell proliferation inhibition. The IDO expression level may be noninvasively evaluated through the newly invented method developed in the present study, which potentially enhances the clinical application of DSCCs as a cell-based product for regenerative medicine.

In general, passage number in DSCCs correlates with certain morphometric parameters and, consequently, with IDO expression. However, this relationship is not absolute, and we observed an exception in our test samples (lot 11–20). Specifically, lot 20 was a late-passage DSCC lot but exhibited morphometric characteristics more similar to early-passage lots. Phase-contrast microscopy images (Supplementary Fig. S4a) qualitatively suggested that lot 20 more closely resembled early-passage lots (e.g., lot 19) than other late-passage lots, and PCA based on morphometric parameters (Supplementary Fig. S4b) placed lot 20 closer to the early-passage cluster. Consistent with these features, the model predicted a high IDO expression level for lot 20, which closely matched the measured value. This observation also relates to the minimal difference in IDO expression observed between lot 19 and lot 20. We speculate that the donor corresponding to these lots inherently exhibited high IDO expression, which may have mitigated the typical passage-related decline, as has been reported for MSCs where donor-to-donor variability can influence immunomodulatory properties including IDO expression [48]. While this is a single example and cannot definitively rule out the influence of passage number on the model’s predictions, it supports the interpretation that the model predicts IDO expression based on morphometric parameters that can vary independently of passage number.

For the clinical application of our model, it is necessary to identify diseases in which IDO plays a key role in the regulation of inflammation. Furthermore, because our model is based on a limited dataset, its validation with a larger sample size is necessary to further delineate its applicability. Nevertheless, the noninvasive evaluation of the cell product quality using in vitro images, which can be achieved during the manufacturing process, is consistent with the Quality by Design concept, emphasizing the importance of understanding the relationship between manufacturing process parameters and quality attributes [49].

Conclusion

DSCCs suppress the proliferation of allogeneic T cells through IDO expression. They lack antigen-presenting capabilities and exhibit low immunogenicity. Moreover, the immunomodulatory capacity of DSCCs can be predicted noninvasively using morphometric parameters, which can predict the efficacy of cell-based treatment for PHL, particularly in females. Our findings provide new insight into the etiopathogenesis/pathophysiology of PHL involving microinflammation and suggest the importance of a pretreatment assessment of the immunoregulatory capacity of DSCCs in addition to their hair-inductive ability, which can be conducted by the novel approach established in this study.

Abbreviations

DSCCs

Dermal sheath cup cells

PHL

Male and female pattern hair loss

MSCs

Mesenchymal stromal/stem cells

PBMCs

Peripheral blood mononuclear cells

IDO

Indoleamine 2, 3-dioxygenase 1

HF

Hair follicle

DS

Dermal sheath

DSC

Dermal sheath cup

DP

Dermal papilla

αCD3/28

Anti-CD3/28 antibody

PD-L1

Programmed cell death ligand 1

L-1MT

L-1-methyltryptophan

Tregs

Regulatory T cells

PCA

Principal component analysis

FOXP3

Forkhead box P3

MFI

Mean fluorescence intensity

Author contributions

Conceptualization: HT Data Curation: HT, HL Formal analysis: HT, SK Investigation: HT, HL, HJ, SK, YN, M Ohyama Project administration: HJ, M Ogo, M Ohyama Supervision: M Ohyama Writing—Original Draft: HT, HL Writing—Review and Editing: HJ, SK, YN, M Ogo, M Ohyama.

Funding

Not applicable.

Data availability

Data are available upon reasonable request.

Declarations

Ethics approval and consent to participate

Ethical approval for the isolation of DSCC from human scalp tissue was granted by the human research ethics committees at the Shiseido Global Innovation Center, under the project titled "Evaluation of Hair Follicle Regeneration Ability of Cultured Cells Derived from Scalp Tissue," granted on 15 March 2023 (#18000030). All donors provided written informed consent prior to participation. The BM-MSC, AD-MSC, and PBMC, were purchased from Promocell and Precision for Medicine. Both companies have confirmed that ethical approval was obtained for the collection of human cells, and that donors provided informed consent. Promocell and Precision for Medicine outline their compliance with ethical and regulatory standards in their official documents (Promocell: https://promocell.com/jp_ja/about-us/compliance, Precision for Medicine: https://www.precisionformedicine.com/about-us/regulatory-information).

Consent for publication

Not applicable.

Competing interests

M Ohyama has earned advisory fees from Taisho Pharmaceutical Co. and ROHTO Pharmaceutical Co. concerning this study topic, along with lecture and advisory fees for unrelated topics from Eli Lilly Co., Pfizer Inc., Sanofi KK., Kyowa Kirin Co., Maruho Co., Bristol-Myers Squibb Japan, and AbbVie GK. He has received research funding for this study topic from Shiseido Co, Ltd., and for unrelated topics from Maruho Co., Advantest Co., and Sun Pharma Japan Ltd. HT, HL, HJ, SK, YN, and M Ogo are employees of Shiseido Co, Ltd.