Mechanical stimuli-induced CCL2 restores adult mouse cells to regenerate hair follicles

Abstract

Aged cells have declined regenerative ability when subjected to environmental insult. Here we elucidate the mechanism by which mechanical stimulus induces hair regeneration at the microenvironmental regulation level using the hair plucking and organoid culture models. We observed that the skin cells harvested from post-plucking day 3 (PPD3) have the best self-organizing ability during skin organoid culture and have the highest hair regeneration upon transplantation. By bulk RNA sequencing (RNA-seq) and single-cell RNA-seq analysis and in situ hybridization, we identified that the chemokine signaling pathway genes including CCL2 are significantly increased in the skin at PPD3 and in skin organoid cultures. Immunostaining shows that the PPD3 skin epithelial cells have increased multipotency, which is verified by the ability to self-organize to form epidermal aggregates during organoid culture. By adding CCL2 recombinant protein to the organoid culture using an environmental reprogramming protocol, we observed the PPD0 adult skin cells, which lose their regenerative ability can self-organize in organoid culture and regenerate hair follicles robustly upon transplantation. Our study demonstrates that CCL2 functions in immune regulation of hair regeneration under mechanical stimulus, and enhances cell multipotency during organoid culture. This provides a therapeutic potential for future clinical application.

Graphical abstract

Wu et al. elucidate the mechanism by which mechanical stimulus induces hair regeneration at the microenvironmental regulation level using the hair plucking and organoid culture models. This work provides a therapeutic potential for future clinical application through mechanical and immune regulation.

Introduction

Skin appendages are essential for protection, defense, thermoregulation, and appearance in mammals. The hair follicle is one of the most important skin appendages that displays visible changes in postnatal life. As more than 30% of people worldwide suffer from hair loss to a certain extent, hair regeneration becomes a booming research hotspot.¹ Locating in the specified niche called bulge, hair follicle stem cells (HFSCs) are responsible for the cyclic regeneration of the hair follicle that experiences multiple rounds of resting phase (telogen), growth phase (anagen), and regression phase (catagen) during the lifespan. However, like other tissues and organs, hair follicles also undergo structural and functional declines in the aging process, leading to hair loss or alopecia eventually.^2,3

Known findings indicate that HFSCs turn to epidermal cell lineages,⁴ escape from the niche,⁵ or enter a long quiescent state during aging,⁶ leading to a decreased regenerative ability. The current concept of hair regeneration suggests that both intrinsic and extrinsic factors dictate stem cell fate.^3,7 The intrinsic regulation of HFSCs includes signals from bulge stem cell niche and dermal papilla, which form a core controlling module that determines the HFSCs activation or silence.⁸ The intrinsic factors involve the epigenetic landscape, and genetic and transcriptional regulation, which are well-reviewed in previous studies.⁸ Increasing evidence shows that the extrinsic regulation by micro-environmental and macro-environmental factors largely influence HFSCs activity. For example, surrounding or attaching to the hair follicle, arrector pili muscle, adipose tissue, blood vessel, lymphatics, nerve, immune cells, and so on, can secrete inhibitory or activating factors that control the telogen-to-anagen transition process of the hair follicle.^1,7 Thus, these cells or tissues act as the micro-environmental regulators to influence HFSCs activity. In contrast, aging, circadian rhythm, chronic or acute stress, circulatory hormone, light, seasonal changes, temperature, and so on, can behave as the macro-environmental regulators that modulate HFSCs activity.^8,9 At the molecular level during the telogen-to-anagen transition, the activating factors include Wnt, fibroblast growth factor (FGF), follistatin, transforming growth factor β, and platelet-derived growth factor subunit A, among others, whereas the inhibitor factors include bone morphogenetic proteins (BMPs), Dkks, Sfrps, and so on.^6,10,11

Residing in the skin, immune cells are activated or increased before or during the telogen-to-anagen transition to influence HFSC activity. Under physiological conditions, the immune system is often maintained in homeostasis in the skin, enabling the hair follicles to have immune privileges. However, under pathological conditions, loss of immune privilege because of a disorder of the skin immune system results in attack from immune cells, which causes hair loss and alopecia eventually. For example, alopecia areata is an autoimmune disease, in which the hair matrix cells were attacked by the reactive cytotoxic CD8⁺ T cells.¹² The immune cells are activated or increased before or during telogen-to-anagen transition. The Cd4⁺Foxp3⁺ regulatory T cells accumulated surrounding the bulge secrete JAG1, which activates the Notch signaling pathway, leading to the activation of HFSCs.¹³ Immune regulation is also involved in large (>1 cm in diameter) wound-induced hair neogenesis in mice.^14,15 During wound healing, FGF-9 secreted by gamma-delta T cells induces Wnt2a activation in fibroblasts, which results in hair neogenesis in the central area of the wound bed.¹⁶ Mild injury also influences hair regeneration. Hair plucking as a mild injury destroys the microenvironment in the inner bulge niche that secretes inhibitory signals such as Fgf18 and Bmp6, also resulting in hair regeneration.⁹ Interestingly, Chen et al.¹⁷ observed that topological hair plucking within a confined skin area causes regeneration of all hair follicle populations, indicating that the microenvironment of HFSCs is also influenced by this micro-injury.

In addition to reactivating hair regeneration in vivo, scientists are now able to generate skin and hair follicles using organoid culture.^{18,19,20,21,22} These three-dimensionally (3D) cultured induced pluripotent stem cells, embryonic stem cells, or primarily isolated progenitor cells can be guided or self-organize to form a fully functional skin with the development of skin appendages including hair follicles. By 3D culture of primarily isolated newborn mouse skin cells in a trans-well culture insert, we observed attractive self-organization behavior in an approximately 10-days period.¹⁸ The single dissociated epidermal cells form a planar layer of the epidermis through a series of morphological transitions, which include aggregation, polarization, coalescence, planarization, and hair primordia formation. We also demonstrated that dermal cells behave actively during culture and are required for the morphological transitions of epidermal cells. Transplantation of the skin organoid into the nude mice results in fully functional skin formation, with abundant of hair follicle regeneration. Thus, this skin organoid culture system provides an appropriate model to study self-organization behavior, tissue development and regeneration, screen drugs, and engineer desired tissues (e.g., scalp skin vs body skin), and so on. Adult cells lose their regenerative ability. By applying the self-organizing principle learned from the newborn mouse progenitor cells, we were able to set up an environmental reprogramming protocol that guides and restores the adult cells to self-organize in organoid culture and improve skin and hair follicle regeneration upon transplantation.¹⁸ However, it is uncertain if the immune environment can be modulated to improve the hair regenerative ability of adult cells based on organoid culture.

In the present study, we first study how external stimuli such as hair plucking arouses stem cell competency that influences hair regeneration. By harvesting skins with hair plucking at different postnatal days and dissociating them into single cells, we tested their regenerative ability using the planar hair-forming assay, and observed that post-plucking day 3 (PPD3) skin cells have the highest hair-forming efficiency. Using transcriptomics analysis and in situ hybridization, we identified that cytokine-related genes are upregulated 3 days after hair plucking. We then performed an organoid culture of the adult skin cells, and observed that the PPD3 skin cells have a better self-organization ability. Importantly, by using the environmental reprogramming strategy, which allows us to apply chemokine (C-C motif) ligand 2 (CCL2) recombinant protein to the organoid culture, we observed a reactivated self-organization behavior of PPD0 adult cells as characterized in that of the newborn mouse cell organoid cultures. Transplantation of these cells into nude mice results in robust hair regeneration. Our study paves the way to apply organoid technology to regenerative biology, which is of general interest and has a high potential for regenerative medicine.

Results

Increased hair regeneration with grafting cells from different PPDs

To test if adult skin cells at different PPDs can regenerate hair follicles during reconstitution, we plucked the hairs from the dorsal back skin of 6-month-old CD1 mice, in which hair follicles are at the telogen phase (Figures 1A and 1B). We harvested these samples at PPD0 to PPD4 daily (Figure 1B), when hair follicles gradually transition from telogen to anagen (Figure 1A). We also harvested samples at PPD10, when the hair follicles enter full anagen. The skin samples were separated into epidermis and dermis, which were then dissociated into single cells using different enzymatic strategies (see materials and methods). The cells were remixed and transplanted into the wound bed created on the dorsal back skin of the nude mice using the planar hair-forming assay (Figure 1B).²³ Six weeks after cell transplantation, hairs grow out from the healed wound bed. Intriguingly, we observed that significantly more hairs grew out from the healed wound bed from the PPD2 (165 ± 10), PPD3 (268 ± 12), and PPD4 (188 ± 6) groups (n ≥ 3, p < 0.01 or 0.001), compared with the PPD0, PPD1, and PPD10 groups in which no or very few hairs regenerate (Figure 1C). These data suggest that adult cells from plucking-stimulated skin can be reactivated to form hairs in the reconstitution assay, and we observed that the PPD3 skin cells have the greatest regenerative ability.

Figure 1.

Hair regeneration by grafting cells from PPD0 and PPD4 skin

(A) Schematic of hair cycling from telogen to early and middle anagen. (B) Experimental design of hair plucking and the timing of sample harvest. Skin samples were harvested at different PPD, dissociated into epidermal and dermal cells and remixed at a ratio of 1:5, and grafted onto the back of the nude mice. The dashed box represents the plucked area. (C) Hair regeneration after grafting skin cells harvested from different PPDs. n ≥ 3, ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05, and # no significant change.

Morphological characterization of adult skin cells-derived hair follicles

Since adult tissues often under structural declines during regeneration, we examined the morphology and structure of the hair follicles regenerated from the PPD3 skin cells. There are four types of hairs developed in the mouse dorsum back skin, including Zigzag, Auchene, Awl, and Guard hairs.²⁴ Using phase-contrast microscopy, we observed adult cells can also regenerate all four types of hairs, although the hairs look a little thinner in diameter than the physiologically developed ones (Figure 2A). In addition, we observed that most of the hairs regenerated from adult cells are Zigzag type (Figure 2A). Scanning electron microscopy shows that the hair fiber in the reconstituted group has normal cuticle structures similar to that of the normal scaly hairs (Figure 2B).

Figure 2.

Morphological and structural characterization of the regenerated hair follicles

(A) Phase-contrast microscope compares the morphology of Zigzag, Auchene, Awl, and Guard hairs from normal development and reconstitution. (B) Scan electron microscope (SEM) image shows the cuticle structure formation in the hair fiber. (C) Hematoxylin and eosin (H&E) staining shows hair follicle and sebaceous gland formation in the reconstituted skin 14 days after transplantation. (D) (Upper) K14 immunostaining shows epithelial structures in the reconstituted skin. (Lower) Frozen sections shows the regenerated epithelia are derived from the K14H2BGFP transgenic mice donor cells. (E) AE13 and AE15 immunostaining shows the formation of the inner root sheath in the reconstituted skin. (F) CD34 and Sox9 immunostaining shows the formation of HFSCs in the bulge region of the reconstituted skin. (G) ALP staining and NCAM immunostaining show the formation of the dermal papilla in the reconstituted skin. N = 8.

Hematoxylin and eosin staining shows complete morphology of the skin with epidermal and dermal layers, as well as hair follicles along with the newly formed sebaceous gland in the reconstituted skin (Figure 2C). K14 immunostaining shows that the skin has essential epithelial layers including the basal epidermis and outer root sheath of the hair follicle (Figure 2D). The serial frozen section reveals that these epithelial cells are mostly derived from the grafted donors, as indicated by the fluorescent cells grafted from K14H2BGFP transgenic mice (Figure 2D), in which the basal epidermal cells where express K14 are inserted with a nucleotide sequence that encodes green fluorescence protein.²⁵ AE13 and AE15 immunostaining shows that the newly formed hair follicles have a normal inner root sheath (Figure 2E), suggesting that the reconstituted hair follicles can undergo differentiation resembling that of the physiologically developed ones.

HFSCs located in the bulge area are required to fuel the next round of hair regeneration. Using immunostaining for HFSC markers including CD34 and Sox9, we observed that HFSCs are correctly located in the bulge area in telogen and anagen hair follicles (Figure 2F). These data suggest that the reconstituted hair follicles using adult cells can form HFSCs and undergo cyclic transition. The dermal papilla is the signaling center required for hair follicle development and regeneration. Alkaline phosphatase (ALP) staining and neural cell adhesion molecule (NCAM) immunostaining reveal that the reconstituted hair follicles have normal dermal papilla and dermal sheath structures (Figure 2G). Together, these results characterized that the adult cells from PPD3 skin can regenerate fully structural and functional hair follicles.

Transcriptome profiling during hair plucking-induced hair regeneration

To identify molecular changes involved in plucking-induced hair regeneration, we performed RNA sequencing (RNA-seq) analysis of the PPD0 and PPD3 samples, which have the least or most hair regenerative ability among all the tested time points in the reconstitution assay. Kyoto Encyclopedia of Genes and Genomes enrichment analysis reveals that the chemokine signaling pathway ranks first among the differential gene ontology terms (Figure 3A). Volcano plots and heatmap show that among the 458 up-regulated genes at PPD3 vs PPD0, many chemokine signaling pathway genes including CCL and chemokine (C-X-C motif) ligands (CXCL) were significantly up-regulated (Figures 3B and 3C).

Figure 3.

RNA-seq compares gene expression between PPD0 and PPD3 skin cells

(A) Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis shows signaling pathways enriched in the PPD3 skin cells. (B) Volcano plot shows CCL and CXCL pathway genes expression in PPD0 and PPD3 skin cells. (C) Heatmap shows the chemokine signaling pathway gene expression in PPD0 and PPD3 skin cells. (D) RNA-seq analysis of inhibitors (BMPs) and activators of hair regeneration in PPD0 and PPD3 skin cells. CPM, counts per million reads mapped. ∗∗p < 0.01, ∗p < 0.05, n = 2. (E) Western blots show BMPs (2, 3, and 4), SHH, and FGF23 expression in PPD0 and PPD3 cells. (F) Statistical analysis of the western blots. ∗∗p < 0.01, ∗p < 0.05, n = 3.

We also examined the extra-follicular inhibitory and excitatory signals that maintain HFSCs quiescence at refractory telogen or activate HFSCs during the telogen-to-anagen transition, respectively. We observed that BMPs (2, 3, and 4) that maintain HFSCs quiescence were significantly decreased in PPD3 vs PPD0 skin cells. Whereas Shh, Nog, FGFs (20, 22, and 23), Fst, and Fstl1 that can activate HFSCs to transition from telogen to anagen were significantly increased in PPD3 vs PPD0 skin cells (Figure 3D). We verified the RNA-seq result by quantitative RT-PCR (See also Figure S1) and western blot (Figures 3E and 3F), which are consistent with each other. These results suggest that hair plucking leads to reduced expression of refractory telogen inhibitors and increased expression of telogen-to-anagen transition activators.

We next examined how these chemokines are changed spatiotemporally in plucking-induced hair regeneration. Among the CCL immune chemokines, we observed that CCLs (2, 3, 4, and 9) and CXCL (2, 11, and 13) are more significantly increased in PPD3 vs PPD0 skin (Figure 4A). We then performed in situ hybridization and immunostaining to verify the expression of representative pro-inflammatory cytokines during plucking-induced hair regeneration. CCL2 expression is up-regulated in the hair follicle keratinocytes soon after plucking at PPD1 and peaked at PPD3 (Figures 4B and 4C; see also Figures S2A and S2B), indicating that hair follicle keratinocytes are the main source of CCL2. CCL9 is less expressed in the plucked skin at PPD0 and PPD1, but is greatly increased in the hair follicle and inter-follicular region at PPD2 and peaks at PPD3 (Figures 4B and 4C). Since CCL2 has the highest expression among all CCL chemokines at PPD3, we focus on studying its role in the present study. In addition, previous study shows that CCL2 and CCL9 can be sensed by CC chemokine receptor 4 (CCR4), which is expressed by M1 macrophages.¹⁷ We observed that CCR4 is also increased in the hair follicle at PPD2 and peaked in the hair follicle and the interfollicular region at PPD3 (Figures 4B and 4C; see also Figures S2C and S2D). We did immunofluorescence staining for both M1 macrophages (F4/80⁺, CD86⁺) and M2 macrophages (CD163⁺, CD206⁺), and observed that M1 macrophages (F4/80⁺) but not M2 macrophages (CD163⁺, CD206⁺) are dramatically increased surrounding the plucked hair follicle in PPD3 vs PPD0 samples (Figures 5A and 5B). Consistent with the previous study,¹⁷ our data confirmed that hair plucking induces CCL2 and CCL9, which can recruit CCR4-expressing M1 macrophages that are required for inducing tumor necrosis factor α (TNF-α) expression (Figure 5C), followed by subsequent hair regeneration (Figure 5D).

Figure 4.

Characterization of CCL2 and CXCL pathway gene expression from PPD0 to PPD4

(A) RNA-seq analysis of CCL and CXCL pathway gene expression between PPD0 and PPD3 skin cells. CPM, counts per million reads mapped. ∗∗p < 0.01, ∗p < 0.05, n = 2.

(B) In situ hybridization shows mRNA expression of CCL2, CCL9, and CCR4 expression in PPD0 to PPD4 skin. (C) Statistical analysis of the in situ hybridization images. ∗∗p < 0.01, ∗p < 0.05, #, no significance, n ≥ 3.

Figure 5.

CCL2-CCR4 axis regulates plucking-induced hair regeneration

(A) F4/80 immunostaining and statistics show recruitment of M1 macrophage at PPD3 after plucking. (B) CD206 immunostaining and statistics show no obvious recruitment of M2 macrophage at PPD3 after plucking. (C) TNF-α immunostaining and statistics show skin inflammation at PPD3 after plucking. (D) K14 immunostaining and statistics show hair regenerate at PPD3 after plucking. (E) K14 immunostaining and statistics show hair regeneration is blocked after inhibition of CCL2 (iCCL2) or CCR4 (iCCR4) at PPD3. (F) F4/80 immunostaining and statistics show recruitment of M1 macrophage is disrupted after inhibition of CCL2 (iCCL2) or CCR4 (iCCR4) at PPD3. (G) CD206 immunostaining and statistics show no obvious recruitment of M2 macrophage at PPD3 in all three groups. (H) TNF-α immunostaining and statistics show significantly decreased skin inflammation after inhibition of iCCL2 or iCCR4 at PPD3. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, #, no significance, n ≥ 3.

To test CCL2 and CCR4 function in plucking-induced hair regeneration, we performed an in vivo assay by injecting a CCL2 inhibitor or CCR4 neutralizing antibody into the plucked skin. By harvesting samples from PPD3 and immunostaining for K14, we observed that the hair regeneration in the CCL2- and CCR4-inhibited groups is significantly blocked compared with the control (Figure 5E). Loss of function of CCL2 or CCR4 results in a decreased recruitment of M1 but not M2 macrophages (Figures 5F and 5G; see also Figures S3A–S3SD), as well as decreased TNF-α expression (Figure 5H). This confirms that loss of function of CCL2 or CCR4 perturbs hair regeneration upon hair plucking.

Activated HFSCs and increased multipotency after hair plucking

We next examined how hair plucking-induced micro-environmental changes influence cellular and molecular events in the hair follicle. Hair plucking results in destroying of the inner bulge niche. Using immunostaining for CD34, we verified that the plucked hair follicles only have one layer of stem cells in the outer bulge region (Figures 6A and 6B). Upon injury caused by plucking, the epithelial cells, including the bulge stem cells, and some of the cells located in the interfollicular region are quickly positive P16 (Figures 6C and 6D), which has been shown to mediate cell apoptosis.²⁶ Interestingly, the cells at the infundibulum region and a few cells at the bulge region start to proliferate at PPD1. Epithelial cell proliferation is more obvious in the bulge and secondary hair germ region at PPD2 and PPD3 (Figures 6E and 6F). Interestingly, P63, which marks the multipotent basal epithelial cells is gradually increased in the infundibulum and bulge region from PPD0 to PPD3 (Figures 6G and 6H). P-cadherin, which marks the secondary hair germ, is also gradually increased in the secondary hair germ and lower bulge regions (Figures 6I and 6J). These data suggest that hair plucking quickly induces cell apoptosis, but can be recovered by proliferation, which may produce more cells with multipotency, resulting in the activation of HFSCs.

Figure 6.

The increased multipotency of the cells in the regenerated hair follicle after hair plucking

(A) CD34 immunostaining shows HFSCs from PPD0 to PPD4. (B) Statistics of percentages of CD34⁺ cells in the plucked hair follicles. (C) P16 immunostaining shows apoptosis from PPD0 to PPD4. (D) Statistics of percentages of P16⁺ cells in the plucked hair follicles. (E) PCNA immunostaining shows cell proliferation from PPD0 to PPD4. (F) Statistics of percentages of PCNA⁺ cells in the plucked hair follicles. (G) P63 immunostaining shows multipotency of the cells in the regenerated hair follicle from PPD0 to PPD4. (H) Statistics of percentages of P63⁺ cells in the plucked hair follicles. (I) P-cadherin immunostaining shows multipotency of the cells in the regenerated hair follicle from PPD0 to PPD4. (J) Statistics of percentages of P-cadherin⁺ cells in the plucked hair follicles. ∗∗p < 0.01, ∗p < 0.05, #, no significance, n ≥ 3.

Self-organization behavior in skin organoid culture of adult skin cells

Adult cells lose their competency to self-organize during skin organoid culture.¹⁸ To examine if the cells from the plucked skin have a distinct cellular behavior during skin organoid culture, we harvested the skin from different PPDs. We dissociated the skin into single cells and cultured them for 1 day in a trans-well culture insert to test their self-organizing ability as indicated by the epidermal aggregate formation, without which hair follicles would not be regenerated upon transplantation. K14 immunostaining and quantification show that the epidermal aggregate size is increased in the PPD2, PPD3, and PPD4 groups, compared with that of the PPD0 and PPD1 groups in which tiny aggregates are formed (Figures 7A and 7B). P-cadherin and P63 were identified to be the markers of basal cells that can maintain the multipotency required for development or regeneration.^27,28 Immunostaining and quantification show that P-cadherin⁺ and P63⁺ cells are significantly increased in the PPD2, PPD3, and PPD4 groups, compared with that of the PPD0 and PPD1 groups (Figures 7C–7F). Notably, P-cadherin and P63 are expressed in most of the epidermal aggregate, indicating a lower differentiated state of the epidermal cells in PPD2, PPD3, and PPD4 groups, compared with that of the PPD0 and PPD1 groups, in which only a proportion of the cells are P-cadherin⁺ and P63⁺ (Figures 7C–7F). Indeed, by immunostaining and quantification of K10, which is an epidermal differentiation marker, we observed that the PPD3 group has the fewest K10⁺ cells, which indicates a low differentiation state in the aggregate, compared with the other groups (Figures 7G and 7H). These data suggest that adult cells from PPD3 can maintain their multipotency best during skin organoid culture.

Figure 7.

Organoid culture shows self-organization using cells from PPD0 to PPD4 skin cells

(A) K14 immunostaining shows epidermal cell cluster formation. (B) Quantification of the cell number in each epidermal cell cluster from K14 immunostaining. (C) P-cadherin immunostaining shows epidermal cell cluster formation. (D) Quantification of the P-cadherin⁺ cell number in each epidermal cell cluster from P-cadherin immunostaining. (E) P63 immunostaining shows epidermal cell cluster formation. (F) Quantification of the P63⁺ cell number in each epidermal cell cluster from P63 immunostaining. (G) K10 immunostaining shows epidermal cell differentiation. (H) Quantification of epidermal cell differentiation in each epidermal cell cluster from K10 immunostaining. ∗∗p < 0.01, ∗p < 0.05, #, no significance, n ≥ 3.

Environmental reprogramming restores adult cells to regenerate hair follicles

Using human embryonic stem cells or induced pluripotent stem cells, Lee et al.²⁰ set up a human skin organoid culture model and observed the cells can be guided to differentiate and form hair-bearing skin. To investigate if chemokine signaling is involved in human skin organoid formation, we analyzed the small conditional RNA sequencing (scRNA-seq) data from D29 and D48 samples when skin cells are primed to generate hair follicles. T-distributed stochastic neighbor embedding (tSNE) plots show 12 cell types with representative gene expression, including basal keratinocytes (CXCL14⁺), intermediate keratinocytes (KRT1⁺), peridermal keratinocytes (KRT4⁺), fibroblasts (PRRX1⁺), cycling cells (MKI67⁺), and neuron cells (PLP1⁺) (Figure 8A). Interestingly, we observed that the chemokine signaling pathway genes including CCL2 and CXCL2 are expressed in the keratinocytes and fibroblasts, and their receptors such as CCR4 and CCR10 are also weakly expressed (Figure 8B). This suggests that these cytokine genes might be important for enhancing skin cell competency to generate hair-bearing skin organoids. We then used adult cells, which lose their regenerative, ability to test this hypothesis.

Figure 8.

ScRNA-seq analysis shows chemokine signaling pathway gene expression in human skin organoid culture

(A) T-distributed stochastic neighbor embedding (tSNE) plots show cell clustering and representative marker gene expression in each cluster. (B) tSNE plots show representative chemokine signaling pathway gene expression.

Our study demonstrated the principles governing morphological transitions in the skin organoid culture of newborn mouse cells result in hair regeneration.¹⁸ We further tested if this principle can be applied to restore adult cells to self-organize during skin organoid culture. By RNA-seq analysis of adult and newborn mouse cells, we observed that chemokine signaling pathway genes including CCL and CXCL genes are significantly decreased in adult cells, compared with the newborn mouse cells (Figure 9A). Particularly, CCL2 and CCR4 expression is significantly decreased in adult mouse cells organoid cultures compared with the newborn mouse cell organoid cultures (see also Figures S4A and S4B). Then we asked if the addition of CCL2 in the organoid culture of adult cells can restore their self-organizing behavior and regenerative ability (Figure 9B). Based on the stepwise procedure that we set up before (Figure 9C),¹⁸ we added CCL2 recombinant protein from D0 to D3, and added Wnt10a recombinant protein from D1 to D4. We then added MMP14 from D3 to D7 to trigger the coalescence of aggregate. In addition, we added protein kinase C (PKC) inhibitors throughout the cultivation period to decrease epidermal cell differentiation, which we identified as the main reason that adult epidermal cells lose their competence to regenerate hairs.²²

Figure 9.

Functional restoration of adult cells to regenerate hair follicles by applying CCL2 to adult cell cultures in organoid culture

(A) Volcano plots of RNA-seq analysis show chemokine signaling pathway genes downregulated in adult skin organoid culture compared to that of the newborn organoid culture. (B) Schematic of organoid culture and transplantation. (C) Environmental reprogramming procedure for adult organoid culture. CCL2 recombinant protein was not added to the control group, but PKC inhibitor (iPKC), Wnt10a, and MMP14 recombinant protein were added to both groups. (D) K14 immunostaining shows self-organization in adult skin organoid culture and hair regeneration upon transplantation. (E) Quantification of aggregate size at D2, coalescence rate at D4, planarization rate at D7, and hair regeneration after addition of CCL2 into the organoid culture of PPD0 cells. ∗∗p < 0.01, n ≥ 3.

Immunostaining and quantification for K14 show that the addition of CCL2 leads to an enlarged aggregate in size, compared with that of the control with small aggregate formation (Figures 9D and 9E). We previously verified that a single addition of Wnt3a can enlarge the aggregate size and induce MMP expression to contribute to aggregate coalescence, but with a less dense hair regeneration upon transplantation, suggesting that additional factors are required to reactivate the adult cells to self-organize. Indeed, in addition to the enlarged aggregate in size, we observed that more dermal cells were attached to the epidermal aggregate, indicating an enhanced epidermal-dermal interaction upon CCL2 addition (Figure 9D). As expected, the addition of MMPs led to the coalescence of the aggregate at D4 followed by the formation of a planar skin at D7, in which the epidermal layer is localized at the culture insert side and the dermal layer faces the air side. In addition, we observed immune cells by using scRNA-seq analysis of the telogen skin (see also Figures S5A and S5B). The restored adult organoids were then grafted onto the dorsum of nude mice, where significantly more hairs were regenerated compared with that of the control group (Figures 9D and 9E), suggesting that the immune signaling might be activated after CCL2 induction in the organoid. These data suggest that the stepwise addition of factors following the environmental reprogramming protocol can successfully reactivate the self-organizing ability of adult cells, which can regenerate hair follicles upon transplantation.

Discussion

Skin appendages such as hairs, feathers, and spines can regenerate after they are subjected to an environmental insult resulting in falling off. This conserved ability in animals raises the possibility that there might be an evolutionary similarity to maintaining the appendages. In the present study, we tested how mechanical stimuli influence HFSC activity using the hair-plucking model, and observed that the PPD3 skin cells can self-organize during skin organoid culture and have the greatest hair regenerative ability upon transplantation. We identified the increased expression of the chemokine signaling pathway genes and enhanced multipotency of the skin epithelial cells after hair plucking. Addition of CCL2 to the organoid culture can reactivate the adult skin cells to regenerate hair follicles upon transplantation.

Molecular mechanism of plucking-induced hair regeneration

It is known that hair plucking can induce hair regeneration, which is attributed to the destruction of the inner bulge niche where inhibitory factors such as FGF18 and BMP6 are secreted to maintain the quiescence of stem cells located in the outer bulge. Interestingly, by establishing a topological patterned hair-plucking model, Chen et al.¹⁷ observed that the immune signals can be activated to respond to the spatial extent of injuries, launching a full-scale regenerative response that is identified as a mechanism of quorum sensing. However, stem cell activation can be stalled without sufficient signaling input. For example, dermal papilla-secreted FGF7 and FGF10 can activate stem cells in hair germ but not bulge stem cells, leading to failed further expansion or down growth of the hair follicle.²⁹ This indicates that subsequent signaling pathways are required to orchestrate new hair growth.

With this perception in mind, we tested the temporal activation of HFSCs that fully acquire the regenerative ability upon hair plucking. We found that the skin cells at PPD3 have the greatest hair regenerative ability compared with those from the other time points, including PPDs (0, 1, 2, 4, and 10). This suggests that a balanced signaling network is achieved in PPD3 skin cells. This balance may include the decreased expression of inhibitors such as BMPs (2 and 4) which are known molecules that maintain stem cell quiescence, and the increased expression of activators Shh, Noggin, FGFs, and follistatin, which can induce anagen reentry. Indeed, our RNA-seq data show decreased inhibitors and increased activators in PPD3 compared with PPD0 skin cells. We also observed that the PPD10 skin cells, in which hair follicles enter full anagen, have a decreased hair regenerative ability compared with those of the PPD3 skin cells during skin reconstitution, which is likely attributed to the high expression of BMPs in dermal adipocytes that contribute to the quiescence of HFSCs.³⁰

CCL2 and multipotency of skin cells

As micro-environmental factors, cytokines and immune cells not only influence physiological HFSC activation, but also contributed to injury-induced hair regeneration. However, our study identified that cytokines including CCL2 have dual roles in plucking-induced hair regeneration. This includes recruiting the immune cells, which has been reported.¹⁷ Another role is enhancing the stemness of the skin cells, which is characterized by our skin organoid culture model.

First, we observed that CCL2 is highly expressed in skin epithelial cells quickly after hair plucking, which is consistent with the previous finding that although used a distinct hair plucking model.¹⁷ CCL2 attracts immune cells including macrophages which secrete TNF-α. Upon accumulating to a threshold, TNF-α stimulates the nuclear factor-κB pathway that ultimately induces hair regeneration through the activation of Wnt signaling. Our study identified that CCL2 expression is significantly increased in PPD3 skin cells, indicating an identical mechanism to that of the topological plucking-induced hair regeneration.

Second, our findings suggest that CCL2 is involved in stimulating stem cell multipotency. We observed that increased stem cell activation coincides with CCL2 expression level. We also found that the addition of CCL2 recombinant protein can enhance the regenerative ability of skin cells using the skin organoid culture system, shown as enlarged cell clusters, increased multipotency, and significantly more hair regeneration after graft. CCL2 has been identified to activate hypoxia-related genes to enhance the pluripotency of human-induced pluripotent stem cells.³¹ CCL2 over-expressed human umbilical cord-derived mesenchymal stem cells (hUC-MSCs) showed better functional recovery relative to naive hUC-MSCs.³² CCL2 can also induce increased capacity for self-renewal, proliferation, and differentiation in neural stem cells.³³ These studies imply that CCL2 correlates with the multipotency of stem cells. Although it is currently unknown how CCL2 enhances the multipotency of skin cells, it is reported that CCL2 regulates breast cancer stem cells by mediating cross-talk between cancer cells and stromal fibroblasts,³⁴ indicating that CCL2 can be a mediator between different cell types, which may enhance cell interaction mutually.

Organoids as a promising strategy to regenerate skin tissues

Organoids are 3D cultures of progenitor cells that can self-organize to form complex tissue patterns, mimicking those generated in development, and become powerful models for investigating the development and regeneration of various organs in regenerative medicine.³⁵ We established the skin organoid using mouse skin progenitor cells, which regenerate skin and hair follicles upon transplantation.¹⁸ This model can be used to investigate the skin developmental process, epithelial-mesenchymal interaction, skin tissue integration, and drug screening, as well as hair follicle morphogenesis. While stem cells or progenitor cells have high plasticity to self-organize during organoid culture, adult cells almost lose their self-organization ability, leading to no hair regeneration when grafted to the back of the nude mice. Combining using the hair plucking and skin organoid culture models, we showed that the PPD3 cells have a higher self-organization ability compared with the cells harvested from other time points. It is reasonable to speculate that this is partially due to increased CCL2 expression by the epithelial cells which may increase the multipotency of the skin cells, since the addition of CCL2 to the adult cell organoid culture can largely restore the adult cells to self-organize and regenerate hair follicles. Another possibility can be the interactions between skin cells and immune cells, which may contribute to the increased multipotency and hair regenerative ability, although this requires further investigation in the organoid culture system. Nevertheless, immune cells are widely involved in physiological hair regeneration¹³ and wound-induced hair neogenesis.^14,15

These findings demonstrate the concept that adult cells can be restored to regenerate when faced with external stimuli or guided by a series of morphological phase transition-like processes. Thus, our results by using in vivo hair plucking assay and in vitro organoid culture of adult mouse cells offer a promise to improve the regenerative ability of human skin cells, which has clear clinical applications. Our study also establishes the regulatory relationship between immune modulation and organoid culture, which is also significant for future clinical transplantation where immunologic rejection occurs, if any.

Materials and methods

Mice

CD1 mice were purchased from Charles River Laboratories. K14H2BGFP mice were kindly gifted by Dr. Elaine Fuchs at Rockefeller University. Mice were housed in a vivarium at a 22°C constant temperature, with a daily food and water supply. All procedures were performed upon approval of the Institutional Animal Care and Use Committee of Chongqing University.

Hair plucking and sampling

Adult mice were anesthetized and the dorsal back skin was smeared with melt wax. After the wax cooled down, hairs were plucked manually from the tail to the head region. The skin samples with hair-plucked regions were harvested at PPD0, PPD1, PPD2, PPD3, PPD4, and PPD10 (Figure 1B). Then the skin samples were fixed with 4% paraformaldehyde to do the following staining or dissociated into single cells to do organoid culture.

Bulk RNA-seq and analysis

RNA-seq was performed on replicate samples (PPD0 and PPD3) from 6-month-old mice. Total RNA was isolated with TRIzol reagent (Invitrogen), the cDNA library was constructed according to TruSeq RNA sample preparation v2 kit (Illumina Biotechnology), and the libraries were sequenced by Hi-seq 2000. Raw data were normalized and then the gene expression values were log2 transformed. For clustering analysis, differentially expressed genes among PPD0 and PPD3 were determined by limma. A false discovery rate of less than 0.05 and log2 fold change or more than 1 were used as a threshold to determine significant differences in gene expression. Differential expression gene enrichment and functional annotation analysis were performed by R package cluster Profiler 4.0.

ScRNA sequencing analysis

The scRNA-seq data of human skin organoid (GSE147206) and telogen skin cells (GSE129218) were download from the GEO database. We reanalyzed the skin organoids (cells were cultured at D29 and D48) data generated using WA25 hESCs cell line. For the quality control, cells with less than 200 genes and genes expressed in fewer than 3 cells we filtered out, then the data were data normalized by Seurat 4.0. tSNE was used for dimension reduction. From all calculated principal components, only the top 20 principal components were used for FindNeighBors, FindClusters, and RunTSNE functions. The differential genes of each cell clusters are carried out through the Findermakers function. We annotated cell types through SingleR package. We checked the expression level of immune-related genes (Ccl4, Ccl9, and Ccr4) in feature plot from organoid data.

Quantitative RT-PCR

RNA extraction and cDNA preparation are the same as the preparation for the bulk RNA-seq. The annealing temperature for PCR is set at 60°C. The primers used this study include: CCLl2 (F: GCATTAGCTTCAGATTTACGGGT, R: GCATTAGCTTCAGATTTACGGGT), CXCL10 (F: CCAAGTGCTGCCGTCATTTTC,R: GGCTCGCAGGGATGATTTCAA), CXCL12 (F: TTAAAAACCTGGATCGGAACCAA, R: GCATTAGCTTCAGATTTACGGGT), CXCL13 (F: TTCTCTGTACCATGACACTCTGC, R: CGTGGAATCTTCCGGCTGTAG), Fst (F: TGCTGCTACTCTGCCAGTTC, R: GTGCTGCAACACTCTTCCTTG), and Shh (F: AAAGCTGACCCCTTTAGCCTA, R: TTCGGAGTTTCTTGTGATCTTCC).

Skin organoid culture and transplantation

Skin organoid culture was established in our previous study.¹⁸ Briefly, skin samples harvested on different PPDs were separated into epidermal and dermal cells using trypsin and collagenase, respectively. Epidermal and dermal cells were remixed at a ratio of 1:5 and cultured on a trans-well culture insert at 37°C PPD0 adult cells were used for the environmental reprogramming experiment in which PKC inhibitor (5 μM, bisindolyl maleimide I, Beyotime), CCL2 (25 ng/mL, BioLegend, 578404), Wnt10a (10 ng/mL, Proteintech, 26238-1-AP), and MMP14 (10 ng/mL, R&D, 918MP010) recombinant proteins, were added (Figure 9C). DMEM/F12 culture medium with the addition of 10% fetal bovine serum was changed every other day. The procedure set up in the planar hair-forming assay was used for the organoid transplantation.²³

ALP staining

For ALP staining, skin samples were washed twice with cold phosphate-buffered saline (PBS) and were fixed in paraformaldehyde at 4°C overnight. The samples were incubated with nitro blue tetrazolium/5-bromo-4-chloro-30-indolyl phosphate solution. The reaction was stopped by washing with PBS. Then 20-μm cryosections were washed with PBS and incubated in NBT/BCIP solution (Roche Diagnostics) for 30 min. The slides were incubated in 100% xylene for 30 s and then mounted using resinene (Thermo Fisher Scientific).

Hematoxylin and eosin staining

The skin tissues were collected, dehydrated, paraffinized, embedded in paraffin, and sliced into 6 μm in thickness. The paraffin sections were rehydrated through xylene and graded ethanol in different concentrations. Then the slides were incubated in hematoxylin for 3 min and eosin for 100 s. The images were taken under a light microscope.

Immunostaining

Tissues were fixed in 4% paraformaldehyde at 4°C overnight. Paraffin sections were deparaffinized and rehydrated in different concentrations of ethanol. The slides were autoclaved or microwaved for 15 min in 10 mM sodium citrate buffer for antigen retrieval. The samples were blocked with 2% BSA in PBS for 1 h at room temperature. The samples were incubated with primary antibodies at 4°C overnight. After wash, the samples were labeled with secondary antibodies at room temperature for 1 h, and counterstained with 4′,6-diamidino-2-phenylindole or propidium iodide for 30 min. The images were taken using an LSM900 confocal microscope (Carl Zeiss Inc.). Primary antibodies include AE13 (ab16113, Abcam), AE15 (gift), CD34 (ab81289, Abcam), Sox9 (ab71762, Abcam), K10 (ab76318, Abcam), K14 (A01432, Boster), CD86 (ab53004, Abcam), F4-80 (sc-25830, Santa Cruz), CD163(sc-33560, Santa Cruz), CD206(sc-58986, Santa Cruz), TNF-α (AF8208, Beyotime), NCAM (from Dr. Cheng-Ming Chuong laboratory), P16 (ab51243, Abcam), P63 (12143-1-AP, Proteintech), P-cadherin (13-2000Z, Thermo), and PCNA (ab92552, Abcam).

In situ hybridization

The DIG-labeled single-stranded RNA probes were used to detect the expression of the gene of interest in paraffin-embedded sections. Samples were deparaffinized and rehydrated. The slides were washed in xylene and different concentrations of ethanol, and digested with 20 μg/mL proteinase K in pre-warmed 50 mM Tris at 37°C for 15 min and rinsed in distilled water. Immerse slides in ice cold 20% (v/v) acetic acid for 20 s and dehydrate the slides by washing for approximately in diverse ethanol. Add 100 μL hybridization solution to each slide and incubate the slides in a humidified hybridization chamber at the desired hybridization temperature for 1 h. The probes were diluted in hybridization solution in PCR tubes and heated at 95°C for 2 min in a PCR block. Then 50 μL diluted probes were added per section and incubated in the humidified hybridization chamber at 65°C overnight. Then washed away excess probe and hybridization buffer with higher temperatures (up to 65°C) for short periods twice in maleic acid buffer containing Tween 20 (MABT) at room temperature for 30 min. Slides were transferred to a humidified chamber blocked for 1.5 h at room temperature. The anti-label antibody anti-ALP (ab108337, Abcam) was added at the required dilution in blocking buffer and incubated at room temperature for 1–2 h then washed with MABT. The slides were washed with pre-staining buffer (100 mM Tris pH 9.5, 100 mM NaCl, 10 mM MgCl₂) and dried in air for 30 min. Wash in 100% ethanol and air dry thoroughly. The slides were sealed with DePeX solution.

Loss-of-function assay

All C57BL/6J mice were fed with Formulab Rodent Diet (PMI Nutrition International) ad libitum. Mice at 8 weeks of age were selected for subcutaneous injection. After the mice were anesthetized with intraperitoneal drug injection, and hairs were plucked manually from the tail to the head region. The dorsal skins were subcutaneously injected with 50 μL iCCL2 (10 μM pirfenidone, AMR69, CAS# 53179, MCE), iCCR4 (10 μM plerixafor octahydrochloride, CAS# 155148, MCE) or saline in the controls at PPD0 and PPD1. The skin samples of PPD3 were fixed with 4% paraformaldehyde for the following examination.

Scanning electron microscopy

The excised samples were fixed with 2% formaldehyde and 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.0) at 4°C. When the biopsies were submitted for processing, they were post-fixed in 1% osmium tetroxide, dehydrated in a graded ethanol series (70%–80% to 95%–100%), and critical-point dried. Dried samples were mounted on aluminum stubs using a double-sided carbon tab, sputter coated with gold. Secondary electron micrographs were recorded using an FEI Quanta 600F scanning electron microscope (ThermoFisher) at an acceleration voltage of 10 kV.

Statistics and reproducibility

The experiments in this study were performed with at least three biological replicates unless specified. All statistical analyses were performed with GraphPad prism 6.0 software or R Bioconductor. The tests used and the p values are listed in the figure legends. The p values less than 0.05 indicated with asterisk (∗p = 0.05–0.01, ∗∗p < 0.01–0.001, ∗∗∗p < 0.001) were considered statistically significant.

Data availability statement

The data are available from the corresponding author on reasonable request.