The human microbiome has a fundamental role in skin homeostasis, barrier function and host immune networks. Specifically, a complex and multispecies microbiome population is required for healthy skin. Acute and chronic skin conditions are often associated with imbalances in host–microbiome–immune interactions. Despite recent advances in skin microbiome research, there are limited models for studying these interactions in healthy and diseased skin; therefore, we have developed a model to investigate how different microbiome profiles alter epidermal homeostasis and T-cell responses. Full-thickness three-dimensional (3D) skin models, developed using primary dermal fibroblasts and keratinocytes isolated from adult skin, were colonized with five human microbiome species: Staphylococcus epidermidis (SE), Staphylococcus capitis (SC), Cutibacterium acnes (CA), Malassezia restricta (MR) and Malassezia globosa (MG), in singular or combined inoculations (5M). Significant changes in epidermal thickness were observed in models colonized with single-species inoculations compared with sterile control models; however, no significant changes in epidermal thickness were observed in models colonised with 5M inoculations. The effect of colonization on epidermal differentiation and barrier formation was observed via the expression of K14, K10, E-cadherin and Ki67. Expression of E-cadherin significantly decreased in models colonized with SC, MR and MG compared with sterile control models. Despite significant changes in epidermal thickness, only MG showed a significant decrease in Ki67 expression compared with sterile control models. No significant changes were observed in K10 and K14 expression following microbial inoculation. Overall, 5M inoculation showed no significant changes in biomarker expression or epidermal thickness compared with sterile control models suggesting a combination of microbes supports epidermal homeostasis as observed in vivo. Commensal microbes have been shown to regulate epidermal barrier function via the aryl hydrocarbon receptor (AhR). Increased nuclear expression of AhR was observed in response to SC, CA and 5M, suggesting these microbes have a greater role in AhR activation. To investigate microbiome–immune interactions in the skin, the expression of CD69, cutaneous lymphocyte antigen (CLA) and HLA-DR on T cells was assessed in response to incubating peripheral blood mononuclear cells with conditioned media from inoculated models. SE caused a significant increase in CD3+/CLA+ cells and 5M significantly reduced CD3+/HLA-DR+ cells. RNA sequencing aims to support current analysis and elucidate any more targets to be explored. Our findings suggest different microbial profiles alter epidermal barrier formation and immune responses.

Background Acute cutaneous inflammation causes microbiome alterations as well as ultrastructural changes in epidermis stratification. However, the interactions between keratinocyte proliferation and differentiation status and the skin microbiome have not been fully explored. Objectives Hypothesizing that the skin microbiome contributes to regulation of keratinocyte differentiation and can modify antimicrobial responses, we examined the effect of exposure to commensal (Staphylococcus epidermidis, SE) or pathogenic (Staphylococcus aureus, SA) challenge on epidermal models. Methods Explant biopsies were taken to investigate species-specific antimicrobial effects of host factors. Further investigations were performed in reconstituted epidermal models by bulk transcriptomic analysis alongside secreted protein profiling. Single-cell RNA sequencing analysis was performed to explore the keratinocyte populations responsible for SA inflammation. A dataset of 6391 keratinocytes from control (2044 cells), SE challenge (2028 cells) and SA challenge (2319 cells) was generated from reconstituted epidermal models. Results Bacterial lawns of SA, not SE, were inhibited by human skin explant samples, and microarray analysis of three-dimensional epidermis models showed that host antimicrobial peptide expression was induced by SE but not SA. Protein analysis of bacterial cocultured models showed that SA exposure induced inflammatory mediator expression, indicating keratinocyte activation of other epidermal immune populations. Single-cell DropSeq analysis of unchallenged naive, SE-challenged and SA-challenged epidermis models was undertaken to distinguish cells from basal, spinous and granular layers, and to interrogate them in relation to model exposure. In contrast to SE, SA specifically induced a subpopulation of spinous cells that highly expressed transcripts related to epidermal inflammation and antimicrobial response. Furthermore, SA, but not SE, specifically induced a basal population that highly expressed interleukin-1 alarmins. Conclusions These findings suggest that SA-associated remodelling of the epidermis is compartmentalized to different keratinocyte populations. Elucidating the mechanisms regulating bacterial sensing-triggered inflammatory responses within tissues will enable further understanding of microbiome dysbiosis and inflammatory skin diseases, such as atopic eczema.

Abstract Disclosure: I. Pelsma: None. R. van den Berg: None. S. Kooijman: None. M. Snel: None. G. Zhu: None. A. Moore: Employee; Self; Unilever. J. Pople: Employee; Self; Unilever. S. Lavrijsen: None. M. Birch-Machin: None. R. Bhogal: Employee; Self; Unilever. A. El Ghalbzouri: None. B. Martinez-Tellez: None. T. Giesbrecht: Employee; Self; Unilever. S. Kersten: None. M. van Weeghel: None. O.C. Meijer: None. A.M. Pereira Arias: None. D. Gunn: Employee; Self; Unilever. P. Rensen: None. N. Biermasz: None. Introduction: Chronic and acute short sleep associate with many adverse effects. Metabolic and neuroendocrine effects during the day following short sleep, however, are unknown. Methods: Randomized cross-over study using short sleep (4h) vs normal sleep (8h) in 31 males (31 years (IQR 23-47, BMI 26.0±4.2 kg/m2 (range 20.1-35.0). Glucose, lipid, incretin, ACTH, and cortisol levels, as well as respiratory exchange ratio (RER; indirect calorimetry) were assessed following standardized mixed meals. RNAseq of morning skeletal muscle biopsies and pathway analysis on differentially expressed genes was performed. Results: Short sleep increased fasting free fatty acids (0.34 mmol/L (IQR 0.26 - 0.38) vs 0.47 mmol/L (IQR 0.38 - 0.70), P<0.0001). Cortisol levels peaked upon early wake-up, which eliminated a peak at normal wake-up. Morning postprandial glucose peaked later (time P<0.001, levels P=0.106, pattern P=0.031), whilst insulin was unaffected. Afternoon postprandial insulin was lowered (time P<0.001, levels P=0.044, pattern P=0.419), whilst glucose was unaffected. Morning and afternoon postprandial TC, and LDL-c were lowered (morning: TC, time P<0.0001, levels P=0.009, pattern P=0.289; LDL-c, time P<0.001, levels P=0.038, pattern P=0.797; afternoon: TC, time P<0.001, levels P=0.007, pattern P=0.92; LDL-c, time P<0.001, levels P=0.055, pattern P=0.991). Pathway analysis of muscle revealed increased fatty acid oxidation (FAO), reflected by increased Oxidative Phosphorylation and Mitochondrial Dysfunction pathways, accompanied by lowered fasting RER (0.87 (IQR 0.86 - 0.92) vs 0.86 (IQR 0.84 - 0.87), P=0.0033). Conclusions: Acute partial sleep restriction subtly and temporarily changed glucose dynamics (morning), and insulin levels (afternoon), and lowered TC and LDL-c levels in the morning and afternoon, with concurrent increased FAO and mitochondrial dysfunction pathways in skeletal muscle, and calorimetric shifting from glucose towards lipid oxidation (fasted and postprandial), potentially highlighting increased delipidation of chylomicrons following sleep restriction. Presentation: Thursday, June 15, 2023

The ability to reliably predict and infer cellular responses to environmental exposures would offer a major advance in the investigation of immune regulation in health and disease. One possible approach is the use of in silico modelling. Design of such a mathematical kinetic model would be based on existing knowledge of a biological system and utilise a partial data set to parameterise. However, the process of parameter estimation, key for the accuracy of the model, is difficult to conduct by hand, and thus a computational alternative is necessary. We report the utility of Genetic Algorithm with Rank Selection (GARS) as a parameter estimation tool on multiple biological models, including heat shock, signal transduction via ERK, circadian rhythm and NFκB systems, where it showed strong accuracy and superiority to the Extended Kalman Filter method, Algebraic Difference Equations, and MATLAB fminsearch approaches. GARS parameter estimation is a valuable tool for biological data because it reliably infers system behaviour from partial data sets, allowing for the prediction of cellular responses to environmental exposures.

Hair follicles (HFs) are immersed within dermal white adipose tissue (dWAT), yet human adipocyte‒HF communication remains unexplored. Therefore, we investigated how perifollicular adipocytes affect the physiology of human anagen scalp HFs. Quantitative immunohistomorphometry, X-ray microcomputed tomography, and transmission electron microscopy showed that the number and size of perifollicular adipocytes declined during anagen‒catagen transition, whereas fluorescence-lifetime imaging revealed increased lipid oxidation in adipocytes surrounding the bulge and/or sub-bulge region. Ex vivo, dWAT tendentially promoted hair shaft production, and significantly stimulated hair matrix keratinocyte proliferation and HF pigmentation. Both dWAT pericytes and PREF1/DLK1+ adipocyte progenitors secreted HGF during human HF‒dWAT co-culture, for which the c-Met receptor was expressed in the hair matrix and dermal papilla. These effects were reproduced using recombinant HGF and abrogated by an HGF-neutralizing antibody. Laser-capture microdissection‒based microarray analysis of the hair matrix showed that dWAT-derived HGF upregulated keratin (K) genes (K27, K73, K75, K84, K86) and TCHH. Mechanistically, HGF stimulated Wnt/β-catenin activity in the human hair matrix (increased AXIN2, LEF1) by upregulating WNT6 and WNT10B, and inhibiting SFRP1 in the dermal papilla. Our study demonstrates that dWAT regulates human hair growth and pigmentation through HGF secretion, and thus identifies dWAT and HGF as important novel molecular and cellular targets for therapeutic intervention in human hair growth and pigmentation disorders.

Adiponectin reportedly stimulates proliferation and elongation of human scalp hair follicles (HFs) ex vivo. In the current study, we investigated how adiponectin oligomers produced by perifollicular dermal white adipose tissue (dWAT), a potent source of adiponectin isoforms, influence human HF proliferation and pigmentation. To do so, we treated microdissected, organ-cultured HFs in the presence or absence of dWAT with a recombinant human adiponectin oligomer mix, or inhibited dWAT-derived adiponectin using a neutralizing antibody. Multiplex qPCR (Fluidigm) revealed that adiponectin oligomers downregulated pigmentation genes KITLG, PMEL and TYRP1 and Wnt genes AXIN2, LEF1 and WNT10B. In situ hybridization showed that adiponectin downregulated AXIN2 and LEF1, and up-regulated DKK1 within the dermal papilla (DP), a highly unusual transcriptional profile for a putative hair growth-promoting agent. Adiponectin oligomers also downregulated protein expression of the HGF receptor c-Met within the matrix and DP. However, adiponectin did not alter hair matrix keratinocyte proliferation within 48 h ex vivo, irrespective of the presence/absence of dWAT; HF pigmentation (Masson-Fontana histochemistry, tyrosinase activity) was also unchanged. In contrast, neutralizing adiponectin isoforms within HF + dWAT increased proliferation, melanin content and tyrosinase activity but resulted in fewer melanocytes and melanocytic dendrites, as assessed by gp100 immunostaining. These seemingly contradictory effects suggest that adiponectin exerts complex effects upon human HF biology, likely in parallel with the pro-pigmentation effects of dWAT- and DP-derived HGF. Our data suggest that dWAT-derived ratios of adiponectin isoforms and the cleaved, globular version of adiponectin may in fact determine how adiponectin impacts upon follicular pigmentation and growth.

In human and rodent skin, the hair follicle (HF) mesenchyme (connective tissue sheath [CTS]), is densely populated with macrophages (Bertolini et al., 2013Bertolini M. Meyer K.C. Slominski R. Kobayashi K. Ludwig R.J. Paus R. The immune system of mouse vibrissae follicles: cellular composition and indications of immune privilege.Exp Dermatol. 2013; 22: 593-598Crossref PubMed Scopus (11) Google Scholar, Christoph et al., 2000Christoph T. Müller-Röver S. Audring H. Tobin D.J. Hermes B. Cotsarelis G. et al.The human hair follicle immune system: cellular composition and immune privilege.Br J Dermatol. 2000; 142: 862-873Crossref PubMed Scopus (235) Google Scholar, Paus et al., 1998Paus R. Van Der Veen C. Eichmüller S. Kopp T. Hagen E. Müller-Röver S. et al.Generation and cyclic remodeling of the hair follicle immune system in mice.J Invest Dermatol. 1998; 111: 7-18Abstract Full Text Full Text PDF PubMed Scopus (100) Google Scholar). Notably, the number of these perifollicular macrophages (pMΦ) fluctuates greatly during the murine hair cycle, being the highest during hair growth (anagen), decreasing in regression (catagen), and reaching the lowest count around resting (telogen) HFs (Paus et al., 1998Paus R. Van Der Veen C. Eichmüller S. Kopp T. Hagen E. Müller-Röver S. et al.Generation and cyclic remodeling of the hair follicle immune system in mice.J Invest Dermatol. 1998; 111: 7-18Abstract Full Text Full Text PDF PubMed Scopus (100) Google Scholar). In rat skin, macrophage-like cells show similar hair cycle-associated fluctuations, along with a switch toward a fibroblast growth factor-5+ phenotype that promotes catagen (Suzuki et al., 1998Suzuki S. Kato T. Takimoto H. Masui S. Oshima H. Ozawa K. et al.Localization of rat FGF-5 protein in skin macrophage-like cells and FGF-5S protein in hair follicle: possible involvement of two Fgf-5 gene products in hair growth cycle regulation.J Invest Dermatol. 1998; 111: 963-972Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar). Only later was it demonstrated that pMΦ can regulate HF cycling, while apoptotic pMΦ around murine telogen HFs secrete Wnt signals that can activate quiescent epithelial HF stem cells inducing anagen (Castellana et al., 2014Castellana D. Paus R. Perez-Moreno M. Macrophages contribute to the cyclic activation of adult hair follicle stem cells.PLOS Biol. 2014; 12: e1002002Crossref PubMed Scopus (57) Google Scholar). However, the role of pMΦ in human HF biology remains obscure, despite abnormalities in their number and activation status being recognized in inflammatory hair diseases (Harries et al., 2013Harries M.J. Meyer K. Chaudhry I. E Kloepper J. Poblet E. Griffiths C.E. et al.Lichen planopilaris is characterized by immune privilege collapse of the hair follicle's epithelial stem cell niche.J Pathol. 2013; 231: 236-247Crossref PubMed Scopus (94) Google Scholar) (Supplementary Text S1a). As a basis for hypothesis building, we asked what happens to the pMΦ number, distribution and polarization during catagen development of healthy organ cultured anagen scalp HFs (Langan et al., 2015Langan E.A. Philpott M.P. Kloepper J.E. Paus R. Human hair follicle organ culture: theory, application and perspectives.Exp Dermatol. 2015; 24: 903-911Crossref PubMed Scopus (56) Google Scholar), as this is clinically the most relevant for hair loss disorders (Paus and Cotsarelis, 1999Paus R. Cotsarelis G. The biology of hair follicles.N Engl J Med. 1999; 341: 491-497Crossref PubMed Scopus (802) Google Scholar) and can be followed ex vivo. To investigate, human HFs were obtained with ethics committee approval, institutional approval and informed, written patient consent. These were organ-cultured for 1 to 6 days collecting anagen, early-, mid- and late HFs for comparison (Kloepper et al., 2010Kloepper J.E. Sugawara K. Al-Nuaimi Y. Gáspár E. van Beek N. Paus R. Methods in hair research: how to objectively distinguish between anagen and catagen in human hair follicle organ culture.Exp Dermatol. 2010; 19: 305-312Crossref PubMed Scopus (78) Google Scholar). HF sections were analyzed by immunofluorescence microscopy for the macrophage marker CD68 (Barros et al., 2013Barros M.H.M. Hauck F. Dreyer J.H. Kempkes B. Niedobitek G. Macrophage polarisation: an immunohistochemical approach for identifying M1 and M2 macrophages.PLOS ONE. 2013; 8: e80908Crossref PubMed Scopus (377) Google Scholar) (for methods, see Supplementary Text S2). Consistent with previous research (Christoph et al., 2000Christoph T. Müller-Röver S. Audring H. Tobin D.J. Hermes B. Cotsarelis G. et al.The human hair follicle immune system: cellular composition and immune privilege.Br J Dermatol. 2000; 142: 862-873Crossref PubMed Scopus (235) Google Scholar), pMΦ predominantly localized to proximal CTS around the proliferative hair bulb as well as the central CTS, with more than half of all CD68+ pMΦ detected in the peribulbar CTS of anagen VI HFs (Figure 1a and b, Supplementary Figure S1a). The number of pMΦ decreased progressively to the infundibulum and was maintained in all the hair cycle stages (Figure 1a and b, Supplementary Figure S1). Immuno-electron microscopy confirmed that these CD68+ cells showed the ultrastructural characteristics of macrophages (Arismendi-Morillo et al., 2010Arismendi-Morillo G. Castellano-Ramírez A. Medina Z. Ultrastructural characterization of macrophage-like mononuclear leukocytes in human astrocytic tumors.Ultrastruct Patho. 2010; 34: 321-326Crossref PubMed Scopus (3) Google Scholar) (Supplementary Figure S2). When human HFs entered catagen, the pMΦ number rapidly decreased to reach a minimum in late catagen (Figure 1a and c), along with a 1.5-fold decrease in CD68 transcript levels (Figure 1d). The fact that freshly isolated, non-cultured HFs in (anagen/catagen/telogen) also showed this cycle-dependent decline in CD68+ pMΦs (Supplementary Figure S3) excludes the possibility that this observation represents an organ culture artifact. Next, we asked whether the catagen-associated decline in human pMΦs resulted from apoptosis, as in mice (Castellana et al., 2014Castellana D. Paus R. Perez-Moreno M. Macrophages contribute to the cyclic activation of adult hair follicle stem cells.PLOS Biol. 2014; 12: e1002002Crossref PubMed Scopus (57) Google Scholar). Indeed, as the HFs progressed through catagen, the number of apoptotic macrophages (CD68+/TUNEL+ immunofluorescence) increased significantly (Figure 1e and f). Therefore apoptosis, rather than emigration from the CTS, is likely responsible for this decline. As murine pMΦ are known to secrete Wnts, we asked whether they do so in the CTS of human HFs, thus contributing to anagen maintenance. Both WNT10a and 7b, which is also produced by murine pfMΦ (Castellana et al., 2014Castellana D. Paus R. Perez-Moreno M. Macrophages contribute to the cyclic activation of adult hair follicle stem cells.PLOS Biol. 2014; 12: e1002002Crossref PubMed Scopus (57) Google Scholar), were highly expressed by macrophages in anagen with protein levels decreasing as they enter catagen (Figure 2a and b). Regarding RNA, in situ hybridization (Supplementary Text S2) demonstrated that LEF1 mRNA, a down-stream Wnt mediator (Supplementary Figure S1), and WLS mRNA, found in Wnt-secreting cells (Supplementary Figure S2), were highly expressed around the anagen bulb. However, the WNT10a mRNA was low in the CTS, being restricted to the hair matrix tips (Supplementary Figure S4). AXIN2 mRNA was also expressed in low levels in the CTS during anagen (Supplementary Figure S4). During catagen, AXIN2 and LEF1 were expressed in the CTS, while WLS and WNT10a were almost absent. These data suggest that the human pMΦ niche is Wnt active and suggests that macropahges secrete Wnt ligands. Given the critical role of Wnt signaling in hair growth control (Geyfman et al., 2015Geyfman M. Plikus M.V. Treffeisen E. Andersen B. Paus R. Resting no more: re-defining telogen, the maintenance stage of the hair growth cycle.Biol Rev. 2015; 90: 1179-1196Crossref PubMed Scopus (89) Google Scholar, Schneider et al., 2009Schneider M.R. Schmidt-Ullrich R. Paus R. The hair follicle as a dynamic miniorgan.Curr Biol. 2009; 19: R132-R142Abstract Full Text Full Text PDF PubMed Scopus (634) Google Scholar), this raises the question whether pMΦ are one important source of WNTs (Castellana et al., 2014Castellana D. Paus R. Perez-Moreno M. Macrophages contribute to the cyclic activation of adult hair follicle stem cells.PLOS Biol. 2014; 12: e1002002Crossref PubMed Scopus (57) Google Scholar) (Supplementary Text S1b). As macrophages exist in different phenotypes, pMΦ polarization was explored using the dual immunofluorescence of CD68 with the M1 marker CD86 or RAGE, which is the receptor for advance glycation end-products (RAGE being upregulated by an M1 phenotype [Jin et al., 2015Jin X. Yao T. Zhou Z.Z. Zhu J. Zhang S. Hu W. et al.Advanced glycation end products enhance macrophages polarization into M1 phenotype through activating RAGE/NF- κ B pathway.Biomed Res. 2015; 2015732450Google Scholar]). Although the CD86 expression was low (Figure 2c) (< 5% of pMΦ), RAGE was highly expressed in anagen in 25% of the pMΦ, decreasing though catagen (Figure 2d) to 10% by late catagen. Interestingly, advance glycation end-products are produced in proliferative tissues such as tumors (Supplementary Figure S4). As RAGE is upregulated and stimulated by high levels of advance glycation end-products (Jin et al., 2015Jin X. Yao T. Zhou Z.Z. Zhu J. Zhang S. Hu W. et al.Advanced glycation end products enhance macrophages polarization into M1 phenotype through activating RAGE/NF- κ B pathway.Biomed Res. 2015; 2015732450Google Scholar), RAGE+ pMΦ might bind to and remove advance glycation end-products from the HF microenvironment during proliferative anagen, consistent with a previous hypothesis that increased advance glycation end-products levels detected by pMΦs may promote catagen induction (Supplementary Figure S4). Conversely, an M2 phenotype (CD206/CD163) (Martinez and Gordon, 2014Martinez F.O. Gordon S. The M1 and M2 paradigm of macrophage activation: time for reassessment.F1000Prime Rep. 2014; 6: 13Crossref PubMed Scopus (2907) Google Scholar) was least present in anagen and early/mid catagen but rapidly increased to 50% (CD206) and 30% (CD163) (Figure 2e and f) of all the macrophages in late catagen, representing a distinct hair cycle associated phenotypic switch. As an M2 phenotype is linked with tissue remodeling (S6), this switch may be important for the extensive catagen-associated HF remodeling. An exception to this was the high levels of CD206 around the anagen bulge region (40% of pMΦ), (Supplementary Figure S1b) where they may contribute to immune privilege (Supplementary Text 1e). Limiting causal investigation, the selective depletion of pMΦs by clodronate (Castellana et al., 2014Castellana D. Paus R. Perez-Moreno M. Macrophages contribute to the cyclic activation of adult hair follicle stem cells.PLOS Biol. 2014; 12: e1002002Crossref PubMed Scopus (57) Google Scholar) failed repeatedly in our ex vivo human model, potentially because of insufficient tissue penetration by these large liposomes (Supplementary Figure S3, Supplementary Figure S4, Supplementary Table S1, and Supplementary Text S1c) Taken together, our pilot study demonstrates that pMΦ dramatically change in number, activity and phenotype during the anagen-catagen transformations of human scalp HFs and suggest that these cells may engage in both Wnt secretion and AGE recognition, particularly in the bulb in human HFs, a region that changes extensively during the anagen-catagen switch (Kloepper et al., 2010Kloepper J.E. Sugawara K. Al-Nuaimi Y. Gáspár E. van Beek N. Paus R. Methods in hair research: how to objectively distinguish between anagen and catagen in human hair follicle organ culture.Exp Dermatol. 2010; 19: 305-312Crossref PubMed Scopus (78) Google Scholar, Oh et al., 2016Oh J.W. Kloepper J. Langan E.A. Kim Y. Yeo J. Kim M.J.M. et al.A guide to studying human hair follicle cycling in vivo.J Invest Dermatol. Elsevier, Inc. 2016; 136: 34-44Abstract Full Text Full Text PDF PubMed Scopus (167) Google Scholar). Therefore, pMΦ deserve to be systematically explored as therapeutic intervention targets in human hair growth control. Research data that is necessary to interpret the information presented here will be made available to any researcher, with minimal reuse restrictions. All the tissue was obtained with written and informed patient consent adhering to the 'Declaration of Helsinki Principles' with ethical and institutional approval from the University of Manchester. All samples, slides and biological material were tracked and stored according the 'Human Tissue Act' guidelines' in an HTA licensed freezer. Jonathan Alan Hardman: https://orcid.org/0000-0002-1653-7908 Ferhan Muneeb: https://orcid.org/0000-0002-2564-9792 Jenny Pople: https://orcid.org/0000-0002-3560-8994 Ranjit Bhogal: https://orcid.org/0000-0003-4835-4871 Asim Shahmalak: https://orcid.org/0000-0003-1411-3666 Ralf Paus: https://orcid.org/0000-0002-3492-9358 This work was funded by Unilever in the form of a research grant awarded to RP. All other authors state no conflict of interest. We would like to thank R Marotta/IIT for support with IM-TEM and acknowledge Kim Mace for reviewing the manuscript. This work was supported by grants from Unilever, UK, and the NIHR Manchester BRC (BRC-1215-20007) "Inflammatory Hair Diseases" program. Conceptualization: JAH, RP, JP, RB; Data Curation: JAH, FM; Formal Analysis: JAH, FM; Funding Acquisition: RP, RB; Project Administration: JAH, RP; Tissue acquisition: AS; Writing – Original Draft Preparation: JAH, RP; Writing – review and editing: JAH, JP, FM.