The current state of knowledge of the immune ecosystem in alopecia areata

Abstract

Alopecia areata (AA) is an autoimmune disease that affects approximately 2% of the general population. Patients with AA most commonly present with one or more patches of hair loss on the scalp in defined circular areas. A fraction of patients progress to more severe forms of the disease, in some cases with involvement of all body surfaces. The healthy anagen stage hair follicle is considered an immune privileged site, described as an environment that suppresses inflammatory immune responses. However, in AA, this immune privileged state collapses and marks the hair follicle as a target for the immune system, resulting in peri- and intrafollicular infiltration by lymphocytes. The complexity of the inflammatory ecosystem of the immune response to the hair follicle, and the relationships between the cellular and soluble participants, in AA remains incompletely understood. Many studies have demonstrated the presence of various immune cells around diseased hair follicles; however, often little is known about their respective contributions to AA pathogenesis. Furthering our understanding of the mechanisms of disease in AA is essential for the novel identification of targeted therapeutics that are efficacious and have few unintended effects.

1. Introduction

The incidence of autoimmune diseases has recently been increasing in the worldwide population [1]. Autoimmune diseases arise when there is a loss of immune tolerance towards self-antigens, resulting in immune responses targeting self-tissue. It is generally understood that interactions between genetic and environmental factors are associated with the risk of developing autoimmunity [2]. Alopecia areata (AA) is a common autoimmune disease of the hair follicle that has an approximate lifetime risk of 2.1% [3]. The mean onset of disease in patients is between 25–36 years of age and is associated with increased rates of psychological disorders, including depression and anxiety, and a diminished quality of life [4]. Treatment options range from local to systemic immunosuppressants and are often unsatisfactory for patients with longstanding and severe disease [5]. However, the complexity of the immune landscape in AA and difficulties in parsing the critical cell types and pathways has complicated therapeutic discovery. Identifying how specific cells and pathways may be contributing to the pathogenesis of AA will help pave the way to developing effective treatment options for patients and provide better outcomes.

With this review, we aim to compile a comprehensive overview of our current state of knowledge of the participants of inflammation, both cellular and soluble, in AA. In particular, we focus on factors thought to be contributing to the immunoregulatory environment of the immune privileged hair follicle, the mechanisms contributing to its collapse, and the immune milieu associated with autoimmunity to the hair follicle in AA. Strides into our understanding of disease pathogenesis have resulted in identification of novel treatments in AA in the last several years, and further identification and addressing gaps in our understanding of disease mechanisms increase the likelihood that further progress will be made.

2. Clinical presentation of disease

AA manifests as non-scarring hair loss. The disease course is relatively unpredictable and can affect any hair-bearing location on the body. In most cases, hair loss develops in well-defined, circular areas on the scalp [6]. On the face, hair loss may emerge within the beard area as well as within the eyebrows and eyelashes [6]. The disease can progress to include total loss of scalp hair, termed alopecia totalis (AT), or complete loss of hair on the entire body, termed alopecia universalis (AU) [7]. Mild cases often undergo spontaneous remission, marked by hair regrowth. This usually occurs within the first year after the development of disease, and spontaneous relapse may occur at any time [4, 7, 8]. Descriptions of “overnight graying” have been used to describe AA due to preferential loss of pigmented hairs and relative sparring of non-pigmented for white/grey hairs [6], although other mechanisms may explain this observation [9]. Furthermore, emergence of non-pigmented hairs in lesions exhibiting regrowth is not uncommon [10]. The baseline hair color is not thought to impact the loss of hair [11]. Nail involvement is another characteristic seen in patients with AA. In particular, pitting and trachyonychia, or a sandpaper-like appearance, are most often described in association with AA [6, 7]. Although the morphology of lesions in AA usually has a distinctive appearance, the severity, location, and course are, in general, highly variable.

3. Genetic risk factors

Numerous studies of various methodologies examining gene relationships with AA have been conducted. Many of the earlier studies used focused or targeted techniques to identify associations with specific loci or genes (Table I) [12–21]. Genome wide assessments have more recently been used to perform more comprehensive and unbiased analyses of susceptible genes in AA patients. The first genome wide association study (GWAS) was performed on a US population of patients [22] and identified eight susceptibility loci/variants. Several of these loci were found to be closely associated with genes implicated in autoimmunity and within T cell pathways including: IL2/IL21 on chromosome 4q27; CTLA4 and ICOS on chromosome 2q33.2, ULBP3 and ULBP6 on chromosome 6q25.1; IL2RA on chromosome 10p15.1, IKZF4 on chromosome 12q13; and closely bundled HLA-DRA/HLA-DQA1/HLA-DQA3/HLA-DQB2/BTNL2/NOTCH4/MICA on chromosome 6p21.32. STX17 on chromosome 9q31.1 and PRDX5 on chromosome 11q13 were also implicated. A follow-up GWAS performed on a European population reproduced five of the susceptibility loci/genes previously identified including: ULBP3/ULBP6, PRDX5, IL2/IL21, IKZF4/ERBB3, and STX17 [23]. Furthermore, this study found two additional loci, closely associated with the genes IL13 and KIAA0350/CLEC16A, with variants demonstrating increased susceptibility to AA. Two additional GWASs, performed by Forstbauer et al and Betz et al, examined patients from the US and Europe and identified SPATA5 on chromosome 4q27-q28, ACOXL/BCL2L11 on chromosome 2q13, and C11ofr30/LRCC32 on chromosome 11q13.5 as additional variants with increased susceptibility in AA patients [24, 25]. In most cases, it is currently unknown how these variants affect gene expression and function, complicating translating these findings into insights into disease pathogenesis.

Table 1.

Single targeted gene studies with variants identified to be associated with AA.

Gene	AA associated variants†
HLA	DQB1*03 (DQ3) 12 DRB1*1104 (DR11) 12 DQB1*0301 (DQ7) 12 DRB1*0401 (DR4) 12
IL1RN	IL-1RN*2 13
IL4	Intron 3 VNTR polymorphism 14
IL17RA	rs879577 SNP 15
IL17A	GG genotype in A7488G polymorphism 16
IL1RN/IL1L1	IL1RN+2018/IL1L1+4734 SNPs 17
AIRE	AIRE961G 18
FOXP3	rs2294020–3675(A) SNP 19
ICOSLG	rs378299–509(C) SNP 19
MICA	MICA*5.1 & MICA*6 20
MIF	MIF-173*C 21

^†Nomenclature as per reference

VNTR= variable number tandem repeat, SNP= single nucleotide polymorphism

4. Mouse models of AA

In 1994, Sundberg et al. reported that C3H/HeJ mice undergo spontaneous hair loss with features similar to that of human AA, including defined regions of hair loss undergoing a waxing and waning cycle, regrowth of depigmented hairs in the regions of alopecia, and mononuclear cell infiltration of the hair follicle bulb and epithelium [26]. Additionally, similarly to the infiltrate around human AA hair follicles, the hair follicle infiltrate of alopecic C3H/HeJ mice was comprised predominantly of CD8 and CD4 T cells. Spontaneous disease development affected approximately 15–20% of C3H/HeJ mice, with a strong preponderance for female mice. In addition, the onset of disease was relatively protracted in male mice; female mice developed disease as early as 2–3 month of age, whereas male mice first began exhibiting hair loss at 7 months of age.

The ability to experimentally induce AA in the C3H/HeJ strain has increased its usefulness as a laboratory model. Transfer of skin from previously-affected mice led to the induction of AA in recipient mice with substantially diminished latency [27]. Engraftment using skin from either the dorsal or ventral side of an affected mouse induced hair loss in as little as 7 weeks post-surgery and in all recipients by 11 weeks. Furthermore, grafted skin from unaffected hairy donors onto AA affected mice resulted in loss of hair within the graft by 21 days post-surgery.

Injection of lymphocytes isolated from AA affected mice has similarly been found to induce AA in previously unaffected recipient mice. Subcutaneous injections of unsorted skin draining lymph node (LN, SDLN) cells, sorted CD4 T cells from SDLNs, and sorted CD8 T cells from SDLNs were able to induce hair loss in recipient mice [28]. In this work, unseparated LN cells induced widespread hair loss in 80% of mice injected, whereas sorted CD8 T cells induced hair loss in about 3 to 5 weeks that was localized to the site of injection in 100% of mice, and CD4 T cells induced diffuse hair loss in 7 weeks in 43% of mice with limited involvement at the site of injection. Recently, Wang et al. reported a similar method of inducing disease using a stimulated population of unsorted lymphocytes derived from SDLNs of AA affected mice [29]. In this study, LN cells co-cultured in vitro with anti-CD3/anti-CD28-coated beads were intradermally injected into eight-week-old recipient C3H mice. Approximately 90% of recipient mice experienced hair loss between 2 to 4 weeks following transfer. It was noted that the route of injection was critically important for disease induction in this model, as subcutaneous injections were less effective at inducing hair loss than intradermal injections. This innovation additionally allowed for a greater disease-inducing capacity from individual donor mice, as the lymphocyte population from a single donor mouse could be used to induce disease in dozens of recipient mice.

Both graft- and cell-induction models are capable of recapitulating characteristics consistent with those observed in AA patients and C3H/HeJ mice with spontaneous disease. Examination of the follicles within lesional skin from mice of each model shows dystrophic hair follicles and perifollicular infiltration of CD4 and CD8 T cells, although the unmanipulated LN cell transfer study [27] solely reported mononuclear infiltration without further identification. The relatively simple implementation of these models makes them particularly attractive for use in the laboratory.

In addition to the C3H/HeJ mouse model, there are several reports of use of xenograft mouse models of AA. In one described model, skin biopsies from severe AA patients were transplanted onto severe combined immunodeficient (SCID) mice and observed for hair regrowth [30]. These grafts were then injected with T cells extracted from AA scalp skin biopsies that had been cocultured with irradiated patient peripheral blood mononuclear cells (PBMCs) and follicular homogenate. In over half of the grafts that received activated T cell injections, there was an induction of hair loss nine to ten weeks following injection, which was not observed in grafts that received injections of uncultured AA scalp T cells, unirradiated PBMCs, or no cells. An additional humanized mouse model of AA has more recently been reported in which skin from a healthy donor was grafted onto SCID mice [31]. After three months, grafts were injected subcutaneously with an in vitro activated PBMC population from the same healthy donor that had been bead-enriched for natural killer group 2 membrane D (NKG2D)⁺/CD56⁺ lymphocytes. In the grafts that received the activated enriched lymphocyte populations, hair loss was appreciated between three to five weeks following transfer.

Another murine model of AA was recently described in the C57BL/6J mouse strain [32]. In this model, Alli et al. surreptitiously discovered a T cell receptor (TCR) that, when used in a retrogenic bone marrow chimera model, could be used to generate a mouse model of autoimmune hair loss. Following transplantation of bone marrow cells transduced to express the TCR, recipient mice developed hair loss as early as six weeks post transplantation. These mice experienced waxing and waning of the extent of disease, with a median of 3 cycles before achieving complete loss of hair. The hair follicles in lesional skin showed intrafollicular CD3⁺ mononuclear infiltration of the anagen follicles, a similar feature of AA in humans. However, these mice exhibited discrepancies with human AA, including the presence of surface epidermal changes as well as scratching behaviors, both of which may be indicative of the presence of other factors not seen in human AA or the C3H AA model. In addition, fixing of the TCR in this model to that associated with CD8 T cells also complicates study of the role of other T cell types, including conventional CD4 T cells and regulatory T cells, in disease development.

5. Immune privileged status of the hair follicle

5.1. Concept of immune privilege and relationship to hair follicle biology

The hair follicle is an immune privileged organ. Immune privilege (IP) is believed to be a state in which inflammation is limited and the recognition of foreign antigens is restricted [33]. Multiple organs have been described as being immune privileged sites including the eyes [34], brain [35], testes [36], fetus [37], liver [38], gut [39], and hair follicles [40]. Experimentally, the first description of immune privilege was made in 1948 by Peter Medawar, where he showed that, in previously unimmunized rabbits, skin grafted into the anterior chamber of the eye or left cerebral hemisphere would not be rejected, in contrast with grafts onto previously immunized rabbits [41]. He proposed that this discrepancy was due to a passive mechanism, where the grafts were rejected only once lymphatic drainage was present through vascularization of the grafts, allowing for an immune reaction to occur. This idea of immune privilege was first described specifically in the context of the hair follicle by Billingham, when he presented experiments in which skin from black guinea pigs grafted onto white guinea pigs [42]. Whereas the grafted skin failed to regain pigment, hairs from within the engrafted skin were retained and emerged as donor black. This observation is consistent with the concept that the follicular bulb in the grafted skin was associated with a protective environment that allowed the melanocytes to survive and avoid immediate destruction by the host immune system.

5.2. Mechanisms of hair follicle IP

Similar to other organs that exhibit IP, the hair follicle possesses an array of factors that lead to its ability to regulate immune cell responses and mitigate potential autoimmunity. There are low levels of major histocompatibility complex (MHC) class I expression in the cells of the lower/proximal hair follicle [43–45], indicating that the presentation of antigens to autoreactive CD8⁺ T cells is decreased. There is also diminished expression of components involved in the formation of the MHC class I complex, including transporter associated with antigen processing-2 (TAP2) and β2 microglobulin (β2m) [44, 45]. Furthermore, there are few MHC class II-presenting cells and few MHC class II-presenting intraepithelial dendritic cells of the hair follicle [45], indicative of reduced capacity to stimulate CD4 T cell responses in the steady state. In AA, collapse of immune privilege at the hair follicle is thought to be a critical event that is needed for the development of disease; this is characterized by an upregulation of MHC class I and class II molecules within and in association with the follicular epithelium, which is seen in human skin and mouse models of AA.

Studies using in vitro cultures of microdissected hair follicles and immunohistological analyses of skin samples have expanded our understanding of how the follicular epithelium and surrounding cells enforce hair follicle IP by identifying a handful of secreted molecules which can impede an immune response. Transforming growth factor beta 1 (TGFβ1), a molecule often associated with diminished immune responses, was identified to be expressed at lower levels around the follicles of AA patients as compared to healthy controls [46]. When added into culture with microdissected human hair follicles, TGFβ1 led to decreased MHC class I expression that had been induced by interferon (IFN)-γ [44]. In addition, in in vitro hair follicle cultures, both alpha-melanocyte stimulating hormone (α-MSH), a neuropeptide known to induce melanogenesis [47] and highly expressed in the bulge region of the hair follicle in healthy skin [48], and insulin growth factor 1 (IGF1) also led to downregulated expression of MHC class I in the follicular epithelium [44]. IGF1 was also shown in these studies to induce downregulation of β2m, TAP2, and MHC class II expression.

Macrophage migration inhibitory factor (MIF) may be contributing to hair follicle IP and its dysregulation may be contributing to AA pathogenesis. Two groups identified strong expression of MIF in healthy control hair follicles [48, 49]. Additionally, it was found that MIF expression is greatly reduced in lesional samples from AA patients [49]. Its published role as a suppressor of natural killer (NK) cell function [50], in combination with these data, supports a potential role in hair follicle IP.

Expression of programmed death ligand-1 (PDL1), an inhibitory molecule that signals through PD-1 present on T cells to induce anergy, may be another mechanism by which the hair follicle regulates T cell activation. Isolated cells of the hair follicle including dermal sheath cup cells (DSCC) and dermal papilla (DP) cells express high levels of PDL1 when compared to non-follicular fibroblasts [51]. When allogenic T cells were placed into a coculture system with DSCC and DP cells, the T cells exhibited a diminished IFNγ response, decreased proliferation, and an increased expression of caspase molecules, as compared to non-follicular fibroblast cells which do not express PDL1. These differences seen could be mitigated by using siRNA technology to knock down PDL1 expression. These data suggest that PDL1 expression by cells of the hair follicle may be contributing to hair follicle IP in part by directly impacting T cell function. Increased expression of PDL1 and PDL2 on fibroblasts in AA lesions in the C3H model has been described; how this relates to AA pathogenesis is unclear [52].

Fas-Fas ligand (FasL) interactions mediate cellular apoptosis and has been proposed as a immunoregulatory mechanism in the IP of the eye [53], raising questions of its potential role in hair follicle IP. In a skin graft-induced AA model, mice deficient in Fas and FasL showed a resistance to developing AA [54]. Additionally, when skin from Fas or FasL was grafted onto an AA affected mice, the grafts remained protected from losing hair. Whether the Fas/FasL pathway is a mechanism of hair follicle IP or is acting in another manner has yet to be determined. Fas is expressed by the cells of the hair follicle in mice with and without AA, and FasL positive cells in the perifollicular region are low in unaffected mice but are increased in AA mice [54]. Additionally, the hair follicles in AA mice show a higher number of apoptotic cells as compared to the unaffected hair follicles. This is suggestive that this pathway occurs during or following the collapse of hair follicle IP, as the perifollicular infiltration contains a higher number of FasL positive cells, which could then induce apoptosis of the hair follicle cells.

The collapse of these mechanisms leaves the hair follicle vulnerable to being targeted by the immune system, creating an environment in which AA can develop. How cells of the immune system and cytokines may be contributing to the pathogenesis of disease is explored below.

6. Cytokines, signaling pathways, and immune cells in AA

6.1. Interferon gamma (IFN-γ)

IFN-γ plays an important role in a range of processes including host defense against pathogens, inflammation, immune cell responses, and tumor immunity [55]. IFN-γ has been implicated as a critical contributor in the pathogenesis of AA. Using hair follicles microdissected from human scalp skin, Ito and colleagues found that low dose IFN-γ treatment induces a significant upregulation of MHC class I in the hair follicle epithelium in areas where they had previously seen low or absent expression [44]. Additionally, IFN-γ co-culture led to upregulated expression of TAP2 and β2m, machinery associated with and required for MHC class I presentation [44]. In this manner, IFN-γ is thought to be driving the collapse of the hair follicle IP, causing upregulation of MHC class I on the follicular epithelium and allowing for recognition by and stimulation of autoreactive T cells.

With an important role in hair follicle IP, multiple groups have asked what role IFN-γ may play in AA pathogenesis and examined AA patient samples for its presence. Microarray analyses conducted on human and C3H/HeJ AA skin identified increased IFN response gene expression signatures in AA-affected lesional skin when compared to skin from non-diseased controls [56]. Studies assessing levels of circulating cytokines have reported higher IFN-γ levels in patients with AA as compared to healthy controls [57–60]. Additionally, the capacity of IFN-γ to induce AA or contribute to AA pathogenesis has been examined in the C3H/HeJ mouse model, with studies showing conflicting findings. In one set of studies, Gilhar and colleagues injected low dose IFN-γ or saline intravenously for three consecutive days and then weekly in C3H/HeJ mice that had been shaved to initiate an anagen hair cycle [61]. They found seven of nine mice who received injections of IFN-γ experienced hair loss, whereas none of the saline injected group did. In contrast, Sundberg et al. injected IFN-γ or phosphate buffered saline (PBS) subcutaneously three times in succession following depilation in recipient mice, and found that only 2 of 10 mice that received IFN-γ injections experienced hair loss; they concluded from this that mice did not exhibit an increased rate of hair loss compared to what would have been expected for spontaneous disease [62]. A more recent study found that subcutaneous injection of IFN-γ in combination with polyinosinic-polycytidylic acid was able to induce hair loss in approximately 80% of recipient mice; in contrast, mice injected with IFN-γ alone did not exhibit hair loss [63]. This last study suggests induction of AA by IFN-γ is enhanced by additional inflammatory signals. In in vivo models, the induction of AA in C3H mice using AA skin grafts could be prevented in the setting of IFNγ-deficiency [64] or antibody-mediated neutralization of IFN-γ [56]. In IFN-γ deficient mice and anti-IFN-γ antibody-treated mice, weak expression of MHC class I was associated with the lack of disease development, indicating IFN-γ may be a vital player in the development of AA through the collapse of hair follicle IP and upregulation of MHC Class I, among other possible contributions.

6.2. Common gamma chain (γ_c) cytokines

Common gamma chain (γ_c) cytokines, so named by their shared use of the γ_c subunit as a part of their respective receptors play an important role in the biological functions and regulations of the adaptive immune system [65]. This family of cytokines, comprised of interleukin (IL) −2, IL-4, IL-7, IL-9, IL-15, and IL-21 are notable for providing critical survival signals and influencing the function of T and B cells. Analysis of lesional AA skin from both human and mice identified a genetic signature from this family of cytokines, supporting a role in AA pathogenesis [56].

Functionally, IL-2 is important in promoting CD8 T cell activity, driving naïve CD4 T cell differentiation, and maintaining T regulatory cells [66], and can therefore have context-dependent pro- or anti- inflammatory roles. Expression of the IL-2 receptor (IL-2R) is recognized in two forms: the low affinity dimeric IL-2R and the high affinity trimeric IL-2R [66]. In lesional skin from AA patients, expression of IL-2 and the IL-2Rα and β subunits were found to be upregulated [56, 67, 68]. Additionally, serum IL-2 levels have been reported to be increased in patients with AA as compared to healthy controls in two studies [69, 70], although a third study found that serum IL-2 levels in AA patients were significantly decreased compared to controls [71]. Furthermore, serum IL-2 levels were found to be the highest in patients with the lowest duration of disease (<1 year) [69].

Several studies have examined the effect of manipulating IL-2 or IL-2 receptor signaling on AA pathogenesis. Neutralization of IL-2 in C3H/HeJ mice that received skin grafts from AA donor mice prevented development of disease and resulted in diminished effector CD8 T cells in the skin [56]. These data are in line with data from mouse models exhibiting reduced IL-2 expression. Heterozygotic IL-2^+/− mice, that exhibit reduced IL-2 expression but do not suffer from the high mortality observed in homozygotic IL-2^−/− mice, developed disease at a decreased rate as compared to IL-2^+/+ controls [72]. Lesional skin from IL-2^+/+ mice showed dense intrafollicular infiltrates of CD8 T cells and perifollicular infiltration by CD4 T cells, while lesional skin from IL-2^+/− showed reduced numbers of CD8 and CD4 T cells. While these studies would indicate that IL-2 acts in a pro-inflammatory manner, small studies examining therapeutic use of IL-2, in order to stimulate the formation of an anti-inflammatory regulatory T cell population, in the clinical setting have been performed. In one pilot study, patients with severe AA that had failed to respond to prior systemic therapies were treated with low dose IL-2 injections [73]. From these treatments, four out of five patients responded with some amount of hair regrowth. Examination of the skin before and after treatment found that those patients who responded to treatment showed a decrease in CD8 T cell infiltration and an increase in the presence of T regulatory cells. However, in a follow-up multicenter randomized trial where patients with severe AA were treated with low dose IL-2 injections, IL-2 treatment failed to show an improvement in hair regrowth [74]. Further studies will be required to determine if IL-2- or IL-2 receptor-based therapies can serve as a treatment for AA.

IL-15 plays a critical role in the maintenance and survival of T cells as well as in the regulation of T cell cytotoxicity and NK cell development [75, 76]. In a manner unique among the γ_c cytokines, IL-15 binds to the IL-15Rα intracellularly and is shuttled to the membrane of the cell. In the context of this complex, IL-15 is presented in trans to cells expressing the IL-2Rβ/IL-15Rβ and γ_c chain heterodimer [77]. Similar to IL-2, messenger RNA (mRNA) of IL-15, and its receptor, were found to be upregulated in the skin of AA patients and AA affected mice [56, 68]. Furthermore, histological analysis of human and mouse AA skin shows upregulation of IL-15 and its receptor around the hair follicles when compared to that of healthy hair follicles [56]. Serum IL-15 levels have been examined across multiple studies and predominantly show increased levels in AA patients as compared to healthy controls [57, 78, 79], although one study failed to describe a difference in serum IL-15 levels between AA and control patients [80]. To determine if IL-15 contributed mechanistically to AA pathogenesis, Xing et al performed experiments in which IL-15 signaling was neutralized in mice following AA skin-grafting [56]. In contrast to AA skin-grafted mice treated with an isotype control antibody that robustly developed hair loss, AA skin-grafted mice that received IL-2Rβ/IL-15Rβ antibody treatments were protected from developing AA. In the context of this model of AA, IL-15 is therefore a necessary factor in AA disease emergence. In particular, IL-15 may be augmenting the function of a pathogenic population of CD8 T cells, discussed further below.

6.3. Janus kinase (JAK)-mediated signaling

Cytokines known to be critical in AA, such as IFN-γ and IL-15, bind to their receptors and activate JAKs, which go on to induce phosphorylation of signal transducer and activator of transcription (STAT) molecules and further transcription. The JAK family, made of JAK1, JAK2, JAK3 and Tyk2, mediate signaling of a wide variety of receptors for cytokines, hormones, and growth factors. For example, IFN-γ signals through JAK1 and JAK2, while IL-15 signals through JAK1 and JAK3. [81]. Spurred by recognition that numerous cytokine pathways that contribute to AA are mediated by JAK/STAT signaling, small molecule JAK inhibitors have been and continue to be studied as therapeutic treatment for AA as well as a variety of inflammatory and autoimmune diseases [82]. The US Food and Drug Administration (FDA) initially approved tofacitinib and ruxolitinib for the treatment of rheumatoid arthritis and myelofibrosis, respectively [83, 84]; these agents were soon thereafter explored for their efficacy in the treatment of AA. Murine AA skin graft recipients systemically treated with these inhibitors starting at the time of grafting were protected from developing hair loss, and mice with established disease demonstrated hair regrowth when these agents were used as a topical treatment [56]. In the same report, treatment of three human subjects with moderate to severe AA with ruxolitinib, an inhibitor of JAK1 and JAK2, orally twice daily resulted in near complete hair regrowth within 5 months of treatment. Other JAK inhibitors have been similarly assessed for their efficacy in the treatment of AA. A study looking at baricitinib, another FDA approved JAK1/2 inhibitor, demonstrated this agent similarly protected C3H AA-skin graft recipients from developing disease, successfully treated established disease in the mouse model, and reversed disease in a patient with AA [85]. Further clinical studies examining the effectiveness of JAK inhibitors in the treatment of moderate to severe AA, AU, and AT have emerged and are ongoing [86–92]. These studies have shown that the use of JAK inhibitors for the treatment of AA may be safe and effective; however, as is the case with immunosuppressive and biologic treatment for other cutaneous autoimmune diseases, JAK inhibitors do not induce a durable remission in patients with AA, as most patients appear to relapse upon discontinuation [91, 92]. Interestingly, AA patients have been shown to harbor a small but detectable population of clonally expanded T cells clones in scalp skin following treatment with tofacitinib [93], consistent with JAK inhibitors suppressing those critical effector functions of autoreactive T cells required for the disease but not completely eliminating them. More complete compilations of the use of specific JAK inhibitors in the treatment of human AA has been extensively reviewed by others [94, 95].

6.4. CD8⁺ cytotoxic T cells (CTL)

CD8 T cells are members of the adaptive immune system with cytotoxic capabilities and are found clustered around the bulb of the hair follicle in lesional AA skin [49]. An early murine AA study found that when CD8 T cells isolated from the SDLNs of AA mice were transferred subcutaneously into recipient mice, those mice exhibited localized hair loss around the injection site [28]. These data underscored that these cells are playing a role in disease, leading to future insights into disease mechanisms.

The first GWAS for AA identified the NKG2D ligands, UL16-binding protein (ULBP) 3 and ULBP6, as disease-associated genes, and complementary experiments showed that ULBP3 is highly expressed within the follicle of AA patients [22]. This work also demonstrated that CD8 T cells localized to the hair follicle in AA patients expressed NKG2D. Growing data support a central role for the NKG2D-expressing CD8 T cell population in disease pathogenesis [56]. NKG2D⁺ CD8 T cells are found in higher frequencies in the skin and SDLNs of affected mice compared with unaffected mice, where they are largely absent [56]. In an in vitro killing assay, CD8 T cells expressing NKG2D from AA mice demonstrated cytotoxicity towards NKG2D-ligand-expressing dermal sheath cells, and this effect could be partially blocked using antibodies specific for NKG2D. Furthermore, transfer of NKG2D⁺ CD8 T cells from the SDLNs of AA-affected mice was sufficient to induce disease in recipient mice following intradermal injection, whereas transfer of NKG2D^neg CD8 T cells from the same lymph nodes was not. These data provide compelling evidence that CD8 T cells expressing NKG2D are critical in autoimmune attack on the hair follicle in murine AA and AA patients. More work is required to identify the mechanisms behind how this cell population is contributing to AA.

As previously mentioned, the emergence of a pathogenic CD8 T cell population may be licensed at least in part from additional stimulatory input in the form of IL-15. IL-15 has been shown to contribute to the generation of lymphokine activated killer (LAK) cells, a heterogeneous population of cells with NK cell-like cytolytic functions [96]. Generation of these LAK cells has been examined through work focused on intraepithelial (IE) CD8 T cells, a population of cells found in the gut, which contribute to the pathogenesis of celiac disease (CD) [96]. Like NKG2D-expressing CD8 T cells in AA, patients with CD harbor a population of NKG2D-expressing CD8 T cells that localize to the intraepithelial region of the targeted tissue. In normal healthy intestines, IE CD8 T cells express low levels of NKG2D [96]. However, if those cells are stimulated with IL-15, they upregulate their expression of NKG2D and their cytolytic capability. In active CD patients, where upregulation of IL-15 in the intestinal epithelium is a hallmark characteristic, the IE CD8 T cells express increased levels of NKG2D as compared to normal IE CD8 T cells [77, 96].

Additionally, single cell sequencing identified a CD8 T cell cluster within the skin and SDLNs of alopecic mice which expressed NKG2D, but also other genes associated with NK cells, like natural killer cell granule protein 7 (NKG7) [97]. This pattern of gene expression is similar to that observed involving a CD8 T cell cluster identified from AA patients, and particularly enriched in lesional skin, in which NKG7 was also expressed [97]. In a recent publication, a group demonstrated that NKG7 expression on CD8 T cells is involved in the secretion of cytotoxic granules by exocytosis and mediating inflammation [98]. Using a model of severe malaria, where antigen-specific CD8 T cells trafficked to the brain, CD8 T cells that were deficient in NKG7 had reduced expression of activation markers, including CD11a, CD49d, and granzyme B, but also were not recruited to the site of inflammation. Furthermore, they found that NKG7 is colocalized with CD107a and contributes to the translocation of CD107a to the surface of the cell [98]. The role of NKG7 expression on the putatively pathogenic CD8 T cell population in AA, however, has yet to be defined.

The trafficking of AA-inducing CD8 T cells to disease sites has begun to be examined. CD8 T cells expressing NKG2D in the SDLNs of AA-affected mice express CXC receptor (CXCR) 3, a chemokine receptor that is associated with a type I inflammatory response in CD4 and CD8 T cells, which are localized to the follicle in lesional skin [99, 100]. Expression of CXCR3 on CD8 T cells is thought to contribute to trafficking of autoreactive CD8 T cells into the skin during disease development. Neutralization of CXCR3 blocked migration of the CD8 T cells into active AA skin, and AA skin graft mediated disease induction was greatly inhibited [99]. Expression of the CXCR3 ligands, CXC ligand (CXCL) 9, CXCL10, and CXCL11, are upregulated in the skin of AA affected mice [99]. These IFN-γ-induced ligands exhibit increased expression at the level of mRNA as early as 5 weeks post skin graft, prior to AA emergence in C3H mice [101]. Additionally, in the chronic phase of disease in AA patients, CXCR3⁺ CD8 T cells were present around the bulb of the hair follicle in high numbers [102]. The previously-mentioned single cell sequencing study provided complementary data confirming high CXCR3 expression by the putatively-pathogenic CD8 T cell population with shared TCR sequences in skin and SDLNs [97].

Although it is unclear which effector mechanisms CD8 T cells are employing to effect disease, there is data indicating that their cytotoxic capabilities play a role. Of the CD8 T cells characterized by single cell sequencing, the cluster of CD8 T cells identified that expressed high levels of CXCR3 also expressed high levels of granzyme B and perforin [97]. Among AA patients treated with squaric acid dibutylester, nonresponding patients exhibited a higher percent of granzyme B producing CD8 T cells in the skin when compared to that of patients that responded with favorable outcomes. In C3H mice that underwent skin graft induction, granzyme B transcripts were found to be increased in the skin 5 weeks post graft, an early timepoint in disease induction, suggesting that granzyme B or granzyme B-expressing cells may have a role in the early stages of disease development [101]. Despite these data, mechanistic studies supporting a role for cytotoxicity as a central mechanism of AA pathogenesis are lacking.

6.5. CD4 T helper cells

CD4 T cells, also called helper T cells, have been identified as perifollicular infiltrating immune cells of the hair follicle in AA patients [103], and there is growing appreciation of the varied composition of CD4 T cell types present (Figure 1). In the C3H mouse model, subcutaneous injection of isolated CD4 T cells from SDLNs of affected mice were capable of inducing systemic hair loss [28]. Furthermore, a potential effector role of CD4 T cells in inducing hair loss was also suggested in the Dundee experimental bald rat (DEBR) model of AA, as these rats experience lesions of hair loss that mimic that of human AA [104]. In this model, lesional hair follicles exhibited a mononuclear cell infiltration that consisted of perifollicular CD4 T cells and intrafollicular CD8 T cells and macrophages [104, 105]. Temporary depletion of CD4 T cells using monoclonal antibodies in this model led to partial hair regrowth, with resumption of hair loss when the CD4 T cell population began re-accumulating [106]. However, how CD4 T cells are contributing to AA, including which effector mechanisms are being employed to induce disease, have yet to be defined.

Figure 1. T cell populations in AA.

T cell populations present in human AA skin. CD4 and CD8 T cells are present in the peri- and intra-follicular regions of the hair follicle, respectively, in AA. CD8 T cells expressing NKG2D and CXCR3 have been appreciated as effector cells contributing to the development of AA. Other studies have suggested that the various CD4 T helper subsets (Th1, Th2, and Th17) are present around the hair follicle in human AA, identified in part by expression of their respective chemokine receptors. Whether these CD4⁺ T cell subsets have a pathogenic role in disease needs to be elucidated further. Additionally, further work to determine the role of regulatory T cells is needed.

6.5.1. T-helper type 1 (Th1)

Th1 cells are an effector helper subtype that produce of IFN-γ, IL-2, and tumor necrosis factor (TNF)-β [107]. T-box protein expressed in T cells (T-bet) is regarded as the master transcription factor for Th1 cells. Their primary function is thought to be the elimination of intracellular pathogens, but they have also been associated with organ-specific autoimmune diseases, including multiple sclerosis [108].

Serologic analysis of AA patients shows an upregulation of Th1 axis cytokines (IFN-γ, IL-2, and IL-12) and chemokines (CXCL10, CC receptor ligand (CCL) 2, and CCL3) in circulation [69, 78]. During acute phase disease (<6 months and patchy), patients have increased numbers of circulating and skin infiltrating CXCR3⁺ CD4 T cells, suggesting that these cells could be contributing to the early stages of disease pathogenesis [102]. Immunohistochemical analysis of skin biopsies showed CD4 T cells expressing CC receptor (CCR) 5, a marker associated with Th1 cells [109], infiltrating the intra- and peri-bulbar regions of the hair follicles, suggesting a potential role in AA [110]. Transcriptomic analyses demonstrate an upregulation of a Th1/IFN-associated genes CXCL9, CXCL10, and STAT1 in lesional AA when compared to either nonlesional AA or healthy control skin, as well as in nonlesional samples compared to healthy controls [68]. Additionally, expression of a similar set of markers were examined in samples from nonlesional and lesional scalp skin before and after treatment with methylprednisolone [67]. At baseline, lesional skin samples showed an upregulation of CXCL9, CXCL10, and IFN-γ when compared with non-lesional skin. Following treatment, there was decreased expression of CXCL9; other Th1 markers such as IFN-γ and CXCL10 trended lower but did not reach statistical significance, and markers of other T helper subsets, such as IL-13, were also decreased following treatment. Although it appears the presence of Th1 cells correlates with disease, further studies are needed to determine their role in AA.

6.5.2. T-regulatory cells (Treg)

Tregs play an important role in maintaining immunologic tolerance, whose function is to aid in the suppression of an immune response. Regulated by Forkhead box P3 (FoxP3), Tregs are often defined in two ways: (1) natural Tregs (nTregs), which are produced in the thymus as FoxP3 expressing cells or (2) induced Tregs (iTregs), which are naïve CD4 T cells that have been induced to differentiate in the periphery under certain circumstances [107]. Their immunoregulatory function raises the question of whether their dysfunction contributes to autoimmune diseases including AA, and they have been investigated as to whether they may be exploited as a potential therapeutic agent [111]. Histological examination of scalp biopsies from AA patients as well as patients with other cutaneous diseases, specifically lichen planopilaris, inflamed seborrheic keratosis, and non-specific folliculitis, showed a lower frequency of CD4⁺ FoxP3⁺ and CD25⁺ FoxP3⁺ Tregs in AA lesions compared to these other entities [112]. This finding is consistent with other work that found few FoxP3⁺ cells localized around the hair follicle in samples from AA patients with varied clinical presentations (multiple patchy, AT, and acute, diffuse, and total alopecia (ADTA)) [113]. Additionally, the percentage of lymphocytes that are FoxP3⁺/CD4⁺ found in peripheral blood are higher in healthy control patients when compared to that of AA patients; similarly, the percentage of lymphocytes that are FoxP3⁺ found in scalp skin samples were higher in healthy control samples compared to that of AA patients [114]. Somewhat paradoxically, the percentages of either FoxP3⁺ CD4 T cells in the circulation or FoxP3⁺ cells in the scalp lesions, of total lymphocytes, were found to be higher in lesions from patients with more severe disease when compared to those from patients with mild forms of disease, according to the same study. In the C3H/HeJ model, mice that received grafts from AA donors two weeks prior had fewer skin infiltrating CD4⁺CD25⁺ cells when compared to mice that received grafts from normal hairy mice [115]. The spleens and SDLNs of these mice also contained a smaller frequency of these Treg cells. In a study examining disease induction by way of transfer of different lymph node populations from AA mice, transfer of CD4⁺CD25⁺ lymphocytes at a 1:2 ratio with either CD25⁻ CD4⁺ or CD8⁺ T cells decreased the rate of disease development when compared to mice that received either CD25⁻ CD4⁺ or CD8⁺ T cells alone [28], indicative that these cells can play an immunosuppressive role in the context of AA. However, a more rigorous assessment of the mechanisms by which they are impacting AA development, are needed.

6.5.3. T-helper type 17 (Th17)

Published data indicates that Th17 cells and the cytokines they produce are present in AA, but their roles in disease pathogenesis, if any, is unknown. Among many reports, serum IL-17A levels are found to be elevated in AA patients when compared to healthy controls [57, 71, 116–118]. Furthermore, both protein and RNA expression of IL-17A are increased in lesional scalp tissue of AA patients [71, 118]. Other Th17-associated cytokines, including IL-21 and IL-22, have been assessed in AA patients. In a few reports, levels of both cytokines were found to be increased in the serum and lesional skin of patients [71, 116, 118]; however, in another study, no difference in serum IL-22 levels among AA and controls patients was observed [118]. RNA expression of the two subunits of IL-23 (p40, shared with IL-12, and IL-23p19), which plays a role in the differentiation of naïve CD4 T cells into the Th17 phenotype, were found to be upregulated in lesional skin of patients [68]. This upregulation was also reported at the protein level in the skin and serum of AA-affected mice [119]. However, neutralization of IL-23, using an antibody specific for the p40 subunit shared with IL-12, did not prevent mice from developing disease, suggesting that IL-23 does not play an active role in disease pathogenesis [119]. Lesional mRNA levels of both IL-17A and IL-22 were found to be positively correlated with disease subtype in patients, with the lowest levels in the healthy control group and highest in those with AU and AT [71]. However, a relationship between serum levels of IL-17A and disease severity is not consistently observed among reports. In one study, serum IL-17 levels were found to be correlated with both clinical type and disease severity [116], but that result was not found in a handful of other studies [57, 117, 118].

Several examinations focused on Th17 cells, rather than IL-17 or Th17-associated cytokines, have been conducted. Histological examination of AA affected hair follicles has shown an increase in infiltrating IL-17 producing lymphocytes as compared to the hair follicles from healthy controls [113, 114, 120]. One study, using CCR6 as an identifying marker, found intra- and peri-bulbar infiltration of Th17 cells in AA, with some cells localized around the bulge of the follicle, an area not typically associated with inflammation in AA [110]. Furthermore, a second study found a higher frequency of IL-17 producing lymphocytes around the hair follicle in patients with less severe disease (patchy subtype) as compared to more severe forms (AT) [113]. Additionally, Th17 cells have been found to be increased in the circulation of AA patients [114]. These cells were more abundant in patients with acute disease (< 6 months) when compared to patients with chronic disease (> 6 months). Whether there are Th17 or IL-17 producing lymphocytes present in the C3H/HeJ mouse model of AA is not known; however, in the MOG TCR mouse model, the pathogenic T cells produced IL-17 during disease development, and these levels correlated with disease progression [32].

Secukinumab, a FDA approved IL-17A inhibitor that is effective at treating plaque psoriasis, has been tested in AA patients for efficacy [121]. In a double-blind, randomized, prospective pilot study, 11 patients were enrolled to examine the potential for secukinumab as an AA treatment [122]. Of the patients enrolled, seven were placed on 300mg secukinumab treatment and four on a placebo treatment. There was a high attrition rate in the study, with 2 patients who received the drug and 1 in the placebo group remaining until the final endpoint, complicating drawing rigorous conclusions from the study. One of the two patients receiving the drug had an improvement in their Severity of Alopecia Tool (SALT) score, a clinical assessment metric that correlates with percent of scalp involvement, while the other had no change; the single patient in the placebo group had a worsening of their score. Multiple case reports of patients being given secukinumab for their psoriasis have shown conflicting results. In one report, after six weeks of treatment, a patient with a 25-year history of AU showed hair regrowth [123]. This patient exhibited hair loss four months following the cessation of treatment. In contrast, two other reports suggested that targeting the IL-17 pathways for psoriasis may have induced AA in the patients. In one report, a patient presented with hair loss in the beard area, which spread to the scalp, six months after starting secukinumab treatment for psoriasis [124]. In the second report, two patients with psoriasis vulgaris underwent treatment, the first receiving secukinumab and the second receiving brodalumab, an IL-17R antagonist; both showed development of AA within months after starting treatment [125]. Taken together, the current data on if and how Th17 cells and their cytokines are affecting the hair follicle and disease progression remain unclear.

6.5.4. T-helper type 2 (Th2)

Recent studies have suggested the presence of a Th2 signature in AA. Transcriptomic analysis of lesional and nonlesional samples from AA patients as well as control patient samples has shown an increase in molecules associated with a Th2 profile, including IL-13, CCL5, CCL13, CCL18, and CCL26 [68]. In a comparison of AA and control scalp skin sections, IL-5 and IL-10 mRNA expression were found to be increased in the deep dermis, defined as below the level of the sebaceous glands and inclusive of where hair bulbs would typically be found, despite the absence of differences in the expression of these cytokines in the upper dermis of AA and control samples [126]. Serologic analysis of AA demonstrated increased levels of the Th2 cytokine IL-13 and Th2-associated chemokines CCL13, CCL17, CCL22, and CCL26 when compared with healthy controls patients [78]. In both acute (<6 months) and chronic (>6 months) phases of disease, CD4 T cells expressing CCR4, a chemokine receptor expressed by Th2 cells, were found at higher frequencies in PBMCs from AA patients when compared to that of healthy controls [102].

A recent double-blind multicenter study looked at the efficacy of dupilumab, a monoclonal antibody that binds the alpha subunit of the IL-4 receptor, in the treatment of AA [127]. Dupilumab is an effective treatment option in patients with type 2 diseases, like atopic dermatitis and asthma, suggesting it may have positive effects in AA, as AA is commonly associated with atopic diseases [128–130]. In total, 60 patients were enrolled in this trial with the criteria of having AA longer than 6 months and with greater than 30% scalp hair loss. These patients were enrolled into two arms at a 2:1 ratio, one being treatment with dupilumab for the duration of the study (drug), and the second being given placebo for the first 24 weeks followed by dupilumab for the remainder of the study (placebo). At 24 weeks, those in the drug group showed a positive change in SALT score while those in the placebo group showed a negative change. Between week 24 and 48, patients in the drug group continued to show a positive change, although those in the placebo group also began to show a positive change in SALT score. At 48 weeks, 32.5% of the patients who had received the treatment showed at least a 30% improvement in SALT score. Additionally, patients with high serum IgE or familial atopy were more likely to respond to the drug. Further studies are required to determine whether the Th2 profile that can be seen in the skin of AA patients is a contributing factor to the disease itself, or whether it is having an indirect effect and whether targeting Th2 cytokines may be an efficacious therapeutic strategy.

6.6. γδ T cells

Observations of healthy hair follicles have identified few γδ T cells to be present; however, in AA skin, the number of γδ T cells is significantly higher when compared to healthy skin, and they appear to be localized to the bulb and suprabulbar epithelium [45, 131]. Because γδ T cells share some functional and phenotypic characteristics with conventional αβ T cells, including expression of NKG2D and secretion of cytotoxic molecules and pro-inflammatory cytokines, there is biological plausibility for their participation in AA pathogenesis [132]. The γδ T cells identified around healthy follicles were γδ1+, expressed skin homing receptors (CXCR3, CXCR4, and CCR2), and displayed a non-activated phenotype (NKG2D^dimCD69⁻) [131]. In contrast, γδ T cells found in AA skin, which were identified in both lesional and nonlesional skin, expressed NKG2D and were capable of producing IFN-γ. Paradoxically, however, there were a higher number of IFN-γ producing γδ T cells present in nonlesional skin when compared to lesional skin. How these cells may be contributing to AA requires further study.

6.7. Dendritic cells

Numerous studies have identified the presence of dendritic cells (DCs) around hair follicles of AA patients [67, 68, 126, 133, 134]. Studies have reported the presence of CD11c⁺ myeloid DCs [67, 68], plasmacytoid DCs (pDC) [133], and CD1a⁺ Langerhans cells [67, 126] in the DC compartment of AA lesions. Furthermore, unbiased single cell RNA-seq performed on murine AA skin identified antigen presenting cell (APC) signatures consistent with monocyte derived DCs (CCR2⁺ CD64⁺) among other cell types [97]. Adoptive transfer of CD11c⁺ cells isolated from the SDLNs of AA affected C3H mice was not sufficient at inducing disease in recipient mice [28]; a definitive role in AA pathogenesis therefore has yet to be defined.

Following activation, DCs are able to induce T cell responses through secretion of cytokines and presentation of antigens through MHC class I and II [135]. IL-12 and IL-23 are cytokines produced by DCs known to skew or maintain polarized T cell responses towards Th1 and Th17 differentiation, respectively. Analysis of AA skin by real-time PCR (RT-PCR) showed an increased expression of the shared IL-12/23 p40 subunit in lesional skin as compared to nonlesional skin [68], and its expression decreased following corticosteroid treatment [67]. In addition, increased protein levels of IL-12/23p40 were detected in the skin of AA-affected mice when compared with unaffected controls [119]. Targeting the IL-12 and IL-23 receptor pathways as a potential treatment for AA, however, has demonstrated conflicting results. In multiple case reports and case series, treatment with ustekinumab, a FDA approved IL-12/23p40 blocker for treating psoriasis, was effective at inducing hair regrowth in patients with AA [136–138]. In contrast, a separate case series of patients treated with ustekinumab failed to respond to treatment [119]. These latter data were in line with similar, aforementioned results found in the C3H AA model, in which treatment with an IL-12/23p40 neutralizing antibody failed to prevent the onset of disease. While these types of clinical reports, including case reports and case series, are often helpful in identifying the potential for efficacy, the high spontaneous resolution rate, especially in patients with patchy-type disease necessitates use of larger, randomized, prospective, well-controlled clinical trials to determine efficacy.

The role of type I interferon, a product of pDCs, has been examined in human tissue and mouse AA models. In human samples, a type I interferon signature has been suggested in AA with studies reporting increased myxovirus-resistance protein 1 (MxA) staining localized to the bulb of inflammatory AA lesions, which were defined as histologically demonstrating peribulbar inflammation, when compared to noninflammatory AA lesions, in which peribulbar inflammation was not appreciated [133, 139]. In vitro treatment of vibrissae microdissected from mice with IFN-α induced upregulation of MHC class I and CXCL10 on the outer root sheath and matrix cells of the follicles [134]. Immunostaining of murine skin for pDCs and IFN-α⁺ pDCs demonstrated an increase in the number of these populations localized around the bulbar region in samples from spontaneous AA mice compared with that from mice that did not develop AA [134]. Somewhat paradoxically, nonlesional skin from spontaneous AA mice showed even higher numbers of these cell types than lesional skin. Furthermore, mice with hair loss induced by transfer of in vitro-expanded T cells from donor AA mice showed increased numbers of skin-infiltrating pDCs in nonlesional samples as compared to lesional samples, potentially suggesting that these cells may play a role in the initiation phase of disease development. In addition, induction of AA was observed when activated pDCs were intradermally injected in three out of three recipient C3H mice. While these studies seem to indicate a role for IFN-α and pDCs in the induction of AA, larger mechanistic studies are likely needed to conclusively dissect the contributions of pDCs in AA pathogenesis.

6.8. NK and NKT cells

Two of the loci linked with AA are associated with MICA and ULBP3, both of which encode ligands of NKG2D that have previously shown to be highly expressed in lesional AA skin [20, 22, 49]. This raises the possibility that NK and NKT cells, both of which may express NKG2D, may participate in the pathogenesis of AA. A classic function of NK cells is to address and clear MHC class I-negative cells. Although low or absent expression of MHC class I has been demonstrated in the inferior region of hair follicles, they do not appear to be normally targeted by NK cells, and NK cells are not thought to commonly localize around these structures [45, 49]. Prior work indicates that healthy hair follicles express increased MIF and diminished levels of MICA, mechanisms that would prevent targeting by NK cells. In AA, however, infiltration of CD56⁺NKG2D⁺ NK cells has been demonstrated in the perifollicular area in lesional AA skin [49], and hair follicles from AA patients exhibit increased MICA expression and downregulated MIF expression [49]. Further, increased frequencies of NK cells, determined by expression of CD16, have been reported in the peripheral blood of patients with AU when compared with normal controls [140]. Following corticosteroid treatment, the frequency of these cells in the circulation were reduced back to levels found in healthy controls. From these studies, however, it is unclear whether the presence of these cells in circulation serves as a biomarker for disease or whether they are contributors to AA pathogenesis.

Experiments in murine AA, however, suggest that NK cells may be playing an immunoprotective role in the hair follicle. Depletion of cutaneous NK cells using an anti-asialo-GM1 treatment during the 4 weeks following AA skin grafting appeared to result in an acceleration of disease, as compared to the mice receiving an isotype control treatment [141]. The mechanism surrounding this is not well understood, but are consistent with reports in other diseases of NK cells having an anti-inflammatory function [142, 143]. Indeed, NK cells have been documented to lyse T cells as well as secrete immunosuppressive cytokines, including IL-10 or TGFβ [144–146], which may inhibit the formation of autoreactive T cells [147]. It is therefore yet unclear if they promote or act to prevent the development of AA.

Using a human-xenograft mouse model of AA, data reported by Ghraieb et al suggested that invariant NKT (iNKT) cells may also have a protective role in AA pathogenesis [148]. Injection of in vitro-stimulated, bead-enriched NKT cells cultured with alpha-Galactosylceramide (α-GalCer), or direct injection of α-GalCer, into donor healthy human scalp skin transplanted onto mice was sufficient to prevent development of PBMC-induced hair loss in the graft. Additionally, depletion or neutralization of NKT cells in PBMC cultures prior to injection in grafts resulted in hair loss. Furthermore, the ability of α-GalCer-treatment on PBMCs to prevent disease onset in skin grafts was lost when the skin grafts were treated with IL-10 blocking antibodies administered daily, indicated that IL-10 is required for NKT cell-mediated protection of hair loss. Interestingly, in vitro proliferation of NKG2D⁺ CD8 T cells was suppressed when cocultured with IL-10 producing iNKT cells, raising the possibility that iNKT cells may be directly suppressing the pathogenic CD8 T cell population in AA. Further studies are needed to identify the role and extent to which of iNKT cells affect autoimmunity to the hair follicle in the native setting and their potential for therapeutic purposes.

6.9. Mast cells

The contributions of mast cells in the pathogenesis of AA remains unknown. Examinations of lesional skin samples from AA patients indicated there are a higher number of mast cells localized around the hair follicle, specifically in the mesenchymal, perivascular, and perifollicular regions, as compared to healthy control skin [46, 126]. As expected, these cells expressed tryptase, a marker of degranulation and activation, suggesting that the increased number of mast cells may be contributing to the inflammatory response around the hair follicle [46, 126]. Additionally, in AA lesions, mast cells are found to be in close proximity to CD8 T cells and strongly express MHC class I, supporting a role in modulating the CD8 T cell response [46, 126]. Indeed, mast cells located around the hair follicle expressed the co-stimulatory molecules OX40L, CD30L, and 4–1BBL, lacked expression of TGFβ1 and failed to produce IL-10, further supporting a role that they may be directly activating surrounding T cells [46]. In the AA skin-induced C3H mouse model, an increased number of mast cells and an increase in mast cells in close proximity with CD8 T cells following disease development was observed. In a human xenograft mouse model, grafted scalp skin induced to lose hair exhibited an increased number of perifollicular mast cells as compared to the uninduced skin grafts. Additionally, an increase in the number of mast cells that were interacting with CD8 T cells in these AA lesions was observed [46]. While in total these data raise the possibility that mast cells may be participating in AA pathogenesis, experimental data from models that directly alter mast cell numbers or function are lacking.

6.10. Autoantibodies and B cell involvement

Early studies examined the role of autoantibodies in AA [149–151]. Two studies found that a greater frequency of AA patients had autoantibodies targeting the outer root sheath and matrix of the hair follicle and were predominantly reacting with different hair follicle antigens of 44, 47, 50, 52, and 57 kDa [149, 150]. To examine a potential mechanistic role of autoantibodies in the development of AA, a xenograft AA model was used wherein skin from AA or control patients was grafted onto nude mice and injected with patient serum [152]. Immunohistological analysis found IgG and C3 deposits around the hair follicles of both AA and healthy control skin that was grafted onto the mice. The introduction of AA patient serum, and the autoantibodies present, had no effect on hair regrowth within grafts of AA skin, as compared with otherwise unmanipulated conditions in this model. In addition, adoptive transfer experiments in C3H mice indicated that splenic B cells from AA mice were not able to induce disease in recipients [28]. In total, the data favor that autoantibodies to the hair follicle appear to be an epiphenomenon and not directly involved in the disease process. The role of B cells as APCs, however, has not been studied.

6.11. Autoantigens and T cell receptor

The identity of a common autoantigen in AA has not been well-defined. The observation that depigmented hair is often spared in clinical disease has led some to hypothesize that melanocyte-derived antigens could be the target of autoimmune attack [153]. T cells isolated from lesional skin of patients with AA have been noted to prevent hair regrowth after stimulation with melanocyte peptides in a mouse xenograft model. Histologically, these grafts showed similar features to that of AA lesions, with dystrophic follicles, infiltration of CD4 and CD8 T cells, and expression of MHC class I and II in the epithelium [153]. In addition, a decreased number of follicular melanocytes have been reported in AA patients, as detected by immunohistochemical staining on scalp samples from AA patients for melanocyte-specific protein melanoma antigen recognized by T cells 1 (MART1) [154]. Alternatively, trichohyalin (THH/TCHH) has been proposed as a potential hair follicle antigen involved in the AA immune response [155]. Immunoprecipitation experiments using serum from AA or control patients found that 100% of AA patients exhibited antibodies to THH, as compared to only 50% control patients. In silico prediction models have also been used to identify candidate hair follicle peptides with high affinity to HLA-A*0201 [156]. Protein epitopes identified to be more highly stabilized onto HLA when compared to their positive control influence peptide included specific TCHH-derived polypeptides, glycoprotein 100 (GP100), and MART1. Pools of mixed TCHH-derived polypeptides were capable of inducing greater CTL activation in PBMCs from AA patients as compared to controls. In addition. using an ELISPOT assay, two additional melanocyte specific antigens, tyrosinase (TYR) and tyrosinase related protein 2 (TYRP2) induced a higher frequency of activation in PBMCs from AA patients. Stimulating PBMCs with a mixed group of TCHH polypeptides showed an increased number of IFN-γ⁺CD8⁺ cells from AA samples, while TYR stimulation showed non-specific cell activation as both AA and control samples showed higher numbers of IFN-γ⁺CD8⁺ cells. Identification of the critical target epitopes would greatly further our understanding of the immune response of AA.

The identification of a clonal T cell population, with a defined T cell receptor, capable of inducing disease would also support the presence of a critical antigen in AA. Sequencing of the T cell receptor repertoire in AA affected mice identified matched sequences of NKG2D⁺ CD8 T cells that were isolated between the SDLNs and lesional skin [93]. Furthermore, paired single cell RNA-sequencing and TCR sequencing identified a population of CD8 T cells with a pathogenic signature and a shared/overlapping TCR repertoire unique to murine AA skin and lymph nodes, indicating that a single population of NKG2D-expressing CD8 T cells travels between these tissues and effects disease [97]. At this time, these data are mostly correlative and indicative of biomarkers of disease rather than causation; further work is therefore needed.

7. Conclusions

Great advancements in our understanding of the immunology of AA have been made in recent years. A variety of immune cells have been described to be present in AA skin and the hair follicle microenvironment, and transfer studies in in vivo models and mechanistic studies have furthered our understanding of AA pathogenesis. Ultimately, these latter types of studies may shed light on the sequence of immunological events that leads to loss of tolerance to hair follicle and its targeting by the immune system, resulting in the emergence of clinical disease in AA. Although newer therapeutic options are being refined, there is a need for a targeted, efficacious treatment with minimal untoward effects, such as immunosuppression or susceptibility to opportunistic pathogens. Identifying how certain cell populations or cytokines are contributing to AA pathogenesis has great potential in pinpointing the next therapeutic target.

Highlights:

• Alopecia areata is characterized by immune infiltration of the hair follicle leading to hair loss.

• Mouse models have aided in our understanding of mechanisms in the immune response of the hair follicle.

• Collapse of immune privilege at the hair follicle is a critical event for AA development.

• There is evidence that many different immune cells may be present around the hair follicle in AA.

• Further studies are needed to dissect the contribution of these various immune cells to the development of AA.

Abbreviations:

AA

alopecia areata

AT

alopecia totalis

AU

alopecia universalis

GWAS

genome wide association study

SDLN

skin draining lymph node

SCID

severe combined immunodeficient

PBMC

peripheral blood mononuclear cells

NKG2D

natural killer group 2 membrane D

TCR

T cell receptor

IP

immune privilege

MHC

major histocompatibility complex

TAP

transporter associated with antigen processing

β2M

β2 microglobulin

TFGβ

transforming growth factor beta

IFN

interferon

α-MSH

alpha-melanocyte stimulating hormone

IGF1

insulin growth factor 1

MIF

macrophage migration inhibitory factor

NK

natural killer

PDL1

programmed death ligand 1

DSCC

dermal sheath cup cells

DP

dermal papilla

FasL

Fas ligand

PBS

phosphate buffered saline

IL

interleukin

mRNA

messenger RNA

JAK

janus kinase

STAT

signal transducer and activator of transcription

FDA

Food and Drug Administration

ULBP

UL16-binding protein

LAK

lymphokine activated killer

IE

intraepithelial

CD

Celiac disease

NKG7

natural killer granule protein 7

CXCR

CXC receptor

CXCL

CXC ligand

DEBR

Dundee experimental bald rat

TNF

tumor necrosis factor

Tbet

T-box protein expressed in T cells

CCL

CC receptor ligand

CCR

CC receptor

nTreg

natural Treg

iTreg

induced Treg

ADTA

acute, diffuse, and total alopecia

SALT

Severity of Alopecia Tool

DC

dendritic cell

pDC

plasmacytoid dendritic cell

RT-PCR

real-time PCR

MxA

myxovirus-resistance protein 1

MICA

major histocompatibility complex class I chain-related gene A

iNKT

invariant NKT

MART1

melanoma antigen recognized by T cells 1

TCHH/THH

trichohyalin

GP100

glycoprotein 100

TYR

tyrosinase

TYRP2

tyrosinase related protein 2