Can the Hair Follicle Become a Model for Studying Selected Aspects of Human Ocular Immune Privilege?

Abstract

Immune privilege (IP) is important in maintaining ocular health. Understanding the mechanism underlying this dynamic state would assist in treating inflammatory eye diseases. Despite substantial progress in defining eye IP mechanisms, becaue of the scarcity of human ocular tissue for research purposes, most of what we know about ocular IP is based on rodent models (of unclear relevance to human eye immunology) and on cultured human eye–derived cells that cannot faithfully mirror the complex cell–tissue interactions that underlie normal human ocular IP in situ. Therefore, accessible, instructive, and clinically relevant human in vitro models are needed for exploring the general principles of why and how IP collapses under clinically relevant experimental conditions and how it can be protected or even restored therapeutically. Among the few human IP sites, the easily accessible and abundantly available hair follicle (HF) may offer one such surrogate model. There are excellent human HF organ culture systems for the study of HF IP in situ that instructively complement in vivo autoimmunity research in the human system. In this article, we delineate that the human eye and HF, despite their obvious differences, share key molecular and cellular mechanisms for maintaining IP. We argue that, therefore, human scalp HFs can provide an unconventional, but highly instructive, accessible, easily manipulated, and clinically relevant preclinical model for selected aspects of ocular IP. This essay is an attempt to encourage professional eye researchers to turn their attention, with appropriate caveats, to this candidate surrogate model for ocular IP in the human system.

It is now obvious that the eye possesses immune privilege (IP) characteristics that are important for eye health. The credit for coining this term belongs to Peter Medawar, ¹ who showed over a half century ago that skin allografts are not rejected by the host's immune system when transplanted heterotopically into defined anatomic sites, such as the rabbit eye or brain. However, even 130 years ago, Van Dooremaal, ² a Dutch ophthalmologist, had already discovered that mouse skin grafts show significantly prolonged survival if transplanted into the anterior chamber (AC) of a dog's eye. These pioneering studies have opened the door for robust research in the field of IP. Consequently, ocular IP has become a subject of major recent interest, and its fundamental importance in inflammatory eye diseases is now widely accepted. ^{3 –15}  

However, as every investigative ophthalmologist painfully experiences sooner or later, as a tissue on which to perform in vitro research, the human eye is an exceptionally rare commodity that is very difficult to come by. In the rare cases in which eyes are enucleated, the damage due to trauma, infection, or tumor growth raises questions as to how useful such tissue is in helping us to understand the physiology of human ocular IP. With very few exceptions (see below), therefore, most currently available data and concepts on ocular IP are based on the systematic analysis of rodent models. ^{16 –26} Given the very substantial immunologic differences between rodent and human systems, ^27,28 it is inherently problematic to extrapolate from rodent to human eyes. Moreover, cultured human eye–derived cells, yet another source of ocular IP research, ^{29 –33} are also problematic, since IP is an in situ state based on complex cell–tissue interactions and not a condition displayed by isolated cell populations in vitro. ^15,34,35  

Therefore, our understanding of human ocular IP remains rather limited, and all extrapolation from rodent and cell culture work must be interpreted with caution. 

Thus, good, clinically relevant human in vitro models are urgently needed to enable the study of the general principles of why and how IP collapses and how it can be protected or even restored in situ. Although the search for such human surrogate models meets with evident obstacles, since no other organ is quite like the eye, at least some aspects of human ocular IP may be studied in other, more accessible and more abundantly available human tissues. 

Searching for a Surrogate Preclinical Model of Ocular IP

The eye is not the only mammalian site of IP. Other sites include parts of the testis and ovary, the adrenal cortex, parts of the brain, the fetomaternal placental unit, the hamster cheek pouch, and probably the proximal nail matrix. ^{34,36 –41} Notably, the human hair follicle (HF) also qualifies as an IP site ³⁴ (for debate, see below). This miniorgan is unique among all other IP sites, in that it is massively distributed over the human body and is highly accessible to experimental analysis and manipulation. Our integument has approximately 5 million HFs and thus a correspondingly vast number of potential (human) IP organs. ⁴² Moreover, there are excellent human HF organ culture systems that can instructively complement in vivo research for the study of autoimmunity in the human system. ^{43 –45}  

Since common mechanisms of IP in different organs have been elucidated, ^{46 –48} an obvious question is whether one IP site can hold lessons for other, less easily explored sites. Although the IP of human HFs has been much less well studied than that of the eye, the relative ease with which human HFs can be obtained, microdissected, organ cultured, and immunologically manipulated in vitro (see below) invites one to exploit human HF organ culture ⁴⁴ as an unusual, but highly instructive and clinically relevant preclinical research model (e.g., for following up in vitro studies of ocular IP in rodent models, before entering into clinical trials on the human eye). 

Following in the footsteps of earlier suggestions along this vein, ^34,49 the current essay attempts to shed some light on both the similarities and important differences between ocular and HF IP and to attract eye researchers and clinical ophthalmologists to systematically employ, with appropriate caution and circumspection, the human HF as an accessible, relatively easily manipulated surrogate model for carefully selected aspects of ocular IP model. 

Immunologic Homeostasis, Ignorance, and Privilege

Before critically exploring the usefulness of human HFs as a candidate surrogate model for ocular IP, it helps to remember that many tissues have their own mechanisms for preserving the immunologic status quo for proper functioning (i.e., immune homeostasis). ⁷ For example, the lung and the gut, which are chronically exposed to foreign antigens that may incite inflammatory responses must allow gas exchange and food processing, respectively, without provoking undesired levels of inflammation. ⁷ Another important concept in this context is immunologic ignorance, which emphasizes the key role of anatomic barriers and peculiarities in a given anatomic site (e.g., the absence of patent lymphatics), which prevents the entry of immune cells resulting in graft rejection. ^8,50  

In contrast, immune privilege classically describes tissue sites within which foreign tissue grafts can survive for extended periods, whereas similar grafts placed in conventional sites are acutely rejected by the host. ^46,50,51 Today, the term IP is generally understood in a much broader sense and indicates the presence of multiple active mechanisms for preventing the induction and expression of both innate and adaptive immune responses. ^{7,8,11,34,36,50}  

Although there are phenomenologic indications that the HF evades some and actively suppresses other potentially autoaggressive immune responses (see below), for evident methodological reasons, rejection and survival of heterologous tissue transplants within this tiny miniorgan have not yet been studied. (In fact, due to insufficiently refined microsurgery and microinjection techniques, previous attempts in our laboratory to inject melanocytes from C57BL/6 mice into the vibrissae hair bulb of white, immunocompetent Balb/c mice have failed miserably.) Thus, one can only lament that Billingham's landmark experiment and its visionary interpretation remain the only currently available functional evidence that HFs can indeed shelter heterotransplants from immune rejection. He observed that, while donor melanocytes within the epidermis are rapidly eliminated, heterologous epidermal melanocytes show long-term survival if they manage to escape into the anagen hair bulbs of (white) host guinea pigs, which thus began to produce black hair shafts. ^34,52,53  

Ocular Immune Privilege: Basic Concepts and Current Perspectives

IP has turned out to be a complex and dynamic tissue state and the list of “players” involved in it is ever-growing (Table 1, Fig. 1). One key mechanism of ocular IP, AC associated immune deviation (ACAID), was identified by Kaplan and Streilein et al. ^{81 –84} ACAID means that injection of antigens into the AC of rodents (and even of monkeys ²⁸ ) produces a stereotypic systemic immune response that is selectively deficient in antigen-specific delayed-type hypersensitivity (DTH), whereas other conventional effector modalities of immunity (such as cytotoxic T cells and non–complement-fixing antibody isotypes) are preserved. ^28,85  

Table 1.

 

Important Characteristics of Ocular and Hair Follicle Immune Privilege

	Eye	Hair Follicle
Specific location of IP	Anterior chamber, iris and ciliary body, subretinal space 7,22,54,55	Hair bulb, hair bulge 34,56,57
Cyclic phenomena	No	Yes
Results of IP collapse	Corneal transplant rejection, immune mediated microbial keratitis and uveitis 7,8,58,59	AA, lichen planopilaris, scleroderma, skin manifestations of SLE, and GVHD 34
ACAID	Yes	No
Strong expression of immunoinhibitory molecules	α-MSH 60 ; TGFβ2 61,62,63 *; TSP1 21,64 *; VIP 65 ; SOM 23 ; MIF 66 ; CGRP 67 ; CRP 32 ; PEDF 19 ; IDO 30 ; IL-1Ra 68 ; GITRL 16 ; TRAIL 25 ; B7–2(CD86) 69 ; cortisol 62	α-MSH, TGFβ1, IL-10, IGF-1 34,43,44 ; MIF, TGFβ2, CD200, IDO 57 ; cortisol 70 ; downregualtion of MICA 71 ; TSP-1? 72 , CGRP? (currently being investigated)
Downregulation of APC molecules	Low MHC class Ia, weak MHC class Ib, no MHC class II cells 24,59,73 –76	Low MHC class Ia, weak MHC class Ib (HLA-E and G), no MHC class II 34,44,56,57,77
Apoptosis induction of lymphocytes	FasL(CD95L) 78 +sFasL(sCD95L) 79 ; B7-H1(CD86) 20,29 ; B7-H3 59	FasL? 80

 AA, alopecia areata; ACAID, anterior chamber-associated immune deviation; APC, antigen presenting cell; GVHD, graft versus host disease; IP, immune privilege; SLE, systemic lupus erythematosus. The abbreviations for proteins are as described in Figure 1.

*  Main inductors of ACAID.

Figure 1.

 

Immune privilege of the eye and anagen VI hair follicle: key features. α-MSH, α-melanocyte stimulating hormone; APC, antigen presenting cell; APM, arrector pili muscle; CGRP, calcitonin gene-related peptide; CRP, complement regulatory proteins; DP, dermal papilla; E, epidermis; FasL, Fas ligand; GITRL, glucocorticoid-induced TNF receptor family-related protein ligand; HLA, human leukocyte antigen; HS, hair shaft; IDO, indolamine 2,3-dioxygenase; IGF, insulin-like growth factor; IL-10, interleukin 10; IL-1Ra, interleukin 1 receptor antagonist; IRS, inner root sheath; MHC, major histocompatibility complex; MICA, MHC class I chain-related A gene; MIF, macrophage migrating inhibitory factor; ORS, outer root sheath; PEDF, pigment epithelium derived factor; RPE, retinal pigment epithelium; sFasL, soluble Fas ligand; SG, sebaceous gland; SOM, somatostatin; TGF, transforming growth factor; TRAIL, TNF-related apoptosis-inducing ligand; TSP-1, thrombospondin 1; and VIP, vasoactive intestinal peptide.

ACAID Has Ocular and Systemic Pathways

It is thought that antigens inoculated into the eye are processed in a distinctive fashion by stromal antigen-presenting cells (APCs) of the iris and ciliary body. ^14,86 This phenomenon appears to be largely under the control of transforming growth factor (TGF)-β, a key immunosuppressive cytokine in the aqueous humor. ^61,62,87 In fact, in mice, a deficit of total TGF-β2 in aqueous humor correlates with loss of ACAID. ⁵¹ These APCs then emigrate from the eye by traversing the trabecular meshwork (TM) and directly enter the blood stream to reach the spleen. ⁸⁸ Access to the blood stream (without an access to the lymphatic system thorough the uveoscleral pathway) seems to be crucial for the induction of ACAID. Indeed, monkeys treated with topical prostaglandins (which redirects a substantial fraction of aqueous humor into the uveoscleral path ⁸⁹ ) failed to induce ACAID. ²⁸  

The eye-derived APCs reach the marginal zone in the spleen. There, a complex dialog involving multiple cells, including B cells, CD4⁺ NKT cells, CD4⁺ T cells, γδ T cells, and CD8⁺ T cells, ensues. ^{58,90 –92} When the latter recognize antigens presented by eye-derived APCs and/or marginal zone B cells, they differentiate into ACAID-inducing regulatory T cells (ACAID-Tregs). ^58,59 CD4⁺ ACAID-Tregs prevent the activation and differentiation of antigen-specific Th1 effector cells, ⁸⁵ whereas CD8⁺ ACAID-Tregs inhibit the local function of effector T cells (Th1 and Th2). ^15,93,94  

Two other elements, are also important in the induction of ACAID: the thymus and the sympathetic nervous system. ⁵⁹ Although the thymus is an important source of the NKT cells that are needed for the splenic phase of ACAID, ⁹⁵ the sympathetic nervous system is believed to have a role in the generation of the NKT-cell population that eventually enters the spleen and participates in the induction of CD8⁺ ACAID Tregs. ^59,96  

Clinical Aspects of ACAID

In mouse models, ACAID has been demonstrated to be involved in various clinical scenarios, such as acceptance of corneal transplants, autoimmune uveitis, acute retinal necrosis (ARN) in a fellow eye that experienced herpes virus infection in the anterior segment, and progression of intraocular malignant melanoma. ³⁹ This mechanism may be the eye's way of protecting its vital functions from immunopathogenic injury. ⁹⁷ ACAID has been widely studied in mice, rats, guinea pigs, and rabbits, implying that it is not a phenomenon restricted to laboratory rodents. ^98,99 Significant progress was made when Eichhorn et al. ²⁸ demonstrated that ACAID can also occur in primates. Even so, it should not be taken for granted that these results (although more convenient than those obtained in rodents) can be extrapolated to conclusions about the human eye. Evidence compatible with the existence of an ACAID-like response in humans indeed has been shown in patients with ARN. ¹⁰⁰ Nevertheless, the evidence that these ACAID mechanisms also apply to the human condition remains rather circumstantial, since definitive experimental proof of the existence of ACAID in humans would require (e.g., the ethically problematic transfer of regulatory T cells from one individual to another). ¹⁰⁰ Thus, whether all the chief characteristics of ACAID established in animal models also occur in humans remains unknown. 

IP and the Posterior Segment of the Eye

If one views the eye as an extension of the brain, it is not surprising that there is also a close relationship between ocular IP and the nervous system. Indeed, ocular IP may mainly be the result of neuroimmune interactions. ^8,101 The retina itself represents a highly organized neuronal tissue that produces many immunosuppressive molecules generated by neurons and glial cells. As a cell layer, RPE can suppress the activation of bystander T cells via soluble inhibitory factors such as TGFβ ^21,102,103 and can induce apoptosis of activated T cells. ¹⁰⁴ Moreover, in mice, the (neonatal) RPE ¹⁰⁵ and the neuronal retina ¹⁰⁶ display inherent IP. It is now known that the vitreous and subretinal space are also IP sites ¹⁰⁷ and that antigens placed in these sites can even lead to the induction of an ACAID-like response. ^22,54,55 In this context, it is interesting to note that (in mice) IP disruption occurs when RPE cells are damaged by laser burns. ¹⁸ This finding has raised the troubling question of whether laser-treated eyes are at increased risk of ocular inflammation, up to the collapse of eye IP. 

Current Perspectives

The concept that “excessive“ ocular IP can facilitate life-threatening infections and the actual presence of autoimmune uveitis and corneal allograft rejection underscores the importance of viewing IP as a relative, rather than absolute status, which can sometimes be bypassed. ⁹⁸ Indeed, it has been proposed that the immune system can “decide” that, sometimes, preservation of life supersedes preservation of vision (e.g., if there is a life-threatening ocular infection, such as trachoma, river blindness, or herpes simplex virus keratitis, resulting in ocular IP collapse and inflammatory eye destruction ⁷ ). 

This example illustrates that IP mechanisms must generally be carefully balanced, since we are continuously exposed to potentially deleterious environmental agents, including microorganisms and antigenic insults; and yet, most humans do not go blind. Moreover, corneal transplants are the least-rejected among all organ transplants, ^39,108 despite the use of HLA-unmatched transplants with minimal immunosuppression. ¹⁰⁹ This is probably due, not only to the anatomic characteristics of the corneal tissue, but also to its low antigenicity. ^59,110  

New frontiers in ocular IP research include the role of ocular IP in eye tumor development and progression. Given that IP can be acquired by tumor cells to evade immune surveillance, ¹¹¹ it is interesting to note that modulating IP by the injection of cytotoxic Fas ligand (FasL) in the vitreous cavity can prevent neovascularization in a mouse model of choroidal neovascularization. ¹¹²  

Thus, instructive experimental models in the human system are urgently needed that allow deeper insights into IP biology and pathology and that facilitate experimental manipulations. This is where the HF, one of the defining features of mammalian species, enters our vision. 

Evidence in Support of HF IP

It has been four decades since Billingham ⁵² discovered that the hair bulb provides a special milieu that permits transplanted allogeneic cells (namely, donor melanocytes) to escape limitation by the host immune system while others are attacked. ⁵² Black skin epidermis transplanted onto skin beds of genetically incompatible white guinea pigs quickly lost its pigmentation (a sign that the foreign melanocytes had been rejected), whereas black hair shafts soon thereafter began to pierce the (now white) epidermis. This result indicates that at least some donor melanocytes had survived in the host hair bulbs and had resumed their transfer of melanosomes to HF keratinocytes. ^34,52,53  

Unfortunately, since then, no additional functional evidence of the existence of HF IP has been published, perhaps because available pointers that HFs are a site of IP have long been ignored by the immunologic and transplantation research communities. Recently, however, HF IP has attracted more widespread interest, ^{34,56,80,113 –116} and there is an increasing community of HF immunology authorities that has embraced the concept of that the HF enjoys a relative IP and that a collapse of HF IP is of critical importance in alopecia areata (AA), one of the most frequent human autoimmune diseases, ^{43,80,114 –116} whereas a collapse of the putative IP of the epithelial stem cell area of the HF (bulge) ⁵⁷ may be important in the pathogenesis of cicatricial alopecia. ¹¹⁷  

Compared with the eye, our understanding of the functional state of resident immune cells within the HF epithelium (Langerhans cells, T cells) and in the HF mesenchyme (macrophages, mast cells) is still very limited, and current HF immunology concepts are largely based on immunophenomenologic analyses. However, a few important facts have surfaced that, taken together, strongly support the concept that defined compartments of the HF represent sites of relative IP. Some key arguments can be summarized as follows (for full discussion, see Ref. 34). 

MHC Class I and β2-Microglobulin Downregulation

A classic feature of IP sites is the downregulation of MHC class Ia expression. ^34,44,116 The lower (proximal) epithelial compartments of the HF epithelium, the anagen hair bulb, shows a striking downregulation of both MHC class Ia and associated β2-microglobulin gene and protein expression. ^44,57,77,118 Since MHC class Ia-stabilization by β2-microgolubin is critical for the proper presentation of MHC class I–dependent antigens, ²⁷ any MHC class I molecules that may still be expressed in the anagen hair bulb probably cannot effectively present autoantigens. In humans and mice, the HF's stem cell zone, the bulge, even expresses nonclassic MHC class I molecules (MHC class Ib molecules, such as Qa-2 and HLA-E), which inhibit, for example, NK cell activities. ^57,77,119 To the best of our knowledge, no healthy mammalian tissue has been described thus far that exhibits this phenomenon without enjoying relative IP. 

Local Generation of Potent Immunosuppressants

A key feature of all recognized IP sites is that they express potent, locally generated immunoinhibitory molecules. ^7,120,121 Therefore, it is important to note that the anagen hair bulb and even the bulge prominently express potent immunosuppressants such as TGFβ, α-melanocyte-stimulating hormone (α-MSH), IL-10, and others. ^{34,44,122,123}  

Functionally Impaired Langerhans Cells

Given the importance of APCs in ocular IP (see above), it is intriguing to note that, in striking contrast to the APCs of the distal HF epithelium (i.e., the upper outer root sheath), the very few intraepithelial Langerhans cells that are detectable ultrastructurally or by CD1a immunohistochemistry in the proximal anagen hair bulb of human scalp HFs do not express detectable MHC class II or I molecules. ^57,123 Even the distal outer root sheath of human HFs harbors immature Langerhans cell populations. ¹²⁴ That at least all professional APCs in the proximal human HF epithelium lack evidence of full antigen-presentation capacity perfectly fits the characteristic impaired antigen presentation by APCs in IP sites. 

Autoreactive CD8⁺ T Cells Are Key Protagonists in HF Autoimmunity

Since AA is a T-cell-mediated, organ-specific autoimmune disease, ¹²⁵ it offers an excellent model for probing the functional relationship between autoreactive T cells and the putative HF IP. AA also promises pointers to the functional relevance of HF IP. 

T lymphocytes isolated from human scalp lesions and expanded in vitro with homogenates of HFs reproduce AA lesions when transferred into scalp explants in SCID mice. ¹²⁶ That CD8⁺ T cells, but not CD4⁺ T cells alone, can produce AA lesions ^43,127 indicates that a prior collapse of HF IP must have occurred, which exposes previously sequestered, MHC class I–presented autoantigens to CD8⁺ T cells. The same is also seen in the best-characterized mouse model of AA ¹²⁸ (see also Ref. 116). This, in turn, suggests that the striking downregulation of MHC class Ia and β2-microglobulin in healthy anagen hair bulbs is functionally important. 

As in most autoimmune diseases, identification of the epitopes that trigger the autoimmune response remains a major goal. In AA, much current interest centers on melanocyte-associated antigens: Melanocyte peptide epitopes (such as Gp100-derived G9–209 and G9–280 and MART-1 (27–35)) injected into autologous lesional human scalp grafts on SCID mice induce AA lesions. ¹²⁹ Moreover, skin-derived CD8⁺ T cells obtained from AA patients co-cultured with MAGE3 show a significant increase in intracellular interferon (IFN)-γ expression compared with the control. ¹³⁰ IFNγ, in turn, is the most potent stimulator of ectopic MHC class Ia expression identified so far. ^34,44  

Since melanocyte-associated, MHC class I–presented autoantigens recognized by CD8⁺ T cells ²⁷ are key immune targets in vitiligo and melanoma, these findings in AA patients and animal models provide functional evidence that the ectopic expression and presentation of MHC class I–presented autoantigens within the HF entails the danger of autoaggressive immune responses against the HF, if cognate, autoreactive CD8⁺ T cells are present. This finding strongly supports the concept that the prominent downregulation of MHC class Ia and β2-microglobulin in healthy anagen HFs is critical to warding off deleterious immune attacks on the HF. ^34,131  

NK Cell Activities Appear to Be Suppressed in Healthy HFs

Since NK cells are primed to recognize and eliminate cells with absent or low MHC class I expression, ^{27,71,132 –138} immunologically privileged HF compartments constitute a basic problem in self-/non–self-discrimination and self-tolerance. ¹³⁹ One would expect that MHC class I-negative or MHC class I “low” anagen HFs are under constant attack by NK cells. However, this is clearly not the case, since very few perifollicular NK cells can ever be found around healthy human anagen HFs. ¹²³ In contrast, in AA, CD56⁺ NK cells prominently aggregate around HFs and show an increased expression of NKG2D (NK cell-activating receptor) and decreased expression of KIR-2D2/2D3 (NK cell-inhibitory receptor). ⁷¹ Moreover, macrophage migrating inhibitory factor (MIF) may suppress NK cells in and around healthy HFs ⁷¹ (just like in the eye ⁶⁶ ). In contrast, excessive NK cell stimulation by NKG2D-activating ligands such as MHC class I chain–related A gene (MICA) by the HF epithelium ⁷¹ and/or other MICA-related NKG2D ligands ¹¹⁴ in and around AA HFs may contribute to HF IP collapse. Therefore, the failure to adequately suppress undesired NK cells activities directed against MHC class I–negative HF cells during anagen may be an important additional element in AA pathogenesis. ⁵⁶ A recent genome-wide association study ¹¹⁴ identified several genetic susceptibility loci for AA. Among these, significant associations include the ULBP genes, which encode activating ligands for NKG2D. Normally, ULBP3 is not present in HFs, but ULBP3 proteins were abundant in and around human HFs affected by AA. NKG2D/ULBP3 engagement, and thus inappropriate NK cell stimulation, may contribute to the development of AA. 

Taken together, this combination of immunophenomenologic in situ observations and functional data from the relevant model disease (AA) constitutes sufficient evidence to suggest that HFs do indeed enjoy relative IP. 

A unique, key feature of IP in the anagen hair bulb is that it is temporary: The HF epithelium rhythmically generates, maintains, and deconstructs an area of relative IP in the region of the HF, which is present only during a defined segment of the hair cycle—that is, the anagen phase (growth stage)—but absent during HF regression (catagen) and the “resting” phase (telogen). ^{34,44,122,123} In contrast, the bulge, the HF's seat of epithelial and melanocyte stem cells, seems to continuously enjoy a relative IP. ^57,140 In this anatomic HF landmark, MHC class Ia, β2-microglobulin, and MHC class II molecules are downregulated, whereas MHCIb (HLA-E) is upregulated. In addition, immunoinhibitory molecules, like the “no danger” signal CD200, ^117,141 α-MSH, MIF, and indoleamine-2,3-dioxygenase (IDO) are markedly overexpressed in the bulge region. ^57,142 All these are features of ocular IP as well. ^{30,60,66,73,143}  

Thus, within the microcosmos of HF immunology, the relative IP of the bulge may actually be more closely related to ocular IP than that of the anagen hair bulb. Bulge IP collapse and the subsequent immunologically mediated destruction of epithelial HF stem cells may play a key role in the pathogenesis of irreversible, scarring alopecia, ¹¹⁷ just as ocular IP collapse can irreversibly destroy the eye (instead, bulb IP collapse in AA typically only induces reversible HF damage ^34,43 ). 

Why Does the HF Need Immune Privilege?

As vision is one of the most important qualities and survival requirements of most living creatures, it is easy to understand conceptually why the eye must be protected against autoaggressive immune attacks, especially in its delicate, apparently nonregenerating compartments (i.e., corneal endothelium and retinal cells). But what about the HF? Which selection advantage might mammals have had during evolution for establishing an area of IP in HF? 

Although this may have been different for humanoids and prehistoric early humans, in our current climates and cultures, hair is clearly dispensable for human survival and propagation of the species. However, during >99% of the total duration of mammalian evolution, an environmentally perfectly adapted hair coat seems to have been vital for numerous of our mammalian ancestors and their reproduction (suffice it here to envision the poor survival and reproduction chances of a hairless polar bear, Arctic fox, or seal). Given that the HF is one of the most frequent targets of immune-mediated tissue injury, ^34,144 the HF may therefore have established its IP as a safeguarding system against immune injury of this important miniorgan. Viewed from this angle, ocular and HF IP may both be necessary (even though this is much less evident for the latter than for the former). 

Similarities and Differences between Ocular and HF IP

Thus, ocular and HF IP show some striking similarities—namely, in the bulge region of the HF:

•  
  Classic MHC class I and β2-microglobulin are downregulated, which renders cells relatively invisible to CD8⁺ cytotoxic T cells. ^{34,44,57,66,74,145}
•  
  APCs are both sparse and functionally impaired (e.g., they lack MHC class II expression). ^{34,44,57,59,74}
•  
  IFN-γ exposure causes ectopic upregulation of MHC class Ia and II expression in cells lining the AC and in the HF. ^{44,59,146,147}
•  
  Nonclassic MHC class I molecules (e.g., HLA-E and -G) are downrebulated. Together with MIF, they are known to inhibit a potential attack of NK cells on MHC class Ia (i.e., HLA-A, -B, and -C)–negative cells. ^57,73,75,119
•  
  There is a strong expression of similar immunoinhibitory molecules (see Table 1 and Fig. 1 for details).

In these specific areas, the human HF may well serve as an attractive surrogate model for ocular IP (see below).

Of course, there are also many important differences between follicular and ocular IP that one needs to keep in mind when studying the HF as a surrogate model. Mainly, an ACAID-like response has not been demonstrated (yet) to be associated with antigens introduced into the HF. Also, Fas–FasL interactions, which seem to be an important element of ocular IP, ⁷⁸ are unlikely to play a major role in HF IP. ^34,49,123 Interestingly, however, FasL is indeed significantly decreased in lesional skin of AA patients, compared with nonlesional skin. ⁸⁰  

Other apparent dissimilarities between ocular and HF IP may also be less pronounced than one may be inclined to think. Since neuroimmune interactions play a key role in ocular IP (see above), it is reasonable to ask whether these have any role in HF IP. Although this question has so far only been addressed very incompletely, the HF does represent a prototypic neuroectodermal–mesodermal tissue interaction system and is one of the most densely and intricately innervated of all peripheral tissues. ¹⁴⁸ In fact, HF biology has multiple neurobiological dimensions along what has been termed the “brain–HF axis” ¹⁴⁹ (for a review, see Ref. 150). Namely, HF development, growth and pigmentation are regulated in part by neurotrophins and neuropeptides, whereas HF-derived neurotrophins control HF innervation. In addition, human scalp HFs even display a fully functional, peripheral equivalent of the central hypothalamic–pituitary–adrenal (HPA) stress response axis. ^44,151 It would be unreasonable to expect that this brain–HF axis plays no role in HF IP. 

Indeed, psychoemotional stress causes prominent perifollicular neurogenic inflammation and hair growth inhibition in mice, which depends on mast cells, substance P, and nerve growth factor (NGF) and goes along with phenomenologic indications of HF IP collapse. ^{152 –156} Mice subjected to noise stress show dense inflammatory cell infiltrates around their HFs, as early as 24 hours after exposure, with an increase in the number and activation of perifollicular mast cells as well as MHC class II–positive inflammatory cell clusters. Noise stress also increases the number of intradermal dendritic cells and induces their maturation. ^{152,155,157,158}  

Of special interest here is substance P, which is released in response to stress by sensory skin nerves, plays a key role in the cutaneous neuroimmune network and influences immune cell functions through the neurokinin-1 receptor (NK-1R). ^153,159 In murine AA, NK-1R is prominently expressed on CD8⁺ lymphocytes and macrophages that accumulate around lesional HFs. Thus, currently available murine data suggest that substance P and NK-1R are important elements in the pathogenesis of in autoimmune hair loss and the associated collapse of HF IP. ¹⁵⁹  

Moreover, in organ-cultured human scalp HFs, substance P directly induces HF IP collapse, as evidenced by ectopic MHC class I and class II expression. ⁴⁵ Although the role of substance P in ocular IP has not yet been systematically explored, the immunoinhibitory neuropeptide calcitonin gene-related peptide (CGRP), which is co-expressed with substance P in sensory skin nerves, is involved in the maintenance of ocular IP. ⁶⁷ Preliminary evidence from our laboratory suggests that CGRP may exert a similar protective function in the context of human HF IP (Kinori et al., manuscript in preparation). Thus, it is quite likely that neuroimmune interactions are an important component, not only of ocular, but also of HF IP. 

How May Insights from Human HF IP Generate Therapeutic Benefits for Clinical Ophthalmology?

Although it would be unreasonable to claim that most aspects of ocular IP can be investigated in human HFs in vitro, we propose here that selected IP-related insights from human HF organ culture could be put to excellent clinical use in ophthalmology. 

Figure 2 illustrates how relatively easy it is to microdissect and organ culture human scalp HFs obtained, for example, from excess scalp skin during routine face-lift surgery. In these, IP collapse can easily be induced by the proinflammatory cytokine IFNγ. This effect induces rapid, massive, ectopic MHC class Ia and II expression in the epithelium of normal anagen (stage VI) HFs ¹⁴⁷ (see Fig. 2), thus seriously endangering maintenance of the HF IP (see above). The same phenomenon can also be induced by adding substance P to the HF culture medium, ⁴⁵ or in vivo by injecting IFNγ into the back skin of mice with HFs in the anagen stage of the hair cycle. ¹⁴⁷  

Figure 2.

 

Hair follicle isolation and culture. (A) Human temporal and occipital uninflamed scalp skin was taken from donors with informed consent during routine face-lift surgery, in compliance with the guidelines in the Declaration of Helsinki. (B) After the skin is shaved, it is cut into thin strips, approximately 5 × 10 mm. (C) Side view of a cut skin strip where the vision of the hair follicle is complete and vertically orientated. (D) The scalpel blade divides the epidermal–dermal part (above) from the subcutaneous (SC) layer of the skin (below). (E) If the cut is successful, a net of white dermal collagen fibers appears, spread all over the SC fat tissue, as shown here. The lower part of the hair follicle with its hair bulb including the dermal papilla resides in the subcutis and is taken for further processing. (The bulge region and the sebaceous gland of the hair follicle remain in the white dermal part) (F) The sides of the fat tissue are pressed carefully with blunt forceps, to partially extrude the upper portion of the hair follicles from the subcutis. At the same time, the tip of the follicle is gently gripped with watchmaker's forceps, and the hair follicle is pulled from the hypodermal fat. (G) It is essential to isolate intact hair follicle bulbs without any visible damage if the successful maintenance of hair follicles is to be achieved. Hair follicles are free-floating in a 24-well multiwell plate (three follicles per well) filled supplemented Williams E medium. (H) Hair follicles are maintained in 500 μL serum-free Williams E medium (Biochrom, Cambridge, UK) supplemented with 2 mM l-glutamine (Invitrogen, Paisley, UK), 10 ng/mL hydrocortisone (Sigma-Aldrich, Taufkirchen, Germany), 10 μg/mL insulin (Sigma-Aldrich) and 1% antibiotic/antimycotic mixture (100×; Gibco, Karlsruhe Germany). Hair follicles are maintained free-floating in the wells at 37°C in an atmosphere of 5% CO₂ and 95% air. This permits detailed measurements to be made on the length of individual hair follicles during the culturing period ^{160 –162} (I) Immunofluorescent staining of a vehicle treated anagen VI hair follicle shows very low or absent MHC class I immunoreactivity in the CTS and proximal ORS. (J) Treatment with 75 IU/mL of IFNγ induces the ectopic MHC class I expression in the DP, the CTS and the proximal ORS. (K) MHC class II expression in the anagen stage VI hair bulb is very low or absent in the CTS and the proximal ORS. (L) Culturing with 75 IU/mL IFNγ prominently induces MHC class II expression in the DP, the CTS, and the proximal ORS keratinocytes. CTS, connective tissue sheath; DP, dermal papilla; IFNγ, interferon γ; MHC, major histocompatibility complex; ORS, outer root sheath.

Thus, human HF organ culture permits one to screen for candidate agents that effectively downregulate IFNγ-induced ectopic MHC class I expression in human anagen HFs. A “protection” or (perhaps more important) “restoration” assay design can be chosen: In the former, the candidate IP protectant is added to the medium before IFNγ is introduced, whereas candidate “IP restoration” agents can be tested by adding them after IFNγ administration. In fact, three immunomodulators known to be locally produced in the anagen hair bulb—α-MSH, TGFβ1, and insulin-like growth factor 1 (IGF-1) ^122,163,164 —are all capable of downregulating ectopic MHC class Ia expression, on both the protein and the mRNA level, in the IP restoration assay design. ⁴⁴ This human organ culture assay, therefore, is well-suited as a clinically relevant preclinical screening system to identify novel candidate IP-restoring or IP-protecting agents. Once identified, these can then further be explored as candidate therapeutics for ocular IP protection and restoration. 

Which Specific Aspects Relevant to Ocular IP Can Be Studied in Human HF Organ Culture?

The human HF is hardly suitable as a surrogate model for studying ACAID or for evaluating the effects of test agents on the delicate retinal neuronal tissue. However, given that the HF may be more dispensable than any other human organ (privileged or not), it offers investigators interested in IP (in the eye and elsewhere) an unparalleled opportunity to directly study and manipulate a complex but easily accessible and widely available human IP site. In this site, the following specific questions that are directly relevant to ocular IP may be studied in situ:

1.  
   How is MHC class Ia, Ib, and II and β2-microglobulin expression regulated in situ in a normal human neuroectodermal–mesodermal interaction unit?
2.  
   How can (experimentally induced) ectopic upregulation of these molecules be effectively downregulated again?
3.  
   Vice versa, how can the local expression of IP-protective, immunoinhibitory molecules that are also of relevance in ocular IP (e.g., IDO; immunoinhibitory neuropeptides, such as α-MSH, vasoactive intestinal peptide [VIP], and CGRP, and TGFβ1, IGF-1, and CD200) be effectively upregulated in a normal human neuroectodermal–mesodermal interaction unit?
4.  
   Psychoemotional stress may be implicated in the relapse of anterior autoimmune uveitis. ¹⁶⁵ Shouldn't it then be possible to exploit human HF organ culture to further explore in this human miniorgan the direct impact of well-defined stress mediators (including neuropeptides like VIP that have not yet been studied in a hair research context, but are important in ocular IP) on key IP characteristics, such as MHC Ia/β2-microglobulin expression and the local generation of immunoinhibitory compounds?
5.  
   What is the relative contribution of human NK cells, NKT cells, intraepithelial T cells, Langerhans cells, and mast cells to IP (e.g., via local tissue interactions with the epithelium)? For example, a recent new concept in murine AA pathobiology suggests that some NK cell subpopulations, as opposed to IFN-γ-secreting CD49b⁺ T-cell subsets, may actually award relative protection from AA development. ¹²⁵
6.  
   How do drugs or operative techniques that are already used in the management of inflammatory eye diseases affect human HF IP? This may help to predict desired and undesired effects on ocular IP, whose study in human eyes would require enucleation and is thus essentially impossible. For example, in mice, retinal laser burn abrogates IP in both the burned and nonburned eye. ¹⁸ Applying laser burns to scalp HFs in vitro may indicate whether laser treatment is likely to exert similar effects on human IP. Also, at least some underlying mechanisms of action could be studied in HF organ culture, but hardly in human eyes.
7.  
   Along the same vein, can HFs from a given patient with autoimmune ocular disease be used to predict the likely response of that patient to the intended therapy, as a potential means to predict whether that therapeutic intervention is likely to restore or protect ocular IP?
8.  
   IP in the proximal HF epithelium is a cyclic phenomenon that appears to be restricted to a defined segment of the hair cycle. Although the human eye is not thought to undergo cyclic transformations in adult life, one wonders whether ocular IP also possesses some cyclic characteristics—for example, being maximal or minimal during certain periods on a circadian or perennial time scale. Could it be that autoimmune uveitis preferentially relapses during such hypothetical periods of constitutively “minimal ocular IP”?

Adult Epithelial Stem Cells and Immune Privilege: Yet Another Eye–Hair Connection?

Stem cells rank among the most exciting current research frontiers in experimental and clinical ophthalmology. Although beyond the scope of this essay, it therefore should at least be mentioned briefly that the study of adult human epithelial stem cells (eSCs) in the bulge region of the HF ^{166 –170} may also benefit ophthalmology research. After all, at least in rodents, HF-derived eSCs can differentiate into cells with a corneal epithelial phenotype when given appropriate stimuli. ^{171 –173}  

On this background, it is interesting to note that physiological concentrations of thyroid hormones enhance expression of CD200 on human HF eSCs. ¹⁷⁴ This important immunoinhibitory and tolerogenic surface molecule ¹⁷⁵ is a crucial element of bulge IP maintenance, since its targeted knockout in mice causes massive inflammation and irreversible HF destruction. ^57,117,141  

Therefore, one might learn from HF-associated eSCs how cell-based therapies for the treatment of ocular disease could be engineered so as to reduce the risk of immune rejection or undesired immune deviation by progenitor cells introduced into the human eye, for example, by promoting their expression of CD200. This could, for example, become important in the treatment of limbal stem cell deficiency (LSCD) ¹⁷³ and age-related macular degeneration (AMD), ¹⁷⁶ especially if one explores the use of autologous, easily accessible human HF eSCs as a potential source for such cell therapy. Do HF-derived human eSCs (or any other epithelial progenitor cell type exploited for cell-based regeneration strategies in experimental ophthalmology) retain the relatively immune-privileged status that they had enjoyed in situ, ¹⁴² once they are isolated, propagated, and treated in cell culture? 

Conclusions and Perspectives

It is now widely accepted that understanding ocular IP will contribute to the development of new therapeutic approaches to tissue transplantation and autoimmune diseases, not only of the eye, but also of other organs. ^4,7,12,59 The same may be claimed for HF IP, and this in a clinically relevant and preclinically much more accessible and available model system. Despite the many evident differences between the eye and the HF, the limited, but persuasive similarities between ocular and follicular IP delineated above raise the possibility that their respective responses to test agents, and the selected aspects of IP listed above also are quite similar—probably at least on the same level of similarity as that of rodent versus human ocular IP. 

Collaboration between ocular and skin scientists in the joint exploration of IP therefore promises to be very fruitful to both communities. Let us remember: Van Dooremaal and Medawar, ^1,2 the pioneers of IP research, made their seminal discoveries on ocular IP by placing skin allografts into the AC, thus paving the way for IP research that combines a cutaneous and an ocular perspective. It is in this tradition that we advocate the study and manipulation of a cutaneous miniorgan, the HF, as an unconventional, but highly instructive, accessible, and clinically relevant preclinical surrogate model for defined aspects of ocular IP. No doubt, the surrogate model we are proposing here has major limitations and cannot fully satisfy a devoted ocular IP researcher; but it is the best preclinical surrogate model that we have so far (and may ever have) in the human system. 

Footnotes

 Supported in part by a “Cluster of Excellence” grant from Deutsche Forschungsgemeinschaft (DFG) (“Inflammation at Interfaces”) and by a DFG Graduate College grant (“Autoimmunity”) (RP).

Footnotes

 Disclosure: M. Kinori, None; J.E. Kloepper, None; R. Paus, None

MK thanks Joseph Moisseiev for his continued encouragement, professional advice, and support.