As social beings humans communicate significantly via their physical appearance, with skin and hair color contributing disproportionately. It is well-accepted that hair growth (a mammal-specific trait) and pigmentation have facilitated evolutionary success in non-human mammals (i.e., via thermal insulation, camouflage, social and sexual communication, and sensory perception) (Wu et al., 2008). However, these traits do not appear to have been critical for survival in humans; indeed much has been written on the evolution of our so-called hairlessness (Schwartz and Rosenblum, 1981; Jablonski, 2010).
Why humans should be unique among primates in exhibiting pigmented scalp hair that is very thick, long, and pigmented is much less clear. On one level our ‘naked ape’ status draws much attention to our facial and scalp hair, and modifications of these may have been critical for the development of optimal communication strategies. The luxuriant nature of human scalp hair growth (at least in youth) is likely to reflect particular evolutionary selective pressures present during the early stages of human evolution.
One possible explanation may be found in our littoral evolution strategies, where a fish-rich diet risked uptake of heavy metals. Strategies to prevent build-up of toxic metals in rapidly growing tissue (like hair follicles) may have been selected for e.g., selective binding of metals to melanin (Liu et al., 2005). In this way, the entrapment of toxins within long, deeply melanized, scalp hair fibers would limit access to the living tissue of the highly vascularized scalp. A recent metal analysis of hair melanin reported significant amounts of various heavy metal ions bound to both brown/black eumelanosomes and red/yellow pheomelanosomes (Liu et al., 2005). For example, the amount of Cu(II) and Zn(II) is similar in both melanocyte types, but Fe(III) content is four times higher in the red-hair pheomelanosomes.
Brown-black (or eumelanic) pigmentation is the default or wildtype for human scalp hair coloration. This environmentally friendly pigment type predominates in >90% of the world’s human population. Why eumelanic skin and hair, which trap radiant heat and provide for significant thermal insulation, should be the skin/hair coloration norm for human evolution in tropical climes (e.g., African) may puzzle the casual observer.
On closer analysis, however, this coloration provides significant protection against sunstroke as eumelanin provides a very efficient and fast exchange of ion transport and efflux to provide adequate salt balance (Wood et al., 1999). The remaining <10% of the world’s population (mostly those with a northern Europe origin) has been bestowed with cutaneous pigmentation genotypes/phenotypes that provide for an eye-catching array of associated hair colors. These range from white blonde, yellow blonde, auburn to red and all shades in between. However, some of the associated skin pigmentation type may afford them poorer protection for life on a UVR-drenched planet (Ha and Rees, 2002).
It may appear a little odd that so much of pigmentation research is focused on the skin color type of a minor fraction of the total world’s population. While this may reflect some residual and misplaced euro-centrism, the finding that this phenotype results from loss of (or reduced) function mutations in the melanocortin-1 receptor (MC1R) gene has yielded hugely valuable information on the regulation of pigmentation and its relevance to skin pathology and melanoma susceptibility (see Rees, 2000).
The power and elegance of the genetics of inbred laboratory strains of Mus Musculus has provided a very appealing vehicle to drive forward our understanding of human pigmentation. Thus, much ‘skin’ pigmentation knowledge derives from the intensive study of murine ‘coat’ color (Nakamura et al., 2002; Steingrímsson et al., 2006). This rodent species exhibits a high level of coat color mutation, and many correlates continue to be discovered in humans.
I would like to suggest we consider some caveats however with an over-reliance on mouse data. Not least of these is the important issue of whether mouse follicular melanocytes are truly the best model for examining human epidermal melanocytes, given the former’s deep and UVR-protected distribution in skin. Moreover, while functional epidermal-melanin and follicular-melanin units co-exist anatomically in human skin (Tobin and Bystryn, 1996); this is strikingly not so for mice, where pelage skin epidermis lacks functional melanocytes (Figure 1).
There continues to be a steady stream of data suggesting that human and mouse pigment cell biology differs in several other important ways (Figure 1, Table 1). Investigators moving from murine to human hair growth studies are commonly struck by the level of autonomy vested in individual hair follicles in the human scalp, reflective in part on the striking mosaic nature of adult human scalp hair growth, which is in marked contradistinction to the highly synchronized nature of murine pelage hair growth. With the latter comes marked hair-cycle-regulated changes in skin anatomy (e.g., dermal/subcutis depth) as well as cycle-associated change in skin vascularisation and innervation (Figures 1 and 2a) (Mecklenburg et al., 2000; Botchkarev et al., 1997).
Moreover, it is also striking that two of the three main melanogenic enzymes may have different regulation/functions in mouse compared with human scalp. For example, there are considerable species-specific differences in the expression and activities of both dopachrome tautomerase (DCT) and tyrosinase-related protein 1 (TRP1) (Commo et al., 2004a,b; Boissy et al., 1998). Moreover, the histologists among us will be impressed by the dramatic ‘ladder-effect’ of murine hair shaft pigmentation, reflecting a preferential transfer of melanin granules into the medulla of murine hair fibers. This contrasts with the almost exclusive transfer of melanin into the cortex of human hair fibers, even in highly medullated beard human hair shafts.
A comparison of hair pigmentation in murine pelage hair versus human scalp reveals a huge difference in the duration of active pigment production in their respective follicular units. Whereas melanogenesis is active for <2 weeks in the mouse pelage, continuous melanogenesis in human scalp hair follicles routinely lasts for up to 100 times longer – and in some unusual cases more than 700 times longer (Thibaut et al., 2010).
Clearly, this extended melanogenesis in human scalp hair follicles is more than could be envisaged by simple comparison of mice and human relative longevities. Despite these important species-specific concerns, we remain hugely indebted to the mouse for several crucial breakthroughs in our understanding of how pigmentation is regulated, and much of the data referred to in this review have been derived at least in part also from mouse-based studies.
Origin and development of the human hair follicle pigmentary unit
The hair follicle and epidermal pigmentary units occupy distinct, although open, cutaneous compartments in adult skin, and are derived from precursor neural crest cells that commit to the melanocyte lineage at the dorsal neural tube of the embryo (reviewed in Thomas and Erickson, 2008) and additionally from Schwann cell precursors through the ventral pathway (Adameyko et al., 2009; Adameyko and Lallemend, 2010).
These cells then migrate under the epidermis to take up residence in the skin throughout the body. Of the more than 100 loci shown to affect hair color in the mouse (Nakamura et al., 2002), mutations in KIT, a tyrosine kinase receptor, and its cognate ligand SCF, together with mutations in endothelin 3 and its receptor Ednrb, have been most informative in the sense that mutant homozygotes for these pathways exhibit an almost complete lack of hair pigmentation.
It is not clear whether the population of human terminal follicle and the mouse pelage follicle with melanocytes follows similar pathways. Histologically, 5–6 different pigment cell subpopulations can be detected in the human hair follicle (Tobin, 2008) (Figure 2). Three follicular compartments contain melanogenically active melanocytes i.e. the infundibulum, sebaceous gland, and hair bulb around and above the dermal follicular papilla.
In addition, undifferentiated amelanotic pigment cells can be found in the upper hair follicle reservoir close to the insertion site of the arrector pili muscle, the outer root sheath along the full length of the hair follicle and another subpopulation in the most peripheral and proximal hair bulb matrix. By comparison, in the murine pelage follicle, the pigment cells are only routinely and readily observable in/around the bulge region (undifferentiated) and hair bulb (differentiated) (Nishimura et al., 1999, 2002).
Pigment cell recognition usually requires detection of differentiation genes/peptides, several of which are associated with melanosomes. Thus, some pigment cells lacking these organelles may escape routine detection. Further discussion of the origin and development of cutaneous melanocytes is beyond the scope of this perspective; interested readers are invited to refer to several excellent recent reviews (Thomas and Erickson, 2008; Ernfors, 2010).
Comparative biology of human epidermal and hair follicle melanocytes
A hallmark of the comparative biology of epidermal and hair follicle melanocytes is the observation that melanogenesis in the latter is stringently coupled to the hair growth cycle while melanogenesis in the former is continuous, although augmentable. Many other differences can be noted at the cell biologic level. Clinical observation also provides ample evidence for the relative independence of the epidermal- and follicular-melanin units.
For example, one can easily appreciate the striking co-expression of white hair on black skin in aging Africans and raven black-hair in white-skinned youthful Europeans. This is further supported clinically, by the selective/preferential targeting of epidermal but not follicular melanocytes in most cases of vitiligo, while follicular melanocytes alone are damaged by immune-mediated pathology in acute alopecia areata (Tobin and Bystryn, 2000, 1996). These findings suggest that follicular and epidermal melanocytes exhibit different antigenic profiles in situ.
Some of this immunologic difference is likely to reflect the fact that the follicular-melanin unit resides in the immune-privileged proximal anagen hair bulb (Westgate et al., 1991; Christoph et al., 2000; Ito et al., 2008), in contrast to the immunocompetent nature of the epidermal-melanin unit. These differences can also be retained in long-term cultures of these cells (Tobin and Bystryn, 1996). My laboratory is interested in elucidating further ways in which these two cutaneous populations differ, particularly in terms of their responses to a range of biological response modifiers. For example, some drug treatments appear to reduce melanogenesis in the epidermal-melanin unit, while potentially increasing this in the follicular-melanin unit (Campbell et al., 2009).
Moreover, both melanocyte compartments may respond variably to stimuli via the MC1R from proopiomelanocortin-derived peptides (e.g., alpha-melanocyte stimulating hormone, ?-MSH), which could account for the co-expression of skin phototype I epidermal pigmentation with black eumelanic hair. To my knowledge, there has been no evidence to suggest that there may be unique polymorphisms of the MC1R gene sequence (or in downstream signaling) between epidermal and follicular melanocytes, although this should be explicitly examined.
Much information on the variable life-histories of cutaneous melanocyte subpopulations can be gained by close examination of their respective cell biologies, which are largely determined by their respective spatio-temporal status in the epidermis and in the cycling hair follicle. Despite their common and shared origin in the early epidermis, the melanogenically active melanocytes in the anagen hair bulb are larger and with longer dendrites than melanogenically active pigment cells in the epidermis.
They also have more extensive Golgi apparatus and rough endoplasmic reticulum, and produce larger melanosomes than epidermal melanocytes (cf. Fitzpatrick et al., 1958; Fitzpatrick and Breathnach (1963). The fate of the melanin product of these cells is also variable. While eumelanin produced by the epidermal melanocytes degrades almost completely in the differentiating layers of the epidermis, eumelanin granules transferred into hair cortical keratinocytes remain minimally digested. Hence, in the latter, the proximal and distal ends of a typical scalp hair shaft are similarly pigmented.
This difference may relate to the significantly larger melanosome size in follicular melanocytes (approximately 0.35 × 1.0 ?m2 for Caucasian eumelanosomes) than epidermal melanocytes (approximately 0.25 × 0.6 ?m2 for Caucasian eumelanosomes). This parameter will influence the nature of their uptake by recipient keratinocytes (Thong et al., 2003) and their relative susceptibilities to enzymatic degradation. This may be analogous to the enhanced survival of larger melanosomes in black skin compared with their white counterparts (Chen et al., 2006).
The above findings indicate that the functionality of melanocytes is influenced significantly by their location in skin and within its appendages, with further influences at the micro-environmental level. Melanotic melanocytes positive for 3,4-dihydroxy phenylalanine (dopa) oxidase activity are readily detectable in the basal layer of the infundibulum and around the upper dermal papilla, while moderately differentiated melanocytes may also be detected in the basal layer of the sebaceous gland (Ito et al., 1974).
Melanin production in the sebaceous gland is a fascinating observation. Its location at this potentially vulnerable site at the interface with the external environment could serve as an important role in innate immunity and defense. Melanin and especially melanogenic intermediates may serve as potential antimicrobial compounds (Mackintosh, 2001). The anagen hair bulb is the only site of pigment production for the hair shaft (Figure 2). Melanogenically active melanocytes are restricted therein to the upper matrix, just below the precortical keratinocytes, a location that facilitates the transfer of melanin to the hair shaft cortex, much less so to the medulla, and very rarely the hair cuticle (Figures 1, 2).
Undifferentiated or poorly differentiated melanocytes (in terms of dendricity, melanosome production and melanisation and dopa-oxidase activity i.e. lack enzyme activity for tyrosinase as assessed histochemically) can be detected in or close to the presumptive stem cell reservoir throughout the outer root sheath as well as in the most peripheral and proximal anagen hair bulb matrix. The latter subpopulation is rather enigmatic, as they commonly exhibit strikingly different expression profiles for a large range of receptors and proteins, including those of neuroimmunoendo crine interest (Tobin and Kauser, 2005). It is not clear whether this subpopulation is continuous with similarly poorly differentiated dopa-oxidase-negative pigment cells located throughout the outer root sheath.
All dopa-positive cells (i.e., active tyrosinase enzyme) and also some dopa-negative melanocytes of the mid outer root sheath contain (pre)melanosomes (i.e., gp100-positive) (Horikawa et al., 1996). Although amelanotic hair follicle melanocytes lack dopa-oxidase activity, low levels of the tyrosinase protein itself may be detected in some cells (even in senile white-hair follicles), as well as KIT and Bcl-2. These melanocytes do not express the melanogenic enzymes TRP1 and TRP2 (DCT) (Horikawa et al., 1996). The role of these amelanotic melanocytes in hair pigmentation is unclear, although it has been speculated that these cells represent a pool of ‘transient’ melanocytes that migrate from precursor melanocyte stores in the upper outer root sheath (Slominski et al., 1994; Tobin et al., 1999). The presence of immature melanocytes (operationally termed melanoblasts here) in fully developed adult anagen hair follicles has also been confirmed in vitro (Tobin and Bystryn, 1996; Tobin et al., 1995).
Hair follicle melanocytes during the hair growth cycle
After full development, the hair follicle enters the hair cycle with the precipitation of the fully formed and hair fiber–producing hair follicle into a regression phase characterized by massive apoptosis (the first catagen). Upon completion of this catagen phase, the terminal scalp hair follicle is reduced to about 30% of its original full (i.e. Stage 8) tissue mass. It thereafter enters a period of relative rest (i.e. telogen) and remains therein until the third and final episode in the hair follicle life history i.e., its entry into the anagen phase of the first cycle. Thereafter, the hair follicle continues with life-long cyclical activity.
Active melanogenesis within melanocytes and melanin transfer from melanocytes to keratinocytes occurs only during the hair growth phase (anagen) (Figure 2), which in human scalp hair can be very long (up to 8 yr or more) (Stenn and Paus, 2001). In rare cases of so-called anagen-block, anagen can remain continuous for over 25 yr (Thibaut et al., 2010) and remarkably these continuously growing hair shafts are similarly pigmented at their proximal and distal ends. This finding suggests that the melanogenic potential of a relatively small number of bulbar melanocytes is much greater than needed for the typical anagen length in scalp hair (approximately 3 yr).
The extended anagen of human scalp hair, together with its mosaic pattern of hair growth, hinders a systematic analysis of melanocyte dynamics during the human hair cycle. By contrast, the C57BL/6 mouse can be used as a model for human hair pigmentation, given its short anagen (15–17 days), synchronous hair growth, restriction of melanogenically active truncal melanocytes to the hair follicles, exclusively eumelanin production, and the similar linkage of murine melanogenesis with anagen (Slominski et al., 1994).
Tracking melanocyte subpopulation dynamics during the hair cycle has tended to rely on the expression of melanogenesis-related proteins (or their mRNA) including tyrosinase, and related proteins TRP1 and DCT (TRP2), in addition to KIT and the proliferation marker Ki67 (Botchkareva et al., 2001). In this way, the extent of melanocyte differentiation can be followed both temporally during the hair cycle and spatially throughout the different hair follicle anatomic compartments. While evidence of continuous melanogenesis can be detected in even the whitest of sun-protected human epidermis, functional melanocytes are lost from the human hair follicle during the relatively quiescent telogen stage. Similarly, telogen C57BL/6 skin lacks evidence of tyrosinase (mRNA or protein), TRP1 protein or melanin (Slominski et al., 1991). Despite this, the telogen germ contains all cell precursors needed to reconstitute a fully developed (and pigmented) anagen VI hair follicle.
To glean a view of the earliest events in the reconstruction of the follicular pigmentary unit, researchers often induce the telogen to anagen transition by depilation (Paus et al., 1990). Although this may also induce a mild wounding reaction which may obscure some physiological events in follicular melanocyte life-histories, this method allows us to see the first 1 or 2 days of anagen induction (anagen I). Here, some cells begin to express tyrosinase message and thereafter some tyrosinase protein, although the tyrosine hydroxylase and dopa-oxidase activities of tyrosinase remain undetectable enzymatically. Some DCT-positive melanocytes begin to express TRP1 at this stage, especially melanocytes located close to the re-forming hair bulb. By contrast, melanocytes residing in the upper outer root sheath (site of the presumptive germ cell reservoir) remain TRP1 negative. A second subpopulation (expressing either TRP1 or DCT together with KIT) begins to show proliferative activity (Botchkareva et al., 2001). Melanocytes in the S-phase of the cell cycle have been reported as early as anagen II and significant proliferation is clearly apparent in anagen III (Sugiyama, 1979).
The anagen-associated stimulation of undifferentiated telogen melanocytes/melanoblasts predates the melanogenic stimulus delivered during anagen III, which in turn is followed by active melanogenesis and the subsequent transfer of mature melanosomes into keratinocytes of the precortical matrix. In cycling human scalp hair follicles, protein expression for tyrosinase and TRP1 is detectable during the reconstruction of the hair follicle pigmentary unit (during anagen phases III/IV to VI) and then only in those differentiating/differentiated melanocytes that take up position in the hair bulb matrix (Commo and Bernard, 2000). At the induction of a new anagen hair follicle in humans, some melanocytes become committed to cell division, when located in the nascent bulb close to the follicular dermal papilla. Melanogenesis linked to anagen is associated with expression of both tyrosinase activity and TRP1, although the functions of the latter protein in human melanocytes are not yet clear (Boissy et al., 1998).
Bulbar melanocytes during the transition from anagen III to anagen VI increase not only in number, they also show increased dendricity, develop more expansive Golgi and rough endoplasmic reticulum, increase the size/number of their melanosomes, and begin to transfer mature melanosomes to precortical keratinocytes. As mentioned earlier, DCT protein and even its mRNA is undetectable in melanogenic melanocytes of the human anagen scalp hair bulb (Commo et al., 2004a,b), highlighting an important species-specific difference with mice. Melanocyte proliferation normally ceases by anagen VI (full anagen), although some proliferation of bulbar melanocytes can be seen in anagen VI hair follicles when maintained in ex vivo (D.J. Tobin, personal observation). Both the expression and activity of tyrosinase remain constant during mid to late anagen VI, but decreases rapidly during late anagen VI to catagen transition phase, to become undetectable or very low in catagen (Slominski et al., 1994; Commo and Bernard, 2000).The expression of other melanogenesis-related proteins follows a similar pattern. This physiologic decrease in follicular melanogenesis may reflect exhaustion of the active signaling system that stimulates melanogenesis, and/or the production of inhibitors of melanocyte activity. Also related may be changes in the redox status of the hair follicle which occur at this stage (see accompanying perspective of K Schallreuter).
Even before catagen-associated structural changes are apparent in the hair bulb, the earliest signs of imminent hair follicle regression include the retraction of melanocyte dendrites and the attenuation of melanogenesis during late anagen VI (Tobin et al., 1999). Limited keratinocyte proliferation continues for a while, so the most proximal telogen hair shaft remains unpigmented – the functional relevance of which remains enigmatic. There is some morphologic evidence for the survival of at least some previously melanogenic melanocytes in the regressing epithelial strand (Commo and Bernard, 2000). In the latter study, gp100 (Pmel17)-positive inactive melanocytes were detected in the catagen-associated epithelial column and also in the telogen sac. Melanocytes are detectable in early anagen as newly recruited immature melanocytes derived from a melanocyte reservoir (Nishimura et al., 2002; Tobin et al., 1999) and do not appear to be re-activated from pre-existing hair bulb melanocytes that were melanogenically active during the previous anagen phase. This is supported by the existence of a population of immature DCT-positive melanocytes unaffected by blocking anti-KIT antibody in the murine bulge (Botchkareva et al., 2001). These melanocyte ‘stem’ cells are located at the base of the permanent part of the hair follicle and are immature, slow cycling, self-maintaining and are fully competent to regenerate progeny at early anagen (Nishimura et al., 2002).
There is evidence of some plasticity in the embryologic segregation of epidermal and follicular melanocytes in the adult cutaneous melanocyte system. This may be revealed by the capacity of hair follicle melanocyte stem cells to enter vacant niches, including (via migration to) the epidermis (Nishimura et al., 2002). Still it may be possible, however, that some ‘new generation’ melanogenically active melanocytes derive from a population of catagen-surviving melanocytes (Commo and Bernard, 2000).
In any event, the majority of the highly melanotic (possibly terminally differentiated melanocytes) hair bulb melanocytes do not survive catagen (Tobin et al., 1998), and deletion of individual melanotic melanocytes by apoptosis was confirmed using well-described ultrastructural features and TUNEL/TRP1 colocalization. A very curious feature of both murine and human late anagen/catagen hair follicles is the finding that some pigment formed during late anagen fails to become incorporated into the hair shaft and instead is transported to the follicular papilla, epithelial strand or connective tissue sheath of catagen hair follicles. This redistribution of follicular pigment is likely to involve phagocytosis by macrophages, Langerhans cells, and follicular papilla fibroblasts, the first two of which increase in number during hair follicle regression (Tobin, 1998).
Regulation of pigmentation in the human hair follicle
The accompanying perspective by Shosuke Ito and Kazumasa Wakamatsu focuses on the biochemistry of melanin formation at enzyme, pH and other levels, while in another Karin Schallreuter considers how oxidative stress and other redox issues relate to the regulation of follicular pigmentation. In his perspective, Ralf Paus examines the neuroendocrinology of human hair follicle as it relates to pigmentation. Pigmentation of human hair fibers is affected by numerous intrinsic factors including hair-cycle-dependent changes, body distribution, racial and gender differences, variable hormone responsiveness, genetic defects, age-associated changes, etc.
The multi-step nature of melanosome biogenesis and melanogenesis involves several positive and negative regulators/factors including growth factors, cytokines, hormones, neuropeptides and neurotransmitters, eicosanoids, cyclic nucleotides, nutrients, microelements, cations/anions, etc. (see in Nordlund and Ortonne, 2006). These factors may act via autocrine, paracrine, and endocrine mechanisms (cf. Slominski et al., 2004), and while much of the available literature pertains to regulatory loops in epidermal melanocytes similar regulators (except UVR direct effects) may also operate in follicular melanogenesis. Recent data, however, suggest different roles for DCT and nitric oxide synthases in human epidermal versus follicular melanocytes (Commo et al., 2004a,b; Sowden et al., 2005), as well for many neurohormones including a-MSH and ACTH (Tobin and Kauser, 2005).
Here I focus on more general cell biological aspects of follicular melanogenesis. Melanogenesis can be divided simply into: (i) melanosome biogenesis and (ii) the biochemical pathway that converts phenylalanine/l-tyrosine into melanin. Both processes are under complex genetic control, which encodes a range of enzymes, structural proteins, transcription factors, receptors, and growth factors. Melanosome structure correlates with the type of melanin produced. Melanocytes in black-hair follicles contain the largest number of, and most electron-dense, melanosomes (eumelanosomes), each with an internal fibrillar matrix.
Brown-hair bulb melanocytes contain eumelanosomes that are somewhat smaller, but phenotypically similar to black-hair melanosomes, while blonde hair bulbs produce weakly melanized melanosomes with often only the melanosomal matrix visible. Red-hair pheomelanosomes are spherical in shape and by contrast, contain a vesicular matrix with melanin deposited irregularly as blotches. Less is known about the events involved in the formation of the pheomelanosome (producing red/yellow melanin), although tyrosinase activity may appear earlier in these melanosomes (Simon et al., 2008). Interestingly, both eumelanogenic and pheomelanogenic melanosomes can exist in the same normal human melanocytes. However, while eumelanosomes maintain structural integrity upon extraction from the hair keratin matrix, pheomelanosomes tend to fall apart (Liu et al., 2005). This may in part relate to their marked difference in protein content. Whereas black-hair eumelanosomes have an average of 14.6% amino acids content, red-hair pheomelanosomes exhibit more than 44%.
The constitutive color of an individual’s hair is because of absolute tyrosinase activities, rather than levels of tyrosinase protein expression. Thus, tyrosinase regulation is critical, being controlled not only by the supply of l-tyrosine but also by the stability/activity of tyrosinase and tyrosinase-related proteins.
Intracellular l-phenylalanine uptake and turnover to l-tyrosine via phenylalanine dehydroxylase (PAH) activity is vital for substrate supply of tyrosinase. While PAH acts in the cytosol of melanocytes, tyrosinase is positioned in the melanosomal membrane. Phenylalanine dehydroxylase activities have been found to correlate positively with skin phototypes (Schallreuter et al., 2004), although it is not clear if a similar correlation exists for hair color. Recently, we observed that similar eumelanin levels can be produced by isolated melanocytes from skin phototype I and IV when grown under identical culture conditions, although importantly the former responds to UVR stimulation with a marked increased in pro-inflammatory PGE2 production (Gledhill et al., 2010).
This finding appears to suggest significant skin phototype involvement in melanocyte behavior above and beyond melanogenesis alone. Also of interest to our laboratory is the co-expression of pale (skin phototype I/II) epidermal pigmentation coexpressed with the darkest of eumelanic hair color. This author speculates that individual hair follicles may have remarkable autonomy in terms of their pigment type, which may even include variable polymorphisms in hair pigmentation.
An increasingly exciting area of focus in pigmentation research is the study of the regulation of melanin transfer between melanocytes and recipient keratinocytes. While again this has been reassessed first in epidermal melanocytes (Van Den Bossche et al., 2006; Singh et al., 2008, 2010), we are now directing attention also to the hair follicle pigmentary unit. Although, similar transfer molecular machinery may be shared by epidermal and follicular-melanin units, the very different states of the main protagonists (i.e. the cyclically active melanocytes and the non-squamating cortical keratinocytes in the hair follicle) is likely to have an influence.
Hair follicle melanocyte behavior with age
Hair color shows striking age-related changes, particularly in those of Eurasian origin. During puberty, there is often a switch from fair intermediate caliber hair to more deeply pigmented, coarser or terminal hair. Furthermore, hair fiber heterochromia may become more apparent with age, most strikingly for scalp and beard (Lee et al., 1996). Follicular melanocytes appear to be more sensitive than epidermal melanocytes to aging influence (Tobin and Paus, 2001).
This can be seen most dramatically in the rapid onset of hair graying/canities (the progressive loss of natural hair pigmentation with age) compared with the very gradual change in skin pigmentation. Canities/graying first appears in our 30s and so is unlikely to have exerted significant evolutionary selective pressure, occurring as it does after reproductive peak age. Moreover, some melanocytes in the epidermis may even become more melanogenically active with age, especially if previously exposed to significant UVR (e.g. in solar lentigo) (Aoki et al., 2007). These findings are likely to reflect significant differences in the epidermal and follicular microenvironments in terms of aging.
For every decade after 30 yr of age, the number of pigment-producing melanocytes in exposed/unexposed epidermis decreases by 10–20% (Whiteman et al., 1999), accounting for much of the loss of skin (epidermis) tone with age. Nevertheless, epidermal melanocytes are relatively long-living cells, protected in part from reactive oxygen species (ROS) (including those generated during melanogenesis) by their high expression of anti-apoptotic cell survival factors, e.g., bcl-2. However, the most dramatic age-related change in hair pigment is the onset of hair graying (Tobin and Paus, 2001), which is the gradual age-dependent dilution of hair color to gray or white, also known as ‘senile’ canities. The increasing longevity of human life inevitably means we will spend an increasing proportion of our lives sporting (or attempting to hide) this sign of lost youth (Schnohr et al., 1998).
The examination of melanocyte aging has only recently been pursued with any particular vigor (Bandyopadhyay et al., 2001; Bennett and Medrano, 2002). Clinical observation suggests that the follicular- and epidermal-melanin units have a different ‘melanogenetic clock’. Accumulation of oxidative damage is an important determinant of the rate of cell aging, although it is unclear whether it is the primary cause of aging (Kauser et al., in press).
It is likely that the antioxidant systems within the hair follicle melanocyte become impaired with age, leading to uncontrolled damage to the melanocyte itself from its own melanogenesis-related oxidative stress. In addition, melanin synthesis, by its very nature, produces mutagenic intermediates. Reactive oxygen species can damage DNA (both nuclear and mitochondrial), result in the accumulation of mutations, and can induce both oxidative stress and antioxidant mechanisms. Thus, the induction of replicative senescence in melanogenic hair bulb melanocytes may be an important protective mechanism against cell transformation.
The extraordinary melanogenic activity of pigmented bulbar melanocytes (up to 10 yr in some scalp hair follicles) is likely to generate large amounts of ROS via the oxidation of tyrosine and dopa to melanin (Hegedus, 2000). A relatively small number of melanocytes (<100 cells per scalp anagen hair follicle) can, in a single hair growth cycle, produce sufficient melanin to intensely pigment up to 1.5 m of hair shaft.
Moreover, they do this within the context of a melanin-laden cell cytoplasm. In this way, hair bulb melanocytes are very different from melanogenically active epidermal melanocytes, which retain few fully mature melanosomes in their cytoplasm at any one time. This intrinsic ability of bulbar melanocytes to ‘pool’ melanin may make them more vulnerable than epidermal melanocytes to the toxic elements of melanogenesis. If not adequately removed, an accumulation of these ROS may generate significant oxidative stress in both the melanocyte itself and in the highly proliferative anagen hair bulb epithelium.
Thus, in these circumstances, melanogenic bulbar melanocytes are perhaps best suited to assume a post-mitotic, terminally differentiated ‘(pre)senescence’ status to prevent cell transformation. Recent work suggests that the follicular-melanin unit of graying hair is associated with increased melanocyte apoptosis and oxidative stress (Arck et al., 2006). Moreover, this study also reported that the ‘common’ deletion in mitochondrial DNA (associated with oxidative stress) occurred more prominently in graying compared to normally pigmented hair follicles. Graying hair follicles were also less well equipped to handle an exogenous oxidative stress, which is likely to be the result of impaired antioxidant mechanisms (Arck et al., 2006).
The synthetic capacity of bulbar melanocytes is greatest during youth when the scalp follicular-melanin unit is only a few cycles old and is able to make use of the full post-puberty hormonal stimulus. On average, an individual scalp hair follicle will experience fewer than 15 melanocyte seedings from the presumptive reservoir in the outer root sheath to the hair bulb in the average fully ‘gray-free’ lifespan of 35 yr for Caucasians (Keogh and Walsh, 1965). Interestingly, repeated plucking of hair from vibrissae follicles leads to the eventual re-growth of gray hair (Ibrahim and Wright, 1978), again suggesting limited capacity of the pigmentary reservoir.
In any event, the onset and progression of hair graying correlates closely with chronological aging and occurs to varying degrees in all individuals, regardless of gender or race. Age of onset also appears to be genetically controlled and inheritable. Thus, the average age for Caucasians is mid-30s; for Asians, late-30s; and for Africans, mid-40s. Similarly, hair is said to gray prematurely if it occurs before the age of 20 in whites, before 25 in Asians, and before 30 in Africans. A good rule-of-thumb is that by 50 yr of age, 50% of people have 50% gray hair. Clearly, the darker the hair, the more noticeable early graying will be.
However, graying can be more extensive in dark hair before total whitening is apparent; the reverse is true for blond hair. Graying usually first appears at the temples and spreads to the vertex and then the remainder of the scalp, affecting the occiput last. Beard and body hair is usually affected later. Graying often follows a wave that spreads slowly from the crown to the occiput.
The last few years have seen significant attempts to increase our understanding of the mechanisms underlying hair graying (at least in mice), and these have subsequently been enthusiastically applied to explain human canities. Thus, Nishimura et al. (2002) published the first definitive report on the location of hair follicle melanocyte stem cells in mice. Very quickly, the view emerged that canities is caused by deficient maintenance of these stem cells in the anatomically distinct ‘bulge’ compartment of the mouse hair follicle outer root sheath. Nishimura’s group along with others followed this with further evidence from mouse studies that these events implicated Pax3, MITF, BCL-2, and ATM (Nishimura et al., 2005, Osawa et al. 2005; Inomata et al. 2009).
However, what these studies mean for human scalp hair graying in the third/fourth decade of life remains unclear, although human canities-affected scalp hair follicle may indeed also show a depletion of melano(blasts)cytes in the upper outer root sheath (Nishimura et al., 2005) and also lower outer root sheath (Commo et al., 2004a,b). Still it is the view of this author that much of what is operationally relevant to graying in human scalp where anagen can last for up to 10 yr pertains to deficiencies also in the hair bulb complement of differentiated melanocytes. Thus, only on the transition from one long-lasting anagen to another may the melanocyte stems cells be called upon to intervene in the hair follicle melanin-production event.
Histopathology of canities in human hair follicles
Canities-affected ‘gray’ hair may in most cases be illusory – an impression of gray provided by the admixture of closely positioned fully white and fully pigmented individual hair fibers. Here unpigmented hair grows in as a white-hair fiber. However, canities can also affect individual hair follicles (Figure 4) part way through anagen VI. Thus, these hair follicle-specific changes may result in either a gradual loss of pigment over time and over several cycles or a gradual loss of pigment along the same hair shaft (i.e., within the anagen phase of a single hair cycle). While few pigment granules are present in truly white-hair shafts, melanin granules can be readily detected within the precortex of gray hair follicles.
Pigment loss in graying hair follicles is because of a marked reduction in melanogenically active melanocytes in the hair bulb of gray anagen hair follicles. A true gray hair bulb shows a much reduced, yet detectable, dopa reaction (tyrosinase activity), while white-hair bulbs are broadly negative. However, there also appears to be a specific defect of melanosome transfer in graying hair follicles, as keratinocytes may fail to contain any melanin granules despite being close to melanocytes with a moderate number of melanosomes (D.J. Tobin, unpublished observations).
Further evidence of a defective melanocyte–keratinocyte interaction is suggested by the presence of melanin debris in the graying hair bulb and surrounding dermis. Moreover, residual hair bulb melanocytes in canities-affected hair follicles often appear hypertrophic, although this may reflect a reduction in dendricity rather than an overall increase in cell volume. Moreover, melanosomes are often packaged within autophagolysosomes suggesting that these melanosomes are defective, perhaps even leaking reactive melanin metabolites. Autophagolysosomal degradation of melanosomes is usually followed by death of the melanocyte itself (Sato et al., 1973).
The involvement of ROS in the histopathology of canities is supported by the observation that melanocytes in graying and white-hair bulbs may be vacuolated, a common cellular response to increased oxidative stress. Degenerative change in canities-affected hair bulbs can resemble apoptosis and is reminiscent of melanocyte degeneration in alopecia areata where pigmented hair follicles are preferentially targeted by an aberrant immune response (Tobin et al., 1990).
Loss of melanocytes can apparently occur very rapidly from canities-affected hair bulbs, as suggested by incontinent/debris melanin located in the follicular papilla and/or connective tissue sheath of hair follicles that lack any morphologic evidence of melanogenesis or melanocytes in their hair bulb. This presence of pigment debris close by/in gray or white-hair follicles would appear to indicate a recent loss of previously melanogenic melanocytes. The removal of melanogenic melanocytes from the hair bulb of graying and white-hair follicles may also be associated with a parallel increase in dendritic cells including Langerhans cells (Tobin, 1998). Re-location of these antigen-presenting phagocytic cells from the upper hair follicle may be in response to antigens released from, or expressed on, degenerating canities-affected melanocytes.
Given their close physical interaction, it is likely that bulbar melanocytes influence precortical keratinocyte behavior in several ways. Melanin transfer to keratinocytes appears to reduce their proliferative potential and increase their terminal differentiation. Indeed, white beard hair has been shown to grow faster than adjacent pigmented hair, and unpigmented hair follicles exhibit a higher rate of hair fiber elongation than matched pigmented hair follicles in ex vivo culture (Arck et al., 2006; Nagl, 1995).
In this way, melanosomes may act as ‘regulatory packages’ (Slominski et al., 1993), e.g., by providing a buffer for calcium with resultant implications for second messenger/cell signaling in melanogenesis, melanosome transfer, and keratinocyte differentiation. Furthermore, the saturation binding of transition metals (e.g., iron, copper) to melanin provides effective antioxidant defense for the melanosome-receiving keratinocyte. Melanocytes may also influence neighboring keratinocytes via the production of various cytokines, growth factors, eicosanoids, adhesion molecules, and extracellular matrix (Tang et al., 1994; Herlyn and Shih, 1994).
Further clinical evidence of melanocyte–keratinocyte interactivity may be the basis of the coarser, wirier, and more unmanageable nature of gray/white hair compared to pigmented hair, reflecting their different chemical and physical properties (Van Neste and Tobin, 2004). Indeed, gray hair is often unable to hold a set and is more resistant to incorporating artificial colorant molecules. Moreover, it appears that the aging hair follicles can reprogram their matrix keratinocytes to increase the production of medullary, rather than cortical, keratinocytes.
Outlook – future developments
Much remains to be established regarding the regulation of hair pigmentation in human health and disease. While murine studies will continue to offer the most direct way of investigating mechanisms underpinning aspects of hair pigmentation, it is likely that we will continue to discover more and more ways in which these two species exhibit important differences. Several rather enigmatic issues remains including the function of amelanotic melanocytes distributed in the outer root sheath of human scalp hair follicles (including those of the so-called stem cell compartment).
We have known for over 50 yr that follicular melanocytes are involved in re-pigmentation/repopulation of the epidermis under emergency circumstances (e.g., after wounding) (Starrico, 1961). Their lack of direct contribution to hair pigmentation may indicate that their status in the senile white-hair follicle is no longer permissive for migration to the melanogenic zone during early anagen, as apparently occurs in pigment-producing hair follicles. Some insights into possible roles for outer root sheath amelanotic melanocytes have been garnered from increasingly convincing anecdotal reports of induced scalp hair re-pigmentation after particular therapy including radiation treatment for cancer (Shetty, 1995). Here, reversal of canities is likely to result from radiation/cytokine-induced activation of outer root sheath melanocytes/melanoblasts. This raises the attractive possibility that these melanocytes may be induced to migrate and differentiate to naturally re-pigment graying hair follicles.
This author remains fascinated by the apparent genetic stability of the melanocytes in the human hair follicle pigmentary unit. The remarkable rarity of melanomas derived from hair follicles suggests that this mini-organ strictly controls melanocyte behavior despite the fact that melanocytes in this location exhibit the most impressive evidence of ‘plasticity’ in the sense that these cells routinely undergo multiple rounds of proliferation, followed by differentiation and death by apoptosis and cell renewal. Clearly, the hair follicle can maintain remarkable control of these cells in this location, and it will be fascinating to dissect the nature of this regulation. This knowledge may be useful if applied to cases where melanocytes lose control, including during melanomagenesis.
original article http://onlinelibrary.wiley.com/doi/10.1111/j.1755-148X.2010.00803.x/full