`Do freckles and red hair help Irishmen catch leprechauns?'1
There are at least two reasons for being interested in the biology of hair and skin colour. First, variation in skin and hair colour is perhaps the most polymorphic of all visible differences between humans, and has historically been of profound social importance. We get used to seeing little people—they are called children—whereas even the most liberal parent shows concern when children start verbalizing their classification of human skin colours. It is likely that the influence of differences in pigmentation rivals or exceeds that of infectious diseases on human history. Second, variation in pigmentation is the most important risk factor for the major forms of skin cancer, both melanoma, and basal and squamous cell carcinoma.2,3 If one were to collect a random sample of people from around the world, their constitutional risk of developing skin cancer would vary over 100-fold, and most of this would be attributable to differences in skin colour.4,5 The interaction of skin colour and ultraviolet radiation provides a timely reminder of the difficulties of viewing nature and nurture as anything but contingent: to an epidemiologist, ultraviolet radiation is the major determinant of skin cancer, whereas for a geneticist, pigment, predominantly under genetic control, is the major determinant. Both views are of course correct.
Of course in between the two reasons given above lie a host of interesting biological questions. Can we explain the variation in human pigmentation solely in terms of protecting against ultraviolet radiation, or alternatively do we imagine that natural selection is only part of the explanation. Why are the Scandinavians taller and blond and the Irish more often red-haired? Is this all natural selection at work, or is there an element of assortive mating depending on choice of mate?
We will concentrate on recent advances in our understanding of the genetics of red hair, and will only briefly describe other aspects of the pigmentary system or the genetics of other pigmentary phenotypes including oculocutaneous albinism. The interested reader is directed to recent reviews on these latter topics.6–9
Melanin is a complex polymer produced in melanocytes in the basal layer of skin. The melanin is packaged into melanosomes and passed into the surrounding keratinocytes. In Caucasian skin, microscopy shows melanin to be largely basal, clustered over the nuclei of the basal keratinocytes as though the nuclei were wearing protective sun hats. The difference in skin colour between Blacks and Caucasians arises not from differences in the number of melanocytes but in the amount of melanin produced and the packaging of that melanin in the melanosomes.9 Because melanin is essentially a mixture of different chemical entities, quantitative analysis has proved limiting in the study of the pigmentary system. There are however within Caucasian populations two types of melanin, eumelanin which is black and phaeomelanin which is red. Although this may well turn out to be an over-simplification, it has to date been useful in understanding skin and hair colour.10–12 It is important to note that hair colour and skin colour arise not from the exclusive production of one type of melanin over another, but that both types are produced together. There are therefore at least two levels of control, the amounts of eumelanin and phaeomelanin, respectively, and the ratio between the two types.13
Balding middle-aged professors of dermatology know that hair provides nature's most effective sun block.14 In one sense this can be viewed as a recapitulation of recent evolution. With loss of body hair, the underlying skin is exposed to the hazards of ultraviolet radiation: hair very efficiently protects against UVR. Thus in the mouse, apart from the tail and the ears, which have only sparse hair, the interfollicular epidermis is devoid of melanocytes. By contrast, in man, cutaneous pigmentation arises from melanin produced by interfollicular melanocytes. For humans, the development of pigmentation as a protection against ultraviolet radiation over most of the body is a relatively recent evolutionary adaptation; in most animals the major function of pigmentation is either as camouflage or as part of sexual display. The bright blue scrotum of the vervet monkey, due to melanin in the dermis, is not there to serve as a sun-block. (Melanin deep with the skin appears blue for the same reason that the sky appears blue. Blue nevi or the Mongolian blue in new-borns result from melanin deep in the dermis).
Nevertheless there continues to be some debate about the purposes of pigmentation in humans.15,16 It has been argued that because there are several competing theories (thermal absorption, cold sensitivity etc., protection against ultraviolet radiation) it is unlikely that any are correct. This seems to the present authors illogical, and we prefer to imagine that most explanations are just wrong, and that the most compelling evidence is in favour of melanin as a useful protection against ultraviolet radiation. Melanin absorbs ultraviolet radiation across a wide spectrum, but it is particularly effective in the shorter wavelengths around 280–320 nm. This of course is the area of maximal absorption of UVR by nucleic acids and proteins. This is not to imagine that the selection for pigmentation is solely accounted for in terms of skin cancer risk. Skin cancer in most human populations occurs well beyond the age of procreation, and avoidance of it may contribute little to evolutionary fitness.17 It seems likely that avoidance of burning, and with it avoidance of the secondary infection and fluid loss that burning would bring, is a far more powerful factor. At the climatic extremes, however, skin cancer is not entirely irrelevant. Albinos in Equatorial regions can develop skin cancer when aged less than three years, so some biological protection will be selected for, given the high mutagenic load of ultraviolet radiation to which skin is exposed.18
The pattern of skin cancer in albinos is however instructive in understanding the role of the various types of melanin.14,19 Several studies have shown that the rate of squamous cell cancer of the skin is greatly elevated in albinos.18 This is to be expected if melanin is playing an important role counteracting the effects of ultraviolet radiation. By contrast, the rate of melanoma in albinos does not seem to be increased as much as one would imagine given the rates of squamous cell carcinomas. Although this could still be accounted for by under-ascertainment of amelanotic melanomas, it is possible that the situation is more complex. Various lines of evidence suggest that melanin, particularly phaeomelanin, may on exposure to ultraviolet radiation release free radicals and be toxic to melanocytes.20,21 In evolutionary terms, there may be a trade-off between the presence of melanin protecting keratinocytes and the damaging effects of melanin on melanocytes. Melanoma may be the evolutionary price paid for protection against burning: in albinism the absence of melanin exposes the individual to the risk of burning and cumulative genetic damage to keratinocytes, but (relatively) spares the individual the risk of melanoma.
We do not know the skin and hair colour of our distant ancestors.22 The majority of authors assume that our African forefathers were darker than present Northern Europeans. Given the apparent advantages of dark skin, what explanations are there for the lightening of skin found in North Europe or for that matter in the Far East and some parts of the Pacific?
The most convincing explanation is that in areas where vitamin D in the diet is marginal there is selection against dark skin because of decreased production of vitamin D in skin.23–25 Dietary change, particularly the advent of agriculture and reliance on cereal crops, would have been associated with decreased dietary vitamin D. It is a matter of recent clinical record that Indo-Asians in the UK, particularly vegetarians, are prone to rickets. In keeping with this explanation is the observation that although skin lightens in proportion to distance from the equator in northern Europe, Eskimos are relatively pigmented. This reversal of the pattern of loss of pigment could be accounted for by their meat and fish diet which is rich in vitamin D.
Whether red hair and the pale skin that is associated with it can be usefully thought of as an extension of pigment lightening is unclear. Some caveats are necessary. First, whilst red-haired individuals usually have pale skin and are very sensitive to the effects of UVR, not all sun-sensitive individuals have red hair. For instance, it is said that individuals with dark hair and very pale skin are particularly common in Wales and Ireland. Secondly, many Scandinavians are relatively pale skinned, and yet tan quite well. By contrast red-haired subjects with a similar skin colour to some Scandinavians may be more sun-sensitive. As discussed above, there is some evidence that melanin may generate harmful radicals under some conditions, and that this is much more likely with phaeomelanin than eumelanin. Thirdly, one might wonder why red hair rather than just further lightening of the skin has been selected for.
Alternative explanations for the development of red hair may be possible. For instance, in the Pacific islands there is a reasonable case for arguing that assortive mating may account for some of the variation in pigment levels between different islands in people who in evolutionary terms are closely related genetically.16,26 It may be possible that related mechanisms influence the pattern of skin and hair colour in Northern Europe. Although much discussed there is little objective data in this area, although it is said that Roman red-haired female slaves fetched a better price at sale!
Skin and hair colour provides perhaps one of the most clear examples of a complex genetic trait. Various methods of measuring hair and skin colour have been proposed, and none is completely satisfactory.27,28 Examination of hair colour shows that in many individuals the colour of individual hairs shows considerable variation; hair colour changes with age with a darkening seen in many fair-haired children and a loss of the bright red colour in redheads in middle age. Although redheads as a group tan poorly, there may be considerable variation within the group and it is possible that for instance the redheads characteristic of Neopolitans may differ in some respects from the redheads seen in the UK. The situation with regard to those with auburn hair is also unclear. It seems that the rate-limiting factor in studying the genetics of skin colour now lies with the assessment of phenotype rather than lack of technical facility in genetics.
Evidence accumulated in the first half of this century that pituitary factors, now known to be peptides including alpha-melanocyte-stimulating hormone (αMSH) could influence pigmentation in a range of animals, particularly amphibians.29,30 The relevance of these peptides in the physiological control of pigmentation in man remained unclear. Although injection of αMSH was reported to increase cutaneous pigmentation in association with UVR,31 some scepticism remained, compounded by the fact that αMSH would not consistently alter pigmentation in cultured melanocytes.32 In man, circulating levels of αMSH are low, and it was not possible to explain pigmentary differences between individuals with different skin types, or between Caucasians and Blacks, or those seen in medical conditions characterized by increased pigmentation, in terms of alterations in the level of circulating αMSH. Analogies can be made with similar concerns about the role of serum levels of αMSH in the control of sebum excretion in man. In rodents, αMSH is a major sebotrophic hormone, although in man the situation still remains unclear. The demonstration that αMSH could be produced locally in skin by Shuster and co-workers was therefore of key importance, suggesting that locally-produced peptides may be important.33
The key role for αMSH or a similar peptide ligand was confirmed following the recent cloning by various groups, based on nucleotide homology, of a family of seven-pass transmembrane G-protein-coupled receptors.30,34–36 There are now known to be five members of the melanocortin receptor family (reviewed in reference 30). MC1R is expressed in melanocytes and is thought to play a key role in controlling the switch from phaeomelanin (red) to eumelanin (black). The MC2R is better known as the ACTH receptor. Although the ligand ACTH, as is well known, can induce pigmentation, this appears to be acting through the MC1R, as individuals with MC2R mutations appear normally pigmented. Both αMSH and ACTH are active at the MC1R and their respective physiological roles in MC1R receptor activation are at present unclear. MC3R and MC4R are both expressed in the brain, and amongst other roles, mediate some of the central roles of αMSH in the regulation of body weight. The final cloned member of this family is the MC5R, which shows a wide expression, particularly in exocrine glands. Knock-outs of mice for the MC5R show a normal phenotype except that sebum production is diminished, resulting in a relative inability to dry out following wetting of the fur and consequent hypothermia.37 This observation is in keeping with earlier studies showing the important sebotrophic role for αMSH in the control of sebum excretion.38 Whether a similar role is played by this receptor and ligand in man is at present unknown.
Signalling through the MC1R is via activation of adenyl cyclase leading to elevation of cAMP. It is not known whether other pathways are activated or inhibited in the melanocyte through the MC1R. There is however evidence for involvement of both protein kinase C (PKC) and diacylglycerol (DAG) in the control of pigmentation.39,40
In the mouse and some other animals, there is clear evidence that there is a physiological antagonist at the MC1R.41,42 In wild-type agouti mice, the hair is predominantly black except for a sub-apical band of yellow hair, a result of transitory production of phaeomelanin. Transplant operations in the mouse show that agouti is produced outwith the hair follicle melanocyte by keratinocytes and that agouti can inhibit or at least act as an inverse agonist at the MC1R receptor. Mutant mice that overexpress agouti, or at least express agouti in a inappropriate range of tissues, including brain, are not only dark-haired but also obese, because of antagonism of the action of the central melanocortin receptors. Agouti is known to play a role in pigmentation in a range of mammals, although no role in the physiological control of pigmentation in man is described.30,43,44
There have been limited studies on the expression of the MC1R in man, although it appears to be expressed in human adult skin, and in cultured keratinocytes and melanocytes, compatible with its role in the control of human pigmentation.45,46 How signalling through the MC1R alters the ratio and production of phaeomelanin to eumelanin is also unclear.
The cloning of MC1R and demonstration that its mutation caused production of yellow-coated mouse mutants (in the mouse, for reasons not completely clear, phaeomelanin is associated with yellow rather than red hair) and that mutations in MC1R were subsequently associated with a red coat in a range of animals47–49 suggested that MC1R was a candidate gene for red hair in humans.
The first study to address this issue employed a case-control design, the reasons being that the mode of inheritance of red hair and the magnitude of the effect of the MC1R locus was unclear.50 We originally compared 30 unrelated Caucasian individuals with red hair who tanned poorly, and for comparison, ethnically-matched individuals who had dark hair and who tanned well. What was immediately striking was the high frequency of coding region variation in the MC1R (70%). Because the study group was not selected at random from the population, it was difficult to ascribe appropriate relative risks to the variant alleles, and because of the number of alleles found, it was difficult to be certain which might be of functional significance. Furthermore the mode of inheritance of the red-hair phenotype was not clear: in some individuals two variants were found, each on a different allele compatible with a recessive model, whereas in others only one variant allele was identified, while in still others multiple changes were found on one allele. Obvious explanations for these findings were that either variants were being missed or that mutations outside the coding region or epigenetic change were important. Finally, this study suggested that auburn hair followed red hair in terms of the genetics of MC1R. It is therefore likely that in subjects with auburn hair, the red element is independent of the fair/dark axis.
Subsequent studies have built on and clarified these findings.51–53 To further define the contribution of the MC1R variants to pigmentation, we looked for alterations in the gene in a consecutive series of individuals from an Irish population.52 Seventy-five per cent harboured a variant in the MC1R coding region, with 30% showing two variants. Three particular variants that would be predicted to impact on protein function were strikingly associated with red hair, the Arg151Cys, Arg160Trp and Asp294His mutations. These associations were highly significant, with odds ratios for red hair of between 9 and 16. In this study, no individual who had two of these changes did not have red hair, and conversely over 60% of the redheads were compound heterozygotes or homozygous for these changes. Associations were also seen with pale skin and freckles, and similar changes noted in a sample of Caucasians from different populations.
More recently we have extended these studies. Based on family studies, the trait seems to behave as a recessive, with the presence of the two variant alleles mentioned above usually being associated with red hair, and only rarely are compound heterozygotes or homozygotes for these three variants not associated with red hair (<5%) (unpublished). It would be surprising if however other loci were not important in determining pigmentary phenotype. For instance, in the study mentioned above, one individual who would have been expected to have red hair secondary to the presence of two of the three key variant alleles, only showed a tint of red insufficient to class him as red according to the operational criteria used.
Sturm's group have shed further light on this in a study of identical and non-genetically identical twins.51 They found that the above three variants were strongly associated with red hair, and that most red-haired individuals harboured two of the three variants, or were homozygous for one variant. They also showed that other factors were clearly important in that some non-identical twins with the same MC1R genotype had different coloured hair. They also suggested that one particular variant, the Val60Leu, may be associated with fair hair, and more recently described an individual with red hair with a nonsense mutation.
A recent report using transient transfection and assessment of cAMP as an endpoint has shown that the Arg151Cys variant produces loss of function.54 We have recently confirmed this, and found that two of the other variants have diminished ability to mediate the effects of MSH on cAMP (unpublished). It therefore seems likely that the majority of red-haired individuals harbour two of three mutations and that these three changes alone account for the majority of red hair in northern European individuals.
Since the MC1R is associated with skin type, it would not be surprising if there were an association with variants showing impaired function and skin cancer. Preliminary studies have indeed suggested that this is the case. One study on melanoma found a greater proportion of variant alleles in melanoma cases than in controls although the difference could be accounted for by an increased frequency of the Asp84Glu variant in the melanoma cases.55 It is not clear at present whether this variant is associated with an alteration in function, and based on population studies, it is extremely rare. Subsequent studies have shown a lower prevalence of this variant in melanoma (unpublished). Other groups have also reported an increased frequency of variant alleles in melanoma and there has been considerable interest in whether the MC1R locus may be a modifying locus for expressivity of the p16 locus in familial melanoma (Gruis N, personal communication).
To date, there is one report showing an increased frequency of the Asp294His variant in subjects with non-melanoma skin cancer.52 Because of the ease of detection of this allele using RFLP, this was the only common variant examined, but it seems likely to us that there will be an over-representation of the other mutant alleles in NMSC reflecting the greater propensity to skin cancer in subjects with red hair and pale skin. None of the reported studies have been able to deal adequately with the possibility of a heterozygote effect.
αMSH exerts a number of effects beyond its involvement in pigmentation.30,56,57 Some of these, for instance its sebotrophic effect or central effects on the control of weight gain, are easily explained in terms of different expression patterns of the various melanocortin receptors, for instance MC5R in sebaceous glands, and MC3R and MC4R in the brain. However, a body of experimental results suggests that αMSH has widespread anti-inflammatory effects in a variety of inflammatory systems, including antagonizing the effects of IL-1 or TNFα (on fever and leukocytosis), the production of cytokines on endothelial cells and inhibition of the sensitization and elicitation phases of contact sensitivity (mouse) and altering the balance between TH1 and TH2 responses.58,59αMSH modulates the production and actions of inflammatory and immunoregulatory cytokines in vitro, and modifies the course of several rodent models of inflammation and autoimmunity, such as adjuvant arthritis in rats. It is well-established that MSH antagonizes IL-1α, IL-1β, IL-6 and TNFα, leading to potent anti-inflammatory effects. In addition, MSH has been shown recently to modulate cytokines which are closely associated with key T lymphocyte effector functions, and both antigen-presenting cells and T lymphocytes possess MSH binding sites. In particular, production of IFNγ is reduced whilst production of IL-10 is increased.58,59 These effects have been linked to the ability of MSH to suppress contact sensitization and induce hapten-specific tolerance in mice, and may potentially have profound effects on determining TH1/2 balance in protective immunity, allergy and autoimmune pathologies.60,61
Expression studies suggest that some of these effects are mediated through the MC1R, as on some of the relevant cell types it is the only melanocortin receptor expressed. There is some poorly developed but interesting supportive evidence in humans. Widespread anecdotal and some experimental evidence suggests that people with different skin types respond to a range of cutaneous inflammatory insults differently (including irritant dermatitis to SDS, sellotape stripping and reaction to anthralin).62–65 These experiments are likely to underestimate the effect of the MC1R locus because of genetic heterogeneity in the sun-sensitive subjects studied.
Although perhaps fanciful, it remains conceivable that because of its cytokine role (which may predate its involvement in pigmentation) MC1R mutants may have been selected for. Many redheads themselves would like to feel that in return for their intolerance of the pleasures of the sun, nature may have furnished them with some compensation. As yet there is no convincing link with cerebral function either.
The inheritance of red hair has long been a subject of great interest for geneticists. The embarrassment has been that until recently the problem was not experimentally tractable. It is salutary to note that although many distinguished geneticists, particularly those of a mathematical persuasion, have been interested in red hair genetics, it is the technical facility of modern genetics coupled with the use of the mouse as a model organism that has allowed progress to be made.66–68 Success has relied on humble data collection rather than sophisticated modelling. What however hampers much current work in understanding the molecular physiology of pigmentation is our relative inability to define the phenotype accurately. Much of the work described in this review has relied on crude assessments of hair colour rather than chemical or physical assessment. Similarly, it has concentrated on hair colour rather than skin colour or more properly the assessment of the cutaneous response to ultraviolet radiation (UVR) because of the absence of a developed methodology for measuring the ability of a pigmented phenotype to protect against UVR. Given the obvious differences between man and mouse in the physiological function of pigmentation, these deficiencies are likely to be rate- limiting in our understanding of the genetic control of skin and hair colour.
We thank other members of our lab for unpublished results including Sion Phillips, Mark Birch-Machin, Carol Todd and Eugene Healy. The unpublished work referred to is supported by grants from the MRC, CRC, DoH, IARC and the Leech Trust.