Age-dependent changes in eumelanin composition in hairs of various ethnic origins


Shosuke Ito, Department of Chemistry, Fujita Health University School of Health Sciences, Toyoake, Aichi 470-1192, Japan. Tel.: +81-562-93-2595; fax: +81-562-93-4595; e-mail:


Hair pigmentation is one of the most conspicuous phenotypes of humans. From a chemical point of view, however, data remain scarce regarding human hair pigmentation characteristics. To determine melanin content and composition in human eumelanic hair from individuals of different ethnic origins and at different ages, we collected hair from 56 subjects with eumelanic hair from each group of African-American, East Asian, and Caucasian origin. The 56 subjects consist of 14, seven each of males and females, each from four age classes of younger than 11, between 12 and 19, between 20 and 45, and older than 46. We analysed hair colour scale, total melanin value, and contents of pyrrole-2,3,5-tricarboxylic acid (PTCA) and pyrrole-2,3-dicarboxylic acid (PDCA). We measured age-dependent increases in the relative quantity of eumelanin in pigmented human hairs in the three ethnic groups. Regarding melanin composition, we observed an increase in the PDCA/PTCA ratio with age in African-American and Caucasian hairs until approaching the quite constant level of the ratio in East Asian hairs in the elderly individuals. Our results evidence differences in the content and composition of eumelanin in human hair among African-American, Caucasian and East Asian individuals. Furthermore, we show evidence of age-dependent changes in the quantity and quality of eumelanin in pigmented human hairs. In particular, the age-dependent modification of the PDCA/PTCA ratio, a marker for 5,6-dihydroxyindole units in eumelanin, suggests a chronological evolution of hair follicle melanocyte phenotype (e.g. decrease in dopachrome tautomerase expression).


Melanocytes produce two chemically distinct types of melanin, black to brown eumelanin and yellow to reddish-brown pheomelanin. It is the quantity and the ratio of eumelanin to pheomelanin that mainly determines the colour of hair, skin and eyes [1]. Both eumelanin and pheomelanin derive from the common precursor dopaquinone that is produced from tyrosine by the action of tyrosinase, the product of the Tyr gene [2, 3]. Dopaquinone, as an ortho-quinone, is a highly reactive intermediate and undergoes intramolecular cyclization to give cyclodopa [4, 5]. Cyclodopa is then rapidly oxidized to give dopachrome. When not accelerated by any additional factors, dopachrome undergoes mostly decarboxylative rearrangement to form 5,6-dihydroxyindole (DHI). However, dopachrome tautomerase (DCT) or tyrosinase-related protein-2, the product of the DCT gene, catalyzes the tautomerization of dopachrome to 5,6-dihydroxyindole-2-carboxylic acid (DHICA)[6]. The dihydroxyindoles DHI and DHICA are then further oxidized to produce the eumelanin polymer. Oxidative polymerization of DHI is catalysed directly by tyrosinase or indirectly by dopaquinone [7], whereas oxidation of DHICA appears to be catalysed by tyrosinase in humans [8].

Quantity of eumelanin can be evaluated as absorbance at 500 nm after solubilization in aqueous Soluene-350 [9–11]. This absorbance value, when reported to the initial hair mass from which melanin was extracted, is called total melanin (TM) value in this study. Another parameter for the eumelanin content is pyrrole-2,3,5-tricarboxylic acid (PTCA), a specific product of DHICA-derived units in eumelanin, produced upon acidic permanganate oxidation [12]. Alkaline hydrogen peroxide oxidation gives, in addition to PTCA, pyrrole-2,3-dicarboxylic acid (PDCA) as a specific degradation product of DHI-derived units in eumelanin, although in lower yields compared with PTCA [13]. As a measure of quality of eumelanin, the ratio of DHI to DHICA appears to be most important to describe the melanin biosynthetic pathway involved. The DHI/DHICA ratio can be assessed by the following two parameters: PDCA/PTCA ratio [13] and PTCA/TM ratio [9, 10, 14]. As the content of DHICA increases, the PDCA/PTCA ratio decreases whereas the PTCA/TM ratio increases.

Hair pigmentation is one of the most conspicuous phenotypes in humans [15], with specific characteristics in comparison with epidermal pigmentation [16]. It is highly variable in colour, ranging from black, dark brown, light brown, and blond to red. The underlying genetic basis for this diversity in human hair colour has been extensively studied in recent years [17–20]. Hence, besides the long-established determinant role of the MC1R locus for the red hair phenotype in humans, recent genomic studies, especially genome-wide association studies, led to the identification of several new genes involved in the normal variation of human hair colour, e.g. TPCN2, MATP/SLC45A2, IRF4 and SLC24A4 [18]. Furthermore, significant differences in genetic variation were identified among Asian, Caucasian, African, and African-American in genes associated with the melanin biosynthetic pathway, e.g. MATP/SLC45A2, SLC24A5, and SLC24A4 [18, 21, 22], some of which were shown to affect melanin content and tyrosinase activity [23]. These genes encode putative ion transporters, and may act in maintaining ion and pH homeostasis in melanosomes, known to be an integral point to modulate tyrosinase activity and melanogenesis [15, 24, 25]. Thus, cumulative data suggest that normal variation of pigmentation in humans relies at least in part on genes acting directly in the melanin biosynthetic pathway with ancestry origin allelic polymorphisms.

Melanin determinations carried-out in mice hair were very informative to match genetic background, melanin biosynthetic pathway and melanocyte phenotypes [15]. In contrast, biochemical studies on human hair melanin are rather scarce. As shown for melanin determination in mice hair, systematic melanin determination in human hair should give insights for understanding molecular control of melanin synthesis and melanocyte phenotype within human hair bulbs, and eventually should help to correlate genetic data to the ‘chemical’ phenotype of pigmentation. In attempt to bring clues to this objective, we examined the quantity and quality of melanin in eumelanic hair in humans in relation to gender, to ancestry origins, and to age, in 168 individuals among African-American, East Asian, and Caucasian.

Materials and methods

Hair samples

We collected pigmented hair from the temples of 56 subjects with eumelanic hairs (black, dark brown, medium brown, to light brown) from each ethnic group of African-American (from U.S.A.), East Asian (from China) and Caucasian (from France) origin living in the corresponding country. From African-American and East Asian, we collected mostly black to dark brown hairs that are representative of hair colour in those populations. From Caucasian, we collected eumelanic hairs, mostly medium brown to light brown hairs, and excluded blond hair in addition to pheomelanic, reddish hair. Thus, our Caucasian hair sample does not represent the diversity of hair colour in that population, but represent the eumelanic subset of Caucasian hair. Those 56 subjects consist of 14 each from four age classes (AC) of younger than 11 (AC1, children), between 12 and 19 (AC2, adolescent), between 20 and 45 (AC3, adult), and older than 46 (AC4, elderly). Fourteen subjects in each age group consist of seven each of males and females. Approximately 50 mg of hair from the base of hair shaft were collected for analyses (cut closely to the scalp). Subjects declared not using hair colour products (to avoid misinterpretation of natural hair colour). The subjects gave an informed consent. Hair colours were visually assessed by a trained experimenter using the L’Oreal’s lightening colour scale. It is a visual scale in 10 points, described by Zviak and Millequant [26]. The natural hair colour or hair tone (HT) represents the darkness of the hair colour from Black (HT = 1) to Pale Blond (HT = 10), irrespective of any kind of shade.

Chemical analyses

Alkaline H2O2 oxidation to measure eumelanin (as PTCA and PDCA) was performed as described in Ito et al. [27]. In brief, approximately 2 mg of hair sample (cut in a length of <5 mm) was taken in a 10-mL screw-capped conical test tube, to which 100 μL water, 375 μL 1 mol L−1 K2CO3 and 25 μL 30% H2O2 (final concentration: 1.5%) were added. The mixture was mixed vigorously at ca. 25°C for 20 h. The residual H2O2 was decomposed by adding 50 μL 10% Na2SO3, and the mixture was then acidified with 140 μL 6 mol L−1 HCl. The reaction mixture was centrifuged at 4000 g for 1 min, and an aliquot (80 μL) of the supernatant was directly injected into the HPLC system. H2O2 oxidation products were analysed with HPLC [27].

HI reductive hydrolysis to measure pheomelanin (as 4-AHP) was performed as described in Wakamatsu et al. [28].

Soluene-350 solubilization to measure TM was performed as described in Ozeki et al. [10] with a minor modification. In brief, approximately 3 mg of hair sample was taken in a 10-mL screw-capped conical test tube, to which 100 μL water and 900 μL Soluene-350 (from Perkin-Elmer, Waltham, MA, U.S.A.) was added. The tube was Vortex-mixed and heated at 100°C (boiling water bath) for 15 min. After cooling, the tube was Vortex-mixed and heated again at 100°C (boiling water bath) for additional 15 min. The mixture was centrifuged at 4000 g for 3 min, and the supernatant was analysed for absorbance at 500 nm (A500). The data were subtracted with absorbance of 0.020 for the background [10]. TM was calculated as the ratio of A500 to hair mass (mg).

Statistical analyses

Students’t-test, Pearson’s product–moment correlation test, Jonkheere-Terpstra test and Mann–Whitney test were employed with SPSS software.


Pigmentation scores in the three ethnic groups

As shown in Table I, HT scale was similar between African-American and East Asian, whereas it was much greater (lighter in colour) in Caucasian. TM value was greatest in African-American hairs, followed by East Asian (ca. 80% of African-American) and then by Caucasian (ca. 40% of African-American). PTCA value showed a similar trend. PDCA/PTCA ratio was greatest in East Asian hairs, followed by African-American and then by Caucasian. Consistent with this observation, PTCA/TM ratio showed an opposite trend. We also examined contents of 4-amino-3-hydroxyphenylalanine (4-AHP), a pheomelanin marker, in randomly selected 12 hair samples each from the three ethnic groups and the four age groups. We did not detect significant levels of 4-AHP accounting for a >1% content of pheomelanin in TM. Therefore, we may exclude any possible contribution of pheomelanin in the hairs examined in this study. Because of the differences between ethnic groups, the following analyses were performed separately in groups of origin.

Table I.   Average and SEM values of the various pigmentation parameters in the three ethnic groups
Ethnic groupHTTM (A500 per mg)PTCA (ng per mg)PDCA/PTCA ratioPTCA/TM (ng per A500)
  1. TM, total melanin.

  2. P ≤ 0.001 (a, b, c, d, e, f, g, h, j, k, m, n); P ≤ 0.005 (i); P ≤ 0.05 (l) (T-test). (a) denotes a significant difference between African-American and Caucasian in hair tone (HT).

African-American2.45 ± 0.15(a)0.290 ± 0.010(c,d)280 ± 14(f,g)0.117 ± 0.005(i,j)976 ± 33(l,m)
East Asian2.41 ± 0.10(b)0.234 ± 0.008(c,e)205 ± 8(f,h)0.138 ± 0.005(i,k)883 ± 25(l,n)
Caucasian4.51 ± 0.10(a,b)0.122 ± 0.006(d,e)136 ± 6(g,h)0.087 ± 0.004(j,k)1143 ± 38(m,n)

Effect of age on quantity and quality of eumelanin

We examined whether there were any differences in quantity and quality of eumelanin based on gender and age in the three groups of origins (Pearson’s product–moment correlation coefficient). We did not observe any statistically significant differences between males and females in the three ethnic groups (data not shown) for all the pigmentation parameters.

As shown in Table II, distinct effects of age on HT, TM, PDCA/PTCA and PTCA/TM appeared in the ethnic groups. HT scale decreased slightly with age (with P < 0.05 for East Asian and Caucasian) and TM value increased with age (with P < 0.05) in hairs from the three ethnic groups. Evolutions of both parameters with age suggest a slight increase in colour intensity that could be due to an increase in the melanin content in hair with age. Further, analyses of TM values among the four age classes of an equal number (n = 14) of children, adolescent, adult and elderly subjects (AC1–4) showed significant differences in the three group of origins: in African-American, a trend of increase with age (P = 0.059) (Fig. 1A); in East Asian, a trend of increase with age (P = 0.060) (Fig. 1B) and in Caucasian hairs, highly significant increases with age (P = 0.002) (Fig. 1C). We also examined correlation of TM values with HT values in the three ethnic groups. As shown in Fig. 1D, TM values correlated well with HT scale in each group (P < 0.01). Interestingly, results showed that African-American hairs had slightly higher relative melanin content (TM) in comparison with East Asian hairs with same HT values.

Table II.   Correlation of the various pigmentation parameters with age in the three ethnic groups
Ethnic groupHair toneTM (A500 per mg)PTCA (ng per mg)PDCA/PTCA ratioPTCA/TM (ng per A500)
  1. TM, total melanin.

  2. x and y denote age and each parameter value, respectively. Values in bold are those with significance (P < 0.05, Jonckheere-Terpstra test).

African-Americany = −0.014x + 2.84
r = −0.219, P = 0.105
y = 0.0014x + 0.249
r = 0.304, P = 0.023
y = 0.56x + 264
r = 0.094, P = 0.489
y = 0.0007x + 0.094
r = 0.346, P = 0.009
y = −2.68x + 1051
r = −0.198, P = 0.144
East Asiany = −0.012x + 2.74
r = −0.311, P = 0.020
y = 0.0010x + 0.204
r = 0.341, P = 0.010
y = 0.78x + 183
r = 0.240, P = 0.075
y = 0.0001x + 0.133
r = 0.074, P = 0.585
y = −0.46x + 896
r = −0.047, P = 0.734
Caucasiany = −0.016x + 4.96
r = −0.422, P = 0.0012
y = 0.0009x + 0.097
r = 0.461, P = 0.0004
y = 0.42x + 124
r = 0.189, P = 0.163
y = 0.0008x + 0.066
r = 0.451, P = 0.0005
y = −4.53x + 1266
r = −0.307, P = 0.021
Figure 1.

 Effect of age on total melanin (TM) in African-American hairs (A), in East Asian hairs (B), and in Caucasian hairs (C). (A) TM value shows a trend of increase with age (P = 0.059, by the Jonckheere-Terpstra test) with significant differences in AC2 vs. AC3 (P = 0.012) and AC2 vs. AC4 (P = 0.019, by the Mann–Whitney test). (B) TM value shows a trend of increase with age (P = 0.060) with a significant difference in AC1 vs. AC4 (P = 0.046). No difference among other age classes. (C) TM value shows a significant increase with age (P = 0.002), with significant differences in AC1 vs. AC3 (P = 0.050), AC1 vs. AC4 (P = 0.005), AC2 vs. AC3 (P = 0.041), and AC2 vs. AC4 (P = 0.025). (D) Correlation of TM value to hair tone scale in hairs from the three ethnic groups. Correlation graphs are y = −0.041x + 0.390 (r = −0.539) for African-American, y = −0.050x + 0.354 (r = −0.608) for East Asian, and y = −0.031x + 0.264 (r = −0.643) for Caucasian hairs (all P < 0.001).

Regarding the quality of melanin, PDCA/PTCA ratio increased with age in African-American and Caucasian hairs (P < 0.01) (Table II), but not in East Asian hairs. PTCA/TM ratio showed a similar trend as PDCA/PTCA ratio (in the opposite direction), but with less significant differences. We then examined differences in PDCA/PTCA ratio among the four age classes (AC1–4). In African-American hairs (Fig. 2A), PDCA/PTCA ratio showed a significant increase (P = 0.006). In East Asian hairs (Fig. 2B), PDCA/PTCA ratio did not show a significant change (P = 0.176). In Caucasian hairs (Fig. 2C), PDCA/PTCA ratio showed highly significant increases with age (P = 0.001). Interestingly, PDCA/PTCA ratios evolved similarly in African-American and Caucasian, and in the elderly individuals tend to reach the quite constant PDCA/PTCA ratio measured in the Asian hair samples (Fig. 2D).

Figure 2.

 Effect of age on PDCA/PTCA in African-American hairs (A), in East Asian hairs (B), and in Caucasian hairs (C). (A) PDCA/PTCA ratio shows a significant increase with age (P = 0.006, by the Jonckheere-Terpstra test) with a significant difference in AC1 vs. AC4 (P = 0.014, by the Mann–Whitney test). (B) PDCA/PTCA ratio does not show any significant differences (P = 0.176). (C) PDCA/PTCA ratio shows significant increase with age (P = 0.001), with significant differences in AC1 vs. AC3 (P = 0.027), AC1 vs. AC4 (P = 0.018), AC2 vs. AC3 (P = 0.007), and AC2 vs. AC4 (P = 0.012). (D) Correlation of age with PDCA/PTCA ratio in hairs from the three ethnic groups. Correlation data are shown in Table II.


In this study, we observed a good correlation of TM value (‘chemical’ phenotype) with HT scale (‘visual’ phenotype). It is interesting, however, that HT values were almost the same between African-American and East Asian hairs, whereas TM and PTCA values were significantly higher in African-American hairs compared with East Asian hairs. This suggests that ‘visual’ phenotype is affected not only by the amount of eumelanin but also by some other factors such as the shape and the thickness of hair shaft, and size of melanosomes.

Our results showed that eumelanin contents (TM values) were significantly higher in African-American hairs, followed by East Asian and Caucasian hairs (Table I). French eumelanic hairs (Caucasian sample in our study) are lighter in comparison with East Asian and African-American eumelanic hairs (Panhard et al., submitted). Our data showed this lighter colour is associated with lower melanin contents. Genetic studies showed polymorphisms in TYR, OCA2/P, SLC45A2/MATP, SLC24A5 and TYRP1 have significant occurrences in European in comparison with African-American and Asian [18, 21, 22, 29], some of which were shown to be associated with reduced tyrosinase activity and melanin content in melanocytes [23]. Such polymorphisms could account for the lower amount of eumelanin observed in Caucasian hair in our study (40% of melanin level of African-American hair). We also observed a small but significantly lower amount of eumelanin in East Asian hairs in comparison with African-American. Polymorphisms in DCT gene evidenced specifically in East Asians could cause the minor (by 20%) reduction in hair pigmentation [17], similarly to the reduction of eumelanin content by 40% observed in the slaty mutation in Dct gene in mice [14]. With other respects, TM values were expressed per mg of hair, and the resulting data reflect the ratio between melanin biosynthetic activity and the rate of hair growth (protein production). Thus, differences in hair growth rate described in a cohort of African, Asian and Caucasian individuals [30] may also account for the differences in melanin content per materials of hair shaft observed in our study among the group of origin.

We observed an increase in TM value with age in pigmented hairs from the three ethnic groups of African-American, East Asian and Caucasian. The increase in melanin content (per mg of hair) with age in eumelanic hair could arise from progressive hair thinning [31] and a reduction in the content of moisture (water) with age. Other possibilities include an increase in tyrosinase expression/activity. Although less likely, this possibility cannot be ruled out.

An increase in PDCA/PTCA ratio with age is seen in African-American and Caucasian hairs but not in East Asian hairs. Interestingly, PDCA/PTCA ratio in African-American and Caucasian hairs increase with age to approach the level of the ratio obtained for the East Asian hairs; in elderly (AC4) individuals, the ratios were 0.139 for African-American and 0.108 for Caucasian and it was 0.139 for East Asian. The increase in the ratio of PDCA/PTCA with age in African-American and Caucasian hairs could be linked to a decrease in DCT activity with age in these ethnic populations. Our previous study [32] showed that in individuals older than 45 years, DCT protein in hair bulbs is neither detectable in brown hair nor in black hair, of African, Asian and Caucasian origins. Thus, our current data on PDCA/PTCA ratio shows that it is plausible that DCT activity is present in human hair at youth and decreases with age in human hair melanocytes because of a reduction of expression level. Lao et al. [22] showed that a polymorphism located in an intron in the DCT gene is specific to East Asian population (Han Chinese). As the polymorphism is located in an intron, it cannot affect DCT catalytic activity. But it may affect DCT level, either by influencing DCT promoter activity or by acting on mRNA processing. Taken together, it seems likely that DCT activity in East Asian hairs remains negligible throughout the life, whereas its activity in African-American and Caucasian hairs decreases gradually with age to the level of East Asian hair.

The ratio of PDCA/PTCA can be used to evaluate the content of DHICA in eumelanin [13]. Based on our calibration curve obtained for synthetic eumelanins prepared from various ratios of DHI to DHICA (our unpublished data), mean DHICA contents in the three ethnic populations are 33% for East Asian, 41% for African-American, and 53% for Caucasian hairs. Also, DHICA content in African-American hairs decreases from 50% for AC1 (children) to 33% for AC4 (elderly). Similarly, it decreases from 58% (children) to 45% (elderly) in Caucasian hairs. Thus, in humans, especially in East Asian, hair melanins consist of rather low (33%) proportions of DHICA (high proportions of DHI), similar to mouse hair (20% in our unpublished data) with slaty mutation at Dct gene [9, 14]. This high level of PDCA/PTCA ratio in human hair was also recently demonstrated in distinct hair samples of various ethnic origins [27]. The PTCA/TM ratio can also serve to evaluate the DHICA contents [9, 14]. However, the PDCA/PTCA ratio appears to be more reliable for this purpose because analysing its ratio is performed on the same sample whereas the PTCA/TM ratio requires separate determinations of PTCA and TM. Thus, our recent study has shown a very high reproducibility of the PDCA/PTCA ratio [27].

In conclusion, our results evidenced differences in the melanin composition in relation to ancestry origins in eumelanic hairs. Furthermore, we have given evidence showing age-dependent changes in the quantity and quality of eumelanin in eumelanic pigmented human hairs. Age-dependent modification of PDCA/PTCA ratio in African-american and Caucasian hairs suggests an evolution of hair follicle melanocyte phenotype with age (e.g. chronological decrease of DCT expression). This evolution could be related to modifications of the factors that control melanocytes within the hair bulb, reflecting an ageing of hair cells, e.g. dermal papilla cells and/or hair keratinocytes, or alternatively could rely on a melanocyte-ageing process leading to an aged melanocyte phenotype in hair bulb.


Authors thank John Wares, Catherine Collaudin and Liu Chen for the collection of hairs, and Philippe Bastien for advices on statistical analyses. This work was supported, in part, by a Japan Society for the Promotion of Science (JSPS) grant (No. 20591357) given to SI and KW.