Isolation and Biophysical Studies of Natural Eumelanins: Applications of Imaging Technologies and Ultrafast Spectroscopy


* Address reprint requests to Prof. John D. Simon,Department of Chemistry, Duke University, Durham, NC 27708-0346, USA. E-mail:


The major pigments found in the skin, hair, and eyes of humans and other animals are melanins. Despite significant research efforts, the current understanding of the molecular structure of melanins, the assembly of the pigment within its organelle, and the structural consequences of the association of melanins with protein and metal cations is limited. Likewise, a detailed understanding of the photochemical and photophysical properties of melanins has remained elusive. Many types of melanins have been studied to date, including natural and synthetic model pigments. Such studies are often contradictory and to some extent the diversity of systems studied may have detracted from the development of a basic understanding of the structure and function of the natural pigment. Advances in the understanding of the structure and function of melanins require careful characterization of the pigments examined so as to assure the data obtained may be relevant to the properties of the pigment in vivo. To address this issue, herein the influence of isolation procedures on the resulting structure of the pigment is examined. Sections describing the applications of new technologies to the study of melanins follow this. Advanced imaging technologies such as scanning probe microscopies are providing new insights into the morphology of the pigment assembly. Recent photochemical studies on photoreduction of cytochrome c by different mass fraction of sonicated natural melanins reveal that the photogeneration of reactive oxygen species (ROS) depends upon aggregation of melanin. Specifically, aggregation mitigates ROS photoproduction by UV-excitation, suggesting the integrity of melanosomes in tissue may play an important role in the balance between the photoprotective and photodamaging behaviors attributed to melanins. Ultrafast laser spectroscopy studies of melanins are providing insights into the time scales and mechanisms by which melanin dissipates absorbed light energy.

Abbreviations – 

atomic force microscopy




5,6-dihydroxyindole carboxylic acid


minimal erythemal dose


pyrrole 3,5-dicarboxylic acid


pyrrole 2,3,5-tricarboxylic acid


reactive oxygen species


retinal pigmented epithelium


scanning electron microscopy


scanning tunneling microscopy


transmission electron microscopy


ultra-high resolution


Melanin refers to a range of biologic pigments found in a variety of locations including the hair, skin, eyes, brain, and inner ear. Melanins are commonly divided into three types: the brown-black eumelanins, the yellow-reddish pheomelanins and the dark brown neuromelanins. These three melanins are distinguished by differences in their molecular precursors (1–3). Eumelanins are composed of indolic units derived from the oxidation of tyrosine. Pheomelanins are composed of benzothiazines derived from the oxidation of cysteinyldopa units. Neuromelanins are derived from the neurotransmitter, dopamine, and have properties of both pheomelanins and eumelanins: containing both indole and benzothiazine units (4–6). In this review, we have limited the discussion to the eumelanins.

A central tenet in biochemistry is the relationship between structure and function. Given that the last few decades have witnessed significant advances in our ability to determine both the structure and function(s) of many classes of biologic molecules (e.g. proteins and nucleic acids), it is somewhat surprising that both the chemical structures and biologic role(s) of melanin are still subject to debate. To appreciate why this is the case, it is instructive to compare melanins with proteins. Features of these two classes of macromolecules are compared in Table 1 and elaborated upon below.

Table 1.  Comparison of fundamental properties of eumelanins with those of proteins
Molecular building blocksTwenty standard amino acidsDHI, DHICA are precursors to molecules present
Primary structureLinear connectivity via peptide bond, some are cross-linked by disulfide bondsUndetermined, probably involves both linear or cross-linked connectivity
3-Dimensional structuresα-Helix, β-sheet, specific folded structures based on primary sequenceUndefined higher order structure
Biogenesis mechanismWell definedMany enzymes for early steps of melanogenesis are known, but the polymerization mechanism is unknown
Synthetic preparationRecombinant and total synthetic protocols exist to make naturally occurring proteinsProtocols for making synthetic melanin, but they are structurally and chemically different than natural materials
Isolation and purificationEstablished protocols for isolation and purificationNo general methods exists, methods exist for specific tissue sources, many procedures used modify the chemical and physical properties of the pigment

1. The building blocks of proteins are the 20 standard α-amino acids. In the case of eumelanin, the basic molecular building blocks are derived from 5,6-dihydroxyindole (DHI) and 5,6-dihydroxyindole carboxylic acid (DHICA). So unlike proteins, only the precursors to the molecular building blocks of melanins have been elucidated.

2. In proteins, the amino acids are connected to one another linearly through formation of peptide bond linkages. In some proteins, inter- and intra-chain disulfide bonds cross-link the polypeptide chains. In melanins, the details of the connectivity between molecular building blocks are not defined. Both linear and cross-linking connectivity have been proposed based on analysis of the early products of synthetic melanin (7, 8), X-ray scattering (9, 10), and scanning tunneling microscopy (STM) imaging (11).

3. Standard and automated methods have been developed to sequence proteins, but no method exists to determine the molecular linkage(s) of melanins. Chemical procedures can degrade proteins into their constituent amino acids, while no such chemical protocols exist to disassemble melanins into their basic molecular units. However, some chemical degradation procedures offer insights into the relative proportions of DHI and DHICA contents in the melanin pigment (12, 13). While useful and revealing, these procedures offer no information about the structure or connectivity of the molecule(s) present in the original pigment.

4. In proteins, polypeptide chains (commonly called the primary structure) adopt secondary structures (α-helix and β-sheet) and generate folded structures (tertiary structure). The three-dimensional structures of many proteins have been determined from analysis of X-ray diffraction patterns generated from a single crystal of the protein. On the contrary, in melanins, the assembly of oligomeric (or polymeric) molecules into the micrometer scale three-dimensional structures observed in vivo is not understood. A hierarchical aggregation of melanin building blocks into the three-dimensional structure has been proposed (14, 15). In this hypothesis, the fundamental units are planar highly cross-linked molecules comprised 6–8 DHI and/or DHICA units. These molecules are proposed to form layered stacks, 10–12 Å in thickness and 10–12 Å in diameter (9–11, 16). Then further lateral aggregation and stacking forms bulk melanin. X-ray scattering, STM imaging, and atomic force microscopy (AFM) imaging provides some evidence supporting this proposed aggregation structural model. Mass spectrometry studies also support the existence of small oligomers (17–19). Nuclear magnetic resonance (NMR) studies counting the number of ‘visible’ H-atoms (aromatic) in melanin also supports the proposed cross-linking between constituent DHI and/or DHICA molecules (20).

5. Standardized methods are available to synthesize proteins, both in vivo (gene recombinant techniques) and in vitro (chemical solid-state synthesis). The generated protein has the same structure and functions as the natural protein. The synthetic methods for melanins generally involve oxidation (catalyzed by enzymes or auto-oxidation) of a specific molecule (e.g. tyrosine, l-Dopa and DHI) or a small set of accepted precursor molecules. The color and solubility of the synthetic melanin mimics that of the natural melanin, however in terms of chemical composition and physical morphology, synthetic melanins are largely non-representative of natural melanins (21, 22).

6. Standardized procedures for isolation and purification of proteins from cells and tissues have been developed. Proteins are easily soluble in aqueous solution or can be solubilized using surfactants. However, melanin is essentially insoluble in any solvent. The lack of a structural knowledge about melanin at the molecular level is partially because of its insolubility and also the lack of appropriate techniques for its study.

Compared with the enormous progress in proteomics, limited progress has been made toward understanding the structure and function of melanin. Unfortunately, many of the powerful technologies used to study proteins are not amenable to the study of melanins. It is important to be aware of the above-described differences and to develop methods suitable for the study of melanin. A first step is to standardize the isolation and purification of melanin from its native environment.

As we will discuss later in this review, it is necessary to work only with well-characterized samples to reconcile the contradictory results in the current literature. These discrepancies, pertaining to the physical properties and biologic function of melanins, could be the result of many different preparative procedures used to isolate melanins, which do not share common chemical or structural characteristics.

Recent progress in the understanding of the genetic control of melanogenesis (23, 24) and the proteomics of the melanosome represent great strides forward in melanin research (25, 26). Enzymes specifically located in melanosomes catalyze many steps of the melanogenesis process. Many other proteins, non-specific to melanosomes, are also found in the organelle, and their function(s) within remain unclear, but cannot be ignored (26).

In this review article, we primarily focus on recent works towards understanding the structure and photochemistry of eumelanin. This subject matter is organized as follows. First, procedures used to isolate natural eumelanins are examined. Collectively, these studies show that the isolation procedure can modify the chemical structure of eumelanins, and so caution must be exercised in relating studies on isolated eumelanins to those in vivo. Secondly, scanning electron microscopy (SEM) and AFM studies on the surface morphology of eumelanins from human hair and the ink sac of Sepia officinalis are summarized. From a comparison of structural features of these two eumelanins, we conclude that both the natural melanin granules from Sepia and human hair and bovine eye melanosomes are aggregates of 10–30 nm substructures, consistent with the ultra-structures revealed by earlier studies of the corresponding premelanosomes. Thirdly, the effect of pigment aggregation on the aerobic photoreactivity of these eumelanins is reviewed. Fourthly, time-resolved spectroscopic studies examining the initial dynamics of the oligomeric molecules present in eumelanin from S. officinalis are discussed. These data reveal the time-scales on which eumelanin dissipates absorbed UV-radiation, and provide an explanation for the low quantum efficiency for oxygen activation observed for melanins, in general. Finally, opportunities made available by recent technical advances will be discussed as related to furthering our understanding of the structure(s) and function of melanins.

Isolation methods and comparison of the pigments they produce

Melanins are synthesized in melanosomes, a specific organelle of pigment-producing cells. Melanosomes are generated in melanocytes located at the basal layer of skin, inside hair bulbs, and in some tissues of the eye. Melanosomes can also be found in the postmitotic cells of the retinal pigment epithelium (RPE) in eyes (27–29). Melanosomes can be delivered to other cells, as evidenced by the transfer of dermal melanosomes from melanocytes to neighboring keratinocytes. In some species, melanosomes are released into an extracellular compartment, e.g. the pigment-producing epithelial cells lining the ink sac of cuttlefish release the melanosomes into the lumen of the ink sac (30).

Melanin in vivo is always associated with proteins, which were present in the premelanosomes. Some of the proteinaceous material present is likely to be the remnants of the enzymes that catalyzed various steps of melanogenesis (31). In the 1960s, several research groups used transmission electron microscopy (TEM) to examine the ultrastructure of premelanosomes (32–36). These studies revealed filaments and sheets-like structures inside the premelanosomes at early stages of melanogenesis. This material was shown not to be lipid membrane and was proposed to serve to localize and control melanin deposition within the organelle. Evidence has suggested this material is protein (35), but its exact structure and composition is not yet known. To study melanin requires separation and purification of the pigment from its biologic environment. The challenge is to develop procedures that do not modify the pigment during isolation, so the pigment obtained is reflective of its native form. Melanins found in different tissues require different isolation procedures. The strengths and weaknesses of various approaches are described below.

Sepia Officinalis

We first consider melanin from the ink sacs of cuttlefish such as S. officinalis. Melanogenesis in the ink gland of S. officinalis was recently reviewed (30). Sepia eumelanin is considered to be pure eumelanin and is used as a standard for natural eumelanins (37, 38). Sepia eumelanin is isolated through iterative washings of the ink with water and centrifugation. The pellet obtained after extensive washings contains largely spherical granules ∼150 nm in diameter (39). This shape and size is consistent with TEM images of granule production within the Sepia melanosome (40). SEM and AFM images of isolated granules are shown in Figs 1 and 2, respectively. Chemical analysis reveals these granules have a protein content of 6–8% by mass (39, 41, 42). In general, no attempt was made to separate melanin from these internal proteins, because such a separation would destroy the granules.

Figure 1.

In the isolation of Sepia eumelanin from its ink sac, different drying processes result in the formation of different aggregated structures. Under high magnification, all aggregates are composed of spherical granule of diameter ∼150 nm, which is the natural morphology. (A) Spray drying prepared by Mel-Co (Whittier, CA) and Sigma (St. Louis, MO) . Air-drying (B), freeze-drying (C), and CO2-supercritical drying (D) prepared from fresh ink sacs. The scale bars are 20 μm on the low magnification images (left) and 2 μm on the higher magnification images (right). Some of the images are reprinted with permission from Liu and Simon (39).

Figure 2.

AFM images of Sepia granules deposited on mica. (A) Height image, (B) phase image. Scan size is 383 × 383 nm. These images reveal the 10–30 nm substructures that comprise the granule.

This isolation procedure serves to illustrate the first caution in melanin research. Much of what we know about the chemical and physical properties of melanins has been determined on synthetic pigments. Such ‘melanins’ may not contain any protein and certainly do not contain all the proteins present in melanosomes. In contrast, natural pigments contain both melanin (the organic component derived from the initial oxidation of tyrosine) and protein. As described above, the protein component is important in defining the assembly of melanin, and it is reasonable to assume through this mechanism the protein that affects the function of the assembled pigment. Furthermore, in the study of natural pigments reported in the literature, the term ‘melanin’ is used loosely and can represent materials with various protein contents.

Caution should not be only limited to the association of proteins with melanin. A second caution concerns the variety of metal cations present in the natural pigment and absent in the synthetic pigments unless added in the reaction mixture. Even if added to the synthetic systems, the coordination of the metal ions in vivo is likely different from that in vitro. The concentration of these metals also varies among natural melanins (43–45), although it is unclear how the isolation procedure affects these results. It is also possible the metals play a significant functional role in the pigment, which would not be manifested in studies of synthetic systems.

A third caution highlighted in the analysis of isolated Sepia granules is that the drying method employed yields different aggregated structures (Fig. 1). The method chosen affects the physical properties of the pigment, e.g. surface area to mass ratio and porosity of the material (39). The drying process does not affect the chemical composition. However, as will be discussed in the following section, the chemical reactivity of the pigment is dependent on aggregation (46). Therefore, care should be taken when comparing the photobiologic data obtained from samples using different preparation procedures.

Sucrose Gradient Purification

In some cases, pigmented cells can be lysed and intact melanosomes can be isolated from other cellular constituents, e.g. RPE and melanoma cells. Ultracentrifugation on a discontinuous sucrose gradient (47) yields high purity samples where the integrity of the melanosomes is preserved. This is exemplified by the AFM images of bovine RPE melanosomes (Fig. 3) isolated in this manner.

Figure 3.

AFM images of bovine eye melanosomes on mica. The melanosomes were isolated from bovine RPE cell lysate using a discontinuous sucrose gradient. The melanosomes were washed with distilled water prior to imaging. (A) Scan size is 3.08 × 3.08 μm. (B) Higher resolution scan of a region shown in (A); scan size 800 × 800 nm. Left panel is height image, and the right panel is phase image. Similar to Sepia, these images reveal the melanosomes is an aggregated assembly of ∼20 nm substructures.

Melanosomes in Complex Environments

Generally, melanosomes are constrained in a biologically complex environment, e.g. in hair or the brain, and their isolation is often a significant challenge. The separation of melanin from such tissues has largely been achieved by using harsh protocols. These protocols often involve repeated treatments with concentrated acid and base breaking down the surrounding tissue to separate the melanin from proteins and other biologic materials. In many such studies melanin was assumed to be resistant to degradation by the chemicals used in the isolation. As hair is the only readily available non-invasive human sample, pigments isolated from hair have been widely used as model for natural human melanin. In hair, the melanin is tightly held within a keratin matrix. Most studies of hair melanins degrade this protein matrix by exposing the hair to harsh chemical environments such as strong alkali (48), hydrazine/ethanol mixtures (49), or thioglycolic acid/phenol mixtures (50). The acid hydrolytic methods induce the protein to form pigment-like artifacts interfering with the isolation of natural melanin pigments, whereas alkaline hydrolytic methods cause transformation of the chemical composition of melanin pigment and associated protein (51). In particular, it has been reported that exposure to acid results in extensive decarboxylation of eumelanins (22). Electron microscopy imaging demonstrated acid/base treatments modify and disrupt the ultrastructure of melanosomes (51, 52). This necessitated changing the approach for degrading the keratin matrix that would leave the hair melanosome intact and chemically unaltered.

A major advance was made in 1956 when Birbeck et al. used urea, papain and sodium metabisulfide to isolate melanosomes from human hair (53). They found the hair melanin granules isolated by enzymatic digestion were more homogenous than both acid and base extracted melanin. In 1973, Bratosin reported a method for isolation of melanosomes from black mouse hair (54). Keratin was removed from the hair through a two-step process consisting of alkaline hydrolysis and subsequent enzymatic (trypsin) digestion. It was observed that the isolated melanosomes retain their intact morphologic structure. In 1981, Arnaud and Bore used proteinase PSF 2019 to isolate human hair melanin (51). Incubation in either DMF or concentrated LiBr was required to accelerate the subsequent enzymatic digestion. The enzymatic digestion went on for 8 d at pH 9. Afterward, the hair was treated in three acid hydrolysis steps to remove external proteins associated with melanin.

All three of these enzymatic methods were used in conjunction with either alkaline hydrolysis or acid hydrolysis, the drawbacks of which drew attention in the mid 1980s (22, 52). In the 1990s, new procedures for enzymatic isolation were developed, mostly in response to desires to study neuromelanin. In the isolation of neuromelanin from the substantia nigra of human brain, proteinase K (55–57), collagenase (6), and pronase E (58) were used.

In 2000, Prota and co-workers introduced another enzymatic procedure for the isolation of melanins from human hair and iris (19). Using proteinase K, papain and protease type XIV consecutively and a detergent Triton-X-100, this method is truly proteolytic and without any harsh chemical reagents. These workers found the chemical and spectral characterization of the pigments from human hair and iris revealed marked structural differences, which challenged the common belief that all melanins are similar independent of their origin. They further suggested this could be due to the diversity of functions and sites of biosynthesis and storage. This conclusion is based on the assumption that the enzymatic isolation procedure produces a pigment representative of melanin in its native environment.

In a recent study (43), we examined eumelanin from a sample of black, Indonesian hair using three different published procedures: two acid/base extractions (59, 60) and an enzymatic extraction (19). The morphology and spectroscopic properties of the isolated pigments differ significantly. The acid/base procedures both yield an amorphous material, whereas enzymatic extraction yields ellipsoidal melanosomes (Fig. 4). Amino acid analysis shows there is still a significant amount of protein associated with the isolated pigments, counting for 52, 40 and 14% of the total mass for the two acid/base extractions and the enzymatic extraction, respectively. Neither of the amino acid compositions correlate with keratin or tyrosinase. Metal elemental analysis by inductively coupled plasma mass spectrometry shows that the acid/base extraction removes the majority of many metal ions bound to the pigment. Chemical degradation analysis by KMnO4 and H2O2/OH indicates significant differences between the pigments by the acid/base extractions and the enzymatic extraction. After correction of the protein mass in the two pigments, the factor of 2–3 lower yields of both pyrrole 2,3,5-tricarboxylic acid (PTCA) and pyrrole 3,5-dicarboxylic acid (PDCA) indicates 45–65% of acid/base-extracted melanin has been chemically modified, consistent with the result of the Soluene solubilization assay. While the optical absorption spectra of the bulk pigments are similar, the spectra of the molecular weight <1000 amu fractions differ significantly. The data clearly indicate pigment obtained from human hair by acid/base extraction does not effectively separate the pigment from amino acids, and the conditions lead to the destruction of the melanosome and alter the molecular structure of melanin. The acid/base extracted hair melanin is not representative of the natural material and is a poor model system for studying the physical and biologic properties of melanins. The enzymatic extraction preserves the integrity of the melanosome, removes most of the external proteins, and therefore should be the preferred choice for isolation of melanin from hair samples. At present, the procedure reported by Prota and co-workers (19) is the best approach for isolating melanosomes from hair and iris samples.

Figure 4.

SEM images of hair eumelanin obtained using different extraction methods. The detailed procedures can be found in (43). (A) and (B) show the results from two different acid/base procedures. (C) shows the results from an enzymatic method. Only the enzymatic method preserves the natural morphology of the melanosome, and is clearly the method of choice for isolating hair eumelanin for physical and photobiological studies. All images are reprinted with permission from Liu et al. (43).

Morphology of melanin assemblies as determined by modern imaging techniques

A variety of mass spectrometry and imaging studies on synthetic eumelanins argue that the pigment is an aggregated system of small oligomeric (∼4–6 monomers) molecules. Supporting mass spectrometry data have been reported for synthetic (61–63) and natural melanins, such as isolated from Sepia (64), human hair (65) and iris (19). Solubilization of natural pigment by detergent solution suggested the dominant forces holding the melanin granules together are not covalent in nature, but rather arise from hydrophobic interactions (54). However, to date the relative contributions of Van der Waal's forces and hydrogen bonding are not clear.

AFM is an imaging technique that provides three-dimensional topographical information. The AFM tip is located at the end of an oscillating cantilever. In tapping mode, this tip is scanned in a two dimensional raster pattern over the sample, and the height is adjusted to maintain the amplitude of the oscillation. In general, the height and lateral resolution of the image are 0.1 and ≤10 nm (depending of the dimensions of the scanning tip), respectively. Phase images can be collected simultaneously with the height image. Phase data measure the phase shift in the cantilever oscillation from that of the driving force. This results from attractive and repulsive interactions between the AFM tip and the sample. The phase signal can be related to the stiffness of the sample and is useful for revealing domains that may otherwise be hidden, overlooked, or difficult to see in the height image.

In a recent AFM study on mass-selected fractions of Sepia eumelanin, many of the observations reported were consistent with the conclusion that the pigment is derived from small oligomeric species (15). First, in experiments utilizing the AFM tip to cut across granules, images revealed that the tip could be used to cut ∼30 nm deep and ∼30 nm wide troughs through a granule. Accompanying the cut, a ∼20-nm ridge is generated, revealing the material displaced by the act of cutting (Fig. 5). These data indicate that the tip displaces constituents that are small compared with its dimensions. Secondly, to gain more information on the dimensions of these molecular components, images were collected on a dried 1000 < MW < 3000 sample. For that sample, significant regions of the mica are covered with a filament structure (Fig. 6). The filaments are 3–6 nm in height, 15–50 nm in width, and often several microns in length. They are easily cut along the 15–50 nm axis using the AFM tip. The data support the hypothesis that these are stacked structures, not extended covalently bound structures. This result is consistent with the mass spectrometry and X-ray scattering studies suggesting the building blocks of the pigments are small oligomeric molecules. These studies also clearly reveal the granules are aggregated structures. When we consider the 3000 < MW < 10000 fraction, height images of this sample dried on mica reveal fractal-like growth patterns (Fig. 7). The fractal aggregation is likely a result of drying a hydrophobic material at low concentration on a hydrophilic surface. The growth involves the re-aggregation of small structures, but we do not observe any evidence for the reassembly of the material into granules. This suggests the granules cannot self-assemble from the constituent oligomers under the conditions used in vitro.

Figure 5.

(A) Tapping mode AFM height images of a Sepia melanin aggregate before (left) and after (right) the AFM tip was used to cut across the aggregate along the direction indicated by the white arrow. The scale bar is 210 nm. (B) Cross-sections along the solid line (perpendicular to the cut) shown in (A) before and after cutting the aggregate. The images are reprinted with permission from Clancy and Simon (15). These images show that the act of cutting displaces molecular constituents that are small compared with the dimensions of the AFM tip.

Figure 6.

Tapping mode AFM phase image of eumelanin filaments formed from the 1000 < MW < 3000 Sepia eumelanin fraction upon drying on mica. The scale bar is 250 nm. The filaments are 3–6 nm in height, 15–50 nm in width, and often several microns in length. This further supports that natural eumelanin structures are not large cross-linked polymers but are aggregates of molecular oligomers, whose structure(s) are yet to be determined. The image is reprinted with permission from Clancy and Simon (15).

Figure 7.

AFM image of mass fraction of 3000 < MW < 10000 of Sigma Sepia melanin deposited on mica. (A) Height image, scan size is 19.6 × 19.6 μm. (B) Height image, scan area is 7.7 × 7.7 μm. (C) Height image, scan size is 2.65 × 2.65 μm. (D) Phase image, scan size is 310 × 310 nm. A series of images of different scan scales are shown to illustrate the fractal pattern, which likely results from drying a hydrophobic material (eumelanin) at low concentration on a hydrophilic surface (mica). The deposit reveals reaggregation of small structures, but the reassembly of this material into its natural morphology (granules) has not been achieved in vitro. The images are reprinted with permission from Liu and Simon (39).

While the complete details of the assembly of the granules in vivo remains unknown, additional insight into the substructure can be obtained from high-resolution SEM and AFM studies (39). Fig. 8A shows an ultra-high resolution (UHR)–SEM image of granules deposited on mica. Two important features are manifested in this image. First, the surface of the granules appears to exhibit some substructures, with constituents being ∼10–20 nm in lateral dimension. Secondly, in regions where granules are in contact, the shape of a granule changes from spherical to hexagonal. This not only represents optimal two-dimensional packing, but also clearly indicates that the granules have a degree of plasticity and can be somewhat deformed without disintegration. The preparation of the sample imaged in Fig. 8A involves coating the materials with an Au/Pd mist. This could give rise to the deposition of Au/Pd colloidal particles on the surface, which could mistakenly be interpreted as substructures. To address this possibility UHR–SEM images of uncoated melanin, deposited on conductive substrate, such as silicon wafer, were examined. The quality of the images is dramatically reduced but one can still discern the roughness of the surfaces (Fig. 8B). Pigment granules without any coating can also be imaged by tapping mode AFM, the variation of the height of the cantilever across the scanning area (height image) gives real three-dimensional topology information and the variation of the phase of the cantilever vibration across the scanning area (phase image) can provide information concerning if boundaries exist between regions of differing hardness. Thus, verification of substructure revealed by the SEM images of coated samples is possible by examining AFM images of uncoated granules, as shown in Fig. 2A,B. The images clearly show similar substructure for the granules. These results suggest that the granules are aggregates of ∼10 nm constituents. UHR–SEM images of sonicated Sepia samples subjected to five cycles of sonication and centrifugation also support the aggregation model (Fig. 9). Thus sonication in a water bath introduces mechanical perturbation strong enough to remove the ∼10 nm constituents from the surface of the granules or cause a partial decomposition of the granules.

Figure 8.

(A) UHR–SEM images of granules from the Sepia ink sac deposited on mica. The sample has been coated with colloidal Au/Pd. (B) UHR–SEM image of uncoated granules deposited on silicon wafer. Comparison of the images of the coated and uncoated samples show the surface substructure is inherent to the melanin and not the result of the protocol used to collect the images. The images are reprinted with permission from Liu and Simon (39).

Figure 9.

UHR–SEM images of supernatant after fifth round sonication and centrifugation, air-dried on mica. The images clearly show sonication can result in deformation of the granules. In addition, a thin layer of material coats some of the granules, which is not melanin, but a remnant of other cellular debris present in the ink sac. This material can be remove from the melanin following more extensive washing. The images are reprinted with permission from Liu and Simon (39).

It is interesting to compare the AFM images of melanin from the Sepia ink sac (Fig. 2) with that of the human hair melanosome isolated enzymatically (Fig. 10). Although the two natural melanin granules have different shapes, spherical vs. ellipsoidal, and different dimensions, roughly 150 nm vs. 450 nm × 1 μm, the surface features look the same. The surfaces of both granules are clearly not smooth but with substructures in the size range of a few nanometer to 30 nm. The similarity between human hair melanosomes with bovine eye melanosome is more striking (Fig. 3). Although they are also of different size (bovine eye melanosomes are ∼900 nm × 2 μm), the shape is similar and the surface features are essentially indistinguishable. Amino acid analysis on the natural melanin indicates there is a significant amount of protein associated with melanin (8% in sepia and 14% in hair by mass). Based on mass, if we approximate the eumelanin oligomer(s) as 1000 amu, then a typical melanosome represents an assembly of more than a million such building blocks. The well-defined structure revealed by imaging experiments strongly suggests that the assembly of the fully pigmented melanosome is a tightly controlled biologic process. Thus, not only are the primary and early oxidation steps of melanogenesis under enzymatic control, but also the assembly of the deposited pigment within the melanosomes must also be controlled, possibly by distribution and localization of the enzymes in the internal membrane structure of the premelanosomes (66, 67).

Figure 10.

AFM images of a single human hair melanosome. The scan size is 1 × 1 μm. Left panel is height image, and the right panel is phase image. Similar to Sepia and bovine eye melanosomes (Figs 2 and 3) these images reveal the hair melanosomes is an aggregated assembly of ∼20 nm substructures.

Aggregation affects photogeneration of reactive oxygen species (ROS)

Photochemical excitation of melanin results in the activation of oxygen, the most important primary photochemical pathway being the formation of the superoxide anion, Oinline image. In an effort to quantify the activation of molecular oxygen by eumelanin, Sarna and co-workers determined the action spectrum for the photoconsumption of oxygen by synthetic and natural eumelanins (68). While eumelanin exhibit absorption throughout the visible and ultraviolet region, oxygen photoconsumption occurs only for wavelengths shorter than 400 nm. In their original report Sarna and co-workers suggested that the chromophore responsible for oxygen photoconsumption differed from that which dominated the absorption spectrum. One explanation would be that eumelanin is composed of a number of molecular entities and one particular constituent has an absorption spectrum matching the action spectrum. Given the above discussion of the structural morphology of the pigment, however, it is reasonable to propose that different sized aggregates have different spectroscopic and photoreactive properties. In this case, a specific arrangement of the oligomeric building blocks (individual or aggregated) would be responsible for the photoconsumption of oxygen.

Different mass-selected fractions of eumelanin solutions were obtained by ultrafiltration, the optical properties of which show that the absorption for the MW < 1000 fraction of eumelanin from both Sepia (69) and black human hair (70) match the action spectrum for photoconsumption of oxygen (Fig. 11). Samples of larger masses (e.g. 1000 < MW < 3000, MW > 10 000) exhibit increased absorption at longer wavelengths, and whether these fractions are aggregates of the oligomers present in the MW < 1000 solution or longer oligomers remains to be determined. Photoacoustic calorimetry revealed the MW < 1000 sample is the only mass fraction showing measurable energy storage following UV-A excitation (71). These data suggest unaggregated oligomers underlie the phototoxic effects of melanin and aggregation mitigates such processes.

Figure 11.

The optical spectra for MW < 1000 fractions of (- - -) hair eumelanin and (…) Sepia eumelanin with the 270-nm absorption feature removed are compared with (-) the action spectrum for the free-radical photogeneration by eumelanin. Quantitative agreement is observed. These data suggests the unaggregated oligomers make the dominant contribution to the photoconsumption of oxygen by eumelanins. The similarity observed among Sepia and human hair suggests that they contain similar if not identical molecular species. The action spectrum is reproduced with permission from Sarna and Sealy (68).

It is important to ask how and if aggregation affects the mechanism and yield of ROS photoproduction by eumelanin. To address this issue, the reduction of cytochrome c (cyt c) by eumelanin excited by UV-B (302 nm) was examined. The following conclusions could be drawn from that study (46).

First, while the initial reduction rate of cyt c does not change with aggregation, the optical density of the solution does. As a result the apparent quantum yield for Oinline image formation decreases with aggregation.

Secondly, studies with biochemical quenchers reveal a quantum efficiency of hydrogen peroxide, H2O2, production by the MW < 1000 fraction of 5.7 × 10−3. The formation of H2O2 is attributed to reaction between Oinline image and semihydroquinone groups on the oligomeric molecules. The quantum yield becomes immeasurable upon aggregation of the oligomers and the mechanisms of this quenching remains to be established. In any event, aggregation clearly reduces the production of this highly toxic oxidant.

The aggregation-dependent generation of ROS by eumelanin presents a framework for understanding contrasting photoprotective and phototoxic roles exhibited by eumelanin. The equivalence between the action spectrum and absorption spectrum along with the aggregation-dependent quantum efficiencies for Oinline image and H2O2 implicate the oligomers as the phototoxic component. Thus, any changes in melanin that lead to a disruption of the aggregated structure could result in increased oxidative stress. The fact that oligomers generate more H2O2 than aggregated pigment may have significant biologic ramifications. H2O2 can react with a variety of cellular components, causing, for example, lipid peroxidation of membrane and hydroxylation of proteins and DNA. Such processes may be important in skin keratinocytes, where melanosomes are partially degraded (72), or in retinal pigment epithelium cells where the structural features of melanosomes are found to change with age (73, 74).

Probing non-radiative relaxation of eumelanins by time-resolved optical spectroscopy

The low emission quantum yield (10−5–10−3 depending on the aggregation) by photoexcited eumelanin suggests the pigment efficiently releases the absorbed light energy through non-radiative means (75). Furthermore, the low quantum yield for oxygen activation (10−4–10−3 depending on the aggregation) suggests that the dominant process is likely rapid relaxation to the ground electronic state of the pigment (46). Such a conclusion is supported by photoacoustic studies of melanins (71), but the time resolution of this technique only reveals that non-radiative relaxation occurs on the sub-nanosecond time scale.

The initial photophysical and photochemical events following UV excitation of melanins can be quantified by using time-resolved optical spectroscopy. Both time-resolved absorption and fluorescence experiments have been performed on melanins. Time resolved absorption studies suggest rapid repopulation of the ground state following UV-A and UV-B excitation (T. Ye and J.D. Simon, unpublished data). The recovery dynamics are complex and three time constants are revealed (Table 2). Despite the complexity of the decay dynamics, it is clear that melanin efficiently and rapidly converts absorbed photon energy into heat. The rapidity of this process mitigates the potential of adverse photochemistry and supports the hypothesis that melanin exhibits, in part, a role in photoprotection.

Table 2.  Time constants and amplitudes (in parentheses) from time-resolved experiments on eumelanin
  1. *The amplitudes of these decay constants vary with probe wavelength.

Transient absorption*0.56 ± 0.8 ps3.2 ± 0.5 ps31 ± 5 ps 
Emission decay58 ± 7 ps (0.54)0.51 ± 0.07 ns (0.22)2.9 ± 0.5 ns (0.16)7 ± 1 ns (0.08)

The decay associated spectrum for each time constant could provide information of the spectral regions associated with a particular decay component. In this manner, it may be possible to decompose the transient spectrum into contributions for different molecular entities, each of which exhibits a particular decay time constant. Unfortunately, in the case of eumelanin, such an analysis does not cleanly assign a specific absorption feature in the transient spectrum to a given time constant. The resolution afforded by the global data analysis suggests that all spectral features present in the transient spectrum contribute to the multi-exponential decay. It is likely that these spectra are derived from a small set of structurally similar chromophores within the eumelanin pigment. This would explain the non-exponentiality of the decays, the similar vibronic structure observed in the transient spectrum, and the differences in the relative intensities of the vibronic bands. Given that the connectivity of the monomer units in the structure of eumelanin is not known, it is currently not possible to speculate on the molecular details that give rise to these spectral differences.

It is interesting to compare these dynamics with those obtained from fluorescence lifetime measurements. The emission dynamics of melanin are non-exponential and require a sum of exponentials to generate functional forms that provide fits to experimental data. In a previous report on Sepia eumelanin (76), four time constants were needed to describe the emission collected at 520 nm (near the maximum of the spectrum) following excitation at 355 nm (see Table 2). It is interesting to note that the fastest lifetime component of the emission is in rough agreement with the slowest time constant revealed by the ultrafast absorption measurements, and thus it is possible that both experimental techniques ‘sense’ the decay of the same molecular species, but the time resolution of the emission experiments is not sufficient to observe the faster decay component.

One must exercise caution in interpreting the transient spectroscopic results of melanins in terms of the photobiologic properties of the pigments. While interesting dynamics are revealed, it remains to be established whether either transient absorption or emission experiments probe the molecules responsible for the photoaerobic reactivity of melanins. To address such issues, it will be important to measure the transient optical properties and the action spectra for photoaerobic processes (oxygen photoconsumption, superoxide formation) on the same set of samples. Comparison between existing published literatures is potentially problematic resulting from the range of methods used to isolate and prepare melanins and the effects these procedures have on the integrity and molecular structure of the pigment.

Future directions

This is an exciting time in melanin research. The last decade has witnessed the development of a diverse set of powerful analytical tools, many of which can be applied to characterize natural pigments. Advances in biochemical techniques are now enabling the isolation of fully assembled melanosomes, and so one can start asking detailed biophysical questions concerning the nature of the assembly. We are optimistic that significant advances will be forthcoming over the next several years. The following topics are presented in a hope to excite and inspire those either working in the field or interested in contributing to biophysical studies of melanins.

Molecular Structure(s) of the Oligomers

While the early steps of melanogenesis are understood (77–80), the molecular structures of the resulting oligomers remain elusive. Mass spectral studies have provided some insights into the molecular weights of constituent oligomers, and oligomers of DHI and/or DHICA with a linear connectivity have been proposed (19, 62–64). Recent NMR studies of melanins suggest cross-linked connectivity between monomers (20). Highly cross-linked planar oligomeric structures have been suggested by X-ray scattering (10, 81) and STM experiments (11, 82). At this point of time, it is unclear whether melanogenesis produces a heterogeneous set of oligomers of appreciable relative concentrations or a single molecular species dominates. Recent advances in high-performance liquid chromatography (HPLC)–MS/MS techniques suggest that this analytical tool could play a powerful role in elucidating the structures of oligomers. While standard MS techniques provide the molecular weight, MS/MS and higher order fragmentation can provide information on the detailed structure. In addition, the recent commercialization of HPLC–NMR may also prove to be a valuable technique for structural determination.

The molecular structure(s) of melanin is one area where theoretical insights currently outpace experimental achievements. Galvao and Caldas were the first to carry out theoretical calculations on model polymers for eumelanins in late 1980s (83, 84). In the last 2 yr, several groups used modern computational chemistry tools to examine the structure and spectroscopy of monomers (85), dimers (86, 87), and oligomers of eumelanin (88, 89). While these papers are of interest in their own right, there is no evidence that the molecular structures explored are actually present in melanin. Experiments are surely needed in this area. With knowledge of naturally occurring structures, the computational tools could be used to provide a deep understanding of the spectroscopy and association of these molecules in the aggregated pigment.


This article has focused on eumelanin. In nature there is a second pigment, pheomelanin, which is yellow-reddish in color. The biogenesis of pheomelanin differs from eumelanin in that cysteine is incorporated into the structure and the building blocks are derived from cysteinyl-dopa. Epidemiological data indicate individuals with fair skin are more susceptible to skin cancers than their darker skin counterparts (3). This observation is commonly associated with the hypothesis that pheomelanin exhibits a greater phototoxicity than eumelanin. In support of this hypothesis, Prota and co-workers explored whether there was a relationship between hair melanin composition and minimal erythemal dose (MED) in a group of red-haired individuals (90). They found a correlation between the eumelanin/pheomelanin ratio and the MED values, suggesting UV sensitivity is associated with high pheomelanin and low eumelanin levels and that the eumelanin/pheomelanin ratio may be a chemical parameter for predicting individuals at high risk for skin cancer and melanoma. Cellular studies support this general concept; UV-A-induced DNA single-strand breaks in human melanocytes differing only in the amount of pigment produced showed photosensitization by intrinsic chromophores, most likely pheomelanin and/or melanin intermediates (91). These results indicate the need to understand the molecular composition, the structure, and the photobiology of pheomelanin. To date, most of our limited knowledge on this pigment comes from the study of synthetic samples. Careful studies on pheomelanin from natural systems are needed, as are quantitative comparisons between such a sample and a related natural eumelanin.

The Role of Metals

Melanin is able to sequester heavy metals ions, such as iron, copper, zinc and lead, in principle protecting the surrounding tissue from their cytotoxicity. However, it is also proposed that when the binding sites of a melanosome become saturated, the integrity of melanosome could be compromised, and heavy metal ions and/or melanin oligomers could be released from melanin and trigger acute damage to the cell (92). Along these lines, a link between Fe(III) binding to neuromelanin and death of pigmented neurons and pathogenesis of Parkinson's disease has been proposed (93, 94).

To what extent melanin can bind metals like Fe(III) and how this binding affects the structure and function of the pigment represent a set of interesting questions, which have not received careful attention. There is clearly an opportunity to carry out significant work in this area, especially if the work can be carried out with well-characterized samples of neuromelanin.

Molecular Aspects of the Aging Melanosome

Sarna and co-workers examined the photochemically induced uptake of oxygen by different age cohorts of human RPE melanosomes (95). These data clearly show the increase of the activation of oxygen by melanosomes with increasing age (Fig. 12). The emission intensity of RPE melanosomes also increases with age (96). It is interesting to note that Schraermeyer and co-workers recently demonstrated that melanin fluorescence from synthetic melanin and melanin isolated from bovine melanosomes increases after oxidation (97). One could therefore hypothesize that the age-dependent aerobic photoreactivity of RPE melanosomes reflect concomitant changes in the molecular structure of the melanin (e.g. oxidative damage) and/or increased concentrations of bound redox-active metal cations. Because of the potential link between changes in the photobiology of retinal melanosomes and cell atrophy of the RPE layer, a molecular understanding of the origin of these effects would be important in the prevention of diseases in this tissue.

Figure 12.

Broadband, blue-light-induced oxygen uptake in suspensions of melanosomes isolated from donors in the following ages: <40 yr (□), 42–60 yr (▵), 61–80 yr (○), >80 yr (◊). Oxygen uptake in the dark was negligible in all samples studied. The concentration of pigment granules was adjusted to 4 × 109 granules/ml. Oxygen uptake was measured with ESR oximetry. The data clearly show the aerobic photoreactivity of melanosomes increases with age. The images are reprinted with permission from Rozanowska et al. (95).


Acknowledgements– This work was partially supported by the National Institute of General Medical Sciences. We thank Unilever Research US for continued support of this work. We also thank the following co-workers for their contributions to the work described herein: Dr Susan E. Forest, Dr Chris M.R. Clancy, Dr J. Brian Nofsinger, Dr Tong Ye, Valerie R. Kempf, Prof. Shosuke Ito, Prof. Kazumasa Wakamatsu, Dr Mark Rudnicki, Emily E. Weinert, Leslie Eibest, Prof. Yuri Il'ichev, and Prof. T. Sarna.