Evolution of hard proteins in the sauropsid integument in relation to the cornification of skin derivatives in amniotes


Dr L. Alibardi, Dipartimento di Biologia evoluzionistica sperimentale, University of Bologna, Italy. E: alibardi@biblio.cib.unibo.it


Hard skin appendages in amniotes comprise scales, feathers and hairs. The cell organization of these appendages probably derived from the localization of specialized areas of dermal–epidermal interaction in the integument. The horny scales and the other derivatives were formed from large areas of dermal–epidermal interaction. The evolution of these skin appendages was characterized by the production of specific coiled-coil keratins and associated proteins in the inter-filament matrix. Unlike mammalian keratin-associated proteins, those of sauropsids contain a double beta-folded sequence of about 20 amino acids, known as the core-box. The core-box shows 60%–95% sequence identity with known reptilian and avian proteins. The core-box determines the polymerization of these proteins into filaments indicated as beta-keratin filaments. The nucleotide and derived amino acid sequences for these sauropsid keratin-associated proteins are presented in conjunction with a hypothesis about their evolution in reptiles-birds compared to mammalian keratin-associated proteins. It is suggested that genes coding for ancestral glycine-serine-rich sequences of alpha-keratins produced a new class of small matrix proteins. In sauropsids, matrix proteins may have originated after mutation and enrichment in proline, probably in a central region of the ancestral protein. This mutation gave rise to the core-box, and other regions of the original protein evolved differently in the various reptilians orders. In lepidosaurians, two main groups, the high glycine proline and the high cysteine proline proteins, were formed. In archosaurians and chelonians two main groups later diversified into the high glycine proline tyrosine, non-feather proteins, and into the glycine-tyrosine-poor group of feather proteins, which evolved in birds. The latter proteins were particularly suited for making the elongated barb/barbule cells of feathers. In therapsids-mammals, mutations of the ancestral proteins formed the high glycine-tyrosine or the high cysteine proteins but no core-box was produced in the matrix proteins of the hard corneous material of mammalian derivatives.

Scales of extant reptiles in comparison with feathers and hairs

Scales in extant reptiles are specialized structures

Most living reptiles are covered by scales of different size, thickness, variety and color, and these characterize different species (Maderson, 1985; Landmann, 1986; Maderson et al. 1998; Alibardi, 2003, 2006c) (Fig. 1A–D). Details on the morphology and cytology of reptilian integument have been reported in the above references, and will not be dealt with in the present review. The main goal of this review is to report the latest data on the present state-of-the-art in the study of the characteristics of keratinization in reptilian epidermis. The present account represents a further step forward from recently published reviews on this rapidly growing topic (Alibardi & Toni, 2006a; Toni et al. 2007b).

Figure 1.

Macroscopic aspect of reptilian scales (A–D) and histology of the epidermis in scales of different reptiles (E–M). (A) Overlapped trunk scale of snake (Natrix natrix). Bar, 0.5 mm. (B) Little overlapping scale of ventral region of midtrunk region of the tuatara (Sphenodon punctatus). Bar, 0.5 mm. (C) Large dorsal scales (arrow) and lateral small scales (arrowhead) of the alligator (Alligator mississippiensis). Bar, 1 mm. (D) Plastron scale (gular, arrowhead) and the darker neck epidermis (arrow) of the turtle Emydura macquarii. The latter shows a rough surface with a scale-like pattern, represented by a rough skin. Bar, 1 mm. (E) Epidermis of the snake N. natrix showing differentiating fusiform beta-cells beneath the shedding line (arrow). Bar, 15 µm. (F) Normal epidermis (resting) of the lizard Podarcis sicula showing the outer beta-layer and the complete alpha-layer. Bar, 15 µm. (G) Epidermis of S. punctatus in resting phase showing the outer beta-layer and inner alpha layer (arrows). Bar, 15 µm. (H) Stratified wound epidermis of S. punctatus showing a thick corneous alpha-layer. Bar, 10 µm. (I) Epidermis of ventral scale in saltwater crocodile (Crocodylus porosus) showing flat cells in the transitional layer before the pale, lower part of the corneous layer. Bar, 15 µm. (J) Tail scale of the turtle Chrysemys picta showing thymidine-labeled nuclei (arrows) 1 day after injection and autoradiography. The arrowhead indicates the transitional layer of the epidermis in the outer scale surface. Bar, 20 µm. (K) Tip of a plastron scute of C. picta showing the shedding line (arrows) along which the outer part of the corneous layer will detach. Bar, 20 µm. (L) Tip of plastron scute of C. picta 3 days post-injection of tritiated histidine. Most autoradiographic labeling (arrows) is present in cells of the intermediate and transitional layer beneath the thick corneous layer. Bar, 20 µm. (M) Autoradiographic detail of the intense silver grains localized in the transitional layer of the carapace 1 day after injection of tritiated histidine in C. picta. Bar, 10 µm. a, alpha-layer; ba, basal layer; be, beta-layer; c, corneous layer; ci, inner corneous layer; co, outer corneous layer; de, dermis; e, epidermis; h, hinge region; in, intermediate layer beneath the wound epidermis; pc, pre-corneous or transitional layer; sb, suprabasal layer; t, tip of scute; tr, transitional layer; w, wound (regenerating) epidermis. Dashes underline the basal layer of the epidermis.

In general terms, from the basic anapsid cotylosaurs a new form of corneous proteins, beta-keratins, was produced from ancestral proteins. The new proteins mainly contributed to the formation of a hard and mechanically resistant corneous layer. The production of these proteins increased the strength of the epidermis but a thick and hard beta-keratinized layer impeded skin elasticity, pliability, and sensitivity in the derived reptiles.

Beta-keratins are produced in the epidermis of most turtles and crocodilians, and in feathers, scales, claw and beak of birds, whereas they tend to disappear in some layers of the epidermis of lizards and snakes, where mainly soft alpha-keratin remains. The evolution of modern reptilian scales from primitive reptiles with an extensive dermaskeleton took place by a specialization of the cornification process. Microscopy of scales in different species of reptiles is presented in Fig. 1(E–M), and gives a direct view of their histological structure and layer stratification. Recent immunological and molecular studies suggest that the reduction in the production of these proteins may be associated with the evolution in lepidosaurian reptiles of a mechanism of shedding, which permits the cyclical loss of the outer part of the corneous layer (Alibardi, 2006c; Alibardi & Toni, 2006a). Lepidosaurians show a multilayered epidermis whose external portion is cyclically shed (Maderson, 1985; Maderson et al. 1998).

Recent immunocytochemical, biochemical and molecular biology studies have suggested that the cytological differences among the six main layers of lepidosaurian epidermis (oberhautchen, beta-, mesos-, alpha-, lacunar and clear-layers; some layers are shown in Fig. 1E,F) are derived from the genetic control of the production of their hard proteins during the shedding cycle (Alibardi & Toni, 2006a–d; Alibardi et al. 2007; Toni et al. 2007a,b). The molecular control of the activation or repression of genes for the synthesis of these proteins still remains to be discovered.

In the tuatara (Sphenodon punctatus), a primitive lepidosaurian, shedding of the outermost generation occurs through the formation of an intermediate region where cells possess mixed alpha and beta characteristics, a precursor of the more specialized shedding layer of lizards and snakes (Fig. 1G–H). The loss of the outer corneous layer occurs in this area, under which a new beta-layer is formed. Finally, in squamates the shedding mechanism reaches perfection: a single layer, the oberhautchen layer, is adjacent to the last alpha-layer of the previous epidermal generation, the clear layer. These two layers form a specialized shedding complex which separates along a shedding line (Fig. 1E). The synchronization of cell differentiation of the epidermis over discrete areas of the skin surface in lizards, or over the entire surface of the skin in snakes, produces a patchy molting in lizards and a single molt in snakes. After shedding, the epidermis is covered by a mature beta-layer and a more or less mature alpha-layer (Fig. 1F).

In crocodilians and turtles a shedding mechanism in the epidermis is absent or limited, and epidermal histology and the control on the production of hard-proteins appear simpler than in lepidosaurians. Crocodilian epidermis comprises a multilayered epithelium with a transitional layer of flat cells that are incorporated into a variably thick corneous layer (depending on the type of scales) (Fig. 1I). In archosaurians (crocodilians and birds), the process of shedding of epidermal layers is not specifically known. However, the presence of a scission layer in avian scales (Cane & Spearman, 1967), made by thinner and lipid-rich corneocytes, suggests that archosaurians may also use this mechanism of a switch in cell differentiation to create a scission layer. A different type of sloughing layer is also formed in regenerating feathers and underneath the sheath before it is lost to reveal the barb ramification.

In turtles, hard scutes are present in the plastron and softer scutes or simple folds in the remaining body regions, with a corneous layer made of soft-keratin (Fig. 1J). The cell proliferation in the basal layer and the intense protein synthesis of transitional cells near the hinge regions of the growing scutes layer allow the growth of scutes (Fig. 1J–M).

In the scutes of the carapace and plastron of some species of turtles (chelonians), the switch from thicker beta- into thinner alpha-keratin cells determines the formation of a shedding layer. The latter permits a slow intra-corneous detachment of the more external layers with the sloughing of the superficial part from the more internal part of the corneous layer (Fig. 1K). A new beta-layer allows for the next stratification of the stratum corneum; this process occurs in shedding turtle but not in non-shedding turtle or in the tortoises where the superficial part of the corneous layer is gradually lost by natural wear.

Scale development and epidermal layers in reptiles and birds

Scales in reptiles derive from the interaction between a likely inductive mesenchyme and a large surface of the epidermis (Fig. 2). Extensive reviews of studies on the embryology in reptilian skin have been presented in previous accounts, and the reader can find further information in these publications (Maderson, 1985; Alibardi, 2003, 2004b, 2005). It has been hypothesized that areas of dermal–epidermal interactions (ADEIs) are relatively large in reptiles, including pre-avian archosaurians (Fig. 2). These interactions are directly or indirectly responsible for the activation of several genes in the epidermis, including those coding for the production of beta-keratins. Beneath an embryonic epidermis made of two to six layers, depending on the reptilian species, the accumulation of beta-keratins determines sloughing at hatching (Fig. 2A-A7). Among reptiles the crocodilians are particularly interesting for their phylogenetic affinity with birds, and the analysis of their embryonic epidermis reveals the highest sequence identity with that of birds. In the latter, considered here to be a lineage of archosaurian reptiles, a special morphogenetic process has evolved for the production of feathers from embryonic skin (Maderson, 1970; Brush, 1993; Sawyer et al. 2005b; Alibardi et al. 2006b). The process has determined the selection of small proteins the size of feather-keratins, which have been utilized for the formation of barbs, as shown in Fig. 2.

Figure 2.

Drawing illustrating the embryonic layers formed on the scale of embryonic crocodilians (A) in comparison with those of bird embryos (B). The arrowed square illustrates the sequence of layers in crocodilian scale (A1) vs. avian beak (C), downfeather (D), claw (E), and scutate scales (F). The same colors indicate homologous layers, including those of downfeathers after the cell displacement within the barb ridge (see details in the text). The arrows pointing down in downfeathers indicate the direction of formation of barb ridges. Whereas feather-keratin form oriented and parallel bundles (arrow to D6), the other beta-keratins form an irregular orientation (arrows to A7, E3, F3. a, alpha (intermediate) layer; ADEI, areas of dermal–epidermal interaction; (B) Definitive (post-hatching and adult) beta-keratin layer containing different beta-keratins in scales, claws, beak and feather (colored with different colors); BA, barb (ramus); BL, barbule cells (equivalent to subperiderm cells); BR, barb ridges; BBK, adult beak beta-keratin with specific epitopes in black; CBK, adult claw beta-keratin with specific epitopes in black; CCBK, crocodilian claw beta-keratin; CSBK, crocodilian scale beta-keratin; CY, cylindrical cells; FBK, feather beta-keratin; FF, forming follicle; G, germinal layer; GFG, growing feather germ; HIS, tritiated histidine autoradiography; IP, inner periderm; OP, outer periderm; RB/PG, reticulate or periderm granules of the periderm; S, sheath (supportive cells corresponding to external layers derived from inner periderm cells); SP, subperiderm (containing the feather-keratin epitope, FBK); SBK, adult scale beta-keratin with specific epitopes in black; T, transitional layer along which the embryonic epidermis is shed around hatching (A3). THY, tritiated thymidine autoradiography; V, barb vane ridge cells of the axial plate of barb ridges (supportive cells equivalent to inner layers derived from inner periderm cells).

Recent studies on the embryonic epidermis of alligator and birds have produced further indications that feathers may derive from a generalized archosaurian epidermis (Sawyer & Knapp, 2003; Sawyer et al. 2005b; Alibardi, 2006b; Fig. 2A–F). During development, different layers with a specific fate are produced in sequence in the epidermis. These studies have suggested that specific cell populations were selected from the generalized archosaurian epidermis to produce specific epidermal structures such as scales, beaks or claws, and for the formation of feathers. In particular, some cell populations containing a specific type of feather-like keratin molecule were selected among others to produce barb and barbule cells and eventually feathers.

In the alligator and in the avian embryonic epidermis, outer and inner periderm layers are initially formed and contain similar periderm granules (Fig. 2B–F). Over most of the body, especially in inter-follicular, apteric and scale skin, only one layer of inner periderm cells is present (Sawyer & Knapp, 2003) (Fig. 2F). In the claw, two to four layers of inner periderm cells are instead present, whereas this increases to six to eight layers in the beak (Kingsbury et al. 1953; Alibardi, 2002, and personal observations). Beneath the inner periderm, one subperiderm layer is present which contains feather-keratin immunoreactivity (Fig. 2C,E). It has been suggested that the displacement of the subperiderm layer into barbule plates and a ramus area eventually generated feathers (Fig. 3D). The subperiderm layer in all archosaurian epidermis contains a 24-amino acid feather-keratin epitope (AVGSTTSAAVGSILSEEGVPINSG) that disappears at hatching except in feathers (Sawyer et al. 2005a,b). Therefore this epitope may be an ancient component of the epidermis of archosaurian reptiles.

Figure 3.

Hypothetical evolution of feathers and their keratins from reptilian scales (A), through coniform scales (B) and a long coniform proto-feather (C). The latter contained a specific feather-keratin box (FK-box) together with the ancestral core-box. The glycine-rich box (GG-box) was lost from B to C. The formation of barb ridges (D) shows the inside view of the base of a feather filament and the different cell displacement (E–G) gives rise to three different branching phenotypes (see text for explanation). Finally, the fusion of barb ridges with the rachidial ridge gives rise to pennaceous feathers (H1, contour feathers; H2, bristles; H3, filoplume). AX, axial plate; BA, barbula; BL, barbule; BR, barb ridge; CO, collar; NA, no axial plate; RA, ramus; RM, ramogenic zone; RR, rachidial ridge; SA, short axial plate.

Feathers as a skin specialization of archosaurian epidermis

It is still believed that feathers might be derived from archosaurian scales (Spearman, 1966; Maderson, 1972; Maderson & Alibardi, 2000), probably by a drastic alteration of the morphogenetic pattern of dermo-epidermal interactions. Reptilian scales are either of the scute type, as in extant crocodilians and chelonian shell, or appear as tuberculate or pebble-like scales, such as those described in fossilized archosaurians, including dinosaurs (Martin & Czerkas, 2000; Prum & Brush, 2002). It is believed that during the evolution of the avian skin the rigid corneous layer of archosaurian scales progressively became restricted to smaller and smaller areas of ADEIs, while most of the skin became non-specialized and softer, and was the origin of the apteric skin (Alibardi, 2004b, 2005). The flat scale in archosaurian ancestors became progressively narrower, tuberculate and the epidermis was nourished from a vascularized mesenchyme, a condition that resembles that of feather filaments (Fig. 3A–C). The latter also exchanged morphoregulative or inductive signals with the epidermis, and this activity determined the formation of specialized layers of cells in both scales and feathers.

The fossil record of feathers is relatively abundant, already showing the typical contour feather morphology (Prum & Brush, 2002; Wu et al. 2004). Simpler forms or filiform appendages are interpreted as both protofeathers (summarized in Prum & Brush, 2002; Wu et al. 2004) or collagenous structures (Lingham-Soliar et al. 2007). The detailed cytological analysis of the process of feather morphogenesis has, however, indicated that some fossilized ‘protofeathers’ may be genuine remnants of barbs with different degrees of branching (Alibardi, 2006a,b,c).

According to the latter studies, the hairy-like elongation was hollow, and when epithelial folds appeared in the epidermis of these hypothesized filaments, the formation of feathers began. The first stage was the formation of barb ridges that could evolve into different morphogenetic mechanisms to produce barbs with extensive ramification (Fig. 3E-E3), short ramification (Fig. 3F-F3) or without any ramification (Fig. 3G-G3) (Alibardi, 2006a,b). Similar morphotypes of downfeathers have been found in fossilized feathers or filamentous structures (summarized in Prum & Brush, 2002; Wu et al. 2004).

The above hypothetical stages are supported by morphological and biochemical data. Reptilian and avian scales and feathers are mainly composed of scale and feather-keratins, and feathers can develop from scales under some natural or experimental circumstances (Dhouailly & Senegel, 1983; Chuong, 1993; Chuong & Widelitz, 1999; Sawyer et al. 2005b). The successive fusion of barb ridges into an axial rachis gave rise to pennaceous feathers of different shapes and sizes (Fig. 3H-H3).

Hairs as a skin specialization of theropsid reptiles

It is believed that the evolution of hairs from basic reptiles resulted from the reduction of a primordial, scaled skin into a hairy and softer skin (Maderson et al. 1972; Alibardi, 2003; Wu et al. 2004). Also in this case, it is hypothesized that the initial expanded ADEIs of reptilian scales became progressively confined to small areas, giving rise to dermal papillae and the formation of hairs among scales, as in the tails of rodents (Alibardi, 2004a,b,c; Fig. 4A-A3). The above references give more information on the possible evolution of hairs, which cannot be considered here. Unfortunately, while some feathers and corneous filaments remain in fossils, no fossilized hairs have so far been reported in the paleontological literature and therefore we do not have any skin evidence about hair evolution. Also the developmental stages of hair development provide few clues about the evolutive transition from a scale to a hair as the two structures are very different (Fig. 4B-B7).

Figure 4 (.

A) Schematic representation of the hypothesis of hair evolution following the reduction of areas of dermal–epidermal interactions (ADEI) (A 1–3). (A1) In a large synapsid scale the area of contact (red layer) between the active dermis (green beads) and epidermis was extended over most of the outer scale surface, forming expanded ADEIs. (A2) ADEIs become smaller and smaller in later mammal-like reptiles (therapsids), producing dermal condensation near the hinge region while scales became reduced or disappeared. The concentration of active dermal cells, a precursor of the dermal papilla, induced the proliferation of a rod of corneous material out of the hinge region, a proto-hair. (A3) In advanced therapsids or true mammals the formation of a (condensed) dermal papilla stimulated the production of longer rods of corneous cells, the hairs. (B) Schematic representation of hair morphogenesis (1–6) to the mature hair (7). From the linear epidermis (1) a hair placode is formed (2) that grows downward into a hair germ (3). The latter forms a hair peg (4 and 4′) that grows into a bulbous peg in which the active dermis starts to penetrate into its cup-like basal epithelium to form a dermal papilla (5–6). From the matrix cell precursor of the hair initially only trichocytic keratins synthesize, and later KAPs are formed at different levels of the differentiating cortical and cuticle cells. Also cells of the inner root sheath express specific keratins and later trichohyalin (7). AD, active dermis (competent to send inductive signals to the epidermis); ADEIs, areas of dermal–epidermal interactions; BU, bulge (stem cell repository); CL, companion layer; CO, cortical cells (containing long bundles of corneous proteins); CU, hair cuticle; DC, dermal condensation; DP, dermal papilla; EP, epidermis; EIRS, elongating inner root sheath; FDP, forming dermal papilla; FIRS, forming inner root sheath; GE, germinal epidermis (matrix); HCU, cuticle of the hair; HE, Henle's layer; HGT, high glycine tyrosine proteins; HL, hair-like primordia; HR, hair; HS, high sulfur proteins; S, high-sulfur proteins; HUX, Huxley layer; ICU, cuticle of the IRS (serrations correspond to those of the IRS); IK, intermediate filament keratins; IRS, inner root sheath; KAPs, keratin-associated proteins; MD, medulla cells; P, periderm; SG, sebaceous gland; SZ, sloughing zone; TH, trichohyalin; TK, trichocyte keratins; UD, undifferentiated dermis (does not produce signaling molecules); UHS, ultra-high sulfur proteins; VE, vesicles.

Figure 5.

Mammalian proteins involved in cornification. (A) Alpha keratins that normally produce an alpha-pattern can be deformed under some conditions (stretching or steaming the corneous material above 70–80 °C). These conditions induce a change of secondary conformation, from alpha-helix to random coil and even beta-sheets, which produce a beta-keratin. (B) Some examples of mammalian KAPs are shown. The key amino acids are colored with different colors to illustrate the three main groups – HGT, HS and UHS matrix proteins. (C) Examples of the main mammalian skin derivatives containing trichocytic keratins and KAPs to form very hard corneous layers. (D) Schematic representation of the progressive phases (1–3) of association between coiled alpha-keratins (trichocytic) and KAPs. In phase 3, KAPs are probably regularly ordered around alpha-keratins and the resulting pattern at X-ray is indicated as alpha pattern.

Figure 6.

Localization of alpha-keratins (A–B) and beta-keratins (C–F) in reptilian corneous layers. (A) AE3-immunogold labeling (arrows) in lizard (Podarcis sicula) beta cells among pale beta-keratin bundles. Bar, 150 nm. (B) Detail of AE3-labeling over filaments (arrow) surrounding the pale beta-keratin bundle in a beta-cell of lizard (P. sicula). Bar, 100 nm. (C) Beta-1 immunolabeled, thick corneous layer of carapace scute in turtle (Chrysemys picta). Dashes underline the basal epidermis. Bar 10 µm. (D) Beta-1 gold immunolabeled corneous layer of scute of C. picta that disappears along the border with the underlying living keratinocyte (arrows). Bar, 250 nm. (E) Beta-lizard (A68B-antibody) immunolabeled keratin bundles in a beta-cell of P. sicula. Bar, 250 nm. (F) Beta-universal (β-U) labeling over keratin bundles of a differentiating beta-cell in snake (Natrix natrix). Bar, 250 nm. (G) Schematic general pattern of epidermal proteins from the epidermis of different species of reptiles (lizard, gecko lizard, turtle, alligator, crocodile and snake) showing the relative amount of alpha- vs. beta-keratins. AE3, immunolabeling positive for alpha-keratin AE3; be, beta-keratin bundles; β1, immunolabeling for beta-keratins using the chick scale β1 antibody; c, corneous layer.

Figure 7.

Genomic organization (A–F) and examples of nucleotide structure (G–I) of avian and reptilian beta-keratin genes. A cluster of genes coding for related proteins is localized in a chromosome locus (A) and they are linearly arranged in multiple copies (B). In the genes of chick (chicken feather-keratin gene, AC J00847, and snake (Elaphe guttata, AC AM404188), a variably-long intron is present in the 5′-non coding region (C,F,G,H). In contrast, in the gecko lizard species studied so far (here exemplified by one sequence of Tarentola mauritanica, AC AM162665), no intron has been found (E,I). ATG (boxed) initiation codon; TAA or TAG (boxed), termination codon; AATAAA (boxed), polyadenylation signal. Uppercase letters, coding region; lowercase letters, 5′ and 3′ untranslated terminal regions; bold lowercase letters, intron sequence. The splice signals of introns are boxed and shaded.

Hairs develop from an invagination of an epidermal placode in contact with a special mesenchymal condensation to form a peg that grows downward into an elongated, club-like epidermal tongue (see stages of hair morphogenesis, Philpot & Paus, 1999). From the neck of the elongated peg some cells become organized in a cone-like layer that grows upward and initiates the hair shaft that grows toward the epidermal surface (Fig. 4B5–6). In the meantime, the mesenchyme at the bottom of the club (the future hair bulb) penetrates into a small cavity that is formed at the base of the club, and forms the dermal papilla (Fig. 4B5–B7). A classic study on developing hairs in monotremes (Poulton, 1894) showed that their hair canal may develop as an open tube, a study that, if confirmed, would provide some evidence of a process of invagination (see references in Alibardi, 2004a,b).

Other hard skin derivatives present in mammals are represented by nails, quills, claws, horns, hooves, corneous teeth in whales, etc., all considered to incorporate hard keratins (Gillespie, 1991). Their embryological origin is partially known but all of them appear to have an intimate dermal–epidermal connection which directly or indirectly induces epidermal cells to produce ‘hard keratins’ (Bragulla & Hirschberg, 2003).

Corneous proteins of hairs, scales and feathers: alpha-keratin vs. keratin-associated proteins of mammals, birds and reptiles

General features of corneous proteins in mammalian epidermis and its derivatives

In the last 25 years the concept of mammalian hard keratins has been more precisely defined. Today, two main components are recognized in the cells (trichocytes) of mammalian hard corneous derivatives (hairs, nails, horns), a fibrous component and a matrix or inter-filament component. The fibrous component consists of a specific set of alpha-keratin proteins, composed of the keratin intermediate filaments (KIFs), with molecular weights of 40–70 kDa. These keratins have been sequenced from their genes cloned from human and mouse (Langbein et al. 1999, 2001; Langbein & Schweizer, 2005), and are classified into two types based on their sequence homology, namely 11 type I (acidic) and nine type II (basic). The spatial-temporal sequence of expression of these hair keratins varies at the different layers and level of developing hairs, suggesting different mechanical roles in differentiating trichocytes (Fig. 4).

The matrix or inter-fibrillar component comprises ‘keratin-associated proteins’ (KAPs), generally with smaller molecular weight (8–30 kDa) (Gillespie, 1991; Powell & Rogers, 1994; Rogers et al. 2006). Over 100 KAPs have been identified and form the matrix among KIFs that effectively results in the hard and resistant corneous material of hairs, nails and horns (Gillespie, 1991). The KAP proteins are not fibrous, but more globular in conformation, but the molecular organization of these proteins and their linkage with KIFs is not well understood (Rogers, 2004; Rogers et al. 2006). X-ray diffraction, electron microscopy and chemical studies have demonstrated that hair KIFs contain about 70% of alpha-helical domains, the keratin intermediate filaments are 7–10 nm thick (Filsie & Rogers, 1961; Fraser et al. 1972), and disulfide links KIFs together and also links KIFs with KAPs (Rogers, 2004; Rogers et al. 2006).

When hair fibers or films of corneous material (produced after chemical extraction or by direct application of X-rays to thin pieces of the corneous material, see Fraser et al. 1972), are stretched or heated, the alpha-type X-ray diffraction pattern changes to what is termed the beta-pattern. The alpha-helical conformation of the keratin chains in the KIFs under stress elongates to form regions of beta-pleated sheets. Early stages of stretching or heating can be reversible (Fig. 5A). Reptilian/avian keratins, in which there are regions containing beta-sheets, are referred to as beta-keratins and are very different from the keratin proteins of intermediate filaments.

Mammalian KAPs (mKAPs) are grouped into three types with a molecular weight range of 6–36 kDa. They are distinguished by their content of particular amino acids, which include glycine-tyrosine-rich (HGTs), high sulfur proteins (HSPs), and ultra high-sulfur proteins (UHSPs). Each of these protein groups is a family of proteins of related sequence. Their regulated expression in the differentiating hair shaft in the follicle follows that of specific acidic/basic trichocytic keratins to which they are associated (Rogers, 2004; Langbein & Schweizer, 2005; Rogers et al. 2006; Plowman et al. 2007). HGT (KAPs 6–8) are mainly synthesized in the lower part of the differentiating zone (pre-cortex), located above the matrix zone, HSP (KAPs 1–3) in the intermediate region, UHSPs (KAPs 4 and 12) in the maturing cortex, and other UHSPs (KAPs 5, 9, 10) in the upper, maturing cells of the hair shaft cuticle (Powell & Rogers, 1994; Rogers, 2004; Rogers et al. 2006) (Fig. 4). Some examples of amino acid sequences of human KAPs are shown in Fig. 5B, where four key amino acids (glycine, cysteine, serine and proline) are colored for comparison with the proteins present in sauropsids (reptiles including birds). The number of amino acids varies from 57 to 90 in HGT proteins, 98 to 298 in HS proteins, and 114 to 205 in UHSs, with molecular weights variable from 5.4 to 36 kDa. The prediction of the secondary structure in mammalian KAPs (mKAPs) using the psipred program (Alibardi et al. 2007; Toni and Alibardi, unpublished data) on 128 mKAPs has shown that only few HSs present strand regions, and a few (less of 5.5% in our analysis) show regions with a core-box structure.

Figure 5D shows schematically that KAPs surround the numerous alpha-keratin filaments and become organized around them, forming the matrix of the corneous material.

The beta-keratins in sauropsid epidermis and its derivatives

The keratin proteins of reptiles have remained poorly known, compared with those of mammalian and avian proteins, for over 30 years. They were considered to consist of soft (alpha)-keratin and hard (beta or phi)-keratins (Maderson, 1985; Landmann, 1986; Alibardi, 2003). The amino acid sequences of keratins present in reptilian corneous structures (scales, claws, ramphotheca, etc.) were unknown until recently. The hard proteins of reptiles and birds are proteins (beta-keratins) with a fibrous structure that gives a beta-pattern by X-ray diffraction and a filament of 3–4 nm in diameter, as first described by Filsie & Rogers (1961).

The amino acid composition and molecular weights of beta-keratins were available for some reptilian species (Baden & Maderson, 1970; Baden et al. 1974; Wyld & Brush, 1979, 1983; Sawyer et al. 2000; Homer et al. 2001). No reptilian alpha-keratins of the softer and harder layers had been sequenced, and only one small keratin from a lizard claw was known in its primary amino acid structure (Inglis et al. 1987). However, it was unclear whether this protein was representative of reptilian scales or merely reflected the composition of a specialized claw keratin. Also, a 20-amino acid sequence localized in an inner region of an alligator beta-keratin was known, and this epitope shows high identity with that of avian beta-keratins (Sawyer et al. 2000, 2005a).

It was known that different proportions of specific monomers of beta-keratins, called phi-keratins, are present in reptilian derivatives of different rigidity (Wyld & Brush, 1979, 1983; Thorpe & Giddings, 1981; Gillespie et al. 1982; Marshall & Gillespie, 1982; Inglis et al. 1987). These proteins are of small molecular weight, mainly in the 10–20 kDa range, but a component of higher molecular weight was also found (30–44 kDa), although in minor amounts. It was not clear whether they were true polymers or aggregates derived from chemical extraction. Different forms (monomers) are present in various skin derivatives such as dorsal and ventral scales, shell scutes, claws, and ramphotheca in turtles, lizards, snakes and crocodilians. According to Wyld & Brush (1983), phi (beta)-keratins of reptiles are believed to be phylogenetically distant from beta-keratins of birds, or even unrelated to feather-keratins (Brush, 1993). The latter study in particular pointed out that feather-keratins and feathers are a completely new innovation of avian skin.

Other studies on crocodilian and avian beta-keratins have indicated that these proteins can be distinguished into a few main types: scale, beak, claw, feather or feather-like keratins (Gregg et al. 1983; Gregg & Rogers, 1986; Presland et al. 1989b; Sawyer et al. 2000, 2005a; Sawyer & Knapp, 2003). Beta-keratins probably vary from species to species in their amino acidic sequences, and those in archosaurian skin (crocodilians) are strictly correlated to those of birds. Scale, claw, and beak keratins showed a relatively higher molecular weight, 14–18 kDa, whereas feather or feather-like keratins possessed a smaller molecular weight, 10–12 kDa (Gregg & Rogers, 1986). Studies on feather genes suggested that at least 30–40 genes alone were present, and other 10–20 genes were found for scale, claw and beak keratins.

Early knowledge on the process of cornification in mammals vs. reptiles/birds indicated that some striking differences were present. In the corneous skin appendages of reptiles/birds, beta-keratin aggregates to form filaments differently from what occurs with alpha-keratins (Gregg & Rogers, 1986; Gillespie, 1991; Brush, 1993). This suggested that a more central region of the molecule of beta-keratin contributed to make the fibrous framework of the filamentous polymer, while the extremities formed the matrix. This was a different mechanism from that present in mammalian corneous appendages. It is not clear why such different mechanisms of cornification originated in reptiles/birds compared to mammals as they have a common reptilian progenitor.

The resulting alpha-keratin pattern of mammals vs. the beta-keratin pattern of sauropsids might be due not to a difference in keratins but instead to a different composition and organization between proteins or the fibrous component (alpha-keratins) vs. those of the matrix component (keratin-associated proteins). This organization is different in the corneous material of mammals compared with that in reptiles/birds, but in both cases the corneous material consists of a fibrous and a matrix component.

Immunocytochemical and biochemical studies suggest that beta-keratins function as keratin-associated proteins in sauropsids

Numerous ultrastructural and immunocytochemical studies on reptilian epidermis have shown the presence of both cytokeratin and beta-keratin immunoreactivity (Alibardi & Sawyer, 2002; Alibardi, 2003, 2005, 2006c; Alibardi & Toni, 2006a–d; see examples in Fig. 6). Controls, omitting the primary antibody, are unlabeled (data not shown). Alpha-keratins in reptilian corneous material were revealed using broad-spectrum cytokeratin antibodies such as AE3 and Pan-cytokeratin antibodies, which recognized different acidic and basic cytokeratins (Alibardi & Toni, 2006a).

The latter indicated that beta-keratins in reptiles and birds have broad cross-reactivity as shown by a chick general antibody (beta-1) (Alibardi & Sawyer, 2002). Later analysis indicated that most reptilian and avian beta-keratins also contain a ‘universal’ antigen found in avian and alligator beta-keratins with the sequence SRVVIQPSPVVVTLPGPILS (Sawyer et al. 2000, 2005a). This sequence, or the homologous sequences, contained in a central region of all known beta-keratins, were indicated as the core-box (Alibardi & Toni, 2006a, 2007). The final confirmation of the precise localization of beta-keratins in reptilian epidermis came after the isolation of antibodies produced in rat and directed to 14–17 kDa proteins isolated from reptilian epidermis (Alibardi & Toni, 2006c,d).

Our proteomic studies on epidermal proteins in reptiles have initially confirmed that beta-keratins generally fall within the 12–20 kDa range (Alibardi & Toni, 2006b, c,d; Toni & Alibardi, 2007a,b,c; Toni et al. 2007ab). The use of general and more specific antibodies for alpha- and beta-keratins after two-dimensional electrophoretic separation has identified mainly acidic and neutral alpha-keratins in the 40–70-kDa range (Fig. 6G). The other large group of proteins, beta-keratins of 10–20 kDa, are mainly basic and neutral, and there is also a small fraction that appears to be acidic (Fig. 6G). Therefore alpha- and beta-keratins may be complementary and thereby associate into keratin filaments, a unique condition of sauropsid keratins.

These immunological proteomic studies indicated that beta-keratins seem to associate with alpha-keratins as mammalian KAPs associate with keratin intermediate filament keratin proteins (KIFs) in mammalian tissues, suggesting that beta-keratins may actually function as reptilian KAPs. Matrix proteins so far known in amniote epidermis (filaggrin and trichohyaline) are basic and they have charged interactions with neutral or negatively charged groups present in keratin filaments (Mack et al. 1993). In avian feathers two basic KAPs, indicated as CAP-1 and CAP-2, and four basic histidine-rich proteins (HRPs) are different from beta-keratins (Gregg & Rogers, 1986; Presland et al. 1989a; Knapp et al. 1991,1993; Barnes & Sawyer, 1995). In summary, all amniotes produce corneous material using alpha-keratins as fibrous material and keratin-associated proteins as a matrix. The peculiarity of reptilian and bird KAPs is that the proteins form filaments that replace or mask those of alpha-keratins. This derives from the special amino acid composition of some key regions of reptilian and avian KAPs, as it will be illustrated in the next section.

Beta-keratins in reptilian and avian keratinocytes of scales and feathers form fibrous polymers and filaments (Fraser et al. 1972; Brush, 1983) as compared to mammalian hard corneous material of hairs, nail and horns, where mKAPs are mainly amorphous. The beta-keratins are also different from intermediate filament proteins in composition, chemical-physical properties, and mechanism of polymerization. Therefore small proteins previously indicated as beta-keratins (fibrous proteins with regions of beta-pleated sheet) should be considered a fibrous type of keratin-associated or matrix proteins (sauropsid KAPs) (Alibardi et al. 2007; Toni et al. 2007a,b). The term beta-keratins will be used in the present review as being equivalent to sauropsid KAPs (sKAPs). However, it is generally agreed that true keratins are defined as belonging to the keratin intermediate filaments family (Steinert & Freedberg, 1991; Fuchs & Weber, 1994; Coulombe & Omary, 2002).

The lack of knowledge about proteins and their genes involved in the production of hard proteins in reptilian scales has impeded a proper evaluation of the evolution of these proteins from reptilian ancestors to birds. The essential, missing information in reptiles was the protein sequence, secondary and tertiary conformation, and the genes coding for these hard proteins. The recent genomic and proteomic studies on reptilian beta-keratins have rapidly clarified this issue and the new information suggests a new interpretation that unifies the mechanism of cornification in all amniotes.

Genomic structure and expression of sauropsid keratin-associated proteins: nucleotide sequences, genes and chromosomal localization

Genes encoding reptilian beta-keratins

The recent molecular biology results on reptilian beta-keratins have been initially obtained by designing degenerate primers from specific regions (core-box) of avian and a lizard claw beta-keratins (Inglis et al. 1987). These primers were then replaced by more specific primers selected on the same reptilian regions (Dalla Valle et al. 2005, 2007a,b, 2008 and 2009). This work has aided the deduction of the amino acid sequences of lepidosaurians (lizards, snakes), chelonian, and crocodilian beta-keratins. The nucleotide and protein sequences of different reptilian keratins are now deposited in GenBank under specific accession numbers.

This initial information has been now integrated with the analysis of the lizard Anolis carolinensis and chicken genomes (Dalla Valle et al. unpublished studies).

These data have allowed, for the first time, the complete comparison of the nucleotide and amino acid sequences and protein structure of different reptilian beta-keratins with avian keratins. In fact, the nucleotide structure of beta-keratin cDNAs so far sequenced from reptilian epidermis (Dalla Valle et al. 2005, 2007a,b) show a general linear organization that resembles the avian nucleotide structure for their scale, claw and feather genes (Gregg et al. 1983, 1984; Presland et al. 1989a,b; Whitbread et al. 1991).

The gene structure in snake, crocodiles and, according to the first data, also in turtle, show the same organization as previously found in avian beta-keratin genes: the coding region is uninterrupted, whereas an intron of variable length is present in the 5′-untranslated terminal region (5′-UTR) (Fig. 7). The first reptilian gene coding for beta-keratin containing an intron has been sequenced in the snake Elaphe guttata (now called Panthoteris guttatus) (Dalla Valle et al. 2007b). In the snake gene an intron of 226 bp is present after 29 bp of the 5′-UTR and it is followed by a short non-translated region of 17 bp. A similar result has been found in the crocodile Crocodylus niloticus, in which the 5′-UTR contains an intron of 621 nt (Dalla Valle et al. 2008).

The presence of intron and the exon-intron borders was identified based on the direct comparison of the cDNA and the genomic sequences obtained by PCR using sense primers located at the beginning of the 5′-UTR and antisense primer covering the stop codon. However, this approach gave a different result in the two species of geckos analysed: a fragment of the same length was amplified with both cDNA and genomic DNA. Sequencing showed that no intron was present in the two genomic fragments obtained (Dalla Valle unpublished data).

The available genome of the American chameleon lizard Anolis carolinensis (at http://genome.ucsc.edu/cgibin/hgGateway?hgsid = 94953969&clade = vertebrate&org = Lizard&db = 0) has allowed analysis of the presence of genes coding for beta-keratin proteins similar to those found in other lizards. In total about 40 lizard genes are present in this species; moreover, as all genes were found in the same scaffold, these genes are probably localized in one single chromosomal locus (Dalla Valle et al. work in progress). The accessibility of A. carolinensis genome allowed direct comparison of the cDNA with the genomic sequences. A preliminary 5′-RACE analysis on some beta-keratin mRNAs expressed in the epidermis of the green anole (Anolis carolinensis), showed that they contain an intron in their 5′-UTR (Dalla Valle et al. unpublished results). At present it is known that one intron of variable length is present in the 5′-untranslated region of beta-keratins in all reptiles: snake, crocodile and in the turtle. It is not known whether the intron may be the regulative region for the transcription. Although in gecko lizard genes so far analysed (from two different species) no intron has been found, these observations suggest that other beta-keratin genes within the lizard genome likely possess introns. The presence of one intron in the 5′-UTR of reptilian beta-keratin gene is probably general but not always constant. More work, such as the 5′-RACE analysis of beta-keratins of the green anole is necessary to clarify this point.

Genomic Southern analysis using reptilian β-keratin cDNA as a probe has indicated the presence of several related genes in the genome of geckos, snake, crocodile and turtle (Dalla Valle et al. 2007a,b, 2008 and 2009). These data have confirmed that also in reptiles β-keratins belong to a multigene family as demonstrated in the chick (Gregg et al. 1983, 1984; Presland et al. 1989a,b; Whitbread et al. 1991). These results are also confirmed after some cloning experiments leading to the sequencing of more then one gene in each analysed reptilian species. When performed, PCR analysis indicates a head-to-tail orientation of these genes in their cluster. However, the analysis of A. carolinensis and avian genomes reveals that, although this is the common orientation, genes are also present that are arranged with tail-to-tail or head-to-head orientation (work in progress).

The in situ localization of messenger RNA for these proteins reveals a strict epidermal expression, mainly localized in the differentiating oberhautchen, in beta-layers of lepidosaurians, and in the pre-corneous layer of crocodilian and turtle scales. (Fig. 8A,B,D–F). The ultrastructural in situ hybridization analysis of mRNA gene expression has shown that transcripts are present in the cytoplasm associated with the initial sites of deposition of beta-keratins (beta-packets; Fig. 8J). However, the periphery of larger bundles of filaments often presents beads of gold particles that identify antisense digoxigenin-labeled probes associated with the filaments (Dalla Valle et al. 2005; Alibardi et al. 2006a). This finding suggests that the protein still in the translation phase can be directly packed in the growing bundle.

Figure 8.

Examples of in situ hybridization localization using antisense probes for specific messengers coding for beta-keratins, and visualized by immunofluorescence (A–D,I), alkaline-phosphatase (E–H), or ultrastructural localization (M). (A) Regenerating scale of gecko (Tarentola mauritanica) with expression (arrows) in cells of the forming beta-layer (arrows). Bar, 10 µm. (B) Beta layer (arrow) of snake (Elaphe guttata). Bar, 10 µm. (C) transitional (arrow) and lower corneous layer of turtle (Pseudemys nelsonii). Bar, 10 µm. (D) transitional layer (arrow) of crocodile (Crocodylus niloticus). Bar, 10 µm. (E) beta-cell layers (arrows) of regenerating scales in lizard (Podarcis sicula). Bar, 20 µm. (F) beta-layer cells (arrow) in scales in renewal stage of snake (E. guttata). Bar, 10 µm. G, fusiform cells of the transitional layer (arrow) of scutes in Chrysemys picta. Bar, 10 µm. H, forming beta-layer (arrow) of scale in the tuatara (Sphenodon punctatus). Bar, 10 µm. I, detail on the spindle-shaped cells of the transitional layer of crocodile (C. niloticus). Bar, 10 µm. J, ultrastructural detail of the cytoplasm of differentiating beta-cell in lizard (P. sicula). Clusters of 5-nm diameter gold particles indicate the position of mRNAs among small keratin material (arrows). The arrowhead points to gold particles associated with a larger keratin bundle. Bar, 50 nm. a, alpha-layer; b, beta-layer; c, corneous layer; d, dermis; e, epidermis.

The genomic structure of avian beta-keratin genes is organized on the reptilian model

Previous information on chick scale, claw, beak and feather-keratins indicated that their genes are clustered in the same locus on chromosomes (Molloy et al. 1982; Gregg et al. 1983, 1984; Gregg & Rogers, 1986; Presland et al. 1989a,b; Whitbread et al. 1991; Brush, 1993; see Fig. 7). However, neither the number of chromosomes containing beta-keratin genes nor their location was indicated. Genes appear to be arranged in multiple copies with the same orientation in the genome. They contain one intron in the 5′-UTR, perhaps with a function involved in transcription regulation. A previous summary of all known avian beta-keratin genes indicated the presence of 18 genes for feather-keratins, three genes for feather-like keratins, four genes for claw keratins and nine genes for scale keratins. No indication was available on the number of genes for beak keratins (summary in Stevens, 1996). This was clearly a large underestimation of avian beta-keratin genes, as initial studies indicated the presence of 100–240 different genes for feather-keratins alone (Kemp et al. 1974). The latest analysis on the chick genome shows that probably around 140 genes for feather, claw, scale and beak keratins are present, not taking into account pseudogenes or incomplete sequencing data (Dalla Valle et al. unpublished results). Among beta-keratin genes, 56 are clustered in chromosome 25, whereas the other 64 genes are present in chromosome 27. A few beta-keratin genes (from a single copy to seven copies) are present in five other chromosomes, but they may be copies of some genes present in chromosomes 25 and 27.

The structure of the chick feather gene (gene C) (Gregg et al. 1984; Gregg & Rogers, 1986) comprises a 5′-UTR of 93 bp, which is interrupted after 37 pb by an intron of 329 bp. Like reptilian genes, the coding region of avian genes is devoid of introns (Fig. 7G). The intron position is conserved in other avian beta-keratin genes sequenced (Gregg & Rogers, 1986) and this is also confirmed by the avian genome analysis. The intron may be of some importance for the regulation of the velocity of transcription of beta-keratin gene (Koltunow et al. 1986). The possibility to increase transcription may explain the rapid production of feather-keratin in 2–3 days, from 13–16 days of development in the chick embryo (Hamburgher-Hamilton stages 37–39; Kemp et al. 1974; Gregg & Rogers, 1986). In a large part of the non-coding region of the beta-keratin gene cluster, is a highly conserved nucleotide sequence, especially nucleotides 1–37, indicating an essential physiological role in the transcription and translation process. The 3′-UTR part of the gene is much less conserved than the other regions.

From the above description and the recent data on the gene organization of crocodilian beta-keratins it is likely that beta-keratins of birds are derived from an archosaurian organization of this family of genes that pre-dated the origin of birds. The latter inherited this basic organization and evolved their own specific sequences for the formation of scale and feather keratins (see description below).

Proteome analysis of sauropsid KAPs: types of proteins in different groups of reptiles and their comparison with those of birds

Comparative analysis of sauropsid KAPs shows that they possess a high homologous beta-pleated region: the core-box

To date about 100 proteins have been completely sequenced and analysed in five species of lizards (Podarcis sicula, Tarentola mauritanica, Hemidactylus turcicus, Gekko gecko, Anolis carolinensis), one snake (Elaphe guttata), a turtle (Pseudemys nelsonii), and the Nile crocodile (C. niloticus) (Dalla Valle et al. 2005, 2007a,b, 2008 and 2009; Hallahan et al. 2008). Partial sequences are known for the alligator (Alligator mississippiensis; (Sawyer et al. 2000), the tuatara (S. punctatus) and the soft-shelled turtle (Trionix spiniferus) (Dalla Valle et al. unpublished results).

The predicted molecular weight and isoelectric point of some deduced proteins correspond to specific spots obtained following two-dimensional separation of epidermal proteins, which confirms molecular findings (Toni et al. 2007a,b and unpublished results). Analysis of the deduced amino acid sequences by specific software can graphically predict secondary structure (Fig. 9). The study of some representative beta-keratins in lizard, snake, turtle, crocodile and birds has demonstrated the presence of a central region where two strands (beta-folded secondary conformation), made up of four to seven amino acids each, are present in-tandem. The core-box represents the region with the highest sequence identity and is present within a 32-amino acid region that contains the strand or beta-pleated sheets in beta-keratin monomeres (Fraser et al. 1972; Brush, 1993; Fraser & Parry, 1996). A recent study on feather and reptilian beta-keratins, based on a different method of analysis of the secondary conformation, reported the presence of four beta-strands in this region (Fraser & Parry, 2008). The predictions show that short beta-sheets made of four to eight amino acids, generally not organized in tandem, are present in other regions of these small proteins (Fig. 9, and data not shown). The role of the extra strands in these proteins remains undetermined but awaits more information from the prediction of their tertiary structures (ongoing analysis in our laboratory).

Figure 9.

Examples of protein prediction using the PSIPRED Protein Structure Prediction Server at http://bioinf.cs.ucl.ac.uk/psipred/ for representative proteins of lizard, snake, crocodile, and turtle. The molecular weight and isoelectric point (pI) of these proteins are indicated. They all present inner amino acid regions with two close strands (beta-folded sheats) indicated as core-box (boxed). Note the variable extension of the alpha-helical region (green rods) in relation to the random-coiled regions (indicated with black lines).

The determination of the amino acid sequence for beta-keratins from lizards, snake, turtle and crocodile, together that for a lizard claw (Inglis et al. 1987), has indicated that some regions of these proteins present high sequence identity with avian keratins (Dalla Valle et al. 2005, 2007a,b). These predictions show some of the structural homologies of reptilian and bird proteins, especially in the core-box or beta-pleated sheet region (Figs 9, 10).

Figure 10.

Comparative representation of some sKAPs from all different reptilian groups. Some key amino acids are specifically colored to reveal their preferred location within these proteins. Note in particular the conservation in the core-box region and the presence of cysteines mainly toward the N- and C-regions in all groups of reptiles and birds, where they probably intervene in cross-linking. Glycine-rich regions in HGPS proteins of lepidosaurians are present toward the N- and C-regions. Cysteine residues in lepidosaurian HCPS proteins are also present toward the N- and C-terminals. Serines are mainly present outside the core-box where they may be involved in phosphorylation or in other post-translational processes. In the chimeric-like chelonian/archosaurian proteins a cysteine-rich region is particularly evident, while the glycine-rich region is mainly present toward the C-terminus. Finally, the glycine-rich region disappears in the specialized, smallest feather proteins (FK). Goanna-claw (Inglis et al. 1987); Anolis Ker 20, 22, 23, 24, 25; Ge-gprp-1 (CAJ44302); Ge-gprp-3 (CAK19321); Ge-gprp-4 (CAJ90467); Ge-gprp-5 (CAK19322); Li-gprp-1 (CAJ67601); Li-gprp-3 (CAJ90483); Li-gprp-5 (CAJ90485); Sn-gprp-1 (CAL49457); Sn-gprp-5 (CA CAL51276); Sn-gcrp-1 (work in progress); Tu-gptrp-1 (CAO78677); Tu-gptrp-2 (CAO78678); Cr- gptrp-1 (CAO78748); Cr-gptrp-2 (CAO78749); Chick-scale (P04459), Chick-beak (AAO85139), Chick-claw (AAA62730), Chick-feather I–IV (P02450, P04458, P20307, P20308); Chick-keratinocyte (NP-001001310); Turkey-vulture-feather (Q98U06); Wood-stork-feather (Q98U05); Pigeon-feather (BAA33471); Duck-feather (PO8335).

Surprisingly, the predicted secondary structure of the deduced proteins from turtle resembles those of crocodilians and birds, which can be included in a subgroup different from those of lepidosaurians. This suggests that turtle and archosaurians have a closer or common basic archosaur progenitor.

While the concentration of proline distorts the polypeptide chains, the presence of numerous valines, isoleucine and leucine in some cases, determines the formation of a beta-sheet structure in this region (Fig. 11, top panel).

Figure 11.

Examples of the specific amino acid sequences that characterize each group of reptiles. (Top panel) Comparative representation of some examples of proteins in which the position of three key amino acids such as tyrosine, valine and isoleucine is indicated. Tyrosine is mainly lateral (hydrophilic interactions?), whereas valine and isoleucine are more concentrated in the core-box and nearby regions, where they contribute to the formation of strand regions. (Other panels) Examples of group-specific (lepidosaurians vs. chelonians/arcosaurians) amino acid sequences found so far in pre- and post-core-boxes of sauropsids (see text).

The beta-strands present in this region of the protein are considered the site of polymerization of beta-keratin monomers into the long polymer that forms the filaments of beta-keratin 3–4 nm in diameter (Fraser et al. 1972; Brush, 1983; Gregg & Rogers, 1986; Fraser & Parry, 1996). This region probably has a three-dimensional conformation acting as an anchoring point for the interaction with similar proteins (monomers). The random coiled regions, generally representing most of the protein sequence, are regions with free rotation of R-groups of amino acids along the fixed -CO-NH- peptide bond.

Our biochemical studies have shown that alpha-keratins are present in hard corneous material of reptilian scales, so that proteins of intermediate filaments are probably masked by the deposition of beta-keratins (sKAPs). It seems that in scales and claws of different reptiles and birds, in the beak and feathers, the amount of sKAPs capable of forming filaments is generally larger than that of alpha-keratins.

The synthesized beta-keratins present in different appendages of sauropsid integument (scales, claws, ramphoteca or beak, and feathers) can form bundles of corneous material as shown in Figs 6, 12 and 13. Alpha-keratin is generally replaced or degraded, whereas sKAPs accumulate and make filaments through the polymerization of their core-boxes. The abundance in proteins containing the beta-strands is probably responsible for the prevalent beta-keratin pattern observed after X-ray diffraction on the corneous layer of reptilian epidermis (Baden & Maderson, 1970; Fraser et al. 1972). It is believed that the stabilization of the filaments of beta-keratin occurs by homophilic interactions in the glycine-rich regions and by intra- and inter-molecular disulfide bonds. The latter are favored by the prevalent localization of cysteines toward the extremities of these proteins, in proteins both with low and high amounts of this amino acid, as indicated in the next section (Fig. 10).

Figure 12.

Schematic drawing illustrating the synthesis of corneous material in beta-keratin cells of various skin derivatives in sauropsids. Beta-cells (B) of claw (A), scale (A1) and beak (A2) accumulate proteins with glycine-rich regions which produce irregularly organized filaments (B2) for high mechanical resistance. Only feather-keratins (A3) do not possess glycine-rich sequences. Beta-keratin filaments (in black in B3 and B4) increase among the decreasing alpha-keratin filaments. The molecular organization of keratin-associated proteins (KAPs) and histidine rich proteins (HRPs) among fibrous proteins is not known. The detail in B5 schematically shows the antiparallel overlapping of keratin monomers (see orientation of arrows inside the core-box) to form the framework of the protein polymer. Numerous inter- and intra-molecular disulfide bonds strengthen the stability of the filament/s.

Figure 13.

Protein predictions using the PSIPRED Protein Structure Prediction Server at http://bioinf.cs.ucl.ac.uk/psipred/ of avian scale and claw keratins (both HTPS proteins) in comparison with the shorter feather protein (HCPS protein). The molecular weight, isoelectric point and type of proteins are indicated. The core-box is present in all these proteins despite the specific amino acid length and composition in other regions. The lack of a glycine-rich region in the feather protein allows its polymerization into long filaments/bundles tending to be parallel, as indicated in the lower right panel (compare Figs 13 and 12). The accumulation of corneous material mainly with a parallel orientation in feather cells during progressive stages of morphogenesis is indicated by the drawing in the bottom right panel. The small amount of alpha-keratin (represented by short coils) is rapidly replaced by the small feather protein (BFK). The detail in the drawing schematically shows the antiparallel overlapping of keratin monomers (see orientation of arrows in the core-box) to form the framework of feather-keratin polymer. Numerous inter- and intra-molecular disulfide bonds strengthen the stability of the filament(s). BL, barbule cells; BCC, barb cortical cells; BMC, barb medullary cells (become vacuolated); HCPS, high cysteine proline serine proteins; HRP, histidine-rich proteins; HTPS, high tyrosine proline serine proteins; HCPS, high cysteine proline serine protein. KAP, keratin-associated proteins; KB, keratin bundles; PF, parallel filaments.

Specific amino acid regions define different sKAPs: lepidosaurians vs. chelonian and archosaurian proteins

The analysis of the predicted secondary structures in lepidosaurian proteins shows that, outside the core-box, most of their molecule contains a random coil conformation and alpha-helix regions are relatively scarce. According to the content of some amino acids, lepidosaurian proteins can be generally divided into high glycine proline serine (HGPS) and high cysteine proline serine (HCPS) (Fig. 10). The latter presumably form numerous disulfide bonds and may be prevalent in the harder corneous material of reptilian scales or in claws.

Further analysis of the primary and secondary structure of glycine-proline-rich proteins of crocodilian and turtle epidermis shows that long regions of the proteins are occupied by a random-coil conformation (Fig. 9).

In the crocodile, after the core-box, from amino acid 57 to 171, near the N-terminal, a region rich in glycine-X and glycine-glycine-X sequences is found. In crocodile glycine-proline-tyrosine-rich protein 1 (Cr-gptrp-1), 19 GGX repeats are present. In particular, from amino acid 100 to 135 alternate repeats of six GGY-GGL are found in Cr-gptrp-1 (Figs 9, 14, Dalla Valle et al. submitted). Even more remarkable is the 23 GGX repeats found in Cr-gptrp-2 protein, the longest so far isolated in any sauropsid protein. As in crocodilian beta-keratins, chick scale (65–74% identity) and claw (66–68 % identity) contain long glycine-rich sequences although they are shorter than in crocodile. Feather-keratins are missing such sequences (Gregg et al. 1984). This progressive reduction of glycine-rich sequences from the C-terminus toward the core-box of archosaurian scales vs. feather-keratins indicates an evolution of the protein toward simplification. This change was somehow favorable for the specialization of feather-keratin to form long bundles for the elongation of barb/barbule cells (Fig. 13). The small proteins form better oriented filaments with fewer cross-links than larger proteins with long glycine-rich regions (Fraser et al. 1972; Gregg & Rogers, 1986). Feather-keratins tend to resist tension better than scale keratins as they are more oriented. Scale proteins present a less organized distribution, which increases mechanical resistance (Brush, 1993).

Figure 14.

Skin derivatives of sauropsids (A), and schematic mechanism of formation of their hard corneous material (B). This occurs as in mammals (compare with panel D in Fig. 5) but, because of the core-box, sauropsid KAPs form themselves into filaments that replace most intermediate filaments. This process determines the prevalence of a beta-keratin pattern over the initial alpha-pattern. The embryo of a crocodilian (C) and its scale (C1) synthesize a protein with a long glycine-rich region that forms irregularly oriented corneous bundles (C2). (D) A bird embryo, with scale keratin (D1) and claw keratin (D2) with a shorter glycine-rich sequence. These keratins form irregular corneous bundles (D3) as for beak keratin (D4). (D5) The short feather-keratin where the glycine-rich regions have disappeared but corneous bundles parallel (D6). One possibility is that an ancestral, crocodilian-like protein gave rise (short arrow on the left) to proteins present in avian scales, claws and beak and, eventually, feathers. The other possibility (long arrow on the right) is that feather-keratin was directly derived from a large deletion of the glycine-rich region from an ancestral protein, and that a specific evolution gave rise to the feather-specific sequence. According to immunological studies (see Fig. 2) the feather-epitope may be very ancient in arcosaurian (embryonic) epidermis but is later eliminated in scale, claw or beak proteins. Color codes: Glycine (red), Serine (yellow), Cysteine (green), Proline (blue).

Turtle proteins resemble those of archosaurian scales and claws, both crocodilians and birds (Figs 9–11, Dalla Valle et al. 2009). Among these characteristics are the richness of tyrosine and the high sequence identity of the core-box region and in surrounding amino acid regions. Therefore, based on sKAPs in the epidermis, turtle, crocodile and birds appear to share a common progenitor, and this reinforces the recent evidence that turtle and archosaurians taxonomically are much closer to each other than to lepidosaurs (Rest et al. 2003).

Some regions of these proteins show sequence identity with mammalian KAPs (Alibardi et al. 2006a) that are present in hard keratinized appendages (hairs, claws, horns, etc., see Gillespie, 1991; Marshall et al. 1991; Powell & Rogers, 1994). However, in comparison with mammalian high glycine-tyrosine proteins (7–13 kDa, see Gillespie, 1991), high glycine-proline-tyrosine proteins of turtle (HGPTs) are larger proteins (12–17 kDa).

The above differences between lepidosaurian and archosaurian/chelonian proteins are also emphasized by the further detection of different amino acid regions, here provisionally indicated as pre- and post-core-boxes, that characterize archosaurians and chelonian vs. lepidosaurian beta-keratins (Fig. 11).

In summary, the central core-box of 20 amino acids contains two beta-strands and has 75–90% identity within archosaurian/chelonian KAPs, whereas a pre-core-box of 17 amino acids, indicated as archosaurian/chelonian-box, has also high identity (Dalla Valle et al. 2007a,b, 2008 and 2009). These values fall to 60–75% when comparing the core-box of crocodilian/chelonians with those of lizards or snakes. The latter contain a pre-core-box called the lepidosaurian-box (Fig. 11). The high sequence identity based on the core-box of their beta-keratins confirms that crocodilians and birds are the closest living archosaurians. Another difference between lepidosaurian and chelonian-archosaurian beta-keratins is that in the former the core-box is localized from halfway to the C-terminus of the protein, whereas in the latter the core-box is moved toward the head to the middle of the protein (Fig. 10). The specific role of these and of other regions of reptilian KAPs remains to be discovered.

Evolution and phylogenetic relationships among sauropsid KAPs

The recent knowledge on genes and protein sequences has addressed for the first time the long-debated problem of the evolution of these small proteins in reptiles and birds (Wyld & Brush, 1979, 1983; Maderson, 1985). A general phylogram obtained using the neighbor-joining method (Saitou & Nei, 1987), showing some examples of sauropsid KAPs, is presented in Fig. 15A. The phylogram shows the initial divergence between lepidosaurian and chelonian/archosaurian proteins, probably during the mid-upper Permian (Pough et al. 2001).

Figure 15 (.

A) Summarizing cladogram showing the affinities of some representative of sKAPs in all groups (see text for further details). Note in particular the initial separation between lepidosaurian and chelonian/archosaurian proteins, and the following separation of HGPS from HCPS in lepidosaurian. In chelonian/archosaurians note the early separation between feather vs. non-feather proteins. The scale bar indicates the percentage of distance. (B) Hypothetical scheme of the point mutations in ancestral genes responsible for the transformation of a small glycine-serine-rich protein into mammalian or sauropsid KAPs, or histidine-rich proteins in arcosaurians (see text for a more detailed explanation). Whereas in sauropsids similar mutations invested a region that originated the core-box (green) in the lineage of synapsid-mammals, the mutations invested another region (blue) from which no core-box was (generally) formed. Also note that the mutated gene might or not belong to that coding for the variable regions of cytokeratins, assumed to be phylogenetically more ancient than KAPs. Finally, in the mammalian line, the enrichment of cysteine in the V1 and V2 regions of cytokeratins may also have contributed to the formation of the E-regions of trichocytic keratins.

The molecular evolution in lepidosaurians apparently produced two main types of proteins, the HGPS and HCPS proteins (lepidosaurian KAPs), which are poor in tyrosine. These two subfamilies contain different members that will be more properly classified when they have all been characterized. The variety of types, in amino acid composition and molecular weight, are probably related to the specializations of the integuments of these modern reptiles (such as different types of scales over different body regions, soft and hard with keels, frills, gastrosteges of snakes for body movement, and claws). Some of the studies reported above have indicated that HGP proteins are present in relatively soft scales of lizards and snakes, while HCP proteins are present in claws and climbing setae.

Proteins of archosaurians/chelonians so far identified differ from those of lepidosaurians in that they are not divided into HGPS and HCPS proteins. Their general structure contains a tail region with HCPS characteristics and a tail region with HGPS characteristics, resembling a chimeric molecule (see Fig. 10). That typical diapsid (archosaurians) and presumed anapsid (chelonians) reptiles possess more similar proteins than the proteins of lizards and snakes (diapsids) was unexpected. The study of a current representative of a basal lineage of diapsid lepidosaurians such as Sphenodon would be very important to complete the whole picture on the evolution of these keratins (Alibardi & Toni, 2006b).

After this initial divergence, in archosaurians two main groups later diversified, that of tyrosine-glycine-rich non-feather proteins and the glycine-tyrosine-poor group of feather proteins (see also Gregg et al. 1983). In chelonian/archosaurian KAPs, the amount of tyrosine and glycine increases somehow in relation to the irregular orientation of the corneous material, which increases resistance (Brush, 1993). These regions were either lost (or never appeared) in the small and specialized proteins of feathers (Figs 13, 14).

Interestingly, the initial archosaurian group includes the proteins of turtle, suggesting that turtle proteins have differentiated from common ancestral proteins with those of archosaurs. Turtle proteins later diverged from crocodilian and avian proteins. Turtles have maintained a quite uniform typology of scutes in the shell, with marginals, coastal and neural, and sternoplastron, hyoplastron, hypoplastron, xyphiplastron, mainly changing their areas and thickness in different species. More variety was generated in crocodilian scutes, where tail keeled scutes, ventral softer scales, dorsal tough scales, etc., are present. The scale proteins of dorsal tough scales are the closest proteins to those of crocodilians. This supports the evolution of scale proteins in birds from an archosaurian protein (Figs 14, 15).

The molecular evolution of archosaurian proteins of scales, claws and ramphoteca, seems to have affected the C-terminus of their molecule, whereas the initial-mid part has remained more similar to that of their possible common ancestors. The molecular data indicate that archosaurian proteins initially possessed extended glycine-rich regions, as in the proteins of extant crocodilians (Fig. 15). Given the ancient lineage of crocodilians, which were present with similar features and adaptations 200 millions years ago, we may speculate that similar long glycine-rich tails were present in beta-keratins of the extinct Mesozoic archosaurians. Glycine-rich proteins may be suited for making hard scales by giving high resistance and hydrophobicity to these proteins. The glycine-rich tails in proteins present in extant crocodiles suggest that similar molecules might have been present in ancient archosaurians from which feathers are believed to be derived (Spearman, 1966; Maderson, 1972; Gregg et al. 1984). In this case, our data support the previous view on the reduction of the glycine-tails in beta-keratins suggesting an evolution from archosaurian scales to feathers (Spearman, 1966; Maderson et al. 1972; Molloy et al. 1982; Gregg et al. 1984; Fig. 14). However, our cladogram in Fig. 15 (upper panel) also shows that feather-keratins separated early from the proteins of avian and non-avian archosaurs, and the cladogram suggests that feather-keratins are very ancient and underwent an independent evolution from that of other skin appendages. This point is at present under detailed phylogenetic analysis (Dalla Valle et al. 2008).

The comparison between feather- and scale keratin showed a deletion of the glycine-rich region (52 amino acids) in feathers from the ancestral sequence present in scales (Gregg & Rogers, 1986). This observation supported the morphological conclusions of the evolution of feathers from scales (Gregg et al. 1984; Presland et al. 1989a,b). Another change between scale-claws-beak and feather-keratins was the appearance of a feather-specific epitope (TVVGSSTSAAVGSILSCEGVPINS, or similar, see Sawyer et al. 2005a). The role of this feather-specific amino acid sequence (underlined in Fig. 14) remains unknown.

The lack of the glycine-tail sequence in feather proteins may, however, also mean that scales, claws and beak beta-keratins evolved differently from those of feathers. The terminal part of this caudal region in feather proteins also contains cysteine residues, probably involved in intra-molecular and/or inter-molecular disulfide bonds for making a strong corneous material. The latter produce keratin bundles more regularly packed than in scales and claws, permitting the axial elongation of barb/barbule cells (Alibardi, 2002, 2006b).

Concluding remarks and future directions: comparison between the process of cornification in sauropsids and that in mammals

During evolution from reptile-like stem amniotes, one evolutive lineage of amniotes leading to mammals (the theropsids) produced mainly non-fibrous keratin-associated proteins. The other lineage of amniotes, the sauropsids, leading to modern reptiles and birds, instead produced fibrous keratin-associated proteins (the ‘beta-keratins’). In particular in the lineage of archosaurian reptiles from which birds originated, a small type of fibrous protein, feather-keratin, was produced and its characteristics permitted the formation of barbs and feathers (Fig. 15A).

As both sauropsids and theropsids probably possess glycine-rich proteins it is hypothesized that some mutations in genes of the ancestral short glycine-rich protein gave rise to HGT proteins in theropsids and HGS in sauropsids. This gene may be derived from that of variable regions of cytokeratins, or by a divergent evolution from a similar gene later incorporated into that of cytokeratins to form their V-regions (Fig. 15B). In sauropsids, the mutation of the first or the second base of coding triplets determined the replacement of serine with proline, cysteine or tyrosine in specific regions of the ancestral protein, giving rise to the core-box and to cysteine-rich regions and tyrosine-rich regions. In theropsids, similar mutation in other parts of the ancestral protein might have produced the HGT and HS/UHS proteins, as indicated in Fig. 15B. Also, in the evolutive lineage from which mammals originated (synapsids and therapsids) other mutations in the triplets of serine might have introduced cysteine in the variable region of cytokeratins, transforming this region in the E-regions of trichocytic keratins.

The increasing knowledge of the nucleotide and amino acid sequences of ‘beta-keratins’ in reptiles, will allow understanding of the molecular evolution of these proteins in archosaurians, chelonians, and lepidosaurians, from common stem reptiles. After a core-box was formed in stem reptiles, where a mechanically resistant and scaled integument was likely present, two main lineages of evolution of beta-keratins began, the one of arcosaurian/chelonians and the other of lepidosaurians (Fig. 15A).

The regional differences in scale keratins vs. claw, beak, and feather-keratins will clarify the specific role of these specialized beta-keratins for the appropriate skin appendages in different species of reptiles. Further studies are needed to compare possible homologies between some regions of sauropsid KAPs with mammalian KAPs to reveal which proteins may have similar roles in hard derivatives of sauropsids and mammalians. The latter goal will be fully realized when reptilian genomes and proteomes are sequenced in representatives of lizards, snakes, crocodilians and turtles.

In conclusion, a broad generalization of the process of cornification in the skin of all vertebrates can be made. In all skin derivatives, the hard material results from the interaction between two main protein components, fibrous keratins and matrix or associated proteins. The latter are now known in amniotes, where they are represented by sKAPs and mKAPs, whereas no molecular information is yet available for KAPs of fishes and amphibians. The latter topic would be a fruitful field of investigation for future studies.


Our studies have been financed with grants from the Universities of Bologna and Padova, and largely by self-support (L. Alibardi). Dr. Vania Toffolo helped in the initial study on lizard beta-keratins. This paper is dedicated to Prof. G. E. Rogers (University of Adelaide, Australia) for his fundamental studies on the sauropsid keratins, for his influence on the start-up of our molecular studies on reptilian beta-keratins, and for checking the language of the manuscript.