Recurrent functional divergence of early tetrapod keratins in amphibian toe pads and mammalian hair

Abstract

Amphibians have invaded arboreal habitats multiple times independently during their evolution. Adaptation to these habitats was nearly always accompanied by the presence or appearance of toe pads, flattened enlargements on tips of fingers and toes that provide adhesive power in these environments. The strength and elasticity of the toe pad relies on polygonal arrayed cells ending in nanoscale projections, which are densely packed with cytoskeletal proteins. Here, we characterized and determined the evolutionary origin of these proteins in the toe pad of the tree frog Hyla cinerea. We created a subtracted cDNA library enriching genes that are expressed in the toe pad, but nowhere else in the toe. Our analyses revealed five alpha keratins as main structural proteins of the amphibian toe pad. Phylogenetic analyses show that these proteins belong to different keratin lineages that originated in an early tetrapod ancestor and in mammals evolved to become the major keratin types of hair. The ancestral keratins were probably already expressed in areas that required skin reinforcement in early tetrapods, and subsequently diverged to support fundamentally different adaptive structures in amphibians and mammals.

1. Introduction

Many frogs and toads (Anura) are capable of sticking with ease to extremely slippery surfaces, such as vertically hanging leaves in trees or moss-covered rocks in torrents. These abilities are achieved by the complex fine structure of toe pads, flattened enlargements on tips of fingers and toes that provide adhesive power by maintaining a good fit with the substrate [1]. The flexibility rises from a highly specialized and conformable epidermis, consisting of a polygonal cell array separated by channels and containing thousands of nanoscale projections on top of these cells. In wet conditions, the channels between the cells facilitate the drainage of water from the toe pads, allowing direct contact between toe pad and substrate [1], whereas in dry conditions, they assist in the dispersal of mucus secreted by glands that open into these channels. The mucus secretions contribute to the adhesive power by a process called wet adhesion, i.e. adhesion by a combination of capillary and viscous forces of the mucus [2,3].

To endure the forces caused by the repetitive contact with the substrate, toe pad cells are filled with cytoskeletal proteins that form a rigid backbone. The projections on top of these cells are packed with cytoskeletal proteins that reach deep into the cell and merge to form a dense matrix, perpendicular to the zone of contact [4]. The typical organization of these proteins is essential both in withstanding mechanical stress and in adjusting to surface irregularities. Despite multiple morphological and histological studies [1,2,4–8], the major structural proteins making up the complex amphibian toe pad remain largely unknown.

To search for these structural proteins, we prepared a subtracted cDNA library of genes expressed in the toe pad, but not in the adjacent toe. We subsequently used phylogenetic analyses to reconstruct their origin and examined expression patterns to determine toe pad specificity.

2. Material and methods

A detailed description of this section is provided in the electronic supplementary material, S1. We created a subtracted cDNA library of genes expressed in the toe pad (complete distal part) versus genes expressed in the same digit minus the toe pad of the tree frog Hyla cinerea. Suppression subtractive hybridization was performed using a PCR-Select Subtractive Hybridization Kit (Clontech). Amplification products were cloned into electrocompetent cells and colonies were picked randomly for sequencing. Expressed sequence tags (ESTs) were assembled using Codoncode Aligner (Codoncode corporation) and screened using BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Relevant toe pad ESTs were extended by rapid amplification of cDNA ends (RACE) PCR, using cDNA created with Smart PCR cDNA Synthesis Kit (Clontech) as PCR template, with one gene specific and one adaptor-specific primer. RT-PCRs used to quantify relative gene expression differences between tissues are described in the electronic supplementary material, S1. Primers are listed in the electronic supplementary material, S2.

To investigate the evolutionary origin of newly identified toe pad keratins (TPKs; see §3), we added them to a representative dataset of vertebrate and outgroup keratins [9]. Type I and type II keratins were analysed separately, using only the central and most conserved helical region. Keratin datasets were aligned using ClustalX [10] and manually corrected in MacClade v. 4.06 [11] (provided as nexus files in the electronic supplementary material, S3 and S4). Phylogenetic relationships were estimated on amino acid sequences using three model-based methods, including heuristic maximum-likelihood (ML) searches and bootstrapping, metapopulation genetic algorithm (MetaGA) and Bayesian inference, as implemented in RAxML [12], MetaPIGA v2 [13] and MrBayes v. 3.1 [14], respectively. Because the Bayesian analyses converged on the JTT + G + I model of AA substitution using a mixed model prior, we used this model for ML and MetaGA analyses (see the electronic supplementary material, S1).

3. Results and discussion

Screening for genes that are specifically expressed in the toe pad yielded 200 unique sequences. None of the ESTs from the ‘reverse’ subtraction experiment (ESTs in the toe but not in the toe pad) corresponded to ESTs from the ‘forward’ subtracted library, indicating an effective enrichment of toe pad genes. Similarity searches showed that alpha keratins make up approximately 33 per cent of ESTs (with a BLAST e-value < 10⁻⁵) in the cDNA library and revealed that these proteins constitute the cytoskeletal backbone of amphibian toe pads. Extension of the partial keratin sequences using RACE PCR yielded five different alpha keratin transcripts (GenBank accession nos KC110649–KC110653).

The three phylogenetic methods, performed on two datasets containing toe pad and other vertebrate alpha keratin sequences (type I and type II separately), yielded similar gene trees consistent with recent hypotheses of keratin diversification in vertebrates [9] and supported similarly divergent origins of the different TPKs in the phylogeny (see figure 1 and electronic supplementary material, S5). Although lineage-specific keratin gene duplications significantly extended the gene repertoires of amphibians in both type I and type II alpha keratins (e.g. Sil 2 to Sil 8 in type I and Sil 26 to Sil 34 in type II; figure 1), none of our retrieved toe pad sequences are closely related to these keratins. Instead, TPKs are recovered in divergent keratin clades, several of which each include a different mammalian keratin type expressed in the hair follicle. Hence, the closest amphibian relatives of the keratins that determine the structure of the mammalian hair fibre, inner root sheath and outer root sheath are expressed in the anuran toe pad.

Figure 1.

Maximum-likelihood phylogenetic inference of toe pad keratins in (a) type I and (b) type II tetrapod keratin evolution. Branches interrupted by two parallel lines correspond to half the length of that branch. Black filled squares represent RAxML bootstrap support greater than or equal to 75% and Bayesian posterior probability greater than or equal to 0.95. MetaGA analyses yielded a similar consensus tree but with low branch support (see the electronic supplementary material, S5). Human hair keratins (red), inner root sheath keratins (IRS; green) and outer root sheath keratins (ORS; blue) are shown in bold. Coloured boxes indicate the smallest inclusive clade that contains the most recent ancestor of a defined human keratin type and its closest related amphibian keratins, and all of its descendants. Gene names represent the species name followed by the number of their position in the genome [9], except for human keratins that follow [15].

Three out of five TPKs fall within the well-supported clades of the so-called ‘hair keratins’ [16,17] of type I (TPK1) and type II (TPK2 and TPK3) (figure 1, red boxes). The human orthologues of these keratins (i.e. the closest related human keratins regardless of their function; figure 1, red boxes, indicated in bold) form the main components of hair fibres, but are also found in other integument appendages like nails [16]. Our semi-quantitative RT-PCRs suggest that expression of these toe pad keratins in H. cinerea is restricted to digits (figure 2). TPK2 and TPK3 show an elevated expression level in the distal parts of digits, but are not restricted to the toe pad. In contrast, TPK1 is the only keratin that was specifically retrieved from the toe pad. To investigate whether this keratin is toe pad-specific or merely expressed distally in amphibian digits, we performed a similar RT-PCR to check for expression of the orthologue in the African clawed frog Xenopus laevis (figure 2, indicated as TPK1*). Our RT-PCR indicates that TPK1* in this species is expressed distally in toes with claws, but not proximally in the same toe, or distally in clawless toes. This pattern suggests that expression of this keratin is tightly linked to the presence of digital appendages. Several orthologous keratins in anole lizards (figure 1, Ano 7, Ano 25) are also mainly expressed in claws, whereas others (figure 1, Ano 8, Ano 23) are known to have a wider expression pattern, including skin and tongue [18].

Figure 2.

Semi-quantitative RT-PCR. Band intensity is an estimate of the relative difference in keratin gene expression between tissues. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a standard. (a) H. cinerea: all toe pad keratins (TPK1 to TPK5) were confined to tissues containing epidermis, and were not expressed in muscle, kidney or tongue tissue. (b) X. laevis: TPK1*, the orthologue of TPK1 in Hyla and Sil 14 in Silurana tropicalis, is only expressed in the distal parts of toes with claws.

Our phylogenetic analyses provide strong support for TPK4 being the orthologue of Sil 20 in Silurana, a keratin for which no functional data are available. In H. cinerea, TPK4 is expressed in all tissues that contain skin but shows no specificity for digits (figure 2). The amphibian keratins cluster with the human outer root sheath keratins K14, K16 and K17 (figure 1, blue box, indicated in bold), which are expressed in the outer layers of the hair follicle. Except for the expression of K14 in the most basal cell layers of the skin, these three keratins mainly exhibit expression patterns in skin appendages (e.g. nails and hairs) and in mammal-specific glands [16]. The restricted expression pattern of K16 and K17 in mammary glands and sweat glands must have evolved along the mammalian stem lineage. We, therefore, hypothesize that the ancestral tetrapod keratin of the outer root sheath clade had an expression pattern similar to those of human K14 and amphibian keratin TPK4, i.e. in the skin and possible associated appendages.

Keratin TPK5 clusters (with limited support) in the group that includes the human keratins K24 (no known function in mammals), K10 (expressed in the skin) and inner root sheath keratins K25 to K28 (figure 1, green box). The expression of the human inner root sheath keratins is confined to hair sensu stricto (e.g. not expressed in nails) [16]. Our RT-PCRs indicate that toe pad keratin TPK5 of H. cinerea is only expressed in digits (figure 2).

Altogether, our study indicates that five alpha keratin genes are highly expressed in the amphibian toe pad. These keratins are not amphibian-specific innovations, but are orthologous to keratins of mammalian hair follicles and, therefore, originated in early tetrapods. In contrast to other cytoskeletal elements, keratins keep their structural backbone even when cells die [19], and leave a network of cross-linked proteins that is essential for the strength of these epidermal structures [20]. The ancestral keratins were probably already expressed in areas that required skin reinforcement in early tetrapods, and subsequently diverged to support fundamentally different adaptive structures in amphibians and mammals.