Scales of fish arise from mesoderm

The trunk of fish is covered by a large variety of morphologically and structurally diverse skeletal elements, such as scales, scutes and bony plates [1]. These elements are formed from intramembraneous ossifications and are part of the integumentary skeleton [2]. Histological and developmental similarities with neural crest-derived teeth in fossil and extant vertebrates, have led to the widely accepted notion that scales and fin rays, which are thought to be a scale modification, primarily derive from neural crest and not from mesodermal sources as the majority of the post-cranial skeleton [2]. Although shortand long-term labeling experiments in zebrafish have suggested a neural crest origin of fin rays [3,4], the contribution of neural crest to the post-cranial integumentary skeleton, and in particular to the scales, has not been thoroughly analyzed. By Cre/loxPbased genetic labeling, transplantation experiments and transposon-mediated clonal analysis, we demonstrate a mesodermal origin of scale-forming osteoblasts. Furthermore, our data do not support an extensive, if any, neural crest contribution to the post-cranial integumentary skeleton. Zebrafish (Danio rerio) possess an integumentary skeleton in the form of elasmoid scales — a highly derived scale type that protects the bodies of most teleost fish [1,5]. To investigate a possible neural crest origin for scale osteoblasts, we used a recently developed sox10:ERT2Cre driver line that allows permanent labeling of neural crest-derived cells in zebrafish [6]. In particular, induction of Cre-mediated recombination in Tg(sox10:ERT2-CRE;bactin:switch) embryos leads to robust marking of cranial skeletogenic tissues that are known to be derived from neural crest. By contrast, we failed to obtain substantial labeling of post-cranial skeletal elements in adults (n = 150), with the exception of a small number of scale clones (n = 31) (Figure 1A–J). Within these scales, we found labeled osteoblasts at the margin, along the grooves of the posterior field (Figure 1A), and on the outer surface, along the growth ridges of the anterior field (Figure 1A,B) and in the posterior field (Figure 1C; Supplemental data). In rare cases, the sox10 promoter used drives transgene expression in future larval and adult somatic muscle fibers (Figure 1D,E). Moreover, the frequency of labeled scale clones (one in every five fish; f = 0,2) is much lower than the frequency of labeled clones within most neural crest-derived cell lineages (for example, there are five pigment cell clones per fish, f = 5). We thus sought to verify whether labeled scale osteoblasts were clonally associated with other labeled tissues, in particular with mesoderm-derived skeletal muscles (Figure 1F–J). Three scale clones (10%) were not associated with other labeled tissues (Figure 1F), 19 (61%) presented exclusive association with labeled muscles (Figure 1G), 7 (23%) were spatially related with both labeled neural crest derivatives and muscles (Figure 1H), and two (6%) were associated exclusively with neural crest derivatives, such as pigment cells (Figure 1I). As in 84% of our clones labeled scales have been found spatially associated with labeled muscles, we conclude that scaleforming osteoblasts do not derive from neural crest, but from sox10:ERT2Cre expressing cells that are also the progenitors of somatic muscles and have thus a mesodermal origin. In order to independently validate our finding, we used an additional transgenic line driving ERT2-Cre expression in the early paraxial mesoderm [Tg(SA1-mCT2aC#HB)] [7]. After embryonic induction (16–48 hpf), we did not observe labeled cells in any of the well-established neural crestderived structures. By contrast, we detected numerous labeled scales (n = 30) clonally associated with labeled, mesoderm-derived muscles and blood vessels in adult fish (n = 50) (Figure 1K–N; Supplemental Information). Remarkably, we found the same cell populations that have been detected after sox10-mediated induction. Within the scales, we found marked cells in the dermis at the anterior margin and osteoblasts along the serrations, on the scale surface, in the grooves and at the scale rim (Figure 1K–L). The spatial association of labeled scales and somatic muscles (Figure 1M–N) and the absence of labeled neural crest-derived tissues as expected from the expression profile of the driver line, strongly support a shared, mesodermal origin for these structures. Further cross-validation has been carried out by means of blastula transplantations of ß-actin:GFP labeled cells into wild-type embryos (Figure 1O,O’) and transposon-based clonal insertion of a vector expressing ubiquitously the DsRed reporter gene (Figure 1P,P’; Supplemental Information), which permits a finer control of clone size. In both cases, the scale–muscle association was confirmed (77% and 89%, respectively). In the light of the ‘new head’ theory, which views neural crest as the source for many vertebrate-specific traits, and on the basis of similarities between dental elements and the ancestral scale type, the entire set of post-cranial integumentary skeletal elements was thought to be neural crest-derived [2]. Recent work supported this view with respect to fin rays [4]. However, we here show that teleost scales are predominantly derived from mesoderm. Despite their diversity, the multiple types of scales and skeletal components of the integument are considered to share an evolutionary origin, the odontode, and use similar developmental mechanisms [1,8,9]. However, a general lack of structural details at the tissue level impedes confirmation of homologies between different integumentary skeletal elements. Elasmoid scales are quite divergent from the ancestral condition, which has persisted almost unchanged in placoid scales of cartilaginous fish [8]. Although it is generally accepted that a reduction led to the loss of some components, the precise identity of the mineralized layers that persisted in the elasmoid scales is debated. In particular, the presence of true dentine, which in the teeth is a product of neural crest-derived odontoblasts, is not widely accepted [5,8]. Our data point to a mesodermal origin of scale-forming osteoblasts in zebrafish and do not support a neural crest contribution to scales in teleosts. However, our analysis does not exclude that more ancestral types of scales with a structure closer to teeth such as placoid scales of cartilaginous fish may include neural crest-derived tissues.