Previously, TEM revealed that erythrophores can be divided into two types: type 1 with carotenoid vesicles and type 2 with markedly larger erythrosomes (up to 1 μm in diameter)4. In the present study, the differences between the two erythrophore types or states were additionally confirmed using TEM and CLEM, which revealed that the inclusion sizes significantly differ (p < 0.001) between the two cell types. However, the present study has revealed that type 1 erythrophores can contain rare larger inclusions/vesicles (with diameters of up to 1 µm), which closely resemble the erythrosomes found in type 2 erythrophores (which, however, have diameters of up to 3.5 µm), and that type 2 erythrophores contain also some smaller inclusions. These findings blur the boundary regarding cell inclusion size, which was the main criteria for distinguishing between the two different erythrophore types in our first study4. Therefore, we here question the validity of defining two types of erythrophores and rather propose describing them as cell subtypes or differentiation states11. To this end, the term subtype, instead of type, is hereafter used to differentiate between erythrophores originating from either black or red spots of brown trout skin.
Subtype 2 erythrophores seem to be stabilized and firmly integrated into red spots by fibroblasts or fibroblast-like cells, which were well connected via numerous desmosomes as well with subtype 2 erythrophores. Subtype 2 erythrophores in red spots were better connected with cell junctions than subtype 1 erythrophores in black spots. This was noticed also during sample preparation for CLEM, where four subtype 2 erythrophores have remained connected whereas subtype 1 erythrophores were always solitary. Moreover, in red spots, subtype 2 erythrophores were surrounded by a prominent network of collagen fibres that presumably hinder the migration of erythrophores out of the red spots.
Multiple genes were found to be upregulated in the red spots compared with other skin regions9. With a more detailed analyses of our previous results and an additional assessment of differential gene expression using qPCR, the present study showed a unique pattern of gene expression in the red spots of brown trout, which are characterised by subtype 2 erythrophores.
Although we speculate that the analysed genes are chromatophore-specific, it must be emphasized that the expression was analysed using the whole skin sample, including the epidermis and dermis with chromatophores and surrounding cells. Thus, both the type of chromatophore and specific homo- and hetero-typic communication between chromatophores and their surrounding cells might lead to specific expression profiles for particular skin regions.
Among the tested genes upregulated in the red spots, ednrba, mitfa, fzd9b, and mc1r are involved in melanogenesis; pax7a and scarb are involved in xanthophore differentiation and carotene metabolism; and sox10 is involved in chromatophore differentiation from neural crest cells12,13,14,15. The lineage relationships between erythrophores (particularly subtype 2) and other cell types in salmonids are unknown. However, our results imply a very specific gene expression profile for erythrophores, with overexpressed genes associated with melanophores and partially xanthophores. Xanthophores have never been found in red spots, whereas individual melanophores are present above the erythrophore layer (4 and this study). Interestingly, mitfa, pax7a, and sox10 are essentially involved in the differentiation of pigment cells and can also be expressed in pigment cell precursors/progenitors in zebrafish12,16,17. A common chromatophore precursor or stem cell has been postulated18 and has been suggested to be located in the dorsal region of adult zebrafish19. It was also demonstrated that sox10 acts cell-autonomously in all pigment cell lineages20. Furthermore, it was hypothesised that in zebrafish, melanoblast- and xanthoblast-specific markers, such as mitfa, csf1r1, xdh, and pax7, are co-expressed in a sub-population of dorsal chromatophore precursors16. Interestingly, many of the genes that were identified in zebrafish as characteristic of chromatophore precursors were co-expressed and upregulated in erythrophore-containing red spots in brown trout.
Erythrophores have rarely been described in salmonid species8 and are poorly characterised. Unlike xanthophores, which are present in all salmonid species and in all developmental stages, erythrophores are not found in all salmonid species nor in all stages of ontogenesis8,21. Interestingly, a similar situation was found in the Danio clade, for which the phylogenetic distribution of species with erythrophores suggests that this cell type may have independently appeared and disappeared multiple times22,23,24. Most adult Danio species have either xanthophores or erythrophores on their bodies23. Recently, Huang et al.24 described adult skin pigment pattern of D. albolineatus, where both cell types are present, but in distinct skin regions (bands) on body and fins, where they are the only pigment cell type present. Similarly, both cell types are present simultaneously also in brown trout, although they do never appear together (no direct cell–cell interactions exist between them4). While xanthophores are distributed across almost all skin regions in salmonids, erythrophores are mainly distributed within spots or lateral lines4,21.
Subtype 2 erythrophores are of special interest, since they are the predominant cell type in red spots, characterised by a particular cell ultrastructure and gene expression profile. However, our analyses could not determine the gene expression profile of subtype 1 erythrophores or xanthophores. These last two cell types are skin-region specific; however, they rarely undergo homotypic interactions but rather undergo heterotypic interactions with other cells (4 and this study). Some of the analysed genes were overexpressed also in black spots, in which subtype 1 erythrophores are present in a small proportion, compared with background colour regions. Together with two additional genes with almost identical expression patterns between black and red spots (gja5 and dct9), the gene analysis in the present study suggests at least partially similar expression patterns in both erythrophore subtypes. Nevertheless, the distinct expression profiles with highly upregulated genes in the red spots indicate a regulatory independence25 of subtype 2 erythrophores.
It has been shown already that pigment cells in zebrafish can differentiate into different cell types or states, based on their micro-environment5. Iridophores obtained a particular morphology (organization of guanine-reflecting platelets) in zebrafish stripes and interstripes during their differentiation that depended on the presence or absence of melanophores. The iridophores of interstripes and stripes also exhibited distinct transcriptomic states, indicating that they represent different cell types5. A similar model of cell differentiation could also be proposed for erythrophores in brown trout. We suppose that cell communication with melanophores and particular paracrine signalling might promote differentiation into subtype 1 erythrophores in black spots, while homotypic cell communication with paracrine/autocrine signalling promotes differentiation into subtype 2 erythrophores in red spots.
The role of the tissue microenvironment in pattern formation and maintenance has been highlighted multiple times26,27,28. To date, many studies have described the diverse mechanisms of cell communication between fish chromatophores and surrounding cells (and their influence on chromatophore distribution and pigment patterns), including gap junctions, potassium channels, long-range communication through cellular extensions (and Delta/Notch signalling), macrophage-mediated communication through airinemes, pseudopodia communication, and communication via diffusing signals (reviewed in27). Although many genes have been characterised as specific to subtype 2 erythrophores, these genes that are upregulated in the red spots of brown trout could also be differentially expressed in the cells surrounding the erythrophores, and this could influence chromatophore distribution and pigment patterning via yet unknown mechanisms of communication. Surrounding cells may also directly affect the differentiation state of the two erythrophore subtypes and subsequently their gene expression. For example, melanophores may affect subtype 1 erythrophores, which may consequently never reach the final cell differentiation state that is presumably present in red spots.
The key to the origin of a new cell type, subtype, or state is the regulatory independence of a cell; that is, the ability of a cell to regulate and evolve gene expression independently of other cells25,29,30. The process of gene duplication is believed to be a major factor contributing to the generation of novel gene functions within a genome31,32. Following genome duplication, the functions of duplicated genes can either be retained, lost, or can diverge. In teleost fish, pigmentation genes have been preferentially retained in duplicate after FSGD, so that teleost fish have 30% more pigmentation genes than tetrapods1,33. It has been shown that large parts of the melanophore and xanthophore pathways are present in two copies in teleost fish2,33. It has also been concluded that FSGD has made an important contribution to the evolution of teleost-specific features of pigmentation, which include novel pigment cell types or the division of existing pigment cell types into distinct subtypes33. Furthermore, salmonids experienced another whole-genome duplication (Ss4R) about 80 million years ago34, resulting in multiple paralogues of pigmentation genes in their current genome. Given the above and based on our results, we suggest that the duplication events FSGD and/or Ss4R may have also promoted the emergence of new cell subtypes in brown trout when duplicated genes or regulatory regions underwent neofunctionalization. Although, subfunctionalization could also be considered as regulatory fate for gene duplicates, neofunctionalization was suggested as predominant evolutionary fate of genes after the whole genome duplication events in salmonids35. Detailed analyses of mitf paralogue gene and protein copies (please see Supplementary Figs. S1–S4 and Table S1 online) are also supporting the hypothesis of duplicated gene neofunctionalizations.
As already mentioned, lineage relationships between erythrophores (regardless of their type, subtype, or differentiation state) and other cell types in salmonids are unknown. However, as we demonstrated, many differentially expressed genes in subtype 2 erythrophores (or in red spots) are involved in melanogenesis, which is characteristic of melanophores, and in xanthophore differentiation. Our detailed analysis of these genes and their transcripts revealed that two or even four paralogues are present in the brown trout genome; however, the expression of only one of them was analysed at first. Paralogues in salmonids are the consequence of two genome duplication events (FSGD and Ss4R). Thus, it is possible that erythrophores (subtype 2 or both subtypes) developed as a consequence of duplicated gene neofunctionalization, either from melanophores, xanthophores, or their common precursors/progenitors derived from neural crest cells. Differentiation into erythrophores might be further triggered by paracrine or autocrine signalling, promoting the expression of duplicated gene copies. Recently, Huang et al.24 showed that erythrophores in the fin of Danio albolineatus share a common progenitor with xanthophores and maintain plasticity in cell fate even after differentiation.
Nevertheless, similarities between the functions of melanophores and xanthophores in metabolic pathways and during development have also been demonstrated36,37. It was even proposed that functional melanin and coloured pteridine pathways are present in both melanophores and xanthophores, and that this could enable chromatophore transdifferentiation (i.e. the transformation of one type into another)8,38. Furthermore, mosaic pigment cells containing more than one type of organelle, and mosaic organelles containing more than one type of pigment can be found in vertebrate skin8,39. For instance, in frogs, a mosaic of chromatophores containing melanosomes, erythrosomes, xanthosomes, and refractosomes was observed40,41. This suggests evolutionary relationships between these cells and possible transdifferentiation of already differentiated chromatophores8. In vitro and interspecies transplantation experiments demonstrated that amphibian xanthophores and iridophores can transform into functional melanophores and vice versa38,42. These findings highlight the proximity of the melanin and pteridine pathways, suggesting that, in teleost fish, established melanophores may also differentiate into xanthophores or erythrophores, or that xanthophores can differentiate into erythrophores, and vice versa. The differentiation of xanthophores into erythrophores would require even fewer changes in the function of genes and metabolism in addition to requirements for several genes in red or yellow coloration, as recently demonstrated in D. albolineatus24. Similarly, studies on medaka (Oryzias latipes) showed that a changed mode of interaction between only two genes (Sox5 and Sox10) affects the specification of pigment cell types and, by affecting the xanthophore progenitor, correlates with the evolution of a novel leucophore pigment cell type43. Differential expression of one or more paralogues might thus promote the differentiation of chromatophore precursors to a novel pigment cell subtype or state, e.g. erythrophores in brown trout.
The expression profile of paralogue genes is shown to be only partially differentiated in differently pigmented brown trout skin regions. However, there is no clear distinction between the expressions of individual paralogues in differently pigmented skin regions, with the exception that multiple paralogues are overexpressed in red spots compared with other skin regions. These results still suggest that duplicate gene copies at least partially adopt new functions (e.g. mitfa coded on Ch:14, more data about mitf nucleotide and amino acid alignments and protein structure is presented in Supplementary information online) and contribute to the transformation of a new chromatophore subtype that is involved in red spot formation. It is hypothesised that erythrophores and/or surrounding cells in red spots partially preserve the expression profile of the precursor/progenitor pigment cell (i.e. sox10 expression, related to the findings of43 mentioned above) or that chromatophore precursors are present in red spots, generating subtype 2 erythrophores. An additional change in the expression of one of the duplicated gene copies might trigger the emergence of a new cell subtype (e.g. an erythrophore). Nevertheless, the surrounding cells, which were analysed together with chromatophores in differently pigmented skin regions, probably also play an important role in chromatophore positioning and pigment patterning. The specific expression profile and interactions of the surrounding cells may also contribute greatly to the transformation/differentiation state of a cell type or the emergence of new cell type/subtype.
In addition to the new hypothesis regarding cell subtype origin, chromatophore transdifferentiation as a mechanism of subtype 2 erythrophore origin in brown trout skin should also be considered. Transdifferentiation of chromatophores in amphibians has already been discussed above. An example of pigment cell transdifferentiation was also demonstrated in zebrafish44. Two classes of leucophores are characteristic of zebrafish skin, and one of them, melanoleucophores, develop directly from melanophores, accumulating white material and losing melanin. Using transcriptomes analyses of individual pigment cells during melanoleucophore development, the authors demonstrated that the differentiation of melanoleucophores from melanophores involves a switch from melanin to purine synthesis44.
The present study has opened a number of hypotheses regarding erythrophore origin and differentiation into two types, subtypes, or states in brown trout and has also provided an important basis for further analyses. Defining cell types and states and tracing their developmental origin requires single-cell assays11. Therefore, single-cell sequencing of all chromatophore types and subtypes, including other cells surrounding the chromatophores, needs to be performed in brown trout. Although single-cell approaches are revolutionizing our understanding of cell structure and gene regulation, it has to be noted that some information is lost when focusing on a single cell, as revealed also in this study using CLEM and in situ cell observations. Furthermore, observing pigment cell emergence and pattern formation during all life stages as well as cell (trans)differentiation in vitro and its dependence on interactions with other cells (i.e. subtype 2 erythrophores with melanophores) could be a promising approach to determine the appropriateness of classifying cells into types or subtypes.
