Molecular and Cellular Bases of Color Pattern Evolution in Cichlid Fishes

Molecular and cellular bases of color pattern evolution in cichlid fishes

Doctoral dissertation for obtaining the

academic degree

Doctor of Natural Sciences

submitted by

Liang, Yipeng

at the

Faculty of Sciences

Department of Biology

Konstanz, 2020

Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-2-rkd9iiw5z5k24

Date of the oral examination: 14th July 2020 1. Reviewer: Prof. Dr. Axel Meyer 2. Reviewer: Prof. Dr. Mark van Kleunen

⁕ Table of contents ⁕

Table of contents

Table of contents ...... i

List of Figures ...... iv

List of Tables ...... vii

Summary ...... ix

Zusammenfassung ...... xii

General Introduction ...... 1 Cellular and molecular basis of vertebrate color patterns ...... 1 Cichlid fishes as a model system for studying color pattern formation ...... 4 Objectives of the Thesis ...... 11 Chapter I Agouti-related Peptide 2 Facilitates Convergent Evolution of Stripe Patterns across Cichlid Fish Radiations ...... 12 Abstract ...... 12 Main text ...... 13 Materials and Methods ...... 20 Supplementary Text ...... 30 Chapter II Evolutionary Dynamics of Structural Variation at a Key

Locus for Color Pattern Diversification in Cichlid Fishes ...... 36 Abstract ...... 36 Introduction ...... 37 Materials and Methods ...... 37 Results ...... 45 Discussion ...... 55 Chapter III The Teleost-Specific Genome Duplication and Color Pattern Evolution in Modern Fishes: Retention of Functional Conservation, Neo-functional Divergence and Sub-functionalization of

Agouti Genes ...... 58 Abstract ...... 58 Main text ...... 59 Materials and Methods ...... 66 – i –

⁕ Table of contents ⁕

Chapter IV Neural Innervation as a Potential Trigger of Morphological Color Change and Sexual Dimorphism in Cichlid Fish

...... 67 Abstract ...... 67 Introduction ...... 68 Results ...... 70 Discussion ...... 82 Methods ...... 86 Chapter V Developmental and Cellular Basis of Vertical Bar Color

Patterns in the East African Cichlid Fish Haplochromis latifasciatus . 91 Abstract ...... 91 Introduction ...... 92 Results ...... 93 Discussion ...... 104 Conclusion ...... 109 Materials and Methods ...... 110

General Discussion ...... 115 Main contribution ...... 115 Future and Outlook ...... 121

References ...... 123

Acknowledgments ...... 149 General acknowledgments ...... 149 Chapter specific acknowledgments ...... 150 Chapter I ...... 150 Chapter II ...... 150 Chapter III ...... 150 Chapter IV ...... 151 Chapter V ...... 151

Author Contributions ...... 152 Chapter I ...... 152 Chapter II ...... 152 Chapter III ...... 152 – ii –

⁕ Table of contents ⁕

Chapter IV ...... 153 Chapter V ...... 153

Supplementary Material ...... 154 Chapter I ...... 154 Chapter II ...... 172 Chapter III ...... 186 Chapter IV ...... 190 Chapter V ...... 201

– iii –

⁕ List of Figures ⁕

List of Figures

Fig. I.1 Convergent evolution of horizontal stripes across African cichlid radiations. . 14 Fig. I.2 Agrp2 controls stripe loss in Lake Victoria cichlids...... 16 Fig. I.3 Regulatory changes of agrp2 are predictive of convergent stripe evolution..... 17 Fig. I.4 A shared genomic basis of stripes across cichlid radiations...... 19

Fig. II.1 Sequence conservation at the agrp2 locus...... 45 Fig. II.2 Isoforms of agrp2...... 46 Fig. II.3 The third exon of agrp2 was tandemly duplicated...... 50 Fig. II.4 Expression bias towards exon 3a and resurrection of exon 3b ...... 52 Fig. II.5 Phylogeny of the agrp2 gene across cichlids...... 53 Fig. II.6 Protein sequence evolution of agrp2 across cichlid fishes...... 54 Fig. II.7 Summary of the evolutionary history of the agrp2 gene...... 55

Fig. III.1 Expression differences of asip1 and agrp2 associate with stripe pattern formation and evolutionary loss across teleost fishes...... 61 Fig. III.2 Evolution of agouti gene function across teleosts and tetrapods...... 64

Fig. IV. 1 Yellow and black morph of Melanochromis auratus...... 68 Fig. IV.2 Chromatophore measurements in scales of yellow and dark morph of M. auratus scales...... 71 Fig. IV. 3 Ultrastructural differences between skin regions of yellow and dark morph...... 74 Fig. IV.4 Differential gene expression analysis...... 77 Fig. IV.5 Increased axon innervation in the epidermis of the dark morph...... 81

Fig. V.1 Vertical bar patterns in Haplochromis latifasciatus...... 94 Fig. V.2 Cellular changes during vertical bar formation...... 96 Fig. V.3 Photographs of scale and skin dissections in bar and interbar region...... 98 Fig. V.4 Chromatophore measurements in adult H. latifasciatus...... 99 Fig. V.5 Expression differences of pigmentation candidate genes...... 101 Fig. V.6 Cellular and genetic mechanism underlying bar pattern formation in H. latifasciatus...... 104

Fig. S.I.1 Fine-mapping of Pnye / H.sau F2 recombinants ...... 154 Fig. S.I. 2 Protein sequences of QTL genes ...... 155

– iv –

⁕ List of Figures ⁕

Fig. S.I.3 Phylogeny of the agouti family...... 156 Fig. S.I.4 Expression of agrp2, its paralogs and neighboring genes of the fine-mapped interval...... 157 Fig. S.I.5 RT-PCR of agrp2, ß-actin, atp6V0d2, agrp1, asip1 and asip2...... 157 Fig. S.I.6 Differential expression along the anterior-posterior axis in bars and interbars (i.e. regions between melanic bars) of Pnye...... 158 Fig. S.I.7 Association-mapping uncovers an intronic 1.1kb interval (enh.a) with alternatively fixed alleles...... 158 Fig. S.I.8 The intronic target interval enh.a shows species-specific cis-regulatory activity...... 159 Fig. S.I.9 Graphical summary of the genetic basis of stripe maintenance and suppression and CRISPR-Cas9 knockout...... 159 Fig. S.I.10 CRISPR-Cas9 mutants genotypes and phenotypes...... 160 Fig. S.I.11 Photographs of species analyzed for gene expression differences of agrp2...... 161 Fig. S.I.12 Enh.a sequence across 29 East African cichlids...... 162 Fig. S.I.13 Summary of shared and non-shared variants in enh.a of Lake Victoria and Malawi cichlids...... 163 Fig. S.I.14 Skin expression in Pdem (Pseudotropheus demasoni) and Pcya (Pseudotropheus cyaneorhabdos)...... 163 Fig. S.I.15 Association between agrp2 locus genotype and stripe phenotype as quantitative measurement...... 164

Fig. S.II.1 5’ RACE results...... 172 Fig. S.II.2 3’ RACE results...... 173 Fig. S.II.3 PCR-based detection of the agrp2 exon duplication using primers DF1 and DR1...... 174 Fig. S.II.4 PCR-based detection of the agrp2 exon duplication using primers DF2 and DR2...... 175 Fig. S.II.5 Long-range-PCR-based detection of the agrp2 exon duplication using primers DF3 and DR3...... 175 Fig. S.II.6 Cladoglam of cichlids with a sequenced agrp2 locus and the likely origins of deletions and insertions...... 176

– v –

⁕ List of Figures ⁕

Fig. S.II.7 PCR-based confirmation of deletion 7 and 8 across Lake Victoria cichlids...... 176

Fig. S.IV.1 Chromatophores in light microscopical and TEM analysis...... 190 Fig. S.IV.2 Melanosome density from the black pigmented regions...... 190 Fig. S.IV.3 Polar chart of the angle of iridosomes in the MLS of yellow morph...... 191 Fig. S.IV.4 Distance between iridosomes and the thickness of iridosomes...... 191 Fig. S.IV.5 Chromatophores organization in skin of yellow and dark morph M. auratus...... 192 Fig. S.IV.6 Axon density on scales across dorsa-ventral axis...... 193 Fig. S.IV.7 Axon density on scales...... 193 Fig. S.IV.8 Immunofluorescence staining of axon in wild-type M. auratus...... 194

Fig. S.V.1 Vertical bar formation is consistent among all three assessed individuals. 201 Fig. S.V.2 Detailed photographs of lateral cluster...... 201 Fig. S.V.3 Changes in melanophore dispersal diameter over time, ...... 202 Fig. S.V.4 Pigment coverage of the three individuals...... 202 Fig. S.V.5 Chromatophore dispersal diameter in the three individuals...... 202 Fig. S.V.6 Chromatophore density in the three individuals...... 203 Fig. S.V.7 Relative expression of melanophore genes corrected by relative expression of mitfa...... 203

– vi –

⁕ List of Tables ⁕

List of Tables

Table IV.1 List of up-regulated genes in skin of dark morph ...... 79 Table IV.2 List of up-regulated genes in skin of yellow morph ...... 80

Table V.1 Selected candidate genes involved in coloration and pigment patterns in fish...... 102

Table S.I.1 Efficiency of gRNAs: Abbreviations: Seq. Clones ...... 165 Table S.I.2 Association tests for agrp2 Locus in Lake Malawi hybrid cross...... 166 Table S.I. 3 Pseudo R2 calculations for Malawi hybrid cross (* above 0.5) ...... 166 Table S.I.4 R2 calculations for Malawi hybrid cross ...... 166 Table S.I. 5 List of primers used in this study ...... 167

Table S.II.1 Position of deletions, the coding sequence (CDS) of agrp2 and a regulatory element (EnhA) of agrp2...... 177 Table S.II. 2 Primers used in this study ...... 178 Table S.II.3 Summary of species and evidence for exon 3 duplications...... 179 Table S.II.4 Repeats and deletions within the agrp2 locus...... 185

Table S.III.1 Relative expression of agrp2 in N. parilus vs. P. pulcher ...... 186 Table S.III.2 Relative expression of agrp2 in D. tinwini vs. D. rerio ...... 186 Table S.III.3 Relative expression levels of target genes in M. auratus...... 186 Table S.III.4 Relative expression levels of target genes in M. cyaneorhabdos...... 187 Table S.III.5 Relative expression levels of target genes in P. pulcher ...... 187 Table S.III.6 Relative expression levels of target genes in T. salvini...... 188 Table S.III.7 Relative expression levels of target genes in B. pi ...... 188 Table S.III.8 Relative expression levels of target genes in D. rerio ...... 189 Table S.III.9 Relative expression levels of target genes in two-striped species ...... 189

Table S.IV.1 Chromatophore measurement and results of the statistical tests (within morph and individual) ...... 194 Table S.IV.2 Statistical tests of chromatophore measurement (across morph) ...... 197 Table S.IV.3 Axon density and results of the statistical tests (within morph/sex and individual) ...... 198 Table S.IV.4 Statistical tests of axon density (across morph/sex) ...... 200

Table S.V.1 Chromatophore measurement and results of the statistical tests...... 203

– vii –

⁕ List of Tables ⁕

Table S.V.2 Melanin synthesis genes expression levels (divide by expression of mitfa) and results of the statistical tests...... 204 Table S.V.3 Relative expression levels of target genes and results of the statistical tests...... 205 Table S.V.4 Melanophore related genes expression levels (standardized by melanophore density)...... 205 Table S.V.5 Primer sequence for qPCR...... 206

– viii –

⁕ Summary ⁕

Summary

Animal coloration and pigmentation vary within and between species and play important roles in natural selection and sexual selection. The diversity and visibility of animal color patterns provide unique opportunities for exploring the processes of divergent and repeated evolution. Whether similar color patterns share common genetic and developmental mechanisms and how gene regulatory complexities translate into diverse color phenotypes are among the long-standing questions in the evolutionary developmental biology of animal color patterns. Considerable efforts have been made in a few model organisms, yet these questions are still just beginning to answer as the general insights into the molecular processes of phenotypic diversity require a wide taxonomic range. Taking advantages of their intrafamilial diversity in color patterns, experimental and genetical tractability and manipulability, the chapters of this thesis use cichlid fishes as an amenable system to integrative investigate the mechanistic basis of color pattern formation. Chapter I investigates the genetic basis underlying the convergent evolution of stripe patterns in cichlid radiations. The melanic horizontal stripes are a notable example of phenotypic convergence and have evolved many times across the cichlid adaptive radiations of Lake Victoria, Malawi, and Tanganyika. Genetic mapping and gene expression analyses suggest that regulatory changes of agouti-related peptide (agrp2), a teleost-specific agouti gene, account for this evolutionarily labile trait: high expression of agrp2 blocks formation of stripe, while low expression permits stripe presence. Experimental manipulation of agrp2 in nonstriped cichlid by CRISPR-Cas9 knockout results in stripes reconstitution further supports the link between this gene and stripe patterns. Yet, cis-regulatory mutations are only predictive of stripes in Lake Victoria, while additional genetic modifiers may exist in species from other lakes, suggesting independent regulatory mechanisms underlying a shared genomic basis (regulatory changes of agrp2) of stripes across East African cichlid adaptive radiations. Chapter II comprehensively characterizes mutational and structural changes at the agrp2 locus and illustrates how these genomic features may promote the functional divergence of agrp2. Using combination of comparative genomic approaches, 5’ and 3’RACE, RNA-sequencing, TaqMan probe assays and molecular evolutionary analyses, several unknown isoform, duplications, insertions and deletions in agpr2 were uncovered.

– ix –

⁕ Summary ⁕

In particular, a tandem duplication of the last exon of agrp2 is described and can date the age to 8-12 million years that likely predates the East African cichlid radiations. Variation in presence/absence patterns and copy numbers between and even within species are observed in this duplication among all African cichlid radiations, suggesting this tandem duplication may facilitate neofunctionalization or even functional degeneration of agrp2, thus serve as a source of incipient diversification in color patterns of cichlid fishes. Striped patterns and dorsa-ventral countershading in cichlids are similar to those seen in the radiation of teleost fishes. In vertebrates, members of the agouti gene family, including agrp2 and agouti-signaling protein 1 (asip1), have been repeatedly linked to the evolution of these two color patterns. Chapter III specifically investigates the functional conservation and diversity of agpr2 and asip1 in color patterns presence- absence and formation across teleost fishes. Comparative gene expression in striped– nonstriped species pairs, together with results from Chapter I, support that regulatory changes of agrp2 act as global molecular switches controlling the presence-absence of stripe patterns across the teleost fish radiation. Spatial expression differences of agrp2 and asip1 across dorsa-ventral axis suggest that the paralogs may have complementary functions in stripe-interstripe patterning. In addition, the asip1 expression patterns are strikingly constrained and conserved across vertebrate and associated with countershading and stripe patterns, suggesting a functional convergence of this gene. Finally, this study gives insights into the functional conservation and sub- and neofunctionalization that occurred during the evolution of agouti gene family and highlight the evolutionary consequence in teleost color patterns. Another remarkable feature of some species is that they change their coloration during ontogeny or even as adults. Color changes are dramatic in some cichlid species, e.g. the Malawi golden cichlid (Melanochromis auratus) which has a prominent sexual dimorphism in coloration. While females and subordinate males are bright yellow with two dark horizontal stripes (yellow morph), dominant males undergo a drastic morphological change in coloration and become dark with two light blue horizontal stripes (dark morph). Chapter IV describes the cellular basis and transcriptional differences between the two morphs. Using light microscope and transmission electron microscope imaging, pronounced changes in cellular and ultrastructural levels are detected, which include changes in chromatophore distribution, pigment dispersal, and

– x –

⁕ Summary ⁕ organelle properties. The transcriptomic analysis uncovers beside differentially regulated pigmentation genes, also several neural genes that are highly expressed in the skin of the dark morph. These results are further supported by immunohistochemistry staining as the density of neuronal fibers increases in the dark morph compared with the yellow morph. Therefore, this chapter gives novel insights into the cellular, physiological and genetic basis of color change in cichlid fishes. Although progress has been made to identify the genetic basis underlying the diversification of cichlid color patterns, the developmental aspects of color phenotypes in cichlid fishes have been understudied. Chapter V investigates the development and cellular and transcriptional correlates of vertical bar patterns of Lake Victoria cichlid Haplochromis latifasciatus. By tracking the pattern formation, the macroscopic appearance of vertical bars is caused by increased differentiation of melanoblasts and accumulation of melanin within bar regions during the first three weeks. In adults, melanic bars are characterized by a higher density of melanophores, which is associated with increased expression of melanin synthesis genes. The marker genes of xanthophore and iridophore are not differentially expressed showing rather homogenous distribution of iridophores and xanthophores across the trunk. Such complement results suggest that the vertical bars of H. latifasciatus are formed through dynamic control of melanophore characteristics and therefore provide a baseline for studying the molecular and cellular properties that contribute to the development of color patterns of cichlid fishes. In summary, this thesis demonstrates how genes generate environments that repress/permit and shape the color pattern formation across the cichlid and teleost radiations, while independent regulatory and structural mutations of the same genes may serve as a source of diversification in color patterns. The detailed morphological description of cichlid color pattern from this thesis, on the other hand, provides a conceptual illustration of how pigment organelles and chromatophores in certain environments acting as paints on a palette to color the body of the fishes. Overall, the integrative analyses within this thesis provide both deeper and broader insights into the developmental and genetic mechanisms underlying color pattern evolution in cichlid fishes.

– xi –

⁕ Zusammenfassung ⁕

Zusammenfassung

Die Färbung und Pigmentierung von Tieren variiert stark sowohl innerhalb als auch zwischen verschiedenen Arten und spielt eine wichtige Rolle innerhalb der natürlichen und sexuellen Selektion. Die Diversität und Sichtbarkeit von Farbmustern in der Tierwelt bieten eine einmalige Möglichkeit um die Prozesse von divergenter und wiederholter Evolution zu erforschen. Die Fragen, ob ähnliche Farbmuster gemeinsame genetische und entwicklungsbiologische Grundlagen teilen und wie komplexe regulatorische Netzwerke von Genen die diversen Farbphänotypen formen, gehören zu den seit langem bestehenden Fragen der evolutionären Entwicklungsbiologie. Viele Studien bezüglich dieser Fragen wurden bereits mithilfe von klassischen Modelorganismen durchgeführt, trotzdem ist unser Verständnis der Vorgänge nur rudimentär. Für ein vollständiges Bild sind Studien in weiteren taxonomischen Gruppen nötig. Aufgrund ihrer hohen Diversität an Farbmustern und der relativ einfachen genetischen und experimentellen Manipulation wurde im Rahmen dieser Doktorarbeit das System der Buntbarsche (Cichlidae) benutzt, um die mechanische Grundlage der Formierung von Farbmustern integrativ zu erforschen. Kapitel I handelt von der genetischen Grundlage konvergent evolviertet Streifenmuster in verschiedenen Buntbarschradiationen. Melanistische, horizontale Streifenmuster sind ein bemerkenswertes Beispiel konvergenter Evolution und sind in den Buntbarschradiationen der Seen Victoria, Tanganjika und Malawi mehrfach entstanden. Genetische Kartierung und die Analyse von Genexpression lassen vermuten, dass regulatorische Modifikationen des Genes agouti related peptide (agrp2), ein Gen spezifisch für echte Knochenfische (Teleostei), verantwortlich für die evolutionäre Labilität dieses Phänotyps sind: eine erhöhte Expression von agrp2 verhindert die Bildung von Streifenmustern, während ein niedrige Expression die Formierung von Streifen erlaubt. Die experimentelle Manipulation des agrp2 Gens in einer nicht gestreiften Bunterbarschart durch Knockout mithilfe von CRISPR-Cas9 resultierte in einer Wiederherstellung von Streifen, was ein weiterer Beweis für den Zusammenhang zwischen diesem Gen und Streifenmustern ist. Jedoch erlauben cis-regulatorische Mutationen nur Schlussfolgerungen für Streifen im Viktoriasee, während zusätzliche genetische Modifikatoren in Arten aus anderen Seen vorhanden sein können, welche unabhängige Regulationsmechanismen zeigen.

– xii –

⁕ Zusammenfassung ⁕

In Kapitel II werden Mutationen und strukturelle Veränderungen von agrp2 charakterisiert und es wird gezeigt, wie diese genomischen Veränderungen die funktionale Divergenz von agrp2 begünstigen können. Mithilfe einer Kombination von komparativer Genomik, 5’ and 3’RACE, RNA Sequenzierung, TaqMan Analyse und Methoden der molekularen Evolution wurden mehrere unbekannte Isoforme, Duplikationen, Insertionen und Deletionen von agrp2 entdeckt. Besonders herauszustellen ist eine Tandemduplikation des letzten Exons von agrp2, deren Entstehung auf etwa 8 bis 12 Millionen Jahre vor unserer Zeit datiert werden kann und die somit wahrscheinlich älter als die ostafrikanischen Buntbarschradiationen ist. Das Vorhandensein dieser Duplikation variiert sowohl zwischen als auch innerhalb verschiedener Buntbarscharten aller afrikanischer Radiationen, was darauf hindeutet, dass die Tandemduplikation Neofunktionalisierung oder sogar die funktionelle Degenerierung von agrp2 begünstigt und somit als eine der frühen Quellen der Diversifizierung von Farbmustern in Buntbarschen gedient haben könnte. Streifenmuster und dorsoventrale Konterschattierung können nicht nur in Buntbarschen sondern auch in anderen Knochenfischradiationen gefunden werden. Schon vorangegangene Studien konnten Mitglieder der agouti Gen Familie [auch agrp2 und agouti-signaling protein 1 (asip1)] mit diesen Farbmustern in anderen Vertebraten Gruppen in Verbindung bringen. Im Rahmen von Kapitel III wird die funktionelle Konservierung und Diversität der Gene agrp2 und asip1 bei der Bildung von Farbmustern in mehreren Gruppen der echten Knochenfische (Teleostei) untersucht. Die vergleichende Analyse von Genexpression in Paaren von gestreiften und nicht gestreiften Arten unterstützt, zusammen mit den Ergebnissen aus Kapitel I, die These, dass regulatorische Veränderung von agrp2 als ein umfassender Mechanismus dient, um das Vorhandensein von Streifenmustern in echten Knochenfischen zu steuern. Räumliche Unterschiede in der Expression von agrp2 und asip1 entlang der dorsoventralen Achse lassen vermuten, dass die beiden Paraloge komplementäre Funktionen bei der Bildung der Streifen- und Zwischenstreifenregionen haben. Außerdem zeigt asip1 ein auffällig konserviertes Expressionsmuster in verschiedenen Wirbeltieren und wurde mit Konterschattierung und der Bildung von Streifenmustern assoziiert, was eine funktionelle Konvergenz dieses Gens nahelegt. Diese Studie gibt Einblicke in die funktionelle Konservierung und die Sub- bzw. Neofunktionalisierung, die im Laufe der

– xiii –

⁕ Zusammenfassung ⁕

Evolution dieser Genfamilie geschehen ist und zeigt die Folgen für die Evolution von Farbmustern in echten Knochenfischen. Ein weiteres bemerkenswertes Phänomen ist die Veränderung der Farbmuster mancher Arten im Laufe der Ontogenese oder sogar im adulten Alter. Die Änderung der Farbmuster können in manchen Buntbarscharten dramatisch sein, so zum Beispiel in der Art Melanochromis auratus, die einen deutlichen Sexualdimorphismus zeigt. Während die weiblichen und untergeordneten Männchen Individuen helle gelbe Farben durchbrochen von zwei schwarzen horizontalen Streifen (gelber Farbmorph) zeigen, durchlaufen die dominanten Männchen einen drastischen morphologischen Wandel, der sie fast vollkommen schwarz nur durchbrochen von zwei hellblauen horizontalen Streifen (dunkler Farbmorph) werden lässt. Kapitel IV beschreibt die zellulären Grundlagen und die transkriptionellen Unterschiede zwischen beiden Farbmorphen. Mithilfe von Lichtmikroskopie und Transmissionselektronenmikroskopie konnten Veränderungen auf dem zellulären und strukturellen Level entdeckt werden, z.B. Veränderungen der Chromatophor Verteilung, der Pigment Ausbreitung und der Organelleigenschaften. Neben mehreren unterschiedlich exprimierten Farbgenen zeigt die Analyse der Transkriptomdaten mehrere hoch exprimierte neurale Gene in der Haut des dunklen Morphs von M. auratus. Diese Ergebnisse werden unterstützt durch immunhistologische Färbungen, die eine erhöhte Dichte von Axonen im dunklen Morph verglichen mit dem gelben Morph zeigen. Dieses Kapitel ermöglicht neue Einblicke in die zelluläre, physiologische und genetische Grundlage der Farbveränderung in Buntbarschen. Obwohl Fortschritte in der Erforschung der genetischen Grundlage der Diversifizierung der Farbmuster in Buntbarschen gemacht wurden, gibt es kaum Studien, die die entwicklungsbiologischen Aspekte der Bildung von Farbmustern in Buntbarschen, untersuchen. Kapitel V beschäftigt sich mit den entwicklungsbiologischen, zellulären und transkriptionellen Aspekten von vertikalen Balkenmustern in der Buntbarschart Haplochromis latifasciatus aus dem Victoriasee. Durch die detaillierte Aufnahme des Entwicklungsprozesses, konnte festgestellt werden, dass das makroskopische Erscheinungsbild der vertikalen Balken durch eine erhöhte Differenzierung von Melanoblasten und eine Anreicherung von Melanin in den Balkenregionen während der ersten drei Wochen entsteht. In adulten Individuen zeichnen sich die melanistischen Balken durch eine höhere Dichte von Melanophoren, die mit einer erhöhten Expression

– xiv –

⁕ Zusammenfassung ⁕ von Genen der Melaninsynthese korrelieren, aus. Zugleich zeigen die Markergene für Iridophore und Xantophore keine differentielle Expression und beide Zelltypen sind gleichmäßig über den gesamten Körper der Tiere verteilt. Die Ergebnisse zeigen, dass die vertikalen Balkenmuster in H. latifasciatus durch die dynamische Kontrolle von Melanophoren geformt werden, und bilden eine Grundlage für die weitere Erforschung der molekularen und zellulären Eigenschaften, die zur Evolution von Farbmustern in Buntbarschen beitragen. Diese Doktorarbeit demonstriert, wie verschiedene Gene die Bedingungen für die Bildung und Unterdrückung von Farbmustern in Buntbarschen und anderen echten Knochenfischen schaffen, während unabhängige regulatorische und strukturelle Mutationen derselben Gene als Grundlage für die Diversifizierung der Farbmuster dienen können. Weiterhin bietet die detaillierte morphologische Beschreibung der Farbmuster in Buntbarschen eine konzeptionelle Darstellung davon, wie pigmentgefüllte Organellen und Chromatophore unter bestimmten Bedingungen als Farbrepertoire zur Einfärbung des Fischkörpers dienen können. Die integrative Analyse dieser Arbeit vermittelt ein breiteres und gleichzeitig ein tiefergehendes Verständnis der entwicklungsbiologischen und genetischen Mechanismen, die der Evolution von Farbmustern in Buntbarschen zugrunde liegen.

– xv –

⁕ General Introduction ⁕

General Introduction

Colorations and pigment patterns visible in animal skin, fur, hairs, feathers, and scales are fascinating to humans. For biologists, the roles of animal color patterns in the “survival of the fittest” (Darwin 1859, Spencer 1896) and the “arrival of fittest” (De Vries 1904, Wagner 2014) are the fundamental aspects of investigation. Since more than a hundred years, the ultimate causes underlying color patterns have been studied and the significance of color patterns (Cott 1940), including concealing coloration (Wallace 1877, Jablonski and Chaplin 2010, 2017), nuptial coloration (Darwin 1888, Seehausen, Mayhew, and Van Alphen 2001), mimicry (Poulton 1890, Mallet and Gilbert Jr 1995), countershading (Thayer 1909, Kapp et al. 2018) and aposematism (Mayr 1999, Ruxton et al. 2019) have been emphasized. Animal color patterns also provide excellent models for interdisciplinarity theoretical studies, for example, the discovery of Turing mechanisms (Turing 1952) that propose the self-organizing reaction-diffusion has received the most attention (Nakamasu et al. 2009, Volkening and Sandstede 2018). Yet, only until the last one to two decades we have the privilege of seeing the proximate causes underlying color patterns — the genetic and cellular mechanisms that drive the development and evolution of the color patterns diversification (Hoekstra and Coyne 2007, Manceau et al. 2010, Irion and Nusslein-Volhard 2019, Patterson and Parichy 2019).

Cellular and molecular basis of vertebrate color patterns

Pigment cells in vertebrates

In vertebrate, similar color patterns appear in both related and divergent taxa. For example, the stripe patterns could be found in birds (Haupaix et al. 2018), mammals (Mallarino et al. 2016) and fishes (Kratochwil et al. 2018). Dorsa-ventral countershading, which is crucial for background matching, UV protection and thermoregulation (Stevens and Merilaita 2011), is one of the most common color patterns in the animal kingdom. Such color patterns, as well as all range of coloration and pigmentation patterns are primary caused by the interaction of light with pigments that selectively reflect and/or absorb certain wavelengths (pigment-based coloration) and structures interfering with

– 1 –

⁕ General Introduction ⁕ light (structural coloration). In vertebrates, pigment cells are the cells responsible for creating color patterns. They are derived from a population of pluripotent cells that call the embryonic neural crest cells. In birds and mammals, melanocyte is the only pigment cell type that can synthesize pigments (black/brown eumelanin and yellow/red pheomelanin) and transfers melanin granules (melanosomes) to developing feathers and hairs. The yellow-orange-red colors are obtained either by the combination of pheomelanin and eumelanin (Hill and Mcgraw 2006) or by carotenoids from dietary sources (Mcgraw et al. 2003, Mcgraw et al. 2004). Blues, greens, whites and iridescent colors that also ordinary showed in homeotherms are caused by certain nanostructure in dermis (Prum and Torres 2003, Prum and Torres 2004) or feathers (Doucet et al. 2006, Hill and Mcgraw 2006, Shawkey and Hill 2006). In contrast to homeotherms, poikilotherms develop multiple types of pigment cells (chromatophores), including melanophores, xanthophores, erythrophores, iridophores, leucophores, and cyanophores (Burton and Burton 2017). Melanophores, which are homologous to melanocytes, synthesize melanin and storage the pigment in melanosomes. Alternations in melanization, melanosome dispersal and melanophore numbers can therefore change the darkness of integument (Cal, Suarez-Bregua, et al. 2017). Yellow-to-red hues are given by a range of pteridine and carotenoid pigments from (yellow-orange) xanthophores and (red) erythrophores (Matsumoto 1965, Bagnara 1966). The cyanophore is a rare pigment cell type that bears blue pigments (Fujii 2000). Iridophores (responsible for blue coloration and iridescent shine) that contain guanine crystals and leucophores (responsible for whites and cream) that contain hypoxanthine crystals are categorized as structural chromatophores (Takeuchi 1976). Besides the variation in cell types, chromatophores of poikilotherms storage their pigment and structure in organelles and are visible in the whole lifetime of the organisms—which makes poikilotherms a good model for investigating the cellular and developmental mechanisms of color pattern formation (Manukyan et al. 2017, Woodcock et al. 2017, Liang et al. 2020).

Molecular pathways underlying vertebrate coloration and pigmentation

The formation of animal color patterns requires a complex pigmentation network that includes multiple molecular signaling pathways and hormone systems. Proper pigment biosynthesis and pigmentation patterning guiding by this network are needed for the

– 2 –

⁕ General Introduction ⁕ overall appearance of color phenotypes. Modulations in any of these processes may cause changes in coloration and pigmentation. For example, changing the process of melanin synthesis by melanocytes/melanophores can lead to substantial differences in darkness over the entire body of an animal. In several species, mutations in the coding region of Melanocortin 1 receptor (Mc1r) are associated with modulation of melanism (Eizirik et al. 2003, Candille et al. 2007, Gross, Borowsky, and Tabin 2009, Rosenblum et al. 2010). Especially in lizard and cavefish, light body coloration that caused by reduction in melanin content and/or melanophore number has crucial adaptive functions and is linked to independent mutations in the coding regions of Mc1r (Rosenblum, Hoekstra, and Nachman 2004, Gross, Borowsky, and Tabin 2009, Rosenblum et al. 2010). However, the underlying functional mechanisms may vary between mutations: the derived mutation in the little striped whiptail (Ancylolomia inornata) dismisses the signaling activity of Mc1r, while a different mutation in the eastern fence lizard (Sceloporus undulatus) disrupts the integration of Mc1r into the melanophore membrane (Rosenblum, Hoekstra, and Nachman 2004, Rosenblum et al. 2010). Color patterns like the orange blotch of guppies, the stripes of zebras, and the eyespots on peacock's tails result from regional differences in number and properties of pigment cells and in synthesis and deposition of pigments. For instance, several studies revealed that regulatory variation in Agouti signaling protein (Asip/asip1, gene name in tetrapods/teleosts), a member from Agouti gene family that can inhibit Mc1r activity, is underlay changes in dorsa-ventral patterning of vertebrates by inhibiting melanin synthesis and/or decreasing melanophore number in ventral areas (Manceau et al. 2011, Linnen et al. 2013, Ceinos et al. 2015). On the other hand, Asip has been also linked to stripe-interstripe patterning in rodent and galliform, where high expression in the light areas between melanic stripes might locally inhibit dark pigmentation and thereby shaping the pattern (Mallarino et al. 2016, Haupaix et al. 2018). Spatial expression of Aristaless-like homeobox 3 (Alx3) in rodents can shape stripes by directly repressing Microphthalmia-associated transcription factor (Mitf), a master regulator of melanocyte differentiation, thereby giving rise to light-colored hair in regions between stripes (Mallarino et al. 2016). Mutation in colony stimulating factor 1 receptor a (csf1ra) of zebrafish (Danio rerio) results in lacking xanthophore thus develop only disorganized stripes anteriorly (Parichy et al. 2000, Parichy and Turner 2003). Tremendous efforts have also been made to identify the causal genes and loci underlying the development

– 3 –

⁕ General Introduction ⁕ and evolution of dorsa-ventral patterning, horizontal stripes, and vertical bars, blotches and spots (Mills and Patterson 2009, Manceau et al. 2010, Irion and Nusslein-Volhard 2019, Patterson and Parichy 2019), the molecular events that governing pigmentation patterning thus have seen considerable progress in recent years, yet still poorly understood.

Cichlid fishes as a model system for studying color pattern formation

To study the mechanistic basis of color pattern development and evolution it is advantageous to focus on organisms that are experimentally and genetically tractable. Teleost fishes are well suited because of their striking beauty and diversity of color patterns, while inspection of pigment patterning and experimental manipulation during development are possible. In particular, the East African cichlid fish adaptive radiations, with their over 1200 species, are a prominent example for repeated, convergent evolution (Meyer 1993, Stiassny and Meyer 1999, Muschick, Indermaur, and Salzburger 2012). One of the best-known features of cichlid fishes is their outstanding intrafamilial diversity in coloration and pigment patterns (Maan and Sefc 2013), which includes coloration ranging across all colors from yellow, orange, red, green, blue to violet hues as well as diverse pigmentation patterns including horizontal/oblique stripes, vertical bars, blotches, spots and complex and ununiformed patterns (Maan and Sefc 2013, Kratochwil et al. 2018, Kratochwil et al. 2019). Owing to the advent of novel techniques, hundreds of cichlid species genomes have been sequenced (Brawand et al. 2014, Malinsky et al. 2018). Protocols for gene expression analyses, including quantitative PCR (qPCR), RNA sequencing, in situ hybridization and immunohistochemistry are well established for cichlid fishes already (Henning et al. 2013, Woltering et al. 2018). A rich and expanding repertoire of experimental approaches including hybrid crosses, transgenesis and CRISPR-Cas9 genome editing (Kratochwil and Meyer 2015a, Juntti 2019) are also available for cichlid fishes. Together with the possibility of developmental inspection (Woltering et al. 2018, Liang et al. 2020), the above characteristics make cichlid fishes an excellent model for identifying the causal genetic variants, genes and molecular mechanisms that influenced evolution of and variation in color patterns. In the

– 4 –

⁕ General Introduction ⁕ following sections, I focus on the current understanding of the molecular and cellular processes that generate complex patterns and how alternation in those processed might produce different color phenotypes in cichlid fishes, and the mechanisms that underlying the convergence and diversification in coloration and pigmentation across teleost and vertebrate more generally.

Color pattern evolution associated with gene expression regulation

As previously mentioned, one of the genetic mechanisms that could drive differences in color patterns is a regulatory change that alters expression level (heterometry), expression pattern (heterotopy) or timing of expression (heterochrony) of the same gene (Jacob 1977, Arthur 2004). Such transcriptional regulation can be induced by the trans- and cis-regulatory variation, as well as variation in untranslated mRNA regions with potential regulatory effects (Gunter et al. 2011, Franchini et al. 2019). For example, a blotched pigmentation polymorphism (orange blotch, OB) can be found in some cichlid species from Lake Malawi and Victoria and mainly restricted to females (Lande, Seehausen, and Van Alphen 2001). The OB phenotype is characterized by dark melanophore blotches (with variation in number and size) on a background of xanthophores. Quantitative trait locus (QTL) mapping and association scans (Streelman, Albertson, and Kocher 2003, Roberts, Ser, and Kocher 2009) suggested the OB phenotype in Lake Malawi species is associated with paired box gene 7 (pax7), a transcription factor that critical for the development of the chromatophores (Minchin and Hughes 2008, Nord et al. 2016). Morphological inspection revealed a “fewer-but-larger” phenotype of melanophores in blotched female is correlated to up-regulation of pax7 expression, whereas no differences in pax7 coding sequence associated with OB phenotype, suggesting that cis-regulatory differences in pax7 account for the OB phenotype in Lake Malawi (Roberts, Ser, and Kocher 2009). Anal fin egg-spots are another evolutionary novelty in haplochromine cichlids, the most species-rich cichlid lineage that forms the adaptive radiations of Lake Victoria and Malawi (Salzburger et al. 2005). Previously, a TATA-box mutation upstream of the colony-stimulating factor 1 receptor (csf1r) was shown to be specific to the haplochromines (Salzburger, Braasch, and Meyer 2007). As csf1r is essential for recruiting xanthophore from precursor cells and exclusively express in egg-spots, it is likely that such cis-regulatory mutations together coding sequence mutations of csf1r

– 5 –

⁕ General Introduction ⁕ might associate with egg-spots evolution (Salzburger, Braasch, and Meyer 2007). Later, another two novel pigmentation genes, the a- and b-paralogs of four and a half LIM domain protein 2 (fhl2a and fhl2b) were highlighted by RNA sequencing in haplochromine cichlid Astatotilapia burtoni (Santos et al. 2014, Santos et al. 2016) and by qPCR in other species (Santos et al. 2016). In particular, comparative genomic analysis identified an insertion of an African cichlid short interspersed repetitive element (AFC-SINE) in the cis-regulatory region of fhl2b in egg-spot bearing haplochromines and functions as a specific enhancer for fhl2b expression in iridophores, the type of chromatophores that potentially establish the egg-spot development (Santos et al. 2014). Although efforts have been made to identify the casual loci that regulate the egg-spot development as above mentioned, the molecular and cellular mechanisms of the evolutionary innovation of this color pattern are not yet understood. Another striking example of convergent evolution in color patterns of cichlid fishes is the horizontal stripes patterns (Seehausen, Mayhew, and Van Alphen 2001). Based on the previous investigation (Henning et al. 2014), Chapter I of this thesis shows that a teleost-specific agouti gene facilitates convergent evolution of horizontal striped patterns across East African cichlid adaptive radiations. Regulatory changes of gene expression can act as molecular switches for gain and loss of stripe patterns during the evolution of this adaptive radiation. Interestingly, cis-regulatory mutations were detected but only associated with Lake Victoria cichlids, while additional genetic modifiers may exist in species from other lakes, indicating independent regulatory mechanisms underlying a shared genomic basis of stripes across cichlid radiations. Meanwhile, results from Chapter III suggest the role of this teleost-specific agouti gene in inhibiting stripe patterns is repeatedly coopted not only for the cichlid fish radiations, but also during teleost fish evolution.

Color pattern evolution and gene structure variation

One of the major sources of genetic variation is alternative splicing—the production of alternative RNA transcripts from a single gene (isoform) that can increase the coding capacity of the genome and thus might contribute to enhance the diversity in color patterns. In Peromyscus mice, the independent evolution of cryptic light pigmentation in different species is driven by a preference isoform of Agouti, a mammalian gene that responsible for the distribution of melanin pigment (Mallarino et al. 2017). In the

– 6 –

⁕ General Introduction ⁕ adaptive radiation of cichlids fishes in the East African Great Lakes, nine isoforms of hagoromo gene were detected and linked to species-specific color pattern characteristics (Terai et al. 2003). Structural changes (e.g. insertion, deletion and exon duplication) within a gene can also contribute to phenotypic evolution (Kratochwil and Meyer 2019, Xie et al. 2019). In particular, the variation of coat coloration and pigmentation in pigs can largely be explained by the structural variations with functional significance at a few major loci (Andersson et al. 2011), including duplications at the tyrosine-protein kinase KIT (KIT) locus (Giuffra et al. 2002, Rubin et al. 2012, Wu et al. 2019), deletion in the coding sequence of Tyrosinase-related protein 1 (TYRP1) gene (Wu et al. 2016) and insertion in MC1R (Jia et al. 2017). There are also instances in cichlid fishes that the color pattern dimorphisms are due to structural variations. Recent work on Lake Malawi cichlid Melanochromis auratus supports that the second exon loss in oculocutaneous albinism II (oca2) leading the trafficking of the Oca2 protein to melanosomes in amelanistic individuals, while the wild-type individuals with integrated oca2 sequence show the “normal” color phenotype (Kratochwil, Urban, and Meyer 2019). Other studies on oca2 of other organisms, including Mexican tetra (Astyanax mexicanus) (Klaassen et al. 2018), Medaka (Oryzias latipes) (Fukamachi et al. 2004) and corn snake (Pantherophis guttatus) (Saenko et al. 2015), illustrating how indels of pigmentation genes facilitate the evolution of color patterns. Although there are many examples for the dynamics and evolutionary significance of alternative splicing, insertion and deletion as above mention, exon duplications have received considerably less attention (Kondrashov and Koonin 2001, Keren, Lev-Maor, and Ast 2010). In particular, no evidence suggests the role of exon duplication in color pattern evolution. Chapter II of this thesis examined whether the above-mentioned teleost-specific agouti gene, a key locus for striped patterns in cichlid fishes, exhibits particular characteristics that facilitate the divergence in color patterns. Isoforms, insertions, deletions and tandem exon duplication were detected in this gene, and such structure variations may promote functional divergence of this locus.

Color pattern evolution and gene duplication

The remarkable diversity of color patterns and chromatophore types in fish is potentially the evolutionary consequence of whole-genome duplication, including the fish-specific

– 7 –

⁕ General Introduction ⁕ genome duplication (FSGD) that increased the number and properties of genes involved in color phenotypes (Braasch, Brunet, et al. 2009). For instance, the molecular signaling pathways and hormone systems of melanocyte/melanophore development and differentiation are very similar among vertebrates. Several gene members of the dark- pigmentation network have been detected a second gene copy in teleost fishes, e.g. the presence of two microphthalmia-associated transcription factor gene (mitfa and mitfb). It is a classic example of subfunctionalization (Ohno 1970) that the ancestral functions of mitf has been divided with mitfa in melanophore development and mitfb in the retinal pigment epithelium development (Lister, Close, and Raible 2001). Whereas additional instances of functional specialization after FSGD could be found in type I integral membrane protein silver (Schonthaler et al. 2005) and tyrosinase-related protein 1(typr1) (Braasch, Liedtke, et al. 2009) of teleost fishes. Another significant consequence of gen(om)e duplication is neofunctionalization (Van De Peer, Maere, and Meyer 2009). Sometimes, one of the paralogs specialized the ancestral functions in pigmentation while another paralog might evolve new functions: only Kit receptor tyrosine kinase a (kita) and kit ligand a (kitla) remain the function in melanophore development as the mammalian Kit ligand–receptor pair after FSGD, while kitb and kitlb have new unknown functions (Mellgren and Johnson 2005, Hultman et al. 2007). In cichlid fishes, the xanthophore linage marker gene csf1ra has a conserved function for pigmentation which is potentially the ancestral function, whereas neofunctionalization occurred in the B paralog of csf1r (csf1rb) (Braasch, Salzburger, and Meyer 2006), yet the function of csf1rb will need to be further addressed. Chapter III of this thesis focusing on the evolutionary history of Agouti gene family, a notable gene family that plays crucial roles in the evolution of vertebrate color patterns (Manceau et al. 2011, Haupaix et al. 2018, Kratochwil et al. 2018). By comparing the gene expression between and within species, Chapter III demonstrates the functional conservation and divergence of the Agouti gene family facilitated the evolution of color patterns in teleost fishes.

Molecular and cellular mechanisms of color change

Color patterns of some animals can change dramatically, whereas the time scales of the color change vary dynamically. Physiological color change (milliseconds to hours) is potentially triggered by neuronal and hormonal signals by changing the intracellular

– 8 –

⁕ General Introduction ⁕ distribution of pigments or reflective molecules. In contrast, changes that involve differences in pigment quantity or chromatophores number are referred to as morphological color change and occur relatively longer periods of time (hours to months) (Fujii 1993, 2000, Skold et al. 2016). A hundred years ago, the mechanism underlying the animal color change started to be investigated (Von Frisch 1910, 1911, Parker 1936, 1937, 1948), yet remain largely puzzling. The color change in some cichlid fishes is dramatic and could be influenced by social and other environmental factors. For example, males of the haplochromine cichlid Astatotilapia burtoni can change reversibly between bright blue and bright yellow overall body which serves as a social signal for mating (Hofmann and Fernald 2001, Korzan and Fernald 2006). It is reported that such polymorphism in coloration may be controlled by the melanocortin system, a neuroendocrine system: the yellowness of the fish body is modulated by regulating exogenous α-melanocyte-stimulating hormone (α-MSH) in dispersal of xanthophore pigments (Dijkstra et al. 2017). Examples of sex-independent color change can be found in the Central American Midas cichlid Amphilophus citrinellus—usually a small proportion of the individuals lose their typical dark pigmentation and obtain a “golden” orange sometimes also white, yellow or red coloration (Barlow 1976, Elmer, Lehtonen, and Meyer 2009, Henning et al. 2013). Although coding sequence mutations were found in melanocortin 1 receptor (mc1r), an important gene for melanin synthesis, yet the sequence polymorphism was not linked to the coloration of Midas cichlids (Dickman, Schliwa, and Barlow 1988, Henning et al. 2010). Ongoing efforts are underway to identify the genetic basis of the gold phenotype in Midas cichlids. Chapter IV of this thesis examines the sexual dimorphism in coloration of Malawi golden cichlid, Melanochromis auratus and revels the cellular and ultrastructural alternations that drive the morphological color change of this cichlid species. Complementing with gene expression analyses, an instructive role of nervous innervation is suggested to orchestrate the color change of cichlid fishes.

Developmental and cellular basis of color pattern formation

For the mechanisms of color pattern formation, the main model organisms for investigation are the zebrafish Danio rerio and the Medaka Oryzias latipes (Meyer, Biermann, and Orti 1993, Kelsh et al. 1996, Meyer, Ritchie, and Witte 1997, Nagao et

– 9 –

⁕ General Introduction ⁕ al. 2014). Especially the horizontal stripe patterns of zebrafish have obtained the most attention (Irion and Nusslein-Volhard 2019, Patterson and Parichy 2019). Recently, studies of Amphiprioninae, the anemone fishes (Salis et al. 2018, Roux et al. 2019) illustrated how iridophores with a specific cytological organization of iridosomes contribute to the white vertical bar pattern formation. Comparison between the development of vertical bars in anemone fishes and horizontal stripes in zebrafish highlights a potentially similar role for iridophores in establishment of pattern orientation, yet the mechanisms of pattern formation are likely different between species (Patterson and Parichy 2019). As previously mentioned, African cichlid fishes with their richness in color patterns are increasingly studied to understand the molecular mechanisms of color pattern formation. Although progress has been made identifying target genes and loci that drive evolutionary diversification in cichlids (Roberts, Ser, and Kocher 2009, Kratochwil et al. 2018) and their roles in adaptation and speciation (Seehausen, Mayhew, and Van Alphen 2001, Elmer, Lehtonen, and Meyer 2009, Maan and Sefc 2013), the developmental and cellular mechanisms of pigmentation phenotypes in cichlid fishes have been understudied. In the convict cichlid Amatitlania nigrofasciata, thyroid hormone signals control the timing of transition between embryonic-larval status and juvenile-adult status thus manipulation of thyroid hormone levels in the early development can lead to variation in vertical bar numbers of adults (Prazdnikov and Shkil 2019), highlighting the role of heterochrony (shifting in developmental timing) in evolution of color patterns. Interestingly, thyroid hormone signals in convict cichlid (Prazdnikov and Shkil 2019), flatfish (Yoo et al. 2000) and zebrafish (Mcmenamin et al. 2014) have similar functions in melanophore differentiation but not so pronounced in xanthophores and iridophores. Another recent study has described the melanic patterns in two Lake Malawi sand- dwelling cichlids, striped Dimidiochromis compressiceps and bared Copadichromis azureus (Hendrick et al. 2019). Despite the variation in pattern orientation, the pattern formation of both species is not driven by rearrangement of existing chromatophores which is the case of zebrafish (Parichy et al. 2009), but by the formation of new chromatophores (especially melanophores). Chapter V of this thesis focuses on the developmental and cellular basis of vertical bar patterns in a Lake Victoria cichlid fish Haplochromis latifasciatus. Oncogenic tracking, cytological quantifications and gene expression analysis

– 10 –

⁕ General Introduction ⁕ demonstrate a direct development of color patterns in cichlid fish (compare to the model organism zebrafish) through dynamic control of melanophore density, melanin synthesis and melanosome dispersal.

Objectives of the Thesis

The overall objective of thesis was to investigate the molecular, genetic and developmental mechanisms underlying the formation and evolution of color patterns in cichlid fishes. The specific objectives were to: 1. To determine the genetic basis underlying the convergent color patterns. Several color patterns, e.g. stripes, bars and spots evolved independently in the radiations of cichlid fishes. Evidences are needed to prove whether convergent color phenotype share similar genetic basis. 2. To characterize the source of genetic variation that contributes to color pattern evolution. Several genes that drive diversification of cichlid color patterns have been identified, yet the underlying mutational and structural changes are not fully detected. 3. To detailed describe how variations in the repertoires of chromatophores contribute to color patterns. Proper distribution, three-dimensional arrangement and properties of chromatophores are critical for the final appearance of color phenotypes and not fully clarified in cichlid fishes. 4. To investigate the transcriptional profile of different color patterns. The cellular correlates of color patterns are controlled by pigmentation networks. Comparative analyses on transcriptional level will be essential to reveal the molecular mechanisms of color pattern formation. The beautiful and diverse color patterns of cichlid fishes give us an unique opportunity to integrative investigate the mechanistic basis of color pattern formation. Understanding these molecular, genetic and developmental mechanisms will help us elucidate how cichlids and — more generally — organisms diversity.

– 11 –

⁕ Chapter I ⁕

Chapter I

Agouti-related Peptide 2 Facilitates Convergent Evolution of Stripe Patterns across Cichlid Fish Radiations

Claudius F. Kratochwil, Yipeng Liang, Jan Gerwin, Joost M. Woltering, Sabine Urban, Frederico Henning, Gonzalo Machado-Schiaffino, C. Darrin Hulsey, Axel Meyer Science (2018) DOI: 10.1126/science.aao6809

Abstract

The color patterns of African cichlid fishes provide notable examples of phenotypic convergence. Across the more than 1200 East African rift lake species, melanic horizontal stripes have evolved numerous times. We discovered that regulatory changes of the gene agouti-related peptide 2 (agrp2) act as molecular switches controlling this evolutionarily labile phenotype. Reduced agrp2 expression is convergently associated with the presence of stripe patterns across species flocks. However, cis-regulatory mutations are not predictive of stripes across radiations, suggesting independent regulatory mechanisms. Genetic mapping confirms the link between the agrp2 locus and stripe patterns. The crucial role of agrp2 is further supported by a CRISPR-Cas9 knockout that reconstitutes stripes in a nonstriped cichlid. Thus, we unveil how a single gene affects the convergent evolution of a complex color pattern.

One Sentence Summary Evolutionary gains and losses of stripe patterns in cichlid fishes are modulated by regulatory variation at the agrp2 locus.

– 12 –

⁕ Chapter I ⁕

Main text

Stephen Jay Gould famously posited that if it were possible to rerun the “tape of life,” outcomes would be different (Gould 1990). The relative importance of determinism and contingency during evolution is still far from settled (Losos et al. 1998, Losos 2017). But for particular groups of organisms, one can now test Gould’s hypothesis. For instance, in less than 8 million to 12 million years, more than 1200 species of cichlid fishes have evolved to form repeated adaptive radiations in the East African Rift Valley lakes, such as Lakes Victoria, Tanganyika, and Malawi (Fig. I.1) (Meyer et al. 1990, Kocher 2004, Turner 2007, Santos and Salzburger 2012, Brawand et al. 2014). These adaptive radiations have given rise to a large diversity of species displaying various color patterns (Fig. I.1C–N), including the repeated occurrence of melanic horizontal stripes (Fig. I.1A and supplementary text). Convergent evolution is prevalent in the East African cichlid radiations (Meyer 1993, Seehausen, Mayhew, and Van Alphen 2001, Muschick, Indermaur, and Salzburger 2012), providing a replicated natural experiment whereby distantly related species from independent adaptive radiations can be used to determine what mechanisms have generated these recurrent phenotypes (Kuraku and Meyer 2008, Protas and Patel 2008, Roberts, Ser, and Kocher 2009, Stern 2013, Kratochwil and Meyer 2015a). More specifically, we address whether horizontal stripes, a convergent phenotype, have an identical, similar, or different molecular bases across the independent adaptive radiations of cichlid fishes. Previously (Henning et al. 2014)—using a genetic mapping panel of two Lake Victoria species, Pundamilia nyererei (Pnye, nonstriped, Fig. I.1E) and Haplochromis sauvagei (Hsau, striped, Fig. I.1C)—we found that horizontal stripes (Fig. I.2C) are inherited as a recessive Mendelian trait mapping to chromosome 18 (Fig. I.2A). This was confirmed by a second cross involving the same nonstriped species and another striped species, H. chilotes (Hchi, striped) (Fig. I.2A and supplementary text). To more precisely isolate the causal genetic interval for stripe presence, we fine-mapped the trait using recombinant F2 individuals of the Pnye × Hsau cross and reduced the causal interval from 600 to 25 kb (Fig. S.I.1). This interval contained the genes agouti-related peptide 2 (agrp2), v-type proton ATPase subunit d 2 (atp6V0d2), and an unknown gene (unk) (Fig. I.2B). The resequencing of all coding regions revealed no fixed missense or

– 13 –

⁕ Chapter I ⁕ nonsense mutations (Fig. S.I.2), suggesting that cis-regulatory variation determines stripe presence. The teleost-specific agrp2 (Fig. S.I.3) is a strong candidate gene for stripes because its paralogs have been previously associated with pigmentation phenotypes (Zhang et al. 2010, Manceau et al. 2011, Ceinos et al. 2015). To test for agrp2 expression differences between nonstriped (Pnye) and striped (Hsau) Lake Victoria cichlids, we performed quantitative polymerase chain reaction (qPCR; Fig. I.2D and Fig. S.I.5) on a number of adult tissues, including skin (supplementary text). Here, agrp2 showed a significantly higher expression in the skin of Pnye (Fig. I.2D and Fig. S.I.4). The lack of consistent expression variation between melanic and nonmelanic regions and generally across dorsoventral and anterior-posterior positions suggests that agrp2 does not shape pigmentation patterns through local expression-level variation but rather acts as a general stripe pattern inhibitor (Fig. S.I.6). Whereas qPCR revealed no such expression differences for paralogs and neighboring genes (supplementary text), qPCRs on

Fig. I.1 Convergent evolution of horizontal stripes across African cichlid radiations. (A) Schematic of melanic horizontal stripes and vertical bars. (B) Map of the African Great Lakes Victoria, Tanganyika, and Malawi and superimposed simplified phylogenetic tree of the adaptive radiations of Lakes Tanganyika (green; not all paraphyletic lineages shown), Malawi (blue), and Victoria (orange). (C–N) Twelve of the focal striped and nonstriped species of this study, including species from Lakes Victoria (C–F), Tanganyika (G–J), and Malawi (K–N). Hlat, Haplochromis latifasciatus; Jorn, Julidochromis ornatus; Lcae, Labidochromis caeruleus; Maur, Melanochromis auratus; Nbri, Neolamprologus brichardi; Ncyl, Neolamprologus cylindricus; Tvit, Telmatochromis vittatus.

– 14 –

⁕ Chapter I ⁕

F2 Pnye/Hsau hybrid individuals confirmed that expression differences are linked to the agrp2 locus and exhibit an allelic dosage effect as expected for cis-regulatory mutations (Fig. S.I.4). To identify causal mutations affecting both agrp2 expression and stripe phenotype, we sequenced the agrp2 locus in individuals (n ≥ 10 individuals per population) from natural populations of the three hybrid-cross species. We screened for alternatively fixed, fully associated variants with the stripe phenotype in pairwise comparisons of each striped species (Hsau or Hchi) versus the nonstriped Pnye. Our analyses indicated that a 1.1-kb interval within the first agrp2 intron (Fig. I.2B) that exhibited shared, alternatively fixed alleles is a strong candidate region for a regulatory element controlling agrp2 expression (Fig. S.I.7). To test whether this 1.1-kb interval [enhancer of agrp2 in Pnye (Pnye.enh.a)] contains cis-regulatory elements that could influence interspecific differences between striped and nonstriped species, we tested the elements of both species in a green fluorescent protein (GFP) reporter assay in vivo (supplementary text). It showed that Pnye.enh.a efficiently modulates GFP expression [Tukey’s honest significant difference (HSD), P < 0.001] and is significantly more potent than the homologous sequence of the striped species Hsau (Tukey’s HSD, P < 0.001) (Fig. I.2E,F and Fig. S.I.8). Together, these results indicate that higher expression of agrp2, and thereby the suppression of stripe patterns, is indeed enhanced by Pnye.enh.a (Fig. S.I.9). Our results reveal agrp2 as a major determinant of stripe presence that might be sufficient to suppress stripe patterns in Pnye. To further test this finding, we used CRISPR-Cas9 genome engineering to manipulate agrp2 and to thereby potentially derepress stripe patterns. Pnye eggs were injected with Cas9 and agrp2 guide RNAs, and we obtained four mutants, all of which had nonsense and frameshift mutations within agrp2 (Fig. S.I.10 and Table S.I1). These CRISPR-Cas9 mutants developed a continuous midlateral stripe (Fig. I.2H and Fig. S.I10) yet no dorsolateral stripe (supplementary text). Because horizontal stripes were never observed in noninjected Pnye individuals (>100 observations; Fig. I.2G), this strongly suggests that although species such as Pnye have no stripes, the genomic and developmental machinery for stripe pattern formation is in place, and stripes can reappear in this nonstriped species by experimental manipulation of agrp2.

– 15 –

⁕ Chapter I ⁕

Fig. I.2 Agrp2 controls stripe loss in Lake Victoria cichlids. (A) In two mapping crosses between Pnye and Hsau and Pnye and Hchi, horizontal stripes map to the same region on chromosome 18. (B) F2 recombinant fine-mapping isolates a 25-kb interval containing agrp2. (C, D) Skin biopsies show differential agrp2 expressions (D) between skins from Hsau and Pnye (C). Error bars indicate means + SD. (E, F) An intronic 1.1-kb element of Pnye(Pnye.enh.a) is regulatory active (E) and shows stronger activity than the homologous sequence of the striped species Hsau (F) in a zebrafish larvae GFP reporter assay. (G, H) CRISPR-Cas9 mediated knockouts of agrp2 in the normally nonstriped cichlid Pnye (G) develop stripes [(H); midlateral stripe indicated by line]. LOD score, logarithm of the odds score; N/N, homozygous for the Pnye agrp2 allele; S/S, homozygous for the Hsau agrp2 allele; C/C, homozygous for the Hchi agrp2 allele; tg, transgenic construct; gRNA, guide RNA; mls, midlateral stripe; dls, dorsolateral stripe; d, dorsal; dm, dorsomedial; vm, ventromedial; v, ventral.

Next, we tested if the expression levels of agrp2 and stripe patterns are generally associated across other cichlid species from the repeated species flocks of Lakes Victoria, Malawi, and Tanganyika, suggesting a shared molecular basis for convergent stripe phenotypes. Using qPCR on adult skins of striped and nonstriped species of each of the three major East African cichlid radiations (in total, 24 species; Fig. S.I.11), we revealed that nonstriped species commonly had higher agrp2 expression levels than striped species (Fig. I.3B). This association was confirmed by comparative phylogenetic analyses (Fig. I.3A), demonstrating a significant evolutionary association between

– 16 –

⁕ Chapter I ⁕ low agrp2 expression and stripe presence [phylogenetic analysis of variance (ANOVA); mean P < 0.001; supplementary text]. To determine if this convergence at the phenotypic and agrp2 gene expression level is also paralleled at the sequence level (Stern 2013), we comparatively analyzed homologous enh.a sequences across cichlids from Lakes Victoria, Malawi, and Tanganyika. A tree of enh.a revealed substantial sequence variation and resolved striped Lake Victoria species as monophyletic, suggesting a single origin of the striped alleles, whereas striped species of other lakes were not monophyletic (Fig. I.3C). None of the nine mutations within enh.a that showed complete association with stripes in Lake

Fig. I.3 Regulatory changes of agrp2 are predictive of convergent stripe evolution. (A) Densitree representation of phylogeny for the 24 examined species, including divergence time estimates (Brawand et al., 2014). mya, million years ago. (B) Skin gene expression analysis of agrp2 across adaptive radiations highlights the strong evolutionary association of stripes with low agrp2 expression levels (phylogenetic ANOVA, P < 0.001). Error bars indicate means + SD. (C) A gene (locus) tree of the cis-regulatory element enh.a supports a single origin of striped alleles in Lake Victoria. Numbers present posterior probabilities > 0.9. Ajac, Aulonocara jacobfreibergi; Ceuc, Cheilochromis euchilus; Dcom, Dimidiochromis compressiceps; Hmel, Haplochromis melanopterus; Hser, Haplochromis serranus; Hthe, Haplochromis thereuterion; Jmar, Julidochromis marlieri; Jreg, Julidochromis regani; Lmul, Lamprologus multifasciatus; Mzeb, Maylandia zebra; Ncau, Neolamprologus caudopunctatus; Onil, Oreochromis niloticus; Pjoh, Placidochromis johnstoni; Ppun, Pundamilia pundamilia.

– 17 –

⁕ Chapter I ⁕

Victoria cichlids showed similar stripe association in cichlids of Lakes Malawi or Tanganyika (Fig. S.I.12, S.I.13). Consequently, independent mutations must be affecting agrp2 expression and thereby stripe patterns across the three major cichlid radiations (Fig. I.4D). Lastly, we tested whether the same locus is responsible for stripe-pattern variation outside of Lake Victoria cichlids using a hybrid cross between the nonstriped Lake Malawi species Pseudotropheus demasoni (Pdem; Fig. I.1M) and striped species Ps. cyaneorhabdos (Pcya; Fig. I.1K) that also differed in their skin agrp2 expression (Fig.

S.I.14). We obtained 270 F2 hybrid individuals that were genotyped at the agrp2 locus and phenotyped regarding their stripe patterns (Fig. I.4B,C; Fig. S.I15; and supplementary text). The results revealed significant linkage between the agrp2 allele and stripe presence (Fisher’s exact test, P = 7.6 × 10−8; Table S.I2). The allelic variation at the agrp2 locus explains more than 50% of the phenotypic variance in stripe patterns [Cox-Snell or Nagelkerke pseudo-R2 from ordered logistic regression; Table S.I.3].

Nevertheless, the phenotypic distribution of F2 individuals (49 nonstriped and 221 striped individuals) differed from the Mendelian 3:1 ratio observed in the Lake Victoria crosses (chi-square test, P < 0.001), providing evidence for additional minor modifier loci. These results strongly suggest that agrp2 acts as a major determinant of stripe pattern absence or presence in Lake Malawi (as in the younger Lake Victoria radiation), but additional minor stripe modifiers have evolved or were recruited in the older Lake Malawi radiation (Fig. I.4D). The repeated evolution of horizontal stripes in East African cichlid radiations is facilitated by cis-regulatory evolution of agrp2. Despite its described role and function in the brain (Zhang et al. 2010), we have discovered a hitherto unknown function for this gene in the skin that highlights notable functional similarities between Agrp2 and the mammalian Agouti (Asip) as well as teleost Asip1 (Manceau et al. 2011, Ceinos et al. 2015). From what is known about proteins of the Agouti family, Agrp2 likely acts as an antagonist for the melanocortin receptors Mc1r and/or Mc5r (Cal, Suarez-Bregua, et al. 2017). Low Agrp2 levels would trigger stripe melanophore proliferation, pigment dispersion, and/or pigment production (stripe patterns present), whereas high levels would block these processes (no stripe patterns) (Cal, Suarez-Bregua, et al. 2017). Expression levels of agrp2 thereby act as a switch controlling stripe presence and absence. In Lake Victoria, expression-level differences seem to be caused by several

– 18 –

⁕ Chapter I ⁕

Fig. I.4 A shared genomic basis of stripes across cichlid radiations. (A) Stripes are associated with agrp2 alleles in Lake Malawi hybrid F2 individuals (Pdem × Pcya). Inheritance is not Mendelian, suggesting that additional genetic modifiers exist in Lake Malawi cichlids. (B, C) F2 hybrids homozygous for the Pcya agrp2 allele (C/C, showing stripes, black lines) and Pdem allele (D/D, no stripes) exhibit clear stripe pattern differences. Parental species are shown in Fig. I.1K (striped) and Fig. I.1M (nonstriped). (D) Summary of the known (black) and unknown (gray) aspects of the genetic control of cichlid horizontal stripes.

mutations in a 1.1-kb intronic regulatory region (enh.a; Fig. I.2B) that push the expression of agrp2 levels above or below a threshold that determines the stripe phenotype. Such a threshold-based molecular on-off switch may have permitted the frequent loss as well as reevolution of stripes within East African cichlids. Although the presence of stripes appears to be controlled by differential expression of the same gene (agrp2), causal genetic variants must differ among the independent radiations of Lakes Victoria, Malawi, and Tanganyika (Fig. I.4D and Figs. S.I.12 and S.I.13). The intermediate phenotypes obtained from the Malawi cross (Fig. I.4A), together with the lack of the dorsolateral stripe in the CRISPR-Cas9 mutants (Fig. I.2H), provide evidence for additional modifier loci determining stripe presence (Fig. I.4D). However, those seem generally less prominent in the young (<15,000 years old) Lake Victoria radiation compared to the older (2 million to 4 million years old) Lake Malawi radiation (supplementary text). Regulatory variation of agrp2 provides a molecular basis for the repeated evolution and loss of stripe patterns across cichlid species flocks. Recurrent regulatory evolution at the agrp2 locus constitutes an example of regulatory tinkering (Kratochwil and Meyer 2015b, Mazo-Vargas et al. 2017) that might have facilitated the ease and speed of the evolution of both converged and diverged phenotypes that characterizes the East African cichlid radiations. The simplicity of such a threshold mechanism might have permitted the phylogenetically observed rapid losses and reevolutions of stripe patterns.

– 19 –

⁕ Chapter I ⁕

Therefore, Stephen Jay Gould’s predictions (Gould 1990) appear questionable at this evolutionary scale, and if one were to replay the evolution of cichlid adaptive radiations, the results might be surprisingly similar: striped and nonstriped cichlids evolving again and again through regulatory evolution at the agrp2 locus.

Materials and Methods

Experimental crosses and quantification of stripes

Victoria mapping panels were obtained by hybridizing a Pundamilia nyererei male and Haplochromis sauvagei female [hybrid cross 1, described in (Henning et al. 2014)] and a P. nyererei female and a H. chilotes male [hybrid cross 2, described in (Henning et al. 2017)]. The Lake Malawi species cross included a Pseudotropheus demasoni female and P. cyaneorhabdos male (hybrid cross 3). Species pairs were chosen, as they clearly differed in stripe patterns and gave fertile offspring in (Henning et al. 2017). Cichlids were obtained from commercial breeders and maintained by full-sibling mating in the animal research facility of the University of Konstanz. To initially form the crosses one male was kept together with five females of the other species. Upon successful fertilization, male and fertilized female were isolated. F1s were raised until sexual maturity and breeding groups were then assembled (6-9 month). Fry of the F1 hybrids

(F2 individuals) were isolated and raised under identical conditions. Horizontal stripes of hybrid cross 1 and 2 were treated as a binary trait (not striped 0, striped 1). Photographs were taken under standardized conditions. Briefly, individuals were anesthetized with MS-222 (Sigma) and subsequently photographed on a copy stand in lateral view. For hybrid cross 1 and 2, QTL mapping and map construction was conducted with R/qtl (Broman et al. 2003). Linkage groups were renamed to match the Oreochromis niloticus genome (Brawand et al. 2014) and therefore differ from previously used names (Henning et al. 2014). For this, we used the scanone function with a binary model. Chromosome-specific and genome-wide significance thresholds were determined using permutation tests (1000 permutations). Thresholds were set at a value of P = 0.05. The max. LODs on Chr.18 were >33 for Pnye/Hsau and >16 for Pnye/Hchi. For the Lake Malawi hybrid cross, as Pcya and Pdem are relatively dark and treatment with MS-222 disperses melanophores, the same type of documentation would have made

– 20 –

⁕ Chapter I ⁕ the scoring of stripe patterns quite unreliable. Therefore, Malawi specimens were photographed in a purpose-built chamber (small aquarium of 25cm length × 10cm height × 5cm depth within white photography tent with black ground) using a Canon 7D with a 100mm objective. Photographs were taken at 3-4 months of age, at a time when both of the parental species already exhibit their adult pigmentation patterns. Stripes in the

Malawi F2 were quantified in three different ways. First, as a binary trait with individuals completely lacking stripes scored as 0, all other individuals as 1 (221 non-striped; 49 striped individuals). Second, as an ordinal trait whereby we focused on the clearly defined midlateral stripe and scored individuals as 0 (non-striped; 49 individuals), 1 (partially striped or interrupted stripe; 132 individuals) and 2 (full and continuous stripe; 89 individuals). And third, a quantitative measurement, where we first used Fiji/ImageJ to draw a line from the operculum to the caudal peduncle and measured the percentage of this line that was melanic (called relative stripe coverage). We then plotted the log2 of the relative stripe coverage against the genotypes of the agrp2 locus. It has to be noted that the quantitative measurement (in contrast to the phenotyping as an ordinal trait) was influenced by the number and size of bars that were crossed by the line. Also, phenotyping as binary trait was not particularly appropriate as almost 50% of the individuals had intermediate phenotypes. Consequently, we used the ordinal phenotyping for all analyses except Fisher’s exact test. Logistic, ordered logistic and linear models as well as Nagelkerke and CoxSnell pseudo R2s (Study 2010, Chettier et al. 2015) were calculated in R using the packages “DescTools” and “MASS”.

Fine-mapping

To narrow down the causal interval for the presence of stripes in the Pnye and Hsau hybrid cross, we screened for F2 individuals with recombination breakpoints between the ddRAD markers with the previous highest associations (RAD_84385 and RAD_90805; distance ~600kb (Henning et al. 2014)). Four informative recombinants (Fig. S.I.1) were found that were subsequently genotyped for microsatellites located within the interval (MS03, MS05, MS06, MS07). We concluded that the causal interval must be between

MS03 and MS06 but had two more recombinants (F2_Rec03 and F2_Rec04) within this interval (~80kb) that allowed further finemapping. Since we found no more informative microsatellites within the interval, we continued using Sanger sequencing to narrow down the causal region. The final breakpoints identified flanked a quite small, ~25kb

– 21 –

⁕ Chapter I ⁕ interval that included agrp2 and MS05. Primers for microsatellite amplifications (P005- P012) and flanking SNPs (P013-P018) are listed in Table S.I.5.

Population mapping

For population mapping, we used DNA extracted from wild-caught samples of Pnye, Hsau and Hchi that were collected in Lake Victoria as previously described (Verheyen et al. 2003, Elmer et al. 2009, Elmer, Thompson, and Meyer 2009). Additionally, we also included DNA from one sample for each species from our stocks as well as the genome sequence of Pnye (Brawand et al. 2014). SNPs were genotyped using PCR amplification followed by Sanger sequencing and assembled to the reference that was extracted from the Pnye genome (Brawand et al. 2014). Gaps of the genome assembly were filled using Sanger sequencing reads from the Pnye QTL parent of the Pnye/Hsau cross and the resulting 25kB scaffold was used as a reference. Primers for amplification and sequencing can be found in Table S.I.5 (P0019-P213). After alignment of the Sanger reads to the reference (using Geneious 10.0.9), we extracted consensus files and produced variant files using custom scripts to perform association mapping between Pnye and Hsau as well as between Pnye and Hchi using Fisher’s exact tests. We excluded variants with genotyping information from less than 10 individuals and SNPs that had more than two alleles or SNPs with minor allele frequencies lower than 15%. Association was calculated using Fisher’s exact test and visualized by plotting log-transformed P-values. Because the number of sequenced individuals differed per SNP (10-26 individuals per species per SNP), genotypes were resampled 1000 times for 10 randomly selected individuals and average P-values were plotted for every SNP to obtain comparable P- values across the interval. The total number of SNPs were 200 for the Pnye/Hchi comparison and 393 for the Pnye/Hsau comparison.

Genotyping of Pseudotropheus demasoni and P. cyaneorhabdos

To associate the stripe phenotype with the Lake Malawi agrp2 locus, we tested if we could use one of the microsatellite markers used for the Pnye/Hsau hybrid cross for genotyping the parents of the cross. MS05 (Fig. S.I.1) was found to be within the interval (first intron of atp6V0d2) and showed a length polymorphism (Pdem female: 322/330;

Pcya male: 250/266) and was therefore used for genotyping the F2 individuals (Primers P007 and P008, Table S.I.5).

– 22 –

⁕ Chapter I ⁕

RNA extraction

To examine the expression of genes in different tissues, muscle, liver, eyes, brain and skin tissue were dissected and kept in RNAlater (Invitrogen) at 4°C overnight and then transferred to -20°C for long-term storage. Tissues were taken from species obtained from cichlid importers. RNAlater was removed prior to homogenization. Skin samples and the appropriate amount of TRIzol (Invitrogen) (1ml TRIzol per 0.1g sample) were homogenized in 2ml Lysing Matrix A tube (MP Biomedicals) using FastPrep-24 Classic Instrument (MP Biomedicals). RNA was extracted according to the manufacturer’s recommendations (Invitrogen) with an additional wash step with 75% Ethanol. Subsequent purification and on-column DNase treatment was performed with the RNeasy Mini Kit (Qiagen) and RNase-Free DNase Set (Qiagen). For all species comparisons, we extracted RNA from whole trunk skin biopsies of one side unless otherwise indicated. For comparison between stripes, bars and interstripe and interbar region we sampled the whole pigmented/non-pigmented region. Other organs were extracted using RNeasy Mini Kit (Qiagen) and digested DNA on column by RNase-Free DNase Set (Qiagen) according to the manufacturer’s protocol. Following extraction and purification, RNA was quantified using Qubit RNA HS Assay Kit (Invitrogen) with Qubit Fluorometer (Life Technologies).

Quantitative reverse transcription PCR (RT-qPCR) and reverse transcription PCR (RT-PCR)

Gene expression analyses were performed on skin, muscle, intestine, liver, eye and brain samples. First strand cDNA was synthesized by using 1μg of total RNA with a GoScript Reverse Transcription System (Promega). qPCRs were performed with 2μl of 100μl synthesized first strand cDNA that was diluted ten times from 20μl of initial reaction volume as a template, 10pmol of each forward primer and reverse primer, and GoTaq qPCR Master Mix (Promega) with nuclease-free water to make the final volume of 20μl in a 96-well plate. Primers for agrp2, agrp1, asip1, asip2, atp6V0d2 and unk are listed under P214-P227 in Table S.I.5. We used 40 cycles of amplification on a CFX96 Real- Time PCR Detection System (Bio-Rad). The amplification program was: initial denaturation at 95°C for 10 min, 40 cycles of 95°C for 20s, 60°C for 60s. At the end of the cycles, melting curves of the products were verified for the specificity of PCR products. Only samples with one peak in the melting curves were processed to analyses.

– 23 –

⁕ Chapter I ⁕

Prior to performing the qPCRs, we tested different housekeeping genes (β-actin, gapdh, rpl8, b2m and ef1a; primer P228-P237, Table S.I.5) on samples from Pnye, Hsau, Mzeb and Nbri. We used NormFinder to choose the best reference gene. β-actin, gapdh and rpl8 showed similar and low stability values and strong correlations, while b2m and ef1a were high and showed poor correlations. We assayed gene expression in triplicate for each sample and normalized the data using the housekeeping gene β-actin (Primer P228 and P229, Table S.I.5). Ct values were defined as the point at which fluorescence crossed a threshold (RCt) adjusted manually to be the point at which fluorescence rose above the background level. Next, we compared the relative expression between samples using the 2-∆∆CT method (Nolan, Hands, and Bustin 2006). For the comparative analysis of agrp2 expression we additionally sequenced the primer binding sites (P020 and P024, Table S.I.5), which were almost fully conserved across all 25 analyzed species. For the 24 species comparisons, we independently extracted left and right skin of three individuals (in total 6 samples per species). Two species had base-pair substitution within the primer site of the reverse primer (Ncyl had two, Ncau one). We confirmed that this SNP has no significant effect on qPCR results by comparing results from qPCRs using reverse primers for each sequence (P215 and P216, Table S.I.5) as well as a degenerated primer (P217, table S.I.5). Results from both Ncyl (P = 0.999) and a species with the other variant Jmar (P = 0.406) showed no significant differences. Furthermore, both species lack stripes. Therefore, even if the real expression levels of agrp2 are higher in those species, it would strengthen the support of our hypothesis that a lack of stripes is linked to high agrp2 expression. For pairwise comparisons of gene expression significance tests were performed using Welch’s two-sample t-test. For group comparisons we used ANOVA followed by Tukey’s HSD. Phylogenetic comparisons were performed using phylogenetic ANOVA (see above). All statistical tests were performed in R. In Fig. I.2D and Fig. S.I.4A expression is shown as fold-change relative to the mean of Hsau skin expressions, in Fig. S.I.4B relative to the genotype with the lowest expression and in Fig. I.3B and S.I.14 relative to mean of whole skin Hsau skin expression. For RT-PCR, cDNA was synthesized as described above. PCR was performed using DreamTaq DNA Polymerase (Thermo Fisher Scientific) using the same primers as above and the following program: initial denaturation at 95°C for 10min, 30 cycles of 95°C for 30sec, 60°C for 30sec, 72°C for 1min 30sec, final extension at 72°C for 7min.

– 24 –

⁕ Chapter I ⁕

Sequencing of enh.a

The region enh.a (agrp2 enhancer) was sequenced in all 24 species that were used for analyzing agrp2 expression, in the parents of all three hybrid crosses (6 individuals), in 10 population samples for Hchi, in 25 population samples for Pnye and 21 population samples for Hsau (see above) and in 3 additional species (see Fig. S.I.12). The sequence was amplified with P103 and P203 (and sequenced with P103, P203 and P068) and P238 and P239 (and sequenced with P238, P239 and P068; Table S.I.5).

Phylogenetic association between agrp2 expression and stripes

To examine the evolutionary association between agrp2 expression and the presence of stripes, phylogenetic relationships among 24 focal cichlid species that either possess or do not possess stripes from Lakes Victoria, Malawi and Tanganyika were reconstructed. Oreochromis niloticus was also included as an outgroup to these species (Brawand et al. 2014). Evolutionary relationships were estimated in a species tree framework using three data sets. First, we estimated the species tree with six loci including ND2, mitfa, mitfb, lws, s7, and rag1. This phylogenetic data matrix of 3474 bp was composed of genetic information available from GenBank combined with newly generated Sanger sequences (using P240-P251, Table S.I. 5) that were used to reconstruct the species tree of these species in BEAST (Bouckaert et al. 2014). This analysis was run five times for 20 million generations each and we discarded the first 10 million trees as burn-in. Because mitfa, mitfb, and lws have the potential to be involved in cichlid color pattern evolution, we also excluded these three loci and reran all of our phylogenetic analyses similarly five times, with 20 million generations each, and a burn-in of the first 10 million trees from each run. Additionally, to obtain an independent phylogenetic hypothesis for testing associations between agrp2 expression and the presence of stripes, we constructed a phylogenetic matrix for 21 of the same species using genomic sequences of 30 additional genes (arnt, azin1a, bcanb, bmp4, cspg5a, ctnnb1, dlx1a, dlx2a, dlx3b, dlx4a, dlx4b, edn2, fzd6, irx1a, irx2a, irx4a, isl1, myg1, ndrg1b, nrp2b, ntrk1, osr2, pitx1, pitx2, pitx3, shh, smo, sostdc1a, tuft1a and wnt4b) from published genomes (Brawand et al. 2014, Meier et al. 2017, Malinsky et al. 2018) as well as from target enrichment data. Target enrichment data was produced using customized 120nt baits with ~3x flexible tiling density for each targeted genomic region from the O. niloticus (NCBI: GCA_000188235.1) genome. DNA was either extracted from muscle tissue or from fin – 25 –

⁕ Chapter I ⁕ clips stored in EtOH following the DNeasy blood & tissue protocol (QIAGEN) or Genaxxon Genomic DNA Purification Mini Spin Column Kit (Genaxxon Bioscience GmbH), respectively. Before lysis the tissue was incubated in TE buffer (pH 9) to reduce ethanol contamination. Lysis was followed by addition of RNAse for RNA digestion. The concentration of the purified DNA was evaluated with fluorescence spectrophotometry Qubit (Invitrogen) and DNA was checked for degradation on a 1 % agarose gel (85V for 40 minutes). For library preparation, we used the Illumina TruSeq Nano HT Library preparation kit (Illumina Inc.) following the manufacturer’s guidelines. All samples were run on a Bioanalyser 12000 Chip to assure a high quality of DNA libraries and afterwards amplified using PCR. We followed the MYBaits manual v3 (http://www.mycroarray.com/pdf/MYbaits-manual-v3.pdf) and hybridized the probes at 65°C for 22 hours. Finally, all sample were checked on a Bioanalyser HS chip before sequencing. Probes were sequenced in paired-end mode on a HiSeq 2500 system. To process the data we trimmed Illumina adapters from the raw fastq reads using picard (http://broadinstitute.github.io/picard) and mapped the samples to the Oreochromis niloticus genome v1 (NCBI: GCA_000188235.1) using bwa mem v0.7.12 (http://bio- bwa.sourceforge.net) (Li and Durbin 2009). The consensus sequences of coding regions were extracted using samtools (Li, Handsaker, Wysoker, Fennell, Ruan, Homer, and Marth 2009), aligned using MAFFT version 7 (Katoh and Standley 2013) and manually trimmed to the Oreochromis niloticus coding sequence annotation of Ensembl (Orenil1.0) using Geneious® 10.2.3. All post burn-in species tree MCMC were checked for convergence initially using the stabilization of trace plots. We also examined the effective sample size of the likelihoods of each gene tree analysis remaining post-burn-in using Tracer v1.5 (Bouckaert et al. 2014) to ensure values were over 200 and the MCMC chains converged to a stationary distribution. To visualize the general patterns recovered from these analyses, 100 randomly drawn post burn-in trees from the six loci, ND2, mitfa, mitfb, lws, s7, and rag1, species tree reconstruction (Fig. I.3) was depicted using densitree (Bouckaert 2010). Then, using 100 randomly drawn phylogenies from each of the combined post burn-in distribution of trees for the three separate species tree reconstructions, we performed three sets of phylogenetic ANOVAs as implemented in the R package phytools (Revell 2012) on the relationship between the presence of stripes and the relative expression of agrp2. As the individuals of Placidochromis johnstoni that

– 26 –

⁕ Chapter I ⁕ we used for gene expression analysis and of which one representative is shown in Figure S.I.11 have a small thin stripe in the dorsal, most anterior skin we also reran the analyses to test if a mis-assignment of P. johnstoni breaks the general association, which was not the case (P = 0.00103) The gene tree for the agrp2 enhancer (enh.a; see under ‘Sequencing of enh.a’) for the same 24 Great Lake cichlid species examined for the association of agrp2 expression and the presence of a stripe as well as Oreochromis niloticus was reconstructed in BEAST. Analyses of relationships were run five times for 20 million generations each to infer the locus history and discarded the first 10 million generations from each run as burn-in samples. We then combined the post burn-in trees to estimate the posterior probabilities of nodes subtending the 24 Great Lake cichlids for this single locus.

In situ hybridization

In situ hybridization was performed as previously described (Di Meglio et al. 2013). Briefly, dissected skin and muscle tissue was fixed in 4% PFA in phosphate buffer (PBS 1x) overnight. Tissues were cryoprotected in 30% sucrose (Sigma) / PBS 1x and embedded in gelatin 12.5% (Sigma) / 30% sucrose / PBS1x before being frozen at -80°C. Cryostat sections (20 μm) were cut on a Microm HM500 OM in coronal orientation. Sections of both species were mounted on the same slide to avoid unspecific differences in staining and kept at -80°C prior hybridization. For in situ hybridization slides were air-dried for 30min fixed in 4% PFA in PBS for 10 min. After several washes in PBS, slides were acetylated in acetylation solution (0.35% acetic anhydride, 0.1M triethanolamine, pH 7.4) before hybridization at 72°C. Antisense riboprobes (agrp2) labeled with digoxigenin(DIG)-11-d-UTP (Roche) was denatured (200ng/ml) in the hybridization buffer (50% deionized formamide, 1x Salts, 10% dextran sulfate, 1mg/ml tRNA, 1x Denhart’s). The 548bp agrp2 probe was cloned using primers P252 and P253 (Table S.I.5). Next, slides were washed in washing solution (0.2x SSC, 50% Formamide, 0,1% Tween 20) and MABT (pH 7.5) (NaOH 200mM, Maleic acid 100mM, NaCl 150mM, Tween 0.1%). They were then blocked in 2% blocking reagent (Roche) in MABT for 2 hours and incubated overnight with anti-DIG antibody conjugated to peroxidase (1/1,000; Roche). After several washes in TBST (pH 7.5), the peroxidase activity was detected using 1/50 Cyanine 3 Plus Amplification Reagent in 1X Plus

– 27 –

⁕ Chapter I ⁕

Amplification Diluent of TSA Plus Cyanine 3 system (PerkinElmer) for 45 min. The development was quenched by 2% hydrogen peroxide in TBST (pH 7.5). The slides were counterstain with 4’,6-Diamidino-2-phenylindole dihydrochloride (DAPI) (1/5000; Sigma) in PBS before we mounted the slides in Mowiol mounting medium (Sigma). Photographs were taken with a ZEISS LSM 880 laser scanning microscope with Airyscan detector (figure S4G) or a Leica DM6B microscope with a Leica DFC3000G camera (all other images).

Molecular cloning

To obtain the reporter assay constructs, target elements were cloned into a Tol2 vector (Kwan et al. 2007) containing GFP and a minimal promoter sequence (minP) as previously described (Di Meglio et al. 2013). For PCR amplification primers with restriction enzyme overhangs for XhoI and EcoRI were used (P254 and P255; Table S.I.5). Both vector and amplicons were cut with both enzymes, purified, ligated and transfected into JM109 competent cells (Promega). Successful cloning was confirmed using restriction digest analysis and Sanger sequencing. DNA for microinjection was purified using Plasmid Midi Kit (Quiagen) and eluted using RNAse-free water.

Microinjection and analysis of transgenic animals

Microinjections were performed as previously described (Fisher et al. 2006, Kratochwil et al. 2017). The injection mix contained 1µl Plasmid (125 ng/µl), 1µl Transposase RNA

(final concentration 125 ng/µl), 0.5µl Phenol Red (2% in H2O) and 7.5 µl RNAse-free water. Zebrafish embryos were raised in 15cm petri dishes and analyzed five days after fertilization. Photographs of all embryos were taken with the same settings under a stereomicroscope (Leica MZ10 F with Leica DFC3000G camera) and a GFP filter. Quantifications were performed in Fiji by selecting the whole trunk with the freehand selection function and measuring the average intensity of the area using the measure function. We quantified F0 and not stable F1 embryos to control for position effect variation. All embryos (including negative embryos) were included in this analysis to fully account for the mosaicism of F0 individuals. Values were plotted in R and statistics were performed using Kruskal-Wallis rank sum test (P < 0.001) followed by Dunn’s test (enh.a_Pnye::GFP vs enh.a_Hsau: P < 0.001; enh.a_Pnye vs ctrl_Pnye P < 0.001; ctrl_Pnye vs enh.a_Hsau: P < 0.001). As an additional statistical test we log2-

– 28 –

⁕ Chapter I ⁕ transformed the average intensity to reach normality (Shapiro-Wilk test on log transformed data: all P > 0.05) an used ANOVA (P < 0.001) followed by Tukey HSD (enh.a_Pnye::GFP vs enh.a_Hsau: P < 0.001; enh.a_Pnye vs ctrl_Pnye P < 0.001; ctrl_Pnye vs enh.a_Hsau: P < 0.001). Embryos injected with enh.a_Pnye vs enh.a_Hsau were raised and outcrossed to wildtypes. A representable F2 transgene for each of the transgenic lines, tg(enh.a_Pnye::GFP) and tg(enh.a_Hsau::GFP) as well as a non- transgenic larva was imaged at five days post fertilization under a Leica DM6B with Leica DFC3000G camera and a GFP filter.

CRISPR-Cas9 gRNAs target sequences in Pnye agrp2 exons 1, 2 and 3 were identified using CHOPCHOP with the Astatotilapia burtoni genome as reference for off-target sequences (Montague et al. 2014, Labun et al. 2016). We designed two gRNAs in exon 1 (gRNA #1 and #2), two gRNAs in exon2 (gRNA #3 and #4) and one gRNA in exon3 (gRNA #5). The PCR primers for gRNA design are given in Table S.I.5. gRNAs were synthesized following (Varshney et al. 2015), using Sp6 and T7 polymerases for different gRNAs depending on their 3’ sequence (indicated in Table S.I.5). Embryos were injected at the 1-4 cell stage with approximately 1-2 nl of a mixture of three gRNAs at a concentration of 20ng/µl each, supplemented with 0.05µl/µl CAS9 protein (NEB M0646T). Two injection mixes were used; one containing gRNAs #1, #3, #5 and one containing gRNAs #2, #4. Mutants CRISPR_mt_03, _04 were injected with gRNA mix (#1, #3, #5), CRISPR_mt _01, _02 with gRNA mix (#2, #4). Photographs were taken at 2 months in the above described photography chamber (Fig. I.2G, H and Fig. S.I.10; CRISPR_mt_04 had stripes but died for unknown reasons before we took photographs). For genotyping, DNA was extracted from fin clips using salt extraction. Exons were amplified using Primer P277 to P284 (Table S.I.5). To confirm the individual knockout genotypes PCR Fragments were cleaned up using Zymo DNA Clean & ConcentratorTM- 5 Kit and cloned into pGEM Vector using the standard protocol. PCR Amplification was directly performed on colonies using P274 and P275 (Table S.I.5) and sequenced using standard sequencing protocols with primer P276 (Table S.I.5). Sequences were aligned in Geneious® 10.2.3 to check for mutations. We found mutations in all individuals with 98% of the clones having mutations (17/19 for gRNA_01, 15/15 for gRNA_02, 13/13 for gRNA_03, 31/31 for gRNA_04 and 11/11 for gRNA_05), suggesting that most of the

– 29 –

⁕ Chapter I ⁕ cells carry in fact mutations many of them inducing frameshifts or early stop codons (79%, 73%, 92%, 35% and 73% for the respective gRNAs; Table S.I.1). Experiments were approved by the German authorities (Permit of the Regierungspräsdium Freiburg, G-17/85).

Supplementary Text

Extended phenotype description

Melanic horizontal stripe patterns can be found in several lineages of African cichlids including the Lake Malawi (Fig. I.1K, L) and Lake Victoria radiations of haplochromine cichlids (Fig. I.1C, D) as well as several lineages of Tanganyika cichlids (Fig. I.1G, H). Based on previous phylogenetic reconstructions it is clear that horizontal stripes have evolved and been lost repeatedly within these cichlid radiations (Stiassny and Meyer 1999, Seehausen, Mayhew, and Van Alphen 2001). In most Rift Lake cichlids, stripe patterns are composed of both a dorsolateral stripe running from the head region towards the posterior end of the dorsal fin and a midlateral stripe that starts at the operculum and runs to the base of the caudal fin (Fig. I.1A, C, D, G, H, K and L). In some species, stripe patterns deviate from this stereotypic pattern that can also vary between sexes and during ontogeny. Several cichlids also exhibit context-dependent chromomotor flexibility that allows stripe patterns to intensify or fade to putatively better match the background or to communicate motivation in social and sexual contexts (Fernald 2012, Nilsson Sköld, Aspengren, and Wallin 2013). Stripe patterns are suggested to have adaptive significance in a range of vertebrates and are implicated in species recognition and intraspecific communication (Kelley, Fitzpatrick, and Merilaita 2013), camouflage through disruptive coloration (Murali, Kodandaramaiah, and Fitzpatrick 2018), and motion camouflage through manipulating speed perception of moving individuals (Fricke 2019). Bar and stripe patterns might be alternative strategies and have been associated to ecology and behavior. While stripes have been shown to evolve with piscivorous feeding and shoaling, bars are often found in species that inhabit rocky habitats and vegetation (Seehausen, Mayhew, and Van Alphen 2001). Presence or lack of stripe (and bar) patterns might therefore play a number of adaptive roles throughout ontogeny, between the sexes, as plastic responses to the environment, and during interspecific interactions, and all of these putative functions of color patterns may have contributed to the exceptional rates

– 30 –

⁕ Chapter I ⁕ of cichlid diversification (Meyer et al. 1990, Seehausen, Mayhew, and Van Alphen 2001, Kocher 2004).

Species names and assignment

Throughout the manuscript we use species abbreviations to improve readability (Pnye instead of Pundamilia nyererei). Several of the analyzed species have different names across the literature. Because no commonly accepted taxonomy of cichlids is available we relayed on the classification of the ‘Catalog of Fishes of the California Academy’ (Fricke 2019). In brief, Pundamilia nyererei and Pundamilia pundamilia are also referred to as Haplochromis nyererei and Haplochromis pundamilia. Haplochromis sauvagei is also referred to as Paralabidochromis sauvagei, Ctenochromis sauvagei or H. spec. ‘rockkribensis’ (Seegers 2008). Haplochromis chilotes is also referred to as Paralabidochromis chilotes. Pseudotropheus cyaneorhabdos is sometimes referred to as Melanochromis cyaneorhabdos, Haplochromis latifasciatus as Astatotilapia latifasciata, Maylandia zebra as Pseudotropheus zebra and Lamprologus multifasciatus as Neolamprologus multifasciatus. Strictly, Haplochromis latifasciatus is not a cichlid from Lake Victoria, but from Lake Kyoga that is connected to the Lake Victoria basin and the species can be considered as part of the Lake Victoria ‘superflock’ (Verheyen et al. 2003).

Additional analyses of Lake Victoria QTLs and populations samples

As reported in the main text, we mapped stripes in a second cross involving the same non-striped species, Pnye and a second striped species, Haplochromis chilotes (Henning et al. 2017). Stripes mapped to the same position providing further evidence that horizontal stripes map to this interval. As in the first cross (Henning et al. 2017), stripes 2 were recessive and the F2 distribution did not differ from 3:1 Mendelian proportions (χ = 0.5291, d.f. = 1, P = 0.467). To exclude ‘causal’ differences in the coding sequence of the parental individuals we re-sequenced all gene-coding regions (agrp2, atp6V0d2, unk) within the interval (Fig. S.I.2) and revealed only a single non-synonymous substitution in agrp2 and no new start/stop codons. The non-synonymous mutation was neither fixed in populations of Hchi, that always exhibit stripes, nor was it associated with striped versus non-striped phenotypes in several closely related Lake Victoria species (Fig. S.I.2).

– 31 –

⁕ Chapter I ⁕

In an effort to identify the causal interval affecting both agrp2 expression and stripe phenotype, we sequenced the agrp2 locus in population samples of the three species from our hybrid crosses (all n≥10). This allowed us to screen for alternatively fixed variants that are fully associated with the stripe phenotype. Pairwise comparisons of two striped species (Hsau or Hchi) with the non-striped Pnye showed that the agrp2 locus exhibits species-specific and alternatively fixed mutations (Fig. S.I.7). In the Pnye – Hsau comparison, alternatively fixed variants spanned an interval of ~11kb (Fig. S.I.7). All alternatively fixed variants of the Pnye – Hchi comparison were contained in a 1.1kb interval that overlapped with the 11kB interval of the Pnye – Hsau comparison (Fig. S.I.7). This interval was located in the first intron, directly upstream of agrp2 exon 2 and thereby provided an excellent candidate region for regulatory elements controlling agrp2 expression.

Additional test for comparative gene expression analysis

For the comparative gene expression analysis our aim was to sample (obtainable) species that were phylogenetically disparate as possible across the radiations of Lake Tanganyika, Malawi and Victoria and to focus on species with the stereotypic stripe pattern (continuous dorso-lateral and mid-lateral stripe) and individuals lacking these stereotypic patterns (in the best case without any pattern remotely similar to horizontal stripes). Individuals of Placidochromis johnstoni and of which one representative is shown in Fig. S.I.11 have a small thin stripe in the dorsal, most anterior skin. After detailed examination of the specimens we obtained, it was in our opinion reasonable to classify P. johnstoni as non-striped (in particular as we were not able to obtain another more clearly nonstriped Non-Mbuna Malawi cichlid). The observation that other individuals/populations of P. johnstoni indeed show more pronounced stripe patterns is very interesting and has been also reported by Ad Konings (Konings 2007). In fact, it might suggest that stripes (and the underlying genomic basis) are polymorphic within P. johnstoni (i.e. standing genetic variation and/or introgression of stripe alleles). To confirm that a potential mis- assignment of P. johnstoni would not break the general association, we classified P. johnstoni as striped species and reran the analysis with all our phylogenetic reconstructions. The analyses revealed a robustly significant association of stripe patterns and agrp2 expression across the sampled species (phylogenetic ANOVA, P = 0.00103).

– 32 –

⁕ Chapter I ⁕

We also repeated the analysis for either a subset of these loci that excluded genes putatively under selection (mitfa, mitfb, lws) as well as when used a larger gene set of 30 genes that we extracted from genomic sequences and target enrichment data for a subset of 21 species and obtained similarly significant results (mean P < 0.001 and P = 0.0012). Down-regulation of agrp2 is therefore reproducibly associated with the stripe phenotype in East African cichlids.

Extended analysis Malawi hybrid cross

As stated in the main text we genotyped 270 F2 hybrid individuals for the agrp2 locus. Stripes were phenotyped in a number of ways: as a binary trait (striped / non-striped), an ordinal trait (striped / partially striped / non-striped; Fig. I.4A) and a continuous trait (percentage of stripe region covered by pigmentation; Fig. S.I.15). Analyses of the stripe phenotype as either an ordinal or continuous trait revealed a pattern consistent with recessive epistasis, whereby the variance in stripe patterns was significantly lower in homozygous individuals for the Pcya allele (C/C; Fig. I.4A and Fig. S.I.15) when compared to heterozygotes (C/D, Levene’s test; P < 0.001) or individuals homozygous for the Pdem allele (D/D, Levene’s test; P < 0.001; Fig. I.4A and Fig. S.I.15). Hence, our data is compatible with the agrp2 locus playing a substantial role in regulating stripes in Lake Malawi. However, phenotypic expression of alleles of one or more independent modifier loci can influence the extent of horizontal stripes in the Malawi cichlids, but appear to be masked in individuals that are homozygous for the agrp2 allele of the striped species, Pcya (C/C). Linear and logistic models (Table S.I.2 to S.I.4) as well as additional information about genotype and allele frequencies as well as odds ratios and penetrance greatly support the notion that agrp2 is a major effect locus within this cross from Lake Malawi (Fig. S.I.15). Odds of C/C and D/D individuals of developing full stripes are reversed and almost absolute (2:67 and 59:1 respectively). Percentage of individuals with full stripes are 97.1% (C/C), 14.8% (C/D) and 1.7% (D/D) (Fig. S.I.15).

Genetic and environmental interactions with the agrp2 pathway

Interestingly, the dorsolateral stripe was not restored in any of the four CRISPR-Cas9 mutants. This ‘only midlateral stripe phenotype’ was never observed in F2 individuals of the Pnye × Hsau cross. The phenotype also does not correlate with the agrp2 expression patterns (Fig. I.2D). This suggests that a combination of recessive alleles that generally

– 33 –

⁕ Chapter I ⁕ suppress pigmentation in the dorsal skin of Pnye likely inhibits the dorsolateral stripe. It also fits the observation that pigmentation (e.g. in bars) of Pnye is greatly reduced dorsally (Figs. I.1N and I.2G). Furthermore, the phenotype matches the stripe pattern of some naturally occurring Lake Victoria cichlids that have a midlateral stripe but lack a dorsolateral stripe (e.g. Haplochromis paropius or Harpagochromis sp. ‘orange rockhunter’). Taken together, our results strongly suggest that although species such as Pnye exhibit no stripes, all mechanisms for stripe pattern formation are in place and stripes can be simply re-revealed in non-striped cichlids by removal or reduction of agrp2 expression. Our results also reveal that additional regulatory mechanisms of the agrp2 locus await discovery as regulatory variation in Lake Malawi and Tanganyika cichlids is likely to be caused by different variants within or variation outside of the enh.a locus (Fig. I.3C). The lack of the dorsolateral stripe in the CRISPR-Cas9 mutants (Fig. I.2H) coupled with the non-Mendelian basis of stripes in the Malawi cross (Fig. I.4A) provide evidence that also epistatic interactions with other loci besides agrp2 remain to be uncovered for a more complete understanding of the genetic mechanisms and the evolutionary process of stripe color pattern diversification in cichlid fishes. It is possible that already known genes and mechanisms from research in other vertebrates contribute to the phenotype (Mallarino et al. 2016, Eom and Parichy 2017). The description of further loci might therefore contribute to a deeper understanding of how and if molecular and cellular building blocks of convergently evolved stripe patterns are shared across larger phylogenetic scales. Also, our study has focused so far only on adult expression patterns and it remains to be investigated what the precise role of ontogenetic processes are in stripe pattern diversification. Further research that addresses why vertical bars or other melanic patterns (Fig. I.1 and (Seehausen, Mayhew, and Van Alphen 2001)) are largely unaffected by regulatory changes or genetic modifications of agrp2 (Figs. I.2 and I.4) and how they are in turn genetically shaped, will provide insights into the evolution and modularity of color patterns. Our results suggest that both patterns are largely independent and are shaped by superimposition of two independent mechanisms, as described for other color patterns in zebrafish (Ceinos et al. 2015). Moreover, stripes are often plastic and can be softened and enhanced through hormonal and/or neural control. Therefore, future efforts that address how neural and endocrine pathways interact with

– 34 –

⁕ Chapter I ⁕ agrp2 will expand our knowledge of socially and environmentally induced phenotypic plasticity in coloration (Fernald 2012).

Cis-regulatory changes at the agrp2 locus

Expression levels of agrp2 in cichlids likely act as an evolutionarily switch controlling the presence and absence of stripe patterns. In Lake Victoria, expression level differences seem to be caused by several mutations in a 1.1 kb intronic regulatory region (enh.a.; Figs. S.I.12 and S.I.13) that push the expression of agrp2 levels above or below a threshold that can ultimately determine stripe presence. Such a molecular on/off switch may have permitted the frequent loss, as well as re-evolution, of stripes. A single locus could therefore account for a substantial amount of the astonishing diversity in color patterns in cichlids, patterns that are thought to play an important role in both natural selection and sexual selection (Kocher 2004, Roberts, Ser, and Kocher 2009, Kelley, Fitzpatrick, and Merilaita 2013, Phillips et al. 2017, Murali, Kodandaramaiah, and Fitzpatrick 2018). More generally, genomic hotspots such as the agrp2 locus might more commonly explain other patterns of the explosive and repeated evolution of cichlid fishes (Meyer 1993, Kocher 2004, Nilsson Sköld, Aspengren, and Wallin 2013). Although the presence of stripes appears to be controlled by the same gene (agrp2), the exact underlying genetic variants must differ among the evolutionarily radiations of Lakes Victoria, Malawi and Tanganyika (Figs. S.I.12 and S.I.13). Different cis-regulatory mutations that convergently influence agrp2 expression variation are apparently generating the same phenotypes (i.e. presence of stripes) but might also be able to drive stripe pattern variation (e.g. through local variation in agrp2 expression). Our result bears resemblance to the genetic basis of color variation found in Heliconius butterflies, where diversification in coloration has been shown to be driven by modular regulatory complexity around a few loci of major effect (Van Belleghem et al. 2017).

Gene Nomenclature

Because gene/protein nomenclature is inconsistent within cichlid research, we adapted the nomenclature of zebrafish when writing about cichlid genes/proteins, and mouse nomenclature when we write about mammalian genes/ proteins.

– 35 –

⁕ Chapter II ⁕

Chapter II

Evolutionary Dynamics of Structural Variation at a Key Locus for Color Pattern Diversification in Cichlid Fishes

Claudius F. Kratochwil†, Yipeng Liang†, Sabine Urban, Julián Torres-Dowdall, Axel Meyer

(† equal contribution) Genome Biology and Evolution (2019) DOI:10.1093/gbe/evz261

Abstract

Color patterns in African cichlid fishes vary spectacularly. Although phylogenetic analysis showed already 30 years ago that many color patterns evolved repeatedly in these adaptive radiations, only recently have we begun to understand the genomic basis of color variation. Horizontal stripe patterns evolved and were lost several times independently across the adaptive radiations of Lake Victoria, Malawi, and Tanganyika and regulatory evolution of agouti-related peptide 2 (agrp2/asip2b) has been linked to this phenotypically labile trait. Here, we asked whether the agrp2 locus exhibits particular characteristics that facilitate divergence in color patterns. Based on comparative genomic analyses, we discovered several recent duplications, insertions, and deletions. Interestingly, one of these events resulted in a tandem duplication of the last exon of agrp2. The duplication likely precedes the East African radiations that started 8– 12 Ma, is not fixed within any of the radiations, and is found to vary even within some species. Moreover, we also observed variation in copy number (two to five copies) and secondary loss of the duplication, illustrating a surprising dynamic at this locus that possibly promoted functional divergence of agrp2. Our work suggests that such instances of exon duplications are a neglected mechanism potentially involved in the repeated evolution and diversification that deserves more attention. – 36 –

⁕ Chapter II ⁕

Introduction

The East African cichlid fish adaptive radiations, with their over 1,200 species, are one of the most prominent examples for repeated, convergent evolution (Meyer 1993, Stiassny and Meyer 1999, Muschick, Indermaur, and Salzburger 2012). In the large species flocks of Lakes Tanganyika, Malawi, and Victoria, body shapes, trophic morphologies, and color patterns evolved dozens of times independently (Kocher 2004, Muschick, Indermaur, and Salzburger 2012, Kratochwil et al. 2018, Salzburger 2018). This exuberant variation demanded the search for genomic explanations for the exceptional rates of diversification and repeated evolution (Brawand et al. 2014, Kratochwil and Meyer 2015a, Salzburger 2018, Kratochwil 2019). Recent work from sticklebacks supports that genomic features such as DNA conformation can promote evolutionary adaptations and do so repeatedly (Xie et al. 2019). A meta-analysis of several cases of parallel evolution also suggests an influence of mutational biases on repeated adaptive evolution (Lynch et al. 2016, Stoltzfus and Mccandlish 2017). Moreover, gene duplications are important molecular mechanisms of diversification (Ohno 1970, Van De Peer, Maere, and Meyer 2009, Musilova et al. 2019) and convergence (Denoeud et al. 2014, Ishikawa et al. 2019). Recently, it has been shown that the evolution of cichlid fish stripe color patterns is facilitated by independent regulatory evolution at the agrp2 locus (Kratochwil et al. 2018). Yet, it remains unclear what genomic features influenced the evolution of the underlying flexible gene regulatory architectures and the evolution of novel cis- regulatory elements that clearly account for diversification and repeated evolution in cichlids. Here, we report on an investigation of the agrp2 locus and asked whether particular genomic features including structural variation might affect the evolutionary dynamics of this important locus for cichlid color pattern diversification.

Materials and Methods

Pairwise global alignment

To perform pairwise global alignments, we extracted ∼200 kb (100 kb for the reference) intervals flanking the agrp2 gene from Ensembl 94 (Cunningham et al. 2019) and NCBI.

– 37 –

⁕ Chapter II ⁕

The Maylandia zebra genome (Ensembl 94, GCA_000238955.5) was used as a reference. Before alignment, the reference was masked using the Tilapia repeat masker (Shirak et al. 2010). Nine genomes were aligned to the reference: Pundamilia nyererei (Ensembl 94, GCA_000002035.4), Neolamprologus brichardi (Ensembl 94, GCA_000239395.1), Oreochromis niloticus (NCBI, MKQE02), Amphilophus citrinellus (Ensembl 94, GCA_000751415.1), Xiphophorus maculatus (Ensembl 94, GCA_002775205.2), Gasterosteus aculeatus (Ensembl 94, BROAD S1), Scophthalmus maximus (Ensembl 94, GCA_003186165.1), Astyanax mexicanus (Ensembl 94, GCA_000372685.2), and Danio rerio (Ensembl 94, GCA_000002035.4). For P.nyererei, we used a curated version (description see below). For alignment, we used the Shuffle-LAGAN algorithm (Brudno et al. 2003) with a RankVISTA probability threshold of 0.5 and providing the following pairwise phylogenetic tree: ((((((((M. zebra P. nyererei) N. brichardi) O. niloticus) Amp. citrinellus) X. maculatus) G. aculeatus) S. maximus)(Ast. mexicanus D. rerio)). For the M. zebra reference, we used an annotation based on the ensemble annotation and manually annotated the previously described cis-regulatory region of agrp2 (Kratochwil et al. 2018). Finally, we extracted a ∼45 kb interval of the Shuffle-LAGAN alignment to be used in Fig. II.1A. Genomic coordinates of agrp2 coding regions, cis-regulatory elements, and deletions in the different genomes are summarized in Table S.II.1.

Cichlid Genome Alignments

For the cichlid genome alignments, we selected an 18 kb region encompassing the agrp2 gene. Sequences from the genomes of M. zebra, P. nyererei, and N. brichardi as well as Astatotilapia calliptera (Ensembl 94, GCA_900246225.3) and Astatotilapi aburtoni (Ensembl 94, GCA_000239415.1). Additionally, we added a Sanger sequencing assembly of the locus from the Lake Victoria species Haplochromis sauvagei (Kratochwil et al. 2018). Sequences were aligned using MAFFT 7 (Katoh, Rozewicki, and Yamada 2019) using automatic strategy choice, and standard settings. Conservation score and annotations were added using Geneious 2019. The alignment was visualized using Adobe Illustrator CC 2018. Dot plots were generated using MAFFT 7 (Katoh, Rozewicki, and Yamada 2019) with a threshold score of 39 (E = 8.4e-11).

– 38 –

⁕ Chapter II ⁕

RNA Extraction and cDNA Synthesis

Fish were obtained from commercial breeders and euthanized with an overdose of MS- 222. Experiments were performed in accordance with animal research regulations (Regierungspräsidium Freiburg, Baden Württemberg, Germany, Reference number: G- 17/110). Skin, brain (both RNA-seq, cDNA sequencing), liver, eye, and muscle tissue (all cDNA sequencing) were dissected and kept in RNAlater (Invitrogen) at 4 °C overnight and transferred to −20 °C for long-term storage. RNAlater was removed prior to homogenization. Skin samples and appropriate amount of TRIzol (Invitrogen) (1 ml TRIzol per 0.1 g sample) were homogenized in 2 ml lysing matrix A tube (MP Biomedicals) using FastPrep-24 Classic Instrument (MP Biomedicals). RNA was extracted according to the manufacturer’s recommendations with additional 75% ethanol wash one time. Subsequent purification and on-column DNase treatment was performed with RNeasy Mini Kit (Qiagen) and RNase-Free DNase Set (Qiagen). The other organs were extracted using RNeasy Mini Kit (Qiagen). DNA was removed using RNase-Free DNase Set (Qiagen) according to the manufacture’s protocol. Following extraction and purification, RNA was quantified using the Qubit RNA HS Assay Kit (Invitrogen) with a Qubit Fluorometer (Life Technologies). First-strand cDNA was synthesized using 1 μg total RNA and the GoScript Reverse Transcription System (Promega).

5′ and 3′ Rapid Amplification cDNA Ends

5′ Rapid amplification of cDNA end (5′RACE) was performed as previously described (Scotto-Lavino, Du, and Frohman 2006b). We performed RT-PCR and nested PCR by two sets of gene specified primers: aR5 + aR4 +aR3 and aR4 + aR3 + aR2 (Fig. S.II.1). To generate “5′ end” partial cDNA, reverse transcription was carried out using gene specified primers aR5 or aR4 by M-MLV Reverse Transcriptase [H–] (Promega) to generate first-strand product. Following the reverse transcription, a poly(A) tail was appended using terminal deoxynucleotidyl-transferase Tdt (Promega) and dATP. First- strand gene-specific cDNA with poly(A) tail was then purified by NucleoSpin Gel and PCR Clean-up (MACHEREY-NAGEL). Amplification was achieved by PCR using Qtotal and Qouter and gene specified primers, aR4 for the first-strand cDNA product of aR5, aR3 for the first-strand cDNA product of aR4. PCR was performed using DreamTaq DNA Polymerase (Thermo Fisher Scientific) in 50 μl reactions with the following program: initial denaturation at 98 °C for 3 min, annealing at 60 °C for 2 min, – 39 –

⁕ Chapter II ⁕ extension of cDNA at 72 °C for 40 min, followed by 30 cycles of 95 °C for 30 s, 60 °C for 30 s, 72 °C for 3 min, final extension at 72 °C for 15 min. A second set of PCR cycles was carried out to increase the yield of specific product using nested primers Qinner and upstream gene specified primers, aR3 for the first-strand PCR product of aR4, aR2 for the first-strand PCR product of aR3. The second PCR was performed using the same condition as for the first-strand PCR without cDNA extension. After the second PCR, bands were purified using NucleoSpin Gel and PCR Clean-up (MACHEREY-NAGEL). 3′RACE was performed as previously described (Scotto-Lavino, Du, and Frohman 2006a). Total RNA was reverse transcribed to first-strand cDNA by M-MLV

Reverse Transcriptase [H–] (Promega) using primer Qtotal in 20 μl reverse transcription reaction. To increase the specificity of the 3′RACE PCR, we performed a nested PCR by two sets of gene specified primers, aF1 + aF2 and aF2 + aF3 (Fig. S.II.2). For the first

RACE PCR, we used the outer primer Qouter with gene-specific primer aF1 or aF2. PCR was performed using DreamTaq DNA Polymerase (Thermo Fisher Scientific) as described for 5′RACE. For the nested PCR, inner primer Qinner was used in combination with gene-specific primers, aF2 for the first PCR product of Qouter and aF1, aF3 for the first PCR product of Qouter and aF2. The second PCR was performed using the same condition as for the first PCR without cDNA extension. Following the second PCR, we gel-purified bands using NucleoSpin Gel and PCR Clean-up (MACHEREY-NAGEL). All DNA fragments from 3′RACE and 5′RACE were processed by Sanger sequencing using Applied Biosystems 3130 Genetic Analyzers. Sanger Sequencing data were analyzed in Geneious 2019. Primer aF3, aF5, aR5, and aR6 were used for sequencing of 3′RACE product. aF1, aF2, aR1, and aR2 were the sequencing primers of the 5′RACE products. To confirm the sequencing result of 3′RACE and 5′RACE, we additionally performed RT-PCRs. For this, cDNA was synthesized as described earlier. Three forward primers (F1, F2, and F3) and one reverse primer (R1–R3) were used in PCR with DreamTaq DNA Polymerase (Thermo Fisher Scientific) with the following program: initial denaturation at 95 °C for 10 min, 30 cycles of 95 °C for 30 s, 60 °C for 30 s, 72 °C for 1 min 30 s, final extension at 72 °C for 7 min. Primers are summarized in Table S.II.2.

– 40 –

⁕ Chapter II ⁕

Tandem Duplication-Specific PCRs

DNA was extracted from fin clips using ethanol precipitation. To identify species and individuals with and without duplication, we used two approaches (Fig. S.II.3A). The first approach included PCRs with a forward primer in the 3′ end of exon 3a and a reverse primer in the 5′end of exon 3 b (Fig. S.II.3, S.II.4 and Table S.II.2). This approach was more sensitive to detect duplications across species (as primer binding sites were inside coding regions). PCR was performed using DreamTaq DNA polymerase (Thermo Fisher Scientific) in a 20 µl reaction with the following program: initial denaturation at 95 °C for 5 min, 32 cycles of 95 °C for 30 s, 60 °C for 30 s, 72 °C for 2 min, final extension at 72 °C for 15 min. The second approach was a long-range PCR that spanned the tandem duplication (Fig. S.II.5 and Table S.II.2). Long-range PCRs failed in many species, likely because of binding site mutations and repeat regions that flank the duplication (Fig. S.II.4). PCR was performed using DreamTaq DNA polymerase (Thermo Fisher Scientific) in a 20 µl reaction with the following program: initial denaturation at 95 °C for 5 min, 35 cycles of 95 °C for 30 s, 62 °C for 30 s, 72 °C for 7 min, final extension at 72 °C for 30 min.

TaqMan Probe Assay

TaqMan probe-based qPCR was used to determine the copy number of exon 1 (as a control) and exon 3 of agrp2. Genomic DNA was extracted from fin clips using DNeasy Blood & Tissue Kit (Qiagen) according to the manufacturer’s protocol. Following extraction, the purity of genomic DNA was checked by Colibri spectrometer (Berthold), the degradation was evaluated with 1% agarose gel and the concentration was quantified using Qubit DNA BR Assay Kit (Invitrogen) with Qubit Fluorometer (Life Technologies). qPCRs were performed with 10 ng genomic DNA, 1 µl of each forward and reverse primer (20 µM stock), 1 µl hydrolysis probe (5 µM stock), and GoTaq Probe qPCR Master Mix (Promega) with Nuclease-Free Water to make the final volume of 20 μl in a 96-well plate. Primers and probes for agrp2 exon 1 and exon 3 are listed in Table S.II.2. We used 40 cycles of amplification on a CFX96 Real-Time PCR Detection System (Bio-Rad) with the program: polymerase activation at 95 °C for 2 min, 40 cycles of denaturation (95 °C for 15 s) and annealing/extension (60 °C for 1 min). Ct values were defined as the point at which fluorescence crossed a threshold adjusted manually to be the point at which fluorescence rose above the background level. We – 41 –

⁕ Chapter II ⁕ generated the standard curve by using different amounts (2.5 ng, 5 ng, 10 ng) of genomic DNA from H. sauvagei (mixture of three individuals) as DNA template. Copy number variation for each sample was detected based on the position of the Ct value on the standard curve. We assayed copy number variation using three technical replicates for each sample and three biological replicates for each species. Primers and probes are listed in Table S.II.2.

Coverage Analysis

To screen for coverage difference in whole genome resequencing data, we used previously published WGS data sets as summarized in Table S.II.3 (Valente et al. 2014, Mcgee, Neches, and Seehausen 2016, Meier et al. 2017, Malinsky et al. 2018). We downloaded the data, and trimmed Illumina adapters from raw fastq reads using picard v2.17.11. Reads were then aligned to a curated genome of P.nyererei (Brawand et al. 2014) as agrp2 was split onto two scaffolds (exons 1 and 2 on scaffold_361/JH419567.1; exon 3 on scaffold_3/JH419209.1). The reverse complement of scaffold_361 was therefore concatenated with scaffold_3 (replacing scaffold_3 in the assembly). The original Scaffold_361 was then removed from the assembly. In addition, Scaffold_6781 (length: 1,095 bp) was removed from the assembly because it caused alignment problems as the nucleotide sequence was identical to a sequence 3′ of agrp2 exon 3 (but not overlapping with the duplication). The whole ∼25 kb agrp2 locus was replaced with a manually Sanger-sequencing-read-curated sequence previously published (Kratochwil et al. 2018). Using this approach, the assembly gap between the scaffolds was closed as well. We confirmed that the individual that has been used for Sanger Sequencing had no duplication of exon 3. Coverage was calculated for each position of the genome for each sample using samtools depth 1.9 (Li, Handsaker, Wysoker, Fennell, Ruan, Homer, Marth, et al. 2009). For this, we counted only reads with a base quality >20 and a mapping quality >30. Further data analysis was performed in R. Relative coverage was calculated by dividing the coverage at each position by the mean coverage of a 110 kb window around the agrp2 gene. Nonoverlapping sliding windows were calculated using the SlidingWindow function of the evobiR package (Blackmon and Adams 2015) with a window and step size of 100 bp. The following additional R packages were used: ggplot2 (Wickham, Chang, et al. 2019) and stringR (Wickham 2019). Transposable element position and identity of the agrp2 locus are summarized in Table S.II.4. We used both,

– 42 –

⁕ Chapter II ⁕ the Tilapia repeat masker (Shirak et al. 2010) and the standard repeat masker (Smit, Hubley, and Green 2015).

Exon Duplicate-Specific Expression

To determine the duplicate-specific expression of agrp2 exon 3a and 3 b, we first used the Pacbio assemblies of A. calliptera (GCA_900246225.3) and M. zebra (GCA_000238955.5) that resolved both exons, performed alignments using MAFFT and screened for paralogous sequence variants. Within the 3′-untranslated region (3′-UTR), we found two SNPs at position 1,048 and 1,115 of the agrp2 transcript (ENSHBUT00000024720) that are alternatively fixed between exon 3a and exon 3b of both species (Fig. II.3). To determine, which paralogous sequence is expressed across other cichlid species, we used Sanger sequencing of cDNA (description of extraction see above; primers see Table S.II.2). Second, we used RNA-seq to determine paralog-specific expression by calculating variant ratios on the two variable positions. RNA-seq libraries were prepared using TruSeq Stranded mRNA Library Prep Kit (Illumina) according to the manufacturer’s protocol with first-strand cDNA synthesis by GoScript Reverse Transcriptase (Promega). The final libraries were amplified using 15 PCR cycles and quantified and quality-assessed by Agilent DNA 12000 Kit with 2100 Bioanalyzer (Agilent). Indexed DNA libraries were normalized then pooled in equal volumes. Libraries were sequenced on an Illumina HiSeq 2500 (P. nyererei skin) or HiSeq X Ten platform (Pseudotropheus demasoni skin, Melanochromis auratus brain). Reads were mapped to the A. burtoni genome (as agrp2 is well annotated and the reference genome has only one copy of exon 3a) using STAR (Dobin et al. 2013). BAM files were extracted for the agrp2 transcript (ENSHBUT00000024720) and genotyped using FreeBayes with –pooled-continuous setting. Ratios between reference and alternative allele were calculated for the two positions using R and plotted as stacked bar plots.

Molecular Evolutionary Analyses

Maximum likelihood phylogenies of the agrp2 gene were based on genomic sequences (outgroups) and Sanger sequencing of agrp2 cDNA (synthesis of cDNA is described earlier; primers for amplification and Sanger sequencing are listed in Table S.II.2). Trees

– 43 –

⁕ Chapter II ⁕ were generated using PhyML (Guindon et al. 2010) with HKY85 substitution model, 100 bootstrap replicates, and Length/Rate optimization. We tested for evidence of the evolution of the agrp2 coding sequence using codon-based models in PAML (Yang 2007). Different random site models (e.g., M0, M1a, M2a) were compared using log-likelihood ratio tests (LRT) to test for the presence of sites classes that differ in dN/dS ratio. Specifically, we first tested for evidence of two site classes (i.e., M1a/M0), one assumed to be evolving under purifying selection (dN/dS < 1) and a second class evolving under neutral selection (dN/dS = 1). Second, we tested for the presence of positively selected sites (i.e., M2a/M1a)(Yang 2014). Then, using clade model C (CmC) in PAML, we tested the hypothesis that the agrp2 gene is evolving divergently between the lineage where the duplication of the third exon occurred (e.g., the radiating cichlids from Lakes Tanganyika, Malawi, and Victoria) compared with lineages without this duplication. The significance of the CmC model was determined conducting a LRT against a null model that allowed for different site classes, but not for divergence among lineages in the alignment (M2a_rel) (Weadick and Chang 2012).

Single Breakpoint Recombination Analysis

Because the functional importance of the third exon of the agrp2 gene, the finding that this exon duplicated, and that this duplicated copy could occasionally become functional through the loss of exon 3a, it is possible that exon 3 shows phylogenetic discordance with the other two exons in the gene. To test this hypothesis, we conducted a single breakpoint recombination analysis (Kosakovsky Pond et al. 2006) as implemented in HYPHY (Pond and Frost 2005, Pond and Muse 2005). Because we found evidence for a recombination breakpoint between exons 1 and 2 and exon 3 (see results), we repeated the molecular evolution analyses described earlier but independently for an alignment including only exons 1 + 2 and one including only exon 3 of the agrp2 gene.

– 44 –

⁕ Chapter II ⁕

Fig. II.1 Sequence conservation at the agrp2 locus. (A) Shuffle-LAGAN global pair-wise alignment of genomic sequences from cichlids as well as representatives from the Order Percomorpha and non-Percomorpha outgroups. The used reference is the genome of the Lake Malawi cichlid M. zebra. While exonic regions of agrp2 and the neighboring genes ripk2, snx16, agrp2, atp6V0d2 and an unknown gene are largely conserved across the Order Percomorpha and partially seemingly even in all teleosts, there are only a few conserved non-coding elements. One of these partially conserved elements [with stickleback and turbot; divergence times 105 – 154 MYA (Kumar et al. 2017) ] is 5’ of a previously described regulatory element of agrp2, enhA (black arrow). (B) Multiple sequence alignments of available assemblies of the agrp2 locus indicate several larger (>50bp) deletions (numbered from 1 to 10).

Results

Evolutionary History of the agrp2 Locus

The agrp2 locus has been previously implicated in regulating stripe patterns in the cichlid radiations of Lakes Victoria, Malawi, and Tanganyika. To gain a deeper understanding of the evolution of this locus, we investigate it in a comparative approach across ten teleost species (Fig. II.1A). The gene agrp2 is flanked by the ripk2 (Receptor interacting serine/threonine kinase 2) and snx16 (sorting nexin-16) on the 5′ side, and atp6V0d2 (v- type proton ATPase subunit d 2) and “unk,” an unknown gene 3′ of agrp2. Beyond cichlids, mainly coding sequences are conserved, with strong signals of conservation – 45 –

⁕ Chapter II ⁕ for ripk2, but also atp6V0d2 and agrp2. Interestingly, alignments show only low conservation 5′ of a skin-specific regulatory region (enha) that has been functionally linked to agrp2 expression (Kratochwil et al. 2018). This possibly suggests a larger regulatory module within this intron that might also contain elements that drive the brain- specific expression that shows greater evolutionary conservation (Zhang et al. 2010, Shainer et al. 2017). Within the family Cichlidae, we find several conserved blocks in the noncoding sequence in 5′ and 3′ as well as the intronic sequences including the enha regulatory sequence (Fig. II.1B).

Description of a Novel Noncoding Exon of agrp2

To be able to better interpret structural variation at the agrp2 locus within African cichlids, we first comprehensively described the transcript annotation of the agrp2 gene. Previously, it has been shown that agrp2 mRNA is mainly expressed in the brain and in the skin (Zhang et al. 2010, Shainer et al. 2017, Kratochwil et al. 2018). For the agouti signaling protein (asip), another member of the agouti family different tissue-specific isoforms have been reported (Mallarino et al. 2017). Therefore, we asked whether agrp2 also has different isoforms that may act—besides the described cis- regulatory variation—as an additional source of differential regulation between tissues (i.e., brain and skin) and species (i.e., striped and nonstriped species). As alternative

Fig. II.2 Isoforms of agrp2. (A) Using 5’ and 3’RACE as well as PCRs on cDNA we characterized two isoforms of agrp2. Both have an identical coding sequences but differ in their 5’ UTR. (B) Isoform 1 has a 72bp 5’UTR. (C) Isoform 2 has an additional non-coding exon with 44bp of the 5’UTR, the second exon has only a short 22bp UTR.

– 46 –

⁕ Chapter II ⁕ splicing and the evolution of novel isoforms could be affected by structural variation in the agrp2 locus, we aimed to identify the full-length transcripts in an effort to discover potentially unknown exons. Based on 5′RACE, we amplified the full length 5′ sequence of agrp2 from both brain and skin tissue of the striped species H. sauvagei and nonstriped species P. nyererei. Sanger sequencing provided evidence for two isoforms (Fig. II.2A). Isoform 1 has three exons: the first exon consists of a 72 bp 5′-UTR and 109 bp of the coding sequence (Fig. II.2B and S.II.1). Isoform 2 has an additional noncoding first exon with 44 bp of the 5′-UTR, the second exon has only a short 22 bp UTR and the 109 bp coding sequence (Fig. II.2C and S.II.1). The isoforms can be found in both species and in both brain and skin tissue, as confirmed by isoform-specific PCR (Fig. S.II.1). Additionally, we performed 3′ Rapid Amplification of cDNA Ends (3′RACE) to amplify the full length 3′-UTR of agrp2 (Fig. S.II.2). Sequencing revealed a total 3′-UTR length of 616 bp in P. nyererei and 579 bp in H. sauvagei, respectively. Within the UTR, we found three potential cleavage sites with an AAUAAA or AUUAAA motif (Fig. S.II.2). Sanger sequencing gave no indication for additional exons between exons 1 and 2 as well as exons 2 and 3 (Fig. S.II.1 and S.II.2) in any of the >30 analyzed species.

Characterization of Insertions and Deletion of the agrp2 Locus

To more comprehensively describe structural variation (i.e., larger insertions and deletions; indels or duplications), we aligned available cichlid fish genomic segments of the agrp2 locus (Brawand et al. 2014, Conte et al. 2017, Kratochwil et al. 2018, Conte et al. 2019) including seven African cichlid species: the geographically widespread O. niloticus, the Lake Tanganyika endemic N. brichardi, nonendemic of the Lake Tanganyika region A. burtoni, nonendemic of the Lake Malawi region A. calliptera, the Lake Malawi endemic M. zebra, and the Lake Victoria endemics P. nyererei and H. sauvagei. In our analyses, we found 12 larger indels (>50 bp) localized in intronic sequence as well as directly upstream and downstream of the agrp2 gene in this set of African cichlid fishes (Fig. II.1B). Based on maximum parsimony most of the indels within the analyzed 10–15 kb locus (10/12) likely arose within the radiating cichlid lineages (Fig. S.II.6). Based on previous studies that performed genome-wide characterizations of indels and structural variations (Fan and

– 47 –

⁕ Chapter II ⁕

Meyer 2014, Penso-Dolfin et al. 2018), density of indels at the agrp2 locus is four to five times higher than genome-wide average. Several of the indels overlap with putatively functional DNA regions. Two independent deletions (nos. 1a and 1 b; compared with the outgroup O. niloticus) are located within the promoter region of agrp2, one in the Lake Malawi species A. calliptera (296 bp deletion; 418 bp 5′ of the start codon) and one in the Lake Tanganyika species N. brichardi (793 bp deletion; 157 bp 5′ of the start codon) (Fig. II.1B). Two further indels (nos. 5 and 6) overlap with a region within intron 1 that has been previously characterized as a cis-regulatory element of agrp2 containing several alternatively fixed alleles associated with stripe presence/absence in Lake Victoria cichlids (Kratochwil et al. 2018) (Fig. II.1B). Element number 5 is a 145 bp short interspersed nuclear element retrotransposon-derived insertion (Table S.II.1 and S.II.4) specific to haplochromine cichlids (and therefore missing in the genomes of N. brichardi and O. niloticus) (Fig. II.1B). The 648 bp element number 6 is only present in O. niloticus (and more distantly related species, compared with the Lake Victoria and L. Malawi species, such as the neotropical cichlid Amp.citrinellus; BLAST E value: 7e-66) therefore suggest that it constitutes a deletion specific to the radiating cichlids including Lake Tanganyika and Haplochromine cichlids (Fig. II.1B). Two of the deletions (451 and 947 bp) were variable even within the extremely young Lake Victoria radiation [∼10,000 years; (Johnson et al. 1996, Elmer et al. 2009)] and are of particular interest as they were specific to the striped species H. sauvagei (Fig. II.1B). To confirm a potential association with the stripe phenotype, we conducted PCRs to assay the variation in length. Indeed, amplicons of H. sauvagei were shorter than those of P. nyererei, also those of Haplochromis chilotes but other striped species such as Haplochromis serranus and Haplochromi sthereuterion as well as all nonstriped species showed no evidence for a deletion (Fig. S.II.7). The lack of association is in line with the previous hypothesis that variation in stripe patterns of Lake Victoria cichlids is mainly associated with the regulatory element in the first intron of agrp2 (Kratochwil et al. 2018).

Tandem Exon Duplication of agrp2 in Several African Cichlids

The largest insertion (1,811 bp) is directly upstream of exon 3 and could be only found in the two long-read chromosome-level assemblies of M. zebra and A. calliptera

– 48 –

⁕ Chapter II ⁕

(Fig. II.1B), yet sequences contained gaps in several of the short-read genomes suggesting assembly problems. To visualize repeats that could cause such assembly problems and to gain insights into the origin of the insertion, we generated syntenic dot plots of the available long-read assemblies (O. niloticus, M. zebra, and A. calliptera). The alignments provide evidence for several duplicated regions in the M. zebra and A. calliptera genomes that are unique in the O. niloticus genome (Fig. II.3A–C). Although most duplications are small (<50 bp), we found one ∼1.6 kb tandem duplication that included exon 3 of agrp2. Using PhyML, we generated a maximum likelihood phylogeny of exon 3 and the two copies (exon 3a and exon 3 b; Fig. II.3D, E) that were found in M. zebra and A. calliptera. Based on the phylogeny, parsimony suggests only one duplication event that gave rise to exon 3 b (bootstrap support 92%). In addition, the phylogeny provides no evidence of gene conversion or concerted evolution. To test if the duplication is common across the radiation of Lake Malawi (where M. zebra and A. calliptera are from) but potentially also Lakes Victoria and Tanganyika, we used three complementary approaches: 1) a PCR-based approach, 2) a TaqMan probe assay, and 3) reanalysis of available genome-resequencing data. In an effort to identify the phylogenetic timing when the exon 3 tandem duplication occurred, we first designed primers that specifically detect the duplication, both by designing primers that amplify the region between the duplicates (Fig. II.3E and F) or by amplifying the whole fragment containing the duplications (Fig. II.3E and G). The results of the PCR approach uncover a surprising degree of standing genetic variation within species as well as substantial variation between species of the flocks of Lakes Victoria, Malawi, and Tanganyika (Fig. II.3F, G; Table S.II.3). In total, we found 16 species without duplication and 25 species with a duplication; and in 2 species from Lake Victoria (P. nyererei and Pundamilia pundamilia), we found individuals with and without duplication suggesting standing genetic variation in some species. More interestingly, duplications could be found across cichlid radiations from Lake Tanganyika, Malawi, and Victoria. Second, as the PCR assay was not sensitive enough to detect heterozygote individuals, or individuals with more than two duplicates, we used a TaqMan probe assay on a subset of species to confirm the results and to determine the exact number of repeats (Fig. II.3H). The assay was performed using probes for exon 1 (that had no evidence of

– 49 –

⁕ Chapter II ⁕

Fig. II.3 The third exon of agrp2 was tandemly duplicated. (A) A syntenic dotplot of the agrp2 loci of O. niloticus against itself shows no evidence for duplicated sequences. (B) In contrast to that the genome of A. calliptera show multiple small duplications (red) and inverted duplications (blue) and a 1.6 kb tandem duplication that gives rise to a duplicated exon 3 (exon 3a and exon 3b). (C) The genome of M. zebra shows a similar pattern of duplications, including a duplicated exon 3. (D) Maximum likelihood phylogeny of the genomic sequences for exon 3 (including 3a and 3b) suggest a common origin of the duplication. (E) PCR design to confirm duplications of exon 3. (F) Short PCRs, amplifying a ~1.6kb sequence between the two copies confirms the duplication in several, but not all species. The presence also varies within populations (e.g. in P. nye and P. pun). (G) Long-range PCRs including sequence of both copies of exon 3 provide additional confirmation for the duplication. (H) TaqMan probe assay to quantify the copy number comparing exon 1 (no evidence for duplication) and exon 3. (I) Copy number quantification based on coverage of genome resequencing data. Thick lines indicate the average for Lake Malawi (blue) and Lake Victoria species (orange). Thin lines represent individuals. Relative coverage is increased in many individuals at a position corresponding to the estimated size of the duplication (dashed lines). Yellow boxes indicate repetitive elements that likely explain the coverage drop 5’ of the duplication in Lake Malawi. (J) Histogram representation of the Lake Victoria species’ mean relative read coverage within the duplicated region. Most species have one or two copies. (K) Same representation for Lake Malawi illustrating that the duplication is more common. Most species have two exon 3 copies, some even more than three. Abbreviations: A. bur, Astatotilapia burtoni; A. cal, Astatotilapia calliptera; C. euc, Cheilochromis euchilus; H. lat, Haplochromis latifasciatus; H. sau, Haplochromis sauvagei; J. orn, Julidochromis ornatus; L. mul, Lamprologus multifasciatus; Lab. cae, Labidochromis caeruleus; M. zeb, Maylandia zebra; Me. aur, Melanochromis auratus; N. bri, Neolamprologus brichardi; N. cau, Neolamprologus caudopunctatus; O. nil, Oreochromis niloticus; O. var, Oreochromis variabilis; P. nye, Pundamilia nyererei; P. pun, Pundamilia pundamilia; Ps. cya, Pseudotropheus cyaneorhabdos; T. vit, Telmatochromis vittatus. duplication in species for which long-read sequence data are available and thereby served as a control) and exon 3. Our analysis confirmed the PCR results and support tandem exon duplications in several species of the Lake Victoria radiation – 50 –

⁕ Chapter II ⁕ including Haplochromis latifasciatus, P. nyererei (1/3 individuals), endemic to Lake Malawi Labeotropheus caeruleus and Pseudotropheus cyaneorhabdos. There is some support for more than two copies in P. nyererei and Pse. cyaneorhabdos (Fig. II.3H). Third, as a further independent approach, we realigned available cichlid genomes (Poletto, Ferreira, and Martins 2010, Mcgee, Neches, and Seehausen 2016, Meier et al. 2017, Malinsky et al. 2018) to the P. nyererei reference genome, a genome without the duplication (Brawand et al. 2014) and analyzed the relative coverage across the agrp2 region (Table S.II.3). Here, we would expect to find a relative coverage of ∼1 for individuals without duplication, a relative coverage of ∼2 for individuals with a homozygous duplication, ∼1.5 for heterozygous individuals and >2 for individuals with more than two copies. Indeed, using this approach, we find support for tandem exon duplications of exon 3 (Fig. II.3I), with some individuals being likely heterozygote (e.g., Haplochromis paucidens and Haplochromis vittatus, both endemics to Lake Victoria) or having even more than two duplications (e.g., Thoracochromis pharyngalis) (Table S.II.3). Generally, duplications and more than two tandem duplicates seem to be more prevalent in the Lake Malawi cichlid radiation (Fig. II.3J and K).

Transcription Bias and Resurrection of a Nontranscribed Exon 3 Copy

To assess potential functional effects, we first tested whether both or only one of the copies of exon 3 are expressed. Using the genomes of M. zebra and A. calliptera, we found copy-specific, paralogous sequence variants at position 1,048 and 1,115 within the UTR region of agrp2 (Fig. II.4A). We synthesized cDNA from RNA extractions from the skin of 33 species. Sanger sequencing shows that 32 of the 33 species, independent of the number of duplicates, showed the exon 3a-specific paralogous sequence variants in the chromatogram trace (Fig. II.4B). This suggests that in species with a duplicated third exon, only the 5′ exon 3 (exon 3a) is transcribed. Interestingly, one species, Mel.auratus, a species with a single copy had the paralogous sequence variant specific for exon 3 b. This suggests a secondary loss of exon 3a and a resurrection of the usually nonexpressed exon 3 b. To confirm the results with a more sensitive approach, we used RNA-seq data for three species (P. nyererei [n = 4], Pse.demasoni [n = 5], and Mel. auratus [n = 4]) and estimated the ratio of the paralogous sequence variants at the cDNA positions 1,048 and 1,115 that are located within the 3′-UTR (Fig. II.4C). The analysis confirmed that in both P. nyererei and Pse. demasoni exon 3a is predominantly

– 51 –

⁕ Chapter II ⁕

Fig. II.4 Expression bias towards exon 3a and resurrection of exon 3b (A) Two paralogous sequence variants identify the two exon copies exon 3a and 3b. The variant of exon 3a corresponds to the ancestral alleles as also found in O. niloticus. (B) In species with two copies (here M. zebra) exon 3a is mainly expressed as indicated by the lack of the exon 3b specific chromatogram trace. Species that lost (or never had) exon 3b have the exon 3a variant. One exception is Melanochromis auratus that only has one copy, but in this species the sequence corresponds to exon 3b, suggesting a secondary loss of exon 3a. (C) RNA-seq data confirms the expression bias towards exon 3a (no evidence of exon 3a in P. nyererei; some weak expression in Ps. demasoni) and the loss of the exon 3a paralogous sequence variants in Me. auratus. Expression ratio was estimated based on two paralog-specific SNPs in the 3’UTR. Error bars indicate standard deviation.

expressed. In Pse. demasoni, two individuals showed some expression of exon 3 b (22% and 25%). We found no sequence variation (i.e., lack of a 3′-UTR cleavage site) that could explain these interindividual differences. In Mel. auratus, only exon 3 b is expressed (as exon 3a has been lost). If we assume that also in Mel. auratus exon 3a was initially the expressed exon, this suggest a resurrection of the nontranscribed exon 3 b.

Signals of Selection

Such a scenario where a nontranscribed, likely neutrally evolving exon copy (exon 3 b), could occasionally become functional again through the loss of exon 3a (as shown for Mel. auratus), but also the observed evolutionary dynamics at this locus could promote the accumulation of genetic variation that positive selection could act upon. Therefore, we analyzed the molecular evolution of agrp2 (always using all three exons of the transcribed sequence across species) to determine if variation is neutral or if it had evolved adaptively (Figs. II.5A and II.6). We found evidence that many codons in the 2 gene (∼33%) are evolving under neutrality (dN/dS ≈1; LRTM1a/M0: X (1)= 60.27, P < 0.0001), yet, we found no evidence for the presence of positively selected sites 2 (LRTM2a/M1a: X (2)= 0, P = 1). In addition, we did not find evidence that the agrp2 gene is evolving under divergent selection pressures in the cichlid lineage with the exon 3

– 52 –

⁕ Chapter II ⁕

duplication compared with other lineages in our alignment (LRTCmC-Great 2 Lakes/M2a_rel: X (1)= 0.46, P = 0.50). Yet, this analysis does not exclude that in some instances amino acid changes contributed to adaptive or nonadaptive pigmentation differences (Figs. II.5B–J and II.6). Given the evolutionary history and characteristics of the third exon of agrp2 in African cichlids (e.g., exon duplication, resurrection of nontranscribed exons, different functional importance of exon 3 compared with exons 1 and 2), it could be expected that this exon has evolved under divergent selection pressures relative to the other exons in the gene, potentially resulting in phylogenetic discordance among exons (Beltrán et al. 2002). We found some support for this prediction. The single breakpoint recombination analysis performed in HYPHY, found evidence for a breakpoint at base-pair position 185(at the border between exons 2 and 3; ΔAIC = 19.78, Model support =1; although the signal was lost when considering cAIC). Thus, we conducted exon-specific analyses of nucleotide substitutions for the agrp2 gene. As for the whole gene, we found evidence 2 for some codons neutrally evolving, both in exons 1 + 2 (LRTM1a/M0: X (1)= 11.48,

Fig. II.5 Phylogeny of the agrp2 gene across cichlids. (A) Maximum likelihood phylogeny using agrp2 cDNA. Points connote the presence/absence of the duplication and origin. Numbers indicate bootstrap values >60. (B-J) Photographs of cichlid fishes studied in this research with information about the duplication and non-synonymous mutations (see Fig. II.6).

– 53 –

⁕ Chapter II ⁕

2 P = 0.0007) and in exon 3 (LRTM1a/M0: X (1)= 40.73, P < 0.0001), but no evidence for positive selection (both P > 0.5). Interestingly, however, we found the proportion of sites evolving neutrally to be larger for exons 1 + 2 (40%) than for exon 3 (9%), suggesting that the functional importance of this exon might result in purifying selection across most 2 codons. We found no evidence that exons 1 + 2 (LRTCmC-Great Lakes/M2arel: X (1)= 2 0.55, P = 0.46) or exon 3 (LRTCmC-Great Lakes/M2arel: X (1)= 1.49, P = 0.22) are evolving divergently in radiating African cichlids compared with lineages where the duplication of the third exon was not found (i.e., there is no evidence of codons evolving under different selection pressures in the compared lineages).

Fig. II.6 Protein sequence evolution of agrp2 across cichlid fishes. Protein alignments show several amino acid substitutions across cichlids (at twelve positions in Lake Tanganyika cichlids, at two positions in Lake Malawi and Victoria cichlids). Putative disulfide bonds that are a characteristic of agouti proteins are indicated at the top. The involved Cysteines are highly conserved.

– 54 –

⁕ Chapter II ⁕

Discussion

It was Susumu Ohno who postulated 50 years ago how gene duplication might serve as a source of evolutionary novelties (Ohno 1970). As one copy is freed from stabilizing selection, the other can accumulate mutations “more freely” and evolve a novel function (neofunctionalization) as one possible consequence. Although there are many examples for the dynamics and evolutionary significance of gene duplications (Perry et al. 2007, Vonk et al. 2013, Denoeud et al. 2014, Ishikawa et al. 2019, Musilova et al. 2019), exon duplications have received considerably less attention (Kondrashov and Koonin 2001, Letunic, Copley, and Bork 2002, Keren, Lev-Maor, and Ast 2010, Lambert et al. 2015, Rogers, Shao, and Thornton 2017). Especially tandem duplications of the terminal exon have been barely investigated in an evolutionary context. A lone exception is a recent study that described a duplication of the terminal exon of the alternative oxidase in oysters that has been involved in adaptation to environmental stress (Liu and Guo 2017). Here, we found a tandem duplication of the third exon of agrp2 that likely occurred >8–12 Ma early in the evolution of the African cichlid fish radiations before the oldest of these, that from Lake Tanganyika, diversified (Fig. II.7). The duplication is not fixed, as its presence varies across the radiations of the Lakes Tanganyika, Malawi, and Victoria and even is polymorphic within some species. Furthermore, we found evidence for copy number variation (up to four or five copies) (Fig. II.7). In previous short-read assembly genomes (Brawand et al. 2014), this duplication was not resolved, illustrating that such instances might have been systematically overlooked as well as their evolutionary and importance went unrecognized. In contrast, long-read assemblies (e.g., using PacBio or Nanopore technologies) have now the power to uncover such

Fig. II.7 Summary of the evolutionary history of the agrp2 gene. Exon 3 was duplicated before the radiations of Lake Tanganyika cichlids more than 8–12 million year ago giving rise to exon 3a and 3b. Exon 3b was partially either lost again or the non- duplicated version was partially maintained through incomplete lineage sorting. Additional duplications occurred generating more copies of exon 3. At least in one species (Me. auratus) exon 3a was subsequently lost.

– 55 –

⁕ Chapter II ⁕ duplications (Fig. II.2B and C). Such assemblies therefore permit studying their evolutionary relevance. Despite not finding any evidence for diversifying selection, evolution of novel isoforms or a link to particular color patterns (Fig. II.5B–J), we discover a surprising dynamic of gains and losses of exon 3 that also includes (at least in one instance) the resurrection of the nonfunctional exon 3 b (Fig. II.7). As we only looked at a small selection of the over 1,200 species of East African cichlids, the dynamics of exon duplication and loss might indeed contribute to the diversity of cichlid color patterns in some instances. Clearly, an increasing availability of RNA-seq data and long-read genome data will permit much more comprehensive analyses in the future. Regulatory variation of agrp2 has been previously shown to be linked to absence and presence of stripe patterns across African cichlids (Kratochwil et al. 2018). High expression of agrp2 represses stripe patterns, while low levels lead to their formation. Moreover, also CRISPR-Cas9 knockouts of agrp2 (which as such constitutes an experimentally introduced coding variation) cause the reappearance of stripe patterns in nonstriped cichlids. This suggests that coding variation might be equally potent in driving diversification in stripe patterns. Loss- but also gain-of-function mutations in both the coding sequences and 3′-UTR of agrp2 could accumulate in the untranslated copy exon 3 b, and be resurrected through subsequent deletion of exon 3a. This, in turn, might lead to phenotypic diversification. Interestingly, the C-terminus encoded by exon 3 shows high evolutionary conservation across all agouti family members (agrp2/asip2b, agrp1, asip1, asip2/asip2a) as it is the part that binds to and antagonizes melanocortin receptor function (Kaelin et al. 2008). Mutations in this exon are therefore likely to have direct effects on the melanocortin pathway and thereby color pattern formation. Our RNA-seq analysis suggests that only the first copy, exon 3a is expressed. This is in contrast to previous work that has shown alternative splicing of terminal exon copies, that can be even regulated through environmental factors (Liu and Guo 2017). As two out of the five Pse. demasoni individuals that we analyzed for relative expression of exon 3a/3b showed a low level of expression of exon 3b, it is possible that cryptic genetic variation or environmental factors might trigger a stronger expression of exon 3 b. As copy number variations occur frequently through unequal crossing over, potentially leading to loss or gain of tandem copies (Hastings et al. 2009) such a tandem duplication could facilitate neofunctionalization or even loss-of-function of agrp2,

– 56 –

⁕ Chapter II ⁕ especially if populations experience strong bottlenecks (e.g., during colonization events). Such a scenario would have implications similar to recent work from sticklebacks (Kratochwil and Meyer 2019, Xie et al. 2019). This study describes how repeated loss- of-function deletions of regulatory elements of pitx1 cause the independent loss of pelvic fin spines in sticklebacks. These recurrent deletions are driven by repeats that result in a more fragile, double-strand breaking DNA-formation (Z-DNA) that in turn results in an elevated mutation rate at this locus. This repeatedly led to loss-of-function mutations of the regulatory element necessary for pelvic fin spine development. In summary, we conclude that the tandem duplication of agrp2 exon 3, or more generally tandem duplications at genomic hotspots are important loci to analyze in light of their evolutionary importance as they might facilitate diversification and adaptation in a similar manner as gene duplications. This analysis crucially depends on state-of-the-art technologies including PacBio and Nanopore sequencing, the so called third revolution in sequencing technology (Van Dijk et al. 2018), sensitive enough to detect such variations. In the future, beyond the characterization of structural variation, insertions, and duplication, functional experiments have to more specifically address the role and phenotypic impact of such mutational events. Many of those will be difficult to assess through comparative approaches, but will need genetic manipulations, either through hybridization experiments or functional approaches including CRISPR-Cas9 genome editing that have become amenable in nonmodel organisms as cichlid fishes (Juntti et al. 2016, Kratochwil et al. 2018) and that steadily become more precise (Anzalone et al. 2019).

– 57 –

⁕ Chapter III ⁕

Chapter III

The Teleost-Specific Genome Duplication and Color Pattern Evolution in Modern Fishes: Retention of Functional Conservation, Neo-functional Divergence and Sub-functionalization of Agouti Genes

Yipeng Liang, Maximilian Grauvogl, Axel Meyer, Claudius F. Kratochwil Manuscript in preparation for submission

Abstract

While color patterns are highly diverse across the animal kingdom, certain patterns as countershading and stripe patterns have evolved many times independently. Across vertebrates, members of the agouti gene family have been repeatedly associated with the evolution of both patterns. Here we study the functional conservation and divergence of the two skin-specific agouti family genes asip1 and agrp2/asip2b across the radiation of teleost fishes. We show that the dorso-ventral expression of asip1 is highly conserved, but additionally we uncover a previously unknown association between stripe–interstripe patterning and the expression of both paralogs asip1 and agrp2. In some species, including the zebrafish Danio rerio, the genes partially show complementary and overlapping patterns suggesting functional redundancy. The paralog agrp2 has been previously linked to the loss of stripe patterns in East African cichlid fishes. Based on species pairs of two additional lineages we find support that regulatory changes of the ‘stripe repressor’ agrp2 might have repeatedly led stripe to pattern losses and gains across the teleost fish radiation.

– 58 –

⁕ Chapter III ⁕

Main text

Pigment patterns as countershading and stripes have evolved repeatedly across vertebrates and constitute powerful anti-predatory strategies (Thayer 1909, Cott 1940, Barlow 1972, Kapp et al. 2018). Within the last decade the genetic basis of such color patterns and mutations driving their variation in the wild have been identified (Manceau et al. 2011, Mallarino et al. 2016, Cal, Megias, et al. 2017, Haupaix et al. 2018, Kratochwil et al. 2018). One of the molecular pathways that has been repeatedly associated with pigment pattern phenotypes is the melanocortin system (Eizirik et al. 2003, Candille et al. 2007, Gross, Borowsky, and Tabin 2009, Linnen et al. 2009, Rosenblum et al. 2010, Manceau et al. 2011, Ceinos et al. 2015, Haupaix et al. 2018, Kratochwil et al. 2018). When bound by the melanocyte-stimulating hormone (MSH), the Melanocortin receptor 1 (Mc1r) modulates pigment cell proliferation or pigment biosynthesis and transport, while this pathway is blocked by antagonists of the agouti gene family including the Agouti-signal protein (Asip) (Mills and Patterson 2009, Manceau et al. 2010, Cal, Suarez-Bregua, et al. 2017). While mutations in the melanocortin receptor cause loss of pigmentation, regulatory variation in the antagonist Asip/asip1 (gene name in tetrapods/teleosts) has been shown to underly changes in dorso- ventral patterning by inhibiting dark pigmentation in ventral areas (Manceau et al. 2011, Linnen et al. 2013, Ceinos et al. 2015). Recently, Asip has been also linked to stripe patterns, where high expression in the light areas between melanic stripes might locally inhibit dark pigmentation and thereby shaping the pattern (Mallarino et al. 2016, Haupaix et al. 2018). More recently, a paralog of Asip/asip1, the agouti-related peptide 2 (agrp2, also called asip2b) has been shown to inhibit stripe patterns in African cichlids, however not by spatial variation in expression but by acting as a global inhibitor of stripe patterns (Kratochwil et al. 2018). To investigate the functional conservation and diversity of the agouti family genes in shaping color pattern formation and color pattern presence–absence we used a comparative approach across the oldest and largest vertebrate radiation: the teleost fishes, Teleostei. We thereby aimed to investigate three questions: 1) To what extent the function and expression pattern of Asip (in teleosts asip1) in shaping dorso-ventral patterning and stripe patterns is conserved across the phylogeny, 2) if the previously reported role of agrp2 in inhibiting stripe patterns was repeatedly coopted during teleost

– 59 –

⁕ Chapter III ⁕ fish evolution and 3) if we find instances where the paralogs agrp2 and asip1 have redundant or complementary functions providing insights into the ancestral mechanism of color pattern formation in vertebrates.

Functional conservation of asip1 across vertebrates

To address the putative role of asip1 across teleost fishes we examined spatial expression differences using quantitative polymerase chain reaction (qPCR) along the dorso-ventral axis of the integument of different lineages of teleost fishes including two East African cichlids (Melanochromis auratus and Pseudotropheus cyaneorhabdos), a West African cichlid (Pelvicachromis pulcher), a Neotropic cichlid (Trichromis salvini), a Betta fish (Betta pi) and the zebrafish (Danio rerio). Previous studies have shown that expression of asip1 varies along the dorso-ventral axis, also in fishes, including the zebrafish and flatfishes (Scophthalmus maximus and Solea senegalensis) (Cerda-Reverter et al. 2005, Kurokawa, Murashita, and Uji 2006, Guillot et al. 2012, Ceinos et al. 2015, Cal, Megias, et al. 2017, Cal et al. 2019), while it is unknown if asip1 is associated with stripe patterns in teleosts. We quantified asip1 mRNA levels from melanic stripes and non-melanic regions from six striped teleosts (Fig. III.1A) using qPCR to be able to assess both a) differences along the dorso-ventral axis and b) differences between melanic and non- melanic regions. In general, asip1 showed differential expression in all tested species (all: P < 0.01, ANOVA; Table S.III.3–8). Compatible with the role of asip1 in countershading, for all species the lowest expression was found in the dorsalmost regions (DOR or DLS) while the highest expression levels of asip1 were found in the most ventral ares (VEN) (Fig III.1B–G; Tables S.III.3–8). This suggest, that the dorso-ventral expression gradient is highly conserved across the teleost radiation. Yet, we did not find a continuous increase of expression from ventral to dorsal, but expression was in most species and across the dorso-ventral axis higher in non-melanic regions than in the adjacent melanic stripes. Because of the relatively low expression values in dorsal regions, these values — although the differences were not consistent across species — were however not significant (Fig. III.1B–G; Table S.III.3–8). However, if we performed statistical test by combining species-specific mean values of the four species with two stripes (M. auratus, Ps. cyaneorhabdos, Pe. pulcher and T. salvini; Fig. III.1B–E), we found asip1 expression to be 3.6 to 9.4-fold and significantly higher in the dorsal interstripe than in the adjacent DLS and MLS (Tukey HSD: P = 0.008–0.032; Fig. III.1H; Table S.III.9). Interestingly,

– 60 –

⁕ Chapter III ⁕

Fig. III.1 Expression differences of asip1 and agrp2 associate with stripe pattern formation and evolutionary loss across teleost fishes. (A) Phylogenetic relationship of focal striped and nonstriped species of this study. Melanochromis auratus and Pseudotropheus cyaneorhabdos are two striped cichlids from East African cichlid radiations. Nonstriped Nanochromis parilus and striped Pelvicachromis pulcher are two West African cichlids. Trichromis salvini is a neotropical cichlid. Please note that all cichlids have two stripes. Betta pi is part of the genus Betta with three horizontal stripes. Danio tinwini (gold-ring danio) is a species of the genus Danio from Myanmar with melanic spot patterns. Danio rerio (zebrafish) is a model organism and has five horizontal stripes. Divergence times were taken from (Kumar et al. 2017, Hughes et al. 2018). (B–G) Horizontal stripe patterns and the spatial expression of asip1 and agrp2 for M. auratus (B), Ps. cyaneorhabdos (C), Pe. pulcher (D), T. salvini (E), B. pi (F) and D. rerio (G). Left panels, skin of striped species showing the characteristic stripe patterns; Middle panels, spatial expression ratio of asip1 along dorsoventral axis; Right panels, spatial expression ratio of agrp2 along dorsoventral axis. P- values are based on ANOVA and Tukey–Kramer post hoc tests (full data see Table S.III.3–8). Error bars indicate means ± SD. ***P < 0.001; **P < 0.01; *P < 0.05. (H) Expression difference of asip1 in the species with two stripes. P-values are based on ANOVA and Tukey–Kramer post hoc tests (full data see Table S.III.9). Each dot represents the mean value of one species. Black/gray lines depict the mean of the four species. ***P < 0.001; **P < 0.01; *P < 0.05. (I, J) Whole skin expression comparison of agrp2 of nonstriped and striped species: nonstriped N. parilus and striped Pe. pulcher (I) and nostriped D.tinwini vs. striped D. rerio (J). Differences in (I, J) were tested by two-tailed t-test, n = 6 in (I) and 5 in (J) (individual dots). Error bars indicate means ± SD. ***P < 0.001; *P < 0.05. Abbreviations: DOR, region dorsal to dorsal-most stripe; DLS, dorsolateral stripe; DIN, dorsal interstripe; MLS, midlateral stripe; VIN, ventral interstripe; VLS, ventralateral stripe; VEN, region ventral to ventral-most stripe. – 61 –

⁕ Chapter III ⁕ asip1 expression was also slightly elevated in interstripes of B. pi, but mainly in dorsal regions of the integument (Fig. III.1F; Table S.III.7). In contrast, such elevation of asip1 mRNA levels in interstripes was only found in the more ventral regions of D. rerio (Fig. III.1G; Table S.III.8). In summary, our results suggest a striking conservation of the dorso-ventral asip1 expression gradient as well as stripe-interstripe differential expression across teleosts and — based on previous findings (Mallarino et al. 2016, Haupaix et al. 2018) — vertebrates more generally. agpr2 is a common “stripe repressor” across teleost radiations

Another member of the agouti gene family, the agouti-related peptide 2 (agrp2), has been recently shown to facilitate loss of stripe patterns in African cichlid fishes: high expression of agrp2 blocks formation of stripe, while low expression permits stripe presence. The expression of agrp2 is ubiquitous across the skin — it therefore has no function in shaping the pattern, but only acts as a global inhibitor of stripe patterns. What remains unclear is, if this function of agrp2 is specific to the analyzed East African cichlids, or if agrp2 also drives stripe loss and gain in other lineages. To test this hypothesis we identified species pairs that are closely related and differ in stripe presence/absence. We found two West-African cichlids of the subfamily Pseudocrenilabrinae, the striped Pe. pulcher and the nonstriped Nanochromis parilus of [divergence time to East African, haplochromine cichlids is about 25 million years (Genner et al. 2007, Schwarzer et al. 2014)], and two Danioninae, the striped D. rerio and the nonstriped D. tinwini (Mccluskey and Postlethwait 2015) (Fig. III.1A). Comparative comparison of agrp2 skin expression using qPCR revealed similar patterns as described in East African cichlids (Kratochwil et al. 2018). Within both Pseudocrenilabrinae and Danioninae agrp2 mRNA levels were significantly higher in the nonstriped species than in the striped species (pairwise t-test, P < 0.001 in Pe. pulcher vs. N. parilus , P < 0.05 in D. rerio vs. D. tinwini; Fig. III.1I, J; Table S.III.1, S.III.2). Our results therefore suggest that stripe loss is repeatedly associated with agrp2 variation, not only in cichlids but across teleost fishes.

Functional divergence of agrp2 in teleost fishes

Although previous evidence showed that agrp2 is ubiquitously expressed across the skin [(as shown for Lake Victoria cichlids (Kratochwil et al. 2018, Liang et al. 2020)], we

– 62 –

⁕ Chapter III ⁕ wanted to test if agrp2 might show more variation in other lineages, suggesting either additional roles in color pattern formation and/or shows more similarities with the paralog asip1, thereby providing evidence into its ancestral function and expression pattern. Hence, we performed qPCRs for agrp2 in the same manner as we did for asip1. We found no expression difference across dorso-ventral axis in B. pi, M. auratus, Ps. cyaneorhabdos and Pe. pulcher (ANOVA: P = 0.168–0.932; Fig. III.1B–D, F; Table S.III.3–S.III.5, S.III.7), indicating that agrp2 — as previously suggested — likely plays no role in shaping the horizontal stripe patterns among these species (Kratochwil et al. 2018, Liang et al. 2020). In contrast, within the neotropic cichlid T. salvini, agrp2 mRNA levels were higher in non-melanic regions than in melanic stripes (ANOVA: P < 0.01; Fig. III.1E; Table S.III.6), similar as observed for asip1. The expression variation of agrp2 in zebrafish is even more intriguing (Fig. III.1G). We observed expression differences between stripes and interstripes (ANOVA: P < 0.001; Fig. III.1G; Table S.III.8). However, stripe–interstripe–differences in agrp2 expression are restricted to the dorsal parts (Fig. III.1G). This is complementary to the stripe–interstripe–differences in asip1 expression that we found in ventral areas suggesting complementary functions of agrp2 and asip1. Such compensatory effects might also possibly explain, why knockouts of asip1 alone did not lead to any effects on zebrafish stripe patterns (but only on dorso- ventral patterning) (Cal et al. 2019), while an agrp2/asip1 double knockout might.

Evolution of agouti gene function

In summary, we investigated the gene expression difference of asip1 and agrp2 in relation to dorso-ventral and stripe patterns across teleost fishes. We suggest that both asip1 and agrp2 have a dual function for color pattern formation (Fig III.2). The agouti family gene asip1 is consistently associated with dorso-ventral countershading and stripe patterns. This is in line with previous findings that have shown dorso-ventral expression differences in zebrafish (Ceinos et al. 2015) and flatfishes (Guillot et al. 2012) as well as of the vertebrate homolog Asip in tetrapods (Linnen et al. 2009, Manceau et al. 2011). Expression differences between between melanic stripes and bright interstripes have previously only been reported for tetrapods (Mallarino et al. 2016, Haupaix et al. 2018). Here we describe similar patterns also within teleost fishes, suggesting deep evolutionary conservation of these expression patterns and their link to the respective color patterns (Fig. III.2).

– 63 –

⁕ Chapter III ⁕

Fig. III.2 Evolution of agouti gene function across teleosts and tetrapods. Illustration of the proposed phylogeny of agouti genes based on previous studies (Braasch and Postlethwait 2011, Vastermark et al. 2012). The agouti genes family encompass the genes Asip and Agrp in tetrapods and asip1, asip2/asip2a, agrp1, and agrp2/asip2b in teleost fishes. The agouti family functions as antagonists of melanocortin receptors in the brain (Agrp/agrp1 and agrp2) and/or skin (Asip/asip1 and agrp2). Asip/asip1 is associated to dorsoventral countershading in mice and teleosts (Manceau et al. 2011, Ceinos et al. 2015). Specific expression in interstripes of rodents (Mallarino et al. 2016), birds (Haupaix et al. 2018) and teleosts (this study) also suggested the association role of Asip/asip1 in periodic/horizontal stripe patterning. Teleost-specific agouti gene agrp2 acts as molecular switch for stripe presence/absence not only in East African cichlid radiations (Kratochwil et al. 2018) but also across teleosts (this study). In some teleosts (e.g. T. salvini and D. rerio from this study), the spatial expression patterns of agrp2 also link to stripe–interstripe patterning. Besides, expression of agrp2 has also a neural function (Shainer et al. 2019). The function of another teleosts specific gene asip2 remains unknown and has been lost in several lineages including the zebrafish. Agrp (in tetrapods) and Agrp1 (in teleosts) are neuropeptides secreted in brain and involved in metabolic homeostasis. R1, R2: vertebrate genome duplication round 1 and 2; TSGD: teleost specific genome duplication.

– 64 –

⁕ Chapter III ⁕

A second agouti family gene, agrp2, has been previously shown to underly the repeated evolutionary losses and gains of stripe patterns in East African cichlid fishes, whereby stripe formation is blocked through high expression of agrp2. Here we show that agrp2 might be also involved in evolutionary losses and gains beyond East African cichlids. In two stripe-nonstriped species pairs, one pair of Pseudocrenilabrinae and one pair of Danioninae, we find the same pattern of differential expression with the nonstriped species having higher agrp2 expression (Fig III.1I, J; III.2). While previous work (Kratochwil et al. 2018, Liang et al. 2020) suggested that agrp2 is rather ubiquitously expressed across the whole skin, we identified spatial expression variation in two species that resemble the patterns observed for asip1. In the Neotropical cichlid T. salvini agrp2 shows higher expression in the non-pigmented regions between the stripes and in the ventral part, in D. rerio agrp2 shows higher expression in dorsal interstripes. The latter is particularly intriguing as agrp2 and asip1 have complementary interstripe expression, with agrp2 being expressed in the dorsal interstripes and asip1 being expressed in the ventral interstripes. Therefore, in some lineages and possibly also ancestrally asip1 and agrp2 might serve and have served redundant or complementary functions (Fig. III.2). The phylogenetic relationship of the agouti family genes in teleosts is complex (Kurokawa, Murashita, and Uji 2006, Braasch and Postlethwait 2011, Schioth, Vastermark, and Cone 2011, Vastermark et al. 2012, Shainer et al. 2019). The accepted phylogenetic tree of the agouti gene family (Fig. III.2) suggests that agouti-related peptide (Agrp/agrp1) and an ancestral agouti-signaling peptide originated at the first (R1) whole-genome duplication (WGD). In the second WGD (R2) this ancestral agouti- signaling peptide duplicated into the agouti-signaling peptide (Asip/asip1) and a gene that got lost in vertebrates and resulted through the teleost-specific WGD into the genes agrp2 (therefore also referred to as asip2b) and asip2 (also referred to as asip2a) (Fig. III.2). Interestingly, agrp2 therefore seems to have retained some properties of asip/asip1 (that itself remained functionally highly conserved across vertebrate evolution) including the skin-specific expression, and in some lineages expression in interstripes and the ventral integument. In D. rerio we find a possible subfunctionalization of asip1 and agrp2 into ventral and dorsal interstripe-expression domains, while agrp2 also underwent neofunctionalization by evolving, potentially even repeatedly, into a ‘stripe- repressor’ within cichlids and — as suggested here also in other lineages. Lastly, agrp2 has also another prominent expression domain in the pineal gland (Shainer et al. 2017,

– 65 –

⁕ Chapter III ⁕

Shainer et al. 2019) that clearly constitutes a further neofunctionalization of agrp2 (Braasch and Postlethwait 2011). In conclusion, our comparative study on the evolutionary dynamics of agouti gene family expression and function provides new insights that demonstrate how Asip/asip1 remained remarkably constraint and conserved in its expression pattern, agrp2 retained functional properties of Asip/asip1 while the gene at the same time repeatedly sub- and neofunctionalized to shape and facilitate the evolution of stripe patterns across teleost fishes.

Materials and Methods

RNA extraction, cDNA synthesis and quantitative real-time PCR were performed as previous descripted (Kratochwil et al. 2018, Liang et al. 2020). Primers for qPCR are listed in Table S.III.10. Experiments were performed in accordance with the rules of the animal research facility of the University of Konstanz, Germany and with the permission of the animal care committee (Regierungspräsidium) Freiburg, Germany (T-16/13).

– 66 –

⁕ Chapter IV ⁕

Chapter IV

Neural Innervation as a Potential Trigger of Morphological Color Change and Sexual Dimorphism in Cichlid Fish

Yipeng Liang, Axel Meyer, Claudius F. Kratochwil Submitted to Scientific Reports

Abstract

Many species can change their coloration through cellular and subcellular changes within their integument. The Malawi golden cichlid (Melanchromis auratus) constitutes an extreme example of such a morphological color change. While females and subordinate males of this species are yellow and white with two prominent black stripes (yellow morph), dominant males undergo morphological color change and completely invert this pattern with the yellow and white regions becoming black, and the black stripes becoming white to iridescent blue (black morph). By comparing the two morphs we uncover substantial changes across multiple levels of biological complexity that underly this phenotypic change. These include changes in pigment cell (chromatophore) number, intracellular dispersal of pigments, and tilting of reflective platelets (iridosomes). On a transcriptional level, we find differences in pigmentation gene expression but surprisingly 80% of the genes overexpressed in the dark morph link to neuronal processes including synapse formation. Nerve fiber stainings confirm that scales of the dark morph are indeed innervated by 1.3 to 2 times more axonal fibers. Our results might suggest an instructive role of nervous innervation orchestrating the complex cellular and ultrastructural changes that drive the morphological color change of the Malawi Golden cichlid.

– 67 –

⁕ Chapter IV ⁕

Introduction

Coloration is an important organismal feature that has long been of interest to biologists. It plays crucial roles in predator avoidance, prey capture, conspecific communication and protection from radiation (Cheney, Grutter, and Marshall 2008, Mills and Patterson 2009). Beside these ecological and evolutionary aspects, the formation of pigment patterns gives insights into the formation of complex tissues. In vertebrate the color of the integument is shaped by the multilayered arrangement and interaction of cells with different pigmentary and structural properties (Singh and Nusslein-Volhard 2015). Variations in the density, shape and properties of pigment cells (chromatophores) and the intracellular organization of pigments and reflective molecules shapes the macroscopic appearance of the integument(Patterson and Parichy 2019, Liang et al. 2020). Several types of pigment-bearing and light-reflecting chromatophores have been identified in teleosts (Fujii 2000). The most common cell types are melanophores (containing the brown to black pigment melanin), xanthophores / erythrophores (containing yellow to red pigments including carotenoids and pteridines), leucophores (white) and iridophores (containing guanine platelet crystals that produce structural coloration) (Fujii 1993).

Fig. IV. 1 Yellow and black morph of Melanochromis auratus. (a-b) Females and subordinate males of M. auratus (a) are brightly yellow with two black stripes (yellow morph). Dominant males transform into the dark morph that has two grey to blue stripes on a black background. (c-d) To comparatively analyze the skin of the yellow (c) and black (d) morph we defined five homologous regions: The dorsolateral stripe (DLS, black in yellow morph, purple/blue in dark morph), the interstripe (INT, white/gray in yellow morph, black in dark morph), the midlateral stripe (MLS, black in yellow morph, blue in dark morph), the dorsal part of the ventral integument (dVEN, white in yellow morph, black in dark morph) and the ventral part of the ventral integument (vVEN, yellow in yellow morph, black in black morph).

– 68 –

⁕ Chapter IV ⁕

Although coloration — as a morphological trait — is often perceived and studied as a static trait, coloration can change very dynamically on different time scales. Rapid changes (milliseconds to hours) are referred to as physiological color change and are triggered by neuronal and hormonal signals. These signals cause changes in the intracellular distribution of pigments or reflective molecules. In contrast, changes that involve differences in pigment quantity or cell number are referred to as morphological color change and occur over longer periods of time (hours to months) (Fujii 1993, 2000, Skold et al. 2016). Teleost fishes are a suitable system for studying the developmental and cellular mechanisms and the molecular bases of color and color change phenotypes (Parichy 2006, Nusslein-Volhard and Singh 2017, Kratochwil et al. 2018, Irion and Nusslein-Volhard 2019). With their richness in pigment patterns, especially African cichlid fishes have become an attractive study system to understand how genomic changes facilitate variation in molecular and developmental mechanisms and thereby the evolution of novel pigmentation phenotypes (Maan and Sefc 2013, Santos et al. 2014, Santos et al. 2016, Kratochwil et al. 2018, Hendrick et al. 2019, Kratochwil, Urban, and Meyer 2019). But they also offer remarkable examples for morphological color change, a trait that has received much less attention — especially from a molecular and cellular perspective. For example, in the Central American Midas cichlid Amphilophus citrinellus, roughly 10% of the individuals lose their typical dark pigmentation and obtain a “golden” orange sometimes also white, yellow or red coloration (Barlow 1976, Henning et al. 2013). Another striking case of morphological color change occurs in the Malawi golden cichlid Melanochromis auratus (Fig. IV.1a, b) (Fryer 1959, Fryer and Iles 1972) from Lake Malawi and was described to be the “perhaps most remarkable case” of “color change reported for any cichlid” (Fryer and Iles 1972). While juveniles, females and submissive males of M. auratus are bright yellow with two melanic horizontal stripes (yellow morph; Fig. IV.1a, c), dominant males undergo a drastic morphological color change and become dark with two light blue horizontal stripes (dark morph; Fig. IV.1b, d). Using light microscopy and transmission electron microscopy (TEM), we describe the cellular and ultrastructural differences between the two morphs. The microscopic investigation is complemented with an analysis of the transcriptomic differences between dark and yellow morphs using RNA- seq and immunohistochemistry that reveal a surprising association of the morphological

– 69 –

⁕ Chapter IV ⁕ differences in coloration with increased neural innervation. Taken together, our results give novel insights into the cellular, and molecular underpinnings of a remarkable case of morphological color change in the East African cichlid fish M. auratus.

Results

Chromatophore number, organization and properties differ between yellow and dark morph of M. auratus

Both yellow morph and dark morph of M. auratus are characterized by an — across cichlids —stereotypic stripe pattern with two longitudinal (horizontal) stripes (Fig. IV.1a, b). To first histologically compare the two morphs we defined five regions across dorsal- ventral axis in both morphs (Fig. IV.1c, d): dorsolateral stripe (DLS, black color in yellow morph — purple/blue color in dark morph), interstripe (INT, white/gray color in yellow morph — black color in dark morph), midlateral stripe (MLS, black color in yellow morph — iridescent blue color in dark morph), the dorsal portion of the ventral region (dVEN, white/gray color in yellow morph — black color in dark morph), and the ventral portion of the ventral region (vVEN, bright yellow color in yellow morph — black color in dark morph). To first test whether and how the morphological color change in M. auratus can be explained by changes in chromatophore number, distribution and characteristics, we compared chromatophore distribution between yellow morph and dark morph using light microscopy of whole-mount scale preparations. In line with previous descriptions in cichlids (Baerends and Baerends-Van Roon 1950, Liang et al. 2020), three types of chromatophores could be detected in both morphs: melanophores with black to dark brown pigmentation, xanthophores with yellow to orange pigmentation, and iridophores that produce iridescent/reflective colors (Fig. S.IV.1). To describe chromatophore distribution and characteristics we measured a) chromatophore coverage (melanophores, xanthophores and iridophores), b) chromatophore density (melanophores and xanthophores), and c) the size of chromatophore (melanophores and xanthophores) in the epidermal layer that cover the scales. Chromatophore coverage was calculated by measuring the percentage of the scale that is covered by dark pigment (melanophores), yellow to orange pigment (xanthophores — detected by autofluorescence) and iridescent substances (iridophores — detected – 70 –

⁕ Chapter IV ⁕

Fig. IV.2 Chromatophore measurements in scales of yellow and dark morph of M. auratus scales. (a-f) Photographs of scales from yellow morph (a-c) and dark morph (d-f). Scale bars correspond to 500 μm. (g-i) Measurements for melanophores in the five homologous regions (DLS, INT, MLS, dVEN and vVEN) of yellow and dark morph including coverage (g), cell density (h) and dispersal (i). (j-l) Same measurements for xanthophores including coverage (j), cell density (k) and dispersal (l). (m) Measurement of iridophore coverage. Differences between morphs in the same homologous regions were evaluated by two-tailed t test, n = 5 (individual dots). Each individual dot represents the mean value of measurement for different fishes. Error bars indicate means + SD. Significant sign: *** P<0.001, ** P<0.01, * P<0.05, n.s. non-significant. (n) Principal component analysis (PCA) using all measurements, demonstrating that similarly colored regions cluster together and that morphological changes of the stripe regins (MLS, DLS) and other regions (INT,dVEN, vVEN) occur in opposite directions. PC1 mainly correlates with melanophore and xanthophore properties (in opposite directions, PC2 with differences in iridophores).

polarized light). Chromatophore coverage is influenced both by number and size of the cells (as well as intracellular dispersal/aggregation of pigments) that we also quantified individually. Size of the pigmented area of melanophores and xanthophores (that is

– 71 –

⁕ Chapter IV ⁕ influenced both by cell size and melanosome/xanthosome dispersal within the cell) was estimated by measuring the diameter of the minimally sized circle that encloses all pigmented parts. For obtaining reliable cell number estimates we treated scales with adrenaline to aggregate the melanosomes, thereby easing both quantifications of xanthophore and melanophore number (Fig. S. IV.1). As cell delimitations could be only visualized for xanthophores and melanophores, for iridophores solely the chromatophore coverage was measured. Consistent with the strong phenotypic differences in pigmentation (Fig. IV.1 and IV.2a–f), melanophore coverage differed significantly between dark and yellow morph (all P < 0.001, two-tailed t test). Coverage was lower in the stripe region of the dark morph [DLS (84.69% in yellow morph to 29.40% in dark morph) and MLS (79.31% to 10.65%)] and higher in the interstripe and belly region of the dark morph [INT (1.34% to 73.24%), dVEN (0.61% to 69.36%), and vVEN (0.22% to 69.88%)] (Fig. IV.2g and Table S.IV.1, S.IV.2). Consistent with the differences in coverage, melanophore cell density also differed significantly (all P < 0.001, two-tailed t test) and was lower in stripes of the dark morph [DLS (377.34 cell/mm2 to 186.36 cell/mm2) and MLS (287.67 cell/mm2 to 81.64 cell/mm2)] and higher in interstripe and belly region [INT (52.83 cell/mm2 to 301.71 cell/mm2), dVEN (22.93 cell/mm2 to 212.81 cell/mm2), vVEN (3.49 cell/mm2 to 134.70 cell/mm2) (Fig. IV.2h and Table S.IV.1, S.IV.2). The average melanosome dispersal diameter also differed significantly (all P < 0.01, two-tailed t test) in all five homologous pigment regions and was lower in the stripes of the dark morph [DLS (Ø71.60 µm to Ø49.71 µm) and MLS (Ø79.16 µm to Ø 38.25 µm)] and higher in the interstripe and belly region of the dark morph [INT (Ø20.56 µm to Ø78.04 µm), dVEN(Ø 17.97 µm to Ø 89.77 µm) and vVEN (Ø11.09 µm to Ø 104.97 µm)] (Fig. IV.2i and Table S.IV.1, S.IV.2). Differences in xanthophore coverage were restricted to the ventral regions, where coverage was higher in the yellow morph [1.68% in dVEN of yellow morph to 0.23% in dVEN of dark morph (P < 0.05), 11.22% to 0.05% in vVEN (P <0.01)]. In the other regions coverage was similar and did not differ significantly (0.25% to 1.37% in DLS, 1.68% to 0.40% in INT, 0.26% to 0.76% in MLS) (Fig. IV.2j and Table S.IV.1, S.IV.2). For Xanthophore cell density differences the patterns were similar (DLS, 30.56 cell/mm2 to 55.77 cell/mm2; MLS, 17.28cell/mm2 to 31.97 cell/mm2; INT, 79.86 cell/mm2 to 27.81 cell/mm2, dVEN, 36.89 cell/mm2 to 16.49 cell/mm2), however only the ventral belly

– 72 –

⁕ Chapter IV ⁕ region showed a significant higher cell density in the yellow morph (99.29 cell/mm2 to 2.38 cell/mm2 in vVEN; P < 0.01) (Fig. IV.2k and Table S.IV.1, S.IV.2). Xanthophore size/dispersal was higher in both dVEN and vVEN of the yellow morph (both P < 0.001; Ø33.23µm to Ø15.62µm in dVEN, Ø64.61µm to Ø18.18µm in vVEN). In the dorsal regions there was no significant difference (Ø16.74µm to Ø15.85µm in DLS, Ø19.76µm to Ø13.10µm in INT, Ø21.083µm to Ø20.45µm in MLS) (Fig. IV.2l and Table S.IV.1, S.IV.2). Although we could identify iridophores by polarized light illumination, we were not able to demarcate individual cells. Therefore, we only measure iridophore coverage but not density and diameter of iridophores. The iridophore coverage increased significantly in the two regions where we observe iridescent white/blue coloration in the dark morph (0.15% to 3.37% in DLS with P < 0.01, 0.02% to 0.76% in MLS with P < 0.05) (Fig. IV.2m and Table S.IV.1, S.IV.2). When taking all data together using principal component analysis, we observe that the five homologous regions of the two morphs largely cluster by color (Fig. IV.2n).

Three-dimensional arrangement of chromatophores and their properties shape the phenotypic differences between morphs

Differences in coloration between yellow and dark morph might likely not only be driven by chromatophore differences on scales but also in the underlying skin. While scales can be easily investigated using light microscopy as there is only one cell layer of chromatophores, within the integument such investigation is hindered by the multilayered, three-dimensional arrangement of chromatophores (Chavin 1969, Hirata et al. 2003). To circumvent this problem, we used transmission electron microscopy (TEM) to examine the multilayered arrangement of chromatophores as well as the ultrastructure of the pigment-bearing organelles. All chromatophores could be classified using TEM. Melanophores were identified by the presence of melanin-containing, black melanosomes (Fig. S.IV.1) (Hawkes 1974, Hirata, Nakamura, and Kondo 2005). Xanthophores could be recognized by their round pigment organelles that contain carotenoid and/or pteridine and appeared grey (in contrast to the dark melanosomes) on the TEM images (Fig. S.IV.1) (Hawkes 1974, Hirata et al. 2003, Hirata, Nakamura, and Kondo 2005). Iridophores are characterized by stacks of white layers (Fig. S.IV.1)(Hawkes 1974, Takeuchi 1976, Gundersen and Rivera 1982, Hirata et al. 2003, Hirata, Nakamura, and Kondo 2005). – 73 –

⁕ Chapter IV ⁕

These areas indicate the position of iridosomes (also referred to as reflecting platelets) that are composed of guanine crystals (guanine crystals are lost during the sample preparation and therefore remain as white areas on TEM images) (Takeuchi 1976, Gundersen and Rivera 1982). Iridosomes reflect light and can therefore, if orientated

Fig. IV. 3 Ultrastructural differences between skin regions of yellow and dark morph. (a-d) Macroscopic appearance of the two homologous regions MLS (a, b) and vVEN (c, d) that were comparatively analyzed between yellow (a, c) and dark (b, d) morph. (e–p) The ultrastructure of the integument was analyzed using Transmission electron microscopy (TEM). Chromatophores were mainly found in three layers: Directly beneath the basal membrane (BM; e–h), within the loose connective tissue of the stratum spongiosum (SP; i–l) and in the deepest layer below the stratum compactum (SC; m–p). The layers were compared between the midlateral stripe (MLS) of the yellow (e, i, m) and dark morph (f, j, n) as well as the ventral part of the ventral integument (vVEN) of the yellow (g, k, o) and dark morph (h, l, p). Scale bars correspond to 2 μm. (q–t) Polar chart of the angle of iridosomes in the stratum compactum of the MLS of yellow (q) and dark morph (r) as well as the vVEN of the yellow (s) and dark morph (t).

– 74 –

⁕ Chapter IV ⁕ horizontally, intensify the colors of chromatophores on top of the platelets (i.e. the yellow to orange color of xanthophores). Tilted iridosome arrangements that cause light interference can result in structural colors (Takeuchi 1976, Gundersen and Rivera 1982, Fujii 2000). We therefore hypothesized that the main differences in dark and yellow coloration will be explained by differences in melanophore and xanthophore number, respectively — or on a subcellular level by differences in melanosome and xanthosomes. The shiny and enhanced appearance of yellow and white regions should be explained by horizontally (or fully randomly) oriented guanine-platelets, while the iridescent blue coloration of the MLS of the dark morph should be explained by a tilted, parallelly oriented and lamellar-like organized iridosomes. Chromatophores of M. auratus are organized in three layers within the integument (Fig. IV.3): a) In the basal membrane layer (BM), the uppermost layer that is directly beneath the basal membrane (Fig. IV.3e–h), b) in the stratum spongiosum layer (SP) that is the middle layer within the loose connective tissue of the upper dermis (Fig. IV.3i–l) and c) in the stratum compactum layer (SC), the bottommost layer between the dense, collagenous layer of the stratum compactum and the vascular layer that delimits the dermis from the muscle (Fig. IV.3m–p). In our results we wish to highlight two main observations: 1) The differences between the similarly dark-colored regions of the yellow and dark morph [MLS (Fig. IV.3a)and vVEN (Fig. IV.3d), respectively] and 2) the differences between homologous, but differently colored regions [i.e. the MLS that is dark in the yellow morph (Fig. IV.3a) and iridescently blue in the dark morph (Fig. IV.3b) and the vVEN area that is brilliantly yellow in the yellow morph (Fig. IV.3c) and dark in the dark morph (Fig. IV.3d)]. In dark regions of both morphs, e.g. MLS of yellow morph (Fig. IV.3e, i, m) and vVEN of dark morph (Fig. IV.3h, l, p), we found melanophores in all three chromatophore layers. The melanophores in dark pigmented regions of both morphs were mostly in dispersed state and could be found in all three chromatophore layers. Melanosome density is higher in the dark pigmented regions of yellow morph (mean ± SD: 4.19 ± 0.38 melanosomes per µm2 cell in SC of MLS; Fig. S.IV.2) than in the dark morph (mean ± SD: 3.17 ± 0.83 melanosomes per µm2 cell in SC of vVen; Fig. S.IV.2) — compatible with the stronger black pigmentation of the MLS in yellow morph (Fig. IV.3a). The melanophores in the SC of the MLS in yellow morphs and vVEN in dark morphs were covered by iridophores with few iridosomes. Iridosomes are almost

– 75 –

⁕ Chapter IV ⁕ arranged parallel to the surface in the MLS of the yellow morph (mean angle to surface 11.47°) (Fig. IV.3q). In contrast, platelets are more tilted in the vVEN of the dark morph (mean angle to surface 21.53°) (Fig. IV.3t). In the BM and SP of the yellow morph MLS, melanophores were more scattered, thereby often overlaying small-sized xanthophores (Fig. IV.3e, i). In the SP of dark morph vVEN, small amounts of iridophores were found beneath the melanophores (Fig. IV.3l). For the vVEN of yellow morph, which is brightly yellow colored, we found stacked and almost parallelly to the surface oriented reflecting platelets (mean angle to surface: 9.12°) in the bottommost SC layer (Fig. IV.3o, s). The two upper layers, BM and SP, had numerous xanthophores (Fig. IV.3g, k). Melanophores were almost absent from all layers (Fig. IV.3g, k, o). The ventral area of the yellow morph therefore had more xanthophores, less melanophores and more and more parallelly oriented iridosomes within iridophores in the SC. The iridosomes were — compared to descriptions in other species (e.g. 5–60 nm in the neon tetra, Paracheirodon innesi)(Hiroshi, Noriko, and Ryozo 1990, Nagaishi, Oshima, and Fujii 1990, Yoshioka et al. 2011) — relatively thick in all regions and both morphs (mean thickness between 94 and 144nm; distance between iridosomes between 76 and 147nm; Fig. S.IV.4) This is in accordance with the hypothesis that such structural characteristic would result in the bright yellow coloration of the vVEN region of the yellow morph. The most noticeable characteristic of the integument of the distinct bluish iridescent MLS of the dark morph (Fig. IV.1b) are several sheets of highly organized iridophores (Fig. IV.3f, j, n). In the upper layers, guanine platelets are arranged parallelly but strongly angled (mean angle to surface in BM: 26.10°; mean angle to surface in SP: 38.63°; Fig. S.IV.3). In the bottom layer (SC) the orientation appears more random and essentially range from 0 to 90° (mean angle to surface: 50.13°; Fig. IV.3r). Some aggregated melanophores (Fig. IV.3n) were also found in the in the iridophore layers, especially in the SC layer. The highly organized ultrastructure of the iridophores in the MLS of the dark morph confirms the hypothesis that the iridescent blue coloration can be explained by iridosome orientation in the integument. Within the other dorso-ventral regions we found comparable patterns that we comprehensively summarized in Fig. S.IV.5. The differences between the INT region of yellow and dark morph resembled those of the vVEN region with the main difference being the much higher number of xanthophores in the vVEN of the yellow morph

– 76 –

⁕ Chapter IV ⁕ compared to the INT region. This is compatible with the whitish appearance of the INT region (Fig. S.IV.5). The DLS region resembles the MLS region, with the main difference being that the DLS of the dark morph has less organized iridosomes likely resulting in the more greyish to whitish appearance of the DLS compared to the bluish iridescent MLS (Fig. S.IV.5).

Comparative transcriptomics of M. auratus color morphs

Fig. IV.4 Differential gene expression analysis. Transcript levels plotted as a function of differential expression (log2 fold-change) in yellow morph versus dark morph skin (n=4). 42 genes show significant differential expression [P < 0.01, lfc>log2(2)]. Yellow dots represent genes overexpressed in yellow morph (13 genes), black dots those genes overexpressed in the dark morph (29 genes). Grey dots indicate genes that do not show significant differential expression. Red circles indicate genes with a reported link to pigmentation, Blue circles indicate genes involved in nervous system function (see Table IV.1 and IV.2 for references).

To identify genes that are associated with the coloration differences between yellow and dark morph, we screened for differentially expressed genes using RNA-sequencing

– 77 –

⁕ Chapter IV ⁕

(RNA-seq) of the skin of yellow (n=4) and dark morph (n=4) individuals. Our comparative transcriptomic approach identified 42 differentially expressed genes: 13 genes that were expressed at significantly higher levels in the skin of the yellow morph and 29 genes expressed at significantly higher levels in dark morph (P < 0.01, lfc>log2(2); Fig. IV.4; Table IV.1, IV.2). Surprisingly, the number of genes previously implicated in teleost coloration where little (6 of 42), all of them were among the genes that were higher expressed in the skin of the yellow morph (Fig. IV.4; Table IV.2). Those six pigmentation genes were hydroxy-delta-5-steroid dehydrogenase, 3 beta- and steroid delta-isomerase 1 (hsd3b1) (Lin et al. 2015), pteridine biosynthesis enzyme GTP cyclohydrolase 2 (gch2) (Lister 2019), carotenoid droplets disperser perilipin 6 (plin6) (Granneman et al. 2017), melanophore-linage cell marker microphthalmia-associated transcription factor a (mitfa) (Lister, Close, and Raible 2001, Elworthy et al. 2003, Johnson, Nguyen, and Lister 2011), tetratricopeptide repeat protein 39B (ttc39b) (Hooper, Griffith, and Price 2019, Salis et al. 2019, Ahi et al. 2020) and oncogenic transcription factor forkhead box Q1 (foxq1a) (Bagati et al. 2018). In the skin of dark morph, we found no coloration genes with significantly higher expression, yet a very high number (80% of the genes) of neuronal genes — 23 of 29 based on literature search (Fig. IV.4; Table IV.1). Two GO-terms were significantly enriched (padj<0.05), syntaxin-1 binding (GO:0017075) and SNARE binding (GO:0000149), both important protein interactions involved in the regulation of synaptic vesicles at the presynaptic zone of neuronal synapses. Among the most significantly differentially expressed genes are the synaptosomal-associated protein 25 (snap25a and snap25b) that is required for the fusion of synaptic vesicles with the presynaptic membrane (Batista, Martinez, and Hengst 2017), the neuronal membrane glycoprotein M6a (gpm6ab) is involved in neurite outgrowth and synaptogenesis (Alfonso et al. 2005, Monteleone et al. 2017) and the calcium sensor protein visinin-like protein 1 (vsnl1a) important for neuronal signaling (Li et al. 2011). The up-regulation of neural and synapse-related genes within the skin suggests differences in neural innervation of the skin that — beside the difference in pigmentation — distinguish the yellow and dark morph of M. auratus. We will elaborate in the discussion how this unexpected link between coloration phenotype and neural innervation can be explained.

– 78 –

⁕ Chapter IV ⁕

Table IV.1 List of up-regulated genes in skin of dark morph Gene ID Description Log2 FC log10(P) Process name Neuronal ENSMZEG00005002732 snap25a synaptosomal-associated protein 25a 6.57 10.76 (Batista, Martinez, and Hengst 2017) Neuronal ENSMZEG00005013660 gpm6ab neuronal membrane glycoprotein M6-a 5.21 10.76 (Alfonso et al. 2005, Monteleone et al. 2017) Neuronal ENSMZEG00005005539 snap25b synaptosomal-associated protein 25b 7.21 6.75 (Batista, Martinez, and Hengst 2017) ENSMZEG00005020450 gdf6 growth/differentiation factor 6-B 4.86 6.75 - Neuronal ENSMZEG00005003539 vsnl1a visinin-like protein 1a 3.46 6.73 (Li et al. 2011) ENSMZEG00005014823 stxbp6l syntaxin-binding protein 6-like 7.58 6.03 -

ENSMZEG00005025841 NA NA 7.91 6.00 - Neuronal ENSMZEG00005027303 napb beta-soluble NSF attachment protein 2.40 5.54 (Woods et al. 2006, Fontenas et al. 2016) N-ethylmaleimide-sensitive factor a Neuronal ENSMZEG00005018814 nsfa 7.29 5.08 (vesicle-fusing ATPase) (Woods et al. 2006) Neuronal ENSMZEG00005001660 syt1a synaptotagmin 1a 5.09 4.79 (Fontenas et al. 2016, Bademosi et al. 2017) Neuronal ENSMZEG00005009208 sncb synuclein beta 5.08 4.52 (Lodygin et al. 2019) Neuronal ENSMZEG00005010543 sez6l2 seizure related 6 homolog like 2 7.08 4.37 (Yaguchi et al. 2017) Neuronal neuronal vesicle trafficking associated ENSMZEG00005002426 nsg2 6.65 4.30 (Barford et al. 2017, Chander 2 et al. 2019) zinc finger and BTB domain Neuronal ENSMZEG00005017078 zbtb20 1.84 4.29 containing 20 (Nagao et al. 2016) Neuronal ENSMZEG00005006423 fabp7 fatty acid binding protein 7 6.99 4.29 (Young, Heinbockel, and Gondre-Lewis 2013) Neuronal solute carrier family 17 (vesicular (Fremeau et al. 2004, ENSMZEG00005010741 slc17a7b 6.96 4.25 glutamate transporter), member 7a Schuske and Jorgensen 2004) Neuronal ENSMZEG00005020088 scg2b secretogranin 2 6.22 4.19 (Tao et al. 2018) adenylate cyclase activating Neuronal ENSMZEG00005020108 adcyap1b 2.07 3.82 polypeptide 1b (Hashimoto et al. 2007) Neuronal (Labouesse et al. 2015, ENSMZEG00005018886 gad2 glutamate decarboxylase 2 5.96 3.54 Davis et al. 2016, Zhang et al. 2017) ENSMZEG00005022306 tuba1b tubulin alpha-1B chain 3.14 3.20 - low molecular weight neuronal Neuronal ENSMZEG00005001648 zgc:65851 6.68 3.19 intermediate filament (Liem 1993) Neuronal ENSMZEG00005017787 eno2 enolase 2 5.77 3.00 (Bai et al. 2007, Bai, Wei, and Burton 2009) solute carrier family 6 Neurona ENSMZEG00005013731 slc6a1b (neurotransmitter transporter), member 6.03 2.80 (Chen, Reith, and Quick 1b 2004) Neuronal ENSMZEG00005025554 ncan neurocan core protein 2.51 2.74 (Sullivan et al. 2018) sodium/potassium-transporting Neuronal ENSMZEG00005025994 atp1a3a 1.42 2.38 ATPase subunit alpha-3 (Doganli et al. 2013) ENSMZEG00005014171 p2ry6 pyrimidinergic receptor P2Y6 1.21 2.29 - Neuronal ENSMZEG00005015980 map1aa microtubule associated protein 1A 1.91 2.29 (Liu, Lee, and Ackerman 2015, Takei et al. 2015) Neuronal ENSMZEG00005012827 vsnl1 visinin-like protein 1 3.98 2.27 (Li et al. 2011) ENSMZEG00005025734 NA NA 3.83 2.06 -

– 79 –

⁕ Chapter IV ⁕

Table IV.2 List of up-regulated genes in skin of yellow morph

ID Gene name Description Log2 FC log10(P) Process Pigmentation ENSMZEG00005025109 foxq1a forkhead box protein Q1 1.99 2.29 (Bagati et al. 2018) ENSMZEG00005008428 tmem130 transmembrane protein 130 1.77 2.30 -

ENSMZEG00005021959 rdh11 retinol dehydrogenase 11 1.69 2.38 - Pigmentation ENSMZEG00005002482 gch2 GTP cyclohydrolase 2 2.94 2.50 (Lister 2019) Pigmentation (Lister, Close, and Raible microphthalmia-associated ENSMZEG00005008959 mitfa 1.15 3.14 2001, Elworthy et al. 2003, transcription factor Johnson, Nguyen, and Lister 2011) ENSMZEG00005002836 map7a ensconsin 1.11 3.18 - scavenger receptor class B member ENSMZEG00005012260 scarb1 1.07 3.18 - 1 dehydrogenase/reductase SDR ENSMZEG00005020078 dhrs12 1.90 3.18 - family member 12 Pigmentation ENSMZEG00005017109 plin6 perilipin 6 1.94 3.62 (Granneman et al. 2017) ENSMZEG00005004808 bscl2 seipin 1.96 3.95 - membrane metalloendopeptidase ENSMZEG00005014299 mmel1 2.70 4.17 - like 1 Pigmentation (Hooper, Griffith, and Price ENSMZEG00005019551 ttc39b tetratricopeptide repeat protein 39B 2.27 4.30 2019, Salis et al. 2019, Ahi et al. 2020) hydroxy-delta-5-steroid Pigmentation ENSMZEG00005019286 hsd3b1 dehydrogenase, 3 beta- and steroid 1.94 4.43 (Lin et al. 2015) delta-isomerase 1

Axonal innervation of the scale epithelium is increased in the dark morph of M. auratus

To find further support for the hypothesis that neural innervation is indeed increased in the dark morph, we performed immunofluorescence staining of axonal fibers on scales of both morphs (Rasmussen, Vo, and Sagasti 2018). We dissected scales from three homologous dorsal-ventral positions and counted the number of nerve fibers (Fig. IV.5, S.IV.6, S.IV.7, see details in the methods section). Axons were visualized using Anti- Acetylated Tubulin antibody (AcTub). In accordance with the transcriptome data, we indeed found more AcTub+ on scales of the dark morph compared to scales of the yellow morph, independent of dorso-ventral level and position on the scale (all P < 0.01, 1.3 to 2 fold change; Fig. IV.5e, S.IV.6, S.IV.7; Tables S.IV.3, S.IV.4). As we were concerned that the lower number of axon counts might be also linked to the increased number of melanophores that could mask the fluorescence and thereby result in an underestimation of axons in the dark morph (Fig. S.IV.8). Therefore, we used a previously described (spontaneous) mutant line that lacks all melanic pigmentation due to a loss of the second exon of oculocutaneous albinism II (oca2) gene (Kratochwil, Urban, and Meyer 2019). Still — despite this loss-of-function mutation, melanophores seem to be present and both, the yellow and dark morph can be found in the amelanistic line as previously described – 80 –

⁕ Chapter IV ⁕

(Kratochwil, Urban, and Meyer 2019) (as the morph names are rather confusing descriptions in the amelanistic strain we use quotations marks, i.e. ‘yellow morph’ and ‘dark morph’). Comparably to the wildtype-like strain of M. auratus, ‘dark morph’ individuals of the amelanistic M. auratus had significantly higher axon density than ‘yellow morph’ individuals (all P < 0.01, 1.3 to 1.8 fold change) (Fig. IV.5a–d, S.IV.6, S.IV.7; Table S.IV.3, S,IV.4). Lastly, as mentioned in the introduction and discussed below it is dominant males that undergo the color transition and become dark morphs, while this has not been reported for female individuals. It is therefore possible that the differences in axon density are linked to sex and not to morphs. To test this possibility, we performed the same quantifications on scales of a closely related species without color change, M. cyaneorhabdos, and compared males and females. No significant differences in axon density could be detected (all P > 0.05) (Fig. S.IV.6, S.IV.7; Table S.IV.3, S.IV.4) suggesting that the difference in neural innervation is linked to the two morphs, but not sex.

Fig. IV.5 Increased axon innervation in the epidermis of the dark morph. (a–d) Immunohistochemistry with Anti-Acetylated Tubulin antibody labeling axonal fibers shows increased innervation of scales of the dark morph (b, d) compared to the yellow morph (a, c). Scale bars are 500 μm in (a, b) and 50 μm in (c, d). (e) Statistical analysis using a line dissect across the scale shows a 1.4 to 2-fold, significant increase of axonal fibers in scales of the dark morph.

– 81 –

⁕ Chapter IV ⁕

Discussion

In this study, we investigated the cellular and transcriptional differences between the yellow and dark morph of M. auratus, a cichlid species endemic to Lake Malawi (Hulsey et al. 2017, Malinsky et al. 2018). Specifically, we asked how the morphological differences can be explained a) by changes in chromatophore number b) chromatophore characteristics, c) multi-dimensional arrangement of chromatophores in the skin and d) if we find transcriptional signatures that provide us more mechanistic insights into genetic and cellular correlates as well as the genetic pathways that might be involved in triggering the morphological color change from yellow to dark morph. Our analysis reveals substantial variation between yellow and dark morph (Fig. IV.1) ranging from transcriptomic differences, to differences in chromatophore morphology, number and three-dimensional organization as well as — unexpectedly — changes in neural innervation of the integument. These dramatic changes must occur during the morphological color change of dominant males that transition over the course of two weeks from the yellow to the dark morph. In the following we therefore i) comprehensively discuss the differences between the yellow and the dark morph, ii) speculate what changes cause the transition and/or what might be merely indirect consequences of the morphological color change. Our study demonstrates that the yellow and dark morph differ greatly on different levels of biological complexity ranging from cellular and ultracellular characteristics of chromatophores as well as cellular composition of the three layers of the integument as well as the scales. The first observation is that the color of phenotypically similar regions is formed through similar mechanisms (Fig. IV.2, IV.3): The dark stripes of the yellow morph (MLS and DLS) and the dark ventral area and interstripe (vVEN, dVEN and INT) of the dark morph are characterized by high density of dispersed melanophores (Fig. IV.2g-i). In contrast, the yellow belly region of the yellow morph (vVEN) has a very low density of melanophores, but high density of large, dispersed xanthophores. A further peculiarity are thick layers of horizontally oriented iridiosomes within the stratum compactum, the deepest layer of the integument. These likely simply serve as a mirror and directly reflect light and thereby intensify the yellow coloration of the overlaying xanthophores. Such mirror-like functionality of iridosomes have been previously described, i.e. in the japanese Koi fish (Cyprinus carpio), where the reflectance is

– 82 –

⁕ Chapter IV ⁕ however generated through a fully random organization of iridosomes (Gur et al. 2017). Lastly, there are several regions with iridescent white (i.e. INT in yellow morph) to blue (MLS in dark morph) coloration. The major differences to the other regions are here beside the lower density of xanthophores and melanophores that iridosomes are not oriented horizontally but in an acute angle of about 25 to 50° (Fig. IV.3r). Such regularly oriented, tilted iridosomes have been previously described in the neon tetra Paracheirodon innesi, that is characterized by a single iridescent blue horizontal stripe. The periodically arranged iridosomes constitute a colour-producing microstructure that cause the blue coloration through multilayer optical interference (Hiroshi, Noriko, and Ryozo 1990, Yoshioka et al. 2011). On a transcriptomic level we find several coloration genes that are differentially expressed. Two of the genes that show higher expression in the yellow morph are gch2 and plin6, two genes that have been associated with carotenoid and pteridine synthesis — the yellow to orange pigments of xanthophores (Fig. IV.4; Table IV.2). The Plin6 protein targets the surface of carotenoid droplets and mediates carotenoid droplet dispersion in xanthophores (Granneman et al. 2017). The gene gch2 encoding a pivotal protein in the pteridine biosynthetic pathway and is required for xanthophore pigmentation in zebrafish (Lister 2019). Also hsd3b1 and ttc39b has been linked to lipid and carotenoid synthesis, and been previously linked to carotenoid-based coloration differences in cichlids (Lin et al. 2015, Hooper, Griffith, and Price 2019, Salis et al. 2019, Ahi et al. 2020). Compared to the dark morph, xanthophores are numerous in the yellow morph and densely filled with xanthosomes as supported by light microscopy (Fig. IV.2) and TEM images (Fig. IV.3), therefore these transcriptional differences are compatible with the morphological differences. We also find two pigmentation genes that are associated to melanophore/melanocyte differentiation: foxq1 and mitfa (Lister, Close, and Raible 2001, Elworthy et al. 2003, Johnson, Nguyen, and Lister 2011, Bagati et al. 2018). As melanophore numbers are at least seemingly lower in the yellow morph, especially in vVEN (Fig. IV.2, IV.3), it is possible that the high expression of these genes indicate large reservoirs of undifferentiated melanoblasts, that only differentiate when the color change occurs. What remains unclear, is how the melanoblasts are mechanistically kept in an undifferentiated state. However, several of the other differentially expressed genes have been previously linked to stem cell maintenance including retinol dehydrogenases (Arregi et al. 2016), ensconsin/map7a (Gallaud et al.

– 83 –

⁕ Chapter IV ⁕

2014), scavenger receptor class B member 1 (Gao et al. 2014) and seipin/bscl2 (Mori et al. 2016). The most striking observation regarding those genes being overexpressed in the dark morph is that a substantial part of these genes (80%) is related to neural processes, particularly to synaptogenesis (Fig. IV.4). This observation is further substantiated by immunohistochemistry for the axonal marker acetylated Tubulin that demonstrate a substantially higher number of neuronal fibers in scales of the dark morph (1.3 to 2-fold change; Fig. IV.5) across the whole skin. To understand this — at first surprising — link between color morphs and nervous innervation there are two facts to consider: First, adult pigment cells derive from postembryonic progenitors that are located in proximity of the dorsal root ganglia. From there these progenitors migrate along peripheral nerves through the myosepta to the skin. Therefore, there is a strong developmental link between differentiation of adult pigment cells and nerve fibers (Dooley et al. 2013, Parichy and Spiewak 2015, Singh et al. 2016). Second, the connection between pigment cell remains even after differentiation. Especially melanophores, but also iridophores and xanthophores are controlled by the nervous system that can trigger the aggregation of melanosomes or xanthosomes, as well as reorientation of iridosomes (Fujii 1993, Yoshioka et al. 2011, Burton and Burton 2017). These fast intracellular changes trigger the fast, so called, physiological color change (Burton and Burton 2017). But neural innervation has been also shown to induce more long-lasting changes (i.e. morphological color change) by triggering remodeling of the cytoskeleton (Sugimoto 2002), apoptosis of chromatophores (Sugimoto, Uchida, and Hatayama 2000, Sugimoto 2002), as well as proliferation of chromatophores and changes in pigment synthesis (Hara et al. 1996). The morphological color change of M. auratus is in so far remarkable as different, inverse changes occur in different parts of the skin (Fig. IV.2). While in the stripes melanophores are reduced and xanthophores are increased in number, we observe the opposite in the interstripe and ventral skin. Additionally, we find increased numbers of iridophores, mainly in the dorsal areas, as well as reorganization of iridosomes, in particular in the MLS. The molecular processes that orchestrate such a complex rearrangement are therefore likely very complex itself. Based on our results we assume that two different processes are occurring: First, we observe changes at the level of pigment cell progenitors, either within the skin or at the postembryonic progenitor niche in vicinity of the dorsal root ganglia, that lead to the generation of new chromatophores.

– 84 –

⁕ Chapter IV ⁕

As cell-cell interactions are an important factor for color pattern formation in fish, it is possible that these changes do not occur everywhere at the same time, but are propagated across the skin (Inoue et al. 2014). Recently it has been shown that long-distance signaling relayed by macrophages trigger tissue color pattern remodeling during postembryonic development (Eom and Parichy 2017), so similar processes could certainly play a role in this instance of morphological color change. Some of the genes overexpressed in the yellow morph indicate that these fish might already have a reservoir of melanoblast that could — triggered by an endocrine, paracrine or neuronal signal — differentiate and thereby realize the morphological color change from yellow to dark morph. Second, the overexpression of neural genes and increased number of neural innervations could be an indication for two processes: The nerve fibers could indicate a stronger neural control of chromatophores (i.e. physiological color change). This interpretation is supported by the fact that the melanic patterns do not fade or enhance very much in the yellow morph, while the dark morph shows substantial physiological color change with the dark melanic parts changing from light grey to black and vice versa. It is also possible that the neural fibers have an instructive role and initiate the color change by promoting migration, differentiation of chromatophore progenitors, cell death as well as changes in pigment production or translocation, as shown in other organisms (Hara et al. 1996, Sugimoto, Uchida, and Hatayama 2000, Sugimoto 2002). One central question is how the color change of the different regions (i.e. DLS, INT, MLS, dVEN and vVEN) is orchestrated. As discussed above, changes happen in different direction, with DLS and MLS becoming lighter (i.e. grey or blue) and INT, dVEN and vVEN becoming darker (i.e. dark grey or black). However, we do not see variation in neural innervation across the dorsoventral axis that could explain these inverse changes and therefore the phenotype remains difficult to explain solely by the innervation. It is possible that this is achieved by a combination of the neural trigger and number of chromatophore precursors that are present in the different regions. Therefore, although our morphological and transcriptomic analyses provide crucial insights into the molecular and cellular correlates of the color change in M. auratus, our analyses cannot indicate whether the nervous system and/or particular genes act directly or indirectly on the distribution, shape and properties of chromatophores and if they are causally involved. An even more detailed analysis will be probably needed that dissects the molecular and

– 85 –

⁕ Chapter IV ⁕ cellular changes in the different skin regions — also at different time points of the transition process — to completely understand this remarkable color change phenotype. In conclusion, our work provides an integrative and comprehensive description of the transcriptomic, ultrastructural and cellular changes that define the morphological color change of M. auratus. It provides therefore the first insights into the massive changes occurring at different levels of biological complexity ranging from gene expression changes, to reorientation of reflecting platelets (iridosomes) in iridophores, to changes in chromatophore density and morphology, and changes in neuronal innervation. The morphological color change M. auratus was described as “perhaps most remarkable case” of “color change reported for any cichlid” almost 50 years ago — the complexity of the changes on that shape and accompany this transition and its evolution are in no way less remarkable.

Methods

Animals

Animals used in this study were raised and kept in the animal research facility of the university of Konstanz. The study was performed in accordance with the rules of the animal research facility of the University of Konstanz and the animal protection authorities of the State of Baden-Württemberg (T-16/13).

Light microscope image acquisition for fish scales

To quantify the change in chromatophores quantity and properties during color transition of M. auratus, we measured the coverage, density and dispersed diameter of chromatophores in scales from five homologous pigmented regions in dorsal-ventral axis. In M. auratus, chromatophores could be found in multiple layers through the integumentary system and overlaid/underlaid each other. The chromatophores only cover the posterior part of the scales and arrange in relatively single layer, which makes the robust chromatophore quantification possible. Five individuals of each color morph were examined for scale chromatophore quantification. Fifteen scales from three dorsal- ventral rows (5 scales each row) were removed from each fish. Scales from most dorsal row covering DLS and INT (Fig. IV.2a, d); scales from middle row covering MLS and

– 86 –

⁕ Chapter IV ⁕ dVEN (Fig. IV.2b, e); the scales from most ventral row (vVEN) (Fig. IV.2c, f). Once being removed from the fish, scales were rinsed by Phosphate Buffered Saline (PBS, pH 7.4) and adhered to HistoBond+ adhesive microscope slides (Paul Marienfeld). Photographs of M. auratus scales were captured on Leica DM6B upright microscope with a Leica DMC 2900 camera. We captured images in three different modes: brightfield for melanophore coverage and dispersed diameter, polarized light for iridophore coverage(Liang et al. 2020), fluorescent with GFP filter for xanthophore coverage and dispersed diameter (Kelsh et al. 1996, Le Guyader and Jesuthasan 2002). To count the number of melanophores and xanthophores, we treated the scales immediately after taking the above photos with 10mg/ml adrenaline (SIGMA- ALDRICH) for 5 min at room temperature on microscope slides to aggregate the melanosomes which permit robust cell number quantification (Fig. S.IV.1). After adrenaline treatment, microscope-slide-adhered scales were rinsed by PBS and photographed using brightfield mode. Leica Application Suite X software was used to capture the photos using the same setting.

Image and data analysis of scale chromatophores

All light microscope photos were analyzed with Fiji (Schindelin et al. 2012). In order to obtain reliable chromatophore coverage quantification, we manually adjusted the color threshold of the brightfield photos for melanophores, polarized light photos for iridophore and fluorescent photos for xanthophore from non-treated scales (Fig. S.IV.1). Following this setting, we executed the “Analyze Particles” function to obtain all three chromatophore coverage. For chromatophore size measurement, we randomly selected 10 melanophores from brightfield photos and 10 randomly xanthophores from fluorescent photos for each pigmented region (DLS, INT, MLS, dVEN, vVEN) and measured the dispersed diameter of pigment covered parts of the chromatophores. The number of melanophores and xanthophores was counted from the same scales after the epinephrine treatment. Although we could identify iridophores and quantified the iridophore coverage in all pigmented regions before and after epinephrine treatment (Fig. S.IV.1), the boundaries of each iridophore could not be identified. Therefore, we were unable to quantify the iridophore number and measure the iridophore size.

– 87 –

⁕ Chapter IV ⁕

Transmission electron microscope image acquisition for fish integument

Melanic and non-melanic skin with scales attached were dissected in buffer pre-fixative (2% formaldehyde, freshly depolymerized from paraformaldehyde (Merch), 2.5% glutardialdehyde (Agar Scientific) in 0.1 M HEPES (AppliChem) with 5mM MaCl2,

5mM CaCl2, 0.125mM MgSO4 and 0.3M sucrose for osmolarity, pH 7.5) and fixed in fresh-made, pre-cooled fixative (same as buffer pre-fixative) at 0 °C for 2h. After osmification in 2% OsO4 (SERVA) (buffered with HEPES, pH 7.5) at 0 °C for 2h, samples were pre-dehydrated in 30% and 50% Ethanol and en-bloc stained the samples by 2% uranyl acetate (Merck) in 70% Ethanol. We dehydrated the samples in graded series of acetone solutions and embedded in SPURR (TedPella) (ERL 4206, DER 736, NSA, DMAE). 50nm ultra-thin sections were cut by a Leica Ultracut microtome with an ultra 45° knife (Diatome, Switzerland). Sections were contrasted with uranyl acetate and lead citrate before examined with a Zeiss TEM 912. Images were processed with Fiji and stitched with TrackEM2 (Cardona et al. 2012). Iridophore orientation was estimated by measuring the angle between iridophore and skin surface.

RNA extraction and purification

RNA extraction and purification were performed as previously described (Kratochwil et al. 2019). Briefly, dissected skin samples were kept in RNAlater (Invitrogen) at -20 °C. RNA from skin samples was first extracted with TRIzol (Invitrogen), following by purification and DNase treatment with RNeasy Mini Kit (Qiagen) and RNase-Free DNase Set (Qiagen). Following extraction and purification, RNA was quantified using the Qubit RNA HS Assay Kit (Invitrogen) with a Qubit Fluorometer (Life Technologies). The RNA integrity number (RIN) was checked using an RNA 6000 Pico Kit (Agilent) on a 2100 Bioanalyzer System (Agilent).

RNA library preparation and sequencing

RNA-seq libraries were prepared using the TruSeq Stranded mRNA Library Prep Kit (Illumina) according to the manufacturer’s protocol. Briefly, 1μg RNA was put into mRNA selection by poly-T oligo attached magnetic beads followed by fragmentation (94°C for 6 min). The cleaved mRNA fragments were reverse transcribed into first-strand cDNA using GoScript Reverse Transcriptase (Promega) and random hexamer primers (Illumina). We used Illumina-supplied consumables to synthesize second-strand cDNA – 88 –

⁕ Chapter IV ⁕ following by adenylating 3’ ends. Barcoded adapters from TruSeq RNA CD Index Plate (Illumina) were ligated to the ends of the double-strand cDNA. The final libraries were amplified using 15 PCR cycles and quantified and quality-assessed using an Agilent DNA 12000 Kit on a 2100 Bioanalyzer (Agilent). Indexed DNA libraries were normalized. Libraries were sequenced on a HiSeq X Ten platform (BGI Genomics, Beijing).

Mapping and data processing

For adapter trimming trimmomatic 0.38 was used (Bolger, Lohse, and Usadel 2014). The processed raw reads (mean: 41.7 million reads per sample) were aligned to the Maylandia zebra genome (M_zebra_UMD2a, INSDC Assembly GCA_000238955.5, Apr 2018) using the STAR RNA-seq aligner (version 2.6.1d) (Dobin et al. 2013). Expression counts were calculated with RSEM (Li and Dewey 2011) using the ENSEMBL annotation. Quality control was done using MultiQC (Ewels et al. 2016). 89% of the reads were uniquely mapped.

Data analysis

The data was analysed in R using the DESEQ2 1.22.1 pipeline (Love, Huber, and Anders 2014) in R (R Development Core Team 2019). The following packages were additionally used: BiocParallel 1.16.2 (Morgan et al. 2019), tximport 1.10.1 (Soneson, Love, and Robinson 2015), stringR 1.4.0 (Wickham 2019), edgeR 3.24.1 (Robinson, Mccarthy, and Smyth 2010), vsn 3.50.0 (Huber et al. 2002), ggplot2 3.1.1 (Wickham, Chang, et al. 2019), RColorBrewer 1.1-2 (Neuwirth 2014), vidger 1.2.0 (Mcdermaid et al. 2018), pheatmap 1.0.12 (Kolde 2019), cowplot 0.9.4 (Wilke 2017) and dplyr 0.8.0.1 (Wickham, François, et al. 2019). GO term analyses was performed by extracting GO terms for the M. zebra ENSEMBL genome annotation using biomaRt 2.38.0 (Durinck et al. 2009). Next, we used goseq 1.34.1 (Young et al. 2010) to correct for transcript length correction and selection-unbiased testing for category enrichment amongst differentially expressed genes. P-values were adjusted with Benjamini & Hochberg correction.

Axon staining and density quantification on scales

As the RNA-seq result showing that the expression of neural genes upregulates in the dark morph, we performed immunofluorescent staining for axon to verify the expression

– 89 –

⁕ Chapter IV ⁕ data. Fish scales (15 scales/fish) were removed from the homologous regions for chromatophores measurement (5 scales from each dorsal-ventral row, 3 rows from each fish). Immunostaining was performed based on previously described with modification (Woltering et al. 2018). Scales were fixed in Dent’s fixative (20% dimethyl sulfoxide (DMSO), 80% Methanol) at 4°C overnight then kept in 100% Methanol at -20 °C. We rehydrate the samples by graded series of Methanol/PBS. After several washes in PBS and PBS 0.1% Triton X-100, we permeabilizated the samples with permeabilization buffer (1% normal goat serum (Sigma), 0.4% Triton X-100 in PBS). Scales were then blocked in 10% normal goat serum with 0.4% Triton 100-X in PBS for minimally 2 hours and incubated at 4°C overnight with mouse monoclonal anti-acetylated-tubulin antibody (1:250; clone 6-11B-1; Sigma-Aldrich # T6793) in PBS supplemented with 10% normal goat serum and 0.1% Triton X-100. Although this primary antibody was raised against acetylated tubulin of sea urchin, it has been used for detection of acetylated tubulin from several tissues of many organisms, including fish scales (Chitnis and Kuwada 1990, Rasmussen, Vo, and Sagasti 2018). The next day scales were washed by PBS supplemented with 0.1% Tween-20 (PBST) several times and the staining was achieved by incubation with secondary antibody (Goat-anti-Mouse IgG (H+L) SuperClonal Secondary Antibody, Alexa Fluor 488 conjugate, Invitrogen # A28175, 1:400) at 4°C overnight. The next day scales were washed by PBST and counterstain with 4’,6- Diamidino-2-phenylindole dihydrochloride (DAPI) (1/5000; Sigma) in PBS. We mounted the scales on HistoBond+ adhesive microscope slides in Mowiol mounting medium (Sigma). Photographs were taken with a Leica DM6B microscope with a Leica DFC3000G black and white camera. Axon density was quantified whit Fiji. We quantified the axon density from both anterior and posterior parts of the scales (Fig. IV.6, S.IV.6, S.IV.7). For axon density in anterior of scales, the two distinct endpoints of measurement line segment were located in the most dorsal and the most ventral edge of epidermis which indicated by DAPI staining. For axon density in the posterior of scales, the two distinct endpoints of the measurement line segment were located in the most dorsal and the most ventral centiis. Following this setting we executed the “Plot Profile” function and “Find Peaks” function of BAR plugin to obtain the axon density. All statistical tests were performed in R.

– 90 –

⁕ Chapter V ⁕

Chapter V

Developmental and Cellular Basis of Vertical Bar Color Patterns in the East African Cichlid Fish Haplochromis latifasciatus

Yipeng Liang, Jan Gerwin, Axel Meyer, Claudius F. Kratochwil Frontiers in Cell and Developmental Biology (2020) DOI:10.3389/fcell.2020.00062

Abstract

The East African adaptive radiations of cichlid fishes are renowned for their diversity in coloration. Yet, the developmental basis of pigment pattern formation remains largely unknown. One of the most common melanic patterns in cichlid fishes are vertical bar patterns. Here we describe the ontogeny of this conspicuous pattern in the Lake Kyoga species Haplochromis latifasciatus. Beginning with the larval stages we tracked the formation of this stereotypic color pattern and discovered that its macroscopic appearance is largely explained by an increase in melanophore density and accumulation of melanin during the first 3 weeks post-fertilization. The embryonal analysis is complemented with cytological quantifications of pigment cells in adult scales and the dermis beneath the scales. In adults, melanic bars are characterized by a two to threefold higher density of melanophores than in the intervening yellow interbars. We found no strong support for differences in other pigment cell types such as xanthophores. Quantitative PCRs for twelve known pigmentation genes showed that expression of melanin synthesis genes tyr and tyrp1a is increased five to sixfold in melanic bars, while xanthophore and iridophore marker genes are not differentially expressed. In summary, we provide novel insights on how vertical bars, one of the most widespread vertebrate color patterns, are formed through dynamic control of melanophore density, melanin synthesis and melanosome dispersal.

– 91 –

⁕ Chapter V ⁕

Introduction

Pigment patterns play important roles in many aspects of animal biology. Yet, until now, only in a few “model” organisms we do have insights into the molecular and developmental underpinnings of color pattern formation and evolutionary diversification. Among teleosts, the zebrafish Danio rerio and the Medaka Oryzias latipes are the main model organisms for investigation of pigmentation (Meyer, Biermann, and Orti 1993, Kelsh et al. 1996, Meyer, Ritchie, and Witte 1997, Nagao et al. 2014, Irion and Nusslein- Volhard 2019, Patterson and Parichy 2019). More recently, African cichlid fishes with their richness in color patterns are increasingly studied to understand the molecular mechanisms of color pattern formation including but not limited to egg spot patterns (Henning and Meyer 2012, Santos et al. 2014, Santos et al. 2016), blotch patterns (Streelman, Albertson, and Kocher 2003, Roberts, Ser, and Kocher 2009), amelanism (Kratochwil, Urban, and Meyer 2019), horizontal stripe patterns (Ahi and Sefc 2017, Kratochwil et al. 2018, Hendrick et al. 2019) and pigment distribution more generally (Albertson et al. 2014). And although progress has been made identifying target genes and loci that drive evolutionary diversification in cichlids (Roberts, Ser, and Kocher 2009, Kratochwil et al. 2018) and play key roles in adaptation and speciation (Seehausen, Mayhew, and Van Alphen 2001, Elmer, Lehtonen, and Meyer 2009, Maan and Sefc 2013), the developmental and cellular mechanisms of pigmentation phenotypes have been barely studied. Pigment patterns are ultimately caused by spatial variation in pigmentary and/or structural tissue properties. Those can be generated by different distribution, density and aggregation state of pigment cells (chromatophores) and their multi-layered arrangement, as well as variation in the synthesis and arrangement of light-absorbing pigments or molecules causing structural coloration (Irion and Nusslein-Volhard 2019, Patterson and Parichy 2019). In teleosts several types of chromatophores, including melanophores, xanthophores, iridophores, erythrophores, leucophores, and cyanophores have been described (Burton and Burton 2017). Melanophores (containing the brown to black pigment melanin), xanthophores/erythrophores (containing yellow to red pigments) and iridophores (containing reflective guanine platelets causing structural coloration) have been also found in cichlids (Maan and Sefc 2013).

– 92 –

⁕ Chapter V ⁕

For the mechanisms of color pattern formation, mainly (horizontal) stripe patterns have received attention because the most commonly studied “model” teleost, the zebrafish, carries this characteristic pattern. Vertical bars are less studied, with the exception of recent detailed description of the convict cichlid Amatitlania nigrofasciata (Prazdnikov and Shkil 2019), studies in Amphiprioninae, the anemone fishes (Salis et al. 2018, Roux et al. 2019) and Lake Malawi cichlids (Hendrick et al. 2019). Bar patterns are presumably adaptive as such disruptive coloration breaks the outline of the individual and thereby constitutes a form of camouflage, in particular in visually complex habitats (Seehausen, Mayhew, and Van Alphen 2001, Maan and Sefc 2013). Additionally color patterns are often involved in species recognition (Hemingson et al. 2019). Here, we focus on the vertical bar pattern of Haplochromis latifasciatus (Fig. V.1E) from Lake Kyoga, a lake north of Lake Victoria. H. latifasciatus is a species of the haplochromine cichlids, the most species-rich cichlid lineage that forms the adaptive radiations of Lake Victoria and Malawi with 500 and 800 species respectively. We describe the formation of the pattern during development and compare it to other teleosts, characterize which cell types and properties underlie this conspicuous pattern, and use a candidate gene approach to obtain insights into the underlying molecular mechanisms.

Results

Developmental Progression of Vertical Bar Formation in H. latifasciatus

Both male and female H. latifasciatus in adult stage are characterized by four (in some strains five) vertical melanic bars. Individuals of our breeding stock had consistently four bars (n > 30; Fig. V.1E): one anterior bar above the operculum (vertical bar 1; vb1), two vertical bars in the trunk area that cover the whole dorso-ventral axis (vb2 and vb3) and a more posterior vertical bar (vb4) at the anterior caudal peduncle that only covers the dorsal part up to the horizontal myoseptum. The regions between the bars (referred to as interbars, ib) are yellow to beige with dominant males often having nuptial colors with stronger yellow but anteriorly also red to orange hues (Fig. V.1Q, R). In contrast to the closely related Lake Victoria cichlids bars are thicker and more pronounced in H. latifasciatus with a clearer demarcation, lower number and without as pronounced

– 93 –

⁕ Chapter V ⁕ physiological color change (Greenwood 1974, Seehausen, Mayhew, and Van Alphen 2001).

Fig. V.1 Vertical bar patterns in Haplochromis latifasciatus. (A–E) Developmental stages of H. latifasciatus at 7 dpf (A), 9 dpf (B), 15 dpf (C), 21 dpf (D) and an adult individual (E). At 7 dpf Melanophores are mainly located on the head and yolk sac (A). At 9 dpf, two melanophore dorsal clusters (dc1 and dc2) are appearing indicating the position of vb1 and vb2, respectively (B). The adult vertical pattern is already fully formed at 15dpf (C) and further increases in contrast until 21dpf (D) and beyond. Adult H. latifasciatus are characterized by four dark vertical bars (E). The region between the bars (interbars) have yellow to orange-red hues. (F–R) High magnification microscopy images showing the development of ib1, vb2, ib2, and vb3 from 9 dpf to 21 dpf (F–P) and adult male (Q) and female (R). Interbar melanophores are often lighter, with parts of the chromatophore or the center being almost pigment-free (F–P). Vertical bar pattern does not differ between sexes in adults (Q, R). Abbreviations: dc: dorsal cluster, lc: lateral cluster, vb: vertical bar, ib: interbar. Scale bars are 1 mm in (A–D, F–P), 2 mm in (Q, R).

– 94 –

⁕ Chapter V ⁕

To investigate the formation of vertical bars (Fig. V.1E), we described the development of H. latifasciatus larvae between 7- and 21-days post-fertilization (dpf; Fig. V.1A–D, F–P and S.V.1). The developmental progression of vertical bar formation is fully consistent in all three individuals examined (Fig. S.V.1, S.V.2). At 7 dpf, melanophores are present in the dorsal head region as well as on the dorsal part of the yolk sac. In the trunk area only a few melanophores have formed without any obvious indication of a bar-like pattern (Fig. V.1A). Starting at 8 dpf vertical bars form in an anterio-posterior sequence (Fig. V.1B–D, F–P and Fig. S.V.1). Trunk melanophore number has increased considerably within an anterior dorsal patch (dc1) forming at 8 dpf (data not shown), followed by a more posterior one at 9 dpf (dc2; Fig. V.1B). The melanophore patches anticipate the position of vb1 and vb2, respectively (Fig. V.1B, C). At 10 dpf a third and fourth dorsal patch (dc3 and dc4) are appearing at the positions where vb3 and vb4 will later form, respectively (Fig. V.1C, G and S.V.1, S.V.2). After the appearance of the dorsal patches they expand dorsally into the dorsal fin and ventrally forming the four bars. One exception is the posterior bar vb4, where the bar only extends to the horizontal myoseptum. The formation of vb3 is furthermore contributed by a second melanophore cluster (lateral cluster; lc) that forms in a more posterior-ventral position and merges with the developing bar between 12 and 13 dpf (Fig. V.1C, I, J and Fig S.V.1, S.V.2). After this time, at around 2 weeks post-fertilization, the complete adult vertical pattern is already fully formed, but the contrast of the bars further increases until 21 dpf (Fig. V.1D, P and S.V.1) and beyond.

Cellular Correlates of Vertical Bar Formation

In order to understand the formation of the characteristic bar pattern of H. latifasciatus, we analyzed the progression of the bar pattern formation over time. Specifically, we analyzed how the darkening of the bar regions is generated at the cellular level. We hypothesized that three processes might contribute to the darker appearance of the bar regions: (a) melanosome dispersal (the aggregation and dispersion of melanosomes, the melanin-containing organelles, within melanophores) as it increases the fraction of the tissue covered by melanin, (b) the density of melanophores, and (c) the darkness of the melanophores, i.e., the concentration of melanin.

– 95 –

⁕ Chapter V ⁕

Fig. V.2 Cellular changes during vertical bar formation. (A–C) Measurement of melanosome dispersal diameter, melanophore density and the darkness of the melanophores focusing on the dorsal portion of vb2, ib2, and vb3 illustrate changes in chromatophore number and characteristics during bar pattern formation between 9 and 21 dpf. No obvious difference could be found in melanophore dispersal diameter between bars and interbars (A). The melanophore density increases in bars compared to interbars (B). Relative gray values of melanophores increase as well (C). (D–M) Time-lapse vertical bar development in the same individual from 9 to 13 dpf as photographs (D–H) and schematics (I–M). The melanophore with the red cross in (D–M) was used to align the images. Dots show the position of melanophores in (I–M). Position on the previous days (based on image overlay) are labeled in gray. Red arrowhead in (D–H) and asterisks in (I–M) indicate where the formation of a new melanophore will occur, red arrows in (D–H) and black dots next to a gray asterisk in (I–M) the forming melanophore on the consecutive day. Scale bars are 0.5 mm in (D–M).

To investigate the development and the importance of these factors we followed the development of three individuals over twelve days of development between 9 and 21 dpf focusing on the dorsal portion of the two bars in the trunk region, vb2 and vb3 and the yellow interbar region in between (ib2). To assess melanosome dispersal, we calculated the diameter of the melanin-covered part of all melanophores (see section Materials and Methods). We found no strong spatial difference in melanosome aggregation (Fig. V.2A and S.V.1), yet dispersal diameters increased with age, likely because cells are still growing during these early developmental stages (Fig. SV.3). Cell density was evenly distributed at 9 dpf, but during the formation of the bars, cells became more densely packed in the bar regions, while cell density decreased in the interbar regions (Fig. V.2B and S.V.1). To measure the darkness of individual melanophores we measured relative gray values (see section Materials and Methods). Here the difference

– 96 –

⁕ Chapter V ⁕ between bar and interbar regions continuously increased suggesting stronger accumulation of melanin in bar melanophores (Fig. V.2C and S.V.1). Indeed, closer observation of melanophores shows that interbar melanophores are often not fully filled with melanin, with parts of the chromatophore, often the center of the cell being poorly pigmented (Fig. V.1N–P and S.V.1). Next, we investigated the cellular behavior of single melanophores between 9 and 13 dpf. To do so, we photographed the vb2 region of the same individual on five consecutive days and tracked cellular migration and the formation of new melanophores (Fig. V.2D–M). The data suggests that melanophores mainly move indirectly through the general expansion of the skin. Newly differentiating melanophores could be found across the whole examined area, but at an increased rate in the forming bars. They grow to the full size (diameter: ∼0.06 mm) within 1–2 days. We found no evidence of proliferating melanin-containing cells. In summary, these results suggest that the bars of H. latifasciatus form through spatial variation in melanophore properties (i.e., melanin content) and melanophore cell density that mainly arise in the second week after hatching (standard length 6–8 mm). The increase in melanophore density is most likely caused by an increased differentiation of melanoblasts within the bar regions.

Adult Patterns in H. latifasciatus

To investigate the distribution of chromatophores in different integument regions, we estimated cell density and size of both, melanophores and xanthophores in the two interbars (ib1, ib2) and the two bars (vb2, vb3) of the trunk region (Fig. V.1E). Here, we only quantified pigment cells in female individuals as the vertical bar pattern does not differ between sexes, but substantial interindividual variation in red coloration of males would have complicated quantifications (Fig. V.1Q, R). Scales and dermis without scales (from hereon called “scales” and “skin”, respectively) were analyzed separately. Three types of chromatophores could be found in both the dark bars and the light interbars: melanophores with black/dark brown pigments, xanthophores with yellow to orange and reddish pigments (we did not differentiate between xanthophores and erythrophores; see discussion in section Materials and Methods), and iridophores with iridescent/reflective properties (Fig. V.3I, L). To quantify pigment cell quantity and properties we used three

– 97 –

⁕ Chapter V ⁕ measurements: (a) pigment cell coverage, (b) pigment cell dispersion, and (c) pigment cell density. Pigment cell coverage is influenced by both cell number and size (or intracellular dispersal of pigments in the cell) and was measured by estimating the percentual coverage of the tissue with pigment using light microscopy for melanophores and fluorescence microscopy to detect the autofluorescence of xanthophores. Consistent with the visual impression, melanophore coverage in bars was significantly higher than in the

Fig. V.3 Photographs of scale and skin dissections in bar and interbar region. (A–H) Photographs of scales from interbar region (A–D) and bar region (E–H). Melanophores from scales of interbars (A) and bars (E) aggregate after epinephrine treatment (B,F) allowing accurate quantification. Insets of (A,B,E,F) show the single melanophores before and after epinephrine treatment. Xanthophores (yellow) from scales of interbars (C) and bars (G) were detected via their autofluorescence (D,H). (I–N) High magnification photographs skin, where scales have been removed. Brightfield images of interbar (I) and bar (L) fluorescent images of interbar (J) and bar (M); brightfield images after epinephrine treatment for interbar (K) and bar region (N). Scale bars are 500 μm in (A, B, E, F); 50 μm in (C, D, G, H); 100 μm in (I–N).

– 98 –

⁕ Chapter V ⁕

Fig. V.4 Chromatophore measurements in adult H. latifasciatus. (A–K) Chromatophore coverage (A, B, G, H), dispersal diameter of cells (C, D, I, J) and cell density (E, F, K) of melanophores (A–F) and xanthophores (G–K) were estimated in scales and skin from two interbars (ib1 and ib2) and two bars (vb2 and vb3). P-values are based on ANOVA and Tukey–Kramer post hoc tests. Each dot represents the mean value of one individual (full data see Fig. S.V.4–6). Black/orange lines depict the mean of the three individuals. ***P < 0.001; **P < 0.01; *P < 0.05.

yellowish interbars in both scales (26.4% in bars, 2.0% in interbars) and skin (75.9% in bars, 13.7% in interbars) (Fig. V.3, V.4A, B; Fig S.V.4 and Table S.V.1). Both melanophore density and melanosome dispersal contributed to the difference in melanophore coverage. The average melanosome dispersal diameter is larger in dark bars (Ø 0.058 mm in scale, Ø 0.082 mm in skin) than light interbars (Ø 0.022 mm in scale, Ø 0.033 mm in skin) (Fig. V.4C, D; Fig. S.V.5 and Table S.V.1). However, variation within the same skin region was quite high suggesting that both dispersed and aggregated melanophores are widely distributed (Fig. S.V.5 and Table S.V.1). The density of melanophores in both scale and skin were significantly higher in the bars (158 cells/mm2 in scale, 333 cells/mm2 in skin) than in interbars (52 cells/mm2 in scale, 115 cells/mm2 in skin) (Fig. V.4E, F; Fig. S.V.6, and Table S.V.1). – 99 –

⁕ Chapter V ⁕

Xanthophore coverage in bars and interbars is similar on scales (15.7% in bars, 18.4% in interbars), but significantly different in the skin (10.9% in bars, 50.4% in interbars) (Fig. V.3D, H, J, M, V.4G, H; Fig. S.V.4, and Table S.V.1). However, xanthophore coverage might be underestimated in the skin of the bar regions due to the high density of melanophores that could aggravate detection of the xanthophore autofluorescence. The dispersal diameter of xanthophores did not differ between interbars (Ø 0.048 mm in scale, Ø 0.025 mm in skin) and bars (Ø 0.044 mm in scale, Ø 0.021 mm in skin), neither in scales nor in skin preparations (Fig. V.4I, J; Figure S.V.5 and Table S.V.1). Cell densities were comparable in scales of interbars (157 cell/mm2) and bars (155 cell/mm2) (Fig. V.4K; Fig. S.V.6 and Table S.V.1).

Gene Expression Associated with the Bar Patterns in H. latifasciatus

Next, we analyzed the molecular correlates of the observed differences in pigment cell density and pigment synthesis. Molecular markers for iridophore and xanthophore were also used as we could not analyze differences in these cells’ number due to the high melanophore density in the bar regions. Therefore, we compared expression levels of twelve candidate genes across the same two bar and interbar regions for the chromatophore measurement (ib1, vb2, ib2, vb3) using quantitative real-time PCR (qPCR; Fig. V.5 and Table V.1). The selected genes are including marker genes for chromatophores, melanin synthesis genes and genes involved in the melanocortin signaling pathway. The latter was a particular focus as they have been previously implicated in color pattern formation of cichlids, teleosts and vertebrates more generally (Table V.1). Sox10 is a marker gene for chromatophore progenitors (Dutton et al. 2001). Expression was slightly higher in bar than in interbar regions (Fig. V.5A) and significantly differed between regions (ANOVA: P < 0.05). However, we found no significant sox10 expression variation between bars and interbars (Tukey HSD: P = 0.067–0.37; Table S.V.2). Expression of the melanophores marker, mitfa (Lister et al. 1999, Béjar, Hong, and Schartl 2003) was higher within the dark vertical bars than in the adjacent light interbars (Fig. V.5B and Table S.V.2), yet differences were not significant (ANOVA: P = 0.19). Similar expression profiles can be also found in pmel, a melanophore specific gene important for melanin deposition in melanosomes (Schonthaler et al. 2005). Also here, we find that pmel is expressed at a higher yet not

– 100 –

⁕ Chapter V ⁕ significantly higher (ANOVA: P = 0.056) level in dark bars compared to interbars (Fig. V.5C and Table S.V.2). The gene slc24a5, a melanosome-specific cation exchanger (Lamason et al. 2005), showed differential expression in some pair-wise comparisons of bar and interbar regions (ANOVA: P < 0.01; Tukey HSD: P = 0.002–0.18; Fig. V.5D and Table S.V.2). The two melanophore-specific genes that express melanogenic enzymes and are essential for the production of melanin, tyr (Hidehito et al. 1994, Camp and Lardelli 2001) and tyrp1a (Braasch, Liedtke, et al. 2009, Krauss et al. 2014) showed significantly higher expression levels in bars (ANOVA: both P < 0.001; Tukey HSD: all P < 0.01; Fig V.5E, F and Table S.V.2).

Fig. V.5 Expression differences of pigmentation candidate genes. (A–K) Quantitative PCR of sox10 (A), mitfa (B), pmel (C), slc24a5 (D), tyr (E), tyrp1a (F), csf1ra (G), ltk (H), asip1 (I), agrp2 (J), mc1r (K), and mc5r (L) mRNA levels along the anterior- posterior axis of adult H. latifasciatus including two vertical bars (gray bar plots, vb2, vb3) and the light interbar regions (yellow bar plots; ib1 and ib2). Differences were tested by ANOVA followed by Tukey–Kramer post hoc test, n = 5 (individual dots). Error bars indicate means + SD. Abbreviations: ***P < 0.001; **P < 0.01; *P < 0.05.

– 101 –

⁕ Chapter V ⁕

Table V.1 Selected candidate genes involved in coloration and pigment patterns in fish. Gene Full name and described function References agrp2 agouti related peptide 2: (Zhang et al. 2010, Shainer A member of the agouti gene family that has been et al. 2017, Kratochwil et al. shown to repress (horizontal) stripe formation in 2018) cichlids. It is also involved in background matching in zebrafish. asip1 agouti signal peptide 1: (Cerda-Reverter et al. 2005, Another member of the agouti gene family that Kurokawa, Murashita, and regulates dorsal-ventral pigmentation in the fish Uji 2006, Guillot et al. 2012, skin. Cal, Megias, et al. 2017, Cal et al. 2019) csf1ra colony stimulating factor 1 receptor a: (Parichy et al. 2000, Parichy A kit-related tyrosine receptor kinase, which is and Turner 2003) essential for migration and survival of xanthophores. ltk leukocyte receptor tyrosine kinase: (Lopes et al. 2008) A tyrosine kinase receptor that is crucial for fate specification of iridophores from neural crest cells. mc1r melanocortin 1 receptor: (Selz et al. 2007, Richardson A G-protein-coupled seven transmembrane helix et al. 2008) receptor that regulate several pigment cell specific processes. Loss-of-function mutation are associated with loss of melanic pigmentation. mc5r melanocortin 5 receptor: (Kobayashi et al. 2010, Another melanocortin receptors expressed in both Kobayashi, Mizusawa, melanophore and xanthophores. Chiba, et al. 2012) mitfa microphthalmia-associated transcription factor a: (Bejar, 2003; J. Lister, A master regulator of melanophore/melanocyte Robertson, Lepage, Johnson, differentiation across vertebrates. & Raible, 1999) pmel premelanosome protein a: (Chakraborty et al. 1996, A melanosome protein that plays an important role Béjar, Hong, and Schartl in the structural organization of premelanosomes 2003, Du et al. 2003, and the formation of intraluminal fibrils during Schonthaler et al. 2005, melanosome biogenesis. Theos et al. 2005) slc24a5 solute carrier family 24 member 5: (Lamason et al. 2005) A transporter protein localized in the melanosomal membrane that is essential for melanin synthesis. sox10 sex determining region Y box 10: (Dutton et al. 2001, A transcription factor that is essential for the Elworthy et al. 2003, Hou, specification of chromatophore progenitors. Arnheiter, and Pavan 2006) tyr tyrosinase: (Korner and Pawelek 1982, An oxidase that can controls the melanogenesis as Hidehito et al. 1994, Camp the rate-limiting enzyme by catalyzing tyrosine and Lardelli 2001) into dopaquinone via L-dopa. tyrp1a tyrosinase-related protein 1a: (Braasch, Liedtke, et al. Another enzyme of the tyrosinase family that 2009, Krauss et al. 2014) catalyzes the melanin biosynthesis.

In order to compare the distribution of iridophores and xanthophores, we used the iridophore lineage-specific marker ltk (Lopes et al. 2008) and the xanthophore marker csf1ra (Parichy and Turner 2003). Notably, both ltk and csf1ra were expressed at similar – 102 –

⁕ Chapter V ⁕ levels across the differently pigmented bar and interbar regions (Fig. V.5G, H and Table S.V.2). This is in support of a rather homogenous distribution of iridophores and xanthophores across the trunk. We also examined the gene expression of Agouti family genes (namely Asip/asip1 across all vertebrates and agrp2 in cichlid fishes) and melanocortin receptors, as they have been previously implicated in pigmentation. Two proteins of the Agouti family, Asip1 and Agrp2, likely act as the antagonists for the melanocortin receptors Mc1r and/or Mc5r in teleost skin (Cal, Suarez-Bregua, et al. 2017). As previous studies suggest, asip1 is a key regulator of dorso-ventral countershading, presumably by regulating melanophore number(Ceinos et al. 2015, Cal, Megias, et al. 2017, Cal et al. 2019). Interestingly we also found significant variation along the anterior-posterior axis (ANOVA: P < 0.01) with asip1 being expressed significantly higher in some pair-wise post hoc comparisons between bars and interbars (Tukey HSD: P = 0.0016–0.8222; Fig. V.5I and Table S.V.2). Previous work in cichlids showed that agrp2 regulates presence/absence of stripe patterns while not contributing to shaping the pigment pattern itself through expression variation across the skin (Kratochwil et al. 2018). In contrast to results from the Lake Victoria species Pundamilia nyererei, we found significant differences between skin regions (ANOVA: P = 0.041; Fig V.5J and Table S.V.2). However, expression mainly differed between anterior and posterior regions (Tukey HSD between ib1 and vb2: P < 0.05) and not consistently between bars and interbars (Tukey HSD: P = 0.046–0.895) (Table S.V.2). Mc1r and Mc5r are two receptors antagonized by Agrp2 and/or Asip1 signaling and mc1r and mc5r have been shown to be expressed in chromatophores. The expression of mc1r has been reported in skin melanophores of zebrafish, barfin, and Japanese flounder (Kobayashi et al. 2010, Kobayashi, Mizusawa, Saito, et al. 2012, Higdon, Mitra, and Johnson 2013) xanthophores of goldfish (Kobayashi et al. 2011) and iridophores of zebrafish (Higdon, Mitra, and Johnson 2013). The expression of mc5r was detected in melanophores and xanthophores of flatfish (Kobayashi et al. 2010, Kobayashi, Mizusawa, Saito, et al. 2012). Here we found no significant expression differences for mc1r (ANOVA: P = 0.296) and mc5r (ANOVA: P = 0.764) which suggest that the melanocortin receptors may not contribute to shaping the vertical bar pattern (Fig. V.5K, L and Table S.V.2).

– 103 –

⁕ Chapter V ⁕

Discussion

Fig. V.6 Cellular and genetic mechanism underlying bar pattern formation in H. latifasciatus. The contrast between bars and interbars is driven by three molecularly likely independent mechanisms: melanosome dispersal (A), density of melanophores (B) and the darkness of individual melanophores (C).

Melanic Pattern Development: H. latifasciatus vs. Other Teleosts

Most previous investigations of teleost pigment pattern formation focused on the horizontal stripe patterns of the model teleost Danio rerio, the zebrafish. In zebrafish, melanoblasts, the progenitors of melanophores, migrate from the dorsal neuroectodermal margin along nerve fibers between the myotomes (Dooley et al. 2013). After they settled at their final position, melanoblasts differentiate into melanophores and accumulate melanin to form the dark horizontal stripes that gave the zebrafish its name. In the haplochromine cichlid H. latifasciatus [divergence time with zebrafish approximately 220 million years (Hughes et al. 2018)] color pattern formation starts with the development of four melanophore clusters that arise in the dorsal rim of the trunk (Fig. V.1B–D). The dorsal clusters initiate the formation of bar patterns later in

– 104 –

⁕ Chapter V ⁕ development, by spreading dorsally into the dorsal fin and ventrally forming bars on the trunk. The third vertical bar (vb3) fuses with the lateral melanophore cluster (Fig V.1C, D, I–P and Fig. S.V.1). In contrast to the zebrafish where adult color pattern develops indirectly during a post-embryonic metamorphic phase (Parichy et al. 2009), the vertical bars in H. latifasciatus develop directly and are already visible at a time when the larvae are still feeding from their yolk. This is in line with previous reports on the direct development of other morphological traits of African cichlid fishes (Woltering et al. 2018). By tracking the behavior of individual melanophores we could observe how new melanophores form over a period of 1–2 days (Fig. V.2 and Fig. S.V.1). It is unclear how the melanoblasts reach this position. However, it is likely that they migrate dorsally from the neuroectodermal margin and then ventrally within the skin, possibly explaining the gradual expansion of bars from dorsal to ventral. This is interesting, because in zebrafish clones of single pigment cell progenitor cells have been shown to mainly spread dorso- ventrally (Singh and Nusslein-Volhard 2015, Nusslein-Volhard and Singh 2017). It is possible that vertical bars are prepatterned at the progenitor level, potentially already at the level of the dorsal neuroectodermal ridge. Position and size of bars could be driven by anterior-posterior differences in progenitor number or transcriptional identity that later influence proliferation and differentiation within the dermis. It is also possible that migratory routes, environmental factors along the migratory pathways and in the skin as well as cell-cell interactions including reaction-diffusion systems contribute to or constitute the basis of this process. The anterior-posterior sequence of bar formation could be a consequence of the anterior-posterior sequence of somitogenesis. Anterior melanoblast would hereby first receive guidance cues from dermamyotome, sclerotome, and ermerging dermis. Anterior neural-crest cells including pigment cell precursors would therefore migrate first as previously described (Bronner and Ledouarin 2012). As soon as melanophores are visible in the skin they do not migrate anymore and grow to the full size within 1–2 days. Hereby the cell-size increases, new dendritic arbors form and melanin levels visibly increase (Fig. V.2 and Fig. S.V.1). Therefore, the appearance of the bar pattern is mainly driven by an increase in cell density in the developing bar region as well as changing cellular characteristics including increased melanosome dispersal and melanin production (Fig. V.2, 6 and Fig. S.V.1).

– 105 –

⁕ Chapter V ⁕

As one of the most common pigment patterns in haplochromine cichlids, vertical melanic bars can vary in contrast and number, both between as well as within species (Witte et al. 1976, Kocher et al. 1993, Seehausen, Mayhew, and Van Alphen 2001). However, compared to most other bared haplochromine cichlids, the vertical bars in H. latifasciatus are fixed in number (mostly four and in some strains five) and are relatively thick. Our study shows that vertical bar number remains constant over the course of development in H. latifasciatus (Fig. V.1F–R, V.2D–M and Fig. S.V.1), which contrasts with a recent study on the Lake Malawi cichlid Copadichromis azureus, a species with unfixed bar number and thinner bars (Hendrick et al. 2019). In C. azureus, vertical bars split during development resulting in an increase of bar number (Hendrick et al. 2019). However, both processes are not driven by rearrangement of excisting pigment cells, but both broadening of bars in H. latifasciatus and increase in bar number in C. azureus is driven by the formation of new melanophores.

Cellular Correlates of Bar Patterns and Underlying Mechanism

Comparisons of juvenile and adult patterns largely demonstrate that the juvenile patterns bear already most cellular and morphological characteristics of adult patterns. Vertical bars have a higher density of melanophores and melanophores are also evidently darker in the melanic regions (Fig. V.4E, F, V.6). In contrast to juveniles, pigments are more dispersed in adult bar melanophores. Xanthophores show no clear differences, yet reliable quantification was only partially possible due to the high density of melanophores. Gene expression analyses of known pigmentation genes gave complementary information as well as mechanistic insights on how the spatial variation of chromatophore properties and densities are achieved (Fig. V.5, V.6). Clearly, melanophore-specific genes such as tyr, tyrp1a, but also slc24a5 are differentially expressed between bars and interbars (Fig. V.5D–F and Table S.V.2). All three genes are essential for melanin production. The two tyrosinases are directly involved in the synthesis of melanin, slc24a5, a potassium-dependent sodium/calcium exchanger is thought to modulate melanosomal pH, which is a crucial parameter for melanin synthesis (Ginger et al. 2008). Also pmel, the premelanosome protein, another melanophore- specific gene is slightly, yet not significantly, upregulated in bar regions compared to interbar regions (Fig. V.5C and Table S.V.2). Overall, expression data show a sixfold

– 106 –

⁕ Chapter V ⁕ higher expression of tyrosinases in bars providing an explanation for the darker appearance of bar melanophores (Fig. V.6 and Table S.V.2). Even, if we standardize the differential expression by the increased cell-number or the melanophore-lineage marker mitfa we still find a 2–4.5-fold (Fig. S.V.7 and Tables S.V.3, S.V.4) increase of tyrosinase expression and therefore likely melanin synthesis. Although we did not find significant differential expression of the transcription factors sox10 and mitfa (Fig. V.5A, B), both crucial factors for melanophore differentiation, expression is slightly elevated in bar regions. As mitfa expression has been reported to be only weak in differentiated melanophores (Johnson, Nguyen, and Lister 2011), this result might indicate an elevated number of melanoblasts. Although this would have to be analyzed in detail, it would explain the increased number of newly forming melanophores in the bars of juveniles (Fig. V.2 and Figure S.V.1) and consequently an increase in melanophore density. Additionally, we used a marker gene for the xanthophore lineage (csf1ra) and the iridophore lineage (ltk). None of them showed any expression differences (Fig V.5G, H), suggesting that there is no obvious difference in xanthophore and iridophore number. Therefore, melanophores seem to be the main cell type that clearly differs both in number and expression of marker genes between bar and interbar regions. Although there is some indication of a decreased number and coverage of xanthophores in the bar regions, this analysis was greatly hindered by the dense melanophore coverage. Use of mutant lines as for example of the oca2 gene that will likely not affect the pattering of xanthophores and iridophores (Kratochwil, Urban, and Meyer 2019) would be a possibility to better assess this question. Still, this result is somewhat surprising as cell-cell interactions between xanthophores, melanophores and iridophores seem to have a completely different dynamic in cichlids compared to zebrafish, where these cell types are spatially segregated (Mahalwar et al. 2014, Patterson, Bain, and Parichy 2014, Singh, Schach, and Nusslein-Volhard 2014, Eom et al. 2015, Eom and Parichy 2017). In zebrafish, especially xanthophores and melanophores are almost mutually exclusive, with a decreased cell number of xanthophores in the melanic stripes. No evidence of such a strong antagonistic interaction could be found in H. latifasciatus. In xanthophore morphology though, zebrafish and H. latifasciatus have some shared featured: xanthophores in melanic regions are more irregular and seem smaller and have fewer and not as evenly distributed

– 107 –

⁕ Chapter V ⁕ dendritic arbors (Fig. V.3, V.4, S.V.5, and Table S.V.1), although we did not find significant differences in xanthophore diameter (Fig. V.4I, J and Table S.V.1). Our result on the expression of mc1r and mc5r shows that the melanocortin receptors likely play no role in shaping the vertical bar patterns in H. latifasciatus (Fig. V.5K, L). The antagonists asip1 and agrp2 show significant differences in expression (Fig. V.5I, J and Table S.V.2). The gene agrp2 has been shown to inhibit the formation of horizontal stripe patterns: species with high agrp2 expression lack any horizontal stripe patterns, while species with horizontal stripes have generally low expression of agrp2 in skin. Yet, spatial variation in agrp2 expression does not seem to be necessary for the formation of stripes (Kratochwil et al. 2018). On the other hand, a gradient in agrp2 expression along anterio-posterior axis could be found in both H. latifasciatus with significant difference from this study (Fig. V.5J and Table S.V.2) and P. nyererei without significant difference from our previous study (Kratochwil et al. 2018). However, based on our previous work that showed no changes in bar patterns in a knockout of agrp2 in P. nyererei (Kratochwil et al. 2018) or association with bar presence and absence (Kratochwil et al. 2019), a role in vertical bar formation seems rather unlikely. The variation in asip1 expression is intriguing. The melanocortin signaling antagonist asip1 has higher expression in interbars (twofold difference; Table S.V.2) with a particularly high expression in the first interbar. This is surprising as asip1 (and also the tetrapod homolog Asip) was previously known to vary along the dorso-ventral axis where it is involved in generating the dorso-ventral countershading that can be seen in many vertebrates (Manceau et al. 2011, Cal, Megias, et al. 2017, Haupaix et al. 2018, Cal et al. 2019, Kratochwil 2019). Here we observed variation along the anterio-posterior axis (Fig. V.5I and Table S.V.2), suggesting that gene expression differences of asip1 might contribute to variation in pigmentation on both axes. As Asip1 acts as Mc1r/Mc5r antagonist, one of the responses of decreased melanocortin signaling would be an increased aggregation of melanosomes. This would be in line with the significant difference in dispersion diameter we observe between bars and interbars. In interbars the diameter is on average 2.8 times smaller (Fig. V.4C,D and Table S.V.1), contributing to the lighter appearance of the interbar region (Fig. V.6). Yet the variation of asip1 expression serves only as a partial explanation as (a) the expression level in int2 is similar to the adjacent bars and (b) the dispersion/aggregation states of melanophores vary greatly within bars and interbars as well as between individuals. Still, based on these

– 108 –

⁕ Chapter V ⁕ findings, it might be an important take-home-message to consider functions of asip1 beyond dorso-ventral patterning (i.e., countershading).

Conclusion

In summary, we investigated the formation of the vertical bar color pattern of H. latifasciatus – a member of the phenotypically diverse East African haplochromine cichlid fish radiations – during embryonic and larval development, how the macroscopic patterns is formed through variation in pigment cell distribution and properties, and how this variation links to known coloration gene expression. Our work provides novel insights into the molecular and cellular properties that contribute to the formation of color patterns in this famously diverse family of fish. More specifically we demonstrate that bar pattern formation is facilitated by three molecularly likely independent mechanisms: increased melanosome dispersal (controlled by melanosome migration along the cytoskeleton), density of melanophores (controlled by proliferation in progenitors) and melanin synthesis (controlled by melanin synthesis pathways and melanosome micro- environment) (Fig. V.6). H. latifasciatus with its – as we describe here – morphologically and transcriptomically well defined bar and interbar regions that can be tracked throughout development at a cellular level provides a unique system to further understand the molecular and cellular underpinnings of color patterns. However, further investigations on the development of vertical bars by assaying gene expression using RNA-seq, in situ hybridization and immunohistochemistry as well as comparative analyses with closely related species will be particularly informative for further understanding the molecular mechanisms underlying vertical bar formation. Cichlids with a rich and expanding repertoire of experimental approaches including hybrid crosses, transgenesis and CRISPR-Cas9 genome editing (Kratochwil and Meyer 2015a, Juntti 2019) will make this species an excellent choice for further investigating the causal genetic variants, genes and molecular mechanisms that influenced evolution of and variation in vertical bar pattern formation.

– 109 –

⁕ Chapter V ⁕

Materials and Methods

Fish Husbandry and Embryo Culture

H. latifasciatus were kept in groups of 10–25 individuals. Fertilized eggs were removed from the gravid females as early as possible. The larvae and juvenile fish were then raised in egg incubators (ZET-E55, Ziss aqua) at room temperature. At 5 days post-fertilization (dpf) three larvae were separated from their siblings and each of them was raised in a single incubator to be able to keep track of a single individual in the following weeks. The hatching fish larvae were provided with nutrients by their yolk sac which lasts for ∼14 days. After 14 dpf the larvae were fed on Artemia nauplii twice a day. Experiments were performed in accordance with the rules of the animal research facility of the University of Konstanz, Germany and with the permission of the animal care committee (Regierungspräsidium) Freiburg, Germany (G18/60 and T16/13).

Fish Larvae Photography and Analysis

Photographs of H. latifasciatus embryos and larvae were captured with a stereomicroscope (Leica MZ10F) with a Leica DMC2900 color camera. Fish were first anesthetized with 0.04% tricaine (MS-222, Sigma-Aldrich). Images were taken as previously described (Kratochwil, Sefton, and Meyer 2015, Kratochwil et al. 2017). In order to capture the color pattern development of H. latifasciatus, photographs were taken from 7 to 21 dpf of three individuals. For quantitative analyses, a comparable area including vb2 and vb3 of photographs of three individuals from stages between 9 and 21 dpf were put together, aligned (using the anterior dorsal fin as landmark), transformed into a black-and-white image and quantified in three different ways using Fiji (Schindelin et al. 2012). Melanosome dispersal was estimated by measuring the diameter of the minimally sized circle that encloses all melanic parts of a melanophore (minimal enclosing circle). It is therefore affected both by the melanophore size and the dispersal/aggregation state of melanosomes within the melanophores. The relative gray value was measured by taking the mean gray value of the same circle and dividing it by the mean of all melanophore gray value measurements of the same individual and stage. This relative value was used to account for differences between images and stages. Melanophore density was calculated from manual melanophore counts. As the position of the bars would shift as – 110 –

⁕ Chapter V ⁕ the fish is growing, we corrected the position values by the growth of the individual (size of individual at X dph/size of individual at 9 dph). The values were plotted using non- parametric regression (locally weighted scatterplot smoothing; LOWESS) in R (f parameter: 0.1). For the density calculation the area was split into ten equally-sized zones. To track individual melanophores we took images of the same individual on five consecutive days (from 9 to 13 dpf; Fig. V.2D–H). The identity of single melanophores across stages was estimated using overlays of the images from multiple days.

Image Acquisition in Adult Fish

Three female individuals (standard length ∼12 cm) were examined for pigment quantification. We only quantified the chromatophores in females as the vertical bar pattern does not differ between sexes, we found more variation in red coloration in males (Fig. V.1Q,R). For each fish, in total 156 photos were obtained from scales and skin with scales removed (referred to as skin). Firstly, we separated the two flanks with scales still attached to skin. Scales from study regions were then carefully removed from the left flank and kept separately in Ringer’s solution (6.5 g/L NaCl, 0.25 g/L KCl, 0.3 g/L CaCl2 and 0.2 g/L NaHCO3) at 4°C before imaging. Five light microscope photographs were taken on each melanic and non-melanic region (the two melanic bar regions vb2 and vb3, and the interbar ib1 which anterior of vb2 and ib3 which between vb2 and vb3) from dorsal to ventral with a Leica MZ10F stereomicroscope equipped with a Leica DMC2900 color camera. Four photos were taken also along the dorsal-ventral axis on both region using Lecia MZ10F equipped with a Leica DFC 3000G black-and-white camera with GFP filter to image the auto-fluoresce of adult xanthophores (Kelsh et al. 1996, Le Guyader and Jesuthasan 2002). For each pigmented region, ten scales were imaged using a Leica DMC2900 camera on Leica DM6B upright microscope. Leica Application Suite X software was used to capture the photos using the same setting. To count the number of melanophores, the right flank including both scales and skin tissue was treated with 10 mg/ml epinephrine (SIGMA-ALDRICH) for 20 min at room temperature to aggregate the melanosomes and thereby permit robust cell number quantifications (Fig. V.3B, F, K, N). After epinephrine treatment, tissue was washed by Phosphate Buffered Saline (PBS, pH7.4) three times and then kept in 4% Paraformaldehyde (PFA) in PBS. Epinephrine-treated scales were removed from skin and photos were taken for both skin

– 111 –

⁕ Chapter V ⁕ and scales as described above. Fluorescence images were not taken as this treatment caused a high autofluorescence background.

Image and Data Analysis of Adult Patterns

All photos were analyzed with the image analysis software Fiji (Schindelin et al. 2012). At the beginning of the analysis we manually adjusted the color threshold to obtain reliable quantification of the bright field photos from non-treated skin and scales. Using this setting we executed the “Analyze Particles” function to obtain the melanophore coverage. For 20 randomly selected melanophores from each skin image and 10 melanophores from each scale image we measured the dispersal diameter of melanin covered parts of the melanophores (as described above). For xanthophore coverage we used the same approach. The dispersal diameter of the pigment filled part of 20 xanthophores of each skin image and 10 xanthophores of each scale image was measured from each photograph. The number of melanophore and xanthophore was counted from epinephrine-treated skin and scale specimens. Single xanthophores on scales could be observed easily by fluorescence microscopy, while not all boundaries of xanthophores in the skin could be easily identified as the cells often overlapped. Therefore, we were able to count the xanthophore number in scales but not in the skin dissections, while dispersal diameter measurements for xanthophores was possible on both scales and skin. Xanthophores and (what likely are erythrophores) was treated as the same cell type. Several studies show that erythrophores also exist in cichlids (Chen et al. 2014, Chen et al. 2015). However, vesicles containing pteridine and carotenoids could be found in the same cells, in which case the overall color depends on the ratio of red and yellow pigments (Matsumoto 1965, Bagnara 1966). Hence, the distinction between xanthophores and erythrophores is not always clear. Therefore, we classified yellow/orange/red colored cells all as xanthophores. Although we could identify iridophores in both dark bars and light interbars before and after epinephrine treatment (Fig. V.3I, K, L, N), iridophores were mostly below or above melanophores hindering reliable measurements. We, therefore, used gene expression of the iridophore lineage marker gene ltk (Lopes et al. 2008).

– 112 –

⁕ Chapter V ⁕

RNA Extraction

To compare the expression of coloration and pigment genes between melanic bars and interbars regions, we sampled the whole melanic and non-melanic skin region. Skin tissue was dissected and kept in RNAlater (Invitrogen) at 4°C overnight and then transferred to −20°C for long-term storage. RNAlater was removed prior to homogenization. Skin samples and the appropriate amount of TRIzol (Invitrogen) (1 ml TRIzol per 100 mg sample) were homogenized in 2 ml Lysing Matrix A tube (MP Biomedicals) using FastPrep-24 Classic Instrument (MP Biomedicals). RNA was extracted according to the manufacturer’s recommendations (Invitrogen) with an additional wash step by 75% Ethanol. Subsequent purification and on-column DNase treatment were performed with the RNeasy Mini Kit (Qiagen) and RNase-Free DNase Set (Qiagen). Following extraction and purification, RNA was quantified using Qubit RNA BR Assay Kit (Invitrogen) with Qubit Fluorometer (Life Technologies).

Quantitative Real-Time PCR (RT-qPCR)

Gene expression analyses were performed on two melanic (vb2 and vb3) and two non- melanic (ib1 and ib2) skin regions. First strand cDNA was synthesized by using 1μg of total RNA with a GoScript Reverse Transcription System (Promega). qPCRs were performed with 2 μl of 100 μl synthesized first strand cDNA that was diluted ten times from 20 μl of initial reaction volume as a template, 10pmol of each forward primer and reverse primer, and GoTaq qPCR Master Mix (Promega) with nuclease-free water to make the final volume of 20 μl in a 96-well plate. Twelve genes were processed to examine the expression level including sox10, mitfa, csf1ra, ltk, pmel, slc24a5, tyr, tyrp1a, asip1, agrp2, mc1r, mc5r (Table V.1). Primers are listed in Table S.V.5. We used 40 cycles of amplification on a CFX96 Real-Time PCR Detection System (Bio-Rad). The amplification program was: initial denaturation at 95°C for 10 min, 40 cycles of 95°C for 20 s, 60°C for 60 s. At the end of the cycles, melting curve of the products was verified for the specificity of PCR products. Only samples with one peak in the melting curves were processed to analyses. We assayed gene expression in triplicate for each sample and normalized the data using the reference genes β-actin and gapdh. Ct values were defined as the point at which fluorescence crossed a threshold (RCt) adjusted manually to be the point at which fluorescence rose above the background level. Next, we compared the relative expression between samples using the 2–ΔΔCT method (Nolan, – 113 –

⁕ Chapter V ⁕

Hands, and Bustin 2006). For group comparisons, we used ANOVA followed by Tukey’s HSD. All statistical tests were performed in R (R Development Core Team 2019).

– 114 –

⁕ General Discussion ⁕

General Discussion

Main contribution

The chapters of this thesis contribute to our understanding of the molecular and cellular mechanisms driving the evolution of color patterns in cichlid fishes and thus provide excellent examples for investigating the origin of biological diversity. The diverse color patterns displaying in the integument of animals are the among the most “obvious” but also among the most “complex” phenotypic traits, even the same coloration (e.g. black/dark color) requires different chemical and cellular components (eumelanin is synthesized by melanocytes and deposited in hairs and feathers of homeotherms; melanin of poikilotherms is synthesized and storaged in melanophores intracellularly), not to mention the differences in pigmentation networks that including numerous molecular signaling pathways and hormone systems. There is no doubt that color patterns have key functions in sexual selection and natural selection (Protas and Patel 2008). Significant process has been made in model organisms identifying the molecular mechanisms controlling the differentiation, migration, proliferation and maintenance of skin chromatophores, and mechanisms of pigment patterning (Parichy and Spiewak 2015, Singh and Nusslein-Volhard 2015, Irion, Singh, and Nusslein-Volhard 2016, Irion and Nusslein-Volhard 2019, Patterson and Parichy 2019). Yet, we are still far from an integrated understanding of the molecular, cellular, developmental mechanisms that shape the remarkable diversity of color patterns across taxa and the underlying genetic changes drive the diversification. Cichlid fishes are famous for their rapid phenotypic diversification, ecological adaptation and speciation (Kocher 2004, Santos and Salzburger 2012). The main focus of this thesis is to study one of the most diverse selection of traits in cichlid fishes, their color patterns (Salzburger, Braasch, and Meyer 2007, Brzozowski et al. 2012, Henning et al. 2013, Maan and Sefc 2013, Kratochwil 2019).

The molecular basis of discrete color patterns across cichlid fishes and teleost fishes

Chapter I of this thesis focuses on the genetic basis of horizontal stripes and seeks to address the question of evolution’s predictability (Gould 1990). Using quantitative trait locus (QTL) mapping, the horizontal stripes of the East African cichlid fishes are found

– 115 –

⁕ General Discussion ⁕ to be inherited as a recessive Mendelian trait mapping to an interval containing agouti- related peptide 2 (agrp2), a teleost-specific agouti gene. Reduced expression of agrp2 is convergently associated with the presence of stripe patterns, suggesting the expression levels of agrp2 serve as a threshold that permits the phylogenetically observed rapid losses and revolutions of melanic stripe patterns in African cichlid radiations. Interestingly, independent regulatory mutations found to control the expression levels of agrp2, which represent the case of mutational divergence but functional convergence: different regulatory mutations in the same gene have similar effects on the alteration of expression and thus result in similar phenotype (Manceau et al. 2010). Chapter III further addresses the role of agrp2 expression in inhibiting stripe patterns was repeatedly cooped during teleost fish evolution. What remains unknown is the developmental origin of stripes and if developmental changes underlie the evolution of stripe patterns. Heterochrony—evolutionary changes in the timing of development (Jacob 1977, Arthur 2004)—is possible to explain the interaction between the different agrp2 alleles and developmental time: low agrp2 expression of striped species is a retention of the juvenile expression pattern (paedomorphosis), whereas agrp2 expression of nonstriped species continues to increase and thus inhibiting stripes formation in the adult form (peramorphosis). Though, the difference of agrp2 expression between striped and nonstriped species may occur already in the very beginning of development, indicating rather the case of heterometry, the evolutionary changes in the amount of development (Abzhanov et al. 2004, Abzhanov et al. 2006). Using a combination of multiple approaches, the evolutionary history, novel noncoding exon, insertions, deletions and tandem exon duplication at agrp2 locus are comprehensively described in Chapter II of this thesis. In particular, the dynamics of tandem duplication and replicate loss of the terminal exon (exon 3) of agrp2 may contribute to the diversity of color patterns in cichlid fishes. Exon 3 was duplicated before the East African cichlid radiations, then the 3’ duplicate was partially lost again or remains as a non-functional duplicate. Additional duplication, deletion of the 5’ duplicate along with the resurrection of the non-functional duplicate also occurred in some species. Tandem duplications of a gene can contribute to the formation of new genes with novel functions, thus serving as a source of genetic novelty (Ohno 1970). Similar manner may also apply to the scenario here: one copy of exon 3 of agrp2 is freed from stabilizing selection, the other copy/copies can accumulate mutations openly thus result in

– 116 –

⁕ General Discussion ⁕ neofunctionalization and phenotypic diversification (Kondrashov and Koonin 2001, Liu and Guo 2017). Yet, to date available data and technology are too limited to verify the above possibility and the link between gene structure variation and divergence of cichlid color patterns. Chapter III is the first study to reveal the functional conservation and divergence of agrp2 and its paralog agouti-signaling protein 1 (asip1) across the radiation of teleost fishes. First, the manuscript again highlights both agrp2 and asip1, two skin-specific agouti family genes, are repeatedly associated with the evolution of convergent color patterns: agrp2 acts as stripe repressor across the teleost fish radiation as above mentioned and asip1 has a highly conserved role in dorsa-ventral countershading of vertebrate (Manceau et al. 2011, Cal, Megias, et al. 2017, Cal et al. 2019). This study is the first to uncover the striking association between stripe-interstripe patterning and the expression of asip1 across teleost fishes and agrp2 in some teleost species, suggesting an even more complex and integral patterning network controlling the presence/absence of stripes and the formation of countershading and stripe patterns (Ceinos et al. 2015). Particularly in zebrafish Danio rerio, the ubiquitous expression of agrp2 is under the threshold allowing the presence of horizontal stripes, meanwhile agrp2 and asip1 have complementary interstripe expression, with agrp2 being expressed in the dorsal interstripes and asip1 being expressed in the ventral interstripes. This finding, though further data is still needed, is particularly informative for understanding the evolutionary events occurred during the evolution of the agouti gene family (Braasch and Postlethwait 2011, Vastermark et al. 2012), e.g. agrp2 might retain functional properties of asip1 in stripe-interstripe patterning, while at the same time repeatedly sub- and neofunctionalized to become a “stripe-repressor” across teleost fishes. Additional investigation on the developmental of horizontal stripes by assaying temporal and spatial expression of both gene will be needed to test the evolutionary tinkering (Jacob 1977) in heterochronic changes, heterometric changes and especially heterotopic changes—the evolutionary changes in the location of development (Guerreiro et al. 2013). The study presented in Chapter IV investigates another major aspect of color pattern diversity among cichlid fishes: color dimorphisms. Several cichlid species undergo complex changes in color patterns at particular life stages or as responses to environmental factors. These morphological color changes include sexual dimorphisms that are linked to sex and dominance (Chapter IV of this thesis), sex-associated color

– 117 –

⁕ General Discussion ⁕ polymorphisms (Streelman, Albertson, and Kocher 2003, Roberts, Ser, and Kocher 2009) and sex-independent polymorphisms (Henning et al. 2010, Henning et al. 2013). In this manuscript, RNA-sequencing was performed on the yellow and dark morph of the Malawi Golden cichlid fish, Melanchromis auratus (Fryer and Iles 1972), to examine the transcriptomic changes that associated with morphological differences. Despite the differences in pigmentation gene expression were found in this study, the number of these genes is little and exclusively higher expressed in the yellow morph. In particular, two melanophore differentiation gene, foxq1a (Bagati et al. 2018) and mitfa (Lister, Close, and Raible 2001, Elworthy et al. 2003, Johnson, Nguyen, and Lister 2011) are at higher expression levels in the skin of yellow morph, indicating large reservoirs of undifferentiated melanoblasts and only differentiate when the color change occurs. Yet, the mechanisms keeping melanoblasts in the undifferentiated state still need to be addressed. One of the most important findings of this study is the association of morphological differences in coloration with increased neural innervation. The increased neural innervation may have an instructive role to initiate the color change by triggering remodeling of the cytoskeleton (Sugimoto 2002), apoptosis of chromatophores (Sugimoto, Uchida, and Hatayama 2000, Sugimoto 2002), and proliferation of chromatophores and changes in pigment synthesis (Hara et al. 1996). However, the increase of neural innervation is rather ubiquitous, thus, the question of how the color change of different regions is orchestrated will need to be further addressed. One goal of Chapter V was to analyze the molecular correlates of the appearance of color patterns in cichlid fishes. In this paper, gene expression analyses of known pigmentation genes were performed in melanic bars and the regions between bars (interbars) of Lake Victoria cichlid Haplochromis latifasciatus to give complementary information and mechanistic insights on how the spatial variation of chromatophore properties and densities (see discussion below) is achieved. Increased expression of asip1 in interbar regions may explain the difference in melanosome dispersal between bars and interbars, as one response of decreased melanocortin signaling is increased aggregation of melanosomes (Cal, Suarez-Bregua, et al. 2017). Although to date data is not yet enough, it would still be important to consider asip1 might have unknown roles in anteroposterior patterning. The most pronounced result is significantly higher expression of melanin synthesis genes in bars, explaining the darker appearance of bar melanophores. This is in line with studies finding the link between skin darkness and melanin related

– 118 –

⁕ General Discussion ⁕ genes expression (Hidehito et al. 1994, Camp and Lardelli 2001, Lamason et al. 2005, Schonthaler et al. 2005, Braasch, Liedtke, et al. 2009, Krauss et al. 2014).

The cellular correlates of cichlid color patterns

Color patterns of teleost fishes result from three-dimensional arrangements of chromatophores within the integument (Chavin 1969, Hirata et al. 2003). Yet, the macroscopically observable color patterns of cichlid fishes are still lacking detailed morphological description [but see (Baerends and Baerends-Van Roon 1950, Lanzing and Bower 1974, Lanzing 1975)]. Another main goal of this thesis (especially Chapter IV and V) is to understand how the color patterns of cichlid fishes are formed by differences in arrangement, number and shape of chromatophores. Chapter IV illustrates the coloration of cichlids at the cellular and ultrastructural levels. In line with studies of other species (Hirata et al. 2003, Kottler et al. 2014), dispersed melanophores could be found in all chromatophore layers from the dark- pigmented regions of both yellow and dark morphs of M. auratus. However, the intracellular melanosomes are varied in density, and this is compatible with the “darkness” of the melanophores and the macroscopic appearance in the integument. Another noticeable variation in those dark pigmented regions is that the chromatophores beneath melanophores are inconsistent in type and number across morphs and/or regions. These distinct three-dimensional distribution of chromatophores in the dark-pigmented regions somewhat demonstrate that the cell-cell interactions between melanophores, xanthophores and iridophores seem to have an incompatible dynamic even within M. auratus, not to mention when comparing to the model organism zebrafish and other taxa (Mahalwar et al. 2014, Patterson, Bain, and Parichy 2014, Singh, Schach, and Nusslein- Volhard 2014, Eom and Parichy 2017). Another remarkable finding of this manuscript is the physical mechanisms (mirror-like functionality and prism-like functionality) underlying iridescent coloration. It has been known that iridophore of teleost fishes containing stacks of periodically arranged light-reflecting platelets (iridosomes) can cause multilayer optical interference phenomena (Sugimoto et al. 2005, Yoshioka et al. 2011, Gur et al. 2013, Gur et al. 2014, Salis et al. 2019). In the regions with iridescent blue coloration of M. auratus (midlateral stripe of dark morph), iridophores with regularly oriented, tilted iridosomes were found and considered establishing a microstructure that allows multilayer optical interference

– 119 –

⁕ General Discussion ⁕ as prisms (Hiroshi, Noriko, and Ryozo 1990, Yoshioka et al. 2011). In some scenarios, iridophores may also serve as a mirror and directly reflect light and thereby intensify the coloration given by the overlaying cells and structures (Gur et al. 2017). For instance, in the yellow belly region of yellow morph M. auratus, iridophores with thick layers of horizontally oriented iridosomes were found in the bottom layer of integument and though to behave mirror-like functionality. In addition, Chapter IV comprehensively describes the morphological color change of M. auratus in cellular and ultrastructural levels, revels pronounced changes in chromatophore number, intracellular dispersal of pigments, and organelle properties. Finally, the study presented in Chapter V focuses on the developmental trajectories and transcriptional codes (as above discussion) of cichlid color patterns. One central question of this paper is how embryonal and juvenile color patterns translate to adult patterns, and if the color patterns are already “pre-patterned” at the embryonal stages. Albeit several previous studies suggested that the diversity in color patterns of cichlid fishes is already visible in the embryonal stage (Baerends and Baerends-Van Roon 1950, Balon 1977, Kratochwil, Sefton, and Meyer 2015), evidence supporting pre- patterning or so called “direct” development (Woltering et al. 2018) in cichlid fishes is still needed. In this study, comparison between the juvenile and adult patterns of Haplochromis latifasciatus uncover that the juvenile patterns bear already most cellular and morphological characteristics of adult patterns: vertical bars have a higher density of melanophores and melanophores are darker in the melanic regions. By tracking the behaviors of individual cells during development, the increase in melanophore density is likely caused by an increased differentiation of melanoblasts within the bar regions, but not because of the migration of mature and visible melanophores. This is in contrast to the model organism zebrafish where early larval melanophores move into prospective adult stripe regions during a post-embryonic metamorphic phase (Parichy et al. 2009, Eom et al. 2015, Eom and Parichy 2017). Yet, whether the vertical bars of H. latifasciatus are pre-patterned already at the chromatophore progenitor level is not clear. Last, by investigating the distribution of chromatophores in different integument regions, this chapter summaries the cellular mechanisms underlying bar pattern formation in H. latifasciatus—dynamic control of melanophore density, melanin synthesis and melanosome dispersal, which can be also apply to other species (Hendrick et al. 2019, Prazdnikov and Shkil 2019) and providing a baseline for further investigation.

– 120 –

⁕ General Discussion ⁕

Conclusion

Taking advantages of their remarkable diversity in color phenotypes, this thesis used cichlid fishes to integrative investigate the molecular, genetic and developmental mechanisms underlying the formation and evolution of color patterns. In particular, the chapters of this thesis provided evidence of shared genomic basis underlying convergent color patterns. For example, the function of asip1 is strikingly conserved in dorsa-ventral countershading and stripe-interstripe patterning across vertebrates. Another evidence would be agrp2, a “stripe repressors” across the East African cichlid adaptive radiations and — more generally — the teleost radiations, but the underlying mutational and structural changes are different between taxa and may contribute to the diversification of color patterns. The studies of this thesis also illustrated how the ultrastructural and cellular correlates contribute to the final appearance of color patterns in cichlid fishes. Pronounced variation in the repertoires of pigment organelles and chromatophores were shown to be accounted for the differences in coloration and pigmentation patterns of cichlid fishes. A close inspection of color patterns development revealed how colors are generated at the cellular level. Comparative analyses on transcriptional level provided novel insights into the physiological and genetic basis underlying the appearance of color phenotypes and color dimorphism in cichlid fishes. In summary, this thesis improved our understanding of the mechanistic basis of color pattern formation in cichlid fishes.

Future and Outlook

This thesis was able to bridge disciplines and methodologies, serving as a foundation for future studies of color patterns in cichlid fishes. The only question now is, what can we do to make cichlid fishes an even better system for investigating the origin of diverse color patterns? Emerging genetic, genomic and imaging technologies (Van Dijk et al. 2018, Tadokoro, Shikaya, and Takahashi 2019) have become feasible at the individual level even in non-model organisms. So, the future will bring about great advances and novel insights into genetic and cellular engines that generate the color patterns. For the genetic basis of different coloration and pigmentation patterns, novel QTL (quantitative trait locus) mappings are underway (Gerwin et al., in preparation). Stripes, bars, spots and meristic variations in these patterns, as well as sexual dimorphisms in

– 121 –

⁕ General Discussion ⁕ color patterns in cichlid fishes (Seehausen, Mayhew, and Van Alphen 2001) can all be the next analyzed traits. Coding/structural mutations detected by QTL mapping can be confirmed using functional test e.g. in vivo morpholinos, transgenesis, and CRISPR-Cas9 genome editing (Kratochwil and Meyer 2015a, Juntti 2019) to compare the differences in chromatophore structure, quantity or location. Regulatory mutations that alter the causal gene expression can be verified the activity profile by chromatin- immunoprecipitation using the histone mark (e.g. H3K27ac and H3K4me3) (Kratochwil and Meyer 2015c). Further stable transgenic line with reporter genes (GFP, tdTomato, mCherry) will allow us to visualize the evolutionary tinkering that alters expression level, patterns and timing (Jacob 1977, Abzhanov et al. 2004, Arthur 2004, Abzhanov et al. 2006, Guerreiro et al. 2013, Kratochwil et al. 2017). For mechanisms of color pattern formation, comparative analyses using histology and microscopy, in juveniles and adults, in all kind of color and pigmentation patterns, are needed to identify the cellular and histological underpinnings. It is now yet clear that how different color patterns are composed through the three-dimensional arrangement of chromatophores and what chromatophores and chromatophore classes generate different patterns color patterns. Screening the ontological expression of chromatophore markers using in situ hybridization and immunohistochemistry can helpful to better molecular- profile the chromatophore cell-types (Woltering et al. 2018). Chemical and ultrastructural analyses that revealing the cell-type-specific chemical compositions and organelle configurations can allow identify novel cell type (Lewis et al. 2019). Using single-cell RNA sequencing, one of the most powerful available techniques today, will decode the transcriptional codes for the developmental trajectories of color pattern formation (Lewis et al. 2019, Saunders et al. 2019, Gur et al. 2020). Animal color pattern is a highly relevant trait for understanding key issues in evolutionary and developmental biology. This thesis — as I hope — firms a firm basis for future research into the evo-devo of color pattern diversity in cichlid fishes. But from a wider perspective, this is only just beginning to more precisely understand how genetic changes that lead to the spectacular diversity of cichlid fishes.

– 122 –

⁕ References ⁕

References

Abzhanov, A., W. P. Kuo, C. Hartmann, B. R. Grant, P. R. Grant, and C. J. Tabin. 2006. "The calmodulin pathway and evolution of elongated beak morphology in Darwin's finches." Nature 442 (7102):563-7. doi: 10.1038/nature04843. Abzhanov, A., M. Protas, B. R. Grant, P. R. Grant, and C. J. Tabin. 2004. "Bmp4 and morphological variation of beaks in Darwin's finches." Science 305 (5689):1462- 5. doi: 10.1126/science.1098095. Ahi, E. P., L. A. Lecaudey, A. Ziegelbecker, O. Steiner, R. Glabonjat, W. Goessler, V. Hois, C. Wagner, A. Lass, and K. M. Sefc. 2020. "Comparative transcriptomics reveals candidate carotenoid color genes in an East African cichlid fish." BMC Genomics 21 (1):54. doi: 10.1186/s12864-020-6473-8. Ahi, E. P., and K. M. Sefc. 2017. "Anterior-posterior gene expression differences in three Lake Malawi cichlid fishes with variation in body stripe orientation." PeerJ 5:e4080. doi: 10.7717/peerj.4080. Albertson, R. C., K. E. Powder, Y. Hu, K. P. Coyle, R. B. Roberts, and K. J. Parsons. 2014. "Genetic basis of continuous variation in the levels and modular inheritance of pigmentation in cichlid fishes." Mol Ecol 23 (21):5135-50. doi: 10.1111/mec.12900. Alfonso, J., M. E. Fernandez, B. Cooper, G. Flugge, and A. C. Frasch. 2005. "The stress- regulated protein M6a is a key modulator for neurite outgrowth and filopodium/spine formation." Proc Natl Acad Sci U S A 102 (47):17196-201. doi: 10.1073/pnas.0504262102. Andersson, L., G. Plastow, M. Rothschild, and A. Ruvinsky. 2011. "Molecular genetics of coat colour variation." The genetics of the pig 38. Anzalone, A. V., P. B. Randolph, J. R. Davis, A. A. Sousa, L. W. Koblan, J. M. Levy, P. J. Chen, C. Wilson, G. A. Newby, A. Raguram, and D. R. Liu. 2019. "Search- and-replace genome editing without double-strand breaks or donor DNA." Nature 576 (7785):149-157. doi: 10.1038/s41586-019-1711-4. Arregi, I., M. Climent, D. Iliev, J. Strasser, N. Gouignard, J. K. Johansson, T. Singh, M. Mazur, H. Semb, I. Artner, L. Minichiello, and E. M. Pera. 2016. "Retinol Dehydrogenase-10 Regulates Pancreas Organogenesis and Endocrine Cell Differentiation via Paracrine Retinoic Acid Signaling." Endocrinology 157 (12):4615-4631. doi: 10.1210/en.2016-1745. Arthur, W. 2004. Biased embryos and evolution: Cambridge University Press. Bademosi, A. T., E. Lauwers, P. Padmanabhan, L. Odierna, Y. J. Chai, A. Papadopulos, G. J. Goodhill, P. Verstreken, B. van Swinderen, and F. A. Meunier. 2017. "In vivo single-molecule imaging of syntaxin1A reveals polyphosphoinositide- and activity-dependent trapping in presynaptic nanoclusters." Nat Commun 8 (1):13660. doi: 10.1038/ncomms13660. Baerends, G. P., and J. M. Baerends-van Roon. 1950. "An introduction to the study of the ethology of the cichlid fishes." Behaviour. Supplement:III-243. Bagati, A., A. Bianchi-Smiraglia, S. Moparthy, K. Kolesnikova, E. E. Fink, M. Kolesnikova, M. V. Roll, P. Jowdy, D. W. Wolff, A. Polechetti, D. H. Yun, B. C. Lipchick, L. M. Paul, B. Wrazen, K. Moparthy, S. Mudambi, G. E. Morozevich, S. G. Georgieva, J. Wang, G. Shafirstein, S. Liu, E. S. Kandel, A. E. Berman, N. F. Box, G. Paragh, and M. A. Nikiforov. 2018. "FOXQ1 controls the induced

– 123 –

⁕ References ⁕

differentiation of melanocytic cells." Cell Death Differ 25 (6):1040-1049. doi: 10.1038/s41418-018-0066-y. Bagnara, J. T. 1966. "Cytology and cytophysiology of non-melanophore pigment cells." In International review of cytology, 173-205. Elsevier. Bai, Q., J. A. Garver, N. A. Hukriede, and E. A. Burton. 2007. "Generation of a transgenic zebrafish model of Tauopathy using a novel promoter element derived from the zebrafish eno2 gene." Nucleic Acids Res 35 (19):6501-16. doi: 10.1093/nar/gkm608. Bai, Q., X. Wei, and E. A. Burton. 2009. "Expression of a 12-kb promoter element derived from the zebrafish enolase-2 gene in the zebrafish visual system." Neurosci Lett 449 (3):252-7. doi: 10.1016/j.neulet.2008.10.101. Balon, Eugene K. 1977. "Early ontogeny of Labeotropbeus Ahl, 1927 (Mbuna, Cichlidae, Lake Malawi), with a discussion on advanced protective styles in fish reproduction and development." Environmental Biology of Fishes. Barford, K., C. C. Yap, N. D. Dwyer, and B. Winckler. 2017. "The related neuronal endosomal proteins NEEP21 (Nsg1) and P19 (Nsg2) have divergent expression profiles in vivo." J Comp Neurol 525 (8):1861-1878. doi: 10.1002/cne.24168. Barlow, G. W. 1976. "The midas cichlid in Nicaragua." In Investigations of the ichthyofauna of Nicaraguan lakes, edited by T. B. Thorson, 333-358. Lincoln, NE: University of Nebraska-Lincoln. Barlow, George W. 1972. "The Attitude of Fish Eye-Lines in Relation to Body Shape and to Stripes and Bars." Copeia 1972 (1):4-12. doi: 10.2307/1442777. Batista, A. F. R., J. C. Martinez, and U. Hengst. 2017. "Intra-axonal Synthesis of SNAP25 Is Required for the Formation of Presynaptic Terminals." Cell Rep 20 (13):3085-3098. doi: 10.1016/j.celrep.2017.08.097. Béjar, J., Y. Hong, and M. Schartl. 2003. "Mitf expression is sufficient to direct differentiation of medaka blastula derived stem cells to melanocytes." Development 130 (26):6545-6553. Beltrán, M., C. D. Jiggins, V. Bull, M. Linares, J. Mallet, W. O. McMillan, and E. Bermingham. 2002. "Phylogenetic discordance at the species boundary: comparative gene genealogies among rapidly radiating Heliconius butterflies." Molecular biology and evolution 19 (12):2176-2190. Blackmon, H, and RA Adams. 2015. EvobiR: tools for comparative analyses and teaching evolutionary biology. Crossref. Bolger, A. M., M. Lohse, and B. Usadel. 2014. "Trimmomatic: a flexible trimmer for Illumina sequence data." Bioinformatics 30 (15):2114-20. doi: 10.1093/bioinformatics/btu170. Bouckaert, R., J. Heled, D. Kuhnert, T. Vaughan, C. H. Wu, D. Xie, M. A. Suchard, A. Rambaut, and A. J. Drummond. 2014. "BEAST 2: a software platform for Bayesian evolutionary analysis." PLoS Comput Biol 10 (4):e1003537. doi: 10.1371/journal.pcbi.1003537. Bouckaert, R. R. 2010. "DensiTree: making sense of sets of phylogenetic trees." Bioinformatics 26 (10):1372-3. doi: 10.1093/bioinformatics/btq110. Braasch, I., F. Brunet, J. N. Volff, and M. Schartl. 2009. "Pigmentation pathway evolution after whole-genome duplication in fish." Genome Biol Evol 1:479-93. doi: 10.1093/gbe/evp050. Braasch, I., D. Liedtke, J. N. Volff, and M. Schartl. 2009. "Pigmentary function and evolution of tyrp1 gene duplicates in fish." Pigment Cell Melanoma Res 22 (6):839-50. doi: 10.1111/j.1755-148X.2009.00614.x.

– 124 –

⁕ References ⁕

Braasch, I., and J. H. Postlethwait. 2011. "The teleost agouti-related protein 2 gene is an ohnolog gone missing from the tetrapod genome." Proc Natl Acad Sci U S A 108 (13):E47-8. doi: 10.1073/pnas.1101594108. Braasch, I., W. Salzburger, and A. Meyer. 2006. "Asymmetric evolution in two fish- specifically duplicated receptor tyrosine kinase paralogons involved in teleost coloration." Mol Biol Evol 23 (6):1192-202. doi: 10.1093/molbev/msk003. Brawand, D., C. E. Wagner, Y. I. Li, M. Malinsky, I. Keller, S. Fan, O. Simakov, A. Y. Ng, Z. W. Lim, E. Bezault, J. Turner-Maier, J. Johnson, R. Alcazar, H. J. Noh, P. Russell, B. Aken, J. Alfoldi, C. Amemiya, N. Azzouzi, J. F. Baroiller, F. Barloy- Hubler, A. Berlin, R. Bloomquist, K. L. Carleton, M. A. Conte, H. D'Cotta, O. Eshel, L. Gaffney, F. Galibert, H. F. Gante, S. Gnerre, L. Greuter, R. Guyon, N. S. Haddad, W. Haerty, R. M. Harris, H. A. Hofmann, T. Hourlier, G. Hulata, D. B. Jaffe, M. Lara, A. P. Lee, I. MacCallum, S. Mwaiko, M. Nikaido, H. Nishihara, C. Ozouf-Costaz, D. J. Penman, D. Przybylski, M. Rakotomanga, S. C. P. Renn, F. J. Ribeiro, M. Ron, W. Salzburger, L. Sanchez-Pulido, M. E. Santos, S. Searle, T. Sharpe, R. Swofford, F. J. Tan, L. Williams, S. Young, S. Yin, N. Okada, T. D. Kocher, E. A. Miska, E. S. Lander, B. Venkatesh, R. D. Fernald, A. Meyer, C. P. Ponting, J. T. Streelman, K. Lindblad-Toh, O. Seehausen, and F. Di Palma. 2014. "The genomic substrate for adaptive radiation in African cichlid fish." Nature 513 (7518):375-381. doi: 10.1038/nature13726. Broman, K. W., H. Wu, S. Sen, and G. A. Churchill. 2003. "R/qtl: QTL mapping in experimental crosses." Bioinformatics 19 (7):889-90. doi: 10.1093/bioinformatics/btg112. Bronner, M. E., and N. M. LeDouarin. 2012. "Development and evolution of the neural crest: an overview." Dev Biol 366 (1):2-9. doi: 10.1016/j.ydbio.2011.12.042. Brudno, M., S. Malde, A. Poliakov, C. B. Do, O. Couronne, I. Dubchak, and S. Batzoglou. 2003. "Glocal alignment: finding rearrangements during alignment." Bioinformatics 19 Suppl 1 (suppl_1):i54-62. doi: 10.1093/bioinformatics/btg1005. Brzozowski, F., J. Roscoe, K. Parsons, and C. Albertson. 2012. "Sexually dimorphic levels of color trait integration and the resolution of sexual conflict in Lake Malawi cichlids." J Exp Zool B Mol Dev Evol 318 (4):268-78. doi: 10.1002/jez.b.22443. Burton, D., and M. Burton. 2017. Essential Fish Biology: Diversity, Structure, and Function: Oxford University Press. Cal, L., M. MegIas, J. M. Cerda-Reverter, J. H. Postlethwait, I. Braasch, and J. Rotllant. 2017. "BAC Recombineering of the Agouti Loci from Spotted Gar and Zebrafish Reveals the Evolutionary Ancestry of Dorsal-Ventral Pigment Asymmetry in Fish." J Exp Zool B Mol Dev Evol 328 (7):697-708. doi: 10.1002/jez.b.22748. Cal, L., P. Suarez-Bregua, J. M. Cerda-Reverter, I. Braasch, and J. Rotllant. 2017. "Fish pigmentation and the melanocortin system." Comp Biochem Physiol A Mol Integr Physiol 211:26-33. doi: 10.1016/j.cbpa.2017.06.001. Cal, L., P. Suarez-Bregua, P. Comesana, J. Owen, I. Braasch, R. Kelsh, J. M. Cerda- Reverter, and J. Rotllant. 2019. "Countershading in zebrafish results from an Asip1 controlled dorsoventral gradient of pigment cell differentiation." Sci Rep 9 (1):3449. doi: 10.1038/s41598-019-40251-z. Camp, E., and M. Lardelli. 2001. "Tyrosinase gene expression in zebrafish embryos." Dev Genes Evol 211 (3):150-3. doi: 10.1007/s004270000125. Candille, S. I., C. B. Kaelin, B. M. Cattanach, B. Yu, D. A. Thompson, M. A. Nix, J. A. Kerns, S. M. Schmutz, G. L. Millhauser, and G. S. Barsh. 2007. "A -defensin – 125 –

⁕ References ⁕

mutation causes black coat color in domestic dogs." Science 318 (5855):1418- 23. doi: 10.1126/science.1147880. Cardona, A., S. Saalfeld, J. Schindelin, I. Arganda-Carreras, S. Preibisch, M. Longair, P. Tomancak, V. Hartenstein, and R. J. Douglas. 2012. "TrakEM2 software for neural circuit reconstruction." PLoS One 7 (6):e38011. doi: 10.1371/journal.pone.0038011. Ceinos, R. M., R. Guillot, R. N. Kelsh, J. M. Cerda-Reverter, and J. Rotllant. 2015. "Pigment patterns in adult fish result from superimposition of two largely independent pigmentation mechanisms." Pigment Cell Melanoma Res 28 (2):196-209. doi: 10.1111/pcmr.12335. Cerda-Reverter, J. M., T. Haitina, H. B. Schioth, and R. E. Peter. 2005. "Gene structure of the goldfish agouti-signaling protein: a putative role in the dorsal-ventral pigment pattern of fish." Endocrinology 146 (3):1597-610. doi: 10.1210/en.2004-1346. Chakraborty, A. K., J. T. Platt, K. K. Kim, B. S. Kwon, D. C. Bennett, and J. M. Pawelek. 1996. "Polymerization of 5,6-dihydroxyindole-2-carboxylic acid to melanin by the pmel 17/silver locus protein." Eur J Biochem 236 (1):180-8. doi: 10.1111/j.1432-1033.1996.t01-1-00180.x. Chander, P., M. J. Kennedy, B. Winckler, and J. P. Weick. 2019. "Neuron-Specific Gene 2 (NSG2) Encodes an AMPA Receptor Interacting Protein That Modulates Excitatory Neurotransmission." eNeuro 6 (1). doi: 10.1523/ENEURO.0292- 18.2018. Chavin, W. 1969. "Fundamental aspects of morphological melanin color changes in vertebrate skin." Am Zool 9 (2):505-20. doi: 10.1093/icb/9.2.505. Chen, N. H., M. E. Reith, and M. W. Quick. 2004. "Synaptic uptake and beyond: the sodium- and chloride-dependent neurotransmitter transporter family SLC6." Pflugers Arch 447 (5):519-31. doi: 10.1007/s00424-003-1064-5. Chen, S. C., M. A. Hornsby, R. M. Robertson, and C. W. Hawryshyn. 2014. "The influence of chromatic background on the photosensitivity of tilapia erythrophores." Biol Open 3 (2):117-20. doi: 10.1242/bio.20146742. Chen, S. C., C. Xiao, N. F. Troje, R. M. Robertson, and C. W. Hawryshyn. 2015. "Functional characterisation of the chromatically antagonistic photosensitive mechanism of erythrophores in the tilapia Oreochromis niloticus." J Exp Biol 218 (Pt 5):748-56. doi: 10.1242/jeb.106831. Cheney, K. L., A. S. Grutter, and N. J. Marshall. 2008. "Facultative mimicry: cues for colour change and colour accuracy in a coral reef fish." Proc Biol Sci 275 (1631):117-22. doi: 10.1098/rspb.2007.0966. Chettier, R., L. Nelson, J. W. Ogilvie, H. M. Albertsen, and K. Ward. 2015. "Haplotypes at LBX1 have distinct inheritance patterns with opposite effects in adolescent idiopathic scoliosis." PLoS One 10 (2):e0117708. doi: 10.1371/journal.pone.0117708. Chitnis, A. B., and J. Y. Kuwada. 1990. "Axonogenesis in the brain of zebrafish embryos." J Neurosci 10 (6):1892-905. Conte, M. A., W. J. Gammerdinger, K. L. Bartie, D. J. Penman, and T. D. Kocher. 2017. "A high quality assembly of the Nile Tilapia (Oreochromis niloticus) genome reveals the structure of two sex determination regions." BMC Genomics 18 (1):341. doi: 10.1186/s12864-017-3723-5. Conte, M. A., R. Joshi, E. C. Moore, S. P. Nandamuri, W. J. Gammerdinger, R. B. Roberts, K. L. Carleton, S. Lien, and T. D. Kocher. 2019. "Chromosome-scale

– 126 –

⁕ References ⁕

assemblies reveal the structural evolution of African cichlid genomes." Gigascience 8 (4):giz030. doi: 10.1093/gigascience/giz030. Cott, H. B. 1940. Adaptive coloration in animals, Adaptive coloration in animals. Oxford, England: Oxford. Cunningham, F., P. Achuthan, W. Akanni, J. Allen, M. R. Amode, I. M. Armean, R. Bennett, J. Bhai, K. Billis, S. Boddu, C. Cummins, C. Davidson, K. J. Dodiya, A. Gall, C. G. Giron, L. Gil, T. Grego, L. Haggerty, E. Haskell, T. Hourlier, O. G. Izuogu, S. H. Janacek, T. Juettemann, M. Kay, M. R. Laird, I. Lavidas, Z. Liu, J. E. Loveland, J. C. Marugan, T. Maurel, A. C. McMahon, B. Moore, J. Morales, J. M. Mudge, M. Nuhn, D. Ogeh, A. Parker, A. Parton, M. Patricio, A. I. Abdul Salam, B. M. Schmitt, H. Schuilenburg, D. Sheppard, H. Sparrow, E. Stapleton, M. Szuba, K. Taylor, G. Threadgold, A. Thormann, A. Vullo, B. Walts, A. Winterbottom, A. Zadissa, M. Chakiachvili, A. Frankish, S. E. Hunt, M. Kostadima, N. Langridge, F. J. Martin, M. Muffato, E. Perry, M. Ruffier, D. M. Staines, S. J. Trevanion, B. L. Aken, A. D. Yates, D. R. Zerbino, and P. Flicek. 2019. "Ensembl 2019." Nucleic Acids Res 47 (D1):D745-D751. doi: 10.1093/nar/gky1113. Darwin, C. 1859. On the Origin of Species by Means of Natural Selection Or the Preservation of Favoured Races in the Struggle for Life: H. Milford; Oxford University Press. Darwin, C. 1888. The descent of man: and selection in relation to sex: John Murray, Albemarle Street. Davis, K. N., R. Tao, C. Li, Y. Gao, M. C. Gondre-Lewis, B. K. Lipska, J. H. Shin, B. Xie, T. Ye, D. R. Weinberger, J. E. Kleinman, and T. M. Hyde. 2016. "GAD2 Alternative Transcripts in the Human Prefrontal Cortex, and in Schizophrenia and Affective Disorders." PLoS One 11 (2):e0148558. doi: 10.1371/journal.pone.0148558. De Vries, H. 1904. Species and Varieties, Their Origin by Mutation: Lectures Delivered at the University of California: Open court publishing Company. Denoeud, F., L. Carretero-Paulet, A. Dereeper, G. Droc, R. Guyot, M. Pietrella, C. Zheng, A. Alberti, F. Anthony, G. Aprea, J. M. Aury, P. Bento, M. Bernard, S. Bocs, C. Campa, A. Cenci, M. C. Combes, D. Crouzillat, C. Da Silva, L. Daddiego, F. De Bellis, S. Dussert, O. Garsmeur, T. Gayraud, V. Guignon, K. Jahn, V. Jamilloux, T. Joet, K. Labadie, T. Lan, J. Leclercq, M. Lepelley, T. Leroy, L. T. Li, P. Librado, L. Lopez, A. Munoz, B. Noel, A. Pallavicini, G. Perrotta, V. Poncet, D. Pot, Priyono, M. Rigoreau, M. Rouard, J. Rozas, C. Tranchant-Dubreuil, R. VanBuren, Q. Zhang, A. C. Andrade, X. Argout, B. Bertrand, A. de Kochko, G. Graziosi, R. J. Henry, Jayarama, R. Ming, C. Nagai, S. Rounsley, D. Sankoff, G. Giuliano, V. A. Albert, P. Wincker, and P. Lashermes. 2014. "The coffee genome provides insight into the convergent evolution of caffeine biosynthesis." Science 345 (6201):1181-4. doi: 10.1126/science.1255274. Di Meglio, T., C. F. Kratochwil, N. Vilain, A. Loche, A. Vitobello, K. Yonehara, S. M. Hrycaj, B. Roska, A. H. Peters, A. Eichmann, D. Wellik, S. Ducret, and F. M. Rijli. 2013. "Ezh2 orchestrates topographic migration and connectivity of mouse precerebellar neurons." Science 339 (6116):204-7. doi: 10.1126/science.1229326. Dickman, M. C., M. Schliwa, and G. W. Barlow. 1988. "Melanophore death and disappearance produces color metamorphosis in the polychromatic Midas cichlid (Cichlasoma citrinellum)." Cell Tissue Res 253 (1):9-14. doi: 10.1007/bf00221733.

– 127 –

⁕ References ⁕

Dijkstra, P. D., S. M. Maguire, R. M. Harris, A. A. Rodriguez, R. S. DeAngelis, S. A. Flores, and H. A. Hofmann. 2017. "The melanocortin system regulates body pigmentation and social behaviour in a colour polymorphic cichlid fish." Proc Biol Sci 284 (1851). doi: 10.1098/rspb.2016.2838. Dobin, A., C. A. Davis, F. Schlesinger, J. Drenkow, C. Zaleski, S. Jha, P. Batut, M. Chaisson, and T. R. Gingeras. 2013. "STAR: ultrafast universal RNA-seq aligner." Bioinformatics 29 (1):15-21. doi: 10.1093/bioinformatics/bts635. Doganli, C., H. C. Beck, A. B. Ribera, C. Oxvig, and K. Lykke-Hartmann. 2013. "alpha3Na+/K+-ATPase deficiency causes brain ventricle dilation and abrupt embryonic motility in zebrafish." J Biol Chem 288 (13):8862-74. doi: 10.1074/jbc.M112.421529. Dooley, C. M., A. Mongera, B. Walderich, and C. Nusslein-Volhard. 2013. "On the embryonic origin of adult melanophores: the role of ErbB and Kit signalling in establishing melanophore stem cells in zebrafish." Development 140 (5):1003- 13. doi: 10.1242/dev.087007. Doucet, S. M., M. D. Shawkey, G. E. Hill, and R. Montgomerie. 2006. "Iridescent plumage in satin bowerbirds: structure, mechanisms and nanostructural predictors of individual variation in colour." J Exp Biol 209 (Pt 2):380-90. doi: 10.1242/jeb.01988. Du, J., A. J. Miller, H. R. Widlund, M. A. Horstmann, S. Ramaswamy, and D. E. Fisher. 2003. "MLANA/MART1 and SILV/PMEL17/GP100 are transcriptionally regulated by MITF in melanocytes and melanoma." Am J Pathol 163 (1):333-43. doi: 10.1016/S0002-9440(10)63657-7. Durinck, S., P. T. Spellman, E. Birney, and W. Huber. 2009. "Mapping identifiers for the integration of genomic datasets with the R/Bioconductor package biomaRt." Nat Protoc 4 (8):1184-91. doi: 10.1038/nprot.2009.97. Dutton, K. A., A. Pauliny, S. S. Lopes, S. Elworthy, T. J. Carney, J. Rauch, R. Geisler, P. Haffter, and R. N. Kelsh. 2001. "Zebrafish colourless encodes sox10 and specifies non-ectomesenchymal neural crest fates." Development 128 (21):4113- 25. Eizirik, E., N. Yuhki, W. E. Johnson, M. Menotti-Raymond, S. S. Hannah, and S. J. O'Brien. 2003. "Molecular genetics and evolution of melanism in the cat family." Curr Biol 13 (5):448-53. doi: 10.1016/s0960-9822(03)00128-3. Elmer, K. R., T. K. Lehtonen, and A. Meyer. 2009. "Color assortative mating contributes to sympatric divergence of neotropical cichlid fish." Evolution 63 (10):2750-7. doi: 10.1111/j.1558-5646.2009.00736.x. Elmer, K. R., C. Reggio, T. Wirth, E. Verheyen, W. Salzburger, and A. Meyer. 2009. "Pleistocene desiccation in East Africa bottlenecked but did not extirpate the adaptive radiation of Lake Victoria haplochromine cichlid fishes." Proc Natl Acad Sci U S A 106 (32):13404-13409. Elmer, K. R., L. R. Thompson, and A. Meyer. 2009. "Similar levels of diversity and population Structure in superflock and non-superflock cichlid fishes from Lake Victoria, Africa." Tropical Freshwater Biology 18 (2):1. Elworthy, S., J. A. Lister, T. J. Carney, D. W. Raible, and R. N. Kelsh. 2003. "Transcriptional regulation of mitfa accounts for the sox10 requirement in zebrafish melanophore development." Development 130 (12):2809-18. doi: 10.1242/dev.00461. Eom, D. S., E. J. Bain, L. B. Patterson, M. E. Grout, and D. M. Parichy. 2015. "Long- distance communication by specialized cellular projections during pigment pattern development and evolution." Elife 4. doi: 10.7554/eLife.12401. – 128 –

⁕ References ⁕

Eom, D. S., and D. M. Parichy. 2017. "A macrophage relay for long-distance signaling during postembryonic tissue remodeling." Science 355 (6331):1317-1320. doi: 10.1126/science.aal2745. Ewels, P., M. Magnusson, S. Lundin, and M. Kaller. 2016. "MultiQC: summarize analysis results for multiple tools and samples in a single report." Bioinformatics 32 (19):3047-8. doi: 10.1093/bioinformatics/btw354. Fan, S., and A. Meyer. 2014. "Evolution of genomic structural variation and genomic architecture in the adaptive radiations of African cichlid fishes." Front Genet 5:163. doi: 10.3389/fgene.2014.00163. Fernald, R. D. 2012. "Social control of the brain." Annu Rev Neurosci 35:133-51. doi: 10.1146/annurev-neuro-062111-150520. Fisher, S., E. A. Grice, R. M. Vinton, S. L. Bessling, A. Urasaki, K. Kawakami, and A. S. McCallion. 2006. "Evaluating the biological relevance of putative enhancers using Tol2 transposon-mediated transgenesis in zebrafish." Nat Protoc 1 (3):1297-305. doi: 10.1038/nprot.2006.230. Fontenas, L., F. De Santis, V. Di Donato, C. Degerny, B. Chambraud, F. Del Bene, and M. Tawk. 2016. "Neuronal Ndrg4 Is Essential for Nodes of Ranvier Organization in Zebrafish." PLoS Genet 12 (11):e1006459. doi: 10.1371/journal.pgen.1006459. Franchini, P., P. Xiong, C. Fruciano, R. F. Schneider, J. M. Woltering, C. D. Hulsey, and A. Meyer. 2019. "MicroRNA gene regulation in extremely young and parallel adaptive radiations of crater lake cichlid fish." Mol Biol Evol 36 (11):2498-2511. doi: 10.1093/molbev/msz168. Fremeau, R. T., Jr., K. Kam, T. Qureshi, J. Johnson, D. R. Copenhagen, J. Storm- Mathisen, F. A. Chaudhry, R. A. Nicoll, and R. H. Edwards. 2004. "Vesicular glutamate transporters 1 and 2 target to functionally distinct synaptic release sites." Science 304 (5678):1815-9. doi: 10.1126/science.1097468. Fricke, R. 2019. Eschmeyer’s catalog of fishes: Genera, species, references (2019). Fryer, G. 1959. "The trophic interrelationships and ecology of some littoral communities of Lake Nyasa with especial reference to the fishes, and a discussion of the evolution of a group of rock ‐ frequenting Cichlidae." Proceedings of the Zoological Society of London. Fryer, G., and T. D. Iles. 1972. Cichlid fishes of the great lakes of Africa. Fujii, R. 1993. "Cytophysiology of Fish Chromatophores." In International Review of Cytology, edited by K. W. Jeon, M. Friedlander and J. Jarvik, 191-255. Academic Press. Fujii, R. 2000. "The regulation of motile activity in fish chromatophores." Pigment Cell Res 13 (5):300-19. doi: 10.1034/j.1600-0749.2000.130502.x. Fukamachi, S., S. Asakawa, Y. Wakamatsu, N. Shimizu, H. Mitani, and A. Shima. 2004. "Conserved function of medaka pink-eyed dilution in melanin synthesis and its divergent transcriptional regulation in gonads among vertebrates." Genetics 168 (3):1519-27. doi: 10.1534/genetics.104.030494. Gallaud, E., R. Caous, A. Pascal, F. Bazile, J. P. Gagne, S. Huet, G. G. Poirier, D. Chretien, L. Richard-Parpaillon, and R. Giet. 2014. "Ensconsin/Map7 promotes microtubule growth and centrosome separation in Drosophila neural stem cells." J Cell Biol 204 (7):1111-21. doi: 10.1083/jcb.201311094. Gao, M., D. Zhao, S. Schouteden, M. G. Sorci-Thomas, P. P. Van Veldhoven, K. Eggermont, G. Liu, C. M. Verfaillie, and Y. Feng. 2014. "Regulation of high- density lipoprotein on hematopoietic stem/progenitor cells in atherosclerosis

– 129 –

⁕ References ⁕

requires scavenger receptor type BI expression." Arterioscler Thromb Vasc Biol 34 (9):1900-9. doi: 10.1161/ATVBAHA.114.304006. Genner, M. J., O. Seehausen, D. H. Lunt, D. A. Joyce, P. W. Shaw, G. R. Carvalho, and G. F. Turner. 2007. "Age of cichlids: new dates for ancient lake fish radiations." Mol Biol Evol 24 (5):1269-82. doi: 10.1093/molbev/msm050. Ginger, R. S., S. E. Askew, R. M. Ogborne, S. Wilson, D. Ferdinando, T. Dadd, A. M. Smith, S. Kazi, R. T. Szerencsei, R. J. Winkfein, P. P. Schnetkamp, and M. R. Green. 2008. "SLC24A5 encodes a trans-Golgi network protein with potassium- dependent sodium-calcium exchange activity that regulates human epidermal melanogenesis." J Biol Chem 283 (9):5486-95. doi: 10.1074/jbc.M707521200. Giuffra, E., A. Tornsten, S. Marklund, E. Bongcam-Rudloff, P. Chardon, J. M. Kijas, S. I. Anderson, A. L. Archibald, and L. Andersson. 2002. "A large duplication associated with dominant white color in pigs originated by homologous recombination between LINE elements flanking KIT." Mamm Genome 13 (10):569-77. doi: 10.1007/s00335-002-2184-5. Gould, S. J. 1990. Wonderful life: the Burgess Shale and the nature of history: WW Norton & Company. Granneman, J. G., V. A. Kimler, H. Zhang, X. Ye, X. Luo, J. H. Postlethwait, and R. Thummel. 2017. "Lipid droplet biology and evolution illuminated by the characterization of a novel perilipin in teleost fish." Elife 6. doi: 10.7554/eLife.21771. Greenwood, P. H. 1974. "Cichlid fishes of Lake Victoria, East Africa: the biology and evolution of a species flock." Bull. British Mus. Nat. Hist.(Zool.) Suppl. 6:1-134. Gross, J. B., R. Borowsky, and C. J. Tabin. 2009. "A novel role for Mc1r in the parallel evolution of depigmentation in independent populations of the cavefish Astyanax mexicanus." PLoS Genet 5 (1):e1000326. doi: 10.1371/journal.pgen.1000326. Guerreiro, I., A. Nunes, J. M. Woltering, A. Casaca, A. Novoa, T. Vinagre, M. E. Hunter, D. Duboule, and M. Mallo. 2013. "Role of a polymorphism in a Hox/Pax- responsive enhancer in the evolution of the vertebrate spine." Proc Natl Acad Sci U S A 110 (26):10682-6. doi: 10.1073/pnas.1300592110. Guillot, R., R. M. Ceinos, R. Cal, J. Rotllant, and J. M. Cerda-Reverter. 2012. "Transient ectopic overexpression of agouti-signalling protein 1 (asip1) induces pigment anomalies in flatfish." PLoS One 7 (12):e48526. doi: 10.1371/journal.pone.0048526. Guindon, S., J. F. Dufayard, V. Lefort, M. Anisimova, W. Hordijk, and O. Gascuel. 2010. "New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0." Syst Biol 59 (3):307-21. doi: 10.1093/sysbio/syq010. Gundersen, R. E., and E. R. Rivera. 1982. "An ultrastructural study of the development of the dermal iridophores and structural pigmentation in Poecilia reticulata (peters)." J Morphol 172 (3):349-359. doi: 10.1002/jmor.1051720309. Gunter, H. M., C. Clabaut, W. Salzburger, and A. Meyer. 2011. "Identification and characterization of gene expression involved in the coloration of cichlid fish using microarray and qRT-PCR approaches." Journal of molecular evolution 72 (2):127-137. Gur, D., E. J. Bain, K Johnson, A. J. Aman, A. Pasoili, J. D. Flynn, M. C Allen, D. D. Deheyn, J. C. Lee, J. Lippincott-Schwartz, and D. M. Parichy. 2020. "In situ differentiation of iridophore crystallotypes underlies zebrafish stripe patterning." bioRxiv. doi: 10.1101/2020.03.25.008664.

– 130 –

⁕ References ⁕

Gur, D., B. Leshem, D. Oron, S. Weiner, and L. Addadi. 2014. "The structural basis for enhanced silver reflectance in Koi fish scale and skin." J Am Chem Soc 136 (49):17236-42. doi: 10.1021/ja509340c. Gur, D., B. A. Palmer, S. Weiner, and L. Addadi. 2017. "Light Manipulation by Guanine Crystals in Organisms: Biogenic Scatterers, Mirrors, Multilayer Reflectors and Photonic Crystals." Advanced Functional Materials 27 (6). doi: 10.1002/adfm.201603514. Gur, D., Y. Politi, B. Sivan, P. Fratzl, S. Weiner, and L. Addadi. 2013. "Guanine-based photonic crystals in fish scales form from an amorphous precursor." Angew Chem Int Ed Engl 52 (1):388-91. doi: 10.1002/anie.201205336. Hara, M., M. Toyoda, M. Yaar, J. Bhawan, E. M. Avila, I. R. Penner, and B. A. Gilchrest. 1996. "Innervation of melanocytes in human skin." J Exp Med 184 (4):1385-95. doi: 10.1084/jem.184.4.1385. Hashimoto, R., H. Hashimoto, N. Shintani, S. Chiba, S. Hattori, T. Okada, M. Nakajima, K. Tanaka, N. Kawagishi, and K. Nemoto. 2007. "Pituitary adenylate cyclase- activating polypeptide is associated with schizophrenia." Molecular psychiatry 12 (11):1026-1032. Hastings, P. J., J. R. Lupski, S. M. Rosenberg, and G. Ira. 2009. "Mechanisms of change in gene copy number." Nat Rev Genet 10 (8):551-64. doi: 10.1038/nrg2593. Haupaix, N., C. Curantz, R. Bailleul, S. Beck, A. Robic, and M. Manceau. 2018. "The periodic coloration in birds forms through a prepattern of somite origin." Science 361 (6408). doi: 10.1126/science.aar4777. Hawkes, J. W. 1974. "The structure of fish skin. II. The chromatophore unit." Cell Tissue Res 149 (2):159-72. doi: 10.1007/bf00222271. Hemingson, C. R., P. F. Cowman, J. R. Hodge, and D. R. Bellwood. 2019. "Colour pattern divergence in reef fish species is rapid and driven by both range overlap and symmetry." Ecol Lett 22 (1):190-199. doi: 10.1111/ele.13180. Hendrick, L. A., G. A. Carter, E. H. Hilbrands, B. P. Heubel, T. F. Schilling, and P. Le Pabic. 2019. "Bar, stripe and spot development in sand-dwelling cichlids from Lake Malawi." Evodevo 10:18. doi: 10.1186/s13227-019-0132-7. Henning, F., J. C. Jones, P. Franchini, and A. Meyer. 2013. "Transcriptomics of morphological color change in polychromatic Midas cichlids." BMC Genomics 14:171. doi: 10.1186/1471-2164-14-171. Henning, F., H. J. Lee, P. Franchini, and A. Meyer. 2014. "Genetic mapping of horizontal stripes in Lake Victoria cichlid fishes: benefits and pitfalls of using RAD markers for dense linkage mapping." Mol Ecol 23 (21):5224-40. doi: 10.1111/mec.12860. Henning, F., G. Machado-Schiaffino, L. Baumgarten, and A. Meyer. 2017. "Genetic dissection of adaptive form and function in rapidly speciating cichlid fishes." Evolution 71 (5):1297-1312. doi: 10.1111/evo.13206. Henning, F., and A. Meyer. 2012. "Eggspot number and sexual selection in the cichlid fish Astatotilapia burtoni." PLoS One 7 (8):e43695. doi: 10.1371/journal.pone.0043695. Henning, F., A. J. Renz, S. Fukamachi, and A. Meyer. 2010. "Genetic, comparative genomic, and expression analyses of the Mc1r locus in the polychromatic Midas cichlid fish (Teleostei, Cichlidae Amphilophus sp.) species group." J Mol Evol 70 (5):405-12. doi: 10.1007/s00239-010-9340-4. Hidehito, I., B. Yoshitaka, K. Akihiko, and H. Hiroshi. 1994. "Expression of the tyrosinase-encoding gene in a colorless melanophore mutant of the medaka fish, Oryzias latipes." Gene 150 (2):319-324.

– 131 –

⁕ References ⁕

Higdon, C. W., R. D. Mitra, and S. L. Johnson. 2013. "Gene expression analysis of zebrafish melanocytes, iridophores, and retinal pigmented epithelium reveals indicators of biological function and developmental origin." PLoS One 8 (7):e67801. doi: 10.1371/journal.pone.0067801. Hill, G. E, and K. J. McGraw. 2006. Bird coloration: mechanisms and measurements. Vol. 1: Harvard University Press. Hirata, M., K. Nakamura, T. Kanemaru, Y. Shibata, and S. Kondo. 2003. "Pigment cell organization in the hypodermis of zebrafish." Dev Dyn 227 (4):497-503. doi: 10.1002/dvdy.10334. Hirata, M., K. Nakamura, and S. Kondo. 2005. "Pigment cell distributions in different tissues of the zebrafish, with special reference to the striped pigment pattern." Dev Dyn 234 (2):293-300. doi: 10.1002/dvdy.20513. Hiroshi, N., O. Noriko, and F. Ryozo. 1990. "Light-reflecting properties of the iridophores of the neon tetra, Paracheirodon innesi." Comparative Biochemistry and Physiology Part A: Physiology 95 (3):337-341. Hoekstra, H. E., and J. A. Coyne. 2007. "The locus of evolution: evo devo and the genetics of adaptation." Evolution 61 (5):995-1016. doi: 10.1111/j.1558- 5646.2007.00105.x. Hofmann, H. A., and R. D. Fernald. 2001. "What cichlids tell us about the social regulation of brain and behavior." J. Aquacult. Aquat. Sci 9:17-31. Hooper, D. M., S. C. Griffith, and T. D. Price. 2019. "Sex chromosome inversions enforce reproductive isolation across an avian hybrid zone." Mol Ecol 28 (6):1246-1262. doi: 10.1111/mec.14874. Hou, L., H. Arnheiter, and W. J. Pavan. 2006. "Interspecies difference in the regulation of melanocyte development by SOX10 and MITF." Proc Natl Acad Sci U S A 103 (24):9081-5. doi: 10.1073/pnas.0603114103. Huber, W., A. von Heydebreck, H. Sultmann, A. Poustka, and M. Vingron. 2002. "Variance stabilization applied to microarray data calibration and to the quantification of differential expression." Bioinformatics 18 Suppl 1 (suppl_1):S96-104. doi: 10.1093/bioinformatics/18.suppl_1.s96. Hughes, L. C., G. Orti, Y. Huang, Y. Sun, C. C. Baldwin, A. W. Thompson, D. Arcila, R. R. Betancur, C. Li, L. Becker, N. Bellora, X. Zhao, X. Li, M. Wang, C. Fang, B. Xie, Z. Zhou, H. Huang, S. Chen, B. Venkatesh, and Q. Shi. 2018. "Comprehensive phylogeny of ray-finned fishes (Actinopterygii) based on transcriptomic and genomic data." Proc Natl Acad Sci U S A 115 (24):6249-6254. doi: 10.1073/pnas.1719358115. Hulsey, C. D., J. Zheng, B. C. Faircloth, A. Meyer, and M. E. Alfaro. 2017. "Phylogenomic analysis of Lake Malawi cichlid fishes: Further evidence that the three-stage model of diversification does not fit." Mol Phylogenet Evol 114:40- 48. doi: 10.1016/j.ympev.2017.05.027. Hultman, K. A., N. Bahary, L. I. Zon, and S. L. Johnson. 2007. "Gene Duplication of the zebrafish kit ligand and partitioning of melanocyte development functions to kit ligand a." PLoS Genet 3 (1):e17. doi: 10.1371/journal.pgen.0030017. Inoue, S., S. Kondo, D. M. Parichy, and M. Watanabe. 2014. "Tetraspanin 3c requirement for pigment cell interactions and boundary formation in zebrafish adult pigment stripes." Pigment Cell Melanoma Res 27 (2):190-200. doi: 10.1111/pcmr.12192. Irion, U., and C. Nusslein-Volhard. 2019. "The identification of genes involved in the evolution of color patterns in fish." Curr Opin Genet Dev 57:31-38. doi: 10.1016/j.gde.2019.07.002.

– 132 –

⁕ References ⁕

Irion, U., A. P. Singh, and C. Nusslein-Volhard. 2016. "The Developmental Genetics of Vertebrate Color Pattern Formation: Lessons from Zebrafish." Curr Top Dev Biol 117:141-69. doi: 10.1016/bs.ctdb.2015.12.012. Ishikawa, A., N. Kabeya, K. Ikeya, R. Kakioka, J. N. Cech, N. Osada, M. C. Leal, J. Inoue, M. Kume, A. Toyoda, A. Tezuka, A. J. Nagano, Y. Y. Yamasaki, Y. Suzuki, T. Kokita, H. Takahashi, K. Lucek, D. Marques, Y. Takehana, K. Naruse, S. Mori, O. Monroig, N. Ladd, C. J. Schubert, B. Matthews, C. L. Peichel, O. Seehausen, G. Yoshizaki, and J. Kitano. 2019. "A key metabolic gene for recurrent freshwater colonization and radiation in fishes." Science 364 (6443):886-889. doi: 10.1126/science.aau5656. Jablonski, N. G., and G. Chaplin. 2010. "Colloquium paper: human skin pigmentation as an adaptation to UV radiation." Proc Natl Acad Sci U S A 107 Suppl 2 (Supplement 2):8962-8. doi: 10.1073/pnas.0914628107. Jablonski, N. G., and G. Chaplin. 2017. "The colours of humanity: the evolution of pigmentation in the human lineage." Philos Trans R Soc Lond B Biol Sci 372 (1724). doi: 10.1098/rstb.2016.0349. Jacob, F. 1977. "Evolution and tinkering." Science 196 (4295):1161-6. doi: 10.1126/science.860134. Jia, Q., C. Cao, H. Tang, Y. Zhang, Q. Zheng, X. Wang, R. Zhang, X. Wang, A. Luo, H. Wei, A. Meng, Q. Zhou, H. Wang, and J. Zhao. 2017. "A 2-bp insertion (c.67_68insCC) in MC1R causes recessive white coat color in Bama miniature pigs." J Genet Genomics 44 (4):215-217. doi: 10.1016/j.jgg.2017.02.003. Johnson, S. L., A. N. Nguyen, and J. A. Lister. 2011. "mitfa is required at multiple stages of melanocyte differentiation but not to establish the melanocyte stem cell." Dev Biol 350 (2):405-13. doi: 10.1016/j.ydbio.2010.12.004. Johnson, T. C., C. A. Scholz, M. R. Talbot, K. Kelts, R. D. Ricketts, G. Ngobi, K. Beuning, I. I. Ssemmanda, and J. W. McGill. 1996. "Late Pleistocene Desiccation of Lake Victoria and Rapid Evolution of Cichlid Fishes." Science 273 (5278):1091-3. doi: 10.1126/science.273.5278.1091. Juntti, S. 2019. "The Future of Gene-Guided Neuroscience Research in Non-Traditional Model Organisms." Brain Behav Evol 93 (2-3):108-121. doi: 10.1159/000500072. Juntti, S. A., A. T. Hilliard, K. R. Kent, A. Kumar, A. Nguyen, M. A. Jimenez, J. L. Loveland, P. Mourrain, and R. D. Fernald. 2016. "A Neural Basis for Control of Cichlid Female Reproductive Behavior by Prostaglandin F2alpha." Curr Biol 26 (7):943-9. doi: 10.1016/j.cub.2016.01.067. Kaelin, C. B., S. I. Candille, B. Yu, P. Jackson, D. A. Thompson, M. A. Nix, J. Binkley, G. L. Millhauser, and G. S. Barsh. 2008. "New ligands for melanocortin receptors." Int J Obes (Lond) 32 Suppl 7 (7):S19-27. doi: 10.1038/ijo.2008.234. Kapp, F. G., J. R. Perlin, E. J. Hagedorn, J. M. Gansner, D. E. Schwarz, L. A. O'Connell, N. S. Johnson, C. Amemiya, D. E. Fisher, U. Wolfle, E. Trompouki, C. M. Niemeyer, W. Driever, and L. I. Zon. 2018. "Protection from UV light is an evolutionarily conserved feature of the haematopoietic niche." Nature 558 (7710):445-448. doi: 10.1038/s41586-018-0213-0. Katoh, K., J. Rozewicki, and K. D. Yamada. 2019. "MAFFT online service: multiple sequence alignment, interactive sequence choice and visualization." Brief Bioinform 20 (4):1160-1166. doi: 10.1093/bib/bbx108. Katoh, K., and D. M. Standley. 2013. "MAFFT multiple sequence alignment software version 7: improvements in performance and usability." Mol Biol Evol 30 (4):772-80. doi: 10.1093/molbev/mst010. – 133 –

⁕ References ⁕

Kelley, J. L., J. L. Fitzpatrick, and S. Merilaita. 2013. "Spots and stripes: ecology and colour pattern evolution in butterflyfishes." Proc Biol Sci 280 (1757):20122730. doi: 10.1098/rspb.2012.2730. Kelsh, R. N., M. Brand, Y. J. Jiang, C. P. Heisenberg, S. Lin, P. Haffter, J. Odenthal, M. C. Mullins, F. J. van Eeden, M. Furutani-Seiki, M. Granato, M. Hammerschmidt, D. A. Kane, R. M. Warga, D. Beuchle, L. Vogelsang, and C. Nusslein-Volhard. 1996. "Zebrafish pigmentation mutations and the processes of neural crest development." Development 123:369-89. Keren, H., G. Lev-Maor, and G. Ast. 2010. "Alternative splicing and evolution: diversification, exon definition and function." Nat Rev Genet 11 (5):345-55. doi: 10.1038/nrg2776. Klaassen, H., Y. Wang, K. Adamski, N. Rohner, and J. E. Kowalko. 2018. "CRISPR mutagenesis confirms the role of oca2 in melanin pigmentation in Astyanax mexicanus." Dev Biol 441 (2):313-318. doi: 10.1016/j.ydbio.2018.03.014. Kobayashi, Y., H. Chiba, K. Mizusawa, N. Suzuki, J. M. Cerda-Reverter, and A. Takahashi. 2011. "Pigment-dispersing activities and cortisol-releasing activities of melanocortins and their receptors in xanthophores and head kidneys of the goldfish Carassius auratus." Gen Comp Endocrinol 173 (3):438-46. doi: 10.1016/j.ygcen.2011.06.019. Kobayashi, Y., K. Mizusawa, H. Chiba, M. Tagawa, and A. Takahashi. 2012. "Further evidence on acetylation-induced inhibition of the pigment-dispersing activity of alpha-melanocyte-stimulating hormone." Gen Comp Endocrinol 176 (1):9-17. doi: 10.1016/j.ygcen.2011.12.001. Kobayashi, Y., K. Mizusawa, Y. Saito, and A. Takahashi. 2012. "Melanocortin systems on pigment dispersion in fish chromatophores." Front Endocrinol (Lausanne) 3:9. doi: 10.3389/fendo.2012.00009. Kobayashi, Y., K. Tsuchiya, T. Yamanome, H. B. Schioth, and A. Takahashi. 2010. "Differential expressions of melanocortin receptor subtypes in melanophores and xanthophores of barfin flounder." Gen Comp Endocrinol 168 (1):133-42. doi: 10.1016/j.ygcen.2010.04.017. Kocher, T. D. 2004. "Adaptive evolution and explosive speciation: the cichlid fish model." Nat Rev Genet 5 (4):288-98. doi: 10.1038/nrg1316. Kocher, T. D., J. A. Conroy, K. R. McKaye, and J. R. Stauffer. 1993. "Similar morphologies of cichlid fish in Lakes Tanganyika and Malawi are due to convergence." Mol Phylogenet Evol 2 (2):158-65. doi: 10.1006/mpev.1993.1016. Kolde, R. 2019. Pheatmap: pretty heatmaps. R package version, 61: 617. Kondrashov, F. A., and E. V. Koonin. 2001. "Origin of alternative splicing by tandem exon duplication." Hum Mol Genet 10 (23):2661-9. doi: 10.1093/hmg/10.23.2661. Konings, Ad. 2007. Lake Malawi Cichlids in Their Natural Habitat. Korner, A., and J. Pawelek. 1982. "Mammalian tyrosinase catalyzes three reactions in the biosynthesis of melanin." Science 217 (4565):1163-5. doi: 10.1126/science.6810464. Korzan, W. J., and R. D. Fernald. 2006. "Territorial male color predicts agonistic behavior of conspecifics in a color polymorphic species." Behavioral Ecology 18 (2):318-323. doi: 10.1093/beheco/arl093. Kosakovsky Pond, S. L., D. Posada, M. B Gravenor, C. H Woelk, and S. D. W. Frost. 2006. "Automated phylogenetic detection of recombination using a genetic algorithm." Molecular biology and evolution 23 (10):1891-1901.

– 134 –

⁕ References ⁕

Kottler, V. A., I. Koch, M. Flotenmeyer, H. Hashimoto, D. Weigel, and C. Dreyer. 2014. "Multiple pigment cell types contribute to the black, blue, and orange ornaments of male guppies (Poecilia reticulata)." PLoS One 9 (1):e85647. doi: 10.1371/journal.pone.0085647. Kratochwil, C. F. 2019. "Molecular mechanisms of convergent color pattern evolution." Zoology (Jena) 134:66-68. doi: 10.1016/j.zool.2019.04.004. Kratochwil, C. F., Y. Liang, J. Gerwin, J. M. Woltering, S. Urban, F. Henning, G. Machado-Schiaffino, C. D. Hulsey, and A. Meyer. 2018. "Agouti-related peptide 2 facilitates convergent evolution of stripe patterns across cichlid fish radiations." Science 362 (6413):457-460. doi: 10.1126/science.aao6809. Kratochwil, C. F., Y. Liang, S. Urban, J. Torres-Dowdall, and A. Meyer. 2019. "Evolutionary Dynamics of Structural Variation at a Key Locus for Color Pattern Diversification in Cichlid Fishes." Genome Biol Evol 11 (12):3452-3465. doi: 10.1093/gbe/evz261. Kratochwil, C. F., and A. Meyer. 2015a. "Closing the genotype-phenotype gap: emerging technologies for evolutionary genetics in ecological model vertebrate systems." Bioessays 37 (2):213-26. doi: 10.1002/bies.201400142. Kratochwil, C. F., and A. Meyer. 2015b. "Evolution: tinkering within gene regulatory landscapes." Curr Biol 25 (7):R285-8. doi: 10.1016/j.cub.2015.02.051. Kratochwil, C. F., and A. Meyer. 2015c. "Mapping active promoters by ChIP-seq profiling of H3K4me3 in cichlid fish - a first step to uncover cis-regulatory elements in ecological model teleosts." Mol Ecol Resour 15 (4):761-71. doi: 10.1111/1755-0998.12350. Kratochwil, C. F., and A. Meyer. 2019. "Fragile DNA contributes to repeated evolution." Genome Biol 20 (1):39. doi: 10.1186/s13059-019-1655-x. Kratochwil, C. F., M. M. Sefton, Y. Liang, and A. Meyer. 2017. "Tol2 transposon- mediated transgenesis in the Midas cichlid (Amphilophus citrinellus) - towards understanding gene function and regulatory evolution in an ecological model system for rapid phenotypic diversification." BMC Dev Biol 17 (1):15. doi: 10.1186/s12861-017-0157-x. Kratochwil, C. F., M. M. Sefton, and A. Meyer. 2015. "Embryonic and larval development in the Midas cichlid fish species flock (Amphilophus spp.): a new evo-devo model for the investigation of adaptive novelties and species differences." BMC Dev Biol 15:12. doi: 10.1186/s12861-015-0061-1. Kratochwil, C. F., S. Urban, and A. Meyer. 2019. "Genome of the Malawi golden cichlid fish (Melanochromis auratus) reveals exon loss of oca2 in an amelanistic morph." Pigment Cell Melanoma Res 32 (5):719-723. doi: 10.1111/pcmr.12799. Krauss, J., S. Geiger-Rudolph, I. Koch, C. Nusslein-Volhard, and U. Irion. 2014. "A dominant mutation in tyrp1A leads to melanophore death in zebrafish." Pigment Cell Melanoma Res 27 (5):827-30. doi: 10.1111/pcmr.12272. Kumar, S., G. Stecher, M. Suleski, and S. B. Hedges. 2017. "TimeTree: A Resource for Timelines, Timetrees, and Divergence Times." Mol Biol Evol 34 (7):1812-1819. doi: 10.1093/molbev/msx116. Kuraku, S., and A. Meyer. 2008. "Genomic analysis of cichlid fish 'natural mutants'." Curr Opin Genet Dev 18 (6):551-8. doi: 10.1016/j.gde.2008.11.002. Kurokawa, T., K. Murashita, and S. Uji. 2006. "Characterization and tissue distribution of multiple agouti-family genes in pufferfish, Takifugu rubripes." Peptides 27 (12):3165-75. doi: 10.1016/j.peptides.2006.09.013. Kwan, K. M., E. Fujimoto, C. Grabher, B. D. Mangum, M. E. Hardy, D. S. Campbell, J. M. Parant, H. J. Yost, J. P. Kanki, and C. B. Chien. 2007. "The Tol2kit: a multisite – 135 –

⁕ References ⁕

gateway-based construction kit for Tol2 transposon transgenesis constructs." Dev Dyn 236 (11):3088-99. doi: 10.1002/dvdy.21343. Labouesse, M. A., E. Dong, D. R. Grayson, A. Guidotti, and U. Meyer. 2015. "Maternal immune activation induces GAD1 and GAD2 promoter remodeling in the offspring prefrontal cortex." Epigenetics 10 (12):1143-55. doi: 10.1080/15592294.2015.1114202. Labun, K., T. G. Montague, J. A. Gagnon, S. B. Thyme, and E. Valen. 2016. "CHOPCHOP v2: a web tool for the next generation of CRISPR genome engineering." Nucleic Acids Res 44 (W1):W272-6. doi: 10.1093/nar/gkw398. Lamason, R. L., M. A. Mohideen, J. R. Mest, A. C. Wong, H. L. Norton, M. C. Aros, M. J. Jurynec, X. Mao, V. R. Humphreville, J. E. Humbert, S. Sinha, J. L. Moore, P. Jagadeeswaran, W. Zhao, G. Ning, I. Makalowska, P. M. McKeigue, D. O'Donnell, R. Kittles, E. J. Parra, N. J. Mangini, D. J. Grunwald, M. D. Shriver, V. A. Canfield, and K. C. Cheng. 2005. "SLC24A5, a putative cation exchanger, affects pigmentation in zebrafish and humans." Science 310 (5755):1782-6. doi: 10.1126/science.1116238. Lambert, M. J., W. O. Cochran, B. M. Wilde, K. G. Olsen, and C. D. Cooper. 2015. "Evidence for widespread subfunctionalization of splice forms in vertebrate genomes." Genome Res 25 (5):624-32. doi: 10.1101/gr.184473.114. Lande, R., O. Seehausen, and J. J. van Alphen. 2001. "Mechanisms of rapid sympatric speciation by sex reversal and sexual selection in cichlid fish." Genetica 112-113 (1):435-43. Lanzing, W. J. R. 1975. "The ultrastructure of erythrophores and melanophores from the skin of Tilapia mossambica (Peters)." Acta Zoologica 56 (3):205-211. Lanzing, W. J. R., and C. C. Bower. 1974. "Development of colour patterns in relation to behaviour in Tilapia mossambica (Peters)." Journal of Fish Biology 6 (1):29- 41. Le Guyader, S., and S. Jesuthasan. 2002. "Analysis of xanthophore and pterinosome biogenesis in zebrafish using methylene blue and pteridine autofluorescence." Pigment Cell Research 15 (1):27-31. Letunic, I., R. R. Copley, and P. Bork. 2002. "Common exon duplication in animals and its role in alternative splicing." Hum Mol Genet 11 (13):1561-7. doi: 10.1093/hmg/11.13.1561. Lewis, V. M., L. M. Saunders, T. A. Larson, E. J. Bain, S. L. Sturiale, D. Gur, S. Chowdhury, J. D. Flynn, M. C. Allen, D. D. Deheyn, J. C. Lee, J. A. Simon, J. Lippincott-Schwartz, D. W. Raible, and D. M. Parichy. 2019. "Fate plasticity and reprogramming in genetically distinct populations of Danio leucophores." Proc Natl Acad Sci U S A 116 (24):11806-11811. doi: 10.1073/pnas.1901021116. Li, Bo, and Colin N Dewey. 2011. "RSEM: accurate transcript quantification from RNA- Seq data with or without a reference genome." BMC bioinformatics 12 (1):323. Li, C., W. Pan, K. H. Braunewell, and J. B. Ames. 2011. "Structural analysis of Mg2+ and Ca2+ binding, myristoylation, and dimerization of the neuronal calcium sensor and visinin-like protein 1 (VILIP-1)." J Biol Chem 286 (8):6354-66. doi: 10.1074/jbc.M110.173724. Li, H, B Handsaker, A Wysoker, T Fennell, J Ruan, N Homer, and G Marth. 2009. "1000 Genome Project Data., Abecasis G., Durbin R.; 1000 Genome Project Data 000 Genome Project Data 000 Genome Project Data Processing Subgroup. The Sequence alignment/map (SAM) format and SAMtools." Bioinformatics 25:2078-2079.

– 136 –

⁕ References ⁕

Li, H., and R. Durbin. 2009. "Fast and accurate short read alignment with Burrows- Wheeler transform." Bioinformatics 25 (14):1754-60. doi: 10.1093/bioinformatics/btp324. Li, H., B. Handsaker, A. Wysoker, T. Fennell, J. Ruan, N. Homer, G. Marth, G. Abecasis, R. Durbin, and Subgroup Genome Project Data Processing. 2009. "The Sequence Alignment/Map format and SAMtools." Bioinformatics 25 (16):2078-9. doi: 10.1093/bioinformatics/btp352. Liang, Y., J. Gerwin, A. Meyer, and C. F. Kratochwil. 2020. "Developmental and Cellular Basis of Vertical Bar Color Patterns in the East African Cichlid Fish Haplochromis latifasciatus." Front Cell Dev Biol 8:62. doi: 10.3389/fcell.2020.00062. Liem, R. K. 1993. "Molecular biology of neuronal intermediate filaments." Curr Opin Cell Biol 5 (1):12-6. doi: 10.1016/s0955-0674(05)80003-1. Lin, J. C., S. Hu, P. H. Ho, H. J. Hsu, J. H. Postlethwait, and B. C. Chung. 2015. "Two Zebrafish hsd3b Genes Are Distinct in Function, Expression, and Evolution." Endocrinology 156 (8):2854-62. doi: 10.1210/en.2014-1584. Linnen, C. R., E. P. Kingsley, J. D. Jensen, and H. E. Hoekstra. 2009. "On the origin and spread of an adaptive allele in deer mice." Science 325 (5944):1095-8. doi: 10.1126/science.1175826. Linnen, C. R., Y. P. Poh, B. K. Peterson, R. D. Barrett, J. G. Larson, J. D. Jensen, and H. E. Hoekstra. 2013. "Adaptive evolution of multiple traits through multiple mutations at a single gene." Science 339 (6125):1312-6. doi: 10.1126/science.1233213. Lister, J. A. 2019. "Larval but not adult xanthophore pigmentation in zebrafish requires GTP cyclohydrolase 2 (gch2) function." Pigment Cell Melanoma Res 32 (5):724- 727. doi: 10.1111/pcmr.12783. Lister, J. A., J. Close, and D. W. Raible. 2001. "Duplicate mitf genes in zebrafish: complementary expression and conservation of melanogenic potential." Dev Biol 237 (2):333-44. doi: 10.1006/dbio.2001.0379. Lister, J. A., C. P. Robertson, T. Lepage, S. L. Johnson, and D. W. Raible. 1999. "nacre encodes a zebrafish microphthalmia-related protein that regulates neural-crest- derived pigment cell fate." Development 126 (17):3757-3767. Liu, M., and X. Guo. 2017. "A novel and stress adaptive alternative oxidase derived from alternative splicing of duplicated exon in oyster Crassostrea virginica." Sci Rep 7 (1):10785. doi: 10.1038/s41598-017-10976-w. Liu, Y., J. W. Lee, and S. L. Ackerman. 2015. "Mutations in the microtubule-associated protein 1A (Map1a) gene cause Purkinje cell degeneration." J Neurosci 35 (11):4587-98. doi: 10.1523/JNEUROSCI.2757-14.2015. Lodygin, D., M. Hermann, N. Schweingruber, C. Flugel-Koch, T. Watanabe, C. Schlosser, A. Merlini, H. Korner, H. F. Chang, H. J. Fischer, H. M. Reichardt, M. Zagrebelsky, B. Mollenhauer, S. Kugler, D. Fitzner, J. Frahm, C. Stadelmann, M. Haberl, F. Odoardi, and A. Flugel. 2019. "beta-Synuclein-reactive T cells induce autoimmune CNS grey matter degeneration." Nature 566 (7745):503-508. doi: 10.1038/s41586-019-0964-2. Lopes, S. S., X. Yang, J. Muller, T. J. Carney, A. R. McAdow, G. J. Rauch, A. S. Jacoby, L. D. Hurst, M. Delfino-Machin, P. Haffter, R. Geisler, S. L. Johnson, A. Ward, and R. N. Kelsh. 2008. "Leukocyte tyrosine kinase functions in pigment cell development." PLoS Genet 4 (3):e1000026. doi: 10.1371/journal.pgen.1000026. Losos, J. B. 2017. Improbable destinies: Fate, chance, and the future of evolution: Penguin. – 137 –

⁕ References ⁕

Losos, J. B., T. R. Jackman, A. Larson, K. Queiroz, and L. Rodriguez-Schettino. 1998. "Contingency and determinism in replicated adaptive radiations of island lizards." Science 279 (5359):2115-8. doi: 10.1126/science.279.5359.2115. Love, M. I., W. Huber, and S. Anders. 2014. "Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2." Genome Biol 15 (12):550. doi: 10.1186/s13059-014-0550-8. Lynch, M., M. S. Ackerman, J. F. Gout, H. Long, W. Sung, W. K. Thomas, and P. L. Foster. 2016. "Genetic drift, selection and the evolution of the mutation rate." Nat Rev Genet 17 (11):704-714. doi: 10.1038/nrg.2016.104. Maan, M. E., and K. M. Sefc. 2013. "Colour variation in cichlid fish: developmental mechanisms, selective pressures and evolutionary consequences." Semin Cell Dev Biol 24 (6-7):516-28. doi: 10.1016/j.semcdb.2013.05.003. Mahalwar, P., B. Walderich, A. P. Singh, and C. Nusslein-Volhard. 2014. "Local reorganization of xanthophores fine-tunes and colors the striped pattern of zebrafish." Science 345 (6202):1362-4. doi: 10.1126/science.1254837. Malinsky, M., H. Svardal, A. M. Tyers, E. A. Miska, M. J. Genner, G. F. Turner, and R. Durbin. 2018. "Whole-genome sequences of Malawi cichlids reveal multiple radiations interconnected by gene flow." Nat Ecol Evol 2 (12):1940-1955. doi: 10.1038/s41559-018-0717-x. Mallarino, R., C. Henegar, M. Mirasierra, M. Manceau, C. Schradin, M. Vallejo, S. Beronja, G. S. Barsh, and H. E. Hoekstra. 2016. "Developmental mechanisms of stripe patterns in rodents." Nature 539 (7630):518-523. doi: 10.1038/nature20109. Mallarino, R., T. A. Linden, C. R. Linnen, and H. E. Hoekstra. 2017. "The role of isoforms in the evolution of cryptic coloration in Peromyscus mice." Mol Ecol 26 (1):245-258. doi: 10.1111/mec.13663. Mallet, J., and L. E. Gilbert Jr. 1995. "Why are there so many mimicry rings? Correlations between habitat, behaviour and mimicry in Heliconius butterflies." Biological Journal of the Linnean Society 55 (2):159-180. Manceau, M., V. S. Domingues, C. R. Linnen, E. B. Rosenblum, and H. E. Hoekstra. 2010. "Convergence in pigmentation at multiple levels: mutations, genes and function." Philos Trans R Soc Lond B Biol Sci 365 (1552):2439-50. doi: 10.1098/rstb.2010.0104. Manceau, M., V. S. Domingues, R. Mallarino, and H. E. Hoekstra. 2011. "The developmental role of Agouti in color pattern evolution." Science 331 (6020):1062-5. doi: 10.1126/science.1200684. Manukyan, L., S. A. Montandon, A. Fofonjka, S. Smirnov, and M. C. Milinkovitch. 2017. "A living mesoscopic cellular automaton made of skin scales." Nature 544 (7649):173-179. doi: 10.1038/nature22031. Matsumoto, J. 1965. "Studies on fine structure and cytochemical properties of erythrophores in swordtail, Xiphophorus helleri, with special reference to their pigment granules (Pterinosomes)." J Cell Biol 27 (3):493-504. doi: 10.1083/jcb.27.3.493. Mayr, E. 1999. Systematics and the origin of species, from the viewpoint of a zoologist: Harvard University Press. Mazo-Vargas, A., C. Concha, L. Livraghi, D. Massardo, R. W. R. Wallbank, L. Zhang, J. D. Papador, D. Martinez-Najera, C. D. Jiggins, M. R. Kronforst, C. J. Breuker, R. D. Reed, N. H. Patel, W. O. McMillan, and A. Martin. 2017. "Macroevolutionary shifts of WntA function potentiate butterfly wing-pattern

– 138 –

⁕ References ⁕

diversity." Proc Natl Acad Sci U S A 114 (40):10701-10706. doi: 10.1073/pnas.1708149114. McCluskey, B. M., and J. H. Postlethwait. 2015. "Phylogeny of zebrafish, a "model species," within Danio, a "model genus"." Mol Biol Evol 32 (3):635-52. doi: 10.1093/molbev/msu325. McDermaid, A., B. Monier, J. Zhao, and Q. Ma. 2018. "ViDGER: An R package for integrative interpretation of differential gene expression results of RNA-seq data." bioRxiv:268896. McGee, M. D., R. Y. Neches, and O. Seehausen. 2016. "Evaluating genomic divergence and parallelism in replicate ecomorphs from young and old cichlid adaptive radiations." Mol Ecol 25 (1):260-8. doi: 10.1111/mec.13463. McGraw, K. J., G. E. Hill, K. J. Navara, and R. S. Parker. 2004. "Differential accumulation and pigmenting ability of dietary carotenoids in colorful finches." Physiol Biochem Zool 77 (3):484-91. doi: 10.1086/383506. McGraw, KJ, MD Beebee, GE Hill, and RS Parker. 2003. "Lutein-based plumage coloration in songbirds is a consequence of selective pigment incorporation into feathers." Comparative Biochemistry and Physiology Part B: Biochemistry and Molecular Biology 135 (4):689-696. McMenamin, S. K., E. J. Bain, A. E. McCann, L. B. Patterson, D. S. Eom, Z. P. Waller, J. C. Hamill, J. A. Kuhlman, J. S. Eisen, and D. M. Parichy. 2014. "Thyroid hormone-dependent adult pigment cell lineage and pattern in zebrafish." Science 345 (6202):1358-61. doi: 10.1126/science.1256251. Meier, J. I., D. A. Marques, S. Mwaiko, C. E. Wagner, L. Excoffier, and O. Seehausen. 2017. "Ancient hybridization fuels rapid cichlid fish adaptive radiations." Nat Commun 8:14363. doi: 10.1038/ncomms14363. Mellgren, E. M., and S. L. Johnson. 2005. "kitb, a second zebrafish ortholog of mouse Kit." Dev Genes Evol 215 (9):470-77. doi: 10.1007/s00427-005-0001-3. Meyer, A. 1993. "Phylogenetic relationships and evolutionary processes in East African cichlid fishes." Trends Ecol Evol 8 (8):279-84. doi: 10.1016/0169- 5347(93)90255-N. Meyer, A., C. H. Biermann, and G. Orti. 1993. "The phylogenetic position of the zebrafish (Danio rerio), a model system in developmental biology: an invitation to the comparative method." Proc Biol Sci 252 (1335):231-6. doi: 10.1098/rspb.1993.0070. Meyer, A., T. D. Kocher, P. Basasibwaki, and A. C. Wilson. 1990. "Monophyletic origin of Lake Victoria cichlid fishes suggested by mitochondrial DNA sequences." Nature 347 (6293):550-3. doi: 10.1038/347550a0. Meyer, A., P. A. Ritchie, and K. E. Witte. 1997. "Predicting developmental processes from evolutionary patterns: a molecular phylogeny of the zebrafish (Danio rerio) and its relatives." Philos T R Soc B 349 (1327):103-111. doi: 10.1098/rstb.1995.0096. Mills, M. G., and L. B. Patterson. 2009. "Not just black and white: pigment pattern development and evolution in vertebrates." Semin Cell Dev Biol 20 (1):72-81. doi: 10.1016/j.semcdb.2008.11.012. Minchin, J. E., and S. M. Hughes. 2008. "Sequential actions of Pax3 and Pax7 drive xanthophore development in zebrafish neural crest." Dev Biol 317 (2):508-22. doi: 10.1016/j.ydbio.2008.02.058. Montague, T. G., J. M. Cruz, J. A. Gagnon, G. M. Church, and E. Valen. 2014. "CHOPCHOP: a CRISPR/Cas9 and TALEN web tool for genome editing." Nucleic Acids Res 42 (Web Server issue):W401-7. doi: 10.1093/nar/gku410. – 139 –

⁕ References ⁕

Monteleone, M. C., S. C. Billi, M. A. Brocco, and A. C. Frasch. 2017. "Neural glycoprotein M6a is released in extracellular vesicles and modulated by chronic stressors in blood." Scientific reports 7 (1):1-12. Morgan, M, V Obenchain, M Lang, R Thompson, and N Turaga. 2019. BiocParallel: Bioconductor facilities for parallel evaluation. R package version 1.18. 1. Mori, E., J. Fujikura, M. Noguchi, K. Nakao, M. Matsubara, M. Sone, D. Taura, T. Kusakabe, K. Ebihara, T. Tanaka, K. Hosoda, K. Takahashi, I. Asaka, N. Inagaki, and K. Nakao. 2016. "Impaired adipogenic capacity in induced pluripotent stem cells from lipodystrophic patients with BSCL2 mutations." Metabolism 65 (4):543-56. doi: 10.1016/j.metabol.2015.12.015. Murali, G., U. Kodandaramaiah, and J. Fitzpatrick. 2018. "Body size and evolution of motion dazzle coloration in lizards." Behav Ecol 29 (1):79-86. doi: 10.1093/beheco/arx128. Muschick, M., A. Indermaur, and W. Salzburger. 2012. "Convergent evolution within an adaptive radiation of cichlid fishes." Curr Biol 22 (24):2362-8. doi: 10.1016/j.cub.2012.10.048. Musilova, Z., F. Cortesi, M. Matschiner, W. I. L. Davies, J. S. Patel, S. M. Stieb, F. de Busserolles, M. Malmstrom, O. K. Torresen, C. J. Brown, J. K. Mountford, R. Hanel, D. L. Stenkamp, K. S. Jakobsen, K. L. Carleton, S. Jentoft, J. Marshall, and W. Salzburger. 2019. "Vision using multiple distinct rod opsins in deep-sea fishes." Science 364 (6440):588-592. doi: 10.1126/science.aav4632. Nagaishi, H., N. Oshima, and R. Fujii. 1990. "Light-Reflecting Properties of the Iridophores of the Neon Tetra, Paracheirodon-Innesi." Comparative Biochemistry and Physiology a-Physiology 95 (3):337-341. Nagao, M., T. Ogata, Y. Sawada, and Y. Gotoh. 2016. "Zbtb20 promotes astrocytogenesis during neocortical development." Nat Commun 7:11102. doi: 10.1038/ncomms11102. Nagao, Y., T. Suzuki, A. Shimizu, T. Kimura, R. Seki, T. Adachi, C. Inoue, Y. Omae, Y. Kamei, I. Hara, Y. Taniguchi, K. Naruse, Y. Wakamatsu, R. N. Kelsh, M. Hibi, and H. Hashimoto. 2014. "Sox5 functions as a fate switch in medaka pigment cell development." PLoS Genet 10 (4):e1004246. doi: 10.1371/journal.pgen.1004246. Nakamasu, A., G. Takahashi, A. Kanbe, and S. Kondo. 2009. "Interactions between zebrafish pigment cells responsible for the generation of Turing patterns." Proc Natl Acad Sci U S A 106 (21):8429-8434. Neuwirth, E. 2014. "RColorBrewer: ColorBrewer palettes. R package version 1.1-2." The R Foundation. Nilsson Sköld, H., S. Aspengren, and M. Wallin. 2013. "Rapid color change in fish and amphibians–function, regulation, and emerging applications." Pigment Cell Melanoma Res 26 (1):29-38. Nolan, T., R. E. Hands, and S. A. Bustin. 2006. "Quantification of mRNA using realtime RT-PCR." Nat Protoc 1 (3):1559-82. doi: 10.1038/nprot.2006.236. Nord, H., N. Dennhag, J. Muck, and J. von Hofsten. 2016. "Pax7 is required for establishment of the xanthophore lineage in zebrafish embryos." Mol Biol Cell 27 (11):1853-62. doi: 10.1091/mbc.E15-12-0821. Nusslein-Volhard, C., and A. P. Singh. 2017. "How fish color their skin: A paradigm for development and evolution of adult patterns: Multipotency, plasticity, and cell competition regulate proliferation and spreading of pigment cells in Zebrafish coloration." Bioessays 39 (3). doi: 10.1002/bies.201600231. Ohno, S. 1970. "Duplication of Regulatory Genes and Receptors." In Evolution by Gene Duplication, 82-88. Berlin, Heidelberg: Springer Berlin Heidelberg. – 140 –

⁕ References ⁕

Parichy, D. M. 2006. "Evolution of danio pigment pattern development." Heredity (Edinb) 97 (3):200-10. doi: 10.1038/sj.hdy.6800867. Parichy, D. M., M. R. Elizondo, M. G. Mills, T. N. Gordon, and R. E. Engeszer. 2009. "Normal table of postembryonic zebrafish development: staging by externally visible anatomy of the living fish." Dev Dyn 238 (12):2975-3015. doi: 10.1002/dvdy.22113. Parichy, D. M., D. G. Ransom, B. Paw, L. I. Zon, and S. L. Johnson. 2000. "An orthologue of the kit-related gene fms is required for development of neural crest- derived xanthophores and a subpopulation of adult melanocytes in the zebrafish, Danio rerio." Development 127 (14):3031-3044. Parichy, D. M., and J. E. Spiewak. 2015. "Origins of adult pigmentation: diversity in pigment stem cell lineages and implications for pattern evolution." Pigment Cell Melanoma Res 28 (1):31-50. doi: 10.1111/pcmr.12332. Parichy, D. M., and J. M. Turner. 2003. "Temporal and cellular requirements for Fms signaling during zebrafish adult pigment pattern development." Development 130 (5):817-33. doi: 10.1242/dev.00307. Parker, G. H. 1936. "Color Changes in Elasmobranchs." Proc Natl Acad Sci U S A 22 (1):55-60. doi: 10.1073/pnas.22.1.55. Parker, G. H. 1937. "Integumentary color changes of elasmobranch fishes especially of Mustelus." Proceedings of the American Philosophical Society:223-247. Parker, G. H. 1948. Animal colour changes and their neurohumours : a survey of investigations, 1910-1943: The University Press. Patterson, L. B., E. J. Bain, and D. M. Parichy. 2014. "Pigment cell interactions and differential xanthophore recruitment underlying zebrafish stripe reiteration and Danio pattern evolution." Nat Commun 5:5299. doi: 10.1038/ncomms6299. Patterson, L. B., and D. M. Parichy. 2019. "Zebrafish Pigment Pattern Formation: Insights into the Development and Evolution of Adult Form." Annu Rev Genet 53:505-530. doi: 10.1146/annurev-genet-112618-043741. Penso-Dolfin, L., A. Man, W. Haerty, and F. D. Palma. 2018. "Analysis of structural variants in four African Cichlids highlights an association with developmental and immune related genes." BioRxiv. doi: 10.1101/473710. Perry, G. H., N. J. Dominy, K. G. Claw, A. S. Lee, H. Fiegler, R. Redon, J. Werner, F. A. Villanea, J. L. Mountain, R. Misra, N. P. Carter, C. Lee, and A. C. Stone. 2007. "Diet and the evolution of human amylase gene copy number variation." Nat Genet 39 (10):1256-60. doi: 10.1038/ng2123. Phillips, G. A. C., M. J. How, J. E. Lange, N. J. Marshall, and K. L. Cheney. 2017. "Disruptive colouration in reef fish: does matching the background reduce predation risk?" Journal of Experimental Biology 220 (11):1962-1974. Poletto, A. B., I. A. Ferreira, and C. Martins. 2010. "The B chromosomes of the African cichlid fish Haplochromis obliquidens harbour 18S rRNA gene copies." BMC Genet 11 (1):1. doi: 10.1186/1471-2156-11-1. Pond, S. L., and S. D. Frost. 2005. "Datamonkey: rapid detection of selective pressure on individual sites of codon alignments." Bioinformatics 21 (10):2531-3. doi: 10.1093/bioinformatics/bti320. Pond, S. L. K., and S. V. Muse. 2005. "HyPhy: hypothesis testing using phylogenies." In Statistical methods in molecular evolution, 125-181. Springer. Poulton, E. B. 1890. The colours of animals: their meaning and use, especially considered in the case of insects. Lodon: Trench, Trubnes & Co. Ltd.

– 141 –

⁕ References ⁕

Prazdnikov, D. V., and F. N. Shkil. 2019. "Experimental evidence of the role of heterochrony in evolution of the Mesoamerican cichlids pigment patterns." Evol Dev 21 (1):3-15. doi: 10.1111/ede.12272. Protas, M. E., and N. H. Patel. 2008. "Evolution of coloration patterns." Annu Rev Cell Dev Biol 24:425-46. doi: 10.1146/annurev.cellbio.24.110707.175302. Prum, R. O., and R. Torres. 2003. "Structural colouration of avian skin: convergent evolution of coherently scattering dermal collagen arrays." J Exp Biol 206 (Pt 14):2409-29. doi: 10.1242/jeb.00431. Prum, R. O., and R. H. Torres. 2004. "Structural colouration of mammalian skin: convergent evolution of coherently scattering dermal collagen arrays." J Exp Biol 207 (Pt 12):2157-72. doi: 10.1242/jeb.00989. R Development Core Team. 2019. R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria. Rasmussen, J. P., N. T. Vo, and A. Sagasti. 2018. "Fish Scales Dictate the Pattern of Adult Skin Innervation and Vascularization." Dev Cell 46 (3):344-359 e4. doi: 10.1016/j.devcel.2018.06.019. Revell, L. J. 2012. "phytools: an R package for phylogenetic comparative biology (and other things)." Methods in ecology and evolution 3 (2):217-223. Richardson, J., P. R. Lundegaard, N. L. Reynolds, J. R. Dorin, D. J. Porteous, I. J. Jackson, and E. E. Patton. 2008. "mc1r Pathway regulation of zebrafish melanosome dispersion." Zebrafish 5 (4):289-95. doi: 10.1089/zeb.2008.0541. Roberts, R. B., J. R. Ser, and T. D. Kocher. 2009. "Sexual conflict resolved by invasion of a novel sex determiner in Lake Malawi cichlid fishes." Science 326 (5955):998-1001. doi: 10.1126/science.1174705. Robinson, M. D., D. J. McCarthy, and G. K. Smyth. 2010. "edgeR: a Bioconductor package for differential expression analysis of digital gene expression data." Bioinformatics 26 (1):139-40. doi: 10.1093/bioinformatics/btp616. Rogers, R. L., L. Shao, and K. R. Thornton. 2017. "Tandem duplications lead to novel expression patterns through exon shuffling in Drosophila yakuba." PLoS Genet 13 (5):e1006795. doi: 10.1371/journal.pgen.1006795. Rosenblum, E. B., H. E. Hoekstra, and M. W. Nachman. 2004. "Adaptive reptile color variation and the evolution of the Mc1r gene." Evolution 58 (8):1794-808. doi: 10.1111/j.0014-3820.2004.tb00462.x. Rosenblum, E. B., H. Rompler, T. Schoneberg, and H. E. Hoekstra. 2010. "Molecular and functional basis of phenotypic convergence in white lizards at White Sands." Proc Natl Acad Sci U S A 107 (5):2113-7. doi: 10.1073/pnas.0911042107. Roux, N., P. Salis, A. Lambert, V. Logeux, O. Soulat, P. Romans, B. Frederich, D. Lecchini, and V. Laudet. 2019. "Staging and normal table of postembryonic development of the clownfish (Amphiprion ocellaris)." Dev Dyn 248 (7):545- 568. doi: 10.1002/dvdy.46. Rubin, C. J., H. J. Megens, A. Martinez Barrio, K. Maqbool, S. Sayyab, D. Schwochow, C. Wang, O. Carlborg, P. Jern, C. B. Jorgensen, A. L. Archibald, M. Fredholm, M. A. Groenen, and L. Andersson. 2012. "Strong signatures of selection in the domestic pig genome." Proc Natl Acad Sci U S A 109 (48):19529-36. doi: 10.1073/pnas.1217149109. Ruxton, G. D., W. L. Allen, T. N. Sherratt, and M. P. Speed. 2019. Avoiding attack: the evolutionary ecology of crypsis, aposematism, and mimicry: Oxford University Press. Saenko, S. V., S. Lamichhaney, A. Martinez Barrio, N. Rafati, L. Andersson, and M. C. Milinkovitch. 2015. "Amelanism in the corn snake is associated with the insertion – 142 –

⁕ References ⁕

of an LTR-retrotransposon in the OCA2 gene." Sci Rep 5:17118. doi: 10.1038/srep17118. Salis, P., T. Lorin, V. Lewis, C. Rey, A. Marcionetti, M. L. Escande, N. Roux, L. Besseau, N. Salamin, M. Semon, D. Parichy, J. N. Volff, and V. Laudet. 2019. "Developmental and comparative transcriptomic identification of iridophore contribution to white barring in clownfish." Pigment Cell Melanoma Res 32 (3):391-402. doi: 10.1111/pcmr.12766. Salis, P., N. Roux, O. Soulat, D. Lecchini, V. Laudet, and B. Frederich. 2018. "Ontogenetic and phylogenetic simplification during white stripe evolution in clownfishes." BMC Biol 16 (1):90. doi: 10.1186/s12915-018-0559-7. Salzburger, W. 2018. "Understanding explosive diversification through cichlid fish genomics." Nat Rev Genet 19 (11):705-717. doi: 10.1038/s41576-018-0043-9. Salzburger, W., I. Braasch, and A. Meyer. 2007. "Adaptive sequence evolution in a color gene involved in the formation of the characteristic egg-dummies of male haplochromine cichlid fishes." BMC Biol 5:51. doi: 10.1186/1741-7007-5-51. Salzburger, W., T. Mack, E. Verheyen, and A. Meyer. 2005. "Out of Tanganyika: genesis, explosive speciation, key-innovations and phylogeography of the haplochromine cichlid fishes." BMC Evol Biol 5:17. doi: 10.1186/1471-2148-5-17. Santos, M. E., L. Baldo, L. Gu, N. Boileau, Z. Musilova, and W. Salzburger. 2016. "Comparative transcriptomics of anal fin pigmentation patterns in cichlid fishes." BMC Genomics 17:712. doi: 10.1186/s12864-016-3046-y. Santos, M. E., I. Braasch, N. Boileau, B. S. Meyer, L. Sauteur, A. Bohne, H. G. Belting, M. Affolter, and W. Salzburger. 2014. "The evolution of cichlid fish egg-spots is linked with a cis-regulatory change." Nat Commun 5:5149. doi: 10.1038/ncomms6149. Santos, M. E., and W. Salzburger. 2012. "Evolution. How cichlids diversify." Science 338 (6107):619-21. doi: 10.1126/science.1224818. Saunders, L. M., A. K. Mishra, A. J. Aman, V. M. Lewis, M. B. Toomey, J. S. Packer, X. Qiu, J. L. McFaline-Figueroa, J. C. Corbo, C. Trapnell, and D. M. Parichy. 2019. "Thyroid hormone regulates distinct paths to maturation in pigment cell lineages." Elife 8:e45181. doi: 10.7554/eLife.45181. Schindelin, J., I. Arganda-Carreras, E. Frise, V. Kaynig, M. Longair, T. Pietzsch, S. Preibisch, C. Rueden, S. Saalfeld, B. Schmid, J. Y. Tinevez, D. J. White, V. Hartenstein, K. Eliceiri, P. Tomancak, and A. Cardona. 2012. "Fiji: an open- source platform for biological-image analysis." Nat Methods 9 (7):676-82. doi: 10.1038/nmeth.2019. Schioth, H. B., A. Vastermark, and R. D. Cone. 2011. "Reply to Braasch and Postlethwait: Evolutionary origin of the teleost A2 agouti genes (agouti signaling protein 2 and agouti-related protein 2) remains unclear." Proc Natl Acad Sci U S A 108 (13):E49-50. doi: 10.1073/pnas.1102120108. Schonthaler, H. B., J. M. Lampert, J. von Lintig, H. Schwarz, R. Geisler, and S. C. Neuhauss. 2005. "A mutation in the silver gene leads to defects in melanosome biogenesis and alterations in the visual system in the zebrafish mutant fading vision." Dev Biol 284 (2):421-36. doi: 10.1016/j.ydbio.2005.06.001. Schuske, K., and E. M. Jorgensen. 2004. "Neuroscience. Vesicular glutamate transporter- -shooting blanks." Science 304 (5678):1750-2. doi: 10.1126/science.1100475. Schwarzer, J., A. Lamboj, K. Langen, B. Misof, and U. K. Schliewen. 2014. "Phylogeny and age of chromidotilapiine cichlids (Teleostei: Cichlidae)." Hydrobiologia 748 (1):185-199. doi: 10.1007/s10750-014-1918-1.

– 143 –

⁕ References ⁕

Scotto-Lavino, E., G. Du, and M. A. Frohman. 2006a. "3' end cDNA amplification using classic RACE." Nat Protoc 1 (6):2742-5. doi: 10.1038/nprot.2006.481. Scotto-Lavino, E., G. Du, and M. A. Frohman. 2006b. "5' end cDNA amplification using classic RACE." Nat Protoc 1 (6):2555-62. doi: 10.1038/nprot.2006.480. Seegers, L. 2008. "The fishes collected by GA Fischer in East Africa in 1883 and 1885/86." Zoosystematics and Evolution 84 (2):149-195. Seehausen, O., P. J. Mayhew, and J. J. M. van Alphen. 2001. "Evolution of colour patterns in East African cichlid fish." Journal of Evolutionary Biology 12 (3):514-534. doi: 10.1046/j.1420-9101.1999.00055.x. Selz, Y., I. Braasch, C. Hoffmann, C. Schmidt, C. Schultheis, M. Schartl, and J. N. Volff. 2007. "Evolution of melanocortin receptors in teleost fish: the melanocortin type 1 receptor." Gene 401 (1-2):114-22. doi: 10.1016/j.gene.2007.07.005. Shainer, I., A. Buchshtab, T. A. Hawkins, S. W. Wilson, R. D. Cone, and Y. Gothilf. 2017. "Novel hypophysiotropic AgRP2 neurons and pineal cells revealed by BAC transgenesis in zebrafish." Sci Rep 7:44777. doi: 10.1038/srep44777. Shainer, I., M. Michel, G. D. Marquart, A. A. Bhandiwad, N. Zmora, Z. Ben-Moshe Livne, Y. Zohar, A. Hazak, Y. Mazon, D. Forster, L. Hollander-Cohen, R. D. Cone, H. A. Burgess, and Y. Gothilf. 2019. "Agouti-Related Protein 2 Is a New Player in the Teleost Stress Response System." Curr Biol 29 (12):2009-2019 e7. doi: 10.1016/j.cub.2019.05.021. Shawkey, M. D., and G. E. Hill. 2006. "Significance of a basal melanin layer to production of non-iridescent structural plumage color: evidence from an amelanotic Steller's jay (Cyanocitta stelleri)." J Exp Biol 209 (Pt 7):1245-50. doi: 10.1242/jeb.02115. Shirak, A., M. Grabherr, F. Di Palma, K. Lindblad-Toh, G. Hulata, M. Ron, T. D. Kocher, and E. Seroussi. 2010. "Identification of repetitive elements in the genome of Oreochromis niloticus: tilapia repeat masker." Mar Biotechnol (NY) 12 (2):121- 5. doi: 10.1007/s10126-009-9236-8. Singh, A. P., A. Dinwiddie, P. Mahalwar, U. Schach, C. Linker, U. Irion, and C. Nüsslein-Volhard. 2016. "Pigment cell progenitors in zebrafish remain multipotent through metamorphosis." Developmental cell 38 (3):316-330. Singh, A. P., and C. Nusslein-Volhard. 2015. "Zebrafish stripes as a model for vertebrate colour pattern formation." Curr Biol 25 (2):R81-R92. doi: 10.1016/j.cub.2014.11.013. Singh, A. P., U. Schach, and C. Nusslein-Volhard. 2014. "Proliferation, dispersal and patterned aggregation of iridophores in the skin prefigure striped colouration of zebrafish." Nat Cell Biol 16 (6):607-14. doi: 10.1038/ncb2955. Skold, H. N., S. Aspengren, K. L. Cheney, and M. Wallin. 2016. "Fish Chromatophores- -From Molecular Motors to Animal Behavior." Int Rev Cell Mol Biol 321:171- 219. doi: 10.1016/bs.ircmb.2015.09.005. Smit, A. F. A., R. Hubley, and P. Green. 2015. RepeatMasker Open-4.0. 2013–2015. Soneson, C., M. I. Love, and M. D. Robinson. 2015. "Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences." F1000Res 4:1521. doi: 10.12688/f1000research.7563.2. Spencer, H. 1896. The principles of biology. Vol. 1: D. Appleton. Stern, D. L. 2013. "The genetic causes of convergent evolution." Nat Rev Genet 14 (11):751-64. doi: 10.1038/nrg3483. Stevens, M., and S. Merilaita. 2011. Animal camouflage: mechanisms and function: Cambridge University Press.

– 144 –

⁕ References ⁕

Stiassny, M. L. J., and A. Meyer. 1999. "Cichlids of the Rift Lakes." Scientific American 280 (2):64-69. doi: 10.1038/scientificamerican0299-64. Stoltzfus, A., and D. M. McCandlish. 2017. "Mutational Biases Influence Parallel Adaptation." Mol Biol Evol 34 (9):2163-2172. doi: 10.1093/molbev/msx180. Streelman, J. T., R. C. Albertson, and T. D. Kocher. 2003. "Genome mapping of the orange blotch colour pattern in cichlid fishes." Mol Ecol 12 (9):2465-71. doi: 10.1046/j.1365-294x.2003.01920.x. Study, The International HIV Controllers. 2010. "The major genetic determinants of HIV-1 control affect HLA class I peptide presentation." Science 330 (6010):1551. Sugimoto, M. 2002. "Morphological color changes in fish: regulation of pigment cell density and morphology." Microsc Res Tech 58 (6):496-503. doi: 10.1002/jemt.10168. Sugimoto, M., N. Uchida, and M. Hatayama. 2000. "Apoptosis in skin pigment cells of the medaka, Oryzias latipes (Teleostei), during long-term chromatic adaptation: the role of sympathetic innervation." Cell Tissue Res 301 (2):205-16. doi: 10.1007/s004410000226. Sugimoto, M., M. Yuki, T. Miyakoshi, and K. Maruko. 2005. "The influence of long- term chromatic adaptation on pigment cells and striped pigment patterns in the skin of the zebrafish, Danio rerio." J Exp Zool A Comp Exp Biol 303 (6):430-40. doi: 10.1002/jez.a.177. Sullivan, C. S., I. Gotthard, E. V. Wyatt, S. Bongu, V. Mohan, R. J. Weinberg, and P. F. Maness. 2018. "Perineuronal Net Protein Neurocan Inhibits NCAM/EphA3 Repellent Signaling in GABAergic Interneurons." Sci Rep 8 (1):6143. doi: 10.1038/s41598-018-24272-8. Tadokoro, R., Y. Shikaya, and Y. Takahashi. 2019. "Wide coverage of the body surface by melanocyte-mediated skin pigmentation." Dev Biol 449 (2):83-89. doi: 10.1016/j.ydbio.2018.04.016. Takei, Y., Y. S. Kikkawa, N. Atapour, T. K. Hensch, and N. Hirokawa. 2015. "Defects in Synaptic Plasticity, Reduced NMDA-Receptor Transport, and Instability of Postsynaptic Density Proteins in Mice Lacking Microtubule-Associated Protein 1A." J Neurosci 35 (47):15539-54. doi: 10.1523/JNEUROSCI.2671-15.2015. Takeuchi, I. K. 1976. "Electron microscopy of two types of reflecting chromatophores (iridophores and leucophores) in the guppy, Lebistes reticulatus Peters." Cell Tissue Res 173 (1):17-27. doi: 10.1007/bf00219263. Tao, B., H. Hu, K. Mitchell, J. Chen, H. Jia, Z. Zhu, V. L. Trudeau, and W. Hu. 2018. "Secretogranin-II plays a critical role in zebrafish neurovascular modeling." J Mol Cell Biol 10 (5):388-401. doi: 10.1093/jmcb/mjy027. Terai, Y., N. Morikawa, K. Kawakami, and N. Okada. 2003. "The complexity of alternative splicing of hagoromo mRNAs is increased in an explosively speciated lineage in East African cichlids." Proc Natl Acad Sci U S A 100 (22):12798- 12803. Thayer, G. H. 1909. Concealing-coloration in the animal kingdom: an exposition of the laws of disguise through color and pattern: being a summary of Abbott H. Thayer's discoveries. New York: Macmillan Company. Theos, A. C., S. T. Truschel, G. Raposo, and M. S. Marks. 2005. "The Silver locus product Pmel17/gp100/Silv/ME20: controversial in name and in function." Pigment Cell Res 18 (5):322-36. doi: 10.1111/j.1600-0749.2005.00269.x. Turing, A. M. 1952. "The chemical basis of morphogenesis." Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 237 (641):37-72. doi: 10.1098/rstb.1952.0012. – 145 –

⁕ References ⁕

Turner, G. F. 2007. "Adaptive radiation of cichlid fish." Curr Biol 17 (19):R827-31. doi: 10.1016/j.cub.2007.07.026. Valente, G. T., M. A. Conte, B. E. Fantinatti, D. C. Cabral-de-Mello, R. F. Carvalho, M. R. Vicari, T. D. Kocher, and C. Martins. 2014. "Origin and evolution of B chromosomes in the cichlid fish Astatotilapia latifasciata based on integrated genomic analyses." Mol Biol Evol 31 (8):2061-72. doi: 10.1093/molbev/msu148. Van Belleghem, S. M., P. Rastas, A. Papanicolaou, S. H. Martin, C. F. Arias, M. A. Supple, J. J. Hanly, J. Mallet, J. J. Lewis, H. M. Hines, M. Ruiz, C. Salazar, M. Linares, G. R. P. Moreira, C. D. Jiggins, B. A. Counterman, W. O. McMillan, and R. Papa. 2017. "Complex modular architecture around a simple toolkit of wing pattern genes." Nat Ecol Evol 1 (3):52. doi: 10.1038/s41559-016-0052. Van de Peer, Y., S. Maere, and A. Meyer. 2009. "The evolutionary significance of ancient genome duplications." Nat Rev Genet 10 (10):725-32. doi: 10.1038/nrg2600. van Dijk, E. L., Y. Jaszczyszyn, D. Naquin, and C. Thermes. 2018. "The Third Revolution in Sequencing Technology." Trends Genet 34 (9):666-681. doi: 10.1016/j.tig.2018.05.008. Varshney, G. K., W. Pei, M. C. LaFave, J. Idol, L. Xu, V. Gallardo, B. Carrington, K. Bishop, M. Jones, M. Li, U. Harper, S. C. Huang, A. Prakash, W. Chen, R. Sood, J. Ledin, and S. M. Burgess. 2015. "High-throughput gene targeting and phenotyping in zebrafish using CRISPR/Cas9." Genome Res 25 (7):1030-42. doi: 10.1101/gr.186379.114. Vastermark, A., A. Krishnan, M. E. Houle, R. Fredriksson, J. M. Cerda-Reverter, and H. B. Schioth. 2012. "Identification of distant Agouti-like sequences and re- evaluation of the evolutionary history of the Agouti-related peptide (AgRP)." PLoS One 7 (7):e40982. doi: 10.1371/journal.pone.0040982. Verheyen, E., W. Salzburger, J. Snoeks, and A. Meyer. 2003. "Origin of the superflock of cichlid fishes from Lake Victoria, East Africa." Science 300 (5617):325-9. doi: 10.1126/science.1080699. Volkening, A., and B. Sandstede. 2018. "Iridophores as a source of robustness in zebrafish stripes and variability in Danio patterns." Nat Commun 9 (1):3231. doi: 10.1038/s41467-018-05629-z. von Frisch, Karl. 1910. "Ueber die Beziehungen der Pigmentzellen in der Fischhaut zum sympathischen Nervensystem." Festschrift R. Hertwig 3:15-28. von Frisch, Karl. 1911. "Beitraage zur Physiologie der pigmentzellen in der fischhaut." Pflugers Arch Gesamte Physiol Menschen Tiere 138:319-387. Vonk, F. J., N. R. Casewell, C. V. Henkel, A. M. Heimberg, H. J. Jansen, R. J. McCleary, H. M. Kerkkamp, R. A. Vos, I. Guerreiro, J. J. Calvete, W. Wuster, A. E. Woods, J. M. Logan, R. A. Harrison, T. A. Castoe, A. P. de Koning, D. D. Pollock, M. Yandell, D. Calderon, C. Renjifo, R. B. Currier, D. Salgado, D. Pla, L. Sanz, A. S. Hyder, J. M. Ribeiro, J. W. Arntzen, G. E. van den Thillart, M. Boetzer, W. Pirovano, R. P. Dirks, H. P. Spaink, D. Duboule, E. McGlinn, R. M. Kini, and M. K. Richardson. 2013. "The king cobra genome reveals dynamic gene evolution and adaptation in the snake venom system." Proc Natl Acad Sci U S A 110 (51):20651-6. doi: 10.1073/pnas.1314702110. Wagner, A. 2014. Arrival of the fittest: solving evolution's greatest puzzle: Penguin. Wallace, A. R. 1877. The colours of animals and plants: Macmillan. Weadick, C. J., and B. S. Chang. 2012. "An improved likelihood ratio test for detecting site-specific functional divergence among clades of protein-coding genes." Mol Biol Evol 29 (5):1297-300. doi: 10.1093/molbev/msr311. Wickham, H. 2019. "R Package ‘stringR’v. 1.4. 0." Cran R. – 146 –

⁕ References ⁕

Wickham, H, W Chang, L Henry, TL Pedersen, K Takahashi, C Wilke, and K Woo. 2019. "R Package ‘ggplot2’v. 3.1. 1." Cran R. Wickham, Hadley, Romain François, Lionel Henry, and Kirill Müller. 2019. "R Package ‘dplyr’ v.0.8.0.1." Cran R. Wilke, C. O. 2017. cowplot: Streamlined Plot Theme and Plot Annotations for “ggplot2”. R package version 0.9. 4. Witte, F., C. D. N. Barel, E. L. Witte-Maas, and M. J. P. VAN OIJEN. 1976. "An introduction to the taxonomy and morphology of the haplochromine Cichlidae from Lake Victoria." Netherlands journal of zoology 27 (4):333-380. Woltering, J. M., M. Holzem, R. F. Schneider, V. Nanos, and A. Meyer. 2018. "The skeletal ontogeny of Astatotilapia burtoni - a direct-developing model system for the evolution and development of the teleost body plan." BMC Dev Biol 18 (1):8. doi: 10.1186/s12861-018-0166-4. Woodcock, M. R., J. Vaughn-Wolfe, A. Elias, D. K. Kump, K. D. Kendall, N. Timoshevskaya, V. Timoshevskiy, D. W. Perry, J. J. Smith, J. E. Spiewak, D. M. Parichy, and S. R. Voss. 2017. "Identification of Mutant Genes and Introgressed Tiger Salamander DNA in the Laboratory Axolotl, Ambystoma mexicanum." Sci Rep 7 (1):6. doi: 10.1038/s41598-017-00059-1. Woods, I. G., D. A. Lyons, M. G. Voas, H. M. Pogoda, and W. S. Talbot. 2006. "nsf is essential for organization of myelinated axons in zebrafish." Curr Biol 16 (7):636-48. doi: 10.1016/j.cub.2006.02.067. Wu, X., Y. Zhang, L. Shen, J. Du, J. Luo, C. Liu, Q. Pu, R. Yang, X. Li, L. Bai, G. Tang, S. Zhang, and L. Zhu. 2016. "A 6-bp deletion in exon 8 and two mutations in introns of TYRP1 are associated with blond coat color in Liangshan pigs." Gene 578 (1):132-6. doi: 10.1016/j.gene.2015.12.011. Wu, Z., Z. Deng, M. Huang, Y. Hou, H. Zhang, H. Chen, and J. Ren. 2019. "Whole- Genome Resequencing Identifies KIT New Alleles That Affect Coat Color Phenotypes in Pigs." Front Genet 10:218. doi: 10.3389/fgene.2019.00218. Xie, K. T., G. Wang, A. C. Thompson, J. I. Wucherpfennig, T. E. Reimchen, A. D. C. MacColl, D. Schluter, M. A. Bell, K. M. Vasquez, and D. M. Kingsley. 2019. "DNA fragility in the parallel evolution of pelvic reduction in stickleback fish." Science 363 (6422):81-84. doi: 10.1126/science.aan1425. Yaguchi, H., I. Yabe, H. Takahashi, M. Watanabe, T. Nomura, T. Kano, M. Matsumoto, K. I. Nakayama, M. Watanabe, and S. Hatakeyama. 2017. "Sez6l2 regulates phosphorylation of ADD and neuritogenesis." Biochem Biophys Res Commun 494 (1-2):234-241. doi: 10.1016/j.bbrc.2017.10.047. Yang, Z. 2007. "PAML 4: phylogenetic analysis by maximum likelihood." Mol Biol Evol 24 (8):1586-91. doi: 10.1093/molbev/msm088. Yang, Z. 2014. Molecular evolution: a statistical approach: Oxford University Press. Yoo, J. H., T. Takeuchi, M. Tagawa, and T. Seikai. 2000. "Effect of Thyroid Hormones on the Stage-specific Pigmentation of the Japanese Flounder Paralichthys olivaceus." Zoolog Sci 17 (8):1101-6. doi: 10.2108/zsj.17.1101. Yoshioka, S., B. Matsuhana, S. Tanaka, Y. Inouye, N. Oshima, and S. Kinoshita. 2011. "Mechanism of variable structural colour in the neon tetra: quantitative evaluation of the Venetian blind model." J R Soc Interface 8 (54):56-66. doi: 10.1098/rsif.2010.0253. Young, J. K., T. Heinbockel, and M. C. Gondre-Lewis. 2013. "Astrocyte fatty acid binding protein-7 is a marker for neurogenic niches in the rat hippocampus." Hippocampus 23 (12):1476-83. doi: 10.1002/hipo.22200.

– 147 –

⁕ References ⁕

Young, M. D., M. J. Wakefield, G. K. Smyth, and A. Oshlack. 2010. "Gene ontology analysis for RNA-seq: accounting for selection bias." Genome Biol 11 (2):R14. doi: 10.1186/gb-2010-11-2-r14. Zhang, C., Y. Song, D. A. Thompson, M. A. Madonna, G. L. Millhauser, S. Toro, Z. Varga, M. Westerfield, J. Gamse, W. Chen, and R. D. Cone. 2010. "Pineal- specific agouti protein regulates teleost background adaptation." Proc Natl Acad Sci U S A 107 (47):20164-71. doi: 10.1073/pnas.1014941107. Zhang, Y., M. Vanmeert, A. Siekierska, A. Ny, J. John, G. Callewaert, E. Lescrinier, W. Dehaen, P. A. M. de Witte, and R. M. Kaminski. 2017. "Inhibition of glutamate decarboxylase (GAD) by ethyl ketopentenoate (EKP) induces treatment-resistant epileptic seizures in zebrafish." Sci Rep 7 (1):7195. doi: 10.1038/s41598-017- 06294-w.

– 148 –

⁕ Acknowledgments ⁕

Acknowledgments

General acknowledgments

Firstly, I would like to thank Prof. Axel Meyer for accepting me as a PhD student and giving me the chance to conduct my research in his lab. Throughout my time in the lab, I enjoyed a lot and learned so much about science and the science community. I am particularly thankful to him as he gave me the opportunity to study coloration and pigmentation in cichlid fishes, one of the most beautiful things in the world. I am thankful to Prof. Mark van Kleunen, for agreeing to referee my thesis and attend my defense. Also, I want to thank the China Scholarship Council as well as the University of Konstanz for financial supports. I want to give eternal thanks to Dr. Claudius Kratochwil for everything he gave me. Without him, I might lose my research direction again and again. Without him, I might make wrong and stupid decisions again and again. His patience and mentoring helped me to survive my PhD. His diligence and enthusiasm will keep inspiring me in the future. I am also thankful to other members of TeamStripe: Dr. Maggie Sefton, Jan Gerwin, Sabina Urban, and all the others. I am very grateful to Dr. Joost Woltering for insightful discussion. I want to thank all my co-authors as well as the whole Meyer lab for the productive and pleasant past few years. Special thanks to Ingrid Bader and Christiane Weber for excellent administrative assistance. Extra thanks to all TAs from Meyer lab and staff from the Electron Microscopy Centre for their helpfulness. I cordially thank my parents for supporting me throughout my whole life. Finally, I genuinely thank Kaijie Liu, who makes my life so much better, for her incredible and invaluable help in all kinds of matters. Thank Dibo for his curiosity, passion, and always being the source of joy.

– 149 –

⁕ Acknowledgments ⁕

Chapter specific acknowledgments Chapter I

This work was supported by the Swiss National Science Foundation (P2BSP3_148629), the Marie Curie Zukunftskolleg Incoming Fellowship Program (291784), the Baden- Württemberg Foundation, the Deutsche Forschungsgemeinschaft (DFG, KR 4670/2-1), and the Young Scholar Fund of the University of Konstanz (all to C.F.K.). Y.L. is funded by the China Scholarship Council (CSC). A.M. and C.D.H. are funded by DFG and the University of Konstanz. Partial funding came from an ERC Advanced Grant GenAdap (no. 293700) to A.M. The authors thank S. Finkernagel for her exceptional technical assistance; A. Kautt, A. Nater, J. Torres-Dowdall, and the anonymous referees for insightful comments that improved the manuscript; and the University of Konstanz bioimaging center and animal facility.

Chapter II

This work was supported by the Baden-Württemberg Foundation (to C.F.K.), the Deutsche Forschungsgemeinschaft (DFG, KR 4670/2-1, KR 4670/4-1, TO914/2-1, ME 1725/20-1, ME 1725/21-1 to C.F.K., J.T.D., and A.M.) a stipend from the China Scholarship Council (CSC, to Y.L.), a bridge funding from the International Max Planck Research School (IMPRS) for Organismal Biology (to S.U.), and an ERC advanced grant (no. 293700 GenAdapt; to A.M.

Chapter III

This work was supported by a stipend from the China Scholarship Council (CSC, to Y.L.), the Baden-Württemberg Foundation (to C.F.K.) grants by the Deutsche Forschungsgemeinschaft (DFG, to A.M. and KR 4670/2-1, KR 4670/4-1 to C.F.K.) and the University of Konstanz (all authors). The authors thank Jan Gerwin and the staff of the animal facility of the University of Konstanz for their valuable help as well as Yoav Gothilf for helpful discussions.

– 150 –

⁕ Acknowledgments ⁕

Chapter IV

This work was supported by a stipend from the China Scholarship Council (CSC, to Y.L.), the Baden-Württemberg Foundation (to C.F.K.) grants by the Deutsche Forschungsgemeinschaft (DFG, to A.M. and KR 4670/2-1, KR 4670/4-1 to C.F.K.) and the University of Konstanz (all authors). The authors thank Jan Gerwin, the staff of the animal facility and the Electron Microscopy Centre of the University of Konstanz for their valuable help.

Chapter V

This work was supported by a stipend from the China Scholarship Council (CSC, to YL), the Baden-Württemberg Foundation (to CK) grants by the Deutsche Forschungsgemeinschaft (DFG, to AM and KR 4670/2-1 and KR 4670/4-1 to CK), an ERC Advanced grant GenAdap number 293700 by the European Research Council (to AM), and the University of Konstanz (YL, JG, AM, and CK). We thank the staff of the animal facility of the University of Konstanz for their excellent help.

– 151 –

⁕ Author Contributions ⁕

Author Contributions

Chapter I

C.F.K. conceptualized the study; provided funding; planned and supervised all experiments; collected, analyzed, and visualized all data, if not otherwise indicated; and wrote the manuscript. Y.L. performed qPCRs and in situ hybridization. J.G. and S.U. conducted the Lake Malawi cross and assisted with the sequencing and enhancer assay. J.M.W. designed and generated the CRISPR-Cas9 knockouts and helped with the enhancer assay. F.H. and G.M.-S. provided data for Lake Victoria crosses and contributed to their analysis. C.D.H. conducted the phylogenetic analysis and edited the manuscript. A.M. conceptualized the study, provided funding, and revised the manuscript. All authors gave comments on the manuscript and approved the final version.

Chapter II

CFK conceived and supervised the study. CFK, YL and SU conducted experiments. CFK, YL, SU and JTD performed analyses. CFK wrote the manuscript. AM edited the manuscript. CFK and AM provided funding. All authors gave comments on the manuscript and approved the final version.

Chapter III

YL and CFK conceived and supervised the study. YL and MG conducted experiments and analyses. YL and CFK wrote the manuscript. AM edited the manuscript. CFK and AM provided funding.

– 152 –

⁕ Author Contributions ⁕

Chapter IV

CFK conceived and supervised the study, CFK analyzed the RNA-seq data, YL conducted all other experiments and analyses. YL and CFK wrote the manuscript. AM edited the manuscript. CFK and AM provided funding.

Chapter V

YL: investigation, visualization, methodology, formal analysis, funding acquisition, validation, writing – original draft, and writing – review and editing. JG: investigation. AM: supervision, funding acquisition, and writing – review and editing. CFK: conceptualization, resources, formal analysis, supervision, funding acquisition, visualization, methodology, writing – original draft, project administration, and writing – review and editing.

– 153 –

⁕ Supplementary Material ⁕

Supplementary Material

Chapter I

Fig. S.I.1 Fine-mapping of Pnye / H.sau F2 recombinants (A and B) Enlarged QTL peaks on chromosome 18, of the two crosses between Pnye / Hsau (A) and Pnye / Hchi (B) (C) To finemap the causal interval we first used first microsatellite markers to reduce the interval, next we sequenced SNPs to further reduce the causal interval. (D) Genotyping information for parents, three F1s and the four recombinants (F2_Rec01-04).

– 154 –

⁕ Supplementary Material ⁕

Fig. S.I. 2 Protein sequences of QTL genes (A to D) To screen for coding mutations within the interval we sequenced agrp2, atp6V0d2 and Exon 66- 71 of unk in the Pnye × Hsau parents to screen for changes resulting in protein sequence differences. We found no non-synonymous mutations in atp6V0d2 (A) unk and atp6V0d2 (B). In contrast, agrp2 revealed an amino-acid substitution at position 63 of 116 (Histidine in Pnye, Tyrosin in Hsau) (C) that was however not fixed in Hchi and showed no association with the stripe phenotype across species (D).

– 155 –

⁕ Supplementary Material ⁕

Fig. S.I.3 Phylogeny of the agouti family. (A and B) Illustration of the proposed phylogeny of agrp2 (Synonym: asip2b) and other asip and agrp genes after (Braasch and Postlethwait 2011)(A) and (Schioth, Vastermark, and Cone 2011) that proposed a second phylogeny that is equally likely (B). The agouti signaling protein family encompass the Agouti-related peptide/protein (Agrp) and Agouti-signaling peptide (Asip) in mammals and asip1, asip2, agrp1 and agrp2 in teleosts (ray-finned fishes). Asip in mammals (short Agouti) and asip1 in teleosts has been causally associated with color pattern variation in mice (Manceau et al. 2011) and teleosts (Ceinos et al. 2015). Agrp (in vertebrates) and Agrp1 (in teleosts) are neuropeptides and involved in metabolic homeostasis. On the other hand, evolutionary relationships for asip2 and agrp2 are controversial, and it is unclear if agrp2 is more closely related to Agrp/agrp1 or Asip/asip1 (Braasch and Postlethwait 2011). In teleosts (zebrafish), agrp2 is mainly expressed in the pineal gland where it binds Melanocortin receptor 1 (Mc1r) and regulates background matching by controlling α-melanocyte stimulating hormone (α-Msh) release indicating it has a function in regulating skin pigmentation in other teleosts (Zhang et al. 2010). It therefore shares regulatory properties with agrp1 (brain specific expression (Shainer et al. 2017)), but has functional similarities with asip1 (as Mc1r antagonist). R1, R2: Vertebrate genome duplication 1 and 2; TSGD: Teleost-specific genome duplication.

– 156 –

⁕ Supplementary Material ⁕

Fig. S.I.4 Expression of agrp2, its paralogs and neighboring genes of the fine-mapped interval. (A) Relative agrp2 expression in the brain. (B) Agrp2 shows differential expression in Pnye F2 individuals with homozygous agrp2 individuals (N/N) > heterozygotes (N/S) > homozygous agrp2Hsau individuals (S/S). Other genes in the interval or paralogs did not show this pattern. (C to G) In situ hybridization reveals agrp2 expression in lower dermis (sc) of Pnye (arrows; C, D, G) but not Hsau (asterisks; E and F). Higher resolution confocal imaging reveals the relatively ubiquitous expression pattern within the sc (G). Bars indicate mean + sd; * P < 0.05; ** P < 0.01; *** P < 0.001. ep, epidermis; ss, Stratum spongiosum; sc, Stratum compactum; hy, hypodermis; mu, muscle. Scale bar: 25µm

Fig. S.I.5 RT-PCR of agrp2, ß-actin, atp6V0d2, agrp1, asip1 and asip2. (A) RT-PCR for agrp2 and ß-actin in Pundamilia nyererei (Pnye) and Haplochromis sauvagei (Hsau) showing strong expression of agrp2 in brain tissue of both species and skin tissue of Pnye but not Hsau. No detectable bands can be seen with cDNA from liver, eye, muscle and intestines. (B) RT-PCR in F2 hybrids of Pnye and Hsau. Only agrp2 shows clear amplification, but only in heterozygotes or in individuals homozygous for Pnye allele of agrp2 and not in individuals homozygous for the Hsau allele of agrp2.

– 157 –

⁕ Supplementary Material ⁕

Fig. S.I.6 Differential expression along the anterior- posterior axis in bars and interbars (i.e. regions between melanic bars) of Pnye. To check for anterior-posterior variation in agrp2 expression we performed qPCR on the first two bar (dark orange) and interbar regions (regions between bars; light orange). No significant differences could be found (ANOVA, P = 0.792). Bars indicate mean + sd

Fig. S.I.7 Association-mapping uncovers an intronic 1.1kb interval (enh.a) with alternatively fixed alleles. (A to B) Association-mapping contrasting Pnye and Hsau population samples (A) and Pnye and Hchi population samples (B). The dotted lines indicate the maximum association, obtained if a variant is alternatively fixed in both species. The comparisons reveal a 1.1kb interval (enh.a; green box) with variants that are perfectly associated with the stripe phenotype in both comparisons.

– 158 –

⁕ Supplementary Material ⁕

Fig. S.I.8 The intronic target interval enh.a shows species-specific cis-regulatory activity. (A) Enh.a sequences from Pnye and Hsau and a control sequence downstream of agrp2 (ctrl_Pnye) were cloned into Tol2-minP-eGFP constructs and injected into zebrafish oocytes. GFP expression in transiently expressing F0 embryos differed significantly at five days post fertilization with enh.a_Pnye::eGFP > enh.a_Hsau::eGFP > ctrl_Pnye::eGFP. Horizontal lines show group means. *** P < 0.001 (B to E) Larvae of stable transgenic F1 individuals showed GFP expression along the trunk, both in enh.a_Pnye::eGFP (B, C; strong fluorescence), enh.a_Hsau::eGFP (D, weak fluorescence) compared to a non-transgenic control (E). Scale bar: 75µm.

Fig. S.I.9 Graphical summary of the genetic basis of stripe maintenance and suppression and CRISPR-Cas9 knockout.

– 159 –

⁕ Supplementary Material ⁕

Fig. S.I.10 CRISPR-Cas9 mutants genotypes and phenotypes. (A to D) Sanger sequencing data for the four obtained mutants showing deletions or insertions in Exon 1 (A and B), Exon 2 (C) and Exon 3 (D) of agrp2, many of which result in frameshifts. Asterisks (*) indicate clones with mutations, red asterisks clones with frame shift mutations or early stop codons. (E and F) Phenotypes of two additional mutants, CRISPR_mt_01 (E) and CRISPR_mt_02 (F).

– 160 –

⁕ Supplementary Material ⁕

Fig. S.I.11 Photographs of species analyzed for gene expression differences of agrp2. Photographs of all species from Lakes Victoria, Malawi and Tanganyika used for the comparative qPCR experiment. Lines indicate horizontal stripes.

– 161 –

⁕ Supplementary Material ⁕

Fig. S.I.12 Enh.a sequence across 29 East African cichlids. Summary of all variants in the enh.a sequence that differ in the parents of any of the three hybrid crosses Pnye × Hsau, Pnye × Hchi or Pdem × Pcya. Most of the variants are Lake Victoria specific, mostly for the allele of striped Lake Victoria species. None of the variants shows perfect association across all lakes, not even across haplochromines of Lake Victoria and Malawi strongly suggesting different regulatory mechanisms. Abbreviations: Horh, Harpagochromis sp. orange rock hunter; Hpar, Haplochromis paropius; Hvon, Haplochromis vonlinnei; Pdem, Pseudotropheus demasoni; Onil, Oreochromis niloticus (all other abbreviations are listed in Fig. S.I.11)

– 162 –

⁕ Supplementary Material ⁕

Fig. S.I.13 Summary of shared and non-shared variants in enh.a of Lake Victoria and Malawi cichlids. Summary of the variants that differ between the parents of the hybrid crosses (blue: Lake Malawi; yellow: Lake Victoria) and that are associated to the stripe phenotype across the sequenced species of Lake Malawi (purple) and Victoria (orange). None of the variants are shared across all analyzed haplochromine species from Lake Malawi and Lake Victoria (red zero).

Fig. S.I.14 Skin expression in Pdem (Pseudotropheus demasoni) and Pcya (Pseudotropheus cyaneorhabdos). Similarly, as shown for other Malawi species, agrp2 is differentially expressed between Pdem and Pcya.

– 163 –

⁕ Supplementary Material ⁕

Fig. S.I.15 Association between agrp2 locus genotype and stripe phenotype as quantitative measurement. (A) To obtain a quantitative measurement we measured the percentage of the midlateral stripe that is covered by pigmentation. (Please note that this measurement is affected by the vertical bar pattern that crosses the stripes, why we mainly treated stripes as ordinal trait) Using this quantitative measure, we obtained significant differences between all genotypes (Kruskal- Wallis rank sum test, P < 0.001; Dunn’s test: C/C - D/D: P < 0.001; C/C - C/D: P < 0.001; C/D - D/D: P < 0.05). Variances of untransformed and log2-transformed values differed between groups (Levene’s test, P < 0.001), with variances being significantly different between C/C and D/D (Levene’s test; P < 0.001), C/C and C/D (Levene’s test; P < 0.001) but not C/D and D/D (Levene’s test; P > 0.5). (B to F) Summary statistics for the analysis of stripes as ordinal trait – this panel is the same as Fig. 4A (B) including genotype frequencies (C), allele frequencies (F) and odds ratios (D and E). Odds ratio for C/C–D/D in E cannot be calculated (NA) because no individuals without stripes exists for C/C (therefore division by zero). Abbreviations: C/C, Homozygous for Pcya agrp2 allele, D/D, homozygous for Pdem agrp2 allele; C/D, heterozygous

– 164 –

⁕ Supplementary Material ⁕

Table S.I.1 Efficiency of gRNAs: Abbreviations: Seq. Clones Total clones sequenced, WT: Wildtype, Ins: Insertion, Del: Deletion, FS/SC: Frameshift or stop codon induced

gRNA 01 (Exon 1) gRNA 02 (Exon 1) Target Sequence GAAGCACAACAAGTGTTTGCCGG GGACACCAAGAAGGATGTGCGG Individual Seq. clones WT Ins Del FS/SC Seq. clones WT Ins Del FS/SC Crisp_mt_01 - - - - - 15 0 1 14 11 Crisp_mt_02 - - - - - NA NA NA NA NA Crisp_mt_03 10 2 1 7 6 - - - - - Crisp_mt_04 9 0 9 0 9 - - - - - Percentage mutated 89% 100% Percentage FS/SC: 79% 73% gRNA 03 (Exon 2) gRNA 04 (Exon 2) Target Sequence GGAGCCAAGCGAAGAATAGGCGG GAATAGGCGGCTGTTTGCAAGG Individual Seq. clones WT Ins Del FS/SC Seq. clones WT Ins Del FS/SC Crisp_mt_01 - - - - - 11 0 0 8 3 Crisp_mt_02 - - - - - 20 0 0 12 8 Crisp_mt_03 5 0 0 5 4 - - - - - Crisp_mt_04 8 0 0 8 8 - - - - - Percentage mutated 100% 100% Percentage FS/SC: 92% 35% gRNA 05 (Exon 3) Total Target Sequence GGTCGCAGCACAGCAGGTGAGGG Individual Seq. clones WT Ins Del FS/SC Seq. clones WT Ins Del FS/SC Crisp_mt_01 - - - - - 26 0 1 22 14 Crisp_mt_02 - - - - - 20 0 0 12 8 Crisp_mt_03 5 0 3 2 5 20 2 4 14 15 Crisp_mt_04 6 0 0 6 3 23 0 9 14 20 Percentage mutated 100% 98% Percentage FS/SC: 73% 64%

– 165 –

⁕ Supplementary Material ⁕

Table S.I.2 Association tests for agrp2 Locus in Lake Malawi hybrid cross. Allelic Dominant Recessive Fisher's exact test (C vs. D) (CD and CC vs DD) (CC vs. CD and DD) Binary (Stripes = 1; No stripes or interrupted stripes = 0) 5.59 × 10-9 1.14 × 10-7 4.45 × 10-6 Binary (Stripes or interrupted stripes = 1; No stripes = 0) 1.03 × 10-26 1.55 × 10-34 4.11 × 10-8

Table S.I. 3 Pseudo R2 calculations for Malawi hybrid cross (* above 0.5)

Model CoxSnell Nagelkerke Pseudo R2 Pseudo R2 Ordered Logistic Regression (additive model) 0.527* 0.605* Logistic Regression striped vs. not striped (additive model) 0.128 0.212 Logistic Regression full striped vs. no full stripe (additive model) 0.651* 0.886* Ordered Logistic Regression (dominant model; CD and CC vs DD) 0.499 0.572* Logistic Regression striped vs. not striped (dominant model) 0.076 0.126 Logistic Regression full striped vs. no full stripe (dominant model) 0.639* 0.869* Ordered Logistic Regression (recessive model; CC vs. CD and DD) 0.158 0.181 Logistic Regression striped vs. not striped (recessive model) 0.092 0.152 Logistic Regression full striped vs. no full stripe (recessive model) 0.127 0.172

Table S.I.4 R2 calculations for Malawi hybrid cross Multiple R2 Adjusted R2 p-value Linear Regression (additive model) 0.326 0.321 < 2.2 × 10-16 Linear Regression (dominant model) 0.304 0.301 < 2.2 × 10-16 Linear Regression (recessive model) 0.098 0.095 1.49 × 10-7

– 166 –

⁕ Supplementary Material ⁕

Table S.I. 5 List of primers used in this study ID Locus & fwd/rev Internal name Sequence Subproject P001 nd2_fwd DH_98_GLN CTACCTGAAGAGATCAAAAC Phylogeny_ND2 P002 nd2_rev DH_97_ASN CGCGTTTAGCTGTTAACTAA Phylogeny_ND2 P003 nd2_fwd JG_005 TAGCCCCTCTTCATGCTTGA Phylogeny_ND2 P004 nd2_rev JG_007 CTGAAAGGGCAGCTAGGGTT Phylogeny_ND2 P005 MS03_fwd LSm03_Hex_F1 CTTTGCCACAGACACAGCAT Finemapping_Microsats P006 MS03_rev LSm03_Hex_R1 CTGGGAGTCACAGGAATGGT Finemapping_Microsats P007 MS05_fwd CK235 GGGTAAGCCTTTTGTATC Finemapping_Microsats P008 MS05_rev CK236 CTCCACTGCTTGTTAGTT Finemapping_Microsats P009 MS06_fwd CK237 ATTCTGTTTGAGTCGCTG Finemapping_Microsats P010 MS06_rev CK238 TTTAATTATGAAGCCGCCTG Finemapping_Microsats P011 MS07_fwd LSm07_Hex_F1 GATGTCTCACAGCCATGCAT Finemapping_Microsats P012 MS07_rev LSm07_Hex_R1 TCGCCGCTGTTTATTGGTTT Finemapping_Microsats P013 SNPs_5_fwd01 CK261 TATACACGAGACACAGACCA Finemapping_SNPs P014 SNPs_5_rev01 CK260 GGGAATTGCTGGATTATGT Finemapping_SNPs P015 SNPs_5_seq CK303 AAATCCTCCACTTACCAC Finemapping_SNPs P016 SNPs_3_fwd01 CK374 CTCTGGATTGTGTTTCCCT Finemapping_SNPs P017 SNPs_3_rev01 CK320 AGGTCCAAACTACACGTT Finemapping_SNPs P018 SNPs_3_seq CK319 ATTTGCAACAATTACCAGGC Finemapping_SNPs P031 Stripe_interval CK295_Agrp2_E1_F1b GCCAGCAGATGTACACAAA Sequencing_Interval P032 Stripe_interval CK296_Agrp2_E3_F3 CTTTTCCACTTCTTCCACT Sequencing_Interval P033 Stripe_interval CK297_Agrp2_E3_R3 CACGTCACCCATCTTTTT Sequencing_Interval P034 Stripe_interval CK298_Agrp2_E3_F4 TTGTTGTGATTTGGCGCT Sequencing_Interval P035 Stripe_interval CK299_Agrp2_E3_R4 CTTTCCGTGCTCTTTTGT Sequencing_Interval P036 Stripe_interval CK300_Agrp2_E3_R3b CTGTGTACCGCATTTACT Sequencing_Interval P037 Stripe_interval CK301_Agrp2_E0_F1 CAGAGAAGCACAAAAGCA Sequencing_Interval P038 Stripe_interval CK302_Agrp2_E0_R1 GCCCACCTCAAAAGAATA Sequencing_Interval P039 Stripe_interval CK303_SI01_F AAATCCTCCACTTACCAC Sequencing_Interval P040 Stripe_interval CK304_SI01_R TCACATTCTCCCTTTTCC Sequencing_Interval P041 Stripe_interval CK305_SI03_F ACACCAACACAGCATCAC Sequencing_Interval P042 Stripe_interval CK306_SI03_R AATGGAAAGCAAGCAGCAA Sequencing_Interval P043 Stripe_interval CK307_SI04_F GGAGCCAAGCGAAGAATA Sequencing_Interval P044 Stripe_interval CK308_SI04_R AAAGGTGACTGGAAAGGA Sequencing_Interval P045 Stripe_interval CK309_SI05_F AAGTAGGGGTGATGTGAG Sequencing_Interval P046 Stripe_interval CK310_SI05_R CCCGAGCCAAAAAAAGAA Sequencing_Interval P047 Stripe_interval CK311_SI06_F TTTTGAGGCGGTAGAGTT Sequencing_Interval P048 Stripe_interval CK312_SI06_R TTTGGTGTCCAGTGCATT Sequencing_Interval P049 Stripe_interval CK313_SIseq01 CTCAGGGCAAGTTTGAAATC Sequencing_Interval P050 Stripe_interval CK314_SIseq02 CTAAAGGAGGAGAAATCTGTG Sequencing_Interval P051 Stripe_interval CK315_SIseq03 TCACAAGCATTCACAAGCA Sequencing_Interval P052 Stripe_interval CK316_SIseq04 AGGGAATTCTACAAGGAACAG Sequencing_Interval P053 Stripe_interval CK317_SIseq05 GAGCCATCTAGCATATTAG Sequencing_Interval P054 Stripe_interval CK318_SIseq06 GCCATGACCTAAAGTGTTT Sequencing_Interval P055 Stripe_interval CK319_SIseq07 ATTTGCAACAATTACCAGGC Sequencing_Interval P056 Stripe_interval CK320_SIseq08 AGGTCCAAACTACACGTT Sequencing_Interval P057 Stripe_interval CK321_SIseq09 CCCAAAATCAACCTGTCAAAGA Sequencing_Interval P058 Stripe_interval CK322_SIseq10 GCCTCTCCATGTTTCACA Sequencing_Interval P059 Stripe_interval CK323_SIseq11 AGGTGTGTACCAGGATGT Sequencing_Interval P060 Stripe_interval CK324_SIseq12 TTAAAACGGTGCCTTACG Sequencing_Interval P061 Stripe_interval CK325_SIseq13 AATGGATGGGAACAATGTG Sequencing_Interval P062 Stripe_interval CK326_SIseq14 ACTTTAGTTGTCGTCTGC Sequencing_Interval P063 Stripe_interval CK327_SIseq15 TGTGGAAGAGAGACAGAGATT Sequencing_Interval P064 Stripe_interval CK328_SIseq16 TATATGTTCAGGCGTTGTG Sequencing_Interval P065 Stripe_interval CK329_SIseq17 CCATCAGCAAAGCAACAGT Sequencing_Interval P066 Stripe_interval CK330_SIseq18 CACAGAGAAGTGGGAGCA Sequencing_Interval P067 Stripe_interval CK331_SI07_F TACACCAACACAGCATCAC Sequencing_Interval P068 Stripe_interval CK332_SI07_R CTAACAAACCGAAAAAGGCA Sequencing_Interval P069 Stripe_interval CK333_SI08_R ACAAAAGAGCACGGAAAGA Sequencing_Interval P070 Stripe_interval CK334_SI08_R AGCACGGGTGAAAAAAGTAA Sequencing_Interval – 167 –

⁕ Supplementary Material ⁕

P071 Stripe_interval CK335_SI09_R TTCTCCTTGCATTTCTCC Sequencing_Interval P072 Stripe_interval CK336_SI09_R CTCTGACTCGTGTGTATT Sequencing_Interval P073 Stripe_interval CK337_SIseq19 GTCTAAGTCTGGCATCTGT Sequencing_Interval P074 Stripe_interval CK338_SIseq20 GGAGTAGACAAGCTGAAA Sequencing_Interval P075 Stripe_interval CK339_SIseq21 GGGGCGCCAAAATAATAAA Sequencing_Interval P076 Stripe_interval CK340_SIseq22 AAGGAAAAGGGTTAGCGT Sequencing_Interval P077 Stripe_interval CK341_SIseq23 CTCAGTGTGTGGATATCTT Sequencing_Interval P078 Stripe_interval CK342_SIseq24 GCAGAGAAGGTAGCCAAC Sequencing_Interval P079 Stripe_interval CK343_SIseq25 ATCGGCCCTTAGATTATGA Sequencing_Interval P080 Stripe_interval CK344_SIseq26 AGTCTAGTAGGAGGAAAGA Sequencing_Interval P081 Stripe_interval CK345_SIseq27 TTTCAGTGACACGCCCTT Sequencing_Interval P082 Stripe_interval CK346_SIseq28 CTTGAGACTGTGTTTTAGCA Sequencing_Interval P083 Stripe_interval CK347_SIseq29 GGCAATTGTGGAAAGTGTATC Sequencing_Interval P084 Stripe_interval CK348_SIseq30 TTGTATGTGAGGAGCAGG Sequencing_Interval P085 Stripe_interval CK349_SIseq31 ATGGAGTTTCACAGGGTT Sequencing_Interval P086 Stripe_interval CK350_SIseq32 GTACTATGAGGTCTTTAGG Sequencing_Interval P087 Stripe_interval CK351_SIseq33 TGACCTGTATTTGTATAGCG Sequencing_Interval P088 Stripe_interval CK352_SIseq34 ATCCAAAACCTAGTGAGG Sequencing_Interval P089 Stripe_interval CK353_SIseq35 TGGGCCATACCAATTTTG Sequencing_Interval P090 Stripe_interval CK354_SIseq36 CGTCAGGAGAAATACTTAG Sequencing_Interval P091 Stripe_interval CK355_SIseq37 TAAAAAGTCCGCTTCACTG Sequencing_Interval P092 Stripe_interval CK356_SIseq38 TCCACTGCTTGTTAGTTTGC Sequencing_Interval P093 Stripe_interval CK357_SIseq39 TTTGCCATGTCCTCTGCT Sequencing_Interval P094 Stripe_interval CK358_SIseq40 AACTATGAAGCGGGCACA Sequencing_Interval P095 Stripe_interval CK359_SIseq41 ATGAAGTAATGGGTGTGAGC Sequencing_Interval P096 Stripe_interval CK360_SIseq42 CTGTGTGCTTTTATGCGTG Sequencing_Interval P097 Stripe_interval CK361_SIseq43 GGGCAGCCTAACAAGAAAA Sequencing_Interval P098 Stripe_interval CK362_SIseq44 GATCGCTTACGCCTCAAA Sequencing_Interval P099 Stripe_interval CK363_SIseq45 TGCTGTGTGGGTTTTTTTGT Sequencing_Interval P100 Stripe_interval CK364_SIseq46 CTCCCTTACCTCAGTCTC Sequencing_Interval P101 Stripe_interval CK365_SIseq47 GGAGAATGAGAAAGGCGA Sequencing_Interval P102 Stripe_interval CK366_SIseq48 GTGTTGTGTGTGTATATCC Sequencing_Interval P103 Stripe_interval CK367_SIseq49 TCTAATTTGTTTCTGCTGCC Sequencing_Interval P104 Stripe_interval CK368_SIseq50 TTTAAGACTTGCTGAGGAC Sequencing_Interval P105 Stripe_interval CK369_SIseq51 GAGAAAGGTGTATATGTGTG Sequencing_Interval P106 Stripe_interval CK370_SIseq52 GGATGTAGAATGAGATGAGGA Sequencing_Interval P107 Stripe_interval CK371_SIseq53 CAAAAACAAAACCCCCCA Sequencing_Interval P108 Stripe_interval CK372_SIseq54 CTCCATGCTCATTACTAC Sequencing_Interval P109 Stripe_interval CK373_SIseq55 GCACAGAAGCAAAAACTAGA Sequencing_Interval P110 Stripe_interval CK374_SIseq56 CTCTGGATTGTGTTTCCCT Sequencing_Interval P111 Stripe_interval CK375_SIseq57 GCACAGTCCATACAAAGAA Sequencing_Interval P112 Stripe_interval CK376_SIseq58 GGTGATGTGAAAGGAGTAG Sequencing_Interval P113 Stripe_interval CK377_SIseq59 CCTCACCCATTGCCTACT Sequencing_Interval P114 Stripe_interval CK378_SIseq60 GAAAGAAAAGACGATGGAGA Sequencing_Interval P115 Stripe_interval CK379_SIseq61 GCAACTACCTCTGAATACT Sequencing_Interval P116 Stripe_interval CK380_SIseq62 CTTACTCTGGGTGGCATT Sequencing_Interval P117 Stripe_interval CK381_SIseq63 CGGCAGCCTATGTCTATT Sequencing_Interval P118 Stripe_interval CK382_SIseq64 TAGGCACATGGCGGAAAA Sequencing_Interval P119 Stripe_interval CK383_SIseq65 AATAAGTAGAGCAGGGCA Sequencing_Interval P120 Stripe_interval CK384_SIseq66 TCCTCGATTATTCAGTCCTT Sequencing_Interval P121 Stripe_interval CK385_SIseq67 AGCTCCTCCTGATACAAAT Sequencing_Interval P122 Stripe_interval CK386_SIseq68 GCAGCATCTTAAAGGACA Sequencing_Interval P123 Stripe_interval CK387_SIseq69 GCCAGCAGATGTACACAA Sequencing_Interval P124 Stripe_interval CK388_SIseq70 GATTCTTTGTATGGACTGTG Sequencing_Interval P125 Stripe_interval CK389_SIseq71 CACGCAAACACATCACAA Sequencing_Interval P126 Stripe_interval CK390_SIseq72 TTTGAATATCTTGCTGCTGG Sequencing_Interval P127 Stripe_interval CK391_SIseq73 CGAGGAGATGCACAGAGA Sequencing_Interval P128 Stripe_interval CK392_SIseq74 TTGTGGTTTGGTTTATGTGG Sequencing_Interval P129 Stripe_interval CK393_SIseq75 TAGAGAAAGTGGGTGAATGTG Sequencing_Interval P130 Stripe_interval CK394_SIseq76 TTACTTTTTTCACCCGTGCT Sequencing_Interval

– 168 –

⁕ Supplementary Material ⁕

P131 Stripe_interval CK395_SIseq77 GAACGTTAGTCAGAGAGG Sequencing_Interval P132 Stripe_interval CK405_SIseq78 TTCCCTGCGGATCTTTTCAC Sequencing_Interval P133 Stripe_interval CK406_SIseq79 TCCACAAGGATCAAATTTCCC Sequencing_Interval P134 Stripe_interval CK407_SIseq80 GGAGTTCCAAACCAGATGAA Sequencing_Interval P135 Stripe_interval CK408_SIseq81 TGAGGATTTAGGGTGACC Sequencing_Interval P136 Stripe_interval CK409_SIseq82 TGGACGGATAAAACTGGGTT Sequencing_Interval P137 Stripe_interval CK410_SIseq83 CCGGACAAGACACTGAACTTA Sequencing_Interval P138 Stripe_interval CK411_SIseq84 GAGGAATGAAAGCTGGGA Sequencing_Interval P139 Stripe_interval CK412_SIseq85 GATACAAAAGGCTTACCC Sequencing_Interval P140 Stripe_interval CK413_SIseq86 TCCCAGCTTTCATTCCTC Sequencing_Interval P141 Stripe_interval CK414_SIseq87 CCCCATCAGGGTCTTTATT Sequencing_Interval P142 Stripe_interval CK415_SIseq88 CCGCTTGTTTGTGACTTT Sequencing_Interval P143 Stripe_interval CK416_SIseq89 CAGCAGAACTACATACCCA Sequencing_Interval P144 Stripe_interval CK417_SIseq90 GGGTATGTAGTTCTGCTG Sequencing_Interval P145 Stripe_interval CK418_SIseq91 TCCCTCCTCACAATGCAC Sequencing_Interval P146 Stripe_interval CK419_SIseq92 ACCCAAACAGACAAACCA Sequencing_Interval P147 Stripe_interval CK420_SIseq93 AAAATGACTCCAAGGTTCC Sequencing_Interval P148 Stripe_interval CK421_SIseq94 GTTTAGGCTTTGTCAGTTCT Sequencing_Interval P149 Stripe_interval CK422_SIseq95 CTGATTTCTCCCAGGTGTG Sequencing_Interval P150 Stripe_interval CK445_SLseq96 GGAACCTTGGAGTCATTT Sequencing_Interval P151 Stripe_interval CK446_SLseq97 GGCCAGACCACTTAAGAT Sequencing_Interval P152 Stripe_interval CK447_SLseq98 AAAGCAGGAGTGAGGGAA Sequencing_Interval P153 Stripe_interval CK448_SLseq99 AACAAGCCAAAGAAAACCCA Sequencing_Interval P154 Stripe_interval CK449_SLseq100 CTACCTTCTCTGCCATTT Sequencing_Interval P155 Stripe_interval CK450_SLseq101 AAGCGTGTTGGTATGTGA Sequencing_Interval P156 Stripe_interval CK451_SLseq102 ACACTGAGAAATTACGGGA Sequencing_Interval P157 Stripe_interval CK452_SLseq103 CCACTTACCACAAACTAAC Sequencing_Interval P158 Stripe_interval CK453_SLseq104 TCCTCATCTCATTCTACATCC Sequencing_Interval P159 Stripe_interval CK454_SLseq105 TCCTCATCTCATTCTACATCC Sequencing_Interval P160 Stripe_interval CK455_SLseq106 TGTGTGAATGTGTGTGTG Sequencing_Interval P161 Stripe_interval CK456_SLseq107 GAGGAGGAAGTGTCTGTGT Sequencing_Interval P162 Stripe_interval CK457_SLseq108 GATGTCAGCCTAATGTGT Sequencing_Interval P163 Stripe_interval CK458_SLseq109 TGGTTTGTCTGTTTGGGT Sequencing_Interval P164 Stripe_interval CK459_SLseq110 TTAATCTCTGGCTCTTGTC Sequencing_Interval P165 Stripe_interval CK460_SLseq111 CCAATAATAATCCGCAGACA Sequencing_Interval P166 Stripe_interval CK462_SLseq112 CTCCACCCTCTCTTGAAT Sequencing_Interval P167 Stripe_interval CK463_SLseq113 ATAGATTGCTTCCAACACC Sequencing_Interval P168 Stripe_interval CK461_SLseq114 AAGGGAAGGAATGTGAAG Sequencing_Interval P169 Stripe_interval CK464_SLseq115 TGTTATGTGCTGTAGGGT Sequencing_Interval P170 Stripe_interval CK465_SLseq116 TATCTCATGCTCCCGTTT Sequencing_Interval P171 Stripe_interval CK466_SLseq117 CTTCACATTCCTTCCCTT Sequencing_Interval P172 Stripe_interval CK467_SLseq118 CTTTTTACCTACGCCTCC Sequencing_Interval P173 Stripe_interval CK468_SLseq119 TTTTGAGAACTGGACCGA Sequencing_Interval P174 Stripe_interval CK469_SLseq120 AAAGAAACGGGAGCATGA Sequencing_Interval P175 Stripe_interval CK470_SLseq121 GCCAGGTGTAGAGAAAAT Sequencing_Interval P176 Stripe_interval CK471_SLseq122 CCCAATGAGCTGAAGCAA Sequencing_Interval P177 Stripe_interval CK472_SLseq123 ACGGATCTCAGGCTAAAA Sequencing_Interval P178 Stripe_interval CK473_SLseq124 GCAGGTAGGATCAGTGAGA Sequencing_Interval P179 Stripe_interval CK474_SLseq125 TGCCCTGCTCTACTTATT Sequencing_Interval P180 Stripe_interval CK475_SLseq126 TGTTTTTCAGTCGTTCCATC Sequencing_Interval P181 Stripe_interval CK476_SLseq127 AAACTTTAGCTGTCAGACC Sequencing_Interval P182 Stripe_interval CK477_SLseq128 CGCAGTGCCTTTGTTTTC Sequencing_Interval P183 Stripe_interval CK478_SLseq129 GCAGCGTTTTTGCATGTT Sequencing_Interval P184 Stripe_interval CK479_SLseq130 ACAACTACATACCCATCTTCC Sequencing_Interval P185 Stripe_interval CK480_SLseq131 TTTGCTTTCGCTCACCTC Sequencing_Interval P186 Stripe_interval CK481_SLseq132 GTGCCACAAAGGACGTAA Sequencing_Interval P187 Stripe_interval CK482_SLseq133 ACAGAGGCCTAAAACATGAA Sequencing_Interval P188 Stripe_interval CK483_SLseq134 TGGAGGTACTCTGGCTTT Sequencing_Interval P189 Stripe_interval CK484_SLseq135 GCTGTTAAATCCTCCTAATGTG Sequencing_Interval P190 Stripe_interval CK485_SLseq136 GAAAGCCAGAGTACCTCCA Sequencing_Interval

– 169 –

⁕ Supplementary Material ⁕

P191 Stripe_interval CK486_SLseq137 ACTTCTCTGGCTCTATGTT Sequencing_Interval P192 Stripe_interval CK487_SLseq138 ACAGCTACATGATCGACAA Sequencing_Interval P193 Stripe_interval CK488_SLseq139 ACCAAAAAGACACCACAG Sequencing_Interval P194 Stripe_interval CK489_SLseq140 GGCCCTGAAGAACAAAGA Sequencing_Interval P195 Stripe_interval CK490_SLseq141 GAGGAATGAAAGCTGGGA Sequencing_Interval P196 Stripe_interval CK491_SLseq142 ACATCTTTGTTCTTCAGGG Sequencing_Interval P197 Stripe_interval CK492_SLseq143 CAGTGAAGCGGACTTTTT Sequencing_Interval P198 Stripe_interval CK493_SLseq144 TTACTTTTTTCACCCGTGCT Sequencing_Interval P199 Stripe_interval CK494_SLseq145 ACGTCAGGAGAAATACTTAG Sequencing_Interval P200 Stripe_interval CK495_SLseq146 TTTTCCTTCCTCTTTGACAG Sequencing_Interval P201 Stripe_interval CK496_SLseq147 TCAGGAGGTAGAGCAGGT Sequencing_Interval P202 Stripe_interval CK497_SLseq148 TTGCTTTTGCCTTCCTGT Sequencing_Interval P203 Stripe_interval CK498_SLseq149 ACAGTGACTTAACCAGCA Sequencing_Interval P204 Stripe_interval CK499_SLseq150 CACTCGCAAACCTGTCAA Sequencing_Interval P205 Stripe_interval CK500_SLseq151 TGGACCACACATTTCCTG Sequencing_Interval P206 Stripe_interval CK501_SLseq152 AGCTGAGCCTCCAAAAGA Sequencing_Interval P207 Stripe_interval CK502_SLseq153 GTCTGGCCCACTTTAGAT Sequencing_Interval P208 Stripe_interval CK503_SLseq154 TCCTTTCCAGTCACCTTT Sequencing_Interval P209 Stripe_interval CK504_SLseq155 GGCCAGACCACTTAAGAT Sequencing_Interval P210 Stripe_interval CK505_SLseq156 CGCTAACTTTCTGCTTCT Sequencing_Interval P211 Stripe_interval CK506_SLseq157 CTGTCTCCCAATCTAAACT Sequencing_Interval P212 Stripe_interval CK507_SLseq158 AATCTAAACTGGAAGCTGG Sequencing_Interval P213 Stripe_interval CK508_SLseq159 GTGTCCCATCCCATGATT Sequencing_Interval P214 Stripe_interval CK509_SLseq160 GTGTTTTCACATCAGCCTA Sequencing_Interval P215 Stripe_interval CK510_SLseq161 CAAAACAAAACACCTCAGTG Sequencing_Interval P216 Stripe_interval CK511_SLseq162 CACACACACATTCACACAC Sequencing_Interval P217 Stripe_interval CK512_SLseq163 GAGTTAGGCCACTGAATA Sequencing_Interval P218 Stripe_interval CK513_SLseq164 ATCCAGCAATTCCCATCC Sequencing_Interval P219 Stripe_interval CK514_SLseq165 ACAACTTTGGACTTCAGG Sequencing_Interval P220 Stripe_interval CK515_SLseq166 CGCAGGTTGATACACTTT Sequencing_Interval P221 Stripe_interval CK516_SLseq167 GTATCTAAACGACCCAGT Sequencing_Interval P222 Stripe_interval CK517_SLseq168 AAACTGTAGAAAGAGGACAC Sequencing_Interval P223 Stripe_interval CK518_SLseq169 AACTATTTGACCCCCCGA Sequencing_Interval P224 Stripe_interval CK519_SLseq170 GGCAATTGTGGAAAGTGT Sequencing_Interval P225 Stripe_interval CK520_SLseq171 AGTGGAAGAAGTGGAAAAG Sequencing_Interval P226 agrp2_Fwd agrp2_Fwd GCGAAGAATAGGCGGCTGTTTG qPCR P227 agrp2_Rev agrp2_Rev CGACGCGCCGGAGTTACGAG qPCR P228 agrp2_Ncyl_Rev agrp2_Ncyl_Rev CGACGTGCCGGAGTTATGAG qPCR P229 agrp2_Rev_deg agrp2_Rev_de CGACGYGCCGGAGTTAYGAG qPCR P230 agrp1_Fwd agrp1_Fwd CCTGCCCTCCTCCCTGTGG qPCR P231 agrp1_Rev agrp1_Rev GCATGGCCCGACCCTGTAG qPCR P232 asip1_Fwd asip1_Fwd GAAGAGCAAGAAACCAAAGAA qPCR P233 asip1_Rev asip1_Rev ACTGGCAGAAAGCACAATAATACA qPCR P234 asip2_Fwd asip2_Fwd GTTTGCCAGGAGCACAGGACATTAT qPCR P235 asip2_Rev asip2_Rev GGATTTGGCCTCACTTTAGCAG qPCR P236 atp6v0d2_Fwd atp6v0d2_Fwd GACTCTTGAGGACCGATTCT qPCR P237 atp6v0d2_Rev atp6v0d2_Rev CCTGCTCCTTAAGTTTGATGTA qPCR P238 unk_Fwd unk_Fwd TACACCTGGCAGATGAAACCTA qPCR P239 unk_Rev unk_Rev GGGGAGAATCGGGAGAAGAGACA qPCR P240 β-actin_Fwd β-actin_Fwd TGACATGGAGAAGATCTGGCACA qPCR P241 β-actin_Rev β-actin_Rev TGGCAGGAGTGTTGAAGGT qPCR P242 gapdh_Fwd gapdh_Fwd CACACAAGCCCAACCCATAGTC qPCR P243 gapdh_Rev gapdh_Rev AAACACACTGCTGCTGCCTACAAT qPCR P244 rpl8_Fwd rpl8_Fwd CCACACCGGACAGTTTA CATCTAC qPCR P245 rpl8_Rev rpl8_Rev GGTCACCGGGCTTCTCCTCA qPCR P246 ef1a_Fwd ef1a_Fwd CAGTCCCCGTAGGTCGTGTGG qPCR P247 ef1a_Rev ef1a_Rev CGGGTAGGGCTTCGCTCAGT qPCR P248 b2m_Fwd b2m_Fwd AGGCGAAATACTCTCCACCAA qPCR P249 b2m_Rev b2m_Rev AATTTCTGCCCCATCCTTCA qPCR P250 Stripe_interval CK555_Agrp2Enh_Fwd CCCAAAACAAAACACCTC Sequencing_Interval

– 170 –

⁕ Supplementary Material ⁕

P251 Stripe_interval CK556_Agrp2Enh_Rev GGGAATGGAAATAAAAGCAG Sequencing_Interval P252 lws_fwd CK_570 ATTGCTGCTCTTTGGTCCCTGAC Phylogeny_other P253 lws_rev CK_571 AGCCAGAGGGTGGAAGGCATA Phylogeny_other P254 mitfa_fwd CK_572 CCTGGCATGAAGCARGTACTGG Phylogeny_other P255 mitfa_rev CK_573 TTGCYAGAGCACGAACTTCRGCAC Phylogeny_other P256 mitfb_fwd mitf.uni.ex4.FOR ATGGAYGAYGTIATYGAYGACA Phylogeny_other P257 mitfb_rev mitf.ex5b.rev CCACAGGRTCCIGRYTTGTCCAT Phylogeny_other P258 rag1_fwd RAG1.Mart.F.L1 AGCTGCAGYCARTAYCAYAAR Phylogeny_other P259 rag1_rev RAG1.Mart.R6 GTGTAGAGCCARTGRTGYTTATGTA Phylogeny_other P260 rag1_fwd JG_001 CAGACGATGAGGAGGAGGAG Phylogeny_other P261 rag1_rev RAG1.Mart.R6 GTGTAGAGCCARTGRTGYTT Phylogeny_other P262 s7_fwd CK_576 CGTGCCATTTTACTCTGGACTK Phylogeny_other P263 s7_rev CK_577 AACTCGTCYGGCTTCTCGCCGC Phylogeny_other P264 Agrp2 agrp2_ISH_Fwd GAAAGCTGCTCCCCTCACC In_Situ_Probe P265 Agrp2 agrp2_ISH_Rev CTCCAGCAGCAAGATATTCAAA In_Situ_Probe P266 Stripe_interval Agrp2Enh_Fwd_XhoI GATCGctcgagCCCAAAACAAAACAG Reporter_assay P267 Stripe_interval Agrp2Enh_Rev_EcoRI GATCGgaattcGGGAATGGAAATAACCTC Reporter_assay P268 Crispr_general jw700gRNAgeneral AAAAGCACCAAAGCAGGACTCGGTGCCAC Crispr P269 Crispr_Agrp2 gRNA Hnye Agrp2- ATTTAGGTGACACTATAGAAGCTT Crispr P270 Crispr_Agrp3 gRNA1_Exon1_Sp6 Hnye Agrp2 - TAATACGACTCACTATAGTTTCAAGTTGATAACGGACTAGACAACAAGTGTTTGCGTTTTAGGACA Crispr P271 Crispr_Agrp4 gRNA2_Exon1_T7 Hnye Agrp2 - CCAAGAAGGATGTGGTTTTAGATAATACGACTCACTATAGGAGCCCTTATTTTAACTTGCTATTTCTAGCTAGAAATAGC Crispr P272 Crispr_Agrp5 gRNA3_Exon2_T7 Hnye Agrp2 - CAAGCGAAGAATAGGGTTTTAGATTTAGGTGACACTATAGAATAGCTAGAAATAGCAGCTCTAAAAC Crispr P273 Crispr_Agrp6 gRNA4_Exon2_Sp6 Hnye Agrp2 - TAATACGACTCACTATAGGTCGGGCGGCTGTTTGCAGTTTTAGAAGCTAGAAATAGC Crispr P277 Stripe_interval CK189_Agrp2_E1_F15_Exon3_T7 CAGCACAGCAGGTGAGTTTTAGCCCACCTCCTCTTTCACAGCTAGAAATAGC Amplifying_Agrp2 P278 Stripe_interval CK190_Agrp2_E1_R1 CCTCCCCCCATAACGATAGCTAGAAATAGC T Amplifying_Agrp2 P279 Stripe_interval CK195_Agrp2_E2_F2 TGCTCTCGCTGTTGGTTA Amplifying_Agrp2 P280 Stripe_interval CK196_Agrp2_E2_R2 ATGTGTTCACTGCTGTCT Amplifying_Agrp2 P283 Stripe_interval CK198_Agrp2_E3_R1 TTTCCATACTTTTCGTCGCT Amplifying_Agrp2 P284 Stripe_interval CK199_Agrp2_E3_F2 CGTGTTTCAGTGAGTGAG Amplifying_Agrp2 P274 pGEM CK071_Sequencing_PCR1 CGGGCCTCTTCGCTATT Sequencing_Crispr P275 pGEM CK072_Sequencing_PCR2 TTAGCTCACTCATTAGG Sequencing_Crispr P276 pGEM CK073_Sequencing_UP-40 GTTTTCCCAGTCACGAC Sequencing_Crispr

– 171 –

⁕ Supplementary Material ⁕

Chapter II

Fig. S.II.1 5’ RACE results. (A) Positions of used primers in Exon 3. (B) The two isoforms and primer used to confirm the two isoforms. F1 is specific for isoform 1, F2 specific for isoform 2. F3 amplifies both isoforms. (C) 5’ RACE amplification. Arrowheads indicate the expected size. (D) Realtime PCRs amplifying both isoforms. Agrp2 is strongly expressed in the brain and shows differential expression between the skin of P. nyererei (arrow) and H. sauvagei (asterisk) as previously reported. (E, F) Realtime PCRs for Isoform 1 (E) and (F) show a similar pattern suggesting similar regulation of the expression of the transcript. Abbreviations: Br, brain; Sk, skin; Mu, muscle; Li, liver; Ey, eye

– 172 –

⁕ Supplementary Material ⁕

Fig. S.II.2 3’ RACE results. (A) Positions of primers used for the 3‘ RACE amplification. (B , C) 3’RACE amplifications using forward primers aF1 and aF2 (B) as well as aF2 and aF3 (C). Arrows indicate expected size. (D) 3’UTR sequence of P. nyererei and H. sauvagei. Sequences labeled in red indicate polyadenylation sequence motifs.

– 173 –

⁕ Supplementary Material ⁕

Fig. S.II.3 PCR-based detection of the agrp2 exon duplication using primers DF1 and DR1. (A) Primer Position and amplicons “PCR 1”, “PCR 1b” and “PCR 2”. Red boxes indicate the duplication. (B, C) PCR assay across 43 cichlid species. Bands (in most cases at ~1.6kB) suggest one or more tandem duplication. Red “X” indicate missing template, black arrowheads are bands shown in Fig. II.2. Asterisks in (C) indicate samples already tested in (B).

– 174 –

⁕ Supplementary Material ⁕

Fig. S.II.4 PCR-based detection of the agrp2 exon duplication using primers DF2 and DR2. Red “X” indicate missing template. Samples are the same as in supplementary Fig. II.3B. Bands (in most cases at ~1.1kB) suggest one or more tandem duplications.

Fig. S.II.5 Long-range-PCR-based detection of the agrp2 exon duplication using primers DF3 and DR3. Larger Bands (in most cases at ~6kB) suggest a tandem duplication, bands >6kB likely more duplications (e.g. in J. ornatus), bands around ~4kB no duplication. Multiple bands are likely individuals heterozygot for the duplication. Black arrowheads indicate bands shown in Fig. II.2.

– 175 –

⁕ Supplementary Material ⁕

Fig. S.II.6 Cladoglam of cichlids with a sequenced agrp2 locus and the likely origins of deletions and insertions. Question marks indicate equally likely scenarios.

Fig. S.II.7 PCR-based confirmation of deletion 7 and 8 across Lake Victoria cichlids. (A) PCR design. (B, C) PCR for deletion 7 (B) and 8 (C). In both cases only the individual of H. chilotes and H. sauvagei show the deletion, suggesting no consistent link to stripe patterns.

– 176 –

⁕ Supplementary Material ⁕

Table S.II.1 Position of deletions, the coding sequence (CDS) of agrp2 and a

regulatory element (EnhA) of agrp2.

167

472

1533

647

DEL

308

776

260

1534

167

108

(Length)

O. niloticus O. niloticus

.1:9,187

9,197,793

9,193,485

9,192,659

9,192,026

9,189,619

9,188,848

9,188,051

9,187,797

9,192,660

9,197,793

9,192,734

9,188,523

alignable

,626

NC_031982.1:9,197

non

,013

NC_031982.1:9,193

,126

NC_031982.1:9,191

,379

NC_031982.1:9,191

,191^9,191,192

NC_031982.1:9,191

,083^9,190,084

NC_031982.1:9,190

,521

NC_031982.1:9,189

,540

NC_031982.1:9,188

,275

NC_031982.1:9,187

,537

NC_031982

,126

NC_031982.1:9,191

,626

NC_031982.1:9,197

,661

NC_031982.1:9,192

,415

NC_031982.1:9,188

(Position; + (Position; Strand) +

O. niloticus O. niloticus

485

3387

464

DEL

728

305

DEL

927

167

108

(Length)

N. brichardi N. brichardi

8,531,101

8,532,123

8,534,505

8,536,550

8,538,953

8,540,139

8,541,080

8,537,831

8,531,940

8,536,904

8,541,205

616

JH422283.1:8,530,

068

JH422283.1:8,532,

118

JH422283.1:8,531,

086

JH422283.1:8,536,

566^8,537,567

JH422283.1:8,537,

757^8,537,758

JH422283.1:8,537,

912

JH422283.1:8,538,

912

JH422283.1:8,538,

411

JH422283.1:8,539,

775

JH422283.1:8,540,

543^8,541,544

JH422283.1:8,541,

543^8,541,544

JH422283.1:8,541,

904

JH422283.1:8,536,

773

JH422283.1:8,531,

830

JH422283.1:8,536,

097

JH422283.1:8,541,

Strand)

(Position; (Position;

N. brichardi N. brichardi

DEL

1800

966

450

DEL

143

2078

1686

856

298

499

DEL

1090

167

108

(Length)

A. calliptera A. calliptera

347,217

18:24,347,216^24,

24,349,981

18:24,348,181

24,352,554

18:24,351,588

24,354,490

18:24,354,040

24,355,523

18:24,355,522

24,355,857

18:24,355,714

24,359,088

18:24,357,010

24,358,696

18:24,357,010

24,360,398

18:24,359,542

24,361,332

18:24,361,034

24,362,306

18:24,361,807

362,056

18:24,362,055^24,

24,355,932

18:24,354,842

24,348,052

18:24,347,885

24,349,795

18:24,349,628

24,354,841

18:24,354,767

24,361,457

18:24,361,349

Strand)

(Position; (Position;

A. calliptera A. calliptera

DEL

1914

364

446

DEL

143

368

DEL

766

298

545

540

1089

167

108

(Length)

A. burtoni

108,

107,047

107,411

110,732

112,243

113,451

114,477

115,744

115,493

110,807

105,005

109,717

114,602

169^104,170

JH425581.1:104,

133

JH425581.1:105,

047

JH425581.1:107,

916^109,362

JH425581.1:

397^110,398

JH425581.1:110,

589

JH425581.1:110,

875

JH425581.1:111,

875^111,876

JH425581.1:111,

685

JH425581.1:112,

179

JH425581.1:114,

199

JH425581.1:115,

953

JH425581.1:114,

718

JH425581.1:109,

838

JH425581.1:104,

643

JH425581.1:109,

494

JH425581.1:114,

Strand)

(Position; (Position;

A. burtoni

DEL

1717

955

452

DEL

143

377

DEL

811

293

791

294

1088

167

108

(Length)

M. zebra M. zebra

19,329,197

19,331,648

19,333,586

19,334,951

19,336,484

19,337,749

19,338,723

19,339,989

19,339,737

19,335,026

19,327,351

19,329,012

19,333,937

19,338,848

6^19,326,517

LG18:19,326,51

LG18:19,327,48

LG18:19,330,69

LG18:19,333,13

6^19,334,617

LG18:19,334,61

LG18:19,334,80

LG18:19,336,10

7^19,336,108

LG18:19,336,10

LG18:19,336,93

LG18:19,338,43

LG18:19,339,19

LG18:19,339,44

LG18:19,333,93

LG18:19,327,18

LG18:19,328,84

LG18:19,333,86

LG18:19,338,74

Strand)

(Position; (Position;

M. zebra M. zebra

DEL

143

338

DEL

850

DEL

545

546

1077

167

108

(Length)

H. sauvagei

5,245

3,954

3,172

6,104

9,720

6,179

1,670

959

1211

10,349^10,350

9,423^9,424

7,922^7,923

6,458^6,459

5,436^5,437

5,102

3,616

3,952^3,953

2,322

1,686^1,687

414

665

5,027

9,553

6,105

1,562

Strand)

(Position; + (Position; +

H. sauvagei

DEL

936

450

DEL

143

338

DEL

850

DEL

797

294

1090

167

108

(Length)

P. nyererei P.nyererei

n; n; +

446,613

444,211

442,538

441,241

440,459

438,498

438,246

443,410

448,408

443,485

438,957

075^449,076

scaffold_3:449,

112^448,113

scaffold_3:448,

677

scaffold_3:445,

761

scaffold_3:443,

729^442,730

scaffold_3:442,

395

scaffold_3:442,

903

scaffold_3:440,

239^441,240

scaffold_3:441,

609

scaffold_3:439,

973^438,974

scaffold_3:438,

701

scaffold_3:437,

952

scaffold_3:437,

320

scaffold_3:442,

241

scaffold_3:448,

411

scaffold_3:443,

849

scaffold_3:438,

Strand)

P.nyererei

#10

(Duplication)

#9 #9

#4b

#4a

#1b

#1a

element

regulatory regulatory

EnhA

Exon 3b

Agrp2 Agrp2 CDS

Exon 3/3a

Agrp2 Agrp2 CDS

Exon 2

Agrp2 Agrp2 CDS

Exon 1

Agrp2 Agrp2 CDS Annotation

– 177 –

⁕ Supplementary Material ⁕

Table S.II. 2 Primers used in this study Internal Name Sequence Used for name aF1 aF1 GGCAGAGGACACCAAGAA 1st PCR(#1) of 3’RACE, 5'RACE sequencing 2nd PCR(#1) and 1st PCR(#2) of 3’RACE, aF2 aF2 GGAGCCAAGCGAAGAATA 5'RACE sequencing GCGAAGAATAGGCGGCTGTT aF3 aF3 3'RACE sequencing TG aF4 aF4 GAAAGCTGCTCCCCTCACC 2nd PCR(#2) of 3’RACE aF5 aF5 ACAAAAGAGCACGGAAAGA 3'RACE sequencing aR1 aR1 TCACATTCTCCCTTTTCC 5'RACE sequencing ACGCGCCGGAGTTACGAGGT 2nd PCR(#2) of 5’RACE, 5'RACE aR2 aR2 TC sequencing aR3 aR3 CGACGCGCCGGAGTTACGAG 2nd PCR(#1) and 1st PCR(#2) of 5’RACE GGGGGTTCATCCTCCAGCAG aR4 aR4 1st PCR(#1) and RT-PCR(#2) of 5’RACE TG aR5 aR5 CTTTCCGTGCTCTTTTGT RT-PCR(#1) of 5’RACE, 3'RACE sequencing aR6 aR6 TTTGAATATCTTGCTGCTGG 3'RACE sequencing CCAGTGAGCAGAGTGACGAG Qt Qtotal GACTCGAGCTCAAGCTTTTTT RT-PCR of 3’RACE and 1st PCR of 5’RACE TTTTTTTTTTT Qo Qouter CCAGTGAGCAGAGTGACG 1st PCR of 3’RACE and 5’RACE Qi Qinner GAGGACTCGAGCTCAAGC 2nd PCR of 3’RACE and 5’RACE F1 YPL058 GTTGTTGTTTTTTTTGCACC RACE result confirmation F2 YPL059 ACAGCCAGAGAGAAACCT RACE result confirmation F3 YPL060 AGGAAGATCACCGGCAAA RACE result confirmation R1-3 YPL061 TTAATCCACAAGAGGCAGG RACE result confirmation DF1 CK400 GAAAGCTGCTCCCCTCACC Duplication Test (PCR1) DR1 CK898 GGGGAGCAGCTTTCCATCAA Duplication Test (PCR1) DF2 CK390 TTTGAATATCTTGCTGCTGG Duplication Test (PCR1b) DR2 CK518 AACTATTTGACCCCCCGA Duplication Test (PCR1b) DF3 CK889 ACCCAAGATGCAGGCACTAC Duplication Test (PCR2) DR3 CK900 GCAGACAGTGTCCCAGACAA Duplication Test (PCR2) Agrp2F YPL350 TTGCACCAYTGTCATCAACA cDNA Amplification AATCCACAAGAGGCAGGCA Agrp2R YPL351 cDNA Amplification A 7F CK345 TTTCAGTGACACGCCCTT Deletion 7 Test 7R CK515 CGCAGGTTGATACACTTT Deletion 7 Test 8F CK519 GGCAATTGTGGAAAGTGT Deletion 8 Test 8R CK520 AGTGGAAGAAGTGGAAAAG Deletion 8 Test CK307 CK307 GGAGCCAAGCGAAGAATA Sequencing cDNA GGGGGTTCATCCTCCAGCAG CK404 CK404 Sequencing cDNA TG YPL352 YPL352 AGGCAGGCAAGTCCAAACAT Sequencing cDNA taqman_e1F YPL104 AGGAAGATCACCGGCAARC Exon 1 forward primer TaqMan taqman_e1R YPL105 CCCTTTTCCGCACATCCTTC Exon 1 reverse primer TaqMan CAAGTCGAACCTCGTAACTC taqman_e3F YPL106 Exon 3 forward primer TaqMan C ACAGATGGTGTTGAAGAGTC taqman_e3R YPL109 Exon 1 reverse primer TaqMan G 6-FAM – exon1_probe exon1_probe CCCAGGACAKCGGCAAGGC Exon 1 TaqMan probe A – TAMRA 6-FAM – exon3_probe exon3_probe AAGCTGCTCCCCTCACCTGC Exon 3 TaqMan probe – TAMRA

– 178 –

⁕ Supplementary Material ⁕

Table S.II.3 Summary of species and evidence for exon 3 duplications. Summary of species and evidence for exon 3 duplications from different approaches (based on RNA-seq coverage or PCRs) as well as presence and absence of the indels #7 and #8 that are variable within the Lake Victoria radiation. horizontal coverage results copy number exon 3a transcribed exon indels stripe species genus lake rel. copy experimental ident # # phenotype coverag source based on number evidence ity 7 8 e Alticorpus Haplochr Malinsky et Malawi no stripes 1.98 >=2 COV(1) ? ? ? geoffreyi omini al. 2017 Alticorpus Haplochr Malinsky et macrocleithr Malawi no stripes 1.78 >=2 COV(1) ? ? ? omini al. 2017 um Alticorpus Haplochr Malinsky et ambigiou Malawi no stripes 1.71 COV(1) ? ? ? peterdaviesi omini al. 2017 s Altolamprolo Lamprolo Tangan Exon gus no stripes ? - 1 PCR(1) cDNA(1) ? ? gini yika 3a calvus Aristochromi Haplochr oblique Malinsky et s Malawi 1.02 1 COV(1) ? ? ? omini stripe al. 2017 christyi Astatotilapia Haplochr partially Malinsky et Malawi 0.92 1 COV(1) ? ? ? bloyeti omini striped al. 2017 Astatotilapia Haplochr partially in in Rivers ? - >=2 PCR(1) ? burtoni omini striped s s Astatotilapia Haplochr Malinsky et in in Malawi no stripes 1.13 >=2 COV(1)|PCR(1) ? calliptera omini al. 2017 s s Astatotilapia Haplochr Malinsky et Malawi unknown 1.05 1 COV(1) ? ? ? sp. 'rujewa' omini al. 2017 Astatotilapia Haplochr partially Malinsky et Malawi 2.15 >=2 COV(1) ? ? ? tweddlei omini striped al. 2017 Aulonocara Haplochr Exon jacobfreiberg Malawi no stripes ? - >=2 PCR(3) cDNA(1) ? ? omini 3a i Aulonocara Haplochr Malinsky et sp. 'blue Malawi no stripes 1.81 >=2 COV(1) ? - ? ? omini al. 2017 chilumba' Aulonocara Haplochr Malinsky et Malawi no stripes 2.01 >=2 COV(1) ? - ? ? sp. ´gold´ omini al. 2017 Aulonocara Haplochr Malinsky et Malawi no stripes 1.99 >=2 COV(1) ? - ? ? sp. ´minutus´ omini al. 2017 Aulonocara Haplochr Malinsky et Malawi no stripes 2.13 >=2 COV(1) ? - ? ? sp. ´yellow´ omini al. 2017 Aulonocara Haplochr Malinsky et Malawi no stripes 1.88 >=2 COV(1) ? - ? ? steveni omini al. 2017 Aulonocara Haplochr Malinsky et Malawi no stripes 1.88 >=2 COV(1) ? - ? ? stuartgranti omini al. 2017 Buccochromi Haplochr oblique Malinsky et s Malawi 1.07 1 COV(1) ? - ? ? omini stripe al. 2017 nototaenia Buccochromi Haplochr oblique Malinsky et s Malawi 1.81 >=2 COV(1) ? - ? ? omini stripe al. 2017 rhoadesii Champsochr Haplochr Malinsky et ambigiou omis Malawi no stripes 1.27 COV(1) ? - ? ? omini al. 2017 s caeruleus Cheilochromi Haplochr Malinsky et Exon s Malawi stripes 1.15 1 COV(1)|PCR(3) cDNA(1) ? ? omini al. 2017 3a euchilus Chilotilapia Haplochr Malinsky et Malawi stripes 0.99 1 COV(1) ? - ? ? rhoadesii omini al. 2017 Copadichrom Haplochr Malinsky et ambigiou Malawi no stripes 1.3 COV(1) ? - ? ? is omini al. 2017 s

– 179 –

⁕ Supplementary Material ⁕

borleyi

Copadichrom Haplochr Malinsky et ambigiou is Malawi no stripes 1.71 COV(1) ? - ? ? omini al. 2017 s mloto Copadichrom is Haplochr Malinsky et ambigiou Malawi no stripes 1.75 COV(1) ? - ? ? quadrimacul omini al. 2017 s atus Copadichrom Haplochr Malinsky et is Malawi no stripes 2.12 >=2 COV(1) ? - ? ? omini al. 2017 trimaculatus Copadichrom Haplochr Malinsky et is Malawi stripes 1.94 >=2 COV(1) ? - ? ? omini al. 2017 virginalis Ctenochromi Haplochr Malinsky et s Malawi no stripes 2.72 >=2 COV(1) ? - ? ? omini al. 2017 pectoralis Ctenopharyn Haplochr Malinsky et x Malawi no stripes 1 1 COV(1) ? - ? ? omini al. 2017 intermedius Ctenopharyn Haplochr Malinsky et x Malawi no stripes 0.9 1 COV(1) ? - ? ? omini al. 2017 nitidus Cyathochrom Haplochr Malinsky et is Malawi stripes 1.87 >=2 COV(1) ? - ? ? omini al. 2017 obliquidens Cynotilapia Haplochr Malinsky et Malawi no stripes 1.94 >=2 COV(1) ? - ? ? afra omini al. 2017 Cynotilapia Haplochr Malinsky et Malawi no stripes 1.84 >=2 COV(1) ? - ? ? axelrodi omini al. 2017 Cyrtocara Haplochr Malinsky et Malawi no stripes 2.31 >=2 COV(1) ? - ? ? moorii omini al. 2017 Dimidiochro mis Haplochr Malinsky et Exon Malawi stripes 2.06 >=2 COV(1)|PCR(3) cDNA(1) ? ? compressicep omini al. 2017 3a s Dimidiochro Haplochr Malinsky et mis Malawi stripes 2.02 >=2 COV(1) ? - ? ? omini al. 2017 kiwinge Dimidiochro Haplochr Malinsky et mis Malawi stripes 2.18 >=2 COV(1) ? - ? ? omini al. 2017 strigatus Diplotaxodon Haplochr Malinsky et ambigiou Malawi no stripes 1.72 COV(1) ? - ? ? apogon omini al. 2017 s Diplotaxodon Haplochr Malinsky et ambigiou Malawi no stripes 1.62 COV(1) ? - ? ? limnothrissa omini al. 2017 s Diplotaxodon Haplochr Malinsky et Malawi no stripes 2.04 >=2 COV(1) ? - ? ? macrops omini al. 2017 Diplotaxodon Haplochr Malinsky et Malawi stripes 2.02 >=2 COV(1) ? - ? ? sp. ´similis´ omini al. 2017 Eclectochrom Haplochr Malinsky et ambigiou is Malawi no stripes 1.59 COV(1) ? - ? ? omini al. 2017 s ornatus Eretmodus Eretmodi Tangan Exon no stripes ? - 1 PCR(1) cDNA(1) ? ? Marksmithi ni yika 3a Fossorochro Haplochr partially Malinsky et mis Malawi 1.93 >=2 COV(1) ? - ? ? omini striped al. 2017 rostratus Genyochromi Haplochr Malinsky et s Malawi no stripes 1.09 1 COV(1) ? - ? ? omini al. 2017 mento Gephyrochro Haplochr Malinsky et mis Malawi no stripes 1.95 >=2 COV(1) ? - ? ? omini al. 2017 lawsi

– 180 –

⁕ Supplementary Material ⁕

Haplochromi Haplochr Victori d d s stripes ? - 1 PCR(3) ? - omini a el el chilotes Haplochromi Haplochr Victori McGee et al. s no stripes 1.05 1 COV(1) ? - ? ? omini a 2016 flavus Haplochromi Haplochr Victori Meier et al. ambigiou s no stripes 1.68 COV(1) ? - ? ? omini a 2017 s gracilior Haplochromi Haplochr Victori Valente et al. TAQ(3)|COV(1)| Exon in in s no stripes 1.79 >=2 cDNA(1) omini a 2014 PCR(3) 3a s s latifasciatus Haplochromi Haplochr Victori Exon in in s no stripes ? - >=2 PCR(1) cDNA(1) omini a 3a s s melanopterus Haplochromi Haplochr Victori McGee et al. ambigiou s no stripes 1.37 COV(1) ? - ? ? omini a 2016 s paucidens Haplochromi Haplochr Victori Exon d d s stripes ? - 1 TAQ(3)|PCR(3) cDNA(1) omini a 3a el el sauvagei Haplochromi Haplochr Victori Exon in in s stripes ? - >=2 PCR(3) cDNA(1) omini a 3a s s serranus Haplochromi s Haplochr Victori McGee et al. stripes 1.88 >=2 COV(1) ? - ? ? sp. omini a 2016 'checkmate' Haplochromi Haplochr Victori Exon in in s stripes ? - >=2 PCR(3) cDNA(1) omini a 3a s s thereuterion Haplochromi Haplochr Victori McGee et al. ambigiou s stripes 1.51 COV(1) ? - ? ? omini a 2016 s vittatus Hemitaenioc Haplochr Malinsky et hromis Malawi stripes 3.4 >=2 COV(1) ? - ? ? omini al. 2017 spilopterus Hemitilapia Haplochr Malinsky et Malawi no stripes 1.17 1 COV(1) ? - ? ? oxyrhyncha omini al. 2017 Iodotropheu Haplochr Malinsky et Malawi no stripes 1.16 1 COV(1) ? - ? ? sprengerae omini al. 2017 Julidochromi Lamprolo Tangan Exon s stripes ? - >=2 PCR(1) cDNA(1) ? ? gini yika 3a dickfeldi Julidochromi Lamprolo Tangan Exon s stripes ? - 1 PCR(3) cDNA(1) ? ? gini yika 3a marlieri Julidochromi Lamprolo Tangan s stripes ? - >=2 PCR(3) ? - ? ? gini yika ornatus Julidochromi Lamprolo Tangan Exon s stripes ? - 1 PCR(3) cDNA(1) ? ? gini yika 3a regani Labeotrophe Haplochr Malinsky et us Malawi no stripes 1.9 >=2 COV(1) ? - ? ? omini al. 2017 fuelleborni Labeotrophe Haplochr Malinsky et us Malawi no stripes 1.98 >=2 COV(1) ? - ? ? omini al. 2017 trewavasae Labidochrom Haplochr Exon is Malawi no stripes ? - >=2 TAQ(3)|PCR(3) cDNA(1) ? ? omini 3a caeruleus Lamprologus Lamprolo Tangan no stripes ? - 1 PCR(1) ? - ? ? meleagris gini yika

– 181 –

⁕ Supplementary Material ⁕

Lamprologus Lamprolo Tangan Exon multifasciatu no stripes ? - 1 PCR(3) cDNA(1) ? ? gini yika 3a s Lamprologus Lamprolo Tangan Exon no stripes ? - >=2 PCR(1) cDNA(1) ? ? similis gini yika 3a Lepidiolampr Lamprolo Tangan partially Exon ologus ? - >=2 PCR(1) cDNA(1) ? ? gini yika striped 3a kendalli Lethrinops Haplochr Malinsky et Malawi stripes 2.1 >=2 COV(1) ? - ? ? albus omini al. 2017 Lethrinops Haplochr Malinsky et Malawi no stripes 2.01 >=2 COV(1) ? - ? ? auritus omini al. 2017 Lethrinops Haplochr Malinsky et ambigiou Malawi no stripes 1.43 COV(1) ? - ? ? gossei omini al. 2017 s Lethrinops Haplochr Malinsky et ambigiou Malawi no stripes 1.75 COV(1) ? - ? ? lethrinus omini al. 2017 s Lethrinops Haplochr Malinsky et ambigiou Malawi no stripes 1.67 COV(1) ? - ? ? longimanus omini al. 2017 s Lethrinops Haplochr Malinsky et Malawi no stripes 2.03 >=2 COV(1) ? - ? ? longipinnis omini al. 2017 Lethrinops Haplochr Malinsky et Malawi unknown 2.16 >=2 COV(1) ? - ? ? sp. 'oliveri' omini al. 2017 Maylandia Haplochr Malinsky et ambigiou Malawi no stripes 1.71 COV(1) ? - ? ? emmiltos omini al. 2017 s Maylandia Haplochr Malinsky et Malawi no stripes 1.85 >=2 COV(1) ? - ? ? fainzilberi omini al. 2017 Maylandia Haplochr Malinsky et Malawi stripes 2.08 >=2 COV(1) ? - ? ? livingstonii omini al. 2017 Maylandia Haplochr Malinsky et polymorp ASS(1)|COV(1)| Exon in in Malawi no stripes 1.42 cDNA(1) zebra omini al. 2017 h PCR(4) 3a s s Melanochro Haplochr Malinsky et Exon RNA-seq(4)| mis Malawi stripes 1.05 1 COV(1)|PCR(2) ? ? omini al. 2017 3b cDNA(1) auratus Mylochromis Haplochr oblique Malinsky et Malawi 1.23 1 COV(1) ? - ? ? anaphyrmus omini stripe al. 2017 Mylochromis Haplochr oblique Malinsky et Malawi 1.83 >=2 COV(1) ? - ? ? ericotaenia omini stripe al. 2017 Mylochromis Haplochr oblique Malinsky et ambigiou Malawi 1.67 COV(1) ? - ? ? lateristriga omini stripe al. 2017 s Neolamprolo Lamprolo Tangan Exon in in gus no stripes ? - 1 PCR(3) cDNA(1) gini yika 3a s s brichardi Neolamprolo Lamprolo Tangan gus stripes ? - 1 PCR(2) ? - ? ? gini yika büscheri Neolamprolo gus Lamprolo Tangan Exon no stripes ? - >=2 PCR(3) cDNA(1) ? ? caudopunctat gini yika 3a us Neolamprolo Lamprolo Tangan Exon gus no stripes ? - >=2 PCR(3) cDNA(1) ? ? gini yika 3a cylindricus Neolamprolo Lamprolo Tangan gus no stripes ? - >=2 PCR(1) ? - ? ? gini yika fasciatus Neolamprolo gus Lamprolo Tangan Exon no stripes ? - >=2 PCR(1) cDNA(1) ? ? longicaudatu gini yika 3a s Neolamprolo Lamprolo Tangan Exon gus no stripes ? - >=2 PCR(1) cDNA(1) ? ? gini yika 3a prochilus Neolamprolo Lamprolo Tangan Exon gus no stripes ? - 1 PCR(1) cDNA(1) ? ? gini yika 3a pulcher – 182 –

⁕ Supplementary Material ⁕

Nimbochromi Haplochr Malinsky et s Malawi no stripes 2.57 >=2 COV(1) ? - ? ? omini al. 2017 linni Nimbochromi Haplochr Malinsky et s Malawi no stripes 2.4 >=2 COV(1) ? - ? ? omini al. 2017 livingstoni Nimbochromi Haplochr Malinsky et s Malawi no stripes 2.45 >=2 COV(1) ? - ? ? omini al. 2017 polystigma Ophthalmotil Tangan Exon apia Ectodini no stripes ? - >=2 PCR(1) cDNA(1) ? ? yika 3a ventralis Oreochromis Oreochro ASS(1)|PCR(1)|g Exon in in Rivers no stripes ? - 1 gDNA niloticus mini DNA(1) 3a s s Oreochromis Oreochro Malinsky et ambigiou Malawi no stripes 1.31 COV(1) ? - ? ? squamipinnis mini al. 2017 s Oreochromis Oreochro Rivers no stripes ? - 1 PCR(1) ? - ? ? variabilis mini Otopharynx Haplochr Malinsky et ambigiou brooksi Malawi unknown 1.67 COV(1) ? - ? ? omini al. 2017 s nkhata Otopharynx Haplochr Malinsky et Malawi no stripes 1.93 >=2 COV(1) ? - ? ? lithobates omini al. 2017 Otopharynx Haplochr Malinsky et ambigiou Malawi no stripes 1.75 COV(1) ? - ? ? speciosus omini al. 2017 s Otopharynx Haplochr Malinsky et Malawi no stripes 1.05 1 COV(1) ? - ? ? tetrastigma omini al. 2017 Pallidochrom Haplochr Malinsky et ambigiou is Malawi no stripes 1.46 COV(1) ? - ? ? omini al. 2017 s tokolosh Petrotilapia Haplochr partially Malinsky et Malawi 2.21 >=2 COV(1) ? - ? ? genalutea omini striped al. 2017 Placidochro Haplochr Malinsky et mis Malawi no stripes 1.8 >=2 COV(1) ? - ? ? omini al. 2017 electra Placidochro Haplochr Malinsky et mis Malawi no stripes 2.1 >=2 COV(1)|PCR(3) ? - ? ? omini al. 2017 johnstoni Placidochro Haplochr Malinsky et mis Malawi no stripes 2.04 >=2 COV(1) ? - ? ? omini al. 2017 longimanus Placidochro Haplochr Malinsky et mis Malawi no stripes 0.98 1 COV(1) ? - ? ? omini al. 2017 milomo Placidochro Haplochr partially Malinsky et ambigiou mis Malawi 1.44 COV(1) ? - ? ? omini striped al. 2017 s subocularis Placidochro Haplochr Malinsky et mis Malawi stripes 2.13 >=2 COV(1) ? - ? ? omini al. 2017 vulgaris Protomelas Haplochr Malinsky et ambigiou Malawi stripes 1.74 COV(1) ? - ? ? similis omini al. 2017 s Pseudosimoc Tangan Exon hromis Tropheini no stripes ? - 1 PCR(1) cDNA(1) ? ? yika 3a babaulti Pseudotrophe us Haplochr Exon Malawi stripes ? - >=2 TAQ(3)|PCR(3) cDNA(1) ? ? cyaneorhabd omini 3a os Pseudotrophe Haplochr Exon RNA-seq(5)| us Malawi no stripes ? - >=2 PCR(1) ? ? omini 3a cDNA(1) demasoni Pseudotrophe Haplochr Malinsky et us Malawi no stripes 1.77 >=2 COV(1) ? - ? ? omini al. 2017 elongatus

– 183 –

⁕ Supplementary Material ⁕

Pundamilia Haplochr Victori 1.65 | Meier et al. polymorp TAQ(3)|COV(2)| Exon RNA-seq(4)| in in no stripes nyererei omini a 1.01 2016 h PCR(3) 3a cDNA(1) s s Pundamilia Haplochr Victori 1.86 | Meier et al. polymorp Exon in in no stripes COV(2)|PCR(3) cDNA(1) pundamilia omini a 0.99 2016 h 3a s s Rhamphochr Haplochr Malinsky et omis Malawi stripes 2.02 >=2 COV(1) ? - ? ? omini al. 2017 esox Rhamphochr Haplochr Malinsky et omis Malawi stripes 2.19 >=2 COV(1) ? - ? ? omini al. 2017 longiceps Rhamphochr Haplochr Malinsky et omis Malawi stripes 2.72 >=2 COV(1) ? - ? ? omini al. 2017 woodi Sciaenochro Haplochr Malinsky et ambigiou mis Malawi no stripes 1.75 COV(1) ? - ? ? omini al. 2017 s benthicola Serranochro Haplochr Malinsky et mis Malawi stripes 3.45 >=2 COV(1) ? - ? ? omini al. 2017 robustus Stigmatochro Haplochr Malinsky et mis Malawi no stripes 1.11 1 COV(1) ? - ? ? omini al. 2017 guttatus Stigmatochro Haplochr Malinsky et ambigiou mis Malawi variable 1.55 COV(1) ? - ? ? omini al. 2017 s modestus Taeniochrom Haplochr Malinsky et is Malawi stripes 2.77 >=2 COV(1) ? - ? ? omini al. 2017 holotaenia Taeniolethrin Haplochr oblique Malinsky et ops Malawi 1.95 >=2 COV(1) ? - ? ? omini stripe al. 2017 furcicauda Taeniolethrin Haplochr Malinsky et ops Malawi no stripes 1.92 >=2 COV(1) ? - ? ? omini al. 2017 praeorbitalis Tanganicodu Lamprolo Tangan Exon s no stripes ? - 1 PCR(1) cDNA(1) ? ? gini yika 3a irsacae Telmatochro Lamprolo Tangan mis stripes ? - >=2 PCR(1) ? - ? ? gini yika bifrenatus Telmatochro Lamprolo Tangan mis stripes ? - >=2 PCR(2) ? - ? ? gini yika vittatus Thoracochro Haplochr Victori Meier et al. mis no stripes 2.46 >=2 COV(1) ? - ? ? omini a 2017 pharyngalis Tramitichrom Haplochr oblique Malinsky et is Malawi 1.94 >=2 COV(1) ? - ? ? omini stripe al. 2017 brevis Trematocran Haplochr Malinsky et us Malawi no stripes 1.01 1 COV(1) ? - ? ? omini al. 2017 placodon Tropheops Haplochr Malinsky et Malawi no stripes 1.78 >=2 COV(1) ? - ? ? tropheops omini al. 2017 Tyrannochro Haplochr Malinsky et mis Malawi stripes 1.88 >=2 COV(1) ? - ? ? omini al. 2017 macrostoma Tyrannochro Haplochr Malinsky et mis Malawi stripes 2.28 >=2 COV(1) ? - ? ? omini al. 2017 nigriventer Variabilichro Lamprolo Tangan Exon mis no stripes ? - 1 PCR(1) cDNA(1) ? ? gini yika 3a moori

– 184 –

⁕ Supplementary Material ⁕

Table S.II.4 Repeats and deletions within the agrp2 locus. Repeats and deletions within the agrp2 locus of Pundamilia nyererei (scaffold3:437297-462903) using http://www.repeatmasker.org (RM) and http://cowry.agri.huji.ac.il/cgi-bin/TilapiaRM.cgi (Cichlid_RM)

score div. del. ins. type position + repeat class/family length 18 15.1 2.7 0 RepeatMasker scaffold_3:438472-438508 + (GTG)n Simple_repeat 36 20 9.8 0 0 RepeatMasker scaffold_3:438796-438828 + (TTGTTT)n Simple_repeat 32 5657 5.9 5.8 0.9 RepeatMasker scaffold_3:439636-440475 C DNA33d_AFC DNA/CMC-EnSpm 839 2163 3.7 0.3 1.7 CichlidRepeatMasker scaffold_3:439656-439954 C F3085379F201HAF7_688165 Unknown 298 362 17.2 0 0 CichlidRepeatMasker scaffold_3:439937-440000 C RN497_frq688_Mariner/Tc1_TZF28C_related_gnl|ti|2139412188 Unknown 63 887 12.3 0 0 CichlidRepeatMasker scaffold_3:439999-440136 C RN81_frq847_gi|29420993|dbj|AB100550.1| Unknown 137 355 18.6 0 0 CichlidRepeatMasker scaffold_3:440115-440184 + F229833F133HAF7_426349 Unknown 69 2022 3.3 9.1 0.3 CichlidRepeatMasker scaffold_3:440159-440466 C M763154M152HAF7_1333 Unknown 307 1943 8.2 3.7 1.4 CichlidRepeatMasker scaffold_3:440926-441221 C RN73_frq898_gi|29421022|dbj|AB100579.1| Unknown 295 2037 6.9 7 1.3 RepeatMasker scaffold_3:440932-441241 C SINE_AFC SINE/MIR 309 611 10.2 0 9.2 CichlidRepeatMasker scaffold_3:441123-441241 C RN72_frq812_gi|29421025|dbj|AB100582.1| Unknown 118 890 10.2 5.2 0.7 RepeatMasker scaffold_3:442405-442552 + SINE_AFC SINE/MIR 147 1059 6.2 0 2.7 CichlidRepeatMasker scaffold_3:442405-442552 + RN73_frq898_gi|29421022|dbj|AB100579.1| Unknown 147 284 19.8 9.9 0 RepeatMasker scaffold_3:443930-444020 C Mariner-N11_EL DNA/TcMar-ISRm11 90 870 25.2 2.9 1.2 RepeatMasker scaffold_3:444905-445330 C L2-6_AFC LINE/L2 425 822 27.2 0.4 1.1 CichlidRepeatMasker scaffold_3:444921-445180 + RN46_frq1453_gi|129307148|gb|EF438175.1| Unknown 259 1315 22.3 3.5 0.9 CichlidRepeatMasker scaffold_3:444976-445319 + RN170_frq2375_gi|6649870|gb|AF016499.1|AF016499 Unknown 343 343 24.9 11.7 5.6 RepeatMasker scaffold_3:445337-445728 C DNA30_AFC DNA 391 624 17.7 2.8 11.5 RepeatMasker scaffold_3:445498-445823 + DNA29_AFC DNA 325 556 21.5 0.6 2.4 CichlidRepeatMasker scaffold_3:445554-445720 C RN449_frq101_hAT-N53_5638-5778_gnl|ti|2140363442 Unknown 166 246 18.7 4.3 3.2 CichlidRepeatMasker scaffold_3:445658-445751 + F1307881F171HAF7_527208 Unknown 93 236 10.3 2.1 24.1 CichlidRepeatMasker scaffold_3:445683-445823 + M1762787M185HAF7_131275 Unknown 140 5594 4.4 0.8 0.7 CichlidRepeatMasker scaffold_3:445934-446681 + RN342_frq1129_CR1-1_2785-3701_gnl|ti|2140164618 Unknown 747 5361 24.8 0.5 0.6 RepeatMasker scaffold_3:445934-447264 + Maui LINE/L2 1330 7608 6 0 0 CichlidRepeatMasker scaffold_3:446102-447093 + RN171_frq4546_gi|6649869|gb|AF016498.1|AF016498 Unknown 991 3411 5.6 0.5 0.2 CichlidRepeatMasker scaffold_3:446981-447428 C RN170_frq2375_gi|6649870|gb|AF016499.1|AF016499 Unknown 447 3189 3.4 5.7 0 RepeatMasker scaffold_3:447258-447674 + L2-6_AFC LINE/L2 416 2447 5.4 0 1.2 CichlidRepeatMasker scaffold_3:447348-447686 + RN171_frq4546_gi|6649869|gb|AF016498.1|AF016498 Unknown 338 492 0 0 0 CichlidRepeatMasker scaffold_3:447637-447697 C RN13_frq1296_gi|115334918|gb|DQ904494.1| Unknown 60 27 0 0 0 RepeatMasker scaffold_3:447675-447700 + (AATTG)n Simple_repeat 25 413 13.6 0 0 CichlidRepeatMasker scaffold_3:447704-447769 C RN175_frq2671_gi|5478227|gb|AF074485.1|AF074485 Unknown 65 419 21.2 1.7 7.8 RepeatMasker scaffold_3:447856-447983 + DNA5_AFC DNA/hAT-Charlie 127 552 10.3 0 0 RepeatMasker scaffold_3:449640-449717 C DNA-4_AFC DNA/CMC-EnSpm? 77 345 10.2 2.9 1.4 CichlidRepeatMasker scaffold_3:449642-449711 C RN384_frq349_DNA-2-16_1-84_gnl|ti|2173509383 Unknown 69 746 7.1 3.7 5.2 CichlidRepeatMasker scaffold_3:449702-449835 C RN70_frq845_gi|29421032|dbj|AB100589.1| Unknown 133 657 12.8 0.8 1.7 RepeatMasker scaffold_3:449711-449829 C SINE_AFC SINE/MIR 118 812 7.9 1.8 0 CichlidRepeatMasker scaffold_3:449722-449835 C RN81_frq847_gi|29420993|dbj|AB100550.1| Unknown 113 3545 7 5.1 0.6 CichlidRepeatMasker scaffold_3:449846-450394 C RN384_frq349_DNA-2-16_1-84_gnl|ti|2173509383 Unknown 548 6087 5.9 3.6 2.8 RepeatMasker scaffold_3:449846-450717 C DNA-4_AFC DNA/CMC-EnSpm? 871 464 15.9 1.6 7.4 CichlidRepeatMasker scaffold_3:450308-450429 C RN385_frq494_SINE_AFC_1-130_gnl|ti|2139378986 Unknown 121 829 4.7 0 0 CichlidRepeatMasker scaffold_3:450447-450552 + M4064051M217HAF7_209843 Unknown 105 1144 5 0 3.6 CichlidRepeatMasker scaffold_3:450552-450717 C RN384_frq349_DNA-2-16_1-84_gnl|ti|2173509383 Unknown 165 558 19.5 7.6 0 CichlidRepeatMasker scaffold_3:450762-450879 + RN469_frq121_Mariner/Tc1_1-342_gnl|ti|2139386163 Unknown 117 836 18.7 12.9 8.4 RepeatMasker scaffold_3:450762-451082 + Tc1-6_AFC DNA/TcMar-Tc1 320 269 25.2 6.5 15.8 CichlidRepeatMasker scaffold_3:450773-450956 + RN471_frq133_Mariner/Tc1_1-1008_gnl|ti|2139464306 Unknown 183 370 27.2 2.3 5.3 CichlidRepeatMasker scaffold_3:451087-451218 + RN483_frq132_Mariner/Tc1_385-1211_gnl|ti|2141265318 Unknown 131 801 19 6.5 3.9 RepeatMasker scaffold_3:451090-451270 + Tc1-13_AFC DNA/TcMar-Tc1 180 501 20.2 0.8 1.6 CichlidRepeatMasker scaffold_3:451119-451239 + RN473_frq256_Mariner/Tc1_11-909_gnl|ti|2141288087 Unknown 120 460 22.5 21.6 4.3 CichlidRepeatMasker scaffold_3:451124-451355 + RN485_frq136_Mariner/Tc1_485-1142_gnl|ti|2139460218 Unknown 231 432 20.8 4.5 1.9 RepeatMasker scaffold_3:451248-451355 + Tc1-6_AFC DNA/TcMar-Tc1 107 19 8.1 0 0 RepeatMasker scaffold_3:451779-451804 + (T)n Simple_repeat 25 16 0 0 0 RepeatMasker scaffold_3:451828-451844 + (A)n Simple_repeat 16 58 0 0 0 RepeatMasker scaffold_3:452356-452407 + (ATCT)n Simple_repeat 51 17 26.8 0 0 RepeatMasker scaffold_3:453633-453681 + (TAG)n Simple_repeat 48 446 18.5 6.1 0 RepeatMasker scaffold_3:455299-455390 C Tc1-6_AFC DNA/TcMar-Tc1 91 416 21 18.4 2.1 CichlidRepeatMasker scaffold_3:455383-455572 C RN477_frq183_Mariner/Tc1_77-944_gnl|ti|2141277805 Unknown 189 672 13.3 16 0 RepeatMasker scaffold_3:455415-455572 C Tc1-6_AFC DNA/TcMar-Tc1 157 297 26.3 10.6 0 RepeatMasker scaffold_3:457195-457312 C Mariner-3_EL DNA/TcMar-Tc1 117 639 16.1 0.8 0.8 CichlidRepeatMasker scaffold_3:457461-457579 + F3293479F203HAF7_319348 Unknown 118 648 10.4 0.9 0.9 CichlidRepeatMasker scaffold_3:457473-457579 C F1681407F182HAF7_257021 Unknown 106 435 5.4 0 0 CichlidRepeatMasker scaffold_3:457607-457662 C F2585091F196HAF6_695488 Unknown 55 280 10.9 1.8 15.6 CichlidRepeatMasker scaffold_3:457661-457769 C F1169603F167HAF7_16361 Unknown 108 275 7.3 2.4 0 CichlidRepeatMasker scaffold_3:457890-457930 + RN607_frq15357_Misc_Short_repeats_generated_by_MIRA Unknown 40 6456 4.7 3.2 2.3 RepeatMasker scaffold_3:457895-458802 + DNA44_AFC DNA/hAT-Ac 907 288 6.1 14.3 0 CichlidRepeatMasker scaffold_3:457911-457959 + M1373773M176HAF7_130583 Unknown 48 547 6.7 0 0 CichlidRepeatMasker scaffold_3:457943-458017 C RN607_frq15357_Misc_Short_repeats_generated_by_MIRA Unknown 74 440 11.4 0 13.7 CichlidRepeatMasker scaffold_3:457992-458093 C M1614293M182HAF7_350417 Unknown 101 831 4.8 0 0 CichlidRepeatMasker scaffold_3:458070-458174 C RN607_frq15357_Misc_Short_repeats_generated_by_MIRA Unknown 104 771 3 1 0 CichlidRepeatMasker scaffold_3:458163-458261 C F2223543F191HAF7_275172 Unknown 98 1072 11.5 2.7 9.5 CichlidRepeatMasker scaffold_3:458225-458445 C RN460_frq555_REX1-1_2-547&Mariner_gnl|ti|2139459883 Unknown 220 401 12.1 0 0 CichlidRepeatMasker scaffold_3:458446-458503 + RN572_frq733_Unknown20_SMAR12planaria_DNA/Mariner_related_MIRA Unknown 57 270 18 0 0 CichlidRepeatMasker scaffold_3:458495-458544 C RN607_frq15357_Misc_Short_repeats_generated_by_MIRA Unknown 49 292 15.4 0 0 CichlidRepeatMasker scaffold_3:458524-458575 C F98723F130HAF7_271348 Unknown 51 1261 2.9 1.7 0.6 CichlidRepeatMasker scaffold_3:458562-458733 C RN596_frq109_Unknown43_MIRA Unknown 171 329 10.3 0 7.9 CichlidRepeatMasker scaffold_3:458690-458752 C F638931F146HAF7_212965 Unknown 62 338 2.2 2.2 0 CichlidRepeatMasker scaffold_3:458762-458806 + M6370538M337HAF7_253011 Unknown 44 844 12.9 8.3 0 RepeatMasker scaffold_3:458873-459027 + Tc1-1_AFC DNA/TcMar-Tc1 154 802 14.6 7.4 0 RepeatMasker scaffold_3:459030-459180 + Tc1-1_AFC DNA/TcMar-Tc1 150 408 3.6 0.5 0 RepeatMasker scaffold_3:459472-459539 C DNA37a_AFC DNA/PIF-Harbinger 67 403 6 13.4 0 CichlidRepeatMasker scaffold_3:459476-459542 C RN386_frq328_DNA-3-5_1-1074_gnl|ti|2140408976 Unknown 66 630 14 1.8 0 RepeatMasker scaffold_3:459541-459647 + Tc1-1_AFC DNA/TcMar-Tc1 106 834 18.4 4.8 0 RepeatMasker scaffold_3:459654-459832 + Tc1-1_AFC DNA/TcMar-Tc1 178 244 18.9 15.9 4.3 RepeatMasker scaffold_3:461182-461275 C piggyBac-N6_DR DNA/PiggyBac 93 272 34.5 0 0 RepeatMasker scaffold_3:461871-461980 C Maui LINE/L2 109 244 23.3 0 0 CichlidRepeatMasker scaffold_3:461921-461980 C F743269F150HAF6_216082 Unknown 59 14 26.9 0 5 RepeatMasker scaffold_3:462507-462569 + (TTTATTC)n Simple_repeat 62

– 185 –

⁕ Supplementary Material ⁕

Chapter III

Table S.III.1 Relative expression of agrp2 in N. parilus vs. P. pulcher Species P value Fold change N. parilus (n=6) P. pulcher (n=6) (two-tailed t test) agrp2 expression 2.80 ± 0.90 0.50 ± 0.36 <0.001 5.57 (%) (Whole trunk)

Table S.III.2 Relative expression of agrp2 in D. tinwini vs. D. rerio Species P value Fold change D. tinwini (n=5) D.rerio (n=5) (two-tailed t test) agrp2 expression 1.14 ± 0.38 0.64 ± 0.10 <0.05 1.78 (%) (Whole trunk)

Table S.III.3 Relative expression levels of target genes in M. auratus. Tissue/Position Expression level (%) (n=5) asip1 agrp2 DLS 0.08 ± 0.10 0.55 ± 0.23 DIN 0.30 ± 0.07 0.75 ± 0.33 MLS 0.13 ± 0.03 1.18 ± 0.50 VEN 3.48 ± 1.99 1.52 ± 1.34

Statistical test for asip1 expression (ANOVA: F-value: 13.86; P value: <0.001) TukeyHSD: DLS 0.985 1.000 <0.001 DIN 0.992 <0.001 MLS <0.001 VEN

Statistical test for agrp2 expression (ANOVA: F-value: 1.69; P value: 0.209) TukeyHSD: DLS 0.972 0.556 0.210 DIN 0.804 0.395 MLS 0.888 VEN

– 186 –

⁕ Supplementary Material ⁕

Table S.III.4 Relative expression levels of target genes in M. cyaneorhabdos. Tissue/Position Expression level (%) (n=5) asip1 agrp2 DLS 0.05 ± 0.08 0.82 ± 0.60 DIN 0.53 ± 0.19 1.11 ± 1.28 MLS 0.09 ± 0.01 1.19 ± 1.38 VEN 3.34 ± 1.02 0.88 ± 0.62

Statistical test for asip1 expression (ANOVA: F-value: 45.59; P value: <0.001) TukeyHSD: DLS 0.492 1.000 <0.001 DIN 0.555 <0.001 MLS <0.001 VEN

Statistical test for agrp2 expression (ANOVA: F-value: 0.143; P value: 0.932) TukeyHSD: DLS 0.971 0.944 1.000 DIN 0.999 0.984 MLS 0.965 VEN

Table S.III.5 Relative expression levels of target genes in P. pulcher Tissue/Position Gene expression level (%) (n=6) asip1 agrp2 DLS 0.10 ± 0.05 0.35 ± 0.17 DIN 0.79 ± 0.44 0.96 ± 0.55 MLS 0.24 ± 0.16 1.33 ± 1.21 VEN 2.87 ± 2.42 1.36 ± 1.31

Statistical test for asip1 expression (ANOVA: F-value: 6.49; P value: <0.01) TukeyHSD: DLS 0.763 0.997 <0.01 DIN 0.863 <0.05 MLS <0.01 VEN

Statistical test for agrp2 expression (ANOVA: F-value: 1.516; P value: 0.241) TukeyHSD: DLS 0.682 0.294 0.273 DIN 0.896 0.877 MLS 1.000 VEN

– 187 –

⁕ Supplementary Material ⁕

Table S.III.6 Relative expression levels of target genes in T. salvini. Tissue/Position Expression level (%) (n=6) asip1 agrp2 DLS 0.00 ± 0.00 0.75 ± 0.70 DIN 0.53 ± 1.17 1.24 ± 0.66 MLS 0.14 ± 0.16 0.05 ± 0.02 VEN 3.39 ± 3.02 1.96 ± 1.12

Statistical test for asip1 expression (ANOVA: F-value: 5.422; P value: <0.01) TukeyHSD: DLS 0.944 0.999 <0.05 DIN 0.977 <0.05 MLS <0.05 VEN

Statistical test for agrp2 expression (ANOVA: F-value: 7.109; P value: <0.01) TukeyHSD: DLS 0.660 0.381 <0.05 DIN 0.050 0.362 MLS <0.01 VEN

Table S.III.7 Relative expression levels of target genes in B. pi Tissue/Position Expression level (%) (n=6) asip1 agrp2 DOR 0.01 ± 0.01 1.31 ± 0.82 DLS 0.10 ± 0.04 0.60 ± 0.41 DIN 0.42 ± 0.19 0.46 ± 0.11 MLS 0.14 ± 0.09 0.40 ± 0.33 VIN 0.56 ± 0.25 1.02 ± 0.68 VLS 2.86 ± 2.72 2.16 ± 2.65 VEN 2.90 ± 2.73 1.05 ± 1.09

Statistical test for asip1 expression: (ANOVA: F-value: 4.721; P value: <0.01) TukeyHSD: DOR 1.000 0.999 1.000 0.994 <0.05 <0.05 DLS 1.000 1.000 0.998 <0.05 <0.05 DIN 1.000 1.000 0.087 0.088 MLS 0.058 0.337 0.327 VIN 0.479 0.465 VLS 1.000 VEN

Statistical test for agrp2 expression (ANOVA: F-value: 1.631; P value: 0.168) TukeyHSD: DOR 0.937 0.867 0.824 0.999 0.872 1.000 DLS 1.000 1.000 0.996 0.274 0.994 DIN 1.000 0.982 0.191 0.976 MLS 0.968 0.159 0.960 VIN 0.632 1.000 VLS 0.661 VEN

– 188 –

⁕ Supplementary Material ⁕

Table S.III.8 Relative expression levels of target genes in D. rerio Tissue/Position Expression level (%) (n=6) asip1 agrp2 Skin-DOR 0.00 ± 0.00 0.62 ± 0.77 Skin-DLS1 0.00 ± 0.00 0.99 ± 0.82 Skin-DIN1 0.00 ± 0.00 3.04 ± 2.26 Skin-DLS2 0.09 ± 0.14 0.68 ± 0.51 Skin-DIN2 0.03 ± 0.04 1.63 ± 0.99 Skin-MLS 0.11 ± 0.09 0.51 ± 0.57 Skin-VIN1 0.38 ± 0.28 1.46 ± 0.78 Skin-VLS1 0.92 ± 0.72 0.51 ± 0.46 Skin-VIN2 2.87 ± 3.60 0.48 ± 0.35 Skin-VLS2 1.37 ± 1.63 0.48 ± 0.37 Skin-VEN 5.22 ± 3.43 0.61 ± 0.91

Statistical test for asip1 expression (ANOVA: F-value: 6.451; P value: <0.001) TukeyHSD: DOR 1.000 1.000 1.000 1.000 1.000 1.000 0.995 0.092 0.917 <0.001 DLS1 1.000 1.000 1.000 1.000 1.000 0.995 0.092 0.917 <0.001 DIN1 1.000 1.000 1.000 1.000 0.995 0.092 0.917 <0.001 DLS2 1.000 1.000 1.000 0.998 0.117 0.946 <0.001 DIN2 1.000 1.000 0.996 0.099 0.926 <0.001 MLS 1.000 0.998 0.121 0.950 <0.001 VIN1 1.000 0.228 0.991 <0.001 VLS1 0.572 1.000 <0.001 VIN2 0.866 0.299 VLS2 <0.01 VEN

Statistical test for agrp2 expression (ANOVA: F-value: 4.152; P value: <0.001) TukeyHSD: DOR 1.000 <0.01 1.000 0.746 1.000 0.903 1.000 1.000 1.000 1.000 DLS1 <0.05 1.000 0.983 0.998 0.999 0.998 0.997 0.997 1.000 DIN1 <0.01 0.282 <0.01 0.149 <0.01 <0.001 <0.001 <0.01 DLS2 0.808 1.000 0.938 1.000 1.000 1.000 1.000 DIN2 0.612 1.000 0.615 0.583 0.582 0.731 MLS 0.810 1.000 1.000 1.000 1.000 VIN1 0.816 0.786 0.786 0.894 VLS1 1.000 1.000 1.000 VIN2 1.000 1.000 VLS2 1.000 VEN

Table S.III.9 Relative expression levels of target genes in two-striped species Tissue/Position Expression level (%) (n=4) asip1 DLS 0.06 ± 0.04 DIN 0.54 ± 0.20 MLS 0.15 ± 0.06 VEN 3.26 ± 0.26

Statistical test for asip1 expression (ANOVA: F-value: 317.32; P value: <0.001) TukeyHSD: DLS 0.008 0.871 <0.001 DIN 0.032 <0.001 MLS <0.001 VEN

– 189 –

⁕ Supplementary Material ⁕

Chapter IV

Fig. S.IV.1 Chromatophores in light microscopical and TEM analysis. (a) Photographs of same part of a scale under different condition: left panels, brightfield images; middle panels, images under polarized light; right panel, fluorescent image. Upper row, images before adrenaline treatment; lower row, images after adrenaline treatment. Red arrows indicate the same melanophore before and after adrenaline treatment. White arrows indicate the iridophore(s) before and after adrenaline treatment. Red arrowheads indicate the same xanthophore under brightfield and fluorescence. (b) Transmission electron microscopy (TEM) images of melanophore (upper row), xanthophore (middle row) and iridophore (lower row). Left panels, low magnification (adjacent cells masked). Middle panels, boxed regions. Right panels, illustrations of chromatophores. Abbreviations: Mel, melanophore; Xan, xanthophore; Iri, iridophore; n, nucleus; m, melanosome; x, xanthosome/pterinosome; rp, reflecting platelet. Scale bars are 20 μm in (a) and 2 μm in (b).

Fig. S.IV.2 Melanosome density from the black pigmented regions. Each individual dot indicates the mean value from the same layer. Orange/gray bars represent the mean values of all three layers from the same regions. Numbers under the plot: mean ± SD.

– 190 –

⁕ Supplementary Material ⁕

Fig. S.IV.3 Polar chart of the angle of iridosomes in the MLS of yellow morph. Result from all three layers (a), SC (b), SP (c) and BM (d).

Fig. S.IV.4 Distance between iridosomes and the thickness of iridosomes. Small dots represent the value of individual measurement. Large black dots indicate the mean values from the same layers. Error bars indicated ± SD.

– 191 –

⁕ Supplementary Material ⁕

Fig. S.IV.5 Chromatophores organization in skin of yellow and dark morph M. auratus. Ultrastructural illustration of chromatophore arrangement in integument: DLS (a), INT (c), MLS (e) and vVEN (g) in yellow morph M. auratus; DLS (b), INT (d), MLS (f) and vVEN (h) in dark morph M. auratus. Illustrations of chromatophores are based on Supplementary Fig S1b: Black dots represent melanosomes, yellow dots represent xanthosomes and blue platelets represent iridosomes/reflecting platelets.

– 192 –

⁕ Supplementary Material ⁕

Fig. S.IV.6 Axon density on scales across dorsa-ventral axis. Each point represents mean value of 5 scales from the same dorsal-ventral position within one individual. Error bars represent +SD. Differences between morphs/sexes in the same species and genotype were evaluated by two- tailed t test, n = 4 (M.auratus AM “yellow morph”) or 5 (M. auratus WT yellow morph and dark morph, M. auratus AM “dark morph”, M. cyneorhabdos female and male). Significant sign: *** P<0.001, ** P<0.01, n.s. non- significant.

Fig. S.IV.7 Axon density on scales. Axon density on scales is quantified based on Acetylated Tubulin staining. Red dash line crossing the scales in the up-right corner of (a) and (b) indicated the measurement position. (a) Axon density in anterior of scales. The two distinct endpoints of measurement are located in the most dorsal and the most ventral edge of epidermis. (b) Axon density in posterior of scales. The two distinct endpoints of measurement are located in the most dorsal and the most ventral centiis. Differences between morphs/sexes in the same species and genotype were evaluated by two-tailed t test, n = 4 (M.auratus AM female) or 5 (M. auratus WT yellow morph and dark morph, M. auratus AM dominant male, M. cyneorhabdos female and male) (individual dots). Each individual dot represents the mean value of measurement for different fishes. Green bars indicate the means. Significant sign: *** P<0.001, n.s. non- significant.

– 193 –

⁕ Supplementary Material ⁕

Fig. S.IV.8 Immunofluorescence staining of axon in wild-type M. auratus. (a, b) Representative images of Acetylated Tubulin staining labeled scales from VEM2 of indicated phenotype. (c, d) High magnification microscope photographs of Acetylated Tubulin staining on scales. Melanophores can be observed in wild-type (WT) yellow morph and dark morph (white arrowhead in c and d). Scale bars are 500 μm in (a, b), 50 μm in (c, d).

Table S.IV.1 Chromatophore measurement and results of the statistical tests (within morph and individual)

– 194 –

⁕ Supplementary Material ⁕

– 195 –

⁕ Supplementary Material ⁕

– 196 –

⁕ Supplementary Material ⁕

Table S.IV.2 Statistical tests of chromatophore measurement (across morph) Mean value ± SD P value Dark/Yellow Measurement position (two-tailed t Dark morph Yellow morph Fold change test) DLS 29.397 ± 7.926 84.690 ± 6.761 <0.001 0.347 Melanophore INT 73.239 ± 8.519 1.343 ± 1.455 <0.001 54.534 coverage MLS 10.653 ± 6.612 79.313 ± 9.873 <0.001 0.1343 (%) dVEN 69.356 ± 9.122 0.607 ± 0.493 <0.001 114.260 vVEN 69.883 ± 5.016 0.215 ± 0.153 <0.001 325.037 DLS 186.36 ± 43.02 377.34 ± 50.27 <0.001 0.494 Melanophore INT 301.71 ± 60.14 52.83 ± 39.78 <0.001 5.711 density MLS 81.64 ± 51.09 287.67 ± 51.11 <0.001 0.284 (cell/mm2) dVEN 212.81± 27.49 22.93 ± 13.71 <0.001 9.281 vVEN 134.70 ± 13.01 3.49 ± 2.88 <0.001 38.596 DLS 49.706 ± 6.172 71.600 ± 6.600 <0.001 0.694 Melanophore INT 78.036 ± 7.634 20.558 ± 4.740 <0.001 3.796 dispersed MLS 38.247 ± 15.999 79.163 ± 5.951 <0.01 0.483 diameter dVEN 89.765 ± 7.410 17.967 ± 6.511 <0.001 4.996 (µm) 104.972 ± 9.464 vVEN 11.092 ± 6.420 <0.001 14.710 DLS 1.369 ± 2.421 0.252 ± 0.185 0.361 5.433 Xanthophore INT 0.402 ± 0.582 1.678 ± 1.838 0.201 0.240 coverage MLS 0.757 ± 1.008 0.262 ± 0.249 0.345 2.889 (%) dVEN 0.225 ± 0.224 1.680 ± 1.051 <0.05 0.134 vVEN 0.054 ± 0.041 11.220 ± 4.827 <0.01 0.005 DLS 55.77 ± 7.41 30.56 ± 11.14 0.510 1.825 Xanthophore INT 27.81 ± 36.67 79.86 ± 50.15 0.102 0.348 density MLS 31.97 ± 49.17 17.28 ± 13.25 0.550 1.850 (cell/mm2) dVEN 16.49 ± 23.55 36.89 ± 34.60 0.312 0.447 vVEN 2.38 ± 1.20 99.29 ± 33.80 <0.01 0.024 DLS 15.854 ± 8.305 16.740 ± 7.298 0.862 0.947 Xanthophore INT 13.104 ± 4.810 19.758 ± 10.848 0.261 0.663 dispersed MLS 20.445 ± 8.004 21.083 ± 0.720 0.868 0.970 diameter dVEN 15.623 ± 5.093 33.232 ± 4.344 <0.001 0.470 (µm) vVEN 18.179 ± 4.866 64.606 ± 13.067 <0.001 0.281 DLS 3.365 ± 1.165 0.150 ± 0.205 <0.01 22.433 Iridophore INT 0.801 ± 0.513 1.136 ± 1.359 0.627 0.7051 coverage MLS 0.757 ± 0.588 0.019 ± 0.027 <0.05 39.842 (%) dVEN 0.150 ± 0.077 0.120 ± 0.127 0.667 1.250 vVEN 0.151 ± 0.200 0.125 ± 0.084 0.800 1.208

– 197 –

⁕ Supplementary Material ⁕

Table S.IV.3 Axon density and results of the statistical tests (within morph/sex and individual)

– 198 –

⁕ Supplementary Material ⁕

– 199 –

⁕ Supplementary Material ⁕

Table S.IV.4 Statistical tests of axon density (across morph/sex)

– 200 –

⁕ Supplementary Material ⁕

Chapter V

Fig. S.V.1 Vertical bar formation is consistent among all three assessed individuals. Scale bar is 1mm for all subpanels.

Fig. S.V.2 Detailed photographs of lateral cluster. A new melanophore cluster (lateral cluster, lc, dotted outline in all panels) as a distinct entity form around 12dpf in all assessed individuals. The lc further merge with dc3 and further develop into vb3.

– 201 –

⁕ Supplementary Material ⁕

Fig. S.V.3 Changes in melanophore dispersal diameter over time, Changes in melanophore dispersal diameter over time suggesting an increase in cell size at early stages (between 9 and 13 dpf)

Fig. S.V.4 Pigment coverage of the three individuals. Each hollow point represents measurement from one region. Solid points represent the mean value of these. Error bars indicate means ± SD.

Fig. S.V.5 Chromatophore dispersal diameter in the three individuals. Each hollow point represents one measured cell. Solid points represent the mean value of these. Error bars indicate means ± SD.

– 202 –

⁕ Supplementary Material ⁕

Fig. S.V.6 Chromatophore density in the three individuals. Each hollow point represents measurement from one region. Solid points represent the mean value of these. Error bars indicate means ± SD.

Fig. S.V.7 Relative expression of melanophore genes corrected by relative expression of mitfa. Differences were tested by ANOVA followed by Tukey–Kramer post-hoc test, n = 5 (individual dots). Error bars indicate means + SD. Abbreviations: ***, P<0.001; **, P<0.01; *, P<0.05.

Table S.V.1 Chromatophore measurement and results of the statistical tests.

– 203 –

⁕ Supplementary Material ⁕

Table S.V.2 Melanin synthesis genes expression levels (divide by expression of mitfa) and results of the statistical tests.

– 204 –

⁕ Supplementary Material ⁕

Table S.V.3 Relative expression levels of target genes and results of the statistical tests.

Table S.V.4 Melanophore related genes expression levels (standardized by melanophore density). Gene expression (Skin) vb / ib fold Gene ib1 vb2 ib2 vb3 change mitfa 0.0113 0.0064 0.0156 0.0059 0.46 pmel 0.0202 0.0107 0.0180 0.0110 0.57 slc24a5 0.0154 0.0220 0.0305 0.0401 1.36 tyr 0.0154 0.0320 0.0168 0.0380 2.17 tyrp1a 0.0149 0.0341 0.0179 0.0375 2.18

Gene expression (Scale) vb / ib fold Gene ib1 vb2 ib2 vb3 change mitfa 0.0298 0.0140 0.0287 0.0119 0.44 pmel 0.0533 0.0234 0.0330 0.0221 0.53 slc24a5 0.0405 0.0483 0.0558 0.0808 1.34 tyr 0.0407 0.0701 0.0307 0.0764 2.05 tyrp1a 0.0393 0.0747 0.0328 0.0756 2.09

Gene expression (Skin + Scale) vb / ib fold Gene ib1 vb2 ib2 vb3 change mitfa 0.0164 0.0088 0.0202 0.0079 0.46 pmel 0.0293 0.0146 0.0233 0.0147 0.56 slc24a5 0.0223 0.0303 0.0394 0.0536 1.36 tyr 0.0224 0.0439 0.0217 0.0507 2.15 tyrp1a 0.0216 0.0468 0.0232 0.0502 2.17

– 205 –

⁕ Supplementary Material ⁕

Table S.V.5 Primer sequence for qPCR. Name Sequence (5’ – 3’) Gene agrp2_Fwd GCGAAGAATAGGCGGCTGTTTG agrp2 agrp2_Rev CGACGCGCCGGAGTTACGAG agrp2 asip1_Fwd GAAGAGCAAGAAACCAAAGAAACA asip1 asip1_Rev ACTGGCAGAAAGCACAATAATCAC asip1 csf1ra_F GCAATGCACGTCTGCCAGTG csf1ra csf1ra_R ACGTCACTCTGGACGGTGT csf1ra gapdh_Fwd CACACAAGCCCAACCCATAGTCAT gapdh gapdh_Rev AAACACACTGCTGCTGCCTACATA gapdh ltk_F GCCACAGCAACCACAGAGTAC ltk ltk_R GAAGTGTCCATCAGACCTTCCCTC ltk mc1r_Fwd TGCTGGGGCCCTTTCTTTCTACAC mc1r mc1r_Rev AAGCGGGTCGATGAGGGAGTTACA mc1r mc5r_Fwd ATCCTGGGTATCATCTCACTGC mc5r mc5r_Rev CATGTCTGCTACTGCCAAACTG mc5r mitfa_Fwd CGACGATGTTCTTGGATTGATGGA mitfa mitfa_Rev CGAGGCCTGGTAGCTGGAGACTT mitfa pmel_F CAGTTCTCCATCACTGATCAAATCCC pmel pmel_R TCGCCCTGTTCTGGATGAAGC pmel slc24a5_F CACCTGAACTAGTCACAGCCTTCC slc24a5 slc24a5_R TGAGACGCCCAGCCATAGA slc24a5 sox10_Fwd AGCGGCGAGGAGGAAACTTG sox10 sox10_Rev GAATGGCCTCTTGTCGGTCTCA sox10 tyr_Fwd TCCCTTAATCCCAACCTCATCAGC tyr tyr_Rev ACCCTCCCCCGTGGCATTACATA tyr tyrp1a_Fwd GCGGGGCAGGCCAGTTTTG tyrp1a tyrp1a_Rev GGCAGGGCGAAGGAAGGGTTTT tyrp1a β-actin_Fwd TGACATGGAGAAGATCTGGC β-actin β-actin_Rev TGGCAGGAGTGTTGAAGGT β-actin

– 206 –