Zebrafish as a vertebrate model to study retinoic acid signaling in head mesoderm and pectoral fin
development and to investigate non-ion channel epilepsies
Dissertation zur Erlangung des akademischen Grades des Doktors der Naturwissenschaften (Dr. rer. nat.)
an der
Universität Konstanz, Mathematisch-naturwissenschaftliche Sektion Fachbereich Biologie
vorgelegt von Yann Gibert, M.Sc.
Konstanz, November 2004
Acknowledgments:
I would like to thank my advisors, Prof. Axel Meyer and Dr. Gerrit Begemann for giving me the opportunity to complete a doctoral degree.
I am grateful of most of the past and current members of the Meyer Lab and friends and especially I thank, with no particular order, Dr Yves van de Peer, Prof. Marc Ekker, Dr. John S.
Taylor, Romulus Abila, Eric Cartmann, Ingo Braasch, Dr. Marta Barluenga, Dr. Walter Salzburger, Sir Paul McCartney, Chiara Reggio, Prof Marie-André Akimenko, Peter McHallaway, Ruth Garcia-Chico, Servietsky, Celine Clabaut, Katharine Webb, Steve-O, Ingrid May, Kathrin Hoffmann and of course, not because I want to but because I have to, Dr. Thierry Wirth.
I would like to thank God if he exists.
I want to thank for financial support the Landesgraduiertenstipendium, for paying for my salary for two years, and the University of Konstanz for funding me for international meetings.
I want to thank my parents
Last but not least I want to thank a couple of people I know, Big Ben, Hou and the two André qué. I do not think I have to explain why I am thanking them because I am sure they know.
To BJ… one last time
“On ne m’a pas aidé, je n’ai pas eu de veine Mais un jour viendra je leur montrerai que j’ai du talent”
C. Aznavour
TABLE OF CONTENTS
CHAPTER I: General Introduction 5
Retinoic acid metabolism
Retinoic acid and hindbrain development Retinoic signaling in pectoral fin development
Non-ion channel epilepsy: new avenues in epileptic diseases LGI genes and the MASS1/VLGR1 genes share EPTP domains Aim of this work
6 7 10 11 12 14 CHAPTER II: Cloning of zebrafish T-box genes tbx15 and tbx18 and
their expression during embryonic development
16
2.1 Abstract
2.2 Materials and methods 2.3 Results and discussion
17 17 18 CHAPTER III: Retinoic acid is essential for antero-posterior patterning
of cranial mesoderm in zebrafish 24
3.1 Abstract 3.2 Introduction
3.3 Materials and methods 3.4 Results
3.5 Discussion
25 25 27 28 33 CHAPTER IV: Temporal and spatial requirement for retinoic acid in
zebrafish pectoral fin development
38
4.1 Abstract 4.2 Introduction
4.3 Materials and methods 4.4 Results
4.5 Discussion
39 40 41 42 52 CHAPTER V: Applying expression and gene-history analysis to predict
the involvement of LGI genes in human epilepsies
56
5.1 Abstract 5.2 Introduction
5.3 Materials and methods 5.4 Results
5.5 Discussion
57 57 59 61 74
CHAPTER VI: Analysis of the very large G-protein coupled receptor
gene (Vlgr1/Mass1/USH2C) in zebrafish. 79
6.1 Abstract 6.2 Introduction
6.3 Materials and methods 6.4 Results and Discussion
80 80 81 84 Summary
Zusammenfassung General References
90 93 96
CHAPTER 1
General Introduction
Retinoic acid metabolism
Vitamin A (retinol) and its derivatives, the retinoids, play several important roles during vertebrate embryonic development. Many organs and tissues are dependent on vitamin A during their development, including the hindbrain (Begemann and Meyer 2001), the spinal cord (Grandel et al., 2002) the heart (Xavier-Neto et al., 2001), the pancreas (Stafford and Prince 2002), the branchial arches (Marc et al., 2004), the eye (Drager et al., 2001) and the forelimb (Lee et al., 2004).
Vertebrates are not able to synthesise vitamin A de novo. Therefore it has to be taken up from dietary sources. This dependency upon vitamin A supply has led to a great storage capacity in the liver, so that the adult organism is not affected by changes in daily vitamin A supply. This does not occur in developing embryos and as a result embryos are highly sensitive to an excess or a deficiency in retinoids, which in both cases lead to defects in patterning and differentiation.
Retinoic acid (RA), the developmentally active derivative of vitamin A, is generated in a two- step metabolic pathway; first retinol is converted to retinaldehyde by alcohol dehydrogenases and it is then subsequently converted to RA by retinaldehyde dehydrogenases. In the amniote embryo, three enzymes, Raldh1 (Aldh1a1), Raldh2 (Aldh1a2) and Raldh3 (Aldh1a3), are capable of synthesising RA. Upon synthesis, signaling through RA and other retinoids is mediated through nuclear ligand-activated transcription factors, the RA receptors (RAR) α, β and γ, which dimerize with the retinoid X receptors (RXRs) α, β and γ, thereby modulating transcription in cells of target tissues (Chambon, 1996).
One of the first opportunities to study RA signaling in amniote embryos was to study nutritional vitamin A deficiency (VAD), which occurs when the mother is on a strict diet depleted of retinol leading to many defects in pattern formation, including specific malformations of the face, eyes, heart, central nervous system (CNS), limbs and pharyngeal arches known as VAD syndrome (Wilson et al., 1953; Stratford et al., 1996). Study of the VAD syndrome shed light on the general function of RA, but it could not tell us about the role of particular enzymes involved in the biosynthesis of RA. To this end individual knock-outs (KOs) in mice were made for all three Raldh enzymes. In zebrafish KO are not yet technically available. Luckily a chemical induced mutation was found by Begemann and colleagues (2001), where the disruption of the raldh2 enzyme leads to a probable loss-of-function of this enzyme.
This neckless (nls) mutant displays most of the characteristics of VAD syndrome embryos, with the exception of a milder hindbrain phenotype.
Figure 1. Approaches to inhibiting retinoic acid mediated gene regulation in the developing hindbrain.
The supply of retinoic acid to the developing embryo can be restricted by depletion of nutritional sources of vitamin A. Subsequent biosynthesis of retinoic acid (RA) involves Retinaldehyde dehydrogenase 2 (Raldh2). Retinoic acid binds to and activates nuclear RARs, which form heterodimers with RXRs to regulate the expression of target genes. Downstream of vitamin A uptake, the signaling cascade leading to RAR activation is blocked in Raldh2 mutants, while activation of RAR signaling can be inhibited by RAR-specific antagonists. Adapted from Begemann and Meyer (2001).
Due to new pharmacological compounds, it is now possible to interfere with RA signaling without working with complex genetic techniques. The RA pathway can be chemically blocked in two ways. Firstly the biosynthesis of RA can be abolished by using diethylaminobenzaldehyde (DEAB), a competitive reversible inhibitor or retinaldehyde dehydrogenases (Perz-Edwards et al., 2001). Secondly, the RARs can be blocked using different pan-RAR antagonists: BMS493 or Ro 62-8175 (Figure 1) leaving the biosynthesis of RA unaffected. Moreover, ectopic RA signaling can easily be performed by adding RA to the liquid medium for zebrafish embryos or by inserting beads soaked with RA into avian embryos.
Retinoic acid and hindbrain development
The developing central nervous system (CNS) is particularly sensitive to variation in the amounts of retinoid signaling (Maden, 2002). Excess retinoid signaling causes patterning defects in the hindbrain and the anterior spinal cord by transforming anterior neural tissue towards a more posterior neural specification, a phenomenon known as posteriorization (Durston et al., 1998). Reduction or absence of RA signaling leads to the anteriorization of a shorter hindbrain (Niederreither et al., 2002; Begemann et al., 2001;Grandel et al., 2002). During normal development, the vertebrate hindbrain transiently forms distinct segments called rhombomeres
(r). Each rhombomere has an individual identity as exemplified by gene expression patterns. In amniote VAD embryos, the caudal hindbrain is transformed towards more anterior fates resulting in enlarged anterior rhombomeres and a truncation of rhombomere posterior to r3 (Maden et al., 1996). In the zebrafish nls mutant, the neural patterning defects are comparable to milder forms of VAD rat embryos (White et al., 2000). It is unlikely that nls is a hypomorphic mutation since the non-neural developmental defects, in the nls branchial arch skeleton and pectoral fins, are stronger or as strong as in full VAD and Raldh2-/- mice, and morpolino- mediated knockdown of the raldh2 enzyme leads to exact phenocopies of nls (Begemann et al., 2001; Grandel et al., 2002). This raises the possibility that zebrafish may possess additional retinaldehyde dehydrogenases that can partially compensate for the loss of RA signaling during neural patterning. In order to study the effect of a loss of RA signaling during neural patterning in zebrafish, Begemann and colleges blocked the RA signaling pathways using different pharmacological compounds: a RAR pan-antagonist (BMS493) and DEAB. The authors observed a much stronger anteriorization phenotype in the hinbrain when wild type embryos were treated with DEAB than in nls embryos (Figure 2). This last finding clearly demonstrates that in zebrafish, the raldh2 enzyme is not the only enzyme involved in the biosynthesis of RA during neuro-ectoderm patterning.
Figure 2. Hindbrain patterning defects in response to alterations in retinoic acid signaling.
Expression of krox-20 in (r) 3 and r5, and hoxb4a in the neural tube, posterior to the r6/r7 boundary, in 20 hpf wild type embryos (A); posterior expansion of r3-5 in nls mutant embryos, hoxb4a expression is present, but often weaker (B); 10 AM of the pan-RAR antagonist BMS493 phenocopies posterior expansion of r3-5 rhombomere boundaries in nls; hoxb4a expression is present, but extends less posteriorly than in wt or nls (not shown) (C). Five-micromolar DEAB anteriorizes the hindbrain further than observed in nls or BMS493-treated embryos (note expansion of r3 and r4), markers of pre- rhombomeric fates posterior to the r4/r5 boundary are reduced (arrows) (D); treatment with 10 nM RA results in variable levels of posteriorization, as indicated by the reduction of r3; at higher doses (not shown), r3 expression is undetectable (E); simultaneous treatment with 10 nM RA and 5 µM DEAB rescues normal hindbrain development, and restores hoxb4a expression in the spinal cord (F). Expression of pax2a delineates the increase in distance (indicated by bracket of identical length in G, H) between the midbrain – hindbrain boundary (mhb) and the otic vesicle (ov) in wt (G) and nls (H) embryos at 20 hpf;
in DEAB-treated embryos (I) expression in the otic vesicles is reduced, sometimes to complete absence, while the mhb appears unaffected. Arrowheads denote visible rhombomere boundaries. Abbreviations:
os: optic stalk, sc: spinal cord; dorsal (A–F) and lateral (G– I) views, anterior to the left. Adapted from Begemann et al. (2001).
By blocking the RA signaling pathway at different times during embryonic development, Grandel et al., (2002) established that RA signaling is necessary during pre-segmentation stages for proper patterning of the posterior hindbrain, as well as proper expression of the spinal cord markers hoxb5a and hoxb6b.
In the zebrafish raldh2 mutant, neckless (nls), the distance between otic vesicle and somites is severely reduced. This phenotype has been contradictorily interpreted as a posterior expansion of the hindbrain, and concommitantly, of the otic vesicle. Indeed, nls embryos do show signs of a mild posterior expansion of neural fates in the posterior hindbrain and anterior spinal cord. Alternatively, the paraxial mesoderm may be affected by changes in retinoid signaling, so that somitic fates may be realized more anteriorly than under wild type levels of RA signaling (Begemann et al., 2001; Grandel et al., 2002).
Loss of RA signaling has an effect on mesodermal derivative tissue of the head, the branchial arches (BA). In mouse, Raldh2 homzygous mutant and in VAD quails BA2 and posterior BAs are absents (Niederreither et al., 1999; Maden et al., 1996) while in the zebrafish nls mutant the formation of BA2 is mildly affected and posterior BAs are lost (Begemann et al., 2001; Grandel et al., 2002). These results demonstrate that head mesoderm is dependant on RA for BAs formation but the contribution of mesodermal cells in BAs development is currently unknown (Mark et al., 2004).
Retinoic signaling in pectoral fin development
Vertebrate limb development is dependent on RA. In Raldh2 homoyzgous mutant in mice and in nls zebrafish embryos, the forelimbs (or pectoral fins) fail to form (Niederreither et al., 2002; Begemann et al., 2001). The timing of RA requirement for pectoral fin development in zebrafish has been proposed to precede to the beginning of somitogenesis (Grandel et al., 2002), while the cellular source of RA in pectoral fin development remains unknown. Additionally, ectopic RA can induce limb duplications in zebrafish and mouse (Vandersea et al., 1998;
Niederreither et al., 1996). Furthermore, local application of RA at the anterior margin of the chick forelimb bud results in mirror image duplications of digits (Helms et al., 1996). The same results are observed if the zone of polarizing activity (ZPA), the mesoderm at the posterior limb margin, whose removal severely truncates limb outgrowth and whose anterior ectopic grafting generates AP pattern duplication, is transplanted into the anterior limb margin (Tickle et al., 1982).
RA is not solely required for forelimb specification but is also involved during limb outgrowth. Along the AP axis, inhibition of RA signaling prevents the formation of a proper ZPA by disrupting the expression of sonic hedgehog (shh), a secreted molecule both colocalized and sufficient to substitute for ZPA function (Riddle et al., 1993) The expression of fgf8 within the ectoderm of the apical ectodermal ridge (AER) was found not to be affected by reduced RA signaling indicating that fgf8 and the development of the AER might not be under the control of RA (Lu et al., 1997). By providing RA within maternal food, Niederreither and colleagues (2002) found both a stage- and dose-dependency for rescue of forelimb growth and patterning.
Following RA supplementation from E7.5 to 8.5, mutant forelimbs are markedly hypoplastic and lack anteroposterior patterning, with a single medial cartilage and 1-2 digit rudiments. RA provided until E9.5 significantly rescues forelimb growth, but cannot restore normal AP patterning. Downregulation or ectopic anterior expression of Fgf4 is also seen. As a result, genes such as Bmp2 or Hoxd genes are expressed symmetrically along the AP axis of the forelimb buds. The authors suggested that RA signaling cooperates with a posteriorly restricted factor
such as dHand, to generate a functional ZPA. Using the same strategy in providing RA within maternal food, limited to E8 during limb field establishment, Mic and collaborators (2004), have shown that the rescued forelimb obtained at E10 display a significant growth defect associated with a smaller AER, referred to as an apical ectodermal mound (AEM). In these RA-deficient forelimbs, a ZPA expressing Shh forms, but it is located distally adjacent to the Fgf8 expression domain in the AEM rather than posteriorly as normal. AER formation in Raldh2 -/-forelimbs is rescued by continuous RA treatment through E10, which restores RA to distal ectoderm fated to become the AER.
These last two papers clearly demonstrate that RA is involved in forelimb development even if both team do not find the same result leading to different roles of RA in AP and proximo- distal patterning.
Non-ion channel epilepsy: new avenues in epileptic diseases
In the second part of my doctoral studies I was interested in non-ion channels epilepsy and to develop the zebrafish as a vertebrate model organism to study such diseases. Until recently, all genes found to be mutated in hereditary idiopathic epilepsies encoded sub-units of ion-channels expressed in the brain. Although, there is considerable variation in the classes of ions channels involved, including voltage-gated sodium or potassium channels as well as acetylcholine and GABA receptors, a classification of idiophatic epilepsies as channelopathies appeared to be warranted (Surtees 2000; Steinlein 2004). The LGI1 (Leucine-rich, glioma inactivated gene 1) gene, apparently not coding for a channel but rather a secreted molecule, was the first reported exception to this rule when different research teams demonstrated that a mutation in the LGI1 gene was the cause of autosomal-dominant partial epilepsy with auditory features (ADPEAF) (Kalachikov et al., 2002; Morante-Redolat et al., 2002: Gu et al., 2002a).
This rare, mild form of familial temporal lobe epilepsy (TLE) is characterised by auditory auras preceding complex partial and secondarily generalized seizures in many, but not all subjects (Winawer et al., 2000). To this date, the disease mechanism of the LGI mutations remains unknown and may reveal new aspect of epilepsy pathologenesis.
LGI1 was originally identified and cloned from the t(10:19) breakpoint of the T98G glioblastoma cell line (Chernova,1998). Human LGI1 maps to 10q24, a region frequently showing loss of heterozygoty in malignant gliomas (Bigner et al., 1990). Neither chromosomal rearrangement nor mutations affecting LGI1 could be identified in most of the glioma cell lines studied so far (Krex 2002; Somerville 2000). Nevertheless, the expression of LGI1 seems to be often reduced in glioma cell lines and malignant brain tumors (Chernova 1998;Krex 2002;
Besleaga 2003).
LGI1 was found to be a member of a gene family, which has four members in human, LGI1, LGI2, LGI3 and LGI4 (Figure 3) (Gu et al., 2002b). LGI genes contain four and a half leucine rich repeats (LRRs) in the N-terminal part of the encoded proteins and align closely with other LRRs containing proteins that bind nerve growth factor and other neurothrophins (Kalachikov et al., 2002; Gu et al., 2002b). Furthermore, all LGI genes in mammals possess seven epitempins (EPTP) repeats in the C-terminal half of the protein. Human LGI proteins share 65-75% of their sequence with each other. Prior to this work, LGI genes had only been identified in mammals.
Figure 3. Dendrogram showing the predicted relationships within the subfamily of LGI genes and between LGI genes and other LRR genes.
Representative genes were chosen from the LRR gene classes I (decorin), II (lumican) and III (opticin).
Horizontal distances of bars are proportional to the evolutionary distances and based upon protein sequences. Adapted from Gu et al. ( 2002b).
The genomic localizations of human LGI2, LGI3 and LG4 are associated with several other epilepsy syndromes and malignancies, while RT-PCR experiments have shown that all human LGI genes are expressed in adult brain (Gu et al., 2002). It was therefore suggested that all LGI genes should be considered as potential candidate for epilepsies. Despite a lot of effort to identify mutations in LGI genes, besides LGI1, in ADPEF, no mutations in LGI2, LGI3 or LGI4 were found to account for families suffering from TLE (Berkovic et al., 2004).
LGI genes and the MASS1/VLGR1 genes share EPTP domains
In 2001, Skradski and colleges isolated the MASS1/VLGR1 (Very Large G-protein coupled Receptor-1) gene, which is mutated in the Frings mouse, a monogenic animal model for epilepsy evoked by auditory stimuli. At 6307 amino acid residues in human, MASS1/VLGR1 is the largest known cell surface protein (McMillan et al., 2002). The MASS1/VLGR1 belong to the
class of G-protein-coupled receptors. There are more than 1000 known genes for such receptors, encompassing as many as 3% of all human genes. MASS1/VLGR1 has a very large putative ectodomain of a unique structure, consisting mostly of 35 copies of a repeat motif of approximately 120 amino acid residues (Nikkila et al. 2000) shared with Na+/Ca2+ exchangers termed Calx- motif (Nicoll et al. 1990). Interestingly, the MASS1/VLGR1 was found to have seven EAR/EPTP repeats, including one highly degenerated, similar to the LGI genes. A third EAR/EPTP protein, an anonymous database entry (GenBank Accession no. BC021197) is characterized by the presence of an N-terminal thrombospondin-N domain (TSP-N). Based on the analysis of EST sequences from various organisms and taking into account the corresponding genomic database entries, it has been speculated that the BC021197 database entry is incomplete. A longer sequence of the BC021197 genes was identified and named TSPEAR by Scheel et al, (2002) to reflect the domain structure. It contains seven adjacent copies of the EAR/EPTP repeat (Figure 4).
LGI (LGI1; LGI2; LGI3; LGI4)
TSPEAR
MASS1/VLGR1 Figure 4. Domain architecture of the three classes of EAR proteins.
The proteins are drawn approximately to scale. Signal peptides are indicated by small boxes at the N termini of the sequences. Other domains are indicated by variably shaded boxes containing abbreviated domain names. Membrane-spanning regions are indicated by black bars. The following domain abbreviations are used: L, leucine-rich repeat; E, EAR repeat; e, highly degenerate EAR repeat; TSP-N, thrombo-spondin N-terminal domain; CX, Calx-b domain; LTP, LamG/TSP-N/Pentaxin homology domain; G, GPS domain. Adapted from Scheel et al. (2002).
Several lines of evidence suggest that MASS1/VLGR1 may play an important role in the development of the vertebrate nervous system. First, it is expressed at high levels in the developing mouse central nervous system, particularly in the ventricular zone, which is the site of neurogenesis for the developing cortex. Expression decreases in concert with the narrowing of the ventricular zone and termination of neurogenesis (McMillan et al. 2002). Second, mutations in VLGR1 have functional consequences in the nervous system including febrile and afebrile
seizures in humans (Nakayama et al. 2002), and Usher syndrome type II C (sensorineural deafness and retinitis pigmentosa) in humans as well (Weston et al. 2004). Prior to this work, MASS1/VLGR1 had only been identified in human and mice.
Aim of this work
Previous studies in chick/quail and zebrafish have shown the importance of RA signaling in posterior hindbrain patterning (Begemann and Meyer 2001) but if RA plays a role in the patterning of the head mesoderm flanking the neural ectoderm had not been investigated. One of the goals of my doctoral studies was to characterised a novel gene expressed in the posterior cranial mesoderm during zebrafish embryonic development and to use this marker and other markers already known to be expressed in posterior head mesoderm to investigate the putative role that RA might play in head mesoderm patterning. Furthermore, the timing of action of RA in head mesoderm patterning was investigated. raldh2, the main enzyme involved in the biosynthesis of RA during embryonic development is widely expressed in the mesendoderm during gastrulation and in somitic mesoderm from early somitogenesis onwards (Begemann et al., 2001, Grandel et al., 2003). By blocking the RA pathway using specific pharmacological compounds I address the question when is RA required to pattern the posterior head mesoderm and compare this finding with previous finding about the timing of RA requirement for neural ectoderm patterning (Grandel et al., 2002).
Another topic addressed during this study was the source and the timing of RA signaling in pectoral fin development. Several authors have shown in a variety of organisms that a lack of RA signaling leads to the loss of forelimbs (see Lee et al., 2004 for review). The temporal requirement of RA signaling for pectoral fin field specification was investigated using two different approaches. One was to rescue the pectoral fin field by the addition of exogenous RA to developing of nls embryos, which lack a functional raldh2 enzyme. The other approach was to inhibit RA signaling in wild type zebrafish embryos with pharmacological compounds, which either antagonised the RAR or abolished the biosynthesis of RA.
When the timing of RA signaling was resolved, the question about the source of RA signaling for pectoral fin development was addressed. At the time when RA is required for pectoral fin development, the raldh2 enzyme, is solely expressed in intermediate and paraxial mesoderm in the trunk. Taking advantage of the spt/ntl double mutant, which lacks paraxial mesoderm but not intermediate mesoderm, I was able to show which cells are the source of RA for pectoral fin field specification. Moreover, by transplanting wild type cell into a nls host embryo I was able to monitor the range of RA signaling to induce pectoral fin specification.
Another aspect of my doctoral dissertation was to study the expression and the evolution of genes involved in non-ion channel epilepsy. In order to better understand the function and the evolution of these new genes involved in epilepsies, I cloned and characterised the expression pattern of all these genes in zebrafish embryonic development. When available, I compared the expression of these genes in fish and mice. Moreover, using different bioinformatic tools, I determined the mode of evolution of the different sub-families of genes belonging to the LGI gene family.
CHAPTER 2
Cloning of zebrafish T-box genes tbx15 and tbx18 and their expression during embryonic development
Published in Mechanisms of Development (2002) 114:137-141.
Cloning of zebrafish T-box genes tbx15 and tbx18 and their expression during embryonic development
2.1 ABSTRACT
Members of the T-box (tbx) gene family encode developmentally regulated transcription factors, several of which are implicated in human hereditary diseases. We have cloned the paralogous genes tbx15 and tbx18 in zebrafish and have characterised their expression in detail. tbx15 is expressed in paraxial head mesenchyme and its derivatives, the extraocular and jaw musculature and the posterior neurocranium. Further areas of tbx15 expression are in the anterior somitic mesoderm, in periocular mesenchyme and in the pectoral fin mesenchyme throughout larval development. Areas of strong tbx18 expression are found in the developing somitic and presomitic mesoderm, in the heart and in pectoral fin mesenchyme, as well as the ventral neuroectoderm and the developing palate. Both genes exhibit particular differences in expression compared to their murine orthologs.
2.2 MATERIALS AND METHODS Fish stocks
Breeding zebrafish of the London wild type strain were reared and staged at 28.5˚C according to Westerfield (1995).
Cloning of T-box genes
Partially degenerate primers were synthesised to a portion of the T-box with regions of amino acid identity between mouse and human Tbx proteins (HEVGTEM: 5'-ccaagacctcga gca t/cga a/ggt igg iac iga a/gat g-3'; PFAKGFR: 5’-gcagttatcga tcg/t a/gaa icc c/ttt igc a/gaa igg-3’) that contained XhoI and ClaI restriction sites, respectively, for subcloning of the 544 bp amplification product into pBluescript SK. RT-PCR was performed with these primers on first strand cDNA, prepared from 48 hpf wild type embryos using a kit (Pharmacia Biotech). PCR conditions: 36 cycles of 1 min. 94˚C, 2 mins. 54˚C, 1.5 mins. 72˚C, followed by an extension step of 7 mins. 72˚C. Sequencing and phylogenetic analyses identified two clones as homologs of tbx15 and tbx18 in humans and mouse. To isolate the 3'-sequence of tbx15 primer 5'-gtt cct act gga gaa gga gtg aag ac-3' was used for 3'-RACE on 24hpf cDNA following the manufacturer's protocol (Smart RACE Kit, Clontech). The tbx18 fragment overlapped with a zebrafish EST (GenBank accession number AI666969) encoding a partial tbx18 clone. The 3'-end of tbx18 was amplified by 3'-RACE with primer 5'- ttc gct ctc cgc aga ctc c-3' from 75hpf cDNA.
In situ hybridisation
Whole mount in situ hybridisation was performed as described previously (Begemann and Ingham, 2000).
Phylogenetic analysis
Amino acid sequences of the alignable T-box domains from members of the Tbx1- subfamily were aligned using the ClustalX program (Thompson et al., 1997). A tree was constructed based on Poisson-corrected distances using the neighbour joining algorithm implemented in TREECON (Van de Peer and De Wachter, 1993).
2.3 RESULTS AND DISCUSSION
T-box (tbx) genes encode a large family of transcription factors, whose importance in development is emphasised by the identification of several mutations in tbx genes linked to human hereditary syndromes (reviewed in Papaioannou, 2001; Smith, 1999). To identify tbx genes that are expressed during limb and head development, we cloned two novel zebrafish T- box genes which we call tbx15 and tbx18 (Fig. 1A). Both genes are members of the Tbx1- subfamily, of which so far only tbx20, also called hrT, had been reported in zebrafish (Ahn et al., 2000; Griffin et al., 2000). Phylogenetic analyses confirmed their orthology to murine and human Tbx15 and Tbx18, respectively (Fig. 1B) (Agulnik et al., 1998; Kraus et al., 2001; Yi et al., 1999).
Figure 1.Sequence alignment and phylogeny of chordate Tbx15/18 genes.
(A) Alignment of predicted amino acid sequence of Tbx15 and Tbx18 orthologs from zebrafish (dr; GenBank accessions AF448504 and AF448503), human (hs; Tbx15 assembled from human contigs AL390796, AL446025 and AL357045; SwissProt 095935), mouse (mm; AAC32316;AAG48598) and Amphioxus (bf; AAG34891). Because the isolated sequences of zebrafish tbx15 lack the 5’-end of the gene, the 171 common amino acids within the T-box were aligned for phylogenetic analyses. The alignment was performed using PILEUP; periods depict identical amino acids. (B) Phylogenetic tree of the Tbx1-subfamily, aligned using amino acid sequences as shown in (A), and rooted with human Tbx2.
Note that zebrafish tbx15 and tbx18 are orthologs of the mammalian genes. Reliability of internal nodes was tested using 1000 bootstrap replications; values are presented above particular branches. dr: Danio rerio; hs: Homo sapiens; mm: Mus musculus; dm: Drosophila melanogaster; bf: Branchiostoma floridae
Both genes, together with human Tbx22 (Laugier-Anfossi and Villard, 2000), constitute a group with a single homolog in amphioxus, tbx15/18/22 (Ruvinsky et al., 2000). Human Tbx15 shows linkage with acromegaloid facial appearance, a dominant human syndrome that affects many of the tissues that express Tbx15 in the mouse (Agulnik et al., 1998).
Expression of tbx15 is first detected at 13 hours post fertilisation (hpf) bilaterally in paraxial head mesenchyme (not shown). Expression increases in the paraxial head mesenchyme by 17 hpf, flanking rhombomere 3 and extending posteriorly to the trunk mesoderm. Expression is excluded from the notochord and otic vesicles (Fig. 2A-C). tbx15 is expressed segmentally in somites in a medial position lateral to the notochord within the first 6-8 somites at 17 hpf (Fig.
2D,E). Expression then expands to the entire somite at 27 hpf, but is absent from tail somites at all developmental stages (Fig. 2F and not shown).
By 24 hpf the the first pharyngeal arch mesenchyme, and a layer of extraocular mesenchyme surrounding the eye, begin to express tbx15 (Fig. 2G,H). During the second day of development expression in the paraxial head mesenchyme extends dorsally, anterior and posterior to the ear and resembles the cartilaginous condensations of the posterior neurocranium (Fig. 2I,J). Fate mapping and extirpation studies in other vertebrates suggest that the posterior
neurocranium is of mesodermal origin (Langille and Hall, 1989; Le Lievre, 1978), thus tbx15 may play a role in the development of the neurocranial commissures and auditory capsules.
Strong expression is detected in extraocular and dorsal jaw muscles as early as 36 hpf, particularly in the medial rectus and adductor mandibulae, and is observed in all pharyngeal muscles of the mandibular and hyoid arches at 60 hpf (Fig. 2K). Within the developing fins, tbx15 is expressed in the pectoral fin bud mesenchyme starting at 27 hpf, prior to distal outgrowth of the fin bud proper. Throughout pectoral fin outgrowth expression remains strong in the entire fin bud mesoderm (Fig. 2L-N). Areas of strong tbx15 expression are found in larval fins at 30 days post fertilisation (dpf) (8.5 mm body length) in the distal tips of lepidotrichs (Fig.
2O). Pelvic and unpaired fins were not assayed for expression.
Figure 2. Expression of tbx15 during wild type zebrafish development.
(A,B) Double in situ hybridisation showing tbx15 expression (purple) at 17 hpf in the paraxial head mesenchyme, extending rostrally underneath and lateral to rhombomere 3 (r3), counterstained with krox-20 in r3 and r5 (red); (B) dorsal view. (C) At 18 hpf expression is restricted to the paraxial mesenchyme, but excluded from r5 and otic vesicles. (D,E) Somitic expression at 17 hpf in 10-11 rostral somites in the region of the horizontal myoseptum (arrows) expands to the entire somite at 27 hpf (F);
(E,F) transverse sections at the level of the fifth somite. (G) Expression is detected in the first pharyngeal arch (at 24 hpf) and (H) in mesenchymal cells separating forebrain and eye. (I,J) tbx15 expression in extraocular and pharyngeal arch muscles, paraxial head mesoderm and pectoral fin bud mesenchyme at (I) 30 hpf and (J) 48 hpf. (K) Strong expression in eye and pharyngeal muscles; ventro-lateral view at 60 hpf. (L-O) tbx15 transcript in developing pectoral fins; expression is exclusively mesodermal, and at 36 h (L) is strong in the medial part of the fin bud and absent from the epidermis (arrowhead); mesodermal expression remains strong throughout the fin bud at (M) 48 hpf and (N) 3 dpf; (O) At 30 dpf tbx15 is strongly expressed in the margins and distal tips of lepidotrichs (arrowheads). (A,D,K-N) lateral views, (B,G,I,J) dorsal views, (C,E,F,H) transverse sections, (O) ventral view. am, adductor mandibulae; eom, extraocular muscles; fb, forebrain; hb, hindbrain; io, inferior oblique; lr, lateral rectus; n, notochord; ov, otic vesicle; pfb, pectoral fin bud; phm, paraxial head mesenchyme; pm, pharyngeal arch mesenchyme; r, rhombomere; sc, spinal cord; Scale bars, 100 µm.
Expression of tbx18 is first detected from 11 hpf onwards, shortly after the onset of somitogenesis, in the somitic and presomitic mesoderm (not shown). At 19 hpf expression in the somitic mesoderm is strong in the anterior half and medial aspect of posterior somites. More mature anterior somites express tbx18 in dorsal and ventral regions of the somites, where they are restricted to the somitic mesoderm proximal to the notochord (Fig. 3A-C). tbx18 expression gradually ceases within the somites in an anteroposterior direction, such that expression at 29 hpf is restricted to the posterior 13-15 somites and is absent from somites by 40 hpf. Interestingly, from 17 hpf onwards, expression of tbx15 gradually appears in place of tbx18 in the somites (compare Figs. 3C and 2E,F). At 24 hpf, tbx18 is expressed in two bilateral domains of single- cell width, extending from within the mesencephalon posteriorly into rhombomere 2. The timing and neuroectodermal location of this expression identify this tissue as the developing medial longitudinal fascicles (Fig. 3D,E). At 30 hpf tbx18 transcript is detected in the paraxial head mesenchyme, flanking the otic vesicles medially and dorsally (Fig.3F,G). This expression is strongly maintained at least up to 3 dpf (Fig. 3L).
tbx18 expression in the heart is detected from 36 hpf onwards in the sinus venosus, atrium and ventricle. In addition, strong expression is found in the septum transversum from 48 hpf onwards (Fig. 3H-J). tbx18 is strongly expressed in the facial area anterior to and surrounding the mouth and in an area lining the olfactory epithelium (Fig. 3 K,K’). From 40 hpf onwards the developing palate strongly expresses tbx18 (not shown and Fig.3 K,L).
In developing pectoral fins, tbx18 expression is not detected prior to 27 hpf. During outgrowth of the apical fold, expression is restricted to the median part of the fin bud mesenchyme, and then expands to the entire mesenchyme. Expression of tbx18 is later restricted to the margin of the pectoral fin buds and is expressed in the interradial zones of the pectoral fins of 30 day old larvae (Fig.3 M-Q).
Figure 3. Expression of tbx18 during wild type zebrafish development.
(A-C) Somitic expression of tbx18. (A) At 19 hpf expression is strong in the anterior halves of posterior somites (arrowhead); more mature somites express tbx18 above and below the horizontal myoseptum (arrow). (B) Higher magnification, 24 hpf; expression is restricted to the median part of each somite; a transverse section (C) at the level indicated by the arrow reveals the restriction of expression to the somites. (D,E) The ventral neuroectoderm expresses tbx18 at 24 hpf in the medial longitudinal fascicles; r3 is indicated by krox-20 expression. (F,G) tbx18 transcript is detected at 30 hpf in the paraxial head mesenchyme, flanking the otic vesicles medially and dorsally. Note expression in the pectoral fin mesenchyme. (H-J) tbx18 transcript in the developing heart at 36 hpf (H), 48 hpf (I) and 4 dpf (J); strong expression is observed in the septum transversum (arrow); arrowheads point to sites of tbx18 expression in atrium and sinus venosus. (K,K’) Ventral views of the head at 40 hpf; tbx18 is expressed anterior to and surrounding the mouth (bracket indicates region depicted magnified in a different embryo in K’) and in an area flanking the olfactory epithelium; (L) Ventral view of the head of a 3 day-old embryo (lower jaw removed), showing tbx18 expression in the palate (arrow) and in paraxial mesoderm between notochord and ear. (M-Q) tbx18 expression in developing pectoral fins; expression is initiated shortly before 30 hpf (M) in the median part of the fin bud mesenchyme and expands to the entire mesenchyme by 48 hpf (N); (O) at 3 dpf expression is stronger in the anterior mesenchyme and the mesenchymal margin; at 4 dpf (P) expression is restricted to cells bordering the non-expressing apical fold; (Q) pectoral fin at 30 dpf; expression of tbx18 in the interradial zone (asterisks); dark spots on the pectoral fin rays are melanocytes. Distal is to the left, anterior to the top.(C,E,G) transverse sections, (A,B,D,H,I,M-P) lateral
views, (J-L,Q) ventral views, (F) dorsal view. a, atrium; hb, hindbrain; m, mouth; mb, midbrain; mlf, medial longitudinal fascicle; n, notochord; oe, olfactory epithelium; ov, otic vesicle; pfb, pectoral fin bud;
r, rhombomere; sep, septum transversum; sc, spinal cord; sv, sinus venosus; Scale bars, 100 µm.
Expression of zebrafish tbx15 in the cranial paraxial mesoderm, the head musculature and the somites are novel features not reported in the mouse. However, expression surrounding the eyes, in the branchial region and in forelimbs are reminiscient of the mouse expression pattern (Agulnik et al., 1998). Likewise, we find both similarities and differences in expression between the orthologs of tbx18. Somitic expression of tbx18 is not restricted to the sclerotome in zebrafish, and cranial paraxial expression is not observed during somitogenesis stages in zebrafish. Moreover, expression of tbx18 was not observed in the central nervous system and in the palate of the mouse (Kraus et al., 2001).
ACKNOWLEDGEMENTS
We thank John S. Taylor for help with the phylogenetic analysis and Rita Hellmann for technical assistance. This work was supported by a Marie-Curie Fellowship (FMBICT960675) to G.B., by the University of Konstanz, Fonds der Chemischen Industrie, and grants from the Deutsche Forschungsgemeinschaft to A.M., and a BBSRC research grant (50/G12140) to P.W.I.
CHAPTER 3
Retinoic acid is essential for antero-posterior patterning of cranial mesoderm in zebrafish
Retinoic acid is essential for antero-posterior patterning of cranial mesoderm in zebrafish
3.1 ABSTRACT
Retinoic acid (RA) has been identified as a key signal involved in the posteriorization of vertebrate neural ectoderm. The major enzyme involved in biosynthesis of RA during embryonic development has been shown to be retinaldehyde dehydrogenase 2 (Raldh2). A zebrafish mutant in raldh2 (neckless; nls), which is devoid of full RA signaling during embryonic development, exhibit anterior-posterior (AP) patterning defects in the neural ectoderm patterning (Begemann et al., 2001). Using the nls mutant we found that loss of RA also affects AP patterning of the cranial mesoderm.
We have previously shown that nls mutants retain some level of RA signaling in the neural tube that can be blocked by treatment with pharmacological antagonists of RA production or RA-mediated gene regulation. We depleted RA signaling in embryos using DEAB, an inhibitor of retinaldehyde dehydrogenases, and blocked activation of RA receptors (RARs) using a pan-RAR antagonist and found that markers of the posterior cranial mesoderm are severely shortened along the AP axis, correlating with the severity of RA depletion.
By blocking RA signaling we determined the timing for requirement of RA to establish the AP-level of the posterior border of head mesoderm. Together with the pattern of raldh2 expression, we conclude that during gastrulation, RA biosynthesis in prospective mesoderm is a key signal for the specification of the AP extent of the posterior cranial mesoderm. Furthermore, RA-antagonist experiments further reveal that AP-patterning processes are coordinated betweeen the neural tube and the paraxial mesoderm, aligning the hindbrain-spinal cord and head-trunk mesoderm boundaries.
3.2 INTRODUCTION
Retinoic acid (RA) acts as a signaling molecule with important roles for vertebrate embryonic development (reviewed by Morris-Kay and Ward, 1999). Interfering with RA signaling affects development of a variety of organs development such as the limbs (Lee et al., 2004), the branchial arches (Marc et al., 2004), the heart (Xavier-Neto et al., 2001), the pancreas (Stafford and Prince 2002) and the central nervous system especially the posterior hindbrain (Begemann and Meyer 2001). All these symptoms are known as vitamin A-deficiency (VAD) syndrome (Morris-Kay and Sokolova, 1996). Ectopic application of RA causes patterning defect in the hindbrain by transforming anterior neural tissue into more posterior ones (posteriorization)
(Durston et al., 1989). In the zebrafish, reduction of RA signaling enlarges the anterior rhombomeric segments of the hindbrain and leads to the loss of rhombomeric segments posterior to rhombomere (r) 7 (Begemann et al., 2001).
RA biosynthesis from retinol (vitamin A) requires two oxidative steps with retinaldehyde as the intermediate, which subsequently is converted into RA by retinaldehyde dehydrogenase.
In amniotes, three retinaldehyde dehydrogenases are involved in the biosynthesis of RA: Raldh1 (Aldh1a1), Raldh2 (Aldh1a2) and Raldh3 (Aldh1a3). In mouse, only Raldh2 seems to be involved in early posteriorization of the neural tube (Niederreither et al., 2002). In zebrafish, raldh2, is the only retinaldehyde dehydrogenase enzyme identified so far, while the puffer fish genome contains orthologs of all three amniote Raldhs (Begemann et al., 2004). Identification of the remaining raldhs in zebrafish should be expected in the near future.
In mouse, a loss of function mutation in Raldh2 mimics the most severe phenotype associated with VAD syndrome, which displays a stronger anteriorization of the hindbrain than does a null mutation of the raldh2 enzyme in the zebrafish. This might be due to other raldh enzymes present in fish or due to different modes of action of the raldh2 enzyme between mammals and fish.
The effects of RA and other retinoids are mediated through nuclear receptors of the RAR and RXR families, which act as ligand-activated transcriptional regulators (reviewed by Mangelsdorf et al., 1995). Inactivation of single receptors in mice has revealed extensive receptor redundancy, while compound mutations in some receptors recapitulate the phenotypic defects observed in VAD, including the disruption of AP patterning in the cranial neural crest and hindbrain (Dupe et al., 1999; Kastner et al., 1997) In mammals, three different RARs, RARα, RARβ and RARγ, are detected in complete genomes while in fish only four RARs seem to be present: two RARα, two RARγ and no RARβ (www.ensembl.org and Escriva H., personal communication).
In the zebrafish raldh2 mutant, neckless (nls), the distance between otic vesicle and somites is severely reduced. This phenotype has been contradictorily interpreted as a posterior expansion of the hindbrain, and concommitantly, of the otic vesicle. Indeed, nls embryos do show signs of a mild posterior expansion of neural fates in the posterior hindbrain and anterior spinal cord. Alternatively, the paraxial mesoderm may be affected by changes in retinoid signaling, so that somitic fates may be realized more anteriorly than under wild type levels of RA signaling. Loss of RA signaling has an effect on mesodermal derivative tissue of the head, the branchial arches (BA). In mouse, Raldh2 homzygous mutant and in VAD quails BA2 and posterior BAs are absents (Niederreither et al., 1999; Maden et al., 1996) while in the zebrafish
nls mutant the formation of BA2 is mildly affected and posterior BAs are lost (Begemann et al., 2001; Grandel et al., 2002). These results demonstrate that head mesoderm is dependant on RA for BAs formation but the contribution of mesodermal cells in BAs development is currently unknown (Mark et al., 2004).
Here we have investigated the possibility and show that formation of the posterior part of the cranial mesoderm is dependent on full RA signaling. We also investigate the timing of RA requirement for paraxial head mesoderm patterning. Finally we show by depleting most of RA signaling that AP-patterning processes are coordinated betweeen the neural tube and the paraxial mesoderm, aligning the hindbrain-spinal cord and head-trunk mesoderm boundaries Together with previous studies, these results implicate a role for RA in patterning processes at the head- trunk boundary involving all three germ layers.
3.3 MATERIALS AND METHODS Fish stocks
Breeding zebrafish of the London and Konstanz wild type strains were reared at 28.5°C and staged as described (Kimmel et al. 1995). Development of endogenous pigments was inhibited by exposing embryos to 1-phenyl-2-thiourea (PTU) at a final concentration of 0.2mM.
In situ hybridisation and photography
Whole mount in situ hybridisation was performed as previously described (Begemann and Ingham, 2000) using the following probes: tbx15 (Begemann et al. 2002); col2A1 (Yan et al.
1995) krox20 (egr2b; www.zfin.org), raldh2 (Begemann et al. 2002), val (Moens et al., 1996) myoD (Weinberg et al., 1996). Stained embryos were examined with a Zeiss Axiophot microscope. Images were processed using Zeiss Axiovision and Adobe Photoshop software.
Pharmacological treatments
Batches of wild type embryos were incubated in the dark at 28.5°C from late blastula onwards in varying dilutions (in embryos medium) of a 10-8 M all-trans RA (Sigma) from a 10-2 M stock solution in DMSO. RAR pan-antagonist (Ro 62-8175, Roche) and RAR-alpha antagonist (Ro 41-5253) were applied at late blastula, unless specified otherwise, at a concentration of 2.5.10-6 M (from a 10-2 M stock in DMSO); 3.10-7 M (from a 10-3 M stock in DMSO) and 10-11 M (from a 10-7 stock in DMSO), respectively. DEAB (4- diethylaminobenzaldehyde; Fluka) was applied at a concentration of 10-5 M from a 10-2 M stock in DMSO. As controls, wild type embryos were treated with equivalent concentrations of DMSO.
3.4 RESULTS
Posterior cranial mesoderm development is dependent on RA signaling
To study the effects of RA signaling on cranial mesoderm development in more detail, we analyzed the expression of col2A1, a collagen specifically expressed in posterior head mesoderm (Jan et al.)(Fig. 1A,C) and tbx15, a transcription factor expressed in the developing paraxial head mesoderm (Begemann et al.)(Fig. 1B,D) in wild type embryos and in embryos with altered levels of RA signaling. These two genes are co-expressed in posterior paraxial mesoderm, but differ in their length of expression within the posterior cranial mesoderm. tbx15 is expressed in the posterior most part of head mesoderm while col2A1 is expressed in more anterior cells (Fig 1A,B). Moreover, at 30 hpf tbx15 is also expressed in the somitic mesoderm, allowing for simultaneous distinction of the border between cranial and somitic mesoderm (Fig.
1D). We therefore assayed gene expression at 30 hpf to determine the posterior extent of cranial mesoderm.
In nls embryos, which exhibit an intermediate level of RA signaling in the neural tube (Begemann et al. 2001; 2004), col2A1 and tbx15 are expressed in a shortened domain along the anteroposterior axis compared to wild type siblings at 30 hpf (compare Fig.1C,D and 1E,F respectively.). This difference of expression is more visible using tbx15 as a marker because it marks the entire posterior cranial mesoderm.
Figure 1. Expression of col2a1 and tbx15 in the head mesoderm in wild type and nls embryos. (A,B) Transverse sections at the level of the oitic vesicle (rhombomere 5) showing col2A1 (A) and tbx15 (B) expression restricted to mesodermal cells (C) Expression of col2A1 in a dorsal view of a 30 hpf wild type embryo. Expression is solely detected in the posterior paraxial head mesoderm. (D) tbx15 expression in posterior head mesoderm and in anterior somitic mesoderm at 30 hpf. tbx15 transcripts are
also detected in the pectoral fin bud (pfb). (E,F) col2A1 (E) and tbx15 (F) expression in nls mutant at 30 hpf. Brackets in (C) and (D) mark the length of col2A1 and tbx15 expression in paraxial head mesoderm respectively. Note that these brackets are reproduced in (E) and (F) to emphasize the reduction in length of col2A1 and tbx15 in head mesoderm in nls mutant. All embryos have been fixed at 30 hpf. (A,B) transverse sections. (C-F) dorsal views. hb, hindbrain; nls, neckless mutant; pfb, pectoral fin bud; ov, otic vesicle; wt, wild type.
Over-expression of RA shows an expanded cranial mesoderm while inhibitions of RA signaling display stronger reduction of posterior head mesoderm than the nls mutant
To study the effects of exogenous RA in head mesoderm patterning, embryos were treated at 70% epiboly with RA (Sigma) and investigated at 30 hpf, when tbx15 staining can be distinguished between the posterior head mesoderm and the anterior somites (Fig. 1B).
Morphologically, these embryos display an enlarged hindbrain and larger pectoral fins. Cardiac oedema develops after 48 hpf and the embryos die at 3-4 days post fertilisation (data not shown).
RA treated embryos show an enlarged domain of expression in head mesoderm for col2A1 (compare Fig. 2A and 2B) and tbx15 (compare Fig. 2F and 2G) than their wild type counterpart.
The nls mutation has been shown to induce a complete loss-of-function of the raldh2 enzyme (Begemann et al. 2001). However, we were previously able to show that nls mutants are not devoid of all sources of RA, but retain a basic level of RA activity from yet undiscovered sources other than raldh2 (Begemann et al., 2004). In order to investigate the fate of cranial mesoderm under successively reduced levels of RA signaling, up to complete absence, we pharmacologically blocked RA activity with Ro 62-8175, a pan-RAR antagonist (panRARant.)(Roche), or with diethylaminobenzaldehyde (DEAB), a competitive reversible inhibitor of retinaldehyde dehydrogenases (Perz-Edwards et al., 2001; Begemann et al., 2004).
Wild type embryos were treated with panRARant. or DEAB starting at 5 hpf and investigated at 30 hpf. Morphologically, embryos treated with either inhibitor display all characteristics of complete RA-depletion, including a reduction of the head along the anteroposterior axis, resulting in close apposition of otic vesicle and trunk somites, and failure to form pectoral fins (not shown). At the gene expression level, embryos treated with panRARant. or with DEAB display a reduction of head mesoderm along the AP axis with a stronger effect observed when the embryos are treated with DEAB (Fig. 2 D,E; I,J). All RA signaling is generally thought to be mediated through RARs, therefore this result might indicate that Ro 62-8175 pan-RAR receptor appear to be not as effective in fish embryos as they are in amniote embryos. To test the involvement of RARαs in head mesoderm patterning, we treated wild type embryos commencing from 5 hpf with the RARα antagonist Ro 41-5253. At 30 hpf, a reduction of the col2A1 and tbx15 expression domains along the AP axis is evident (compared Fig. 2C with 2B and Fig. 2H with 3G). Furthermore, tbx15 expression is lost in pectoral fin buds (Fig 2H). Loss
of RA signaling through RARα results in a less severe reduction of the head mesoderm than blocking all RAR signaling with panRARant. (compare Fig. 2C and 2D and Fig. 2H and 2I).
While this suggests that RARαs are required for cranial mesodermal patterning, a role for the remaining receptors cannot be ruled out. As no RARβ have so far been identified in the zebrafish, we conclude that cranial mesodermal development in zebrafish is dependent on both RARα and RARγ.
Figure 2. Head mesoderm patterning is dependant of RA signaling. (A,F) col2A1 (A) and tbx15 (F) expression in RA treated embryos. (B,G) col2A1 (B) and tbx15 (G) expression in wild type embryos. (C,H) col2A1 (C) and tbx15 (H) expression in RARα-antagonist treated embryos. (D,I) col2A1 (D) and tbx15 (I) expression in pan RAR antagonist treated embryos. (E,J) col2A1 (E) and tbx15 (J) expression in DEAB treated embryos. The two horizontal lines in (F-J) indicate the length of tbx15 expression in the head mesoderm in wild type embryo. All embryos have been treated at 5 hpf and fixed at 30 hpf. All embryos are shown in a dorsal view. ov: otic vesicle; s1: level of the first somite
In order to more rigorously test whether a RA signal plays a role in head mesoderm patterning, we treated wild type embryos with DEAB and RA at 5 hpf. Embryos were fixed at 30 hpf and analysed with head mesoderm markers. Embryos treated with DEAB display a significant reduction in posterior head mesoderm visualized with col2A1 (Fig. 3A) and tbx15 (Fig. 3D). When simultaneously incubated in 10-8 M RA, DEAB treated embryos are morphologically and molecularly undistinguishable from wild type (compare Fig. 3B with 3C and Fig. 3E with 3F). Similar results were obtained with a rescue attempt of nls embryo (not shown), demonstrating that RA is the key signal in the specification of the head mesoderm along the AP axis, rather than disparate molecules whose synthesis may be affected by DEAB.
Figure 3. Rescue of DEAB head mesoderm phenotype with exogenous RA. (A,D) col2A1 (A) and tbx15 (D) expression in 10-5 M DEAB treated embryos starting at 5 hpf. (B,E) col2A1 (B) and tbx15 (E) expression in wild type embryos. (C,F) col2A1 (C) and tbx15 (F) expression in 10-5 M DEAB treated at 5 hpf and 5.10-8 M RA treated at 7 hpf embryos. Note that at these concentrations, a full rescue of posterior head mesoderm is observed with both markers (C and F). All embryos are shown in dorsal view at 30 hpf.
Late gastrulation stage patterning of cranial mesoderm through RA
We examined the expression of raldh2 by whole mount in situ hybridisation. raldh2 is first detected at 30% epiboly at the blastoderm margin (Begemann et al., 2001; Grandel et al., 2002) (not shown). At 75% epiboly, raldh2 is expressed in involuting cells at the margin that will form mesendoderm (Fig. 4A) but is excluded from the most dorsal cells of the embryonic shield. Expression at this stage is strong is the lateral hypoblast but is excluded from the epiblast (Fig. 4 A’).
Patterning of both the zebrafish neural ectoderm and the endoderm through RA occurs during late gastrulation stages (Grandel et al., 2002; Kudoh et al, 2002, Stafford and Prince, 2002). We therefore were interested to determine whether the requirement for cranial mesodermal patterning through RA occurs concommitantly with the other germ layers. To this end wild type embryos where treated with panRARant. during various time windows, and investigated for cranial mesoderm development at 30 hpf. Treatment at 80% epiboly (8 hpf) with panRARant.
produces a strong head mesoderm phenotype as visualized by col2A1 expression (Fig. 4B; B’) (100%, n=30). In contrast, treatment commencing at 10 hpf (tail bud stage) and later (11 hpf) appears to have no effect on cranial mesoderm development along the AP axis (93%, n=30 and 100%, n=30 respectively), as such post-gastrulation treated embryos exhibit the same length of col2A1 expression cells along the AP axis compared to wild type (fig. 4C-E’). Treatment with DEAB or with RARα antagonist alone achieve the same results (data not shown). We conclude that late gastrulation stages, after 8 hpf and prior to 10 hpf, are critical for the co-ordinated patterning of all three germ layers at the head-trunk boundary.
Figure 4. Timing of action of RA in head mesoderm patterning. (A) Lateral view of a wild type embryo showing raldh2 expression in the involuting mesendoderm at 75% epiboly. (A’) Higher magnification confirming that raldh2 expression is excluded from the epiblast. (B-E) dorsal views; (B’- E’) lateral views of pan-antagonist treated embryos (B-D’) and wild type embryos (E,E’). The age of the embryos when they have been treated is indicated in the pictures. (B,B’) col2A1 expression in head mesoderm is highly shorter than in wild type embryos (E,E’), while staining appears normal when embryos are treated during, or soon after, the end of gastrulation. Compare C,C’ and D,D’ with E,E’. d;
dorsal side.
Posterior head mesoderm is aligned with the posterior neural tube in DEAB treated embryos In the hindbrain, the expression domain of krox20 (egr2b) at the 20 hpf stage serves as landmarks of segmentation. krox20 is expressed in r3 and r5 (Oxtoby and Jowett, 1993). At 20hpf, tbx15 is expressed in the paraxial head mesoderm posterior to r3 (Fig. 5A) (Begemann et al., 2002). In absence of RA signaling, like in DEAB treated embryos, expression of krox20 is greatly reduced in r5 and only remains in its most anterior part (Fig. 5B, arrowhead). Begemann and colleges have shown that in DEAB treated embryos, rhombomeres posterior to r5 do not form (Begemann et al., 2004). Interestingly, in DEAB treated embryos, the posterior most expression of tbx15 in paraxial head mesoderm matches the end of rhombomeric expression (Fig. 5B). This result suggests that neural tube cells and head mesoderm cells behave in the same way along the AP axis in absence of RA signaling. In other words, the posterior limit of the neural tube along the AP axis is the same as the one of the head mesoderm. When using a somitic mesoderm marker like myoD and a posterior hindbrain marker for r5 and r6, valentino (val), on embryos treated with 10-6 M of DEAB, we observed that the posterior border of val, marked by an arrow coincides with the anterior border of the somitic mesoderm (Fig. 5C,D).
These RA-antagonist experiments reveal that AP-patterning processes are coordinated betweeen the neural tube and the paraxial mesoderm, aligning the hindbrain-spinal cord and head-trunk mesoderm boundaries.
Figure 5. Head mesoderm and neural ectoderm are coordinately patterned by RA along the AP axis. Double in situ hybridisation showing tbx15 expression in paraxial head mesoderm extending rostrally underneath to r3 stained with krx20. (A) 22 hpf wild type embryo with tbx15 expression found in the entire head mesoderm; the posterior most end of tbx15 expression is marked by an arrow. (B) DEAB treated embryos with an enlarged r3 and only the most anterior r5 stained with krx20. Most posterior expression of tbx15 in head mesoderm in DEAB treated embryos is concomitant with the anterior expression of krx20 in r5. (C,D) valentino (val) expression in r5/r6 and somitic mesoderm stained with myoD. (C) Dorsal view of a 14 hpf DEAB-treated (low concentration, 10-6 M) embryo, showing that the posterior border of val, marked by an arrow, coincides with the anterior border of the somitic mesoderm. (D) Same embryo as C in lateral view. A and B lateral view; r3: rhombomere 3; r5: rhombomere 5.
3.5 DISCUSSION
The studies reported here describe the involvment of RA in posterior cranial mesoderm patterning. We have shown that a lack of RA shortened the head mesoderm along the AP axis while exogenous RA expands head mesoderm along the AP axis and that this RA involvement occurs during gastrulation when raldh2, the main RA synthesizing enzyme, is widely expressed in the mesendoderm but excluded from the neural ectoderm. Moreover we were able to show using RA-deficient embryos that the hindbrain-spinal cord boundary is aligned with the cranial mesoderm-somitic mesoderm boundary.
RA patterns the posterior head mesoderm
While it has been shown in several vertebrates that RA is a key signal for posterior hindbrain patterning (Dupe and Lumsden 2001; Begemann et al., 2001; Begemann and Meyer 2001; Grandel et al., 2002), its role in posterior cranial mesoderm patterning remained unclear.
Taking advantage of the nls mutant, which lacks a functional raldh2 enzyme, our study has revealed a reduction, in the AP axis, of the posterior paraxial head mesoderm. When compared
with the Raldh2 homozygous mutant (Niederreither et al., 2000) or with VAD in bird and rats (Gale et al., 1999; Maden et al., 1996) the severity of the posterior hindbrain phenotype in nls resembles a state of reduced but not total loss of RA signaling activity. Application of specific alpha RAR antagonist or pan-RAR antagonist to zebrafish embryos lead to a severe reduction of posterior head mesoderm (Fig. 2B-D,G-I). We have previously shown that an alternative agent, which inhibits RA signaling may be required to investigate the full extent of RA function during neural patterning in zebrafish in a similar fashion as seen in full VAD in amniotes (Begemann et al., 2004). Therefore we applied DEAB to block RA synthesis to study its implication in cranial mesoderm patterning. DEAB is a competitive, reversible inhibitor antagonist that act as substrate of retinaldehyde dehydrogenases, but not on Aldh2 or Aldh3 (Russo et al., 1988; 2002).
Our analysis shows a much severe head mesoderm phenotype when DEAB was applied (Fig. 2E,J), which correlates with the observed neural ectoderm phenotype in zebrafish embryos treated with that chemical (Begemann et al., 2004). When treated with DEAB early on during embryonic development, posterior rhombomeres are enlarged and hindbrain posterior to the anterior border of r5 are not specified anymore (Begemann et al., 2004). To explain the difference observed in pan-RAR antagonist and DEAB treated embryos, the most obvious explaination is that, like when treated with BMS493 (Begemann et al., 2004), not all RA signaling can be blocked using Ro 62-8175. We therefore suggest that Ro 62-8175 is not as effective in fish embryos as it is in amniote embryos leading to some remaining RA activity. The use of more concentrated Ro 62-8175 in the embryonic medium lead to either teratogenic phenotypes or even death of the embryos.
These results clearly demonstrate that nls embryos retain levels of RA signaling in posterior head, which is sufficient for an almost normal posterior cranial mesoderm patterning.
Two hypothesis have been proposed to explain this. First, maternal supply of RA in nls embryo;
the second proposed hypothesis is that other retinaldehyde dehydrodenases may compensate for a dramatic loss of RA signaling (Begemann et al., 20004). It will be therefore of high interest to identify new RA producing enzyme in zebrafish and to investigate the role they play during posterior head mesoderm patterning.
RA signaling is required prior to somitogenesis to pattern the cranial mesoderm
In zebrafish, raldh2 is highly expressed in the paraxial mesoderm during gastrulation (Fig. 4A, A’) and in paraxial and intermediate mesoderm during early somitogenesis (Begemann et al., 2001; Grandel et al., 2002). To investigate the timing of RA siganlling, we used pan-RAR antagonist to inhibit RA signaling within the embryo. By inhibitor treatments during gastrulation stages we are able to phenocopy the posterior cranial mesoderm defects observed in embryos
treated at early gastrulation with the same chemical. In contrast, inhibitor treatments initiated after gastrulation do not noticeably affect posterior head mesoderm markers as observed with col2a1 expression. Prior to this study, RA signaling was found to pattern the neural ectoderm in the posterior hindbrain and the endodermal organs such as the pancreas and the liver prior to somitogenesis (Grandel et al., 2003; Stafford and Prince 2002). These findings implicate RA as an early, global regulator of development that influences, prior to the end of gastrulation, different structures and tissues as neural ectoderm, anterior endoderm and cranial mesoderm.
Reduce RA signaling shows that hindbrain-spinal cord and head-trunk mesoderm boundaries are aligned
Neuromeres are not visible caudal to r7 (Hannemann et al., 1988) suggesting that the caudal hindbrain is composed of two additional segments, yet immunohistochemical labels fail to detect a segmented pattern in this region (Trevarrow et al., 1990). The coordination of segmentation processes in the caudal hindbrain and in the paraxial mesoderm appears to be complex and is not well understood (Kimmel et al., 1988).
A model for the mechanism of early AP patterning of the head mesoderm, based on the observations described in this paper, is presented in Figure 6. A key feature of this model is that patterning of posterior cranial mesoderm is dependent on RA signaling during pre-segmentation stages (Fig. 6A). This proposed model fits well with the model on posterior neuroectoderm proposed by Grandel and collaborators (2003). Our finding demonstrates that AP-patterning processes are coordinated, betweeen the neural tube and the paraxial mesoderm (Fig. 6B).
Figure 6. Model for patterning of the head by RA. (A) Left panel shows the expression of raldh2 of a at 75% epiboly wild type embryo in a lateral view. Schematic diagram of RA produced in the paraxial mesoderm at the blastoderm margin during pre-segmentation stages for anteroposetrior patterning of both the posterior cranial mesoderm and the posterior hindbrain. (B) Schematic diagram of the head/trunk boundary in a wild type and in a RA-deficient embryo at 30 hpf in dorsal view. Our model proposed that the hindbrain-spinal cord boundary is aligned with the cranial mesoderm-somitic mesoderm boundary. cm: cranial mesoderm; sc: spinal cord; sm: somatic mesoderm. The numbers in blue boxes in B indicate the number of individual rhombomere.
One question then remains: which germ layer specify the posterior boder of the head? In other word is the AP extent of the head mesoderm specified by RA and then patterns the AP extent of the neuroectoderm through a sencondary unknown signal or is the AP extebdof neuroectoderm dependent of RA and then specifys the AP extent of the head mesoderm? Or maybe both germ layers can equally respond to RA signaling for the specification of their respective posterior border. Unfortunately we cannot provide yet a definitive answer to this question but as shown in Fig. 6A, raldh2 is solely expressed in the hypoblast implicating that RA is solely produced in the paraxial mesoderm. In the future, a precise localisation of RA a this stage will give answers to these questions.
ACKNOWLEDGMENTS
The authors would like to thank Séverine André and Hector Escriva and for their help during the elaboration of this manuscript. We also thank Kathrin Hoffmann for general technical assistance and Ingrid May for fish maintenance. We are grateful to Roche for the kind gifts of
the Ro 62-8175 and Ro 41-5253 antagonists. This work was supported by a Landesgraduiertenstipendium to Y.G. and grants from the Deutsche Forschungsgemeinschaft to G.B. (BE 1902/3-1) and A.M.
CHAPTER 4
Temporal and spatial requirement for retinoic acid in zebrafish pectoral fin development
