Ceramides constitute the central unit of the diverse class of sphingolipids, which are involved in a variety of cellular and physiological functions (Mullen et al., 2012), e.g., cell proliferation, cell differentiation, apoptosis, and lipid metabolism. There are two major pathways leading to ceramide, namely the de novo synthesis and the salvage pathway, both of which require functions of the ceramide synthase (Cers) gene family.
De novo synthesis occurs in the endoplasmatic reticulum, where serine palmitoyl transferase (SPT) catalyzes the condensation of activated palmitate (palmitoyl-CoA) with L-serine (Mullen et al., 2012), followed by synthesis of sphinganine, mediated by activity of 3-ketosphinganine reductase. Condensation of a CoA activated fatty acid with sphinganine gives rise to dihydroceramide, a reaction catalyzed by ceramide synthases. Finally dihydroceramide desaturase catalyzes conversion of dihydroceramide to ceramide. In the second pathway of ceramide generation, the salvage pathway, existing sphingolipid species are degraded to sphingosine, which serves as a substrate for ceramide generation by ceramide synthases.
Phylogenetic studies have shown that sphingolipids and the enzymes involved in sphingolipid metabolism, like ceramide synthases are evolutionarily highly conserved. Ceramide synthases are transmembrane proteins with five to eight predicted transmembrane helices (Teufel et al., 2009), depending on the prediction algorithm applied. They possess several functional domains and can be subdivided into six groups (Cers1–6), which is reflected by overall sequence homologies as well as protein architecture. A single Cers protein has been identified in yeast (Egilmez et al., 1989) and in fruit fly (Bauer et al., 2009), whereas the mouse and human genomes encode for six Cers protein groups (Teufel et al., 2009). Cers groups 2–6 contain a so called homeobox (HOX) domain (Holland et al., 2007), whereas the Cers1 group and Cers from yeast and plants are devoid of a HOX domain. As part of transcription factors, HOX domains have been characterized as sequence-specific regulators of transcription (Svingen and Tonissen, 2006; Noyes et al., 2008). Although the functional role of the HOX domain in Cers proteins remains elusive, there is a remarkable conservation of each HOX motif between Cers2–6 homologs and there is a high sequence similarity between the Cers HOX motif to the canonical Antennapedia/Bicoid (Antp/Bcd)-derived consensus motif (Voelzmann and Bauer, 2009). The catalytic activity of all Cers resides in the Lag1 motif, a stretch of 52 amino acids (Spassieva et al., 2006), which localizes within the TLC (Tram, Lag and CLN8) domain, a region of approximately 200 C-terminal residues (Winter and Ponting, 2002). Catalytical activity of the Lag1 motif is also influenced by a stretch of amino acids flanking the HOX domain and preceding the TLC domain (Mesika et al., 2007).
Ceramide synthases studied so far harbor functional specificities, which is reflected on two major levels. (i) Each mammalian Cers studied so far has a unique expression profile (reviewed in Mullen et al., 2012) in terms of tissue-specific and temporal control of expression. Moreover, activity of some Cers proteins, e.g., Cers1 is controlled posttranslationally in response to external stress stimuli like chemotherapy or UV irradiation (Sridevi et al., 2009). (ii) Each of the Cers displays a preference toward acyl CoAs of different chain length (Lahiri and Futerman, 2005), although some biochemical redundancies remain. For instance, Cers1 shows a preference for C18 ceramide, whereas Cers2 synthesizes ceramide with longer acyl chains and both Cers5 and Cers6 can generate C14 and C16 ceramide (reviewed in Mullen et al., 2012). Of note, ceramides with different chain lengths have different biophysical and molecular characteristics, e.g., very-long-chain ceramides CoAs reduce membrane fluidity (Grösch et al., 2012).
Apart from serving as structural components of biological membranes, sphingolipids have also been reported to dynamically cluster with sterols to form lipid microdomains or lipid rafts, which were proposed to function as platforms for cellular signal transduction and protein sorting. Cellular stimuli triggered by cytokines, growth factors or other secreted molecules, e.g., tumor necrosis factor-α, interleukin-1, Fas ligand, or by induction of heat or oxidative stress were shown to trigger the synthesis of ceramides regulating proliferation, differentiation, cell survival, and apoptosis. Understanding the complex regulatory mechanisms that control the formation of ceramides and their glucosylated derivatives requires a detailed knowledge of temporal and spatial synthesis of these sphingolipid entities. The six human and murine Cers genes have been extensively studied in vitro, in tissue culture, and recently also by using knockout models in mouse (for review see Mullen et al., 2012). The functional studies on mammalian cers genes have, however, mainly focused on their requirement in adult tissue physiology and homeostasis. A comprehensive analysis of the expression of cers family members during embryonic development has not been performed in any vertebrate, yet.
Because the zebrafish Danio rerio offers several advantages for expression analysis like the transparency of embryo and larval stages of the life cycle, we decided to identify and characterize the cers gene homologs in this genetic model system. Based on a bioinformatics approach, we identified nine cers gene homologs in the zebrafish genome, compared with six cers genes in mouse and human. Based on our bioinformatic analysis and the nomenclature given by the ZFIN database (www.zfin.org), we refer to the newly identified cers homolog genes in the following as cers genes. Phylogenetic analysis indicates that the zebrafish cers homologs are highly conserved and can be arranged into two phylogenetic groups. We have further analyzed the expression pattern of all the nine cers gene homologs throughout embryonic development by in situ hybridization. Our data show that the expression profiles of the zebrafish cers are highly dynamic and may indicate tissue-specific profiles of ceramides and their derivatives.
Identification and Molecular Cloning of the Zebrafish ceramide synthases
Human and murine Cers sequences were assembled using ENSEMBL and then used for identification of homologs in the Zebrafish Genome (www.ensembl.org/danio_rerio). Nine different cers homologs were identified in the Ensembl Zv9 genome assembly. Using a reverse transcriptase-based polymerase chain reaction (PCR) approach, the complete coding sequences of all zebrafish ceramide synthases (Supp. Table S1, which is available online) were cloned and confirmed by sequencing. We named the zebrafish homologs of the vertebrate cers genes cers1 (chromosome 22), cers2a (chromosome 19), cers2b (chromosome 16), cers3a (chromosome 7), cers3b (chromosome 18), cers4a (chromosome 22), cers4b (chromosome 2), cers5 (chromosome 22), and cers6 (chromosome 9).
Compared with Mus musculus and Homo sapiens Cers genes cers2, cers3, and cers4 are duplicated in the Danio rerio genome, whereas the cers1, cers5, and cers6 are present as single copy genes. The predicted zebrafish cers proteins contain the two major protein domains: the HOX domain and the TLC domain, including the 52 aa long Lag1 motif (Fig. 1A), except for cers1, which is devoid of a HOX domain like in mammals. Zebrafish cers proteins show a high conservation with respect to the mouse and human homologs in a multi alignment (Supp. Fig. S1), ranging from 65–91% amino acid similarity and 46–79% identity (Supp. Table S2). The high degree of sequence similarity may reflect a positive evolutionary selection pressure acting on ceramide synthases, which may also indicate a conservation of their functions across vertebrate evolution.
To further address the evolutionary conservation of zebrafish cers proteins, a phylogenetic analysis was performed (Fig. 1B). The full-length amino acid sequences of yeast, fruit fly, nematode, zebrafish, mouse, and human Cers proteins available in ENSEMBL and NCBI databases were used to generate a phylogenetic tree based on the maximum likelihood method using PHYLIP 3.69 and bootstrap values from 1,000 replicates. With the same method, also the single putative HOX domain and TLC domain of all the above mentioned proteins were analyzed for evolutionary conservation (Supp. Figs. 2, 3). As indicated by the topology of the tree, all vertebrate ceramide synthases show the division into six groups, namely Cers1 to Cers6, each of them corresponding to a different type of Cers.
Cers1, Cers5, and Cers6 are defined by clearly separated phylogenetic branches. Zebrafish cers1, which is structurally different from the other cers, diverged relatively early during evolution, as is reflected by its localization on a completely separate branch of the phylogenetic tree. Homologs of cers1 are already present in the genome of teleost ancestors, like Petromyzon marinus and Latimeria cholumnae, from which zebrafish diverged 110–160 million years ago (Mya). Zebrafish cers5 and cers6 homologs are also members of distinctive branches of the cers5 and cers6 homologs respectively and most probably emerged from a duplication of a common ancestor. On the other hand, cers2, cers3, and cers4 homologs diverged more recently in evolution and therefore closely related on the molecular level. The two copies of cers2, cers3, and cers4 in the zebrafish genome are coherent with a whole genome duplication event which occurred at the base of the teleost radiation approximately 300–450 Mya (Taylor et al., 2001). Obviously the paralogous gene pairs of zebrafish cers2, cers3, and cers4 survived during evolution, and they are probably subject to the three major fates of duplicated genes (Holland et al., 1994, Postlethwait, 2007). In principal, duplicated genes can undergo neo-functionalization or sub-functionalization. Neo-functionalization suggests acquisition of novel functional properties, whereas sub-functionalization leads to a physical subdivision of the original protein functions. Both fates suggest a positive selection pressure on both cers paralogues and both copies would be preserved. Due to functional redundancy, one or both of the novel cers gene copies can also accumulate deleterious mutations and degenerate into pseudogenes.
Interestingly, the genome of Petromyzon marinus, one of the very ancestral representatives of the fish lineage, only contains cers1, cers5, and cers6 homologs (Fig. 1C). The genome of Latimeria cholumnae, which diverged approximately 100 million years after Petromyzon marinus, already contains single gene homologs of cers2 and cers3. The genome tetraploidization event, which occurred 300–450 Mya, probably led to the two gene copies observed in zebrafish Danio rerio. The single copy of the cers3 paralogous gene in Tetraodon nigrovirids suggests that, after the whole genome duplication (Postlethwait, 2007), one copy quickly disappeared. Of note, medaka (Oryzias latipes) has lost both gene copies of cers3, which encodes the Cers associated with synthesis of C22–C26 ceramide species in mammals (Levy and Futerman, 2010). Loss of cers3 in medaka may indicate functional redundancy by another cers gene or a medaka-specific sphingolipid requirement rendering cers3 function dispensable. However, it is possible that some of the medaka genes are not annotated yet, because the genome sequencing project is still ongoing (see http://www.shigen.nig.ac.jp/medaka/).
Expression of Zebrafish cers Genes During Embryogenesis
Due to the different sphingolipid structures generated by ceramide synthases directly (ceramides with different fatty acid chain length) or indirectly (complex glycosphingolipids like gangliosides which act as major components of axonal membranes of neurons and serve stabilization of cell-to-cell contacts), temporal- and tissue-specific control of ceramide synthase expression is pivotal to meet the requirements on sphingolipid metabolism in a distinct cell type, at a given developmental stage or upon a physiological situation. For example, during apoptosis ceramide containing C16 fatty acid (palmitic acid) is specifically generated upon apoptotic stimuli (Kroesen et al., 2003) and de novo synthesis of this sphingolipid is only catalyzed by Cers5 or Cers6 in mammals.
To analyze the spatial and temporal expression pattern of zebrafish ceramide synthases homolog genes during embryogenesis, whole-mount in situ hybridization (WISH) was performed with embryos at the gastrula stage, the 6-somite stage, 18-somite stage, 24 hours postfertilization (hpf), 48 hpf and 5 days post fertilization (dpf) (Figs. 2–4, Supp. Fig. S4; Table 1; Supp. Table S5). Cers RNA in situ hybridizations were validated by using sense probes (Supp Fig. S5) as negative control. In general, zebrafish cers homolog genes are widely expressed with elevated transcript levels in the developing central nervous system in all stages analyzed, whereas there are differential expression patterns in other tissues.
Table 1. Overview of Prominent Tissue-specific cers Genes Expression During Early Embryogenesisa
Gastrula stage (8 hpf)
6-somite stage (12 hpf)
18-somite stage (18 hpf)
n.s.r., not spatially restricted; n.e., not expressed.
trigeminal placodes Rohon-Beard sensory neurons intermediate cell mass
trigeminal placodes Rohon-Beard sensory neurons intermediate cell mass
notochord rhombomere 5
rhombomere 5 paraxial mesoderm pronephric duct or gut cloaca
Like most of the cers in zebrafish the expression of cers1 shows a localized distribution from the 18-somite stage to 24 hpf. Cers1 RNA was detected in the trigeminal placodes and Rohon-Beard sensory neurons of the nervous system (Fig. 2A), which undergo differentiation and axon extension in this developmental stage (Kimmel et al., 1995). The process of axon extension involves complex membrane remodeling, guidance, and cell adhesion events that are dependent on ceramide synthesis (Schwarz and Futerman, 1997). Cers1-specific ceramides may be involved in this process in the zebrafish nervous system. In addition cers1 is present in the intermediate cell mass of mesoderm (ICM) that is part of the hematopoietic system. Before the onset of blood circulation in the embryo at 24 hpf, this tissue mainly differentiates into embryonic erythrocytes and epithelial cells (Chen and Zon, 2009). The zebrafish ICM has been shown to also contain multipotent progenitor cells that do not only generate primitive erythrocytes, but also early thrombocytes and neutrophils (Warga et al., 2009). In the blood of mature mice and humans, high ceramide levels have been reported especially for thrombocytes and leukocytes, where the majority of the ceramides contain C24 or C24:1 fatty acyl chains, but other ceramides were also abundant (Dahm et al., 2006). Our expression analysis points to a role of Cers1 in zebrafish hematopoiesis. In later stages, cers1 expression is still localized in the central nervous system and at 48 hpf is present also in the pectoral fin bud (Fig. 2A), whereas at 5 dpf, there is an upcoming expression in the intestinal bulb and in the mid intestine (Supp. Fig. S4A).
The expression of cers2a (Fig. 2B) and cers2b (Fig. 2C) has clearly distinct features from the 6-somite stage on with an overlap in the expression of both genes at the 18-somite stage in the paraxial mesoderm and in the lens of the eye at 24 hpf. This could point to a greater need for cers2 derived ceramides in these tissues during proliferation and differentiation. At 6-somites, cers2a is present in rhombomere 5 and the notochord, a pattern which has also been described by Thisse and Thisse (2004). The rhombomere expression persists to the 18-somite stage, whereas at 24 hpf cers2a is detected in the midbrain–hindbrain boundary. At 18-somites and 24 hpf, cers2a is particularly switched on in the eye, cloaca, and endoderm of the embryo, but also has a basal ubiquitous distribution which is prolonged also in later stages (Supp. Fig. S4B). Cers2b on the other hand has the strongest ubiquitous expression of all cers at 80% epiboly and is strongly expressed in the anterior polster and paraxial mesoderm at the 6-somite stage (Fig. 2C). At 18-somites and 24 hpf, the mesodermal expression of cers2b is reduced and subsequently restricted to the posterior somites, which shows similarity to the anterior to posterior pattern of somite specification. Within the somite in this developmental stage the fusion of myoblasts and formation of mature muscle fibers occurs (Moore et al., 2007). The expression can also be observed at later stages when it is strongly expressed in the myotomes (Supp. Fig. S4C). In the anterior part of the embryo, cers2b is specifically found in the lens which might point to a function of cers2 in the lens that reflects the high level of C24 ceramide found in mammalian lenses (Borchman and Yappert, 2010). Cers2b is moreover expressed in the branchial arches, which later give rise to the gills.
The two zebrafish cers3 copies share a weak basic expression in early developmental stages and an expression in the eye and brain at 18-somites to 24 hpf. Cers3a (Fig. 3A) is also present in the intermediate cell mass, whereas cers3b (Fig. 3B) is additionally switched on prominently in the ear placodes and the pronephric duct. The expression in the ear placode is maintained also in later stages. At 5 dpf, a strong expression in visceral organs (liver, intestine, pancreas) is detected (Supp. Fig. S4D,E).
In zebrafish, cers4a (Fig. 3C) seems to be more relevant for early embryonic development, because cers4b RNA was not detected in the analyzed stages (Fig. 4A). cers4a is present in a ubiquitous distribution from early stages on, with an emphasis on expression in the brain and tail bud. At later stages, the expression is still ubiquitous, with a localized pattern in the brain, more strongly for cers4a than for cers4b (Supp. Fig. S4F,G).
Cers5 is strongly expressed in a ubiquitous manner at gastrula and 6-somite stage in the enveloping layer (Fig. 4B). At 18-somite to 24 hpf, the level of ubiquitous expression is reduced and cers5 is prominently expressed in the eye, particularly near the posterior margin, as well as in the intermediate cell mass. At 48 hpf the expression of the gene is elevated in the brain, in the myotomes and in the pectoral fin bud and this pattern can be detected also at 5 dpf stage (Supp. Fig. S4H), during which cers5 is additionally expressed in gill arches and some of the gastrointestinal organs (liver, intestinal bulb, mid intestine).
Cers6 is detectable from 6-somite stage onward and is restricted to tissues of neural origin including the neural plate (Fig. 4C). At 18-somites and 24 hpf, the cers6 is expressed in neural tissues, where expression domains in the brain, trigeminal placodes, and spinal cord can be identified at these developmental stages. Cers6 keeps on high expression levels in the brain also in later stages (Supp. Fig. S4I).
Cers genes encode key regulators of sphingolipid metabolism found in all eukaryotes and these enzymes are required for diverse biological processes such as proliferation, differentiation, apoptosis, stress response, cancer, and neurodegeneration (for Review, see Mullen et al., 2012). This study provides the first comprehensive cloning and in situ expression analysis of the cers genes during embryogenesis in a vertebrate model system, the zebrafish Danio rerio. Our data show that the zebrafish contains nine cers genes which are evolutionarily conserved and that their expression profiles during embryonic development are highly dynamic. Because the Cers studied so far in mammals and in Drosophila were shown to display a preference toward acyl CoAs of different chain length (Lahiri and Futerman, 2005), the unique expression profiles of the zebrafish cers during embryonic development are likely to result in tissue-specific profiles of ceramides and their derivatives.
The expression patterns we have revealed by our analysis are of particular interest. This includes the expression patterns of cers2a and cers3b in the embryonic pronephros because high expression levels of Cers2 have been found in the adult mouse kidney (Laviad et al., 2008), especially in the collecting ducts and renal papilla (Imgrund et al., 2009). Despite this strong expression in the kidney, a Cers2 knockout did not reveal a corresponding renal phenotype in mouse (Imgrund et al., 2009; Pewzner-Jung et al., 2010a,b). However, sphingolipids have been shown to be mediators in several renal diseases (Shayman, 1996). The embryonic zebrafish pronephros develops early in the intermediate mesoderm from cells that undergo a mesenchyme-to-epithelium transition. To establish the pronephric tube, the epithelial cells undergo polarization and form tight cell–cell adhesions (Drummond, 2003). In this process, a high content of very long chain ceramides which are produced by Cers2 or Cers3 in mammals could ensure a physiologically required membrane rigidity. In invertebrate models, an epithelial expression of ceramide synthase has also been reported for Drosophila schlank (Voelzmann and Bauer, 2011). In C. elegans, RNAi knockdowns of several lipid-biosynthetic enzymes including SPTL-1 and ceramide glucosyltransferases and the inhibition of ceramide synthases by fumonisin B result in a loss of cell polarity in the developing gut epithelium (Zhang et al., 2011). The effect was tracked down to hydroxylated glucosylceramide with a saturated long chain fatty acid of C21–C26 chain length, which might affect the vesicular transport of apical membrane components (Zhang et al., 2011). As demonstrated for the invertebrate intestine, a similar mechanism involving cers2a and cers3b could be active in the establishment of epithelial polarity and formation of a lumen in the zebrafish pronephric ducts.
Our data support the hypothesis that there are tissue-specific requirements for specific cers genes during zebrafish embryogenesis, e.g., in the pronephric duct, whereas there is a need for the entire repertoire of cers activities in the developing brain, presumably reflecting the extensive need for sphingolipid species for neuronal membrane synthesis and physiology. During embryogenesis, the brain reveals massive plasticity and undergoes morphological changing, with the proliferation and differentiation of the neuronal cell pool. For this reason, this particular tissue requires synthesis of a large amount of membranes, of which ceramides and sphingolipids are one of the most important constitutive elements and ceramides also provide the signals which sustain these processes. Studies in mouse have shown a key role of ceramides and sphingolipids for regulation of the interaction between the two major cell types in the brain, neurons and glia, which finally ensures brain development and energy homeostasis (Hirabayashi and Furuya, 2008). Moreover, ceramide synthesis is necessary for cell cycle control (survival and death) in neurons as well as for dendrite morphogenesis (Hirabayashi and Furuya, 2008). Our observation that all zebrafish cers genes are expressed in the developing brain of the embryo may indicate that de novo synthesis of all cers genes is required to generate the spectrum of sphingolipid species involved in neuronal differentiation and organogenesis of the brain during embryogenesis. In mouse knockout models, a function of Cers-dependent sphingolipid synthesis has been shown for postembryonic stages only. Cers1, Cers2, and Cers4 show strong and Cers5 and Cers6 show moderate expression in the brain and other neuronal tissues of Mus musculus and Homo sapiens (Laviad et al., 2008; Levy and Futerman, 2010; Mullen et al., 2012). Moreover, Cers-dependent sphingolipid metabolism is functionally required in the brain, indicated by the neurodegeneration phenotype caused by loss of murine Cers1 and Cers2, respectively. Loss of Cers1 is associated with depletion of Cers1-dependent C18 sphingolipid species and finally leads to accelerated aging process in the brain and loss of Purkinje cells (Zhao et al., 2011, Jennemann et al., 2012). Neurodegeneration is also observed in brains of Cers2 knock-out mice (Imgrund et al., 2009). Of interest, in both Cers knockout models, secondary perturbations of sphingolipid metabolism are observed. Loss of Cers1 gene function is associated with an increase in the steady state levels of C14 and C16 sphingolipid species, whereas loss of Cers2 gene function is associated with an up-regulation of Cers5 activity, which is capable to generate C16 sphingolipid species (Imgrund et al., 2009). Because C16-ceramides have been shown to act as potent bioactive sphingolipids mediating antiproliferative responses such as cell cycle arrest, senescence, and apoptosis (Senkal et al., 2011), these modulations of sphingolipid species might significantly contribute to the neuronal phenotypes described, and it would be instructive to analyze multiple loss of Cers gene functions, e.g., Cers2 and Cers5 loss of function situations, in which a secondary up-regulation of Cers5 expression is prevented. Because it is technically challenging to generate multiple knockout mouse models for cers genes, the morpholino technology to target gene functions in Danio rerio offers a straight forward approach to uncover redundant, yet unknown cers functions.
Apart from neuronal expression, most of zebrafish cers genes are expressed in the liver (Supp. Fig. S5). In mammals, Cers2 is the predominant and essential hepatic ceramide synthase activity and loss of Cers2 is associated with hepatocarcinomas, apoptosis (Imgrund et al., 2009), and a perturbed lipid and glucose metabolism (Pewzner-Jung et al., 2010a, 2010b). Because C22 and C24 ceramide are the major ceramide species synthesized by Cers2, this observation may indicate a predominant function of these ceramides in mouse liver. In contrast, most of the nine zebrafish cers genes are expressed in the liver, which might reflect a requirement of a rather broad repertoire of sphingolipids in the zebrafish liver.
A similar observation was obtained for the myotome, the somitic compartment that gives rise to skeletal muscle. Here, all zebrafish cers with the exception of cers1 are expressed. In murine skeletal muscle, de novo synthesis of sphingolipids is predominated by cers1 expression. In the intermediate cell mass, giving rise to the various zebrafish blood cell lineages, cers1, cers3, and cers5 are expressed. Although a comparable expression analysis is not available for murine embryogenesis, it might be interesting that in 6- to 8-week-old mice, the predominating cers expressed in the analogous hematopoietic organ, the bone marrow, is cers2 (Laviad et al., 2008).
Strikingly, the localized and organ-specific expression pattern of the various zebrafish cers genes described above is paralleled by a highly similar expression pattern of the key enzyme of de novo ceramide synthesis, namely serine palmitoyl transferase (SPT, see Thisse and Thisse, 2004). In mammals, SPT is composed of a multiprotein complex composed of the subunits SPTLC1, SPTLC2, and SPTLC3 (Hornemann et al., 2007). Whereas expression profiles of SPTLC1 and SPTLC2 are currently unknown in zebrafish, SPTLC3 is expressed ubiquitously at early stages and beyond 18-somite stage, expression localizes in the pronephric duct, the gut, the pectoral fin bud and in the brain, highly reminiscent of the tissue-specific cers gene expression (Thisse and Thisse, 2004).
The temporal and spatial cers expression profiles identified here suggest a differential requirement of sphingolipids in different tissues and organs and at specific stages of development. To address functional analysis of the nine cers, single and multiple cers loss of function experiments have to be performed which will be subsequently used for ceramide-specific enzyme activity assays and for determination of the amount and the spectrum of sphingolipid species affected by mass spectrometry. The molecular and expression analysis of the cers genes provided by this study will be the basis of future functional cers analysis in zebrafish. Because cers are considered as prime candidate targets for anticancer and metabolic therapies, the zebrafish model system may contribute valuable information about the regulation and function of cers genes in vivo.
Zebrafish wild-type strain AB was used and raised under standard conditions. (Westerfield, 2000). Embryos were obtained by natural mating and raised at 28.5°C in a ZebTec standalone system (Tecniplast).
Cloning of Zebrafish cers Genes
Zebrafish cDNAs were synthesized from total RNA from zebrafish embryos (24 hpf) using oligo-dT primers with superscriptII reverse transcriptase (Invitrogen). The zebrafish cers cDNA fragments were amplified by PCR from the embryo cDNA using primer pairs listed in Supp. Table S2. PCR was performed with phusion hot start high fidelity DNA polymerase (New England Biolabs). The conditions were as follows: 30 sec at 98°C and then 26 to 32 cycles of 7 sec at 98°C, 20 sec at 60°C, and 40 sec at 72°C, with final elongation of 10 min at 72°C. The PCR products were cloned into the pCRII vector (Invitrogen) and then subjected to sequence analysis.
The accession numbers of zebrafish cers homologs members and all the other sequences used in phylogenetic analysis are listed in Supp. Table S1. The sequences of primers used for cloning the zebrafish cers genes are listed in Supp. Table S2.
Ceramide synthases sequences from mouse (M. musculus), human (H. sapiens), fruit fly (D. melanogaster), nematode (C. elegans), and yeast (S. cerevisiae) were used to identify homologs in zebrafish by mean of ENSEMBL Genome Browser (http://useast.ensembl.org/Danio_rerio/Info/Index - 67th release, May 2012) and later compared with the zebrafish genome using the tBLASTn algorithm on the NCBI BlastServer (http://www.ncbi.nlm.nih.gov/genome/seq/BlastGen/BlastGen.cgi?taxid=7955). The protein sequences were aligned using ClustalW (Thompson et al., 1994) with default parameters. For the analysis of the alignment of zebrafish and mouse proteins, Jalview alignment editor (Waterhouse et al., 2009) was used, highlighting conserved residues according to BLOSUM62 matrix. A phylogenetic tree was constructed by the maximum-likelihood method using PHYLIP (phylogeny inference package), ver. 3.69 (Felsenstein, 2005). Bootstrap resampling analysis from 1,000 replicates was conducted to estimate the reliability of the tree. Putative functional domains were predicted using prosite database (http://prosite.expasy.org/ - Hulo et al., 2006). Notation of HOX domains was then refined by using known HOX domain sequences as template (Holland, 2007).
Whole-Mount In Situ Hybridization
Whole-mount in situ hybridization (WISH) was performed according to the protocol from Thisse and Thisse (2008) with the following modifications. RNA sense and antisense probes were produced from the full length zebrafish cers cDNA fragments in pCRII vector using the DIG RNA labeling kit from Roche. The probe in hybridization buffer was heated to 94°C before use. The duration of proteinase K digestion of embryos was 30 sec (gastrula and 12-somite stage), 1 min (18-somite stage and 24 hpf embryos), 5 min (48 hpf embryos), and 20 min (5 days postfertilization embryos), respectively. The probe hybridization and following washing steps were performed in reaction tubes on a heating block. Staining solution contained 2% polyvinyl alcohol (Sigma Aldrich, cat. no. 363138) that was dissolved at 94°C. The amount of probe used for each staining ranged between 30 ng and 140 ng. Embryos were imaged in 80–100% glycerol using an SZX16 stereobinocular microscope (Olympus) and a DP21-SAL CCD camera (Olympus).
Thanks to H.M. Pogodah and M. Hammerschmidt for providing zebrafish lines. This work was supported by a fellowship of the International Graduate School LIMES Chemical Biology (M.B.) and grants of the SFBs 645 and TR83 (M.H.).