Ascl1b and Neurod1, instead of Neurog3, control pancreatic endocrine cell fate in zebrafish
© Flasse et al.; licensee BioMed Central Ltd. 2013
Received: 25 April 2013
Accepted: 28 June 2013
Published: 8 July 2013
Skip to main content
© Flasse et al.; licensee BioMed Central Ltd. 2013
Received: 25 April 2013
Accepted: 28 June 2013
Published: 8 July 2013
NEUROG3 is a key regulator of pancreatic endocrine cell differentiation in mouse, essential for the generation of all mature hormone producing cells. It is repressed by Notch signaling that prevents pancreatic cell differentiation by maintaining precursors in an undifferentiated state.
We show that, in zebrafish, neurog3 is not expressed in the pancreas and null neurog3 mutant embryos do not display any apparent endocrine defects. The control of endocrine cell fate is instead fulfilled by two basic helix-loop-helix factors, Ascl1b and Neurod1, that are both repressed by Notch signaling. ascl1b is transiently expressed in the mid-trunk endoderm just after gastrulation and is required for the generation of the first pancreatic endocrine precursor cells. Neurod1 is expressed afterwards in the pancreatic anlagen and pursues the endocrine cell differentiation program initiated by Ascl1b. Their complementary role in endocrine differentiation of the dorsal bud is demonstrated by the loss of all hormone-secreting cells following their simultaneous inactivation. This defect is due to a blockage of the initiation of endocrine cell differentiation.
This study demonstrates that NEUROG3 is not the unique pancreatic endocrine cell fate determinant in vertebrates. A general survey of endocrine cell fate determinants in the whole digestive system among vertebrates indicates that they all belong to the ARP/ASCL family but not necessarily to the Neurog3 subfamily. The identity of the ARP/ASCL factor involved depends not only on the organ but also on the species. One could, therefore, consider differentiating stem cells into insulin-producing cells without the involvement of NEUROG3 but via another ARP/ASCL factor.
The pancreas is a mixed gland of the digestive tract composed of an exocrine compartment (acini and ducts), releasing digestive enzymes into the duodenum, and an endocrine compartment, secreting hormones into the bloodstream in order to control glucose homeostasis. Loss or dysfunction of endocrine insulin-secreting β-cells leads to diabetes, a widespread disease affecting more than 370 million people worldwide.
Outstanding progress has been made to set up new therapies for diabetes through cell therapy (reviewed by [1–5]). Recent efforts have been focused on directing stem cells to differentiate in vitro into pancreatic β cells that could be transplanted to diabetic patients . To achieve that goal, it is essential to understand in detail the molecular mechanisms controlling pancreatic endocrine cell differentiation.
Although much of our knowledge on pancreas organogenesis relies on mouse genetic studies, the use of zebrafish has also significantly contributed to the deciphering of mechanisms involved in the earliest phases of pancreas development [7–12]. In this fish, the endoderm forms two converging sheets of cells by the end of gastrulation (10 hours post fertilization, hpf). Subsequently, these cells condense at the midline to form the endodermal rod which will give rise to the digestive tract and the associated organs [9, 13]. Early in development, at the 10 somite stage (10s, 14 hpf), the homeobox Pdx1 factor starts to be expressed in the endodermal region located between the first and the fourth somite . As in mammals, a dorsal and a ventral pancreatic bud will emerge from this pdx1+ region and will later coalesce to form the pancreas . In zebrafish, the first hormone-expressing cells that appear from the dorsal bud are the insulin-producing β-cells, detected from 15 hpf onward. Next appear the somatostatin-secreting δ-cells (17 hpf), the ghrelin ϵ-cells (18 hpf) and finally the glucagon-producing α-cells (21 hpf). This first wave of endocrine cells is followed by a second wave coming from the ventral bud that forms from 32 hpf onwards [15–17]. After that stage, the increase of the endocrine cell mass is believed to result from the differentiation and proliferation of late forming ventral bud-derived endocrine cells . Therefore, while at 2 days post fertilization (dpf), the vast majority of the endocrine cells is generated from the dorsal bud, at 12 days, a majority seems to derive from the ventral bud .
Both in zebrafish and mice, the Notch signaling pathway tightly controls pancreatic cell differentiation. Notch prevents commitment to the endocrine cell fate, thereby reserving a population of undifferentiated precursor cells for ongoing proliferation and generation of later-appearing cell lineages [18–21]. Notch pathway is a fundamental and evolutionarily conserved process in metazoan development. There are numerous core players that participate in this process. Briefly, via the Hairy Enhancers-of-split proteins, Notch signaling represses the expression of genes of the Achaete-Scute like (ASCL) family or of the Atonal related protein (ARP) family, this latter being subdivided into Atonal, Neurogenin and Neurod subfamilies. These genes encode basic helix-loop-helix (bHLH) transcription factors and can be classified in two categories, ‘cell fate determinant’ factors and ‘cell differentiation’ factors . The cell fate determinants are transiently expressed at early stages and are both necessary and sufficient to initiate the development of a specific cell lineage. The ‘cell differentiation’ factors are expressed at later stages and implement the differentiation program initiated by the cell fate determinants. For example, in the murine pancreas, NEUROG3 is the cell fate determinant of the pancreatic endocrine lineage  as its transient expression initiates the endocrine differentiation program of all endocrine cells [24–26]. Indeed, almost no endocrine pancreatic cells were detected in the Neurog3 knock-out mice . NEUROG3 triggers the sustained expression of the ‘cell differentiation gene’ Neurod1 that maintains the endocrine cell differentiation program [27, 28]. Homozygous Neurod1 null mice notably have a striking reduction in the number of insulin-producing β cells and fail to develop mature islets .
The neurog3 gene is found in the zebrafish genome but, surprisingly zebrafish neurog3 mRNAs were not detected in the developing pancreas while they were detected in the hypothalamus and intestine [20, 30]. In this study, we extensively analyzed neurog3 expression during pancreas development and could not detect any expression at any stages in this tissue. The lack of Neurog3 function in the zebrafish pancreas was further confirmed by analyzing the phenotype of the recently identified sa211 neurog3 null mutant. As neuronal or endocrine cell-fate commitment controlled by Notch is classically carried out via ARP/ASCL factors [31–33], we next searched for other ARP/ASCL factors acting downstream of Notch signaling that would promote the formation of pancreatic endocrine cells. Among the 14 ARP/ASCL factors identified in the zebrafish genome, only ascl1b and neurod1 were found to be strongly expressed at early stages of endocrine cell differentiation. Knock-down analysis reveals that these factors have complementary roles in endocrine cell differentiation and that their simultaneous inactivation leads to a loss of all hormone-secreting cells. These two bHLH factors are, therefore, playing together a role analogous to that described for murine NEUROG3.
The expression of the 14 ARP/ASCL genes was analyzed by WISH at different time points during pancreas development (6 to 8s, 12 to 14s, 18 to 20s, 24 hpf, 30 hpf, 48 hpf, 72 hpf). We observed high expression in the pancreatic region at early stages for only two factors, namely ascl1b and neurod1 (Figure 2B-C). In contrast, only weak expression was detected at early stages for ascl1a, neurog1 and neurod6b within the pancreatic area [see Additional file 2: Figure S2, A-C], while atoh8 and neurod6a genes were detected only at late stages, that is, around two dpf [see Additional file 2: Figure S2, D-E].
neurod1 expression has already been reported in the dorsal pancreatic bud in zebrafish [20, 36–38] but its precise expression profile was not determined. In this study, we show that pancreatic neurod1 expression starts around 6s (12 hpf) in a few cells. At 10s (14 hpf), neurod1 is expressed in two rows of cells located on both sites of the midline (Figure 2B). Over the next four hours, the number of neurod1+ cells progressively increases and they start to cluster to form the islet which is completely formed at 30 hpf. neurod1 remains expressed in the pancreas of four dpf larvae as well as in adults (data not shown).
ascl1b is detected before neurod1 as its expression starts as early as 10 hpf (bud stage) in two rows of cells in the prospective pancreatic region (Figure 2C). Over the following two hours, the number of ascl1b-expressing cells increases to reach its maximal level around 12 hpf. Then, ascl1b expression progressively decreases and is turned off at 17 hpf.
To further define neurod1 function in early steps of endocrine cell differentiation, we analyzed the expression of sox4b at early stages of development. While the initiation of sox4b expression is not perturbed in neurod1 morphants (Figure 6M and O) its expression is strongly reduced at 18 hpf (Figure 6Q and S) and is not detectable anymore at 24 hpf (Figure 6U and W). All together, these data show that, in the absence of Neurod1, sox4b expression is correctly initiated. Then, around 17 to 18 hpf, the stage when most β-cells and a minority of δ-cells are already differentiated (see diagram in Figure 3K), sox4b expression is no longer maintained and the differentiation of late-appearing endocrine cells is blocked, leading to a loss of ϵ- and α-cells as well as to a reduction of δ-cells.
Intriguingly, although neurod1 is expressed from 12 hpf (6s) in the pancreatic endocrine precursors, neurod1 morphants do not display any apparent pancreatic defects before 17 hpf. We, therefore, hypothesized that, at early stages, Ascl1b could complement the loss of Neurod1 function. To test this hypothesis, we first defined Ascl1b function in endocrine differentiation by injecting an antisense morpholino targeting the translational start site of ascl1b mRNA as described in [44, 45].
Altogether, our data indicate that the simultaneous loss of ascl1b and neurod1 expression abolishes the formation of all pancreatic hormone-expressing cells by interfering with the initiation of their differentiation process as shown by the loss of the pancreatic expression of sox4b, isl1, pax6b, mnx1 and arx genes.
In this study, we show that, in zebrafish, Neurog3 does not control pancreatic endocrine cell fate as it does in the mouse. This function is fulfilled by two ASCL/ARP factors, Ascl1b and Neurod1.
The role of Neurod1 as a ‘cell differentiation factor’ required for the maintenance of the endocrine differentiation program in zebrafish is similar to the murine situation where NEUROD1 pursues the endocrine differentiation program initiated by NEUROG3 and participates in the maintenance of the mature islet cells (reviewed by [51, 52]). However in zebrafish, if Ascl1b is absent, Neurod1 remains expressed and acts as a cell fate determinant inducing the formation of the sox4b+ precursors. Indeed, in ascl1b morphants, sox4b+ precursor cells are detected only after neurod1 onset and sox4b expression is exclusively induced in neurod1 expressing cells while, in the control morphants, the majority of sox4b+ cells do not express neurod1. Consequently, the pool of endocrine precursors is reduced in Ascl1b morphants, leading to a reduced number of all mature endocrine cells at 30 hpf (Figure 12C). Thus, in zebrafish, neurod1 has the capacity to act as a ‘cell fate determinant’ or a ‘cell differentiation’ factor. In mice, NEUROD1 has been reported to act only as a cell differentiation factor. However, when NEUROD1 is ectopically expressed in the murine pancreatic anlagen under the control of the Pdx1 promoter, it has the same intrinsic capacity as Neurog3 to induce endocrine differentiation . Such functional equivalence highlighted by a gain-of-function approach can probably be explained by the fact that these two factors regulate largely overlapping sets of genes .
This dual capacity to promote the selection of precursors and to regulate some differentiation steps is not restricted to Neurod1 but is shared by many ASCL/ARP factors. For example, in the olfactory epithelium, Mash1 and Neurog1 are expressed sequentially in the sensory neuron precursors. MASH1 acts as a cell fate determinant by inducing the olfactory precursors, while Neurog1 acts as a cell differentiation gene by allowing the differentiation of these precursors into sensory neurons. However, NEUROG1 can partly compensate for the loss of the determination function of MASH1 in the olfactory placode, suggesting that NEUROG1 also has the intrinsic capacities to act as a cell determinant or differentiation factor in the same tissue . This suggests that the function of some ARP/ASCL factors is not intrinsically determined by their protein sequence and their structure but could be simply dictated by the context (cell type and timing of expression). This flexibility could be explained by the strong conservation within the bHLH domain. Nine of the ten residues predicted to contact DNA are identical among ARP and ASCL proteins  and, based on in vitro studies, there is no indication of DNA binding differences between ARP and ASCl proteins [55, 56]. Therefore, these factors can regulate overlapping sets of genes as described for NEUROG3 and NEUROD1  or for the proneural bHLH factors Xath5 and XNeuroD .
However, in other cases, the protein sequence can be crucial for the intrinsic properties of ARP/ASCL proteins acting as ‘cell fate determinants’ versus ‘cell differentiation factors.’ For example, in the context of muscle differentiation, the ‘cell differentiation gene’ Myogenin (Myog) is less efficient than the ‘cell fate determinant gene’ Myogenic factor 5 (Myf5) at remodeling chromatin and activating transcription at previously silent loci . Such specificities can result from their interactions with specific cofactors. For example, functional divergence between Ato and Neurog proteins in Drosophila is encoded by three non-conserved residues in the basic domain of these bHLH which are responsible for the differential interaction with Senseless or MyT1 . Such cofactors can affect the interaction of the bHLH with their DNA binding sites and/or modulate their transcriptional activity. Identifying such cofactors will be an important issue to understand better the transcriptional properties of ARP/ASCL factors.
The lack of pancreatic defects in the zebrafish neurog3 mutant contrasts with the crucial function of NEUROG3 in the mouse. This raised the question whether the neurog3 gene studied here is the actual ortholog of the murine Neurog3 gene. This seems to be the case as extensive searches in the zebrafish genome revealed the presence of only two neurogenin genes in zebrafish, neurog3 and neurog1, neurog3 being the most closely related to murine and human NEUROG3 and the locus displaying a conserved synteny with the region of human chromosome 10 containing NEUROG3. As the zebrafish genome is almost completely sequenced , it is highly unlikely that another neurogenin gene exists in zebrafish but this possibility cannot be totally excluded. Rather, we show here that the crucial role of murine NEUROG3 in pancreatic endocrine cell differentiation is fulfilled in zebrafish by two bHLH transcription factors, Ascl1b and Neurod1. Indeed, the simultaneous inactivation of these two bHLH factors leads to a complete loss of all hormone-secreting cells, as in Neurog3 null mice. The endocrine cells remain at the precursor stage as they still express the ascl1b transcripts but none of the later pancreatic transcription factors, such as sox4b, pax6b, mnx1, isl1 and arx. During the second endocrine wave coming from the ventral bud, Ascl1b and Neurod1 probably also play an important role as we have observed a strong expression of these two factors in the intrapancreatic ducts after Notch inhibition.
This indicates that NEUROG3 is not the only cell fate determinant that can promote pancreatic endocrine cell differentiation. Even in the mouse, some rare endocrine cells can be produced in a NEUROG3-independent manner as Wang et al. reported a small, yet significant, number of glucagon expressing cells in the Neurog3-/- pancreas before e15.5 . One could, therefore, consider differentiating stem cells into insulin-producing cells without the involvement of NEUROG3 but via another ARP/ASCL factor. Such an assumption is strengthened by the fact that in the murine stomach, it is ASCL1 which is required for the differentiation of all endocrine cell types  whereas NEUROG3 is only involved in the differentiation of a subset of these cells [63, 64]. In contrast, ASCL1 is only expressed in a few cells of the murine pancreas and its knock-out does not disrupt the pancreatic endocrine differentiation . These results highlight the diversity found in the selection of the ARP/ASCL factors involved in the determination and differentiation of the endocrine cells and show that the choice of these factors depends not only on the organ considered but also on the species.
Zebrafish (Danio rerio) were raised and cared for according to standard protocols . Wild-type embryos from the AB strain were used and staged according to Kimmel (Kimmel et al., 1995). Homozygous mutants were obtained by mating heterozygous fish for the mind bomb (mibta52b) allele  and for the neurog3sa211 allele (ZMP, Zebrafish Mutation Project ). The genotyping of neurog3sa211 embryos were done on DNA extracted from tails of WISH stained embryos by performing a PCR using primers 0230 (CCAACACATACCCAGTACCTC) and O233 (TGATTTGACCTCTGTCGAAC) followed by a nested PCR, 0231 (GCTTGCAAGAGGTAAGCATC) and O232 (TGTAATTATGCGCGAATCTC) and by subsequent sequencing of the PCR products. The LY411575 treatment was performed by incubating the embryos during the indicated period with a 10 μm LY411575 solution (Medchemexpress) and replacing the media every day.
In order to discover all neurog genes present in the zebrafish genome, we screened the Zv9 Genome assembly using the ensembl genome browser  that covers 1.357 Gb in scaffolds placed on chromosomes 1 to 25 and 55 Mb in unassigned scaffolds. This zebrafish sequence is considered to be nearly complete. Indeed, out of a non-redundant set of 21,471 zebrafish cDNAs from ENA/Genbank, only 120 (0.6%) are not currently found in Zv9 (that is, these 120 do not have a match at 90% identity covering at least 10% of their length) ). Furthermore, we also screened the nucleotide collection (nt/nr) from the National Center for Biotechnology Information (NCBI)  that consists of GenBank + EMBL + DDBJ + PDB + RefSeq sequences. These databanks have been screened by the program ‘tblastn’ with the consensus amino acid sequence of the basic domain of vertebrate Neurog3 (RRXKANDRERNR) or of vertebrate Neurog (RRXKA(D/N)DRERNR), the amino acid sequence of the bHLH domain of murine NEUROG3 and of zebrafish Neurog3 and with the consensus amino acid sequence of the bHLH of all Neurog proteins
(RRXKANXRERXRMHXLNXALDXLRXXLPXFPXDXKLTKIETLRFAXNYIWALXXTXR). All these searches identified only 2 neurog genes in the zebrafish genome, neurog3 (AF181996) and neurog1 (AF017301).
Antisense riboprobes were made by transcribing linearized cDNA clones with SP6, T7 or T3 polymerase using digoxigenin or DNP labeling mix (Roche) according to the manufacturer’s instructions. They were subsequently purified on NucAway spin columns (Ambion) and ethanol-precipitated. The zebrafish ascl1a, ascl1b, sox4b, isl1, neurod1, pax6b, neurog3, atoh1a, atoh1b, neurod6a, neurod6b, atoh1c, atoh7, atoh8, mnx1/hb9, arx, pdx1, insulin, somatostatin 2 (PPS2), ghrelin and glucagon probes have been described elsewhere. The neurod2, neurod4 and neurog1 probes were obtained from Imagene clones MAGp998E0411982Q (Pst1, SP6), IRBOp991C31D (Pst1, T7) and IRBOp991BO232D (EcoRI, T7), respectively. Single whole-mount and double fluorescent in situ hybridizations were carried out as described .
For visible WISH, quantifications of the number of cells were performed by counting the cells under the microscope by focusing successively on each layer of stained cells. For that purpose, the NBT/BCIP staining was carefully monitored in order to avoid an overstaining which would have prevented us from visualizing the individual cell boundaries. This method has been validated by counting the number of isl1-expressing cells at 24 hpf after a visible WISH or after a fluorescent WISH analyzed by confocal microscopy; very similar results were found (a mean of 49 cells counted by the first method and of 47 by the second) (see Figure 7 in ). For fluorescent WISH, quantifications were performed by determining the volume occupied by specific cells in control and morphants, 100% being arbitrarily fixed as the mean of the volume occupied by these cells in control morphants. Cell volume has been calculated by the program Imaris (Bitplane) on the confocal images of each embryo.
WISH was performed using the following antibodies: rabbit polyclonal antibody against GFP (1:500; Chemicon), guinea pig polyclonal antibody against Insulin (1:500; Biomeda), rabbit polyclonal antibody against Somatostatin (1:1000; MP Biomedicals). The embryos were fixed with 2% formaldehyde in 0.1 M PIPES, 1.0 mM MgSO4 and 2 mM EGTA overnight at 4°C and washed three times with PBS/0.3% Triton X-100. The yolk was manually removed, the embryos permeabilized for one hour with PBS/1% Triton X-100 and then blocked for two hours in PBS/4% BSA/0.3% Triton X-100 at room temperature. Both primary and secondary antibodies were incubated overnight in the same blocking buffer. Washes were done with PBS/0.3% Triton X-100 for three hours to remove excess antibodies.
The fragmented DNA of apoptotic cells was identified by the TUNEL method, using the Apoptag Apoptosis in situ detection kit (Chemicon) in which the fragmented DNA fragments are labeled with digoxigenin-nucleotide, subsequently recognized by peroxidase coupled anti-digoxigenin antibody and revealed by incubating with tyramide-fluorescein isothiocyanate (FITC) substrate. Briefly, after the WISH, the endogenous peroxidases were inactivated by incubation for one hour in 3% H2O2 at room temperature, washed three times in PBS/1% Triton X-100 and incubated in Equilibration Buffer for one hour. The terminal deoxynucleotidyl transferase (TdT) reaction was performed by incubating the embryos in a mixture of 18 μl reaction buffer and 6 μl TdT enzyme overnight at 37°C. The reaction was stopped with 500 μl of stop/wash buffer for 10 minutes. Then, the embryos were rinsed 3 times for 10 minutes with PBS/0.3% Triton X-100. The detection was performed by incubating the embryos with 50 μl anti-digoxigenin conjugate (peroxidase) overnight at 4°C, rinsing three times for 10 minutes with PBS/0.3% Triton X-100 and incubating in 50 μl amplification reagent (Perkin Elmer). + tyramide-FITC (1/5000) for one hour at room temperature. After three 15-minute washes in PBS/0.3% Triton X-100, samples were mounted in Prolong (Invitrogen) and imaged.
Microscope pictures were obtained with an Olympus DP70 photocamera fixed on a BX60 Olympus microscope. Confocal imaging was performed using a Leica TCS SP2 inverted confocal laser microscope (Leica Microsystems, Germany) Digitized images were acquired using a 63X (NA 1.2) Plan-Apo water-immersion objective at 1024 X 1024 pixel resolution. For multicolor imaging, FITC was visualized by using an excitation wavelength of 488 nm and the emission light was dispersed and recorded at 500 to 535 nm. Cy3 was detected by using an excitation wavelength of 543 nm and the fluorescence emission was dispersed and recorded at 555 to 620 nm. The acquisition was set up to avoid any cross-talk of the two fluorescence emissions. Series of optical sections were carried out to analyze the spatial distribution of fluorescence, and for each embryo, they were recorded with a Z-step ranging between 1 and 2 μm. Image processing, including background subtraction, was performed with Leica software (version 2.5). Captured images were exported as TIFF and further processed using Adobe Photoshop and Illustrator CS2 for figure mounting.
All the morpholinos (Mo) were designed by Gene Tools. The neurog3 morpholinos are complementary to either the ATG (Mo1: 5′-GGATCTTGGAGTCATTCTCTTGCAA-3′) or to the 5′UTR (Mo2: 5′-GCTCGCTCAGTAAAACCGAGGTACT-3′). The neurod1 morpholinos are complementary to either the ATG (Mo1: 5′-TTTCCTCGCTGTATGACTTCGTCAT-3′) or the 5′UTR (Mo2: 5′-CCTCTTACCTCAGTTACAATTTATA-3′). The ascl1b morpholino, as described by [44, 45], is complementary to the ATG (5′-TCGTAGCGACGACAGTTGCCTCCAT-3′). A standard control Mo, having the sequence 5′-CCTCTTACCTCAGTTACAATTTATA 3′ has also been designed by Gene Tools in a way that it should have no target and no significant biological activity. A morpholino directed against p53 mRNA was used to prevent nonspecific apoptosis , observed with the ascl1b morpholino. They were dissolved at a concentration of 3 μg/μl in 1× Danieau buffer containing 0.5% of rhodamine dextran (to follow the microinjection process) and microinjected at the 1 to 2 cells stage at a dose of 2.5 ng for Mo1-neurog3, 3 ng for Mo2-neurog3, 3 ng for Mo1 and Mo2 neurod1 and 6 ng for Mo-ascl1b together with 3 ng Mo-p53. For the double knock-down of Ascl1b and Neurod1, a mix with 6 ng of Mo-ascl1b, 3 ng of Mo2-neurod1 and 3 ng of Mo-p53 was injected. Injected embryos were then grown in the presence of 0.003% 1-phenyl-2-thiourea until the desired stage, fixed overnight in 4% paraformaldehyde and stored in 100% methanol before analysis.
Atonal related protein
Bovine serum albumin
days post fertilization
Green fluorescent protein
hours post fertilization
Polymerase chain reaction
Terminal deoxynucleotidyl transferase dUTP nick end labeling
Whole-mount in situ hybridization
We thank M. Hammerschidt (ascl1b), G. Bellipanni (ascl1a), B.B. Riley (atoh1a and atoh1b), S. Wilson (atoh5), Gong Zhiyuang (neurod6a and neurod6b), M. Hibi (atoh1c) and W. Jia (atoh8) for sending probes. We thank the Sanger Institute Zebrafish Mutation Resource for providing the zebrafish neurog3 knock-out allele sa211. The zebrafish lines were raised with the help of M. Winandy (Zebrafish Facility and Transgenics Platform, GIGA, University of Liège). The results of confocal imaging were obtained thanks to S. Ormenese and G. Moraes (Imaging Platforms, GIGA, University of Liege). The results of sequencing were obtained thanks to Genomic-Sequencing Platform, GIGA, University of Liège. L.C.F was supported by the FNRS-FRS and the Leon Fredericq fund and J.L.P and D.G.S by the FRIA. I.M. was supported by the FNRS-FRS and by the Action de Recherches Concertées (University of Liège). B.P. and M.L.V. are Chercheur qualifié FNRS. This work was funded by the Belgian State’s ‘Interuniversity Attraction Poles’ Program (SSTC, PAI).
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.