The putative Notch ligand HyJagged is a transmembrane protein present in all cell types of adult Hydra and upregulated at the boundary between bud and parent

Background The Notch signalling pathway is conserved in pre-bilaterian animals. In the Cnidarian Hydra it is involved in interstitial stem cell differentiation and in boundary formation during budding. Experimental evidence suggests that in Hydra Notch is activated by presenilin through proteolytic cleavage at the S3 site as in all animals. However, the endogenous ligand for HvNotch has not been described yet. Results We have cloned a cDNA from Hydra, which encodes a bona-fide Notch ligand with a conserved domain structure similar to that of Jagged-like Notch ligands from other animals. Hyjagged mRNA is undetectable in adult Hydra by in situ hybridisation but is strongly upregulated and easily visible at the border between bud and parent shortly before bud detachment. In contrast, HyJagged protein is found in all cell types of an adult hydra, where it localises to membranes and endosomes. Co-localisation experiments showed that it is present in the same cells as HvNotch, however not always in the same membrane structures. Conclusions The putative Notch ligand HyJagged is conserved in Cnidarians. Together with HvNotch it may be involved in the formation of the parent-bud boundary in Hydra. Moreover, protein distribution of both, HvNotch receptor and HyJagged indicate a more widespread function for these two transmembrane proteins in the adult hydra, which may be regulated by additional factors, possibly involving endocytic pathways.


Background
Early embryonic development is regulated in all animals by common signalling pathways such as the Wnt-, FGF-, Hedgehog-, TGFβ/BMP-and Notch pathways. They have been shown to be present in basal metazoans such as Cnidarians [1][2][3][4][5]. The fresh water polyp Hydra consists of a simple tube-like body made of two epithelial layers, the inner endoderm and the outer ectoderm separated by an acellular mesoglea. It further possesses a hypostome and a ring of tentacles at the apical end and a foot at its basal end. Interstitial cells are located in the spaces between the epithelial cells of both cell layers. They constitute a pluripotent stem cell lineage that differentiates nerve cells, gland cells, germ cells and nematocytes. The latter carry nematocyte capsules used for catching prey and are unique to Cnidarians [6][7][8][9].
The Notch signalling pathway has recently been shown to be involved in nematocyte and female germ cell differentiation, as well as in budding [5,10]. Whilst an inductive function of Notch for terminal differentiation of nematoblasts and oocytes has been suggested the precise mechanisms are completely unclear. During budding Notch signalling was shown to be important for the formation of a sharp boundary between the bud and the parent animal. This seemed to be the preposition for the bud to make a constriction at its base and form a foot. As has been shown at many other developmental boundaries Notch signalling here appears to have the potential to create patterns where two adjacent cells adopt different fates. These are signal receiving cells, which have the Notch receptor, and signal sending cells carrying a Notch ligand. Notch ligands possess a characteristic domain, the DSL (Delta, Serrate, Lag) domain, followed by several EGF repeats and a transmembrane domain. Serrate/Jagged-like ligands have an extra domain, the cysteine rich von Willebrand type C (VWC) domain, between the EGF repeats and the transmembrane domain. Interaction with Notch receptors takes place via the DSL domains of the ligand and the EGF repeats of the receptor [11].
We have now identified a potential Notch ligand in Hydra. Here we present its sequence and expression pattern. Comparison of its amino acid sequence with that of other Notch ligands of the DSL family shows similarity to Serrate/Jagged-like ligands. We thus named it HyJagged. Its mRNA expression levels in adult animals were not detectable by in situ hybridisation with the exception of a remarkable upregulation at final stages of budding in cells of the parent animal at the site where the bud detaches. However, immunofluorescence studies with anti-HyJagged antibody revealed the presence of HyJagged in all cells of adult animals where it is localised at membranes and in endosomes.

Results
Isolation and characterization of the Hydra Jagged gene Database searches of the Hydra genome [12] led to a core sequence prediction for a DSL protein with similarity to Jagged/Serrate-like Notch ligands. After performing 5' and 3' RACE experiments the full-length Hyjagged sequence was amplified from Hydra cDNA [GenBank:JN036823]. The resulting HyJagged protein sequence is shown in Figure 1 in an alignment with mouse Jagged1. It contains a signal peptide (1-22aa), a DSL (Delta, Serrate, Lag) domain (181-252aa), which is characteristic for Notch ligands ( Figure 2) [13], and five EGF repeats (255-540aa). A cysteine rich domain, the von Willebrand type C (VWC) domain (549-616aa), is located N-terminally of the transmembrane domain (763-785aa), followed by a short non-conserved intracellular region. Comparison with Notch ligands from worms, insects and vertebrates further shows that the modular structure is conserved in HyJagged ( Figure 3). The presence of the von Willebrand type C domain, which is characteristic for Serrate/Jagged-like ligands, led us to name the Hydra protein HyJagged.
Interestingly, comparison of HyJagged with the predicted sequence of a Notch ligand from the sea anemone Nematostella vectensis, another Cnidarian, shows a generally similar domain structure between the two proteins but also an important difference ( Figure 3). The Nematostella gene encodes a protein with 17 EGF repeats, whereas HyJagged has only six, which in both cases nearly matches the number of EGF repeats in their respective Notch receptors. The DSL domains between the Notch ligands from Nematostella and Hydra are also highly conserved ( Figure 2).
HyJagged consists of 835 amino acids that are encoded by 16 exons on a stretch of 11.452 nucleotides. Comparison of the gene organisation of mouse jagged1 with Hyjagged is shown in Figure 4. All exon-intron boundaries are flanked by the conserved GT-AG sequences, 14 of these boundaries appear to be conserved between mouse and Hydra ( Figure 4, indicated by arrows). These reflect the modular organisation of Jagged proteins in all animals. Most EGF repeats and the transmembrane domain are encoded by single exons. However the Hydra DSL domain and the mouse VWC domain, as well as the signal peptides of both proteins are encoded within two exons, respectively.

mRNA expression of Hyjagged
To look for expression of the Hyjagged gene in adult hydra in situ hybridisation experiments were performed. These showed no signal, indicating that Hyjagged is expressed at very low levels in contrast to the expression of HvNotch, which shows very high levels of mRNA all over the animal [5]. This is in accordance with the fact that no ESTs encoding HyJagged could be identified in the libraries.
However, when we looked for Hyjagged mRNA during budding we found a very strong expression of the gene during budding stage 9. This is when the bud begins to form foot cells. The signal was present in cells of the ectoderm of the parent animal and coincided with the expression zone of the Hydra FGF-receptor homolog kringelchen ( Figure 5).

Protein localisation of HyJagged
In order to study the subcellular distribution of HyJagged in hydra cells the protein was overexpressed as a GFP fusion protein in single cells of intact animals and observed in vivo ( Figure 6). For detection of plasma membranes and endosomes the membrane-selective dye FM4-64 was injected into the gastric cavity and the animals were imaged at different time points afterwards. FM4-64 is usually detected at membranes first but is then increasingly enriched in endosomes, which become larger over time [14]. In Figure 6 we show the distribution of FM4-64 in cells of live imaged animals over a period of 20 hours. HyJagged-GFP hardly co-localised with FM4-64 at membranes at the time points indicated. However, it was seen in punctuate and ring-like structures in the cytoplasm. Some of these clearly corresponded to FM4-64 positive structures and thus most likely constituted early and late endosomes. We then compared this HyJagged-GFP distribution with the localisation of HvNotch-RFP in epithelial cells. Figure 7 shows that both proteins are localised in punctuate and ring-like structures, however they only partially colocalised.
In the absence of detectable mRNA for the putative Notch ligand Hyjagged we thought to look for the endogenous HyJagged protein. We therefore raised an antibody in rabbit against a synthetic peptide corresponding to 18 amino acids of the intracellular domain of HyJagged. The antibody recognised the intracellular domain of HyJagged when it was expressed in E.coli after SDS-PAGE and Western blotting (Additional file 1A). In addition, on a Western blot with a vesicle fraction from Hydra the antibody recognised a band at an apparent molecular weight of 130 kDa. This is larger than the HyJagged primary sequence suggests (96 kDa), most probably due to post-translational modification of mature HyJagged. Heavy glycosylation of EGF repeats is well documented for Notch ligands (recently reviewed in [15]).
The anti-JAG-IC antibody also detected HyJagged in immunofluorescence experiments, when we expressed it as GFP fusion protein in human HEK293T cells. This is shown in Additional file 1 D-G. In a transient transfection experiment in HEK293T cells HyJagged-GFP did not reach the membrane but was instead trapped in the ER as can be seen by strong staining of the nuclear envelope. This incorrect distribution of HyJagged is most probably due to incomplete post-translational processing of the hydra protein in human cells [16].
Nevertheless this experiment clearly shows that the antibody recognises HyJagged specifically in immunofluorescence. It did not react with any other human protein in HEK293T cells. We therefore used this antibody for immunofluorescence on Hydra whole mounts. Here it stained membranes and punctuate structures in the cytoplasm of epithelial cells all over the animal ( Figure  8). A weak increase of the staining in comparison with the surrounding tissue was detected in budding animals at stage 9 at the base of the bud in the ectoderm of the parental animal ( Figure 9). This is in accordance with the in situ hybridisation data, which indicated strong mRNA expression in this region.
In addition to epithelial cells, cells of the interstitial cell lineage were also stained with the antibody. This was seen especially in nerve cells of the basal disc (Figure 10). Co-immunofluorescence experiments with the anti-JAG-IC-and anti-Nv1-antibodies were performed in order to relate HyJagged-positive cells to a specific cell type. The anti-Nv1-antibody is specific for sensory neurons in the tentacles and ganglion cells in the peduncle [17]. Co-staining with anti-JAG-IC showed that these ganglion cells also contained HyJagged, but in addition, HyJagged was also present in Nv-1 negative neurons ( Figure 11).
In order to identify HyJagged proteins in cells of the nematocyte line we used a rat monoclonal antibody against HyZic protein since Hyzic mRNA is known to Figure 2 Alignment of DSL domains from predicted NvJagged (Nematostella), mJagged1 (mouse) and HyJagged (Hydra) in comparison with the DSL domain consensus sequence suggested by Tax et al. [13]. Consensus sequence of DSL domains according to [13]. Identical amino acids are marked in grey. be expressed in proliferating nematocyte precursors [18]. Rats were immunized with HyZic protein that we produced in E.coli. The resulting monoclonal antibody anti-ZIC7A12 detected the recombinant protein on Western blots. Moreoever, it stained a band of 46 kDa, which corresponds to the predicted size of HyZic in Western blots from hydra lysates (Additional file 2). In addition, in immunofluorescence experiments it recognised a HyZic-GFP fusion protein when expressed in hydra cells (see Additional file 2). On Hydra whole mounts the anti-ZIC7A12 antibody strongly stained single and double interstitial cells and nests of nematoblasts before their exit from mitosis. Weaker staining was observed in post-mitotic nematoblasts, which already had formed a small vacuole. In further developed nematocytes HyZic was not detected anymore. This distribution of HyZic protein in Hydra is in good agreement with published data about the localisation of its mRNA in proliferating nematoblasts [18]. Figure 12 shows that single and pairs of interstitial cells highlighted with the anti-ZIC7A12 antibody also contained HyJagged in a punctuate pattern. This was also the case in proliferating nematoblasts. However, in finally differentiated nematocytes negative for HyZic very strong HyJagged staining was observed ( Figure 12 Q-T, nc). In conclusion, while all cells of the nematocyte line seem to have HyJagged protein, its abundance increases in terminally differentiated cells. Finally, we wanted to compare the cellular localisation of HyJagged with that of HvNotch. To this end we raised an antibody in chicken against HvN ΔE (the Hydra Notch receptor HvNotch without its extracellular  domain) after its expression in E.coli. It detected HvN ΔE after expression in E.coli by SDS-PAGE and Western blotting down to a concentration of 8, 9 ng (see Additional file 3A). Furthermore it recognised overexpressed HvNotch-GFP fusion protein in human HEK293T cells in immunofluorescence experiments (see Additional file 3). We thus used it for co-immunofluorescence staining on Hydra whole mounts together with the anti-JAG-IC antibody. Both antibodies stained membranes of epithelial cells and cells in the interstitial spaces, however, the patterns did not completely overlap. Moreover, both antibodies stained punctuate structures in the cytoplasm. From co-localisation experiments of live images where HvNotch and HyJagged clearly co-localised with the membrane selective dye FM4-64 we concluded that these punctae most likely correspond to endosomes. However, HyJagged-and HvNotch antibody signals in endosomes do not generally co-localise. Moreover, HyJagged signals appear stronger localised on the membranes in comparison with Notch, this is especially obvious in interstitial cells ( Figure 13).

Discussion
In this work we have identified a potential ligand for HvNotch in Hydra, HyJagged. It has the same modular structure as Notch ligands from higher animals and the closely related Cnidarian Nematostella, which is also reflected in the conserved exon-intron structure of its gene between Hydra and mouse. The protein includes a DSL domain, which is responsible for the interaction with the extracellular EGF repeats of the Notch receptors in flies and mammals [11]. Moreover, HyJagged possesses a relatively small number of EGF repeats and a transmembrane domain. Comparison of the abundance of EGF repeats in Notch ligands and receptors shows some interesting similarities: Species with large extracellular domains in their Notch receptors (e.g. Nematostella, Drosophila and mouse) also have a high number of EGF repeats in their Serrate/Jagged-like proteins (however not in Delta) and vice versa (Figure 3 and [5]).
Thus, although we have not provided experimental evidence, we suggest that HyJagged constitutes the only canonical Notch ligand in Hydra. A second gene encoding a DSL domain is also present in Hydra. It has been identified previously and was shown to be expressed in gland cells [19]. The resulting protein is 332 amino acids long, has a signal peptide and a DSL domain and thus shows some structural homology with DSL2, a soluble Notch ligand from C.elegans. It would be interesting to see whether this protein is involved in Notch signalling in Hydra.
In situ hybridisation showed that Hyjagged was not visibly expressed on the mRNA level in intact Hydra with the exception of a very strong upregulation at budding stage 9. We have previously described that Notch signalling is important for late budding stages in Hydra. Whilst as early as at budding stage 5-6 HvWnt8 and the Hydra FGFR-homolog kringelchen demarcate the boundary between the parent animal and the bud [20,21] budding cannot be completed without Notch signalling. The gene expression zone of kringelchen remains diffuse and a crucial metalloprotease, MMP-A3 is not induced. Thus the bud cannot constrict and does not form a foot. Two stripes of gene expression are required for correct patterning of the bud, one leading to constriction involving MMP-A3 and the other leading to differentiation of the buds foot cells involving expression of foot specific genes such as ppod-1. For these to form we proposed a model including mutual inhibition and lateral activation on the basis of Notch signalling at this boundary [10].
Our model required that cells which express kringelchen at the boundary express a Notch ligand. The data  shown here are partially in accordance with this assumption because Hyjagged is expressed in cells that also express kringelchen at the boundary between parent and bud. However, the timing of this upregulation is unexpected. While constriction begins in budding stage 8, Hyjagged mRNA could only be detected at the beginning of budding stage 9. Hyjagged expression during earlier budding stages probably occurs in too low amounts to be detectable by in situ hybridisation experiments. Antibody staining clearly shows the presence of  HyJagged at the basis of buds, however, only a weak increase in its level at this position in comparison with the remaining tissue is detectable (Figure 9). At this point we clearly have not used techniques suited to uncover precise quantitative differences in the amounts of HyJagged mRNA and protein at the cellular level during the final stages of budding. Therefore we cannot exclude the possibility that the observed requirement for Notch signalling at the boundary between parent and bud involves a mechanism that does not include a direct HyJagged-HvNotch interaction.
In contrast to the generally undetectable levels of mRNA expression outside the boundary between parent and bud, HyJagged protein is clearly present in all cells of the animals. The same is true for HvNotch. Although this makes it difficult to conclude whether there is active Notch signalling going on, the presence of the proteins is not unexpected. Notch signalling is highly regulated by factors involved in post-translational modification of the receptors (e.g. Fringe) [22], by the activity of metalloproteases like ADAM [23] and of presenilin as catalytic subunit of the γ-secretase complex and also by endocytosis of both, the Notch receptors and the ligands [24,25]. Endocytosis is part of a mechanism that leads to mutual inhibition of the signal sending and signal receiving cells. This means that signalling cells have higher levels of ligand but inhibit the propagation of Notch signals from adjacent cells and vice versa [26]. That way signalling can be induced by small differences in the concentration of receptors and ligands in this  system, which might be hard to detect in immunofluorescence experiments [27]. We have shown that GFP/ RFP-tagged HyJagged and HvNotch are indeed localised in endosomes (this work and [5]). However, the fluorescently tagged protein did not completely co-localise with each other within these structures. Differences in the localisation of the endogeneous proteins were also seen in immunofluorescence experiments. HyJagged appeared stronger localised at membranes whereas HvNotch showed a slightly stronger staining of punctae in the vicinity of the membranes. In interstitial cells membrane staining was completely lacking with the HvNotch antibody.
In Hydra, Notch signalling might also be regulated through post-translational modification of receptor and ligand as genes encoding metalloproteases of the ADAM family, the glycosyltransferase Fringe and the ubiquitin ligase Mindbomb, which regulates endocytosis of Notch ligands, are present in the genome [12]. Their expression patterns might be conclusive in the future in order to understand the precise molecular mechanism by which Notch signalling exerts its profound impact on pattern formation and stem cell differentiation in Hydra.

Conclusions
We report here the identification of the Hyjagged gene, which encodes a protein with great similarity to canonical Notch ligands from worms, insects and vertebrates. Its mRNA is especially strongly expressed at the constriction site of budding animals, confirming the conserved role for Notch ligands during boundary formation. Furthermore, both HvNotch and HyJagged proteins are found in all cells of adult animals indicating a general role for Notch signalling in regulating cellular communication in Hydra. The differential distribution of both proteins in endosomes indicates that endosomal pathways may play an important role in modulating Notch signalling.

Whole mount in situ hybridisation experiments
Single whole mount in situ hybridisation experiments were carried out using Digoxigenin-labeled RNA probes as described before [17,28].

Construction of GFP and RFP fusion proteins
Full-length Hvnotch was cloned into the NheI and SmaI sites of the HoTRed expression vector. Full-length Hyjagged as well as Hyzic were cloned into the modified EGFP expression vector HoTG(reen) [29] into the NheI and SmaI sites for overexpression from the Hydra actin promoter in hydra cells.
For expression in mammalian cells HvNotch and Hyjagged genes were cloned into the EcoRI and SmaI sites of the expression vector pEGFP (Clontech), leading to expression of C-terminally GFP-tagged proteins.

Transfection of Hydra cells
GFP-and RFP-constructs were introduced into hydra cells using a particle gun (Biorad) as described previously [29]. After 1-2 days cells expressing the GFP and RFP fusion proteins were clearly visible.

Production of recombinant proteins
To allow expression in E.coli the coding sequences for HvN ΔE (HvNotch without its extracellular domain), HvNICD , Hyjagged-ICD, Hyzic, HyHes and Cnash were cloned into the vector pRSET (Invitrogen) using the BamHI and EcoRI sites for HvN ΔE , HvNICD, Hyjagged-IC and Hyzic, and the EcoRI and XhoI sites for Cnash and HyHes.

Antibody production
Anti-Notch antibody was produced in chicken by Davids Biotechnologie GmbH by immunisation with recombinant HvN ΔE . A portion of the resulting polyclonal antibody was then affinity purified.
For the anti-JAG-IC antibody a peptide corresponding to a stretch of the intracellular domain of HyJagged (VNKDNLKKGIFKTISRKS) was produced by Davids Biotechnologie GmbH and used for immunisation of rabbit. The resulting antibody was then affinity purified.
For the anti-HES antibody a peptide corresponding to a stretch of the region shortly before and the begin of the basic domain of HyHES (RHPMKEKRRANKPLLER) was produced by Davids Biotechnologie GmbH and used for immunisation of rabbit. The resulting antibody was then affinity purified.
Monoclonal anti-ZIC7A12-and anti-CNA7B10-antibodies were produced in rat as described in [30]. Recombinant HyZic and CnASH were used for immunisation.

Antibody staining of HEK293T
HEK293T cells at 70-80% confluence were seeded on glass coverslips in 6-well culture plates and incubated for 24 h at 37°C. After that the cells were transfected with the specific DNA. After 24 h the coverslips were washed twice with PBS, and the cells were fixed in 4% PFA in PBS for 15 minutes at room temperature. After washing the cells again, they were permeabilized for 15 minutes with 1% Triton X-100 in PBS, washed twice with PBS and incubated in blocking solution (10% FCS, 0,2% Tween 20, PBS) for one hour at room temperature. Cells were then incubated with primary antibody (anti-Notch antibody (dilution 1:100) or anti-JAG-IC antibody (dilution 1:50)) for further 60 minutes at room temperature. After that, they were washed three times for 10 minutes in washing solution (1% BSA, 0,2% Tween 20, PBS) before they were incubated in secondary antibody for one hour at room temperature. Finally, cells were washed three times for 10 minutes with washing solution, incubated with TO-PRO 3 (Molecular probes, 1 μg/ml in PBS) for 5 minutes and mounted on slides with Vectashield (Alexis Biochemicals).
Other immunofluorescence staining experiments were performed by relaxing animals with 2% urethane and fixation with 4% PFA in hydra medium for 1 hour. Hydra were then washed three times for 10 minutes with PBS before they were permeabilized for 15 minutes with 0,5% Triton X-100 in PBS. Animals were blocked for 15 minutes with 0,1% Triton X-100, 1% BSA in PBS and incubated in anti-Notch antibody (dilution 1:100), anti-JAG-IC antibody (dilution 1:50), anti-Nv1 (dilution 1:2) or anti-ZIC7A12 (dilution 1:1) for 1 h at room temperature. Hydra were washed three times 10 minutes in PBS followed by incubation in secondary antibody for 2 h at room temperature. Before the animals were stained with TO-PRO3 (Molecular probes, 1 μg/ml in PBS) they were washed in PBS three times 10 minutes. Finally they were mounted on slides with Vectashield (Alexis Biochemicals). cells (lanes Hydra Ly) probed with monoclonal rat antibody anti-ZIC7A12 (A). Control with a rat IgG that had been raised against CnASH (B). This antibody recognises a band below 43 kDa in hydra lysates. The anti-ZIC7A12 recognises a band at 46 kDa, which corresponds to the predicted size for HyZic. (C-F) Single hydra epithelial cell ectopically expressing HyZic-GFP fusion protein from HotG after transfection of the animals with particle gun. (C) DNA staining with TO-PRO3, (D) HyZic-GFP, (E) anti-ZIC7A12 staining, (F) merged images in false colours: DNA (blue), HyZic-GFP (green), anti-ZIC7A12 staining (red); Confocal sections; scale bar: 2 μm.