NvDclk1 is expressed in scattered cells from blastula stage on
We identified two doublecortin-like kinase genes in the Nematostella genome, which we termed NvDclk1 and NvDclk2. Both contain two dcx domains, typical of the Dclk1/2 subgroup and different from most Dclk3s that possess only one dcx domain [9, 64]. Two additional predicted genes encoding two N-terminal dcx domain, but no kinase domain, were most similar to vertebrate Dcdc2 genes in BLAST searches (Fig. 1A, [9, 11]). For NvDclk1, we cloned two splice variants, one encoding the dcx and kinase domains (NvDclk1- long or NvDclk1l); and one that lacks the kinase domain (NvDclk-short or NvDclk1s, Fig. 1A). NvDclk1s is not a truncated version of NvDclk1l, instead the entire sequence after the second dcx domain is derived from exons that are not present in the mature NvDclk1l transcripts (Fig. 1A). Using a probe that recognizes both splice variants, we found that NvDclk1 is first detectable at mid-blastula stage in a patch of tissue (Fig. 1B). At late blastula stage, this tissue thickens, identifying it as the future oral region that will give rise to the endoderm (Fig. 1C, [49, 50]). In addition, expression becomes visible in individual cells distributed throughout the blastoderm (Fig. 1C). The number of these cells increases during gastrulation, whereas the expression in the invaginating endoderm ceases (Fig. 1D, E). During planula stages, expression of NvDclk1 remains detectable in the scattered ectodermal and endodermal cells of the body column and in particular around the oral pole, the site where the tentacles with a high density of cnidocytes form (Fig. 1F, G). NvDclk2 is expressed throughout the Nematostella tissue, with higher levels in the pharynx during planula stages (Fig. 1H-J). Since the expression pattern of NvDclk1 resembles the distribution of neural cells, we focused our subsequent analyses on this gene.
A NvDclk1::GFP reporter line labels cnidocytes and neurons
To gain better insight into a potential expression of NvDclk1 in neural cells, we first analysed existing transcriptome data. NvDclk1 transcripts are enriched in transcriptomes derived from NvElav1::mOrange+ neurons (3.4 fold) and from NvNCol3::mOrange2+ cnidocytes (2.1 fold), both generated at primary polyp stage [61, 63, 65, 66]. Next, we generated a transgenic reporter line in which a 2.9kb genomic DNA fragment immediately upstream of the NvDclk1 start codon drives the expression of a membrane-tethered green fluorescent protein (GFP). We used double fluorescence in situ hybridization (DFISH) in this NvDclk1::GFP transgenic line to assess how well the line represents NvDclk1 expression. At gastrula and planula stages, the GFP expressing cells were also expressing NvDclk1, but we also observed many NvDclk1 expressing cells that did not express GFP (Fig. 2A-F). The transgenic line can thus be used for tracing NvDclk1-expressing cells, though it does not identify all of these cells and their progeny. GFP-positive cells were detectable from gastrula stage on (Fig. 2G) and remained visible throughout the body column at planula and tentacle bud stages (Fig. 2H-K). From mid-planula stage on, strong GFP signal was visible at the oral end, in the developing tentacles (Fig. 2I-K). While the signal in the tentacles was readily visible in animals derived from crosses of NvDclk1::GFP and wildtype animals, the signal in the body column was clearly visible only in a fraction of planulae and polyps derived from incrosses of NvDclk1::GFP animals. We assume that animals with visible signal in the body column contain two alleles of the transgene. Notably, we did not observe GFP expression in the pre-endodermal plate and the invaginating endoderm, despite clear signal in the NvDclk1 in situ hybridizations.
To test whether NvDclk1::GFP labels cnidocytes, we labelled cnidocysts (the extrusive capsule inside cnidocytes) with an antibody against the minicollagen NvNCol3 [67]. Both at gastrula and planula stages, many of the GFP-positive cells also stained for NvNCol3 (Fig. 3A, B). At gastrula stage, many of the NvNCol3 - labelled cnidocysts were small (Fig. 3A), suggesting that the transgene identifies cnidocytes at an early stage in their development. At planula stage, NvDclk1::GFP-positive cnidocytes were visible both in the pharynx and in the tentacle buds (Fig. 3B). Crossing the NvDclk1::GFP animals to a NvNCol3::mOrange2 line [65] showed that many of the labelled cnidocytes possess multiple long processes with bifurcations and varicosities, morphologically resembling neurites (Fig. 3C, D), as also described by Karabulut et al., 2022 [68] and partially visible in Weir et al., 2020 [69]. This network of cell processes was visible both in the tentacles and in the body column.
Observation of polyps mosaic for the NvDclk1::GFP transgene (i.e. F0 animals injected with the NvDclk1::GFP plasmid) identified differences in the morphology of labelled cells in different parts of the animals. Cnidocytes around the mouth opening and at the tentacle tips possessed long stalks connecting the cell body to the base of the epithelium, with the stalk branching into several shorter processes at the base (Fig. 3E-G). Cnidocytes with more proximal positions in the tentacle (closer to the mouth) and cells in the body column extended several processes, but they lacked prominent stalks towards the base of the epithelium (Fig. 3E, H, I).
Since NvDclk1 transcripts are enriched in NvElav1::mOrange+ neurons, we next tested whether the NvDclk1::GFP transgene can be observed in these neurons. In double transgenic animals (with one allele of each transgene), GFP signal in the body column of planulae and primary polyps was weak, whereas the NvElav1::mOrange transgene labelled a large number of neurons and their neurites (Fig. 4A-C), as previously described [63]. In older polyps, starting at around six weeks of age, the NvDclk1::GFP transgene was visible in most of the NvElav1::mOrange-expressing neurons (Fig. 4D-F).
Taken together, the NvDclk1::GFP transgene labels cnidocytes that form an extensive network of neurite-like processes, and it is visible in a large part of the NvElav1::mOrange+ neurons of juvenile polyps.
NvDclk1s localizes to microtubules in cnidocytes
Dcx-domain containing proteins have been shown to bind to microtubules, and microtubules support the development of the cnidocyst in Hydra and other cnidarians [70–72]. To visualize the localization of microtubules in cnidocytes, we generated a transgene in which the regulatory elements of NvDclk1 drive the expression of the microtubule-binding domain of human ensconsin fused to three copies of GFP (NvDclk1::ensconsin-GFP, [73]). Injection of this construct labelled patches of cells in varying areas of the animals. In the tentacle tips, the GFP signal was prominent in filamentous structures at the apical end of cnidocytes and in the cnidocil, the sensory structure protruding from the cnidocytes into the exterior (Fig. 5A). In the body column, we found cells with labelling of a convoluted thread-like structure connected to an oval capsule with more prominent GFP signal on one side of the long axis of the capsule (Fig. 5B, C). The non-invaginated thread suggests that these cells are immature cnidocytes and the observations with the NvDclk1::ensconsin-GFP transgene are in line with descriptions of microtubules in cnidocytes based on electron microscopy and immunohistochemistry [70–72]. To test whether NvDclk1 also localizes to microtubules in cnidocytes, we next generated a transgene in which a NvDclk1s-mCherry fusion protein is expressed under the control of the NvDclk1 regulatory elements (NvDclk1::Dclk1s-mCherry). We injected this construct together with the NvDclk1::ensconsin-GFP construct to observe potential colocalization. Though we did not recover double positive patches in the tentacle tips, cells in the body column showed near complete colocalization of the ensconsin-GFP and Dclk1s-mCherry signals (Fig. 5E-G). This suggests that NvDclk1 indeed is a microtubule binding protein.
NvDclk1 mutants lack mature cnidocytes
To analyze the function of NvDclk1 in cnidocyte development, we generated a mutant using the CRISPR/Cas9 system [74, 75]. The mutant has a 53bp deletion located at the beginning of the first dcx domain. In the predicted protein resulting from this mutation, the first 14 amino acids of the first dcx domain are as in the wildtype allele, followed by 23 amino acids not found in the wildtype allele and a premature stop codon. The predicted protein is thus truncated before the middle of the first dcx domain and the mutation affects both splice variants (Additional Fig. 1A-C).
In crosses of heterozygous animals, we observed approximately 25% of primary polyps in which cnidocysts (the capsule inside the cnidocytes) were not visible in the tentacle tips (Fig. 6A, B) or the body cloumn (Fig. 6C, D) by light microscopy of living polyps. The cnidocyst phenotype in these animals was confirmed by a lack of signal from a modified DAPI staining protocol that labels the content of mature cnidocysts (Fig. 6E, F [76, 77]). PCR confirmed that these animals represent homozygous mutants for NvDclk1 (Additional Fig. 1D). Histological staining of animals with this phenotype showed that a small number of cnidocyst-like structures were present in the tentacles, but they did not develop into the elongated capsules with an internalized thread that are observed in mature cnidocytes (Fig. 6G, H). Transmission electron microscopy revealed bent tentacle cnidocysts in the mutants that had not invaginated the thread (Fig. 6I, J) and often had an uneven surface of the capsule (Fig. 6K-M).
To analyze whether the nervous system is affected in NvDclk1 mutants, we crossed the mutant allele into the NvElav1::mOrange line. Crosses of animals carrying both the transgene and one mutant allele resulted again in offspring that lacked cnidocysts in the tentacle tips. In contrast to the absence of mature cnidocytes, we could not detect defects in the NvElav1::mOrange+ nervous system (Fig. 6E, F). We note, however, that subtle changes in the development or function of the nervous system would be difficult to detect with the available tools.
Taken together, the data show that NvDclk1 is required for the development of cnidocytes, a derived neural cell type specific to cnidarians.