Characterization of NvLWamide-like neurons reveals stereotypy in Nematostella nerve net development

Cnidarian nervous systems are traditionally described as diffuse nerve nets lacking true organization. However, there are examples of stereotypical structure in the nerve nets of multiple cnidarian species that suggest nerve nets are organized. We previously demonstrated that the NvashA target gene NvLWamide-like is expressed in a small subset of the Nematostella nerve net and speculated that observing a few neurons within the developing nerve net would provide a better indication of potential stereotypy. Here we document NvLWamide-like expression more systematically. NvLWamide-like is initially expressed in the typical neurogenic salt and pepper pattern within the ectoderm at the gastrula stage, and expression expands to include endodermal salt and pepper expression at the planula larval stage. Expression persists in both ectoderm and endoderm in adults. We generated an NvLWamide-like::mCherry transgenic reporter line to visualize the neural architecture. NvLWamide-like is expressed in six neural subtypes identifiable by neural morphology and location. Upon completing development the numbers of neurons in each neural subtype are minimally variable between animals and the projection patterns of each subtype are consistent. Between the juvenile polyp and adult stages the number of neurons for each subtype increases. We conclude that cnidarian nerve nets are organized, develop in a stereotyped fashion, and that one aspect of generating the adult nervous system is to modify the juvenile nervous system by increasing neural number proportionally with size.

aspect of generating the adult nervous system is to modify the juvenile nervous system by increasing neural number proportionally with size.
Cnidarians (e.g. jellyfish, corals, anemones, Hydra, etc.) represent the sister taxon to the bilaterians, and their nervous systems are comprised of a nerve net wherein neurites emanating from neural soma scattered throughout the body form a "net" encompassing the organism (Dunn et al., 2008;Hejnol et al., 2009;Rentzsch et al., 2016a). The phylogenetic position of cnidarians in relationship to bilaterians together with the fact that they have a nerve net, puts this group in a unique position to inform and potentially test hypotheses about the origin and evolution of bilaterian centralized nervous systems. However, very little is known about the developmental patterning of cnidarian nerve nets, which makes it difficult to compare components of cnidarian nerve nets to their putative bilaterian counterparts. In particular, it isn't clear if stereotypical development of cnidarian nerve nets occurs. Thus, determining the stereotypy of nerve net development and identifying distinct neural subtype markers is important to then assess how nerve nets are patterned.
Cnidarian nerve nets have traditionally been described as being diffuse and disorganized in the sense that there is minimal reproducible architecture that is obvious from animal to animal within a given species. However, a number of observations suggest cnidarian nerve nets are more organized than usually appreciated. First, there are obvious structures identified in many cnidarian nervous systems. For example, the sea anemone Nematostella vectensis has clearly visible longitudinal tracts that run the length of the oral--aboral axis and are stereotypically positioned over each mesentery structure (Layden et al., 2016b;Marlow et al., 2009;Nakanishi et al., 2012;Rentzsch et al., 2016a). In Hydra the foot of the animal has been identified as an organizing center for distinct behaviors, which indicates unique functions are regionally assigned within the nerve net. There are distinct subsets of neurons with particular spatial locations that are identifiable in Hydra and Nematostella (Anderson et al., 2004;Grimmelikhuijzen and Spencer, 1984;Koizumi et al., 2004;Marlow et al., 2009). A nerve ring containing at least 4 different neuronal subsets has been identified at the base of tentacles in Hydra oligactis and thought to be involved in feeding behaviors (Koizumi et al., 2015).
Two marginal nerve rings have been identified in the jellyfish Aglantha (Mackie, 2004;Donaldson et al., 1980;Roberts and Mackie, 1980). Hydrozoan nerve rings are visualized with various different antibodies that do not overlap/colocalize, suggesting that distinct neuronal subpopulations populate the nerve ring, although the number of neurons within and organization of these subpopulations is not clear (Koizumi et al., 2015). Altogether, these observations across many cnidarian species hint at more patterning and organization within a typical cnidarian nerve net than once thought.
The sea anemone Nematostella vectensis is an anthozoan cnidarian model system that has grown in popularity due to its ease of culture and the ever--growing repertoire of genomic tools available (Darling et al., 2005;Layden et al., 2016b;Putnam et al., 2007). While Cnidarians posses both ectodermal and endodermal nerve nets, the developmental origin varies between species. In Nematostella, neurogenesis occurs in both the ectoderm and endoderm (Nakanishi et al., 2012).
Studies primarily conducted in Nematostella suggest that cnidarian and bilaterian neural patterning likely share a common evolutionary origin. (Layden et al., 2016b;Rentzsch et al., 2016b). For instance, MEK/MAPK , Wnt (Leclère et al., 2016;Marlow et al., 2013;Sinigaglia et al., 2015), BMP (Saina et al., 2009;Watanabe et al., 2014) and Notch (Layden and Martindale, 2014;Richards and Rentzsch, 2015), all key regulators of bilaterian neurogenesis, have similar roles in Nematostella. Additionally, key neurogenic transcription factors such as NvsoxB (2) and Nvath--like are expressed in proliferating neural progenitor cells (Richards and Rentzsch, 2015;. The bHLH proneural transcription factor NvashA is necessary and sufficient to promote neurogenesis including the formation of individual neuronal subtypes (Layden et al., 2012). It has been suggested that the conserved neurogenic programs described above coordinate with axial patterning to generate distinct neural subtypes, similar to mechanisms that pattern bilaterian central nervous systems (Layden et al., 2012;Leclère et al., 2016;Rentzsch et al., 2008;Watanabe et al., 2014). However, an understanding of cnidarian neural architecture or the patterning of distinct neural subtypes in these organisms is lacking, and thus with few exceptions there are not clear neural subtypes that can be used to screen for neural patterning defects.
Our previous efforts identified the NvLWamide--like::mcherry transgenic line . We speculated that the neural number and neurite projection patterns were consistent from animal to animal, which if true would suggest development of nerve nets is at least in some cases stereotyped. Here we characterize the NvLWamide::mcherry transgenic line in more detail in order to better determine the development and organization of the juvenile and adult nervous system. We identified at least six neuronal subtypes described by NvLWamide::mcherry that are consistently found at particular locations in the polyp and with distinct neurite morphologies. By the juvenile polyp stage the number of each neural class is predictable from animal to animal and these subtypes persist through adult stages. Interestingly, the number of neurons for each individual neuronal subtype increases with age, and can in some cases be easily quantified revealing a positive correlation between neural number and body length. These combined data imply that the Nematostella nervous system develops and is modified in a stereotypical fashion, which contradicts the notion that nerve nets are loosely organized structures.

Animal care
All animals were maintained in 1/3X artificial seawater (ASW) with a pH of 8.1--8.2, with weekly water changes. Embryos were either grown at 25°C, 22°C or 17°C to the desired stages, and juvenile polyps were maintained at room temperature in the dark. Adult Nematostella were housed in the dark at 17°C, were fed brine shrimp 4 times a week, and were given pieces of oyster 1 week prior to spawning. Spawning was induced by changing the light cycle (Fritzenwanker and Technau, 2002;Hand and Uhlinger, 1992). Generation of NvLwamide--like transgenic animals was described by Layden et al, 2016. All transgenic animals were from a stable F1 line, and since there was no visible difference in mCherry expression in heterozygous versus homozygous animals both were utilized in these studies.

Quantification of mCherry/dsred and NvLWamide--like positive cells
To quantify the number of mCherry/α−dsred positive cells that colabel with NvLWamide--like positive cells the embryos were mounted to capture z--stacks of lateral views with a 1uM step size using a Zeiss LSM 880 (Carl Ziess) confocal microscope. Cells were then scored as single or double positive using the Imaris (Bitplane) imaging software.
Quantification of NvLWanide--like::mcherry--expressing neurons/ neuronal subtypes NvLWamide--like::mCherry expressing juvenile polyps were relaxed in 7.14% (wt/vol) MgCl2 in 1/3 X ASW for 10 minutes at room temperature. Animals were either examined and quantified live, subjected to live imaging, or imaged following a light fixation and phalloidin stain. Animals were fixed in 4% paraformaldehyde and 0.3% glutaraldehyde (in 1/3 X ASW) for 1 minute, followed by fixation in 4% paraformaldehyde (in 1/3 X ASW) for 20 minutes at room temperature on a rocker.
Animals were then washed in PBS + 0.2% Triton X--100 for 5x5 minute washes, then

NvLWamide--like expression is scattered throughout the ectoderm and endoderm during development.
NvLWamide--like is expressed in a subset of the nervous system, but its detailed spatiotemporal distribution has not been described (Layden et al., 2012).
We used mRNA in situ hybridization to better characterize expression of NvLWamide--like is also expressed in small number of endodermal cells.
Endodermal expression is first detected in early planula stage animals ( Figure 1C, yellow arrowhead). Endodermal expression is maintained throughout development ( Figure 1C--E, yellow arrowheads) and into adult stages (see below). Additionally, endodermal cells are associated with the mesenteries and longitudinal tracks that run the length of the oral--aboral axis ( Figure 2D) (Layden et al., 2016b;Nakanishi et al., 2012).

NvLWamide--like::mCherry expression recapitulates endogenous NvLWamide-like mRNA expression during polyp stages
To better determine what cell types express NvLWamide, we generated a stable F1 NvLWamide::mCherry transgenic reporter line, and either live imaged animals, or imaged animals that were lightly fixed and co--stained with phalloidin (see materials and methods) .
We found that mCherry expression followed the expression dynamics of the endogenous NvLWamide transcript (compare Figures 1 and 2) with the exception that mCherry positive cells are not readily detected in gastrula and early planula stage animals (Figure 2A  However, by adult stages the same neural subtypes identified juvenile polyps were present but were increased in number (Figure 7). Below we describe each neuronal class in detail. We conclude that the neurons present in the juvenile polyp represent the complement of neurons specified during embryonic and larval development and that additional neurons are not born until after the polyp begins to grow once it starts to feed.

Pharyngeal neurons
NvLWamide--like pharyngeal neural somas are located within the pharyngeal ectoderm ( Figure 4A, orange arrow; Figure 4B; Figure 5A, arrowheads). The neurites project out of the basal surface of the soma and encompass/wrap around the pharynx ( Figure 5A and B, arrows). NvLWamide--like pharyngeal cell bodies span the pharyngeal ectoderm ( Figure 5A, Supplemental Figure 1A, and Supplemental movie 1) and their apical surface appears to be exposed to the pharyngeal lumen. A view of the basal surface of the NvLWamide--like pharyngeal neurons suggests that initially the neurons are formed in a row ( Figure 5B, arrowhead), each with 2 neurites that project orthogonally to the oral--aboral axis ( Figure 5B, arrow).
NvLWamide--like neurons contribute to the pharyngeal nerve mesh that surrounds the pharynx in the adult Nematostella ( Figure 5C). By adult stages many NvLWamide--like pharyngeal neurons are detected throughout the pharynx and they do not appear to be restricted to a particular location. It is difficult to observe pharyngeal neurites in adult animals because the adult pharynx has robust red auto fluorescence. However, the NvLWamide--like pharyngeal neurites appear to maintain a mesh around the pharynx and do not appear to condense into a pharyngeal ring, which has been previously hypothesized to exist in Nematostella (Marlow et al., 2009). Based on the NvLWamide--like mRNA in situ and developmental time course in NvLWamide--like::mcherry transgenic animals (Figures 1 and 2) the pharyngeal neurons are born in the pharynx at early planula stages, coincident with or shortly after pharynx formation. As the animal matures and the pharynx increases size the number of pharyngeal neurons also grows.

Mesentery neurons
NvLWamide--like mesentery neurons are aptly named for their location in the mesentery ( Figure 4B; Figure 5D--F). Initially the juvenile polyp possesses few mesentery neurons ( Figure 5D), but in feeding adults NvLWamide--like mesentery neurons are located along the entire length of the mesentery, ( Figure 5E--F, arrowheads, and Supplemental Figure 1C). In adult Nematostella the soma seem to be in pairs ( Figure 5E, arrowheads) mirrored on either side of the tract of neurites generated along the mesentery by these neurons (Figures 5E, 5F, and Supplemental Figure 1C). We observe both unipolar (a single projection) ( Figure 5F and Supplemental Figure 1C, arrows) NvLWamide--like mesentery neurons and bipolar (two projections that project 180 o opposite one another) ( Figure 5F and Supplemental Figure 1C, arrowheads) NvLWamide--like mesentery neurons. The projections emanate from the basal pole and fasciculate into bundles that project along the long axis of the mesentery ( Figure 5F and Supplemental Figure 1C).
Because the tracts of neurites are immediately contacting the mesentery neural soma, it is difficult to estimate what percentage of the mesentery neurons are either uni--or bipolar. While we first observe NvLWamide--like mesentery neurons in the polyp stage, we speculate that their development occurs earlier, but that the delay in mCherry maturation makes identifying them in larval stages difficult.

Longitudinal Neurons
The  Table 1). Although the number is somewhat variable from animal to animal, the total number of neurons in each tract positively correlates with body length (Figure 7C), suggesting that the neuronal number is being controlled relative to the overall body length.

Tentacular Neurons
NvLWamide--like tentacular neurons are first detected shortly after tentacle bud formation ( Figure 4A, red arrow, and Figure 6A). Cell bodies are initially located both in the ectoderm ( Figure 6A and B, white arrowhead) and endoderm ( Figure 6A, yellow arrowhead) and are observed along the entirety of the proximal--distal axis of the tentacle. It is important to note that observing endodermal tentacular neurons is rare. A typical juvenile polyp tentacle has 1 to 3 cell bodies per tentacle. However, we have in one case observed seven cell bodies in one tentacle, suggesting that the exact number of neurons is somewhat variable. By adult stages the number of tentacular neurons dramatically increases and nearly all NvLWamide--like tentacular neurons are found in the ectoderm. It is unclear whether endodermal neurons migrate into the ectoderm or are lost by adult stages, but to date no trans--body layer migration has been described for Nematostella neurons. Tentacular neurites project in a fashion that appears to be similar to the tripolar neurons (described below) ( Figure 6B--C).

Tripolar and quadripolar neurons
The last two subtypes of neurons we identified are the NvLWamide--like  Figure 7C).
At the early juvenile polyp stage we also observe a single NvLWamide+ quadripolar neuron per animal ( Figure 6E). Quadripolar neurons have four neurites that emanate from the soma and are similar to the tripolar in that they are located in the ectoderm. When detectable, quadripolar neurons can be first identified as early as 12 days post fertilization (dpf). Interestingly, unlike the tripolar neurons, these neurons appear to be more restricted to the oral portion of polyp trunk. While quadripolar neurons are also observed in the adult ( Figure 6G) their numbers are not significantly increased, and their presence in the adult is difficult to detect. We cannot rule out that tripolar and quadripolar neurons are not the same neural subtype, but that there is some variability in the number of projections with three being significantly more common. We currently suspect that the tripolar and quadripolar neurons are the first neural cell types born at gastrula stages as that is the time we observe the most nascent ectodermal expression of NvLWamide ( Figure   1). Interestingly, the number of NvLWamide+ ectodermal cells decreases after gastrula stage suggesting that tripolar/quadripolar cells might be initially specified in excess and some of those cells undergo apoptosis. However, we cannot currently rule out that the early NvLWamide--like cells are unrelated to the tri and quadripolar cells, and that the tri and quadripolar cells are born later in larval stages.

NvLWamide::mCherry is expressed in Nematosomes
Nematosomes are a Nematostella specific feature, comprised of a collection of both cnidocyte (stinging) cells and non--cnidocyte cells whose identity is unclear (Babonis et al., 2016). We detect NvLWamide::mCherry expression in the nematosomes along the entire length of the tentacle lumen (data not shown), in the nematosomes harvested from an egg mass ( Figure 6K and L), and moving throughout the gut cavity of the polyp (not shown). Even though we did not detect nematosomes via mRNA in situ, previous reports detected NvLWamide expression in nematosomes by RNAseq analysis (Babonis et al., 2016).

Summary of results:
Our observations support stereotypy in the developing Nematostella nerve net. The juvenile polyp contains specific subtypes of NvLWamide--like+ neurons, whose position, number, and neurite projection patterns show minimal variation from animal to animal. These data argue that although the nerve net appears unstructured when viewed in its entirety, it is comprised of neural subtypes that display significant organization when observed individually. Additionally, our data suggest that the neural cell types patterned during embryonic and larval development pioneer the neurite architecture. During adult growth the nervous system is modified in part by increasing numbers of individual subtypes in accordance with increasing body size.

Discussion
The Nematostella nervous system is organized and stereotyped at the subtype level The Nematostella nervous system is often described as a "diffuse nerve net." Here we examine NvLWamide--like expressing neurons, which represent a small subset of the nervous system to better visualize potential organization that might otherwise be masked if too many neurons are labeled. We observed reproducible neuronal subtypes: longitudinal, tripole, quadpole, pharyngeal, tentacular, and mesenetery. Moreover, we showed that the relative number of neurons for each neuronal subtype is similar from animal to animal upon completion of development, implying that neural number is patterned during development. These developmentally specified neural subtypes persist throughout the adult stage.
However, the number of neurons in each subtype increases, and in the case of the longitudinal and ectodermal tripolar/quadripolar neurons there is a correlation between the numbers of neural bodies and body column length ( Figure 7C). We hypothesize that many neuronal subtypes are stereotyped in terms of number of neurons and projections from those neurons. Together our data support the idea that the Nematostella nerve net is not a "structurally simple, diffuse net", but instead specific patterns of neuronal subtypes and their projections are established during development and then neuronal number is modified in adult stages. Further characterization of the development and structure of other cnidarian developmental model nerve nets such as Clytia hemisphaerica and Hydractinia is necessary to confirm that this is a widespread phenomenon in cnidarian neurogenesis.

Putative functions of subtypes expressing NvLWamide--like
We hypothesize that the tripolar, quadripolar, many of the tentacular and pharyngeal neurons are likely sensory neurons. They are all present in the ectoderm or ectodermally derived structures and have a cell body that spans the depth of the epithelial layer. The morphology of their soma is also consistent with previously predicted ectodermal sensory neurons (Marlow et al., 2009;Nakanishi et al., 2012).

Mesentery and longitudinal neurons likely function as interneurons. Their
projections fasciculate into either the mesentery or the previously described longitudinal neural tracts respectively (Marlow et al., 2009;Nakanishi et al., 2012). However, we remain agnostic in regards to either hypothesis for the mesentery neurons.

Development and patterning of the neuronal subtypes
The developmental mechanism(s) that pattern the NvLWamide--like subtypes remains unknown. Some evidence suggests that in Nematostella regional patterning likely interacts with neural programs to generate specific neural subtypes (Layden et al., 2012;Leclère et al., 2016;Marlow et al., 2013;Watanabe et al., 2014). This is most evident in the requirement of proper directive axis formation to pattern asymmetrically distributed GLWamide+ neurons (Watanabe et al., 2014), and in the expression of neural markers in the apical tuft, which requires proper establishment and maintenance of oral--aboral patterning programs (Leclère et al., 2016;Sinigaglia et al., 2015;. Additional work has shown that regional boundaries with distinct gene expression profiles are established along the oral--aboral axis, which could be integrated into the generic neural patterning mechanisms previously identified to generate specific neural subtypes (Kusserow et al., 2005;Marlow et al., 2013). The tissue layer in which the neural cell resides and the temporal window in which the neural progenitor is born are additional factors that may contribute to neural subtype specification. To date our understanding of the temporal patterning in Nematostella is poor, but future work aimed at understanding how distinct neural fates arise should not ignore this component.
Our data also speak to the cellular mechanisms related to generating the appropriate number of each neural subtype from neural progenitor cells. Vertebrate neurogenesis occurs by overproducing neurons and neurons that fail to properly synapse with target cells do not survive. Drosophila ventral nerve cords form from hardwired neuroblasts that generate neural subtypes in an exact lineage so that there is essentially zero variability in regards to neural number between animals (Schmid et al., 1999). Nematostella neurogenesis appears to be in between these approaches. We observe stereotyped numbers of each neural subtype, but typically there is a 1--2 cell number variability for each cell type between animals and even between radial segments in the same animal. These data suggest that neural progenitor lineages are not hardwired. Additionally, with the exception of the reduction in ectodermal NvLWamide--like positive cells in the aboral ectoderm, it does not appear that there is a massive over production of NvLWamide--like neurons that then compete for a limited number of targets. We hypothesize the neural progenitor lineages are not hardwired in Nematostella.
NvLWamide--like::mcherry expression describes a subset of the Nematostella nervous system. To fully understand the molecular and cellular programs that drive the development of diverse neural cell types, it will be necessary to generate new transgenic lines that identify additional subtypes of neurons in the Nematostella nervous system. The distinct neural subtypes will then need to be mapped to unequivocally establish their spatiotemporal origin during development, as well as the molecular program that specifies each cell type. Once we better understand the molecular programs controlling development of various neural cell types in cnidarians, we will then be poised ask questions regarding the relationship of bilaterian nervous systems to some or all of the cnidarian nerve net.        We were able to score two radial segments in each animal. N = 13, 5, 8, 14, 14, 9, 9, 4, 2, 1 for > 0.5 to 12 mm respectively.