Homologs of all bilaterian glial TFs, including Gcm, are found in a cnidarian model organism, Nematostella vectensis, but not in ctenophores (Fig. 1b). Moreover, among basal metazoans, the Gcm domain is conserved only in the Cnidaria (Supplementary Fig. S1). This raises an intriguing possibility of glial birth coincident with the appearance of neurons, utilizing classical chemical transmitters like glutamate.
To develop an unbiased view of NvGcm targets in N. vectensis, we knocked it down using NvGcm-specific siRNA 30 and performed transcriptomic analysis (Fig. 2a). KD efficiency of NvGcm-targeted siRNA was ~ 75% for samples selected for RNA-seq analysis (Fig. 2b). To identify genes affected by NvGcm KD, we performed differential gene expression (DE) analysis and identified genes that were downregulated or upregulated in NvGcm-depleted larvae (Fig. 2c).
We next analyzed the composition of all DE genes (p-adj < 0.2, 214 genes) with respect to their pathway groups, as defined in the KEGG PATHWAY Database. All candidate genes were grouped into the same eleven functional categories to characterize NvGcm candidate target genes in Drosophila31 (Fig. 2d). Functions of about a fourth of the identified genes are unknown. As in Drosophila, transcription factors and genes encoding ion transporters and channels constitute 8% and 7% of affected genes respectively, suggesting a possible function in cell fate specification and homeostatic control of NvGcm. A large fraction (15%) of NvGcm candidate target genes are structural and cell adhesion molecules, whereas 4% are signaling-related molecules. Another 10% of these genes are enzymes involved in general metabolism. All DE genes and their functional classifications are listed in Supplementary Table S2.
To further validate NvGcm candidate target genes, we selected several genes from each of the major GO groups for RT-qPCR analysis (Fig. 3a). We ignored those with functional groups that would not contribute to understanding the NvGcm-regulated program (unknown function, ncRNA/miscRNA, RNA splicing, ribosome, RT/retrotransposons, etc.). NvGcm KD efficiency was assessed with RT-qPCR for each experimental batch (Fig. 3b). We selected 33 genes for RT-qPCR validation and confirmed that 21 of them are up/downregulated in NvGcm-depleted larvae (Fig. 3a). In addition to these 21 genes identified in RNA-seq data, we confirmed that two genes encoding GABA transporters (Gats) are affected by NvGcm KD. We grouped validated NvGcm candidate target genes into nine functional categories, described below (Fig. 3a). All RT-qPCR validated candidate genes, their functional classifications, and RNA-seq attributes are listed in Supplementary Table S3.
Interestingly, NvGcm downstream targets in Nematostella included genes essential in bilaterian neuronal and glial functions. First, are genes encoding proteins involved in glucose and lipid metabolism: Sorl1 (sortilin-related receptor 1), Pla2s (phospholipases A2), P2Y6-like (purinoceptor P2y6-like), Sglt5-like (sodium-dependent glucose cotransporter 5-like), and Pkm (pyruvate kinase) (Fig. 3c). Second, there are post-translational protein modification enzymes required for axonogenesis and axonal support: RPTPsigma-like (receptor-type tyrosine-protein phosphatase S-like) and Ogt (UDP-N-acetylglucosamine–peptide N-acetylglucosaminyltransferase) 32 (Fig. 3c). Third, there are TFs involved in development of the nervous system and in driving glio/neurogenesis: Ets-related and Zic4 (zinc finger) (Fig. 3c). Fourth, several ion channels and chemical neurotransmitter transporters are among the candidate genes: Eaat1 (glutamate transporter), Lrrc8a (volume-regulated anion channel), Gats (GABA transporters), Pkd1l2 (transient receptor potential polycystic (TRPP) channel), and Trpa1 (transient receptor potential cation channel) (Fig. 3c). Finally, candidate genes include cell adhesion and signaling molecules involved in organizing synapses and promoting glio-neuronal interactions: Caspr2 (contactin-associated protein-like 2), Adgrf5 (adhesion G protein-coupled receptor F5), Sdk1 (sidekick 1), ADGRV1-like (adhesion G protein-coupled receptor V1-like), an integrin ligand Edil3 (EGF-like repeat and discoidin I-like domain-containing protein, Notch1, and Mpeg1 (Fig. 3c).
We next analyzed expression profiles of NvGcm and its downstream genes using single-cell transcriptomic (SCT) data from adult N. vectensis to furnish an overview of functionally fully differentiated cell repertoires 10. We found that NvGcm and most candidate genes are predominantly expressed in neuronal cell clusters (Fig. 4a). NvGcm is broadly expressed across neuronal cell clusters, with the highest expression in C34, C40, and C57 (Fig. 4b). Expression of NvGcm mRNA in a relatively broad range of cell types suggests several possible explanations. If it indicates the breadth of cell types in which this transcription factor actually functions, NvGcm may be involved in regulating expression of these downstream genes in various neurons, rather than deciding the fate of specific neuronal cell types, such as glia. Alternatively, alteration of gene expression by NvGcm depletion may be due, at least in part, to non-cell-autonomous effects. Among candidate genes, a glutamate transporter, NvEaat1, is of particular interest, as the highest expression of NvEaat1 mRNA is found in C34 and C57, clusters enriched with NvGcm (Fig. 4b). This is in sharp contrast to neuropeptides, e.g., PRGamide, that tends to be expressed only in specific neuronal clusters (Fig. 4c). In Drosophila, expression of excitatory amino acid transporters, Eaat1 and Eaat2, depends on Gcm34,35, suggesting that Eaat1 is a conserved Gcm target. Although these data reveal NvGCM-dependent gene expression, it is helpful to confirm protein expression to gain further insight into cells in which downstream genes function. Transcriptomes and proteomes tend to be correlated overall, but there are known exceptions 36,37.
To obtain information on morphological and physiological features of cells expressing glial genes, we first tried to identify cells expressing NvGCM protein using a specific antibody. However, our NvGCM antibody identified multiple bands in western blotting and no clear signal in immunostaining, probably due to the low expression level of NvGcm mRNA (Fig. 4a). We next tried commercially available EAAT1 antibodies. An antibody from German Research Products gave a strong, specific signal in western blotting, while an antibody from Bio-Techne recognized multiple proteins of various molecular sizes (Supplementary Fig. S4a). Nevertheless, both EAAT1 antibodies detected a band with the expected size of NvEAAT1 (65 kDa) and identical staining patterns in immunostaining (Fig. 5a).
EAAT1-positive cells began to be detected at late gastrula stage in outer ectoderm of the aboral region (Fig. 5b). At late gastrula and early planula stages, only a few cells were detected. Numerous EAAT1-positive cells became visible in a broad aboral region from planula through primary polyp stages. The salt-and-pepper-like distribution of NvEAAT1-expressing cells in aboral epithelium is reminiscent of expression patterns of other genes expressed in neurons 38,39. However, EAAT1-positive cells exhibited flat morphology without clear protrusions (neurites), a characteristic different from that of cnidarian interneurons with spherical or flask-shaped cell bodies (Fig. 4b). In fact, immunostaining for PRGamide, a neuropeptide most abundantly expressed in the NvGcm-enriched neuronal cluster during the larval stage (Supplementary Fig. S3), demonstrated typical morphology of neuropeptide-positive neurons at the aboral side of planula larvae (Fig. 6).
Taken together, our functional analysis of NvGcm identified a number of Gcm downstream genes including neuronal/glial genes involved in glutamate quenching in N. vectensis. Although mRNAs of NvGcm and many of its downstream genes, including EAAT1, were widely expressed in various neuronal clusters, EAAT1 protein was visualized mainly in a population of cells with non-neuronal morphology.