Reviewer #2 (Public Review):

Kelly et al. strategically leverage state-of-the art scRNA-seq methods combined with unique strengths of the zebrafish larval model to identify gene expression patterns that underlie the different functional output of different neuronal circuits that converge on similar muscle groups. The results lead to the identification of ion channel and synapse associated genes that distinguish the neuronal components of a fast circuit mediating escape behavior from a rhythmic circuit mediating graded swimming.

The authors develop methods for isolation of single spinal cord neurons from 4 day post fertilization (dpf) zebrafish larvae. The 4 dpf neuronal circuits mediating escape vs. rhythmic swimming behavior have been extensively characterized allowing knowledge of the specific motor neuron and interneuron populations involved in one vs. the other circuit. (Work from the authors' research group has contributed to this strong starting point for this study.)

The transcriptomic analyses lead to the identification of clusters of cells sharing significant gene expression that distinguishes them from other clusters. Using well-known neuron subtype specific markers, the authors are able to assign a specific neuronal identity to about 2/3 of the cluster. Moreover, one other cluster results in the recognition in zebrafish of a neuronal cell type identified in the mammalian spinal cord, v0c, that they confirm to be present in zebrafish using solid markers. In addition, the results show that the zebrafish v0c population expressed markers of both cholinergic and glutamatergic neurons, while the mammalian v0c population is known to be cholinergic. (It is not clear whether the possibility that mammalian v0c neurons also express glutamatergic markers has been specifically tested, but it seems, at present, there is no evidence to suggest that might be the case.)

To zoom in on the question of molecular differences between the fast vs. rhythmic circuits, the authors focus on motor neurons as two different populations of neurons are involved in each circuit. (Along the way, they also identify markers that mark different subtypes of motor neurons.) They find that primary motor neurons (PMNs) involved in the fast circuit express a distinguishing cassette of ion channel and synapse associated genes. Moreover, the cassette of genes also is expressed by interneurons that function in the fast circuit. The results are illuminating and set the stage for many future exacting experiments.

As is true for significant work, the results open up and permit yet more rigorous and strategic analyses, running the gamut from specific molecules to behavior, of the circuit mechanisms underlying unique behaviors.

Overall, the work is carried out to high rigorous standards and the vast majority of conclusions are strongly supported by the results. However, there are a few instances of potential over-interpretation and points that could be further clarified/discussed:

1 - lines 412-414. The authors conclude that "Most importantly, and as detailed below, our scRNA seq revealed the ion channel and synaptic genes that serve to match specific neuronal function to behavior." That the authors have identified a gene cassette that distinguishes neurons of the fast escape circuit is a laudable finding. However, at this stage, to say that this gene cassette is the basis for unique circuit function and resultant behavior is a well-supported hypothesis that requires rigorous testing and not yet a solid conclusion. (Maybe that is what the authors meant, and I have misinterpreted the sentence.)

2 - lines 323-324: Given that ~ 6 hrs separates PMN from SMN birthdates (Myers et al. 1986) and that the study was done using 4dpf larval tissue, the possibility that the higher level of expression of transcription factors and RNA-biding factors in SMNs reflects "the less well differentiated state that accompanies the later birthdate of the SMns" seems unlikely.

3 - Fig 5 and Sup Fig 1:The authors mention that the unidentified cluster in the motor neuron set shares markers with non-skeletal muscle. I realize that this cluster is tangential to their focus. However, given that this cluster predominantly arises from the FACS sorted cells, it is worth considering that the cells might correspond to the pancreas.

4 - lines 113-115 and Fig. 1: The authors indicate that three clusters reflect cells that have mixed glial and neuronal cell expression. Is there any possibility that in a few instances, in the final single cell capture, that two rather than one cell were collected? (Again, not a major focus of the study but the cluster is commented on.)

Finally, as the transcriptomic information about glial cells will be of interest to many in the field, the authors are to be commended for depositng the data in congratulations to the authors for depositing the data in the publicly accessible Gene Expression Omnibus.

Figures and data