Cell-type-specific origins of spinal rhythmicity at different locomotor speeds in larval zebrafish

Reviewer #1 (Public Review):

Summary:
In this very interesting study, Agha and colleagues show that two types of Chx10-positive neurons (V2a neurons) have different anatomical and electrophysiological properties and receive distinct patterns of excitatory and inhibitory inputs as a function of speed during fictive swimming in the larval zebrafish. Using single-cell fills they show that one cell type has a descending axon ("descending V2as"), while the other cell type has both a descending axon and an ascending axon ("bifurcating V2as"). In the Chx10:GFP line, descending V2as display strong GFP labeling, while bifurcating V2as display weak GFP labeling. The bifurcating V2as are located more laterally in the spinal cord. These two cell types have different electrophysiological properties as revealed by patch-clamp recordings. Positive current steps indicated that descending V2as comprise tonic spiking or bursting neurons. Bifurcating V2as comprise chattering or bursting neurons. The two types of V2a neurons display different recruitment patterns as a function of speed. Descending tonic and bifurcating chattering neurons are recruited at the beginning of the swimming bout, at fast speeds (swimming frequency above 30 Hz). Descending bursting neurons were preferentially recruited at the end of swimming bouts, at low speeds (swimming frequency below 30 Hz), while bifurcating bursting neurons were recruited for a broader swimming frequency range. The two types of V2a neurons receive distinct patterns of excitatory and inhibitory inputs during fictive locomotion. In descending V2as, when speed increases: i) excitatory conductances increase in fast neurons and decrease in slow neurons; ii) inhibitory conductances increase in fast neurons and increase in slow neurons. In bifurcating V2as, when speed increases: i) excitatory conductances increase in fast neurons but do not change in slow neurons; ii) inhibitory conductances increase in fast neurons and do not change in slow neurons. The timing of excitatory and inhibitory inputs was then studied. In descending V2as, fast neurons receive excitatory and inhibitory inputs that are in anti-phase with low contrast in amplitude and are both broadly distributed over the phase. The slow neurons receive two peaks of inhibition, one in anti-phase with the excitatory inputs and another just after the excitation. In bifurcating V2as, fast neurons receive two peaks of inhibition, while slow ones receive anti-phase inhibition.

Strengths:
This study focuses on the diversity of V2a neurons in zebrafish, an interesting cell population playing important roles in locomotor control and beyond, from fish to mammals. The authors provide compelling evidence that two subtypes of V2as show distinct anatomical, electrophysiological, and speed-dependent spiking activity, and receive distinct synaptic inputs as a function of speed. This opens the door to future investigation of the inputs and outputs of these neurons. Finding ways to activate or inhibit specifically these cells would be very helpful in the years to come.

Weaknesses:
No major weakness was detected. The experiments were carefully done, and the data were of high quality.

Reviewer #2 (Public Review):

Summary:
Animals exhibit different speeds of locomotion. In vertebrates, this is thought to be implemented by different groups of spinal interneurons and motor neurons. A fundamental assumption in the field has been that neural mechanisms that generate and sustain the rhythm at different locomotor speeds are the same. In this study, the authors challenge this view. Using rigorous in vivo electrophysiology during fictive locomotion combined with genetics, the authors provide a detailed analysis of cellular and synaptic properties of different subtypes of spinal V2a neurons that play a crucial role in rhythm generation. Importantly, they are able to show that speed-related subsets of V2a neurons have distinct cellular and synaptic properties and may utilize different mechanisms to implement different locomotor speeds.

Strengths:
The authors fully utilize the zebrafish model system and solid electrophysiological analyses to study the active and passive properties of speed-related V2a subsets. Identification of the V2a subtype is based directly on their recruitment at different locomotor speeds and not on indirect markers like soma size, D-V position etc. Throughout the article, the authors have cleverly used standard electrophysiological tests and analysis to tease out different neuronal properties and link it to natural activity. For example, in Figures 2 and 4, the authors make comparisons of V2a spiking with current steps and during fictive swims showing spike rates measured with current steps are physiologically relevant and observed during natural recruitment. The experiments done are rigorous and well-controlled.

Weaknesses:
The authors claim that a primary result of their study is that reciprocal inhibition is important for rhythmogenesis at fast speeds while recurrent inhibition is key at slow speeds. This is shown in Figure 6, however, the authors do not show any statistical tests for this claim. The authors also do not show any conclusive evidence that reciprocal inhibition is required for rhythmogenesis at fast speeds and vice versa for slow speeds. Additional experiments or modeling studies that conclusively show the necessity of these different inhibitory sources to the generation of different rhythms would be needed to strengthen this claim.

The authors do a great job of teasing out cellular and synaptic properties in the different V2a subsets, however, it is not clear if or how these match the final output. For example, V2aD neurons are tonic or bursting for fast and slow speeds respectively but it is not intuitive how these cellular properties would influence phasic excitation and inhibition these neurons receive.

It is not clear from the discussion why having different mechanisms of rhythm generation at different speeds could be an important circuit design. The authors use anguilliform and carangiform modes of swimming to denote fast and slow speeds but there are differences in these movements other than speed, like rostrocaudal coordination. The frequency and pattern of these movements are linked and warrant more discussion.

Reviewer #3 (Public Review):

The manuscript by Agha et al. explores mechanisms of rhythmicity in V2a neurons in larval zebrafish. Two subpopulations of V2a neurons are distinguishable by anatomy, connectivity, level of GFP, and speed-dependent recruitment properties consistent with V2a neurons involved in rhythm generation and pattern formation. The descending neurons proposed to be consistent with rhythm-generating neurons are active during either slow or fast locomotion, and their firing frequencies during current steps are well matched with the swim frequency they firing during. The bifurcating (patterning neurons) are active during a broader swim frequency range unrelated to their firing during current steps. All of the V2a neurons receive strong inhibitory input but the phasing of this input is based on neuronal type and swim speed when the neuron is active, with prominent in-phase inhibition in slow descending V2a neurons and bifurcating V2a neurons active during fast swimming. Antiphase inhibition is observed in all V2a neurons but it is the main source of rhythmic inhibition in fast descending V2a neurons and bifurcating neurons active during slow swimming. The authors suggest that properties supporting rhythmic bursting are not directly related to locomotor speed but rather to functional neuronal subtypes.

This is a well-written paper with many strengths including the rigorous approach. Many parameters, including projection pattern, intracellular properties, inhibition received, and activity during slow/fast swimming were obtained from the same neuron. This links up very well with prior data from the lab on cell position, birth order, morphology/projections, and control of MN recruitment to provide a comprehensive overview of the functioning of V2a interneuronal populations in the larval zebrafish. The overall conclusions are well supported by the data. Weaknesses are relatively minor and were largely related to terminology for some of the secondary conclusions.

1. The assumption is made that all in-phase inhibition is recurrent and out-of-phase inhibition is reciprocal. The latter is likely true but the definition of recurrent may be a bit loose as could be multisegmental feed-forward inhibition as well.

2. In a few places, it is mentioned that the properties of the V2a-D neurons are consistent with pacemakers. This could be true of both the V2a-D and -B neurons that burst in response to depolarizing steps but the properties of the remaining (fast) V2a-D neurons do not seem to be consistent with pacemakers, based on the properties shown. Tonic firing at a frequency related to the locomotor speed the neuron is active during and strong antiphase inhibition may instead suggest a stronger network component driving the rhythmicity.