Escribir un comentario

Summary and overall impression:

In this preprint, Shaulsky et al. make a valuable contribution to our understanding of how the conductance of a voltage-gated potassium (Kv) channel in the visual system can contribute to neuronal spiking during coherent looming stimuli and infer how this conductance influences predator escape behaviors. The authors identified Shaker channels (homologous to the mammalian Kv1 channel family) in Lobula Giant Movement Detector (LGMD) neurons of grasshoppers as a key channel that selectively suppresses responses to incoherent black (looming) stimuli. The application of robust techniques applied provides solid evidence that the Shaker channel generates a 4-aminopyridine (4-AP) sensitive conductance that contributes to discerning between coherent and incoherent dark (or looming) stimuli. Additional evidence is required to establish whether only slowly activating and inactivating Shaker channels contribute to these neural computations or if other conductances from voltage-gated channels that can be similarly blocked by 4-AP are also involved. The authors propose that the biophysical properties of Shaker channels expressed in specific compartments of LGMD neurons enable computations at physiologically relevant timescales that could support predator escape behaviors. However, more details about the activation and inactivation of Shaker channels at the different stages of looming behavior responses are needed to support this claim. In addition, more details are necessary to determine whether Shaker channels are exclusively expressed in the dendrites of field A of LGMD neurons versus in field C of LGMD neurons or in other neurons within the circuit. Whether the 4-AP-sensitive conductance drives coherent looming computations and predator escape behaviors remains to be directly determined.

Strengths:

• Single-cell electrophysiology data combined with visual stimulation, pharmacology, and RNAi manipulations within a live animal and intact neural circuit are robust.

• Phylogenetic analyses show strong evolutionary conservation of Shaker channels in relation to other families of Kv channels, indicating a high likelihood of physiological importance across species.

Weaknesses:

• Subcellular patterns of Shaker channel expression between fields A and C of LGMD neurons were not determined experimentally.

• RNA interference (RNAi) approach utilized reduced global Shaker channel expression within the animal, limiting the specificity and interpretation of their results.

• Pharmacology experiments with 4-AP, a nonselective Kv channel inhibitor, could also have off target effects in other voltage-gated channels within the neural circuit. This also limits the specificity and interpretation of their results.

Major points:

• The authors suggest that the biophysical properties calculated from the 4-AP-sensitive currents, which include half-maximal voltages and time constants of the voltage dependence of activation and inactivation, are sufficient to explain how slowly activating and inactivating potassium conductances in neuronal dendrites of LGMD neurons contribute to visual object segmentation. However, their data do not fully support this conclusion for the following reasons:  

  • Although the pharmacology experiments with 4-AP from this and a previous study (PMID: 29667927) implicate Shaker channels, 4-AP can also block other Kv channels like the mammalian Kv2.1 and Kv3.1 channels. Other homologous channels in addition to Shaker channels could therefore contribute to the observed functional effects. To address this, the authors could provide additional evidence from their scRNA-seq data to indicate whether other Kv channels that can be blocked by 4-AP are expressed in LGMD neurons. This, in addition to Shaker protein level quantification in LGMD neurons, would strengthen their main claim.

  • Given that the RNAi experiments did not specifically target Shaker channels expressed in field A and that Shaker expression could have been depleted in other neurons within the circuit, their electrophysiology data do not specifically point to field A Shaker channels as the main effector channel. To determine how much Shaker channel expression was decreased in LGMD neurons after RNAi injections, the authors could repeat their RNAi experiments, pool single LGMD neurons, and perform qPCR and western blot analyses. Together with the authors’ scRNA-seq data, which indicated that the K+ channel gene that was most highly expressed in the LGMD encodes Shaker channels, these data could improve the robustness of their results. To address whether the expression of Shaker channels in other circuit neurons could contribute to the observed effects, the authors could collect sagittal sections of neural tissue from the head of the grasshopper and perform immunolabelling and confocal microscopy.

  • There could be additional conductances active from other voltage-gated channels with distinct biophysical properties that, when combined at the depolarized potentials where the effects of Shaker inactivation are more pronounced, could also drive the observed effects. The authors could take advantage of their scRNA-seq dataset to discuss why this is unlikely within their system given the voltage-gated channels that are expressed in LGMD neurons.

  • Visual object segmentation was not assayed directly in this study, as it is a different, more complex visual computation than neural computations made to discern between coherent and incoherent stimuli. The authors could expand upon their definition of object segmentation in the introduction and/or discussion section(s) to clarify that spatial coherence alone is not sufficient for segmentation but that it is functionally equivalent to object segmentation in the context of this study. Perhaps the authors could rephrase their claim and pivot their discussion on how suppression of black incoherent looming stimuli are influenced by the Shaker conductance and how visual object segmentation computations made by LGMD neurons can lead to prey escape behaviors.

• Wild-type Shaker channels exhibit both fast N-type inactivation and slow C-type inactivation. The authors should consider elaborating on why they emphasize the role of slow inactivation and not fast inactivation in this study.

• The RNAi experiments only partially reduced mRNA expression (~24% reduction) and net protein levels were not assessed after this global reduction of Shaker mRNA expression. Given that the Shaker mRNA was not specifically reduced in field A of LGMD neurons, this limits the interpretation that can be made between the 4-AP experiments and the RNAi + 4-AP experiments. In line 498, the authors compare Figures 3E and 4E. Rather than instructing readers to make comparisons between these different figures, the authors could plot an additional supplemental figure comparing the two figures directly for clarity.

• Because the low level of mRNA depletion without protein abundance confirmation confounds the authors’ results, the limitations with the global RNAi approach could be addressed in the discussion for clarity.

• Though the authors indicate that coherence discrimination computations are due to differences in the subcellular patterns of Shaker channel expression between fields A and C of LGMD neurons, the expression density and/or distribution of Shaker channels was not assessed directly in these regions and therefore limits the structure-function relationship conclusions that can be extracted from this study. The authors could consider rephrasing this claim and/or clarifying their rationale for saying this is so.

• The density and distribution of Shaker channels in the dendritic arbors of LGMD fields A and C were not assessed in this study directly but the presence of these channels was functionally characterized with pharmacology and single-cell electrophysiology in this study and a previous one (PMID: 29667927). The authors could make a more compelling structure-function argument if they  address this question experimentally with immunolabelling and imaging. This would support their conclusions by complementing their functional and computational modeling data.

• Given that the authors have experience recording the probability of escape jumps in unrestrained animals during a looming stimulus, the authors could replicate their experiment in RNAi injected animals. This would provide additional confirmation linking a single ion channel’s conductance to behavior and could elevate the impact of the study, with the caveat that RNAi effects are global and not specific to LGMD neurons. The contributions from other neurons in the circuit in the RNAi experiments thus remain unclear.

• Although the authors provide a reasonable rationale for not studying the effect on coherence selectivity for white stimuli in their RNAi + 4-AP experiments (i.e. Figure 4), one could argue that, blocking ion channel currents and silencing the translation of ion channels are mechanistically distinct processes, therefore these two experiments address different questions.

• As the authors alluded to in their discussion section, there are cases where genetic silencing of ion channels can elicit compensation by other ion channels. In other cases, interfering with the RNA expression of one ion channel can exert effects on other ion channel families, as seen in the Eichel et al. 2019, eLife (PMID: 31670657) study. Co-translational mechanisms could therefore contribute to different effects between the 4-AP pharmacology and RNAi experiments. Given the potential for these off-target effects in the RNAi experiments, an expanded discussion of what these could be in relation to Shaker expression patterns across the circuit being studied would also be helpful to readers.

• One way to definitively demonstrate that Shaker channels do not contribute to coherence selectivity computations in LGMD field C would be to label and image Shaker channels in the LGMD field C dendritic arbors in control and RNAi-injected animals and show that there are no changes in the expression and/or distribution of Shaker channels.

• The authors point out that the responses to loom stimuli could be lower in the RNAi-injected animals than the control animals due to a decrease in animal health following the physical stress of the injections. If this were the case, this could confound the results obtained in this study. To address this concern, the authors can state why multiple injections are needed to knock down Shaker channel expression and/or cite previous studies that used this methodology. Additional experiments could assay the viability of the animals and/or neuronal circuit orthogonally by using behavioral or immunolabelling and imaging approaches, respectively, to support the methodological rigor of this experiment.

• In Figure 6, there are significant differences between individual animal replicates in the inactivation voltage-dependence plot. The authors should consider clarifying why the same number of animals/cells were not utilized to generate both the activation and inactivation curves. Alternatively, the authors could increase their total number of animals/cells assayed in the inactivation curve analysis to a comparable number as those utilized to generate the activation voltage-dependence plot. This could also reduce the variability of the calculated inactivation time constants.

• The NEURON simulations of the LGMD do not completely reproduce the trends observed in their experimental data, as the modelled coherence selectivity for white stimuli appears to have an effect, but it is not clear if this effect is significant. To aid interpretation, the authors should consider providing further details about the statistical analysis of this data or elaborate on this discrepancy in their discussion.

Minor points:

• The temperature used for the electrophysiology experiments is not stated in this study. To improve clarity, the authors can include this important parameter, as the biophysical properties of Kv channels, including Shaker, are influenced by temperature.

• In line 360 of the results section, the authors could specify that genome assembly iqSchAmer2.1 was utilized in their analysis for additional clarity.

• In Figure 3E, the highlighted thick orange line in the 4-AP condition (indicating the representative trace that was chosen in Figure 3A) does not seem to represent the average value for the group. The authors could choose another representative trace and highlight the corresponding line in the Figure 3E plot with a thick orange line.

• In Supplemental Figure 4, the authors indicate that the percent expression levels of Shaker mRNA in wild-type and Shaker-RNAi animals was quantified relative to wild-type average. Given that the expression levels of Shaker mRNAs are inferred to be normalized to Gapdh mRNAs, it would be helpful to clarify this point in their methods section.

• A ~24% reduction in Shaker mRNA after RNAi seems low and is surprising given the robust functional effects observed in the responses to looming stimuli. Could the authors elaborate on why such a low reduction in Shaker mRNAs may be expected in this system compared to mammalian organisms where dsRNA interference can achieve a higher mRNA reduction?

• References indicating previously reported biophysical properties of Shaker channels are missing in the authors’ discussion of their activation (PMID: 9450944 and PMID: 9450946) and inactivation (PMID: 2122519) half-maximal voltages and time constant results in Figure 6. This will help clarify the authors’ interpretation for readers.

• The “percent permissive at rest” described in line 319 of the Methods section and utilized in Figure 6A is not defined in the manuscript. To improve clarity, the authors can elaborate on what this metric is and how it is calculated.

• In Figure 7 the plots corresponding to the LGMD resting membrane potential at the base of field A, before and after 4-AP in the RNAi-injected cells are missing. The authors should consider including these data and the corresponding statistical analyses in their supplemental materials, similar to Supplemental Figure 3.

• The statistics for Figure 7B are missing. The authors could include this analysis for reader clarity.

• Although it is clear in Figure 7A that a +6 pA depolarizing step to the base of LGMD field A results in a 10mV depolarization (as shown by the scale bar), indicating what the baseline membrane voltages are for the control and 4-AP representative current step traces shown in Figure 7A would further aid interpretation.

• Although the authors state that “Less is known about the role of slowly inactivating dendritic potassium channels besides their involvement in delaying spiking (Storm 1988),” a lot is known about the roles of Kv1 channels in setting the spike threshold, regulating the action potential shape, and limiting the backpropagation of action potentials, among other neural computations.  The authors could expand upon this to further contextualize their findings.

• Shaker channels are known to be abundantly expressed in the axons of central and peripheral neurons, but whether Shaker is expressed in the axonal compartment of LGMD neurons is not clear. The authors could expand upon how a high density of Shaker channels in LGMD axons would be expected to contribute in terms of neural computations.

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they did not use generative AI to come up with new ideas for their review.