Programmed Cell Death Modifies Neural Circuits and Tunes Intrinsic Behavior

  1. Alison Kochersberger
  2. Dongyeop Lee
  3. Mohammad Mahdi Torkashvand
  4. Sandeep Kumar
  5. Saba Baskoylu
  6. Titas Sengupta
  7. Noelle Koonce
  8. Chloe E. Emerson
  9. Nandan V. Patel
  10. Daniel Colón-Ramos
  11. Steven Flavell
  12. Andrew M. Leifer
  13. Vivek Venkatachalam
  14. H. Robert Horvitz
  15. Marc Hammarlund

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

Programmed cell death (PCD) is a common feature of animal development. During development of the C. elegans hermaphrodite, programmed cell death eliminates 131 cells in stereotyped positions in the cell lineage, mostly in neuronal lineages. Blocking cell death results in supernumerary “undead” neurons. We find that undead neurons can be wired into circuits, can display activity, and can modify specific behaviors. The two undead RIM-like neurons participate in the RIM-containing circuit that computes movement. The presence of these two extra neurons results in animals that initiate fewer reversals and lengthens the duration of those reversals that do occur. We describe additional behavioral alterations of cell-death mutants, including in locomotory turning angle and pharyngeal pumping. These findings indicate that physiological or evolutionary variations in PCD might reveal latent neuronal elements that the nervous system can incorporate to modify nervous system function and animal behavior.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Programmed cell death is prominent in developing nervous systems across evolution, but its function remains obscure. Recent work suggests that it might impact behavior, but an examination of its effects on behavior and underlying neuronal circuits in intact organisms has not been determined. In this manuscript we report that programmed cell death sculpts the developing nervous system and shapes innate behavior. Using synaptic labeling, *in vivo *calcium imaging, targeted rescue of programmed cell death, and automated high-resolution analysis of cell death mutants, we find that loss of programmed cell death alters animal behavior. These findings reveal that neuronal cell death during development provides a reservoir of fates and circuit connections that could be accessed on evolutionary time scales to modify innate behavioral programs. Our manuscript thus answers one of the major outstanding questions in developmental neuroscience—why programmed cell death is so prevalent—by identifying consequences for brain function at the subcellular, cellular, circuit, and behavioral level. This study will be of interest to those interested in evolution of the nervous system and behavior, developmental biology, and neural circuit development.

    We thank the reviewers for their careful attention to the manuscript. Both reviewers were enthusiastic about the work. Here we address their suggestions. As noted below, we have already addressed most of their points, and we discuss in detail the remaining point—whether it is possible to perform experiments for a more specific targeting of the undead RIM cell death event to provide additional evidence for its role in altering reversal behavior.

    2. Description of the planned revisions

    *Reviewer 1: “1. The argument that that differences in reversal behavior are likely attributable to the difference in RIM neuron numbers in the ced-3 rescue studies is very plausible. Nonethless, there remains the possibility that for some reason in animals with 4 RIMs there may be a more global effect on the fate of cells slated to die, unrelated to the number of RIMs. I think there are two ways to test this. (1) quantify the behavior in 2- vs. 4- RIM neurons in animals also containing a marker for other undead neurons, and see if there is any correlation between 4 RIMs and survival of unrelated neurons (but preferably reasonably closely related by lineage- in case that's the issue). (2) Since the authors are able to distinguish the undead cells, can they perform laser ablations on these cells and assess whether behavior is restored to normal values?” *

    • *We agree that this point is already very plausible. We also appreciate the reviewer’s suggestions on how to extend this conclusion.

    Regarding suggestion (1): Unfortunately there is not a reliable marker for undead neurons (although a current project in the lab is indeed to develop one). However, we note that the undead RIM sister cells adopt a RIM neuron fate in 96% of *ced-3 *mutants, while with other undead cells investigated neuron fate adoption ranged from 59% (ASEL) to 77% (ASER). This suggests that the undead RIM fate adoption is not strongly correlated with the fates of other undead cells.

    Regarding suggestion (2): We attempted to perform laser ablation of undead RIM neurons in *ced-3 *mutants, but we could not overcome the technical hurdles (despite our lab’s expertise in laser axotomy). We found that we could not reliably remove both undead RIMs without damaging the wildtype RIM that is in close proximity, especially in the quantities of animals necessary for behavioral experiments.

    As an alternative, we plan to perform more targeted experiments to manipulate cell death in the undead RIM to address the points raised by both reviewers. Our goal is to generate two strains. In one, programmed cell death is prevented specifically in the RIM neurons in wild type animals. We hope to achieve this by either transgenic expression of a gain-of-function mutation of *ced-9, *or else by RIM-specific RNAi against *egl-1, ced-3 *or ced-4. To do this we will use the RIM promoter tdc-1, which is confined to RIM and RIC. The second strain will allow cell death to occur only in RIM (and RIC) in animals that otherwise have no cell death. Here, we will drive wild-type *ced-3 *or *ced-4 *under the *tdc-1 *promoter in the corresponding mutant background.

    We note 2 caveats for both of these approaches: 1) RIC also has an undead sister; 2) Most probably, the *tdc-1 *promoter will not be active in time to block cell death. Caveat #2 is actually the reason why we did not do these experiments initially (instead we used the most specific promoter we could find that is expressed early in the RIM lineage, before RIM is born).

    However, we agree that if successful these experiments would complement the existing experiments, and we will build all these strains.

    Reviewer 2: “Mosaic rescue of RIM via stochastic loss of a rescue array helped demonstrate the contribution RIMu have to the locomotor phenotype. As the authors emphasise these animals have many other undead cells (outside of the reverse network). A conditional rescue of only the RIMu would greatly improve the strength of the claims made. Would a conditional RIM egl-1 knockdown (via RNAi) be possible to selectively inhibit apoptosis in those neurons. This experiment should be considered OPTIONAL. It may be that such specific promoters do not allow for egl-1 RNAi to function at the right time to rescue death.”

    • *We appreciate the reviewer’s suggestion. As stated above, we are working to perform an expanded version of these exact experiments, as well as their converse. However, as the reviewer notes, it is very possible that the timing of expression will prevent these approaches from working (Caveat #2 above).

    Reviewer 2: There is a slight issue with interpretation of the data with the mosaic GLR-1::tagRFP Fig 2M which reveals the postsynaptic compartment of one RIM even though there are two present. There seems to be no obvious apposition between pre/post and they somewhat seem to be floating in space. Why is this the case? One would have imagined that the structures in Fig 2L would be tiled composites of both AIB & RIM pre and postsynaptic elements coalescing. Can the authors provide an alternative explanation for this phenotype. Nevertheless, the data on Fig 2L seems solid.. that is animals with extra undead RIM cells have additional cell-type specific synaptic terminals

    We have selected a different micrograph that is more representative of the RIM post-synapses in *ced-3 *mutants. In this animal, the array labeling the post-synapses in RIM has been lost from one of the two RIM neurons, making it easier to discern that the post-synapses are apposed to the AIM pre-synaptic marker (Fig 1M).

    Reviewer 2: Clarity should be improved around the use of 'expected number' in figure 1. The description of the metric 'The 'expected number' is defined as the number of neurons of the type present in wild-type animals, plus the number of lineage-proximate undead cells.' suggests that expected (blue) regions of pie charts represents lineages with expected sum total of wt and extra undead cells. However, in reference to panel H 'The wild-type animal has two RIM neurons, and the ced-3(n717) animal has two additional RIMlike cells and is counted as contributing to the orange "more than expected" sector in panel (A)' it is said that the animals with 2 WT accompanied by each undead sister contributes to more than expected (orange) region. These appear inconsistent. Can you qualify?

    We thank the reviewer for this point and have added a schematic to clarify the quantification of undead cell fates (Fig. 1).

    Reviewer 2: Specific observations shown in supplemental data SI-L despite being cited in the text is not explained or formally referenced. The details of these panels should either be briefly explained/their inclusion qualified in the text or simply remove from the figure

    We have added reference to these figures in the main text “Undead cells are even capable of producing complex morphology, such as the highly branched dendrites of the PVD neurons (Figure S1I-L).” (p. 3)

    Reviewer 2: The dual image photomicrographs could be in green/ magenta or red/cyan to make colourblind friendly.

    We have updated micrograph colors to be colorblind friendly (Fig 1K-M, S1L).

    Reviewer 2: Do the authors have data with the pRIMtagRFP egl-nucGFP. If they do it would be useful to show it.

    We have added a micrograph of egl-1::GFP and RIM labeled using NeuroPAL (Fig. S2A).

    Reviewer 1: 2. The authors speculate, if I understand correctly, that the mechanism by which reversal frequencies are decreased in 4 RIM animals may be that the reversal state is stabilized, resulting in longer reversals and consequently fewer reversal events. This is a nice model that is testable. The authors could, for example, examine the connections of RIM neurons to the AVA neuron, a main command interneuron for reversal initiation, and assess whether there are indeed more such synapses. Furthermore, the authors can assess whether the frequency of AVA firing is decreased. Of course, there are other plausible mechanisms involving connectivity of other neurons onto AVA which could explain the phenomenon. The authors may wish to add a comment regarding this in the discussion.

    We thank the reviewer for this suggestion. There are multiple postsynaptic receptors expressed in AVA for RIM neurotransmitters and the contribution of each to reversal behavior is still being debated, making it challenging to dissect the contribution of each of these to the effects on reversal behavior mediated by the undead RIM. Given this, we believe that addressing this point experimentally is beyond the scope of this paper. We have added a sentence in the discussion commenting on this as a future direction for this work “The mechanism of the downstream circuit mediating the effects of the undead RIM could be determined through quantification of AVA postsynaptic receptors and examining reversal behavior of cell death mutants with knockouts of AVA receptors.”

  2. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    Summary:

    The blockade of programmed cell death (PCD) results in the generation of supernumerary neurons from the rescue of normally dying progeny within a lineage. Many of these are the product of a lineage branch that produce a single surviving neuronal daughter.

    The authors show that for many such divisions the rescued sibling expresses fate markers associated with their normally surviving sibling. The authors use the RIM neurons and their role in reversal behaviour as a model. By generating supernumerary RIM interneurons, they show that these undead neurons correlate with an increase in synaptic terminals, with their normal presynaptic partners (AIBs), and that these same supernumerary neurons have an activity pattern preceding reversal events.

    Using mosaic experiments, based around the stochastic differences in the segregation of an extrachromosomal array, the authors are able to demonstrate that a strong component to alterations in locomotor behaviour is likely due to the supernumerary undead RIMu neurons.

    To address the impact of blockade of PCD more broadly the authors assayed 4 independent PCD blocking alleles (2x ced-3, 2x ced-4) and show that the rescue of many neurons alters normal turning dynamics associated with foraging and free feeding behavioural states.

    Interestingly, subtle differences in the behaviours of ced-3 mutant animals and ced-4 mutant animals may also hint at the broader significance of these genes beyond simply controlling the on/off PCD in these lineages. One possible issue with work published in this paradigm is that by generating undead neurons, by removing caspase activity, the intervention may also impact non-apoptotic caspase function.

    The study presented does not pose any issues relating to reproducibility or associated statistical analysis.

    Major comments:

    • Mosaic rescue of RIM via stochastic loss of a rescue array helped demonstrate the contribution RIMu have to the locomotor phenotype. As the authors emphasise these animals have many other undead cells (outside of the reverse network). A conditional rescue of only the RIMu would greatly improve the strength of the claims made. Would a conditional RIM egl-1 knockdown (via RNAi) be possible to selectively inhibit apoptosis in those neurons. This experiment should be considered OPTIONAL. It may be that such specific promoters do not allow for egl-1 RNAi to function at the right time to rescue death.
    • The changes in the RIM/AIB synaptic organisation and the correlation of observed RIM/RIMu activity coincident with reverse locomotor bouts strongly supports the assertion that these two features are causally linked.
    • One interpretation is that when the additional undead RIM neuron is present the cell-type specific connectivities are recapitulated i.e. many of the pre and post-synaptic elements of RIM/AIB look apposed see Fig 2L.
    • There is a slight issue with interpretation of the data with the mosaic GLR-1::tagRFP Fig 2M which reveals the postsynaptic compartment of one RIM even though there are two present. There seems to be no obvious apposition between pre/post and they somewhat seem to be floating in space. Why is this the case? One would have imagined that the structures in Fig 2L would be tiled composites of both AIB & RIM pre and postsynaptic elements coalescing. Can the authors provide an alternative explanation for this phenotype. Nevertheless, the data on Fig 2L seems solid.. that is animals with extra undead RIM cells have additional cell-type specific synaptic terminals.

    Minor comments:

    • Clarity should be improved around the use of 'expected number' in figure 1. The description of the metric 'The 'expected number' is defined as the number of neurons of the type present in wild-type animals, plus the number of lineage-proximate undead cells.' suggests that expected (blue) regions of pie charts represents lineages with expected sum total of wt and extra undead cells. However, in reference to panel H 'The wild-type animal has two RIM neurons, and the ced-3(n717) animal has two additional RIMlike cells and is counted as contributing to the orange "more than expected" sector in panel (A)' it is said that the animals with 2 WT accompanied by each undead sister contributes to more than expected (orange) region. These appear inconsistent. Can you qualify?
    • Specific observations shown in supplemental data SI-L despite being cited in the text is not explained or formally referenced. The details of these panels should either be briefly explained/their inclusion qualified in the text or simply remove from the figure
    • The dual image photomicrographs could be in green/ magenta or red/cyan to make colourblind friendly.
    • Do the authors have data with the pRIMtagRFP egl-nucGFP. If they do it would be useful to show it.

    Significance

    The work presented here complements similar studies performed on insects, illustrating that this biological motif holds true outside of that class. Showing this is a more general feature within other taxa broadens the appeal and will be of great interest to those in developmental biology, neural circuit development/function and evolutionary biology.

    Although the work does not illustrate a completely novel finding, it is rigorous and well-conceived, adding support to previous studies in the field and is an important jumping off point for future studies. The authors present compelling independent support that undead developmentally 'doomed' neurons retain the ability to differentiate, show molecular hallmarks of sibling fate, can integrate within networks and function.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    This is an interesting paper describing the neural circuit and behavioral consequences of blocking programmed cell death in C. elegans with a mutation in the apoptotic caspase ced-3. The authors survey several reporters for neurons whose sister cells normally die, and observe that undead cells can express markers of the normally living sister cell. They also show that ced-3 mutants exhibit a large variety of behavioral defects. They interrogate the effects of inappropriate survival of two RIM sister neurons in animals carrying a complement of 4 instead of 2 RIMs. The authors demonstrate that the presynaptic neuron AIB makes synapses onto the undead RIM neurons and show GCaMP activity in these neurons that correlates with reversal behavior, a normal function of RIM neurons. Animals with 4 RIM neurons have a reduced number of reversal events compared to wild type animals, suggesting circuit defects. Importantly, restoring ced-3 expression only in the RIM lineage partially restores reversal behavior frequency. The authors conclude that undead neurons interfere with the normal function of the nervous system by making aberrant connections that interfere with circuit activity.

    This paper is beautifully written. The logic is crystal clear, and the experiments are appropriate and rigorously executed. The conclusions are generally appropriate. I have a couple of comments that may be useful for the authors to consider:

    1. The argument that that differences in reversal behavior are likely attributable to the difference in RIM neuron numbers in the ced-3 rescue studies is very plausible. Nonethless, there remains the possibility that for some reason in animals with 4 RIMs there may be a more global effect on the fate of cells slated to die, unrelated to the number of RIMs. I think there are two ways to test this. (1) quantify the behavior in 2- vs. 4- RIM neurons in animals also containing a marker for other undead neurons, and see if there is any correlation between 4 RIMs and survival of unrelated neurons (but preferably reasonably closely related by lineage- in case that's the issue). (2) Since the authors are able to distinguish the undead cells, can they perform laser ablations on these cells and assess whether behavior is restored to normal values?
    2. The authors speculate, if I understand correctly, that the mechanism by which reversal frequencies are decreased in 4 RIM animals may be that the reversal state is stabilized, resulting in longer reversals and consequently fewer reversal events. This is a nice model that is testable. The authors could, for example, examine the connections of RIM neurons to the AVA neuron, a main command interneuron for reversal initiation, and assess whether there are indeed more such synapses. Furthermore, the authors can assess whether the frequency of AVA firing is decreased. Of course, there are other plausible mechanisms involving connectivity of other neurons onto AVA which could explain the phenomenon. The authors may wish to add a comment regarding this in the discussion.

    Significance

    This is an interesting paper describing the neural circuit and behavioral consequences of blocking programmed cell death in C. elegans with a mutation in the apoptotic caspase ced-3. The authors survey several reporters for neurons whose sister cells normally die, and observe that undead cells can express markers of the normally living sister cell. They also show that ced-3 mutants exhibit a large variety of behavioral defects. They interrogate the effects of inappropriate survival of two RIM sister neurons in animals carrying a complement of 4 instead of 2 RIMs. The authors demonstrate that the presynaptic neuron AIB makes synapses onto the undead RIM neurons and show GCaMP activity in these neurons that correlates with reversal behavior, a normal function of RIM neurons. Animals with 4 RIM neurons have a reduced number of reversal events compared to wild type animals, suggesting circuit defects. Importantly, restoring ced-3 expression only in the RIM lineage partially restores reversal behavior frequency. The authors conclude that undead neurons interfere with the normal function of the nervous system by making aberrant connections that interfere with circuit activity.

    The paper should be of interest to researchers studying neural circuit assembly and function, programmed cell death, behavior, and evolution of behavior and the nervous system.

  4. Version published to 10.1101/2023.09.11.557249 on bioRxiv