Dendrite regeneration in C. elegans is controlled by the RAC GTPase CED-10 and the RhoGEF TIAM-1

Neurons are vulnerable to physical insults, which compromise the integrity of both dendrites and axons. Although several molecular pathways of axon regeneration are identified, our knowledge of dendrite regeneration is limited. To understand the mechanisms of dendrite regeneration, we used the PVD neurons in C. elegans with stereotyped branched dendrites. Using femtosecond laser, we severed the primary dendrites and axon of this neuron. After severing the primary dendrites near the cell body, we observed sprouting of new branches from the proximal site within 6 hours, which regrew further with time in an unstereotyped manner. This was accompanied by reconnection between the proximal and distal dendrites, and fusion among the higher-order branches as reported before. We quantified the regeneration pattern into three aspects–territory length, number of branches, and fusion phenomena. Axonal injury causes a retraction of the severed end followed by a Dual leucine zipper kinase-1 (DLK-1) dependent regrowth from the severed end. We tested the roles of the major axon regeneration signalling hubs such as DLK-1-RPM-1, cAMP elevation, let-7 miRNA, AKT-1, Phosphatidylserine (PS) exposure/PS in dendrite regeneration. We found that neither dendrite regrowth nor fusion was affected by the axon injury pathway molecules. Surprisingly, we found that the RAC GTPase, CED-10 and its upstream GEF, TIAM-1 play a cell-autonomous role in dendrite regeneration. Additionally, the function of CED-10 in epidermal cell is critical for post-dendrotomy fusion phenomena. This work describes a novel regulatory mechanism of dendrite regeneration and provides a framework for understanding the cellular mechanism of dendrite regeneration using PVD neuron as a model system.


Major Revisions
1. PVD dendritic branching is drastically reduced in tiam-1 mutants which show limited secondary and tertiary branch outgrowth. It thus seems plausible that the overall reduction in PVD dendritic branching in tiam-1 mutants could also hinder regeneration and that this effect would also be observed for mutants in other genes (e.g., hpo-30, lect-2, act-4, dma-1) that drive PVD branching. The authors need to test at least one additional PVD dendritic branching mutant to rule out this possibility.

Response:
We agree with Reviewer #1 on the above point that the effect on dendrite regeneration in the tiam-1 mutant could be influenced by the dendrite branching defect in this mutant. Thus we checked dendrite regeneration phenotype in a few mutants with developmental branching defect in PVD. For example, the hpo-30(ok2047) mutant showed a developmental defect in PVD dendrites (Tang et al 2019, eLIFE). In hpo-30 mutant, the extent of regeneration (territory length) was reduced as similar to tiam-1 mutant (13<n<15, N>2, unpaired student t test, considering p<0.05 * , 0.01 ** , 0.001 *** ).
hpo-30 acts upstream to tiam-1 in developmental branching by controlling actin dynamics (Tang et al 2019, eLIFE). It is possible that regeneration of PVD dendrites may utilize a similar mechanism for the new regrowth thus dependent on hpo-30 upstream to tiam-1.
Additionally, we tested the mec-3 mutant, in which only primary dendrite are present ( Figure S4A-B,) (Smith et al 2013 Neuron). To our surprise, unlike the hpo-30 dendrite arborization mutant, there was a robust regrowth response from the primary dendrite following dendrotomy. Both the territory length and reconnection frequency with the distal dendrite were similar to the wild type control ( Figure S4 E-F). Though the number of regrowing branches was reduced, the overall regenerative capacity of the mec-3 mutant neuron was comparable to the wildtype ( Fig S4F). Therefore, this addresses the point that the mutants with diminished higher-order branches are not necessarily compromised in their potential for dendrite regeneration. Moreover, to find the specific role of tiam-1 in dendrite regeneration, we expressed the version of TIAM-1 with compromised GEF activity (TIAM-1 T548F). This mutant version could rescue the developmental branching phenotype ( Figure S4A-B) as seen before (Tang et al., eLIFE, 2019), but failed to rescue the dendrite regeneration phenotype in the tiam-1 mutant. This indicated a specific role of TIAM-1 in dendrite regeneration. Please see the Line 344-357, Page 16 in the revised manuscript.

If TIAM-1 GEF activity is required for activating CED-10-dependent regeneration, then a tiam-1 point mutation that specifically eliminates TIAM-1 GEF activity should impair dendritic regeneration (Demarco et al., 2012). This question is important because a recent paper showed that TIAM-GEF activity is apparently not required for PVD dendritic branching (Tang et al., eLIFE, 2019).
Response: We generated the GEF dead tiam-1 transgene (T548F) in tiam-1(0 mutant background, which rescues the dendrite arborization defect in the tiam-1(0) mutant ( Figure S4A-B) as seen before (Tang et al., eLIFE, 2019). However, the regeneration defects including the reduced territory length and the number of branches were not rescued by this transgene in tiam-1(0) mutant. This indicated that the GEF activity of TIAM-1 is required for dendrite regeneration ( Figure 6A-B). This also strengthened the hypothesis that the impaired dendrite regeneration in the tiam-1 mutant is not a secondary effect of the aberrant dendrite arborization in this mutant. Please see the Line 350-357, Page 16 in the revised manuscript.
3. The authors report that severed PVD dendrites regrow and ultimately fuse with each other to restore a contiguous dendritic arbor. The evidence of fusion is limited to the observation that the tips of apposing regenerated, GFP-labeled dendrites appear continuous in the light microscope. This observation does not rule out the alternative explanation that the regenerated dendrites are overlapping each other or touching but not actually fused. I'm not requiring an experiment to distinguish between these possibilities since this does not seem be a convention in the field but the authors need to address this caveat at the very least in the manuscript. This question is actually quite significant since the long term goal of this work on regeneration in model organisms is to discover pathways that can restore function to injured circuits.

Response:
We concur with the reviewer that it is difficult to determine correctly whether the proximal and distal parts of the injured dendrites fuse during regeneration. For this reason, we described the contacts between the proximal and distal primary dendrites as 'reconnection' events. To address whether the two tips actually contact or appeared to be connected in the Z-projected image, we represented the confocal planes in the Z-projected image as depth-coded images. We made sure that the contact between the proximal and distal dendrites are at the same optical depth while counting the event as 'reconnection'. This way, in a false reconnection event, the proximal and distal end would  (Fig S1G). By this method, 60% of events were counted as 'reconnection events' as opposed to 80% counted when the projected image was not depthcoded (Fig S1H).
We have now revised the quantification of the reconnection events based on their respective depthcoded images. Another correlation that has been considered into account is that in the cases of false reconnection events, the distal part of the dendrites tends to degenerate ( Figure S1I). Similarly, the menorah-menorah fusion events ( Figure S1H) were also judged in this procedure. Please see the Line 128-138, Page 6-7 in the revised manuscript. Also please see the method section "Dendrite Regeneration Analysis and Quantification" line 512-522, page 24 in the revised manuscript.

Minor revisions
1. Minor grammatical and stylistic errors are scattered throughout the text. The use of the article "the" is problematic in several instances (e.g., "triggers elevation in the Cyclic Adenosine Monophosphate (cAMP)…" should be "triggers elevation of Cyclic…"

Response:
We have incorporated the changes that you have suggested and reduced the usage of "the" wherever inappropriate. We have revised our manuscript text to avoid the occurrence of any grammatical errors to the best of our abilities. Response: We observe a significant amount of autofluorescence punctae from the gut that are visible in both the green and red fluorescence channels. We have revised this experiment with a strategy to remove the autofluorescence puncta in the background. We have provided the newly acquired images with background correction using auto-fluorescence acquired in the near UV channel. Please see the revised Figure 3B.

National Brain Research Center
Reviewer #2: In this manuscript Brar et al. perform a basic characterization of dendrite regeneration in C. elegans PVD neurons, compare dendrite and axon regeneration in the same cell, and identify two genes required for dendrite regeneration. This work provides an important foundation for understanding mechanisms that allow neurons to respond to dendrite injury. Dendrite regeneration (with the exception of fusion, which is specific to C. elegans) has been studied almost exclusively in Drosophila. Having another model system in which to study this process is a substantial advance. As in Drosophila, none of the core axon regeneration machinery regulates dendrite regeneration, and these are compared in the same cell. Moreover, the authors find two regulators of dendrite regeneration. This is a nice foundational story that provides good footing to use C. elegans to investigate dendrite regeneration. There are just a few points that would strengthen the manuscript further: Response: Thanks for appreciating the strength of this study. We have addressed all the points below.
1. Axon regeneration in PVD looks quite subtle; growth difficult to see in control image in figure  3D. Perhaps a different image might help? It looks like the axon does not regrow to reach its former length; is this the case? If so, it would be good to discuss what this might mean for function.