Voltage-gated calcium channels act upstream of adenylyl cyclase Ac78C to promote timely initiation of dendrite regeneration

Abstract

Most neurons are not replaced after injury and thus possess robust intrinsic mechanisms for repair after damage. Axon injury triggers a calcium wave, and calcium and cAMP can augment axon regeneration. In comparison to axon regeneration, dendrite regeneration is poorly understood. To test whether calcium and cAMP might also be involved in dendrite injury signaling, we tracked the responses of Drosophila dendritic arborization neurons to laser severing of axons and dendrites. We found that calcium and subsequently cAMP accumulate in the cell body after both dendrite and axon injury. Two voltage-gated calcium channels (VGCCs), L-Type and T-Type, are required for the calcium influx in response to dendrite injury and play a role in rapid initiation of dendrite regeneration. The AC8 family adenylyl cyclase, Ac78C, is required for cAMP production after dendrite injury and timely initiation of regeneration. Injury-induced cAMP production is sensitive to VGCC reduction, placing calcium upstream of cAMP generation. We propose that two VGCCs initiate global calcium influx in response to dendrite injury followed by production of cAMP by Ac78C. This signaling pathway promotes timely initiation of dendrite regrowth several hours after dendrite damage.

Fig 1. Calcium influx during axon and dendrite injury to class IV (ddaC) neurons.

(A) Diagram of location of ddaC neurons in larvae and location of laser cuts of the neurons. We severed either axons or dendrites of ddaC neurons with a UV laser and monitored calcium levels in the soma simultaneously with the GCaMP6f calcium sensor. Example time series of GCaMP6F fluorescence during dendrite (B) and axon (C) injury. Cut site is labeled with a white arrow. Axonal compartment is labeled with a red dashed line, and the dendritic compartments are labeled with blue dashed lines. GCaMP6f fluorescence increased 4–5 fold within 10–15 seconds after injury. Average normalized fluorescence intensities of the cell body are shown (D). The peak fluorescence of each injured neuron is graphed (E), and this is not different between axon and dendrite injuries with a Mann-Whitney test. Error bars for (D) are omitted for clarity, but relevant error information is contained in (E), where bars represent standard deviation.

Fig 2. Voltage-gated calcium channels (VGCCs) are required for injury-induced calcium influx and dendrite regeneration.

We performed GCaMP6f live imaging while knocking down VGCC subunits with RNAi in ddaC neurons (see Reagent table (S2 Table) for specific tester line and RNAi lines). Average GCaMP6f fluorescence is plotted over time for each genotype after dendrite injury (A) and axon injury (B). Peak fluorescence values are plotted in (C). Peak values of Ca-α1D A, Ca-α1T, Ca-β and stj RNAi neurons are significantly lower than control for the dendrite cut but not axon cut condition. (D) Total dendrite regeneration length at 24HPD is shown for each RNAi line used. (E) Example dendrite regeneration images are shown 24h after dendrite removal. All statistics are Kruskal-Wallis one way analysis of variance (ANOVA), with Dunn’s multiple comparisons test. *, p<0.05, **, p<0.01, ***, p<0.001. Error bars in (C) and (D) are standard deviations. Error bars are omitted in (A) and (B) for clarity, but relevant error information is contained in (C).

Fig 3. VGCCs are required for early initiation of dendrite regeneration.

We assayed ddaC dendrite regeneration at 6HPD and 72HPD in control and VGCC RNAi neurons. Example images are shown for Control (A) and VGCC subunit RNAi (B-D) conditions. Green arrows highlight new dendrites and red arrows show remaining degenerating dendrites. (E) The fraction of neurons with at least 1 new dendrite branch at 6HPD is shown. All three RNAi conditions are significantly lower than control with individual Chi-square tests. Total regeneration length and total regenerated branch points at 6HPD are quantified in (F) and (G) and compared with a Kruskal-Wallis one way analysis of variance (ANOVA), with Dunn’s multiple comparisons test. (H) We assayed dendrite regeneration with these same RNAi lines at 72HPD and found little difference between conditions; example images for control and Ca-β RNAi are shown (H and I). Red dashed lines show the approximate area originally occupied by dendrite. Total regeneration length and total regenerated branch points 72HPD are quantified in (J) and (K) and both are compared with Kruskal-Wallis one way analysis of variance (ANOVA), with Dunn’s multiple comparisons test. For all statistics, *, p<0.05, **, p<0.01, ***, p<0.001. Error bars are standard deviations.

Fig 4. Class I (ddaE) neurons require VGCCs for dendrite injury signaling.

We performed dendrite and axon cuts in ddaE neurons expressing GCaMP6f. (A) Diagram of a ddaE neuron. (B) Average GCaMP6f fluorescence intensity in the cell body over time after injury is shown for dendrite (blue) and axon (red) cuts. Peak fluorescence values are plotted in (C) and are not significantly different when compared with a Mann-Whitney test. (D and E) Blue and red dashed lines show dendrites and axons respectively, while white arrows denote location of laser severing. Example time series images are shown for 0, 2, 5, 60, and 120 seconds post cut. (F) The dorsal comb dendrite of ddaE neurons was severed (red arrow) at time 0 and regeneration was assayed by counting added branch points (green arrows) to remaining dendrites 24HPD. (G) Number of added branches at 24HPD is quantified for control, Ca-α1D and Stj RNAi genotypes. Both Ca-α1D and Stj RNAi regenerated significantly fewer branch points than control. The statistical test used was a Kruskal–Wallis one-way analysis of variance (ANOVA) with Dunn’s multiple comparisons test. *, p<0.05, **, p<0.01. All error bars are standard deviations. (H and I) A regeneration assay similar to that in F and G was performed, except that instead of assaying added branches at 24 HPD, the branches were counted at 72 HPD. Only a subset of new branches are indicated with green arrows to avoid obscuring the dendrite. Statistical analysis in I is the same as in G.

Fig 5. Dendrite regeneration screen with adenylyl cyclase (AC) candidates.

(A) Dendrites of ddaC neurons expressing RNAi hairpins to target candidate ACs were severed at time 0. Maximum regeneration diameter was measured 24HPD. (B) Representative images are shown for control, Ac78C A and Rut B RNAi lines. **, p < .01; ***, p < .001 with a Kruskal–Wallis one-way analysis of variance (ANOVA), each condition compared with the control with Dunn’s multiple comparisons test. (C) Morphology of uninjured ddaC neurons is shown for control, Ac78C A and Rut B RNAi lines. Dendrites of Rut B RNAi neurons did not reach the edge of the territory, had sparse coverage and extra branching near the soma. Error bars in (A) are standard deviations.

Fig 6. cAMP is produced after axon and dendrite injury and Ac78C is required after dendrite injury.

cAMP levels were measured after injury of ddaC neurons for five minutes with the Flamindo2 biosensor. (A) Example images of time points within the 5-minute videos are shown for control sham, dendrite, and axon cuts as well as Ac78C RNAi dendrite cuts. Percent loss of fluorescence is shown for each image in the top right corners. (B) Average fluorescence intensity over time is plotted for axon, dendrite, and sham cut conditions. Note the deep trough and subsequent fluorescence recovery after axon and dendrite cuts (red and blue traces respectively) compared to a sham injury (gray trace). (C) Neurons expressing Ac78C RNAi were tested for dynamic cAMP levels after injury. Note that the dendrite cut condition closely mirrors the sham cut condition, but axon cut response is largely unaltered. (D) Minimum fluorescence values are shown for sham, dendrite, and axon cut conditions, Kruskal Wallis with Dunns multiple comparisons was used to compare control sham to control dendrite cut and control sham to control axon cut. To compare control dendrite cut to Ac78C dendrite cut and control axon cut to Ac78C axon a Mann Whitney test was used. Error bars for (B and C) are omitted for clarity, but relevant error information is contained in (D), which are standard deviations.

Fig 7. Ac78C is required for early initiation of dendrite regeneration.

Dendrite regeneration of ddaC neurons was assayed with two independent Ac78C RNAi hairpins at 6, 24, and 72HPD. Example images for each time point are shown in (C), (A) and (F) respectively. (B) Total regenerated dendrite length was quantified for control and Ac78C RNAi conditions at 24HPD from the same images used to calculate diameter in Fig 6. Fraction regenerating and total regenerated dendrite length were quantified at 6HPD in (D) and (E). Total new branch points and total regenerated dendrite length were quantified at 72HPD in (G) and (H). Note that control data in (B) is the same set as displayed in Fig 2, and control data in (D) through (H) comprises the control data for 6HPD and 72HPD in Fig 3 and an additional set of control data; the additional control data was obtained at the time of the second Ac78C RNAi data (Ac78C B) as it was done later than the other 6h and 72h experiments. Statistics in (D) is a Chi-square test (all ns) and tests in all other graphs are Kruskal-Wallis ANOVAs with Dunn’s multiple comparisons test. *, p<0.05, **, p<0.01, ***, p<0.001. All error bars are standard deviations.

Fig 8. VGCCs are required for cAMP production after dendrite but not axon injury.

Flamindo2 fluorescence was monitored in ddaC neurons expressing RNAi hairpins to target VGCC subunits. (A) Example time series for control dendrite cut, and both dendrite and axon cuts with Ca-α1D RNAi, are shown. Average fluorescence intensity over 5 minutes is graphed for dendrite cut in (B) and axon cuts in (C). Note that control dendrite and axon cuts (black traces in (B) and (C) are the same data sets displayed in blue and red, respectively, in Fig 5B and 5C. (D) Minimum fluorescence value is displayed for each genotype and compared to its respective control with Kruskal-Wallis one way analysis of variance (ANOVA), with Dunn’s multiple comparisons test, **, p<0.01. Error bars in (D) are standard deviations. Error bars are omitted in (B) and (C) for clarity, but relevant error information is contained in (D).