Centrally expressed Cav3.2 T-type calcium channel is critical for the initiation and maintenance of neuropathic pain

Reviewer #1 (Recommendations for the authors):

The recommendation is to provide details of the statistics used and rationale for analysis of multiple observations in single animals. It is not clear if they are treated as independent observations or corrected for in the analyses.

We considered each unit as an independent observation. We did so since the number of recorded PV+-units per animal (identified with the PINP method) was small and varied greatly between animals, from 2 to 6 units. We are not aware of statistical methods using a nested design for multiple cells in animals that could be used in such condition.

Since the measured variables did not follow normal distributions, we performed unpaired comparison with the Wilcoxon sum rank test. This is now stated in the section ‘quantification and statistical analysis’ (page 22, line 680). P-values are now included in the figures and the result section when appropriate.

Reviewer #2 (Recommendations for the authors):

1. Figure 1, does SNI alter the expression of Cav3.2 and the overlap between Cav3.2 and PV in APT?

We did not perform a systematic comparison of the Cav3.2 and PV colocalization between control and SNI mice but we did not observe a change in the number of GFP-expressing neurons per slice after SNI.

2. Figure 2, please clarify whether light-evoked bursts are included when calculating the mean firing rate? If yes, what's the mean firing rate in the absence of evoked bursts, and is it different between naïve and SNI mice? Does SNI alter mean firing and/or evoked bursts in the ipsilateral APT? Sham mice should be used as the control of SNI.

Light-evoked bursts were not included when calculating the mean firing rate. We apologize for this omission in the description of the analysis protocol. This point is now specified in the Materials and methods section (page 22, line 677).

We did not record in the ipsilateral APT.

We fully agree that we should have used sham animals in first instance and we accordingly performed a new set of experiments. Now, all control animals underwent sham surgery. Statistically significant differences between sham operated and SNI animals were observed similarly to what was previously reported for the naïve/SNI comparison (see new Figure 2 and Results section page 3, lines 98-116).

3. Figure 3, are the amplitude of Ni-sensitive current, the number of spikes in rebound burst and the spike-current relationship altered by SNI? Are they attenuated in APT-KO mice after sham/SNI?

We could not directly assess the impact of SNI on Cav3.2 current amplitude for technical reasons. Indeed, at the beginning of our study, we tried to record low-threshold calcium currents in APT neurons. Although we choose to record these currents in very young mice (P12 -14) to minimize the space clamp problem, it was clearly impossible to clamp the membrane potential even at this age when the dendritic arborization is not fully mature. Many experimental and theoretical studies have shown that in non-spherical cells, voltage-dependent currents, such as Cav3, are severely distorted due to the lack of space clamp (see for example Bar-Yehuda and Korngreen J. neurophysiol 2008). Recording Ni²⁺ sensitive currents in sham versus SNI mice, approximately 100 days old, cannot be performed with the minimal accuracy required for rigorous comparison.

Therefore, we performed a new set of experiments by recording rebound burst responses in PV+ APT neurons in sham and SNI mice. By comparing the rebound bursts, we observed that the distribution of the maximum number of spikes in the rebound bursts is shifted to larger values in SNI compared to sham mice. To evaluate whether Cav3.2 channels are indeed responsible for this increase in bursts, we applied 100µM Ni²⁺ and observed that this difference in distribution disappeared. These results, which clearly show a Cav3.2 related enhance rebound burst in APT neurons after SNI are now presented page 4, line 150-158, in the new figure 4 (panel D) and outlined as one of the main result of the present study at the end of the introduction (page 2, line 71).

We could not perform similar experiments in APT-KO mice. Given that Cav3.2-expressing neurons represent 20% of the APT neuron population, comparing bursting activity in KI and APT-KO mice after sham/SNI would require identification of this specific subpopulation in living tissue, which is clearly not possible in APT-KO mice. As far as KI mice are concerned, although a GFP tag is attached to the Cav3.2 channels in the Cav3.2^eGFP-flox knock-in, its fluorescence is not detectable without antibody preventing direct identification in living tissue (François et al., 2015).

4. Figure 4B-C, in SNI mice that receive AAV-Cre, does blocking peripheral/spinal Cav3.2 further attenuate/normalize mechanical allodynia?

At the peripheral level, deletion of Cav3.2 in a subset of sensory neurons altered light-touch perception and noxious mechanical, cold and chemical sensation (François et al., Cell Reports, 2015). In the SNI model, such deletion also induced a marked reduction in the mechanical allodynia similar to that reported here. To date, no data are available regarding the effect of a specific Cav3.2 KO at the spinal level. Indeed, the multiple sites where Cav3.2 seems to play a crucial role in chronic pain open very interesting question about the respective roles of centrally versus peripherally expressed Cav3.2 channels in allodynia. However, such studies would require a large number of experimental investigations that are well beyond the scope of the present work.