Veratridine produces distinct calcium response profiles in mouse Dorsal Root Ganglia neurons

Nociceptors are a subpopulation of dorsal root ganglia (DRG) neurons that detect noxious stimuli and signal pain. Veratridine (VTD) is a voltage-gated sodium channel (VGSC) modifier that is used as an “agonist” in functional screens for VGSC blockers. However, there is very little information on VTD response profiles in DRG neurons and how they relate to neuronal subtypes. Here we characterised VTD-induced calcium responses in cultured mouse DRG neurons. Our data shows that the heterogeneity of VTD responses reflects distinct subpopulations of sensory neurons. About 70% of DRG neurons respond to 30–100 μM VTD. We classified VTD responses into four profiles based upon their response shape. VTD response profiles differed in their frequency of occurrence and correlated with neuronal size. Furthermore, VTD response profiles correlated with responses to the algesic markers capsaicin, AITC and α, β-methylene ATP. Since VTD response profiles integrate the action of several classes of ion channels and exchangers, they could act as functional “reporters” for the constellation of ion channels/exchangers expressed in each sensory neuron. Therefore our findings are relevant to studies and screens using VTD to activate DRG neurons.


Veratridine produces four distinct calcium response profiles that occur at different frequencies.
We classified VTD responses into four distinct profiles, Fig. 2A. Three profiles were characterised by a single peak with a rapid rise in intracellular calcium [Ca]i levels but differed in their decay rate. The Slow Decay (SD) profile returned to baseline over 15-40 min following VTD washout, Fig. 2A (blue). In the Intermediate Decay (ID) profile, calcium responses returned to baseline within 2-5 min, Fig. 2A (purple). The Rapid Decay (RD) profile was characterised by a transient [Ca]i peak which returned to baseline during the application period, Fig. 2A (green). The fourth profile was characterised by a multi-peak response which we thus called the Oscillatory profile (OS), Fig. 2A (red). In some OS neurons the response ended prior to the end of VTD application while in others it continued to oscillate during the wash. The four VTD response profiles varied in their frequency of occurrence. The OS profile was the most frequent accounting for 48% of all VTD-responsive neurons, followed by the SD profile (24%) and equal frequencies for the ID and RD profiles (13% each), Fig. 2B. Furthermore, the four profiles differed in their onset, calculated as the latency to peak from the start of VTD application. The RD profile had the longest latency which was twice that of the other three profiles, Fig. 2C. The onset difference was inherent to neurons and not due to variation in the rate of perfusion since all neurons responded with the same latency to high potassium. A very small number of VTD-responsive neurons had a "ramp" like profile characterised by a very slow rise and decay of [Ca]i levels (not shown). These neurons were classified as "undefined" in further experiments and were not included in our characterisation because they were too infrequent (less than 1%) to include in our analysis and draw any conclusions about. The OS profile is the most frequent (48 ± 4%), followed by the SD profile (24.4 ± 4%). The RD (13.4 ± 4%) and ID (12.7 ± 2%) profiles have similar frequencies. (C) Mean response latency for the four VTD response profiles (N = 3 mice, 143 cells). Only neurons that have the same onset to KCl were included in this analysis to exclude differences due to rate of perfusion. The RD (115.4 ± 15 s) profile has the longest latency while the SD (34.2 ± 5 s), ID (55.8 ± 13 s) and OS (43.5 ± 4 s) profiles have similar latencies. Data shown are mean ± SEM. One-way analysis of variance with Tukey's post-test, ***P < 0.001 and ****P < 0.0001.
Scientific RepoRts | 7:45221 | DOI: 10.1038/srep45221 The Veratridine SD response profile is enriched in large diameter neurons. DRG neurons are classified according to their soma size into small, medium and large diameter neurons [16][17][18] . Soma size of DRG neurons correlates with functional modalities. Small and medium sized neurons tend to be nociceptors and innocuous thermoreceptors, while large diameter neurons are more likely to be proprioceptors and low threshold mechanoreceptors 19,20 . Neuronal soma size in our DRG cultures ranged from 10 to 45 μ m. Analysis of soma size and VTD response profiles showed that although the SD profile occurs in all neuronal sizes (mean 25 μ m), it was the most prevalent profile in neurons larger than 30 μ m in diameter (82% of these neurons had the SD profile). Neurons with the RD, OS and ID profiles had the narrowest size range with almost all smaller than 30 μ m in diameter (100, 99 and 91%, respectively). The mean soma diameter of neurons with the RD, OS and ID VTD response profiles was smaller than that of SD neurons ( We next investigated how the four VTD response profiles correlate with a commonly used functional marker of nociceptors. Capsaicin is a TRPV1 agonist used to activate peptidergic nociceptors 21 . In order to examine the relationship between VTD response profiles and sensitivity to capsaicin we applied them sequentially in a single calcium imaging protocol as illustrated in Fig. 4A. The order in which the two agonists were applied did not influence the proportion of responding neurons nor the frequency of occurrence of VTD response profiles, Supplementary Fig. S1. Therefore, we combined results from both experiments in subsequent analysis, Fig. 4B. The frequency of occurrence of the four response profiles in these neurons was in the same order as obtained without capsaicin, Fig. 4C. The percentage of capsaicin sensitive neurons in our cultures (32%) is comparable to published reports 22 . Capsaicin sensitive neurons were approximately twice as likely to respond to VTD (156 of 223, 70%) than not (67 of 223, 30%), Fig. 4B. We next examined VTD-responsive neurons for the relationship between each of the four VTD response profiles and capsaicin sensitivity. The proportions of capsaicin-sensitive and insensitive neurons within the OS and ID profiles were not significantly different. However, there was a clear bias in neurons with the SD and RD profiles towards capsaicin insensitivity, Fig. 4D.
As the SD profile occurred in a wide range of neuronal sizes, we examined the relationship between neuronal size and capsaicin sensitivity in the 111 neurons with the SD profile. Capsaicin-sensitive SD neurons had a smaller mean diameter than capsaicin-insensitive SD neurons, Fig. 4E, which is consistent with the known tendency of capsaicin sensitive neurons to be small to medium in diameter. Interestingly, capsaicin activated 35% (67 of 193) of VTD-unresponsive neurons which were all smaller than 30 μ m, Fig. 4F. From all the above, the SD and RD profiles are enriched in capsaicin-insensitive neurons.
The Veratridine SD and RD response profiles are under-represented in α, β-methylene ATP sensitive neurons. We next investigated how the four VTD response profiles correlate with sensitivity to another commonly used marker for nociceptors. α , β -methylene ATP is a specific agonist for P2× 3 receptors that marks the non-peptidergic population of nociceptors [23][24][25][26] . A preliminary experiment showed that responses to 1 μM α , β -methylene ATP (the EC50 for P2×3) can be completely inhibited by TNP-ATP, a selective P2×3 antagonist 27 (data not shown). Therefore, we applied 1 μ M α , β -methylene ATP with VTD to examine the relationship between the four VTD response profiles and sensitivity to α , β -methylene ATP. We applied the two agonists in alternate order, Fig 5A. As with capsaicin, the order in which VTD and α , β -methylene ATP were applied did not significantly influence the proportion of responding neurons nor the frequency of occurrence of VTD profiles, Supplementary Fig. S2. The frequency of occurrence of the four VTD response profiles was as obtained previously, Fig. 5C. The percentage of α , β -methylene ATP-responsive neurons (26%) was comparable to percentage of P2×3 positive neurons reported in immunolabelling studies 23,26 . α , β -methylene ATP-responsive neurons are approximately five times more likely to respond to VTD (289/344, 84%) than not (55/344, 16%), Fig. 5B. We next In (E,F), dotted lines represent the mean. Two-tailed unpaired Student's t-test, *P < 0.05 and **P < 0.01. examined the 912 VTD-responsive neurons for a relationship between each of the four VTD response profiles and α , β -methylene ATP sensitivity. The proportions of α , β -methylene ATP-sensitive and insensitive neurons within the OS and ID profiles were not significantly different, Fig. 5D. However, like with capsaicin, there was a clear bias in neurons with the SD and RD profiles towards α , β -methylene ATP insensitivity.
The Veratridine OS, ID and RD response profiles are enriched in nociceptors. Both the SD and RD response profiles were enriched in neurons insensitive to the two nociceptive markers; capsaicin and α , β -methylene ATP. This led us to hypothesise that these two response profiles could be functional markers for non-nociceptors, while the OS and ID response profiles could be functional markers for nociceptors. However, The OS and ID profiles did not occur preferentially in neurons sensitive to the nociceptive markers capsaicin (Fig. 4D) and α , β -methylene ATP (Fig. 5D). Nevertheless, it is possible that the OS neurons insensitive to capsaicin are the same neurons that are sensitive to α , β -methylene ATP and vice versa. To test this hypothesis, it was necessary to apply VTD, capsaicin and α , β -methylene ATP sequentially in the same protocol. We also wanted to profile VTD responses in C-low threshold mechanoreceptors because it was reported that this population is required for injury induced mechanical hypersensitivity 28 . This population expresses TRPA1 but not TRPV1 nor P2×3 29 ; therefore we included allyl isothiocyanate (AITC, a specific TRPA1 agonist [30][31][32] ) in our protocol. We applied the five agents in the calcium imaging protocol in two orders, Fig. 6A. The order of agents in the  (Figs 4 and 5), there was no significant difference in the proportion of responding neurons nor the frequency of occurrence of VTD profiles in both protocols, Supplementary Fig. S3.
Neurons sensitive to any of the three agonists were three times more likely to respond to VTD (333/442, 75%) than not (109/442, 25%), Fig. 6B. The frequency of occurrence of the four response profiles, Fig. 6C, was similar to previous values in Figs 3C, 4C and 5C. We next examined the 511 VTD-responsive neurons for a relationship between each of the four VTD response profiles and sensitivity to any of the three agonists. The three agonists divided the four VTD profiles into two groups. In one group is the SD response profile which continued to show a bias towards insensitivity to any of the three agonists, Fig. 6D. In the other group are the OS, ID and now the RD response profiles all showing a significant bias towards sensitivity to any of the three agonists, Fig. 6D.
The SD subpopulation of neurons sensitive to any of the three agonists have a smaller mean size (21 μ m) compared to the subpopulation insensitive to any of the three agonists (26 μ m), Fig. 6E. Interestingly, all neurons unresponsive to VTD but sensitive to any of the three agonists are smaller than 30 μ m in diameter, Fig. 6F. The percentage of responsive neurons to 100 μ M AITC (54%) is consistent with published findings 31 . The relationship between VTD response profiles and sensitivity to AITC alone matches that for the three agonists combined (Supplementary Fig. S4). From all the above, the OS, ID and RD response profiles are enriched in neurons sensitive to the nociceptive markers whereas the SD response profile is enriched in neurons insensitive to the nociceptive markers, Fig. 7.   We also examined the distribution of sensitivity to the three agonists among neurons with the four VTD profiles, Fig. 8. AITC sensitivity was the commonest feature of neurons that responded to one or more of the three agonists with the OS (88%), RD (86%), ID (81%) profiles but not the SD response profile (52%). Neurons with the SD response profile were approximately equally as likely to be capsaicin-sensitive (56%) or AITC-sensitive (52%).

Discussion
The present study examined VTD-elicited calcium responses in cultured adult mouse DRG neurons. VTD responses were heterogeneous. We categorised VTD responses into four distinct profiles based on the number of peaks and their decay rate. We report that VTD response profiles correlated with soma size and with commonly used pharmacological markers of nociceptor. To our knowledge this is the first detailed characterisation of VTD responses in DRG neurons.
What do VTD response profiles represent? Most of the studies that looked at the effect of VTD on voltage-gated ion channels used patch-clamp electrophysiology to examine the effect of VTD on sodium 8,9 , calcium 33 and potassium 13 currents separately. In calcium imaging the measured response occurs in non-clamped neurons and is indirect to sodium entry. Therefore, the observed VTD response profiles are the net effect of VTD action on VGSCs at the resting membrane potential and the subsequent activation of other ion channel/exchanger classes. Therefore, it is important to keep in mind that although the response is initiated by VTD's action on VGSCs, the overall response is shaped by; 1) Qualitative and quantitative differences in the "constellation" of sodium, potassium and calcium voltage-gated ion channels subtypes in each DRG neuron. 2) Variations in calcium buffering mechanisms within each neuron which includes calcium efflux pathways, intracellular stores as well as calcium binding proteins 34 .
Our data shows that 30 μ M VTD evoked robust responses in approximately 70% of sensory neurons. A higher dose of 100 μ M VTD did not significantly increase the percentage of responding cells, Fig. 1D. Considering that higher doses of VTD are reported to have inhibitory effects on potassium channels, 30 μ M seems to be the most suitable concentration to use for an action primarily on VGSCs. VTD predominantly activates TTX-S VGSCs and our results are in agreement with this as TTX blocked most VTD responses (Fig. 1E-G). Voltage-clamp experiments on rat DRG neurons showed that VTD binds TTX-R VGSCs but dissociates at much faster rate than with TTX-S VGSCs 10 , which might explain why the TTX-R VTD responses we observed are transient in nature, Fig. 1F. It remains to be determined which of the TTX-R channels expressed in DRG neurons (i.e. Nav1.8 and Nav1.9) underlie the observed TTX-R VTD responses. We speculate that Nav1.9 is the most likely candidate because its expression is more restricted in DRG than Nav1.8 and this fits with low incidence of TTX-R VTD responses. Secondly, Nav1.9 activation potential is hyperpolarised allowing it to be open at the resting membrane potential which would allow VTD to act on the open channel. In contrast Nav1.8 activation potential is more depolarised and is unlikely to be open at resting membrane potential for VTD to affect it. In fact, it was reported that VTD did not activate a stable cell line expressing Nav1.8 35 . Therefore, the contribution of Nav1.8 to VTD responses will be dependent upon prior action of VTD on TTX-S subtypes to depolarise the membrane. Since this would not happen in the presence of TTX; it is thus unlikely that Nav1.8 produces the TTX-R VTD responses in our experiment.
Importantly, the inability of VTD to activate stable cells expressing Nav1.8 may explain the about 25-30% of DRG neurons that are unresponsive to up to 100 μ M VTD, Fig 1D. We suggest that VTD-unresponsive neurons are so because they express mostly TTX-R and little TTX-S channels. In support of this the majority of neurons in this population are less than 30 μ m in diameter, Figs 3D, 4F, 5F and 6F. Additionally, the majority of neurons in this population (109/160, 68%) responded to at least one of the three nociceptor markers; capsaicin, AITC and α , β -methylene ATP, Fig. 6B. These 109 neurons constitute 25% of nociceptors (109/442, Fig. 6B). These neurons are likely to represent the small diameter, Nav1.8-rich and high-threshold neurons known as "silent nociceptors" 36,37 . Interestingly, silent nociceptors were estimated to be about 30% of all DRG neurons which is about the same percentage as the VTD-unresponsive population. This finding has significant implications for studies using veratridine to activate DRG neurons as it means that up to 25% of nociceptors would not be assayed and their contribution would be unknown.
We categorised VTD calcium responses in cultured DRG neurons into four profiles which we named SD, ID, RD and OS. The four profiles were observed in similar relative frequencies from cultures prepared over many months, from different patches of mice and by different researchers. Furthermore, the frequency of occurrence of VTD profiles was not affected by the order of applications of agonists in our experiments, Supplementary Figs 1-3. The stability of these profiles supports their suitability for use as functional signatures of subpopulations of DRG neurons in drug screens.
The most abundant response type is the OS (47-51% of VTD-responsive neurons) where VTD elicited oscillatory changes in [Ca]i levels. Interestingly, this is the only response profile VTD elicited in N2a and SH-SY5Y cells, Fig. 1B,C. A similar response profile was also reported in bovine chromaffin cells where VTD induced [Ca]i oscillations were dependant on the activation of TTX-S VGSC 25 . The oscillatory response in bovine chromaffin cells were long lasting (up to 40 min) even after VTD wash. A proportion of the OS responses in DRG neurons were persistent but most returned to baseline after VTD wash (not shown). Although SH-SY5Y, N2a and chromaffin cells 12,38,39 all express Nav1.7, it is unlikely that the OS profile is a characteristic of Nav1.7 expressing cells as Nav1.7 is expressed in all DRG neurons whereas the OS profile occurs in about 50% of DRG neurons.
The other three types of VTD response profiles shared a profile characterised by a single peak but differed in their decay rate. The difference in the rate of signal decay could be due to a difference in the disassociation rates of VTD from the different VGSC subtypes. In addition, it has been shown that DRG neurons (particularly large-diameter neurons) produce calcium responses with slow decay rates (SD-like responses) when treated with VGKC blockers 40 . Therefore, differences in the decay rates of the VTD response profiles might be in part due to some inhibitory effect of 30 μ M VTD on VGKCs. However, this is unlikely to be the main determinant of decay rates as it takes a much higher concentration of VTD (200 μ M) to inhibit VGKCs 13 .
A recent paper classified DRG neurons into 11 groups based on their mRNA profile 29 . Using their data (Figs 2 and 4 in ref. 29) we predicted the distribution of sensitivity of DRG neurons to the three agonists used in this study based on the expression of their receptors, Fig. 9A. We compared this to the distribution of sensitivity we obtained from applying the concentrations of agonists we used, Fig. 9B. The two distributions have many similarities. For example, all non-peptidergic neurons (represented by groups NP1-3 in ref. 29) are predicted to respond to AITC and in agreement we found that almost all α , β -methylene -ATP responded to 100 μ M AITC. Another similarity is that the TRPA1 positive population is the largest of the three agonists being 64% of DRG neurons by mRNA expression of TRPA1 and 54% by response to AITC. The percentage of AITC responders (54%) is almost identical to that reported by Barabas, M.E. et al. 31 (55%). Therefore, the 10% higher percentage of TRPA1 expressing neurons must be due to the higher sensitivity of mRNA detection compared to functional imaging. The same reason is likely to explain the 9% higher percentage of neurons not responding to any of the agonists in our hands compared to what is predicted by mRNA expression. Capsaicin activated 25% DRG neurons in our experiments as opposed to a predicted 17% based on expression of TRPV1 receptor. The percentage of capsaicin responders in our experiments is in agreement with others (e.g.ref. 22). The higher percentage in our and others' experiments could be due increased TRPV1 expression in cultured DRG neurons (> 1 day) compared to disassociated but non-cultured neurons as used for the mRNA sequencing study.
The data presented here raises several interesting questions. One question is what are the key molecular determinants of each of the four VTD response profiles? Answering this allows the use of VTD profiles as readout for these ion channels and exchangers in high throughput functional screens on primary DRG neurons. Considering that 25-30% of nociceptors do not respond to 30 μ M VTD and that these are likely to be "silent" nociceptors, a second question is does the sensitization of DRG neurons by inflammation and nerve injury affect the number of VTD responsive neurons and their response profiles? Answering this question would allow the use of VTD  Fig. 6. The classification of neuronal subpopulations is as per that in ref. 29. NP is non-peptidergic, PEP is peptidergic, NF is neurofilament heavy chain and TH for tyrosine hydroxylase.
Scientific RepoRts | 7:45221 | DOI: 10.1038/srep45221 response profiles as readout of drugs' ability to cause or reverse sensitisation of DRG neurons in high throughput screens. Such screens have the advantage of showing the net effect of a drug on the various classes of ion channels that give rise to VTD response profiles.
In conclusion, the present study shows that VTD induces calcium responses with heterogeneous profiles in TTX-S rich sensory neurons. VTD response profiles reflect distinct subpopulations of sensory neurons. These subpopulations overlap but are not identical to the subpopulations identified by classical functional nociceptive markers, Fig. 7. The OS and RD profiles are particularly enriched in nociceptors (neurons sensitive to at least one of the three agonists), while the SD profile is enriched in non-nociceptors (neurons insensitive to any of the three agonists). Our findings provide a detailed characterisation of VTD action on the different subsets of DRG neurons. Our work is relevant to studies and screens using VTD to activate DRG neurons.

Methods
DRG culture. Adult male C57BL/6 mice were sacrificed according to Schedule 1 of the Animal (Scientific procedure) Act 1986. DRG from all spinal levels were isolated and collected in PBS. PBS was then replaced with 1 mL Dulbecco's Modified Eagle's Medium/F12 (DMEM/F12) with Glutamax medium (Gibco) containing Dispase (1 mg/mL, Sigma) and Collagenase Type XI (0.6 mg/mL, Sigma) and left for 60 min at 37 °C and 5% CO 2 . DRGs were then triturated with a P1000 pipette tip. The cell suspension was carefully layered on top of 15% Bovine Serum Albumin (Melford) in DMEM/F12 and centrifuged at 800 g, for 10 min at room temperature with the minimum deceleration speed. The cell pellet was then washed in culture medium composed of DMEM/F12 plus 10% Fetal Bovine Serum (FBS, Gibco), 100 units/ml penicillin and 100 μ g/ml streptomycin (Gibco). Cells were pelleted again, re-suspended in DRG culture medium and plated on glass coverslips coated with polyornithine (20 μ g/mL, Sigma). Cells were imaged 24 h after plating.
Calcium imaging. DRG neurons were loaded with 2 μ M Fura-2, AM (Molecular Probes) in standard extracellular solution (140 mM NaCl, 4 mM KCl, 2 mM CaCl 2 , 10 mM HEPES, 5 mM glucose, pH = 7.4 with NaOH) and incubated for 30 min at 37 °C. Coverslips were then washed with standard extracellular ringer solution and left for 15 min at 37 °C and 5% CO 2 and then for another 15 min at room temperature. Cells were excited with 350 and 380 nm for ratiometric measurement of intracellular calcium using Cairn Dual OptoLED system. Cells were viewed using a 40X oil immersion objective. Images were acquired using a Hamamatsu C4742-95 camera. The cells were perfused with standard extracellular solution for at least 5 min to establish stable baseline. All recordings were performed at room temperature (23 ± 1 °C). Drugs were perfused at a flow rate of 3 mL/min. High potassium extracellular solution (104 mM NaCl, 40 mM KCl, 2 mM CaCl 2 , 10 mM HEPES, 5 mM glucose, pH = 7.4 with NaOH) was perfused at the end of the recordings, unless mentioned otherwise, as a depolarizing agent to identify viable neurons. Simple PCI 6 software was used for data acquisition, background subtraction and Fura-2,AM ratiometric measurement (F350/380 nm). Data and statistical analysis. Neurons were identified by their responsiveness to 40 mM KCl. We defined a response as an increase in (F350/380) ratio of > 6 SD above the baseline. Differences in fluorescence (Δ F/F 0 ) were calculated according to the following formula: F350/380 ratio in the presence of drug (F) during drug application -the mean of F350/380 ratio of the 2.5 min prior to drug application (F 0 ). Statistical analysis was performed by calculating the mean percentage of responsive neurons (n) from the indicated number of mice for each figure (N). Sample mean of each set of experiments was calculated from multiple independent experiments (as specified in the result section) and compared to each other by one-way analysis of variance (ANOVA) with Sidaks' post-test. Cell diameter measurements were performed using ImageJ software; cell area was obtained from a hand-drawn line delineating the soma. Soma diameter was then calculated using the following formula, Diameter = √ (4.Area/π ). For cell size comparison, two-tailed unpaired Student's t-test was used for comparisons. All statistical analysis and comparisons were performed by GraphPad Prism software (version 7.00 for Windows). Area-proportional Venn diagrams were generated by BioVenn software 41 and colour edited by CorelDRAW X8 software.