1. Neuroscience
Download icon

Voltage-gated Na+ currents in human dorsal root ganglion neurons

  1. Xiulin Zhang
  2. Birgit T Priest
  3. Inna Belfer
  4. Michael S Gold  Is a corresponding author
  1. The Second Hospital of Shandong University, China
  2. Lilly Research Laboratories, United States
  3. National Institutes of Health, United States
  4. University of Pittsburgh School of Medicine, United States
Research Article
  • Cited 21
  • Views 1,734
  • Annotations
Cite this article as: eLife 2017;6:e23235 doi: 10.7554/eLife.23235

Abstract

Available evidence indicates voltage-gated Na+ channels (VGSCs) in peripheral sensory neurons are essential for the pain and hypersensitivity associated with tissue injury. However, our understanding of the biophysical and pharmacological properties of the channels in sensory neurons is largely based on the study of heterologous systems or rodent tissue, despite evidence that both expression systems and species differences influence these properties. Therefore, we sought to determine the extent to which the biophysical and pharmacological properties of VGSCs were comparable in rat and human sensory neurons. Whole cell patch clamp techniques were used to study Na+ currents in acutely dissociated neurons from human and rat. Our results indicate that while the two major current types, generally referred to as tetrodotoxin (TTX)-sensitive and TTX-resistant were qualitatively similar in neurons from rats and humans, there were several differences that have important implications for drug development as well as our understanding of pain mechanisms.

https://doi.org/10.7554/eLife.23235.001

Introduction

It has long been appreciated that voltage-gated Na+ channels (VGSCs) underlie the upstroke of the action potential, and therefore play an essential role in the propagation of action potentials along axons (Hodgkin and Huxley, 1952a, 1952b). VGSCs consist of an alpha subunit, responsible for all essential features of a functional channel and up to two beta subunits that influence channel density and/or gating properties (Yu et al., 2005). Nine alpha subunits and four beta subunits have been identified. VGSCs have remained a target for the development of novel therapeutics because it is now also appreciated that changes in the biophysical properties (Cantrell and Catterall, 2001), distribution (Cusdin et al., 2008; Kuba et al., 2010), and/or expression (Aptowicz et al., 2004; Qiao et al., 2013) of these channels contributes to the dynamic regulation of neuronal excitability. Such changes have been particularly well documented in the context of pain, where both phosphorylation-dependent increases in Na+ current, as well as more persistent increases in VGSC expression, have been shown to underlie both the acute and persistent increases in nociceptor excitability associated with inflammation (Gold et al., 1996; Gould et al., 1998). Similarly, both the ongoing pain and hypersensitivity associated with peripheral nerve injury are associated with changes in the pattern of VGSC expression (Gold et al., 2003; Waxman et al., 1994; Hunter et al., 1997) as well as distribution of channels in peripheral nerves (Gold et al., 2003; Henry et al., 2007; Tseng et al., 2014). From a therapeutic perspective, what has been particularly exciting about the evidence implicating VGSCs in inflammatory and neuropathic pain, is that several of the VGSC alpha subunits shown to contribute to the injury-induced increases in afferent excitability are preferentially expressed in the peripheral nervous system in general, and nociceptive afferents in particular (Gold and Gebhart, 2010). This has raised the intriguing possibility that a VGSC subtype specific blocker would provide effective pain relief with minimal side effects.

Evidence in support of a role of VGSCs in a variety of pain states obtained with pre-clinical, largely rodent models, has been confirmed in pain patients. Similarly, genetic and anatomical evidence suggests that the VGSC alpha and beta subunits, and their pattern of expression, are similar in rodents and man. Nevertheless, virtually all that is known about the biophysical and pharmacological properties of VGSCs comes from the study of these channels in heterologous expression systems and in isolated rodent sensory neurons. The potential problem with this situation is highlighted by evidence that both the biophysical and pharmacological properties of channels are influenced by the expression system and species differences. For example, co-expression of VGSC beta1 and beta2-subunits with the alpha subunit NaV1.2 in frog oocytes not only increases current density, but the rate of current inactivation (Patton et al., 1994). In contrast, these beta-subunits have no influence on either current density or inactivation when co-expressed with NaV1.2 in tsA-201 cells (Qu et al., 2001). Rather, in tsA-201 cells, the beta-subunits drive a rightward shift in the voltage-dependence of channel activation. Similarly, the putatively NaV1.8 selective blocker A-803467 is over three orders of magnitude more potent against heterologously expressed human NaV1.8 than four of the other nine human alpha subunits tested, yet is only half as potent against the current believed to reflect activation of NaV1.8 channels, natively expressed in rat dorsal root ganglion neurons (Jarvis et al., 2007). Thus, the purpose of the present study was to determine the extent to which the biophysical and pharmacological properties of VGSCs described in rodent sensory neurons reflect the properties of the VGSCs in human sensory neurons. We focused on the two major classes of current that have been most extensively studied in rodent sensory neurons, those historically referred to as the low threshold rapidly activating, rapidly inactivating tetrodotoxin (TTX) sensitive current, and the high threshold more slowly activating and slowly inactivating TTX-resistant current, although voltage-protocols were used to isolate the putative TTX-resistant current in the majority of experiments described.

Whole cell patch clamp techniques were used to study acutely dissociated dorsal root ganglion (DRG) neurons from rats and humans. Our results suggest that while the two major Na+ current types in neurons from rats are similar to those in neurons from humans, there are several potentially important pharmacological and biophysical differences that could contribute to the limited success in the development of novel pain therapeutics.

Results

Neurons included in this study were obtained from 21 donors (13 males and 8 females). The average age of the donors was 45.2 years with a range of 13 to 77. Additional demographic data are summarized in Table 1. At least one data point was obtained from a total of 226 neurons that met inclusion criteria for clamp control and holding current. In order to maximize the amount of data collected from each neuron, we first determined whether it was possible to use voltage-clamp protocols to isolate the rapidly activating rapidly inactivating putative TTX sensitive (TTX-S) Na+ current from the more slowly activating and slowly inactivating TTX-resistant (TTX-R) Na+ current, as we have previously done in the rat (Gold et al., 2003). We subsequently confirmed that currents isolated in this manner were identical to those isolated with TTX (Figure 1). Based on these results, we refer to the slowly activating and slowly inactivation current resistant to steady-state inactivation as TTX-R current even though voltage steps rather than TTX was used to isolate this current in all subsequent experiments.

Table 1

Donor demographics and Na+ current density.

https://doi.org/10.7554/eLife.23235.002
SexAge (yrs)TTX-R INa Current Density (pA/pF)TTX-S INa Current Density (pA/pF)
Male (n = 13)40.9 ± 4.9 (1 Latin American, 1 African American, 11 Caucasian )−42.5 ± 2.7 (n = 69)−66.2 ± 5.0 (n = 51)
Female (n = 8)52.3 ± 5.4 (all Caucasian)−46.1 ± 5.3 (n = 50)−54.4 ± 8.2 (n = 31)
  1. Data are mean ± SEM. Differences between males and females are not statistically significant (p > 0.05).

Separation of tetrodotoxin (TTX) sensitive (TTX-S) and resistant (TTX-R) voltage gated Na+ currents in human DRG neurons.

(A) A steady-state availability protocol was used to assess the voltage-dependence of inactivation of TTX-S and TTX-R currents. Left panel: The protocol consisted of a 500 ms pre-pulse to potentials between −100 and +10 mV, followed by test pulse to 0 mV. Right panel: The fast component of the current evoked at 0 mV (open circles), was inactivated over a range of test potentials more hyperpolarized than the range of test potential over which the slow component of the current evoked at 0 mV (closed circles) was inactivated. (B) Left Panel, top traces: Current evoked during the test pulse in A (left panel), plotted on a shorter time scale to more clearly illustrate the fast and slow components of the current evoked at 0 mV. Because the fast component of the current was completely inactivated with a test pulse more hyperpolarized than that at which the slow component began to inactivate, it was possible to digitally isolate the fast component by subtracting the slow component (purple trace), from the total current. Conversely, because the fast component was completely inactivated within ~10 ms of the start of the test pulse, it was possible to generate an availability curve for the slow component across the entire range of pre-pulse potentials. The bottom traces are those of the fast component digitally isolated from the slow component. Right panel, top traces: In another neuron, the steady-state availability protocol used in A, was used to inactivate the fast component of the current evoked at 0 mV. Application of 300 nM TTX removed the same component of the total current as the test pulse to −40 mV. Bottom traces: The difference between the total current and the current evoked at −40 mV, or that evoked in the presence of 300 mM TTX is virtually identical. (C). Current-voltage (I-V) protocols were used to assess current activation, with pre-pulse potentials that were based on steady-state availability data. Thus, total current (top traces) was evoked following a 500 ms pre-pulse to a potential at which currents were fully available for activation (i.e., −100 mV). TTX-R currents (middle traces) were evoked following a 500 ms pre-pulse to a potential at which TTX-S currents were completely inactivated, but TTX-R currents were fully available for activation (i.e, −35 mV). It was then possible to digitally isolate TTX-S currents (Bottom traces) from the total current by subtracting TTX-R current from the total current.

https://doi.org/10.7554/eLife.23235.003

We initially focused on the small to medium diameter DRG neurons from both human and rat in this study for two main reasons. First, because the study of Na+ channels in sensory neurons has largely been in the context of pain. In this context, data from guinea pig and rodents suggest that neurons with a small to medium cell body diameter are more likely to give rise to slowly conducting axons (Lawson, 2002), which are, in turn, more likely to be nociceptive. Second, it was more difficult to maintain clamp control over the currents evoked from human sensory neurons with a larger cell body diameter. Our decision about the size range in which to consider a human DRG neuron small- or medium-diameter was based on the distribution of cell body sizes observed in crysections of whole ganglia obtained from three donors (Figure 2A, inset). Interestingly, in contrast to the skewed, clearly bimodal distribution of DRG neuron cell body sizes previously described in rodents (i.e., see [Lawson et al., 1993]), human DRG neuron cell body sizes were relatively normally distributed. We therefore included a subpopulation of larger neurons in subsequent experiments to further test the association between cell body diameter and phenotype in human DRG neurons. The average membrane capacitance of all the neurons included in this study was 116 pF (or ~60 ± 0.8 μm in diameter), with a range from 23 to 255 pF (or 26 to 90 μm). A histogram of the size distribution is plotted in Figure 2. As expected, human DRG neurons are significantly larger than small to medium (cell body diameter < 35 μm) rat DRG neurons, which had an average cell body capacitance of 36 pF, with a range of 22 to 70 pF (Figure 2A). Consistent with previous evidence from the rat suggesting that TTX-R currents are enriched in nociceptive afferents (Djouhri et al., 2003), which tend to have a small cell body diameter (Lawson, 2002), no TTX-R current was detected in rat neurons with a cell body capacitance >55 pF. Conversely, there was no correlation (p>0.05) between human DRG neuron cell body diameter and TTX-R current density, or the ratio of TTX-S to TTX-R current, and TTX-R currents were present in the largest neuron studied. Furthermore, the average capacitance of the nine neurons studied in which only TTX-S current was detected, 83.2 ± 18 pF, was, if anything, smaller (p=0.059), than that of the population of neurons with both TTX-S and TTX-R currents.

The size distribution of human and rat DRG neurons.

(A) A histogram of the size distribution of all of the human (n = 226) and rat (n = 44) DRG neurons included in this study. Membrane capacitance was used as an indirect measure of cell size, since the capacitance is a reflection of the cell surface area. Capacitance was determined with amplifier circuitry. The average membrane capacitance of human DRG neurons was 116 pF, while that of rat DRG neurons was 36 pF. Inset: The cell size distribution of human DRG neurons from cryosections (20 μm) of paraformaldehyde post-fixed DRG. Data are from 20 sections, collected at 200 μm intervals, from each of three donors. Data from neurons studied with patch-clamp (Recording) have been replotted for comparison, where membrane capacitance was used to estimate cell body diameter based on the assumption that the capacitance of human DRG neurons is one μF/cm2. An example of the tissue counted for this analyses is shown, where the scale bar is 50 μm. (B) In the neurons studied in which peak TTX-S and TTX-R were determined, the ratio of TTX-S to TTX-R current density was significantly greater in human (n = 71) than rat (n = 26) DRG neurons. ** is p<0.01.

https://doi.org/10.7554/eLife.23235.005

Despite normalizing for the differences in cell body capacitance, the density of both TTX-R and TTX-S current were significantly larger in neurons from human (−53.2 ± 6.8 pA/pF (n = 114) and −62.2 ± 4.3 pA/pF (n = 78), respectively) compared to those in rat neurons (−47.7 ± 3.3 pA/pF (n = 26) and −42.6 ± 4.6 pA/pF (n = 26), respectively). Furthermore, the ratio of TTX-S to TTX-R current density was significantly greater in human than rat DRG neurons (Figure 2B).

Inflammatory mediator-induced increase in both TTX-R and TTX-S currents in human DRG neurons

As the acute inflammatory mediator-induced potentiation of TTX-R currents in rat DRG neurons was one of the first observations driving a focus on these currents in the context of pain (Gold et al., 1996), we sought to determine the impact of inflammatory mediators on Na+ currents in human DRG neurons. An inflammatory soup, consisting of bradykinin (10 μM), histamine (1 μM) and prostaglandin E2 (1 μM), was applied to neurons following establishment of stable recordings. Consistent with previous results from rat and subsequently mouse sensory neurons (Yang and Gereau, 2004), TTX-R currents were increased within seconds following bath application of inflammatory soup (Figure 3). This increase saturated within ~90 s. Comparable results were obtained in 13 of 17 human neurons tested, resulting in an average increase in peak current of 15 ± 1%. Concomitantly, rates of current activation and inactivation were increased, 9 ± 0.8% and 12 ± 3%, respectively. Both the increase in peak current and the increase in current activation rate would contribute to an increase in excitability.

Inflammatory mediators (IM) increase Na+ current in human DRG neurons.

A combination of bradykinin (10 μM), histamine (1 μM) and prostaglandin E2 (1 μM) was bath applied to neurons after establishing the stability of evoked currents. (A) TTX-R currents were increased within seconds of IM application, and this increase was largely saturated within 90 s. In 13 of 17 neurons tested, the average increase in peak TTX-R current was 15 ± 1%. TTX-S currents were increased by 30 ± 6% in four of 10 neurons tested. Insets: Typical traces of TTX-R (above) and TTX-S (below) current evoked with a voltage step to 0 mV before (black) and after (red) IM application, following a 500 ms prepulse to −40 mV or −90 mV, respectively.

https://doi.org/10.7554/eLife.23235.007

Changes in TTX-S currents were assessed in 10 of the 17 neurons in which inflammatory soup-induced changes in TTX-R currents were assessed. TTX-S currents were increased in four of these neurons by 30 ± 6% (Figure 3). However, in contrast to TTX-R currents, the increase in TTX-S current was not associated with changes in rates of current activation or inactivation.

Biophysical properties of TTX-R currents in rat and human DRG neurons

Because, as noted above, the vast majority of what is known about the biophysical and pharmacological properties of TTX-S and TTX-R currents in sensory neurons was derived from the study of rodent sensory neurons, we next sought to compare the biophysical and pharmacological properties of currents in human DRG neurons with those of currents in rat DRG neurons. Data were collected from DRG neurons obtained from eight rats under conditions identical to those used for the study of human DRG neurons with the same series of protocols. Data collection from these eight rats was interleaved with the collection of the last ~third of the human DRG neuron data, enabling the use of the same stock solutions of test reagents. The steady-state properties of TTX-R currents in rat DRG neurons were qualitatively similar to those in human DRG neurons. However, in contrast to previous results indicating that persistent TTX-R currents were larger in human DRG neurons (Han et al., 2015), we observed the opposite (Figure 4A and B), where the persistent current in rat neurons at −10 and 0 mV was significantly larger than that in human neurons (Figure 4C, p<0.01 two-way ANOVA with Holm-Sidak post hoc test). There were also small but significant differences in the V0.5 of both inactivation (Figure 4D), and activation (Figure 4E), which were more hyperpolarized in human neurons (p<0.01). In contrast, recovery from inactivation (Figure 4F) was comparably rapid in neurons from both human and rat.

Steady-state biophysical properties of TTX-R currents in human and rat DRG neurons.

(A) A 500 ms test pulse to potentials between −40 and 0 mV from a holding potential of −40 mV was used to assess the presence of persistent TTX-R current in human (A) and rat (B) DRG neurons. (C) The mean (± SEM) of persistent current analyzed as the % of the peak TTX-R current, is plotted relative to the voltage at which the current was evoked in neurons from rat (n = 30) and human (n = 34). (D) Steady-state inactivation of TTX-R current was assessed with the protocol shown in Figure 1. Availability curves were fitted with a modified Boltzmann equation to determine Imax, the slope and the voltage at which TTX-R current were half inactivated (V0.5). Data for each neuron were then normalized to the calculated Imax. There was small but significant difference in the V0.5 of inactivation of TTX-R current in neurons from rat (n = 25) and human (n = 99, although data plotted were 25 neurons to facilitate comparisons with rat data). (E) G-V curves were generated from I-V data and fitted with a modified Boltzmann equation to determine Gmax, the slope, and the voltage (V0.5) at which conductance was half of Gmax. Data for each neuron were normalized to the calculated Gmax. There was a small but significant difference in the V0.5 of activation of TTX-R current in neurons from rat (n = 25) and human (n = 123, although data plotted were again from 25 neurons to facilitate comparisons with rat data). (F) Recovery from inactivation of TTX-R current was assessed with a two pulse protocol shown at the inset, the extent of recovery from inactivation was determined by comparing the peak inward current evoked during the test pulse(second) to that evoked during the conditioning pulse(first). This ratio is plotted relative to the interpulse duration. Recovery curves were fitted with a double exponential. Recovery from inactivation of TTX-R current in neurons from rat (n = 10) and human (n = 13) were comparable.

https://doi.org/10.7554/eLife.23235.009

The kinetic properties of TTX-R currents in neurons from rat and human were also qualitatively similar (Figure 5A and B) with comparable rates of current activation (Figure 5C) and inactivation (Figure 5D).

Kinetic properties of TTX-R current in human and rat DRG neurons.

TTX-R I-V data from human (A) and rat (B) DRG neurons was used to assess the voltage-dependence of the rates of current activation (C) and inactivation (D). Currents were evoked with 15 ms voltage steps to potentials ranging between −40 mV and +40 mV, following a 500 ms pre-pulse to a potential at which TTX-S currents were completely inactivated and TTX-R currents were fully available for activation. That potential was −40 mV for the neurons studied in A, and B. The rising phase of TTX-R current at each potential was fitted with a single exponential to determine the rate of current activation. The falling phase of the TTX-R current during each depolarizing voltage step was fitted with a single exponential to determine the rate of current inactivation. Data in each plot are from 25 rat neurons and 27 human neurons.

https://doi.org/10.7554/eLife.23235.011

Pharmacological properties of TTX-R currents in rat and human DRG neurons

Two experiments were performed to enable comparison of the pharmacological properties of TTX-R currents in rat and human DRG neurons. In the first, the impact of the putatively NaV1.8 selective channel blocker, A-803467, was assessed. While this compound was shown to be both highly selective and potent against human NaV1.8 in heterologous expression systems (Jarvis et al., 2007), we were unable to obtain any evidence of an A-803467-induced decrease in TTX-R current in human DRG neurons at concentrations between 3 and 100 nM (n = 3–7 per concentration, data not shown). It was only at the very highest concentration tested (1 μM), that small but consistent block of TTX-R current was observed (Figure 6A). The same concentration of A-803467 blocked over 50% of TTX-R current in rat DRG neurons (Figure 6B). The difference in fractional block between human and rat obtained with this concentration was statistically significant (Figure 6C, p<0.01).

Impact of A-803467 on TTX-R current in human and rat DRG neurons.

TTX-R current was evoked with a 15 ms depolarizing voltage step to 0 mV every 10 s, following a 500 ms prepulse to −65 mV in (A), and −40 mV in (B) Current evoked before (black) and after (grey) application of A-803467 (1 μM) to a human (A) and a rat (B) DRG neuron. Of note, as a prepulse to −65 mV would not inactivate all the TTX-S current in the human DRG neuron, the more rapid current activation likely reflects the contaminating presence of this faster current. Nevertheless, there is only a small reduction in current at the end of the test pulse. (C) The mean block of current in rat neurons (n = 4) was significantly greater than that in human neurons (n = 4). Of note, no detectable block was observed in eight other human neurons tested with lower concentrations of A-803467 (30–300 nM).

https://doi.org/10.7554/eLife.23235.013

In the second pharmacological experiment, we assessed both resting and use-dependent block of TTX-R currents with lidocaine. The potency of lidocaine was comparable against resting TTX-R currents from rat and human DRG neurons (Figure 7A,B and C). However, not only did TTX-R currents from human DRG neurons demonstrate little use-dependent inactivation in the absence of lidocaine, there was no detectable use-dependent block of these currents in the presence of lidocaine (Figure 7A and D). This was in contrast to the use-dependent inactivation and block of TTX-R currents in rat DRG neurons in the absence and presence of lidocaine, respectively (Figure 7B and D, p<0.01 two-way ANOVA).

Resting and use-dependent block of TTX-R currents with lidocaine in human and rat DRG neurons.

TTX-R current was evoked twenty times with a voltage step to 0 mV at 1, 2 and 5 Hz before and after the application of lidocaine. The responses to the first five voltage steps evoked at 2 Hz before and after 100 µM lidocaine are shown for a human (A, evoked from a holding potential of −40 mV) and rat (B, evoked from a holding potential of −35 mV) DRG neuron. (C) Lidocaine-induced steady-state block of TTX-R current was assessed ~three minutes after application of each concentration of lidocaine, prior to the initiation of the use-dependent block protocols. The steady-state block of currents in human (n = 9) and rat (n = 10) neurons were comparable. (D) Use-dependent block, calculated as the fraction of current evoked at the 20th pulse relative to that evoked with the first (P20/P1). Use-dependent block of TTX-R current observed in rat DRG neurons was increased in the presence of lidocaine. In contrast, there was little use-dependent block of TTX-R currents in human DRG neurons in the absence of lidocaine, and the only evidence of a lidocaine-induced increase in use-dependent block was observed at a stimulation frequency of 1 Hz.

https://doi.org/10.7554/eLife.23235.015

Biophysical properties of TTX-S currents in rat and human DRG neurons

As with TTX-R currents, TTX-S currents in rat and human DRG neurons were qualitatively similar. There was also no significant difference between species with respect to the steady-state inactivation of TTX-S currents (Figure 8A). However, the voltage-dependence of TTX-S activation was significantly more hyperpolarized in human than in rat DRG neurons (Figure 8B, p<0.01). Furthermore, while over 80% of TTX-S current from human DRG neurons recovered from inactivation with a fast time-constant, only 50% of TTX-S current from rat DRG neurons recovered as rapidly (Figure 8C, p<0.01).

Steady-state biophysical properties of TTX-S current in human and rat DRG neurons.

TTX-S currents were isolated as described in Figure 1. Steady-state availability (A), activation (B), and recovery from inactivation (C) data were collected for TTX-S currents as described for TTX-R currents in Figure 4, except that the holding and recovery potential were −90 mV. Availability data were from 29 rat neurons and 99 human neurons (although data from only 29 neurons are plotted, to facilitate comparisons between human and rat). G-V data were also from 29 rat neurons, and from 128 human neurons (although data from only 29 neurons is plotted). Recovery data are from the 8seven rat neurons and 8 human neurons. The faction of current recovered with a fast time constant in human neurons was significantly greater than that in rat neurons.

https://doi.org/10.7554/eLife.23235.017

TTX-S current kinetics in neurons from rat and human were also qualitatively similar (Figure 9A - D). However, current activation rate in human DRG neurons was faster than that in rat DRG neurons across the voltage range of current activation that was tested (Figure 9C).

Kinetic properties of TTX-S current in human and rat DRG neurons.

TTX-S I-V data from human (A) and rat (B) DRG neurons was used to assess the voltage-dependence of the rates of current activation (C) and inactivation (D). Current was isolated as described in Figure 1, where in the neurons shown in (A), and (B), the pre-pulse potential was −90 mV to evoke total current, and –35 and –40 mV to inactivate TTX-S currents in the human and rat neuron, respectively. Activation and inactivation rates were determined as described in Figure 5. The activation of currents in human DRG neurons (n = 128, although data from only 29 are plotted to facilitate comparisons with rat data) was significantly faster, and demonstrated significantly less voltage-dependence than TTX-S currents from rat DRG neurons (n = 29). However, inactivation rates and the voltage-dependence of this process was comparable in currents from rat and human neurons.

https://doi.org/10.7554/eLife.23235.019

Pharmacological properties of TTX-S currents in rat and human DRG neurons

Four pharmacological experiments were also performed on TTX-S currents. The first was prompted by the initial observation that TTX-S currents in human DRG neurons were less sensitive to TTX than was our previous experience from rat DRG neurons. Consistent with this impression, there was a significant difference between TTX-S currents from rat and human DRG neurons with respect to the fraction of current blocked by 30 nM TTX. That is, only ~30% of the TTX-S current in human DRG neurons was blocked by this concentration of TTX (Figure 10A) in contrast to the >90% of current blocked in rat DRG neurons (Figure 10B). This difference was statistically significant (Figure 10C, p<0.01).

The potency of TTX block of TTX-S currents was lower in human than in rat DRG neurons.

TTX-S currents isolated as described in Figure 1 where in the neurons shown in (A) (human), and (B) (rat), the pre-pulse potential was −90 mV to evoke total current, and −40 mV to inactivate TTX-S currents in both the human and rat neurons shown. Currents were evoked before and after 30 nM TTX application. (C). The fractional block of TTX-S current produced by 30 nM was significantly (p<0.01) greater in rat (n = 9) than in human (n = 8) DRG neurons.

https://doi.org/10.7554/eLife.23235.021

Lidocaine was again used in the second experiment. In contrast to the results obtained with TTX-R currents, there were differences between rat and human DRG neurons with respect to the resting block of TTX-S current, where lidocaine was less potent in the rat than the human (Figure 11A,B and C, p<0.01). Furthermore, while TTX-S currents in neurons from both rat and human were subject to use-dependent inhibition in the absence of lidocaine (Figure 11A,B and D), the inhibition was significantly (p<0.01) greater in rat neurons, at least at 1 and 2 Hz (Figure 11D). In both human and rat neurons, lidocaine caused significant (p<0.01) use-dependent block (Figure 11A,B and D).

Resting and use-dependent block of TTX-S currents with lidocaine in human and rat DRG neurons.

Currents were evoked in neurons from human (A) and rat (B), as described in Figure 7 with a voltage step to −25 mV from a holding potential of −90 mV. (C). Lidocaine-induced steady-state block of TTX-S currents was also determined as in Figure 7. However, in contrast to TTX-R currents, the potency of lidocaine-induced block of TTX-S currents in human DRG neurons (n = 8) was significantly higher than that in rat DRG neurons (n = 13). (D). Lidocaine was associated with a significant increase in use-dependent block of TTX-S currents, assessed at 1, 2 and 5 Hz, in both human and rat neurons.

https://doi.org/10.7554/eLife.23235.023

The third and fourth experiments we designed to begin to assess the channel subtypes underlying the TTX-S current, given a growing body of evidence pointing to NaV1.7 as a therapeutic target for the treatment of pain (Vetter et al., 2017). As an initial foray into this question, we assessed the presence of a low threshold ‘ramp current’ in human DRG neurons based on rodent and heterologous expression data suggesting that because of the relatively slow development of closed-state inactivation of NaV1.7 channels, this subunit was responsible for large, low threshold currents evoked with a ramp depolarization. Despite the presence of TTX-S currents in every neuron included in this analysis, we detected no evidence of a low threshold ramp current in human DRG neurons in response to a depolarizing ramp from −100 mV to 0 mV over 500 ms, following a 500 ms pre-pulse to −90 mV (Figure 12A). Rather, peak inward current evoked in response to the ramp was at −5.9 ± 0.6 mV (n = 74). Because closed state-inactivation of NaV1.7 in human neurons may develop more rapidly than in rodent sensory neurons, we next assessed the impact of the NaV1.7 selective blocker Pro-Tx II (Schmalhofer et al., 2008) on TTX-S current in human DRG neurons. Little if any detectable suppression of TTX-S currents was observed in response to concentrations as high as 30 nM in the 27 neurons tested (Figure 12B). While we tested three different lots of the toxin on human neurons, we did not confirm the efficacy of any of these on rat DRG neurons. Thus, because of the notorious difficulty in working with large peptides, it is possible that negative results with this toxin were due to experimental errors. Nevertheless, we did test a fourth lot of the toxin, handled identically to the previous three, on rat DRG neurons, and observed a 50.5 ± 13.7% reduction in TTX-S current in the five neurons tested with 10 nM Pro-Tx II (Figure 12B).

The contribution of NaV1.7 to TTX-S currents in human DRG neurons.

(A) Ramp currents were evoked with a voltage-clamp protocol consisting of a depolarization from −100 mV to 0 mV over 500 ms following a 500 ms voltage-step to −90 mV. Peak inward current evoked in response to the ramp was seen at −5.9 ± 0.6 mV (n = 74). (B) Top traces: TTX-S currents evoked at 0 mV, isolated as described in Figure 1 from a human DRG neuron with 500 ms voltage-steps to −90 mV and −35 mV prior to the voltage-step to 0 mV, to activate total current and inactivate TTX-S current, respectively. Currents were evoked before and after 10 nM Pro-Tx II. Comparable data were obtained in 8 other neurons tested. Bottom traces: The same concentration of toxin blocked a fraction of TTX-S currents in a rat DRG neuron, isolated with 500 ms voltage-steps to −90 mV and −40 mV prior to the voltage-step to −5 mV. While the fraction of block was variable, a fraction of TTX-S current was blocked in the five other neurons studied. (C) Pre-incubating human DRG neurons with 100 nM PF-05089771 for 30 min prior to study (left traces) was associated with an almost complete block of TTX-S currents (top traces), as well as a significant reduction in TTX-R currents (bottom traces). TTX-S current isolated from TTX-R current as described in Figure 1, with pre-pulse potential to −90 mV and −35 mV, for both the control neuron (left traces) and the treated neuron (right traces). Control neurons were run in parallel with pre-incubation times in bath solution identical to the neurons pre-incubated with PF-05089771. (D) The inhibition of both TTX-S and TTX-R currents by PF-05089771 was concentration dependent, were the average current density in control neurons run in parallel with treated neurons was used to assess the magnitude of current block. Pooled data are from three donor, with 5–6 neurons per concentration and 13 neurons in the control group. Data were fitted with a modified Hill equation, which yielded an IC50 of 6 nM for block of TTX-S current and 47 nM for block of TTX-R current.

https://doi.org/10.7554/eLife.23235.025

While the ramp current and Pro-Tx II data argued against the presence of NaV1.7 in human DRG neurons, recent results with a putatively NaV1.7 selective small molecule blocker, PF-05089771 suggested that the majority of TTX-S current in human DRG neurons is carried by NaV1.7 (Alexandrou et al., 2016). Thus, in the fourth set of experiments, we assessed the impact of PF-05089771 on human DRG neurons. Consistent with previous results (Alexandrou et al., 2016), we observed complete inhibition of TTX-S currents in human DRG neurons following a 30 min pre-incubation with PF-05089771 with concentrations as low as 30 nM (Figure 12C and D). Concentration-response data indicated that the IC50 for TTX-S current block was ~6 nM, close to that previously reported (Alexandrou et al., 2016). However, in contrast to the previous study of PF-05089771 on human DRG neurons, which was performed in the presence of A-803467, we assessed the impact of this compound on TTX-R currents. In striking contrast to the heterologous expression data indicating NaV1.8 is resistant to PF-05089771 at concentrations as high as 10 μM, we observed a significant reduction in TTX-R currents in neurons treated with this compound. In fact, the calculated IC50 for inhibition of TTX-R currents in human DRG neurons was ~50 nM (Figure 12C and D).

Discussion

The purpose of the present study was to compare the properties of voltage-gated Na+ currents in DRG neurons from rat and human. As with currents in the rat, the major types of current, those classically described as the rapidly activating and rapidly inactivating TTX-S current, and the more slowly activing and slowly inactivating TTX-R current, were well isolated with voltage clamp protocols. In fact, an issue potentially relevant to the interpretation of our pharmacological experiments, voltage protocols were used to isolate these currents in all but our initial characterization experiments. Both types of current were potentiated by inflammatory mediators, although TTX-S currents were increased in only a subpopulation of neurons in which TTX-R currents were increased. The biophysical properties of TTX-R currents in human DRG neurons were qualitatively similar to those in rat DRG neurons, with small, but significant differences in the V0.5 of current inactivation and activation. However, in contrast to TTX-R current in rat DRG neurons, there was little evidence of use-dependent inactivation of currents in human DRG neurons. There were also significant differences in the extent of the block produced by the putatively NaV1.8 selective blocker A-803467, as well as the use-dependent block produced by lidocaine, both of which were significantly smaller on currents from human DRG neurons. As with TTX-R currents, TTX-S currents in rat and human DRG neurons were qualitatively similar. However, the V0.5 of current activation was more hyperpolarized, a larger fraction of current recovered rapidly from inactivation, and the currents activated significantly faster in human DRG neurons. The potency of TTX was significantly lower against currents in human DRG neurons. The lidocaine-induced use-dependent block of TTX-S currents was also smaller in human DRG neurons. Finally, there was no evidence of low threshold ramp currents in human DRG neurons, the TTX-S currents were resistant to the NaV1.7-selective blocker Pro-Tx II, and while TTX-S currents were blocked by PF-05089771, this putatively selective small molecule inhibitor of NaV1.7 also blocked TTX-R currents in human DRG neurons. These results have important implications for drug development, as well as our understanding of pain mechanisms.

With respect to drug development, the results of the present study are consistent with those from previous studies indicating that both expression systems and species differences may have a significant influence on the pharmacological properties of the protein in question. This may be particularly true for VGSC because of extensive post-translational modifications (Laedermann et al., 2015). Potentially more problematic is that these modifications may not only be cell type specific, but specific to location(s) within a given cell (Harriott and Gold, 2008). In the context of pharmacology, where side effects are often due to off-target actions of a drug, any decrease in relative potency increases the likelihood of side effects. We observed a loss of selectivity in human DRG neurons for both A-803467 and PF-05089771. In the case of the former, we observed a decrease in TTX-S current in human DRG neurons in response to 1 μM A-803467 which was close to that of TTX-R currents (data not shown), while in the case of the latter, we observed a block of TTX-R currents with an IC50 less than an order of magnitude higher than that for block of TTX-S currents. These results suggest that despite the very high degree of selectivity observed with these compounds on heterologously expressed VGSC, the therapeutic window for A-803467 would likely have been quite limited in the clinical setting, while any therapeutic effect of PF-05089771 may also reflect block of NaV1.8 channels. Similarly, the demonstration in heterologous expression systems that the potency of VGSC blockers may be state-dependent served as the rationale for the development of state-dependent VGSC blockers for the treatment of pain (Dick et al., 2007). That is, this observation suggested it may be possible to use such drugs to preferentially block spontaneous or aberrant activity, and thereby reduce spontaneous pain while preserving normal sensation. However, our observation that TTX-R currents in human DRG neurons, currents thought to play a critical role in spike initiation in nociceptive afferents, are subject to little if any use-dependent block in either the absence or presence of a prototypical use-dependent VGSC blocker could explain the limited therapeutic efficacy of local anesthetics (at least when delivered at concentrations below that needed to block action potential propagation) in pain patients (Finnerup et al., 2015).

There are several implications of the results of the present study with respect to our understanding of pain mechanisms. First, consistent with results from another recent study of human DRG neurons, our data confirm the utility of this model for the analysis and/or confirmation of second messenger systems implicated in the sensitization of putative nociceptive afferents. That is, it was recently demonstrated that application of inflammatory mediators to isolated human sensory neurons resulted in an increase in excitability (Davidson et al., 2014). Furthermore, activation of group II metabotropic glutamate receptors, shown to block the inflammatory mediator-induced sensitization of mouse sensory neurons, was also shown to block the sensitization of human sensory neurons (Davidson et al., 2016). Results of the present study confirm previous results from the study of rodent sensory neurons, suggesting that increases in both TTX-R (Gold et al., 1996) and TTX-S (Cardenas et al., 1997) currents are likely to contribute to this increase in excitability. The observation that TTX-S currents were only increased in a subpopulation of neurons in which TTX-R currents were increased, is interesting for at least two reasons. One is that it suggests that there is relatively tight coupling between G-proteins and effector molecules, even within the isolated sensory neurons. This would underscore the importance of getting a therapeutic intervention to the right place. Another reason the observation about the modulation of TTX-S and R currents is interesting is that it suggests that it may be necessary to reduce both TTX-R and TTX-S currents to achieve maximal pain relief.

Another implication of our results concerns assumptions about the relationship between cell body diameter and afferent function. That is, there is compelling evidence to suggest that at least for cutaneous afferents, nociceptors are enriched in the subpopulation with a small cell body diameter while low-threshold afferents are enriched in a subpopulation of neurons with a large cell body diameter (Lawson, 2002; Djouhri et al., 2003). Consistent with association between cell body size and function, NaV1.8 is not only enriched in nociceptive afferents (Djouhri et al., 2003), but is preferentially expressed in sensory neurons with a small cell body diameter. In contrast, neither the size distribution nor the distribution of TTX-R currents among human DRG neurons suggests that there is a comparable relationship between cell body size and function in the human as has been documented in the rodent. Given that we did not assess the currents present in the very largest human DRG neurons, it is possible that we missed the subpopulation of putative non-nociceptive afferents comparable to that in the rodent, but this would still suggest TTX-R currents are far more widely distributed in human neurons than in the rodent. In this regard, it is also possible that TTX-R currents play a more important role in non-nociceptive human than rodent neurons. However, our inflammatory mediator data, as well as previous excitability data from human DRG neurons (Davidson et al., 2014) suggests the association between TTX-R currents and the nociceptive phenotype is preserved in human DRG neurons. Minimally, these data suggest that additional criteria will be needed to identify putative subtypes of human sensory neurons in future studies.

A third implication of our results concerns the present model of spike initiation and the emergence of sustained activity following injury in putative nociceptive neurons. That is, data from the study of rodent sensory neurons suggests that spike initiation involves the initial activation of NaV1.7 followed by the activation of NaV1.8 (Cummins et al., 1998). This suggested NaV1.7 as a viable target to attenuate nociceptor activity. However, our data from human DRG neurons suggest that while the threshold for activation of TTX-S currents is more negative than that of TTX-R currents, the failure of TTX to block ramp currents in the nine neurons tested (data not shown) suggests that ramp currents are dominated by TTX-R currents. This suggests that TTX-R currents are responsible for both spike initiation and the action potential over-shoot in human DRG neurons. Similarly, while use-dependent block of TTX-R currents in rodent sensory neurons has been argued to contribute to the slow adaptation observed in nociceptive afferent in response to prolonged stimulation (Choi et al., 2007), the absence of use-dependent block of TTX-R currents in human DRG neurons would enable these channels to underlie sustained neural activity such as that associated with ongoing pain. Furthermore, the relatively fast and complete recovery from inactivation observed for human TTX-S currents suggests that even in the absence of injury, these currents would be able to contribute to a relatively high level of sustained activity. This is in contrast to current models of injury-induced changes in VGSC expression, where the upregulation of NaV1.3, a channel with a relatively rapid rate of recovery from inactivation (Cummins and Waxman, 1997), is thought to contribute to the increase in sensory neuron excitability observed following peripheral nerve injury.

Our data on the channel subunits underlying the TTX-S current in human DRG neurons also has important implications given the recent focus on NaV1.7 as a potential therapeutic target for the treatment of pain based on the human channelopathy data (Dib-Hajj et al., 2013). One interpretation of our PF-05089771 data is that NaV1.7 underlies most, if not all of the TTX-S current in human DRG neurons. This interpretation would be consistent with that proposed by Alexandrou and colleagues based on the selectivity of this compound on human Na+ channel subtypes expressed in HEK293 cells, as well as the impact of this compound on TTX-S current in human DRG neurons (Alexandrou et al., 2016). If this interpretation was correct, however, it would suggest that the biophysical properties of NaV1.7 in human DRG neurons, at least with respect to the relatively slow entry into an inactivated state, are different from that of human NaV1.7 channels expressed in HEK293 cells, or NaV1.7 channels in rodent sensory neurons (Cummins et al., 1998). That is, because closed state inactivation develops so much more slowly in heterologously expressed NaV1.7 channels than other subtypes such as NaV1.4, NaV1.7 was proposed to account for the large low threshold TTX-S current evoked with ramp depolarization of small rodent DRG neurons (Cummins et al., 1998). However, only high threshold ramp currents were evoked in human DRG neurons that were resistant to TTX. Furthermore, the interpretation that NaV1.7 is the dominant channel subtype underlying TTX-S currents in human DRG neurons would also suggest that the potency of Pro-Tx II is different for the block of NaV1.7 in human DRG neurons than in HEK 293 cells (Schmalhofer et al., 2008) or rodent sensory neurons (Laedermann et al., 2014).

An alternative interpretation of our results is that NaV1.7 may not be the dominant subunit underlying TTX-S currents, let alone significantly contribute to these currents in human DRG neurons. That is, while it is possible that PF-05089771 retains its selectivity for NaV1.7 over other natively expressed TTX-S channels, our observation that this compound blocks TTX-R currents in human DRG neurons with an IC50 of ~50 nM, yet has no activity at NaV1.8 channels in HEK293 cells at concentrations as high as 10 μM (Alexandrou et al., 2016) raises the possibility that there is a more generalized loss of selectivity of this compound against Na+ channels present in their native environment. Importantly, while the channel block produced by PF-05089771 was shown to be highly state-dependent in heterologous expression experiments, blocking inactivated channels with a potency almost four orders of magnitude higher than that of channels in a resting or closed state (Alexandrou et al., 2016), this property is unlikely to account for the apparent block of TTX-R currents observed in human DRG neurons. That is, the pre-pulse potential used to evoke TTX-R currents would have left the majority of channels underlying the TTX-R current in a closed or resting state (i.e., fully available for activation), and therefore should have decreased, rather than increased the potency of this compound. A relative dearth of NaV1.7 in human DRG neurons would also account for the absence of a low threshold TTX-S ramp current, as well as the absence of channel block with ProTx-II. A limited expression of NaV1.7 in a subpopulation of human DRG neurons would also explain why the inflammatory mediator-induced increase in TTX-S current was only detected in a subpopulation of neurons. It is important to point out that even in rodent sensory neurons, other TTX-S channel subunits are expressed at levels comparable to that of NaV1.7, at least in subpopulations of neurons. For example, RNAseq analysis of mouse DRG neurons suggests that the number of copies of NaV1.6 (31.1) is comparable to that of NaV1.7 (54), at least in the subpopulation of neurons not in the TRPV1 lineage (Goswami et al., 2014). Comparable results were obtained with a quantitative PCR analysis of single DRG neurons, where in those larger than 30 μm in diameter, NaV1.6 expression was comparable to that of NaV1.7 (Ho and O'Leary, 2011). It is also worth noting that despite evidence that NaV1.7 is distributed throughout sensory neurons (Black et al., 2012), recent evidence suggests that the contribution of NaV1.7 to even rodent sensory neurons may have been overestimated. That is, while the activity of the putatively NaV1.7 selective spider venom peptide Pn3a on native TTX-R currents was not well described, results with this peptide suggests that NaV1.7 does not even account for 50% of the TTX-S current in the majority of rat DRG neurons (Deuis et al., 2017). It should also be noted that recent data from the NaV1.7 null mutant mice suggest that this subunit may, in fact play a more important role in mediating Na+ influx and associated changes in gene expression than it does in the electrical properties of these neurons (Minett et al., 2015). And while compensation may account for the relatively limited impact of the NaV1.7 knock-out on TTX-S currents in mouse small diameter DRG neurons, the ~25% decrease in TTX-S current density observed in these neurons is consistent with a relatively limited contribution of NaV1.7 to the TTX-S current (Nassar et al., 2004). More relevantly, the suggestion that NaV1.7 contributes little to the TTX-S current in human sensory neurons is consistent with the relatively limited impact of gain of function mutations on the excitability of human nociceptive afferents (Namer et al., 2015).

While several potentially important differences were observed in the Na+ currents in human and rat DRG neurons, it is important to consider the extent to which these differences were due to experimental variables rather than species differences. Neurons were collected from both male and female human donors with an average age of 45 years, at least an hour after cross clamp. In contrast, neurons were obtained from relatively young (~60 day old) male rats that were deeply anesthetized at the time of tissue collection. Unfortunately, we were unable to rule out the time between cross-clamp and tissue collection as a factor contributing to the differences observed, as preliminary results indicated that the viability of rat DRG neurons fell off precipitously 15 min after death, even in rats perfused with ice-cold saline. However, we suggest that neither of the other two main factors were likely to contribute to the differences observed, as there was no detectable influence of either age (at least comparing neurons from young (<30) and old (>65) donors) or donor sex on the biophysical properties of Na+ currents assessed (data not shown). It is also possible that the larger currents and larger cell body size of human DRG neurons resulted in a decrease in clamp control that contributed to the hyperpolarizing shift in both TTX-R and TTX-S current G-V curves as well as the faster rates of current activation observed in human neurons. Arguing against this possibility, however, is the fact that while the slopes of the G-V curves for both current types were steeper in human than in rat neurons (4.8 ± 0.1 mV and 5.4 ± 0.2 mV vs 5.0 ± 0.1 mV and 6.3 ± 0.1 mV, for TTX-R and TTX-S currents in human and rat neurons, respectively), these small differences could not account for the observed differences in the V0.5 of activation, nor are they consistent with a loss of clamp control, which would be predicted to result in a much larger decrease in the slope of the G-V curve. In addition, differences in the V0.5 of activation persisted when data from only the smallest human and largest rat neurons were compared. For example, the V0.5 of current activation for TTX-R in the overlapping subpopulation of rat and human sensory neurons were −2.1 ± 0.4 mV and −10.8 ± 1.7 mV, respectively.

We also suggest that it is unlikely that the recording conditions used, which were necessary to both minimize voltage-gated Ca2+ currents while maximizing patch stability had a significant influence on the biophysical properties of the currents recorded. That is, the properties of the currents observed in rat neurons recorded under conditions identical to those used for the study of human neurons, were comparable to those we have previously observed (Gold et al., 2003; Flake et al., 2004; Vaughn and Gold, 2010). That said, we have always used a relatively low concentration of extracellular Na+ to help maintain clamp control, and minimized Ca2+ current contamination with a reduction of extracellular Ca2+ and/or the use of extracellular Ca2+ channel blockers such as Cd2+. In contrast, others, who have reported considerably more hyperpolarized potentials for the activation and inactivation of both TTX-S and TTX-R currents in rat DRG neurons have used higher concentrations of extracellular Na+ as well as intracellular solutions containing fluoride (Cummins and Waxman, 1997; Roy and Narahashi, 1992; Rush and Elliott, 1997). It remains to be determined whether recording conditions have comparable influences on the biophysical properties of Na+ currents in human DRG neurons. Finally, it is important to point out that while we attempted to control for the impact of state-dependent properties of the compounds tested, our pharmacological results should be interpreted with caution because of the potentially confounding interaction between these properties and the voltage protocols used to isolate TTX-S from TTX-R currents.

In summary, we have described important similarities and differences between human and rat dorsal root ganglion neurons with respect to the biophysical and pharmacological properties of the two major classes of voltage-gated Na+ current. We would argue that the similarities were sufficient to justify the continued use of rodent sensory neurons as a model system with which to explore the potential contribution of a variety of channels critical to both pain and analgesia. However, we would also argue that the differences were sufficient to contribute to what continues to be a tremendously high failure rate in the development of novel analgesics. Use of human DRG neurons for the screening of potentially novel therapeutics targeting the primary afferent and/or proteins in the primary afferent may help rectify this situation, by enabling the identification of compounds unlikely to achieve the intended effect prior to the initiation of expensive clinical trials.

Materials and methods

Human tissue

Request a detailed protocol

L4 and L5 DRG were collected from organ donors with the consent of family members for the use of their loved one’s tissue for research purposes. The protocol for the collection and study of tissue from organ donors was approved by the University of Pittsburgh Committee for Oversight of Research and Clinical Training Involving Decedents. As previous data on the biophysical and pharmacological properties of Na+ currents in human DRG neurons was extremely limited, no a priori analyses were performed to estimate the number of donors needed to complete this study. Rather, we studied tissue from each donor entered into the study with the same basic protocol needed to extract basic features of Na+ current types from each neuron, and then added additional analyses with the goal of obtaining data for each endpoint with neurons from at least three donors. Post-hoc analysis was then performed to rule out an influence of sex or age, the only major demographic features sufficiently represented in our data-set to perform such an analysis, on any of the biophysical properties determined for the majority of neurons from all donors.

Rat tissue

Request a detailed protocol

Adult (250–320 g) male Sprague-Dawley rats (Envigo, Indianapolis, IN)) were used for all experiments. Rats were housed two per cage in a temperature and humidity controlled, Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC) accredited animal housing facility on a 12 hr:12 hr light:dark schedule. Food and water were available ad libitum. All procedures were approved by the University of Pittsburgh Institutional Animal Care and Use Committee and performed in accordance with National Institutes of Health guidelines for the use of laboratory animals in research. The number of rats used in this study was based on estimates made from variability observed in our previous biophysical and pharmacological analysis of Na+ currents in rat sensory neurons as well as previous experience with both the amount of time needed to complete each protocol and consequently the number of neurons that could be studied from each rat in the time window in which neurons were considered ‘acutely’ dissociated. As our previous data indicates that there is considerably more heterogeneity between neurons within the same rat that between rats, the use of three rats for any given endpoint enables us to detect any potential issues associated with any single preparation of neurons. Thus, neurons from at least three rats were used for each endpoint.

Isolation and plating of human sensory neurons

Request a detailed protocol

DRG were obtained from organ donors following collection of tissue needed for transplantation purposes as previously described (Zhang et al., 2015). Following surgical isolation of ganglia, they were placed in ice cold collection media composed of 124.5 mM NaCl, 5 mM KCl, 1.2 mM MgSO4, 1 mM CaCl2, and 30 mM HEPES, and had been filter sterilized after the pH had been adjusted to 7.35 with NaOH. As with our previous study, the time between cross-clamp and the harvest of ganglia was generally under 45 min, and the time between tissue collection and initiating the dissociation protocol was less than three hours. The same protocol and combination of solutions was employed in the present experiments, as described previously, except that the complete media used for plating the neurons consisted of basal media (500 ml bottle of L-15 media containing: 60 mg imidizole, 15 mg aspartic acid, 15 mg glutamic acid, 15 mg cystine, 5 mg β-alanine, 10 mg myo-inositol, 10 mg cholineCl, 5 mg p-aminobenzoic acid, 25 mg fumaric acid, 2 mg vitamin B12 and 5 mg of lipoic acid (which was first dissolved in methanol at a concentration of 1g/2.5 ml)) diluted with fetal bovine serum (1:10) and then supplemented to yield a final concentration of 50 ng/ml nerve growth factor (NGF 2.5S, Invitrogen), 0.3 mg/ml glutamine (Invitrogen), 4.5 mg/ml glucose (Sigma-Aldrich), 0.525 mg/ml ascorbic acid (Sigma-Aldrich), 2.4 μg/ml glutathione (Invitrogen), and 0.2% (w/v) NaHCO3 (Sigma-Aldrich). Cells were plated onto poly-L-lysine coated glass coverslips (Invitrogen) placed in 35 mm culture dishes and stored in a CO2 (5%) incubator at 37°C for 2–4 hr prior to flooding the culture dishes with Complete Media. Neurons were studied within 12 hr (acute) of plating.

Isolation and plating of rat sensory neurons

Request a detailed protocol

Adult rat sensory neurons were surgically obtained, enzymatically treated and mechanically dissociated as previously described (Lu et al., 2006). Cells were also plated on poly-L-lysine coated glass coverslips (Invitrogen), which were placed in 35 mm culture dishes and stored in a C02 (3%) incubator at 37°C for 2 hr prior to flooding with Complete Media. Neurons were studied within 12 hr of plating.

Whole cell patch clamp

Request a detailed protocol

Whole-cell patch-clamp recordings were performed with an Axopatch 200B controlled with pClamp (v 10.2) software (Molecular Devices, Carlsbad, CA) used in combination with a Digidata 1320A A/D converter (Molecular Devices). Unless otherwise noted, data were acquired at 20 kHz and filtered at 5 kHz. Borosilicate glass (WPI, Sarasota, FL) electrodes were 0.75–2 MΩ when filled with an electrode solution that contained (in mM): Cs- Methansulphonate 100, TEA-Cl 40, NaCl 5, CaCl2 1, EGTA 11, HEPES 10, Mg-ATP 2, and GTP 1; pH was adjusted to 7.2 with Tris-base and osmolality was adjusted to 310 mOsm with sucrose. The bath solution consisted of (in mM): NaCl 35, Choline-Cl 65, TEA-Cl 30, CaCl2 0.1, MgCl2 5, CdCl2 0.1, HEPES 10, and glucose 10; pH was adjusted to 7.4 with Tris-Base, and the osmolality adjusted to 320 with sucrose.

Capacitative currents were minimized with amplifier circuitry. Series resistance compensation was always employed, and if it was not possible to achieve compensation greater than 75%, neurons were not included for further analysis. Similarly, if estimated voltage errors were greater than 5 mV, data were not included for further analysis, where voltage errors were estimated based on the peak inward current across the uncompensated series resistance. Data was also excluded from neurons in which the holding current was >500 pA. Preliminary experiments indicated that it was rarely possible to maintain clamp control of currents evoked in neurons in culture for more than 24 hr, even with extracellular Na+ reduced to 20 mM. Thus, all data included in this data set were from neurons <24 hr in culture. A p/−4 leak subtraction was employed from a holding potential of −80 mV. Steady-state availability curves were determined for each neuron in which both TTX-S and TTX-R currents were detected, so as to confirm the pre-pulse potential amplitude necessary for relief of steady-state inactivation, as well as the pre-pulse potential at which TTX-S currents were completely inactivated (Figure 1). A 500 ms pre-pulse was used to drive changes in channel availability, followed by a voltage step to a potential that enabled visualization of both TTX-S and TTX-R components of the total current (generally between −5 and 0 mV). The pre-pulse was increased by 10 mV increments every 5 s. The pre-pulse potentials needed for full channel availability and for inactivation of TTX-S currents were used for the generation of current-voltage (I-V) curves for total current, and for TTX-R current. I-V curves were generated with a series of 15 ms test pulses between −60 and +40 mV, evoked every 5 s. It was then possible to subtract TTX-R currents from total current to obtain TTX-S currents in isolation (Figure 1). We subsequently confirmed that currents isolated in this manner were identical to those isolated with TTX (Figure 1).

Reagents

Unless otherwise noted, all reagents were obtained from Sigma-Aldrich. TTX was dissolved in distilled water (dH2O) as a 1 mM stock solution, and stored at 4°C until use. A-803467 was dissolved in dimethylsulfoxide (DMSO) at a concentration of 10 mM, immediately before use, and subsequently diluted in bath solution. Bradykinin was dissolved in 0.1% acetic acid as a 10 mM stock solution, and stored at −20°C until use. Prostaglandin E2 was dissolved in ethanol as a 10 mM stock solution and stored at −20°C until use. Histamine was dissolved in dH2O as a 10 mM stock solution and stored at −20°C until use. Lidocaine was prepared as a 10 mM stock solution with 5 mM Na-HEPES buffered dH2O, with a pH adjusted to 7.0 with TEA-OH, and stored at −20°C until use.

Data analysis

Request a detailed protocol

Amplifier circuitry was used to estimate membrane capacitance which was used to estimate current density. Na+ current reversal potential (Vr) was determined for each neuron from the linear phase of the I-V curve. Conductance-voltage curves (G-V) were determined by dividing current evoked at each test potential (Vt) by the driving force on the current (Vr – Vt). A two-pulse protocol was used to determine the recovery from inactivation, where the first pulse to 0 mV, was used to completely inactivate Na+ currents, and the second pulse to −50 mV or −90 mV of increasing duration was used to drive recovery of TTX-R and TTX-S currents from inactivation, respectively. The second pulse was followed by a final test pulse to 0 mV. Steady-state inactivation and G-V curves were fitted with modified Boltzmann equations so as to determine maximal conductance (Gmax), the voltage at which current were either half inactivated or half activated (V0.5), as well as the slopes of the two curves. Recovery from inactivation data were fitted with a double exponential. Current activation, inactivation, and when possible, deactivation, were determined with a single exponential fitted to the rising phase (activation), falling phase (inactivation), and tail currents (deactivation) of currents evoked with an I-V protocol. Use-dependent block was determined with 20 pulses to 0 mV delivered at 1, 2, or 5 Hz. Fractional use-dependent block was estimated by dividing current evoked after the 20th pulse by the current evoked after the first. Finally, concentration-response curves were fitted with a modified Hill equation to enable estimation of the concentration needed to block 50% of evoked current.

A t-test was used for statistical comparisons between human and rat data, where a difference with p<0.05 was considered statistically significant. If comparisons were made between human and rat across voltage, a two-way ANOVA was used. Prior to use of a t-test of ANOVA, data to be compared was assessed for normality and equal variance. Linear regression and a one-way ANOVA were used to assess the potential impact of age on evoked currents.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
    5HT4 receptors couple positively to tetrodotoxin-insensitive sodium channels in a subpopulation of capsaicin-sensitive rat sensory neurons
    1. CG Cardenas
    2. LP Del Mar
    3. BY Cooper
    4. RS Scroggs
    (1997)
    Journal of Neuroscience 17:7181–7189.
  6. 6
  7. 7
    Slow closed-state inactivation: a novel mechanism underlying ramp currents in cells expressing the hNE/PN1 sodium channel
    1. TR Cummins
    2. JR Howe
    3. SG Waxman
    (1998)
    Journal of Neuroscience 18:9607–9619.
  8. 8
    Downregulation of tetrodotoxin-resistant sodium currents and upregulation of a rapidly repriming tetrodotoxin-sensitive sodium current in small spinal sensory neurons after nerve injury
    1. TR Cummins
    2. SG Waxman
    (1997)
    Journal of Neuroscience 17:3503–3514.
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
    Redistribution of na(V)1.8 in uninjured axons enables neuropathic pain
    1. MS Gold
    2. D Weinreich
    3. CS Kim
    4. R Wang
    5. J Treanor
    6. F Porreca
    7. J Lai
    (2003)
    Journal of Neuroscience 23:158–166.
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
    Differential regulation of tetrodotoxin-resistant (PN3) and tetrodotoxin-sensitive (PN1) sodium channels following peripheral nerve injury in rats
    1. JC Hunter
    2. E Tzoumaka
    3. M Haraguchi
    4. L Sangameswaran
    5. K Wong
    6. KR Gogas
    7. SD Novakovic
    (1997)
    Society Neuroscience Abstracts 23:912.
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
    The adult rat brain beta 1 subunit modifies activation and inactivation gating of multiple sodium channel alpha subunits
    1. DE Patton
    2. LL Isom
    3. WA Catterall
    4. AL Goldin
    (1994)
    The Journal of Biological Chemistry 269:17649–17655.
  41. 41
  42. 42
  43. 43
    Differential properties of tetrodotoxin-sensitive and tetrodotoxin-resistant sodium channels in rat dorsal root ganglion neurons
    1. ML Roy
    2. T Narahashi
    (1992)
    Journal of Neuroscience 12:2104–2111.
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
    NaV1.7 as a pain target - From gene to pharmacology
    1. I Vetter
    2. JR Deuis
    3. A Mueller
    4. MR Israel
    5. H Starobova
    6. A Zhang
    7. LD Rash
    8. M Mobli
    (2017)
    Pharmacology & Therapeutics, 172, 10.1016/j.pharmthera.2016.11.015, 27916648.
  49. 49
    Type III sodium channel mRNA is expressed in embryonic but not adult spinal sensory neurons, and is reexpressed following axotomy
    1. SG Waxman
    2. JD Kocsis
    3. JA Black
    (1994)
    Journal of Neurophysiology 72:466–470.
  50. 50
  51. 51
  52. 52

Decision letter

  1. Indira M Raman
    Reviewing Editor; Northwestern University, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Voltage-Gated Na+ Currents in Human Dorsal Root Ganglion Neurons" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Gary Westbrook as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Bruce P Bean (Reviewer #1); Stephen Cannon (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This manuscript describes the biophysical properties and pharmacology of voltage-dependent sodium channels in dorsal root ganglion neurons from adult humans and provides a comparison of the characteristics of those sodium currents to the more common experimental model of rats. The work identifies both similarities and differences between the human and rat preparations.

Essential revisions:

Both reviewers found the work to be very important, informative, and well done. Their compiled favorable comments are quoted verbatim below in "General Comments." Several key points came up that require addressing, however, relating to pharmacology, which was seen as the weakest area of the manuscript, and a few other aspects of voltage protocols and cell selection. The points that require addressing are summarized here and elaborated upon in the reviewers' words under "Detailed Points for Essential Revisions."

1) Verify that the very surprising results indicating that the primary TTX-S channels in humans (a) are not 1.7 and (b) are not blocked by the broad-spectrum Protoxin II are not influenced by plausible sources of error, e.g., those relating to complications of peptide perfusion.

2) Omit or address the very unexpected result that TTX increases ramp current, which also raises questions about the efficacy/specificity of drug perfusion.

3) Clarify voltage protocols used and (a) use the information to strengthen/clarify/curtail conclusions based on pharmacology and (b) validate (test and/or rationalize) the prepulse approach as opposed to a pharmacological approach to separate the currents in humans.

4) Test and/or clarify ambiguities arising from size differences and size-based selection to verify that differences between species weren't based on clamp artifacts from the different sizes of neurons.

5) Address or omit the deactivation time constant measurements, about which specific questions were raised.

We recognize that the availability of human tissue may be limited. If it is not possible or practical to address the issues directly, a (less desirable) alternative is to tone down or qualify the conclusion that Nav1.7 may not contribute much to the TTX-S current, indicating that further pharmacological work is necessary for a definitive conclusion.

General comments

This is important experimental information, because almost all previous knowledge about the properties and functional roles of sodium channels in DRG neurons has come either from studies on rodent DRG neurons or on cloned human channels in heterologous expression systems. Since an important goal of the research area, especially on the pharmacology, is to help guide development of new treatments for pain, possible differences in the pharmacology of native human channels compared to those in rodent neurons or to human channels in heterologous systems are of great interest. Consequently, the surprising results in this manuscript suggesting that the pharmacological properties of both TTX-resistant and TTX-sensitive populations of sodium channels in human neurons differ from those in rat neurons give the manuscript considerable interest. Beyond this, the work shows, very surprisingly, that obtaining electrophysiologically healthy neurons acutely-dissociated from human DRGs is actually easier (or at least has a higher success rate once the DRGs are obtained) than with rat DRG neurons of fully adult animals. The technical quality of the voltage-clamp studies in the manuscript are excellent and the volume of data for many of the figures leaves no doubt of the possibility of obtaining statistically impressive data from human neurons. Thus an important secondary message of the manuscript will be to encourage others to study human DRG neurons.

An excellent feature of the work is a detailed comparison with the properties of rat DRG neurons studies with the same protocols. This gives confidence that the differences seen are genuine and not the result of slightly different protocols (although as noted below, it would helpful to have a statement about the extent to which the experiments were performed in parallel, especially in regard to the pharmacologic experiments).

In general, the biophysical description of both the TTX-R and TTX-S components of sodium current are thorough and convincing (although as noted below, it would be helpful to have a more complete description of some of the voltage protocols in some cases).

The repertoire of voltage gated sodium channels (VGSCs) in dorsal root ganglia (DRG) and their modulation by inflammatory mediators are important for the perception of pain and the hypersensitivity of pain syndromes associated with tissue injury. Much of this knowledge has come from studies in rodent DRG neurons or from human VGSCs expressed in heterologous systems. This practical constraint leaves open the question of whether the VGSC properties of human DRG neurons are comparable. Zhang and colleagues have recorded Na currents from human DRG neurons acutely isolated from organ donors. In broad terms, the two major classes of Na currents (TTX-R and TTX-S) were both present and comparable in rat and human DRG neurons. Several differences, however, were identified and these may have important consequences for the pathogenesis of pain syndromes and for therapeutic strategies. For example, the putative NaV1.8-selective blocker, A803467, did not cause a detectable reduction of human TTX-R currents until 1 μm was applied, whereas rat TTX-R currents are much more sensitive to block. Human TTX-R currents also had no evidence of use-dependent cumulative inactivation or use-dependent lidocaine block (1-5 HZ) whereas rat TTX-R current had substantial use dependence for both. Differences of intrinsic gating were also observed. Most striking were faster activation and recovery from inactivation for human TTX-S currents as compared to rat TTX-S. Overall, this is a carefully designed and conducted study that demonstrates while the VGSC properties in DRG neurons are overall similar for rat and human, there are clear differences that may have important implications for development of therapeutics.

Detailed points for essential revisions:

1) The area in which the results in the manuscript could be stronger is the most important – the conclusion that the TTX-S channels in the human neurons are not primarily carried by Nav1.7 channels. If true, this is very important, since multiple companies are developing Nav1.7 inhibitors with the assumption that these channels are critical for the function of nociceptors. Also, the conclusion that Nav1.7 do not contribute to the TTX-S currents is opposite to a previous study of human DRG neurons in which it was concluded that the majority of the TTX-S current was carried by Nav1.7 channels, based on inhibition of 75% of the current by 30 nM PF-05089771, a fairly selective blocker of Nav1.7 channels. The authors point out this previous study used neurons in culture for up to 9 days, which could cause an upregulation of Nav1.7 channels – a valid point. They also suggest that the result could have been from imperfect selectivity of PF-05089771, which while possible seems a lot less likely, since Alexandrou et al., 2016 saw big effects of 30 nM PF-05089771, which would be expected to have minimal effects on any of the other channel types.

Here the authors present two pieces of evidence arguing against mediation of TTX-S by Nav1.7. One is the lack of effect of 10 nM Protoxin II. It seemed very puzzling that the authors chose protoxin II as a blocker, because a previous review of the properties of this toxin stated "ProTx-I and ProTx-II inhibit all sodium channel (Nav1) subtypes tested with similar potency." (Priest et al. Toxicon. 2007 49:194-201). This is certainly not clear from the presentation, where a reader would naturally infer that the toxin was used because it had selectivity for Nav1.7. The evidence against Nav1.7 would be far stronger if the authors had used PF-05089771, which is much more selective. This compound has been commercially available (from both Σ and Tocris) for some time.

Given the efficacy of protoxin-II against all sodium channels, the lack of effect is very surprising whatever channels underlie the TTX-S current. A possible experimental problem with potent peptide toxins of this type is that they can bind to plastic tubing and glassware and be difficult to remove, so that subsequent application appears to have no effect because of presence of the agent under "control" conditions. I raise that because the lack of any effect of the toxin is unexpected whatever channels the sodium currents are coming from and because we have experienced that insidious problem several times in my own lab with similar toxins – a problem that could only be resolved by completely replacing tubing and chambers.

2) The second piece of evidence that authors present against the contribution pf Nav1.7 channels is the most puzzling result in the paper – that TTX not only did not inhibit a current induced by ramp of voltage but actually enhanced the current and shifted its voltage dependence to more hyperpolarizing voltages. The shift of current to more hyperpolarized voltage makes no sense at all and suggests some sort of experimental problem. It would suggest that TTX enhances sodium current for small depolarizations which has never been described for any channels. At a minimum, the authors need to follow up the ramp experiment with step depolarization experiments to characterize this effect in more detail. Even if TTX simply had no effect on the ramp current, this is only weak evidence against mediation by Nav1.7. It could reflect for example faster or more complete closed-state inactivation in human Nav1.7 channels versus rodent Nav1.7 channels, which would be interesting. The authors did not compare closed-state inactivation in the human vs rat neurons so this seems like an open possibility.

3) It would be helpful if all the figures had clear statements of exactly what the voltage protocols were, which were not always clear to me. In Figures 47 on TTX-R current, it seems that the records were obtained from a steady holding voltage of -80 mV and a 500 ms prepulse, probably to -40 mV based on Figure 1, but this is not stated explicitly and the prepulse potential is not given in each case. Materials and methods says "The pre-pulse potentials needed for full channel availability and for inactivation of TTX-S currents were used for the generation of current-voltage (I-V) curves for total current, and for TTX-R current" which suggests that the same prepulse was not always used. The exact protocol in each figure should be stated. This is especially an issue for interpreting the effects of lidocaine, since the potency is so strongly sensitive to resting membrane potential. It seems possible that the potency of A-803467 could also be sensitive to the exact voltage protocol.

Additionally, the separation of TTX-R and TTX-S components by using a conditioning voltage pulse (instead of the more laborious method of TTX block) is a key aspect of the analysis and interpretation. Deficiencies in this separation technique, which may be different for rat versus human DRG neurons, could result in apparent differences or underestimate differences. With that in mind, how was it possible to determine the availability curve for TTX-R specific currents in Figure 4D? The conditioning pulse range extended from -60 mV to 0 mV and yet Figure 1 shows a prepulse of -35 mV is needed to isolate the TTX-R component (via inactivation of TTX-S). How was availability data obtained for conditioning voltages more negative than -35 mV?

A great feature of the work is the comparison to studies in rat neurons done in the same lab. If these were actually done in parallel, using for example exactly the same batches of pharmacological reagents and by the same experimenter, this might be mentioned explicitly as it further strengthens the point that the differences are genuine and not the result of differences in procedures or reagents.

4) A missing piece of information is exactly how the authors selected which cells to record from. From the fact that they all had TTX-R current, the experimenter presumably chose smaller-diameter cells to record from, but what exactly was the selection criterion? Do the authors think the cells all correspond to cell bodies of C-fibers? The authors have discussed the challenges of high Na current density and series resistance that often plague studies of VGSCs. Because human DRG soma are much larger than rat, the demands on clamp quality would be systematically more challenging in recordings from human than rat neurons. This might cause a systematic difference in the apparent gating behavior. The distribution of cell capacitance did, however, have a region of overlap around 50 pF. Perhaps a subset of rat and human DRG neurons with comparable size could be analyzed to provide reassurance that the observed species differences were still detectable in a population of similar-sized neurons?

5) The interpretation and discussion of VGSC deactivation was confusing (Figure 5E, which shows plots of deactivation tau versus voltage). First, deactivation time constants are usually presented as a function of the deactivation pulse potential, not the preceding voltage to activate the channel as shown in Figure 5E. However, from the text it seems that this is not a standard plot of tail time constant against voltage as one might assume at first but rather a measurement of the tail time constant at -60 mV following steps to different voltages. This should be made clear in the figure legend. Were there any measurements of deactivation time constants at other voltages? These would be helpful to someone who wanted to make a Hodgkin-Huxley type model from the data. Second, deactivation rates are usually measured over a range of test potentials. The more hyperpolarized ones are often regarded as more representative of "true" deactivation (O->C transition) but the fast kinetics are technically demanding. Responses at more depolarized potentials are more easily resolved, but may represent other state transitions (not just O -> C). The canonical view is that a true deactivation rate should be independent of the pulse potential used to open the channel. The atypical behavior in Figure 5E suggests a pulse to -60 mV was not measuring deactivation in isolation. Moreover, deactivation data are not presented for TTX-S currents. Perhaps the deactivation data should be omitted?

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Voltage-Gated Na+ Currents in Human Dorsal Root Ganglion Neurons" for further consideration at eLife. Your revised article has been evaluated by Gary Westbrook (Senior editor), a Reviewing editor, and two reviewers.

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

"TTX-S" and "TTX-R" are used as shorthand for prepulse subtraction, which is validated for control solutions in Figure 1. This procedure may complicate the interpretation of some of the results, however, particularly the pharmacology leading to the conclusion that 1.7 is not the primary TTX-S channel. Please edit the text (especially the Introduction and Discussion) for the following:

1) To make sure that readers are clear that "TTX-S" and "TTX-R" actually refer to currents inactivated at -40 mV and available at -40 mV,

2) To ensure that the prepulse subtraction procedure is kept in mind in interpreting the effects of the blockers, and

3) To limit or constrain claims that NaV1.7 is not the primary TTX-S current, as appropriate.

Please also include a statement about informed consent (as appropriate) for collection of human tissue.

The full comments pertaining to the requested revisions are below and have been agreed upon by all participants in the review.

General assessment and major comments:

The most important change in the manuscript is new data with the Pfizer Nav1.7 inhibitor PF-05089771, which the authors find inhibits the TTX-S current quite potently, with an IC50 of about 10 nM. In addition, the authors have clarified some the details of the voltage protocols and cell selection.

I was very surprised that with the new data showing potent block of the TTX-S current by PF-05089771, the authors still end up concluding that Nav1.7 channels make little contribution to the TTX-S current in human DRG neurons. If it were my data, my interpretation of the overall set of the results the authors present would be the opposite, that the TTX-S current is likely to be primarily from Nav1.7 channels based on its sensitivity to PF-05089771, but that it is (interestingly) less sensitive to Protoxin-II than rat Nav1.7 channels. The observation that PF-05089771 also seems to inhibit TTX-R currents in human DRG neurons at higher concentrations is interesting and important. So, like most small molecules, the selectivity of PF-05089771 is imperfect. That is important to know, and constitutes another of the many interesting points in the manuscript about different pharmacology of the native currents in human DRG neurons compared to rat DRG neurons or heterologously-expressed human channels. However, it is still true that PF-05089771 inhibits the TTX-S current with significantly higher potency, and about what is expected from the effects on cloned human Nav1.7 channels. So my interpretation would be that the results with PF-05089771 imply that the TTX-S current is mostly from Nav1.7, but that the compound is less selective than previously realized and at higher concentrations also inhibits native human Nav1.8 channels – a very important point that adds to the significance of the results in the manuscript.

To me, the flow of the revised manuscript now seems strange in that final set of experimental results in Results showing potent block by PF-05089771 gives the reader the impression that the TTX-S current is likely mainly Nav1.7, but then in the Discussion the first mention of this issue suggests the opposite without explaining why. The first part of the Discussion summarizing the pharmacology is very reasonable and straightforward (the paragraph starting, "With respect to drug development, the results of the present study are consistent with those from previous studies indicating that both expression systems and species differences may have a significant influence on the pharmacological properties of the protein in question." This paragraph nicely summarizes why it is not shocking that the pharmacology of channels might differ between rat and human cells and between native currents and those from heterologous expression.

The first suggestion in the Discussion that the authors conclude that Nav1.7 does not make the major contribution to TTX-S current comes obliquely, in a sentence in a paragraph about the up-regulation of currents by the inflammatory soup "As noted above, data from rodent DRG neurons suggest that NaV1.7 accounts for ~70% of the TTX-S current in small diameter neurons and NaV1.6 accounts for the remainder (Laedermann et al., 2013). However, the suggestion that the contribution of NaV1.7 varies across subpopulations of human sensory neurons is also consistent with the suggestion that NaV1.7 is not the dominant, let alone only VGSC subtype in human sensory neurons." In the original version, this sentence referred to a stated conclusion in Results based on the protoxin-II results that Nav1.7 was not the dominant current. But in the revised version, there has been no explanation yet of why the authors would conclude this, and it seems to come out of nowhere.

The authors' argument about Nav1.7 comes later, in the paragraph starting "The possibility that NaV1.7 may not be the dominant subunit underlying TTX-S currents, let alone significantly contribute to these currents in human DRG neurons, as suggested by our inflammatory mediator results, was further supported by our results with Pro-Tx II and PF-05089771." This is where the authors argue that the block by PF-05089771 is not good evidence for Nav1.7 mediating the TTX-S because they find that PF-05089771 at higher concentrations also inhibits the TTX-R current. To me, the argument in this paragraph was not convincing, as it essentially argues that the lack of effect of protoxin-II is more convincing than the potent block by PF-05089771. (The lack of ramp currents from TTX-S currents is another argument, but as the authors point out when presenting this data, this could be easily explained by different gating kinetics between human and rat TTX-S currents.).

There is a reason to be cautious about interpreting the results with PF-05089771 and Protoxin-II, because both interact with their target channels in a state-dependent manner. The authors did not study the effects of the blockers on isolated currents from TTX-R or TTX-S channels, but rather used a prepulse procedure to distinguish them by inactivation. They show that this protocol works in the absence of drugs, but obviously it may not in the presence of drugs that can modify the voltage-dependence of inactivation. In my view, that should introduce a lot of caution in interpreting these results. In fact, in reading the paper quickly, it would be very easy for a reader to miss the important point that when the authors refer to "TTX-S" and "TTX-R" currents, this is really shorthand for "currents available from -80 but inactivated by a prepulse to -40 mv" and "currents remaining after a prepulse to -40 mV". Although the authors clearly state this in the course of Results, I think it is an important point to reiterate in the Discussion, because it is an important qualification for interpreting the pharmacological results (not only for protoxin and PF-05089771, but also lidocaine and A-803467, which are also state-dependent blockers).

Because of these issues, I did not find the authors' suggestion that Nav1.7 accounts for little if any of the TTX-S current convincing and it would not be my interpretation. Of course, the authors should be free to interpret their own data, but they should probably consider at least qualifying their conclusion.

Whatever the interpretation given, the experimental results are very interesting and leave no doubt that the detailed pharmacology of the channels in human cells is quantitatively different in interesting ways compared to the currents in rat neurons, probably most importantly in the relative lack of use-dependence with lidocaine and the apparent significant sensitivity of TTX-R channels to PF-05089771.

I did not see a statement about informed consent for the collection of the human tissue, which seems important. Presumably this is the authors' 2015 paper that is cited for the Methods, but it seems that it should be here also.

https://doi.org/10.7554/eLife.23235.027

Author response

Essential revisions:

Both reviewers found the work to be very important, informative, and well done. Their compiled favorable comments are quoted verbatim below in "General Comments." Several key points came up that require addressing, however, relating to pharmacology, which was seen as the weakest area of the manuscript, and a few other aspects of voltage protocols and cell selection. The points that require addressing are summarized here and elaborated upon in the reviewers' words under "Detailed Points for Essential Revisions."

1) Verify that the very surprising results indicating that the primary TTX-S channels in humans (a) are not 1.7 and (b) are not blocked by the broad-spectrum Protoxin II are not influenced by plausible sources of error, e.g., those relating to complications of peptide perfusion.

We agree with reviewers that this is an important concern, and as noted below, have addressed this issue with additional data and further discussion. The additional data include results from the application of Protoxin II to rat DRG neurons, as well as data from human DRG neurons in response to the putative NaV1.7 selective small molecule channel blocker PF-05089771. With respect to the former, we were able to demonstrate that Protoxin II blocked a fraction of the TTX-S current in rat DRG neurons, arguing against potential complications with handling the peptide. With respect to the latter, we were able to demonstrate complete block of TTX-S currents in human DRG neurons with PF-05089771 with an IC50 of ~6 nM, remarkably close to that reported by Alexandrou and colleagues (Alexandrou et al., 2016). However, in marked contrast to heterologous expression results reported by Alexandrou and colleagues, we observed a significant reduction in TTX-R current in human DRG neurons treated with PF-05089771, with an IC50 of ~50 nM. Given that Alexandrou and colleagues reported that heterologously expressed NaV1.8 was resistant to PF-05089771 at concentrations as high as 10 μM, these observations underscore the need to validate results from heterologous expression studies on native proteins expressed in their native environment. The observation that PF- 05089771 has limited utility for the differential block of TTX-S and TTX-R currents in human DRG neurons raises the possibility that this compound also has limited utility for the differential block of NaV1.7 relative to other TTX-S channels likely present in human DRG neurons. Thus, we would argue that when considered in the context of our results with Protoxin II and our ramp current data, the limited selectivity of PF-05089771 on native channels lends further support to the suggestion that NaV1.7 is not the only, or even dominant channel in human DRG neurons.

We have expanded our discussion of these issues to address several points. The first concerns the selectivity of Protoxin II. As noted below, the reviewers are correct to note the sentence in the 2007 Priest paper, in which the authors stated that the toxin blocked all NaV1 isoforms (Priest et al., 2007). However, the Priest et al., 2007 paper represents a literature review of limited available data. Importantly, they went on to characterize this toxin in more detail, reporting in 2008, that the toxin was at least 100 time more potent against NaV1.7 than other NaV1 isoforms (Schmalhofer et al., 2008). Thus, in our revised manuscript, we acknowledge that our negative results with this toxin in human DRG neurons may have been due to technical problems, and the results with PF- 05089771 may still reflect a significant contribution of NaV1.7 to the TTX-S current in human DRG neurons. However, we also suggest an alternative hypothesis, which is that NaV1.7 is not the primary channel underlying TTX-S currents in human DRG neurons, where the negative results with Protoxin II and the altered pharmacological profile of PF- 05089771 reflect the impact of the native environment on the pharmacological properties of the channels present. Consistent with the suggestion that the relative contribution of NaV1.7 to TTX-S currents in rodent sensory neurons may have been over-estimated are the recent results from Deuis and colleagues (2017), who reported that a putative NaV1.7 selective spider venom peptide blocked no more than 50% of the TTX-S current in every subpopulation of sensory neurons tested (Deuis et al., 2017). A relatively limited role for NaV1.7 in human DRG neurons would also be consistent with our observation that a low threshold ramp current is absent in human DRG neurons as well as the relatively limited impact gain of function mutations in NaV1.7 have on the excitability of human nociceptive afferents (Namer et al., 2015).

2) Omit or address the very unexpected result that TTX increases ramp current, which also raises questions about the efficacy/specificity of drug perfusion.

We agree with the reviewers that this was a very perplexing observation. Per the reviewer’s suggestion, we have omitted these data. We still describe the high threshold ramp current in these neurons, however.

That said, we have pursued this issue further in the literature, and came across a couple of relevant studies. Most relevant is a study by Farmer and colleagues (2008), who observed a TTX-induced increase in TTX-R currents in rat DRG neurons that was associated with a leftward shift in the voltage-dependence of activation (Farmer et al., 2008). These authors concluded that the modulation of TTX-R currents by TTX was due to the relief of a tonic block of these channels by La3+ (used to block Ca2+ currents in their recording solution). We used Cd2+ rather than La3+ to block Ca2+ currents in our experiments, but there is also evidence of a Cd2+-induced block of TTX-R currents in DRG neurons (Kuo et al., 2002). Thus, it is possible that the leftward shift in the ramp current we observed was due to a TTX-induced relief of Cd2+ block. Importantly, such an explanation would support the conclusion that ramp currents in human DRG neurons are largely carried by TTX-R currents. If the reviewer’s agree that the evidence is sufficient to support this speculation, we would happily include our TTX ramp results in the manuscript, but leave this decision to the reviewers/editor.

3) Clarify voltage protocols used and (a) use the information to strengthen/clarify/curtail conclusions based on pharmacology and (b) validate (test and/or rationalize) the prepulse approach as opposed to a pharmacological approach to separate the currents in humans.

We have provided additional details about the voltage protocols used throughout the study. We have acknowledged the possibility that the use of a voltage- protocol to isolate TTX-S and TTX-R currents introduced a source of error in the characterization of current properties, particularly for pharmacological experiments where the drugs used could have influenced the biophysical properties of the channels such that isolation of current with voltage-steps was no longer possible. We have also added further justification for our decision to rely on voltage protocols for current separation. This decision was primarily based on the facts that 1) there was a finite time- frame over which stable recordings were obtained, 2) the use of a voltage-protocol was faster than a pharmacological approach, at least with respect to the reversal of the pharmacological block of TTX-S channels, and 3) we wanted to maximize the amount of data collected from each neuron. With respect to this last point, the use of voltage protocols to isolate TTX-S and TTX-R currents enabled us to monitor both currents in each neuron.

4) Test and/or clarify ambiguities arising from size differences and size-based selection to verify that differences between species weren't based on clamp artifacts from the different sizes of neurons.

This is an important point. We have added a section to the revised manuscript specifically addressing this concern. We have laid out the data arguing against clamp artifacts on the biophysical parameters obtained. We have also acknowledged the limitations associated with assumptions about the phenotype of the human neurons.

5) Address or omit the deactivation time constant measurements, about which specific questions were raised.

Done

We recognize that the availability of human tissue may be limited. If it is not possible or practical to address the issues directly, a (less desirable) alternative is to tone down or qualify the conclusion that Nav1.7 may not contribute much to the TTX-S current, indicating that further pharmacological work is necessary for a definitive conclusion.

The reviewer’s suggestion about the use of other NaV1.7 blockers prompted us to go back into the analysis of compounds that we originally tested for Eli Lilly when we got this project up and running. We were blinded to the compounds used, but in fact had a complete data set with PF-05089771, which included the activity of this compound on TTX-R currents in human DRG neurons, an observation not previously reported.

Importantly, and consistent with our original suggestion about the interpretation of the previous data with this compound on human DRG neurons, this drug does not appear to be as selective on native channels in human neurons as it was reported to be against heterologously expressed channels. We have also included additional data with protoxin II against rat TTX-S currents. Thus, while we have now acknowledged the limitations of the conclusions drawn from our pharmacological and biophysical data, we suggest that our results in the context of data available in the literature strongly support the possibility that NaV1.7 is not the dominant channel underlying TTX-S currents in human DRG neurons.

[…]

Detailed points for essential revisions:

1) The area in which the results in the manuscript could be stronger is the most important – the conclusion that the TTX-S channels in the human neurons are not primarily carried by Nav1.7 channels. If true, this is very important, since multiple companies are developing Nav1.7 inhibitors with the assumption that these channels are critical for the function of nociceptors. Also, the conclusion that Nav1.7 do not contribute to the TTX-S currents is opposite to a previous study of human DRG neurons in which it was concluded that the majority of the TTX-S current was carried by Nav1.7 channels, based on inhibition of 75% of the current by 30 nM PF-05089771, a fairly selective blocker of Nav1.7 channels. The authors point out this previous study used neurons in culture for up to 9 days, which could cause an upregulation of Nav1.7 channels – a valid point. They also suggest that the result could have been from imperfect selectivity of PF-05089771, which while possible seems a lot less likely, since Alexandrou et al., 2016 saw big effects of 30 nM PF-05089771, which would be expected to have minimal effects on any of the other channel types.

Here the authors present two pieces of evidence arguing against mediation of TTX-S by Nav1.7. One is the lack of effect of 10 nM Protoxin II. It seemed very puzzling that the authors chose protoxin II as a blocker, because a previous review of the properties of this toxin stated "ProTx-I and ProTx-II inhibit all sodium channel (Nav1) subtypes tested with similar potency." (Priest et al. Toxicon. 2007 49:194-201). This is certainly not clear from the presentation, where a reader would naturally infer that the toxin was used because it had selectivity for Nav1.7. The evidence against Nav1.7 would be far stronger if the authors had used PF-05089771, which is much more selective. This compound has been commercially available (from both Σ and Tocris) for some time.

The reviewer raises important points. We had, in fact tested PF-05089771 on human DRG neurons and observed a complete block of TTX-S currents with 30 minutes of incubation with an IC50 of 6 nM. However, the IC50 for the block of TTX-R currents was 50 nM. Admittedly, these data support the possibility that NaV1.7 does in fact underlie the majority of TTX-S currents in human DRG neurons. However, given that heterologous expression data in the Alexandrou paper indicated that heterologously expressed NaV1.8 is insensitive to PF-05089771 at concentrations as high as 10 μM, our data with this compound at support the possibility that the TTX-S current blocked reflects more than activity in NaV1.7. Of note, Alexandrou and colleagues studied TTX-S currents in the presence of A-803467 to block NaV1.8, and therefore would not have detected an influence of PF-0589771 on TTX-R currents. These data have been added to the revised manuscript.

By way of an explanation for our failure to include results with PF-05089771 in our original manuscript, these data were collected in a series of experiments in which other proprietary compounds from Eli Lilly were also tested. We had not broken the code on the compounds assessed until this question was raised (results were submitted to Lilly in the blinded manner in which they were obtained). These data have now been added to the revised manuscript.

On the other hand, the choice of Pro-Tx II was based on evidence that this toxin is a potent and selective blocker of NaV1.7 as documented by Schmalhofer and colleagues (Schmalhofer et al., 2008), and subsequently others. The toxin is sold as an NaV1.7 selective blocker through companies such as Σ and Tocris. The Priest et al. 2007 paper referred to contains a sentence in the Abstract that is admittedly misleading, based largely on the results of a study by Middleton and colleagues (2002), in which results from variety of assay protocols were combined (Middleton et al., 2002). The Schmalhofer paper was a follow-up to this 2007 review which contained a more careful characterization of Pro-Tx II, in which the selectivity of the toxin was clearly documented (Schmalhofer et al., 2008).

The negative data with Pro-Tx II combined with the considerably lower level of selectivity of PF-05089771 in human DRG neurons than that reported in heterologous expression system and the absence of low threshold ramp currents in human DRG neurons all point to the possibility that NaV1.7 may not be the dominant TTX-S channel in human DRG neurons. This suggestion is further substantiated by the relatively minor impact of NaV1.7 gain of function mutations on the excitability of human nociceptive afferents documented in microneurography studies (Namer et al., 2015). While the activity of the spider venom peptide Pn3a on native TTX-R currents was not well described, recent results with this putatively NaV1.7 selective blocker would also suggest that NaV1.7 does not even underlie the majority of TTX-S current in rat DRG neurons (Deuis et al., 2017). It should also be noted that recent data from the NaV1.7 null mutant mice suggest that this subunit may, in fact play a more important role in mediating Na+ influx and associated changes in gene expression than it does in the electrical properties of these neurons (Minett et al., 2015), while a more prominent role in transmitter release from the central terminals (Alexandrou et al., 2016), may also contribute to a more limited role for this channel in the TTX-S currents detected in the afferent cell body.

Nevertheless, because these lines of evidence are indirect, these possibilities are only suggested in the revised manuscript.

Given the efficacy of protoxin-II against all sodium channels, the lack of effect is very surprising whatever channels underlie the TTX-S current. A possible experimental problem with potent peptide toxins of this type is that they can bind to plastic tubing and glassware and be difficult to remove, so that subsequent application appears to have no effect because of presence of the agent under "control" conditions. I raise that because the lack of any effect of the toxin is unexpected whatever channels the sodium currents are coming from and because we have experienced that insidious problem several times in my own lab with similar toxins – a problem that could only be resolved by completely replacing tubing and chambers.

We appreciate the reviewer’s suggestion, but tried a number of different strategies, including siliconizing glass tubes and the application of the toxin via siliconized glass pipettes. We suggest that the issue is really the concentration range over which the drug was tested. While we did use concentrations as high as 300 nM, the majority of experiments involved concentrations of 30 nM or lower. More importantly, we were able to demonstrate a block of TTX-S currents in rat DRG neurons of 50.5 ± 13.7 (n = 5)% with 10 nM toxin. Thus, while we have acknowledged that our failure to detect an effect of the toxin in human DRG neurons may reflect problems with working with the toxin, we do not think that was the primary source of the negative results obtained.

2) The second piece of evidence that authors present against the contribution pf Nav1.7 channels is the most puzzling result in the paper – that TTX not only did not inhibit a current induced by ramp of voltage but actually enhanced the current and shifted its voltage dependence to more hyperpolarizing voltages. The shift of current to more hyperpolarized voltage makes no sense at all and suggests some sort of experimental problem. It would suggest that TTX enhances sodium current for small depolarizations which has never been described for any channels. At a minimum, the authors need to follow up the ramp experiment with step depolarization experiments to characterize this effect in more detail. Even if TTX simply had no effect on the ramp current, this is only weak evidence against mediation by Nav1.7. It could reflect for example faster or more complete closed-state inactivation in human Nav1.7 channels versus rodent Nav1.7 channels, which would be interesting. The authors did not compare closed-state inactivation in the human vs rat neurons so this seems like an open possibility.

We completely agree with the reviewer that this was a perplexing and unexpected observation. There was no suggestion from the step depolarization data that TTX-R currents were facilitated by the presence of TTX. Nor was there evidence of a shift in the gating of TTX-R currents in the presence of TTX, as there was no significant change in either the V1/2 of activation (which changed 0.07 ± 0.04 mV after the application of TTX) or the slope of the GV curve (which changed by 0.03 ± 0.03 mV after the application of TTX). Furthermore, with a decrease in total current following application of TTX, a decrease in voltage error should have resulted in a depolarizing shift in the ramp current. We did not observe a similar shift in rat DRG neurons, suggesting that whatever the underlying mechanism, it was only detected in human neurons. Nevertheless, because we did not assess the onset of closed state inactivation, we have removed these data from the manuscript.

3) It would be helpful if all the figures had clear statements of exactly what the voltage protocols were, which were not always clear to me. In Figures 47 on TTX-R current, it seems that the records were obtained from a steady holding voltage of -80 mV and a 500 ms prepulse, probably to -40 mV based on Figure 1, but this is not stated explicitly and the prepulse potential is not given in each case. Materials and methods says "The pre-pulse potentials needed for full channel availability and for inactivation of TTX-S currents were used for the generation of current-voltage (I-V) curves for total current, and for TTX-R current" which suggests that the same prepulse was not always used. The exact protocol in each figure should be stated. This is especially an issue for interpreting the effects of lidocaine, since the potency is so strongly sensitive to resting membrane potential. It seems possible that the potency of A-803467 could also be sensitive to the exact voltage protocol.

We apologize for the confusion and have provided the information requested – at least for the raw traces included in each figure because the reviewer is correct, we did vary the protocol in some neurons to facilitate our ability to isolate TTX-S from TTX-R currents.

Additionally, the separation of TTX-R and TTX-S components by using a conditioning voltage pulse (instead of the more laborious method of TTX block) is a key aspect of the analysis and interpretation. Deficiencies in this separation technique, which may be different for rat versus human DRG neurons, could result in apparent differences or underestimate differences. With that in mind, how was it possible to determine the availability curve for TTX-R specific currents in Figure 4D? The conditioning pulse range extended from -60 mV to 0 mV and yet Figure 1 shows a prepulse of -35 mV is needed to isolate the TTX-R component (via inactivation of TTX-S). How was availability data obtained for conditioning voltages more negative than -35 mV?

We apologize for failing to make this more clear. As can be seen in Figure 1, it was possible to use a combination of voltage and time to separate TTX-S from TTX-R currents. That is, in every neuron in which the TTX sensitivity of the TTX-S current was confirmed with TTX, the current evoked at potential ranging between -5 mV and +5 mV was completely inactivated by 10 ms after the start of the voltage step. Thus, it was possible to monitor TTX-R current availability across a full range of pre-pulse potentials. We have further clarified this point in the revised manuscript.

A great feature of the work is the comparison to studies in rat neurons done in the same lab. If these were actually done in parallel, using for example exactly the same batches of pharmacological reagents and by the same experimenter, this might be mentioned explicitly as it further strengthens the point that the differences are genuine and not the result of differences in procedures or reagents.

We initiated the rat experiments toward the end of the collection of data from human DRG neurons. However, the same stocks of test agents were used in both rats and humans. We have added these details to the revised manuscript.

4) A missing piece of information is exactly how the authors selected which cells to record from. From the fact that they all had TTX-R current, the experimenter presumably chose smaller-diameter cells to record from, but what exactly was the selection criterion? Do the authors think the cells all correspond to cell bodies of C-fibers? The authors have discussed the challenges of high Na current density and series resistance that often plague studies of VGSCs. Because human DRG soma are much larger than rat, the demands on clamp quality would be systematically more challenging in recordings from human than rat neurons. This might cause a systematic difference in the apparent gating behavior. The distribution of cell capacitance did, however, have a region of overlap around 50 pF. Perhaps a subset of rat and human DRG neurons with comparable size could be analyzed to provide reassurance that the observed species differences were still detectable in a population of similar-sized neurons?

We thank the reviewer for raising this issue. We did, in fact start this study with a focus on small to medium diameter human DRG neurons, based on our interest in pain and the rodent data indicating an association between afferent cell body diameter and function. However, when we noticed that the vast majority of even the medium diameter neurons also contained TTX-R currents, those generally associated with nociceptors, we also started to record from larger neurons. And while we generally avoided the largest of neurons because of the higher likelihood for larger currents and consequently greater difficulty with clamp control, in the end, we sampled a relatively broad distribution of cell body sizes, relative to the total distribution observed in cut sections. We have included data from cut sections in the revised manuscript. In doing so, we were able to expand on the implications of the cell body size distribution observed as well as the distribution of TTX-S currents with respect to our understanding of the relationship between cell body size and function in the Discussion of the revised manuscript.

While we have acknowledged the potential impact of clamp control problems on our characterization of current properties, we suggest errors associated with clamp control had a minimal impact on differences observed. This suggestion is based on fact that despite the more hyperpolarized voltage-dependence of activation of both TTX-R and TTX-S currents in human DRG neurons, there were only small differences in the slope of the G-V curves for the currents evoked from rat and human neurons.

Finally, with respect to the suggestion that it might be possible to record from rat and human neurons of comparable size, as shown in Figure 2, there was a small overlap in the size of the neurons studied from the two species. To address the reviewer’s concern, we have analyzed current properties of the largest rat neurons to compare to those in the smallest human neurons (in the 40-70 pF range). It should be acknowledged from the outset that this comparison is not ideal, as the largest rat neuron in which TTX-R currents were detected had a membrane capacitance of 54.3 pF. The observation that only TTX-S currents were detected in larger neurons is consistent with our previous experience and the suggestion that to the extent to which TTX-R currents are a reflection of neurons with nociceptive properties, this subpopulation of neurons tends to have a smaller cell body diameter. That said, even when these subpopulations of neurons were analyzed, differences in TTX-S and TTX-R currents in rat and human neurons were evident. For example, the V1/2 of current activation for TTX-R in the overlapping subpopulation of rat and human sensory neurons were -2.1 ± 0.4 mV and -10.8 ± 1.7 mV, respectively.

These details have been added to the revised manuscript.

5) The interpretation and discussion of VGSC deactivation was confusing (Figure 5E, which shows plots of deactivation tau versus voltage). First, deactivation time constants are usually presented as a function of the deactivation pulse potential, not the preceding voltage to activate the channel as shown in Figure 5E. However, from the text it seems that this is not a standard plot of tail time constant against voltage as one might assume at first but rather a measurement of the tail time constant at -60 mV following steps to different voltages. This should be made clear in the figure legend. Were there any measurements of deactivation time constants at other voltages? These would be helpful to someone who wanted to make a Hodgkin-Huxley type model from the data. Second, deactivation rates are usually measured over a range of test potentials. The more hyperpolarized ones are often regarded as more representative of "true" deactivation (O->C transition) but the fast kinetics are technically demanding. Responses at more depolarized potentials are more easily resolved, but may represent other state transitions (not just O -> C). The canonical view is that a true deactivation rate should be independent of the pulse potential used to open the channel. The atypical behavior in Figure 5E suggests a pulse to -60 mV was not measuring deactivation in isolation. Moreover, deactivation data are not presented for TTX-S currents. Perhaps the deactivation data should be omitted?

We regret that tail current data were not collected in a more traditional and appropriate manner. Unfortunately, we only realized this after the fact, despite the consistent observation that the TTX-R tail current in human DRG neurons looked consistently different from those evoked in the rat. Consequently, we attempted to describe the basis for this difference. Nevertheless, we agree with the reviewer that in the absence of a protocol enabling us to clearly describe deactivation, these data should be omitted and we have done so in the revised manuscript.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

"TTX-S" and "TTX-R" are used as shorthand for prepulse subtraction, which is validated for control solutions in Figure 1. This procedure may complicate the interpretation of some of the results, however, particularly the pharmacology leading to the conclusion that 1.7 is not the primary TTX-S channel. Please edit the text (especially the Introduction and Discussion) for the following:

1) To make sure that readers are clear that "TTX-S" and "TTX-R" actually refer to currents inactivated at -40 mV and available at -40 mV,

We appreciate the reviewers concern. As this terminology is widely used throughout the literature to describe these two general current types in DRG neurons, that are not only readily distinguishable based on their sensitivity to TTX, but by their biophysical properties (both steady-state and kinetic) we have largely retained the use of the terminology in the Introduction. Although we did add a clause indicating that voltage- protocols were used in the majority of experiments described in this study. To further address the reviewer’s concern, however, we have included the following sentence in the Results section after the demonstration in Figure 1 that voltage and TTX can be used to isolate the same currents: “Based on these results, we refer to the slowly activating and slowly inactivation current resistant to steady-state inactivation as TTX-R current even though voltage steps rather than TTX was used to isolate this current in all subsequent experiments”. We go on clarify in the Discussion that we refer to this slow-activating and inactivating current in human DRG neurons as the TTX-R current, even though voltage was used for current isolation in all but the experiments described in Figure 1.

2) To ensure that the prepulse subtraction procedure is kept in mind in interpreting the effects of the blockers, and

We have included this caveat in the discussion of results obtained.

3) To limit or constrain claims that NaV1.7 is not the primary TTX-S current, as appropriate.

We have attempted to present the data for and against this proposal as objectively as possible. The possibility that NaV1.7 may not be the primary TTX-S current is suggested as an alternative possibility. We hope we have left it to the reader to decide, acknowledging that even our attempt at a balanced consideration of the data still favors this possibility.

Please also include a statement about informed consent (as appropriate) for collection of human tissue.

This information was included in the original version of our manuscript

The full comments pertaining to the requested revisions are below and have been agreed upon by all participants in the review.

General assessment and major comments:

The most important change in the manuscript is new data with the Pfizer Nav1.7 inhibitor PF-05089771, which the authors find inhibits the TTX-S current quite potently, with an IC50 of about 10 nM. In addition, the authors have clarified some the details of the voltage protocols and cell selection.

I was very surprised that with the new data showing potent block of the TTX-S current by PF-05089771, the authors still end up concluding that Nav1.7 channels make little contribution to the TTX-S current in human DRG neurons. If it were my data, my interpretation of the overall set of the results the authors present would be the opposite, that the TTX-S current is likely to be primarily from Nav1.7 channels based on its sensitivity to PF-05089771, but that it is (interestingly) less sensitive to Protoxin-II than rat Nav1.7 channels. The observation that PF-05089771 also seems to inhibit TTX-R currents in human DRG neurons at higher concentrations is interesting and important. So, like most small molecules, the selectivity of PF-05089771 is imperfect. That is important to know, and constitutes another of the many interesting points in the manuscript about different pharmacology of the native currents in human DRG neurons compared to rat DRG neurons or heterologously-expressed human channels. However, it is still true that PF-05089771 inhibits the TTX-S current with significantly higher potency, and about what is expected from the effects on cloned human Nav1.7 channels. So my interpretation would be that the results with PF-05089771 imply that the TTX-S current is mostly from Nav1.7, but that the compound is less selective than previously realized and at higher concentrations also inhibits native human Nav1.8 channels – a very important point that adds to the significance of the results in the manuscript.

We agree with the reviewer that one interpretation of our data is that NaV1.7 underlies the majority of TTX-S current in human DRG neurons. Our data with PF- 05089771 are indeed consistent with this interpretation. This is now clearly stated in the revised manuscript.

However, several lines of data support an alternative possibility, which is that NaV1.7 is not the primary subunit underlying TTX-S currents. First, while PF-05089771 was an order of magnitude more potent in the block of NaV1.7 than NaV1.2 or NaV1.6 expressed in HEK293 cells, with IC50 values of 0.011 μM, 0.11 μM and 0.16 μM respectively, our data suggest this compound is less selective against channels expressed in their native environment. That is, PF-05089771 had no activity at NaV1.8 channels expressed in HEK cells at concentrations as high as 10 μM, while TTX-R currents in human DRG neurons were blocked with an IC50 of ~60 nM. A comparable loss of selectivity at other channel subtypes would not enable PF-05089771 to differentiate between NaV1.7, NaV1.2 or NaV1.6. Importantly, semiquantitative PCR analysis of NaV subunit expression levels in whole mouse DRG indicates that NaV1.2 is mRNA is the most highly expressed TTX-S subunit (Laedermann et al., 2014). RNAseq analysis of DRG neurons suggests that the number of copies of NaV1.6 (31.1) is comparable to that of NaV1.7 (54), at least in the subpopulation of neurons not in the TRPV1 lineage (Goswami et al., 2014). Comparable results were obtained with a quantitative PCR analysis of single DRG neurons, where in those larger than 30 μm in diameter, NaV1.6 expression was comparable to that of NaV1.7 (Ho and O'Leary, 2011). Probably most importantly, there was only a 25% reduction in TTX-S current in DRG neurons from the NaV1.7 knock-out mouse (Nassar et al., 2004). Second, there was no evidence of a low threshold ramp current in human DRG neurons suggesting that if NaV1.7 was the dominant subunit, it would have very different biophysical properties than the current observed in rodent sensory neurons or in heterologous expression systems (Cummins et al., 1998). Third, there was no evidence of Protoxin II block of the TTX-S current in human DRG neurons suggesting that if NaV1.7 was the dominant subunit, it would also have very different toxin sensitivity than that of NaV1.7 expressed in heterologous expression systems (Schmalhofer et al., 2008) or in rodent sensory neurons (Laedermann et al., 2013). And fourth, inflammatory mediator-induced modulation of TTX-S currents was only observed in a subpopulation of neurons in which TTX-S currents were observed. While this could be explained by differences in second messenger signaling between neurons, it may also reflect a differential expression of TTX-S channel subtypes.

To me, the flow of the revised manuscript now seems strange in that final set of experimental results in Results showing potent block by PF-05089771 gives the reader the impression that the TTX-S current is likely mainly Nav1.7, but then in the Discussion the first mention of this issue suggests the opposite without explaining why. The first part of the Discussion summarizing the pharmacology is very reasonable and straightforward (the paragraph starting, "With respect to drug development, the results of the present study are consistent with those from previous studies indicating that both expression systems and species differences may have a significant influence on the pharmacological properties of the protein in question." This paragraph nicely summarizes why it is not shocking that the pharmacology of channels might differ between rat and human cells and between native currents and those from heterologous expression.

The first suggestion in the Discussion that the authors conclude that Nav1.7 does not make the major contribution to TTX-S current comes obliquely, in a sentence in a paragraph about the up-regulation of currents by the inflammatory soup "As noted above, data from rodent DRG neurons suggest that NaV1.7 accounts for ~70% of the TTX-S current in small diameter neurons and NaV1.6 accounts for the remainder (Laedermann et al., 2013). However, the suggestion that the contribution of NaV1.7 varies across subpopulations of human sensory neurons is also consistent with the suggestion that NaV1.7 is not the dominant, let alone only VGSC subtype in human sensory neurons." In the original version, this sentence referred to a stated conclusion in Results based on the protoxin-II results that Nav1.7 was not the dominant current. But in the revised version, there has been no explanation yet of why the authors would conclude this, and it seems to come out of nowhere.

We have reworked the Discussion to address the reviewers concerns. The suggestion that NaV1.7 might not be the dominant subunit in human DRG neurons is now only raised in the context of a Discussion of the possibility that NaV1.7 may not be the dominant subunit underlying the TTX-S currents in human DRG neurons. And this possibility is only raised after a Discussion of the possibility that NaV1.7 does underlie the TTX-S current.

The authors' argument about Nav1.7 comes later, in the paragraph starting "The possibility that NaV1.7 may not be the dominant subunit underlying TTX-S currents, let alone significantly contribute to these currents in human DRG neurons, as suggested by our inflammatory mediator results, was further supported by our results with Pro-Tx II and PF-05089771." This is where the authors argue that the block by PF-05089771 is not good evidence for Nav1.7 mediating the TTX-S because they find that PF-05089771 at higher concentrations also inhibits the TTX-R current. To me, the argument in this paragraph was not convincing, as it essentially argues that the lack of effect of protoxin-II is more convincing than the potent block by PF-05089771. (The lack of ramp currents from TTX-S currents is another argument, but as the authors point out when presenting this data, this could be easily explained by different gating kinetics between human and rat TTX-S currents.).

We appreciate the reviewers concerns and agree that this argument could and should have been made better on the one hand, and that the possibility that NaV1.7 could still be the dominant subunit made more clearly on the other. With respect to the reviewers primary concern about the selectivity of PF-05089771, we acknowledge that it is possible that the compound retained its selectivity for NaV1.7 over NaV1.2 and NaV1.6 (as well as the rest of the TTX-S subunits present in sensory neurons), while dramatically losing selectivity for NaV1.8 in human DRG neurons. However, a greater than three orders of magnitude increase in potency would at least raise the possibility that the compound loses all selectivity against human channels in their native environment.

There is a reason to be cautious about interpreting the results with PF-05089771 and Protoxin-II, because both interact with their target channels in a state-dependent manner. The authors did not study the effects of the blockers on isolated currents from TTX-R or TTX-S channels, but rather used a prepulse procedure to distinguish them by inactivation. They show that this protocol works in the absence of drugs, but obviously it may not in the presence of drugs that can modify the voltage-dependence of inactivation. In my view, that should introduce a lot of caution in interpreting these results.

We completely agree with the reviewer that it is important to consider the potential implications of state-dependent block. However, we disagree with the implication that the voltage protocol used to isolate TTX-R from TTX-S could account for the observation that TTX-R currents were blocked by PF-05089771. This compound blocks inactivated channels more potently that channels in a resting or closed state. The protocol used to evoke TTX-R currents should have minimized any inactivated-state block of these channels, and consequently any detectable block of these channels.

Furthermore, the TTX-R currents remaining in the presence of PF-05089771 had biophysical properties comparable to the TTX-R currents recorded in the absence of the compound, with a comparable voltage-dependence of inactivation, as well as kinetics of activation and inactivation. Conversely, any contamination of TTX-R currents observed in the presence of PF-05089771 by TTX-S currents whose gating properties were somehow modified by the presence of the compound would suggest that the potency of the compound was lower in human DRG neurons than in HEK cells.

In fact, in reading the paper quickly, it would be very easy for a reader to miss the important point that when the authors refer to "TTX-S" and "TTX-R" currents, this is really shorthand for "currents available from -80 but inactivated by a prepulse to -40 mv" and "currents remaining after a prepulse to -40 mV". Although the authors clearly state this in the course of Results, I think it is an important point to reiterate in the Discussion, because it is an important qualification for interpreting the pharmacological results (not only for protoxin and PF-05089771, but also lidocaine and A-803467, which are also state-dependent blockers).

We completely agree with the reviewer and have reiterated this point in the Introduction, Results, and Discussion as suggested.

Because of these issues, I did not find the authors' suggestion that Nav1.7 accounts for little if any of the TTX-S current convincing and it would not be my interpretation. Of course, the authors should be free to interpret their own data, but they should probably consider at least qualifying their conclusion.

As noted above, we have tried to lay out the evidence both for and against the contribution of NaV1.7 to the TTX-S currents in human DRG neurons more clearly.

Whatever the interpretation given, the experimental results are very interesting and leave no doubt that the detailed pharmacology of the channels in human cells is quantitatively different in interesting ways compared to the currents in rat neurons, probably most importantly in the relative lack of use-dependence with lidocaine and the apparent significant sensitivity of TTX-R channels to PF-05089771.

We appreciate the supportive comments.

I did not see a statement about informed consent for the collection of the human tissue, which seems important. Presumably this is the authors' 2015 paper that is cited for the Methods, but it seems that it should be here also.

We agree with the reviewer that this is an important point. This information was in the first sentence of Materials and methods section of this manuscript.

https://doi.org/10.7554/eLife.23235.028

Article and author information

Author details

  1. Xiulin Zhang

    Department of Urology, The Second Hospital of Shandong University, Jinan Shi, China
    Contribution
    XZ, Data curation, Formal analysis, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared.
  2. Birgit T Priest

    Lilly Research Laboratories, Indianapolis, United States
    Contribution
    BTP, Conceptualization, Methodology, Writing—review and editing
    Competing interests
    BTP: Employee of Eli Lilly
  3. Inna Belfer

    Office of Research on Women’s Health, National Institutes of Health, Bethesda, United States
    Contribution
    IB, Conceptualization, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared.
  4. Michael S Gold

    Department of Neurobiology, University of Pittsburgh School of Medicine, Pittsburgh, United States
    Contribution
    MSG, Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Project administration, Writing—review and editing
    For correspondence
    msg22@pitt.edu
    Competing interests
    MSG: Has received grant support from Eli Lilly and Grunenthal and has served on an advisory panel for Grunenthal and the Global Pain Foundation.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2083-6206

Funding

National Institutes of Health (R01DE018252)

  • Michael S Gold

Eli Lilly and Company

  • Michael S Gold

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported in part by a Grant from Eli Lilly, as well as Grants from the National Institutes of Health (R01DE018252).

Ethics

Human subjects: DRG were collected from organ donors with the consent of family members for the use of their loved one's tissue for research purposes. The protocol for the collection and study of tissue from organ donors was approved by the University of Pittsburgh Committee for Oversight of Research and Clinical Training Involving Decedents. CORID ID #358

Animal experimentation: All procedures were approved by the University of Pittsburgh Institutional Animal Care and Use Committee and performed in accordance with National Institutes of Health guidelines for the use of laboratory animals in research. IACUC Protocol #12121265

Reviewing Editor

  1. Indira M Raman, Northwestern University, United States

Publication history

  1. Received: November 12, 2016
  2. Accepted: April 11, 2017
  3. Version of Record published: May 16, 2017 (version 1)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 1,734
    Page views
  • 378
    Downloads
  • 21
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)