Redox-Dependent Modulation of T-Type Ca2+ Channels in Sensory Neurons Contributes to Acute Anti-Nociceptive Effect of Substance P

Abstract Aims: Neuropeptide substance P (SP) is produced and released by a subset of peripheral sensory neurons that respond to tissue damage (nociceptors). SP exerts excitatory effects in the central nervous system, but peripheral SP actions are still poorly understood; therefore, here, we aimed at investigating these peripheral mechanisms. Results: SP acutely inhibited T-type voltage-gated Ca2+ channels in nociceptors. The effect was mediated by neurokinin 1 (NK1) receptor-induced stimulation of intracellular release of reactive oxygen species (ROS), as it can be prevented or reversed by the reducing agent dithiothreitol and mimicked by exogenous or endogenous ROS. This redox-mediated T-type Ca2+ channel inhibition operated through the modulation of CaV3.2 channel sensitivity to ambient zinc, as it can be prevented or reversed by zinc chelation and mimicked by exogenous zinc. Elimination of the zinc-binding site in CaV3.2 rendered the channel insensitive to SP-mediated inhibition. Importantly, peripherally applied SP significantly reduced bradykinin-induced nociception in rats in vivo; knock-down of CaV3.2 significantly reduced this anti-nociceptive effect. This atypical signaling cascade shared the initial steps with the SP-mediated augmentation of M-type K+ channels described earlier. Innovation: Our study established a mechanism underlying the peripheral anti-nociceptive effect of SP whereby this neuropeptide produces ROS-dependent inhibition of pro-algesic T-type Ca2+ current and concurrent enhancement of anti-algesic M-type K+ current. These findings will lead to a better understanding of mechanisms of endogenous analgesia. Conclusion: SP modulates T-type channel activity in nociceptors by a redox-dependent tuning of channel sensitivity to zinc; this novel modulatory pathway contributes to the peripheral anti-nociceptive effect of SP. Antioxid. Redox Signal. 25, 233–251.


Introduction
S ubstance P (SP) is an 11-amino-acid neuropeptide (RPKPQQFFGLM) of a tachykinin family that is produced in both peripheral and central nervous systems (CNS). SP is mainly an excitatory neuromodulator that is involved in pain transmission and smooth muscle contraction, and it modulates inflammatory and immune responses (69). Increased SP release or elevated expression of its receptors has been linked to many diseases such as acute and chronic inflammation and infections, asthma, affective and addictive disorders, and some forms of cancer [reviewed in Steinhoff et al. (62)]. In the peripheral somatosensory system, SP is abundantly produced by a subset of transient receptor potential cation channel subfamily V member 1 (TRPV1)-positive, pain-sensing (nociceptive) neurons and released in response to noxious stimulation both centrally and peripherally (40).
Peripheral SP released from nociceptive nerve endings causes so-called ''neurogenic inflammation'' (54); whereas in the spinal cord, SP functions as an excitatory co-transmitter alongside glutamate (12,38). Intracellular effects of SP are mediated by the G-protein-coupled neurokinin (NK1 -3) receptors. The strong excitatory role of SP in spinal nociceptive pathways has triggered a massive, industry-wide quest for new analgesics based on NK1 receptor antagonists. However, despite strong pre-clinical data, clinical trials of NK1 receptor antagonists failed (22,59).
The reasons for the lack of analgesic efficacy of the systemically applied NK antagonists are still unclear. However, one hypothesis is that, though excitatory in CNS, SP may exert paradoxical inhibitory/analgesic effects in the peripheral nociceptive pathways (32,33). Thus, SP augments an inhibitory M-type K + current in nociceptors via the atypical G i/o -mediated signaling cascade that utilizes reactive oxygen species (ROS) as second-messenger molecules (32,33).
Here, we discovered another component of the antiexcitatory action of SP in nociceptors: inhibition of T-type (Ca V 3) Ca 2+ channels. Ca V 3 are encoded by CACNA1G, CACNA1H, and CACNA1I genes, which give rise to Ca V 3.1, Ca V 3.2, and Ca V 3.3 pore-forming a subunits, respectively (8). T-type channels are activated at voltages near or below -60 mV and may display a significant window current at the neuronal resting membrane potential (E rest ) (51). Ca V 3s are abundantly expressed in cell bodies, axons, and peripheral terminals of nociceptive dorsal root ganglia (DRG) neurons, with Ca V 3.2 being the dominant isoform (9,11,18,56,63). Accordingly, peripherally applied T-type channel inhibition produces anti-nociceptive effects (2,65).
Here, we show that (i) SP acutely inhibits T-type Ca 2+ channels in DRG neurons via an ROS-dependent intracellular signaling cascade; (ii) this redox-mediated inhibition operates through the modulation of Ca V 3.2 channel sensitivity to ambient extracellular zinc; and (iii) a hind-paw injection of SP produces anti-nociceptive effects that can be reduced by the in vivo knock-down of Ca V 3.2 in DRG and partially mimicked by the peripheral injection of a specific T-type channel blocker. Thus, our study establishes a dual mechanism by which SP exerts its peripheral anti-nociceptive effect: a simultaneous inhibition of excitatory T-type Ca 2+ current and an enhancement of inhibitory M-type K + current.

SP inhibits T-type Ca 2+ current in sensory neurons via endogenous ROS generation
SP has been shown to trigger endogenous ROS release in nociceptive DRG neurons in an ''atypical'' G i/o -mediated signaling cascade stimulating mitochondrial ROS generation (33). Since T-type Ca 2+ channels are expressed in nociceptors and are redox sensitive (49,65), we investigated whether SP affects the activity of low-voltage-activated (LVA, T-type) Ca 2+ currents in DRG neurons.
Whole-cell recordings were made from cultured, smalldiameter, and mostly TRPV1-positive neurons. In 301 of 710 (42%) such DRG neurons, depolarization to -40 mV induced LVA currents, with kinetic properties similar to those of recombinant Ca V  . In our earlier experiments, we used MIB to confirm current identity but later switched to Z944, when it became available.
The mean LVA current amplitude was -161.7 -11.3 pA. The NK1-specific agonist [Sar 9 ]-Substance P (S9SP; 1 lM) acutely inhibited LVA current in the majority of the LVAexpressing DRG neurons (Fig. 1A 1 , B); the mean inhibition was 22.5% -2.3% (n = 84; p < 0.001), although there was a degree of scattering in the magnitude of a response and approximately a third of neurons tested displayed minimal response (e.g., 28/ 84 or 33% neurons had <10% inhibition), which was suggestive of a low level of NK1 receptor expression in these neurons.
Although NK1 receptor signaling is generally associated with the G q/11 type of signaling cascade (37,43), recent data suggest that at least some effects of SP in DRG neurons, such as ROS release and redox-dependent M channel augmentation, occur via the G i/o -associated pathway (33). To test which signaling cascade is used in the S9SP-mediated inhibition of LVA current in DRG neurons, we used small interfering RNA (siRNA) knock-down of either G i/o or G q/11 Ga subunits in DRG cultures ( Fig. 2A). Simultaneous knock-down of G i1-3 and G o subunits abolished the S9SP-mediated LVA current inhibition, whereas simultaneous knock-down of G q and G 11 subunits was without an effect (as compared with nontargeting ''scrambled'' siRNA control; Fig. 2A, B). All siRNA oligos reduced the abundance of their corresponding target messenger RNA (mRNA) by 60%-80% (Fig. 2C).
In accord with these findings, overnight pre-treatment of DRG cultures with the G i/o inhibitor, pertussis toxin (PTX; 400 ng/ml) completely abolished the S9SP-mediated T-type current inhibition (Fig. 2D 1 , E). A protein kinase C inhibitor, bisindolylmaleimide I (BIM I, 200 nM) was without a significant effect (Fig. 2D 2 , E). These data indicate that, similar to their effect on the M channels, NK1 receptors use the G i/omediated pathway to inhibit T-type channels in DRG neurons. Immunohistochemical localization of Ca V 3.2, NK1, and a major M channel subunit, KCNQ2 revealed that, indeed, Ca V 3.2 co-localizes with NK1 and KCNQ2 in many DRG neurons, particularly in those of a small diameter ( Supplementary Fig. S2).
Washout of S9SP (up to 5 min) did not restore LVA current amplitude (not shown) but the inhibition was completely reversed by the reducing agent dithiothreitol (DTT; 1 mM),

Innovation
We discovered a novel endogenous pathway of ''tuning'' the activity of T-type Ca 2+ channels via the modulation of its sensitivity to ambient zinc. We further suggest that simultaneous reciprocal modulation of two different ion channels by a single signaling cascade can sum up to produce a cumulative, endogenous anti-nociceptive effect. These findings describe and explain an unanticipated inhibitory effect of neuropeptide substance P (SP) in peripheral nociceptive pathways and shed new light on the mechanisms of endogenous analgesia.
FIG. 1. NK1 receptor activation inhibits T-type calcium channels in small-diameter DRG neurons. All recordings were performed from small (*20 lm) DRG neurons with a mean whole-cell capacitance of 23.0 -0.9 pF (n = 100); these neurons are predominantly (*70%) capsaicin sensitive (13,35 , similar to that shown in (B). ***Significantly different from the control, p < 0.001 (unpaired t-test). The data for the S9SP inhibition in control conditions (black circles) are the same as in Figure 1B and are included for comparison. BIM I, bisindolylmaleimide I; MIB, mibefradil; PTX, pertussis toxin; siRNA, small interfering RNA; VGCC, voltage-gated Ca 2+ channel. even in the continued presence of S9SP (Fig. 3A); DTT on its own did not significantly affect LVA current amplitude, but it rendered S9SP ineffective when both compounds were applied together (Fig. 3B). S9SP had a similar effect on recombinant Ca V 3.2 (main T-type channel subunit in DRG) that was transiently overexpressed in human embryonic kidney (HEK293) cells together with NK1 receptors (Supplementary Fig. S1A). The inhibition of T-type channels by S9SP was accompanied by a modest leftward shift of the Ca V 3.2 voltage dependence ( Supplementary Fig. S3A 1 ) and a small but significant slowing of the inactivation kinetics ( Supplementary Fig. S3A 2 ). In accord with previous findings (58), in DRG neurons, S9SP had no significant effect on the high-voltage-activated (HVA) Ca 2+ currents that are mainly conducted by N-, P/Q-, and Ltype VGCCs (55) (Supplementary Fig. S3B).
Reversibility of the S9SP-induced T-type current inhibition by DTT suggested a redox-related mechanism. Thus, we Asterisks indicate significant difference from the group indicated by the line connector with *p < 0.05, **p < 0.01, or ***p < 0.001 (paired t-test). In all bar charts, data are shown as mean -SEM. Ant.A, antimycin A; DTT, dithiothreitol; SP, substance P.
investigated whether exogenous or endogenous ROS can produce action similar to S9SP on the channels. Bath application of H 2 O 2 (300 lM) inhibited LVA current in cultured DRG neurons by *20% (Fig. 3C, D; scatter plots for inhibition in each individual neuron in this and other similar experiment series are shown in Supplementary Fig. S4). The effect of H 2 O 2 was also reversed by DTT. Application of S9SP after (and in the presence of) H 2 O 2 did not produce a significant further inhibition of the LVA current (Fig. 3D). H 2 O 2 also inhibited recombinant Ca V 3.2, an effect that was reversed by DTT ( Supplementary Fig. S1B). The mitochondrial electron transport chain (ETC) is one of the main sources of intracellular ROS (70), and SP was shown to induce ROS release from mitochondria in DRG neurons (33). Inhibition of the ETC complex III with antimycin A has been shown to cause a burst release of mitochondrial ROS (76), and antimycin A-induced ROS release augmented the activity of redox-sensitive M-type K + channels in DRG neurons (33). Consistently, application of antimycin A (1 lM) inhibited LVA current in DRG by *25% (Fig. 3E), and this effect was completely reversed by DTT (Fig. 3E 1 ). S9SP applied after the pre-application and in the continuous presence of antimycin A only had a marginal further effect (Fig. 3E 2 ).
These experiments suggest that a major component of S9SP-mediated T-type channel inhibition is mediated by endogenous ROS (although a contribution from additional mechanisms cannot be excluded).

SP acts by increasing Ca V 3.2 sensitivity to ambient zinc
Recently, we showed that hydrogen sulfide (H 2 S) inhibits Ca V 3.2 by increasing channel sensitivity to zinc (15), a trace metal that inhibits Ca V 3.2 at sub-micromolar concentrations and is present in sufficient amounts in biological fluids to affect Ca V 3.2 activity (27,66). Therefore, we tested whether a similar mechanism might account for the SP-mediated T-type channel inhibition.
Application of 750 nM ZnCl 2 caused inhibition of LVA current in DRG neurons by *25% ( p < 0.001; n = 9; Fig. 3A), but when S9SP was applied in the presence of zinc, it had no further inhibitory effect (Fig. 4A). Furthermore, S9SP did not produce any effect in the presence of the zinc chelator N,N,N¢,N¢-Tetrakis(2-pyridylmethyl)ethylenediamine (TPEN; 10 lM; Fig. 4B and Supplementary Fig. S4). S9SPinduced inhibition was completely reversed by bath application of TPEN (still in the presence of S9SP, Fig. 4C and Supplementary Fig. S4). Similar to DRG neurons, SP-and H 2 O 2 -induced inhibitions of recombinant Ca V 3.2 were prevented by TPEN ( Supplementary Fig. S1C, D).
To accept a hypothesis that SP promotes T-type channel inhibition by ambient zinc and that this effect is mediated by a redox-dependent mechanism, two questions need to be addressed. (i) Is there enough ambient zinc to mediate the observed effects (as all our solutions are nominally zinc free, unless stated otherwise)? (ii) Does SP induce any changes to extracellular or intracellular zinc levels in DRG neurons?
First, we measured the zinc content of extracellular solutions and media using the atomic absorption spectroscopy (AAS; Fig. 5A). Both the DRG culture medium and the extracellular bath solution contained micromolar concentrations of zinc (*10 lM in Dulbecco's modified Eagle medium [DMEM] and *5 lM in extracellular bath solution), which were similar to the reported plasma zinc levels in humans (44). There were no detectable changes in zinc content in the extracellular media after up to 10 min incubation of DRG cultures with S9SP ( Fig. 5A). Notably, AAS detects total zinc content and the free zinc is likely to be significantly lower than the 5-10 lM; this can explain the significant inhibition of LVA current observed in the extracellular solution supplemented with 750 nM ZnCl 2 (Fig. 4A). Next, we measured intracellular zinc levels using confocal fluorescence microscopy of DRG cultures loaded with the zinc indicator FluoZin Ô -3AM. Application of S9SP did not produce detectable changes in intracellular zinc levels; however, large responses were recorded in response to 1.5 mM ZnCl 2 or zinc ionophore zinc pyrithione (ZnPy; 10 lM; Fig. 5B, D). In a separate experiment, we detected significant intracellular zinc transients in response to acidification (pH 5.5; Supplementary  Fig. S5). Acidification has been previously shown to release intracellular zinc from cytosolic zinc-cysteine complexes (28) and, thus, acidification-induced zinc transients are likely to represent intracellular zinc release.
We then tested whether DTT can chelate zinc, as has been suggested (30,48). We first loaded DRG cultures with zinc by the application of ZnPy and then applied, in sequence, 1 mM DTT and 20 lM TPEN (still in the presence of ZnPy; Fig. 5C, E). Cell-permeable (19) DTT did not alter the elevated zinc levels, whereas TPEN completely reversed the increase in fluorescence produced by ZnPy (Fig. 5C, E). This suggests that TPEN effectively competed for zinc binding with the zinc-sensitive dye at a molar ratio of 2:1, whereas DTT was unable to produce such an effect even at a molar ratio of 200:1. The experiments presented in Figure 5 and Supplementary Figure S5 established that (i) ambient zinc is present in nominally zinc-free solutions and media at levels that are In all bar charts, data are shown as mean -SEM. DMEM, Dulbecco's modified Eagle medium; ZnPy, zinc pyrithione. To see this illustration in color, the reader is referred to the web version of this article at www.liebertonline.com/ars ‰ MODULATION OF T-TYPE CA 21 CHANNELS BY SUBSTANCE P sufficient to affect Ca V 3.2 activity (27,66); (ii) we could not detect significant changes in intracellular or extracellular zinc levels in DRG neurons in response to SP; and (iii) we could not detect zinc chelator activity of DTT. Although this last observation cannot rule out zinc binding by DTT, in principle, it argues against zinc chelation by DTT as the reason underlying the reversibility of the SP effect on Ttype channels by DTT in our experiments. Instead, we hypothesize that oxidation increases and reduction decreases channel sensitivity to zinc. We further suggest that SP induces T-type channel inhibition in DRG neurons by producing ROS-mediated modification of T-type channel protein, which, in turn, alters the channel sensitivity to zinc.
To directly test whether SP treatment changes zinc sensitivity of Ca V 3.2, we investigated concentration dependency of zinc effect on the recombinant Ca V 3.2 overexpressed in HEK293 cells together with NK1 receptors with and without S9SP treatment. We applied 10 lM TPEN in the nominally zinc-free bath solution at the beginning of each of such experiments to achieve near-zero extracellular free zinc levels and then applied increasing concentrations of ZnCl 2 (now without TPEN; Fig. 6A, B).
The concentration dependency was best fit with the twocomponent logistic function ( Fig. 6C; see Materials and Methods section). The data obtained in control conditions could be fit well with the one-component function as well [in agreement with previous reports (27,66); not shown], as the inhibition at low nanomolar zinc concentrations was small. However, a far better fit was achieved with the twocomponent logistic function, which clearly separated two zinc effects (Fig. 6C). Such bi-phasic concentration dependence is likely to indicate that Ca V 3.2 uniquely possesses high-affinity (sub-micromolar) and low-affinity (micromolar) zinc binding sites. The latter is likely to be shared with other Ca V 3 subunits and other types of VGCC channels, as these are also inhibited by micromolar (but not nanomolar) zinc levels (6,27,66).
S9SP treatment strongly increased the efficacy of zincmediated inhibition via the high-affinity mechanism while having no effect on the inhibition at higher concentrations ( Fig. 6C). The maximal inhibition at the high-affinity site in the control and S9SP-treated cells was 9.9% -11.3% and 44.1% -5.1%, respectively; the IC 50 values for high-and low-affinity sites did not change significantly. We estimated that free zinc concentrations in our solutions are in the low nanomolar range, since S9SP inhibited Ca V 3.2 in nominally zinc-free solutions by *40% (Supplementary Fig. S1A and Fig. 6); however, TPEN produced only marginal augmentation of Ca V 3.2 currents in control conditions ( Supplementary  Fig. S1C, D and Fig. 6A, B). These experiments also explain why application of 750 nM extracellular zinc abolished the S9SP effect (Fig. 4A), as at concentrations higher than 100 nM the difference in zinc-induced inhibition between control and S9SP-treated cells started to disappear (Fig. 6C). The presence of low-nanomolar free zinc in nominally zincfree solutions introduces a slight systematic error to the concentration-dependency measurements; however, we believe that the overall conclusion that SP treatment enhances inhibition of Ca V 3.2 by zinc acting at its high-affinity site is valid.
The unique high-affinity metal-binding site of Ca V 3.2 has been elucidated (27): It is composed of an Asp-Gly-His motif in the extracellular S3-S4 linker of the domain I and a separate Asp in the S1-S2 linker of the same domain ( Fig. 7A 1, A 2 ); this site is also critical for channel modulation by redox agents (26,27,47,48). His191 plays a central role in the zincmediated channel inhibition. This residue is absent in Ca V 3.1 and Ca V 3.3; accordingly, these channels are far less sensitive to zinc.
To test whether SP-mediated Ca V 3.2 channel inhibition requires this high-affinity zinc-binding site, we tested the SP sensitivity of two reciprocal mutants: Ca V 3. mutant, in contrast, displayed small but significant inhibition by S9SP (10.4% -1.6%; p < 0.05; n = 8; Fig. 7C 2 , C 3 ). In accord with previous reports, Q172H substitution in Ca V 3.1 only partially restored channel sensitivity to sub-millimolar zinc, as the entire binding site is required for the full effect (27).
We next tested the effect of Z944 and ST101 on DRG neuron excitability using the perforated current clamp technique. Z944 decreased (Fig. 8C 1 ) and ST101 increased (Fig. 8C 2 ) small DRG neuron excitability; these effects are further analyzed in Supplementary Figure S7: Z944 significantly increased the rheobase, decreased action potential (AP) firing, and hyperpolarized small-diameter DRG neurons without significantly affecting AP amplitude or duration. In contrast, ST101 significantly increased AP firing and depolarized E rest ; AP amplitude or duration was not affected (Supplementary Fig. S7A-C).
S9SP produced a pattern of effects similar to that of Z944 ( Supplementary Fig. S7A-C), although only the effect on rheobase (Supplementary Fig. S7A) reached significance. Of note, mean values and the statistical significance of the effects shown in Supplementary Figure S7 are underestimated, since not all DRG neurons express all the necessary components of the signaling cascade. Thus, *40% of small DRG neurons in our cultures express LVA currents; *40% to 50% of small neurons express functional NK1 receptors [(33), present study]; and *70% of LVA-expressing neurons respond to S9SP. Therefore, in each series of experiments, there was a subpopulation of ''unresponsive'' neurons. We further tested whether SP or a T-type Ca 2+ channel blocker would attenuate the excitatory effect of the proalgesic peptide bradykinin (BK) (36,52). BK (250 nM) significantly reduced the rheobase, increased the AP number, and depolarized the E rest ( Supplementary Fig. S7A-C). When co-applied with BK, both S9SP and Z944 completely reversed the effects of BK on the rheobase and the AP number ( Supplementary Fig. S7A, B). The BK-induced depolariza-tion in the presence of either S9SP or Z944 was more variable (Supplementary Fig. S7C), perhaps reflecting the complexity of the depolarizing effects of BK (25,36,52), but in both cases the effect of BK was no longer statistically significant.
To confirm that the effects of Z944 and ST101 on sensory neuron excitability are, indeed, mediated by T-type channels, we repeated some of the current clamp experiments in DRG neurons in which Ca V 3.2 expression has been downregulated with siRNA against Cacna1h. The procedure (see Materials and Methods section) reduced the Cacna1h transcript levels by *50% while having no effect on the housekeeping transcript (Gapdh; Supplementary Fig. S8A, B). In the Cacna1h knocked-down neurons, neither Z994 nor ST101 produced any significant effect on E rest , AP number, or rheobase (Supplementary Fig. S8C-E); whereas in scrambled oligonucleotidetransfected neurons, both drugs were still effective (cf. Supplementary Figs. S7 and S8C-E).
Taken together, the experiments reported in Figure 8 and Supplementary Figures S7 and S8 suggest that (i) T-type channels modulate excitability of many small-diameter DRG neurons; (ii) acute inhibition of T-type channels produces an anti-excitatory effect that can partially offset the excitatory action of BK; (iii) SP produces an anti-excitatory effect that is qualitatively similar to that produced by a T-type channel blocker; and (iv) effects of Z944 and ST101 on DRG neuron excitability are, indeed, mediated by T-type Ca 2+ channels.

T-type Ca 2+ channel inhibition contributes to the peripheral anti-nociceptive effect of SP in vivo
We next investigated the influence of T-type channels on peripheral sensory afferents in vivo. First, we tested whether T-type channel augmentation is, indeed, pro-nociceptive. An injection of ST101 (50 ll; 0.001-10 nmol/site) into the rat hind paw induced a moderate but significant protective effect (''nocifensive'' behavior, see Materials and Methods section) that was indicative of moderate pain. The pro-algesic effect of ST101 was smaller than that induced by BK (50 ll; 10 nmol/site), but it was significantly stronger than that of saline (Fig. 9A).
Time courses of the nociceptive effects of BK and ST101 are shown in Supplementary Figure S9. The nocifensive behavior induced by BK usually tailed out after 15 min postinjection ( Supplementary Fig. S9A); whereas the effects of ST101 showed a distinct pattern, with lower concentrations causing a delayed response and higher doses producing more immediate and shorter effects (Supplementary Fig. S9B). Such different kinetic patterns may indicate that excitations produced by BK and ST101 have distinct mechanisms.
We then evaluated the effects of Z944, ST101, and an Mtype K + channel activator, retigabine (RTG), on animal sensitivity to noxious heat using the Hargreaves' method. RTG has well-known peripheral analgesic activity (13,21,36,50), and M-channel potentiation also contributes to the endogenous anti-nociceptive effect of SP (33). A peripheral hind-paw injection of Z944 (50 ll; 0.005-0.5 nmol/site) significantly increased paw withdrawal latency on radiant heat exposure, suggesting decreased heat sensitivity (Fig. 9B). RTG (50 ll; 0.1-10 nmol/site) produced a similar effect. In contrast, ST101 significantly reduced withdrawal latency, indicating increased heat sensitivity (Fig. 9B); time courses of these effects of Z944 and RTG are given in Supplementary Figure S9C. To see this illustration in color, the reader is referred to the web version of this article at www.liebertonline.com/ars Ca V 3.2 is expressed in subpopulations of Ad and C lowthreshold mechanoreceptors (LTMRs) and plays a role in tuning sensitivity to light touch (18). Thus, we tested the effects of local application of Z944, ST101, and S9SP on a range of mechanical stimulations.
In von Frey test (see Materials and Methods section), saline-injected rats had a 60% withdrawal threshold at 9.2 -1.0 g (n = 25). Injections of Z944 (50 ll; 0.5 nmol/site) and S9SP (50 ll; 0.5 nmol/site) significantly impaired their ability to respond to stimulation, ranging from innocuous to noxious; Z944 was more efficacious in the noxious range (Fig. 9C). In contrast, an injection of ST101 (50 ll; 0.001 nmol/site) greatly increased response rates (Fig. 9C). The cotton swab assay (see Materials and Methods section) also showed significantly reduced sensitivity to light touch after the S9SP injection (50 ll; 0.5 nmol/site); Z944 (50 ll; 0.5 nmol/site) produced a similar effect (Fig. 9D). An injection of ST101, however, was ineffective in this type of test.
We then tested the contribution of T-type channel modulation to the peripheral effect of S9SP. Consistent with previous reports (33), a local hind-paw injection of S9SP (0.5-5 nmol/site) did not produce measurable nocifensive behavior; whereas BK induced a strong nocifensive response (Fig. 9E).
We then pre-injected RTG, Z944, or S9SP into the rat hind paw 5 min before the BK injection (together with the same pre-injection drug; Fig. 10A, D). RTG served as a positive control, as it was shown to offset excitatory effects of BK and to reduce the BK-induced nociception (33,34,36). Preinjections of both RTG (50 ll; 1-10 nmol/site) and Z944 (at 0.5 nmol/site but not at 0.05 nmol/site) significantly attenuated the BK-induced nociception. A pre-injection of S9SP (0.5 nmol/site) also strongly suppressed the BK-induced nocifensive behavior (Fig. 10A). Importantly, a mixture of RTG (1 nmol/site) and Z944 (0.5 nmol/site) was as efficacious as S9SP in reducing BK-induced nocifensive behavior To see this illustration in color, the reader is referred to the web version of this article at www.liebertonline.com/ars and showed a tendency to be more efficacious than RTG or Z944 injected alone ( p = 0.059 vs. RTG alone; Fig. 10A); this finding is consistent with a hypothesis that M-channel augmentation and T-type channel inhibition can be cumulative components of peripheral anti-nociceptive effects of SP. Finally, we performed in vivo knock-down of Ca V 3.2 in DRG neurons by an intrathecal injection of antisense (AS) phosphorodiester oligodeoxynucleotides (ODNs) using the protocol that had been successfully used earlier for the same purpose [ (3,41); Materials and Methods section]. In salineand mismatch ODN-injected animals, both Z944 (0.5 nmol/ site) and S9SP (0.5 nmol/site) still significantly attenuated nocifensive behavior produced by the hind-paw BK injection. However, in Ca V 3.2 AS ODN-injected animals, the nocifensive behavior produced by both Z944 and S9SP was significantly reduced (Fig. 10B). The levels of Ca V 3.2 transcript in lumbar DRGs of Ca V 3.2 AS ODN-injected animals were reduced by *70% as compared with the saline-injected controls and mismatch ODN-injected animals (Fig. 10C).

Discussion
Here, we described a hitherto unknown endogenous signaling pathway linking NK1 receptors with Ca V 3.2 channel activity and demonstrated that this pathway is involved in the paradoxical peripheral anti-nociceptive effect of SP. T-type Ca 2+ channels are sensitive to redox modulation; oxidizing agents or ROS inhibit channel activity (47,65), whereas reducing agents such as L-cysteine or DTT reverse such inhibition and can augment channel activity above the basal level in some cases [(46, 65); Fig. 3 and Supplementary Fig. S1].
SP induces ROS release in immune (64) and epithelial (60) cells and in DRG neurons (33). Accordingly, SP-induced inhibition of T-type Ca 2+ channels observed in this study  displayed the following features: (i) It was fully reversed or prevented by DTT (Fig. 3); (ii) mimicked by exogenously applied H 2 O 2 ( Fig. 3 and Supplementary Fig. S1); and (iii) stimulated by endogenous ROS release (Fig. 3E). The mechanism of the redox sensitivity of T-type Ca 2+ channels is not entirely understood, but it requires the extracellular His191 in Ca V 3.2; the other Ca V 3 isoforms do not have a corresponding histidine and are largely redox insensitive (47). The same H191 mediates Ca V 3.2 inhibition by sub-micromolar concentrations of zinc (26,27). H191 is a general metal-binding site, and it was hypothesized that it could participate in the metal-catalyzed oxidation (MCO) reaction accelerated by transition metals (47,61).
It is also possible that an oxidative modification (possibly but not necessarily at H191) can promote zinc binding or potentiate its action on the channel, thus sensitizing it to ambient concentrations of this trace metal. The following observations are in favor of this latter mechanism. (i) SP induced a large increase in zinc-mediated inhibition at the high-affinity site in Ca V 3.2 (Fig. 6). (ii) A mutation of H191 in Ca V 3.2 rendered the channels insensitive to SP (Fig. 7). (iii) Ca V 3.1 is insensitive to SP but insertion of His into the position equivalent to H191 in Ca V 3.2 renders the Ca V 3.1 channel sensitive to SP (Fig. 7). (iv) The zinc chelator TPEN prevents or reverses SP-induced T-type Ca 2+ current inhibition ( Fig. 4 and Supplementary Fig. S1). (v) Extracellular zinc inhibits LVA currents in DRG neurons, an effect that is similar in amplitude to that produced by SP (Fig. 4). (vi) Extracellular zinc occludes any further inhibitory actions of SP (Fig. 4). (vii) We could not detect any changes in intracellular or extracellular zinc levels induced by SP (Fig. 5 and Supplementary Fig. S5). (viii) Nominally zinc-free media contain low-nanomolar concentrations of free zinc, which are sufficient to affect channel activity when the channel is sensitized by SP (Fig. 6); however, these levels are too low to significantly inhibit T-type channels in their tonic state. The last observation would explain why TPEN produced only a marginal effect on T-type currents under basal conditions (Figs. 4B and 6B and Supplementary Fig. S1C, D).
Despite previous suggestions (30, 48), we could not detect zinc chelation by DTT under our experimental conditions (Fig. 5). Therefore, we suggest that the reversal of the SPmediated T-type channel inhibition by DTT is due to the reducing activity of the latter.
Interestingly, there are three extracellular cysteines (residues that can be easily modified by redox mechanisms) in the IS1-IS2 linker of Ca V 3.2, a domain that contributes to the high-affinity zinc-binding site (27). We hypothesize that cumulative oxidation of such extracellular cysteines may result in a graded increase in channel sensitivity to zinc. In such a scenario, Ca V 3.2 tonically exists in a partially ''oxidized'' state with intermediate zinc sensitivity, which is, however, too low for inhibition by ambient zinc. This could be why in our hands TPEN or DTT had only marginal effects on channel activity under control conditions.
The partially oxidized state can then be converted to a fully oxidized state, with the highest zinc sensitivity by SP or ROS, or to a fully reduced state, with the lowest zinc sensitivity by DTT. This could account for the augmentation of tonic Ca V 3.2 currents by DTT (65) and TPEN (15) that are observed in some labs, as different experimental conditions may result in different ''tonic'' redox states of the channel (e.g., DTT or TPEN would potentiate fully oxidized channels). This hypothesis requires further verification.
The levels of NK receptor expression in peripheral sensory neurons are reportedly low. Thus, one immunohistochemical study found little evidence for NK receptor immunoreactivity in rat sensory neurons (5); whereas recent RNAseq experiments found very low levels of NK receptor transcripts in mouse DRG (39,68), although the levels were higher in human DRG (16). However, in both mouse (39) and human (16) DRG, the NK receptor transcript levels were comparable (or even higher) to those of the brain or lung, tissues where prominent SP signaling is well characterized. Moreover, the DRG NK receptor levels are comparable to those of BK B 2 receptors (16,39), which exert robust effects in DRG neurons [reviewed in Petho and Reeh (52)].
Perhaps low transcript levels and expression in only specific cell populations are sufficient for functional activity of such G-protein-coupled receptors. In support of this notion, in vitro and, particularly, in vivo evidence presented here and elsewhere (32,33,73) clearly demonstrate the presence of functional SP signaling in DRG. In addition, NK1 receptor immunoreactivity was found throughout DRG neuron somata and often co-localized with that of Ca V 3.2, especially in small-diameter neurons ( Supplementary Fig. S2).
There is also some discrepancy regarding cellular effects of SP in cultured DRG neurons. Thus, some potentially excitatory effects, including depolarization (1, 74), potentiation of Nav1.8 Na + channels (7), L-and N-type Ca 2+ channels (58), and sensitization of TRPV1 (33,57,73), were reported. However, there is no acute pain produced by local SP injections [(33); present study]. Application of SP to human skin by dermal microdialysis produces no painful sensation or primary afferent excitation (72). Thus, at present, the excitatory effects of SP in cultured neurons are difficult to reconcile with the lack of behavioral responses to local applications of SP. However, excitatory actions of SP may underlie responses of neurons innervating some internal organs, for example, bladder (74).
Opposing modulation of T-type channel activity by the latest generation of selective T-type channel modulators, ST101 (potentiator) and Z944 (inhibitor) confirmed the role of this channel in the control of sensory neuron excitability ( Fig. 8 and Supplementary Figs. S7 and S8). Accordingly, a peripheral injection of ST101 induced moderate ''spontaneous'' nociception and increased sensitivity to both light and noxious touch, whereas Z944 reduced heat and touch sensitivity (Fig. 9); these effects are consistent with the analgesic efficacy of Z944 (31).
Importantly, we now show that both S9SP and Z944 significantly attenuate nociception induced by a peripheral injection of BK and in vivo knock-down of Ca V 3.2 in DRG reduces such attenuation (Fig. 10). Co-application of Z944 with M-channel opener RTG produced a similar antinociceptive effect as that of S9SP (Fig. 10).
Since both M-type (33) and T-type (present study) channels are modulated by SP (although in opposite directions) via a signaling cascade that shares common initial steps, we suggest that both M-channel augmentation and T-type channel inhibition contribute to the redox-mediated antinociceptive action of SP in peripheral nerves (Fig. 11). We hypothesize that since peptidergic nociceptive fibers release SP locally in response to noxious stimulation, this neuropeptide may exert local endogenous analgesia in an autocrine/paracrine fashion by simultaneous inhibition of T-type Ca 2+ channels and augmentation of M channels in nociceptors.
Although we have previously shown that SP does induce mitochondrial ROS release in DRG neurons (33), the release of ROS from other, non-neuronal NK1-expressing cells surrounding the site of tissue damage cannot be excluded (particularly in the in vivo experiments), especially given the fact that the redox-sensitive site in T-type channels (but not M-type channels) is located extracellularly. Additionally, activation of NK receptors in DRG may result in the release of other substances that can modulate T-type channel activity, such as thioredoxin (4) or H 2 S (14,15). Interestingly, both factors inhibit Ca V 3.2 channels by interfering with their extracellular zinc-binding site (4,14,15).
Finally, although we could not detect any changes in intracellular or extracellular zinc levels in response to SP in vitro, one cannot exclude that some degree of redoxmediated zinc liberation from intracellular ligands, such as metallothioneins (28), or even extracellular zinc-bound proteins could occur (especially in the in vivo situation). Such zinc release could potentially contribute to the SP-mediated T-type channel inhibition, although in the case of intracellular zinc, it would need to be transported to the extracellular space to be able to affect T-type channel activity.
Clearly, further research is needed to address these intriguing questions; nevertheless, the present study firmly establishes redox-dependent modulation of T-type Ca 2+ channels in nociceptive neurons as a mechanism contributing to the acute anti-nociceptive action of SP in the peripheral nociceptive system.

Materials and Methods
Cell culture, transfections, complementary DNA constructs, and chemicals DRG neurons from adult male Sprague-Dawley rats (170-180 g) were dissociated as previously described (13,29) and cultured in DMEM supplemented with GlutaMax I, 10% fetal calf serum, penicillin (50 U/ml), and streptomycin (50 lg/ml) on glass coverslips that were coated with poly-d-lysine for 2-5 days in a humidified incubator (37°C, 5% CO 2 To knock down Ca V 3.2, DRGs were dissociated as described (13); immediately after dissociation, DRG pellets were resuspended in 900 ll DMEM (without GlutaMax I, serum or antibiotics), to which a mixture of 9 ll Lipofectamine RNAi-MAX (Invitrogen) and either a mix of three anti-Cacna1h Stealth siRNAs (RSS350285, RSS350286, RSS350287; Life Technologies; 200 nM each) or a scrambled control oligo (Stealth RNAi Ô siRNA Negative Control; Life Technologies; 200 nM) were added immediately before resuspension. Cultures were than plated in 50 ll droplets onto glass coverslips that were coated with poly-d-lysine and incubated for 5-6 h; the transfection media were then replaced with full DRG culture medium. Patch-clamp recordings were conducted 24-48 h after the transfection. The following primers were used for reverse transcription polymerase chain reaction (RT-PCR) analysis: Cacna1h sense: 5¢-TGCCCACGGAGTCTATGAGT-3¢ Cacna1h AS: 5¢-GTTGTAGGGGTTCCGGATGT-3¢ Gapdh sense: 5¢-GACATGCCGCCTGGAGAAAC-3¢ Gapdh AS: 5¢-AGCCCAGGATGCCCTTTAGT-3¢ To knock down G i/o or G q/11 Ga subunits, Silencer Select RNAi siRNA (Invitrogen) oligonucleotides were used; the FIG. 11. Schematic of the proposed peripheral antinociceptive signaling via NK1 receptors. Local tissue damage excites peptidergic nociceptors that release SP into the tissue damage area. Acting at the NK1 receptors expressed in the neighboring fiber endings (potentially including the SPreleasing fibers as well), SP triggers mitochondrial ROS release and redox-dependent inhibition of T-type Ca 2+ channels and potentiation of M-type K + channels, resulting in hyperpolarization and reduced excitability of the fibers. ROS, reactive oxygen species. To see this illustration in color, the reader is referred to the web version of this article at www.liebertonline. com/ars Electrophysiology All recordings were made using Multiclamp 700B amplifier in combination with pCLAMP 10.4 software (Molecular Devices). Voltage-clamp recordings from DRG neurons were made using whole-cell patch clamp. The standard bath solution contained the following (in mM): 150 TEA-Cl; 2.5 CsCl; 2.5 CaCl 2 ; 10 HEPES; 0.5 MgCl 2 ; and 10 glucose (pH 7.4 adjusted with CsOH). The standard pipette solution contained the following (in mM): 110 CsCl; 3 MgCl 2 ; 10 EGTA; 10 HEPES; 3 Mg-ATP; and 0.6 GTP (pH 7.4 adjusted with CsOH). The access resistance was typically within 4-8 MO. T-type calcium currents (LVA) were measured by 50 ms square voltage pulses to -40 mV from a holding potential of -90 mV. HVA currents were measured by 200 ms square voltage pulses to 0 mV from a holding potential of -90 mV.
In recordings from HEK 293 cells, amphotericin Bperforated patch-clamp technique was used (13). The pipette solution contained the following (in mM): 155 CsCl; 10 HEPES; 1 EGTA; and 4 MgCl 2 supplemented with amphotericin B (250 lg/ml), pH 7.4 adjusted with CsOH. The bath solution was the same as for DRG neuron recordings. The access resistance was typically within 8-12 MO.
Voltage-clamp recordings were sampled at 4 kHz. Series resistance was compensated online by 50%; the voltage error due to series resistance in our recordings did not exceed 6 mV and typically was within *2 mV.
In experiments with H 2 O 2 , 3% agar-salt (3 M KCl) bridges were used. Concentration dependence of Ca V 3.2 channel inhibition by zinc was fit to the two-component logistic function: where N 0 is the total amount of zinc-binding sites, N 0(max1) is the amount of high-affinity binding sites, N 0(max2) is the amount of low-affinity binding sites, I max is the maximal Ca V 3.2 peak current amplitude, IC 50 is the [Zn 2+ ] necessary for half-maximal inhibition, and nH is the Hill coefficient. The concentration dependency of the S9SP effect was fit to the one-component logistic function: where y is the current inhibition; A 1 and A 2 are the maximum and minimum effects, respectively; x is the S9SP concentration; and nH is the Hill coefficient. Neuronal excitability was studied using the amphotericin B-perforated patch-clamp current technique (13). The extracellular solution contained the following (in mM): 160 NaCl, 2.5 KCl, 5 CaCl 2 , 1 MgCl 2 , 10 HEPES, and 8 Glucose. The intracellular solution contained the following (in mM): 145 KCl, 15 NaCl, 5 MgCl 2 , 10 HEPES, and 250 lg/ml Amphotericin B. The access resistance was typically within 8-12 mO. APs were generated by an injection of 200 pA current for 1 s from a holding current of 0 pA. Rheobase was determined from a train of current steps from 0 to 400 pA delivered in 20-pA increments. Current clamp recordings were sampled at 12.5 kHz. Bridge balance adjustments were not applied, since only relative changes in E rest , AP number, and rheobase were investigated.

Measurement of zinc concentrations
The concentrations of zinc in experimental solutions were determined using an automated Ion Channel Reader 8000 flame atomic absorption spectrometer (Aurora Biomed) according to the manufacturer's instructions. To test whether SP induces zinc release from the cultured DRG cells, DRG cultures were grown on 10 mm round coverglass in 24-well plates; cultures were washed with extracellular solution (EC) and incubated in 0.5 ml EC that was supplemented with 1 lM S9SP for 2-10 min; and supernatants were collected and analyzed for zinc content.

Zinc imaging
Changes in the intracellular zinc levels were evaluated using fluorescence imaging. DRG cultures were loaded with FluoZin-3AM (5 lM for 30 min at 37°C in the presence of 0.02% pluronic F-127). Cells were washed with extracellular bath solution composed of the following (in mM): 160 NaCl; 2.5 KCl; 2 CaCl 2 ; 1 MgCl 2 ; and 10 HEPES (pH 7.4 adjusted with NaOH); they were imaged using a Leica SP5 fluorescence imaging system assembled on a Leica DMI6000 microscope and illuminated with 488 nm light for 1200 ms (400 Hz) with a 2 s interval. Solutions were applied using a gravity perfusion system.
For measurement of thermal hyperalgesia, the change in the latency of hind-paw withdrawal in response to noxious heat was recorded using the Hargreaves' plantar method (Ugo Basile) at 8 min after the hind-paw injection of drugs; the heat source was set to 25% of the maximal intensity.
For measurement of sensitivity to gentle touch and mechanical hyperalgesia, two methods were used. (i) Cotton swab test was performed as previously described (53). Briefly, rats were allowed to acclimate in a cage with a wire mesh bottom for at least 30 min and then were manually probed with a ''cotton swab'' pulled from a Q-tip (to approximately three times the original size). A swab was applied in a sweeping motion underneath the rats' paws, and the withdrawal responses were recorded. Five sweeps were applied in each test, with at least 10 s between each sweep; withdrawal score was calculated as a percentage of positive responses. (ii) The von Frey test was performed using a calibrated range of filaments with different bending forces (2, 4, 6, 8, 10, and 15 g; Stoelting Co.). A single filament was applied perpendicularly to the plantar surface of the hind paw for five times, with an interval of 5 s, and it was scored for withdrawal responses. Each force was probed five times, the number of withdrawals out of five trials was counted, and the percentage of withdrawal responses was calculated. Both types of assays were performed 8 min after the hind-paw drug injection.

Intrathecal administration of ODNs targeting Ca V 3.2 subunit
The in vivo Ca V 3.2 knock-down was performed as described earlier (3) by using the same ODN sequences that were synthesized by SBS Genetech. AS ODNs had the following sequence: CCACCTTCTTACGCCAGCGG. The mismatch ODNs of the following sequence were used as a negative control: TACTGTACTTGCGAGGCCAC. The mismatch ODN was not complementary to any known nucleotide sequence. ODNs were reconstituted in saline before injection. ODNs (12.5 lg/rat in 10 ll of saline) or saline were injected intrathecally (i.t.) between the L5 and L6 dorsal spinous processes as described earlier (3,42) under isoflurane anesthesia. The treatment was repeated twice daily for 4 days. The pain tests were performed on the next day after the last injection.
After behavioral testing, animals were humanly sacrificed; lumbar DRGs were extracted and analyzed for the knockdown efficiency by comparing the Cacna1h mRNA levels in lumbar (L4 to L6) DRGs of the saline-injected, mismatch ODN-injected, and AS ODN-injected animals using RT-PCR. The same Cacna1h and Gapdh primers as in siRNA experiments were used for RT-PCR.

Statistics
All data are given as meanstandard error of the mean. Differences between groups were assessed by Student's t test (paired or unpaired, as appropriate) or one-way analysis of variance (ANOVA) with Bonferroni correction. Behavioral experiments were analyzed using Kruskal-Wallis ANOVA; difference between groups was analyzed with Mann-Whitney test. The effects of the drugs on neuronal excitability were analyzed using either paired t test or paired Wilcoxon signed ranks test (for data sets that failed normality test). The differences were considered significant at p £ 0.05. Statistical analyses were performed using Origin 8.6 (OriginLab Corporation).