PKCε phosphorylation of the sodium channel NaV1.8 increases channel function and produces mechanical hyperalgesia in mice (original) (raw)
Identification of PKCε substrates in lumbar DRGs. To identify PKCε substrates with high specificity, we generated an ATP analog-specific mutant of PKCε, _AS-_PKCε (9). We engineered this mutation (M486A) to be functionally silent with respect to kinase activity and substrate specificity but to allow use of an ATP analog to deliver a chemical tag to substrates. The analog contains two modifications: a side group at the N6 position of the adenine moiety, which allows preferential binding to an analog-specific kinase (_AS-_kinase), and a phosphate mimetic (thiophosphate) at the γ-phosphate of ATP to generate the kinase-transferable tag. The thiophosphate tag is unique in that it resists phosphatases and can be chemically distinguished from other functional groups by alkylation, followed by detection with a specific antibody that recognizes thiophosphate esters (14).
Incubation of mouse DRG lysates with N6-benzyl ATP-γS and _AS_-PKCε resulted in thiophosphorylation of several proteins identified by Western blot analysis. Incubation of lysates in the absence of _AS_-PKCε or in the presence of a specific _AS_-kinase inhibitor, 1-naphthyl-4-amino-1-ter-butyl-3-(p-methylphenyl)pyrazolol[3,4-d]pyrimidine (1Na-PP1), prevented thiophosphorylation, indicating that it was mediated by _AS_-PKCε (Figure 1A). The general PKC activator phorbol 12-myristate, 13-acetate (PMA) did not increase thiophosphorylation, further suggesting that endogenous activators present in the lysate were sufficient to activate _AS_-PKCε. To reduce sample complexity and enrich for low-abundance proteins, we performed solution-phase isoelectric focusing (15) and then separated proteins by SDS-PAGE for Western blot analysis (Figure 1B). We focused attention on proteins of molecular mass greater than 100 kDa to identify large membrane proteins that might be substrates. Coomassie blue–stained bands (Figure 1B, left) that matched immunoreactive bands (Figure 1B, right) were excised, and proteins in excised gels were identified by tandem mass spectrometry. From an excised band at approximately 200 kDa in the pH 6.2–7.0 fraction, we identified Nav1.8 as a potential PKCε substrate.
Screening of PKCε substrates. (A) Western blot analysis with anti-thiophosphate ester antibody, showing that thiophosphorylation of lumbar DRG lysates by _AS_-PKCε was blocked by the _AS_-kinase inhibitor, 1Na-PP1 (1-Na). (B) Thiophosphorylated proteins were separated by solution-phase isoelectric focusing into 5 pools (isoelectric point [pI] ranges of each pool are shown at bottom). Proteins separated by SDS-PAGE were detected by Western blot analysis with anti-thiophosphate ester antibody (right); parallel gels were stained with Coomassie blue (left). The asterisk indicates a band at approximately 200 kDa in the 6.2–7.0 pI pool that was subsequently identified as NaV1.8 by mass spectrometry.
Native PKCε colocalizes with native Nav1.8. If PKCε phosphorylates Nav1.8 in vivo, then both proteins should be expressed in the same cells. In adult rat DRG neurons, we found that PKCε immunoreactivity was mainly distributed in small- to medium-diameter neurons, while NaV1.8 immunoreactivity was mostly found in small- and medium-diameter neurons together with PKCε immunoreactivity (Figure 2A). We also found that endogenous PKCε could be coimmunoprecipitated from DRG lysates by an anti-NaV1.8 antibody (Figure 2B). These results indicate that PKCε colocalizes and interacts with Nav1.8 in DRG neurons.
Native Nav1.8 sodium channels are colocalized with and phosphorylated by PKCε. (A) Nav1.8 (green) and PKCε (red) colocalize in small- and medium-sized DRG neurons (yellow). Areas labeled 1 and 2 in the merged image are shown at higher magnification in the bottom panels. Scale bar: 200 μm (top); 25 μm (bottom). (B) Lumbar DRG lysates were immunoblotted directly (Input) or immunoprecipitated (IP) with PKCε, IgG, or anti-Nav1.8 antibodies, and then the immunoprecipitates were subjected to Western blot (WB) analysis with anti-PKCε antibody. (C) PKCε-mediated thiophosphorylation of proteins immunoprecipitated from lumbar DRG lysates with anti-Nav1.8 or control IgG and assayed in the presence or absence of the PKC inhibitor bisindolylmaleimide I (BIS; top). The Western blot in the bottom panel shows that equal amounts of immunoprecipitated Nav1.8 were used in the kinase assay. Thiophos. ester, thiophosphate ester.
Identification of a PKCε phosphorylation site in Nav1.8. To determine whether PKCε phosphorylates native NaV1.8 channels, we immunoprecipitated NaV1.8 from rat lumbar DRG lysates (Figure 2C, bottom) and incubated the immunoprecipitate with recombinant PKCε and ATP-γS. PKCε phosphorylated Nav1.8 in vitro, and this phosphorylation was inhibited by the general PKC inhibitor bisindolylmaleimide I (Figure 2C, top). To identify sites of phosphorylation, we expressed and purified all intracellular domains (Figure 3A) of rat Nav1.8 as 6xHis-tagged fusion proteins (Figure 3B) for in vitro phosphorylation by recombinant PKCε. The L3 loop appeared to be the best PKCε substrate (Figure 3B). Although several other phosphorylated bands could be detected in samples of N terminus, L1, and L2 fusion proteins, the molecular masses of these phosphoproteins did not match those of the fusion proteins, as determined by Western blot analysis with an anti-His tag antibody (Figure 3B), suggesting that they were bacterial proteins and not intracellular domains of NaV1.8.
PKCε phosphorylates the third intracellular loop of Nav1.8. (A) Schematic diagram illustrating the structural topology common to all eukaryotic sodium channels. (B) Intracellular domains of NaV1.8 were expressed in bacteria as 6xHis-tagged fusion proteins, and their expression was confirmed by Western blot analysis (left) with an anti-6xHis antibody (N terminus [N], ~24 kDa; L1, ~38 kDa; L2, ~39 kDa; L3, ~11 kDa; C terminus [C], ~35 kDa). Fusion proteins were used in a PKCε assay to determine whether any were PKCε substrates (right). An autoradiogram illustrates that the L3 loop (~11 kDa) is a likely PKCε substrate. Similar amounts of each fusion protein were used in Western blots (left) and kinase assays (right).
As shown in Figure 4A, PKCε phosphorylated the intracellular Nav1.8/L3 loop at a rate similar to phosphorylation of the major intracellular loop of the GABAA γ2S subunit, which contains a PKCε phosphorylation site at S327 (9). Similar to GABAA γ2S, the Nav1.8/L3 loop was phosphorylated to a maximal stoichiometry of 0.95 ± 0.08 (n = 3), suggesting that it is a true PKCε substrate. Since there are only 2 potential PKC phosphorylation sites, T1437 and S1452, in the L3 loop, we generated 2 alanine substitution mutants, L3-T1437A and L3-S1452A, and examined their phosphorylation by PKCε in vitro. The L3-S1452A mutation markedly decreased PKCε-mediated phosphorylation, whereas the L3-T1437 mutation did not (Figure 4, B and C). This result indicates that S1452 in the L3 loop can be phosphorylated by PKCε in vitro. We noticed that the S1452A mutation did not completely block phosphorylation of the L3 fusion protein (Figure 4C). This may have been due to weak phosphorylation of non-loop residues within the 6xHis tag, which contains 5 serine residues (MGSSHHHHHHSSGLVPRGSHM).
Identification of PKCε phosphorylation sites in the L3 loop. (A) The top 2 panels show an autoradiogram of phosphorylated intracellular L3 loop (p-L3) and a scanned image of a Coomassie blue–stained gel before autoradiography (L3). The bottom 2 panels are an autoradiogram and gel of PKCε phosphorylation of the large intracellular loop of the GABAA γ2 subunit (p-γ2s), a known PKCε substrate (9). (B) An autoradiogram and Western blot with anti-6xHis antibody, showing that PKCε phosphorylation of the NaV1.8 L3 loop was substantially reduced by alanine substitution at S1452 but not at T1437. (C) Quantification of L3 loop phosphorylation by PKCε. Results were calculated as the ratio of optical density values for phosphorylation (determined by autoradiography) and immunoreactivity (determined by Western blot analysis with anti-6xHis antibody) and were normalized to values obtained for native L3 loop run in parallel (mean ± SEM values from 3 experiments). *P = 0.0002 compared with a theoretical mean of 1.0 by 1-sample t test.
PKCε phosphorylation of S1452 enhances Nav1.8 channel function. To determine whether PKCε phosphorylation of S1452 regulates the function of NaV1.8, we functionally expressed NaV1.8 in ND7/23 cells, which are a hybrid cell line derived from rat DRG neurons and mouse N18TG2 neuroblastoma cells (16) and were previously used to express Nav1.8 (16–21). We conducted these studies in the presence of 300 nM tetrodotoxin (TTX) to block endogenously expressed, voltage-gated, TTX-sensitive (TTX-S) sodium channels (Figure 5A). As shown in Figure 5B, we detected a TTX-R voltage-gated sodium current in Nav1.8-transfected cells (peak current, 2,279 ± 411 pA; n = 22). In cells expressing wild-type Nav1.8, activation of PKCε with the ψεRACK peptide (n = 18 cells) increased the current density by 76% over that of the control condition (n = 39 cells), while a scrambled ψεRACK peptide (n = 19 cells) had no effect (H = 11.09, P = 0.0039). Likewise, in cells that expressed the T1437A mutant, ψεRACK (n = 24 cells) increased the current density by 59% over that of the control condition (n = 24 cells), while the scrambled ψεRACK peptide (n = 17 cells) was ineffective (H = 7.03, P = 0.0298). In contrast, in cells expressing the S1452A mutant, ψεRACK (n = 20 cells) failed to increase the current density over the current measured in the control condition (n = 19 cells), and the scrambled peptide (n = 16 cells) again had no effect (H = 0.033, P = 0.9836). These findings indicate that phosphorylation at S1452 is required for PKCε to increase Nav1.8 function.
PKCε enhances Nav1.8 currents in ND7/23 cells. (A) ND7/23 cells express an endogenous Na+ current (left) that can be blocked with TTX (middle), while NaV1.8-transfected ND7/23 cells express a TTX-R Na+ current (right). (B) Alanine substitution at S1452 prevents PKCε enhancement of Nav1.8 current density. Histograms show mean ± SEM values from 18 to 39 recordings for each condition from 4–6 independent experiments. Treatment with ψεRACK in the patch pipette increased the current density in cells expressing native NaV1.8 or the T1437A mutant but not in cells expressing the S1452A mutant. *P < 0.05 compared with control and scrambled ψεRACK-treated cells by Dunn’s multiple comparison tests.
PKCε increases Nav1.8 currents in DRG neurons. Adult small-diameter DRG neurons express at least 2 TTX-R sodium channels, Nav1.8 and Nav1.9 (12, 22, 23), which can be separated by applying different holding potentials and further identified by their inactivation kinetics (24). We were able to isolate slowly inactivating NaV1.8 currents by using a holding potential (Vh) of –70 mV (Figure 6A, right). Persistent Nav1.9 TTX-R Na+ currents (Figure 6A, left) could be recorded in DRG neurons from Scn10a–/– mice, which lack Nav1.8 channels, when the Vh was –120 mV (n = 5 neurons), whereas no TTX-R Na+ currents (n = 20 neurons) could be recorded when the Vh was –70 mV (Figure 6A, middle), consistent with published data (24). In Prkce+/+ DRG neurons, administration of ψεRACK increased the peak NaV1.8 current density by approximately 65%, while the scrambled ψεRACK peptide had no effect (H = 9.256, P = 0.0098; Figure 6, B and C). In contrast, neither ψεRACK nor the scrambled ψεRACK peptide altered the Nav1.8 peak current density in Prkce–/– DRG neurons (H = 0.935, P = 0.6265).
ψεRACK enhances Nav1.8 current in wild-type but not Prkce–/– DRG neurons. (A) Families of current traces recorded in the presence of 300 nM TTX from cells depolarized to a range of voltages (–70 to +50 mV) from a holding potential of –120 mV (left) to elicit NaV1.9 currents or from –70 mV (middle and right) to elicit NaV1.8 currents from Scn10a–/– (middle) or Scn10a+/+ (right) neurons. (B) Voltage-clamp recordings in Prkce+/+ or Prkce–/– neurons incubated with ψεRACK or scrambled ψεRACK using a holding potential of –70 mV. (C) Compared with that in untreated control Prkce+/+ neurons (n = 23), administration of ψεRACK (n = 30) increased the peak sodium current, while scrambled ψεRACK (n = 20) had no effect (*P < 0.05 versus Prkce+/+ control cells or Prkce+/+ cells treated with scrambled ψεRACK by Dunn’s multiple comparison test). Neurons were obtained from 5 to 8 mice of each genotype.
PKCε modifies the voltage dependence of NaV1.8 channel activation and inactivation. NaV1.8 currents in Prkce+/+ and Prkce–/– DRG neurons showed similar activation and inactivation kinetics, which were not significantly altered by ψεRACK (Supplemental Figure 1, A and B; supplemental material available online with this article; doi:10.1172/JCI61934DS1). However, ψεRACK caused a significant leftward shift (11 mV) in the activation curve in Prkce+/+ neurons (Figure 7A and Supplemental Table 1), decreasing the voltage that elicits half-maximal activation (_V_1/2) in Prkce+/+ neurons but not in Prkce–/– neurons, while scrambled ψεRACK had no effect. Analysis of these data showed no significant main effect of genotype (Fgenotype [1,29] = 3.74, P = 0.0629), but there was a significant effect of treatment (Ftreatment [2,29] = 12.05, P = 0.0002) and a significant interaction between genotype and treatment (Fgenotype × treatment [2,29] = 11.07, P = 0.0002). In addition, ψεRACK produced a small (3.7 mV) but statistically significant depolarizing shift in the steady-state inactivation curve in Prkce+/+ neurons (Figure 7B and Supplemental Table 1), increasing the voltage that elicits half-maximal inactivation (_V_1/2) in Prkce+/+ but not in Prkce–/– neurons (_F_genotype [1,44] = 92.85, P < 0.0001; _F_treatment [1,44] = 5.99, P = 0.0185; _F_genotype × treatment [1, 44] = 15.38, P = 0.0003). Thus, activation of PKCε shifted Nav1.8 voltage dependence of activation, permitting the channel to open in response to weaker depolarizations, and inactivation to increase channel availability in the potential domain between –50 mV and –15 mV.
PKCε alters the voltage dependence of activation and steady-state inactivation of NaV1.8 channels. (A) ψεRACK shifted the voltage dependence of activation to more negative voltages in Prkce+/+ (WT) neurons but not in Prkce–/– (KO) neurons. (B) ψεRACK shifted the voltage dependence of steady-state inactivation to more positive voltages. Best-fitted curves of activation and steady-state inactivation were generated by the Boltzmann distribution equation.
ψεRACK peptide evokes a PKCε-dependent hyperalgesia in mice. Intraplantar injection of ψεRACK in rats produces mechanical hyperalgesia that can be antagonized by the general PKC inhibitor bisindolylmaleimide I or by the selective PKCε inhibitor peptide εV1-2 (3). To determine whether ψεRACK can also be used to model PKCε-dependent pain responses in mice, we compared mechanical hyperalgesia, thermal hyperalgesia, and nocifensive behaviors in Prkce+/+ and Prkce–/– mice after intraplantar injection of this peptide.
Thermal hyperalgesia produced by ψεRACK was measured using the Hargreaves method 1 hour after intraplantar injection and was compared with baseline measurements made in the same paws 1 day prior (Figure 8A). ψεRACK reduced the paw withdrawal latency compared with baseline responses in Prkce+/+ mice but not in Prkce–/– mice (_F_genotype [1,16] = 7.43, P = 0.015; _F_treatment [1,32] = 4.49, P = 0.0501; _F_genotype × treatment [1,16] = 5.07, P = 0.0387). There was no difference in baseline thermal sensitivity between Prkce+/+ and Prkce–/– mice. These data demonstrate that ψεRACK produces a thermal hyperalgesia that is PKCε dependent.
PKCε-dependent mechanical hyperalgesia is substantially reduced in Scn10a–/– mice, which lack Nav1.8 channels. (A) Pretreatment with ψεRACK reduced the latency to withdraw the paw upon thermal stimulation in Prkce+/+ mice (n = 8) but not in Prkce–/– mice (n = 10) (*P < 0.05 compared with wild-type baseline or Prkce–/– mice treated with ψεRACK). (B) ψεRACK increased the response to von Frey filament stimulation in wild-type mice (n = 6) but not in Prkce–/– mice (n = 6) (*P < 0.05 compared with other conditions). (C) Nocifensive behavior lasted longer after administration of ψεRACK than after administration of saline in wild-type mice (n = 5) but not in Prkce–/– mice (*P < 0.05 compared with wild-type mice treated with saline or Prkce–/– mice treated with ψεRACK). (D) ψεRACK reduced the latency to withdraw the paw upon thermal stimulation in both wild-type mice (n = 12) and Scn10a–/– mice (n = 12). (E) ψεRACK increased the response to von Frey filament stimulation in wild-type mice (n = 14) but not in Scn10a–/– mice (n = 12) (*P < 0.01 compared with other conditions). (F) ψεRACK elicited nocifensive behavior that lasted for a similar amount of time in wild-type mice (n = 6) and in Scn10a–/– (n = 3) mice; the scrambled ψεRACK peptide did not elicit spontaneous pain in either genotype.
Mechanical hyperalgesia produced by ψεRACK was measured as the change in response frequency to a 0.4 g von Frey filament, tested 1 hour after intraplantar injection of either ψεRACK or scrambled ψεRACK peptide (Figure 8B). There was a greater response frequency in Prkce+/+ mice treated with ψεRACK than in Prkce+/+ mice treated with the scrambled peptide but no difference between Prkce+/+ and Prkce–/– mice treated with the scrambled peptide or between Prkce–/– mice treated with ψεRACK or the scrambled peptide (H= 8.75, P = 0.0328). These results indicate that ψεRACK produces a mechanical hyperalgesia that specifically requires PKCε.
In addition to limb withdrawal reflexes, we examined spontaneous pain by timing the duration of licking, biting, or lifting of the injected paw after administration of ψεRACK (Figure 8C). Injection of saline elicited brief nociceptive behavior in both Prkce+/+ and Prkce–/– mice. In contrast, in Prkce+/+ mice, treatment with ψεRACK produced robust nocifensive behavior that persisted much longer than that after saline injection, while in Prkce–/– mice there was no difference in nocifensive behavior elicited by saline or ψεRACK (_F_genotype [1,16] = 4.63, P = 0.0471; _F_treatment [1,16] = 8.50, P = 0.0101; _F_genotype × treatment [1,16] = 5.26, P = 0.0357). These data indicate that local injection of ψεRACK produces spontaneous pain that is fully dependent on PKCε.
PKCε-dependent mechanical hyperalgesia is reduced in Scn10a–/– mice. To determine whether NaV1.8 is necessary for PKCε-induced hyperalgesia, we administered ψεRACK or the control scrambled ψεRACK peptide by intraplantar injection in Scn10a–/– mice and measured thermal hyperalgesia 1 hour later (Figure 8D). Although there was an overall effect of treatment (Ftreatment [1,44] = 12.45; P < 0.001), there was no significant effect of genotype (Fgenotype [1,44] = 1.68; NS) and no genotype-by-treatment interaction (Fgenotype × treatment [1,44] = 0.1810; NS). However, the response frequency to von Frey filament stimulation was differentially altered in Scn10a+/+ and Scn10a–/– mice after administration of ψεRACK (H= 16.70, P = 0.0008). In wild-type mice, there was a significant increase in paw withdrawal frequency after injection of ψεRACK compared with that after injection of the scrambled peptide, whereas the responses in Scn10a–/– mice were not significantly different after these treatments (Figure 8E). In contrast, ψεRACK elicited a similar increase in nocifensive behavior in both genotypes (Figure 8F), with a main effect of treatment (_F_1,26 = 21.94; P < 0.0001) but no effect of genotype (_F_1,26 = 0.16; NS) and no genotype-by-treatment interaction (_F_1,26 = 0.25; NS). These findings indicate that NaV1.8 contributes to PKCε-stimulated mechanical hyperalgesia but not to thermal hyperalgesia or nocifensive behavior.