KCNQ/M Currents in Sensory Neurons: Significance for Pain Therapy

Cell culture

Dorsal root ganglion and superior cervical ganglion neurons. Dorsal root ganglion (DRG, from all spinal levels) and superior cervical ganglion (SCG) neurons were dissected from 17-d-old Sprague Dawley rats killed by CO₂ asphyxiation and prepared using a standard enzymatic dissociation procedure as described elsewhere (Owen et al., 1990). Briefly, after incubation in collagenase (500 U/ml for 15 min) and then trypsin (1 mg/ml for 30 min), the ganglia were mechanically triturated with a fire-polished glass Pasteur pipette. The ganglia were then centrifuged and resuspended in Leibovitz' L-15 supplemented with 10% fetal calf serum (FCS), 2 mm glutamine, 24 mm NaHCO₃, 38 mm glucose, 2-3% penicillin and streptomycin, and 25 ng/ml nerve growth factor. For electrophysiological recording, dissociated neurons were plated on 35 mm plastic dishes (Nunc, Roskilde, Denmark) coated with laminin and used either 1-2 (SCG) or 1-8 (DRG) d in culture. For immunofluorescence, dissociated DRG neurons were plated onto laminin-coated glass coverslips contained within four-well sterile tissue culture plates and used 1 d in culture.

Chinese hamster ovary cells. Chinese hamster ovary (CHO) cells were cultured and transfected as described elsewhere (Selyanko et al., 1999). Briefly, CHO cells were grown at 37°C and 5% CO₂ in α-MEM supplemented with 10% FCS, 1% l-glutamine, and 1% penicillin and streptomycin. Cells were plated in 35 mm plastic dishes and transfected 1 d later using LipofectAMINE Plus according to the manufacturer's instructions (Invitrogen, San Diego, CA). KCNQ and CD8 cDNA plasmids, driven by the cytomegalovirus promoter, were cotransfected in a 10:1 ratio. For expression of heteromultimers, equal amounts of human KCNQ2 and rat KCNQ3 cDNAs were used (obtained from Dr. D. McKinnon, State University of New York, Stony Brook, NY), as described by Wang et al. (1998). Transfected cells were identified by adding CD8-binding Dyna-beads (Dynal, Oslo, Norway) before recording.

Perforated patch whole-cell recording

Solutions. The extracellular solution contained (in mm): 144 NaCl, 2.5 KCl, 2 CaCl₂, 0.5 MgCl₂, 5 HEPES, and 10 glucose, pH adjusted to 7.4 with Tris base. Pipettes were filled with an intracellular solution containing (in mm): 80 K acetate, 30 KCl, 40 HEPES, 3 MgCl₂, 3 EGTA, and 1 CaCl₂, pH adjusted to 7.4 with NaOH. Amphotericin B was used to perforate the patch (Rae et al., 1991). Pipette resistance was 2-3MΩ, and the series resistance was compensated (60-90%). Recordings were made at room temperature (20-22°C).

Data acquisition and analysis. Data were acquired and analyzed using pClamp software (version 8.0; Axon Instruments). Currents were recorded using an Axopatch 200A (or 200B) patch-clamp amplifier, filtered at 1 kHz, and digitized at 4-8 kHz. I_K(M) amplitude and its inhibition by K⁺ channel blockers were measured from deactivation relaxations at -50 mV. Results are expressed as mean ± SEM. In general, half-inhibitions were calculated using the Hill equation: y/y_max = 1/(1 + (x/x₀)^p), where y is the fractional reduction of the relaxation amplitude; y_max is the maximum reduction; x is the drug/blocker concentration; x₀ is the IC₅₀ (the concentration at which y/y_max = 0.5), and p is the power (equivalent to the Hill slope). However, for tetraethylammonium (TEA) sensitivity, some of the inhibition curves were best fit using a two-component equation: y/y_max = q/(1 + x/x₀) + (1 - q)/(1 + x/x₁), where x₀ and x₁ are the IC₅₀ values for two channel populations with proportional contributions of q and (1 - q), respectively (other definitions are as above). The program Origin (version 5.0; Microcal Software Inc.) was used for creating the figures.

PCR analyses

Reverse transcription-PCR. Sequence-specific oligonucleotide primers were designed to the 3′ coding and or noncoding regions of all members of the KCNQ gene family. Alignments of rat, where available, or human KCNQ cDNA sequences allowed the design of KCNQ subtype-specific primers. The primer sequences were as follows: KCNQ1, forward, AGGATCGGAGGCCAGACCAT; reverse, TCATATCAGGCCTTCAAGAG; KCNQ2, forward, AAGCTAGACTTCCTGGTGAG; reverse, ACTGTATGTGCTAAGGAACC; KCNQ3, forward, AAGACAGGTTCACGACATGG; reverse, CTAGAAGAGACTAACAGTGC; KCNQ4, forward, GCCGGATCAAGAGCCTGCAA; reverse, CCAAGCAGCCTGAGACCAGCT; and KCNQ5, forward, CTGTCATTCGAGCTATCAGA; reverse, GCTTGACTGTGCATAGTAGG.

Rat DRG total RNA was purified from whole DRG by homogenization in Trizol reagent (Invitrogen, Paisley, UK) according to the manufacturer's instructions. Rat total RNA samples from heart and skeletal muscle were obtained from Clontech (Palo Alto, CA). Total RNA was reverse transcribed using the Superscript kit (Invitrogen). KCNQ cDNA fragments were amplified using the Advantage GC melt kit (Clontech, Basingstoke, UK). Amplified cDNA fragments were cloned using the TOPO cloning kit (Invitrogen, Groningen, The Netherlands) and sequenced to confirm their identity.

Single-cell PCR. Single-cell PCR analysis was performed as previously described (Shah et al., 2002). Briefly, cultured neurons were collected into 7.5 μl of recording solution and eluted into an Eppendorf tube containing 2.5 μl of first-strand buffer [a 2 mm concentration of each DTP, 20 μm oligo-dT₁₅, 40 mm dithiothreitol, and 20 U of RNase inhibitor (Roche Molecular Biochemicals, Indianapolis, IN)]. Reverse transcription (RT) of mRNA transcripts was initiated by addition of 100 U of Moloney murine leukemia virus reverse transcriptase RNase H(-) point mutant (Promega, Madison, WI) followed by incubation at 37°C for 1 hr. A multiplex PCR protocol was then used to amplify cDNA for KCNQ2-5 simultaneously. Primers were designed to be intron-spanning (on the basis of human KCNQ genes) and have been described fully by Shah et al. (2002).

Immunofluorescence

DRG neurons cultured on glass coverslips were initially rinsed three times for 5 min/rinse with PBS. Cells were fixed with freshly prepared 4% paraformaldehyde for 30 min at room temperature and then quenched twice with 0.37% glycine and 0.27% ammonium chloride in PBS for 10 min. After several rinses with PBS, the cells were permeabilized with 0.1% Triton X-100 in PBS for 15 min and then incubated for 60 min in a blocking solution containing 2% bovine serum albumin (BSA) plus 2% FCS in PBS. After three rinses with 1% BSA in PBS, the cells were incubated overnight at 4°C with primary antibody diluted in 1% BSA in PBS. The primary antibodies used were goat anti-KCNQ2 (1:100; Santa Cruz Biotechnology, Santa Cruz, CA), rabbit anti-KCNQ3 (1:100; a gift from S. Burbidge, GlaxoSmithKline), goat anti-KCNQ3 (1:100; Santa Cruz Biotechnology), and rabbit anti-KCNQ5 (1:500; a gift from A.Villaroel, Instituto Cajal-Consejo Superior de Investigaciones Científicas, Madrid, Spain).

Cells were rinsed six times for 5 min/rinse with 1% BSA in PBS and then incubated with tetramethylrhadomine isothiocyanate- or FITC-coupled secondary antibodies (1:1000; Molecular Probes, Eugene, OR) for 60 min. After six additional washes with 1% BSA in PBS and a final wash with PBS alone, the coverslips were mounted on ethanol-cleaned slides using a fluorescence mounting medium (Dako, High Wycombe, UK) and visualized using a confocal microscope. Images were obtained using either a 40 or 100× objective with the sequential acquisition setting at 1024 × 1024 pixel resolution. Control experiments in which the primary antibody was omitted or preincubated with its relevant immunogenic peptide were performed to determine antibody specificity.

In vivo spinal cord electrophysiology

Model of neuropathy. Male Sprague Dawley rats, initially weighing 130-150 gm, were used for in vivo electrophysiological studies. All experimental procedures were approved by the Home Office and followed the guidelines of the International Association for the Study of Pain (Zimmerman, 1983). The spinal nerve ligation model of neuropathic pain was performed as first described by Kim and Chung (1992). Briefly, under gaseous halothane anesthesia (3.5% for induction and 1.5% for maintenance) in N₂O and O₂ (50:50), the L5 and L6 spinal nerves were exposed as follows. The rat was placed in a prone position, and a midline incision was made from L4 to S2. A little of the left paraspinal muscles and left spinous process of the L5 lumbar vertebra were removed to expose the L4 and L5 spinal nerves. L6 was identified lying just under the sacrum. Using 6-0 silk thread, the left spinal nerves L5 and L6 were tightly ligated in the section between their dorsal root ganglion and their conjunction to form the sciatic nerve. Hemostasis was confirmed; the wound was sutured; and the animal recovered from anesthesia.

Behavioral testing. For 2 weeks after surgery, the rats were housed in groups of four in plastic cages under a 12 hr day/night cycle, and their general health was monitored. Successful reproduction of the neuropathic model was confirmed by behavioral testing (postoperative days 2, 3, 5, 7, 9, 12, and 14). Rats were placed in transparent plastic cubicles on a mesh floor, and the sensitivities of both the ipsilateral and contralateral plantar surfaces of the hindpaws to normally non-noxious punctate mechanical stimuli, using von Frey filaments (bending forces, 1, 5, and 9 gm) were assessed as described by Matthews and Dickenson (2001).

Neuronal characterization. Electrophysiology was performed 14-17 d after surgery and on nonoperated naive rats of similar size (Dickenson and Sullivan, 1986). Briefly, anesthesia was induced with 3% halothane in N₂O and O₂ (66:33); a cannula was inserted into the trachea; and the rat was secured in a stereotaxic frame. A laminectomy was performed (vertebrae L1-L3) to expose segments L4 and L5 of the spinal cord, and the level of halothane was reduced to 1.8%. Extracellular recordings of single convergent neurons, located deep within the dorsal horn (>500 μm), receiving input from the toe region ipsilateral to the spinal nerve ligation (when performed), were made using a parylene-coated tungsten electrode. Neurons selected responded to both noxious (pinch) and non-noxious (touch) stimuli. Spikes evoked by natural stimuli applied constantly over 10 sec were quantified by the application of both punctate mechanical (von Frey filaments, 9 and 75 gm) and thermal (constant water jet at 45°C) stimuli applied to the center of the receptive field of the neuron. The thermal response to 45°C was determined by subtracting the response to 32°C (a non-noxious temperature to ascertain any mechanical response evoked by the water jet) from the response to 45°C. All responses to natural stimuli were normalized by the subtraction of any spontaneous activity measured before the application of each stimulus. Response of the neuron to transcutaneous electrical stimulation was established by insertion of two fine needles into the center of its peripheral receptive field. A test consisted of a train of 16 stimuli (2-msec-wide pulse at 0.5 Hz at three times the threshold required to evoke a C fiber response), and a poststimulus histogram was constructed. Electrically evoked spikes were separated on a latency basis into Aβ fibers (0-20 msec), Aδ fibers (20-90 msec), C fibers (90-300 msec), and postdischarge spikes (300-800 msec). The “input” represents the number of spikes (90-800 msec) evoked by the first stimulus of the train. “Excess spikes” are measures of “windup,” which is increased NMDA receptor-mediated neuronal excitability to repeated constant stimulation (Dickenson, 1995). Excess spikes were calculated as the total spikes (90-800 msec) after a 16-stimulus train-input × 16. Windup graphs for individual neurons show how the combined number of evoked C fiber and postdischarge action potentials (i.e., 90-800 msec) increase with each repeated electrical stimulation.

Pharmacological studies. The testing protocol, initiated every 10 min, consisted of an electrical test followed by the natural stimuli, as described. Stabilization of the neuronal responses was confirmed with at least three consistent predrug responses (<10% variation), for all measures. These values were then averaged to generate predrug control values with which to compare the effect of retigabine administration on subsequent evoked responses. Retigabine was dissolved in saline and applied directly onto the spinal cord in 50 μl volumes. Each dose (10, 30, 60, and 90 μg) was followed until maximum effects were exerted (a minimum of 60 min), when the next dose would be applied cumulatively. The results were calculated as maximum percentage of inhibition from the averaged predrug value for each neuron, and the overall results for each dose were expressed as mean ± SEM of the normalized data. Statistical analysis of maximal drug effects at each dose compared with the averaged predrug value was determined by paired t test on raw data. An unpaired t test on the normalized data was used for the comparison of drug effects between different experimental groups. The level of significance was taken as p < 0.05.

Animal model of nociceptive and inflammatory hypersensitivity

The methods for induction and assessment of carrageenan-induced hyperalgesia have been described previously (Clayton et al., 2002).

Induction of carrageenan-induced hyperalgesia. Briefly, male random-bred hooded rats were injected intraplantar into the left hind paw with 100 μl of 2% carrageenan. To assess the analgesic effect of retigabine and its antagonism by XE991, animals were dosed orally at 5 mg/kg with a combination of vehicle plus vehicle, vehicle plus retigabine, vehicle plus XE991, or retigabine plus XE991 2 hr after the carrageenan dose and 1 hr before assessment of hyperalgesia.

Behavioral assessment of hyperalgesia. In normal rats, body weight is distributed equally between the two hindpaws. However, body weight is redistributed when one hindpaw is inflamed or painful so that less weight is placed on the affected paw. Thus, redistribution of body weight after induction of inflammation may be used to assess the development of hyperalgesia. Weight bearing was examined using a dual-channel weight averager (Bioengineering; GlaxoSmithKline). The two hindpaws were placed on separate sensors, and the percentage of weight distribution was calculated over 7 sec (Clayton et al., 1997).

Data analysis. All data are expressed as mean ± SEM. In all experiments, there were seven animals per group. Statistical analysis was performed to compare control responses with test responses using ANOVA followed by a post hoc Kruskal-Wallis test on raw data. The level of significance was taken as p < 0.05.

Drugs and chemicals

Retigabine and XE991 (10, 10-bis (4-pyridinyl-methyl)-9(10H)anthracenone) were gifts from GlaxoSmithKline (Stevenage, UK) and DuPont (Wilmington, DE), respectively. Linopirdine was purchased from Research Biochemicals (Natick, MA, USA). Margatoxin (MgTX) was obtained from Peptide Institute. TEA was purchased from Lancaster Synthesis (Morecambe, UK). WAY-123,398 ([(4-methylsulfonyl) amido]benzene-sulfonamide) was a gift from Wyeth-Ayerst Research (Princeton, NJ). Nerve growth factor was purchased from Tocris (Bristol, UK). All other drugs and chemicals were obtained from Invitrogen, Sigma (Gillingham, UK), or BDH (Poole, UK).

Identification of the M current

The I_K(M) was identified by using a standard deactivation voltage protocol, in which the cell is held at a steady depolarized potential to activate the current and then deactivated by intermittent hyperpolarizing steps (Fig. 1A); the contribution of I_K(M) to the outward current is then diagnosed from the slow deactivation tail current (Fig. 1B) (Brown and Adams, 1980). The reason for using this protocol is that, because the M current does not inactivate, contamination by other voltage-gated currents is minimized.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

M currents are expressed in both small and large DRG neurons. A, Standard deactivation protocol for recording I_K(M). B, Representative current recorded from a small (17.3 pF) DRG neuron. C, Mean instantaneous (open circles) and leak-subtracted steady-state (filled circles) I-V relationships for I_K(M) recorded from small DRG neurons (n = 17; capacitance, 19.4 ± 1.2pF), obtained by measuring the current at the beginning and end of the voltage pulse, respectively. Leak subtraction was performed by extrapolation of the linear portion of the I-V curve negative to -70 mV. D, Semilogarithmic plot of the voltage dependence of the fast (filled circles) and slow (open squares) deactivation time constants (τ) of I_K(M) recorded from small DRG neurons (n = 9; capacitance, 19.0 ± 1.2 pF). E, F, Representative current recorded from a large (>100 pF) DRG neuron before (E) and after (F) block of I_Ins and I_h with 100 nm α-DaTX and 1 mm Cs⁺, respectively. Note the gain increase in F. G, Voltage protocol used for recording I_K(M) in the absence and presence of various concentrations of K⁺ channel blockers and M channel blockers and activators. H, I, Enhancement (H) and inhibition (I) of I_K(M) recorded from a large DRG neuron with 10 μm retigabine and 0.03-10 μm XE991, respectively.

An M current so defined was identified in each of 30 small neurons tested (capacitance, 20.4 ± 1.1 pF; range, 11.5-34 pF). Of 22 cells so tested, 16 also responded to 1 μm capsaicin with an inward current (204 ± 97 pA at -50 mV) and an increase in conductance (data not shown), whereas the remaining 6 of 22 cells showed no response. Capsaicin-sensitive and -inse-nsitive cells had the same size (mean capacitance, 19.9 ± 1.4 and 17.3 ± 3.3 pF, respectively; p > 0.05). The small size of cells expressing I_K(M) and their sensitivity to capsaicin suggest that they were probably nociceptors.

The I-V relationship for I_K(M), computed from the initial amplitude of the deactivation tail currents (Adams et al., 1982), showed a negative threshold of activation (-60 mV), close to the resting membrane potential, and a reversal potential for its tail of approximately -80 mV, close to E_K (Fig. 1C). Deactivation of I_K(M) was best-fitted by two exponentials, with the time constants, τ_fast and τ_slow, equal to 76.4 ± 9.9 and 583 ± 134 msec, respectively, at -50 mV (n = 9). These time constants were very close to τ_fast and τ_slow for I_K(M) in rat sympathetic neurons and those for the KCNQ2 plus KCNQ3 current (I_KCNQ2+KCNQ3) in transfected CHO cells (Table 1). Deactivation showed the characteristic acceleration with increasing hyperpolarization; both τ_fast and τ_slow shortened, with τ_fast falling e-fold for a 52.6 mV hyperpolarization (Fig. 1D).

I_(K)M was also detected in 9 of 10 large cells tested (capacitance, >100 pF). However, in contrast to small cells, I_K(M) was not clearly visible in the initial macroscopic current recordings because it was masked by the presence of a large “instantaneous” outward current (I_Ins) and a slow hyperpolarization-activated inward current (I_h), which were activated positive and negative to V_rest, respectively (Fig. 1E). Nevertheless, the presence of I_(K)M became apparent after inhibition of I_Ins and I_h with 100 nm α-dendrotoxin (α-DaTX) or 1-50 nm MgTX and 1 mm Cs⁺, respectively (Fig. 1F, note the increased gain). Confirmation of this current as I_(K)M was achieved using retigabine (10 μm; Fig. 1H) and XE991 (0.03-10 μm; Fig. 1I), which, as anticipated, activated and blocked the current, respectively (see below).

K⁺ channel pharmacology

Linopirdine has previously been demonstrated to be a potent inhibitor of the M current in hippocampal and rat sympathetic neurons (Aiken et al., 1995; Lamas et al., 1997). I_K(M) in DRG neurons showed a high sensitivity to linopirdine (Fig. 2B,C;IC₅₀, 2.1 ± 0.2 μm; n = 8) and to its analog XE991 (Fig. 2C;IC₅₀, 0.26 ± 0.01 μm; n = 6) (Wang et al., 1998), as well as to Ba²⁺ (data not shown; IC₅₀, 0.3 ± 0.04 mm; n = 4). One to 40 μm MgTX and αDaTX (blockers of slowly inactivating K_v1.1, 1.2, and 1.3 and 1.1, 1.2, and 1.6 channels, respectively; Kaczorowski and Garcia, 1999), 1 mm 4-AP, or 10 μm WAY-123,398 (a blocker of ether-a-go-go-related gene potassium channels, which are capable of generating M-like currents; Selyanko et al., 1999) had no effect. Thus, I_K(M) in DRG neurons closely resembled that in sympathetic neurons and also the currents generated by its presumed KCNQ2 and KCNQ3 genes when coexpressed in Xenopus oocytes or mammalian cell lines (Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Pharmacology of I_K(M) in small DRG neurons. A, Voltage protocol used for recording I_K(M) in the absence and presence of various concentrations of K⁺ channel blockers and M channel blockers and activators. B, D, F, I_K(M) recorded in response to various concentrations of linopirdine (B), TEA (D), and retigabine (F). C, concentration dependence of inhibition of I_K(M) by linopirdine (closed circles) and XE991 (open circles). E, Concentration dependence of inhibition of I_K(M) by TEA in individual DRG neurons. G, Steady-state I-V relationship under control conditions (filled circles) and in the presence of 10 μm retigabine (open circles).

View this table:

• View inline
• View popup

Table 2.

Effects of K⁺ channel blockers on I_K(M) in DRG and SCG neurons and I_KCNQ2+KCNQ3 in mammalian cells or frog oocytes

TEA sensitivity

KCNQ2-5 subunits vary in their sensitivity to TEA (Yang et al., 1998; Wang et al., 1998; Kubisch et al., 1999; Hadley et al., 2000; Lerche et al., 2000; Schroeder et al., 2000), with KCNQ2 having the highest affinity arising from the presence of a tyrosine just downstream of the GYG pore sequence (Hadley et al., 2000). When measured in individual DRG neurons, the TEA sensitivity was variable (Fig. 2D,E), indicating different proportions of TEA-sensitive and -insensitive subunits. Thus, in some cells (two of seven), the TEA sensitivity was very high (IC₅₀, ∼0.2-0.6 mm), suggesting expression of KCNQ2 subunits only, whereas in other cells (three of seven), the TEA sensitivity was intermediate and similar to that expected for KCNQ2/3 heteromers (IC₅₀, ∼3.9-4.7 mm). Two of seven cells displayed biphasic inhibition by TEA, and a two-component Hill equation gave an improved fit (lower IC₅₀ values, 0.26 and 0.41 mm, upper IC₅₀ values, 8.55 and 3.28 mm), suggesting expression of a mixture of homomeric and heteromeric subunits.

Activation by retigabine

In seven of seven cells tested, the neuronal KCNQ channel opener retigabine (see Introduction) enhanced I_K(M) (Fig. 2F) and produced a characteristic slowing of its deactivation, consistent with the negative shift in activation of neuronal KCNQ channels (Main et al., 2000; Rundfeldt and Netzer, 2000; Wickenden et al., 2000; Tatulian et al., 2001). The effect of retigabine was concentration- and voltage-dependent, being stronger at -20 than at -50 mV. Steady-state I-V relationships constructed in the presence and absence of retigabine showed that this outward current was reduced by membrane hyperpolarization and dissipated at approximately -80 mV, close to E_K (e.g., see Fig. 2G). This was accompanied by a negative shift (by 12.3 ± 3.3 mV) in zero current (resting) potential.

M current modulation and cell excitability

M current can act as a brake on repetitive firing in neurons (Brown, 1988). Thus, M channel blockers enhance repetitive firing in sympathetic (Wang et al., 1998) and hippocampal (Aiken et al., 1995) neurons. We therefore tested what effect linopirdine and retigabine might have on the firing properties of small DRG neurons during long (1 sec) depolarizing current pulses. The resting potential was close to -55 mV and was maintained at this level, if necessary, by injecting DC current. The cells were highly refractory and fired only once in response to currents of 150-200 pA or more (Fig. 3A). In five of five neurons, the inhibition of the M current by 30 μm linopirdine reduced the threshold of firing, and in two of them, it produced a brief burst of multiple firing (Fig. 3B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Modulation of excitability in small DRG neurons by inhibiting I_K(M) with linopirdine and by activating I_K(M) with retigabine. A, Single firing was produced by a 400 pA depolarizing current pulse in the control, whereas two or three spikes were produced in response to the same current injection in the presence of 30 μm linopirdine (B). C, Under control conditions, electrotonic potentials and a single action potential (top) were produced by hyperpolarizing and depolarizing current pulses (bottom), respectively. D, Ten micromolar retigabine hyperpolarized the membrane by 12 mV and abolished the spike. E, A depolarizing current injection restores the membrane potential to the initial value in presence of 10 μm retigabine. The action potential remains absent and reduced voltage responses are produced, owing to an enhanced membrane conductance. F, The effects of retigabine on spike initiation and membrane conductance were reversed by 10 μm linopirdine. The data in panels A and B and C-F were obtained from two different cells.

In contrast, retigabine (10 μm) hyperpolarized the membrane and increased the threshold of firing (Fig. 3D). When the membrane potential was restored to the initial level by injecting the depolarizing DC current, the voltage changes associated with current pulses were much smaller than in the control, implying an increase in membrane conductance (Fig. 3E). The increase in conductance and inhibition of firing produced by retigabine could then be reversed by adding 10 μm linopirdine (Fig. 3F). Similar effects of retigabine were obtained from five other cells; in two of two cells tested, the effect of retigabine was reversed by 10 μm XE991.

Expression of KCNQ subtypes in DRG neurons

RT-PCR and single-cell PCR analyses were used to determine the expression of the family of KCNQ genes in rat DRG. KCNQ2-5 but not KCNQ1 were expressed at detectable levels in whole rat DRG (Fig. 4A). Several controls to support our observations are included. First, no amplified bands were detected in the absence of reverse transcriptase (second lane), thus confirming that the amplified bands were of cDNA and not genomic origins. Second, each of the amplified fragments was cloned and sequenced to confirm its identity. Thirdly, either brain (third lane) or heart tissue (fifth lane) was used as a positive control, allowing confirmation of a negative signal in the DRG. These results confirm that all of the potential molecular correlates of I_K(M) are expressed in the DRG. A low level of KCNQ1 expression in the brain was also detected, as previously reported (Kubisch et al., 1999). Interestingly, KCNQ4 and low levels of KCNQ5 expression were also detected in the heart (Fig. 4A, fifth lane). In the case of KCNQ4, this confirms previous data from Northern analysis (S. A. Burbidge, D. Crowther, and P. Sanseau, personal communication).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

KCNQ subunit mRNA expression in rat dorsal root ganglion neurons. A, Reverse transcription-PCR analysis of RNA isolated from adult rat DRG using primer pairs for KCNQ subunits 2-5. B, Single cell PCR analyses on cultured rat DRG small and large neurons using primer pairs for KCNQ subunits 2-5. Arrows point to KCNQ subunit-specific PCR products. Control reactions were performed using primer pairs on plasmid constructs containing the coding sequence for each of the KCNQ subunits 2-5. PCR products were separated by electrophoresis through 2% Metaphor agarose (FMC Bioproducts, Rockland, MD) and a 1 kb plus DNA ladder (Invitrogen).

Single-cell PCR was used to determine the KCNQ subunit expression in individual small and large neurons when cultured as for the electrophysiological experiments (Fig. 4B). PCR products of the predicted size from cDNA were obtained for KCNQ2, KCNQ3, and KCNQ5 from both small and large DRG neurons (n = 12). Neurons isolated from both acutely dissociated cultures and 3 d primary cultures gave identical results. KCNQ4 mRNA was not detected in either small or large DRG neurons, indicating that its expression in whole ganglia (Fig. 4A) refers to other cell types within the ganglia. Control reactions performed in parallel using plasmid DNA containing the coding sequence for KCNQ2-5 gave specific PCR products for each primer pair. Contamination by genomic DNA can be excluded because the primer pairs were intron spanning and because longer PCR products were not evident in the single cell PCRs.

Immunofluorescence results

Confocal immunofluorescence microscopy of cultured DRG neurons revealed variable expression of KCNQ2, 3 and 5 immunoreactivity on the somata and neuronal processes of both small and large neurons (Fig. 5). Thus, although some cells expressed both KCNQ2 and KCNQ3, others expressed KCNQ2 in the absence of KCNQ3 (Fig. 5C). Likewise, although some cells expressed both KCNQ3 and KCNQ5, others expressed KCNQ3 in the absence of KCNQ5 (Fig. 5I). Colocalization of KCNQ2 and KCNQ5 (Fig. 5F) was also detected, although it is likely that such cells also express KCNQ3 because KCNQ2 and KCNQ5 do not form heteromeric channels (Lerche et al., 2000; Schroeder et al., 2000). Costaining for KCNQ2, 3, and 5 subunits could not be performed because of species limitations of the secondary antibodies.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Immunocytochemistry with antibodies to KCNQ2, KCNQ3, and KCNQ5 reveals variable expression of KCNQ subunits in small and large DRG neurons. Confocal images of KCNQ2, KCNQ3, and KCNQ5 immunostaining in cultured DRG neurons are shown. A, B, D, E, G, H, Immunostaining images for individual antibodies. C, F, I, Overlays of immunostaining for KCNQ2 and KCNQ3 (C), KCNQ2 and KCNQ5 (F), and KCNQ3 and KCNQ5 (I). Scale bar, 25 μm.

In vivo spinal cord electrophysiology and pharmacology in a model of neuropathy

Rats subjected to spinal nerve ligation (see Materials and Methods) exhibited abnormal foot posture ipsilateral to nerve injury, whereby toes were held together in a “guarding” behavior. Successful replication of the nerve injury model was confirmed by the development of mechanical allodynia in the injured hindpaw, displayed as a brisk withdrawal, in some cases accompanied by shaking and licking of the foot, to normally innocuous von Frey filaments (1, 5, and 9 gm bending forces). Increased frequency of foot withdrawal was observed with increased bending force. Mechanical allodynia was evident at postoperative day 2, reached a maximum at days 7-12, and was still present at day 14 (Matthews and Dickenson, 2001). Consistent withdrawal responses were never exhibited by the contralateral hindpaw.

Recordings were obtained from ipsilateral dorsal horn neurons in 10 animals after spinal nerve ligation and in 11 naive animals. No significant differences were found between the two experimental groups in the mean values of recorded neuron depth and the responses evoked by electrical and natural stimulation. However, neurons of spinal nerve-ligated animals showed appreciable ongoing spontaneous activity (mean rate, 2.75 ± 1.09 Hz), whereas no such activity was observed in naive animals; this difference was significant (p < 0.02).

Application of retigabine (10-90 μg) produced statistically significant dose-related inhibitions of both the electrically and naturally evoked neuronal responses from the determined predrug control values in both experimental groups (Fig. 6; p < 0.05; n = 5-10). Clear effects were seen at ∼60 min after application of the drug (Fig. 6I). In naive and nerve-injured rats, the Aβ fiber response was least affected by spinal retigabine, with mean maximal inhibitions from predrug control values at the top dose only reaching 23 ± 3 and 33 ± 8%, respectively (Fig. 6C). The other electrically evoked neuronal responses were more susceptible to the inhibitory actions of retigabine. Thus, in nerve-injured animals, the C fiber, Aδ fiber, and input responses reached similar mean maximal observed inhibitions in the range of 61 ± 4 to 73 ± 9% (Fig. 6A,B, respectively). Greater inhibitory effects of the high dose of retigabine were observed on the postdischarge and excess spike measurements (Fig. 6D,E, respectively), such that, at 90 μg, retigabine maximally inhibited both measures in ligated animals by 81 ± 5%, whereas in naive rats, the effect was 69 ± 11 and 65 ± 7% from predrug control values for postdischarge and excess spikes, respectively. This is further emphasized by the examples in Figure 6, G and H, in which windup (the increase in the number of spikes per stimulus over the train of electrical impulses, which generates the postdischarge and excess spikes; Dickenson, 1995) is clearly reduced more after nerve injury, as shown by the flattening of the windup slope. Overall, there was no difference in the predrug values of these measures of neuronal excitability for the naive and nerve-injured groups.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Effect of retigabine on electrically evoked dorsal horn responses. A-F, Effect of spinally applied retigabine on the electrically evoked dorsal horn neuronal responses recorded from naive (open squares; n = 5-10) and spinal nerve-ligated (filled circles; n = 5-8) rats at postoperative days 14-17. Data are expressed as maximal mean percent inhibition of the predrug values ± SEM. A, C fiber responses; B, Aδ fiber responses; C, Aβ fiber responses; D, after discharge; E, excess spikes; F, input. G, H, Inhibitory effect of spinally applied retigabine (open circle, control; filled triangle, 10 μg of retigabine; open square, 30 μg of retigabine; filled circle, 90 μg of retigabine) on individual neurons exhibiting windup recorded from naive (G) and spinal nerve-ligated (H) rats. I, Time course of the effect of increasing doses of spinally applied retigabine on the evoked response of a typical dorsal horn neuron recorded from a spinal nerve-ligated rat. Examples of the effect on the Aβ fiber (filled triangle), C fiber (open circle), and postdischarge (filled circle) measurements are shown with the cumulative dose indicated, and the data are expressed as percent of the predrug control value.

Retigabine (90 μg) also inhibited the naturally evoked neuronal responses to innocuous (von Frey, 9 gm) and noxious (von Frey, 75 gm) punctate mechanical stimuli and noxious heat (45°C) in both experimental groups by amounts within the range of 65 ± 13 to 87 ± 5% from the predrug control (data not shown).

Analgesic properties of retigabine

To assess the analgesic effect of retigabine, we used a model for chronic pain in which the irritant carrageenan is injected into one hindpaw. This leads the animal to distribute its weight unevenly between the two legs. Thus, 3 hr after intraplantar administration of carrageenan (2%, 100 μl), there was a substantial and significant decrease in the weight bearing on the left hindpaw, such that animals distributed only 21 ± 3% of their hind leg load onto the inflamed paw (Fig. 7), compared with the normal 50% after injection of vehicle alone. Retigabine [5 mg/kg, orally (p.o.)] strongly reduced this asymmetry, so that the weight borne on the inflamed leg increased to 41 ± 2%. This effect was antagonized by coadministration of XE991 (5 mg/kg, p.o.), resulting in a weight distribution (28 ± 3%) similar to that in animals treated with vehicle alone. Interestingly, XE991 itself had no effect on weight distribution.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Analgesic activity of retigabine and block by XE991: effect of M channel activators and blockers on weight bearing in a behavioral model of inflammatory pain. Retigabine (5 mg/kg, p.o.), administered 2 hr after intraplantar carrageenan, significantly increased the weight placed on the inflamed paw. Coadministration of XE991 (5 mg/kg, p.o.) reversed the analgesic effects of retigabine, whereas treatment with XE991 (5 mg/kg, p.o.) alone had no effect on weight distribution. Bars indicate mean ± SEM of seven animals in each case. ^**p < 0.05.