Hyperpolarization-activated Cyclic Nucleotide-gated Channels May Contribute to Regional Anesthetic Effects of Lidocaine

VOLTAGE-GATED sodium channel is conventionally regarded as the main molecular target for local anesthetics.¹  By blocking sodium channels, local anesthetics produce nonselective sensory blockade, nociceptive blockade, motor paralysis, and sympathetic nerve blockade.¹  If local anesthetics could specifically impair sensory or even nociceptive functions, they would be widely used in various medical procedures, such as treatment of chronic pain.

Besides voltage-gated sodium channel, increasing evidences have indicated that local anesthetics at clinical relevant concentrations could interact with a large spectrum of receptors and/or ion channels, such as potassium channels,²  calcium channels,³  and transient receptor potential vanilloid-1 channels.⁴  For recent years, hyperpolarization-activated cyclic nucleotide-gated cation (HCN) channel has been identified as an underlying target of local anesthetics.^5–7  Previous studies demonstrated that lidocaine could inhibit HCN channel currents^6,7  and ZD7288 (HCN channel blocker) could enhance the effects of lidocaine and ropivacaine.^8,9  Interestingly, many regional anesthetic adjuvants, such as clonidine,^8,10  dexmedetomidine⁹  and ketamine,^11,12  have been found as HCN channels inhibitors at their regional anesthetic relevant concentrations (the concentrations used in regional anesthesia rather than systemic administration). Thus, recent studies demonstrated that the enhancement of clonidine and dexmedetomidine in regional anesthesia resulted from their modulation on HCN channels rather than adrenergic receptors.^8,9  However, the exact role of HCN channels in regional anesthesia in vivo is still elusive.

HCN channel (consists of four subtypes, known as HCN1–4)^13–16  is widely expressed in various tissues, including central nervous system, heart, and peripheral nervous system^13,17  from both animals and human beings. Ivabradine, an HCN blocker, has been used in the clinical setting for the treatment of heart failure.^18–21  For peripheral nervous system, HCN1 and HCN2 are the most common subtypes,¹³  which have been found in dorsal root ganglia (DRG)^22–25  and nerve fibers.^26,27  Previous studies indicated that the functions of nerve axon were modulated by HCN channels,^28,29  which might be involved in peripheral nerve block. Interestingly, HCN1 subtype is predominated in large diameter DRG neurons, whereas HCN2 subtype mainly in small ones.^22–24,30  Because small-diameter nerve fibers are widely recognized for high-threshold sensory transmission,^31,32  HCN2 subtype might contribute to nociceptive sensation.

In the current study, we hypothesized that in addition to voltage-gated sodium channel, HCN channels contribute to regional anesthetic effects of lidocaine in vivo.

With the approval of the Institutional Animal Experimental Ethics Committee of Sichuan University (Chengdu, Sichuan, China), male adult wild-type C57BL/6J mice and HCN1 knockout (HCN1^−/−) mice were used. The mice were housed in cages on a 12-h light-dark cycle with free access to food and water. HCN1^−/− mice were bred, as previously described.¹¹  HCN1^−/− mice were normal in body size and longevity, and no obvious behavioral abnormalities were found.¹³  All the study mice were placed on the experimental platform 1 h/day for consecutive 3 days (acclimation to the experimental environment) before formal experiments and were randomly assigned into each group based on a SPSS-generated random number (SPSS Inc., Chicago, IL). For behavioral assessment of mice, observers were blinded to the treatment and group assignment of mice.

Lidocaine was purchased from Fortune Zhaohui Pharmaceutical Co. Ltd. (Shanghai, China). ZD7288 was purchased from Tocris Bioscience (Ellisville, MO). Lidocaine and ZD7288 were prepared with normal saline. Stock solution of ZD7288 was 20 mM. Forskolin was purchased from Sigma-Aldrich Co. Ltd. (Shanghai, China) and diluted with dimethyl sulfoxide to stock solution at concentration of 5 mg/ml and then prepared with normal saline to final concentration. Isoflurane was provided by Abbott Pharmaceutical Co. Ltd. (Shanghai, China). All the study solutions were prepared immediately before injection.

DRG and sciatic nerves from wild-type and HCN1^−/− knockout mice were homogenized, and BCA kit (Pierce, Rockford, IL) was used to determine protein concentrations of the supernatant. Sodium dodecyl sulfate–polyacrylamide gel electrophoresis was applied to separate the proteins of supernatant and then transferred to a nitrocellulose membrane. Primary antibodies of HCN1 (#ab65706, Abcam, Cambridge, MA), HCN2 (#ab65704, Abcam), and β-actin (#ab1801, Abcam) were used. After incubation with primary antibody, the membrane was incubated with horseradish peroxidase–conjugated anti-rabbits secondary antibody (1:10,000; Pierce) for 1 h. Supersignal chemiluminescence detection kit (Pierce) was used for the development of blot. Immunoblotting was visualized by Kodak X-ray Processor 102 (Eastman Kodak, Rochester, NY) and analyzed by the software Quantity One (Bio-Rad Laboratories, Hercules, CA).

Under inhalation of 2% isoflurane, 50 μl study solution was injected into popliteal fossa of mice. All the mice received two injections with an interval of 20 min. For instance, ZD7288 (5.5 mM⁹) or forskolin (62.5 μg/ml⁸) or normal saline was injected 20 min before injection of lidocaine. After injection of lidocaine, tactile sensation of injected foot was evaluated by von Frey fibers (37450-277; Ugo Basile, Comerio, Italy), and tactile blockade was defined as no aversive reactions to 4 g von Frey fiber; thermal sensation was determined by heat stimulus (Plantar test 37370; Ugo Basile), and thermal sensory blockade was defined as reaction latency of mice to heat stimulus increased by more than 50% as previously described³³ ; pinprick sensory blockade was evaluated by a 24G needle (Terumo Medical Corp., Tokyo, Japan), and pinprick blockade was defined as no aversive responses to pinprick. Motor function of mice was scored according to a 4-point scale³⁴ : 1 = normal; 2 = intact dorsiflexion of foot with impaired ability to splay toes when elevated by the tail; 3 = toes and foot plantar flexed with no splaying ability; 4 = loss of dorsiflexion, flexion of toes, and impairment of gait. Score of motor function up to 2 (≥2) was defined as motor blockade.

Under inhalation of 2% isoflurane, study solutions (saline or ZD7288 or lidocaine) at 5 μl were intrathecally injected at L5–L6 of mice by microinjector, as previously described.^35–37  After injection, tactile sensory blockade, pinprick sensory blockade, and motor blockade of mice foot were evaluated as described in sciatic nerve block model. Left or right foot was randomly chosen for each mouse. For thermal sensory blockade, heat stimulus was applied to tails of mice with tail-flick unit (7360; Ugo Basile). Thermal sensory blockade was defined as reaction latency of mice tail to heat stimulus increased by more than 50%.

Estimated EC₅₀ (median effective concentration) of lidocaine in both sciatic nerve block and intrathecal anesthesia was determined by the up-and-down method.³⁸  Briefly, EC₅₀ of lidocaine for tactile blockade and pinprick blockade were determined, respectively. Tactile blockade and pinprick blockade were defined as no aversive responses to 2 g von Frey fiber and pinprick, respectively. Based on our preliminary experiments, initial concentration of lidocaine was 0.4% for the first mouse. After each successful blockade, lidocaine concentration was decreased by 0.8-fold (e.g., 0.32%); after each failure, it was increased by 0.8-fold (e.g., 0.5%). The up-and-down procedure was repeated until six crossovers points appeared.

Acute isolated DRG neurons were used for electrophysiologic recordings. The isolation procedure of DRG neurons was modified from the method as previously described.³⁹  Briefly, adult mice were anesthetized by ketamine/xylazine (40/15 mg/kg), and DRG was removed and digested in papain (1 mg/ml) for 40 to 50 min and in collagenase I (1 mg/ml) for 30 to 40 min. Large DRG neurons (>30 μm) and small DRG neurons (<20 μm) were recorded, respectively.

For whole-cell voltage clamp recording of HCN channel currents, pipettes (3 to 5 MΩ) were filled with KCl 130 mM, NaCl 4 mM, MgCl₂ 1 mM, CaCl₂ 0.5 mM, HEPES 10 mM, EGTA 10 mM, MgATP 3 mM, and GTP-Tris 0.3 mM (pH 7.3, osmolarity 310 to 315 mOsmol/l). Standard bath solution contained NaCl 140 mM, KCl 3 mM, HEPES 10 mM, CaCl₂ 2 mM, MgCl₂ 2 mM, and glucose 10 mM (pH 7.4, osmolarity 300 to 310 mOsmol/l). Tetrodotoxin (0.5 μM; Alomone Labs, Jerusalem, Israel) was added to bath solution routinely for inhibition of action potential. Recordings were obtained by Axopatch 200B amplifier (Axon Instruments, Inc., Sunnyvale, CA) at room temperature (approximately 25°C). HCN channel currents in mice DRG neurons were recorded by a series of hyperpolarizing step voltage pulses (−60 to −140 mV with 10 mV increments) from a holding potential of −60 mV, followed by a voltage step of −100 mV to record tail currents. Tail currents were normalized, plotted as a function of the preceding hyperpolarization step voltage, and fitted with Boltzmann curves for derivation of half-activation voltage (V_1/2a) as: G/Gmax = 1/(1 + e(V_1/2a − V)/k), G/G_max = normalized conductance; G_max = maximum conductance; V_1/2a = voltage activation of half-maximum; and k = slope factor. Concentration–response relationship for lidocaine on maximal current amplitude was fitted to the Hill equation: Y = 1/[1 + 10(logIC₅₀ − X)×h], where Y = inhibitory effect; X = concentration of lidocaine; h = Hill coefficient.

For whole-cell voltage clamp recording of voltage-gated sodium channel currents, holding potential was −90 mV. The extracellular solution contained NaCl 110 mM, NaHCO₃ 26 mM, KCl 5.6 mM, NaH₂PO₄ 1 mM, CaCl₂ 0.1 mM, MgCl₂ 5 mM, tetraethylammonium 20 mM, and glucose 11 mM, pH 7.4, and osmolarity 300 to 310 mOsmol/L. The pipette electrode solution contained CsCl 140 mM, EGTA 5 mM, MgCl₂ 1 mM, HEPES 10 mM, and MgATP 3 mM, pH 7.3, and osmolarity 310 to 315 mOsmol/L. Currents were sampled at 20 kHz and filtered at 1 to 3 kHz and recorded using a series of depolarizing step voltage pulses from −60 to 40 mV with 10 mV increments. Activation curves of sodium channel currents were fit to the Boltzmann equation: G/G_max = 1/(1 + e (V_1/2a − V)/k), G/G_max = normalized conductance; G_max = maximum conductance; V_1/2a = voltage activation of half-maximum; and k = slope factor. Sodium channel conductance (G_Na) was calculated as: G_Na= I_Na/(V_t − V_r), I_Na = peak sodium current; V_t = test potential; and V_r = sodium channel reversal potential. Concentration–response curves of lidocaine were calculated by Hill equation: Y = 1/[1 − 10(logIC₅₀ − X)×h], where Y = inhibitory effect; X = concentration of lidocaine; h = Hill coefficient.

Neural membrane properties of DRG neurons including resting membrane potential (RMP) and input resistance (R_in) were measured by current clamp recording, from voltage responses to hyperpolarizing current injection. RMP was the membrane potential at I = 0, and R_in was calculated with V/I relationship by the Ohm law (R = U/I) based on hyperpolarizing current injection (from 20 to −100 pA with 20 pA increments) and amplitudes of voltages were the stable value for each current injection on hyperpolarization state. Pipette solution contained KCl 17.5 mM, potassium gluconate 122.5 mM, HEPES 10 mM, EGTA 0.2 mM, NaCl 9 mM, MgCl₂ 1 mM, MgATP 3 mM, and GTP-Tris 0.3 mM (pH 7.3). Standard bath solution contained NaCl 140 mM, KCl 3 mM, HEPES 10 mM, CaCl₂ 2 mM, MgCl₂ 2 mM, and glucose 10 mM (pH 7.4).

SPSS version 16.0 (SPSS Inc.) was used for all the statistical analysis, except where noted. By preliminary test (n = 4) on difference of pinprick blockade durations between wild-type saline/lidocaine and wild-type ZD7288/lidocaine (7.5 ± 2.9 vs. 12.5 ± 2.9 min) groups, calculated minimal sample size was 7.0 (α = 0.05; β = 0.10). Therefore, sample size of 10 was chosen for in vivo assay. Of note, data of the preliminary test were not included in the final data; therefore, the power analysis based on preliminary test was not an interim analysis for final analysis and P values were not adjusted. Data were presented as mean ± SD or median (interquartile range) for normal distribution data and skewed data, respectively. Latency data (e.g., durations) were compared by the Kruskal–Wallis test followed by post hoc of the Dunn test (software Prism 6.0; GraphPad, San Diego, CA). For “survival curves” of pinprick blockade and motor blockade, Kaplan–Meier method with log-rank test was used for comparison among groups. EC₅₀ of lidocaine was calculated by averaging concentration points of crossovers (12 points) and compared with one-way ANOVA with post hoc of Games–Howell. For comparison of HCN and sodium channel currents, two-way ANOVA (drugs and genotypes as principal factors) was performed, with the Tukey correction of the t test for post hoc pairwise comparisons. Activation curves of HCN and sodium channel currents were fit to the Boltzmann equation. Two-way ANOVA (drugs and genotypes as principal factors) was performed for the comparison of membrane properties (RMP and R_in). For all the cases, P < 0.05 was considered as statistically significant.

Both HCN1 and HCN2 channels were detected in DRG and sciatic nerves (fig. 1A). Generally, expression levels of HCN1 and/or HCN2 channels were higher in DRG than sciatic nerves, and expression level of HCN1 was higher than that HCN2 in DRG (fig. 1B). No expression of HCN1 channel protein (with expected molecular weight) was found in HCN1^−/− mice.

Tactile sensation (by von Frey fibers, fig. 2A) and thermal sensation (by foot-flick test, fig. 2B) were similar between wild-type and HCN1^−/− mice (n = 15 and 18 for each genotype). For sciatic nerve block model in wild-type mice, ZD7288 alone did not produce any anesthetic relevant actions: tactile (n = 10, fig. 2C) and thermal (n = 10, fig. 2D) sensations were unchanged after injection of 5.5 mM ZD7288, which was similar to the effect of saline (n = 10 and 10). For intrathecal anesthesia, ZD7288 alone did not produce tactile (n = 10, fig. 2E) or thermal (n = 10, fig. 2F) sensory blockade.

EC₅₀ (median effective concentration) of lidocaine for tactile sensory blockade (2 g von Frey fiber) and pinprick sensory blockade were determined by the up-and-down method (fig. 3 and table 1). In sciatic nerve block, EC₅₀ of lidocaine for tactile blockade was significantly increased in HCN1^−/− mice than wild-type mice (0.22 ± 0.04% vs. 0.14 ± 0.03%, P < 0.001). EC₅₀ of lidocaine for pinprick blockade was similar between HCN1^−/− and wild-type mice (0.36 ± 0.07% vs. 0.35 ± 0.05%, P = 0.807). Meanwhile, EC₅₀ of lidocaine was significantly decreased both in tactile blockade (P = 0.033 vs. lidocaine alone) and in pinprick blockade (P<0.001 vs. lidocaine alone) with preinjection of 5.5 mM ZD7288 (table 1). In intrathecal anesthesia, EC₅₀ of lidocaine for tactile blockade was significantly increased in HCN1^−/− mice than wild-type mice (0.38 ± 0.08% vs. 0.28 ± 0.06%, P = 0.008), whereas EC₅₀ of lidocaine for pinprick blockade was similar between HCN1^−/− and wild-type mice (0.47 ± 0.07% vs. 0.45 ± 0.05%, P = 0.878).

In sciatic nerve block model, durations of 1% lidocaine were shortened in HCN1^−/− mice (n = 10) than that in wild-type mice (n = 10), while maximal effects of lidocaine were unchanged (fig. 4). Durations of tactile sensory blockade decreased from 15.0 (6.3) to 9.0 (5.0) min (fig. 4A, P = 0.008); durations of motor blockade decreased from 17.5 (11.3) to 10.0 (6.3) min (fig. 4B, P = 0.021); durations of thermal sensory blockade decreased from 10.0 (5.0) to 5.0 (6.0) min (fig. 4C, P < 0.001). However, durations of pinprick sensory blockade were similar (fig. 4D, P = 0.721) between the two genotypes. All the results were list in table 2. In intrathecal anesthesia, durations of 1% lidocaine were also decreased in HCN1^−/− mice (n = 10) than in wild-type mice (n = 10) (fig. 5). Durations of tactile sensory blockade and motor blockade were significantly shortened in HCN1^−/− mice (P = 0.022 and 0.039, respectively), whereas durations of thermal sensory blockade and pinprick sensory blockade were similar (P = 0.490 and 0.898, respectively). All the results were list in table 3.

ZD7288 (HCN channel blocker, nonselectively inhibits HCN1 and HCN2 subtypes) could prolong durations of 1% lidocaine. In wild-type mice (n = 10), durations of tactile sensory blockade increased from 15.0 (6.3) to 20.0 (1.3) min (fig. 6A, P = 0.011); durations of motor blockade increased from 17.5 (11.3) to 25.0 (10.0) min (fig. 6B, P = 0.024); durations of thermal sensory blockade increased from 10.0 (5.0) to 15.0 (5.0) min (fig. 6C, P = 0.016); and durations of pinprick sensory blockade increased from 6.0 (6.3) to 12.5 (11.3) min (fig. 6D, P = 0.018). In HCN1^−/− mice (n = 10), durations of tactile sensory blockade increased from 9.0 (5.0) to 12.0 (5.0) min (fig. 6A, P = 0.035); durations of motor blockade increased from 10.0 (6.3) to 15.0 (6.3) min (fig. 6B, P = 0.006); durations of thermal sensory blockade increased from 5.0 (6.0) to 10.0 (5.0) min (fig. 6C, P = 0.038); and durations of pinprick sensory blockade increased from 6.0 (5.0) to 10.0 (11.3) min (fig. 6D, P = 0.039). Of note, durations of 1% lidocaine in HCN1^−/− ZD7288/lido group were similar to wild-type saline/lido group (fig. 6). Both HCN1 and HCN2 subtypes contributed to tactile sensory blockade, thermal sensory blockade, and motor blockade of lidocaine, whereas HCN2 subtype contributed to pinprick sensory blockade. All the results were list in table 2.

In wild-type and HCN1^−/− mice (n = 10 and 10), pinprick blockade durations of 1% lidocaine were significantly shortened by forskolin (fig. 7 and table 2). Although durations of tactile sensory blockade, thermal sensory blockade, and motor blockade were shorter than that of 1% lidocaine alone, no significance was found.

For wild-type mice, activation kinetics of HCN channel currents were slower in small neurons than large neurons (time constant at −120 mV: 62 ± 11 vs. 382 ± 46 ms, P < 0.001). For HCN1^−/− mice, HCN channel currents were almost diminished in large neurons, whereas HCN channel currents in small neurons were unaffected (fig. 8A). I-V (current-voltage) curves of HCN channel currents in small neurons were similar between wild-type and HCN1^−/− mice (fig. 8B). We further determined the effects of lidocaine on HCN channel currents in both large and small DRG neurons from wild-type mice. Lidocaine reversibly inhibited HCN channel currents without selectivity between large and small neurons (estimated IC₅₀ = 82 ± 10 and 88 ± 9 μM, respectively). Maximal inhibition of lidocaine on HCN channel currents was 53 ± 9% and 48 ± 10%, respectively, for large and small neurons (fig. 8C). At a concentration of 100 μM, lidocaine significantly hyperpolarized the activation potential of HCN channel currents in large neurons but not in small neurons (fig. 8D).

Sodium channel currents and inhibition of lidocaine on sodium channel currents were unaffected by HCN1 deletion. Sodium channel currents were similar between wild-type and HCN1^−/− mice (fig. 9, A and B). Lidocaine could reversibly inhibit sodium channel currents without selectivity between large and small neurons (fig. 9, C and D). For sodium channel currents in large DRG neurons, estimated IC₅₀ of lidocaine were 86 ± 6 and 99 ± 6 μM (P = 0.092), respectively, for wild-type and HCN1^−/− mice (fig. 9D). For sodium channel currents in small DRG neurons, estimated IC₅₀ of lidocaine were 103 ± 7 and 111 ± 7 μM (P = 0.098), respectively, for wild-type and HCN1^−/− mice (fig. 9D).

As shown in figure 10, RMP of large neurons was significantly hyperpolarized by HCN1 deletion (−67 ± 10 vs. −56 ± 9 mV, P < 0.001) and R_in of large neurons was significantly increased (72 ± 17 vs. 140 ± 18 MΩ, P < 0.001). For large neurons from wild-type mice, both lidocaine 100 μM and ZD7288 100 μM could hyperpolarize RMP of neurons and increase their R_in. The modulation of lidocaine on RMP and R_in was diminished in large neurons by HCN1^−/− deletion (fig. 10, A and C). However, RMP and R_in of small neurons were unaffected by HCN1^−/− deletion (fig. 10, B and D). For small neurons from wild-type mice and HCN1^−/− mice, lidocaine 100 μM and ZD7288 100 μM could significantly hyperpolarize RMP of neurons and increase their R_in (fig. 10, B and D). All the membrane properties data were listed in table 4.

It is widely believed that sodium channel is the main target for local anesthetics, and by blocking sodium channels, local anesthetics could inhibit propagation of action potential.¹  Besides sodium channel, previous studies have demonstrated that lidocaine could also inhibit HCN channels,^6,7  and enhancement of clonidine and dexmedetomidine on regional anesthesia was achieved by inhibiting HCN channels.^8,9  The current study indicates that HCN channel contributes to regional anesthetic effects of lidocaine. This finding not only provides an innovative regional anesthetic mechanism but also indicates a molecular target for designing novel local anesthetics.

HCN channel is widely expressed on various tissues in both animals and human beings.^13–17  For peripheral nervous system, HCN1 and HCN2 subtypes have been found in DRG^22–25  and nerve fibers.^26,27  Previous study indicated that function of nerve axon was modulated by HCN channels,^28,29  which might contribute to peripheral nerve block. In addition, it has been demonstrated that HCN1 subtype is predominant in large diameter DRG neurons, whereas HCN2 is predominant in small ones.^22–24  This differential distribution between HCN1 and HCN2 subtypes has also been confirmed by electrophysiologic recordings.^22,25  Small DRG neurons are widely known as nociceptor, and most HCN1-positive neurons are A-type neurons.^31,32  We confirmed the differential distribution between HCN1 and HCN2 subtypes in this study because HCN channel currents were almost diminished in large DRG neurons by HCN1 deletion, while these currents were unaffected in small neurons.

In the current study, durations of lidocaine were shortened in HCN1^−/− mice except for pinprick blockade. By blocking HCN2 subtype in HCN1^−/− mice, ZD7288 could prolong durations of lidocaine, including pinprick sensory blockade. Contrarily, forskolin, a cyclic adenosine monophosphate (cAMP) enhancer, could significantly shorten duration of lidocaine for pinprick blockade but with no significant effect on tactile sensory blockade, thermal sensory blockade, and motor blockade. By previous studies, activation of HCN channels could be enhanced by cAMP, and HCN2 subtype is more sensitive than HCN1 subtype.^13,25  Comparing with HCN1 subtype, HCN2 subtype is activated at hyperpolarized potential.^13,40  Thus, it has been demonstrated that cAMP analogs could depolarize the activation potential of HCN channels in C-type neurons but not in A-type.²⁶  For this study, the effect of forskolin on pinprick blockade indicates that HCN2 subtype contributes to high-threshold sensory blockade. Interestingly, previous studies found that HCN1 subtype was more critical than HCN2 for the function of motor axon, and the HCN channels expressed on motor axon were activated at the resting membrane potential.^29,41,42  Therefore, selectively inhibit HCN2 subtype in regional anesthesia might induce a nociceptive-preferred blockade.

By electrophysiologic recordings, we found that HCN1 deletion did not affect the functions of sodium channels. Lidocaine could inhibit sodium channel currents without selectivity between large and small DRG neurons. Meanwhile, lidocaine also inhibited HCN channel currents in both large and small DRG neurons at relevant concentrations. Of note, maximal inhibition of lidocaine on HCN channel currents was merely approximately 50% in the current study while previous study found that lidocaine could almost completely inhibit HCN channel currents when concentration of lidocaine reached 600 μM (while IC₅₀ of lidocaine was 72 ± 7 μM, which was similar to the current study) in central nervous system.⁷  This discrepancy might result from different recording samples and protocols. We further investigated the effect of lidocaine on neural membrane properties. Like ZD7288, lidocaine could hyperpolarize RMP and increase input resistance of neurons at regional anesthetic relevant concentrations. Therefore, HCN channels might contribute to regional anesthesia, and effects of lidocaine could be enhanced by further inhibiting HCN channels via affecting neural membrane properties.

There are still some limitations in the current study. First, the exact role of HCN2 subtype could not be directly determined. Because HCN2 knockout mice are prone to epileptic phenomena and sinoatrial node dysrhythmia,¹³  these abnormality might affect behavioral evaluations of regional anesthesia in vivo. Second, ZD7288 is a nonselective blocker between HCN1 and HCN2 subtypes. Further studies would be more interesting if selective blockers such as MEL57A and/or EC18¹⁵  are applied. Meanwhile, although forskolin has been widely used as an enhancer of HCN channels by raising level of cAMP, cAMP might induce other effects because it is a second messenger. Other cAMP enhancer like lamotrigine and gabapentin might also induce these effects. Third, electrophysiologic recordings were performed on acute isolated DRG neurons in this study, which were of uncertain significance for regional anesthesia/peripheral nerve block. Further study that recorded on isolated peripheral nerve might be even more relevant to in vivo sciatic nerve block.

In summary, the current study indicates that modulation of lidocaine on HCN channel at its clinical relevant concentrations may contribute to regional anesthetic effects of lidocaine. Regional anesthetic durations of lidocaine are prolonged by HCN blocker. Selective inhibition of HCN2 subtype might induce nociceptive-preferred sensory blockade. This finding provides a molecular target for designing long-lasting nociceptive-preferred local anesthetics.

This study was supported by the grant 81371249 (to Dr. Chen) from the National Natural Science Foundation of China (Beijing, China); the grant 81401139 (to Dr. Zhou) from the National Natural Science Foundation of China (Beijing, China); and the grant 2014M552361 (to Dr. Zhou) from the National Postdoctoral Foundation of China (Beijing, China).

The authors declare no competing interests.