Lipid Rescue Reverses the Bupivacaine-induced Block of the Fast Na⁺ Current (I_Na) in Cardiomyocytes of the Rat Left Ventricle

ACCIDENTAL systemic toxicity of local anesthetics (LA) is a rare but potentially life-threatening complication of regional anesthesia.^1,2  The symptoms usually involve the central nervous system and the heart, ranging from rather harmless signs such as mild discomfort, nausea, dizziness, or minor electrocardiogram changes up to seizures, coma, ventricular fibrillation, and asystole.¹  Therapy of severe systemic LA intoxication consisted of advanced cardiac live support and has been a challenge because no specific therapy was available. This situation started to change some 15 yr ago, when infusion of a commercially available lipid emulsion was shown to effectively treat intoxications with the lipophilic LA bupivacaine in rats³  and later in dogs also.⁴  Only a few years later, the concept was transferred to the clinic, when patients with LA intoxication refractory to conventional treatment were successfully treated with infusion of a lipid emulsion.^5,6  Since then, numerous case reports have been published reporting the successful use of lipid emulsion therapy in intoxications with LAs and, more recently, other lipophilic drugs⁷  and, moreover, lipid infusion has meanwhile been included in resuscitation guidelines.⁸ 

To date, the mechanism of action of lipid resuscitation is not well understood. Among different concepts, the so-called lipid-sink hypothesis suggests an accumulation of the lipophilic LA in the lipid phase of the lipid emulsion, thereby decreasing the concentration of the LA in the water phase and hence in the tissue.⁹  According to this concept, there are experimental observations that lipid emulsion infusion is effective in the treatment of intoxications with the lipophilic LA bupivacaine, but not with the less lipophilic LAs mepivacaine¹⁰  and ropivacaine.^11,12  Moreover, with increasing concentrations of the lipid, a decrease in the myocardial concentration of bupivacaine was observed.¹³  However, lipid resuscitation has also been successfully applied to intoxications with the substantially less lipophilic LA lidocaine,¹⁴  suggesting additional modes of action.

Although the exact mechanism of action is not clearly understood, a common clinical observation reported during the application of the lipid emulsion in patients is a rapid improvement in electrocardiogram alterations, i.e., shortening of the QRS complexes, decrease in QTc-interval, or termination of ventricular arrhythmias.^15,16  This suggests a rather rapid action of the lipid emulsion on the effects of the LA on the cardiac action potential (AP) and hence, the underlying ion channels.

In the current study, we therefore asked whether the application of a clinically used lipid emulsion might reverse the bupivacaine-induced changes in the AP and the fast Na⁺ current (I_Na) of left ventricular cardiomyocytes and compared the results with those obtained with the hydrophilic LA mepivacaine. We show that a lipid emulsion partially reverses the effects of bupivacaine on the AP and the Na⁺ current at the cellular level. Moreover, we provide evidence for a lipid-sink mechanism. Finally, by investigating the effect of lipid emulsion on the LA action on heterologously expressed hKv4.2 + hKChIP2b channels in Xenopus laevis oocytes, we show that the effect of the lipid emulsion is not limited to the AP and the Na⁺ current in cardiomyocytes, but appears as to be a general mechanism on ion channels.

The investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the U.S. National Institutes of Health (National Institutes of Health Publication No. 85-23, revised 1996) and was approved by local government authorities (Government of Middle Franconia, Ansbach, Germany). Myocytes were isolated from the cardiac left ventricular free wall of female Wistar rats (weighing approximately 220 g) as described previously.^17,18  Briefly, after induction of deep anesthesia by intraperitoneally injecting thiopental-sodium (100 mg/kg body mass; Inresa Arzneimittel GmbH, Freiburg, Germany), the heart was quickly excised and placed into cold (4°C) Tyrode solution where it stopped beating immediately. Subsequently, the heart was perfused with modified Tyrode solution containing 4.5 mM Ca²⁺ and 5 mM EGTA (approximately 1 µM free Ca²⁺) via the ascending aorta at 37°C for 5 min. The perfusion was continued for 19 min, recirculating 25 ml of the same solution containing collagenase (CLS type II, 160 U/ml; Biochrom KG, Berlin, Germany) and protease (type XIV, 0.6 U/ml; Sigma-Aldrich Corporation, St. Louis, MO). Finally, the heart was perfused with storage solution¹⁹  containing 100 µM Ca²⁺ for 5 min. Using fine forceps, myocytes were carefully dissected from the subepicardial and the subendocardial layer of the left ventricular free wall and placed in cell culture dishes containing the same solution. Tissue pieces were minced and gently agitated to obtain single cardiomyocytes. After adaption to physiological Ca²⁺ levels, cells were transferred to cell culture dishes containing storage solution supplemented with 100 U/ml penicillin and 0.1 mg/ml streptomycin, stored at 37°C in a water saturated atmosphere containing 5% CO₂ and used for experiments for up to 36 h. Only quiescent single rod-shaped cells with clear cross striations were used for experiments.

The ruptured-patch whole cell configuration was used as described previously.^18,20  Currents were recorded using an EPC-10 amplifier (HEKA Elektronik Dr. Schulze GmbH, Lambrecht, Germany), controlled by the PULSE-Software (HEKA). Membrane capacitance (C_m) and series resistance (R_s) were calculated using the automated capacitance compensation procedure of the EPC-10 amplifier. During AP measurements, R_s averaged 6.8 ± 2.9 MΩ (mean ± SD, n = 95) and C_m was 130.2 ± 35.0 pF (mean ± SD, n = 95). During I_Na measurements, R_s averaged 5.2 ± 1.7 MΩ (mean ± SD, n = 88) and was compensated by 80%. This resulted in an average effective R_s of 1.0 MΩ, leading to an average voltage error of 5.5 mV (average current −5.3 ± 2.4 nA [mean ± SD], n = 88) for the native current, 2.5 mV (average current −2.4 ± 1.0 nA [mean ± SD], n = 41) when blocked with bupivacaine and 4.0 mV (average current −3.8 ± 1.8 nA [mean ± SD], n = 47) when blocked by mepivacaine (see also the section on limitations in the Discussion). C_m averaged 115.8 ± 24.3 pF (mean ± SD, n = 88). Effective R_s and C_m resulted in an average time constant of 120 µs for charging of the membrane capacitance. Together with the relatively slow kinetics of I_Na at −40 mV at room temperature, this allowed the peak of the capacitive artifact to be clearly separated from the peak of the Na current. The capacitive artifact was subtracted using a P/4 leak subtraction protocol. Pipette potentials were corrected for the liquid junction potential of 13 or 9 mV for the solution with reduced Na⁺ concentration, respectively. All experiments were performed at room temperature (22°–24°C). For each set of experiments, myocytes obtained from two to four rats were used. Some control experiments were performed on cells from one rat only. All reported potentials are pipette potentials.

Trains of 150 APs were elicited at 1 Hz in cells with a resting membrane potential negative to −80 mV (a V_m positive to −80 mV was considered to be secondary to a leaky electrical access to the myocyte) by depolarizing current pulses of 5 ms duration. After 50 APs, the solution was switched to the solution containing the LA (bupivacaine or mepivacaine). After another 50 APs, the solution was switched to the mixture of lipid emulsion and the LA. The last AP under each condition was evaluated.

To elicit I_Na, cells were clamped to −40 mV for 20 ms and then for 80 ms to 0 mV. Holding potential was −90 mV and cycle length was 1 s. I_Na was measured at the pulse to −40 mV to assess the Na⁺ current during the upstroke of the AP while the pulse to 0 mV served to simulate the plateau phase of the AP. The standard protocol consisted of 200 pulses: 50 under control conditions, 50 with LA, 50 with LA plus lipid emulsion, and another 50 under control conditions. The last current under each condition was evaluated. Leak subtraction was applied using a P/4 protocol, and R_s and C_m were automatically readjusted before each pulse. The extracellular Na⁺ concentration was reduced to 20 mM by replacing 118 mM Na⁺ with Cs⁺. Moreover, this inhibited the inward-rectifying K⁺ current,²¹  thus rendering the leak subtraction possible. All experiments were conducted under continuous perfusion of 7 ml/min. Current and voltage recordings were low-pass filtered at 5 kHz and sampled at 25 kHz. To compensate for variability in cell size, currents were divided by the cell capacitance and are thus given as current densities in pApF^-1.

Female X. laevis were anesthetized by immersing in tap water containing 0.2% MS-222 for 10 min. Ovarian lobes were surgically removed, and oocytes were isolated by enzymatic digestion using collagenase (CLS type II, 260 U/ml; Biochrom KG, Berlin, Germany) in Ca²⁺-free OR2 solution at 10°C for 3–4 h. We used full-length complimentary DNA transcripts encoding human Kv4.2 (hKv4.2) inserted in pGEM and human KChIP2b (hKChIP2b) included in pGEM-HJ. Linearized plasmids were used as templates for coding RNA synthesis using the mMessage mMachine Transcription Kit T7 (Life Technologies, Grand Island, NY). Defolliculated stage V and VI oocytes were injected with 0.1 ng hKv4.2 + 0.5 ng hKChIP2b coding RNA. Coding RNAs were dissolved in RNAse-free water, and the total volume injected was 50 nl per oocyte. After injection, oocytes were maintained in ND96 solution and were studied 2 days after injection.

Oocytes were transferred to a perfusion chamber, which is continuously superfused with NaCl-95 solution, and impaled with electrodes (0.1–1.5 MΩ) filled with 3 M KCl. To increase the Lipovenös® (Fresenius Kabi AG, Bad Homburg, Germany) content to more than 10% in the bath solution, a stock of a modified bath solution was designed which was diluted as needed with Lipovenös® or control to assure a constant K⁺ concentration of 4 mM and Ca²⁺ and Mg²⁺ concentrations of 1 mM each. Whole cell currents were measured at room temperature (19°–22°C) with the two-electrode voltage clamp technique using an OC-725C amplifier (Warner Instruments LLC, Hamden, CT) controlled by the Pulse-software (HEKA) via an LIH-1600 interface (HEKA). An Ag–AgCl pellet placed directly in the bath solution served as a reference electrode for the current injection circuit, whereas an additional Ag–AgCl pellet located close to the oocyte was used to sense the bath potential to minimize series resistance errors. Currents were elicited by voltage steps to +40 mV from a holding potential of −90 mV. Pulsed current data were filtered at 1 kHz and sampled at 5 kHz.

Bupivacaine and mepivacaine concentrations were measured by gas chromatography–mass spectrometry using an Agilent model 6890plus gas chromatograph and an MSD 5973 in the electron impact selected ion monitoring mode (Agilent Technologies Inc., Santa Clara, CA). Extraction of the samples was adopted from the study by Colin et al.²²  Shortly, samples (50 µl) were extracted by fluid–fluid extraction using hexane:ethyl acetate (70:30, 4 ml) plus 0.05% Lipofundin® (20%; B. Braun Melsungen AG, Melsungen, Germany) and glycine buffer at pH 9.0 (250 µl), dried, solved in toluol:methanol (20:1, 250 µl) and injected into the gas chromatograph (0.2 µl, split mode 20:1 at 285°C). Ropivacaine (m/z = 126.1 and retention time = 3.97 min) was used as internal standard for bupivacaine (m/z = 140.1 and retention time = 4.18 min, r² = 0.99) and mepivacaine (m/z = 98.1 and retention time = 3.65 min, r² = 1.00) quantification. Separation was achieved on a Phenomenex Zebron ZB-1ms column (30 m × 0.25 mM × 0.25 µM; Phenomenex Inc., Aschaffenburg, Germany) at a starting temperature and period of 200°C and 2 min followed by a linear temperature gradient (60 K/min) up to 300°C. Linearity of the detector was given over the whole concentration range with relative recovery rates of 95.7 ± 3.7% (mepivacaine, n = 17) and 100.0 ± 6.3% (bupivacaine, n = 17) and lower limits of quantification of 1.43 µM (mepivacaine) and 1.03 µM (bupivacaine).

Modified Tyrode solution was used for cell isolation and as bath solution and contained NaCl, 138 mM; KCl, 4 mM; MgCl₂, 1 mM; NaH₂PO₄, 0.33 mM; CaCl₂, 2 mM; glucose, 10 mM; and HEPES, 10 mM (pH 7.30 with NaOH). The Na⁺ concentration was reduced to 20 mM by replacing 118 mM Na⁺ with Cs⁺ for measuring I_Na. Myocyte storage solution contained NaCl, 130 mM; NaH₂PO₄, 0.4 mM; NaHCO₃, 5.8 mM; MgCl₂, 0.5 mM; CaCl₂, 1 mM; KCl, 5.4 mM; glucose, 22 mM; and HEPES, 25 mM (pH 7.40 with NaOH in the presence of 5% CO₂) and supplemented with 1 mg/ml bovine serum albumin (albumin fraction V; Merck KGaA, Darmstadt, Germany), 100 U/ml of penicillin, and 0.1 mg/ml of streptomycin. The pipette solution contained glutamic acid, 120 mM; KCl, 10 mM; MgCl₂, 4 mM; EGTA, 10 mM; HEPES, 10 mM; and Na₂ATP, 2 mM (pH 7.20 with KOH). ND96 solution contained NaCl, 96 mM; KCl, 2 mM; MgCl₂, 1 mM; CaCl₂, 1.8 mM; and HEPES, 5 mM (pH 7.40 with NaOH), supplemented with 100 U/ml of penicillin and 0.1 mg/ml of streptomycin (Sigma-Aldrich Corporation). NaCl-95 solution contained NaCl, 95 mM; KCl, 4 mM; MgCl₂, 1 mM; CaCl₂, 1 mM; and HEPES, 10 mM (pH 7.40 with Tris). The ×5 stock of the modified bath solution contained NaCl, 60 mM; KCl, 18.25 mM; MgCl₂, 5 mM; CaCl₂, 5 mM; and HEPES, 20 mM (pH 7.4 with NaOH).

Bupivacaine and mepivacaine were obtained from Sigma and were freshly dissolved in the bath solution on each day of experiments. Bupivacaine was used at a concentration of 10 µM in most experiments, a plasma concentration that can be reached in patients.¹⁶  Mepivacaine was used at a concentration of 40 µM, a concentration that was reported to cause similar cardiotoxicity as 10 µM bupivacaine.¹⁰  Experiments in which the lipid emulsion (Lipovenös® MCT 20%; Fresenius Kabi AG) was applied were started with a bath solution containing 10% of a control solution adapted to the Na⁺ content (approximately 2 mM) and osmolarity (290 mOsm, adapted with mannitol in the control solution) of Lipovenös®. Eventually, the solution was exchanged to the same solution containing 10% of the lipid emulsion instead of control. Ultracentrifugation was performed at 40,000 rpm (110,000g) in an Optima L60 ultracentrifuge (Beckman Coulter Inc., Brea, CA) using a Ti70.1 rotor for 2 h at 4°C.

Data were analyzed using the PULSE-FIT software (HEKA), IGOR Pro (WaveMetrics Inc., Lake Oswego, OR), and Microsoft Excel (Microsoft Corporation, Redmond, WA) as described previously.¹⁸  To indicate the precision of the mean values, average data are given together with the SEM (mean ± SEM) and the number of experiments if not stated otherwise. To report the variability of the data, means together with their respective SDs are given in the table, Supplemental Digital Content 1, https://links.lww.com/ALN/A968, and 95% CIs are detailed in the table, Supplemental Digital Content 2, https://links.lww.com/ALN/A969.²³  To estimate the reduction in the bupivacaine concentration by the lipid emulsion, concentration–response analysis was performed in the relevant range of concentrations assuming the previously described complete block of I_Na at high concentrations of bupivacaine.²⁴  Concentration–response curves of Kv channels were calculated at +40 mV from charge rather than current amplitude because bupivacaine and mepivacaine are open-channel blockers of Kv4 channels.^25,26  In both cases, a Hill function with variable slope was fitted to the data. Statistical significance was evaluated by paired or unpaired Student t test or paired or unpaired one-way ANOVA followed by a Newman–Keuls multiple comparison test when more than two groups were compared using Prism 5 (GraphPad Software Inc., La Jolla, CA). A two-tailed P value of less than 0.05 was considered statistically significant.

Action potentials were recorded to assess the overall effect of lipid rescue on the cardiac cellular electrophysiology. Because of the heterogeneity of AP duration (APD) within the rat left ventricular free wall,²⁷  myocytes from epicardial and endocardial regions were analyzed separately. Figure 1 depicts the effect of lipid rescue on the epicardial AP; 10 µM of the lipophilic bupivacaine, a concentration reportedly reached during intoxication,¹⁶  significantly reduced the overshoot (fig. 1D; P < 0.001) and maximal upstroke velocity (by 56%; fig. 1E; P < 0.001) of the AP (table 1). Lipid rescue with 10% Lipovenös® partially reversed the effect on the overshoot (P < 0.001) and increased maximal upstroke velocity by 31% (P < 0.001; table 1). Similar effects were observed in endocardial myocytes (table 1). Bupivacaine also significantly increased APD to 50% (APD₅₀) and 90% (APD₉₀) repolarization, which was completely reversed by lipid rescue (fig. 1, F and G; table 1) in epicardial myocytes. In endocardial myocytes, AP prolongation by bupivacaine was less profound and, surprisingly, lipid rescue shortened APD₉₀ even below its duration measured under control conditions (table 1). Because the hydrophilic LA mepivacaine is reportedly less cardiotoxic than bupivacaine,¹⁰  a fourfold higher dose was used. Similar to bupivacaine, 40 µM mepivacaine significantly reduced AP overshoot and upstroke velocity (P < 0.001 each; fig. 2, A–E; table 2). However, it did not significantly alter repolarization (fig. 2, F and G; table 2). As expected for a relatively hydrophilic LA, lipid rescue did not reverse the effects on overshoot and upstroke velocity. In fact, upstroke velocity was even slightly but significantly reduced by Lipovenös® (fig. 2, A–E; table 2). Surprisingly, Lipovenös® somewhat prolonged APD₅₀ (P < 0.001) and APD₉₀ (P < 0.05) in the presence of mepivacaine (fig. 2, F and G; table 1). Both LAs tended to slightly hyperpolarize the resting membrane potential. This effect was significant in some of the experiments (table 1). Lipovenös® did not have a consistent effect on the resting membrane potential. The tendency of Lipovenös® to prolong epicardial and to shorten endocardial APD was tested in a series of control experiments. Lipovenös® alone marginally but significantly prolonged APD₉₀ in epicardial myocytes (46.8 ± 4.5 ms, n = 10 vs. 43.8 ± 4.3 ms, n = 10; P < 0.05) and tended to shorten the endocardial APD₉₀ (87.5 ± 15.4 ms, n = 5 vs. 105.9 ± 23.1 ms, n = 5, not significant). Resting membrane potential, maximal upstroke velocity, and overshoot were not significantly altered (not shown).

The results on the upstroke velocity and overshoot of the AP demonstrate that Lipovenös® reversed the block of cardiac I_Na induced by bupivacaine but not or to a substantially lesser degree by mepivacaine. The AP prolongation induced by bupivacaine is consistent with an additional block of repolarizing K⁺ currents by this LA²⁵  which was also reversed by Lipovenös®. To further characterize the effect of lipid rescue therapy on the level of ionic currents, we decided to focus on I_Na because it is the primary target of LAs in the heart. Moreover, we will also detail some findings on heterologously expressed repolarizing K⁺ channels.

Figure 3, A and B, depicts representative Na⁺ currents from left ventricular cardiomyocyte, and figure 3C depicts average normalized I_Na amplitudes. The inset of figure 3A shows the pulse protocol. Bupivacaine at 10 µM reduced I_Na by 54% (−19.3 ± 1.9 pApF⁻¹vs. −42.3 ± 4.3 pApF⁻¹; n = 17; P < 0.001). Lipovenös® of 10% increased I_Na by 37% in the presence of bupivacaine (−26.4 ± 2.8 pApF⁻¹; n = 17; P < 0.001). Washout with control solution returned I_Na to control values.

A “lipid sink,” i.e., the reduction in the free LA concentration by absorption of the LA by the lipid, has been proposed as a possible mechanism of lipid rescue therapy.²⁸  To directly assess the contribution of the lipid sink, we removed the lipid phase from the solution containing both 10 µM bupivacaine and 10% Lipovenös® by ultracentrifugation. If a lipid sink contributes to the effects of Lipovenös®, the water phase of the centrifuged solution should, due to the reduced bupivacaine concentration, have a similar effect as the whole solution. Figure 3, D–F, illustrates the effects of the water phase: I_Na was increased by 17% (−27.3 ± 2.0 pApF⁻¹vs. −23.0 ± 1.7 pApF⁻¹; n = 17; P < 0.001, control: −50.1 ± 4.0 pApF⁻¹), demonstrating the contribution of the lipid-sink effect. However, the total effect of Lipovenös® seems to be larger than the effect of the water phase alone, indicating that part of the effect is dependent on the presence of the lipid (“direct lipid effect”).

According to the lipid-sink hypothesis, the effect of Lipovenös® on the mepivacaine-induced block should be markedly smaller than in the case of bupivacaine. Lipovenös® increased I_Na blocked by 40 µM mepivacaine by 17% (−35.1 ± 3.9 pApF⁻¹vs. −30.0 ± 3.5 pApF⁻¹; n = 24; P < 0.001, control: −44.3 ± 5.4 pApF⁻¹, fig. 4, A–C). Surprisingly, also the water phase of the solution containing Lipovenös® and mepivacaine increased I_Na by 8% (−37.9 ± 3.6 pApF⁻¹vs. −35.1 ± 3.2 pApF⁻¹; n = 23; P < 0.01, control: −48.2 ± 4.4 pApF⁻¹) compared with mepivacaine alone (fig. 4, D–F).

The total effect of Lipovenös® on the LA blocked I_Na was significantly (P < 0.05) larger for bupivacaine than for mepivacaine (fig. 5A). As expected from the higher degree of lipophilicity of bupivacaine, this was due to a significantly (P < 0.01) higher lipid-sink effect (fig. 5B), whereas the direct lipid effect was similar for both LAs (fig. 5C). For this comparison, the direct lipid effect was assessed by dividing the effect of the complete Lipovenös® solution in each experiment by the mean effect of the water phase and normalized to the first pulse after perfusion with the Lipovenös® solution. Consistent with a direct lipid effect, Lipovenös® also significantly (P < 0.01) increased I_Na in the absence of LA (fig. 5D).

To estimate the bupivacaine concentration in the water phase of the centrifuged solution, the relative increase in I_Na was plotted to a concentration–response curve obtained for the relevant range of concentrations (fig. 5E). Interpolation using a Hill function indicated that the increase in the blocked I_Na by the lipid-sink effect corresponded to a reduction in the bupivacaine concentration in the water phase to approximately 5.9 µM. Notably, the total increase in the blocked I_Na after the application of bupivacaine + Lipovenös® (total lipid effect) was as large as if the bupivacaine concentration had been reduced to approximately 3.9 µM.

To investigate whether lipid rescue of ionic currents is a general mechanism, additional experiments on hKv4.2 + hKChIP2b currents in X. laevis oocytes were performed, thereby investigating the lipid effect on a different ion channel in a different experimental system. Kv4.2 and the β subunit KChIP2 are ion channel subunits that underlie the repolarizing transient outward K⁺ current (I_to) in cardiomyocytes. Figure 6, A and B, depicts representative hKv4.2 + hKChIP2b currents measured at +40 mV from an oocyte exposed to 0, 100, and 300 µM concentrations of bupivacaine (fig. 6A) and to the same concentrations of bupivacaine in the presence of 10% Lipovenös® (fig. 6B). The lipid emulsion significantly increased the IC₅₀ of bupivacaine (calculated from the charge at +40 mV) from 126.6 ± 17.9 µM (n = 14) to 218.2 ± 29.4 µM (n = 15; P < 0.05; fig. 6C). The hillslope was unaffected (1.1 ± 0.1, n = 14 vs. 1.3 ± 0.1, n = 15, n.s.). The IC₅₀ for mepivacaine was considerably higher (1,688 ± 235 µM, n = 13) and tended to be lower rather than higher in the presence of the lipid (1,203 ± 146 µM, n = 19; P = 0.07; fig. 6D). Also for mepivacaine, the hillslope was unaffected (1.0 ± 0.0, n = 13 vs. 0.9 ± 0.0, n.s., n = 19).

In a separate series of experiments, using a highly concentrated stock of the bath solution, we were able to record the concentration–response of hKv4.2 + hKChIP2b currents blocked by 300 µM bupivacaine to increased Lipovenös® concentrations. The relative charge increased quite linearly up to 40% of Lipovenös® concentration and was nearly twice as high at 80% lipid emulsion as at 10% (82.5 ± 1.2% vs. 42.9 ± 2.3%, n = 13; P < 0.001; fig. 6E).

The contribution of a lipid-sink mechanism was also evaluated in this series of experiments (fig. 6F); after ultracentrifugation, the water phase of the lipid and 100 µM bupivacaine-containing solution was as effective as the whole mixture in reversing the bupivacaine-induced block (whole solution: +34.4 ± 3.9%, n = 10, P < 0.001; water phase: +39.2 ± 3.6%, n = 10; P < 0.001; effect whole solution vs. effect water phase: n.s.). Together with the mepivacaine data, this indicates that at 10%, Lipovenös® did not exert a direct lipid effect on hKv4.2 + KChIP2b channels.

The data presented indicate the presence of a lipid-sink effect. To directly assess this effect, LA concentrations were measured by gas chromatography followed by mass spectrometry (table 3). After ultracentrifugation of the solution containing bupivacaine and 10% Lipovenös®, the concentration of bupivacaine (original concentration: 10 µM and 100 µM) was significantly reduced in the water phase by approximately 30% (P < 0.001; table 3), demonstrating the lipid-sink effect. The concentration of bupivacaine in the water phase was remarkably close to the concentration estimated by concentration–response analysis. Concomitantly, the concentration in the lipid-enriched phase was increased. Also for mepivacaine, we noted a small decrease in the concentration in the water phase, which was significant for 10 µM only. The decrease in LA concentration was larger for bupivacaine than for mepivacaine (at 10 µM: P = 0.05 and at 100 µM: P < 0.001). These data confirm our interpretation of the electrophysiological experiments.

We demonstrate lipid rescue of the cardiac AP and I_Na after intoxicating cardiomyocytes with bupivacaine. Lipid rescue did not reverse the effects of mepivacaine on the AP, although it increased I_Na. Part of the lipid effect was attributable to a lipid-sink mechanism. Lipid rescue was also effective on hKv4.2 + KChIP2b currents in X. laevis oocytes resulting in a rightward shift of the concentration–response curve for bupivacaine, but not for mepivacaine.

Local anesthetics block inactivated Na⁺ channels;^24,29  therefore, the block exhibits use-dependence and depends on the membrane potential.^1,30  We concentrated on the clinically relevant use-dependent block using APs and a voltage protocol mimicking the physiological resting membrane potential and AP shape of the cardiomyocyte. The effects of bupivacaine on AP upstroke and I_Na are in good agreement with previous findings where the IC₅₀ of bupivacaine for cardiac Na⁺ currents was between 3 and 8.6 µM, depending on model and enantiomer.^24,29,31  For mepivacaine, only limited information on cardiac I_Na is available.¹  In a study conducted on closely related³²  tetrodoxin-resistant neuronal Na⁺ currents,³³  the IC₅₀ of use-dependent block was 13–15 µM for bupivacaine and 70–90 µM for mepivacaine.

The APD prolongation by bupivacaine indicates a block of repolarizing K⁺ channels. Cardiac repolarization in rat is achieved by the transient outward K⁺ current (I_to), slowly or noninactivating currents, and the inward-rectifying K⁺ current I_K1.²⁷  For I_to charge block, Castle²⁵  reported an IC₅₀ of 23 µM for bupivacaine and 790 µM for mepivacaine. I_K1 was not affected by bupivacaine and blocked approximately 50% by 3 mM of mepivacaine. These data explain why bupivacaine but not mepivacaine prolonged APD in our study. Our IC₅₀ for hKv4.2 + KChIP2b channels expressed in X. laevis oocytes is somewhat higher than that reported for I_to.²⁵  This is consistent with findings from our group showing greater IC₅₀ values for the aminoquinolines primaquine and chloroquine in oocytes expressing hKv4.2 + hKChIP2b compared with native I_to.³⁴ 

We examined the effects of a clinically used lipid emulsion on single cardiomyocytes at a concentration that can be reached in lipid resuscitation in patients.¹⁶  Bupivacaine concentrations and the experimental design were chosen to replay the situation in the patient and to provide insight into the underlying mechanisms. Experiments, in which the lipid phase of the solution containing Lipovenös® and bupivacaine was removed by ultracentrifugation, indicated that a part of the effect of Lipovenös® on AP and I_Na was independent of the presence of the lipid. This is an experimental evidence for the contribution of a lipid-sink effect to the effects of Lipovenös® in our model. However, another part of the effect was dependent on the presence of the lipid. This effect was also present in the absence of the LA and was largely responsible for lipid rescue of I_Na after block by mepivacaine. We could not demonstrate lipid rescue from mepivacaine of the AP. This was probably due to a combination of the small-sized effect on I_Na together with a slight depolarization of the resting membrane potential by the lipid emulsion which likely led to an increased inactivation of the Na⁺ channels, sensitizing I_Na for mepivacaine. This depolarization was not present in the AP experiments with bupivacaine.

Lipid rescue was also assessed for hKv4.2 + KChIP2b channels. Concentration–response experiments demonstrated an increased IC₅₀ of bupivacaine but not of mepivacaine in the presence of the lipid emulsion. Moreover, we also demonstrated a concentration dependence of the effect of the lipid emulsion. At 10%, the lipid itself did not influence the Kv currents.

The cardiac I_Na is a large current with rapid activation and inactivation kinetics, potentially driving the whole cell patch clamp technique to its limits. To minimize voltage errors, we lowered the extracellular Na⁺ concentration to reduce I_Na and used low-resistance pipettes and R_s compensation to reduce the effective R_s. This resulted in an average time constant of 120 µs for the capacitive artifact and an average voltage error between 5.5 (control conditions) and 2.5 mV (bupivacaine block). This voltage error was small but nonetheless different between the control and blocked conditions and could possibly affect our results. To assess its consequences, we evaluated the degree of block induced by bupivacaine and mepivacaine separately for the 25% of cells with the largest control currents (median bupivacaine: −8.2 nA, mepivacaine: −8.9 nA) and the 25% with the smallest control currents (median bupivacaine: −3.1 nA, mepivacaine: −2.4 nA). Block was similar in these groups both for bupivacaine (large I_Na: 55 ± 1%, n = 10, small I_Na: 53 ± 2%, n = 11, n.s.) and for mepivacaine (large I_Na: 30 ± 2%, n = 11, small I_Na: 28 ± 2%, n = 12, n.s.). Moreover, our data on I_Na correlate well with the AP upstroke velocity, which is largely unaffected by R_s. Taken together, a distortion of our results by the voltage error is unlikely.

Lipid resuscitation has been effectively used to treat intoxications with lipophilic LAs and other lipophilic drugs. Moreover, there are reports in which lipid rescue has also successfully been used in intoxications with less lipophilic drugs such as mepivacaine.¹⁵  In the current study, the effectiveness of lipid rescue of the cardiac AP correlated with the lipophilicity of the LA, and we could demonstrate that a large part of this effect was due to a lipid-sink mechanism. However, the slight reversal of the mepivacaine-induced block of I_Na might provide encouragement to further investigate lipid rescue of less lipophilic compounds also.

Recently, the contribution of ion channel block to cardiac LA toxicity and the clinical importance of the lipid-sink effect have been questioned, and alternative explanations have been proposed.²⁸  In this context, it is helpful consider the specific models and symptoms examined. In most experimental studies, bupivacaine was applied at a dose that resulted in asystole. However, many case reports imply that arrhythmias, including bradycardia, supraventricular, and ventricular tachycardia, are more common in the clinical situation than primary asystole.¹⁵  In the electrocardiogram, the typical early sign of bupivacaine intoxication is broadening and deformation of the QRS complex, indicating a delayed conduction of cardiac excitation consistent with Na⁺ channel block.^35–37  The QTc time is mildly prolonged^36,37  due to K⁺ channel block.^2,26,38–41  The main reason for ventricular tachycardia and consecutive ventricular fibrillation in these patients is likely a slowing of cardiac conduction by Na⁺ channel block.²  At higher concentrations, bupivacaine additionally induces a reduction of contractility and asystole, attributable to complete Na⁺ channel block and interference with pacemaking, Ca²⁺ handling, and mitochondrial function.^1,2  Consequently, the partial reversal of the bupivacaine effects on I_Na and the AP may well contribute the clinical effectiveness of lipid resuscitation.

We present evidence for a dual mode of action of the lipid emulsion, the lipid-sink effect and a direct lipid effect. The lipid-sink effect will be most important at plasma concentrations of near IC₅₀ (i.e., at the concentrations used in the current study), where the concentration–response relation is steepest. In fact, plasma concentrations of approximately 10 µM of bupivacaine have been reported in successful lipid resuscitation.¹⁶  The direct lipid effect is unlikely to depend on the bupivacaine concentration because it is also present in the absence of LA. Considering the physiology of the microcirculation, both effects may act sequentially; the lipid emulsion consists of artificial chylomicrons of approximately 500 nM diameter,⁴²  which cannot easily leave the blood vessels (pore size approximately <30 nM),^43,44  and are broken down by lipoprotein lipase to free fatty acids which then leave the capillary by diffusion.^42,45  The first action of lipid rescue in vivo may be a reduction of the free bupivacaine concentration in the plasma and consequently in the interstitial fluid (the lipid-sink effect), before a sufficient concentration of fatty acids reach the myocytes for an additional direct lipid effect. Here also a direct effect of fatty acids on bupivacaine binding by the Na⁺ channel may be important.⁴⁶  Although the current data and results from others^9,11,47  give clear evidence for a lipid sink in vitro and in vivo, it has been difficult to demonstrate the lipid-sink effect in other experiments in vivo and in the clinical setting.^48,49  For example, Litonius et al.⁴⁸  observed a relevant reduction in the free bupivacaine concentration (from >20 µg/l to <13 µg/l) after lipid application in human, which was statistically not significant but nevertheless is consistent with a lipid-sink effect and warrants further examination. Our results exclude neither other beneficial effects of the lipid emulsion that improve metabolic function or contractility nor effects present only at the whole body level such as accelerated redistribution of the LA.²⁸ 

Taken together, we demonstrate lipid rescue of cardiac electrophysiology from bupivacaine intoxication at the cellular level and identify a lipid-sink effect and a direct lipid effect as underlying mechanisms.

The authors gratefully thank the expert technical assistance of Céline Grüninger, M.Sc., Jessica Rinke, and Ralf Rinke, M.Sc., all from the Institut für Zelluläre und Molekulare Physiologie, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany. The authors are indebted to Angelika Lampert, M.D., Ph.D., Institut für Physiologie und Pathophysiologie, Friedrich-Alexander-Universität Erlangen-Nürnberg, for help with designing the pulse protocols, and to Olaf Pongs, Ph.D., Zentrum für Molekulare Neurobiologie Hamburg, Hamburg, Germany, for providing the plasmids coding for hKv4.2 and hKChIP2b.

Support was provided by Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany; Universitätsklinikum Regensburg, Regensburg, Germany; and the Johannes und Frieda Marohn-Stiftung, Erlangen, Germany.

The authors declare no competing interests.