Modulation of the effects of class Ib antiarrhythmics on cardiac NaV1.5-encoded channels by accessory NaVβ subunits

Native myocardial voltage-gated sodium (NaV) channels function in macromolecular complexes comprising a pore-forming (α) subunit and multiple accessory proteins. Here, we investigated the impact of accessory NaVβ1 and NaVβ3 subunits on the functional effects of 2 well-known class Ib antiarrhythmics, lidocaine and ranolazine, on the predominant NaV channel α subunit, NaV1.5, expressed in the mammalian heart. We showed that both drugs stabilized the activated conformation of the voltage sensor of domain-III (DIII-VSD) in NaV1.5. In the presence of NaVβ1, the effect of lidocaine on the DIII-VSD was enhanced, whereas the effect of ranolazine was abolished. Mutating the main class Ib drug-binding site, F1760, affected but did not abolish the modulation of drug block by NaVβ1/β3. Recordings from adult mouse ventricular myocytes demonstrated that loss of Scn1b (NaVβ1) differentially affected the potencies of lidocaine and ranolazine. In vivo experiments revealed distinct ECG responses to i.p. injection of ranolazine or lidocaine in WT and Scn1b-null animals, suggesting that NaVβ1 modulated drug responses at the whole-heart level. In the human heart, we found that SCN1B transcript expression was 3 times higher in the atria than ventricles, differences that could, in combination with inherited or acquired cardiovascular disease, dramatically affect patient response to class Ib antiarrhythmic therapies.


Introduction
subunits near the domain-III VSD (DIII-VSD) (10)(11)(12). However, the recent structure of Na V 1.5 revealed that Na V β1 interacts with the predominant cardiac Na V α subunit at a distinct site or sites that are characterized by weaker binding and an unresolvable Na V 1.5-Na V β1 complex (13). This difference in comparison with channels encoded by other Na V α subunits is partially due to the unique N-linked glycosylation of Na V 1.5 that hinders its interaction with the Ig domain of Na V β1 (13). Intriguingly, Na V β1 and Na V β3 are highly homologous except in the Ig domains. Previously, optical tracking of the Na V 1.5 VSDs using voltage-clamp fluorometry (VCF) revealed that the Na V β3 subunit modulates both the DIII and the DIV-VSDs, whereas Na V β1 only modulates the DIV-VSD conformational dynamics (14,15). Fluorescence quenching experiments showed that the DIII-VSD is in close proximity to Na V β3 but not Na V β1 (14). These results suggest that Na V β1 and Na V β3 regulate the Na V 1.5 DIII-VSD differently.
The conformational changes in the VSDs are not only important for regulating channel gating; they are also essential for modulating channel interactions with drugs, including those that bind to the pore domain, such as local anesthetics (16). Previously, VCF and gating current recordings showed that when lidocaine blocks Na V 1.4 channels, it stabilizes the DIII-VSD in its activated conformation (17). Moreover, we recently demonstrated that alteration of DIII-VSD conformational changes caused by long QT syndrome 3 variants leads to channels with different mexiletine sensitivities (18,19).
Class I antiarrhythmics modulate cardiomyocyte excitability via Na V channel targeting. Class Ib molecules, such as lidocaine, ranolazine, and mexiletine, specifically modulate the late component of I Na , resulting in shortening of the action potential duration in ventricular cardiomyocytes (20). Lidocaine has long been used to manage ventricular arrhythmias in hospital settings (21). Ranolazine has been shown to be effective in controlling various cases of atrial fibrillation (AF) (22)(23)(24), particularly paroxysmal AF (25,26).
Recently, the RAID trial demonstrated that ranolazine also marginally lowered the risk of recurrent ventricular tachycardia and ventricular fibrillation in high-risk patients with implanted cardioverter-defibrillators (27). Although both drugs are commonly prescribed for several arrhythmias, their efficacies are highly variable. Thus, it remains an important task to understand the determinants of channel-drug interactions that contribute to this variability.
In the experiments presented here, we aimed to understand the molecular mechanisms whereby noncovalently bound Na V β subunits modulate the interaction of class Ib antiarrhythmics with myocardial Na V 1.5 channels. We further investigated the physiological significance of this modulation by assessing ranolazine and lidocaine drug blockade of native Na V currents in mouse ventricular myocytes, probing the mRNA expression levels of Na V β subunits in human hearts and detailing the in vivo electrophysiological phenotypes evident in the cardiac-specific Scn1b-null mouse (28). Our results showed a critical role for β subunits in differentially modulating the efficacy of lidocaine and ranolazine, implying that patient-to-patient differences in β subunit expression are likely to have a significant impact on therapeutic outcomes.

Results
Both lidocaine and ranolazine alter Na V 1.5 DIII-VSD dynamics. Previous studies demonstrated that lidocaine shifts the activation of the DIII-VSD in rat Na V 1.4 channels encoded by Scn4a and prominent in skeletal muscle in the hyperpolarizing direction (16,29,30). Recent findings showed that a class Ib antiarrhythmic, mexiletine, which is similar in structure to lidocaine, also affects the DIII-VSD conformation in Na V 1.5 channels (18,19). The DIII-VSD effect also determines the tonic and use-dependent properties of class Ib drugs (18,19). Taken together, these observations suggest that factors that alter drug effects on the DIII-VSD would be expected to have an impact on therapeutic efficacy.
To explore this hypothesis, we first used VCF to assess the effects of 2 class Ib antiarrhythmics, lidocaine and ranolazine ( Figure 1A), on the DIII-VSD in heterologously expressed human Na V 1.5 channels, which are encoded by SCN5A, the predominant Na V α subunit expressed in the mammalian heart ( Figure  1, B and C). When we expressed the Na V 1.5 α subunit alone in Xenopus oocytes, we observed a hyperpolarizing shift (ΔV 1/2 = -24.8 ± 9.4 mV, P = 0.03) in the DIII fluorescence-voltage (F-V) curve on exposure to 10 mM lidocaine, and a similar shift (ΔV 1/2 = -30.7 ± 7.5 mV, P = 0.05) on application of 4 mM ranolazine, suggesting that both lidocaine and ranolazine stabilize the DIII-VSD in its activated conformation ( Figure  1, B-D). The observation of similar effects on the DIII-VSD caused by both drugs is not surprising because they share similar molecular structures ( Figure 1A), shown previously to interact with residue F1760 in DIV-S6 (20,31). In addition, however, the effects of lidocaine and ranolazine are not identical. Lidocaine, for example, induced a hyperpolarizing shift in the DIV F-V curve, an effect not observed with ranolazine JCI Insight 2021;6(15):e143092 https://doi.org/10.1172/jci.insight.143092 ( Figure 1, B-D), suggesting that despite sharing common binding motifs on the Na V 1.5 α subunit, the distinct chemical structures of lidocaine and ranolazine ( Figure 1A) uniquely regulate DIV-VSD dynamics.
Na V β1 and Na V β3 differentially modulate lidocaine/ranolazine effects on the DIII-VSD. We have previously shown that both Na V β1 and Na V β3 alter DIII-VSD dynamics during Na V 1.5 channel gating (14). Thus, we hypothesized that these Na V β subunits will also alter the effects of class Ib antiarrhythmics on the DIII-VSD. To test this hypothesis, we coexpressed Na V 1.5 with the Na V β1 or Na V β3 subunit and measured DIII-VSD and DIV-VSD conformational changes before and after lidocaine or ranolazine application.
When we coexpressed Na V 1.5 with Na V β1, we observed distinct DIII-VSD responses to lidocaine and ranolazine. Lidocaine induced a greater hyperpolarizing shift (ΔV 1/2 = -57.6 ± 10.2 mV, P = 0.01) in DIII-FV ( Figure 2A) when Na V β1 was present compared with the Na V 1.5 α subunit expressed alone. Exposure to ranolazine, in marked contrast, did not result in a significant DIII-FV shift (ΔV 1/2 = -12.8 ± 16.8 mV, P = 0.53) ( Figure 2C), suggesting that the DIII-VSD was free to move in Na V 1.5 channels in the presence of Na V β1 to recover to the resting state. Although the presence of Na V β1 increased the lidocaine effect on the DIII-VSD, Na V β1 coexpression eliminated the ranolazine effect.
Additionally, coexpression of Na V β1 or Na V β3 with Na V 1.5 both eliminated the hyperpolarizing shift in the DIV F-V curve that was observed with the Na V 1.5 α subunit expressed alone ( Figure 2, A and B), suggesting that the Na V β1 and Na V β3 subunits similarly altered lidocaine's effect on the DIV-VSD.
These results demonstrated that Na V β subunits differentially regulated lidocaine and ranolazine interactions with the DIII-VSD in heterologously expressed Na V 1.5 channels. Specifically, Na V β1 enhanced the effect of lidocaine but decreased the effect of ranolazine on the DIII-VSD activation, whereas Na V β3 coexpression had the opposite effects on both drugs. The altered drug interactions with the DIII-VSD resulted in an enhanced block by lidocaine and reduced block by ranolazine when the Na V 1.5 α subunit was coexpressed with Na V β1 compared with Na V β3 ( Figure 3C).
To determine whether the differential modulation of lidocaine and ranolazine block by Na V β1 and Na V β3 is dependent on the main local anesthetic binding site F1760 (20, 31) ( Figure 3A), we assessed drug blockade of the F1760A-mutant Na V 1.5 channel in the presence of Na V β1 or Na V β3. As expected, the F1760A-mutant channels exhibited much reduced block by lidocaine and ranolazine compared with the WT channels ( Figure 3, C and D). However, application of 10 mM lidocaine or 4 mM ranolazine still caused significant tonic block (TB) and use-dependent block (UDB) of the F1760A channels ( Figure 3, D and E). TB reflects resting-state drug block, while UDB requires prior channel opening (32). In contrast to the WT channels, the hyperpolarizing shift in the DIII F-V upon lidocaine or ranolazine block was not observed with the F1760A-mutant channels ( Figure 3B). The F1760A mutation also eliminated Na V β1 and Na V β3 modulation of TB by lidocaine and ranolazine as well as UDB by lidocaine (Figure 3, D and E). However, despite the absence of a major drug-binding site, coexpression of Na V β3 still caused stronger UDB by ranolazine compared with Na V β1 ( Figure  3E). These results suggest that the effects of Na V β1/β3 on lidocaine and ranolazine block are affected by the F1760 anesthetic binding site but are not completely dependent on it.
Loss of Scn1b expression in mouse cardiomyocytes does not affect Na V channel gating. To further investigate how noncovalent Na V β1/β1B subunits affect the cardiomyocyte response to class Ib antiarrhythmics, we utilized the cardiac-specific Scn1b-null mouse model (Scn1b fl/fl /Myh6-cre) described previously (28). First, we compared I Na in left ventricular (LV) myocytes acutely dissociated from adult Scn1b cardiac-specific null and WT mice. Peak I Na density was increased by 28% in Scn1b-null compared with WT LV myocytes (Scn1b-null: 81.3 ± 3.6 pA/pF, WT: 63.9 ± 5.2 pA/pF, P = 0.017). An increase in I Na density in cardiac-specific Scn1b-null isolated from juvenile mice was previously reported (28). Consistent with the increase in current density, we also observed increased Scn5a transcript expression in the ventricles (and atria) of the Scn1b-null compared with WT mice (Supplemental Figure 1; Supplemental material available online with this article; https:// doi.org/10.1172/jci.insight.143092DS1). Other than increasing peak current density, Scn1b deletion did not measurably alter other Na V channel gating properties in ventricular cardiomyocytes ( Figure 4A kinetics of channel recovery from inactivation ( Figure 4C). Notably, deleting Scn1b did not measurably alter the expression of other Na V β subunits (Supplemental Figure 1). These results, although contrary to previously reported effects of Na V β1 on I Na in heterologous expression systems, are consistent with results obtained in studies on global and cardiac-specific Scn1b-null mice (28,33).
Increased block of I Na by ranolazine but reduced block by lidocaine in adult Scn1b-null mouse ventricular myocytes. Even though Na V channel gating was not measurably affected in cardiac-specific Scn1b-null myocytes, we went on to determine whether the loss of Scn1b affects the responses of native Na V channels to class Ib antiarrhythmics. We examined the effects of lidocaine and ranolazine on TB and UDB of I Na in LV myocytes isolated from WT and cardiac-specific Scn1b-null mice.
The TB produced by 100 μM lidocaine was similar in WT and Scn1b-null LV myocytes ( Figure 5A). In marked contrast, the block of late I Na by lidocaine was significantly reduced in Scn1b-null LV myocytes ( Figure 5B). There was also an approximately 3-fold reduction in lidocaine UDB in Scn1b-null compared with WT LV myocytes (WT: EC 50UDB = 9.3 μM, Scn1b-null EC 50UDB = 24.8 μM) ( Figure 5E). Conversely, ranolazine increased TB, late I Na block, and UDB in Scn1b-null compared with WT adult mouse LV myocytes ( Figure 5, C, D, and F) (WT: EC 50UDB = 53.3 μM, Scn1b-null EC 50UDB = 36.0 μM). The differences in UDB by lidocaine between WT and Scn1b-null myocytes depended on the frequency and duration of the depolarizing pulses ( Figure 5G). In response to 10 μM lidocaine, I Na from Scn1b-null showed decreased UDB compared with WT myocytes at 10 Hz (25 ms duration) and 2 Hz (400 ms duration) but not 5 Hz (25 ms duration) ( Figure 5G). In contrast, in response to 10 μM ranolazine, I Na in WT myocytes showed increased UDB compared with Scn1b-null myocytes at all 3 frequencies ( Figure 5H). Both lidocaine and ranolazine are known to cause a hyperpolarizing shift in the voltage dependence of steady-state inactivation of cardiac I Na (24,34), indicating that drug binding promotes channel inactivation at more hyperpolarized membrane potentials. Therefore, we also compared the voltage dependence of I Na inactivation in WT and Scn1b-null LV  In the presence of Na V β1, the hyperpolarizing shift in the DIII F-V curve produced by lidocaine was enhanced compared with the Na v 1.5 α subunit expressed alone. In marked contrast, the DIV F-V curve was not affected by lidocaine with Na V β1 present. (B) In contrast with Na V β1 (A), the hyperpolarized shift in the DIII F-V curve induced by lidocaine was eliminated when Na V β3 was coexpressed. Similar to Na V β1, however, the DIV F-V curve was minimally affected by lidocaine. (C) In the presence of Na V β1, the effect of ranolazine on the DIII F-V curve was eliminated, whereas the DIV F-V was slightly hyperpolarized. (D) In contrast with Na V β1 (C), the hyperpolarized shift in the DIII F-V curve caused by ranolazine was enhanced when Na V β3 was coexpressed. In the presence of Na V β3, ranolazine also caused a small hyperpolarizing shift in the DIV F-V curve. Each data set represents mean ± SEM values from 4-6 cells.
JCI Insight 2021;6(15):e143092 https://doi.org/10.1172/jci.insight.143092 myocytes before and after lidocaine or ranolazine application. These experiments revealed 100 μM lidocaine induced a hyperpolarizing shift in I Na inactivation in WT LV myocytes and a smaller shift in Scn1b-null LV myocytes (Supplemental Figure 2A). Conversely, 100 μM ranolazine induced a comparable leftward shift in the voltage dependence of inactivation of I Na in WT and Scn1b-null LV myocytes (Supplemental Figure 2B).
Overall, these cellular studies revealed that in adult mouse LV myocytes, the cardiac deletion of Scn1b resulted in reduced lidocaine UDB but increased ranolazine UDB. These results are consistent with our In contrast to WT, Na V β1 and Na V β3 no longer exerted a significant effect on lidocaine and ranolazine TB. (E) UDB by lidocaine and ranolazine in F1760A channel coexpressed with Na V β1 or Na V β3. There was no change in lidocaine UDB comparing coexpression with Na V β1 and Na V β3. However, the presence of Na V β1 caused a reduced ranolazine UDB compared with Na V β3, a phenomenon that is similar to Na V β1's effects on the WT channel. Each data set represents mean ± SEM values from 3-6 cells. Unpaired 2-tailed Student's t test was used to test significance (C-E). *P < 0.05.
JCI Insight 2021;6(15):e143092 https://doi.org/10.1172/jci.insight.143092 VCF data (Figures 1 and 2), suggesting that the presence of Na V β1 subunits enhanced lidocaine's effects but reduced ranolazine's effects on the Na V 1.5 DIII-VSD. The reduced effects on the DIII-VSD are also consistent with the decreased UDB of I Na observed in LV myocytes ( Figure 5).
Ranolazine and lidocaine induced distinct ECG phenotypes in WT and Scn1b-null mice. To understand how Na V β1/β1B modulate antiarrhythmic responses at the whole-heart level, we measured surface ECGs in anesthetized WT and Scn1b-null mice before and after i.p. injection of lidocaine or ranolazine ( Figure 6 and Supplemental Table 1). From the raw ECG data, we quantified several parameters that describe overall heart electrical functioning, including RR intervals, providing a measure of heart rates; P wave intervals, representing atrial conduction; PR intervals, characterizing atrial-ventricular conduction; QRS intervals, revealing ventricle conduction; and QT and ST intervals, corresponding to ventricular repolarization.
We found that 20 mg/kg ranolazine caused QRS prolongation in both WT and Scn1b-null mice ( Figure  6, A and B), but that P wave and PR interval prolongation only occurred in the Scn1b-null mice ( Figure  6B). The QT interval, but not the ST interval, was also prolonged by ranolazine in Scn1b-null mice ( Figure  6B and Supplemental Figure 3). These results suggest that Scn1b deletion enhanced the inhibitory effect of ranolazine on cardiac conduction. Similar to the effects observed at the single myocyte level (Figures 4 and  5) that the loss of Scn1b enhanced TB and UDB of I Na by ranolazine, loss of Na V β1/β1B in Scn1b-null mice promoted ranolazine block, manifesting as P wave, PR, and QRS interval prolongation.
We observed that 30 mg/kg lidocaine administration increased P wave duration in WT and Scn1b-null mice ( Figure 6, A and C). In addition, lidocaine induced prolongation of RR, QT, and ST intervals in WT but not Scn1b-null mice ( Figure 6C and Supplemental Figure 3). In contrast, lidocaine increased PR and QRS intervals in Scn1b-null mice ( Figure 6C). Lidocaine injection, therefore, resulted in distinct functional effects in the 2 genotypes. We conducted control experiments in which we measured ECGs before and after injection of PBS solution. Comparison of baseline and post-PBS data showed that ECG parameters remained constant (Supplemental Figure 4).
SCN1B is differentially expressed in human atria and ventricles. The observation that loss of Scn1b alters the ability of class Ib antiarrhythmics to block Na V channels in the mouse heart suggests that the differential expression of SCN1B might play an important role in regulating antiarrhythmic drug responses in humans.

Figure 4. I Na gating is similar in Scn1b-null and WT mouse LV myocytes. (A) Representative recordings of I Na in WT and
Scn1b-null mouse LV myocytes revealed similar kinetics of activation and inactivation. However, the average peak current density was slightly (~28%) higher in Scn1b-null compared with WT. (B) Loss of Na V β1 in Scn1b-null mouse LV myocytes did not affect the voltage dependences of I Na activation or steady-state inactivation. (C) Loss of Na V β1 in Scn1b-null mouse LV myocytes also did not affect the time course of I Na recovery from inactivation. Each data set represents mean ± SEM values from 6-9 cells. To begin to explore this hypothesis, we examined mRNA expression levels of the genes SCN1B, SCN2B, SCN3B, and SCN4B encoding Na V β subunits in human heart tissue in a recently published RNA-Seq library (35). These analyses revealed that in the human heart, SCN1B is the most abundant of the Na V β subunits at the transcript level and that SCN1B transcript expression is much higher in atria than in ventricles (Figure 7A). In contrast, the expression levels of the SCN2B and SCN4B transcripts are higher in the ventricles than in the atria ( Figure 7A). To validate the RNA-Seq findings and to determine whether both SCN1B splice variants, SCN1BA (Na V β1) and SCN1BB (Na V β1B), are differentially expressed in human atria and ventricles, we performed quantitative RT-PCR analyses on the same tissue samples used in the RNA-Seq analyses. These experiments revealed that the relative expression levels of the 2 SCN1B splice variants were significantly higher in the atria compared with the ventricles ( Figure 7B). Additional analyses revealed that, although expression of SCN1BA was approximately 100-fold higher than SCN1BB, the expression levels of the 2 (SCN1BA and SCN1BB) splice variants were similar in human right and left atria and in the LV and right ventricle (RV) (Supplemental Figure 5).

Discussion
Although class Ib antiarrhythmics have considerable therapeutic potential, they are not broadly prescribed because of proarrhythmic risks in some patients and ineffectiveness in others (36,37). Patient or disease variability in class Ib drug response suggests that there are external factors that modulate drug interactions with the channel (37). In this study, we investigated the role of Na V channel accessory subunits Na V β1 and Na V β3 in regulating class Ib antiarrhythmic-mediated effects on Na V 1.5 channels. We demonstrated that at a molecular level, Na V β1 or Na V β3 subunit coexpression differentially altered the effects of lidocaine and ranolazine on the Na V 1.5 DIII-VSD. Na V β1 enhanced lidocaine but inhibited ranolazine modulation of the DIII-VSD. Conversely, Na V β3 eliminated lidocaine modulation but increased the effect of ranolazine on the DIII-VSD. Differential molecular interactions between Na V 1.5 DIII-VSD and class Ib antiarrhythmic drugs caused by Na V β1 subunit expression in a heterologous system translated to distinct drug blockade of Na V channels in WT versus Scn1b cardiac-specific null mouse cardiomyocytes. We further demonstrated differential effects of lidocaine and ranolazine on the ECG phenotypes of WT and Scn1b-null mice.
Na V β1 and Na V β3 subunits alter Na V 1.5 channel pharmacology via the DIII-VSD. The DIII-VSD plays an important role in regulating Na V 1.5 channel gating. It is involved in both activation and inactivation of Na V channels (38,39). Recent studies showed a correlation between DIII-VSD deactivation and the slow component of recovery from inactivation, which suggests that an activated form of the DIII-VSD stabilizes inactivation (39). Given that class Ib antiarrhythmics promote DIII-VSD activation, they may subsequently promote inactivation to induce greater levels of UDB. We recently demonstrated that DIII-VSD activation determines Na V 1.5 channel blockade by mexiletine and affects the sensitivity of LQT3 variants to this drug (18), demonstrating a clear connection between the conformation of the DIII-VSD and class Ib drug potency.
We previously demonstrated that Na V β3 directly modulates the DIII-VSD, whereas Na V β1 does not (14). Recent channel structures suggest that Na V β1 associates with Na V 1.7 and Na V 1.4 through the DIII-VSD, an interaction that is not conserved in Na V 1.5 (10,12,13). In light of these structural and functional data, it is plausible that Na V β3 interacts with Na V 1.5 through similar sites as illustrated in the Na V 1.4/ Na V 1.7-β1 complex, while Na V β1 occupies a different site. Two distinct interaction mechanisms will result in the differential modulation of the DIII-VSD by Na V β1 and Na V β3. Here, we showed that coexpression of the Na V β1 or Na V β3 subunit differentially modulated the ability of class Ib antiarrhythmics to stabilize the DIII-VSD in the activated conformation, providing further evidence that in contrast to other Na V channel α subunits, Na V β1 and Na V β3 have distinct interactions with Na V 1.5. The effect of lidocaine on the DIII-VSD has been previously shown (14) to regulate UDB, a critical feature of class Ib drugs, which renders them most potent when myocytes are being excited repeatedly during an arrhythmic event. We observed that ranolazine and lidocaine similarly affected Na V 1.5 α subunit-encoded channels expressed in the absence example traces (bottom) for UDB of I Na by ranolazine. In contrast to lidocaine, the EC 50 for UDB by ranolazine was higher in WT compared with Scn1b-null, suggesting Na V β1 reduces the effects of ranolazine. (G) Frequency-dependent UDB block of I Na by 10 μM lidocaine in WT and Scn1b-null LV myocytes. UDB was assessed by measuring I Na evoked by repetitive depolarizing pulses at 5 Hz (25 ms, 40 pulses), 10 Hz (25 ms, 40 pulses), and 2 Hz (400 ms, 8 pulses). Normalized currents indicate I Na(last-pulse) /I Na(first-pulse) . (H) Frequency-dependent UDB block of I Na by 10 μM ranolazine in WT and Scn1b-null LV myocytes. Each data set represents mean ± SEM of data from 3-5 cells. Unpaired 2-tailed Student's t test. *P < 0.05; **P < 0.01; ***P < 0.001. JCI Insight 2021;6(15):e143092 https://doi.org/10.1172/jci.insight.143092 of the Na V β subunits. However, with the Na V β1 subunit present, the lidocaine effect on the DIII-VSD was enhanced, whereas the ranolazine effect was blunted. Conversely, Na V β3 enhanced ranolazine-induced DIII-VSD stabilization while inhibiting the effect of lidocaine. The differential regulation of the DIII-VSD resulted in distinct effects on the potencies of lidocaine and ranolazine depending on which Na V β was present. Thus, despite the similarity of the chemical structures (Figure 1), the therapeutic responses to lidocaine and ranolazine were differentially modified by the coexpression of the Na V β1 or the Na V β3 subunit.
Na V β1/β1B modulates class Ib antiarrhythmic responses from molecular to the whole-heart level. In Xenopus oocytes, with VCF and cut-open voltage clamp recordings, we demonstrated that Na V β1 coexpression enhanced lidocaine's but inhibited ranolazine's effect on the DIII-VSD, which resulted in an increased lidocaine but decreased ranolazine block. In addition, we observed increased UDB and late I Na block by lidocaine but opposite effects with ranolazine in WT compared with Scn1b-null mouse LV myocytes. This cellular difference further led to distinct phenotypes of the in vivo ECG recordings in response to lidocaine and ranolazine injections. For example, in response to ranolazine injection, the P wave duration and PR interval were prolonged in the Scn1b-null mice but not in the WT mice. This difference in ECG parameters reflects the cellular phenotype of enhanced ranolazine block of I Na in Scn1b-null compared with WT LV myocytes. However, not all the ECG changes can be explained by the differences observed in myocyte I Na recordings of atrial and ventricular myocytes, and they may be caused by Na V β1-mediated effects on other regions of the myocardium, such as the sinoatrial and atrioventricular nodes (40). Alternatively, Na V β1 may affect drug interactions with other ion channels that regulate cardiac excitation, as discussed further below.
Differential expression of Na V β1/β1B in human atria and ventricles and chamber-specific drug responses. Ranolazine was proposed as a candidate for atrial specific therapy for AF (24,41). Studies in the canine heart showed that atrial and ventricular cardiomyocytes have distinct responses to ranolazine (24). In the atria, ranolazine prolongs the action potential duration measured at 90% repolarization (APD 90 ) and effective refractory period (24). In contrast, in the ventricle, ranolazine shortens the APD 90 . Here, we demonstrated higher SCN1Ba (Na V β1) and SCN1Bb (Na V β1B) subunit mRNA expression levels in human atria compared with ventricles. While we showed that Na V β1 coexpression attenuated ranolazine block of I Na in both Xenopus oocytes and mouse LV myocytes, higher levels of Na V β1 in human atria may result in similar attenuation of the effects of ranolazine, thus contributing to decreased ranolazine blockade compared with ventricles. Notably, ranolazine is also a blocker of KCNH2-encoded (HERG-encoded) repolarizing I Kr channels (22). If ranolazine blockade of I Na is reduced in atria, modulation of I Kr may dominate, resulting in prolongation of APD specifically in atria. Thus, the heterogeneous expression of Na V β1 may play a role in the chamber-specific ranolazine response.
Na V β subunit modulation of class Ib drug effects may underlie disease-specific drug responses. A search of NCBI's Gene Expression Omnibus (GEO) profile database (42) revealed that SCN1B is upregulated in ischemic cardiomyopathies in human heart [GEO GDS651 and GDS1362 (43)] and mouse heart failure models [GEO GDS411, GDS427 (44), and GDS3660 (45)] (Supplemental Figure 6). These data suggest that the expression of Na V β subunits can be altered in disease remodeling of the cardiac tissue. Since late I Na was found to be increased in the failing heart, class Ib drugs have become potential therapeutic approaches in targeting heart failure-related arrhythmias (46). As we have demonstrated that Na V β can differentially modulate class Ib effects, upregulation of Na V β1 in failing tissue may alter the patient response to Na V channel-targeting antiarrhythmic therapies.
Differential effects of Na V β subunits on lidocaine and ranolazine interactions reflect distinct molecular drug-pore interactions. Previous work postulated that both lidocaine and ranolazine bind to common residues in the Na V 1.5 channel pore (20,31). However, we observed that lidocaine modulated both the DIII-and DIV-VSDs of Na V 1.5, whereas ranolazine only affected the DIII-VSD when Na V 1.5 was expressed without Figure 6. ECG recordings from WT and Scn1b-null mice before and after ranolazine or lidocaine injections. (A) Representative ECG recordings obtained from WT and Scn1b-null mice at baseline, postranolazine, and postlidocaine are presented. The postranolazine and postlidocaine data were recorded 10 minutes after the i.p. injections of ranolazine or lidocaine. P wave durations, PR, QRS, and QT intervals were measured as indicated in the insets. (B) Comparison of ECG parameters measured in WT (left panel) and Scn1b-null (right panel) mice at baseline and 10 minutes after i.p. injections of ranolazine injection. Ranolazine markedly prolonged the P wave duration and the PR interval in Scn1b-null but not in WT mice. (C) Comparison of ECG parameters measured in WT (left panel) and Scn1b-null (right panel) mice at baseline and 10 minutes after i.p. injections of lidocaine. Lidocaine markedly prolonged the RR interval, P wave duration, and QT interval in WT mice. In Scn1b-null mice, lidocaine also prolonged the P wave duration and resulted in marked prolongation of the PR and QRS intervals. Each data set represents data from 4-7 mice. The ECG parameters and statistical comparisons are shown in Supplemental Table 1. JCI Insight 2021;6(15):e143092 https://doi.org/10.1172/jci.insight.143092 Na V β subunits, suggesting that these compounds may present in different orientations in the channel pore. Recent molecular dynamics simulations shed light on the detailed binding conformations of lidocaine and ranolazine within the Na V 1.5 channel (47). Interestingly, these studies revealed that 2 lidocaine molecules can bind to the pore concurrently, one to the F1760 site and the other to the central pore (47). In contrast, ranolazine binds to the F1760 residue and possesses a more flexible linear structure, allowing it to interact with a larger area ranging from the fenestration to the selectivity filter (47). Moreover, ranolazine has a pK a of 7.2 (48) and lidocaine has a pK a of 7.9 (49). At physiological pH, therefore, a higher percentage of ranolazine molecules are expected to be uncharged compared with lidocaine. This difference will determine the relative percentages of the drug molecules entering in the hydrophobic pathway through the fenestration versus the hydrophilic pathway via the intracellular gate. The presence of Na V β1 or Na V β3 modulates DIII-VSD and DIV-VSD dynamics, which can allosterically affect the conformation of the pore and fenestrations. We demonstrated that Na V β1/β3's modulation of ranolazine block was not entirely dependent on the main local anesthetic binding site, F1760, suggesting the changes in the DIII-VSD conformation due to Na V β modulation are essential for determining drug accessibility to the pore, independent of binding. In contrast, eliminating the F1760 binding site abolished the effects of Na V β1/β3 coexpression on lidocaine block, further suggesting that lidocaine acts through mechanisms distinct from those of ranolazine. Because lidocaine and ranolazine have different stoichiometries, orientations within the pore, and entrance pathways, it is plausible that changing channel VSD and pore conformations in the presence of Na V β subunits can result in opposite effects on DIII-VSD interactions with these drugs.
Na V β subunit modulation of antiarrhythmic drug outcome beyond Na V channels. Aside from Na V channels, Na V β subunits have been shown to modulate expression and gating of various potassium channels, including the voltage-gated K V 4.3 and the inward-rectifying K ir 2.1 channels, resulting in modifications of 2 essential cardiac currents, the fast transient outward K + current (I to,f ) and the inward rectifier current (I K1 -) (50)(51)(52). Although unexplored to date, it also seems highly likely that the presence of Na V β subunits can affect K V and K ir channel pharmacology as well. Here, we showed that although single cardiac-specific Scn1b-null LV myocytes displayed attenuated inhibition of I Na by lidocaine compared with WT cells, enhanced PR and QRS interval prolongation was observed in Scn1b-null animals with lidocaine, suggesting contributions from other cardiac ionic currents.
Limitations. To assess the drug effects on VSD dynamics, we conducted the VCF experiments in Xenopus oocytes. Because this is a heterologous expression system and the experiments were performed at 19°C, extrapolating the results to mammalian physiology can be difficult. However, we observed consistent drug-mediated modulatory effects in oocytes and in mouse LV myocytes, observations that support the hypothesis that the molecular mechanisms we identified with VCF are operative in mammalian systems.  (44). The SCN1B transcript was the most abundant of the Na V β subunits expressed in human atria and ventricles. In addition, SCN1B expression was approximately 3-fold higher in human atria compared with ventricles, whereas both SCN2B and SCN4B were approximately 2-fold higher in human ventricles than atria. (B) The differential expression of SCN1B in human atria and ventricles was confirmed by quantitative PCR (qPCR) analyses of the same paired human atrial and ventricular tissue samples analyzed by RNA-Seq. In addition, qPCR analyses using primers that distinguish the 2 SCN1B variants, SCN1Ba and SCN1Bb, revealed that the relative expression levels of both SCN1Ba and SCN1Bb transcripts were higher in the atria than the ventricles. Paired 2-tailed Student's t test was used to test significance. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.
JCI Insight 2021;6(15):e143092 https://doi.org/10.1172/jci.insight.143092 We were only able to show that the expression of SCN1B was enriched in the atria compared with ventricles in the human heart at the transcript level. Although we attempted to examine protein expression levels directly, none of the available anti-Na V β1 antibodies detect Na V β1 proteins in mouse or human cardiac tissues. Since it is currently not possible to perform gene editing or to use siRNA-mediated knockdown strategies in native human myocytes, we were not able to explore the functional effects of Na V β1 on class Ib drugs in human myocytes directly.
Conclusions. In summary, we have demonstrated roles for noncovalently linked Na V β subunits in regulating antiarrhythmic drug effects from molecular interactions to whole-heart phenotypes. Our results elucidated the differential regulation of Na V 1.5 channels by 2 class Ib agents, lidocaine and ranolazine, by Na V β1 and Na V β3 subunits. The unique expression profile of Na V β subunits in the human heart suggests chamber-dependent responses to these 2 compounds. Our findings provide crucial insights into strategies for improving the clinical outcomes of patients treated with class Ib agents for different forms of arrhythmias. Moreover, Na V β1 expression is upregulated in heart failure, and it remains unexplored whether Na V β subunit expression is affected in other heart pathologies. This knowledge will be highly valuable in establishing disease-specific approaches to personalize arrhythmia treatment with lidocaine and ranolazine because known changes in β subunit expression will have predictable effects on therapeutic outcomes.

Methods
Experimental animals. Adult (8-15 weeks old) male and female WT and cardiac-specific Scn1b-null C57BL/6J mice were used in the experiments here. Cardiac-specific Scn1b-null mice were generated by crossing Scn1b fl mice (27) with B6.FVBTg(Myh6-cre)2182Mds/J mice (The Jackson Laboratory), which express Cre recombinase driven by the α-myosin heavy chain promoter. Mice were genotyped by PCR analyses of genomic (tail) DNA using primers targeting sequences external to the loxP sites, as well as primers targeting Cre recombinase, as previously described (27). Because of the breeding strategy required to generate cardiac-specific Scn1b-null C57BL/6J mice, WT littermates were not generated. The WT mice used in the experiments presented here, therefore, were not littermate controls; rather they were WT C57BL/6J mice from our colony. Additional control experiments, however, were conducted on Scn1b fl and B6.FVBTg(Myh6-cre)2182Mds/J mice, which were determined to be indistinguishable electrophysiologically from WT C57BL/6J animals. Xenopus oocyte harvests were performed as described previously (53).
Cut-open VCF. VCF experiments were conducted using 4 previously developed Na V 1.5 channel constructs (DI: V215C, DII: S805C, DIII: M1296C, DIV: S1618C) (53). Capped mRNAs were synthesized with the mMESSAGE mMACHINE T7 transcription kit (Life Technologies, Thermo Fisher Scientific) from the linearized pMAX vectors. VCF construct mRNA was injected alone or coinjected with SCN1B or SCN3B mRNA into Xenopus oocytes as previously described (14). VCF experiments were performed 4-6 days after injection. The recording setup and labeling protocol used were described previously (14,(53)(54)(55). Lidocaine hydrochloride and ranolazine dihydrochloride were dissolved in extracellular recording solution and then further diluted to 10 mmol/L and 4 mmol/L, respectively. Both drugs were manually perfused into the extracellular solution chamber in the cut-open voltage clamp setup. Fluorescence signals and currents were analyzed as previously described (18). V 1/2 values reported were quantified from the Boltzmann function fit, y = 1/(1 + exp [V -V 1/2 ]/k). Because the DIII F-V curve, especially under drug treatment conditions, did not saturate at the lowest voltage recorded (-160 mV), the fit was performed by fixing 0 at -200 mV. Because of the lack of saturation at the most negative potentials measured, the estimated V 1/2 for DIII F-V is likely to be higher than the actual V 1/2 value. ECG recordings. Surface ECG recordings were obtained as previously described from mice anesthetized by i.p. injection of avertin (0.25 mg/kg; MilliporeSigma) (56). Baseline ECGs were recorded, and animals were weighed. For injections, drugs were dissolved in 250 μL PBS. Lidocaine or ranolazine was then injected at a dosage of 30 mg/kg or 20 mg/kg, respectively. Different animals were used for lidocaine or ranolazine injections. Between recordings, mice were kept on a heating pad maintained at 37°C ± 0.5°C. Postinjection ECGs were recorded at 5 minutes, 10 minutes, 15 minutes, 20 minutes, and 30 minutes. Peak responses were observed at 10 minutes, which was subsequently selected as the time point for ECG analysis.
RR, PR, and QT intervals, as well as P and QRS durations, were measured and compiled using Clampfit 10.3 (Molecular Devices) and GraphPad Prism. Note that QT intervals shown in figures were not corrected because several recent studies have shown that QT intervals in anesthetized mice do not vary with heart rate (57, 58). Similar differences were revealed, however, when corrected QT intervals were compared (Supplemental Table 1).
JCI Insight 2021;6(15):e143092 https://doi.org/10.1172/jci.insight.143092 Isolation of adult mouse cardiomyocytes. Myocytes were isolated from adult (8 to 12 weeks old) WT or Scn1b-null mice as previously described (59). Briefly, hearts were isolated and perfused retrogradely through the aorta with a Ca 2+ -free Earle's balanced salt solution containing 0.8 mg/mL type II collagenase (Worthington). After perfusion, the LV free wall was dissected and minced. The tissue pieces were then triturated to provide individual LV myocytes. Dispersed cells were then filtered and resuspended in Medium199 (Gibco, Thermo Fisher Scientific), plated on laminin-coated (MilliporeSigma) glass coverslips, and maintained in a 95% air/5% CO 2 incubator at 37°C.
Electrophysiological data were acquired at 10-20 KHz, and signals were low-pass-filtered at 5 kHz before digitization and storage. After the formation of a giga-seal (>1 GΩ) and establishment of the whole-cell configuration, brief (10 ms) ± 10 mV voltage steps from a holding potential (HP) of -70 mV were presented to allow measurements of whole-cell membrane capacitances (C m ), input resistances (R in ), and series resistances (R s ). In each cell, C m and R s were compensated electronically by approximately 85%; voltage errors resulting from uncompensated R s were less than 2 mV and were not corrected. Leak currents were always less than 50 pA and were not corrected. Whole-cell I Na were evoked in response to 40 ms voltage steps to potentials between -60 and +40 mV from an HP of -100 mV in 10 mV increments at 15-second intervals.
Electrophysiological data were compiled and analyzed using Clampfit 10.3 (Molecular Devices) and GraphPad Prism.
Quantitative reverse-transcription PCR. Total RNA (2 μg) isolated from individual matched (n = 6) human right atria, left atria, RV, and LV tissue samples was reverse-transcribed into cDNA with a high-capacity cDNA kit. Transcript analysis was conducted with SYBR green using a 7900HT Fast Real-Time PCR system (all from Applied Biosystems, Thermo Fisher Scientific). Data were analyzed using the Ct relative quantification method using the GAPDH and hypoxanthine guanine phosphoribosyl transferase I (HPRT) genes as endogenous controls.
Statistics. Results are presented as mean ± SEM. The number of animals and the number of cells used in each experiment are provided in the figure legends. Comparisons of differences between WT and Scn1bnull cells/animals under control conditions and before and after drug treatments were performed using a paired 2-tailed Student's t test (Microsoft Excel). In comparisons of more than 2 groups, 1-way ANOVA was used followed by multiple comparisons. The P values shown were corrected for multiple hypothesis testing using Dunnett's correction method. P values less than 0.05 were considered significant.
Study approval. All animals were handled in accordance with the NIH Guide for the Care and Use of Laboratory Animals, and all experimental protocols were approved by the Washington University IACUC.

Author contributions
WZ contributed to designing studies, conducting experiments, acquiring data, analyzing data, and writing the manuscript. WW, PA, and RLM contributed to conducting experiments and acquiring data. LLI contributed to providing experimental animals and editing the manuscript. JMN and JRS contributed to designing studies and editing the manuscript.