Delayed Ventricular Repolarization and Sodium Channel Current Modification in a Mouse Model of Rett Syndrome

Rett syndrome (RTT) is a severe developmental disorder that is strongly linked to mutations in the MECP2 gene. RTT has been associated with sudden unexplained death and ECG QT interval prolongation. There are mixed reports regarding QT prolongation in mouse models of RTT, with some evidence that loss of Mecp2 function enhances cardiac late Na current, INa,Late. The present study was undertaken in order to investigate both ECG and ventricular AP characteristics in the Mecp2Null/Y male murine RTT model and to interrogate both fast INa and INa,Late in myocytes from the model. ECG recordings from 8–10-week-old Mecp2Null/Y male mice revealed prolongation of the QT and rate corrected QT (QTc) intervals and QRS widening compared to wild-type (WT) controls. Action potentials (APs) from Mecp2Null/Y myocytes exhibited longer APD75 and APD90 values, increased triangulation and instability. INa,Late was also significantly larger in Mecp2Null/Y than WT myocytes and was insensitive to the Nav1.8 inhibitor A-803467. Selective recordings of fast INa revealed a decrease in peak current amplitude without significant voltage shifts in activation or inactivation V0.5. Fast INa ‘window current’ was reduced in RTT myocytes; small but significant alterations of inactivation and reactivation time-courses were detected. Effects of two INa,Late inhibitors, ranolazine and GS-6615 (eleclazine), were investigated. Treatment with 30 µM ranolazine produced similar levels of inhibition of INa,Late in WT and Mecp2Null/Y myocytes, but produced ventricular AP prolongation not abbreviation. In contrast, 10 µM GS-6615 both inhibited INa,Late and shortened ventricular AP duration. The observed changes in INa and INa,Late can account for the corresponding ECG changes in this RTT model. GS-6615 merits further investigation as a potential treatment for QT prolongation in RTT.


Introduction
Rett syndrome (RTT; OMIM #312750) is a severe X-chromosome-linked developmental disorder, first identified in the 1960s [1,2]. It is characterized by arrested development (typically between 6-18 months of age), cognitive and motor skill deficits, microencephaly and seizures [3][4][5]. RTT patients also exhibit autonomic dysfunction, with altered heart rate control and respiratory difficulties (including breath holding and hyperventilation [3,5]). The sporadic "typical" or "classical" form of RTT is strongly associated with mutations to the MECP2 gene, which encodes the X-linked transcriptional regulator Methyl-Cp-binding protein 2 [4,[6][7][8][9]. 95-97% of patients with typical RTT exhibit MECP2 mutations [3,10]. Patients with RTT are nearly all female; males exhibit more severe cardiac and respiratory abnormalities and the majority die within a year of birth [4]. However, whilst X-linked dominance has long been considered to be characteristic of RTT, MECP2 mutations have infrequently also been reported in males that exhibit a classical RTT phenotype; such cases are associated with somatic mosaicism or possession of an extra X chromosome [11].
representative ECG traces from WT (upper panel) and Mecp2 Null/Y (lower panel) mice, with the QT interval for a single ECG complex from each strain highlighted in Figure 1(Aii). Table 1 summarizes mean ECG data from 12 WT and 12 Mecp2 Null/Y mice. Heart rate and R-R interval values were similar between WT and mutant strains. However, there was a small but statistically significant increase in QRS interval duration in the RTT model strain (12.2 ± 0.5 ms vs 10.8 ± 0.2 ms in WT; p < 0.01). Uncorrected QT interval duration was also significantly greater in RTT (56.9 ± 1.8 ms) than in control (51.9 ± 0.9 ms) mice (p < 0.05). We applied two different methods of rate correction to the QT interval ( [18,25,26]; see Materials and Methods), which resulted in different absolute QT c values, each of which demonstrated that the mean QT c interval of Mecp2 Null/Y mice was significantly greater (by 5.3-5.6 ms) than that of WT controls (Table 1). Respiratory rate was also monitored during ECG measurement and observed to be significantly slower in RTT than control animals ( Table 1). Our ECG measurements are consistent with the original findings of McCauley et al. [18] that Mecp2 Null/Y mice develop prolonged QT intervals. AP measurements were made at 1 Hz from isolated ventricular myocytes, as described in the Methods. Figure 1(Bi,Bii) show representative APs from WT and Mecp2 Null/Y myocytes respectively, whilst Table 2 summarizes mean AP parameters from 27 myocytes from 9 WT animals and 29 myocytes from 13 Mecp2 Null/Y mice. Resting membrane potential (RMP) was~2.7 mV less negative in RTT myocytes (p < 0.05). The threshold amplitude of the fixed duration (3 ms) depolarizing current stimulus required to elicit APs was significantly reduced in myocytes from RTT animals. However, whilst there was a tendency for both AP overshoot potential and overall AP amplitude to be slightly lower for APs from RTT myocytes these differences did not attain statistical significance. Similarly, mean AP upstroke velocity showed a non-statistically significant trend to be higher in WT than RTT myocytes. AP repolarization parameters were quantified at 10%, 25%, 50%, 75% and 90% of complete repolarization and significant differences were observed for both APD 75 and APD 90 (Table 2). Mean APD 90 was 166.0 ± 10.8 ms and 115.6 ± 9.5 ms in RTT and WT myocytes respectively (p < 0.01). AP triangulation was measured as the APD 25 to APD 90 difference, which was also shown to be significantly different between WT (113.5 ± 9.4 ms) and RTT (163.5 ± 10.8 ms; p < 0.01) myocytes. The Poincaré plot in Figure 1C shows representative beat-to-beat variability (BVR) over 15 successive APs from individual WT and Mecp2 Null/Y myocytes. BVR was quantified using Equation (3) and mean values for WT and Mecp2 Null/Y mice are plotted in Figure 1D: BVR was significantly larger for Mecp2 Null/Y than WT myocytes, indicating increased instability of APD 90 in myocytes from the RTT model. Upper and lower panels of (Ai) show respectively ECG records from WT and Mecp2 Null/Y mice. The periods of high frequency noise in each trace represent breathing interference. In (Aii) an individual expanded ECG cycle is shown for each of WT (upper) and Mecp2 Null/Y (lower) conditions, with labelling of different parts of the ECG complex and the QT intervals shown. (B) shows representative recordings of APs (elicited at 1 Hz) from WT (Bi) and Mecp2 Null/Y (Bii) isolated ventricular myocytes. (C) Poincaré plots of APD 90 from 15 consecutive APs illustrating increased beat-to-beat variability of repolarization (BVR) in Mecp2 Null/Y compared to WT myocytes. (D) Mean data showing BVR increase in Mecp2 Null/Y myocytes (n = 23 cells from 12 mice) compared to WT myocytes (n = 23 cells from 9 mice). BVR of APD 90 was calculated by using equation 3. * denotes p < 0.05, unpaired t test, unequal variances assumed. Unrestrained whole body plethysmography was used separately from ECG measurements to monitor occurrence of respiratory apnoea as described previously [27]. In measurements from 34 WT and 23 Mecp2 Null/Y animals, no significant difference was found in numbers of apnoea episodes (Figure 2A, apnoea count). The mean duration of apnoea episodes ( Figure 2B, apnoea length) was, however, significantly greater in Mecp2 Null/Y mice. For 11 mice (4 WT and 7 Mecp2 Null/Y ) for which ventricular myocytes were isolated following body plethysmography, we plotted mean APD 90 values against apnoea length ( Figure 2C). However, no significant correlation between these two parameters was found (R = −0.07; p = 0.84). Mecp2 Null/Y animals. These did not differ significantly. An apnoea was counted when the expiration time was longer than four times the average of the expiration time of minute running average. Apnoea was counted for 20 min. (B) shows mean apnoea length (duration of each episode) for the same 34 WT and 23 Mecp2 Null/Y animals. ** represents p < 0.01; unpaired t-test. (C) shows a plot of mean ventricular AP duration from 11 mice (4 WT and 7 Mecp2 Null/Y ) against apnoea length observed for the same animals. There was no significant correlation between the two parameters (R = −0.07 and p = 0.84).
2.2. Late Sodium Current, I Na,Late , Enhancement in Mecp2 Null/Y Mice I Na,Late was studied using the recording solutions described in the Methods and a voltage clamp protocol comprised of a 1 s depolarization from −120 to −20 mV (applied at a frequency of 0.1 Hz). With this protocol, the initial rapid I Na component gave way to a small, sustained current component during the applied voltage command. Figure 3(Ai,Aii) respectively show representative records of the Na-sensitive (NMDG-subtraction) current expanded to focus on the sustained component. It is evident that the sustained current, representing I Na,Late is larger for the Mecp2 Null/Y myocyte than the WT example. Three complementary approaches were used to quantify I Na,Late , each of which excluded the large initial I Na . First, we measured I Na density (pA/pF) at 300 ms into the voltage test command; second, similar measurements were made at a later time-point of 600 ms. The third analysis method was to evaluate the current integral (pC/pF) between 350 and 800 ms into the applied voltage command. Mean data for each of these approaches are plotted in Figure 3B (for 41 cells from 20 WT mice and for 28 cells from 17 Mecp2 Null/Y mice). Each of the three methods used showed a similar result: I Na,Late was significantly augmented in ventricular myocytes from Mecp2 Null/Y mice. and Mecp2 Null/Y I Na,Late . Three different measures were taken: I Na,Late density at 300 ms following the start of the applied voltage command; I Na,Late density at 600 ms following the start of the applied voltage command; I Na,Late integral between 350 and 800 ms following the start of the applied voltage command. Data were obtained from 41 myocytes from 20 WT mice and 28 cells from 17 Mecp2 Null/Y mice. ** denotes p < 0.01 from comparison with unpaired t test assuming unequal variances. (C) shows records of sodium-sensitive current elicited by depolarization from −120 mV to −20 mV (protocol shown under currents), expanded to focus on the persistent, late current component I Na,Late to illustrate the effect of the Na v 1. Although cardiac I Na is largely attributable to channels for which the major poreforming subunit is encoded by SCN5A [28] (scn5a in mice [29]), the neuronal Na v 1.8 isoform (encoded by SCN10A) may also be present and potentially contribute to I Na,Late [30,31]. A-803467 has been shown to be potent and highly selective for Na v 1.8 over Na v 1.5 over a wide range of concentrations (30-1000 nM; [31]). Therefore, we evaluated the effects of 100 nM A-803467 on I Na,Late under our recording conditions. Figure 3(Ci,Cii) show representative traces in the absence and presence of A-803467 for WT and Mecp2 Null/Y myocytes respectively. Traces in the absence and presence of the drug were closely overlain in each case. Figure 3(Di,Dii) respectively show mean data for WT (eight cells from six mice) and Mecp2 Null/Y (six cells from five mice) myocytes. Irrespective of the analysis method used to quantify I Na,Late there was no significant effect of A-803467. As inhibition of Na v 1.8 is near complete at 30 nM [31], this observation precludes Na v 1.8 from contributing significantly to I Na,Late under the conditions of this study.

Alterations to Fast I Na in Mecp2 Null/Y Mice
In this study and the earlier investigation from McCauley and colleagues [18], QRS duration differed significantly between WT and Mecp2 Null/Y mice. This observation and the trends towards changes in ventricular AP overshoot/upstroke raise the possibility that differences in fast I Na exist between WT and Mecp2 Null/Y mice. McCauley et al. suggested there is little difference in I Na between the strains, but that was based on measurements from the fast current component under conditions optimised for I Na,Late [18]. The speed and amplitude of I Na make accurate voltage clamp of the current difficult with high [Na + ] o and cardiac I Na measurement is facilitated by a combination of the use of low [Na + ] o to reduce I Na amplitude and experimentation at room temperature (e.g., [32][33][34]). Accordingly, measurements of I Na were made at room temperature with pipette and external solutions giving rise to symmetrical 5 mM [Na + ] (see Methods). Figure 4(Ai,Aii) show families of I Na elicited from WT and Mecp2 Null/Y myocytes, by the protocol shown as insets to these panels. From a holding potential of −80 mV and following a 2 s pre-pulse to −140 mV a series of 250 ms command steps (increasing by 10 mV increments) between −80 and 0 mV were applied (pulse frequency of 0.2 Hz). As shown in Figure 4(Ai,Aii) there was a decrease in current amplitude in the Mecp2 Null/Y condition. Mean current voltage (I-V) relations for I Na from 11 WT myocytes (from 5 mice) and 9 Mecp2 Null/Y myocytes (from 5 mice) are shown in Figure 4B. The I-V plot for Mecp2 Null/Y I Na exhibited statistically significantly decreased I Na density between −50 and −30 mV. I Na conductance values were calculated at each potential for each cell (using Equation (4), Materials and Methods) and G/G max data were then plotted as shown in Figure 4C and fitted with a Boltzmann equation (Equation (5), Materials and Methods) to derive activation V 0.5 and k values. For WT (11 cells from 5 mice) I Na , the derived V 0.5 value was −47.4 ± 1.4 mV and k was 3.6 ± 0.2 mV; for Mecp2 Null/Y (9 cells from 5 mice) I Na V 0.5 was −45.4 ± 1.0 mV and k was 3.9 ± 0.5 mV; neither I Na activation parameter differed significantly between the two strains. Fast I Na inactivation time-course (τ inact ) was quantified through exponential fitting of the time-course of current decline of I Na elicited by between −40 and −20 mV (data from experiments shown in Figure 4). At −40 mV, but not −30 or −20 mV, the τ inact was significantly larger for Mecp2 Null/Y than WT I Na (WT and Mecp2 Null/Y τ inact at −40 mV of 2.6 ± 0.2 ms and 3.1 ± 0.1 ms; p < 0.05; n = 11 myocytes from 5 mice and n = 9 myocytes from 5 mice for WT and Mecp2 Null/Y , respectively). Voltage dependence of I Na inactivation was evaluated using a paired pulse protocol (shown schematically as insets to Figure 5(Ai,Aii). From a holding potential of −80 mV, 1.5 s conditioning pulses were applied to a range of potentials between −150 and −50 mV in 10 mV increments. Each conditioning pulse was followed by a 40 ms duration test command to −40 mV. The protocol frequency was 0.1 Hz. Figure 5(Ai,Aii) respectively show I Na elicited from WT and Mecp2 Null/Y myocytes during the −40 mV test command, following conditioning pulses to the membrane potential values shown. For each cell studied, the currents during the −40 mV step following the different conditioning potential were normalized to the maximal current observed during the protocol (I/I max ) and then plotted against conditioning voltage as shown in Figure 5B. Most I/I max values were similar between WT and Mecp2 Null/Y myocytes, although those at −90 and −80 mV were larger in myocytes from the RTT model. Inactivation V 0.5 and k values were obtained from Boltzmann fits to the data with equation 6. For 9 WT cells (from 5 mice) mean V 0.5 and k values were respectively −83.1 ± 1.7 mV and 6.5 ± 0.3 mV. For 8 Mecp2 Null/Y myocytes (from 4 mice), mean V 0.5 and k values were respectively −79.6 ± 0.7 mV (not significantly different from WT) and 5.1 ± 0.3 mV (p < 0.01 vs WT). The mean activation and inactivation V 0.5 and k values were then used to calculate the fast I Na "window current" (calculated at 2 mV as the product of activation and inactivation variables, maximal conductance and driving force), which is plotted in Figure 5C. The calculated I Na window was reduced in the Mecp2 Null/Y compared to the WT condition. Consequently, the increased I Na,Late observed in the RTT model cannot be attributed to an increased steady state "window" for fast I Na . Nor can the modest depolarization of RMP shown in Table 2 be attributed to increased steady state "window" current. I Na calculated at 2 mV intervals, utilizing the product of the mean activation and inactivation variables, G max and driving force at each test potential (see Methods). Window current was decreased for Mecp2 Null/Y compared to WT I Na . The negative sign was removed in order to visualize the window in the positive direction. (D) Recovery from inactivation (or 'reactivation') of I Na obtained using paired pulse protocol described in the Methods/Results text. I Na elicited by a second command at varying inter-pulse intervals was normalized to that elicited by the first pulse and plotted against inter-pulse interval. Mean data from 8 cells from 4 mice for each of WT and Mecp2 Null/Y are plotted. τ fast was 11.0 ± 0.7 ms for WT I Na and 8.6 ± 0.2 ms for Mecp2 Null/Y I Na (p < 0.05, unpaired t test, unequal variances assumed); τ Slow was 362.0 ± 53.4 ms for WT I Na and 311.5 ± 50.2 ms for Mecp2 Null/Y . The fraction of fast reactivation was 0.85 ± 0.02 for WT and 0.85 ± 0.03 for Mecp2 Null/Y . * p < 0.05, ** p < 0.01, 2-way repeated measures ANOVA with Bonferroni post-test.
Recovery of fast I Na from inactivation was determined using a paired pulse protocol in which a 1 s command from a holding potential of -120 mV was applied to −40 mV (P 1 ; to activate and then inactivate I Na ) was followed at varying time-intervals (0.1, 0.3, 1, 3.2, 10, 31.6, 100, 316.2, 1000 and 3162 ms) by a second 40 ms duration command (P 2 ) to −40 mV to monitor recovery of I Na amplitude from inactivation (protocol frequency 0.1 Hz). The I Na elicited by P 2 was expressed as a fraction of that elicited by P 1 and mean values were plotted against the inter-pulse interval. Recovery from inactivation time-course was quantified through biexponential fitting (Equation (7), Methods). Figure 5D shows mean recovery time-course data, plotted on a logarithmic time-scale (data from eight cells from four mice for each of WT and Mecp2 Null/Y conditions are plotted) [34]. The fast time constant of recovery from inactivation (τ fast ) was 11.0 ± 0.7 ms for WT I Na and 8.6 ± 0.2 ms for Mecp2 Null/Y I Na (p < 0.05), whilst the slow time constant of recovery from inactivation (τ Slow ) was 362.0 ± 53.4 ms for WT I Na and 311.5 ± 50.2 ms for Mecp2 Null/Y I Na (no significant difference). The proportion of recovery from inactivation fitted by τ fast did not differ significantly between the two strains (0.85 ± 0.02 for WT and 0.85 ± 0.03 for Mecp2 Null/Y ). Thus, fast I Na from Mecp2 Null/Y myocytes exhibited slightly faster recovery from inactivation than did WT I Na due to acceleration of the rapid component of I Na reactivation.

Effects of the I Na,Late Inhibitors Ranolazine and Eleclazine (GS-6615)
Ranolazine (Ranexa) was developed as an anti-anginal agent but has undergone extensive investigations for repurposing as an antiarrhythmic drug [35][36][37][38][39]. It is structurally related to lidocaine and is an effective inhibitor of cardiac I Na,Late [40,41]. Figure 6(Ai,Aii) respectively show exemplar traces of the effect of ranolazine (30 µM) on I Na,Late from WT and Mecp2 Null/Y myocytes: in each case the current was reduced following ranolazine exposure. Figure 6B shows the mean the effect of ranolazine evaluated using the three measures described in Figure 3, each plotted as percentage inhibition values (for nine WT myocytes from four mice and seven Mecp2 Null/Y myocytes from four mice). Irrespective of the measurement chosen, ranolazine reduced I Na,Late by a similar percentage in WT and RTT myocytes.
GS-6615 (also known as eleclazine) is a second generation selective I Na,Late inhibitor with an improved structure-activity relationship profile to ranolazine [42] and effectiveness against LQT3 Na channel mutations [43]. Figure 6(Di,Dii) respectively show exemplar traces of the effect of GS-6615 (10 µM) on I Na,Late from WT and Mecp2 Null/Y myocytes. Similar to ranolazine, in each case the current was reduced following GS-6615 exposure. Figure 6E shows the mean effect of GS-6615 evaluated using the three measures used for ranolazine in Figure 6B, each plotted as percentage inhibition values (for 16 WT myocytes from 8 mice and 8 Mecp2 Null/Y myocytes from 6 mice). Irrespective of the measurement chosen, GS-6615 reduced I Na,Late by a similar percentage in WT and RTT myocytes. Figure 6C,F display mean control I Na,Late integral values for WT and Mecp2 Null/Y myocytes together with the Mecp2 Null/Y myocyte integral from the same myocytes following treatment with ranolazine ( Figure 6C) and GS-6615 ( Figure 6F). Consistent with Figure 3, the Mecp2 Null/Y myocyte I Na,Late integral was significantly larger than that for WT myocytes; however, in the presence of ranolazine or GS-6615 this was no longer the case. Thus, application of 30 µM ranolazine or 10 µM GS-6615 to myocytes from the RTT model restored I Na,Late to levels not significantly different from those in WT myocytes. and Mecp2 Null/Y (Bii) myocytes. In each case, GS-6615 led to AP abbreviation, consistent with the compound's I Na,Late inhibitory action; however, ranolazine lengthened rather than shortened AP duration, as would have been predicted from I Na,Late inhibition. Figure 7C shows the mean percentage changes in APD 90 produced by each agent, with statistically similar effects on WT and Mecp2 Null/Y myocytes in each case, with APD 90 lengthening by ranolazine and abbreviation by GS-6615 (the plots show data from seven cells from four WT mice and six cells from three Mecp2 Null/Y mice for the ranolazine groups, and six cells from three WT mice and eight cells from five Mecp2 Null/Y mice for GS-6615 groups). Furthermore, in Mecp2 Null/Y myocytes GS-6615 reduced AP triangulation (from 156.0 ± 14.4 ms to 125.9±9.1 ms; p < 0.01). Mean BVR was also reduced by GS-6615, although this difference was not statistically significant (from 8.6 ± 2.8 ms to 5.4 ± 1.4 ms; p = 0.09). (C) shows mean data indicating the mean % change in AP duration at 90% repolarization (APD 90 ) for each of WT and Mecp2 Null/Y conditions. Plots show data from 7 cells from 4 WT mice and 6 cells from 3 Mecp2 Null/Y mice for the ranolazine groups, and 6 cells from 3 WT mice and 8 cells from 5 Mecp2 Null/Y mice for GS-6615 groups. There was no significant difference between the magnitude of response between WT and Mecp2 Null/Y myocytes for either drug: ranolazine prolonged APD 90 and GS-6615 abbreviated APD 90 .

Summary of Main Findings
The principal findings of this study are summarized in diagrammatic form in Figure 8A. ECG measurements from Mecp2 Null/Y animals showed statistically significant increases in QRS interval and in both uncorrected QT and rate-corrected QT c intervals, indicative of delayed ventricular repolarization in this model ( Figure 8A and Table 1). Repolarization delay was also demonstrated directly through AP measurements from WT and Mecp2 Null/Y ventricular myocytes. Time points up to and including APD 50 were not significantly different between WT and Mecp2 Null/Y myocytes, but both APD 75 and APD 90 were increased ( Table 2). In consequence, the difference between APD 25 and APD 90 , which provides an index of AP triangulation, was significantly increased in RTT myocytes (Table 2 and Figure 8A). Instability of APD 90 , measured as BVR, was also greater in Mecp2 Null/Y myocytes ( Figures 1D and 8A). I Na,Late was significantly increased in myocytes from the RTT model (Figures 3 and 8A). In contrast, peak fast I Na amplitude was significantly decreased in Mecp2 Null/Y myocytes without significant change to voltage-dependent activation of the current (Figures 4 and 8A). Comparison of voltage dependent inactivation of I Na from WT and RTT myocytes showed no significant difference in inactivation V 0.5 , but the slope of the inactivation relation was steeper (evidenced by a significant decrease in the slope factor, k; Figure 5B). Consistent with the combination of decreased I Na amplitude over some voltages and a steeper inactivation relation, simulated "window" current was smaller for the Mecp2 Null/Y than WT condition ( Figures 5C and 8A). Minor additional changes to I Na were a modest slowing of inactivation time course at −40 mV and a decrease in the fast time constant of recovery from inactivation (Section 2.3; Figure 5D). Our experiments have also shown that I Na,Late from Mecp2 Null/Y myocytes retains sensitivity to inhibition by ranolazine and GS-6615 (Figures 6 and 8A); this was associated with AP abbreviation for GS-6615 but, unexpectedly, with AP prolongation for ranolazine (Figures 7 and 8A). Figure 8B illustrates, in schematic form, the relationship between each of fast I Na and I Na,Late and the ventricular AP (shown with a high plateau phase as occurs in humans). Fast I Na flows during the AP upstroke, whilst I Na,Late provides sustained inward current during the AP plateau phase. This schematic illustration may aid contextualization of aspects of our findings discussed below. studied; the right-hand column shows the direction of alteration in the RTT model compared to WT controls. Upward and downward arrows represent increase and decrease, respectively. The horizontal arrow indicates no change (to I Na activation). For the two drugs studied "inhibition retained" highlights the observation that Mecp2 Null/Y I Na,Late retained the ability to be inhibited by ranolazine and GS-6615. (B) Schematic diagram showing relationship between each of I Na and I Na,Late and the ventricular AP (for an AP with a high plateau phase, as occurs in humans). Fast I Na flows during the AP upstroke; I Na,Late is a sustained low amplitude sodium current that flows during the AP plateau phase.

QT Interval and AP Prolongation
Although the prevalence of QT c interval prolongation in RTT patients varies between studies, it clear is that this ECG change does occur in some patients [13][14][15][16][17][18]. Moreover, the finding that serial ECG measurements within a single cohort have revealed increases during follow up in the number of those exhibiting prolonged QT intervals [17] indicates that a lack of QT c prolongation in a single ECG measurement may not be definitive, and therefore repeated testing over time may be prudent. The results of the present study have shown very similar ECG changes in Mecp2 Null/Y animals to those found in the study of McCauley et al. [18]: QRS widening and QT/QT c interval prolongation. The lack of QT c prolongation in Mecp2 Null/Y mice in the study by Hara et al. [24] is at variance with these findings, but measurements were made between 6 and 8 weeks of age and it is possible that the use of animals from earlier timepoints may have influenced the outcome. Two further studies have reported QT prolongation in RTT mice [19,20] and it is notable that this phenomenon has also been seen in a primate model [21]. An observation of QT c prolongation implies that underlying ventricular APs are also prolonged, but this is the first study to provide an explicit demonstration of delayed ventricular AP repolarization in cardiomyocytes from an experimental RTT model. The results of our experiments showed APD prolongation at measurement timepoints (APD 75 and APD 90 ) beyond APD 50 . The lack of significant AP prolongation at early timepoints during repolarization thus resulted in increased AP triangulation. This is significant because augmented AP triangulation is considered to represent a marker of increased proarrhythmic risk [44]. The larger APD 90 BVR in Mecp2 Null/Y than WT myocytes is indicative of increased AP instability, which can also act as a marker of proarrhythmic risk [44,45]. Our AP measurements also revealed two unexpected alterations that cannot be detected in ECG measurements: the threshold current required to evoke APs was lower in Mecp2 Null/Y myocytes; there was also a small (<3 mV) depolarization of resting membrane potential. These alterations suggest that differences are likely to exist between WT and Mecp2 Null/Y myocytes in addition to alterations in Na channel function (e.g., these two observations could be explained by a reduction in resting potassium conductance (s)). Whilst it was beyond the intended scope of this study to pursue the underlying ionic basis of those additional changes, future work to that end is warranted.

Changes to Fast I Na and I Na,Late
To our knowledge, this is the first study to identify changes to fast I Na in the Mecp2 Null/Y RTT model. A previous study included limited data on fast I Na amplitude at a single test voltage, under recording conditions that were optimised for I Na,Late , but not for fast I Na [18] and saw no difference in current amplitude. In contrast, our experiments, performed under conditions that facilitate accurate recording of fast I Na [32][33][34], revealed significant reductions in I Na amplitude over the voltage range that encompassed maximal I Na magnitude, without any significant alteration to voltage dependent activation V 0.5 and with only a modest effect on the slope of voltage-dependent inactivation (Figures 4 and 5). The reduction in I Na amplitude may help explain QRS complex widening in the Mecp2 Null/Y mouse ECG. The modest delay to I Na inactivation time course may be supplemental to this effect.
However, changes to fast I Na are unable to account for altered I Na,Late in the Mecp2 Null/Y RTT model. The fast sodium channel "window" current, denoted by the area of overlap between steady-state activation and inactivation relations, occurs over a range of membrane potentials within which the probability of availability and activation both exceed zero [46]. Noble and Noble have previously highlighted that whilst the I Na window may contribute to a proportion of I Na,Late , it cannot account for all of it; they noted that I Na,Late flows at membrane voltages at which Hodgkin-Huxley models of I Na would predict virtually no current [47]. The experiments in the present study have shown that the I Na window was not increased (rather it was modestly decreased) in Mecp2 Null/Y myocytes, and in consequence this cannot explain the increased I Na,Late in myocytes from the RTT model. Fast I Na and I Na,Late can both be carried by Na v 1.5 channels [46] and our experiments with A-803467 ruled out a role for Na v 1.8 in the I Na,Late recorded in this study. Previous single channel investigations of I Na,Late have revealed "scattered" and "burst mode" channel openings [48][49][50] and it is notable that "scattered" and "burst mode" single channel behaviours have both been seen in recordings from recombinant Na v 1.5 channels [50].
Children with RTT exhibit sympathovagal imbalance, with parasympathetic underactivity and sympathetic overactivity [51]. McCauley et al. reported that QT prolongation and I Na,Late augmentation were observed both in a global Mecp2 Null/Y model and in mice with selective nervous system Mecp2 knockout [18]. This suggests that the cardiac changes linked to repolarization delay are secondary to neural effects in this RTT model. It is therefore striking that ventricular AP prolongation and changes to I Na and I Na,Late can be measured from isolated ventricular myocytes. This indicates that, whatever the identity of the trigger(s) for these changes in the RTT model, they persist in myocytes that are not subject to acute neural modulation. In both humans and mice with RTT, it is thought that chronic intermittent desaturations and re-oxygenations caused by the respiratory abnormalities induce mitochondrial oxidative stress [52] and it is also known that I Na.Late enhancement can occur in pathological states such as hypoxia and ischaemia, with reactive oxygen species (ROS) activation of CaMKII suggested to mediate current enhancement in such circumstances [46,53]. However, although the RTT mice in this study showed an altered respiration rate and the mean duration of apnoea episodes was significantly greater in Mecp2 Null/Y mice, there was no correlation between APD 90 values and apnoea length in our experiments. Mucerino and colleagues have reported alterations to the carnitine cycle in this mouse model of RTT, with upregulation of carnitine palmitoyl transferase 1A/B and carnitine acylcarnitine translocase in the hearts of Mecp2 +/− mice [20]. Whether or not resultant changes in fatty acid metabolism could influence I Na,Late is unknown at this time. Consequently, although our experiments have characterized the changes to I Na and I Na,Late in the Mecp2 Null/Y RTT model, the underlying bases for these alterations remain to be elucidated.

Actions of Ranolazine and GS-6615
In the earlier study by McCauley and colleagues, β adrenoceptor blocker administration to this model of RTT did not reduce QT c interval or protect against arrhythmia provoked by programmed electrical stimulation (PES) [18]. By contrast, Mecp2 Null/Y myocyte treatment with the anticonvulsant drug phenytoin decreased I Na,Late , whilst injection of Mecp2 Null/Y mice with phenytoin reduced QT c interval duration and ventricular arrhythmia susceptibility [18]. A subsequent study confirmed the beneficial effects of phenytoin Mecp2 Null/Y mice (without any accompanying I Na.Late experiments), but it also found that it worsened abnormal breathing patterns [19]. This compromises the suitability of phenytoin as a treatment for prolonged cardiac repolarization in RTT. Retrospective analysis of patients enrolled in the RTT Syndrome Natural History Study found 10 patients with prolonged QT c intervals that shortened after receiving antiepileptic drugs with Na + channel inhibitory properties [19]. Our data constitute independent direct evidence of augmented I Na,Late in a RTT model and taken together with this earlier work they highlight the utility of I Na,Late inhibition as a potential therapeutic strategy for cases in which QT c prolongation is observed.
Both ranolazine and GS-6615 are of interest as I Na,Late -targeting antiarrhythmic agents [35][36][37][38]42,43]; both drugs were observed in our experiments to inhibit I Na,Late in ventricular myocytes from WT and RTT animals and to restore I Na,Late in Mecp2 Null/Y myocytes to control levels. The binding of both ranolazine and GS-6615 to Na v 1.5 channels is sensitive to mutation of aromatic amino-acid residues that contribute to the local anaesthetic (LA) binding site on the channel [43,54]. The retained sensitivity of I Na,Late from Mecp2 Null/Y myocytes to both agents seen here suggests that the LA binding site is unaffected in this RTT model. The finding of prolongation rather than abbreviation of APs by ranolazine suggests that in addition to I Na,Late inhibition, the drug exerted effect(s) on additional ionic current(s) that influence murine AP repolarization. An independent study has also reported murine ventricular AP prolongation with ranolazine, but did not pursue the underlying reason for this [55]. Ranolazine is known to inhibit the hERG-mediated rapid delayed rectifier K + current, I Kr [56][57][58]. However, I Kr does not contribute to murine ventricular repolarization [59] and so cannot account for the observed AP prolongation. Indeed, known differences between the K + currents that mediate human and mouse ventricular repolarization [59] make it difficult to extrapolate pharmacological murine AP prolongation to humans. It is notable, however, that LQT3 patients respond to ranolazine with QT c interval shortening [60] and so it remains possible that in humans and species with a human-like complement of repolarizing currents ranolazine could abbreviate repolarization in the RTT setting.
With its enhanced I Na,Late selectivity and demonstrated effectiveness at inhibiting LQT3 mutant Na channels (e.g., [43]), GS-6615 is a potentially attractive agent for mitigating effects of delayed ventricular repolarization. Its actions are abrogated by mutations (F1760A/Y1767A) at the local anaesthetic binding site on Na v 1.5 [43]. In our experiments it both inhibited I Na,Late and abbreviated ventricular AP duration. APD abbreviation was associated with a reduction in AP triangulation in Mecp2 Null/Y myocytes. These findings highlight a potential for GS-6615 to be used in RTT-associated QT c prolongation. Further work is warranted to determine its ability to inhibit provocation of ventricular arrhythmias in the intact heart of this RTT model.

Limitations and Conclusions
Although most human RTT patients are female with few boys surviving beyond one year of age [4], this study was conducted on male Mecp2 Null/Y mice. Female mice from this strain are heterozygous for Mecp2 deletion (Mecp2 +/− ) and in consequence their clinical phenotype develops rather more slowly than in males: QT c prolongation in Mecp2 +/animals was absent at 4 months, but present at 10 months in the study by McCauley and colleagues [18]. A more recent study found that at 11 months some Mecp2 +/females display QT c prolongation and others do not [20]. These features make characterization of repolarization and repolarization-linked changes I Na /I Na,Late much more challenging and costly in females than in males. The reported similarity between ECG changes in older Mecp2 +/and younger Mecp2 Null/Y mice [18], gives confidence in the use of Mecp2 Null/Y mice for studying repolarization changes in the model. Both the study of McCauley et al. and our own have identified changes to QRS interval width as well as QT c interval in the Mecp2 Null/Y model. Our data on fast I Na highlight changes to this in RTT myocytes that may underpin or substantially contribute to QRS changes. However, whilst the earlier study investigated arrhythmias induced by programmed electrical stimulation, neither investigation involved conduction mapping and it would thus be useful in future work to perform such mapping to establish whether alterations to I Na in the model result in altered conduction. Our experiments on GS-6615 and ranolazine focused on acute application to isolated ventricular myocytes. Further work is now required at the intact heart and chronic in vivo exposure levels to establish the ability of GS-6615 to protect against QT c -linked arrhythmia susceptibility. In the case of ranolazine, further exploration in a non-murine RTT model with ventricular AP repolarization mechanisms closer to those in humans would be useful to explore further its utility in RTT. While it is important that such limitations are acknowledged, they do not diminish the principal findings of this investigation in relation to ECG and AP changes indicative of delayed repolarization in this model of RTT and the changes to I Na , I Na,Late and drug responses that we report. As has been highlighted elsewhere, QT/QT c interval prolongation in RTT is of interest not only due to the association of the syndrome with sudden death, but also because of the use in RTT patients of drugs linked to QT c interval prolongation (serotonin-selective reuptake inhibitors, SSRIs) and a need to monitor QT c prolongation in drug trials [61]. The present study extends information on ionic mechanisms underlying delayed repolarization in RTT. Future work is clearly now needed to elucidate the underlying mechanism for I Na,Late augmentation in this RTT model and to establish the therapeutic value of the potential approaches to mitigating QT c prolongation suggested by our findings.

Mouse Model of RTT Used
All experiments were conducted in accord with UK Home Office legislation and were approved by the University of Bristol Animal Welfare Ethical Review Body (AWERB). Studies were performed on male Mecp2 Null/Y mice with deletions of the third and fourth exons (the so-called "Bird" strain [62]), at between 8 and 10 weeks of age, and on wildtype age-matched male littermates. Mice were genotyped using a standard protocol (P Protocol 24870: Standard PCR Assay-Mecp2<tm1.1Bird> Version 6.2). https://www.jax. org/Protocol?stockNumber=003890&protocolID=24870 (accessed/link confirmed live on 17 April 2022). The following primers were used: common-AAA TTG GGT TAC ACC GCT GA; mutant reverse-CCA CCT AGC CTG CCT GTA CT; wild-type reverse-CTG TAT CCT TGG GTC AAG CTG.

Electrocardiogram (ECG) Measurement
Mice were anaesthetized by 1.5% isoflurane and ECG measurements were obtained 5 min after anaesthesia had been established. Surface lead II ECG measurements were made using subcutaneous needle electrodes, with recording using an A-M Systems (Sequim, WA, USA) differential AC amplifier model 1700 and an CED (Cambridge, UK) Micro 1401-3 AD/DA converter. Signals were high-pass filtered at 10 Hz, with a low-pass setting of 1 kHz. The following ECG parameters were analysed: RR interval (and from this heart rate); PR interval; QRS interval; QT interval (and from this, rate-corrected QT (QT c ) interval); respiration rate (breathing was evident as brief periods of high frequency interference on the ECG recording). Mean values for each ECG parameter for each mouse were obtained from 5 consecutive ECG complexes, avoiding complexes upon which breathing noise was superimposed. The QT interval duration was determined as the interval between onset of the QRS complex and time point after the T-wave peak [18]. QT c interval values were obtained using two distinct rate correction methods (for Equation (1) [25,26]; for Equation (2)

Unrestrained Whole Body Plethysmography
Respiratory patterns were monitored using unrestrained whole-body plethysmography (Emka Technologies, Paris, France). Unanaesthetised individual mice were placed in a mouse chamber (396 mL) with a bias flow of 0.7-0.8 L/min. Chamber pressure, temperature and humidity were measured to allow accurate calculation of respiratory flow. Mice were allowed to adapt for the first 20 min, and data was analysed in the subsequent 20 min of recording. Analysis was automated and manually checked by an experienced researcher. Timeseries respiratory flow data was analysed using published custom written 'algorhythms' in Spike 2 (V8.22, Cambridge Electronic Design, Cambridge, UK) [63]. A running average of the total expiration time for each breath was taken every minute. If an expiration time was longer than 4 times this average, it was counted as an apnoea. Both apnoea count (number of episodes in 20 min) and length (duration of each episode) were recorded. Comparisons were made between Mecp2 Null/Y and WT control mice.

Ventricular Myocyte Isolation
Animals were killed by cervical dislocation, and the heart was then rapidly excised and placed in ice-cold isolation solution supplemented with 0.1 mM CaCl 2 and 10 U/mL heparin. The isolation solution contained (in mM) 130 NaCl, 5.4 KCl, 0.4 NaH 2 PO 4 , 4.2 HEPES, 10 glucose, 1.4 MgCl 2 , 20 taurine, and 10 creatine (pH 7.4 with NaOH) [64]. The heart was Langendorff perfused at 37 o C at constant pressure of gravity (~80-100 cm H 2 O) with isolation solution for 3 min followed by enzyme solution (isolation solution plus 0.1 mM CaCl 2 , 0.07 mg/mL protease (Sigma, Type XIV), and 0.7 mg/mL collagenase (Worthington, Type 1)) for 15 min. The ventricles were removed and shaken in enzyme solution for 5 min before being filtered and centrifuged. Cells were resuspended in isolation solution plus 0.1 mM CaCl 2 and stored at room temperature for using within 10 h.

Cellular Electrophysiology
Ventricular myocytes were placed in an experimental chamber mounted on an inverted microscope (Nikon Eclipse TE2000-U) and were superfused with a Tyrode's solution containing (in mM): 140 NaCl, 4 KCl, 1.5 CaCl 2 , 1 MgCl 2 , 10 glucose, 5 HEPES (pH 7.4 with NaOH). During experimental recordings, solutions were applied to the cell under investigation using a home-built device that was able to exchange local superfusate within 1 s [65]. Patch pipettes were made from borosilicate glass (A-M Systems Inc, Sequim, WA, USA) pulled and fire polished to resistances of 2-3 MΩ (PP-830 and MF83, Narishige, Tokyo, Japan). For whole cell current recording, series resistance values were typically compensated by >70%. Protocols were generated and data recorded online with pClamp 10 and a Digidata 1440A interface (Molecular Devices, San Jose, CA, USA). The digitization rate was 10 kHz. Further details of specific electrophysiological experiments are given below.
In order to quantify the voltage dependence of activation, I Na amplitude data obtained from current voltage (I-V) protocols (see Figure 4, Results) were used to obtain conductance (G) values: where V m is the command voltage potential and E rev is the reversal potential extrapolated from the ascending limb of the I Na I-V relation. Activation relations for I Na were obtained by plotting G/G max (where G max is the maximum value of G obtained during the I-V protocol) against V m values across the range from −80 mV to −20 mV. The resultant plots were then fitted with a Boltzmann equation of the form: G/G max = 1/[1 + exp((V 0.5 − V m )/k)] where V 0.5 = half-maximal activation voltage and k = slope factor for the fitted relation.
To quantify voltage-dependent inactivation of I Na , the following equation was used: where I = measured I Na magnitude during a command to −40 mV following a test pulse to voltage V m ( Figure 5, Results). I max = maximal I Na observed during the −40 mV command during the protocol, V 0.5 = half-maximal inactivation voltage and k = slope factor for the fitted relationship, between −150 and −50 mV. The I Na window was obtained by using the experimentally derived activation and inactivation V 0.5 and k values and equations 4 and 5 to calculate activation and inactivation variables at 2 mV intervals between −120 mV and +20 mV. The product of activation and inactivation variables was then multiplied by the experimentally obtained G max and driving force at each test voltage. The resulting plots represented the I Na window under WT and RTT conditions.
In order to evaluate recovery of I Na from inactivation, a paired pulse protocol was used in which two depolarizing commands from −120 mV to −40 mV were applied. The first (1 s duration) pulse (P 1 ) activated then inactivated I Na . The second voltage command (to −40 mV for 40 ms; P 2 ) was separated from the first by interpulse intervals of differing durations (0.1, 0.3, 1, 3.2, 10, 31.6, 100, 316.2, 1000, 3162 ms). The data were fitted with the following equation: I Na (P 2 /P 1 ) (t) = 1 − (A f exp(−t/τ fast ) + A s exp(−t/τ slow )) where I Na (P 2 /P 1 ) is the ratio of I Na elicited by P 2 applied after a time interval 't' following P 1 . A f represents the proportion of recovery fitted by a fast time constant τ fast and A s represents the proportion of recovery fitted by a fast time constant τ slow .

Data Analysis and Statistics
Data are presented as mean ± SEM. The numbers of myocytes and animals from which particular datasets were derived are given in the relevant Results text and figure legends. Statistical analysis and fits to datasets were performed using Microsoft Excel (Microsoft Corporation, USA), Prism 8.4.3 (Graphpad Software Inc., San Diego, CA, USA) Origin 7.0 (OriginLab, Northampton, MA, USA), and Clampfit of pClamp 10.7 (Molecular Devices, San Jose, CA, USA). Statistical comparisons were made using paired t-test, unpaired t-test, one-way or two-way ANOVA followed by a Tukey/Bonferroni post-test, with equal or unequal variances as appropriate. p < 0.05 was taken as statistically significant.

Drugs
Ranolazine dihydrochloride was obtained from Sequoia Research products Ltd., and 30 mM stock solution was made in distilled water. GS-6615 (also known as eleclazine) was obtained from SYNthesis Med Chem, and 10 mM stock solution was made in DMSO. A-803467 was obtained from Sigma-Aldrich, and 1 mM stock solution was made in DMSO. All stock solutions were stored at −20 • C. Stock solutions were diluted with the external solutions to obtain the final concentrations as given in the Results.