Role of Sodium Channel Deglycosylation in the Genesis of Cardiac Arrhythmias in Heart Failure

We investigated the cellular and molecular mechanisms underlying arrhythmias in heart failure. A genetically engineered mouse lacking the expression of the muscle LIM protein (MLP -/- ) was used in this study as a model of heart failure. We used electrocardiography and patch-clamp techniques to examine the electrophysiological properties of MLP -/- hearts. We found that MLP -/- myocytes had smaller Na + currents with altered voltage-dependencies of activation and inactivation and slower rate of inactivation than control myocytes. These changes in Na + currents contributed to longer action potentials and to a higher probability of early afterdepolarizations in MLP -/- than in control myocytes. Western blot analysis suggested that the smaller Na + current in MLP -/-myocytes resulted from a reduction in Na + channel protein. Interestingly, the blots also revealed that the alpha subunit of the Na + channel from the MLP -/- heart had a lower average molecular weight than in the control heart. Treating control myocytes with the sialidase neuraminidase mimicked the changes in voltage dependence and rate of inactivation of Na + currents observed in MLP -/- myocytes. Neuraminidase had no effect on MLP -/- cells thus suggesting that Na + channels in these cells were sialic acid-deficient. We conclude that deficient glycosylation of Na + channel contributes to Na + current-dependent arrhythmogenesis in heart failure. caused by deficient glycosylation of the Na + channel protein during HF. We conclude that incomplete glycosylation during post-translational processing contributes to Na + -channel-dependent arrhythmogenesis in HF.


INTRODUCTION
Cardiac arrhythmias are a leading cause of death among patients with HF (1). However, the cellular and molecular changes in cardiac myocytes with contractile dysfunction leading to arrhythmias are not clear (2). Recent work suggests that action potential (AP) prolongation is a major contributing factor to the increased probability of cardiac arrhythmias observed in HF. The cellular and molecular processes that underlie AP prolongation have been attributed to: (1) altered K + channel currents (3)(4)(5)(6)(7)(8)(9)(10)(11) (2) changes in Na + channel currents (12)(13)(14)(15) and (3) altered Ca 2+ signaling (16)(17)(18). These alterations are often brought about by changes in the amount or identity of specific ion channels or Ca 2+ signaling proteins (11,19,20). In addition, it has been suggested that alterations in post-translational modification of proteins, specifically reduced glycosylation, could also contribute to HF and related conditions (21). The molecular mechanisms underlying these perturbations in protein expression and processing during HF have remained unclear.
The purpose of this study was to investigate the molecular causes of AP prolongation and arrhythmias in HF. The muscle LIM protein "knockout" mouse (MLP -/-; 2, [22][23][24] was used as a model of HF because it has been examined extensively by multiple groups and found to have reproducible HF. Furthermore, the MLP -/mouse reproduces many of the morphological and clinical features of HF in humans (22,25). We found that during HF, Na + currents (I Na ) are modified in ways that cause action potential prolongation and arrhythmogenesis. Our data suggest that these changes in I Na are

Isolation of Cardiac Myocytes
Adult animals (25 g) were euthanized with an intraperitoneal injection of pentobarbital (100 mg/kg) in strict accordance to the guidelines established by the Institutional Animal Care and Use Committee, which follows all applicable state and federal laws. Single mouse ventricular myocytes were isolated as previously described (26) and stored at RT (22-25 °C) in Dulbecco's modified Eagle's medium (Sigma Chemical Co., St. Louis, MO) until used.

ECG measurements
ECGs were recorded from diazepam-sedated (15 mg/kg) WT and MLP -/mice using previously described methods (27). A differential amplifier (Warner Instruments, Hamden, CT) was used to record any one of the six bipolar limb leads in standard fashion. Data was digitized at 2 kHz with a Digidata 1320 A/D converter (Axon Instruments, Foster City, CA) and stored in a computer running Axotape software (Axon Instruments, Foster City, CA). Analysis was performed in signal-averaged beats. To do this the first 20 beats recorded were aligned with respect to their R peak and then averaged. QT intervals were calculated using the lead with the most prominent T wave (usually lead 1). Correction of the QT interval for heart rate (QT c ) was performed using the formula (28): Analysis of APs and I Na was performed using Clampfit 8 software (Axon Instruments, Foster City, CA). Normalization of ventricular Na + currents was performed by dividing them by the capacitance (in pF) of the cell they were recorded from. Cell capacitance was measured using the membrane test module of Clampex 8.0. I Na conductance was dete rmined according to the following formula: Where G is the conductance of sodium, I is the measured current, V test is the test potential at which the current was measured and V rev is the calculated Nernst equilibrium potential of Na + (0 mV).
Analysis was performed on I Na that met three criteria. First, the amplitude of I Na could not exceed 3 nA. Second, there were no abrupt increments in I Na during the increasing phase of the current-voltage relationship. Third, the series resistance was not larger than 3 MΩ.
Protein samples were loaded onto 5 -6% discontinuous gels and resolved with SDS-by guest on March 24, 2020 http://www.jbc.org/ Downloaded from Arrhythmogenic Na + currents in Heart Failure 8 PAGE. Equal amounts of MLP -/and WT proteins were loaded in multiple lanes. Two control and two MLP -/mice were used in these experiments. Proteins were transferred to PVDF-plus membranes (Osmonics, Westborough, MA) at 45mA for 4 hours using a semidry electroblotter (Owl Scientific). Membranes were used for western blot analysis as previously described (29). Polyclonal antibodies against the α subunit of the Na + channel were obtained from Upstate biotechnology (Lake Placid, NY) and Alomone labs (Jerusalem) and used at the recommended concentration. Both antibodies recognized the same Na + channel smear whose lower boundary was the predicted molecular weight of the α subunit of the Na + channel protein (~220kD). The Na + channel smears produced by both antibodies could be blocked by pre-absorption of the antibody with the peptide antigen. In addition, each antibody produced a unique set of lower molecular weight bands that could not be blocked by pre-absorption with the antigen. The intensity of these non-specific bands did not vary between control and mutant mice, and served as convenient markers for normalization of the Na + channel signal. Optical densities for antibody signals were determined with Gel Expert software (Nucleotech, Hayward, CA).

Statistics
Data is presented as mean ± SEM. Two-sample comparisons were performed using Student's t-test. When multi-group comparisons were necessary they were made using a one-way ANOVA followed by a Tukey's test. In all tests a p value smaller than 0.05 was considered an indicator as a significant difference amongst groups.

Action potential prolongation and early afterdepolarizations in MLP -/cells
We examined the electrical activity of ventricular myocytes with HF and contractile dysfunction using whole animal electrocardiograph (ECG) measurements and single cell AP's. The electrical properties of the MLP -/mouse heart are compared with those of WT mouse hearts in Fig. 1. ECG's taken from adult mice showed signs of abnormal electrical activity in the MLP -/animals ( Fig. 1A). Specifically, MLP -/mice had QT intervals (87.56 ± 6.56 ms, n=5) that were significantly longer than those observed in WT mice (46.57 ± 4.76 ms, n=7; p = 0.01). This difference in QT interval between WT and MLP -/mice was still significant after correcting it for the heart rate (QT c interval) of each animal (40.24 ± 4.78 ms in WT vs. 77.50 ± 6.56 in MLP -/-, p = 0.02).
Because long QT intervals are caused largely by prolonged AP's our data suggest that MLP -/myocytes have longer AP's than WT myocytes. To test this hypothesis AP's were Indeed, Fig. 1D illustrates that the probability of observing early afterdepolarizations in MLP -/cells (25 ± 3%; n=15) was significantly higher than in WT cells (2 ± 1%, n=25, p < 0.001).

Altered membrane current in MLP -/myocytes
The longer QT intervals we observed in MLP -/mice resemble the long QT syndrome (LQTS) in humans. In some LQTS, mutations in Na + channel proteins lead to alterations in I Na that have been associated with increases in AP duration and cardiac arrhythmias (13,16). Though there is no mutation in the Na + channel gene in MLP -/mice, the changes in AP shape that we observed experimentally in Fig. 1 suggested to us that a change in I Na could account for the changes in the AP and arrhythmias seen in the MLP -/heart.
If changes in I Na contribute to the development of the LQTS in the MLP -/model of HF, then specific changes in I Na should be clear and reproducible. Fig. 2 shows the results of a series of experiments that compared the biophysical properties of I Na in WT and MLP -/myocytes. The current-voltage (IV) relationship for I Na of ventricular myocytes isolated from WT and MLP -/hearts is shown in Fig. 2A. The data indicate a lower I Na density in MLP -/than in WT myocytes. The peak I Na was reduced by 35 ± 5 % (n = 6, p < 0.001) in the MLP -/myocytes with a shift in the IV relationship towards more positive potentials. This shift is more clearly revealed in Fig. 2B, where the normalized conductance (G/G max ) is plotted as a function of voltage. The half-maximal conductance is significantly shifted by 12 mV from -49.64 ± 0.13 mV (n = 6) to -37.76 ± 0.28 mV (n = 6; p < 0.001). Additionally, the voltage dependence of the steady-state inactivation of I Na in MLP -/myocytes is shifted 7 mV in the negative direction from -80.66 ± 0.49 mV to -86.20 ± 0.36 mV (p = 0.01) as is shown in Fig. 2C. Fig. 2C also shows that there was a clear slowing of inactivation of I Na in the MLP -/myocytes.
Indeed, the inactivation time constant (τ inactivation ) of I Na in MLP -/myocytes was about 50% greater than in WT myocytes.
These alterations in I Na in the MLP -/mouse ventricular myocyte raise two important questions. First, how are the changes in I Na produced? Second, are the changes in I Na sufficient to produce the observed changes in AP shape?

Deglycosylation of Na + channels in HF
The Na + channel gene is not mutated in MLP -/mice. Thus, if the changes in I Na in MLP -/are due to gene expression and protein processing, the Na + channel could be modified in many different ways. Indeed, down regulation of Na + channel expression levels could account for the 35% reduction in the density of I Na (see above), but not for the shifts in its voltage dependence. On the other hand, improper processing of the channel protein could account for one or both of these effects.
To determine if Na + channel expression levels were altered in MLP -/mice, we performed western blot analysis of the α subunit of the cardiac Na + channel. Fig. 3 illustrates that there is a statistically significant 34.27 ± 0.07% (n=12, p < 0.001) reduction in the intensity of the Na + channel band in MLP -/hearts relative to WT. These data suggest that there is significant reduction in Na + channel protein during HF and that the smaller I Na density observed in MLP -/cells (Fig. 2) results from a lower number of functional Na + channels expressed in these cells.
Close examination of the western blots showed that the α subunit of Na + channels is heavily glycosylated, as the signal appears as a smear above the predicted molecular weight of this glycoprotein (≈ 220 kD), rather than a single sharp band (Fig. 3). The blots also revealed that the width of the Na + channel band was significantly narrower in MLP -/than in WT hearts. Indeed, the width at 50% of the amplitude of the Na + channel band was significantly smaller (21 ± 6%, p < 0.05, n=7) in MLP -/than in WT hearts (Fig. 3).
The reduction in width reflects a loss in the higher molecular weight species. Thus, these data suggest that in the MLP -/heart, heavily glycosylated, high molecular weight Na + channels are less abundant than in the WT heart, resulting in a reduction in the average molecular weight of the Na + channel protein during HF.
Interestingly, the peak and leading edge of the Na + channel from the MLP -/heart tended to migrate further in the gel (12 ± 5%, n = 7) than the Na + channel of the WT heart. However, this shift in the peak and leading edge of the Na + channel band was not statistically significant, a finding that, based on the work of others, is not unexpected (30). The position of the peak reflects the molecular weight of the most abundant species. Thus, these data suggest that the most abundant species of Na+ channel proteins have similar molecular weights in WT and MLP -/myocytes. However, the position of the peak provides little information about the average molecular weight of the Na + channel protein, which depends directly on the average glycosylation of these channels in these cells.
Our western blot analysis suggests that a reduction in the glycosylation of Na + channel proteins during post-translational processing could underlie the changes in I Na observed in MLP -/myocytes. Indeed, the extent of Na + channel glycosylation has been shown to broadly affect this channel's function (30,31). To investigate if a reduction in glycosylation of Na + channel proteins during post-translational processing was responsible for the changes in I Na observed in MLP -/myocytes, both control and MLP -/cells were exposed to neuraminidase, an enzyme known to reduce extracellular glycosylation when applied to the solution bathing cells (30). Myocytes were treated for 3 hours at room temperature (23-25 °C) with 0.3 U/ml of neuraminidase. We found that neuraminidase produced a shift of about 10 mV in the voltage at which the peak I Na occurred in WT cells (Fig. 4a). This shift in the current-voltage relationship of I Na in neuraminidase-treated WT cells is similar to the one observed in MLP -/cells. Note, however, that neuraminidase did not produce any shift in the current-voltage relationship of I Na in MLP -/myocytes. As expected from these results, neuraminidase shifted the voltage-dependence of the relative conductance of I Na in WT cells, but not in MLP -/cells (Fig. 4B). This positive shift in the conductance-voltage relationship of I Na in by guest on March 24, 2020 http://www.jbc.org/ Downloaded from neuraminidase-treated WT cells is identical to that in MLP -/cells. Fig. 4C shows that neuraminidase also produced a negative shift in the voltage-dependence of steadystate I Na inactivation of WT, but not MLP -/cells. This change in the steady-state I Na inactivation of neuraminidase-treated cells is similar to the one observed in MLP -/myocytes. Finally, the rate of inactivation of I Na in WT cells is slowed by more than 50% at negative potentials by the application of neuraminidase as shown in Fig. 4D. This slowing of I Na inactivation in WT cells is identical to the one observed in MLP -/cells.
Neuraminidase did not affect the rate of inactivation of I Na in MLP -/cells. These experiments show that, with respect to I Na , neuraminidase treatment alone on WT cells produces results similar to those in HF. These data suggest that glycosylation of Na + channels is altered in HF.

Discussion
In this paper we have examined the electrical activity of hearts with contractile dysfunction. We found that in HF Na + channel currents have shifted voltage dependencies of activation and inactivation and slower rate of inactivation. Our results suggest that it is the reduction in Na + channel glycosylation (α subunit) that is responsible for the changes of I Na properties. These provocative results raise a difficult question: Are the changes in I Na consistent with the AP prolongation that is observed?

AP Prolongation.
The data presented in this paper show that the acute reduction of Na + channel glycosylation by neuraminidase can reproduce all of the changes in I Na observed in heart failure except one: The I Na current density is not reduced. Our data suggest that this reduction in Na + channel density in HF is produced by a reduction in the expression of functional Na + channels. Most of the changes in I Na that are observed in the MLP -/heart cell tend to reduce I Na and thereby shorten AP duration. The one change that does tend to promote inward current (i.e. slowly inactivating I Na ) is over in less than about 10 ms and therefore has little direct effect on the APD. Thus, how can the AP prolongation observed in Fig. 1 be compatible with the characteristics of I Na shown in Figs. 2? Wehrens et al. (16) addressed an identical question for a Na + channel mutation that they were stud ying: the D1790G mutation.
In this case the D1790G mutation was found in patients who had a Na + channeldependent LQTS known as LQT-3. With the exception of this mutation, Wehrens et al. (16) noted that all other mutations of the cardiac Na + channel that associated with LQTS, resulted in an increase in the sustained component of I Na . The D1790G mutation, however, produces a reduction in the sustained I Na as was shown for the MLP -/heart cells. To examine this conundrum, Wehrens et al. (16) modified the Luo/Rudy model (32,33) of the cardiac AP to include changes in I Na consistent with the D1790G mutation. They discovered that changes like those found in the D1790G mutation can lead to AP prolongation despite reduced AP "overshoot" and reduced maximum dV/dt of the AP upstroke. The model revealed that such alterations in I Na can lead to increases in [Ca 2+ ] I , and the prolongation of the AP was produced by the changes in Ca 2+ . Thus, altered I Na was paradoxically the cause of the AP prolongation in LQT-3. The changes in I Na in HF are similar to those produced by the D1790G mutation in LQTS.
Additionally, Fig. 5 shows that, in agreement with the model proposed by Wehrens et al. (16), the AP overshoot is less positive in MLP -/mouse ventricular myocytes (WT, 49.43 ± 2.36 mV, n=25; MLP -/-, 32.20 ± 4.15 mV, n=15; p < 0.05) and the maximum rate of rise of the AP (dV/dt) is reduced in MLP -/myocytes. Ca 2+ entry is increased when the AP's have a slower rate of depolarization and a less positive peak because these changes increase the driving force for Ca 2+ . A direct consequence of this increase in Ca 2+ influx is that it will augment the inward Na + /Ca 2+ exchanger current during the late stage of the AP and thereby prolong the AP (34). Taken together with the investigation of Wehrens et al. (16), these results suggest that the changes in I Na seen in the MLP -/model of HF could significantly contribute to the AP prolongation observed and the associated EAD's.

Post-translational modification of the cardiac Na + channels during heart failure
Data presented in this paper demonstrate that the deficient glycosylation of Na + channels during heart failure produces a I Na that has a slower rate of inactivation and shifted voltage-dependencies of activation and inactivation. A mechanistic explanation of how negatively charged sialic acid particles on the external portion of Na + channels contribute to the channel's voltage dependence is provided by the surface potential model (35,36). According to this model, fixed, negatively charged particles (phospholipids and sugar residues) on the surface membrane of a cell produce a surface potential. The functional implication of this surface potential is that its magnitude determines the intra-membrane voltage, which is the voltage "sensed" by the Na + channel's voltage-sensor: the smaller the surface potential the smaller the intramembrane voltage and vice versa. Thus, reductions in the number of negative charges in the outer face of the plasma membrane, as would be produced by ion channel deglycosylation or synthesis of sialic acid-deficient glycoproteins, has the effect of reducing the magnitude of the surface potential and hence increase the intra-membrane potential. With a higher intra-membrane potential a greater membrane depolarization will have to occur for the channel to "sense" the same trans-bilayer field. The surface potential model thus predicts that a reduction in the number of negatively charged sialic acid residues linked to Na + channels should produce a positive shift in the channel's threshold for activation.
It is intriguing to speculate on the possible mechanism by which Na + channels are deficiently glycosylated during HF. One possibility is that the activity of sialyltransferase, the enzyme responsible for linking sialic acid to Na + channels proteins in the Golgi apparatus, is reduced in MLP -/mice, which would lead to the production of sialic aciddeficient Na + channels. Interestingly, the activity of sialyltransferase has been found to be reduced in a model of HF (21). Future experiments should examine if sialyltransferase activity is down regulated in MLP -/myocytes, and if this correlates with ion channel dysfunction. Our data suggest that the reduction in I Na density observed in MLP -/cells results from a reduction in the number of functional Na + channels expressed in these cells. Indeed, this conclusion is consistent with the observation that deglycosylation of cardiac Na + channels does not modify their unitary conductance (31). It would be interesting to examine whether this reduction in Na + channel number results from a reduction in the abundance or translation of Na + channel transcripts and/or a defect in the post-translational processing of these channels that promotes their rapid degradation.
We conclude that in the MLP -/model of HF, changes in the glycosylation of Na + channel proteins during post-translational processing appear to be responsible for the changes in I Na that lead to AP prolongation and the development of EAD's. Finally, it is intriguing to speculate that other ion channels and regulatory proteins in the heart may be subject to modification like the one observed for Na + channels in HF.

FOOTNOTES
The abbreviations used are: HF, heart failure; AP, action potential; APD 90 , action potential duration at 90% repolarization; I Na , sodium current; MLP, muscle LIM protein; WT, wild type; MLP -/-, MLP knockout; G, conductance; G max , maximal conductance; ECG, electrocardiogram; EAD, early afterdepolarization; QT c , heart rate-corrected QT interval; V test , test potential; V rev , Nernst equilibrium potential; EGTA, ethylene glycol-    The graph in the right shows a normalized densitometric scan of the specified region of the western blot. The insets are bar plots of the mean ± sem of the relative intensity (right) and relative width at 50% of the amplitude (left) of the Na + channel band. The asterisk in these plots represents a statically significant difference between the WT and MLP -/-.