Kv4.3-Encoded Fast Transient Outward Current Is Presented in Kv4.2 Knockout Mouse Cardiomyocytes

Gradients of the fast transient outward K+ current (Ito,f) contribute to heterogeneity of ventricular repolarization in a number of species. Cardiac Ito,f levels and gradients change notably with heart disease. Human cardiac Ito,f appears to be encoded by the Kv4.3 pore-forming α-subunit plus the auxiliary KChIP2 β-subunit while mouse cardiac Ito,f requires Kv4.2 and Kv4.3 α-subunits plus KChIP2. Regional differences in cardiac Ito,f are associated with expression differences in Kv4.2 and KChIP2. Although Ito,f was reported to be absent in mouse ventricular cardiomyocytes lacking the Kv4.2 gene (Kv4.2-/-) when short depolarizing voltage pulses were used to activate voltage-gated K+ currents, in the present study, we showed that the use of long depolarization steps revealed a heteropodatoxin-sensitive Ito,f (at ~40% of the wild-type levels). Immunohistological studies further demonstrated membrane expression of Kv4.3 in Kv4.2-/- cardiomyocytes. Transmural Ito,f gradients across the left ventricular wall were reduced by ~3.5-fold in Kv4.2-/- heart, compared to wild-type. The Ito,f gradient in Kv4.2-/- hearts was associated with gradients in KChIP2 mRNA expression while in wild-type there was also a gradient in Kv4.2 expression. In conclusion, we found that Kv4.3-based Ito,f exists in the absence of Kv4.2, although with a reduced transmural gradient. Kv4.2-/- mice may be a useful animal model for studying Kv4.3-based Ito,f as observed in humans.


Introduction
The fast transient outward potassium current (I to,f ) modulates action potential profile, Ca 2+ handling, contractility and hypertrophy in cardiomyocytes [1,2]. I to,f is also developmentally regulated [3] and its levels are invariably reduced in heart disease. Regional differences in I to,f are major determinants of the transmural gradient of ventricular repolarization in many species [4,5] and these gradients are often altered in diseased hearts [6]. The pore-forming αsubunits of I to,f channels are comprised of Kv4.2 and Kv4.3 in rodents [7], but apparently only Kv4.3 in humans [8]. In both case, they form oligomeric complexes with the obligatory auxiliary subunit KChIP2, as well as either DPP6 or DPP10 [7]. While the I to,f gradient in humans and dogs are linked to regional differences in KChIP2 expression [4,9], the basis for regional differences in mouse I to,f is not clear with studies concluding that either Kv4.2 or KChIP2 or both underlie the transmural gradient [10].
In rodents, the molecular basis of I to,f is less clear. As I to,f was undetected in mice lacking the Kv4.2 gene, a previous report concluded that the native cardiac I to,f requires obligatory heterotetramers of Kv4.2 and Kv4.3 [11], even though I to,f currents are present in mice lacking Kv4.3 [12] and can be generated in heterologous expression systems by either Kv4.2 or Kv4.3 α-subunits alone [5]. While several molecular explanations could account for these conflicting findings, the previous studies [11,12] used short depolarization pulses (lasting~5 sec) to quantify I to,f , which do not allow accurate dissection and quantification of K + currents in mouse cardiomyocytes [13,14]. Recently, we showed that long depolarization pulses (lasting~25 sec) are required to accurately dissect and quantify the different K + currents in mouse cardiomyocytes [12]. Therefore, we re-examined the existence of I to,f in mice lacking the Kv4.2 gene (Kv4.2-/-) and found clear evidence for the presence of I to,f in Kv4.2-/-hearts with a markedly reduced transmural gradient.

Transgenic animals and genotyping
All animal experiments were performed in accordance with protocols approved by the Faculty of Medicine and Pharmacy Animal Care Committee, University of Toronto, and conformed to the standards of the Canadian Council on Animal Care. Mice lacking of Kv4.2 gene (K V 4.2-/-) [11] were bred into a C57/B6 background. Heterozygous-heterozygous breeding was used to generate Kv4.2-/-and wild-type (WT) littermates. Genotyping was performed using the following primers: K V 4.2 (613 bps; forward, GTGGAT GCC TGT TGC TTC; reverse, CCC ACA AGG CAG TTC TTT TA) and aminoglycoside phosphotransferase (neomycin-resistance) (500 bps; forward, AGG ATC TCC TGT CAT CTC ACC TTGCTC CTG; reverse, AAG AAC TCG TCA AGA AGG CGA TAG AAG GCG) [15].

Isolation of Adult Mouse Ventricular Myocytes
As previously described [13], hearts were rapidly removed from anesthetized (4% isoflurane oxygen mixture inhalation) mice and retrogradely perfused with Ca 2+ -free Tyrode's solution, which contained (in mM): 137 NaCl, 5.4 KCl, 1.0 MgCl 2 , 0.33 NaH 2 PO 4 , 10 D-glucose, 10 HEPES, pH 7.4, at 37°C through aorta for 3-4 min. Then it was perfused with collagenase (Type II, 1.0 mg/mL, Worthington) for 10-12 min. The EPI and ENDO heart cells were collected with a sharp forceps under a dissecting microscope. After heart was digested, the right ventricle was removed and an incision was made into the ventricular septum in order to expose the endocardium (ENDO) of the left ventricular free wall. A small amount of ENDO tissue (~0.05mm 3 ) was collected from left ventricular free wall inner surface at the level of the tip of papillary muscle. Then the outside surface of left ventricular was washed with Krebs bicarbonate solution and a small amount (~0.05mm 3 ) of EPI tissue was picked up from the middle point of left ventricular free wall. The digested ventricular tissue was gently triturated in the cell storage solution, which contained (in mM): 120 potassium glutamate, 20 KCl, 20 HEPES, 1.0 MgCl 2 , 10 D-glucose, 0.5 K-EGTA, and 0.1% bovine serum albumin.

Patch Clamp Electrophysiology & Data Analysis
Voltage-activated K + currents from isolated ventricular myocytes were recorded with the whole-cell patch clamp technique (Axopatch 200B and Clampex8 software, Axon Instrument, CA, USA) at room temperature (24°C). Measurements were made at room temperature in order to facilitate comparisons with previous studies in which voltage-gated K + currents were quantified in mouse cardiomyocytes [4,7,11,16,17] and in particular to allow comparisons with our previous study wherein a method was developed to reliably dissect voltage-gated K + currents in mouse cardiomyocytes [13]. Cells were perfused with bath solution for 15 minutes before electrophysiological recording. Ca 2+ tolerant rod-shape cardiomyocytes were selected and examined. The bath solution contained (in mM): 140 NaCl, 4 KCl, 1 MgCl 2 , 1.2 CaCl 2 , 10 HEPES, 10 D-glucose, and 0.3 CdCl 2 (pH 7.4). The pipette resistance ranged between 1.2~2.0 MO when filled with a pipette solution, which contained (in mM): 120 potassium aspartate, 20 KCl, 10 NaCl, 1 MgCl 2 , 5 MgATP, 10 HEPES, and 10 EGTA (pH 7.2). Cell capacitance and series resistance were electronically compensated by 85%. Voltage-gated K + currents were induced with a 25 second depolarization from a holding potential of -80 mV to +60 mV, at which membrane potential K + channels are fully activated and also minimize the contributions of potential overlapping Na + currents. A double-pulse protocol was used for characterizing K + currents recovery from inactivation kinetics. Cells were first depolarized to +60 mV for 20 sec (pre-pulse), subsequently hyperpolarized to the holding potential at -80 mV for various times ranging from 10 ms to 6 sec, and then stepped to +60 mV for another 5-20 sec (testpulse) to activate the currents and assess the extent of recovery; sweep interval was set at 45 sec for full recovery of all K + currents and allow sufficient time to adequately check and adjust (as required) the series-resistance compensation and resting membrane potential before each subsequent depolarization. HpTx-2 (5 μM) was used in bath solution for certain experiments.
Electrophysiological data were analyzed using pClamp software (Clampfit9, Axon, CA, USA). Inactivation of voltage-dependent outward K + currents was fitted with a sum of three exponentials: Here, A is the current amplitude, τ is inactivation time constant, C is non-inactivate current component (I ss ). The goodness of fitting was evaluated by the correlation coefficients (R value).
To quantify the recovery of I to in mouse cardiomyocyte, I to in both pre-pulse and test-pulse were dissected from I Kslow1 , I Kslow2 and I ss with 3-exponential curve fitting. I to in test-pulse was normalized to that in pre-pulse, and the ratio was plotted as a function of pulse interval time and fit with a biphasic equation: Here A is percentage of each component in whole current (A 1 + A 2 = 1), τ is the recovery time constant.
corresponding to amino acids 415-636 in the cytoplasmic C-terminus of a rat Kv4.3 protein [19]. A secondary Alexa Fluor 488 goat anti-mouse IgG (Invitrogen) was subsequently used for visualization. Images were acquired with an argon laser beam equipped confocal microscope (excitation 488nm, emission 519nm; Olympus). To control for background and cell auto-fluorescence, parameters were normalized based on fluorescence intensities measured in control samples containing no antibody, primary antibody only or secondary antibody. The Kv4.3 expression was quantified by the average fluorescence signals in 3-dimensional reconstructions of immunofluorescence images of isolated myocytes obtained using optical Z sections (0.5 μm thickness) with a 60X magnification oil objective in conjunction with a Kalman filter (n = 3). The distribution of fluorescence density across the cells in the middle layer image was analyzed. 3D construction was performed using ImagePro Plus (Media Cybernetcis).

Quantitative RT-PCR
Total RNA was isolated with TRIZOL reagent (Life Technologies), and then cDNA was synthesized from 1.0 μg of RNA using Superscript II reverse transcriptase (Life Technologies). The samples from different regions of the heart (i.e. left ventricular epicardium and endocardium) were obtained by dissection from the left ventricular free wall using a scalpel with the aid of a dissecting microscope. Gene expression assay was conducted on 10 ng of template cDNA by Quantitative PCR (qPCR) using Taqman and SYBR green PCR methods equipped with ABI 7900HT (Applied Biosystems). Primers for mouse Kcnd2 (Kv4.2), Kcnd3 (Kv4.3) and Kcnip2 (KChIP2) are as follows: Kcnd2 forward, 5'-GTGTCGGGAAGCCATAGAGGC-3', Reverse, 5'-TTACAAGGCAGACACCCTGA-3'; Kcnd3 forward, 5'-CTCCCGTCGTAGCAAGAAGA-3', Reverse, 5'-GGTGGGGATGCTGATGATG-3'; and Kcnip2 forward, 5'-GGCTGTATCAC GAAGGAGGAA-3', Reverse, 5'-CCGTCCTTGTTTCTGTCCATC-3'. The relative expression of these genes was determined using the normalized comparative CT method [20] with Gapdh as the background gene, which did not vary between regions. The total cycle numbers for all experiments was 35 and the products were verified using both melt tests and agarose gels. Thresholds were determined from the linear regions of the amplification curves and the efficiency of the probes was estimated using serial dilutions.

Statistics
All results were expressed as Mean ± S.E.M. Statistical significance of differences were determined by Student's t test (paired or unpaired) and ANOVA. Differences at P< 0.05 were considered statistically significant. Calculations and statistical tests were performed using Prism5.03 (GraphPad Software, San Diego, CA, USA).

I to, f is present in Kv4.2-/-endocardial myocytes (ENDO)
Voltage-dependent K+ currents in mouse cardiomyocytes are generated by Kv4 [13]. Fig 1A shows K + currents recorded using 25 sec depolarizing pulses (to +60mV) from a holding potential of -80mV recorded in wild-type (WT) and Kv4.2-/-cardiomyocytes. As reported previously, the outward K + currents in WT cardiomyocytes (isolated from the ENDO) are well fit with triexponential functions (R-value range was 0.976-0.995, 88% of R-values were > 0.98). Cardiomyocytes from Kv4.2-/-mouse hearts could be similarly fit well with tri-exponential functions and we found no differences in either the amplitudes or decay (i.e. inactivation) kinetics of the I k,slow1 , I k,slow2 and I sus components between the groups ( Fig 1B and  (A) Typical outward K + currents measured in both the WT and Kv4.2-/-are shown following step depolarization to +60mV (for 25s) from a holding potential of -80mV. The 2 nd and 4 th panels are 50-folds time expanded relative to the area in the boxes on their left side adjacent panels (1 st and 3 rd panels). Voltage-dependent K + currents recorded from both groups of cardiomyocytes could be well fit with tri-exponential functions (show in blue). The three green lines in each panel represent the three individual components of the tri-exponential functions, and their time constant (t) are shown in the inset tables. The residuals (i.e. the differences between the red and blue curves) are also shown in black and, as can be seen, these residuals are essentially zero consistent with the very high correlation coefficients (R-values) typically estimated in the fits of all our traces (which ranged from 0.976-0.995 with 88% of R-values being > 0.98). (B) Summary of the predicted inactivation time constants of the three individual exponential components of the tri-exponential functions that provided the best fit for all K + current traces analyzed. These three components were labeled based on their kinetics as: I Kslow2 (slowest), I Kslow1 (intermediate) and I to (fastest). We previously demonstrated that tri-exponential fits provide optimal fits to the outward K + currents thereby allowing the different underlying current to be quantified in detail 21   The normalized current is defined as the amplitude of the I to density (i.e. the fastest component) estimated from fits to the currents elicited following depolarization during the recovery period (as described in Fig 1) [22,28,29], as discussed further below.

Expression of Kv4.3 α-subunit at the plasma membrane of myocytes
Although as expected Kv4.2 protein was not detected in Kv4.2-/-myocardium [11], Kv4.3 protein is expressed in Kv4.2-/-myocardium even though I to,f was failed to be detected using the short depolarization pulses (lasting~5 sec) method [11], and these observations were considered as key to conclusion that Kv4.2 is required for functional integration of Kv4.3-based channels into the cardiomyocyte membrane [11]. However, in heterologous expression systems Kv4.3 alone generates I to,f and is inserted into cell membrane [23,30]. We therefore performed immunofluorescent staining with Kv4.3 antibodies. As can be seen, from longitudinal and transverse three-dimensional reconstructions of isolated cells labeled with Kv4.3 antibodies (with the appropriate secondary antibodies), Kv4.3 α-subunits are clearly present at the membrane as indicated by the higher fluorescent density at the periphery of the cell than cell interior in both WT and Kv4.2-/-ventricular myocytes (Fig 4). The average fluorescent density in the periphery were comparable (P = 0.4886) between WT (99.17 ± 5.32, n = 9) and Kv4.2-/-(95.20 ± 2.91, n = 13).

I to heterogeneity in the ventricular free wall of Kv4.2-/-
The results above support the conclusion that Kv4.3 α-subunits can form homotetrameric functional channels and generate I to,f in mouse cardiomyocytes when Kv4.2 is absent. Since the transmural I to,f gradients in mouse myocardium have been linked to variations in the expression of both Kv4.2 and KChIP2 [10], we further assessed regional variations of I to,f in the Kv4.2-/-hearts. In order to evaluate transmural heterogeneity of I to,f , K + currents were recorded from epicardial (EPI) cardiomyocytes from the left ventricular free wall. The analysis revealed, that the density of I to was larger in the EPI cardiomyocytes than ENDO cardiomyocytes in both WT and Kv4.2-/-hearts ( Fig 5A and Table 1). More importantly, this transmural I to pattern was entirely due to differences in the fast components of I to,f ( Table 2). After separating I to,f and I to,s as shown above in Fig 5B, I to,f density in EPI was~4-fold bigger than in ENDO in WT, while it only~1.2-fold bigger in EPI than ENDO in Kv4.2-/- (Fig 5C). Thus, while Kv4.2 appears to be a major determinant of the I to,f transmural gradient in mouse myocardium, a residual transmural I to,f gradient remains in Kv4.2-/-and is associated with a corresponding the gradient (P = 0.0418) of KChIP2, but not (P = 0.7214) Kv4.3, mRNA expression. Interestingly, the transmural Kv4.3 mRNA gradient in the Kv4.2-/-hearts did not differ (P = 0.8633) from that in the WT hearts (Fig 6).

Discussion
Our results establish that, when long depolarization pulses (25 sec), I to,f can be readily identified. This conclusion clearly contrasts with a previous study [11] concluding that I to,f is absent in Kv4.2-/-hearts. We believe the lack of agreement can be readily explained by differences in membrane repolarization (to -80mV). (G) This panel shows the normalized I to current as a function of the pulse interval duration during the recovery from inactivation protocol with 5 μM HpTx-2. The normalized current is defined as the amplitude of the I to density (i.e. the fastest component) estimated from fits to the currents elicited following depolarization during the recovery period (as described in Fig 1) divided by the I to density (i.e. fastest component) estimated from fits of the current measured following depolarization during the initial conditioning pulse. Notice that, the recovery of I to in both WT and Kv4.2-/-here are similar and are best fit with mono-exponential equation without the fast recovery component (that is, I to,f ) showed in Fig 2B. Moreover, they were comparable to the slow components (that is, I to,s ) dissected in Fig 2B without HpTx-2.
the voltage protocols and methods of quantifying the time dependent K + current traces. Specifically, in a previous study [13] we showed that, when K + current traces are lasting~25 s, K + current traces are best fit with tri-exponential functions. This conclusion can be readily illustrated by showing that the distribution of the τ inact values obtained from tri-exponential fits are much "tighter" (i.e. smaller variance) than when either bi-exponential or quadri-exponential fits to the traces were used [13]. Our results further showed that the estimates of the magnitude and time constants of the different current components become increasingly inaccurate as the length of the traces (used for the fits) are shortened. These observations are consistent with our analysis showing that tri-exponential functions gave the best statistical fits to the K + current traces [31]. We confirmed these findings in the Kv4.2-/-cardiomyocytes (data not shown) and found further that only the component with fastest time constant (i.e. τ inact for I to ) differed with WT cardiomyocytes. One potential complication when using long depolarization pulses is K + ion depletion which could impact on the measured kinetics of K + current decay during the depolarizing pulses. However, the resting membrane potentials did not differ immediately before and after the long depolarizing pulses (i.e. before -72.38 ± 0.29 mV, n = 200 vs. after -72.38 ± 0.29 mV, n = 200; S1 Table). These observations are not unexpected since the amount of K + that is able to diffuse from the pipette (1.2~2 MΩ, 1~3 μm tip) into the cell is predicted, from the application of Fick's Law of Diffusion [32,33], to be at least 10-fold greater than the amount of K + that leaves the cardiomyocytes through K + channels (i.e. 2.92 ± 0.09 × 10 −7 Coulombs) during the depolarizing pulses under our experimental conditions. We are confident that the long depolarization pulses did not significantly distort the various K + currents or the conclusions derived from our analysis. Although, as expected, the estimated density of I to,f was reduced in the Kv4.2-/-cardiomyocytes compared to WT, there were also notable difference in both the kinetics of inactivation (τ inact ) and recovery from inactivation (τ rec ) of I to,f between cardiomyocytes from Kv4.2-/-(τ inact =~105ms and τ tec =~87ms) versus for WT (τ inact =~58ms and τ rec =~34ms) hearts. The deceleration in the rates of inactivation and recovery from inactivation observed in Kv4.2-/-cardiomyocytes are consistent with slower intrinsic inactivation properties of Kv4.3-based currents compared to Kv4.2-based currents observed in heterologous expression systems [28]. Taken together, these results support the conclusion that I to,f in mouse cardiomyocytes can be formed by Kv4.3 alone, and does not require co-assembly of Kv4.2 and Kv4.3 [11]. This conclusion is further supported by our toxin studies showing that the highly selective blocker of Kv4 currents, HpTx-2, was reduced K + currents in Kv4.2-/-cardiomyocyte during the first 400 ms after depolarization only, which is the period during which I to,f is expected to be active (i.e. not inactivated). Moreover, the densities of the HpTx-2-sensitive current were This shows the average I to density (I to = I to,f + I to,s ) estimated in EPI and ENDO cardiomyocytes isolated from WT and Kv4.2-/-hearts (n = 12 for all groups) using the tri-exponential fitting protocol (described in Fig 1). As we can see, I to density in Kv4.2-/-was reduced compared to WT for both ENDO and EPI cardiomyocytes. What is striking here is that the transmural gradients of I to (EPI/ENDO) exist in left ventricular free walls for both WT and Kv4.2-/-but this gradient is markedly reduced in the Kv4.2-/-compared to WT. (B) Again I to recovery from inactivation curve (n = 12 in each group) was used to separate I to,f and I to,s in I to using the method described in Fig 2. (C) This panel quantifies the I to,f measured in ENDO and EPI cardiomyocytes from WT and Kv4.2-/-hearts (n = 6 for all groups).  Kv4.3-Encoded K + Current in Mouse Cardiomyocyte indistinguishable from the I to,f estimated from the analysis of our curve fits to the voltage-gated K + currents. In addition, the HpTx-2-sensitive currents in Kv4.2-/-cardiomyocytes had decay properties that were also slower than in WT cardiomyocytes, as expected if I to,f in the Kv4.2-/cardiomyocytes is comprised of Kv4.3-based channels [28]. HpTx-2 also eliminated the fast recovering component of I to in our recover from inactivation measurements (i.e. I to,f ) in both WT and Kv4.2-/-, which uncovered mono-exponential slow recovering component (i.e. I to,s ) in both groups with identical magnitudes and kinetics. Our conclusion that I to,f does not require Kv4.2/3 co-assembly in mouse myocardium is supported by a recent study showing that I to,f is detectable in cardiomyocytes from mice lacking Kv4.3, which showed accelerated rates of inactivation compared to wild-type [12]. This conclusion is also consistent with the observation in Western Blots that the Kv4.3 protein (and mRNA) are present in Kv4.2-/-hearts at levels equivalent to that in WT hearts [11]. We confirmed that no significant differences in either Kv4.3 mRNA expression and or Kv4.3 protein in the membranes as measured immunohistochemically between Kv4.2-/-and WT cardiomyocytes.
Another novel finding of our studies is the identification of a transmural gradient in the Kv4.2-/-hearts. Previous studies have established that the gradients of I to,f underlie regional differences in the amplitude and kinetics of contraction in rodent [34,35] and dog [36] hearts, which are predicted to synchronize contraction across the ventricular wall [1]. Regardless, the I to,f gradient in the Kv4.2-/-hearts was associated with a gradient in KChIP2, as we can see its expression seems increase a little in Kv4.2-/-in Fig 6, but not Kv4.3 expression. These observations suggest that the KChIP2 gradient can contribute to the transmural I to,f gradient in mouse hearts, as appears to be the case in dogs [9] and humans [37] which both do not express Kv4.2 [24,29]. However, a recent study has linked Kv4.2 mutations to cardiac arrhythmias [38]. Consistent with previous mouse studies [10] and unlike dog hearts [37], we did not see a transmural gradient of Kv4.3 expression in the mouse heart, although a trend (P = 0.1984) did exist for a gradient. Taken together, the observation that the transmural I to,f gradient exists, albeit with a reduced magnitude, in Kv4.2-/-hearts compared to WT hearts, supports the conclusion that both Kv4.2 and KChIP2 contribute to the I to,f gradient in the mouse heart.
Since the Kv4.2-/-heart expresses Kv4.3-dependent I to,f like in humans cardiomyocytes, these mice might be useful for studying the regulation and consequences of I to,f in humans who do not appear to express Kv4.2, although Kv4.2 mutation was associated with a single patient with J-wave syndrome [38]. This suggestion is bolstered by the observation that I to,f levels and I to,f channel gene expression change similarly between humans and mice in both heart disease and during heart development [3,[39][40][41]. Moreover, despite the fact that the action potentials durations may differ by~10-fold between these species, the effects of changes in I to,f are primarily limited to the early notch of the action potential in human cardiomyocyte which are predicted to be associated with alterations in I Ca,L current profiles and thereby excitation-contraction coupling as seen in mice and other species [1,[42][43][44][45][46]. Future studies will be required to assess whether the Kv4.2-/-mice are a useful model for further understanding I to,f in humans, and how the change of I to,f affect the expression and function of Kv1.4, Kv2.1 and Kv1.5-encoded channels and currents.
Supporting Information S1 Table. Effect of 25 s stimulation pulse on resting potential in cardiomyocytes. Resting potential in Kv4.2-/-and WT ventricular myocyte were measured with whole-cell patch-clamp technique before and immediately after 25s stimulation pulse at +60 mV. The resting potential did not changed immediately after stimulation pulse which suggests no significant intracellular K + concentration changed by the stimulation pulse. (XLSX)