Fibroblast growth factor 23 dysregulates late sodium current and calcium homeostasis with enhanced arrhythmogenesis in pulmonary vein cardiomyocytes

Fibroblast growth factor 23 (FGF23), elevated in chronic renal failure, increases atrial arrhythmogenesis and dysregulates calcium homeostasis. Late sodium currents (INa-Late) critically induces ectopic activity of pulmoanry vein (the most important atrial fibrillation trigger). This study was to investigate whether FGF23 activates the INa-Late leading to calcium dysregulation and increases PV arrhythmogenesis. Patch clamp, western blot, and confocal microscopy were used to evaluate the electrical activities, calcium homeostasis, and mitochondrial reactive oxygen species (ROS) in PV cardiomyocytes with or without FGF23 (0.1 or 1 ng/mL) incubation for 4~6 h. Compared to the control, FGF23 (1 ng/mL, but not 0.1 ng/mL)-treated PV cardiomyocytes had a faster beating rate. FGF23 (1 ng/mL)-treated PV cardiomyocytes had larger INa-Late, calcium transients, and mitochondrial ROS than controls. However, ranolazine (an inhibitor of INa-Late) attenuated FGF23 (1 ng/mL)-increased beating rates, calcium transients and mitochondrial ROS. FGF23 (1 ng/mL)-treated PV cardiomyocytes exhibited larger phosphorylation of calcium/calmodulin-dependent protein kinase II (CaMKII). Chelerythrine chloride (an inhibitor of protein kinase C) decreased INa-Late in FGF23 (1 ng/mL)-treated PV cardiomyocytes. However, KN93 (a selective CaMKII blocker) decreased INa-Late in control and FGF23 (1 ng/mL)-treated PV cardiomyocytes to a similar extent. In conclusion, FGF23 increased PV arrhythmogenesis through sodium and calcium dysregulation by acting protein kinase C signaling.


INTRODUCTION
Atrial fibrillation (AF) is the most often seen cardiac arrhythmia to cause heart failure and cardiovascular events in clinical practice. [1,2] A previous study reported that fibroblast growth factor 23 (FGF23), a regulatory hormone secreted by osteocytes for phosphate-calcium homeostasis, can increase the incidence of AF. [3][4][5] High circulating levels of FGF23 in chronic kidney disease (CKD) linked to an elevated prevalence of AF. [6][7][8][9] FGF23 was shown to induce atrial arrhythmogenesis due to calcium dysregulation with increased phosphorylation of calcium/ calmodulin-dependent protein kinase II (CaMKII). [10][11][12][13] Pulmonary vein (PV) is the most important focus of initiating AF. [14][15][16] Our previous animal study showed that CKD PVs had increased arrhythmogenesis. [17] Accordingly, FGF23 is hypothesized to induce the high incidence of AF in CKD patients through increasing PV electrical activity.
FGF23 is known to produce cardiac hypertrophy by activating calcineurin with concomitant promotion of protein kinase C (PKC), which can increase late sodium current (I Na-Late ) in the cardiomyocyte. [11,[18][19][20][21][22] Moreover, FGF23 increases cellular reactive oxygen species (ROS). [23] Mitochondria ROS are the mian source of cellular ROS in the cardiomyocyte. [24] Up-regulation of I Na-Late can lead to overproduction of mitochondrial ROS and oxidation of CaMKII, resulting in abnormalities of calcium handling. [25] The purpose of this study was to investigate whether FGF23 increases PV arrhythmogenesis and the potential underlying mechanisms.

Figure 3:
The I Na-Late in control and FGF23-treated PV cardiomyocytes and the effect of ranolazine on the spontaneous beating rates of control and FGF23-treated PV cardiomyocytes. A. Tracings and average data of the I Na-Late from control (n = 18) and FGF23 (1 ng/mL)-treated PV cardiomyocytes (n = 18). B. An example and average data show that control (n = 8) and FGF23 (1 ng/mL)-treated PV cardiomyocytes (n = 8) after ranolazine (10 μM) had slower beating rates than those before ranolazine. * p < 0.05, *** p < 0.005. www.impactjournals.com/oncotarget A. Tracings and average data of calcium transients in control (n = 10) and FGF23 (1 ng/mL)-treated PV cardiomyocytes (n = 10) before and after ranolazine (10 μM) administration. B. Tracings and I-V relationship of the NCX from control (n = 10) and FGF23 (1 ng/mL)treated PV cardiomyocytes (n = 10). C. Tracings and I-V relationship of the I Ca-L from control (n = 10) and FGF23 (1 ng/mL)-treated PV cardiomyocytes (n = 10). The inset in the current trace shows the clamp protocol. D. Tracings and average data of caffeine-induced NCX and SR calcium contents from integrating NCX currents in control (n = 11) and FGF23 (1 ng/mL)-treated PV cardiomyocytes (n = 11). * p < 0.05, *** p < 0.005. www.impactjournals.com/oncotarget DISCUSSION FGF23 was shown to enhance arrhythmogenesis in mice atrial cell lines (HL-1 cells) with increased triggered activity of DADs and calcium transient. [10] In the present study, we found that FGF23 at a concentration of 1 ng/ mL increased PV spontaneous activity and amplitude of DADs, which may have been caused by FGF23-increased NCX currents. [26,27] In addition, the shorter APD in FGF23 (1 ng/mL)-treated PV cardiomyocytes may be caused by their larger NCX currents (reverse mode) and faster beating rates (rate-dependent shortening). [28] Our results imply that FGF23 at 1 ng/mL may enhance atrial arrhythmic burden through its arrhythmogenic potential in PV cardiomyocytes.
The increase of I Na-Late can induce calcium overload through calcium entry from the NCX in PV cardiomyocytes and lead to arrhythmogenesis. [29][30][31] We found that FGF23 increased the I Na-Late in PV cardiomyocytes, and ranolazine inhibited the effects of FGF23 on PV arrhythmogenesis. Therefore, FGF23induced I Na-Late may contribute to FGF23-related PV arrhythmogenesis. Ranolazine has been shown to prolong ventricular APD and QTc interval by the inhibitory effect on rapidly activating delayed rectifier potassium current blockers. [32,33] In this study, as the tracing shown in Figure 3B, ranolazine shortened the APD in control and FGF23-treated PV cardiomyocytes. Those findings are similar to the results in the previous studies of ranolazine on ventricular middle myocardium, Purkinje fibers, and PV cardiomyocytes. [29,32] The ranolazine-induced APD shortening may be caused by the decrease in net inward current due to I Na-Late inhibition.
Similar to those found in mice atrial cell lines (HL-1 cells), we found that FGF23 increased the L-type calcium current and calcium transients in PV cardiomyocytes. [10,11] FGF23-increased calcium transients in PV cardiomyocytes may be induced by enhanced I Ca-L , NCX, and I Na-Late currents. [34,35] Moreover, FGF23-treated PV cardiomyocytes had increased phosphorylation of CaMKII, which can hyperphosphorylate the RyR resulting in increased calcium leakage. [13,36] We found that FGF23 induced mitochondrial ROS genesis in PV cardiomyocytes. These results indicated that FGF23 can contribute to the ROS overproduction in CKD and result in calcium dysregulation via increased phosphorylation of CaMKII. [18,25,37,38] The increase of I Na-Late was demonstrated to enhance mitochondrial oxidative stress. [25] We found that ranolazine could block the effects of FGF23 on mitochondrial ROS. Accordingly, as speculated in Figure 7C, FGF23 may produce ROS resulting in calcium dysregulation through activation of the I Na-Late . Our findings imply the therapeutic potential of ranolazine on CKD-related arrhythmogenesis.
Phospholipase Cγ-PKC and phospholipase Cγcalcineurin pathways has been shown to mediate FGF23induced ventricular hypertrophy. [38] However, this study found similar expressions of calcineurin in control and FGF23-treated PV cardiomyocytes. Therefore, phospholipase Cγ-calcineurin pathway is not involved in FGF23-triggered PV arrhythmogenesis. Increased activity of PKC or CaMKII was proven to enhance the I Na-Late leading to calcium overload. [20,22,39] In this study,  an inhibitor of PKC only reduced the I Na-Late in FGF23treated cells which suggests that FGF23 may increase the I Na-Late by PKC ( Figure 7C). However, a selective CaMKII blocker reduced the I Na-Late in control and FGF23treated cells. Therefore, PKC and CaMKII inhibitors both potentially can reduce AF risk in CKD patients.
This study has some limitations. We studied the acute effects of FGF23 on PV electrical activity, but the long-term effects of FGF23 on the cardiovascular system may be different. Although this study tried to dissect the molecular mechanism underlying FGF23-induced PV arrhythmogenesis, interactions of ROS, I Na-Late , PKC, and CaMKII were not completely elucidated. Moreover, we studied the effects of FGF23 on PV cardiomyocytes in healthy animals. However, it is not clear whether our findings can be translated to pathological settings.
In conclusion, FGF23 induces PV arrhythmogenesis with enhanced NCX and calcium transient through increasing the I Na-Late , oxidative stress, and PKC activation.

Isolation of single cardiomyocytes and cell preparations
The investigation was approved by the local ethics review board and conformed to the institutional Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health. Male rabbits (2.0~3.0 kg) were anesthetized with an intraperitoneal injection of sodium pentobarbital (100 mg/kg) and heparin (2500 units). Single cardiomyocytes from rabbit PVs were enzymatically dissociated through a previously described procedure. [40] In brief, a mid-line thoracotomy was performed, and the heart and lungs were removed. PVs were perfused in a retrograde manner via polyethylene tubing cannulated through the aorta and left ventricle into the left atrium. The free end of the polyethylene Figure 7: Signaling pathway and proposed mechanism of FGF23 on the modulation of I Na-Late leading to calcium handling abnormalities. A. Tracings and B. average data of the I Na-Late from control (n = 8) and FGF23 (1 ng/mL)-treated PV cardiomyocytes (n = 8) before and after chelerythrine chloride (2.5 μM) or KN93 (1 μM) administration. *** p < 0.005. C. The I Na-Late was activated by enhanced FGF23 through the signaling of PKC or the phosphorylation of CaMKII, which can be stimulated by increased mitochondrial ROS. FGF23related PV arrhythmogenesis can be induced by calcium overload, which is produced by increased L-type calcium current (LTCC), NCX, and phosphorylation of CaMKII resulting in RyR dysfunction in the SR. tube was connected to a Langendroff perfusion column for perfusion with oxygenated normal Tyrode's solution (containing (in mM): NaCl 137, KCl 5.4, CaCl 2 1.8, MgCl 2 0.5, HEPES 10 and glucose 11; with pH adjusted to 7.4 by titrating with 1 N NaOH). The perfusate was replaced with oxygenated Ca 2+ -free Tyrode's solution containing 300 units/ml collagenase (Sigma Type I) and 0.25 units/ ml protease (Sigma, Type XIV) for 8~12 min. Proximal PVs were cut away from the atrium and lung, and were gently shaken in 5~10 ml of Ca 2+ -free oxygenated Tyrode's solution until single cardiomyocytes were obtained. The solution was then gradually changed to oxygenated normal Tyrode's solution. Cells were allowed to stabilize in the bath for at least 30 min before the experiments. Single cardiomyocytes with spontaneous activity were identified by the presence of constant beating during perfusion with Tyrode's solution. PV cardiomyocytes were pretreated in the control and FGF23 (0.1 and 1 ng/mL) incubated for 4~6 h with or without ranolazine (10 μM

Patch clamp experiments
A whole-cell patch-clamp was used on isolated PV cardiomyocytes with an Axopatch 1D amplifier (Axon Instruments, Foster City, CA, USA) at 35±1C. [40] The ionic currents and action potentials (APs) were recorded at an approximately similar period (3~5 min) after rupture or perforation to avoid decay of ion channel activity over time. The micropipette resistance was 3~5 MΩ. A small hyperpolarizing step from a holding potential of -50 mV to a testing potential of -55 mV for 80 ms was delivered at the beginning of each experiment. The area under the capacitative current curve was divided by the applied voltage step to obtain the total cell capacitance. Normally, 60%~80% series resistance (R s ) was electronically compensated for. APs were recorded in the current-clamp mode and ionic currents in the voltage-clamp mode. Spontaneous beating rate was defined as the constant occurrence of spontaneous APs in the absence of any electrical stimuli. DADs were defined as the presence of spontaneous depolarization of the impulse after full repolarization had occurred. The DADs were selected from consistent deflections without abrupt changes of the resting membrane potential and AP morphology.
The I Na-Late was recorded at room temperature with an external solution containing (in mM): NaCl 130, CsCl 5, MgCl 2 1, CaCl 2 1, HEPES 10, and glucose 10 at a pH of 7.4 with NaOH by a step/ramp protocol (-100 mV stepped to +20 mV for 100 ms, then ramped back to -100 mV over 100 ms). The I Na-late was measured as the tetrodotoxin (30 µM)-sensitive portion of the current traces obtained when the voltage was ramped back to -100 mV.
The I Ca-L was measured as an inward current during depolarization from a holding potential of -50 mV to test potentials ranging from -40 to +60 mV in 10-mV steps for 300 ms at a frequency of 0.1 Hz using a perforated patch clamp with amphotericin B. NaCl and KCl in the external solution were respectively replaced with tetraethylammonium chloride and CsCl. In order to avoid 'run-down' effects, the I Ca-L was measured at 5~15 min after rupturing the membrane patch in each PV cardiomyocyte.
The NCX current was elicited by test pulses of between -100 and +100 mV from a holding potential of -40 mV for 300 ms at a frequency of 0.1 Hz. The amplitudes of the NCX current were measured as 10-mM nickel-sensitive currents. The external solution (in mM) consisted of NaCl 140, CaCl 2 2, MgCl 2 1, HEPES 5, and glucose 10 with a pH of 7.4, and contained strophanthidin (10 μM), nitrendipine (10 μM), and niflumic acid (100 μM).

Measurement of intracellular and SR calcium contents
As described previously, PV cardiomyocytes were loaded with fluorescent Ca 2+ (10 μM, fluo-3/AM) for 30 min at room temperature. [41,42] The Fluo-3 fluorescence was excited with a 488-nm line of an argon ion laser. The emission was recorded at > 515 nm. Cells were repetitively scanned at 2-ms intervals. Fluorescence imaging was performed with a laser scanning confocal microscope (Zeiss LSM 510, Carl Zeiss, Jena, Germany) and an inverted microscope (Axiovert 100, Carl Zeiss). The fluorescent signals were corrected for variations in dye concentrations by normalizing the fluorescence (F) against the baseline fluorescence (F 0 ), to obtain reliable information about transient intracellular Ca2+ (Ca2+i) changes from baseline values ((F -F 0 )/F 0 ) and to exclude variations in the fluorescence intensity by different volumes of injected dye. The Ca 2+ i transient, peak systolic Ca 2+ i, and diastolic Ca 2+ i were measured during a 2-Hz field-stimulation with 10-ms twice-threshold strength square-wave pulses.
After achieving steady-state Ca 2+ transients with repeated pulses from -40 to 0 mV (1 Hz for 5 s), the total amount of charge crossing the membrane SR Ca 2+ content (Ccaff) was estimated by integrating the NCX current rapid application of 20 mM caffeine during rest with the membrane potential clamped to -40 mV to cause SR Ca 2+ release. The NCX current was measured by wholecell patch-clamp experiments. The total SR Ca 2+ content (expressed as mM of cytosol) was determined using of the equation: SR Ca 2+ content = [(1 + 0.12) (Ccaff/F × 1000)]/ (Cm × 8.31 × 6.44), where Cm = membrane capacitance, F = Faraday's number, and the cell surface to volume ratio = 6.44 pF/pL.

Measurement of intracellular ROS
We used the mitochondria-specific ROS indicator MitoSOX Red to assess ROS production in mitochondria of control and FGF23-treated PV cardiomyocytes. Experiments were also performed using a laser scanning confocal microscope (Zeiss LSM 510, Carl Zeiss) and an inverted microscope (Axiovert 100) with a 60x1.4 numerical aperture oil immersion objective as described previously. [25] Cardiomyocytes were maintained in oxygenated normal Tyrode's solution (containing (in mM): NaCl 137, KCl 5.4, CaCl 2 1.8, MgCl 2 0.5, HEPES 10 and glucose 11; with the pH adjusted to 7.4 by titrating with 1 N NaOH) supplemented with the appropriate fluorescent dye of 5 μM MitoSOX Red (life Technologies, Grand Island, NY, USA). MitoSOX Red was excited at 488 nm and fluorescence signals were acquired at wavelengths of > 505 nm in the XY mode of the confocal system. Fluorescent images were analyzed using Image-Pro plus 6.0 and Sigmaplot 12.3 software.

Statistical analysis
All continuous parameters are expressed as the mean ± standard error of the mean (SEM). A one-way or two-way repeated analysis of variance (ANOVA) with a Bonferroni post-hoc test or an unpaired t-test was used to compare differences in control and FGF23-treated PV cardiomyocytes before and after ranolazine, KN93, KN92, or chelerythrine chloride when appropriate. A p value of < 0.05 was considered statistically significant.