Chronic Kidney Disease Induces Pro-Arrhythmic Remodeling

Abstract

Background:

Patients with chronic kidney disease (CKD) are at increased risk of developing cardiac arrhythmogenesis and sudden cardiac death, however the basis for this association is incompletely known.

Methods:

Here, using murine models of CKD, we examined interactions between kidney disease progression and structural, electrophysiological, and molecular cardiac remodeling.

Results:

C57BL/6 mice with adenine supplemented in their diet developed progressive CKD. Electrocardiographically, CKD mice developed significant QT prolongation and episodes of bradycardia. Optical mapping of isolated-perfused hearts using voltage-sensitive dyes revealed significant prolongation of action potential duration (APD) with no change in epicardial conduction velocity. Patch-clamp studies of isolated ventricular cardiomyocytes revealed changes in sodium and potassium currents consistent with APD prolongation. Global transcriptional profiling identified dysregulated expression of cellular stress response proteins RNA-binding motif protein 3 (RBM3) and cold inducible RNA-binding protein (CIRP) that may underlay the ion channel remodeling. Unexpectedly, we found that female sex is a protective factor in the progression of CKD and its cardiac sequelae.

Conclusion:

Our data provide novel insights into the association between CKD and pathologic proarrhythmic cardiac remodeling. Cardiac cellular stress response pathways represent potential targets for pharmacologic intervention for CKD-induced heart rhythm disorders.

Graphical Abstract

Introduction

Renocardiac syndrome (RCS) is a heterogeneous set of disease states in which renal dysfunction induces or exacerbates cardiac dysfunction¹. Indeed, individuals with non-dialysis-dependent chronic kidney disease (CKD) exhibit a 10-fold risk of cardiovascular (CV) mortality when compared to the age-matched general population, and dialysis-dependent patients are at a further increased risk.² In particular, mounting evidence shows a strong connection between CKD and sudden cardiac death (SCD), which may be due in part to the association between CKD and heart rhythm disorders, including atrial fibrillation, long QT syndrome, and ventricular arrhythmias^3–8. Potential mechanisms linking impaired renal function to cardiac arrhythmogenesis include blood electrolyte disturbances resulting from dysregulation of electrolyte handling in the renal tubules and cardiac remodeling due to circulating uremic toxins, systemic inflammation, and sympathetic overactivity⁹. However, the precise molecular mediators and substrates underlying SCD in CKD are largely unknown. Given that SCD accounts for up to 60 percent of CV-related deaths in CKD⁸, increasing our understanding of its pathogenesis at a molecular level may provide important insights into the development of new therapies to benefit the more than 1 billion people suffering from CKD worldwide¹⁰.

In this study, we sought to further characterize the effect of CKD on cardiac conduction and identify potential molecular drivers of CKD-induced heart rhythm dysfunction using a well-established murine model of CKD which induces renal failure through the addition of adenine to the animals’ diet¹¹. Previous studies have shown that adenine fed mice exhibit several important markers of renal pathology, including tubulointerstitial fibrosis, decreased glomerular filtration rate, increased serum creatinine, and increased blood urea nitrogen (BUN) levels^11–13. Furthermore, this model has been shown to recapitulate several features of RCS, including cardiac hypertrophy and fibrosis, reduced ejection fraction, and alterations in blood electrolyte levels important to cardiac conduction and myocyte function^{5, 14, 15}. However, no studies to our knowledge have examined the utility of the adenine mouse model of CKD for testing the association between renal dysfunction and cardiac electrophysiology at a global, cellular, and molecular level. Here we establish adenine-induced CKD as a useful model for studying cardiac electrophysiology in CKD and begin to explore the molecular underpinnings of arrhythmogenesis in adenine-induced CKD using cardiac global transcriptional profiling.

Methods

The data, methods used in the analysis, and materials used to conduct the research are available from the corresponding author upon reasonable request.

Murine CKD Models

To induce CKD, 8-week-old male and female C57BL/6 wildtype mice (Jackson Laboratories) were fed a standard chow diet (Teklad Global 18% Protein Rodent Diet, Envigo) supplemented with 0.25% adenine (formulated by Research Diets) alternating with standard chow for 6 weeks (Figure 1A) with the goal of sustaining median BUN between 40 and 70 mg/dL (Supplementary Figure 1C) and limiting median body weight loss to 20% of baseline (Supplementary Figure 1B) This protocol is referred to as the high-dose regimen. Control animals were fed standard chow for 6 weeks. To complement the high-dose regimen, we also established a low-dose regimen, in which 8-week-old male and female C57BL/6 wildtype mice (Jackson Laboratories) were fed a standard chow diet (Teklad Global 18% Protein Rodent Diet, Envigo) supplemented with 0.15% adenine (formulated by Research Diets) continuously for 16 weeks. Control animals were fed standard chow for 16 weeks. All animal studies were performed in accordance with institutional guidelines.

Figure 1:

Dietary adenine induces chronic kidney disease (CKD) in C57BL/6 mice. (A) Schematic representing the timeline of dietary adenine exposure for CKD mice in high-dose regimen over 6-week experiment. (B) Representative images for whole left kidneys (mid-coronal slices) and renal cortex stained with H&E and representative images of renal cortex stained with Masson’s Trichrome for control and CKD mice in high-dose regimen. (C) Quantification of fibrosis in control (n=8) and CKD (n=8) kidneys exposed to high-dose regimen using percent of area stained with Masson’s Trichrome. (D) Blood urea nitrogen (BUN) in control and CKD mice for high- (0.25%) and low-dose (0.15%) regimens (n=19 control and 26 CKD mice for high-dose regimen; n=14 control and 14 CKD mice for low-dose regimen). (E) Whole blood chemistries, including sodium (Na), potassium (K), ionized calcium (iCa), and hemoglobin (Hgb), in control (n=15) and CKD mice (n=14) for high-dose regimen. Box-and-whisker plots show the median, first and third quartiles, minimum, and maximum values; *P<0.05; **P<10⁻²; ***P<10⁻³; ****P<10⁻⁴ for CKD versus control using the Mann-Whitney U test for non-parametric comparison of unpaired samples.

Measurement of blood chemistries

Blood urea nitrogen (BUN) levels were measured in serum collected from study animals using the QuantiChrom Urea Assay Kit from BioAssay Systems (DIUR-100) according to the manufacturers standard protocol. Hemoglobin and electrolyte levels were measured in whole blood using the i-STAT 1 Blood Analyzer with the Chem8+ Cartridge from Abbott (09P31–26) according to the manufacturers standard protocol.

Quantification of fibrosis

Mid-coronal sections from adult mouse kidneys as well as short- and long-axis sections of adult mouse hearts were stained with Masson’s Trichrome and percent of stained area was calculated using the Automated Fibrosis Analysis Toolkit¹⁶ in Python (v3 10).

Electrocardiogram measurements

Electrocardiogram (ECG) measurements were acquired in awake study animals using the emka TECHNOLOGIES ecgTUNNEL for rodents. ECG intervals were calculated in an unbiased fashion across 50 contiguous beats using LabChart 7 Pro version 7.3.1 (ADInstruments, Inc) software set to Mouse ECG parameters.

Telemetry experiments

DSI PhysioTel^® (Harvard Bioscience, Inc) transmitters were surgically implanted into 7-week-old male and female C57BL/6 wildtype mice (Jackson Laboratories). The mice were given 1 week to recover from the procedure prior to initiating the high-dose dietary adenine regimen as described above. 24-hour telemetric ECG recordings were captured once per week for the first 4 weeks of the study after which recordings were made twice per week through the duration of study. Poincaré plots were generated for each animal across 12-hour light and dark cycles using 500 contiguous beats per animal per time point. ECG parameters were calculated in an unbiased fashion across a minimum of 200 contiguous beats per time point using LabChart 7 Pro version 7.3.1 (ADInstruments, Inc) software using Mouse ECG settings.

RNA sequencing and RT-qPCR

RNA was isolated from adult mouse ventricles using Qiagen RNeasy Plus Mini Kit according to manufacturer’s protocol. RNA-seq experiments were performed at the NYU School of Medicine Genome Technology Center. Samples were sequenced as 50 bp paired-end reads at 10 million to 20 million reads per replicate on an Illumina HiSeq2500 instrument Quality control (QC) was assessed using FastQC and MultiQC software (http://www.bioinformatics.babraham.ac.uk/projects/fastqc)^{17, 18}. Low-quality base noise in the first 10 basepairs of the reads was trimmed using FASTX-Toolkit fastx_trimmer (v 0.0.13, http://hannonlab.cshl.edu/fastx_toolkit/). All the reads were mapped to the mouse reference genome (mm10) using the STAR aligner (v2.6.1d)¹⁹. The read count tables were generated using featureCounts (Subread, v1.6.3)²⁰ and normalized based on their library size factors using DESeq2 (v1.30.1)²¹ and differential expression analysis was performed. Differential expression analysis on CKD versus control mice was also performed with DESeq2 using the default Wald test and accounting for differences based on sex (design: ~ sex + group). Multiple testing correction was done on DESeq2 using the Benjamini and Hochberg method by default. For further quantification of transcript levels, RNA samples were reverse-transcribed to cDNA using the Maxima First Strand cNDA Synthesis Kit (Thermo Fisher Scientific). qPCR was performed using the PowerSYBR© Green PCR Master Mix (Applied Biosystems) on the StepOne Real-Time PCR System (Applied Biosystems). qPCR probes were purchased from Origene and experiments were performed according to the manufacturer’s instructions.

Western blot antibody reagents

The antibody target, dilution, species, company and product number used for Western blot analysis are as follows: Cirbp 1:500 (rabbit, Proteintech, 10209–2-AP); GAPDH 1:2000 (mouse, Invitrogen, AM4300); Vinculin 1:1000 (mouse, Abcam, ab130007). Secondary antibodies were goat anti-rabbit (LI-COR IRDye 800CW, 926–32211) and goat anti-mouse (LI-COR IRDye 680RD, 926–68070).

Western blot analysis

Adult mouse ventricles were homogenized in RIPA lysis buffer with phosphatase and protease inhibitors (Thermo Scientific) and then incubated in the same buffer on ice for 1 hour. The samples were then centrifuged and the supernatant (whole protein lysate) was collected. Samples were run on pre-cast AnyKD) Mini-PROTEAN TGX gels (Bio-Rad), and then transferred to nitroceullulose (Bio-Rad) overnight at 4 °C. Nitroceullulose membranes were incubated in blocking buffer consisting of 5% nonfat dry milk in TBST for 1 hour. Membranes were then incubated with specific primary antibodies diluted in 5% nonfat dry milk in TBST overnight at 4°C followed by wash steps and secondary antibodies. Antigen complexes were visualized and quantified using the Odyssey Imaging System (Li-Cor).

Immunohistochemistry

Tissues sections were deparaffinize in 2 changes of xylene (10 minutes each) then hydrate in 2 changes of 100% ethanol (5 minutes each), 590% ethanol (5 minute),70% ethanol (5 minute). Antigen Retrieval was performed by steaming in 1x Citrate Buffer (10x Citrate Buffer, sigma C9999) for 30 min. Endogenous peroxidase activities were blocked by quenching the tissue sections with 3% Hydrogen Peroxide in Methanol for 10 min. Red Blood Cell autofluorescence was quenched by incubation of TruBlack in 70% Ethanol (Biotium 23007). Antibody staining was done by blocking the tissues (10% Normal Donkey Serum, 0.1%Triton in PBS) for 1 hour at Room Temp, followed by Cntn2 antibody (R&D AF4439) incubation overnight at 4°C. Slides were incubated with AlexaFluor 488 Donkey anti-goat secondary (Fisher Scientific A-11055) for 1 hour at room temperature, then mount with VECTASHIELD^® PLUS Mounting Medium with DAPI (Vector Labs H-2000-10). Images were acquired with Confocal Leica SP5.

Optical mapping

Mice were heparinized (500 U/kg) and euthanized by CO₂ inhalation followed by cervical dislocation. Hearts were quickly removed through a midline sternotomy and rinsed in a modified Tyrode’s solution containing (in mM): NaCl 130, NaHCO 24, KH₂PO₄ 1.2, MgCl₂ 1.0, glucose 11.1, KCl 4.7, and CaCl₂ 1.8, equilibrated with a 95% O₂–5% CO₂ gas mixture. Hearts were rapidly cannulated and perfused with a constant pressure (50–60 mmHg) in a retrograde fashion via an aortic cannula with warm (37–39 °C) oxygenated modified Tyrode’s solution. Once connected to the Langendorff perfusion system, hearts were immersed in modified oxygenated Tyrode’s in a jacketed perfusion chamber where the temperature was controlled (37–39 °C) to ensure the absence of transmural temperature gradients^{22, 23}. The excitation–contraction uncoupler Blebbistatin (Enzo Life Sciences,13.75 μM/L) was added to the perfusate to limit motion artifacts during optical recordings. Voltage-dependent fluorescent signals were recorded using a modified microscope (MVX10 Olympus) equipped for epifluorescent illumination. A 530 nm mounted LED (ThorLabs) was used for excitation, combined with a 593 ± 20 nm band pass emission filter. Images were acquired with a CMOS camera (SciMedia MiCAM ULTIMA) at 1000 frames per second with 14-bit resolution from a 100 × 100-pixel array. Hearts were loaded with the Voltage indicator Di-4-Anepps as previously described²². Images were acquired from the RV/LV junction on the anterior surface while pacing at 100 ms BCL. Baseline fluorescence images were acquired before dye loading and subtracted from the movies prior to data analysis. Activation maps for voltage transients were generated using custom software. Movies were signal averaged to improve signal-to-noise ratio and pixels with low signal-to-noise ratio were excluded from analysis. APD was determined on a pixel-by-pixel basis from the time of 50% maximum fluorescence during the rising phase to the time point of 70% recovery to its original baseline.

Cardiomyocyte dissociation

Murine ventricular myocytes were obtained by enzymatic dissociation following standard procedures²⁴. Briefly, mice were heparinized (500 U/kg) 20 min before heart excision and anaesthetized by inhalation of CO₂. Deep anesthesia was confirmed by lack of response to otherwise painful stimuli. Hearts were quickly removed from the chest and placed in a Langendorff column. The isolated hearts were perfused sequentially with Ca²⁺-free solution containing (in mm): NaCl 113, KCl 4.7, MgSO₄ 1.2, Na₂HPO₄ 0.6, KH₂PO₄ 0.6, NaHCO₃ 12, KHCO₃ 10, HEPES 10 and taurine 30 (pH 7.45 with NaOH) and an enzyme solution (collagenase type II; Worthington, Lakewood, NJ, USA). Ventricles were cut into small pieces, and gently minced with a Pasteur pipette. The Ca²⁺ concentration was increased gradually to 1.0 mM. Cardiomyocytes were kept in Tyrode’s solution containing (in mm): NaCl 148, KCl 5.4, MgCl₂ 1 0, CaCl₂ 1.0, NaH₂PO₄ 0.4, HEPES 15 and glucose 5.5; pH 7.4. Cells were used for patch clamp experiments on the same day of isolation.

Measurement of depolarization-activated outward potassium currents

To measure the whole cell depolarization-activated outward potassium (Kv) currents at room temperature, the recording pipette solution contained (in mmol/l): KCl 135, MgCl₂ 1, EGTA 10, HEPES 10, and glucose 5, pH 7.2 with KOH The bath solution contained (in mmol/l): NaCl 136, KCl 4, CaCl₂ 1, MgCl₂ 2, CdCl₂ 0.2, HEPES 10, tetrodotoxin 0.04 and glucose 10, pH 7.4 with NaOH Kv currents were recorded in response to 4.5 s voltage steps to potentials between −40 and +70 mV from a holding potential (HP) of −70 mV; voltage steps were presented in 10 mV increments at 15 s intervals Peak Kv currents at each test potential were defined as the maximal outward current recorded during the 4.5 s voltage steps The decay phases of currents at +50 mV were described by the sum of two exponentials: y(t) = A₁ * exp(−t/t₁) + A₂ exp(−t/t₂) + B, where t is the time, t₁ and t₂ are the decay time constants, A₁ and A₂ are the amplitudes of the inactivating current components (I_to,fast and I_to,slow), and B is the amplitude of the non-inactivating current component, I_ss. Fitting residuals and correlation coefficients were determined to assess the quality of fits.

Measurement of sodium currents

To measure the whole cell sodium current (I_Na) at room temperature, the recording pipette solution contained (in mmol/l): NaCl 5, CsF 135, EGTA 10, MgATP 5 and HEPES 5, pH 7.2 with CsOH. The bath solution contained (in mmol/l): NaCl 5, CsCl 112.5, TEACl 20, CdCl₂ 0.1, MgCl₂ 1, CaCl₂ 1, HEPES 20, glucose 11, pH 7.4, with CsOH. I_Na recordings were conducted by holding the cell at −120 mV followed by stepping to voltages between −90 and +20 mV in 5 mV steps for 300 ms with 3 s interpulse intervals. All recordings were obtained three times to verify reproducibility and within 15 minutes after establishing whole-cell configuration using an Axon multiclamp 700B Amplifier coupled to a pClamp system (v10 2, Axon Instruments).

Statistics

Endpoints were compared in RStudio environment (v1 3 959) (http://www.r-project.org/) using appropriate statistical tests as indicated in the figure legends. P or adjusted P<0.05 was considered statistically significant.

Results

Dietary adenine induces CKD in C57BL/6 mice

Previous studies of dietary adenine-induced CKD in C57BL/6 mice report a number of different dosing regimens leading to renal dysfunction of varying severity.^{11, 15, 25–27}. Initially, we selected an experimental diet with an adenine concentration of 0.25%, which has been reported to induce renal dysfunction within 2 weeks.^{25, 27}. In our pilot experiment, we observed significant renal dysfunction among CKD mice as measured by blood urea nitrogen (BUN) at 4 weeks; however, we also observed a median body weight loss of 31.6% among CKD mice (Supplementary Figure 1A). Although weight loss is a known limitation of the dietary adenine mouse model of CKD,^{11, 15, 25} we sought to identify a treatment regimen that would result in significant renal dysfunction while minimizing weight loss given its potential as a confounder.

We therefore developed a new regimen of intermittent 0.25% dietary adenine exposure by trending BUN and body weight measurements to achieve significant renal dysfunction while limiting weight loss (Figure 1A; Supplementary Figure 1B–C). Over the course of 6 weeks, our regimen resulted in significant structural damage to the kidneys of CKD mice, including diminished kidney size with tubular adenine crystal formation as well as tubular dilation, diffuse interstitial immune cell infiltrate, and diffuse interstitial fibrosis (Figure 1B–C), pathologies analogous to those observed in humans with CKD²⁸. We also observed significant renal dysfunction comparable to humans with CKD in the form of elevated BUN, electrolyte disturbances, and anemia (Figure 1D–E), while attaining a considerably more modest median body weight decrease of 11.1% (Supplementary Figure 1D).

In parallel with our experiments using the intermittent 0.25% dietary adenine regimen illustrated in Figure 1A, we adopted a more indolent version of the model, first reported by Kieswich et al., which uses continuous exposure to a low-dose 0.15% adenine diet to further limit weight loss while still resulting in significant renal and cardiac dysfunction over a much longer study period¹⁵. Indeed, we found that continuous exposure to the 0.15% adenine diet resulted in significant renal dysfunction with elevated BUN after 16 weeks (Figure 1D), while allowing for an even more favorable median body weight change of +4.6% (r Figure 1D). Throughout the remainder of the paper, we refer to our 6-week intermittent 0.25% dietary adenine regimen as the “high-dose regimen” and the 16-week continuous 0.15% dietary adenine regimen as the “low-dose regimen.” Replicating our results across these two dietary regimens gave us more confidence that the CKD-associated changes in cardiac physiology reported here were due to renal failure, rather than a primary effect of weight loss or toxin exposure.

CKD mice display functional deficits in cardiac electrophysiology

Following validation of the high- and low-dose dietary adenine regimens, we sought to identify functional changes in cardiac electrophysiology associated with CKD. Towards this goal, we compared awake electrocardiogram (ECG) recordings in control and CKD animals and found that, consistent with ECG data from human studies^{5, 29}, CKD mice exhibited significant bradycardia for the high-dose regimen and QT prolongation across both the high- and low-dose regimens (Figure 2A–C). To further characterize the cardiac conduction phenotype of CKD mice independent of blood electrolyte disturbances (Figure 1D) and autonomic effects, we isolated hearts from control and CKD animals and performed optical mapping using a voltage-sensitive dye.

Figure 2:

CKD mice exhibit ECG changes and prolonged action potential duration (APD). (A) Representative ECG tracings for control and CKD mice. (B) RR interval in control and CKD mice for high- (0.25%) and low-dose (0.15%) regimens (n=31 control and 32 CKD mice for high-dose regimen; n=14 control and 14 CKD mice for low-dose regimen). (C) QT interval in control and CKD mice for high- and low-dose regimens (n=31 control and 32 CKD mice for high-dose regimen; n=14 control and 14 CKD mice for low-dose regimen). (D) RR interval in control (n=14) and CKD (n=14) isolated-perfused hearts for high-dose regimen. (E) Representative APD₇₀ maps and quantification of APD₇₀ in control (n=14) and CKD (n=14) mice for high-dose regimen. (F) Representative activation maps and quantification of conduction velocity (CV) in control (n=14) and CKD (n=14) mice for high-dose regimen. (G) RR interval in control (n=14) and CKD (n=14) isolated-perfused hearts for high-dose regimen. Box-and-whisker plots show the median, first and third quartiles, minimum, and maximum values; *P<0.05; **P<10⁻²; ***P<10⁻³; ****P<10⁻⁴ for CKD versus control using the Mann-Whitney U test for non-parametric comparison of unpaired samples.

Consistent with our in vivo ECG recordings, we found that CKD hearts displayed significant bradycardia (Figure 2D). In addition, we observed a significant increase in action potential duration (APD) in CKD hearts, with no change in conduction velocity (CV) or atrioventricular (AV) conduction delay (Figure 2E–G), suggesting that intrinsic cardiac electrical remodeling is likely to play a role in the CKD-associated delay in cardiac repolarization observed on ECG.

Next, we sought to identify functional deficits in cardiac ion channels that might explain the ECG and optical mapping phenotypes observed in CKD mice. Using whole cell patch-clamp recordings of ventricular myocytes isolated from control and CKD mice, we found that CKD was associated with significantly increased sodium current (I_Na) density (Figure 3 A–B) with no significant difference in steady state inactivation (Figure C–D). In addition, we found that the current-voltage (IV) relationship for the transient outward potassium current (I_to) was significantly altered in CKD mice, primarily due to a decrease in the slow inactivating component of I_to (Figure 3E–F). Together, these abnormalities represent likely factors contributing to ventricular APD prolongation in CKD mice.

Figure 3:

CKD mice display functional disturbances in cardiac ion channels. (A) Current-voltage (I-V) relationship of I_Na in control versus CKD ventricular myocytes. (B) Comparison of peak I_Na density and maximum conductance in control versus CKD ventricular myocytes. (C) Steady state inactivation curve of I_Na in control versus CKD ventricular myocytes. (D) Comparison of voltage at I_Na half-inactivation in control versus CKD ventricular myocytes. (E) I-V relationship of Ito in control versus CKD ventricular myocytes. (F) Comparison of inactivating (I_{to, slow} and I_{to, fast}) and non-inactivating (I_ss) components of Ito in control versus CKD ventricular myocytes. Control n=33 cells and CKD n=27 cells from mice exposed to high dose regimen in panels A-D. Control n=28 cells and CKD n=30 cells from mice exposed to high-dose regimen in panels E-F. Box-and-whisker plots show the median, first and third quartiles, minimum, and maximum values; *P<0.05; **P<10⁻²; ***P<10⁻³; ****P<10⁻⁴ for CKD versus control using the Mann-Whitney U test for non-parametric comparison of unpaired samples.

CKD mice die from bradyarrhythmias progressing to asystole

Among the 36 CKD mice used to validate the high-dose regimen and characterize CKD-associated functional deficits in cardiac electrophysiology, 4 died prior to the completion of the 6-week experiment. Given the arrhythmogenic cardiac phenotype exhibited by these mice and the known association between CKD and SCD in humans, we hypothesized that these animals may be dying from lethal arrhythmias. To test this hypothesis, we performed continuous telemetric ECG recordings using surgically implanted devices in a small cohort of control (n=3) and CKD mice exposed to the high-dose regimen (n=3). Initial inspection of the tracings revealed that all CKD mice displayed extensive sinus arrhythmia. To better visualize the extent of this aberrant behavior, we generated Poincaré plots for each animal across light and dark cycles, which confirmed that in contrast to control mice, CKD animals exhibited marked variation in heart rate during both light and dark cycles (Figure 4A and Supplementary Figure 2). Furthermore, consistent with a recent prospective telemetric study of SCD in CKD patients²⁹, we found that all three of the CKD mice under telemetric observation experienced SCD on days 35–38 of the high-dose regimen from a similar mechanism, characterized by worsening sinus bradycardia progressing to third degree AV block and eventual asystole (Figure 4B–D).

Figure 4:

CKD mice die from bradyarrhythmias progressing to asystole. (A) Representative Poincaré plots across 12-hour light and dark cycles for control and CKD mice on day 32 of the high-dose regimen (≥3 days prior to SCD events for CKD mice); SD1 corresponds to the minor radius of each ellipse calculated as the standard deviation (SD) perpendicular to the line of identity; SD2 corresponds to the major radius of each ellipse calculated as the SD along the line of identity. (B) Kaplan-Meier survival curves for control (n=3) and CKD mice (n=3) undergoing telemetric ECG recordings plotted against days following initiation of control or high-dose dietary adenine regimen; P=0 025 using log-rank test. (C) Hourly mean HR in the 3 CKD mice spanning the 24 hours leading up to the sudden cardiac death (SCD) event for each animal. (D) Representative ECG tracings from CKD mouse in hours/minutes leading up to SCD; AV=atrioventricular; bpm=beats per minute; HR=hear rate; NSR=normal sinus rhythm; VEB=ventricular escape beat. Box-and-whisker plots show the median, first and third quartiles, minimum, and maximum values; *P<0.05; **P<10⁻²; ***P<10⁻³; ****P<10⁻⁴ for CKD versus control using the Mann-Whitney U test for non-parametric comparison of unpaired samples.

CKD is associated with significant changes in ventricular global gene expression

With the knowledge that dietary adenine-induced CKD in C57BL/6 mice produces a phenotype consistent with many of the abnormalities in cardiac conduction observed in humans with CKD, we sought to identify structural and transcriptional changes that might form the basis of cardiac arrhythmogenesis in these patients. Examining cardiac histology, we found no evidence of ventricular hypertrophy or fibrosis in CKD mice exposed to the high-dose regimen compared with controls on H&E and trichrome staining (Supplementary Figure 3A–B). When specifically examining the AV node and bundle of His, we also observed no evidence of increased fibrosis or gross structural damage in CKD mice (Supplementary Figure 3C). We then performed global transcriptional profiling of ventricles isolated from control and CKD mice exposed to the high-dose regimen using RNA-seq and found 378 differentially expressed genes (DEGs; adjusted p<0.05). Adopting an unbiased approach to selecting transcripts for future study, we sorted the list of DEGs by adjusted p-value and found that RNA-binding motif protein 3 (Rbm3) was significantly upregulated in CKD ventricles with the lowest adjusted p-value of any DEG in our dataset. Moreover, cold inducible RNA-binding protein (Cirp), a close homologue of Rbm3, was significantly upregulated as well (Figure 5A). Interestingly, we also observed that 13 of the 25 most significant DEGs in our sample were previously implicated in cellular response to hypoxia (Apln, Egln3, Eif4ebp1, Fmo2, Gnmt, Lrrc55, Mr1, Mycn, Nqo1, Nt5e, P4ha1, Prps1, and Trnp1; Figure 5A)^30–43. Of these 13 hypoxia-related genes, 10 shared a connection to the well-known hypoxia-inducible factor (HIF) pathway (Apln, Egln3, Eif4ebp1, Fmo2, Gnmt, Lrrc55, Mycn, Nqo1, Nt5e, and P4ha1)^{30–33, 35, 36, 38–40, 42}. A subset of these HIF-related transcripts are illustrated in Figure 5B.

Figure 5:

Global gene expression profiling reveals upregulation of cold-inducible RNA binding proteins RBM3 and CIRP in CKD mouse ventricles. (A) Gene expression profile in whole ventricles of CKD (n=8) mice compared with controls (n=8); blue dots represent Rbm3 and Cirp; red dots represent genes linked to cellular response to hypoxia in previous studies. (B) Pathway diagram including select hypoxia inducible factor (HIF)-related transcripts; red text highlights transcripts that are differentially expressed in CKD mice. (C) RT-qPCR comparison of expression levels of Rbm3 and Cirp transcripts in control versus CKD ventricles exposed to high (0 25%) and low-dose (0 15%) regimens (n=8 control and 8 CKD mice for high-dose experiment; n=10 control and 10 CKD mice for low-dose experiment). (D) Quantification of Western blot analysis comparing expression levels of CIRP in control (n=8) versus CKD ventricles (n=8) exposed to the low-dose regimen. Box-and-whisker plots show the median, first and third quartiles, minimum, and maximum values; *P<0.05; **P<10⁻²; ***P<10⁻³; ****P<10⁻⁴ for CKD versus control using the Mann-Whitney U test for non-parametric comparison of unpaired samples.

RBM3 and CIRP have known roles in physiologic stress response⁴⁴. Moreover, CIRP has been implicated in modulating ventricular repolarization via post-transcriptional regulation of I_to channel subunits Kv4.2 and Kv4.3⁴⁵, making it a particularly attractive candidate linking CKD to cardiac electrical remodeling. Accordingly, we next validated differential expression of Rbm3 and Cirp in CKD mouse ventricles using quantitative reverse transcription polymerase chain reaction (RT-qPCR). Consistent with the RNA-Seq data, we found significantly increased Rbm3 and a trend towards increased Cirp (p=0.052) in CKD ventricles exposed to the high-dose regimen as well as significantly increased Rbm3 and Cirp in CKD ventricles exposed to the low-dose regimen (Figure 5C). We also measured protein levels of CIRBP and RBM3 in whole cell lysates from control and CKD ventricles and found a significant increase in CIRBP in CKD mouse ventricles. We were unable to detect a signal for RBM3 in either control or CKD samples using available antibodies (Figure 5D).

Changes in gene expression and cardiac electrophysiology are correlated with the sex-linked severity of renal dysfunction

Interestingly, we observed significant heterogeneity in the treatment effect of dietary adenine for both renal and cardiac parameters. This allowed us to explicitly test the relationship between the severity of renal dysfunction and our cardiac phenotype. We accomplished this task by first selecting control and CKD animals exposed to either high- or low-dose regimens for whom we had collected data on renal dysfunction (i.e., BUN), cardiac electrophysiology (i.e., ECG), and ventricular gene expression all within the same animal. We then measured the correlation between BUN and cardiac parameters of interest, including RR interval and QT interval as well as Rbm3 and Cirp transcript levels. We found that BUN was significantly correlated to RR interval, QT interval, Rbm3, and Cirp, suggesting that the severity of cardiac dysfunction in CKD mice is directly related to the severity of renal dysfunction. We also found that the level of Rbm3 and Cirp expression was significantly correlated with our ECG findings, further suggesting a role for these proteins in regulating cardiac electrophysiology (Figure 6A).

Figure 6:

Changes in cardiac electrophysiology and gene expression in CKD mice are associated with sex-linked severity of renal dysfunction. (A) Correlation matrix showing Pearson’s correlation coefficient (r) for pairwise comparisons of BUN, RR interval, QT interval, Rbm3 expression level, and Cirp expression level for control (n=15) and CKD mice exposed to high- (n=8) or low-dose regimen (n=6); shading of circles corresponds to the magnitude of r; size of circles corresponds to the level of significance *P<0 05; **P<10–2; ***P<10–3; ****P<10–4. (B) BUN for control and CKD mice separated by sex for high (0 25%) and low-dose (0 15%) regimens (n=8 female control, 14 female CKD, 11 male control, and 12 male CKD mice in high-dose experiment; n =7 for all groups in low-dose experiment). (C) RR interval for control and CKD mice separated by sex for high and low-dose regimens (n=13 female control, 14 female CKD, 18 male control, and 18 male CKD mice for high-dose experiment; n =7 for all groups in low-dose experiment). (D) QT interval for control and CKD mice separated by sex for high and low-dose regimens (n=13 female control, 14 female CKD, 18 male control, and 18 male CKD mice for high-dose experiment; n =7 for all groups in low-dose experiment). (E) Ventricular APD₇₀ for control and CKD mice separated by sex in high-dose experiment (n=6 female control, 7 female CKD, 8 male control, and 7 male CKD mice). (F) Inactivating slow component of I_to in control versus CKD ventricular myocytes separated by sex in high-dose experiment (n=15 female control, 16 female CKD, 13 male control, and 14 male CKD cells). (G) Cirp and Rbm3 transcript levels in control and CKD ventricles separated by sex for low-dose experiment. Box-and-whisker plots show the median, first and third quartiles, minimum, and maximum values; *P<0 05; **P<10–2; ***P<10–3; ****P<10–4 for within sex comparisons of CKD versus control using Dunn’s Kruskal-Wallis multiple comparisons test; #P<0 05; ##P<10–2 for within treatment comparisons of male versus female using Dunn’s Kruskal-Wallis multiple comparisons test.

Next, we sought to identify factors that may be driving the heterogeneity in renal dysfunction and in turn influencing cardiac outcomes in our study. We could exclude many of the genetic and environmental risk factors traditionally associated with CKD as possible effect modifiers in our study given we used animals with homogeneous genetic backgrounds and tightly controlled laboratory conditions. We next considered sex as a potential modulator of treatment effect as several recent reports have delineated female sex as a protective factor in the progression of CKD to end-stage renal disease (ESRD) in humans and murine models^46–49. Indeed, when we re-analyzed our data examining treatment effects in males and females separately, we found the treatment effect of dietary adenine on renal and cardiac function was more pronounced in males across many of the variables we studied (Figure 6B–F), supporting the hypothesis that female sex is protective against CKD. Moreover, CKD-associated differential expression of Rbm3 and Cirp in ventricular tissue was more pronounced in males as well (Figure 6G).

Discussion

Dietary adenine-induced CKD models cardiac conduction deficits observed in humans with CKD

The association between CKD and abnormalities in heart rhythm, including sudden cardiac death is well known, however the mechanisms underlying these associations are less clear. Here we establish dietary adenine-induced CKD in C57BL/6 mice as a promising model for studying the effects of renal failure on cardiac electrophysiology. Similar to the surgery-based 5/6 nephrectomy mouse model⁵⁰ and the male Cy/+ rat model of CKD⁵¹, we found that mice with dietary adenine-induced CKD exhibit slowed ventricular repolarization in vivo and ex vivo independent of electrolyte abnormalities, as evidenced by QT interval prolongation on ECG and APD prolongation on optical mapping, respectively. Furthermore, we show that cardiac myocytes isolated from CKD mice exhibit functional changes in I_Na and I_to, consistent with the phenotype of prolonged cardiac repolarization, and that CKD mice experience SCD characterized by significant sinoatrial (SA) and AV node dysfunction as displayed by the progression of sinus bradycardia to complete AV block to eventual asystole on telemetric ECG recordings. Through an unbiased approach, we implicate cold-inducible RNA binding proteins RBM3 and CIRP as potential regulators of pathologic cardiac remodeling in the setting of CKD. Finally, we show that female sex is a protective factor for the progression of renal and cardiac dysfunction in the adenine mouse model of CKD.

The validity of dietary adenine induced CKD in mice as a model for studying the molecular mechanisms of heart rhythm disorders in CKD patients is supported by the observation that our findings parallel results from human studies. For example, recent clinical studies have highlighted the strong association between CKD and acquired long QT syndrome^{5, 52}. Sherif et al report a dose-response relationship between CKD staging and the prevalence of QT interval prolongation in humans⁵². Similarly, we show a significant correlation between BUN and QT prolongation in mice exposed to dietary adenine. Additionally, our finding that mice with CKD die from progressive bradycardia and asystole bears a striking resemblance to the outcome of a recent prospective telemetric study of SCD in patients undergoing outpatient hemodialysis which found that bradycardia progressing to asystole was the predominant mechanism of SCD in humans with CKD²⁹. Several other recent studies also indicate a clinically significant relationship between renal dysfunction and bradyarrythmias in humans, showing that CKD and ESRD patients undergoing continuous cardiac monitoring exhibit intermittent episodes of sinus bradycardia, AV node block, long pauses, and asystole.^53–55 Interestingly, although we observed progressive AV block in CKD mice during episodes of SCD, we found no increase in AV delay or apparent underlying structural changes to the AV node in isolated hearts from CKD animals, suggesting that fluctuations in blood electrolytes or autonomic tone may play a role in CKD-associated cardiac conduction abnormalities in vivo.

CKD drives differential expression of cellular stress pathways in mouse ventricles

While recent studies have connected the fibroblast growth factor (FGF) 23-Klotho axis to the cardiac repolarization abnormalities and ventricular arrhythmias observed in CKD^{50, 56}, our study suggests that stress inducible proteins, such as CIRP and perhaps RBM3, may also play significant roles in the regulation of cardiac electrophysiology⁵⁷. Rat knock-out (KO) models have shown that CIRP loss of function shortens ventricular APD through post-transcriptional regulation of I_to channel subunits Kv4.2 and Kv4.3⁴⁵. Our data, in which enhanced expression of CIRP is associated with lengthening of ventricular APD, are consistent with these findings. Moreover, CIRP has been implicated in the control of heart rate through its regulation of phosphodiesterase 4 (PDE4) mRNA in SA nodal cells and modulation of intracellular concentrations of cyclic adenosine monophosphate (cAMP)⁵⁸. While we did not directly measure CIRP or RBM3 abundance in SA nodal cells in this study, it is conceivable that induction of these two RNA-binding proteins in this restricted population of pacemaker cells contributes to CKD-associated bradycardia and sinus arrhythmia.

Previous studies have implicated cold inducible RNA binding proteins in a variety of systemic pathologies, including cancer, sepsis, hemorrhagic shock, ischemia-reperfusion injury, and aneurysmal disease,⁴⁵ but no studies to our knowledge have identified a role for CIRP and RBM3 in the pathophysiology of CKD. Mild hypoxia can be a potent stimulus for CIRP and RBM3 expression⁵⁹, and it is plausible that systemic or organ-level hypoxia resulting from anemia due to impaired renal erythropoietin (EPO) production in CKD triggers increased expression of CIRP and RBM3 in the heart. Indeed, results from our own study and others reveal significantly reduced blood hemoglobin levels in mice with dietary adenine-induced CKD²⁶. Interestingly, among the 25 most significant DEGs from our global transcriptional profile of CKD mouse ventricles, 13 have been linked to cellular response to hypoxia. This list of 13 genes includes myofibrillogenesis regulator-1 (Mr-1), which has previously been shown to regulate apoptosis in the setting of ischemia-reperfusion injury of cardiac myocytes,⁴¹ 5 potential upstream regulators of the HIF pathway (Egln3, Gnmt, Mycn, Nqo1, and P4ha1), and 7 downstream targets of HIF (Apln, Egln3, Eif4ebp1, Fmo2, Lrrc55, P4ha1, and Nt5e)^{30–33, 35, 36, 38–40, 42}. Although no studies have demonstrated a direct link between the HIF pathway and cardiac arrhythmias, several lines of experiments have revealed that the HIF pathway regulates EPO production and iron transport in the setting of CKD-associated hypoxia,⁶⁰ and HIF prolyl hydroxylase enzyme inhibitors, which have been shown to stimulate EPO production in patients with CKD by stabilizing the HIF complex, are currently under study in phase 2 and phase 3 clinical trials for therapeutic management of anemia in CKD⁶¹. As with the increased expression of CIRP and RBM3, the confluence of hypoxia-related genes at the top of our DEG list is consistent with the notion that tissue hypoxia may play a mechanistic role driving the pathologic transcriptional remodeling observed in this model of CKD.

In combination with hypoxia, uremic toxins and inflammatory cytokines circulating in the blood stream of CKD mice may also interact with cardiac RBM3 and CIRP to influence tissue remodeling. In particular, CIRP has been identified as an important mediator of systemic inflammatory response to sepsis and has even been shown to potentiate inflammation leading to tissue injury by acting extracellularly as a damage-associated molecular pattern molecule (DAMP) through the Toll-like receptor 4 (TLR4)-myeloid differentiation factor 2 (MD2) complex on macrophages⁶². In addition, circulating FGF-21, which increases drastically with worsening uremia,⁶³ has been implicated alongside β-klotho as a regulator of RBM3 in neuronal cell culture models and clinical studies of stroke,^64–66 hinting at a connection between the CKD-associated changes in RBM3 expression reported here and the FGF-Klotho axis, which is currently under study as a mediator of cardiac arrhythmogenesis in CKD^{50, 56}.

Female sex plays a protective role in the progression of CKD and its cardiac sequelae

Consistent with the sex-specific differences in renal and cardiac dysfunction observed in our study (Figure 6), previous literature in humans and animals has identified an important role for female sex as a protective factor in the progression of renal failure and CKD-associated CV mortality^{48, 67–70}. Although social factors including health care utilization and risk factor modification undoubtedly play a role in the sex-specific progression of CKD in humans, the biological mechanisms contributing to this discrepancy are not fully understood⁴⁶. Previous research has focused primarily on the effects of endogenous sex hormones on kidney function via various signaling cascades with estrogen demonstrating a possible protective effect on the progression of CKD⁷¹ and testosterone playing a potentially harmful role⁷². The relationship between sex hormones and heart rhythm disorders independent of CKD has also been a topic of recent investigation, and there is evidence to suggest that estrogen, progesterone, and testosterone may regulate cardiac action potential duration and propagation by exerting downstream effects on cardiac ion channels; however, the magnitude and nature of these effects as well as their underlying mechanisms remain controversial⁷³. In general, females exhibit faster ambient heart rates, faster AV nodal and ventricular conduction, and slower ventricular repolarization when compared to males⁷⁴. No studies to our knowledge have explicitly examined the interaction between sex, renal disease, and cardiac arrhythmogenesis on a molecular level. Our results suggest that sex-specific changes in CKD-associated cardiac gene expression may indeed play an important role in the protective effect of female sex on CV mortality in CKD.

Conclusion

In summary, our study establishes dietary adenine-induced CKD as a useful model for studying the molecular underpinnings of SCD in CKD with several important parallels to humans, including the development of prolonged QT interval and SCD by progressive bradyarrhythmias. Using an unbiased approach, we identify a potential role for cellular stress response proteins RBM3 and CIRP in the dysregulation of cardiac ion channels in response to CKD and implicate the HIF pathway as another possible mediator of CKD-associated cardiac remodeling. In addition, our patch-clamp and telemetric ECG data suggest a potential mechanistic relationship between cold inducible RNA binding proteins and cardiac sodium channels as well as a potential future line of inquiry examining the vulnerability of CKD patients to SA and AV node dysfunction, a concept which has been highlighted by several recent clinical case reports of the newly established BRASH (bradycardia, renal failure, AV nodal blockers, shock, and hyperkalemia) syndrome.^75–77 Finally, our results support the hypothesis that female sex plays a protective role in the progression of renocardiac syndrome on a molecular level. Further elicudating the mechanisms behind this effect will be an important strategy for identifying therapeutic targets for the treatment of CKD and its cardiac sequelae.

Limitations

The adenine mouse model of CKD has several important limitations. As with any animal model of disease that relies on exposure to an exogenous agent, we cannot rule out the possibility that the CKD-associated changes in cardiac physiology reported here were due in part to a cardiotoxic effect of adenine. However, the murine adenine model of CKD is well established as a method for studying the cardiovascular sequelae of renal disease and no direct cardiotoxic effect of adenine has previously been identified.⁷⁸ Additionally, the renotoxic effect of adenine relies on crystallization of adenine and its metabolite 2,8-dihydroxyadenine in renal tubules, a process which is readily visible on histology.⁷⁹ No analogous process has been observed in cardiac tissue from animals exposed to dietary adenine.^{15, 78} Moreover, we find a significant dose-response relationship between the cardiac deficits and the degree of renal failure we observe, and the changes we report in cardiac electrophysiology resemble changes seen in humans with CKD. A second potential limitation of this model is the body weight loss associated with dietary adenine. We address this issue by utilizing two separate dietary adenine regimens designed to minimize weight loss, including a low-dose regimen that results in an overall median net weight gain among dietary adenine-exposed animals. Our results replicate across these regimens and in some cases, show even greater precision for the low-dose regimen, suggesting that our findings are likely independent of body weight change.

What is known

• The burden of heart rhythm disorders and SCD among CKD patients is substantial.

• The molecular mechanisms of arrhythmogenesis in CKD are largely unknown.

What the study adds

• CKD results in functional deficits in cardiac ion channels leading to prolonged cardiac repolarization and increased susceptibility to SCD via progressive bradyarrhythmias.

• CKD induces pathologic cardiac remodeling involving cellular stress response pathways, which represent possible targets for pharmacologic treatment of heart rhythm disorders in CKD.

• Female sex protects against the progression of CKD and its cardiac sequelae by attenuating CKD-associated changes in cardiac electrophysiology and gene expression.

Non-standard Abbreviations and Acronyms

CKD

Chronic kidney disease

APD

Action potential duration

SCD

Sudden cardiac death

RCS

Renocardiac syndrome

BUN

Blood urea nitrogen

ECG

Electrocardiogram

DEG

Differentially expressed gene

HIF

Hypoxia-inducible factor

ESRD

End-stage renal disease