The most apparent phenotype in the Pkd2-KO hearts was hypertrophy and interstitial fibrosis. The Pkd2-KO left ventricular (LV) myocardial thickness (Figures 1A,B) and heart-to-body-weight ratio (Figure 1C) were significantly increased in Pkd2-KO compared to control WT hearts. The extensive diffuse and reactive interstitial fibrosis of LV were significantly higher in Pkd2-KO, compared to WT hearts (Figures 1D,E). The size of LV myocardiocytes was significantly larger in Pkd2-KO than WT hearts (Figures 1F,G), supporting the myocyte hypertrophy in Pkd2-KO hearts. The sinoatrial node (SAN) and atrioventricular node (AVN) were also studied to examine fibrosis in the cardiac conduction system in Pkd2-KO mice. Upregulation of fatty and fibrotic tissues was observed in SAN and surrounding tissues in Pkd2-KO mice (Figure 2A). Although significant fibrosis was detected in SAN, fibrosis was not statistically significant in AVN between Pkd2-KO and WT hearts (Figure 2B). Because no significant fibrosis was found in AVN, cell numbers were further analyzed (Figure 3). AVN was identified by both its position within the heart and localization of hyperpolarization-activated cyclic nucleotide-gated channel 4 (HCN4), which is known to localize in heart conduction systems. AVN significantly had a lower cell number in Pkd2-KO than WT hearts.

Because heart conduction systems including SAN and AVN have an important role in electro-cardiac propagation (28, 29), electrocardiac properties were further analyzed in Pkd2-KO mice. The heart rate based on the in vivo electrocardiogram (ECG) was 583 ± 5 beats per minute for the WT mice and 625 ± 10 beats per minute for Pkd2-KO mice; we did not observe an apparent irregularity in ECG (Figure 4A). In order to evaluate the electrical activity of the heart in the absence of neurohumoral effects, we obtained an ex vivo ECG from isolated hearts (Figure 4B). Without neurohumoral effects, WT hearts showed a regular rhythm with 125 ± 20 beats per minute, and the Pkd2-KO hearts had arrhythmic heart rate of 127 ± 22 beats per minute evidence from atrial fibrillation or atrioventricular block characterized by irregular R-R intervals. Cardiac arrhythmia was also evidence in Pkd2-KO hearts at higher preloads.

Based on these analyses, our studies indicated that the Pkd2-KO hearts were hypertrophied with interstitial fibrosis and abnormal SAN and AVN. The impact of cardiac and conduction system fibrosis might thus result in arrhythmogenic hearts in Pkd2-KO mice.

The heart-specific Pkd2 mutant knockout did not show any behavior issue, at least not within the end-point of our studies at 6-months of age. However, because these mice had cardiac hypertrophy and fibrosis (Figure 1), heart functions were evaluated independently from the autonomic neuronal system by analyzing the left ventricular pressure-volume relationship (PV loop). PV loop allows a more precise analysis of heart functions by plotting the changes in left ventricular pressure and volume during each cardiac cycle. In order to quantify the changes in heart functions during physiologic stresses, LV parameters in both WT and Pkd2-KO were further evaluated in response to adrenaline (4 μg/L) or diltiazem (0.08 μg/L; Figure 5A). As expected, while adrenaline increased heart rate and contractility via β₁-receptors, diltiazem had a negative inotropic and negative chronotropic effects in both WT and Pkd2-KO hearts (Supplementary Table 1). The end-systolic pressure volume relationship (ESPVR) is a relationship between LV pressure (LVP) and volume (LVV) at the end of the systole, and the ESPVR is considered a marker for LV systolic contractile function and elastance (30–32). Our results demonstrated that the LV ESPVR, stroke volume (SV), and ejection fraction (EF) didn't change significantly after adrenaline or diltiazem in Pkd2-KO hearts (Figure 5B), which could be a result of decreased heart muscle contractility due to fibrosis and systolic dysfunction of LV. Despite no changes in the ESPVR, SV, and EF in Pkd2-KO hearts, the cardiac output (CO) was significantly altered by adrenaline or diltiazem. Importantly, the CO in responses to the pharmacological agents was significantly less in Pkd2-KO than WT hearts primarily due to contractility dysfunction in Pkd2-KO hearts. The LV Pmax (maximum pressure) and LV ESP (end-systolic pressure) were higher in Pkd2-KO hearts before and after adrenaline and diltiazem, which could be due to LV fibrosis, stiffness, structural changes, and narrow LV chamber (Supplementary Table 1).

Preload is used to express EDV (end-diastolic volume); therefore, the higher the preload is, the greater the EDV is. EDV depends on left ventricular compliance during diastole. The 5, 10, and 15 mmHg preload pressures were therefore used to study EDV and other cardiac functions (Figure 6A). In Pkd2-KO mice, the EDV decreased and did not change with increasing preload due to abnormal LV expansion or compliance and filling impairment in the fibrotic hearts (Supplementary Table 2). Decreased EDV subsequently led to decrease in SV and CO significantly in Pkd2-KO compared to WT hearts (Figure 6B). Furthermore, dP/dt_min which is a valuable marker for diastolic function (25) was significantly decreased in Pkd2-KO than WT hearts, indicating Pkd2-KO hearts had a diastolic failure due to compliance and filling impairment. Finally, the ESPVR was significantly higher in Pkd2-KO than the WT hearts in all preloads (Figure 6B). The steep and leftward of ESPVR slope was most likely not due to the improved myocardial function, but it was due to a narrow chamber in the hypertrophic and fibrotic LV. In Pkd2-KO hearts, the EF could not compensate in response to different preloads due to abnormal contractility in the fibrotic LV. Higher preload did not affect ex vivo ECG in either rhythmic WT or arrhythmic Pkd2-KO hearts. The ex vivo heart rate was also not significantly altered by higher preloads in both mice (Figure 4B, Supplementary Table 1).

The Pkd1 mutant myocytes have been previously shown to have abnormal contractile calcium (16, 33). To investigate if it was also the case in our Pkd2 mutant, we used lentivirus shRNA knockdown system to examine the effects of Pkd2 knockdown (KD) in contractility and calcium. The contractile efficacy in myocytes was calculated by measuring the changes in muscle displacement (cell shortening) for each myocyte beat. The infection and blocking efficiencies of different Pkd2 shRNAs were first verified in a kidney cell line and were further examined in primary cultured myocytes (Supplementary Figure 1). The Pkd2 shRNA-D was selected and used, because it showed the highest knockdown efficiency. The myocyte contractions and calcium oscillations in Pkd2 knockdown myocytes occurred more frequently than in control myocytes (Supplementary Figure 2; Supplementary Videos 1, 2). Further, the intracellular contractile calcium levels from Pkd2-KD myocytes were significantly lower than those from the control. The Pkd2-KD myocytes consistently and significantly showed a decrease in contractility. We also confirmed the knockdown studies in cardiomyocytes using a mouse model with complete Pkd2 knockout (Supplementary Figure 3; Supplementary Videos 3, 4). Mouse Pkd2 knockout cardiomyocytes were characterized by increased contractile oscillation frequency, decreased contractile calcium and dampened contractile strength.

The contractile calcium and contractility were abnormal in single Pkd2-KO myocytes (Supplementary Figures 2, 3), and the ESPVR and SV among others were functionally impaired in Pkd2-KO hearts (Figures 4–6). Together with the arrhythmogenic characterization of Pkd2-KO hearts (Figure 4), we speculated that interstitial and SAN fibrosis together with hypertrophy and abnormal contractile calcium could play important roles in Pkd2-KO hearts, leading to impaired cardiac functions (Figures 1, 2).

To examine if/how fibrosis pathways were activated, TGF-β₁ expression in the circulating plasma and heart itself was analyzed with ELISA. While no difference was observed in the plasma level of TGF-β₁, ELISA quantifications showed a significant increase of TGF-β₁ in Pkd2-KO compared to WT hearts (Figure 7A). ELISA result was further supported by the Western blot data, which showed a significant increase in TGF-β₁ expression level in Pkd2-KO compared to WT hearts (Figure 7B). We subsequently evaluated TGF-β₁ receptor expression in the WT and Pkd2-KO hearts. The expression level of TGF-β₁ receptor was significantly higher in Pkd2-KO than WT hearts (Figure 7C). To study the downstream effect of TGF-β₁, we also evaluated phosphorylated-Smad3 (pSmad3), total-Smad3 (tSmad3), and β-catenin expression levels in the hearts. Our results showed higher expressions of pSmad3, tSmad3, and β-catenin in WT compared to Pkd2-KO hearts (Figures 7D,E, Supplementary Figure 4). While pSmad3 and tSmad3 were higher in WT hearts, the fractionation study showed that pSmad3 and tSmad3 were localized in nuclear fraction of Pkd2-KO myocytes (Supplementary Figure 5).

To confirm the Western blot data, immunohistochemistry studies were performed to evaluate TGF-β₁ and TGF-β₁ receptor expressions. Consistent to our immunoblot results, expressions of both TGF-β₁ and TGF-β₁ receptors were significantly higher in Pkd2-KO compared to WT hearts (Figures 8A,B). The expression levels of pSmad3 and β-catenin were also analyzed with immunofluorescence studies. Consistent with our Western blot data, expressions of both pSmad3 and β-catenin in cytosol were significantly lower in Pkd2-KO compared to WT hearts (Figures 9A,B). The nucleus-to-cytosol ratio of both pSmad3 and β-catenin were significantly higher in Pkd2-KO than WT hearts, indicating greater the nuclear localizations of pSmad3 and β-catenin in Pkd2-KO hearts.

Because the Pkd2-KO hearts are characterized with a higher TGF-β₁ expression, we next analyzed a potential inflammation in the Pkd2-KO hearts. We investigated potential infiltration in heart tissues by macrophage M1 and M2, which is known to be a local source for TGF-β₁ (34). The immunofluorescence analyses revealed that macrophage M1 and M2 infiltration was significantly higher in the mouse Pkd2-KO compared to WT hearts (Figures 10A,B).

To evaluate myocadiac fibrosis and hypertrophy, heart tissues were collected and fixed in 10% formalin. The tissues were dehydrated in ethanol and xylene, embedded in liquid paraffin, and cut with a thickness of 5 μm. Cut sections were stained with Masson's trichrome kit (Cat# 25088-1; Polysciences, Inc.), and images were visualized and captured using KEYENCE BZ-X710 or Nikon Eclipse Ti microscope.

To examine protein localization and expression, the formalin-fixed tissues were de-paraffinized. After deparaffinization and dehydration of paraffin sections, heat-induced epitope retrieval was performed using a pressure cooker and sodium citrate buffer (10 mM sodium citrate, 0.05% tween-20, pH 6.0). Once boiled, slides were transferred from phosphate-buffered saline (PBS) to the sodium citrate buffer in pressure cooker for 10 min, and then cooled to room temperature for 30 min. To permeabilize the tissues, slides were washed with permeabilization buffer containing 0.3% triton-100 in PBS for 10 min. After permeabilization, slides were blocked with 5% bovine serum albumin and 0.1% triton-100 in PBS solution and subsequently processed to detect specific proteins. All washing steps were done three times with PBS-T (tween-20, 0.05%).

For cell surface staining and cell size measurement, tissue slides were incubated with wheat germ agglutinin (WGA) (1:1,000; Cat# FL-1021; Vectorlabs) for 20 min. To detect specialized conduction cells in the atrioventricular node, tissues were incubated with rat anti-HCN4 antibody (1:1,000; Cat# ab32675; Abcam, Inc.) overnight at 4°C in a humidified chamber followed with AlexaFluor-488 goat anti-rat fluorescence secondary antibody (1:500; Cat# ab6840; Abcam, Inc). To study fibrotic pathways, anti-β-catenin (1:50; Cat# sc-133240; Santa Cruz, Inc.) and anti-pSMAD3 (1:1,000; Cat# sc-517575; Santa Cruz, Inc) antibodies were used in the same manner as anti-HCN4 antibody followed with AlexaFluor-488 secondary goat anti-mouse antibody (1:1,000; Cat# ab150113; Abcam, Inc). To evaluate macrophages infiltration, tissues were stained with anti-NOS2 antibody (1:50; Cat# sc-7271; Santa Cruz, Inc), and anti-CD-86 (1:50; Cat# sc-52448; Santa Cruz, Inc) for macrophages M1detection, anti-CD163 antibody (1:50; sc-20066; Santa Cruz, Inc), and anti-CD-206 (1:100; Cat# ab91992; Signaling, Inc) antibodies for macrophages M2 detection, followed with horse anti-mouse antibody Texas Red (1:1,000; Cat# NC9634446; Thermo Fisher Scientific), and AlexaFluor-488 goat anti-rat fluorescence secondary antibody (1:500; Cat# ab6840; Abcam, Inc), AlexaFluor-488 secondary goat anti-mouse antibody (1:1,000; Cat# ab150113; Abcam, Inc), and AlexaFluor-488 goat anti-rabbit fluorescence secondary antibody (1:500; Cat# ab91992; Signaling, Inc). In all cases, we used 4′,6-diamidino-2-phenylindole (DAPI) as a nuclear binding dye. All images were taken with the confocal A1R⁺ Nikon microscope (Version 4.30).

To examine protein localization and expression, the formalin-fixed tissues were de-paraffinized. After deparaffinization and dehydration of paraffin sections, heat-induced epitope retrieval was performed using a pressure cooker and sodium citrate buffer (10 mM sodium citrate, 0.05% tween-20, pH 6.0). Once boiled, slides were transferred from PBS to the sodium citrate buffer in pressure cooker for 10 min. Slides were then cooled to room temperature for 30 min, and permeabilized with permeabilization buffer containing 0.3% triton-100 in PBS for 10 min. To block endogenous peroxidase activity, slides were incubated in 3% hydrogen peroxide for 10 min and blocked with animal-free blocking solution (Cat# 15019; Cell Signaling, Inc.). After blocking, slides were incubated overnight at 4°C with mouse monoclonal anti-TGF-β₁ (1:50; Cat# sc-130348; Santa Cruz, Inc.) or rabbit polyclonal anti-TGF-β₁ receptor (1:100; Cat# ab235178; Abcam, Inc.) and subsequently incubated with SignalStain Boost Detection Reagent (HRP mouse; Cat# 8125; or HRP Rabbit; Cat# 8114; Cell Signaling, Inc.) for 30 min at room temperature. Slides were then incubated for 2–10 min with SignalStain DAB (Cat# 8059; Cell Signaling, Inc.), immersed in distilled H₂O, stained with hematoxylin (Cat# 14166; Cell Signaling, Inc.) and mounted with coverslips. All washing steps were done three times with PBS-T (tween-20, 0.05%).

To evaluate protein expressions, proteins were extracted from the tissues using radioimmunoprecipitation assay (RIPA) lysis buffer (Cat# 89901, Thermo Fisher, Inc.) containing Halt protease inhibitor cocktail (Cat# 78425, Thermo Fisher, Inc.). Total protein was quantified using the Bradford assay. Extracted proteins (25–50 μg) were loaded into 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel and transferred onto nitrocellulose membrane afterward. For nuclear fractionation, immunoprecipitation with anti-Histone H3 was used and loaded to the gels. The membrane was then blocked with 5% dry milk and subsequently processed to detect specific proteins using a standard method.

We used the following primary antibodies: rabbit polyclonal anti-TGF-β₁ (1:1,000; Cat# ab92486; Abcam, Inc.), rabbit polyclonal anti-TGF-β₁ receptor (1:1,000; Cat# ab31013; Abcam, Inc.), mouse monoclonal anti-β-catenin (1:100; Cat# sc-133240; Santa Cruz, Inc.), rabbit monoclonal pSmad3 (1:1,000; Cat# 9520s; Cell Signaling, Inc.), rabbit monoclonal Smad3 (1:1,000; Cat# 9523s; Cell Signaling, Inc.), rabbit polyclonal histone H3 (1:1,000; Cat# 9715s; Cell Signaling, Inc.), and anti-GAPDH (1:100; Cat# sc-365062; Santa Cruz, Inc.). Except for anti-GAPDH antibody which was already tagged with horseradish peroxidase (HRP), HRP-tagged secondary antibodies were used accordingly, followed with chemiluminescent substrate (Cat# 34577; Thermo Fisher Scientific, Inc.). Protein detection was carried out with Bio-Rad imager (ChemiDoc™ XRS+ System with Image Lab™ Software; Cat# 1708265; Bio-Rad, Inc.).

Lentiviral vectors containing shRNA to Pkd2 knockdown (Origene; pGFP-C-shLenti clone ID: TL310397) were transfected into HEK293T cells. Viral supernatants were collected and pooled at 24- and 48-h post-transfection. Primary mouse myocytes were then incubated with pseudoviral particles and 8 μg/ml polybrene for 48 h prior to analysis. Transduction efficiency was observed via GFP reporter fluorescence, and Pkd2 knockdown was verified through Western blot analysis. The shRNA sequences used in our experiments are shown below.

To observe cytosolic calcium in cultured cardiomyocytes, cells were incubated with 10 μM fura-2AM (Invitrogen) and protected from light for 30 min. Calcium was measured by comparing fura-2 excitation at 340 nm (bound calcium) and 380 nm (free calcium). After each experiment, free cytosolic calcium was determined by perfusing ionomycin (10 μM final concentration) to obtain the maximum calcium signal followed by the addition of EGTA (2 mM final concentration) to observe the minimum calcium signal. All experiments were performed at 37°C in a heated microscope chamber.

To detect TGF-β₁ in the circulating plasma, blood was collected from a submandibular vein (cheek punch) using a heparinized tube. Within 30 min of collection, blood was centrifuged for 15 min at 1000 × g at 4 °C, and supernatant was collected for the ELISA assay. To detect TGF-β₁ in the heart, the heart tissues were collected and homogenized in RIPA buffer containing Halt protease inhibitor cocktail, and total protein was quantified with Bradford assay from a commercially available kit (Cat# PI23227; ThermoFisher Scientific, Inc.).

The TGF-β₁ level was measured using a commercially available mouse TGF-β₁ ELISA kit (Cat# LS-F5184, LSBio, Inc.). The assay was based on the sandwich ELISA method. Each well had been pre-coated with TGF-β₁ antibody. Tissue lysate samples were added to the wells, and the target antigen (TGF-β₁) was bound to the antibody. The unbounded proteins were washed away, and biotin-conjugated detection antibody was added, which bound to the captured antigen. Avidin-tagged HRP conjugate was added to bind with biotin. The TMB (3,3′,5,5′-tetramethylbenzidine) detection substrate was added to react with the HRP enzyme leading to color development, which was proportional to total TGF-β1 bound. A stop solution (sulfuric acid) was added to terminate the color development reaction, and the optical density of each well was measured at a wavelength of 450 nm.

To study isolated heart functions, the 6-month-old mouse was injected with heparin (100 units; IP) to prevent blood coagulation and anesthetized with ketamine 8–10 min later (200-350 mg/kg; IP). Mouse's chest was wiped clean with water, dried, and shaved to obtain in vivo ECG. The ultrasonic gel was warmed to 37°C to reduce stress to the mouse. The electrodes were then placed onto the mouse's chest. One electrode was placed around the xiphoid process of the left side of (the) sternum, and the other electrode at the 4th or 5th intercostal space on the right side. The ECG reading was taken at 5-s interval.

Immediately after obtaining in vivo ECG, an incision was made in the mid-abdomen toward the diaphragm. The diaphragm was opened, the thoracic cage was cut bilaterally, and the heart was dissected out. Immediately after the heart dissection, the aorta was cannulated and perfused with Krebs-Ringer solution (125 mM NaCl, 2.5 mM KCl, 1.25 mM NaH2PO4, 2 mM CaCl2, 1 mM MgCl₂, 25 mM NaHCO₃, and 25 mM glucose), which was continuously bubbled with carbogen (95% O₂ and 5% CO₂) to reach a pH of 7.4 at 38.0°C.

The ex-vivo perfusion of the mouse's heart was performed, the left atrium was cannulated, and cardiac function parameters were recorded using the working heart system (Supplementary Video 5; Emka Technology, Inc.). This approach provided quantitative measurements and heart functions that other methods cannot measure, including elastance, contractility, and stroke work. For evaluation of cardiac electrical activity in the absence of neurohumoral factors, an ex vivo ECG was obtained from the software after placing ECG electrodes on the right atrium and apex of the heart. Other cardiac parameters, including left ventricle pressure (LVP), left atrial pressure (LAP), aortic pressure (AP), and aortic flow (AoF) were obtained from the software. The preload and afterload were adjusted manually as needed.

The heart's responses to the external stresses were examined with a pharmacological stress test. The stress test was performed by using epinephrine (4 μg/L) or diltiazem (0.08 μg/L). End-systolic pressure-volume relationship (ESPVR), left ventricle pressure maximum (LVP max), and end-systolic pressure (ESP), left ventricle end-systolic volume (LV ESV), stroke volume (SV), and ejection fraction (EF) were obtained from the software or pressure-volume analysis. In some cases, preload was adjusted from 5 mmHg to either 10 or 15 mmHg. We used preload to indicate end-diastolic volume (EDV); therefore, the higher the preload was, the greater the EDV was. Preload was changed manually on the preload reservoir graduated cylinder, located on the isolated working heart system. After increasing the preload, LV pressure max (LV Pmax), end-systolic pressure (ESP), end-diastolic pressure (EDP), end-systolic volume (ESV), end-diastolic volume (EDV), stroke volume (SV), and ejection fraction (EF) were obtained from the software data and pressure-volume analysis.

Image analyses were performed in Nikon NIS-Element for Advanced Research software (Version 4.51). This software was also used for image capture, segmentation, 3D object reconstruction, and automatic object recognition. A Photometric Coolsnap EZ CCD Monochrome Digital Camera was used to capture images with a 1392 × 1040 imaging array to resolve fine details of the images. Binary masking was used to calculate cell size, fibrosis, ventricular thickness, and image intensity for protein expression and localization based on granularity, shape, size, and pixel intensity. Quantitation of Western blots was done with the NIH Fiji ImageJ (Version 2.1). After a box of the intended proteins was drawn using the “gel” analysis function, the area under the curve was measured for the intensity of each protein band. All images were finalized on a six-core Mac Pro, 3.9 GHz to facilitate complete data extraction. Scale bars are provided in all figures and videos to indicate the actual image size.

A list of heart functions was measured or calculated using Emka Technology software. The software captured and recorded real-time data for electrical heart propagation, cardiac contractility, heart rate, left ventricle pressure and volume, preload, afterload, aortic pressure, and aortic outflow. These data were further used to calculate stroke volume, cardiac output, end-diastolic/systolic volume, and left atrial pressure rise and fall. Additional analyses were performed in Microsoft Excel for macOS (Version 16.48).

Our statistical analyses were performed using GraphPad Prism for macOS (Version 9.1). To compare two groups of unpaired datasets, a non-parametric student t-test was used with two-tail assuming no Gaussian distribution. For comparison within 3 or more groups with no matching datasets, we used non-parametric or mixed ordinary one-way ANOVA. The mean of each group was then compared with the mean of every other group using Tukey post-hoc multiple comparison test with multiplicity adjusted P-value for each comparison. P < 0.05 was considered significant in our studies. A more precise P-value was reported separately in the graphs. The total measurements and N values were also reported independently in each figure legend.