Circulating Extracellular Vesicles in the Pathogenesis of Heart Failure in Patients with Chronic Kidney Disease

Abstract

Background:

Cardiovascular disease (CVD) causes more than 50% of deaths in patients with advanced chronic kidney disease (CKD). Clinical studies suggest that kidney-derived factors contribute to CVD development in CKD, independently of co-morbidities. However, to date, no kidney-specific humoral risk factor that triggers direct cardiotoxicity has been identified. In this cross-sectional study, we investigate how, in CKD patients, circulating extracellular vesicles (EVs) facilitate pathological kidney–heart communication, thereby causing cardiotoxicity, impairing cardiac function, and contributing to heart failure (HF) progression.

Methods:

We investigated the function of EVs from CKD patients and adenine diet-induced CKD mice on cardiomyocyte and cardiac contractility. MiRNA cargo of EVs was identified by small RNA sequencing and qRT-PCR, and their cardiotoxicity was tested by using miRNA-mimics. Tissue and cellular origin of CKD-EV-miRNAs were determined from their corresponding primary miRNA expressions in mice.

Results:

EVs from plasma of CKD patients, but not from healthy controls, were cardiotoxic; they significantly induced apoptosis both in vitro and in vivo and impaired contractility of adult rat primary cardiomyocytes in vitro. Likewise, EVs from both plasma and kidneys of CKD mice were cardiotoxic. Pharmacologically depleting circulating EVs in CKD mice significantly recovered cardiac function and ameliorated HF, improvements that suggest CKD-EVs play a causal role in HF pathogenesis. Both human and mouse CKD-EVs were enriched in distinct miRNAs, compared to control-EVs. CKD-EV-miRNA mimics were cardiotoxic, impairing contractility and downregulating contractile gene expression in human-iPSC-derived cardiomyocytes. Interestingly, levels of endogenous primary miRNAs corresponding to circulating CKD-EV-miRNAs were significantly higher in CKD-kidney tissues, specifically in CD45^−veCD31^−ve renal cells, but not in CKD-hearts, CKD-livers or CKD-PBMCs from peripheral blood, a result that indicates that CKD-EV-miRNAs originate renally. Remarkably, CKD-EV-miRNA levels correlated with established markers of cardiac injury, thus uncovering the presence of sub-clinical heart disease and demonstrating heterogeneity in reno-cardiac disease.

Conclusion:

Collectively, our human subject and mouse studies show that circulating CKD-EVs, carrying distinct renal-derived miRNAs, mediate the molecular crosstalk that contributes to the pathogenesis of HF in CKD. Consequently, CKD-EVs hold promise as diagnostic and prognostic biomarkers for early disease detection and as targets for novel therapeutic interventions in chronic reno-cardiac disease.

INTRODUCTION

Chronic kidney disease (CKD) is increasingly recognized as a major global public health concern, affecting more than 800 million individuals and resulting in over 3 million deaths worldwide in 2019¹. While CKD is a direct cause of morbidity and mortality, it is also an important risk factor for cardiovascular disease (CVD)². Indeed, the degree of CVD correlates with CKD severity. In patients with early-stage CKD (Stages 1–2), CVD incidence is significantly higher than in the general population³. CVD mortality risk doubles and triples during CKD Stages 3 and 4, respectively. Moreover, CVD is the leading cause of death during end-stage CKD (Stage 5)^{3, 4}.

The relationship between heart failure (HF) and CKD is close and complex. A better understanding of HF pathophysiology is necessary to develop prevention and treatment strategies to reduce the high morbidity and mortality in CKD patients. Traditional risk factors, such as hypertension, hyperlipidemia, smoking, and diabetes, are prevalent in CKD patients; however, these comorbidities do not fully explain the elevated cardiovascular risk in CKD patients⁵. Pathophysiological dysfunctions associated with CKD can induce other nontraditional risk factors, such as inflammation, oxidative stress, and abnormal calcium-phosphorus metabolism, which may contribute to HF progression². Nevertheless, kidney-specific humoral risk factors that cause early functional and structural cardiac damage have yet to be determined, largely because populations with reno-cardiac disease remained understudied. Thus, there is an urgent need to identify novel mediators of HF onset in CKD patients.

Emerging evidence shows that circulating EVs facilitate long-distance cell-to-cell communication and mediate organ crosstalk⁶. The molecular signature of EVs mirrors the pathophysiological status of their originating cells, thus enabling the bioactive EV cargo to serve as a biomarker for early disease detection⁷. EVs are known to play a central role in the pathological dissemination of cancer⁸, cardiovascular disease⁹, and neurodegeneration¹⁰. Recently, CKD-derived EVs were shown to induce smooth muscle cell calcification¹¹ and vascular remodeling¹², and several EV miRNA signatures discovered in a rodent model of CKD have been linked to vascular calcification¹¹. Nonetheless, significant knowledge gaps remain regarding whether circulating CKD-EVs from kidneys directly affect cardiomyocyte (CMs) and cardiac function, as well as the mechanism by which CKD-EVs cause HF in CKD patients.

Our goal in this study was to characterize the molecular underpinnings of EV-mediated pathological communication from the kidney to the heart in CKD patients. To address this, we i) collected plasma samples from CKD patients and healthy controls and from a CKD mouse model of HF; ii) evaluated the direct cardiotoxic effects of circulating/kidney-derived CKD-EVs on human and rodent cardiomyocytes in vitro and mice hearts in vivo; iii) investigated the impact of CKD-EVs on cardiomyocyte and cardiac contractile functions both in vitro and in vivo; iv) used small-RNA sequencing to identify the miRNA cargo responsible for CKD-EV cardiotoxicity and validated the function of CKD-EV-miRNA risk factors in cardiomyocyte apoptosis and contractility; v) determined the renal and cellular origins of cardiotoxic CKD-EV-miRNAs. Our findings have the potential to define cardiotoxic CKD-EV contents as biomarkers in early diagnosis of and novel therapeutics for HF in CKD patients.

METHODS

Data supporting the findings of this study are available from the corresponding author upon reasonable request. Expanded methods are provided in the Supplemental Material.

Study Design

This study aimed to investigate the role of circulating EVs in the pathogenesis of HF in CKD patients. We first compared plasma EVs from CKD patients (hCKD-EVs) to plasma EVs from healthy controls (hCtr-EVs) to evaluate cardiotoxic function in cardiac cell death and contractile dysfunction, both in vitro and in vivo. To gain mechanistic insights and to define EV-mediated kidney-cardiac pathological communication, we used a mouse model of adenine diet-induced CKD that instigated HF with reduced cardiac function. Molecular cargo of hCKD-EVs was identified by small RNA sequencing and qPCR, and the top differentially expressed miRNAs were functionally quantified by using human AC16 cardiomyocytes (hAC16-CMs) and human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs). The renal and cellular origins of cardiotoxic CKD-EV-miRNAs were determined from the corresponding endogenous pri-miRNA expressions in different tissues and kidney cells in CKD or Ctr mice.

Human subjects

The study was approved by the institutional review boards of the Mount Sinai Hospital and University of Virginia (approved IRB at UVA: HSR#20550). The participants gave written informed consent. Blood samples were collected from patients with moderate or advanced CKD and healthy controls (Tables S1 and S2). History of CKD was defined based on patient’s medical history and hospital records. In the CKD patient cohort, participant inclusion used the following criteria: (i) outpatients at UVA Medical Center nephrology clinic from 2018 to 2025, (ii) age range >18 years and (iii) CKD diagnosed according to the Kidney Disease Improving Global Outcomes (KDGIO) guidelines¹³, based on cause, glomerular filtration rate (GFR) category (G1–G5), and albuminuria. The KDIGO defines CKD as abnormalities of kidney structure or function, present for a minimum of 3 months, with implications for health. Subjects with heart transplant were excluded. Blood from patients receiving hemodialysis was obtained prior to dialysis treatment. The following were exclusion criteria for all subjects: active infections, active malignancy, immunomodulatory therapy, bleeding disorder, pregnancy/nursing, sickle cell disease, or anemia. Healthy control subjects had no chronic conditions and were recruited based on clinical history and self-report. In this proof-of-concept study, our sample size was based on availability of patient and control samples rather than formal power calculations.

Statistical analysis

Groups were first assessed for normality to determine whether parametric or nonparametric tests would be used. Assumptions of normality were tested with Shapiro-Wilk test. Levene’s test was used for equality of variances. Normally distributed data with equal variances were statistically analyzed with either two-tailed Student’s t-test (two groups) or one-way ANOVA (three or more groups with one experimental factor) with Tukey multiple comparison test or Dunnett’s multiple comparison test. Normally distributed data with two experimental factors were analyzed with two-way ANOVA followed by pairwise comparisons with Tukey multiple comparison test. Non- normally distributed data were analyzed using Mann-Whitney U test (two groups) or Kruskal-Wallis test (three or more groups). Grubbs test was used to detect outliers. All data are shown either as the mean ± SEM (normally distributed data) or median ± interquartile range (not normally distributed data). For human subject characteristics, data are given as median with interquartile range for all. p-value is calculated using non-parametric Kruskal-Wallis test. To evaluate the association between miRNA expression and HF biomarkers, a Pearson correlation analysis on log-transformed data was performed. All data were graphed using Prism 9 (GraphPad software, version 9.5.0) or R and statistical analyses were performed in Prism 9 or R. The statistical details, sample size and significance levels for each experiment are specified in the figure legends.

RESULTS

Study Population

The functional and pathological roles of circulating EVs from CKD patients (hCKD-EVs) in reno-cardiac communication, cardiac contractility, remodeling and HF have not been studied. To investigate these dynamics, we recruited 35 well-characterized CKD patients with stable moderate or advanced chronic kidney disease (CKD stages 3–5) plus 18 comparable healthy controls (Tables S1 and S2). Compared to healthy controls, the CKD cohort had significantly lower median estimated glomerular filtration rate (eGFR) and significantly higher urinary albumin to creatinine ratio (UACR) levels, thus indicating impaired renal function. Demographics, patient characteristics, and other risk factors like obesity, uncontrolled hypertension, and diabetes mellitus did not significantly differ between the CKD and control groups. All CKD patients received evidence-based medical management adherent to established guidelines (GDMT), to ensure standardized therapeutic interventions across the cohort. All CKD patients were in a steady state, with no ongoing acute kidney injury (AKI). Higher medication utilization within the CKD cohort reflected the intense therapeutic regimen necessitated by their disease state.

Circulating EVs from CKD patients (hCKD-EVs) induce cardiomyocyte apoptosis and contractile dysfunction

We collected plasma samples from patients with moderate or advanced CKD (stages 3–5) and healthy controls. Because larger microvesicles (MV) in plasma have been shown to have platelet or erythrocyte origins^14,15, we focused on smaller EVs. To this end, small EVs were isolated from platelet-poor and MV-depleted human plasma by using ultracentrifugation, a precipitation or size-exclusion chromatography (SEC) method as previously described^16,17 (Figure 1A; Figure S1). Plasma EVs isolated from CKD patients (hCKD-EVs) versus healthy controls (hCtr-EVs) did not significantly differ in either size (~90nm v ~97nm respectively, determined using dynamic light scattering, DLS; Figure 1B) or concentration (quantified using nanoparticle tracking analysis, NTA; Figure S2A–S2C). Western blot determined the presence of EV-surface marker proteins Flotilin-1, CD63 and CD81, as well as EV intraluminal marker protein Alix (Figure 1C) and the absence of EV-negative marker GM130 (Figure S2D), thus confirming EV purity. Single EV phenotyping by nano-flow cytometry revealed known EV-surface tetraspanin marker proteins (CD9, 63 and 81) (Figure 1D), plus significantly increased number of tetraspanin^+ve hCKD-EVs compared to hCtr-EVs (Figure 1D; Figure S3), consistent with the literature^18,19. Cryo-electron microscopy (EM) (Figure 1E) showed the presence of EVs with known nano-size and morphology. Our EV characterization data confirmed successful isolation of EVs from human plasma of CKD patients and healthy controls.

Figure 1. Circulating hCKD-EVs are cardiotoxic in vitro and in vivo.

A, Study design for EV isolation and characterization. B, Dynamic light scattering (DLS) analysis of EV size. C, Western blotting of EV-positive markers (Alix, Flot1, CD81, CD63) for EVs (n=6). D, Nanoflow cytometry analysis of tetraspanins (CD9, CD63, and CD81) for EVs (n=3). E, Cryo- electron microscopy (EM) of human plasma EVs. F, Cell viability assay for hAC16-CMs treated with equal amounts of protein (20ug/ml) from EV-enriched or EV-depleted plasma after 48h (n=6–12). G, Study design for EV functional analysis in vitro. H and I, Representative image of Live/dead assay and TUNEL assay of hAC16-CMs treated with hCKD-EVs and hCtr-EVs. J and K, Quantification of Live/dead assay and TUNEL assay (n=5–12). L and M, IonOptix analysis showing calcium transients and contractility of rat primary cardiomyocytes treated with hCKD-EVs and hCtr-EVs (n=5–6). N, Study design for EV functional analysis in vivo. O, Intramyocardial injection (IMCI) of hCKD-EVs labeled with PKH67 dye showing EV uptake in NODScid mice heart. P, Representative image of TUNEL assay in NODScid hearts injected with hCKD-EVs and hCtr-EVs. Q, Quantification of TUNEL+ve cells in mice hearts injected with hCKD-EVs or hCtr-EVs (n=3–4). Data in panel D are presented as median with interquartile range and were analyzed with Mann-Whitney U test. Data in panel F, J, K, M, Q are presented as mean ± SEM and were analyzed using Student’s t test.

To investigate hCKD-EV cardiotoxicity, we separated platelet- and MV-free CKD plasma into EV-enriched or EV-depleted fractions, treated each fraction (20μg protein equivalent to 1.6e9 particles/100μL for EVs) with hAC16-CMs in culture and then compared cell viability between the fractions after 48h (Figure 1F). Interestingly, hCKD-EVs, but not the hCtr-EVs, hCKD-EV-depleted, or hCtr-EV-depleted fractions significantly decreased hAC16-CM viability in vitro (Figure 1F). These data suggest that circulating hCKD-EVs are a key cardiotoxicity mediator in CKD patient plasma.

Further characterization of hCKD-EVs’ cardiotoxic effects show significant increases in hAC16-CM death and TUNEL-positive apoptotic cells treated with hCKD-EVs compared to hCtr-EVs (Figure 1G –1K). To study if hCKD-EVs impact CM function, we treated adult rat primary CMs with hCKD-EVs or hCtr-EVs and quantified their contractile function and Ca²⁺ handling by using IonOptix Myocyte Calcium and Contractility System (Figure 1G). IonOptix analysis demonstrated that compared to hCtr-EVs, hCKD-EVs significantly impair contractile function and Ca²⁺ handling in adult primary CMs. (Figure 1L and 1M; Figure S4). Next, to evaluate hCKD-EVs cardiotoxicity in vivo, we administered hCKD-EVs or hCtr-EVs (100μg/8e9 particles) to the left ventricle (LV) of NOD-Scid mice via intramyocardial injection (Figure 1N). Troponin^+ve CMs internalized PKH67-labeled hCKD-EVs (Figure 1O). Administration of hCKD-EVs, compared to hCtr-EVs, significantly expanded TUNEL-positive apoptotic cells in the LV (Figure 1P and 1Q). Taken together, these data suggest that circulating plasma EVs from CKD patients detrimentally affect CM viability, contractile function, and Ca²⁺ handling.

Mouse adenine diet-induced CKD model led to heart failure

To investigate the role of kidney-derived EVs in the pathogenesis of CKD-induced HF, we used a previously reported adenine (Ade) diet-induced CKD mouse model of reno-cardiac syndrome²⁰. This model mimics the pathophysiology of human CKD-induced HF, with tubular and glomerular damage as well as interstitial fibrosis. However, varying amounts of adenine in diet generate different levels of cardiomyopathy²¹. To study the pathomechanism of reno-cardiac disease, we employed a 0.25% Adenine diet to create severe CKD, which was sufficient to induce cardiac dysfunction. Our CKD mouse model developed HF at 8w. We thoroughly characterized the HF phenotype by assessing cardiac function, hypertrophy, fibrosis, and biochemistry at 4w and 8w by performing Echocardiogram (Echo), Magnetic Resonance Imaging (MRI) and histology (Figure 2A).

Figure 2. Adenine-induced CKD mouse model shows HF and EVs from Ade-CKD mice are cardiotoxic.

A, Generation of CKD mouse model. B, Quantification of systolic blood pressure (BP) (n=9) and creatinine (n=4) in Ade-CKD (in red) and control (Ctr, in blue) mice at 8w. C and D, Magnetic Resonance Imaging (MRI) quantification of LV longitudinal strain and LV mass/LVEDV ratio at 8w (n=3–5). E, Echocardiographic quantification of % ejection fraction (EF) and % fractional shortening (FS) to assess cardiac function at 8w (n=10). F, Masson’s trichrome staining of LV showing fibrosis and Wheat Germ Agglutinin (WGA) staining showing hypertrophy in CKD vs. Ctr mice at 8w (n=6). G, Schematic of mouse plasma EV (mCtr/CKD-pEV) functional study. H, Representative calcium amplitude transients through one contraction-relaxation cycle of rat primary cardiomyocytes treated with mCtr/CKD-pEV as measured by IonOptix analysis. Quantification of calcium transients and contractility in primary cardiomyocytes treated with mCtr/CKD-pEV (n=3). I, Viability of H9C2 cells treated with equal protein amounts of Ctr/CKD kidney- (n=8–9) or liver- (n=3) derived EVs. J and K, Live/dead cell assay and TUNEL assay of H9C2 cells treated with mCtr-kEVs or mCKD-kEVs (n=4). L Schematic of IonOptix analysis of rat primary CMs treated with mCtr/CKD-kEVs. M, Representative calcium amplitude transient through one contraction-relaxation cycle, and quantification of calcium transients and contractility of primary CMs treated with mCtr/CKD-kEVs as measured by IonOptix analysis (n=4). N, Schematic of mCtr/CKD-kEVs functional assays in vivo. O, Representative image of TUNEL assay of C57BL/6J mouse hearts injected with equal protein from mCKD-kEVs and mCtr-kEVs. Quantification of TUNEL^+ve cells in mice hearts injected with mCtr/CKD-kEVs (n=3–4). Data are presented as mean ± SEM and were analyzed using Student’s t test.

CKD mice exhibited significantly higher plasma blood urea nitrogen (BUN) and creatinine level, increased blood pressure (BP), early signs of systolic dysfunction at 4 weeks (Figure S5), and established HF at 8 weeks (Figure 2B; Figure S6). Cardiac MRI revealed CKD mice had depressed LV systolic function with distinctly reduced LV longitudinal strain and mitral annular plane systolic excursion (MAPSE) (Figure 2C and 2D; Figure S6F). In agreement, echo analysis showed significantly worsened LV systolic dysfunction (quantified by LV ejection fraction and LV fractional shortening) in CKD mice (Figure 2E). Echo and MRI data also demonstrated substantial LV dilatation and hypertrophy with notably higher LV end-diastolic (LVEDV) and LV end-systolic volumes (LVESV, p=0.06); thicker LV anterior and posterior walls (LVAW and LVPW); indexed LV mass, septal, and lateral walls; expanded relative wall thickness (RWT), and right ventricle (RV) dysfunction in CKD mice (Figure S6D–S6I). Histological data confirmed adverse cardiac remodeling in CKD mice with significantly more fibrosis in both the kidney (Figure S6B) and heart (Figure S6C), and increased cardiomyocyte size/area indicating hypertrophy (Figure 2F). Additionally, LV tissues from CKD hearts had elevated markers of heart failure (BNP and ANP), fibrosis (measured by Sircol and hydroxyproline assays), oxidative stress (higher malondialdehyde and lower activity of the antioxidant enzyme superoxide dismutase), and apoptosis (Caspase-3 activity) (Figure S7).

Circulating plasma and kidney-tissue-derived EVs from CKD mice are cardiotoxic

To test the role of circulating EVs in HF in CKD mice, we isolated plasma EVs from CKD (mCKD-pEVs) and control (mCtr-pEVs) mice at 8w by using the precipitation method (Figure S8A). The EV-marker proteins Alix and Flotillin1 had significantly higher expression in mCKD-pEVs compared to mCtr-pEVs (Figure S8B), whereas the size distribution and total particle counts were comparable (Figure S8C and S8D). To assess cardiotoxicity, we treated isolated adult rat primary CMs with either mCKD-pEVs or mCtr-pEVs and then quantified their contractile function (Figure 2G). Similar to hCKD-EVs from patients, mCKD-pEV treatment significantly decreased systolic and intracellular Ca²⁺ and % cell shortening and increased systolic sarcomere length in adult rat primary CMs, compared with mCtr-pEV treatment (Figure 2H). These data suggest that mCKD-pEVs are cardiotoxic and induce contractile dysfunction in primary CMs in vitro.

Next, to examine whether CKD-kidney-derived EVs directly affect the heart, we isolated kidney-EVs from digested, cell-free kidney- supernatants of both 8w CKD and control mice by using a modified tissue-EV isolation protocol²². Debris and contamination originating from intracellular sources were removed by floating the EVs up through an iodixanol-density gradient ultracentrifugation (Figure S9). Kidney-derived EVs from both CKD (mCKD-kEVs) and control mice (mCtr-kEVs) were ~80–100nm in size (Figure S10A and S10B), positive for EV-markers (Alix and Flotilin-1), and negative for a known non-EV marker (Cyc1) (Figure S10C), thus demonstrating purity. Interestingly, mCKD-kEVs had significantly more EV marker protein Alix (quantified from Western Blot, Figure S10C), a higher % tetraspanin-positive EVs (quantified by single-EV flow cytometry, Figure S10E), but significantly fewer total particles per kidney (measured by NTA, Figure S10D). These results suggest mCKD-kEVs qualitatively and quantitatively differ from mCtr-kEVs.

When administered (10μg, ~1.2e10 particles/100μL) to rat myocytes (H9C2 cells)²³ in culture, mCKD-kEVs significantly lowered their viability compared to mCtr-kEVs, thus indicating the cardiotoxic effects of mCKD-kEVs (Figure 2I). In contrast, mCKD-liver-EVs from CKD mice were not cardiotoxic (Figure 2I), suggesting the cardiotoxicity is limited to the CKD kidney EVs. Moreover, in live/dead and TUNEL assays, mCKD-kEVs significantly induced H9C2 cell death and apoptosis compared to mCtr-kEVs (Figure 2J and 2K). Our IonOptix analysis of adult rat primary CMs showed that mCKD-kEVs significantly impaired adult rat primary CMs contractility and Ca²⁺ handling compared with mCtr-EVs (Figure 2L and 2M). mCKD-kEVs injected intramyocardially into mice hearts (20μg (2.3e10 particles)/injection) were taken up by cardiac cells and significantly increased the apoptosis, compared to mCtr-kEVs (Figure 2N and 2O). Taken together, these findings illustrated that kidney-derived EVs exert direct cardiotoxic effects on both cardiomyocytes and the heart.

Depleting circulating EVs ameliorated heart failure in a CKD mouse model

To verify if circulating CKD-EVs contribute to the pathogenesis of HF in vivo, we depleted circulating EVs in CKD mice by injecting them with GW4869 (GW), a systemic inhibitor of nSMase, an enzyme involved in EV biogenesis^11,24,25, for 10 weeks (Figure 3A). Circulating particle numbers significantly decreased and tetraspanin^+ve EV numbers dropped to less than half in both kidney and plasma of GW-injected CKD mice compared to vehicle-injected CKD mice (Figure 3B; Figure S11). Neither plasma BUN level nor blood pressure were altered in either GW- or vehicle-injected CKD mice and remained elevated compared to controls (Figure 3C and 3D). In contrast, plasma creatinine level dropped significantly in GW-CKD mice compared to vehicle-CKD mice and was significantly higher than in GW-Ctr mice (Figure 3C), thus indicating the kidney injury in GW-CKD mice (Figure 3C and 3D). Remarkably, cardiac function, as measured by %EF and %FS, was substantially improved in the GW-CKD mice compared to vehicle-CKD mice (Figure 3E and 3F). Additionally, GW-CKD mice showed significantly ameliorated cardiac fibrosis (Figure 3G) and hypertrophy (Figure 3E–3H). The heart-to-body weight ratio was significantly decreased in GW-CKD mice compared to Vehicle-CKD mice (Figure S11D). Results from this loss-of-function experiment demonstrated that depleting CKD-EVs rescues cardiac function even in the presence of kidney injury, high blood pressure and other co-morbidities circulating in the plasma. Our data therefore established that circulating CKD-EVs play a causal role in and significantly contribute to CKD-induced HF.

Figure 3. Depletion of EVs improves cardiac function in CKD mice.

A, Schematic of study design. C57BL/6J mice fed an adenine diet received an intraperitoneal injection of 2.5 mg/kg (3 times/ week) of either GW4869 or vehicle (7.5% DMSO in saline) for 10 weeks. B, Quantification of mouse plasma particle counts from CKD mice receiving either vehicle or GW4869 as measured by NTA (n=4). C and D, Measurement of blood urea nitrogen (BUN), plasma creatinine level, systolic and mean blood pressure in Ctr and CKD mice treated with either vehicle or GW4869. E, Echocardiographic image of ventricle wall motion in Ctr and CKD mice treated with either vehicle or GW4869. F, Echocardiography of heart function showing %EF, %FS and wall thickness (n=10), groups as indicated. G, Representative Masson-trichome staining of cross sections of the whole heart and LV, and quantification of % fibrosis in the groups as indicated (n=3). H, WGA staining and quantification of myocyte cross-sectional area (CSA) of the hearts from the groups as indicated (n=3). Data are presented as mean ± SEM. Data in panel B are presented as median with interquartile range and were analyzed with Mann-Whitney U test. Data in panel C–H were analyzed using two two-way ANOVA followed by Tukey multiple comparison test.

Circulating EVs from CKD patients carry a distinct miRNA signature

The molecular components of hCKD-EVs that are responsible for cardiac dysfunction are largely unknown. EVs are selectively enriched for small RNAs that can provide mechanistic insights into HF etiology and progression in CKD patients. To identify CKD-EV risk factors, we examined the transcriptomic profiles of hCKD-EV-RNA (from 13 CKD patients) and compared them to hCtr-EV-RNA (from 6 healthy subjects) by using deep small-RNA sequencing followed by qPCR validation (Figure 4A).

Figure 4. miRNA signatures of human plasma CKD-EVs.

A, Schematic of study design. B, Principal component analysis (PCA) of miRNA sequencing data from hCtr/CKD EVs. C and D, Volcano plot and heatmap showing differentially expressed miRNAs in hCKD-EVs compared with hCtr-EVs (control, n=6; CKD, n=13). 16 upregulated and 36 downregulated miRNAs in hCKD-EVs were identified. E, Analysis of significantly expressed miRNAs with high logFold change and high abundance in hCKD-EVs compared with hCtr-EVs. Among the 16 upregulated miRNAs in hCKD-EVs, three exhibited the highest log-Fold changes (miR-320a-3p, 4454, 2110) and seven (miR-22–5p, 423–5p, 320b, 484, 193a-5p, 130b-3p and 378a-3p) showed the most abundant expression in hCKD-EVs. F, KEGG pathway analysis showing the top six most enriched pathways in highly expressed hCKD-EV-miRNAs. G, qPCR analysis of miRNA expressions in hCKD-EVs compared with hCtr-EVs (Ctr, n=15; CKD, n=20). Data in panel D are present as median with interquartile range and were analyzed with Mann-Whitney U test.

Principal component analysis (PCA) revealed that the miRNA expression profiles of cardiotoxic hCKD-EVs were distinct from those of non-cardiotoxic hCtr-EVs (Figure 4B). The clustered hCtr-EVs exhibited low variability; while the hCKD-EVs displayed irregular distributions, both clustering with hCtr-EVs and showing discrete patterns, suggesting the heterogeneity within the CKD group (Figure 4B).

The transcriptomic profiles of hCtr-EVs and hCKD-EVs differed significantly, as depicted in the volcano plot (Figure 4C; p-value < 0.05, log-Fold change >1.0); heatmap (Figure 4D) and list of top 51 differentially expressed miRNAs (Figure S12). For further validation, we selected i) 10 upregulated miRNAs exhibiting the highest significant log-Fold changes (miR-320a-3p, 4454, and 2110) and most abundant expression levels (miR-22–5p, 423–5p, 320b, 484, 193a-5p, 130b-3p, and 378a-3p) in hCKD-EVs (Figure 4E); and ii) four downregulated miRNAs common between our differentially expressed list and previously reported downregulated CKD miRNAs^11,12 conserved between species (Figure S12 and S13). Of note, the only study with unbiased small RNA sequencing of clinical CKD-EV samples (from children post-kidney transplant) reported mostly downregulated miRNAs. Kyoto encyclopedia of genes and genomes (KEGG) analysis showed the significantly enriched pathways, including apoptosis, autophagy, and cell senescence (Figure 4F), a result that aligns with the pro-apoptotic and cardiotoxic properties of the hCKD-EVs. Our qPCR validation confirmed that nine of the 10 upregulated miRNAs (miR-130b-3p, 193a-5p, 2110, 22–5p, 320a-3p, 320b, 378a-3p, 423–5p, and 484) had significantly higher expression in hCKD-EVs than in hCtr-EVs, while miR-4454 remained unchanged. qPCR analysis also confirmed significant downregulation of two CKD-EV-miRNAs (miR-142–5p and 17–5p) and no change in the other two miRNAs (miR-224–5p and 16–5p), compared to hCKD-EVs (Figure 4G).

To examine differences between CKD patients with and without HF (CKD(+)HF and CKD(−)HF) in their miRNA expressions, we compared miRNA levels in their plasma EVs. Interestingly, expression of several CKD-miRNAs (130–3p, 2110, 320a-3p, 320b, 378a-3p, and 423–5p) was significantly higher in EVs from both CKD(+)HF and CKD(−)HF patients compared with Ctr; but was comparable between these two patient groups (Figure S14). The lack of significant difference between CKD(−)HF and CKD(+)HF could be because the CKD(−)HF group (%EF>50; Table S2) is heterogeneous, and likely include patients with HFpEF or sub-clinical or asymptomatic HF. To confirm this, we assayed for ANP, NT-ProBNP, and hsTNI, known markers of myocardial injury and HF, in CKD(−)HF, CKD(+)HF groups and healthy Ctr groups using quantitative immunoassays (similar to those in clinics). Remarkably, all three markers were significantly higher in plasma of both CKD(+)HF and CKD(−)HF cohorts compared to healthy Ctr (Figure S15), suggesting the presence of cardiac stress and sub-clinical HF in the CKD(−)HF cohort. Moreover, expression of many (9 out of the total 14 tested) CKD-EV-miRNAs correlated with these HF markers (Figure S16–S18), thus underscoring the need to analyze a well-characterized CKD patient population for novel biomarker and therapeutic target discovery for HF.

miRNAs enriched in hCKD-EVs are cardiotoxic to human cardiomyocytes

To test whether the above miRNAs enriched in hCKD-EVs are cardiotoxic, we treated hAC16-CMs and hiPSC-CMs with nine miRNA mimics and quantified their effects on viability and contractility by using the MTT assay and IonOptix measurements, respectively. The MTT assay revealed that four miRNA mimics (miR-2110, 320b, 484, and 130b-3p) significantly decreased hAC16-CMs viability (Figure 5A). IonOptix analysis of beating hiPSC-derived CMs showed that seven miRNA mimics (miRs-2110 and 320b (both human-specific), 484, 130b-3p, 193–5p, 22–5p, and 320a-3p) significantly reduced intracellular Ca²⁺ and four of these also lowered the systolic Ca²⁺ (Figure 5B and 5C; Figure S19), compared to control-miRNA mimic treatment. Additionally, these miRNA mimics significantly increased the T90 (time for the Ca²⁺ signal (F₃₄₀/F₃₈₀) to rise from 10% to 90%) and Tau (time constant for calcium decay) in treated hiPSC-derived CMs. In addition, inhibition of downregulated CKD-EV-miRs by using anti-miR-142–5p and anti-miR-17–5p did not affect hAC16-CM viability. However, anti-mir-17–5p significantly reduced Ca²⁺ transients in treated hiPSC-CMs (Figure S20), thereby demonstrating the cardioprotective function of downregulated CKD-EV-miRNAs. These findings suggest that the miRNAs enriched in hCKD-EVs are cardiotoxic to human CMs in vitro.

Figure 5. miRNAs enriched in hCKD-EVs contribute to their cardiotoxicity.

A, Viability of hAC16-CMs transfected with nine selected miRNA mimics and one negative control (n=6–16). B, Schematic of calcium handling analysis in hiPSC-CMs treated with miRNA mimics. hiPSC-CMs were transfected with miRNA mimics every three days for a total of three times. Representative calcium amplitude transients through contraction-relaxation cycles of hiPSC-CMs treated with miR-2110, miR-320b, miR-484 mimics as measured by IonOptix analysis. C, Quantification of calcium transients and contractility of hiPSC-CMs transfected with the same nine miRNA mimics and one negative control (n=12). Quantification of systolic Ca²⁺ (F₃₄₀/F₃₈₀), intracellular Ca²⁺ (F₃₄₀/F₃₈₀), Tau, and T90(s). D, Schematic of study design evaluating uptake and gene regulation of hCKD-EVs. E, Confocal microscopy of PKH67-labeled hCKD-EV uptake in hAC16-CMs at 4h. F and G, qPCR analysis of miRNA expressions in hAC16-CMs treated with equal counts of hCKD/Ctr-EVs (2e5 particles /cell) (n=3–6). H, Viability of hAC16-CMs treated with hCKD-EVs and anti-miRNA, as indicated. hAC16-CMs were transfected with anti-miRNA (scramble control, miR-2110, or miR-320b) and then treated with hCKD-EVs after 6h. Cell viability MTT assay was performed 24 hours after treatment. Data in panel A-C are presented as mean ± SEM. Data in panel H are presented as box-and-whisker plots. Data in panel A–C and H were analyzed with one-way ANOVA with Dunnett’s multiple comparison test. P < 0.05 (*). Data in panel F and G are presented as median with interquartile range and were analyzed with Mann-Whitney U test.

To determine whether CMs effectively internalized hCKD-EVs to subsequently increase miRNA expression, we treated hAC16-CMs with hCKD-EVs and studied first EV uptake and then miRNA expression (Figure 5D). Uptake analysis of PKH67 dye-labeled hCKD-EVs confirmed their internalization into the cytoplasm of hAC16-CMs after 4 hours of treatment in vitro (Figure 5E). PKH67-hCKD-EVs intravenously injected into mice via tail vein (Figure S21A) showed higher cardiac uptake than dye-only control injections, as determined using in vivo imaging system (IVIS) (Figure S21B). Immunofluorescence analysis showed the PKH67-hCKD-EVs were internalized by Troponin^+ve CMs in the heart (Figure S21C). These data indicate that CMs take up the CKD-EVs both in vitro and in vivo. To study whether CKD-EV internalization induces CKD-miRNA expression, we incubated equal numbers of hCtr/CKD EVs and hAC16-CMs (2e5 particles/cell) for 4 hours and then measured cellular miRNA expression. qPCR quantification revealed that four miRNAs (miR-2110, 320b, 130b-3p, 484) had significantly upregulated expression in hCKD-EV-treated cells compared to hCtr-EV-treated cells (Figure 5F), thereby corroborating effective EV internalization. Interestingly, expression of other five miRNAs remained unchanged between hCKD-EV- and hCtr-EV-treated cells (Figure 5G), a finding that suggests these miRNAs have either lower abundance in hCKD-EVs or higher endogenous levels in hAC16-CMs. Collectively, these data demonstrate that CMs can internalize miRNA-carrying hCKD-EVs that contribute to the cardiotoxicity through pro-apoptotic effects and by disrupting CM contractility and Ca²⁺ handling.

Some of the novel enriched cardiotoxic miRNAs in CKD-EVs (e.g. miR-2110 and miR-320b) are human-specific and have not previously been characterized for function on cardiomyocytes before. To determine their function in response to CKD-EVs, we used antisense oligonucleotides (anti-miRs) that bind to target miRNAs and negate the function of CKD-EVs carrying those miRNAs. Specifically, we treated hAC16-CMs with each anti-miRNA followed by co-treatment with CKD-EVs, which suppressed EV-miRNA activity. As expected, hAC16-CMs co-treated with CKD-EVs and a scrambled control anti-miR showed significantly reduced viability. Remarkably, co-treating CKD-EVs with anti-miRs-2110/320b notably improved cell viability (Figure 5H and Figure S22). These data demonstrate that EV-miRNAs contribute to the cardiotoxic responses induced by CKD-EVs.

Cardiotoxic CKD-miRNAs impaired contractile and calcium-handling gene expression

For a mechanistic understanding of our findings, we sought to uncover gene expression changes in cardiomyocytes induced by novel cardiotoxic miRNAs uniquely enriched in CKD-EVs. We transfected the hiPSC-CMs with mimics of three miRNAs associated with contractile dysfunction (miR-2110, miR-320b, miR-484) or a scramble control for 48h, followed by RNA sequencing (Figure 6A). This analysis revealed suppression of several contractile and calcium-handling genes (Figure 6B; Figure S23–S24). Gene Set Enrichment Analysis (GSEA) of GO Biological Process pathways revealed that contractile function, calcium handling, and metabolic processes were among the top five significantly suppressed pathways based on differentially expressed genes (e.g., negative regulation of calcium channel activity, suppressed by miR-2110; myofibril assembly, suppressed by miR-320b; and mitochondrial ATP synthesis, suppressed by miR-484; Figure 6C–6E), which correlated with the miRNAs’ function. Moreover, qPCR analysis in hiPSC-CMs chronically treated (three times, once every three days) with miR-2110, miR-320b, or miR-484 mimics showed downregulation of several calcium handling and contractile function associated genes, such as PLN (miR-2110, miR-484), MYH6 (miR-320b, miR-484) and KLF (miR-2110) (Figure 6F). Likewise, expression of PLN and Atp2a2 were found to be downregulated in the hearts of the CKD mice (Figure S25), suggesting that cardiotoxic miRNAs affect cardiac function by altering the expression of genes crucial for Ca²⁺ cycling and muscle contraction. Together, these data indicate numerous CKD-miRNAs synergistically limit cardiomyocyte and cardiac contractility

Figure 6. miRNAs enriched in hCKD-EVs impaired the contractility and calcium handling gene expression in human iPSC-derived cardiomyocytes.

A, Schematic of study design for mRNA sequencing of hiPSC-CMs transfected with miRNA mimics. B, Heatmap showing expression of differentially expressed genes in hiPSC-CMs transfected with miRNA mimics (miR-NC, miR-2110, miR-320b, miR-484); the top 50 significantly differential expressed genes are listed in Tables S5–S7. C, Gene Set Enrichment Analysis (GSEA) of GO Biological Process pathways in hiPSC-CMs transfected with miR-2110, miR-320b (D), and miR-484 (E). The plot displays significantly enriched pathways grouped by activation or suppression. Dot size represents the number of genes involved (Count), and the x-axis shows the Gene Ratio. F, Expression of the calcium handling and contractile function-associated genes was assessed by qPCR in hiPSC-CMs treated with miR-2110, miR-320b, or miR-484 mimics every three days for a total of three doses (n=6). Data in panel F are presented as median with interquartile range and were analyzed with Mann-Whitney U test.

miRNAs associated with circulating CKD-EVs originate in the kidney

To assess the role of CKD-plasma EVs in reno-cardiac crosstalk, we sought to determine whether circulating-EV-miRNAs originate from the kidney. First, we tested whether mCKD-circulating- and kidney-derived EVs harbor the same miRNAs signatures as hCKD-EVs (Figure 7A and 7B). Of the nine key miRNAs detected in hCKD-EVs, seven are evolutionarily conserved across mammals (Figure S26). qPCR analysis revealed that four of those miRNAs (miR-130b-3p, 22–5p, 320–3p, and 423–5p) were significantly upregulated in both circulating and kidney mCKD-EVs compared to mCtr-EVs (Figure 7A and B); miR-484 showed a similar trend but did not reach significance. miR-378a-3p had higher expression in plasma-CKD-EVs but not in kidney-CKD-EVs, compared to controls, whereas miR-193–5p levels were unchanged. These data suggest both shared and distinct circulating miRNA cargo between mouse and human CKD-EVs. Further, these findings indicate the circulating CKD-miRNAs can originate renally.

Figure 7. CKD-EV-miRNAs originate renally in CKD mice.

qPCR analysis of miRNA expressions in A, mCKD-pEVs compared with mCtr-pEVs (n=5–10); B, mCKD-kEVs compared with mCtr-kEVs (n=7). C, Schematic of study design tracing the cellular origin of EV-miRNAs in circulation: expression of corresponding primary miRNA (pri-miRNA) was used to predict the cellular origin of mature EV-miRNAs. D, Expression of corresponding pri-miRNA transcripts in kidney, heart, liver, and PBMCs was assessed by qPCR in CKD versus control mice (n=3–8). Data in panel A and B are presented as median with interquartile range and were analyzed with Mann-Whitney U test. Data in panel D are presented as box-and-whisker plots and were analyzed with Mann-Whitney U test with multiple comparisons; P values are unadjusted.

Next, we investigated the kidney as a potential source of circulating CKD-EV-miRNAs. Transcription of primary miRNAs (pri-miRNAs), which give rise to mature miRNAs, is cell type- and tissue-specific²⁶ and their abundance in cells and tissues can predict the origin of circulating EV-miRNAs. To determine the origin of circulating mCKD-EV-miRNAs, we quantified the expression of their corresponding endogenous pri-miRNAs²⁷ in different tissues, including the kidney, heart, liver, blood (PBMCs), lungs, spleen, and skeletal muscle, from CKD and control mice, by qPCR (Figure 7C and Figure S27). Interestingly, pri-miRNAs corresponding to four miRNAs (miR-320, −130, −484 and −22) (pri-miR320 correspond to miR-320–3p in mice; since mice lack miR-320b) that were significantly upregulated in mCKD-kidney- and mCKD-plasma-EVs had significantly higher expression in mCKD-kidneys than in mCtr-kidneys (Figure 7D). Importantly, these pri-miRNAs did not increase in mCKD-heart, -liver, -lungs, -spleen, -PBMC, or -skeletal muscle (Figure 7D and Figure S27; except for pri-miR-22 in mCKD-PBMCs). These data show the circulating cardiotoxic CKD-EV miRNAs have a predominantly renal origin, with PBMCs serving as an additional source.

To extend these findings to human data and to underpin their translational value, we performed a tissue deconvolution analysis²⁸ to determine if the hCKD-EV miRNAs are unique to the diseased kidney. We estimated the relative tissue enrichment of all differentially expressed hCKD-EV-miRNAs and compared them to the publicly available, curated, tissue-specific, bulk-tissue miRNA expression profiles from the Genotype-Tissue Expression (GTEx) database²⁹ as a reference. Our deconvolution analysis found significantly higher hCKD-EV-miRNA relative enrichment in kidney cortex compared to its relative enrichment in control. These data indicate that kidney cortex, a major site of CKD pathology, contribute significantly to circulating EV-miRNAs in CKD patients, rather than liver, heart, lung, or other organs (Figure S28).

To identify the specific cell types contributing to CKD-EV-miRNAs, we prepared single cell suspensions from the control- and CKD-kidneys and separated them into CD45^+ve immune cells, CD31^+ve endothelial cells, and the remaining CD31^−veCD45^−ve renal cells (the latter group includes renal glomerular and tubule cells, podocytes, and fibroblasts) (Figure 8A–D). We then examined CKD-pri-miRNA expressions in those cell groups. Four CKD-pri-miRNAs (pri-miR-130, −22, −320, and −484; Figure 7D) increased in renal tissue had significantly higher expression in CD31^−veCD45^−ve renal cells (Figure 8E). Two CKD-pri-miRNAs (pri-miR-130b and −484) exhibited significantly higher expression in CD31^+ve endothelial cells (Figure 8E). Interestingly, none of the CKD-pri-miRNAs were enriched in CD45^+ve immune cells. These data indicated that CKD-EV-miRNAs within the CKD-kidney have heterogeneous origins. Remarkably, although CD45^+ve leukocytes numbers are markedly elevated in the CKD-kidney (Figure 8B; Figure S29), they are not a substantial source of CKD-pri-miRNAs or CKD-EV-miRNAs. This is the first direct evidence that circulating CKD-EV-miRNAs have renal origins.

Figure 8. CKD-EV-miRNAs may have heterogeneous cellular origins in CKD-kidneys.

A, Schematic of study plan to detect the origins of CKD-EV-miRNAs (via expression of their corresponding pri-miRNAs) from different cell types in the kidney. B, Representative flow cytometry analysis of CD45^+ve and CD31^+ve cells isolated from CKD kidneys compared to Ctr kidneys. C, Purity of CD45⁺ and CD31^+ve cell population was assessed by flow cytometry following magnetic-activated cell sorting (MACS). D, Quantification of %CD45^+ve cells and %CD31^+ve cells in CKD kidneys compared with Ctr kidneys (n=5). E, Expression of the corresponding pri-miRNA transcripts of selected CKD-EV-miRNAs in CD45^+ve leukocytes, CD31^+ve endothelial cells, and double-negative CD45^−veCD31^−ve renal cells were assessed by qPCR in CKD vs. Ctr kidneys. F, Schematic illustrating the proposed mechanism by which EV-associated miRNAs secreted from CKD kidneys enter the circulation and contribute to heart failure development. Data in panel D are presented as median with interquartile range and were analyzed with Mann-Whitney U test. Data in panel E are presented as box-and-whisker plots and were analyzed with Mann-Whitney U test with multiple comparisons; P values are unadjusted.

DISCUSSION

Our study reports several new findings. First, we present the evidence that circulating EVs from CKD patients, but not the EV-depleted fraction thereof, are cardiotoxic. CKD-EVs induce apoptosis, impair contractile function, and restrict Ca²⁺ handling in treated cardiomyocytes. Second, we show that depleting circulating EVs in CKD mice recovers cardiac function and ameliorates heart failure, even in the presence of CKD co-morbidities. Third, we demonstrate that CKD-EVs are enriched in cardiotoxic miRNAs that originate from kidney cells. Taken together, these results outline a novel mechanism through which, during CKD, renal cells secrete CKD-EVs carrying cardiotoxic miRNAs to the circulation, thereby altering cardiac contractile gene expression and, with chronic exposure, causing heart failure (Figure 8F). Circulating CKD-EVs thus play a causal role in the pathogenesis of HF, via a reno-cardiac communication axis.

CKD and HF have a complex and interdependent relationship. Independently of comorbidities and other traditional and nontraditional CVD risk factors in CKD patients, kidney dysfunction triggers the release of bioactive components that contribute to cardiac damage³. In more than 70 studies of non-dialyzed subjects with CKD, correcting for classical cardiovascular risk factors, such as hypertension, diabetes, and dyslipidemia, did not neutralize the impact of CKD on cardiovascular risk^3,30,31. That these traditional risk factors only partially explain the excess risk of CVD in CKD patients points to additional kidney-derived molecular factors affecting cardiac function. However, no kidney- or plasma-derived risk factor that triggers HF at either a cellular or molecular level in CKD patients has yet been identified, primarily because there is little systematic investigation in patients with reno-cardiac disease. Our study is the first to address this critical knowledge gap by exploring the direct cardiotoxic effects of CKD-associated molecular factors on cardiomyocyte and/or cardiac contractile functions.

EVs are an important component of circulating plasma. Most prior EV studies involving reno-cardiac axis are observational, and/or correlational in nature. For example, a CKD-mouse study that reported proarrhythmic remodeling of the heart did not study reno-cardiac crosstalk²¹. Another acute kidney injury (AKI) murine study reported interleukin-33 signaling from kidney to heart³² but did not investigate the EVs. CKD-EVs from mice¹¹ and patients³³ have been shown to exacerbate vascular calcification, to limit angiogenesis¹², and to serve as a surrogate marker of endothelial dysfunction in ESRD patients^34–36. Collectively, these previous studies implicate circulating factors or EVs as contributors to CKD-induced HF but do not explicate the direct role CKD-EVs play in cardiotoxicity, contractility and HF pathogenesis. Additionally, none of these studies investigated reno-cardiac interactions or renal cell contributions to altered circulating EV cargo in CKD.

To bridge these knowledge gaps, we recruited CKD patients and healthy control subjects, utilized a mouse model of modified adenine diet-induced CKD with heart failure, and employed hiPSC-derived cardiomyocytes to study the causal roles and mechanisms of CKD-EVs. We fully characterized hCKD-EVs and demonstrate that they are cardiotoxic and carry several pro-apoptotic and anti-contractile miRNAs, including the human specific miRs-2110 and 320b. In addition, several hCKD-miRs are known to be associated with renal inflammation and fibrosis (miR-193a-5p³⁷), apoptosis (miR-130b-3p^{38, 39}, miR-320–3p^40,41, miR-378a-3p⁴²), oxidative stress (miR-423–5p^43,44) and cardiac function (miR-320–3p^41,45). Overall, our findings confirm that CKD patients have significantly altered circulating CKD-EV and -EV-miR compositions. Besides, cardiotoxic effects and miRNA signatures of human and mouse CKD-plasma-EVs were largely comparable. Additionally, we demonstrated that pharmacologically depleting CKD-EVs by using GW4869 ameliorated HF in CKD mice, even in the presence of other co-morbidities, thus providing a direct proof-of-concept evidence for the causal role of EVs in HF. Although GW4869 reduced EV secretion from CKD-kidneys (possibly via its effect on preventing intraluminal vesicle formation), it can impact lipid-metabolism and inflammation²⁵. These and other side effects of GW4869 may have partially contributed to improved cardiac contractility in our CKD mouse model. Future studies with endogenous labeling of kidney-EVs may reveal in-depth EV-based mechanisms in more detail.

Kidney-enriched miRNAs, which can be packaged into EVs and transported into circulation, mediate organ crosstalk^7,46. A proteomic analysis of cardiac EVs showed that some proteins in these EVs may originate from the kidney⁴⁷. To date, few direct experimental methods can define the tissue or cellular origin of circulating EV miRNA cargo. Pri-miRNAs are precursors to mature microRNAs (miRNAs) and are not typically found in circulating EVs²⁷. While pri-miRNAs are not a primary marker of EV origin, the tissue and cellular presence of pri-miRNAs can be used to infer the origins of circulating EV-miRNAs. Interestingly, high expression of many CKD-plasma-EV-miRNAs correlated with CKD-kidney-EV-miRNAs and their corresponding endogenous pri-miRNAs from the CKD-kidneys, but not with pri-miRNAs from other organs such as the CKD-heart, -liver, -lungs, -spleen, or -PBMCs from the blood (Figures 7D and S27). Moreover, CD31^−veCD45^−ve renal cells including renal tubular epithelial cells, podocytes, and fibroblasts, but not CD45^+ve immune or CD31^+ve endothelial cells, exhibited concomitantly increased CKD-EV-miRNA levels, which suggest these cells could be an origin of cardiotoxic miRNA-carrying EVs. We thus provided cumulative evidence of renal origin of CKD-EV-miRNAs using pri-miRNA expression from organs and sorted kidney cells from mice and in silico deconvolution data from humans. This is the first direct evidence that during CKD, altered circulating CKD-EV miRNAs originate from kidney cells. Unequivocally demonstrating that kidney-derived EVs directly transfer to the heart via circulation would require elaborate kidney cell-specific genetic mouse models with endogenously labeled fluorescent EV-markers; this may be a future line of inquiry.

Remarkably, in our study, CKD-EV-miRNA expressions did not significantly differ between our CKD(−)HF and CKD(+)HF cohorts, possibly due to heterogeneity within the CKD(−)HF group (%EF>50). In concurrence, many CKD-EV-miRNA expressions correlated with traditional myocardial injury and mortality markers, such as hsTNI, NT-proBNP, and ANP, that suggest cardiac stress and sub-clinical cardiac damage in our CKD(−)HF cohort. These data are supported by previous observations that HFpEF is more prevalent than HFrEF in CKD patients⁴⁸. NT-proBNP, hsTnI, and ANP levels in CKD can be elevated due to both decreased renal clearance and increased production in response to cardiac stress⁴⁹. Yet, cutoffs for these prognostic markers, specifically in patients with HFpEF and CKD are still under active investigation⁴⁹. While elevated levels of these markers are valuable in assessing subclinical cardiac damage in CKD, they are not standalone diagnostics. Accurate diagnosis in this context requires integration with additional clinical assessments, such as echocardiography, cardiac MRI, LV filling pressures, exercise capacity (6-minute walk test or peak oxygen consumption), symptoms (e.g. NYHA functional class evaluation), quality of life, and other objective measures. This underscores CKD-miRNAs’ potential in diagnosing HF early in CKD patients even before the symptoms are clinically apparent, identifying individuals at higher risk of developing HF, or predicting disease severity and risk stratification of CKD-induced HF. That CKD-EV-miRNAs correlate with markers of myocardial injury and cluster in similar patterns between CKD +/− HF patients (Figures S16–18) could be attributed to their decreased renal clearance in CKD (low eGFR). Yet, we have identified several EV-miRNAs with both increased and decreased expression in CKD patients (Figures 4G and S12), suggesting other factors, including increased production, uremic toxicity, and/or inflammation, or interplay between these factors, may contribute to their dysregulation. Higher levels of primary miRNAs in CKD kidneys in our study (Figure 7D) also confirmed increased miRNA biogenesis. To address the clearance mechanisms and the CKD-specificity of the EV-miRNAs, future studies should systemically correlate CKD-EV-miRNA expressions with established HF markers in longitudinal comparisons of CKD patients with and without HF, CKD patients with HFpEF, as well as in HF patients who do not have CKD. Moreover, patients with HF before and after a kidney transplant for ESRD can provide an ideal human model to investigate the functional relevance of EVs. Such hybrid clinical and basic-science work can also map the complex relationship between HF and CKD and define the different disease phenotypes of the reno-cardiac axis.

Going forward, investigations expanding on our findings may address several important questions. While our functional and miRNA expression data provided valuable insights, our significantly younger control group may have introduced age-related bias. In addition, many of our controls were healthy according to their medical history and clinical metadata were not uniformly available. On the other hand, all patients in our CKD cohort were clinically stable, received GDMT and had higher medication utilization that reflected the intensified therapeutic regimen necessitated by their disease state. 30% of the CKD patients in our functional contractility study (Figure 1L and M) received Ca²⁺ channel inhibitors and 50% received beta blockers. It is unlikely that these medications were transferred from plasma via isolated CKD-EVs (which were thoroughly washed and processed) in quantities that could significantly impact their cardiotoxic function. The cardiotoxic and anti-contractile function of plasma- and kidney-EVs from CKD-mice also confirmed the pathological role of CKD-EVs play in HF, independent of the variations arising from co-morbidities, age differences or concomitant medications in human samples. Future studies to identify HF biomarkers in CKD patients may need to recruit larger study populations with age-matched cohorts and utilize regression analysis for adjusted data sets.

EVs contain diverse cargo encompassing other RNA species, proteins, and lipids that reflect the pathophysiology of their originating cell. To comprehensively uncover the cardiotoxic factors in CKD-EVs, the identities and functions of those biomolecules must be investigated. CKD is a systemic disease with broad adverse effects, including on the endocrine and immune systems⁵⁰. The molecular identities of EVs produced by various cell types need to be determined in order to comprehensively characterize pathogenic factors, including the systemic effect of uremic toxins on the biogenesis of EVs, contributing to HF in CKD. Moreover, while our study evaluated the cardiotoxic effects of circulating and kidney CKD-EV-miRNAs on cardiomyocytes, to fully understand the molecular mechanisms underlying renal–cardiac crosstalk requires clarity regarding how EVs influence non-cardiomyocytes, including immune cells, endothelial cells, and cardiac fibroblasts, within the heart.

Collectively, using human, mouse, and cell culture models, the work we report here demonstrated that CKD-EVs, via their renal-derived miRNA cargo, are key contributors to humoral cardiotoxicity in CKD patients. Our study revealed CKD-EV-mediated molecular crosstalk between the kidney and the heart that plays a causal role in the pathogenesis of HF in CKD. Future studies on the cardiotoxic cargo of CKD-EVs could identify robust biomarkers for early HF diagnosis, prognosis, and monitoring as well as new therapeutic targets to treat chronic reno-cardiac disease.

Supplementary Material

Supplemental_Publication_Material_1

Expanded Supplemental Methods

Table S1 to Table S7

Figure S1 to S30

Reference 51–73

Supplemental_Publication_Material_2

MINSEQE Checklist

Supplemental_Publication_Material_3

STROBE Checklist

Supplemental_Publication_Material_4

ARRIVE Checklist

CLINICAL PERSPECTIVE.

What is new?

• Using human, mouse and cell culture models, we show that circulating CKD-EVs are cardiotoxic; they induce apoptosis, impair contractility, and restrict calcium handling in treated cardiomyocytes.

• CKD-EVs carry distinct CKD-EV-miRNAs which are primarily of renal origin.

• This CKD-EV-mediated molecular crosstalk between the kidneys and the heart plays a causal role in the pathogenesis of heart failure in patients with CKD.

What are the clinical implications?

• CKD-EV-miRNAs correlate with established markers of cardiac injury, thus uncovering the presence of sub-clinical heart disease and heterogeneity in CKD patients not yet diagnosed with HF.

• Our work aids in defining the complex relationship between CKD and HF and the different disease phenotypes along the reno-cardiac axis.

• CKD-miRNAs are promising biomarkers for early disease detection and potential targets for novel therapeutic interventions in chronic reno-cardiac disease.

Nonstandard Abbreviations and Acronyms

CMs

Cardiomyocytes

GW

GW4869

hAC16-CM

Human AC16 cardiomyocyte

hiPSC-CM

Human induced pluripotent stem cell-derived cardiomyocyte

hCKD-EV

EVs from CKD patients

hCtr-EV

EVs from healthy control

Pri-miRNA

Primary miRNA

KDGIO

Kidney disease improving global outcomes

mCtr-pEV

Plasma EVs from control mice

mCKD-pEV

Plasma EVs from CKD mice

mCtr-kEV

Kidney EVs from control mice

mCKD-kEV

Kidney EVs from CKD mice

SEC

Size exclusion chromatography