Abstract
Background/Aims: Indoxyl sulfate (IS) is a protein-bound uremic toxin that accumulates in patients with chronic kidney disease (CKD). We explored the effect of IS on human early endothelial progenitor cells (EPCs) and analyzed the correlation between serum IS levels and parameters of vascular function, including endothelial function in a CKD-based cohort. Methods: A cross-sectional study with 128 stable CKD patients was conducted. Flow-mediated dilation (FMD), pulse wave velocity (PWV), ankle brachial index, serum IS and other biochemical parameters were measured and analyzed. In parallel, the activity of early EPCs was also evaluated after exposure to IS. Results: In human EPCs, a concentration-dependent inhibitory effect of IS on chemotactic motility and colony formation was observed. Additionally, serum IS levels were significantly correlated with CKD stages. The total IS (T-IS) and free IS (F-IS) were strongly associated with age, hypertension, cardiovascular disease, blood pressure, PWV, blood urea nitrogen, creatine and phosphate but negatively correlated with FMD, the estimated glomerular filtration rate (eGFR), hemoglobin, hematocrit, and calcium. A multivariate linear regression analysis also showed that FMD was significantly associated with IS after adjusting for other confounding factors. Conclusions: In humans, IS impairs early EPCs and was strongly correlated with vascular dysfunction. Thus, we speculate that this adverse effect of IS may partly result from the inhibition of early EPCs.
Introduction
Cardiovascular diseases are the primary complications in patients with chronic kidney disease (CKD) [1,2]. In the advanced stages of CKD, endothelial dysfunction is accelerated [3], which participated in the development of atherosclerosis and arterial stiffening of the large arteries. These factors have been shown to be independent predictors of cardiovascular-related mortality in CKD. Thus, the maintenance of endothelial integrity is a major step in delaying the progression of vascular diseases in CKD [4].
Indoxyl sulfate (IS) is a protein-bound uremic toxin that is markedly affected by liver, kidney and diet [5,6,7] and has been regarded as a nontraditional cardiovascular risk factor in CKD. Previous in vitro studies showed that IS had a specific toxic effect on vascular endothelial cells [8], smooth muscle cells [9], and renal tubular cells [10,11]. Clinical outcome studies also demonstrated that IS may play a major role in cardiovascular and all-cause mortality in CKD patients, including those on hemodialysis [5,12,13,14,15]. These adverse effects may be directly or indirectly attributed to endothelial dysfunction as a result of exposure to retained uremic toxins. Recent studies showed that endothelial integrity is maintained by the proliferation and migration of resident cells, as well as by endothelial progenitor cells (EPCs), which are primarily derived from bone marrow or of myeloid origin [16,17]. One early report suggested that circulating EPC levels may be a biological marker for vascular function and cumulative cardiovascular disease [18].
However, whether IS could directly impair EPC function remained unclear. To this end, the present study explored the in vitro effects of IS on early EPCs. Additionally, the correlation between serum IS levels and parameters of endothelial function and vascular function were analyzed in a CKD cohort.
Materials and Methods
This study (12MMHIS143) was approved by the Institutional Review Board of the Mackay Memorial Hospital and performed in accordance with the principles of the Declaration of Helsinki. Healthy donors and patients with stable CKD stages 1-5 were recruited, and written informed consent was obtained from each.
Isolation and cultivation of EPCs
The isolation and cultivation of early EPCs was reported previously [19]. In brief, after the collection of peripheral blood (80 mL), the PBMCs were fractionated from other blood components via centrifugation on Ficoll-Paque plus (Amersham Biosciences, Uppsala, Sweden) according to the manufacturer's instructions. Isolated PBMCs were resuspended in an endothelial Growth Media-2 (EGM-2) BulletKit system (Lonza, Walkersville, MD, USA) consisting of endothelial basal medium, 2% fetal bovine serum (FBS), human epidermal growth factor, VEGF, human fibroblast growth factor-B, insulin-like growth factor-1, ascorbic acid, and heparin. Cells (1×106 cells/cm2) were seeded on fibronectin-coated dishes or coverslips (both from BD Biosciences, Bedford, MA, USA) supplemented with the EGM-2 BulletKit system and incubated in a 5% CO2 incubator at 37°C. Under daily observation, the first medium change was performed 3 days after plating. Thereafter, the medium was changed every 3 days.
Characterization of early EPCs
After 7 days of cell culture, acetylated-low density lipoprotein labeled with the fluorescent probe 1,1'-dioctadecyl-3,3,3',3'-tetramethyl-indocarbocyanine perchlorate (Dil-acLDL; 1:100; Molecular Probes, Paisley, UK) was added to the medium. Two hours later, the early EPCs were washed with phosphate-buffered saline (PBS) and fixed using 4% paraformaldehyde (Sigma) for 10 minutes. After 3 PBS washes, the early EPCs were treated with fluorescein isothiocyanate (FITC)-labeled Ulex europaeus agglutinin-1 lectin (UEA-1; Sigma) in 0.5% BSA (1:100) at 37°C for 1 hour. After 3 washes in PBS, the cells were then counterstained with bisbenzamide (1 µg/ml; Sigma) for 15 minutes and mounted.
Colony-forming assay
Isolated PBMCs were resuspended in EndoCult medium (STEMCELL Technologies, Vancouver, Canada) or plus different amounts of IS dissolved in PBS and plated in fibronectin-coated 6-well plates. The mean concentration of IS in CKD patients from the data of European Uremic Toxin (EUTox) is: 37.07 ± 26.50 mg/L (range from 0.23 to 53.58 mg/L). Thus, in order to test the toxic effect of IS on early EPC function, we chose various IS concentrations from 2, 10, 50 to 250 mg/L in our study. The cells in the control group were treated with PBS only. After 2 days, the nonadherent cells were collected, and 1x 106 cells were replated onto a fibronectin-coated 24-well plate. Five days later, the number of colony-forming units (CFU) per well was counted. A colony- forming unit of EPCs was defined as a central core of round cells with elongated sprouting cells at the periphery.
Early EPC Chemotaxis Assay
The chemotactic motility of early EPCs was evaluated using Transwell inserts (Costar, Cambridge, USA) with 6.5-mm-diameter polycarbonate filters (8 µm pore size). The upper filter surface was coated with 10 µg gelatin (Merck, Darmstadt, Germany). Early EPCs treated with IS (2, 10, 50 or 250 mg/L) for 5 days was trypsinized and suspended in pure MV2 basal medium (1x105 cells/100 µl ). A total of 100 µl pure cell suspension was loaded into each upper well, and pure MV2 basal medium plus 2% FBS (0.6 mL) was placed in the lower wells. The chamber was incubated at 37°C for 4 hours. The cells were fixed with ice-cold methanol and stained with bisbenzamide (1 µg/mL) for 15 minutes. The migrated cells were quantified with optical microscopy (Leica, Heidelberg, Germany) at 40x magnification using QWIN image analysis software (Leica) by counting the cells that migrated to the lower portion of the filter. Four randomly selected fields in each well were counted. Each test was performed in triplicate, and assays were repeated 3 times.
CKD Patients
Patients with stable CKD (n=128) were recruited from the outpatient department in a medical center (Table 1). Those with acute infection, cardiovascular events in the past 3 months, malignancy, or younger than 18 years were excluded. The etiology of CKD in the study patients included chronic glomerular nephritis (cGN), diabetic nephropathy, and polycystic kidney disease. The patient characteristics were recorded and biochemical parameters measured.
Human endothelial function measurement
FMD on the brachial artery was performed using an ALOKA ultrasound machine (Hitachi Aloka Medical, Ltd, Japan) according to the guidelines set by the International Brachial Artery Reactivity Task Force. The brachial artery was longitudinally imaged above the elbow using a 11.3-MHz probe. The image was recorded for 2 minutes, followed by the induction of forearm ischemia by inflating a cuff below the elbow to 200 mmHg (or 50 mmHg above SBP, whichever was higher) for 5 minutes and rapidly deflating. The resulting reactive hyperemia was recorded for a further 2 minutes. Analyses of all FMDs were performed on the Brachial Analyzer version 5.0 software by a single trained investigator.
Measurement of Artery stiffness
The ABI is calculated by dividing the systolic blood pressure in the right or left ankle by the blood pressure in the right or left arm using a device with an oscillometric method. Brachial-femoral PWV (m/s) is evaluated with two pressure probes and calculated by measuring the pulse transit time and the distance traveled between two selected sites (brachial artery and femoral artery). In the study, the right and left ABI and brachial-ankle PWV were measured by a fixed physician with a dedicated, validated device (VaSera® VS-1000, Fukuda Denshi, Tokyo, Japan) 20 minutes prior to hemodialysis. The measurements of ABI and PWV were performed once for each study patient.
Laboratory Assessment
All blood samples were obtained after overnight fasting or just prior to the dialysis procedure for those on regular hemodialysis. The following analyses were performed: BUN (mg/dL), Cr (mg/dL), Hb (g/dL), Ht (%), sodium (mEq/L), K (mEq/L), Ca (mg/dL), P (mg/dL), total cholesterol (mg/dL), triglyceride (mg/dL), bicarbonate (mmol/L), albumin (g/dL), total IS (mg/L) and free IS (mg/L).
IS measurement
Serum IS was analyzed with LC-MS/MS (4000 QTRAP, USA). Briefly, serum samples were prepared and deproteinized via heat denaturation. The free concentrations of IS were measured in serum ultrafiltrates, which were obtained using Microcon YM-30 separators (Millipore, Billerica, MA, USA) followed by the same sample preparation and analysis that was performed for the serum IS. HPLC was performed at room temperature using a dC18 column (3.0 × 50 mm, Atlantis, Waters). The buffers used were (A) 0.1% formic acid and (B) 1 mM NH4OAc plus 0.1% formic acid in 100% acetonitrile. The flow rate was 0.6 mL/min with a 3.5-min gradient cycling from 90% A/10% B to 10% A/90% B. Under these conditions, IS was eluted at 2.34 min. Standard curves for IS were set at 1, 5, 10, 50, 250, 500 and 1,000 µg/L and correlated with the serum samples with average r2 values of 0.996 ± 0.003. These samples were diluted if the IS concentration exceeded a standard curve. Quantitative results were obtained and calculated in terms of their concentrations (mg/L). The sensitivity of this assay was 1 µg/L for IS.
Statistical analysis
In the human demographic and in vitro data, continuous variables were expressed as the mean ± SD or median and Interquartile Range (IQR) (if variables were not normally distributed), and categorical variables were expressed as the number (percentage) for each item. One-way ANOVA with Tukey's post hoc test was used to compare the differences between the control and study groups in vitro.
To search for bivariate correlations, the correlation between uremic toxins and clinical independent variables was analyzed using Spearman's rank correlation analysis and the Mann-Whitney U test for dichotomous variables. First, a univariate linear analysis was performed with one predictive variable at a time. Multivariate linear analyses were then performed to obtain the predictive factors of the clinical outcomes to analyze the relationship (we put all significant variables in a univariate linear analysis). In a multivariate hierarchical linear analysis, each predictor for FMD was controlled for the participants' main variables and was then controlled for the baseline variables. In the final model, each predictor was controlled for other covariates. Finally, we assessed the normality of the standardized residuals using the Kolmogorov-Smirnov test (p>0.05).
Clinical vascular parameters and uremic toxin levels in different CKD stages were analyzed using the Kruskal-Wallis test with the Mann-Whitney U post hoc test between 7 groups (Bonferroni's correction, α/21=0.0024). All variables were detected to have normality via the Kolmogorov-Smirnov test. The p values were based on two-sided tests and were considered statistically significant if they were less than 0.05. All statistical analyses were conducted using IBM SPSS release 21.0 (IBM, Armonk, New York).
Results
In vitro study
Peripheral blood mononuclear cells (PBMCs) from venous blood exhibited CFUs after 3 to 5 days of culture. The early EPCs were positive for KDR, CD31, CD14, and CD45 (data not shown) and elongated and in a spindle shape with a typical uptake of acetylated low density lipoprotein (Dil-acLDL) and binding of Ulex europaeus agglutinin-1 (UEA-1) lectin (Figure 1A).
Concentration-dependent effects of IS on human early EPC function. (A) Characterization of human early EPCs. Cultured PBMCs uptake of DilacLDL and binding of UEA-1 lectin at day 5 and 7. Bar, 100 µM. (B) EPC migration. Example images of migrated early EPCs treated with various dose of IS as evaluated by transwell method. Bar, 100 µM. (C) Chemotactic motility. Histogram *, p<0.05 compared to 0 mg/L, # , p<0.05 compared to 2 mg/L. 8 subjects were used in each group. (D) Colony-forming units (CFUs). Histogram *, p<0.05 compared to 0 mg/L, # , p<0.05 compared to 2 mg/L. 8 subjects were used in each group.
Concentration-dependent effects of IS on human early EPC function. (A) Characterization of human early EPCs. Cultured PBMCs uptake of DilacLDL and binding of UEA-1 lectin at day 5 and 7. Bar, 100 µM. (B) EPC migration. Example images of migrated early EPCs treated with various dose of IS as evaluated by transwell method. Bar, 100 µM. (C) Chemotactic motility. Histogram *, p<0.05 compared to 0 mg/L, # , p<0.05 compared to 2 mg/L. 8 subjects were used in each group. (D) Colony-forming units (CFUs). Histogram *, p<0.05 compared to 0 mg/L, # , p<0.05 compared to 2 mg/L. 8 subjects were used in each group.
Early EPCs chemotaxis and colony-forming assay
To understand the effect of IS on the migration and growth of human early EPCs, 0, 2, 10, 50 or 250 mg/L of IS were added to the PBMCs cultured in the EndoCult medium. IS inhibited the chemotactic motility of human early EPCs in a concentration-dependent manner. Figure 1B shows the chemotactic motility of early EPCs after treatment with various concentrations of IS. The number of transwell-migrating cells significantly decreased by approximately 30% at 250 mg/L compared with those without treatment (p<0.05) (Figure 1C). In addition to cell migration, the colony-forming ability of early EPCs was assessed by adding different concentrations of IS (0, 2, 10, 50 or 250 mg/L). The morphology of the CFUs of early EPCs in control (no IS) and at various doses of IS were examined. Overall, the CFU number was significantly decreased after treatment with 50 and 250 mg/L of IS (both p<0.05 compared with the control) (Figure 1D).
Human study: Endothelial function in patients with CKD
Table 1 shows the baseline characteristics and biochemistry of 128 CKD patients. The data are presented as the means ± SD or median and interquartile range (IQR) (if values are not normally distributed). The vascular parameters were as follows: flow-mediated dilation (FMD) levels: 2.94±1.69 %; augmentation index (AI): 11.30 (18.87); right pulse wave velocity (R-PWV): 14.99±3.22 m/s; left pulse wave velocity (L-PWV): 15.13±4.47 m/s; right ankle brachial index (R-ABI): 1.05 (0.14); left ankle brachial index (L-ABI): 1.05 (0.13); and β value: 28.41 (20.67). The total IS (T-IS) and free IS (F-IS) were 1.14 (3.02) and 0.01 (0.05) mg/L, respectively. The association between IS and clinical independent parameters is shown in Table 2. T-IS was positively associated with hypertension (p<0.001), cardiovascular disease (CVD) (p=0.039), age (p<0.001), systolic blood pressure (SBP) (p<0.001), diastolic blood pressure (DBP) (p<0.001), blood urea nitrogen (BUN) (p<0.001), creatine (Cr) (p<0.001), potassium (K) (p=0.03), phosphate (P) (p<0.01), R-PWV (p=0.001) and L-PWV (p<0.001) and negatively correlated with estimated glomerular filtration rate (eGFR) (p<0.001), hemoglobin (Hb) (p<0.001), hematocrit (Ht) (p<0.001), FMD (p<0.001) and Ca (p=0.004). Moreover, F-IS had a positive correlation with hypertension (p<0.001), CVD (p=0.024), age (p=0.001), SBP (p<0.001), DBP (p<0.001), BUN (p<0.001), Cr (p<0.001), P (p<0.01), R-PWV (p=0.003) and L-PWV (p<0.001) and a negative correlation with eGFR (p<0.001), Hb (p<0.001), Ht (p<0.001), FMD (p<0.001) and Ca (p=0.001).
Table 3 shows the association between FMD with IS and clinical variables analyzed using univariate and multivariate linear regression models. Independent variables with significant differences in univariate analysis were put into a multivariate analysis using three models. In model 1, FMD was associated with T-IS (B=-0.06, p=0.049). In model 2, FMD was correlated with T-IS (B=-0.08, p=0.005) and age (B=-0.05, p<0.001). In model 3, there were no significant correlations between FMD and independent parameters when all variables were put into a multivariate analysis. This demonstrated that FMD was independently related with serum T-IS after adjusting for other confounding factors. Table 4 shows clinical parameters and uremic toxins levels at different stages of CKD (Kruskal-Wallis test with the Mann-Whitney U post hoc test). FMD, PWV, T-IS and F-IS were associated with CKD stages. Figure 2 shows the agreement between FMD and independent parameters including IS and PWV (Pearson's correlation, p<0.05 for all). Human endothelial function marker - FMD levels gradually decreased as kidney function deteriorated. Patients with hemodialysis had lower FMD levels than those with early CKD (Figure 3).
Demographic factors, clinical vascular parameters and serum IS levels associated with FMD by univariate and multivariate hierarchical linear regression models

Clinical vascular parameters and uremic toxins levels in different stages of CKD analyzed by Kruskal Wallis with post hoc test

Agreement among FMD, IS, and PWV, as analyzed by Pearson's correlation. (A) FMD vs T-IS, r=-0.375, p<0.01. (B) FMD vs F-IS, r=-0.253, p<0.01. (C) FMD vs R-PWV, r=-0.369, p<0.01. (D) FMD vs L-PWV, r=-0.193, p<0.05.
Agreement among FMD, IS, and PWV, as analyzed by Pearson's correlation. (A) FMD vs T-IS, r=-0.375, p<0.01. (B) FMD vs F-IS, r=-0.253, p<0.01. (C) FMD vs R-PWV, r=-0.369, p<0.01. (D) FMD vs L-PWV, r=-0.193, p<0.05.
Human endothelial function, as evaluated using FMD, gradually decreases, as the stage of CKD advances by Kruskal Wallis with post hoc test.
Human endothelial function, as evaluated using FMD, gradually decreases, as the stage of CKD advances by Kruskal Wallis with post hoc test.
Discussion
Our findings showed that IS was not only capable of inhibiting the activity of early EPCs but that it was negatively correlated with FMD, a surrogate of endothelial function, and positively correlated with pulse wave velocity, which reflected arterial stiffness, in patients with CKD. We speculate that this deleterious effect of IS on vascular function, including endothelial function, may result from an imbalance related to the increased injury and decreased repair capacity of EPCs.
The endothelium is a key regulator of vascular homeostasis. Endothelial dysfunction may occur in patients with CKD [20,21] and contributed to a high risk of cardiovascular morbidity and mortality in this population [22]. EPCs are a group of circulating immature cells from the bone marrow that can incorporate into the endothelium and are responsible for endothelial repair and angiogenesis [23]; thus, EPC function is essential in maintaining vascular integrity. However, there is an imbalance between endothelial injury and repair in CKD. One study reported that EPC count and functionality were decreased in patients with renal failure [24]. Moreover, endothelial function was improved and EPC number was increased after the initiation of dialysis [25]. In the present study, the dose-dependent ability of IS to inhibit colony formation and chemotactic motility in EPCs indicated that IS could directly impair early EPCs. Additionally, a previous report showed that indole acetic acid, a uremic toxin similar to IS generated from the gastrointestinal tract, could induce apoptosis in CD133+ cells. In the same study, the indole acetic acid level is also negatively correlated with CD34+CD133+ immature progenitor cell numbers in patients on dialysis [26]. Taken together, these results demonstrated a toxic effect of IS on EPCs.
Patients with CKD were reported to display impaired endothelium-dependent vasodilatation [27]. This phenomenon was also confirmed in the present study, which further showed verse effect could be restored by atorvastatin [30]. This evidence indicated that IS is a vascular toxin. However, few reports have explored the correlation between IS and FMD. We assume that poor endothelium-dependent vasodilatation can be observed in patients with higher IS levels. Notably, IS was an alternative important factor with detrimental effects on human endothelial function. In vitro, IS directly induced endothelial dysfunction through many pathways, including increasing ROS production via an increase in NAD(P)H oxidase activity [31] and inhibiting endothelial proliferation and wound healing [8]. In addition to the direct inhibition of endothelial function mentioned above, our findings indicate that IS-induced EPC dysfunction could further attenuate the repair capacity and angiogenesis for endothelial injury in CKD patients.
The novelty and advantage of the present study is to provide more evidence of the toxicity of IS from in vitro to in vivo. However, this study has some limitations. First, the sample size was small, and all of the subjects were enrolled at one medical center. Second, circulating endothelial microparticles (EMP), an essential marker reflecting endothelial damage, have been reported to correlate with FMD and PWV [32]. In our study, we only showed the adverse effects of IS on early EPCs and the relationship between independent variables and human FMD. It remains unclear whether human FMD, EMP or EPC function could be restored after the reduction of serum IS level using AST 120, an activated charcoal capable of absorbing a precursor of IS, to delay renal function progression [33]. Third, we did not measure the EPC number in patients at different CKD stages. Whether the IS level is correlated with EPC numbers also remains unclear. Fourth, there is lack of direct evidence indicating whether IS-induced EPC dysfunction could be linked to endothelial dysfunction. More prospective investigations are required to answer these questions.
Conclusion
We propose that the dysfunction of early EPCs induced by IS may play a role in the vascular dysfunction observed in patients with CKD. Further studies are required to provide evidence indicating whether the lowering of serum IS has beneficial effects on EPCs and FMD in CKD patients.
Disclosure Statement
The authors of this manuscript state that they do not have any conflict of interests and nothing to disclose.
Acknowledgments
We would like to thank all the patients who were involved in this study and Ting-Yi Tien for technical assistance. This study was supported in part by a Grant from Mackay Memorial Hospital (MMH-104-06 and MMH-105-06).
References
C.-J. Wu and H.-I. Yeh contributed equally to this article and thus share corresponding authorship.