Protein-bound uremic toxin clearance as biomarker of kidney tubular function in diabetic kidney disease

Abstract

Kidney tubular damage is an important prognostic determinant in diabetic kidney disease (DKD). A vital homeostatic function of the proximal tubule is active tubular secretion of waste products via organic anion transporters (OATs), including protein-bound uremic toxins (PBUTs) that accumulate in plasma in tubular dysfunction. We here hypothesize that PBUT clearance may be a sensitive tubular function marker, and tested this in a DKD mouse model and in type 2 diabetic patients. Among the PBUTs with the highest OAT affinity (i.e., indoxyl sulfate (IS), hippuric acid (HA) and kynurenic acid (KA)), plasma concentrations were higher and urinary excretions were lower 6 and 8 months after DKD induction in mice. These parameters correlated better with tubular atrophy, f4/80 scores and tubular injury markers than conventional filtration markers. In patients, the clearance of IS, HA, KA and p-cresyl sulfate (PCS) was associated with urinary neutrophil gelatinase-associated lipocalin (NGAL) and kidney injury molecule-1 (KIM-1), independent of eGFR. In multiple regression analysis, additionally adjusted for relevant risk factors for tubular injury, the clearance of IS, HA and PCS remained significantly associated with urinary NGAL. In conclusion, IS, HA, KA and PCS clearance may represent a biomarker of kidney tubular function in DKD.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-07248-3.

Introduction

Diabetic kidney disease (DKD) is the leading cause of chronic kidney disease (CKD) and subsequent kidney failure requiring renal replacement therapy, and is associated with comorbidities such as cardiovascular diseases and hypertension¹. Early identification of the disease would allow timely intervention to delay its progression to the advanced stages and the onset of concurrent medical conditions. Since DKD is often asymptomatic at early stages, the American Diabetes Association (ADA) and the 2022 Kidney Disease: Improving Global Outcomes (KDIGO) Clinical Practice Guidelines recommend screening diabetic patients for estimated glomerular filtration rate (eGFR) and urinary albumin creatinine ratio (UACR) at least once per year². Still, the lack of sensitivity and reliability of these measures in the diagnosis of DKD is recognized as well³. Recent evidence indicates that the efficiency of eGFR and albuminuria in risk prediction is limited in patients with type 2 diabetes (T2D)^4,5. Many diabetic patients with kidney disease have minimal or no albuminuria, and DKD could progress without albuminuria⁶. Furthermore, both markers primarily represent either loss of function (decrease in eGFR) or damage (albuminuria) at the glomerular level, which has long been thought to be the primary site of injury of diabetes-related kidney impairment. New studies highlighted the involvement of the proximal tubule in the pathogenesis of DKD as well and suggested that tubular lesions correlate better than glomerular lesions with kidney dysfunction^7,8. Tubular injury markers appeared elevated earlier than microalbuminuria and were associated with an early decline in kidney function and progression of DKD⁹.

Proximal tubular secretion is an active process that can be influenced by various conditions and may not parallel GFR. Hyperglycemia in diabetic patients directly affects tubular cells via various mechanisms, including reactive oxygen species (ROS) overproduction and apoptosis, which are linked to DKD progression¹⁰. Protein overload in DKD may induce proximal tubular epithelial cell inflammation and tubulointerstitial fibrosis⁹. Tubular damage is an essential prognostic determinant of DKD¹¹, and eGFR and albuminuria-based diagnostic methods are inefficient in detecting tubular damage¹². Several tubular injury markers, such as kidney injury molecule-1 (KIM-1)¹³ and neutrophil gelatinase-associated lipocalin (NGAL)¹⁴, have been identified for the detection of tubular injury in DKD. Although highly sensitive in detecting acute kidney failure, these markers may be less suitable for determining chronic tubular dysfunction.

Protein-bound uremic toxins (PBUTs) are small molecules (< 500 Da) derived from dietary proteins produced by gut microbial metabolism. Within the systemic circulation, these toxins are bound to plasma proteins, predominantly albumin, and are primarily eliminated by the kidneys through active tubular secretion. This process involves uptake via organic anion transporters (OAT1 and OAT3) at the proximal tubule cell basolateral membrane, followed by luminal efflux through transport proteins including multidrug resistance protein 4 (MRP4) and breast cancer resistance protein (BCRP) at the apical membrane^15,16. Once tubular function is impaired, PBUTs are inefficiently cleared and accumulate in plasma. Increased plasma indoxyl sulfate (IS) concentration has been reported at early stages of DKD, and correlated with renal function in diabetic patients^17,18. Growing evidence has been pointing towards the relation between PBUT plasma concentration and the progression of CKD^19–21 and kidney failure in DKD patients²². Furthermore, reduced clearance of PBUTs is associated with the progression of kidney disease and all-cause mortality²³ and increased intestinal uptake of their precursor (p-cresol) was associated with cardiovascular events independent of eGFR²⁴.

Here we hypothesize that plasma concentration and clearance of PBUTs may serve as biomarkers of kidney tubular function in DKD. To this end, we measured six prototypical PBUTs (IS, hippuric acid (HA), kynurenic acid (KA), p-cresyl glucuronide (PCG), p-cresyl sulfate (PCS) and indole-3-acetic acid (IAA)) in samples available from an experimental long term DKD mouse model and from diabetic patients, and investigated the associations of PBUT parameters with histological and/or biochemical markers of kidney tubular injury.

Results

Preclinical study

Evaluation of tubular damage in DKD mice

High-dose streptozotocin (STZ) induced diabetes mellitus in C57BL/6 mice, ensured by structural and functional features of DKD as documented previously²⁵. For additional data, including morphological details, the reader is referred to that study. The characteristic features of kidney tubular damage were evident after 6 and 8 months following the induction of diabetes (Table 1). Examination of tubular histology using tubular atrophy scores (TAS) showed a tenfold increase after 6 months and 7.5-fold after 8 months (Table 1 and Suppl. Figure 1). Compared to the control mice, diabetic mice showed a significant increase in plasma and urinary markers of tubular injury. In contrast, markers for glomerular filtration (plasma Cr and Cys C) and glomerular damage (albuminuria) were not significantly elevated except for albuminuria after 8 months (Table 1).

Table 1.

Characteristics of the control and DKD mice after 6 and 8 months.

Parameters	6 months		8 months
	Control	DKD	Control	DKD
N (% male)	15(40)	16 (44)	19 (47)	23 (61)
Urine volume (ml)	1.52 ± 0.96	15.3 ± 10.5***	1.57 ± 1.09	21.6 ± 26.07***
Plasma creatinine (μg/ml)	1.97 ± 0.83	2.55 ± 1.10	1.55 ± 0.53	1.68 ± 0.43
Plasma cystatin C (μg/ml)	2.18 ± 1.59	2.50 ± 1.72	0.95 ± 0.53	1.51 ± 1.76
Albuminuria (mg/g Cr)	326 ± 91	420 ± 141	268 ± 158	413 ± 174**
Kim-1 gene expression	0.04 ± 0.05	0.04 ± 0.04	0.01 ± 0.02	0.03 ± 0.03*
Plasma NGAL (ng/ml)	107 ± 40.5	174 ± 51.8***	76.5 ± 13.0	145 ± 14.8***
Urine NGAL (mg/g Cr)	0.23 ± 0.09	2.02 ± 2.02***	0.46 ± 0.31	39.1 ± 44.5**
Tubular atrophy scores	0.17 ± 0.10	1.72 ± 0.50***	0.19 ± 0.15	1.43 ± 0.49***
Creatinine clearance (ml/24 h)	246 ± 147	280 ± 122	337 ± 164	537 ± 432

Data are mean ± SD. The non-parametric Mann–Whitney U test or parametric unpaired t-test was performed for statistical analysis, based on the data normality determined by the Kolmogorov–Smirnov test. ***P < 0.001, **P < 0.01, *P < 0.05. Kim-1: kidney injury molecule-1; NGAL: neutrophil gelatinase associated lipocalin.

Significant values are in [bold].

DKD mice show increased PBUT plasma concentrations and decreased clearance

Among the PBUTs with high OAT affinity, i.e., IS, HA and KA²⁶, plasma concentrations were 1.7-, 2.4- and 1.9-fold higher after 6 months and 1.9-, 2.4- and 2.2-fold higher after 8 months in DKD, respectively (Fig. 1A–C). Urine PBUT/creatinine ratios normalized for plasma PBUT, used as a surrogate for PBUT clearance, were 2.4-, 5.4- and 2.3-fold lower after 6 months and 2.3-, 1.6-, 2.1-fold lower after 8 months (Fig. 1D-F), and fractional excretions were 1.6-, 4.4-, and 1.8- fold lower after 6 months and 1.7-, 1.4-, 1.7-fold lower after 8 months in DKD (Fig. 1G-I), respectively. Other PBUTS, i.e., PCG, PCS and, IAA were not significantly affected (Suppl. Figure 2).

Fig. 1.

PBUTs plasma concentration, surrogate clearance and fractional excretion in DKD mice and healthy controls. Data are presented as mean ± SD. The non-parametric Mann–Whitney U test or parametric unpaired t-test was performed for statistical analysis based on the data normality determined by the Kolmogorov–Smirnov test. ***P < 0.001, **P < 0.01, *P < 0.05. DKD: diabetic kidney disease; IS: indoxyl sulfate; HA: hippuric acid; KA: kynurenic acid.

PBUTs correlate better with tubular atrophy than commonly used filtration markers

Next, we evaluated whether the PBUT parameters correlate with the histopathology. All animals were included in the analysis (i.e., controls and DKD mice, sacrificed at 6 and 8 months). Plasma concentration of IS (r = 0.49; P < 0.001), HA (r = 0.68; P < 0.001) and KA (r = 0.40; P = 0.002) positively and surrogate clearance of IS (r = -0.44; P = 0.001), HA (r = -0.37; P = 0.007) and KA (r = -0.44; P = 0.11) negatively correlated with TAS (Suppl. Figure 3). The mice were then grouped based on their TAS: 0 to 0.1 (n = 7), 0 to 0.3 (n = 21), 0 to 0.5 (n = 27), 0 to 1.0 (n = 32), 0 to 1.3 (37), 0 to 1.5 (n = 48), 0 to 2.0 (n = 55) to evaluate the relative sensitivity among commonly used biomarkers and PBUTs to various grades of tubular atrophy. Plasma HA correlated with tubular atrophy starting already at TAS ≤ 0.5 and became stronger with higher TAS. For plasma IS and KA, the correlation became significant at TAS ≤ 1.0, for HA surrogate clearance at TAS ≤ 1.3 and for albuminuria, IS and KA surrogate clearance at TAS ≤ 1.5 (Table 2). Of note, the plasma concentration and surrogate clearance of IS and KA correlated with TAS already at a very low level (TAS ≤ 0.1) but not significantly due to the small sample size (Table 2). In comparison, plasma Cys C, plasma Cr and Cr clearance did not correlate with TAS at any level. PBUTs with non-significant changes in plasma concentration, surrogate clearance and fractional excretion did not correlate with tubular atrophy (Suppl. Table 1).

Table 2.

Correlation of PBUTs and commonly used markers with tubular atrophy scores.

Markers	Tubular atrophy scores
	TAS ≤ 2.0	TAS ≤ 1.5	TAS ≤ 1.3	TAS ≤ 1.0	TAS ≤ 0.5	TAS ≤ 0.3	TAS ≤ 0.1
Plasma hippuric acid	0.71***	0.74***	0.67***	0.55**	0.45*	0.33	0.06
Plasma indoxyl sulfate	0.54***	0.58***	0.53***	0.37*	0.16	0.25	0.63
Plasma kynurenic acid	0.46***	0.52***	0.37*	0.38*	0.23	0.25	0.71
Hippuric acid surrogate clearance	− 0.47***	− 0.54***	− 0.40*	− 0.27	− 0.16	− 0.15	− 0.08
Kynurenic acid surrogate clearance	− 0.48***	− 0.54***	− 0.32	− 0.37	− 0.19	− 0.17	− 0.70
Indoxyl sulfate surrogate clearance	− 0.47***	− 0.53***	− 0.26	− 0.16	− 0.22	− 0.47*	− 0.91
Albuminuria	0.48***	0.48**	0.19	0.31	0.33	0.28	0.37
Plasma creatinine	0.14	0.20	0.12	0.13	− 0.04	− 0.07	0.20
Plasma cystatin C	− 0.03	0.14	0.00	− 0.05	0.10	0.16	− 0.34
Creatinine clearance	0.09	0.07	0.03	0.16	− 0.02	0.01	− 0.03

Analyses include both control and diabetic animals, sacrificed at 6 and 8 months. Non-parametric Spearman correlation was performed and the data are presented as Spearman rank, where ‘- ‘ indicates negative correlation. ***P < 0.001, **P < 0.01, *P < 0.05. PBUTs: protein-bound uremic toxins; sur.cl: surrogate clearance; cl: clearance.

Significant values are in [bold].

PBUTs correlate with interstitial inflammation in the kidney

Renal interstitial inflammatory cell infiltration, one of the common features in DKD^27,28, was evident by an increase of f4/80 positive macrophages after 6 months. Of note, 8 months data were not available. Diabetic mice showed a 1.9-fold increase in f4/80 positive cells in the kidney (141.9 ± 10 vs. 75.5 ± 3.4, P < 0.01) per high power field compared to control mice. This f4/80 score, representing interstitial macrophage infiltration, positively correlated with the plasma concentration and negatively with surrogate clearance and fractional excretion of IS and HA, but not with KA and the conventional markers (Fig. 2, Suppl. Table 2). PBUTs with non-significant changes in plasma concentration, surrogate clearance and fractional excretion did not correlate with tubular atrophy (Suppl. Table 1).

Fig. 2.

Correlation of f4/80 score with plasma concentration, clearance and fractional excretion of IS (A–C), HA (D–F), Cr clearance (G), plasma Cr (H), plasma Cys C (I). Non-parametric spearman correlation was performed, and the data are presented as spearman rank, where ‘-’ indicates negative correlation. **P < 0.01, *P < 0.05. IS: indoxyl sulfate; HA: hippuric acid; KA: kynurenic acid; Cr: creatinine; Cys C: cystatin C; uCr: Urinary creatinine.

PBUTs correlate better with tubular damage markers than commonly used filtration markers

Plasma and urinary NGAL have been demonstrated as early markers for acute kidney injury (AKI)^29–32 and in particular for tubular damage³³. Systematic review and meta-analyses suggested both serum and urinary NGAL as sensitive biomarkers of kidney disease in DKD patients, with pooled sensitivity and specificity ranging from 0.90 to 0.98^14,34. Cys C is filtered by the glomerulus but catabolized completely after uptake by proximal tubule cells. Impaired reabsorption by the tubules may lead to its increased urinary excretion in DKD^35–37 Kim-1 is a transmembrane tubular protein only detectable in acute events of kidney injury³⁸, and studies indicated that urinary Kim-1 reflects lesions of proximal tubule at early stage of DKD³⁹.

Plasma PBUT concentrations correlated better with tubular injury (plasma and urinary NGAL) and function (urinary Cys C) markers than the markers for glomerular filtration (plasma Cr and Cys C) and damage (albuminuria) both at 6 and 8 months, except for KA with plasma NGAL after 6 months and IS with urinary Cys C after 8 months (Table 3, Suppl. Figures 4–11). Similarly, PBUTs surrogate clearance and fractional PBUT excretion showed a significant negative correlation with the tubular markers at 6 months. However, after 8 months the correlation between surrogate clearance and fractional excretion of IS and HA and tubular markers was lost. PBUTs with non-significant change in plasma concentration, clearance and fractional excretion did not show notable correlation with tubular markers, except for surrogate clearance of PCG, PCS and IAA with plasma NGAL after 6 months (Suppl. Table 1). Plasma and urinary Kim-1 were undetectable in our samples; however, the expression of Kim-1 gene (Havcr1) in kidney tissues was significantly higher in diabetic mice after 8 months (Table 1) but did not correlate with any of the markers evaluated here.

Table 3.

Correlation of PBUTs and commonly used filtration markers with tubular markers.

Markers	Plasma NGAL		uNGAL/uCr		uCysC/uCr
	6 months	8 months	6 months	8 months	6 months	8 months
Plasma indoxyl sulfate	0.50**	0.44**	0.50*	0.36*	0.54**	0.32
Plasma hippuric acid	0.54**	0.65***	0.63**	0.66***	0.74***	0.45*
Plasma kynurenic acid	0.23	0.58***	0.43*	0.48**	0.53*	0.40*
Indoxyl sulfate surrogate clearance	− 0.45*	− 0.09	− 0.64**	− 0.12	− 0.45*	− 0.28
Hippuric acid surrogate clearance	− 0.58**	− 0.30	− 0.68***	− 0.23	− 0.71***	− 0.14
Kynurenic acid surrogate clearance	− 0.40*	− 0.50**	− 0.44*	− 0.56**	− 0.47*	− 0.36*
Indoxyl sulfate fractional excretion	− 0.57**	0.06	− 0.48*	0.09	− 0.38	− 0.14
Hippuric acid fractional excretion	− 0.48*	− 0.13	− 0.58**	0.02	− 0.59**	− 0.00
Kynurenic acid fractional excretion	− 0.39	− 0.40*	− 0.36	− 0.41*	− 0.41	− 0.23
Plasma creatinine	0.23	− 0.05	0.31	0.24	0.05	0.17
Creatinine clearance	0.31	− 0.11	0.17	0.45*	0.40	0.10
Plasma Cystatin C	0.33	− 0.11	0.41	− 0.14	0.05	− 0.14
Albuminuria	0.39	− 0.05	0.27	0.51**	0.36	0.52**

Non-parametric Spearman correlation was performed, and the data are presented as Spearman rank, where ‘-’ indicates negative correlation. ***P < 0.001, **P < 0.01, *P < 0.05. PBUTs: protein-bound uremic toxins; cl: clearance; NGAL: neutrophil gelatinase-associated lipocalin; Cr: creatinine; Cys C: cystatin C; uCr: Urinary creatinine.

Significant values are in [bold].

Clinical study

Patient characteristics

Baseline characteristics and medications (including ACEi and ARB) of the study population (N = 33) are summarized in Table 4 and Suppl. Table 3, respectively. The patients included in this study were at an early stage of kidney function decline with an eGFR between 50–90 ml/min/1.73 m2 in the majority of patients (Suppl. Figure 12). Among PBUTs, the clearance (mean ± SD) ranged from 6.1 ± 6.3 ml/min for IAA to 377.0 ± 271.5 ml/min for HA (Table 5).

Table 4.

Patient demographic and characteristics.

Characteristic	Total (N = 33)
Age (years)	61 ± 9
Sex (female/male) (n)	8/25
Caucasian (n, %)	30 (91)
Body weight (kg)	96 ± 22
Body mass index (kg/m2)	31 ± 5.5
Systolic blood pressure (mmHg)	142 ± 13
Diastolic blood pressure (mmHg)	78 ± 6.7
Smoking status (n, %)	9 (27.3)
Diabetes duration (years)	10 ± 7
HbA1c (mmol/mol)	59 ± 15
Cardiovascular disease history (n, %)a	10 ± 30
Estimated GFR (ml/min/1.73 m2)	72 ± 19
Serum creatinine (µmol/L)	91.3 ± 20.5
Urea (mmol/L)	6.8 ± 2.2
Plasma cystatin C (ng/ml)	325 ± 8
UAER (mg/24 h)*	613 (208–1102)
UACR (mg/mmol)*	51.1 (22.2–89.8)

Data are given as mean ± SD and *median (25th to 75th percentile); ^aCardiovascular disease history defined as history of coronary artery disease or peripheral vascular disease; UAER: urinary albumin excretion rate; UACR: urinary albumin creatinine ratio.

Table 5.

Plasma concentration and clearance of PBUTs.

Toxins	Plasma concentration (ng/ml)	Clearance (ml/min)
Indoxyl sulfate	2860.3 ± 1924.8	46.2 ± 35.3
Hippuric acid	1870.3 ± 1154.4	377.0 ± 271.5
Kynurenic acid	12.0 ± 4.4	133.6 ± 53.8
p-cresyl glucuronide	78.4 ± 75.5	275.8 ± 221.9
p-cresyl sulfate	9715.8 ± 6957.3	12.8 ± 5.8
Indole-3-acetic acid	1003.7 ± 1028.5	6.1 ± 6.3

Data are given as mean ± SD.

PBUT clearance correlates better with tubular damage markers than commonly used filtration markers

Table 6 depicts comparative ranks of Pearson correlation of log-transformed PBUT clearance and commonly used biomarkers with urinary markers of kidney tubular injury (urinary NGAL and KIM-1) and function (urinary Cys C). The clearance of IS, HA, KA and PCS negatively correlated with urinary NGAL and KIM-1, whereas conventional markers, including eGFR and UAER, did not show any correlation except for creatinine clearance with urinary NGAL. Urinary Cys C only showed a negative correlation with IAA clearance (Table 6). However, in contrast to our preclinical findings, plasma PBUT concentrations did not correlate with tubular injury markers.

Table 6.

Comparative rank of (Pearson) correlation of (log-transformed) PBUT clearance and commonly used markers with kidney tubular injury markers.

Markers	Correlation with uNGAL/uCr	Correlation with uKIM-1/uCr	Correlation with uCysC/uCr
Indoxyl sulfate clearance	− 0.51**	− 0.48**	0.05
Hippuric acid clearance	− 0.54**	− 0.49**	− 0.15
Kynurenic acid clearance	− 0.52**	− 0.51**	− 0.08
p-cresyl sulfate clearance	− 0.56**	− 0.45*	− 0.32
p-cresyl glucuronide clearance	0.01	0.05	0.24
Indole-3-acetic acid clearance	− 0.33	− 0.27	− 0.44*
Serum creatinine	− 0.06	− 0.11	0.16
Urea	− 0.23	− 0.32	− 0.04
Plasma cystatin C	− 0.11	− 0.24	0.01
eGFR	− 0.11	− 0.00	− 0.15
UAER	− 0.13	− 0.18	0.34
Creatinine clearance	− 0.41*	− 0.33	− 0.31

cl, clearance; uCr, urine creatinine; NGAL, neutrophil gelatinase-associated lipocalin; KIM-1, kidney injury molecule-1; eGFR, estimated glomerular filtration rate; UAER, urinary albumin excretion rate.

Significant values are in [bold].

Association of PBUT clearance with tubular damage markers

Lower clearances of IS, HA, KA and PCS were associated with increased concentration of urinary NGAL and KIM-1 in the unadjusted model (Model 1), with an estimated decrease in (log-transformed) clearance of 6.87–12.45 ml/min and 0.45–0.89 ml/min for every µg/mmol Cr increase in urinary NGAL and KIM-1, respectively (Tables 7 and 8). These associations remained significant after adjustment for eGFR (Model 2) (Tables 7 and 8). Further adjustment for age, sex and race (Model 3) resulted in a slight attenuation of the estimates, however, the clearance of IS, HA and PCS remained significantly associated with urinary NGAL and KIM-1. In Model 4, after adjustment with all available risk factors, lower clearance of PCS remained significantly associated with urinary NGAL (Table 7). In contrast, UAER and creatinine clearance showed no significant association with urinary KIM-1 and NGAL, except for creatinine clearance with urinary NGAL in model 1 and 2, while their estimates were very low (Tables 7 and 8). Next, we tested these associations with an average clearance of 4 PBUTs: IS, HA, KA and PCS, which was significantly associated with urinary NGAL and KIM-1 in all 4 models, with an estimated decrease in (log-transformed) clearance of 8.03–12.99 ml/min and 0.56–0.90 ml/min for every µg/mmol Cr increase in urinary NGAL and KIM-1, respectively (Table 9). The sensitivity of these associations was further evaluated in a subgroup of patients stratified for eGFR. The average clearance of IS, HA, KA and PCS was significantly associated with urinary NGAL and KIM-1 in all 4 models with high eGFR subgroup (> 70 mL/min/1.73 m2) (Suppl. Table 4).

Table 7.

Association of PBUT clearance (log-transformed) with (creatinine normalized) urinary NGAL (uNGAL/uCr) in multiple regression model.

Biomarker	Model 1		Model 2		Model 3		Model 4
	Estimate (Standard error)	P value	Estimate (Standard error)	P value	Estimate (Standard error)	P value	Estimate (Standard error)	P value
Indoxyl sulfate clearance	− 7.53 (2.34)	0.004	− 7.57 (2.49)	0.005	− 5.62 (2.22)	0.018	− 4.75 (2.73)	0.096
Hippuric acid clearance	− 6.87 (2.00)	0.002	− 7.01 (2.11)	0.003	− 5.04 (1.83)	0.011	− 3.86 (2.12)	0.083
Kynurenic acid clearance	− 12.45 (3.83)	0.003	− 12.40 (3.88)	0.003	− 7.41 (3.75)	0.059	− 4.03 (5.04)	0.433
p-cresyl sulfate clearance	− 11.75 (3.23)	0.001	− 11.78 (3.37)	0.002	− 9.33 (2.77)	0.002	− 8.38 (3.47)	0.025
p-cresyl glucuronide clearance	0.13 (1.66)	0.940	0.12 (1.68)	0.943	0.12 (1.42)	0.932	1.39 (1.52)	0.373
Indole-3-acetic acid clearance	− 2.91 (1.55)	0.070	− 2.91 (1.66)	0.090	− 2.71 (1.27)	0.043	− 2.38 (1.37)	0.098
UAER	0.00 (0.00)	0.483	0.00 (0.00)	0.343	0.00 (0.00)	0.223	0.00 (0.00)	0.221
Creatinine clearance	− 0.01 (0.01)	0.021	− 0.01 (0.01)	0.023	− 0.01 (0.00)	0.054	− 0.01 (0.01)	0.127

#UAER was not included as a confounder.

cl, clearance; UAER, urinary albumin excretion rate; Cr, Creatinine; uCr, urine creatinine; NGAL, neutrophil gelatinase-associated lipocalin; Model 1: Unadjusted; Model 2: Model 1 + eGFR; Model 3: Model 2 + age, sex and race; Model 4: Model 3 + UAER, smoking status, BMI and blood pressure.

Significant values are in [bold].

Table 8.

Association of PBUT clearance (log-transformed) with (creatinine normalized) urinary KIM-1 (uKIM-1/uCr) in multiple linear regression models.

Biomarker	Model 1		Model 2		Model 3		Model 4
	Estimate (Standard error)	P value	Estimate (Standard error)	P value	Estimate (Standard error)	P value	Estimate (Standard error)	P value
Indoxyl sulfate clearance	− 0.52 (0.17)	0.006	− 0.54 (0.18)	0.006	− 0.41 (0.16)	0.021	− 0.36 (0.20)	0.085
Hippuric acid clearance	− 0.45 (0.15)	0.006	− 0.48 (0.16)	0.005	− 0.35 (0.14)	0.018	− 0.26 (0.15)	0.103
Kynurenic acid clearance	− 0.89 (0.28)	0.003	− 0.89 (0.28)	0.004	− 0.51 (0.28)	0.086	− 0.25 (0.37)	0.505
p-cresyl sulfate clearance	− 0.68 (0.25)	0.011	− 0.71 (0.26)	0.011	− 0.52 (0.23)	0.030	− 0.37 (0.27)	0.194
p-cresyl glucuronide clearance	0.03 (0.12)	0.784	0.03 (0.12)	0.788	0.01 (0.11)	0.900	0.10 (0.11)	0.378
Indole-3-acetic acid clearance	− 0.17 (0.11)	0.144	− 0.19 (0.12)	0.133	− 0.17 (0.10)	0.092	− 0.15 (0.10)	0.144
UAER	0.00 (0.00)	0.314	0.00 (0.00)	0.291	0.00 (0.00)	0.301	0.00 (0.00)#	0.270
Creatinine clearance	− 0.00 (0.00)	0.061	− 0.00 (0.00)	0.066	− 0.00 (0.00)	0.126	− 0.00 (0.00)	0.233

#UAER was not included as a confounder.

cl, clearance; UAER, urinary albumin excretion rate; Cr, Creatinine; uCr, urine creatinine; KIM-1, kidney injury molecule-1; Model 1: Unadjusted; Model 2: Model 1 + eGFR; Model 3: Model 2 + age, sex and race; Model 4: Model 3 + UAER, smoking status, BMI and blood pressure.

Significant values are in [bold].

Table 9.

The association of average clearance of 4 PBUTs with urinary tubular injury markers in multiple linear regression models.

Biomarker	Model 1		Model 2		Model 3		Model 4
	Estimate (Standard error)	P value	Estimate (Standard error)	P value	Estimate (Standard error)	P value	Estimate (Standard error)	P value
Association of (log-transformed) average clearance of 4 PBUTs (IS, HA, KA and PCS) with uNGAL/uCr
Average PBUT clearance	− 12.48 (2.63)	< 0.001	− 12.99 (2.77)	< 0.001	− 9.10 (2.68)	0.002	− 8.03 (3.59)	0.036
Association of (log-transformed) average clearance of 4 PBUTs (IS, HA, KA and PCS) with uKIM1/uCr
Average PBUT clearance	− 0.83 (0.20)	< 0.001	− 0.90 (0.21)	< 0.001	− 0.65 (0.20)	0.004	− 0.56 (0.26)	0.046

uCr, urine creatinine; IS, indoxyl sulfate; HA, hippuric acid; KA, kynurenic acid; PCS, p-cresyl sulfate; NGAL, neutrophil gelatinase-associated lipocalin; KIM-1, kidney injury molecule-1; Model 1: Unadjusted; Model 2: Model 1 + eGFR; Model 3: Model 2 + age, sex and race; Model 4: Model 3 + UAER, smoking status, BMI and blood pressure.

Similar to Pearson correlations, only IAA clearance was associated with urinary Cys C after adjustment with eGFR and this association remained significant in all models (Suppl. Table 5).

Contrary to the preclinical findings (Table 3), but like the Pearson correlations (Table 6), the plasma concentration of PBUTs were not associated with kidney tubular injury markers (Suppl. Table 6).

Discussion

In the present study, we hypothesized that PBUTs plasma concentration and clearance may be sensitive tubular function markers in DKD. This was tested in a long-term streptozotocin-induced DKD mouse model and in type 2 diabetic patients. Major finding of our study is that, in both experimental and human DKD, reduced clearance of PBUTs with high OAT1 affinity, i.e., IS, HA and KA, and PCS in human DKD, were related to tubular injury, and this remained significant after adjustment for glomerular filtration (using fractional PBUT excretion in experimental DKD and multiple regression in human DKD). The relationship was present even for mild tubular atrophy in experimental DKD and mild human DKD (eGFR > 70 mL/min/1.73 m²), suggesting that PBUT clearance is a sensitive marker for tubular function.

Although DKD is traditionally viewed primarily as a glomerular disease, tubulointerstitial injury is a major feature of DKD with important prognostic significance^9,40,41. Tubular injury was shown to be an early event in the pathogenesis of DKD, which may precede albuminuria⁴⁰. In diabetic kidneys, increased SGLT2-mediated glucose uptake in proximal tubular cells induces oxidative stress and inflammation^9,40–42 leading to tubulointerstitial histological changes, including hypertrophy, thickening of the tubular basement membrane and interstitial inflammation, subsequently contributing to renal structural modifications such as tubular atrophy and interstitial fibrosis. These alterations correlate with kidney function decline and ultimately play a significant role in the development and progression of DKD^11,43–46. Early detection of incipient DKD is important as timely initiation of adequate treatment may slow progression and improve patient outcomes. This study suggests that PBUT clearance may play a role in this.

PBUTs are dietary metabolites generated by gut microbial metabolism and eliminated primarily by active tubular secretion in the renal proximal tubule via influx transporters at the basolateral side, such as OAT1 and 3, and efflux transporters at the apical side, such as MRP4 and BCRP^16,47. Numerous studies conducted with various (animal and human) models demonstrated the uptake of PBUTs by these transporters, with IS, HA and KA having the highest affinities^48–51. In tubular dysfunction, particularly impairment of the secretory mechanisms in the proximal tubule, this secretion process is compromised. As a result, PBUTs cannot be effectively transported into the urine and, consequently, accumulate in plasma^15,16. Previous studies emphasized the importance of plasma concentration and clearance of PBUTs in the assessment of the secretory function of kidneys^12,51. However, PBUTs as indicators of secretory function of kidney tubules has not yet been investigated in DKD. Our data suggest that PBUT clearance reflects renal tubular dysfunction in DKD and may be used for early detection.

In experimental DKD plasma PBUT concentration was also related to tubular injury but in human DKD this was not the case. The PBUTs studied here are tryptophan and phenol derivatives, generated endogenously by gut microbial metabolism of dietary proteins. Their plasma concentrations may, therefore, be influenced by various factors, including variability in protein intake, colonic microbial metabolism and hepatic metabolism process^52–54. In uremic condition, the composition and function of gut microbiome may change, leading to inter-individual and inter-species variation in colonic metabolism of dietary proteins and subsequent metabolites⁵³. Due to the experimental setup, these factors are comparable among the mice thus variation in plasma PBUT is primarily determined by tubular function, while in humans these factors may have caused large inter-individual variability in plasma PBUTs. In principle, fractional excretions should represent renal tubular function more accurately than clearance. However, the correlation strengths of fractional excretions in our preclinical data were weaker than those of clearances, possibly due to a small proportion of creatinine that is excreted via tubular secretion which may attenuate the association (when both denominator and numerator of the ratio partly represent tubular function). However, this was not found in diabetic patients where fractional excretion was not affected by creatinine secretion (Suppl. Table 7).

On the other hand, the efficiency of PBUTs in assessing tubular function might be affected by, for example, saturation of or competition for in- and efflux transporters, especially in DKD patients on multiple drugs that excrete via tubular secretion^55,56.

Tubular injury (such as NGAL, KIM-1) and function (such as Cys C) markers, showed promising efficacy in early detection of DKD in various preclinical and clinical studies^5,39,57–59. In our study, PBUT parameters demonstrated better relations with these markers than conventional markers such as creatinine clearance and albuminuria. This is in accordance with kidney physiology since PBUTs are primarily excreted at the tubular level while tubular secretion only represents a minor fraction of creatinine clearance, and both glomerular and tubular injury may cause albuminuria. The independent associations of PBUT clearance with urinary NGAL and KIM-1 suggest that PBUT clearance may -like these markers- be used for early DKD detection. However, while NGAL and KIM1 are induced by injured proximal epithelial cells, PBUT clearance reflects function of the secretory transport system of these cells. In our study, we were unable to assess transporter expression due to insufficient sample material. Literature suggested that these transporters are downregulated under diabetic conditions, likely as a consequence of hyperglycemia induced oxidative stress and impaired insulin signaling^60–62. Notably, such downregulation may theoretically precede the increase in levels of tubular injury markers. Remarkably, clearance of IS, HA and KA was not associated with urinary Cys C reflecting dysfunction of the cubilin and megalin mediated endocytic uptake of proteins in proximal tubular cells^63,64.

Also for prediction of progression of DKD, urinary NGAL and KIM-1 appear to be the most promising among the available markers, although some longitudinal studies have reported a suboptimal performance of KIM-1 in predicting renal failure in long-term diabetes^5,57. In a prospective cohort study of 3416 CKD patients PBUT clearance, but not PBUT plasma concentration, was associated with CKD progression and all-cause mortality²³. A case–control study in DKD patients showed an association between plasma concentration of pCS, but not IS and HA, and the risk of kidney failure²². Long term prospective studies should elucidate whether PBUT clearance also has prognostic significance in DKD.

While our findings support PBUTs as potential biomarkers of early DKD, our study has important limitations. Firstly, since this was a retrospective study, we did not have all samples available for statistical analyses and for investigating the expression of renal transporters involved in PBUT clearance. Secondly, diabetes was induced in mice by streptozotocin, which may have contributed to damage to tubular epithelial cells via activation of p53 signaling in acute phases⁶⁵, however, this was unexpected to influence tubular damage after 6 months. Thirdly, we could not use 24 h urine excretion of PBUTs to calculate clearance in mice due to a suspected systemic error during urine collection from the metabolic cages. Finally, the cohort of diabetes patients was rather small, and we could not evaluate the association of PBUT clearance with histology score of tubular atrophy from patient’s biopsies or with the gold standard measurements of tubular function, such as the clearance of para-amino hippuric acid.

In conclusion, we demonstrated that plasma concentration and clearance of IS, HA and KA correlate with histological scores of kidney tubular damage and with available tubular markers, stronger than with conventional markers creatinine clearance and albuminuria in a long-term experimental mouse DKD model. In a cohort of type 2 diabetic patients, clearance of IS, HA, and PCS associated with kidney tubular injury markers independent of eGFR and other relevant confounders. Together, these findings suggest that the clearance of IS, HA, KA and PCS may serve as biomarkers for tubular dysfunction in DKD. The evaluation of tubular function in addition to the filtration function could facilitate detecting site-specific injury and may help in early detection of DKD. Future studies should focus on validation in large cohorts of DKD patients.

Materials and methods

Reagents

IS, KA, HA, IAA and acetic acid (HAc; LC–MS grade) were purchased from Sigma-Aldrich (St. Louis, MO, USA). PCS and PCG were purchased from AlsaChim (Illkirch-Graffenstaden, France). Water (LC–MS grade), acetonitrile (ACN; HPLC-S grade) and methanol (HPLC grade) were obtained from Biosolve (Valkenswaard, The Netherlands). Formic acid (analytical grade) was obtained from Merck (Darmstadt, Germany). Ultrapure water was produced by a Milli-Q^® Advantage A10 Water Purification Systems (Merck, The Netherlands). The isotopically labeled internal standards d5-KA and d5-IAA were purchased from CDN Isotopes (Pointe-Claire, Quebec, Canada), 13C6-IS and d5-HA and were purchased from Cambridge Isotope Laboratories (Andover, MA, USA), d7-PCS (potassium salt) was purchased from IsoSciences (Ambler, PA, USA) and d7-PCG was purchased from Toronto Research Chemicals (North York, Ontario, Canada).

Diabetic nephropathy mouse model

We used a long-term streptozotocin (STZ)-induced murine DKD model previously documented by Falke et al.²⁵. Briefly, diabetes mellitus was induced in C57BL6/J mice by a single intraperitoneal injection of 200 mg/kg STZ, concentration 30 mg/ml, dissolved in 100 mM sodium citrate buffer, pH 4.5 (Sigma-Aldrich, St. Louis, MO, USA). The buffer alone was used for the control animals. 24 h urine was collected by metabolic cages. Mice were sacrificed 6 and 8 months post-STZ administration using a KXA injection (an anesthetic combination of ketamine, xylazine, and acepromazine), and plasma and kidney tissues were harvested. All experiments were approved of the Experimental Animal Ethics Committee of the University of Utrecht (DEC number 2009.II.11.129), and performed in accordance with their guidelines and regulations. This study used samples (stored at − 80 °C) based on availability (6 months: 15 control vs. 16 DKD; 8 months: 19 control vs. 23 DKD). The methods for tubular atrophy (TA) scoring and macrophage infiltration were documented previously²⁵. The study is reported in accordance with ARRIVE guidelines.

Tubular atrophy and F4/80 scoring

Renal tissues were paraffin-embedded and sectioned at a thickness of 3 μm. Periodic acid–Schiff (PAS) staining was performed to evaluate tissue morphology. Tubular atrophy was assessed independently by two experienced observers who were blinded to the slide identities. The scoring was done in 10 randomly selected cortical fields at × 100 magnification using a semiquantitative scale. The average score across the ten fields was used for subsequent statistical analysis. Images were captured using a Nikon Eclipse E800 microscope equipped with a Nikon DXM1200 digital camera and Nikon ACT-1 software (version 2.70; Nikon Netherlands, Lijnden, the Netherlands).

Macrophage infiltration was assessed in frozen kidney sections. Sections were fixed in acetone, blocked, and incubated with a rat anti-mouse F4/80 antibody (Serotec Benelux, Oxford, UK) to detect macrophage-specific antigen. Subsequently, sections were incubated with horseradish peroxidase (HRP)-conjugated rabbit anti-rat secondary antibody (Dako, Glostrup, Denmark), followed by goat anti-rabbit Powervision-HRP (Klinipath, Duiven, The Netherlands). Detection was performed using Nova Red substrate (Vector Laboratories, Burlingame, CA, USA), and sections were counterstained with hematoxylin. F4/80-positive cells were counted per high-power field (field area 0.245 mm²) in a blinded manner. Only cells showing F4/80 positivity in both the nucleus and cytoplasm were considered truly positive.

Type 2 diabetic patient cohort

The samples of type 2 diabetic patients were obtained from a prospective, randomized, double-blind, placebo-controlled cross-over clinical trial, which was registered in the Netherlands Trial Register (NTR 4439), was approved by the Medical Ethics Committee of the University Medical Center Groningen, the Netherlands (METc 2014/111), and was conducted in accordance with their guidelines and regulations, the Declaration of Helsinki and Good Clinical Practice Guidelines⁶⁶. All participants (N = 33) provided written informed consent before any study procedure started. This clinical trial involved 3 consecutive cross-over treatment periods of 6 weeks with washout periods of 6 weeks in between. The current study utilized the baseline samples from this trial when patients were unaffected by the test drugs.

PBUTs quantification in plasma and urine by LC–MS/MS

Equipment and LC–MS/MS condition

An Accela LC system (quaternary pump and autosampler), coupled to a TSQ Quantum Ultra triple quadrupole mass spectrometer with heated electrospray ionization (ESI; Thermo Fischer Scientific (San Jose, CA, USA)), was used for this study. The software for controlling, data recording and processing (Xcalibur version 2.07) was supplied by Thermo Fischer Scientific (San Jose, CA, USA). A Waters ACQUITY UPLC HSS T3 column (100 mm × 2.1 mm, 1.8 µm particles) combined with an ACQUITY UPLC HSS T3 VanGuard pre-column (5 mm × 2.1 mm, 1.8 µm particles) was kept at a temperature of 40 °C. Mixed isocratic and gradient elution at a flow rate of 0.45 mL/min was applied for analyte separation. In the mobile phase, solvent A consisted of 0.2% (v/v) HAc, and solvent B was ACN. The final method was an initial isocratic composition of 5% B held for 1 min, followed by a linear increase to 15% B for the next 1 min, and to 20% B in another 1 min. Then, B was increased linearly to 80% over the next minute to flush the column. To re-equilibrate, B was reduced back to the initial isocratic composition of 5% and was held for 2 min.

Sample pre-treatment and quantification

Samples were pretreated prior to LC–MS/MS analysis. The plasma samples were diluted two times, and urine samples ten (mice) or four (patients) times with ultrapure water. Then, 20-µL of (diluted) plasma or urine was pipetted into a 1.5-mL Eppendorf tube. Ultra-pure water was used as a surrogate matrix for calibration curve and quality control samples. Subsequently, 30 µL of cold (4 °C) ACN with stable-isotope internal standards were added. Samples were vortexed for 5 min on a plate vortexer and centrifuged for 5 min at 4000 g. A 35-µL sample of the supernatant was then collected in a 1-ml round-bottom well of a polypropylene 96-deep well plate and diluted with 200 µl of ultra-pure water. Finally, the plate was gently shaken before placing in the autosampler for analysis.

The calibration was obtained by eight different concentrations of standard for each analyte using linear regression analysis. The peak area was normalized with the respective labeled internal standards for each analyte before extracting the quantified values using the calibration curve. The reliability of quantification was further ensured by four quality control samples with different concentrations in every measurement. We then measured the total plasma concentration and urinary concentration of PBUTs.

RNA extraction, cDNA synthesis and Real-Time qPCR

Total RNA from homogenized mouse kidneys was isolated by the following method. After homogenizing the kidney tissue, 5 min of incubation was allowed to ensure complete dissociation of nucleoproteins complex. Then 200 µl of chloroform per 1 mL of homogenate was added and centrifuged for 15 min at 12,000 g at 4 °C following a gentle mix by pipetting and 2–3 min of incubation. The colorless aqueous layer of the supernatant containing the RNA was then treated with 500 µl of isopropanol and incubated for 10 min after a gentle mix before the next centrifugation for 10 min at 12,000 g at 4 °C. After discarding the supernatant, the precipitated total RNA was resuspended in 1 mL of 75% ethanol ((v/v) with water) before a brief vortex and centrifuge for 5 min at 7500 g at 4 °C. Then the obtained pellet was air-dried for 5–10 min and resuspended in 30µL of RNase-free water, followed by incubation in a heat block set at 55–60 °C for 15 min before quantification using the NanoDrop^® ND-1000 spectrophotometer (Thermo Fisher Scientific, DE, USA).

The synthesis of cDNA was performed using 1000 ng of total mRNA per sample and using the iScriptTM Reverse Transcription Supermix (Bio-Rad Laboratories, Hercules, CA, USA) and T100™ Thermal Cycler (Bio-Rad Laboratories). Subsequently, Real-Time PCR was performed using the iQ SYBR^® Green Supermix (Bio-Rad Laboratories) as indicated in the manufacturer’s protocol and by means of CFX96TM Real-Time PCR Detection System (Bio-Rad Laboratories). The data were analyzed using Bio-Rad CFX ManagerTM Software version 3.1 (Bio-Rad Laboratories) and expressed as relative gene expression. β-actin was used as housekeeping gene for normalization. Specific sense and anti-sense primers for β-actin (forward: TAAGGCCAACCGTGAAAAG; reverse: ACCAGAGGCATACAGGGACA), and Kim-1 (forward: ATGAATCAGATTCAAGTCTTC; reverse: TCTGGTTTGTGAGTCCATGTG) were synthesized by Biolegio (Nijmegen, The Netherlands).

Biochemical analyses

Blood glucose level in mice was measured (Medisense Precision Xtra; Abbott, Bedford, IN) 3 days after the intraperitoneal STZ injection to determine hyperglycemia. After 5 days, the condition of diabetic mice was stabilized by slow-release insulin pellets (Linshin, Scarborough, Canada). Creatinine concentrations were determined by enzymatic assays (J2L Elitech, Labarthe Inard, France).

The concentration of NGAL, KIM-1 and cystatin C (Cys C) in plasma and urine samples were determined by Enzyme-Linked Immunosorbent Assay (ELISA) using Duoset kit (R&D Systems, Minneapolis, USA) for humans or mice. Mouse plasma samples were diluted 250 and 500 times, and mouse urine samples 1000 and 500 times for NGAL and Cys C measurement, respectively. The urine samples from diabetic patients were diluted 20, 50 and 50 times for measuring KIM-1, NGAL and Cys C, respectively, before following the manufacturer’s instructions. The absorbance was determined using the Microplate Absorbance Reader (Bio-Rad Laboratories, Hercules, CA, USA) set to 450 nm. Each sample was measured in duplicates and quantification was done using Microplate Manager software (version 6.0, Bio-Rad Laboratories).

Calculations

For the mice, we used the urinary PBUT/ creatinine ratio normalized for plasma PBUT concentration (Eq. 1) as a surrogate for PBUT clearance (instead of the 24h PBUT clearance calculated with the 24h urine volume) since we suspected a systematic error in the 24h urine collection with relatively more loss of urine for the non-diabetic mice than for the polyuric (Table 1) diabetic mice. This was based on the observation that 24h creatinine excretion (calculated with 24 h urine volume) for diabetic mice was 50% higher than that of nondiabetic mice, i.e., 701 ± 60 vs. 418 ± 39 µg/d (P = 0.001) after 6 months and 932 ± 343 vs. 469 ± 48 µg/d (P = 0.016) after 8 months, respectively. In theory it is very unlikely that the 24h creatinine excretion reflecting muscle mass is significantly higher in diabetic mice than in healthy controls. We therefore suspect that a higher fraction of the 24h urine was left in the metabolic cages for the healthy controls than for the polyuric diabetic mice, causing this systematic error. Accordingly, urinary tubular injury markers were expressed as creatinine-normalized and not as absolute amounts excreted in 24h. The fractional excretion of PBUTs in mice was calculated by Eq. 2. The clearance of PBUTs for diabetic patients was calculated from 24 h urine samples in a standard manner, as stated in Eq. 3. The MDRD equation was used to calculate eGFR from serum creatinine. The fractional excretion of PBUTs in patients was calculated by normalizing total clearance with eGFR (not adjusted for body surface area), as indicated in Eq. 4 instead of Eq. 2, to minimize the influence of secretion fraction of creatinine. Urinary tubular markers were normalized using creatinine concentration in urine, consistent with the approach used in the preclinical study.

1

2

3

4

where and represent the urine and plasma concentration of a specific PBUT; and represent the urine and plasma concentration of creatinine, respectively. Urine concentrations were measured in the 24h urine.

Statistical analyses

The statistical analyses were performed using GraphPad Prism software for Mac, version 9 (GraphPad Software, San Diego, CA, USA) and R studio 2021.09.1 (Build 372^© 2009–2021 RStudio, PBC). Data are expressed as mean ± SD, mean (SD), and median (IQR), as appropriate. In the preclinical data, differences between groups were analyzed by nonparametric Mann–Whitney U test based on the data normality, determined by the Kolmogorov–Smirnov test. Spearman correlation coefficients were used to test correlations between variables. For patients, PBUT clearance values were log-transformed to attain normal distribution. Then the correlation of tubular injury markers with all variables were tested using Pearson correlation. The estimates of association were calculated by multiple linear regression models. The univariate model (Model 1) was followed by multivariate regression models (Models 2–4), where the estimates were adjusted for eGFR in Model 2, additionally adjusted for age, sex and race in Model 3, and finally also UAER, smoking status, BMI and blood pressure in Model 4. The level of significance was set at P < 0.05 in all cases.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

KGFG, a nephrologist specializing in chronic kidney disease and hypertension, and RM, an experimental nephrology expert, contributed to the study’s conception, design, supervision, and manuscript editing. TQN and RG, senior pathologists, assisted with the animal study and related analyses. RB, a technician, aided with the animal study. SMM provided data handling expertise. RWMV supervised patient data analysis, and RWS offered analytical support. SA, a PhD student, conducted the experiments, performed data analysis, and wrote the manuscript. All authors contributed to the manuscript review and agreed upon it.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval and consent

The clinical study was registered in the Netherlands Trial Register (NTR 4439), complied with the Declaration of Helsinki and Good Clinical Practice Guidelines, and was approved by the Medical Ethics Committee of the University Medical Center Groningen, the Netherlands (METc 2014/111) (28). All participants (N = 33) provided written informed consent before any study procedure started. All animal experiments were performed with the approval of the Experimental Animal Ethics Committee of the University of Utrecht (DEC number 2009.II.11.129).