Urinary podocyte-associated mRNA levels correlate with proximal tubule dysfunction in early diabetic nephropathy of type 2 diabetes mellitus

A total of 76 patients with type 2 DM (28-normoalbuminuria; 27-microalbuminuria; 21-macroalbuminuria), attending the Outpatient Department of Diabetes and Metabolic Diseases and 20 healthy age- and gender-matched control subjects, recruited from the office of a general practitioner, were enrolled in a cross-sectional study.

The inclusion criteria were duration of diabetes longer than 5 years, normoalbuminuria [urinary albumin:creatinine ratio (UACR) <30 mg/g], moderately increased albuminuria or microalbuminuria (UACR 30–300 mg/g), and severely increased albuminuria or macroalbuminuria (UACR >300 mg/g), therapy with oral antidiabetic drugs, angiotensin-converting enzyme inhibitors and/or angiotensin receptor blockers, and statins. The exclusion criteria were on-going urinary tract infections and other types of acute infection which could bias the interpretation of data related to albuminuria, urinary excretion of the biomarkers studied, as well as to renal function.

The County Emergency Hospital Timisoara Ethical Committee (Board of Human Studies) approved the protocol (approval number 36/10th September 2015), and every patient provided written informed consent before enrolment.

All patients were assessed concerning: urinary alpha₁-microglobulin and urinary kidney injury molecule-1 (KIM-1) as biomarkers of PT dysfunction; urinary nephrin and urinary VEGF, as markers of podocyte damage; plasma and urinary AGE; UACR and serum cystatin C; urinary podocyte-associated mRNA for nephrin, podocin, alpha-actinin 4, CD2AP, GLEPP1, ADAM 10, and NFκB.

Serum and urinary biomarkers were determined in specimens frozen at −80 °C and thawed before assay. Urinary biomarkers were assessed in the first morning urine (midstream urine), except for urinary AGE, which were determined in the 24-h sample. All study variables were assessed in triplicate on aliquots from the same first morning urine sample, or from the 24-h urine collection, as appropriate. The ELISA assessments were performed as per protocol indicated by the manufacturer in triplicate assessments for each patient from the same aliquot. The inter- or intra-assay coefficients of variance (CVs) were indicated according to the data provided by the manufacturer’s brochure. Podocytes were assessed in cell cultures performed from the first morning urine (midstream) in all patients.

Chronic kidney disease was defined according to the KDIGO Guideline for the Evaluation and Management of Chronic Kidney Disease by calculating the eGFR with the formula which included serum creatinine and serum cystatin C, in order to increase diagnosis accuracy of CKD staging (2009 CKD-EPI creatinine–cystatin C equation) [27, 28].

Alpha ₁ -microglobulin was evaluated in the first morning urine (midstream) specimen with N α1-microglobulin kit (Siemens Healthcare Diagnostics, Marburg, Germany) through particle-enhanced immunonephelometry using the BNProSpec System. The reference interval was 12 mg/l or 0.07–5 mg/g creatinine. The intra-assay precision was 2.9–5.2% CV, while the inter-assay precision was 7.4–13.2% CV.

KIM-1 was assessed in the first morning urine (midstream) specimen by KIM-1 ELISA test kit for the detection of KIM-1 in human urine, Cat. No. H-RENA-E-001, Bio Assay Works, Ijamsville, MD, USA. A human KIM-1 antibody was utilised and the detection level was set at urinary KIM-1 <0.150 ng/ml.

Cultures of urinary podocytes were performed as previously described [2, 29]. Midstream urine samples of 30 ml were collected in sterile tubs and centrifuged at 700g for 5 min. The pelleted cellular material was washed twice with PBS, suspended in appropriate medium (RPMI medium supplemented with 10% fetal bovine serum, insulin–transferrin–selenium G—Gibco, Cat. No. 41400045, NY, USA) and 1% penicillin/streptomycin (Life technologies, Cat. No. 15140-122, USA), and cultured on cell culture flasks coated with type I collagen (Gibco, Cat. No A10483-01, NY, USA). Samples were incubated at 37 °C with 5% CO₂ overnight. After 12 h, the cells were detached with trypsin, suspended in PBS, and cytocentrifuged at 700g for 5 min. After centrifugation, the cells were fixed with 4% paraformaldehyde in PBS for 30 min at room temperature and 10 min in ice-cold methanol (−20 °C) followed by permeabilization with 0.2% Triton X-100 in PBS for 10 min. Nonspecific binding sites were blocked with 2% FBS and 2% BSA in PBS overnight. Podocalyxin and synaptopodin were used as podocyte markers [4, 30, 31].

The slides were incubated with primary antibody anti-podocalyxin (Podocalyxin Mouse Monoclonal Antibody, clone 3D3, Cat. No. 39-3800, Invitrogen, USA) (1:200 dilution) for 1 h at room temperature and with primary antibody anti-synaptopodin (Polyclonal rabbit anti-synaptopodin antibody, Cat. No. ab101883, Abcam, Cambridge, MA, USA) (1:300 dilution) for 2 h at room temperature. After washing with PBS, the slides were incubated with the secondary antibody (Life Technologies, fluorescein goat anti-mouse IgG-(H+L), F2761, USA) (1 µg/ml in phosphate buffered saline with 0.2% BSA) for 1 h, examined by immunofluorescence microscopy and counted. Urinary podocytes were expressed as cells/ml.

For mRNA preparation and RT-PCR Assay, the total urine pellet RNA was isolated using the TaqMan Gene Expression kit. Relative quantification of the nephrin NPHS1, podocin NPHS2, CD2AP, alpha-actinin-4, ADAM 10, GLEPP1, and NFκB was realised using the 7900 HT Fast Real-Time PCR System. For normalization in gene expression we used GAPDH as internal control. The relative amounts of the gene were expressed as 2^−∆CT (∆CT = CT value target gene − CT value internal control).

Clinical and biological data are presented as medians and IQR, as for variables with skewed distribution. Podocytes were expressed as mean ± standard deviation (SD). If medians and IQR were utilised for podocytes median of the first 3 groups would be 0, while that one of group 4 would be 6. Thus, podocytes were not expressed as medians and IQR, as for variables with skewed distribution, but were expressed as mean ± SD in order to be utilised in analyses. Podocytes which reached levels above 0 were coded with “1” and logistic regression was performed.

Depending on the distribution of the values, differences between subgroups were analyzed with the Mann–Whitney U test for comparison of two groups and the Kruskal–Wallis test for comparison of four groups. Univariable regression analyses were carried out to evaluate the significance of the relation between continuous variables. Only significant variables yielded by univariable regression analysis were introduced in the models for multivariable regression analysis. The p values for all hypothesis tests were two-sided, and statistical significance was set at p < 0.05. All analyses were conducted with Stata 9.2 (Statacorp, Texas, USA).

Demographic, clinical, and biological data of patients with type 2DM and healthy controls are presented in Tables 1 and 2. The target genes studied showed significantly increased levels of expression in patients with type 2 DM as compared with healthy controls. Also, there were significant differences between the mRNA of podocyte-associated molecules among the groups studied. The expression levels increased with progression of DN staged by albuminuria and renal function decline.

Linear regression analysis showed significant correlations between podocyte-associated genes and albuminuria, eGFR, and podocytes shed into urine. Thus, mRNA of nephrin, podocin, alfa-actinin-4, CD2AP, ADAM 10, GLEPP-1, and NFκB correlated directly with urinary podocytes (p < 0.0001), UACR (p < 0.0001), and negatively with eGFR (p < 0.0001), respectively.

Furthermore, podocyte-associated genes showed correlations with podocyte damage biomarkers, such as urinary nephrin and urinary VEGF, and with biomarkers of PT dysfunction, such as urinary alpha₁-microglobulin and urinary KIM-1. The mRNAs of nephrin, alfa-actinin-4, CD2AP, ADAM10, GLEPP-1, and NFκB also correlated with urinary AGE (Table 3).

After adjustment for potential confounders, such as lipid profile, HbA1c, and high-sensitive C-reactive protein, multivariable regression analysis yielded models of which the deriving data revealed significant correlations between nephrin mRNA and podocytes, UACR, KIM-1, nephrin, and eGFR (p < 0.0001; R² = 0.808); podocin mRNA and podocytes, UACR, KIM-1, nephrin, and eGFR (p < 0.0001; R² = 0.825); alpha-actinin-4-mRNA and podocytes, UACR, and eGFR (p < 0.0001; R² = 0.805); CD2AP mRNA and podocytes, UACR, VEGF, KIM-1, and eGFR (p < 0.0001; R² = 0.663); ADAM10 mRNA and podocytes, UACR, alpha-1 microglobulin, and eGFR (p < 0.0001; R² = 0.726); GLEPP-1 mRNA and podocytes, UACR, VEGF, KIM-1, and eGFR (p < 0.0001; R² = 0.724) (Table 4).

The detection of increased urinary mRNA of podocyte-specific molecules in DN is a useful and non-invasive tool to assess podocyturia and podocyte damage, as markers of renal involvement and disease activity in DN [2, 23].

The increased mRNA levels of podocyte-associated molecules in urinary sediments may constitute direct evidence of the presence of detached podocytes originating in the injured glomerulus in the course of DN, and may correlate with albuminuria, as well as with activity and progression of DN [5, 32].

In our study we found increased levels of urinary podocyte-associated molecules such as nephrin, podocin, alpha-actinin-4, CD2AP, ADAM10, GLEPP-1, and NFκB, which varied across the studied groups in relation with albuminuria and renal function decline. The increased levels of podocyte-associated genes in the urinary sediment of patients with type 2 DM were detected for all seven genes studied, a fact which could be indicative of an increased podocyte excretion into the urine of patients with diabetic kidney disease, even in the normoalbuminuric stage.

These data are in agreement with other studies conducted on patients with type 2 DM, in whom urinary nephrin, podocin, synaptopodin, Wilm’s tumor-1, and alpha-actinin-4 were increased in the urinary sediment [33]. Also, Zheng et al. revealed increased levels of synaptopodin, podocalyxin, CD2AP, alpha actinin-4, and podocin, which increased with progression of DN [32].

Moreover, in a study performed in patients with prediabetes and type 2 DM, podocyte marker expression was higher in diabetic subjects as compared to those with prediabetes with regard to mRNA of nephrin, podocin, podocalyxin, synaptopodin, TRPC6, alpha actinin-4, and TGF-β1 [34]. A recent study carried out in normoalbuminuric patients with type 2 DM revealed increased levels of urinary podocyte markers, such as synaptopodin, nephrin, and podocin, an observation which reflects early podocyte injury [35].

In our study we identified three more podocyte markers with high gene expression such as GLEPP-1, ADAM10, and NFκB, which could be considered as additional candidate molecules of urinary podocyte biomarkers in early DN. This observation points to significant damage to the podocytes and their subsequent loss into the urine in the course of type 2 DM. All seven target genes correlated with urinary podocytes, a fact which reinforces the diagnostic significance of these podocytes-associated molecules as biomarkers of podocyturia. The relation of the podocyte-associated proteins with AGE was also investigated. The target genes studied correlated with urinary AGE in patients with type 2 DM, even in those with normoalbuminuria.

Urinary expression of nephrin correlated with AGE, thus showing a potential implication of hyperglycemia on podocytes and their cytoskeletal integrity. Also, alpha-actinin-4 had an increased excretion in patients with type 2 DM and correlated with AGE, a fact which points to significant podocyturia and impaired interactions between podocytes and their cytoskeletal proteins, data which is in keeping with the study by Ha et al. [36].

In another study by Ha et al. high glucose and AGE-added condition decreased CD2AP protein amount and its mRNA expression as compared to normal glucose or osmotic conditions. This phenomenon was related to relocalization and concentration of CD2AP at internal and perinuclear areas of podocytes [37].

By contrast, in our study urinary mRNA of CD2AP was increased and correlated with AGE, most likely related to severe damage to podocytes, followed by their excretion in the urine. Urinary AGE could induce disruption of podocyte structural proteins and suppressed expression of podocyte-specific proteins in renal tissue, but this decreased intra-renal expression may be associated with increased expression of podocyte-associated genes in urinary sediment, as reported by other studies, performed on patients with type 2 DM [5].

The activity of ADAM10 was increased as shown by the elevated levels of mRNA-ADAM10 expression in the urinary sediment of the patients with type 2 DM studied. This observation is in keeping with the results reported by Gutwein et al., who showed that ADAM10 is expressed in podocytes and displays increased levels in the urine of patients with various glomerular diseases [19]. In our study we also found a significant correlation of urinary ADAM10 with urinary AGE, an observation which may underline the role of AGE in the expression of podocyte-specific genes. The same observation applies for GLEPP-1, which was also found increased in the urinary sediment and correlated with urinary AGE.

The activation of NFκB reflects underlining inflammatory processes in the kidney. In our study, the increased levels of mRNA of NFκB show a significant involvement of the inflammatory gene in podocyte damage, as has been demonstrated in human DN [25]. Of note, urinary expression of NFκB correlated with urinary AGE, a fact which shows that the hyperglycemic milieu favours rearrangement of gene expression profiles in the course of DN.

In the present study, urinary podocyte mRNAs were associated to biomarkers of podocyte damage. Thus, nephrin mRNA correlated with urinary nephrin and urinary VEGF. In a study conducted by Pätäri et al. in the urine of patients with type 1 DM, nephrin was evidenced by western blotting in 30% of normoalbuminuric patients, in 17% of those with microalbuminuria, in 28% of those with macroalbuminuria, and in 28% of those with recent onset albuminuria. These results pointed to the fact that the detection of nephrinuria in normoalbuminuric patients is highly predictive for the development of DN [12]. In another study performed in type 2 DM patients, nephrinuria was detected in 54% of normoalbuminuric patients, suggesting its potential role as an early biomarker of DN. In this report, nephrinuria was assessed by enzyme-linked immunosorbent assay and correlated significantly with albuminuria and showed a negative correlation with eGFR [10]. The other podocyte-associated molecules assessed in our study also correlated with the biomarkers of podocyte damage studied, indicating severe injury to the podocytes, even in normoalbuminuric patients.

Moreover, podocyte-associated genes correlated with biomarkers of PT dysfunction in the early stages of DN. These data are in keeping with the results provided by a study in an animal model which showed that in CD2AP knockout mice, changes in glomerular permeability could not explain increases in intact albumin excretion. The authors conclude that albuminuria could be related to inhibition of the tubule degradation pathway associated with degrading filtered albumin. They also stated that factors released by the structurally impaired podocytes could influence the PT degradation pathway downstream [38]. In a previous study we showed that in patients with type 2 DM, there is an association of PT dysfunction with podocyte damage biomarkers in normoalbuminuric patients, observation which raised the possibility of a putative role of the PT in urinary nephrin and urinary VEGF excretion in early DN. Also, we forwarded the hypothesis according to which in early DN, sequentially, PT dysfunction biomarkers may be elevated before the increase in podocyte damage biomarkers in the normoalbuminuria stage. Most likely, the PT interferes with the expression of glomerular injury in early DN [13].

Due to the fact that in our study the podocyte-associated proteins correlated with biomarkers of PT dysfunction in normoalbuminuric patients we may reinforce the idea that PT dysfunction could precede the expression of glomerular injury, as assessed by podocyte biomarkers and by podocyte-associated genes.

In our study, all seven podocyte-associated molecules correlated directly with albuminuria even in normoalbuminuric patients, similarly with the results provided by other studies [32‐35]. This correlation is highly indicative of significant podocyturia and allows for a non-invasive diagnosis of podocyte loss into urine in early DN.

It should be underlined that in multivariable analysis, all seven target genes remained independent predictors of progressive albuminuria. Moreover, the target genes studied showed an indirect correlation with eGFR. This observation is in agreement with the results released by other studies in which podocyte-associated genes, such as podocalyxin mRNA [32], and nephrin and synaptopodin [33] correlated inversely with eGFR. By contrast, in another study performed in patients with prediabetes and type 2 DM, the authors did not find a correlation between podocyturia as assessed by podocyte-associated genes, and eGFR [34].

In our study, the diagnostic power of renal function stratification increased by using eGFR calculated by the CKD-EPI formula which includes serum creatinine and serum cystatin C, and implies a more accurate evaluation of renal function decline [28].

Our study has several limitations. Firstly, the reduced sample size affects the statistical power of the study. However, given the fact that the distribution of the variables was not too grossly non-normal, the regression analyses provided good approximations. Second, this is a cross-sectional study which requires validation by a longitudinal study in order to prove correlation of podocyte-associated molecules, biomarkers of podocyte damage, and of PT dysfunction with DN progression. Third, despite significant correlations in multivariable analysis of podocyte-associated proteins with albuminuria, nephrinuria, VEGF, alpha-1 microglobulin, KIM-1, AGE, and eGFR, residual confounding factors could have interfered interpretation of data. Discrepancy between correlations of podocyte-associated mRNAs with the parameters studied in multivariable analysis is probably related to high variability of expression of these genes in the course of DN. One biomarker has usually a high variability. Multiple biomarkers are needed in order to increase diagnostic accuracy and predictive value for disease activity and progression.

There are also several strengths of our study. We demonstrated a direct correlation of urinary podocytes isolated in cell cultures with their damage biomarkers and with podocyte-specific molecules in normoalbuminuric patients with type 2 DM. We found additional podocyte-associated mRNAs, such as ADAM10, GLEPP-1, and NFκB, highly indicative of podocyte loss into urine in the early stages of DN. Furthermore, the novelty of the study resides in the demonstration of a correlation of podocyte-associated genes with PT dysfunction, the latter being well known to intervene in albumin processing in the course of DN [39]. Moreover, serum cystatin C was utilised in the calculation of eGFR, thus increasing the diagnostic power of the correlation between podocyte-associated genes and eGFR.

In conclusion, in patients with type 2DM there is an association between urinary mRNA of podocyte-associated molecules, biomarkers of podocyte damage, and of PT dysfunction. GLEPP-1, ADAM 10, and NFκB may be considered additional candidate molecules useful in the diagnosis of podocyturia in early DN. AGE could be involved in this association.

Podocyte-associated molecules, in conjunction with podocyte damage biomarkers and PT dysfunction biomarkers allowed for a stratification of the patients studied staged by albuminuria and renal function decline. These data should be validated by further prospective studies on larger cohorts.