BDNF: mRNA expression in urine cells of patients with chronic kidney disease and its role in kidney function

Abstract Podocyte loss and changes to the complex morphology are major causes of chronic kidney disease (CKD). As the incidence is continuously increasing over the last decades without sufficient treatment, it is important to find predicting biomarkers. Therefore, we measured urinary mRNA levels of podocyte genes NPHS1, NPHS2, PODXL and BDNF, KIM‐1, CTSL by qRT‐PCR of 120 CKD patients. We showed a strong correlation between BDNF and the kidney injury marker KIM‐1, which were also correlated with NPHS1, suggesting podocytes as a contributing source. In human biopsies, BDNF was localized in the cell body and major processes of podocytes. In glomeruli of diabetic nephropathy patients, we found a strong BDNF signal in the remaining podocytes. An inhibition of the BDNF receptor TrkB resulted in enhanced podocyte dedifferentiation. The knockdown of the orthologue resulted in pericardial oedema formation and lowered viability of zebrafish larvae. We found an enlarged Bowman's space, dilated glomerular capillaries, podocyte loss and an impaired glomerular filtration. We demonstrated that BDNF is essential for glomerular development, morphology and function and the expression of BDNF and KIM‐1 is highly correlated in urine cells of CKD patients. Therefore, BDNF mRNA in urine cells could serve as a potential CKD biomarker.


| INTRODUCTION
The "Greifswald Approach to Individualized Medicine (GANI_MED)" aims at the development of individualized diagnosis, prevention and therapy strategies for common diseases. 1,2 Therefore, patient cohorts were recruited and investigated under standardized and routine conditions. The cohort investigated in this report consisted of patients diagnosed with chronic kidney disease (CKD). 3 CKD mostly affects risk groups like patients suffering from diabetes mellitus or hypertension. [4][5][6] The disease is characterized by a decrease and subsequently by a loss of kidney function named endstage renal disease. Loss of renal function can only be compensated by renal replacement therapies like haemodialysis or transplantation.
Until today, CKD is not reversible and it is therefore important to identify predictive biomarkers and possible target molecules allowing early detection and prevention.
As it became obvious that podocytes are associated with the development of CKD, this specific cell type was brought into the focus of research. 7,8 Podocytes, a post-mitotic cell type, maintain the glomerular filtration barrier by their unique cellular structure, which includes major processes and foot processes covering the glomerular basement membrane (GBM) in a zipper-like fashion. 9,10 Changes in these structures lead to an impairment of glomerular function and are related to several kidney diseases like focal segmental glomerulosclerosis (FSGS), minimal change disease (MCD) and diabetic nephropathy (DN).
As podocytes share certain structural and molecular biological characteristics with neurons, proteins involved in neuronal structural and physiological maintenance 11,12 are of great interest for podocyte research and might play a potential role as biomarkers. One of those neuron-specific proteins is brain-derived neurotrophic factor (BDNF), a neurotrophic factor which is involved in neurogenesis, neuronal survival, 13,14 branching 15,16 and synaptic growth. 17 Thus, Ernfors et al 18 have already shown that BDNF plays a key role in neuronal development, because heterozygous knockout (KO) mice showed decreased neuronal development and homozygous KO mice often die directly after birth. It has already been shown that BDNF binds to 2 different receptors-TrkB and p75, that are involved in cell survival and differentiation processes. 19,20 Recently, it has been reported that BDNF and TrkB are expressed in podocytes in vivo, being essential for actin polymerization and cell survival. 21 As the actin cytoskeleton plays an important role for podocyte morphology and adhesion in vivo, and podocyte detachment is a major event in glomerulopathies, we investigated the expression of BDNF in cells appearing in the urine of patients suffering from CKD in an attempt to find out whether BDNF could be a suitable marker for the detection of DN.
As a second potential biomarker for glomerulosclerosis, we chose Hepatitis A virus cellular receptor 1 (HAVCR1) or kidney injury molecule-1 (KIM-1), a transmembrane protein that is not or at very low levels expressed in healthy kidneys. [22][23][24][25] Interestingly, Zhao et al 26 found KIM-1 being up-regulated in parietal epithelial cells and dedifferentiated podocytes of diabetic rats. Furthermore, recent findings show that the expression of KIM-1 reduces the negative effects of acute kidney injury by inducing phagocytosis. 27 Therefore, we selected KIM-1 to proof whether it could be used as a potential glomerular biomarker.
To study the influence of BDNF on podocyte development and glomerular morphology in vivo, we took larval zebrafish as a wellestablished model organism. The zebrafish larva is ideal for podocyte research [28][29][30] as it develops a functioning glomerulus during 48-56 hours post-fertilization (hpf), 31,32 which can be studied directly in living larvae by 2-photon microscopy (2-PM). 28,33,34 Moreover, by the use of the morpholino technology, specific proteins can easily be knocked down.
Zebrafish express a bdnf orthologue, whose amino acid sequence is 91% identical to human BDNF. 35 Although the BDNF sequence is rather conserved among these species, little is known about the function of bdnf in the zebrafish pronephros. A recent study has shown that there is a beneficial, microRNA-mediated effect on actin polymerization in adriamycin-induced podocyte damage emphasizing the important role for BDNF in kidney homoeostasis. 21 Our study shows that the mRNA expressions of BDNF, a newly identified podocyte gene, and of KIM-1, an injury-induced protein, are highly correlated in urine cells of CKD patients and secondly that the expression is associated with DN. Moreover, we show the importance of BDNF for glomerular function in zebrafish larvae and in isolated murine glomeruli.

| Study participants
Participants were recruited in the GANI_MED nephrology cohort. 3 All participants signed informed written consent forms. The study is consistent with the principles of the declaration of Helsinki, as reflected by an a priori approval of the Ethics Committee of the University of Greifswald.

| Clinical sample collection
A total of 120 urine samples were collected from participants who had known CKD with or without hypertension and/or diabetes. We used 50-100 mL morning urine. Only in 5% of cases, the urine volume was <50 mL. Unfortunately, the HbA1c value of 1 patient was not available. The time period from urine voiding until processing never exceeded 4 hours, as we found out that in this time viable cells could still be cultivated.

| Urine processing
Urine was centrifuged in a 50 mL centrifuge tube at room temperature (RT) for 3 minutes at 2100 g. The urine pellet was resuspended in 1 mL phosphate-buffered saline (PBS), transferred to a 1.5 mL centrifuge tube and then centrifuged at 12 000 g for 1 minute at RT. The supernatant was discarded. The washed urine pellet was resuspended in 900 μL Phenol/Guanidine-based Qiazol lysis reagent (Qiagen, Hilden, Germany) and then short-term stored at −20°C until use.

| Kidney specimens
Kidney tissue for immunofluorescence was obtained by percutaneous renal biopsy from patients undergoing diagnostic evaluation.
Biopsies from 2 subjects with diagnosed DN were investigated. The histopathological diagnosis included the following: diabetic glomerulosclerosis (patient 1) and FSGS with tubular changes (patient 2).
Control kidney tissue was taken from normal kidney parts of a renal tumour surgery patient. The clinical-functional diagnosis included the following: slight restriction in GFR and arterial hypertrophy. An informed consent was obtained from the donor.

| Podocyte de-/differentiation assay
All animal experiments were performed in accordance with national animal protection guidelines that conform to the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the local governmental authorities. The podocyte dedifferentiation assay was performed as described by Kindt et al. 36 Glomeruli were treated with ANA-12 (1-100 μmol/L, Sigma-Aldrich). RNA sequencing was performed as previously described. 37
The morpholinos were diluted to 1 mmol/L. A volume of approximately 3 nL per zebrafish was injected into 2 to 4-cell stage fertilized eggs using a microinjector (Transjector 5246, Eppendorf, Hamburg, Germany).

| Zebrafish in vivo microscopy
In vivo imaging was performed as previously described. 28,34,40

| Statistical analysis
Urine expression data were log-transformed for all correlation analyses. Associations between potential biomarkers were assessed using Pearson correlation followed by the Benjamini-Hochberg procedure. Comparisons between groups were performed as indicated.
All comparisons between 2 groups concerning zebrafish experiments were done with the Mann-Whitney U test. All statistical analyses were performed using LABMAT version 2013 and SPSS V. 21.

| Baseline characteristics of patients
To identify individual prognostic biomarkers for CKD, we analysed a panel of potential urinary biomarkers in 120 GANI_MED renal study participants. The baseline characteristics are shown in Table 1 There was also no statistically significant difference concerning | 5267 the uACR, which was 1531 mg/g in diabetics and 1149 mg/g in non-diabetics. The group of diabetics included 32 patients with an eGFR lower than 60 mL/min/1.73 m 2 and 16 dialysis patients. The group of non-diabetics included 77 patients with an eGFR lower than 60 mL/ min/ 1.73 m 2 and 59 dialysis patients. Again, these differences were not statistically significant.

| Correlations between urine mRNA levels
To investigate whether the mRNA levels were interrelated, we performed a correlation analysis ( Figure 1A) followed by the Benjamini-Hochberg procedure to determine statistical significance ( Figure 1B).
We found a strong positive correlation between the expressions of Figure 1C). Interestingly, we also observed significant correlations between the expres-

| Correlation analysis between urine mRNA levels and clinical parameters revealed a negative correlation between BDNF and uACR
To identify possible correlations between the investigated expressions of urinary cell pellet mRNA species and clinical parameters, we applied the Pearson correlation analysis corrected by the Benjamini-Hochberg procedure for the expression of the 3 podocyte markers NPHS1, NPHS2, PODXL, the 3 biomarker candidates BDNF, KIM-1 and CTSL-1 and the clinical parameters eGFR, HbA1c and uACR.
These calculations revealed a significant negative correlation between BDNF and uACR (P = .0046; Figure 1E). BDNF and eGFR were not correlated with each other. Additionally, there was no significant difference in BDNF expression between patients of different CKD stages ( Figure 1F). Furthermore, we did not find any significant correlation between any of the other parameters in the observed patient group (data not shown).

| Sex-specific expression of BDNF and KIM-1 in diabetic and non-diabetic patients
As our cohort shows a typical frequency distribution of nephropathies with the largest proportion suffering from diabetes, which is respectively, in non-diabetic females. The associations of BNDF and KIM-1 with DN were not detected in males ( Figure S1). There was a statistically significant difference in BDNF mRNA expression between female and male diabetics but not between female and male nondiabetics. In contrast, KIM-1 mRNA expression did not significantly differ between female and male diabetics or non-diabetics.

| BDNF stabilizes podocyte differentiation in cultured glomeruli
We recently established a novel assay to study podocyte de-/differentiation in isolated glomeruli of mice expressing CFP under control of the Nphs1 promoter. 36 As podocytes of isolated glomeruli spontaneously dedifferentiate in cell culture accompanied by a decrease in

| The knockdown of BDNF induces proteinuria in zebrafish larvae
To study the function of bdnf in vivo, we performed a bdnf knockdown (KD) with specific morpholinos. To this end, we generated a translation-blocking morpholino (bdnfMO). Three days after injection of the bdnfMO, 82.7 ± 2.0% of the zebrafish larvae had developed severe pericardial oedema (arrow in Figure 5A), a hallmark of impaired kidney function. In contrast, only 1.3 ± 0.8% of CtrlMO-injected larvae developed pericardial oedema ( Figure 5A,B).
In addition, bdnfMO-injected larvae were less viable (72.9 ± 7.1%) compared with larvae injected with CtrlMO (93.6 ± 6.0%) ( the blood which cannot pass the intact filtration barrier. In contrast to CtrlMO-treated larvae, we observed a nearly complete loss of eGFP fluorescence in the blood of bdnf KD larvae at 3 and 6 dpf ( Figure 5A), indicating leakage of the filtration barrier due to bdnf KD.

| BDNF is important for the proper morphology of zebrafish glomeruli as well as for the expression of nephrin and podocin
The KD of bdnf was verified by RT-PCR showing reduced intensities of the specific amplicon in bdnfMO-injected larvae compared with CtrlMO-injected larvae ( Figure 5C). Furthermore, bdnf KD larvae showed reduced expression of nphs2 (podocin) in RT-PCR analysis ( Figure 5C), which was confirmed by qRT-PCR ( Figure 5D). To study the glomerular morphology, we stained cryosections of zebrafish larvae, utilizing the ET strain that expresses eGFP specifically in podocytes. 28,30 After staining the F-actin cytoskeleton with Alexa-546 phalloidin, we observed significant changes in the morphology of the glomeruli in response to the KD of bdnf. In addition to an enlargement of the glomerular tuft and Bowman's space, we observed a reduced number of podocytes in bdnf KD larvae (3 dpf) in contrast to CtrlMO-treated zebrafish larvae ( Figure 6A). Moreover, immunohistological staining for nephrin revealed a significant reduction of the slit membrane protein due to the KD of bdnf in the zebrafish larvae ( Figure 6A).  Figure 6A).

| DISCUSSION
Biomarkers from non-invasive sources like urine are of growing interest in CKD research, as they seem to be a promising diagnostic tool for early detection of the disease. Conventional parameters like proteinuria, cystatin C and creatinine are established markers of kidney function, which are positively correlated with kidney dysfunction. [41][42][43][44][45] Because they are only measurable at a relatively progressed disease state, the establishment of markers for an earlier disease state is needed. In this study, we investigated the mRNA expression levels of potential predictive biomarkers for CKD in the urine sediment. Previous studies identified a variety of predictive biomarkers from urine sediment mRNA for non-malignant kidney diseases. [46][47][48][49] In the present study, we found a strong and highly significant positive correlation between BDNF mRNA levels and KIM-1 mRNA levels. KIM-1, which is endogenously expressed at a very low levels, is a well-established urinary biomarker for acute kidney injury and is also known to be positively correlated with tubular and tubulointerstitial injury as well as with glomerular damage. 26,[50][51][52] This is in a very good agreement with our findings, as we could also detect glomerular expression of KIM-1 by immunofluorescence staining and in murine glomeruli of our podocyte dedifferentiation assay. Therefore, BDNF mRNA levels might serve as a new potential biomarker for glomerular kidney injury. Interestingly, we observed a statistically significant up-regulation of both mRNA species in diabetic patients compared with non-diabetic patients and for BDNF in a sex-specific manner. Sex specificity in kidney injury has been described earlier, 53,54 but seems to be relatively underestimated. As podocytes and neurons share some common features, it was predictable that BDNF may also be expressed in both cell types, as is the case for other proteins. 11,12 In keeping with this notion, we found that BDNF was expressed in the cell body and in major processes of podocytes in human kidney biopsies. Surprisingly, we did not see any co-localization with synaptopodin, a podocyte-specific protein which is exclusively expressed in podocyte foot processes, indicating that the localization of BDNF is restricted to the cell body and the major processes.
Looking at sectioned kidney biopsies of patients, which were diagnosed with DN, we observed less BDNF-expressing podocytes, but podocytes that still expressed BDNF showed an increased ENDLICH ET AL. As podocyte dedifferentiation is a critical step in the progression of DN, we applied our well-established dedifferentiation assay to study the influence of BDNF on podocyte differentiation. 36 We could show that the inhibition of the TrkB receptor, which mediates for BDNF signalling, led to decreased nephrin promotor activity and therefore to increased podocyte dedifferentiation in murine glomeruli.
Only little is known about the effect of a BDNF KO on kidney homoeostasis in vivo. As BDNF KO mice die directly after birth and no kidney-specific phenotypical impacts have been published, 18 we selected the zebrafish larva as model organism. Zebrafish larvae are relatively easy to breed and show a glomerular morphology similar to that of mammals with one glomerulus connected to two tubules in their first functional state, the pronephros. 31,32 Another advantage is their applicability for in vivo microscopic techniques like 2-PM which can track changes in morphology and function. 28,34 bdnf KD larvae developed pericardial oedema as an indicator for an impairment of the glomerular filtration barrier. This finding could be confirmed by a decrease in the intravascular eGFP intensity in bdnfMOtreated CADE larvae, also indicating a leaky filtration barrier. 28,30 We also found a reduced expression of the podocyte markers podocin and nephrin, and a disrupted F-actin structure in bdnfMO-treated larvae, which are suggestive of podocyte loss.
A positive influence of BDNF on the expression of podocyte markers like nephrin has been shown before. 21  Here, we show that BDNF mRNA may potentially serve as a new prognostic urinary biomarker for CKD. We also show that the expression of BDNF in human podocytes is locally restricted to the cell body and major processes. Furthermore, we give first insights into the involvement of BDNF in podocyte dedifferentiation and into its deregulation in DN. The KD of bdnf leads to impaired glomerular filtration as well as to podocyte loss and/or hindered migration of glomerular progenitors in an in vivo zebrafish model.

ACKNOWLEDG EMENTS
The

CONFLI CT OF INTEREST
The authors disclose no conflict of interests.

AUTHOR CONTRI BUTIONS
N.E. and K.E. designed the study. S.S. and J.K. contributed to urine processing. J.K. performed urine mRNA PCR experiments. T.L. and P.K. F I G U R E 6 Bdnf knockdown in zebrafish larvae leads to morphological changes in the glomerulus and the down-regulation of podocyte marker proteins. BdnfMO and CtrlMO were injected in ET eggs expressing eGFP specifically in podocytes. Cryosections were counterstained for F-actin by phalloidine (red) and nuclei by Hoechst (blue) (A; a-h). bdnfMO-treated larvae show an enlarged glomerular tuft (A; b arrow) and Bowman's space (A; b asterisk) compared with the normal morphology of the CtrlMO-treated larvae (A; e-h). Counterstaining of the slit diaphragm protein nephrin (red) and nuclei by Hoechst (blue, A; i-p) reveals a down-regulation of nephrin due to bdnfMO treatment (A; k) compared to CtrlMO-treated larvae (A; o). [Scale bars = 20 μm] In vivo microscopy reveals 2 different phenotypes of bdnfMO-treated larvae. Phenotype 1 is characterized by unfused glomeruli (B; a white line and b), a reduced number of podocytes and a dilatation of Bowman's space and the glomerular tuft (B; a asterisks). The second phenotype is characterized by the absence of podocyte major processes, a reduced podocyte number and a dilatation of Bowman's space and the glomerular tuft (B; c asterisk and d). CtrlMO-injected larvae show a normal glomerular morphology with well-shaped major processes (B; e and f). [Scale bars = 20 μm]