Identification of cell and disease specific microRNAs in glomerular pathologies

Abstract MicroRNAs (miRs) are small non‐coding RNAs that regulate gene expression in physiological processes as well as in diseases. Currently miRs are already used to find novel mechanisms involved in diseases and in the future, they might serve as diagnostic markers. To identify miRs that play a role in glomerular diseases urinary miR‐screenings are a frequently used tool. However, miRs that are detected in the urine might simply be filtered from the blood stream and could have been produced anywhere in the body, so they might be completely unrelated to the diseases. We performed a combined miR‐screening in pooled urine samples from patients with different glomerular diseases as well as in cultured human podocytes, human mesangial cells, human glomerular endothelial cells and human tubular cells. The miR‐screening in renal cells was done in untreated conditions and after stimulation with TGF‐β. A merge of the detected regulated miRs led us to identify disease‐specific, cell type‐specific and cell stress‐induced miRs. Most miRs were down‐regulated following the stimulation with TGF‐β in all cell types. Up‐regulation of miRs after TGF‐β was cell type‐specific for most miRs. Furthermore, urinary miRs from patients with different glomerular diseases could be assigned to the different renal cell types. Most miRs were specifically regulated in one disease. Only miR‐155 was up‐regulated in all disease urines compared to control and therefore seems to be rather unspecific. In conclusion, a combined urinary and cell miR‐screening can improve the interpretation of screening results. These data are useful to identify novel miRs potentially involved in glomerular diseases.

the 3′ untranslated region (3′UTR) of a target messenger RNA and inhibits RNA translation. 4 Several miRs are enriched in human kidney and miRs seem to play a role in the glomerular homeostasis. 5 Mice with podocyte-specific alteration in miR-expression by deletion of Dicer or Drosha display progressive glomerular damage with proteinuria and podocyte defects. 6,7 MiRs can be secreted in body fluids and therefore could possibly serve as biomarkers in various glomerular diseases. For example, patients with focal segmental glomerulosclerosis (FSGS) had 10-times elevated concentrations of miR-3d and miR-10a in the urine compared to healthy controls. 8 Urinary expression of miR-200a, miR-200b and miR-429 were downregulated in patients with IgA glomerulonephritis (IgA-GN) and this down-regulation correlated with severity of the disease and rate of progression. 9 MiRs offer some important advantages over other markers as they are stable and protected from endogenous RNase because of their small size and by packaging within exosomes. 10,11 However, although suggested as potential biomarkers the origin of miRs has rarely been defined and cell type specific miRs have not yet been reported in the kidney. In the past miR-screenings in body fluids or tissue samples of patients with glomerular diseases were compared to healthy controls. [12][13][14] By this approach, comparing samples of one disease to control samples makes the specificity of the findings elusive. In addition, it is not clear if the miRs are 'bystanders' or might originate from cells and tissues involved in the disease.
Urinary miRs might be filtered from the blood, excreted by tubular cells or derived directly from glomerular cells affected by the disease process.
In this study, we described the advantages of a combined miRscreening in urine as well as in cultured renal cells and explained different ways of data normalization and interpretation. We identified renal cell type-specific miRs and miRs specifically regulated by TGF-β in these cells. Hereby, we were able to investigate miRs in urines from patients with different glomerular diseases and could compare them to controls. In addition, we could assign the urinary miRs to the different renal cell types.

| Cell culture
Under permissive conditions at 33°C, podocytes proliferate. When cultured at 37°C, the SV40 T-antigen was inactivated for cell dif-

| MiR-isolation and miR-screening
Purification of total RNAs including miRs from cultured renal cells and cell-free urine from patients was done with miRNeasy Kit

| Data analysis
We used the delta-delta cycle threshold (CT) method to normalize our miR-screening data and to generate fold changes in miR-expression after TGF-β stimulation and fold changes in miR-expression in urine samples from patients with glomerular diseases compared to control.
Delta-delta CT = delta CT (sample) -delta CT (reference) with delta CT (sample) = CT value for sample normalized to endogenous housekeeping gene and delta CT (reference) = CT value for calibrator normalized to endogenous housekeeping gene. The CT is defined as the number of cycles required for the fluorescent signal to cross the threshold (ie exceeds background level). CT levels are inversely proportional to the amount of target nucleic acid in the sample.
The maximum allowable CT value in our study was set to 35.
MiR-samples with bad intensity quality were excluded in the analysis. The normalization of the miR-screening data of the different cell lines was done with the housekeeper U6 snRNA-001973. Other like RNU48 or RNU44 was not used because they were regulated in our cell-screening. This is in line with published data showing that these small-nucleolar RNAs like RNU44, RNU48, RNU43 and RNU6B commonly used for miR-normalization are regulated in tumours. 15 The housekeeper for the miR-analysis in urines was cel-39 that was spiked into pooled urines of each disease before miR-isolation. Normalized CT values were then transformed into relative quantity (RQ) value according to formula 2 −(delta-delta CT) . MiRs with RQ values >1.5 were considered up-regulation and <0.5 were considered down-regulation.

| MiR-screening setup
We performed a Q-PCR based miR-screening (TaqMan ® Array Human MicroRNA Card Set v3.0) in cultured human podocytes,   Table 1. Urine samples of two women and two men (except PREEC: urine samples of four women) with an active form of their disease were used for the miR-screening.

| Renal cell type-specific miRs
By comparing the individual expression levels of each miR in the different cell types to the mean expression level of this miR in all cell types, we were able to detect cell type-specific miRs (Table S1).
TA B L E 2 Cell type-specific miRs in cultured human glomerular endothelial cell, mesangial cells, podocytes and tubular cells  16 Examples for glomerular endothelial cell and amesangial cell-specific miRs are miR-126 and miR-206, respectively. MiR-9 was exclusively expressed in tubular cells.
We examined the overlap of miR-expression in the different cell types. Twenty-nine different miRs were only expressed in cultured human mesangial cells, 32 miRs were specific for human glomerular endothelial cells, 65 miRs were only detectable in human podocytes and 38 miRs were specific for proximal tubular cells ( Table 2).
Nineteen different miRs could be detected in all four renal cell types ( Figure 2).
Next, we compared the expression level of each miR and cell type to the global mean of all 754 miRs from the screening to categorize them in rather high (fold change >10) or low (fold change <0.5), expressed (Table S2). This analysis revealed that miR-126, miR-126#, miR-531 and mir-346 were not only glomerular endothelial cell-specific but also highly expressed in this cell type (fold change >10 compared to the global mean of all miRs). MiR-106a# and miR-302a were highly expressed only in mesangial cells. Furthermore, miR-200b, miR-1225, miR-221, miR-1267 and miR-331were specific for podocytes and expressed more than 10-fold in this cell type compared to the global mean. Four different miRs were rather high expressed and specific for tubular cells: miR-1305, miR-499-3p, let-7b and miR-454 (Table S2).
A third way of analysis is comparing the mean miR-expression levels of all cell types to the global mean of miR-expression level of all miRs, Again, we looked for miRs that where only expressed more than 10-fold in one cell type. This method gave rather similar but different results than the data analysis above. MiR-126, miR-581, miR-1274A and miR-126# were endothelial cell-specific whereas miR-106a#, miR-484 and let 7b were specific for mesangial cell. For podocytes the same miRs as in the analysis above came up: miR-200b, miR-1225, miR-221, miR-1267 and miR-331. MiR-1305, miR-520b and miR-486 were only up-regulated in tubular cells more than 10-fold (Table S3).

| MiRs regulated by TGF-β in cultured renal cells
To identify cell stress inducible miRs we compared miR-profiles from cultured renal cell lines before and after stimulation with TGF-β. We generated fold changes in miR-expression levels after TGF-β stimulation (Table S4) and looked at the context cell type-specific up-or down-regulation of miRs. Figure 3A-D gives the number of miRs upregulated (fold change >1.5, red), down-regulated (fold change <0.5, green) or unregulated (fold change >0.5 and <1.5, overlap) after stimulation with TGF-β in human mesangial cells ( Figure 3A), human glomerular endothelial cells ( Figure 3B), human podocytes ( Figure 3C) and human tubular cells ( Figure 3D).
Of note, most miRs were down-regulated following the stimulation with TGF-β in all cell types. Up-regulation of miR-expression after stimulation with TGF-β was most prominent in cultured glomerular endothelial cells and podocytes. Moreover, up-regulation of miRs after TGF-β was cell type-specific for most miRs. For example, miR-378a-3p was specifically up-regulated in cultured human podocytes. We previously described the importance of miR-378a-3p for glomerular filter function and ultrastructure. 17 Interestingly, the regulation of some miRs was concordant for more than one cell type. For example, miR-199a-3p was up-regu- pathways. It is striking that we found no miR significantly up-regulated in more than two cell types, indicating that the miR-regulation after TGF-β is clearly cell type-specific.

| MiRs in cell-free urines from patients with different glomerular diseases compared to control
Next, we examined in a screening experiment urinary miR-profiles from pooled urine samples from patients with different glomerular diseases and compared them to those from healthy controls (Table   S5). Interestingly, except for ANCA and IgA-GN, most miRs were higher expressed in disease urines compared to controls. Figure 4 gives  Table S4).
We were able to identify disease-specific miRs that were only detectable in one glomerular disease and not present in controls.

| Assigning urinary miRs to the different renal cell types
By combining the cell and urinary miR-screenings, we could assign the miRs found in the urines from patients with glomerular diseases to the miRs identified in TGF-β stressed cultured renal cells ( Figure 5A-H). The combination of cell and urine miR-screenings enabled us to identify cell type-specific miRs in the different glomerular diseases (Table 3 and Figure 5A-H).

| D ISCUSS I ON
It is estimated that 60% of the total human proteome is regulated by about 2000 known miRs. 18 MiRs also seem to play an important role in gene regulation in disease processes.  Most studies on urinary miRs used the urine sediment obtained after low-speed centrifugation. However, a large number of lowquality and degraded RNA was recently detected in the urinary cell pellet. 19 Therefore, we decided to use the cell-free supernatants of pooled patient urines in our miR-screening. We first wanted to identify potentially cell type-specific miRs. We could find miR-126 as one of the glomerular endothelial cell-specific miRs in our screening.
Well in line with this finding, miR-126 was already described as an endothelial cell-specific miR that governs vascular integrity in other tissue. [20][21][22][23] MiR-143-3p was predominantly expressed in podocytes in our miR-screening. We have confirmed the importance of this miR for the maintenance of a functional glomerular filtration barrier in the zebrafish model. 16 MiR-30 family members were also highly expressed in our cultured human podocytes. In line with this, a role of the miR-30 family in podocyte homeostasis was previously described in mice. 24 The comparison of cellular miR-profiles before and after stimulation with TGF-β revealed cell stress-induced miRs. Most miRs up-regulated after TGF-β were cell type-specific and none was regulated in more than two renal cell types. Some cell type-specific regulations were consistent with previous findings. For example, miR-143-3p and miR-378a-3p are known to be regulated by TGF-β in non-renal cells. 25,26 Both miRs were also up-regulated in cultured human podocytes after stimulation with TGF-β in our miR-screening. Circulating miR-210 predicts survival in critically ill patients with acute kidney injury. 27 Interestingly, miR-210 was up-regulated after TGF-β in human mesangial cells in our study. MiR-199a-3p was upregulated in human mesangial cells and podocytes after stimulation with TGF-β. A known target of miR-199a-3p is versican also produced by podocytes and mesangial cells. 28,29 Most miRs were down-regulated following the stimulation with TGF-β in our miR-screening in cultured human renal cell lines.
TGF-β is associated with the increase in many pro-fibrotic and The miRs down-regulated in our cultured renal cell lines after stimulation with TGF-β may give cues to look for interesting up-regulated targets to prove TGF-β mediated fibroses and inflammation regulated through miRs. Elevating miRs that are decreased by TGF-β by miR-mimics might have therapeutically potential.
Regarding the urinary miR-screening, most miRs were higher expressed in patients with glomerular diseases compared to healthy controls. This might be due to the increased leakiness of the glomerular filtration barrier in glomerular disease that also allows more micro particles and RNA-binding proteins associated with miRs to pass. 33  Nevertheless, the origin of these urinary miRs is unknown. 35 In our screening, miR-21 was expressed in cultured human me- In animal models of kidney injury, miR-21 expression was found to be increased as well. However, its function remains controversial, because it has been implicated in promotion as well as protection from tubule-interstitial and glomerular injury. [36][37][38][39] MiR-21 was described to ameliorate TGF-β and hyperglycemiainduced glomerular injury through repression of pro-apoptotic signals. 34 In contrast to this, murine models of tubule-interstitial kidney injury demonstrated that miR-21 contributes to fibrogenesis and epithelial injury. 36 In line with these findings, Our findings and data from the literature suggest that miR-21 might have diverse functions in different glomerular disease contexts and its regulation might be much more complex than initially suggested.
MiR-378a-3p was found in urines samples from patients with MGN, FSGS and MCD and up-regulated in cultured human podocytes in response to TGF-β treatment. In a previous study we could confirm the importance of miR-378a-3p for glomerular function as it is a regulator of the glomerular matrix protein nephronectin. 17 Thus, our data confirm that urinary miRs seem to be markers for renal injury.
MiR-155 was the only miR that was up-regulated in all disease urines compared to control. In all different analysis strategies in cells including the analysis after TGF-β-stimulation, miR-155 was highly up-regulated in glomerular endothelial cells. TGF-β-regulation of miR-155 has been described before. 41 Well in line with our observation miR-155 was previously described by others to play an important role in endothelial cell activation. 42,43 F I G U R E 4 (Contiunued) Patients with diabetic nephropathy displayed reduced levels of serum miR-31. Moreover, miR-31 levels were positively correlated with leucocyte rolling velocity and negatively associated to leucocyte adhesion, TNFα, IL-6 and ICAM-1 levels. 44 We did not found miR-31 among the top 10 miRs down-regulated in diabetic nephropathy compared to control in our miR-screening. This highlights the importance of the type of bio-fluid investigated in miR-studies.
Overexpression of miR-370 was shown to promote mesangial cell proliferation and extracellular matrix accumulation by suppressing CNPY1 in a rat model of diabetic nephropathy. However, this miR was not among the top 10 miRs regulated in human renal cells or urines from patients. This indicated that a species-specific miR-analysis seems to be important as there might be differences in miR-expression in humans and rats. 45 Combining the cellular and urinary miR-screening, we were able to define disease-specific and cell type-specific miR-profiles and could assign the urinary miRs to the different renal cell types. To