Early involvement of cellular stress and inflammatory signals in the pathogenesis of tubulointerstitial kidney disease due to UMOD mutations

Autosomal dominant tubulointerstitial kidney disease (ADTKD) is an inherited disorder that causes progressive kidney damage and renal failure. Mutations in the UMOD gene, encoding uromodulin, lead to ADTKD-UMOD related. Uromodulin is a GPI-anchored protein exclusively produced by epithelial cells of the thick ascending limb of Henle’s loop. It is released in the tubular lumen after proteolytic cleavage and represents the most abundant protein in human urine in physiological condition. We previously generated and characterized a transgenic mouse model expressing mutant uromodulin (TgUmodC147W) that recapitulates the main features of ATDKD-UMOD. While several studies clearly demonstrated that mutated uromodulin accumulates in endoplasmic reticulum, the mechanisms that lead to renal damage are not fully understood. In our work, we used kidney transcriptional profiling to identify early events of pathogenesis in the kidneys of TgUmodC147W mice. Our results demonstrate up-regulation of inflammation and fibrosis and down-regulation of lipid metabolism in young TgUmodC147W mice, before any functional or histological evidence of kidney damage. We also show that pro-inflammatory signals precede fibrosis onset and are already present in the first week after birth. Early induction of inflammation is likely relevant for ADTKD-UMOD pathogenesis and related pathways can be envisaged as possible novel targets for therapeutic intervention.


Results
Transcriptional profiling of kidneys from Tg UmodC147W and Tg Umodwt mice. We previously showed that Tg UmodC147W mice already display mild tubulointerstitial damage at 3 months of age, and that renal damage progresses into severe interstitial fibrosis, inflammatory cell infiltrate, tubular atrophy and renal failure by 6 months. Up to this time point Tg Umodwt mice, i.e. transgenic mice for wild type uromodulin serving as an expression-matched control, were similar to control, non-transgenic mice 25 .
Aiming at identifying early events triggered by the expression of mutant uromodulin, we performed transcriptional profiling of kidneys from 1 month-old Tg UmodC147W and Tg Umodwt mice (males and females). At this age mutant uromodulin is already strongly accumulating in the ER of TAL cells in the kidneys of Tg UmodC147W mice (Supplementary Figure 1), while no renal damage can be observed at the histological level, except for an increase in the number of tubular casts formation ( Fig. 1 and Supplementary Figure 2). We also carried out transcriptional profile analysis on kidneys from female mice at a later time point (2 month-old), with the goal of identifying pathways related to the progression of the disease. In kidneys from 1 month-old mice we obtained positive hybridisation signal for 19,026 (males) and 18,523 (females) probes, corresponding to 13,662 and 13,357 genes expressed in males and females respectively. At 2 months of age 18,302 probes, corresponding to 13,187 genes, were expressed. Multivariate analyses performed on expressed genes using hierarchical clustering and Principal Component Analysis (PCA) allowed the separation of Tg UmodC147W sample groups from Tg Umodwt ones (Fig. 2). This demonstrates that mutant uromodulin expression is inducing significant transcriptional changes in the kidney at a time point in which renal structure is still preserved. Differential expression analysis was assessed in each of the 3 age-and sex-matched comparisons between Tg UmodC147W and Tg Umodwt mice, using a cut-off adjusted P value 0.05 and fold change 2. The number of genes up-regulated in 1 month-old Tg UmodC147W mice were 92 and 68 while the down-regulated ones were 179 and 89 in males and females respectively. In the 2 month-old Tg UmodC147W mice we found 41 up-regulated and 15 down-regulated genes ( Table 1). We validated results from the microarray analysis by real-time RT-qPCR, selecting genes that were differentially expressed (adjusted P < 0.05) in all comparisons. In particular, we chose transcripts highly (fold change > ± 2; Col1a1 and Slc25a17), moderately (fold change ~ ±1.5; Ntn4 and Slc13a3) or only slightly (fold change ~ +1.1; Panx1) up-or down-regulated in Tg UmodC147W mice. Differential expression was confirmed for all analysed genes ( Table 2).
Pathways related to inflammation, fibrosis and lipid metabolism are altered in the kidneys of 1 month old Tg UmodC147W . We carried out pathway analysis on transcriptional data from 1 month-old transgenic mice by combining results from GSEA (Gene Set Enrichment Analysis) (FDR or q value < 0.05) and DAVID functional annotation tools.
Clustering analysis with GSEA using Kyoto Encyclopedia of Genes and Genomes (KEGG) database shows up-regulation of pathways related to inflammation and fibrosis, and a decrease of fatty acid and amino-acid metabolism (Tables 3 and 4). Consistently, clustering of differentially expressed genes according to their cellular localization shows up-regulation of components of the extracellular matrix and down-regulation of peroxisomal and mitochondrial genes (data not shown). Very similar results were obtained performing analysis with DAVID (data not shown). Identification of fibrosis and inflammation as the most represented signatures in 1 month-old Tg UmodC147W mice is interesting, as no clear signs of renal fibrosis or inflammation were detected at the histological level in these mice (Fig. 1). In order to confirm up-regulation of inflammatory pathways we assessed by RT-qPCR the expression of chemokines (i.e. Ccl5, Ccl12 and Ccl19) that are up-regulated at later time points in Tg UmodC147W mice (Supplementary Figure 3). Interestingly, transcripts of these chemokines are indeed already increased in the kidneys of mutant mice at 1 month of age (Fig. 3). Also, we confirmed by RT-qPCR the increased expression of genes involved in fibrosis (i.e. Tgfb1, Vim, Col6a1 and Acta2) and the reduced expression of genes belonging to lipid metabolism pathways (i.e. Acox3, Ehhadh and Cyp4b1) (Fig. 3).
We further characterised inflammation in the kidneys of 1 month-old Tg UmodC147W mice by assessing the expression of markers of inflammatory cells by RT-qPCR analysis. While expression of Cd5 (T cells) and Cd19 (B cells) was barely detectable (Cp > 35), we found robust expression of Cd45 (Ptprc) (common leukocyte marker), Cd68 (macrophage) and Cd15 (Fut4) (granulocyte). Relative expression analysis indicates increased infiltrating inflammatory cells, essentially macrophages, in the kidneys of Tg UmodC147W mice (Fig. 4a). This was confirmed by immunohistochemistry analysis showing areas of focal macrophage invasion in the kidneys of 1 month-old Tg UmodC147W mice (Fig. 4b).
Inflammatory pathways are associated with renal disease progression in Tg UmodC147W mice. Pathway analysis on data from kidneys of 2 month-old mice confirmed findings on 1 month-old mice, with enrichment of genes related to inflammation and fibrosis (up-regulated) and to metabolism of fatty acids and amino acids (down-regulated) (data not shown). In order to identify pathways associated with disease progression, we compared transcriptional profiles derived from transgenic female mice at 1 and 2 months of age. We identified 477 genes that are significantly up-(196) or down-regulated (281) in Tg UmodC147W mice at 2 months of age, regardless of their fold change, and that were not differentially expressed at 1 month of age. Pathway Immunohistochemistry analysis for total uromodulin shows that the protein is mainly distributed at the apical membrane of TAL cells in Tg Umodwt mice, while it is intracellularly enriched in Tg UmodC147W mice (left panels, scale bar 100 µm). Representative images of kidney sections (right panels, PAS, scale bar 200 µm) and quantification of histological parameters (bottom) are shown. Data are expressed as mean ± s.d. (n = 9 Tg Umodwt and 6 Tg UmodC147W ). Tg UmodC147W mice show a relatively well preserved kidney structure. Only tubular casts, mostly uromodulin-positive (left panel, arrowhead) resulted to be increased. **P < 0.01 (Mann-Whitney test). enrichment analysis (clusterProfiler Bioconductor package, see Methods) identified up-regulated pathways related to inflammation and immune response, highlighting their growing activation along with disease progression. Noteworthy, pathways related to fatty acid and lipid metabolism were significantly enriched among down-regulated ones (Fig. 5). We also identified few genes showing opposite behaviour in 1 and 2 month-old mice: 5 genes that resulted significantly up-regulated in 1 month-old (A630097A12Rik, Adnp, Ren1, Spin, Ugt1a13) and significantly down-regulated in 2 month-old Tg UmodC147W mice and 5 genes (4931408A02Rik, Gpx4, Igh-VJ558, LOC277193, Slc9a8) that showed opposite direction of age-related change. Future work will be needed to establish the significance of these findings for disease progression.
TAL stress and inflammatory signals represent an early event in ADTKD-UMOD. Our data showed induction of pathways related to inflammation and fibrosis in the kidneys of Tg UmodC147W mice already at  1 month of age, demonstrating that expression of mutant uromodulin induces release of pro-inflammatory and pro-fibrotic signals well before the presence of any sign of renal damage at the histological level. To gain further insight into the potential role of inflammatory/fibrotic signals in the disease onset, we analysed kidneys of mice 8 days postpartum (p8). Interestingly, ER retention of mutant uromodulin is already evident at this early time point (Fig. 6).
In line with what observed at 1 month of age, evaluation of histological parameters reveals that kidneys of Tg UmodC147W mice at p8 are similar to the ones of Tg Umodwt mice (Supplementary Figure 4). We then analysed the transcript level of cytokines/chemokines, genes related to fibrosis and inflammatory cell markers in these kidneys. Interestingly, Ccl5 and Ccl12 genes were significantly increased in Tg UmodC147W mice, while no differences were detected in expression of Ccl19 and of transcripts related to fibrosis or lipid metabolism (Fig. 7). Upregulation of chemokines in p8 kidneys was associated with higher expression of markers of inflammatory cells, indicating increased macrophage infiltrate (Fig. 8a-c). These results suggest that the induction of pro-inflammatory signals is likely one of the first events occurring in the kidneys of Tg UmodC147W mice following mutant uromodulin expression.
We also checked the expression of Lcn2 (lipocalin 2 or Neutrophil gelatinase-associated lipocalin (NGAL)) and Hacvr1 (Hepatitis A virus receptor or Kidney injury molecule-1 (Kim1)) genes. Lcn2 and Havcr1 are well-established markers of distal and proximal tubule damage respectively that were found to be significantly up-regulated already at 1 month of age in the transcriptome analysis of Tg UmodC147W kidneys. At p8 we detected both at transcript and protein level significant up-regulation of Lcn2 while expression of Kim1 was not different in Tg UmodC147W mice (Fig. 9a,c). These results suggest that renal damage occurs in the distal tubules first, where uromodulin is expressed and is then spread to neighbouring proximal tubules. Data from microarray analysis show up-regulation of Atf3 transcript. ATF3 is a transcription repressor that was shown to play a protective role in the kidney after ischemia/reperfusion injury through anti-inflammatory and anti-apoptotic effects in renal tubular epithelial cells 31 . It can be induced by a large variety of stresses, including ER stress, cytokine, genotoxic and cytotoxic agents 32 . Interestingly, we found increased expression of Atf3 at transcript and protein level (Fig. 9a,b), specifically in TAL nuclei at p8 (Supplementary Figure 5). These data suggest early induction of cellular stress in TAL cells of Tg UmodC147W mice.

Discussion
In this study, we aimed at identifying early renal pathways triggered by expression of mutant uromodulin in a mouse model of ADTKD-UMOD. Our results derived from transcriptional profiling of kidneys from young adult mice show that inflammation and fibrosis are two of the most up-regulated pathways, while down-regulated ones are mostly related to lipid and amino acid metabolism. In line with this, clustering of DEGs according to their localization indicates up-regulation of components of the extracellular matrix and down-regulation of peroxisomal and mitochondrial genes. Moreover, the comparison of transcriptional data from 1 and 2 month-old Tg UmodC147W mice indicate that inflammatory pathways may play a role in the progression of the disease.
Deregulation of some of the identified pathways has already been observed in other models of renal damage. By using a genetic model of nephronophthisis, i.e. mice lacking Glis2, it was shown that tubulointerstitial infiltrating cells and fibrosis are already present in kidneys of young animals 33 . Also, in a genetic model of Alport syndrome tubulointerstitial nephritis associated with presence of inflammatory cells is one of the major histological features 34 . Fibrotic pathways have also been shown to be up-regulated at 3 weeks of age in a rat model of polycystic kidney disease 35 . More recently, down-regulation of amino-acid and lipid metabolism has been associated with renal damage progression in humans and in mouse models of tubulointerstitial fibrosis 36 . Markers of lipid metabolism were indeed strongly reduced in fibrotic human kidneys. Moreover, restoring fatty acid metabolism by genetic or pharmacological methods protected mice from tubulointerstitial fibrosis 36 . Interestingly, also reduction of mitochondrial activity has been linked to kidney diseases 37 . Moreover, decreased number of functional peroxisomes was shown to worsen tubulointerstitial damage 38 . Overall, these data suggest that the main pathways that are dysregulated in kidneys of Tg UmodC147W mice may reflect common features of chronic kidney disease onset and progression.
By using two different mouse lines carrying Umod mutations (Umod C93F , Umod A227T ) induced by N-ethyl-N-nitrosourea (ENU), Kemter et al. showed that inflammation could play a role in models of ADTKD-UMOD, via activation of NF-kB pathway in TAL segments 39 . Consistently, pathway analysis of transcriptome data from female Tg UmodC147W mice shows up-regulation of NF-kB pathway (Biocarta database, data not shown), suggesting that this pathway has a role also in Tg UmodC147W mice. Moreover, Horsch et al. performed transcriptional profiling of kidneys from young-adult Umod A227T mice (17 weeks, mild disease model) and aged  Table 2. Fold change of kidney gene expression in Tg UmodC147W mice relative to age-and sex-matched Tg Umodwt . *P < 0.05; **P < 0.01; ***P < 0.001 (unpaired t-test). § P < 0.05; § § P < 0.01; § § § P < 0.001 (Benjamini Hochberg adjusted).  Umod mutations, mouse strain, microarray platform (in-house produced versus commercially available, covering 17,346 genes and 48,804 transcripts respectively). Despite such differences, it is interesting to note that some of the main pathways identified in our study, as inflammation and lipid metabolism, were also detected in the study by Horsch et al. It is also of note that, similarly to what we observed, also in ENU models the number of DEGs decreases with disease progression. Our data suggest that inflammation has relevance in the onset and progression of the renal disease in Tg UmodC147W mice as we identified increased levels of chemokines in kidneys of mutant mice already at p8. At this time point, we were also able to find presence of infiltrating cells. Real time analyses showed that inflammatory cells are mainly represented by cells of the innate immunity type (i.e. macrophages) instead of cells of the adaptive immunity (i.e. T and B cells). This is in line with what observed in an Alport syndrome mouse model, that    develops a chronic renal disease 34 . Although partial, our data also indicate that inflammation in Umod mutant mice precedes development of fibrosis. Indeed, transcript levels of Tgfb, a key factor regulating the fibrotic process that is found up-regulated in virtually any type of CKD in human and mouse models 40 , Vim, Col6a1 and Acta2 are identical in mutant and wild type transgenic mice at one week of age. Similarly to Tg UmodC147W mice, inflammation preceding fibrosis was also described in several mouse models of induced renal damage 41 . Overall, these data suggest that inflammation could be a target of future pharmacological intervention directed at treatment of the pathology. Early development of inflammation is coupled with tubular damage, as demonstrated by increased expression of Lcn2 in kidneys from Tg UmodC147W mice at p8. Lipocalin-2 transcript remains at high levels in adult mice, and it results as one of the most up-regulated DEGs in our transcriptome analysis. Lipocalin-2 is an established marker of distal tubule damage 42 that regulates progression of CKD and cyst formation 43 . It was found to be highly up-regulated in renal tissues of other genetic models of progressive kidney damage 44 . Havcr1, a marker of proximal tubular damage 45 , is found to be up-regulated in adult Tg UmodC147W mice used for the transcriptional analysis while is identical to Tg Umodwt mice at p8. These results might indicate that tubular damage takes place in distal tubules, where uromodulin is expressed and then propagates to proximal tubules, possibly via a cross-talk between TAL and proximal tubular segments.
Starting from 8 days after birth, we also identified in kidneys of Tg UmodC147W mice up-regulation of transcriptional repressor Atf3 that remains at high levels in adult animals. ATF3 belongs to the activating transcription factor/cAMP responsive element-binding protein (ATF/CREB) family. It can be activated by several stimuli, including ER stress and unfolded protein response (UPR), chemokines and cytokines, genotoxic and cytotoxic agents. ATF3 plays a protective role after renal ischemic injury 31, 46, 47 through anti-inflammatory and anti-apoptotic effects on epithelial cells 46 . The anti-inflammatory effect of ATF3 is thought to be exerted via chromatin remodelling and subsequent transcriptional inhibition of gene targets of NF-kB pathway 46,48 . In Tg UmodC147W mice Atf3 is specifically induced in the nuclei of TAL cells, likely exerting a protective role in mutant mouse kidneys. Given the primary effect of uromodulin mutation, it is likely that Atf3 is induced in TAL cells because of ER stress. Indeed, mutations in UMOD leading to protein misfolding and accumulation in the ER likely elicit ER and oxidative stress pathways that play a main role in the disease pathogenesis. Evidence for UPR induction in Tg UmodC147W mice is limited in our dataset, as it was identified only in the group of male mutant mice (Reactome database, Unfolded_Protein_Response, ATF4 and PERK, data not shown). This is likely due to the fact that such cell stress pathways are expected to be increased in TAL cells only. The use of RNA from total kidneys may have masked their induction due to a dilution effect. Nevertheless, very recent data indicate UPR induction 28,49 and derangement of mitochondria 49 in TAL segments of ADTKD-UMOD mouse models.
The induction of UPR could also be upstream of inflammation, and the coupling of these responses in specialized cells and tissues is now thought to be fundamental in the pathogenesis of inflammatory diseases 50 . In kidneys from p8 mutant mice we identified increased expression of Ccl5 (Rantes) and Ccl12 (MCP-5), that is a structural and functional homologue of the human CCL2 (MCP-1) 51 . Further studies will be need to assess if these chemokines are produced by TAL epithelial cells, possibly downstream of ER stress signals, and play a role in the onset and progression of tubulointerstitial disease 52 .
In conclusion, our study identifies renal induction of inflammatory signals as an early event in the pathogenesis of ADTKD-UMOD. Further characterisation of these pathways is warranted to assess their relevance in the disease and as potential targets of novel therapeutic intervention.

Materials and Methods
Transgenic mice. A detailed description of the methods for the generation of Tg UmodC147W and Tg Umodwt lines was previously reported 25 . Both mutant (Cys147Trp) and wild type transgenes comprise a 2.9 Kb fragment of the mouse Umod gene promoter, non-coding exon 1, intron 1, the coding sequence from exon 2 to 11 and the entire 3′ UTR. An HA-tag was inserted at the uromodulin N-terminus after the leader peptide. Full constructs were injected in FVB mouse. All animal procedures were carried out at San Raffaele Scientific Institute, Milan, Italy, according to, and approved by, the San Raffaele Institutional Animal Care and Use Committee (IACUC 571). All animal studies were performed in adherence to the Guide for the Care and Use of Laboratory Animals as published by the US National Institutes of Health.
Tissue collection and preparation. To collect renal tissues, mice were sacrificed by decapitation after anaesthesia with avertin. Kidneys were taken and immediately homogenized in TRIzol reagent (Invitrogen, Thermo Fisher Scientific, Waltham, MA) for RNA extraction or in appropriate lysis buffer for protein extraction. Renal tissues were fixed in 4% paraformaldehyde and paraffin-embedded for immunohistochemical and histological analysis or embedded in killik embedding medium (Bio-Optica, Milan, Italy) for immunofluorescence staining.
Microarray Analysis. Total RNA was extracted from whole kidney with TRIzol reagent, quantified using the Nanodrop 8000 spectrophotometer (Thermo Fisher Scientific) and qualitatively analysed by agarose gel electrophoresis. Extracted RNA was treated with DNase (Qiagen, Hilden, Germany) and purified using RNeasy Mini Kit columns (Qiagen), according to the manufacture's protocol. Four animals for each experimental group were analysed. Nanodrop 8000 Spectrophotometer and its quality was evaluated with RNA 6000 NanoChip on the Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA). cRNA (1,500 ng) was then hybridized to the BeadChip Array and stained with streptavidin-Cy3. All procedures were performed following the manufacturer's instructions. BeadChips have been imaged using the IlluminaBeadArray Reader, a two-channel 0.8-μm-resolution confocal laser scanner, and the Illumina BeadScan software. The software IlluminaGenomeStudio v.2011.1 was used to elaborate the fluorescence signal to a value whose intensity corresponds to the quantity of the respective transcript in the original sample. The same software was used to assess quality controls, including the biological specimen control, hybridization controls, signal generation controls, and negative controls. All quality controls were satisfactory. Raw data from the SampleProbeProfile txt file were exported from GenomeStudio software. Gene expression data were normalized using the cubic spline algorithm implemented in the Illumina ® GenomeStudio software. An initial selection of "expressed" probes was done by filtering on the "detection P value" parameter. Transcripts with an intensity value significantly different from that of the background (detection P value < 0.01) in at least one sample of an analysed series (i.e. experimental groups taken into consideration) were considered "expressed" genes. Transcriptome data reported in this publication have been deposited in NCBI's Gene  Figure 6). Data are expressed as mean ± s.d. *P < 0.05 (unpaired t-test).
Expression Omnibus 53 and are accessible through GEO Series accession number GSE97093 (https://www.ncbi. nlm.nih.gov/geo/query/acc.cgi?acc=GSE97093). Subsequent multivariate analyses and differential expression were performed on the subset of expressed genes in the groups defined by the experimental design. Principal Component Analysis (PCA) was done on expressed genes using scripts in Rstudio 54 . Hierarchical clustering was carried out on the 150 genes showing the greatest inter-group variance among compared groups. Genes and samples have been clustered using Euclidean distance and average clustering algorithm (pheatmap; https:// cran.r-project.org/web/packages/pheatmap/index.html). Determination of differentially expressed genes and processes was done using LIMMA Bioconductor package 55 . Pair-wise comparisons (i.e. "contrasts") were defined among the experimental groups considered in a specific design. Benjamini Hochberg multiple comparison correction was applied to the P value and genes passing a cut-off adjusted P value < 0.05 were considered differentially expressed in the specified comparison. Gene Set Enrichment Analysis (GSEA) (http://www.broadinstitute. org/gsea/index.jsp), which identifies groups of genes enriched towards the top or bottom of a ranked list of genes based on a running sum statistics, was used to identify functionally related groups of genes whose expression pattern was correlated with the template, defined by the hallmark curated gene sets from MSigDB 56,57 . Pathway enrichment analysis was also performed by using DAVID software (https://david.ncifcrf.gov/).
Pathways enriched during the progression of renal disease in Tg UmodC147W mice were identified by carrying out a biological Term Enrichment Analysis using KEGG Pathways database 58 and clusterProfiler Bioconductor package 59,60 . We compared the prevalence of gene annotations (KEGG Pathways) among differentially expressed genes (i.e. significantly up-or down-regulated in Tg UmodC147W female mice at 2 months of age, regardless of their fold change, and not differentially expressed at 1 month of age in sex-matched mice) to their prevalence in a background defined by all the expressed genes (as defined above). Fold enrichment was defined as the ratio gene frequency/background frequency (ratio of the number of genes annotated to a given biological term in the list of DEGs over the number of DEGs)/(ratio of the number of genes annotated to that biological term in the entire background set over the total number of genes in the background set).

Real-time qPCR analysis.
Total RNA from mouse whole kidney was extracted by homogenization in TRIzol reagent. RNA sample concentrations were determined by using Nonodrop-8000 and RNA quality was assessed on agarose gel. For each sample 1 μg of extracted RNA was reverse-transcribed using iScript kit (BioRad Laboratories, Hercules, CA) according to the manufacturer's protocol. Expression of target genes was analysed by RT-qPCR on LightCycler 480 (Roche, Basel, Switzerland) using qPCR Core kit for SYBR Assay (Eurogentec, Liège, Belgium). Specific primers were designed by using Primer 3 (a list of all primers used in this study is reported in Supplementary Table 1) and the amplification efficiency for each couple of primers was determined by dilution curves. Expression of genes of interest was normalized to Hprt1. The relative mRNA expression was calculated following the ΔΔCT method 61 .
Immunofluorescence staining. Immunofluorescence was carried out using standard protocol on 7 µm-thick kidney sections. Briefly, after 1 h in blocking solution, slides were incubated over-night with specific primary antibody (see below) and then with appropriate AlexaFluor-labeled secondary antibody (Life Technologies, Thermo Fisher Scientific). All slides were viewed under a DM 5000B fluorescence upright microscope (Leica DFC480 camera, Leica DFC Twain Software, 40X/0.75 lens) (Leica Microsystems, Wetzlar, Germany) or under an UltraVIEW ERS spinning disk confocal microscope (UltraVIEW ERS-Imaging Suite Software, Zeiss 63X/1.4; PerkinElmer Life and Analytical Sciences Boston, MA). Identical acquisition parameters were ensured for the visualization of the same antibody in different kidney sections. All images were imported in Photoshop CS (Adobe Systems, Mountain View, CA) and adjusted for brightness and contrast.