Podocyte-Selective Deletion of Dicer Induces Proteinuria and Glomerulosclerosis

Abstract

Dicer is an enzyme that generates microRNA (miRNA), which are small, noncoding RNA that function as important regulators of gene and protein expression. For exploration of the functional roles of miRNA in glomerular biology, Dicer was inactivated selectively in mouse podocytes. Mutant mice developed proteinuria 4 to 5 weeks after birth and died several weeks later, presumably from kidney failure. Multiple abnormalities were observed in glomeruli of mutant mice, including foot process effacement, irregular and split areas of the glomerular basement membrane, podocyte apoptosis and depletion, mesangial expansion, capillary dilation, and glomerulosclerosis. Gene profiling revealed upregulation of 190 genes in glomeruli isolated from mutant mice at the onset of proteinuria compared with control littermates. Target sequences for 16 miRNA were significantly enriched in the 3′-untranslated regions of the 190 upregulated genes. Further suggesting validity of the in silico analysis, six of the eight top-candidate miRNA were identified in miRNA libraries generated from podocyte cultures; these included four members of the mir-30 miRNA family, which are known to degrade target transcripts directly. Among 15 upregulated target genes of the mir-30 miRNA, four genes known to be expressed and/or functional in podocytes were identified, including receptor for advanced glycation end product, vimentin, heat-shock protein 20, and immediate early response 3. Receptor for advanced glycation end product and immediate early response 3 are known to mediate podocyte apoptosis, whereas vimentin and heat-shock protein-20 are involved in cytoskeletal structure. Taken together, these results provide a knowledge base for ongoing investigations to validate functional roles for the mir-30 miRNA family in podocyte homeostasis and podocytopathies.

MicroRNA (miRNA) are a group of noncoding small RNA that are present in lower through higher organisms and function to regulate gene expression by translation repression or transcript degradation of target genes.¹ A miRNA gene is transcribed by RNA polymerase II² or III³ to form a primary transcript. This primary transcript is then cleaved by Drosha,⁴ an RNase III enzyme, in the nucleus to form an approximately 70-nt-long product that is characteristic of a stem-loop structure and is termed precursor miRNA (pre-miRNA). Pre-miRNA is transported to the cytoplasm by exportin 5, a nuclear membrane transporter.⁵ In the cytoplasm, all pre-miRNA are further processed by Dicer, another RNase III enzyme, to form approximately 22-nt mature double-stranded miRNA.⁶

The importance of miRNA in development and tissue homeostasis is now well documented. Abrogating Dicer to prevent production of functional miRNA in mice resulted in embryonic lethality at approximately embryonic day 7.5.⁷ Studies of conditional ablation of Dicer in immune cells,^8,9 developing oocytes,^10,11 limb buds,¹² Purkinje cells,¹³ chondrocytes,¹⁴ and heart¹⁵ all have suggested critical roles of miRNA in multiple biologic processes in those target cells, including proliferation, differentiation, and maintenance of cell structure and function. In addition to Dicer-deficient models of global miRNA inactivation, single miRNA have been shown to be required in regulating a variety of cellular processes, for example, inactivation of miR-208 and miR-155 in mice has revealed their roles in mediating cardiomyocyte hypertrophy in response to stress¹⁶ and in regulating T helper cell differentiation and the germinal center reaction,^17,18 respectively. Besides, miRNA are implicated in other biologic processes.^19–22 Whereas recent observations suggested a role for mouse miR-192 in glomerular TGF-β and hyperglycemia-induced expression of collagen I α2 in glomeruli,²³ the roles of miRNA in kidney remain largely unknown.

Here, we report functional, morphologic, and molecular alterations induced by conditional inactivation of Dicer selectively in podocytes of mice. The mutant mice manifest a sequelae of podocyte and glomerular abnormalities, including foot process (FP) effacement, glomerular basement membrane (GBM) alterations, proteinuria, podocyte apoptosis, podocyte depletion, mesangial expansion, and glomerulosclerosis consistent with the podocyte depletion paradigm of progressive glomerulosclerosis.^24,25 Thus, miRNA control expression of genes and gene products that are essential for structural homeostasis and survival of mature podocytes in mice. The identification and characterization of specific miRNA and their target genes that are critical for podocyte maintenance are ongoing and are expected to produce novel insights into mechanisms of glomerular diseases.

RESULTS

NPHS2-Cre Efficiently Deletes Dicer and Eliminates miRNA

To confirm that podocin promoter-controlled NPHS2-Cre transgene efficiently deleted loxP-flanked sequences in podocytes, we crossed Nphs2-Cre transgenic and Rosa26 reporter mice followed by β-gal staining of kidney sections. NPHS2-Cre mediated successful recombination at the Rosa26 reporter locus as demonstrated by positive β-gal staining in glomeruli of Rosa26/+:NPHS2-Cre mice at 2 wk of age (Figure 1, A and B). A schematic representation depicts the Dicer alleles before and after Cre-mediated recombination (Figure 1C). PCR of genomic DNA obtained from isolated glomeruli confirmed Cre-mediated recombination of Dicer alleles in podocytes of glomeruli from conditional knockout (mutant) mice (Dicer^F/F:NPHS2-Cre) with and without proteinuria (Figure 1D). To confirm that Dicer deletion resulted in loss of miRNA in podocytes, we examined the abundance of mature form of miR-30d and miR-23b, two miRNA that are known to be expressed in podocytes (smiRNAdb [http://www.mirz.unibas.ch/smiRNAdb/]). Compared with control (Dicer^F/+:NPHS2-Cre), levels of both miRNA were significantly reduced in glomeruli of mutant mice and the miRNA reduction was inversely correlated with the absence or presence of proteinuria (Figure 1E). Residual levels of miRNA present in glomeruli of mutant mice were likely due to variation of efficiency of Dicer deletion in different podocytes/glomeruli and/or to expression of the miRNA in endocapillary cells. To demonstrate further that Dicer was inactive in precursor miRNA processing, we conducted quantitative real-time PCR to examine whether precursor miR-30a, 30c-2, and 30d accumulated in mutant podocytes. The result showed the abundances of these precursor miRNA in mutant podocytes were indeed significantly higher than that in control podocytes (Figure 1F).

Figure 1.

NPHS2-Cre transgene efficiently deleted Dicer floxed region, leading to downregulation of mature miRNA and accumulation of precursor miRNA in podocytes of conditional knockout (mutant) mice. (A) A glomerulus of a control mouse after β-gal staining. (B) Two glomeruli from a 2-wk-old Rosa26^F/+:NPHS2-Cre mouse after β-gal staining. The arrows point to the blue spots depicting podocytes with deletion at Rosa26 allele locus in a single glomerulus. (C) Schematic representation of Dicer alleles: Wild-type, loxP floxed, and deleted. D1 and D3 are the primers for PCR. E20 and E21 depict exons 20 and 21 of Dicer gene. loxP sites are denoted with open arrowheads. (D) Electrophoretic agarose gel analysis of PCR products from glomerular genomic DNA of mice. Lane 1, Dicer^F/+ (control); lane 2, Dicer^F/F:NPHS2-Cre (mutant) without proteinuria; lane 3, Dicer^F/F:NPHS2-Cre (mutant) with proteinuria; lane 4, Dicer^F/+:NPHS2-Cre; lane 5, no DNA (negative control). Molecular weight (MW) 1 kb⁺ DNA ladder (Invitrogen). (E) Histogram shows the abundances of miR-30d and miR-23b in mutant mice without (−PU) and with (+PU) proteinuria relative to that in control mice. The bars represent average ± SD (n = 3). (F) Histogram shows the abundances of premiR-30a, 30c-2, and 30d in mutant without (−PU) and with (+PU) proteinuria relative to that in control mice. Bars represent average ± SD (n = 3).

Conditional Knockout (Mutant:Dicer^F/F:NPHS2-Cre) Mice Developed Progressive Proteinuria, Wasting, and Death

The distribution of observed genotype frequencies among offspring of heterozygotic crosses was consistent with the expected genotype distribution on the basis of Mendelian ratio (data not shown). The mutant mice were indistinguishable from wild-type littermates within the first 5 wk of life, suggesting that Dicer deletion driven by NPHS2-Cre did not cause embryonic lethality or overt postnatal developmental phenotypes.

Progress of proteinuria was monitored periodically in cohorts of mutant (Dicer^F/F:NPHS2-Cre) and control (Dicer^F/+:NPHS2-Cre, Dicer^F/F, and Dicer^F/+) littermate mice by urine dipsticks. The average age at proteinuria onset was 4.7 ± 2.0 wk (SD), ranging from approximately 2 wk to 10 mo of age (n = 50). Once proteinuria was detectable, it progressed subsequently to high-grade proteinuria and the mice died within several weeks. Proteinuria progression was associated with weight loss and runted and pale appearance followed by death. The average lifespan of mutant mice was 7.89 ± 1.97 wk (SD), ranging from 4 wk to 1 yr (n = 26). There was no gender dimorphism in lifespan and any of the observed phenotype manifestations (15 males, 11 females; data not shown). Mice heterozygous for Dicer deletion in podocytes (Dicer^F/+:NPHS2-Cre) were indistinguishable from wild-type mice with regard to growth, urinary protein excretion, appearance, and lifespan (n = 13).

Dicer Mutant Mice Developed Progressive Glomerulosclerosis and Tubulointerstitial Fibrosis

Kidneys of mutant mice (Dicer^F/F:NPHS2-Cre) with severe proteinuria appeared smaller, paler, and denser when compared with control kidneys (Figure 2A). Histopathology of kidney in mutant mice was assessed by examination of periodic acid-Schiff–stained sections. Low-power microscopy revealed protein casts in tubular lumens of degenerated nephrons, the frequency of which was proportional to the level of proteinuria determined by dipstick (Figure 2B). In glomeruli, progressive sclerosis was observed with increased extracellular matrix deposition, mesangial expansion, reduced cellularity, and capillary dilations (Figure 2, D through F).

Figure 2.

Dicer deficiency in podocytes resulted in progressive glomerular defects. (A) Kidneys from a Dicer^F/+:NPHS2-Cre mouse (control, left) and a Dicer^F/F:NPHS2-Cre mouse (mutant, right) that had severe proteinuria and was dying. (B) Periodic acid-Schiff (PAS) staining of a kidney section from a mutant mouse with massive proteinuria. Protein casts were present in most of tubular lumens. (C through F) Glomeruli from PAS-stained kidney sections of control (C) and mutant (D through F) mice. Glomeruli of mutant mice were sclerotic (D through F) with mesangial expansion (white arrow), extracellular deposition (white arrowhead), capillary dilation (▪), and reduced cellularity. In addition, tubulointerstitial fibrosis (black arrow), tubular dilation, and protein casts (⋄) are seen in kidney section of mutant mice (F). (G and H) Electronic microscopy examination showing FP effacement (white arrowhead), GBM irregularity (white arrow), and splitting (red arrow) of mutant mice (6-wk-old with proteinuria of 300 mg/L). Magnifications: ×2.5 in B; ×63 in C through E; ×20 in F.

Ultrastructural analysis of glomeruli by electron microscopy demonstrated normal podocyte FP and GBM morphology in mutant mice without proteinuria (n = 4) and control mice (n = 5). In contrast, FP effacement, splitting of GBM, and irregular GBM were observed in glomeruli of mice with proteinuria (Figure 2, G and H). FP effacement was first detectable in glomeruli of mutant mice with new onset of proteinuria, whereas GBM alterations were observed in mutant mice with advanced proteinuria. These results indicate that FP effacement was the earliest morphologic defect in mutant glomeruli.

Immunofluorescence labeling for podocyte differentiation markers synaptopodin and podocin expression was comparable in control and mutant glomeruli without proteinuria, but both markers were reduced in mutant mice with onset of proteinuria (Figure 3). In contrast, components of GBM, including collagen IV, laminin, and nidogen/entactin,^26–30 were indistinguishable between control and mutant mice, irrespective of the absence or presence of proteinuria (Supplemental Figure 1S). Together, these observations suggest that podocyte dedifferentiation coincides with onset of FP effacement and proteinuria in mutant mice without detectable difference of GBM components.

Figure 3.

Synaptopodin and podocin expressions both were downregulated in podocytes of mutant mice with proteinuria (+PU). Different degrees of synaptopodin and podocin downregulations are shown in the middle panels (milder) and right panels (more severe). Magnification, ×40.

Dicer Deletion in Podocyte Leads to Podocyte Apoptosis and Depletion

Podocyte apoptosis has been identified as one cause of podocyte depletion in experimental kidney diseases in mice.^31–33 Rates of apoptotic podocytes were determined by triple-immunofluorescence labeling including terminal deoxynucleotidyl transferase–mediated digoxigenin-deoxyuridine nick-end labeling (TUNEL) assay, WT1, and DAPI (Figure 4, A through D). Apoptotic podocytes were rare or absent in glomerular sections of control mice or mutant mice without proteinuria, respectively. In contrast, rates of apoptotic podocytes per 100 glomerular sections were increased in mutant mice with low-grade proteinuria or high-grade proteinuria, respectively (Figure 4E). In addition, WT1-positive cells were detectable in tubular lumens of mutant mice with proteinuria but not in control or mutant mice without proteinuria (data not shown). Rates of apoptosis per 100 glomerular sections of WT1-negative cells were also increased in mutant mice with low-grade and high-grade proteinuria, respectively. Apoptosis was detectable in tubular epithelial cells, in particular in mutant mice with high-grade proteinuria (data not shown). Taken together, these results demonstrate that apoptosis of podocytes was closely correlated with the degree of proteinuria in Dicer mutant mice.

Figure 4.

Mutant podocyte underwent apoptosis, and podocyte number of mutant mice decreased as proteinuria progressed. (A through D) TUNEL assay showing an apoptotic podocyte in the glomerulus of a mutant mouse. WT1, green; TUNEL, red; and DAPI, blue. (E) Rates of apoptotic nuclei in four groups of mice: (1) control mice (n = 9), one apoptotic podocyte found in 930 glomeruli examined; (2) mutant mice without proteinuria (−; n = 4), no apoptotic podocyte in 379 glomeruli; (3) mutant mice with low-grade proteinuria (+; n = 6), three apoptotic podocytes in 590 glomeruli; (4) mutant mice with high-grade proteinuria (++; n = 7), 19 apoptotic podocytes in 874 glomeruli. (F) Podocyte number among four groups of mice as described in E shows podocyte numbers of mutant mice with proteinuria were reduced significantly. **P < 0.01.

WT1 immunostaining was performed to examine whether podocyte number was altered in mutant mice. WT1 and DAPI double-positive cells were counted as podocytes as described previously.³² Five-week-old mutant mice were divided into three groups according to severities of proteinuria: (1) None, (2) low-grade proteinuria (approximately 100 μg/ml), and (3) high-grade proteinuria (≥300 μg/ml). Podocyte counts per glomerular section were comparable between mutant mice without proteinuria (n = 4) and age-matched control mice (n = 11; Figure 4F). In contrast, podocyte counts per glomerular section were significantly decreased (P < 0.01) in mutant mice with low-grade proteinuria (n = 5) and in mutant mice with high-grade proteinuria (n = 5; Figure 4F).

Global Gene Expression Profiling of Glomeruli of Mutant Mice

Inactivation of Dicer/miRNA in podocytes was expected to alter expression of specific miRNA target genes. To identify these genes, we performed microarray analyses on isolated glomeruli from three groups of 4- to 5-wk-old mice: (1) Control (Dicer^F/+:NPHS2-Cre, 5 mice), (2) mutant without proteinuria (5 mice), and (3) mutant at proteinuria onset (10 mice with proteinuria: 100 to 300 mg/L). Kidneys of mutant mice with new-onset proteinuria were characterized by tubular protein casts and evidence of segmental glomerulosclerosis and mesangial expansion in a few glomeruli, whereas the majority of glomeruli appeared normal by light microscopy. The results from microarray analyses showed that gene expression profiles of control and mutant mice without proteinuria were indistinguishable (data not shown). In contrast, 190 transcripts were significantly upregulated (q < 10%) in mutant mice with new-onset proteinuria compared with control (Supplemental Table 1S). Among those genes, 68 were >1.5-fold elevated (Table 1). A total of 121 genes were downregulated >1.5-fold (Supplemental Table 2S).

Table 1.

68 upregulated transcripts with >1.5-fold change in mutant podocytes

Gene Symbol	Gene Name	q (%)	Log Ratio	Fold Change
AU067747	–	0.00	2.96	7.79
Plod2	Procollagen lysine, 2-oxoglutarate 5-dioxygenase 2	0.00	2.25	4.77
Hspb6	Heat shock protein, alpha-crystallin-related, B6	0.00	2.18	4.54
Mm0.116781	Transcribed locus	0.00	2.09	4.27
Mm0.375137	MRNA similar to angiopoietin-like factor	0.00	2.02	4.06
Dusp14	Dual specificity phosphatase 14	0.00	1.92	3.78
Scel	Sciellin	0.00	1.83	3.56
C030009O12Rik	RIKEN cDNA C030009O12 gene	0.00	1.73	3.32
Tagln	Transgelin	5.71	1.62	3.08
Tnfrsf12a	TNF receptor superfamily, member 12a	7.87	1.60	3.04
Gabrb1	γ-Aminobutyric acid (GABA-A) receptor, subunit β1	0.00	1.56	2.95
Spp1	Secreted phosphoprotein 1	3.77	1.55	2.93
AI790298	Expressed sequence AI790298	0.00	1.50	2.84
Mt2	Metallothionein 2	9.69	1.46	2.75
D5Buc30e	DNA segment, Chr 5, Bucan 30 expressed	0.00	1.40	2.63
Fgfbp1	Fibroblast growth factor binding protein 1	0.00	1.36	2.57
Krt1–18	Keratin complex 1, acidic, gene 18	5.36	1.33	2.51
F3	Coagulation factor III	2.86	1.31	2.48
Gadd45b	Growth arrest and DNA-damage-inducible 45 beta	0.00	1.29	2.45
Serpinb6b	Serine (or cysteine) proteinase inhibitor, clade B, 6b	0.00	1.26	2.40
S100a6	S100 calcium binding protein A6 (calcyclin)	9.32	1.21	2.31
2810417M05Rik	RIKEN cDNA 2810417M05 gene	4.01	1.10	2.14
Maff	v-maf musculoaponeurotic fibrosarcoma oncogene family, F	5.56	1.03	2.05
Mm0.71633	0 day neonate skin cDNA, RIKEN clone:4632424N07	0.00	1.03	2.04
D0H4S114	DNA segment, human D4S114	3.26	1.03	2.04
Masp1	Mannan-binding lectin serine protease 1	0.00	0.99	1.98
Sbk1	SH3-binding kinase 1	0.00	0.99	1.98
Pp11r	Placental protein 11 related	0.00	0.97	1.97
BB369191	–	4.34	0.92	1.89
2900041A09Rik	RIKEN cDNA 2900041A09 gene	8.4	0.92	1.89
Ifi30	IFN-γ inducible protein 30	0.00	0.89	1.86
AI323359	–	4.83	0.88	1.84
Ier3	Immediate early response 3	4.34	0.88	1.84
Ager	Advanced glycosylation end product-specific receptor	1.93	0.87	1.83
Bhlhb2	Basic helix-loop-helix domain containing, class B2	7.87	0.87	1.83
Prnp	Prion protein	4.91	0.86	1.81
Gmpr	Guanosine monophosphate reductase	0.00	0.85	1.80
Acvr2b	Activin receptor IIB	0.00	0.84	1.79
Stx11	Syntaxin 11	5.10	0.84	1.79
Tspan4	Tetraspanin 4	2.62	0.82	1.77
Ppm1l	Protein phosphatase 1 (formerly 2C)-like	2.95	0.81	1.75
Npepl1	Aminopeptidase-like 1	3.41	0.80	1.74
Sema3b	Sema domain, immunoglobulin domain, (semaphorin) 3B	2.62	0.80	1.74
Cldn12	Claudin 12	0.00	0.79	1.72
MGI:2446326	Suprabasin	5.36	0.76	1.69
B3galt2	UDP-Gal:βGlcNAc β1,3-galactosyltransferase, polypeptide 2	4.34	0.76	1.69
9530018I07Rik	RIKEN cDNA 9530018I07 gene	6.55	0.75	1.68
Tsrc1	Thrombospondin repeat containing 1	3.26	0.73	1.66
Ppp1r3c	Protein phosphatase 1, regulatory (inhibitor) subunit 3C	6.30	0.73	1.66
Efemp1	EGF-containing fibulin-like extracellular matrix protein 1	2.99	0.73	1.66
Mthfd1l	Methylenetetrahydrofolate dehydrogenase 1-like	3.32	0.70	1.63
AI506321	–	6.89	0.70	1.62
Hmgb3	High mobility group box 3	0.00	0.69	1.62
E030024M05Rik	RIKEN cDNA E030024M05 gene	3.26	0.69	1.62
5730499H23Rik	RIKEN cDNA 5730499H23 gene	4.25	0.68	1.60
Sulf2	Sulfatase 2	5.10	0.68	1.60
Eno1	Enolase 1, α non-neuron	2.39	0.68	1.60
Chst12	Carbohydrate sulfotransferase 12	3.77	0.67	1.59
0610010O12Rik	RIKEN cDNA 0610010O12 gene	0.00	0.66	1.58
MGI:1917275	Brain acyl-CoA hydrolase	3.32	0.65	1.57
Mvp	Major vault protein	0.00	0.65	1.57
Hlf	Hepatic leukemia factor	8.40	0.64	1.56
Aplp1	Amyloid β (A4) precursor-like protein 1	7.87	0.64	1.55
Mustn1	Musculoskeletal, embryonic nuclear protein 1	9.69	0.63	1.55
Vim	Vimentin	0.00	0.63	1.55
Gclc	Glutamate-cysteine ligase, catalytic subunit	8.40	0.61	1.53
Mm0.361409	Transcribed locus	2.79	0.60	1.51
NM_011937	–	2.94	0.59	1.51

We selected 14 of the 68 upregulated genes for validation of microarray measurements by quantitative real-time PCR. The upregulation of 13 of 14 genes was confirmed (Supplemental Figure 2S). To confirm that the upregulation of gene expression was specifically in podocytes but not in other types of glomerular cells, we conducted immunofluorescence staining using available antibody against SM22α (Transgelin), a smooth muscle differentiation marker, and found that SM22α signal co-localized with synaptopodin (Supplemental Figure 3S), indicating that the upregulation of SM22α was observed in podocytes of Dicer mutant mice with proteinuria.

Identification of miR-30 Family miRNA and Their 15 Upregulated Target Genes as Candidate miRNA–Target Gene Pairs in Dicer Mutant Podocytes

miRNA have been shown to degrade transcripts directly, resulting in negative regulation of steady-state mRNA levels. Thus, we hypothesized that genes with increased mRNA levels in mutant glomeruli may be direct targets for miRNA-mediated RNA degradation.

If some of the 190 upregulated transcripts of mutant podocytes are indeed miRNA degradation targets, then select miRNA target sequences would be expected to be overrepresented in the 3′ untranslated regions (UTR) of the 190 transcripts. To examine this hypothesis, we used miRBase (http://microrna.sanger.ac.uk) to search for miRNA target sequences in the 3′ UTR of the 190 transcripts, and we found target sequences for 294 miRNA. The target sequence number of individual miRNA ranges from 1 to 19. Next, we performed permutation analyses on the occurrence of miRNA–target gene pairs to determine eight miRNA whose target sequences were significantly overrepresented (false discovery rate <5%) in the 190 upregulated transcripts (Table 2). Among them, six were previously shown to be represented in miRNA library cloning from podocyte culture according to smiRNAdb (http://www.mirz.unibas.ch/smiRNAdb/). These included miR-28; miR-34a; and four members of the miR-30 family, miR-30c-1, miR-30b, miR-30d, and miR-30c-2. Among 68 upregulated genes with >1.5-fold change, 15 contained miR-30 family target sequences, including 0610010O12Rik, Ager (receptor for advanced glycation end products [RAGE]), Aplp1, Chst12, Gabrb1, Gclc, Hspb6, Ier3, Ifi30, Mthfd1l, Npepl1, Scel, Sulf2, Tnfrsf12a, Tspan4, and Vim (Supplemental Table 4S). The miR-30 family members fulfilled all criteria for candidate miRNA. First, miR-30d was regulated in mutant glomeruli. Second, all four miR-30 miRNA are normally present in podocytes. Third, miR-30 miRNA are known RNA target–degrading miRNA. Fourth, miR-30 target sequences were highly enriched in upregulated genes in mutant glomeruli.

Table 2.

miRNA with target sequence significantly enriched in 190 upregulated transcripts

miRNA	Hits	FDR	Podocyte Expressiona
mmu-mir-30c-1	13	0.0297	x
mmu-mir-30b	19	0.0359	x
mmu-mir-28	16	0.0359	x
mmu-mir-34a	14	0.0359	x
mmu-mir-30d	13	0.0359	x
mmu-mir-762	13	0.0359
mmu-mir-30c-2	12	0.0359	x
mmu-mir-711	11	0.0359

^aThe miRNA present in podocytes according to smiRNAdb (http://www.mirz.unibas.ch/smiRNAdb/) based on limited podocyte miRNA library sequencing (only 888 clones sequenced) are marked with “x.’

DISCUSSION

Mice mutant for Dicer in podocytes manifested variable expression of progressive proteinuria, structural glomerular defects and glomerulosclerosis consistent with the podocyte depletion paradigm,^24,25 and tubulointerstitial fibrosis, demonstrating that miRNA exert essential genomic controls in podocytes. In an attempt to delineate the temporal sequence of the observed endophenotypes, we obtained evidence for podocyte dedifferentiation and FP effacement with GBM defects as the earliest detectable defects, followed by podocyte apoptosis, podocyte depletion, mesangial expansion, and segmental or global glomerulosclerosis. The age of onset and the extent of the observed defects were highly variable among mutant mice. This is likely attributable to variation in expression of transgenic Cre recombinase and Cre-dependent inactivation of Dicer in podocytes in different glomeruli within an individual mouse and/or between different mice, as has been previously observed with the NPHS2-Cre transgenic model.³⁴ Interestingly, histologic and ultrastructural glomerular defects detectable in all mutant mice were coincident only with the onset and progression of proteinuria but not before proteinuria onset.

In an unbiased approach to narrow the spectrum of molecular targets and cognate miRNA responsible for these defects, microarray analysis of glomerular gene expression profiles combined with in silico interrogation of genes with significant upregulation in mutant glomeruli identified candidate molecular mechanisms that may underlie the observed defects. The validity of our applied microarray and in silico approach to identify target-degrading miRNA for which Dicer deficiency-induced loss of expression would be expected to result directly in increased expression of their respective target transcripts is supported by the following observations. First, six of the eight most significant candidate target-degrading miRNA are known to be present in podocytes, on the basis of limited sequencing information available in the smiRNAdb database (http://www.mirz.unibas.ch/smiRNAdb/), on miRNA libraries established from murine podocytes. Second, the identification of four known target mRNA–degrading miRNA of the miR-30 family,³⁵ all of which are expressed in podocytes, among the six most significant candidate podocyte miRNA is highly significant. For example, the observed frequency of hits among the 190 upregulated genes for miR-30 family miRNA indicated between 12 and 19 targets per miR-30 miRNA, compared with the expected frequency of five hits for random miRNA among the 190 upregulated transcripts. Moreover, only a few target-degrading miRNA have been identified (let-7b,^36,37 miR-125b,³⁶ miR-1,³⁸ miR-124,³⁸ miR-122,³⁹ miR-16,⁴⁰ miR-98,⁴¹ and miR-106b⁴²) to date in addition to the miR-30 family.

Among the list of 15 target genes of the mir-30 family in our study, several are known to be expressed in injured podocytes in experimental models and human kidney disease and have functional roles in apoptosis. For example, RAGE is barely detectable in podocytes in normal control mice but is strongly expressed in the streptozotocin murine models of diabetic kidney disease and of adriamycin-induced nondiabetic podocyte injury associated with proteinuria and glomerulosclerosis.^43,44 Furthermore, RAGE is increased in podocytes of patients with diabetic nephropathy and several nondiabetic glomerular diseases.⁴⁵ Intriguingly, the known RAGE ligand S100 calcium-binding protein A6 (calcyclin) was among the top 20 upregulated genes in glomeruli of mutant mice with proteinuria (see Table 1). S100A6 binding with RAGE is known to induce apoptosis in epithelial cells through activation of Jun N-terminal kinase.⁴⁶ Thus, aberrant coincident upregulation of RAGE and its proapoptotic cognate ligand S100A6 may be caused by loss of podocyte mmu-mir-30b and possibly mmu-mir-431, respectively, and may directly contribute to podocyte injury and/or podocyte apoptosis observed in proteinuric mutant mice. The Ier3 is a predicted target of both mmu-mir-30c and mmu-mir-431. Ier3 also functions as positive regulator of apoptosis in keratinocytes⁴⁷ and has been identified in renal cDNA libraries as protein that interacts with modulators of apoptosis.⁴⁸ Podocyte apoptosis is an early glomerular manifestation leading to podocyte depletion in TGF-β1 transgenic mice and CD2AP^−/− and diabetic mouse models with albuminuria and glomerulosclerosis.^31–33 Observations in this study provide a rationale for a role of select miRNA–target gene pairs, such as miR-30 and miR-431 and RAGE and S100A6, in progressive glomerular disease.

Podocytes possess major processes containing microtubules and intermediate filaments with vimentin, and FP containing actin filaments as core cytoskeletal elements.⁴⁹ Heat-shock protein 20 (Hsp20) and the intermediate filament protein vimentin both are shared targets of mmu-mir-30c-1 and 30c-2 and interact with cytoskeletal components. The intermediate filament vimentin is upregulated in podocytes of nephrotic glomeruli in puromycin aminonucleoside model in rat.⁵⁰ Hsp20 binds to actin in vitro and in vivo, and the association with filamentous actin is dependent on the phosphorylation state of Hsp20.⁵¹ Interestingly, phosphorylated Hsp20 has a direct role in smooth muscle contractility and relaxation.⁵² In addition, Tagln (SM22α and transgelin) is an actin cross-linking protein.^53,54 The de novo expression of SM22α in podocytes of mutant glomeruli observed in our studies may disrupt the integrity of the actin cytoskeleton in podocytes and lead to FP effacement. In addition, Plod2 (lysyl hydroxylase 2),⁵⁵ is one of the three lysyl hydroxylases that mediate lysine hydroxylation of collagens, including collagen IV, which is essential for their proper assembly and cross-linking of collagen fibrils.⁵⁶ Upregulation of lysyl hydroxylase 2 may therefore underlie alterations of the collagen IV network and thus GBM defects in Dicer mutant mice. Pathway enrichment analysis based on gene expression data identified activation of integrin signaling as top pathway, consistent with the observed disruption of podocyte adhesion to GBM and GBM alterations. Integrin α-3 (Itga3) is the major α subunit of β1 integrin receptors expressed by podocytes, and podocytes in kidneys of integrin α-3 knockout (Itga3^−/−) mice are unable to maintain normal podocyte structure, including the elaboration of mature FP along the GBM.⁵⁷ Indeed, Itga3 mRNA was increased in mutant mice, consistent with integrin signaling activation. Functional validation of the discussed candidate miR-30–target gene pairs requires an expansive experimental approach and is ongoing in our laboratory.

In conclusion, our results demonstrate that miRNA are important genomic regulators of molecular podocyte homeostasis. Disruption of miRNA-controlled gene networks results in progressive proteinuria and glomerulosclerosis. These findings warrant further studies to examine whether the dysregulation of miRNA-controlled gene networks may contribute to the pathogenesis of glomerular diseases. The identification of candidate miRNA–target gene pairs in our study may provide a starting point for further study to validate their role in glomerular diseases.

CONCISE METHODS

Mouse Strain

The use of the mice in this study adhered to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The mouse line carrying a Dicer allele with two lox P sites that flank RNase III domain was created as described previously.⁸ NPHS2-Cre was a gift from Dr. Holzman.³⁴ Rosa26 reporter mice were purchased from Jackson Laboratory (Bar Harbor, ME).

Antibodies

The antibodies used in this study were as follows: WT1 (Santa Cruz Biotechnology, Santa Cruz, CA), SM22α (Abcam, Cambridge, MA), laminin (Sigma, St. Louis, MO), entactin/nidogen (Chemicon Int., Temecula, CA), collagen IV (Southern Biotech, Birmingham, AL), and podocin (Santa Cruz, Santa Cruz, CA). Monoclonal synaptopodin antibody was prepared from hybridoma cells. Fluor-conjugated second antibodies, Alexa Fluor 488 goat anti-rabbit IgG (H+L) and Alexa Fluor 568 goat anti-mouse Ig G1, both were purchased from Invitrogen (Carlsbad, CA).

β-Galactosidase Staining of Kidney

The kidneys were excised from the mice, imbedded in OCT and snap-frozen. Five-micrometer sections were cut and stained for the expression of β-galactosidase using β-galactosidase Staining Kit (Specialty Media, Phillipsburg, NJ) following the manufacturer's instructions.

PCR for Genotyping

PCR to detect the deletion product in the podocytes was conducted using the primers D1 (5′-AGTAATGTGAGCAATAGTCCCAG-3′) and D3 (5′-CTGGTGGCTTGAGGACAAGAC-3′), which locate outside the loxP floxed region. PCR reaction was performed with standard protocol using TaqDNA polymerase (Invitrogen). The thermal cycling condition was 94°C for 2 min, followed by 35 cycles of 94°C for 15 s, 57°C for 15 s, and 72°C for 30 s. The predicted size of the PCR product derived from the allele undergone deletion is 309 bp.

Proteinuria Measurement

Strips of Albustix (Bayer Corp., Mishawaka, IN) were used to measure the proteinuria of mice at approximately 6 p.m. on the day of measurement.

Histology

Mice were perfused with 4% paraformaldehyde followed by 18% sucrose PBS solution. The kidneys were cut into two halves, one imbedded with OCT and snap-frozen, another fixed with 10% formalin overnight for paraffin embedding. Four-micrometer sections were cut for both frozen and paraffin-embedded kidneys. Periodic acid-Schiff staining was performed following the protocol recommended by the Animal Models of Diabetic Complications Consortium (http://www.amdcc.org). Immunostaining was conducted briefly as follows: OCT-embedded frozen sections were cut (4 μm), air-dried, and incubated with blocking solution at room temperature for 30 min. Then the sections were incubated with the first antibody for 1 h at room temperature. After PBS wash for three times, second antibody was added and incubated at room temperature for 1 h. The sections were then washed, stained with DAPI, and sealed with Fluoromount-G (Southern Biotech).

Podocyte Number Counting

OCT-embedded frozen kidneys were used. Four-micrometer-thick sections were cut, and rabbit polyclonal WT1 antibody (C-19; Santa Cruz) and the second antibody goat anti-rabbit IgG (H+L) Alex Fluor 488 (Invitrogen, Carlsbad, CA) were used to stain podocytes, followed by DAPI staining. WT1 and DAPI double-positive cells were counted as podocytes. All WT1 and DAPI double-positive cells across the whole section were counted, added up, and then divided by the total number of glomeruli on the section, resulting in the average podocyte number per glomerular cross-section.

TUNEL Assay

The apoptotic cells were detected using ApopTag Red In Situ Apoptosis Detection Kit (Chemicon Int., Temecula, CA) on frozen kidney sections. For identification of apoptotic podocytes, the kidney sections were co-stained with WT1 antibody and DAPI. The triple-positive cells, WT1 (green), TUNEL (red), and DAPI (blue), were considered to be apoptotic podocytes. The ones positive for WT1 and DAPI but negative for TUNEL were nonapoptotic podocytes.

Electron Microscopy

A small piece of kidney from cortex was collected, sliced into 1-mm³ cubes, and fixed in 2% glutaraldehyde solution. The kidney samples were then processed following standard protocol.

Isolation of Glomeruli

The procedure for isolation of glomeruli was based on a published method with modifications.⁵⁸ Briefly, mice were perfused with 2.5 mg/ml iron oxide solution in PBS. Then the two kidneys from a mouse were diced into 1-mm³ pieces. A total of 100 μl of collagenase A solution (10 mg/ml) and 100 μl of DNase I (1000 U/ml) were added to the tissues, and the mix was incubated at 37°C for 30 min with rotation. After passing the disrupted tissues through a 100-μm cell culture strainer, several times of magnetic concentration of glomeruli were performed to obtain purified glomeruli.

Preparation of Total RNA and miRNA-Enriched Small RNA from Glomeruli

Glomerular total RNA and small RNA (200 bp) fraction with miRNA enriched were prepared using mirVana miRNA isolation kit (Ambion, Austin, TX) according to the manufacturer's instructions.

Microarray

Affymetrix Genechip Mouse Genome 430 2.0 Array containing approximately 39,000 gene probes was used for gene expression profiling of the glomeruli of mutant and control mice using total RNA from purified glomeruli. The whole procedure followed the manual instruction of the product. The Affymetrix Genechip data analysis was performed using R (R-Group) and Bioconductor⁵⁹ statistics packages in combination with other widely known GeneChip analysis tools. Briefly, for preprocessing of the chip data, the raw data were normalized across the chips by log scale robust multi-array analysis (GCRMA) method,⁶⁰ which has been shown to give more accurate results than the standard algorithm supplied by Affymetrix, and the probe sets with absent calls will be eliminated for downstream analysis. For genes with multiple probe sets on the chip, the median of the expression values of duplicated probe sets was calculated to represent the expression value for corresponding genes. For identification of differentially expressed genes between two groups, the statistical package Significance Analysis of Microarrays⁶¹ was used.

Quantitative Real-Time PCR

For mature miRNA quantitative real-time PCR, Ncode miRNA First-Strand cDNA Synthesis kit and Ncode SYBR Green miRNA quantitative real-time PCR kit (Invitrogen) were used following the manufacturer's instructions. For precursor miRNA quantitative real-time PCR, reverse transcriptions using pre-miRNA–specific primers (premiR-30a, 5′-GCAGCTGCAAACATCCGACTGAA-3′; premiR-30c-2, 5′-AAGGCAGAGAGAGTAAACAGCCTT-3′; and miR-30d, 5′-GAGCCAGTAGCAGCAAACATCT-3′) and SuperScript reverse transcriptase III (Invitrogen) were conducted, followed by quantitative PCR using these precursor-specific primers and the primers corresponding to mature miRNA sequences. For mRNA quantitative real-time PCR, SYBR Green PCR master mix (Applied Biosystems). The primers used were as follows: RAMP1-AU067747 (forward 5′-GGGATGAGAAGAGCATCCAGAAT-3′, reverse 5′-CGGGTACTGATGCGAGCTTTGTC-3′); Ccbe1(AV264768; forward 5′-GAC CGCGAGAGACACCAAAAGC-3′, reverse 5′-CCCATCGTCCTCAAGGATGTAG-3′); Hsp20 (forward 5′-CCCCAGTGTGGCGTTACCCACAG-3′, reverse 5′-GCGGTGGAACTCTCGAGCAAT-3′); Scel (forward 5′-AACAGGGTTTTCAGGACGTG-3′, reverse 5′-TGTTTGCATTGAAGGAGCTG-3′); Plod2 (forward 5′- GCCGTCTGGTCCAGCAGTGGAAT-3′, reverse 5′-GGCCTGGAAAATTTTGCATTTG-3′); Dusp14 (forward 5′-GGCTCCAG CTCCCTGGAA ATCCT-3′, reverse 5′-GGGAGCGTGCTGTGACCTCTGGA-3′); AI790298 (forward 5′- CGGCAAGCAGCTGCTGGTGATC-3′, reverse 5′-CCGGTGCTCAGTCACATCAAAGT-3′); Gadd45b (forward 5′-GGCGAGCGACAACGCGGTTCAG-3′, reverse 5′-CATCCTCCTCTTCTTCGTCTATGG-3′); Serpinb6b (forward 5′-CAAGAAATGCCTTTCAATGTCAC-3′, reverse 5′-ATTTCTTATAAGTTATCTCCTTTTCC-3′); Sbk1 (forward 5′-CCCTTCATCATCAAGGTCTTTGAC-3′, reverse 5′-CTGCCTGCTGTGCATGAAGTCCA-3′); Ager (forward 5′-CCCTCCTCAGGTCCACTGGATAA-3′, reverse 5′-TGTGACCCTGATGCTGACAGGA-3′); Acvr2b (forward 5′-CGGCCTGGCTGTTCGGTTTGAG-3′, reverse 5′-GCCCATGGCGTACATGTCGATAC-3′); Tagln (forward 5′-CCTTCCAGTCCACAAACGACCAA-3′, reverse 5′-GGCCACACTGCACTACAATCCA-3′); Nphs2 (forward 5′-CGGTGGAAGCTGAGGCACAAAGA-3′, reverse 5′-GCGACTGAAGAGTGTGCAAGTAT-3′). ABI real-time PCR machine 7900HT was used for all kinds of quantitative PCR described.

Identification of miRNA with Enriched Target Sequence in Upregulated Transcripts

To identify the potential miRNA candidates whose suppression by dicer knockout caused upregulation of their targeting genes, we used Sanger's miRNA target database (http://www.miRNA.org) to search for miRNA targeting 3′ UTR sequences of significantly upregulated genes and the occurrence for each miRNA was obtained. For determination of the significance of miRNA occurrence, the same number of genes as miRNA targets in upregulated genes were randomly chosen from the chip gene list and the occurrence of a miRNA was counted. Such permutation steps were repeated 10,000 times. The P value was calculated as the percentage of occurrences of a miRNA for 10,000 randomly generated gene lists greater than that for upregulated genes and then adjusted by false discovery rate.

DISCLOSURES

None.