Global polysome analysis of normal and injured podocytes

Abstract

Podocyte injury is a key event for progressive renal failure. We have previously established a mouse model of inducible podocyte injury (NEP25) that progressively develops glomerulosclerosis after immunotoxin injection. We performed polysome analysis of intact and injured podocytes utilizing the NEP25 and RiboTag transgenic mice, in which a hemagglutinin tag is attached to ribosomal protein L22 selectively in podocytes. Podocyte-specific polysomes were successfully obtained by immunoprecipitation with an antihemagglutinin antibody from glomerular homogenate and analyzed using a microarray. Compared with glomerular cells, 353 genes were highly expressed and enriched in podocytes; these included important podocyte genes and also heretofore uncharacterized genes, such as Dach1 and Foxd2. Podocyte injury by immunotoxin induced many genes to be upregulated, including inflammation-related genes despite no infiltration of inflammatory cells in the glomeruli. MafF and Egr-1, which structurally have the potential to antagonize MafB and WT1, respectively, were rapidly and markedly increased in injured podocytes before MafB and WT1 were decreased. We demonstrated that Maff and Egr1 knockdown increased the MafB targets Nphs2 and Ptpro and the WT1 targets Ptpro, Nxph3, and Sulf1, respectively. This indicates that upregulated MafF and Egr-1 may promote deterioration of podocytes by antagonizing MafB and WT1. Our systematic microarray study of the heretofore undescribed behavior of podocyte genes may open new insights into the understanding of podocyte pathophysiology.

INTRODUCTION

Accumulating evidence indicates that podocyte injury plays a key role in the progressive destruction of kidney architecture. Regardless of the nature of the primary injury, once a substantial number of podocytes are lost, the kidneys are irreversibly and progressively injured, leading to end-stage renal failure (38).

Previously, we established a mouse model of podocyte injury (NEP25), which expresses human (h) CD25 selectively on podocytes (25). Injection of the hCD25-targeted immunotoxin LMB2 induces podocyte injury dose-dependently through inhibition of protein synthesis in a similar manner to puromycin but strictly selective to podocytes. A minimum dose [0.625 ng/g body weight (BW)] of LMB2 causes NEP25 mice to develop moderate proteinuria, which peaks 1 wk after injection. Although injected LMB2 is rapidly cleared from the circulation (half-life of 35 min) (19), podocyte injury progresses over weeks and involves neighboring cells. The glomeruli show focal segmental glomerulosclerosis (FSGS) 3 wk after LMB2 injection. Using the chimeric mice made up with podocytes with and without hCD25, we have demonstrated that the initially damaged hCD25 podocytes lead to other non-hCD25 podocytes becoming damaged (24). These indicate that podocytes injured by LMB2 may damage other initially intact podocytes.

To investigate the molecular events in injured podocytes, we sought to analyze the RNA profiles of podocytes from NEP25 mice. To isolate podocyte mRNA, we utilized the RiboTag mice that express the ribosomal protein L22 (Rpl22) tagged with the hemagglutinin (HA) epitope only in cells expressing Cre recombinase (33). Polysomes specific to Cre-expressing cells are able to be obtained from tissue homogenate by immunoprecipitation with an anti-HA antibody without cellular dissociation. RiboTag mice have previously been used for cell-specific translatome analyses in various cell types (4, 9, 10, 12, 15, 32, 35, 36, 41).

In this study, we combined the RiboTag line with a podocyte-specific Cre-expressing line (Nphs1-Cre) (Fig. 1A), then combined with our NEP25 line. Utilizing these mice, we performed global translatome analyses in both normal and injured podocytes.

Fig. 1.

Strategy of podocyte polysome analysis. A: transgene structure of Nphs1-Ribotag mice. RiboTag mice (top) were crossed with Nphs1-Cre mice (middle). In podocytes of RiboTag^TG/TG Nphs1-Cre^+/WT (Nphs1-Ribotag) mice, Cre recombinase deletes the wild-type exon 4 (E4) of the Rpl22 gene and replaces with HA epitope-fused exon 4. B: HA immunostaining of Nphs1-RiboTag mice. HA was selectively expressed in the podocytes of Nphs1-RiboTag mice. Left and right panels show a low power field (scale bar = 100 μm) and a high-power field (scale bar = 25 μm), respectively. C: schematic outlining the method for podocyte RNA isolation. Glomeruli were isolated from Nphs1-RiboTag mice, homogenized, and immunoprecipitated with an anti-HA antibody to obtain podocyte specific polysomes, then purified to collect the podocyte specific mRNAs. D: RNA levels from the anti-HA antibody (Bound) fraction and the initial glomerular (Input) fraction. Although podocyte-specific RNAs, Nphs1 and Nphs2, were concentrated (left), a mesangial cell specific RNA, Des, and endothelial cell specific RNAs, Tek and Kdr, were diluted (right) in the bound fraction compared with the input fraction. Error bars represent SE (n = 8 for Nphs1 and Nphs2, n = 6 for Des, and n = 7 for Tek and Kdr). HA, hemagglutinin.

MATERIALS AND METHODS

Animal experiments.

The protocols for animal experiments were approved by the Animal Experimentation Committee of Tokai University School of Medicine, in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health.

RiboTag transgenic mice were obtained from Jackson Laboratory (no. 011029, Bar Harbor, ME) and mated with the Nphs1-Cre and NEP25 mice, obtaining mice carrying RiboTag^TG/TGNphs1-Cre^+/WT (designated as Nphs1-RiboTag) and mice carrying RiboTag^TG/TGNphs1-Cre^+/WTNEP25^TG/WT (designated as Nphs1-RiboTag-NEP25).

Nphs1-RiboTag-NEP25 mice (n = 4 for each time point, 12–50 wk of age) were injected with 0.625 ng/g BW of LMB2, and glomeruli were isolated 4 and 7 days later. The concentrations of albumin and creatinine in the urine were determined. Kidney sections were immunostained with following antibodies: Dach1 (10914-AP, Proteintech, Tokyo, Japan), Desmin (M0760, Dako, Tokyo, Japan), Egr-1 (4153, Cell Signaling Technology, Tokyo, Japan), HA (11867423001, Roche, Tokyo, Japan), MafF (12771-1-AP, Proteintech), Nephrin (GP-N2, Progen, Heidelberg, Germany), and WT1 (sc-192, Santa Cruz Biotechnology, TX). For MafF and desmin double staining, the desmin antibody was conjugated to Alexa594- anti-mouse IgG1 (Zenon Z 25007, Thermo Fisher Scientific, Yokohama, Japan).

Isolation of glomeruli.

Under anesthesia with pentobarbital sodium (60 μg/g BW ip) and buprenorphine (50 ng/g BW sc) kidneys were perfused through the abdominal aorta with 7 ml saline containing 5 U/ml heparin followed by 5 ml saline containing 35 μl of Dynabeads M450 Tosyl-activated (Thermo Fisher) and 100 μg/ml cycloheximide. Kidneys were dissected, minced, and incubated in Hanks’ balanced salt solution containing 1 mg/ml collagenase A (Roche), 250 U/ml DNase I (Worthington, Lakewood, NJ), and 100 μg/ml cycloheximide at 37°C shaking (100 revolutions/min) for 30 min. The digested kidneys were sieved through a 106-μm metal mesh. Glomeruli were collected from the flow through with a neodymium magnet after washing several times with PBS containing 100 μg/ml cycloheximide.

Immunoprecipitation of podocyte-specific polysomes.

Glomeruli were suspended in 200 μl of a homogenization buffer [50 mM Tris, pH 7.4, 100 mM KCl, 12 mM MgCl₂, 1% Nonidet P-40, 1 mM DTT, 200 U/ml RNasin (Promega, Tokyo, Japan), 1 mg/ml heparin, 100 μg/ml cycloheximide, 1% Protease Inhibitor Cocktail (Sigma-Aldrich, Tokyo, Japan)] and vortexed for 1 min. After centrifugation at 10,000 revolutions/min for 10 min at 4°C, the supernatant was separated and 2.5 μg of anti-HA.11 epitope tag antibody (16B12, Convance, Tokyo, Japan) was added, and the sample was rotated for 4 h at 4°C. Dynabeads Protein G (Thermo Fisher) was then added to the samples, and they were rotated overnight at 4°C. On the following day, the samples were placed on a neodymium magnet on ice, and the supernatant was separated. The remaining pellet was washed 3 times for 10 min in a high-salt buffer (50 mM Tris, pH 7.4, 300 mM KCl, 12 mM MgCl₂, 1% Nonidet P-40, 1 mM DTT, 100 μg/ml cycloheximide). QIAGEN RLT buffer was added to the remaining pellet. Total RNA was prepared using the RNeasy Micro kit (QIAGEN, Tokyo, Japan) from the pellet and the separated supernatant and quantified with a NanoDrop spectrophotometer (Thermo Fisher). The supernatant contained unprecipitated RNAs.

Gene expression array and quantitative reverse-transcription PCR assays.

RNA samples were labeled using Low Input Quick Amp Labeling Kit (Agilent Technologies, Hachioji, Japan), hybridized to SurePrint G3 Mouse GE 8x60K arrays (Agilent), and scanned. The microarray data are accessible from the Gene Expression Omnibus repository under the accession number GSE108629.

For Nphs1, Nphs2, Wt1, Des, Kdr, and Gapdh mRNAs, TaqMan primer probe sets (Thermo Fisher) were used in quantitative reverse-transcription PCR (qRT-PCR). Primers for other genes are shown in Table 1. Relative amounts of mRNAs were determined using the ΔΔCT method.

Table 1.

Primers for quantitative reverse-transcription PCR

Gene	Forward	Reverse
C1qtnf7	aaaggagagaagggggaaaa	gctggaccgacttctcctt
Cxcl1	gactccagccacactccaac	tgacagcgcagctcattg
Dach1	tctacaatgactgcaccaacg	tgagagttctctggggaagtg
Dpp4	aagctctgaccagcgattatct	cacatttgtgtggtcagtaagttg
Egr1	cctatgagcacctgaccaca	tcgtttggctgggataactc
Foxd1	gcagccgcttcccttacta	gttgagcgacaggttgtgac
Foxd2	accgcttttctgtttgagca	gactgtgaaaggcgcaaag
Mafb	gcaacggtagtgtggaggac	acctcgtccttggtgaagc
Maff	ggctgtggatcccttatctagc	atcagcgcttcatccgaca
Pecam1	agccagtagcatcatggtca	agcaggacaggtccaacaac
Podxl	atcctgtggggcagcatac	agaggtgcccaatggtttg
Ptpro	aggctagtctccatgaacgaag	tcctggggtgagttgtatcc
Relb	gtgacctctcttccctgtcact	tgtattcgtcgatgatttccaa
Synpo	tctggagaaggttgccagtg	gcgacaggtgtagcccattg
Tek	cctgaactgtgatgatgaggtg	aatgatggtctctcataaggcttc
Tie1	cggcatgcagtaccttagtg	ctctccgaccagcacatttc
Tnfrsf12a	ccgccggagagaaaagtt	actggatcagtgccacacc
Vwf	acgccatctccagattcaag	tggctcttctcttcccgata

Experiments using primary cultured podocytes.

Glomeruli were harvested using the bead method and cultured on a collagen I-coated dish for 7 days in DMEM/F12 containing 5% FBS supplemented with 0.5% insulin-transferrin-selenium A liquid supplement. Cellular outgrowths were detached and passaged after removing remaining glomeruli. Immunostaining with anti-podoplanin antibody (PMab-1, kindly gifted by Dr. Umetsu in Tohoku University, Japan) revealed that more than 95% of these cells were podocytes. The cultured podocytes were used for experiments within 14 days after the first passage.

Podocytes were transfected with siRNAs against Dach1, Foxd1, Foxd2, Maff, and Egr1 or control siRNA (Sigma-Aldrich) using Lipofectamine RNAiMAX Reagent (Thermo Fisher). Subsequently, qRT-PCR analyses, proliferation assays, or 5-ethynyl-2′-deoxyuridine (EdU) assays were performed. For the proliferation assay, 2,000 podocytes were seeded into wells of a 96-well plate. Cell number was evaluated using a Cell-Counting Kit-8 (Dojindo Laboratories, Kumamoto, Japan) 5 days after siRNA transfection. For the EdU assay, 1.5 × 10⁴ podocytes were seeded on coverglass. The day after siRNA transfection, EdU was added at 1 μM to the medium. After 22 h, the incorporated EdU was calculated by staining with the Click-iT EdU Imaging kit (Thermo Fisher).

For examining the efficiency of immunoprecipitation, podocytes were cultured from Nphs1-RiboTag mice (n = 7). RNA was extracted from the polysome of cultured cells by immunoprecipitation with an anti-HA antibody. The yield of polysome RNA was, on average, 13.3 ± 9.5 (SD) % of the total cellular RNA.

Statistical analysis.

For microarray analysis, qRT-PCR analyses in Tables 2 and 6, and gene suppression assays unpaired t-tests were used, and the P values were adjusted by the Benjamini-Hochberg procedure. Unpaired t-test was used in the analysis of Table 4. Mann-Whitney U-test was used for quantitative analyses of histology in Fig. 3, B and C. ANOVA and Dunnett’s test were used for urinary albumin/creatinine ratio (UACR) (logarithmically transformed), qRT-PCR analyses in Figs. 3A and 4A, quantitative analyses of histology in Fig. 3D, and cell counting assays.

Table 2.

Representative data of microarray and qRT-PCR analysis for mRNAs obtained by RiboTag method compared with those of glomerular cells

	Microarray			qRT-PCR
Gene Symbol	Fold Enrichment	Corrected P Value	Normalized Value of Array Signal in Podocytes	Fold Enrichment	Corrected P Value
Podocyte genes
Nphs1	5.00	0.00609	8.02	9.19	1.23E-5
Nphs2	1.34	0.0235	9.44	6.27	1.88E-5
Podxl	5.19	0.00322	7.98	5.00	0.00173
Synpo	2.91	0.00744	7.89	3.29	0.00268
Wt1	4.93	0.00304	8.85	7.93	1.61E-5
Nonpodocyte genes
Des	0.161	0.00262	−1.30	0.0940	5.35E-4
Tie1	0.00418	0.00400	−0.890	0.109	0.00359
Tek	0.0616	0.00681	−1.01	0.212	0.00386
Kdr	0.0791	0.00677	3.42	0.0443	0.0139
Vwf	0.133	0.00286	−4.24	<0.100
Pecam1	0.0621	0.00293	−1.41	0.0767	0.00707

qRT-PCR, quantitative reverse-transcription PCR.

Table 6.

Representative data of microarray and qRT-PCR analyses in podocytes injured by immunotoxin compared with normal podocytes

	Microarray				qRT-PCR
	4 Days After LMB2		7 Days After LMB2		4 Days After LMB2		7 Days After LMB2
Gene Symbol	Fold Change	Corrected P Value	Fold Change	Corrected P Value	Fold Change	Corrected P Value	Fold Change	Corrected P Value
Podocyte genes
Nphs1	0.675	0.0700	0.366	0.00271	0.383	2.21E-5	0.312	8.72E-6
Nphs2	0.869	0.284	0.874	0.402	0.706	0.0352	0.417	9.32E-4
Podxl	0.626	0.068	0.390	0.00170	0.575	0.0149	0.567	0.00955
Synpo	0.751	0.181	0.486	3.80E-4	0.611	0.0147	0.558	0.00695
Wt1	0.890	0.214	0.504	3.80E-4	0.562	9.36E-4	0.309	5.27E-5
Other genes
Des	1.10	0.660	8.09	3.54E-4	1.39	0.779	5.63	2.94E-4
Relb	6.23	0.0341	6.94	7.51E-4	27.2	3.57E-4	4.37	0.590
Tnfrsf12a	4.00	7.80E-4	11.3	1.15E-4	5.11	0.0317	9.18	7.56E-4
Cxcl1	199	0.00449	159	0.00163	247	4.40E-5	67.5	0.0593

qRT-PCR, quantitative reverse-transcription PCR.

Table 4.

RiboTag method into primary cultured podocytes

Gene	Fold Enrichment*
Highly podocyte-enriched genes
Dach1	3.45
Foxd1	2.46
Foxd2	3.45
Tcf21	2.63
Foxc2	3.45
Foxl1	2.78
Pdpn	2.65
Sema3g	2.30
(Average)	2.90 (P < 0.05 vs. housekeeping genes)
Housekeeping genes
Fau	1.11
Cox5a	1.67
Cox8a	1.29
Hspa8	1.02
Rpl8	0.99
Sod1	1.12
(Average)	1.20

^*Mean ratio of immunoprecipitated polysomes per whole podocyte RNAs.

Fig. 3.

Foxd1, Foxd2, and Dach1 in podocytes. A: quantitative reverse-transcription (qRT-PCR) analysis. Compared with normal glomeruli, normal podocytes highly express Foxd1, Foxd2, Dach1, Foxc1, and Foxc2 mRNAs. Foxd1, Foxd2, Dach1, and Foxc2 but not Foxc1 mRNAs were decreased in podocytes after injection with LMB2. Error bars represent SE (n = 4 each). B: immunostaining for Dach1. Dach1 protein is intensely stained in nuclei of podocytes of normal mice (a), coinciding with WT1 staining on the adjacent section (b). Dach1 staining diminishes in injured podocytes of NEP25 [8 days after injection of LMB2 (5 ng/g body weight) (c), similarly to that of WT1 on the adjacent section (d). Scale bars = 25 μm. Quantitative analysis demonstrated that Dach1-positive podocytes (e) and WT1-positive podocytes (f) decreased in the kidney of NEP25 mice compared with that of normal mice (WT). Each dot shows the number of Dach1-positive or WT1-positive podocytes per glomerulus (2,500–5,500 μm²) in total of 40 glomeruli from 2 mice. C: Dach1 immunostaining in various mouse models. Dach1 staining was diminished in injured glomeruli from mice with Adriamycin nephropathy (ADR) (a), human immunodeficiency virus-1 associated nephropathy (HIVAN) (c), and α-actinin4 nephropathy (Actn4) (e) compared with their controls (b, d, f). Dach1 staining was not diminished in diabetic Akita mice (DM) without nephropathy (g), compared with their control (h). Scale bars = 25 μm. Quantitative analysis confirmed these changes (i–l). Each dot shows the number of Dach1-positive podocytes per glomerulus in total of 40 glomeruli (2,500–5,500 μm²) from 2 mice (i, k, l) or 20 glomeruli from 1 mouse (j). D: DACH1 and WT1 immunostaining in human kidneys. Serial sections were taken from normal areas in a resected kidney with renal carcinoma (a and b) and biopsy samples with minimal change disease (c and d), diabetic nephropathy (e and f), lupus nephropathy (g and h), and immunoglobulin A (IgA) nephropathy (i and j). These were stained for DACH1 (a, c, e, g, and i) and WT1 (b, d, f, h, and j). DACH1 staining in the glomerulus coincided with WT1 staining in normal glomeruli. In glomeruli with diabetic nephropathy, lupus nephropathy, or IgA nephropathy, the number of DACH1-stained cells was markedly decreased compared with normal glomeruli. Scale bars = 50 μm. Quantitative analysis confirmed these changes (k and l). Each dot shows the number of Dach1-positive (k) or WT1-positive (l) podocytes per glomerulus (15,000–35,000 μm²) [n = 20 in normal section, n = 16 in minimal change disease (MCD), n = 20 in diabetic nephropathy (DN), n = 28 in lupus nephropathy (LN), and n = 23 in IgA nephropathy (IgAN)]. E: 5-ethynyl-2′-deoxyuridine (EdU) assays. Compared with podocytes treated with control siRNA, those treated with siRNA for Foxd2 or Dach1 but not Foxd1, showed less EdU incorporation. The incorporation rate of EdU-positive cells (green) per DAPI-positive cells (blue) is shown above each picture. Scale bars = 25 μm. F: gene knockdown assays. Suppression of Foxd1, Foxd2, or Dach1 by siRNAs decreased Dpp4 and Podxl mRNAs. Error bars represent SE (n = 3 each).

Fig. 4.

MafF and Egr-1 in podocytes. A: qRT-PCR analysis. Maff and Egr1 expression was very low in normal podocytes and markedly upregulated after LMB2 injection as early as 4 days after LMB2 injection. Mafb and Wt1 were highly expressed in normal podocytes, and these mRNAs were downregulated after LMB2 injection. Error bars represent SE (n = 4 each). B: immunostaining for MafF. Nuclear MafF staining was increased in the glomerulus of NEP25 mice after LMB2 injection (1.25 ng/g body weight) (top). Double immunostaining revealed that some of MafF-positive cells were also intensely stained for desmin (bottom, arrows), indicating that injured podocytes can express MafF protein. Scale bars = 25 μm. C: immunostaining for Egr-1. Egr-1 protein was not visible in normal glomeruli (top). In NEP25 mice injected with LMB2, Egr1 stained injured podocytes. Injured podocytes were identified by diminished nephrin staining (middle, arrow) or intense expression of desmin (bottom, arrow). Scale bars = 25 μm. D: EGR-1 staining in human kidneys. Serial sections of normal areas in a resected kidney with renal carcinoma (a and b) and kidney biopsy samples with minimal change disease (c and d), membranous glomerulonephropathy (e and f), IgA nephropathy (g and h), and diabetic nephropathy (i and j). These were stained for EGR-1 (a, c, e, g, and i), and nephrin (b, d, f, h, and j). EGR-1 was not observed in normal podocytes. EGR-1 staining in podocytes was positive in membranous glomerulonephropathy and IgA nephropathy but not in minimal change disease and diabetic nephropathy. EGR1 was expressed even in podocytes with normal nephrin staining. Scale bars = 50 μm. E: gene knockdown assays in murine primary podocytes. Maff siRNA suppressed Maff mRNA and upregulated Nphs2 and Ptpro mRNAs without a change in Mafb mRNA (left). Egr1 siRNA suppressed Egr1 mRNA and upregulated Ptpro, Sulf1, and Nxph3 mRNAs, with a slight change of Wt1 mRNA (right). Error bars represent SE (n = 5 each). qRT-PCR, quantitative reverse-transcription PCR.

RESULTS

Isolation of podocyte mRNA in Nphs1-RiboTag mice.

The HA-tag was selectively expressed in podocytes in the Nphs1-RiboTag mice, indicating that the Rpl22 transgene was recombined as expected (Fig. 1B). Nphs1-RiboTag and Nphs1-RiboTag-NEP25 mice without LMB2 showed a normal UACR of 0.085 mg/mg [95% confidence interval (CI) 0.076–0.096] and normal renal morphology.

Glomeruli were isolated from Nphs1-RiboTag mice, homogenized, and immunoprecipitated with an anti-HA antibody (Fig. 1C). Initial analysis by qRT-PCR showed that the podocyte-specific RNAs Nphs1 and Nphs2 were concentrated, whereas a mesangial cell-specific RNA Des and endothelial cell-specific RNAs Tek and Kdr were diluted, in the immunoprecipitated polysomes compared with glomerular RNAs. This indicates that podocyte mRNAs were successfully isolated (Fig. 1D).

Gene array analyses of normal and injured podocytes.

We immunoprecipitated podocyte-specific polysomes from Nphs1-RiboTag mice (n = 4) and compared them with the glomerular RNAs by microarray analysis.

Additionally, we injected LMB2 (0.625 ng/g BW) into Nphs1-RiboTag-NEP25 mice. At 4 and 7 days after LMB2 injection, glomeruli were harvested (n = 4 each) from which podocyte RNAs were obtained and analyzed by microarray. The UACR was increased to 0.19 mg/mg (95% CI 0.10–0.36) 4 days after injection and 76 mg/mg (95% CI 74–78) 7 days after injection (Fig. 2A).

Fig. 2.

Polysome analysis of normal and injured podocytes. A: urinary albumin/creatinine ratios. Without LMB2, all mice showed normal urinary albumin/creatinine ratios. Nphs1-RiboTag-NEP25 mice displayed albuminuria both 4 days and 7 days after LMB2 injection. B: principal component analysis. Normal glomeruli (Glo), normal podocytes (Podo), and podocytes from day 4 and day 7 after LMB2 injection (Podo day 4 and Podo day 7, respectively) (each n = 4) showed good clustering within groups in the first three components with small intersample variability. C: heat maps. Representative podocyte and nonpodocyte genes in normal glomeruli, normal podocytes, and podocytes 4 days and 7 days after LMB2 injection (each n = 4). D: logarithmic scatterplots of gene probe intensities of normal podocytes against those of glomerular cells. Distinct differences in translated RNA levels between normal podocytes and glomerular cells are demonstrated. Each dot represents one probe and selected probes of known podocyte genes and non-podocyte genes are highlighted. Greater than twofold up- and downregulated probes in podocytes are shown in red and green, respectively. E: logarithmic scatterplots of gene probe intensities of injured podocytes against those of normal podocytes. Distinct differences in translated RNA levels between normal podocytes and those 7 days after LMB2 injection are demonstrated. Each dot represents one probe and selected probes are highlighted. Greater than twofold up- and downregulated probes in injured podocytes are shown in red and green, respectively.

We performed principal component analysis on normal glomeruli, normal podocytes, and podocytes 4 and 7 days after LMB2 injection (each n = 4). These showed good clustering within groups in the first three components, with small intersample variability (Fig. 2B).

Normal podocytes have specifically enriched RNAs.

In normal podocytes compared with the glomerulus, many well-known podocyte-specific genes, including Nphs1, Podxl, Synpo, and Wt1, were highly expressed and significantly enriched (Table 2, Fig. 2, C and D). However, the mesangial cell-specific gene Des and endothelial cell-specific genes Tie1, Tek, Kdr, Vwf, and Pecam1 were diluted. These results were validated using qRT-PCR. Mircoarray analysis showed Nphs2 to be enriched only by 1.34-fold in normal podocytes, probably because of saturation of the array signal; however, analysis using qRT-PCR showed that Nphs2 was enriched by 6.3-fold in normal podocytes. The majority of podocyte-enriched genes reported in previous studies (2, 3, 13, 16, 37) were also concentrated in immunoprecipitated RNAs from Nphs1-RiboTag mice (Table 3).

Table 3.

Enriched genes in podocytes

Author	GEO ID	Method	Mouse	Comparison Subject	Number of Podocyte-Enriched Genes*	Number (%) of Genes Commonly Enriched in Ribotag Podocytes†
Takemoto et al. (37)	None	Flow cytometry	Podocin-Cre;Z/EG mice	Nonpodocyte glomerular cells	49	38 (77.6)
Brunskill et al. (3)	GSE17142, GSE17143, GSE17145	Flow cytometry	MafB-GFP mice	Renal cortex	138	132 (95.7)
Boerries et al. (2)	GSE39441	Flow cytometry	Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J:hNPHS2Cre mice	Nonpodocyte glomerular cells	864	735 (85.1)
Grgic et al. (13)	GSE53156	Immunoprecipitation with anti-eGFP antibody	Col1α1-eGFP-L10a (PodoTRAP) mice	Renal cortex	1,012	671 (66.3)

GEO, Gene Expression Omnibus; eGFP, enhanced GFP; GFP, enhanced green fluorescent protein.

^*Takemoto et al., Brunskill et al., Boerreis et al., and Grgic et al. defined >2, >5, >2, and >3-fold changed genes in podocytes compared with each comparison subject as podocyte-enriched genes, respectively.

^†Significantly (>2-fold) enriched in these genes in podocytes of Nphs1-RiboTag mice.

Given that the contribution of podocytes to the whole glomerular RNA content is one-third, the maximum rate of podocyte versus glomerular RNA is expected to be around three. However, 127 genes showed more than sixfold enrichment in podocyte polysomes. We tested whether mRNAs of these genes were concentrated in polysomes using primary cultured podocytes of Nphs1-RiboTag mice (n = 6). Relative amounts of mRNAs were determined for eight highly enriched podocyte genes and six housekeeping genes in immunoprecipitated polysomes and whole cell lysate. The results showed that podocyte mRNAs highly enriched within RiboTag glomeruli were significantly more concentrated in polysomes than those of the housekeeping genes (Table 4).

Podocytes had 353 genes significantly enriched (>3-fold) and highly expressed (>4 average normalized log2 signal) (Supplemental Table S1; Supplemental Material for this article is available online at the Journal website). Enrichment analyses of these highly expressed podocyte genes with Enrichr (http://amp.pharm.mssm.edu/Enrichr) (20) are summarized in Table 5 (whole data available in Supplemental Table S1). PodoNet genes and FSGS-related genes were significantly enriched, indicating that this gene set contains genes important for maintaining normal podocyte function.

Table 5.

Enrichment analyses in normal podocytes

Gene-Set Library	Term	P Value
WikiPathway 2016	PodNet: protein-protein interactions in the podocyte_Mus musculus_WP2310	5.8E-36
	XPodNet - protein-protein interactions in the podocyte expanded by STRING_Mus musculus_WP2309	3.5E-22
	Primary Focal Segmental Glomerulosclerosis FSGS_Homo sapiens_WP2572	1.5E-17
ChEA 2016	WT1_25993318_ChIP-Seq_PODOCYTE_Human	2.5E-43
	WT1_20215353_ChIP-ChIP_NEPHRON_PROGENITOR_Mouse	4.6E-19
MGI Mammalian Phenotype 2017	MP:0005325_abnormal_renal_glomerulus_morphology	2.6E-8
Human Phenotype Ontology	Proteinuria (HP:0000093)	6.1E-6
	FSGS (HP:0000097)	1.7E-5
	Glomerulosclerosis (HP:0000096)	1.6E-5

ChEA, ChIP Enrichment Analysis; FSGS, focal segmental glomerulosclerosis.

Gene array analyses of injured podocytes.

After induction of podocyte injury, the gene expression profile dramatically changed (Fig. 2, C and E). Four and 7 days after podocyte injury induction, 1,478 (2.7%) and 1,938 (3.5%) probes were significantly (>2-fold) downregulated, respectively. These genes included typical podocyte-specific genes such as Nphs1, Podxl, Synpo, and Wt1 (Table 6). Among the 353 highly podocyte-enriched genes (Supplemental Table S1), 146 genes (41.3%) were significantly (>2-fold) downregulated 7 days after induction of podocyte injury (Supplemental Table S2). On the other hand, 1,881 (3.4%) and 3,130 (5.6%) probes were significantly (>2-fold) upregulated 4 days and 7 days after induction of podocyte injury, respectively (Fig. 2E and Table 7). These included the well-established podocyte injury marker Des and inflammation-related genes such as Relb, Tnfrsf12a, and Cxcl1 (Table 6).

Table 7.

Top 20 upregulated genes in injured podocytes 4 days after immunotoxin injection

Gene Symbol	Description	Fold Change	Corrected P Value
Epha8	Mus musculus Eph receptor A8 (Epha8), mRNA (NM_007939)	6,450	0.00124
Thbs4	Mus musculus thrombospondin 4 (Thbs4), mRNA (NM_011582)	4,540	0.00513
Tmprss9	Mus musculus transmembrane protease, serine 9 (Tmprss9), mRNA (NM_001081688)	2,770	0.00563
Maff	Mus musculus v-maf musculoaponeurotic fibrosarcoma oncogene family, protein F (avian) (Maff), mRNA (NM_010755)	874	0.00464
P2rx7	Mus musculus purinergic receptor P2X, ligand-gated ion channel, 7 (P2rx7), transcript variant 1, mRNA (NM_011027)	435	0.00390
Cxcl1	Mus musculus chemokine (C-X-C motif) ligand 1 (Cxcl1), mRNA (NM_008176)	199	0.00449
Prr7	Mus musculus proline rich 7 (synaptic) (Prr7), mRNA (NM_001030296)	159	0.00196
Zmynd15	Mus musculus zinc finger, MYND-type containing 15 (Zmynd15), mRNA (NM_001029929)	131	0.00572
Birc5	Mus musculus baculoviral IAP repeat-containing 5 (Birc5), transcript variant 3, mRNA (NM_001012273)	83.5	0.0169
Egr1	Mus musculus early growth response 1 (Egr1), mRNA (NM_007913)	70.1	0.00493
Cirbp	Cold inducible RNA binding protein (Source:MGI Symbol;Acc:MGI:893588) (ENSMUST00000054666)	40.2	0.00229
Gadd45b	Mus musculus growth arrest and DNA-damage-inducible 45 beta (Gadd45b), mRNA (NM_008655)	29.7	0.00530
Ermn	Mus musculus ermin, ERM-like protein (Ermn), mRNA (NM_029972)	25.2	0.0252
Rhox8	Mus musculus reproductive homeobox 8 (Rhox8), mRNA (NM_001004193)	23.7	0.0222
Gab3	Mus musculus growth factor receptor bound protein 2-associated protein 3 (Gab3), mRNA (NM_181584)	23.3	0.0188
Gcm1	Mus musculus glial cells missing homolog 1 (Drosophila) (Gcm1), mRNA (NM_008103)	23.2	0.0221
Gm11974	Mus musculus predicted gene 11974 (Gm11974), long non-coding RNA (NR_045893)	22.0	0.00885
Ifrd1	Mus musculus interferon-related developmental regulator 1 (Ifrd1), mRNA (NM_013562)	21.1	0.00530
Dcc	Mus musculus deleted in colorectal carcinoma (Dcc), mRNA (NM_007831)	20.9	0.0188
Slc25a25	Mus musculus solute carrier family 25 (mitochondrial carrier, phosphate carrier), member 25 (Slc25a25), transcript variant 1, mRNA (NM_146118)	14.6	0.00563

Representative enrichment analyses from podocytes 7 days after immunotoxin injection of 1,344 downregulated genes (<0.5-fold change, P < 0.05) and 1,179 upregulated genes (>4-fold change, P < 0.05) are summarized in Table 8 (whole data available in Supplemental Tables S3 and S4).

Table 8.

Enrichment analyses in injured podocytes 7 days after immunotoxin injection

Gene-Set Library	Term	P Value
Downregulated genes
KEGG 2016	Protein processing in endoplasmic reticulum	2.3E-5
WikiPathway 2016	PodNet: protein-protein interactions in the podocyte_Mus musculus_WP2310	3.8E-5
Reactome 2016	MAP2K and MAPK activation Homo sapiens_R-HSA-5674135	0.012
ChEA 2016	WT1_25993318_ChIP-Seq_PODOCYTE_Human	1.4E-12
Upregulated genes
KEGG 2016	Cytokine-cytokine receptor interaction	2.5E-6
	TNF signaling pathway	6.0E-6
WikiPathways 2016	TWEAK Signaling Pathway Homo sapiens WP2036	1.3E-4
Reactome 2016	Extracellular matrix organization Homo sapiens R-SHR-1474244	6.8E-7
ChEA 2016	RELA 24523406 CHIP-Seq FIBROSARCOMA Human	2.6E-9

KEGG, Kyoto Encyclopedia of Genes and Genomes; ChEA, ChIP Enrichment Analysis.

The gene expression changes induced in podocytes by immunotoxin injection were remarkably similar to those observed in the podocytes of Actn4-knockout (KO) mice (13). Among the significantly (>2-fold) up- and downregulated genes in the Actn4-KO mice, 37.9% and 33.8% genes were similarly up- or downregulated in our study (>2 or <0.5-fold change 7 days after podocyte injury, P < 0.05) (Supplemental Table S5).

Functional analysis of selected genes in primary cultured podocytes.

The above set of highly expressed podocyte genes (Supplemental Table S1) contained many genes that are well characterized to have important roles in normal podocyte maintenance. We focused on Dach1 and Foxd2, which do not have an established function in podocytes. DACH1 polymorphism has been shown to be associated with chronic kidney disease (CKD) (18, 30) and may modulate or collaborate with Fox transcription factors (40). Foxd2 belongs to the Fox family, and Foxc1 and Foxc2 collaboratively maintain normal podocyte integrity (27). Our microarray and qRT-PCR data showed that Dach1 and Foxd2 mRNAs are highly expressed in podocytes similarly to Foxc1 and Foxc2 and are downregulated after podocyte injury (Fig. 3A). Foxd1 also showed a similar expression pattern, although the expression level was lower.

Immunostaining showed that Dach1 was intensely localized in the nucleus of podocytes, and this was diminished in the NEP25-injured podocytes in a parallel pattern to WT1 (Fig. 3B). Dach1 staining was also diminished in the glomerulus of other podocyte injury models but not in diabetic mice without nephropathy (Fig. 3C). Dach1 staining was also observed in normal human podocytes and was diminished in injured glomeruli in parallel with WT1 staining (Fig. 3D).

We next studied the function of Dach1, Foxd1, and Foxd2 in primary cultured podocytes. Compared with podocyte RNAs obtained by immunoprecipitation from RiboTag mice, Dach1, Foxd1, and Foxd2 mRNAs in primary podocytes cultured for 13 days were decreased by 0.025-, 0.21-, and 0.018-fold, respectively. Nevertheless, the suppression of Dach1 and Foxd2 but not Foxd1 by siRNA decreased cell proliferation (0.723, 0.741, and 0.953 relative ratio versus control siRNA, P = 7.04E-5, 5.30E-5, and 0.702, respectively) as well as the incorporation of EdU (Fig. 3E). We quantified the mRNAs of several Foxc-target genes (27, 37). Foxd1, Foxd2, or Dach1 siRNA decreased Dpp4 and Podxl mRNAs (Fig. 3F).

Among the genes upregulated after podocyte injury, we focused on two transcription factors, MafF and Egr-1. These may antagonize MafB and WT1, respectively, both of which are essential for maintaining normal podocyte functions. Microarray and confirmatory qRT-PCR data demonstrated that baseline levels of Maff and Egr1 mRNAs were very low in podocytes but that they increased markedly after LMB2 injection (Fig. 4A). Interestingly, Maff and Egr1 mRNAs were upregulated (15- and 4.2-fold, respectively) in podocytes freshly isolated from normal glomeruli by flow cytometry compared with podocyte RNAs obtained by immunoprecipitation from the RiboTag mice. This indicates that stress during cellular dissociation may stimulate expression of these genes. Maff and Egr1 mRNAs remained higher (9.0- and 5.2-fold) in primary cultured podocytes compared with RiboTag podocyte RNAs.

Immunostaining showed that nuclear MafF emerged in the glomerulus of NEP25 mice after LMB2 injection (Fig. 4B). Some of the MafF-positive cells were also intensely stained for desmin, indicating that injured podocytes express the MafF protein. Egr-1 was also shown to emerge in damaged podocytes in NEP25 mice, which were identified by diminished nephrin staining and enhanced desmin staining (Fig. 4C). Additionally, EGR-1 staining was negative in podocytes of normal human kidney but positive in membranous glomerulonephropathy and immunoglobulin A nephropathy but not in minimal change disease and diabetic nephropathy (Fig. 4D).

We next studied the effects of Maff and Egr1 siRNA on their respective MafB and WT1-target genes (7, 34) in primary podocytes. Maff knockdown increased Nphs2 and Ptpro mRNAs without altering Mafb mRNA. Egr1 knockdown increased Ptpro, Nxph3, and Sulf1 mRNAs with a slight change in Wt1 mRNA (Fig. 4E).

DISCUSSION

In this study, we obtained podocyte RNA by immunoprecipitation in glomerular homogenate utilizing RiboTag mice. Another method used to isolate podocyte RNA is flow cytometry. As injured cells are fragile and often lost during the dissociation process, it is difficult to apply the flow cytometry method to models with substantial podocyte injury. In addition, cellular dissociation and sorting can have significant effects on gene expression. In fact, our data as well as data from other laboratories have shown that Maff and Egr1 genes are highly increased in normal podocytes obtained by the flow cytometry method (2, 16). Therefore, as the RiboTag method does not need dissociation of glomerular cells, we found that it is more suitable for the purposes of our study.

Many genes in normal podocytes displayed a higher enrichment in podocyte polysomes than what would be expected from the podocyte population in the glomerulus. These genes also showed a higher enrichment in polysomes of primary cultured podocytes compared with housekeeping genes. This finding implies that mRNAs of these genes are concentrated in polysomes by posttranscriptional regulation and are preferentially translated. Shigeoka et al. (35) performed global translatome analysis of the retinal ganglion cell axon utilizing RiboTag mice. They demonstrated that mRNAs are concentrated in the polysome when and where necessary; therefore, it is conceivable that these highly enriched genes have functionally important roles for podocyte maintenance.

Grgic et al. (13) described the profile of podocyte-enriched mRNAs obtained from the Col1α1-eGFP-L10a (Podo^TRAP) mice, which displayed eGFP tagged ribosomes. We found that 66.3% of the podocyte-enriched genes in their study were also enriched in podocytes in our present study. This difference may be caused by different comparison subjects; their study used the renal cortex whereas we have used the glomeruli. Additionally, they analyzed the gene expression changes using an FSGS model (Actn4-KO mice). Interestingly, many genes were commonly up- or downregulated in Actn4-KO mice and NEP25 mice. Genes for the TNF signaling pathway are highly enriched in the commonly regulated genes. This indicates that common inflammation-like mechanisms could underlie the development of FSGS after podocyte damage. It should also be noted that the Col1α1 promoter of Podo^TRAP mice may not be always specific to podocytes. Pericytes and myofibroblasts also have Col1α1 promoter activity (14). In addition, the promoter activity may be changed in injured podocytes. However, in Nphs1-RiboTag mice the HA tag is added to Rpl22 irreversibly and specifically in podocytes, and the HA-Rpl22 expression is driven by the endogenous Rpl22 promoter.

Our polysome analysis identified highly podocyte-enriched genes, many of which were downregulated after LMB2 injection. These genes included many that are known to play important roles in podocytes. We speculated that genes with similar expression patterns may include some functionally important genes that were previously unrecognized. In this regard, we found that Foxd2 mRNA was highly expressed in normal podocytes and downregulated after LMB2 injection similarly to Foxc2, which is known to play an important role in maintaining normal podocyte integrity (27). No abnormal phenotype was reported for podocytes of Foxd2 knockout mice (21). Nevertheless, it remains conceivable that Foxd2 may collaborate with other Fox members. In this study, we found that suppression of Foxd2 by siRNA in primary cultured podocytes decreased DNA synthesis and proliferation. Interpretation of this phenomenon is difficult because in vivo, podocytes do not proliferate; however, it suggests that Foxd2 could maintain podocyte integrity. Foxd2 knockdown also decreased Podxl and Dpp4 mRNAs, which are Foxc2 targets, further supporting a potentially important role of Foxd2 in podocytes.

We were also interested in Dach1, because DACH1 has been reported to be associated with CKD (18, 30). Dach1 is a putative transcription factor and has been reported to be related to cell fate determination by collaborating with Pax2, Six1, Six2, Eya1, and Eya2 (1). Homozygous mutation in the Dach1 gene leads to postnatal lethality in mice (6). DACH1 protein is highly expressed in adult podocytes and is decreased in kidneys with immunoglobulin A nephropathy, idiopathic membranous nephropathy, or minimal change disease (22). We confirmed in this study that Dach1 mRNA and protein were abundant in normal podocytes but that they decreased after LMB2 injection. Endlich et al. (8) reported that knockdown of the zebrafish ortholog Dachd resulted in downregulation of nephrin and foot process formation in glomeruli. They also showed that the transfection of parietal epithelial cells with a plasmid encoding for Dach1 induced the expression of synaptopodin; however, decrease of synaptopodin mRNA was not observed in our knockdown assay of murine primary podocytes. Cancer research has shown that Dach1 restrains cell proliferation and migration by inhibiting Cyclin D1 expression (5, 23, 39). As podocytes do not proliferate even after LMB2 induced injury, Dach1 may have a completely different role in podocytes. In support of this notion, we found that Dach1 suppression in primary cultured podocytes decreased proliferation. Zhou et al. (40) reported that Dach1 may bind to the same site as Fox. In this study, we tested the effect of Dach1 knockdown on several Fox target genes in primary cultured podocytes and found that Dach1 siRNA decreased Dpp4 and Podxl mRNAs. This suggests that Dach1 may modulate or collaborate with Fox transcription factors in podocytes.

MafF and MafB belong to the same Maf protein family, have highly conserved DNA binding domains, and recognize the same DNA-binding motif (17); however, unlike MafB, MafF lacks the transcriptional activation domain. Likewise, Egr-1 and WT1 belong to the same Egr family and share an almost identical binding motif (26, 28). The transactivation domains are not conserved between Egr-1 and WT1. Both WT1 and MafB are intensely expressed in podocytes and are indispensable for maintaining the differentiation state of podocytes. Therefore, it is conceivable that the induction of MafF and Egr-1 in injured podocytes can actively antagonize MafB and WT1, respectively. In this study, we found that Egr1 and Maff mRNAs were already increased in murine primary cultured podocytes under basal conditions. Maff knockdown increased Nphs2 and Ptpro mRNAs, which are MafB targets, and Egr1 knockdown increased WT1 targets Ptpro, Nxph3, and Sulf1 mRNAs in primary podocytes (7, 34). In this regard, competition between other Maf protein members or between Egr-1 and WT1 has been reported (11, 29, 31).

NEP25 mice 4 days after LMB2 injection displayed only slightly decreased Wt1 levels with no decrease in Mafb, whereas Maff and Egr1 increased by 874- and 70.1-fold, respectively. This suggests that in the early phase of podocyte injury, MafB and WT1 signals are actively shut down, not by downregulation of the transcription factors themselves but by rapid induction of their counteracting factors. It is interesting to know whether such a mechanism would have a biological merit. The possibility exists that dedifferentiated podocytes may in some situations help the repair processes, such as cellular migration and hypertrophy. MafF and Egr-1 are well-known early response gene members. Thus, rapid activation of these transcription factors by stress in podocytes expands podocyte damage. Inhibition of MafF and Egr-1 may prevent this auto-dedifferentiation of injured podocytes; therefore, MafF and Egr-1 may be novel drug targets for preventing progressive podocyte deterioration.

In conclusion, the RiboTag method was a useful tool for isolating podocyte specific mRNAs, and we obtained podocyte-specific translatomes in normal and damaged conditions. These translatomes may reveal the mechanism of podocyte injury, leading to cessation of CKD progression.

GRANTS

This study was supported by a Grant-in-Aid for Scientific Research by the Japan Society for the Promotion of Science.

DISCLOSURES

I. Pastan is an inventor on several patents on immunotoxins that have all been assigned to the NIH. None of the other authors have any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

M.O. and T.M. conceived and designed research; M.O. performed experiments; M.O. and T.M. analyzed data; M.O. and T.M. interpreted results of experiments; M.O. and T.M. prepared figures; M.O. drafted manuscript; M.M., Y.M., I.P., T.Y., and T.M. edited and revised manuscript; T.M. approved final version of manuscript.