Wnt9a Promotes Renal Fibrosis by Accelerating Cellular Senescence in Tubular Epithelial Cells

BASIC RESEARCH www.jasn.org

Congwei Luo,1 Shan Zhou,1 Zhanmei Zhou,1 Yahong Liu,1 Li Yang,1 Jiafeng Liu,1 Yunfang Zhang,2 Hongyan Li,2 Youhua Liu ,1,3 Fan Fan Hou,1 and Lili Zhou1

1State Key Laboratory of Organ Failure Research, National Clinical Research Center of Kidney Disease, Division of Nephrology, Nanfang Hospital and 2Department of Nephrology, Huadu District People’s Hospital, Southern Medical University, Guangzhou, China; and 3Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

ABSTRACT Cellular senescence is associated with renal disease progression, and accelerated tubular cell senescence promotes the pathogenesis of renal fibrosis. However, the underlying mechanism is unknown. We assessed the potential role of Wnt9a in tubular cell senescence and renal fibrosis. Compared with tubular cells of normal subjects, tubular cells of humans with a variety of nephropathies and those of several mouse models of CKD expressed high levels of Wnt9a that colocalized with the senescence-related protein p16INK4A. Wnt9a expression level correlated with the extent of renal fibrosis, decline of eGFR, and expression of p16INK4A. Furthermore, ectopic expression of Wnt9a after ischemia-reperfusion injury (IRI) induced activation of b-catenin and exacerbated renal fibrosis. Overexpression of Wnt9a exacerbated tubular senescence, evidenced by increased detection of p16INK4A expression and senescence-associated b-galactosidase activity. Conversely, shRNA-mediated knockdown of Wnt9a repressed IRI-induced renal fibrosis in vivo and impeded the growth of senescent tubular epithelial cells in culture. Notably, Wnt9a- induced renal fibrosis was inhibited by shRNA-mediated silencing of p16INK4A in the IRI mouse model. In a human proximal tubular epithelial cell line and primary renal tubular cells, Wnt9a remarkably upregulated levels of senescence-related p16INK4A,p19ARF, p53, and p21 and decreased the phosphorylation of retinoblastoma protein. Wnt9a also induced senescent tubular cells to produce TGF-b1, which promoted proliferation and activation in normal rat kidney fibroblasts. Thus, Wnt9a drives tubular senescence and fibroblast activation. Furthermore, the Wnt9a–TGF-b pathway appears to create a reciprocal activation loop between senescent tubular cells and activated fibroblasts that promotes and accelerates the pathogenesis of renal fibrosis.

J Am Soc Nephrol 29: 1238–1256, 2018. doi: https://doi.org/10.1681/ASN.2017050574

CKD is highly prevalent worldwide, especially in the 1 elderly. Compared with younger counterparts, the Significance Statement elderly have higher rates of CKD, and mortality and CKD displays a number of features of accelerated senescence. Specifically, tubular cell senescence is Received May 25, 2017. Accepted January 3, 2018. strongly associated with progression of renal fibrosis. However, the underlying mechanisms of tubular se- Published online ahead of print. Publication date available at nescence are poorly understood. This study examines www.jasn.org. the potential role of Wnt9a in tubular cell senescence fi Correspondence: Dr. Lili Zhou or Dr. Fan Fan Hou, Division of and renal brosis in human biopsy material and in ex- Nephrology, Nanfang Hospital, Southern Medical University, perimental models. The authors found that Wnt9a 1838 North Guangzhou Avenue, Guangzhou 510515, China. knockdown in vivo with shRNA suppressed fibrosis, E-mail: [email protected] or [email protected] suggesting a decisive role for this pathway in tubular senescence and fibroblast activation. Copyright © 2018 by the American Society of Nephrology

1238 ISSN : 1046-6673/2904-1238 JAmSocNephrol29: 1238–1256, 2018 www.jasn.org BASIC RESEARCH renal function decline are worse in older than in younger pa- RESULTS tients.1,2 Interestingly, CKD-affected kidneys have many characteristics associated with natural aging, such as glomerular Wnt9a Increases in Multiple Types of Clinical sclerosis and interstitial fibrosis,3 tubular atrophy,4 local re- Nephropathy and Experimental CKD Models and nin-angiotensin-aldosterone system activation,5 loss of repair Associates with Tubular Senescence capabilities6 in renal tubular cells, and intracellular deposition Totest the clinical relevance of Wnt ligands in the pathogenesis of lipofuscin, a strong indicator of senescence.7,8 These find- of CKD, we conducted immunostaining in kidney biopsy ings suggest CKD is a clinical manifestation of premature ag- specimens from 35 patients with various nephropathies, ing or accelerated senescence.9 including minimal change disease, IgA nephropathy (IgAN), Cellular senescence, a stress-induced growth arrest, has re- diabetic nephropathy, membranous nephropathy, lupus ne- cently been found in multiple kidney diseases2,7,10,11 and tu- phritis, and others. Most of these diseases are highly associated bulointerstitial nephritis.11 The accumulation of senescent with accelerated cellular senescence.4,11,12 Compared with the cells has been detected in early clinical nephropathy, when absence of signal in normal human kidneys, the signal for patients have mild proteinuria and normal GFR.4,10 Accelera- Wnt9a strikingly increased in all biopsy specimens from pa- ted senescence may promote more pathologic alterations in tients with CKD (Figure 1A). Notably, Wnt9a was predomi- kidneys7,8,12 and even reduce the healthy lifespan.13 nantly localized in the renal tubular epithelium in human The key characteristics of cellular senescence are growth diseased kidneys (Figure 1A, arrows). Furthermore, immu- arrest and loss of DNA replication, breakdown of DNA double nostaining on sequential sections of human IgAN biopsy spec- strands,14 and accumulation of senescence-related proteins imens confirmed the localization of Wnt9a in tubular cells mainly through p16INK4A-retinoblastoma (Rb) and ARF- with Ki67-negative nuclei, a feature of senescent cells (Figure p53-p21 pathways.15,16 These pathways collectively halt cell 1B, yellow arrows).14 Because p16INK4A is also a typical senes- proliferation and accelerate cellular senescence. Senescent cence-related protein marker,13 we examined the colocaliza- cells can be identified by senescence-associated b-galactosidase tion of Wnt9a and p16INK4A expression. As shown in Figure (SA–b-gal) activity,4,13 and these cells produce components of 1C, the expression of Wnt9a nearly completely colocalized the senescence-associated secretory phenotype (SASP), includ- with p16INK4A-positive tubules. Furthermore, among 30 pa- ing proinflammatory cytokines such as IL-6 and matrix-syn- tients with sequentially sectioned biopsy specimens available thesizing molecules such as TGF-b1.17,18 for additional analysis, there was a positive correlation be- Cellular senescence can be caused by DNA damage, mito- tween Wnt9a and p16INK4A expression levels (Figure 1D). Ad- chondria dysfunction, inflammation, oxidative stress, and epi- ditionally, Wnt9a expression correlated with renal fibrosis and genetic alterations, each of which is a common characteristic decline of eGFR (Figure 1, E and F). The demographic and feature in CKD.4,13,19,20 Of note, among somatic cells of the clinical data of the patients with CKD are presented in Sup- kidney, tubular cells are the most likely to transition to the se- plemental Table 1. nescent phenotype.4,10,11,21 However, the underlying molecular To further evaluate the role of Wnt9a in tubular senescence mechanism of tubular senescence has not been elucidated. and progression of CKD, we assessed the expression of Wnt9a in Wnt/b-catenin signaling is an evolutionarily conserved different CKD models, including mice subjected to ischemia- pathway involved in organ development and tissue repair.6 reperfusion injury (IRI), unilateral ureteral obstruction (UUO), In normal adults, Wnt/b-catenin signaling is silent, but it is andadriamycin (ADR)nephropathy. AsshowninFigure 2, Aand reactivated after kidney injury in a wide range of CKD mod- B, Wnt9a expression increased as early as 1 day after severe IRI, els.5,22,23 Recent publications show the aberrant expression of although this increase was not yet significant. Three days after Wnt/b-catenin is highly associated with progression of renal severe IRI, the key transitional time point at which AKI prog- fibrosis5,23 and tubular atrophy.3 Such pathologic effects may resses to CKD through sustained activation of Wnt(s) signal- result from increased hyperactivity of local renin-angiotensin- ing,24 Wnt9a expression was significantly upregulated, as was the aldosterone system5 and repressed expression of Klotho, an expression of histone gH2AX. gH2AX is a selective marker of antiaging protein,22 caused by the increase in Wnt/b-catenin DNA double-strand breaks, and an increase in the number of signaling. These findings suggest the intimate correlation be- gH2AX foci is an important feature of senescence.25,26 We also tween Wnt activation and cellular senescence. However, whether assessed the expression of the antiaging protein Klotho, the loss Wnt ligands directly promote renal tubular cell senescence and of which derepresses Wnt signaling and serves as a marker of the progression of renal fibrosis remains unknown. tubular damage and CKD.22 As shown in Figure 2, A and C, as In this study, we examined the role of Wnt9a in CKD- Wnt9a expression increased after IRI, Klotho expression corre- associated renal fibrosis and tubular senescence in vivo and spondingly decreased. We then checked the expression of Wnt9a in vitro, which led to the identification of a Wnt9a-induced in a mouse model of UUO, inwhich p16INK4A is highly expressed reciprocal activation loop between senescent tubular cells and in tubules.27 Our immunostaining results demonstrated that activated interstitial fibroblasts. Our results suggest that Wnt9a Wnt9a expression was remarkably high in renal tubules of these promotes renal fibrosis through the induction of tubular senes- mice (Figure 2D, arrow). Furthermore, western blot analysis cence and crosstalk between tubular cells and interstitial fibroblasts. confirmed the upregulation of Wnt9a and gH2AX, as well as

J Am Soc Nephrol 29: 1238–1256, 2018 Wnt9a Accelerates Tubular Cell Senescence 1239 BASIC RESEARCH www.jasn.org

Figure 1. Wnt9a is induced specifically in renal tubular epithelium in human CKD. (A) Representative micrographs show the expression and localization of Wnt9a in various types of human CKD. Nontumor kidney tissue from the patients who had renal cell carcinoma and underwent nephrectomy was used as normal controls. Arrows indicate Wnt9a-positive tubules. Scale bar, 50 mm. (B) Representative micrographs show Wnt9a localized in those tubules with nuclei negative for expression of Ki67. Sequential paraffin-embedded kidney sections from patients with IgAN were immunostained for Wnt9a and Ki67. Yellow arrows indicate Wnt9a-positive tubules lack Ki67 nuclear expression. Scale bar, 50 mm. (C) Colocalization of Wnt9a and p16INK4A in tubules of patients with IgAN. Sequential paraffin-embedded kidney sections were immunostained for Wnt9a and p16INK4A. Colocalization of Wnt9a and p16INK4A in tubules is indicated by arrows. Scale bar, 50 mm. (D and E) Scatter plots with linear regression show significant correlation between Wnt9a expression levels and (D) p16INK4A expression levels and (E) extent of renal fibrosis. Expression of p16INK4A and fibrosis were quantitatively assessed by analysis of immunostaining and Masson trichrome staining, respectively, by three individuals who were blinded to Wnt9a data. The Spearman correlation coefficient (R) and P value are shown. (F) Linear regression shows an inverse correlation between Wnt9a expression level and eGFR. The Spearman correlation coefficient (R)andP value are shown. LN, lupus nephritis; MCD, minimal change disease; MN, membranous nephropathy.

1240 Journal of the American Society of Nephrology J Am Soc Nephrol 29: 1238–1256, 2018 www.jasn.org BASIC RESEARCH

Figure 2. Wnt9a is commonly induced and accompanied by increasing expression of gH2AX and loss of Klotho expression in various animal models of CKD. (A) Representative western blots showing the renal expression of Wnt9a, gH2AX, and Klotho in sham and injured kidneys 1 or 3 days after IRI. (B and C) Graphical representations of (B) Wnt9a, gH2AX, and (C) Klotho expression levels in three groups, as indicated. *P,0.05 versus sham controls (n=5–6). (D) Representative staining showing induction of Wnt9a protein expression in tubules in the UUO group. Scale bar, 50 mm. (E) Representative western blots showing the renal expression of Wnt9a, gH2AX, and Klotho in sham and obstructed kidneys 7 days after UUO. (F and G) Graphical representations of (F) Wnt9a, gH2AX, and (G) Klotho protein expression levels in two groups, as indicated. *P,0.05 versus sham controls (n=5–6). (H) Representative western blots showing renal Wnt9a, gH2AX, and Klotho expression in ADR nephropathy 3 weeks after ADR injection. CTL, control. (I and J) Graphical representations of (I) Wnt9a and gH2AX and (J) Klotho protein expression levels in two groups, as indicated. *P,0.05 versus normal controls (n=5–6). (K) Colocalization of Wnt9a and p16INK4A in ADR nephropathy tubules. Sequential parafﬁn-embedded kidney sections were immunostained for Wnt9a and p16INK4A. Colocalization of Wnt9a and p16INK4A intubulesisindicated by arrows. Scale bar, 50 mm.

the downregulation of Klotho, in UUO mice (Figure 2, E–G). In area, which suggests that these cells are subjected to unique the ADR nephropathy model, Wnt9a expression was high 3 disease-speciﬁc stresses, as previously reported.11 Wnt9a ex- weeks after ADR injection, and this effect was also accompanied pression did not correlate with tubular cell apoptosis in se- by increased expression of gH2AX and loss of Klotho (Figure 2, quential sections of ADR-injected kidney (Supplemental Fig- H–J). In sequential sections of kidneys from ADR mice, Wnt9a ure 1), consistent with the notion that senescent cells are expression largely colocalized with p16INK4A expression generally resistant to apoptosis.28 These results suggest the in- (Figure 2K, arrows). In these sections, we observed expres- timate correlation between upregulation of Wnt9a and tubular sion of p16INK4A in the nucleus as well as the tubular cytoplasmic senescence in kidney injury.

J Am Soc Nephrol 29: 1238–1256, 2018 Wnt9a Accelerates Tubular Cell Senescence 1241 BASIC RESEARCH www.jasn.org

Figure 3. Overexpression of Wnt9a promotes tubular injury and impairs renal function in IRI. (A) Experimental design. Green arrow indicates the injection of pcDNA3 or pFlag-Wnt9a plasmid. Red arrows indicate the timing of renal IRI surgery. (B) Representative western blots showing renal expression of Wnt9a in three groups, as indicated. (C) Graphical representation of Wnt9a protein expression levels in three groups. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (D) Representative micrographs show kidney morphology after IRI injury in different groups of mice, as indicated. Images of PAS staining are shown. Boxed areas are enlarged in the bottom panels. Arrows indicate injured tubules. Scale bar, 50 mm. (E) Quantitative analyses of injured tubules in three groups, as indicated. Kidney sections were subjected to PAS staining. At least 20 randomly selected ﬁelds were evaluated under 3400 magniﬁcation and results were averaged for each kidney. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (F) Urinary NAG level, expressed as international units per gram creatinine, in different groups. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (G and H) Ectopic expression of Wnt9a increased Scr and Ualb levels in IRI mice. Quan- titative analysis of (G) Scr and (H) Ualb levels in three groups, as indicated. Scr was expressed as milligrams per deciliter, and Ualb was expressed as micrograms per milligram urinary creatinine. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). Ucr, urinary creatinine; UNx, unilateral nephrectomy.

We also detected the expression of Wnt9a in a develop- association with the progression of aging-associated kidney mental senescence model, consisting of mouse embryonic ﬁbrosis in adults. day 14.5 (E 14.5) mesonephric tubules29 and kidneys from 2-, 12-, and 24-month-old mice (Supplemental Figure 2). In Vivo Expression of Wnt9a Augments Kidney Injury Although we detected p16INK4A expression in the E 14.5 me- after IRI sonephric tubules, we did not detect the expression of To establish the role of Wnt9a in renal ﬁbrosis, we delivered a Wnt9a. However, the expression of Wnt9a increased in Flag-tagged Wnt9a-expression vector (pFlag-Wnt9a) to mice

1242 Journal of the American Society of Nephrology J Am Soc Nephrol 29: 1238–1256, 2018 www.jasn.org BASIC RESEARCH

Figure 4. Ectopic expression of Wnt9a aggravates renal fibrosis in IRI mice. (A) Representative micrographs show that Wnt9a exacerbated renal interstitial fibrosis. Kidney sections were subjected to Masson trichrome staining. Arrow indicates positive staining. Scale bar, 50 mm. (B) Graphical representation of kidney fibrotic lesion area in different groups after quantitative determination. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (C) Representative western blots show renal expression of fibronectin and a-SMA in three groups, as indicated. (D and E) Graphical representations of (D) fibronectin and (E) a-SMA expression levels in different groups. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (F) Graphical representation of the relative abundance of collagen I mRNA in different groups, as indicated. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (G) Representative immunostaining micrographs show fibronectin, collagen I, and Fsp-1 expression in different groups. Boxed areas are enlarged and presented in the right column. Arrows indicate positive staining. Scale bar, 50 mm.

by hydrodynamic-based gene delivery, an approach that is rou- hydrodynamic-based gene transfer technique22 4daysafter tinely used in our laboratory for in vivo expression of a variety UIRI. Western blot analyses revealed that Wnt9a expression of genes.22,24 As shown in Supplemental Figure 3, Wnt9a was was induced at 7 days after a single injection of the Wnt9a signiﬁcantly induced in kidneys 24 hours after intravenous expression plasmid (11 days after UIRI) (Figure 3, B and C). injection of naked pFlag-Wnt9a plasmid. We then investigated We further detected the pathologic morphology by periodic the effect of exogenous Wnt9a in a mouse model of unilateral acid–Schiff (PAS) staining, a method that allows discernment IRI (UIRI). As shown in Figure 3A, pFlag-Wnt9a or empty of injured tubules through the detection of glycogen content. vector (pcDNA3) was administered intravenously by the As shown in Figure 3, D and E, in vivo expression of exogenous

J Am Soc Nephrol 29: 1238–1256, 2018 Wnt9a Accelerates Tubular Cell Senescence 1243 BASIC RESEARCH www.jasn.org

Figure 5. In vivo expression of exogenous Wnt9a promotes the expressions of b-catenin and targeted genes. (A) Immunohisto- chemical staining showing the renal induction of b-catenin and MMP-7 after injection of Wnt9a expression plasmid. Arrows indicate positive staining. (B and C) Graphical representations of the relative abundance of (B) MMP-7 and (C) PAI-1 mRNA in different groups, as indicated. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (D) Representative western blots show renal expression of b-catenin, MMP-7, PAI-1, and Snail 1 in three groups, as indicated. (E–H) Graphical representations show expression of (E) b-catenin, (F) MMP-7, (G) PAI-1, and (H) Snail 1 in three groups. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6).

Wnt9a significantly exacerbated the morphologic injury in- that the aberrant expression of Wnt9a accelerates pathologic duced by IRI alone. This injury was characterized by tubular kidney damage and functional decline. dilation, hyaline casts, and tubular atrophy with thickened basement membranes, as well as detached epithelial cells in In Vivo Expression of Wnt9a Aggravates Renal Fibrosis tubular lumens. We then examined urinary N-acetyl-b-D- and Activates b-Catenin Signaling after IRI glucosaminidase (NAG), a well known marker of tubular We further investigated the effects of exogenous Wnt9a on damage.30 As shown in Figure 3F, urinary NAG level was sig- matrix gene expression and renal fibrotic lesions after IRI. nificantly elevated by IRI injury but further increased by Analysis of Masson trichrome staining indicated that in vivo expression of exogenous Wnt9a. Consistent with this result, expression of exogenous Wnt9a aggravated renal fibrotic le- levels of serum creatinine (Scr) (Figure 3G) and urinary albu- sions (Figure 4, A and B) and further induced fibronectin min (Ualb) (Figure 3H) were further increased by in vivo ex- and a–smooth muscle actin (a-SMA) protein expression in pression of exogenous Wnt9a after IRI. These data indicate IRI mice (Figure 4, C–E). Consistently, exogenous Wnt9a

1244 Journal of the American Society of Nephrology J Am Soc Nephrol 29: 1238–1256, 2018 www.jasn.org BASIC RESEARCH

Figure 6. Overexpression of Wnt9a after IRI accelerates cellular senescence in renal tubules. (A) Expression of p16INK4A mRNA in different groups was assessed by real-time PCR. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (B) Rep- resentative western blots show that Klotho expression was further downregulated after injection of Wnt9a expression plasmid. (C) Graphical representations of Klotho protein expression levels in different groups, as indicated. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (D) Representative staining micrographs show p16INK4A and TGF-b1 expression and SA–b-gal activity in different groups. Paraffin sections were immunostained with antibodies against p16INK4A and TGF-b1. Frozen kidney sections were stained for SA–b-gal activity, which appears as bright-blue granular staining in the cytoplasm of tubular epithelial cells. Boxed areas are enlarged and presented in the right column. Arrows indicate positive staining. Scale bar, 50 mm. further upregulated collagen I mRNA levels in kidneys after results showed that in vivo expression of the Wnt9a gene in IRI (Figure 4F). Similar results were observed when kidney mice further aggravated the expression of b-catenin and sections were immunostained with antibodies against fibro- MMP-7, a transcriptional target of canonic Wnt/b-catenin nectin and collagen I (Figure 4G). Furthermore, as shown in signaling and a pathogenic mediator of kidney fibrosis.3,32 Figure 4G, exogenous Wnt9a strikingly increased the ex- Similarly, quantitative real-time PCR results showed exoge- pression of fibroblast-specificprotein–1(FSP-1),atarget nous Wnt9a expression further upregulated the expression gene of b-catenin that is also known as S100A4 protein of PAI-1, another transcriptional target of b-catenin (Figure and is a strong indicator of activated myofibroblasts in dis- 5C).33 Furthermore, as shown in Figure 5, D–G, western blot eased kidneys.5,23 These data indicate that activation of analyses of whole-kidney lysates revealed that expression of Wnt9a is a major driver of kidney fibrogenesis. exogenous Wnt9a promoted the expression of b-catenin, We next analyzed the expression of b-catenin signaling, MMP-7, and PAI-1 proteins. In addition, exogenous Wnt9a which is activated by Wnt ligand(s) and highly associated induced the expression of Snail 1, a key transcription factor of with the progression of renal fibrosis.23,31 As shown in Figure b-catenin signaling that is regarded as a “master gene” in the 5, A and B, immunostaining and quantitative real-time PCR control of renal fibrogenesis (Figure 5, D and H).34

J Am Soc Nephrol 29: 1238–1256, 2018 Wnt9a Accelerates Tubular Cell Senescence 1245 BASIC RESEARCH www.jasn.org

Figure 7. Knockdown of Wnt9a ameliorates renal injury after IRI. (A) Experimental design. Green arrow indicates the injection of pLVX-shRNA (Ctrl-shR) or pLVX-shWnt9a (Wnt9a-shR) plasmids. Red arrows indicate the timing of renal IRI surgery. (B) Repre- sentative western blots show expression of Wnt9a was nearly completely abolished after injection with Wnt9a-shR plasmid. (C) Graphical representation of Wnt9a protein expression levels in different groups, as indicated. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (D and E) Knockdown of Wnt9a significantly decreased Scr and secretion of Ualb. Quantification of (D) Scr and (E) Ualb levels in three groups, as indicated. Ualb level was expressed as micrograms per milligram urinary creatinine. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (F) Representative immunofluorescence micrographs show collagen I and fibronectin expression in three groups. Arrows indicate positive staining. Scale bar, 50 mm. (G) Representative western blots show fibronectin and a-SMA expression in three groups, as indicated. (H and I) Graphical representations of (H) fibronectin and (I) a-SMA protein expression levels in different groups, as indicated. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (J) Representative micrographs show collagen deposition in different groups, as indicated. Paraffin sections were used for Masson trichrome staining. Arrow indicates positive staining. Scale bar, 50 mm. (K) Graphical representation of the extent of fibrosis. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). Ucr, urinary creatinine; UNx, unilateral nephrectomy.

1246 Journal of the American Society of Nephrology J Am Soc Nephrol 29: 1238–1256, 2018 www.jasn.org BASIC RESEARCH

Figure 8. Knockdown of Wnt9a inhibits tubular cell senescence in IRI model. (A) Representative micrographs show staining for p16INK4A expression, SA–b-gal activity, and Klotho expression in different groups. Parafﬁn sections were immunostained with antibodies against p16INK4A and Klotho. Frozen kidney sections were stained for SA–b-gal activity. Arrows indicate positive staining. Scale bar, 50 mm. (B) Western blot analyses show p16INK4A,TGF-b1, and Klotho expression in three groups, as indicated. (C–E) Graphical representations of (C) p16INK4A, (D) TGF-b1, and (E) Klotho protein expression levels in different groups, as indicated. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6). (F) Representative micrographs show knockdown of Wnt9a reversed expression of E-cadherin after IRI, as indicated. Frozen sections were immunostained with antibody against E-cadherin. Scale bar, 50 mm. Ctrl-shR, injection of pLVXshRNA; Wnt9a-shR, injection of pLVX-shWnt9a.

In Vivo Expression of Wnt9a Accelerates Tubular findings strongly suggest that Wnt9a-induced tubular senes- Senescence after IRI cence could be a mediator of renal fibrosis (Figure 6D). To further clarify the mechanism by which Wnt9a promotes To further investigate the role of Wnt9a in tubular senes- renal fibrosis, we analyzed tubular senescence. As shown in cence, we assessed tubular senescence and renal fibrosis in Figure 6A, quantitative real-time PCR results indicated that 9-month-old mice subjected to exogenous Wnt9a expression IRI induced the renal expression of p16INK4A mRNA, and in without additional renal damage. As shown in Supplemental vivo expression of exogenous Wnt9a dramatically aggravated Figure 4, overexpression of Wnt9a in the absence of disease this effect. Moreover, in vivo expression of exogenous Wnt9a was sufficient to accelerate tubular senescence and renal fibro- further inhibited the expression of Klotho after IRI (Figure 6, sis. These data suggest the important role of Wnt9a in tubular B and C). We next assessed the protein expression levels of p16 senescence and development of renal fibrosis. INK4A and TGF-b1, which is secreted by tubular cells that have adopted the SASP.17,18 Immunostaining results indicated that Knockdown of Wnt9a Protects Renal Function and in vivo expression of exogenous Wnt9a significantly aggra- Kidney Fibrogenesis after IRI vated the IRI-induced increase in p16INK4A and TGF-b1ex- To further confirm the role of Wnt9a in kidney fibrogenesis, pression and SA–b-gal activity in tubules (Figure 6D). These mice were intravenously injected with an shRNA vector

J Am Soc Nephrol 29: 1238–1256, 2018 Wnt9a Accelerates Tubular Cell Senescence 1247 BASIC RESEARCH www.jasn.org

Figure 9. Knockdown of p16INK4A inhibits Wnt9a-induced renal fibrosis in IRI mice. (A) Experimental design. Red arrow indicates the injection of pcDNA3, pFlag-Wnt9a, or pFlag-Wnt9a with pLVX-shp16INK4A (p16-shR) plasmid. Black arrows indicate the timing of renal IRI surgery. (B) Representative western blots show expression of p16INK4A was largely abolished after injection with p16-shR plasmid. (C) Graphical representation of p16INK4A protein expression levels in different groups, as indicated. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6); #P,0.05 versus IRI injected with pFlag-Wnt9a (n=5–6). (D) Knockdown of p16INK4A pre- vented the increase in Scr levels induced by Wnt9a in IRI mice. Quantitative assessment of Scr level (mg/dl) in four groups, as indicated. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6); #P,0.05 versus IRI injected with pFlag-Wnt9a (n=5–6). (E) Representative micrographs show collagen deposition in different groups, as indicated. Paraffin sections were used for Masson trichrome staining. Arrow indicates positive staining. Scale bar, 50 mm. (F) Graphical representation of the extent of fibrosis after quantitative determination. (G) Representative western blots show fibronectin and a-SMA expression in four groups, as indicated. (H and I) Graphical representations of (H) fibronectin and (I) a-SMA protein expression levels in different groups, as indicated. *P,0.05 versus sham controls (n=5–6); †P,0.05 versus IRI alone (n=5–6); #P,0.05 versus IRI injected with pFlag-Wnt9a (n=5–6). (J) Representative immunostaining micrographs show fibronectin expression in different groups. Arrow indicates positive staining. Scale bar, 50 mm. UNx, unilateral nephrectomy.

1248 Journal of the American Society of Nephrology J Am Soc Nephrol 29: 1238–1256, 2018 www.jasn.org BASIC RESEARCH

Figure 10. Wnt9a induces cellular senescence in cultured proximal tubular epithelial cells. (A) Expression of p16INK4A mRNA in different groups was assessed by real-time PCR in HK-2 cells transfected with empty vector (pcDNA3) or Wnt9a expression plasmid (pFlag-Wnt9a) for 24 hours. *P,0.05 versus pcDNA3 group (n=3). (B) Representative micrographs of HK-2 cells immunostained for gH2AX (red) and counterstained with DAPI (blue) after transfection with pcDNA3 or pFlag-Wnt9a. Arrow indicates positive staining for gH2AX. Scale bar, 10 mm. (C) Representative western blots show Wnt9a transfection decreases the phosphorylation of Rb and promotes expression of p14ARF, p53, and p21. (D–G) Graphical representations of the levels of phosphorylated (D) Rb and (E) p14ARF,(F) p53, and (G) p21 protein expression in two groups, as indicated. *P,0.05 versus pcDNA3 group (n=3). (H) Representative western blots show incubation with recombinant Wnt9a for 24 hours dose-dependently promotes secretion of TGF-b1. (I) Graphical representations of TGF-b1 protein expression levels in four groups, as indicated. *P,0.05 versus medium alone (n=3). (J) Representative western blots show ICG-001inhibits Wnt9a-induced p53 and TGF-b1 expression. HK-2 cells were preincubated with ICG-001 (5 mM) or neutralizing antibody against type II receptor of TGF-b1(5mg/ml), then treated with Wnt9a (50 ng/ml) for 24 hours. (K and L) Graphical representations of (K) p53 and (L) TGF-b1 protein expression levels in four groups, as indicated. *P,0.05 versus medium alone (n=3); †P,0.05 versus Wnt9a alone (n=3). (M) Representative western blots show interference of p16INK4A blocks Wnt9a-induced increase in a-SMA and TGF-b1 expression. HK-2 cells were transfected with control (CTL) or p16INK4A-speciﬁc siRNA, then treated with Wnt9a (50 ng/ml) for 24 hours. (N–P) Graphical representations of (N) p16INK4A,(O)a-SMA, and (P) TGF-b1 protein expression levels in three groups, as indicated. *P,0.05 versus CTL siRNA alone (n=3); †P,0.05 versus Wnt9a alone (n=3). siRNA, small interfering RNA.

J Am Soc Nephrol 29: 1238–1256, 2018 Wnt9a Accelerates Tubular Cell Senescence 1249 BASIC RESEARCH www.jasn.org

Figure 11. Wnt9a induces cellular senescence in primary cultured tubular cells. (A) Representative micrographs show freshly isolated tubules and cultured primary tubular cells. Renal tubules were isolated from mouse kidneys and cultivated for use as primary tubular cells. (B) Representative micrographs show staining of SA–b-gal activity induced by Wnt9a. The primary cultured tubular cells were treated with Wnt9a (50 ng/ml) for 8 days before staining. Arrow indicates positive staining. (C) Representative micrographs of cells immunostained for gH2AX (red) and E-cadherin (green) and counterstained with DAPI (blue). Primary tubular cells were treated with Wnt9a or medium alone for 8 days. Arrows indicate positive staining. Scale bar, 10 mm. (D) Representative western blots show p16INK4A,p19ARF, phosphorylated Rb, and PCNA expression levels in two groups, as indicated. (E and F) Graphical representation of (E) p16INK4A and p19ARF, and (F) phosphorylated Rb and PCNA expression levels in two groups, as indicated. The primary cultured tubular cells were treated with Wnt9a (50 ng/ml) for 4 days before assessment. *P,0.05 versus medium alone (n=3). Ctrl, control.

encoding the interference sequence for Wnt9a (pLVX- Knockdown of Wnt9a Inhibits Tubular Senescence and shWnt9a) through a hydrodynamic-based gene delivery ap- Epithelial Injury after IRI proach (Figure 7A). As shown in Figure 7, B and C, renal We next investigated the relevance of Wnt9a knockdown in expression of Wnt9a was almost completely abolished in tubular senescence and cellular injury. As shown in Figure 8A, IRI mice by intravenous injection of Wnt9a shRNA plasmid. staining analyses revealed that knockdown of Wnt9a after IRI The IRI-induced increase in Scr and Ualb levels was also largely blocked the IRI-induced increase in p16INK4A expres- significantly blocked by knockdown of Wnt9a expression sion and SA–b-gal activity in tubules. Conversely, the tubular (Figure 7, D and E). Immunofluorescence revealed that expression of Klotho was remarkably preserved by knock- knockdown of Wnt9a also reduced the deposition of major down of Wnt9a after IRI (Figure 8A). As shown in Figure 8, interstitial matrix proteins, including fibronectin and colla- B–E, knockdown of Wnt9a significantly inhibited genI(Figure7F),andwesternblotanalysisshowedthat IRI-induced expression of p16INK4A and TGF-b1 and reversed knockdown of Wnt9a largely inhibited the expression of the expression of Klotho in kidneys. To further confirm the fibronectin and a-SMA in IRI kidneys (Figure 7, G–I). Con- role of Wnt9a in tubular damage, we examined the expression sistently, Masson trichrome staining revealed that knock- of E-cadherin, an epithelial marker that maintains normal down of Wnt9a attenuated the extent of renal fibrotic lesions epithelial integrity.35 As shown in Figure 8F, knockdown of in these mice (Figure 7, J and K). Wnt9a after IRI largely preserved the expression of E-cadherin

1250 Journal of the American Society of Nephrology J Am Soc Nephrol 29: 1238–1256, 2018 www.jasn.org BASIC RESEARCH

Figure 12. Wnt9a induces communication between senescent tubules and activated fibroblasts. (A) Representative western blots showing expression of PCNA and cyclin D1 after NRK-49F cells were treated with or without Wnt9a recombinant protein (50 ng/ml) for 24 hours. (B and C) Graphical representation of (B) PCNA and (C) cyclin D1 protein expression levels in two groups, as indicated. *P,0.05 versus medium alone (n=3). (D) Graphical representation shows that Wnt9a promoted NRK-49F cell proliferation as assessed by a colorimetric MTT assay. *P,0.05 versus medium alone (n=3). (E) Flow chart shows the experimental design and procedures. HK-2 cells were transfected with the Wnt9a expression plasmid for 24 hours to induce cellular senescence or with empty vector as controls. The supernatant was collected (Wnt9a-TCM or CTL-TCM) and used to stimulate NRK-49F cells. (F) Representative micrographs of NRK- 49F cells immunostained for fibronectin (red) and counterstained with DAPI (blue). NRK-49F cells were plated on coverslips and preincubated with neutralizing antibody against TGF-b1receptorII(5mg/ml) for 1 hour, then treated with Wnt9a-TCM or CTL-TCM for 24 hours before immunostaining. Scale bar, 20 mm. (G) Representative western blots show PCNA and a-SMA expression in three groups, as indicated. (H and I) Graphical representations of (H) PCNA and (I) a-SMA protein expression levels in different groups, as indicated. *P,0.05 versus CTL-TCM group (n=4); †P,0.05 versus Wnt9a-TCM alone (n=4). (J) Representative western blots show TGF- b1 promoted Wnt9a expression in cultured fibroblasts. NRK-49F cells were stimulated with TGF-b1 (5 ng/ml) for 24 hours. Whole-cell kidney lysates were analyzed for Wnt9a. (K) Graphical representation of Wnt9a levels in (J). *P,0.05 versus control (n=3). (L) Schematic presentation of reciprocal activation loop between senescent tubular cells and activated fibroblasts. In pathologic conditions, Wnt9a is dramatically upregulated. Wnt9a induces tubular cell senescence and promotes fibroblast activation. Tubular cells exhibiting the SASP secrete TGF-b1, which aggravates fibroblast proliferation and facilitates matrix protein production and deposition. The activated fibroblasts also produce Wnt9a, which further promotes tubular cell senescence. All of these lead to the pathogenesis of renal fibrotic foci. CTL-TCM, the supernatant collected from tubular cells with empty vector transfection; ECM, excellular matrix.

J Am Soc Nephrol 29: 1238–1256, 2018 Wnt9a Accelerates Tubular Cell Senescence 1251 BASIC RESEARCH www.jasn.org in tubules. These data further suggest that Wnt9a has an im- against TGF-b receptor II did not. Similarly, ICG-001 re- portant role in tubular senescence and cellular injury. pressed the expression of p21 (Supplemental Figure 6).15 These results suggest Wnt9a induces tubular senescence through activation of b-catenin but not TGF-b1. INK4A Knockdown of p16 Inhibits Wnt9a-Induced Renal We then examined the effect on fibrogenic responses of Fibrosis in IRI Mice siRNA-mediated inhibition of p16INK4A in the Wnt9a-treated To establish a causative link between Wnt9a-induced renal tubular cell line. As shown in Figure 10, M–P, the knockdown fi in brosis and senescence, we examined shRNA-mediated of p16INK4A expression significantly blocked the Wnt9a- vivo INK4A silencing of p16 in the model of UIRI mice injected induced expression of a-SMA and TGF-b1. These data further fi with pFLAG-Wnt9a (Figure 9A). The silencing ef cacy of re- confirm that tubular senescence has an important role in the INK4A fi nal p16 was con rmed by western blotting. Exogenous Wnt9a-induced fibrogenic response. Wnt9a-induced expression of p16INK4A in IRI mice was mostly abolished by intravenous injection of p16INK4A shRNA plasmid (Figure 9, B and C). As shown in Figure 9D, knock- Wnt9a Leads to Cell Senescence in Primary Cultured down of p16INK4A tended to prevent the increase in Scr in- Tubular Cells duced by IRI and aggravated by exogenous Wnt9a, but this In primary cultured mouse renal tubular cells (Figure 11A), finding was not statistically significant (P=0.07; n=5–6). incubation with recombinant Wnt9a triggered the upregula- Knockdown of p16INK4A also reduced the renal fibrosis tion of SA–b-gal activity (Figure 11B). Consistently, recombi- observed after IRI and Wnt9a overexpression (Figure 9, E nant Wnt9a induced the accumulation of nuclear gH2AX foci and F) and significantly attenuated Wnt9a-induced expres- and repressed the expression of E-cadherin (Figure 11C). As sion of fibronectin and a-SMA (Figure 9, G–J). These data shown in Figure 11, D–F, treatment with recombinant Wnt9a suggest that Wnt9a could induce tubular senescence, which significantly induced the expression of p16INK4A and p19ARF has a causative link to development of renal fibrosis. and inhibited the phosphorylation of Rb and the expression of proliferating cell nuclear antigen (PCNA). These data pro- vide striking evidence that Wnt9a induces renal tubular cell In Wnt9a Induces Cellular Senescence in Tubular Cells senescence. Vitro We further examined the cellular senescence in a cultured tubular epithelial cell line (HK-2). Quantitative real-time Wnt9a Induces Cell Communication between PCR results indicated that transfection of the Wnt9a expres- Senescent Tubular Cells and Activated Fibroblasts sion plasmid significantly induced the mRNA expression of We next examined the role of Wnt9a in cell-cell communica- p16INK4A (Figure 10A). Moreover, Wnt9a induced a signifi- tion between tubular cells and fibroblasts. As shown in Figure cant increase in gH2AX nuclear foci in these cells 12, A–D, treatment of a cultured fibroblast cell line (NRK-49F) (Figure 10B). We next examined other senescence-related pro- with recombinant Wnt9a induced protein expression of the teins p14ARF; p53; p21, a cell-cycle inhibitor and a target of the proliferation-related genes PCNA and cyclin D1 and dose- p53 gene that is often expressed by senescent cells; and hypo- dependently induced cell proliferation. phosphorylated Rb, which represses E2F-targeted DNA repli- To further confirm the role of Wnt9a in communication cation-related enzymes.36,37 As shown in Figure 10, C–G, between tubular cells and fibroblasts, we transfected HK-2 cells overexpression of Wnt9a significantly induced the expression with the Wnt9a expression plasmid and collected the super- of p14ARF, p53, and p21, and reduced the level of Rb phos- natant as conditioned medium (Wnt9a-TCM; Figure 12E). phorylation at S807/811. These results were also confirmed by Treatment of NRK-49F cells with Wnt9a-TCM induced the treatment with recombinant human Wnt9a protein (data not secretion of fibronectin, as detected by immunofluorescence. shown). As shown in Supplemental Figure 5, the addition of Similarly, Wnt9a-TCM enhanced the expression of PCNA, as recombinant Wnt9a to the standard growth media signifi- well as a-SMA, which suggests fibroblast-to- cantly inhibited the growth of HK-2 cells from passage 3 myofibroblast transdifferentiation (Figure 12, G–I).38 Of (P3) through P6. note, TGF-b receptor II blockade inhibited the Wnt9a- Administration of recombinant Wnt9a protein also signif- TCM–induced secretion of fibronectin and expression of icantly induced the expression of TGF-b1 in a dose-dependent PCNA and a-SMA (Figure 12, F–I). Interestingly, treatment manner (Figure 10, H and I). Tofurther assess the downstream of NRK-49F cells with TGF-b1 induced the expression of targets of Wnt9a, we pretreated HK-2 cells with ICG-001, a Wnt9a in these cells (Figure 12, J and K). These findings sug- small molecule that inhibits b-catenin–mediated gene tran- gest that Wnt9a has different functions in tubular cells and scription,5 or a neutralizing antibody against TGF-b receptor fibroblasts. Furthermore, the Wnt9a–TGF-b pathway may II. As shown in Figure 10, J–L, administration of ICG-001 regulate a reciprocal activation loop between senescent tubu- significantly inhibited the Wnt9a-induced expression of p53 lar cells and activated fibroblasts that accelerates the patho- and TGF-b1, but administration of the neutralizing antibody genesis of renal fibrosis (Figure 12L).

1252 Journal of the American Society of Nephrology J Am Soc Nephrol 29: 1238–1256, 2018 www.jasn.org BASIC RESEARCH

DISCUSSION deletion could be a plausible strategy for therapeutic interven- tion of cellular senescence and renal fibrosis. CKD exhibits characteristics of a premature senescence syn- Supporting our findings, it was previously reported that drome,9,11 including many features of aging9 such as increased Wnt/b-catenin signaling could accelerate senescence in bone oxidative stress, persistent inflammation, and loss and skin,42 and cellular senescence of bone marrow–mesen- of Klotho.5,22,39 Accelerated senescence has been observed in chymal stem cells and hematopoietic stem/progenitor cells.43 patients with advanced clinical impairment, and even in pro- It is noteworthy that Wnt9a has different functions in tu- teinuric patients with normal GFR,4 suggesting that cell bular cells and fibroblasts. Although we demonstrated that senescence is an early event in CKD. In multiple types of Wnt9a induces senescence in tubular cells, it has a proliferative CKD, an accelerated senescence phenotype has been observed role in fibroblasts, suggesting Wnt9a functions through differ- in defined cell types, mainly tubular cells.4,7 ential mechanisms in these cells to jointly reinforce the path- Cellular senescence is an irreversible fate of damaged cells ogenesis of renal fibrosis. The other interesting observation in that features growth arrest, DNA damage, and the enhanced this study is that Wnt9a evokes cell communication between expression of senescence-related proteins such as p16INK4A and senescent tubular cells and interstitial fibroblasts. Consistent p21. Emerging evidence suggests that accumulation of chronic with other studies,18 we observed that Wnt9a induced senes- senescent cells could promote pathologies and shorten healthy cent tubular cells to secrete the SASP component TGF-b1, lifespan.4,13 Senescent cells not only lose growth and repair which led to interstitial fibroblast proliferation and transition ability, but also secrete SASP components, including proin- to myofibroblasts. flammatory cytokines and growth factors, that attract Of note, TGF-b1 reinforces paracrine senescence in normal inflammatory cells and affect the neighboring cells. Thus, un- cells by regulating p15INK4B and p21,18 and has a potent mi- derstanding the mechanism that triggers cellular senescence is togenic role in fibroblasts and induces myofibroblast transi- of great importance. Manipulating senescence by clearance of tion.44 In our study, we found TGF-b1 also induces Wnt9a in p16INK4A-positive cells and calorie restriction can significantly fibroblasts, thereby creating a vicious cycle of tubular senes- reduce renal fibrosis and tubular atrophy and increase sur- cence, fibroblast activation, and renal fibrosis. The evidence vival.10,13,14 However, the underlying mechanism that pro- further suggests that signaling crosstalk between tubular epi- motes cellular senescence has not been clarified. The results thelial cells and interstitial fibroblasts is an important mech- of this study demonstrate that Wnt9a is involved in premature anism in accelerating progressive disease.45 senescence in tubular cells and in promoting renal fibrosis. In summary, we have shown that aberrantly expressed Wnt/b-catenin signaling is silent in normal adult kidneys Wnt9a is critically involved in mediating tubular senescence but reactivated in a wide variety of nephropathies.24,40,41 Of and renal fibrosis. Wnt9a also directly induces fibroblast pro- note, cellular senescence has been shown to have an important liferation and activation. Importantly, TGF-b1andWntsig- role in the pathologic features of these conditions.4,14 In this naling set up a vicious activation loop that promotes renal study, we showed that Wnt9a has a decisive role in driving fibrosis. Although more studies are needed, these results pro- tubular senescence and renal fibrosis. vide proof of principle that targeted inhibition of tubular Although barely detectable in normal adult kidneys, Wnt9a Wnt9a protects against fibrosis through breakdown of the re- expression is clearly induced, predominantly in tubular epi- ciprocal activation loop involving senescent tubular cells and thelial cells, in humans and mice after kidney injury and in activated interstitial fibroblasts. aging-associated kidney fibrosis. In addition, Wnt9a colocal- izes with p16INK4A-positive tubules. Here, we observed the expression of p16INK4A not only in nuclei but also in the tu- CONCISE METHODS bular cytoplasmic area. This is consistent with results from other groups, suggesting the dysregulation of the nuclear Animal Models transport mechanism or mislocalization owing to overexpres- Male C57BL/6 mice, weighing approximately 20–22 g, were sion.11 We also found the expression of Wnt9a positively cor- purchased from Southern Medical University Animal Center relates with p16INK4A expression, renal fibrosis, and decline of (Guangzhou, China) and housed in a standard environment on a reg- eGFR, and is accompanied by an increase of gH2AX foci and ular light/dark cycle with free access to water and chow. Renal IRI was loss of Klotho, a marker of tubular damage and CKD,22 sug- established as described.24 Briefly, bilateral renal pedicles were clipped gesting the intimate correlation between activation of Wnt9a for 32 minutes using microaneurysm clamps. During the ischemic and cellular senescence in kidney injury. Our additional in vivo period, body temperature was maintained between 37°C and 38°C and in vitro studies showed that renal expression of Wnt9a using a temperature-controlled heating system. After removal of the induces b-catenin signaling activation and promotes renal fi- clamps, reperfusion of the kidneys was visually confirmed. Mice were brosis and accelerates tubular senescence. Notably, these ef- euthanized 7 days after IRI, and kidney tissues were collected for var- fects require p16INK4A expression. These data undoubtedly ious analyses. show a causative role of Wnt9a in controlling tubular senes- The male C57BL/6 mouse model of UUO was established by dou- cence and renal fibrosis in vivo and suggest that Wnt9a gene ble-ligating the left ureter using 4–0 silk after a midline abdominal

J Am Soc Nephrol 29: 1238–1256, 2018 Wnt9a Accelerates Tubular Cell Senescence 1253 BASIC RESEARCH www.jasn.org incision. Sham-operated mice had their ureters exposed and manip- were grown in cell culture dishes for 4–8 days until they reached ulated but not ligated. Mice were euthanized 7 days after UUO. For 60%–80% confluency. Medium was changed on days 2 and 5, then the ADR nephropathy model, male BALB/c mice were administered a every 3 days. single intravenous injection of ADR (doxorubicin hydrochloride; Sigma, St. Louis, MO) at 10.5 mg/kg body weight. Mice were eutha- Western Blot Analysis nized 3 weeks after ADR injection. For the UIRI model, male C57BL/ Protein expression was analyzed by Western blot analysis as described 5 6 mice were subjected to unilateral renal IRI by an established pro- previously. The primary antibodies used were as follows: anti-p16 INK4A tocol, as described previously.24 Briefly, left renal pedicles were clip- (ab189034; Abcam), anti-p21 (sc-397; Santa Cruz Biotechnol- ped for 35 minutes using microaneurysm clamps for IRI injury. After ogy), anti-PCNA (sc-56; Santa Cruz Biotechnology), anti-cyclinD1 ARF removal of the clamps, reperfusion of the kidneys was visually con- (sc-753; Santa Cruz Biotechnology), anti-p14 (sc-8613; Santa – firmed. Ten days later, the intact right kidney was removed via aright Cruz Biotechnology), anti p-Rb (sc-16670; Santa Cruz Biotechnol- ARF flank incision. For studying the effects of Wnt9a, four sets of exper- ogy), anti-p19 (ab202225; Abcam), anti-p53 (sc-126; Santa Cruz – – iments were performed. The detailed experimental designs are pre- Biotechnology), anti Flag-tag (KM3002; Sungene Biotech Co.), anti – sented in Figures 3A, 7A, and 9A and Supplemental Figure 4A. In vivo TGF-b1 (sc-146; Santa Cruz Biotechnology), anti b-catenin fi expression or knockdown of Wnt9a in mice was carried out by a (610154; BD Biosciences), anti- bronectin (F3648; Sigma-Aldrich), – – hydrodynamic-based gene delivery approach, as described previ- anti a-SMA (ab5694; Abcam), anti MMP-7 (GTX104658; GeneTex, – ously.22 Briefly, the human Wnt9a and p16INK4A siRNA sequences Inc.), anti-snail1 (ab180714; Abcam), anti PAI-1 (AF1179; R&D Sys- (59-GCTTCAAGGAGACTGCCTT-39 and 59-CACCAGAGGCA- tems), anti-Klotho (AF1819; R&D Systems), anti-Wnt9a (ab125957; – – GUAACCAUTT-39, respectively) were ligated into an shRNA expres- Abcam), anti g-H2AX (ab26350; Abcam), anti a-tubulin (RM2007; – sion plasmid (pLVX-shRNA). Groups of mice were administered hu- Ray Antibody Biotech, Beijing, China), anti b-actin (RM001; man Wnt9a expression plasmid (pFlag-Wnt9a) or shRNA expression Ray Antibody Biotech), and anti-GAPDH (RM2002; Ray plasmid (pLVX-shWnt9a or pLVX-shp16INK4A) by rapid injection Antibody Biotech). of a large volume of DNA solution through the tail vein. Immunofluorescence Staining Kidneycryosectionswerefixedwith3.7% paraformalin for15minutes Determination of SCr, Ualb, and NAG Level at room temperature. Primary cultured mouse tubular cells, HK-2 Scr level was determined by an automatic chemistry analyzer (AU480; cells, or NRK-49F cells cultured on coverslips were fixed with cold Beckman Coulter, Pasadena, CA). Ualb was measured by using a methanol/acetone (1:1) for 15 minutes at room temperature, then mouse Albumin ELISA Quantitation kit, according to the manufac- blocked with 10% normal donkey serum in PBS. Slides were incubated turer’s protocol (Bethyl Laboratories, Inc., Montgomery, TX). Urine with antibodies against fibronectin (F3648; Sigma-Aldrich), E- creatinine was determined by a routine procedure, as described pre- cadherin (3195s; Cell Signaling Technology), collagen I (BA0325; viously.22 Ualb was standardized to urine creatinine and expressed as Boster Biotechnology), or g-H2AX (ab26350; Abcam). After washing, mg/mg urinary creatinine. Urinary NAG level was analyzed by a com- the slides were incubated with Cy3- or Cy2-conjugated donkey anti- mercial kit (CSB-E07444m; CUSABIO Life Science, Wuhan, China) mouse or donkey anti-rabbit IgG (Jackson Immuno-Research and expressed as IU/g urinary creatinine. Laboratories,WestGrove,PA).NucleiwerestainedwithDAPI (Sigma-Aldrich) according to the manufacturer’s instructions. Images Cell Culture and Treatment were captured by fluorescence microscopy (Leica DMi8; Leica Micro- Human proximal tubular epithelial cells (HK-2) and normal rat kid- systems, Buffalo Grove, IL). ney interstitial fibroblast cell lines (NRK-49F) were obtained from the American Type Culture Collection (Manassas, VA) and maintained RT and Real-Time PCR per routine protocol. HK-2 cells or NRK-49F cells were treated with Total RNA isolation was carried out using the TRIzol RNA isolation human recombinant Wnt9a (R & D Systems, Minneapolis, MN) at system (Life Technologies, Grand Island, NY) according to the man- the indicated concentration or transfected with Wnt9a expression ufacturer’s instructions. The first strand of complementary DNA was plasmid (pFlag-Wnt9a). Whole-cell lysates were prepared and sub- synthesized using 1 mgofRNAin20ml of reaction buffer containing jected to Western blot analyses. Some cells were also detected by AMV-RTand random primers at 42°C for 60 minutes. Real-time PCR immunofluorescence. was performed using the Platinum SYBR Green qPCR SuperMix- Primary mouse tubular cells were isolated and cultivated as pre- UDG kit (Invitrogen). The sequences of the primer pairs are shown viously described.46 Briefly, the kidneys were peeled off and minced, in Supplemental Table 2. then digested in 0.75 mg/ml collagenase for 40 minutes at 37°C, after which the mashed tissue was sieved in PBS. The tubular tissues were Histology and Immunohistochemistry isolated using 31% Percoll gradients, resuspended, and washed twice Paraffin-embedded mouse kidney sections (4-mmthickness)were with PBS. Finally, tubules were suspended in DMEM supplemented prepared by a routine procedure. PAS and Masson trichrome staining with 10% bovine calf serum, 50 U/ml penicillin, 50 mg/ml strepto- were conducted by standard protocol. Immunohistochemical stain- mycin, and hormone mix (5 mg/ml insulin, 1.25 ng/ml prostaglandin ing was performed using routine protocol. Antibodies used were as PG E1, 34 pg/ml triiodothyronine, 5 mg/ml transferrin, 1.73 ng/ml follows: anti-Wnt9a (ab189010; Abcam), anti-Ki67 (ab16667; Abcam), sodium selenite, 18 ng/ml hydrocortisone, and 25 ng/ml EGF). Cells anti-p16INK4A (ab189034; Abcam), anti-fibronectin (F3648; Sigma-

1254 Journal of the American Society of Nephrology J Am Soc Nephrol 29: 1238–1256, 2018 www.jasn.org BASIC RESEARCH

Aldrich), anti–FSP-1 (S100A4; Merk MILLIPORE), anti–b-catenin Youths Development Scheme of Nanfang Hospital, Southern Med- (ab15180; Abcam), anti–MMP-7 (3801s; Cell Signaling Technology), ical University (JQ201401). anti–TGF-b1 (sc-146; Santa Cruz Biotechnology), and anti-Klotho (AF1819; R&D Systems). Human biopsy sequential sections were obtained from diagnostic renal biopsies performed at the Nanfang Hospital DISCLOSURES and Huadu District People’s Hospital, and stained with rabbit polyclonal None. anti-Wnt9a (Abcam) and mouse monoclonal anti-p16INK4A (Abcam). All studies involving human kidney sections were approved by the In- stitutional Ethics Committee at the Nanfang Hospital and Huadu Dis- REFERENCES trict People’s Hospital, Southern Medical University. 1. Minutolo R, Borrelli S, De Nicola L: CKD in the elderly: Kidney senes- SA–b-gal and TUNEL Staining cence or blood pressure-related nephropathy? Am J Kidney Dis 66: Frozen sections (3 mm) were used for detection of SA–b-gal activity 184–186, 2015 according to the manufacturer’s instructions (9860; Cell Signaling 2. Clements ME, Chaber CJ, Ledbetter SR, Zuk A: Increased cellular se- Technology). Paraffin sections (4 mm) were fixed and stained for nescence and vascular rarefaction exacerbate the progression of kidney fibrosis in aged mice following transient ischemic injury. PLoS One TUNEL staining by a commercial kit (KGA703; KeyGEN BioTECH, 8: e70464, 2013 Nanjing, China). 3. Zhou D, Tian Y, Sun L, Zhou L, Xiao L, Tan RJ, et al.: Matrix Metal- loproteinase-7 is a urinary biomarker and pathogenic mediator of kid- Renal Immunohistochemical Analyses ney fibrosis. JAmSocNephrol28: 598–611, 2017 Semiquantitation ofWnt9a and p16INK4A staining was performed by the 4. Verzola D, Gandolfo MT, Gaetani G, Ferraris A, Mangerini R, Ferrario F, method previously reported.47 Briefly, digital images at 244831920- et al.: Accelerated senescence in the kidneys of patients with type 2 di- – pixel resolution were captured at 3400 magnification by the DP 27 abetic nephropathy. Am J Physiol Renal Physiol 295.: F1563 F1573, 2008 5. Zhou L, Li Y, Hao S, Zhou D, Tan RJ, Nie J, et al.: Multiple genes of the CCD camera (Olympus, Japan) coupled to an Olympus BX-51 mi- fi renin-angiotensin system are novel targets of Wnt/b-catenin signaling. croscope (Olympus). Images of randomly selected elds were taken J Am Soc Nephrol 26: 107–120, 2015 INK4A and the staining scores of Wnt9a and p16 were calculated by the 6. Zhou D, Tan RJ, Fu H, Liu Y: Wnt/b-catenin signaling in kidney injury and Image-Pro Plus software (version 6.0; Media Cybernetics, Silver repair: A double-edged sword. Lab Invest 96: 156–167, 2016 Spring, MD) and expressed as the ratio of integrated optical density 7. Kakoki M, Kizer CM, Yi X, Takahashi N, Kim HS, Bagnell CR, et al.: Se- (IOD) to observed area (IOD/area), representing the relative expres- nescence-associated phenotypes in Akita diabetic mice are enhanced – sion levels of Wnt9a and p16INK4A. by absence of bradykinin B2 receptors. JClinInvest116: 1302 1309, 2006 8. Bobkova NV, Evgen’ev M, Garbuz DG, Kulikov AM, Morozov A, Semiquantitative Scoring of Renal Fibrosis Samokhin A, et al.: Exogenous Hsp70 delays senescence and improves Masson trichrome staining was conducted by routine procedures in cognitive function in aging mice. Proc Natl Acad Sci U S A 112: 16006– kidney biopsy specimens from patients with CKD with different cau- 16011, 2015 sesetiologies. Stained slides were observed under a microscope by the 9. Stenvinkel P, Larsson TE: Chronic kidney disease: A clinical model of DP 27 CCD camera (Olympus, Japan), and images at 244831920- premature aging. Am J Kidney Dis 62: 339–351, 2013 pixel resolution captured at high-powered (3400) fields from ran- 10. Zhu K, Kakehi T, Matsumoto M, Iwata K, Ibi M, Ohshima Y, et al.: domly selected fields were taken. Each image per section was split into NADPH oxidase NOX1 is involved in activation of protein kinase C and premature senescence in early stage diabetic kidney. Free Radic Biol 100 squares. Tissue fibrosis as defined by blue staining was scored by Med 83: 21–30, 2015 three experienced observers in a blinded fashion. The percentage of 11. Sis B, Tasanarong A, Khoshjou F, Dadras F, Solez K, Halloran PF: blue staining in each image was calculated, and the mean values of the Accelerated expression of senescence associated cell cycle inhibitor fibrosis scores determined by the three observers were reported. p16INK4A in kidneys with glomerular disease. Kidney Int 71: 218–226, 2007 12. Liu J, Yang JR, He YN, Cai GY, Zhang JG, Lin LR, et al.: Accelerated Statistical Analyses senescence of renal tubular epithelial cells is associated with disease All data examined were expressed as mean 6SEM. Statistical analyseis progression of patients with immunoglobulin A (IgA) nephropathy. – of the data were carried out using SPSS 13.0 (SPSS Inc., Chicago, IL). Transl Res 159: 454 463, 2012 13. Baker DJ, Childs BG, Durik M, Wijers ME, Sieben CJ, Zhong J, et al.: Comparison between groups was made using one-way ANOVA fol- Naturally occurring p16(Ink4a)-positive cells shorten healthy lifespan. – – ’ lowed by Student Newman Keuls test or Dunnett s T3 procedure. Nature 530: 184–189, 2016 P,0.05 was considered significant. 14. Braun H, Schmidt BM, Raiss M, Baisantry A, Mircea-Constantin D, Wang S, et al.: Cellular senescence limits regenerative capacity and allograft survival. JAmSocNephrol23: 1467–1473, 2012 ACKNOWLEDGMENTS 15. Sturmlechner I, Durik M, Sieben CJ, Baker DJ, van Deursen JM: Cellular senescence in renal ageing and disease. Nat Rev Nephrol 13: 77–89, 2017 16. Campisi J, d’Adda di Fagagna F: Cellular senescence: When bad things This work was supported by the National Natural Science Foun- happen to good cells. Nat Rev Mol Cell Biol 8: 729–740, 2007 daion of China grants 81722011, 81570620, 81521003, and 17. Melk A, Schmidt BM, Takeuchi O, Sawitzki B, Rayner DC, Halloran PF: 81370014; Guangdong Science Foundation grant 2014A030312014; Expression of p16INK4a and other cell cycle regulator and senescence Guangzhou Projects grant 201504010001; and the Outstanding associated genes in aging human kidney. Kidney Int 65: 510–520, 2004

J Am Soc Nephrol 29: 1238–1256, 2018 Wnt9a Accelerates Tubular Cell Senescence 1255 BASIC RESEARCH www.jasn.org

18. Acosta JC, Banito A, Wuestefeld T, Georgilis A, Janich P, Morton JP, 34. Simon-Tillaux N, Hertig A: Snail and kidney fibrosis. Nephrol Dial et al.: A complex secretory program orchestrated by the inflammasome Transplant 32: 224–233, 2017 controls paracrine senescence. Nat Cell Biol 15: 978–990, 2013 35. Liu Y: Epithelial to mesenchymal transition in renal fibrogenesis: Path- 19. Mo H, Wu Q, Miao J, Luo C, Hong X, Wang Y, et al.: C-X-C chemokine ologic significance, molecular mechanism, and therapeutic in- receptor type 4 plays a crucial role in mediating oxidative stress- tervention. JAmSocNephrol15: 1–12, 2004 induced podocyte injury. Antioxid Redox Signal 27: 345–362, 2017 36. Zhao L, Zhang Y, Gao Y, Geng P, Lu Y, Liu X, et al.: JMJD3 promotes 20. Li SY, Susztak K: The long noncoding RNA Tug1 connects metabolic changes SAHF formation in senescent WI38 cells by triggering an interplay with kidney disease in podocytes. JClinInvest126: 4072–4075, 2016 between demethylation and phosphorylation of RB protein. Cell Death 21. Satriano J, Mansoury H, Deng A, Sharma K, Vallon V, Blantz RC, et al.: Differ 22: 1630–1640, 2015 Transition of kidney tubule cells to a senescent phenotype in early ex- 37. Ding G, Franki N, Kapasi AA, Reddy K, Gibbons N, Singhal PC: Tubular perimental diabetes. Am J Physiol Cell Physiol 299: C374–C380, 2010 cell senescence and expression of TGF-beta1 and p21(WAF1/CIP1) in 22. Zhou L, Li Y, Zhou D, Tan RJ, Liu Y: Loss of Klotho contributes to kidney tubulointerstitial fibrosis of aging rats. Exp Mol Pathol 70: 43–53, 2001 injury by derepression of Wnt/b-catenin signaling. J Am Soc Nephrol 38. LeBleu VS, Taduri G, O’Connell J, Teng Y, Cooke VG, Woda C, et al.: 24: 771–785, 2013 Origin and function of myofibroblasts in kidney fibrosis. Nat Med 19: 23. He W, Dai C, Li Y, Zeng G, Monga SP, Liu Y: Wnt/beta-catenin signaling 1047–1053, 2013 promotes renal interstitial fibrosis. J Am Soc Nephrol 20: 765–776, 2009 39. Zhou LL, Cao W, Xie C, Tian J, Zhou Z, Zhou Q, et al.: The receptor of 24. Xiao L, Zhou D, Tan RJ, Fu H, Zhou L, Hou FF, et al.: Sustained activation advanced glycation end products plays a central role in advanced ox- of Wnt/b-catenin signaling drives AKI to CKD progression. JAmSoc idation protein products-induced podocyte apoptosis. Kidney Int 82: Nephrol 27: 1727–1740, 2016 759–770, 2012 25. Cardus A, Uryga AK, Walters G, Erusalimsky JD: SIRT6 protects human 40. von Toerne C, Schmidt C, Adams J, Kiss E, Bedke J, Porubsky S, et al.: endothelial cells from DNA damage, telomere dysfunction, and se- Wnt pathway regulation in chronic renal allograft damage. Am J nescence. Cardiovasc Res 97: 571–579, 2013 Transplant 9: 2223–2239, 2009 26. d’Adda di Fagagna F: Living on a break: Cellular senescence as a DNA- 41. Zhou T, He X, Cheng R, Zhang B, Zhang RR, Chen Y, et al.: Implication of damage response. Nat Rev Cancer 8: 512–522, 2008 dysregulation of the canonical wingless-type MMTV integration site 27. Wolstein JM, Lee DH, Michaud J, Buot V, Stefanchik B, Plotkin MD: (WNT) pathway in diabetic nephropathy. Diabetologia 55: 255–266, 2012 INK4a knockout mice exhibit increased fibrosis under normal condi- 42. Liu H, Fergusson MM, Castilho RM, Liu J, Cao L, Chen J, et al.: Aug- tions and in response to unilateral ureteral obstruction. Am J Physiol mented Wnt signaling in a mammalian model of accelerated aging. Renal Physiol 299: F1486–F1495, 2010 Science 317: 803–806, 2007 28. Romero Y, Bueno M, Ramirez R, Álvarez D, Sembrat JC, Goncharova 43. Gu Z, Tan W, Feng G, Meng Y, Shen B, Liu H, et al.: Wnt/b-catenin EA, et al.: mTORC1 activation decreases autophagy in aging and idi- signaling mediates the senescence of bone marrow-mesenchymal opathic pulmonary fibrosis and contributes to apoptosis resistance in stem cells from systemic lupus erythematosus patients through the IPF fibroblasts [published online ahead of print August 26, 2016]. Aging p53/p21 pathway. Mol Cell Biochem 387: 27–37, 2014 Cell doi:10.1111/acel.12514 44. Meng XM, Tang PM, Li J, Lan HY: TGF-b/Smad signaling in renal fi- 29. Muñoz-Espín D, Cañamero M, Maraver A, Gómez-López G, Contreras brosis. Front Physiol 6: 82, 2015 J, Murillo-Cuesta S, et al.: Programmed cell senescence during mam- 45. Tan RJ, Zhou D, Liu Y: Signaling crosstalk between tubular epithelial malian embryonic development. Cell 155: 1104–1118, 2013 cells and interstitial fibroblasts after kidney injury. Kidney Dis (Basel) 2: 30. Mise K, Hoshino J, Ueno T, Hazue R, Hasegawa J, Sekine A, et al.: 136–144, 2016 Prognostic value of tubulointerstitial lesions, urinary N-acetyl-b-d- 46. Bernhardt A, Fehr A, Brandt S, Jerchel S, Ballhause TM, Philipsen L, glucosaminidase, and urinary b2-microglobulin in patients with type 2 et al.: Inflammatory cell infiltration and resolution of kidney in- diabetes and biopsy-proven diabetic nephropathy. Clin J Am Soc flammation is orchestrated by the cold-shock protein Y-box binding Nephrol 11: 593–601, 2016 protein-1. Kidney Int 92: 1157–1177, 2017 31. Zhou L, Li Y, He W, Zhou D, Tan RJ, Nie J, et al.: Mutual antagonism of 47. Wang J, Liang M, Xu J, Cao W, Wang GB, Zhou ZM, et al.: Renal ex- Wilms’ tumor 1 and b-catenin dictates podocyte health and disease. J pression of advanced oxidative protein products predicts progression Am Soc Nephrol 26: 677–691, 2015 of renal fibrosis in patients with IgA nephropathy. Lab Invest 94: 966– 32. He W, Tan RJ, Li Y, Wang D, Nie J, Hou FF, et al.: Matrix metal- 977, 2014 loproteinase-7 as a surrogate marker predicts renal Wnt/b-catenin activity in CKD. JAmSocNephrol23: 294–304, 2012 33. He W, Tan R, Dai C, Li Y, Wang D, Hao S, et al.: Plasminogen activator inhibitor-1 is a transcriptional target of the canonical pathway of Wnt/ This article contains supplemental material online at http://jasn.asnjournals. beta-catenin signaling. JBiolChem285: 24665–24675, 2010 org/lookup/suppl/doi:10.1681/ASN.2017050574/-/DCSupplemental.

1256 Journal of the American Society of Nephrology J Am Soc Nephrol 29: 1238–1256, 2018