The Ser/Thr kinase p90RSK promotes kidney fibrosis by modulating fibroblast–epithelial crosstalk

Healthy kidney structure and environment rely on epithelial integrity and interactions between epithelial cells and other kidney cells. The Ser/Thr kinase 90 kDa ribosomal protein S6 kinase 1 (p90RSK) belongs to a protein family that regulates many cellular processes, including cell motility and survival. p90RSK is predominantly expressed in the kidney, but its possible role in chronic kidney disease (CKD) remains largely unknown. Here, we found that p90RSK expression is dramatically activated in a classic mouse obstructive chronic kidney disease model, largely in the interstitial FSP-1–positive fibroblasts. We generated FSP-1–specific p90RSK transgenic mouse (RSK-Tg) and discovered that these mice, after obstructive injury, display significantly increased fibrosis and enhanced tubular epithelial damage compared with their wt littermates (RSK-wt), indicating a role of p90RSK in fibroblast–epithelial communication. We established an in vitro fibroblast–epithelial coculture system with primary kidney fibroblasts from RSK-Tg and RSK-wt mice and found that RSK-Tg fibroblasts consistently produce excessive H2O2 causing epithelial oxidative stress and inducing nuclear translocation of the signaling protein β-catenin. Epithelial accumulation of β-catenin, in turn, promoted epithelial apoptosis by activating the transcription factor forkhead box class O1 (FOXO1). Of note, blockade of reactive oxygen species (ROS) or β-catenin or FOXO1 activity abolished fibroblast p90RSK-mediated epithelial apoptosis. These results make it clear that p90RSK promotes kidney fibrosis by inducing fibroblast-mediated epithelial apoptosis through ROS-mediated activation of β-catenin/FOXO1 signaling pathway.

Regardless of the etiology, interstitial fibrosis, characterized by extensive interstitial fibroblast activation and excessive extracellular matrix deposition, is a hallmark of chronic kidney disease (CKD) 4 (1)(2)(3). The extent of interstitial fibrosis generally predicts the prognosis of patients with CKD (1,2,4). Both interstitial fibroblasts and tubular epithelial cells play essential roles in fibrogenesis and CKD progression. Interstitial fibroblasts are considered to be the primary matrix-producing cells and principal mediators of renal fibrosis associated with progressive renal failure (2,5,6). In response to injury, epithelial cells, especially proximal tubular epithelial cells, not only initiate inflammatory response by producing proinflammatory chemokines, but also undergo apoptotic death, leading to kidney parenchymal destruction. Structurally, fibroblasts reside in the renal interstitium surrounding the tubules formed by epithelial cells. This proximity facilitates interstitial-epithelial communication and interactions that are fundamental in maintaining the integrity of the kidney structure and environment, as well as fine-regulated process of adaption to pathogenic cues (7,8). However, the exact mechanisms of these interactions and their roles in CKD are largely unknown.
The 90 kDa ribosomal S6 kinases (RSKs) are a group of serine/threonine kinases that regulate diverse cellular process, such as cell growth, cell motility, and cell survival (9,10). There are four RSK isoforms (RSK1-4), of which RSK1 is also designated as p90RSK and is predominantly expressed in the kidney (9,10), suggesting that p90RSK may play a unique role in the pathogenesis of CKD. Recently, p90RSK has been shown to promote endothelial cell dysfunction and atherosclerosis in a diabetes model (11). However, the role of p90RSK in CKD development and progression remains unknown.
In the present study, we investigated the role of p90RSK in CKD pathogenesis in a novel FSP-1-specific p90RSK transgenic mouse. Our data demonstrated that p90RSK promotes kidney fibrosis by inducing fibroblast-mediated epithelial apoptosis through a novel signaling mechanism involving reactive oxygen species (ROS), ␤-catenin, and forkhead box class O1 (FOXO1).

p90RSK is activated during chronic kidney disease
To evaluate the activation of p90RSK during CKD, we examined the phosphorylation of p90RSK in the fibrotic kidneys from C57BL/6J mice with unilateral ureter obstruction (UUO), a classic CKD model, for 3, 7, and 14 days. We found that with the progression of kidney fibrosis, as indicated by increased fibronectin deposition (Fig. 1, A and C), p90RSK phosphorylation was dramatically induced (Fig. 1, A and B). However, there was little change of total-p90RSK abundance during kidney fibrosis (Fig. 1, A and B). Of note, phospho-p90RSK was largely induced in the interstitial FSP-1-positive cells (Fig. 1D).

FSP-1-specific p90RSK accelerates kidney fibrosis
Based on the above finding that p90RSK is activated, primarily in FSP-1-positive interstitial cells, during CKD ( Fig. 1), we generated FSP-1-specific p90RSK transgenic mice (RSK-Tg) by cross-breeding p90RSK-Tg flox mice (11) and S100A4 (FSP-1)-Cre mice (The Jackson Laboratory). Although other types of cells, such as macrophages (12) and podocytes (13), also express FSP-1, it has been verified in various organ systems, including kidneys, that S100A4 (FSP-1)-Cre specifically targets fibroblasts (14 -17). In particular, Inoue et al (14), using enhanced GFP reporter mice, have further confirmed that the FSP-1 promoter only drives Cre recombinase expression in interstitial fibroblasts in kidneys. Consistently, we also confirmed the overexpression of p90RSK in primary kidney fibroblasts from these transgenic mice ( Fig. 2A) compared with their littermates (RSK-wt), and RSK-Tg mice displayed dramatically increased phosphorylation of p90RSK in FSP-1-positive fibroblasts after fibrotic kidney injury (Fig. 2B). Under physiological condition, RSK-Tg mice had a phenotype similar to RSK-wt mice (data not shown). However, after receiving UUO for 7 days, RSK-Tg mice displayed significantly more severe tubular epithelial damage (Fig. 2, C and D) and increased interstitial fibronectin and collagen deposition (Fig. 2, E-G).

p90RSK induces fibroblast-mediated tubular epithelial apoptosis
We further evaluated apoptosis in the fibrotic kidneys from RSK-Tg and RSK-wt mice using TUNEL and cleaved caspase-3 immune staining. It was shown that there were more apoptotic cells in the obstructed kidneys from RSK-Tg mice, and these apoptotic cells were almost exclusively lectin-positive proximal tubular epithelial cells (Fig. 3, A-C), consistent with the histological assessment of epithelial damage (Fig. 2, C and D). Because FSP-1-positive cells are largely fibroblasts in the kidneys, it is presumable that fibroblast-specific p90RSK mediates epithelial damage. We established an in vitro fibroblast-epithelial coculture system to test our hypothesis (Fig. 3D). Briefly, primary kidney fibroblasts extracted from RSK-Tg and RSK-wt mice were seeded onto Transwell inserts in a 6-well plate with HKC-8 cells on the bottom. After serum-free overnight, low concentration of two classical apoptosis inducers, staurosporine and H 2 O 2 , were added to the lower chamber for additional 4 or 16 h, respectively, and followed by apoptosis assay. Consistent with our in vivo studies (Fig. 3, A-C), RSK-Tg fibroblasts dramatically enhanced epithelial apoptosis compared with RSK-wt fibroblasts (Fig. 3, E-H), suggesting a pathogenic role of p90RSK in mediating fibroblast-induced epithelial damage.

p90RSK induces fibroblast-mediated oxidative stress and triggers tubular epithelial apoptosis
To clarify the underlying mechanism, we examined the level of 4-hydroxynonenal (4-HNE), a specific marker for oxidative stress, in the fibrotic kidneys from RSK-Tg mice and their littermates. We found that there was remarkably p90RSK promotes kidney fibrosis increased 4-HNE expression largely in tubular epithelial cells of the fibrotic kidneys of RSK-Tg mice compared with control littermates (Fig. 4A). Notably, positive 4-HNE staining concentrated at the basal aspect of tubular epithelial cells, as well as in the interstitium (Fig. 4A), strongly implicating an interstitial origin of oxidative stress. We also measured the ROS level in mouse kidney homogenates and found that obstruction-induced renal ROS production was dramatically increased in RSK-Tg mice (Fig. 4B). To confirm the role of oxidative stress in fibroblast-mediated epithelial damage, we further measured the H 2 O 2 content in the medium of our coculture system. We found that RSK-Tg fibroblasts consistently released remarkably higher levels of H 2 O 2 into the coculture medium (Fig. 4C). To clarify whether ROS is necessary to fibroblast p90RSK-induced epithelial cell death, ROS-specific inhibitor YCG063 was added into the coculture. We found that YCG063 remarkably decreased the effects of fibroblast-specific p90RSK in both staurosporine-and H 2 O 2 -induced epithelial apoptosis (Fig.  4, D-G). Thus, it is clear that ROS mediates fibroblast-specific p90RSK-induced epithelial cell death.

␤-Catenin acts as a downstream effector of ROS and mediates fibroblast p90RSK-induced tubular epithelial apoptosis
We further found that H 2 O 2 , at a concentration comparable with that in the medium of fibroblast-epithelial coculture system, was able to induce epithelial ␤-catenin nuclear translocation (Fig. 5A). We also uncovered that ROS mediated epithelial induction of ␤-catenin, because YCG063, a specific ROS inhibitor, largely abolished fibroblast-specific p90RSK-induced epithelial ␤-catenin accumulation (Fig. 5, B and C), as well as active ␤-catenin (Fig. 5D), in our fibroblast-epithelial coculture system. Moreover, knockdown of epithelial ␤-catenin by siRNA, as indicated by Western blotting in HKC-8 nuclear extracts (Fig.  5E), almost completely abolished p90RSK-enhanced epithelial apoptosis (Fig. 5, E and F). Thus, we conclude that ROS-activated epithelial ␤-catenin mediates fibroblast-specific p90RSKinduced tubular epithelial apoptosis.

FOXO1 mediates fibroblast p90RSK-induced tubular epithelial apoptosis
FOXO1, a downstream transcription factor of oxidative stress pathway (18,19), has been shown to bind to ␤-catenin (20) and mediate TGF-␤1-induced myofibroblast activation (21,22). Given the prominent role of TGF-␤1 and ␤-catenin in kidney fibrosis, we further examined the role of FOXO1 in fibroblast-specific p90RSK-induced epithelial apoptosis. First, we checked the effect of H 2 O 2 , at a concentration comparable to that in the medium of fibroblast-epithelial coculture system, on epithelial FOXO1 activity. We found that H 2 O 2 dramatically induced FOXO1 nuclear translocation in epithelial cells (Fig.  6A). We further examined the epithelial FOXO1 activation level in our fibroblast-epithelial coculture system using two complementary approaches such as measuring FOXO1 abundance in epithelial nuclear extracts (Fig. 6, B and C) and evaluating FOXO1 activity by a TransAM FKHR Transcription Factor ELISA Kit (Fig. 6D). We found that RSK-Tg fibroblasts triggered a dramatically higher level of epithelial FOXO1 activation than RSK-wt fibroblasts (Fig. 6, B-D). Intriguingly, we also discovered that FOXO1 was dramatically up-regulated, predominantly in tubular epithelial nuclei, in the fibrotic kidneys from RSK-Tg mice (Fig. 6E), indicating an in vivo significance of FOXO1 in fibroblast-epithelial communication and kidney fibrosis. We further found that ROS-specific inhibitor YCG063 eliminated RSK-Tg fibroblast-induced FOXO1 activation in epithelial cells (Fig. 6F), suggesting that FOXO1 acted in the downstream of ROS pathway. Additionally, FOXO1-specific inhibitor AS1842856 (Fig. 6, G and H) or siRNA (Fig. 6, I-K) abolished fibroblast-specific p90RSK-induced epithelial apoptosis in fibroblast-epithelial coculture system. Thus, FOXO1 clearly mediates fibroblast-specific p90RSK-induced tubular epithelial death.

Discussion
Although p90RSK has been shown to be predominantly expressed in the kidney (9, 10), its in vivo role in CKD pathogenesis remains unknown. Our previous in vitro work has demonstrated that p90RSK is a key control point that regulates the size of the interstitial fibroblast population (23-25), which largely determines the outcome in patients with chronic kidney

p90RSK promotes kidney fibrosis
injury (1,2,4,6,26,27). p90RSK plays an important role in the Ras mitogen-activated protein kinase (MAPK) signaling cascade and is the direct downstream effector of Ras-Erk1/2 signaling. Erk1/2 activation directly phosphorylates and activates p90RSK (9,10,28), which in turn activates various signaling events through selection of different phosphorylation substrates including GSK3␤ and BAD (9, 10), suggesting a pivotal role of p90RSK in tissue fibrosis. This notion is supported by  p90RSK promotes kidney fibrosis recent findings that p90RSK is involved in the pathogenesis of atherosclerosis of various causes (11,29,30). In the present study, we found that the abundance of total p90RSK was little changed during progressive CKD. However, phosphorylation or activation of p90RSK was dramatically induced and correlated with the extent of fibrotic injury (Fig. 1, A and B), suggesting that activation status of p90RSK, rather than its abundance, determines the severity of kidney fibrosis. Consistently, RSK-Tg mice, under physiological condition (lower p90RSK activation level), had a similar phenotype as their littermates, whereas, after obstructive injury (higher p90RSK activation level), these mice displayed a significantly enhanced fibrosis (Fig. 2). It is known that other types of cells, such as macrophages and podocytes, also express FSP-1. Previous work strongly indicates that FSP-1-Credriven cells are primarily fibroblasts in various organs, including kidneys (14 -17). Particularly, recent work from Inoue et al. (14) specifically confirms that the promoter for FSP-1 driving Cre recombinase only expresses in the interstitial fibroblasts in the same obstructive CKD model using enhanced GFP reporter mice. However, we cannot exclude the possible contribution from macrophages to p90RSK-induced epithelial apoptosis because some kidney macrophages also express FSP-1. Future investigations in this aspect are warranted. In summary, we have found that fibroblast-specific p90RSK induces tubular epithelial apoptosis and promotes kidney fibrosis. As summarized in Fig. 7, fibroblasts with p90RSK overexpression produce and release excessive H 2 O 2 , causing ROS accumulation and ␤-catenin nuclear translocation in the surrounding epithelial cells. Excessive ␤-catenin interacts with transcription factor FOXO1 to promote tubular epithelial apoptosis, leading to kidney structural destruction and eventually fibrosis.
The current results have illustrated a novel mechanism and the pivotal role of p90RSK-mediated fibroblast-epithelial communications in CKD development and progression. The juxtaposition of fibroblasts and epithelium facilitates their interactions. Under physiological condition, interstitialepithelial communication plays a fundamental role in maintaining the integrity of the kidney structure and environment (7,8). As an essential part of wound healing process, renal interstitial cells, including fibroblasts, also provide a supportive environment to promote tubular epithelial regeneration in response to transient injury (31). However, persistent pathogenic stimuli will cause extensive activation of signaling mediators, such as p90RSK (Fig. 1), and disrupt the delicate healing process, resulting in progressive tissue destruction and loss of function. The present results have

p90RSK promotes kidney fibrosis
shown that RSK-Tg fibroblasts, after chronic obstructive injury, acquire substantially enhanced ability to generate sustained H 2 O 2 that not only induces epithelial injury (Fig.  7), but also further triggers the activation of p90RSK (32,33) and forms a vicious loop of amplification, eventually leading to irreversible kidney fibrosis. Several factors, including mitochondria and NADPH oxidases, may contribute to the generation of ROS (34). It is possible that p90RSK induces oxidative stress through modulating the functions of these factors. However, future investigations are required to elucidate the exact mechanisms.
In summary, our results demonstrate that p90RSK plays an essential role in promoting kidney fibrosis through a novel mechanism of fibroblast-epithelial communication involving ROS, ␤-catenin, and FOXO1. These finding indicate that p90RSK may be a promising therapeutic target for CKD treatment.

Animal model
Animal studies were performed using a protocol (45872) approved by the Institutional Animal Care and Use Committee at the Pennsylvania State University College of Medicine. FSP-1-specific p90RSK transgenic mice (RSK-Tg) and their littermates (RSK-wt) were generated by cross-breeding p90RSK-Tg flox mice (11) and S100A4 (FSP-1)-Cre mice (The Jackson

p90RSK promotes kidney fibrosis
Laboratory). The original p90RSK-Tg flox and FSP-1-Cre mice had been backcrossed for at least 10 generations to the C57BL/6 background. UUO was performed in 20 -22 g sex-matched mice (five animals per group) using established procedures described previously (42)(43)(44). The right unobstructed kidneys served as controls. At day 7 after UUO, mice were sacrificed and kidneys were harvested for analyses.

TUNEL and immunofluorescence staining
Snap-frozen kidney was cryosectioned at 5-m thickness and subjected to TUNEL or immunofluorescence staining as described previously (25,45). TUNEL was performed using Roche Cell Death Detection kit. Fluorescence-conjugated lectins from Tetragonolobus purpureas (BioWorld and Vector Laboratories) were used to localize the proximal tubules. Images were acquired by an Olympus IX81 fluorescence microscope (Olympus America Inc.). Sections stained without primary antibodies served as negative controls. TUNEL-positive epithelial cells were counted from five non-overlapping 400ϫ microscopic fields per slide, and the percentage of TUNELpositive epithelial cells in total epithelial cell population was calculated, five mice per group.

Immunohistochemistry and Sirius Red staining
Paraffin-embedded kidney tissue was sectioned at 4 m and then subjected to Sirius Red or immune staining (23). Briefly, tissue sections were deparaffinized, hydrated, and antigen-retrieved, followed by Sirius Red staining using Picrosirius Red Stain kit or by incubation with primary and secondary antibodies for immune staining. Sections stained without primary antibodies served as negative controls. Sections were then incubated with ABC reagents, followed by NovaRED or DAB staining, and counterstained with methyl green (Vector Laboratories).

Fibroblast-epithelial coculture
Human kidney proximal tubular epithelial cells (HKC-8) were maintained as described previously (48). Primary kidney fibroblasts were isolated from fibroblast-specific p90RSK transgenic mice (RSK-Tg) and WT (RSK-wt) mice and maintained as reported (49). Primary fibroblasts were validated by morphology and positive staining of FSP-1 and vimentin. An in vitro fibroblast-epithelial coculture system was established using above primary kidney fibroblasts and HKC-8 cells. Briefly, primary fibroblasts (3 ϫ 10 5 cells) were seeded onto Transwell inserts (0.4 mm PET, 4.5 cm 2 ) in a 6-well plate with HKC-8 cells (3 ϫ 10 5 cells) on the bottom. After serum-free overnight, 50 nM staurosporine or 150 M H 2 O 2 was added into the lower chamber for an additional 4 or 16 h, respectively, followed by apoptosis assay. In some experiments, epithelial cells were treated with various chemical inhibitors, such as FOXO1 inhibitor (AS1842856, 1 M, EMD Millipore, Billerica, MA) (46) or ROS inhibitor (YCG063, 50 nM, Calbiochem).

Western blot analysis
Tissue or cell lysates were prepared and separated on SDSpolyacrylamide gels as described previously (23,25,45). The PVDF membrane with transferred proteins was probed with various antibodies, and the signals on the membrane were visualized by the SuperSignal West Pico Chemiluminescent Substrate kit (Fisher Scientific).

FOXO1 activity
Primary kidney fibroblasts were cocultured with HKC-8 cells in serum-free medium for 24 h, then HKC-8 cells were harvested at 1, 2, 4, 12, and 24 h. In some experiments, 50 nM ROS-specific inhibitor YCG063 was added into the coculture medium for 24 h. Then, nuclear fractions were isolated from cells using a TransAM Nuclear Extract Kit (Active Motif, Carlsbad, CA), and FOXO1 activities were determined using a TransAM FKHR Transcription Factor ELISA Kit (Active Motif, Carlsbad, CA).

Oxidative stress
Mouse kidney ROS level was assessed by a Mouse Reactive Oxygen Species ELISA kit (Neo Scientific). Hydrogen peroxide level in coculture medium was evaluated by a hydrogen peroxide assay kit from Bioassay Systems.

Hydrogen peroxide production assay
HKC-8 and fibroblasts were cocultured in serum-free medium for 4 h, followed by measurement of hydrogen peroxide content in the coculture medium by a hydrogen peroxide assay kit (Bioassay Systems, Hayward, CA).

Statistical analysis
All the experimental data were presented as mean Ϯ S.E. Statistical analysis of data were performed using SigmaStat 3.5 software (Jandel Scientific Software). Comparison between multiple groups was performed by using one-way analysis of variance (ANOVA) followed by the Student-Newman-Keuls test or Student's t test between two groups. A p value of less than 0.05 was considered statistically significant.