Hippuric Acid Promotes Renal Fibrosis by Disrupting Redox Homeostasis via Facilitation of NRF2–KEAP1–CUL3 Interactions in Chronic Kidney Disease

Chronic kidney disease (CKD) is characterized by the accumulation of protein-bound uremic toxins (PBUTs), which play a pathophysiological role in renal fibrosis (a common pathological process resulting in CKD progression). Accumulation of the PBUT hippuric acid (HA) is positively correlated with disease progression in CKD patients, suggesting that HA may promote renal fibrosis. Oxidative stress is the most important factor affecting PBUTs nephrotoxicity. Herein, we assessed the ability of HA to promote kidney fibrosis by disrupting redox homeostasis. In HK-2 cells, HA increased fibrosis-related gene expression, extracellular matrix imbalance, and oxidative stress. Additionally, reactive oxygen species (ROS)-mediated TGFβ/SMAD signaling contributed to HA-induced fibrotic responses. HA disrupted antioxidant networks by decreasing the levels of nuclear factor erythroid 2-related factor 2 (NRF2), leading to ROS accumulation and fibrotic responses, as evidenced by NRF2 activation and knockdown. Moreover, NRF2 levels were reduced by NRF2 ubiquitination, which was regulated via increased interactions of Kelch-like ECH-associated protein 1 with Cullin 3 and NRF2. Finally, renal fibrosis and redox imbalance promoted by HA were confirmed in rats. Importantly, sulforaphane (NRF2 activator) reversed HA-promoted renal fibrosis. Thus, HA promotes renal fibrosis in CKD by disrupting NRF2-driven antioxidant system, indicating that NRF2 is a potential therapeutic target for CKD.


Introduction
Approximately 10-15% of the world population suffers from chronic kidney disease (CKD), making it a serious public health and social issue [1]. CKD is a progressive disease in which sustained damage to the kidney structure ultimately results in irreversible kidney failure, defined as the end-stage renal disease (ESRD) [2]. Renal fibrosis, a common pathological process of CKD, results in the multiple types of CKD progress to ESRD [3,4], in which original functional tissue is replaced by connective tissue, resulting in loss of the kidney architecture [5]. During CKD, renal fibrosis progression is closely related

Animal Experiments
All procedures and protocols involving rats were approved by the Ethical Committee of Experimental Animal Care at China Agricultural University and complied with ARRIVE guidelines (ethical code: KY19034). Male Sprague Dawley rats at the age of 7 weeks were housed in a pathogen-free environment. The environment was that with a humidity of 50-70%, temperature of 22-24 • C, and the day and night cycle was 12/12 h. These rats were given a normal laboratory diet (Shanghai Puluteng Biological Technology Co., Ltd., Shanghai, China) with ad libitum access to water. After 1 week of adaptation, fifty rats were subjected to either sham surgery (kidney exposure operation only) or 5/6NX surgery [30]. The 5/6NX model commonly used to simulate typical CKD symptoms and study PUBT-promoted renal fibrosis, such as IS and PCS [7,15,17,18]. Two weeks after surgery (experimental week 1), excluding the rats that died during the operation, rats in the sham group (n = 9) were treated with vehicle (0.5% DMSO), and 5/6NX rats were randomized into four groups (n = 9 each) and treated as follows: (1) a 5/6NX group, 0.5% DMSO; (2) a 5/6NX + SFN group, 2.5 mg/kg SFN; (3) a 5/6NX + HA group, 100 mg/kg HA; (4) a 5/6NX + HA + SFN group, 100 mg/kg HA and 2.5 mg/kg SFN. All treatments were administered intraperitoneally, five times per week for 10 weeks.

Cell Viability
The viability of HK-2 cells was determined using Cell Counting Kit-8 (Beyotime, Shanghai, China) at 450 nm on a microplate reader (Bio-Rad, Hercules, CA, USA), according to the manufacturer's recommendations. The baseline level was measured on a microplate containing the cell culture medium, DMSO (0.1%), and CCK-8 solution but no cells (n = 6).

Quantitative RT-PCR
Total RNA of HK-2 was obtained by TRIzol solution (Invitrogen, Carlsbad, CA, USA). Reverse transcription was carried out by 5× All-In-One RT MasterMix (Applied Biological Materials, Inc., Richmond, BC, Canada). The quantitative polymerase chain reaction was performed on a 7900HT instrument (Applied Biosystems, Foster City, CA, USA) by TB Green Premix Ex Taq (Takara Biomedical Technology Co., Ltd., Shiga, Japan) following the manufacturers' protocol. The sequences of primers were shown in Supplementary Materials (Table S1).

Western Blotting (WB)
Total proteins of cells were extracted by the ice-cold RIPA solution (Beyotime). Nuclear and Cytoplasmic Protein Extraction kit (Beyotime) was applied to obtain the nuclear and cytoplasmic proteins. Equal-concentration protein samples were separated using SDS-PAGE. Then the samples were shifted to polyvinylidene difluoride membranes (EMD Millipore, Billerica, MA, USA). The membranes were blocked at 4 • C overnight and then were incubated using the indicated primary antibody. Next, the membranes were incubated with a secondary antibody. The enhanced chemiluminescence (EMD Millipore) was used to develop the signals.

Immunofluorescence Assays
HK-2 cells grown in 6 well plates were fixed for about 10 min using paraformaldehyde (4%) and then permeabilized about 10 min using Triton X-100 (0.25%) at 25 • C. Then, the samples were blocked for 1 h using blocking buffer (Beyotime). After washing, the samples were incubated with primary antibodies and then secondary fluorescent antibodies. The nuclei of cells were stained by DAPI solution (Solarbio, Beijing, China) for 5 min.

ROS and Hydrogen Peroxide (H 2 O 2 ) Assays
Total intracellular ROS and H 2 O 2 concentrations were determined using commercial ROS and H 2 O 2 assay kits (Beyotime) in accordance with the manufacturer's instruction manuals.

NRF2 Knockdown
The cells were transfected with a NRF2-specific small interfering RNA (siRNA) or with a siRNA control (Table S2) by Lipofectamine 2000 (Thermo Fisher Scientific, Waltham, MA, USA) in accordance with the manufacturer's instruction manuals.

Coimmunoprecipitation (Co-IP)
NRF2 ubiquitination and protein-protein interactions were determined by Co-IP using a kit of Pierce™ Crosslink IP (Thermo Fisher Scientific) following the manufacturers' protocol. Co-IP products were boiled in loading buffer and subjected to WB with the appropriate antibodies.

Molecular Docking
The HA structure was obtained from the PubChem database (CID 464). The KEAP1 structure (PDB 4N1B) was obtained by the Protein Data Bank. Water molecules, acetate ion, and the original ligand molecules 2FS were deleted. Subsequently, HA was docked into KEAP1 active sites using Autodock Vina software 1.1.2 (Scripps Research Institute, La Jolla, CA, USA). The interaction model of minimum binding energy was selected.

Malondialdehyde and Antioxidant Enzyme Assays
Malondialdehyde (MDA) in kidney tissues was assessed with an MDA assay kit (Beyotime) following the manufacturers' protocol. The activity of superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GSH-Px) in renal samples were quantified with enzyme-specific assay kits (Beyotime) following the manufacturers' protocol.

Statistical Analyses
All values were expressed as mean ± standard deviation and analyzed using SPSS 21.0 (SPSS, Inc., Chicago, IL, USA). The Gaussian distribution of data was analyzed by the Shapiro-Wilk test. Statistical comparison was measured by one-way ANOVA followed by Tukey's post-test (data with Gaussian distribution) or Kruskal-Wallis followed by Dunn's post-test (data without Gaussian distribution). The data that does not conform to the normal distribution are the WB results of MMP9 ( Figure 1B Figure 6C). Differences were considered significant when p < 0.05.

HA Displays Strong Potential to Cause Fibrotic Responses In Vitro
To evaluate the potential role of HA in promoting fibrotic responses, HK-2 cells were incubated with HA for 24 h at concentrations of 0-1000 µM, based on mean serum HA concentration in patients on hemodialysis (610.78 µM; EUTox data). There were no significant differences in cell viability at 0-1000 µM HA ( Figure S1). Next, renal fibrosis markers were assessed, including COL1A1, VIM, CDH1, and ACTA2. HA significantly increased COL1A1, VIM, and ACTA2 mRNA and protein levels, and reduced those of CDH1 in a concentration-dependent manner ( Figure 1A,B). Immunofluorescence staining for CDH1 and ACTA2 revealed similar results, with the CDH1 intensity significantly decreased and that of ACTA2 increased at 250 and 1000 µM HA compared to the control ( Figure 1C). Furthermore, MMP9 and TIMP1 expression levels were assessed. HA significantly attenuated MMP9 mRNA and protein levels, whereas it significantly increased those of TIMP1 ( Figure 1A,B), suggesting that HA induced extracellular matrix (ECM) imbalance. These results indicate that HA caused fibrogenic responses in HK-2 cells.

TGFβ/SMAD Signaling is Involved in HA-Induced Fibrotic Responses In Vitro
To elucidate the underlying molecular mechanism of the effects of HA on fibrotic responses, TGFβ/SMAD signaling was investigated in HK-2 cells. Phosphorylation of SMAD2 and SMAD3, as well as the protein expression levels of SMAD4, significantly increased by HA treatment in a concentration-dependent manner ( Figure 2). Moreover, expression of the transcription factor SNAI1 , alpha-smooth muscle actin (ACTA2), matrix metalloproteinase-9 (MMP9), and tissue inhibitor of metalloproteinases-1 (TIMP1) after treatment with 0-1000 µM HA. Quantification was performed by densitometry. (C) Immunofluorescence staining for CDH1 (red) and ACTA2 (green) after treatment with 250 and 1000 µM HA. Nuclei (blue) were stained with DAPI. Bar = 100 µm. Data are presented as the mean ± standard deviation. * p < 0.05 vs. control (Ctrl) cells. Experiments were repeated three to four times.

TGFβ/SMAD Signaling Is Involved in HA-Induced Fibrotic Responses In Vitro
To elucidate the underlying molecular mechanism of the effects of HA on fibrotic responses, TGFβ/SMAD signaling was investigated in HK-2 cells. Phosphorylation of SMAD2 and SMAD3, Antioxidants 2020, 9, 783 7 of 20 as well as the protein expression levels of SMAD4, significantly increased by HA treatment in a concentration-dependent manner ( Figure 2). Moreover, expression of the transcription factor SNAI1 was significantly enhanced in renal tubular cells ( Figure 2). These data suggest that HA induced fibrotic responses by activating TGFβ/SMAD signaling.
Antioxidants 2020, 9, x FOR PEER REVIEW 7 of 21 was significantly enhanced in renal tubular cells ( Figure 2). These data suggest that HA induced fibrotic responses by activating TGFβ/SMAD signaling.

HA Induces Fibrotic Responses Via ROS-Activated TGFβ/SMAD Signaling In Vitro
To investigate whether oxidative stress is involved in HA-induced fibrotic responses, the levels of oxygen free radicals were analyzed in HK-2 cells. Increasing HA concentrations not only significantly augmented the production of ROS and H2O2 but also promoted the mRNA and protein expression of NOX4, a known enzymatic source of ROS in the renal system ( Figure 3A,B). These data indicate that HA induced oxidative stress in tubular cells.
Further, HK-2 cells were pretreated with the free radical scavenger NAC (1000 µM) and NOX4 inhibitor DPI (50 µM), followed by incubation with HA (1000 µM). NAC and DPI suppressed ROS and H2O2 production in HK-2 cells treated with 1000 µM HA ( Figure S2). Moreover, significant decreases in COL1A1, VIM, and ACTA2 protein levels and an increase in the CDH1 level were observed in both the HA + NAC and HA + DPI groups, compared with those in the HA-only group ( Figure 3C). Consistent with the WB results, immunofluorescence revealed that the two antioxidants (NAC and DPI) significantly reversed the reduction in CDH1 expression and enhancement of ACTA2 expression ( Figure 3D). Additionally, HA-induced elevation in the levels of key proteins (p-SMAD2, p-SMAD3, SMAD4, and SNAI1) in the TGFβ/SMAD signaling pathway was reversed by NAC and DPI pretreatment of HK-2 cells ( Figure 3E). Together, these data indicated that the HA-induced fibrotic responses were mediated by ROS activation of the TGFβ/SMAD pathway in renal tubular cells.

HA Induces Fibrotic Responses via ROS-Activated TGFβ/SMAD Signaling In Vitro
To investigate whether oxidative stress is involved in HA-induced fibrotic responses, the levels of oxygen free radicals were analyzed in HK-2 cells. Increasing HA concentrations not only significantly augmented the production of ROS and H 2 O 2 but also promoted the mRNA and protein expression of NOX4, a known enzymatic source of ROS in the renal system ( Figure 3A,B). These data indicate that HA induced oxidative stress in tubular cells.
Further, HK-2 cells were pretreated with the free radical scavenger NAC (1000 µM) and NOX4 inhibitor DPI (50 µM), followed by incubation with HA (1000 µM). NAC and DPI suppressed ROS and H 2 O 2 production in HK-2 cells treated with 1000 µM HA ( Figure S2). Moreover, significant decreases in COL1A1, VIM, and ACTA2 protein levels and an increase in the CDH1 level were observed in both the HA + NAC and HA + DPI groups, compared with those in the HA-only group ( Figure 3C). Consistent with the WB results, immunofluorescence revealed that the two antioxidants (NAC and DPI) significantly reversed the reduction in CDH1 expression and enhancement of ACTA2 expression ( Figure 3D). Additionally, HA-induced elevation in the levels of key proteins (p-SMAD2, p-SMAD3, SMAD4, and SNAI1) in the TGFβ/SMAD signaling pathway was reversed by NAC and DPI pretreatment of HK-2 cells ( Figure 3E). Together, these data indicated that the HA-induced fibrotic responses were mediated by ROS activation of the TGFβ/SMAD pathway in renal tubular cells.

Effects of HA Treatment on the NRF2-Driven Antioxidant Pathway In Vitro
Under ROS stress, the NRF2-mediated antioxidant system can be activated and eliminate ROS, thereby preventing the initiation of oxidative stress [32,33]. To assess the effect of HA on the antioxidant network, NRF2 and its downstream antioxidant enzymes (HO1 and NQO1) levels were measured. The expressions of mRNA and protein of HO1 and NQO1 were notably reduced with increasing concentrations of HA ( Figure 4A,B), suggesting that the NRF2-driven antioxidant pathway was disrupted. Interestingly, HA induced a concentration-dependent increase in NRF2 mRNA levels, accompanied by a significant decrease in cytoplasmic and nuclear NRF2 protein expression ( Figure 4A,B). Consistent with the WB data, NRF2 immunofluorescence significantly decreased at 250 and 1000 µM HA compared to the control ( Figure 4C). These results suggested that HA inhibited the antioxidant pathway by downregulating NRF2 protein.   increasing concentrations of HA ( Figure 4A,B), suggesting that the NRF2-driven antioxidant pathway was disrupted. Interestingly, HA induced a concentration-dependent increase in NRF2 mRNA levels, accompanied by a significant decrease in cytoplasmic and nuclear NRF2 protein expression ( Figure  4A,B). Consistent with the WB data, NRF2 immunofluorescence significantly decreased at 250 and 1000 µM HA compared to the control ( Figure 4C). These results suggested that HA inhibited the antioxidant pathway by downregulating NRF2 protein.

NRF2-Driven Antioxidant Network Mediates HA-Induced Fibrotic Responses In Vitro
To investigate the potential effect of NRF2 on fibrotic toxicity of HK-2 cells caused by HA, we used the NRF2 activator SFN (0-20 µM) to reverse HA-induced NRF2 protein downregulation and then measured the changes in the expression of key genes related to renal fibrosis. Compared with that in HA-only treatment, ROS and H 2 O 2 production significantly suppressed following SFN + HA treatment in a concentration-dependent way, and NRF2, HO1, and NQO1 levels significantly upregulated ( Figure S3 and Figure 5A), suggesting that HA-induced ROS accumulation and dysfunction of the antioxidant network were suppressed by SFN. Furthermore, NRF2 activation significantly reversed HA-induced elevation of COL1A1, VIM, and ACTA2 levels and reduction of CDH1 levels ( Figure 5B), suggesting that HA-induced fibrotic responses were alleviated by NRF2 activation. Antioxidants 2020, 9, x FOR PEER REVIEW 11 of 21  To better clarify the role of NRF2 in HA-induced fibrotic responses, an NRF2-specific siRNA was introduced into HK-2 cells. NRF2 silencing exacerbated HA-induced ROS and H 2 O 2 generation and downregulated the NRF2-dependent antioxidant pathway ( Figure S4 and Figure 5C). Moreover, HA-induced upregulation of COL1A1, VIM, and ACTA2 levels, and downregulation of CDH1 was further augmented by NRF2 knockdown ( Figure 5D). In these experiments, the siRNA control group was not significantly different from the HA-only control ( Figure 5C,D), displaying no additional siRNA effects on HK-2 cells. Overall, these data indicate that the NRF2 antioxidant defense system, which was disrupted by downregulation of NRF2 protein, played a vital role in HA-induced fibrotic responses.

HA Increases NRF2 Ubiquitination by Increasing E3 Ubiquitin Ligase Activity In Vitro
To explore the mechanism of the HA-induced reduction of NRF2 protein, we first determined whether HA causes NRF2 ubiquitination in HK-2 cells. The levels of ubiquitinated NRF2 significantly increased after treatment with 1000 µM HA, and similar results were obtained in the presence of 10 µM MG132 ( Figure 6A), suggesting that HA promoted the ubiquitination of NRF2.
Further, to investigate whether HA affects E3 ubiquitin ligase activity, KEAP1 and CUL3 expression levels were analyzed in HK-2 cells. Interestingly, no significant changes in the mRNA and protein expression of the KEAP1 and CUL3 subunits were observed following HA treatment ( Figure 6B,C). It has been reported that E3 ubiquitin ligase activity also is regulated by protein-protein interactions [34]. Therefore, Co-IP was performed to reveal such interactions. At 1000 µM HA, immunoprecipitated KEAP1 pulled down more NRF2 compared to control-treated cells, and similar results were obtained for immunoprecipitated KEAP1 and CUL3 ( Figure 6D), suggesting that the interaction between KEAP1 and NRF2 or CUL3 had been strengthened. These results indicate that HA enhanced the activity of E3 ubiquitin ligase by strengthening the NRF2-KEAP1-CUL3 interactions.
To identify the HA target responsible for HA-induced NRF2 ubiquitination, a docking experiment was performed to model potential interactions between HA and KEAP1, a core component of the E3 ligase complex. HA appeared to dock into the cavity of human KEAP1 protein ( Figure 6E). The benzene ring of HA formed π-π interactions with the benzene ring of Tyr334 of KEAP1 and π-cation interactions with -NH2+ of Arg415 of KEAP1, whereas the terminal carboxyl oxygen of HA formed a salt bridge with -NH2+ of Arg415 of KEAP1 ( Figure 6F,G). These results suggest that HA stably binds to KEAP1 to exert its biological functions.
Collectively, these data suggest that HA directly binds to KEAP1, thereby enhancing the interactions of KEAP1 with CUL3 and NRF2, increasing the activity of E3 ligase complex, and ultimately increasing NRF2 ubiquitination and degradation.

HA Promotes Renal Dysfunction and Fibrosis by Disrupting the Antioxidation Function of Nrf2 In Vivo
To further explore the ability of HA to promote renal fibrosis in CKD in vivo, we generated a 5/6NX rat model to mimic typical CKD symptoms, which can simulate the actual pathogenic environment of HA-promoted renal fibrosis in patients with CKD more accurately. Compared with sham rats, 5/6NX rats showed significantly increased levels of SCr, BUN, and HA, as well as increased fibrosis area, glomerular sclerosis index, and fibrotic marker levels (COL1A1, VIM, and ACTA2), suggesting that the 5/6NX rat model was effective.
Serum HA levels in 5/6NX + HA rats were significantly higher than those in the 5/6NX rats ( Figure 7A), suggesting that HA effectively accumulated in vivo. Compared to the 5/6NX rats, SCr and BUN contents increased in the 5/6NX + HA group ( Figure 7A), suggesting that HA accumulation worsened renal function. Moreover, SFN treatment (5/6NX + HA + SFN group) attenuated the increases in SCr, BUN, and HA levels, and significant changes were observed for SCr and HA ( Figure 7A).

HA Promotes Renal Dysfunction and Fibrosis by Disrupting the Antioxidation Function of Nrf2 In Vivo
To further explore the ability of HA to promote renal fibrosis in CKD in vivo, we generated a 5/6NX rat model to mimic typical CKD symptoms, which can simulate the actual pathogenic environment of HA-promoted renal fibrosis in patients with CKD more accurately. Compared with sham rats, 5/6NX rats showed significantly increased levels of SCr, BUN, and HA, as well as increased 5/6NX + HA rats, compared to 5/6NX rats ( Figure 7B). These data indicate that HA promoted the progression of renal fibrosis in the rat model. Furthermore, SFN treatment (5/6NX + HA + SFN group) significantly reduced the fibrosis area, glomerular sclerosis index, and fibrotic protein levels (COL1A1, VIM, and ACTA2) compared with those in the 5/6NX + HA group ( Figure 7B).
Overall, these data suggest that HA promotes renal fibrosis by disrupting the antioxidation function of Nrf2, consistent with the results obtained in HK-2 cells.  HA-injected rats (5/6NX + HA group) exhibited increased tubulointerstitial fibrosis and glomerulosclerosis, compared with those in the 5/6NX group ( Figure 7B). Consistent with these histological findings, the COL1A1-, VIM-, and ACTA2-positive areas were significantly larger in 5/6NX + HA rats, compared to 5/6NX rats ( Figure 7B). These data indicate that HA promoted the progression of renal fibrosis in the rat model. Furthermore, SFN treatment (5/6NX + HA + SFN group) significantly reduced the fibrosis area, glomerular sclerosis index, and fibrotic protein levels (COL1A1, VIM, and ACTA2) compared with those in the 5/6NX + HA group ( Figure 7B).
Overall, these data suggest that HA promotes renal fibrosis by disrupting the antioxidation function of Nrf2, consistent with the results obtained in HK-2 cells.

HA Induces Redox Imbalance by Disrupting the Antioxidation Function of Nrf2 In Vivo
To investigate whether the redox imbalance induced by HA occurs in vivo, we determined the levels of MDA, NRF2, and antioxidant genes, including SOD, CAT, and GSH-Px. Immunohistochemical analysis revealed that 5/6NX + HA rats had much lower NRF2 levels compared to 5/6NX rats ( Figure 8A). Moreover, the activities of SOD, CAT, and GSH-Px were decreased, whereas MDA levels were increased in 5/6NX + HA rats compared with those in 5/6NX controls ( Figure 8B), suggesting that redox homeostasis was disrupted by HA. Furthermore, SFN treatment (5/6NX + HA + SFN group) attenuated the reductions in NRF2 and its downstream antioxidant genes levels, and elevation in MDA contents, compared to the 5/6NX+HA group, with significant differences observed for NRF2, SOD, and MDA ( Figure 8A,B). These findings indicate that in the rat model, HA induced a redox imbalance by disrupting the antioxidation function of Nrf2, consistent with the results obtained in HK-2 cells.

HA Induces Redox Imbalance by Disrupting the Antioxidation Function of Nrf2 In Vivo
To investigate whether the redox imbalance induced by HA occurs in vivo, we determined the levels of MDA, NRF2, and antioxidant genes, including SOD, CAT, and GSH-Px. Immunohistochemical analysis revealed that 5/6NX + HA rats had much lower NRF2 levels compared to 5/6NX rats ( Figure 8A). Moreover, the activities of SOD, CAT, and GSH-Px were decreased, whereas MDA levels were increased in 5/6NX + HA rats compared with those in 5/6NX controls ( Figure 8B), suggesting that redox homeostasis was disrupted by HA. Furthermore, SFN treatment (5/6NX + HA + SFN group) attenuated the reductions in NRF2 and its downstream antioxidant genes levels, and elevation in MDA contents, compared to the 5/6NX+HA group, with significant differences observed for NRF2, SOD, and MDA ( Figure 8A,B). These findings indicate that in the rat model, HA induced a redox imbalance by disrupting the antioxidation function of Nrf2, consistent with the results obtained in HK-2 cells.

Discussion
The nephrotoxicity of HA and its pathogenic mechanism have not been widely evaluated. Our results indicate that HA (1) promoted renal fibrosis progression; (2) activated ROS-mediated TGFβ/SMAD signaling; (3) caused a redox imbalance by impairing the antioxidant system; and (4) increased the degradation of ubiquitinated NRF2 by facilitating NRF2-KEAP1-CUL3 interactions, in

Discussion
The nephrotoxicity of HA and its pathogenic mechanism have not been widely evaluated.
Our results indicate that HA (1) promoted renal fibrosis progression; (2) activated ROS-mediated TGFβ/SMAD signaling; (3) caused a redox imbalance by impairing the antioxidant system; and (4) increased the degradation of ubiquitinated NRF2 by facilitating NRF2-KEAP1-CUL3 interactions, in HK-2 cells and 5/6NX rats. To our knowledge, this paper is the first to disclose a profibrotic role for HA, which acts by disrupting redox homeostasis.
In healthy kidneys, PBUTs are cleared through tubular secretion mediated by organic anion transporters (OATs) controlling the uptake of free PBUTs to tubular cells and excretion of PBUTs into the urine [35]. In patients with CKD, PBUTs accumulation exerts unique physicochemical properties and displays toxicity in the development of renal fibrosis. As important PBUTs, IS and PCS can promote the upregulation of fibrotic proteins and cytokines, epithelial mesenchymal-like transition, collagen accumulation, and renal fibrosis, in both nephrectomy animal model and HK-2 cells [17,[36][37][38][39][40]. In this study, similar to IS and PCS, HA increased the levels of fibrotic proteins and induced an ECM imbalance in vitro, with similar results obtained in vivo. Moreover, we previously reported that Eggerthella lenta, an HA-producing bacterium, increased HA production and aggravated the progression of renal fibrosis in a 5/6NX rat model [41]. These findings provide compelling evidence that HA accumulation promotes the development of renal fibrosis in CKD.
A typical feature of renal fibrosis is high oxidative stress, a state involving uremic metabolite accumulation [42,43]. Previous studies reported that IS and PCS have strong prooxidant properties and can promote kidney injury and organ fibrosis by augmenting oxidative stress [17,44]. Another study reported that o-hydroxyhippuric acid induces ROS formation in renal tubular cells, inhibiting cell proliferation [45]. In this study, HA also showed prooxidant properties in cells and rats. Furthermore, the ROS elimination by antioxidants (NAC and DPI) significantly reversed HA-induced fibrotic marker expression in vitro, confirming that ROS play a direct role in HA-induced fibrotic responses. Numerous studies have reported that ROS, as intracellular second messengers, can promote organ and tissue fibrosis by activating redox-sensitive pathways, such as TGFβ/SMAD signaling [46,47], which is responsible for the initiation of fibrosis progression and activation of downstream fibrotic genes [48,49]. Here, antioxidants (NAC and DPI) significantly prevented HA-induced activation of TGFβ/SMAD signaling, suggesting that oxidative stress is the primary contributing factor to the HA-induced activation of TGFβ/SMAD signaling. Concurrent with our results, IS and PCS have been shown to activate TGFβ/SMAD signaling via increased oxidative stress [18]. Hence, our results indicate that HA promotes renal fibrosis in CKD by activating ROS-mediated TGFβ/SMAD signaling.
Oxidative stress occurs upon excessive ROS accumulation. The NRF2-driven antioxidant network can eliminate accumulated ROS, thereby preventing the initiation of oxidative stress [50]. Previous studies of PBUTs showed that the accumulation of IS, PCS, and indole-3-acetic acid is positively correlated with decreased NRF2 levels in patients with CKD [51]. Our results showed that HA suppressed the antioxidant network by downregulating NRF2 protein expression, indicating that HA attenuates ROS elimination and aggravates the redox imbalance. Concurrent with our results, another study reported that IS decreased the renal expression of NRF2 protein, leading to ROS accumulation [52]. Several studies have suggested that NRF2 is a critical target for renal failure. Cui et al. [53] and Shin et al. [54] emphasized that SFN attenuates cyclosporin A-induced nephrotoxicity and diabetic nephropathy by activating the antioxidative function of NRF2. Yoh et al. [55] demonstrated that knockdown of NRF2 aggravates oxidative stress and renal disease progression in diabetic mice. However, the association between PBUT-promoted renal fibrosis and loss of the NRF2-driven antioxidant system remains largely unclear. Through NRF2 gain-and loss-of-function experiments with SFN and NRF2-specific siRNA, respectively, we found that HA-induced fibrotic toxicity was inhibited by SFN and aggravated by NRF2 knockdown in vitro. Similarly, in vivo, SFN alleviated HA-promoted renal fibrosis. These consequences suggest that the dysfunction of the Nrf2 antioxidant defense network, caused by downregulation of NRF2, plays a significant role in HA-promoted renal fibrosis.
Currently, the mechanism underlying PBUT-mediated NRF2 downregulation remains unclear. Numerous studies have reported that the reduction in NRF2 levels is mediated by protein ubiquitination via the E3 ubiquitin ligase system [56,57]. In this system, KEAP1 is the substrate adaptor, CUL3 is the scaffold, and NRF2 is the substrate [58]. KEAP1 and CUL3 are negative regulators of NRF2 [58,59]. In this study, we found that HA accelerated NRF2 ubiquitination. However, KEAP1 and CUL3 expression levels were not significantly altered, suggesting that KEAP1 and CUL3 expression are not associated with HA-induced NRF2 ubiquitination. Two models of protein-protein interactions have been proposed to be responsible for E3 ubiquitin ligase activity [60]. The first is a hinge-latch model, suggesting that binding and dissociation between KEAP1 and NRF2 regulate the strength of ubiquitination, whereas the second model is a KEAP1-CUL3 dissociation model, suggesting that the association between KEAP1 and CUL3 regulates NRF2 ubiquitination. Herein, we found that HA strengthened the associations of KEAP1 with NRF2 and CUL3, indicating that HA increases E3 ligase activity by facilitating NRF2-KEAP1-CUL3 interactions. Concurrent with our results, Zhang et al. [61] reported that PAQRS strengthens NRF2 binding to KEAP1 to enhance E3 ubiquitin ligase activity, thus increasing the ubiquitination and degradation of NRF2. Villeneuve et al. [24] revealed that USP15 causes KEAP1 incorporation in the E3 ligase complex more efficiently, thereby increasing E3 ligase activity. Furthermore, enhanced protein-protein interactions in the E3 ligase complex potentially prevent NRF2 dissociation upon stimulation by ROS or electrophiles, ultimately preventing activation of the antioxidant system. Numerous studies have focused on how disturbing the interactions between Keap1 with Nrf2 or Cul3 result in the dissociation of NRF2 and inactivation of E3 ubiquitin ligase. However, our study revealed the opposite results, contributing to further improvements in relevant models. Overall, these findings indicate that HA increased E3 ligase activity by enhancing NRF2-KEAP1-CUL3 interactions, resulting in degradation of ubiquitinated NRF2.
Accumulating evidence indicates that KEAP1 is a direct site for small-molecule binding to the E3 ubiquitin ligase complex [59]. KEAP1 is a core component in the regulation of NRF2 ubiquitination and serves at least three functions by acting as a scaffold to anchor NRF2 in cytoplasm, as a sensor responding to oxidative/electrophilic stimulation, and as a substrate adaptor of E3 ubiquitin ligase to incorporate NRF2 into the KEAP1-CUL3 complex [61][62][63]. In this study, computer simulations indicated that HA directly docks into the active site of KEAP1 by interacting with the Tyr334 and Arg415 residues, which agrees with a previous report [64]. These results suggest that KEAP1 is a direct HA target responsible for HA-induced NRF2 ubiquitination. However, further studies are required to verify these results.

Conclusions
This study revealed that HA promotes the progression of renal fibrosis by disrupting redox homeostasis, which is maintained by the NRF2 antioxidant network. This negative impact is attributed to the ability of HA to induce NRF2 ubiquitination and degradation by enhancing NRF2-KEAP1-CUL3 interactions. These findings provide insight into the underlying mechanisms of PBUT-promoted renal fibrosis in CKD and suggest that the antioxidant system plays a key role in this mechanism. Furthermore, targeting the NRF2-driven antioxidant system may be an effective therapeutic strategy for preventing renal fibrosis progression and alleviating CKD deterioration.

Conflicts of Interest:
The authors declare no conflict of interest.