A network pharmacological analysis was applied to study related genes in treating Salvia miltiorrhiza on AKI. We searched the keyword “Danshen” in the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP http://tcmspw.com/tcmsp.php) (Ru et al., 2014) to obtain active ingredients and their ADME (absorption, distribution, metabolism, and excretion) information. OB (oral bioavailability) and DL (drug-likeness) were recorded. The criteria of OB ≥ 30% and DL ≥ 0.18 suggest that the ingredients with good absorption into blood were screened as potential active ingredients. Then, the corresponding target proteins of the active ingredients were collected from the TCMSP database, and corresponding target genes were analyzed via String database (Von Mering et al., 2003) UniProt databases (The UniProt Consortium, 2018). The target genes of AKI were obtained from the GeneCards (https://www.genecards.org/), OMIM (https://omim.org/), and DisGeNET database (https://www.disgenet.org/). We used the Venn online analysis tool (https://www.omicshare.com/) to study the intersection target genes of Salvia miltiorrhiza and AKI. These genes are considered potential targets of Salvia miltiorrhiza for treating AKI. With those intersection target genes, we performed GO and KEGG pathway enrichment analyses by using the ClueGO plug-in of Cytoscape 3.6.0. The screening was based on p < 0.01 and a κ score ≥ 0.53 to visualize the results of GO and KEGG enrichment (Bindea et al., 2009). A flowchart of network pharmacology study is shown in Supplementary Figure S3. The original data in the network pharmacological analysis are available on Figshare (https://doi.org/10.6084/m9.figshare.17895506.v1)

TanIIA and cisplatin were HPLC-purified products purchased from Sigma-Aldrich. The chemical structure of TanIIA was constructed by ChemOffice (CambridgeSoft, Cambridge, MA) as shown in Figure 7A. TanIIA was dissolved in 150 μl saline with 0.01% DMSO. Male C57BL/6 mice at eight weeks of age (weighting 20–25 g) were housed under 23 ± 2°C, 55 ± 5% humidity, and 12/12-hour light/dark cycle environment with free access to water and food. The number of mice was calculated and assigned to groups following an allocation process (Charan and Kantharia, 2013; Bespalov et al., 2020). The mice were divided into 4 groups (n = 5): normal control (CTRL), AKI model, AKI+TanIIA12.5 (12.5 mg/kg/day, i.p.) low-dose treatment, and AKI+TanIIA25 (25 mg/kg/day, i.p.) high-dose treatment group. The doses were chosen based on previous publications (Jiang et al., 2015; Zhang et al., 2020). The AKI treatment groups received a single cisplatin injection (20 mg/kg, i.p.). CTRL received the same volume of control saline intraperitoneally. The body weight of all mice was recorded before and after 72 h of the cisplatin injection. The mice were killed on AKI day 3. Serum, kidneys, and other organs were collected after ultrasonic assessment for renal blood flow. The experimental protocol was approved by the Institutional Animal Care and Use Committee of Guangdong Provincial Hospital of Chinese Medicine (Approval No. 2020020).

Renal ultrasonography was performed to visualize renal blood vessels using a high-frequency ultrasound imaging system (Vevo 2100, Visual Sonics Inc., Toronto, Canada) as previously described (Okumura et al., 2018). All measurements were done on the right abdomen. Peak systolic and end-diastolic blood flow velocities (mm/s) were measured in the renal artery to obtain the pulsatility index (PI). An experienced technician who was blinded to the study protocol repeated the measurement three times. According to the manufacturer’s instructions, serum creatinine was measured with commercial kits (C011-2-1, Nanjing Jiancheng Bioengineering Institute, China).

Human proximal tubular cell line HK-2 was purchased from American Type Culture Collection (CRL-2190, ATCC, VA, United States). HK-2 cells were cultured in DMEM/F12 medium supplemented with 10% (v/v) FBS (Gbico, United States) under a humidified atmosphere with 5% CO₂ at 37°C. Cells were trypsinized and harvested when the confluence reached 80–90%.

Cytotoxicity activity was assayed by using a Cell Counting Kit-8 (CCK-8) (Dojindo, Japan) according to the manufacturer’s instructions. Dimethyl sulfoxide (DMSO) was used as the vehicle and did not exceed 0.1%. Each cell line would be seeded into a new flask and treated with 5% DMEM/F12 medium with DMSO (0.1% v/v) and TanIIA at different concentrations. HK-2 cells were, respectively, seeded into 96-well plates at a density of 5 × 10³ cells/well, and cultured in the DMEM/F12 with a low FBS concentration for 24 h. Then they were divided into control groups, DMEM/F12 with DMSO (0.1% v/v) (100 μl/well) and TanIIA group at various final concentrations of 3.125, 6.25, 12.5, 25, 50, and 100 μM (100 μl/well). After incubating for 6, 12, and 24 h, the medium was replaced with 10 μl 10% CCK-8 solution dissolved in 90 μl PBS, and all the cells were incubated for another 1h at 37°C in the dark. Absorbance at 450 nm was measured using a Sunrise™ Microplate Reader (TECAN, Switzerland). Each concentration was repeated five times in the same plate, and all experiments were performed at least three times. The percentage of cell viability was normalized to control group cells to calculate the inhibition rate: I n h i b i t i o n   r a t e   ( I R )   ( % ) = 1 − S a m p l e   s o l u t i o n   O D   v a l u e C o n t r o l   O D   v a l u e × 100 % . IC₅₀ was calculated with the Excel add-in tool ED50plus (created by Mario H. Vargas, MD).

In total, 1 × 10⁶ HK-2 cells were transfected with the hPXR small-interfering RNA (siRNA) or the control siRNA (Santa Cruz Biotechnology, CA, United States) using Lipofectamine 3000 transfection reagent (Thermo Fisher Scientific Inc., United States) according to the manufacturer’s instructions. The control siRNA was used as a negative control. RIF (rifampin) (Sigma-Aldrich, United States) (10 µM) was used as the positive control. Cells were treated with DMSO (0.1% v/v), RIF (10 µM), and TanIIA (5 µM) and were stimulated with/without TNF-α (20 ng/ml) for 24 h. The cells were then subjected to the NF-κB luciferase reporter assay.

The pSG5-hPXR expression plasmids and pGL3-CYP3A4-XREM luciferase reporter construct, containing the basal promoter (−10466/+53) with the proximal PXR response element (ER6) and the distal xenobiotic responsive enhancer module (XREM, −7836/−7208) of the CYP3A4 gene 5′-flanking regions inserted to pGL3-basic reporter vector, were kind gifts from Dr. Sridhar Mani (Albert Einstein College of Medicine, Bronx, NY, United States) (Huang et al., 2007). The pRL-TK Rotylenchulus reniformis control vector was purchased from Promega, Madison, United States.

HK-2 cells were seeded at 60–70% confluence into opaque 96-well plates at a density of 1 × 10⁵ cells/well in DMEM/F12 containing 10% FBS and cultured for 12 h. The cells were transiently transfected with pGL3-CYP3A4-XREM, pSG5-hPXR nuclear receptor expression vector, and pRL-TK control vector in a particular ratio with Lipofectamine 3000 transfection reagent according to the manufacturer’s instructions. After 6 h of incubation, the transfected HK-2 cells were exposed to TanIIA at 1, 2.5, 5, and 10 μM; RIF (10 μM); and control (0.1% DMSO) for an additional 24 h. Following the treatments, HK-2 cells were rinsed with PBS, and the luciferase activities were determined using a Dual-Luciferase Reporter Assay System (Promega, Madison, United States) and an Infinite F500 multimode microplate reader (TECAN, Switzerland).

HK-2 cells were transiently transfected with pGL4.32 [luc2P/NF-κB-RE/Hygro] NF-κB reporter, pCMV6-entry or pCMV6-XL4-hPXR, and pRL-TK (Promega, Madison, United States) with Lipofectamine 3000 transfection reagent according to the manufacturer’s instructions. After overnight transfection, the cells were treated as control (0.1% DMSO), TanIIA (5 μM), and RIF (10 μM) for 24 h, and additional incubation with/without TNF-α (20 ng/ml) for another 24 h. A standard dual luciferase assay was performed on cell lysates and detected using an Infinite F500 multimode microplate reader.

Transfection efficiency was expressed as fold induction of firefly to Renilla luciferase activities relative to control. Each concentration of the test substances was repeated five times in the same plate, and the transfection experiments were performed at least three times. Data are expressed as mean ± SEM from three independent experiments.

Tissues were fixed with 4% buffered paraformaldehyde for 48 h, and they were cut into 3-μm paraffin sections for hematoxylin and eosin (HE), periodic acid–Schiff (PAS) and Masson’s trichrome staining. The morphology of renal tubules with prominent necrosis were recorded. Patchy or diffuse denudation with loss of brush border detachment of cells with intratubular cast formation was identified. The degree of tubular injury was scored as score 0: no tubular injury; score 1: ≤10% tubules injured; score 2: 11–25% tubules injured; score 3: 26–50% tubules injured; score 4: 51–74% tubules injured; and score 5: ≥75% tubules injured.

Immunohistochemistry (IHC) sections were de-waxed and proceeded by a microwave-based antigen retrieval technique following previous procedures (Gu et al., 2018). Antibodies used in IHC included PCNA Rabbit mAb IKKβ Rabbit mAb (Cell Signaling Technology, MA, United States), anti-NF-κB p105/p50 Rabbit mAb (Abcam, MA, United States). The kidney sections were stained with primary and secondary antibodies. The sections were developed with diaminobenzidine (DAB) to produce a brown color. The percentage of positive staining areas of PCNA, NF-κB p105/p50, and IKKβ expression was measured by ImageJ software (NIH, Bethesda, MD, United States).

HK-2 cells were seeded, incubated, and treated with 0.1% DMSO, TanIIA (5 µM), and/or TNF-α (20 ng/ml) for 24 h. Cells were fixed with 4% paraformaldehyde for 30 min, permeabilized with 0.5% Triton X-100 for 20 min, and blocked with 5% BSA for 30 min at room temperature. Primary antibodies against NF-κB p65 (1:100, Cell Signaling Technology, MA, United States) were incubated with cells at 4°C overnight. After washing with PBS three times, an FITC-conjugated secondary antibody was added for 60 min at room temperature. All slides were mounted with DAPI-containing mounting medium and then analyzed with a fluorescence microscope (Leica Microsystems, Wetzlar, Germany)

Mouse serum was collected and centrifuged at 3,000 rpm for 15 min at 4°C. The level of neutrophil gelatinase-associated lipocalin (NGAL), IL-6, and TNF-α in blood was detected by using ELISA kits (R&D Systems Inc., Minneapolis, United States), following the manufacturer’s instructions.

Total RNA was isolated from the kidney tissue with the TRIzol reagent (Thermo Fisher Scientific Inc., United States) according to the manufacturer’s protocol. Total RNA was quantified by the absorbance ratio (260/280 nm) and reverse-transcribed into cDNA using PrimeScript™ RT reagent (Takara Bio Inc., Japan). All the PCR reactions were carried out using SYBR® Premix EX Taq™ II Kit (Takara Bio Inc., Japan) in 96-well optical plates on an ABI ViiA™ 7 PCR System (Applied Biosystems, Thermo Fisher Scientific, United States). The quantity of each transcript was calculated as described in the instrument manual and normalized to the housekeeping gene β-actin. The ratio of target to β-actin was calculated as ΔC_t=C_t (target)-C_t (β-actin), ratio (target) = 2^−ΔΔCt. The primers used for real-time PCR are listed in Supplementary Table S5.

The proteins from renal cortical tissues were extracted with RIPA lysis buffer. The samples were subjected to Western blot analysis with primary antibodies against phospho-NF-κB p65, NF-κB p65, IKKβ, NF-κB p105/p50 (Cell Signaling Technology, MA, United States), RXRα (Abcam, MA, United States), PXR, and GAPDH (Santa Cruz Biotechnology, CA, United States) overnight at 4°C and subsequently incubated with the conjugated secondary antibody of horseradish peroxidase-labeled anti-rabbit IgG or anti-mouse IgG (Cell Signaling Technology, MA, United States) at room temperature. Target protein expression was detected by using an Image Lab System (Bio-Rad Laboratories Inc., Hercules, CA, United States) and analyzed with ImageJ software.

The ligand and the receptor recognize each other and bind together by geometry and energy matching. Molecular docking provides new insights into the ligand–receptor interactions and structural features of the two molecules within the active site. To predict the protein–ligand interaction, we performed the CDOCKER algorithm and Libdock module in Discovery Studio (DS) 2.5 (Accelrys Software Inc., San Diego, United States) to evaluate the potential interaction of ligands and the proteins, respectively. The 3D crystal structure of TanIIA (CID: 164676) was obtained from the PubChemProject (http://pubchem.ncbi.nlm.nih.gov/compound/). The receptor PXR was selected from Protein Data Bank (PDB) (http://www.rcsb.org/pdb/). The program was run by using a localhost9943 server on Microsoft Windows 7.

The calculation of root mean square deviation (RMSD) was carried out. RMSD indicates the reliability of the results from the above modules. Docking procedures on DS 2.5 were as follows: the water molecules in the targeted protein PXR were removed, and the hydrogen atoms were added. Then, PXR and the ligand were refined with CHARMM. The active sites of PXR were defined automatically according to the internal ligand binding site through a series of algorithms by DS2.5. A three-dimensional pharmacophore model of the ligand with the targeted protein was generated in the CDOCKER module and Libdock module.

Data are expressed as mean ± SEM. All data were analyzed with GraphPad Prism 5 (GraphPad Software, San Diego, CA) by one-way analysis of variance (ANOVA) for single-variable analysis or two-way ANOVA for two independent variables, followed by Turkey’s multiple comparisons test. Student’s t-test was used for paired data.

Overall, 65 eligible active ingredients (Supplementary Table S1) and 139 target proteins (Supplementary Table S2) were collected from TCMSP. Among these proteins, 133 significant target genes were collected from the String and Uniprot database (Supplementary Table S3); 7123 AKI-related target genes were collected from GeneCards, OMIM, and DisGeNET databases; and 122 intersection genes (Supplementary Table S4, Supplementary Figure S1) of Salvia miltiorrhiza and AKI were screened out by the Venn analysis tool. A total of 15 KEGG pathways were enriched, as shown in Supplementary Figure S2. A total of 22 GO pathways were enriched. Figure 1A has indicated the pathways associated with Salvia miltiorrhiza treatment in AKI. The numbers of the target genes in the pathway were also highlighted. A pie chart was used to classify 22 GO pathways into 7 significant signaling pathways and percentage of genes related to AKI (as shown in Figure 1B): nuclear receptor activity (4.55%), extrinsic apoptotic signaling pathway (4.55%), response to alkaloid (13.64%), integral component of synaptic membrane (27.27%), regulation of blood vessel diameter (13.64%), regulation of smooth muscle cell proliferation (22.73%), and neurotransmitter biosynthetic process (13.64%) (Figure 1zc C).

Among these pathways, we put great interest in the process of nuclear receptor activity. The nuclear receptor family has shown a percentage of 19.3% association on Salvia miltiorrhiza treatment in AKI with 11 target genes involved (AKR1B1, AR, CYP1A1, ESR1, ESR2, NR1I2, NR3C1, PGR, PPARG, RXRA, and STAT3). Up to date, the underlying mechanisms of how nuclear receptors, especially PXR, affect kidney injury are still poorly understood. Following this clue, we tried to reveal the role of PXR in AKI in vivo and in vitro and seek an alternative possibility in developing Salvia miltiorrhiza as a renal protective drug.

On AKI day 3, the bodyweight of both AKI model and AKI+TanIIA25 (mg/kg/day) was significantly decreased compared to the CRTL group (Figure 2A). As shown in Figure 2B, serum creatinine levels were increased in all cisplatin-treated groups compared to CTRL. However, the creatinine level was reduced in both TanIIA treatment groups compared with the cisplatin-induced AKI model. The neutrophil gelatinase-associated lipocalin (NGAL) and kidney injury molecule-1 (KIM-1) are important biomarkers related to AKI (Schrezenmeier et al., 2017). Consistent with the creatinine results, the serum and mRNA levels of NGAL (Figures 2C,D) and mRNA of KIM-1 were significantly reduced by the TanIIA treatment compared to the AKI model (Figure 2E).

Morphological changes of renal damage were shown by HE and PAS staining. Tubular damages such as tubular dilation, atrophy, cast formation, and sloughing of TECs, or loss of the brush border in cisplatin-treated groups were observed (marked as arrowheads and asterisks). TanIIA reduced tubular necrosis, restored the brush border, and formed fewer casts in the tubule, therefore alleviating tubular injuries in a dose-dependent manner compared to the AKI group (Figure 3A). Masson’s trichrome staining indicated an improvement in fibrosis in TanIIA treatment groups, with less collagen formation within the renal interstitial area (Figure 3B). IHC staining has shown an increased staining area of PCNA (cell proliferation marker) in both TanIIA treatment groups compared to the AKI model (Figure 3C).

It is well-demonstrated that AKI may lead to endothelial dysfunction and impaired vascular autoregulation, which causes vasoconstriction and changes renal hemodynamics. Therefore, the reduction in medullary blood flow can be observed in AKI (Ozkok and Edelstein, 2014). Ultrahigh-frequency ultrasound is a non-invasive and sensitive method that helps monitor renal function. The color Doppler and pulsed wave Doppler images could be used to show renal blood supply and arterial flow velocity patterns (Kelahan et al., 2019). In the present study, renal ultrasonography was used to observe renal blood flow at AKI day 3. The upper panel in Figure 4A has shown the representative image, showing the anatomical relationship of renal vein and artery in normal mouse kidneys. The color Doppler-based sonographic assessment had shown an increase in the pulsatility index (PI) in the AKI model, while PI of the TanIIA treatment was decreased compared to the AKI group (Figure 4B). The results had indicated better renal perfusion in TanIIA-treated mice under AKI conditions.

To investigate how TanIIA affects renal inflammation, we hypothesize that TanIIA inhibits the activation of the NF-κB-activated inflammatory response. Therefore, the expressions of key mediators in NF-κB signaling were measured. The expression of NF-κB p105/p50 and IKKβ was significantly reduced by TanIIA treatment when compared to the AKI model (Figures 4C–E), and the results were confirmed by the NF-κB p105/p50 and IKKβ protein expression measured by Western blot (Figures 4K–M). Similarly, serum levels of pro-inflammatory cytokines in kidneys, such as TNF-α and IL-6, mRNA level of IL-1β, and TGF-β were also declined in TanIIA treatment groups compared to the AKI model (Figures 4F–J). These findings had suggested an anti-inflammatory effect of TanIIA in AKI by inhibiting the production of pro-inflammatory cytokines and the activation of NF-κB signaling.

We studied the mechanisms of TanIIA that affected the PXR/NF-κB signaling pathway in AKI mouse and human TEC cell line HK-2. The expression level of mouse PXR and RXRα was significantly decreased in the kidneys by cisplatin treatment (Figures 5A–C). TanIIA treatment significantly induced the expression of mouse PXR, coactivator RXRα, and expression of cytochrome P3A11 (Cyp3a11, mouse homolog of CYP3A4). On the other hand, TanIIA suppressed the activation of NF-κB signaling by inhibiting the phosphorylation of NF-κB p65 (Figure 5D). Moreover, reduced mRNA expression of mouse PXR and downstream Cyp3a11 were also observed in AKI mouse kidneys, while TanIIA treatment upregulated the mRNA expression of mouse PXR and Cyp3a11 significantly compared to the AKI model (Figures 5E,F).

A transfection luciferase reporter assay was performed in HK-2 cells to evaluate the effect of TanIIA on human PXR and PXR-mediated CYP3A4 transcription activity. The optimal concentration for TanIIA was chosen by cell viability assay. The IC₅₀ values for TanIIA in HK-2 cells were 17.16 ± 1.83 μM (6 h) (Figure 6A). HK-2 cells were transiently transfected with pGL3-CYP3A4-XREM, pSG5-hPXR nuclear receptor expression vector and pRL-TK control vector to investigate whether TanIIA could transcriptionally activate human PXR-mediated CYP3A4 expression. A strong hPXR agonist, rifampin (RIF = 10 µM), was used as the positive control. Cells were incubated with/without TanIIA at different concentrations (TanIIA = 1, 2.5, 5, 10 µM). TanIIA activated human PXR and induced CYP3A4 promoter activity (Figure 6B). We confirmed the optimal concentration of TanIIA (5 µM) to the activation of human PXR.

To explore the role of human PXR on the repression of NF-κB by TanIIA, HK-2 cells were transfected with hPXR small-interfering RNA (siRNA) or the control siRNA. pGL4.32 [luc2P/NF-κB-RE/Hygro], pCMV6-entry or pCMV6-XL4-hPXR, and pRL-TK control vector were transfected into the cells. HK-2 cells were stimulated by TNF-α (20 ng/ml) to activate NF-κB signaling. As shown in Figure 6C, the pro-inflammatory cytokine TNF-α had significantly induced the luciferase activity of NF-κB. When compared to the groups treated with TNF-α alone, TanIIA significantly reduced TNF-α-stimulated NF-κB luciferase expression. In cells treated with TanIIA and stimulated by TNF-α (TNF-α+ TanIIA), NF-κB luciferase activities were significantly reduced in cells cotransfected with human PXR plasmid (pCMV6-XL4-hPXR). The result had suggested TanIIA repressed PXR-mediated NF-κB activity. Besides, in HK-2 cells stimulated by TNF-α, immunofluorescence staining also revealed an inhibitory effect of TanIIA treatment on nuclear translocation of NF-κB p65, suggesting TanIIA alleviated renal inflammation by repressing the activation of NF-κB signaling (Figure 6D).

RMSD values are mainly used for analyzing the stability of protein and predicting conformational changes of the protein. RMSD values depend upon the binding interaction and energy between the protein and ligand. The optimized protein has the lowest RMSD values (Carugo and Pongor, 2001). Generally, an RMSD value less than 1Å represents good reproduction of the correct pose. The molecular interactions like hydrogen bonds (H-bonds), hydrophobic interactions (such as π–π interaction and electrostatic interaction), Van der Waals interaction, and ionic bonds are also playing an important role in assessing the reliability of results and the binding stability between two modules (Patil et al., 2010).

TanIIA (chemical and 3D structure are shown in Figures 7A,B) could be docked into the active sites (binding pocket with LBD) of PXR (PDB ID: 1SKX). Inside the binding pocket, PXR consists of an endogenous ligand that had been confirmed as a PXR agonist. This ligand was used as a reference confirmation to compare docked TanIIA (Figure 7C). The RMSD of PXR (PDB ID: 1SKX) was 0.6720 Å (Table 1). Therefore, CDOCKER program was performed to analyze the interactions and binding features of PXR and TanIIA. Inside the binding pocket, TanIIA was surrounded by amino acids of the PXR protein (1SKX). Amino acids SER²⁴⁷ and GLN²⁸⁵ have formed H-bonds, and PHE²⁸⁸ and TRP²⁹⁹ formed π–π interactions between TanIIA and PXR (Figure 7D).