PO (potassium oxante, 98.0%), HX (hypoxanthine, 99%), allopurinol (98%), and benzbromarone (98%) were brought from Aladdin Reagent Co. (Shanghai, China). Cordycepin (99.5%) was obtained from Target Molecule Corp. (Boston, USA). TRIZOL reagent was offered by Invitrogen Co. (USA). By Nanjing Jian-Cheng Bioengineering Institute (Nanjing, China), Uric acid assay kits were supplied. BUN and creatinine Kits were purchased from Mindray Medical Corp. (Shenzhen, China). From R&D System Inc. (USA), XOD and URAT1 Elisa Kits were obtained.

All animal experiments were approved by and performed in Guangdong Institute of Microbiology (approved ID: GT-IACUC20170228; Guangzhou, China). From the Guangdong Provincial Medical Laboratory Animal Centre (Guangzhou, China), Kunming mice (20 ± 2 g) were brought. Mice were allowed to have food and water freely for adapting the laboratory conditions before experiment for 1 week. Primarily, mice were separated into normal, hyperuricemic, allopurinol, and benzbromarone controls and cordycepin groups of 15, 30, and 60 mg/kg, respectively.

The protocol reported by us (Yong et al., 2016) was exploited for model establishment with allopurinol and benzbromarone as positive controls. Specifically, mice were dosed with PO (100 mg/kg) intraperitoneally and HX (600 mg/kg) orally simultaneously for models. Meanwhile, normal control mice were dosed with physiological saline (0.9%) at the same time.

For drug administration, animals were treated at 1 h after model building at the frequency at once per day for 7 days. Allopurinol and benzbromarone controls were drugged at 5 and 7.8 mg/kg correspondingly. For cordycepin groups, mice were dosed at a 15, 30, 60, mg/kg. Using physiological saline (0.9%), normal and hyperuricemic controls were treated at the same time.

To test uric acid, BUN, and creatinine according to the manufactures' protocols, serum, and urine were gathered. For XOD activity and URAT1 protein assay by ELISA kits, liver and kidney were collected and tested following the manufactures' protocols.

Liver, kidney, and spleen were washed with saline (0.9%) and sucked with normal filters, followed by weighting. Organ coefficients were computed by dividing the weight of organ by that of corresponding mouse.

Total RNA extractions were performed using TRIZOL reagent. After homogenation of kidney tissue, the obtained liquid was added with chloroform and centrifuged, followed by precipitating aqueous phase with volume of isopropanol. After washed by ethanol (75%), the total RNA pellets were suspended using DEPC water. Using RNA (1 μg) together with M-MLV reverse transcriptase, Reverse transcription was conducted. The obtained cDNA was diluted with DNase free water and PCR amplification was performed using primers at appropriate conditions (Table 1) with GAPDH as external standard. Finally, products were quantified by electrophoresis.

After washing with PBS for three times, kidney cortexes were homogenized with 10-folds of RIPA Lysis Buffer (adding 1 mM PMSF; protease inhibitor) in ice bath. Following bathed on ice for 30 min, the mixtures were centrifuged (12000 g, 10 min). The supernatants were gotten as the total proteins, determined by BCA Protein Assay Kit (Tiangen Biotech Co., China). Before electrophoresis, the total proteins were denatured by incubating for 5 min in boiling water. Samples (5 μg) were separated by 10% SDS-PAGE. And then onto PVDF membrane (Millipore, USA) they were electrophoretically transferred. Non-specific binding sites of obtained membranes were blocked in TBST (Tris-buffered saline with 0.1% Tween-20) mixed with 5% skimmed milk powder. After that, they were then incubated overnight individually with specific antibodies (Table 2) diluted in TBST, consisting of URAT1 (1:2,000), GLUT9 (1:2,000), OAT1 (1:2,000), and GAPDH (1:4,000). After washed for three times with TBST, they were incubated with HRP-conjugated goat anti-rabbit IgG (1:3,000) as the secondary antibody diluted in TBST for 30 min. Consequently, they were washed for three times with TBST and then mixed with ECL (Enhanced Chemiluminescence, Servicebio Co., China). Following that, they were exposed to X-ray film. The contents of target protein were analyzed via densitometry using Alpha Innotech (AlphaEaseShop, USA) and normalized by the respective blotting from GAPDH.

Data were analyzed by ANOVA and showed as mean ± standard error. Significance of difference were at P < 0.05 or P < 0.01, compared by two-tailed Student's t-tests.

The molecular docking modeling study was performed with the CDOCKER (Wu et al., 2003). The structure of URAT1 was homology modeled previously (Yong et al., 2016). The binding pocket was determined by alignment with its template. The 3D structures were down from PubChem Compound (ID: 6303) and optimized through energy minimization with the CHARMm force field. Docking calculations were performed using the default parameters.

Simulation was performed with GROMACS 5.0.4 (Hess et al., 2008; Wennberg et al., 2015). The docked structure of URAT1 with cordycepin was utilized as the start. Geometric optimization and electrostatic potential of cordycepin were calculated at the DFT/6-31G^* level. The corresponding topology files of the ligands were generated by the Automated Topology Builder (ATB) server with the charge distribution calculated by DFT/6-31G^* level quantum method. The Charmm36 force field was used for URAT1, which were imbedded in POPC membrane and then solvated using the TIP3P water model. To salt and neutralize charges, 0.15 M NaCl were added. Then, the built system was minimized and then equilibrated with cononical (NVT) ensemble. Following that, a further equilibration simulation with the isobaric-isothermic (NPT) ensemble was conducted. The production was carried out in the NPT ensemble. van der Waals (vdW) and long-range electrostatic interactions were ignored beyond 1.2 nm.

MM-PBSA (Kumari et al., 2014) was used for binding free energy calculation (ΔG_binding, Equation 1).

The G-value (G_x) can be represented as Equation (2).

where E_mm represents the molecular mechanics energy, TS denotes the entropic contribution, (T, temperature; S, entropy) and G_solvation represents the solvation free energy.

Molecular mechanics energy E_mm includes the energy of bonded (E_bonded), electrostatic (E_{electrostatic}), and van der Waals (E_vdW) interactions (Equation 3).

G_solvation includes electrostatic (G_polar) and nonelectrostatic (G_apolar) solvation free energy (Equation 4).

G_apolar was calculated with SASA model (Kumari et al., 2014).

Snapshots extracted from MD trajectory at intervals of 4 ns were used for calculation of MM-PBSA binding free energy. All calculations were performed by use of the g_mmpbsa package developed for GROMACS (Kumari et al., 2014).

As shown in Figure 2A, potassium oxonate and hypoxanthine were further demonstrated to induce hyperuricemia in mice successfully, elevating SUA from normal control (202 μmol/L) to 337 μmol/L for hyperuricemic control (P < 0.01). The treatment of allopurinol (5 mg/kg, positive control) in hyperuricemic mice caused significantly declines in SUA to 78 μmol/L (P < 0.01), much lower than normal control (P < 0.01). For benzbromarone at 7.8 mg/kg, it evoked a significant decline in SUA to 226 μmol/L (P < 0.01), nearing normal control. Remarkably, cordycepin produced decreases in SUA to 216, 210, and 203 μmol/L (P < 0.01) at 15, 30, and 60 mg/kg correspondingly, closing to normal control.

To elucidate the possibilities that the reductions of SUA by cordycepin were caused by the elevation of renal uric acid excretion, its effects on UUA were detected (Figure 2B). In hyperuricemic control, treatment of PO and HX combindely induced a uric acid excretion decline in mice (241 μmol/L) in comparison with normal control (371 μmol/L, P < 0.01). Furthermore, allopurinol induced a further decrease (215 μmol/L, P < 0.01), given rise by inhibiting XOD. In contrast, cordycepin at 15, 30, and 60 mg/kg effectively elevated UUA (303, 374, and 403 μmol/L for three doses, P < 0.01).

In order to estimate renal functions, BUN and creatinine were examined (Figures 2C,D). BUN for hyperuricemic control (16.61 mmol/L) was higher than that of normal control (11.96 mmol/L, P < 0.01), indicating some impairment by amount PO. BUN in allopurinol control (16.67 mmol/L, P < 0.01) was 139% of normal control and no significant difference was observed between allopurinol and hyperuricemic controls. Moreover, benzbromarone elevated that to 19.95 mmol/L (P < 0.01). In contrast to that, groups of cordycepin demonstrated BUN levels at 15.93, 15.82, and 16.27 mmol/L. Significant increases in serum creatinine were observed in hyperuricemic (61.6 μmol/L, P < 0.05) and allopurinol (61.2 μmol/L, P < 0.05) controls, contrasting to normal control (57.9 μmol/L, Figure 2D). However, cordycepin decreased that to 56.8, 52.3, and 55.7 μmol/L (P < 0.01) in comparison with hyperuricemic control, closing to normal control.

Mice body weight data were shown in Figure 3. After acclimatization for 1 week and the establishment of hyperuricemic for another 1 week, normal, hyperuricemic, and benzbromarone controls showed similar body weights. However, allopurinol significantly inhibited the weight growth (P < 0.05).

Organ coefficients were recorded for evaluation of effects of cordycepin on inner organ functions (Figure 4). Allopurinol increased liver coefficient in comparison with normal control (P < 0.05, Figure 4A). But cordycepin groups showed similarities to normal control. Kidney coefficients for all groups were in neighborhood (Figure 4B). Spleen coefficients for hyperuricemic control (0.40%) were higher than normal control (0.34%, P < 0.01, Figure 4C). Allopurinol (0.38%) and benzbromarone (0.45%) controls and cordycepin at three doses (0.47, 0.43, and 0.44%) showed similarity to hyperuricmic control in that.

XOD activities in hyperuricemic control were ascertained to be 118% fold (14.31 U/L) of normal mice (12.17 U/L, P < 0.05, Figure 5A). In contrast to hyperuricemic control, allopurinol control (5 mg/kg) reduced liver XOD activities to 12.33 U/L significantly (P < 0.05), closing to that of normal control (12.17 U/L). The none XOD inhibitor, benzbromarone, had not impact on XOD activity (14.39 U/L) comparing with that of hyperuricemic group. Cordycepin at 15, 30, 60 mg/kg showed similar or even enhanced activities to 14.14, 15.94, and 17.54 U/L (P > 0.05, P < 0.05, and P < 0.05) respectively, comparing to hyperuricemic control.

Affections of cordycepin on URAT1 levels of renal organ were examined by Elisa test primarily (Figure 5B). Renal URAT1 of hyperuricemic (14.56 pg/mL) and allopurinol (14.80 pg/mL) groups increased significantly (P < 0.01) when compared them with normal control (10.96 pg/mL). The URAT1 levels in hyperuricemic mice treated by cordycepin at various doses were 11.48, 11.05, and 8.98 pg/mL and all was much lower than that of hyperuricemic control (14.56 pg/mL, P < 0.01).

To further examine the mechanisms of anti-hyperuricemic actions of cordycepin, its affections for GLUT9, URAT1, and OAT1 mRNAs and proteins were examined (Figure 6). For hyperuricemic control, PO down-regulated GLUT9 mRNA (P < 0.01) and protein (P < 0.05, Figures 6A–C). It also up-regulated URATl and OAT1 mRNA but down-regulated their proteins. Allopurinol and benzbromarone regulated down the GLUT9, URAT1, and OAT1 mRNA expressions and GLUT9 and URAT1 proteins compared with hyperuricemic group. Significantly, 15, 30, and 60 mg/kg cordycepin significantly decreased the URATl mRNA and protein in a dose-dependent pattern. Thirty and sixty milligrams/kilograms cordycepin elevated the GLUT9 and OAT1 protein expressions in kidney, respectively, compared with hyperuricemic control. According to the above, cordycepin may interacts mainly with URAT1 to lower serum uric acid levels. Thus, we tried a computational protocol, including docking, molecular dynamics, and energy calculations, to gather an insight into the binding mode of cordycepin and gain some structural demands on cordycepin inhibitory.

Figure 7 shows the initial pose of the cordycepin in the active pocket. Figure 7a depicts the two dimensional diagram of the docked cordycepin-protein complex. Apparently, cordycepin bind with the tunnel shaped pocket surrounded by CYS21, ARG22, TYR85, MET147, PHE174, PHE293, and LEU415 via three hydrogen bonds specifically to MET147, ARG22, and TYP85. Wherein, hydrogen donor of the amine on nucleoside of cordycepin bonds with sulfur atom of MET147. Hydrogen and oxygen atoms of two hydroxyl groups of sugar ring bind with oxygen atom of hydroxyl group of TYP85 and hydrogen atom of amine group of ARG22, respectively.

In simulation, the RMSDs for the protein backbone and the binding pocket were monitored (Figure 8). Accordingly, RMSD for the binding pocket was fluctuant initially and stable after 6 ns. Both RMSDs for protein backbone and binding pocket converged to equilibrium during the last 5 ns. Hence, 500 snapshots extracted from the 5 ns were utilized for structural and energy analysis.

The hydrogen-bond occupancies were computed to examine the binding behaviors of cordycepin (Table 3). Significant hydrogen bond forces were observed for cordycepin with URAT1, where GLY411 as the acceptor for cordycepin occupied up to 11.6% and TRP290 as the donor up to 29.9%.

The energy distributions of binding free energy were depicted (Table 4), where the prediction (ΔG_predict = −83.324 ± 1.122 kJ/mol) supported the experiment result of binding. Overall, the remarkable nonpolar actions (ΔG_nopolar = ΔG_vdw + ΔG_sa; −127.33 kJ/mol) implied the key roles of the hydrophobic residues in pocket for binding; conversely, the polar term (ΔG_polar = ΔG_ele + ΔG_pb; 43.99 kJ/mol) was opposite for the integration of URAT1-cordycepin complex. The polar solvation free energies (ΔG_pb) exhibited an adverse effect on binding.

From the averaged structure (Figure 9A), the hydroxymethyl group on the sugar moiety and amine group on the nucleoside part of cordycepin could form HBonds to TRP200 and GLY411, respectively, which made a stable interaction between cordycepin and URAT1. In aim to screen the pivotal residues, energy decomposition was conducted to gain energy contributions of each residue. Obviously, the interaction spectra (Figure 9B) showed that the highlighted favorable residues were TRP290, ARG410, ARG420, GLY411, MET147, ARG17, ALA408, LEU415, PHE293, GLY412, MET148, MET407, and LYS78 at the binding pocket. Consistent with the binding free energy analysis, most of them were hydrophobic, implying for the key role of hydrophobic interaction. Of which, the residues TRP290, ARG17, ALA408, GLY411, and MET147 had important roles in binding, which mainly reflected that the nonpolar interactions (Figure 9C) were driving the complexing process with cordycepin. Besides that, the favorable polar interactions were generated by ARG420 and ARG410, LYS78, and ARG398 (Figure 9D). Thus, the residues ARG420, ARG410, LYS78, and ARG398 contributed strong polar forces to stabilize the URAT1-cordycepin complex, and TRP290, LEU404, GLY411, and ARG17 had an adverse effect on the polar binding.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.