To explore the possible impact of GTW on spermatogenesis and reproductive function, four groups of 4-week-old mice were administered intragastrically with GTW suspension at doses of 7.5 mg/kg/day (group 1), 15 mg/kg/day (group 2), 22.5 mg/kg/day (group 3) and 45 mg/kg/day (group 4). Group 5 is control. Pregnancy rates of all five groups, including control (group 5) were examined after treatment for 40 days. Pregnancy rates decreased with increasing doses of GTW treatment. At 45 mg/kg/day, the pregnancy rate even dropped to zero, which indicates a strong detrimental effect of GTW on fertility (data were shown by He et al. ^[21]).

There were no significant changes in body size and weight of mice after treatment with low dose (7.5 mg/kg/day) GTW. However, the hair of the mice became withered and vertical after high dose (45 mg/kg/day) GTW treatment for 2 weeks. Control animals appeared normal. Testes of the treated mice were separated and examined morphologically. Size of testes decreased with increasing GTW dose (Figure 2A (a)). Histological analysis observed that the number of spermatozoa in the seminiferous tubules decreased, and this phenomenon was even more obvious in the high dose treatment (Figure 2A (b) – (f)). Many cavities were visible in group 4, and the arrangement of germ cells was disordered especially in group 4.

To analyze the influences of GTW in vitro, SSCs were isolated from 6-day-old mice. Morphological profiles of SSCs after cultivation for 12 h are shown (Figure 2B (a)). In accordance with previous studies ^[22], ^[23], serum containing GTW was added into the SSC medium at the concentration of 10% or 20%. SSCs were then cultured in these media for 12 h. After 12 h, some cell fragments began to appear, but most SSCs were relatively intact in morphology (Figure 2B (c) – (d)).

Using an in situ Apoptosis Detection Kit, apoptotic cells were stained brown with diaminobenzidine. There were few apoptotic cells in group 1, which was similar to the control group (Figure 3A (b)). However, in group 2, apoptotic cells significantly increased (Figure 3A (c)). When the dose of GTW was increased to 22.5 mg/kg/day (group 3), apoptosis-related cavities began to appear (Figure 3A (d)). For animals exposed to 45 mg/kg/day (group 4), although seminiferous tubules were pathologically atrophied and the majority of germ cells was depleted, there still were some cells that showed apoptotic traits among those that survived (Figure 3A (e)). We randomly chose 20 fields of view of each slide and calculated the ratio of apoptotic (Figure 3C (a)).

Noticeably, most of the apoptotic cells were distributed near basement membrane of the seminiferous tubules, where early stage germ cells including SSCs are located. Previous observations in vitro have demonstrated the great morphological impact of GTW on these cells. To find out if GTW induced SSC apoptosis in a direct manner, SSCs isolated from wild-type mice were cultured with serums containing different concentrations of GTW. The apoptosis rate of SSCs was analyzed after 12 h treatment (Figure 3B (a) – (c)). After 12 h, the number of apoptotic cells in the 10% GTW group was lower than in the 20% GTW group. However, they were both much higher than in the controls (Figure 3C (d)), which indicates an elevated level of GTW-induced apoptosis in SSCs.

To explore the temporal expression pattern of H3K9me2, testes of different ages in untreated mice were analyzed by immunofluorescence. Expression levels of dimethylated H3K9 was the highest in adult mice. However, it decreased gradually with increasing age, and there were few H3K9 positive cells in 360-day-old mice (Figure S1). Compared with controls, a few H3K9 dimethylated positive cells moved from the edge towards the lumen in groups 1–3 (7.5 mg/kg/day to 22.5 mg/kg/day), while there was no significant difference among the three groups (Figure 4B). When the dose was further increased, dimethylated cells decreased significantly (P<0.05). In group 4, although cells were almost destroyed, there were still a few scattered methylated positive cells in the tubules (Figure 4A).

Similarly, we focused on the dimethylated state of H3K9me2 in SSCs. Gonocytes transformed into SSCs between 0 and 6 days postpartum ^[24], ^[25]. Protein from testes of 6-day-old mice was analyzed by western blotting. H3K9me2 was expressed in SSCs (Figure S2). Furthermore, SSCs were treated for 12 h with different concentrations of GTW or Triptolide, after which dimethylation levels of H3K9me2 were determined. The number of H3K9me2 positive cells in treated groups was lower than in the controls, and there was a significant difference between control and higher dosage groups (Figure 5, Figure S3, P<0.05), which is in agreement with the in vivo results.

Triptolide is one of the primary active compounds in the GTW and was used to investigate the regulation mode. 5.0 ns molecular dynamics simulations were performed for the complex of triptolide and EED. To examine the variations in the intramolecular conformations of EED, the root-mean-squared deviation (rmsd) with respect to the initial structure was calculated. Simulation time versus rmsd of the backbone of the protein during the full simulation (5.0 ns of molecular dynamics) is presented in Figure 6A. The rmsd variations of EED model were about 1.3 Å. The results show that the complex became dynamically equilibrated after 3 ns of simulation.

The principal hydrophobic and hydrogen bonding interactions for H3K9-EED from the crystal structure and triptolide-EED from the nearest conformers to their average conformers were identified with Ligplot ^[26] and are represented in Figure 6B and 6C, respectively. For H3K9-EED, there were two important hydrogen bonds between the oxygen (O) atom in H3K9 and the nitrogen atom in residue Arg414(O…NH1), with a distance of 2.99 Å, and between the oxygen (O) atom in H3K9 and the nitrogen atom in residue Arg414 (O…NH2) with a distance of 2.74 Å. For triptolide-EED, there was one hydrogen bond between the oxygen (O22) atom in triptolide and the nitrogen atom in residue Arg414 (O…NH1), with a distance of 3.08 Å. Figure 6D (a) illustrates the variation of distance between O22 of GTW and NH1 of Arg414. The average distance of O22…NH1 (Arg414) is about 3.0 Å. In addition, there was another margin hydrogen bond between O22 of triptolide and oxygen atom in residue Trp364. This suggests that there are two hydrogen bonds for H3K9 and triptolide. Furthermore, there was a common hydrogen bond with Arg414 for methylated H3K9 and triptolide.

The hydrophobic contacts were defined by a distance of less than 6.5 Å between the hydrophobic center of EED and the ligand ^[27], ^[28]. The hydrophobic contacts for H3K9-EED and triptolide-EED are also shown in Figure 6B and 6C. Four hydrophobic interactions can be found for H3K9-EED: H3K9/Trp364, H3K9/Tyr365, H3K9/Tyr148, and H3K9/Phe97. For triptolide-EED, two stable hydrophobic contacts can be found in Figure 6D: triptolide/Trp364, and triptolide/Ile363. The hydrophobic contact of triptolide/Phe97 is rather weak. Comparing with H3K9, we found that there is also a common hydrophobic contact with Trp364 for H3K9 and triptolide.

Figure 6E illustrates the superposition of H3K9 and triptolide within an EED hydrophobic pocket. The dimethyl of triptolide and methylated H3K9 are mostly aligned well. It demonstrates that triptolide and H3K9 might have similar interaction mechanisms with EED. That is, Arg414 and Trp364 are key residues for triptolide and H3K9 effective binding.

BALB/c mice and CD1 mice used in this study were purchased from SLAC Laboratory Animal Co. LTD, Shanghai China. Mice were housed in the specific pathogen free (SPF) facility at the Laboratory Animal Center, Shanghai University of Traditional Chinese Medicine. Four-week-old males were administered with GTW by gavage. Six-day-old males were used for the isolation of SSCs. All procedures involving animals were approved by the Institutional Animal Care and Use Committee of Shanghai, and were conducted in accordance with the National Research Council Guide for Care and Use of Laboratory Animals [SYXK (Shanghai 2007-0025)].

GTW was purchased from Shanghai Fudan Fuhua Pharmaceutical Industry Company Limited. Four-week-old BALB/c male mice were randomly divided into 5 groups of 10 animals in each. In groups 1–4, GTW was given by gastric gavage for 40 days. Group 1 received GTW 7.5 mg/kg/day; group 2 received GTW 15 mg/kg/day; group 3 received GTW 22.5 mg/kg/day; group 4 received GTW 45 mg/kg/day; group 5 was fed with distilled water instead of GTW for control.

Wild-type BALB/c male mice were administered with 45 mg/kg/day GTW or 17 µg/kg Triptolide ^[41] for 3 days. Two hours after the intragastric administration, blood was collected and centrifuged at 13,400×g for 15 min. Supernatant was precipitated at 4°C overnight and the serum was carefully moved to a new Eppendorf tube. Later the concentration of Tripterygium wilfordii and Triptolide in the serum was dected with spectrophotomer. After filtering for sterilization, the serum was stored at −20°C for further study.

Testes from mice were fixed with 4% paraformaldehyde (PFA) and 0.2% glutaraldehyde in PBS overnight at 4°C. After being dehydrated through 50%, 70%, 85%, 95% and 100% ethanol, the testes were vitrificated in xylene and later embedded in paraffin. Sections of 6 µm thickness were mounted on slides, de-waxed with xylene and stained with haematoxylin for histological analysis. Pictures were taken using a Nikon Eclipse E600 microscope equipped with a Nikon DXM 1200 digital camera (Tokyo, Japan).

For immunohistochemistry, tissue sections were deparaffinized, equilibrated in phosphate-buffered saline (PBS) for 10 min, digested by 0.05% trypsin at 37°C for 15 min, and then washed with PBS and blocked in 10% normal goat serum at room temperature for 10 min. Slides were subsequently incubated with primary antibodies at 4°C overnight (mouse anti-H3K9me2; 1∶100; Abcam). Sections were washed with PBS before incubation with fluorescein isothiocyanate (FITC) conjugated secondary antibody (goat anti-mouse immunoglobulin G; 1∶150; Invitrogen) for 30 min at 37°C in darkness to detect H3K9me2. Sections were then washed with PBS, and incubated in DAPI (1∶10000, Sigma) for 18 min at 37°C in darkness.

After treatment with serum containing GTW, SSCs were fixed with 4% PFA in PBS for 20 min at room temperature, washed with PBS and incubated in 3% H₂O₂ for 15 min at room temperature. Finally, SSCs were blocked with goat serum and incubated with mAb of mouse anti-H3K9me2 (1∶100) and polyantibody of rabbit anti-PLZF (1∶50) for 1 h, washed by PBS, then incubated with secondary antibody (goat anti-mouse IgG conjugated TRITC, 1∶150; goat anti-rabbit IgG conjugated FITC, 1∶150, Abcam) for 30 min at 37°C in darkness, stained with DAPI (1∶10000) for 20 min, obscured under the fluorescence microscopic after washing by PBS three times.

CD1 mice at 6 days of age were used as donor mice for establishing SSC cultures. The two-step enzymatic digestion of Nagano et al. and Wu et al. was used ^[42]–^[48]. Seminiferous tubules were collected from decapsulated testes and dissociated enzymatically in D-Hanks buffer containing 1 mg/ml collagenase (type IV, Sigma) at 37°C for 15 min. Then the testicular tissue was digested by 0.05% trypsin prepared with D-Hanks containing 1 mM EDTA at 37°C for 10 min. After neutralization by DMEM with 10% fetal bovine serum (FBS), lysates were centrifuged at 300×g for 5 min, the precipitated cells were mixed with BD™ IMaganti-mouse CD 90.2 (Thy-1.2) particles and purified according to instructions of the kit. Finally, the isolated cells were transferred onto STO cell feeder layers (mouse SIM embryonic fibroblasts, strain SIM, 5×10⁴ cells/cm², ATCC). The composition of germ cells medium was as follows: minimum essential medium α medium (MEM-α) supplemented with 15% FBS, 1 mM sodium pyruvate, 1 mM non-essential amino acids, 2 mM L-glutamine, 0.1 mM β-mercaptoethanol (Sigma), 10 ng/ml LIF (Santa Cruz Biotechnology) and 15 mg/l penicillin. SSCs were cultivated at 37°C for 12 h under 5% CO₂, and then treated with different doses of GTW or Triptolide for 12 h.

For protein analyses, testes were isolated from 6-day-old or adult mice and homogenized on ice with lysis buffer containing protease and phosphatase inhibitors (1% Triton X-100, 50 mM Tris–HCl, 5 mM EDTA, 1% aprotin and 1% leupeptin). The cell lysates were centrifuged (13400 g/min for 30 min), and total protein was qualified by the BCA Protein Assay Kit (Pierce, Rockford, IL), then separated by 8% SDS-PAGE and transferred to nitrocellulose membrane. The expression of H3K9me2 was probed by western blotting with mouse anti-H3K9me2 primary antibody (1∶100; Chemicon). Secondary detection included incubation with HRP conjugated goat anti-mouse IgG antibody (1∶20,000; Amersham Pharmacia Biotech). Signals were detected using ECL western blotting Kit (Amersham Pharmacia Biotech).

The samples were analyzed by terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay using in situ Apoptosis Detection Kit (TaKaRa, Japan) according to the instructions.

Molecular modeling was used to investigate the mechanism behind the decrease in methylated levels of H3K9 by triptolide. The complex of EED (embryonic ectoderm development) and methylated H3K9 was extracted from Brookhaven Protein Databank (PDB code: 3K27) ^[49]. The structure of triptolide is shown in Figure 1. The structure was optimized using Tripos force field. All calculations were performed on the in-house Xeon (1.86GHz) cluster.

The active site was determined based on the co-crystallized H3K9/EED complex. First, the methylated H3K9 was extracted from the EED complex, and then the remaining EED structure was completed by adding the missing polar hydrogen atoms. This structure was optimized with constrained molecular dynamics. Partial atom charges of EED were calculated with Kollman-all-atom approximation ^[50]. Gasteiger-Hückel charges were calculated for tripterygium glycoside, as recommended in the AutoDock 3.0 package, which was used for performing automated docking of inhibitors to EED. The development and the principle of this software have been described elsewhere ^[51].

MD simulations and energy minimizations were performed using the AMBER8.0 simulation package ^[52] and the parm99 force field ^[53] with the TIP3P water model ^[54] The initial coordinates of the complex for triptolide and EED were extracted from Autodock. Hydrogen atoms were added using the LEAP module of AMBER8.0. Antechamber ^[55] was used to handle the force field of triptolide. Ten Cl ions were placed around the complex to maintain the system's neutrality. The SHAKE algorithm ^[56] was used to constrain bonds involving hydrogen atoms. The complex was solvated in a rectangular box of water, with the shortest distance between any protein atom and the edge of the box approximately 10 Å. Particle Mesh Ewald (PME) ^[57] was employed to calculate long-range electrostatic interactions. Then the complex was minimized using the PMEMD module of AMBER8.0. This minimization consisted of 1000 steps with the steepest descent method. The systems were equilibrated at 298 K for 20 ps. After 20 ps equilibration, MD simulation was employed to record time trajectory. The complex was simulated for 5.0 ns at 298 K. The time step used in all calculations was 2.0 fs. Coordinates were saved every 1 ps for the purpose of subsequent analysis.

All data are expressed as the mean ± standard error of the mean (SEM). Results were analyzed by Chi-square test. Data between groups were compared using partition analysis. P-values less than 0.05 (P<0.05) were considered significant.