Critical Role of Hepatic Cyp450s in the Testis-Specific Toxicity of (5R)-5-Hydroxytriptolide in C57BL/6 Mice

Low solubility, tissue accumulation, and toxicity are chief obstacles to developing triptolide derivatives, so a better understanding of the pharmacokinetics and toxicity of triptolide derivatives will help with these limitations. To address this, we studied pharmacokinetics and toxicity of (5R)-5-hydroxytriptolide (LLDT-8), a novel triptolide derivative immunosuppressant in a conditional knockout (KO) mouse model with liver-specific deletion of CYP450 reductase. Compared to wild type (WT) mice, after LLDT-8 treatment, KO mice suffered severe testicular toxicity (decreased testicular weight, spermatocytes apoptosis) unlike WT mice. Moreover, KO mice had greater LLDT-8 exposure as confirmed with elevated AUC and Cmax, increased drug half-life, and greater tissue distribution. γ-H2AX, a marker of meiosis process, its localization and protein level in testis showed a distinct meiosis block induced by LLDT-8. RNA polymerase II (Pol II), an essential factor for RNA storage and synapsis in spermatogenesis, decreased in testes of KO mice after LLDT-8 treatment. Germ-cell line based assays confirmed that LLDT-8 selectively inhibited Pol II in spermatocyte-like cells. Importantly, the analysis of androgen receptor (AR) related genes showed that LLDT-8 did not change AR-related signaling in testes. Thus, hepatic CYP450s were responsible for in vivo metabolism and clearance of LLDT-8 and aggravated testicular injury may be due to increased LLDT-8 exposure in testis and subsequent Pol II reduction.

Lack of aqueous solubility, general toxicity, and tissue accumulation limited the clinical use of triptolide (Ye et al., 2010;Sun et al., 2013;Zhuang et al., 2013;Kong et al., 2015;Ma et al., 2015;Patil et al., 2015;Qi et al., 2015;Wang et al., 2016aWang et al., ,b, 2017Ruan et al., 2017). Overexposure and subsequent toxicity of triptolide lead to the clinical trial termination of F6008, a prodrug of triptolide (Fidler et al., 2003;Kitzen et al., 2009). While the incomplete cleavage, loss of drug bioactivity and interpatient variability stopped the clinical trials of other prodrugs of triptolide, PG490-88, and omtriptolide (Kitzen et al., 2009). Therefore, clarifying the connection between toxicity and pharmacokinetics of triptolide and its derivatives may help us overcome some limitations of triptolide and allow the clinical use of triptolide and its derivatives .
Compared to triptolide (Xue et al., 2011), LLDT-8 has a better safety profile and does not induce abnormalities in the epididymis, liver, kidney, spleen, or circulation (Qi et al., 2016). The testicular injury is the main adverse effect of LLDT-8 in rodents, and recently we reported that spermatocytes are the primary target for LLDT-8 in the testes. Dephosphorylating TGFβ activated kinase (Tak1) Ser412 contributes to this selectivity (Qi et al., 2016). However, the interaction between toxicity and pharmacokinetics of LLDT-8 remained unknown.
The functional redundancy of CYP450 makes it difficult to determine the isoforms involved in drug metabolism and toxicity. Wu et al. developed a liver-specific knockout mouse model of cytochrome P450 reductase (Cpr), the sole electron donor of CYP450s, to overcome these limitations and reduced almost 95% hepatic CYP450 activity (Wu et al., 2003). Cpr knockout inhibited the hepatic P450-dependent metabolism of monocrotaline, aristolochic acid, and triptolide (Xiao et al., 2008;Xue et al., 2011;Yao et al., 2014).
Here, this study compared the toxicity and tissue distribution of LLDT-8 between Cpr knockout and wildtype mice. Inactivation of hepatic cyp450s increased the exposure of LLDT-8 and blocked meiosis in the testes by selectively downregulating γ-H2AX and RNA polymerase II in spermatocytes.

Animal Treatments
Animal use protocols were approved by the Institutional Animal Care and Use Committee of the Shanghai Institute of Materia Medica (Shanghai, China), IACUC No. 2016-10-RJ-136. The Cpr knockout (KO) mice were a gift from Professor Xinxin Ding (Wadsworth Center, Albany, NY, USA). Procedures for animal breeding and genotyping were reported previously . Eight to twelve weeks old male KO mice and their WT littermates on a mixed C57BL/6 and 129/Sv genetic background were used in this study. All animals were provided with a certified standard diet and tap water ad libitum during experiments. All animals were maintained under controlled temperature with a 12 h:12 h light/dark cycle. All efforts were made to minimize animal discomfort and illness, and mice were anesthetized with pentobarbital sodium (150 mg/kg, i.p.) before sample collection. No animals died during the experiment.

Toxicological Study
The LD 50 of LLDT-8 was 9.3 mg/kg (p.o.) in mice (Zhou et al., 2005), and we used a more clinically relevant dose in our animal experiments. For toxicity experiments, mice (n = 3/group) were treated with LLDT-8 (0.5 or 1.0 mg/kg) consecutively administered by gavage for 15 days. Controls received saline. Mice were killed on the 15th-day post-administration and blood, liver, kidney, spleen, testes, and epididymis were collected. About 500 µL blood samples were collected from the heart. The main lobe of the liver, kidney, spleen, and epididymis was fixed in 10% neutral buffered formalin for histological examination, and the left testicle was fixed in modified Davidson's buffer (30% of a 37-40% solution of formaldehyde, 15% ethanol, 5% glacial acetic acid, and 50% distilled H 2 O) for 16 h followed by 10% neutral buffered formalin for 24 h (Latendresse et al., 2002). Tissue sections were stained with hematoxylin and eosin (H&E). Remaining tissues were stored at −80 • C for RNA and protein extraction.

Toxicokinetics of LLDT-8 in Mice
For toxicokinetic experiments, mice (n = 5-6/group for each time point) were treated with a single dose of LLDT-8 at (0.5 or 1.0 mg/kg) by oral gavage, creating 4 treatment groups: 0.5 mg/kg LLDT-8-treated WT and KO mice, and 1.0 mg/kg LLDT-8treated WT and KO mice. After LLDT-8 treatment, 20 µL blood samples were collected at 5,10,20,30,45,60,90,120, and 180 min from tail vein. Plasma was separated by centrifugation at 3,000 rpm for 5 min and kept at −80 • C until analysis. Tissues including the liver, kidney, testes, and epididymis were also collected from individual mice at 30 and 180 min after dosing and were weighed and homogenized in saline (1.0 g wet weight/mL) on ice. LLDT-8 was then extracted from the plasma or the tissue homogenates with an equal volume of ethyl acetate and dried under nitrogen. The residues were reconstituted in 50 µL of mobile phase for analysis.
The quantification of LLDT-8 in samples was performed on Shimadzu 20A HPLC system (Shimadzu, Kyoto, Japan) equipped with an autosampler coupled to Shimadzu 8030 triple quadrupole mass spectrometer (Shimadzu, Kyoto, Japan). Separations were conducted under isocratic conditions. The mobile phase consisting of acetonitrile and water with 0.1% formic acid (50:50, v/v) was set at a flow rate of 0.2 mL/min. An electrospray interface in negative ionization mode was used. ESI source parameters were as followed: high purity dryinggas (N2) flow rate 8 L/min, temperature 400 • C, nebulizer pressure 25 psi. Multiple reaction monitoring (MRM) was used to quantify  [M+COO]-→ 45.10, fragmentor 110 eV, collision energy −20 eV). Analytical data were processed using the labsolution software package (Shimadzu, Kyoto, Japan) consisting of qualitative and quantitative software.
Standard curves for LLDT-8 were prepared by spiking known amounts of the LLDT-8 standard into plasma or tissue homogenate samples prepared from untreated mice. LLDT-8 concentrations in biological samples were determined by comparisons with standard curves. Pharmacokinetics were calculated using Kinetica software (version 4.4.1; Thermo Fisher Scientific Inc, Woburn, MA).

Metabolic Profile of LLDT-8
Metabolic profiles of LLDT-8 in liver microsomes from WT and KO mice were compared. In brief, 1 mM NADPH (Cat. No: 10107824001, Lot. No: 20595625, Sigma, USA), 5 mM MgCl 2 , 10 µL mouse liver microsomes (20 mg/mL, microsomes were prepared from the liver of mouse pretreated with 80 mg/kg dexamethasone for 3 days), 4 µL LLDT-8 (12.5-50 µM) in a 100 mM phosphate buffer (pH 7.4). There was a 3 min preincubation step at 37 • C before initiating the reaction by adding the NADPH into the microsomal suspension. The reaction was stopped with ice-cold acetonitrile after 60 min incubation. Identification of LLDT-8 metabolites was performed on an Agilent 1200 HPLC system (Agilent technologies Inc, Palo Alto, CA) equipped with a CTC PAL autosampler coupled to an API4000 QTRAP LC-MS/MS system (Thermo Fisher Scientific, USA). Separations were conducted under gradient conditions. The mobile phase consisting of 0.1% formic acid in acetonitrile and 0.1% formic acid in water (50:50, v/v) was set at a flow rate of 0.6 mL/min. An electrospray interface in negative ionization mode was used. ESI source parameters were as followed: high purity drying-gas (N 2 ) flow rate 8 L/min, temperature 550 • C, and nebulizer pressure 25 psi. Single ion monitoring with EMS/EPI scan mode was used to quantify LLDT-8 (positive: m/z 377.1, collision energy 60 eV; Negative: m/z 375.1, collision energy 60 eV; negative: m/z 375.1), and mono-hydroxylated LLDT-8 was selected with EPI scan mode (CE: −45 v, CES: −15).

Immunofluorescence
For immunofluorescence analysis, the paraffin-embedded testis tissue samples were deparaffinized three times in exchanges wash of xylene. The testis sections were then gradually rehydrated using decreasing ethanol concentrations followed by PBS (pH = 7.4). After antigen retrieval in boiling 0.1M citrate buffer (pH = 6.0), the tissue sections were placed in humidity chambers and incubated with 4% BSA in PBS for 2 h at room temperature.

Statistical Analysis
Statistical analysis was performed using two-tailed Student's t-tests or one-way ANOVA followed by LSD's post hoc test (p < 0.05 was considered statistically significant).

Knockout of Hepatic Cpr Aggravated Testicular Injury Induced by LLDT-8
The dosage of LLDT-8 (0.5-1 mg/kg) in this study was clinically relevant. Chronic administration of LLDT-8 did not alter the body weight of WT and KO mice, but dosedependently reduced the testes (Figures 1A,B). The epididymis weight was not changed by LLDT-8 treatment LLDT-8 treatment ( Figure 1C). Compared to WT mice, the testes in KO mice showed reduced germinal cell layers (Figures 1D,E, Star), severe vacuolar degeneration (Figure 1E, asterisk) and absence of spermatids development (Figure 1E, arrow). TUNEL assay found the number of TUNEL positive foci in KO mice testis were much higher than that in WT mice (Figures 2A,B) after LLDT-8 weight after LLDT-8 treatment in WT mice, consistent with another experiments using C57BL/6 mice (Supplementary Figure  S1A). Noticeably, compared to WT mice with saline, TUNELpositive foci in the testes of WT mice with LLDT-8 treatment apparently increased (Figures 2A,C). LLDT-8 also induced hematological change and enlarged spleen in KO mice (Supplementary Figure S2). LLDT-8 did not cause liver injury and kidney damage (Supplementary Figures S3, S4). Inactivation of Hepatic CYP450 Increased the Exposure of LLDT-8 WT and KO mice were treated with a single dose of LLDT-8 at 0.5/1.0 mg/kg by oral gavage. There is a significant increase in half-life (1.9 folds to WT mice), C max (1.9 folds to WT mice) and AUC (3.1 folds to WT mice) of LLDT-8 in KO mice, accompanied by a marked decrease in the clearance, compared to those in WT mice (Figures 3A,B; Table 2). LLDT-8 levels in liver, kidney, testis, and epididymis of KO mice were about 2-3 folds higher than those of WT mice at 30 min following a dose of LLDT-8 treatment at 1.0 mg/kg ( Figure 3C).
Hepatic microsomes isolated from the liver of WT and KO mice were used to identify the potential metabolites of LLDT-8. Based on our previous findings on triptolide (Xue et al., 2011), we hypothesized hydroxylation is an elimination route of LLDT-8 in vivo. After 60 min-incubation, the recovered samples were subjected to API400 QTRAP LC-MS/MS system, single ion monitoring (SIM) with EMS/EPI scan mode was used to identify LLDT-8 (Negative: m/z 375.1), mono-hydroxylated LLDT-8 (Negative: m/z 391.1). Two mono-hydroxylated metabolites M1 and M2 (m/z = 391.1) were identified in the microsomal samples from WT mice, but not in those of KO mice (Figures 3D-F). The response of mass spectrometric response of M1 and M2 increased with the concentration of LLDT-8 (Table 3). There were no di-hydroxylated metabolites of LLDT-8 (m/z = 407.1) were identified in both WT and KO mice.

LLDT-8 Decreased the Expression and Changed the Distribution of γ-H2AX, a Marker of Meiosis
To clarify the potential mechanisms of LLDT-8 induced testicular injury in KO mice, we examined the expression and distribution   Table 3. Inhibitory effects of CYP inhibitors on the metabolism of LLDT-8 in rat liver microsomes (G). LLDT-8 (2 µM) was incubated with rat liver microsomes for 0.5 and 1 h, the residual LLDT-8 was analyzed by LC-MS/MS, mean ± SD, all incubations were carried out in three independent experiments in triplicate. # p < 0.05, ## p < 0.01, ### p < 0.001 vs. 0 h. of γ-H2AX, a marker of meiosis and XY body formation in the testes of mice (Hamer et al., 2003;Noguchi et al., 2008;Ahmed et al., 2013). In WT and KO mice with saline, the nuclei distribution of γ-H2AX indicated the presence of leptotene or zygotene spermatocytes, characterized by the ringlike property of seminiferous tubules, which was reduced in WT mice or even disappeared in the testes of KO mice after LLDT-8 treatment (Figures 4A,B). The XY bodies indicated by γ-H2AX foci (cloud like, yellow arrow) were normal in the testes of WT and KO mice with saline, while decreased in the testes of WT mice with 1.0 mg/kg LLDT-8 and disappeared in the testes of KO mice treated with LLDT-8 (Figures 4E,F).
Immunoblotting also confirmed the reduction of γ-H2AX protein level in the testes but not in the liver (Figures 4C,D).
These results strongly suggested that accumulated LLDT-8 may induce a severe meiotic block in the testes of KO mice. We also detected other germ cell markers in the testes of WT and KO mice (Supplementary Figures S5, S6) and found a marked decrease of Dazl (a marker for type B spermatogonia and primary spermatocytes), Prm1 (a marker for spermatids) and a mild decrease of Hspa2 in the testes of WT mice (Supplementary Figure S5).

LLDT-8 Selectively Decreased RNA Polymerase II of Spermatocytes
In spermatogenesis, the formation of RNA granule and specific transcription activation at some genomic foci (including histone, piRNA.etc) are entirely dependent on RNA Polymerase II (Pol II) (Voronina et al., 2011;Pandey and Pillai, 2014). In KO mice, LLDT-8 decreased Pol II protein in the testes and did not change its level in the livers (Figures 5A,B). In WT mice, the expression of Pol II in the testes and liver did not change significantly (Figures 5A,B). Three immortalized germ cell lines derived from mouse testes, spermatogonialike GC-1spg, Sertoli-like TM4, and spermatocyte-like GC-2spd were used to evaluate the effects of LLDT-8. LLDT-8 dose-and time-dependently reduced Pol II protein in GC-2spd cells, but not in GC-1spd and TM4 cell lines (Figures 5C-E).

LLDT-8 Did Not Lessen the Expression of AR-Related Genes in Sertoli Cells
Spermatogenesis is tightly regulated by androgen from Leydig cells and androgen receptor (AR) signaling in Sertoli cells (SCs) (Tan et al., 2005). Here, we selected AR-dependent two genes: Rhox5 and Cldn11 as the key indexs of ARrelated signaling in sertoli cells (Willems et al., 2010). We also selected AR-independent three genes: Cst12, ABP and FABP as the indexs of sertoli cell function (Johnston et al., 2004;Li et al., 2005;Tan et al., 2005). LLDT-8 treatment did not reduce the expression of all these genes including AR in the testes of WT mice (Figures 6A-F), while increased the expression of Cldn11, Cst12 and FABP, particularly in 1.0 mg/kg group.

DISCUSSION
Many efforts have been made to modify triptolide and overcome three major problems: low solubility, toxicity and tissue accumulation . One of the outcomes is LLDT-8, (5R)-5-hydroxytriptolide, which shows a favorable safety profile and is well-tolerated in Phase I and II clinical trials in a female population (Supplementary Figure S1). The testicular injury is the major adverse effects in male rodents and dogs (data not shown). As the sole triptolide derivative entering Phase II clinical trial, it's well deserved to clarify the potential contribution of the pharmacokinetics of LLDT-8 to its toxicity, which will be helpful to its Phase III clinical trial in various populations.
Compared to triptolide, LLDT-8 has better pharmacokinetics properties than those of triptolide: lengthened Tmax and non-cumulative property (Xue et al., 2011). Unlike triptolide (Xue et al., 2011), LLDT-8 did not induce hepatotoxicity in the liver of KO mice, while attenuating the liver injury caused by hepatic Cpr knockout (Supplementary Figure S3). Hepatic Cpr knockout inhibits Cyp51 activities, a key enzyme of cholesterol synthesis, interferes hepatic lipid metabolism and induces hepatocytes injuries (Lorbek et al., 2015). Thus, LLDT-8 has a wider therapeutic window. Potential mono-hydroxylated metabolites (m/z 391.1) were identified by single-ion monitor method with API4000 QTRAP LC-MS/MS. The samples recovered from the incubation of LLDT-8 in liver microsomes from WT and KO mice were subjected to LC-MS/MS. The retention time and peak area of each metabolite were summarized here. M1 and M2 were detected at three LLDT-8 concentration levels, and their mass spectrometric response levels (Metabolite peak area) increased with increasing LLDT-8 concentration (12.5-50 µM).
FIGURE 4 | LLDT-8 dose-dependently decreased γ-H2AX in testis but not in liver. The localization of γ-H2AX in the testes of WT (A,E) and KO (B,F) mice were shown (A,B ×20, Scale bar: 100 µm; E,F ×63, Scale bar: 20 µm); The protein level of γ-H2AX in the testes (C), liver (D) of WT and KO mice. Yellow arrow heads indicated leptotene or zygotene spermatocytes marked by the nuclei distribution of γ-H2AX, and the yellow arrows pointed the XY bodies in the process of meiosis. The protein level was quantified with ImageQuant software (C,D). Beta-Actin was used as a loading control. Significant difference was determined by one way ANOVA, mean ± SD, n = 3, *p<0.05, ***p<0.001 vs. Saline group (0 mg/kg).
Hydroxylation via CYP450s and the subsequent glucuronidation, sulfation, and glutathione conjugation is the primary phase I/II elimination pathways of triptolide (Du et al., 2011). We only found positive signal of potential mono-hydroxylated metabolites of LLDT-8 (m/z: 391.1) in WT mice, actively supporting the role of hepatic cyp450s in the FIGURE 5 | LLDT-8 decreased RNA Pol II in vivo and in vitro. The protein levels of Pol II were detected by western blot in the testes (A), liver (B), spermatogonia-like GC-1spg cells (C), sertoli-like cells (D) and spermatocyte-like cells (E). The protein level was quantified with ImageQuant software. Beta-Actin was used as a loading control. Significant difference was determined by one way ANOVA, mean ± SD, n = 3, *p < 0.05, **p < 0.01, ***p < 0.001 vs. control group. hydroxylation of LLDT-8. Under the context of inactivation of hepatic cyp450s, two pathways may be responsible for the clearance of LLDT-8 in KO mice. Firstly, glucuronidation et al phase II metabolism may be responsible for the elimination of LLDT08; Secondly, the remnant 5% the hepatic cyp450s activity via cytochrome b5 as electron donor may also promote the elimination of LLDT-8 in KO mice (Riddick et al., 2013). Hydroxylation decreased the bioactivity of triptolide (Li et al., 2009). It is reasonable to deduce that hydroxylation could subdue the activities of LLDT-8. Our recent work has excluded the potentiality of hydroxylation at position 6 and 20 on LLDT-8. More work is needed to confirm the exact hydroxylation sites on LLDT-8. Additionally, chemical inhibition assay showed that Cyp3a, Cyp2c, Cyp2d and Cyp2e1 may be the main CYP450 isoform responsible for the metabolism of LLDT-8 in the liver of rodents. Here, clomethiazole, an inhibitor of Cyp2e1 seems to completely inhibit the metabolism of LLDT-8, which may be due to its less inhibition on other Cyp450s including Cyp2c and Cyp3a (Stresser et al., 2016).
Serine 139 phosphorylation H2AX (γ-H2AX), a marker for DNA double-strand-breaks, distributes mainly at the leptotene, zygotene, pachytene, diplotene stages of spermatocytes, and indicates the meiotic process, synapsis and XY body formation (a γ-H2AX-positive foci) (Rogakou et al., 1998). RNA polymerase II is essential for the formation of RNA granules in spermatogenesis (Pandey and Pillai, 2014). In this study, LLDT-8 decreased the protein levels of γ-H2AX and RNA pol II in the testes of KO mice. It was a reasonable assumption that the reduction of γ-H2AX and Pol II in the testes of KO mice might be due to the loss of germ cells. Here, we observed a noticeable reduction of γ-H2AX by immunofluorescence, but no apparent loss of spermatocytes in many seminiferous tubules. Thus, we suggested that LLDT-8 may disrupt meiosis by decreasing Pol II and γ-H2AX in KO mice, then induce the spermatocytes apoptosis. In WT mice, immunoblotting did not find a statistically significant reduction of γ-H2AX and Pol II in the testes of WT mice, but the reduction of γ-H2AX in the testes of WT mice was up to 40%. As a derivative of triptolide, LLDT-8 may induced transcription arrest through two mechanisms: (1) LLDT-8 covalently binds XPB, inhibits its DNA-dependent ATPase activity (IC50: 2,900 nM), and leads to the inhibition of Pol II-mediated transcription (Titov et al., 2011); (2) LLDT-8 may trigger CDK7-dependent degradation of Pol II (Manzo et al., 2012). Thus, the reduction of γ-H2AX by LLDT-8 and potential transcription inhibition via XPB binding may mediate the mild but significant damage in the testes of WT mice. Additionally, unlike the normal spermatogenesis in vivo, GC-2spd and GC-1spg cell lines are absent of the meiosis process, and the genome is intact under normal culture condition. The increase of γ-H2AX induced by LLDT-8 in vitro may be due to the DNA damage caused by LLDT-8 (Supplementary Figure S7).
In our study, histology investigation did not find any abnormality in Leydig cells. Importantly, LLDT-8 treatment could not reduce the expression of AR-dependent and independent genes in sertoli cells. In special, Rhox5, a lead gene for searching SCs and a major androgen response gene, was not influenced by LLDT-8 treatment, strongly supporting that AR-related signaling may be not involved in the testis injury of LLDT-8 (Supplementary Figure S8). We used mouse testis derived GC-1spg (spermatogonia-like), TM4 (Sertoli cell-like) and GC-2spd (spermatocyte-like) in vitro. Some previous reports have shown that GC-2spd cells are more sensitive to apoptosis (McKee et al., 2006;Lizama et al., 2011;Qi et al., 2016). Here, the selective reduction of RNA Pol II by LLDT-8 in GC-2spd cells may partly explain this sensitivity difference. However, the detailed mechanisms for this selectivity are still unclear.
In summary, our results indicate that hepatic P450s inactivation can influence the pharmacokinetics and distribution of LLDT-8 in vivo and may increase the risk of LLDT-8-induced testicular toxicity. Inactivation or inhibition of P450s is often caused by genetic polymorphism and drugdrug interactions, which contribute to individual differences in xenobiotic metabolism and drug toxicity. Personalized prescription based on the blood concentration could be employed to maximize the therapeutic efficacy of LLDT-8 and reduce its adverse effects in the clinic use, especially in therapies that other drugs with P450 inhibitory property must be utilized along with LLDT-8.