The Natural Diterpenoid Isoforretin A Inhibits Thioredoxin-1 and Triggers Potent ROS-Mediated Antitumor Effects

Aberrant expression of thioredoxin 1 (Trx1) plays an important role in cancer initiation and progression and has gained attention as an anticancer drug target. Here we report that the recently discovered natural diterpenoid isoforretin A (IsoA) signi ﬁ cantly inhibits Trx1 activity and mediates anticancer effects in multiple preclinical settings. The inhibitory effect of IsoA was antagonized by free radical scavengers polyethylene glycol-catalase, polyethylene glycol superoxide dismutase, thiol-based antioxidants N -acetylcysteine and glutathione. Mass spectrometry analysis revealed that the mechanism of action was based on direct conjugation of IsoA to the Cys32/Cys35 residues of Trx1. This conjugation event attenuated reversible thiol reduction of Trx1, leading to ROS accumulation and a broader degradation of thiol redox homeostasis in cancer cells. Extending these in vitro ﬁ ndings, we documented that IsoA administration inhibited the growth of HepG2 tumors in a murine xenograft model of hepatocellular carcinoma. Taken together, our ﬁ ndings highlight IsoA as a potent bioactive inhibitor of Trx1 and a candidate anticancer natural product. Cancer Res; 77(4); 926 – 36. (cid:1) 2016 AACR.


Introduction
Thioredoxin 1 (Trx1; approved symbol: TXN; synonyms: TRX) is a critical antioxidant protein and participates in a wide range of cellular processes, such as cell proliferation, apoptosis, and aging (1,2).Trx1 contains the active site Cys-Gly-Pro-Cys, which reduces the target proteins by cysteine thiol-disulfide exchange (3).Upon oxidation, the two conserved cysteine residues in the active site form a disulfide bond, and the oxidized Trx (Trx-S 2 ) is then recycled to the reduced form Trx-(SH) 2 through the action of thioredoxin reductase and NADPH (4).
So far, several Trx1-specific inhibitors have been identified, such as semisynthetic unsymmetrical 2-imidazolyl disulfides, which interact with Trx1 as its substrates (3,19).Of note, 1-methylpropyl 2-imidazolyl disulfide (PX12) has been in clinical trials, but little success has been observed (20,21).Currently, no natural compounds have shown specific inhibitory effect on Trx1 by binding to Trx1.Therefore, it is of great value to explore novel natural Trx1-specific inhibitors for effective cancer therapy.
Isodon forrestii var.forrestii is a traditional Chinese medicinal herb of the Isodon species distributed in the Southwest China and used for centuries in China.The Isodon plants are rich in entkaurane diterpenoids, which attract considerable interest for their well-known anticancer activities (22)(23)(24)(25)(26)(27).IsoA is a novel entkaurane constituent isolated from the leaves of I. forrestii var.forrestii.The anticancer activity of IsoA has not been fully characterized.
In this study, we report that IsoA inhibited the growth of cancer cells both in vitro and in vivo.Mechanically, IsoA suppressed Trx1 activity through covalently binding to the Cys32/Cys35 residues of the activation sites of Trx1 and triggered ROS accumulation, resulting in DNA damage and apoptosis in cancer cells.Our work suggested, for the first time, that IsoA, as a Trx1 inhibitor, is selectively cytotoxic to cancer cells and has potential to be a novel agent for cancer therapy.

Cell lines and cultures
Human fetal lung fibroblasts HFL1, hepatocellular carcinoma (HepG2, BEL7402, and QGY7701), breast cancer (MCF7 and MDA-MB-231), cervical carcinoma (HeLa), lung cancer (A549), and colon cancer (Caco2) cell lines were purchased from Cell Bank of Shanghai Institute of Biochemistry and Cell Biology in June 2013.Human mammary epithelial cell line MCF 10A was kindly provided by Stem Cell Bank, Chinese Academy of Sciences in November 2015.Human osteosarcoma (U2OS), melanoma (A375), and normal hepatic (LO2) cell lines were purchased from Cell Bank of Kunming Institute of Zoology, Chinese Academy of Sciences in June 2013.All the cell lines were kept within 10 passages and preserved in liquid N2 after receipt.The used cells were resuscitated within 1 month.Cell lines were authenticated by the above cell banks through short tandem repeat (STR) analysis.HepG2, MCF7, MDA-MB-231, and Caco2 cells were cultured in MEM supplemented with 10% FBS.HeLa, A549, U2OS, A375, and LO2 cells were cultured in DMEM supplemented with 10% FBS.BEL7402 and QGY7701 cells were cultured in RPMI1640 containing 10% FBS.HFL1 cells were grown in F12K medium supplemented with 10% FBS.MCF 10A cells were cultured with MEGM Kit (Lonza/Clonetics, CC-3150) supplemented with cholera toxin (100 ng/mL) and 10% FBS.

Cell proliferation and apoptosis analysis
Cell proliferation and viability were determined by MTT assay and Trypan blue exclusion.Apoptotic cells were stained with Hoechst 33258, and also quantified using a FACScan laser flow cytometer (Guava easyCyteHT; Millipore) after staining with Annexin V and 7-AAD.

ROS determination
The level of cellular ROS was analyzed using DCFH-DA as a fluorescent probe.Cells were incubated with 10 mmol/L DCFH-DA for 30 minutes at 37 C in a 5% CO 2 humidified environment.The labeled cells were washed with PBS for three times.To quantify ROS, cells were harvested and the DCFH-DA fluorescence was measured using a FACScan laser flow cytometer (Guava easyCyteHT; Millipore).

Assessment of GSH levels
The intracellular levels of GSH were measured using GSH and GSSG Assay Kit (Beyotime).According to the manufacturer's instructions, the IsoA-treated cells were collected and homogenized.The protein concentrations were quantified using the BCA method.The total GSH levels were carefully measured by the enzymatic recycling method using glutathione reductase and 5 0 , 5 0 -dithio-bis (2-nitrobenzoic acid).The sulfhydryl group of GSH reacts with DTNB and produces a yellow-colored 5-thio-2-nitrobenzoic acid, which has an absorbance at 405 to 414 nm.Oxidized glutathione (GSSG) levels were accomplished firstly by first derivatizing GSH with 2-vinylpyridine.The concentrations of reduced GSH were calculated by subtracting the GSSG levels from the total GSH (GSH ¼ total GSH À 2 Â GSSG).The intracellular levels of GSH were determined on the basis of cellular protein concentrations.

Western blot analysis
For Western blot analysis, cells were lysed with RIPA to extract total proteins.Equal amounts of proteins were separated by SDS-PAGE and transferred onto a PVDF membrane (Millipore).Membranes were blocked with 5% nonfat milk at room temperature, and probed with primary antibodies at 4 C overnight, followed by IRDye-conjugated secondary antibodies.The membranes were scanned with an Odyssey infrared fluorescent scanner (LI-COR Biosciences).

8-Oxoguanine quantification by flow cytometry
The IsoA-treated cells were fixed in 1% paraformaldehyde and permeabilized by 0.5% Triton X-100 dissolved in PBS, followed by incubating with a FITC-conjugated 8-oxoguanine (8-oxoG) antibody at room temperature.Then, the cells were stained with propidium iodide to determine DNA content.Fluorescence was analyzed using a FACScan laser flow cytometer (Guava easyCyte HT; Millipore).
For in vitro Trx1 activity, this assay was performed using a modified Thioredoxin Activity Fluorescent Assay Kit (Caymen Chemical Company).The samples containing 0.02 mmol/L hTrx1 (the final concentration in the assay) were prepared according to the manufacturer's instructions.Next, 2 mL of IsoA at various concentrations (0.195-100 mmol/L) was added to each sample, followed by addition of 5 mL of b-NADPH to all samples and incubation at 37 C for 30 minutes.Finally, the fluorescent substrate was added, and the Trx1 activity was recorded as the emission at 518 nm after 488 nm excitation for 30 to 60 minutes, with a Thermo Scientific Varioskan Flash fluorescent microplate reader.

Determination of glutaredoxin activity in vitro
The in vitro Grx activity assay was performed using a modified Fluorescent Glutaredoxin Assay Kit (Caymen Chemical Company).The samples containing 1.5 nmol/L of hGrx-1 (the final concentration in the assay) were prepared according to the manufacturer's instructions.Next, 2 mL of IsoA at various concentrations (0.195-100 mmol/L) was added to each sample, followed by adding 10 mL of fluorescent substrate to each well, and recording the emission at 545 nm after excitation at 520 nm for 15 to 30 minutes with a Thermo Scientific Varioskan Flash fluorescent microplate reader.

Covalent docking
The crystal structure of human Trx1 was obtained from the protein data bank (PDB ID: 4PUF_C).The protein structure was prepared with Protein Preparation Tool (ProPrep) in the Schrodinger 2014-4 suit software.Hydrogen atoms were added at pH 7.0 by the PROPKA tool in Maestro with optimized hydrogenbond network.OPLS_2005 force field was used for restrained minimization with converge heavy atoms to 0.30 Å.The ligand IsoA was sketched in Maestro and subjected to Ligand Preparation Tool (LigPrep) using OPLS_2005 at pH 7.0 to generate low-energy conformation.Covalent docking was performed using Glide Docking module.The reactive residues Cys32 and Cys35 were supposed to form a covalent bond with the ketene of IsoA through Michael addition reaction.

MS/MS analysis
A stock solution (5 mg/mL) of human Trx1 protein (Sino Biological Inc.) was prepared in HPLC-grade water.Next, 4 mL of IsoA (40 mmol/L stock solution) was incubated with 140 mL of human Trx1 stock solution in 25 mmol/L Tris-HCl buffer solution (pH 7.4) for 2 hours at room temperature.In this reaction, the final concentrations of IsoA and Trx1 were about 80 mmol/L and 291 mmol/L, respectively.The samples were reduced according to the methods of Shimadzu Biotech Proteome Kit, trypsin digested by MonoSpin Trypsin, and desalted using a MonoSpin C18 column, before being analyzed using MALDI 7090 (SHIMADZU).

Murine models
All animal experiments were conducted under protocols approved by the Animal Care and Use Committee of Jiangsu Province Academy of Traditional Chinese Medicine.Nude BALB/c mice (13 AE 2 g) were raised in air-conditioned pathogen-free environment.HepG2 cells (2 Â 10 6 ) were injected subcutaneously into the right flank of nude mice.When tumors became palpable, the mice were randomized into two groups (n ¼ 6, per group), and treated with IsoA (15 mg/kg) or vehicle control for 14 days.The body weight and tumor size were measured every 3 days.Tumor volumes were calculated according to following formula: a 2 Â b Â 0.4, where a is the smallest diameter and b is the diameter perpendicular to a.At the end of the experiments, all animals were sacrificed and the tumors were excised, weighed, snap-frozen in liquid nitrogen, and stored at À80 C, or fixed in 4% paraformaldehyde for further analysis.
To explore the toxicity of IsoA to main organs, nude BALB/c mice (13 AE 2 g) were treated with IsoA (15 mg/kg) or vehicle control for 14 days (n ¼ 5, per group).At the end of the experiments, all animals were anesthetized using isoflurane (3%) inhalation, and blood samples were collected from the retro-orbital plexus.Complete blood counts were done on Auto Hematology Analyzer BC-5380 (Mindray) and plasma biochemical indices were analyzed on Cobas C311 (Roche Diagnostics).

Immunohistochemical staining and in situ terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling assay
The immunostaining assay was performed on tumor specimens that were fixed in 4% paraformaldehyde and embedded with paraffin.Tissue sections were deparaffinized and rehydrated, followed by antigen retrieval and endogenous peroxidase blocking.Primary antibodies against 8-oxoG and nitrotyrosine were added and incubated overnight at room temperature.Immunosignals were detected by the two-stage peroxidase-based EnVision (Dako) method.Apoptosis index in the tumor samples was assessed by in situ terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) analysis using the One Step TUNEL Apoptosis Assay Kit (Beyotime).

Statistical analysis
The difference between two different treatments was assessed by unpaired Student t test using PRISM software.ANOVA was used to compare multiple treatment groups and the nontreatment group.The intensity of the immune-reactive bands in Western blots was quantified by ImageJ software (NIH, Bethesda, MD).P < 0.05 was considered as statistically significant.
To investigate the antitumor potential of IsoA, 10 human cancer cell lines, including hepatic, breast, lung, colon, and cervical carcinoma, osteosarcoma, melanoma cells, along with nonmalignant human hepatic cell line LO2, fetal lung fibroblasts HFL1, and mammary epithelial cell line MCF 10A were tested for the growth-inhibitory effect of IsoA.Compared with the nonmalignant cells, IsoA displayed a preferential antiproliferative activity against cancer cells, with IC 50 values of <30 mmol/L (Fig. 1B; Supplementary Table S2).By clonogenic assay, we further found that IsoA drastically inhibited the colony formation of all tested cancer cells at 10 mmol/L (Supplementary Fig. S8).
Of all the tested cancer cell lines, HepG2 cells were the most sensitive to IsoA, with an IC 50 value of 15.83 AE 1.10 mmol/L.Thus, this cancer cell line was chosen as a model to further investigate the anticancer activity and the underlying mechanisms of IsoA.MTT assay demonstrated that IsoA reduced the viability of HepG2 cells in a concentration-dependent manner, but did not obviously alter the viability of LO2 cells even at 40 mmol/L (Fig. 1C).This was further verified by Trypan blue exclusion assay (Fig. 1D).Hoechst staining revealed that IsoA was able to induce chromatin condensation and fragmentation (Fig. 1E), suggesting induction of apoptosis.Flow cytometry analysis further showed that the pan-caspase inhibitor Z-VAD-FMK significantly attenuated IsoAinduced apoptosis (Fig. 1F).These data indicated that the IsoAinduced apoptosis was at least partially mediated by the activation of caspase cascade.

IsoA induced oxidative stress in cancer cells
Induction of ROS has been reported to preferably induce cancer cell apoptosis (29,30).Next, we therefore investigated the effects of IsoA on the induction of ROS using a fluorescent probe DCFH-DA.As shown in Fig. 2A and B, the induction of ROS was detected in HepG2 cells after exposure to IsoA.Similar results were obtained from another hepatocellular carcinoma cell line BEL7402 (Fig. 2C and D).To rule out the changes in ester cleavage, uptake, or efflux of DCFH-DA, we further detected IsoA-induced ROS in HepG2 cells using the oxidation insensitive analogue DCF as a control.IsoA treatment caused a remarkable shift in the fluorescence signal in cells loaded with DCFH-DA, but not with DCF (Supplementary Fig. S9).These results suggested that induction of ROS might play a critical role in inhibiting cancer cell growth.
Because GSH is an important antioxidant in cells to defend oxidant damage and regulate redox homeostasis (31), we next examined the association between intracellular GSH levels and the IsoA-induced ROS elevation.As shown in Fig. 2E, IsoA treatment of HepG2 cells resulted in a swift depletion of GSH compared with that of the control cells.Furthermore, treatment of HepG2 cells with IsoA for 24 hours also led to a dramatic depletion of reduced GSH (Fig. 2F).
The main cytotoxicity of excessive ROS is through DNA damage induced by base oxidation and double-strand breaks (DSB).Using a fluorescent antibody that recognizes 8-oxoG, we characterized and quantified DNA damage by flow cytometry.As shown in Fig. 2G, a significant increase of 8-oxoG was detected in IsoAtreated HepG2 cells.To examine whether the DNA damage involves DSBs, we performed Western blot analysis for the phosphorylated form of H2AX (gH2AX), a specific marker for DSBs (32).As shown in Fig. 2H and I, IsoA induced a dose-and timedependent expression of gH2AX in HepG2 cells.Taken together, our results suggested that ROS accumulation might be a general mechanism of IsoA in inhibiting cancer cell growth, and IsoA triggered both DSBs and oxidative DNA lesions in cancer cells.ROS scavengers attenuate IsoA's activity in HepG2 cells.A-D, HepG2 cells were pretreated with NAC (A), GSH (B), PEG-Catalase (C), and PEG-SOD (D) at indicated concentrations for 30 minutes, and then incubated with indicated concentrations of IsoA for 24 hours.The cell viability was determined by MTT assay.E, HepG2 cells were pretreated with 2 mmol/L of NAC for 30 minutes and then incubated with 15 mmol/L of IsoA for 16 hours, followed by apoptosis assay using Annexin V staining and flow cytometry.F, Quantification of Annexin V staining results.G, HepG2 cells were pretreated with 2 mmol/L of NAC for 30 minutes and then incubated with 15 mmol/L of IsoA for 1 hour.ROS levels were measured by flow cytometry and quantified.H, HepG2 cells were pretreated with 1,000 U of PEG-Catalase for 30 minutes, followed by incubating with 15 mmol/L of IsoA for 16 hours, and cell apoptosis was analyzed by flow cytometry and quantified.I, HepG2 cells were pretreated with 1,000 U of PEG-Catalase for 30 minutes, followed by incubating with 15 mmol/L of IsoA for 1 hour, and ROS levels were analyzed by flow cytometry and quantified.

Antioxidant alleviated IsoA-induced ROS accumulation and apoptosis
To further understand the contribution of ROS to IsoA-induced cell growth inhibition and apoptosis, we investigated whether the thiol-based antioxidant agents NAC or GSH could antagonize IsoA.For this, HepG2 cells were pretreated with NAC or GSH for 30 minutes, and then exposed to IsoA (0-30 mmol/L) for 24 hours, followed by MTT assay.NAC (2 mmol/L) or GSH (2 mmol/L) markedly attenuated the inhibitory effect of IsoA on cell viability (Fig. 3A and B).To further clarify whether ROS or some other oxidants were induced, we evaluated the effect of the powerful free radical scavenger PEG-Catalase or PEG-SOD on IsoA-induced cytotoxicity.The results showed that PEG-Catalase or PEG-SOD also antagonized IsoA-induced cytotoxicity in HepG2 cells (Fig. 3C and D).Our additional experiments further demonstrated that NAC and PEG-Catalase abrogated the IsoAinduced apoptosis and intracellular accumulation of ROS in HepG2 cells (Fig. 3E-H).Collectively, our results demonstrated that the IsoA-induced apoptosis was primarily induced by oxidative stress.
Mitochondria are known to be the major place of intracellular ROS generation during electron transport flow.To determine whether IsoA induction of ROS was through disrupting the oxidative metabolism of mitochondria, we generated A549 Rho0 cells, which lacked mitochondrial function due to depleting mitochondrial DNA.IsoA treatment induced similar levels of ROS in both normal A549 and A549 Rho0 cells (Supplementary Fig. S10A).In line with this, IsoA treatment also resulted in similar reduction of cell viability in both A549 and A549 Rho0 cells (Supplementary Fig. S10B).These results suggested that mitochondria should not be the main site of IsoA-induced ROS formation.

IsoA inhibited Trx1 activity
Chemical structure analysis revealed that IsoA contained an a, b-unsaturated carbonyl group, which can readily interact with biological molecules by forming covalent bonds with free thiol of cysteine or by acting as an electrophilic center in redox reactions.Hence, we speculated that IsoA might be a novel inhibitor of Trx1.For this, the levels of Trx1 in all cell lines were detected by Western blotting.As shown in Supplementary Fig. S11, Trx1 was extensively overexpressed in cancer cells relative to nonmalignant cells, which was consistent with previous reports (5 -10).Next, we investigated the inhibition potency of IsoA toward Trx1 using an in vitro Trx1 activity fluorescent assay.Our results demonstrated that IsoA effectively inhibited Trx1 activity with an IC 50 of approximately 5.177 mmol/L, whereas it had little inhibitory effect on Grx (Fig. 4A).To investigate the specificity of IsoA for Trx1, we analyzed the in vitro-inhibitory effects of IsoA on a spectrum of thiol-containing enzymes including glutathione reductase, thioredoxin reductase, glutathione-S-transferase, and protein disulfide isomerases.The results showed that IsoA inhibited thioredoxin reductase activity with an IC 50 of approximately 130.1 mmol/L, approximately 25 times of IC 50 for Trx1 (Supplementary Fig. S12A).In addition, 100 mmol/L of IsoA inhibited the activity of protein disulfide isomerases only by 10% (Supplementary Fig. S12B), and almost had no inhibitory effects on glutathione reductase and glutathione-S-transferase (Supplementary Fig. S12C and S12D).Using the insulin reduction assay, we found that IsoA suppressed Trx1 activity in cancer cells in a dose-dependent manner (Fig. 4B and C), despite no effects on the protein levels of Trx1 (Fig. 4D).These results indicate that IsoA is a potent inhibitor of Trx1.
Trx1 interacts with ASK1, and inhibits its kinase activity and ASK1-mediated apoptosis (13).ROS releases Trx1 from ASK1 partly through oxidizing Trx1 and activates ASK1/JNK death signaling cascade (33).To determine whether IsoA-mediated cell apoptosis is associated with the activation of ASK1 signaling, we examined the effect of IsoA on the phosphorylation levels of ASK1 and JNK.The Western blot analysis results showed that IsoA indeed increased the phosphorylation levels of ASK1 and JNK (Supplementary Fig. S13), suggesting that the activation of ASK1/JNK may contribute to IsoA-induced cell apoptosis.
To assess whether the effects of IsoA are Trx1 dependent, we used siRNAs to knockdown Trx1 expression.Trx1 protein level was reduced in the Trx1 knockdown cells (Fig. 4E and F), which rendered cancer cells more sensitive to IsoA (Fig. 4G).These results indicated that Trx1 levels influence cell sensitivity to IsoA.

IsoA covalently conjugated to Trx1 at Cys32 and Cys35 residues
To investigate whether IsoA directly interacts with the catalytic site of Trx1, we predicted the binding mode between IsoA and Trx1 through molecular docking.As shown in Fig. 5A and B, the docking poses of IsoA were almost identical to each other.The free sulfhydryl of reactive residues Cys32 and Cys35 acted as a nucleophile to attack a, b-unsaturated carbonyl group of IsoA, forming strong covalent bond between the receptor and the ligand through Michael addition in each complex.Meanwhile, the hydrogen bond formed between the acetyl of IsoA and Met74 also contributed to the stable binding interaction in both complexes.
To further confirm whether IsoA binds to Trx1 directly, we incubated IsoA with Trx1 protein in vitro, and examined the products by MS.We detected two major components at m/z 1624.79 and m/z 2694.12,respectively.The molecular weight of IsoA was 534.6.The component at m/z 1624.79 represented the peptide LVVVDFSATWCGPCK, corresponding to the 22 to 36 residues of Trx1.The molecular weight of this peptide plus two IsoA molecules was 2694.12,equal to the molecular weight of another component at m/z 2694.12,indicating that a covalent reaction occurred between IsoA and Trx1 at the ratio of 2:1 (Fig. 5C).Further MS/MS analysis demonstrated that IsoA covalently conjugated to both Cys32 and Cys35 residues of Trx1 together (Fig. 5D).

In vivo antitumor efficacy of IsoA in a xenograft mouse model
We next evaluated the in vivo antitumor efficacy of IsoA.In the xenograft model, HepG2 cells were inoculated subcutaneously into the nude mice.The mice were then treated by intraperitoneal injection with vehicle or IsoA (15 mg/kg/d) for 14 days.We found that treatment with IsoA did not show significant toxicity on the basis of stable body weights (Fig. 6A).However, treatment Means AE SD; ÃÃÃ , P < 0.001; ÃÃ , P < 0.01.D, The DNA oxidative damage marker 8-oxoG and protein oxidative marker nitrotyrosine were evaluated by immunostaining with anti-8-oxoG and anti-nitrotyrosine antibodies.E, Apoptosis was examined using the in situ TUNEL assay.F, Reduced GSH levels were evaluated by the GSH Detection Kit (means AE SD of three experiments; ÃÃÃ , P < 0.001).Trx1 activity in the xenograft was assessed using a Thioredoxin Activity Fluorescent Assay Kit (means AE SD of three experiments; Ã , P < 0.05).
with IsoA significantly inhibited the growth of HepG2 xenograft (Fig. 6B) and reduced the weights of tumors compared with the vehicle-treated group (Fig. 6C).The DNA damage marker 8-oxoG and protein oxidative marker nitrotyrosine (Fig. 6D) were obviously increased in IsoA-treated tumors, suggesting that IsoA induced oxidative stress in the xenografts.Further analysis revealed that there was a significant increase of TUNEL-positive cells in the xenografts treated with IsoA, compared with the vehicle control (Fig. 6E).Besides, IsoA significantly suppressed the activity of Trx1 and depleted the GSH levels in the tumors (Fig. 6F), consistent with our in vitro observations.These results demonstrated that IsoA exhibits potent antitumor activity in vivo.
To explore the safety of IsoA further, we screened its toxicity to bone marrow (blood counts), liver (aspartate aminotransferase, AST; alanine aminotransferase, ALT; alkaline phosphatase, ALP), and kidney (creatinine, and blood urea nitrogen, BUN) in mice.All the indices of routine blood test including red blood cell count, white blood cell (WBC) count, lymphocyte count, platelet count, and hemoglobin level remained in the normal ranges after IsoA treatment (Fig. 7A).There were no significant differences in blood biochemical parameters (ALT, AST, ALP, BUN, and creatinine) between IsoA-treated and the control groups (Fig. 7B-F).In addition, after collecting blood samples for hematology, the vital organs (liver, spleen, and kidney) were collected, fixed in formalin, and processed for hematoxylin and eosin (H&E) staining.Histopathologic evaluation did not reveal any significant differences between the vehicle and IsoA-treated groups (Supplementary Fig. S14).

Discussion
In this study, we evaluated anticancer potentials of IsoA, a key component and a novel ent-kaurane diterpenoid from I. forrestii var.forrestii.Our data established IsoA as a Trx1 inhibitor, which inhibited tumor cell growth both in vitro and in vivo.
The inhibitory effect of IsoA on Trx1 in cancer cells was demonstrated with the following lines of evidence.First, IsoA induced a rapid increase of ROS in cancer cells, and the collapse of redox homeostasis may be responsible for IsoA's anticancer activity.This result suggested that IsoA regulated cellular antioxidant systems.Second, IsoA suppressed in vitro and in vivo activity of Trx1.In the in vitro reaction, IsoA inhibited Trx1 activity with the IC 50 of approximately 5.177 mmol/L, but had little effects on several thiol-containing enzymes, such as glutaredoxin, glutathione reductase, thioredoxin reductase, glutathione-S-transferase, and protein disulfide isomerases (Fig. 4A; Supplementary Fig. S12).At the in vivo level, IsoA showed an inhibitory effect on Trx1 both in tumor cells and xenograft tumors (Figs.5B and C  and 6F).Third, downregulation of Trx1 in cancer cells significantly enhanced IsoA-induced growth-inhibitory effects.Importantly, IsoA-Trx1 adducts were detected by MS/MS.Two residues, Cys32 and Cys35, at the catalytic sites of Trx1 were identified to be critical amino acids for the IsoA-Trx1 interaction, probably through a Michael addition reaction.Taken together, these results indicated that IsoA directly targeted the Cys32/Cys35 amino acids to form a covalent complex to inhibit Trx1 activity.This oxidative-like modification inhibited the reversible reduction of Trx1 by TrxR and NADPH.Consequently, cellular redox balance was disrupted and ROS-mediated cell oxidative damage and apoptosis ensued.
modification of cysteine residues of Trx1 may be a direct and effective way to suppress its activation.
As IsoA showed stronger suppression effect against Trx1 than several other thiol-containing enzymes (glutaredoxin, glutathione reductase, thioredoxin reductase, glutathione-S-transferase, and protein disulfide isomerases) tested (Fig. 4A; Supplementary Fig. S12), we propose that IsoA is a relatively specific inhibitor of Trx1.The results that IsoA triggered ROS and activated ASK1/JNK cascade associated with apoptosis (Supplementary Fig. S13) and knockdown of Trx1 sensitized the cytotoxicity of IsoA (Fig. 4F) also demonstrated a critical role of Trx1 in IsoA-mediated apoptosis of cancer cells.Currently, we could not rule out the possibility that IsoA may interact with other thiol-containing enzymes or proteins.Nevertheless, at least, Trx1 is one of the important targets for IsoA.Accumulating evidence clearly showed that the reactivity of thiolcontaining proteins was not determined solely by their -CXXCmotif, but also by several residues spatially close to the active sites (41,42).The formation of a hydrogen bond or a salt bridge in the active site has a dramatic effect on the function of the protein.In our study, besides the Cys32/35 in the active sites, molecular docking showed that IsoA also formed a hydrogen bond with Met74, which also contributed to the stable binding interaction in both complexes (Fig. 5A and B).Consequently, we reasoned that the relatively specific inhibitory effect of IsoA on Trx1 might attribute to the extended active-site motif as well as residues distant in sequence but spatially close to the active site (Cys-Gly-Pro-Cys) of Trx1.Further research is needed to explore the mechanism by which IsoA had a relatively higher affinity with Trx1.
Recently, numerous ROS-inducing agents have been described to selectively kill cancer cells but spare normal cells (29)(30)(31)(43)(44)(45).In this work, IsoA induced an intracellular burst of ROS in cancer cells (Fig. 2A-D).Antioxidants NAC, GSH, PEG-SOD, or PEG-Catalase markedly abolished IsoA-induced growth inhibition, ROS generation, and apoptosis (Fig. 3).Therefore, ROS elevation may play a central role in mediating antitumor activity of IsoA through blocking Trx1.Consistent with the ROS threshold theory, our findings suggest that IsoA may be a novel Trx1 inhibitor with minimal toxicity to normal cells.
In conclusion, here we present evidence that IsoA is a novel Trx1 inhibitor through covalently binding to its catalytic sites Cys32 and Cys35 and inhibits tumor cell growth both in vitro and in vivo.The selective cytotoxic activity of IsoA to cancer cells can be explored further to develop novel antitumor agents.

Figure 1 .
Figure 1.Effects of IsoA on various cells.A, The chemical structure of IsoA.B, The IC 50 values of IsoA for indicated cell lines.Cells were treated with IsoA at various concentrations for 24 hours and processed for MTT assay.C, Cytotoxicity of IsoA to hepatocellular carcinoma cell line HepG2 and normal human hepatic cell line LO2.Cells were incubated with IsoA for 24 hours, and the viability was determined by MTT assay.D, Selective inhibitory effects of IsoA on cancer cells.LO2 cells and HepG2 cells were treated with IsoA at indicated concentrations.Viable cell number was assessed by Trypan blue exclusion assay.E, HepG2 cells were exposed to IsoA for 16 hours, and apoptosis was determined by Hoechst 33258 staining analysis.F, HepG2 cells were pretreated with Z-VAD-FMK for 2 hours followed by treatment with IsoA for 16 hours, and the apoptotic cells were evaluated and quantified by Annexin V/7-AAD staining and flow cytometry.The mean AE SD of three experiments is shown.ÃÃÃ , P < 0.001.

Figure 2 .
Figure 2. IsoA induces ROS and DNA damage in human cancer cells.A, HepG2 cells were treated with indicated concentrations of IsoA for 2 hours, and the ROS level was determined by flow cytometry (means AE SD of three experiments).B, HepG2 cells were treated with 15 mmol/L of IsoA for the indicated time, and then ROS levels were measured (means AE SD of three experiments).C, BEL7402 cells were treated with indicated concentrations of IsoA for 5 hours, and ROS levels were measured (means AE SD of three experiments).D, BEL7402 cells were treated with 20 mmol/L of IsoA for the indicated time, and ROS levels were measured (means AE SD of three experiments).E, IsoA decreased intracellular GSH levels.HepG2 cells were treated with 15 mmol/L of IsoA for the indicated time.The GSH levels were measured by GSH and GSSG Assay Kit and normalized to protein level (means AE SD of three experiments; ÃÃ , P < 0.01).F, IsoA depleted GSH levels dosedependently.HepG2 cells were treated with IsoA at indicated concentrations for 24 hours.The GSH levels were measured by GSH Detection Kit and normalized to protein level.G, The cells were incubated with 15 mmol/L of IsoA for 16 hours.8-oxoG DNA lesion was detected and quantified by flow cytometry using a FITC-conjugated antibody.H and I, Concentration-and time-dependent induction of gH2AX in response to IsoA.HepG2 cells were treated with IsoA at indicated times or concentrations.gH2AX was detected by Western blot analysis.GAPDH served as a loading control.The histograms represent means AE SD (of three experiments) of Western blot quantification of gH2AX/ GAPDH.

Figure 3 .
Figure 3.ROS scavengers attenuate IsoA's activity in HepG2 cells.A-D, HepG2 cells were pretreated with NAC (A), GSH (B), PEG-Catalase (C), and PEG-SOD (D) at indicated concentrations for 30 minutes, and then incubated with indicated concentrations of IsoA for 24 hours.The cell viability was determined by MTT assay.E, HepG2 cells were pretreated with 2 mmol/L of NAC for 30 minutes and then incubated with 15 mmol/L of IsoA for 16 hours, followed by apoptosis assay using Annexin V staining and flow cytometry.F, Quantification of Annexin V staining results.G, HepG2 cells were pretreated with 2 mmol/L of NAC for 30 minutes and then incubated with 15 mmol/L of IsoA for 1 hour.ROS levels were measured by flow cytometry and quantified.H, HepG2 cells were pretreated with 1,000 U of PEG-Catalase for 30 minutes, followed by incubating with 15 mmol/L of IsoA for 16 hours, and cell apoptosis was analyzed by flow cytometry and quantified.I, HepG2 cells were pretreated with 1,000 U of PEG-Catalase for 30 minutes, followed by incubating with 15 mmol/L of IsoA for 1 hour, and ROS levels were analyzed by flow cytometry and quantified.

Figure 4 .
Figure 4. IsoA inhibits Trx1 activity.A, In vitro assays for Trx1 and Grx activity.For Trx1 acticity assay, 0.02 mmol/L of hTrx1 protein (the final concentration) was subjected to the assay in the presence of indicated concentrations of IsoA.For Grx activity assay, 1.5 nmol/L of hGrx-1 (the final concentration) was subjected to the assay in the presence of indicated concentrations of IsoA (means AE SD of three experiments).B, HepG2 cells were treated with IsoA at the indicated concentrations for 8 hours, and Trx1 activity in cell lysates was determined by insulin endpoint assay (means AE SD of three experiments).C, BEL7402 cells were treated with 20 mmol/L of IsoA for 8 hours, and Trx1 activity in cell lysates was measured by insulin endpoint assay (means AE SD of three experiments).D, HepG2 cells were incubated with 15 mmol/L of IsoA for the indicated time, and Trx1 and thioredoxin reductase levels were detected by Western blotting.GAPDH served as a loading control.Bar graphs represent the quantified results of protein levels (Trx1 and TrxR), which were normalized to corresponding GAPDH protein level and expressed as fold of control (mean fold of control AE SD of three experiments).E, HepG2 cells transfected with the scrambled siRNA (control) and Trx1specific siRNA for 48 hours, and the knockdown efficiency of Trx1 levels was analyzed by Western blotting.The histogram showed means AE SD (of three experiments) of Western blot quantification of Trx1/GAPDH.ÃÃÃ , P < 0.001.F, The Trx1 activity in Trx1-silenced HepG2 cells was determined by insulin endpoint assay.G, Trx1-silenced HepG2 cells and control cells were treated with IsoA for 24 hours, and cell viability was measured by MTT assay (means AE SD of three experiments).

Figure 5 .
Figure 5. IsoA covalently binds with Cys32 and Cys35 in Trx1.A and B, Molecular docking for binding models of IsoA in Cys32 and Cys35 residues of Trx1.C, IsoA (200 mmol/L) was incubated with Trx1 protein (70 mg) at 37 C for 2 hours, and the adduct forms were assayed by MS.The molecular weight of peptide P1 was 1624.85,representing 22 to 36 amino acids of Trx1, and the molecular weight of the adduct product between peptide P1 and two IsoA, which was the detected peptide P2, was 2694.14.C Ã represents the cysteine binding with IsoA.D, The identified peptide P2 was further analyzed by MS/MS.b1, b2, b3, b4, b5, b6, and b7 represent the dissociated fragments of peptide P2.C Ã represents the cysteine binding with IsoA.b2, b4, and b5 represent the fragments conjugated with IsoA.

Figure 6 .
Figure 6.IsoA treatment suppresses tumor growth in vivo.A-C, Mice implanted with HepG2 cells were administered IsoA or the vehicle control for 14 days.Body weights (A), tumor volume (B), and tumor mass (C) of IsoA-treated or vehicle-treated animals were monitored as described in Materials and Methods.Means AE SD; ÃÃÃ , P < 0.001; ÃÃ , P < 0.01.D, The DNA oxidative damage marker 8-oxoG and protein oxidative marker nitrotyrosine were evaluated by immunostaining with anti-8-oxoG and anti-nitrotyrosine antibodies.E, Apoptosis was examined using the in situ TUNEL assay.F, Reduced GSH levels were evaluated by the GSH Detection Kit (means AE SD of three experiments; ÃÃÃ , P < 0.001).Trx1 activity in the xenograft was assessed using a Thioredoxin Activity Fluorescent Assay Kit (means AE SD of three experiments; Ã , P < 0.05).