Tanshinones induce tumor cell apoptosis via directly targeting FHIT

The liposoluble tanshinones are bioactive components in Salvia miltiorrhiza and are widely investigated as anti-cancer agents, while the molecular mechanism is to be clarified. In the present study, we identified that the human fragile histidine triad (FHIT) protein is a direct binding protein of sodium tanshinone IIA sulfonate (STS), a water-soluble derivative of Tanshinone IIA (TSA), with a Kd value of 268.4 ± 42.59 nM. We also found that STS inhibited the diadenosine triphosphate (Ap3A) hydrolase activity of FHIT through competing for the substrate-binding site with an IC50 value of 2.2 ± 0.05 µM. Notably, near 100 times lower binding affinities were determined between STS and other HIT proteins, including GALT, DCPS, and phosphodiesterase ENPP1, while no direct binding was detected with HINT1. Moreover, TSA, Tanshinone I (TanI), and Cryptotanshinone (CST) exhibited similar inhibitory activity as STS. Finally, we demonstrated that depletion of FHIT significantly blocked TSA’s pro-apoptotic function in colorectal cancer HCT116 cells. Taken together, our study sheds new light on the molecular basis of the anti-cancer effects of the tanshinone compounds.

Here, we report that STS directly engaged to FHIT protein and inhibited FHIT Ap3A hydrolase activity through the competitive binding to the FHIT substrate binding site. Similar effects were observed with other tanshinone analogs, including TSA, TanI, and CST, although STS and TSA were very different on cell toxicity. Finally, we demonstrated that FHIT depletion partially blocked the pro-apoptotic function of TSA.

STS directly binds to FHIT protein.
To identify small molecule directly binding to FHIT in cellular content, we took the advantage of microscale thermophoresis (MST) assay that allows determining the direct engagement between small molecule and cellular target proteins in biological liquid 39,40 . We first created HEK293T cells containing GFP-tagged FHIT and prepared the cell lysates for single-point MST binding check screening. In a FDA-approved drug library containing 1971 compounds, several small molecules were found directly binding to FHIT (Fig. 1). Among these candidates, MST binding affinity check assay revealed small molecule with the highest binding affinity is STS, a water-soluble derivative of TSA, and the Kd value between STS and FHIT was 268.4 ± 42.59 nM (Fig. 2B). To further verify the direct interaction between STS and FHIT, we purified FHIT protein in E.coli. Interestingly, we also detected a binding curve using the purified FHIT, but the binding affinity (Kd value = 1.82 ± 0.3 µM) was 5 times lower that the cellular FHIT (Fig. 2B). The higher binding affinity of cellular FHIT and STS supports the advantage of drug screening using cellular protein.
STS binds to the FHIT substrate-binding site. The overall structure of FHIT is displayed as a general α + β type, and the N terminal is the HIT domain which accounts for Ap3A binding and hydrolysis (Fig. 2C). The FHIT protomer contains two helices, A and B, and seven β strands 41 . Substrate binds on the surface of the pocket consisted of seven β strands in the crystal structure of the FHIT-IB2 complex 7 . To determine the binding site of STS, we constructed three GFP-tagged FHIT truncations (Fig. 2C). In cellular MST assay, robust binding curve was determined with STS and truncation region 1 (TR1) and TR2, but no direct binding was observed with TR3, which represented C-terminal only (Fig. 2D). The occupation of STS to the HIT domain implies that STS potentially attaches to the substrate-binding pocket of FHIT. To validate the possibility, molecule docking was proposed to assess the interaction between STS and FHIT using the published crystal structure of the FHIT-AMP complex [Protein Data Bank (PDB): 3FIT]. Indeed, the prediction showed that STS adopts a similar three-dimensional (3D) conformation upon docking to FHIT as AMP (Fig. 2E). Moreover, STS could be well fitted into the 'active site' of FHIT in the surface model (Fig. 2F). STS competitively inhibits FHIT hydrolase activity. Next, we attempted to assess whether STS inhibits the hydrolase activity of FHIT. It has been established that Ap3A can be rapidly hydrolyzed to AMP and ADP by FHIT 2 . Therefore the production of ADP in the reaction system could represent the hydrolase activity of FHIT. With this principle, we determined the IC50 value of suramin (a known inhibitor of FHIT) is 0.26 ± 0.08 µM, which was comparable to previous report 17 . Using the same method, we found that STS inhibits the hydrolase activity of FHIT with an IC50 value of 0.34 ± 0.09 µM (Fig. 3A). To further investigate whether STS inhibits FHIT hydrolase activity via competitive engagement with the substrate-binding site of FHIT, we purified an FHIT mutant that substituted His to Asn at the central histidine (H96N mutant), which is catalytically inactive with no effect on substrate binding 7,8 . Consistently, in a fluorescence polarization assay with the H96N mutant utilizing a fluorogenic substrate ApppBODIPY, STS significantly disrupted the interaction between H96N mutant and ApppBODIPY with an IC50 value of 2.2 ± 0.05 µM (Fig. 3B). Similar result was observed for suramin with an IC50 value of 2.6 ± 0.044 µM (Fig. 3B). All those results indicated that STS is a competitive inhibitor for FHIT.   (Fig. 4A). Notably, no direct binding curve was observed between STS and HINT1 (Fig. 4A).
We also tested whether STS inhibits Ap3A hydrolyze activity of ENPP1. In the same method of Ap3A hydro- www.nature.com/scientificreports/ lase activity detection as described above (Fig. 3A), STS showed much less inhibition to ENPP1 compared to suramin (Fig. S1B). This was confirmed by the thin layer chromatography (TLC) assay with ApppBODIPY, in that suramin inhibited ENPP1 hydrolase activity at 6.25 µM, whereas STS did not show inhibition until to 125 µM (Fig. S1C). Altogether, These results indicate that STS specifically inhibits FHIT hydrolase activity compared with ENPP1. The inhibition of STS on the enzyme activity of DCPS and GALT needs further investigation.  with similar biological effects as TSA, the major bioactive compound of Danshen 44 . Moreover, the structures of these three compounds are very similar, except that CST has a single bond at the C-15 position of the furan ring and TanI having a methylbenzene instead of dimethyl cyclohexanes at the left compared with TSA (Fig. 5A).
To investigate whether these three compounds inhibit FHIT hydrolase activity or not, we firstly determined the binding affinity between those compounds and FHIT via MST assay. As expected, all the compounds displayed similar binding curves with the Kd value around 100-150 nM except TanI (Fig. 4B). The Kd value of TanI increased by four-folds (466 ± 60 nM), which may be due to the different structure of TanI at the left compared with others. Furthermore, in TLC assay with ApppBODIPY 45 , TSA, TanI, and CST exhibited similar inhibition on FHIT hydrolase activity with IC50 values of 4 ~ 6 µM, and STS displayed slightly more effective inhibition than others (Fig. 5B).

The effect of TSA inducing tumor cell apoptosis is blocked by FHIT depletion. It has been
reported that the FHIT-Ap3A complex may mediate the anti-tumor function of FHIT by inducing apoptosis 7,8 .
To investigate whether tanshinone compounds inhibit tumor cell growth via FHIT-dependent apoptosis pathway, we created FHIT knockout (KO) colorectal HCT116 cells by CRISPR/Cas9 mediated genome editing, genomic DNA sequence revealed that exon5 to exon6 of the FHIT gene was deleted (Fig. S2A), and the depletion of FHIT protein was confirmed by western blotting (Fig. S2B). In ordinary culture, no clear change of FHIT knockout cells was observed regarding the epithelial morphology and growth rate compared to wild-type (WT) cells (Fig. S2D). Next, we determined the cytotoxicity of STS and TSA in WT HCT116 cells. CCK8 assay shown STS had no cytotoxicity, even the concentration is as high as 100 µM (Fig. S3A). TSA significantly inhibits the growth of HCT116 cells at 2 µM (Fig. S3A). A previous study reported that TSA, TanI, and CST induced tumor cell apoptosis 46 . To ask if FHIT knockout disrupt the pro-apoptotic effects of TSA and TanI, apoptosis was detected by flow cytometry with Annexin V-fluorescein isothiocyanate (V-FITC) and propidium iodide (PI) staining. When cells were treated with 5 µM TSA for 24 h, the apoptosis rate (4.9%) in FHIT KO cells was significantly lower than the rate (13.5%) in WT cells (Fig. 6A). Similar to previous report 47 , when cells were treated with camptothecin (CPT) at 1 µM that served as a positive control, apoptosis was significantly lower in KO cells (5.3%) than in WT cells (24.9%) as well (Fig. 6C). Notably, TanI led more striking difference in apoptosis assay between WT and KO cells, in that TanI at 10 µM induced 31.66% apoptosis in WT cells and 12.7% apoptosis cells in KO cells (Fig. 6B), a significant difference was also presented at 5 µM (Fig. 6D). All these findings indicated that tanshinone compounds induction of tumor cell apoptosis is partially dependent on FHIT.  (  e  n  o  n  i  h  s  n  a  t  o  t  p  y  r  C  )  I  n  a  T  (  I  e  n  o  n  i  h  s  n  a  T  )  A  S  T  (  A  I  I  e  n  o  n  i  h  s  n  . The experiment was repeated three times in each group. Data were analyzed using two-way ANOVA, one asterisk stands for P < 0.05, four asterisks stand for P < 0.0001. (D) Quantification of V-FITC/PI positive apoptotic cells in (B). The experiment was repeated 3 times in each group. Data were analyzed using two-way ANOVA, two asterisks stand for P < 0.001, three asterisks stand for P < 0.0005.

Discussion
Tanshinones are natural terpenoids and the main bioactive ingredients of Salvia miltiorrhiza 23,24 . Previous studies showed that tanshinones inhibit tumor via induction of various apoptosis signaling pathways 46 . Our study demonstrated that tanshinone compounds directly bind to FHIT and inhibit the hydrolase activity. We found that STS is a competitive inhibitor of FHIT by occupying the 'active site' with a IC 50 value of 2.2 ± 0.05 µM. As the substrate-binding activity of FHIT is essential for its anti-tumor function 7,8 , we speculate that STS might mimic the function of Ap3A. Given that FHIT-Ap3A complex was immediately abolished when Ap3A was hydrolyzed 48 , while STS functions as a nonhydrolyzable substrate, the stabilized FHIT-STS complex could enhance the antitumor function of FHIT. In addition, it has been reported that a nonhydrolyzable analog of Ap4A binded to FHIT and stabilized its' flexible loop which was important for its tumor suppression activity 49 . Therefore, STS may function in a similar way to enhances the anti-tumor function of FHIT as the nonhydrolyzable substrate. Moreover, pro-apoptotic induction by TSA and TanI's was partially blocked by FHIT depletion. Considering that FHIT knockout did not fully abolish tanshinones' anti-cancer function, there should be multiple targets in multiple steps involved to the effects of tanshinones inhibition of tumor progression 50 . Recently, Li et al.reported that TSA docked into the ATP-binding pocket of Aurora B, and was a potential ATP competitive kinase inhibitor 51 . However, that study lacked evidence of direct binding between TSA and Aurora B kinase. It will be interesting to clarify the engagement between TSA and Aurora B and the potential connection to FHIT in the future work. We also found that STS and TSA are very different on cytotoxicity for HCT116 cells, in that TSA significantly inhibited cell growth at 2 µM, while STS did not show any cytotoxicity even at 100 µM (Fig. S3A), although they are structurally similar and function interchangeable. This was also confirmed by the cloning formation assay, TSA at 1 µM clearly abolished the colony formation of HCT116 cells while no effect was detected with STS up to 10 µM (Fig. S3B). This difference is consistent to previous studies, in that Song et al.reported that TSA inhibited the proliferation, migration, and invasion of colorectal cancer 52 , osteosarcoma cells 53 , non-small cell lung cancer cells 54 , while few report indicated the anti-cancer activity of STS 38 , and Liu et al. claimed that STS was not effective on the growth and metastasis of Lewis carcinoma 55 . One explanation for this difference could be that the higher lipophilicity of TSA facilitates its uptake by cancer cells 56 .
Finally, an important finding in our study is that STS inhibits FHIT Ap3A hydrolase activity. Diadenosine polyphosphates (ApnAs) have been proposed to have several intracellular functions as signaling stress responses 18 . Indeed, Ap3A plays an essential role in platelet regulation, which provides a source of long-lasting ADP at the site of vascular injury 57 . What's more, the intracellular concentration of Ap4A increases upon oxidative stressors, and aminoglycoside antibiotics induce the production of Ap4A in bacteria, which in turn enhances their bactericidal activity 58 . In addition, Ap4A attenuates STING-dependent signaling and inflammatory 59 . Given that tanshinone compounds have similar activities, such as anticoagulant, anti-oxidant, and anti-inflammatory 24 , tanshinone compounds directly bind on FHIT 'active site' and inhibit its hydrolase activity may shed light on the molecular mechanism of the function for those compounds. The role of FHIT and its hydrolase activity in cardiovascular, ROS, and STING signaling is also worth to be anticipated in future studies.
Protein purification. The gene encoding human FHIT (P49789) was synthesized (Genscript, China) and cloned into pGEX-6P-1 such that the expressed protein would have an N-terminal glutathione S-transferase (GST)-tagged protein. The construct was transformed into expression host E.coli BL21 (DE3), and cells were grown in LB medium at 37 °C until the OD600 reached 0.6. The cells were then induced by adding 0.1 mM isopropyl-β-D-l-thiogalactopyranoside (IPTG) and grown at 18 °C overnight. Cells were harvested by centrifugation. The cell pellet was resuspended in a buffer containing 20 mM Tris (pH 7.5), 500 mM NaCl, and 2 mM DTT and lysed by sonication. After centrifugation, the clarified cell lysate was incubated with glutathione-sepharose 4B beads, and GST-tagged FHIT was eluted with the buffer consisting of 50 mM reduced glutathione. GST tag was cleaved by prescission protease, and the released FHIT was further purified to homogeneity by successive chromatographic steps involving glutathione-sepharose 4B beads, and gel-filtration columns. Pure protein fractions in buffer containing 20 mm Tris (pH 7.5), 150 mM NaCl, and 2 mM DTT were collected and used for the indicated assays. The single-site mutants H96N of FHIT was also constructed in the same vector using two primers, with required mutation using PCR. The mutant was purified with the same protocol as the wild-type FHIT. The gene encoding human HINT1 (P49773), GALT (P07902), DCPS (Q96C86) were synthesized (Genscript, China) and cloned into pET-28a such that the expressed proteins would have an N-terminal 6 × His-tag. Those plasmids were expressed in E. coli BL21(DE3) strain, purified as recommended before, by Co-NTA beads, then size exclusion (Superdex200 increase 10/300GL), and Anion exchange (ENrich Q) columns case by case. Finally, all the proteins were concentrated in 20 mM Tris (pH 7.5), 150 mM NaCl, 2 mM DTT, and stored at − 80 °C.
Microscale thermophoresis (MST). Ten million 293 cells overexpressing GFP-tagged FHIT or free GFP were lysed in 0.5 mL M-PER mammalian protein extraction reagent (Thermo Fisher Scientific). Cell lysates were FHIT hydrolase activity detection assay. 5

Molecular docking.
To find STS binding site against FHIT, the X-ray crystal structure of FHIT with AMP (PDB ID: 3FIT) was downloaded from Protein Data Bank (PDB). The protein structure was prepared using AutoDockTools-1.5.6, including deleting AMP and water molecules, and adding hydrogens, before the corresponding protein grid files were generated for docking. Then the structure file of STS was pretreated in the Ligand module of AutoDockTools-1.5.6, and docking was performed with AutoDock Vina software as mentioned before 60 . The docking pose for the FHIT-STS complex was then submitted to binding mode analysis and figure generation using PyMOL1.8.
Fluorescence polarization assay. The binding of FHIT (H96N) to fluorescein-labeled substrates App-pBODIPY was analyzed by fluorescence polarization assay in 384-well format for dose-response curve studies using spark instrument (TECAN) according to the protocol reported before 61 . And 5 µM FHIT (H96N) protein was incubated with 1 µM fluorescein-labeled substrates in buffer A (25 mM Tris-HCl, pH 7.5, 5 mM MgCl2 and 2 mM DTT) with series diluted inhibitors as indicated. Each sample was allowed to equilibrate in the solution for 30 min. After 30 min, the steady-state fluorescence anisotropy (mP) was measured. A second reading was taken after 10 min, to ensure that the mixture was well-equilibrated and stable. Less than 5% change was observed between 30 and 40 min measurements. The IC50 value was determined by fitting the binding curves using GraphPad Prism 7.0.
Western Blot analysis. Cells were lysed using RIPA (Cell Signaling Technology) containing protease inhibitor (Thermo Fisher Scientific) at 4 °C. Proteins were quantified by BCA protein quantification kit (Thermo Fisher Scientific). And proteins (10-30 μg per lane) were separated by SDS-PAGE and then transferred onto polyvinylidene fluoride membranes (Millipore). Following this, the membranes were blocked with 5% non-fat milk for an hour. Later on, the membranes were incubated with primary antibodies anti-FHIT (#ab170888, Abcam), anti-GAPDH (#5174, CST) overnight at 4 °C. After that, the membranes were incubated with HRPgoat anti-mouse secondary antibodies (1:5000, CST) for 1 h, Enhanced chemiluminescence detection reagent (NCM biotech) was used to visualize the signal strength of the bands.