Frenolicin B targets peroxiredoxin 1 and glutaredoxin 3 to trigger ROS/4E-BPl-mediated antitumor effects

SUMMARY

Peroxiredoxin 1 (Prxl) and glutaredoxin 3 (Grx3) are two major antioxidant proteins that play a critical role in maintaining redox homeostasis for tumor progression. Here, we identify the prototypical pyranonaphthoquinone natural product frenolicin B (FB) as a selective inhibitor of Prx1 and Grx3 through covalent modification of active site cysteines. FB-targeted inhibition of Prx1 and Grx3 results in a decrease in cellular glutathione levels, an increase of reactive oxygen species (ROS), and concomitant inhibition of cancer cell growth, largely by activating the peroxisome-bound tuberous sclerosis complex to inhibit mTORC1/4E-BP1 signaling axis. FB structure-activity relationship studies reveal a positive correlation between inhibition of 4E-BP1 phosphorylation, ROS-mediated cancer cell cytotoxicity and suppression of tumor growth in vivo. These findings establish FB as the most potent Prx1/Grx3 inhibitor reported to date and also notably highlight 4E-BP1 phosphorylation status as a potential predictive marker in response to ROS-based therapies in cancer.

Graphical Abstract

eTOC Blurb

Ye, Zhang et al. identify Frenolicin B as a potent and selective inhibitor of Prx1 and Grx3, leading to generation of ROS and subsequent repression of mTORC1/4E-BP1-mediated translational control of tumor growth with the potential to be developed into a new class of anticancer agents.

INTRODUCTION

Altered redox status is a common feature of many cancers wherein deregulation of cell signaling and metabolism by multiple genetic alterations often leads to increased generation of intracellular reactive oxygen species (ROS) (Trachootham et al., 2009). Increased intracellular ROS can contribute to tumorigenesis and cancer progression by amplifying the genomic instability and activation of tumor-promoting signaling pathways. However, excess oxidative stress, due to further elevated ROS levels beyond the cellular tolerability threshold or weakened antioxidant defenses of the cell, can cause oxidative damage to DNA, lipids and proteins, which in turn results in pathophysiological changes such as apoptosis, necrosis and cell cycle disruption (Trachootham et al., 2009). As such, discovery of pharmacological agents that either promote ROS generation or disable the cellular antioxidant system to induce ROS-mediated damage in cancer cells has been proposed as a potential selective anticancer therapeutic strategy.

Peroxiredoxins (Prxs), thioredoxins and glutaredoxins (Grxs) have been characterized as antioxidants to protect cells from oxidative insults and maintain the cellular redox homeostasis (Hanschmann et al., 2013). These antioxidant proteins regulate redox processes by the reversible formation of disulfides through the oxidation of active site cysteine thiols (Aslund and Beckwith, 1999). Prxs catalyze peroxide reduction to balance cellular hydrogen peroxide (H₂O₂) levels, and play essential roles in oxidative stress and cell signaling. Based on the number and position of conserved cysteine residues involved in the peroxidase reaction, mammalian Prxs are classified into typical 2-Cys (Prx1-4), atypical 2-Cys (Prx5) and 1-Cys Prxs (Prx6) (Park et al., 2016). Prx1 is overexpressed in many cancers associated with tumor recurrence and poor prognosis (Park et al., 2016). Expression of Prx1 enhances cancer cell survival from oxidative stress-induced cell death and promotes malignant transformation (Bae et al., 2007; Jiang et al., 2014). Thus, Prx1 has gained attention as an attractive target for cancer therapy. While several small molecule inhibitors targeting Prxd1 and Prx2 have been identified (Liu et al., 2012; Nguyen et al., 2013), the isoform selectivity and potency of these inhibitors remain to be further developed.

Grxs protect cells against oxidative stress by catalyzing S-glutathionylated protein (protein-SSG) mixed disulfide and can be regenerated by the reduced form of glutathione (GSH) (Lillig et al., 2008). Grxs can be divided into two main groups based phylogeny, active site motif and domain structure: the ‘classical’ dithiol Grxs containing the active site consensus sequence Cys-Pro-Tyr-Cys and the monothiol Grxs with a Cys-Gly-Phe-Ser active site consensus sequence (Lillig et al., 2008). Although monothiol Grxs contain all functional residues and substrate binding motifs to catalyze the reduction of mixed disulfides between proteins and GSH, for most their enzymatic activities remain unresolved (Herrero and de la Torre-Ruiz, 2007). Monothiol Grxs can be further separated into single-domain Grxs that consist of only one Grx domain, and multi-domain Grxs that contain an N-terminal thioredoxin-like domain and one to three C-terminal monothiol Grx domains (Isakov et al., 2000). The later Grxs are restricted to eukaryotic cells and includes human Grx3 (also known as TXNL2 and PICOT). Human Grx3 is a non-redox active 2Fe-2S center-containing multi-domain monothiol Grx that contains two monothiol Grx domains (Haunhorst et al., 2010), and is overexpressed in colon, lung and breast cancers (Cha and Kim, 2009; Qu et al., 2011). Recent studies demonstrated Grx3 to play a crucial role in tumor growth and metastasis through regulating ROS levels (Li et al., 2018; Qu et al., 2011), implicating Grx3 as an important target for cancer treatment. As inhibitors of Grx3 function have yet to be reported, the discovery of novel Grx3-specific inhibitors/probes are of great interest.

Dysregulation of cap-dependent translation through redundant phosphorylation of the translational repressor 4E-BP1 by multiple oncogenic pathways, such as PI3K/AKT/mTOR and RAS/RAF/MEK/ERK, is associated with malignant progression and therapeutic resistance (Bhat et al., 2015; Boussemart et al., 2014; Cai et al., 2014; Ducker et al., 2014; Hsieh et al., 2010; Mi et al., 2015; She et al., 2010; Ye et al., 2014). During a screen to identify potential novel natural product-based inhibitors of cap-dependent translation, we unexpectedly identified the pyranonaphthoquinone (PNQ) natural product, frenolicin B (FB) as an inhibitor of 4E-BP1 phosphorylation and cap-dependent translation. Subsequent mechanistic studies revealed FB as a potent covalent Prx1/Grx3 inhibitor that functions via selective modification of Prx1 Cys83/Cys173 and Grx3 Cys159/Cys261. These studies also revealed inhibition of Prx1 and Grx3 by FB to indirectly repress the mTORC1/4E-BP1-mediated translational activity through a ROS-dependent mechanism, resulting in growth inhibition of cancer cells in vitro and in vivo. Furthermore, FB structure-activity relationship (SAR) studies implicate 4E-BP1 phosphorylation status as a potential sensor for ROS-mediated cancer cell cytotoxicity.

RESULTS

FB Inhibits Cap-Dependent Translation in 4E-BP1-Dependent Manner

Recent work suggests that targeting cap-dependent translation may overcome intra-tumor heterogeneity-mediated resistance and provide a promising strategy for improving cancer therapy (Bhat et al., 2015; Boussemart et al., 2014; Cai et al., 2014; Ducker et al., 2014; Hsieh et al., 2010; Mi et al., 2015; She et al., 2010; Ye et al., 2014). To identify microbial natural products capable of targeting cap-dependent translation, crude extracts of prioritized microbes from the Ruth Mullins underground coal mine fire site (O’Keefe et al., 2010) were screened using a cap-dependent translation-based luciferase reporter assay (She et al., 2010; Wang et al., 2017). This initial screen revealed extracts of Streptomyces sp. RM-4-15, a strain previously identified to produce a series of known and novel PNQs (Wang et al., 2013), to contain compounds capable of inhibiting cap-dependent translation (Figure 1A). Identical assays with purified metabolites (Wang et al., 2013) from Streptomyces sp. RM-4-15 demonstrated FB and the related PNQ metabolite UCF76-A to effectively inhibit cap-dependent translation (Figures 1B and 1C).

Figure 1. FB Inhibits Cap-Dependent Translation Mediated by 4E-BP1.

(A) Inhibition of cap-dependent translation by Streptomyces sp. RM-4-15 bacterial extract. HCT116 CRC cells were transfected with a bicistronic luciferase reporter (upper diagram) for 24 h, followed by treatment with different concentrations of bacterial extract for 12 h. Cap-dependent renilla luciferase activity was normalized with cap-independent firefly luciferase activity. The results are expressed as the inhibition of cap-dependent translation relative to the untreated controls. (B) Structures of FB and UCF76-A.

(C) Inhibition of cap-dependent translation by representative pure metabolites (RM1-RM7) of Streptomyces sp. RM-4-15. RM1, UCF76-A; RM2, FB.

(D) HCT116 cells were treated with 1 μM MK2206 and 100 nM PD0325901 alone and in combination, 100 nM rapamycin, 0.5 μM AZD8055, 2 μM UCF76-A, 2 μM FB or DMSO control for 12 h followed by western blot analysis for the indicated proteins.

(E and F) HCT116 cells with stable expression of two different sets of 4E-BP1 shRNAs or control shRNA (ShCtrl) were analyzed by western blot for 4E-BP1 and β-actin (E) or determined for cap-dependent translation activity after treatement with 2 μM FB or DMSO control for 12 h (F).

Data are shown as mean ± SEM (n=3). *p < 0.001; NS, not significant. See also Figure S1.

To further investigate the function of PNQs within the context of cap-dependent translation, the ability of FB and UCF76-A to modulate 4E-BP1 and p70S6 kinase phosphorylation was compared to that of representative mTOR inhibitors. The mTOR kinase complex 1 (mTORC1), a downstream target of both AKT and ERK signaling, is a well-characterized activator of cap-dependent translation through phosphorylation of 4E-BP1 and p70S6 kinase (Laplante and Sabatini, 2012). Rapamycin is an allosteric inhibitor of mTORC1 and can effectively inhibit p70S6K phosphorylation, but only weakly inhibits 4E-BP1 phosphorylation (Choo and Blenis, 2009). Alternatively, second generation ATP-competitive mTOR kinase inhibitors such as AZD8055 inhibit both mTORC1 and mTOR complex 2 (mTORC2) are more effective than rapamycin in inhibiting 4E-BP1 phosphorylation (Feldman et al., 2009). Like AZD8055 but distinct from rapamycin, FB and UCF76-A effectively inhibited 4E-BP1 phosphorylation in HCT116 colon cancer cells (Figure 1D). Both rapamycin and AZD8055 potently inhibited phosphorylation of p70S6K, and AZD8055 also inhibited phosphorylation of the mTORC2 substrate AKT (Laplante and Sabatini, 2012). Similarly, FB or UCF76-A also inhibited p70S6K phosphorylation, but both compounds had no inhibitory effect on AKT phosphorylation (Figure 1D). While FB was previously reported to inhibit AKT activity in vitro (Toral-Barza et al., 2007), no detectable inhibition of AKT phosphorylation or that of its substrate PRAS40 was observed in HCT116 cells treated with FB or UCF76-A (Figure 1D). In addition, the highly selective pan-AKT-1/2/3 inhibitor MK2206 (Yap et al., 2011) led to negligible modulation of 4E-BP1 phosphorylation (Figure 1D), consistent with our previous findings that simultaneous inhibition of both AKT (MK2206) and MEK/ERK (PD0325901) signaling is required to inhibit 4E-BP1 phosphorylation (Figure 1D) and repress cap-dependent translation in colorectal cancer (CRC) cells (She et al., 2010). Similar effects by FB and UCF76-A were also observed in other CRC (DLD-1) and breast (MDA-MB-231) cancer cell lines (Figure S1). Furthermore, Invitrogen SelectScreen® Kinase Profiling revealed no effect on mTOR kinase activity by representative PNQs (unpublished data). Notably, silencing 4E-BP1 expression by short hairpin RNAs (shRNAs) in HCT116 cells completely prevented the inhibitory effect of FB on cap-dependent translation (Figures 1E and 1F). Taken together, these data highlighted a previously unknown function of PNQs as potent inhibitors of cap-dependent translation through a 4E-BP1-dependent manner. Our findings further suggested that the inhibition of 4E-BP1 phosphorylation by PNQ-based natural products is mechanistically distinct from that of known mTOR, AKT and/or MEK/ERK inhibitors.

FB and Active PNQs Preferably Induce Cancer Cell Cytotoxicity That Correlates with Inhibition of 4E-BP1 Phosphorylation

To determine the antitumor potential of FB, 8 human cancer cell lines, including colon, breast and lung cancer cells, along with non-malignant human lung epithelial cell line BEAS-2B, fetal lung fibroblasts IMR-91 and TIG-1, and porcine aortic endothelial cell line PAE were tested for the growth-inhibitory effect of FB. Compared with the nonmalignant cells, FB displayed a preferential cancer cell cytotoxicity with IC₅₀ values of <150 nM (Figure 2A). Prompted by the potential mechanistic novelty of PNQs, we evaluated additional FB-based PNQ synthetic analogs (Figure S2A) for HCT116 CRC cell cytotoxicity (Figure 2B) and 4E-BP1 phosphorylation inhibitory potential (Figure 2C). Enabled by our recently reported divergent FB synthetic strategy (Zhang et al., 2013b), preliminary SAR analysis highlighted a clear correlation between cancer cell cytotoxicity, induction of cleaved PARP (an apoptotic marker), and inhibition of 4E-BP1 phosphorylation (Figures 2B and 2C), where improvements of up to 4-fold in potency over FB were observed (e.g., 12, Figure 2D). Consistent with the ability of dephosphophorylated 4E-BP1 to bind to the eIF4E-mRNA cap complex and suppress cap-dependent translation (She et al., 2010), active PNQ analogs displayed similar effects (Figures S2B and S2C) as exemplified by the correlation between inhibition of 4E-BP1 phosphorylation (Figure 2D) and inhibition of cap-dependent translation (Figures S3A and S3B) and protein synthesis (Figures S3C and S3D), HCT116 growth inhibition (Figure 2B) and induction of apoptosis (Figure 2E). Importantly, depletion of 4E-BP1 by shRNA or knockout largely attenuated 12-induced apoptosis and cell growth inhibition (Figures 2F and 2G). These data further established the central role of 4E-BP1 phosphorylation in response to PNQ-mediated cancer cell death.

Figure 2. FB and Active PNQs-Induced Cancer Cell Cytotoxicity Correlates with Inhibition of 4E-BP1 Phosphorylation.

(A) Growth for the indicated cell lines was assessed after 72 h of treatment with various concentrations of FB. Results are expressed as the half-maximal growth inhibitory concentration (IC50) of FB.

(B) HCT116 cells were treated with various concentrations of the indicated compounds for 72 h. The viable cells were counted, and the results are presented as a percentage of viable cell number relative to DMSO-treated control cells.

(C and D) HCT116 cells were treated with 2 μM of the indicated compounds (C) or with various concentrations of FB and 12 (D) for 12 h followed by western blot analysis for the indicated proteins.

(E) HCT116 cells were treated with vatious concentrations of FB and 12 for 72 h. Apoptotic cells were analyzed by flow cytometry, and the results are expressed as the increased levels of apoptosis by subtracting each of the DMSO-treated controls.

(F) HCT116 cells with stable knockdown of 4E-BP1 and control cells were treated with 2 μM 12 for 72 h followed by measuring apoptosis.

(G) 4E-BP1/2 WT and DKO MEFs were treated with various concentrations of 12 for 72 h and the results are presented as a percentage of viable cell number relative to DMSO-treated control cells.

Data are shown as mean ± SEM (n=3). *p < 0.001. See also Figures S2 and S3.

FB Directly Targets Prx1 and Grx3 via Covalent Conjugation to Their Active Cysteine Residues

To identify the PNQ molecular target(s), we employed a comparative affinity pulldown-based target identification strategy. Specifically, guided by the SAR studies described, a set of structurally-related but functionally distinct FB-based biotinylated probes were synthesized. Probe 1 retained FB-like activity (inhibition of 4E-BP1 phosphorylation and CRC cell line cytotoxicity), while the structurally similar ‘caged’ probe 2 lacked comparator activity (Figure 3A). Parallel incubation of probes 1 and 2 with the Hela cell lysates followed by comparative affinity pulldown and mass spectrometry-based proteomic analysis revealed Prx1 (22 kDa) and Grx3 (38 kDa) as the two highest MOWSE scoring proteins in the probe 1 pulldown sample unique to the probe 2 pulldown sample (Figure S5). As further support of the target identification, free excess FB could effectively compete with probe 1 binding to Prx1 or Grx3 using HCT116 crude extracts (Figure 3B), and probe 1 also efficiently labeled pure recombinant Prx1 and Grx3 (Figure 3C). Furthermore, using intact HCT116 cells treated with probes 1 and 2, only probe 1 was found to bind to Prx1 or Grx3 based on the subsequent pulldown assay (Figure 3D). Consistent with this, immunofluorescence co-staining with a target-selective Prx1 or Grx3 antibody (Figure S4) and a probe-selective (biotin) antibody also revealed co-localization of probe 1 with Prx1 or Grx3 in HCT116 cells (Figures 3E and 3F).

Figure 3. FB Binds Prxl and Grx3 and Potently Inhibits Prxl Activity.

(A) Structures of active probe 1 and inactive probe 2.

(B) HCT116 cell lysates were incubated with various concentrations of probe 1 or probe 2 in the absence or presence of a ten-fold excess of FB for 1 h, followed by pulldown with streptavidin agarose (SA) and Western blot analysis for the indicated proteins.

(C) The recombinant Prx1 (top) or Grx3 (bottom) was incubate with 1 μM probe 1 or probe 2 for 1 h, followed by SA pulldown and Western blot analysis for the indicated proteins.

(D-F) HCT116 cells were treated with 25 μM probe 1, 25 μM probe 2 or DMSO as control for 5 h, followed by SA pulldown and then western blot analysis for the precipitated proteins and whole cell lysates (WCL) (D), or by confocal sections of the cells staining for Prx1 (E) or Grx3(F) (red), biotin (green) and DAPI (blue). Scale bars, 20 μm.

(G) The catalytic activity of Prx1 was measured by glutamine synthase (GS) protection assay with various concentrations of FB and conoidin A.

(H) Kinetic measurement of FB against Prx1 catalytic activity. The k_obs values for interaction of Prx1 were determined with FB at various incubation times and then plotted the k_obs values with varying concentrations of FB.

Data are shown as mean ± SEM (n=3). See also Figures S4 and S5.

Prx1 and Grx3 are antioxidant proteins known to regulate oxidative stress. Using an in vitro Prx1-catalyzed H₂O₂ reduction assay, our results revealed that FB is a notably more potent inhibitor of Prx1 (apparent IC₅₀ 0.2 μM) than the commercial pan-Prx inhibitor conoidin A (Nguyen et al., 2013) (apparent IC₅₀ >20 μM) (Figure 3G). In addition, the FB inhibition kinetics (k_i 0.65 μM, k_inact 0.02 min⁻¹, Figure 3H) also exceeds that of the recently reported Prx1 inhibitor adenanthin (k_i 3.9 μM, k_inact 0.11 min⁻¹) (Liu et al., 2012). Human Grx3 is a non-catalytic essential 2Fe-2S center-containing protein involved in the regulation of signal transduction in response to redox signals and is also important to iron homeostasis (Gallogly et al., 2009; Haunhorst et al., 2010; Haunhorst et al., 2013; Herrero and de la Torre-Ruiz, 2007). Similar to the reported change in the GSH/GSSG ratio by depletion of Grx3 expression (Qu et al., 2011), a significant decrease in GSH levels and an increase in GSSG levels in HCT116 cells was observed upon treatment with FB and its active analogs (Figures S5A and S5B).

Consistent with the putative structural/functional role of Prx1/Grx3 cysteines (Gallogly et al., 2009; Haunhorst et al., 2010; Lee et al., 2007; Rhee et al., 2001), replacement of each cysteine with serine in Prx1 and Grx3 revealed that only Prx1 Cys83/Cys173 and Grx3 Cys159/Cys261 were essential for FB ligand-binding (Figures 4A–4D). Notably, these Cys residues are known active-site residues of Prx1 (Lee et al., 2007; Rhee et al., 2001) and the redox sensing 2Fe-2S ligating residues of Grx3 (Haunhorst et al., 2010). Consistent with this result, covalent modification of Prx1 Cys173 by adenanthin was also recently reported (Liu et al., 2012). FB C-4 alkylation by the side-chain thiol of Prx1 Cys173, Grx3 Cys159 or Grx3 Cys261 was further supported via mass spectrometry-based proteomic analysis and chemical model studies (Figures 4E, 4F and Supplemental Methods). Given that human Prx1 and Prx2 are 91% homologous and 78% identical in their amino acid sequences (Lee et al., 2007), we also tested the ability of FB to inhibit Prx2 in vitro. Based on this analysis, FB was found to be ~20-fold more potent against Prx1 than Prx2 (apparent IC₅₀ 4.2 μM; Figure S5C). It is noteworthy that Prx2 lacks the active-site Cys83 (Figure S5D) modified by FB in Prx1. Cumulatively, these studies provide strong validation of Prx1 and Grx3 as key molecular targets of PNQs in cancer cells, the inhibition of which occurs via selective covalent modification.

Figure 4. FB Directly Targets the Active Cysteine Residues of Prxl and Grx3.

(A-D) HCT116 cells were tranfected with Myc-tagged wild-type (WT) Prx1 (A and C) or Grx3 (B and D), their mutants, or control vector for 36 h. Cell lysates were incubated with 1 μM probe 1 for 1 h, followed by pulldown with streptavidin-agarose (A and B) or by immunoprecipitation with Myc-tag antibody (C and D), and then western blot analysis for the precipitated proteins and whole cell lysates (WCL).

(E) MS/MS analysis of the Cys173-containing tryptic peptide from recombinant Prx1 incubated with FB for 2h. The red character C represents the Cys bound by FB.

(F) MS/MS analysis of a FB-modified tryptic peptide from recombinant Grx3 incubated with FB for 2h. Since trypsin digestion produces the identical peptide CGFSK for both Cys159 and Cys261, the red character C represents the FB-modified Cys (Cys159 and/or Cys261) within this consensus sequence.

See also Figure S5.

FB And Active PNQs Induce ROS-Based Inhibition of mTORC1/4E-BP1 Signaling and Cancer Cell Growth

Consistent with Prx1 and Grx3 as the molecular targets, we found that FB and active surrogates (Figures 2B, 2C and S2) induced a marked increase in cellular H₂O₂ and ROS (Figures 5A and 5B). Reminiscent of the effects observed with FB treatment, direct treatment of HCT116 cells with H₂O₂ led to a concentration-dependent inhibition of 4E-BP1 phosphorylation and induction of cleaved PARP (Figure 5C), whereas pretreatment with the thiol-based antioxidant agent N-acetyl-L-cysteine (NAC) in HCT116 cells almost completely prevented ROS induction, the inhibitory effects on 4E-BP1 phosphorylation and cell growth, and induction of cleaved PARP by FB and active PNQs (Figures 5D–5F). Notably, overexpression of exogenous Prx1 or Grx3 also markedly reversed inhibition of 4E-BP1 phosphorylation and cap-dependent translation induced by FB (Figure S6). In contrast, silencing Prx1 or Grx3 expression increased cellular ROS levels and led to a concomitant suppression of cell growth and 4E-BP1 phosphorylation, and induction of cleaved PARP in HCT116 cells (Figures 5G–5J). The ROS accumulation and cell growth inhibition were further enhanced by knockdown of both Prx1 and Grx3 (Figures 5I and 5J). H₂O₂ and ROS have been reported to inhibit mTORC1 signaling by activating peroxisome-bound tuberous sclerosis complex (TSC) (Chen et al., 2009; Zhang et al., 2013a). Consistent with this, inhibition of 4E-BP1 phosphorylation and induction of cleaved PARP by FB or 12 were almost completely reversed in both peroxisome-deficient human Zellweger (GM13627) fibroblasts (Zhang et al., 2013a) and TSC2-knockdown HCT116 cells (Figures 5K and 5L). Together, these data are consistent with a model in which the inhibition of Prx1 and Grx3 by PNQs elevates intracellular ROS to a level sufficient to inhibit mTORC1-mediated 4E-BP1 phosphorylation and cancer cell growth (Figure 6E).

Figure 5. Inhibition of Prx1/Grx3 by FB and Active PNQs Increases ROS Accumulation Leading to Repression of mTORC1/4E-BP1 Signaling and Cancer Cell Growth.

(A) The extracellular H₂O₂ level was determined in HCT116 cells treated with 2 μM of the indicated compounds or DMSO control for 3 h.

(B) Flow cytometry analysis of ROS levels in HCT116 cells treated with 2 μM of the indicated compounds or DMSO control for 1 h.

(C) Western blot analysis of HCT116 cells treated with various concentrations of H₂O₂ for 2 h.

(D) Flow cytometry analysis of ROS levels in HCT116 cells pre-treated with 400 μM N-acetyl-L-cysteine (NAC) for 1 h before treatment with 2 μM FB or 12, or DMSO control for 1 h.

(E) Western blot analysis of HCT116 cells pre-treated with 400 μM mM NAC for 1 h before treatment with 2 μM of the indicated compounds or DMSO control for 6 h.

(F) HCT116 cells were pretreaated with 1 mM NAC for 1 h, and then incubated with various concentrations of FB or 12 for 72 h, followed by counting the viable cells.

(G and H) Western blot analysis of HCT116 cells with stable expression of two different sets of Prx1 (G), Grx3 (H) shRNAs or control shRNA.

(I and J) Flow cytometry analysis of ROS levels (I) and cell growth assay (J) in HCT116 cells with stable expression of Prx1 shRNA and Grx3 shRNA alone and in combination or control shRNA.

(K) Western blot analysis of human fibroblasts obtained from a Zellweger (GM13267) or a corresponding control patient (with Ehlers-Danlos syndrome (GM15871)) treated with 2 μM of FB or 12 for 6 h.

(L) Western blot analysis of HCT116 cells with stable expression of control shRNA or TSC2 shRNA treated with 2 μM of FB or 12 for 6 h.

Data are shown as mean ± SEM (n=3). *p < 0.01 versus DMSO; **p < 0.001; NS, not significant. See also Figures S2 and S6

Figure 6. Treatment with An Active FB Analog Suppresses Tumor Growth In Vivo.

(A) Synthesis of soluble prodrug 14 from 12.

(B) Mice bearing HCT116 or DLD-1 CRC xenograft tumors were treated with 14 (14 mg/kg, five times/week) or vehicle control (n=8 mice/group).

(C) The mouse body weight was measured twice per week over the time couse of the treatment in (B).

(D) Western blot analysis of representative HCT116 tumors collected from mice in (B) 6 h after the final treatment with 14 or vehicle control.

(E) A proposed model for the anticancer mechanism of FB through targeting Prx1/Grx3 to induce ROS accumulation leding to inhibition of mTORC1/4E-BP1-mediated tarnaltion and tumor growth.

Data are shown as mean ± SEM. *p < 0.001 for 14 versus vehicle control; NS, not significant. See also Figure S2

An Active FB Analog Effectively Suppresses Tumor Growth In Vivo

To assess whether this unique anticancer mechanism translates to in vivo efficacy and to improve formulation for the in vivo study, we utilized an aqueous soluble phosphate prodrug 14 synthesized from 12 in two steps (65% overall yield, Figure 6A). As exemplified by approved drugs such as osamprenavir, prednisolone sodium phosphate and fosphenytoin sodium salt, such phosphate prodrugs are cleaved via endogenous phosphatases (particularly abundant on the apical surface of enterocytes and in plasma) to afford rapid release of the active agent (Huttunen et al., 2011). Nude mice bearing established HCT116 or DLD-1 CRC xenografts were treated with 14 at a predetermined maximum tolerated dose, 14 mg/kg, daily for five days a week or saline control for 2 weeks. Administration of 14 was able to suppress overall tumor progression and caused 56% (DLD-1) or 64% (HCT116) tumor reduction without significant weight loss (Figures 6B and 6C). Western blot analysis of tumor extracts revealed effective inhibition of 4E-BP1 phosphorylation associated with induction of cleaved PARP by 14 (Figure 6D). These findings highlight the potential effectiveness of 14 in a suitable animal model for CRC and, consistent with the in vitro studies, implicate the inhibition of 4E-BP1 phosphorylation as a contributor to the mechanism of action in vivo (Figure 6E).

DISCUSSION

FB is a prototypical PNQ-based natural product first reported in the late 60’s (Ellestad et al., 1968) and has since been demonstrated to function as an effective anticoccidial and antimalarial (Fitzgerald et al., 2011), the fundamental mechanism(s) for which was unknown. The current study highlights that FB and PNQ-based analogs function as ligands for Prx1 and Grx3 where the active PNQs represent the most potent Prx1/Grx3 inhibitors reported to date.

Our mass spectrometry-based proteomic analysis clearly revealed FB to covalently modify Cys173 of Prx1 and the 2Fe-2S-binding Cys residues Cys159/Cys261 of Grx3 within its two Grx domains. In Grx3, the 2Fe-2S centers contribute to a redox-induced dissociation mechanism in response to reactive oxygen and nitrogen species where the equilibrium of apo-Grx3 (lacking the 2Fe-2S center) versus holo-Grx3 (containing the 2Fe-2S center) depends on cellular redox state (Haunhorst et al., 2010). Mutational analysis further identified the Cys83 of Prx1 as another target site of FB, although we were not be able to detect the Cys83-FB adduct ion under the mass spectrometry conditions that were tested. Prx1 Cys83 is critical to forming the catalytically active Prx1 dimer via a key interfacial Cys83-Cys83 disulfide bond. Despite its high sequence conservation with Prx1, Prx2 lacks Cys83 (Figure S5D) and this distinction may be an important contributor to the selectivity of FB for Prx1 (Figure S5C).

Covalent modification of Prx1 and Grx3 by FB leads to cellular effects that mimic silencing human Prx1 (Bajor et al., 2018) and Grx3 (Haunhorst et al., 2013; Qu et al., 2011); namely, a lower GSH/GSSG ratio, induction of ROS and, in the case of Grx3, may also alter iron uptake and/or utilization. Accordingly, FB and active PNQs induced an increase of the intracellular ROS to a level sufficient to inhibit cancer cell growth and survival associated with inhibition of mTORC1-mediated 4E-BP1 phosphorylation and cap-dependent translation. Blocking ROS induction by NAC markedly abolished FB and active PNQs-induced growth inhibition, ROS generation, and repression of mTORC1-mediated 4E-BP1 phosphorylation. While Prx1 and Grx3 are the predominate targets for FB, we cannot exclude the possibility that FB may also interact with other thiol-containing enzymes or proteins and inhibit other signaling pathways. Nevertheless, the corresponding mTORC1/4E-BP1 signaling axis is a major downstream pathway in response to FB-induced oxidative stress on growth inhibition, as both Prx1 and Grx3 knockdowns presented similar effects to those of FB on CRC cells, and overexpression of Prx1 and Grx3 or depletion of 4E-BP1 markedly or completely rescued inhibition of cap-dependent translation by FB and largely attenuated FB-induced apoptosis and growth inhibition. Collectively, our data strongly support Prx1 and Grx3 inhibition by FB as the primary mediator of ROS/4E-BP1-mediated cancer cell death. Given that active PNQs-induced ROS generation and cancer cell cytotoxicity correlate with inhibition of 4E-BP1 phosphorylation, our study also highlights the utility of 4E-BP1 phosphorylation as a potential predictive correlative sensor for guiding development of ROS-inducing agents and their anticancer activities.

Overexpression of Prx1 and Grx3 often occurs in a variety of cancers, and is associated with redox adaptation that promotes tumor progression and resistance to many anticancer agents and radiation (Cha and Kim, 2009; Iwao-Koizumi et al., 2005; Park et al., 2016; Qu et al., 2011; Trachootham et al., 2009; Woolston et al., 2011). In contrast, knockdown of Prx1 and Grx3 expression in cancer cells increases ROS levels, resulting in inhibition of proliferation, survival, invasion and metastasis and sensitization of cancer cells to chemotherapy and radiation (Bajor et al., 2018; Chen et al., 2006; Jiang et al., 2014; Poschmann et al., 2015; Qu et al., 2011). Agents that abrogate the adaptation mechanism to intrinsic oxidative stress in cancer cells, such as the PNQs described herein, thereby hold potential as an attractive new strategy to improve therapeutic outcomes, and can be explored further to develop a new class of anticancer agents .

SIGNIFICANCE

Frenolicins are prototypical pyranonaphthoquinone (PNQ) natural products known for their potent anticancer, anticoccidial and antimalarial activities, but the molecular mechanism of their action remains undefined. In this study, we discovered FB as the most potent Prx1/Grx3 inhibitor reported to date. Mass spectrometry and mutational analyses reveal that FB functions via selective covalent modification of Prx1 Cys83/Cys173 and Grx3 Cys159/Cys261. Our data reveal that the covalent modification of Prx1 and Grx3 by FB results in increased levels of intracellular ROS to induce apoptosis and suppress tumor growth by largely through inhibition of mTORC1-mediated 4E-BP1 phosphorylation. FB SAR studies further highlight the utility of 4E-BP1 phosphorylation as a potential predictive marker to guide development of ROS-inducing agents and their anticancer activities. Overexpression of Prx1 and Grx3 in a variety of cancers is associated with cancer cell adaption to intrinsic oxidative stress, tumor progression, and drug resistance. Our findings suggest that inhibitors of Prx1/Grx3 such as PNQs hold potential as a new class of anticancer agents to improve therapeutic outcomes. Given the longstanding anticoccidial use and recent demonstrated anti-malarial efficacy of FB, this work is also expected to stimulate new mechanistic studies and implicate new potential targets beyond the scope of cancer.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Qing-Bai She (qing-bai.she@uky.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Cell Lines

Human male colon (HCT116, DLD-1, HCT15, SW480), female colon (RKO), female breast (MDA-MB-231, MDA-MB-468) and female lung (H1975) cancer cell lines, and the BEAS-2B human male lung epithelial cell line were obtained from the American Type Culture Collection (ATCC) and cultured at 37°C/5% CO₂ in the appropriate medium with supplements as recommended by ATCC. The IMR-91 (male) and TIG-1 (female) human fetal lung fibroblasts, and fibroblast cells derived from the Zellweger male patient (GM13267) and control fibroblasts from the male patient with Ehlers-Danlos syndrome (GM15871) were obtained from Coriell Institute for Medical Research and cultured at 37°C/5% CO₂ in Eagle’s minimum essential medium supplemented with 15% FBS. The human male porcine aortic endothelial (PAE) cells (Guo et al., 2013) were kindly provided by Dr. Craig W. Vander Kooi and were grown in F12 medium containing 10% FBS. The wild-type 4E-BP1/4E-BP2 and 4E-BP1/4E-BP2 double knockout (DKO) female mouse embryonic fibroblasts (MEFs) (Dowling et al., 2010) were kindly provided by Dr. Nahum Sonenberg and were grown in DMEM medium containing 10% FBS. HCT116 cells with stable knockdown of 4E-BP1, TSC2 and its control stable transfectants were from our previous studies (Cai et al., 2014; Ye et al., 2014).

Animal Studies

Male athymic nude mice (5-6 weeks old) were purchased from Taconic. Experiments were carried out under a protocol approved by the University of Kentucky Institutional Animal Care and Use Committee. HCT116 and DLD-1 xenograft tumors were established by subcutaneously injecting 2 × 10⁶ cells in a 1:1 mixture of media and Matrigel (BD Biosciences) into the right flank. For efficacy studies, mice were randomized among control and treated groups (n=8 per group) when tumors were well-established (~120 mm³). Compound 14 was prepared freshly in saline and administered by intraperitoneal injection at 14 mg/kg once per day, Mon-Fri per week. Control mice received saline solution. Tumor dimensions were measured using a caliper and tumor volumes were calculated as mm³ = π/6 × larger diameter × (smaller diameter)². Tumors were excised and snap frozen in liquid nitrogen, homogenized in 2% SDS lysis buffer and then processed for Western blot analysis (She et al., 2010; Ye et al., 2014).

METHOD DETAILS

DNA Constructs and Transfection

The human Prx1 and Grx3 were amplified by PCR using a HCT116 cDNA library, and then subcloned into the pCMV6-Entry expression vector with C-terminal Myc-Flag Tag (OriGene #PS100001). Using the pCMV6-Prx1-Myc-Flag or pCMV6-Grx3-Myc-Flag as a template, the Prx1 mutant (C51S, C71S, C83S, C173S) and Grx3 mutant (C46S, C146S, C159S, C229S, C261S, C159S/C261S) constructs were generated using the QuikChange lightning site-directed mutagenesis kit (Agilent Technologies). The primers used are listed in Table S1. All constructs were confirmed using enzyme digestion and automated DNA sequencing. For transient transfection, cells were transiently transfected with DNA using Lipofectamine 3000 according to the manufacturer’s protocol (Thermo Fisher Scientific).

shRNAs and Lentiviral Infection

The lentiviral-based shRNA (pLKO.1 plasmids) used to knock down expression of human Prx1 and Grx3, and the Non-Target Control shRNA were purchased from Sigma-Aldrich. 293T cells were used for packaging of lentiviral shRNA-expressing viruses, and subsequent infection of HCT116 cells were performed (Wang et al., 2017). In brief, medium with secreted viruses was collected three times at 48 h after transfection. After filtering through 0.45 μm filters, viruses were used to infect cells in the presence of 8 μg/ml polybrene (Sigma-Aldrich). Cells with stable knockdown of Prx1 and/or Grx3 were selected using puromycin (2 μg/ml) and/or hygromycin (250 μg/ml) for 7-10 days.

Western Blot Analysis and Immunoprecipitation

Cells were lysed in NP-40 lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% NP-40, 10% glycerol, protease and phosphatase inhibitor cocktail). Protein concentrations were measured using the BCA protein assay reagent (Thermo Fisher Scientific). Equal amounts of protein were resolved by SDS-PAGE, transferred to PVDF membranes, immunoblotted with specific primary and secondary antibodies, and detected using chemiluminescence (GE Healthcare). For immunoprecipitations analysis, cell lysates (500-1000 μg protein) were incubated with 2 μg of the indicated antibody overnight followed by incubation with a 50% slurry of protein G sepharose beads for 3 h at 4°C. The beads were washed three times with the lysis buffer, and the immunoprecipitated protein complexes were resuspended in 2× Laemmli sample buffer followed by western blot analysis.

Cap-Dependent Translation Assay

Cells (8×10⁴) were transfected with a bicistronic luciferase reporter plasmid (0.5 μg), pcDNA3-rLuc-PolioIRES-fLuc, which directs cap-dependent translation of the Renilla luciferase gene and cap-independent Polio IRES-mediated translation of the firefly luciferase gene (She et al., 2010; Wang et al., 2017). After 24 h transfection, cells were treated with indicated compounds for 12 h, and cell lysates were assayed for Renilla and firefly luciferase activities using a dual-luciferase assay kit (Promega). Cap-dependent Renilla luciferase activity was normalized against cap-independent firefly luciferase activity as the internal control. The ratio of Renilla/firefly luciferase activity was calculated for cap-dependent translational activity (She et al., 2010; Wang et al., 2017). Each experiment was performed in triplicate and repeated at least three times.

Cap-Binding Assay

Cap-binding assay was performed as described previously (She et al., 2010). Cell lysates (500 μg protein) as prepared in the NP-40 lysis buffer were incubated at 4 °C overnight with m⁷GTP Sepharose beads (GE Healthcare Life Sciences) to capture eIF4E and its binding partners. The beads were washed three times with the lysis buffer, and the precipitated protein complexes were resuspended in 2× Laemmli sample buffer followed by western blot analysis.

SUnSET Assay

The SUnSET assay was used to monitor the rate of protein synthesis as described (Schmidt et al., 2009). Cells were incubated with 1 μM puromycin (Sigma) for 30 minutes followed by washing twice with ice cold PBS and lysing with the NP-40 lysis buffer. The lysates were subjected to western blot analysis using mouse anti-puromycin monoclonal antibody (Kerafast). Signal was normalized against Coomassie blue staining.

Cell Viability and Apoptosis Assays

Cells (5×10⁴/well) were seeded in 6-well plates in triplicate. After 24 h, cells were treated with the indicated compounds for 3 days, and the number of viable cells was counted using the Vi-CELL XR 2.03 (Beckman Coulter). To measure apoptosis, both adherent and floating cells were harvested after treatment with the indicated compounds for 72 h, followed by flow cytometric analysis using the annexin V-FITC apoptosis detection kit (BD Biosciences) according to the manufacturer’s protocol. Each experiment was performed in triplicate and repeated at least three times.

Pulldown and MS Analysis of FB-Bound Proteins

To identify the target protein for FB, FB-based biotinylated inactive probe 1 and active probe 2 were synthesized as described below. HeLa cell pellets were purchased from National Cell Culture Center and lysed in 20 ml of the NP-40 lysis buffer as indicated above. The cell lysates (250 mg protein) were pre-cleared with 200 μl streptavidin beads (Thermo Fisher Scientific) at 4°C for 1 h. Binding reactions were performed by incubating the pre-cleared cell lysates (125 mg proteins/40 ml) with 2 μM probe 1 or probe 2 at 4°C for 3 h, followed by adding 100 μl streptavidin beads and incubating the mixtures overnight at 4°C. After incubation, the beads were washed four times with the lysis buffer, and the bead-bound proteins were eluted, followed by digestion, LC-MS/MS analysis and MASCOT protein identification as previously reported (Yang et al., 2014). A decoy database was built and searched to determine the false discovery rates (FDR) and peptides with FDR lower than 0.01 were assigned as high confidence identification. The probability-based MOWSE scores calculated by MASCOT were used to determine the predominant protein in the sample.

MS/MS Analyses of Prx1-FB and Grx3-FB Complexes

Recombinant Prx1 (60 μg, Thermo Fisher Scientific) or Grx3 (60 μg, MyBiosource) was reduced using an excess of dithiothreitol (DTT, 60 mM, Sigma-Aldrich) in a 25 mM Tris-HCl buffer (pH 7.5) in a total volume of 300 μl at 37°C for 1 h. DTT was removed using a G25 Sephadex column (GE Healthcare Life Sciences) and the DTT-free, reduced Prx1 (30 μg) or Grx3 (30 μg) was incubated with 200 μM FB or buffer alone in a total volume of 300 μl at 37°C for 2 h. The unreacted FB was subsequently removed by chromatography over a G25 Sephadex column and the recovered Prx1 or Grx3 was digested by trypsin and analyzed by LC-MS/MS.

Prx1 and Prx2 Activity Assays

The activity of Prx1 and Prx2 was determined using slight modifications of a previously reported glutamine synthetase (GS) protection assay (Haraldsen et al., 2009; Kim et al., 1988). Specifically, varying concentrations of FB or conoidin A (0.032 μM-30 μM) were incubated with 0.2 μg DTT-reduced Prx1 or Prx2 (Thermo Fisher Scientific) as indicated above in a 100 mM HEPES buffer (pH 7.4) in a total volume of 16 μL at room temperature for 15 min. GS (1 μl of 1 unit/μl stock in HEPES buffer, pH 7.4) and 3 μl of inactivation solution (50 mM DTT, 5 μM FeCl₃, 100 mM HEPES, pH 7.4) were added and the mixture was incubated at 30°C for 20 min. Initiation solution (150 μl total; 100 mM HEPES, 10 mM KH₂AsO₄, 20 mM NH₂OH·HCl, 0.4 mM ADP, 0.5 mM MnCl₂, 100 mM glutamine, pH 7.0) was subsequently added and the mixture incubated at 30°C for an additional 30 min. Finally, the reaction was terminated by the addition of termination solution (80 μl total; 5.5% FeCl₃·6H₂O, 2% TCA, 2% concentrated HCl), after which the absorption at 540 nm was measured and normalized to a corresponding blank lacking inhibitor. Assays were conducted in triplicate and the IC₅₀ values were calculated using Prism software from the curve fits highlighted in Figure 3G. Using the described conditions, linear signal range was observed with assays containing 0.2-0.4 μg Prx1 and FB was also confirmed to have no direct effect on GS activity (highest concentration tested 30 μM). k_i and k_inact were measured from identical assays but variable incubation times following standard methods (0.5-15 min) (Singh et al., 1997). Assays were conducted in triplicate and k_i values were calculated using Prism software from the curve fits highlighted in Figure 3H.

Immunofluorescence

Cells grown on glass bottom culture dishes were incubated with 25 μM probe 1, 25 μM probe 2 or DMSO as control for 5 h. After treatment, cells were fixed with 4% paraformaldehyde in PBS for 15 min, permeabilized in 0.2% Triton X-100 and 0.5% BSA in PBS for 5 min and then blocked with 4% BSA in PBS for 10 min. The cells were incubated overnight at 4 °C with the rabbit polyclonal antibody against Prx1 (1:200, Abcam) and mouse monoclonal anti-Biotin-FITC (1:200, Jackson ImmunoResearch), or with the mouse monoclonal antibody against Grx3 (1:200, R&D Systems) and rabbit polyclonal anti-Biotin-FITC (1:200, Abcam). After three washes with 0.05% Triton X-100 in PBS, cells were incubated with anti-rabbit secondary antibody conjugated with Texas-Red for Prx1 (1:500, Jackson ImmunoResearch) or anti-mouse secondary antibody conjugated with Texas-Red for Grx3 (1:500, Jackson ImmunoResearch) for 1 h. Cells were washed, mounted with DAPI containing mounting medium (Vector Laboratories), viewed, and photographed under a Nikon A1⁺-Ti2 confocal microscope.

Cellular GSH Assay

The cellular GSH level was determined using the glutathione fluorometric assay kit (BioVision Research Products) according to the manufacturer’s protocol. Cells were treated with 2 μM of the indicated compounds or DMSO as control for 5 h. After treatment, a total number of 1×10⁶ cells were lysed in 100 μl of ice-cold lysis buffer for 10 min. The lysate was centrifuged for 10 min and the supernatant was used for GSH assay. The total amount of GSH was measured using a fluorescence plate reader at Ex./Em. = 380/460 nm. Each experiment was performed in triplicate and repeated at least three times.

Cellular GSSG Assay

The cellular GSSG level was determined using a microplate assay kit for GSH/GSSG (Oxford Biomedical Research) according to the manufacturer’s protocol. Cells were treated with 2 μM of the indicated compounds or DMSO as control for 5 h. After treatment, a total number of 0.5×10⁶ cells were collected in 1.5 ml centrifuge tubes containing ice-cold lysis buffer with the thiol scavenger to remove GSH and keep GSSG in its oxidized form. The cells were homogenized with a Teflon pestle and the cell suspension sonicated in icy water for 2-3 minutes. Ice-cold metaphosphoric acid was added to deproteinate the samples. The samples were centrifuged at 1000×g at 4°C and the supernatants were used for determining the GSSG concentration using a microplate reader with 405 nm filter. Each experiment was performed in triplicate and repeated at least three times.

Measurement of H₂O₂ Production

The H₂O₂ released from cells was determined using the amplex red hydrogen peroxide assay kit (Thermo Fisher Scientific) according to the manufacturer’s protocol. In the presence of peroxidase, the amplex red reagent (10-acetyl-3,7-dihydroxyphenoxazine) reacts with H₂O₂ in a 1:1 stoichiometry to produce the red-fluorescent oxidation product, resorufin. Cells (2×10⁴/well) were seeded in 96-well plates in triplicate. After 24 h, media were removed and cells were treated with 2 μM of the indicated compounds or DMSO as control for 3 h. Fifty microliters of the amplex red reaction mixture (100 μM amplex red reagent and 0.2 U/ml horseradish peroxidase in phosphate buffered saline, pH 7.4) was added to each well, followed by incubation at 37°C for 45 min. Amplex red conversion to resorufin was measured for absorbance at ~560 nm using Molecular Devices The SpectraMax® Plus 384 microplate reader. Each experiment was repeated at least three times.

Measurement of ROS Production

The ROS production was determined using the CellROX deep red flow cytometry assay kit (Thermo Fisher Scientific) according to the manufacturer’s protocol. Cells (5×10⁵) were treated with 2 μM of the indicated compounds or DMSO as control for 1 h, followed by incubation with 0.5 μM CellROX deep red reagent at 37°C for 45 min. The fluorescence of the CellROX deep red reagent upon oxidation was analyzed by flow cytometry using a FACSDiva (BD Biosciences). Each experiment was performed in triplicate and repeated at least three times.

General Chemistry

¹H (400 MHz) and ¹³C (100 MHz) NMR spectra were recorded on a Varian Unity Inova 400 MHz instrument. The chemical shifts were reported in δ (ppm) using the δ 7.26 signal of CDCl₃, δ 1.94 signal of CD3CN and δ 2.50 signal of DMSO-d₆ (¹H NMR), the δ 77.16 signal of CDCl₃, δ 1.32 signal of CD₃CN and δ 39.52 signal of DMSO-d₆ (¹³C NMR) as internal standards. The following abbreviations were used to explain the multiplicities: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet. HR-ESI-MS experiments were carried out using AB SCIEX TripleTOF® 5600 System. HPLC analyses were performed using an Agilent 1260 system equipped with a DAD detector and a Phenomenex C18 column (4.6 × 150 mm, 0.5 μm). Semipreparative/preparative HPLC separation was performed using a Varian Prostar 210 HPLC system equipped with a PDA detector 330 using a Supelco C18 column (25 × 21.2 mm, 10 μm; flow rate, 10 ml/min). Enantiomeric excess was determined by HPLC with a Chiralpak IC column, compared with racemic isomer. All commercially available reagents were used without further purification, purchased from Sigma-Aldrich, TCI America and Alfa-Aesar. The progress of the reactions was monitored by analytical thin-layer chromatography (TLC) from EMD Chemicals Inc. with fluorescence F₂₅₄ indicator. Silica gel (230–400 mesh) for column chromatography was purchased from Silicycle. FB and UCF76-A were isolated from Streptomyces sp. RM-4-15 as previously described (Wang et al., 2014; Wang et al., 2013). The syntheses of compounds 1-14 followed previously reported strategies (Zhang et al., 2013b) and are detailed below. Compound purity for all studies was ≥95% based on HPLC and all compound stock solutions were standardized to reference standards based on HPLC and UV-vis.

Synthesis of Compounds 1-14

(3aR,5S,11bR)-5-(3-Azidopropyl)-7-methoxy-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromene-2,6,11-trione (6). To a solution of (3aR,5S,11bR)-5-(3-azidopropyl)-6,7,11-trimethoxy-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromen-2-one (reported intermediate 6h (Zhang et al., 2013)) (41 mg, 0.1 mmol) in a mixture of water (0.5 ml) and acetonitrile (1 ml) at 0 °C, a solution of cerium ammonium nitrate (126 mg, 0.2 mmol) in H₂O (0.5 ml) was added in dropwise fashion with stirring. The reaction mixture was stirred for 10 min before the addition of water (5 ml). The mixture was extracted with EtOAc (10 ml × 2) and the combined organic layers were washed with brine, dried over Na2SO4, filtered, and concentrated. The residue was purified on silica gel using hexane/EtOAc (1/1) to afford 6 as a yellow solid (31 mg, 80% yield). ¹H NMR (400 MHz, CDCl₃): δ = 7.77-7.69 (m, 2H), 7.32 (d, J = 8.0 Hz, 1H), 5.30 (s, 1H), 4.77 (d, J = 9.6 Hz, 1H), 4.34 (d, J = 7.2 Hz, 1H), 4.01 (s, 3H), 3.27 (t, J = 6.4 Hz, 2H), 2.91 (dd, J = 4.4, 17.6 Hz, 1H), 2.73 (d, J = 17.6 Hz, 1H), 2.17-2.13 (m, 1H), 1.99-1.92 (m, 1H), 1.76-1.65 (m, 2H); ¹³C NMR (100 MHz, CDCl₃) δ = 183.4, 182.3, 174.5, 159.6, 151.3, 135.6, 133.7, 133.3, 120.3, 119.5, 118.3, 72.1, 71.3, 69.7, 56.7, 51.3, 37.4, 30.6, 24.7 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₁₉H₁₈N₃O₆ 384.1196, found 384.1199.

(3aR,5S,11bR)-5-(3-Azidopropyl)-7-hydroxy-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromene-2,6,11-trione (1). Compound 6 (25 mg, 0.065 mmol) was dissolved in CH₂Cl₂ (1 ml) and then cooled to −78 °C under argon. A solution of BCl₃ (0.1 ml, 0.1 mmol, 1 N in CH₂Cl₂) was added to the mixture with stirring for 2 hours at −78 °C. After quenching with saturated aqueous NH₄Cl solution (1 ml), the reaction was diluted with H₂O (5 ml) and EtOAc (5 ml). The organic layer was washed with brine, dried over Na₂SO₄, filtered, and concentrated. The residue was purified on silica gel using hexane/EtOAc (2/1) to afford 1 as an orange solid (19 mg, 80% yield). ¹H NMR (400 MHz, CDCl₃): δ = 11.71 (s, 1H), 7.71-7.66 (m, 2H), 7.31 (dd, J = 2.4, 7.2 Hz, 1H), 5.28 (t, J = 2.0 Hz, 1H), 4.80-4.79 (m, 1H), 4.35 (dd, J = 2.4, 4.4 Hz, 1H), 3.30 (t, J = 6.4 Hz, 2H), 2.88 (dd, J = 4.4, 17.6 Hz, 1H), 2.75 (d, J = 17.6 Hz, 1H), 2.27-2.25 (m, 1H), 2.05-2.03 (m, 1H), 1.75-1.72 (m, 1H), 1.64-1.62 (m, 1H); ¹³C NMR (100 MHz, CDCl₃) δ = 188.6, 181.4, 174.2, 161.9, 148.9, 137.3, 136.7, 131.4, 125.1, 119.9, 115.1, 71.6, 71.1, 69.7, 51.2, 37.3, 31.1, 24.5 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₁₈H₁₆N₃O₆ 370.1039, found 370.1041.

epi-Frenolicin B (2). The synthesis and characterization data of 2 were previously reported (Zhang et al., 2013).

(3aS, 5S, 11bS)-7-Hydroxy-5-propyl-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromene-2,6,11-trione (3). This compound was synthesized according to the reported procedures (Zhang et al., 2013). ¹H NMR (400 MHz, CDCl₃): δ = 11.85 (s, 1H), 7.71-7.65 (m, 2H), 7.30 (dd, J = 2.0, 8.0 Hz, 1H), 5.25 (t, J = 3.2 Hz, 1H), 4.91 (dd, J = 3.2, 10.4 Hz, 1H), 4.62 (dd, J = 2.8, 5.2 Hz, 1H), 2.96 (dd, J = 5.2, 17.6 Hz, 1H), 2.70 (d, J = 17.6 Hz, 1H), 1.71-1.64 (m, 6H), 1.03 (t, J = 7.2 Hz, 3H); ¹³C NMR (100 MHz, CDCl₃) δ = 188.2, 181.6, 174.0, 162.0, 149.4, 137.3, 135.3, 131.6, 124.9, 119.8, 114.9, 69.7, 68.8, 66.3, 36.9, 33.8, 19.6, 13.6 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₁₈H₁₇O₆ 329.1025, found 329.1025.

(3aR,5S,11bR)-8-Bromo-6,7,11-trimethoxy-5-propyl-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromen-2-one (4a) and (3aR,5S,11bR)-10-bromo-6,7,11-trimethoxy-5-propyl-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromen-2-one (5a). NBS (53 mg, 0.3 mmol) was added by portion to (3aR,5S,11bR)-6,7,11-trimethoxy-5-propyl-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromen-2-one (reported intermediate 6aα (Zhang et al., 2013)) (111 mg, 0.3 mmol) in CH₂Cl₂ (3 ml) at room temperature and the resulting mixture was stirred overnight. After evaporating the volatiles, the residue was purified on silica gel using hexane/EtOAc (6/1) to afford the 4a (40 mg, 29%) and 5a (90 mg, 65%).

(3aR,5S,11bR)-8-Bromo-6,7,11-trimethoxy-5-propyl-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromen-2-one (4a). ¹H NMR (400 MHz, CDCl₃): δ = 7.75 (d, J = 8.0 Hz, 1H), 7.64 (d, J = 8.0 Hz, 1H), 5.57 (s, 1H), 5.05 (d, J = 6.0 Hz, 1H), 4.35 (s, 1H), 4.09 (s, 3H), 3.88 (s, 3H), 3.74 (s, 3H), 2.91 (dd, J = 4.0, 18.4 Hz, 1H), 2.77 (d, J = 17.6 Hz, 1H), 2.15-2.13 (m, 1H), 2.03-1.99 (m, 1H), 1.76-1.65 (m, 2H), 0.98 (t, J = 7.2 Hz, 3H); ¹³C NMR (100 MHz, CDCl₃) δ = 175.9, 153.7, 152.5, 147.7, 130.7, 129.8, 124.8, 120.2, 117.5, 107.3, 107.0, 73.2, 73.0, 71.1, 65.0, 62.2, 56.4, 38.5, 37.4, 18.4, 14.1 ppm.

(3aR,5S,11bR)-10-Bromo-6,7,11-trimethoxy-5-propyl-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromen-2-one (5a). ¹H NMR (400 MHz, CDCl₃): δ = 7.72 (d, J = 8.0 Hz, 1H), 6.74 (d, J = 8.0 Hz, 1H), 5.60 (s, 1H), 5.05 (d, J = 6.0 Hz, 1H), 4.34 (s, 1H), 4.11 (s, 3H), 3.99 (s, 3H), 3.96 (s, 3H), 2.90 (d, J = 18.4 Hz, 1H), 2.77 (d, J = 17.6 Hz, 1H), 2.15-2.13 (m, 1H), 1.99-1.96 (m, 1H), 1.76-1.66k (m, 2H), 0.91 (t, J = 7.2 Hz, 3H); ¹³C NMR (100 MHz, CDCl₃) δ = 175.9, 156.0, 152.9, 149.5, 133.9, 129.6, 127.1, 126.7, 122.1, 107.8, 107.0, 73.1, 72.8, 71.1, 65.8, 61.8, 56.7, 38.7, 37.8, 18.5, 14.0 ppm.

(3aR,5S,11bR)-8-Bromo-7-hydroxy-5-propyl-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromene-2,6,11-trione (4). Following the above deprotection protocol for 1, 4a (40 mg, 0.09 mmol) was used to obtain compound 4 (21 mg, 60% yield). ¹H NMR (400 MHz, CDCl₃): δ = 12.38 (s, 1H), 7.95 (d, J = 8.0 Hz, 1H), 7.56 (d, J = 8.0 Hz, 1H), 5.26 (t, J = 2.0 Hz, 1H), 4.77-4.75 (m, 1H), 4.33 (dd, J = 2.4, 4.4 Hz, 1H), 2.90 (dd, J = 4.4, 17.6 Hz, 1H), 2.74 (d, J = 17.6 Hz, 1H), 2.00-1.90 (m, 2H), 1.44-1.42 (m, 1H), 1.28-1.25 (m, 1H), 0.90 (t, J = 7.2 Hz, 3H); ¹³C NMR (100 MHz, CDCl₃) δ = 188.8, 180.9, 174.4, 158.3, 149.6, 140.4, 136.7, 130.4, 120.1, 119.9, 115.5, 72.0, 71.0, 69.7, 37.4, 36.0, 18.4, 14.1 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₁₈H₁₆BrO₆ 407.0130, found 407.0117.

(3aR,5S,11bR)-10-Bromo-7-hydroxy-5-propyl-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromene-2,6,11-trione (5). Following the above deprotection protocol for 1, 5a (90 mg, 0.2 mmol) was used to obtain compound 5 (43 mg, 53% yield). ¹H NMR (400 MHz, CDCl₃): δ = 12.28 (s, 1H), 7.83 (d, J = 9.2 Hz, 1H), 7.11 (d, J = 9.2 Hz, 1H), 5.33 (t, J = 2.0 Hz, 1H), 4.75-4.73 (m, 1H), 4.37 (dd, J = 2.4, 4.4 Hz, 1H), 2.94 (dd, J = 4.4, 17.6 Hz, 1H), 2.73 (d, J = 17.6 Hz, 1H), 2.03-2.00 (m, 1H), 1.89-1.85 (m, 1H), 1.44-1.42 (m, 1H), 1.29-1.27 (m, 1H), 0.88 (t, J = 7.2 Hz, 3H); ¹³C NMR (100 MHz, CDCl₃) δ = 187.9, 180.0, 174.4, 162.0, 148.6, 143.9, 136.9, 128.2, 125.6, 116.5, 113.7, 71.6, 71.1, 69.5, 37.2, 35.7, 18.4, 14.0 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₁₈H₁₆BrO₆ 407.0130, found 407.0118.

9-Methyl-frenolicin B (7). The synthesis of 7 was previously reported (Zhang et al., 2013). ¹H NMR (400 MHz, CDCl₃): δ = 7.81 (dd, J = 1.2, 8.0 Hz, 1H), 7.72 (t, J = 8.0 Hz, 1H), 7.32 (dd, J = 1.2, 8.4 Hz, 1H), 5.25 (d, J = 2.8 Hz, 1H), 4.87 (m, dd, J = 3.2, 10.8 Hz, 1H), 4.60 (dd, J = 3.2, 5.2 Hz, 1H), 4.03 (s, 3H), 2.94 (dd, J = 5.2, 17.6 Hz, 1H), 2.69 (d, J = 17.6 Hz, 1H), 1.83-1.81 (m, 1H), 1.66-1.57 (m, 3H), 1.00 (t, J = 7.2 Hz, 3H); ¹³C NMR (100 MHz, CDCl₃) δ = 182.7, 182.3, 174.3, 160.2, 150.6, 135.8, 133.8, 132.6, 119.7, 119.5, 118.3, 70.4, 69.1, 66.3, 56.7, 37.0, 19.7, 13.6 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₁₉H₁₉O₆ 343.1182, found 343.1180.

2-(1R,3R,4R)-4,9-Dihydroxy-5,10-dioxo-1-propyl-3,4,5,10-tetrahydro-1H-benzo[g]isochromen-3-yl)acetic acid (8). A solution of frenolicin B (10 mg, 0.03 mmol) in DMSO (1 ml) was added to HEPES buffer (8 ml, 50 mM, pH = 9.5) in a dropwise fashion at room temperature. The resulting mixture was incubated at 37 °C overnight with brief shaking. After neutralization with 1N HCl to pH = 7, the mixture was extracted with Et₂O (10 ml × 2). The collected organic layers were washed with brine, dried over Na₂SO₄, concentrated and purified on preparative HPLC (40%-100% CH₃CN/H₂O, 20 min, then 100% CH₃CN, 5 min) to give 8 as a yellow solid (8 mg, 80%). ¹H NMR (400 MHz, CDCl₃): δ = 11.91 (s, 1H), 7.65-7.63 (m, 2H), 7.28-7.62 (m, 1H), 4.86 (dd, J = 2.0, 10.8 Hz, 1H), 4.69 (d, J = 2.4 Hz, 1H), 4.31 (s, 1H), 3.49-3.46 (m, 1H), 2.91-2.89 (m, 2H), 2.65 (s, 1H), 1.74-1.65 (m, 6H), 1.02 (t, J = 7.2 Hz, 3H); ¹³C NMR (100 MHz, CDCl₃) δ = 189.0, 183.5, 175.4, 161.8, 146.6, 141.0, 136.8, 131.6, 125.0, 119.5, 114.9, 71.0, 67.1, 60.3, 35.4, 33.0, 19.8, 13.7 ppm; HRMS (ESI) m/z [M - H]⁻ calcd for C₁₈H₁₇O₇ 345.0974, found 345.0971.

(3aR,5S,11bR)-7-hydroxy-5-propyl-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromene-6,11-dione (9). To a stirred solution of (3aR,5S,11bR)-6,7,11-trimethoxy-5-propyl-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromen-2-one (reported intermediate 6aα (Zhang et al., 2013)) (93 mg, 0.25 mmol) in CH₂Cl₂ (2.5 ml) at −78°C was added DIBAL-H (0.5 ml, 1 M in toluene). The reaction was quenched with saturated aqueous potassium sodium tartrate (1 ml) after 1 hour. The mixture allowed warm to room temperature with stirring, extracted with CH₂Cl₂ (10 ml × 2) and the combined organics washed with brine, dried over Na₂SO₄, and concentrated. The residue was dissolved in CH₂Cl₂ containing trifluoroacetic acid (58 μL, 0.75 mmol) and cooled to −78 °C to which triethylsilane (119 μL, 0.75 mmol) was added in dropwise fashion. The resulting mixture was allowed to warm to room temperature with stirring overnight. After evaporating the volatiles, the residue was purified on silica gel using hexane/EtOAc (15/1) to obtain the tetrahydrofuran 9a (65 mg, 73% for two steps) as a colorless oil.

Following the above deprotection protocol for 1, 9a (65 mg, 0.18 mmol) was used to obtain 9 (28 mg, 50% yield). ¹H NMR (400 MHz, CDCl₃): δ = 11.84 (s, 1H), 7.68-7.59 (m, 2H), 7.26-7.24 (m, 1H), 4.71-4.68 (m, 1H), 4.59 (t, J = 2.0 Hz, 1H), 4.16-4.10 (m, 3H), 2.24-2.20 (m, 1H), 2.04-1.99 (m, 1H), 1.51-1.46 (m, 1H), 1.32-1.28 (m, 1H), 0.91 (t, J = 7.2 Hz, 3H); ¹³C NMR (100 MHz, CDCI3) δ = 189.7, 182.7, 161.5, 148.0, 139.6, 136.7, 131.9, 124.4, 119.4, 115.1, 74.9, 72.2, 70.2, 67.7, 36.1, 33.6, 18.3, 14.2 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₁₈H₁₉O₅ 315.1232, found 315.1224.

(3aR,11bR)-7-Hydroxy-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromene-2,6,11-trione (10). Following the above deprotection protocol for compound 1, (3aR,11bR)-6,7,11-trimethoxy-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromen-2-one (reported intermediate 14 (Zhang et al., 2013)) (33 mg, 0.1 mmol) was used to obtain 10 (16 mg, 54% yield). ¹H NMR (400 MHz, CDCl₃): δ = 11.71 (s, 1H), 7.70-7.68 (m, 2H), 7.30 (dd, J = 2.0, 7.6 Hz, 1H), 5.26 (t, J = 2.0 Hz, 1H), 4.95 (d, J = 18.8 Hz, 1H), 4.47-4.39 (m, 2H), 2.95 (dd, J = 2.0, 17.6 Hz, 1H), 2.76 (d, J = 17.6 Hz, 1H); ¹³C NMR (100 MHz, CDCl₃) δ = 188.0, 173.8, 161.9, 146.1, 137.5, 137.4, 135.8, 131.6, 124.9, 120.0, 114.7, 72.5, 69.1, 61.4, 37.0 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₁₅H₁₁O₆ 287.0556, found 287.0546.

(3aR,5S,11bR)-5-Ethyl-7-hydroxy-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromene-2,6,11-trione (11). Following the above deprotection protocol for compound 1, (3aR,5S,11bR)-5-ethyl-6,7,11-trimethoxy-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromen-2-one (reported intermediate 6b (Zhang et al., 2013)) (36 mg, 0.1 mmol) was used to obtain 11 (19 mg, 61% yield). ¹H NMR (400 MHz, CDCl₃): δ = 11.75 (s, 1H), 7.68-7.66 (m, 2H), 7.30-7.28 (m, 1H), 5.27 (s, 1H), 4.74 (s, 1H), 4.34 (s, 1H), 2.90 (dd, J = 4.0, 17.6 Hz, 1H), 2.75 (d, J = 17.6 Hz, 1H), 2.14-2.03 (m, 2H), 0.89 (t, J = 7.2 Hz, 3H); ¹³C NMR (100 MHz, CDCl₃) δ = 188.7, 181.5, 174.5, 161.8, 149.5, 137.2, 136.7, 131.5, 124.9, 119.7, 115.1, 72.8, 70.9, 69.8, 37.4, 27.1, 9.2 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₁₇H₁₅O₆ 315.0869, found 315.0857.

(3aR,5S,11bR)-10-Chloro-7-hydroxy-5-(3,3,3-trifluoropropyl)-3,3a,5,11b-tetrahydro-2H-benzo[g]furo[3,2-c]isochromene-2,6,11-trione (12). To a solution of (4R,5R)-4-hydroxy-5-(1,4,5-trimethoxynaphthalen-2-yl)dihydrofuran-2(3H)-one (reported intermediate 5 (Zhang et al., 2013)) (192 mg, 0.6 mmol) and 4,4,4-trifluorobutanal (152 mg, 1.2 mmol) in anhydrous CH₂Cl₂ at 0 °C, Cu(OTf)₂ (108 mg, 0.3 mmol) was added with stirring. The temperature was allowed to rise to room temperature and the mixture was stirred for 16 hours. After evaporating the volatiles, diastereoselectivity of the crude mixture was evaluated via NMR and then purified on silica gel using hexane/EtOAc (3/1-2/1) to give 12a as a colorless solid (200 mg, 81% yield, >20:1 dr ratio). ¹H NMR (400 MHz, CDCl₃): δ = 7.73 (dd, J = 1.2, 8.4 Hz, 2H), 7.46 (t, J = 8.4 Hz, 1H), 6.95 (d, J = 7.6 Hz, 1H), 5.58 (d, J = 2.4 Hz, 1H), 5.12-5.10 (m, 1H), 4.37 (dd, J = 2.4, 4.0 Hz, 1H), 4.09 (s, 3H), 4.02 (s, 3H), 3.75 (s, 3H), 2.92 (dd, J = 4.4, 17.2 Hz, 1H), 2.77 (d, J = 17.6 Hz, 1H), 2.63-2.60 (m, 1H), 2.31-2.20 (m, 2H), 2.06-2.02 (m, 1H); ¹³C NMR (100 MHz, CDCl₃) δ = 175.5, 156.3, 153.4, 149.6, 130.6, 127.0, 126.2, 126.0, 121.7, 119.4, 115.2, 107.7, 72.9, 71.9, 71.3, 64.6, 61.7, 56.4, 38.3, 29.6 (q), 27.7 ppm.

N-chlorosuccinamide (68 mg, 0.51 mmol) was added to a CH₂Cl₂ solution of 12a (200 mg, (0.47 mmol) at room temperature. The resulting mixture was heated to 80 °C with stirring for 24 hours. After evaporating the volatiles, the residue was purified on silica gel using hexane/EtOAc (5/1) to afford 12b (200 mg, 93% yield). ¹H NMR (400 MHz, CDCl₃): δ = 7.48 (d, J = 8.4 Hz, 2H), 6.83 (d, J = 8.4 Hz, 1H), 5.58 (d, J = 2.0 Hz, 1H), 5.12-5.09 (m, 1H), 4.35 (dd, J = 2.4, 4.0 Hz, 1H), 3.98 (s, 3H), 3.97 (s, 3H), 3.7 (s, 3H), 2.92 (dd, J = 4.4, 17.2 Hz, 1H), 2.76 (d, J = 17.6 Hz, 1H), 2.62-2.56 (m, 1H), 2.28-2.24 (m, 2H), 2.06-2.00 (m, 1H); ¹³C NMR (100 MHz, CDCl₃) δ = 175.4, 171.2, 155.4, 153.2, 149.9, 130.2, 127.4, 126.8, 123.5, 122.0, 120.7, 107.7, 72.7, 71.6, 71.5, 65.7, 61.8, 56.7, 38.3, 29.5 (q), 28.0 ppm.

Following the above deprotection protocol for compound 1, 12b (200 mg, 0.43 mmol) was used to obtain 12 (120 mg, 67% yield). ¹H NMR (400 MHz, CDCl₃): δ = 12.28 (s, 1H), 7.65 (d, J = 9.2 Hz, 1H), 7.25 (d, J = 9.2 Hz, 1H), 5.35 (t, J = 1.6 Hz, 1H), 4.79-4.77 (m, 1H), 4.39 (dd, J = 2.8, 4.4 Hz, 1H), 2.96 (dd, J = 4.4, 17.6 Hz, 1H), 2.76 (d, J = 17.6 Hz, 1H), 2.53-2.48 (m, 1H), 2.20-2.01 (m, 3H); ¹³C NMR (100 MHz, CDCl₃) δ = 187.7, 179.6, 173.8, 161.7, 146.6, 141.1, 137.9, 127.3, 126.7, 126.0, 115.9, 71.4, 70.3, 69.2, 37.1, 29.5 (q), 26.3 ppm; HRMS (ESI) m/z [M + NH₄]⁺ calcd for C₁₈H₁₆ClF₃NO₆ 434.0618, found 434.0615.

(3aR,5S,11bR)-10-Chloro-2,6,11-trioxo-5-(3,3,3-trifluoropropyl)-3,3a,5,6,11,11b-hexahydro-2H-benzo[g]furo[3,2-c]isochromen-7-yl diethyl phosphate (13). To a solution of 12 (84 mg, 0.2 mmol) and Na₂CO₃ (106 mg, 1 mmol) in acetone (1 ml) was added diethyl chlorophosphate (44 μL, 0.3 mmol). The resulting mixture was stirred at 35 °C for 6 hours. Upon completion, the reaction mixture was directly loaded to silica gel using hexane/EtOAc (1/1) to obtain 13 as a pale yellow liquid (100 mg, 91% yield). ¹H NMR (400 MHz, CDCl₃): δ = 7.72 (d, J = 9.2 Hz, 1H), 7.66 (d, J = 9.2 Hz, 1H), 5.38 (t, J = 2.0 Hz, 1H), 4.74-4.72 (m, 1H), 4.37-4.32 (m, 1H), 4.31-4.23 (m, 4H), 2.95 (dd, J = 4.8, 17.6 Hz, 1H), 2.74 (d, J = 17.6 Hz, 1H), 2.38-2.34 (m, 1H), 2.25-2.23 (m, 2H), 1.99-1.96 (m, 1H), 1.40-1.34 (m, 6H); ¹³C NMR (100 MHz, CDCl₃) δ = 182.2, 179.8, 173.8, 148.8, 148.2, 148.1, 138.5, 135.4, 131.3, 129.2, 128.3 (t), 125.7 (t), 71.8, 70.6, 68.7, 65.5 (d), 65.4 (d), 37.0, 29.7 (q), 25.6, 16.2, 16.1 ppm.

(3aR,5S,11bR)-10-cChloro-2,6,11-trioxo-5-(3,3,3-trifluoropropyl)-3,3a,5,6,11,11b-hexahydro-2H-benzo[g]furo[3,2-c]isochromen-7-yl dihydrogen phosphate (14). Iodotrimethylsilane (29 μL, 0.2 mmol) was added to a solution of 13 (100 mg, 0.18 mmol) in anhydrous CH₂Cl₂ (360 μl) with stirring under argon. The reaction was stirred at room temperature and monitored by HPLC analysis. Upon completion (typically 8-12 hours) the volatiles were evaporated and the residue was taken up in the mixture of Et₂O (5 ml) and H₂O (5 ml). The aqueous phase was collected and lyophilized to afford 14 as a yellow amorphous powder (64 mg, 72%). ¹H NMR (400 MHz, DMSO-d₆): δ = 7.91 (d, J = 8.8 Hz, 1H), 7.68 (dd, J = 1.2, 9.2 Hz, 1H), 5.37 (t, J = 2.0 Hz, 1H), 4.79-4.78 (m, 1H), 4.43 (dd, J = 2.8, 4.8 Hz, 1H), 3.21 (dd, J = 4.8, 17.6 Hz, 1H), 2.54 (d, J = 17.6 Hz, 1H), 2.40-2.24 (m, 3H), 1.93-1.86 (m, 1H); ¹³C NMR (100 MHz, DMSO-d₆) δ = 181.7, 180.0, 175.2, 149.1, 148.8, 137.5, 134.8, 128.7, 127.8, 125.8, 125.7, 71.5, 69.7, 69.1, 36.5, 28.4 (q), 24.6 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₁₈H₁₄ClF₃P 497.0016, found 497.0008.

Synthesis of Probes 1 and 2

N-(5-((7bR, 10aR, 12S)-7-methoxy-2,2-dimethyl-9-oxo-7b, 10,10a, 12-tetrahydro-9H-furo[2″,3″:5′,6′]pyrano[3′,4′:2,3]naphtho[1,8-de][1,3]dioxin-12-yl)pentyl)-6-(5-((3aS,4S,6aR)-2-oxohexahydro-1H-thieno[3,4-d]imidazol-4-yl)pentanamido)hexanamide (probe 2). To a solution of (4R,5R)-4-hydroxy-5-(6-methoxy-2,2-dimethylnaphtho[1,8-de][1,3]dioxin-5-yl)dihydrofuran-2(3H)-one (reported intermediate 5 (Zhang et al., 2015)) (66 mg, 0.2 mmol) and 6-azidohexanal (64 mg, 0.4 mmol) in anhydrous CH₂Cl₂ (2 ml) at 0 °C, Cu(OTf)₂ (36 mg, 0.1 mmol) was added with stirring. The temperature was allowed to rise to room temperature and the mixture was stirred for 16 hours. After evaporating the volatiles, the crude mixture was purified on silica gel using hexane/EtOAc (5/1) to give prob-int as a colorless solid (76 mg, 84% yield, >20:1 dr ratio). ¹H NMR (400 MHz, CDCl₃): δ = 7.66 (d, J = 8.4 Hz, H), 7.45-7.41 (m, 1H), 6.91 (d, J = 7.6 Hz, 1H), 5.57 (d, J = 2.4 Hz, 1H), 4.96-4.93 (m, 1H), 4.34-4.33 (m, 1H), 4.09 (s, 3H), 3.23 (t, J = 7.2 Hz, 2 H), 2.90 (dd, J = 4.4, 17.2 Hz, 1H), 2.75 (d, J = 17.6 Hz, 1H), 2.20-2.10 (m, 2H), 1.66 (s, 3H), 1.64 (s, 3H), 1.59-1.55 (m, 2H), 1.42-1.25 (m, 4H); ¹³C NMR (100 MHz, CDCl₃) δ = 175.8, 151.0, 148.3, 139.5, 127.6, 127.1, 120.2, 118.4, 115.2, 114.6, 110.6, 101.7, 73.0, 72.8, 71.6, 64.5, 51.4, 38.5, 33.8, 28.8, 26.8, 25.9, 24.8, 23.9 ppm.

Water (10 μL) and 1 M trimethylphosphine solution in THF (320 μl, 0.32 mmoL) were added to a THF solution (1.6 ml) of 15a (72 mg, 0.16 mmol). The resulting mixture was stirred at room temperature for 2 hours and then removed the volatiles. The residue was dissolved in DMF (1.6 ml) and treated with trimethylamine (46 μl, 0.32 mmol) and NHS-LC-biotin (90 mg, 0.2 mmol; Chempep, Wellmington, Florida, USA) with stirring overnight at room temperature. Upon completion, the crude mixture was purified on silica gel using CH₂Cl₂/MeOH (10/1-8/1) to afford probe 2 as a colorless solid (60 mg, 49% yield). ¹H NMR (400 MHz, CD₃OD): δ = 8.10-8.04 (m, 2H), 7.83 (d, J = 8.4 Hz, 1H), 7.66-7.62 (m, 1H), 7.10-7.08 (m, 1H), 5.81 (s, 1H), 5.12 (s, 1H), 4.65-4.62 (m, 1H), 4.57 (s, 1H), 4.45-4.42 (m, 1H), 4.22 (s, 3H), 3.53-3.48 (m, 2H), 3.08-3.04 (m, 1H), 2.87-2.76 (m, 4H), 2.36-2.30 (m, 6H), 1.90-1.88 (m, 2H), 1.80 (s, 6H), 1.64-1.48 (m, 18H); ¹³C NMR (100 MHz, CD₃OD) δ = 178.5, 176.0, 175.9, 166.0, 152.1, 149.5, 140.6, 128.6, 128.1, 121.8, 119.9, 116.2, 115.7, 111.6, 102.9, 74.6, 73.9, 73.1, 64.5, 63.3, 61.5, 56.9, 41.0, 40.5, 40.2, 39.1, 37.0, 36.8, 34.8, 30.4, 30.1, 29.7, 29.4, 28.0, 27.5, 26.9, 26.7, 25.8, 25.2, 25.1 ppm; HRMS (ESI) m/z [M + H]⁺ calcd for C₄₀H₅₅N₄O₉S 767.3690, found 767.3671.

N-(5-((3aR,5S,11bR)-7-hydroxy-2,6,11-trioxo-3,3a,5,6,11,11b-hexahydro-2H-benzo[g]furo[3,2-c]isochromen-5-yl)pentyl)-6-(5-((3aS,4S,6aR)-2-oxohexahydro-1H-thieno[3,4-d] imidazol-4-yl)pentanamido)hexanamide (probe 1). A solution of probe 2 (42 mg, 0.05 mmol) in acetonitrile (2 ml) at 0 °C was treated with a precooled solution of ammonium cerium nitrate (63 mg, 0.1 mmol) in water (2 ml). The reaction was quenched after 10 min with the addition of dichloromethane (10 ml) and water (10 ml). The organic layer was separated, washed with brine, dried with Na₂SO₄. The solution was concentrated and then purified on silica gel using CH₂Cl₂/MeOH (10/1-8/1) to afford probe 1 as an orange solid (20 mg, 56% yield). ¹H NMR (400 MHz, DMSO-d₆): δ = 11.51 (s, 1H), 7.78 (t, J = 8.0 Hz, 1H), 7.70-7.66 (m, 2H), 7.60 (d, J = 7.2 Hz, 1H), 7.38 (t, J = 8.4 Hz, 1H), 6.41-6.34 (m, 2H), 5.29 (s, 1H), 4.76-4.75 (m, 1H), 4.37 (s, 1H), 4.31-4.28 (m, 1H), 4.12-4.11 (m, 1H), 3.19-3.09 (m, 2H), 2.99-2.95 (m, 3H), 2.81 (dd, J = 4.2, 12.8 Hz, 1H), 2.27 (d, J = 12.8 Hz, 1H), 2.04-2.00 (m, 4H), 1.86-1.82 (m, 1H), 1.61-1.57 (m, 1H), 1.48-1.20 (m, 16H); ¹H NMR (400 MHz, CD₃OD): δ = 7.74 (t, J = 8.0 Hz, 1H), 7.66 (dd, J = 1.2, 7.6 Hz, 1H), 7.32 (dd, J = 1.2, 8.4 Hz, 1H), 5.33 (t, J = 2.0 Hz, 1H), 4.83-4.81 (m, 1H), 4.80 (dd, J = 4.4, 8.0 Hz, 1H), 4.41 (dd, J = 2.4, 4.4 Hz, 1H), 4.29 (dd, J = 4.4, 8.0 Hz, 1H), 3.20-3.12 (m, 5H), 2.92 (dd, J = 4.8, 12.8 Hz, 1H), 2.70 (d, J = 12.8 Hz, 1H), 2.60 (d, J = 17.6 Hz, 1H), 2.20-2.14 (m, 4H), 2.05-1.99 (m, 1H), 1.49-1.29 (m, 20H), 0.92-0.88 (m, 2H); ¹³C NMR (100 MHz, CD₃OD) δ = 190.2, 182.9, 177.6, 176.0, 175.9, 166.0, 162.7, 150.7, 138.1, 137.9, 133.0, 125.5, 120.1, 116.5, 73.0, 72.5, 71.9, 63.3, 61.6, 57.0, 41.0, 40.6, 40.1, 38.1, 36.9, 36.8, 34.6, 30.2, 30.1, 29.7, 29.4, 27.8, 27.5, 26.9, 26.7, 25.7; HRMS (ESI) m/z [M + H]⁺ calcd for C₃₆H₄₇N₄O₉S 711.3064, found 711.3035.

Reaction of FB with N-Acetyl-Cysteine

N-Acetyl-S-((1R,3R)-3-(carboxymethyl)-9-hydroxy-5,10-dioxo-1-propyl-3,4,5,10-tetrahydro-1H-benzo[g]isochromen-4-yl)cysteine (FB-NAC). HMBC correlations were highlighted in red. A solution of frenolicin B (3.2 mg, 0.01 mmol) in water (0.5 ml) and acetonitrile (0.5 ml) at room temp was treated N-acetyl-cysteine (8.1 mg, 0.05 mmol). The resulting mixture was stirred for 2 h and purified on Supelco C18 reverse HPLC (25 cm × 21.2 mm, 10 uM) to give the titled product FB-NAC as an orange solid (100% conversion on HPLC). ¹H NMR (400 MHz, CDCl₃): δ = 11.91 (s, 1H), 7.57-7.54 (m, 3H), 7.23-7.21 (m, 1H), 4.86 (s, 1H), 4.74 (s, 1H), 4.35 (s, 1H), 4.13 (s, 1H), 3.20-3.11 (m, 3H), 2.90-2.85 (m, 1H), 2.14 (s, 3H), 1.74-1.52 (m, 4H), 0.97 (t, J = 6.8 Hz, 3H); ¹³C NMR (100 MHz, CD₃OD) δ = 190.1, 183.0, 175.6, 174.4, 173.2, 162.6, 145.1, 143.8, 137.6, 133.3, 125.3, 120.2, 116.1, 72.5, 68.5, 54.0, 42.2, 38.1, 35.5, 35.0, 22.4, 20.8, 13.9; HRMS (ESI) m/z [M + H]⁺ calcd for C₂₃H₂₆NO₉S 492.1328, found 492.1267.

QUANTIFICATION AND STATISTICAL ANALYSIS

GraphPad Prism software was used to calculate the statistics as described in the methds and legends. Replicates are indicated in the figure legends with each experiment repeated at least twice. Data between two groups were compared using a two-tailed Student’s t-test. All data are presented as mean ± SEM. Differences between groups were considered statistically significant at p < 0.05.

DATA AND SOFTWARE AVAILABILITY

Spectroscopic data of compounds 1-14, probes 1 and 2, and FB-NAC are available at Mendeley Data (http://dx.doi.org/10.17632/r3b47hrv3h.1).

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti-phospho-Akt (Ser473)	Cell Signaling Technology	Cat# 4060, RRID:AB_2315049
Rabbit monoclonal anti-phospho-p70 S6 Kinase (Thr389)	Cell Signaling Technology	Cat# 9234, RRID:AB_2269803
Rabbit monoclonal anti-phospho-4E-BP1 (Thr37/46)	Cell Signaling Technology	Cat# 2855, RRID:AB_560835
Rabbit monoclonal anti-phospho-4E-BP1 (S65)	Cell Signaling Technology	Cat# 13443, RRID:AB_2728761
Rabbit monoclonal anti-phospho-4E-BP1 (Thr70)	Cell Signaling Technology	Cat# 13396
Rabbit monoclonal anti-4E-BP1	Cell Signaling Technology	Cat# 9644, RRID:AB_2097841
Rabbit monoclonal anti-eIF4E	Cell Signaling Technology	Cat# 2067, RRID:AB_2097675
Mouse monoclonal anti-Myc-Tag	Cell Signaling Technology	Cat# 2276, RRID:AB_331783
Rabbit monoclonal anti-Cleaved PARP (Asp214)	Cell Signaling Technology	Cat# 5625, RRID:AB_10699459
Rabbit polyclonal anti-Peroxiredoxin 1	Abcam	Cat# ab15571, RRID:AB_2170316
Mouse monoclonal anti-PICOT	Santa Cruz Biotechnology	Cat# sc-100601, RRID:AB_2110379
Mouse monoclonal anti-Glutaredoxin 3	R&D Systems	Cat# MAB7560
Rabbit polyclonal anti-Biotin	Bethyl Laboratories	Cat# A150-109A, RRID:AB_67327
Mouse monoclonal anti-β-Actin	Sigma-Aldrich	Cat# A5441, RRID:AB_476744
Mouse monoclonal anti-Puromycin	Kerafast	Cat# EQ0001, RRID:AB_2620162
Mouse monoclonal anti-Biotin-FITC	Jackson ImmunoResearch Labs	Cat# 200-092-211, RRID:AB_2339020
Rabbit polyclonal anti-Biotin-FITC	Abcam	Cat# ab53469, RRID:AB_867862
Alexa Fluor® 594 AffiniPure Goat Anti-Rabbit IgG	Jackson ImmunoResearch Labs	Cat# 111-585-144, RRID:AB_2307325
Alexa Fluor® 594 AffiniPure Goat Anti-Mouse IgG	Jackson ImmunoResearch Labs	Cat# 115-585-146, RRID:AB_2338881
Chemicals, Peptides, and Recombinant Proteins
MK-2206	Selleck Chemicals	S1078; CAS: 1032350-13-2
Rapamycin	Selleck Chemicals	S1039; CAS: 53123-88-9
AZD8055	Selleck Chemicals	S1555; CAS: 1009298-09-2
PD0325901	Selleck Chemicals	S1036; CAS: 391210-10-9
Human Prx1 Recombinant Protein	Thermo Fisher Scientific	Cat# LF-P0002
Human Prx2 Recombinant Protein	Thermo Fisher Scientific	Cat# LF-P0007
Human Grx1 Recombinant Protein	Abcam	Cat# ab86987
Human Grx2 Recombinant Protein	Abcam	Cat# ab95471
Human Grx3 Recombinant Protein	MyBioSource	Cat# MBS203427
Human Grx5 Recombinant Protein	Abcam	Cat# ab95352
Critical Commercial Assays
QuikChange Lightning Site-Directed Mutagenesis Kit	Agilent Technologies	Cat# 210518
Dual-Luciferase® Reporter Assay System	Promega	Cat# E1910
Annexin V-FITC Apoptosis Detection Kit	BD Biosciences	Cat# 556547
Glutathione Fluorometric Assay Kit	Biovision	Cat# K251-100
Microplate Assay Kit for GSH/GSSG	Oxford Biomedical Research	Cat# GT40
Amplex Red Hydrogen Peroxide Assay Kit	Thermo Fisher Scientific	Cat# A22188
CellROX Deep Red Flow Cytometry Assay Kit	Thermo Fisher Scientific	Cat# C10491
Deposited Data
Spectroscopic data	This paper; Mendeley Data	http://dx.doi.org/10.17632/r3b47hrv3h.1
Experimental Models: Cell Lines
Human: HCT116 cells	ATCC	Cat# CCL-247, RRID:CVCL_0291
Human: DLD-1 cells	ATCC	Cat# CCL-221, RRID:CVCL_0248
Human: HCT15 cells	ATCC	Cat# CCL-225, RRID:CVCL_0292
Human: RKO cells	ATCC	Cat# CRL-2577, RRID:CVCL_0504
Human: SW480 cells	ATCC	Cat# CCL-228, RRID:CVCL_0546
Human: MDA-MB-231 cells	ATCC	Cat# HTB-26, RRID:CVCL_0062
Human: MDA-MB-4681 cells	ATCC	Cat# HTB-132, RRID:CVCL_0419
Human: H1975 cells	ATCC	Cat# CRL-5908, RRID:CVCL_1511
Human: BEAS-2B cells	ATCC	Cat# CRL-9609, RRID:CVCL_0168
Human: IMR-91 cells	Coriell	Cat# I91L-16, RRID:CVCL_5420
Human: TIG-1 cells	Coriell	Cat# AG06173, RRID:CVCL_0560
Human: GM13267 cells	Coriell	Cat# GM13267, RRID:CVCL_4F67
Human: GM15871 cells	Coriell	Cat# GM15871, RRID:CVCL_AM76
Human: PAE cells	Guo et al., 2013	N/A
Mouse: wild-type 4E-BP¼E-BP2 cells	Dowling et al., 2010	N/A
Mouse: 4E-BP¼E-BP2 double knockout cells	Dowling et al., 2010	N/A
Experimental Models: Organisms/Strains
Mouse: sp/sp CrTac:NCr-Foxn1nu	Taconic	NCRNU-M
Oligonucleotides
Primers for human Prx1 and Grx3 and their mutants, see Table S1	Sigma-Aldrich; This paper	N/A
Recombinant DNA
pCMV6-Entry	OriGene	Cat# PS100001
pcDNA3-rLuc-PolioIRES-fLuc	She et al., 2010	N/A
Plasmids: pCMV6-Prx1-Myc-Flag and its mutants (C51S, C71S, C83S, C173S); pCMV6-Grx3-Myc-Flag and its mutants (C46S, C146S, C159S, C229S, C261S, C159S/C261S)	This paper	N/A
pLKO control shRNA	Sigma-Aldrich	Cat# SHC002
pLKO human Prx1 shRNA #1	Sigma-Aldrich	TRCN0000029511
pLKO human Prx1 shRNA #2	Sigma-Aldrich	TRCN0000029512
pLKO human Grx3 shRNA #1	Sigma-Aldrich	TRCN0000064666
pLKO human Grx3 shRNA #2	Sigma-Aldrich	TRCN0000064667
Software and Algorithms
Graphpad Prism 7	Graphpad	RRID:SCR_002798 https://www.graphpad.com/
FACSDiva v8.0.1	BD Biosciences	http://www.bdbiosciences.com/
CellQuest Pro v6.1	BD Biosciences	http://www.bdbiosciences.com/
Xcalibur 3.1	Thermo Fisher Scientific	https://www.thermofisher.com/
Proteome Discoverer	Thermo Fisher Scientific	https://www.thermofisher.com/
Mascot	Matrix Science	http://www.matrixscience.com/

Highlights.

• Frenolicin B directly targets Prxl and Grx3 to induce ROS

• Frenolicin B-derived ROS inhibits 4E-BP1 phosphorylation and cancer cell growth

• 4E-BP1 phosphorylation status potentially predicts the cytotoxic level of ROS

• Optimized inhibitor effectively suppresses tumor growth in vivo