Phomoxanthone A Targets ATP Synthase

Abstract

Phomoxanthone A is a naturally occurring molecule and a powerful anti-cancer agent, although its mechanism of action is unknown. To facilitate the determination of its biological target(s) we used affinity-based labelling using a phomoxanthone A probe. Labelled proteins were pulled down, subjected to chemoproteomics analysis using LC-MS/MS and ATP synthase was identified as a likely target. Mitochondrial ATP synthase was validated in cultured cells lysates and in live intact cells. Our studies show sixty percent inhibition of ATP synthase by 260 μM phomoxanthone A.

Introduction

Phomoxanthone A (1) is a naturally occurring dimeric tetrahydroxanthone that possesses both an enticing molecular architecture and impressive biological properties (Figure 1).^1-10 For example, phomoxanthone A is known to have antibacterial, antimalarial, and anticancer activity. It has been observed that 1 is equally as effective at killing both cisplatin sensitive and cisplatin resistant cancer cells.^3,5

Figure 1.

Photoaffinity labelling studies of anticancer agent phomoxanthone A.

We were initially drawn to phomoxanthone A in 2014 because of its demanding synthetic framework. The challenges associated with a total synthesis of 1 include the formation of difficult tertiary ether stereocenter (highlighted in purple). As we studied and advanced synthetic methods to enable the construction of phomoxanthone A and its structurally related analogs,^8,11-13 we became more enamored with both its recognized anticancer properties and the emerging collection of evidence suggesting that 1 may affect cancer cells through a unique mechanism of action. A 2018 study from Bohler and coworkers found that phomoxanthone A (1) causes rapid disintegration of the inner mitochondrial membrane, but it did not identify the biomolecular targets.⁴ Follow up studies by Monti identified carbamoyl-phosphate synthase 1 as a possible target of phomoxanthone A.⁶ A separate study by Chen and Luo identified phomoxanthone A as a protein tyrosine phosphatase inhibitor.⁷ However, the role of these two biomolecular targets in affecting mitochondrial disintegration is unclear leaving open the possibility of phomoxanthone A having more direct target(s) in the mitochondria. This gap in knowledge inspired us to use our synthetic experience to enable the investigation of the biomolecular targets of phomoxanthone A. Herein, we detail the identification of ATP synthase as a biomolecular target of phomoxathone A using a photoaffinity labelling approach.

Results and Discussion

Probe Optimization.

The project was initiated with the thoughtful selection of a photoaffinity label (PAL).^{5, 14} Common photoactivatable chemical probes for PAL studies include diazirines, benzophenones and arylazides. After weighing the pros and cons of each photoreactive functional group, we chose diazirines due to several factors, including their small size, anticipated ease of installation on the phomoxanthone A scaffold, and activation at a wavelength appropriate for cellular work.^15-19

There is little literature on the structure-activity relationship of phomoxanthone A. The few reports available show that phomoxanthone A analogs containing various substitution patterns on the oxygen substituents can retain anticancer activity.^3,8 Based on these observations we decided to conjugate a minimalistic diazirine probe to the parent compound via an ether or ester linkage. With this in mind, we prepared two known diazirine probes for our studies (3a and 3b Scheme 1).^20,21 Due to the limited availability of phomoxanthone A for test reactions, we optimized the PAL on test chromenone 2. Diazirine 3a underwent Sn2 reaction with 2 to give rise to 70% of desired PAL-labelled chromenone 4 where the triisopropylsilyl group (TIPS) was removed under the reaction conditions (see pink H in 4, Scheme 1). Enthused by the synthesis of 4, we immediately tested the breakdown of the photoaffinity label when exposed to 356 nm UV light. Unfortunately, even after 30 minutes less than 50% of the starting material had undergone photolysis. Thus, it was concluded that the photoaffinity label 3a would not be a suitable PAL for phomoxanthone A because of the extensive exposure of the cells to UV light, which may cause harm to the biomolecules in the cell that are attempting to be studied.

Scheme 1.

Photoaffinity label optimization on chromenone model system.

With the knowledge that 3a was not suitable, we moved on to photoaffinity label 3b. The reaction of chromenone 2 with 3b gave rise to 5 with the PAL connected to the chromenone through an ester linkage. Importantly, the reaction was high yielding (65%), executed under mild reaction conditions appropriate for a complex natural product, and clean (spot to spot conversion by TLC, see SI for details). Our testing revealed that complete photolysis of the PAL was observed after just 5 minutes of exposure to 356 nm UV light.

We were delighted to find that diazirine 3b readily reacted with phomoxanthone A to give rise to PAL-labelled phomoxanthone A PAL-1, the compound envisioned to be used for our protein pull down experiments (Scheme 2A). The reaction was determined to be successful by both ¹H NMR analysis as well as HRMS. Notably, the PAL easily attached to the highly functionalized secondary metabolite on the non-phenolic oxygen. While the probe conjugated phomoxanthone A PAL-1 showed somewhat lower anticancer activity toward Jurkat cells (IC₅₀ = 6.9 μM) than phomoxanthone A (IC₅₀ = 53 nM), this activity was deemed sufficient to carry out the subsequent studies (Scheme 2B).

Scheme 2.

(A) Photoaffinity labelling of phomoxanthone A and (B) Comparison of cytotoxicity of 1 and PAL-1 in Jurkat cells.

PAL-Phomoxanthone A in Cells.

With PAL-1 in hand we moved on to evaluate its labelling performance in cellular studies. First, we sought to collect evidence that PAL-1 was capable of selectively labelling biomolecules. Experimentally, the selective labelling abilities of PAL-1 were studied by comparing control compound 5 to the probe-conjugated phomoxanthone A (PAL-1). Each compound (e.g., 5 and PAL-1) was separately incubated with Jurkat cell lysates at both 10 μM and 25 μM concentrations for both 1 hour and 3 hours. The resultant mixtures were then exposed to 10 minutes of 356 nM light, lysed by sonication and conjugated with TAMRA-azide by copper-catalyzed alkyne-azide cycloaddition (CuAAC).^22-24 The proteins labelled in this method were then subjected to SDS-PAGE separation and visualized by in-gel fluorescence scanning (Figure 2).

Figure 2.

Live cell protein labelling with PAL-1. (A) Structures of PAL-1 and 5, the compounds used to probe the selective protein labelling abilities of PAL-1. (B) Experimental process for bioconjugating and tagging PAL-1 and 5. (C) SDS-PAGE separation and visualization by in-gel fluorescence scanning.

We were thrilled to find that PAL-1 is capable of selectively labelling proteins (Figure 2C). At concentrations as low as 10 μM with 1 hour of incubation new bands were observed in the 20-25 kDa range. After 3 hours of incubation, the new band is even more visible, along with a few additional bands in the 37-75 kDa range. In comparison to the control compound 5, which gave rise to a very streaky gel, it is clear that the minimalistic probe labelled phomoxanthone A possessed selective protein labelling capabilities as marked by an absence of smear in the case of PAL-1. In support of this conclusion, a Coomassie gel was used to verify that equal amounts of protein were introduced into the gel wells (see SI for details).

Encouraged by these results, we set out to determine which biomolecules were being pulled out with PAL-1 using proteomic analyses. To achieve this, the labelling experiment was scaled up and changed slightly to enable protein purification by affinity chromatography. We also incorporated co-incubation experiments that were ultimately subjected to reductive dimethylation (ReDiMe) procedures to allow for quantification of protein concentrations in the final proteomics analysis.²⁵ Specifically, in the first experiment, 25 μM of PAL-1 was incubated with Jurkat cells that were pre-treated with 2.5 mM of phomoxanthone A for 30 minutes. A second set of experiments was executed with cells that were only treated with 25 mM PAL-1 and corresponding control experiments where cells were only treated with DMSO were also included for comparison. After the 3 hour incubation period, all three experiments were subjected to UV light for 10 minutes so as to enable the bonding of PAL-1 to the nearby biomolecules. Instead of using TAMRA-azide to visualize the biomolecular targets, we “clicked” on biotin-azide so the protein targets could be isolated. The biotin conjugate proteins were purified by affinity chromatography using streptavidin-agarose beads. The purified conjugated proteins were first exposed to on-bead trypsin digestion and then labeled by reductive methylation (ReDiMe) using heavy (for PAL-1 preincubation experiment), medium (phomoxanthone A alone) and light (for DMSO treated) sodium cyanoborohydride. Samples with heavy methyl tags and medium methyl tags were mixed, co-analzed by LCMS/MS. Relative protein abundances are quantified by comparing the ion chromatogram peak areas of heavy and medium labeled versions of the constituent peptide extracted from the full MS spectra.

The resultant digested and labelled proteins were followed by proteomic analysis using a Lumos Orbitrap mass spectrometer. The stable isotope labelling combinations described above served to eliminate non-specific labelling and to further identify phomoxanthone A specific labelling. The identified proteins were then analyzed by corresponding volcano plots as a log2 of the competition ratio (probe/probe with excess competitors) against statistical significance (−log10 p value) (Figure 3). Hits identified from co-incubated experiments can be considered non-specific interactions as phomoxanthone A will likely preclude PAL-1 from interacting with the primary targets. Hits identified only from PAL-1 treated samples compared to DMSO and co-incubated experiments were then considered possible on target interactions. A total of 132 such unique targets were identified, sorted by log2 values and plotted against statistical significance as a volcano plot (Figure 3). Based on this screen, and considering phomoxanthone A is a known mitochondrial toxin, the search was narrowed for primarily mitochondrial hits. About ten mitochondrial targets emerged from the analysis and the beta subunit of ATP synthase was considered a highly probable target.

Figure 3.

Volcano plots depicting hits derived from phomoxanthone A proteomics.

In vitro validation of ATP synthase.

We proceeded to validate ATP synthase as a possible target for phomoxanthone A. Quantitative measurement of the activity of mitochondrial ATP synthase was performed using the commercially available ATP Synthase Enzyme Activity Microplate Assay Kit (Abcam, ab109714).^26-27 In this assay, ATP synthase activity is measured through the conversion of the nicotinamide adenine dinucleotide cofactor from its reduced form (NADH) to its oxidized form (NAD+). Activity of ATP synthase is monitored through loss of absorption at 340 nm, the wavelength that NADH absorbs light energy. As the conversion of NADH to NAD+ proceeds, the light absorption subsequently decreases, and the decrease in absorption over a 120 minute time frame is used to attain a relative rate of ATP synthase activity. As ATP synthase activity is inhibited, the rate of NADH to NAD+ conversion is slowed (or completely stopped).

Experiments were carried out to determine appropriate concentrations of phomoxanthone A and incubation periods needed for the assay using commercially available rat mitochondrial lysate (ab 110347). Under the optimized condition, 60 percent inhibition of ATP synthase activity was observed with the greatest concentration of phomoxanthone A used (260 μM). The well recognized ATP synthase inhibitor oligomycin A was used as a reference. A 2 μM concentration of oligomycin A produced an 89% inhibition of ATP synthase activity (Figure 4A).

Figure 4.

(A) Relative rates of ATP synthase activity after treatment with oligomycin (2 mM) or phomoxanthone A (260 μM) in comparison to the control (DMSO). ****For both phomoxanthone A and oligomycin A, p≤0.0001. (B) Effect of phomoxanthone on ATP synthase activity. Each data point is the mean of triplicate measurements. The error bars represent calculated standard deviations.

A dose dependent effect was seen in inhibition of ATP synthase activity with phomoxanthone concentrations of 26 micromolar or greater (Figure 4B). Lower concentrations of phomoxanthone A that were tested did not yield significant decreases in ATP synthase activity in comparison to the control (p> 0.05).

Validation of ATP synthase in cells.

With experimental evidence that phomoxanthone A inhibits ATP synthase in a biological assay, we proceeded to validate the target in live cells. The method chosen for this analysis required measuring the autofluorescence lifetime of HeLa cells upon treatment with various doses of phomoxanthone A.^28-29 HeLa cells were used in this experiment because they were a readily available cancer line and they are easily studied with fluorescence microscopy.

The chemistry behind the experiment is centered on the idea that metabolic activities in normal cells rely on mitochondrial oxidative phosphorylation (OXPHOS) to generate ATP with ATP synthase for energy. The creation of ATP requires the conversion of NAD+ to NADH, which are two forms of the key cofactor nicotinamide adenine dinucleotide (NAD). The NAD+/NADH ratio plays a crucial role in regulating the intracellular redox state, especially in the mitochondria. Fluorescence-lifetime imaging microscopy (FLIM) can be used to measure NADH autofluorescence, and this in turn can be used to determine the engagement of ATP synthase with phomoxanthone A. To determine the influence of phomoxanthone A on ATP synthase in live cells, HeLa cells we treated with different concentration of phomoxanthone A and incubated for 24 hours (Figure 5). A decrease in NADH lifetime in phomoxanthone A treated HeLa cells was observed, suggesting that phomoxanthone A engages with ATP synthase in live HeLa cells.

Figure 5.

Phomoxanthone A decreases NADH lifetime in HeLa cell studies. (A) Representative HeLa cells treated with DMSO or 5μM or 50 SymbolM of phomoxanthone (left to right) showing autofluorescence corresponding to primarily to NADH/NAD+ redox levels. (B) Phasor representation of raw lifetimes values of each image pixel showing heterogeneity. The distinct regions on the phasor plots are highlighted by colored circles indicating distinct lifetimes (red lifetime center= 1.57 ns, orange lifetime center= 1.39 ns, yellow lifetime center= 1.24 ns, green lifetime center= 1.11 ns, blue lifetime center= 1 ns). The pixels underlying these circles are false colored and overlaid on grayscale cell images. The cells have been treated with DMSO or 5 mM or 50 mM of phomoxanthone (left to right, respectively) where the asterisks represented significant differences * = p ≤0.05, **=p≤0.01, ***=p≤0.001, ****=p≤0.0001. Error bars denote standard deviation. (D) Decrease in lifetime (%) at 5 μM and 50 μM phomoxanthone A when compared to the control experiment. (E and F) Decrease in autofluorescence of HeLa cells treated with 5 mM of phomoxanthone after 24 hours when compared to the control (DMSO treatment), and a similar study showing the immediate decrease. Note that decrease in the lifetime is proportional to the dose of phomoxanthone

Conclusion

In conclusion, we identified ATP synthase as a cellular target of phomoxanthone A. This finding emerged as a result of first incorporating a photoactivatable diazirine based crosslinker onto the natural product phomoxanthone A. Once the phomoxanthone A probe was determined to be suitable for labelling proteins, cell-based probe incubation assays were conducted followed by chemoproteomics experiments using ReDiMe. The results of these experiments suggested that ATP synthase is a probable target of phomoxanthone A. We then proceeded to validate the ATP synthase in both in vitro and cell based assays. The identification of this target is an exciting and critical step in the development of rationally designed ATP synthase inhibitors for anticancer drug discovery, which is a topic of current interest in our laboratories. While ATP synthase was the only biomolecular target validated in this study, the validation of additional cellular targets with which phomoxanthone A may engage is underway.

Experimental Section

Photoaffinity Labelling.

Commercially available phomoxanthone A (1 mg, 0.0013 mmol) was dissolved in 1 mL of DCM and cooled to 0°C. In to this mixture was added DCC (0.60 mg, 0.0.0029 mmol), DMAP (0.031 mg, 0.00025 mmol) and 3b (0.24 mg, 0.0016 mmol). Reagents were added from freshly prepared stock solutions. The mixture was stirred at 0°C for 24 hours. The mixture was then warmed up to room temperature and condensed using rotary evaporation. Compound PAL-1 was purified on silica Prep TLC plates as a yellow oil.

Live Cell Protein Labelling.

Jurkat cells were cultured in 225 mL flasks as described in the supporting information. Once cells reached 90% density, culture was replaced with Serum free medium. Cells were incubated with 10 μM concentration of PAL-1 for 3 hours at 37°C. Cells were then harvested and washed 1 time with PBS. Cell lysates were then transferred into UV-transparent microplate and subjected to UV irradiation for 10 minutes, followed by spin down at 3000 RPM for 3.5 minutes and then the cells were lysed by sonication. Cell supernatant was collected by spinning at 15000 RPM for 10 minutes.

Sample Preparation for Proteomic Studies

5 million Jurkat cells in 20 mL RPMI media without serum in T125 flasks were treated as follows:

A) 30 minutes preincubation with 2.5μM Phomoxanthone A followed by 3 hours co-incubation with 25μM RA664

B) 3 hours incubation of cells with 25μM RA664 (50μL of 10mM stock)

C) 3 hours incubation of cells with DMSO (50μL)

Cells were washed 1 time with PBS followed by UV exposure for 10 minutes. Cells were harvested and flash frozen and stored at −80°C. Cells were lysed using a probe sonicator, protein quantified, click chemistry performed, biotinylated proteins purified by streptavidin beads, purified peptides digested, sample combinations labeled by reductive dimethylation using stable isotope labeled forms to tag primary amines (see below), desalted and lyophilized.

Sample A: DMSO (light) + Phomoxanthone A (RA664) (Medium) + RA664 (Heavy)

Sample B: DMSO (light) + RA664 (Heavy)

Sample C: DMSO (light) + RA664 (Heavy) (technical replicates)