In situ imaging and proteome profiling indicate andrographolide is a highly promiscuous compound

Natural products represent an enormous source of pharmacologically useful compounds, and are often used as the starting point in modern drug discovery. Many biologically interesting natural products are however not being pursued as potential drug candidates, partly due to a lack of well-defined mechanism-of-action. Traditional in vitro methods for target identification of natural products based on affinity protein enrichment from crude cellular lysates cannot faithfully recapitulate protein-drug interactions in living cells. Reported herein are dual-purpose probes inspired by the natural product andrographolide, capable of both reaction-based, real-time bioimaging and in situ proteome profiling/target identification in live mammalian cells. Our results confirm that andrographolide is a highly promiscuous compound and engaged in covalent interactions with numerous previously unknown cellular targets in cell type-specific manner. We caution its potential therapeutic effects should be further investigated in detail.

fluorescently quenched activity-based probes based on peptide acyloxymethylketones which were capable of real-time imaging of protease activities in mammalian cells 17 . Herein, by successfully uniting natural product drug discovery, in situ drug profiling and real-time bioimaging of target-drug interaction for the first time, we report dual-purpose probes based on andrographolide (a natural product; Fig. 1). Detailed in situ bioimaging and proteome profiling confirmed that andrographolide is a highly promiscuous compound engaged in covalent interactions with numerous previously unidentified cellular targets in cell type-specific manner.
Andrographolide (henceforth referred to as WT) is a bicyclic diterpenoid isolated from leaves of Andrographis paniculata 18 , which is used extensively in traditional Chinese medicine (TCM). It possesses a wide spectrum of biological activities including antibacterial, anti-inflammatory, antimalarial, anticancer and others 19 , and is currently used in clinic in China for the treatment of paediatric pneumonia and respiratory tract infection 20 . The compound contains an α -alkylidene γ -butyrolactone moiety, two olefinic C-8/17 and C-12/13 bonds and three hydroxyls at C-14/C-3/C-19 positions. Like many known natural products that possess electrophilic moieties 15 , the α ,β -unsaturated ester in WT is essential in maintaining the compound's cellular activities by engaging in covalent reactions with its targets 19 . Previous structure-activity relationship (SAR) studies show attachment of potential leaving groups at C-14 OH of WT, via an ester or ether linkage, could improve its cellular potency, presumably due to facilitated nucleophilic attack by the target protein ( Fig. 1) 19,21 . In our probe design ( Fig. 2 & Table 1), we took advantage of this key feature by linking fluorogenic dyes of one-and two-photon properties (4-methyl-7-aminocoumarin or AMC, and 6-acetyl-2-naphthylamine or AAN) via a self-immolative carbamate 22 , giving probes APCM/AP1CM/AP2CM and APNP/AP1NP/AP2NP, respectively. Michael addition at the C-12 position of these probes by a nucleophilic residue (i.e. cysteine) from the intended target would result in the formation of a covalent protein-probe adduct, which then undergoes rapid and spontaneous β -elimination to release the highly fluorescent AMC and AAN ( Fig. 2 & Supplementary Information). Since the Turn-ON fluorescence effect is position-specific and reaction-driven 23 , these probes are ideally suited for accurate, real-time reporting of target-probe interaction in living cells. SAR studies indicate modifications at C-3/C-19 OHs of WT were less predictable, but in many cases resulted in analogues that retained most of WT's biological activities 19,24 . We therefore introduced 5-hexynoic acid to C-19 and C-3, via an ester linkage, providing the corresponding AP1/AP1CM/AP1NP and AP2/AP2CM/AP2NP, respectively. Molecular "tagging" of WT with such a clickable "minimalist" linker was not expected to adversely affect its intended cellular activities and target recognition 6,12,25 , and at the same time allowed in situ proteome profiling and subsequent in vitro affinity protein enrichment/ target identification via the use of Cu(I)-catalyzed azide-alkyne cycloaddition (CuAAC) 26 . WT was the natural product of choice in our dual-purpose probe design, not only because of its unique chemical structure/reactivity, diverse biological activities and easy access via semi-synthesis 21,24 , but also of the ongoing controversy surrounding its true cellular targets. Several reports have indicated that, in addition to intracellular glutathione (GSH) 27 , and p50 protein of the transcription factor complex NFκ B 28 , WT might have other unknown cellular targets 12,18,19,29,30 . In total, 9 different probes of WT were designed for the current study (Table 1).
Bioimaging and in situ proteome profiling. We first determined the cellular activities of these probes by XTT antiproliferation assay in HepG2 and A549 cancer cells (Supplementary Fig.  S2). All probes showed similar activities as WT in inhibiting cell growth at micromolar concentrations. NC did not inhibit cell growth as expected. With the hexynoic acid-modified probes, AP1/AP2/AP3/AP1NP/AP2NP/AP1CM/AP2CM, additional in situ proteome profiling could be performed. Upon probe labeling, cell lysis and click chemistry with Rh-PEG-N 3 , cell lysates were further analyzed by SDS-PAGE and in-gel fluorescence scanning ( Supplementary Fig. S3). Interestingly, by comparing the fluorescent labeling profiles (overall intensity and specificity), AP1/AP1NP consistently outperformed other probes. Surprisingly, distinctly different fluorescence profiles were observed between the two mammalian cell lines (see Fig. 3a-c & Supplementary Fig. S4; boxed in red), indicating the presence of cell type-specific probe targeting. Since our earlier in vitro results also showed AP1/AP1NP were the best-performing probes in Turn-ON fluorescence experiments (Table 1), they were chosen as the de facto dual-purpose set for subsequent imaging and proteome profiling experiments. We first surveyed different mammalian cells by in situ labeling with AP1 at varied probe concentrations and incubation time (Fig. 3a & Supplementary Fig. S4); at 10 μ M and 3 h incubation (conditions similar to most published protocols with WT 12,[18][19][20][21]24,[27][28][29][30][31] ), all tested cell lines except THP1/HEK293 showed robust fluorescence profiles. We again observed a highly distinctly labeled band at ~60 kDa in A549 cells, but in most other cell lines including HepG2, the strongest labeled band was at ~50 kDa (* in Fig. 3a-c), which might be that of labeled p50 from NFκ B 28 . Subsequent Western blotting (WB) analysis indicated endogenous p50 expression levels in all cell lines were similar. Competitive labeling with excessive WT (up to 10× ) further confirmed the two highly cell type-specific 50-/60-kDa bands were indeed specific endogenous targets of WT ( Supplementary Fig. S5). As the most abundant cellular target of WT, intracellular GSH would effectively quench most of WT/AP1/AP1NP upon their cellular uptake, rendering them inactive toward other cellular protein targets which might be more biologically relevant. We therefore developed an NEM-treatment protocol to effectively block intracellular GSH activities prior to subsequent in situ profiling and imaging experiments. A previously reported GSH-specific imaging probe G2 was used to quantitatively monitor intracellular GSH levels (Fig. 3d) 32 . Pre-treatment of cells with NEM (1 mM, 20 min) completely abolished detectable intracellular GSH activities without causing significant cell death ( Supplementary Fig. S6), and led to dramatic improvement of in situ proteome labeling profiles in both HepG2 and A549 cells (Fig. 3b); significantly more fluorescently labeled protein bands were detected even at lower AP1 concentrations (0.1-2 μ M). Upon click chemistry with Rh-Biotin-N 3 , the labeled lysates were subsequently pulled-down (PD) followed by WB analysis, which unequivocally confirmed the 50-kDa labeled band in HepG2 cells was indeed p50 of NFκ B. In A549 cells, p50 was also positively, but much more weakly, labeled by AP1 (bottom gels). Both AP1 and AP1NP were shown to label cells similarly, with or without NEM treatment (Fig. 3c). AP1NP was thus used for subsequent reaction-driven, Turn-ON fluorescence imaging experiments in real-time ( Fig. 3d & Supplementary video SI_IV).
AP1NP (a bifunctional WT mimic) was capable of real-time imaging of target/probe reaction through its C-14-linked AAN reporter (see Supplementary video). In addition, with the C-19-attached 5-hexynoic acid, it could provide information about the sub-cellular location of labeled protein targets (Fig. 3d), and when needed, be used for target identification by large-scale PD/LC-MS/MS experiments 8,9,16 . As a red-fluorescing GSH sensor, G2 was used together with AP1NP for simultaneous two-color imaging. Again, cells were pre-treated with NEM to suppress intracellular GSH activities (Fig. 3d); without NEM treatment, intense fluorescence was observed in both AP1NP and G2 channels (panels 1/2 & 9/10 for HepG2 & A549 cells, respectively). While signals from the G2 channel were completely NEM-dependent (panels 6/14) and could be readily washed away, weaker AAN signals from the AP1NP channel persisted after NEM treatment (panels 5/13), indicating they were likely from covalent reactions between AP1NP and cellular protein targets. This was further confirmed by click chemistry of labeled cells with Rh-PEG-N 3 followed by imaging (panels 3/7/11/15), with merged images of NEM-treated cells showing excellent overlaps between AP1NP and Rh-PEG-N 3 channels (panels 8/16). During imaging, care was taken to ensure minimal diffusion of these AAN signals, which could also be removed by repeated washes of the cells. Significantly higher fluorescence signals from the Rh-PEG-N 3 channel were detected in NEM-treated cells (compare panels 7/15 with 3/11), indicating blocking intracellular GSH activities had led to more effective labeling of endogenous cellular proteins, which is consist with our earlier in situ proteome profiling results (Fig. 3b,c). Of note, most fluorescence signals from both AP1NP and Rh-PEG-N 3 channels were cytosolic, indicating the presence of unknown cellular targets in addition to the nuclear-localized p50 from NFκ B (vide infra). Finally, to confirm APNP/AP1NP/AP2NP might be useful in deep-tissue imaging experiments with two-photon fluorescence microscopy (TPFM), two-day-old Drosophila brains were treated with APNP and successfully imaged with good resolution at a depth of 100 μ m (Supplementary Fig. S7). We thus conclude that these novel self-reporting, natural product-inspired imaging probes are capable of sensitive, real-time detection of WT's cellular activities in live cells and tissues.

Cellular Targets Validation. With a long history of remedial applications in TCM, Andrographis
paniculata is currently a World Health Organization (WHO)-listed herb 33 . Its active ingredient, WT, despite having relatively poor cellular activities (in potency and selectivity), is being pursued by many research labs as a promising drug candidate 18,19 . We were intrigued by this natural product because most of its reported cellular and pharmacological properties point to the likelihood that it belongs to a group of molecules called pan-assay interference compounds (PAINS), which are highly promiscuous and appear active in many biological assays 34 . The in situ proteome profiling capability of AP1 made it possible to test this hypothesis (Fig. 4). We were also interested to positively identify the 60-kDa labeled band, which appeared to be the most prominent cellular target of WT in A549 cells (a human lung adenocarcinoma epithelial cell line). Large-scale PD/LC-MS/MS experiments were performed with AP1-labeled cells and the gel slice at ~60-kDa region was cut, tryptically digested and analyzed (Fig. 4a,b); a total of six high-confidence candidate proteins were identified, and three cytosolic proteins were further confirmed to be true cellular targets of WT (Fig. 4c-e). NAMPT is a pleiotropic enzyme involved in a number of human diseases 35 . ALDH1B1 and GSR are two key proteins required to balance endogenous reactive oxygen species (ROS) 36,37 . PD/WB results indicate all three proteins were endogenously labeled in A549 cells even without NEM treatment. p50 on the other hand was positively labeled only in NEM-treated A549 cells. In HepG2 cells, they were labeled only after NEM treatment ( Supplementary  Fig. S10). In a cellular thermal shift assay (CETSA) 38 , all three proteins were efficiently stabilized by WT in A549 cells, indicating they were physically engaged in binding to WT in intake cells (Fig. 4d). p50 was similarly stabilized by WT in HepG2 cells. Cellular imaging of AP1-labeled A549 cells further indicates fluorescence signals from all three newly identified WT targets completely colocalized with those from the Rh-PEG-N 3 channel (Fig. 4e, panels 3/6/9), but not vice versa, indicating there were additional endogenous AP1 targets. Finally, 6 recombinant proteins known to possess nucleophilic cysteine residues were randomly chosen and labeled by AP1 (Fig. 4f); results showed all but one were positively labeled at 1-10 μ M (a concentration lower than most published WT protocols 18,19 ), but negated in the presence of excess WT. These results thus unequivocally confirm WT is indeed a highly promiscuous compound.

Discussion
In conclusion, the development of natural product-inspired dual-purpose probes has led to the successful reaction-based, real-time imaging of andrographolide activities in live mammalian cells and subsequent in situ proteome profiling and target identification. Our finding indicates andrographolide is a highly promiscuous compound, and engaged in covalent interactions with numerous previously unknown cellular targets at its pharmacologically relevant concentrations in cell type-specific manner. Because of this, the potential of these probes for live-cell imaging of specific cellular targets might be limited. The mechanism of cell type-specific targeting by andrographolide remains unknown as well, and is the subject of our ongoing investigation. Notwithstanding, we caution this bioactive compound should be further scrutinized before being seriously considered as a potential drug candidate.

Methods
Chemical synthesis. Information in detail was provided in Supporting Information.
In Situ Proteome Profiling. These experiments were carried out mostly based on previously published protocols 8,9 , with necessary modifications as shown below. Briefly, cells were grown to > 90% confluence in 6-well cell culture plates. After washing with PBS, cells were treated with different probes in 0.8 mL growth medium with a final DMSO concentration of < 0.1%. Probes were stored in DMSO at − 20 °C. AP1 and other probes were added into cell medium and incubated for 3 h or other indicated time. After incubation, the medium was removed and cells were washed with PBS for three times to remove excessive probes. In competitive labeling experiments, cells were pre-incubated for 1 h at 37 °C with WT (1 ~ 10-fold excess) before addition of 10 μ M probe. In NEM-competitive labeling experiments, cells were pre-incubated with NEM (1 mM or indicated concentrations) for 20 min at 37 °C, before addition of the probe (0 ~ 10 μ M). After treatments, cells were detached and lysed in 300 μ L PBS containing 0.02% Triton X-100. A freshly premixed click cocktail (4 eq. of Rh-PEG-N 3 , 4 eq. of TBTA (tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine), 40 eq. of TCEP (tris(2-carboxyethyl)phosphine) and 40 equivalents of CuSO 4 ) was added. The reaction was incubated at room temperature for 2 h with gentle shaking before termination by addition of five-fold volumes of pre-chilled acetone. Incubation was continued overnight at − 20 °C. Precipitated proteins were subsequently collected by centrifugation (13,000 rpm × 10 min at 4 °C). The supernatant was discarded and the residue pellet was washed with pre-chilled methanol, air-dried until the pellet started to shrink. Then the residue was resuspended in 1× standard SDS-loading buffer, sonicated for 10 min, heated for 10 min at 95 °C with gentle mixing, separated by SDS-PAGE, followed by in-gel fluorescence scanning and/or coomassie staining.

General procedures for one-and two-photon fluorescence imaging of live cells and tissue.
Cells were seeded in glass-bottom dishes (Mattek) and grown till 70 ~ 80% confluence. Subsequently, for imaging results shown in Fig. 3d & Supplementary Fig. S7, cells were incubated with APNP/AP1NP/ AP2NP (1 or 10 μ M in fresh growth medium) for 3 h. For experiments where G2 was used to confirm successful blocking of in-cell GSH activitie 32 , G2 was incubated with the cells for 20 min. Cells were next washed once with PBS, and then imaged with the Leica TCS SP5X Confocal Microscope System (panels 1/5/9/13 in Fig. 3d), as previously described [39][40][41] . NC or DMSO was used as imaging background controls ( Supplementary Fig. 6 & 7). For imaging of cells pre-treated with NEM, cells were incubated with NEM (1 mM) for 20 min prior to addition of the probes. Subsequently, for probes-treated cells, they were fixed with 3.7% formaldehyde in PBS for 1 h at 37 °C in CO 2 , washed twice again and permeabilized with 0.02% Triton X-100 in PBS for 10 min at room temperature, washed twice with PBS again. Subsequently, cells were treated with a freshly premixed click chemistry reaction in 200 μ L (1 eq. of Rh-PEG-N 3 , 2 eq. of TBTA, 20 eq. of TCEP, 20 eq. of CuSO 4 ) for 1 h at room temperature with gentle shaking. Cells were washed with 2× PBS, several times PBS containing 0.02% Triton X-100 and 0.1 mM EDTA (until there was no small crystal under microscopy), 2× PBS, 2× methanol, 2× PBS. The cells were incubated with nucleus stain (Hoechst, 0.2 μ g/mL final concentration) for 10 min at 37 °C and washed twice with PBS. Images were taken where indicated, as previously described 22 . For Drosophila 2P imaging, whole brains were prepared from a series of 2-day-old wild type Drosophila. Brains were incubated with APNP (50 μ M) in DMEM under a humidified atmosphere of 5:95 (v/v) of CO 2 /air at 37 °C for 3 h. Treated brains were subsequently transferred to poly-L-lysine-coated coverslips. The images were taken at 100-μ m depth by changing the z-axis thickness ( Supplementary Fig. 8) 22 . For immunofluorescence of NAMPT, ALDH1B1 and GSR, AP1-treated A549 cells were fixed with 3.7% formaldehyde and permeabilized with 0.02% Triton X-100, blocked with 2% BSA in PBS for 30 min at room temperature, followed by incubation with primary antibody anti-NAMPT (1:200), anti-ALDH1B1 (1:500) and anti-GSR (1:500), respectively, for 4 h at room temperature or 4 °C overnight. Upon washing with PBST (with 0.02% Triton X-100) twice for 5 min each and incubation with Cy5-conjugated Goat anti-Rabbit IgG (1:1000) for 2 h at room temperature, the cells were then washed with PBST (with 0.02% Triton X-100) three times for 5 min each and washed with PBS Scientific RepoRts | 5:11522 | DOi: 10.1038/srep11522 for 5 min with gentle agitation and a final wash with deionized water (1 ~ 2 min with gentle agitation) before images were acquired (Fig. 4e). All images were processed with Leica Application Suite Advanced Fluorescence (LAS AF) 22 . A real-time imaging video clip (time: 0 to 3 h) of HepG2 cells upon treatment with AP1NP (10 μ M; 3 h) are shown in SI_IV.
Target Identification/Validation. For cellular thermal shift assay (CETSA), experiments were carried out similarly as published protocols 38 . A549 cells were seeded in 10 cm cell culture dishes and grown till ~90% confluence. Subsequently, the cells were incubated with WT (100 μ M in fresh growth medium) for 3 h in the CO 2 incubator at 37 °C. The same volume of DMSO was used to serve as negative control in a separate dish. The medium was subsequently aspirated, and cells were washed with PBS (10.0 mL), then harvested by scraping in fresh PBS (20.0 mL). Cell pellets were isolated by centrifugation (1200 rpm, 5 min, 4 °C), resuspended in 1.2 mL PBS and distributed into 12 different 1.5 mL tubes with 100 μ L of cell suspension in each tube for both DMSO control and WT-treated cells. The tubes were heated at their designated temperature endpoints (40 ~ 73 or 52 ~ 85 °C) for 3 min on a heating block. Immediately after heating, the tubes were removed and incubated at room temperature for another 3 min. After that, cells were freeze-thawed twice by using liquid nitrogen and a heating block set at 25 °C in order to ensure a uniform temperature between tubes after keeping at − 80 °C overnight. 50 μ L of supernatant were collected by centrifugation (eppendorf centrifuge 5415R, 13, 000 rpm × 1 h) from each resulting cell lysate at 4 °C. Then the residue was mixed with 10 μ L 6× standard SDS-loading buffer, sonicated for 10 min, heated for 10 min at 95 °C with gentle mixing, separated by SDS-PAGE, followed by transfer to a PVDF membrane for WB analysis (Supplementary Fig. S12).
Other methods. Additional experimental information and results (chemistry and biology) are provided in Supplementary Methods.