Exploring resveratrol dimers as virulence blocking agents – Attenuation of type III secretion in Yersinia pseudotuberculosis and Pseudomonas aeruginosa

Bacterial infections continue to threaten humankind and the rapid spread of antibiotic resistant bacteria is alarming. Current antibiotics target essential bacterial processes and thereby apply a strong selective pressure on pathogenic and non-pathogenic bacteria alike. One alternative strategy is to block bacterial virulence systems that are essential for the ability to cause disease but not for general bacterial viability. We have previously show that the plant natural product (-)-hopeaphenol blocks the type III secretion system (T3SS) in the Gram-negative pathogens Yersinia pseudotuberculosis and Pseudomonas aeruginosa. (-)-Hopeaphenol is a resveratrol tetramer and in the present study we explore various resveratrol dimers, including partial structures of (-)-hopeaphenol, as T3SS inhibitors. To allow rapid and efficient assessment of T3SS inhibition in P. aeruginosa, we developed a new screening method by using a green fluorescent protein reporter under the control of the ExoS promoter. Using a panel of assays we showed that compounds with a benzofuran core structure i.e. viniferifuran, dehydroampelopsin B, anigopreissin A, dehydro-δ-viniferin and resveratrol-piceatannol hybrid displayed significant to moderate activities towards the T3SS in Y. pseudotuberculosis and P. aeruginosa.

inhibition of yopE expression and YopH secretion. The compounds were tested for inhibition of the T3SS in the combined yopE-lux and YopH phosphatase assay for dose-dependent activity as described previously 8 . In addition, inhibition of bacterial growth was measured to allow identification of T3SS selective inhibitors with little or no effect on bacterial viability. The results are compiled in Table 1. The direct half of (-)-hopeaphenol (1) i.e. ampelopsin A (2) and B (3) as well as dehydroampelopsin B (4), which all contain a central 7-membered ring structure, showed no or modest inhibition of the T3SS. Similar data was obtained for the related opened form compounds ε-viniferin (5) and ω-viniferin (7) (IC 50 > 50 μM, yopE). The related benzofuran based viniferifuran (8), on the other hand, showed good inhibition of the T3SS (Fig. 1) with an IC 50 of 10 μM (yopE) and 4 μM (YopH), without any effect on bacterial growth, i.e. the inhibition profile is similar to that of the original lead (-)-hopeaphenol. The fact that the inhibition profile of the resveratrol tetramer (-)-hopeaphenol is maintained by the resveratrol dimer viniferifuran is encouraging since it has a simplified structure that can be prepared in gram quantities in only four steps 11 . The resveratrol-piceatannol hybrid (10) that differs from viniferifuran in its hydroxylation pattern has a similar inhibition profile as viniferifuran in vitro. δ-Viniferin (9) with the stilbene substitution at position 6 displayed a moderate activity (IC 50 = 29 μM, yopE) and the corresponding benzofuran analog dehydro-δ-viniferin (10) was found to be equally potent as viniferifuran. Anigopreissin A (11), with the stilbene group in position 7, inhibits the T3SS, but its detrimental effect on bacterial growth indicates that the inhibition is non-selective (Fig. S1, Table 1). All compounds with a benzofuran core structure i.e. viniferifuran, dehydroampelopsin B, anigopreissin A, dehydro-δ-viniferin and the resveratrol-piceatannol hybrid displayed significant to moderate activities (IC 50 = 5-40 μM, yopE).
Viniferifuran is an irreversible inhibitor of T3SS. Viniferifuran was further tested for reversibility using a Western blot analysis as described previously 8 . Vinferifuran (8) and (-)-hopeaphenol (1) were added to the bacteria and after induction of T3SS, the compounds were washed away and the bacterial solution was divided into www.nature.com/scientificreports www.nature.com/scientificreports/ Continued two, where one was treated again with compound under T3SS inducing conditions (-Ca 2+ ) and one was used as a control by adding dimethyl sulfoxide (DMSO) under inducing conditions. Both compounds irreversibly block secretion of effector molecules, i.e. the T3SS could not be reactivated by washing (Fig. 2).

Binding of viniferifuran to Yersinia pseudotuberculosis. Propagylation of viniferifuran with prop-
argyl bromide resulted in three mono-propargylated compounds that could be separated by chromatography (Supplementary information). These fractions were tested for their biological activity and compound 13 was selected for further experiments due to its retained biological activity (Table 1) as well as the fact that it could be produced in the highest amounts of all of the mono-propargylated products. Y. pseudotuberculosis (YPIII (pIB102, pFU95gfp)), which constitutively expresses green fluorescent protein (GFP), was treated with compound 13 or DMSO followed by cupper catalyzed click chemistry of the propargyl functionality with the azide group of sulfo-cyanine azide. The bacteria were analyzed using confocal spinning disc microscopy (The CellASIC ONIX2 Microfluidic System, Merck). All bacteria display green fluorescence from the constitutively expressed GFP and the bacteria that has bound propagylated viniferifuran labeled with the azide fluorophore appear red. The results showed a clear binding of viniferifuran to the bacteria ( Fig. 3b in top panel). Competition experiments showed that the binding could be outcompeted by adding five times more of free viniferifuran compared to the propargylated compound 13 (Fig. 3). Furthermore, the results showed that the compound binds directly to the bacteria without causing aggregation of the bacteria.
inhibition of exoS expression in P. aeruginosa. The compounds were further tested for efficacy on the opportunistic pathogen P. aeruginosa. A reporter-gene was constructed where the promoter of the P. aeruginosa T3S-toxin, ExoS, was transcriptionally fused to the GFP reporter gene. The gene construct was then cloned into the high copy plasmid pUCP18, together with orf1, coding the chaperon of ExoS 15 . To be able to select for the plasmid, the gene coding for gentamicin resistant was inserted into the plasmid, creating plasmid pCS500  www.nature.com/scientificreports www.nature.com/scientificreports/ (Supporting information). The resulting plasmid was inserted into wild-type P. aeruginosa strain PAK and the T3SS activator mutant PAKexsA, resulting in strains that expresses the GFP when the ExoS promoter is activated. The T3SS is induced by incubating the bacteria in calcium depleted medium at 37 °C. The wild-type PAK and the PAKexsA strains with the reporter plasmid were grown under induced conditions and the compounds were added to the growth medium at different concentrations. After 16 h of incubation the GFP signal was measured in a plate reader (Biotek Synergy) and the inhibition of ExoS expression was calculated. The PAK(pCS500) control (no compound added) showed a high GFP level corresponding to a high expression of exoS, while the PAKexsA(pCS500) control strain showed a low level of GFP corresponding to a low level of exoS expression. The level of GFP seen by the PAKexsA(pCS500) strain was equal to a full inhibition of exoS by a fully active compound ( Figure S3). Simultaneously, the inhibition of bacterial growth was measured to select for selective T3SS inhibitors. All compounds with a benzofuran core structure i.e. viniferifuran (8), dehydroampelopsin B (4), anigopreissin A (11), dehydro-δ-viniferin (10) and the resveratrol-piceatannol hybrid (12) exhibited significant activities by   Table 2). However, compounds containing the central 7-membered ring structure displayed higher IC 50 values, as exemplified by comparison of dehydroampelosin B (4, IC 50 33 µM) and viniferifuran (8, IC 50 11 µM) ( Table 2). None of the compounds showed inhibition of bacterial growth and could therefore be considered as selective T3SS inhibitors in P. aeruginosa.

Inhibition of T3S-mediated cytotoxicity of eukaryotic cells. The T3SS is important for P. aeruginosa
infection of eukaryotic cells. A T3SS activator mutant or a mutant lacking the T3SS tip protein PcrV is not able to infect eukaryotic cells 16,17 . The J774 cells (ATCC) were infected with the wild-type PAK strain and the T3SS mutant with or without compounds. The T3SS mediated cytotoxic effect on the eukaryotic cells was measured by using a viability staining method, Uptiblue (Interchim), after 4, 5 and 6 h of infection. Simultaneously, uninfected cells were treated with compounds and toxicity towards eukaryotic cells was determined using the same assay. The cells were also studied by microscopy after 6 h to assess cell morphology. The results showed a clear dose-dependent rescue effect of (-)-hopeaphenol (1), with complete inhibition of cytotoxicity at 50 µM, which is in agreement with previous results 8 (Fig. 5). Additionally, dehydro-δ-viniferin (10) and resveratrol-pinceatannol hybrid (12) showed a significant dose-dependent rescue effect on eukaryotic cells, while viniferifuran (8), (±)-ε-viniferin (5) and anigopreissin A (11) showed modest or no efficacy (Fig. 5). Viniferifuran (8) and dehydro-δ-viniferin (10) showed substantial toxicity towards eukaryotic cells at 100 µM, while some toxicity could be seen for the resveratrol-pinceatannol hybrid (12, Figure S2). Toxicity towards eukaryotic cells limits the possibility to study and accurately assess virulence blocking properties of these compounds.

Discussion
Most antibiotics in clinical use today are either natural products of microbial origin or improved derivatives obtained by synthesis. Development of new antibiotics based on new synthetic scaffolds have proved to be challenging and this holds true also for identification of chemical entities that block virulence systems by mechanisms distinct from conventional antibiotics e.g. attenuation of T3SS in Gram-negative pathogens. A substantial number of synthetic T3SS inhibitors have been identified by predominantly phenotypic screening, and a selection of compounds have entered medicinal chemistry programs as summarized in recent review articles 18,19 . Several studies indicate that natural sources also have the potential to provide T3SS inhibitors for further exploration. Screening of marine invertebrate extracts led to the identification of the glycolipid caminoside A as an inhibitor of T3SS in  (8), dehydro-δ-viniferin (10), anigopreissin A (11) and resveratrol-piceatannol hybrid (12). Wild-type P. aeruginosa PAK(pCS500) with GFP under the control of the exoS promoter was treated with different concentration of compounds under T3SSinducing contitions. As controls the the untreated PAK(pCS500) (wt) and PAKexsA(pCS500) (exsA − ) were used. The expression of exoS was calculated as percenteage of the untreated wild-type PAK(pCS500). Samples were run in triplicates and the experiment was repeated three times. The graph shows the mean value (n = 3) +/−SD.   20 and later caminocide B-D were also found to be T3SS inhibitors 21 . In addition to activity against T3SS in EPEC, the caminosides also display antimicrobial activity against vancomycin resistant Enterococcus and methicillin resistant Staphylococcus aureus, suggesting that these compounds act as general antibiotics rather than selective T3SS inhibitors 20,21 . In another study guadinimine A-F, isolated from a soil-derived Streptomyces broth, were found to block T3SS dependent hemolysis of erythrocytes by EPEC 22,23 .
Aurodox is a known actinomycete derived antibiotic that inhibits protein biosynthesis by binding to bacterial elongation factor Tu 22,23 . Recently it was reported that aurodox inhibited EPEC T3SS-mediated hemolysis and protected mice infected with a lethal dose of Citrobacter rodentium 24 . The mechanism for its T3SS inhibition has however not been established. Pseudoceramines A-D and spermatinamine produced by the marine sponge Pseudoceratina spp. were identified as putative inhibitors of T3SS in Y. pseudotuberculosis, but the compounds also exhibited a detrimental effect on bacterial growth, indicating general antibacterial properties 25 . Phenolic compounds involved in plant defense signaling have been reported to block the T3SS in P. aeruginosa via the GacS-GacA two-component signal transduction system 26 . Cytosporone B, a fungal metabolite, and analogues selectively block the secretion of Salmonella pathogenicity island 1 (SPI-1)-associated effector proteins, without significant toxicity against bacteria 27 and similar effects have been reported for prenylated plant flavonoids, of which licoflavonol proved to be the most promising compound 28 . It was recently reported that the T3SS of Y. pseudotuberculosis is effectively blocked by piericidin A1, a metabolite produced by a marine actinobacterium 29 . The compound was identified by screening of a library composed of marine-derived extracts and its virulence blocking activity is not attributed to general toxicity towards the bacteria. Based on our finding that the plant derived complex resveratrol tetramer (-)-hopeaphenol (1) is an irreversible T3SS inhibitor we tested a series of resveratrol dimers that can be obtained by biomimetic or total synthesis. We identified the resveratrol dimer viniferifuran (8) and analogues as inhibitors of T3SS that target both Y. pseudotuberculosis and P. aeruginosa. Since we could observe inhibition of T3SS-toxins in both Y. pseudotuberculosis and P. aeruginosa it is not likely that the resveratrol dimers act on the toxins per se, since they are different between the bacterial species. The compounds rather act on a homologous function of the T3SS as the needle like structure or tip-proteins on the bacterial surface or some common regulatory system. Viniferifuran was found to irreversibly block the T3SS, indicating that (-)-hopeaphenol and viniferifuran might act by the same mechanism, although the chemical structures are substantially different. The efficacy of viniferifuran in cell infection experiments could not be accurately assessed due to its toxicity toward J774 cells.
Due to its ease of preparation, we still consider viniferifuran a promising scaffold amendable for medicinal chemistry to optimize its potency toward diverse Gram-negative pathogens and reduce its toxicity. The resveratrol-piceatannol hybrid (12) and the dehydro-δ-viniferin (10) showed better efficacy to rescue eukaryotic cells from infection, however they both show toxicity towards the eukaryotic cells at high concentrations, even though the toxicity of the resveratrol-piceatannol hybrid is minor compared to viniferifuran. (-)-Hopeaphenol is non-toxic toward both bacteria and eukaryotic cells and possibly this can be attributed to its size, which likely prevents cell permeability and suggest that (-)-hopeaphenol acts on the bacterial surface.
Incubation of Y. pseudotuberculosis with the monopropargylated viniferifuran derivative 13, followed by click chemistry using sulfo-cyanine azide and subsequent confocal spinning disc microscopy, suggested that Figure 5. Effect of the (-)-hopeaphenol analogues on P. aeruginosa infection. J774 cells were infected with wild-type P. aeruginosa strain PAK and compounds were added and the T3SS dependent cell survival was analysed using a viability staining method. As controls uninfected cells, cells infected with the PAKpcrV-mutant and cells infected with the wild-type strain (PAK) were used. The survival was calculated as percenteage of the uninfected control. Samples were run in triplicates and the experiment were repeated three times. Viniferifuran (8) and dehydro-δ-viniferin (10) were very toxic at 100 µM, while some toxicity could be seen for resveratrolpinceatannol hybrid (12). The graph shows the mean value (n = 3) +/−SD after 6 h of infection.

Scientific RepoRtS |
(2020) 10:2103 | https://doi.org/10.1038/s41598-020-58872-0 www.nature.com/scientificreports www.nature.com/scientificreports/ viniferifuran might be located at the outer membrane of the bacteria but an intracellular localization cannot be excluded. Importantly, the pictures showed that the compound binds to the bacteria and the T3SS inhibition is not due to e.g. aggregation of bacteria.
Interestingly, all the most promising compounds in the present study share the unsaturated benzofuran scaffold as opposed to parent compound (-)-hopeaphenol that contains the related dihydrobenzofuran scaffold. Other compounds with the dihydrobenzofuran core e.g. ampelopsin A (2) and B (3) were essentially void of activity.
In conclusion, we have identified resveratrol dimers as novel inhibitors of T3SS in Y. pseudotuberculosis and P. aeruginosa. Compared to the parent compound, (-)-hopeaphenol, these compounds are amendable to medicinal chemistry and are thus useful as starting points to develop lead compounds and initiate drug discovery programs.
inhibition of bacterial growth. For growth control, the Y. pseudotuberculosis serotype III strain YPIII(pIB102E-lux) ( Table 3) was grown overnight in LB supplemented with chloramphenicol, diluted to an OD 600 of 0.2 in LB medium with 2.5 mM CaCl 2 and added into 96-well plate 100 µl/well (Nunc flat bottom, transparent). Compounds were then added and the plate was incubated at 37 °C for 24 h, shaking. The absorbance at OD 600 was measured every hour for 8 h and then after 24 h in a microplate reader (Wallac Viktor 2 1420 Multilabel counter).  www.nature.com/scientificreports www.nature.com/scientificreports/ Western blot analysis of supernatant samples of Y. pseudotuberculosis for studying reversibility.
The experiment was performed essentially as described earlier 8 . A final concentration of 40 µM of the compounds were added to a 1:20 diluted overnight culture of YPIII(pIB102). Control cultures received only DMSO. The cultures were incubated at 26 °C for 30 min followed by 2 h at 37 °C to induce T3SS. The samples were then centrifuged and the pellet washed once with LB where after it was suspended in LB complemented with 20 mM MgCl 2 and 5 mM EGTA for Ca 2+ depletion (induction of T3SS) or 2.5 mM CaCl 2 . The samples were divided and 40 µM compound or DMSO was added followed by incubation for 45 min at 37 °C. The cultures supernatant was mixed with sample buffer and loaded on to a 12% sodium dodecyl sulfate (SDS) polyacrylamide gel and blotted onto a PVDF membrane. Polyclonal rabbit anti Yop-antiserum and polyclonal goat anti-rabbit immunoglobulin conjugated with horseradish peroxidase (HRP) (Dako Denmark) were used together with Millipore Immobilon Western chemiluminescent HRP substrate to determine Yop proteins in the supernatant. The chemiluminiscense signals were detected with a charge-coupled device (CCD) camera.
Visualizing viniferifuran binding to bacterial cells. The Y. pseudotuberculosis serotype III strain YPIII (pIB102; pFU95) ( Table 3) was grown overnight in LB supplemented with carbenicillin. 100 µl YPIII (pIB102; pFU95) without compound in calcium depleted LB medium at an OD 600 of 0.11, was mixed with propagylated viniferifuran (13) to a final concentration of 10 µM. To investigate selectivity of the compound viniferifuran (8) was added to a final concentration of 50 µM together with propagylated viniferifuran (13). The DMSO concentration was kept to 1% in all samples and bacteria treated only with 1% DMSO were used as control. The samples were incubated in 96-well plates for 1 h at 26 °C and 2 h in 37 °C with shaking. After incubation, the bacteria were diluted 1/5 in phosphate buffered saline (PBS) and 50 µl was loaded into a CellASIC ONIX B04A-03 Microfluidic Bacteria Plate of the CellASIC ONIX2 Microfluidic System (Merck), with a flow rate of 15 s at 4 psi twice to get them into the plate layers. After additional 5-10 times at 6 psi the bacteria reached the correct position (trap heights of 0.7 µm).
The click chemistry reaction mixture was made with a final concentration of CuSO 4 (0.1 mM), tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (TBTA, 128 µM), sodium ascorbate (5 mM), sulfo-cyanine5 azide (100 µM, Lumiprobe, ex/em 646/662) in PBS. The plate was washed with for PBS 10 min at 4 psi, where after the click reaction mix was added for 5 min at 8 psi and 60 min at 4 psi. Finally, the plate was washed with PBS for 10 min at 8 psi and 20 min at 4 psi and the results were analyzed using a Confocal spinning disc microscope, far red 647 laser. The plate was incubated at 37 °C during the experiment. cloning and exoS expression assay. The pRIC-GFP plasmid was cut with PstI and EcoRI and the GFP fragment was purified and ligated into pTS103 plasmid 15 cut with NsiI and EcoRI. The resulting plasmid was purified and cut with BamHI and EcoRI and ligated into the pUCP18 plasmid cut with BamHI and EcoRI. The resulting plasmid was cut with ScaI (blunt, cutting in the amp gene) and the gentamicin (Gm) gene was inserted by cleaving the p34S-Gm plasmid with Ecl136II (SstI) (blunt). The resulting plasmid, pCS500, was resistant to gentamicin and the gfp-gene is transcriptionally fused with the exoS promoter. pCS500 was transformed into the P. aeruginosa strain PAK and the T3SS activator mutant PAKexsA.
Overnight cultures of PAK(pCS500) and PAKexsA(pCS500) were diluted 1:30 in LB supplemented with 3 µg/ ml Gm and incubated at 37 °C for 6 h on a rotary shaker. The cultures were then diluted to an OD 600 of 0.0004 in calcium deleted LB supplemented with 3 µg/ml Gm and added into a flat-bottom 96-well transparent plate (Nunc). Compounds were added to give a final concentration of 1-100 µM. As a positive control 1% DMSO was added to the PAK(pCS500) strain and as a negative control 1% DMSO was added to the PAKexsA (pCS500). The fluorescence was measured after 16 h of incubation on a rotary shaker (ex/em, 488/509) in a plate reader (Biotek Synergy). The value for the wt PAK(pCS500) strain with no additives was set to 100% GFP expression (i.e. 100% ExoS expression). The IC 50 values were calculated using GraphPad Prism software. Inhibition of bacterial growth was also measured after 16 h using the same plate by measurement of the absorbance at OD 600 . All experiments were carried out in triplicates.
infection and toxicity assay. J774 cells (ATCC) was seeded to a concentration of 5 × 10 4 cells/well (100 µl/ well) in 96-well plates and incubated 18 h prior to infection in Dulbecco's modified eagle medium (DMEM, GlutaMAX (Glu), 3 µg/ml Gentamicin (Gm) + 10% FCS). Overnight cultures of PAK and PAKpcrV were diluted 1:10 in DMEM (Glu) medium and incubate at 37 °C, 250 rpm for 1 h. Meanwhile, the cells were wash once with PBS pH 7.2, where after 30 µl of compound solutions was added diluted in DMEM (Glu, 10% FCS). 1% DMSO was added to the control wells. 30 µl of the induced PAK bacteria solution, OD 600 = 0.0008, was then added to the wells giving a final OD 600 of 0.0004. PAKpcrV mutant, deficient for translocation of exoenzymes, was used to control that the effect on the cells was T3SS dependent. In a parallel plate, the compound solutions were added without bacteria to screen for toxicity. A viability staining method to investigate cell survival, after 2.5 h, 10 µl Uptiblue (Interchim, France) was added to each well, efficacy and toxicity of the compounds was measured 4, 5 and 6 h after infection (ex/em, 535/595) in a plate reader (Biotek Synergy). For efficacy, the compounds were tested at 50, 25, 12.5 and 6.25 µM in triplicates.