Metabolic fingerprinting of banana passion fruits and its correlation with quorum quenching activity Author Castellanos, Leonardo ∗ & Universidad Nacional de Colombia - Sede Bogotá - Facultad de Ciencias - Departamento de Química, Carrera 30 # 45 - 03, Bogotá, D. C., 111321, Colombia & ∗ & Natural Products Laboratory, Institute of Biology, Leiden University, Sylviusweg 72, 2333 BE, Leiden, the Netherlands Author Naranjo-Gaybor, Sandra Judith Author Forero, Abel M. Author Morales, Gustavo Author Wilson, Erica Georgina Author Ramos, Freddy A. Author Choi, Young Hae text Phytochemistry 2020 112272 2020-04-30 172 1 13 http://dx.doi.org/10.1016/j.phytochem.2020.112272 journal article 10.1016/j.phytochem.2020.112272 1873-3700 8294125 2.3. Bioprospecting studies of Passiflora species related to QQ activity Having completed the metabolic profiling of the studied Passiflora species , the next step was to relate those profiles with the biological activity observed for the extracts in order to identify the compounds responsible for such an activity ( Wu et al., 2015 ). The selected bioactivity was quorum sensing inhibitory activity (QSI activity) because the search of anti-pathogenic compounds seemed to be a better strategy than the search for antibiotics, in terms of reducing the damage in the host, without generating induced resistance in the pathogen. Several small molecules including C -glycoside flavonoids, vanillin, 3-indolyacetonitrile, among others have been reported to be quorum sensing inhibitors ( Grandclément et al., 2016 ), ( Brango-Vanegas et al., 2014 ). The MeOH/H 2 O extracts of Passiflora species were tested for the inhibition of violacein production using Chromobacterium violaceum ATCC 31532 as a biosensor (supporting info table 4). Results showed that P. uribei , P. lehmannii and P. cumbalensis exerted a strong activity (inhibition halo> 40 mm ) (Fig. 17 supporting information) while other Passiflora samples showed less or no activity at all. The complete results are summarized in Supp. Table 4. The metabolites that had been detected by 1 H-NMR were correlated with the bioactivity (QSI) by applying the orthogonal projection to latent structures (OPLS-DA), using the coded QSI activity (20 mm-inhibition zone was coded as 1;> 30 mm of inhibition zone was coded as 3) as the Y-variable. Separation of the active groups is observed in the OPLS-DA score plot (R 2 = 0.425 and Q 2 = 0.302, pareto scaling), with the active groups on the negative side along OPLS1 ( Fig. 7A ). Passiflora cumbalensis clustered as a well-defined active group, while the other species did not show a clear clustering tendency. Three active groups were identified along the OPLS2 axis, one being on the negative side for P. lehmannii and P. uribei , one on the positive side for P. cumbalensis and a third one for the other species spread out in the middle of the plot, suggesting that the active compounds for these three groups were different. Fig. 6. PCA plot: The score plot of the principal component analysis (PCA) of 8 different species of banana passion fruits species shows a separation into four main groups. Using two S-plots , one excluding P. lehmanii samples ( Fig. 7B ) and the other excluding P. cumbalensis samples ( Fig. 7C ) it was possible to identify the active compounds. The variables important for the projection (VIPs) were selected, and the chemical shifts responsible for the QSI activity were highlighted. These highlighted chemical shifts were found to correspond mostly to the glycosylated flavonoids because the signals could be assigned to aromatic protons such as those of the A and B rings from flavonoids as well as signals for sugar moieties, including those of the anomeric protons close to 5 ppm (Tables 5 and 6, Supporting info). The quality and robustness of the OPLS-DA model was validated by a permutation test (n = 100). The Q 2 intercept value was −0.504 (below 0.05), showing that the original model was statistically effective (Fig. 18 Supporting info). The model was validated by calculating the area under the receiver operating characteristic ( ROC ) curve. The value of the area under the curve (AUC) was 0.9565 providing added confidence to the model (Fig. 18B supporting info). Pure compounds 1 and 2 were tested for their QS inhibition against C. violaceum at five concentrations in the range of 50 μM–400 μM in a 96 well-plate. The QS inhibition of compound 1 and compound 2 was detected at concentrations of 100 μg/mL (0.13 mM) and 300 μg/mL (0.47 mM) respectively. In order to establish whether the observed inhibition was due solely to QS inhibition and not to growth inhibition, samples were submitted to a growth inhibition test (Fig. 19 supporting information). Results of the assays showed not only the absence of growth inhibition but an increase in bacterial cell densities, indicating that the flavonoids likely inhibited cell communication. A second model, Burkholderia glumae , a well-known phytopathogen that causes rice grain rot and wilt in various field crops was also used to evaluate QSI ( Compant et al., 2008 ). In B. glumae , the production of toxoflavin (a bright yellow pigment) is known to be one of the major virulence factors ( Jeong et al., 2003 ; J. Kim et al., 2004 ). The biosynthesis of toxoflavin is controlled by ToxR, a LysR-type transcriptional regulator and this toxin also activates the expression of the tox operons (J. Kim et al., 2004 ). For this reason, the search for compounds that are able to inhibit toxoflavin production is an important target for the control of this phytopathogen. Two strains were chosen to determine the toxoflavin inhibitory activity of extracts and pure compounds. Burkholderia glumae COK 71, is a biosensor strain, that is highly specific for toxoflavin based on β- galactosidase activity on X gal substrate that produces a blue pigment, and the B. glumae ATCC 33617 strain as a toxoflavin producer. In this test, the levels of the blue pigment are used to determine toxoflavin inhibitory activity ( Choi et al., 2013 ). Our results indicated that toxoflavin productions was inhibited by concentrations of 6.76 μM and 7.87 μM of compounds 1 and 2, respectively, while the positive control, 2- n -propyl-9-hydroxy-4H-pyrid [1,2-a] pyrimidin-4-one was active at 80 μM, showing the potential of these flavonoids to control toxin production by the phytopathogen, B. glumae (Fig. 20, supporting information). The presence of flavonoids in plant extracts has been previously related to their QS inhibition activity. Phytochemical screening of Centella asiatica has revealed that flavonoids can disrupt AHL-mediated QS-controlled systems in C. violaceum and P. aeruginosa while major constituents such as the triterpene, asiatic acid, did not show an anti-QS activity ( Vasavi et al., 2016 ). Concentrations of 100 μg/mL of quercetin and kaempferol have been reported to exhibit anti-QS activity against C. violaceum and P. aeruginosa PAO 1. The anti-QS activity of Psidium guajava leaf extract has been determined with a biosensor bioassay using Chromobacterium violaceum CV 026, and quercetin and quercetin 3- O -arabinoside were identified as the QQ compounds in the extract, against C. violaceum 12,472, at concentrations of 50 and 100 μg/mL, respectively ( Vasavi et al., 2014 ). Similarly, Paczkowski et al. studied the QS inhibition mechanism of flavonoids, establishing that they are inhibitors of the QS transcriptional regulator LasR and that they specifically inhibit quorum sensing via antagonism with the transcriptional regulator LasR/RhlR. Further structure-activity relationship analyses suggest that the presence of two hydroxyl moieties in the flavone A-ring backbone are essential for potent inhibition of LasR/RhlR. Biochemical analyses also revealed that flavonoids function non-competitively to prevent LasR/RhlR DNA-binding. The administration of the flavonoids to P. aeruginosa was found to alter transcription of the quorum-sensing controlled target promoters and suppress virulence factor production, confirming their potential as antimicrobials which do not function by traditional bactericidal or bacteriostatic mechanisms ( Paczkowski et al., 2017 ).