Materials and methods

Plant material. Plants were collected in December 2016 in the west-central area of the State of Mexico. T. recurvata and T. schiedeana were collected in "La Pedrera", Ixtlahuaca (19° 10. 662' N; 100° 15.303' W), altitude 1,368 m asl, and T. fasciculata was collected in "El Pedregal", Santo Tomás de los Plátanos (19° 11.167' N; 100° 15.912' W), altitude 1,216 m asl. The specimens were identified by Dr. María Flores Cruz (Figure 1 A-C) and deposited in the Collection of Living Bromeliads at the University Center for Conservation and Research of Mexican Bromeliads (CUCIBROM).

Extract preparation. The dried and powdered tissues of the whole plant (10 g) were de-fatted with 200 mL of hexane (JT Baker). They were extracted sequentially with similar volumes of dichloromethane (JT Baker) and methanol (JT Baker), evaporated under reduced pressure (Buchi-R114, Switzerland) and kept refrigerated at 4 °C. Organic extracts were subjected to chemical testing following the methods described by ^{Soto-Hernández et al. (2019)} for identification of terpenes, flavonoids, phenols and tannins.

Fractionation and chemical identification. The samples of T. recurvata (9.39 g) and T. schiedeana (200 g) were sequentially extracted (1:2 w/v) with hexane (to defatted, × 1), dichloromethane (D, × 3) and methanol (MeOH, × 3). Evaporation was carried out with a rotary evaporator under low pressure, yielding T. recurvata-DCM (Tr-D, 131 mg), T. recurvata-MeOH (Tr-MeOH, 488 mg), T. schiedeana-DCM (Ts-D, 2.79 g) and T. schiedeana-MeOH (Ts-MeOH 14.79 g) of crude extracts.

The fractionation of the extracts was performed by low pressure column chromatography (silica gel 60, 70-230 mesh, Merck^®) with increasing polarities of hexane and ethyl acetate (J.T. Baker^®) (Figure S1 and S2). In Tr-D, 1.6 mg of a solid (Tr-D4) was obtained, while in Tr-MeOH, 2.5 mg (Tr-MeOH3) and 5 mg (Tr-MeOH6) of other compounds were precipitated.

In Ts-D, 37 fractions were obtained, which were grouped into four final fractions according to their thin layer chromatographic profile: Ts-D13-15 (13.3 mg), Ts-D18-23 (15 mg), Ts-D33-35 (1.3 mg) and Ts-D36-37 (1.3 mg). In Ts-MeOH, the mother liquor was separated by low-pressure column chromatography (silica gel 60, 70-230 mesh, Merck^®), obtaining 61 fractions, from which it was possible to isolate Ts-MeOH57-61 (5.8 mg), Ts-MeOH31 (2.1 mg), Ts-MeOH11 (traces) and Ts-MeOH9 (21.5 mg). In the case of T. fasciculata, the biological material available was not sufficient to carry out a fractionation.

Finally, the identification by physical methods was performed by ¹H NMR, using a Bruker-Avance 300 spectrometer (300 MHz), and dissolving each sample in CDCl₃. Tetramethylsilane (TMS) was used as internal reference. Chemical shifts were expressed in parts per million (ppm) and coupling constants (J) were reported in Hz.

Inhibition of violacein production. Overnight cultures of C. violaceum ATCC 553 were adjusted to OD_620nm = 0.1 (Dynamica Halo MPR-96), and 200 μL per condition were added in a 96-well microplate (Corning^® Costar). The extracts were dissolved in dimethyl sulfoxide (DMSO, Merck) and 5μL was added to obtain final concentrations of 62.5, 125, 250 and 500 μg mL^-1. Subsequently, they were incubated at 28 °C, 150 rpm. DMSO was used as a negative control and a mixture of anacardic acids (AA) as a positive control (^{Castillo-Juárez et al. 2013}).

A plate count was performed to determine the effect on bacterial growth at 48 h. The determination of violacein production was carried out according to ^{Castillo-Juárez et al. (2013)}. The pigment was extracted with ethyl acetate (J.T. Baker) and evaporated at room temperature for 24 h. The dried violacein was solubilized in 200 μL of 80 % ethanol and quantified at 620 nm. Experiments were performed in triplicate. The results were converted to percentage of violacein produced with respect to the untreated control (DMSO).

Inhibition of QS-regulated factors in P. aeruginosa. Overnight cultures of P. aeruginosa PA14 strain (37 °C, 200 rpm) were adjusted to OD_{600 nm} = 0.05 in 5 mL of LB medium (Spectronic^® Genesys 5) and incubated until the turbidity reached OD_600nm = 1.0, and then when 62.5μL of each sample was added to obtain a final concentration of 250 μg mL^-1. The cultures were incubated for 6 h more and production of virulence factors and growth at 600 nm was quantified.

Pyocyanin production was determined as follows: cultures were centrifuged at 13,000 rpm for 5 minutes. To the supernatant (800 μL) of each culture, 450 μL of chloroform was added and vigorously stirred for 2 minutes. It was again centrifuged to recover the organic phase, to which 800 μL of 0.2 N hydrochloric acid was added. Subsequently, 650 μL of the aqueous phase was separated and the same volume of water was added. Finally, absorbance was measured at 520 nm (Spectronic^® Genesys 5). Bacterial growth was measured at 600 nm. DMSO was used as the negative control, and production of pyocyanin was normalized by growth.

Elastolytic activity in the supernatants was measured by the Elastin-Congo red (ECr) (Sigma Aldrich) assay. Five milligrams ECr was mixed with 950 μL buffer (100 mM Tris-HCl, 1 mM CaCl₂, pH 7.5) and 50 μL of the supernatant previously diluted with buffer (1:10, v/v). This mixture was incubated 2 h (37 °C at 200 rpm) and centrifuged for 5 minutes at 13,000 rpm. Absorbance of the supernatant was measured at 495 nm (Spectronic^® Genesys 5). At each value, the absorbance of the control was subtracted (LB medium diluted 1:10). Activity was expressed in 495 nm absorbance/growth (600 nm).

Alkaline exoprotease production was determined using Hide-Remazol Brilliant Blue R (HRBBR) (Sigma Aldrich) as substrate. To 875 μL of reaction buffer (20 mM Tris-HCl, 1 mM CaCl₂, pH 8.0) were added to 125 μL of bacterial culture supernatant (diluted 1:10 in buffer) and 5 mg of HRBBR. The mixture was incubated (35 minutes at 37 °C and 200 rpm) and was centrifuged at 13,000 rpm for 5 minutes to remove the unhydrolyzed portion. Absorbance was determined at 595 nm (Spectronic^® Genesys 5).

Biofilm formation was evaluated by microtiter plate method. Overnight culture (15 h at 37 °C, 200 rpm) of PA14 was adjusted to an OD_600nm = 0.05 (Spectronic^® Genesys 5) and 200 μL per sample was added in 96-well plates (Corning^® Costar). Subsequently, 5 μL of the extracts were incorporated to obtain concentrations of 62.5, 125, 250 and 500 μg mL^-1. DMSO was used as negative control and furanone C-30 at 50 μM was a positive control. The mixture was incubated without shaking for 24 h at room temperature. The biofilm adhered to the plate was stained with 200 μL of crystal violet (0.1 %, w/v) and washed three times with distilled water. The adhered dye was solubilized with 80 % ethanol and absorbance was measured at 570 nm (Thermo Scientific).

Soft-agar motility assay was performed in order to examine swarming motility of bacteria. Extracts (250 and 500 µg mL^-1) were added to M8 minimal medium supplemented with 0.2 % glucose, 0.5 % casamino acids, 1 mM MgSO₄ and 0.5 % agar (^{Ha et al. 2014}). The assay was performed in 6-well plates that were inoculated with 2.5 μL of an overnight culture of PA14 (12 h). After 12 h of incubation at 37 °C (Riossa E-71 culture oven), motility diameters were measured, and the results were transformed to swarming percentage with respect to the control (DMSO). The assay was performed with three independent cultures.

Statistical analysis. Data were processed with SigmaPlot software version 10 (Systat Software, Inc., San Jose California USA). Violacein production and biofilm formation were analyzed through ANOVA and the Dunnett test (α = 0.05) to compare the means of the treatments with DMSO (control). Pyocyanin, protease, elastase production and motility were compared using the Student´s t-test (P ≤ 0.05). Both tests were performed in the Statistic software V.9 Analytical Software, Tallahassee, FL, USA.

Results

Quorum sensing inhibition in Chromobacterium violaceum. The organic extracts of the three species reduced violacein production without affecting bacterial growth, suggesting that they contain metabolites whit quorum quenching activity. At the lowest dose evaluated (62.5 μg mL^-1), the CH₂Cl₂ extracts reduced pigment production by 43 to 52 %, but only T. recurvata and T. schiedeana were able to inhibit 68 to 76 % at 250 μg mL^-1 (Figure 2A). This effect is similar to that exhibited by the AA used as positive control.

The CH₃OH extracts showed a dose-response effect. T. recurvata at 500 μg mL^-1 inhibited the production of violacein by 85 %, whereas T. schiedeana and T. fasciculata inhibited only 58 % to 65 % (Figure 2B).

Quorum sensing inhibition in Pseudomonas aeruginosa. Extracts exhibited a moderate effect on inhibition of QS-regulated virulence factors in this bacterium. Although none inhibited pyocyanin production, the CH₂Cl₂ extracts of the three species reduced alkaline protease activity from 20 to 33 %. T. recurvata and T. schiedeana also reduced elastase activity by about 23 % (Table 1). It should be mentioned that for the case of CH₃OH extracts that affected elastase inhibition, the effect was not statistically significant (Table 1).

Inhibition of biofilm formation and swarming in Pseudomonas aeruginosa. CH₂Cl₂ extracts of the three species inhibited biofilm formation (32 to 51 %), although this effect was less than that exhibited by the positive control, furanone C-30 (50 μM), which reduced it by 79 % (Figure 3A). On the other hand, the CH₃OH extracts had an opposite effect; they stimulated biofilm formation. T. schiedeana (500 μg mL^-1) promoted an increase of 37 % (Figure 3B).

Regarding bacterial motility, the CH₂Cl₂ extracts of the three species inhibited P. aeruginosa swarming. T. recurvata showed the best effect by reducing motility by 50 % at 500 μg mL^-1, similar to that shown by furanone C-30 used as the positive control. T. schiedeana reduced motility by 28 and 39 %, and T. fasciculata did so by 7 and 43 % at the concentrations of 250 and 500 μg mL^-1, respectively (Figure 4). There was no effect on inhibition of motility by the CH₃OH extracts.

Preliminary phytochemical analysis indicated that dichloromethane extracts of the three species contain mainly terpenoid compounds and in a smaller proportion of phenolic type metabolites. In the case of methanol extracts, they mainly contain compounds of a phenolic nature, such as flavonoids and to lesser extent terpenoids. Interestingly, none of the samples contained tannins (Table 2. Figure S3). For the chemical identification of the constituents that conform the extracts, separation was carried out using conventional chromatography techniques and identification by ¹H NMR (Table 3).

For the compounds and fractions of T. recurvata in Tr-D4, at δ_H 7.6 and 6.5, signals of aromatic protons were identified and attributed to the flavonoid backbone. Similarly, signals between δ_H 12-13 indicate the presence of three distinct phenolic protons at C-5, chelated with the carbonyls at C-4 in the flavonoid framework (Table 3. Figure S4) (^{VH et al. 2014}). Compounds of this nature such as 5,3'dihydroxy-6,7,8,4'-tetramethoxyflavanone have been isolated from T. recurvata (^{de Queiroga et al. 2004}). In Tr-MeOH3, a pentacyclic type-triterpene was identified as the principal constituent of the fraction, which is supported by signals δ_H 5.36, 3.9 and 0.68, 0.82 that are characteristic of this class of molecules (Table 3. Figure S5) (^{Woo et al. 2015}). In the Tr-MeOH6 fraction, it was identified as a mixture of methoxylated flavonoids, which is supported by the signals of aromatic protons (δ_H 7.7 to 6.5), that of a C-5 phenolic proton (δ_H 12.19) and that a methoxy group (δ_H 3.98) (Table 3. Figure S6).

For T. schiedeana, Ts-D13-15 was identified as a mixture of fatty acids that is supported by the δ_H 1.26 signal (Figure S7). Ts-MeOH31 was identified a compound related to β-sitosterol, which supported by signals attributable to sterols such as δ_H 2.3 and 3.9 (Table 3. Figure S8) (^{Bulama et al. 2015}). While in Ts-MeOH11 the characteristic signals of β-sitosterol (δ_H 2.2, 3.5, 5.3) are observed (^{BMRB 2019}), indicating that in this mixture there is a majority presence of this sterol or a compound with a very similar structure (Table 3. Figure S9). In Ts-MeOH57-61, signals at δ_H 4.5 and 3.2 among others were identified (Table 3. Figure S10), as characteristic of sugar residue protons (^{Primahana & Darmawan 2017}). This indicates the presence of glycosides in both flavonoids (δ_H 7.0-7.5), as well as triterpenes (δ_H 2.5-1.0) (^{Barbosa et al. 2010}, ^{Ibrahim et al. 2019}). Finally, in Ts-MeOH9 the presence of two doublets at δ_H 0.33 (J = 4.2 Hz) and at δ_H 0.55 (J = 4.2 Hz), is indicative of cyclopropane protons in the cycloartane framework (Table 3. Figure S11) (^{Maatooq et al. 2002}). This class of triterpenes has been previously isolated from T. recurvata (^{Cabrera & Seldes 1995}). Likewise, signals are observed in the aromatic region, indicating that flavonoids are also present in this fraction (^{Primahana & Darmawan 2017}).

Discussion

Natural products are the main source of antimicrobial compounds. Specifically, those produced by microorganisms have been the main suppliers of the antibiotics that are currently used in the clinic (^{Bérdy 2012}). However, the constant emergence of resistance has decreased effectiveness of antibiotics and has compromised their use against antibiotic-resistant strains (^{López-Jácome et al. 2019}). Faced with this situation, anti-virulence therapies have been proposed for treatment of bacterial infections, without inducing the appearance of resistance since they do not directly affect viability of microorganisms (^{Castillo-Juárez et al. 2017}). This class of molecules acts by inhibiting the production of virulence factors that participate in the establishment and generation of damage in infectious processes (^{Muñoz-Cazares et al. 2017}). However, although its use is feasible, up to now the low efficiency of the molecules in in vivo assays and the lack of clinical studies have limited its application (^{García-Contreras 2016}). Moreover, canonical quorum quenching compounds such as furanone C-30 and 5-fluorouracil are highly toxic and are not effective against several clinical strains (^{García-Contreras et al. 2015}, ^{Guendouze et al. 2017}).

Due to the growing problem of resistance to antibiotics, plants have become an important source in the search for metabolites with mechanisms of action that are not bactericidal (^{Silva et al. 2016}). In this respect, the greatest number of molecules capable of inhibiting virulence, mainly QS and biofilm formation, have been identified in plant species (^{Ta & Arnason 2015}, ^{Silva et al. 2016}). It should be noted that they are not new chemical structures, and they can be found in different genera of plants distributed in different parts of the world. Some reports have focused on investigating the anti-virulence properties of local species or of traditional remedies, but in Mexico studies are still scarce. In this work, we focused on investigating the anti-virulence properties of three species of Tillandsia distributed in Mexico as potential source of this class of molecules.

Chromobacterium violaceum is one of the main biosensors used to evaluate QS inhibition because it produces a violet pigment which facilitates visualizing the effect on the QS inhibition, which is in relation to the amount of pigment inhibited (^{McClean et al. 1997}). Although generally Tillandsia extracts do not completely prevent violacein production, this bioassay suggests that these species contain metabolites that potentially inhibit QS. This effect is similar to other reported extracts, which do not completely inhibit pigment production and potentiate their activity only with purified active compounds (^{Castillo-Juárez et al. 2013}). On the other hand, the only study that has been carried out on the anti-virulence potential of these species was conducted with ethanolic and aqueous extracts of T. recurvata, which did not reduce violacein production (^{Adonizio et al. 2006}). Because we identify activity in extracts of lower polarity, the active compounds may not be extracted in large quantities with ethanol or water.

Pseudomonas aeruginosa QS systems have also been extensively investigated. This microorganism contains at least three identified systems that are regulated hierarchically to produce several virulence factors, such as pigments (pyocyanin and pyoverdine), enzymes (protease and elastase), swarming and biofilms (^{Lee & Zhang 2015}). It has been reported that the aqueous extract of Ananas comosus (Bromeliaceae) inhibits proteolytic and elastolytic activity, as well as biofilm formation and, partially, pyocyanin production (^{Musthafa et al. 2010}). However, the extracts of the three species of Tillandsia evaluated had no effect on pyocyanin production and only the CH₂Cl₂ extract slightly reduced protease and elastase activity. Similarly, swarming and biofilm formation decreased with less polar extracts. Interestingly, when the extract is more polar, as in the case of CH₃OH, the inhibitory activity on swarming is lost and biofilm formation is stimulated. In this regard, it has been reported that different molecules with capacity to inhibit or stimulate QS can coexist in a plant extract.

In addition, it has been reported that certain phenolic compounds inhibit virulence factors, while others stimulate their production, depending on their concentration (^{Plyuta et al. 2013}). Similarly, in the case of some essential oils, structural isomerism is also important for the oil to stimulate or inhibit the violacein and pyocyanin production (^{Ahmad et al. 2015}). ^{Varposhti et al. (2013)} indicates that Dianthus orientalis and Origanun majorana extracts also increase biofilm production depending on the type of extraction. On the other hand, stimulation of biofilm cannot be considered a negative effect because it has been shown that its stimulation by certain extracts or metabolites induce defective formation, making it more sensitive to destruction by antimicrobial agents.

Chemical composition of the Bromeliaceae family is characterized by the presence of abundant terpenes, flavonoids and phenols (^{Manetti et al. 2009}). However, of the species studied, only T. recurvata and T. fasciculata have been reported to contain some compounds (^{Estrella-Parra et al. 2019}). In this study, we corroborate the presence of terpenes, flavonoids and phenols in the three Tillandsia species and in the case of T. schiedeana, it is the first report describing the kinds of specialized metabolites derived from the species. However, that it was not possible to obtain pure compounds, the analysis of the fractions allowed us to identify various chemical groups such as glycosylated compounds (flavonoids and cycloartane-type triterpenes). Flavonoids isolated from plant species are the main group of molecules that have been identified as having anti-virulence properties on different bacterial species (^{Silva et al. 2016}). However, of those reported with this activity for the species analyzed in this study, only quercetin has been identified in T. fasciculata (^{Williams 1978}). On the other hand, the low polarity extract of Tillandsia has an inhibiting effect on QS-regulated virulence factors of P. aeruginosa and represent an important source of anti-virulence molecules to be explored.

Supplemental data

Supplemental material for this article can be accessed here: https://doi.org/10.17129/botsci.2380