Tannins Possessing Bacteriostatic Effect Impair Pseudomonas aeruginosa Adhesion and Biofilm Formation

Plants produce many compounds that are biologically active, either as part of their normal program of growth and development or in response to pathogen attack or stress. Traditionally, Anadenanthera colubrina, Commiphora leptophloeos and Myracrodruon urundeuva have been used by communities in the Brazilian Caatinga to treat several infectious diseases. The ability to impair bacterial adhesion represents an ideal strategy to combat bacterial pathogenesis, because of its importance in the early stages of the infectious process; thus, the search for anti-adherent compounds in plants is a very promising alternative. This study investigated the ability of stem-bark extracts from these three species to control the growth and prevent biofilm formation of Pseudomonas aeruginosa, an important opportunistic pathogen that adheres to surfaces and forms protective biofilms. A kinetic study (0–72 h) demonstrated that the growth of extract-treated bacteria was inhibited up to 9 h after incubation, suggesting a bacteriostatic activity. Transmission electron microscopy and fluorescence microscopy showed both viable and nonviable cells, indicating bacterial membrane damage; crystal violet assay and scanning electron microscopy demonstrated that treatment strongly inhibited biofilm formation during 6 and 24 h and that matrix production remained impaired even after growth was restored, at 24 and 48 h of incubation. Herein, we propose that the identified (condensed and hydrolyzable) tannins are able to inhibit biofilm formation via bacteriostatic properties, damaging the bacterial membrane and hindering matrix production. Our findings demonstrate the importance of this abundant class of Natural Products in higher plants against one of the most challenging issues in the hospital setting: biofilm resilience.


Introduction
Adhesion and colonization are prerequisites for the establishment of bacterial infection and pathogenesis. Once adhesion has taken place, on implanted medical devices or damaged tissue, microorganisms may undergo specific molecular changes to become pathogenic and to establish biofilms [1]. It is well known that biofilm formation involves the attachment and accumulation of microbial cells, within a self-produced extracellular matrix, on a solid surface [2]. The inherently defensive character of the biofilm is demonstrated by enhanced persistence of bacteria grown in the sessile mode model versus bacteria grown planktonically, which makes most biofilm-associated infections difficult to eradicate, thus contributing to disease chronicity [3,4]. Pseudomonas aeruginosa, a ubiquitous bacterium in the nature, is an opportunistic pathogen that adheres to surfaces and forms protective biofilms [3]. In addition, multidrug-resistant P. aeruginosa is a leading cause of nosocomial infection worldwide, ranking first among all nosocomial pathogens related to pneumonia in intensive care units in Brazil [5] and in the United States [6].
The challenge and difficulty in finding novel antibacterial agents with innovative mechanisms of action, including anti-adherent compounds, drive the search for antimicrobials toward vegetable sources. This is an appropriate choice because plants play an important role in the biosynthesis of natural products, providing chemical defense against environmental microbes through secondary metabolism, and because they can be considered as a therapeutic alternative in primary health care (ethnopharmacological knowledge). The Caatinga, a xeric shrub-dominated biome of northeastern Brazil, supports a high diversity of plant resources used as folk medicine. This region is known as an area of low economic development, which reflects poor access of the population to pharmaceutical drugs and, consequently, determines the treatment of illnesses based on the use of medicinal plants. The limited scientific basis for the biological properties of these plants prompted an interest in investigating species widespread in the Caatinga in more detail. Our ongoing efforts to evaluate their biological potential have revealed antibacterial and antibiofilm activities against the Gram-positive bacterium Staphylococcus epidermidis for several plant species [7] and against P. aeruginosa for a reduced number of plants (unpublished data).
In this context, tannins have attracted attention due to their biological and physiological properties [8]. The tannins are complex mixtures of oligomers widely distributed in plants and foods which are yielded by polymerization of flavan-3-ol units (condensed tannins) or by esterification of several gallic acid residues to a carbohydrate core, mainly glucose (hydrolysable tannins) [9,10]. The polymerization degree and oxygenation pattern of tannins are related with their biological properties [9,11], making relevant their structural identification. In several cases, the isolation of tannins is very intricate, therefore mixtures are being successfully analyzed by mass spectrometry (MS), such as Matrix-Assisted Laser Desorption/Ionization (MALDI) [12,13]. The couplings between different flavan-3-ol units have been proposed with the MS data and there are few studies based in MS/MS data to identify tannins, even though the structural identification seems to be more reliable using the fragmentation pathway [14,15].
This study aimed to investigate the activity of stem-bark extracts of Anadenanthera colubrina, Commiphora leptophloeos and Myracrodruon urundeuva and their ability to control the growth and prevent biofilm formation of P. aeruginosa using bioguided fractionation. The bioactive compounds were purified and further analyzed by MALDI-MS/MS in order to identify the structures.

Plant material
Stem barks were collected at a national park, Parque Nacional do Catimbau (PARNA do Catimbau), located in the state of Pernambuco, northeastern Brazil, between July and August 2009, under authorization of the responsible authority Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio) using the license SISBIO 16.806. The taxonomic identification was confirmed at the herbarium of Instituto Agronômico de Pernambuco (IPA), where the vouchers were deposited ( Table 1). As reported by official authorities in Brazil the species used in this study are not endangered or protected in the state of Pernambuco, where the sampling occurred. The extracts were prepared as previously described [7].

Bacterial strain and culture conditions
Pseudomonas aeruginosa ATCC 27853 was grown overnight on Mueller-Hinton (MH) agar (Oxoid Ltd., England, UK) at 37uC. A bacterial suspension of 3610 8 colony-forming units (CFU)/mL in 0.9% NaCl was used in the assays.

Minimum inhibitory concentration (MIC) and bacterial viability
Bacterial growth was assessed as the difference between optical density at 600 nm (OD 600 ) at the end (6 h) and at the beginning (0 h) of incubation time, in 96-well microtiter plates (Costar 3599, Corning, Inc., USA). In each well, 80 mL of the bacterial suspension, 80 mL of the aqueous extract (concentration ranging from 0.5 to 4.0 mg/mL in the wells) and 40 mL of tryptone soya broth (TSB) (Oxoid Ltd., England, UK) were added. MIC was defined as the lowest concentration of samples able to restrict bacterial growth to a level lower than 0.04 at OD 600 . Serial dilutions of the MIC wells were performed and spread on MH agar plates. After incubation (37uC, overnight), the CFU/mL was obtained, in order to determine the viability of bacterial cells. As references for bacterial growth and viability, the extracts were replaced with water (negative control) or with gentamicin sulfate (Sigma-Aldrich Co., USA) (positive control).

Antibacterial activity kinetics
A kinetic study was performed to assess the effect of extracts (at concentrations of 1/4xMIC, 1/2xMIC, MIC, and 2xMIC) upon P. aeruginosa according to the incubation time, as previously described in Section 2.3. OD 600 was measured at 0, 3,6,9,24,30,48,52, and 72 h after incubation (37uC). Samples were replaced with sterile water as a control for bacterial growth. To avoid the interference of sample color in all results obtained by OD evaluations, the samples were incubated in TSB and sterile saline (without inoculum) and the arithmetic means of OD readings were corrected for each extract (by subtracting OD without inoculum from OD with inoculum for each incubation time). Erythromycin (Sigma-Aldrich Co., USA) was used as control for bacteriostatic action. The results are expressed as mean 6 standard deviation (SD) of 4 wells for each extract concentration and for each incubation time.

Biofilm formation assay
Biofilm formation was evaluated using the crystal violet assay in 96-well microtiter plates [7]. The incubation period at 37uC was 6, 24, and 48 h. To represent 100% of biofilm formation (untreated sample), the extracts were replaced with sterile water. Values higher than 100% represented stimulation of biofilm formation in comparison with the untreated sample. Since does not exist a commercially available non-biocidal compound possessing antibiofilm activity, we can not apply a positive control to antibiofilm activity.

Microscopic analysis
Scanning electron microscopy (SEM). Pseudomonas aeruginosa biofilms were grown in 96-well microtiter plates (37uC during 6, 24, and 48 h) with a piece of Permanox TM slide (Nalge Nunc International, USA), as described in Section 2.3. The samples were prepared and examined according to Trentin et al [7].
Transmission electron microscopy (TEM). Tubes containing 800 mL of the bacterial suspension, 800 mL of the aqueous extract (MIC concentration) and 400 mL of TSB were incubated (37uC, 6 h). In untreated samples, sterile water was added instead of samples. The bacteria were harvested by centrifugation, fixed in glutaraldehyde and paraformaldehyde solution, and, subsequently, in 2% osmium tetroxide. The pellet was dehydrated in an ascending series of acetone concentrations and cells were embedded in acetone:EmBed TM resin, homogenized by rotation and polymerized. Ultrathin sections were contrasted with uranyl acetate and lead citrate and imaged with a JEOL JEM-1200 EX II electron microscope (JEOL Ltd, Tokyo, Japan).
Fluorescence microscopy. Samples were prepared as described in Section 2.6.2. After incubation, they were stained with LIVE/DEAD BacLight Bacterial Viability Kit (Life Technologies, USA). In this assay, the SYTO-9 and propidium iodide (PI) stains compete for binding to the bacterial nucleic acid. SYTO-9 labels cells with both damaged and intact membranes (green cells), whereas PI penetrates only in the cells with damaged membranes (red cells), reducing the fluorescence of SYTO-9. The samples were observed in the AxioVert 200 fluorescence microscope using the AxioVision AC software (Carl Zeiss MicroImaging Inc, Germany), and image overlays were obtained using ImageJ software.

Hemolytic assay
This assay was performed as previously described [16], using human venous blood. All the blood donors were healthy researchers and students who signed specific form for consent to participate in the study. The Universidade Federal do Rio Grande do Sul Ethical Committee approved all documents, procedures and project under authorization number 19346. All forms are stored in specific place in the laboratory. The extracts were tested at MIC and, as reference samples, we used water (for baseline values) and Quillaja saponaria saponins (Sigma-Aldrich Co., USA) at 0.25 mg/mL (for 100% hemolysis). To avoid the interference of sample color, a blank sample of extracts and phosphate-buffered saline (PBS) (without erythrocytes) was developed. The assay was calculated as follows: (Abs extracts -Abs blank=Abs saponin -Abs water)|100:

MALDI-MS and MALDI-MS/MS analyses
High-resolution mass spectrometry (MS) analyses were performed using an UltrafleXtreme MALDI-TOF/TOF equipment (Bruker Daltonics, Bremen, Germany). A mixture of peptides was used for external and internal calibration (peptide calibration standard II [Bruker Daltonics]: bradykinin 1-7, angiotensin II and I, substance P, bombesin, renin substrate, ACTH clip 1-17, ACTH clip 18-39, and somatostatin 28). The ions were generated by irradiation with a nitrogen laser (337 nm) and accelerated at 20 kV. For MS analyses, the experimental conditions were: pulsed ion extraction of 100 ns, laser frequency of 1000 Hz, reflectron mode, positive ion mode, and 600 laser shots were averaged to record a mass spectrum. In addition, the selected ions were accelerated to 19 kV in the LIFT cell for MS/MS analyses. The matrix of choice was DHB (2,5-dihydroxybenzoic acid) at 20 mg/ mL (in 30% acetonitrile [ACN] and 70% H 2 O with 0.1% trifluoroacetic acid). All samples were suspended in ACN:H 2 O (3:7) and mixed with DHB containing 0.1 M solution of NaCl. These mixtures (1 mL) were spotted onto a ground stainless steel MALDI target. The compounds were identified by MS data, fragmentation pathway and accurate mass measurements using the internal calibrant (standard peptide mixture).

Statistical analysis
Biological assays were carried out in triplicate. Data differences in relation to the untreated samples were analyzed by the Student t test, and p#0.05 was considered to be significant.

MIC and viability determinations for stem-bark extracts
The MIC of all three stem-bark aqueous extracts against P. aeruginosa was determined (Table 1). CFU counting was used to determine the viability of 6 h-treated cells at MIC. Statistical analysis indicated that previously treated and untreated bacterial suspensions were equivalent regarding CFU/mL, suggesting bacteriostatic activity, i.e., when bacterial growth is inhibited (at MIC), the cells present viability (Table 1).

Kinetic analysis of antibacterial activity
To observe the effect of extracts upon P. aeruginosa growth according to incubation time, a kinetic study was performed (Fig. 1). In this set of experiments, there was a significant decrease in bacterial growth within a short period of exposure to all extract concentrations (except for 1/4xMIC of A. colubrina and C. leptophloeos - Fig. 1A-B). Bacterial growth remained inhibited or low up to 9 h after incubation. After 24 h, extract-treated bacteria started to grow, achieving values similar to those obtained with untreated cells. Reinforcing our former result, a similar dynamic profile was observed with erythromycin, a bacteriostatic agent against P. aeruginosa.

Activity upon biofilm formation
The effect of extracts at MIC upon P. aeruginosa biofilm formation on the polystyrene surface, using crystal violet assay, is shown in Table 1. Biofilm formation was strongly prevented in 6 h-treated bacteria in relation to untreated cells. After 24 h of treatment, the extracts remained able to inhibit biofilm formation, although with a loss in their absolute antibiofilm activity by 20% when compared to inhibition at an earlier stage (6 h). According to this assay, no antibiofilm effect was observed at 48 h of incubation. Unlikely, C. leptophloeos and M. urundeuva extracts stimulated biofilm formation (Table 1). These results were corroborated by morphological features observed using different microscopic techniques. SEM images showed rod-shaped cells of untreated P. aeruginosa grown on the Permanox surface, forming a dense and uniform biofilm (untreated biofilms) (Fig. 2 images 1A-C), covered with an extracellular matrix (Fig. 2 images 1B and 1C). In contrast, 6 htreated biofilms displayed lesser adherent bacteria and only slight aggregation, reducing bacterial agglomerates to small clusters ( Fig. 2 images 2A, 3A, and 4A). At 24 h of incubation, we could still observe a low number of bacterial clusters deficient in matrix production compared to the untreated sample ( Fig. 2 images 2B and 4B), while for C. leptophloeos-treated cells a larger amount of aggregated cells was observed, but without the presence of a matrix ( Fig. 2 image 3B). After 48 h of treatment, matrix production was extremely poor; the bacterial cluster architecture varied among samples and was different from the typical biofilm shape of the untreated P. aeruginosa. However, it was possible to observe that cell agglomerates had completely covered the surface ( Fig. 2 images 2C, 3C, and 4C).

Transmission electron microscopy (TEM) and fluorescence microscopy (FM)
On TEM micrographs, untreated cells exhibited an undisturbed cytosol and intact cell envelope (cytoplasmic membrane and cell wall) (Fig. 3A). In contrast, extracts at MIC induced ultrastructural modifications in P. aeruginosa cells (Fig. 3B-D). At 150006 magnification, images revealed that all three extracts were able to promote intense vacuolization in several cells (as signaled by black arrows with white outline), although some cells with normal morphology remained present (white arrows with black outline). Regarding A. colubrina and M. urundeuva, cell deformation and disrupted cell wall could also be observed ( Fig. 3B and D, in the inserts). In C. leptophloeos-treated cells, in addition to an injured cell wall, we could observe vacuoles dispersed throughout the cytoplasm and within the periplasm of cells (Fig. 3C, in the inserts). FM images reinforced these results: at MIC, both viable

Hemolytic activity
We carried out a simple model to assess injury in human cells and preliminary toxicity of extracts. At MIC, A. colubrina and M. urundeuva caused 4.361.9% and 13.160.5% of hemolysis, respectively while C. leptophloeos was not hemolytic (061.2%). The microscopic analysis showed erythrocyte integrity and absence of erythrocyte aggregation (data not shown).

Bioguided fractionation
The aqueous extracts were fractionated on Sephadex LH-20. The eluted fractions (F1-F9) were tested for P. aeruginosa antibiofilm and antibacterial activities at three concentrations: MIC, 1/2xMIC and 1/4xMIC of the crude extracts, at 6 h of incubation. For all plants, the F7 fraction provided the same activity observed for the crude extracts (Fig. 5). The fraction obtained from A. colubrina allowed a biofilm formation of 14, 20, and 32% at 2.5, 1.25, and 0.625 mg/mL, respectively, with growth inhibition in all tested concentrations (Fig. 5A). The same profile was observed for F7 obtained from M. urundeuva at all concentrations tested, in which biofilm formation was limited to 23%, accompanied by absence of growth (Fig. 5C). Regarding F7 from C. leptophloeos, at 1.0 mg/mL, bacterial growth was suppressed, with biofilm formation of about 35%. In C. leptophloeos, when fraction concentration decreased to 0.25 mg/mL, P. aeruginosa began to grow and biofilm inhibition decreased (54% of biofilm formation) (Fig. 5B). The number of CFU/mL after F7 treatments was determined and bacteriostatic effects were observed in all concentrations tested for the three plants evaluated (Fig. 5A-C). Figure 6 shows the MALDI-TOF spectra of the polymeric tannin mixture in the F7 fractions from A. colubrina, C. leptophloeos, and M. urundeuva. The molecular formulae of tannins were obtained from measured accurate masses ( Fig. S3 and Fig. S15), and the MS/MS data were important to establish and confirm the units of flavan-3-ol or galloyl bound in oligomers. All ions and fragment ions were cationized by sodium. Proanthocyanidins (or condensed tannins) were identified from A. colubrina fraction yielding units mainly of profisetinidin type (Fig. 7) up to 10 units (Table 2 -series B), and from C. leptophloeos fraction, in which the units are mainly of prorobinetinidin type (Fig. 7) reaching up to 13 units (Table 2 -  S5) represented the loss of 152 u due to Retro Diels-Alder (RDA) fission (C-ring fragmentation pathway) and confirmed the B-ring substituents of the proposed units (procyanidin and profisetinidin). This loss was also observed for the other ions of series B (Fig. S10,  Fig. S11 and Fig. S13). Regarding the minority series (A, C and D), is possible to observe that series A is comprised of oligomers formed only by profisetinidin units (Table 2 and Fig. S1, Fig. S4, Fig. S7 and Fig. S9). Differently of series B, the series C presents, additionally, one procyanidin residue (288 u) in the structure of its oligomers ( Table 2 and Fig. S12 and Fig. S14). The series D is composed of up to four units in the following sequence: profisetinidin, procyanidin, prorobinetinidin and profisetinidin (Table 2, Fig. S6 and Fig. S8), where the loss of 168 u due to RDA fission from ion at m/z 873 confirmed the B-ring substituents of prorobinetinidin unit (Fig. S6). The polymeric series of A. colubrina showed differences in 16 u, such as m/z 841, 857 and 873, which represents oxygen (hydroxyl) due the mass difference between 288 u monomers (procyanidin and/or prorobinetinidin) and 272 u    monomers (prodelfinidin); additionally, all compounds presented B type linkage (Fig. 7A).

MALDI study of tannin fractions
In the bioactive fraction (F7) from C. leptophloeos three polymeric series (A, B and C) were found, being rich in prorobinetinidin units as demonstrated by consecutive losses of 288 u (Table 2, Fig. 6B and Fig. S2). The main polymeric series was yielded with exclusively prorobinetinidin units ( Fig. 6B and Table 2 -series A) exhibiting B-type linkage pattern (Fig. 7A) and their compounds were determined through MS/MS data (Fig. S17, Fig. S19, Fig.  S22, Fig. S23). The minority series B differs from the others since A-type linkage is found. The starter unit in this series is a profisetinidin linked to repeat units of prorobinetinidin. The peak m/z 1159 yielded the fragments m/z 1007, 989, 871, 719, 581, 429, and 311 in the MS/MS spectrum (Fig. S18), similarly to what was observed for the fragmentation pathway of the ion m/z 871 (Fig. S16) and for the other peaks from this series (Fig. S21). The compound with m/z 871 was produced by coupling one profisetinidin and two prorobinetidins, the second unit with an A-type linkage (Fig. 7B). Thus, the compounds of polymeric series B present a difference of 2 u compared with similar tannin showing B-type linkage. Series C was formed by repeat units of prorobinetinidin and only one prodelphinidin, which was confirmed by MS/MS data, where the loss of 168 u due to RDA fission confirmed the prodelphinidin unit (Fig. S20).
MS/MS analysis of the bioactive fractions from A. colubrina (Fig.  S4-S14) and from C. leptophloeos (Fig. S16-S23) showed the same loss patterns for structural characterization of condensed tannins, such as the losses due to RDA fission and flavan-3-ol units; therefore only the classical fragmentations are addressed in this work.
The bioactive fraction (F7) from M. urundeuva was also analyzed by MALDI-TOF, and hydrolyzable tannins were observed in its composition. The MS spectrum showed peaks with increments of 152 u (Fig. 6C) and MS/MS spectra showed consecutive losses of gallic acid units attached to glucose (Fig. S24) confirming the presence of gallotannins in this fraction, as observed in a recent study [17]. However, the structural elucidation of gallotannin series was not possible, since the MS/MS data were inconclusive.

Discussion
The ability to impair bacterial adhesion represents an ideal strategy to combat bacterial pathogenesis, given its importance in the early stages of the infectious process. In addition, bacterial adhesion blockade is suitable as a prophylactic intervention to prevent infection [18]. Therefore, the use of natural agents that can successfully inhibit cell attachment is a promising tool for reduction of bacterial colonization on several surfaces [19]. In recent years, studies have reported anti-adhesive or antibiofilm activities of compounds, which are related to their antimicrobial properties [20][21][22][23][24].  Plants produce many compounds that are biologically active, either as part of their normal program of growth and development or in response to pathogen attack or stress. Traditionally, A. colubrina, C. leptophloeos and M. urundeuva have been used by communities in the Caatinga to treat infectious diseases, such as cough, bronchitis, influenza, urinary/liver diseases, ulcerative external lesions, and ovarian inflammation [25,26].
In this study, we demonstrated that a highly complex mixture (Fig. 6, Fig. 7 and Table 2) of proanthocyanidins (mostly composed of profisetinidin in A. colubrina and prorobinetinidin in C. leptophloeos) and hydrolyzable tannins (consisting of gallic acid units in M. urundeuva) induced P. aeruginosa damage, providing bacteriostatic and anti-adhesive effects. Bacterial growth kinetic experiments revealed that inhibition of bacterial growth persisted up to 9 h post-incubation (Fig. 1). The bacteriostatic property was confirmed by cell counting (CFU/mL), after both aqueous extract and the tannin-enriched fraction F7 treatments (Table 1 and Fig. 5), and by recovering bacterial growth of extract-and erythromycin-treated cells, as indicated by OD 600 measurements in the kinetic experiment ( Fig. 1). At MIC, the extracts were also able to inhibit bacterial adhesion and prevent biofilm formation on the polystyrene surface for 6 and 24 h after incubation. However, this activity was lost after 48 h of incubation (Table 1). SEM was employed to improve understanding of the qualitative impact of the extracts upon the behavior of bacteria. We could observe that 6 h-treated cells presented morphological changes, such as stalked nubs (white arrows with black outline in Fig. 2), a phenotype that is also induced by amikacin and oxytetracycline [27]. This finding supports the hypothesis that morphological changes were due to the mechanisms of bacteria involved in protection against aggression from extracts. After 24 and 48 h of incubation (the latter characterizing the mature stage of biofilm development), untreated P. aeruginosa was enclosed by extracellular matrix while just the 48 h-treated cells were surrounded by a very discrete matrix (Fig. 2). Flemming and Wingender [28] demonstrated that matrix is essential for biofilm formation, which allows a lifestyle that is entirely different from the planktonic state, and concluded that there is no biofilm without a matrix. Based on these observations, we may suggest that P. aeruginosa agglomerates as visualized by SEM at 24 and 48 h of treatment could not be considered biofilm structures.
The level of bacterial membrane dysfunction could potentially result in cell death and may explain the rapid loss of viability observed in the kinetic experiments. Although TEM and FM showed no viable cells, these analyses also indicated the presence of cells without morphological alterations and with an intact membrane. These findings are consistent with Pankey and Sabath report [29], who highlighted that most agents characterized as bacteriostatic agents are able to kill some bacteria -often more than 90-99% of the inoculum, but this is not sufficient (.99.9%) to characterize them as bactericidal agents. Although in vitro bacteriostatic/bactericidal data may provide information on the potential action of antimicrobial agents, this is only one of many factors required to predict a favorable clinical outcome. Bacteriostatic agents have been effectively used in the treatment of endocarditis, meningitis, and osteomyelitis -indications that are often considered to require bactericidal activity [29]. Additionally, the bacterial membrane may be compromised during antimicrobial treatments, such as due to exposure to a bacteriostatic agent. The ultrastructural analysis of P. aeruginosa (Fig. 3) showed that extract-exposed cells presented excess vacuoles and a disrupted cell wall, when compared to untreated samples. FM results (Fig. 4) suggest that the extracts have anti-membrane activity, resulting in the disturbance of membrane structure in a large amount of cells, while displaying absence or very low toxicity against human erythrocytes, which corroborates the idea of a selective action of these tannins upon P. aeruginosa membrane.
The programmed death of some damaged cells may be beneficial to a multicellular bacterial community [30]. Thus, the occurrence of both viable and dying cells after extracts treatment could be understood as a suicide mission that would contribute to the maintenance of a population. There is strong evidence that genomic DNA released during P. aeruginosa lysis is a structural component of the biofilm matrix, supporting the idea that cell lysis contributes to the stability of the biofilm structure [31]. These findings are in agreement with our data, which show the development of P. aeruginosa clusters at 48 h of incubation, at the same time point when the potential of extracts to impair bacterial adhesion is decreased. The immune system is capable of eliminating pathogens that would otherwise persist in the presence of bacteriostatic agents, although the elimination of persister cells from biofilms by the immune system has not been specifically studied yet [30].
As a Gram-negative bacterium, P. aeruginosa has a cell wall consisting of a peptidoglycan layer and an additional outer membrane [32]. It should be noted that, in order to reach the cell membrane, tannins must cross the bacterial cell wall. Scalbert [33] has suggested that the cell wall probably fixes part of tannins, contributing to increase their MIC values. This observation is in agreement with our results, since MIC values for the extracts ranged from 1.0 to 4.0 mg/mL (Table 1) and for purified tannins (F7) the value was 1.0 mg/mL for C. leptophloeos and at least 0.625 mg/mL and 1.0 mg/mL for A. colubrina and M. urundeuva, respectively (Fig. 5).
This study represents an unprecedented report on phytochemical analysis, identifying tannins from three important plants used as folk medicine in Brazil: A. colubrina, C. leptophloeos, and M. urundeuva. It is also the first work to elucidate the tannin structure of plants from the genera Anadenanthera and Commiphora. Considering the origin of the plant material (stem-barks), the extraction method (aqueous maceration) and the data about the tannin content of these species [34], the achievement of tannins by bioguided fractionation was as expected. It is worth mentioning that, as previously reported by Almeida et al [35], Caatinga plants are exposed to high solar radiation in a semiarid environment, which favors the synthesis of phenolic derivatives, reinforcing the medicinal potential of plants from this biome. In summary, we propose that tannins are able to inhibit biofilm formation by damaging the bacterial membrane and hindering matrix production, displaying bacteriostatic properties. Therefore, biofilm formation is prevented during minimal bacterial growth; when cells start to grow, they are able to attach to the surface and develop matrix-deficient cell clusters. The bacteriostatic activity against P. aeruginosa observed in this study provides a scientific basis which may justify some uses of these plants in traditional medicine. To the best of our knowledge, this is the first study to show the modulation of P. aeruginosa biofilm formation by the herein described bacteriostatic tannins and the first to report the identification of tannins from A. colubrina and C. leptophloeos. This fact highlights the importance of this widespread and abundant class of secondary metabolites in higher plants against one of the most challenging issues in the hospital setting: biofilm resilience.

Supporting Information
Table S1 Yield (mg/%; w/w) of each fraction obtained from Sephadex LH-20 column in relation to the powdered plant material. (TIF)