The antibiofilm activities of the tannins were determined using 96-well plates with a cover lid with pegs for growing the bacterial biofilms in presence of several concentration of the assayed tannins. These assays, divided in a 1st exploratory screening (Section “Antibiofilm screening”) and a 2nd validation experiment (Section “Validation anti-biofilm screening: Experimental design and statistical analysis”), are explained in detail in the following section.

To determine the effect of phenolic hydroxyl (OH) content on the antimicrobial and antibiofilm effect of the different tannins, a simple linear regression between the phenolic OH content of the tannins and the level of biofilm formation (or planktonic growth) was performed. To better calculate this correlation, we used the different tannin assayed concentrations and the obtained mmol of phenolic OH per gram of material to calculate the mmol of phenolic OH present in the system for each tannin at each assayed concentration, and we correlated this value with the respective percentage of biofilm inhibition and planktonic growth inhibition. The phenolic OH content of the tannins was determined via ³¹PNMR as described elsewhere (Zhen et al., 2021a).

A diverse range of five commercially obtained tannins, and five types of derivatives, were selected and screened for anti-biofilm activity. The condensed tannins were previously identified as mixtures of (epi) catechins and fisetinidols (Zhen et al., 2021a) and consisted of (i) the essentially monomeric Vv-20, (ii) the low oligomeric Vv and (iii) the higher oligomeric Am. The hydrolysable tannins consisted of two large tannins: (i) the “typical” tannic acid Ta-01 and (ii) the galloquinic acid derivative Ta-04. All tannins were previously functionalized with a positive charge introducing ammonium salt hydroxy-N,N,N-trimethylpropanyl-3-aminium chloride (C₃NMe3Cl-0.1 and C₃NMe3Cl-0.5), an acidifying hydroxypropyl-1-carboxylic acid (C₃COOH-0.1 and C₃COOH-0.1-0.5) motive and a polymerizing oligomeric ethylene glycol polyether (PEG₅₀₀-0.05; Figure 1). The preventive activity of both the natural and derivatized tannins against the biofilm formation and planktonic growth of the Gram-negative species S. Typhimurium, P. aeruginosa, E. coli and the Gram-positive species S. aureus was evaluated by means of the Calgary biofilm device. In a first set of exploratory experiments, two-fold serial dilutions (from 600 till 4.96 mg/L) of the tannins were evaluated in order to obtain a first glance on activity spectrum and active concentration range. Activities against all four bacterial species were observed, with BIC₅₀ values ranging from 4.69 to 545.8 mg/L and IC₅₀ values ranging from 37.5 to 459.2 mg/L (Supplementary Table S1). As such this experiment allowed to determine the test concentrations for future validation experiments: 9.38 mg/L was the lowest assayed concentration in the preliminary screening and the most active tannins exhibited BIC₅₀ equal to or below that value; 150 mg/L was the concentration at which almost all tannins with antibiofilm effect were active; and 79.69 mg/L is the average of those two concentrations. Such validation experiments were required because the exploratory experiments only had three independent repeats and this did not provide sufficient statistical power to distinguish the levels of activity of the different tannins and delineate the relationship between the chemistry and the antibiofilm level of the tannins. Furthermore, there was a statistically significant plate-to-plate variation between the controls of each plate (see Supplementary Figures S1, S2).

To allow a multivariate analysis considering tannin scaffold, derivatization and concentration as well as bacterial target species, the previous antibiofilm and antimicrobial experiments were repeated in one experiment with eight repeats per condition, but only focusing on the three tannin concentrations that could capture best the antibiofilm effect of the tannins: 9.38, 79.69, and 150 mg/L. In order to minimize previously observed effects of plate-to-plate variation, these experiments were designed in a randomized way, i.e., all the tannins were distributed through all the plates in a random fashion. This allowed to decrease the random error and the possibility of confounding factors, a necessary step for doing a complex statistical analysis that allows to link the different chemical characteristics. In what follows we will first focus on the unmodified tannins, after which we will elaborate on the effect of chemical derivatization.

To determine the percentual inhibition of biofilm formation and planktonic growth of the assayed tannins against the selected bacteria, the bacterial growth and biofilm development in absence of tannin was taken as reference. As can be seen in Supplementary Figure S3, the assayed bacteria exhibited differences in their biofilm formation and bacterial growth, Staphylococcus aureus being the bacterium that reached the highest planktonic population density and formed most biofilm. On the contrary, the assayed Gram-negative bacteria showed less biofilm and less planktonic growth in comparison with S. aureus.

In Figure 2 it can be seen that the commercially available natural tannins showed different antibiofilm activities against the four assayed bacterial species. All unmodified tannins can be considered to have “broad spectrum activity,” since all of them exhibited statistically significant antibiofilm activity against Gram-positive and Gram-negative bacteria at least at the concentration of 150 mg/L, according to an ANOVA test with Tukey post-hoc analysis. Also, most of the assayed tannins exhibited a concentration-dependent activity, depending on the assayed tannin and the tested bacteria.

However, there were clear differences in the degree of antibiofilm activity of each tannin. Starting from the condensed tannins, the monomeric Vv-20 exhibited significant antibiofilm activity against all the assayed bacteria at all the assayed concentrations in a clear dose-dependent way, a dose of 79.69 mg/L being sufficient for inhibiting biofilm formation more than 50% against the four assayed bacteria. Also, Vv-20 was the most effective tannin against S. Typhimurium, with more than 75% of biofilm inhibition at 79.69 and 150 mg/L, and against E. coli, with more than 80% biofilm inhibition at 79.69 mg/L. Regarding the low oligomeric Vv, the antibiofilm activity was preferential against P. aeruginosa and S. aureus, for which it exhibited potent dose-dependent antibiofilm activity. This compound was able to inhibit more than 50% of biofilm formation by S. aureus both at 79.69 and 150 mg/L, and is the unmodified tannin with highest effect against P. aeruginosa, displaying antibiofilm activity of more than 90% at 79.69 mg/L and more than 95% at 150 mg/L. On the contrary, Vv was less effective against S. Typhimurium and E. coli and was able to inhibit <30% of biofilm formation of both bacterial species, regardless of the assayed concentration. Contrary to the previous two unmodified condensed tannins, the high oligomeric Am had preferential antibiofilm activity against S. Typhimurium and S. aureus, but also had significant antibiofilm activity against P. aeruginosa. Moreover, Am is the unmodified tannin with the highest antibiofilm activity against S. aureus, with more than 85% of antibiofilm activity at 79.69 and 150 mg/L, also inhibiting more than 50% of biofilm formation of S. Typhimurium both at 79.69 and 150 mg/L, and inhibiting P. aeruginosa biofilm formation in more than 30% at 79.69 mg/L and more than 50% at 150 mg/L. On the contrary, Am had unnoticeable inhibitory activity against E. coli biofilms at any of the assayed concentrations.

With respect to the hydrolysable tannins, tannic acid, Ta-01 exhibited preferential activity against P. aeruginosa and S. Typhimurium. The biofilm inhibitory activity of Ta-01 ranged from 40% at 9.38 mg/L to 60% at 150 mg/L against S. Typhimurium, and from 15% at 9.38 mg/L to more than 90% at 150 mg/L against P. aeruginosa. On the contrary, the galloquinic acid derivative, Ta-04 exhibited preferential activity against S. aureus. While the antibiofilm activity of Ta-04 against S. Typhimurium and E. coli was dose dependent (reaching a maximum of 50% biofilm inhibition against both bacteria at a concentration of 150 mg/L), the antibiofilm activity of Ta-04 against S. aureus was not dose dependent, with more than 70% biofilm inhibition at the three assayed concentrations.

Importantly, the unmodified tannins were in general not found to have antibacterial activity against the planktonic bacteria, except for the low oligomeric condensed Vv against S. Typhimurium and the hydrolysable tannic acid Ta-01 against E. coli at the highest concentration of compound (Figure 2). In previous reports, it was found that hydrolysable tannins similar to galloylquinic acid, hence similar to Ta-04, exhibit broad spectrum antibiofilm activity (Widsten et al., 2014; Ekambaram et al., 2016; Puljula et al., 2020). Those same reports, however, also suggest that hydrolysable tannins have antibacterial effects against planktonic Gram positive and Gram bacteria, which was not observed in our experiment. On the other hand, tannic acid (Lee et al., 2013) and 1,2,3,4,6-penta-O-galloyl-β-d-glucopyranose (Lin et al., 2011), both hydrolysable tannins, have been reported to inhibit biofilm formation of S. aureus without inhibiting planktonic growth, which is consistent with the results observed for tannic acid Ta-01. In our assay, also galloylquinic acid Ta-04 showed such activity. This selective activity against biofilms offers opportunities for potential applications, such as the titanium-tannin composite coating for implants developed by Shukla and Bhathena (2015), that allows sustained release of the tannin.

The aim of the modifications was to partially functionalize via ether linkages the phenolic OH-groups of each tannin molecule and to add tannin-alien functionalities at various levels to test the possibility to modulate the native activity of tannins toward biofilms and planktonic bacteria.

Figure 3 shows the effect of derivatization on the antibiofilm activity. The differences in antibiofilm activity between the different modifications were assessed via the Tukey test for multiple comparisons, by comparing if there were differences in the maximum activity (i.e., the antibiofilm activity at the highest assayed concentration). If there were no differences in the maximum activity, the activity at lower concentrations was also evaluated, thus allowing to determine if a derivatization was able to obtain the same effect as the unmodified tannin, but at a lower concentration.

As a general conclusion, it could be established that most of the chemical derivatizations, but especially positive charge introducing C₃NMe₃Cl-0.1 reduce the antibiofilm activity of the tannins, while some of them can shift the activity spectrum toward preferential activity against Gram-positive or Gram-negative bacteria. More importantly, it must be noted that, in general, the chemical derivatizations increased the activity against planktonic bacteria and decreased the activity against bacterial biofilms of the unmodified tannins, thus decreasing the biofilm specificity of the assayed tannins (see Supplementary Figures S4–S7 for a direct comparison between growth inhibitory effect and biofilm inhibitory capability of the assayed tannins against each bacterial species). This effect will be discussed in more detail in Section “Biofilm selectivity of the antibiofilm effect of the assayed tannins.”

Two derivatizations generally shifted the antibiofilm spectrum toward the Gram-negative group of bacteria: polymerizing PEG₅₀₀-0.05 and positive charge-introducing C₃NMe₃Cl-0.5. Contrarily, derivatization with acidifying C₃COOH-0.1 and C₃COOH-0.5 in general decreased the antibiofilm activity against Gram-negative bacteria, while it retained activity against S. aureus for larger tannins Am, Ta-01 and Ta-04, and increased activity against S. aureus for monomeric and low oligomeric Vv-20 and Vv. The other derivatizations in general decreased the activity against S. aureus. However, this effect of derivatizations on the spectrum and potency was highly dependent on the specific tannin submitted to the derivatization.

In more detail, it can be seen that the blank reaction (both in water -Blank-W- and in dimethylformamide -Blank-D-) already modified the antibiofilm effect of the assayed tannins, a phenomenon that could be attributed to removal of impurities that affect the antibiofilm effect of the tannins. It can be seen that Blank-W conditions increased the maximum antibiofilm effect of Vv against Salmonella Typhimurium, but decreased the maximum antibiofilm effect of Vv-20 and Vv against P. aeruginosa and decreases the antibiofilm effect of Vv-20 against E. coli at 79.69 mg/L without affecting its maximum antibiofilm effect. It can also be seen that Blank-D conditions only affected the antibiofilm activity of Ta-04, by reducing its antibiofilm effect against Gram-negative bacteria.

Regarding actual substitutions, it can be seen that the effect of a derivatization with positive charge introducing ammonium groups, i.e., C₃NMe₃Cl, was different depending on the equivalents of chemical substitution.

On the one hand, C₃NMe₃Cl-0.5, the high equivalent derivatization, decreased the maximum antibiofilm effect of condensed tannins against S. aureus but did not affect the anti-staphylococcal effect of hydrolysable tannins. However, C₃NMe₃Cl-0.5 derivatization had a tannin-dependent effect against Gram-negative bacteria. It did not affect the antibiofilm effect of monomeric Vv-20 against any Gram-negative bacteria but increased the maximum antibiofilm effect of low oligomeric Vv against Salmonella Typhimurium while decreasing the maximum antibiofilm effect of Vv against P. aeruginosa. It increased the effect of high oligomeric Am against all the Gram-negative bacteria at 79.69 mg/L without significantly changing the maximum effect compared to the unmodified tannin. It increased the maximum effect of tannic acid Ta-01 against E. coli and Salmonella Typhimurium but decreased the effect against P. aeruginosa at 79.69 mg/L and it significantly decreased the maximum antibiofilm effect of galloquinic acid derivative Ta-04 against Salmonella Typhimurium. On the other hand, C₃NMe₃Cl-0.1, the low equivalent derivatization, reduced the maximum effect of all tannins (with the exception of the galloquinic acid derivative Ta-04, which was not affected) against Gram-negative bacteria without affecting the antibiofilm effect against S. aureus.

Contrary to C₃NMe₃Cl, there were no big differences between different levels of derivatizations with acidifying motif CH₃COOH, i.e., CH₃COOH-0.1 and CH₃COOH-0.5, in terms of their impact on the activities compared to the unmodified tannins. For S. aureus, neither derivatization level affected the antibiofilm effect exerted by larger tannins Am, Ta-01, and Ta-04, comparable to the C₃NMe₃Cl-0.1 derivatization, while CH₃COOH-0.1 increased the maximum effect of low oligomeric Vv. CH₃COOH-0.5 had a similar effect on essentially monomeric Vv-20. For the Gram-negative bacteria, both derivatizations equally decreased the maximum antibiofilm effect of condensed tannins Vv-20 and Am, as well as for hydrolysable tannin Ta-04 against S. Typhimurium, but only CH₃COOH-0.1 significantly decreased the maximum antibiofilm effect against S. Typhimurium of Vv. While both derivatization levels drastically decreased the maximum antibiofilm effect of condensed tannins against P. aeruginosa, only CH₃COOH-0.1 decreased the maximum antibiofilm effect of tannic acid Ta-01. Neither derivatization level significantly affected the antibiofilm effect of hydrolysable Ta-04. This effect was similar regarding E. coli because neither derivatization level affected the maximum antibiofilm effect of hydrolysable tannins, while CH₃COOH-0.1 decreased the maximum antibiofilm effect of all condensed tannins. CH₃COOH-0.5 only significantly decreased the maximum antibiofilm effect of Vv-20 against E. coli.

Regarding derivatization with polymerizing PEG₅₀₀, it can be seen that PEG₅₀₀-0.05 in general did not affect the antibiofilm effect of the assayed tannins. The only exceptions are the increase in the maximum antibiofilm effect of low oligomeric Vv against Salmonella Typhimurium and the decrease in the antibiofilm effect of monomeric Vv-20 against S. aureus at 79.69 mg/L without changing the maximum antibiofilm effect. It has to be taken in account that, due to technical issues, it was not possible to analyze the effect of polymerizing PEG₅₀₀-0.05 derivatization on Ta-04. Polymerizing PEG₅₀₀-0.05 derivatization proofed difficult to analyze in terms of loading for Ta-04 (Zhen et al., 2021b).

No literature data are yet available that would describe the effects of chemical modifications of tannins on their antibiofilm, or more generally antibacterial activity. With the aim of elucidating a hypothesis about the reason behind the effect of the derivatizations on our assayed tannins in comparison to the parent tannins and the blanks, we decided to look into other non-tannin organic compounds.

One of these examples targets the effect of the high equivalent derivatization C₃NMe₃Cl-0.5 in shifting the anti-biofilm spectrum toward Gram-negative bacteria, which may be associated with addition of positive charges in the form of ammonium groups to the tannin molecules. This allows for comparison of our results with data obtained by Dalcin et al. (2017), who discovered that nanoencapsulation of dihydromyricetin within the polycationic polymer Eudragit RS 100^® not only increased its antibiofilm activity against P. aeruginosa, but that the polymer itself had antibiofilm activity. This finding goes in accordance with a previous publication of Campanac et al. (2002) which states that cationic quaternary ammonium compounds were more effective against P. aeruginosa than S. aureus biofilm. Also, Gao et al. (2019) showed a decreased biofilm formation in E. coli and S. aureus using positively charged nanoaggregates based on zwitterionic pillar-[5] arene, requiring a 10 times lower concentration of nanoaggregate to decrease biofilm formation in E. coli compared to S. aureus. These results suggest that equipping the tannins with positive charge-introducing ammonium groups may give preferential action against Gram-negative bacteria. However, tannins should be derivatized with enough positive charge-introducing ammonium groups to obtain this shift toward inhibition of Gram-negative bacteria since the low equivalent derivatization C₃NMe₃Cl-0.1 appeared to lower the activity against Gram-negative bacteria.

Regarding the effect of acidifying CH₃COOH derivatizations, there is a precedent of the effect of several substitutions on the antibiofilm effect of anthraquinones against methicillin-resistant S. aureus (MRSA; Song et al., 2021). This study shows that a carboxyl group at position 2 of the anthraquinone molecule increases both the antibiofilm and the antimicrobial activity, which is partially in agreement with our data that shows that CH₃COOH derivatizations increase the antibiofilm activity of low oligomeric Vv-20 and essentially monomeric Vv without affecting the bacterial growth. Relatedly, Warraich et al. (2020) found that the acidic d-amino acids d-aspartic acid and d-glutamic acid were effective in dispersing and inhibiting biofilm formation in S. aureus, and they attributed this effect to the negative charges introduced by carboxyl groups of the molecules under growth conditions.

Finally, regarding polymerizing PEG derivatizations, there are several studies about the potential of PEG cross-linked hydrogels for wound healing because of their antimicrobial, pro-angiogenesis and pro-epithelization capabilities (Chen et al., 2019), but there is no indication regarding the effect of an introduction of a PEG-motif on the antibiofilm capability of an organic compound.

In the last years, there has been increasing research on non-lethal antimicrobial targets against several bacterial species, from virulence factors to biofilm formation, including inhibition of regulatory mechanisms such as quorum sensing and production of public goods (Defoirdt et al., 2013; Defoirdt, 2016; Maura et al., 2016; Totsika, 2016). The rationale behind this research is the assumption that if bacterial viability is not affected, the selective pressure will be lower and thus the risk for emergence of antimicrobial resistance will be lower too (Kalia et al., 2014; Dieltjens et al., 2020; Hemmati et al., 2020). However, there is still discussion about the effectiveness of this “resistant-proof” approach (García-Contreras et al., 2016).

The heatmap of Figure 4 shows the anti-planktonic activity of the tannins at the assayed concentrations. We defined that the antibiofilm activity of a particular tannin was biofilm specific in case the tannin did not exhibit significant anti-planktonic effect at that concentration (Dos Santos Goncalves et al., 2014; Vijayakumar and Thirunanasambandham, 2021). Most of the tannins with antibiofilm activity did not exhibit anti-planktonic effects, indicating that they are biofilm specific. This is particularly true for unmodified tannins, of which only Vv and Ta-01 exhibited anti-planktonic activity against Salmonella Typhimurium and E. coli, respectively. Also, unmodified tannins re-isolated from blank reactions, i.e., Blank-W or Blank-D, with the exception of Blank-D against Salmonella Typhimurium, did not reduce the planktonic growth of the assayed bacteria.

Regarding the effect of derivatizations on the biofilm specificity of the tannins, this was both derivatization and species dependent. Derivatization with positive charge introducing C₃NMe₃Cl-0.1 not only generally decreased the antibiofilm activity of the assayed tannins, but also increased the anti-planktonic activity specifically against Gram-negative bacteria. Antibiofilm specificity of tannins derivatized with C₃NMe₃Cl-0.5 was highly dependent on both the assayed bacteria and the derivatized tannin: the antibiofilm effect was non-specific when the derivatization was applied to Vv-20, but it was biofilm-specific for Vv, Am and Ta-01. Also, tannins derivatized with C₃NMe₃Cl-0.5 most strongly affected the growth of P. aeruginosa except for Am-C₃NMe₃Cl-0.5, which did not affect the growth of P. aeruginosa at any concentration. Only a derivatization of Ta-04, and here especially with C₃NMe₃Cl-0.5, led to a dose-dependent anti-planktonic effect, thus exhibiting non-specific antibiofilm activity at higher concentrations. With respect to a functionalization with the acidifying element CH₃COOH at various concentrations, it can be stated that these in this study did not significantly change the anti-planktonic behavior of the derivatized Vv, Am and Ta-01 against the assayed bacteria, but increased the antibacterial effect of Vv-20 and Ta-04 against planktonic bacteria. However, one notable exception are the tannins that were modified with crosslinking PEG₅₀₀-0.05, whose antibiofilm activity against Gram negative bacteria was highly correlated with the ability to inhibit the planktonic growth (Figure 4).

As a general summary, tannin derivatization did not affect biofilm specificity against S. aureus but affected the biofilm specificity against Gram-negative bacteria in a tannin-specific and derivatization specific manner. More importantly, there was no correlation in the assayed tannins between the degree of inhibition of planktonic growth and the degree of antibiofilm effect, a situation that goes in accordance with some previous reports (Janecki and Kolodziej, 2010; Payne et al., 2013; Hricovíniová et al., 2021).

One of the potential consequences of the chemical derivatization of the tannins are changes in the content of free phenolic hydroxyl groups present in the chemical structure of the tannins, since functionalization occurs at these hydroxyl groups. This is potentially important, because in previous literature it has been described that the biological activity of polyphenols could be mediated by their phenolic hydroxyl groups (Ultee et al., 2002; Ito et al., 2005; Ben Arfa et al., 2006; Veldhuizen et al., 2006; Alves et al., 2013; d’Avila Farias et al., 2014; Salar et al., 2017). Particularly regarding hydrolysable tannins, Taguri linked the degree of antibacterial activity to the presence of galloyl groups (Taguri et al., 2004). Because of this, we aimed to determine if there was a significant impact of the derivatizations on the phenolic hydroxyl (OH) content of the tannins, and if the phenolic OH content affected the antibiofilm.

In a first approach, we studied the impact of derivatizations on the phenolic OH content of the tannins. As can be seen in Table 1, derivatizations with C₃NMe₃Cl-0.1 decreased the phenolic OH content of the unmodified tannin. This is important, since derivatization with C₃NMe₃Cl-0.1 significantly decreased the antibiofilm effect of tannins, mostly against Gram negative bacteria. In a similar trend, derivatizations with CH₃COOH tended to decrease the phenolic content of condensed tannins: derivatization with CH₃COOH decreased the antibiofilm effect against Gram negative bacteria of condensed tannins, but, interestingly, not of still larger hydrolysable tannins.

In a second approach, we hence studied potential correlations between the phenolic OH content and the antibiofilm activity and we observed a weak correlation between the phenolic OH content and the antibiofilm activity against Gram negative bacteria and, to a lesser extent, against S. aureus (see Supplementary Figure S8). These data, combined with the previously mentioned effect of the different derivatizations on the phenolic OH content of the tannins, suggest that the phenolic OH content of the tannins is more important for the antibiofilm effect against Gram negative bacteria than against S. aureus. However, we did not observe a correlation between the antibacterial activity against planktonic bacteria and the phenolic OH content of the assayed tannins (see Supplementary Figure S9), which is in accordance with the study Kim et al. (2020), who did not find a significant correlation between the total phenolic content of several plant extracts from Chinese traditional medicine and the antimicrobial activity against S. aureus. However, this contradicts a previous study from Vattem et al. (2004), who found a linear correlation between the phenolic content of cranberry pomace and the antimicrobial activity against Listeria monocytogenes, Vibrio parahaemolyticus and E. coli.

It is important to note that the previous studies are focused on the antibacterial effect against planktonic bacteria, since there are no previous studies pointing to the effect of the hydroxyl or phenolic content on the capability of tannins to inhibit biofilms. However, research on other classes of polyphenolic compounds showed that the quantity and the position of the hydroxyl groups affects the capabilities of polyphenolic compounds to disperse biofilm, which could explain the results obtained in this research with the assayed tannins. In more detail, the research of Cho et al. on the capability of flavonoids to disperse S. aureus biofilms showed that there is a direct correlation between the quantity of hydroxyl substitutions of the flavonoid molecules and their antibiofilm capability (Cho et al., 2015), and the research of Kim et al. with phorbaketals against S. aureus biofilms indicated that the position and quantity of hydroxyls substitutions in the molecule affects their antibiofilm capacity (Kim et al., 2021), which is also consistent with the results of Arshia et al. (2017).