American Civil War plant medicines inhibit growth, biofilm formation, and quorum sensing by multidrug-resistant bacteria

A shortage of conventional medicine during the American Civil War (1861–1865) spurred Confederate physicians to use preparations of native plants as medicines. In 1863, botanist Francis Porcher compiled a book of medicinal plants native to the southern United States, including plants used in Native American traditional medicine. In this study, we consulted Porcher’s book and collected samples from three species that were indicated for the formulation of antiseptics: Liriodendron tulipifera, Aralia spinosa, and Quercus alba. Extracts of these species were tested for the ability to inhibit growth in three species of multidrug-resistant pathogenic bacteria associated with wound infections: Staphylococcus aureus, Klebsiella pneumoniae, and Acinetobacter baumannii. Extracts were also tested for biofilm and quorum sensing inhibition against S. aureus. Q. alba extracts inhibited growth in all three species of bacteria (IC50 64, 32, and 32 µg/mL, respectively), and inhibited biofilm formation (IC50 1 µg/mL) in S. aureus. L. tulipifera extracts inhibited biofilm formation (IC50 32 µg/mL) in S. aureus. A. spinosa extracts inhibited biofilm formation (IC50 2 µg/mL) and quorum sensing (IC50 8 µg/mL) in S. aureus. These results support that this selection of plants exhibited some antiseptic properties in the prevention and management of wound infections during the conflict.

Antibiotic resistance in pathogenic microbes poses a significant threat to human health 1 ; antibiotics are critical not only in treating bacterial diseases but also in enabling surgery and other procedures with high risks of infection. Given the great genetic diversity and capacity for evolution present in bacteria, a rise in antibiotic resistance is an inevitable response to antibiotic use. For example, in 1940, even before penicillin was widely used, penicillin resistance was observed. Any single antibiotic, then, is not a permanent solution but another step in the struggle against infection.
Several factors complicate the relationship between antibiotics and bacteria. For example, the innate immune system plays a role in fighting infections with or without the use of antibiotics. Further, commensal members of the microbiome may compete with pathogenic bacteria or may themselves become pathogenic under certain circumstances. Relevant to this study, bacterial community effects such as biofilms and quorum sensing produce resistance and virulence phenotypes not necessarily observed in vitro 2,3 . Biofilms are extracellular mixtures of polysaccharides and proteins that can physically protect bacterial populations from antibiotics and immune responses 2,4 . Consequently, biofilms are associated with chronic infections, especially in the cases of indwelling medical devices and implants, and there is currently a lack of effective treatments for these conditions 4 . Quorum sensing is a system by which toxin production or other pathogenic activity is initiated when extracellular communication indicates achievement of a threshold population of bacteria. Inhibition of quorum sensing and biofilm formation, then, can be therapeutic but not bactericidal 3 . In the absence of new antibiotics, multidrug-resistant infections may be treatable by administering biofilm inhibitors or quorum quenchers to increase the vulnerability of bacteria to the immune system or conventional antibiotics. www.nature.com/scientificreports www.nature.com/scientificreports/ tissue structures. Multidrug-resistant bacteria were used in all experiments to examine the potential use of these plant compounds to combat emerging resistance in species commonly found in wound infections today.
Results extract yield. Extraction in MeOH yielded six crude extracts, representing Q. alba bark and galls, A. spinosa leaves, and L. tulipifera leaves, root inner bark, and branch bark (Table 1). Extract yield was highest (27.1% of dry mass) in extract 620 (Q. alba galls). Other crude extracts had yields ranging from 8-11%. Masses of partitions and fractions of crude extracts varied from <0.1% to 4% relative to dry plant matter (Supplementary  Table S1). Partitions were labelled B, C, D, and E for solvents hexane, ethyl acetate, n-butanol, and water, respectively; non-tannin fractions were labelled F1 and tannin fractions were labelled F2. The non-tannin fraction of L. tulipifera leaves (616-F1) was more than 10 times as massive as the tannin fraction, suggesting that tannin content is not high in L. tulipifera leaves. The tannin and non-tannin fractions of Q. alba bark were similar in mass.
Extracts which displayed strong activity against S. aureus, A. baumannii, K. pneumoniae, and P. aeruginosa (619, 619-F2, and 620) were tested for growth inhibition of S. epidermidis and additional strains of A. baumannii and K. pneumoniae. All three of these Q. alba extracts inhibited growth in the strains of A. baumannii (IC 50 32-256 μg/mL), but not in the additional K. pneumoniae strains tested (Table 3). Q. alba extracts 619 and 619-F2 were found to inhibit growth of S. epidermidis (IC 50 256 and 64 μg/mL, respectively).
Quorum sensing inhibition. Transcription of S. aureus agr types I, II, and III was inhibited by several Civil War extracts (Fig. 3). L. tulipifera extract 617 C, A. spinosa extract 618 C, and Q. alba extract 619-F1 exhibited the most activity in these assays, primarily against agr III (IC 50  Cytotoxicity. Of the 19 extracts studied, 13 were recognized to have potential antibiotic activity and were tested with human keratinocytes (HaCaT) to counter test for cytotoxicity. L. tulipifera root bark extracts 617 and 617 C displayed high levels of cytotoxicity (IC 50 16 μg/mL in each case). Q. alba extracts displayed no significant cytotoxicity at test concentrations (2-256 μg/mL). Chemical analysis. Q. alba extracts 619-F2 and 620 were selected for chemical analysis because of their strong antibacterial activity both in growth inhibition and in adjuvant assays and because of their lack of toxicity towards human cells. Initial HPLC indicated a wealth of early eluting compounds, so the chromatographic conditions were adjusted for LC-FTMS to achieve greater separation in that region. LC-FTMS revealed that 619-F2 and 620 have few compounds in common (Fig. 5).

Discussion
Extracts of L. tulipifera, A. spinosa, and Q. alba displayed inhibitory activity against bacteria that cause skin and soft tissue infections, substantiating their use as antiseptics during the American Civil War. These medicinal plants may be useful in modern medicine as treatments for antibiotic-resistant bacteria. Of particular interest are 618B and 620 as S. aureus biofilm inhibitors and 619, 619-F2, and 620 as growth inhibitors of carbapenem-resistant Klebsiella pneumoniae.
While a 1947 survey of antibacterial properties of plants found no activity in A. spinosa and L. tulipifera 15 , the positive results in this experiment may be explained by differences in a number of factors. The previous study used H 2 O extracts whereas this experiment used MeOH extracts 15 ; L. tulipifera bark was historically prepared for treatment by dissolving in EtOH 5 , which produces an extraction profile similar to MeOH 16 . Additionally, given the role of endophytic microorganisms in the synthesis of secondary metabolites, the chemical composition of plant extracts can vary based on differences in the plant microbiome 17 . Other possible sources of variation include collection date and location, assay method, and extract concentration tested. Finally, given the variability in how different laboratories may perform one type of extraction, results can vary between related studies. For example, of two studies that evaluated Aralia nudicaulis root (a traditional Native American remedy ingredient) for growth inhibition of mycobacteria, only one reported moderate antibacterial effects while the other reported little activity 18,19 .
In his report, Porcher recommended the entire genus Quercus as a source of antiseptics 5 . This activity is confirmed not only by the results of the experiments reported herein, but also by multiple other studies showing antibiotic effects by Quercus spp. extracts [20][21][22][23][24] . A European herbal remedy referred to as Quercus cortex (originating from Q. robur, Q. petrea, and Q. pubescens bark) has shown weak antibacterial and quorum sensing inhibition effects 25 . Acorn extract from a variety of oaks has shown inhibition of both Gram-positive and Gram-negative bacteria 26 .
However, the activity of various Quercus spp. extracts is far from uniform. For example, the Q. alba gall extract (620) in this study inhibited growth of drug-resistant K. pneumoniae whereas a study of Q. infectoria galls found no significant inhibition of drug-resistant K. pneumoniae 24 .
Antibacterial activity in oak extracts is frequently attributed to tannins 27 , compounds that typically interfere with biological processes by binding to proteins 28 . In Quercus, tannin content is typically highest in galls, with a reported 70% tannin content in Q. infectoria galls 27 . In this experiment, higher activity in 620 (gall crude extract) over 619 (bark crude extract) and 619-F2 (bark tannin fraction) over 619-F1 (bark non-tannin fraction) suggests that Q. alba's growth inhibitory activity is due to tannins. However, quorum sensing inhibition by 619-F1 suggests that additional compounds could contribute to the antibacterial activity of crude oak extract, the medicine used during the Civil War.
LC-FTMS analysis of 619-F2 and 620 confirmed the existence of a variety of tannins in both extracts (Supplementary Tables S6 and S7). Of particular interest are ellagitannin isomers, 2, found in 620; as well as related ellagitannins 12a and 12b. Ellagitannins have been reported to have antibiotic activity against antibiotic-resistant S. aureus 9 . While only three MS peaks were found in common between 619-F2 and 620, both extracts are rich in tannins. 619-F2 is enriched in procyanidin condensed tannins and 620 contains many ellagitannins and triterpenes. Amp www.nature.com/scientificreports www.nature.com/scientificreports/ Tannins have been shown to inhibit growth in a wide range of bacteria, fungi, and viruses. Suggested mechanisms of action include inactivation of microbial enzymes, inhibition of membrane transport, and sequestering essential metal ions in complexes 28 . Tannins may also act as biofilm inhibitors by binding to matrix proteins 29 . However, tannins have also been found to bind with digestive enzymes and nutrients such as proteins and starches, and as such are generally considered as anti-nutritive; a variety of animals have shown gastrointestinal distress and decreased growth when fed on high-tannin diets 28 . Because of this nondiscriminatory binding, external applications of Q. alba extracts would be preferable to internal or systemic applications; Porcher recommended that powdered oak bark be applied in a wash for gangrene and a poultice for wounds 5 .
Leaves of several Quercus species (Q. cerris, Q. ilex, Q. virginiana, Q. incana) have also shown antibacterial properties, including biofilm and quorum sensing inhibition 20,22,30 . One future research direction could be to compare the antibacterial properties of Q. alba leaves with the activity identified in bark and gall extracts.
While A. spinosa has several reported uses in traditional medicine 31 , it has not frequently been studied for medicinal properties. The most notable results of this experiment for A. spinosa are significant biofilm inhibition by 618B (leaf hexane partition) and quorum sensing inhibition by 618 C (leaf ethyl acetate partition). The presence of these adjuvant properties rather than simple growth inhibitory activity in A. spinosa leaves may explain the 1947 report of no significant antibiotic activity in A. spinosa 15 .
Other Aralia species have exhibited antibacterial activity in roots 18 and aerial parts (flowers, leaves, and stems) 32 , including biofilm inhibition by A. cachemirica 32 . In his list, Porcher also ascribed antiseptic activity to A. racemosa 5 .
L. tulipifera has been widely studied and its various parts have exhibited a variety of medicinal effects including antibacterial 33 , anti-malarial 34 , and anti-cancer 35,36 activity. The other species of Liriodendron, L. chinense, is used in Chinese traditional medicine and has been shown to have antibacterial effects 37 . Additionally, an extract www.nature.com/scientificreports www.nature.com/scientificreports/ from a hybrid of L. tulipifera and L. chinense has been shown to exhibit inhibition of biofilm production and quorum sensing 38 .
In the present study, L. tulipifera extracts demonstrated activity in the inhibition of growth, biofilm production, and quorum sensing. However, the root bark extract (617), which is generally more bioactive than the leaf extract (616) and branch bark extract (621) in our models, displayed significant mammalian cytotoxicity (IC 50 : 16 µg/mL). It may therefore be ill-suited for medicinal use, or at least dose-limited. A study of L. tulipifera for antiplasmodial activity also found high cytotoxicity in active fractions but it has been suggested that, given the use of L. tulipifera in traditional medicine, toxicity may not be problematic in vivo at therapeutic doses 34 . Porcher recommended root bark as the medicinal part of L. tulipifera to be harvested 5 ; perhaps preparation techniques or dosage made the potency/toxicity trade-off worthwhile in a wartime context. Interestingly, Porcher also suggested L. tulipifera bark as a substitute for Cinchona bark in malaria treatment, an application supported by recent research 34 .
Perhaps the most notable L. tulipifera extract with low toxicity is 616-F1 (leaf non-tannin fraction), which displayed little growth inhibition but significant biofilm and quorum sensing inhibition-an adjuvant effect similar to the A. spinosa extracts tested.
Further study should focus on bioassay-guided fractionation, a recursive process of fractionation and bioassay to identify individual active compounds and synergistic relationships. Of the extracts tested, 616-F1, 618B, 618 C, 619-F2, and 620 exhibit the most promise for antibiotic NCEs and are good candidates for this process. Specifically, the HPLC methods developed for 619-F2 and 620 could be used to produce further fractions with adaptation to preparative liquid chromatography.
In vivo testing of the antibacterial properties of extracts active in vitro is also a logical next step in this research. Given the potential of some of these extracts as adjuvants rather than direct antibiotics, they may be tested as adjuvants with existing, FDA-approved antibiotics for the potentiation of antibacterial activity in wound infections.
Finally, given the activity seen in the extracts tested in this study, it may be worthwhile to investigate the antibacterial properties of other plants recorded as antiseptics in Porcher's book. In total, 37 plant species were described as having antiseptic applications 5 . As the global spread of antibiotic-resistant strains of bacteria continues, it is increasingly important to consider all possible sources of new, and perhaps old, treatments. www.nature.com/scientificreports www.nature.com/scientificreports/ Methods plant material. Samples of Liriodendron tulipifera, Aralia spinosa, and Quercus alba were identified and collected in May 2015 from Lullwater Preserve on the Emory University campus in Atlanta, Georgia. Leaves were gathered manually and a handsaw was used to cut segments of roots and branches for bark collection. Vouchers (Accession numbers 20338-20341) were deposited in the Emory University Herbarium (GEO) in Atlanta and digital copies of the specimens are accessible for viewing online via the SERNEC web portal 39 . Samples were dried and ground into powder by either a Wiley mill equipped with a 2 mm mesh or coffee grinder (Table 1). www.nature.com/scientificreports www.nature.com/scientificreports/ Extraction, partitioning, and fractionation. All ground material (Table 1)

was sonicated in MeOH
(1 g/10 mL). After 20 minutes the sample was filtered sequentially with Whatman filter paper 8 and 2, and then fresh MeOH was added to the plant material for a second round of sonication. The two filtrates were combined and dried in vacuo at ≤40 °C. The resulting residue was suspended in H 2 O, frozen, and lyophilized. The dried extract was collected and 20 mg of each extract was dissolved in DMSO (10 mg/mL) for biological testing.
Growth inhibition assays. Assays were carried out under CLSI M100-S23 guidelines 40 . A working culture was created by standardizing liquid culture using a BioTek Cytation3 and inoculating into CAMHB to a concentration of 5.0 × 10 5 CFU/mL. Working culture was added to extracts and controls in 96-well microtiter plates (Grenier-Bio 655-185) such that each well contained a total volume of 0.2 mL. Vehicle controls and antibiotic controls (ampicillin, kanamycin, and vancomycin for Staphylococcus spp. assays, gentamicin, tetracycline, and meropenem for other species, 0.5 to 64 µg/mL) were included for each strain. Extracts and vehicle were tested at a concentration range of 2.0 to 256 µg/mL, using 2-fold serial dilution. Plates were incubated at 37 °C, with S. aureus, S. epidermidis, and P. aeruginosa for 18 hours and A. baumannii and K. pneumoniae for 22 hours. Optical density (OD 600 ) was measured using a BioTek Cytation3 plate reader at initial and final time points, to account for extract colour. The IC 50 for growth was defined as the lowest concentration at which an extract displayed ≥50% inhibition and MIC (IC 90 ) at ≥90% inhibition.
Extracts active against multidrug-resistant A. baumannii (OIFC143) and K. pneumoniae (NR-15410) were tested for growth inhibition of S. epidermidis and additional strains of A. baumannii and K. pneumoniae.
Biofilm inhibition assays for S. aureus. Biofilm inhibition of S. aureus was performed as described previously 41 . Briefly, supplemented TSB with 3% NaCl, 0.5% dextrose, and 2% human plasma was used in 96-well microtiter plates (Falcon . Working cultures of UAMS-1 (wt) and UAMS-929 (isogenic ΔsarA mutant of UAMS-1) were standardized to a concentration of 5 × 10 5 CFU/mL and the final well volume was 0.2 mL. Extracts were assessed at sub-IC 50 concentrations for growth, ranging from 2.0 to 256 µg/mL. The vehicle and positive control, 220D-F2, were assessed from 2.0 to 256 µg/mL. All experiments were incubated statically at 37 °C for 22 hours. Optical density (OD 600 ) was measured using a BioTek Cytation3 plate reader at initial and final time points, to account for extract colour. Biofilms were rinsed twice with 1X PBS, fixed with 100% EtOH, and stained with crystal violet. The dry stain was eluted with ethanol, diluted in PBS, and quantified at 595 nm using a BioTek Cytation 3 plate reader. The MBIC 50 (minimum biofilm inhibitory concentration) was defined as the lowest concentration at which an extract displayed ≥50% inhibition and MBIC 90 at ≥90% inhibition. www.nature.com/scientificreports www.nature.com/scientificreports/ Quorum quenching assays for S. aureus. Examination of the quorum quenching potential of extracts against S. aureus was conducted as previously described 3 . Briefly, all agr fluorescent reporter strains were maintained in chloramphenicol (10 µg/mL) supplemented TSA and TSB. The assay was conducted in tissue culture-treated clear bottom, black-sided 96-well microtiter plates (Costar 3603) with a final well volume of 0.2 mL. Extracts were assessed at sub-MIC 50 concentrations, ranging from 0.5 to 64 µg/mL. Vehicle and positive control, 224C-F2, were also assessed from 0.5 to 64 µg/mL. Plates were incubated at 37 °C in a humidified chamber, shaking at 1200 rpm (Stuart SI505). OD (600 nm) and fluorescence (493 nm excitation, 535 nm emission) readings were taken at initial (0 hr) and final (22 hr) time points.