Identification of plants with antibacterial activities

Literature search was conducted using the search string: ‘Plant OR extract AND antibacterial OR antimicrobial OR activity AND Escherichia coli OR Klebsiella pneumoniae’. The search was done on PubMed^®, Web of Science^™ and Google Scholar databases, targeting full research articles published between 1948 and 2018 in English.

Target plant species were identified by virtue of having crude extracts with moderate to high activities (Minimum Inhibitory Concentrations (MIC) of ≤256 μg/mL) against E. coli or K. pneumoniae, as determined by broth dilution assays [15].

Materials

Acetone, dimethyl sulfoxide (DMSO), ethanol, ethyl acetate, n-hexane, methanol, petroleum ether, iodonitrotetrazolium chloride (INT) were purchased from Sigma-Aldrich Chemie, (Schnelldorf, Germany); Mueller Hinton Broth (MHB), Lysogeny Broth (Lennox) (LB) and agar were purchased from Carl Roth (Karlsruhe, Germany); gentamicin sulfate from AppliChem (Darmstadt, Germany) and demineralized water.

Preparation of crude extracts

All materials were collected from fully matured plants in March, May and August 2018 from the Botanical garden of the University of Wuerzburg. Collected plant materials were kept in open paper bags and transported to the laboratory within 2 hours. The materials were then chopped into small pieces and air-dried under shade at room temperature for one to two weeks.

Dried plant materials were size reduced into coarse powders using an electric blender (Braun, Germany). Extraction was performed using 72 h maceration at room temperature and magnetic stirring. Extraction solvent types and sequences were reproduced as reported in the cited articles as indicated in Tables 2 and 3. Crude extracts solutions were obtained upon filtration and were dried under vacuum at 40 °C. Extracts obtained from solvents composed of alcohols and water (80% v/v) were further dried in a freeze dryer (Martin Christ Gefriertrocknungsanlagen, Germany) at -60 °C and 0.03–0.13 mbar. Dried extracts were weighed and stored at -15 °C until they were further used.

Preparation of stock and working extract test solutions

Two sets of stock solutions were prepared by dissolving the extracts in 10% dimethyl sulfoxide (DMSO) and acetone, respectively, followed by ultrasonication (Bandelin electronic, Germany) for 15 minutes. Clear solutions were then obtained by centrifugation of the sonicated samples at 13000 RPMs for 10 minutes (Heraus pico 17, Thermo Fisher Scientific, Germany). 2 mL of working solutions were prepared at concentrations equivalent to 2048 μg/mL by diluting the respective volumes of stock solutions with MHB media. Following the dilutions, working solutions were mixed for 10 minutes on a lateral shaker, followed by centrifugation for 10 minutes. The concentrations of DMSO and acetone in the working solutions were 2.048 and 20.48% (v/v), respectively.

Qualitative phytochemical screening

Qualitative phytochemical screening was done on extracts from identical plant species whose phytochemical profiles were reported in the respective publications. Qualitative tests for the presence of alkaloids, terpenes, sterols, flavonoids, phenols and tannins were carried out as per methods described by Harbone [16]. Moreover, we used LC-MS to study the profiles of an extract which phytochemical profile was established using GC-MS and LC-MS techniques. The procedure for LC-MS analysis is described in the section below.

Solubility testing and comparative LC-MS analysis

The extents to which the extracts had dissolved in the respective solvents were semi-quantitatively evaluated based on the amount sediments at the bottom of the Eppendorf tubes after each centrifugation, as described above.

LC-MS analysis was done on selected extracts to further investigate the differences in compositions between the working solutions prepared using acetone and DMSO. Sampling of the extracts was done to include extracts with better solubility in acetone (C. longa and G. tinctoria) as well as in 10% DMSO (F. carica) (Table 4).

In that, stock solutions equivalent to 10 mg/mL were prepared from the extracts in acetone and 10% DMSO, followed by 10 min. ultrasonication and centrifugation at 13000 RPMS for 10 min. To prepare working solutions, 1 mL of the stock solutions diluted with 4 mL of each distilled water, with subsequent 10 min ultrasonication and centrifugation at 13000 RPMS for 10 min.

To remove the solvents, 2 mL of the supernatant from each working solution were transferred to 10 mL preparation glasses and were freeze dried for 24 h. Solution for LC-MS analysis were thereafter obtained by reconstituting the dried residues with 1 mL of methanol. Analysis was conducted by injecting 10 μL of each obtained solution into an LCMS-2020 (Shimadzu, Japan) system, using previously reported chromatographic conditions [17].

Antimicrobial susceptibility testing

Since a broad range of bacterial strains were used in the referred studies, we opted to use reference Escherichia coli (ATCC 25922) and Klebsiella pneumoniae (ATCC 10031) strains. The two strains have no resistance mechanisms to antibiotics and were obtained from the American Type Culture Collection (ATCC).

Overnight bacteria cultures were prepared by suspending one colony from LB agar plates into 2 mL of autoclave sterilized LB broth followed by 24 h incubation at 37 °C under lateral shaking (Edmund Bühler GmbH, Germany). Fresh cultures were then prepared by transferring 200 μL of the overnight culture into 20 mL of LB broth with a subsequent incubation for 6 h at 37 °C under lateral shaking. The number of Colony Forming Units (CFU) in the fresh cultures was determined using the optical densities (Eppendorf BioPhotometer Plus, Eppendorf AG, Germany) of ten-times diluted fresh cultures and Newman’s correlation curve. Bacterial suspensions containing 10⁶ CFU/mL of each bacteria were prepared by drawing the appropriate volume of the fresh culture into the corresponding amount of MHB media. These suspensions were used within 30 minutes from their preparation [18, 19].

Antimicrobial susceptibility testing was done using broth microdilution assays on 96 wells microtiter plates (Greiner Bio-One, Austria). Using a multichannel pipette (Eppendorf AG, Germany), 100 μL of autoclave MHB media was added in triplicates into wells in rows 2 to 12, followed by 200 μL of the extract test solution in the first row. A serial two-fold dilution was then done by drawing 100 μL of the extract solution from the first row and mixing with 100 μL of MHB media in the second row. The procedure was repeated to the last row, whereby the final 100 μL was discarded [18, 19].

Following the serial dilution, 100 μL of 10⁶ CFU/mL bacteria suspension in MHB media were added into the respective wells on the microtiter plate in triplicates. Each extract was tested in a range of 0.5–1024 μg/mL. The highest concentration of DMSO and acetone, respectively in the first wells were 1.024 and 10.24%(v/v). Gentamicin sulfate was used as a positive control in a range of 0.0156–32 μg/mL. Negative control involved solutions of DMSO and acetone, respectively, in MHB media at 1.28%(v/v) and 10.24%(v/v). Other controls in place included sterility controls for crude extracts and MHB media, as well as bacteria growth controls. All extracts and controls were tested in triplicates and the experiments were repeated twice in accordance to references [13, 20, 21].

The loaded plates were incubated for 18 h at 37 °C (Hera Cell incubator, Heraeus, Germany); then 40 μL of 0.2 mg/mL solution of iodonitrotetrazolium chloride (INT) was added into all wells and further incubated for 30 minutes. The MICs were determined by visual observation for wells with no formation of pinkish coloration, indicating the absence of actively diving bacteria.

Accounting for aspects likely to affect reproducibility

Upon evaluation of selected aspects in the articles used to determine our choices of the studied plants, several inconsistencies were observed:

1. On aspects related to practices during the collection of plant materials; 13 out of 20 articles did not specify the time of the year/season in which the collection was done. Additionally, the growth state of the plant at the time of collection (e.g. maturity, flowering) was not indicated in all articles. On the other hand, 18 out of 20 articles adhered well to the reporting of the location(s) from which the studied plant(s) were collected (Fig 4a).
2. Moreover, 11 out of 20 articles indicated to have used reference bacterial strains from sources such as the American Type Culture Collection (ATCC), Microbial Type Culture Collection (MTCC), National Collection of Industrial Microorganisms (NCIM), Center of Institut Pasteur (CIP) or PMFKg. Also, 5 out of 20 articles reported the use of only clinical isolates. Further, 4 out of 20 articles did not specify the sources of the bacteria used in conducting the AST studies (Fig 4b).
3. Regarding methodological aspects in the conduction of antimicrobial susceptibility tests; 6 out of 20 articles did not specify the solvent(s) used in dissolving the crude extracts before testing. Only 5 out of 20 articles cited the Clinical and Laboratory Standards Institute (CLSI), formerly called the National Committee for Clinical Laboratory Standards (NCCLS). The remaining articles (16 out of 20) indicated to have cited other journal articles or a textbook as a reference for the applied test methods (Fig 4b).
4. Furthermore, we noted varying methods used in evaluating the MIC values of the studied extracts. These included unaided visual observation for turbidity, colored indicator-aided visual observation, and the use of spectrophotometric devices. Additionally, the positive and/or negative control(s) used in the AST experiments were missing in 8 out of 20 evaluated articles (Fig 4b).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4.

a. Frequency of missing key information/aspects in the referred articles (n = 20). b. Variations in methodological approaches in antimicrobial susceptibility testing with respect to identities of studied bacteria, origins of test methods and visualization techniques for ascertaining the MIC values (n = 20).

https://doi.org/10.1371/journal.pone.0255437.g004

Antibacterial activity of identical plant species

Low chances of reproducing previously reported antibacterial activities can be anticipated even upon ensuring the use of the same plant, bacteria species, extraction solvents, studied plant parts, and testing methods. We could not reproduce any of the previously reported MICs of the tested plants against the tested bacteria. The MICs we observed in Cinnamomum verum L. and Piper betle L. were at least four folds higher than those previously reported (Table 2). This was irrespective of ensuring the use of same plant species and extracting solvents, using recommended solvents for dissolving extracts, and employing reference bacterial strains with no known resistance mechanisms. With some variations in the extraction and MIC reading methods, a small degree of variation in the MIC values was expected. However, our results showed relatively high deviations from those previously reported (Table 2).

On the one hand, reproducibility challenges are due to solubility issues of crude extracts and varying composition of phytochemicals due to differences in geographical locations, sampling, climatic conditions, and ecological factors [9, 10, 42, 43]. On the other hand, this might be due to unstated plant-related and experimental details such as antibacterial testing methods and used bacterial strains [13, 14]. To account for climatic conditions, plants used in this study were grown in conditions simulating their usual environments in terms of temperature and relative humidity (Table 1). Other challenges will be discussed in more details.

Solubility of plant extracts

Crude extracts from extracting solvents of varying polarities dissolve better in acetone as compared to the commonly used DMSO 10% solution. For instance, 17 out of 24 extracts showed better solubilities in acetone during the preparation of stock solutions (Table 4). However, the MIC values of extracts dissolved in acetone were generally not different from those dissolved in DMSO (Tables 2 and 3).

Since most of antibacterial plant-derived compounds are of low to intermediate polarities [14, 44–46], acetone is well suited for uses in extracting and dissolving of crude extracts from plant matrices. This is as well due to its good miscibility with water and non-toxicity to bacteria at higher concentrations (25% v/v) [20, 21, 47]. On the other hand, the use of DMSO offers better compounds’ stability in solution, as well as lower vapor pressure. These features are crucial when prolonged storage or testing times are needed [48].

Insufficient solubility of the many extracts in 10% DMSO prompted our undertaking to use acetone in order to avoid missing out compounds with antibacterial activity. LC-MS analysis showed qualitatively similar compositions of the working solutions prepared from the two solvents. Further, a better solubility of an extract in a particular solvent was related to the presence of higher quantities of corresponding phytochemicals in the resulting working solution (Figs 1–3b). However, some extracts indicated additional compounds in acetone dissolved extracts’ solutions, and most likely did not miss any compounds present in 10% DMSO dissolved solutions.

Since the solvent used for extraction as well as the diluent have an influence on the composition of the resulting test/working solution, a careful selection of the two is necessary. Based on quantitative benefits, our findings are suggestive of favoring acetone as a diluent when handling extracts obtained from less polar solvents and using 10% DMSO for extracts from more polar organic solvents.

Plant materials, sampling and phytochemicals composition

Reporting of essential details about plants used in the screening for antibacterial activities is inadequate in a big proportion of published research articles. This is demonstrated by the missing information on the season and location of collection, as well as the maturity state of plants at collection in a big number of referred studies (Fig 4a). This challenge is aggravated by the observed low reporting of both qualitative and quantitative phytochemical profiles of the studied extracts. Since plants contain different types and quantities of phytochemicals in different seasons and at different maturity stages, stating of these details is crucial [9, 11, 42, 43]. The lack of this information results in less objective plant collection, and contributes to low reproducibility.

However, the amount and types of phytochemicals in plants are likely to vary even during different hours of a day [49]. The availability of details of season and plant’s stage of maturity at collection may therefore not address this challenge in full. Factors such as the amount of sunlight and water received, soil type, and predators or pathogens induction of phytochemicals production may also largely vary in different geographical locations even during a similar season [9–12]. This is examplified by the negative tests for sterols and caffeic acid in methanol leaves extract of V. album and acetone leaves extract of O. europaea respectively, as opposed to the corresponding previous reports. The absence of sterols in the V. album extract may have contributed to the observed discrepancies in antibacterial activities. However, this cannot be stated with certainty because the initial study did associate the observed antibacterial activities to any of the phytochemicals found in the extract. On the other hand, caffeic acid was shown to have inhibitory activity againt E. coli and K. pneumoniae among other bacteria. The differences in antibacterial activities observed in O. europaea leaves extracts can therefore be related to the missing sole or synergistic role of caffeic acid [22, 24].

Taken together, it is of great necessity to indicate the season and state of maturity alongside location(s) of plant collection. This will enable others to make all possible adjustments towards conforming to the previously reported conditions. Moreover, doing qualitative and/or quantitative fingerprint profiling of tested extracts using techniques like Thin Layer Chromatography (TLC), High Performance Liquid Chromatography (HPLC), or Liquid Chromatography-Mass Spectrometry (LC-MS) among others, should be considered necessary. Using fingerprint profiles enables an objective comparison on the extents at which the extracts to be studied are similar to those used previously.

Testing for antibacterial activities

There is a limited use of reference bacteria isolates in carrying out antimicrobial susceptibility testing of crude extracts from plants. Moreover, uncomprehensive reporting of methodological aspects and the use of varying references for methods of antimicrobial susceptibility testing are common.

Our evaluation revealed that 9 out of 20 referred articles either indicated the use of clinical isolates or did not give any reference(s) of the studied bacteria (Fig 4b). The use of clinical isolates largely limits the reproducibility of obtained results by researchers elsewhere. Even upon successfully reproducing other factors, the genetic and phenotypic variations among clinical isolates of a particular bacterium may hinder the objective screening of plant extracts for antibacterial activity [14]. Lower susceptibilities to plant extracts have been observed in tests that involved the use of clinical isolates of bacteria or those with known resistance phenotypes [50–52].

Furthermore, we observed missing details on the solvent(s) used in the preparation of extracts’ test solutions as well as on the applied positive and/or negative controls used (Fig 4a). The non-disclosure of solvents used in preparing test samples disguises a proper choice of solvent(s) in the follow-up studies. Moreover, the use of solvents toxic to bacteria such as methanol and ethanol, or using recommended solvents above optimal concentrations precipitates the reporting of false-positive results [13, 14, 53]. The lack of information on the used negative controls aggravates the magnitude of this challenge.

Additionally, the approaches in the testing methods may be widely varying due to the observed diversity of references of methods for antimicrobial susceptibility testing. Among the 20 articles we referred to, 16 cited other published research papers as a reference for the used AST method (Fig 4b). Since each cited article might have done slight modifications of a previously reported method(s), this is also likely to affect the reproducibility [13, 14].

To warrant good reproducibility, it is therefore necessary to ensure the use of reference bacterial strains, especially during initial screening of plant extracts for antibacterial activities. This will enable others to select objectively the type of bacterial strains to use during the follow-up studies. Additionally, a thorough reporting of key aspects of the AST experiments and the use of standard methods from bodies like CLSI and EUCAST is a commendable approach in safeguarding the reproducibility of reported results [13, 18, 19].

Antibacterial activity of related plant species

Most interestingly, we found more extracts with higher antibacterial activity in the species related to those primarily identified in our literature review. This indicates that, screening of related species for antibacterial activities is a good approach in the search for antibacterial hit compounds from plants. As shown in Table 3, antibacterial activities were observed in the extracts of G. spicata and P. officinalis. While the MIC values exhibited by G. spicata extracts were at least 6.5 higher than those previously reported in G. smeathmannii; that of P. officinalis extract against K. pneumoniae (128 μg/mL) was about two times lower than that previously reported in P. broteroi (250 μg/mL). Additionally, the antibacterial activity of the acetone extract of P. officinalis against E. coli (MIC = 256 μg/mL), was not seen previously.

Expanding the search perimeter by screening of species related to those with previously reported activities is therefore noteworthy [14, 54]. This approach is generally better than the common practice of selecting plants based on their traditional uses in the treatment of bacterial infections. This is underscored by the reports that antibacterial activities observed among randomly collected plants did not significantly differ from those being collected on the bases of their usage by traditional healers [14].