The seeds of T. fortunei were collected from the campus of Huazhong Agricultural University (Wuhan, China) and were dried and stored at room temperature until further experiment. The seeds were taken from palm trees alone two sides of Xueyuan Road, beginning from the West entrance of the university to the crossing of Zhuyuan Road. The plants were identified by Prof. Xuebo Hu (Huazhong Agricultural University) to be T. fortunei.

The plant material was extracted as described by Chatterjee et al. (2011) with some modifications. The seed coat of T. fortunei (2 kg) was exhaustively extracted three times with ethanol/water solution (70:30, v/v) for 2 h at 80°C at a revolution of 80 per minute under reflux. After 2 h, the solution was allowed to be cooled for 4 h with continuous agitation. Then the extract was removed from the apparatus, followed by paper filtration. After extraction, the ethanol in the solution was removed by rotary evaporator at 50°C and then sought to vacuum freeze drying. The extracts were weighed and then stored in a clean, dry and air-tight container with proper labeling in refrigerator at 4°C until further use.

Previously prepared extracts were further fractionated by column chromatography using ethanol and water solvent system Mbukwa et al. (2007). The extract was loaded on a glass column (10 cm × 250 cm) packed with the D101 macroporous resin (Sangon Biotech, China), and eluted with 0, 20, 40, 60, 80, and 100% ethanol (5 L each), to get different fractions of extract, respectively. Finally, each eluent fractions was evaporated under vacuum to obtain according fractions.

Different fractions of seed coat extract of T. fortunei were screened against a collection of Gram-positive and Gram-negative bacterial strains, including Staphylococcus aureus, S. epidermidis, Bacillus cereus, B. subtilis, Escherichia coli, and Pseudomonas aeruginosa, which were obtained from the China General Microbiological Culture collection center (Beijing).

Antibacterial activity was evaluated by two different methods. At first, extracts activity was determined by measuring the inhibition zone against selected bacteria. Once antibacterial activity was confirmed, the minimum inhibitory concentration (MIC) was checked. MIC of different fractions of extract of T. fortunei plant were evaluated on the selected bacteria using rapid p-iodonitrotetrazolium chloride (INT) colorimetric assay according to method described by Kuete et al. (2008). Briefly, we dissolved different plant fractions in dimethyl sulfoxide (DMSO) methyl-4-hydroxybenzoate (MHB) with a ratio of (10:90). In a 96-well we added 100 μL of this solution to each well. Then we serially diluted two-fold for two times, followed by addition of 100 μL of bacterial inoculum that we already prepared in Mueller Hinton broth. Afterward, sealed plates were placed in an incubator for a period of 20–22 h at 37°C. The final concentration of inoculum was 1.5 × 10⁶ colony-forming unit /mL and it was less than 2.5% for DMSO in each well. Wells with inoculum and DMSO (2.5%) only were declared as negative control; whereas, streptomycin and tobramycin were used as positive control. After 20 h, we added 40 μL of INT (0.2 mg/mL) to each well and re-incubated for a period of 30 min. MIC is defined as the least concentration of extract that can inhibit the growth of bacterial species.

For the determination of MMCs, a portion of liquid (5 μL) from each well that showed no growth of microorganism was placed on MHA or NA and incubated at 37°C for 24 h. The lowest concentrations that yielded no growth after this sub-culturing were taken as the MMCs (Nyaa et al., 2009).

Antimicrobial activity was explored by disk diffusion method as proposed by Bauer et al. (1966). Mueller Hinton Agar (MHA) and Nutrient Agar (NA) were used for the screening of in vitro antimicrobial activity. The MHA and NA plates were prepared by pouring 15 ml of molten media into sterile petri plates. These plates were allowed to solidify for 5 min, then swabbed uniformly with 0.1% inoculum suspension, and allowed to dry for 5 min. The 6 mm sterile disks were loaded with different concentrations (2.5 and 5 mg per disk) of extracts. Then these disks were placed on the surface of prepared medium and the compound was allowed to diffuse for 5 min and the plates were incubated for 22 h at 37°C. Then inhibition zones were measured with transparent ruler in millimeter around the disks. Each sample was analyzed in triplicate and values were presented as mm.

Organic fraction exhibiting the most antimicrobial behavior (20% ethanol fraction) was selected for identification of metabolites by performing UPLC/MS analysis. Purified fractions were filtered with filter papers of 0.45 μm pore size. Each of 50 μg of the sample was dissolved in 1 ml methanol (HPLC grade). The sample was then filtered through 0.2 μm filter paper (GH polypropylene 0.2 μm pore size, 47 mm diameter filter paper). Sample reaction solutions were transferred to UPLC vials (LC-MS certified 12 × 32 mm clear presilt combo, Waters) suitable for the UPLC-MS autosampler. A definite amount (30 μL) of filtered material was taken in a vial. Further, the samples were immediately analyzed by performing UPLC/MS analysis. Sample was prepared as described by Matsuura et al. (2008). Prepared samples were stored at 4°C before undergoing UPLC-MS operation. This precautionary measure was adopted to protect material from degradation.

The mass spectrometer was tuned according to the manufacturer’s manuals for optimal parameters for ion lenses, detector voltage and other settings. Sample reaction solutions were transferred to glass vials suitable for the UPLC/MS auto sampler. Each sample tube was closed immediately with crimps that contain a Teflon rubber seal. After 2 h, samples were injected into the UPLC/MS. The UPLC/MS analysis was carried out in a Waters’ (UPLC) coupled with Waters Q-ToF Premier Mass Spectrometer. UPLC conditions were used as described by (Matsuura et al., 2008). The LC column used was ACQUITY UPLC BEH C18 RF 1.7 μm, 2.1 × 50 column. Data analysis was carried out using MassLynx (Waters, Milford, MA, United States). Compounds were detected by comparing MS patterns with standard NIST Library. Metabolite peaks were quantified by area of targeted ion traces, and result of peak tables of all chromatograms were exported to MS Excel 14.0.

Obtained data were analyzed by performing both univariate statistics (t-test in Excel, ANOVA in MatLab) and multivariate statistics. The results of multivariate statistics were processed using minitab 17.1 (LEAD Tachnologies, Charlotte, NC, United States). Each experiment was repeated three times.

In an effort to explore antibiotic substances, we have screened up to 200 kinds of plant extracts and found T. fortunei was one of the strongest one (data will be published separately). In this study, we aimed to screen the active compounds from T. fortunei. After extraction of the palm seed coat by ethanol/water solution (70:30, v/v), chemicals in the solution were dried out and then subjected to macroporous resin column separation by elution with 0, 20, 40, 60, 80, and 100% ethanol. The increasing ethanol concentration is adopted to differentiate the chemicals by polarity. Fractions were dried up by vacuum evaporation. It was found that fractions eluted by 80 and 100% ethanol yielded little chemicals, therefore chemicals eleuted by of 0, 20, 40, and 60% ethanol were used for further study.

Disk diffusion method was firstly used for determination of antibacterial activity of previously prepared fractions against various pathogens, i.e., S. aureus, S. epidermidis, E. coli, P. aeruginosa, B. subtilis and B. cereus. Resulted antibacterial inhibitory zones of all the fractions against different bacterial strains were listed in Figure 1. It can be seen that all fractions had antibiotic activity against B. cereus, E. coli, S. aureus, S. epidermidis and but not P. aeruginosa and B. subtilis. Furthermore, 20% ethanol fraction showed the highest anti-bacterial activity against most of the bacterial strains used in this study.

Minimum inhibitory was performed through a series of double dilution beginning from 1.25 mg/ml. E. coli growth was inhibited by almost all the ethanol and aqueous fractions of sample plant material but it showed the maximum inhibition from 20% ethanol fraction where the MIC = 156.25 μg/mL. All the fractions inhibited the growth of S. epidermidis with the same potency (MIC = 312.25 μg/mL). Inhibitory activity of S. aureus was the same as E. coli. Likewise, 20% ethanol also revealed the lowest value of MIC against B. cereus (312.55 μg/ml) among all the tested fractions (Table 1).

Data in Table 2 represented the minimum microbicidal concentration (MMC) values of the different fractions of seed coat extract of T. fortunei against reference bacterial strains. Although results of MIC and MMC showed variations among the tested bacterial strains but both the values were equivalent to each other in most of the cases, showing anti-bactericidal effect. Of all fractions of sample plant, the lowest MMC value (312.5 μg/mL) was recorded by 20% ethanol fraction against S. aureus and S. epidermidis.

In order to search for the responsible chemicals for the antibiotic activity, we conducted UPLC/MS for all extractions. The Compounds that were identified from different ethanol aqueous fractions are compound 1–29. All the compounds were identified using NIST compound library on the basis of their mass spectra and their names are shown in Table 3.

The identified compounds from different fractions were then paired with antibacterial activity. Principal Component Analysis (PCA) divided all the compounds into two different groups: (I) with positive concentration coefficient; and (II) negative concentration coefficient. Positive coefficient indicates the increased extracted yields of specific compound with more concentrated ethanol solvents. However, compounds with negative concentration coefficient were more likely to be efficiently isolated with solvents of lesser ethanol concentrations. Bioactive compounds recorded varied antibacterial coefficient against S. aureus according to PCA. Compound 9 and compound 12 exhibited the highest antimicrobial coefficient (0.78 and 0.75, respectively), but compound 12 had the positive value (0.48) of concentration coefficient. Whereas, compound 9 had a negative concentration coefficient (-0.49) revealing the opposite extraction behavior than compound 12. Compound 10 was the third most active compound against S. aureus. Keeping in view only the compounds with positive concentration coefficient, compound 7 was the second most active compound against S. aureus, while compound 2 was the least active in this group. Among the compounds with negative concentration coefficient, compound 8 showed the least antibacterial activity against S. aureus. There were 13 compounds which stood neutral against the bacterial species, however, a total of 5 compounds were possessing negative coefficient values for controlling bacterial growth (Figure 2A).

Compounds 8, 9, and 10 were the most active compounds against E. coli among the group with negative concentration coefficient. Whereas, compound 7 recorded a comparable antibacterial coefficient to the compound 9, but with positive concentration coefficient value. Only one compound (29) didn’t affect the bacterial growth, while a total of 10 compounds exhibited negative antibacterial coefficient. Among those 10 compounds 3 had positive concentration coefficient and 7 exhibited negative concentration coefficient. The maximum negative antibacterial coefficient was shown by the compound 6 closely followed by compound 2. Collectively, the compounds showing substantial antibacterial coefficient were 7, 8, 9, and 10 (Figure 2B).

Compound 10 was the leading compound in controlling bacterial species S. epidermidis closely followed by compound 9. Compound 8 was the third compound among the group of compounds with negative concentration coefficient; however, it occupied overall fourth position in terms of antibacterial coefficient closely following compound 7. Keeping in view the low antibacterial activity, compound 6 showed the least coefficient, which was close to the values of compounds 1, 2, and 16. However, compound 16 was categorized among the compounds with negative concentration coefficient (Figure 2C).

Compound 10 occupied the highest antibacterial coefficient value with reference to B. cereus controlling activity, among all the detected compounds. Moreover, compound 9 and compound 8 possessed the second and third highest antibacterial activity in the group with negative concentration coefficient. However, compound 7 recorded the overall second position in comparison to all other compounds; but, its position among group of compounds with positive concentration coefficient was first. Compound 1, 2, and 6 had the lowest coefficient values for controlling bacterial species B. cereus in the positive concentration coefficient group. Whereas, among the compounds with negative concentration coefficients, compound 16 and 19 occupied the eminent position due to their lowest antibacterial activity (Figure 2D).

No compound detected in UPLC/MS could affect the growth of bacterial species B. subtilis and P. aeruginosa, hence producing zero coefficient values for their antibacterial activities. The only distribution of compounds could be seen on the axis of concentration coefficient revealing the differential extraction behavior of compounds against ethanol concentration (Figures 2E,F).

Based on the above analysis, all the compounds were grouped into four different categories, i.e., α, α, ф, λ (Figure 3). The highest anti-bacterial activity was shown by α group. This group contains four different compounds having activity coefficient ≤ 0.03. Based on PCA analysis, compound 10 possess maximum anti-bacterial activity followed by compounds 7, 9, and 8. Four of these compounds (among three of them were not previously reported for antibacterial activity) were selected to test anti-bacterial activity by direct exposure to six bacterial strains. The β group showed closely lower anti-bacterial activity than α, and it also contained four member compounds 4, 5, 11, and 12. These compounds can be potentially used as anti-bacterial agents. However, they were screened and not tested for confirmation of their anti-bacterial potential in the current study. The 16 members of ф group accounted for slightly positive or negative behavior against bacterial species. Meanwhile, five compounds fall under λ category and they attributed negative anti-bacterial coefficient value suggesting their neutral to supporting behavior for bacterial pathogens according to the results of PCA.

Principal component analysis generated antimicrobial profiles of UPLC/MS identified four different compounds, i.e., 7, 8, 9, and 10 having significant antimicrobial activity against most of the important bacterial pathogens. Among these four, three compounds are not previously reported for their antibacterial activity, while compound 8 anti-microbial activities has been previously reported (Murthy et al., 2011). Therefore, we used rest of the three compounds 7, 9, and 10 that were not previously reported for further study (Figure 5). To ensure the exact identification of the compounds UPLC/MS run of their individual standards were carried out. The peaks of the UPLC/MS run of standard chemicals were the same as found in UPLC/MS chromatogram of 20% ethanol fraction (Figures 4, 5).

Further, we evaluated antibacterial activity of pure bio-active compounds of fraction 20% ethanol. Compounds selection was made based on their peak area by UPLC/MS and activity of different fractions against bacterial strains. PCA was used for selection of compounds with significant antibacterial potential.

Compound 10, isolated from fraction 20% ethanol showed maximum antibacterial activity against all the bacterial strains followed by compound 9, whereas compound 7 was unable to inhibit the growth of any bacterial strain (Table 4). Compound 10 showed varied antimicrobial activity against all test bacterial strains. The maximum inhibitory zone (14 mm) was seen against S. epidermidis. This was followed by B. cereus by showing an inhibitory zone of 13 mm as shown in Figure 6. Likewise, compound 9 also showed maximum inhibitory potential against S. epidermidis and provided inhibition zone of 13 mm. In the same way, inhibition zone of 11 mm was observed when this compound was used against B. cereus as shown in Table 4. Altogether, the inhibitory activity of pure compounds was found less as compared to the parental 20% ethanol fraction.

The compound 10 exhibited a minimum MIC value (39.06 μg/mL) against S. epidermidis followed by B. cereus and S. aureus with MIC value of 156.25 μg/mL for both bacterial strains. In case of compound 9 the MIC value was a little higher for most of the bacterial strains, i.e., (312.5 μg/mL) as shown in Table 5.

MMC values of the pure compounds of active fraction 20% ethanol against reference bacterial strains were shown in Table 6. Although results of MIC and MMC showed variations among the tested bacterial strains but both the values of MIC and MMC were equivalent with each other in most of the cases. Compound 10 showed the least value of MMC (78.25 μg/mL) against S. epidermidis followed by B. cereus with MMC value of 625 μg/mL. Whereas, compound 9 showed its antibacterial potential against all the tested microbial species at a constant MMC value of 312.5 μg/mL.