A clinical isolate of NDM-1 E. coli strain used in this study was a generous gift from Dr. David M. Livermore, Health Protection Agency Centre for Infections, London, United Kingdom (Kumarasamy et al., 2010). Plasmid DNA was isolated according to the protocol described by Sambrook et al. (1989) and a region of NDM-1 gene was amplified using NDM-1 specific primers F-(5′-GGGCAGTCGCTTCCAACGGT-3′) and R-(5′-GTAGTGCTCAGTGTCGGCAT-3′) (Yong et al., 2009; Manchanda et al., 2011). 16S rDNA was amplified by colony PCR using the universal forward and reverse primers (5′-GAGTTTGATCATGGCTCAG-3′) and (5′-CTACGGCTACCTTGTTACG-3′), respectively (Grifoni et al., 1995). Sequencing of 16S rRNA and NDM-1 gene was done for confirmation of the strain. New Delhi metallo-β-lactamase 1 E. coli resistance to different antibiotics was determined according to the performance standards for antimicrobial disk susceptibility testing given by Clinical and Laboratory Standards Institute for E. coli (Clinical and Laboratory Standards Institute, 2014). The presence of MBL activity was tested using the double disk synergy test in the presence of ethylenediaminetetraacetic acid (EDTA) as the chelating agent and meropenem as the β-lactam antibiotic as previously described by Arakawa et al. (2000).

Two hundred and forty species from 183 genera and 75 families, listed as medicinal plants by the National Medicinal Plants Board of India, were selected. Leaf samples were collected from different parts of Southern India including Karnataka, Kerala, and Tamil Nadu. Voucher specimens from all the collections were identified using local floras and were deposited in the SRM University Herbarium (Supplementary Table S1).

Fresh mature leaves were thoroughly rinsed with running water followed by 70% ethanol, dried, and stored at room temperature. The dried leaves were ground to a powder in a laboratory blender, labeled, and stored in air tight containers at room temperature until use. The powder (10 g) was mixed with 100 ml of ethanol, and the mixture was macerated in a shaker incubator at 37°C and 150°g for 24 h. This process was repeated twice, and the macerated samples were filtered through Whatman No. 1 filter paper. Residual solvent from the extracts was removed using a rotary evaporator under reduced pressure, and stored at 4°C.

The plant extracts were dissolved in dimethyl sulfoxide (DMSO), and screened for bactericidal activity against an NDM-1 E. coli strain. The cells from the agar slants were recovered by inoculating into nutrient broth containing 100 μg/ml cefotaxime, for the maintenance of plasmid-borne NDM-1 gene (Kumarasamy et al., 2010). The culture was incubated at 37°C and 150 g until its growth reached 0.5 McFarland standard (10⁸ cfu/ml). Plant extract was added to yield final concentrations of 5–25 mg/ml (in 5 mg increments) and inoculated with 100 μl of NDM-1 E. coli culture. Dimethyl sulfoxide (4%) was used as a negative control. The cultures were grown in a shaker incubator at 37°C and 150 g for 24 h. The cultured cells were streaked on MHA plates, incubated at 37°C for 16 h, and observed for bacterial growth.

Minimum inhibitory concentration (MIC) was determined using the Resazurin Microplate Assay as described by Palomino et al. (2002) with minor modifications. Stock solutions of extracts were prepared in DMSO and NDM-1 E. coli cells were grown to 0.5 McFarland standard. Serial dilutions were made by microbroth dilution technique, and the total volume was brought up to 100 μl with Muller Hinton broth. Colistin and DMSO were used as positive and negative controls, respectively. Sealed microtiter plates were incubated overnight at 37°C. After the incubation period, 3.0 μl of 0.03% resazurin was added to the wells as an indicator of cell viability, and then incubated at 37°C for 4 h.

Antibacterial activity of plant extracts was determined by measuring the zones of inhibition using the disk diffusion method (Bauer et al., 1966). Bacterial cells were grown to 0.5 McFarland standard, swabbed on Mueller Hinton Agar (MHA) media, and allowed to dry for 10 min. The disks loaded with 5–20 mg (in 5 mg increments) of plant extracts were then placed on the MHA plates. The disks loaded with colistin (10 μg/disk) and DMSO (20 μl/disk) were used as positive and negative controls, respectively. The culture plates were incubated at 37°C for 24 h, and the zones of inhibition were measured in triplicate for each plant extract.

Synergy testing of plant extracts and three antibiotics was performed in 96-well microtiter plates by following the checkerboard method (Bayer and Morrison, 1984). New Delhi metallo-β-lactamase 1 E. coli cells were inoculated in nutrient broth and incubated at 37°C for 16 h. Combinations of plant extract (10–5120 μg/ml) with colistin (0.5–8 μg/ml), meropenem (0.5–32 μg/ml), and tetracycline (1–16 μg/ml) were tested. Interactions between the antibacterial agents were determined by calculating the fractional inhibitory concentration (ΣFIC) index:

where MIC E+D is the MIC of extract in combination with antibiotic and MIC D+E is the MIC of antibiotic in combination with extract. Based on the ΣFIC, interaction between the antibiotic and the extract was inferred to have a synergistic (ΣFIC < 0.5), additive (0.5 > ΣFIC < 1), indifferent (1 > ΣFIC < 4), or antagonist (ΣFIC > 4) effects (dos Santos et al., 2015).

The NDM-1 enzyme inhibition assay was carried out using nitrocefin as a substrate as described by King et al. (2014) with minor modifications. Nitrocefin and the plant extracts were dissolved in DMSO. In a microplate, 5 nM recombinant NDM-1 enzyme (Cusa Bio, China) supplemented with 10 μM ZnSO₄ was incubated with plant extract (0.02–5.12 ng/μl) or 10 μl of DMSO as negative control for 10 min. After the incubation, nitrocefin was added to a final concentration of 60 μM, and absorbance was measured at 490 nm and 30°C temperature using a multi-mode reader (BioTek Synergy, United States). The percentage inhibition was estimated, and IC50 values were determined for each extract.

Membrane integrity of the NDM-1 E. coli cells after treatment with plant extracts was analyzed as described by Kennedy et al. (2011) using the LIVE/DEAD^® BacLight^TM Kit (Thermo Scientific, United States). Flow cytometry was optimized by preparing a standard by mixing live and dead cells in phosphate buffer. Bacterial cell density in the mid-exponential growth phase was adjusted to 1 × 10⁸ cfu/ml, and treated with 2× MIC of plant extracts at 37°C for 1 h. The treated bacterial cells were incubated with 5 μM SYTO 9 in the dark for 15 min, and propidium iodide (PI) was added to a final concentration of 30 μM. Colistin (4 μg/ml) and DMSO (4%) were used as positive and negative controls, respectively. The cells were analyzed in a flow cytometer and the signals were captured using FL1 and FL3 channels (BD FACS Calibur, United States).

Morphological changes in the NDM-1 E. coli cells after treatment with plant extracts were observed following the method described by Hartmann et al. (2010) using a Vega 3 scanning electron microscope (SEM; Tescan, United States). Cells grown in nutrient broth with or without DMSO were used as negative controls and colistin as a positive control. Bacterial cell density in the mid-exponential growth phase was adjusted to 1 × 10⁸ cfu/ml and treated with 2× MIC of plant extracts for 1 and 4 h at 37°C. The cells were centrifuged and the pellet was resuspended in 100 μl of phosphate buffer (pH 7.0). The cell suspension was spread on a glass slide, and fixed using 2.5% glutaraldehyde. The fixed cells were serially washed in ethanol ranging from 10 to 90%, dried, and observed under the microscope.

Qualitative analysis of the phytochemicals present in the plant extracts was carried out following standard protocols (Harborne, 1973). Quantitative assessment of total glycosides, phenolic compounds, saponins, steroids, flavonoids, alkaloids, and terpenoids was carried following the protocols described by Masuko et al. (2005), Zhang et al. (2007), Xu and Chang (2009), Daksha et al. (2010), Herald et al. (2012), Li et al. (2014), and Lin et al. (2015), respectively.

The NDM-1 E. coli IR-6 strain used in this study was confirmed by DNA sequencing, and the antibiotic susceptibility and double disk synergy tests. The size of NDM-1 gene amplified from plasmid DNA was 475 bp and that of the 16S rDNA amplified from genomic DNA was 1500 bp. BLAST analysis of 16S rDNA showed 100% identity with E. coli, and that of NDM-1 gene showed 100% identity with the NDM-1 gene reported from many bacteria including E. coli (Acc. No. CP021210.1). Antibiotic susceptibility test was conducted using different classes of antibiotics, which included aminoglycosides, cephalosporins, fluoroquinolones, β-lactams, extended-spectrum β-lactams, polymyxins, and carbapenems. The NDM-1 E. coli IR-6 strain was susceptible to all the antibiotics tested, except colistin (Table 1). There were no zones of inhibition for most of the antibiotics tested indicating high levels of resistance against these antibiotics. In the double disk synergy test, the zone of inhibition for meropenem was 12.6 ± 0.57 mm, and this zone increased to 24.3 ± 1 mm in the presence of EDTA due to chelation of metal ions, which indicated that this strain harbors MBL.

Ethanol extract from the leaves of 240 medicinal plants was tested against NDM-1 E. coli at concentrations 5–25 mg/ml (in 5 mg increments) to identify extracts that have antibacterial activity. The estimation of minimum bactericidal concentration (MBC) showed antibacterial activity for only 12 medicinal plant extracts against NDM-1 E. coli. Among them, the extracts from six plants, which showed MBCs between 5 and 15 mg/ml were taken for further studies. Minimum bactericidal concentration for the extract from Hibiscus cannabinus (HC) and Tamarindus indica (TI) was 5 mg/ml, Combretum albidum (CA) and Hibiscus acetosella (HA) was 10 mg/ml, and Hibiscus furcatus (HF) and Punica granatum (PG) was 15 mg/ml (Table 2).

Minimum inhibitory concentrations of the plant extracts against NDM-1 E. coli were estimated to be 2.56 mg/ml for CA, HA, HC, and TI, and 5.12 mg/ml for HF and PG. Antimicrobial activity of the extract was also determined by measuring the zone of inhibition of bacterial growth around the disks that were loaded with plant extracts at concentrations of 5–20 mg/disk. Zones of inhibition were observed against NDM-1 E. coli at all the concentrations of the plant extracts tested. At the lowest concentration of plant extract tested (5 mg/disk), the HC plant extract showed the highest zone of inhibition (11.6 ± 0.57 mm) followed by the plant extracts of TI (10.3 ± 0.57 mm), PG (9.6 ± 0.57 mm), CA (8.3 ± 0.57 mm), HF (7.6 ± 0.57 mm), and HA (7.3 ± 0.57 mm). Hibiscus cannabinus also exhibited the highest zone of inhibition at all the concentrations. Colistin (10 μg/disk), which was used as a positive control, showed a zone of inhibition of 10.8 ± 0.57 mm, and DMSO (20 μl/disk), which was used as a negative control, did not show any zone of inhibition (Figure 1).

Phytochemical analysis of these extracts showed variations in the quantity of total phenolic compounds, terpenoids, alkaloids, glycosides, steroids, flavonoids, and saponins (Table 3). All of the extracts contained very low (<1.0 μg/ml) to low quantity (1.0–3.0 μg/ml) of steroids and saponins. While CA, HA, HC, and HF contained flavonoids as the major secondary metabolite (7.34 ± 0.02 to 9.69 ± 0.25 μg/ml), PG contained phenolic compounds as the major secondary metabolite (11.02 ± 0.49). Notably, TI contained very low (<1.0 μg/ml) or low quantity (1.0–3.0 μg/ml) of all secondary metabolites that were tested.

Combinatory effects were determined by calculating the ΣFIC index for colistin, meropenem, and tetracycline with all six plant extracts. Different concentrations of antibiotics and plant extracts were combined to check for synergistic activity. The observed ΣFIC values when the plant extracts were combined with colistin, meropenem, and tetracycline antibiotics were 0.12–0.31, 0.09–0.31, and 0.12–0.37, respectively. This indicated that the plant extracts have synergistic effects against NDM-1 E. coli when combined with all of the three tested antibiotics (ΣFIC ≤ 0.5). Hibiscus cannabinus showed consistently lower ΣFIC, and the smallest ΣFIC was observed when it was combined with meropenem (0.09). Punica granatum showed consistently higher ΣFIC, and the highest ΣFIC was observed when it was combined with tetracycline (0.37). In terms of reduction in MIC of antibiotics when combined with plant extracts, we have observed between 4- and 16-fold reduction for individual combinations. The highest reduction in MIC against NDM-1 E. coli was observed in combinations of colistin with CA and HC; meropenem with HA, HC, and HF; and tetracycline with HC and TI (Table 4).

The ability of the six plant extracts to inhibit the activity of NDM-1 enzyme was tested in vitro using nitrocefin as a chromogenic substrate. The NDM-1 enzyme hydrolyzes the nitrocefin to form a red colored product. Recombinant NDM-1 enzyme activity in the reaction, which contained only nitrocefin in DMSO, was considered 100%. Treatment with the plant extract reduced the activity of NDM-1 enzyme in a time-dependent manner. When treated with 5.12 ng/μl plant extract for 4 h, the highest percentage of enzyme inhibition was observed with HC (77%), followed by CA (67%), PG (59%), HF (57%), TI (53%), and HA (40%). The calculated IC50 in terms of nanograms per microliter was the lowest with HC (0.50), followed by CA (0.73), PG (0.76), HF (0.77), TI (0.78), and HA (1.2).

Integrity of the NDM-1 E. coli cells after treatment with the plant extracts was studied using the LIVE/DEAD permeability assay and SEM analysis. A large number of NDM-1 E. coli cells were found to be dead after the 4 h treatment period, which showed that the plant extracts were highly effective against NDM-1 E. coli. All the extracts caused an increase in cell permeability as indicated by PI uptake. Viability of NDM-1 E. coli cells was initially 88.3% with DMSO (negative control), which was drastically reduced to as low as 0.3% when treated with plant extracts. This was lower than the viability of the cells treated with colistin, which was 1.8%. The order of plant extracts based on the viability of NDM-1 E. coli cells after the 4 h treatment was TI < HC < HF < CA < PG < HA (Figures 2A,B). The percentage of cell death ranged between 79.4 and 89.7% (HA and TI, respectively) when treated with the plant extracts. In the SEM analysis, untreated NDM-1 E. coli cells displayed smooth and intact cell surfaces, while the cells treated with the plant extracts exhibited corrugated, wrinkled, shrunken, and deformed surface morphologies (Figure 3).