To investigate inactivation effect and mechanism of zinc oxide nanoparticles (ZnO NPs) activity against Campylobacter jejuni biofilms.
To investigate inactivation effect and mechanism of zinc oxide nanoparticles (ZnO NPs) activity against Campylobacter jejuni biofilms.
ZnO NPs with concentrations of 0, 0·6, 1·2 and 6 mmol l−1 were employed in antimicrobial tests against Camp. jejuni planktonic cells and biofilms. Campylobacter jejuni sessile cells in biofilms were more resistant to a low concentration of ZnO NPs when compared to planktonic cells. The ZnO NPs penetrated the extracellular polymeric substance (EPS) without damage to the EPS and directly interacted with the sessile bacterial cells, as determined using infrared spectroscopy and scanning electron microscopy. Raman spectroscopy shows alterations in quinone structures and damage to nucleic acids following Camp. jejuni treatment with ZnO NPs. The mechanism of DNA damage is most likely due to the generation of reactive oxygen species (ROS). Spectroscopic-based partial least squares regression (PLSR) models could predict the number of surviving sessile cell numbers within a bacterial biofilm (≥log 4 CFU, root mean square error of estimation <0·36) from Fourier transform infrared (FT-IR) spectral measurements.
ZnO NPs were found to have antimicrobial activity against Camp. jejuni biofilms. ZnO NPs penetrated the biofilm EPS within 1 h without damaging it and interacted directly with sessile cells in biofilms. Alterations in the DNA/RNA bases, which are owing to the generation of ROS, appear to result in Camp. jejuni cell death.
ZnO NPs may offer a realistic strategy to eliminate Camp. jejuni biofilms in the environment.
Metal oxide nanoparticles (NPs) are <100 nm in one dimension, have a higher surface to volume ratio compared with conventional metal oxide particles and are stable under high temperature and pressure (Stoimenov et al. 2002). Zinc oxide (ZnO) NPs can inactivate a wide range of both Gram-positive and Gram-negative bacteria cells, including Listeria monocytogenes, Staphylococcus aureus, Campylobacter jejuni, Escherichia coli O157:H7 and Salmonella spp. (Jones et al. 2008; Liu et al. 2009; Xie et al. 2011). Importantly, ZnO NPs have minimal effect on human cells (Reddy et al. 2007). Owing to their antimicrobial activity, ZnO NPs are used to extend the shelf life of fresh produce and meat (Stoimenov et al. 2002).
Campylobacter jejuni is a foodborne pathogen that is responsible for approximately 1·4 million cases of human illness (campylobacteriosis) in the United States annually (Young et al. 2007). Although campylobacteriosis is generally classified as a self-limiting disease, it can result in life-threatening sequelae, including Guillain-Barré syndrome (Young et al. 2007). The paradox of Camp. jejuni is that the microbe is extremely prevalent in the environment – however, its microaerobic growth requirements mean that the organism does not multiply in the natural aerobic environment. Under these conditions, Camp. jejuni may maintain itself within a biofilm microenvironment. A biofilm is an assemblage of micro-organisms in which cells are associated with each other and adhere to a surface (Joshua et al. 2006). Even though bacteria can exist as separate cells, most bacteria aggregate and grow in biofilms in nature (Davies 2003; Stewart and Franklin 2008). Those adherent sessile cells are embedded within a matrix called extracellular polymeric substance (EPS). The EPS comprises up to 90% of the particulate fraction of the biofilm and is composed of polysaccharides, proteins, nucleic acids, lipids and humic-like substances. The EPS matrix mediates cell/cell communication (i.e. quorum sensing), keeps the microbes within the biofilm hydrated and protects against environmental stress and antimicrobial agents (Davies 2003).
Infrared and Raman spectroscopies provide a method to examine the biochemical composition of microbial cells (Jarvis and Goodacre 2004) and biofilms (Ivleva et al. 2008, 2010). Unique and wide spectral patterns are generated across specific wavenumbers to identify and differentiate microbial cells, particularly when infrared and Raman spectroscopies are coupled together (Lu et al. 2011a,b,c). Bacterial cell growth and various metabolic characteristics can be verified within 6 h by Raman spectroscopy and 8 h by infrared spectroscopy (Lu et al. 2011c).
A previous study showed that ZnO NPs have a significant antimicrobial effect against Camp. jejuni planktonic cells (Xie et al. 2011). The investigators demonstrated that the antimicrobial mechanism of ZnO NPs was owing to disruption of the cell membrane and oxidative stress.
The sessile cells within a biofilm are afforded greater protection from antibiotics than planktonic cells owing to the protective nature of the EPS (Davies 2003). Moreover, the EPS matrices of bacterial biofilms have been shown to impede the transport of selective metal oxide NPs (Peulen and Wilkinson 2011). To date, little research has been conducted on Camp. jejuni biofilms, specifically how metal oxide NPs interact with Camp. jejuni biofilm EPS and subsequently inactivate sessile cells. We examined the mode of ZnO NPs action against Camp. jejuni biofilms using the complementary biophysical techniques of infrared and Raman spectroscopies. These two spectroscopic methods are noninvasive, allowing direct examination of the EPS and sessile cells. To our knowledge, this is the first study to examine the interaction between metal oxide NPs and Camp. jejuni biofilms.
ZnO NPs with size of 70 ± 15 nm were purchased from Alfa Aesar (Ward Hill, MA, USA). The original ZnO NP suspension (12 mol l−1) was diluted with sterile deionized water or broth to make a series of concentrations from 0 to 120 mmol l−1. Further, an aliquot (20 ml) of ZnO NPs suspension was filtered through an aluminium oxide membrane filter (20 nm pore size, 25 mm optical density) (Anodisc; Whatman Inc., Clifton, NJ, USA), resulting in a NPs-free solution containing surfactants and water (control) (Liu et al. 2009; He et al. 2011).
Campylobacter jejuni strains F38011 (a human clinical isolate) and NCTC 11168 (a sequenced strain) were cultured in a microaerobic environment (85% N2, 10% CO2 and 5% O2) at 37°C on Mueller-Hinton agar plates supplemented with 5% citrated bovine blood (MHB agar plates) or in MH broth with constant shaking. The Camp. jejuni cultures were passaged onto fresh media every 24–48 h.
Overnight cultures of Camp. jejuni isolates were diluted to an OD540 of 0·03 (c. 107 CFU), and 100 μl of the diluted suspension was added to the surface of sterile cellulose-nitrate membrane (0·45 μm pore size, 47 mm diameter) on MHB agar plate and incubated in a microaerobic environment at 37°C (Lu et al. 2012). The membrane was aseptically transferred to a new MHB agar plate every 24 h for up to 3 days to form a discernible and uniform Camp. jejuni biofilm with the area of approximately 1 cm2.
For planktonic cell experiments, a mixed culture of the Camp. jejuni F38011 and NCTC 11168 strains were prepared by combining equal volumes of each culture growing at the late log phase. The bacterial mixtures in MH broth were then treated with different concentrations (0, 0·6, 1·2 and 6 mmol l−1) of ZnO NPs for 0, 1, 3, 5, 7 and 10 h with constant shaking in a microaerobic environment at 37°C to determine the antibacterial effect (n = 3). At each sampling time, a sample of the treated bacterial mixture was serially diluted and spread onto the surface of MHB agar plate (antibiotic free) in duplicate for viable bacteria enumeration. For the biofilm experiments, each biofilm sample was treated with concentrations (0, 0·6, 1·2 and 6 mmol l−1) of ZnO NPs for same treatment times as listed previously. Campylobacter jejuni biofilm coated on cellulose-nitrate membrane were aseptically removed from MHB agar plate, placed into 20 ml MH broth with the concentrations of ZnO NPs and incubated with gentle shaking in a microaerobic environment at 37°C. At the end of each treatment time, the Camp. jejuni biofilm was rinsed three times with phosphate buffer saline (PBS) followed by a treatment with a solution of 0·1% trypsin (25 ml) for 15 min statically at room temperature. Preliminary experiments indicate that a 0·1% trypsin treatment could detach biofilm from cellulose-nitrate membrane and maintain bacterial sessile cell viability (data not shown). Following incubation, the trypsin solution was recovered, serially diluted and plated onto MHB agar plates for bacteria enumeration.
The minimum inhibitory concentrations (MICs) of ZnO NPs against Camp. jejuni planktonic cells were determined using a broth microdilution method (Xie et al. 2011). Briefly, serial two-fold dilutions of ZnO NPs ranging from 0·12 to 12 mmol l−1 were prepared in a 96-well microtiter plate using MH broth. Freshly grown bacterial cells (log phase) were inoculated into each well to give a final concentration of c. 105 CFU ml−1. After the inocula were incubated in a microaerobic environment for 24 h at 37°C, the growth of the bacteria in each well were monitored and compared with that of the positive control (a well containing no ZnO NPs) by measuring the absorbance at OD540. The MIC was recorded as the lowest concentration of ZnO NPs that completely inhibited bacterial cell growth.
Fourier transform infrared (FT-IR) spectra (range of 4002–399 cm−1) were collected using a Nicolet 380 FT-IR spectrometer (Thermo Electron Inc., San Jose, CA, USA). The cellulose-nitrate membrane covered with Camp. jejuni biofilm was placed in direct contact with the diamond crystal cell (30 000–200 cm−1) of an attenuated total reflectance (ATR) cell. Determination of spectral interference from cellulose-nitrate membranes has been previously studied (Wang et al. 2011) and can be avoided with proper experimental technique and data preprocessing tools. Thirty-two consecutive scans were averaged with an individual spectrum collection time of 32 s. FT-IR spectral resolution was 4 cm−1. Eight spectra were individually collected from different locations of a biofilm for each treatment (n = 3).
Raman spectroscopic analyses were performed using a WITec alpha300 Raman microscope (WITec, Ulm, Germany) equipped with a UHTS-300 spectrometer and a 532 nm laser delivering ~2 mW on samples via a 100× objective (Nikon, Melville, NY, USA). Raman scattering spectral features were collected by a 1600-by-200-pixel charge-coupled-device (CCD) detector (pixel size was 16 × 16 μm). WITec Control v1.5 software (WITec) was employed for instrument control, data collection and export. Collection of Raman spectra was performed over a simultaneous wavenumber shift range of 3700–200 cm−1 in the extended mode. For measurement at a single location, each full spectral collection was conducted with a 3-s integration time with 20 spectral accumulations (total integration time 60 s).
Klarite™ (Renishaw Diagnostics, Glasgow, UK) surface enhanced Raman scattering (SERS)-active substrates were used for Camp. jejuni planktonic cells to enhance the intensities of Raman signals (Lu et al. 2011a,b). Treated and untreated planktonic Camp. jejuni cells (10 μl) were deposited onto the substrate, and Raman measurements were collected after 1-h drying under a fume hood at room temperature. For the biofilm experiments, the cellulose-nitrate membrane coated with bacterial biofilm was directly put under the microscope for focus adjustment and employed for Raman scattering spectral collection. Eight spectra were collected for each filter (n = 3).
Full area raster scans were performed to create Raman maps, with single spectrum integration times of 3 s, saving 600 spectra acquired over a regularly spaced array of sample locations in a grid pattern (30 by 20 arrays). As for image resolution, the numerical aperture was 0·95, which left a spatial resolution of approximately 345 nm. A computer-controlled xyz motorized stage was employed during Raman mapping. This allowed scanning of different vertical and horizontal layers of the sample. Subsequent band analysis for the collected spectra was applied to generate intensity-correlated maps for relevant Raman bands (Ivleva et al. 2010). Thus, chemical images corresponding to optical images of the biofilm were generated based upon the area beneath the baseline-corrected specific bands. To distinguish different biochemical components of the biofilm in a single map, the selective images generated by the fit procedure were illustrated with different colours and combined (overlapped) in one image (n = 10).
Infrared and Raman spectra were initially preprocessed by OMNIC (Thermo Electron, Inc., Lafayette, CO, USA). Automatic baseline correction was employed to flatten the baseline, followed by smoothing using a Gaussian function with a bandwidth of 9·462 cm−1 (Lu et al. 2011c). The height and area of spectral bands were measured and calculated by OMNIC and Matlab. Second-derivative transforms using a 9-point Savitzky-Golay filter and wavelet transforms with a scale of 7 were performed in Matlab (Lu et al. 2011a). The reproducibility of the vibrational spectra was investigated by calculating Dy1y2 (Lu et al. 2011a,b).
Partial least squares regression (PLSR), a supervised chemometric model, was established based upon processed Raman spectra (noise removed) and employed for quantitative analysis in Matlab. Fourteen spectra of each sample were used to establish the calibration model and six spectra were used for model cross-validation. The other four spectra were employed for prediction. This selection and validation process was repeated three times independently, and the average relevant values derived from PLSR chemometric model were determined. The suitability of the developed PLSR models for predicting live Camp. jejuni sessile cell numbers in biofilm was assessed by regression coefficient (R), latent variables, the root mean square error of estimation (RMSEE), the root mean square error of cross-validation (RMSECV) and the residual prediction deviation (RPD) values (Lu et al. 2011a,b,c).
Scanning electron microscopy (SEM) was performed to examine changes of Camp. jejuni biofilm untreated and treated with 1·2 mmol l−1 ZnO NPs for 6 h in a microaerobic environment at 37°C. The Camp. jejuni biofilm was fixed with 2% glutaraldehyde and 2% paraformaldehyde in 0·1 mol l−1 phosphate buffer overnight at 4°C. The samples were then washed with PBS, postfixed in 1% osmium tetroxide for 2 h at 20°C and rinsed twice with 0·1 mol l−1 buffer prior to dehydration in an ethanol series. They were then incubated with hexamethyldisilizane (HMDS) overnight and air-dried. The samples were mounted onto SEM stubs and sputter coated with a thin layer of gold. The coated samples were observed under an FEI Quanta 200F scanning electron microscope using an accelerating voltage of 30 kV.
Each experiment was repeated a minimum of three times to ensure reproducibility in triplicate. The significant difference (P < 0·05) between band heights of raw and second-derivative transformed spectra was determined by one-way analysis of variance (anova) following t-test in Matlab.
The MIC of ZnO NPs against Camp. jejuni planktonic cells was determined to be between 0·6 and 0·3 mmol l−1. As shown in Fig. 1(a), a 4-log10 reduction was achieved after Camp. jejuni planktonic cultures were treated with 0·6 mmol l−1 ZnO NPs for 10 h. A Camp. jejuni planktonic culture was eliminated after exposure to 1·2 and 6 mmol l−1 ZnO NPs for 10 and 5 h, respectively. The sessile cells within the Camp. jejuni biofilm exhibited similar susceptibility as planktonic cells at 1·2 and 6 mmol l−1 ZnO NP treatments with inactivation of 9-log10 cells observed after 10 and 5 h, respectively (Fig. 1b). However, the Camp. jejuni biofilm was significantly (P < 0·05) more resistant at lower concentrations (0·6 mmol l−1 ZnO NPs) compared with planktonic counterparts (Fig. 1a), with only 1-log10 reduction being observed after 10-h treatment.
Figure 2 shows a single combined map from selective images generated by the fit procedure as illustrated using different colours for band intensities at monosaccharide/polysaccharide (868 cm−1), protein secondary structure (1302 cm−1) and DNA/RNA (1320 cm−1) collected from an area of 15 μm × 10 μm. Acquisition times of 60 s for a single spectrum and spectral region from 1800 to 400 cm−1 were employed. This Raman map visualizes the distribution of monosaccharide/polysaccharide, protein secondary structure and DNA/RNA in the upper biofilm layer showing that the biofilm surface is heterogeneous.
The reproducibility of Raman spectra from independent experiments and various sample locations were calculated using Pearson coefficient (D value). The D value for Raman spectra was 7·95 ± 1·36 to 9·53 ± 1·89 using wavenumbers from 1800 to 400 cm−1 after 72 h of cultivation on MHB agar plates, demonstrating good reproducibility (Lu et al. 2011a,b).
In preliminary studies, we found that FT-IR spectral features of planktonic cells and biofilms are distinctly different at wavenumbers 1280, 1162 and 829 cm−1. These three bands represent the functional groups of EPS proteins and polysaccharides (X. Lu and M. E. Konkel, unpublished data). Here we used FT-IR spectroscopy to monitor the variations in the biofilm EPS following treatment of Camp. jejuni biofilms with ZnO NPs (0·6, 1·2 and 6 mmol l−1). We found that treatment of the Camp. jejuni biofilms with ZnO NPs did not result in readily discernible change in the EPS, as demonstrated by no significant changes in the intensities of the 1280, 1162 and 829 cm−1 bands (Fig. 3). The integrity of the biofilm EPS remained intact over the entire course of the 10-h experiment. In contrast, the Camp. jejuni sessile cells in the biofilm began to die following 1 h of treatment with ZnO NPs and were eradicated at 10 h by 1·2 and 6 mmol l−1 ZnO NPs treatments (Fig. 1).
We used SEM to examine the biofilm for potential structural alterations following treatment with ZnO NPs (Fig. 4). Compared with untreated biofilms (Fig. 4a,d), no structural abnormalities were observed in the EPS of the biofilms treated with 1·2 mmol l−1 ZnO NPs for 6 h (Fig. 4b,c). However, the SEM analysis revealed that treatment of the biofilms with ZnO NPs resulted in the sessile cells undergoing morphological alterations to a coccoid shape (Fig. 4c,e) when compared to sessile cells to a rod shape of the untreated biofilm (Fig. 4d). Direct attachment of ZnO NPs on sessile cell membrane was observed (Fig. 4e,f). These results together demonstrated that ZnO NPs can rapidly penetrate through Camp. jejuni biofilm EPS with little adsorption by chemical components within EPS resulting in cell injury and death.
The interactions between ZnO NPs and Camp. jejuni planktonic cells and/or sessile cells in biofilm were examined using second-derivative transformed Raman spectra (Fig. 5). Significant band variations (P < 0·05) were calculated and evaluated using Matlab to study biochemical variations on the bacterial cell membrane. Raman scanning was initially established (negative 5 μm on z-axis) to collect spectra directly from sessile cells to avoid any interference from intact EPS. Raman spectral features of EPS are different from those of the sessile cells (data not shown), indicating that Raman scattering signals were derived from the sessile cells and not the EPS. For both planktonic cells and sessile cells in biofilm, several of the same band variations appeared. The band at 760 cm−1 was from amino components (Movasaghi et al. 2007). The band at 1130 cm−1 indicated phospholipid structural changes (Lu et al. 2011c). The band at 1323 cm−1 was from guanine (Naumann 2001).
A unique spectral feature of the cells is with the band at 1593 cm−1 and it was assigned to C=N and C=C stretching in quinoid ring (Movasaghi et al. 2007). The band at 1373 cm−1 was assigned to ring breathing modes of DNA/RNA base (Movasaghi et al. 2007). The variation of this band was not shown in Fig. 5(b), which clearly indicated that DNA/RNA base is the major target for ZnO NPs to perform bactericidal effect while sessile cells were more resistant to the treatment of low concentration of ZnO NPs (0·6 mmol l−1). This validates the resistance of biofilm sessile cells to low concentration of ZnO NPs (0·6 mmol l−1) in Fig. 1(b). The band at 1300 cm−1 was related to CH2 twisting of lipids (Naumann 2001). The intensity of this band decreased when the sessile cells were treated with low concentration (0·6 mmol l−1) of ZnO NPs (Fig. 5b) and increased when the sessile cells were treated with high concentration (6 mmol l−1) of ZnO NPs (Fig. 5a).
To determine whether alterations in DNA structures were owing to a direct interaction between DNA and ZnO NPs or to an indirect interaction between DNA and radical oxygen species (ROS) generated by ZnO NPs, the following experiments were performed (Fig. 6a). Briefly, overnight Camp. jejuni planktonic culture (~109 CFU ml−1) was treated with 100 μg ml−1 erythromycin for 10 h following an additional 7-h treatment of 6 mmol l−1 ZnO NPs in a microaerobic environment at 37°C. The purpose of treating the bacterial planktonic cells with erythromycin prior to treatment with ZnO NPs was to kill all of the bacteria, thereby preventing the generation of reactive oxygen species (ROS) from cell metabolism. Erythromycin does not alter the structure of a bacterial cell (Alfredson and Korolik 2007).
Raman spectroscopy was then employed to determine the variations of DNA bases. DNA bases showed significant (P < 0·05) variation in this Raman spectral region (1373 cm−1) when Camp. jejuni planktonic cells were treated with ZnO NPs for 7 h compared with nontreated planktonic cells. We treated the Camp. jejuni planktonic cells with 6 mmol l−1 of ZnO NPs to ensure that the viable bacteria would be killed by the ZnO NPs (see Fig. 1a). Interestingly, there was no significant (P > 0·05) structural variation in the DNA bases when Camp. jejuni planktonic cells were treated with erythromycin following ZnO NPs treatment (Fig. 6b), suggesting that these changes in spectral features reflect oxidative damage to nucleic acids. In addition, we treated Camp. jejuni chromosomal DNA with 6 mmol l−1 ZnO NPs in a microaerobic environment at 37°C for 2 h and applied FT-IR spectroscopy to determine whether structural variations occurred in the DNA. No significant (P < 0·05) variation in the DNA spectral features was observed following treatment (data not shown). This indicates that there is no direct interaction between Camp. jejuni DNA and ZnO NPs and that alteration in nucleic acid is from other effects, such as ROS generation by ZnO NPs. Taken together, these results support the hypothesis that ZnO NPs alter the structure of quinones, resulting in ROS generation and the damage of Camp. jejuni DNA.
The number of viable sessile bacteria was predicted within a biofilm following the treatment with ZnO NPs by developing a chemometric model to correlate Raman spectral features with bacterial cell survival numbers. The PLSR model using a wavenumber below 1800 cm−1 as x and a loading plot for differentiation between the Camp. jejuni strains of injured levels as y-axis was performed for Raman spectroscopy. All the parameters related to this supervised chemometric model are summarized in Table 1. Each PLSR model was constructed using 70% of the values for model calibration and the remaining 30% for model validation and prediction. Figure 7 shows a representative PLSR model predicting the survival of Camp. jejuni sessile cell numbers in biofilm. PLSR models provided consistent model behaviour and good prediction ability on the basis of R, RPD and RMSEE (Table 1) (Lu et al. 2011a,b,c). The validated PLSR model could be applied to monitor the levels of viable cells in a Camp. jejuni biofilm during the course of antimicrobial agent treatment using a noninvasive method.
|Range||No. of samples||Latent variables||R-cal||RMSEE-cal||RPD-cal||R-val||RMSEE-val||RPD-val|
In the present study, the antimicrobial effect of ZnO NPs on both Camp. jejuni planktonic cells and sessile cells in biofilms was studied. Researchers have previously examined suspensions of ZnO NPs by transmission electron microscopy and found that they are 70 ± 15 nm and have either rod or spherical morphology (Liu et al. 2009; He et al. 2011). Liu et al. (2009) demonstrated that the ZnO NPs were uniformly dispersed with little agglomeration in aqueous systems containing surfactants. In preliminary work, we determined that the ZnO NP-free solution (containing only surfactants) had no effect on Camp. jejuni growth (data not shown). Based on published work and our preliminary studies, which provided the foundation of our experimental approach, we examined the antimicrobial effect of ZnO NPs on Camp. jejuni planktonic cells and sessile cells in biofilms.
Campylobacter jejuni biofilm was more resistant to lower concentration (0·6 mmol l−1) ZnO NPs compared with planktonic counterparts. Xie et al. (2011) found that Camp. jejuni planktonic cells were sensitive to the treatment of ZnO NPs and determined the MIC to be 0·05–0·025 mg ml−1 (equivalent to 0·6–0·3 mmol l−1). The investigators also observed changes in cell morphology following treatment with ZnO NPs, and concluded that these changes were associated with membrane damage and oxidative injury. In other studies, gold (Zhao et al. 2010) and silver (Choi et al. 2010; You et al. 2011) NPs also showed antimicrobial effect against Gram-negative bacteria planktonic cells, such as E. coli O157:H7 and Pseudomonas aeruginosa. Further, Pseudomonas putida, E. coli and Staphylococcus aureus biofilms were reported to show susceptibility to nanosilvers (Fabrega et al. 2009; Choi et al. 2010; Seil and Webster 2011). Hetrick et al. (2009) validated the anti-biofilm efficacy of nitric oxide-releasing silica NPs. Recently, ZnO NPs were reported to show antifungal effect against Botrytis cinerea and Penicillium expansum (He et al. 2011) and antibacterial effect against E. coli O157:H7 planktonic cells (Liu et al. 2009).
To evaluate how sessile cell inactivation may be occurring, spectroscopic studies were conducted. Spectral features of sessile and planktonic cells and biofilm EPS were compared.
Campylobacter jejuni planktonic cell FT-IR and Raman spectral features were determined with band assignments selected from earlier studies (Moen et al. 2005; Lu et al. 2011b). Here, the distribution of different chemical components on Camp. jejuni biofilm EPS were obtained using confocal Raman mapping (Fig. 2). Raman maps of the relevant band intensities correlated with optical images of biofilms showing the distribution of saccharides, proteins and nucleic acids in the upper biofilm EPS layer.
High dimensional Raman scattering spectra contains much useful information, but interference (e.g., noise and fluorescence) must be reduced using various data preprocessing techniques such as binning, smoothing and second-derivative transformations prior to chemometric analyses (Bocklitz et al. 2011). Wavenumber selection and cultivation time are critical to obtaining reproducible Raman spectra for biofilms, similar to what has been previously observed for bacterial planktonic cells (Moen et al. 2005; Lu et al. 2011a,b). In the present study, we chose the wavenumbers from 1800 to 400 cm−1 and 72-h cultivation as the parameters as these provided good reproducibility for biofilm Raman spectra. Physiological heterogeneity in biofilm is ubiquitous (Stewart and Franklin 2008) but we validated that the heterogeneity did not significantly (P > 0·05) affect biofilm Raman spectral reproducibility, making it possible to quantify treatment differences. High reproducibility is critical for the development of reliable chemometric analyses (i.e. PLSR) (Moen et al. 2005).
We employed FT-IR spectroscopy and SEM to validate the integrity of Camp. jejuni biofilm EPS by ZnO NPs treatment on the basis of chemical and morphological profiles. ZnO NPs could rapidly (within an hour) penetrate through Camp. jejuni biofilm EPS without damage and directly inactivate sessile bacterial cells. Antimicrobial agents, including antibiotics (ciprofloxacin and erythromycin) and plant-derived bioactive compounds (organosulfur compounds and polyphenolic antioxidants) typically destroy the biofilm EPS before reaching the sessile cells. It takes many antimicrobial agents several hours to gain access to the sessile cells because of the need to breakdown the biofilm EPS (Lu et al. 2012). Ciprofloxacin transport to the biofilm-substratum interface is naturally impeded owing to the biofilm EPS itself in a flowing system (Suci et al. 1994). Similarly, slow diffusion of piperacillin into Ps. aeruginosa biofilms has also been observed (Hoyle et al. 1992).
The mode of action of ZnO NPs is likely different from antibiotics, as the integrity of EPS is maintained in ZnO NPs treatment during which time the sessile cells are injured or killed. This indicates that ZnO NPs may be more effective antimicrobial because of their ability to penetrate the biofilm EPS. Recently, fluorescence correlation spectroscopy was used to investigate the diffusion of selective NPs into bacterial biofilms, including biofilms composed of Lactococcus lactis, Pseudomonas fluorescens and Streptococcus mutants (Habimana et al. 2011; Peulen and Wilkinson 2011). The diffusion coefficients relate to biofilm surface properties, radius of NPs and charges of NPs (Zhang et al. 2011). Further, rapid penetration of NPs through biofilm EPS was found in these studies. In the present study, we show that ZnO NPs penetrate through Camp. jejuni biofilm within an hour leaving the EPS intact using noninvasive FT-IR spectroscopy measurements and confirmed by SEM.
Second-derivative transformed Raman spectra were studied to analyse the interaction modes of ZnO NPs with planktonic cells and sessile cells in biofilms (Fig. 5). Variations in the biochemical components within bacterial cells were observed, including changes to phospholipid and protein secondary structure. Two critical findings relating to bacterial survival and inactivation are associated with quinone derivatives (1593 cm−1) and DNA/RNA bases (1373 cm−1).
Quinone derivatives are constituents of biologically relevant molecules and serve as electron acceptors in electron transport chains such as those in photosystems I & II of photosynthesis and aerobic respiration, which perform a critical role in maintaining bacterial survival, including for the microaerophile Campylobacter spp. (Georgellis et al. 2001). Importantly, quinones are highly redox active molecules that can cycle with their conjugate semiquinone and hydroquinone radicals, leading to the formation of ROS, including superoxide and hydrogen peroxide (Bolton et al. 2000; Yu et al. 2002). The variations of Raman spectral bands derived from quinoid compounds (1373 cm−1) may indicate the importance of quinones for maintenance of Camp. jejuni cell (both planktonic and sessile forms) viability. A previous study reported that the expression levels of two oxidative stress genes (i.e. katA and ahpC) in Camp. jejuni planktonic cells significantly increased in response to ZnO NPs treatment (Xie et al. 2011). While our results are consistent with the notion that oxidative stress is responsible for the antibacterial effect of ZnO NPs, we further show changes in the structure of quinones (Fig. 5).
Variations of DNA/RNA bases appeared except when sessile bacteria in biofilms were treated with low concentration of ZnO NPs (0·6 mmol l−1), confirming resistance of sessile cells in biofilms to inactivation at this ZnO NPs concentration (Fig. 1b). Taken together, it indicated DNA/RNA base is the primary target for ZnO NPs providing bactericidal effect. Further, we demonstrated that the variation of nucleic acid base is owing to ROS generation by ZnO NPs rather than direct interaction between ZnO NPs and nucleic acids. The production of ROS in bacterial cells owing to metal oxide NPs have been proposed in previous studies (Stoimenov et al. 2002). Lipovsky et al. (2011) demonstrated that ZnO NPs can generate ROS, including hydroxyl radicals and singlet oxygen, which cause fungal cell death. Recently, Premanathan et al. (2011) displayed that the mechanisms underlying the toxicity of ZnO NPs on Gram-negative bacteria (E. coli and Ps. aeruginosa) is the involvement of the generation of ROS. Raghupathi et al. (2011) employed Northern analyses of various ROS specific genes and confocal microscopy to demonstrate that the antibacterial activity of ZnO NPs might involve both the production of ROS and the accumulation of NPs in the cytoplasm or on the outer membranes. Applerot et al. (2009) employed electron-spin resonance measurements and revealed that aqueous suspensions of small NPs of ZnO could produce increased levels of ROS, namely hydroxyl radicals. In addition, a remarkable enhancement of the oxidative stress was detected following the ZnO NPs antibacterial treatment (Applerot et al. 2009).
Raman spectroscopic-based PLSR models were established and validated to predict actual survival numbers of sessile bacterial cells in biofilms (Fig. 7 and Table 1) during the course of ZnO NPs treatment. This may provide a novel method to noninvasively and accurately determine the effectiveness of antimicrobial agents.
Antimicrobial activity of ZnO NPs against Camp. jejuni sessile cells within a biofilm was demonstrated and inactivation mechanisms were proposed. We found that Camp. jejuni sessile cells were more resistant to low concentrations (0·6 mmol l−1) of ZnO NPs compared with their planktonic counterparts. ZnO NPs rapidly penetrated the biofilm EPS without damaging the EPS and interacted directly with sessile cells in a biofilm. Quinone derivatives play a critical role in aerobic respiration and the structure of these were altered in both planktonic cells and sessile cells in biofilm with the treatment of ZnO NPs. Alterations in the DNA/RNA bases, which are owing to the generation of ROS, appear to result in Camp. jejuni cell death. Raman spectroscopic-based PLSR models can predict actual survival numbers of sessile bacterial cells in biofilms, resulting in accurately determining the effectiveness of antimicrobial agents.
We thank Dr Valerie Jean Lynch-Holm who aided us with electron microscope work in the Franceschi Microscopy and Imaging Center at Washington State University, Pullman. This study was supported from funds awarded to M.E.K. from the NIH (R56 AI088518-01A1) and funds awarded to B.A.R. from National Institute of Food and Agriculture (AFRI 2011-68003-20096). S.W. is supported from funds from the China ‘863’ Program, Ministry of Science and Technology, project no. 2011AA100806.