J. R.
Lawrence
*a,
G. D. W.
Swerhone
a,
J. J.
Dynes
b,
A. P.
Hitchcock
c and
D. R.
Korber
d
aEnvironment Canada, 11 Innovation Blvd., Saskatoon, SK S7N 3H5, Canada. E-mail: John.Lawrence2@canada.ca
bCanadian Light Source, University of Saskatchewan, 44 Innovation Blvd., Saskatoon, SK S7N 5A8, Canada
cBrockhouse Institute for Materials Research, McMaster University, Hamilton, ON L8S 4M1, Canada
dFood and Bioproduct Sciences, University of Saskatchewan, 51 Campus Drive, Saskatoon, SK S7N 5A8, Canada
First published on 14th December 2015
Little is known about the fate of carbon nanotubes in aquatic environments. To investigate their interactions with dissolved organics and microorganisms, carbon nanotubes (CNTs) were exposed to natural river water in microcosms for 50 days. CNTs and developed biofilms were recovered and examined using scanning transmission X-ray microscopy (STXM). CNTs underwent extensive aggregation, forming bundles and flocs. CNTs were also integrated into developing biofilm and either became associated with the biology or acted as a scaffold for growth. STXM examination revealed the development of an extensive complex coating consisting of lipids, proteins, polysaccharides and carbonates. Single walled-CNT had a significantly greater affinity, on a per-unit material basis, for protein (2× more) and polysaccharides (10× more) relative to multi walled-CNTs. These bio-molecular coatings constitute a highly-modified surface chemistry influencing a range of functional properties, including the toxicity of these nanomaterials. Reactive oxygen species (ROS) are linked to CNTs toxicity; therefore, we compared ROS production in bacteria by “as-manufactured” with that by biofilm-coated CNTs. Fluorescence confocal laser microscopy using carboxy-H2DCFDA demonstrated a significant reduction in ROS production in bacteria exposed to coated CNTs. CNT-based toxicity will likely be rapidly attenuated when these materials enter aquatic environments. These observations contribute to understanding the fate and potential effects of CNTs in natural systems, and confirm the need to evaluate the impact of complex bio-molecular coatings on environmental effects of nanomaterials.
Nano impactThe environmental impact of nanoparticles in aquatic environments may be greatly influenced by adsorbed organic molecules or macromolecules on the NP surface. The interactions with microbial biofilm matrix components such as proteins, polysaccharides, and lipids are anticipated to impact bioavailability and toxicity of nanoparticles. This study demonstrates that nanoparticles such as carbon nanotubes undergo extensive material specific interactions including the development of complex coatings featuring polysaccharides, proteins, lipids and minerals. Furthermore, we show that these changes impact the toxicity of carbon nanotubes through reduction of reactive oxygen species production. Therefore, our study highlights the importance of examining micro and nanoscale events in understanding the fate and effects of these materials in the environment. |
Sacchetti et al.19 have detected the coating of CNTs by organic and inorganic materials in the environment, as well as by plasma proteins in blood. Microscopy-based approaches are critical to achieving the goals of detection and assessment of fate for a broad range of nanomaterials. Scanning transmission X-ray microscopy (STXM)20 is a technique that can be used effectively to assess interactions of nanomaterials and their fate in complex systems such as microbial biofilms. As we have demonstrated, STXM can provide detailed in situ information on the biochemistry, structure and composition of biofilms, bioflocs and their associated materials in hydrated environments at a minimum resolution scale of ∼30 nm.21–25 It is essential to study the behaviour of NPs in the context of complex microbial communities and at the appropriate scale. STXM combines the chemical sensitivity of NEXAFS spectroscopy at the C-1s edge with high spatial resolution (30 nm or better) and has been proven to be capable of mapping the biochemical composition of bacteria and biofilms at a subcellular scale, as well as the speciation of metals.13,22–24 Thus, STXM is well suited to test hypotheses regarding the fate of nanomaterials in the environment.
Carbon nanomaterials have been shown to inhibit the growth and activity of a range of microorganisms in pure culture and mixed microbial community exposures, including bacteria and protozoa.11,26–29 A major mechanism for toxicity of carbon-based nanomaterials is the generation of reactive oxygen species (ROS) that can damage cells.30,31 These may be generated by exposure to sunlight, or even by ozonation and ultrasonication.3 Some have suggested that physical damage may also occur via direct physical puncturing by multi-walled carbon nanotubes (MWCNTs) and single-walled carbon nanotubes (SWCNTs). However, Krishnamoorthy et al.31 demonstrated that this mechanism was not significant relative to ROS production. It has also been noted that the release of toxic metals from the catalyst materials used to synthesize CNTs may account for toxicity in some studies.32,33
We have focused on the fate and effects of two main types of CNTs: i) SWCNTs, which are single-layered graphitic cylinders having diameters on the order of a few nanometers, and ii) MWCNTs, which contain between 2 and 30 concentric cylinders with outer diameters commonly between 30 and 50 nm. CNT lengths vary substantially and often range from 100 nanometers to 10 or more micrometers.3 Following exposure to natural river water and developing complex microbial communities, the corona/coating on SWCNTs and MWCNTs was analysed by STXM at the C 1s edge, and ROS production of conditioned and “as-manufactured” CNTs was compared using a fluorescence-based CLSM assay.
Parameter | Spring | Summer |
---|---|---|
Conductivity (μmhos cm−1) | 451 | 429 |
pH | 8.13 | 8.46 |
Turbidity (NTU) | 2.7 | 5 |
Ammonia (mg N l−1) | 0.04 | 0.03 |
Nitrate–nitrite (mg N l−1) | 0.75 | 0.31 |
Orthophosphate (mg P l−1) | 0.01 | 0.01 |
Dissolved organic carbon (mg C l−1) | 3.5 | 3.0 |
Total suspended solids (mg l−1) | 1 | 1 |
The reactors were maintained at 21 ± 2 °C. River water was pumped through the reactors at a rate of 500 ml per day (one reactor volume) by using a multichannel peristaltic pump (Watson Marlow, Wilmington, MA). Treatments involved the addition of 500 μg L−1 of MWCNTs or SWCNTs to reactor influent waters during the biofilm development period. Microcosms receiving 0.1 mg L−1 CNTs were also operated to create biofilms with sufficient CNT material for the reactive oxygen species production assay (see below). In addition, control reactors were operated that received river water alone. Biofilms were grown under treatment and control conditions on polycarbonate coupons in replicate/triplicate bioreactors over a period of 50 days, at which time coupons were removed for immediate analysis. After incubation, the biofilm (∼40 μm thick) was aseptically scraped from the surface with a sterile silicone spatula and placed in a sterile 1 mL centrifuge tube. No homogenization of the sample was undertaken, as the goal was to look at as intact a biofilm as possible.
STXM at the C 1s edge was performed on the spectromicroscopy (SM) beamline 10ID-1 at the Canadian Light Source (CLS), Saskatoon, SK, Canada.38 The beamline was operated at an energy resolving power of E/ΔE ≥ 3000. All samples were analyzed in 1/3 atmosphere of He. After each analytical measurement, an image was recorded at 289 eV, an energy which readily visualizes radiation damage to polysaccharides, the most easily damaged chemical component. The extra-cellular matrix polysaccharide signal was reduced by less than 20% as a consequence of beam damage in the worst case of the measurements reported. The microscope energy scale was calibrated to an accuracy of ±0.05 eV using sharp gas phase signals, typically the Rydberg peaks of CO2. STXM was used analytically by measuring image sequences at specific energies39 or from image difference maps, which are the differences between on- and off-resonance images. Representative absorption spectra for the target species (MWCNT, SWCNT, protein, lipid, exopolysaccharide, CaCO3) were obtained by placing these species on Si3N4 windows and performing C 1s scans using STXM at the SM beamline. Results of these spectral scans are shown in Fig. 2A and B.
The as-measured transmitted X-ray signals were converted to optical densities using the incident flux measured in the Si3N4 window devoid of sample to correct for absorbance by the window. In analytical mode STXM was used to collect spectro-microscopic datasets consisting of thousands of spectra representing the sum of the contributions of all species present. This data was then converted to quantitative component maps of for example, protein, lipid, polysaccharide, CO3 and CNTs derived from image sequences measured between 280 eV and 320 eV by using singular value decomposition to fit the spectrum at each pixel to a linear combination of reference spectra (Fig. 2A and B) recorded from pure materials of the components suspected to be present. The analysis methods have been described extensively elsewhere.20 Data analysis was performed using aXis2000.40 A total of twenty one locations (n = 21) were quantitatively analysed for CNT, protein, lipid, polysaccharide and calcium carbonate content. Analyses of variance (ANOVA) as well as principle components analyses (PCA) were used to assess significant differences between treatments (p < 0.05).
To date most research has used extracted NOM for environmental studies12 or EPS from pure cultures,45 rather than the complex array of macromolecules present in natural systems. Limited characterization of the protein corona formed on CNTs following exposure to fetal bovine serum and human blood plasma has been reported based on extractive methods.19,47,48 However, Lowry et al.12 noted that there are an “endless number of biomacromolecules” that may interact with nanomaterials in the environment. These include proteins, polysaccharides, peptidoglycans, lipids, humic/fluvic acids, and a wide range of poorly defined products of the decomposition of bacteria, algae, and plants in the aquatic environment.12,49,50
The review of Petersen and Henry51 indicates that there are a range of methods that may allow detection of fullerenes and CNTs, including electron energy loss spectroscopy, coherent anti-stokes Raman scattering microscopy (CARS), UV/vis spectroscopy, which may be interfered with by the presence of other biomolecules such as lipids. Fourier-transformed infra-red (FTIR) spectroscopy may be effective if the spectra are unique and Raman spectroscopy can be used to qualitatively detect CNTs in bio-matrices. Infra-red (IR) spectroscopy has also been used to detect coatings on nanomaterials, although the resolution of the technique is a limitation.52 Transmission electron microscopy is highly effective for characterization of nanomaterials, although detection of the associated corona may present a challenge due to its transparency. Thus, it appears there is a need for direct analyses of naturally-developed nanoscale coatings or coronas associated with nanomaterials. STXM provides a useful analytical and imaging technique for environmental samples, including biofilms and flocs.52,53 The resolution, sample thickness (up to 10 μm hydrated material), use of the X-ray absorption properties of the sample (imaging/spectromicroscopy), identity and mapping of metals, macromolecules, biomacromolecules (polymers, proteins, lipids), speciation of metals, reduced radiation damage relative to electron microscopy21,23,54 all make STXM a valuable but under-utilized tool for the study of nanomaterials.13
Initially, we determined C 1s absorption spectra for the target CNT species under investigation, following the approach described previously.21,22 The results indicated well-defined spectra similar to those reported elsewhere25 for MWCNTs and SWCNTs, allowing them to be detected in a complex sample and differentiated from associated species detected at the C 1s edge, including proteins, exopolysaccharides, lipids, etc. (Fig. 2A) as well as other carbon nanotubes such as those modified by carboxylation or hydroxylation21 (Fig. 2B).21 Furthermore, examination of “as-manufactured” CNT’s suspended in deionized water as reference materials did not reveal the presence of any organic coatings. Following these preliminary studies, we sought to map the presence of the target CNTs when they had been exposed to complex microbial communities and unfiltered natural river waters.
Fig. 4 and 5 show large-scale C 1s maps of exposed/coated MWCNTs and SWCNTs, respectively. The CNTs were detected in the context of complex microbial biofilms and organic coatings. Given the concentration upon addition, the CNTs likely underwent extensive homoaggregation, as well as heteroaggregation, before undergoing integration into the biofilm communities. At expected environmental concentrations, heteroaggregation would likely be the more dominant process, leading to greater interactions with dissolved organic matter and suspended solids. This may favour integration into microbial biofilms and bioaggregates. Each CNT exhibited a unique general distribution pattern with the MWCNT forming a meshwork (Fig. 4) upon which biofilm and organics accumulated versus the SWCNT which tended to occur as coatings on biofilm organisms and materials (Fig. 5). These differences may reflect the differences in size and possibly the stronger tendency for homoaggregation of the SWCNTs. All of these events would likely limit transportation/redistribution of CNTs in aquatic environments but may enhance their potential for entering the food web.
The morphology of the community is evident in the optical density component of the data stack, showing algae (diatoms, non-carbon = blue, Fig. 4), bacterial cells/colonies, while the chemistry of the macromolecular structures, including MWCNTs, SWCNTs, protein, polysaccharide, lipid, and calcium carbonate, are mapped. High-resolution mapping based on fitting the reference X-ray absorption spectra (Fig. 2) to the image sequences to identify pixels with matching spectra was carried out within defined sub-regions of the large-scale maps to enable analyses of all components. STXM analyses of biofilms confirmed the tendency for different CNTs to accumulate a halo, or corona, of various macromolecules including lipids, EPS and proteins. The results of analysis of multiple representative stacks taken at biofilm locations are summarized in Table 2, showing significant differences (p < 0.05, ANOVA) in the quantitative amounts of protein, lipid, polysaccharide and calcium carbonate detected in association with MWCNTs or SWCNTs. The differences in composition of the associated corona are also evident in the proportional graph (Fig. 6A) which illustrates the differences in amounts of the major macromolecules associated with each type of CNT clearly showing the increase in polysaccharide contribution in the SWCNT. The results of principle component analyses (PCA) of the data sets (Fig. 6B) show the nature of the drivers of the differences in the two CNT-exposed biofilm systems, illustrating the importance of polysaccharide and protein components in differentiating the coatings on SWCNTs from those on MWCNTs. It is also apparent that the quantity of polysaccharide detected associated with SWCNT is variable as indicated by the scatter in these points (Fig. 6B) and the variance (Table 2).
MWCNT | 27 ± 4aa | SWCNT | 12 ± 4b |
---|---|---|---|
a Thickness in nanometers. Values followed by the same letter are not significantly different p < 0.05 ANOVA. n = 21 (##) = nm per unit of MWCNT or SWCNT. | |||
Protein | 31 ± 12a (1.2 ± 0.6b) | 28 ± 6a | (2.7 ± 1.4a) |
Lipid | 7 ± 10a (0.26 ± 0.38a) | 2 ± 3a | (0.14 ± 0.19a) |
Polysaccharide | 8 ± 10b (0.37 ± 0.52b) | 43 ± 6a | (3.9 ± 1.3a) |
CaCO3 | 27 ± 9a (0.99 ± 0.27a) | 15 ± 10b | (1.6 ± 1.4a) |
The STXM analyses suggest that both types of CNTs undergo relatively extensive conditioning, with the development of a complex macromolecular coating or corona. Of particular interest is that although the CNTs are exposed to identical conditions, the resultant coating is, in fact, significantly different (p < 0.05 ANOVA), presumably dictated by the underlying surface chemistry and various reactive mechanisms, including electrostatic, hydrogen bonding, and hydrophobic interactions. MWCNTs and SWCNTs accumulated similar amounts of protein (Table 2) but on a per-unit CNT basis SWCNTs had twice as much material. The amounts of lipid, and calcium carbonate accumulated by each CNT on a per-unit basis were very similar but SWCNTs had significantly (p < 0.05) more protein. The major difference was the apparently-significant high affinity of SWCNTs for polysaccharides where there was a more than 10× greater accumulation (Table 2).
These coatings presumably dominate the properties of the modified CNT, altering composition, size, electrophoretic mobility and aggregation.11 The coatings may also control dissolution and short-range toxicity mechanisms, such as ROS production and potential palatability and entry into the food web. This is particularly true when interactions with flocs and biofilms are considered, since these represent foci for biogeochemical activity and a major source of carbon and energy in aquatic systems.17,55 Although Kang et al.11 reported that coating with NOM did not influence bacterial cytotoxicity of SWCNTs, the presence of coatings, in general, reduced the toxicity of nanomaterials and may reduce ROS production.3 Krishnamoorthy et al.31 demonstrated that the major source of bacterial toxicity for CNTs was generation of ROS by CNTs exposed to light. We hypothesized that a reduction in ROS production would occur for SWCNTs and MWCNTs after incubation with river water and during biofilm formation. To assess this, we used a well-defined, biofilm-forming organism Pseudomonas fluorescens strain 840406E (ref. 41) as a reporter strain in the ROS assay. After staining P. fluorescens cells with carboxy-H2DCFDA they were placed in contact with “as-manufactured” and biofilm-coated SWCNTs and MWCNTs. A coupon was created where contact between test material and stained cells were ensured, and the system observed with CLSM. The results (Fig. 7) clearly indicate that the short range ROS activity was attenuated since cells directly associated with the aged CNT masses did not fluoresce, indicating no ROS-induced stress. In contrast, the “as-manufactured” CNTs generated an ROS-positive signal.
Further, when the common inducer of ROS production, tert-butyl hydroperoxide, was applied (Fig. 3), a ROS-positive fluorescence signal was generated. Although there was evidence that bundling of CNTs may result in internal attenuation of ROS production to some degree, the similarity of sample preparations would mitigate against this mechanism. These observations confirm that short-range ROS generation is attenuated by these organic coatings. Similarly, Liu et al.56 indicated that exposure to ambient urban air resulted in organic coatings that significantly decreased cytotoxicity of SWCNTs. Metal contamination has also been proposed as a mechanism for bacterial cytotoxicity of CNTs;44,51 however, metal edge scans (data not shown) using STXM did not reveal detectable concentrations of nickel or other metals, and neither did ICP-MS analyses (data not shown) of water and biofilm materials. Thus, close CNT contact appeared to mediate generation of ROS within our test system. It is important to note that we have only examined two of a wide variety of CNTs, and that the properties of CNTs can vary considerably with even batch-to-batch variability in physiochemical properties, implying that while valid in the context of this study results may not be generalized to all CNTs.
This journal is © The Royal Society of Chemistry 2016 |