Influence of CNTRENE^® C100LM carbon nanotube material on the growth and regulation of Escherichia coli

Bacterial growth and media

Escherichia coli K12 strain SMG 123 (ATCC PTA-7555) was grown in lysogeny broth (LB) or M9 minimal salts medium with the addition of 1 mM thiamine and 2% glucose (hereafter M9 medium) at 37 °C and 200 rpm in a Thermo Scientific MaxQ 400 incubator unless stated otherwise.

Carbon nanotubes

Pristine carboxyl functionalized CNTRENE^® C100LM CNT material, supplied by Brewer Science, Inc., was suspended in distilled water at 135 µg/mL (pH 7.0) and were of the same lot as previously used by Woodman et al. (2016), wherein the physical characterization was described. Briefly, the CNTs had a total metal ion content of less than 25 ppb. Carboxyl functionalization was estimated to be at 2–6% and mainly observed at the open ends of the CNTs. The CNTs had a length range of 0.3–1.5 µm (90% of CNTs) and a diameter range of 0.7–3 nm (95% of CNTs), with the average length of 0.87 µm and width of 1.56 nm. The pristine product was made up of SWCNT, DWCNT, and MWCNT at 70%, 25%, and 5%, respectively.

Carbon nanotube aging process and spectroscopic characterization

To simulate environmental weathering, pristine CNTRENE material was aged as supplied in distilled water in a QUV Accelerated Weathering chamber (Q-Lab Corp, Cleveland, OH, USA) as previously described (Woodman et al., 2016). Briefly, 14 mL pristine CNTRENE material was exposed to 3–4 h alternating ultraviolet and condensation cycles for 12 days. The UV cycle and condensation cycle temperatures were 68 ± 0.5 °C and 47 ± 0.5 °C, respectively. The current of the lamps were 0.5–0.6 amperes and the condensation cooling fan set point was 15. Distilled water was used to provide moisture and served as temperature control bath. The Raman, FTIR, and UV-vis spectra of both pristine and aged CNT material were quantified after sonication for one minute to disperse the CNTs. For Raman spectroscopy, 2–10 µL of the sample was dropped onto an aluminum foil wrapped around a microscope glass slide. The sample was air-dried in a clean environment free from any dust or other contaminants. A Horiba LabRAM HR800 spectrometer equipped with a 50 mW 532 nm excitation laser with a detection capability in the range of 200 to 4,000 nm was used for Raman Photoluminescence spectroscopy at ambient temperature. For FTIR spectroscopy, 20 µL of the CNTs were put into separate pre-cleaned dry mortars preheated at 80 °C. The sample was dried on the mortar and approximately 400 µg of dry potassium bromide was added and ground into fine powder to make a KBr pellet. The pellet was used for IR measurements using a Bruker IR spectrophotometer. The background data was obtained using a KBr pellet without CNTs. For UV-Vis analysis, CNTs were diluted 10-fold with deionized water and spectra taken with a PerkinElmer Lambda 650 UV-Vis spectrometer at room temperature.

Minimal Inhibitory Concentration (MIC)

A broth microdilution MIC assay on a 96-well transparent C-bottom plate was performed with a standard inoculum of 5.0 × 10⁵ CFU/mL in a final reaction volume of 200 µL as previously described (Wiegand, Hilpert & Hancock, 2008). The MIC was determined for CNT concentrations from 1.05 µg/ml to 6.44 × 10⁻⁵ µg/ml and were assayed using a minimum of three replicates from three independent cultures (n = 9). Plates were incubated at 37 °C in a BioTek EL808 plate reader with shaking at the medium shake rate and optical density at 595 nm was monitored for 24 h. The Gen5 software (BioTek, Winooski, VT, USA) was used for data collection and growth curves and doubling times were used to evaluate growth effects of CNT exposure.

Antibacterial plate counts

The effect of pristine CNTRENE material concentrations greater than 5 µg/mL on the growth of E. coli K12 was evaluated by the spot-plate technique using a modified method of (Gaudy Jr, Abu-Niaaj & Gaudy, 1963). E. coli K12 was inoculated to 5.0 × 10⁵ CFU/ml in 96-well transparent C-bottom plates and grown in LB (pH 7) with addition of pristine CNTRENE material (0 µg/mL, 8.44 µg/mL, 16.88 µg/mL, and 33.75 µg/mL) in a final volume of 200 µL in triplicate. Plates were incubated for 24 h as described above. After incubation, serial ten-fold dilutions of the overnight cultures were performed in 96-well plates and 10 µl of the dilutions were spotted in triplicate on LB plates and incubated at 37 °C for 16–18 h. Colony forming units per milliliter were calculated for each concentration of CNTs and compared to that of the unexposed control group.

Electron microscopy

Morphological change of bacterial cells exposed to CNTs was evaluated by scanning electron microscopy (SEM) and atomic force microscopy (AFM). E. coli K12 was inoculated to a concentration of 5.0 × 10⁵ CFU/ml in a 200 µl LB culture in a 96-well transparent C-bottom plate containing 0 µg/mL, 33.75 µg/mL, 16.88 µg/mL, and 1.05 µg/mL pristine CNTRENE material. E. coli K12 was grown at each CNT concentration in triplicate for 24 h at 37 °C. Replicates were combined, pelleted by centrifugation at 5,000 x g for 5 min, and then washed three times in 0.1 M phosphate buffer (pH 7.2) to remove growth medium. Cells were dehydrated by an ascending ethanol wash series (50%, 70%, 80%, 90%, 95%, 100%, and 100%) with a 5 min exposure to each concentration of ethanol. Samples were transferred onto silicon wafers in 10 µL volumes and either frozen in liquid nitrogen and freeze dried overnight for SEM or allowed to air dry overnight for AFM. SEM was done with a JEOL JSM-7600F field emission scanning electron microscope under vacuum (9.6 × 10⁻⁵ Pascal). SEM images were captured using an accelerating voltage of 1.00 kV and a working distance (WD) between 5.1 mm and 5.2 mm at total magnification ranging from 10,000× to 20,000×. AFM imagines were captured utilizing the Veeco Dimension 3100 with a Nanoscope IIIA Controller under atmospheric conditions (1.01 × 10⁵ Pascal). AFM imagines were captured on tapping mode (scan size 5.000 µm or 20.0 µm, scan rate 1.001 Hz, 512 samples) using a silicon tip with a nominal radius of 8.0 nm.

RNA sequencing

Differential gene expression of CNTRENE material-exposed bacterial cells was evaluated with RNA sequencing. An overnight E. coli K12 culture was inoculated 1:100 in 5 mL of M9 medium. Pristine CNTRENE material was added at 1.05 µg/mL to experimental cultures, with all cultures set up in triplicate and incubated to mid-log phase. RNA was extracted using the Qiagen RNeasy Plus Mini Kit (Qiagen, Hilden, Germany), including an on-column DNase treatment (Qiagen RNase-Free DNase Set) according to the manufacturer’s instructions.

Extracted RNA was initially quantified using an IMPLEN nanophotometer followed by analysis with an Agilent Technologies 2100 Bioanalyzer system and 2100 Expert software to confirm RNA quality. High integrity samples (RIN > 8) were depleted of ribosomal rRNA using the bacterial Ribo-Zero rRNA Removal Kit (Illumina, San Diego, Ca, USA) according to the manufacturer’s instructions. The SMARTer Stranded RNA-Seq kit (Clontech, Mountain View, Ca, USA) was used to prepare cDNA from RNA samples according to the manufacturer’s instructions. An Illumina MiSeq instrument (single-end 50 bp read length) was used for RNA sequencing and sequencing data was analyzed using the DNAstar Lasergene Suite Qseq software by the University of Wisconsin- Milwaukee Great Lakes Genomics Center. Sequencing was done from three independent RNA preparations for each sample type. The sequence reads were mapped and analyzed using the RNA-Seq pipeline default parameters using E. coli K12 MG1655 as a reference genome. Differential gene expression was analyzed using the student t-test with the false discovery rate restricted to 0.05 as the p-value adjustment method (Benjamini & Hochberg, 1995).

Data analysis

All statistical analyses were done in using the GraphPad Prism 5.0 software.

Effect of aging on CNTRENE^® C100LM CNT material

Exposure to environmental conditions is known to change the physiochemical properties of CNTs, which has the potential to influence toxicity (Valsami-Jones & Lynch, 2015). To mimic the effect of weathering on CNTs, the pristine CNTRENE material was aged by exposure to UVA at 340 nm using a QUV Accelerated Weathering chamber that has been used previously to simulate outdoor weathering. Accelerated aging of materials in the QUV Accelerated Weathering chamber for between 1,000 h–1,800 h has been equated to a year of Florida summer sunlight exposure (University of Delaware, CfCM, 2002) and has been used to investigate the stability of polymers such as polymer-bound hindered amine light stabilizers (Macleay, 1989), to predict the service life of exterior automotive coatings (Shanbhag, 2012), the photodegradation of wood-plastic composites (Peng et al., 2014), and carbon fiber-reinforced polymer composites (Tcherbi-Narteh, Hosur & Jeelani, 2013), among others. The UV-Vis spectra of pristine CNTRENE material had a peak at about 250 nm that is due to the first interband transition of the nanotubes (Fig. 1). As the nanotubes age, the above peak appears to experience a redshift to 270 nm. Additionally, another peak emerges at about 210 nm. This is attributed to the loss of π-structure of the nanotubes in the aging process. In the Raman spectra, the major observation is the difference in the ratios of characteristic CNT bands (Fig. 2). In general, the aged CNTRENE material exhibited lower D/G and RBM/G band ratios compared to pristine CNTs. The decreasing ratio of bands, especially RBM/G, implies diminishing CNT character as the CNTs age. Previous research on aging of nanocarbons under ambient conditions indicated a decrease in the net structural defects with aging (Yang et al., 2009). This is in line with the results observed from the decrease in the intensity of the D band compared to the G band, leading to a decrease in the D/G ratio giving an indication of decreased disorder or defects in the CNTs with aging. From the FTIR spectra, it can be inferred that there is an O–H peak (∼3,430 cm⁻¹), C–H stretching as observed in alkanes (and possibly aldehydes or carboxylic acids), and a carbonyl stretch at 1,653 cm⁻¹–1,701 cm⁻¹ confirmed the presence of a carboxylic acid group (–COOH) (Chen et al., 1998; Liao & Zhang, 2012) (Fig. 3). Deviations were observed for the intensities of the C =O (with a likelihood of H-bonding as that observed for diketones) and O–H peaks for the aged CNTs being higher than those of pristine CNTs. It can be further inferred that the diketone functionalities are dominant in the aged CNTs compared to the pristine CNTs based on the intensities of the peaks. The band at ∼1,653 cm⁻¹ may be attributed to C=C stretching (Chang et al., 2006). The peaks at 1,385 cm⁻¹ and 1,090 cm⁻¹ correspond to the expected C–O–H and the C–C–C bending. The peaks at ∼2,920 cm⁻¹ suggest the presence of a C–H stretch in C–CH₃ that can be attributed to protonation of the CNTs as a result of their interactions with water in extreme aging conditions. It is possible that such interactions could occur in the environment because of exposure to favorable reaction conditions such as humidity, presence of hydrogen, oxygen, and heat. The differences in peak intensities suggest probable oxidation of the CNTs with aging, especially evident from the increase in O–H groups. Additionally, aged CNTs had increased coiling and bundling compared to pristine CNTs (Fig. 4). Together these data imply morphological and functional changes as the CNTs are aged in conditions that mimic prolonged environmental exposure.

Cytotoxicity of CNTRENE^® C100LM CNT material exposure

The inhibition and cytotoxicity of an increasing gradient of CNTRENE material up to 1.05 µg/ml on E. coli K12 was examined by a MIC assay. Growth curves for E. coli K12 in LB (pH 7) in the presence of pristine CNTRENE material over the tested concentration range were very similar to the growth observed in the unexposed control group, with all growth curves overlaying (Fig. 5A). The same trend was observed for E. coli K12 in LB (pH 7) in the presence of aged CNTRENE material (Fig. 5B). Cultures grown in LB (pH 7) had anaverage doubling time of 22.4 min (±1.2 min) after exposure to pristine CNTs, and an average doubling time of 22.3 min (±1.1 min) after exposure to aged CNTs (Fig. 6A). In comparison, the unexposed control groups in the pristine and aged CNT assays had doubling times of 22.7 min (±0.8 min). Doubling times observed for E. coli K12 in LB (pH 7) were similar between pristine and aged CNTRENE material concentrations regardless of the concentration used. Indeed, doubling times were not correlated to increasing exposure to either pristine (r = 0.0995, n = 9 for each concentration, p = 0.4873) or aged (r = 0.2581, n = 9 for each concentration, p = 0.0675) CNTRENE material.

The influence of CNTRENE material on E. coli growth was analyzed in M9 medium, in which E. coli have a slower growth rate and a lower final cell density. In this minimal medium, growth with pristine or aged CNTRENE material was similar to the unexposed control group (Figs. 5C–5D). Unexposed cells had an average doubling time of 60.7 min (±0.7 min). In comparison, E. coli exposed to pristine CNTRENE material had an average doubling time of 60.4 min (±2.3 min), and when exposed to aged CNTRENE material an average doubling time of 59.9 (±0.7 min) was observed (Fig. 6B). Considering these data, it is not surprising that there was not a correlation between doubling times and the concentration of pristine (r = 0.0109, n = 9 for each concentration, p = 0.9414) or aged (r =  − 0.2688, n = 9 for each concentration, p = 0.1381) CNTRENE material exposure.

Lastly, the effect of pH on CNTRENE material toxicity was examined using LB medium at pH of 5. As for the other conditions, the growth of E. coli was not inhibited by either pristine or aged CNTRENE material exposure, having growth curves that superimposed (Figs. 5E–5F). For pristine CNTRENE material treatment, an average doubling time of 30.3 min (±0.8 min) was observed. Similarly, aged CNTRENE material treatment produced an average doubling time of 29.6 (±0.9 min) upon exposure of up to 1.05 µg/ml CNTRENE material. These doubling times are similar to that of the untreated cells of 30.3 ± 0.5 min (Fig. 6C). Again, no correlation was seen between doubling time and treatment with increasing concentrations of pristine (r =  − 0.1103, n = 9 for each concentration, p = 0.5490) or aged (r = 0.2342, n = 9 for each concentration, p = 0.1896) CNTRENE material.

Antibacterial plate counts

Significant optical interference was observed at CNTRENE material concentrations over 1.05 µg/ml, making MIC microtiter assays infeasible for the assessment of cytotoxic effects. Therefore, growth effects of pristine CNTRENE material at concentrations higher than 1.05 µg/mL on E. coli K12 were evaluated by a modified spot plate technique (see Methods). The CFU/mL were calculated after 24 h of exposure to pristine CNTRENE material at final concentrations of 0 µg/ml (control), 8.44 µg/ml, 16.88 µg/ml, and 33.75 µg/ml. The calculated CFU/mL were 1.47 × 10⁹ ± 7.87 × 10⁸ CFU/ml (n = 9), 1.24 × 10⁹ ± 9.26 × 10⁸ CFU/ml (n = 9), 1.20 × 10⁹ ± 6.24 × 10⁸ CFU/ml (n = 9), and 1.87 × 10⁹ ± 5.00 × 10⁸ CFU/ml (n = 9), respectively (Fig. 7). These data were found to be Gaussian by a D’Agostino-Pearson normality test, and the differences between groups were not significant as determined by one-way ANOVA (p = 0.2138). Taken together with the MIC assays described above, these data imply that CNTRENE material exposure up to 33.75 µg/ml does not significantly impact the growth of E. coli.

Electron microscopy imagining

Morphological changes of E. coli K12 grown in LB pH 7 and exposed to pristine CNTRENE material at concentrations at and above 1.05 µg/mL were evaluated by electron microscopy. With SEM, control cells visualized at 10,000× and 25,000× appeared as morphologically normal bacilli, with intact outer membranes and lengths ranging between 1 µm and 2 µm (Figs. 8A–8B). After 24 h exposure to pristine CNTRENE material (33.75 µg/mL and 16.88 µg/mL), cells had similar morphologies to control samples were within the typical 1 µm to 2 µm length range of E. coli (Figs. 8C–8F).

As was observed with SEM, AFM images of control E. coli K12 cells appeared as morphologically normal bacilli with intact outer membranes, lengths between 1 µm to 2 µm, and diameters of 0.5 µm (Fig. 9A). After 24 h exposure to pristine CNTRENE material (33.75 µg/mL, 16.88 µg/mL, and 1.05 µg/mL), cells had normal morphological features including cell length (1 µm–2 µm) and diameter (0.5 µm), similar to control cells (Figs. 9B–9D).

In SEM images, CNTRENE material was primarily observed to be adjacent to bacterial cells and not in direct contact with the cell surface. Although, some CNTRENE material was in direct contact with outer membranes of the cells, no damage to outer membranes, such as physical puncturing, was observed. Due to the resolution limitations of the instrument, no CNTRENE material structures could be positively identified in AFM images, so no association between cells and CNTRENE material was directly observable. However, cells appeared intact, without abnormalities in cell morphology, which corresponds to the SEM images. Taken together, these data suggest that this CNTRENE material does not physically damage E. coli, which is in agreement with the data demonstrating normal growth upon CNTRENE material exposure.

RNA sequencing

Global transcriptional changes were examined by RNA sequencing to examine if any regulatory changes were responsible for the tolerance of E. coli to CNTRENE material. E. coli K12 cultures were grown to mid-log phase in M9 medium containing 1.05 µg/ml CNTs and compared to cultures without CNTRENE material exposure. Control samples had an average of 2,199,975 reads (6,599,925 total reads) and experimental samples had an average of 2,592,180 reads (7,776,541 total reads). All control and experimental samples had sequence lengths of 35–51 bp with a GC content of approximately 54%, closely mirroring the genomic GC percentage of 50.8% (Riley et al., 2006). Gene expression of E. coli K12 exposed to pristine CNTRENE material was compared to native gene expression with the E. coli K12 MG1655 reference genome used for mapping sequencing reads. Of the 4464 open reading frames (ORFs) (NCBI accession NC_000913), 4,314 genes were mapped indicating that 96.6% of all genes were expressed. Of the 4,314 genes mapped, 186 genes were differentially expressed using a 2-fold change between control and experimental samples as a threshold. Of these 186 genes, 26 genes were upregulated in the experimental CNT exposed samples, and 160 genes were downregulated (Fig. S1). However, only three genes (pptA, alpA, and mgtL) were expressed at a significantly different level in the experimental samples after correcting for false discovery rate of 0.05 using the Benjamini–Hochberg method (Benjamini & Hochberg, 1995). The pptA and alpA genes were considered significantly downregulated with a 2.5-fold change (p = 0.0272) and 35.1-fold change (p = 0.0227), respectively. The mgtL gene was the only gene to be upregulated, with an 85.3-fold increase in expression in experimental samples (p = 1.87 × 10⁻⁷). The pptA gene (COG 1942) is predicted to encode a 4-oxalocrotonate tautomerase that functions in degradation pathways for xenobiotic aromatic compounds (Kanehisa & Goto, 2000). The alpA gene (COG 3311) is a prophage regulatory protein that is part of a group of DNA transcription regulators within the MerR superfamily, and regulators within this family have been shown to regulate in response to environmental stressors, such as heavy metals (Marchler-Bauer et al., 2015). However, the genes downstream of this ORF encode proteins associated with the cryptic prophage CP4-57 and are not known to be associated with stress responses. The mgtL gene acts as a leader sequence to the downstream mgtA gene (COG 0474), which is a Mg²⁺ transport protein (Park et al., 2010). The mgtL leader sequence serves a riboswitch for the Mg²⁺ porin, which allows expression of the porin to be regulated by the availability of proline and Mg²⁺ (Park et al., 2010). Regardless, the role of all three genes identified by RNA-seq in response to CNTRENE material exposure is unclear as all three genes have dissimilar function and are not part of a general stress response that would be expected from known mechanisms of CNT toxicity, such as physical interaction (e.g., cell envelope damage), ROS generation, or oxidative stress (Nel et al., 2006; Kang et al., 2008).