Carboxylation of MWCNT
As we discussed earlier, surface modification of CNTs are essential not only for binding them to other materials, but also to improve their solubility. In our case, MWCNTs were functionalized by oxidation through acid treatment, which enabled us to incorporate carboxyl groups as binding sites to the linearized plasmid. Among different oxidation treatments available, the H2SO4/NO3 (3:1, v:v) method has been demonstrated to yield carboxyl groups with complete oxidation of side walls. We used this method with some minor modifications, as reported in our previous study. The obtained fCNTs were soluble, and they did not aggregate in MES buffer (100 mM, pH 5.0) up to a concentration of 2 mg mL−1.
Additionally, characterization of the oxidized MWCNTs was carried out by comparing them with the unmodified MWCNTs using X-ray photoelectron spectroscopy. An increase in intensity was clearly observed by comparing the changes in O(1s) spectral envelope. According to the C(1s) spectrum, no significant structural damage was observed even in the presence of strong oxidizing agent (Figure 2). For further characterization and visualization of the morphologies, TEM images of the modified and unmodified MWCNTs were taken. Compared to pristine CNTs (Figure 3a), the reduced dimensions of fCNTs (Figure 3b) resulting from the acidic treatment were clearly observed.
Effect of fCNT on cell viability
Even though carbon nanotubes are widely used for their intrinsic properties, there is a debate for their cytotoxicity in biomedical applications. Their toxic properties result from surface metal residues (Fe, Co, Ni) originating from synthesis process, as well as their low solubility in unmodified form. However, it is possible exclude these disadvantages using a proper modification method.
When we tested the toxic effect of fCNT in solid medium, we did not observe any significant changes in colony numbers. Even in the presence of 1 mg fCNT, E. coli cells were able to grow on agar plates (Figure 4). We also carried out the experiments using chemically competent cells in liquid medium, as described in methods. Considering that competent cells are readily open to cytotoxic effects, the effect of fCNT in transformation medium was expected to be greater. We observed a significant decrease in colony numbers, when fCNT was present in amounts >150 μg mL−1 (Table 1). Thus, we did not exceed the 100 μg fCNT limit in further experiments.
Figure 4. The effect of fCNT on cell viability was tested in solid medium.
E. coli cells were incubated in the absence of fCNT, or in the presence of 1 mg fCNT. Colony forming units (CFU) were inspected manually on agar plates.
Download figure to PowerPoint
Table 1. The Effect of Different Concentrations of fCNT on E. coli Cell Viability was Tested in Liquid Medium (see the text for details)
Several studies evaluated the toxicity of CNTs using different strategies.[24-26] Differentiation of the methods yielded contradictive results because of different structure, agglomeration state, and perhaps most importantly, surface functionalization parameters. In a recent study, time-dependency of cytotoxic effect has been investigated at different incubation times (15 min, 1 h, 5 h and overnight). It has been shown that a 1-h incubation with CNTs results in a 50% viability loss of E. coli cells. The main reason of this toxicity could be attributed to cell membrane damage. In our study, a 15-min of incubation time was sufficient for transformation experiments. Therefore, it is unlikely to see a dramatic cytotoxic effect of MWCNTs in our study.
Effect of the modified nucleotide ratio on PCR yield
Intrinsic properties of DNA molecules make them ideal scaffolds for preparation of nanoparticles and immobilization onto nanomaterials. As reviewed elsewhere, DNA oligonucleotides have been modified with various groups that have affinity for proteins of interest, and noncovalent conjugates have been prepared. Covalent conjugation is mostly based on thiol chemistry, which has been used for conjugation of a 500-bp DNA fragment with cytochrome c protein. In that study, the DNA fragment has been PCR-amplified using 2-aminoallyl-dUTP, and then a second reaction has been performed to introduce free thiol groups. In a similar approach, we introduced primary amino groups into pQE-70/gfp plasmid using commercially available N6-(6-Amino)hexyl-2′-deoxyadenosine-5′-triphosphate (NH2-dATP). The circular plasmid DNA was linearized, and modified with amino groups in a semirandom manner. The length of the PCR product was ∼4100 bp, which was then covalently conjugated with fCNT.
To optimize the PCR conditions, two parameters were investigated. Molar ratio of the modified nucleotide (NH2-dATP), and primer annealing temperatures were tested. Three PCR mixtures were prepared as follows: (a) without NH2-dATP, (b) NH2-dATP:dATP at 1:9 molar ratio, and (c) NH2-dATP:dATP at 1:3 molar ratio. For each reaction mixture, PCRs were performed at six different annealing temperatures ranging between 50 and 64°C. The intensities of the amplified fragments were examined under UV light, which revealed that adding NH2-dATP at 1:9 molar ratio had only a slight effect on PCR efficiency. As shown in Figure 5, increasing the concentration of NH2-dATP resulted in relatively weak DNA bands, which indicated a lower yield in case of the 1:3 molar ratio. It has been reported that PCR yield is strongly dependent on chemical nature of the modified nucleotide and DNA polymerase used. Therefore, the PCR conditions were determined experimentally. As shown in Figure 5, the annealing temperatures of 62 and 64°C were found to be optimum for primer binding.
Figure 5. Gradient PCR with annealing temperatures of 50, 55, 58, 60, 62, and 64°C was performed in the absence of NH2-dATP (lanes a1-a6).
For comparision of the PCR yield, gradient PCR was also performed using NH2-dATP at molar ratios of 1:9 (lanes b1-b6) or 1:3 (lanes c1-c6). The amplification products were visualized on agarose gel under UV light. DNA marker (NanoHelix, South Korea) is indicated by M.
Download figure to PowerPoint
EDC reaction and agarose gel analysis of the reaction products
The well established EDC chemistry was used to bind fCNT and mpDNA covalently. The binding reaction was continued at 4°C overnight. Before removal of the pellet by centrifugation, an aliquot from the reaction mixture was taken. Then, the reaction mixture was centrifuged, and another aliquot was taken from the supernatant fraction. The free mpDNA was used as a control, which gave a bright DNA band with an expected size of ∼4100 bp (Figure 6). Staining of the agarose gels with ethidium bromide (EtBr) enabled detection of DNA bands under UV light. Interestingly, the samples taken before (R) and after centrifugation (Rs) were retarded, and moved slower on agarose gel. These samples yielded DNA bands of approximately equal intensity. As it has been reported, MWCNTs show different water dispersion properties depending on surface chemistry and length. We assumed that a certain amount of fCNTs were highly dispersed in the reaction medium, and they could not be removed by centrifugation at 8,000 rpm. We concluded that the unbound mpDNA might have interacted with the highly dispersed fCNTs. The pellet obtained by centrifugation was also examined on agarose gel under UV light, which yielded no DNA bands. It has been reported that non-covalent NH2-MWCNT-DNA complex with a tendency to aggregate could not move through the agarose gel, and it is not visible on agarose gel. Similarly, we could not observe the presence of bound DNA. Nevertheless, the brightness of samples taken from the soluble fraction of the reaction mixture (Figure 6, lanes R and Rs) were weaker than that of free mpDNA (Figure 6, lane C). Therefore, it was deduced that a significant amount of mpDNA was bound by fCNT. Using Tris-EDTA buffer in washing steps enabled a net negative charge on mpDNA. Thus, we assumed that the non-covalently adsorbed mpDNA could be eliminated by electrostatic repulsion forces between the negatively charged DNA and the carboxylate groups of fCNT. The supernatants obtained from washing with Tris-EDTA buffer were subjected to agarose gel electrophoresis (Figure 6, lanes 1′, 2′, and 3′), but no DNA bands were observed. This could be attributed to the absence of unbound mpDNA, or that the samples were below detection limit. EtBr has a detection limit of 2–10 ng, and we introduced 4.9 µg mpDNA into the reaction. Thus, regarding the binding yield of mpDNA, the presence of DNA at ng levels could be negligible compared to µg level.
Figure 6. Agarose gel visualization of free mpDNA and samples taken from the reaction mixture.
Lane 1, DNA marker (Thermoscientific, Rockford, IL); lane 2, free mpDNA that was introduced into the reaction; lane 3, the pellet fraction obtained by centrifugation (Rp); lane 4, an aliquot taken from the reaction mixture before centrifugation (R); lane 5, an aliquot taken from the reaction mixture after centrifugation (Rs). Lanes 6, 7 and 8 indicate samples taken from first, second and third washing steps, respectively. The weak DNA bands in lanes 4 and 5 are indicated by white arrows.
Download figure to PowerPoint
Transformation of chemically competent E. coli cells
The negatively charged lipopolysaccharide layer of E. coli prevents the entry of foreign DNA. Thus, E. coli cells are artificially made “competent” for uptake of DNA. Chemical transformation includes treatment of the cells with chloride salts of calcium, magnesium, or rubidium, which is followed by a brief heat-shock at 42°C. Addition of CaCl2 is required to neutralize the unfavorable interactions between DNA and the negatively charged outer layer. When we used non-competent cells for transformation of fCNT-mpDNA bioconjugate, we could not observe any transformants. For this reason, we treated the cells with CaCl2, and we aimed to reduce the electrostatic repulsion between fCNT-mpDNA and lipopolysaccharide layer. Computational studies revealed the cell penetration mechanism of single walled carbon nanotubes (SWCNTs). Therefore, we hypothesized that fCNT-mpDNA including the surrounding Ca2+ ions would be able to penetrate into the cells, and the heat-shock step at 42°C would not be required. In our approach, we excluded the heat-shock step for transformation of fCNT-mpDNA bioconjugate into E. coli cells. It has been reported that heat-pulse and cold-shock steps are responsible for formation of pores on cell surface and lowering of membrane potential, which facilitate DNA to cross inner membrane of E. coli.
As a control, we performed transformation of chemically competent cells with the purified pQE-70/gfp plasmid DNA using heat-shock step at 42°C. When we used NH2-dATP:dATP at 1:9 molar ratio, we could not observe any transformants. We assumed that wide distribution of NH2-dATP on the plasmid DNA resulted in a high degree of fCNT conjugation, and that steric hindrance interfered with gene expression in E. coli cell. Thus, we decided to reduce the number of binding points by decreasing the molar ratio of NH2-dATP. Cell transformation experiments were achieved using the 1:19 ratio, which corresponded to 5% substitution by NH2-dATP. E. coli cells, which were transformed with either control plasmid or with the bioconjugate were spread on agar plates supplemented with ampicillin. Because pQE-70 carries the ampicillin resistance gene, only the cells that acquired the plasmid were able to survive. As shown in Figure 7, the number of CFU was roughly same for both control and bioconjugate transformations. Presence of a cloned gfp gene on the plasmid enabled us to detect GFP fluorescence manually. When the agar plates were inspected under UV light, the cells that expressed GFP were identified manually. Compared with the control, GFP fluorescence of the cells transformed with the bioconjugate was slightly weaker. This could be attributed to the behavior of the linearized plasmid DNA in E. coli cell. When the linearized pBR322 plasmid DNA has been transformed into Ca2+-treated E. coli cells, deletions or rearrangements have been observed. Because of intracellular DNA ligase activity, linearized DNA could recircularize and form the functional plasmid in the cell. In our case, we assumed that recircularization occurred depending on the steric hindrance of the covalently bound fCNT. Therefore, rearrangement of the linearized plasmid, or steric effects could be responsible for the hindered expression of GFP.
Figure 7. Chemically competent E. coli cells were transformed with pQE-70/gfp (as control) and fCNT-mpDNA bioconjugate.
The green fluorescence expressed by E. coli cells was visualized under UV light.
Download figure to PowerPoint
We achieved transformation of the linearized plasmid DNA using covalently bound fCNT. Although plasmids are widely used for gene transfection, the use of linear DNA fragments avoids the need for plasmid propagation in bacterial cells and antibiotic resistance genes. Thus, linear DNA transfection is preferred for medical purposes. However, nuclease attack on the foreign DNA fragment and low transfection efficiency are limiting factors. Besides its cell penetrating properties, single walled carbon nanotubes (SWCNTs) have been shown to protect nucleic acid strands from nuclease digestion. To accomplish this, the SWCNT has been wrapped non-covalently with a 30-base-paired single-stranded oligonucleotide. This study differs in two ways from our strategy. First, we used a double stranded plasmid DNA, which was linearized by PCR. Second, our strategy was based on a covalent chemical reaction between the carboxylated fCNT and the amino-modified plasmid DNA. The linearized plasmid DNA with a cloned β-galactosidase gene has been transfected into Vero cells using a cationic lipid, albeit with a 10–15% β-galactosidase expression yield. In a more recent study, micro-linear vector has been developed for gene transfection, and GFP expression has been achieved. Considering the need for linear gene delivery systems, the developed fCNT-mpDNA bioconjugate could give insights into the use of CNTs for this aim. CNT-based gene delivery systems have been described in the literature,[40-42] which involve functionalization of CNTs with positively charged groups and electrostatic interactions with DNA. In contrast to these noncovalent conjugates, we developed a covalent conjugate between the linearized plasmid DNA and fCNT.
As previously reported, transformation yield depends on number of the cells, especially when a carrier or auxiliary molecule is present. We performed the transformation experiments using E. coli XL1 Blue cells at OD600 = 200, which required a high cell density. E. coli XL1 Blue is one of the most widely used strains for cloning and transformation.
In another study, transformation has been achieved by microwave exposure to a non-covalent mixture consisting of MWCNT, E. coli DH5α cells and pUC19 plasmid DNA. Compared with this study, our approach relied on a covalent fCNT-mpDNA bioconjugate instead of a noncovalent interaction. Moreover, we did not require additional steps, such as microwave activation or heat-shock step. In the previous study, E. coli cells have been used following washing with phosphate buffered saline buffer. In our study, we washed E. coli cells with CaCl2 in order to make the cells chemically competent.
Plasmid DNA can be delivered into competent E. coli cells using electroporation. Compared to chemical transformation, this method enables a higher transformation efficiency (108−1011 CFU µg−1 DNA). However, it is limited by high rate of cellular mortality, and electroporation conditions should be carefully optimized. As mentioned above, the method of microwave activation of MWCNTs has been developed as an alternative to electroporation, which has resulted in a transformation efficiency ranging between 9 and 32 CFU µg−1 circular pUC19 plasmid DNA. MWCNT-aided plasmid delivery is not as efficient as electroporation, but it eliminates some disadvantages of electroporation, such as cell damage and the need for instrumentation. It should also be considered that electroporation of linear plasmid DNA results in a 1,000-fold drop in transformation efficiency. In our approach, we used fCNT for delivery of PCR-linearized plasmid DNA. Following transformation of the bioconjugate, we investigated GFP expression rather than evaluation of transformation efficiency. Nevertheless, based on the cell growth on the plates, we estimated that transformation efficiency of linear DNA was comparable with that of circular plasmid DNA. We assume that our approach could be particularly useful for delivery of linear DNA, albeit with a lower transformation efficiency than that of electroporation.
Cellular uptake mechanism of CNTs has been discussed in a recent study, which concludes that several internalization pathways might simultaneously operate. To our knowledge, there is no study that could reveal possible interactions between CNT and cellular components inside the cell. However, it has been reported that MWCNTs could induce pro-inflammatory responses depending on their length and cell type. Further experiments are necessary to characterize the effects of E. coli cell type and DNA length on transformation. Moreover, the uptake mechanism of the developed covalent bioconjugate needs to be revealed. EDC chemistry has been used for functionalization of CNTs with single-stranded short DNA oligonucleotides[48-50] and proteins. However, in those studies, delivery of DNA-functionalized CNT into cells has not been attempted. We employed a similar EDC chemistry, but in order to attach the double-stranded plasmid DNA. We also achieved the transformation of E. coli cells with the developed covalent bioconjugate, which allowed the expression of GFP in E. coli cells.