Immobilization of DNA at Glassy Ccarbon Electrodes: A Critical Study of Adsorbed Layer

In this work we present a critical study of the nucleic acid layer immobilized at glassy carbon electrodes. Different studies were performed in order to assess the nature of the interaction between DNA and the electrode surface. The adsorption and electrooxidation of DNA demonstrated to be highly dependent on the surface and nature of the glassy carbon electrode. The DNA layer immobilized at a freshly polished glassy carbon electrode was very stable even after applying highly negative potentials. The electron transfer of potassium ferricyanide, catechol and dopamine at glassy carbon surfaces modified with thin (obtained by adsorption under controlled potential conditions) and thick (obtained by casting the glassy carbon surface with highly concentrated DNA solutions) DNA layers was slower than that at the bare glassy carbon electrode, although this effect was dependent on the thickness of the layer and was not charge selective. Raman experiments showed an important decrease of the vibrational modes assigned to the nucleobases residues, suggesting a strong interaction of these residues with the electrode surface. The hybridization of oligo(dG)21 and oligo(dC)21 was evaluated from the guanine oxidation signal and the reduction of the redox indicator Co(phen)33+. In both cases the chronopotentiometric response indicated that the compromise of the bases in the interaction of DNA with the electrode surface is too strong, preventing further hybridization. In summary, glassy carbon is a useful electrode material to detect DNA in a direct and very sensitive way, but not to be used for the preparation of biorecognition layers by direct adsorption of the probe sequence on the electrode surface for detecting the hybridization event.


Introduction
Since the first publication about the electrochemistry of nucleic acids [1], there has been a growing interest in the development of electrochemical affinity biosensors based on the use of nucleic acids as biorecognition element. Different strategies for DNA-based biosensors preparation as well as different transduction modes have been proposed, as it was largely reviewed [2][3][4][5][6][7][8][9][10][11][12][13].
Brabec et al. [35] studied the electrochemical behavior of DNA at carbon electrodes by using voltammetric techniques. They found that the oxidation currents are higher for smaller DNA molecules and that low-molecular weight components of DNA present in DNA samples can influence the DNA electrooxidation. Pang et al. [36] found that DNA can be strongly adsorbed on glassy carbon surfaces only if the DNA solution is evaporated to dryness on the electrode. They evaluated the presence of the nucleic acid layer from the signal of the redox indicator Co(phen) 3 3+ . Oliveira Brett et al. [27][28][29][30] reported on the use of glassy carbon pretreated as a substrate for preparing DNA modified electrodes (subjected to oxidation processes) to evaluate the voltammetric behavior of nitroimidazoles. Wang, Z. et al. [37] reported on the adsorption and oxidation of denatured and double stranded DNA at glassy carbon electrodes by using differential pulse voltammetry (DPV) and "in situ" FTIR spectroelectrochemistry. They found that after accumulation of denatured calf thymus DNA, and depending on the denaturation procedure, a partly reversible voltammetric behavior is obtained and a multilayer is formed on the electrode surface.
Oliveira Brett and Matysik [38] reported the sonovoltammetric detection of all nucleobases at GCE. These results were obtained in alkaline solutions and pyrimidine concentrations ten times higher than those of purines. Recently, Oliveira Brett et al. [39] proposed the detection of all deoxyribonucleotides at GCE using DPV. They reported that an adequate conditioning of GCE (polishing, sonication and potentiostatic and potentiodynamic electrooxidation schemes) made it possible a better peak separation and enhancement of the oxidation currents for all the DNA bases, as well as the simultaneous detection of the oxidation of pyrimidine residues in ssDNA in phosphate buffer solutions pH 7.4.
In this work we report on a critical study about the characteristics of the DNA layer immobilized at the surface of GCE by using different techniques. The main interest was focused on the determination of the type of interaction of DNA with the electrode surface (electrostatic versus hydrophobic), as well as the compromise of the bases in this interaction.

2.1.Reagents
The oligonucleotides were purchased from Life Technologies (Grand Island, New York, USA) as their ammonium salts: Probe sequence: Oligo(dG) 21 3'-GGG GGG GGG GGG GGG GGG GGG-5' Target sequence: Oligo (dC) 21    Repetitive measurements were carried out by polishing the surface of GCE or gCPE and repeating the above assay. All experiments were performed at room temperature.

Raman experiments
The reference spectrum of DNA was taken from solid calf thymus dsDNA. Samples obtained by immobilization of DNA solutions, were prepared in two ways. In one case, by casting the glassy carbon disk (CH Instruments or Alfa AEsar) with 10 µL of 1000 ppm dsDNA solution and waiting until dryness. In the other case, it was obtained by depositing the nucleic acids under potential controlled conditions ensuring full coverage of the surface. In the latter, the dsDNA was deposited at 0.200 V from a 10.0 ppm solution for 5.0 min, while the oligo(dG) 21 was accumulated from a 2 ppm solution for 3.0 min. Some experiments were performed once the DNA layer was oxidized by performing a chronopotentiometric stripping between 500 and 1250 mV at a constant current of 8.0 µA. respectively. An increase in the peak potentials separation for all the redox markers was also obtained, indicating that the DNA layer makes more difficult the electron transfer. The peak potential separation increased by 51, 83 and 156 mV for potassium ferricyanide, catechol and dopamine, respectively.

Results and Discussion
Another interesting fact is that once the DNA layer was oxidized, in general, the oxidation and reduction currents were slightly smaller than those obtained at dsDNA-GCE without oxidation, while peak potentials separations increase just a little. and dopamine (C) at bare GCE (black) and at dsDNA-modified-GCE (one cycle without oxidizing the DNA (red), another cycle during the DNA oxidation (green) and a last cycle after DNA oxidation (blue)). In the presence of the DNA layer the reduction current for potassium ferricyanide and oxidation currents for dopamine and catechol decreased around 80 %. The peak potential separation showed a large increase in the case of catechol (186 mV) and dopamine (196 mV) and a small increase in the case of potassium ferricyanide (24 mV). Another important aspect to be considered is the change in the currents and peak potential separation with the number of cycles. After the large decrease in peak currents observed for the cycle one, a gradual increase was observed in the consecutive ones in all cases. Once the DNA was electrooxidized, the signal for potassium ferricyanide decreased, while those for the other two compounds, slightly increased.
In summary, these results indicate that the DNA layer obtained by both of the two procedures employed forms some kind of islands leaving channels or uncovered areas that allow the redox markers reach the electrode surface and perform the electrons exchange. It is clear that both types of DNA layers work like a non-charge selective physical barrier for the diffusion of the redox markers, although it does not completely block their diffusion towards the GC surface.
Regarding the different increase in the peak potential separation for the three redox markers at the GCE modified with thick DNA films, it can be attributed to some interaction of catechol and dopamine with the DNA immobilized close to the electrode surface. Once these aromatic compounds reach the electrode surface through the pores of the DNA layer, they could establish hydrophobic interactions with the nucleobases located close to the electrode surface through the π system. Since potassium ferricyanide does not have aromatic structure, it is not able to establish the same kind of interaction with DNA as catechol and dopamine and the peak potential separation change is smaller.
The increase in the currents observed in the second cycle can be assigned to a reorganization of the DNA layer that allows more redox marker molecules reach the electrode surface. At the thin layers, a similar behavior is obtained, although the changes are not as important as in the case of a GCE modified with a thick DNA layer. conditions like direct exposure to air, potentiodynamic and potentiostatic treatments in sulfuric acid, sodium hydroxide, phosphate and acetate buffers gave a response considerably smaller than that obtained at the freshly polished electrode. Therefore, the adsorption and electrooxidation of nucleic acids is more difficult at this pretreated glassy carbon surface, at variance with graphite paste electrode, where a pretreatment is necessary to obtain a good guanine oxidation signal. The origin (procedence) of the glassy carbon has also demonstrated to be an important variable. We have evaluated the adsorption and electrooxidation of dsDNA at three different electrodes: CH Instruments, BAS (having the same geometric area) and V-10 Carbone-Lorraine (with a considerably larger geometric area). The response obtained at the CH and BAS glassy carbon electrodes was similar. On the contrary, no response was obtained at the V-10 Carbone-Lorraine glassy carbon electrode even after covering the electrode with 5 µL of a 4900 ppm solution. Therefore, the different content of oxygenated functions, related to the preparation conditions of glassy carbon, results very important for further adsorption and electrooxidation of nucleic acids. A similar behavior was reported [22] in the case of carbon fiber electrodes, where very different guanine oxidation signals were obtained after nucleic acids adsorption and electrooxidation depending on the origin of the fiber.
It is known that the adsorption of oligo and polynucleotides on the surface of glassy carbon is independent on the adsorption potential between -0. 30   breathing and stretching of the bases (600-800 and 1200-1400 cm -1 , respectively) [41][42]. The spectra for dsDNA immobilized at the glassy carbon (in any of the immobilization procedures) showed less bands than that for solid dsDNA. There was a shifting in the frequencies of some bands and several disappeared as a consequence of the interaction and orientation of the macromolecule with the electrode surface. This effect is more pronounced if the immobilization of DNA is performed under potential controlled conditions. In this case, optical microscopy reveals an inhomogeneous surface coverage with the formation of islands probably due to the drying step previous to the Raman experiments (not shown). The main difference between the solid dsDNA reference spectrum and that of the dsDNA immobilized at the electrode surface can be observed in Table 1. In this table we  It is important to remark that the spectra of oligo(dG) 21 and dsDNA showed similar bands and the spectra obtained before and after electrochemical oxidation of the DNA layer immobilized under potential controlled conditions, were very similar. These results suggest that the absence of guanine electrooxidation signal for a DNA immobilized on the same surface where a stripping step was previously performed (not shown), is indicative of some kind of electrode passivation by the oxidized DNA residues that remain adsorbed to the surface after the stripping step. Table 1. Comparison and frequency assignment between Raman bands of the solid dsDNA spectrum, taken as reference (first column), and those of the dsDNA immobilized at the electrode surface by casting the glassy carbon electrode with 10 µL of 1000 ppm dsDNA solution (second column), or by adsorbing the polynucleotide under controlled-potential conditions for full surface coverage (5 min, at 0.20 V from a 10.0 ppm solution) (third column), and the reference spectrum bands informed in the literature [42,43] (fourth column). Abbreviations: dA, deoxyadenosine; dT, deoxythimidine; A, adenine; T, thymine, G, guanine; bk, backbone; s, stretching; b, breathing; df, deformation; rk, rock; ip, in plane.
The obvious question after these results is the feasibility to use glassy carbon modified with oligonucleotides for sequence-specific detection, since, on one hand, the DNA immobilized layer is very stable, which is highly desirable when developing a hybridization biosensor; but on the other hand, there is a strong interaction of the bases with the glassy carbon surface, that could not leave them available to recognize the complementary sequence. Two strategies were used to evaluate the hybridization event. One of them was based on the guanine oxidation signal and the other on the signal of a redox indicator able to preferentially bind to dsDNA. The probe was oligo(dG) 21 while the target was oligo(dC) 21 . Figure 5 shows  We also study the hybridization event using Co(phen) 3 3+ to make sure that the expected decrease in guanine oxidation signal did not occur because the compromise of the bases in the interaction with the electrode surface and not due to problems with the electron transfer between guanine residues and the electrode. This redox indicator presents the advantage that the reduction occurs at very low potentials, offering, in this way, a more defined hybridization signal. This redox indicator has been successfully used as hybridization redox marker [43][44][45]. The interaction between this metallic complex and DNA was studied at the solution [46] and at the electrode surface [47][48]. A very interesting discussion about the nature of its interaction with dsDNA immobilized at gold electrodes as a function of the ionic strength was proposed by Abruña et al. In our case, the experimental conditions for performing the interaction of the adsorbed DNA with Co(phen) 3 3+ were selected according to previous results using this redox indicator for detecting the hybridization event [43][44][45]. It is important to remark that in our case the dsDNA is oligo(dG) 21 oligo(dC) 21 , at variance with others where use calf-thymus dsDNA and the supporting electrolyte was Tris.HCl without NaCl. Since this compound can interact with single stranded DNA by association at the grooves and it can also intercalate into the double helix, the analytical response was obtained as the  Figure 6 shows this analytical signal as a function of the hybridization time in the presence of the target (oligo(dG) 21 ) at probe (oligo(dC) 21 )-modified GCE and gCPE. No changes in the chronopotentiometric reduction signal of Co(phen) 3 3+ were observed at the GCE modified electrode, confirming that the hybridization event could not take place. At probe-gCPE the signal increases with the hybridization time due to the increase in the accumulated amount of the double helix at the surface of the electrode and the consequent increase in the amount of bond redox indicator.
In summary, considering the information given in the different publications, it is clear that if GCE is used for the detection of nucleic acids, there are several factors that need special consideration.
The adsorption and further electrooxidation of DNA largely depends on the nature of the GCE, that is, on the pyrolisis conditions during the preparation. The pretreatment of the surface is another important parameter to take into account, and it is directly related to the GCE preparation conditions. In some cases a pretreatment is necessary, while in others, as in our case, is it absolutely prejudicial. The presence of halides changes the conformation of the nucleic acid in solution, thus, affecting the conformation in the adsorbed state, especially in the case of bromide and iodide even at very small concentrations. The results presented here, confirm that the interaction between DNA and glassy carbon electrode is mainly hydrophobic with an important compromise of the bases that decreased their availability for further hybridization.

Conclusions
The knowledge of the characteristics of the DNA layer immobilized on the surface of a transducer is critical for the adequate functioning of a biosensor. Electrochemical and spectroscopic experiments demonstrated that the interaction between DNA and glassy carbon surface is mainly hydrophobic. There is an important compromise of the bases in the adsorption, which makes difficult further hybridization. Therefore, glassy carbon is a useful electrode material to quantify DNA in a direct and very sensitive way, but not to be used for the preparation of biorecognition layers for detecting the hybridization event by direct adsorption of the probe sequence on the glassy carbon surface. It is necessary to perform some modification on the surface of this material previous to immobilize the DNA layer in order to decrease the compromise of the bases in the adsorption. Our group is currently working in this direction.