Ternary DNA chip based on a novel thymine spacer group chemistry

Colloids and Surfaces B: Biointerfaces 125 (2015) 270–276

Contents lists available at ScienceDirect

Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb

Ternary DNA chip based on a novel thymine spacer group chemistry Yanli Yang, Umit Hakan Yildiz, Jaime Peh, Bo Liedberg ∗ Center for Biomimetic Sensor Science, School of Materials Science and Engineering, Nanyang Technological University, 50 Nanyang Drive, Singapore 637553, Singapore

a r t i c l e

i n f o

Article history: Received 17 April 2014 Received in revised form 24 October 2014 Accepted 30 October 2014 Available online 7 November 2014 Keywords: Ternary DNA chip Thymine Spacer groups Surface chemistry Hybridization Electrochemistry

a b s t r a c t A novel thymine-based surface chemistry suitable for label-free electrochemical DNA detection is described. It involves a simple two-step sequential process: immobilization of 9-mer thymine-terminated probe DNAs followed by backfilling with 9-mer thymine-based spacers (T9). As compared to commonly used organic spacer groups like 2-mercaptoethanol, 3-mercapto-1-propanol and 6-mercapto-1-hexanol, the 9-mer thymine-based spacers offer a 10-fold improvement in discriminating between complementary and non-complementary target hybridization, which is due mainly to facilitated transport of the redox probes through the probe-DNA/T9 layers. Electrochemical measurements, complemented with Surface Plasmon Resonance (SPR) and Quartz Crystal Microbalance (QCM-D) binding analyses, reveal that optimum selectivity between complementary and non-complementary hybridization is obtained for a sensing surface prepared using probe-DNA and backfiller T9 at equimolar concentration (1:1). At this particular ratio, the probe-DNAs are preferentially oriented and easily accessible to yield a sensing surface with favorable hybridization and electron transfer characteristics. Our findings suggest that oligonucleotide-based spacer groups offer an attractive alternative to short organic thiol spacers in the design of future DNA biochips. © 2014 Elsevier B.V. All rights reserved.

1. Introduction With the advancement in technology and the need for compact, miniaturized, hand held analytical instrumentation, biosensor concepts based on electrochemical interrogation have gained increasing popularity because of ease-of-use, low fabrication cost, possible non-labeling and fast target detection [1–5]. Electrochemical biosensors have been used extensively for the detection of various biomolecular targets including enzymes [6], antigens [7], oligonucleotides/DNA [8,9], and metabolites [10,11]. The single-stranded probe-DNA density on the transducer surface, often gold, has proven to be an important design parameter in order to attain biosensors with high sensitivity and selectivity. While it is commonly assumed that the sensitivity of the biosensor increases with the number of immobilized probe-DNAs, it is today widely accepted that a closed-packed probe DNA layer can have a detrimental effect on the DNA hybridization efficiency [4]. When the probe-DNAs are too densely packed, DNA hybridization decreases, because the complementary target DNA is unable to penetrate into the probe-DNA layer for steric reasons [12]. The change

∗ Corresponding author. Tel.: +65 6316 2957; fax: +65 6791 2274. E-mail address: [email protected] (B. Liedberg). http://dx.doi.org/10.1016/j.colsurfb.2014.10.058 0927-7765/© 2014 Elsevier B.V. All rights reserved.

in electrochemical response upon DNA hybridization may also be negligible as the closely-packed probe-DNAs would already have blocked the electrode surface for efficient electron transfer [13]. On the other hand, if the probe-DNA density is too low, the detection efficiency also will be low. Furthermore, while some groups reported an increase in electron transfer due to the enhanced accessibility of the redox probes to the electrode surface, some reported a decrease in electron transfer due to the increased charge repulsion with the higher interfacial negative charges. The improved accessibility of the redox probes to the electrode surface was believed to be due to the formation of the more rigid and rod-like doublestranded DNA, which exposed the electrode surface for efficient electron transfer [13,14]. The increase in the number of negatively charges at the electrode surface due to the additional amount of DNA after hybridization, on the other hand, result in an increased in electrostatic repulsion with the negatively-charged redox probes, hence reducing electron transfer [15,16]. Thus, favorable conditions for hybridization, and thereby improved sensitivity and selectivity, can be obtained by optimizing the surface probe density and orientation/presentation on the electrode surface. Short organic spacers groups, often thiolated, have been extensively employed to modulate the probe-DNA density on the transducer surface. Besides reducing steric hindrance, the spacer groups also help removing non-specific interaction between

Y. Yang et al. / Colloids and Surfaces B: Biointerfaces 125 (2015) 270–276

the nucleotide groups of the probe-DNA and the gold surface [2,17,18]. Thus, one can say that the spacer assists the probe-DNA to orient favorably on the surface [17,19–22]. Commonly used spacer groups include 2-mercaptoethanol (ME), 3-mercapto-1propanol (MP) and 6-mercapto-1-hexanol (MCH). These short spacer groups are known to effectively reduce steric hindrance and non-specific interactions [18,23]. However, due to the differences in hydrophilicity between these organic spacer groups and probeDNAs, multiphase separation might occur which affects the overall immobilization efficiency of the sensing layer [24]. Furthermore, electron tunneling through the ME layer has also shown to be relatively easy due to the presence of surface imperfections and pinholes [13,25,26] and this causes an insignificant change in redox signal after the backfilling process [24]. Dharuman and co-workers have been extensively involved in developing different electrode designs composed of binary and ternary mixed spacers based on short alkanes and probe-DNAs [24,27,28]. They have successfully showed that, while hydrophobic spacer groups are required for the formation of a compact sensing layer, hydrophilic and charged groups are also needed to introduce a favorable distribution/orientation of the probe-DNAs on the electrode surface. X-ray photoelectron analysis showed that the hydrophobic MCH spacer effectively fills up the exposed gold surface after probe-DNA immobilization. The mercaptopropanoic acid (MPA) spacer also can be used to effectively reduce the non-specific interactions between the bases in probe-DNAs and Au surface because of electrostatic repulsion [27]. In addition, by observing the voltammetric reductive desorption pattern, it was shown that MPA causes phase separation when co-adsorbed with probe-DNAs [24]. Dharuman furthermore observed that a ternary sensing surface prepared by co-adsorbing two different spacer groups and the probe-DNAs offered greater discrimination capacity between complementary and non-complementary targets. For example, an improved discrimination efficiency of 1.79 was obtained for a ternary probe-DNA/MCH/MPA surface as compared to using binary sensing surface (probe-DNA/MCH) [27]. Although highly relevant to our work, their main focus is on DNA hybridization discrimination in the presence of cationic intercalators (e.g. methylene blue), which can introduce problems related to non-specific interactions with other negatively charged groups present on the transducer surface. Unwanted charge compensation between proflavine (a cationic intercalator) and the probe DNAs also has proven to result in false electrochemical response [29]. Besides the use of conventional linear organic thiols, Campuzano et al. also studied a ternary sensing surface that was prepared by co-immobilizing probe-DNA and a linear dithiol (e.g. hexanedithiol), followed by backfilling with MCH. Such a sensing surface improves the signal-to-nose ratio upon hybridization with the target (1 nM), as compared to binary sensing surface [30]. However, this finding is based on chronoamperometric measurements using a sandwich hybridization method, which is indirect and requires a multiple-step assembly procedure. An oligonucleotide-based spacer is employed in this study as an alternative to conventional organic alkyl spacer group to reduce steric hindrances between neighboring probe-DNAs and to improve target recognition/hydridization. Demers and co-workers have shown that the binding strengths of DNA bases on gold are generally weaker than that of an alkanethiol, and among the DNA bases, thymine was shown to interact weakly with gold as compared to the other three bases (cytosine, adenine and guanine) [31] resulting in a small footprint of the probe-DNA on the surface. A small footprint of the attached thymine bases on gold is an important property that favors a vertically oriented assembly layer on gold. Moreover, the thymine bases which contain both the charged phosphate backbone and a methyl group are likely to satisfy the requirement of spacer groups having both hydrophobic and hydrophilic properties, as suggested by Dharuman et al.

271

[24,27,28]. A higher probe-DNA coverage and orientation has also been achieved when thymine bases are present near the anchoring site used for immobilization of the probe-DNA on gold substrate [32]. This is an important feature in electrochemical biosensing, since compactness of the SAM formed on the surface will affect the measured current flow before and after hybridization. The overall aim of this contribution is to introduce an alternative oligonucleotide-based surface chemistry for label-free DNA detection. The proposed surface chemistry is designed for use in combination with redox probes for electrochemical interrogation. However, SPR and QCM-D are also employed as complementary techniques to establish the most effective immobilizing conditions of probe-DNA and thymine spacer. We show that the capacity in discriminating between complementary and non-complementary DNA is significantly improved by employing compositionally optimized layers based on thymine spacers, as compared to sensing surfaces based on short organic spacer groups. Sensitivity optimization of the developed sensing surface is subject for future studies using different electrode configurations. 2. Experimental 2.1. Materials 2-mercaptoethanol (ME), 3-mercapto-1-propanol (MP), 6-mercapto-1-hexanol (MCH), Potassium Ferricyanide (III) (K3 [Fe(CN)6 ]), Sodium Chloride (NaCl), Magnesium Chloride (MgCl2 ) and Sulphuric acid (H2 SO4 ) were obtained from Sigma–Aldrich. Potassium Hexacyanoferrate (II)-3-hydrate (K4 [Fe(CN)6 ]·3H2 O) was obtained from Riedel-deHaën. 10 X Phosphate Buffer Saline (PBS) solution was obtained from 1st BASE Pte. Ltd., Singapore. Saline sodium citrate (SSC) buffer dry blend was obtained from Fluka, USA. Ammonia solution (25% NH4 OH), Hydrogen peroxide (30% H2 O2 ) and Hydrochloric Acid (HCl) were obtained from Merck-Chemical.1, 0.3 and 0.05 ␮m alumina (Al2 O3 ) suspensions were obtained from Allied High Tech Products, Inc., Singapore. Planar gold electrodes (diameter 2 mm), platinum wire and Ag/AgCl reference electrodes were obtained from CH Instruments, Inc., USA. 5 MHz AT-cut piezoelectric quartz crystal coated with 100 nm gold was obtained from Q-Sense. SPR gold chips were obtained from GE-Healthcare, Biacore division. All DNA sequences were bought either from 1st BASE Pte. Ltd., Integrated DNA Technologies (IDT), Inc. Probe-DNA sequence is complementary for MRSA (Methicillin-resistant Staphylococcus aureus) detection. (Please note that no experiment was done with genomic DNA or RNA from Staphylococcus aureus.) Sequence of disulphide probe-T9 DNA (from now on referred

to as probe-DNA): ATG ATT ATG GCT CAG GTA CTG CTA TCC ACC–TTT TTT TTT–O–(CH2 )3 –S–S–(CH2 )3 –OH Sequence of disulphide T9 spacer backfiller groups: TTT TTT TTT–O–(CH2 )3 –S–S–(CH2 )3 –OH Sequence of complementary MRSA target: (cDNA) 5 –GGT GGA TAG CAG TAC CTG AGC CAT AAT CAT–3 Sequence of non-complementary DNA: (NcDNA) 5 –CAA CCT CAA ACA GAC ACC ACG G–3 2.2. Solution preparation All solutions were prepared using deionized water (18 M  cm resistivity) from a Millipore Milli-Q system. Incubation Buffer (IB) comprised of 1 M NaCl. Hybridization Buffer (HB, pH 8.0) comprised of 4X SSC buffer. 0.01 M Phosphate Buffer (1XPBS, pH

272

Y. Yang et al. / Colloids and Surfaces B: Biointerfaces 125 (2015) 270–276

7.4) contained 137 mM NaCl and 2.7 mM KCl, diluted from 10XPBS. Electrolyte used for all electrochemical measurements comprised of 5 mM [Fe(CN)6 ]3−/4− (Redox probes) in 1X PBS. 2.3. Cleaning of planar gold surface Planar gold electrodes were first cleaned twice in RCA solution comprised of DI water, ammonia solution and hydrogen peroxide solution, in the ratio 5:1:1 at 80 ◦ C for 5 min. After rinsing thoroughly with DI water, the electrodes were polished sequentially with 1, 0.3 and 0.05 Al2 O3 suspensions and sonicated in DI water for 10 min. Finally the electrodes undergo electrochemical cleaning with potential cycling from 0 to 1.57 V (vs. Ag/AgCl) for 10 cycles at a scan rate of 0.06 V/s in 1 M H2 SO4 . After sonicating in DI water for about 2 min, the electrodes were dried in N2 gas and used immediately for DNA immobilization. 2.4. Cleaning of quartz crystal/SPR gold surface Quartz crystals and SPR gold surfaces were first cleaned twice in RCA solution comprised of DI water, ammonia solution and hydrogen peroxide solution, in the ratio 5:1:1 at 80 ◦ C for 5 min. The cleaned surfaces were then washed thoroughly in DI water, dried in nitrogen gas and used immediately. 2.5. DNA immobilization and hybridization Solutions for immobilization for both probe and spacer groups were diluted in IB into the required concentration from a stock concentration of 1 mM. For all experiments, probe-DNA concentration was fixed at 5 ␮M and incubation of gold substrates were undertaken overnight (>16 h) at room temperature in dark. Backfiller T9 concentrations were varied from 0 to 12.5 ␮M and the incubation time was 1 h at room temperature in dark. For sensing surfaces backfilled by organic spacer groups, 1 mM of ME, MP or MCH was used. This is a commonly-used concentration for backfilling with organic spacer groups [18,33,34], since Tarlov et al. reported a controlled surface coverage of probe-DNA with 1 mM of MCH which has shown to maximize hybridization efficiency [35]. Complementary and non-complementary DNA solutions were diluted to 1 ␮M in HB from a stock concentration of 100 ␮M and hybridization was performed for 2 h at 55 ◦ C. (Optimum hybridization temperature (Thyb = Tm − 25oC) calculated based on online Tm calculator – IDT Oligo Analyzer.) 2.6. Electrochemistry All cyclic voltammetry measurements were performed using Autolab PGSTAT 302N electrochemical analyzer. At each step, 7 ␮l of the probe or spacer group solution was dropped onto the working electrode using a micropipette and covered with an inverted Eppendorf tube to prevent drying of solution. After each incubation step, electrodes were rinsed in 1XPBS before cyclic voltammetry and impedance measurements. Cyclic voltammetry scans were performed from 0.7 V to −0.6 V at a scan rate of 0.06 V/s. AC impedance measurements were scanned from 105 Hz to 0.01 Hz at an amplitude of 0.01 V amplitude at 0.20 V. 2.7. QCM-D QCM-D measurements were done using Q-Sense E4 (Q-Sense, Sweden) with 4-channels IPC pump (Ismatec SA, Switzerland). A stable baseline was first established in buffer before the injection of DNA solution into the QCM-D chambers. 300 ␮l of probe-DNA or spacer group solution was then added and left overnight at room temperature in the chambers for self-assembly process. This was

followed by the injection of 300 ␮l of spacer group solution at the respective concentration and the backfilling process was allowed to occur for 1 h. NcDNA solution was first circulated for about 10 min or until the frequency change was stable. After rinsing with buffer for about 2–3 min, cDNA solution was added and circulated for about 10 min or until the frequency change was stable, followed by buffer rinsing. Solutions were allowed to circulate at a flow rate of 50 ␮l/min at all incubation steps. 2.8. SPR SPR measurements were done using Biacore 3000 (GEHealthcare, Uppsala, Sweden). 40 ␮l of probe or spacer group solution was dropped onto the SPR gold chip using a micropipette and allowed to assemble overnight at room temperature in a selfbuilt humidifying chamber to prevent drying of solution. Spacer group backfilling with the respective concentration was incubated for 1 h. After rinsing with 1xPBS, the SPR chip was loaded into the SPR setup and a stable baseline was first established in buffer. The SPR setup was programmed to allow successive flow of buffer, NcDNA and cDNA solutions to the sensing chip, with adequate rinsing time with buffer between each flow. Generally, 5 ␮l of buffer will be injected first, followed by 50 ␮l of NcDNA solution and finally 50 ␮l of cDNA solution. Solution was circulated at a flow rate of 5 ␮l/min and a dissociation time of 120 s was set at the end of cDNA hybridization. 3. Result and discussions Initial experiments were performed to optimize the two-step sequential assembly processes of probe-DNA and T9 (both in their disulphide form) on gold. Fig. 1 shows in (A) a schematic illustration of the different surface architectures obtained by sequential adsorption of probe-DNA and T9 at different relative concentrations ([probe-DNA]:[T9]), as well as the average current change upon hybridization (B) (See Supplementary Information for typical CV data in Figs. S1 and S2). The large jump in current change upon complementary DNA hybridization at surfaces prepared from solutions 1:0.5 to 1:1 (see Fig. 1B) is attributed to a change in orientation of the immobilized probes on the surface due to an increasing amount of spacer groups, which not only affects the accessibility of the probes to hybridization but also the efficiency of subsequent electron transfer. More explicitly, with an increasing amount of T9, the average distance between the probe-DNAs increases at the same time as they rearrange from a flat to an upright orientation. These two phenomena produce a more open sensing surface with favorable electron transfer characteristics. This is in agreement with literature articles [13,14,28,30] that report an increase in current flow after complementary DNA hybridization, which most likely is due to the formation of the more rigid and rod-like double-stranded DNA structure that exposes the electrode surface for efficient electron transfer. Furthermore, with the use of disulphide DNA strands where each probe-DNA and T9 spacer group are accompanied by a short organic propanol strand, separation of neighboring probe strands is likely to increase. This is advantageous in terms of reducing steric hindrances for better target recognition/hybridization and for improving the channeling properties of the sensing layers by the redox probes, which otherwise may not be possible in the usual protocol where linear thiols are used, since the sensing surface formed are likely to be too dense for efficient channeling. In order to show the capacity of each ratio of probe-DNA and T9 spacer to distinguish between the cDNA hybridization and NcDNA interaction, the signal-to-noise ratio (S/N) is being calculated (See Supplementary Information for actual calculations). Note that in

Y. Yang et al. / Colloids and Surfaces B: Biointerfaces 125 (2015) 270–276

273

Fig. 1. (A) Schematic illustration of the sensing surface sequentially immobilized with different relative concentrations of probe-DNA and backfilling thymine (T9) spacers. Generally, when there is no T9 spacer group, probes (black lines) are not aligned and some of the probes can lie on the gold surface (a). As T9 concentration increase, probes become more aligned as shorter T9 spacer group (long /short straight lines) are newly immobilized on the exposed gold surface ((b) and (c)). (B) Average current change due to hybridization for both NcDNA and cDNA on surfaces with varying ratios of disulphide probe-DNA and disulphide T9 spacer group. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

this paper, “noise” is defined as the signal obtained after NcDNA interaction, since this signal change is expected to reveal the nonspecific interactions on the gold surface. Fig. 2 shows that the S/N ratio increases from a negative value at 1:0 via values near zero for intermediate ratios to a maximum value at the ratio 1:1 (S/N = 5.9), after which the S/N drops significantly and levels out at a plateau for the ratios 1:1.5–1:2.5. Such a trend has also been observed by Boozer et al., who reported maximum SPR response at a specific probe DNA:OEG (oligo(ethylene glycol)) ratio [36]. Based on their results, a higher or lower concentration of OEG backfiller were shown to decrease the SPR response, which they attributed to a too high surface coverage of the backfillers at high OEG concentration, and to a poor orientation of the probe-DNA, that restricted hybridization, at low OEG concentration. In Fig. 3, there is a switch from negative S/N values at 1:0 and 1:0.25, to a positive one at 1:0.5. The regime between 1:0.25 and 1:0.5 defines the minimum relative concentration of T9 spacer groups required to cause a change in the alignment/presentation of the probe on the electrode surface. When the surfaces are incubated with ratios 1:0.5 and above, there is a sufficient amount of T9 spacer groups available to “lift” the probes to an upright orientation by removing

non-specific interaction and filll empty areas of the gold surface. The probe-DNAs are not only more aligned than those formed from lower ratios (e.g. 1:0 and 1:0.25), they are also, on average, separated further apart from the neighboring probes by the T9 spacer. This reduces steric hindrances and results in an increased accessibility of the target cDNA to hybridize [37–39], as well as for NcDNA to interact with the aligned probes. The negative S/N values seen at ratio 1:0 and 1:0.25 are due to the negative current change caused by interaction with NcDNA, Fig. 1. This observation suggests that NcDNA has very little or no direct interaction with the unaligned probes. However, since the probes are unaligned, with possibly higher steric hindrance, the actual hybridization with cDNA is also reduced at these ratios. Even though a negative current change for NcDNA binding is more desirable than a positive one, the S/N values in Fig. 2 show that the best ratio to discriminate between the NcDNA binding and cDNA hybridization occurs at 1:1. This is confirmed by SPR analysis, where the highest response corresponding to the greatest amount of cDNA remaining on the surface after rinsing, occurs at ratio 1:1 (see Supplementary Information for typical SPR sensorgrams in Fig. S3 and overall change in SPR response in Fig. S4). QCM-D studies furthermore show that the dissipation changes

274

Y. Yang et al. / Colloids and Surfaces B: Biointerfaces 125 (2015) 270–276

Fig. 2. Signal-to-noise ratio (see Supporting Information for details) upon exposure to a 1 ␮M cDNA in 4XSSC for varying ratios of disulphide probe DNA and disulphide T9 spacer group.

remain low (close to zero) after the exposure of NcDNA to surfaces prepared with ratios up to 1:1 (see Supplementary Information for typical QCM-D graphs in Fig. S5 and overall dissipation changes in Fig. S6). Upon cDNA hybridization, the formed rigid double helices are likely to extend out into the solution where they can trap solvent causing an increase in viscoelasticity that result in higher dissipation changes. This is unlike the observations seen at higher ratios (e.g. 1:1.5 and above), where dissipation changes are high even after NcDNA injection. High frequency changes (f ≈ −25 Hz) solely upon cDNA hybridization observed across the whole ratio range show surfaces that are highly selective, since negligible frequency changes are observed after NcDNA injections (Fig. S7). In order to compare the performance of DNA-based spacer group with the commonly used organic spacer groups, the hybridization experiments were conducted and compared for sensing surfaces first immobilized with probe-DNA and backfilled with (i) ME, (ii) MP, (iii) MCH, and (iv) T9 (1:1). Results from cyclic voltammetry scans show varying average current change after the Au surfaces are immobilized with the probe-DNA and the respective spacer groups, as seen in Fig. 3. The DNA surface that is incubated with T9 displays the lowest average current change. (Please note that the real time binding data from SPR and QCM-D reveals lower affinity of T9 to Au surface, as compared to using an organic spacer group.) This signifies a

Fig. 4. Signal-to-noise ratio (see Supporting Information for details) with 1 ␮M cDNA in 4XSSC incubated at optimized hybridization conditions (55 ◦ C for 2 h) on sensing surface immobilized with disulphide probe-DNA and backfilled with various spacer groups.

sensing surface with enhanced electron transfer across the immobilized layer. MCH, having the longest alkyl chain among the three organic spacer groups, is known to form a sensing surface which blocks the accessibility of the redox probes to the gold surface most efficiently, resulting in the highest average current change. From Fig. 4, the addition of organic spacer groups to the probeDNA surface generally results in a higher average current change as compared to T9. This higher surface blockage after addition of the organic spacer groups is partly due to their–OH head groups, which recently was reported to be responsible for the removal of non-specific interactions between the DNA bases and gold surface by repelling the negatively charged DNA bases from the surface [24,34]. As mentioned above, MCH has the longest alkyl chain with 6 carbons, while MP has a 3 and ME 2 carbons, respectively. Short-chain spacers, such as ME, are known to promote the direct diffusion of the redox probes through the SAM layer to yield increased electron transfer across the surface [13]. It is therefore reasonable to conclude that the decreasing average current change, which is interpreted as an increasing probability of electron transfer across the interface, reflects the differences in length of MCH, MP and ME. This is also in agreement with the difference in the peak-to-peak potential (Epp ) for the different sensing surface, where the largest Epp , which is indicative of a highly compact layer with reduced access of redox probes and low electron transfer [26,40], is obtained for MCH, Table 1. With the use of T9, both the average current change (Fig. 3) and Epp (Table 1) points toward a less compact sensing surface with high electron transfer. This can be attributed to the difference in the charge and hydrophobicity of the T9 with respect to the other organic spacer groups. Dharuman et al. showed that the highly charged and hydrophilic COOH head group in 3mercaptopropionic acid is likely to repel the negatively-charged Table 1 Average peak-to-peak potential (Epp ) obtained from cyclic voltammetry scans after incubation of Au electrode surfaces with immobilized probe-DNA in solutions containing various spacer groups.

Fig. 3. Average current of immobilized disulphide probe DNA after backfilling with various types of spacer group on planar gold surface.

Type of spacer group

Average Epp (mV)

ME MP MCH T9

312 394 467 331

Y. Yang et al. / Colloids and Surfaces B: Biointerfaces 125 (2015) 270–276

probe-DNA more effectively than the more hydrophobic MCH, thus creating more exposed (open) Au surface for electron transfer via the [Fe(CN)6 ]3−/4− redox probes [24]. This results in a much smaller peak current change and Epp value than that obtained by MCH. Similarly, when T9 is used, smaller values of the average current change and Epp are obtained. The negatively-charged T9 spacer layer, which is highly hydrophilic, is likely to repel the probe-DNAs from the surface more effectively than the other organic spacer groups used, thus successfully removing most of the non-specific interactions between the probe-DNAs and gold. At the same time this opens up the surface and exposes easily accessible domains between the erected probe-DNAs. This is less achievable by the organic spacer groups due to their lower efficiency in repelling the DNA bases from the gold surface as compared to the T9 spacer. Moreover, since both the probe-DNA and T9 are of similar hydrophilicity, multiphase separation is less likely to occur, which is advantageous as it leads to a more homogenous sensing surface. Even though a more compact surface is commonly more desirable for an electrochemical biosensor, it can be seen in Fig. 4, that the use of T9 as backfilling spacer group results in an increase of the S/N ratio by approximately a factor of 10 as compared to those that are backfilled with commonly used organic spacer groups. This is a significant improvement that points toward a more selective sensing surface when T9 is used as backfilling agent. We also evaluated surfaces prepared from probe-DNA and filling molecules having short thymine strands (T6) and it was found that shorter strands had a detrimental effect on the trend of the current changes upon incubations in cDNA and NcDNA. This is most likely due to increasing interactions between non-thymine DNA bases and the gold surface leading to a larger probe-DNA footprint and a less accessible probe for hybridization. As mentioned earlier, T9 results in a less compact and highly charged surface. Since the redox probes are also negatively charged, electrostatic repulsion between the redox probes and the highly charged surface may be regarded as highly undesirable for the electron transfer and electrochemical detection. This appears, however, to be contrary to the results obtained herein. Thus, there must be another contribution that results in the favorable electron transfer process for the T9 surface. If one considers a case where the surface is compactly filled with organic spacer groups and the amount of intrinsic defect sites in the layer is small. When a large fraction of the probes on the surface are hybridized with the cDNA targets, the formation of oriented and rigid double helices will expose easy accessible domains for efficient channeling and electron transfer. With NcDNA interaction, some strands may also be lifted (oriented) to increase the electron transfer, though to a much smaller extent. At a more charged surface where defect sites are likely to be more abundant (due to in-plane repulsive interactions), as deduced experimentally when using T9 as spacers, the channeling is enhanced via these defects post cDNA hybridization. Once again, NcDNA interaction may lift some of the strands, but the increase in electron transfer will be lower with respect to that caused by the cDNA hybridization. Overall, the calculated S/N ratio for a charged and defect rich surface will be much higher than that of the uncharged and more densely packed surface architecture. A prerequisite for efficient channeling is that the probe-DNA is readily accessible for hybridization with cDNA to adopt an erected “vertically aligned” orientation. Ideally the probe-DNA should appear well-separated from each other as in a statistically mixed assembly. It is well-know from numerous studies of self-assembled monolayers (e.g. thiols on gold) that the larger the difference is between the molecules used (structurally and chemically) in mixing and backfilling experiments the harder it is to control the final outcome of the experiments in terms of phase behavior and composition. In a mixing experiment the on rate

275

varies with the size and the chemical nature of the two molecules competing for the binding site on the surface. In a backfilling experiment the exchange and filling rates are also strongly dependent on the size and chemical properties of the molecules involved and the sequence by which they are exposed to the solid support. Thus, the final outcome piles down to adsorption or exchange/filling kinetics. A mixing(backfilling) experiment between a long oligonucleotide (as our probe-DNA) and a short alkyl thiol will for sure result in a phase segregated monolayer of unpredictable composition and domain size. On the other hand, if the probe-DNA is mixed(backfilled) with a short spacer molecule, and if the initial (leading) sequence of the probe-DNA is identical with the one of the short filling spacer (T9 in our case), then the likelihood of obtaining a statistically mixed (or at least less phase segregated) layers increases substantially. Thus, the use T9 oligonucleotide spacers as backfillers to reduce the risk of forming large-scale phase segregated layers seems from that point of view reasonable and justified. It is important to emphasize though that we have very little information of the exact phase behavior of the domain size and phase behavior of our T9 backfilled layers. However, as can be seen from our functional test the hybridization efficiency (S/N) is a factor or 10 higher for the T9 layers as compared to those containing short organic spacers, Fig. 4. We interpret this observation as supporting evidence for a less phase segregated layer with more accessible probe-DNA for hybridization with the cDNA. Thus, despite the general idea that a highly charged surface can deter electron transfer and reduce the selectivity of the sensing surface, the use of T9 at the optimized [probe-DNA]:[T9] ratio of 1:1 offers improved capability in discriminating between NcDNA binding and cDNA hybridization, making DNA-based spacer groups a better alternative than the short organic spacers. 4. Conclusion A sensing surface for label-free detection of DNA has been prepared using thymine-based spacer groups. Sensing layers based on such spacer groups offer an attractive alternative to commonly used organic spacer groups. A layer with optimized performance (S/N ratio) is obtained for a sensing surface prepared from a two-step sequential process of immobilizing the probe-DNA and backfilling with T9 spacer group at a 1:1 ratio. At this optimized ratio, the ternary sensing layer, comprising of T9-terminated probe-DNA, T9 spacer groups and propanol, is demonstrated to be adequately packed and spaced apart, having probe-DNAs sufficiently aligned toward the contacting target solution for favorable discrimination between complementary and non-complementary hybridization. Compared with sensing surfaces backfilled with commonly used organic spacer groups, namely 2-mercaptoethanol, 3-mercapto1-propanol or 6-mercapto-1-hexanol, T9 forms a sensing surface that is approximately 10 times more selective. Our findings provide insights into the use of oligonucleotide-based spacer groups, specifically thymine groups, for the design of future DNA sensing architectures Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.colsurfb. 2014.10.058. References [1] [2] [3] [4]

F. Lucarelli, et al., Biosens. Bioelectron. 19 (2004) 515. J. Chen, et al., Biosens. Bioelectron. 26 (2010) 144. X. Li, et al., Anal. Chim. Acta 582 (2007) 158. J. Wang, Anal. Chim. Acta 469 (2002) 63.

276 [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22]

Y. Yang et al. / Colloids and Surfaces B: Biointerfaces 125 (2015) 270–276 E. Palecek, Talanta 56 (2002) 809. J.M. Gong, et al., Biosens. Bioelectron. 24 (2009) 2285. A.C. Barton, et al., Biosens. Bioelectron. 24 (2009) 1090. Y.K. Ye, et al., Biosens. Bioelectron. 18 (2003) 1501. E.E. Ferapontova, et al., Chem. Commun. 46 (2010) 1836. M. So, et al., Chem. Commun. 36 (2008) 4354. L.M. Wang, et al., Biosens. Bioelectron. 26 (2011) 2835. E. Huang, et al., Langmuir 17 (2001) 1215. S.L. Pan, L. Rothberg, Langmuir 21 (2005) 1022. V. Dharuman, et al., Biosens. Bioelectron. 21 (2005) 645. H. Peng, et al., Biosens. Bioelectron. 20 (2005) 1821. R. Jin, et al., J. Am. Chem. Soc. 125 (2003) 1643. R. Levicky, et al., J. Am. Chem. Soc. 120 (1998) 9787. G.D. McEwen, et al., Anal. Chim. Acta 643 (2009) 26. W. Xu, et al., Small 3 (2007) 2011. A.A. Karyakin, et al., Anal. Chem. 72 (2000) 805. A. Kausaite-Minkstimiene, et al., Anal. Chem. 82 (2010) 6401. A.B. Steel, et al., Anal. Chem. 70 (1998) 4670.

[23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40]

M.A. Keniry, et al., Nucl. Acids Res. 25 (1997) 4389. V. Dharuman, J.H. Hahn, Sens. Actuators B: Chem. 127 (2007) 536. M.D. Porter, et al., J. Am. Chem. Soc. 109 (1987) 3559. A.-E. Radi, et al., Anal. Chem. 77 (2005) 6320. v. Dharuman, et al., Biosens. Bioelectron. 25 (2010) 2129. V. Dharuman, J.H. Hahn, Biosens. Bioelectron. 23 (2008) 1250. N. Park, J.H. Hahn, Anal. Chem. 76 (2004) 900. S. Campuzano, et al., Biosens. Bioelectron. 26 (2011) 3577. L.M. Demers, et al., J. Am. Chem. Soc. 124 (2002) 11248. M. Ostblom, et al., J. Phys. Chem. B 109 (2005) 15150. A.W. Peterson, et al., J. Am. Chem. Soc. 124 (2002) 14601. J. Wu, et al., Anal. Chem. 82 (2010) 8830. T.M. Herne, M.J. Tarlov, J. Am. Chem. Soc. 119 (1997) 8916. C. Boozer, et al., Langmuir 22 (2006) 4694. F. Xu, et al., Thin Solid Films 516 (2008) 8634. Z. Gagnon, et al., Electrophoresis 31 (2010) 666. J. Zhang, et al., Langmuir 28 (2012) 6494. T.H. Degefa, J. Kwak, J. Electroanal. Chem. 612 (2008) 37.