Enhanced nucleotide mismatch detection based on a 3D silicon nanowire microarray

Melania Banu*a, Monica Simiona, Attila C. Ratiub, Marian Popescua, Cosmin Romanitana, Mihai Danilaa, Antonio Radoia, Alexandru Al. Ecovoiub and Mihaela Kuskoa
aNational Institute for Research and Development in Microtechnologies – IMT Bucharest, 126A Erou Iancu Nicolae Street, 077190, Bucharest, Romania. E-mail: ​melania.banu@imt.ro
bDepartment of Genetics, University of Bucharest, 1-3 Intrarea Portocalelor Street, 060101, Bucharest, Romania

Received 21st July 2015 , Accepted 27th August 2015

First published on 27th August 2015


Abstract

We explored the capacity of a nanostructured silicon substrate to be used in microarray technology for nucleotide mismatch detection. The aim of our study was to investigate single nucleotide polymorphisms (SNPs) with pathological significance within breast cancer 1, early onset gene (BRCA1). Herein, we report an easy, one step and low cost method to obtain a layer of silicon nanowires (SiNWs) as a starting point for the development of a microarray platform. The experimental analyses indicate that, in comparison with the usually employed commercial glass support, the new microarray support significantly improves the hybridization signal quality and also the sensitivity of SNP detection. More accurately, the developed platform allows a statistically significant discrimination of the hybridization signal intensities when the DNA probes differ by the presence of one mismatch in a selected position. These achievements are attributed to the facilitated target molecules' access to the attached probes due to the three-dimensional (3D) architecture of the support and to the antireflective properties of the SiNWs substrate.


1. Introduction

Genetic modifications which trigger different types of cancer tend to affect mainly three categories of genes: proto-oncogenes, tumour-suppressor genes and genes that are involved in DNA repair. BRCA1 is a tumour suppressor gene, being one of the genes responsible for inherited breast cancer syndromes. It encodes a nuclear protein, which contributes to maintaining the stability of the genetic material.1 Recently, genome-wide association studies were conducted using a large number of common SNPs in order to identify the links between them and disorders which rely upon different patterns of linkage disequilibrium in the human genome.2

Routine investigation methods, such as gel electrophoresis assays, sequencing, polymerase chain reaction (PCR) assays for mutational analyses3–5 and high resolution melting analyses (HRM) were used for BRCA1 and BRCA2 genotyping in clinical practice.6,7 PCR – restriction fragment length polymorphism (PCR – RFLP) technique is not suitable for the simultaneous analysis of a large number of point mutations, due to the requirement of a specific primer pair and restriction enzyme used for each SNP.8 The next generation sequencing technique is time and resource-consuming and the alignment algorithms should be robust enough to overcome the inherent sequencing errors.9 Undoubtedly, HRM is a simple, rapid and inexpensive technique, but it strongly depends on high-quality PCR outcome, fluorescent dyes and specific instruments.10

The SNP arrays represent an alternative to sequencing and were initially developed in 1998 to genotype human DNA, having a major role in identification of DNA variants which cause different disorders.11 Due to the complex genetic alterations in cancer cells, the SNP array analysis has found application in this area. Researchers from Affymetrix have developed whole-genome sampling method for SNP genotyping.12 Several SNP arrays also have applications in forensics and represent a suitable method to analyze many single nucleotide polymorphisms in parallel.13 DNA microarrays consist in large numbers of probes, represented by single-stranded DNA (ssDNA) molecules which are attached to a solid surface and hybridize to target nucleic acids, labelled with fluorescent dyes, usually from cyanine family (Cy3 and Cy5). Cy3 and Cy5 have sharp absorption bands and high extinction coefficients. Cy3, in contrast to Cy5, has a better resistance to photobleaching. Thus, the Cy3-labelled DNA oligomers are highly fluorescent and even single hybridized molecules can be observed.14,15

The main advantages held by arrays are: low cost, high-throughput capabilities, miniaturization, speed and parallelism.16 Nevertheless, a drawback of this method is given by the use of a flat support, which limits the amount of probes covalently attached to a chemically modified surface, leading to a poor detection. Even nowadays, relatively little is known about how the hybridization process is influenced by the surface configuration and various simulations were conducted in comprehending the microarray platforms' contribution.17–19

Galbiati et al.,20 have reported the development of a microarray platform on oxidized silicon (Si) substrate for detection of KRAS and BRAF mutations, where the optical properties of the substrate have improved the detection sensitivity by enhanced fluorescent signals. Other types of platforms for both DNA and protein microarrays were described, such as nano- and macro-porous silicon surfaces, fabricated by classical electrochemical etching of Si. These platforms have been used for studying the C-reactive protein (CRP) immobilization21 for Human Papilloma Virus (HPV) detection,22 respectively. It was demonstrated that an improved quality of the microarray spots has been achieved by reducing the frequent “coffee-ring effects”. Hence, the 3D porous structures have been successfully used to detect low concentrations of analytes, in the pM range. Murthy et al.23 has used ordered, high aspect ratio nanopillar arrays on Si surfaces and demonstrated the higher probe immobilization capacity and improved target accessibility. Later on, Dawood et al.24 have reported an enhanced signal-to-noise ratio detection of DNA targets, by using nanostructured silicon-nanowire microarray platforms.

Metal assisted chemical etching (MACE) of Si was reported in 1997 and represents a low cost and versatile method for fabricating large scale silicon nanowire (SiNWs) arrays.25,26 The morphological characteristics of SiNWs, such as orientation, length and diameter can be easily controlled by this method.27,28 MACE includes two processes, the nucleation of metal catalysts and anisotropic etching. In one-step MACE (1-MACE), the etchant solution contains both metal salts and hydrofluoric acid (HF), allowing these two steps to take place simultaneously. In two-steps MACE (2-MACE), the metal catalysts are firstly deposited onto the wafer surface by electroless plating and the etching occurs in HF/H2O2 solution. 1-MACE has not been widely used, even though it's a more simple operation, due to the fact that the doping level of silicon represents a strict prerequisite. Li et al.25 reported the fabrication of p-type porous SiNWs through 1-MACE while earlier, To et al.29 have claimed the fabrication of mesoporous SiNWs, using the same technique.

In this paper, we provide experimental data for the sensitivity and specificity of hybridization on the novel nanopatterned silicon substrate obtained by 1-MACE process – 3D microarray platform – in contrast with the 2D commercial glass substrate. As an experimental model, we performed an analysis of the DNA sequences which correspond to BRCA1 polymorphisms in the rs28897696 locus. Since another issue in microarray is given by a poor reliability due to hybridization signal intensity variation,30,31 we focused our study on assessing the discrimination efficiency between the hybridizations involving perfect-matched (PM) and mismatched (MM) DNA sequences relative to the canonical probe. We restricted the possibility of secondary structure formation which may hinder the hybridization efficiency, by carefully choosing the probes and target sequences. Regarding the effects of the mismatches in terms of specificity, the canonical probe sequence (containing the common C allele) has perfect complementarity with the target molecule. The other three variants of BRCA1 probe sequences were designed to have the C nucleotide replaced with another nucleotide (C > A, C > G and, respectively, C > T). We demonstrate that in comparison with the usually employed commercial glass support, the nanostructured silicon platform presents an enhanced sensitivity – high signal-to-noise ratio (SNR) – and qualitative nucleotide change discrimination for SNP detection.

2. Experimental section

2.1 Materials and reagents

Hydrofluoric acid (HF), nitric acid (HNO3), silver nitrate (AgNO3), ethanol, phosphate buffered saline (PBS), (3-aminopropyl)triethoxysilane (APTES), glutaraldehyde (GA), sodium dodecyl sulfate (SDS), citrate sodium saline (SSC) 20×, bovine serum albumin (BSA) and Aluma seal 96 film were provided by Sigma-Aldrich (Germany). HPLC water and coverslips were supplied by Roth. The sterile microplates were obtained from BRAND (Germany).

The p-type silicon wafers were ordered from SILTRONIX (France) and Si-Mat (Germany).

The Superaldehyde 3 Premium Microarray Substrates and UniHyb hybridization buffer were provided by ArrayIt Corporation (Sunnyvale, USA).

The design of the probes and target oligonucleotides was performed starting from the BRCA1 gene sequence indexed in NCBI and were purchased from Biomers.net (Germany).

2.2 Technological workflow for microarray platform fabrication

The fabrication flow of the 3D microarray platform on Si was accomplished by obtaining firstly the SiNWs substrates, followed by surface functionalization in order to promote a covalent attachment of the DNA sequences.32 The BRCA1 oligonucleotide variants (probes) were bound to the commercial glass supports and to the experimental Si support. After probe tethering, the hybridization was carried out using target oligonucleotides, and consecutively the fluorescent signal was detected and analyzed. For a better overview, the technological workflow is schematically presented in Fig. 1.
image file: c5ra14442f-f1.tif
Fig. 1 Technological workflow.
2.2.1 Silicon nanowires (SiNWs) preparation and functionalization of the substrate. p-Type Si (100) wafers with 1–3 Ω cm resistivity were used as starting substrate for 3D Si based supports and diced into 2 × 2 cm2 samples. The silicon nanowires (SiNWs) were fabricated through one-step metal assisted chemical etching (1-MACE) in 0.06 M AgNO3 and 4.5 M HF, for 40 minutes (min) under dark conditions.

After the etching process, the structures were dipped into 60% v/v HNO3 solution, for 30 min, in order to remove the generated silver dendrites and, as a final step, immersed in 5% v/v HF solution for 1 min. The nanostructured substrates were firstly hydroxylated in H2SO4[thin space (1/6-em)]:[thin space (1/6-em)]H2O2 3[thin space (1/6-em)]:[thin space (1/6-em)]1 v/v (Piranha solution) for 5 min, rinsed thoroughly with deionized water, and then placed in 5% v/v APTES solution prepared in anhydrous ethanol for 6 hours (h) to allow the chemical surface modification to take place. The structures were afterwards rinsed thoroughly in anhydrous ethanol, dried under a N2 stream and cured for 1 h at 110 °C.

Based on our previous results on functionalization of porous silicon substrates,33 we slightly modified the APTES functionalization protocols. Additional functionalization was done by immersing the samples in 2.5% GA/ultrapure water solution for 5 hours, in order to provide –CHO functional groups.22,34 The substrates were rinsed thoroughly with deionized water, dried under a N2 stream and kept into N2 atmosphere until further use.

2.2.2 Design of the microarray experiment, probe immobilization and hybridization. For the present study, ssDNA molecules and complementary target oligonucleotides corresponding to BRCA1 were employed, the sequences being presented in Table 1. The oligonucleotides used in this work were chosen in accordance with a SNP locus containing normal and putative pathogenic alleles. Rs28897696 is a SNP known for a relatively rare missense variant (C > A) which is a single nucleotide change that results in a codon which encodes for a different aminoacid (Ala > Glu). This SNP was among the top 10 likely to lead to breast cancer.35 The single nucleotide changes simulated in our experiments were written in the following format: C > A, C > G and C > T. Herein, we focus on the detection of the effects of A/G, G/G and T/G base pairs mismatches on hybridization efficiency, comparative to the canonical C/G pair.
Table 1 Probe and target sequences
BRCA1
Normal/perfect match sequence (PM) 5′–C6NH2–CTAGGAATTGCGGGAGGAAAATGGG–3′
Sequences with 1 mismatch (MM) C > A 5′–C6–NH2–CTAGGAATTG[A with combining low line]GGGAGGAAAATGGG–3′
C > G 5′–C6–NH2–CTAGGAATTG[G with combining low line]GGGAGGAAAATGGG–3′
C > T 5′–C6–NH2–CTAGGAATTG[T with combining low line]GGGAGGAAAATGGG–3′
Complementary sequence (C) 5′–Cy3–CCCATTTTCCTCCCGCAATTCCTAG–3′


The probes and target sequences were designed to avoid as much as possible the formation of secondary structures (such as hairpins, self-dimers or hetero-dimers), using IDT OligoAnalyzer 3.1 software36 and the probability of the PM probe to form hairpin structure is presented in Fig. S1 (ESI). Also, the data concerning the probabilities of probe self-dimer formation are summarized in Table S1 (ESI).

The BRCA1 probes were designed to have a C6 aminolink modification at 5′ end and to contain up to three different types of single mismatches, placed near the centre of the sequence. The probes containing mismatches were designed in accordance with the SNPs with pathogenic and likely pathogenic allele, submitted in “The Single Nucleotide Polymorphism Database (dbSNP) of nucleotide Sequence Variation”.37,38 The complementary target sequence has a Cy3 dye attached to the 5′ end. Cyanine dyes are used in hybridization detection by labeling a target molecule. Besides that, their second function is to stabilize the formed DNA duplex.14

The design of the microarray experiments took into consideration probe concentrations ranging from 1 to 150 μM, linked to the commercial support, in order to optimize the hybridization protocol. Four replicates were spotted for each probe at a given concentration, in order to have a reliable analysis of the fluorescent intensities, after hybridization. The solution containing the target's complementary sequences was diluted to a concentration of 10 μM.

After analyzing the hybridization signal intensities obtained with the commercial support, we further analyzed the hybridization behaviour on the SiNWs support. In order to immobilize the probes on the aforementioned support, four concentrations were chosen: 1 μM, 10 μM, 25 μM and 50 μM, respectively. A solution of 10 μM target sequences was further used for hybridization.

The probe solution was prepared in phosphate buffered saline (PBS), 7.4 pH, 10 mM, the spotting being achieved with an Omni Grid Micro contact printer (Genomic Solutions). The humidity in the printer chamber was 80% and the temperature was set to 25 °C. The diameter of the pin tips used for spotting was 200 μm. For controlling the pin contact force and time, the following motion parameters of the plotter setup were established: acceleration at 2000 mm s−2 and the velocity ranged between 10 and 170 mm s−1.

After deposition, the ssDNA molecules were allowed to attach covalently to the surfaces, by incubating them in humid atmosphere (80%) at room temperature, overnight. After immobilization, an essential step was achieved by the removal of the unbound probes and blocking of the unreacted sites. The washing steps were performed for three times in solutions containing: (i) 2× SSC (saline sodium citrate)/0.1% w/v SDS (sodium dodecyl sulfate), (ii) 1× SSC, (iii) DIW (deionized water). The blocking step is needed for two main reasons: (i) to prevent the unspecific binding of the labelled targets and (ii) to lower the background auto-fluorescence. It was performed by immersing the microarray platforms for 1 h, at 42 °C and 450 rpm, in 1% w/v BSA solution prepared in 5× SSC and 0.1% w/v SDS, and preheated, for 30 min, at 42 °C.

DNA hybridization was achieved in a humid case to prevent the target evaporation. The labelled DNA strand molecules were prepared in UniHyb buffer/ultrapure water. Each microarray platform was encased with a coverslip to obtain a uniform dispersion of the solution. Prior to scanning, the microarrays were incubated at room temperature, for 3 hours, and washed to discard the un-hybridized target sequences, with the same solutions, at the same parameters used after the immobilization step.

2.3 Investigation methods

Morphological characterization and measurements were performed on the nanostructured silicon substrates employing a high resolution Field Emission Gun-Scanning Electron Microscopy (FEG-SEM), FEI-NOVA NanoSEM 630.

The crystalline structure of the nanostructured silicon samples was investigated by X-ray diffraction methods using a 9 kW RigakuSmartLab triple-axis rotating anode thin film diffractometer operated at U = 45 kV and I = 200 mA, using a parallel beam. The crystallinity state of the SiNWs was studied using two X-ray diffraction methods (XRD). The GIXRD method uses a fixed small incidence angle (ω = 0.5°) and a detector scan (2θ-scan).

The hybridization detection was carried out using a laser scanning confocal fluorescence system (GeneTAC UC4 Microarray Scanner, Genomic Solutions). The arrays were scanned at a photomultiplier tube (PMT) gain range of 35–45% for Cy3 (532 nm) and 40–50% for Cy5 (635 nm) at 5–10 μm resolution. The raw images were imported into GeneTac Integrator software for spot detection and quantification of the hybridization signals. The generated intensity values for each hybridized spot were normalized by log2. The normalized spot intensities were then statistically analyzed in GraphPad Prism 5,39 using two-tailed Student's t-test with Welch's correction, and OriginPro 8.0 softwares.

3. Results and discussions

SEM images of the SiNWs samples produced by 1-MACE method are illustrated in Fig. 2, where an approximate length of 3.5 μm and a diameter ranging from 30 to 95 nm were estimated for the silicon nanowires. The upper extremities tend to bundle, and this phenomenon might be determined by their increased length or by the capillary and short-range van der Waals forces,24 when the sample is left to dry after etching.
image file: c5ra14442f-f2.tif
Fig. 2 (a) Cross sectional and (b) surface SEM images of SiNWs.

The GIXRD scans have highlighted the polycrystalline character of the SiNWs (Fig. 3). Additionally, the texture of the samples was studied by IP-PF measurements. Different lattice planes are present in the IPXRD spectra: (220), (004), (111) and (113). The (220) SiNWs in-plane texture correspond to the vertically grown Si NWs. The additional (111) and (113) SiNWs lattice planes surface correspond to bended SiNWs which are randomly orientated and tend to get conjoined. The XRD results are consistent with the SEM characterization, pointing to a uniform surface of strongly ordered (mostly vertical) SiNWs.


image file: c5ra14442f-f3.tif
Fig. 3 XRD pattern of the SiNWs substrate.

3.1 Calibration tests on commercial functionalised glass support

In the first set of experiments carried with BRCA1 oligonucleotides, preliminary calibration tests were conducted on commercial glass supports in order to optimize the hybridization protocol. In Fig. 4, the hybridization results can be examined together with the array configuration, with the immobilized probe concentrations and types. For each probe concentration, four technical spot replicates were printed, revealing the equivalent number of hybridization signals. More exactly, the left columns correspond to the perfect-matched (PM) canonical hybridized probes, the second ones with the C > A mismatched (MM) probes, the third ones to the C > G MM tethered and hybridized probes, and the last ones to the C > T MM hybridized probes. The experiments were performed using six different probe concentrations tethered on the platform (1 μM, 25 μM, 50 μM, 75 μM, 100 μM and 150 μM) and two different concentrations of target molecules: 0.1 μM – Fig. 4(a) and 10 μM – Fig. 4(b), respectively.
image file: c5ra14442f-f4.tif
Fig. 4 Hybridization results of (a) arrays hybridized with 0.1 μM target molecules; (b) arrays hybridized with 10 μM target molecules.

Fig. 4(a) illustrates the hybridization results with 0.1 μM target concentration. At 1 μM probe concentration, the spots corresponding to the hybridized probe variants which carry one mismatch appear to be smaller than those assigned to PM hybridized probes. For the other concentrations of tethered probes (25 μM, 50 μM, 75 μM, 100 μM and 150 μM), large variations between the diameters of the spot replicates can be observed. Moreover, the statistical analyses (t tests) clearly confirmed these observations, showing that these concentrations did not offer reliable results.

Fig. 4(b), illustrates the hybridization results with 10 μM target concentration. One can observe that for 1 μM of immobilized probes, the hybridization intensities are too low, whereas the diameters for the spot replicates appear to be similar. For a concentration of 25 μM probes, the signal intensity for the C > A technical spot replicates seems to be higher than for the hybridized PM spot replicates. An increase in probe concentration to 50 μM leads to the amplification of PM signal intensity. However, the differences between the intensity values generated after the hybridization of target molecules with PM and C > A probes are not significant (p = 0.9073). Nevertheless, by using probes' concentration of 75 μM, the signal intensities are unreliable, because the signal intensity is supersaturated for the hybridized C > A probes. A higher probe concentration leads to a decrease of signal intensity, being a consequence of the steric hindrance which arose on the microarray surface.40 This phenomenon is consistent with the coarse-grain simulation studies of Welling and Knotts.31

The rows of spots highlighted with red borders in Fig. 4(a) and (b), where further subjected to statistical analyses. The log2 signal intensities for different probe-target concentrations are shown in Fig. 5.


image file: c5ra14442f-f5.tif
Fig. 5 Assessment of the signal intensity trends between the C:G (PM probe) base pairs and the impairs represented by A:G (C > A probe), G:G (C > G probe) and T:G (C > T probe). Results of (a) arrays hybridized with 0.1 μM target molecules; (b) arrays hybridized with 10 μM target molecules.

The results clearly demonstrate that the degree of hybridization depends on the concentration ratio of probes and target molecules. Therefore, it has been revealed that the use of a 1[thin space (1/6-em)]:[thin space (1/6-em)]10 ratio between the probe concentrations and target concentrations does not indicate a significant variation between the mismatched probe variants – Fig. 5(a). However, by working with a low concentration of target molecules (0.1 μM), the hybridization signal intensities allow a good discrimination between the PM probes and the ones with different mismatch variants, the differences being also statistically significant Table 2(a).

Table 2 Student's t-test results for glass microarray: (a) 1 μM probe sequences hybridized with 0.1 μM target molecules. (b) 50 μM probe sequences hybridized with 10 μM target molecules
PM vs. C > A PM vs. C > G PM vs. C > T C > A vs. C > G C > A vs. C > T C > G vs. C > T
P Values
0.0012 0.0010 0.0013 0.3357 0.7994 0.5037

PM vs. C > A PM vs. C > G PM vs. C > T C > A vs. C > G C > A vs. C > T C > G vs. C > T
P Values
0.9073 0.0040 0.0483 0.0015 0.0136 0.0003


Meanwhile, by selecting a 1[thin space (1/6-em)]:[thin space (1/6-em)]5 ratio of probe concentration versus target concentration generates variations among the SNPs. The results from the graphical data analysis correlated with the statistical analysis Fig. 5(b) have shown the clear distinction of the different mismatch types. Hence, the hybridization with the C > G and C > T probes displayed signal intensities significantly lower in contrast with the PM probes. Furthermore, the differences between them are statistically assessable. Instead, the signal intensities of PM and C > A probes have close values and the differences between the two categories are not statistically significant Table 2(b).

The calibration tests on commercial functionalised glass supports revealed that the probe concentrations can cause different spatial features which complicate the discrimination between mismatched base pairing on flat substrate. Herein, the use of a 2D microarray platform for SNP discrimination has led to steric limitations when a higher concentration of probes was spotted, these being overcomed by handling smaller probe concentrations. Also, when the effect of probe mismatches in the hybridization reaction was analysed, we noticed that the mismatch types are more sensitive to different experimental concentrations, this observation being in accordance with the observations of Duan et al.41

In order to further investigate if the 3D architecture of the platform can overcome the aforementioned drawbacks, the hybridization sensitivity and specificity was compared between the standard glass support and newly fabricated silicon based support.

3.2 Hybridization tests on SiNWs platform

In Fig. 6, the scanned fluorescent image of the hybridization results achieved on SiNWs is presented, where the immobilized probe concentrations (1 μM, 10 μM, 25 μM and 50 μM) and the probe types (PM, C > A, C > G and C > T) are also indicated. As it can be observed, detectable spots are only at probe concentrations at the threshold level of 50 μM. This is not necessarily an impediment as long as we further achieve a good discrimination between the different types of mismatches.
image file: c5ra14442f-f6.tif
Fig. 6 Hybridization results of arrays hybridized on SiNWs support with 10 μM target molecules.

In Fig. S2 (ESI), for a better overview we have comparatively analysed the hybridization results from Fig. 4(b) and from Fig. 6. It is obvious that for probe concentrations of 50 μM, the spots' hybridization signals on SiNWs appeared more uniform than those on glass support. It has to be mentioned also that we used 20% reduced PMT gain settings for scanning the SiNWs platform instead of the PMT settings of 43% used for commercial glass substrates, since, at the higher levels of the laser voltage power, the scanned images were supersaturated. The smaller PMT gain has the advantages of reducing the signal background and has led to an improved spot detection.

The improved performance of the fluorescence measurements is due to the antireflective properties over the entire range of visible spectrum of the SiNWs layer42,43 and therefore it acts like a trapping layer for the excitation radiation. In this way, the interaction of the excitation beam with the fluorophore is enhanced leading to an increased fluorescence signal. We also calculated that the targets hybridized on SiNWs had an average diameter of 400 μm whereas the complementary sequences hybridized on commercial substrates had an average diameter of 200 μm. The diameter differences of the spots are ascribed to the SiNWs host matrix, which allows the capillary diffusion of both the immobilized probes and the hybridization solutions. We did not focus on optimization studies in terms of controlling the microarray spots, our present interest being to investigate the capacity of the novel SiNWs substrate to be used as platform for mismatch detection. In the future, additional plasma treatments can resolve this issue, by producing spatially selective hydrophilic areas.44

The spots intensities from the SiNWs microarray were also normalized by log2 and the data corresponding to the spots resulted after 50 μM of probes hybridized with 10 μM of target molecules were used for a comparative analysis between both types of supports. While the glass substrate data have not shown reliable differences between all the analysed hybridization series, from the graphical and statistical data analysis of the hybridization results on SiNWs support, statistically significant differences between all the analyzed series are observed (Fig. 7, Table 3). Hence, by using the 3D silicon support in the hybridization experiments of BRCA1 probes without mismatch (PM) and the probe variants with one mismatch (MM), one can clearly differentiate through the types of probes by means of hybridization signal intensities.


image file: c5ra14442f-f7.tif
Fig. 7 Assessment of the signal intensity trends between the C:G base pairs and the impairs represented by A:G, G:G and T:G on SiNWs support with 10 μM target molecules.
Table 3 Student's t-test results for 50 μM probe sequences hybridized with 10 μM target molecules on SiNWs platform
PM vs. C > A PM vs. C > G PM vs. C > T C > A vs. C > G C > A vs. C > T C > G vs. C > T
P Values
0.0048 0.0005 0.0037 0.0032 0.0071 0.0323


The hybridization thermodynamics has been calculated for the sequences with IDT Biophysics software45 and the expected hybridization thermodynamics is shown in Table S2 (ESI). There are differences of: −2.32 (C > G), −2.54 (C > T), −3.05 (C > A) kcal mole−1, between the Gibbs free energies of the PM and MM sequences. Between the different types of mismatches, there are observed changes of 0.73 (C > A vs. C > G), 0.51 (C > A vs. C > T) and 0.22 (C > G vs. C > T) kcal mole−1 between the Gibbs free energies. It is noteworthy that, although the theoretical calculations indicate relatively small differences between PM and MM, and furthermore between every type of MM, the actual differences are noticeable according to both visual and statistical analyses.

Theoretically, the most stable mismatched base pair is A/G, evidences pointing out that the A/G base pair fits into a B-form standard DNA duplex with little distortion, as a consequence of the purine–purine stacking and of the hydrogen bonding interactions.46,47 The G/G mismatched base pair has shown both stabilizing and destabilizing effects on DNA duplex, depending on the adopted conformation.48 On the opposite pole, the T/G mismatched base pair adopts a wobble configuration, which has a surprisingly destabilizing effect due to the unfavourable steric interactions involving the thymine methyl group.49

Also, the results confirm the experiments regarding the stability in solution of different mismatched base pairs revealing that the G/G base pair has a stability close to A/G base pair, whereas, the T/G bond has a lower stability than A/G and G/G duplexes.50

The results obtained on SiNWs were very promising, demonstrating that, in contrast with flat 2D substrates, the experimental 3D microarray platform significantly improves both the hybridization efficiency and the qualitative analyses of the fluorescent hybridization intensities when the probes differ through the presence of different mismatch variants in the same position. This improvement is an effect of the uniform and tridimensional distribution of the silicon nanowires which promoted a better dispersion of the immobilized probe and, consequently, has led to a more uniform hybridization signal.

3.3 Signal to both standard deviations ratio assessment

The signal-to-noise ratio (SNR) has been widely used in the microarray technology to define a positive spot24,51 and the sensitivity of the experiment. The smaller the background noise is, the better the SNR will be. To assess the performance of the newly developed microarray supports, we assayed the dependence of probe concentrations with SNR intensities, and further compared them to the commercial supports. The signal-to-noise ratio calculation, called the signal-to-both-standard-deviations ratio (SSDR), was defined by He and Zhou51 for a target spot as:
 
image file: c5ra14442f-t1.tif(1)
where [S with combining macron] signifies the mean intensity of the spot, [B with combining macron] represents the mean background of the local signal, while σS and σB represent the standard deviations of signal and background, respectively. Spots that have a SSDR equal or greater than 0.7 can be considered valid.

In order to assess which type of support has a lower background signal, the overall SNR of the commercial support was compared with the SNR calculations from the results generated by the hybridizations on the SiNWs platform. Furthermore, to test if the obtained SSDR results are reproducible, we have repeated the test on an additional set of commercial glass support versus the SiNWs one. The SSDR was calculated by considering the raw hybridization intensities of 16 spots corresponding to 50 μM of probes hybridized with 10 μM of target molecules. A SSDR comparison between the replicates of 2D commercial and 3D SiNWs supports is presented in Fig. 8(a).


image file: c5ra14442f-f8.tif
Fig. 8 Assessment of signal to both standard deviation ratio for (a) two chip duplicates for each type of platform and (b) each type of platform, statistically interpreted.

The two-tailed Student's t-tests have demonstrated that for the calculated SSDR corresponding to the SiNWs support replicates the differences are not statistically significant (p = 0.5341), and neither for the glass support replicates (p = 0.6541). The aforementioned statistical results confirm that the experimental conditions have conducted to robust hybridizations and regardless of the investigated substrates, the SSDR results were reproducible.

By comparing the SSDR data obtained on SiNWs versus commercial glass platform, the differences are statistically significant (p = 0.0012) and furthermore, for SiNWs platform, the SSDR value was approximately 1.6 fold higher than on commercial glass – Fig. 8(b). These results demonstrate once more that the nanostructured silicon substrate used in fabricating microarray platforms is superior to the commercial support. The increased signal-to-noise ratio might be assigned to: (i) the increased active area on which more probes can be attached and a higher density on spot level is achieved; (ii) the capacity of excluding the unwanted effects of a planar surface, caused by the limited diffusion of the probes in a spot;24 (iii) the anti-reflectivity property of the 3D support52 which leads to a low background signal and improves the hybridization signal reading.

4. Conclusions

In summary, we have successfully developed a silicon-based substrate as a microarray platform for the investigation of SNPs with pathological significance. In order to increase the active area of the support, the silicon substrate was subjected to a one step MACE process and a uniform array of silicon nanowires was obtained.

The sensitivity and the SNP discrimination performance of the new experimental supports were analysed in comparison with the commercial glass substrates using synthetic oligonucleotides. In contrast to the standard glass support, the developed platform allowed the discrimination of the hybridization signal intensities when the probes differ by the presence of one mismatch (C > A, C > G and C > T) in the same position, the results being confirmed by statistical analyses. Reliable differences were actually achieved between the fluorescent signal intensities and confirmed by statistical analyses even if the theoretical calculations indicate relatively small differences between the Gibbs free energies of the hybridized oligonucleotide pairs. These achievements are mainly attributable to the fact that the target molecules' access at the attached probes is facilitated when a tridimensional architecture of the support is used. The SSDR values were further calculated for the both tested platforms and the results have indicated a 1.6 fold higher SSDR for the SiNWs support, mainly due to the layer of SiNWs with antireflective properties, the differences being also statistically significant. Based on these results, it is reasonable to assume that the functionalised SiNWs substrate represents an adequate microarray platform to discriminate the hybridization results depending on the identity of mispairs.

Acknowledgements

The authors acknowledge the support of the Romanian Ministry of Education and Research through the contracts HRCarrays no. 4-1/2012 and MultiplexGen no. 36/2014. The authors are grateful to Dr Sabin Cinca for stimulating discussions regarding the hybridization experiments within the workshops, and to Marioara Paznicu for technical support.

References

  1. W. D. Foulkes, Clin. Genet., 2014, 85(1), 1 CrossRef CAS PubMed.
  2. D. Fanale, V. Amodeo, L. R. Corsini, S. Rizzo, V. Bazan and A. Russo, NPG Asia Mater., 2012, 31(17), 212 Search PubMed.
  3. I. Hernan, E. Borràs, M. de Sousa Dias, M. J. Gamundi, B. Mañé, G. Llort, J. A. G. Agúndez, M. Blanca and M. Carballo, J. Mol. Diagn., 2012, 14(3), 286 CrossRef CAS PubMed.
  4. L. Zhang, T. Kirchhoff, C. J. Yee and K. Offit, J. Mol. Diagn., 2009, 11(3), 176 CrossRef CAS PubMed.
  5. A. Juwle and D. Saranath, Med. Oncol., 2012, 29(5), 3272 CrossRef CAS PubMed.
  6. A. Riahi, M. Kharrat, I. Lariani and H. Chaabouni-Bouhamed, Fam. Cancer, 2014, 13, 603 CrossRef CAS PubMed.
  7. D. Berzina, M. Nakazawa-Miklasevica, J. Zestkova, K. Aksenoka, A. Irmejs, A. Gardovskis, D. Kalniete, J. Gardovskis and E. Miklasevics, BMC Med. Genet., 2013, 14, 61 CrossRef CAS PubMed.
  8. H. B. Rasmussen, in Restriction Fragment Length Polymorphism Analysis of PCR-Amplified Fragments (PCR-RFLP) and Gel Electrophoresis - Valuable Tool for Genotyping and Genetic Fingerprinting, Gel Electrophoresis - Principles and Basics, ed. S. Magdeldin, InTech, Croatia, 2012, ch. 18, pp. 315–334 Search PubMed.
  9. W. Wang, Z. Wei, T. W. Lam and J. Wang, Sci. Rep., 2011, 1(55), 1 Search PubMed.
  10. C. T. Wittwer, Hum. Mutat., 2009, 30(6), 857 CrossRef CAS PubMed.
  11. T. LaFramboise, Nucleic Acids Res., 2009, 37(13), 4181 CrossRef CAS PubMed.
  12. X. Mao, B. D. Young and Y. J. Lu, Curr. Genomics, 2007, 8, 219 CrossRef CAS.
  13. B. Sobrino, M. Brion and A. Carracedo, Forensic Sci. Int., 2005, 154, 181 CrossRef CAS PubMed.
  14. B. G. Moreira, Y. You and R. Owczarzy, Biophys. Chem., 2015, 198, 36 CrossRef CAS PubMed.
  15. J. Worley, K. Bechtol, S. Penn, D. Roach, D. Hanzel, M. Trounstine and D. Barker, in A Systems Approach to Fabricating and Analyzing DNA Microarrays, ed M. Schena, Eaton Publishing, USA, 2000, ch. 4, pp. 65–85 Search PubMed.
  16. G. Ventimiglia and S. Petralia, J. Bionanosci., 2013, 3(4), 428 CrossRef.
  17. R. M. Elder and A. Jayaraman, J. Chem. Phys., 2014, 140(15), 155103 CrossRef PubMed.
  18. A. Jayaraman, C. K. Hall and J. Genzer, J. Chem. Phys., 2007, 127(14), 1 CrossRef PubMed.
  19. S. Monti, I. Cacelli, A. Ferretti, G. Prampolini and V. Barone, Mol. BioSyst., 2010, 6(11), 2230 RSC.
  20. S. Galbiati, F. Damin, P. Pinzani, I. Mancini, S. Vinci, M. Chiari, C. Orlando, L. Cremonezi and M. Ferrari, PLoS One, 2013, 8(3), 1 Search PubMed.
  21. M. Simion, L. L. Ruţă, M. Matache, I. Klepş, M. Miu, C. C. Paraschivescu, A. Brăgaru and T. Ignat, Mater. Sci. Eng., B, 2010, 169(1–3), 67 CrossRef CAS PubMed.
  22. M. Simion, L. Savu, A. Rădoi, M. Miu and A. Brăgaru, J. Nanosci. Nanotechnol., 2011, 11(10), 9102 CrossRef CAS PubMed.
  23. B. R. Murthy, J. K. K. Ng, E. S. Selamat, N. Balasubramanian and W. T. Liu, Biosens. Bioelectron., 2008, 24(4), 723 CrossRef CAS PubMed.
  24. M. K. Dawood, L. Zhou, H. Zheng, H. Cheng, G. Wan, R. Rajagopalan, H. P. Too and W. K. Choi, Lab Chip, 2012, 12(23), 5016 RSC.
  25. S. Li, W. Ma, Y. Zhou, X. Chen, Y. Xiao, M. Ma, F. Wei and X. Yang, J. Solid State Chem., 2014, 213, 242 CrossRef CAS PubMed.
  26. Y. Qi, Z. Wang, M. Zhang, X. Wang, A. Ji and F. Yang, AIP Adv., 2014, 4(3), 031307 CrossRef PubMed.
  27. H. Han, Z. Huang and W. Lee, Nano Today, 2014, 9, 271 Search PubMed.
  28. Z. Huang, N. Geyer, P. Werner, J. de Boor and U. Gösele, Adv. Mater., 2011, 23, 285 Search PubMed.
  29. W. K. To, C. H. Tsang, H. H. Li and Z. Huang, Nano Lett., 2011, 11(12), 5252 CrossRef CAS PubMed.
  30. U. Mueckstein, G. G. Leparc, A. Posekany, I. Hofacker and D. P. Kreil, BMC Bioinf., 2010, 11, 35 CrossRef PubMed.
  31. R. C. Welling and T. A. Knotts, J. Chem. Phys., 2015, 142(1), 015102 CrossRef PubMed.
  32. S. Nimse, K. Song, M. Sonawane, D. Sayyed and T. Kim, Sensors, 2014, 14(12), 22208 CrossRef PubMed.
  33. M. Simion, M. Kusko, I. Mihalache and A. Brăgaru, Mater. Sci. Eng., B, 2013, 178, 1268 CrossRef CAS PubMed.
  34. S. Libertino, V. Aiello, A. Scandurra, M. Renis, F. Sinatra and S. Lombardo, Sensors, 2009, 9(5), 3469 CrossRef CAS PubMed.
  35. SNPedia, http://snpedia.com/index.php/Rs28897696, accessed April 2015 Search PubMed.
  36. IDT SciToolsOligoAnalyzer 3.1., http://eu.idtdna.com/calc/analyzer, accessed April 2015 Search PubMed.
  37. A. Kitts and S. Sherry, in The Single Nucleotide Polymorphism Database (dbSNP) of Nucleotide Sequence Variation, ed. J. McEntyre and J. Ostell, The NCBI Handbook, Bethesda, MD, 2002, ch. 5 Search PubMed.
  38. dbSNP, http://www.ncbi.nlm.nih.gov/snp?db=snp, accessed April 2015 Search PubMed.
  39. D. Radushev, GraphPad Prism (5.03), GraphPad Software Inc., San Diego, California, 2009 Search PubMed.
  40. L. Zhang, M. F. Miles and K. D. Aldape, Nat. Biotechnol., 2003, 21(7), 818 CrossRef CAS PubMed.
  41. F. Duan, M. A. Pauley, E. R. Spindel, L. Zhang and R. B. Norgren, BioData Min., 2010, 3(1), 1 CrossRef PubMed.
  42. M. Khorasaninejad, S. Patchett, J. Sun, N. O and S. S. Saini, J. Opt. Soc. Am. B, 2012, 29(11), 3063 CrossRef CAS.
  43. Z. Guo, J.-Y. Jung, K. Zhou, Y. Xiao, S.-W. Jee, S. A. Moiz and J.-H. Lee, Proc. SPIE Conf., 2010 DOI:10.1117/12.860397.
  44. S. Yokoyama, Y. Kamei, T. S. Matsui and S. Deguchi, J. Bioanal. Biomed., 2015, 7(3), 081 Search PubMed.
  45. IDT Biophysics 1.02, http://biophysics.idtdna.com/, accessed April 2015 Search PubMed.
  46. Y. Li, G. Zon and W. D. Wilson, Proc. Natl. Acad. Sci. U. S. A., 1991, 88, 26 CrossRef CAS.
  47. H. T. Allawi and J. SantaLucia, Biochemistry, 1998, 37, 2170 CrossRef CAS PubMed.
  48. M. Sharma, A. V. Predeus, S. Mukherjee and M. Feig, J. Phys. Chem. B, 2013, 117, 6194 CrossRef CAS PubMed.
  49. H. T. Allawi and J. SantaLucia, Biochemistry, 1997, 36, 10581 CrossRef CAS PubMed.
  50. S. Pan, X. Sun and J. K. Lee, Int. J. Mass Spectrom., 2006, 253, 238 CrossRef CAS PubMed.
  51. Z. He and J. Zhou, Appl. Environ. Microbiol., 2008, 74(10), 2957 CrossRef CAS PubMed.
  52. Y. L. Deng and Y. J. Juang, Biosens. Bioelectron., 2014, 53, 37 CrossRef CAS PubMed.

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra14442f

This journal is © The Royal Society of Chemistry 2015
Click here to see how this site uses Cookies. View our privacy policy here.