Biosensors and Bioelectronics 56 (2014) 278–285

Contents lists available at ScienceDirect

Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios

Microdevices for detecting locus-specific DNA methylation at CpG resolution Kevin M. Koo a,b,1, Eugene J.H. Wee a,1, Sakandar Rauf a, Muhammad J.A. Shiddiky a, Matt Trau a,b,n a Centre for Biomarker Research and Development, Australian Institute for Bioengineering and Nanotechnology (AIBN), The University of Queensland, Brisbane, QLD 4072, Australia b School of Chemistry and Molecular Biosciences, The University of Queensland, Brisbane, QLD 4072, Australia

art ic l e i nf o

a b s t r a c t

Article history: Received 12 November 2013 Received in revised form 13 January 2014 Accepted 17 January 2014 Available online 25 January 2014

Simple, rapid, and inexpensive strategies for detecting DNA methylation could facilitate routine patient diagnostics. Herein, we describe a microdevice based electrochemical assay for the detection of locusspecific DNA methylation at single CpG dinucleotide resolution after bisulfite conversion of a target DNA sequence. This is achieved by using the ligase chain reaction (LCR) to recognize and amplify a C to T base change at a CpG site and measuring the change electrochemically (eLCR). Unlike other electrochemical detection methods for DNA methylation, methylation specific (MS)-eLCR can potentially interrogate any CpG of interest in the genome. MS-eLCR also distinguishes itself from other electrochemical LCR detection schemes by integrating a peroxidase-mimicking DNAzyme sequence into the LCR amplification probes design which in turn, serves as a redox moiety when bound with a hemin cofactor. Following hybridization to surface-bound capture probes, the DNAzyme-linked LCR products induce electrocatalytic responses that are proportional to the methylation levels of the gene locus in the presence of hydrogen peroxide. The performance of the assay was evaluated by simultaneously detecting C to T changes at a locus associated with cancer metastasis in breast cancer cell lines and serum-derived samples. MS-eLCR required as little as 0.04 pM of starting material and was sensitive to 10–15% methylation change with good reproducibility (RSD ¼7.9%, n ¼3). Most importantly, the accuracy of the method in quantifying locus-specific methylation was comparable to both fluorescence-based and Next Generation Sequencing approaches. We thus believe that the proposed assay could potentially be a low cost alternative to current technologies for DNA methylation detection of specific CpG sites. & 2014 Elsevier B.V. All rights reserved.

Keywords: Locus-specific DNA methylation CpG resolution DNAzyme Ligase chain reaction Electrocatalytic reaction Breast cancer cell line

1. Introduction DNA methylation is an epigenetic modification process that generally involves the addition of a methyl group (–CH3) to the fifth carbon position of the pyrimidine ring of cytosine at cytosine– guanine (CpG) dinucleotides (Robertson, 2005). Aberrant DNA methylation at specific CpG sites has been reported to cause genome instability and thus promoting cancer (Feinberg and Tycko, 2004). Therefore, the analysis of locus-specific methylation is of aid in early cancer diagnosis and predisposition. Current locus-specific methylation assays at single base resolution are generally PCR-based and involve bisulfite modification of DNA to convert methylation events to base changes (C to T base changes). This is usually followed by either mass spectrometric (Ehrich et al.,

n

Corresponding author. Tel.: þ 61 7 33464178; fax: þ 61 7 33463973. E-mail address: [email protected] (M. Trau). 1 Authors contributed equally.

0956-5663/$ - see front matter & 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bios.2014.01.029

2005) or fluorescence (Wang et al., 2009; Palanisamy et al., 2011) readouts. However, such readout methods are too expensive, tedious, and time-consuming for routine clinical diagnostics. The development of detection methods that are faster, more sensitive, and less expensive than currently available counterparts in mass spectrometry and fluorescence readouts will fill important needs in clinical applications. Electrochemical approaches offer simple, economical and sensitive alternatives to conventional readout methods for biosensing applications (Zhu et al., 2004; Wang et al., 2012). Existing electrochemical methylation assays either detect the overall DNA methylation at a single locus (Kato et al., 2008; Wang et al., 2010; Kato et al., 2011; Wang et al., 2012; Wang et al., 2013) or depend on the limited availability of restriction enzyme sites at the locus of interest to infer CpG methylation (Dai et al., 2012; Li et al., 2012; Dai et al., 2013). None however, are able to provide methylation information for both locus specificity and CpG resolution at virtually anywhere in the genome. The eLCR (Wee et al., 2012a) assay, a novel electrochemical technique based on the ligase chain

K.M. Koo et al. / Biosensors and Bioelectronics 56 (2014) 278–285

reaction (LCR) to rapidly detect single DNA base changes at specific sequences on microfabricated devices (Wee et al., 2013), may be able to provide such methylation information. However, the use of horse radish peroxidase (HRP) labeling for enzymatic signal amplification in eLCR is limited by higher running costs and susceptibility to enzyme degradation. A simple solution to these limitations could be to replace HRP with a peroxidase-mimicking DNAzyme (as a redox moiety) for electrochemical quantification of LCR products. DNAzymes have gained recent interest as electrocatalytic labels for signal amplification in electrochemical biosensors (Niazov et al., 2004; Guo et al., 2010; Pelossof et al., 2010). In particular, the hemin/G-quadruplex-based DNAzyme is a widely used HRPmimicking compound in electrochemical biosensing (Travascio et al., 1998; Willner et al., 2008b; Kosman and Juskowiak, 2011). This is because DNAzymes can potentially lower reagent cost and provide higher chemical stability. Moreover, the synthesis of hemin/G-quadruplex-based DNAzyme is much easier and cheaper compared to protein enzymes. After adopting a spontaneous G-quadruplex structure consisting of layers of G tetrads, a high affinity binding site is formed for the redox active hemin molecule, thus resulting in a functional hemin/G-quadruplex peroxidasemimicking DNAzyme (Willner et al., 2008b; Cheng et al., 2009; Kong et al., 2010; Kosman and Juskowiak, 2011). While various DNAzyme-based strategies have been used to detect single base changes (Willner et al., 2008a; Wang et al., 2011; Tang et al., 2013; Zhou et al., 2013) via chemiluminescence and colorimetric means, its application in quantifying DNA methylation electrochemically has yet to be demonstrated. Microfabricated devices have the potential advantages of integration, portability, speed, cost, and minimal solvent/reagent requirements (Shiddiky and Shim, 2007; Rauf et al., 2013). Additionally, devices with micro- or nano-scaled footprint working areas have the high signal-to-noise ratio (Soleymani et al., 2009). In this study, we present a microdevice-based DNA methylationspecific (MS) electrochemical assay using DNAzymes (i.e. MSeLCR). Following DNA bisulfite PCR conversion, site-specific methylation events (C or T base) in target sequences were amplified by a duplexed LCR to generate two different species of LCR products representing methylated or unmethylated events respectively. Finally, the relative proportion of two LCR product species was then detected via DNAzyme-mediated electrocatalytic reduction of hydrogen peroxide (H2O2) at electrodes surfaces to estimate methylation levels. With this approach, MS-eLCR was able to accurately measure the methylation status of a specific CpG site in a breast cancer associated gene in three breast cancer cell lines. We also extended the application to a serum-derived DNA sample and finally, the analytical performance of the device was validated with fluorescence- and Next Generation Sequencing

279

(NGS)- based findings. MS-eLCR is, to our knowledge, the only electrochemical method allowing DNA methylation interrogation potentially anywhere in the genome while achieving locus-specific single CpG resolution.

2. Experimental 2.1. Reagents All reagents were of analytical grade and purchased from Sigma Aldrich (Australia). UltraPure™ DNase/RNase-free distilled water (Invitrogen, Australia) was used throughout the experiments. DNA oligonucleotides were purchased from Integrated DNA Technologies (USA) and sequences are shown in Table 1. MDA-MB-231, MDA-MB-468 and HCC1937 breast cancer cells lines were purchased from ATCC (USA).

2.2. Fabrication of microdevices Microdevices were fabricated as described previously (Wee et al., 2013). It contains five independent sample wells connected by a microchannel and each well contains four independentlywired electrodes (diameter¼ 20 mm). A PDMS cover containing five sample reservoirs connected to an inlet and outlet by the microchannel was bonded permanently onto the device to complete the fabrication process. This design could either facilitate five simultaneous sample measurements for a given biomarker by addressing each well individually or simultaneously measure five biomarkers for a given sample.

2.3. Bisulfite PCR and sequencing Genomic DNA was extracted from cell lines using the DNeasy Blood and Tissue DNA extraction kit (Qiagen, Australia). Bisulfite conversion was done using the MethylEasy™Exceed kit (Human Genetic Signatures, Australia). Finally the miR200b P2 promoter sequence was PCR amplified from all three cell lines and serum DNA samples using GoTaqs PCR system (Promega, Australia) using protocols previously described.(Wee et al., 2012b; Wee et al., 2013) Primer sequences (PCR Fwd and PCR Rev) are given in Table 1. The PCR products derived from MDA-MB-231, MDA-MB-468 and HCC1937 were then outsourced to the Australian Genome Research Facility (AGRF, Brisbane Node) for NGS on the 454platform.

Table 1 Sequences of DNA oligonucleotides used in 5' to 3' orientation. DNA base being interrogated is highlighted in underlined font. For fluorescent experiments, the TYE 653 and TYE 665 dyes replaced the DNAzymes sequences in P3 and P6 respectively. Oligos

5'-Sequence-3'

Capture 1 Capture 2 P1 P2 EAD2 DNAzyme-P3 P4 P5 EAD2 DNAzyme-P6 Methylated target Unmethylated target PCR Fwd PCR Rev

GAACATGACGATCTGTAACTGG-C3-SH GTACATTGGTGCAGACAGAATT-C3-SH CCAGTTACAGATCGTCATGTTCAACGCCCCTCGACGCACCTAACCG Phos-CCCTACCCGCCTACCTAACCGA AGGGAGGGAGGGAGGGTC-C3-TCGGTTAGGTAGGCGGGTAGGGC Phos-GGTTAGGTGCGTCGAGGGGCGTT AATTCTGTCTGCACCAATGTACAACGCCCCTCGACGCACCTAACCA AGGGAGGGAGGGAGGGTC-C3-TCGGTTAGGTAGGCGGGTAGGGT GGGATTCGGTTAGGTAGGCGGGTAGGGCGGTTAGGTGCGTCGAGGGGCGTTTTTA GGGATTCGGTTAGGTAGGCGGGTAGGGTGGTTAGGTGCGTCGAGGGGCGTTTTTA GTCGGGCGTTTTTATTTTATTTTAGTT AAAACCGCCCCAACAAAAAA

280

K.M. Koo et al. / Biosensors and Bioelectronics 56 (2014) 278–285

2.4. LCR protocol

2.6. Electrochemical measurements

A 25 mL reaction mix (Ampligases DNA Ligase Kit, Epicentre) consisting of 2.5 mL 10x reaction buffer, 0.1 mM of each DNA probes, 2.5 units Ampligase DNA Ligase, 50 ng salmon sperm DNA as blocking agent and 1 μL of initial DNA target was used for each LCR reaction. The thermocyling used for LCR reactions was 1 cycle of 95 1C for 3 min, 19 cycles of 95 1C for 30 s followed by 63 1C for 2 min. For sensitivity studies, designated concentrations of synthetic DNA target were prepared for each reaction. To generate the methylation calibration plot, synthetic targets of various known methylation levels were amplified by LCR using P1 to P6 probes (0.1 mM each, Table 1) in duplexed reactions to simultaneously amplify both methylated and unmethylated DNA targets. C and T base specificity was achieved with P1/P3 probes which recognized methylated DNA (left panel in Scheme 1) while P5/P6 recognized unmethylated DNA (right panel in Scheme 1). For breast cancer cell line experiments, methylation events (CpG12 of the miR200b P2 promoter) were amplified by LCR using a 1 in 100 dilution of bisulfite PCR samples generated as described above. Following thermal cycling, 5 mL of LCR products were electrophoresed through 2% agarose and stained with ethidium bromide before being visualized on an UV transilluminator to verify amplification. After verification, an equal volume (20 mL) of DNAzyme buffer (50 mM Tris–HCl, 20 mM KCl, 150 mM NaCl, 1% [v/v] DMSO, 0.05% [w/v] Triton X-100, pH 7.4) was added to each LCR product tube, and heated for 3 min at 95 1C. The tubes were then allowed to cool back down to room temperature for DNAzyme sequences to spontaneously fold into correct G-quadruplex structures. For the fluorescence-based detection, the same P1-P6 probes were used but with P3 and P6 labeled with TYE653 and TYE665 dyes respectively to distinguish between the methylation states. After amplification with the same LCR protocol as the DNAzyme approach, LCR products were visualized on the Typhoon 9100 scanner and processed using the ImageJ densitometry plugin. Methylation was quantified using a similar approach as the electrochemical method (see Supplementary for details).

All electrochemical measurements were done on a CH1040C potentiostat (CH Instruments) with a three-electrode system consisting of DNA-modified gold microelectrode as working electrode, Pt counter electrode, and Ag quasi reference electrode (QRE). Electrocatalytic reduction of H2O2 was performed using cyclic voltammetry from 100 mV to  700 mV at a scan rate of 50 mV/s. For sensitivity studies, the increase in electrocatalytic response corresponding to the long “knife” motifs (methylated or unmethylated) was calculated as follows: % increase in catalytic response ¼ Δjcat ¼ ½ð jf inal –jNoT Þ=jNoT 100% ð1Þ where jNoT is the mean current density at 0 M target (no target control) concentration, jfinal is the mean current density at any concentration of methylated targets or unmethylated targets. jfinal and jNoT were calculated by averaging the current density of the four independent electrodes present in the same well of the device (current density ¼ electrocatalytic response/effective area of the electrode). For methylation detection, % methylation in a duplexed assay was defined as: %Methylation ¼ ½M=ðM þ UMÞ  100% ¼ jratio ¼ ½ jcapture 1 =ð jcapture 1 þ jcapture 2 Þ  100%

ð2Þ

where M and UM are methylated and unmethyated events respectively, jcapture 1 and jcapture 2 are the mean current densities in Capture 1 probe-modified well (methylated event) and Capture 2 probe-modified well (unmethylated event) respectively. The mean current density was calculated by averaging the current density from four independent electrodes present in the same well of the device.

3. Results and discussion 3.1. MS-eLCR principle

2.5. Functionalization of device The device was cleaned by rinsing with distilled water and dried under a flow of nitrogen. A 10 mL solution containing 2 mM thiolated DNA probes in 50 mM PBS (pH 7.4) was dropped onto the electrodes in each well and sealed overnight at room temperature for self-assembling DNA monolayer (SAM) formation. Subsequently, 1 mM 6-mercapto-1-hexanol in 50 mM PBS containing 2.5% glycerol solution was added to each well for 30 min at room temperature to block bare surfaces. After washing with 50 mM PBS, 20 mL of DNAzyme-linked LCR products were added into each well, sealed and allowed to hybridize at 37 1C for 2 h. For methylation quantification and cell line experiments, designated sample was loaded into the inlet of the device and directed to sampling wells by suction. To discriminate between methylation events, sample wells were either modified with Capture 1 or Capture 2 probes which specifically recognized P1 (methylated events, left panel in Scheme 1) or P5 (unmethylated events, right panel in Scheme 1) respectively. Following hybridization at 37 1C for 2 h, the wells were gently washed with DNAzyme buffer to remove unbound products. Finally, 10 mM hemin in DNAzyme buffer was added to each well and incubated for 30 min at room temperature to form hemin/G-quadruplex DNAzymes. Prior to electrochemical measurements, the hemin solution was replaced with 0.1% H2O2 in DNAzyme buffer.

Scheme 1 shows the principle for our duplexed electrochemical assay to quantitatively determine the DNA methylation status at a specific locus. Firstly, bisulfite-PCR products are LCR-amplified with six probes that simultaneously interrogate the methylation status at the site of interest (C: methylated; T: unmethylated). In the first LCR cycle, probes complementary to either the C (P1) or T (P5) bases ligate with probe P2. These probe pairings then served as targets for probes P4 and either P3 or P6 in the next amplification cycle. Through further thermal cycling, LCR exponentially amplifies the DNA base changes, generating two different ligated long “knife” motif products to represent C or T methylation status. In contrast, excess unligated probes will lead to short “knife” motif products. P1 and P5 probes have different barcode sequences which are recognized by pre-immobilized capture probes 1 or 2 on two independent gold electrodes respectively. Electrochemical currents generated by the electrocatalytic reduction of H2O2 by DNAzymemodified sensors provide a quantitative measure of the amount of long “knife” motifs, which in turn, is proportional to the amount of C or T events in the initial bisulfite-treated DNA sample. Finally, the level of % methylation at the site of interest is estimated by the electrochemical signal ratios generated from Capture 1 probesmodified and Capture 2 probes-modified wells. The use of LCR also has two advantages. Firstly, LCR potentially allows interrogation of virtually anywhere in the genome by simply designing locusspecific probes. Secondly, the ligating junction of the probes

K.M. Koo et al. / Biosensors and Bioelectronics 56 (2014) 278–285

281

Scheme 1. Schematic representation of MS-eLCR. (A) Microdevice platform consisting of five independent sample wells with four independently wired electrodes in each well. (B) Duplexed LCR with six probes (P1 to P6) to amplify methylation events (C: methylated; T: unmethylated) in bisulfite-PCR samples. Successful ligation generates long “knife” motifs and excess unligated probes result in short “knife” motifs. (C) Hybridization of “knife” motifs to pre-immobilized capture probes specific for methylation events (Capture Probe 1: C; Capture Probe 2: T) on independent gold electrodes. In the presence of H2O2, DNAzymes on long “knife” motifs generate electrocatalytic response at the electrode surface. The catalytic response is proportional to the amount of the long “knife” motifs present in the samples.

overlaps a CpG of interest, thus enabling CpG-level methylation detection. Both these characteristics have not been described in any electrochemical method for DNA methylation. 3.2. Detection sensitivity and specificity The sensitivity of the DNAzyme-based eLCR assay was first evaluated with different amounts of LCR-amplified products. Fig. 1

shows the cyclic voltammetric responses obtained with initial target concentrations from 0 pM to 40 pM. The magnitude of the catalytic currents represented the amount of long “knife” motifs, which in turn, was proportional to the amount of DNAzymes available to electrocatalyze the reduction of H2O2. These data were also verified by gel electrophoresis (Supporting Information, Fig. S1) which showed a positive association between the long “knife” product band and initial target concentration. At 0.04 pM,

282

K.M. Koo et al. / Biosensors and Bioelectronics 56 (2014) 278–285

calibration plot was first generated by using synthetic standards of various known methylated levels (0%, 5%, 10%, 25%, 50%, 75% and 100%). Moreover, to demonstrate the multiplexing capability of our microdevices, each target was amplified in a duplexed LCR reaction i.e. simultaneous amplification of both methylated and unmethylated events. Methylation events were distinguished by the different barcode sequences on probes P1 (methylated) and P5 (unmethylated) and recognized by pre-immobilized Capture 1 or Capture 2 probes respectively (Table 1 and Scheme 1) in microdevice wells. As such, we would expect targets with higher % methylation to display higher current responses in Capture 1 probes-modified wells and vice-versa for Capture 2 probesmodified wells. As expected, we observed the predicted trends in both Capture 1 and 2 probes-modified wells (Supporting Information, Fig. S2A,S2B). By taking the ratio of current responses, jratio (i.e. jratio is the proportion of jcapture 1 to the total current density obtained from both Capture 1 and Capture 2 modified wells, as defined in Eq. 2), between the two wells, a calibration plot showing response change at different methylation levels was established. Assay reproducibility was determined with a RSD of 7.9% (n ¼3) and the calibration plot was derived by taking the mean jratio at each methylation level (Fig. 2). As shown in Fig. 2, the assay was quantitative of DNA methylation as jratio showed a positive association with % methylation. By representing the data points with a line-of-best-fit, an assumed linear regression relationship (R2 ¼ 0.984) between electrochemical response and % methylation was established: % Methylation ¼ ð jratio –2:506Þ=0:926

Fig. 1. (A) Cyclic voltammograms corresponding to different concentrations of initial target. (B) % response changes within the range of 0–40 pM LCR initial target concentration. Each data point represents the average of the three separate trials (n¼3) and error bars represent standard error within each experiment.

the associated response change was 90% higher than that of the notarget (NoT) control, indicating that the assay was sensitive to 0.04 pM of DNA even in a background of 50 ng unrelated salmon sperm DNA, consistent with our previous attempts using the HRP enzyme (Wee et al., 2012a, Wee et al. 2013). This detection limit of the assay was comparable to fluorescent-based detection methods (0.05 pM) (Benjamin et al., 2003) but requiring lesser amounts of starting material as compared to several existing electrochemistrybased DNA methylation techniques (Tanaka et al., 2007; Kato et al., 2008; Wang et al., 2010). The low current response from the NoT control indicated the specificity of the capture probes on the electrode surface (Fig. 1A). In addition, the absence of a long “knife” product band in the NoT control after gel electrophoresis (Supporting Information, Fig. S1) also supported the specificity of LCR. The RSD over three independent assays is 7.1% (Fig. 1B), demonstrating good assay reproducibility. Taken together, the good sensitivity and specificity of this assay has potential for detecting single DNA base changes in complex biological samples.

ð3Þ

This relationship could then be used to estimate the methylation level of an unknown sample such as that of cell line or serum DNA samples shown in this work. Furthermore, the hybridization specificity of the assay was also demonstrated by the low cross-reactivity between noncomplementary barcodes and capture sequences (Supporting Information, Fig. S2). For instance, in Capture 1 probes-modified wells specific for methylated DNA, non-complementary unmethylated DNA did not hybridize and no significant electrochemical response was detected (Supporting Information, Fig. S2A). To determine the lower detection limit for % methylation changes, samples corresponding to 5% and 10% methylation were compared to 0% methylation. Since only the response from the 10% methylated sample was an appreciable 2-fold higher (Fig. 2) with the RSD of 7.9% (n ¼ 3), we estimated that MS-eLCR was sensitive to between 10–15% methylation changes. While more established commercial methodologies like the GoldenGates Methylation Assay and Sequenoms EpiTYPER Assay have demonstrated 2.5% (Bibikova and Fan, 2009) and 5% (Ehrich et al., 2005) methylation

3.3. Quantitative detection of DNA methylation by MS-eLCR Aberrant DNA methylation is common in cancers and measuring these changes may be clinically useful (Ehrich et al., 2005; Bibikova and Fan, 2009; Vrba et al., 2011; Wang et al., 2012; Wee et al., 2012b). Changes in DNA methylation can be detected by a C to T base change after bisulfite treatment of DNA (Frommer et al., 1992; Clark et al., 2006). With the aim of using MS-eLCR for detecting methylation levels in biological samples, a methylation

Fig. 2. Calibration plot for MS-eLCR responses at 0%, 5%, 10%, 25%, 50%, 75% and 100% methylation. jratio is calculated using eqn. 2. Each data point represents the average of the three separate trials (n ¼3) and error bars represent standard error within each experiment.

K.M. Koo et al. / Biosensors and Bioelectronics 56 (2014) 278–285

283

Fig. 3. % methylation in breast cancer cell line MDA-MB-231, MDA-MB-468, HCC1937 and a serum-derived DNA samples as determined after a duplexed assay and use of calibration plot (Fig. 2). Inset shows the corresponding cyclic voltammograms for the three cell lines and serum-derived DNA samples using M (Capture 1) and UM (Capture 2) probes. Each bar graph represents the average of the three separate trials (n¼ 3) and error bars represent standard error within each experiment.

detection limits respectively, we believe that our MS-eLCR is performing relatively well as a proof-of-concept approach. We next validated our electrochemical assay using an analogous fluorescence-based MS-LCR. Using a similar approach, we also found a positive relationship between fluorescence intensities and methylation levels (Supplementary Information, Fig. S3), thus confirming our electrochemical results. The fluorescence-based assay was sensitive to 5% methylation changes with a high reproducibility of 4.6% RSD (n ¼ 3) in contrast to the DNAzyme approach. The fluorescence data also suggested that LCR conditions were optimal since the observed sensitivity approached the theoretical 2% limit for a duplexed ligase reaction (inherent 1% error per LCR reaction (Wiedmann et al., 1994). The slightly poorer performance of the DNAzyme method may have resulted from variability in immobilizing LCR products on the electrode surface unlike the fluorescence approach which was a direct measurement of the LCR products, hence was less prone to handling errors. Fluorescence approaches, however, incurred higher assay costs as compared to electrochemical means due to the need for costly fluorescent probes and specialized readout equipment. Hence, in terms of assay cost, speed and simplicity, our electrochemistrybased MS-eLCR assay is still a viable option despite the lower methylation detection sensitivity. Moreover, since it is not known whether low level differences (less than 10%) in methylation are of any biological or clinical significance, MS-eLCR's current methylation detection limit could still be sufficient for general locusspecific methylation analysis. We believe future optimization to the protocol and device design could help to improve performance to that comparable of current established technologies. 3.4. Performance of MS-eLCR on complex biological samples The aberrant methylation of miR200b P2 promoter is associated with cancer metastasis and is linked to poor breast cancer prognosis (Korpal et al., 2008; Nicoloso et al., 2009; Wee et al.,

2012b). We hence chose the methylation state at CpG12 of the P2 promoter as a model system to demonstrate the performance of MS-eLCR assay on biologically complex samples. To this end, MSeLCR was first used to determine the methylation statuses in a panel of breast cancer cell lines representing a range of methylation levels (Fig. 3). CpG12 methylation was found to be 90%, 5% and 45% for cell lines MDA-MB-231, MDA-MB-468 and HCC1937 respectively. We also tested MS-eLCR on a clinically normal serum DNA sample which was found to be moderately methylated at 60%. The RSD of 6.7% over three independent runs suggested good assay reproducibility on complex samples. As an electrochemical technique, MS-eLCR has many advantages such as high sensitivity, reproducibility and low cost. To the best of our knowledge, MS-eLCR is a novel electrochemical methylation assay for achieving CpG level resolution without any use of restriction enzymes. Existing electrochemical methylation assays mostly detect regional DNA methylation levels (Kato et al., 2008; Wang et al., 2010; Kato et al., 2011; Wang et al., 2012; Wang et al., 2013) and are limited by availability of restriction enzyme sites at the locus of interest (Dai et al., 2012; Li et al., 2012; Dai et al., 2013). In contrast, MS-eLCR can potentially be designed for any site of interest. Furthermore, we have demonstrated MS-eLCR on real biological samples such as cell line and serum DNA samples whereas most existing electrochemical techniques were demonstrated on synthetic oligonucleotide sequences. To further validate our electrochemical data, the same panel of cell lines (MDA-MB-231, MDA-MB-468 and HCC1937) were analyzed using the analogous fluorescence based LCR assay (Supporting Information, Fig. S4). In addition, we also compared the electrochemical results with NGS, the preferred approach for quantifying methylation at CpG resolution. Our MS-eLCR results were found to be in good agreement with the results generated from both fluorescence and NGS approaches (Fig. 4). The consensus between all three approaches coupled with the consistency of methylation levels reported in the literature (Vrba et al., 2011;

284

K.M. Koo et al. / Biosensors and Bioelectronics 56 (2014) 278–285

demonstrating the feasibility of our method. We believe that this device-based approach could have wide applications as a viable alternative to current technologies in detecting DNA methylation and possibly other single DNA base changes for diagnostics.

Acknowledgments

Fig. 4. Comparison of estimated methylation in three breast cancer cell lines using MS-eLCR (blue), analogous fluorescence approach (red) and 454-Next Generation Sequencing (gray) techniques. Each data point represents the average of the three separate trials (n ¼3) and error bars represent standard error within each experiment. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Wee et al., 2012b) confirmed the accuracy of the MS-eLCR approach and its feasibility as an alternative to current technologies. In addition, a similar methylation trend in the cell line samples was observed in our previous studies (Wee et al., 2013) where MDA-MB-231 was highly methylated and MDA-MB-468 was of low methylation. Despite the wide variety of fluorescence-based approaches developed and established for locus-specific methylation analysis (Eads et al., 2000; Dugast-Darzacq and Grange, 2009; Goedecke et al., 2009), MS-eLCR offers a more cost-effective and simpler alternative assay by avoiding the need for expensive labeling and high background signals commonly associated with fluorescence assays. NGS is an extremely comprehensive methylation quantification technique with single CpG resolution (Potapova et al., 2011) However, it requires expensive instruments as well as timeconsuming experimental procedures and complex data analysis. In contrast, MS-eLCR greatly reduces assay complexity, duration and cost while achieving single CpG resolution. Through crossvalidating MS-eLCR with current established methods, especially one as extensive as NGS, we believe that MS-eLCR may be applicable for clinical applications. While the combined use of microdevices with DNAzymes potentially allows MS-eLCR to be a cost effective alternative approach, a current limitation of MS-eLCR is the lack of massively high throughput, multiplexed detection of multiple CpGs similar to that of current established technologies. However, the duplexed assays used in these experiments highlights its potential for multiplexed measurements of DNA methylation at clinically important loci.

4. Conclusions We have developed a simple microdevice approach for the detection and accurate quantification of locus-specific DNA methylation in both breast cancer cell lines and serum-derived DNA samples. The combined use of LCR amplification, DNAzyme-based electrochemical detection, and microdevice could allow for a simpler and more economical approach than existing traditional technologies for DNA methylation readout. We have demonstrated the feasibility of using peroxidase-mimicking DNAzymes as an electrocatalytic label in detecting as low as 0.04 pM of starting DNA sample and 10–15% DNA methylation at single CpG resolution in complex biological samples such as breast cancer cell lines and serum DNA with good reproducibility (RSD ¼ 6.7%, n ¼3). Most importantly, the analytical accuracy of MS-eLCR was validated using an analogous fluorescent approach and NGS thus

We gratefully acknowledge funding received by our laboratory from the National Breast Cancer Foundation of Australia (CG-08-07 and CG-12-07). These grants have significantly contributed to the environment to stimulate the research described here. This work was also supported by the ARC DECRA (DE120102503) to MJAS and UQ Post-doctoral Research Fellowship (RM ♯ 2009001818) to S. R. The fabrication work was performed at Queensland node of the Australian National Fabrication Facility (Q-ANFF). We also thank Darren Korbie and Erica Lin for supplying the purified serum DNA sample.

Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.bios.2014.01.029. References Benjamin, W.H., Smith, K.R., Waites, K.B., 2003. Methods Mol. Biol. 226, 135–150. Bibikova, M., Fan, J.B., 2009. Methods Mol. Biol. 507, 149–163. Cheng, X.H., Liu, X.J., Bing, T., Cao, Z.H., Shangguan, D.H., 2009. Biochemistry 48, 7817–7823. Clark, S.J., Statham, A., Stirzaker, C., Molloy, P.L., Frommer, M., 2006. Nat. Protoc. 1, 2353–2364. Dai, Z., Cai, T., Zhu, W.Y., Gao, X.Y., Zou, X.Y., 2013. Chem. Commun. 49, 1939–1941. Dai, Z., Hu, X., Wu, H., Zou, X.Y., 2012. Chem. Commun. 48, 1769–1771. Dugast-Darzacq, C., Grange, T., 2009. Methods Mol. Biol. 507, 281–303. Eads, C.A., Danenberg, K.D., Kawakami, K., Saltz, L.B., Blake, C., Shibata, D., Danenberg, P.V., Laird, P.W., 2000. Nucleic Acids Res. 28. Ehrich, M., Nelson, M.R., Stanssens, P., Zabeau, M., Liloglou, T., Xinarianos, G., Cantor, C.R., Field, J.K., van den Boom, D., 2005. Proc. Natl. Acad. Sci. USA 102, 15785–15790. Feinberg, A.P., Tycko, B., 2004. Nat. Rev. Cancer 4, 143–153. Frommer, M., McDonald, L.E., Millar, D.S., Collis, C.M., Watt, F., Grigg, G.W., Molloy, P.L., Paul, C.L., 1992. Proc. Natl. Acad. Sci. USA 89, 1827–1831. Goedecke, S., Schlosser, S., Muhlisch, J., Hempel, G., Fruhwald, M.C., Wunsch, B., 2009. Electrophoresis 30, 1412–1417. Guo, Q.P., Bao, Y.Y., Yang, X.H., Wang, K.M., Wang, Q., Tan, Y.Y., 2010. Talanta 83, 500–504. Kato, D., Goto, K., Fujii, S., Takatsu, A., Hirono, S., Niwa, O., 2011. Anal. Chem. 83, 7595–7599. Kato, D., Sekioka, N., Ueda, A., Kurita, R., Hirono, S., Suzuki, K., Niwa, O., 2008. J. Am. Chem. Soc. 130, 3716–3717. Kong, D.M., Yang, W., Wu, J., Li, C.X., Shen, H.X., 2010. Analyst 135, 321–326. Korpal, M., Lee, E.S., Hu, G.H., Kang, Y.B., 2008. J. Biol. Chem. 283, 14910–14914. Kosman, J., Juskowiak, B., 2011. Anal. Chim. Acta 707, 7–17. Li, W., Wu, P., Zhang, H., Cai, C., 2012. Anal. Chem. 84, 7583–7590. Niazov, T., Pavlov, V., Xiao, Y., Gill, R., Willner, I., 2004. Nano Lett. 4, 1683–1687. Nicoloso, M.S., Spizzo, R., Shimizu, M., Rossi, S., Calin, G.A., 2009. Nat. Rev. Cancer 9, 293–302. Palanisamy, R., Connolly, A.R., Trau, M., 2011. Anal. Chem. 83, 2631–2637. Pelossof, G., Tel-Vered, R., Elbaz, J., Willner, I., 2010. Anal. Chem. 82, 4396–4402. Potapova, A., Albat, C., Hasemeier, B., Haeussler, K., Lamprecht, S., Suerbaum, S., Kreipe, H., Lehmann, U., 2011. BMC Biotechnol. 11. Rauf, S., Shiddiky, M.J.A., Trau, M., Dimitrov, K., 2013. RSC Adv. 3, 4189–4192. Robertson, K.D., 2005. Nat. Rev. Genet. 6, 597–610. Shiddiky, M.J.A., Shim, Y.B., 2007. Anal. Chem. 79, 3724–3733. Soleymani, L., Fang, Z., Sargent, E.H., Kelley, S.O., 2009. Nat. Nanotechnol. 4, 844–848. Tanaka, K., Tainaka, K., Kamei, T., Okamoto, A., 2007. J. Am. Chem. Soc. 129, 5612–5620. Tang, S., Tong, P., Li, H., Gu, F., Zhang, L., 2013. Biosens. Bioelectron. 41, 397–402. Travascio, P., Li, Y.F., Sen, D., 1998. Chem. Biol. 5, 505–517. Vrba, L., Garbe, J.C., Stampfer, M.R., Futscher, B.W., 2011. Genome Res. 21, 2026–2037. Wang, H.-Q., Liu, W.-Y., Wu, Z., Tang, L.-J., Xu, X.-M., Yu, R.-Q., Jiang, J.-H., 2011. Anal. Chem. 83, 1883–1889. Wang, P., Chen, H.B., Tian, J.Y., Dai, Z., Zou, X.Y., 2013. Biosens. Bioelectron. 45, 34–39.

K.M. Koo et al. / Biosensors and Bioelectronics 56 (2014) 278–285

Wang, P., Mai, Z., Dai, Z., Zou, X., 2010. Chem. Commun. 46, 7781–7783. Wang, P., Wu, H., Dai, Z., Zou, X.Y., 2012. Chem. Commun. 48, 10754–10756. Wang, X.L., Song, Y.L., Song, M.Y., Wang, Z.X., Li, T., Wang, H.L., 2009. Anal. Chem. 81, 7885–7891. Wee, E.J.H., Shiddiky, M.J., Brown, M.A., Trau, M., 2012a. Chem. Commun. 48, 12014–12016. Wee, E.J.H., Peters, K., Nair, S.S., Hulf, T., Stein, S., Wagner, S., Bailey, P., Lee, S.Y., Qu, W.J., Brewster, B., French, J.D., Dobrovic, A., Francis, G.D., Clark, S.J., Brown, M.A., 2012b. Oncogene 31, 4182–4195. Wee, E.J.H., Rauf, S., Koo, K., Shiddiky, M.J.A., Trau, M., 2013. Lab Chip 13, 4385–4391.

285

Wiedmann, M., Wilson, W.J., Czajka, J., Luo, J., Barany, F., Batt, C.A., 1994. PCR Methods Appl. 3, S51–S64. Willner, I., Cheglakov, Z., Weizmann, Y., Sharon, E., 2008a. Analyst 133, 923–927. Willner, I., Shlyahovsky, B., Zayats, M., Willner, B., 2008b. Chem. Soc. Rev. 37, 1153–1165. Zhou, L., Du, F., Zhao, Y., Yameen, A., Chen, H., Tang, Z., 2013. Biosens. Bioelectron. 45, 141–147. Zhu, D.B., Xing, D., Shen, X.Y., Liu, J.F., 2004. Biochem. Biophys. Res. Commun. 324, 964–969.

Microdevices for detecting locus-specific DNA methylation at CpG resolution.

Simple, rapid, and inexpensive strategies for detecting DNA methylation could facilitate routine patient diagnostics. Herein, we describe a microdevic...
2MB Sizes 2 Downloads 0 Views