Contents lists available at ScienceDirect

Biosensors and Bioelectronics

journal homepage: www.elsevier.com/locate/bios

Optical biosensing strategies for DNA methylation analysis

Md. Nazmul Islama,b, Sharda Yadava,b, Md. Hakimul Haquec, Ahmed Munazb, Farhadul Islamc,Md Shahriar Al Hossaind, Vinod Gopalanc, Alfred K. Lamc, Nam-Trung Nguyenb,Muhammad J.A. Shiddikya,b,⁎

a School of Natural Sciences, Griffith University, Nathan Campus, Nathan, QLD 4111, Australiab Queensland Micro, and Nanotechnology Centre (QMNC), Griffith University, Nathan Campus, Nathan, QLD 4111, Australiac School of Medicine and Cancer Molecular Pathology laboratory in Menzies Health Institute Queensland, Griffith University, Gold Coast, Australiad Institute for Superconducting and Electronic Materials, Australian Institute for Innovative Materials, University of Wollongong, NSW 2519, Australia

A R T I C L E I N F O

Keywords:DNA methylationEpigenetic biomarkerMethylation assaysOptical biosensorsBisulfite treatment

A B S T R A C T

DNA methylation is an epigenetic modification of DNA, where a methyl group is added at the fifth carbon of thecytosine base to form 5 methyl cytosine (5mC) without altering the DNA sequences. It plays important roles inregulating many cellular processes by modulating key genes expression. Alteration in DNA methylation patternsbecomes particularly important in the aetiology of different diseases including cancers. Abnormal methylationpattern could contribute to the pathogenesis of cancer either by silencing key tumor suppressor genes or byactivating oncogenes. Thus, DNA methylation biosensing can help in the better understanding of cancerprognosis and diagnosis and aid the development of therapies. Over the last few decades, a plethora of opticaldetection techniques have been developed for analyzing DNA methylation using fluorescence, Raman spectro-scopy, surface plasmon resonance (SPR), electrochemiluminescence and colorimetric readouts. This paper aimsto comprehensively review the optical strategies for DNA methylation detection. We also present an overview ofthe remaining challenges of optical strategies that still need to be focused along with the lesson learnt whileworking with these techniques.

1. Introduction

DNA methylation is one of the important epigenetic alterations,which results from the enzymatic addition of a methyl group at the fifthcarbon of the cytosine ring (5mC) (Bird, 2002; Tucker, 2001). AlthoughDNA methylation does not change the sequence of the DNA strands,the level and distribution patterns of 5mC across the genome regulatethe genomic imprinting (Li et al., 1993; Constancia et al., 1998) and X-chromosome inactivation (Mohandas et al., 1981; Robertson andJones, 2000). DNA methylation is also essential in maintaininggenomic stability to prevent the onset of somatic mutations (Jonesand Gonzalgo, 1997; Robertson and Jones, 2000; Taleat et al., 2015).The methyl group of 5mC can be removed via demethylation pathways,which enables cells to manage their own 5mC levels as well as tocontrol gene expression and protein turnover depending on theirneeds. In this way, information encoded by the methylation patternacross the genome defines the cell's identity and function, whichbecomes crucial in the study of embryonic development, cell differ-entiation, and onset of many diseases including cancer (Gonzalgo andJones, 1997; Robertson and Jones, 2000; Kulis and Esteller, 2010).

Detection of 5mC in eukaryotic DNA was first described byHotchkiss in 1948 and later by Wyatt in 1951 (Hotchkiss, 1948;Wyatt, 1951). In cancers, a gradual loss in cytosine methylation levelsthroughout the genome (i.e., global hypomethylation) affects 5mC levelthat are evenly distributed (average 1 CpG in every 150 bp) (Ehrlich,2002; Kulis and Esteller, 2010). Despite this loss, a hallmark of thecancer development is the simultaneous acquisition of increasing 5mClevels in the genome known as hypermethylation (Aggerholm et al.,1999; Ehrlich, 2002; Kulis and Esteller, 2010). Hypermethylationoccurs in cytosines that are clustered in CpG rich regions located atthe promoter and the first exon regions of various tumor suppressorgenes and proto-oncogenes and can in turn causes their transcriptionalsilencing (i.e., regional hypermethylation). However, hypomethylationactivates transcription of proto-oncogenes, proteins involves in geno-mic instability and metastasis in cancer (Herman and Baylin, 2003).Generally, the presence of 5mC in a genomic DNA undergoes multipleepigenetic changes such as chromatin condensation, stabilization ofchromosomes, genomic imprinting and gene silencing (Reik and Dean,2001). Also, by altering methylation regionally at non-clustered sites,cancer cells can reprogram their signalling pathways to ensure their

http://dx.doi.org/10.1016/j.bios.2016.10.034Received 31 August 2016; Received in revised form 5 October 2016; Accepted 18 October 2016

⁎ Corresponding author at: School of Natural Sciences, Griffith University, Nathan Campus, Nathan, QLD 4111, AustraliaE-mail address: m.shiddiky@griffith.edu.au (M.J.A. Shiddiky).

Biosensors and Bioelectronics 92 (2017) 668–678

Available online 19 October 20160956-5663/ © 2016 Elsevier B.V. All rights reserved.

MARK

own survival and acquire the ability to invade other tissues. Thus, theDNA methylation landscapes for cancer and healthy genomes are verydifferent, not only in their global levels but also in their distributionacross the genome (Ehrlich, 2002; Kulis and Esteller, 2010).

There is an increasing evidence that the methylation status ofcertain genes in different cancers provide independent information onpatient prognosis (Tang et al., 2000). For instance, Zou et al. (2002)provided strong evidences that the detection of p16 methylation in theserum samples of patients with colorectal carcinoma was significantlyassociated with advanced pathological stages (stage 3 or 4). Also, incolorectal carcinoma, methylation is a common cause of microsatelliteinstability (MSI) as detected by loss of immunohistochemistry.Sporadic MSI+ colorectal carcinomas result from methylation ofmismatch repair genes (Pakneshan et al., 2013). Similarly, methylationof SHOX2 in lung cancers (Dietrich et al., 2012) and PITX2 promotermethylation in prostate (Dietrich et al., 2013) and breast carcinomas(Nimmrich et al., 2008) were correlated with patient prognosis.Therefore, accurate quantification of methylation level at wholegenome and regional levels not only provides a characteristics signa-ture of gene function but also gives a better understanding about theclinical and prognostic roles of DNA methylation in different cancers.

Over the past several decades, various approaches have beenextensively developed to quantify the level of the DNA methylation inresearch and clinical applications (Fig. 1.) (Singer-Sam et al., 1990;

Frommer et al., 1992; Herman et al., 1996; Gonzalgo and Jones, 1997;Stach et al., 2003; Gao et al., 2005; Weber et al., 2005; Kristensen et al.,2008; BonDurant et al., 2011; Hibi et al., 2011). For example, in 1992Frommer et al. (1992) introduced the bisulfite conversion method forthe analysis of 5mC. Since then, this method has been widely used inclinical applications. Several other methods such as methylationspecific (MS) polymerase chain reaction (MS-PCR) (Herman et al.,1996), real-time PCR (BonDurant et al., 2011), quantitative MS-PCR(Hibi et al., 2011), methylation-sensitive single nucleotide-primerextension (MS-SnuPE) (Gonzalgo and Jones, 1997), methylight (Eadset al., 2000), sensitive melting analysis after real-time MS-PCR(Kristensen et al., 2008) and microarray based (Yan et al., 2001; Gaoet al., 2005) methods have also been adopted with the bisulfiteconversion method to achieve better quantification of 5mC. Most ofthese methodologies typically require large sample volumes due toDNA degradation during bisulfite conversion and are often limited bylow amplification efficiency and PCR bias. Microarray based methodsovercome the need for sequencing but still rely on bisulfite conversionand PCR amplification. To avoid the use of bisulfite conversion andPCR amplification, enzyme-linked immunosorbent assay (ELISA)based methylation assays (So et al., 2014; Kurdyukov and Bullock,2016) were proposed. These assays were relatively simpler but affectedby low sensitivity and require multiple external controls for quantita-tive analysis. To address these issues, methyl binding domain (MBD)

Fig. 1. An overview of the development of DNA methylation analysis techniques.

Md. Nazmul Islam et al. Biosensors and Bioelectronics 92 (2017) 668–678

669

based assays were developed, where MBD protein or other specificantibody (i.e., MeDIP-seq) were used to selectively detect the methy-lated CpG sites (Weber et al., 2005) within the target DNA sequence.

Besides bisulfite conversion and MBD based methods, high perfor-mance liquid chromatography (HPLC) and mass spectroscopy (MS)have also been used for the direct quantification of DNA methylation(Ramsahoye, 2002; Liu et al., 2009; Armstrong et al., 2011; Lisantiet al., 2013). Although HPLC and MS based methods detect DNAmethylation directly, they also require a large amount of input DNAwhich limit their use in routine clinical applications. Other MS basedstrategies including matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) (Hillenkamp et al., 1991) have also beenreported for the simultaneous quantification of DNA methylation(Friso et al., 2002; Tost et al., 2003; Le et al., 2011; Hu et al., 2013;Le et al., 2011; Humeny et al., 2003).

In recent years, a number of fluorescence readout based methods(Duan et al., 2010; Taleat et al., 2015; Ma et al., 2015; Wang et al.,2016), have successfully been used to quantify DNA methylation.Among these fluorescence readouts, fluorescence resonance energytransfer (FRET) is most commonly used optical readout method inDNA methylation analysis due to its superior sensitivity (Bailey et al.,2010; Cao and Zhang, 2012; Chen and Wang, 2013; Dadmehr et al.,2014; Duan et al., 2010). Semiconductor quantum dots (QDs) have alsobeen adapted in fluorescence based methylation assays (e.g., methyla-tion sensitive (MS)-qFRET) to further improve the assay performanceconsidering the fact that QDs offer broader absorption spectra andshow higher detection efficiency when compared to those of conven-tional fluorophores (Clapp et al., 2006; Resch-Genger et al., 2008;Bailey et al., 2009; Ma et al., 2015). Other optical readout methodssuch as Raman spectroscopy (Wang et al., 2016), surface plasmonresonance (SPR) (Sina et al., 2014), electrochemiluminescence (ECL)(Zhao et al., 2015) and colorimetry (Wee et al., 2015a) have also beenwidely used for analyzing DNA methylation (Kurita et al., 2012; Liet al., 2012; Zhao et al., 2015).

The advancements of DNA methylation detection strategies havebeen reviewed recently (Zhang et al., 2015; Taleat et al., 2015,Hernandez, 2013). For instance, Zhang et al. (2015) reviewed the useof conventional approaches in methylation analysis whereasHernandez et al. (2013) discussed PCR based strategies. Later, Taleatet al. (2015) reported a more comprehensive review on the currentmethylation detection techniques evolved till 2014. To the best of ourknowledge, till date, no review has solely covered the available opticaltechniques for DNA methylation detection. The present review gives abrief outline of the bisulfite sequencing based methylation studiesfollowed by a comprehensive overview of the recent advancements inoptical detection strategies for methylation analysis. More importantly,we highlights the technical challenges that have been addressed tomake each of the techniques suitable for DNA methylation analysis. Inthe end, we also put our perspectives on the future direction of thefield.

2. DNA methylation analysis strategies

2.1. Sodium bisulfite treatment based techniques

Until 1992, the regular DNA sequencing technologies were unableto distinguish cytosine and 5mC. To solve this problem, complexsequencing strategies based on different digestion and chemicalcleavage steps were proposed (Church and Gilbert, 1984; Saluz andJost, 1989). Most of these strategies were limited by the low sensitivity.Additionally, the sequencing results were obtained by averaging apopulation of DNA. Therefore, these methods failed to detect methyla-tion level in a small volume of sample. The bisulfite sequencingmethod, introduced by Frommer and colleagues in 1992, can quantita-tively distinguish 5mC from cytosine (Frommer et al., 1992). Themethod was based on the sodium bisulfite-mediated conversion of

cytosine to uracil in single-stranded DNA, where cytosine is convertedto uracil but 5mC remains unchanged. After PCR amplification andsequencing, uracil derived from the cytosine is read as thymidine. Onthe other hand, as 5mC is immune to bisulfite conversion, it is countedas cytosine. This method has been successfully employed to developseveral bisulfite sequencing based methylation assays such as MS-PCRand Ms-SnuPE have been developed (Herman et al., 1996; Gonzalgoand Jones, 1997). In MS-PCR, bisulfite treated DNA was PCR amplifiedusing two methylation specific primers. As amplification was done by5mC specific primers, the presence of amplified 5mC in the PCRproduct confirmed the methylation of target sequence. MSPCR suc-cessfully eliminated the common PCR bias, however, it could notanalyze the methylation quantitatively (Herman et al., 1996). A followup strategy known as Ms-SnuPE was able to quantify DNA methylationby electrophoretic separation and radiation analysis of sample(Gonzalgo and Jones, 1997).

Bisulfite treated DNA was further studied for methylation analysisby (i) pyrosequencing technology (Tost and Gut, 2007), (ii) meltingtemperature based sequence-free methods (e.g., methylation-specificdenaturing gradient gel electrophoresis (MS-DGGE, Aggerholm et al.,1999), methylation-specific melting-curve analysis (MSMCA, Wormet al., 2001) and high-resolution melting based assay (Wojdacz andDobrovic, 2007)) and (iii) treating with restriction enzymes (e.g.,combined bisulfite-restriction analysis (COBRA), Xiong and Laird,1997). Pyrosequencing technology was later adapted in next generationsequencing (NGS) technology (e.g., reduced representation bisulfitesequencing (RRBS)) for high throughput and cost-effective analysis(Meissner et al., 2005; Taylor et al., 2007; Hirst and Marra, 2010; Guet al., 2010; Li et al., 2010; Gu et al., 2011).

2.2. Methyl CpG binding domain (MBD) based techniques

Bisulfite sequencing of DNA derived from heterogeneous samplecan result in an altered representation of original methylation pattern(Warnecke et al., 1997). This limitation was overcome by the successfuluse of MBD proteins in methylation assays (Hiraoka et al., 2012;Shiraishi et al., 2004). The MBD based methods are also less vulnerableto DNA damage (one of the disadvantages of bisulfite sequencingmethod) (Seguin-Orlando et al., 2015). Generally, MBD proteins (i.e.,MeCP2, MBD1, MBD2, MBD3 and MBD4 in mammals) consist of aspecial domain (i.e., MBD) which has very strong affinity for 5mCregion of DNA. Thus, MBD proteins recognize and bind methylatedCpG region in double-stranded DNA (dsDNA) with high affinity(nanomolar KD) (Taleat et al., 2015; Kurdyukov and Bullock, 2016).In 2004, Shiraishi and colleagues used MBD protein to detect 5mC bycolumn chromatography, and demonstrated that the binding efficiencybetween MBD protein and 5mC is directly proportional to the level ofthe 5mC present in the genome (Shiraishi et al., 2004). However, thismethod was limited by the large nonspecific binding to MBD proteins.This nonspecific binding can be avoided using restriction enzymes inMBD based assays. For example, in a restriction enzyme based assayreferred to as COMPARE-MS (combining methylated-DNA precipita-tion with methylation-sensitive restriction enzyme), target DNA wasfirst digested by methylation-sensitive restriction enzymes and thenprecipitated by MBD polypeptides immobilized on a magnetic beadfollowed by RT-PCR for quantitative detection (Yegnasubramanianet al., 2006).

MBD proteins or MBD specific antibodies have also been used toenrich the CpG islands for large scale analysis of global methylation.For example, for detecting DNA methylation in genomic DNA, CpGregions were enriched by MBD specific antibodies (e.g., MeDIP-seq)(Weber et al., 2005) or methyl binding domain protein (e.g., MBD-seq)(Aberg et al., 2012), which were then detected using microarray or NGSbased readouts. The accuracy of these approaches was, however,affected by several downstream amplifications steps. This drawbackwas avoided in an improved protocol, where a DNA-mediated floccula-

Md. Nazmul Islam et al. Biosensors and Bioelectronics 92 (2017) 668–678

670

tion assay was coupled with the MBD enrichment step for naked eyeevaluation of DNA methylation (Wee et al., 2015b).

2.3. Enzymatic digestion based assays

Enzymatic digestion based methylation assays have been developedlong before the inception of bisulfite conversion based assays (Bird,1978; Cedar et al., 1979). Generally in an enzymatic digestion basedmethylation assay, the region of interest of DNA (i.e., 5mC) can bedigested selectively using specific restriction enzymes (e.g., endonu-clease MspI), followed by the quantification via a suitable readoutmethod. In 1979, Cedar and colleagues reported the first approach todetect DNA methylation directly in eukaryotes by using radiolabeledMspI and HpaII restriction enzymes (Cedar et al., 1979). In thismethod, target DNA was first digested with MspI and HpaII enzymes,where MspI cleaved the target region of DNA (irrespective of whetherthe sequence is methylated or not) and HpaII cleaved only theunmethylated target region. These digests were then run on gelelectrophoresis for the visual separation of methylated and unmethy-lated DNA.

Due to the ability of sensitive selection of methylated region,enzymatic digestion based assays have been adapted in many widelyused regional methylation detection platform including COBRA (Xiongand Laird, 1997), Methylight (Eads et al., 2000) and ECL (Li et al.,2012). The enzymatic digestion based assays have also been developedfor the analysis of global methylation. In 1999, Hatada et al. haveshown that restriction landmark genomic scanning (RLGS) methodcould be used to analyze global DNA methylation (Hatada et al., 1991).Since then the RLGS has been extensively used in analyzing globalDNA methylation (Costello et al., 2002; Rush and Plass, 2002). Thoughenzymatic digestion based assays are very promising, their applicationsin point-of-care diagnostics are limited by radiation hazards andrequirements of expertise.

2.4. Major drawbacks of bisulfite based techniques and currentrequirements

Despite being the most widely used DNA methylation analysismethod, bisulfite based methods have some common drawbacks. Oneof the drawbacks is the DNA fragmentation which could result chimericproduct, making amplification of long fragments difficult. Anotherdrawback is their uncertainty of complete conversion. This could be theresult of either failure of deamination of cytosine or an inappropriateconversion of 5mC to thymine (Genereux et al., 2008). The failure ofdeamination of cytosine in bisulfite conversion thus gives an alteredassessment of DNA methylation. On the other hand, due to theinappropriate conversion of 5mC to thymine, 5mC is mistaken asunmethylated which also gives biased representation of the methyla-tion status. These limitations of bisulfite conversion along with the useof radioactive materials, limited read length of DNA, the need forcumbersome procedures and specialized instruments limit their wideapplications in screening of DNA methylation in clinics. Althoughsequencing based method such as NGS is highly cost-effective and canpotentially analyze methylation at the nucleotide level in the wholegenome, it has significant bioinformatic challenges (i.e., due to thelarge sequencing data management) (Hirst and Marra, 2010). Variousmelting temperature based sequence-free methods including MS-DGGE and MSMCA are unable to precisely detect methylation patternat single base level. This drawback was resolved by the development ofthe high-resolution melting (HRM) method. However, HRM method isnot suitable for methylation screening of heterogonous samples.Generally, clinical samples contain a heterogeneous population of cells.This diverse variability in cellular composition can heavily interfere themethylation analysis, leading to high differences between presumablytwo identical tissue samples. Therefore, a method that uses smallvolumes of sample (low number of homogenous cells) is highlydesirable for overcoming complication in extracting accurate methyla-tion information. On the contrary, analysis of large cohort samples isalso required to understand prognosis and diagnosis of many diseasesincluding cancer. Therefore, robust and cost-effective methods areneeded for analyzing a large cohort of samples. Currently a significant

Table 1Optical detection strategies for DNA methylation.

Detection strategy Startingmaterials

Note on methylationbiosensing

Detectionlimit

Advantages and limitations Ref.

Fluorescence Synthetic DNA andhuman lung cancercell line

Circular padlock probes formationin presence of 5mC, hyperbranchedrolling circle amplification.

0.8 fM Highly sensitive, alternative PCR freeamplification, restriction enzyme free.Complex operation, multi-step process.

(Cao andZhang,2012)

Colorimetry WGA DNA andHeLa and DuCapcell line

MBD based magnetic enrichment,streptavidin coated horse radishperoxidase as color reporter.

37.5 picogram

Naked eye assessment, both global and sitespecific methylation detection, low samplevolume requirement. Require multiplecontrols, require methyl binding domain(MBD).

(Wee et al.,2015a)

Quantum dots (QDs) in FRET Synthetic DNA andlung cancer tissue

Water soluble QDs as FRETdonormethylation sensitive restriction,alexa Fluor −647 (FRET acceptor)giving FRET signal in presence of5mC.

Notmentioned

Multiplexing capacity, simultaneous assaymonitoring, photostable sensor. PCR bias,restriction digestion required, expertiserequired.

(Ma et al.,2015)

Surface Enhanced RamanSpectrophotometry (SERS)

Synthetic DNA Gold nanoparticle modified captureprobe as SERS substrate, cyaninedye as SERS label, single baseextension reaction (in presence of5mC).

3 pM Highly sensitive and accurate, multiplexingability, real time analysis. Expertise required,toxicity of SERS label, larger fluorophores

(Hu andZhang,2012)

Surface Plasmon Resonance (SPR) Synthetic DNA Sandwich recognition by antibodyand MBD protein, increasedrefractive index in presence of 5mC.

5 pM Real time and label free detection, explainsample kinetics and thermodynamics duringbinding. Large sample volume required,nonspecific bindings

(Pan et al.,2010)

Electrochemiluminescence based Synthetic DNA andDNA bacteriophage

Acetylcholine esterase labeled antimC antibody, ECL emission from Ru(bpy)3

2+ luminophore.

0.18 pmol Highly sensitive, PCR-free, digestion-free,bisulfite conversion-free. Antibodyrequirement.

(Kuritaet al.,2012)

Md. Nazmul Islam et al. Biosensors and Bioelectronics 92 (2017) 668–678

671

level of advances has been made along this line and as a result, anumber of optical based DNA methylation readout strategies have beendeveloped which offer several advantages over conventional methodsincluding low sample requirement, high-throughput screening, sensi-tive limit of detection (LOD), real-time analysis, label-free detection,PCR bias free approach, heterogeneous methylation screening, nakedeye evaluation etc. (Table 1).

3. Optical detection of DNA methylation

Fluorescence, electrochemiluminescence, SERS, SPR, and colori-metric readouts are the most commonly used optical detection methodsin methylation assays (Fig. 2, Table 1). In this section, these develop-ments will be discussed.

3.1. Fluorescence based techniques

Among many other fluorescence detection methods, FRET is themost widely used method for detecting DNA methylation as FRETbased readouts offer high sensitivity in methylation detection due totheir less susceptibility to the nonspecific interferences. In FRET, adonor fluorophore from an electronically excited state gives electron toan accepter chromophore resulting change of fluorescence intensity.This energy transfer happens due to the dipole–dipole interactionsbetween a donor and an acceptor, where the FRET probability isinversely proportional to the distance between donor and acceptor bythe sixth power (Miyawaki, 2003; Freeman and Willner, 2012). In DNAmethylation analysis, several donor and acceptor fluorophores includ-ing cationic conjugated polymers (CCPs) (Feng et al., 2008; Yang et al.,2012), quantum dots (QDs) (Ma et al., 2015) and upconversionnanoparticles (UCNPs) (Wu et al., 2016) have been used. Usually, in

FRET based methylation assays, donor fluorophores specifically canbind the target (i.e., 5mC). When donors are excited at their absorptionspectra, they transfer energy to the acceptor molecules. By obtainingtransferred energy, acceptor molecules then results a significantfluorescence amplification. Since unmethylated DNA do not bind to5mC specific donor molecules, they do not transfer energy to theacceptor molecules resulting no FRET signal (Fig. 2).

Over the past several years a number of FRET based methylationanalysis methods have been developed. In most of these methods, CCPsare used as FRET donors as they act as powerful transducer whichtriggers electron transfer from acceptor fluorescein through FRET.Since fluorescence intensity is obtained from both CCP and fluorescein,it gives a strong optical amplification allowing sensitive detection ofDNA methylation (Feng et al., 2008; Duan et al., 2010; Yang et al.,2012). For instance, it was successfully employed to detect as low as 1%methylation status of CpG sites in human colon cancer cells (Fenget al., 2008). Duan et al. (2010) demonstrated another CCP basedFRET assay for the detection of DNA methylation (Fig. 3A.). In thismethod, bisulfite converted DNA was initially amplified by PCR. Singlebase extension of the amplified DNA samples were then performedusing Taq polymerase in the presence of fluorescein labeled deoxygua-nosine triphosphate (dGTP-Fl) and methylation-specific probe. ThedGTP-Fl specifically recognized and bound complementary probe ofmethylated DNA but did not attach with the unmethylated probe.Consequently, after the addition of cationic polymer PFP1a, onlymethylated DNA sample resulted a strong FRET readout leavingunmethylated counterpart irresponsive. This assay was validated inthe genomic DNA sample extracted from human colon cancer cell line,HT29.

In recent years, quantum dots (QDs) based FRET donors have alsobeen used to improve the sensitivity of the fluorescence based

Fig. 2. Typical optical detection strategies for DNA methylation analysis.

Md. Nazmul Islam et al. Biosensors and Bioelectronics 92 (2017) 668–678

672

methylation assays (Bailey et al., 2010; Ma et al., 2016;). QD is asemiconductor nanocrystal which stays in a core shell structure. Thefluorescence emission from QDs is dependent on the size of the crystals

and is unaffected by photobleaching. Additionally, while many otherorganic fluorophores are affected by the narrow absorption spectra,QDs provide wide absorption spectra along with narrow emission

Fig. 3. (A) Schematic representation of FRET based assay. Bisulfite converted and PCR amplified-methylated DNA was extended with Taq polymerase in the presence of dGTP-Fl. Inthe detection step, methylation-specific probe and PFP1a resulted strong FRET signal whereas unmethylated sample produced no FRET response. (B) An ECL based DNA methylationassay. The acetylcholinesterase labeled anti methylcytosine antibody was used to quantify DNA methylation by measuring ECL emission from Ru(bpy)3

2+ luminophore. Reproducedfrom Duan et al. (2010) and Kurita et al. (2012) with permission from The American Chemical Society.

Md. Nazmul Islam et al. Biosensors and Bioelectronics 92 (2017) 668–678

673

spectra and high extinction coefficient (Clapp et al., 2006; Resch-Genger et al., 2008; Ma et al., 2015; Ma et al., 2016). In 2009, Baileyand colleagues developed a QD based FRET analysis for DNAmethylation referred to as MS-qFRET which could detect as small as15 pg of methylated DNA from large alleles (Bailey et al., 2009). In thismethod, bisulfite treated DNA was amplified with biotin taggedforward primers and fluorophore labeled reverse primer. Due to thestrong streptavidin- biotin interaction, the labeled-PCR product wascaptured by streptavidin functionalized QDs. Then FRET stimulationbetween the QD donor and the fluorophore acceptor resulted theemissions of fluorophores (accompanied by quenching of QDs) whichgave the relative level of DNA methylation. Although, this method wasuseful for multiplexed analysis and long-time monitoring of themethylation, it was also limited by high background noise and non-specific PCR amplicons. These drawbacks could easily be avoided usingFRET linker probes (FLPs) as outlined by Keeley et al. (2012). In FLPsbased assay, after bisulfite conversion and PCR amplification withmodified primers, the FLP distinguished the target amplicon, and thuseliminated the background noise from the non-specific PCR amplicons.

Ma et al., (2015) reported an advanced QD-FRET based DNAmethylation assay, where enzymatically digested and PCR-amplifiedDNA was incorporated with a fluorophore to get the fluorescencereadout. Signal amplification from QDs to the fluorophores uponinteraction with FRET allowed quantitative determination of methy-lated DNA (unmethylated DNA is completely insensitive to FRET).More recently, this method has been extended to quantify DNAmethylation in specimens derived from lung cancer patients (Maet al., 2016). The advantages of using QDs in methylation assays liein their inherent sensitivity and multiplexing capacity as well as theirversatile ability in detecting methylated targets via other readouts. Forexample, QDs based assays have been well adapted in an electroche-mical detection, known as methylation-specific ligation-detection reac-tion (EmsLDR), where multiple QDs have been used for the simulta-neous detection of methylation levels (Dai et al., 2013).

Upconversion nanoparticles (UCNPs, e.g., Ln3+, Ti2+, Ni2+) havealso been introduced as a new class of fluorophores in FRET basedmethylation assays (Kim et al., 2016; Wu et al., 2016). UCNPs haveexcellent optical properties including large anti-stokes shift, highphotochemical stability and minimal background signal. UCNPs alsoemit light in the visible wavelength region enabling easy opticaldetection. Wu et al. (2016) used combination of lanthanide-dopedUCNPs and gold nanorods for the detection of DNA methylation, wherea DNA duplex complementary to the target region worked as the linkerbetween donor and acceptor fluorophores. Importantly, this methodavoided bisulfite conversion and PCR amplification. Beside these FRETbased assays, membrane-based near-infrared fluorescence assay hasalso been developed (Chen and Wang, 2013). This assay is capable ofdetecting both transcription and methylation status of a hypermethy-lated gene without the use of bisulfite treatment and PCR amplification,and thereby offering a relatively simpler methylation detection plat-form.

In 2012, rolling-circle amplification (RCA) process, as an alter-native to PCR, has been integrated with fluorescence readout forquantitative analysis of DNA methylation (Cao and Zhang, 2012).Under isothermal conditions, RCA can produce hundreds of DNAcopies within a very short time using a single primer (Lizardi et al.,1998). In the method described by Cao and Zhang, circular padlockprobes were first produced via ligating the bisulfite treated methylatedDNA samples with linear padlock probes. Upon purification, thecircular probes were amplified by hyperbranched RCA process anddetected by the fluorescence readout. Beside these advances influorescence based direct detection strategies, a significant advance-ment has also been made in fluorescence based indirect detectionstrategies based on DNA methyltransferase ( Liu et al., 2015a; Weiet al., 2015; Xue et al., 2015; Cui et al., 2016).

3.2. Techniques based on electrochemiluminescence

In electrochemiluminescence detection system, oxidizing or redu-cing potentials are applied to chemical systems (containing target) inpresence of luminophores. Depending on the presence of target,underlying chemical reaction is triggered causing emission of detect-able lights from the luminophores (Liu et al., 2007; Wei and Wang,2011; Doeven et al., 2015; Liu et al., 2015b). A number of materials(e.g., transition metal complexes, organic molecules and nanomater-ials) have been studied as ECL active luminophores. Among these, tris(2, 2′‐bipyridyl) ruthenium(II) (Ru(bpy)3

2+) and its analogues alongwith co-reactants tri-n‐propylamine (TPrA, and its alternatives) havebeen mostly used (Wei and Wang, 2011) as they offer relatively betterstability, good water solubility, spatial and temporal controllability, andwide dynamic range while producing luminescence. Generally, in DNAmethylation assays (Fig. 2), a methylation specific chemical reactionpathway is incorporated on the electrode surface which only occurs inthe presence of methylated DNA. Thus, if 5mC is present this reactionoccurs and excites the luminophore resulting ECL light that allowsquantification of 5mC level present in the sample (Kurita et al., 2012;Zhao et al., 2015). Surprisingly, until 2012, there was no report on ECLbased DNA methylation detection assays. This could be due to the factthat conventional intercalation or hybridization based DNA methyla-tion assays are believed to be not suitable to selectively confirm thatluminescence are derived only from the 5mC present in the targetsample.

In 2012, Kurita et al. (2012) have reported that ECL can be used tomeasure the level of DNA methylation. In this method, they usedacetylcholinesterase labeled anti 5mC antibody to selectively capturethe methylated DNA. Thiocholine was then produced from enzymaticreaction between acetylcholine and acetylcholinesterase, which wasadsorbed on gold surface by gold-thiol affinity binding. Upon applica-tion of potential to the electrode, ECL emission from Ru(bpy)3

2+

luminophore gave the quantitative level of the methylation present intarget DNA (Fig. 3B). This method was successfully challenged in48,502 bp-long DNA bacteriophage. A similar ECL method for thedetection of DNA methylation and methyltransferase activity has beendescribed by Li et al. (2012) where gold electrode was modified withthe thiol-linked single strand DNA (ss-DNA) and ruthenium complex.Upon hybridization with the complementary ssDNA, this was treatedwith HpaII endonuclease. When cytosine was methylated, HpaII couldnot cleave the cytosine region whereas it cut between C-C in unmethy-lated cytosines, resulting a decrease in ECL signal.

Click chemistry based ECL assay has been developed to quantifyDNA methylation (Zhao et al., 2015), where GO/AgNPs/luminolcomposite offers an extra specificity by producing amplified-ECLsignal. This composite was functionalized with alkene to facilitate clickchemistry. First azide-terminated duplex DNA modified electrode wasformed upon hybridization of azide-terminated complementary DNA.Alkynyl GO/AgNPs/luminol composite was then accumulated on theelectrode via click chemistry. When methylation-responsive restrictionendonuclease DpnI was added on the electrode, GO/AgNPs/luminolwas released from the electrode providing decreased ECL signalthereby confirming the presence of methylation. Unlike bisulfitesequencing and restriction enzyme based methodologies, the ECLbased methylation assays avoid the use of bisulfite conversion, PCRamplification and enzymatic digestion steps. Other advantages of ECLbased methylation assays include high sensitivity, rapid analysis, lowvolume of sample, and wide dynamic range. However, development ofECL high-throughput assay based on point-of-care testing with highsensitivity and good stability are required for routine clinical screeningof DNA methylation. Though the production of multiple emitters in acommon co-reactant has already been developed (Doeven et al., 2015)for multiplexed ECL detection of biological targets, it is yet to beincorporated in DNA methylation analysis.

Md. Nazmul Islam et al. Biosensors and Bioelectronics 92 (2017) 668–678

674

3.3. Surface-enhanced Raman scattering (SERS) based techniques

Recent demonstrations have shown that SERS can be used inbiosensing by two different ways- (i) the label-free detection of targets(direct detection of SERS fingerprint of adsorbed targets) via measur-ing the SERS intensity and (ii) indirect detection of the target speciesusing SERS labels such as enzymes, dyes, quantum dots and fluor-ophores (Doering et al., 2007; Schlucker, 2009). It has also demon-strated that the target species do not always need to be directlyadsorbed onto the metallic surface for obtaining the SERS spectra(Laing et al., 2016). To observe the most intense SERS response, targetmolecules must be within few nanometers distance from the metalsurface (the degree of SERS signal enhancement is dependent on thedistance between metal surface and the target, Laing et al., 2016).Using these distinct SERS characteristics, DNA methylation can beassessed by immobilizing methylation specific capture probe on themetallic substrate of SERS along with the use of methylation specificSERS labels (Wang et al., 2015; Hu and Zhang, 2012). In recent years,SERS based readout has successfully been used in developing methyla-tion analysis methods. For instant, Hu and Zhang (2012) developed ahighly sensitive readout method using SERS labels which could detect1% methylation level in p16 tumor suppressor gene. In this study, goldnanoparticle (modified with capture probe) was used as a SERSsubstrate and cy5-dGTP, a guanine base with cy5 label (cyanine dye),was used as a SERS label. When methylated DNA was attached to thegold nanoparticle-modified capture probe, it triggered a single baseextension reaction between gold substrate and cy5-dGTP resultingtheir hybridization. For an unmethylated DNA sample, no baseextension reaction occurred. This binding event showed a high SERSreadout with a detection limit of 3 pM DNA. In another approach,Wang and colleagues have enhanced the sensitivity of the SERS assayby coupling a ligase chain reaction (LCR) amplification with a SERSmethylation assays (Wang et al., 2015). In their assay, bisulfiteconverted DNA was first amplified with PCR and then with LCR. Inunmethylated LCR products, cytosines converted to thymines whereasin methylated LCR products cytosines remained unchanged. Followingadsorption of LCR products on the SERS surface, specific Raman peakswere observed for methylated and unmethylated DNA. The method wassuccessfully tested in a panel of breast cancer cell lines. More recently,this study has been extended to detect global DNA methylation usingSERS nanotags and MBD. This advanced method is able to distinguishas minimum as 6.25% methylation changes in genomic DNA (Wanget al., 2016).

While classical Raman scattering is unable to detect molecularsample at low molar concentration (or low sample volume), SERS cansuccessfully detect these targets with excellent sensitivity. The accuracyof the SERS detection lies on its unique electromagnetic enhancementmechanism, which is absent in conventional Raman and fluorescencereadouts (Stiles et al., 2008). Additionally, compared to other availablemethods for DNA methylation analysis including conventional fluor-escence, chemiluminescent and electrochemical approaches, SERSbased DNA methylation assays have distinct advantages. In variousfluorescence based assays, fluorophores usually suffer from largespectral overlap when the sample contain multiple targets thus makingthe detection difficult. This issue can easily be avoided by SERS as itprovides sharp and narrow spectra allowing multiplexed analysis of thetarget (Harper et al., 2013; Laing et al., 2016). Although, it has beenreported that SERS based detection of oligonucleotide is relativelymore sensitive than conventional fluorescence based readout method,no comparative studies between SERS and fluorescence based DNAmethylation detection have been reported. SERS based methylationassays require minimal sample preparation and avoid issues withinterference, long term stability and non-specific adsorption whichare common in conventional electrochemical sensing systems (Wanget al., 2008).

3.4. Techniques based on SPR

SPR can analyze the binding event of target molecules (in thesolution) on the metallic surface (a thin metal film substrate) uponadsorption. If target binds the metallic sensor surface, refractive indexof the metal substrate changes in real time. Thus, the amount of boundtarget and rate of the association and disassociation of targets can becalculated from the respective spectral shift (McDonnell, 2001;Homola, 2008). In SPR based methylation assays, the detection canbe obtained by the direct detection mode where the methylationspecific capture probe is immobilized on the SPR sensor surface.Methylated target in solution binds to the surface-bound captureprobe, resulting a refractive index change detected by the SPR sensor.Methylation specific recognition elements (e.g., MBD protein or 5mCantibody) can also be used to further recognize the methylation sites(resulting a further refractive index change). This has been demon-strated by Pan et al. (2010), where a double recognition mechanismwas used to enhance the analytical performance of the methylationassay. A complementary target DNA was first captured by themethylation specific probe immobilized onto the SPR sensor chip,which was further recognized by the recombinant MBD protein.Interaction between the methylated DNA and the MBD protein onthe chip surface resulted in an increase in the refractive index, whichgenerated a detectable optical signal. This method was capable ofdetecting 5 pM of methylated adenomatous polyposis coli (APC)promoter gene.

The level of 5mC can also be detected without the use of methyla-tion specific capture probe, MBD protein, or 5mC antibody, wheremethylated DNA can directly be adsorbed on SPR sensor chip (e.g.,gold surface) resulting a refractive index changes compared to that ofthe unmethylated control. This has been experimentally demonstratedby Sina et al. (2014). The underlying principle of their method relies onthe base dependent interaction affinity interaction between individualDNA bases and unmodified gold substrate (Sina et al., 2014; Koo et al.,2014; Koo et al., 2015). Bisulfite treatment converts unmethylatedcytosines into uracils but methylated cytosine remains unchanged. Asubsequent asymmetric PCR step amplifies the antisense strand of thebisulfite-treated DNA to generate ssDNA products. This step alsoconverts uracils into adenines in the antisense strand of unmethylatedDNA whereas guanines are generated in the methylated DNA. SinceDNA–gold affinity interaction follows the trend A > C >G > T, theadenine-enriched unmethylated DNA leads to a larger level of adsorbedDNA on the SPR sensor surface in comparison to the guanine-enrichedmethylated DNA. This different adsorption patterns resulted twodifferent spectral shifts for methylated and unmethylated sampleallowing the methylation detection (Fig. 4). The sensitivity of thismethod was tested in genomic DNA extracted from breast cancer celllines such as MCF7 and MDA-MB-231.

In recent years, several other DNA methylation methods have beendeveloped based on SPR readout. For instance, an end-to-end nanorodassembly enhanced-SPR system has been developed for the detectionof both DNA methylation and adenine methylation methyltransferase(DamMTase) activity (Li et al., 2015). The method showed a goodrelationship between the SPR spectral shifts. SPR readout has also beensuccessfully adapted in microfluidics (Kurita et al., 2015) and mole-cular inversion probe (MIPs) (Carrascosa et al., 2014) based methyla-tion analysis methods. For example, Kurita et al. (2015) have devel-oped an SPR based microfluidic DNA methylation assay, wheresequence-selective immunochemical discrimination of the 5mC statusin a genomic sample has been detected within 45 min. This methodavoids the use of conventional bisulfite treatment, PCR amplification,sequencing and enzymatic digestion steps. In the MIP based SPR assay(Carrascosa et al., 2014), bisulfite treated genomic DNA was investi-gated with a MIP using a fill-in approach (Palanisamy et al., 2011),where MIP binds to the DNA and generates a wide gap of several basesbetween the MIP recognition ends. This represents the methylated

Md. Nazmul Islam et al. Biosensors and Bioelectronics 92 (2017) 668–678

675

region to be investigated. This gap was then filled by a polymerase andsubsequently ligated by a ligase, leading to a circularized DNA probecarrying the imprinted information from the target region. Finally, theMIP circularized product was amplified by asymmetric PCR to generatea ssDNA amplicon, which was captured by a receptor oligo sequencepreviously immobilized onto the SPR sensing surface. This method wassuccessfully applied to detect regional methylation in the cancer celllines.

Besides the real-time and label-free nature of the SPR readout, ithas other advantages. For example, (i) it offers the analysis ofreceptor–target interactions with a wide range of molecular weightsand affinities, (ii) it is also highly compatible with different chemicalenvironments (e.g., temperature, ionic strength and pH) (McDonnell,2001), and (iii) it can explain the DNA molecular recognition events viaanalyzing binding kinetics and thermodynamics under different con-ditions. For example, in a microarray based DNA methylation assay,SPR was used to measure the binding kinetics MBD protein anddsDNA (Yu et al., 2010). Although these current demonstrations offerreal-time, label-free analysis of DNA methylation, they are limited bythe poor sensitivity of SPR readout (Green et al., 2000).

3.5. Colorimetric detection techniques

Colorimetric assays for the detection of DNA methylation hasgained recent interest as they are simple, reliable, cost effective,portable and targets can be detected with the naked eye. Since nativenucleic acids do not show absorption in the visible region, various colorreporting groups such as small organic dyes, conjugated polymers,enzymes coupled to a chromogen and metallic nanoparticles have beenused to develop colorimetric biosensors (Liu, 2009). Among these

groups, metallic (e.g., gold) nanoparticles functionalized with either aprobe DNA or an aptamer for capturing the target analyte(Kanjanawarut and Su, 2009; Elghanian et al., 1997) have beencommonly used to develop colorimetric biosensors. When target DNAcontains complementary sequences of the probes, it causes particles'aggregation through sandwich hybridization that can be detected ascolor change of the colloidal solution. Another popular approach usessalt induced aggregation of gold nanoparticles to visualize the targets(driven by the London-van der Waals attractive force) (Sato et al.,2003; Li and Rothberg, 2004). These nanoparticle aggregation basedassays have been used for naked eye detection of DNA methylation (Geet al., 2012; Lin and Chang, 2013). In this assay, anti 5mC antibodymodified with magnetic microspheres (MMPs) were selectively at-tached with the CpG region of APC (Ge et al., 2012). An APC specificprobe was then added which captured the MMPs-APC complex. Afterthe capture event, the hybridized complex was denatured by heating.The released capture probe was then mixed with unmodified goldnanoparticle (AuNP) and salt solutions for colorimetric detection. Inthis system, the presence of DNA methylation hinders the salt-inducedaggregation of AuNP. In contrast, unmethylated DNA facilitates theaggregation, leading to a quick color change from red to purple. Themethod was sensitive to detect 80 fM of target DNA. Lin and Chang(2013) have developed another colorimetric assay for DNA methylationdetection based on AuNPs aggregation. The applicability of the methodhas been tested to analyze methylation level in nasopharyngealcarcinoma cells lines.

To improve the sensitivity of the colorimetric assay, MBD proteinbased enrichment method can be adapted in methylation assay. Forexample, Wee and colleagues have developed a colorimetric assay forthe rapid detection of both global and regional methylation, where

Fig. 4. Bisulfite treated and asymmetric PCR-amplified adenine enriched (blue) or guanine-enriched (red) antisense DNA strands represent unmethylated and methylated templates,respectively. The adsorption pattern on SPR sensor was monitored in real-time and label-free manner using SPR readout. Reprinted from Sina et al. (2014) with permission from TheAmerican Chemical Society.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Md. Nazmul Islam et al. Biosensors and Bioelectronics 92 (2017) 668–678

676

enzymatically digested-genomic DNA was labeled with biotin (via a fill-in reaction with Klenow polymerase and biotin-dNTPs (Fig. 5) (Weeet al., 2015a). Magnetic beads functionalized-MBD were then selec-tively bound to the biotin conjugated methylated region. Next strepta-vidin coated horseradish peroxidase (SA-HRP) was used to recognizebiotinylated methylated region. The methylation level was then visuallyevaluated via HRP/H2O2 system coupled to 3,3′,5,5′-tetramethylben-zidine. It is evident that unlike many other optical readout methods,colorimetric assays require less sophisticated optical setup to detectDNA methylation which can make on-site detection of DNA methyla-tion more manageable.

4. Conclusion and perspectives

We have thoroughly reviewed the recent advances in opticaldetection strategies for DNA methylation. We have also addressedthe major methodological shortcomings of these strategies. It is nowevident that most of these strategies are not suitable for routine clinicalapplications as these methods typically rely on cumbersome chemicaltreatment (e.g., bisulfite conversion), amplification (e.g., PCR) andenzymatic digestion of target samples. We believe that the mostsuitable method for diagnostics or clinical applications should not betoo much technically complicated. In our opinion the method with theability of detecting DNA methylation directly with an inexpensivedetection tool will ultimately be the future of methylation basedepigenetic research.

In optical detection strategies, development in laser, filters andother optical set up of these techniques can contribute to the develop-ment of more suitable commercial methylation biosensors. The on-going research should try to resolve interference created from samplepopulation variation. Researchers should also exploit different optionsto improve multiplexing capability of DNA methylation. Differentsandwich model with various transduction steps can be designed inthe sensors to increase the target response. Moreover, innovativeapproaches should be established to remove false negative result andnonspecific binding. It is obvious that a considerable amount of workneeds to be performed to meet the remained challenges and to ensurefurther enhancement of the methylation biosensors for achievingdesired level of accuracy and sensitivity. In our opinion, what havebeen achieved so far will serve as the foundation for the development of

novel strategies for DNA methylation. We forefeel that in the nearfuture, systematic research will translate many of these proof-of-concept ideas into the suitable point-of-care DNA methylation biosen-sors.

Acknowledgement

This work was supported by the NHMRC CDF (APP1088966) andGriffith University New Researcher Grant Scheme to M.J.A.S., andhigher degree research scholarships (GUIPRS and GUPRS scholarshipsto. M.N.I., S.Y., M.H.H., A.M. and F.I.) from the Griffith University.

References

Aberg, K.A., McClay, J.L., Nerella, S., Xie, L.Y., Clark, S.L., Hudson, A.D., et al., 2012.Epigenomics 4 (6), 605–621.

Aggerholm, A., Guldberg, P., Hokland, M., Hokland, P., 1999. Cancer Res. 59 (2),436–441.

Armstrong, K.M., Bermingham, E.N., Bassett, S.A., Treloar, B.P., Roy, N.C., Barnett,M.P., 2011. Biotechnol. J. 6 (1), 113–117.

Bailey, V.J., Easwaran, H., Zhang, Y., Griffiths, E., Belinsky, S.A., Herman, J.G., Baylin,S.B., Carraway, H.E., Wang, T.H., 2009. Genome Res. 19 (8), 1455–1461.

Bailey, V.J., Keeley, B.P., Zhang, Y., Ho, Y.P., Easwaran, H., Brock, M.V., Pelosky, K.L.,Carraway, H.E., Baylin, S.B., Herman, J.G., Wang, T.H., 2010. ChemBioChem 11 (1),71–74.

Bird, A., 2002. Genes Dev. 16 (1), 6–21.Bird, A.P., 1978. J. Mol. Biol. 118 (1), 49–60.BonDurant, A.E., Huang, Z., Whitaker, R.S., Simel, L.R., Berchuck, A., Murphy, S.K.,

2011. Gynecol. Oncol. 123 (3), 581–587.Cao, A., Zhang, C., 2012. Anal. Chem. 84 (14), 6199–6205.Carrascosa, L.G., Sina, A.A.I., Palanisamy, R., Sepulveda, B., Otte, M.A., Rauf, S.,

Shiddiky, M.J.A., Trau, M., 2014. Chem. Commun. 50 (27), 3585–3588.Cedar, H., Solage, A., Glaser, G., Razin, A., 1979. Nucleic Acids Res., 6.Church, G.M., Gilbert, W., 1984. Proc. Natl. Acad. Sci. 81 (7), 1991–1995.Chen, Y., Wang, J., 2013. Anal. Biochem. 442 (2), 196–204.Clapp, A.R., Medintz, I.L., Mattoussi, H., 2006. ChemPhysChem 7 (1), 47–57.Constancia, M., Pickard, B., Kelsey, G., Reik, W., 1998. Genome Res. 8 (9), 881–900.Costello, J.F., Plass, C., Cavenee, W.K., 2002. Restriction landmark genome scanning. In:

Mills, K.I., Ramsahoye, B.H. (Eds.), DNA Methylation Protocols. Humana Press,Totowa, New Jersey, 53–70.

Cui, W., Wang, L., Jiang, W., 2016. Biosens. Bioelectron. 77, 650–655.Dadmehr, M., Hosseini, M., Hosseinkhani, S., Ganjali, M.R., Khoobi, M., Behzadi, H.,

Hamedani, M., Sheikhnejad, R., 2014. Biosens. Bioelectron. 60, 35–44.Dai, Z., Cai, T., Zhu, W., Gao, X., Zou, X., 2013. Chem. Commun. 49 (19), 1939–1941.Dietrich, D., Hasinger, O., Liebenberg, V., Field, J.K., Kristiansen, G., Soltermann, A.,

2012. Diagn. Mol. Pathol. 21, 93–104.Dietrich, D., Hasinger, O., Banez, L.L., Sun, L., van Leenders, G.J., Wheeler, T.M.,

Bangma, C.H., Wernert, N., Perner, S., Freedland, S.J., et al., 2013. J. Mol. Diagn.

Fig. 5. Assay schemes. (A) Strategy for total genomic methylation detection where magnetic beads functionalized with MBD specifically bound the target biotinylated DNA. Furtherinclusion of streptavidin coated SA-HRP caused the biotin- streptavidin interaction. Then methylation was analyzed by measuring the color intensity of HRP signal (B) Strategy forregional methylation detection followed same principle but used gene specific primer for detecting regional methylation. Reproduced from Wee et al. (2015a) with permission from TheClinical Epigenists Society.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Md. Nazmul Islam et al. Biosensors and Bioelectronics 92 (2017) 668–678

677

15, 270–279.Doering, W.E., Piotti, M.E., Natan, M.J., Freeman, R.G., 2007. Adv. Mater. 19 (20),

3100–3108.Doeven, E.H., Barbante, G.J., Hogan, C.F., Francis, P.S., 2015. ChemPlusChem 80,

456–470.Duan, X., Liu, L., Feng, F., Wang, S., 2010. Acc. Chem. Res. 43 (2), 260–270.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 (8), E32.Ehrlich, M., 2002. Oncogene 21 (35), 5400–5413.Elghanian, R., Storhoff, J.J., Mucic, R.C., Letsinger, R.L., Mirkin, C.A., 1997. Science 277

(5329), 1078–1081.Feng, F., Wang, H., Han, L., Wang, S., 2008. J. Am. Chem. Soc. 130 (34), 11338–11343.Freeman, R., Willner, I., 2012. Chem. Soc. Rev. 41 (10), 4067–4085.Friso, S., Choi, S.-W., Dolnikowski, G.G., Selhub, J., 2002. Anal. Chem. 74 (17),

4526–4531.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.Gao, L., Cheng, L., Zhou, J.N., Zhu, B.L., Lu, Z.H., 2005. Coll. Surf. B: Biointerfaces 40,

127–131.Ge, C., Fang, Z., Chen, J., Liu, J., Lu, X., Zeng, L., 2012. Analyst 137 (9), 2032–2035.Genereux, D.P., Johnson, W.C., Burden, A.F., Stöger, R., Laird, C.D., 2008. Nucleic Acids

Res. 36 (22), (e150-e150).Gonzalgo, M.L., Jones, P.A., 1997. Nucleic Acids Res. 25 (12), 2529–2531.Green, R.J., Frazier, R.A., Shakesheff, K.M., Davies, M.C., Roberts, C.J., Tendler, S.J.B.,

2000. Biomaterials 21 (18), 1823–1835.Gu, H., Bock, C., Mikkelsen, T.S., Jager, N., Smith, Z.D., Tomazou, E., Gnirke, A., Lander,

E.S., Meissner, A., 2010. Nat. Methods 7 (2), 133–136.Gu, H., Smith, Z.D., Bock, C., Boyle, P., Gnirke, A., Meissner, A., 2011. Nat. Protoc. 6 (4),

468–481.Harper, M.M., McKeating, K.S., Faulds, K., 2013. Phys. Chem. Chem. Phys. 15 (15),

5312–5328.Hatada, I., Hayashizaki, Y., Hirotsune, S., Komatsubara, H., Mukai, T., 1991. Proc. Natl.

Acad. Sci. 88 (21), 9523–9527.Herman, J.G., Graff, J.R., Myohanen, S., Nelkin, B.D., Baylin, S.B., 1996. Proc. Natl.

Acad. Sci. 93 (18), 9821–9826.Herman, J.G., Baylin, S.B., 2003. N. Engl. J. Med. 349, 2042–2054.Hernandez, Hernan G., et al., 2013. Biotechniques 55 (4), 181–197.Hibi, K., Goto, T., Shirahata, A., Saito, M., Kigawa, G., Nemoto, H., Sanada, Y., 2011.

Cancer Lett. 311 (1), 96–100.Hillenkamp, F., Karas, M., Beavis, R.C., Chait, B.T., 1991. Anal. Chem. 63 (24),

1193–1203.Hiraoka, D., Yoshida, W., Abe, K., Wakeda, H., Hata, K., Ikebukuro, K., 2012. Anal.

Chem. 84 (19), 8259–8264.Hirst, M., Marra, M.A., 2010. Brief. Funct. Genom. 9 (5–6), 455–465.Homola, J., 2008. Chem. Rev. 108 (2), 462–493.Hotchkiss, R.D., 1948. J. Biol. Chem. 175, 315–330.Hu, C.-W., Lee, H., Chen, J.-L., Li, Y.-J., Chao, M.-R., 2013. Anal. Bioanal. Chem. 405

(27), 8859–8869.Hu, J., Zhang, C.Y., 2012. Biosens. Bioelectron. 31 (1), 451–457.Humeny, A., Beck, C., Becker, C.M., Jeltsch, A., 2003. Anal. Biochem. 313 (1), 160–166.Jones, P.A., Gonzalgo, M.L., 1997. Proc. Natl. Acad. Sci. 94 (6), 2103–2105.Kanjanawarut, R., Su, X., 2009. Anal. Chem. 81 (15), 6122–6129.Keeley, B., Yi, Z., Ye, Z., Stark, A., Wang, T.H., 2012. In: Quantum dot FRET linker

probes for highly sensitive DNA methylation detection. In: Proceedings of the 12thIEEE Conference on Nanotechnology: 20–23 Aug 2012, Brihimhham (UnitedKingdom), pp. 1–4.

Kim, S., Hwang, S.H., Im, S.G., Lee, M.K., Lee, C.H., Son, S.J., Oh, H.B., 2016. Sensors16 (8), e1259.

Koo, K.M., Sina, A.A.I., Carrascosa, L.G., Shiddiky, M.J.A., Trau, M., 2015. Anal.Methods 7 (17), 7042–7054.

Koo, K.M., Sina, A.A.I., Carrascosa, L.G., Shiddiky, M.J.A., Trau, M., 2014. Analyst 139(23), 6178–6184.

Kristensen, L.S., Mikeska, T., Krypuy, M., Dobrovic, A., 2008. Nucleic Acids Res. 36 (7),e42.

Kulis, M., Esteller, M., 2010. In: DNA Methylation and Cancer 70. Academic Press Inc.Elsevier Science, San Diego, California, 27–56.

Kurdyukov, S., Bullock, M., 2016. Biology 5 (1), 3.Kurita, R., Arai, K., Nakamoto, K., Kato, D., Niwa, O., 2012. Anal. Chem. 84 (4),

1799–1803.Kurita, R., Yanagisawa, H., Yoshioka, K., Niwa, O., 2015. Anal. Chem. 87 (22),

11581–11586.Laing, S., Gracie, K., Faulds, K., 2016. Chem. Soc. Rev. 45 (7), 1901–1918.Le, T., Kim, K.-P., Fan, G., Faull, K.F., 2011. Anal. Biochem. 412 (2), 203–209.Li, E., Beard, C., Jaenisch, R., 1993. Nature 366 (6453), 362–365.Li, H., Rothberg, L., 2004. Proc. Natl. Acad. Sci. 101 (39), 14036–14039.Li, N., Ye, M., Li, Y., Yan, Z., Butcher, L.M., Sun, J., Han, X., Chen, Q., Zhang, X., Wang,

J., 2010. Methods 52 (3), 203–212.Li, Y., Huang, C., Zheng, J., Qi, H., 2012. Biosens. Bioelectron. 38 (1), 407–410.Li, X., Song, T., Guo, X., 2015. Analyst 140 (18), 6230–6233.Lin, Y.-Z., Chang, P.-L., 2013. ACS Appl. Mater. Interfaces 5 (22), 12045–12051.Lisanti, S., Omar, W.A., Tomaszewski, B., De Prins, S., Jacobs, G., Koppen, G., Mathers,

J.C., Langie, S.A., 2013. PLoS One 8 (11), e79044.Liu, X., Shi, L., Niu, W., Li, H., Xu, G., 2007. Angew. Chem. Int. Ed. 46 (3), 421–424.Liu, J., Cao, Z., Lu, Y., 2009. Chem. Rev. 109 (5), 1948–1998.Liu, W., Lai, H., Huang, R., Zhao, C., Wang, Y., Weng, X., Zhou, X., 2015a. Biosens.

Bioelectron. 68, 736–740.Liu, Z., Qi, W., Xu, G., 2015b. Chem. Soc. Rev. 44 (10), 3117–3142.Lizardi, P.M., Huang, X., Zhu, Z., Bray-Ward, P., Thomas, D.C., Ward, D.C., 1998. Nat.

Genet. 19 (3), 225–232.Ma, Y., Zhang, H., Liu, F., Wu, Z., Lu, S., Jin, Q., Zhao, J., Zhong, X., Mao, H., 2015.

Nanoscale 7 (41), 17547–17555.Ma, Y., Bai, Y., Mao, H., Hong, Q., Yang, D., Zhang, H., et al., 2016. Biosens. Bioelectron.

85, 641–648.McDonnell, J.M., 2001. Curr. Opin. Chem. Biol. 5 (5), 572–577.Meissner, A., Gnirke, A., Bell, G.W., Ramsahoye, B., Lander, E.S., Jaenisch, R., 2005.

Nucleic Acids Res. 33 (18), 5868–5877.Miyawaki, A., 2003. Dev. Cell 4 (3), 295–305.Mohandas, T., Sparkes, R.S., Shapiro, L.J., 1981. Science 211 (4480), 393–396.Nimmrich, I., Sieuwerts, A.M., Meijer-van Gelder, M.E., Schwope, I., Bolt-de Vries, J.,

Harbeck, N., Koenig, T., Hartmann, O., Kluth, A., Dietrich, D., et al., 2008. BreastCancer Res. Treat. 111, 429–437.

Pakneshan, S., Salajegheh, A., Smith, R.A., Lam, A.K., 2013. Pathology 45 (4), 346–356.Palanisamy, R., Connolly, A.R., Trau, M., 2011. Anal. Chem. 83 (7), 2631–2637.Pan, S., Xu, J., Shu, Y., Wang, F., Xia, W., Ding, Q., Xu, T., Zhao, C., Zhang, M., Huang,

P., Lu, S., 2010. Biosens. Bioelectron. 26 (2), 850–853.Ramsahoye, B.H., 2002. Methods 27 (2), 156–161.Reik, W., Dean, W., 2001. Electrophoresis 22, 2838–2843.Resch-Genger, U., Grabolle, M., Cavaliere-Jaricot, S., Nitschke, R., Nann, T., 2008. Nat.

Methods 5 (9), 763–775.Robertson, K.D., Jones, P.A., 2000. Carcinogenesis 21 (3), 461–467.Rush, L.J., Plass, C., 2002. Anal. Biochem. 307 (2), 191–201.Saluz, H., Jost, J.P., 1989. Proc. Natl. Acad. Sci. USA 86, 2602–2606.Sato, K., Hosokawa, K., Maeda, M., 2003. J. Am. Chem. Soc. 125 (27), 8102–8103.Schlucker, S., 2009. ChemPhysChem 10 (9–10), 1344–1354.Seguin-Orlando, A., Gamba, C., Sarkissian, C.D., Ermini, L., Louvel, G., Boulygina, E.,

Sokolov, A., et al., 2015. Sci. Rep. 5, 11826.Shiraishi, M., Sekiguchi, A., Oates, A.J., Terry, M.J., Miyamoto, Y., Sekiya, T., 2004. Anal.

Biochem. 329 (1), 1–10.Sina, A.A.I., Carrascosa, L.G., Palanisamy, R., Rauf, S., Shiddiky, M.J.A., Trau, M., 2014.

Anal. Chem. 86 (20), 10179–10185.Singer-Sam, J., LeBon, J.M., Tanguay, R.L., Riggs, A.D., 1990. Nucleic Acids Res. 18 (3),

687.So, M.Y., Tian, Z., Phoon, Y.S., Sha, S., Antoniou, M.N., Zhang, J., Wu, R.S.S., Tan-Un,

K.C., 2014. PLoS One 9 (5), e96404.Stach, D., Schmitz, O.J., Stilgenbauer, S., Benner, A., Döhner, H., Wiessler, M., Lyko, F.,

2003. Nucleic Acids Res. 31 (2), e2.Stiles, P.L., Dieringer, J.A., Shah, N.C., Van Duyne, R.P., 2008. Annu. Rev. Anal. Chem.

(1), 601–626.Taleat, Z., Mathwig, K., Sudhölter, E.J.R., Rassaei, L., 2015. TrAC Trends Anal. Chem.

66, 80–89.Tang, X., Khuri, F.R., Lee, J.J., Kemp, B.L., Liu, D., Hong, W.K., Mao, L., 2000. J. Natl.

Cancer Inst. 92, 1511–1516.Taylor, K.H., Kramer, R.S., Davis, J.W., Guo, J., Duff, D.J., Xu, D., Caldwell, C.W., Shi,

H., 2007. Cancer Res. 67 (18), 8511–8518.Tost, J., Schatz, P., Schuster, M., Berlin, K., Gut, I.G., 2003. Nucleic Acids Res. 31 (9),

e50.Tost, J., Gut, I.G., 2007. Nat. Protoc. 2 (9), 2265–2275.Tucker, K.L., 2001. Neuron 30 (3), 649–652.Wang, Y., Xu, H., Zhang, J., Li, G., 2008. Sensors 8 (4), 2043–2081.Wang, Y., Wee, E.J., Trau, M., 2015. Chem. Commun. 51 (54), 10953–10956.Wang, Y., Wee, E.J.H., Trau, M., 2016. Chem. Commun. 52 (17), 3560–3563.Warnecke, P.M., Stirzaker, C., Melki, J.R., Millar, D.S., Paul, C.L., Clark, S.J., 1997.

Nucleic Acids Res. 25 (21), 4422–4426.Weber, M., Davies, J.J., Wittig, D., Oakeley, E.J., Haase, M., Lam, W.L., Schubeler, D.,

2005. Nat. Genet. 37 (8), 853–862.Wee, E.J., Ngo, T.H., Trau, M., 2015a. Clin. Epigenet. 7 (1), 65.Wee, E.J.H., Ha Ngo, T., Trau, M., 2015b. Sci. Rep. 5, 15028.Wei, H., Wang, E., 2011. Luminescence 26 (2), 77–85.Wei, W., Gao, C., Xiong, Y., Zhang, Y., Liu, S., Pu, Y., 2015. Talanta 131, 342–347.Wojdacz, T.K., Dobrovic, A., 2007. Nucleic Acids Res. 35 (6), e41.Worm, J., Aggerholm, A., Guldberg, P., 2001. Clin. Chem. 47 (7), 1183–1189.Wu, M., Wang, X., Wang, K., Guo, Z., 2016. Chem. Commun. 52, 8377–8380.Wyatt, G.R., 1951. J. Biochem. 48, 581–584.Xiong, Z., Laird, P.W., 1997. Nucleic Acids Res. 25 (12), 2532–2534.Xue, Q., Lv, Y., Xu, S., Zhang, Y., Wang, L., Li, R., Yue, Q., Li, H., Gu, X., Zhang, S., Liu,

J., 2015. Biosens. Bioelectron. 66, 547–553.Yan, P.S., Chen, C.M., Shi, H., Rahmatpanah, F., Wei, S.H., Caldwell, C.W., Huang,

T.H.M., 2001. Cancer Res. 61 (23), 8375–8380.Yang, Q., Dong, Y., Wu, W., Zhu, C., Chong, H., Lu, J., Yu, D., Liu, L., Lv, F., Wang, S.,

2012. Nat. Commun. 3, 1206.Yegnasubramanian, S., Lin, X., Haffner, M.C., DeMarzo, A.M., Nelson, W.G., 2006.

Nucleic Acids Res. 34 (3), (e19-e19).Yu, Y., Blair, S., Gillespie, D., Jensen, R., Myszka, D., Badran, A.H., et al., 2010. Anal.

Chem. 82 (12), 5012–5019.Zhang, L., Xu, Y.-Z., Xiao, X.-F., Chen, J., Zhou, X.-Q., Zhu, W.-Y., Dai, Z., Zou, X.-Y.,

2015. TrAC Trends Anal. Chem. 72, 114–122.Zhao, H.F., Liang, R.P., Wang, J.W., Qiu, J.D., 2015. Biosens. Bioelectron. 63, 458–464.Zou, H.Z., Yu, B.M., Wang, Z.W., Sun, J.Y., Cang, H., Gao, F., Li, D.H., Zhao, R., Feng,

G.G., Yi, J., 2002. Clin. Cancer Res. 8, 188–191.

Md. Nazmul Islam et al. Biosensors and Bioelectronics 92 (2017) 668–678

678