electrochemical and piezoelectric dna biosensors for hybridisation detection

R

Eh

FMa

b

a

A

R

R

1

A

P

K

D

G

A

H

E

P

C

0d

a n a l y t i c a c h i m i c a a c t a 6 0 9 ( 2 0 0 8 ) 139–159

avai lab le at www.sc iencedi rec t .com

journa l homepage: www.e lsev ier .com/ locate /aca

eview

lectrochemical and piezoelectric DNA biosensors forybridisation detection

austo Lucarelli a, Sara Tombelli a, Maria Minunnia, Giovanna Marrazzaa,arco Mascinia,b,∗

Department of Chemistry, University of Florence, 50019 Sesto Fiorentino, Florence, ItalyINBB, Unit of Florence, Italy

r t i c l e i n f o

rticle history:

eceived 15 October 2007

eceived in revised form

9 December 2007

ccepted 21 December 2007

ublished on line 8 January 2008

eywords:

NA biosensor

enosensor

mplicon

ybridisation

lectrochemical

iezoelectric transduction

a b s t r a c t

DNA biosensors (or genosensors) are analytical devices that result from the integration of a

sequence-specific probe and a signal transducer. Among other techniques, electrochemical

and piezoelectric methods have recently emerged as the most attractive due to their

simplicity, low instrumentation costs, possibility for real-time and label-free detection and

generally high sensitivity.

Focusing on the most recent activity of worldwide researchers, the aim of the present

review is to give the readers a critical overview of some important aspects that contribute

in creating successful genosensing devices. Advantages and disadvantages of different

sensing materials, probe immobilisation chemistries, hybridisation conditions, transducing

principles and amplification strategies will be discussed in detail. Dedicated sections will

also address the issues of probe design and real samples pre-treatment. Special emphasis

will be finally given to those protocols that, being implemented into an array format, are

already penetrating the molecular diagnostics market.

© 2008 Elsevier B.V. All rights reserved.

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1402. Probe design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

2.1. Linear oligonucleotide probes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1412.2. Hairpin oligonucleotide probes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1412.3. Peptide nucleic acids (PNAs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1422.4. Locked nucleic acids (LNAs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142

3. Immobilisation methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1423.1. Carbon-based electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

3.1.1. Direct physisorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1433.1.2. Attachment of biotinylated probes to avidin-coated surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1433.1.3. Attachment to an electropolymerised layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1433.1.4. Self-assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

∗ Corresponding author. Tel.: +39 055 4573283; fax: +39 055 4573384.E-mail address: [email protected] (M. Mascini).

003-2670/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.aca.2007.12.035

mailto:[email protected]

dx.doi.org/10.1016/j.aca.2007.12.035

140 a n a l y t i c a c h i m i c a a c t a 6 0 9 ( 2 0 0 8 ) 139–159

3.2. ITO electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1443.2.1. Covalent attachment to a self-assembled monolayer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1443.2.2. Attachment by electro-copolymerisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

3.3. Silicon oxide electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1443.4. Platinum electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1443.5. Gold electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

3.5.1. Electrochemical sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1453.5.2. Piezoelectric sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

4. Hybridisation reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1474.1. Sample pre-treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

4.1.1. Genomic DNA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1474.1.2. PCR-amplified samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

4.2. Passive vs. active hybridisation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1485. Hybridisation detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

5.1. Reagentless methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1495.1.1. Fully electronic detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1495.1.2. Electronic beacons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1495.1.3. Polypyrrole electrochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505.1.4. Alternative approaches in QCM sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

5.2. Label-free methods (solution-phase metal complexes and organic dyes). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505.3. Label-based methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1525.4. Signal amplification. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152

5.4.1. Nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1525.4.2. Enzymes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1525.4.3. Enzyme multilabelling (dendritic-like architectures). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1535.4.4. Ferrocene–streptavidin conjugates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1535.4.5. Polymeric osmium complex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1545.4.6. Oligonucleotide-loaded gold nanoparticles and [Ru(NH3)6]3+ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1545.4.7. Enhancement of gold nanoparticles by electrocatalytic deposition of silver. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1545.4.8. Bio-metallization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1545.4.9. Streptavidin-conjugated nanoparticles for amplified piezoelectric sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

5.5. Commercially available tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1545.5.1. Electrocatalytic oxidation of guanine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1545.5.2. DNA-mediated charge transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1555.5.3. Redox cycling the product of an enzymatic reaction (eMicroLISA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1565.5.4. Minor groove binder (GenelyzerTM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1565.5.5. Ferrocene-labelled signalling probes (CMS eSensorTM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1565.5.6. HRP enzyme label (CombiMatrix ElectraSenseTM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

6. Conclusions and future prospects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

1. Introduction

DNA biosensors (or genosensors) are analytical devices thatresult from the integration of a sequence-specific probe (usu-ally a short synthetic oligonucleotide) and a signal transducer.The probe, immobilised onto the transducer surface, actsas the biorecognition molecule and recognises the targetDNA, while the transducer is the component that convertsthe biorecognition event into a measurable signal. Assem-bly of numerous (up to a few thousand) DNA biosensorsonto the same detection platform results in DNA microar-rays (or DNA chips), devices which are increasingly used

are in development whose promise is to provide flexible andeconomical alternatives for applications that require relativelyfewer measurements. Among other techniques, electrochem-ical and piezoelectric transductions are the most attractivedue to their simplicity, low instrumentation costs, possibil-ity for real-time and label-free detection and generally highsensitivity.

In the past few years, several excellent reviews have beenpublished on both electrochemical and piezoelectric DNAsensing [1–7]. Focusing on the most recent activity of world-wide researchers, the aim of the present review is to givethe readers a critical overview of those important aspects

for large-scale transcriptional profiling and single-nucleotidepolymorphisms (SNPs) discovery.

As clinical diagnostics and other applications (e.g., envi-ronmental screening) do not generally need the massive dataaccumulation typical of gene chips, alternative technologies

that contribute in creating successful genosensing devices.
For the sake of clarity, this review will specifically focuson devices in which the biological recognition elements aresurface-immobilised, and therefore in intimate contact withthe transducer. Hence, the whole branch of literature recently

t a 6

pc

rtbicte

2

Adotpa

ei(q

esptatssc

fsb

Fc

a n a l y t i c a c h i m i c a a c

ublished on (electrochemical) geno-magnetic sensing con-epts will be deliberately ignored.

Advantages and disadvantages of different sensing mate-ials, probe immobilisation chemistries, hybridisation condi-ions, transducing principles and amplification strategies wille discussed in detail. Dedicated sections will also address the

ssues of probe design and real samples pre-treatment. Spe-ial emphasis will be finally given to electrochemical protocolshat, being implemented into an array format, are already pen-trating the molecular diagnostics market.

. Probe design

s the specificity of the hybridisation reaction is essentiallyependent on the biorecognition properties of the captureligonucleotide, design of the capture probe is undoubtedlyhe most important pre-analytical step. Thus, a number ofrobes, variable for chemical composition and conformationalrrangement, have been used to assemble DNA biosensors.

Most commonly, the probes are linear oligonucleotides,ither synthesised in situ or pre-synthesised and afterwardsmmobilised onto the sensor surface. However, structuredhairpin) oligonucleotides are being used with increasing fre-uency.

The experimental variables affecting the hybridisationvent at the transducer–solution interface are referred to astringency. Such variables typically include hybridisation andost-hybridisation-washing buffers composition and reactionemperature. When dealing with more than a single probet the time, the basic requirement for a functional system ishe ability of all the different probes to hybridise their targetequences with high affinity and specificity under the sametringency conditions [8]. This aspect makes the design ofomplex sets of probes even more difficult.

The design of probes for the analysis of samples susceptibleor degradation (such as RNA) requires additional attention if aandwich hybridisation scheme is chosen. Selection of probesinding sites in close proximity to each other results partic-

ig. 1 – eBeacon (ferrocene-labelled hairpin probe): (A) before hybonformation).

0 9 ( 2 0 0 8 ) 139–159 141

ularly convenient [9], as it minimises the adverse effects ofsample degradation on the success of the sandwich hybridis-ation reaction.

2.1. Linear oligonucleotide probes

Design of linear probes takes now great advantage of decadesof experience, which has led to many commercially availablesoftware. A particular challenge is, however, still representedby design of full sets of arrayed probes for the screeningidentification of closely related and unrelated pathogens [10].Upon retrieval of the genomic sequences from specific databanks and their assembly and alignment using dedicated soft-ware, PCR primers and reporter oligonucleotides are typicallyselected within highly conserved regions of each bacterialgenome. According to the requirements of each specific appli-cation, capture oligonucleotides can be then designed withineither hypervariable or highly conserved regions. Candidatesequences are finally tested for theoretical melting tempera-ture (Tm), hairpins and dimers formation and for homologiesusing a Basic Local Alignment Search Tool (BLAST) search.

Use of capture probes in the order of 18–25 nucleotidesusually confers higher levels of specificity to the hybridisa-tion reaction. However, excessively longer capture oligonu-cleotides often exhibit particularly unfavourable hybridisationspecificity and yields. While a single or a few mismatches areunlikely to significantly destabilise even a 30-mer probe–targetduplex over a wide range of experimental conditions [11], thegeneral hybridisation efficiency of similar or longer probesmight be particularly low, due to intramolecular hydrogenbonding and consequent formation of non-reactive hairpinstructures.

2.2. Hairpin oligonucleotide probes

Hairpin probes are structured oligonucleotides that combine adouble-stranded stem region (due to intramolecular base pair-ing) and a single-stranded loop, which contains the capturesequence (e.g., Fig. 1A). Similarly to molecular beacons, the

ridisation (folded state); (B) after hybridisation (unfolded

a c t a
142 a n a l y t i c a c h i m i c a
hairpin probes used for electrochemical sensing were func-tionalised at both 3′- and 5′-end. One of these functionalities(e.g., an alkylthiol spacer arm) allowed for surface immobil-isation of the biomolecule, while the other consisted of aredox active compound (e.g., a ferrocene derivative). In theabsence of a target, the stem–loop configuration hold theredox moiety in proximity to the electrode. Hence, because ofthe short distance and easy electron transfer, the label gener-ated large faradic currents. The conformational change whichfollowed hybridisation displaced the label from the electrodesurface, resulting in strong diminution of the electrochemicalaccessibility of the redox moiety. The extent of signal suppres-sion thus reflected presence and concentration of the targetsequence. The major advantage of using redox labelled hair-pin probes was that hybridisation could be measured directly,obviating the need for labelling the target through additionaland time-consuming steps and also avoiding use of furtherreagents.

Proper design of the hairpin probes is obviously crucial,as functionality, selectivity and sensitivity of such captureoligonucleotides are reported to strongly depend on the ampli-tude of the loop and the length of the stem region [12,13].In particular, the hairpin probe used by Miranda-Castro et al.exhibited superior selectivity with respect to analogous linearoligonucleotides. Such an improved discrimination capabilitywas attributed to the loop–stem structure of the probe, whichstabilised the dissociated state of the probe–analyte duplexespecially in the presence of mismatched base pairs [13]. How-ever, other authors [12,14] showed different hairpin probes toyield non-specific hybridisation with a number of sequenceswhich presented a certain degree of homology with their tar-get. These results raised the question of whether the hairpinsare truly advantageous in terms of selectivity. Also, most ofthe electrochemical studies were limited to the use of rela-tively short synthetic oligonucleotide targets (<52 bases), thuslacking the true challenge of real sample analysis.

2.3. Peptide nucleic acids (PNAs)

PNAs are DNA mimics in which the nucleobases are attachedto a neutral N-(2-aminoethyl)-glycine pseudopeptide back-bone (Fig. 2B). Following the Watson–Crick rules, PNA probes
form hybrids stabilised by hydrogen bonding and base stack-ing. However, because of the neutral backbone of PNA,PNA/DNA hybrids display melting temperatures higher thanthe corresponding DNA/DNA duplexes (generally +1 ◦C bp−1),
Fig. 2 – DNA (A), PNA (B) an

6 0 9 ( 2 0 0 8 ) 139–159

stability against nucleases and proteases, relative insensitivityto ionic strength and usually higher selectivity against single-base mismatches [15].

If compared to the conventional oligonucleotide probes,PNAs have appeared particularly interesting for the devel-opment of electrochemical genosensing concepts, the mainreason being the drastically different electrical characteristicsof their molecular backbone.

The supporters of PNAs claim that these probes are able tobind their complementary sequence with comparably higheraffinity and specificity than usual oligonucleotide probes,while discriminating mismatched targets to a larger extent.However, despite careful control of the hybridisation condi-tions, different authors observed relatively strong non-specificsignals for a number of SNP-containing targets [16,17].

An example of how the selectivity of PNA probes can befurther improved is offered by the work of Marchelli’s group,which exploited the effect exerted by the introduction of chiralaminoacids into the PNA backbone [18]. A group of three adja-cent d-lysine chiral monomers (referred to as “chiral box”) wasintroduced in the middle of a 15-mer PNA specific for the cysticfibrosis W1282X point mutation. When compared to analo-gous achiral PNAs or oligonucleotides, such a modified probeexhibited enhanced mismatch recognition capability, allowingeasy differentiation of healthy, heterozygous and homozygousmutant patients.

2.4. Locked nucleic acids (LNAs)

Locked nucleic acids (LNATM, www.exiqon.com) are a class ofnucleic acid analogues in which the ribose ring is “locked”by a methylene bridge connecting the 2′-O atom with the4′-C atom (Fig. 2C). DNA oligos incorporating LNA nucleo-sides show increased thermal stability (+2 to 6 ◦C/includedmonomer) and further improved discriminative power withrespect to single-base mismatched targets (�Tm up to 8 ◦C). Tothese authors’ knowledge, LNA probes have not, yet, been usedfor electrochemical or piezoelectric sensing of DNA. However,the outstanding properties of such DNA mimics are likely tofind massive utilisation for the future development of SNPstyping assays.

3. Immobilisation methods

The ability to immobilise the probe in a predictable mannerwhile maintaining its inherent affinity for the target DNA is

d LNA (C) monomers.

http://www.exiqon.com/

t a 6

cdpccsmftiaabopciebiim

tttbepnafsu

peiAtarab

3

Dgsldbbfipbbo


rucial to the overall device performance. Hence, indepen-ently on the molecular identity and conformation of therobe, some general aspects must be taken into account whenonsidering its surface attachment. Being obvious that thehoice of the most appropriate immobilisation protocol istrictly dependent on the characteristics of the transducingaterial, robust immobilisation chemistries are usually pre-

erred, in order to prevent desorption of the probes fromhe sensing layer. As it will be illustrated below, retentionn a polymeric matrix, covalent attachment on a function-lised support, affinity immobilisation and self-assemblingre, to date, the most successful approaches. This is mainlyecause each of these immobilisation strategies can lead tordered sets of end-point attached and properly orientedrobes. Moreover, such chemistries also allow to control theonformational freedom of the probes and the correspondingnterchain space through the modulation of the surface cov-rage. Hybridisation efficiencies as high as 100% have eveneen reported in the most favourable cases, thus demonstrat-

ng that the resulting bio-architectures are fully accessible fornteraction with the solution-phase target, while offering only

inimal steric impediments.While the electrodes applied onto piezoelectric crystals are

ypically made of gold (with platinum being a rare exception),he availability of a range of conducting materials have madehe family of electrochemical genosensors and related immo-ilisation chemistries much broader. Besides classical bulkylectrodes, detection platforms that are suitable for massroduction, miniaturisation and possibly allow for simulta-eous multi-site immobilisation and detection are the mostttractive. In fact, such devices combine the capability toully characterise a given sample in a single assay (e.g., whencreening for several single-nucleotide polymorphisms) withse of minimal sample volumes and short analysis times.

Functionalisation of the sensing interface with the capturerobe must be done following two basic criteria. As a gen-ral rule, the probe has to be immobilised in such a way thatts biorecognition capability is preserved as much as possible.dditionally, as a particular need of electrochemical detec-

ion schemes, the attached bio-layer does not have to behaves a total insulator, thus allowing for electrochemical inter-ogation of the surface. The most interesting immobilisationpproaches, specific for each electrodic material, are describedelow.

.1. Carbon-based electrodes

uring the past decade, carbon electrodes have experiencedreat popularity, mainly because of favourable characteristicsuch as suitability for mass production (via screen-printing),ow cost, wide electrochemically accessible potential win-ow, etc. Such a popularity has been renewed and reinforcedy the advent of carbon nanotubes [19]. However, carbon-ased transducers are unlikely to play, at least in the nearuture, an important role in DNA diagnostic devices. Thiss because on bare carbon surfaces immobilisation of the
robe usually proceeds randomly, via multiple interactionsetween the transducer surface and both the phosphateackbone and the hydrophobic nitrogenous bases of theligonucleotides. Random adsorption probably accounts for a
0 9 ( 2 0 0 8 ) 139–159 143

significant fraction of the surface-confined strands also whenattempting carbodiimide/N-hydroxy-succinimide (EDC/NHS)covalent coupling of derivatised oligos to activated surfaces.Besides being non-oriented, physisorbed molecules are lim-ited in their mobility, which makes the hybridisation eventsterically inhibited. Efficient blockage of the surface (e.g., toprevent the non-specific adsorption of DNA sequences, redoxlabels, etc.) can additionally be non-trivial.

Along with an example of immobilisation by simpleadsorption, the most relevant strategies developed to circum-vent the drawbacks of carbon surfaces are described below.The one exploited by Gorodetsky and Barton [20] appears,between the others, of particular interest.

3.1.1. Direct physisorptionKerman et al. [15] choose physisorption as a rapid and sim-ple method to modify the surface of an electrochemicallypre-treated glassy carbon electrode (GCE) with PNA probes.However, confirming the validity of the above-mentionedconcerns, the non-specific signal from the Co(NH3)63+ labelwas difficult to be minimised. Significant signals were alsoobtained only for micromolar concentrations of the tar-get, which reflected the low hybridisation efficiency of thephysisorbed PNA molecules.

3.1.2. Attachment of biotinylated probes to avidin-coatedsurfacesMetfies et al. [9] used a layer of neutravidin adsorbed ontoa carbon paste electrode as the platform for immobilisingbiotin-labelled probes. Even prior to attach such captureoligonucleotide, the electrode surface was blocked usingcasein, which effectively minimised the non-specific adsorp-tion of the components of the bioassay.

Similarly, Hernandez-Santos et al. modified the surface ofdisposable screen-printed carbon electrodes with streptavidinprior to immobilisation of the biotinylated capture sequence.Blocking of the surface with BSA was also found important inorder to reduce non-specific interactions [21].

3.1.3. Attachment to an electropolymerised layerTan and co-workers [22] used a monolayer of 4-aminobenzoicacid (4-ABA) electropolymerised on the surface of a glassycarbon electrode for the covalent immobilisation of an amino-modified capture oligonucleotide. The capture probe wascovalently attached to the carboxylic functionalities of the4-ABA monolayer via EDC/NHS cross-linking reaction. Voltam-metric transduction of the hybridisation events (accomplishedin the presence of methylene blue as the redox label) revealed,however, modest performances for such a biorecognitionlayer.

3.1.4. Self-assemblyGorodetsky and Barton recently demonstrated that pyrene-functionalised oligonucleotides can be conveniently self-assembled onto highly oriented pyrolytic graphite electrodes(Fig. 3) [20]. Characterisation of these films evidenced the fea-
tures previously observed for thiol-tethered oligos on gold sur-faces: closely packed oligonucleotides with helices oriented ina nearly upright conformation. This technique might, there-fore, provide the oligonucleotides immobilised on carbon sur-

144 a n a l y t i c a c h i m i c a a c t a

Fig. 3 – Pyrene-functionalised oligonucleotide immobilised
(self-assembled) onto a highly oriented pyrolytic graphiteelectrode. Adapted from Ref. [20].
faces of unprecedented biorecognition efficiencies, thus givinga renewed impulse to the use of carbon-based transducers.

3.2. ITO electrodes

Tin-doped indium oxide (ITO) is the material of choice forfabrication of a number of optoelectronic devices. However,ITO electrodes have also found application for DNA sequence-specific detection due to their particular amenability forelectro-catalytic oxidation of guanine residues [23].

3.2.1. Covalent attachment to a self-assembled monolayerWhile earlier papers described the use of randomly adsorbedoligo- or poly-nucleotides sequences, ordered and orientedimmobilisation of specific probes was subsequently demon-strated for these metal-oxide surfaces. According to themethod of Popovich et al., the oligonucleotide probes werebound to an ITO electrode by firstly forming a self-assembledmonolayer (SAM) of 12-phosphonododecanoic acid [24].The terminal carboxylic moiety of the monolayer was thenactivated with carbodiimide and reacted with an amine-terminated oligonucleotide probe. The immobilisation ofpreformed PNA–silane conjugates was also demonstrated inthe same paper.

3.2.2. Attachment by electro-copolymerisationAn array of individually addressable ITO electrodes was usedas the transduction element in an integrated analytical deviceused for the multiplexed detection of E. coli and B. subtilis cells[25]. The choice of an electrochemically driven immobilisa-tion strategy (electrochemical copolymerization of pyrrole and
pyrrole–oligonucleotide conjugates) allowed individual posi-tioning of the predefined probes at selected ITO surfaces.Notably, such immobilisation chemistry provided the probessufficient stability to tolerate the repeated thermal cycling ofin situ PCR amplification of the target DNA.
6 0 9 ( 2 0 0 8 ) 139–159

3.3. Silicon oxide electrodes

Contrasting with the widespread use of glass in optical sens-ing, silicon-based electrodes have found, up to now, limitedapplication in electrochemical genosensors.

Ingebrandt et al. [26] used, for example, an ultra-thin silicon dioxide layer as the dielectric gate in a fieldeffect transistors (FETs) DNA microarray (eight transistorgates/chip). The bio-modification of the SiO2 gates involvedprior cleaning and activation followed by silanisation with3-aminopropyltriethoxysilane (APTES). The amino-silanisedSiO2 surfaces were then modified using succinic anhydride,which allowed cross-linking of amino-modified oligonu-cleotide probes.

Gao et al. employed a related approach for modifying arraysof highly ordered n-type silicon nanowires (SiNW), fabricatedusing a complementary metal-oxide semiconductor (CMOS)compatible technology [27]. The arrays (100 wires/array, witheach wire being 5–50 nm wide, several micrometer long) werefirst cleaned with HF, silanised using trimethyloxysilane alde-hyde and then interacted with an amino-modified PNA probe.The non-reacted aldehyde moieties were blocked using buty-lamine and, finally, imine bonds were reduced using sodiumborohydride.

3.4. Platinum electrodes

Until recently, platinum electrodes have only found limitedapplication in DNA sensing technologies. One rare exam-ple was the work of Guiseppi-Elie and Hang, which used3-glycidoxypropyl-trimethoxysilane as a coupling reagent toimmobilise 5′-amine-terminated oligonucleotides onto theoxidized platinum electrodes of a quartz crystal microbalance(QCM) and the exposed glass surface between the Pt digits ofmicrofabricated electrodes [28].

Contrasting with this rather unpopular past, arrays ofplatinum microelectrodes are the heart of a breakthroughtechnology that is likely to revolutionise the current equi-librium between optical and electrochemical detection inmolecular diagnostic devices [29]. A general limitation forelectrochemical detection has been the technical difficultyto obtain an output from an array of electrodes, due towiring complexity and requirement for a multichannel reader.Overcoming such a fundamental drawback, the core tech-nology developed by CombiMatrix was based on a speciallymodified CMOS semiconductor containing 12,544 individuallyaddressable platinum microelectrodes (44 �m in diameter).Outstandingly, the semiconductor logic circuitry even allowedto direct the in situ synthesis of different oligonucleotidesat thousands of electrodes simultaneously. The probes weregrown within a proprietary “porous reaction layer” (lowblockage membrane) that coated each microelectrode within1 �m of the surface. Upon digital activation, each featureon the array selectively generated acid by means of anelectrochemical reaction. The generated acid, in turn, con-trolled detritylation of the growing oligonucleotide chain, thus
activating it for binding of the next nucleotide. Since a dif-ferent probe can be synthesized at each microelectrode, thistechnology enabled design of microarrays of any desired con-figuration.

t a 6

hcrtmsoa

3

Iuaef

Spttnebt“ponrweAtrdosp

Fr


The massive availability of test sites allowed synthesis ofundreds of unique 35–40-mer probes (each in tens of repli-ates), thus reaching levels of parallelisation that were, tillecently, prerogative of fluorescence-based arrays. The advan-ageous chemistry used for the bio-modification of the Pt

icroelectrodes allowed the microarrays to be chemicallytripped of hybridised targets and re-used. No memory effectsr significant loss of sensitivity were observed after re-use ofmicroarray up to three times.

.5. Gold electrodes

mmobilisation by self-assembly onto gold electrodes, widelysed for both electrochemical and piezoelectric sensing, takesdvantage of the strong interaction (chemisorption) which isstablished between thiolated molecules and this metal sur-ace.

While earlier works described preliminary formation of aAM of a reactive thiol followed by carbodiimide coupling ofhosphate, amino or carboxyl-terminated oligonucleotides,he most recent approaches involved direct use of thiol-ethered oligos (e.g., [12,16,30–33]). Besides reducing theumber of steps required for immobilisation, the latter strat-gy also featured another important advantage. As pioneeredy the studies of Herne and Tarlov [34], post-treatment ofhe probe-modified surface with a secondary thiol acting asdiluent” (e.g., mercaptohexanol, MCH) represented a sim-le yet effective means to control density and availabilityf the capture probe. The secondary thiol displaced theon-specifically adsorbed probe molecules, while leaving theemaining ones in an upright position. Hence, consistentlyith the primary attachment of the probe through the thiol-

nd group, the hybridisation efficiency was greatly enhanced.n additional feature of this immobilisation chemistry is

he stability of the surface-attached biolayer. With the onlyequirement of storing the genosensors in the dark and in
ry conditions to prevent possible oxidation and breakagef the thiol-gold bonds, the performance characteristics ofuch DNA biosensors were reported to be unaltered over aeriod of at least 2 months [35]. Self-assembled monolay-
ig. 4 – Enzyme- and electro-amplified detection of unmodified seaction. Adapted from Ref. [35].

0 9 ( 2 0 0 8 ) 139–159 145

ers of thiol-modified probes also possessed notable thermalstability, being unaffected by temperature gradients up to70 ◦C [36].

With an exception being CombiMatrix technology, immo-bilisation via chemisorption of thiolated probes is the methodof choice for all commercially available electrochemical arrays.This fact additionally confirms the enormous advantages ofsuch a protocol in terms of simplicity, robustness and reli-ability. In the following, specific technologies and surfacemodifications related to the use of gold electrodes (for bothelectrochemical and piezoelectric sensing) will be furtherillustrated.

3.5.1. Electrochemical sensing3.5.1.1. Chemisorption onto interdigitated microelectrodesarrays. Elsholz et al. [35] described their analytical platformas a chip carrying 16 electrode positions (500 �m in diameter),each consisting of arrays of interdigitated gold electrodes. Asthe detection mechanism of choice relied on redox cycling theproduct of an enzymatic reaction (Fig. 4), the comb-like struc-tures were designed so that gaps of just 400 nm separatedthe 800 nm wide anode and cathode fingers. To demon-strate the multiplexing capability of their commerciallyavailable analytical device (“eMicroLISA” automated ArrayAnalyzer), thiol-tethered oligonucleotide probes specific forfive pathogens were spotted in triplicate onto the chip usinga piezoelectric nanoplotter.

3.5.1.2. Electrochemical lithography. Plaxco’s group exploitedalternate chemisorption and electrochemically driven des-orption steps to selectively immobilise different captureoligonucleotides onto closely spaced gold electrodes [37].

Inspired by photolithography, in which a light-sensitiveresist that covers a surface is selectively removed to exposespecific regions to further modifications, the electrodes werefirst passivated with a self-assembled monolayer of 11-
mercapto-1-undecanol. Site-specifically, the monolayer couldbe then stripped off by electro-oxidation, thus regeneratinga bare gold surface available for immobilisation of a specificprobe. Interestingly, immobilisation and oxidative desorption
equences: redox recycling the product of an enzymatic

a c t a
conditions were so gentle that sequential labelling of multiple,closely spaced electrodes (95 �m separation), was possi-ble. Unlike photolithography or dip-pen nanolithography,the “electrochemical lithography” did not require expensivemicromechanical devices, obviating the need for having directphysical access to the electrode surface. However, the lengthof the process (which was limited by thiol adsorption kineticsand the need for extensive washing steps) probably limited toabout 100, the number of electrodes that could be modified ina reasonable time frame.

3.5.1.3. Immobilisation onto sandwiched assemblies. With theaim to improve the electrical communication between aredox enzyme and the electrode surface, Dominguez et al.exploited electrostatic self-assembly as a means to sandwichthe anionic horseradish peroxidase between two layers of apositively charged osmium-based redox polymer [38]. Plat-form for such multi-layer assembling was a gold electrodemodified with a negatively charged monolayer of 3-mercapto-1-propane sulfonic acid. The oligonucleotide probes were thencovalently attached through their 3′ phosphate groups to thefree amino functionalities of the redox polymer. Embeddedinto the redox complex, peroxidase exhibited enhanced com-munication with the electrode, thus efficiently supporting thebio-catalysed reduction of the hydrogen peroxide generatedby an oligonucleotide-conjugated glucose oxidase label.

3.5.1.4. Immobilisation onto an electropolymerised film. Gar-nier et al. used a low-density array of gold electrodes(100 �m in diameter) to develop a DNA chip based on anelectropolymerised film of polypyrrole bearing captureoligonucleotides [39]. Two monomers (3-acetic acid pyrroleand 3-N-hydroxyphthalimide pyrrole) were first electro-copolymerised onto the gold surfaces. Covalent attachmentof the capture probes was then obtained by chemical substi-tution of the easy leaving ester group N-hydroxyphthalimideby the amino-substituted oligonucleotides. The system wasused for monitoring the interaction between the surface-immobilised probes and the synthetic targets in a label-freeand real-time mode.

3.5.2. Piezoelectric sensingWhile in the case of electrochemical genosensors directchemisorption of thiol-modified probes onto the gold elec-trode is one of the most popularly employed methods, inthe case of piezoelectric sensing the gold electrode surfaceis typically modified using more complex immobilisationchemistries (e.g., thiol/dextran/streptavidin modification).Although longer (taking a few days instead of a few hours), thelatter strategies provide the immobilised bio-layers of supe-rior analytical performances (robustness for repeated use,higher sensitivity, etc.). Use of an immobilisation chemistrywhich leads to sensing interfaces resistant to non-specificadsorption phenomena is also of particular importance. Sen-sitivity and specificity of QCM devices are, in fact, essentiallylimited by the non-specific adsorption of the constituents of
real matrices.
3.5.2.1. Direct chemisorption of thiol-tethered probes. As forelectrochemical genosensors, the influence of the interfacial

6 0 9 ( 2 0 0 8 ) 139–159

design on efficiency and kinetics of the hybridisation reac-tion has been widely investigated over the past decade. Morerecently, Gooding et al. [40] elucidated the effect of the lengthof the diluent thiol used to create mixed self-assembled mono-layers. Wu et al. [41] also demonstrated that the addition of a12-deoxythymidine-5′-monophosphate (12-dT) spacer at the5′-end of the 30-mer immobilised thiolated probe enhancedby twofold the hybridisation efficiency.

3.5.2.2. Attachment of biotinylated probes to (strept)avidin-coated surfaces. Functionalised alkane thiols form stableself-assembled layers on planar surfaces and act, therefore,as ideal linkers to achieve single-point attachment of the DNAprobe at either 5′- or 3′-end [42].

A procedure widely employed on QCM devices exploitsthe well-known affinity between biotin and (strept)avidin.This method relies on four basic steps: formation of a self-assembled monolayer of functionalised alkane thiols ontothe gold surface; subsequent binding of a dextran hydrogel;covalent attachment of (strept)avidin; affinity immobilisationof biotin-labelled probes. This immobilisation chemistry, ini-tially introduced for optical (surface plasmon resonance, SPR)sensing, was successfully transferred to piezoelectric trans-ducers by Tombelli et al. [43]. The good performances of suchbiotinylated probes (immobilised onto dextran-bound strep-tavidin) have been widely demonstrated for targets of clinical[44–46], food and environmental interest [47,48]. Tombelli etal. [43] also compared the performance of QCM genosensingplatforms obtained using either the thiol/dextran/streptavidinchemistry or thiolated probes directly. The two immobilisa-tion procedures exhibited similar analytical characteristics interms of selectivity, sensitivity and reproducibility.

Several alternative immobilisation methods have also beenused to create the streptavidin layer at the sensor surface. Oneof the most frequently used is the covalent coupling of the freeamino groups of streptavidin to carboxyl-functionalised sur-faces [49]. This attachment procedure does not influence theability of streptavidin to bind biotin, thus leaving the proteinavailable for the immobilisation of the biotinylated oligonu-cleotide probes.

The influence of streptavidin packing density and filmstructure on binding and orientation of the DNA probe wasstudied by Knoll and co-workers [49]. How these parametersinfluenced the hybridisation efficiency of the probe was inves-tigated comparing two different methods for immobilisinga streptavidin layer onto the gold surface: direct interactionof streptavidin with a biotin-terminated SAM and covalentattachment of the protein onto an 11-mercaptoundecanoicacid SAM via amine coupling. Streptavidin films assembledonto the biotin-containing thiols appeared to be highly rigidwith a well-ordered structure, while the streptavidin filmformed via amine coupling was less ordered. Interestingly,the biotinylated probes immobilised onto the well-organisedstreptavidin layer allowed a higher sensitivity to be obtained.

The same group exploited such biotin/streptavidin/biotin-probe bridges as platforms to detect single-base mutations,
insertions and deletions [50]. The gold electrodes of the QCMwere first modified with a mixed layer of a biotin-terminatedthiol and an OH-terminated thiol; streptavidin was thenbound to this film. Using a biotinylated probe, DNA mutations

t a 6

iamd

Sao

4

Ihosgo

custhsyd

4

Mpts(emdobps

4APff(iHdpdaogHso


n oligonucleotide targets were detected by the use of MutS,particular protein which binds DNA sequences containingispaired and unpaired basis, but not the perfectly matched

uplex.Use of biotin-doped supported lipid bilayers (b-SPB) on

iO2-coated QCM crystals was also reported by Larsson etl. for immobilising streptavidin first and then biotinylatedligonucleotides [51].

. Hybridisation reaction

f, on one hand, RNA-based assays benefit from the typicallyigh number of ribosomes per cell, on the other the sensitivityf most genoassays do not generally allow the direct analy-is of sequences that occur as a single copy within organismsenome. Detection thus depends on the PCR pre-amplificationf the target sequence.

Specific aspects crucial to the pre-analytical and analyti-al phase will be illustrated in the next sections. Along withse of capture probes with improved selectivity and hybridi-ation efficiency (see Section 2), specific pre-treatments ofhe samples can be used to greatly enhance the yield of theeterogeneous hybridisation events and, consequently, theensitivity of detection. The reasons that make direct anal-sis of genomic DNAs (i.e., unamplified sequences) extremelyifficult will be also discussed.

.1. Sample pre-treatment

any analytical protocols require the samples to bere-treated prior to hybridisation. The aims of such pre-reatments include: (a) the reduction of the complexity of theample; (b) the separation of possible interfering moleculessuch as unincorporated primers and nucleotides); (c) the gen-ration of the target in a single-stranded form (or at least theinimisation of amplicon sister strand re-annealing); (d) the

isruption of thermodynamically stable secondary structuresf the sequence (which might severely inhibit the interfacialiorecognition process). As a general rule, the number of sam-le pre-treatments should be kept as small as possible, as theyignificantly slow down the whole analytical process.

.1.1. Genomic DNAlthough high levels of fidelity generally characterise theCR amplification step, this process have always to be care-ully optimised. Non-specific annealing of the primers can,or example, lead to the generation of a wrong producti.e., a false positive amplicon), while the presence of annhibitor of polymerase can lead to a false negative result.aving the possibility to directly sense genomic DNAs, theserawbacks would be all overcome and the number of sam-le pre-treatments reduced, thus significantly facilitating theetection of nucleic acids. Nevertheless, scientists who havettempted the analysis of unamplified genomic DNA haveften faced insurmountable difficulties. The huge size of
enomic samples constitutes the first of a series of obstacles.ence, in order to facilitate the interfacial hybridisation of
uch targets, fragmentation of the samples by sonication [52]r enzymatic digestion [53,54] has been suggested. Although

0 9 ( 2 0 0 8 ) 139–159 147

simpler and cheaper, an undesired effect of sonication (whichis a random process) might be disruption of the sequence tar-geted by the surface-immobilised probe. Nevertheless, evenenvisioning an optimal fragmentation, additional problemsarise. If one considers that the genome of many organismshas a size of about 108 to 109 bp and that fragments in theorder of 500 bp on average would probably be the most suitedfor hybridisation, as few as a single copy of the target would be“lost” between 105 and 106 non-relevant strands. Diffusion ofthe fragment containing the target toward the probe-modifiedsurface would be significantly slowed down by the concomi-tant presence of such sequences and further hindered by theirelectrostatic repulsion when accumulated in close proximityof the sensing layer. Effective match of the capture probe withthe target sequence (the process that leads to duplex forma-tion) is, therefore, a rather unlikely event. One has additionallyto consider that, due to the complexity of genomic DNA,several of the obtained fragments could include sequencesthat present substantial homology with the true target. Yield-ing non-specific hybridisation with the capture probe, suchsequences would be, therefore, responsible for false positivesignals. The development of sufficiently sensitive methodsrepresents an additional analytical challenge when levels ofselectivity are required that single-nucleotide polymorphismscan be unambiguously discerned. To these authors knowl-edge, the work of Patolsky et al. still constitute the uniqueexample describing the electrochemical detection of a single-nucleotide polymorphism in unamplified genomic samples[53]. QCM sensing of genomic sequences related to the P35promoter in transgenic plants (i.e., Nicotiana glauca) was alsoobtained after fragmentation with restriction enzymes anduse of an optimised sample denaturation step [55]. Viscoelas-tic effects coupled to the increase of mass at the surface ofthe sensor combined with a denaturation protocol aimed toprolong the lifetime of the target in its single-stranded statecould account of this sensitive detection. Similar results wereobtained by Evans and co-workers for bacterial DNAs [56].

The double role of the PCR amplification thus clearlyappears from the above-mentioned considerations. Firstly, ithelps reducing the complexity of the target DNA; secondly, itincreases the concentration of the sample.

A case where the amount of genomic sample is unusuallyabundant is represented by highly repeated (microsatellite)sequences. In contrast to single copy targets per aploidgenome, microsatellite sequences appear in high numbers ofcopies, thus making their detection significantly more acces-sible. Minunni et al. [54] reported, for example, the detectionof microsatellite bovine DNAs using a QCM device. Again, thepossibility to successfully detect such a genomic sample usingQCM mainly relied on the development of an accurate sam-ple pre-treatment (fragmentation with restriction enzymesand proper thermal denaturation). Similarly, Mascini et al.detected this microsatellite bovine genomic DNA through alabel-free electrochemical genoassay [57].

4.1.2. PCR-amplified samples
Analysis of PCR amplicons is definitely simpler than anal-ysis of genomic DNA. However, even when dealing withPCR-amplified sequences, detection can be anything butstraightforward. Inconsistent results along with a consider-

a c t a
able number of false positive and false negative responseswere, for example, found by Barlaan et al. [10] when workingwith relatively long (533 bp) amplicons. Shorter PCR products(200 and 350 bp) gave much better results.

Amplified sequences are usually in a double-stranded formthat have to be thermally or chemically denatured. Even doingso, the strand complementary to the amplified target (stillpresent in solution) competes with the immobilised captureprobe for hybridising the target itself. Upon decreasing theamount of the single-stranded sequence available for surfacehybridisation, this phenomenon is thus responsible for gener-ally lower hybridisation yields.

In order to overcome this problem three main strategieshave been developed. The first one relies on use of asymmet-ric PCR [15,25], a process that naturally leads to samples inwhich an excess of the single-stranded target is present. Themain disadvantage of this method is, however, that amplifi-cation of the target proceeds linearly with the number of PCRcycles instead of exponentially as in the case of usual PCR pro-tocols [58]. Relatively lower amplification factors are thereforeobtained when using this strategy.

Single-stranded targets can also be obtained by chemicaldenaturation of the duplex followed by affinity binding of thebiotinylated interfering strand [59] or through specific enzy-matic digestion of the undesired sequence [60]. Both methodsare, however, labour-intensive and time-consuming.

A simple, rapid and cost-effective alternative to thesemethodologies is represented by the use of the so-called“helper” or “blocking” oligonucleotides [9,35,61]. When consid-ering heterogeneous hybridisation reactions, the accessibilityof the target sequence by the oligonucleotide probe is oneof the key conditions determining the overall hybridisationefficiency. The helper oligonucleotides are unmodified oligoswhich, in the most efficient format, are hybridised to thesample upstream and downstream the site targeted by thecapture probe (Fig. 5). The resulting single-stranded bulge is
thus easily accessible to the immobilised probe, determin-ing substantially higher heterogeneous hybridisation yields.Similar unmodified “blocking” oligos were also used as ageneral means to inhibit re-annealing of double-stranded
Fig. 5 – Use of “helper” oligonucleotides to facilitate the heterogethermodynamically stable secondary structures. Adapted from R

6 0 9 ( 2 0 0 8 ) 139–159

amplicons [15,47,48,61], to break thermodynamically stablesecondary structures of the target [35] and even to controlthe extent of chemical fragmentation of RNA targets by form-ing relatively short stretches of cleavage-resistant RNA/DNAduplexes [35]. Notably, helper oligonucleotides can be addedto the sample solution just prior to its thermal denaturation.Hence, heterogeneous hybridisation of the target and bindingof helper oligonucleotides occurs in a single step, eliminat-ing the need to perform successive rounds of hybridisation.All these characteristics thus render use of this strategy par-ticularly amenable to solve a range of hybridisation-relatedproblems.

4.2. Passive vs. active hybridisation

The hybridisation reaction can be carried out in both a pas-sive and an active mode. Passive hybridisation takes placewhen temperature, buffer composition (salt concentration,formamide, etc.) and washes are used to control the stringencyand rate of the reaction. These and other experimental vari-ables were, for example, recently explored [11,62] to improvethe selectivity of single-base mutation detection.

In contrast to the passive approach, active hybridisationuses electric fields on a microelectronic device to regulatenucleic acids transport, hybridisation and stringency [10].This technology, developed and implemented by Nanogen Inc.for arrays of 100 individually addressable microelectrodes, isreported to possess several advantages over passive method-ologies. An electric field is firstly used to address differentprobes at specific sites of the microarray. The applied positivepotential can also be used to increase the concentration of thetarget locally at the functionalised electrodes. Hence, whilepassive hybridisation is limited by diffusion, electrochemicalpreconcentration of the target can ultimately lead to increasedhybridisation rates and efficiencies. Finally, reversing of thepotential (i.e., the application of negative voltages) imparts
to the system an extremely high stringency for mutationdetection based on the preferential destabilisation of mis-matched hybrids by means of purely electrostatic (repulsive)forces.
neous hybridisations of targets prone to fold intoef. [35].

t a 6

5

Osdcb

iaHrftcriotttotlt(hfiaacctwbafm

coteTap

5

5W(gtamab


. Hybridisation detection

nce the target DNA has been captured onto the sensorurface, a range of different approaches can be used for trans-ucing the biorecognition event. The transducing principlesan be broadly divided into reagentless, label-free and label-ased schemes.

Reagentless sensing concepts are the simplest, as noth-ng but the sample solution itself is needed to perform thenalysis. Reagentless methods are not necessarily label-free.owever, when both conditions are met (i.e., the method is

eagentless and label-free), one can take advantage from theact that undesired effects, such as steric impediments tohe hybridisation reaction due to the reporter molecule, areompletely absent. Such schemes have thus attracted intenseesearch efforts due to their promise to provide analyticalnformation in a faster, cheaper and safer mode compared tother strategies. As an additional advantage, transduction inhe reagentless mode can also give access to the kinetics ofhe biorecognition event. Piezoelectric transduction is essen-ially reagentless and label-free, with only a limited numberf exceptions being described. Indeed, much more difficult iso draw net boundary lines between the concepts reagentless,abel-free and label-based when considering electrochemicalransducers. Those methods that use solution-phase reagentse.g., metal complexes or organic dyes) as markers of theybridisation process will be referred to as label-free. Label-ree approaches typically rely on the measurement of changesn the electrical characteristics of sensing layer, before andfter the hybridisation reaction. By contrast, when organicnd organometallic electroactive compounds, nanoparticles,atalytic and redox enzymes are permanently bound (e.g.,ovalently or via (strept)avidin–biotin interactions) to one ofhe constituents of the surface-tethered duplex, the methodill be considered as label-based. Sensitivity and relia-ility of label-based approaches are often still unrivalled,s also witnessed by the choice of these methodologiesor the electrochemical microarray platforms now on the

arket.As one of the most attractive features of electrochemi-

al and piezoelectric sensing, electrochemical reactions onne hand and mass changes on the other, give an elec-ronic output directly, which makes the signal transductionquipment particularly compact and relatively inexpensive.hose mature technologies which have led to commerciallyvailable DNA sensors will be discussed within a dedicatedaragraph.

.1. Reagentless methods

.1.1. Fully electronic detectionith the introduction of the differential transfer function

DTF) method, Ingebrandt et al. substantially improved theeneral reliability of the label-free DNA hybridisation detec-ion based on silicon FET microarrays [26]. Enabling fast
nd fully electronic readout of ex situ hybridisations, such aethod relied on the intrinsic charge of the DNA molecules
nd/or on changes of the interfacial impedance that followedinding of the target sequence. Ingebrandt’s approach over-

0 9 ( 2 0 0 8 ) 139–159 149

came the drawbacks of the systems that used a time-resolveddc readout, where the FET signal suffers from sensor drift,temperature drift, changes in electrolyte composition or pH,influence of the reference electrode. Channels modified with afully non-complementary probe and three and two bases mis-matched capture oligonucleotides provided reference signalsthat were subtracted from the outputs generated in the pres-ence of single-base mismatched or perfectly matched probes.The resulting differential readout of the transfer-function can-celled out any signal change due to non-specific binding ofthe target DNA, thus allowing easy detection of the SNP in lowionic strength buffers when using low ac frequencies. The sen-sitivity of the method (�M range) required, however, furtherimprovement.

Gao et al. described real-time and label-free electricaldetection of DNA at femtomolar levels using an array ofhighly ordered SiNW [27]. A monolayer of neutral PNA captureprobes, assembled onto the individual SiNWs via silane chem-istry, acted as the biosensing interface. When exposed to thenegatively charged target sequence, the PNA-functionalisedSiNWs showed a concentration-dependent increase of resis-tance. As with other SiNW biosensing devices, the sensingmechanism could be understood in terms of the change incharge density at the SiNW surface after hybridisation, theso-called “field effect”. The reason for using PNA insteadof DNA probes was twofold. Due to the neutral characterof the PNA backbone, hybridisation could take place in lowionic strength buffers, thus minimising the background con-tribution from dissolved salt ions while emphasising thehybridisation-related resistance changes. The SiNW array alsosatisfactorily discriminated mismatched target DNAs, withthe 1 bp mismatch yielding only 15% of the signal observed forthe perfectly matched target. Besides being ultrasensitive andselective, an additional advantage of the SiNW biosensor arraydescribed by Gao was the capability to detect the hybridisationevents in situ, in real-time and in a reagentless and label-freemode. Nonetheless, as pointed out by the authors themselves,the analytical signal intensity (in terms of signal/unit concen-tration) was very low, while being superimposed on a highbackground.

5.1.2. Electronic beaconsJenkins et al. developed a reagentless assay for the detection ofAgrobacterium tumefaciens strain C58 using a mixed monolayerof a ferrocene-labelled hairpin probe and �-mercaptoethanolassembled onto the gold surface of disposable screen-printed electrodes [12]. Upon hybridisation, the hairpin probeunderwent a conformational change which displaced theredox-active label from the electrode surface. This processresulted in a diminished efficiency for electron transfer fromthe label, which was detected as a drop in peak redox currentmeasured by cyclic voltammetry (Fig. 1).

The unexpectedly high electrochemical signals allowedthe oligonucleotide target to be quantified over seven ordersof magnitude (100 fM to 1 �M). However, when performingthe assay at room temperature, the hairpin probes yielded
non-specific hybridisation with a number of oligonucleotidesequences which presented a certain degree of homologywith the target. Notably, the perfectly matched target wasdistinguished from the single-base mismatched sequence by

a c t a
recording in situ melting profiles. While increasing the tem-perature of the hybridisation solution, the ferrocene oxidationsignals were periodically measured, evidencing a �Tm of about5 ◦C between the target and the SNP-containing sequence.Such results were consistent with the behaviour theoreticallypredicted for such oligonucleotide duplexes.

Using a methylene blue-modified hairpin probe covalentlyattached to a gold electrode, Ricci et al. [14] characterised howsignalling properties, specificity, and response time of suchsystems depended on probe surface density, target length, andother aspects of interfacial molecular crowding. The highestsignal suppression (i.e., the highest sensitivity) was obtainedwith the highest probe density investigated; however, theresponse time also increased at high probe coverages. Thekinetics of the redox process was further investigated usingac voltammetry. This allowed to suggest that the signallingmechanism of electrochemical hairpins might essentially relyon hybridisation-induced changes in the rate with which theredox moiety collides with the electrode and transfers elec-trons.

Most of the methods based on the use of redox-labelledhairpin probes typically rely on a “signal-off” mechanism.Thus, they suffer from the need to measure small vari-ations (diminutions) of the signal, when the backgroundvalue is the highest possible. With the aim to overcomesuch a key drawback of hairpin probes, Grinstaff and co-workers [32] developed an alternative sensing scheme inwhich a thiol-tethered capture probe and a ferrocene-labelledreporter sequence were interconnected through a flexiblepoly(ethylene glycol) spacer. This system exploited the confor-mational change that occurred when the surface-immobilisedtriblock assembly bound to the target strand. When the macro-molecular assembly and the mercaptopropionic acid spacerwere co-immobilised onto the gold electrode, the 5′-terminalredox label was electrochemically inaccessible because elec-trostatically repelled far from the anionic surface. Uponbinding of the target DNA, folding of the oligo-PEG-oligomacromolecule decreased the distance between the ferrocenemoiety and the electrode, thus affording an electrochemicalsignal.

5.1.3. Polypyrrole electrochemistryGarnier et al. achieved reagentless, label-free and real-timeelectrochemical sensing of DNA sequences using captureoligonucleotides covalently immobilised onto an electropoly-merised thin film of polypyrrole [39]. Such a conjugatedpolymer film possessed an electrochemical signature whichunderwent a modification upon hybridisation of the graftedprobes with their complementary targets. This effect wasrelated to the increase of steric hindrance which followedhybridisation at each probe immobilisation site. Because of thebulkiness of the hybridised DNA pending along the polypyrrolechains, the structural reorganisation of the macromolecularpolymer (which normally imposes the complete planarity ofthe whole pyrrole system) was forbidden. This was reflectedinto a modification of the electrochemical properties of the
polymer film, thus offering direct measurement of the hybridi-sation kinetics. Interestingly, the authors showed that thequantity of probes immobilised onto the polypyrrole governedthe detection threshold of the method. By tuning the amount
6 0 9 ( 2 0 0 8 ) 139–159

of the surface-attached capture oligonucleotides to the quan-tity of target present in solution, a detection limit of about1 pM was obtained.

5.1.4. Alternative approaches in QCM sensingTill recently, most of QCM studies have been conducted usingcontinuous resonance devices. However, in the past few yearsdiscontinuous resonance methodologies, mostly the QCM-D technique, have shown their potential for monitoring thehybridisation reaction. In comparison with traditional vari-ants, the key feature of QCM-D technique is that, in addition tochanges in resonant frequency, f, it also provides simultaneousmeasurement of energy dissipation, D, induced by interfacialreactions [63]. If, on one hand, the resonance frequency f isrelated to the mass m bound at the sensor surface, on theother the energy dissipation D depends on the viscoelasticproperties of the interface.

The measurement of the energy dissipation (the so-calledD factor) has become popular to indicate the energy loss ofthe viscous (immobilised) bio-layer against the total vibra-tion energy of the QCM plate. The D factor has been reportedto indicate the energy loss between viscous molecules andthe media in cases of heterogeneous DNA hybridisation [51],adsorption and subsequent structural modification of largevesicles [64] and proteins [65]. The information contained incombined energy dissipation and frequency measurementshas an added value since it gives much richer informationthan a mass measurement alone. Of particular value is thatstructural changes of surface-bound molecules (e.g., uponantigen–antibody binding or vesicle decomposition into a sup-ported lipid bilayer) yield signals that, on the basis of existingtheories, can be converted into well-known quantities suchas film thickness, shear viscosity and elastic modulus of theadlayers [66].

Use of QCM-D to study nucleic acid interactions and tocompare different immobilisation procedures and probes (i.e.,DNA and PNA probes) has been reported by Hook et al. [66].Combined measurements of dissipation energy and frequencychanges by using a 27-MHz piezoelectric quartz crystal havealso been recently reported for monitoring conformationalchanges of proteins, single-stranded and double-strandedDNAs [67].

5.2. Label-free methods (solution-phase metalcomplexes and organic dyes)

Indicators such as cationic metal complexes [e.g., Ru(NH3)63+,[Fe(CN)6]3−/4−, Co(phen)33+ and Ru(bpy)32+], or organic com-pounds, like Hoechst 33258, daunomycin and methylene blue,recognise the hybridised DNA binding to the grooves of thehelix, selectively oxidising the guanine moiety, intercalat-ing selectively and reversibly into the double-stranded DNA.Being their use for electrochemical sensing investigated sincethe pioneering work of Millan and Mikkelsen in the early1990s [68], most of them have already been comprehensivelydescribed in Ref. [6] and other reviews. Research on these
classes of compounds is, however, still open, with recognitionmechanisms spanning from the simple yet effective electro-static attraction/repulsion to the more complex intercalation,in simple and threading mode.

t a 6

cabbsPosttimaaspcohdm

fIctehptP(sb

1rgTfrocPwaestttre

Depaa


Among the different electrochemical techniques, electro-hemical impedance spectroscopy (EIS) has shown to ben effective and sensitive tool for the characterization ofio-functionalised electrodes and for monitoring interfacialiocatalytic reactions [53]. Knoll and co-workers [16] employeduch a powerful technique for in situ characterisation ofNA/DNA hybridisation kinetics. Impedimetric sensing reliedn the dramatic change of electrical charge at the electrodeurface that resulted from hybridisation of the polyanionicarget DNA to the neutral PNA probe. Buffered solutions con-aining both the unmodified target and the anionic redoxndicator [Fe(CN)6]3−/4− were interacted with the PNA/MCH

odified electrodes. As hybridisation proceeded, increasingmounts of the negatively charged target were specificallyccumulated at the electrode surface. The resulting electro-tatic repulsion between the redox-active molecules and theolyanionic backbone of the targets thus led to increasingharge transfer resistance (Rct) values. By recording the changef Rct with time, EIS allowed in situ characterisation of theybridisation kinetics of PNA/DNA duplexes. Association andissociation kinetics of a fully matched and a single-base mis-atched duplex were obtained.Steichen et al. [17] also used a 16-mer PNA probe and ac

aradic admittance for monitoring the hybridisation reaction.n contrast to Knoll’s paper, these authors used the highlyharged Ru(NH3)63+ cation as the redox probe. Transductionhus relied on the evaluation of the amount of Ru(NH3)63+

lectrostatically associated to the anionic backbone of theybridised target DNA. When compared to an analogous DNArobe (also 16-mer), the PNA exhibited higher levels of selec-ivity against a single-base mutant target. Analysis of twoCR-amplified sequences from the 23S rRNA gene of H. pylori100 and 400 bp fragments, respectively) evidenced a higherensitivity for the longer amplicon, as it possessed much moreinding sites (phosphate groups) for the Ru complex.

Following an analogous strategy, Kerman et al. [15] used an1-mer PNA probe immobilised onto a GCE and the Co(NH3)63+

edox probe to detect a single-base mutation within theene encoding the human alcohol dehydrogenase (ALDH).he selectivity of the electrochemical method directly derived

rom the selectivity of the PCR protocol used to amplify theeal samples. Selective amplification of either the wild-typer mutant ALDH target was obtained by means of a “PCRlamping” strategy. This method was based on the ability ofNA/DNA complexes to block the formation of a PCR producthen a PNA (also introduced into the PCR mix) was targeted

gainst one of the PCR primer sites. Interestingly, this blockagevent allowed selective amplification/suppression of targetequences that differed by only one base pair. Once generated,he double-stranded amplicons were thermally denatured andhen interacted with the PNA capture probe immobilised athe surface of a GCE. After accumulation of the Co(NH3)63+

edox probe, differential pulse voltammetry was used for thelectrochemical interrogation of the surface.

Gorodetsky and Barton demonstrated that reduction ofNA-bound intercalators (e.g., methylene blue) at highly ori-
nted pyrolytic graphite electrodes (HOPG) modified withyrene-functionalised duplexes also proceeded being medi-ted by the DNA helix [20]. The signal from the intercalator wasttenuated in the presence of the single-base mismatches CA
0 9 ( 2 0 0 8 ) 139–159 151

and GT. As on gold, this sensitivity to single-base mismatcheswas enhanced when methylene blue reduction was coupledin an electrocatalytic cycle with ferricyanide. The extendedpotential range afforded by the HOPG surface also allowedto investigate the electrochemistry of previously inaccessiblemetallointercalators [Ru(bpy)2dppz2+ and Os(phen)2dppz2+]which proved to be suitable labels for DNA-mediated chargetransfer reactions.

Following the concept of long-range charge transferthrough a DNA duplex developed by Barton’s group, Wong andGooding [33] developed an improved and real-time electro-chemical genoassay to detect single-base pair mismatches.Gooding’s approach overcame the one limitation of Barton’stechnology, namely the need to create closely packed arraysof DNA molecules to ensure the electrochemical response wasdue to charge transfer through the double helix only. TheDNA recognition interface consisted of a mixed monolayer ofloosely packed thiolated oligonucleotide probes and mercap-tohexanol immobilised onto a gold electrode surface. Such adesign of the sensing layer thus ensured a higher hybridisa-tion efficiency at the transducer–solution interface. An anionicintercalator, anthraquinonemonosulfonic acid (AQMS), wasalso selected. Hence, the electrostatic repulsion between theanionic AQMS molecules and the hydroxyl-terminated MCHlayer limited the direct interaction of the redox reporter withthe metal surface. In situ and real-time monitoring of thehybridisation events was allowed, as AQMS and the targets(perfectly matched and mismatched) coexisted within thesolution hybridised at the electrode surface. Interestingly, theelectrochemical behaviour AQMS changed according to its sur-rounding environment, with the freely diffusing compoundbeing reduced at lower potentials than the intercalated one.This shift in the reduction potential of AQMS actually pro-vided the means for monitoring the hybridisation event in situ(as no washing steps were required) and in real-time. Con-firming the good selectivity of the assays based on long-rangecharge transfer, such an improved detection scheme allowedclear differentiation of the perfectly matched target from C·Aor G·A mismatched sequences. Compared to previous works,the operational simplicity of this in situ assay constituted themajor advantage.

Tansil et al. [69] described synthesis, characterisation,and analytical applications of a novel imidazole-substitutednaphthalene diimide (PIND) threading intercalator function-alised with electrocatalytic redox moieties. Two Os(bpy)2Cl2complexes were grafted onto each PIND unit through coor-dinative bonds with the imidazole groups, thus yielding a[Os(bpy)2Cl]+–PIND–[Os(bpy)2Cl]+ compound (PIND–Os). Inter-estingly, the naphthalene diimide bound to the double helixin a “classical” threading intercalation mode, while the twoOs(bpy)2Cl+ pendants interacted with DNA via electrostaticinteraction, thus reinforcing the intercalation by “locking up”the naphthalene diimide group in place. The experimentalresults suggested that PIND-Os was highly selective to ds-DNA, while presenting a higher binding constant with respectto its parent compound PIND. The combination of the Os-
labelled threading intercalator with a sacrificial electron donor(ascorbic acid) generated highly enhanced electrocatalytic cur-rents, thus providing a platform for ultrasensitive detection ofDNA.

a c t a
5.3. Label-based methods

Shen and co-workers [70] developed a novel strategy to detectsingle-point mutations by combining a DNA ligase assay witha reverse molecular beacon concept. A detection probe con-taining a phosphoryl group at the 5′-end, a ferrocene tag atthe 3′-end, and a six-base sequence close to the 3′-end com-plementary to the 5′-end region of the thiol-tethered captureprobe was used. After “sandwiching” the target DNA betweenthe surface-immobilised capture probe and the detectionprobe, a ligation reaction was carried out in the presenceof E. coli DNA ligase. As a result of the subsequent ther-mal denaturation (which dissociated the target sequencefrom the surface), the ligated assembly assumed a molecu-lar beacon-like hairpin structure which brought the ferrocenelabel in closer proximity to the electrode surface. By thismethod, the target DNA could be determined down to thepicomolar range. Additionally, because of the high fidelityof E. coli ligase, the genosensor exhibited excellent capabil-ity to discriminate single-base mutant targets. In contrastto existing methods based on the conformational change ofredox-labelled oligonucleotides, Shen’s strategy offered sev-eral substantial advantages, such as negligible backgroundcurrent and “signal-on” mechanism.

Flechsig and Reske extended the simple and effective tech-nology developed by Palecek and co-workers for labellingthymine-rich sequences to develop a heterogeneous hybridis-ation assay [36]. Specifically, the method relied on the covalentmodification of the target sequence using [OsO4(bipy)]. Thiscomplex preferentially reacted with thymine bases forming adiester of osmium(VI) acid through the oxidation of the C–C–double bond of the pyrimidine ring. Interestingly, the reac-tion of the osmium tetroxide complex was strand-selective,proceeding at a reasonable rate only when the oligos werein their single-stranded form. The Os-modified strands areknown to lose their hybridisation capability. However, withthe simple requirement of masking the hybridisation sitesof the sequences using short “protective” oligonucleotides,this method allowed convenient and easy synthesis of redox-labelled strands.

Interesting aspects of this approach are as follows. Thelabelling reaction results particularly effective when thesequences are provided of a non-hybridising poly-T tail, andmultiple redox markers can be simultaneously introduced.Additionally, if compared to other metal complexes (e.g.,ferrocene derivatives), one can take advantage of the two elec-trons reaction of the reversible osmium couple [Os(VI/IV)] tofurther improve the sensitivity of the assay. As the last consid-eration, the redox potential of the Os label is extremely facileto be tuned, by simply changing the organic ligand into thecomplex. This strategy thus offers the possibility to developa series of redox-labelled signalling probes each with a dif-ferent formal potential and, therefore, suitable for use in amultiplexed assay.

5.4. Signal amplification

One of the limiting factors for the development of DNA sen-sors and arrays is the sensitivity. When the specific targetgene is present as a single copy in the organism genome, the

6 0 9 ( 2 0 0 8 ) 139–159

amount of DNA that has to be detected is, in fact, at the atto-molar to femtomolar level. To date, as the possibility to directlysense such low concentrations via heterogeneous hybridisa-tion to surface-immobilised probes has not reliably achieved(see Section 4.1.1 for more details), PCR amplification partiallysolves the problem, increasing the amount of target DNA to beidentified. Intense research efforts are, nevertheless, devotedto further improve the sensitivity of the analytical protocols.Interestingly, the most efficient signal amplification pathsare now approaching sub-femtomolar detection limits. Thesuccessful use of these ultrasensitive bioelectronic detectionschemes requires, however, proper attention to non-specificadsorption issues, as they commonly control the sensitivityof such bioaffinity assays [5,71].

5.4.1. NanoparticlesUse of nanoparticle labels has proved to be particularly advan-tageous, due to the fact that single biorecognition eventsare typically translated into the specific accumulation ofa multitude of detectable atoms (or ions) at the electrodesurface. Gothelf’s group described a multiplexed metal sul-phide nanoparticle-based electrochemical detection schemethat provided sensitivities down to the femtomolar level [72].Semiconductor CdS, ZnS, and PbS nanoparticles were synthe-sised and then conjugated with distinct 5′-thiolated reportersequences. A competitive hybridisation assay was subse-quently adopted. The oligonucleotide-conjugated nanoparti-cles were first hybridised to specific capture probes arrangedon the surface of a gold electrode. Addition of a defined targetsequence (which acted as a competitor, being complemen-tary to one of the oligos used for labelling the nanoparticles)led to selective dissociation of the corresponding metal sul-phide from the surface. Transduction of the biorecognitionevent was finally accomplished by means of acidic dissolutionof the remaining nanoparticles and subsequent detection ofthe metal signals by anodic stripping voltammetry. Comparedto previously reported sandwiched assays, the competitiveassay described by Gothelf and co-workers was, in principle,reagentless. However, the long time required for competi-tion (5–6 h) discourages routinely use of such an analyticalprotocol.

5.4.2. EnzymesThe successful and widespread use of enzymes as labelsin affinity bioassays is also essentially due to their abilityto convert single hybridisation events into a multitude ofdetectable molecules. Enzymes of choice are usually alkalinephosphatase (AP), horseradish peroxidase (HRP) and oxidases(such as glucose oxidase [GOx]), all having the commonadvantages of being relatively stable, cheap and to possessgenerally high turnover rates. Such enzymes are typicallyused as avidin (or related proteins) conjugates and are, there-fore, coupled to the duplex (that must be biotinylated, forexample using a biotin-labelled signalling probe) by takingadvantage of the avidin–biotin affinity reaction. The mostcommon enzymes are commercialised as avidin, streptavidin,
neutravidin or extravidin conjugates. Hence, the selection ofthe most suited for a given assay must be done carefully.For example, when comparing avidin–alkaline phosphataseand streptavidin–alkaline phosphatase conjugates, Carpini et

t a 6

asc

grgborasaraea

jildmAbd

Aashrdawwwa

iLiwopsbLfoso

5Thct

a


l. [73] found the former to give much higher levels of non-pecific signals, due to the higher isoelectric point of avidinompared to streptavidin.

Besides these commercially available enzymatic conju-ates (which are exceptionally versatile, easy to use andelatively cheap), direct use of oligonucleotide–enzyme conju-ates has also been reported. The synthetic pathway describedy Dominguez et al. [38] for labelling of GOx with syntheticligonucleotides relied on the oxidation of the glycosidicesidues of the enzyme and their covalent binding with 5′-mino-terminated oligos. Upon use of such GOx-conjugatedignalling probes in a sandwich hybridisation assay andddition of glucose, the H2O2 generated by the enzymaticeaction readily diffused through a multilayer composed byn Os-based redox polymer and HRP sandwiched onto a goldlectrode. The resulting electrocatalytic current (evaluated bymperometry) was used as the analytical signal.

Possible drawbacks of these oligonucleotide–GOx con-ugates included lower hybridisation kinetics and labour-ntensive synthesis and purification. However, this way ofabelling the reporter probe coupled with use of different oxi-ases could represent a route for designing affinity-basedultidetection systems with no need for spatial resolution.lso, their potential for use as building blocks to assem-le more sensitive analytical devices will be subsequentlyescribed.

Metfies et al. reported the detection of the toxic algae. ostenfeldii, responsible for shellfish poisoning, using

digoxygenin/anti-digoxygenin–HRP conjugate labellingcheme [9]. Algal rRNA was detected by sandwich-ybridisation in the presence of a digoxygenin-labelledeporter oligonucleotide. Labelling with horseradish peroxi-ase was achieved using a HRP-conjugated anti-digoxigeninntibody. Catalysing the reduction of hydrogen peroxide toater, the enzyme was first oxidised. The reduced form of HRPas, however, readily regenerated by p-aminodiphenylamin,hich acted as a mediator. Reduction of the oxidised mediator

t −150 mV vs. Ag/AgCl provided the electrochemical readout.Miranda-Castro et al. exploited the features of a surface-

mmobilised hairpin probe for the selective identification ofegionella pneumophila sequences [13]. A thiolated hairpin,mmobilised onto a gold electrode surface, was combined

ith a sandwich-type hybridisation assay in the presencef a biotinylated signalling probe and streptavidin–alkalinehosphatase as the reporter enzyme. The sensor allowedelective discrimination between L. pneumophila and L. long-eachae (52-mer sequences) with a detection limit of 340 pM of. pneumophila DNA. Direct comparison of the analytical per-ormance of this assay with that of an analogous test basedn the use of linear oligonucleotide probes, clearly demon-trated the superior sensitivity and selectivity of the hairpinligonucleotide.

.4.3. Enzyme multilabelling (dendritic-like architectures)raditionally, a single enzyme label accounted for eachybridisation event. However, more sensitive detection con-
epts exploited the transduction of the hybridisation reactionhrough multi-labelling strategies.
The basic idea behind one of such approaches is to takedvantage from the considerable length of the real samples

0 9 ( 2 0 0 8 ) 139–159 153

(PCR products) to introduce multiple labels onto the targetsequence [35,74]. In Del Giallo et al. binding of more than onebiotinylated oligonucleotide per target sequence offered mul-tiple anchoring points for a streptavidin–alkaline phosphataseconjugate. Simultaneous use of two biotinylated probes led to+82% enhanced signals, while a +126% (relative to the caseof a single signalling probe) was observed when using threebiotin-labelled sequences. Such a non-linear growth of the sig-nal with the number of labels was explained as a consequenceof the steric hindrance of streptavidin–alkaline phosphatase,the footprint of which is in the order of 134 nm2. The stericand electrostatic interference from the conjugates bound toneighbouring duplexes was likely to further inhibit the simul-taneous binding of three enzymatic labels per target strand.

Each of the oligonucleotide-conjugated GOx labels devel-oped by Dominguez et al. [38] carried an average of threeoligos. The authors thus suggested that, through successiverounds of hybridisation and labelling, the resulting enzyme-rich architectures could provide a means to significantlyenhance the sensitivity of their genoassay. This attractive con-cept was, however, only partially demonstrated, hybridising abiotinylated sequence onto the primary GOx-oligonucleotidelabel and using an avidin-modified GOx as the secondary enzy-matic conjugate.

Mascini and co-workers [71] also described a genosens-ing scheme based on a dendritic-like signal amplificationpath. The analytical strategy relied on labelling the biotiny-lated hybrid obtained at the electrode surface by alternateexposure to streptavidin and biotinylated alkaline phos-phatase, thus self-assembling nanoarchitectures rich inenzyme labels. Compared to the commercially availablestreptavidin–alkaline phosphatase conjugates, a single gen-eration of the streptavidin–biotinylated alkaline phosphataseassembly readily allowed a 15–20-fold enhancement of theelectroanalytical signals. Interestingly, after the first bindingevent, each biotinylated alkaline phosphatase still possessedseveral free biotins that could efficiently cross-link a sec-ond “layer” of streptavidin. By sequentially repeating sucha self-assembling process, one could in principle generatedendritic-like nanoarchitectures with the desired content ofalkaline phosphatase labels. Indeed, the experimental resultsdemonstrated that the response increased linearly with thenumber of protein–enzyme generations, confirming that eventhe �-naphthol produced by alkaline phosphatase moleculesbound on the top of the network could easily diffuse throughthe protein layers becoming electrochemically accessible.

5.4.4. Ferrocene–streptavidin conjugatesKnoll and co-workers [75] employed a modified streptavidin(Strept) labelled with multiple ferrocene (Fc) units to amplifyelectrochemical and SPR response of short biotinylatedsequences. Interaction of the surface-immobilised PNA probeswith the biotinylated targets resulted in hybrid duplex for-mation. Subsequent labelling with the Fc-conjugated Strept,allowed SPR and electrochemical monitoring of the hybridisa-tion events. The amount of hybridised target was estimated
by cyclic voltammetry, chronocoulometry and square wavevoltammetry. Due to the long linker between the Fc groupsand the Strept (six glycine arms), electron transfer betweenthe metal complex and the underlying gold electrode occurred

a c t a
efficiently. Additionally, each Strept molecule carried at leastnine Fc moieties. As a result, a detection limit of 10 pM wasobserved for the 12-mer oligonucleotide target.

5.4.5. Polymeric osmium complexLiu and Anzai [76] developed a high performance redoxindicator based on a polymeric chain bearing numerous Oscomplexes. When interacted with the surface-immobilisedduplex, the poly(4-vinylpyridine) labelled with [Os(5,6-dmphen)2Cl]2+ exhibited 1000 times higher sensitivity thanthe monomeric analogue, [Os(5,6-dmphen)3]2+. This wasexplained as a probable consequence of the polymeric struc-ture of the indicator. Unlike monomeric labels, intercalationof the first Os complex into the double helix facilitated theinteraction of other complexes adjacent in the same polymericchain. Since about 120 Os residues were connected to a singlepolymer, the improved sensitivity allowed detection of as fewas 1 pM of a 25-mer synthetic target.

5.4.6. Oligonucleotide-loaded gold nanoparticles and[Ru(NH3)6]3+

Following the concepts based on the quantitation of the[Ru(NH3)6]3+ electrostatically associated to the anionic back-bone of DNA, Fan and co-workers [77] described a novelapproach which exploited the signal amplification feature ofgold nanoparticles (AuNPs). The DNA sensor was based on a“sandwich” detection strategy, which involved hybridisationof the target sequence between the surface-immobilised cap-ture probe and a reporter sequence-loaded gold nanoparticle.Hybridisation thus brought the AuNPs proximal to the elec-trode surface. Since a single nanoparticle carried hundredsof reporter oligonucleotides, the chronocoulometric interro-gation of [Ru(NH3)6]3+ allowed amplification of the targetsignal by two to three orders of magnitude. As a result, thegenosensor could detect as low as femtomolar concentra-tions of the target. Additionally, because of the sharp meltingprofiles of such nanoparticle-loaded sequences in low ionicstrength buffers, an excellent selectivity against single-basemismatches was demonstrated.

5.4.7. Enhancement of gold nanoparticles byelectrocatalytic deposition of silverYeung et al. demonstrated the multiplexed detection of E.coli and B. subtilis cells implementing sample preparation,DNA amplification and electrochemical detection in a single-microchamber device [25]. Detection of the asymmetric PCRamplification products was accomplished by labelling the tar-gets with streptavidin-conjugated gold nanoparticles (5 nm),electrocatalytic deposition of silver (from a AgNO3 solution)and chronopotentiometric quantitation of Ag at ITO elec-trodes. As few as 100 cells sample−1 could be detected usingsuch an integrated DNA biochip.

5.4.8. Bio-metallizationHwang et al. [30] demonstrated the amplified detectionof a target DNA through the bio-catalysed deposition of
silver (bio-metallization). In this method, the target DNAand a biotinylated detection probe hybridised to a 16-mer capture probe tethered onto a gold electrode. Uponbinding of a neutravidin-conjugated alkaline phosphatase
6 0 9 ( 2 0 0 8 ) 139–159

label, the non-electroactive substrate of the enzyme, p-aminophenyl phosphate, was converted into a reducing agent,p-aminophenol. The latter, in turn, reduced Ag+ ions alsopresent in the enzymatic substrate solution, thus leading todeposition of the metal onto the electrode surface and theDNA backbone. Linear sweep voltammetry was finally usedfor stripping the accumulated silver, with the correspondingcurrent taken as the analytical signal.

In contrast to previously developed methods where thesensitivity depended on the size of the metal nanoparticletracer, the sensitivity of this protocol essentially depended onthe condition of the enzymatic reaction only. As a result, thestripping signal of the metal was amplified by the enzymeuntil alkaline phosphatase was completely buried into theprecipitated silver. The bio-metallization protocol also tookadvantage from detecting the signal of a compound whichdid not diffuse away from the electrode surface but, rather,continuously accumulated onto the sensing layer as the reac-tion proceeded. This analytical strategy, coupled with carefuldesign of the capture oligonucleotide, allowed sensitive andselective detection of as few as 100 aM (10 zmol) of the 31-mer perfectly matched target, while efficiently discriminatinga single A·C mismatch.

5.4.9. Streptavidin-conjugated nanoparticles for amplifiedpiezoelectric sensingMao et al. [78] described a QCM biosensing strategy basedon the use of streptavidin-conjugated Fe3O4 nanoparticleslabels. A thiolated probe specific for E. coli 0157:H7 eaeA genewas hybridised with the biotinylated target (both a syntheticoligonucleotide and a PCR product). The amplification featuresof the streptavidin-conjugated nanoparticles (average diam-eter = 145 nm) allowed detection of as low as 2.7 × 102 colonyforming units mL−1 of E. coli 0157:H7 cells.

5.5. Commercially available tools

In the past few years the most promising electrochemicaldetection concepts have also been implemented into commer-cially available genosensing platforms. Their features will bediscussed in the following paragraphs. While some of theseproducts were not successful and rapidly disappeared, someother are likely to play a significant role in the future moleculardiagnostics market.

5.5.1. Electrocatalytic oxidation of guanineOxidation of guanine residues using Ru(bpy)32+ as the medi-ator provided a sensitive and simple method for detectingunmodified nucleic acids at ITO electrodes. Originally devel-oped by Armistead and Thorp [23], this concept involveduse of immobilised synthetic oligonucleotide probes in whichguanine was substituted with the non-physiological baseinosine, in order to minimize the background signal. Afterhybridisation of the sample and an appropriate washingstep, ruthenium tris(2,2′-bipyridine) was added and cyclicvoltammetric scans performed. In the absence of surface-
confined guanines (i.e., when the capture oligonucleotide didnot find its complementary sequence in solution), each medi-ator molecule was oxidised only once. On the contrary, whenthe guanine-containing target was hybridised at the surface,

a n a l y t i c a c h i m i c a a c t a 6 0 9 ( 2 0 0 8 ) 139–159 155

Fig. 6 – Electrocatalytic oxidation of guanine by means of Ru(bpy)32+: (A) simple reaction in the presence of thei presf

taRd

etAaahb

5Trdsws

Fdlfb

nosine-modified probe only; (B) electrocatalytic cycle in therom Ref. [23].

he oxidised form of the metal complex was able to abstractn electron from such purine moieties, thus regenerating theu(II) form. The regenerated reduced mediator was again oxi-ised at the electrode, thus completing a catalytic cycle (Fig. 6).

This technology, implemented in an array of 96 workinglectrodes (200-�m diameter) at Xanthon Inc., was supposedo be commercialised with the name of Xanthon Xpressionnalysis System. Unfortunately, for a number of economicalnd technical reasons (including difficulties in manufacturingnd modifying the ITO chips by the DNA probes) the companyas not survived. The technology was subsequently acquiredy Motorola.

.5.2. DNA-mediated charge transporthe ability of the DNA duplex to conduct electrical cur-ents was first exploited by Barton and co-workers [79], who
eveloped an elegant electrochemical assay format to detectingle-nucleotide polymorphisms. Chips of gold electrodesere used as platform for the covalent immobilisation of
ynthetic oligonucleotide probes corresponding to mutational

ig. 7 – Exploiting DNA-mediated charge transport for the detectuplex. When applying a negative potential, the intercalated MB

eucomethylene blue (LB). MB was then re-generated through theerricyanide. (B) Mismatched duplex. Any mismatch located witheing the first electro-reduction step impeded. Adapted from Ref

ence of the hybridised (guanine-containing) target. Adapted

“hot spots”. After hybridisation with clinically relevant sam-ples, the redox-active intercalator methylene blue (MB), wasinteracted with the surface-tethered hybrids along with ferri-cyanide. Upon applying a negative potential, the intercalatedMB underwent electrochemical reduction. The original formof the intercalator was then re-generated (by-oxidation) bythe solution-phase ferricyanide, thus entering a catalyticcycle (Fig. 7). Interestingly, a mismatch located within theimmobilised duplexes caused a striking decrease in theelectrochemical accessibility of the redox-active intercalator.Mismatch detection could be accomplished irrespectively ofDNA sequence and mismatch identity, being depended onelectronic coupling within the base pair stack, rather than thethermodynamics of base pairing.

GeneOhm Sciences Inc. implemented such an elegant prin-ciple of detection as the core technology of a microchip-based
platform (GeneOhm’s ePlexTM electrochemical array). How-ever, according to our recent visit to GeneOhm’s website(www.bd.com/geneohm/english/), electrochemical genosens-ing is not in the products list anymore.
ion of single-base mismatches. (A) Perfectly matchedunderwent electrochemical reduction, generatingchemical oxidation of LB by the solution-phase

in the immobilised duplexes inhibited the whole process,. [79].

http://www.bd.com/geneohm/english/

a c t a
5.5.3. Redox cycling the product of an enzymatic reaction(eMicroLISA)Besides simple enzymatic, dendritic and nanoparticle-basedamplification routes, further amplification of the analyti-cal signal can be obtained by pure electrochemistry, e.g.,redox cycling the product of an enzymatic reaction (Fig. 4).Based on this concept, Elsholz et al. demonstrated thespecies-specific identification and quantitation of 16S rRNAsof five pathogens using a low-density electrical microarray[35]. Recognition of the rRNA samples was accomplishedthrough a sandwich hybridisation scheme in the pres-ence of biotinylated signalling probes, followed by labellingof the surface-confined hybrid with an extravidin–alkalinephosphatase conjugate. The enzymatic hydrolysis of theelectro-inactive p-aminophenyl phosphate (p-APP) producedp-aminophenol (p-AP) which was detected by amperometryusing a 16 channels multipotentiostat. Redox recycling of thep-AP product was obtained polarising the cathodic fingers ofthe interdigitated array of electrodes at −150 mV (vs. Ag/AgCl)while the anodic ones were polarised at +350 mV. The smartdesign of the interlocking comb-like structures combined theadvantageous features of ultramicroelectrodes (which expe-rience enhanced lateral diffusion compared to macroscopicelectrodes) with an efficient redox regeneration of the elec-troactive product (which underwent oxidation and reductionmore than 10 times on average). A detection limit of 0.5 ng �L−1

of unamplified total RNA from E. coli was determined using a15 min hybridisation time.

This technology, also adapted for the identification andquantitation of proteins and haptens, is commercialised byeBiochip Systems GmbH (Germany).

5.5.4. Minor groove binder (GenelyzerTM)Another electrochemical DNA chip, namely the GenelyzerTM

(originally developed by Toshiba) is now commercialised byAntara BioSciences Inc. (USA). On GenelyzerTM transductionof the hybridisation events relied on use of Hoechst 33258, anelectrochemically active dye that specifically bound the minorgroove of double-stranded DNAs. An example of this tech-nology is offered by the paper of Goto et al. [80]. Goto et al.employed a microfabricated disposable chip consisting of 40probe-modified gold working electrodes to trace the presenceof several bacterial and viral strains. The corresponding DNAsamples were derived from the 16S rRNA of each strain afterreverse transcription to cDNA, a primary PCR amplificationand a secondary (nested) PCR. Upon hybridisation, the sur-face was then exposed to Hoechst 33258, whose anodic signalwas measured by cyclic voltammetry. Hybridisation reaction(at controlled temperature), washing step, indicator accumu-lation and electrochemical detection were all performed usinga proprietary DNA detection system.

Since this method relied on the measurement of a dyewhich accumulated into the minor groove of the duplex, a par-ticular strategy was chosen for allowing interaction betweenthe surface-immobilised probes and the amplicons whilemaximising the length of the double helix. On one hand,
the second pair of primers used for nested PCR includedconventional and phosphorothioate (exonuclease-resistant)nucleotides. On the other hand, the probe sequence wasdesigned to largely overlap the forward primer. Hence, the
6 0 9 ( 2 0 0 8 ) 139–159

samples were not thermally or chemically denatured, butrather, selectively digested with T7 Gene6 exonuclease, whichcreated a discrete single-stranded site where the surface-immobilised capture probes could accommodate.

5.5.5. Ferrocene-labelled signalling probes (CMSeSensorTM)In 2001 Motorola reported on CMS eSensorTM chips, low-density arrays for the bioelectronic detection of DNAsequences. This technology, now acquired by Osmetech (USA),rests on the generation of an electrochemical signal throughthe reversible oxidation and reduction of a ferrocene complex,used as an electrochemical label for the nucleic acid targets.

eSensorTM chips have up to 36 individually addressablegold electrodes affixed to printed circuit boards. The elec-trodes are modified with a mixed monolayer of thiol-tetheredcapture probes (25-mer) and alkylthiol molecules (Fig. 8). Sucha modification of the electrode surfaces not only inhibitsany non-specific binding but also blocks electrochemical sig-nals from both unbound redox labels and extraneous redoxcompounds. Bioelectronic detection proceeds via a sand-wich hybridisation assay where the target is simultaneouslybound by the capture probes on the electrode surface andthe signalling probes labelled with ferrocene derivatives. Thesignalling probes are designed to be complementary to a por-tion of the target different from, but adjacent to, the regionbound by the capture probe. When the target hybridises toboth capture and signalling probes, the ferrocene moieties arebrought into sufficient proximity to the electrode surface, thusbecoming electrochemically accessible for detection. Use of acvoltammetry for interrogating the sensors allows for repeatedcollection of the electrons from the ferrocene labels. Hence,this detection method is not truly electrocatalytic, but doeshave a built-in signal amplification that results from the inter-rogation step [81]. The generated current is detected usingan electronic detection platform called the eSensorTM DNADetection System, which can analyze 36 electrodes at a time.

5.5.6. HRP enzyme label (CombiMatrix ElectraSenseTM)A major breakthrough in electrochemical DNA biosensorshas been achieved by CombiMatrix (USA) which developedarrays of 12,544 individually addressable microelectrodes anddemonstrated their use for genotyping and gene expressionanalysis [29]. Thousand of different oligonucleotide probes,immobilised onto porous membranes, specifically capturedtheir biotinylated targets from the hybridisation solution.After labelling of the hybrids using an extravidin–horseradishperoxidase conjugate, the enzyme catalysed the oxidationof 3,3′,5,5′-tetramethylbenzidine (TMB). Due to the closeproximity of the HRP label to the Pt microelectrode sur-face, the oxidised TMB was readily reduced, thus generatingthe electrical output. All 12,544 electrodes were sequen-tially interrogated in approximately 25 s using a specificdevice (ElectraSenseTM Reader). Such a fast reading time(<2 ms electrode−1) ensured that HRP retained a high level ofactivity throughout measuring of the whole microarray. The
performance of the ElectraSenseTM platform was also com-pared to that of the standard fluorescent detection. Goodconsistency (and similar detection limits, about 1 pM) wasobserved between these techniques. However, the opera-

a n a l y t i c a c h i m i c a a c t a 6 0 9 ( 2 0 0 8 ) 139–159 157

F MS e

tfftb1

6

Tbsdmt

snamfeeysotIcaa

r

ig. 8 – Molecular architecture and strategy of detection of C

ional convenience, the lower cost and the high potentialor further improvements make the ElectraSenseTM platformavoured over the conventional fluorescent detection. Notably,he authors suggested that the density of spots onto the Com-iMatrix chip could be realistically increased from the actual7,778 up to 1.5 × 106 electrodes cm−2.

. Conclusions and future prospects

he rapid progresses of DNA-based biosensors highlightedy the elegant sensing concepts summarised by this review,uggest the major impact these technologies may have uponetection of pathogens, genetic mutations and targets of phar-acogenomic and industrial interest (e.g., transgenic crops) in

he near future.Electrochemical and piezoelectric methods thus emerge as

imple, accurate and inexpensive means for a range of diag-ostic applications. Electrochemical genosensing schemesppear particularly suitable to accomplish this task. Suchethodologies have demonstrated the potential to allow

or very low detection limits when all of the analyte isfficiently delivered to the electrode surface and the het-rogeneous hybridisation reaction proceeds with reasonableields. Having established that detection cannot presently pre-cind from prior PCR amplification of the sample (becausef the low abundance and extreme complexity of most ofhe unamplified targets), other important hurdles remain.
mproved probe designs and sample pre-treatments are bothrucial for allowing efficient biorecognition events to occurt the transducer–solution interface. Use of synthetic DNAnalogues like PNA or LNA oligonucleotides will probably
SensorTM chips (Osmetech [USA]). Adapted from Ref. [81].

lead to unprecedented levels of selectivity, crucial for a reli-able screening of clinically relevant SNPs and closely relatedpathogenic species. Systematic use of helper oligonucleotidesis also likely to provide a general and inexpensive meansto minimise any interference associated to thermodynami-cally stable secondary structures of the target sequence. Theexamples of “active” (electrochemically driven) hybridisationalso provide an alternative route to significantly enhance thehybridisation rates and yields, along with the selectivity ofdetection.

Of great significance, the outstanding technologicaladvance afforded by CombiMatrix chips have finally answeredthe question of whether electrochemical detection may rivalthe level of parallelisation of fluorescence-based arrays. Solv-ing the major engineering challenge of individually addressthousands of electrodes at a time, such electrical detec-tion scheme also possesses the potential to be miniaturisedbeyond the present limits imposed by optical readouts. This,in turn, could ultimately led to an increased number of DNAsequences that can be simultaneously interrogated. Hence,despite the versatility and advances made with optical meth-ods for nucleic acid detection, there is reason to believe thatelectrochemical methods may ultimately be preferable for dis-tributed use in the field or even in microarray applications.

e f e r e n c e s

[1] W. Vercoutere, M. Akeson, Curr. Opin. Chem. Biol. 6 (2002)816.

[2] T.G. Drummond, M.G. Hill, J.K. Barton, Nat. Biotechnol. 21(2003) 1192.

a c t a
[3] K. Kerman, M. Kobayashi, E. Tamiya, Meas. Sci. Technol. 15(2004) R1.

[4] R. Moeller, W. Fritzsche, IEEE Proc. Nanobiotechnol. 152(2005) 47.

[5] J. Wang, Biosens. Bioelectron. 21 (2006) 1887.[6] F. Lucarelli, G. Marrazza, A.P.F. Turner, M. Mascini, Biosens.

Bioelectron. 19 (2004) 515.[7] M. Minunni, S. Tombelli, M. Mascini, Springer Series on

Chemical Sensors and Biosensors 5 (Piezoelectric Sensors),Springer Berlin Heidelberg, Berlin, 2007, p. 211.

[8] http://www.affymetrix.com.[9] K. Metfies, S. Huljic, M. Lange, L.K. Medlin, Biosens.

Bioelectron. 20 (2005) 1349.[10] E.A. Barlaan, M. Sugimori, S. Furukawa, K. Takeuchi, J.

Biotechnol. 115 (2005) 11.[11] P. Abad-Valle, M.T. Fernandez-Abedul, A. Costa-Garcia,

Biosens. Bioelectron. 22 (2007) 1642.[12] D.M. Jenkins, B. Chami, M. Kreuzer, G. Presting, A.M. Alvarez,

B.Y. Liaw, Anal. Chem. 78 (2006) 2314.[13] R. Miranda-Castro, P. de-los-Santos-Alvarez, M.J.

Lobo-Castanon, A.J. Miranda-Ordieres, P. Tunon-Blanco,Anal. Chem. 79 (2007) 4050.

[14] F. Ricci, R.Y. Lai, A.J. Heeger, K.W. Plaxco, J.J. Sumner,Langmuir 23 (2007) 6827.

[15] K. Kerman, M. Vestergaard, N. Nagatani, Y. Takamura, E.Tamiya, Anal. Chem. 78 (2006) 2182.

[16] J. Liu, S. Tian, P.E. Nielsen, W. Knoll, Chem. Commun. (2005)2969.

[17] M. Steichen, Y. Decrem, E. Godfroid, C. Buess-Herman,Biosens. Bioelectron. 22 (2007) 2237.

[18] R. Corradini, G. Feriotto, S. Sforza, R. Marchelli, R. Gambari, J.Mol. Recognit. 17 (2004) 76.

[19] S. Daniel, T.P. Rao, K.S. Rao, S.U. Rani, G.R.K. Naidu, H.-Y. Lee,T. Kawai, Sens. Actuators B 122 (2007) 672.

[20] A.A. Gorodetsky, J.K. Barton, Langmuir 22 (2006) 7917.[21] D. Hernandez-Santos, M. Dıaz-Gonzalez, M.B.

Gonzalez-Garcıa, A. Costa-Garcia, Anal. Chem. 76 (2004)6887.

[22] H.F. Teh, H. Gong, X.-D. Dong, X. Zeng, A.L.K. Tan, X. Yang,S.N. Tan, Anal. Chim. Acta 551 (2005) 23.

[23] P.M. Armistead, H.H. Thorp, Anal. Chem. 72 (2000) 3764.[24] N.D. Popovich, A.E. Eckhardt, J.C. Mikulecky, M.E. Napier, R.S.

Thomas, Talanta 56 (2002) 821.[25] S.-W. Yeung, T.M.-H. Lee, H. Cai, I.-M. Hsing, Nucleic Acids

Res. 34 (2006) e118.[26] S. Ingebrandt, Y. Han, F. Nakamura, A. Poghossian, M.J.

Schoning, A. Offenhausser, Biosens. Bioelectron. 22 (2007)2834.

[27] Z. Gao, A. Agarwal, A.D. Trigg, N. Singh, C. Fang, C.-H. Tung,Y. Fan, K.D. Buddharaju, J. Kong, Anal. Chem. 79 (2007)3291.

[28] T.C. Hang, A. Guiseppi-Elie, Biosens. Bioelectron. 19 (2004)1537.

[29] A.L. Ghindilis, M.W. Smith, K.R. Schwarzkopf, K.M. Roth, K.Peyvan, S.B. Munro, M.J. Lodes, A.G. Stover, K. Bernards, K.Dill, A. McShea, Biosens. Bioelectron. 22 (2007) 1853.

[30] S. Hwang, E. Kim, J. Kwak, Anal. Chem. 77 (2005) 579.[31] F. Lucarelli, G. Marrazza, M. Mascini, Biosens. Bioelectron. 20

(2005) 2001.[32] C.E. Immoos, S.J. Lee, M.W. Grinstaff, J. Am. Chem. Soc. 126

(2004) 10814.[33] E.L.S. Wong, J.J. Gooding, Anal. Chem. 78 (2006) 2138.[34] T.M. Herne, M.J. Tarlov, J. Am. Chem. Soc. 119 (1997) 8916.[35] B. Elsholz, R. Worl, L. Blohm, J. Albers, H. Feucht, T.

Grunwald, B. Jurgen, T. Schweder, R. Hintsche, Anal. Chem.
78 (2006) 4794.
[36] G.-U. Flechsig, T. Reske, Anal. Chem. 79 (2007) 2125.[37] R.Y. Lai, S.-H. Lee, H.T. Soh, K.W. Plaxco, A.J. Heeger,

Langmuir 22 (2006) 1932.

6 0 9 ( 2 0 0 8 ) 139–159

[38] E. Dominguez, O. Rincon, A. Narvaez, Anal. Chem. 76 (2004)3132.

[39] F. Garnier, B. Bouabdallaoui, P. Srivastava, B. Mandrand, C.Chaix, Sens. Actuators B 123 (2007) 13.

[40] E.L.S. Wong, E. Chow, J.J. Gooding, Langmuir 21 (2005)6957.

[41] V.C.H. Wu, S. Chen, C. Lin, Biosens. Bioelectron. 22 (2007)2967.

[42] M.A. Cooper, Anal. Bioanal. Chem. 377 (2003) 834.[43] S. Tombelli, M. Minunni, M. Mascini, Methods 37 (2005) 48.[44] M. Minunni, S. Tombelli, R. Scielzi, I. Mannelli, M. Mascini, C.

Gaudiano, Anal. Chim. Acta 481 (2003) 55.[45] D. Dell’Atti, S. Tombelli, M. Minunni, M. Mascini, Biosens.

Bioelectron. 21 (2006) 1876.[46] D. Dell’Atti, M. Zavaglia, S. Tombelli, G. Bertacca, A.O.

Cavazzana, G. Bevilacqua, M. Minunni, M. Mascini, Clin.Chim. Acta 383 (2007) 140.

[47] M. Minunni, S. Tombelli, J. Fonti, M.M. Spiriti, M. Mascini, P.Bogani, M. Buiatti, J. Am. Chem. Soc. 127 (2005) 7966.

[48] S. Tombelli, M. Minunni, A. Santucci, M.M. Spiriti, M.Mascini, Talanta 68 (2006) 806.

[49] X. Su, Y. Wu, R. Robelek, W. Knoll, Langmuir 21 (2005) 348.[50] X. Su, R. Robelek, Y. Wu, G. Wang, W. Knoll, Anal. Chem. 76

(2004) 489.[51] C. Larsson, M. Rodahl, F. Hook, Anal. Chem. 75 (2003) 5080.[52] T.L. Mann, U.J. Krull, Biosens. Bioelectron. 20 (2004) 945.[53] F. Patolsky, A. Lichtenstein, I. Willner, Nat. Biotechnol. 19

(2001) 253.[54] M. Minunni, I. Mannelli, M.M. Spiriti, S. Tombelli, M. Mascini,

Anal. Chim. Acta 526 (2004) 19.[55] M. Minunni, S. Tombelli, J. Fonti, M.M. Spiriti, M. Mascini, P.

Bogani, M. Buiatti, JACS 127 (2005) 7966.[56] R.B. Towery, N.C. Fawcett, P. Zhang, J.A. Evans, Biosens.

Bioelectron. 16 (2001) 1.[57] M. Mascini, M. Del Carlo, M. Minunni, B. Chen, D.

Compagnone, Bioelectrochemistry 67 (2005) 163.[58] U.B. Gyllensten, H.A. Erlich, Proc. Natl. Acad. Sci. U.S.A. 85

(1988) 7652.[59] M. Takahashi, J. Okada, K. Ito, M. Hashimoto, K. Hashimoto,

Y. Yoshida, Y. Furuichi, Y. Ohta, S. Mishiroc, N. Gemma,Analyst 130 (2005) 687.

[60] M. Wojciechowski, R. Sundseth, M. Moreno, R. Henkens,Clin. Chem. 45 (1999) 1690.

[61] R. Wang, M. Minunni, S. Tombelli, M. Mascini, Biosens.Bioelectron. 20 (2004) 598.

[62] Y. Cho, S. Kim, Y.A. Kim, H.K. Lim, K. Lee, D. Yoon,G. Lim, Y.E. Pak, T.H. Ha, K. Kim, J. Colloid Interf. Sci. 278(2004) 44.

[63] F. Hook, B. Kasemo, Springer Series on Chemical Sensorsand Biosensors 5 (Piezoelectric Sensors), Springer BerlinHeidelberg, Berlin, 2007, p. 425.

[64] A. Graneli, J. Rydstrom, B. Kasemo, F. Hook, Langmuir 19(2003) 842.

[65] F. Hook, M. Rodahl, B. Kasemo, P. Brezezinski, Proc. Natl.Acad. Sci. U.S.A. 95 (1998) 12271.

[66] F. Hook, A. Ray, B. Norden, B. Kasemo, Langmuir 17 (2001)8305.

[67] T. Ozeki, M. Morita, H. Yoshimine, H. Furusawa, Y. Okahata,Anal. Chem. 79 (2007) 79.

[68] K.M. Millan, S.R. Mikkelsen, Anal. Chem. 65 (1993) 2317.[69] N.C. Tansil, H. Xie, F. Xie, Z. Gao, Anal. Chem. 77 (2005) 126.[70] Z.-S. Wu, J.-H. Jiang, G.-L. Shen, R.-Q. Yu, Hum. Mutat. 28

(2007) 630.[71] F. Lucarelli, G. Marrazza, M. Mascini, Langmuir 22 (2006)

4305.
[72] J.A. Hansen, R. Mukhopadhyay, J.Ø. Hansen, K.V. Gothelf, J.
Am. Chem. Soc. 128 (2006) 3860.[73] G. Carpini, F. Lucarelli, G. Marrazza, M. Mascini, Biosens.

Bioelectron. 20 (2004) 167.

http://www.affymetrix.com/

t a 6

S. Hongo, N. Gemma, N. Hayashimoto, T. Itoh, A. Takakura, J.Microbiol. Methods 69 (2007) 93.


[74] M.L. Del Giallo, F. Lucarelli, E. Cosulich, E. Pistarino, B.Santamaria, G. Marrazza, M. Mascini, Anal. Chem. 77 (2005)6324.

[75] J. Liu, S. Tian, L. Tiefenauer, P.E. Nielsen, W. Knoll, Anal.Chem. 77 (2005) 2756.

[76] A. Liu, J. Anzai, Anal. Chem. 76 (2004) 2975.[77] J. Zhang, S. Song, L. Zhang, L. Wang, H. Wu, D. Pan, C. Fan, J.

Am. Chem. Soc. 128 (2006) 8575.[78] X. Mao, L. Yang, X. Su, Y. Li, Biosens. Bioelectron. 21 (2006)

1178.

0 9 ( 2 0 0 8 ) 139–159 159

[79] E.M. Boon, D.M. Ceres, T.G. Drummond, M.G. Hill, J.K. Barton,Nat. Biotechnol. 18 (2000) 1096.

[80] K. Goto, H. Horiuchi, H. Shinohara, K. Motegi, K. Hashimoto,

[81] C.J. Yu, Y. Wan, H. Yowanto, J. Li, C. Tao, M.D. James,C.L. Tan, G.F. Blackburn, T.J. Maede, J. Am. Chem. Soc. 123(2001) 11155.

electrochemical and piezoelectric dna biosensors for hybridisation detection

Documents

ito electrodes

platinum electrodes

gold electrodes

analyt ica ch im ica

t of chemistry

specic probe

shortthe probe

covalent attachment