Philosophy

A fully electronic sensor for the measurement of cDNA hybridization kinetics

Description
A fully electronic sensor for the measurement of cDNA hybridization kinetics
Categories
Published
of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Biosensors and Bioelectronics 22 (2007) 2108–2114 A fully electronic sensor for the measurement of cDNAhybridization kinetics L. Bandiera a , ∗ , G. Cellere a , S. Cagnin b , A. De Toni a ,E. Zanoni a , G. Lanfranchi b , L. Lorenzelli c a  DEI, Department of Information Engineering, Padova University, Italy b CRIBI Biotechnology Center, Padova University, Italy c  ITC-IRST Microsystems Division, Trento, Italy Received 24 April 2006; received in revised form 8 September 2006; accepted 19 September 2006Available online 7 November 2006 Abstract Ionsensitivefieldeffecttransistors(ISFET)arecandidatesforanewgenerationoffullyelectricalDNAsensors.Tothispurpose,wehavemodifiedISFET sensors by adsorbing on their Si 3 N 4  surface poly- l -lysine and single (as well as double) stranded DNA. Once coupled to an accurate modelof the oppositely charged layers adsorbed on the surface, the proposed sensor allows quantitatively evaluating the adsorbed molecules densities,as well as estimating DNA hybridization kinetics.© 2006 Published by Elsevier B.V. Keywords:  DNA sensors; Silicon nitride; Hybridization kinetic; dsDNA; cDNA 1. Introduction Surface-based methods are promising alternatives for thedetection of biological molecules such as DNA and proteins(Schena et al., 1995; Han et al., 2006). In particular, micro-electronic devices can play a major role in this arena, thanksto extremely high repeatability and process control availableby using even relatively old technologies. One of the mostinteresting properties of this class of sensor is the ability todetect unlabelled sample molecules, thus avoiding bias intro-duced by enzymatic labelling used, for example, in microarraytechnology. During the last years, several fully electronic andlabel-free DNA sensors have been developed. Some sensors arebasedonthemodificationofredoxreactionduringhybridization(Ye and Ju, 2003), while others are based on variation of doublelayer capacitance after hybridization, on variation of surfacepotential (such as ISFET based sensors, e.g. Pouthas et al.,2004; Kim et al., 2004), and/or on the detection of the intrinsiccharge of molecules adsorbed onto the sensors surface (Fritz etal., 2004; Uslu et al., 2004; Sakata et al., 2005; Purushothaman ∗ Corresponding author. Tel.: +39 049 8277664; fax: +39 049 8277699.  E-mail address:  leonardo.bandiera@gmail.com (L. Bandiera). et al., 2006). In literature, ISFET structures have been widelyreported for several types of sensors, ranging form the moreconventionalforthesedevicespHsensors(Bergveldetal.,1998;Chin et al., 2001; Grattarolla and Massobrio, 2002; Martinoiaand Massobrio, 2000; Martinoia et al., 2001), or biochemicalsensors (Lauwers et al., 2001), to more exotic ones, such assensors for detection of micro-organisms in water (Cambiasoet al., 1996), for cell population measurements (Martinoia etal., 2001), and so on. The ISFET sensitive layer is here made of silicon nitride (Si 3 N 4 ) (Siu and Cobbold, 1979). The majorityof studies in literature have considered SiO 2  (Pouthas et al.,2004) or Au (Kim et al., 2004) surfaces as interfaces between silicon-based devices and DNA. Si 3 N 4  /electrolyte interfacefeatures not only silanol groups (Yates et al., 1974; Siu andCobbold, 1979) (typical of SiO 2  surface) but also amino groups(Grattarolla and Massobrio, 2002; Martinoia and Massobrio,2000) that play an important role in the control of the chargeat sensor/electrolyte interface. Another important advantageof ISFET is the possibility to evaluate the DNA charge bymeasuring a simple parameter such as the variation in the tran-sistor threshold voltage  V  th . In this work we are experimentallystudyingtheadsorptionofpoly- l -lysine(PLL),doublestrandedDNA (dsDNA), single stranded DNA (ssDNA) onto Si 3 N 4 sensitive area of ISFETs and the hybridization kinetic of a DNA 0956-5663/$ – see front matter © 2006 Published by Elsevier B.V.doi:10.1016/j.bios.2006.09.025   L. Bandiera et al. / Biosensors and Bioelectronics 22 (2007) 2108–2114  2109 target to a probe linked to the sensitive surface. The electricalcharge associated with these molecules can be evaluated byusing an accurate model based on the site binding theory of Si 3 N 4  (Cambiaso et al., 1996; Grattarolla and Massobrio,2002; Martinoia and Massobrio, 2000; Martinoia et al., 2001).Differently from other proposed models (e.g. Landheer et al.,2005; Erickson et al., 2003) this one can be easily includedin spice simulator. We therefore propose the surface-modifiedISFET as the base for a fully electronic DNA hybridizationsensor. 2. Materials and methods 2.1. Experimental devices All the ISFET sensors used throughout this work are man-ufactured with a well-established technology (Cambiaso etal., 1996; Martinoia and Massobrio, 2000; Martinoia et al.,2001) by ITC-IRST (Trento, Italy). The ion sensitive fieldeffect transistors/complementary metal nitride oxide semicon-ductor (ISFET/CMNOS) technology consists in a modified4  m CMOS process with Al gate, realized on p-well substrate.In particular, a SiO 2  layer is at first grown on silicon surface,over which a stoichiometric Si 3 N 4  is deposited at 775 ◦ C in aLPCVD system. The stack of the two materials is thus used as agate dielectric, but only the nitride is actually used as the chemi-calsensinglayer.TheISFETchannellengthandwidthare18and804  m, respectively, and the equivalent oxide thickness, withrespecttoSiO 2 ,is84nm.TheISFETsensitivityrangesbetween20and58mVperpHunits,itslong-termdriftis600  V/h(mea-suresnotshownhere),andthetemperaturedependenceisaround1.5mV/K at constant pH.The exposed Si 3 N 4  layer is initially covered by PLL, thenthe sensor is dipped in the electrolyte solution (see Fig. 1a).The external gate voltage applied between the electrolyteand the common source contact with an Ag/AgCl electrode(Crison, 5242) (Lautor et al., 2000) controls the currentbetween drain and source contacts, similarly to what happensfor a conventional MOSFET. The trans-characteristics of the ISFET/PLL and ISFET/PLL/DNA systems are studieswith a semiconductor parameter analyzer HP4156A to applya variable voltage  V  gs  between the Ag/AgCl reference andsource/bulk contact, a constant  V  ds  between the drain andthe source contacts, and to measure the drain current  I  ds . Alldevices are characterized with  V  gs  ranging between 0 and3V while  V  ds  is kept at 100 or 300mV. All measurementswere carried out in a temperature-controlled bath ( T  =300K)to ensure the same environmental conditions for all theexperiments. 2.2. Experimental procedures Several steps need to be carried out for each sensor inorder to detect the DNA charge. After each of them, the trans-characteristics (  I  ds – V  gs  measurements) of the ISFET are mea-sured to control the evolution of charge distributions on thesurface. In detail: Fig. 1. (a) Sketch of n-channel ISFET with its active gate area covered by bio-logical molecules, those are PLL, ds/ss DNA. (b) Schematic representation of thechargedistributionassumedintheone-dimensionalmodeloftheISFET/PLLlayer/electrolyte system, and the equivalent electric circuit of the structure. Theinterfaces are schematized by equivalent double layer capacitance. (i) Step“0” issurface cleaning.First of all,devicesare cleanedwith HF (2%), then with a wash solution (280ml of 6.25MNaOH with 420ml of 95% ethanol) for 1h in an orbitalshaker. The sensors are then rinsed by plunging in ddH 2 Ofive times (using fresh H 2 O for each rinsing), and finallydried by centrifuging for 1min at 100rpm. At this point,they are measured first time: this is what we will call a“fresh” device.(ii) Step “1” is the PLL-adsorption: the sensitive gate areais functionalized with 0.2mg/ml PLL (pH 7, PLL-hydrobromide, MW 70,000–150,000, Sigma), rinsed withddH 2 O, and dried for 5min at 45 ◦ C. The sensors are thenstored at room temperature in a dust-free rack.(iii) After PLL coating, the ISFET surface is suitable for step“2”, that is, the adsorption and immobilization of the unla-belled dsDNA or ssDNA probe. The sensors are soaked for30min into a solution containing 400base pair (bp) longdsDNA amplified by a PCR reaction or with a solution  2110  L. Bandiera et al. / Biosensors and Bioelectronics 22 (2007) 2108–2114 containing20baseslongssDNAHPLCpurifiedsuppliedbyInvitrogen. The PCR products have been purified using theGenEluite TM PCR Clean-up Kit (Sigma–Aldrich). Purifi-cation enables to remove PCR components (dNTP s , buffer,Taq polymerase) that interfere with the DNA immobiliza-tion onto the sensors surface. The adsorbed dsDNA orssDNA is then immobilized by using UV cross-linking(Stratagene ® UV Stratalinker) at 60mJ intensity. UV expo-sure allows some covalent bonds to be formed betweenthe surface-adsorbed PLL and the DNA. In the microarrayapplicationstheaminosilaneandpoly- l -lysineslidesurfaceprovides available amine groups for initial ionic attachmentof the negatively charged phosphate groups in the DNAbackbone. The DNA can subsequently be attached cova-lently to the slide by either baking or by UV irradiation(Wang et al., 2003). After DNA immobilization devices arerinsedwithwaterforatleast2mintoremovenon-covalentlybonded DNA.(iv) Step “3” is the denaturation of dsDNA into ssDNA, bymeans of brief soaking into hot water. We performed thisstepontoISFETfunctionalizedwith20baseslongtoo.Thispermits to have the same conditions with all probe types(dsDNA and ssDNA), but in this latter case the denaturisingstep is not necessary because the oligonucleotide is alreadyin single strand.(v) The prepared sensors obtained with this process are thenusedtodetectthepresenceofcomplementarystrandtoDNAprobe in a low ionic concentration solution. Low ionic con-centrationpermitstoavoidnonspecifictargetbindinggivingto the hybridization reaction the necessary stringency. Thiscould have been done using denaturizing solutions (for-mamide) adding difficulties in the electric measurements. 3. Results and discussions 3.1. A model for the ISFET interface AschematicrepresentationoftheISFETsystemisillustratedin Fig. 1a. The ISFET operation is very similar to that of con-ventionalMOSFET,exceptthatareferenceelectrode/electrolytesolution gate is used instead of metal or polysilicon gate. Theelectrical equivalent circuit and the relationships between volt-ages and currents are rather similar for the two device types.In detail, when a voltage is applied by the Ag/AgCl refer-ence electrode through the electrolyte solution the followingequation holds true to describe the drain-to-source current fornon-saturated region (Grattarolla and Massobrio, 2002): I  ds  = µC ox W L [( V  gs − V  th ) V  ds − V  2ds / 2](1)where  µ  is the electron mobility,  C  ox  the oxide capacitance,  W  and  L thechannelwidthandlength,respectively, V  gs  thegate-to-source voltage,  V  th  the threshold voltage, and finally  V  ds  is thedrain-to-source voltage. The major difference between ISFETand MOSFET is their threshold voltage ( V  th ) expression. Whendealing with an ISFET,  V  th  contains additional terms account-ing for the interface between the liquid and the gate dielectric(Grattarolla and Massobrio, 2002). In the case of a pure salinesolution as electrolyte, without any molecular adsorption onthe surface, the Si 3 N 4  /electrolyte interface can be describedby assuming that only silanol and basic primary amine sitesplay a role in the surface/electrolyte equilibrium. Under thishypothesis the system can be described by the site-binding the-ory (Grattarolla and Massobrio, 2002; Yates et al., 1974) andthe potential drop  φ ei  per unit area at the electrolyte/insulatorinterface can be written as: φ ei  = qC eq [ N  sil f  a ( φ ei , pH) + N  nit f  b ( φ ei , pH)] (2)where  f  a ( φ ei , pH) and  f  b ( φ ei , pH) are suitable setting functionswhose complete expressions can be found in Grattarolla andMassobrio (2002).  N  sil  and  N  nit  are the surface densities of thesilanolandoftheprimaryaminesites,respectively, C  eq  thetotalcapacitance of the insulator/electrolyte interface per area, and  q is the elementary charge. Eq. (3) states that the potential  φ ei  canbe modelled as a non-linear function of pH (and of   φ ei  itself).After adsorption of macromolecules on the ISFET sensitivearea,thepotentialattheinsulator/solutioninterfacechangesandinduces a  V  th  shift. Thus, the threshold voltage can be writtenas: V  th (ISFET , mol) = V  th (MOSFET) + E ref  + φ lj + χ e − ϕ ei , mol − ϕ mol − φ m q (3)where V  th (MOSFET)isthethresholdvoltageoftheMOSequiv-alent system with a metal gate instead of the electrolyte,  E  ref  the potential of the reference electrode,  φ lj  the liquid junctionpotential between the reference solution and the electrolyte,  χ e the surface dipole potential,  ϕ ei,mol  represents the potential of Si 3 N 4  /solution interface after adsorption, and  ϕ mol  takes intoaccount the potential drop across the adsorbed molecule layer.Fig. 1b sketches the resulting charge distribution:  ρ mos  is thecharge density in the semiconductor,  ρ s  the charge density atthe electrolyte/insulator interface, and ρ pll  is the positive chargedue to PLL monomers, assumed to be constant and located at adistance  d  pll  from the ISFET surface. The charge neutrality of the structure leads to the following expression for the surfacecharge densities (Grattarolla and Massobrio, 2002): ρ s + ρ mos + ρ mol + ρ d,mol + ρ d,s  = 0 (4)where  ρ d,s  and  ρ d,mol  are the charge densities in Sternand Gouy-Chapmann layers of Si 3 N 4  /electrolyte and elec-trolyte/PLL/electrolyte interfaces, respectively (these are notshowninFig.1bforclarity).WithreferencetoFig.1b,thedouble layer capacitances ( C  dl ) take into account the electrolyte/Si 3 N 4 and electrolyte/PLL plane interfaces. As a simplified condition,weassumethatthepotentiallinearlydropsinsidetheelectrolytelayer separating the PLL/DNA virtual plane from the sensorsurface.TheBoltzmanndistributionpredictsthattheprotoncon-centration at the insulator surface is related to its concentration just outside the adsorbed molecular layer. Hence, the pH at theinsulatorsurfacedependsonthepotentialthroughthePLL/DNA   L. Bandiera et al. / Biosensors and Bioelectronics 22 (2007) 2108–2114  2111 layer following (Grattarolla and Massobrio, 2002):pH ohp1 − 2  = pH ext + ϕ mol q 2 . 3 kT  (5)where pH ohp1 − 2  is the pH between the double layers atthe Si 3 N 4  /electrolyte and PLL/DNA/electrolyte interfaces andpH ext  is the pH on the outermost surface of the PLL layer. Byusing Eqs. (2)–(4), we can evaluate the actual potential drop across the PLL/DNA layer  φ mol , the potential across the insu-lator/electrolyte interface  φ ei,mol , and the corresponding chargedistribution. 3.2. The PLL adsorbed layer  The above described model holds true under the hypothesisthat the adsorbed molecules form a uniform layer over the gatearea.Forthisreason,itisimportanttounderstandifandhowtheadsorbed molecules physically (other than chemically) modifythe ISFET sensitive area.Atomic Force Microscopy (AFM) was used to characterizethe ISFET surface after PLL adsorption to better understand themorphological characteristics of the PLL layer. Fig. 2a showsan AFM height mode image obtained for PLL film on the Si 3 N 4 surface, prepared following the procedure described in Section3. The image was taken in air to avoid the distortion inducedby the strong attraction between the film and the AFM tip whenimaging the film in liquid. The image lateral dimensions are5  m and the maximum  z -range is 18nm. After adsorption, theself-assembled monolayer of PLL was locally ablated by a laserpulsetoproduceanabruptstepalongwhichthethicknesscanbe Fig. 2. (a) AFM height mode images in air obtained for PLL films built at pH7. The hole in the AFM image corresponds to the laser ablated silicon. (b) Theprofile drawn by AFM image along the hole. The thickness of PLL layer isestimated as 12nm. estimated by AFM measurements. Fig. 2a shows the PLL layerafter laser treatment: the black “crater” in the AFM image cor-responds to the laser-ablated Si 3 N 4  region. Note that the depthof the crater is lower than the gate stack thickness, so that theunderlying silicon is not exposed. Around this crater there is anannular region corresponding to the non-damaged Si 3 N 4  level,that is, the pristine sensor gate area. In this region, PLL hasbeen ablated by the laser, but the Si 3 N 4  layer is still intact.The outer region corresponds to a sensor area covered by thenon-damaged PLL layer. Therefore, we can evaluate the PLLthickness by looking at the AFM reading across the “border”between the last two regions (Fig. 2b). The arrow puts in evi-denceaPLL-peakclosetotheedge,probablyduetotheresidualPLLbankedupagainsttheholeedgeasdirectconsequenceofthelaser shot. The AFM profile confirms the low average thickness( ≈ 12nm)ofthePLLlayer,similartothosefoundinotherworks(West et al., 1997; Richert et al., 2004). Under the hypothesisthat PLL layer is uniformly distributed charged layer, we canapproximate PLL as a charged plane, whose average distancefrom the surface is one half of the actual layer thickness, on theotherhandthedistancebetweenthePLLchargecentroidandtheSi 3 N 4  /electrolyteinterfaceisestimatedas d  pll  =6nm.Notethat,as required by the sketch in Fig. 1b, this value is greater than theDebye length, which is 1nm for saline solution (Grattarolla andMassobrio, 2002).This result can be used in our model to quantitatively calcu-late the amount of adsorbed PLL. Before doing this calculation,the densities of amine and silanol sites need to be quantified.This information can be gained from the variation of the ISFET V  th  as a function of the pH of the solution following the pro-cedure described in Grattarolla and Massobrio (2002) based onEq. (2). Using this calibration curve (not shown here) the amineand silanol site densities are evaluated as  N  nit  =2 × 10 18 and  N  sil  =3 × 10 18 m − 2 , respectively.Once coupled with measurements of the  V  th  shifts after eachstepandwithlayerthickness,thesedatacanbeusedtoquantita-tively evaluate the charge of molecules adsorbed on the surface.ThisisillustratedinFig.3:thecurvessteps“0”and“1”showthe Fig. 3.  I  ds – V  gs  curves taken after adsorption of PLL and after probe dsDNAimmobilization compared with the step 0 curve.  2112  L. Bandiera et al. / Biosensors and Bioelectronics 22 (2007) 2108–2114  I  ds – V  gs  characteristics before and after PLL adsorption, taken at V  ds  =300mV. Experimentally, we observe a rigid shift of the  I  ds – V  gs  curve to lower voltages after PLL deposition, which isexpected for adsorption of positive charge to n-channel ISFET.Such shift indicates that the PLL adsorption on the Si 3 N 4  gatesurface modifies both the Helmholtz and diffused layers, butno interface state is generated; in fact no variation in sub-threshold slope is observed (the log scale used for the  Y  -axisemphasizesthesub-thresholdbehaviour).Thethresholdvoltagevariation   V  th  generated by the PLL layer can then be writtenas   V  th  = V  th (ISFET, pll) − V  th (ISFET) and it is about 150mV. 3.3. Electrical detection of cDNA hybridization The DNA molecules carry a negative charge in neutralaqueous solution, i.e., opposite to PLL and Si 3 N 4  aminogroups, that are both positively charged. The super-aminesubstrate due to the PLL layer allows for DNA attachment tothe substrate by weak ionic bonds only; however, after UVtreatment the pyrimidine bases engage covalent bonds with thePLL molecules. Fig. 3 shows how the dsDNA immobilizationmodifiestheISFET  I  ds – V  gs  curve,leadingtoarigidshifttowardhigher gate voltages. Fig. 4 shows the average variation of thethreshold voltage   V  th  = V  th (step  n ) − V  th (step 0), during steps“1”–“3”(i.e.,denaturation).Eachpointcorrespondsto15exper-iments at least. After PLL adsorption   V  th  is around 120mVcorresponding to a PLL charge of (1.59 ± 0.65) × 10 − 4 C/m 2 ,that is about 20lysines/nm 2 , equivalent to roughly  N  occ ∼ 10 15 PLL occupied sites (  N  occ ) that is much lower than the totalnumber of silanol sites (  N  sil ). It is interesting to note that  C  sat,pll is comparable to the saturation concentration observed for PLLadsorption on glass microspheres (West et al., 1997). Whileafter dsDNA adsorption and immobilization (step “2”),   V  th withrespectthestep“1”curveisabout+500mV,correspondingto  − (36 ± 14.5) × 10 − 4 C/m 2 of immobilized charge. Noticethat after step “3” the negative charge decreases to roughly − (18 ± 8.5) × 10 − 4 C/cm 2 , probably due either to a net loss of DNA probe or to the loss of several hydrogen bonds (Bard and Fig. 4. Variation of the threshold voltage  V  th  = V  th (step  n ) − V  th (fresh) duringseveral ISFET coating steps (for their definition see text).Fig.5. Relativevariationofthethresholdvoltageduringhybridizationofperfectmatching cDNA and non-matching DNA at room temperature. Faulker, 1980) between the PLL amino groups and the Si 3 N 4 surface silanol groups during high temperature treatment.Which of the two possible events should be considered moreimportant is still an open question. In fact, the denaturation stepis routinely used in microarray experiments (Campanaro et al.,2002; Duggan et al., 1999; Chenug et al., 1999), but there areno dynamic methods to evaluate the DNA probe loss inducedby this process.Hybridizationexperimentsarethencarriedoutwithperfectlymatching cDNA and compared with non-matching DNA. Fig. 5shows the relative change of the ISFET threshold voltage dur-ing such hybridization experiments. The perfect matching DNAhybridization kinetics shows an initial increase during the first10 3 s(similartothatfoundbyfluorescencemicroscopyinsimilarconditions, e.g. Fixe et al., 2004), then   V  th  approaches a satu-ration value at around 16% with respect to the threshold voltageafter denaturation (i.e., step 3). The corresponding total chargeis around − 35 × 10 − 4 C/m 2 , twice that after step “3”, indicat-ing that practically all probe ssDNA molecules are bound to thecDNA in the solution. On the contrary, the non-matching DNAkinetics curve does not show any significant variation, show-ing only a slight drift, confirming that functionalized ISFET isreceptive only to a specific DNA sequence. 3.4. PLL/DNA adsorption kinetic One of the major advantages of the DNA ISFET-based sen-sors is their capability for real time measurements, such as thestudy of hybridization kinetics (Erickson et al., 2003; Tawa andKnoll, 2004). In fact, by using our ISFET sensors, it is also pos-sible to study the kinetics of both PLL adsorption on the gatearea and dsDNA adsorption on the PLL layer. Fig. 6 points outthesecharacteristics,infact,thefigureshowstheconcentrationsof the hybridized cDNA,  C  cDNA ( t  ), as a function of time ( t  ) dur-ing experiments (measured at room temperature). As first orderapproximation (an extensive description of the kinetics can befound in Erickson et al., 2003), the curve follows a Langmuir-like kinetic equation (Bard and Faulker, 1980): C cDNA ( t  ) = C sat  1 − exp  − t τ    (6)
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks