Anal. Chem. 1994,66, 2943-2948
Voltammetric DNA Biosensor for Cystic Fibrosis Based on a Modified Carbon Paste Electrode Kelly M. Mlllan, Angela Saraullo, and Susan R. Mlkkelsen’ Department of Chemistry and Biochemistry, Concordia University, 1455 de Maisonneuve Boulevard West, Montreal, Canada, H3G 1M8
Carbon paste electrodes modified by the inclusion of either octadecylamine or stearic acid were used as solid phases to which DNA was covalently bound. Immobilized DNA was detected by voltammetry of solutions containing submillimolar quantities of Co(bpy)3(c104)3, Co(p~n)3(clOs)j,4W ~ P h-Y C12 (bpy = 2,2‘-bipyridine; phen = 1,lO-phenanthroline), all of which associate reversibly with immobilized DNA and yield increased peak currents at DNA-modified electrodes. Immobilization onto octadecylamine-modified electrodes was performed using a water-soluble carbodiimide, and at high DNA concentrations in the reaction mixture, it resulted in visible polymerization of DNA on the surface. Optimization of the deoxyguanosine- (dG-) selective immobilization reaction for stearic acid-modified electrodes, using water-soluble carbodiimide and N-hydroxysulfosuccinimidereagents to activate carboxylate groups on the surface, yielded conditions of 4.5% (w/w) stearic acid and 10 pg/mL DNA. Polythymidylic acid of 4000-base average length (poly(dT)m) was immobilized at stearic acid-modified electrodes following enzymatic elongation with dG residues at the 3‘-end. These DNA-modified electrodes were used to study hybridization with analyte poly( d A ) m by insitu voltammetry of 60 pM Co(bpy)3(ClOd)jat low ionic strength (20 mM NaCl), and by voltammetry of the same complex, following exposure of the electrode to poly( d A ) m in a separate hybridization step conducted at high ionic strength (0.5 M NaCl). Results indicate slow ( 1 1 h) hybridization at low ionic strength and fast (110 min) hybridization at high ionic strength. At high ionic strength, a detection limit of 2.5 ng of poly(dA)m was obtained, which corresponds to 250 pg of a typical 400-base polymerase chain reaction product. The results are applied to the selective detection of the cystic fibrosis AF508 sequence in an 18-base oligodeoxynucleotide sample. Increased knowledge of the normal base sequence of human deoxyribonucleic acid (DNA) has led to the discovery of many mutations that are responsible for, or linked to, inherited d i s e a s e ~ l - These ~ mutations vary greatly in complexity. A well-defined point mutation, in which a single base is substituted for another, causes the amino acid substitution in @-globin that results in sickle-cell anemiaa4 A three-base deletion in the gene that codes for a transmembrane regulatory protein is associated with 70% of cystic fibrosis patients and ( I ) Jameson, J. L.; Hollengerg, A. N. Harm. Merab. Res. 1992, 24, 201. (2) Baird. P. A,; Anderson, T. W.; Ncwcombe, H. B.; Lowry, R. B. Am. J. Hum. Genet. 1988, 42, 677. (3) Skogerboc, K. J. Anal. Chem. 1993, 65, 416R. (4) Conner, B. J.; Reyes, C. M.;Morin, C.; Itakura, K.; Teplitz, R. L.; Wallace, R. B. Proc. Narl. Acad. Sci. U.S.A. 1983, 80, 278. 0003-2700/94/0366-2943$04.50/0 0 1994 American Chemlcal Society
carriers.’~~Elongation of a healthy gene by the insertion of repetitive multibase units occurs in Huntington’s disease.’ The detection of such disease-related mutations is becoming increasingly important in areas related to genetic screening and therapy. Genetic testing methods commonly use DNA isolated from lymphocytes and employ the polymerase chain reaction (PCR)* to amplify or increase the concentration of the region of interest from the 3 billion base-pair human genome. The PCR product can then be subjected to electrophoresis or adsorbed directly onto a membrane, which is then exposed to a solution containing a DNA probe. The DNA probe is a synthetic, single-stranded oligodeoxynucleotide,about 20 bases long, that has been chemically or enzymatically labeled with a radioisotope, fluorophore, chemiluminophore, or hapten/ ligand such as biotin.9 The sequence of the probe is complementary to the sequence of interest in the PCR product, so that hybridization results in retention of the label on the membrane after rinsing, and the presence of the label implies the presence of the sequence of interest. Stringent control of hybridization conditions allows the detection of point mutat i o n ~ .Elegant ~ modern variations of this general protocol exploit the sequence selectivity of the PCR amplification reactionI0J1 or the susceptibility of mismatched bases to chemical degradation.I2 Recent interest in the development of a biosensor capable of detecting a known sequence of DNA has led to reports of transduction methods that allow discrimination between immobilized single- and double-stranded DNA. These methods are either direct or indirect and employ singlestranded DNA immobilized on the transducer surface. The sequence-recognition event, hybridization to form immobilized ( 5 ) Riordan, J. R.;Rommens, J. M.;Kerem,B.; Alon,N.;Rozmahel, R.;Grzelczak, Z.; Zielenski, J.; Lok, S.;Plavsic, N.; Chou, J. L.; Drumm, M.L.; Iannuzzi, M.C.; Collins, F. S.;Tsui, L. C. Science 1989, 245, 1066. (6) Fanen, P.; Ghanem, N.; Vidaud. M.; Besmond, C.; Martin, J.; Castes, B.; Plassa, F.; Goossens, M. Genomics 1992, 13, 770. (7) MacDonald, M. E.; Ambrose, C. M.; Duyao, M.P.; Myers, R. H.; Lin, C.; Srinidhi, L.; Barnes, G.;Taylor, S.A.; James, M.;Groot, N.; MacFarlane, H.; Jenkins, B.; Anderson, M. A.; Wexler, N. S.;Gusella, J. F.;Bates, G. P.; Baxendale, S.;Hummerich, H.; Kirby, S.;North, M.; Youngman, S.; Mott, R.; Zehetner, G.; Sedlacek, Z.; Poustka, A.; Frischauf, A.-M.; Lehrach, H.; Buckler, A. J., Church, D.; DoucettcStamm, L.; ODonovan, M. C.; RimaRamirez, L.; Shah, M.; Stanton, V. P.; Strobel, S.A.; Draths, K. M.; Wales, J. L.; Dervan, P.; Housman, D. E.; Altherr, M.;Shiang, R.; Thompson, L., Fiedler, T.; Wasmuth, J. J.; Tagle, D.; Valdes, J.; Elmer, L.; Allard, M.; Castilla, L.; Swaroop, M.;Blanchard, K.; Collins, F.; Snell, R.; Holloway, T.; Gillespie, K.; Datson, N.; Shaw, D.; Harper, P. S.Cell 1993, 72, 971. (8) Timmer, W.C.; Villalobos, J. M.J. Chem. Educ. 1993, 70, 273. (9) Guesdon, J. L. J. Immunol. Methods 1992, 150, 33. (IO) Wenham, P. R.; Newton, C. R.; Price, W. H. Clin. Chem. 1991, 37, 241. ( I I ) Jalanko, A.; Kere, J.; Savilahti, E.; Schwartz, M.; Syvanen, A. C.; Ranki, M.; Soderlund, H. Clin. Chem. 1992, 38, 39. (12) Dianzani, I.; Camaschella, C.; Saglio, G.; Forrest, S.M.; Ramus, S.;Cotton, R. G. H. Genamics 1991, J J , 48.
Analflical Chemistty, Vol. 66, No. 18, September 15, 1994 2843
double-stranded DNA, can be followed directly, as is the case with piezoelectric13and surface acoustic wave14.15transducers, which are sensitive to added surface mass. An indirect detection method uses a fluorophore that binds only to doublestranded DNA at the surface of an evanescent wave detector waveguide.16 Our research toward a sequence-selective biosensor for DNA employs a probe sequence that is covalently bound to the surface of an amperometric e l e ~ t r o d e . ~Hybridization ~J~ of the immobilized probe sequence with its dissolved complement yields an immobilized double strand that can be detected using redox-active metal/polypyridine complexes that associate selectively and reversibly with double-stranded DNA. The presence of the immobilized double strand causes preconcentration of the metal complex in the DNA layer near the surface of the electrode, and it results in much larger voltammetric peakcurrents than are observed for immobilized single-stranded DNA. We have studied the complexes tris(1 ,IO-phenanthroline)cobalt(III) perchlorate and tris(2,2’bipyridyl)cobalt(III) perchlorate, which are reversibly reducible to their cobalt(I1) forms with formal potentials of 0.137 and 0.085 V vs SCE, respectively, well within the window of +1.2 to -0.9 V vs SCE over which DNA is electroinactive. Our studies with oxidized glassy carbon electrodes have shown that oligo- and polydeoxynucleotides can be covalently immobilized onto carboxylic acid groups using water-soluble carbodiimide and N-hydroxysuccinimide coupling reagents and that this reaction is selective for immobilization through deoxyguanosine residues.18 We have continued this work by investigating carbon paste electrodes for use in a DNA biosensor. The graphite/mineral oil mixture of carbon paste is readily modified with additives, such as octadecylamine or stearic acid, to provide functional groups for the covalent attachment of DNA. In the presence of a carbodiimide reagent, the primary amine group of octadecylamine will form a phosphoramidate bond with the 5’-terminal phosphate group of DNA,19 while carbodiimide and N-hydroxysuccinimide reagents allow DNA to couple to stearic acid-modified carbon paste electrodes through deoxyguanosine residues. Hybridization of immobilized DNA was detected using tris( 1,lo-phenanthroline)cobalt(III), tris( 2,2’bipyridyl)cobalt(III), and tris(2,2’-bipyridyl)osmium(II).A model sensor for the AF504 deletion that has been linked to cystic fibrosis was constructed using an 18-base probe sequence immobilized through deoxyguanosine residues that were enzymatically added20 to the 3’-end of the probe. An investigation of the hybridization of 20-base oligodeoxynucleotide probes for &globin with their complementary sequences has shown that the presence of a short (1 5-base) oligo(dA) tail does not interfere with selective hybridization,21 so the (13) Fawcett, N. C.; Evans, J. A,; Chen, L.-C.; Flowers, N. Anal. Lett. 1988,21, 1099. (14) Andle, J. C.; Vetelino, J. F.; Lade, M. W.; McAllister, D. J. Sens. Actuators E 1992, 8, 191. ( I 5 ) SU,H.; Kallury, K. M. R.; Thompson, M.; Roach, A. Anal. Chem. 1994.66, 769. ( 1 6) Squirrel, D. J. Measurement of Nucleic Acid Hybridization by Total Internal Reflection Fluorescence. PCT Int. Patent W09306241, 1993. (17) Millan, K. M.; Spurmanis, A. J.; Mikkelsen, S.R. Electroanalysis 1992, 4, 929. (18) Millan, K. M.; Mikkelsen, S . R. Anal. Chem. 1993, 65, 2317. (19) Rasmussen, S.R.; Larsen, M. R.; Rasmussen, S . E. Anal. Eiochem. 1991,198, 138. (20) Deng, G.; Wu, R. Methods Enzymol. 1983, 100, 96.
2944
Analytical Chemistty, Vol. 66, No. 18,
September 15, 1994
addition of dG residues to the normal and cystic fibrosis sequences are similarly not expected to interfere with selectivity. Conditions for the selective detection of the disease sequence are reported. The reusability of the DNA biosensors are investigated, and results with modified carbon paste electrodes are compared to previous results obtained with glassy carbon electrodes.
EXPER IMENTAL SECT1ON Materials. Poly(dG)poly (dC), terminal deoxynucleotidyl transferase, DNA molecular weight markers for electrophoresis, and proteinase K were obtained from Boehringer Mannheim. Poly(dA), poly(dT), oligo(dT)zo,deoxyguanosine triphosphate (dGTP), l-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC), guanidinium hydrochloride, and tris(hydroxymethy1)aminomethane (Tris) were purchased from Sigma. The four oligonucleotides used for the cystic fibrosis sensor were obtained from Queens University DNA Synthesis Laboratory (Kingston, Ontario) and had the following sequences: S-GAA-ACA-CCA-ATG-ATA-TTT3’ (AF508 probe), 5’-GAA-ACA-CCA-AAG-ATG-ATA-3‘ (normal probe), 5’-AAA-TAT-CAT-TGG-TGT-TTC-3’ (complement to AF508 probe), and 5’-TAT-CAT-CTT-TGGTGT-TTC-3’ (complement to normal probe). Octadecylamine, stearic acid, potassium hexachloroosmate( IV) chloride, 2,2’-bipyridine, 1-methylimidazole, mineral oil, and glycerol were obtained from Aldrich. Fisher supplied reagent grade cobalt(I1) chloride, bromine, hydrochloric and perchloric acids, diethyl ether, ethanol, chloroform, phenol, potassium iodide, sodium perchlorate, sodium chloride, and mono- and dibasic potassium phosphates. 1,lo-Phenanthroline was obtained from GFS Chemicals. N-Hydroxysulfosuccinimide (NHS) was purchased from Pierce. Bio-Rad supplied acrylamide, bis(acrylamide), N,N,N’,N’-tetramethylethylenediamine, bromophenol blue, urea, and silver stain. Carbon powder used for carbon paste electrodes was supplied by Johnson Matthey Electronics. Electrode holders and Ag/AgCl reference electrodes were obtained from Bioanalytical Systems, and platinum wire (0.5-mm diameter, Fisher) was used as the auxilliary electrode. Tris(1,lO-phenanthroline)cobalt(III) perchlorate (Co(phen) 3 (C104)3), and tris( 2,2’- bipyridyl)cobalt( 111) perchlorate (Co(bpy)3(ClO&), were prepared from cobalt(I1) chloride, bromine, perchloric acid, and the appropriate ligand according to published procedures.22 They were recrystallized twice from water prior to use. Tris(2,2’-bipyridyl)osmium(11) chloride (Os(bpy)sC12) was also prepared by a published method23 and was recrystallized from a minimum volume of 50% ethanol. Methods. Preparation of Carbon Paste Electrodes. Carbon powder (500 mg) and mineral oil (300 pL) were combined in the presence of either octadecylamine or stearic acid modifier, using a mortar and pestle, so that the final quantity of modifier ranged from 0.4 to 10% (w/w). The mixture was packed tightly into an electrode holder and polished to a smooth finish. (21) Woodhead, J. L.; Malcolm, A. D. B. Eiochem. Soc. Trans. 1986.14, 1168. (22) Dollimore, L. S.; Oillard, R. D. J. Chem. Soc., Dalton Tram. 1973, 933. (23) Creutz, C.; Chou, M.; Netzel, T. L.; Okumura, M.; Sutin, N. J. Am. Chem. SOC.1980. 102, 1309.
Catalytic Elongation of Oligodeoxynucleotides. For immobilization through deoxyguanosine (dG) residues onto stearic acid-modified electrodes, oligodeoxynucleotides were tailed at the 3'-end with dG residues. A 100-pg aliquot of the oligodeoxynucleotide was incubated with a 200-fold molar excess of dGTP and terminal deoxynucleotidyl transferase (250 units total, in aliquots of 50 units at t = 0, 4, and 8 h, and 100 units at 12 h) in 500 pL of sodium cacodylate buffer at 37 OC for 24 h, as recommended by Boehringer Mannheim. Products were purified by treatment with proteinase K, followed by phenol/chloroform extraction, and reconstituted to 1 mg/mL in 0.02 M phosphate buffer, pH 6.9. Elongation was confirmed by comparison of electrophoretic migration distances for the reactant oligodeoxynucleotides and the isolated products by polyacrylamide gel electrophoresis as described previously.18 Immobilization of DNA. ( A ) Octadecylamine-Modified Carbon Paste Electrodes. DNA was coupled to the prepared electrode surfaces in a single step in which 100 pL of a solution containing 0.1 M EDC and a given DNA or oligodeoxynucleotide concentration in 0.01 M methylimidazole buffer, pH 8.2, was evaporated to dryness at ambient temperature. Modified electrodes were rinsed with and stored in 5 mM Tris, pH 7.0, with 0.02 M NaCl at 4 "C. ( B )Stearic Acid-Modified Carbon Paste Electrodes. The prepared electrode was inverted and activated by the evaporation to dryness of 50 pL of a solution containing 5 mM EDC and 8 mM NHS in 0.02 M phosphate buffer, pH 6.9. After the surface was rinsed, DNA or oligodeoxynucleotide was coupled to NHS ester groups during the evaporation to dryness of 50 pL of a solution of DNA in the same phosphate buffer. The concentration of DNA in this solution was varied from 1 pg/mL to 1 mg/mL during optimization studies. For poly(dG)poly(dC) immobilization, a denaturation step preceded the immobilization: the solution was heated to 100 OC for 10 min and then rapidly cooled in an ice bath. The modified electrodes were then rinsed and stored in 5 mM Tris, pH 7.0, with 0.02 M NaCl at 4 "C. Hybridization of Immobilized DNA. For hybridization reactions studied by in situ voltammetry, the DNA sensor was immersed in a solutioncontaining 60 pM Co(bpy)3(C104)3 in 5 mM Tris, pH 7.0, with 20 mM NaCl at 25.0 OC. At time t = 0, an aliquot of a solution of DNA possessing a sequence complementary to the immobilized DNA was added. M a g netic stirring was continued throughout the experiment, except during voltammetric scans. Hybridizations were also performed at high ionic strength in a solution containing 0.5 M NaCl and the desired complementary DNA concentration. Sensors were periodically removed from this solution, rinsed, and placed in 60 pM Co(bpy)3(C104)3 in 5 mM Tris, pH 7.0, with 20 mM NaCl for the voltammetric scans; the sensors were then rinsed and returned to the hybridization solution. More stringent conditions for the selective hybridization of the cystic fibrosis target sequence were obtained by applying a 25-pL drop of analyte DNA to the surface of the inverted sensor and varying the incubation temperature during hybridization. Cyclic Voltammetry. Cyclic voltammetry was performed in a standard, three-electrode, water-jacketed cell at 25.0 f
0.2 OC, using a BAS lOOA potentiostat. Unless otherwise stated, solutions of electroactive species were prepared in 5 mM Tris, pH 7.0, with 0.02 M NaCl, and voltammetry was performed at 50 mV/s. RESULTS AND DISCUSSION Two reactions were tested for the coupling of singlestranded DNA (ssDNA) to carbon paste electrode surfaces. One employed octadecylamine-modified carbon paste, and the carbodiimide-mediated formation of a phosphoramidate bond between the 5'-terminal phosphate of the DNA and the terminal amino group of octadecylamine. The second involved stearic acid-modified carbon paste, with immobilization through deoxyguanosine residues using carbodiimide and N-hydroxysuccinimide reagents as previously described for immobilizations onto oxidized glassy carbon surfaces.18 The octadecylamine (ODA) content of carbon paste electrodes was varied from 0.4 to 4% (w/w), while the concentration of denatured poly(dG)poly(dC) in the immobilization mixture was held constant at 1 mg/mL. Following modification of the surface, the electrodes were examined by cyclic voltammetry of a solution of 0.12 mM O~(bpy)3~+, and anodic peakcurrents of 5.1,8.8, and 7.0 pA were obtained for 0.4, 0.8, and 4% ODA, respectively, after a 30-min incubation. Over this ODA range, unmodified electrodes yieldedpeakcurrentsof 1 . 4 f 0 . 1 pAfor O~(bpy)3~+ oxidation, but no voltammetric peaks were detected for Co( b ~ y ) 3 ~ +C0(phen)3~+reduction 0r under the same conditions. Holding ODA constant at 0.4% and varying the poly(dG)poly(dC) concentration yielded increased peak currents at the higher concentrations: 0.010,0.10, 1.0, and 2.0 mg/mL poly(dG)poly(dC) gave anodic peak currents of 3.7, 3.5,7.0, and 6.9 pA, respectively. At the two highest DNA concentrations, gelatinous strands were apparent on the surface of the electrode after rinsing with buffer. Polyacrylamide gel electrophoresisshowed that DNA incubated with carbodiimide at 4 OC for 24 h migrated much more slowly than DNA not exposed to the carbodiimide. It is likely that phosphodiester bonds are formed between 5'-terminal phosphates and 3'terminal hydroxyl groups to yield much higher molecular weight immobilized species. Because of this polymerization during the reaction, the octadecylamine-modifiedcarbon paste electrodes were not studied further. Optimum conditions for the coupling of denatured poly(dG)poly(dC) to stearic acid-modified carbon paste electrodes were obtained by varying the concentration of stearic acid in the electrode material and by varying the DNA concentration. Figure 1 shows cathodic peak currents for 0.12 mM Co( ~ h e n ) 3 ~as+a function of percent stearic acid in the carbon paste mixture, before and after the immobilization of poly(dG)poly(dC) from a 1 mg/mL solution. A clear optimum is observed at 4.5% stearic acid, where the DNA-modified electrode yields a peak current of 6.0 pA at 50 mV/s. Using 4.5% stearic acid in the carbon paste mixture, the DNA concentration in the poly(dG)poly(dC) solution was varied from 1 pg/mL to 1 mg/mL, and the results are shown in Figure 2. The absolute magnitudes of the peak currents at the DNA-modified electrodes varied considerably from one day to the next. For example, at a poly(dG)poly(dC) concentration of 10 pg/mL in the immobilization solution, Analytical Chemistty, Vol. 66,No. 18, September 15, 1994
2945
2
6
i C
2
5 0
4
Y
6
L
2 '
01 0
I 7
1
2 3 4 5 6 Percent Stearic Acid (w/w) F@re 1. Cathodicpeakcurrentfor0.12mMCo(phenh3+asafunction of percent stearic acM in carbon paste electrode (0)before and g) after modification with denatured poly(dG)poly(dC). Conditions: 5 mM Tris, pH 7.0, wkh 20 mM NaCI; scan rate 50 mV/s.
100
t 0
-1.00'
o
30
.
'
1000
'
J
'
zoo0 aooo t, a
60 90 Time (mid
4000
120
6001
150
Flgure 3. Cathodic peak current for 60 pM Co(bpyh3+ as a function of time at a poly(mmodified carbon paste electrode, where t = 0 corresponds to the addiiion of 10 pg/mL poly(dA) to the voltammetry solution. Inset: firstorder kinetic plot for these data. Conditions as in Figure 1.
90 80
70 60
50
40 0.0
1.o
2 .o
3 .O
4.0
5.0
Log[Poly(dG)Poly(dC), ug/mL Flgure 2. Normalizedcathodic peak current for 0.12 mM Co(phenh3+ as a function of the concentration of denatured poly(dG)poly(dC)used during immobilizationonto a 4.5% stearic acibmodified carbon paste electrode. Inset: voltammogram obtained after modlficatlon with 10 Rg/mL DNA. Conditions as in Figure 1.
and a 0.12 mM solutionof C0(phen)3~+,cathodicpeakcurrents of 17.1,6.8,6.7, and 16.8 pA were obtained on four different days. Reactions performed at the same time using different electrode housings, however, yielded the same trends, with maximum peak currents obtained at DNA concentrations of 10 pg/mL. For this reason, peak currents were normalized to the maximum value obtained on a given day, which was assigned a value of 100%. The inset to Figure 2 shows the voltammogram obtained at the poly(dG)poly(dC)-modifiedelectrode under these optimized conditions. The shape of the voltammogram obtained at the DNA-modified electrode suggests adsorption rather than diffusion of the electroactive species. While the expected asymmetric voltammetric peaks are observed at the unmodified electrode, nearly symmetric peaks were obtained after DNA immobilization, indicating that the complexes are bound to DNA very close to the electrode surface and behave like adsorbed electroactive species. Measurements of peak current as a function of scan rate over the 5-200 mV/s range indicate behavior intermediate between diffusing and adsorbed species for all three complexes. Plots of peak current against scan rate tend toward a plateau at high scan rates, while plots of peak current against the square root of scan rate show upward curvature. The smaller signals obtained at higher DNA concentrations in Figure 2 may indicate steric or conformational factors 2946
AnalyticalChemistty, Vol. 66, No. 18, September 15, 1994
influencing the immobilization, but may also result from rehybridization of the denatured (single-stranded) poly(dG) and poly(dC) strands, leaving fewer dG residues available for covalent binding. DNA sensors for all further experiments were constructed by use of 4.5% stearic acid mixtures and 10 pg/mL DNA solutions. Detection limits for analyte DNA with the DNA biosensor are expected to depend on hybridization time and ionic strength. To study these variables, single-stranded poly(dT) of 4000-base average length (poly(dT)4W) was catalytically elongated with dG residues using terminal deoxynucleotidyl transferase, and the elongated product was bound to stearic acid-modified carbon paste electrodes. Three elongation reactions were performed, each with a different batch of enzyme, yielding three products (1-111). Following electrode modification, the hybridization of the immobilized poly(dT) with its complement poly(dA) was studied by voltammetry, in a solution containing the desired poly(dA) concentration along with 60 pM C0(bpy)3~+,20 mM NaCl, and 5 mM Tris, pH 7.0. Stirring was stopped periodically to allow the voltammetric run, and cathodic peak currents were recorded as a function of time over several hours. Figure 3 shows the peak currents obtained during hybridization of the immobilized poly(dT) (product I) with 10 pg/ mL poly(dA) in 3.0-mL total volume. As hybridization proceeds, the peak currents increase because the increased local concentration of double-stranded DNA near the electrode surface allows greater quantities of C0(bpy)3~+to bind near the surface of the electrode. The hybridization reaction appears complete after about 1 h, when a plateau or maximum value of the peak current is obtained. The inset to Figure 3 shows a first-order plot of -ln(i,f - ip,t) against time, where i , f is the final, or maximum value of the peak current and ip,t is the peak current measured at time t . Pseudo-first-order kinetics are expected to apply to the hybridization of a small quantity of immobilized DNA with analyte DNA in solution, since the concentration of analyte DNA (poly(dA)) is not expected to change significantly as a result of the surface hybridization reaction. From the slope of this plot, a pseudofirst-order rate constant of 8.8 X 10-4 s-* was obtained. Similar in situ hybridization experiments were performed over a range
of analyte poly(dA) concentrations using immobilized poly(dT) (product 11). Over the concentration range of 1-100 pg/mL, the pseudo-first-order rate constants did not vary substantially from 0,001 s-1, indicating that the interaction of immobilized poly(dT) with soluble poly(dA) is not the ratelimiting step in the hybridization process under these conditions. To observe an increased peak current during hybridization, it is necessary to use a hybridization medium of low ionic strength, since the binding of transition metal polypyridine complexes is at least partly electrostatic in nature, and high ionic strength media result in lower association constants.24 However, hybridization of single-stranded DNA is ordinarily carried out at high ionic strength, with [NaCl] 1 0.4 M, and at a temperature 25 OC below the melting temperature of the double-stranded form (53 - 25 = 28 OC for 4000-base-pair poly(dA)poly(dT)). Under these condition^,^' the pseudofirst-order rate constant may be calculated from eq 1, where
k,, = (3.5 X 10S)(L'/*)[poly(dA)]/N
(1)
L is the probe strand length in bases, N is the total number of base pairs in a nonrepeating sequence, the target DNA concentration is in molar units, and kohis in reciprocal seconds. Using this expression with L = 4000, N = 1, and poly(dA) =8 X M (10 pg/mL), a value of 0.2 s-l is obtained for kob. Therefore, we expected that the hybridization process could be made to occur much faster, or at lower analyte DNA concentrations, if it took place in a separate step at high ionic strength and in the absence of C0(bpy)3~+. A series of batch hybridization experiments were carried out, where the hybridization of immobilized poly(dT) (product 111) with soluble poly(dA) in 0.5 M NaCl was periodically interrupted by the removal of the sensor for voltammetry in a solution containing 60 pM C0(bpy)3~+,20 mM NaCl, and 5 mM Tris, pH 7.0. In these experiments, maximum peak currents were always obtained after the first 10-min exposure to the target DNA solution, for all [poly(dA)] values over the 1 ng/mL to 1 pg/mL range, and the peak currents tended to decrease slightly at longer hybridization times. Interestingly, in both the batch and in situ hybridization experiments, the magnitude of the signal change during hybridization depends on the poly(dA) concentration in the hybridization solution. The covalent poly(dT) immobilization step is believed to contribute to significant variability in the absolute values of the initial and final peak currents; we have found that factors such as evaporation rates, pretreatment of carbon prior to carbon paste preparation (heating at 100 OC overnight), and electrochemical pretreatment affect peak currents, but this subject requires further study. Because of the variability in the absolute peak currents, the percent increase in current upon hybridization was determined for each sensor and is related to poly(dA) concentration as shown in Figure 4. It is interesting to note that hybridization reactions carried out at high ionic strength yield lower detection limits for poly(dA), by about 2 orders of magnitude, for a 30% increase in peak current. The 4000-base analyte sequence is (24) Barton, J. K.; Goldberg,J. M.; Kumar, C. V.; Turro, N. J. J . Am. Chem. SOC. 1986, 108, 2081. (25) Meinkoth, J.; Wahl, G . Anal. Biochem. 1984, 138, 267.
8
1 120 100 80 60 40 140
20 0
,;- - - - _ _ _0'_'.-
'.
,,,**,*,**'*
m -
/
/
Flgure 4. Percent increase in cathodic peak current for 60 pM Co(bpyh3+ at a poly(dT)-modified carbon paste electrode as a function of analyte poty(dA) concentration used during hybridization, using )(. indtuvottammetry in the presence of poly(dA), 5 mM Tris, pH 7.0, with 20 mM NaCl and (0)batch hybridization performed separately In 0.5 M NaCI.
thus readily detectable at a concentration of 100 ng/mL. The same total quantity of immobilized double-stranded DNA would be formed by 10 ng/mL of a typical PCR product of 400-base length, since the sequence of interest is present at 10-fold higher concentration in this shorter species. The volume of analyte DNA solution used during the batch hybridization step can be as low as 25 pL, so that a total quantity of 250 pg of target DNA is required for each hybridization assay. This value is almost 1 order of magnitude lower than the recently reported value of 2 ng obtained with a surface acoustic wave transducer.15 DNA sensors for the cystic fibrosis-linked AF508 sequence and the corresponding normal sequence were constructed by first elongating the two synthetic 18-base oligodeoxynucleotides at the 3'-hydroxyl end using terminal deoxynucleotidyl transferase and dGTP. Electrophoresis of the products of these reactions indicated total lengths of 42-44 bases, so that about 25 dG residues had been added to each probe sequence. Sensors were constructed by binding these products to the surface of stearic acid-modified carbon paste electrodes. Exposure of each sensor to its complementary oligodeoxynucleotide (18-mer) sequence resulted in signal increases of about 2 pA. At ambient temperature, both the normal and AF508 sensors yielded increased voltammetric peak currents for C0(phen)3~+reduction at the two sensors. When the temperature of hybridization was raised to 42 "C, however, only the completely complementary analyte sequences yielded the expected signal increases. Voltammograms obtained under these conditions with the AF508 sensor are shown in Figure 5.
Attempts to regenerate the single-stranded oligodeoxynucleotide on the surface of the carbon paste electrode were made by rinsing thesensors in hot, deionized water (this method was successful in previous studies employing glassy carbon electrodes's) or by incubating the sensor in 8 M urea or 6 M guanidinium hydrochloride. The hot-water rinse method yielded satisfactory results only over the first two complete cycles of hybridization and regeneration, and the third cycle of hybridization yielded a near-doubling of the 2.0-2.5-pA cathodic peak currents for C0(phen)3~+observed after the first two hybridizations. Attempted regeneration using the Analytical Chemistry, Vol. 66, No. 18, September 15, 1994
2947
A
*
tion of the immobilized, single-stranded oligodeoxynucleotide on the electrode surface, it therefore appears that the mechanical stability of DNA sensors based on carbon paste electrodes does not permit their reuse.
& +0.5* E, V vs AglAgCl
Figure 5. Cyclic voltammograms of 0.12 mM Co(phenh3+ In 5 mM Tris, pH 7.0, with 20 mM NaCI, obtained at a AF508 oligodeoxynucleotide-modifled carbon paste electrode (A) before hybridization and (B) after hybrldirationwith the complementaryoligodeoxynucleotide 18-mer.
denaturants urea and guanidinium were unsuccessful and resulted in destruction of the carbon paste electrodes. In the absence of alternative, less extreme methods for the regenera-
2948
AnalyticalChemistry, Vol. 66, No. 18, September 15, 1994
ACKNOWLEDGMENT Research grants from the Natural Sciences and Engineering Research Council of Canada and from the Fonds pour la Formation des Chercheurs et 1'Aide a la Recherche of Quebec are gratefully acknowledged. Valuable discussions with Professor Oswald S.Tee (Concordia) are also acknowledged. Received for review April 13, 1994. Accepted June 1, 1994.' Abstract published in Aduance ACS Abstracts, July 15. 1994.
