High-Density, Covalent Attachment of DNA to Silicon Wafers for

Sequenom Inc., 11555 Sorrento Valley Road, San Diego, California 92121, Department of Biochemistry and Molecular. Biology, University of Hamburg, ...
0 downloads 0 Views 193KB Size
Anal. Chem. 1997, 69, 2438-2443

High-Density, Covalent Attachment of DNA to Silicon Wafers for Analysis by MALDI-TOF Mass Spectrometry Maryanne J. O’Donnell,*,†,‡ Kai Tang,† Hubert Ko 1 ster,†,§ Cassandra L. Smith,‡ and Charles R. Cantor‡

Sequenom Inc., 11555 Sorrento Valley Road, San Diego, California 92121, Department of Biochemistry and Molecular Biology, University of Hamburg, Martin-Luther-King Platz, 20146 Hamburg, Germany, and Center for Advanced Biotechnology, Boston University, Boston, Massachusetts 02215

A method is described for the covalent attachment of DNA to a solid surface at high density for hybridization detection by mass spectrometry. A silicon wafer is functionalized to place an amino group on the surface; a heterobifunctional cross-linking agent is then reacted with the primary amine to incorporate an iodoacetamido group. An oligodeoxynucleotide containing a 3′- or a 5′-disulfide is treated with a reducing agent, resulting in a terminal free thiol, which is then coupled to the iodoacetamido surface. Analysis of the surface reveals that the amount of covalently bound oligodeoxynucleotide is 250 fmol of DNA/mm2 with ∼40% of the immobilized oligodeoxynucleotides available for hybridization. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometric (MALDI-TOF MS) analysis reveals that the covalent linkage to the support remains intact, only the annealed strand is desorbed by the laser, and the amount of DNA hybridized to the array is sufficient for detection. In the fields of molecular biology and biochemistry, as well as in the diagnosis of diseases, nucleic acid hybridization has become a powerful tool for the detection, isolation, and analysis of specific oligonucleotide sequences. Typically, such hybridization assays utilize an oligodeoxynucleotide probe that has been immobilized on a solid support; such is the case for reverse dot blots1 for example. More recently, arrays of immobilized DNA probes attached to a solid surface have been developed for sequencing by hybridization2,3 (SBH). SBH uses an ordered array of immobilized oligodeoxynucleotides on a solid support. A sample of unknown DNA is applied to the array, and the hybridization pattern is observed and analyzed to produce many short bits of sequence information simultaneously. An enhanced version of SBH, termed positional SBH (PSBH), has been developed which uses duplex probes containing single-stranded five base 3′ overhangs to capture the target rather than the simple singlestranded probes.4 It is now possible to combine a PSBH capture †

Sequenom Inc. Boston University. § University of Hamburg. (1) Saiki, R. K.; Walsh, P. S.; Levenson, C. H.; Erlich, H. A. Proc. Natl. Acad. Sci. U.S.A. 1989, 86, 6230-6234. (2) Drmanac, R.; Labat, I.; Brukner, I.; Crkvenjakov, R. Genomics 1989, 4, 114128. (3) Strezoska, Z.; Paunesku, T.; Radosavljevic, D.; Labat, I.; Drmanac, R.; Crkvenjakov, R. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 10089-10093. (4) Broude, N. E.; Sano, T.; Smith, C. L.; Cantor, C. R. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 3072-3076. ‡

2438 Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

approach with conventional Sanger sequencing to produce sequencing ladders detectable by gel electrophoresis.5 For the arrays utilized in these schemes, there are a number of criteria that must be met for successful performance. For example, the immobilized DNA must be stable and not desorb during hybridization, washing, or analysis. The density of the immobilized oligodeoxynucleotide must be sufficient for the ensuing analyses; however, there must be minimal nonspecific binding of nontarget DNA to the surface. The immobilization process should not interfere with the ability of immobilized probes to hybridize. For the majority of applications, it is best for only one point of the DNA to be immobilized, ideally a terminus. In recent years, a number of methods for the covalent postsynthesis immobilization of DNA to “flat” (essentially twodimensional) solid supports have been developed which attempt to meet all the criteria listed above. For example, appropriately modified DNA was covalently attached to flat surfaces functionalized with amino acids,6-8 carboxyl groups,9 epoxy groups,10,11 or amino groups.12,13 Although many of these methods were quite successful for their respective applications, the density of oligonucleotide bound to silicon (maximum of ∼20 fmol of DNA/mm2 of surface)10,11 was far less than the theoretical packing limit of DNA. Therefore, methods were explored to achieve higher densities of immobilized DNA on a silicon surface; this surface may subsequently be used for hybridization and analysis by any detection method, including matrix-assisted laser desorption/ ionization time-of-flight mass spectrometry (MALDI-TOF MS). MALDI-TOF MS enables the separation and detection of a mixture of biomolecules in a fraction of a millisecond without gel electrophoresis and labeling.14-17 For DNA molecules, the detec(5) Fu, D.; Broude, N. E.; Ko ¨ster, H.; Smith, C. L.; Cantor, C. R. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 10162-10166. (6) Running, J. A.; Urdea, M. S. Biotechniques 1990, 8, 276-277. (7) Newton, C. R.; Holland, D.; Heptinstall, L. E.; Hodgson, I.; Edge, M. D.; Markham, A. F.; McLean, M. J. Nucleic Acids Res. 1993, 21, 1155-1162. (8) Nikiforov, T. T.; Rogers, Y. H. Anal. Biochem. 1995, 227, 201-209. (9) Zhang, Y.; Coyne, M. Y.; Will, S. G.; Levenson, C. H.; Kawasaki, E. S. Nucleic Acids Res. 1991, 19, 3929-3933. (10) Lamture, J. B.; Beattie, K. L.; Burke, B. E.; Eggers, M. D.; Ehrlich, D. J.; Fowler, R.; Hollis, M. A.; Kosicki, B. B.; Reich, R. K.; Smith, S. R.; Varma, R. S.; Hogan, M. E. Nucleic Acids Res. 1994, 22, 2121-2125. (11) Eggers, M. D.; Hogan, M. E.; Reich, R. K.; Lamture, J. B.; Ehrlich, D. J.; Hollis, M. A.; Kosicki, B. B.; Powdrill, T.; Beattie, K. L.; Smith, S. R.; Varma, R. S.; Gangadharan, R.; Mallik, A.; Burke, B. E.; Wallace, D. BioTechniques 1994, 17, 516-524. (12) Rasmussen, S. R.; Larsen, M. R.; Rasmussen, S. E. Anal. Biochem. 1991, 198, 138-142. (13) Guo, Z.; Guilfoyle, R. A.; Theil, A. J.; Wang, R.; Smith, L. M. Nucleic Acids Res. 1994, 22, 5456-5465 S0003-2700(96)01007-4 CCC: $14.00

© 1997 American Chemical Society

Figure 1. Covalent attachment of oligodeoxynucleotides to a silicon dioxide surface. The silicon dioxide was reacted with (3-aminopropyl)triethoxysilane to produce a uniform layer of primary amino groups on the surface. A heterobifunctional cross-linking agent was then reacted with the primary amine to incorporate an iodoacetamido group. An oligodeoxynucleotide containing a 3′- or a 5′-disulfide (shown as the 5′) was treated with TCEP to reduce the disulfide to a free thiol, which was then coupled to the iodoacetamido surface.

tion range of MALDI-TOF MS has been extended to 500 bases long.18 Because of these advantages, mass spectrometry of DNA is now emerging as tool for analysis,19,20 sequencing,21 and diagnostics.22,23 It is now possible to envision a combination of an array of immobilized DNA probes, PSBH, and Sanger sequencing with analysis by MALDI-TOF MS to produce a plethora of sequence information extremely quickly compared to electrophoretic analysis. However, a limitation in the implementation of this procedure has been the amount of DNA that can be detected by MALDI-TOF MS. Current array-making techniques produce densities of immobilized probe and annealed hybrid that are below the detection limit of MALDI-TOF MS. Because of this lack of sensitivity, it was necessary to devise a new scheme for array making that would place a higher density of DNA on a silicon surface to which a greater amount of complement could be (14) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301. (15) Cotter, R. J. Anal. Chem. 1992, 64, 1027A-1039A. (16) Vestling, M. M.; Fenselau, C. Anal. Chem. 1994, 66, 471-477. (17) Nordhoff, E.; Cramer, R.; Karas, M.; Hillenkamp, F.; Kirpekar, F.; Kristiansen, K.; Roepstorff, P. Nucleic Acids Res. 1993, 21, 3347-3357. (18) Tang, K.; Tarenenko, N. I.; Allman, S. L.; Chang, L. Y.; Chen, C. H. Rapid Commun. Mass Spectrom. 1994, 8, 727-730. (19) Tang, K.; Fu, D. J.; Kotter, S.; Cotter, R. J.; Cantor, C. R.; Ko¨ster, H. Nucleic Acids Res. 1995, 23, 3126-3131. (20) Jurinke, C.; van der Boom, D.; Jacob, A.; Tang, K.; Wo¨rl, R.; Ko¨ster, H. Anal. Biochem. 1996, 237, 174-181. (21) Ko ¨ster, H.; Tang, K.; Fu, D-J.; Braun, A.; van der Boom, D.; Smith, C. L.; Cotter, R. J.; Cantor, C. R. Nature Biotechnol. 1996, 14, 1123-1128. (22) Braun, A.; Little, D. P.; Ko ¨ster, H. Clin. Chem., submitted. (23) Jurinke, C.; Zo ¨llner, B.; Feucht, H.-H.; Jacob, A.; Kirchhu ¨ bel, J.; Lu ¨ chow, A.; van der Boom, D.; Laufs, R.; Ko ¨ster, H. Genet. Anal. 1996, 13, 67-71.

hybridized. Although array making by in situ synthesis places a large quantity of DNA on the surface,24-27 it is difficult to monitor the quality of each immobilized oligodeoxynucleotide, and the cost of synthesis may be substantially increased compared to postsynthesis immobilizations. Therefore, DNA was synthesized using automated techniques,28,29 and postsynthetic modifications were performed to immobilize the probes on a surface. In this case, a silicon wafer was functionalized with (3-aminopropyl)triethoxysilane to place amino groups on the surface. A variety of heterobifunctional cross-linkers were explored, but that linker producing the best results incorporated an iodoacetamido group on the surface. An oligodeoxynucleotide containing a 3′- or a 5′disulfide was treated with tris(carboxyethyl)phosphine to reduce the disulfide to a free thiol; this reactive molecule was then coupled to the iodoacetamido surface (Figure 1). Analysis of the surface revealed that the amount of covalently bound oligodeoxynucleotide was 250 fmol of DNA/mm2, thus a ∼12.5-fold increase over previously reported postsynthesis immobilization methods on silicon. Approximately 40% of the immobilized oligodeoxy(24) Maskos, U.; Southern E. M. Nucleic Acids Res. 1992, 20, 1679-1684. (25) Southern, E. M.; Case-Green, S. C.; Elder, J. K.; Johnson, M.; Mir, K. U.; Wang, L.; Williams, J. C. Nucleic Acids Res. 1994, 22, 1368-1373. (26) Williams, J. C.; Case-Green, S. C.; Mir, K. U.; Southern, E. M. Nucleic Acids Res. 1994, 22, 1365-1367. (27) Pease, A. C.; Solas, D.; Sullivan, E. J.; Cronin, M. T.; Holmes, C. P.; Fodor, S. P. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 5022-5026. (28) Sinha, N. D.; Biernat, J.; Ko¨ster, H. Tetrahedron Lett. 1983, 24, 58435846. (29) Sinha, N. D.; Biernat, J.; McManus, J.; Ko ¨ster, H. Nucleic Acids Res. 1984, 12, 4539-4557.

Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

2439

nucleotides was available for hybridization. MALDI-TOF MS analysis revealed that the amount of DNA annealed to the support was sufficient for detection and that the covalent linkage was stable to laser desorption. MATERIALS AND METHODS All reagents, unless otherwise noted, were obtained from Aldrich Chemical, Milwaukee, WI. Silicon Surface Preparation. Silicon wafers were washed with ethanol, flamed over a bunsen burner, and immersed in an anhydrous solution of 25% (by volume) (3-aminopropyl)triethoxysilane in toluene for 3 h. The silane solution was then removed, and the wafers were washed three times with toluene and three times with dimethyl sulfoxide (DMSO). The wafers were then incubated in a 10 mM anhydrous solution of N-succinimidyl(4iodoacetyl)aminobenzoate (SIAB; Pierce Chemical, Rockford, IL) in anhydrous DMSO. Following the reaction, the SIAB solution was removed, and the wafers were washed three times with DMSO. Since it was impossible to monitor the condensation of SIAB and the amino group while on the solid support of the wafer, the reaction was performed in solution to determine the optimal reaction time. Thin-layer chromatography (TLC) (glass-backed silica plates with a 254 nm fluorescent indicator) (Baker, Phillipsburg, NJ) was employed using 95:5 chloroform/methanol (Baker) which enabled separation of the two starting materials. It was possible to visualize the SIAB starting material under long-wave ultraviolet light (302 nm); (3-aminopropyl)triethoxysilane was not active under ultraviolet light; therefore, the plate was sprayed with a solution of ninhydrin, which reacts with primary amines to reveal a purple spot upon heating. A microscale reaction was run in chloroform/DMSO using a slight molar excess of SIAB in comparison to (3-aminopropyl)triethoxysilane and monitored with the above mentioned TLC conditions. Oligonucleotide Modifications. Reduction of the disulfide from 3′- or 5′-disulfide-containing oligodeoxynucleotides (Operon Technologies, Alameda, CA or Oligos Etc., Wilsonville, OR) was monitored using reversed-phase FPLC (Pharmacia, Piscataway, NJ); a shift can be seen in the retention time of the oligodeoxynucleotide upon cleavage of the disulfide. Various reduction methods were investigated to determine the optimal conditions. In one case, the disulfide-containing oligodeoxynucleotide (31.5 nmol, 0.51 mM) was incubated with dithiothreitol (DTT; Pierce Chemical) (6.2 mmol, 100 mM) at pH 8.0 and 37 °C. With the cleavage reaction essentially complete, the free thiol-containing oligodeoxynucleotide was isolated using a Chromaspin-10 column (Clontech, Palo Alto, CA) since DTT may compete in the subsequent reaction. Alternatively, tris(2-carboxyethyl)phosphine (TCEP; Pierce Chemical) has been used to cleave the disulfide. The disulfide-containing oligodeoxynucleotide (7.2 nmol, 0.36 mM) was incubated with TCEP in pH 4.5 buffer at 37 °C. It is not necessary to isolate the product following the reaction since TCEP does not competitively react with the iodoacetamido functionality. Varying concentrations of TCEP were used for the cleavage reaction to determine the optimal conditions for the conjugation reaction. Probe Coupling. To each wafer that had been derivatized to contain the iodoacetamido functionality as described above was added a 10 mM aqueous solution of the free thiol-containing oligodeoxynucleotide in 100 mM phosphate buffer, pH 8; the reaction was allowed to proceed for a minimum of 5 h at room 2440

Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

temperature in 100% relative humidity. Following the reaction, the oligodeoxynucleotide solution was removed, and the wafers were washed two times in 5 × SSC buffer (75 mM sodium citrate/ 750 mM sodium chloride, pH 7) with 50% formamide (USB, Cleveland, OH) at 65 °C for 1 h each. Radiochemical Determination of Probe Density. In order to determine the amount of DNA covalently attached to a surface or the amount of a complementary sequence hybridized, radiolabeled probes were employed. In cases where a 5′-disulfidecontaining oligodeoxynucleotide was to be immobilized, the 3′ terminus was radiolabeled using terminal transferase enzyme and a radiolabeled dideoxynucleoside triphosphate. In a standard reaction, 15 pmol (0.6 µM) of the 5′-disulfide-containing oligodeoxynucleotide was incubated with 50 µCi (16.5 pmol, 0.66 µM) of [R-32P]dideoxyadenosine 5′-triphosphate (ddATP) (Amersham, Arlington Heights, Il) in the presence of 0.2 M potassium cacodylate, pH 7.0/4 mM magnesium chloride/0.2 mM mercaptoethanol. Upon the addition of 40 units of the terminal deoxynucleotidyl transferase enzyme (USB), the reaction was allowed to proceed for one hour at 37 °C. After this time, the reaction was stopped by immersion of the vial in a 75 °C water bath for ten minutes, and the product was isolated using a Chromaspin10 column (Clontech). Similarly, a 5′-disulfide-containing oligodeoxynucleotide was radiolabeled with 35S. In cases where a 3′-disulfide-containing oligodeoxynucleotide was to be immobilized, the 5′ terminus was radiolabeled using T4 polynucleotide kinase and a radiolabeled nucleoside triphosphate. For example, 15 pmol (0.6 µM) of the 3′-disulfidecontaining oligodeoxynucleotide was incubated with 50 µCi (16.5 pmol, 0.66 µM) of [γ-32P]adenosine 5′-triphosphate (ATP; Amersham) in the presence of 50 mM Tris-HCl, pH 7.6/10 mM MgCl2/ 10 mM 2-mercaptoethanol. Following the addition of 40 units of T4 polynucleotide kinase, the reaction was allowed to proceed for 1 h at 37 °C. The reaction was stopped by immersion of the vial in a 75 °C water bath for 10 min; the product was then isolated using a Chromaspin-10 column. To determine the density of covalently immobilized probe, the disulfide-containing oligodeoxynucleotide of choice was added to a trace amount of the same species that had been radiolabeled as described above. The disulfide was cleaved, the probe was immobilized on iodoacetamido-functionalized wafers, and the wafers were washed and then exposed to a phosphorimager screen (Molecular Dynamics, Sunnyvale, CA). For each different oligodeoxynucleotide utilized, reference spots were made on polystyrene in which the molar amount of oligodeoxynucleotide was known; these reference spots were exposed to the phosphorimager screen as well. Upon scanning the screen, the quantity (in moles) of oligodeoxynucleotide bound to each chip was determined by comparing the counts to the specific activities of the references. Hybridization and Efficiency. To a wafer that had been functionalized with an immobilized probe was added a solution of a complementary sequence (10 µM) in 1 M NaCl and TE buffer. The wafer and solution were heated to 75 °C and allowed to cool to room temperature over 3 h. After this time, the solution was removed, and the wafer was washed two times with TE buffer. To determine the amount of oligonucleotide hybridized, immobilization of the probe was first carried out as described above except that the probe was labeled with 35S rather than 32P. The density of immobilized probe was determined with the phosphor-

imager. Next, the same wafer was incubated in TE buffer, 1 M NaCl, and its complementary strand (10 µM) which had been radiolabeled with 32P. Hybridization was carried out as previously described. Following a wash to remove nonspecific binding, the wafer and reference were exposed to a phosphorimager screen with a piece of copper foil between the screen and the wafer. The copper foil serves to block the signal from 35S, while allowing the 32P signal to pass freely. The molar amount of hybridized oligonucleotide was then be determined, thus revealing the percent of covalently immobilized probe that is available for hybridization. MALDI-TOF Mass Spectrometric Analysis. As described above, wafers containing nonradiolabeled immobilized oligodeoxynucleotide (name, TCUC; sequence, GAATTCGAGCTCGGTACCCGG; molecular weight, 6455) were synthesized, and a complementary sequence (name, MJM6; sequence, CCGGGTACCGAGCTCGAATTC; molecular weight, 6415) was hybridized. The wafers were washed in 50 mM ammonium citrate buffer for cation exchange to remove sodium and potassium ions on the DNA backbone.30 A matrix solution of 3-hydroxypicolinic acid31 (3-HPA; 0.7 M in 50% acetonitrile/10% ammonium citrate) was spotted onto the wafer and allowed to dry at ambient temperature. The wafers were attached directly to the sample probe of a Finnigan MAT (Bremen, Germany) Vision 2000 reflectron TOF mass spectrometer using conducting tape. The reflectron possesses a 5 keV ion source and 20 keV postacceleration, a nitrogen laser was employed, and all spectra were taken in the positive ion mode. RESULTS AND DISCUSSION A variety of surfaces have been examined as supports for DNA immobilizations including microparticles,32-34 nylon membranes,35 and silicon.10,11 However, when a surface that would be appropriate for launching samples in mass spectrometry is to be chosen, the choices quickly become limited. Streptavidin-coated magnetic microbeads have been used as launches for mass spectrometry,19 but in this case, the beads were applied directly to the sample probe of the instrument. Thus, this technique is not amenable to an array format. To expedite the process of mass spectrometric analysis of DNA, an essentially flat launching surface is quite desirable since this can translate easily into an array. Additionally, the current mass spectrometric resolution is such that it is necessary to obtain the highest possible densities of immobilized and hybridized DNA on the surface in order to resolve the hybridized strand. For these reasons, silicon wafers were investigated to achieve high loading densities of immobilized DNA on the surface, resulting in high hybridization densities, and eventually producing successful launching pads for mass spectrometric analysis. Recent results have illustrated that mass spectrometric data acquired from silicon surfaces are superior to spectra acquired from alternative “flat” surfaces, such as glass.36,37 (30) Pieles, U.; Zu ¨ rcher, W.; Schar, M.; Moser, H. E. Nucleic Acids Res. 1993, 21, 3191-3196. (31) Wu, K. J.; Steding, A.; Becker, C. H. Rapid Commun. Mass Spectrom. 1993, 7, 142-146. (32) Ghosh, S. S.; Musso, G. F. Nucleic Acids Res. 1987, 15, 5353-5372. (33) Kremsky, J. N.; Wooters, J. L.; Dougherty, J. P.; Meyers, R. E.; Collins, M.; Brown, E. L. Nucleic Acids Res. 1987, 15, 2891-2909. (34) Yershov, G.; Barsky, V.; Belgovskiy, A.; Kirillov, E.; Kreindlin, E.; Ivanov, I.; Parinov, S.; Guschin, D.; Drobishev, A.; Dubiley, S.; Mirzabekov, A. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 4913-4918. (35) Zhang, Y.; Coyne, M. Y.; Will, S. G.; Levenson, C. H.; Kawasaki, E. S. Nucleic Acids Res. 1991, 19, 3929-3933.

Surface Chemistry. Employing standard silicon dioxide modification chemistry, a silicon wafer was reacted with (3aminopropyl)triethoxysilane to produce a uniform layer of primary amino groups on the surface. As shown in Figure 1, the surface was then exposed to a heterobifunctional cross-linker resulting in iodoacetamido groups on the surface. It was possible to determine the optimal reaction time of this reaction in solution using TLC. The SIAB cross-linker was visualized under long-wave ultraviolet light (302 nm) to reveal a spot with an Rf value of 0.58. (3-Aminopropyl)triethoxysilane was not active under ultraviolet light; therefore, ninhydrin was used to reveal a purple spot indicating the presence of a primary amine at the baseline. A microscale reaction was run using a slight molar excess of SIAB in comparison to (3-aminopropyl)triethoxysilane; TLC analysis after ∼1 min revealed a new spot visible under long-wave ultraviolet light with an Rf value of 0.28. There was no evidence of a purple spot upon spraying with ninhydrin; thus, all the (3aminopropyl)triethoxysilane starting material had been consumed in the reaction. UV light also revealed the excess SIAB that remained following the reaction. From these results, it was determined the reaction is complete after ∼1 min. In all cases, the iodoacetamido-functionalized wafers were used immediately to minimize hydrolysis of the labile iodoacetamido functionality. Additionally, all further wafer manipulations were performed in the dark since the iodoacetamido functionality is light sensitive. Disulfide reduction of the modified oligonucleotide was monitored by observing a shift in retention time on reversed-phase FPLC. It was determined that after 5 h in the presence of DTT (100 mM) or TCEP (10 mM), the disulfide was fully reduced to a free thiol. If the DTT reaction was allowed to proceed for a longer time, an oligonucleotide dimer formed in which pairs of free thiols had reacted. Such dimerization was also observed when the DTT was removed following the completion of the cleavage reaction. This dimerization was not observed when TCEP was employed as the cleavage reagent since this reaction is performed at pH 4.5; thus the free thiols were fully protonated, inhibiting dimerization. Immediately following disulfide cleavage, the modified oligonucleotide was incubated with the iodoacetamido-functionalized wafers. To ensure complete thiol deprotonation, the coupling reaction was performed at pH 8.0. The probe surface density achieved by this chemistry on silicon wafers was analyzed using radiolabeled probes and a phosphorimager. The probe surface density was also monitored as a function of the TCEP concentration used in the disulfide cleavage reaction (Figure 2). The results illustrated in Figure 2 are reproducible; however, work is currently in progress to determine the cause of the decreased conjugation between 20 and 40 mM TCEP. Using 10 mM TCEP to cleave the disulfide and the other reaction conditions described above, it was possible to reproducibly yield a surface density of 250 fmol/ mm2 of surface. Identical experiments as described above were performed except that the oligonucleotide probe lacked a thiol modification; surface densities of less than 5 fmol/mm2 of surface proved that nonspecific binding is minimal and that probe coupling most likely occurred as proposed in Figure 1. It is believed that (36) Little, D. P.; Cornish, T.; O’Donnell, M. J.; Braun, A.; Cotter, R.; Ko ¨ster, H., in preparation. (37) O’Donnell, M. J.; Little, D. P.; Cantor, C. R.; Ko ¨ster, H., in preparation.

Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

2441

Figure 2. Conjugation of oligodeoxynucleotide probes to a silicon surface as a function of TCEP concentration used in the disulfide reduction. The amount of conjugation was determined using radiolabeled probes and a phosphorimager.

the resultant improved conjugation density of this reaction is due to the high reactivity of the iodoacetamido groups on the surface combined with the large amount of free thiol-containing DNA generated by TCEP in the absence of a competitor such as DTT. Hybridization. After attaching 35S-labeled probes to the surface of wafers and determining conjugation density as described above, hybridization of 32P-labeled oligonucleotides was carried out; hybridization efficiency and density were determined using the phosphorimager and copper foil. It was determined experimentally that copper foil blocks 98.4% of a 35S signal, while fully allowing a 32P signal to be detected (data not shown). The complementary sequence reproducibly hybridized to yield 105 fmol/mm2 of surface; this corresponds to ∼40% of the conjugated probes available for hybridization. Similarly, a noncomplementary sequence was employed in this scheme yielding less than 5 fmol/ mm2 of surface in nonspecific binding. It is hypothesized that steric interference between the tightly packed oligonucleotides on the flat surface inhibits hybridization efficiencies higher than 40%. With this in mind, a spacer molecule was incorporated between the terminus of the hybridizing region of the oligonucleotide and the support. The chosen spacers were a series of poly(dT) sequences ranging in length from 3 to 25. Upon examination of these samples with radiolabels and the phosphorimager, it was determined that 40% was still the maximum hybridization that could be achieved. MALDI-TOF MS Analysis. Wafers were functionalized with probes, complementary sequences were hybridized, and samples were analyzed under standard MALDI conditions as described above. Previous findings have shown that double-stranded DNA is denatured in the MALDI process,38,39 and these experiments confirmed this result. Analysis revealed that only the annealed (38) Tang, K.; Allman, S. L.; Chen, C. H.; Chang, L. Y.; Schell, M. Rapid Commun. Mass Spectrom. 1994, 8, 183-186.

2442 Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

Figure 3. MALDI-TOF mass spectrum of a silicon wafer with the oligodeoxynucleotide TCUC (GAATTCGAGCTCGGTACCCGG) covalently bound as described in Figure 1 and the oligodeoxynucleotide MJM6 (CCGGGTACCGAGCTCGAATTC) hybridized. Matrix solution of 3-hydroxypicolinic acid (3′-HPA; 0.7 M in 50% acetonitrile/10% ammonium citrate) was spotted onto the wafer and allowed to dry at ambient temperature. The silicon wafer was attached to the sample probe with conducting tape, and the spectrum was obtained in the positive ion mode with a reflectron. Only MJM6 (theoretical m/z of 6415) was observed with a trace amount of depurination (m/z of 6262.0); the wafer-conjugated strand TCUC (theoretical m/z of 6455) was not desorbed; thus the iodoacetamido linkage was stable enough to withstand the laser and remain intact.

strand (MJM6) was observed in the mass spectrum with an experimental mass-to-charge ratio (m/z) of 6415.4; the theoretical m/z is 6415 (Figure 3). Since there was no signal at an m/z of 6455, it was determined that the wafer-conjugated strand (TCUC) was not desorbed; thus, the iodoacetamido linkage was stable enough to withstand the laser and remain intact. There was an additional signal observed at an m/z of 6262.0. This signal results from a depurination of guanosines since it is known that DNA is susceptible to the loss of purine bases during the MALDI process.40 It has been established that matrix crystals are often not homogeneously distributed on a surface.19 Such was the case in these experiments as well; thus it was necessary to hunt for a good spot. Because of this nonhomogeneity, the mass resolution varied, but it generally ranged from 200 to 300 for the desorbed oligonucleotide in the mass spectra. In one set of experiments, noncomplementary sequences were hybridized to the wafer; following a wash as previously described, analysis by MALDITOF MS revealed that minimal nonspecific annealing had taken place since no signal was detected.

CONCLUSIONS The use of silicon wafers, a heterobifunctional cross-linker, and thiol-modified oligodeoxynucleotides has resulted in a high density of DNA molecules immobilized on a flat surface with minimal nonspecific binding. The resultant density of 250 fmol of DNA/ mm2 of surface is ∼12.5-fold higher than other postsynthesis immobilization techniques on silicon previously reported. A (39) Bai, J.; Liu, Y.; Lubman, D. M.; Siemieniak, D. Rapid Commun. Mass Spectrom. 1994, 8, 687-691. (40) Nordoff, E.; Ingendoh, A.; Cramer, R.; Overberg, A.; Stahl, B.; Karas, M.; Hillencamp, F.; Crain, P. F. Rapid Commun. Mass Spectrom. 1992, 6, 771776.

hybridization efficiency of 40% places ∼100 fmol of complementary DNA/mm2 of surface. These functionalized wafers possess a sufficient density of DNA to be successful launching pads for MALDI-TOF MS. Only the annealed strand is desorbed by the laser and detected, while the covalent linkage to the support remains intact. The advances described herein present a new possible strategy for DNA analysis, sequencing, and diagnostics. The potential exists to combine PSBH and Sanger sequencing on these silicon arrays with detection by MALDI-TOF MS. The advantage of this approach is that, when translated to an array format with detection

by mass spectrometry, the speed of analysis could be increased by a few orders of magnitude. ACKNOWLEDGMENT We gratefully acknowledge the helpful discussions and comments on the manuscript provided by Dr. Daniel P. Little. Received for review October 1, 1996. Accepted March 24, 1997.X AC961007V X

Abstract published in Advance ACS Abstracts, May 1, 1997.

Analytical Chemistry, Vol. 69, No. 13, July 1, 1997

2443