Capillary Affinity Gel Electrophoresis for Combined Size - American

An interesting new approach to capillary affinity gel electrophoresis (CAGE) has been developed for the selec- tive capture and separation of homopoly...
0 downloads 0 Views 80KB Size
Anal. Chem. 1998, 70, 1419-1424

Capillary Affinity Gel Electrophoresis for Combined Size- and Sequence-Dependent Separation of Oligonucleotides Angelika Muscate,†,| Franc¸ ois Natt,‡ Aran Paulus,*,†,§ and Markus Ehrat†

Bioanalytical Research, Drug Metabolism and Pharmacokinetics, and Nucleotide Chemistry, Pharma Research, Novartis Pharma AG, 4002 Basel, Switzerland

An interesting new approach to capillary affinity gel electrophoresis (CAGE) has been developed for the selective capture and separation of homopolymer and heteropolymer oligonucleotides. The combination of selectivity of bioaffinity recognition and high-resolution power of capillary gel electrophoresis allows the on-line sequenceand size-specific separation of oligonucleotides. Both rigid gel formulations and viscous replaceable polymer solutions having user-defined, single-stranded oligonucleotides covalently attached as recognition sequences are used. Contrary to most known affinity systems in capillary electrophoresis, which operate in a continuous mode, binding and release are accomplished in two steps, effectively separating the affinity from the separation step. At low temperature, oligonucleotides with complementary sequences in the analyte solution will bind to the immobilized recognition sequence while unrelated oligonucleotides will continue to migrate. This step is a preseparation, removing all nonspecific solutes from the sample. The release of the bound solutes is achieved at elevated temperature, allowing a probe of cross-reactivity for a given biorecognition element. Applications for highresolution separations of short oligonucleotides and their mismatches are shown, and the potential for on-line preconcentration and separation of dilute analyte solutions, thus effectively enhancing the sensitivity, is demonstrated. Affinity capillary electrophoresis (ACE) is a relatively new technique combining the selectivity of bioaffinity recognition with the high separation power of capillary electrophoresis (CE). This combination holds great potential for increased selectivity and sensitivity, which may be particularly important for the separation and detection of highly complex samples containing solutes of similar structure or the analysis and identification of compounds of interest in biological matrixes. Potential applications include immunoassays, analysis of drugs in serum samples, screening, and gene analysis. * Corresponding author (e-mail) [email protected]. † Bioanalytical Research. ‡ Nucleotide Chemistry. § Present address: Soane Biosciences Inc., 3906 Trust Way, Hayward, CA 94545. | Present address: Evotec Biosystems GmbH, Grandweg 64, D-22529 Hamburg, Germany. S0003-2700(97)01057-3 CCC: $15.00 Published on Web 02/28/1998

© 1998 American Chemical Society

Different modifications of ACE have been explored, including homogeneous assay formats such as the addition of the biospecific recognition element or its ligands to the electrophoresis buffer1, 2 and preincubation of the sample with the receptor.3,4 An alternative approach is the use of a heterogeneous assay by covalent immobilization of the biospecific recognition element. Birnbaum and Nilsson5 developed stationary affinity phases for chiral separations by cross-linking bovine serum albumin (BSA) with glutaraldehyde. With these phases, it was possible to separate the optical isomers of tryptophan. Sun et al.6 described the covalent binding of BSA to a high-molecular-weight dextran (Mr 2 000 000). This affinity polymer was used as a replaceable chiral phase for the separation of the enantiomers of leucovorin. Baba et al. were the first to describe a continuous process for the separation of oligodeoxynucleotides by a technique termed capillary affinity gel electrophoresis (CAGE).7,8 Using conjugated polyacrylamide-poly(9-vinyladenine) as the affinity phase, the electrophoretic mobility of oligothymidylic acids (pd(T)n) was selectively reduced compared to oligodeoxyadylenic acids (pd(A)n). It was postulated that adenine serves as the recognition element for oligothymidylic acids, which are specifically retained by the formation of hydrogen bonds. The affinity system resolved mismatch mixtures such as 5′-TTTATT-3′, 5′-TTTTAT-3′, and 5′TTTTA-3′9 and separated dA12-18 from dT15. The extent of hydrogen bonding was controlled via the concentration of urea, the temperature, and the concentration and length of poly(vinyladenine). All three immobilization approaches described above are continuous systems. The separation is based on a weak interaction of the analytes of interest with the affinity phase. An increase in the strength of the interaction is usually undesirable because it results in an increase in the migration time and significant band (1) Avila, L. Z.; Chu, Y. H.; Blossey, E. C.; Whitesides, G. M. J. Med. Chem. 1993, 36, 126. (2) Guttman, A.; Cooke, N. Anal. Chem. 1991, 63, 2038. (3) Schultz, N. M.; Kennedy, R. T. Anal. Chem. 1993, 65, 3161. (4) Chen, J. W.; Cohen, A. S.; Karger, B. L. J. Chromatogr. 1991, 559, 295. (5) Birnbaum, S.; Nilsson, S. Anal. Chem. 1992, 64, 2872. (6) Sun, P.; Barker, G. E.; Hartwick, R. A.; Grinberg, N.; Kaliszan, R. J. Chromatogr. 1993, 652, 247. (7) Baba, Y.; Tsuhako, M.; Sawa, T.; Akashi, M.; Yashima, E. Anal. Chem. 1992, 64, 1920. (8) Baba, Y.; Tsuhako, M.; Sawa, T.; Akashi, M. J. Chromatogr. 1993, 632, 137. (9) Akashi, M.; Sawa, T.; Baba, Y.; Tsuhako, M. J. High Resolut. Chromatogr. 1992, 15, 625.

Analytical Chemistry, Vol. 70, No. 7, April 1, 1998 1419

broadening. Therefore, the system operates well under lowselectivity conditions, inherently limiting the scope of applications. The CAGE approach of Baba et al. is designed for the separation of homomeric oligonucleotides and their mismatches only. In fact, only the separation of oligoadenylic acids has been described, severely limiting applications to the retardation of homomeric pd(T)n sequences. It is not a two-dimensional system consisting of an independent affinity and separation step, and therefore, the coupling cannot fully exploit the separation potential of each method. In addition, the affinity polymer is not exchangeable, requiring relatively clean samples in order to avoid memory effects of unwanted matrix components from run to run. Ensing et al. described the first two-step affinity system, coupling an affinity step with a capillary zone electrophoretic separation.10 After the antibodies were covalently immobilized to the inner wall of a capillary, specific binding of the antigen could be demonstrated. Only by drastically changing the buffer composition, using methanol in the elution buffer, was the antigen released and migrated toward the detector. A related approach is the immobilization of antibodies on beads which are subsequently injected in a capillary.11 The beads carrying the reactive functional groups are attached to the inner wall of an open capillary via a cross-linking agent. In a second step, biospecific recognition elements can be coupled to the beads via the reactive groups and the affinity capillary used as such. An inherent problem of these two approaches is the small surface area of the inner wall of the capillary. It limits the number of bound receptors, resulting in an overall low loadability. If the immobilized antibody denatures upon aging or harsh elution conditions or if memory effects start to arise, a new affinity capillary has to be used because the system is not replaceable. Antisense technology plays an important role in the development of novel drugs.12 A crucial step in a drug development program is the analysis of parent drug and its metabolites in biological samples. In the case of antisense compounds, single base resolution for oligonucleotides ranging from 10 to 30 bases in size is required for both the purity control of synthetic lots and metabolic studies. Different established methods exist for the high-resolution separation of oligonucleotides, including capillary gel electrophoresis13 and ion pair reversed-phase HPLC14 for size separation analysis. Nevertheless, the complexity of the samples in biological matrixes demands an improvement of the selectivity and resolution of the separation methods. In addition, the high pharmacological efficiency of antisense compounds allows low dosing in clinical trials thereby requiring analysis schemes with ever more decreasing sensitivity. On the basis of the experiences in the area of CGE and ACE, we have developed an improved use of CAGE. Representing a true two-step system, our approach to CAGE couples the high resolution power of CGE with the high specificity and sensitivity of probe hybridization. This is accomplished by using singlestranded oligonucleotides as the biospecific recognition elements (10) Ensing, K.; Oroszlan, P.; Paulus, A.; Effenhauser, C. S. EP 671626 A1 950913, 1995. (11) Gjerde, D. T.; Yengoyan, L., WO 95 10344 A1 950420, 1995. (12) Altmann, K. H.; Dean, N. D.; Fabbro, D.; Freier, S. M.; Geiger, T.; Ha¨ner, R.; Hu ¨ sken, D.; Martin, P.; Monia, B. P.; Mu ¨ ller, M.; Natt, F.; Nicklin, P.; Phillips, J.; Pieles, U.; Sasmor, H.; Moser, H. E. Chimia 1996, 50, 168. (13) Paulus, A.; Ohms, J. I. J. Chromatogr. 1990, 507, 113. (14) Huber, C. G.; Oefner, P. J.; Bonn, G. K. Anal. Biochem. 1993, 212, 351.

1420 Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

immobilized to a polymer matrix. The replaceable nature of the affinity polymer minimizes memory effects of complex biological samples and allows multitarget processing by sequentially filling a single capillary with different affinity polymers. We will show applications of this technology for high-resolution and specificity separations and demonstrate its potential for on-line preconcentration. EXPERIMENTAL SECTION Instrumentation. All CE experiments were performed on a P/ACE 5010 Beckman instrument (Fullerton, CA). The control of the temperature during the CE separations is a crucial part of the experimental setup. In the Beckman PACE system, the capillary temperature is controlled by circulating liquid in a cartridge. Unfortunately, both ends of the capillary extend out of the cartridge by 3.9 cm and are not temperature controlled. However, for the approach described here, exact temperature control at the injection end is critical. Since the injection end of the capillary rests in a buffer vial during the CE run, its temperature can be set indirectly by thermostating the inlet buffer vial, using an external water bath (Julabo, Merck ABS). For oligonucleotide hybridizing, the temperature of the water bath was set to 10 °C, for release to 60 °C. To quickly change the temperature during the binding and elution stages of the assay, the collection tray was connected to two water baths which were kept at 10 and 60 °C, respectively, with a three-way valve. Temperature calibration curves and frequent vial changes of the nontheromstated buffer vials were necessary for an acceptable temperature control because the temperature at the capillary ends slowly starts to change when the sample vials are lifted out of the thermostated sample tray. The sample vials only reach the capillary ends when they are in the “up” position. Data collection and processing were done with the P/ACE and Caesar software package (Analytical Devices, Alameda, CA), respectively. Materials. CElect N capillaries from Supelco (Buchs, Switzerland) were used in all experiments unless mentioned otherwise. The capillaries had an inner diameter of 77 µm, a total length of 24 cm, and a distance from the inlet to the detection window of 17 cm. The PVA-coated capillary with 100-µm i.d. was ordered from Hewlett-Packard (Waldbronn, Germany). All chemicals were purchased from Fluka, unless mentioned otherwise. All aqueous solutions were prepared with HPLC-quality water. The oligonucleotides were custom synthesized at MWG-Biotech (Munich, Germany), HPLC purified, and shipped in a lyophilized form. Stock solutions were prepared by redissolving the lyophilized powder in HPLC water to a final concentration of 1 × 10-2 OD/mL and stored at -20 °C. Synthesis of the Oligonucleotide Building Block. The amino oligonucleotide was prepared according to the standard protocol by the β-cyanoethylphosphoramidite strategy on an automated ABI 394 B DNA synthesizer.15 In the last stage, the support-bound amino oligonucleotide was treated with a solution of 0.25 mmol of methacrylic acid anhydride and 0.25 mmol of N-methylmorpholine in 1 mL of dichloromethane for 30 min at room temperature. The column material was then washed with dichloromethane and acetonitrile. After vacuum-drying, the support material was placed in an Eppendorf tube, 1 mL of concen(15) Beaucage, S. L.; Iyer, R. P. Tetrahedron 1992, 48, 2223.

Figure 1. Synthetic strategy for affinity polymer.

trated aqueous ammonium hydroxide was added, and the mixture was allowed to stand for 2 h at room temperature. The supernatant liquid was quickly evaporated to remove excess ammonia. The solution obtained was subjected to a chromatographic purification to remove monomers and shorter failure fragments using a NAP 10 column (Pharmacia, Upsala, Sweden). The acrylamidooligonucleotide was checked by CGE and matrixassisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS). The salt-free solution was lyophilized at room temperature and used directly in the polymerization procedure. For the preparation of the heteromeric affinity polymer, acrylic acid anhydride was used in the synthesis of the monomer block. Synthesis of the Affinity Polymer. The polymers were synthesized by polymerization of acrylamide and the monomer building block (Figure 1) in the presence of ammonium persulfate (APS) and N,N,N′,N′-tetramethylethylenediamine (TEMED). In a typical synthesis, 2.477 mL of a 3% T acrylamide solution was degassed for 15 min by bubbling a stream of helium through the solution; 7.5 mL of the dT15-modified acrylamide building block (0.1 mg/mL) was added, followed by 8 mL of a 10% (w/v) of APS and 10% (v/v) TEMED. The vial was capped, gently shaken, and placed into a sonication bath for 5 min. The reaction was allowed to proceed for 30 min at room temperature and the resulting solution transferred to a dialysis cassette (MW cutoff 10 000, Pierce, Rockford, IL), where it was dialyzed with stirring against 1 L of water for 48 h with three changes of water. The polymer was transferred to a 10-mL Falcon tube, frozen in a dry iceacetone bath and lyophilized. The weight ratio of acrylamide to monomer was 100:1. For the synthesis of the heteromeric affinity polymer, acrylic acid anhydride was used in the synthesis of the monomer building block. Preparation of a Nonreplaceable Affinity Capillary. The procedure was a modification of the method of Paulus et al.16 for the synthesis of gel capillaries for CGE. 496 µL of a 3% T solution of acrylamide solution in 100 mM (trihydroxymethyl)aminomethane] (Tris), 100 mM boric acid, pH 8.3 (100 mM TB) were degassed. Then, 1.5 µL of dT15 monomer (0.1 mg of monomer/ mL), 1.3 µL of a 10% (w/v) solution of APS, and 1.3 µL of a 10% solution (v/v) of TEMED, each in water, were added. The polymerizing solution was immediately injected by syringe into a (16) Paulus, A.; Gassmann, E.; Field, M. Electrophoresis 1990, 11, 702.

fused-silica capillary, previously silylated with methacrylic acid3-(trimethoxysilyl)propyl ester. Polymerization was allowed to continue for 24 h at room temperature. After equilibration with 100 mM TB for 30 min at low voltage (3-5 kV), a stable UV baseline and a stable current were observed. Preparation of a Replaceable Affinity Capillary. The dry polymer was dissolved at 40 °C in a buffer, typically 100 mM TB with 0-50% of organic solvents [dimethylformamide (DMF) or dimethyl sulfoxide (DMSO)] and centrifuged (model 5417R Eppendorf, Hamburg Germany) at 14 000 rpm for 5 min to remove any undissolved fine particles and air bubbles. The final concentration of the polymer was usually between 2 and 6% T. The supernatant was carefully collected and passed through the inlet site into a coated capillary (CElect N) using a pressure of 50 psi for 2 min. Unless mentioned otherwise, the same buffer composition used to dissolve the polymer was employed for the electrophoresis run. The polymer solution was equilibrated at -10 kV for 5 min to obtain a stable UV baseline and current. Melting Point Determination in Polymer Solution. The melting curve was measured at 260 nm with a UV-visible spectrophotometer (Cary 3, Varian, Palo Alto, CA) equipped with a temperature controller. The change in UV absorbance was recorded while the temperature of the sample was changed. Using an equimolar mixture of complementary oligonucleotides in 1-5% T polyacrylamide, dissolved in 100 mM TB, the samples were slowly heated from -1 to 95 °C, cooled to -1 °C, and heated again to 95 °C at a rate of 0.5 °C/min. RESULTS AND DISCUSSION Principle of CAGE. The general technique for the analysis of oligonucleotides by capillary electrophoresis is based on the use of a capillary filled with an affinity polymer having the complementary sequence of the target oligonucleotide covalently immobilized to the polymer backbone (Figure 1). The oligonucleotide mixture, containing the target oligonucleotide, is injected at low temperatures. The temperature and the electrophoresis buffer are chosen in such a way that the target oligonucleotide hybridizes to the recognition sequence while unmatched oligonucleotides continue to migrate through the polymer network. In a second step, hybridized oligonucleotides are released by raising the temperature. Due to the inherent sieving character of this affinity matrix, the migrating oligonucleotides are separated on the basis of their size differences. With temperature gradients, it is also possible to use the affinity polymer as a continuous system, separating the oligonucleotides on the basis of their degree of complementarity to the immobilized recognition sequence (Figure 2). The specificity of the probe hybridization is controlled via the stringency of the electrophoresis buffer. Typically, monovalent and divalent ions, dextran sulfate, spermine, or organic solvents are used. In addition, the temperature, concentration, and length of the complementary oligonucleotide and the length of the spacer also influence the hybridization. Considering the large electric field inherent to a CE method, the stringency was best controlled via the temperature and addition of a neutral organic modifier such as DMF or DMSO. In the absence of an electroosmotic flow (EOF), DMF or DMSO does not migrate, and subsequently, concentrations as high as 50% can be used. In contrast, the use of 100 mM to 1 M concentrations of monovalent ions is not feasible Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

1421

due to the high conductivity and consequently Joule heating of such buffers. Design Criteria of the Affinity Polymer. Immobilization procedures are often very time intensive and have to be specifically developed for each receptor. Our synthetic approach represents a general method for the synthesis with affinity-derivatized oligonucleotide polymers. A monomer building block consisting of a vinyl functionality, a spacer, and a biospecific recognition element is copolymerized with a monomer such as acrylamide or derivatives thereof. The vinyl moiety enables the incorporation of the building block into long polymer chains (Figure 1). By adjusting the ratio of acrylamide to monomer building block, the number of biospecific recognition elements in the polymer can be controlled, thereby defining capacity. A spacer was placed between the 5′ end of the oligonucleotide and the acrylamide moiety because the length of a spacer arm significantly determines the degree of hybridization of the target oligonucleotide to the probe.17 This synthetic strategy is open to all customer-defined target oligonucleotides that can be prepared on a DNA synthesizer. The spacers employed are commercially available, and the synthesis of the monomer building block follows standard protocols. This CAGE approach allows the preparation of either nonreplaceable or replaceable affinity capillaries. To manufacture nonreplaceable affinity capillaries, the polymerization is directly carried out in methacrylic acid-3-(trimethoxysilyl)propyl ester silylated capillary. The replaceable affinity polymer is prepared batchwise and redissolved in the electrophoresis buffer before use, allowing filling of the capillary with fresh polymer after each run. The analysis of complex biological samples may become feasible because of minimal carry-over. Also, the migration times in a replaceable sieving system exhibit better reproducibility because a minor but steady current decrease is seen in nonreplaceable systems due to the formation of small bubbles. The handling of the system is easy, and air bubbles are no longer a problem as often observed for gel capillaries.18 The method is flexible and easy to optimize by varying the temperature and the amount of organic solvents. Solubility problems only appeared

with use of DMF in concentrations above 30% (v/v). However, by using DMSO or mixtures of DMF and DMSO, solubility was sufficient. The lyophilized polymer is stable at room temperature for at least 12 months. Synthesis of the Affinity Polymer. The polymerization protocol is simple and based on published procedures with some minor modifications.19 By varying the 2-propanol, ammonium persulfate, and TEMED concentration, duration of sonication, temperature, reaction time, and reaction solvent, the yield of the affinity polymer could be maximized. Using the procedure as outlined in the Experimental Section, yields of 70-80% with respect to the amount of starting monomer were obtained. MALDI-TOF and CGE analysis confirmed that the oligonucleotide of the monomer building block were unaffected by the radical polymerization (data not shown). Preparation of an Affinity Capillary. Typically, a CElect N capillary was filled from the inlet site with a 3% T affinity polymer solution for 2 min at 50 psi. An equilibration of the affinity polymer was necessary before each separation. All polymer chains and unreacted monomer not completely retained in the polymer matrix start to migrate at the same time and are detected as a sharp decrease in UV absorbance at 260 nm after a few minutes. The larger the voltage applied, the faster the equilibration of the affinity polymer. After the equilibration, the UV baseline is stable for hours, indicating a stable polymer filling despite the charged nature of the polymer. It can be speculated that due to the low charge-to-mass ratio of the affinity polymer the polymer does not move at a significant speed at the time scale of the affinity separation. The CElect N coating proved to be superior to the PVA coating for this application. Several separations could be performed with a single polymer filling while in PVA capillaries the sieving power significantly decreased with an increasing run time. Mechanism of Binding. By measuring a melting curve of free dA15 and dT15 oligonucleotides in a polymer solution of a viscosity comparable to the viscosity in a replaceable matrix, the temperature necessary for binding and release of dA15 on a dT15derivatized affinity polymer could be determined. The melting curve suggests that complete binding occurs below 17 °C, complete dehybridization above 32 °C. In the CAGE system, complete hybridization (no signal!) was found up to 17 °C and elution of a sharp signal at 36 °C. In a range of 1-5% T of the polymer, the viscosity had no influence on the hybridization as determined by melting curve experiments. These data demonstrate that the interaction between the immobilized recognition element and the injected complementary target sequence is based on hybridization. Hybridization of an oligonucleotide to an immobilized complementary probe is not influenced by the high electric field of 200-600 V/cm. This finding is supported by work of Heller and Evans, who successfully used an electric field to control the hybridization of target oligonucleotides in microdevices.20 Control of Selectivity. Initially, a nonreplaceable affinity capillary was prepared. Using capillary gel electrophoresis without affinity binding for the separation of dC12-18 and dA15, overlapping signals were observed at 8 min (Figure 3a). The feasibility of

(17) Natt, F., unpublished results. (18) Bruin, G. J. M.; Paulus, A. Anal. Methods Instrum. 1995, 2, 3.

(19) Patel, D. Gel Electrophoresis, Essential Data; John Wiley & Sons: New York, 1994. (20) Heller, M.; Evans, G., WO 95 12808 A1 950511, 1995.

Figure 2. Principle of the CAGE method.

1422 Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

Figure 3. Electropherograms showing the separation of dC12-18 from dA15 in (a, top) a polyacrylamide gel-filled capillary and in (b, bottom) a nonreplaceable affinity capillary, temperatures 17 and 40 °C. CAGE conditions: 3% T of a dT15 affinity polymer, dissolved in 100 mM TB; co-injection of dC12-18 (5 × 10-4 OD/µL) each, dA15 (5 × 10-4 OD/ µL) for 5 s at -6 kV at 16 °C; applied voltage, 5 kV.

the approach was tested with the separation of dA15 from dC12-18, using dT15 as the immobilized recognition sequence. At low temperatures, dC12-18 migrated freely through the capillary and was detected after 7.1 min while dA15 was tightly bound to the affinity matrix (Figure 3b). Only upon raising the temperature of the capillary to 40 °C did the dA15 oligonucleotide start to migrate; it eluted at 3.8 min, counted after the temperature raise (Figure 3c). The separation of dA15 from a sequence carrying three mismatches is shown in Figure 4a. Due to the small dependence of the electrophoretic mobility of an oligonucleotide on the base composition,21,22 four signals were observed in a sieving matrix with conventional CGE. Using CAGE with a temperature gradient, the resolution of all four oligonucleotides is greatly improved (Figure 4b). The mismatched oligonucleotides do not bind tightly to the affinity matrix and are eluted on the basis of their decreasing affinity for the dT15 recognition sequence. The oligonucleotide with the greatest number of mismatches had the least interaction with the dT15 affinity polymer and migrated the fastest. The oligonucleotide with only one mismatch migrated the slowest because of its higher affinity to the complementary immobilized oligonucleotide. Considering that the difference of the melting points between these oligonucleotides is only 1 and 2 °C, respectively, a high selectivity can be achieved. d(TA6TA6T), d(A7TA6T), and d(A7TA7) were eluted after 7.7, 9.9, and 10.5 min, respectively. dA15 is fully complementary to the immobilized oligonucleotide and completely retained under these electrophoretic conditions. When the temperature is raised to 50 °C, the dA15 oligonucleotide is finally released, and migrates through the affinity matrix, and is detected after 4.7 min, counted after the final temperature rise. Depending on the sample composition, the temperature gradient can be adjusted to optimize the resolution of the analytes. As shown in Figure 4c, the resolution of the three mismatches could be further improved by changing the temperature gradient. (21) Guttman, A.; Nelson, R. J.; Cooke, N. J. Chromatogr. 1992, 593, 297. (22) Satow, T.; Akiyama, T.; Machida, A.; Utagawa, Y.; Kobayashi, H. J. Chromatogr. 1993, 652, 23.

Figure 4. Electropherograms showing the separation of dA15 from oligonucleotides with 1, 2, or 3 mismatches, using conventional CGE (a) and the CAGE approach (b, c). CAGE conditions: (a) 3% T of a dT15 affinity polymer, dissolved in 100 mM TB; concentration of each oligonucleotide, 5 × 10-4 OD/µL; capillary, CElect N; (b) co-injection of dA15 for 6 s at -6 kV and a mixture of d(TA6TA6T), d(A7TA6T), and d(A7TA7) for 9 s at -8 kV; (c) co-injection of dT10, dT15, and dT20 for 4 s at -2 kV and a mixture of d(TA6TA6T), d(A7TA6T), and d(A7TA7) for 9 s at -8 kV; applied voltage, 5 kV; temperature program 6.5 min at 23 °C, 6.5 min at 29 °C, 50 °C.

CAGE was also tested with heteromeric recognition sequences. An affinity polymer was prepared with the complementary sequence of 5′-AAT GCA TGT CAC AGG CGG GA-3′. The melting point of the heteromeric 20-mer duplex is 58 °C as determined by melting point measurements in a 3% T polyacrylamide solution. However, the temperature control system of the Beckman P/ACE units operates only in the range of 15-50 °C. Therefore, it was necessary to lower the duplex melting point by the addition of organic solvents to the polymer buffer. Measurements in 3% polyacrylamide solution showed that each increase in the DMF concentration by 10% (v/v) lowered the melting point by 7 °C. The separation of the heteromeric 20-mer, dT10, dT15, and dT20 is performed using 56 mM TB in 20% (v/v) DMSO and 24% (v/v) DMF as the electrophoresis buffer (Figure 5). After 9.1, 9.6, and 10.4 min, dT10, dT15, and dT20 are detected, respectively. After 13 min, the temperature is raised to 50 °C and the heteromer is eluted after a total migration time of 17.2 min. The elution of the heteromer strongly depends on the concentration of DMF and DMSO in the polymer buffer. Increasing amounts of the organic solvent lowered the melting point of the duplex, resulting in sharper elution signals. Increase of Sensitivity. Preliminary studies demonstrate the feasibility of CAGE for the preconcentration of dilute oligonucleotides. Because CAGE is a two-dimensional system, the immobilized recognition sequences concentrate the dilute target molecules in a small band as they enter the capillary during an Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

1423

filled with a number of different affinity polymers. Alternatively, the combination of an optimized size separation polymer with an affinity polymer may be desirable.

Figure 5. Electropherograms showing the separation of a heteromeric oligonucleotide from a mixture of dT10,15,20. Affinity polymer, 3.4% T in 70 mM Tris, 70 mM boric acid, pH 8.3 in 70% water (v/v) and 30% DMF (v/v); co-injection of dT10, dT15, dT20 (5 × 10-4 OD/µL each), and 5′-TCC CGC CTG TGA CAT GCA TT-3′ (1 × 10-3 OD/ µL) for 10 s at -4 kV and 20 s at -10 kV, respectively; applied voltage, 13 min at -15 kV, followed by 9 min at -8 kV.

extended injection. First experiments using a dT15 affinity polymer showed that a 1.1 × 10-8 M solution of dA15 was still detectable upon binding at low temperature and elution at 50 °C while a solution of dT15 having the same optical density as the dA15 solution was not observed. Multtarget Processing. When cross-reactivity of highly complementary sequences is not an issue, a small plug of affinity polymer at the inlet site may be sufficient. The remainder of the capillary is filled with regular polyacrylamide polymer. Using only a 1.5-cm-long plug of dT15 affinity polymer, it is possible to bind and elute dA15. This way, the method is greatly simplified because only the inlet temperature needs to be controlled. The lengths of the plugs can be adjusted with the instrument pressure system if the polymer viscosity is known. This approach has the potential for multitarget processing in a single capillary using a capillary (23) Fujimoto, C.; Fujise, Y.; Matsuzawa, E. Anal. Chem. 1996, 68, 2753.

1424 Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

CONCLUSION The results presented here suggest that CAGE with userdefined covalently bound oligonucleotide recognition sequences extends the scope of CGE with respect to selectivity and sensitivity. Future work may show that the method is not limited to the analysis of short oligonucleotides such as antisense compounds but can also be applied to the analysis of larger single-stranded DNA or RNA fragments. Our data on nonreplaceable affinity capillaries suggests that it may be feasible to develop affinity matrixes carrying biospecific recognition elements and create an EOF in an electric field. This is further supported by recent work on polyacrylamide polymers with covalently bound alkyl and sulfonate groups which were shown to exhibit an EOF and to separate neutral analytes on the basis of their hydrophobicity.23 Such approaches would greatly widen the range of applications and allow screening of DNA binding agents, determination of their binding constants, or separation of peptidic nucleic acids. ACKNOWLEDGMENT We thank Mrs. Iris Barme´, Novartis Pharma AG, for stimulating discussions on the affinity purification with oligonucleotide samples and Drs. Gerard Bruin and Jon Hall, both at Novartis Pharma AG in Basel, for critical reading and helpful suggestions to the manuscript. Received for review September 24, 1997. January 12, 1998. AC971057F

Accepted