Affinity Improvement of a VEGF Aptamer by in Silico Maturation for a

Dec 13, 2012 - List of Evaluated Oligonucleotides .... information on the composition and distribution of putative quadruplex-forming, guanine-rich se...
0 downloads 0 Views 577KB Size
Download PDF
Article pubs.acs.org/ac

Affinity Improvement of a VEGF Aptamer by in Silico Maturation for a Sensitive VEGF-Detection System Yoshihiko Nonaka,† Wataru Yoshida,† Koichi Abe,† Stefano Ferri,† Holger Schulze,‡ Till T. Bachmann,‡ and Kazunori Ikebukuro*,† †

Department of Biotechnology and Life Science, Tokyo University of Agriculture and Technology, 2-24-16 Naka-cho, Koganei, Tokyo 184-8588, Japan ‡ Division of Pathway Medicine, College of Medicine and Veterinary Medicine, The University of Edinburgh, Chancellor’s Building, Little France Crescent, Edinburgh EH16 4SB, U.K. S Supporting Information *

ABSTRACT: Systematic evolution of ligands by exponential enrichment (SELEX) is an efficient method to identify aptamers; however, it sometimes fails to identify aptamers that bind to their target with high affinity. Thus, post-SELEX optimization of aptamers is required to improve aptamer binding affinity. We developed in silico maturation based on a genetic algorithm1 as an efficient mutagenesis method to improve aptamer binding affinity. In silico maturation was performed to improve a VEGF-binding DNA aptamer (VEap121). The VEap121 aptamer is considered to fold into a Gquadruplex structure and this structure may be important for VEGF recognition. Using in silico maturation, VEap121 was mutated with the exception of the guanine tracts that are considered to form the G-quartet. As a result, four aptamers were obtained that showed higher affinity compared with VEap121. The dissociation constant (Kd) of the most improved aptamer (3R02) was 300 pM. The affinity of 3R02 was 16-fold higher than that of VEap121. Moreover, a bivalent aptamer was constructed by connecting two identical 3R02s through a 10-mer thymine linker for further improvement of affinity. The bivalent aptamer (3R02 Bivalent) bound to VEGF with a Kd value of 30 pM. Finally, by constructing a VEGF-detection system using a VEGF antibody as the capture molecule and monovalent 3R02 as the detection molecule, a more sensitive assay was developed compared with the system using VEap121. These results indicate that in silico maturation could be an efficient method to improve aptamer affinity for construction of sensitive detection systems.

V

also as unique reagents with their own properties.13−18 In previous study, some researchers reported the production of VEGF-binding aptamers19−22 and the construction of the VEGF-detection system using the aptamers.23−25 VEGF is secreted as several isoforms that are formed by alternative exon splicing. The most basic isoform is VEGF165, which has strong biological activity. A biologically weaker, secondary basic isoform of VEGF is VEGF121. The receptorbinding domain is common to both VEGF165 and VEGF121, but only VEGF165 has a heparin-binding domain. Hereafter, the VEGF referred to in this study is the VEGF165 isoform. In a previous study, we reported the production of two VEGFbinding aptamers: one aptamer specifically bound to VEGF16526 and the other (VEap121) bound to both VEGF165 and VEGF121.27 Using VEap121, we constructed a VEGF-detection system with a lower limit of detection (LOD) of 15 nM.28 The binding ability of aptamers to the target molecule is one of the most important features of aptameric biosensor production, and an improved binding ability of VEGF-binding

ascular endothelial growth factor (VEGF) is one of the key regulators of angiogenesis and vascular permeabilization. In 1989, Ferrara and Henzel identified VEGF as a diffusible endothelial mitogen from a medium conditioned by bovine pituitary follicular cells.2 VEGF is expressed in spatial and temporal association during physiological development in vivo.3 Moreover, inhibition of the VEGF activity results in the inhibition of neovascularization and tumor growth, which suggests that VEGF is a major initiator of tumor angiogenesis.4,5 VEGF is considered to be an important biomarker of several diseases such as cancer or age-related macular degeneration. Several anti-VEGF treatments have been approved by the U.S. Food and Drug Administration (FDA) for the treatment of VEGF-related diseases. However, unexpected side effects experienced during anti-VEGF treatment require management by monitoring VEGF levels in body fluids. Thus, the construction of sensitive VEGF-detection systems is important.6,7 In these studies, the cutoff value of VEGF detection was indicated to be in the pM order.8 Aptamers are selected from random sequence libraries in vitro by systematic evolution of ligands by experimental enrichment (SELEX).9,10 Aptamers have been developed for use as biosensors, not only as alternatives to antibodies11,12 but © 2012 American Chemical Society

Received: October 17, 2012 Accepted: December 13, 2012 Published: December 13, 2012 1132

dx.doi.org/10.1021/ac303023d | Anal. Chem. 2013, 85, 1132−1137

Analytical Chemistry

Article

through a 10-mer thymine linker for further improvement of aptamer affinity. Several studies were reported for the improvement of aptamers by designing bivalent or multivalent aptamers.32,33 Besides, we reported an improvement to VEGFbinding aptamers by designing a bivalent aptamer27,34 in our previous studies. Therefore, in this study, we also attempted to improve the VEGF-binding aptamer by designing a bivalent aptamer. The improved aptamer was used to construct a VEGF-detection system.

aptamers is necessary to achieve sensitive VEGF detection. In general, LOD is related to the dissociation constant (Kd) value of the molecular recognition element. A VEGF-detection system, which can detect pM levels of VEGF, is required for practical usage. Therefore, to construct such a detection system, VEGF aptamers with Kd values in the pM order are required. Systematic evolution of ligands by experimental enrichment is an efficient method of obtaining target-binding aptamers. However, SELEX may fail to screen for aptamers with highbinding ability because of the following reasons. The diversity of the randomized sequence pool is limited in experimental manipulation during SELEX.29 In many cases, the concentration of the sequence pool is less than 1 μM (equal to 1.0 × 10−6 M) and the experimental manipulation is performed in a 1.5 mL (equal to 1.5 × 10−3 L) sampling tube. Therefore, the manipulated sample could contain 1.5 × 10−9 mol (equal to 1.0 × 1015 molecules) of nucleotides. On the other hand, the diversity of the 30-mer random library is 430 (approximately equal to 1018). Thus, only one-thousandth of all candidates is evaluated in common SELEX operation. In addition, the amplification efficiency of PCR is affected by the secondary structure of oligonucleotides.30,31 Aptamers usually have a particular secondary structure that is necessary for recognition of their specific target molecules, and oligonucleotides with higher order structures may not be easily amplified by PCR in SELEX. Thus, the efficiency of obtaining such target-binding sequences is reduced through SELEX. Therefore, post-SELEX screening methods are required to obtain aptamers, which bind to their target molecules with high affinity. Several post-SELEX optimization methods have been reported, including bivalent aptamers,27,32−34 photoaffinity aptamers,35 and labeling site characterization.36−38 Using these methods, a number of sequence mutations or chemical modifications were randomly introduced into aptamers, and the functions of the mutated aptamers were evaluated. The binding affinities of aptamers improved using these post-SELEX optimization methods, which suggested that the post-SELEX optimization could overcome the experimental limitations of SELEX. We also reported a post-SELEX screening method that we designated as “in silico maturation.”1,39−42 In silico maturation involves the following two steps. The first step is the generation of mutant aptamers from parent aptamers using a genetic algorithm, and the second step is the in vitro analysis of the aptamer function to select parent aptamers for the next generation. Improved aptamers can be obtained by repeating the two steps. In the first step, mutant aptamer sequences are identified using the following four steps: (1) the parent aptamers are replicated in different ratios according to the functional ability of their parent aptamers; (2) a pair of sequences is randomly selected from the parental sequences; (3) the pair of sequences is crossed at one or more random points; (4) point mutations are randomly introduced into the obtained sequences. In this method, all of the candidate sequences are produced in silico, whereas the synthesis and functional evaluation of all candidates are performed in vitro. There is no PCR bias in this method because all the candidates can be synthesized chemically without PCR amplification. Thus, in silico maturation facilitates the efficient exploration of a large sequence space. In this study, the affinity of a VEGF-binding aptamer was improved by in silico maturation. Moreover, a bivalent aptamer was produced by connecting two identical improved aptamers

■

EXPERIMENTAL SECTION Materials. All synthetic oligonucleotides were purchased from Greiner Bio-One (Japan). Recombinant human VEGF165 (expressed in Sf21 insect cells) and biotinylated anti-VEGF antibody (BAF293) were purchased from R&D systems (U.S.A.). Bovine serum albumin (BSA) was purchased from SIGMA-Aldrich (Japan). Nitrocellulose membrane (Amersham Hybond-ECL) was purchased from GE Healthcare (U.S.A.). Immobilon Western chemiluminescent Horseradish peroxidase (HRP) substrate was purchased from Merck Millipore (U.S.A.). All other chemical reagents used were of analytical grade. Production of Sequential Mutants by in Silico Maturation. During the first cycle of sequence control with in silico maturation, several mutations were introduced into the sequence of the VEGF-binding aptamer (VEap121: 5′TGTGGGGGTGGACGGGCCGGGTAGA-3′) that was expected to fold into a G-quadruplex structure.27 The sequence of VEap121 was randomly mutated, with the exception of the guanine bases in order to maintain the G-quadruplex structure. Ten VEap121 mutants were produced (Figure S-1, Supporting Information). The VEGF-binding abilities of these 10 mutants were evaluated by a surface plasmon resonance (SPR) assay as described below. Seven of the mutants were found to bind to VEGF and these mutants were ranked according to their Kd values. To produce the second generation by in silico maturation, the top five mutants from the first generation were selected, and the sequences of these mutants were replicated with different appearance rates depending on the Kd ranking. Random pairs of sequences were chosen from among the replicated sequences. Each pair of sequences was crossed over at one random point and two single-base mutations were randomly introduced to produce a set of 20 new sequences for the next generation (Figure S-2, Supporting Information). To produce the third generation, the top five mutants were selected from among the mutants evaluated in the first and second generation. The five mutants were crossed with VEap121 at a random point. Two single point mutations were randomly introduced into the sequences. Repetition of this process produced a third generation of 20 sequences (Figure S-3, Supporting Information). Surface Plasmon Resonance Measurement. The binding ability of oligonucleotides to VEGF was evaluated by measuring SPR on CM5 sensor chips in a Biacore T200 (GE Healthcare Japan, Japan). VEGF was dissolved in 10 mM acetate buffer (pH 6.0) and VEGF (approximately 1000 RU) was covalently attached to a CM5 sensor chip using an aminecoupling kit. Nonlabeled oligonucleotides (1 μM) were dissolved in Tris-buffered saline (TBS: 10 mM Tris-HCl, 100 mM NaCl, 5 mM KCl, pH 7.4), heated at 95 °C for 5 min, and then gradually cooled. Under these conditions, each oligonucleotide would fold into its specific secondary structure. Various concentrations of oligonucleotides were diluted in TBS and 1133

dx.doi.org/10.1021/ac303023d | Anal. Chem. 2013, 85, 1132−1137

Analytical Chemistry

Article

injected onto the VEGF-immobilized CM5 chip at a flow rate of 20 μL/min at 20 °C. The binding between the immobilized VEGF and the oligonucleotides was monitored. The CM5 chip was regenerated by injection of NaCl (1 M). The Kd values were calculated by fitting the association and dissociation rates using the Biacore evaluation software. Evaluation of VEGF-Binding Aptamers by Circular Dichroism. The aptamer structures were analyzed by measuring the circular dichroism (CD) spectra using a J-720 spectropolarimeter (JASCO, U.S.A.). All aptamers [final concentration (f.c.) 10 μM] were diluted in TBS and folded by heat treatment as described above. The spectra were recorded in a 0.1 cm cell at 20 °C, and each spectrum was the average of 20 scans. Construction of VEGF-Detection System Using Improved Aptamers. A VEGF-detection system was constructed using the improved aptamers. The aptamers were modified by the addition of FITC at the 5′-end. Biotinylated anti-VEGF antibody, BAF293 (supplied in a solution of 20 mM Tris buffer containing 150 mM NaCl, pH 7.3) was prepared with TBS, at a f.c. of 40 ng/mL. The diluted BAF293 (100 μL) was added to a streptavidin-immobilized 96-well polystyrene plate (Nunc Immobilizer #436015, Nunc). The polystyrene plate was incubated with gentle shaking for 30 min, and the supernatant was removed. After washing, each well was filled with 100 μL of 5-fold-diluted blocking reagent, N101 (Nitiyu, Japan), containing 500 nM of biotin, and incubated for 30 min. The blocking reagent (a synthetic polymer-based reagent) was added to reduce nonspecific absorption of proteins on the polystyrene plate. Biotin was added in order to block excess amounts of streptavidin on the polystyrene plate. The blocking reagent was then removed from each well. After washing, a total of 100 μL of various concentrations of proteins (VEGF or BSA) was added to the well and incubated for 30 min. The well was then washed three times. The FITC-labeled aptamer (100 μL, 10 nM), prepared in TBS after heat treatment for folding, was added to each well and incubated for 30 min. After three washes, HRP-conjugated anti-FITC antibody was added to each well and incubated with gentle shaking for 30 min. Each well was washed, and Immobilon Western chemiluminescent Horseradish peroxidase (HRP) substrate was added. The HRP activity was measured by a multilabel plate counter (Wallac 1420 ARVO MX, Perkin-Elmer). Wells were washed with TBST (TBS containing 0.05% (v/v) Tween 20) throughout this assay. All experimental procedures were performed at room temperature.

produced by introducing random mutations into the VEap121 sequence. Guanine bases were left unaltered because these were considered necessary for VEap121 to fold into a G-quadruplex structure (The production scheme for the first generation is shown in Figure S-1 (Supporting Information) and the sequences of all mutants produced are listed in Table 1.). After synthesis, the VEGF-binding ability of the mutants was evaluated by SPR. The oligonucleotides were ranked according to their Kd value. The VEGF-binding ability of each oligonucleotide in each generation is shown in Figure 1. Seven of the ten first generation VEap121 mutants showed VEGF-binding ability. The top five mutants were selected to act as the parents for the second generation of in silico maturation. To produce the second generation, the top five sequences from the first generation were replicated into 40 sequences with the difference ratio dependent on the ranking of VEGF-binding ability (The production scheme for the second generation is illustrated in Figure S-2, Supporting Information). The ratios are as follows: Top is twelve fortieth, second is ten fortieth, third is eight fortieth, fourth and fifth are five fortieth. Two sequences were randomly chosen from the replicated pool and then paired. A random point was chosen on the sequences and the sequences were crossed at this point. Finally, a sequence was randomly chosen from the sequence pair and two singlebase random mutations were introduced. Repetition of this process yielded 20 sequences as the second generation. Evaluation by SPR revealed that five of the mutants showed VEGF-binding ability whereas the remainder did not. However, the binding abilities of the five mutants were weaker than that of VEap121. These results indicated that VEap121 could be relatively optimized to interact with VEGF. Thus, 20 sequences were produced by crossing VEap121, and the top five mutants, which showed binding ability in the first and second generation, were selected for the third generation (The production scheme for the third generation is illustrated in Figure S-3, Supporting Information). Each of the mutants was replicated into two sequences and the replicated sequence was crossed with VEap121 at a random point. Following this, two single-base mutations were introduced into each of the crossed sequences. By repeating this process, 20 sequences were produced as the third generation. From the third generation, four oligonucleotides were identified (3R02, 3R03, 3R08, and 3R09), which bound to VEGF more strongly than VEap121 (Figure 1, Table 1). The strongest binding ability to VEGF was shown by 3R02 (5′TGTGGGGGTGGACTGGGTGGGTACC-3′). The Kd value for 3R02 was 300 pM, indicating that it had a 16-fold stronger binding ability than VEap121. The dissociation rate constant (koff) for 3R02 was 1.92 × 10−4 (M−1), whereas the koff for VEap121 was 4.99 × 10−3 (M−1) (Figure 2). The binding rate constant (kon) for 3R02 was 6.39 × 105 (M−1 s−1), whereas that for VEap121 was 1.05 × 106 (M−1 s−1). Therefore, the improvement of VEGF-binding ability was accomplished by reducing the dissociation rate constant. By evaluating a limited number of candidates, new aptamers were obtained which were missed by SELEX. This indicates that the post-SELEX screening procedure, which is based on in silico maturation is an efficient method to improve the function of aptamers. The TGTG motif was found in the 5′ terminal sequences of 3R02, 3R03, 3R08, 3R09, and VEap121. This motif was also found in 3R01 but affinity of the aptamer was weaker than that of VEap121. Other oligonucleotides that did not contain the TGTG motif also showed lower VEGF-binding ability than

■

RESULTS AND DISCUSSION Improvement of VEGF Aptamer-Binding Ability by in Silico Maturation. Aptamers with high VEGF-binding ability are required for use as recognition elements in VEGF-detection systems. The Kd value of a VEGF aptamer (VEap121) that was identified in a previous study27 was re-evaluated by an SPR assay (the condition is described in the Experimental Section). As the calculated Kd value of VEap121 was 4.7 nM, it was evident that the binding ability of VEap121 required improvement in order to construct a sensitive VEGF-detection system. Therefore, we attempted to improve the binding ability of VEap121 by in silico maturation. For in silico maturation, several sequences are required to act as parents (templates) from which the mutants will be produced. Mutated VEap121 sequences were used as parents for the first generation of in silico maturation: a set of 10 oligonucleotide mutants was 1134

dx.doi.org/10.1021/ac303023d | Anal. Chem. 2013, 85, 1132−1137

Analytical Chemistry

Article

Table 1. List of Evaluated Oligonucleotidesa name

sequence (5′ to 3′)

Kd

VEap121 1R01 1R02 1R03 1R04 1R05 1R06 1R07 1R08 1R09 1R10 2R01 2R02 2R03 2R04 2R05 2R06 2R07 2R08 2R09 2R10 2R11 2R12 2R13 2R14 2R15 2R16 2R17 2R18 2R19 2R20 3R01 3R02 3R03 3R04 3R05 3R06 3R07 3R08 3R09 3R10 3R11 3R12 3R13 3R14 3R15 3R16 3R17 3R18 3R19 3R20

TGTGGGGGTGGACGGGCCGGGTAGA TGAGGGGGCGGCGGGGCCGGGCAGG GGGGGGGGTGGCTGGGCGGGGTGGG AGAGGGGGGGGTCGGGGTGGGTCGC TGAGGGGGTGGGTGGGACGGGTTGG AGGGGGGGGGGAGGGGTCGGGGTGG CGAGGGGGTGGAAGGGCGGGGCTGT AGAGGGGGTGGCCGGGGCGGGCTGC TGCGGGGGCGGCAGGGCAGGGTAGG AGCGGGGGGGGCTGGGCTGGGTGGG TGAGGGGGGGGGAGGGCGGGGCTGC TGAGGGGGCGGAGTGGCCGGGGTGG TGAGGGGGTGGGCCGGCCGGGTAGA TGTCGGGGTGGACGCGCCGGGTAGA AGGGAGGGGGGCTGGGCTGGGTGTG TGAGGGGGTGGACAGGCCGGGTAGT AGGGGGGGGGGACGGGACGGGTATA AGGGGGGGGGGTGGGGGTGGGTCCC TGTGGGGTTGGACGGACCGGGTAGG AGAGGGGCGGGACGCGCCGGATAGA TGTGGGGTTGGACGGGCCGTGTAGA TGTGGGGGTGCGCGGGCCGGGCAGG TGGGGGGTTGGACGGGCGGGGCTGC TATGGGGGTGGACGGGCCGGGCAGC TGAGGGGGCGGCCGGGCGGGGCAGC AGAGTGGGCGGCGAGGCCGGGCAGG TGACGGGGGGGACGGGCCGGGTAGT TGAGGGGGGGGGAGGTCGGGCCTGC AGAGGGGGGGGTGGGGGTGGCCTGC TCTGGGGGTGTACGGGCCGGGTAGA TATGGGGGTGGACGGGGCGGGTCGC TGTGGGGGAGGACGGGCGGGGATGC TGTGGGGGTGGACTGGGTGGGTACC TGTGGGGGTGGACGGGCCGAGTACG TGTGCGGGGGGTGGGGGTGGTTCCC TGTTGGGGCGGACGGGCCGGGCAGG TGTGGGGGGGGTGGGGGTGGACCGC TGTGCGGGTGGCCGGGGTGGCCTGC TGTGGGGATGGATGGGCCGGGCTGC TGTGGGGGTGGTTGGGCGGGGCTGC TGTGGGGCTGGGAGGTCGGGCCTGC TGGAGGGTTGGACGGGCGGGGCAAA AGGGGGGGGGGTAAGGGTGGGTAGA AGGGGGTGTGGACGGGCCGGGTAAA AGGGGGAGGGGTGGGGCCGGGCAGA TAAGGGGGCGGCGGGGCCGGGCATA AGTGGGGGGGGTGGAGCCGGGTAGA AGAGGGGGGGGACGGCCCGGGTAGA AGAGGGGGGGGTGGAGGGGGCTAGA TGATGGGGGGGACGGGCCGGGTATA TGAGCGAGGGGGCGGGCCGGGTAGA

4.7 nM 24 nM 81 nM 56 nM 180 nM 40 nM n.d. n.d. n.d. 60 nM 31 nM n.d. n.d. n.d. n.d. n.d. 36 nM 23 nM n.d. n.d. n.d. n.d. 14 nM n.d. n.d. n.d. n.d. 27 nM 36 nM n.d. n.d. 8.8 nM 300 pM 1.7 nM n.d. n.d. n.d. n.d. 1.5 nM 2.4 nM n.d. n.d. 76 nM n.d. 27 nM n.d. 790 nM 5.0 μM 360 nM n.d. n.d.

Figure 1. VEGF-binding ability of the oligonucleotides from three generations of in silico maturation.

Figure 2. Kinetic plot of VEGF-binding aptamers.

QGRS Mapper to predict which guanine bases were important for folding into the G-quadruplex structure43 (Table S-1, Supporting Information). QGRS Mapper is an online tool used to generate information on the composition and distribution of putative quadruplex-forming, guanine-rich sequences. As the result of this analysis, VEap121, 3R01, 3R02, 3R08, and 3R09 were expected to fold into similar G-quadruplex structures (G1: 5′-NNNNGGNNNGGNNNGGNNNGGNNNN-3′). The guanine bases, which are shown as bold text, were predicted to be involved in the formation of the G-quadruplex. The oligonucleotide 3R03 also bound to VEGF but it was expected to fold into a different G-quadruplex structure (G2: 5′NNNGGNGGNGGNNGGNNNNNNNNNN-3′). Oligonucleotides, which had the TGTG motif but did not bind to VEGF, were expected to fold into G-quadruplex structures that differed from G1 or G2. From these results, we concluded that the 5′-TGTGGGNNNGGNNNGGNNNGGNNNN-3′ motif may be important for the VEGF-binding ability of the aptamers. The X-axis shows the ranking of oligonucleotides in each generation based on Kd. The Y-axis shows the binding constant for each oligonucleotide. Oligonucleotides with no binding ability for VEGF are not shown in this figure. All kinetic parameters were calculated by measuring the SPR signal. The X-axis shows the dissociation rate constant and the Y-axis shows the association rate constant. Improvement of Binding Ability of the VEGF Aptamer by Dimerization. In previous studies, we reported an improvement to VEGF-binding aptamers by designing a bivalent aptamer.27,34 Therefore, in this study, we also attempted further improvement of 3R02 by designing a bivalent aptamer. The sequence of the designed bivalent aptamer (3R02 Bivalent) is 5′- TGTGGGGGTGGACTGGGTGGGTACCTTTTTTTTTTTGTGGGGGTGGACTGGGTGGGTACC-3′. 3R02 Bivalent is composed of two monomeric 3R02 oligonucleotides and a 10-mer thymine linker sequence. The Kd value of 30 pM for 3R02 Bivalent was

a

Sequences of evaluated oligonucleotides and their dissociation constants (Kd) against VEGF. “n.d.” means that no binding signal was observed in the measurement. The dissociation constants were calculated by measuring SPR signal using VEGF-immobilized sensor chip.

VEap121. Therefore, it was assumed that the TGTG motif was important for the VEGF-binding ability of the aptamers. In this study, 50 sequences were evaluated, 12 of which had the TGTG motif in their 5′ end. These 12 sequences were analyzed by 1135

dx.doi.org/10.1021/ac303023d | Anal. Chem. 2013, 85, 1132−1137

Analytical Chemistry

Article

on the concentration of VEGF. LOD of this assay was 6.7 nM (Figure 4a) and was calculated as follows: LOD = 3(SD/slope).

calculated by measuring SPR. A 10-fold improvement in binding ability was achieved by dimerization of 3R02. This is the highest binding ability among previously reported VEGFbinding DNA aptamers, and approaches the binding ability of the VEGF-binding RNA aptamer Macugen, which is known for the anti-VEGF therapeutic drug.44 CD Spectrum Measurement of 3R02 and 3R02 Bivalent. The CD spectra are useful for gaining information concerning the quadruplex structures of DNAs. The CD spectra of VEap121, 3R02, and 3R02 Bivalent were obtained: VEap121 had two negative peaks at 239 and 280 nm and two positive peaks at 262 and 293 nm (Figure 3a); 3R02 had two

Figure 4. Detection of VEGF by plate assay using 3R02 (a) or VEap121 (b).

“SD” was defined as the standard deviation of the response at 0 nM of VEGF. “Slope” was defined as the slope of the calibration curve. No signal increase was observed when BSA was added onto the plate. The plate assay was also performed using VEap121 (Kd = 4.7 nM) (Figure. 4b). Horseradish peroxidase chemiluminescence signals were obtained which indicated binding between the aptamer and VEGF, but the signal was one-eighth of that obtained with 3R02. Furthermore, the signal appeared to saturate at a VEGF concentration of 10 nM, and the error bars of each data set overlapped. Thus, it was not possible to calculate LOD for VEap121. These results indicated that the sensitivity of the VEGF-detection system was improved by using 3R02. A VEGF-detection system was also constructed using 3R02 Bivalent (Kd = 30 pM) but the results were similar to those obtained for 3R02 (data is shown in Figure S-5). It was assumed that the antibody would interrupt the binding between 3R02 Bivalent and VEGF. Therefore, we believe that we could improve the sensitivity of the detection system by using a VEGF antibody that binds to a different epitope. The anti-VEGF antibody was immobilized to each well of a polystyrene plate by avidin−biotin interaction. Various concentrations of VEGF in TBS were added to each well. A solution of FITC-labeled 3R02 or VEap121 was added to each well. The chemiluminescence from the HRP-conjugated antiFITC antibody, which represented the binding of FITC-labeled 3R02 against VEGF, was detected (n = 3). BSA was used as a control protein. Counts per second (CPS) is an expression of chemiluminescence intensity.

Figure 3. CD spectra of VEGF-binding aptamers.

negative peaks at 238 and 271 nm and two positive peaks at 256 and 288 nm (Figure 3a). The intensity of the positive peak, located near 260 nm in the CD spectrum for 3R02, was twice the height of that for VEap121. Both VEap121 and 3R02 were expected to have the same G-quadruplex motif (G1: 5′NNNNGGNNNGGNNNGGNNNGGNNNN-3′) from analysis using QGRS Mapper; however, data obtained from the CD spectra suggested that their G-quadruplex structures were different. The CD spectrum of 3R02 was similar to the spectrum for d(GGGGTCAGGCTGGGGTTGTGCAGGTC), which folds into an antiparallel G-quadruplex.45 Therefore, it was assumed that 3R02 may adopt a similar antiparallel Gquadruplex structure. A CD spectrum was also obtained for 3R02 Bivalent. Because 3R02 Bivalent had two 3R02 monovalent sequences and a 10-mer thymine linker (T10), the T10 spectrum was subtracted from the 3R02 Bivalent spectrum (raw spectra are shown in Figure S-4, Supporting Information). The subtracted spectrum is shown in Figure 3b. The subtracted spectrum showed a similar band pattern to that of 3R02 but the peak intensity of the spectrum was twice as high as that of 3R02. Therefore, 3R02 Bivalent may contain two independent monomeric 3R02 G-quadruplex structures. Figure 3a shows the difference between the parent sequence (VEap121) and the sequences improved by in silico maturation (3R02). Figure 3b shows the difference between spectra for 3R02 and 3R02 Bivalent. The raw CD spectrum of T10 was subtracted from the raw CD spectrum of 3R02 Bivalent. The subtracted spectrum is shown in this figure. The X-axis shows wavelength and the Y-axis shows mean residue molar ellipticity. Construction of a VEGF-Detection System. We attempted to improve the sensitivity of the VEGF-detection system by using the improved aptamers. A VEGF-detection system was constructed using an anti-VEGF antibody and 3R02 (Kd = 300 pM). The antibody was immobilized on the plate and then VEGF was added. Next, FITC-labeled 3R02 was added to the plate. The FITC-labeled 3R02 was detected using an HRP-conjugated anti-FITC antibody. This resulted in the production of an HRP chemiluminescence signal that represented the binding of 3R02 to VEGF and was dependent

■

CONCLUSIONS In this study, in silico maturation of a VEGF-binding aptamer was carried out in order to obtain an improved VEGF-binding aptamer that bound more strongly to VEGF. As a result of the improvement process, we obtained an aptamer that showed a 16-fold higher VEGF-binding ability than that of the parent aptamer (VEap121, Kd = 4.7 nM) that was selected by SELEX. The Kd value of the improved aptamer (3R02) was calculated as 300 pM by SPR. The 3R02 Bivalent aptamer with a Kd value of 30 pM was also designed. A sensitive VEGF-detection system was constructed using 3R02, indicating that in silico maturation is an efficient method to improve aptamer affinity for the production of a sensitive detection system.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. 1136

dx.doi.org/10.1021/ac303023d | Anal. Chem. 2013, 85, 1132−1137

Analytical Chemistry

■

Article

(27) Nonaka, Y.; Sode, K.; Ikebukuro, K. Molecules 2010, 15, 215− 225. (28) Nonaka, Y.; Abe, K.; Ikebukuro, K. Electrochemistry 2012, 80, 363−366. (29) Klug, S. J.; Famulok, M. Mol. Biol. Rep. 1994, 20, 97−107. (30) Polz, M.; Cavanaugh, C. Appl. Environ. Microb. 1998, 64, 3724− 3730. (31) Kanagawa, T. J. Biosci. Bioeng. 2003, 96, 317−323. (32) Kim, Y.; Cao, Z.; Tan, W. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 5664−5669. (33) Mallikaratchy, P. R.; Ruggiero, A.; Gardner, J. R.; Kuryavyi, V.; Maguire, W. F.; Heaney, M. L.; McDevitt, M. R.; Patel, D. J.; Scheinberg, D. A. Nucleic Acids Res. 2011, 39, 2458−2469. (34) Hasegawa, H.; Taira, K.; Sode, K.; Ikebukuro, K. Sensors 2008, 8, 1090−1098. (35) Smith, D.; Collins, B. D.; Heil, J.; Koch, T. H. Mol. Cell. Proteomics. 2003, 2, 11−18. (36) Famulok, M. J. Am. Chem. Soc. 1994, 116, 1698−1706. (37) Tsiang, M.; Gibbs, C. S.; Griffin, L. C.; Dunn, K. E.; Leung, L. L. K. J. Biol. Chem. 1995, 270, 19370−19376. (38) Zhang, D.; Lu, M.; Wang, H. J. Am. Chem. Soc. 2011, 133, 9188−9191. (39) Ikebukuro, K.; Okumura, Y.; Sumikura, K.; Karube, I. Nucleic Acids Res. 2005, 33, e108. (40) Noma, T.; Ikebukuro, K. Biochem. Biophys. Res. Commun. 2006, 347, 226−231. (41) Noma, T.; Sode, K.; Ikebukuro, K. Biotechnol. Lett. 2006, 28, 1939−1944. (42) Ikebukuro, K.; Yoshida, W.; Noma, T.; Sode, K. Biotechnol. Lett. 2006, 28, 1933−1937. (43) Kikin, O.; D’Antonio, L.; Bagga, P. S. Nucleic Acids Res. 2006, 34, W676−682. (44) Ruckman, J.; Green, L. S.; Beeson, J.; Waugh, S.; Gillette, W. L.; Henninger, D. D.; Claesson-Welsh, L.; Janjić, N. J. biol. Chem. 1998, 273, 20556−20567. (45) Wen, J. D.; Gray, D. M. Biochemistry 2002, 41, 11438−11448.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Fax: +81-42-388-7030. Phone: +81-42-388-7030. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the Industrial Technology Research Grant Project in 2009 from the New Energy and Industrial Technology Development Organization (NEDO) of Japan for K.I. This work was also supported by Grant-in-Aid for JSPS (Japan Society for the Promotion of Science) Fellows for Y.N.

■

REFERENCES

(1) Savory, N.; Abe, K.; Sode, K.; Ikebukuro, K. Biosens. Bioelectron. 2010, 26, 1386−1391. (2) Ferrara, N.; Henzel, W. J. Biochem. Biophys. Res. Commun. 1989, 161, 851−858. (3) Peirce, S. M.; Price, R. J.; Skalak, T. C. Am. J. Physiol. Heart Circ. Physiol. 2004, 286, H918−925. (4) Kim, K. J.; Li, B.; Winer, J.; Armanini, M.; Gillett, N.; Phillips, H. S.; Ferrara, N. Nature 1993, 362, 841−844. (5) Kondo, S.; Asano, M.; Matsuo, K.; Ohmori, I.; Suzuki, H. Biochim. Biophys. Acta 1994, 1221, 211−214. (6) Stacker, S. A.; Baldwin, M. E.; Achen, M. G. FASEB J. 2002, 16, 922−934. (7) Rini, B. I.; Michaelson, M. D.; Rosenberg, J. E.; Bukowski, R. M.; Sosman, J. A.; Stadler, W. M.; Hutson, T. E.; Margolin, K.; Harmon, C. S.; DePrimo, S. E.; Kim, S. T.; Chen, I.; George, D. J. J. Clin. Oncol. 2008, 26, 3743−3748. (8) Tolentino, M. Surv. Ophthalmol. 2011, 56, 95−113. (9) Tuerk, C.; Gold, L. Science 1990, 249, 505−510. (10) Ellington, A. D.; Szostak, J. W. Nature 1990, 346, 818−822. (11) Ikebukuro, K.; Kiyohara, C.; Sode, K. Anal. Lett. 2004, 37, 2901−2909. (12) Ikebukuro, K.; Kiyohara, C.; Sode, K. Biosens. Bioelectron. 2005, 20, 2168−2172. (13) Potyrailo, R. A.; Conrad, R. C.; Ellington, A. D.; Hieftje, G. M. Anal. Chem. 1998, 70, 3419−3425. (14) Lee, M.; Walt, D. R. Anal. Biochem. 2000, 282, 142−146. (15) Bruno, J. G.; Kiel, J. L. Biotechniques 2002, 32, 178−183. (16) Xiao, Y.; Lubin, A. A.; Heeger, A. J.; Plaxco, K. W. Angew. Chem., Int. Ed. 2005, 44, 5456−5459. (17) Yoshida, W.; Sode, K.; Ikebukuro, K. Anal. Chem. 2006, 78, 3296−3303. (18) Ogasawara, D.; Hachiya, N. S.; Kaneko, K.; Sode, K.; Ikebukuro, K. Biosens. Bioelectron. 2009, 24, 1372−1376. (19) Carrasquillo, K. G.; Ricker, J. A.; Rigas, I. K.; Miller, J. W.; Gragoudas, E. S.; Adamis, A. P. Invest. Ophthalmol. Visual Sci. 2003, 44, 290−299. (20) Burmeister, P. E.; Lewis, S. D.; Silva, R. F.; Preiss, J. R.; Horwitz, L. R.; Pendergrast, P. S.; McCauley, T. G.; Kurz, J. C.; Epstein, D. M.; Wilson, C.; Keefe, A. D. Chem. Biol. 2005, 12, 25−33. (21) Ng, E. W. M.; Shima, D. T.; Calias, P.; Cunningham, E. T.; Guyer, D. R.; Adamis, A. P. Nat. Rev. Drug Discov. 2006, 5, 123−132. (22) Nick, T. J.; Darugar, Q.; Kourentzi, K.; Willson, R. C.; Landes, C. F. Biochem. Biophys. Res. Commun. 2008, 373, 213−218. (23) Zhao, S.; Yang, W.; Lai, R. Y. Biosens. Bioelectron. 2011, 26, 2442−2447. (24) Zhao, J.; He, X.; Bo, B.; Liu, X.; Yin, Y.; Li, G. Biosens. Bioelectron. 2012, 34, 249−252. (25) Freeman, R.; Girsh, J.; Jou, A. F.; Ho, J. A.; Hug, T.; Dernedde, J.; Willner, I. Anal. Chem. 2012, 84, 6192−6198. (26) Hasegawa, H.; Sode, K.; Ikebukuro, K. Biotechnol. Lett. 2008, 30, 829−834. 1137

dx.doi.org/10.1021/ac303023d | Anal. Chem. 2013, 85, 1132−1137