Unimolecular Beacons for the Detection of DNA-Binding Proteins

Jan 22, 2004 - 1402 South Grand Boulevard, St. Louis, Missouri 63104. A new methodology for detecting sequence-specific DNA- binding proteins has been...
0 downloads 0 Views 220KB Size
Anal. Chem. 2004, 76, 1156-1164

Unimolecular Beacons for the Detection of DNA-Binding Proteins Eric Knoll and Tomasz Heyduk*

Edward A. Doisy Department of Biochemistry and Molecular Biology, St. Louis University Medical School, 1402 South Grand Boulevard, St. Louis, Missouri 63104

A new methodology for detecting sequence-specific DNAbinding proteins has been recently developed (Heyduk, T.; Heyduk, E. Nat. Biotechnol. 2002, 20, 171). The core feature of this methodology is protein-dependent association of two fluorochrome-labeled DNA fragments, which allows generation of a fluorescence signal reporting the presence of the target protein. Previous kinetic experiments identified the association of the two DNA fragments as the rate-limiting step of the assay. Here we report on a variant of the assay, in which components of the assaysfluorescent DNA fragmentsswere covalently tethered by a non-DNA linker with the goal of increasing the rate of association of the two fragments. We investigated the effect of the tether on the performance of the assay under a variety of conditions using a model DNAbinding protein. Quantitative titrations and rapid kinetic stopped-flow experiments were conducted to validate the molecular model that describes the two linked equilibria: oscillation of the tethered construct between the open and closed states and the exclusive association of the protein with the closed state. Experiments were also performed to demonstrate the ability of these tethered constructs to signal when attached to a solid surface. The major advantage of this new assay format is the faster response time for the detection allowing the higher throughput of the analysis. Additionally, it will be possible to attach tethered beacons to other solid surfaces, thus allowing the preparation of arrays containing molecular beacons for many different DNA-binding proteins. The cellular levels of specific proteins are commonly used as diagnostic markers for many diseases, which makes determination of the level or activity of a specific protein one of the most common experimental procedures performed by biomedical researchers. Current methodologies for detecting the levels and activities of proteins rely heavily on the use of antibodies developed to recognize specific proteins. These methodologies often involve multistep lengthy assay protocols and are thus not amenable to high-throughput analyses. Alternative ideas for detecting and quantifying specific proteins are clearly needed. We have previously described a novel methodology, molecular beacons for * Corresponding author. E-mail: [email protected]. Fax: (314) 577-8156.

1156 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

detecting DNA-binding proteins,1 which allows for the generation of reporter molecules capable of detecting the presence of a specific DNA-binding protein in a rapid and sensitive manner compatible with high-throughput analyses. The molecular beacon assay for detecting DNA-binding proteins involves splitting a DNA sequence defining a protein-binding site into two fragments, each containing about half of the proteinbinding site.1 The fragments are generated such that they include short complementary overhangs, which allow for some propensity for the fragments to associate. However, conditions of the assay are selected such that in the absence of the protein very little association between the fragments is observed. High affinity of the protein for the intact DNA-binding site serves as the thermodynamic driving force for the association between the two DNA “half-sites”, and this protein-dependent association is used to generate a signal that corresponds to the presence of the protein (Figure 1A). We have previously studied the physical mechanism of the assay illustrated in Figure 1A.2 The obtained results showed that the behavior of this assay could be adequately described by a simple model involving two linked equilibria: an association between the two DNA fragments and an exclusive binding of the protein to the product of this association reaction. These experiments also showed that the bimolecular association reaction between the two DNA fragments was the overall rate-limiting step of the assay.2 Based on this observation we hypothesized that the response time of the assay could be significantly reduced by creating a unimolecular beacon in which the DNA fragments would be covalently joined by a long flexible linker. In this work, we describe the design of such a unimolecular beacon. Additionally, we propose the model describing its function and we present experimental data in support of the model. The new unimolecular beacon for detecting DNA-binding proteins will allow for decrease of the assay time and will be also useful for generating beacon arrays capable of detecting multiple target DNA-binding proteins by coupling multiple unimolecular beacons to a solid surface. MATERIALS AND METHODS: Materials. Oligonucleotides were prepared using standard phosphoramidite chemistry and were purified using reversed(1) Heyduk, T.; Heyduk, E. Nat. Biotechnol. 2002, 20, 171-176. (2) Heyduk, E.; Knoll, E.; Heyduk, T. Anal. Biochem. 2003, 316, 1-10. 10.1021/ac034985p CCC: $27.50

© 2004 American Chemical Society Published on Web 01/22/2004

Figure 1. (A) Overall design of the original bimolecular beacon for DNA-binding proteins (1). (B) Overall design of unimolecular beacons for detecting proteins with DNA-binding activity. The arrow indicates that the tether is composed of several units of the flexible linker. (C) Proposed model describing the behavior of the unimolecular beacon.

phase chromatography as previously described.2,3 The following oligonucleotides were used (F, dT-Fluorescein (Glen Research, Sterling, VA); D, dT-Dabcyl (Glen Research); X, Spacer18 (Glen Research); Z, Amine Virtual Nucleotide (Clontech, Palo Alto, CA): 5′-CDAGATCACATTTATTGCGTT- 3′ (UT2), 5′-GGTGCCTAAAATGTGAT-3′ (UT3), 5′-CFAGATCACATTTTAGGCACCXXXXXXXXAACGCAATAAATGTGAT-3′ (UT1A), 5′-CFAGATCACATTTTAGGCACCXXXXXXXXXXXXAACGCAATAAATGTGAT3′ (UT1), 5′-CFAGATCACATTTTAGGCACCXXXXXXXXXXXXXXXXAACGCAATAAATGTGAT-3′ (UT1B), 5′-CFAGATCACATTTTAGGCACCXXXXXXXXXXXXXXXXXXXXAACGCAATAAATGTGAT-3′ (UT1C), 5′-TCDACTTCACATTTATTGCGTT-3′ (UT7), 5′-GGTGCCTAAAATGTGA-3′ (UT6), 5′-AGFAGATCACATTTTAGGCACCXXXXXXXXXXXXXXXXXXXXAACGCAATAAAT GTGA (UT45)-3′, 5′-TCTACTTCACATTTATTGCGTT-3′ (UT8), 5′-AACGCAATAAATGTGA (UT4)-3′, 5′-AAGFAGATCACATTTTAGGCACCXXXXXXXXXXXXXXXXXXXXAACGCAATAAATGTG (UT109)-3′, 5′-ATCDACTTCACATTTATTGCGTT-3′ (UT11), 5′-GGTGCCTAAAATGTG-3′ (UT12), and 5′AGFAGATCACATTTTAGGCACCXXXXXXXXXXZXXXXXXXXXXAACGCAATAAATGTGA-3′ (UT45AVN). To obtain the DNA duplexes, equimolar amounts of the appropriate complementary oligonucleotides were mixed at 10 µM concentration in 100 µL of 50 mMTris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA, heated for 1 min at 95 °C, and (3) Heyduk, E.; Heyduk, T. Anal. Biochem. 1997, 248, 216-227.

Table 1. Assembled Beacon Constructs beacon name CAP1 CAP2 CAP3 CAP4 CAP5 CAP6 CAP7

A

B

C

no. of bases in overhang region

UT1A UT1 UT1B UT1C UT45 UT109 UT45AVNb

UT2 UT2 UT2 UT2 UT7 UT11 UT7

UT3 UT3 UT3 UT3 UT6 UT12 UT6

4 4 4 4 6 8 6

paired oligonucleotidesa

no. of Spacer 18 moieties added 8 12 16 20 20 20 20

a

A, fluorescein and Spacer18 containing oligonucleotides. B, dabcylcontaining oligonucleotides. C, unmodified oligonucleotides. b Contains amine virtual nucleotide.

cooled to 25 °C over 2 h. Table 1 lists the various the beacon constructs used throughout this work and the oligonucleotides used to create these constructs. Hybridization of UT4 and UT8 produced an unlabeled, competitor half-site with 6-bp overhangs. Catabolite activator protein (CAP)6,7 was purified as previously described.4 DNA-BIND 96-well microplates were purchased from CorningLife Sciences (Acton, MA), and attachment of DNA was done per manufacturer’s protocol. Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1157

Fluorescence Measurements. All fluorescence measurements were performed at 25 °C (unless specified) and at a 200µL reaction volume in quartz cuvettes using an Aminco-Bowman Series 2 spectrofluorometer (Spectronic Instruments, Rochester, NY). The measurements were performed at 50 mM Tris-HCl (pH 8.0), 1 mM EDTA, 200 µM cAMP, 0.1 mg/mL BSA, and with NaCl and CAP concentrations indicated in the text. The intensity of fluorescence emission was measured using excitation at 490 nm. The equilibrium titrations were performed by mixing components of the binding reaction in Eppendorf tubes. Separate 200-µL reaction mixtures were prepared for each data point. Reaction mixtures were incubated for 1 h at room temperature and were transferred to the cuvette for fluorescence intensity determination. The time course of the fluorescence signal change upon mixing of the unimolecular beacon with the protein or with higher salt concentrations was measured using the rapid mixing stopped-flow technique in an Aminco-Bowman Series 2 spectrofluorometer equipped with the stopped-flow accessory. Data Analysis. All nonlinear regression fitting and simulations were performed with Scientist (MicroMath, Salt Lake City, UT) using the numerical solution to the mass conservation relationship defined by a particular model. Equilibrium salt titrations of the unimolecular beacon were fitted to

F ) (fo + fcK)/(1 + K)

(1)

Equilibrium titrations of the unimolecular beacon with the protein were fitted to

F ) fo[DNAo] + f([DNAc] + [DNAc-Protein])

(4)

where F is the observed fluorescence intensity, fo is the fluorescence intensity associated with the open state of the beacon, fc is the fluorescence intensity associated with DNA species in the closed state of the beacon. It was assumed in eq 4 that the protein binding to the closed state does not affect the fluorescence intensity. This assumption was previously validated using a bimolecular beacon.2 During the fitting, concentrations of all DNA species were calculated numerically at each point by solving for the following set of mass conservation equations:

Kc ) [DNAc]/[DNAo] K2 ) [DNAc-Protein]/([DNAc][Protein]) [DNA]total ) [DNAc] + [DNAo] + [DNAc-Protein] [Protein]total ) [Protein] + [DNAc-Protein]

(5)

Kinetic experiments in which the unimolecular beacon was mixed with the protein were fitted to eq 4. During the fit, time-dependent concentrations of all DNA species were calculated numerically for each point by solving the following set of differential equations:

d([DNAo])/dt ) k[DNAc] - k[DNAo] where K is the equilibrium constant generated from the known logarithmic dependence of the stability of duplex DNA on salt concentration,5 F is the observed fluorescence intensity, fo is the specific fluorescence intensity of the open state of the beacon, and fc is specific fluorescence intensity of the closed state of the beacon beacon. The competition equilibrium titration of the unimolecular beacon was fitted to

F ) fc[DNAc] + fo([DNAo] + [DNAo-DNAx])

d([DNAc])/dt ) kc*[DNAo] - k2[DNAc][Protein] + k-2[DNAc-Protein] - ko[DNAc] d([Protein])/dt ) k-2[DNAc-Protein] k2[DNAc][Protein] d([DNAc-Protein])/dt ) k2[DNAc][Protein] k-2[DNAc-Protein] (6)

(2) Rate constants in eq 6 correspond to specific steps of the reaction as illustrated in Figure 1C.

where F is the observed fluorescence intensity, fc is the fluorescence intensity associated with the closed state of the beacon, fo is the fluorescence intensity associated with DNA species in the open state of the beacon (DNAx ) competitor DNA). During the fitting, concentrations of all DNA species were calculated numerically at each point by solving for the following set of mass conservation equations:

Kc ) [DNAc]/[DNAo] Kx ) [DNAo-DNAx]/([DNAo][DNAx]) [DNAx]total ) [DNAx] + [DNAo-DNAx] [DNA]total ) [DNAo] + [DNAc] + [DNAo-DNAx] (3) 1158

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

RESULTS AND DISCUSSION Design of the Unimolecular Beacon for Detecting DNABinding Proteins. Figure 1B illustrates the design of the unimolecular beacon for detecting DNA-binding proteins. The two DNA fragments containing protein-binding site “halves” were covalently joined using a flexible poly(ethylene glycol)-based linker (Spacer 18, Glen Research) introduced during oligonucleotide synthesis. The length of the linker can be easily modulated by varying the number of Spacer 18 phosphoramidite units, each of which adds ∼23 Å to the overall length of the linker (Figure 1B). (4) Heyduk, T.; Lee, J. Biochemistry 1989, 28, 6914-6924. (5) Record, M. T.; Anderson, C.; Lohman, T. Q. Rev. Biophys. 1978, 11, 103178. (6) Ebright, R. H.; Ebright, Y. W.; Gunasekera, A. Nucleic Acids Res. 1989, 17, 10295-10305. (7) Busby, S.; Ebright, R. H. J. Mol. Biol. 1999, 293, 199-213.

Figure 2. Tethering of two DNA half-sites forming a unimolecular species to which CAP is able to bind. (A) Electrophoretic mobility shift assay analysis of CAP binding to the unimolecular beacon or control 46-bp duplex DNA containing an intact CAP-binding site.6 (B) Electrophoretic mobility shift assay of CAP binding to the CAP5 beacon or bimolecular (untethered) beacon.

Assembly of the unimolecular beacon illustrated in Figure 1B requires synthesizing three oligonucleotides: one in which two oligonucleotides corresponding to the two halves are joined with the linker and two oligonucleotides complementary to the corresponding halves. Table 1 lists the various beacon constructs used throughout this report. We hypothesized that a simple physical model illustrated in Figure 1C should be able to describe the behavior of the unimolecular beacon. This model involves an oscillation of the unimolecular beacon between the open and closed states. The closed conformation corresponds to a beacon in which complementary overhangs of DNA half-sites are annealed, whereas in the open conformation, the complementary ends of DNA half-sites are free (Figure 1B). This oscillation between two conformational states is described by a dimensionless equilibrium constant (Kc), and the formation of the complex between the closed state and the protein is described by an equilibrium constant, K2. Experimental Validation of the Physical Model for the Unimolecular Beacon. We performed a series of experiments to validate the model illustrated in Figure 1C using wellcharacterized sequence-specific DNA-binding protein (CAP). The goal of the first set of experiments was to demonstrate that introduction of the flexible linker connecting the two DNA halfsites did not interfere with the ability of the protein to form a complex with DNA. Electrophoretic mobility shift assays8 were performed in which the unimolecular beacon, CAP5, was titrated with an increasing concentration of CAP (Figure 2A). Formation of a protein-DNA complex was apparent, and similar concentrations of the protein were required for complex formation in comparison to the control in which a 46-bp duplex containing the consensus CAP-binding site was used.6 Formation of protein(8) Fried, M. G.; Crothers, D. M. Nucleic Acids Res. 1981, 9, 6505-6525.

DNA complex by the unimolecular beacon was also compared to CAP binding to a bimolecular beacon (Figure 2B). Complex formation in this case also required similar concentrations of CAP. These results showed that it was possible to covalently link the two DNA half-sites and that the presence of the linker did not greatly affect the binding of CAP. The ability of the unimolecular beacon to exist in equilibrium between two states (closed and open) is a fundamental element of the model illustrated in Figure 1C. We used perturbation with salt to test if, indeed, these molecules exist in solution in an equilibrium between closed and open states (Figure 3). The isomerization reaction illustrated in Figure 3A should be sensitive to salt concentration because the annealing of two complementary overhangs, which drives the transition from open to the closed state, should have the expected dependency on salt concentration (i.e., a marked shift from open to closed conformation with the increase of salt concentration). A shift of equilibrium toward closed state with the increase of salt concentration should be detected as fluorescence quenching since specific fluorescence of the closed state is low (donor and acceptor probes are in proximity). Salt titrations were performed with unimolecular beacons CAP4, CAP5, and CAP6 that contain four, six, or eight base pair overhangs, respectively, at a 5 nM DNA concentration. The expected decrease of fluorescence with the increase of salt concentration was observed (Figure 3B). Experimental data could be fitted to eq 1 and such analysis provided cyclization equilibrium constants and their salt dependence. Observed cyclization constants increased with the length of the complementary overhang and with the increase in salt concentration (Figure 3C). Such behavior was expected based on a known logarithmic dependence of the stability of duplex DNA on salt concentration and duplex length.5 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1159

Figure 3. (A) Association of the tethered donor and acceptor labeled DNA half-sites in the absence of the protein. (B) The change in fluorescence as a function of salt concentration for 4-bp overhang construct, CAP4 (filled circles), 6-bp overhang construct, CAP5 (open squares), and 8-bp overhang construct, CAP6 (open triangles). (C) Dependence of Kc on salt concentration for 4-bp overhang construct, CAP4 (short dashed line), 6-bp overhang construct, CAP5 (solid line), and 8-bp overhang construct, CAP6 (medium dashed line).

Figure 4. (A) Proposed model describing the behavior of the assay in the presence of unlabeled competitor DNA half-sites. (B) Dependence of fluorescence increase on the concentration of competitor DNA half-site added. Competitor DNA half-site duplex was prepared with the same sequence as the tethered acceptor half-site of the unimolecular beacon. The CAP5 beacon was titrated with increasing amounts of competitor half-site (0-20 µM) at 350 mM NaCl.

Additional confirmation that unimolecular beacons exist in solution in equilibrium between open and closed states was obtained in an experiment in which the 6-bp tethered construct, CAP5, was titrated with the unlabeled competitor DNA. As illustrated in Figure 4A, unlabeled competitor was designed to bind exclusively to the open state. Thus, we expected that, with the increase of competitor concentration, the equilibrium would be shifted to the open state resulting in an increase of fluores1160

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

cence. A representative example of such a titration curve obtained at 350 mM NaCl is shown in Figure 4B. The concentration of 350 mM NaCl was selected because the previous titration (Figure 3A) indicated that this was the concentration at which the CAP5 beacon was mostly in the closed state. The expected increase in fluorescence with the increase of competitor concentration was observed (Figure 4). Experimental data were fitted to eqs 2 and 3 (Figure 4, solid line). The value of Kc obtained from such analysis was very similar to Kc obtained for the same salt concentration in salt titration experiments (not shown). The salt titration and competitor titration experiments demonstrated that unimolecular beacons existed in solution in equilibrium between open and closed states. The final set of experiments aimed at testing the model for the unimolecular beacon shown in Figure 1C involved the analysis of the response of the unimolecular beacon to increasing concentrations of the target protein. We used the tethered construct CAP5 to examine the dependence of the fluorescence intensity on the protein concentration under a variety of assay conditions. A representative example of such a titration curve obtained at 150 mM NaCl is shown in Figure 5A. The applicability of the model illustrated in Figure 1C for describing the behavior of the assay in the presence of the protein was tested by fitting the data in Figure 5A to eqs 4 and 5. During the fitting, Kc was fixed at the value that was obtained from the titration of the CAP5 beacon with increasing concentrations of NaCl described previously. An excellent fit of the data was obtained (Figure 5A, solid line), indicating that the model shown in Figure 1C is sufficient to describe the assay’s response in the presence of the protein. Furthermore, K2 obtained from the fit was in good agreement with the published value for consensus CAP-binding site under similar conditions of salt concentration and temperature.6 Titrations with

Figure 5. (A) Example of a titration curve of the molecular beacon containing both donor and acceptor labeled halves with the protein. The data shown are from the experiment using the CAP5 beacon. The solid line represents the fit of the data to the model illustrated in Figure 1B. (B) DNA concentration dependence of the apparent affinity constant for CAP to CAP5. DNA concentrations ranging from 2 to 100 nM were used.

the protein at various salt concentrations also yielded excellent fits (not shown), thus demonstrating the applicability of the model at different solution conditions. Next, we examined the dependence of the apparent affinity constant (Kapp) on the concentration of the unimolecular beacon. The Kapp values were obtained by fitting protein titration curves (such as the one illustrated in Figure 5A) by assuming that a simple 1:1 complex between the protein and unimolecular beacon was formed. In our previous analysis of bimolecular beacon design, it was demonstrated that as the total concentration of DNA halfsites was increased, the Kapp of the protein was also increased.2 This increase in Kapp was observed because the increase in DNA concentration shifted the equilibrium toward the formation of a full-length duplex to which the protein bound with high affinity. Our working model for the unimolecular beacon predicts that an increase in beacon concentration should not increase the Kapp of the protein since an increase of beacon concentration should not affect the unimolecular isomerization reaction between open and closed states. Figure 5B clearly shows that an increase in total CAP5 concentration did not increase the apparent binding constant, which is in agreement with the model illustrated in Figure 1C. In summary, the obtained data demonstrated the feasibility of unimolecular beacon design and provided experimental evidence validating the proposed model of its action. Effect of the Linker Length. Two most important variable elements of unimolecular beacon design are the length of the complementary overhang and the length of the flexible linker connecting the two DNA half-sites. Length of the overhang plays a relatively obvious role in determining the affinity between the two DNA half-sites2 and will thus determine the value of Kc. The effect of linker length is more difficult to predict. Therefore, a series of experiments designed to test the role of linker length was conducted. Figure 6A shows the effect of increasing the linker length on the apparent affinity of the protein for the beacon. Protein titrations were preformed with unimolecular beacons of various linker lengths containing 4-bp overhangs (CAP1-CAP4) at a 25 nM DNA concentration and 150 mM NaCl. The obtained data showed that, the longer the total linker length, the higher the apparent affinity

of the protein for the beacon. CAP beacons with 16 and 20 linker units exhibited essentially identical affinity for the protein. Thus, although longer linkers were not tested, it is unlikely that a higher affinity with linkers longer than 20 units would be observed. In fact, one would expect that the dependence of the affinity on the linker length should be represented by a bell-shaped curve since with very long linkers the local concentration of DNA half-sites should be decreased to a point that the advantage of linking DNA fragments would disappear. Decreased apparent affinity of the protein for CAP beacons with 8 and 12 linker units suggests that these linker lengths are somewhat too short to allow 100% efficient formation of the closed state of the beacon. To test whether changes in apparent affinity of the protein as a function of linker length could be explained by the effects of linker length on the equilibrium between open and close states, we examined the effects of temperature and linker length on the closed-open equilibrium of the beacon over a range of salt concentrations. First, the salt titrations were preformed with CAP4 at a 5 nM DNA at 15, 25, and 30 °C. An example of such a titration curve and the fit of the experimental data to eq 1 is shown in Figure 6B. Observed cyclization constants increased with the decrease in temperature as was expected based on the known effects of temperature on DNA annealing.5 Figure 6C shows the effect of linker lengths on beacon cyclization at a constant temperature. Formation of the closed conformation of the beacon was more efficient with the increase of linker length, which is in agreement with the trend observed in studies of protein-binding dependence on linker length (Figure 6A). Taken together, it appears that if the linker is long enough, such that the stability of the closed conformation is not negatively affected, a relatively wide range of linker lengths can be used in the unimolecular beacon design. In the case of the CAP beacon tested in this work, a linker of 16 units of Spacer 18 or more (corresponding to a length of ∼370 Å or more) appeared to be optimal. Knowledge of the physical behavior of the model in Figure 1B will also allow the use of the assay for quantitative assessment of the DNA-binding activity of the protein of interest. Titration of the beacon with the increasing protein concentration (for example, Figure 5) can be used to obtain Kapp, the apparent equilibrium Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1161

Figure 6. (A) Effect of linker length on the apparent affinity constant for CAP binding to the unimolecular beacon (all constructs contain 4-bp overhangs). Eight, 12, 16, and 20 linker units were used for a total linker length of 187.2, 280.8, 374.4, and 468 Å, respectively. (B) Effect of temperature on Kc using CAP4; 15 (filled circles), 25 (open squares), and 30 °C (open triangles). (C) Effect of the linker length on the ability of the tethered beacon to cyclize at 15 °C; 8 linker units, CAP1 (open circles), 12 linker units, CAP2 (filled circles), 16 linker units, CAP3 (open squares), and 20 linker units, CAP4 (open triangles).

binding constant. Using the model describing unimolecular beacon (Figure 1C), the following relationship between the Kapp and K2 can be obtained:

K2 ) Kc(Kapp/(1 + Kc))

(7)

Thus, the true equilibrium binding constant for the protein-DNA complex formation can be obtained from measured Kapp and Kc (which can be obtained from salt titration experiments as in Figure 3). It should be possible to estimate the value of Kc from the predicted stability of DNA duplex corresponding to the complementary overhang of the construct. To accomplish this, local concentration of tethered DNA half-sites can be estimated from the volume of solvent accessible to the tethered DNAs (which is determined by the length of the two duplexes and the flexible linker joining the DNA half-sites). Such a calculation produces a local concentration in the micromolar range. This estimate of the local concentration is consistent with the observation that micromolar concentrations of the competitor half-site were required to affect open-close equilibrium of the unimolecular beacon (Figure 4B). The product of local concentration of DNA half-sites and the equilibrium binding constant for the two complementary overhangs should be the Kc. Such calculations result in Kc values that are only severalfold greater than the measured values (data not 1162

Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

shown), suggesting that the flexible linker has a relatively minor, negative impact (for example, through some steric constraints) on the association between the DNA half-sites in the context of the unimolecular beacon. Kinetic Analysis of Unimolecular Beacon. The aim of the next set of experiments was to understand the kinetics of the assay and, in particular, to identify the rate-limiting step determining the overall time necessary for performing the assay. Previous kinetic experiments with the bimolecular beacon revealed that bimolecular association reaction between the two DNA half-sites was the rate-limiting step of the overall reaction.2 We hoped that one of the advantages of the unimolecular beacon design would be a greatly increased rate of association reaction between the two tethered half-sites, which would reduce time necessary to perform the assay. To understand the kinetic behavior of the assay, two sets of experiments were performed. The first set of experiments involved rapid changes of salt concentration to perturb the equilibrium between open and closed forms of the beacon. This was accomplished by rapidly mixing (in a stopped-flow apparatus) the CAP5 beacon in the buffer containing 50 mM NaCl with a buffer containing higher salt concentration. The rate of change between the existing equilibrium and subsequent new equilibrium can be measured by examining the change of fluorescence over time. Examination of the obtained data (Figure 7A) revealed that the

Figure 7. (A) Kinetics of cyclization of the unimolecular beacon (CAP5) in the absence of the protein. Salt perturbation experiments were performed by rapidly mixing 50 nM DNA in 50 mM NaCl with buffer containing salt concentrations of 50 (red circles), 200 (green circles), 500 (yellow triangles), and 800 mM (blue triangles) NaCl. (B) An example of a kinetic curve obtained by rapidly mixing the unimolecular beacon (25 nM) containing both donor and acceptor labeled half-sites with 300 nM CAP. The data shown are for the experiment with the CAP5 beacon at 150 mM NaCl. The solid line represents the best fit to the model illustrated in Figure 1B. Inset: A family of kinetic curves obtained by mixing 25 nM DNA with CAP at 150, 225, 300, 375, 450, and 525 nM concentrations.

relaxation reaction to a new equilibrium following rapid change of salt concentration was completed during mixing of the reagents (estimated dead time of the instrument was ∼1 ms). “Salt jumps” to increasing salt concentrations produced a progressive decrease of the fluorescence value corresponding to the end point of the reaction (Figure 7A), which is in agreement with higher expected proportions of closed beacon conformation at higher salt concentrations. However, under the applied experimental conditions, none of the relaxation reactions could be captured. This indicated that oscillations between the open and closed states of the unimolecular beacon were very rapid (in submillisecond time range). The second set of kinetic experiments involved the rapid mixing of the protein with the CAP5 beacon (Figure 7B). Previous work involving rapid mixing between the bimolecular beacon and the protein yielded kinetic curves that were ∼60 times slower than compared to association between two DNA half-sites in the absence of the protein.2 Upon mixing of the protein with the unimolecular beacon, fluorescence decreased rapidly and the reaction was completed in ∼0.5 s. This is ∼1200-fold faster than the time scale of the reaction completion in the case of the bimolecular beacon (∼600 s) under similar conditions.2 Thus, linking the two DNA half-sites by a long flexible linker greatly reduced the time necessary for completion of the assay. The reaction between the protein and the unimolecular beacon, although fast, was much slower than the relaxation in the absence of the protein. One possible interpretation could be that protein binding to the full-length duplex (k2) was now the rate-limiting step of the overall reaction illustrated in Figure 7B. If this were true, then one would expect that the rate of fluorescence change would depend on protein concentration since the rate constant k2 describes a bimolecular reaction. However, kinetic curves obtained at a wide range of protein concentrations revealed that the apparent rate constant was clearly not dependent on protein concentration (Figure 7B, inset). The alternative interpretation, which we favor, is that the rate constant kc is the rate-limiting step in the presence of the protein and governs the overall kinetics

observed. The explanation for the difference in apparent rates between Figure 7A and B would be that, in the presence of the protein, the isomerization of the unimolecular beacon to the closed state is essentially irreversible since all of the full-length duplex is rapidly bound by the protein and this complex dissociates very slowly. Thus, in the presence of the protein, ko does not contribute significantly to the kinetics of the overall process. In contrast, in the absence of the protein (“salt jump” experiments), the overall kinetics of relaxation to a new equilibrium is governed by the sum of ko and kc. Immobilization of the Unimolecular Beacon on a Solid Support. Preparation of the unimolecular beacon opens the possibilities of preparing arrays of immobilized beacons designed to recognize different target proteins. To test the feasibility of attaching the unimolecular beacon to a solid support, we introduced an amine virtual nucleotide (AVN) into the linker region between linker units 10 and 11 of the CAP5 beacon to create CAP7 (Figure 8A), which was then covalently attached to the wells of an activated 96-well plate (Figure 8B). The functionality of the covalently attached CAP7 beacon was first examined by salt equilibrium titrations. An example of such a titration curve is shown in Figure 8C. It appeared that the conditions for attachment of the beacon to the solid surface do not hinder its ability to cyclize. The next experiment was performed to determine the ability of the protein to gain access to the beacon and cause a measurable signal change. Protein concentration-dependent quenching of fluorescence was observed (Figure 8D), illustrating that attachment of the beacon to the solid support did not significantly affect its ability to signal the presence of the target protein. CONCLUSIONS We describe here a design of a unimolecular beacon for detecting DNA-binding proteins. By linking the two DNA halfsites (components of the beacon) with a long flexible linker, a major improvement in the response time of the beacon was obtained. Furthermore, the unimolecular beacon design eliminates a step involving association of two separate DNA half-sites that is Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

1163

Figure 8. (A) Design of the CAP7 beacon used for coupling to a solid surface. The arrow points to the structure of the amine virtual nucleotide that was used to couple to solid surfaces. (B) A representation of a single well from a 96-well DNA-BIND plate, which shows the amine group from AVN covalently attached to the N-oxysuccinimide surface. (C) Change in fluorescence as a function of salt concentration for CAP7 that is covalently attached to a solid surface. (D) An example of CAP binding to the unimolecular beacon attached to a 96-well plate. The data shown are for the experiment with the CAP7 beacon at 150 mM NaCl. The solid line represents the best fit to the model illustrated in Figure 1B.

dependent on total DNA concentration. Thus, the apparent affinity of the protein for binding the unimolecular beacon is not dependent on DNA concentration. Unimolecular beacons remain functional when immobilized on a solid support, opening the possibilities of their use in preparing arrays for detecting multiple DNA-binding proteins.

1164 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004

ACKNOWLEDGMENT This work was supported by NIH Grant 1R21CA94356. Received for review August 22, 2003. Accepted November 30, 2003. AC034985P