Molecular assembly for high-performance bivalent nucleic acid inhibitor
- Youngmi Kim,
- Zehui Cao, and
- Weihong Tan*
- Center for Research at the Bio/nano Interface, Department of Chemistry, University of Florida Genetics Institute, Shands Cancer Center, and McKnight Brain Institute, University of Florida, Gainesville, FL 32611-7200
-
Edited by Richard N. Zare, Stanford University, Stanford, CA, and approved February 7, 2008 (received for review December 14, 2007)
Abstract
It is theorized that multivalent interaction can result in better affinity and selectivity than monovalent interaction in the design of high-performance ligands. Accordingly, biomolecular engineers are increasingly taking advantage of multivalent interactions to fabricate novel molecular assemblies, resulting in new functions for ligands or enhanced performance of existing ligands. Substantial efforts have been expended in using small molecules or epitopes of antibodies for designing multifunctional or better-performing ligands. However, few attempts to use nucleic acid aptamers as functional domains have been reported. In this study, we explore the design of bivalent nucleic acid ligands by using thrombin and its aptamers as the model by which to evaluate its functions. By assembling two thrombin-binding aptamers with optimized design parameters, this assembly has resulted in the successful development of a nucleic acid-based high-performance bivalent protein inhibitor. Our experimentation proved (i) that the simultaneous binding of two aptamers after linkage achieved 16.6-fold better inhibition efficiency than binding of the monovalent ligand and (ii) that such an improvement originated from changes in the kinetics of the binding interactions, with a k off rate ≈1/50 as fast. In addition, the newly generated aptamer assembly is an excellent anticoagulant reagent when tested with different samples. Because this optimized ligand design offers a simple and noninvasive means of accomplishing higher performance from known functional aptamers, it holds promise as a potent antithrombin agent in the treatment of various diseases related to abnormal thrombin activities.
In contrast to monovalent interaction, multivalent, or polyvalent, interactions involve the binding of multiple ligands of a biological entity, such as small molecules, oligosaccharides, proteins, nucleic acids (NAs), lipids, or aggregates of these molecules, to multiple binding pockets or receptors of a target, e.g., a protein, virus, bacterium, or cell (1). Polyvalency is ubiquitous in biology and has a number of benefits over monovalent interactions. For instance, polyvalent interactions collectively possess higher binding affinity than the corresponding monovalent interactions. That is, polyvalency results in a “cooperative” configuration in which the probability of rebinding of a dissociated monomer to the target is increased by the presence of other monomers bound to the same target. A classical example of this is demonstrated by the binding of galactose-terminated oligosaccharides to C-type mammalian hepatic lectins (2). In addition to increased binding affinity, polyvalent interactions also stand a better chance of providing higher selectivity in target recognition. Noticeably, a multivalent binder, despite being composed of weak homo- or heterogeneous ligands, can still have stronger binding property because of multiple binding events. A well known example of this phenomenon is taken from the biology of gene regulation by oligomeric transcription factors. Specifically, the retinoid X receptor (RXR) functions as a transcription factor in the presence of its ligand (3). Each RXR–ligand complex (RXR-L) binds to a single-stranded DNA called the cellular retinol-binding protein II element (CRBP-II element). Interestingly, although the intrinsic affinity of one or more units of RXR-L for one CRBP-II element (i.e., di-, tri-, or tetravalent interaction) is insufficient to initiate transcription, more than five of these complexes adjacent to CRBP-II elements can, in fact, initiate the transcription. As a result, transcriptive response is well regulated depending on the concentration of the transcription factor. Furthermore, as noted above, this activity demonstrates the cooperative configuration, as noted above, which gives polyvalent interactions the potential for considerably increased binding affinity.
A number of recent studies have reported the unique properties of multivalent interactions. Investigators have attempted to mimic the mechanisms underlying such interactions to design new therapeutic entities, particularly those using repetitive epitopes of antibodies (1). By designing more efficient targeting reagents with potentially viable therapeutic applications, all of these attempts have shown promising results. A typical example is the single-chain variable fragment (scFv) constructed by linking the antigen-binding VH and VL domains of an antibody with a flexible polypeptide linker (4). The combinatory configurations of scFvs can be designed and investigated to optimize the functionality. Another successful therapeutic design, which takes advantage of polyvalent interactions, is the bi-specific T cell engager molecule (BiTE) (5). A BiTE molecule is a bi-specific antibody that is constructed by linking the binding domains of two antibodies with different specificities with short, flexible peptides and is, therefore, expressed as a single polypeptide chain. The typical working principle is that BiTEs bind with one arm to a target cell and the other arm to a T cell, consequently activating the T cell. This unique mode of action results in increasing the cytotoxic potency of BiTE molecules at least 10,000-fold higher than that of conventional human IgG1 antibodies (6). These two achievements demonstrate how biomolecular engineers have exploited the potential of multivalent binding motifs. However, the genetic engineering required to mimic the mechanisms underlying multivalent interactions is time consuming and prone to many technical difficulties. For instance, expressed proteins may not fold into the expected tertiary structures, leading to nonfunctional products. Also, the high molecular weight of the final product can be a limitation for future therapeutic applications. For these reasons, alternative ligands that have functionality similar to that of antibodies, but without the limitations, are clearly attractive. In the present study, we demonstrate how the NA aptamer can be a strong candidate well suited for such multivalent applications.
Aptamers are NA sequences selected by the Systematic Evolution of Ligands by Exponential Enrichment (SELEX) method (7, 8). They specifically recognize targets ranging from ions and small organic or inorganic molecules to proteins and living cells, with binding affinity and selectivity for targets comparable to those of antibodies. NA aptamers have markedly lower molecular weights (usually below 20,000) than antibodies, and their secondary structures are easily predictable. In addition, aptamers are not prone to the irreversible denaturation that affects antibodies. Moreover, they can be synthesized efficiently and reliably by using established phosphoramidite chemistry. Thus far, many aptamers have been identified, and some of them are very close to becoming marketable drugs (9–13). Nevertheless, engineering these aptamers for enhanced performance or new functions remains virtually unexplored. One of the most significant advantages of aptamers in molecular assembly for multivalent binding is the ease of conjugating aptamers together through NA chemistry. This capability presents great potential in making aptamer conjugates with greatly enhanced functions or new and unique properties.
Here, we demonstrate that rationally designed aptamer assemblies can combine the functionality and binding affinity of different aptamers to achieve greatly enhanced enzymatic inhibition with potential medical significance. Using fluorescence resonance energy transfer (FRET)-based molecular probes (14), we have also revealed the kinetic properties of the aptamer–assembly-protein interaction that underlie this improved inhibition. Based on the molecular assembly of two separate aptamers, we can expect the following advantages: stronger binding affinity, enhanced inhibitory function, intrinsic high selectivity, minimal host immune response, and low to no cytotoxicity (15). In addition, a unique feature of DNA-based therapy is that the antidotes of aptamer drugs are readily available from their target DNAs (16). We have demonstrated the use of these antidotes to effectively reverse the activity of the aptamer assembly because of the strong and fast aptamer/DNA hybridization. With this set of investigations, we have proved that the careful and rational design of aptamer assemblies can exploit multivalent interactions with the protein target to produce enhanced enzymatic inhibition in a simple, effective, and practical way.
Results and Discussion
Thrombin Aptamers and Their Properties.
Thrombin is a multifunctional protease involved in the regulation of homeostasis. As an initiator of blood clot formation, thrombin hydrolyzes fibrinogen and activates platelets and some blood coagulation factors that have procoagulant activity. Disorders in blood clotting are tightly linked to many serious health issues, including heart attack and stroke. Therefore, thrombin is typically the target in anticoagulation therapy for these diseases. However, anticoagulant drugs currently on the market often suffer from indirect inhibition and suboptimum selectivity, which could lead to side effects, including bleeding (17, 18).
There are two known thrombin NA aptamers. One is 15 bases long (15Apt) and binds to exosite 1, whereas the other, called 27Apt, is 27 bases long and interacts with exosite 2 (11, 19) as shown in Fig. 1. The dissociation constant K d of 15Apt tends to be very high (up to 450 nM), depending on measurement methods (9, 20, 21), and K d of 27Apt is ≈0.7 nM (19) As a potential anticoagulant, only 15Apt should have the enzymatic inhibitory functions required for thrombin-mediated coagulation because it occupies the fibrinogen-binding exosite 1. However, efforts to explore the anticoagulant effect of this aptamer have shown only limited progress because of the lack of sufficient strength of binding to the exosite 1 on the target protein. Modifications on the existing aptamer itself have also been explored to generate an enhanced functional NA ligand of thrombin. However, this type of construct is likely to be even more unpredictable, because its unique conformational structure can be disrupted, resulting in the loss of its binding property (22). Therefore, instead of modifying the aptamer sequence itself, we have linearly assembled two existing NA aptamers of thrombin to form a molecular assembly of thrombin aptamers. This assembly specifically improves the inhibitory function of thrombin because of the multivalent interaction mentioned above.
Schematics of the working principles of monovalent and bivalent NA ligands. (a) 15Apt, a monovalent ligand, has constant ON and OFF and diffuses into bulk solution immediately after dissociation from thrombin, resulting in low inhibitory function. (b) In contrast, when linked to 27Apt to form a bivalent ligand, 15Apt can rapidly return to the binding site after dissociation because of molecular diffusion confined by 27Apt that is still bound to thrombin. As a result, the equilibrium of the reaction is shifted to the left side.
The Design of Bivalent NA Inhibitor.
We hypothesized that linear molecular assembly of two monovalent NA aptamers would result in a superior functional NA inhibitor of enzymatic reactions with multivalent binding properties. However, first we need to find out whether we could achieve enhanced inhibition even in the absence of multivalent interactions by simply mixing these two aptamers without covalent linkage. The second important question would be whether these two aptamers could interact with each other and subsequently cause the loss of inhibition after assembly. To address these issues, we performed typical clotting test using no aptamer, 15Apt alone, 27Apt alone, and a mixture of 15Apt and 27Apt. As shown in Fig. 2, 15Apt alone delayed the coagulation time by nearly 3-fold, whereas 27Apt had no significant observed inhibition. This result was expected because 27Apt does not target the critical exosite 1. The nonlinked 15Apt/27Apt mixture showed an inhibitory effect similar to the 15Apt alone. These findings indicate that 15Apt and 27Apt are independent of each other and thus do not promote or interfere with each other's activity.
Comparison of the normalized clotting times of thrombin bound to different NA inhibitors. Clotting time of thrombin alone was defined as 1, and the relative values based on it are plotted. 15Apt alone showed a 3-fold increase of the clotting time, but no delay was observed from the 27Apt-treated sample. Among bivalent NA candidates (Bi-xS), Bi-8S is the best inhibitor with the longest clotting time. The replacement of 27Apt by a random sequence, as in Bi′-8S, causes almost complete loss of anticoagulant function, mainly due to the lack of bivalency and interaction between the scrambled sequence and 15Apt.
Next, several candidate bivalent NA ligands were designed and evaluated with the purpose of optimizing the distance between the two different NA aptamers. This step is particularly critical in designing bivalent ligands, and we initially assumed that a shorter distance between the two aptamers would result in disruption of their simultaneous binding and, hence, less effective binding and inhibition. Therefore, we designed several potential bivalent NA ligands with linkers of different lengths composed of 4, 6, 8, or 10 spacer phosphoramidites and designated Bi-xSs, as shown in supporting information (SI) Table S1. Considering that one spacer is ≈2.1 nm long and that the inner diameter of thrombin is several nanometers, this represents a sufficiently ample range of lengths; i.e., from 8.4 to 21.0 nm. Then a clotting test was carried out to first evaluate the effect of linker length on thrombin inhibition; the results were compared with the clotting times of the monovalent aptamer. In this test, the thrombin converts soluble fibrinogen into insoluble strands of fibrin, resulting in increasing turbidity and decreasing fluidity. The recorded time when it became completely nonfluidic was normalized and compared among the series of potential bivalent ligands. The summary is shown in Fig. 2. Briefly, as the number of spacers increased, the inhibition activity first increased and then maximized with eight spacers. After that, a decline of inhibition was seen. Interestingly, the Bi-4S displayed a worse anticoagulation efficiency than free 15Apt. Competition between the two aptamers readily explains this phenomenon. In other words, if the distance between 15Apt and 27Apt is not sufficiently long, the binding of 27Apt will surpass that of 15Apt because of its stronger affinity to thrombin. This binding, in turn, will prevent 15Apt from reaching the binding pocket of thrombin and lead to a drastically reduced anticoagulation capability. However, with a sufficient linker length, Bi-8S offered a 9-fold better thrombin inhibition than 15Apt alone. Contrary to our initial assumption, and considering that thrombin is ≈3–4 nm in diameter, it was unexpected to see that a linker as long as ≈16 nm (eight spacers) was needed for the best inhibition. Based on these findings, we therefore hypothesized that some extra linker length was required for both aptamers to wrap around thrombin and take the optimum three-dimensional position and orientation to interact with thrombin, probably in order for minimizing interference originating from thrombin/linker contact. Despite these results, an even longer linker, such as that offered in Bi-10S, was not as efficient as a shorter linker, and it gave a decreased clotting time. Taken together, these results support the fact that simultaneous binding through bivalent interaction does take place and does result in improved inhibitory effect. Additional evidence was obtained from the clotting test using Bi′-8S, in which 15Apt was linked to a scrambled DNA sequence with eight spacers. The result showed very limited thrombin inhibition. Therefore, the 9-fold increase of clotting inhibition by Bi-8S over 15Apt alone is a direct result of bivalent interaction realized by molecular assembly, implying, in turn, that aptamers are ideal for the design of multivalent ligands through molecular assembly.
Monitoring Inhibitory Functions by Using Light Scattering.
To obtain real-time kinetics of coagulation, we designed a more quantitative measurement based on the fact that gel formation increases the turbidity of the reaction mixture and scatters more light. Light scattering changes during coagulation were monitored on a spectrofluorometer, and the results are shown in Fig. 3. The background scattering of the thrombin/inhibitor mixture was stable until the addition of fibrinogen. Increase of the light scattering reflected the net rate of the coagulation reaction, as this was the direct result of fibrinogen cleavage. The relative inhibition strengths of the monovalent and bivalent NA ligands were estimated from the initial reaction rates, with higher rates representing weaker inhibition and the reverse for lower rates. As expected, Bi-4S produced an initial rate of 2,903 cps/sec calculated from the early slope of the scattering profile, 2.8 times faster than that of 15Apt alone, 1049 cps/sec, which means that 27Apt interferes with 15Apt in the binding process. In contrast, the initial reaction rates of Bi-8S and Bi-6S were much slower than the others, at 63 and 97 cps/sec, respectively. Because Bi-8S demonstrates an initial rate close to 6.0% of that for free 15Apt, it also represents a considerable improvement of antithrombin efficacy. Thus, the anticoagulation trend among the tested inhibitors correlates very well with the results from the clotting tests. As shown in the clotting test, Bi-10S did not function as well as Bi-8S, resulting in an increased initial reaction rate. Bi′-8S, whose sequence of 27Apt domain is replaced by the scrambled one, also showed no improvement in inhibiting the clotting process (data not shown). Therefore, based on the evidence gathered from both clotting test and turbidity measurement using light scattering, we conclude that Bi-8S is the best design for improved thrombin inhibition.
Real-time monitoring of light scattering generated by the coagulation process in the presence of different monovalent or bivalent NA ligands (Bi-xSs). After coagulation is initiated by adding fibrinogen to each sample, the reaction kinetics varied depending on the ligands. The initial reaction rate of each sample was calculated (scattering signal increase divided by time, cps/sec) and then plotted in the Inset. This result is consistent with the clotting test. As the number of spacers increased, the reaction rate went down and then up (Inset). Results show that Bi-8S is the best design of bivalent NA inhibitor.
Binding Kinetics Studies.
Because bivalent interaction of the aptamer assembly with thrombin increases overall binding affinity, it is proposed as the mechanism for the enhanced inhibition. Because the binding affinity is directly related to kinetic parameters, such as k on and k off, of the thrombin–inhibitor interaction, we carried out experiments to investigate the impact of the molecular assembling on k on and k off of the reaction to reveal what actually causes the improved inhibition. One important feature of aptamers is their binding to the target is often accompanied by changes in tertiary structures. This allows researchers to build various signal transduction mechanisms, such as FRET, into aptamers for sensitive target detection. In fact, 15Apt was among the first aptamers to be built into molecular beacon aptamers (MBAs) for protein detection based on FRET (14). Here, we labeled 15Apt and the 15Apt domain of Bi-8S with a fluorophore and quencher pair to form 15Apt MBA and Bi-8S MBA shown in Table S1. The compact structure of the 15Apt bound to thrombin was expected to differ considerably from the random coil structure in solution, thus giving different fluorescence intensity. We then used these modified aptamers to study the kinetics of interactions with thrombin under different conditions.
To compare k off of the 15Apt MBA and 15Apt domain of Bi-8S MBA 1, the full complementary target DNA of 15Apt (T-15Apt) was used to make 15Apt, when released from thrombin, inactive by forming a duplex with it (Fig. 4 b). At the same time, opening of the MBA should give intensive fluorescence. Each MBA probe (100 nM) and 500 nM thrombin were preincubated for half an hour to complete the binding reaction. Then, T-15Apt was added to the mixture while the fluorescence signal was monitored. To normalize the fluorescence signal, MBAs fully opened by T-15Apt were used as the reference. One might question whether T-15Apt could induce the release of 15Apt from the binding pocket. Our investigation of concentration effects of T-15Apt (Fig. S1) shows that this was not the case. In our tests, regardless of the concentration of T-15Apt, the reaction kinetics was consistent, which means that T-15Apt does not induce the release, but rather captures the released 15Apt as a separate step. Finally, the initial rate of each reaction was calculated by using the linear part of the slope. Even though values calculated this way are not the absolute k′ off rates, they can still be useful for comparing kinetic parameters among different thrombin ligands. Determination of k′ on for each MBA was done in a similar way (Fig. 4 a). Competition between thrombin and a short target DNA of 15Apt, called T′-15Apt, was studied. Briefly, each MBA was incubated with T′-15Apt, and then thrombin was added while the fluorescence signal was monitored. Because of the weak binding affinity between T′-15Apt and 15Apt, thrombin would compete with T′-15Apt for binding to 15Apt or the 15Apt domain of the bivalent ligands. This study was based on three assumptions: (i) that T′-15Apt interacted only with 15Apt, (ii) that the binding affinity of the T′-15Apt is identical for both MBAs, and (iii) that binding of 15Apt to thrombin is more favorable than binding to T′-15Apt. Optimization of the T′-15Apt sequence revealed that 10 complementary bases gave the best results in terms of binding to 15Apt and detectability of the kinetic parameters. The reason that we concluded that T′-15Apt works the best is the disassociation between 15Apt and T′-15Apt is not the rate-limiting step but the association between 15Apt and thrombin target is, so that what we observed is the association rate between 15Apt and thrombin (Fig. S2). The experiment result showed that such dissociation was very fast (a few tens of seconds for the competition). Specifically, 100 nM each MBA and the 10-base T′-15Apt were preincubated to complete the hybridization. Then, a 5 times excess of thrombin was added to the mixture while the fluorescence was monitored. Immediately after that, there was sharp fluorescence signal decay. The obtained plot was normalized, and the data points for the first 100 sec were used to calculate relative reaction rates.
Comparison of binding kinetics. (a) Cartoon to describe the k′ on measurement. (b) Cartoon to describe the k′ off measurement. (c) Real-time fluorescence signal change of k′ on measurement. After thrombin was added, each sample mixture showed fluorescence decay. The decreasing rate was comparable in both cases. According to the calculation of the initial reaction rate, Bi-8S exhibited a 1.2 times faster k′ on than did 15Apt. (d) Real-time fluorescence signal change of k′ off measurement. Free 15Apt MBA (green line) showed very rapid hybridization kinetics with its target DNA. Thrombin-bound 15Apt MBA (blue) showed slower hybridization kinetics compared with the free form. Interestingly, thrombin-bound Bi-8S MBA (red) showed a dissociation rate only 1.9% of that for unbound Bi-8S MBA. The K′ a of the 15Apt domain of Bi-8S is ≈62 times stronger than that of free 15Apt.
The relative k′ off values obtained for monovalent and bivalent 15Apts were 1.5% and 0.029%/sec, respectively (Fig. 4 d) (all measured by percentage of changes in fluorescence intensity). This means that k′ off of monovalent 15Apt is 51.7 times faster than that of bivalent 15Apt. 15Apt MBA and Bi-8S MBA 1 have relative k on values of approximately −0.00424% and 0.00498%/sec, respectively (Fig. 4 c). In other words, the k′ on of bivalent 15Apt is similar to that of monovalent 15Apt. Finally, we obtained the relative K′ a values by dividing k′ on by k′ off, which revealed that binding affinity of bivalent 15Apt to thrombin is ≈51.7 or more times higher than that of monovalent 15Apt. These results agree with the binding properties of other reported multivalent ligands. It is believed that, whereas multivalent interaction does not affect k′ on significantly, it does alter k′ off considerably. By using the FRET strategy, we were able to measure the changes in kinetics of a single domain rather than the whole molecule, observing 62 times higher binding affinity for the 15Apt domain. This observation confirms that the increased thrombin inhibition potency of the aptamer assembly originated from the kinetic changes caused by cooperative binding. As a byproduct of the study, we have also demonstrated that aptamer-based multivalent ligands make the study of kinetics by using FRET quite convenient compared with antibody or small molecule ligands.
The Antidote Effect of the Binding Aptamer.
One of the unique properties of NA aptamers is that the binding can be readily regulated by using the complementary sequences (16). Strong binding affinity of a target is critical, but reversibility of binding is equally, if not more, important. Reversibility of binding directly impacts the pharmacology of drug treatments in the sense that the side effects of drugs could be quenched by antidotes. Based on their binding affinity as measured by the Watson–Crick base paring of complementary sequences, NA ligands and their complementary NAs are shown to be the most effective drug/antidote pairs. To demonstrate this, the antidote effects of two aptamers' target sequences, T-15Apt and T-27Apt, were investigated. The clotting mixture, including thrombin and fibrinogen with Bi-8S, was treated with excess T-15Apt and T-27Apt separately while the scattering was monitored (Fig. 5). With the treatment of T-15Apt, an immediate scattering increase was seen, and the extent of final scattering intensity was comparable to that observed without any thrombin inhibitors (Fig. 5), indicating that the activity of thrombin is readily recovered by inactivation of 15Apt, and the response is rapid. In contrast, the clotting mixture treated with target DNA of 27Apt showed a slower change in the scattering signal (data not shown), suggesting that its effectiveness in reversing the inhibition of thrombin is limited. We later used the MBA to confirm that the hybridization of 27Apt to its target was much slower than that of 15Apt and its target. It is clear that the secondary structure of 27Apt is quite stable, leading to a slower duplex formation. The second reason is that 15Apt domain was still active even when 27Apt was hybridized to its target. In conclusion, the target DNA of 15Apt is an effective antidote, even for aptamer assembly-based therapy.
Reversible inhibitory function. Red: T-15Apt was added at ≈200 sec to the incubation of Bi-8S, thrombin, and fibrinogen. Black: fibrinogen was added to thrombin at 0 sec in the absence of any inhibitors. Blue: Bi-8S incubated with thrombin and fibrinogen (no T-15Apt).
Antithrombin Potency of Bi-8S.
Recent studies related to its biological functions found that thrombin is critical both in blood clotting disorders and in influencing tumor angiogenesis (23). Thus, after we demonstrated that Bi-8S is the best inhibitor of thrombin in buffer system, we tested Bi-8S in human plasma samples to further demonstrate the potency of the anticoagulant. Standard activated partial thromboplastin time (aPTT) (24) and prothrombin Time (PT) (25) tests were used as described in Materials and Methods. The aPTT and PT are performance indicators measuring the efficacy of the contact activation pathway and extrinsic pathway of coagulation, respectively, as well as the common coagulation pathways. In each test, a different amount of each inhibitor was treated, and the obtained results were plotted by using sigmoid fit. The dosage dependence is shown in Fig. 6. The enhancements in delaying the coagulation triggered by both the contact activation pathway and the extrinsic pathway were consistently observed. The results show that the plasma samples treated with Bi-8S had approximately 5–6 times longer PT and aPTT than those without any treatments, whereas 15Apt alone was able to delay only 2–3 times longer. Even though the enhancement obtained by using human plasma samples was not as great as the one in a buffer system, it proved that the bivalent NA ligand can still function well in human biological fluid and give enhanced anticoagulation efficacy.
Comparison of anticoagulant potency of Bi-8S and 15Apt by using human plasma and aPTT and PT measurements. (a) Dosage-dependent aPTT plotted for each NA inhibitor, and the maximal aPTT is shown inside the figure. (b) Dosage-dependent PT, and the maximal PT recorded appears inside the figure.
Conclusion
In summary, by assembling two thrombin-binding aptamers with optimized linker and linker length, we have developed a NA-based high-performance bivalent ligand, which can be applied as a potential anticoagulant. This design has the combined strengths of both ligands and has achieved enhanced thrombin inhibition capability, indicating its potential in biomedical applications for treating various diseases related to blood clotting disorders. Moreover, the molecular assembly approach offers a simple and noninvasive way to accomplish high performance with known protein inhibitors. It is worth noting that aptamers are ideal candidates by which to construct such assemblies because they can be easily linked in a predictable way, using reliable NA synthetic chemistry. In addition, binding of aptamers to different epitopes of one protein is quite common, and aptamers, theoretically, can be obtained by using a variety of suitable selection strategies for targets ranging from small molecules to complex organisms (7, 8, 26–28). Advanced functions and unique properties can be achieved through simple conjugation of existing aptamers that otherwise would not have the desired performance and capability.
Materials and Methods
Chemicals and Reagents.
All DNA synthesis reagents, including 6-fluorescein phosphoramidite, 5′-Dabcyl phosphoramidite [Dabcyl is 4-(4-dimethylaminophenylazo)benzoic acid], spacer phosphoramidite 18, and d-deoxyphosphoramidite, were purchased from Glen Research. All reagents for buffer preparation and HPLC purification were from Fisher Scientific. A buffer resembling physiological conditions was used for the buffer experiment and contained 25 mM Tris·HCl at pH 7.4, 150 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1 mM CaCl2, and 5% (vol/vol) glycerol. Human α-thrombin was purchased from Haematologic Technologies. Fibrinogen and the sulfated hirudin fragment were obtained from Sigma-Aldrich. Universal coagulation reference plasma (UCRP) and thromboplastin-DL for human sample testing were purchased from Pacific Hemostasis. The aPTT assay reagent was from Trinity Biotech. The bivalirudin was obtained from The Medicines Company.
Synthesis and Purification of Mono- and Bivalent NA Ligands and Their Targets.
To optimize the design of bivalent NA ligand, multiple candidates were designed and prepared (shown in Table S1). All of them were synthesized by using an ABI 3400 DNA/RNA synthesizer (Applied Biosystems) at 1-μmol scale with the standard synthesis protocol. After the complete cleavage and deprotection and ethanol precipitation, the precipitates were dissolved in 0.5 ml of 0.1 M triethylammonium acetate (TEAA, pH 7.0) for purification with HPLC. The HPLC was performed on a ProStar HPLC Station (Varian Medical Systems) equipped with a fluorescence detector and a photodiode array detector. A C18 reverse-phase column (Alltech, C18, 5 μm, 250 × 4.6 mm) was used.
Clotting Time Tests.
To evaluate the inhibitory potency of each NA ligand, we measured the clotting time of each sample containing only thrombin, each NA ligand, and fibrinogen substrate in physiological buffer. The theory behind the experiment is that the mixture of sample becomes nonfluidic when the fibrinogen is digested by thrombin. As a result, the different time points of this transition can be used as an indicator. Briefly, 1 μl of 10 μM thrombin and 1 μl of 100 μM monovalent or bivalent NA ligand were added to a disposable transparent plastic cuvette (Fisher Scientific) containing 200 μl of physiological buffer and then incubated for 15 min. In the case of nonlinked 15Apt and 27Apt mixture, 1 μl of 100 μM each probe was applied. Following that, 4 μl of 20 mg/ml fibrinogen was added, and samples in the cuvette were carefully examined by tilting the cuvette to record the time when the sample became nonfluidic. Each experiment was performed in tandem. A reaction mixture containing only thrombin and fibrinogen was always tested together with other samples as an internal standard. All clotting times were normalized based on the internal standard and compared with it.
Real-Time Monitoring of the Clotting Reaction.
To monitor the clotting event in real time, we used light scattering. When fibrinogen is digested by thrombin, the sample mixture becomes not only nonfluidic but also cloudy. Turbidity of samples can be measured by using either absorption or light scattering. To monitor clot formation, we chose light scattering. Briefly, reaction mixtures were prepared in the same way as the clotting tests described above, except that the reaction took place in a 100-μl quartz fluorescence cuvette (Starna Cells), and the light scattering was monitored on a Fluorolog-3 spectrofluorometer (Jobin Yvon). For monitoring scattering, the excitation and emission wavelengths were both set at 580 nm, and the emission was detected at a right angle relative to the light excitation so that the excitation light did not interfere with the light-scattering signal. The initial rate of scattering increase represented the relative thrombin-inhibition strength of the tested sample. Initial rates were calculated from the linear range of the early slopes of the scattering profiles.
Monitoring of the Apparent kon and koff.
Monitoring the binding kinetics was accomplished by modifying each inhibitor with fluorescein and Dabcyl. The sequence of
each probe is shown in Table S1. To compare the k′
on of each inhibitor with thrombin, we preincubated 100 nM each inhibitor with 100 nM T′-15Apt. The fluorescence decay, after
a 5-fold excess of thrombin (500 nM) was added, was monitored on a Fluorolog-3 spectrofluorometer. Obtained results were used
to calculate the kinetic parameters in the following way:
To monitor the k′
off of each inhibitor, 500 nM T-15Apt was added to the preincubated mixture of thrombin (500 nM) and each ligand (100 nM). To
generate the values of k′
off, the following equation was applied:
Reversible Binding Reaction Using Target DNAs.
To test the reversible binding of Bi-8S, we treated the sample mixture with target DNAs of 27Apt or 15Apt. The sample mixture, including fibrinogen, was prepared in the same way as indicated for the clotting time. About 500 sec after fibrinogen was added to the reaction mixture, each target DNA of either 15Apt or 27Apt was added to reach a final concentration of ≈5 times that of Bi-8S.
Human Plasma Tests.
To evaluate the feasibility of the bivalent NA ligand as a potential anticoagulant reagent, we determined aPTT and PT for each ligand by using human plasma samples. Procedures applied were those recommended by the supplier. For aPTT determination, 50 μl of UCRP was preincubated at 37°C with a different amount of each ligand for 2 min; then 50 μl of aPPT-L was added and incubated for another 200 sec. Next, 50 μl of prewarmed CaCl2 was added to initiate the intrinsic clotting cascade. Finally, the scattering signal was monitored until the signal was saturated. For PT determination, 50 μl of UCRP was preincubated at 37°C with a different amount of each ligand for 2 min; then 50 μl of thromboplastin-L was added to initiate the extrinsic clotting cascade. Finally, the scattering signal was monitored until the signal was saturated. For the calculation of aPTT and PT, the end time was determined to be the point where scattering signal reached half maximum between lowest and maximum points. It was repeated twice, and each set of experiments was done with one batch of plasma.
Acknowledgments
We thank many group members for discussion. This work was supported by National Institute of General Medical Sciences Grant R01 GM079359, National Institute of Neurological Disorders and Stroke Grant U54 NS058185, and Office of Naval Research Grant N00014-07-1-0982.
Footnotes
- *To whom correspondence should be addressed. E-mail: tan{at}chem.ufl.edu
-
Author contributions: Y.K. and Z.C contributed equally to this work; Y.K., Z.C., and W.T. designed research; Y.K., Z.C., and W.T. performed research; Y.K. and Z.C. contributed new reagents/analytic tools; Y.K., Z.C., and W.T. analyzed data; and Y.K., Z.C., and W.T. wrote the paper.
-
The authors declare no conflict of interest.
-
This article is a PNAS Direct Submission.
-
This article contains supporting information online at www.pnas.org/cgi/content/full/0711803105/DCSupplemental.
- © 2008 by The National Academy of Sciences of the USA
