The single-stranded DNA aptamer-binding site of human thrombin.

A new class of thrombin inhibitors based on sequence-specific single-stranded DNA oligonucleotides (thrombin aptamer) has recently been identified. The aptamer-binding site on thrombin was examined by a solid-phase plate binding assay and by chemical modification. Binding assay results demonstrated that the thrombin aptamer bound specifically to alpha-thrombin but not to gamma-thrombin and that hirudin competed with aptamer binding, suggesting that thrombin's anion-binding exosite was important for aptamer-thrombin interactions. To identify lysine residues of thrombin that participated in the binding of the thrombin aptamer, thrombin was modified with fluorescein 5'-isothiocyanate in the presence or absence of the thrombin aptamer, reduced, carboxymethylated, and digested with endoproteinase Arg-C. The digestion products were analyzed by reversed-phase high performance liquid chromatography and the peptide maps compared. Four peptides with significantly decreased modification in the presence of the aptamer were identified and subjected to N-terminal sequence analysis. Results indicated that B chain Lys-21 and Lys-65, both located within the anion-binding exosite, are situated within or in close proximity to the aptamer-binding site of human alpha-thrombin. The thrombin aptamer binds to the anion-binding exosite and inhibits thrombin's function by competing with exosite binding substrates fibrinogen and the platelet thrombin receptor.

Thrombin is a serine protease responsible for the conversion of fibrinogen to fibrin, platelet activation, and the cleavage of coagulation factors V, VIII, XI, and XI11 (Fenton, 1981;Shuman, 1986;Furie and Furie, 1988;Mann et al., 1990;Gailani and Broze, 1991). In addition to its role in blood coagulation, thrombin can act as a potent mitogen (Chen and Buchanan, 1975;Carney et al., 1985) and can also exert a chemotactic effect on monocytes (Bar Shavit et al., 1983). Because of its pivotal role in both thrombosis and hemostasis, thrombin is a major target for anticoagulation and cardiovascular disease therapy. Using a novel in vitro selection/amplification technique, a new class of thrombin inhibitors based on single-stranded DNA (ssDNA)l oligonucleotides has been recently identified (Bock et al., 1992). These thrombin inhibitors are the first example of the use of this technique to * This work was supported in part by National Institutes of Health Small Business Innovative Research Grant 1-R43 HL48431. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed Gilead Sciences, Inc., 353 Lakeside Dr.,Foster City, The abbreviations used are: ss, single-stranded; FITC, fluorescein 5'-isothiocyanate; HPLC, high performance liquid chromatography. obtain ssDNA oligonucleotides that bind a target protein with no known specificity for nucleic acids. One oligonucleotide GGTTGGTGTGGTTGG (thrombin aptamer) was capable of nanomolar inhibition of fibrinogen cleavage in vitro (Bock et al., 1992) and was shown to inhibit clot-bound thrombin and reduce arterial thrombus formation in an ex vivo whole artery angioplasty model.' Recent in vivo studies in cynomolgus monkeys have shown the thrombin aptamer to be a potent anticoagulant with a rapid onset of action and a short halflife (Griffin et al., 1993). The three-dimensional structure of thrombin has been solved by x-ray crystallography (Bode et al., 1989), and, recently, the tertiary structure of the thrombin aptamer has been elucidated by NMR spectroscopy (Macaya et al., 1993;Wang et al., 1993). Identification of the aptamer binding site on thrombin will help define the structural basis of this novel ssDNA-protein interaction and guide further development of a thrombin aptamer with an improved therapeutic profile.
Unlike hirudin and other active site thrombin inhibitors, the thrombin aptamer does not inhibit the cleavage of small chromogenic amide substrates3 (Wu et al., 1992), indicating that the aptamer does not bind directly to the active site of thrombin. However, distinct from the catalytic center of thrombin are two highly basic regions that form secondary binding sites on the surface of the molecule. These sites are important for thrombin's specificity in interactions with several macromolecular substrates and receptors. One site, the anion-binding exosite, contributes to the formation of a tight specific complex with fibrinogen (Fenton et al., 1988, Church et al., 1989, thrombomodulin (Wu et al., 1991), hirudin (Chang, 1989;Grutter et al., 1990;Rydel et al., 1990), the platelet thrombin receptor (Vu et al., 1991)) and heparin cofactor I1 (Sheehan et al., 1991;Rogers et al., 1992). The second site, the putative heparin recognition site, contributes to the significant increase of thrombin inactivation by antithrombin I11 in the presence of heparin (Chang et al., 1979;Messmore et al., 1979). Considering the polyanionic nature of the nucleic acid phosphodiester backbone, these basic sites on thrombin, shown to interact with acidic regions of other thrombin-binding molecules, are likely targets for thrombin aptamer binding.
In this paper the area of thrombin involved in aptamer binding was investigated by both a solid-phase binding assay and by chemical modification of the thrombin-aptamer complex. The thrombin aptamer was shown to bind within the anion-binding exosite, and two of the lysine residues4 that participate in aptamer binding were identified.

EXPERIMENTAL PROCEDURES
Materials-Human a-thrombin and human y-thrombin were from Haematologic Technologies (Essex Junction, VT). Deoxyoligonucleo-W. X. Li and L. L. K. Leung, personal communication. Louis Bock, personal communication. ' Amino acid residues are numbered sequentially beginning with the first residue of the human a-thrombin B chain. tides, GGTTGGTGTGGTTGG (thrombin aptamer) and GGTGGT-GGTTGTGGT (15-mer scramble sequence) were prepared by solidphase phosphoramidite chemistry on a Biosearch 8800 synthesizer using standard methods and assayed as described previously (Bock et al., 1992). Biotinylated oligonucleotides were prepared by coupling biotin-X-X-NHS ester (Glen Research) to the 5' amine of both the 5' modified (hydroxyl to amine) thrombin aptamer and the 15-mer scramble sequence. Avidin DN and anti-avidin alkaline phosphatase were from Vector Laboratories. Hirudin was from American Diagnostica, Inc. Fluorescein 5'-isothiocyanate (isomer 1) (FITC) was from Molecular Probes, Endoproteinase Arg-C was from Boehringer Mannheim, NICK (Sephadex G-50) and NAP-5 (Sephadex G-25) disposable gel filtration columns were from Pharmacia LKB Biotechnology Inc.
Thrombin Aptamer Binding Assay-A 96-well enzyme-linked immunosorbent assay plate (Corning) was coated overnight at 4 "C with 250 nM thrombin, human albumin, or chymotrypsin in selection buffer (Bock et al., 1992). The excess sites on the plates were blocked with 1% bovine serum albumin, 0.05% Tween 20 in selection buffer. Biotinylated thrombin aptamer or the biotinylated 15-mer scramble sequence were added at increasing concentrations and incubated for 2 h at room temperature. The plate was washed with 0.1% bovine serum albumin, 0.05% Tween 20 in selection buffer. Binding was detected by incubating the plates with Avidin DN (250 pg/ml) for 1 h at 4 "C followed by anti-avidin-alkaline phosphatase (1:250) for 1 h at 4 "C. The plate was washed in between each incubation step. The substrate, p-nitrophenyl phosphate, was added and incubated at 37 "C for an additional hour. The absorbance at 405 nm was measured using a Molecular Devices plate reader.
For demonstrating competition, 500 nM biotinylated thrombin aptamer was preincubated for 1.5 h at room temperature with increasing concentrations of competitors. The incubation mixture was then added to a plate previously coated with 250 nM thrombin, and binding was detected as described above.
Modification of Thrombin and Aptamer-Thrombin Complex with FITC-The derivatization of thrombin was performed essentially as described (Chang, 1989) except FITC was used in place of 4-N-Ndimethylaminoazobenzene-4' isothiocyano-2'-sulfonic acid (S-DA-BITC). Briefly, the glycerol was removed by gel filtration chromatography (NICK) and the thrombin (37.5 pM) was resuspended in selection buffer adjusted to pH 8.4 for conjugation. An equal volume of 4 mM FITC in selection buffer was added, and the mixture was incubated for 3 h at room temperature. The reaction was terminated by passing the sample over a NAP-5 (Sephadex G-25) column that was equilibrated in 50 mM NH4HC03, pH 8.3. The amount of derivatization (mol of FITC/mol of protein) was calculated based on a molar extinction coefficient of the fluorescein group of 64,000 M" cm" at 495 nm. The protein concentration was based on a BCA (Pierce Chemical Co.) protein assay using thrombin as the standard.
To derivatize the aptamer-thrombin complex, 37.5 p~ thrombin was incubated with a 5-fold molar excess (187.5 p~) of the thrombin aptamer in selection buffer for 10 min at room temperature. An equal volume of 4 mM FITC in selection buffer was then added, and the mixture was incubated for an additional 3 h. Removal of unincorporated FITC and determination of the extent of modification were as described above for thrombin alone.
Structural Analysis of FITC-modified Protein and DNA-Protein Complex-The derivatized thrombin and aptamer-thrombin complex were lyophilized and then resuspended in 400 ~1 of 5 M urea, 250 mM Tris-HC1,5 mM EDTA, pH 8.5, containing 20 mM dithiothreitol. The samples were incubated overnight at room temperature. Iodoacetic acid (1 M in 1 M NaOH) was added to a final concentration of 100 mM. Incubation was at room temperature for 30 min, and the reaction was stopped by the addition of 100 mM dithiothreitol. Reduced and carboxymethylated thrombin and the thrombin-aptamer complex were desalted using a NAP-5 column equilibrated in 50 mM NH,HCO,, pH 8.3, and then lyophilized. The samples were then digested with endoproteinase Arg-C (1:60 enzyme/substrate weight ratio) overnight at 37 "C according to manufacturer's instructions. The digestion products were separated by reversed-phase HPLC using a Hewlett Packard 1090 M system and a Vydac CI8 (250 X 4.6 mm) column. The column was equilibrated in 0.1% trifluoroacetic acid in 5% acetonitrile (solvent A) at 40 "C. Following sample injection, a linear gradient at 1 ml/min from 7-65% solvent B (0.08% trifluoroacetic acid in 90% acetonitrile) was developed over 80 min. The eluent was monitored at both 214 and 440 nm, and the peaks of coincident absorbance were manually collected. Aliquots of each peak were subjected to N-terminal sequence analysis using a Hewlett Packard GlOOA system.

RESULTS AND DISCUSSION
Competition of Thrombin Aptamer Binding to Immobilized a-Thrombin by y-Thrombin and Hirudin-To characterize the aptamer-thrombin interaction, a solid-phase plate binding assay was developed using the biotinylated thrombin aptamer and a biotinylated 15-mer scramble sequence. To assess the effect of biotinylation on thrombin inhibitory activity, these oligonucleotides were assayed for their anticoagulant activity in a plasma thrombin time clotting assay. The biotinylated thrombin aptamer was only slightly less active than the unmodified thrombin aptamer, requiring approximately 1.7fold higher concentration to double the thrombin clotting time. The biotinylated 15-mer scramble sequence, like the unmodified 15-mer scramble sequence, was completely inactive (data not shown). Increasing amounts of the biotinylated thrombin aptamer or the control biotinylated 15-mer scramble sequence were added to immobilized a-thrombin (Fig. hi).
The thrombin aptamer bound a-thrombin in a dose-dependent manner, whereas the 15-mer scramble sequence did not bind. The thrombin aptamer did not bind chymotrypsin or albumin, demonstrating the specificity of binding.
To identify the thrombin aptamer binding site, y-thrombin, an active site competent autoproteolyzed derivative of athrombin, was assayed for its ability to compete with immobilized a-thrombin for aptamer binding. In addition, hirudin, a potent and specific thrombin anion-binding exosite inhibitor, was tested for its ability to compete with the aptamer for thrombin binding. The results are displayed in Fig. 1B. The unmodified thrombin aptamer competed with the biotinylated thrombin aptamer for thrombin binding, whereas no competition was observed with the unmodified 15-mer scramble sequence. a-Thrombin effectively competed with the immobilized a-thrombin with maximum inhibition observed at 500 nM. However, even at a 2000-fold excess concentration, ythrombin, which contains a cleavage within the anion-binding exosite as well as two additional cleavages to produce four noncovalently associated fragments, did not compete with immobilized a-thrombin for aptamer binding. Therefore, the regions that differ between a-thrombin and y-thrombin may participate in thrombin aptamer binding. Furthermore, hirudin, which is known to bind in the anion-binding exosite, effectively competed with the aptamer for thrombin binding. Taken together, these data suggest that the anion-binding exosite is important for aptamer-thrombin interactions.
Identification of Thrombin Lysine Residues Involved in Aptamer-Thrombin Complex by FITC Derivatization and Microsequencing-To identify the essential lysyl residues in human a-thrombin that participate in thrombin aptamer binding, both the aptamer-thrombin complex and thrombin alone were derivatized with FITC, a reagent for t-amino group modification. a-Thrombin was incubated in the presence or absence of an excess of the thrombin aptamer. The extent of modification was 3.9 mol of FITC/mol of thrombin in the absence of aptamer. In the presence of the aptamer, the derivatization was decreased approximately 30% to yield 2.7 mol of FITC/ mol of thrombin.
Following the FITC modification, the two samples were reduced, carboxymethylated, and digested with endoproteinase Arg-C. The digestion products were analyzed by reversed-phase HPLC with detection at 440 nm to identify the lysine residues in thrombin that were protected by the aptamer from FITC modification (Fig. 2). In the absence of the aptamer, approximately 20 FITC-labeled peptides were generated (Fig. u). As a result of aptamer binding, the recovery (0). This incubation mix was then added to the coated plate and incubated for an additional hour at room temperature. Binding was detected as described under "Experimental Procedures." of four peaks was drastically reduced (Fig. 2B). These peaks, Pl-P4, were selected for further analysis, because they represented significant absorbance at 214 nm and consistently showed a greater than 50% decrease in absorbance at 440 nm. The four peaks were collected and subjected to N-terminal sequence analysis (Table I). Peaks P1 and P2 both contain peptides with the identical sequence, Ile-Gly-(Lys)-His-Ser-Arg, which corresponds to Arg-C peptide 63-68. FITC-labeled lysine is not detected in the sequencer; therefore, the lack of lysine at the third position in these peptides demonstrated that Lys-65 was modified in the absence of the aptamer and protected by the aptamer from FITC derivatization. Peaks P3 and P4 were both found to contain the Arg-C peptide 21-36 (Table I). The extremely low recovery of Lys at the Nterminal position of this peptide indicated that Lys-21 was partially protected by the aptamer, but interpretation of these results was complicated by the presence of unrelated coeluting

P1
Ile ( The underlined residues represent fluoresceinated lysyl residues. Quantitative yields of amino acid residues in pmol are given in parentheses following the identified residue. NQ, not quantitated ND, not detected.
peptides. The identification of Lys-21 as the second protected residue was confirmed by the sequencing results obtained with peak 4, which contained no contaminating peptides and no detectable Lys at the N-terminal position of Arg-C peptide The major peptides present in peaks P1 and P2 have identical amino acid sequences and compositions but different retention times, as do the major peptide components of peaks P3 and P4. This may be due to a rearrangement of a portion of the fluorescein group, which alters the hydrophobic properties of the peptides and, hence, their retention times. This phenomenon has been observed for several other FITC-labeled peptides (Mitchinson et al., 1982;Farley et al., 1984;Phillips, 1988).
A basic premise of these experiments is that in the aptamerthrombin complex, amino acids in the thrombin binding site are shielded and become relatively inaccessible to FITC modification. These results indicate that both Lys-21 and Lys-65, which are in the anion-binding exosite of a-thrombin, are located either within or in close proximity to the thrombin aptamer binding site. The possibility exists that ssDNA binding induces conformation changes that interfere with the chemical modification of amino acid residues that are not in the combining site. However, this is unlikely, since the lack of competition by y-thrombin and the effective competition 21-36.
of aptamer binding to immobilized thrombin by hirudin indicated that the anion-binding exosite is important for aptamer binding (Fig. 1).
Lys-21 and Lys-65 also participate in fibrinogen-thrombin interactions (Church et al., 1989). These residues, along with four other lysyl residues of a-thrombin, were shown to be protected from chemical modification following complex formation with hirudin (Chang, 1989). Recently, a thrombin mutant that contained an Arg to Glu substitution at position 70 (Arg-70 + Glu) in the anion-binding exosite was shown to no longer bind the thrombin aptamer (Wu et al., 1992). Using a complementary approach, our results extend this observation that the thrombin aptamer binds to the anion-binding exosite to include that Lys-21 and Lys-65 play a role in that interaction. Finally, Griffin et al. (1993) and others (Wu et al., 1992) have observed that the thrombin aptamer inhibits thrombin-induced platelet activation. Binding of thrombin's anion-binding exosite by the platelet thrombin receptor has been shown to be involved in platelet activation (Vu et al., 1991). Thus, the thrombin aptamer binding site appears to overlap with the binding sites of hirudin, fibrinogen, thrombomodulin, and the platelet thrombin receptor.
Recently, the tertiary structure of the thrombin aptamer has been elucidated by NMR spectroscopy (Macaya et al., 1993;Wang et al., 1993). The ssDNA molecule forms a highly compact, partially symmetrical structure in solution containing two syn-anti-syn-anti G-tetrads, a loop which is partially folded over one tetrad and two T-T loops folded under the other tetrad. The stability and rigidity of this sequencedependent structure probably contributes to the formation of a tight complex with thrombin. With the identification of the protected lysines in the anion-binding exosite, the thrombin aptamer can be "docked" onto the surface of thrombin by molecular modeling. Elucidation of the structural basis of aptamer-thrombin interaction will greatly aid medicinal chemistry efforts to improve the therapeutic profile of the thrombin aptamer.