The Solvent Isotope and Ionic Effects on Thrombin-Hirudin Interactions

John Paul Sheehy

Biochemistry Comprehensive Submission

April, 24 2008

Abstract

Hirudin, found naturally in medicinal leaches, is the most potent reversible inhibitor of thrombin known to exist. It interacts non-covalently over a large area, especially on the fibrinogen-binding exosite of the enzyme. As hydrogen-bonding interactions play a major role, this study targeted the dependence of rate constants and the inhibition constant, Ki, on ionic strength (), sodium ion concentration, nature of buffer, and the presence of D2O in buffers. Kinetic measurements were conducted spectro-photometrically and spectro-fluorometrically in the presence of a chromophoric or fluorophoric substrate, respectively. The dissociation constant and the second-order-rate constant for the slow-tight-binding inhibition of human -thrombin with r-hirudin1,3 are 1.9 ± 0.4 pM and (7.14 ± 0.07) x 107 M-1 s-1 at [Na+]=  = 0.19 M in pH 8.06, 0.05 M barbital buffer, and 0.56 ± 0.07 pM and (1.50 ± 0.06 ) x 108 M-1 s-1 at Na+= 0.30,  = 0.31 M in pH 8.12, 0.02 M Tris buffer at 25.0  0.1 C. There is a strong buffer molecule dependence, with results in Tris buffers differing consistently from those in barbital buffers. When this is considered, Na+ ion concentration appears to slightly lower the forward kinetic constant, k1. The solvent isotope effects on Ki are 1.3 ± 0.5, 0.9 ± 0.1, 1.4 ± 0.7, 1.03 ± 0.7 between pH 8.1 and 8.3 at ionic strength 0.19-0.31 at 25.0 ± 0.1 C. The solvent isotope effect on the second-order rate constant, k1, is typically lower than unity:1.04 ± 0.15, 0.71 ± 0.30, 0.64 ± 0.16, 0.68 ± 0.05 under the same conditions.

Introduction

Thrombin (EC 3.4.21.5) is a serine protease which plays an important role in blood coagulation. It is activated from prothrombin in the final steps of coagulation, and then catalyzes the conversion of fibrinogen into fibrin, which actually forms blood clots.1 α-thrombin has two polypeptide chains, a thirty-six residue A-chain and a 259 residue B-chain, connected by a disulfide bridge. The longer B-chain also includes several internal disulfide bridges.1 α-thrombin is prone to proteolytic cleavage by trypsin or autolysis at an Arginine 77, and the resulting form is known as β-thrombin. Other, similar breaks form other versions of thrombin, including γ-thrombin and ε-thrombin.1 The function of thrombin in the clotting cascade is shown in Figure 1 below.

Figure 1: Thrombin’s role in clotting cascade.2

Hirudin is one of many substances in the saliva of leeches which helps prevent the clotting of ingested blood in the digestive tract.3 In fact, it is the most potent natural inhibitor of thrombin. Its primary structure is shown in Figure 2. It is sixty-five amino acid residues in length, with glutamine at the C-terminus, and isoleucine as the N-terminal residue.1 The structure contains three disulfide bridges, between cysteines at the sixth and fourteenth, the sixteenth and twenty-eighth, and the twenty-second and 39thpositions. The latter two bonds form a “double-loop.”1From the fortieth to sixty-fifth residue runs the C-terminal tail, which contains no disulfide bonds, but does have one peculiarity at the tyrosine in the sixty-third position, which is sulfonated in the natural version of the molecule. Unfortunately, the recombinant form does not have this sulfate group. Some variant molecules exist with differing amino acids at a number of positions. In contrast to the long, unbound tail, the N-terminal domain is compact due to the double loops.1

Grutter et al. report that the interactions between hirudin and thrombin are quite different from those of other serine protease inhibitors and their targets. Typically such inhibitors make contact primarily in the active site region, but hirudin interacts with thrombin over a comparatively larger area (1400 Å2). This large area of interaction is probably one of the factors contributing to the extreme tightness of the complex.3 The loop region from residues 5-45 of hirudin is, for the most part, not in contact with thrombin,3 except for two salt bridges, two hydrogen bonds, and thee hydrophobic contacts.1 Both N-terminal head and the 18 residue C-terminal tail region of hirudin do make contact.3 The N-terminal three residues of hirudin occupy the active site of thrombin, but in a novel way. Most serine-protease inhibitors derive their specificity from their ability to bind the primary specificity pocket of thrombin.3 Aside from the presence of hirudin’s second residue, a threonine, near the entrance to this pocket, the inhibitor leaves it free, and derives its specificity from numerous interactions both inside and outside of the active site.3 Fifteen of the first 48 residues of hirudin make some contact with thrombin, totaling 103 interactions of less than 4Å.

The C-terminal chain of hirudin binds with the fibrinogen exosite of thrombin.1 This chain has an unusually long extended conformation, and makes several hundred contacts within less than 4Å with thrombin.1 A site-directed mutagenesis study by Betz, Stone, and Hofsteenge showed that, at least for the acidic residues 53, 55, 57, 58, 61, and 62, the residues contribute to the total binding energy (75 kJ/mol) in relatively similar amounts (from 2.3 to 5.9 kJ/mol).4 An earlier site-directed mutagenesis study of several basic residues in both domains of hirudin found that only the Lysine 47 made a major contribution to thrombin binding.5 An image of the interactions can be seen in Figure 3 below.

Given the large number of interactions between thrombin and hirudin, it is worth while to consider whether any short strong hydrogen bonds (SSHB), exist in the thrombin-hirudin complex. Such bonds, shorter than 2.6Å, can be important contributors to enzyme catalysis, and thus also sometimes appear in enzyme-inhibitor complexes.6They can be as much as six times stronger than an ordinary hydrogen bond. SSHBs typically form in the absence of a hydrogen bonding solvent (water) and when the two atoms involved have similar pKa’s and are of the same element (Oxygen or Nitrogen).6 One example occurs in the catalytic activity of triosephosphate isomerase, one of the most efficient enzymes known to exist.6,7 Evidence has been presented that serine proteases, such as thrombin, trypsin, and chymotrypsin, have SSHBs between the catalytic aspartate and histidine in the active site.”6,7

Hirudin is both a slow-binding and tight-binding inhibitor. “A reversible tight-binding inhibitor is one that exerts its reversible inhibitory effect on an enzyme-catalyzed reaction at a concentration comparable to that of the enzyme. Therefore, allowance must be made for the change in the concentration of free inhibitor that occurs as a result of it undergoing interaction with one or more forms of the enzyme.”8 This differs from classical inhibitors, which only cause inhibition at concentrations far greater than that of the target enzyme. This means that such inhibition cannot be quantitatively described with the Michealis-Menten equation, “since the assumption that the free inhibitor concentration is equal to the total inhibitor concentration is not valid.”8

Hirudin is known as a slow-binding inhibitor because the strength of enzyme-inhibitor interactions lead to a delay in the establishment of a steady state velocity.8 When the equilibrium for the interaction is not established rapidly, a “transient or pre-steady-state” phase occurs in a plot of product formation vs. time.8

The reaction scheme chosen to describe the inhibition of thrombin by hirudin includes reversible inhibition of the free enzyme in the presence of substrate. The substrate reaction is described by two steps: the formation of the enzyme-substrate complex, and the subsequent formation of product and free enzyme.

The formation of product vs. time is described by the equation

P= vst + (vs-vo)(1-d)/(dk)*log((1-de-kt)/(1-d)) eq.19,8

where vs is the steady state reaction velocity, vo is the reaction velocity in the absence of inhibitor, t is time, P is the amount of product formed, d is a function equivalent to (Ki+Et+It-[(Ki+Et+It)2-4EtIt]1/2)/(Ki+Et+It+[(Ki+Et+It)2-4EtIt]1/2)

where Ki is the dissociation constant, Etis the concentration of all enzyme present in solution, It is the concentration of all inhibitor present in the solution, and k is a sort of second-order rate constant equivalent to

k1[(Ki+Et+It)2-4EtIt]1/2/(1+S/Km)

where k1is the forward kinetic constant for inhibition and S is the concentration of substrate.8

According to a study by Stone and Hofsteenge, the dissociation constant of the hirudin-thrombin complex depends greatly on ionic strength, increasing approximately twenty fold between the ionic strengths of 0.1 and 0.4. This increase in dissociation constant is largely caused by a decrease of k1. In fact, k-1 also decreases with ionic strength, which, by itself, would actually contribute to tighter interactions, but the dependence of k1on ionic strength is much more significant.9According to a later study of Stone, Hofsteenge, and Dennis, each negatively charged residue in contact with thrombin makes an approximately equal contribution of -4 kJ/mol to the binding energy of the hirudin-thrombin complex. “For native hirudin, ionic interactions accounted for 32% of the binding energy at a [theoretical] ionic strength of zero.”10

The goal of this studywas to determine the deuterium solvent isotope effect on the interactions of α-thrombin and recombinant hirudin at a number of differing ionic strengths and sodium concentrations. The data obtained could be useful for better understanding the strength of the thrombin-hirudin interactions and the potential existence of SSHBs.

Materials and Methods

Materials Anhydrous dimethyl sulfoxide (DMSO), heavy water with 99.9 % deuterium content and anhydrous methanol, were purchased from Aldrich Chemical Co. All buffer salts were reagent grade and were purchased from either Aldrich, Fisher, or Sigma Chemical Co. H-D-Phe-Pip-Arg-4-nitroanilide.HCl (pNA) (S-2238) 99% (TLC) was purchased from Diapharma Group Inc. and Boc-Val-Pro-Arg-7-Amino-4-Methyl Coumarin (VPR-7AMC) was from Sigma Chemical Co.. Human -thrombin, MM 36,500 d, 3010 NIH u/mg activity in pH 6.5, 0.05 M sodium citrate buffer, 0.2 M NaCl, 0.1% PEG-8000 was purchased from Enzyme Research Laboratories, OH. R-hirudin was purchased from Pentapharm, Lot 40907401/126-05 and Lot 405383/126-05.

Instruments Spectroscopic measurements were performed with a Perkin-Elmer Lambda 6 UV-Vis Spectrophotometer connected to a PC. The temperature was monitored using a temperature probe connected to a digital readout device. Either a Neslab RTE-4 or a Lauda 20 circulating water bath was used for temperature control. Fluorometric measurements were performed with Cary Eclipse fluorometer with a Peltier temperature control system and PC. Positive displacement Gilson and Rainin Microman pipettes with plastic tips were used for the delivery of enzyme solution, substrate solution and inhibitor solution.

Solutions Buffers were prepared by weight from Tris-base and Tris-HCl, Sodium Barbiturate, and Barbital. The concentrations were0.02 M Tris, 0.05 M Barbital, or 0.01 M Sodium Barbital, 0.3 M, 0.15 M, or .03 M NaCl, and 0.1% PEG4000 in H2O or D2O. The pH was adjusted to near 8.0. Water was distilled from a copper-bottom distiller, run through an ion-exchanger and distilled before use. Buffer solutions in D2O were made identically to their counterpart in H2O, although one buffer, 0.01 Sodium Barbital, 0.03 NaCl, was prepared by diluting an H2O 1:5 in D2O to form an 80% D2O buffer. The negative log of the deuteron concentration, pD, in these solution was measured by the same pH electrode and pD was calculated by adding 0.4 to the electrode reading.11 The pD was typically near 8.7. Buffer pH and pD was monitored during storage. Thrombin and hirudin were typically obtained from stock solutions stored at -20° C. Small stocks of thrombin were prepared identically in advance, and were further diluted in two steps to about 4 nM in the buffer with the parameters under study. Hirudin stock solutions were similarly prepared at 180 nM, and just before use were diluted with the buffer under study to stock solutions of approximately 36 nM and 3.6 nM respectively. The substrate was prepared in DMSO at a concentration of approximately 5 mM and stored at 0°C.

Protocol A single set of experimental data was collected in a single session, and the temperature was maintained at 25.0±0.1°Cand monitored with a thermistor in all experiments.

In the spectrophotometric studies absorbance was measured at 405 nm to monitor pNA release from S-2238. Glass cuvettes were placed in the cell compartment before the reaction was begun so that they would acquire the proper temperature.

The fluorometric studies were performed with excitation set to 365 nm and emission to 445 nm with the PMT sensitivity set to 600 V and both slits to 5 nm. It was determined that 0.57 Arbitrary Units were equal to 1 nM of 7AMC in a fluorometric cell.

Cells were prepared as follows: 10-25 uL of the approximately 4 nM thrombin solution, 5-50 uL of either the 36 or 3.6 nM hirudin and 10-20 uL of the 5-mM substrate solution were added to enough of the buffer under study to have a total volume of 2 mL. The order of addition was always the same: buffer, substrate, inhibitor, enzyme. Before the addition of enzyme, the instrument reading was be “zeroed.” Upon the addition of enzyme, the number of seconds between this moment and the time the instrument began recording data was measured with a stopwatch. Change in fluorescence in the cell was recorded ten times per second for a given length of time, usually 1000-2000 seconds. Initial velocities were measured without any inhibitor in order to measure enzyme activity. These were typically performed in the corresponding H2O buffer during D2O experiments. If enzyme activity dropped significantly, a new stock solution was obtained and the added volume of thrombin would be changed if necessary to maintain the same activity.

The spectrophotometric experiments were performed similarly, although the cells were brought to a total volume of 1 mL, no stopwatch was used to measure initial delay, and the instrument recorded absorbance only once every 1-3 seconds, depending upon the particular length of the run.

Data Processing The time course data curves were first processed in Microsoft Excel. The fluorometric calibration curve conversion, 0.57 AU/nM or the pNA extinction coefficient, 9920 M-1cm,-1was applied, so that the concentrations were listed in nM. Also, the number of seconds recorded on the stopwatch was added to the time values of the ordered pairs, shifting the curve slightly to the right, and manual curve smoothing was applied. Once the data curves were thus prepared, they were transferred into the fitting program Graphit and the steady state velocity (the linear slope which curves ultimately reach, was calculated by means of fitting to a simple version of the progress curve equation (eq. 1). The actual input in the Graphit equation software was as follows:

P=vs*(t)+((((vo-vs)*(1-(d)))/((d)*k))*log((1-(d)

*exp(-k*(t)))/(1-(d))))+h eq. 2

The symbols in eq. 2 are as were defined above, but for this level of fitting, d was simplified, left merely as d, and h was added as an offset (used because the measurement of the delay between the start of the reaction and the start of data collection was not always perfect, and occasionally there was a question as to whether or not a run had been “zeroed.”) P was the dependent variable, t the independent variable, and d, vo, vs, k, and h were found by means of a non-linear least squares fit. The Graphit fitting style was “Simple” without “Robust,” and this fitting was not sensitive to initial estimates within at least one order of magnitude. A full set of data curves for a single determination may be seen in Figure4 below.

The resulting vs values were then collected and plotted against the labeled hirudin concentration based on the activity determined by Pentapharm, and calculated with the assumption that one anti-thrombin unit (ATU) was equal to 8.5 pmol of r-hirudin. An additional data point was obtained at the intersection of 0 pM hirudin with the average velocity of the runs performed without r-hirudin in the experimental buffer.

The dependence of steady state velocity on inhibitor concentration was modeled in Graphit:

vs=(vo/(2*E))*((((Ki+(I*x)-E)^2+(4*Ki*E))^(1/2))-(Ki+(I*x)-E))+h eq.3

In this formula, vs and vo are as previously defined. E is enzyme concentration, I is the labeled inhibitor concentration, x is a factor which represents the ratio of the true inhibitor concentration to the concentration calculated by Pentapharm, Ki is the apparent dissociation constant Ki’, and h is an offset which represents activity believed to be due to contaminating β-thrombin and γ-thrombin which are less susceptible to r-hirudin. Such activity can be eliminated by using an irreversible covalent inhibitor PPACK, but even substantially large concentrations of r-hirudin do not eliminate it. In this fitting, vs was the dependant variable, and I was the independent variable. The offset, h, was arbitrarily decided to be 80% of the activity of the lowest point measured, and E was calculated using the measured kcat of 73.8 s-1 for VPR-7AMC, and 95 s-1 for S-2238 in pH 8.4-8.6, 0.02 Tris buffer, 0.3 M NaCl, 1% PEG400, and 1-2% DMSO at 25.0 ± 0.1° C.12 Values of kcatin other buffers or at other pH were also determined by measuring the difference in activity when the same volume of the same solution of thrombin was added to the different buffers. For the 0.1 M sodiumbarbital in D2O, the kcat was assumed to be 31.5 s-1, one third of the corresponding H2O value,due to a large effect of D2O. Ki, vo and x were then fitted using a non-linear least squares fit with “Simple Robust” fitting. This fit was also not sensitive to estimates differing within an order of magnitude. A sample fitting is shown in Figure 5 below.

The apparent Ki’ could then be corrected for substrate concentration using the formula:

Ki (corrected) = Ki’ (apparent)/(1+(S/Km)) eq.4

where S is the substrate concentration for the run (determined from the fluorescence of completed runs) and Km is the Km for that substrate under the experimental conditions.

The next level of fitting involves the full version of equation 1, the complete progress curve equation, as described above. The actual Graphit input was as follows:

P=vs*(t)+((((vo-vs)*(1-((Ki+E+(I)-(((Ki+E+(I))^2)-(4*E*I))^(1/2))/(Ki+E+(I)+(((Ki+E+(I))^2)-(4*E*I))^(1/2)))))/(((Ki+E+(I)-(((Ki+E+(I))^2)-(4*E*I))^(1/2))/(Ki+E+(I)+(((Ki+E+(I))^2)-(4*E*I))^(1/2)))*k))*log((1-((Ki+E+(I)-(((Ki+E+(I))^2)-(4*E*I))^(1/2))/(Ki+E+(I)+(((Ki+E+(I))^2)-(4*E*I))^(1/2)))*exp(-k*(t)))/(1-((Ki+E+(I)-(((Ki+E+(I))^2)-(4*E*I))^(1/2))/(Ki+E+(I)+(((Ki+E+(I))^2)-(4*E*I))^(1/2))))))+h eq.5