J. BIOMEI). MATElI. RE%
VOL. 9, PP. 511-536 (1975)
The Interactions of Thrombin with Borosilicate Glass Surfaces DAVID F. WAUGH, L I S D A J. ASTHOSY, and HELES SG,* Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts
Summary Borosilicate glass (G) and commercial poly(methy1 methacrylate) (PMM) surfaces were examined a t pH 7 , monovalent ionic strength 0.14 and 22°C. Thrombin concentrations ranged from 6 x 10-9 to 5.2 X 10-8M. Decreases in thrombin concentration in 10 ml vessels were det,ermined by clotting t,ime assay. Adsorption to PMM vessels is small. These were examined to establish the validity of the method. For G-vessels, the terminations of initial rapid decreases, associated with adsorption equilibrium, are accounted for mainly by an average isotherm having an association constant of 1.8 X lo7 l./mol. Adaorption equilibrium is accompanied by a slower rate of decrease consistent with the surface denaturation of 2.5yo/min of adsorbed molecules. Molecules remaining adsorbed to the surfaces of 1 mm i.d. G-capillaries were examined using the sequence of thrombin adsorption, buffer washing for times t,, removal of buffer and introduction of a fibrinogen aliquot for determination of effective surface thrombin concentration by a capillary clotting time. Most adsorbed molecules ( 8%) desorb in t, 30 min, and members of additional classes are still present after t, = 1440 min. For all of the properties examined, there is a small variance between vessels taken a t the same time from the same batch, and a larger variance between groups of vessels taken a t different times from the same batch or from different batches.
-
INTRODUCTION I n recent years there has been a steadily increasing interest in devices whose surfaces will contact blood. As a result of the frequency with which thrombi appear subsequent to contact, attention has
* Present address: University of Hong Kong, Faculty of Medicine, Lady Ho Tung Hall, Pokfulam Road, Hong Kong. 51 1 @ 1975 by John Wiley & Sons, Inc.
512
WAU(;H, ANTHONY, ANI) N(i
been directed toward the types of contributory surface interaction. Among these is the type which involves macromolecules in solution, and evidence is available to show that such precede and may mediate surface interaction with formed e l e m e n t ~ . l - ~The nell known complexity of the problem of accounting for macromolecular interaction at the solution-solid interface, and of relating interaction to thrombosis, is pointed out in recent r e v i e n ~ . ~ -A~ further expectation is that, contrary to uncomplicated systems in which solution components are in simple competition with each other for surface sites, surface cooperativity in interaction contributes critically to the initiation (nucleation) of a thrombus.6 As a step toward studies of competitive and cooperative surface interaction it was felt that information more extensive than currently available should first be obtained regarding the surface behavior of some single component which might be of importance in thrombus development. Thrombin was chosen as a consequence of its unique position: it plays a crucial role in blood coagulation not only in the conversion of fibrinogen to monomer fibrin, but also as a result of its interaction with other solution coagulation components and its effects on platelets. For studies of adsorption it has additional highly desirable attributes. The clotting time assay for thrombin concentration has both high sensitivity and discrimination, and thrombin is available in essentially pure and stable form. There are two areas regarding molecule-surface interaction which should eventually be specified in some detail. First, that regarding the forces and energies which are responsible for interaction, and secondly that regarding the effects of interaction on the properties of adsorbed molecules. The results reported here are concerned with the latter. Studies have been made of the interaction of thrombin with a number of plastic surfaces, but particularly with poly(methy1 methacrylate) (PMM). The latter results from the fact that thrombin adsorption to commercial P M M is small, and P M M is easily fabricated into vessels to contain dilute thrombin. Certain plastic surfaces appear to have special properties, with respect to paucity of surface sites and cooperativity of surface interaction. As a n aid in understanding these effects it was felt that studies of a surface having structural detail at a level small compared to the thrombin molecule (radius -22 A) might be informative. Borosilicate glass was chosen partly on this basis and partly on the basis that it is a “high energy”
A1)SOItPTION OF THItOMBIN TO GLASS SURFACES
513
surface and as a consequence would have a high probability of displaying the full complement of surface effects. If so, the glass surface would provide a reference with which other surfaces could be compared. Finally, phenomena which appear when blood is allowed t o pass over the surfaces of glass beads are now used as a means for investigating platelet reactivity.' Characterization of glass surfaces appeared desirable also from the standpoint of this application. Our purpose a t this time is to provide information regarding the extent, homogeneity and reversibility of adsorption of thrombin to glass surfaces, and to examine the responses of adsorbed molecules, such as their tendency toward surface denaturation. MATERIALS AND METHODS All chemicals were reagent grade. Two types of distilled water were used. For the first, the general supply of distilled water was passed through a mixed bed ion exchange resin and then through an HA grade Millipore filter. The second type, used for all buffers and critical washing, was obtained by redistilling the first from permanganate in Pyrex containers. Distilled water was stored in polyethylene containers. Absorbance.-A Beckman DU spectrophotometer and a 1 cm path length were used. One absorbance unit of a substance is that amount per ml which gives unit absorbance a t pH 7, 1 cm path and X = 280 nm. Standard environmental conditions.-So far as possible, stock solutions of thrombin and fibrinogen were diluted t o final conditions of pH 7, 0.028 A4 sodium barbital and 0.122 M as the sum of sodium plus potassium chlorides. Final solutions contained mainly sodium chloride. Total ionic strength was 0.142. Reservoirs and adsorption vessels were at room temperature (22OC). Clotting time assays were carried out a t 29°C. Syringe calibrations.-Syringes were calibrated gravimetrically t o deliver within 0.1% of the stated volume. Thrombin.-The method of Baughman and Waugh as modified by Rosenberg and Waugh9 was used to obtain gel filtered thrombin (Gel-T) from fresh bovine plasma. Gel-T preparations contain 0.2 to 0.7 absorbance units/ml (330 t o 1170 N.I.H. units/ml) in an environment containing 1.0 ionic stIength (0.1 due to sodium phosphate
514
WAUGH, ANTHONY, ANI) Nf;
and the remainder to sodium chloride). The high degree of purity and reproducibility of single animal Gel-T have been examined elsewhere,'" where the equivalence of one S.I.H. unit of thrombin to absorbance unit is obtained. Using an absorbanve of 1.97 6 x at 1 rng/ml," and a molecular weight of 36,000,*a thrombin concentration of 1 S.I.H. unit per ml corresponds to a concentration of S.64 ill. FibYinogen.-Stock fibrinogens were prepared, each from the plasma of a single animal. Essentially, a number of clotting vomponents are first removed from plasma anticoagulated with sodium oxalate by treatment with BaS04. A cold ethanol precipitate is obtained by a modification of the procedure of Cohn et a1.I2l 3 This is brought into solution and fractionated by ammonium sulfate precipitation according to the procedure of La1~i.I~The details of the preparative procedure, which must be carefully controlled, will be published elsewhere. Stock fibrinogen is at a concentration of about 20 mg/ml in 0.3 M KCl a t pH 7. Preparations contained above 9.5% clottable protein determined by absorbance. A lot of stock fibrinogen is subdivided into -3 ml aliquots in polycarbonate tubes and these are quick frozen in liquid nitrogen and stored a t -90°C. A r-tube will refer to a clear plastic cuvet and fibrinogen content in which a clotting time is to be determined. To establish a set of .r-tubes, aliquots of stock fibrinogen are thawed a t 29"C, air bubbles removed by centrifugation and an aliquot delivered by pipet to a volume of buffer designed t o yield standard environmental conditions and a concentration of 1.26 mg/ml. The resulting solution is incubated for 16 hr a t 29OC, subdivided into 0.5 ml aliquots in r-tubes by an instrument which holds constant all interface which contacts fibrinogen solution. After filling, r-tubes are held a t 15°C. A 7-tube is returned to 29°C 10 min before it is to be used for a clotting time assay. Details of the procedure will be published elsewhere. Calculations.-S.D. is the standard deviation. Where a range is specified, the numbers indicate the 95% confidence limits. Where linear relations are given, they were calculated by linear least squares approximation. S , is the coefficient of variance, the S.D./mean.
Measurement of Thrombin Adsorption by Solution Depletion Clotting times were used to assay for thrombin concentration. A thrombin reservoir provided a known thrombin concentration, which
ADSORPTION O F THROMBIN TO GLASS SURFACES
515
was delivered into a series of small adsorption vessels. Depletion in each of the latter was determined by the difference in concentration between reservoir and vessel. The operation of the procedure is indicated in Figure 1. Reservoir.-In Figure 1, A represents a thrombin reservoir constructed of PMM and having a bottom drain. Initially the drain is empty and closed off a t the inside of the reservoir by a stainless steel stake having an O-ring a t its end. The reservoir, of internal radius 3.75 cm, can accommodate a volume up to 800 ml. The actual volumes used ranged from about 260 to 560 ml, depending on the convenience of establishing accurately the desired thrombin concentration. To establish a thrombin concentration, the closed off reservoir is first filled with standard buffer and a small volume of distilled water so that, after the addition of Gel-T, standard conditions are
- Fig. 1. Measurement of adsorption by solution depletion. (A) represents a large thrombin reservoir, its bottom drain, and a stake with an O-ring a t its end which is used to close off the exit on the inside of the chamber. The bottom drain is used either to fill the side arm thrombin delivery syringe (B) or a set of adsorption vessels (C). After filling, the 10 ml contents of each adsorption vessel is sampled a t a different time by siphoning into a second matched side arm delivery syringe. The thrombin delivery syringe is used to inject a thrombin aliquot into a 7-tube containing fibrinogen for a clotting time determination. Average clotting time is converted into thrombin concentration.
516
WAUGH, ANTHONY, AND NG
met (except for the small amount of phosphate which is introduced with the Gel-T). Gel-T is measured b y calibrated syringe. The contents of the reservoir are mixed slowly for 10 min by a teflon coated magnetic stirring bar and thereafter remain quiescent during a n experiment. After a thrombin concentration is established, the drain (.066 in. i.d. x .095 in. 0.d. polyethylene tubing, Intramedic) is filled with solution by removing the stake and it is then closed a t its open end by a stainless steel plug. After a 1 hr incubation period the drain is briefly purged and is then used to fill either a side arm syringe, or a set of adsorption vessels. The initial thrombin volume is chosen so that, a t termination of a protocol, less than 50y0is removed. Reservoirs are washed after their fabrication with detergent and alkali as described below. Then, after each use, reservoirs are emptied, rinsed with distilled water and allowed to air dry a t room temperature. Side arm syringe.-The syringe is indicated schematically by B in Figure 1. The thrombin delivery syringe, first described as a device to be filled by piercing the surface of the thrombin ~ o l u t i o n , 'was ~ modified t o provide entry into the cavity of the syringe through a hole in the side of the barrel.I6 At this time, the glass plunger was grooved so that its rotation acted as a valve, permitting control of filling and emptying (see Fig. 2 in ref. 12). More recently the barrel has been ground to a more constant internal diameter and glass replaced by 6, closely fitting stainless steel plunger. This plunger has at its side a groove 0.020 in. wide, 0.010 in. deep and just sufficiently long t o accommodate the side arm hole to the total excursion of the plunger. This permits filling without introduction of a n air bubble. The groove leads into a n axial hole which exits a t the end of the plunger. At its other end the plunger threads into a stainless steel disc whose excursion on emptying is limited by the aluminum block to which the barrel is cemented. The barrel of the syringe is cut off t o allow the disc to contact the block. The excursion of the disc on filling is limited by a calibrating stop. A major advantage of the side arm syringe is that it avoids surface piercing, and thus a n attendant alteration in source thrombin concentration. A second advantage is equilibration. When attached to the reservoir, for example, the thrombin in the syringe can be equilibrated with that in the reservoir by a through flow of -0.5 ml. It is noted that when the syringe solenoid is activated, back pressure
ADSORPTION OF THROMBIN TO GLASS SURFACES
517
will tend t o drive thrombin a t equilibrated concentration into connecting spaces, such as that between barrel and plunger. Adsorption vessels.-These are indicated by C in Figure 1. Two types will be examined here: those made from commercial P M M (mainly Lucite) and those of borosilicate glass, G. Vessel dimensions were chosen so that a volume of 10 ml was accommodated. For P M M vessels Lucite tubing, 0.625 in. i.d. X 0.750 in. o.d., was cut into lengths of 2.5 in. One end was machined and t o this was attached a 0.125 in. thick base, using Lucite dissolved in ethylene dichloride as a cement. Tubes were machined t o uniform 0.d. so that their ends fitted into a stirring device. Borosilicate glass test tubes 0.7 in. 0.d. were cut to be 3 in. long. Each vessel was marked a t the position where it contained 10 ml. P M M vessels were washed in Lakeseal detergent in distilled water a t room temperature. They were carefully rinsed and allowed t o soak in distilled water for -3 hr. A set of 36 t o 100 tubes was then rapidly immersed in 6 1. of 1.0 M NaOH brought t o 75°C and the system was gently stirred and allowed to cool for 1 hr (to -55°C). Vessels were then removed and drained, rinsed thoroughly with distilled water, and allowed to air dry. As will be shown elsewhere this treatment so far has been found most effective in reducing day-to-day variance with P M M vessels. Borosilicate glass vessels were washed in the same manner. For a determination of adsorption, a vessel was filled to the 10 ml level by using the reservoir bottom drain held so that a minimum air-solution interface was encountered by the thrombin solution. Thereafter, mixing to facilitate contact between thrombin in solution and the vessel wall was accomplished by vortex free vessel rotation. A single stirring occupied 1 min and consisted of the following. At 10 see intervals a vessel was subjected t o a single rotation of 8 revolutions in 1 sec. The sequence was alternately clockwise and counterclockwise, and vessels were stationary for 9 see between each rotation. Vessels were stirred just after filling, a t 30 and 60 min after filling, and in any case just before assay. Clotting time assays were initially carried out by surface piercing, then by siphoning into a side arm syringe. Usually, a reservoir and 8 adsorption vessels were involved in one experiment. Customarily, a reservoir was examined a t the beginning, the middle and the end of a n experiment. For each examination, 16 clotting times were determined. Each adsorption vessel was
518
WAUGH, ANTHONY, AND NG
examined once by carrying out 8 clotting times. Averages and variances were calculated. Vessel characterjstics were determined from average clotting time, 7,as will be described below. Clotting time assay.-The end point was determined using a BioDynamics Model CA550 coagulation analyzer modified so that the contents of the 7-tube are automatically mixed by magnetic stirrer for 3 sec just after thrombin addition. For assay, 0.1 ml of a solution containing thrombin is added by thrombin delivery syringe to 0.5 ml of fibrinogen solution held a t 29°C. The Bio-Dynamics instrument measures the interval between the addition of thrombin and the moment when the time differential of increasing turbidity reaches a preselected value. This is the clotting time, 7 , which is determined t o the nearest 0.1 see. Before each clotting time determination, a t least 0.5 ml of the thrombin solution to be examined is allowed t o flow through the thrombin delivery syringe and equilibrate i t with the thrombin source. Frequently, the reservoir and adsorption vessels differ significantly in thrombin concentration. To eliminate possible cross contamination, side arm syringes matched by gravimetric and clotting time assay were used: one for the reservoir and one for each set of adsorption vessels. The reference system-The reservoir and set of r-tubes constitute the reference system with which adsorption vessels and the same set of tubes are to be compared. Within the reference system, contributions to clotting time regression can come from botJh the reservoir and the set of r-tubes. Thrombin can disappear in the reservoir, for example, through interaction with the surface or through spontaneous solution denaturation. I n connection with reservoir losses the following are noted. Previous studiess have shown that at 'i = 20 sec the maximum clotting time decay is about 2% of 7 per day in polyethylene containers. The current loss could be smaller, since present reservoirs have a lower surface t o volume ratio (0.65 cm-l) than those used previously and they are sampled through a bottom drain rather than by surface piercing. These previous studies have shown also that concentrated thrombin is stable for long periods of time a t 4°C and that in a large reservoir a thrombin concentration can be reproduced to within 0.5% from day t o day. As a result of thrombin stability compared to fibrinogen, the reservoir concentration of thrombin, based on the concentration of thrombin in Gel-T
ADSORPTION O F THROMBIN TO GLASS SURFACES
519
and a dilution factor, was used as the standard by which the days assay characteristics were established. The instability of dilute fibrinogen is well known.l5,l6 For a large thrombin reservoir and a single set of r-tubes, previous resultss,l0 and present reference system response (an example of which is indicated as R in Fig. 2) show that during an experiment the clotting time-thrombin concentration relationship can be expressed as log ( r
+ at) = y log Th + C
(1)
where r is the clotting time, Th is the thrombin concentration, t is laboratory time, and a, y and C t o a first approximation are constants. Of the constants, y appears to be an intrinsic characteristics of clotting systems in that it has a small variance between sets of r-tubes, and is relatively independent of the source of fibrinogen. For fibrinogens prepared from Armour Fraction-I, and using a visual compaction endpoint for determining clotting time, 7 = -0.679 when T h is in N.I.H. units/ml and r is in sec. The S , of y is 1.9%.lo The turbidimetric endpoint now employed yields 7 = -0.719, and this value is used throughout. With respect to C , when a thrombin concentration giving a n average 7 of 30 sec is reproduced, the variance in F between lots of fibrinogen, each prepared from the plasma of a single animal, corresponds t o a n S , of just less than a%, and a similar value is obtained when sets of r-tubes from the same lot of stock fibrinogen are compared with each other. The procedure for stabilizing r-tubes, described under Fibrinogen, is designed to make a as small as possible, and a n average a could be used. However, the determination of Q for each experiment establishes the satisfactory operation of and permits the individual use of each reference system. When two known thrombin concentrations are compared using one set of r-tubes, elimination of C in eq. (1) gives
Thi Thz
y log-
=
+ +
51 ad log ___ 72 azt
and since the left side is a constant for all t, including t = 0, a1 - -a2 _ -
11
52
(3)
520
WAUGH, ANTHONY, AND NG
14C
cn c
12c
T)
0
0 W
cn
z
.c .-Z I O O -0u
G
c
R
80 200
:
II
t
,t
G
PMM
300
I
I
400
500
t i m e , min Fig. 2. Results of a single solution depletion experiment. Zero time on the abscissa corresponds to 8:40 a.m. laboratory time. The reservoir, R, was established a t 50 min to contain 389 ml of thrombin a t 1 N.I.H. unit/ml. The set of 7-tubes was filled with fibrinogen between 95 and 146 min. The arrows on the abscissa a t G and PMM indicate the times when 10 ml glass or poly(methy1 methacrylate) adsorption vessels, respectively, were filled. The ordinate gives average clotting times and 95% confidence limits of the mean. Regressions are calculated as linear. I n this experiment the thrombin reservoir was examined twice, usually 3 examinations were made.
For any one experiment the regression slope of a reference system is predicted to be proportional to its clotting time. The total number of clotting times required to estimate (Y and 7 for a reservoir, and to determine T for each of a set of adsorption vessels is an important factor in placing an upper limit on the clotting time, thus a lower limit on thrombin concentration, which can be handled during a single experiment. Clotting times between 20 and 300 sec have been used. I n the clotting assay these correspond to 1 and 0.0225 N.I.H. unit thrombin/ml. The sources from which
ADSORPTION O F THROMBIN TO GLASS SURFACES
521
aliquots are taken are 6 times as concentrated. Thus the range of thrombin concentration examined in the reservoirs and adsorption vessels is from 5.18 X lo-* to 1.17 X M.
Examination of Adsorbed Molecules For this purpose we have used Pyrex capillaries of 1 mm. i.d. Lengths of 5 cm were cut and these were cleaned in hot chromic acid cleaning solution, then carefully rinsed in distilled water and air dried. Clean capillaries were used according t o the following procedure. Thrombin solution was introduced into a horizontal capillary a t one end (end A) for a distance of -3.0 cm. After standing for 30 min the capillary was oriented vertically (end A down) and inserted gently into a tube from which buffer was flowing a t a rate of 9 ml/min for washing times less than 30 min and at a rate of -3 ml/mm for longer washing times. After a desired buffer washing the capillary was removed and while vertical was drained of buffer, but not dried. Placed zgain in a horizontal position an aliquot of fibrinogen solution was introduced, through end B, so as just to fill the capillary. Fibrinogen solution thus first passed over a section of the capillary which had not contacted thrombin. A capillary clotting time was taken as the time when the demarkation between untreated and thrombin treated sections of the capillary become visible, with indirect illumination, as a result of the development of turbidity in the treated section. It is well known that, for a solution containing thrombin and fibrinogen, turbidity measured by apparent absorbance remains close to that of the fibrinogen solution until, over a comparatively short time period, it abruptly increases. Turbidity in the capillary has a similar dependence. The capillary clotting time in a suitably washed capillary is taken to reflect the surface concentration of active thrombin, according to a relationship similar t o that of eq. (1).
RESULTS Solution depletion.-Figure 2 gives results obtained in a n experiment where normalized regressions, slope/?, are close t o representing group averages. Average clotting times for reservoir and glass and poly(methy1 methacrylate) adsorption vessels are plotted on the ordinate against laboratory time. I n each case dotting time increases
522
WAUGH, ANTHONY, AND NG
with laboratory time, thus thrombin concentration decreases. The slopes of the clotting time regressions will be indicated as a,SC and SPMM, respectively. The reference system has a = 0.011 sec/min. The time of filling and the succession of T for glass vessels show that there is initially a high rate of increase of clotting time, followed by a slower rate of increase. I n some experiments the rapid initial increase was verified. The slower increase gives SG = 0.21 sec/min. PMM-vessels are similar to glass vessels except that the extent of the rapid increase is less, as is SPMM,.whichis 0.022 (about twice the (Y of the reference system). I n what follows the rapid disappearance of thrombin is associated with approach to adsorption equilibrium, and slower disappearance with progressive surface denaturation. Two aspects of preliminary results became evident. The first is that (as indicated in Fig. 2 ) for any one experiment the adsorption vessels, which come a t the same time from a single batch, appear t o be responding relatively uniformly, as judged by their positions with respect to their lines of regression. For vessels taken a t different times, or from batch-to-batch, adsorption and slow disappearance both have a noticeable variance. This will become evident from what follows. The second aspect is that as the thrombin concentration is increased, the error in clotting time determination makes it increasingly difficult first t o determine SG and SPMM', and a t higher thrombin concentrations the extent of adsorption. Both effects are expected since all values eventually are determined from small differences between large thrombin concentrations. Eighteen final experiments were carried out at initial reservoir thrombin concentrations ranging from 6.0 to 0.7 N.I.H. units/ml (from 5.2 X to 6 X lO-OM). Equilibrium adsorption from solution depletion.-That most of the adsorption t o glass can be described b y a simple average adsorption isotherm is the conclusion of what follows. For a simple isotherm
where 6 is the fraction of maximum (saturation) adsorption, K is a n association constant and The is the equilibrium thrombin concentration. An important problem is introduced by slow thrombin disappearance, which requires a n evaluation of the time when the experi-
ADSORPTION OF THROMBIN TO GLASS SURFACES
,523
mentally determined ambient thrombin concentration, Th, is equal to The. From Th,, 0 is calculated using
e=
(Tho - T h J V A . Us
-
X
(5)
where Tho is the total initial thrombin concentration, V is the adsorption vessel volume (10 ml), A is the solution-wall interfacial area (taken as the apparent area of 27 em2), Us is the surface concentration of thrombin at saturation and X is the amount of thrombin adsorbed a t the solution-air interface. Hexagonal close packing of spherical thrombin molecules a t the solution-solid interface yields a molecular area of -1700 A 2 . Us would then be 5.9 X 10l2mol/cm2 or 1.13 N.I.H. units thrombin/cm2. By eq. ( 5 ) , 0 is sensitive t o the product A . Us. Choice of a n obvious minimum in A and maximum in Us seemed most appropriate. I n adsorption vessels having the geometry used here, the solution-air interface is 2 cm2and adsorption a t this interface according to previous results is about 0.2 N.I.H. units/cm2 (9 X 10" mol/cm2),6'16thus a total of 0.4 N.I.H. units. Adsorption t o the 27 cm2 of solution-glass interface is always in excess of 12 times this value. Since the concentration dependence of adsorption a t the air-solution interface is not known, X = 0.4 N.I.H. units (or its equivalent) was used as a constant. If X = 0.2 is used instead of 0.4, 0 values are increased uniformly by 0.006. Regression in the reference system was examined and taken into consideration as follows. According to eq. ( 3 ) , such regression leads ?~ to slopes at two constant thrombin concentrations related by ( Y ~ / = C Y ~ / T ~If. regression in the reference system were t o account completely for SPMM and S G ,reservoir and vessel regression slopes should be correlated through this relationship. The average ratio for reservoirs is always small, 1.42 f 1.38 X low4min-I (range). Measured S P ~and M SG and the ? a t the midpoints of regressions were used to calculate ratios for a preliminary examination of adsorption vessels. I n all 5 experiments, the ratio for PMM vessels is greater than the a/? of the reference reservoir, and is 3.1 f 3.8 X min-1 (range). I n all 18 experiments, the ratio for glass vessels, much greater than min-' (range). the (Y/T for each partner reservoir, is 1.9 f 1.4 X Evidently, reference system regressions do not account for slow decay in adsorption vessels and for each a progressive thrombin loss
524
WAUGH, ANTHONY, AND NG
(denaturation) is indicated. It is noted that thrombin loss alters the total concentration of active thrombin, Tho, in eq. (5). Progressive thrombin loss in adsorption vessels is related to a residual clotting time regression, X P M M ( , . ) or S G ( ~ =) ( d ~ l d t residual, ) where, for example,
~ )X X 'ivesse~ sec/min. This includes Average S P M ~is( 1.3 effects which have their origins in manipulations to which the reservoir is not subjected, such as filling and stirring, as well as possible denaturation at the PMM surface. The fact that SPMM(~) is small and less than & that for glass suggests that denaturation resulting from the presence of the glass surface is by far the most important effect contributing to progressive thrombin loss. Returning to eqs. (4)and ( 5 ) , it was evident that S G ( ~should ) be back extrapolated to a time such that measured Th corresponds to The. Extrapolation to vessel filling time would neglect the expectation that surface effects occur mainly after thrombin adsorbs. To ~ ) back extrapolated to an arbiapproach this problem, each S G (was trary time of 60 min after vessel filling, and 0 calculated from the observed T h using eq. (5). I n Figure 3, 0 / ( l - 0) is plotted against T h for this time. As seen, there is a small positive intercept a t Th = 0. The slope of the regression of Figure 3 has a value of K = 2 X lo7l./mol, which corresponds to an adsorption free energy close to -10 kcal/mol. Back extrapolation of to less than 60 min produces an increase in Th and this in turn leads to a decrease in both the intercept and slope in Figure 3. At 30 min after vessel filling, the intercept is -0 and the slope is 1.8 x lo7 l./mol, which also is close to -10 kcal/mol. Capillary studies, to be considered below, show that some thrombin molecules are much more strongly bound than average. This observation suggests that the relationship of Figure 3, based on average behavior, should have a small positive intercept. An intermediate value of K = 1.9 X lo7 l./mol has been selected. When concentration is expressed in N.I.H. units/ ml, K = 0.18 ml/unit. The scatter evident in Figure 3 is, on the basis of reference system stability and the small values of S p M M ( , . ) , to be attributed t o variance
ADSORPTION OF THROMBIN TO GLASS SURFACES
,
I
2
I
3
4
525
1 5
Th, 10' M
Fig. 3. Adsorption to borosilicate glass vessels. The abscissa records molar ambient thrombin concentration 60 min after filling adsorption vessels. The ordinate records O / ( l - e), where e is the fraction of the nominal surface area calculated to be covered by adsorbed thrombin. The line, obtained by linear least squares approximation, has the equation e/(l - e) = 2.05 X 10' (Th) 0.04. Scatter is attributed largely to variance between groups of adsorption vessels.
+
between groups of glass adsorption vessels. It is within this variance that adsorption can be described by an average isotherm. Surface inactivation.-Equation (6) yields Sacr) which are considered as follows. For a residual regression, as clotting time increases the thrombin concentration in the assay clotting mixture decreases according to log 7 = y log Th C
+
which on differentiation yields, with y
dTH ~-
at
-
-1.4
=
Th 7
-0.719,
SG(~)
(7)
Multiplying by 60/27 converts this solution rate of decrease to a corresponding rate of decrease a t the vessel surface, R , in N.I.H. units/cm2.min. The specific rate is then
526
WAUGH, ANTHONY, AND NG
Specific rates were calculated for two thrombin-clotting time pairs, the first 60 rnin after vessel filling and the second a t the midpoints of regressions, thus a t about 115 min after vessel filling. Experiments a t high thrombin concentrations, where residual regressions become obscured as a result of short clotting times (
VOL. 9, PP. 511-536 (1975)
The Interactions of Thrombin with Borosilicate Glass Surfaces DAVID F. WAUGH, L I S D A J. ASTHOSY, and HELES SG,* Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts
Summary Borosilicate glass (G) and commercial poly(methy1 methacrylate) (PMM) surfaces were examined a t pH 7 , monovalent ionic strength 0.14 and 22°C. Thrombin concentrations ranged from 6 x 10-9 to 5.2 X 10-8M. Decreases in thrombin concentration in 10 ml vessels were det,ermined by clotting t,ime assay. Adsorption to PMM vessels is small. These were examined to establish the validity of the method. For G-vessels, the terminations of initial rapid decreases, associated with adsorption equilibrium, are accounted for mainly by an average isotherm having an association constant of 1.8 X lo7 l./mol. Adaorption equilibrium is accompanied by a slower rate of decrease consistent with the surface denaturation of 2.5yo/min of adsorbed molecules. Molecules remaining adsorbed to the surfaces of 1 mm i.d. G-capillaries were examined using the sequence of thrombin adsorption, buffer washing for times t,, removal of buffer and introduction of a fibrinogen aliquot for determination of effective surface thrombin concentration by a capillary clotting time. Most adsorbed molecules ( 8%) desorb in t, 30 min, and members of additional classes are still present after t, = 1440 min. For all of the properties examined, there is a small variance between vessels taken a t the same time from the same batch, and a larger variance between groups of vessels taken a t different times from the same batch or from different batches.
-
INTRODUCTION I n recent years there has been a steadily increasing interest in devices whose surfaces will contact blood. As a result of the frequency with which thrombi appear subsequent to contact, attention has
* Present address: University of Hong Kong, Faculty of Medicine, Lady Ho Tung Hall, Pokfulam Road, Hong Kong. 51 1 @ 1975 by John Wiley & Sons, Inc.
512
WAU(;H, ANTHONY, ANI) N(i
been directed toward the types of contributory surface interaction. Among these is the type which involves macromolecules in solution, and evidence is available to show that such precede and may mediate surface interaction with formed e l e m e n t ~ . l - ~The nell known complexity of the problem of accounting for macromolecular interaction at the solution-solid interface, and of relating interaction to thrombosis, is pointed out in recent r e v i e n ~ . ~ -A~ further expectation is that, contrary to uncomplicated systems in which solution components are in simple competition with each other for surface sites, surface cooperativity in interaction contributes critically to the initiation (nucleation) of a thrombus.6 As a step toward studies of competitive and cooperative surface interaction it was felt that information more extensive than currently available should first be obtained regarding the surface behavior of some single component which might be of importance in thrombus development. Thrombin was chosen as a consequence of its unique position: it plays a crucial role in blood coagulation not only in the conversion of fibrinogen to monomer fibrin, but also as a result of its interaction with other solution coagulation components and its effects on platelets. For studies of adsorption it has additional highly desirable attributes. The clotting time assay for thrombin concentration has both high sensitivity and discrimination, and thrombin is available in essentially pure and stable form. There are two areas regarding molecule-surface interaction which should eventually be specified in some detail. First, that regarding the forces and energies which are responsible for interaction, and secondly that regarding the effects of interaction on the properties of adsorbed molecules. The results reported here are concerned with the latter. Studies have been made of the interaction of thrombin with a number of plastic surfaces, but particularly with poly(methy1 methacrylate) (PMM). The latter results from the fact that thrombin adsorption to commercial P M M is small, and P M M is easily fabricated into vessels to contain dilute thrombin. Certain plastic surfaces appear to have special properties, with respect to paucity of surface sites and cooperativity of surface interaction. As a n aid in understanding these effects it was felt that studies of a surface having structural detail at a level small compared to the thrombin molecule (radius -22 A) might be informative. Borosilicate glass was chosen partly on this basis and partly on the basis that it is a “high energy”
A1)SOItPTION OF THItOMBIN TO GLASS SURFACES
513
surface and as a consequence would have a high probability of displaying the full complement of surface effects. If so, the glass surface would provide a reference with which other surfaces could be compared. Finally, phenomena which appear when blood is allowed t o pass over the surfaces of glass beads are now used as a means for investigating platelet reactivity.' Characterization of glass surfaces appeared desirable also from the standpoint of this application. Our purpose a t this time is to provide information regarding the extent, homogeneity and reversibility of adsorption of thrombin to glass surfaces, and to examine the responses of adsorbed molecules, such as their tendency toward surface denaturation. MATERIALS AND METHODS All chemicals were reagent grade. Two types of distilled water were used. For the first, the general supply of distilled water was passed through a mixed bed ion exchange resin and then through an HA grade Millipore filter. The second type, used for all buffers and critical washing, was obtained by redistilling the first from permanganate in Pyrex containers. Distilled water was stored in polyethylene containers. Absorbance.-A Beckman DU spectrophotometer and a 1 cm path length were used. One absorbance unit of a substance is that amount per ml which gives unit absorbance a t pH 7, 1 cm path and X = 280 nm. Standard environmental conditions.-So far as possible, stock solutions of thrombin and fibrinogen were diluted t o final conditions of pH 7, 0.028 A4 sodium barbital and 0.122 M as the sum of sodium plus potassium chlorides. Final solutions contained mainly sodium chloride. Total ionic strength was 0.142. Reservoirs and adsorption vessels were at room temperature (22OC). Clotting time assays were carried out a t 29°C. Syringe calibrations.-Syringes were calibrated gravimetrically t o deliver within 0.1% of the stated volume. Thrombin.-The method of Baughman and Waugh as modified by Rosenberg and Waugh9 was used to obtain gel filtered thrombin (Gel-T) from fresh bovine plasma. Gel-T preparations contain 0.2 to 0.7 absorbance units/ml (330 t o 1170 N.I.H. units/ml) in an environment containing 1.0 ionic stIength (0.1 due to sodium phosphate
514
WAUGH, ANTHONY, ANI) Nf;
and the remainder to sodium chloride). The high degree of purity and reproducibility of single animal Gel-T have been examined elsewhere,'" where the equivalence of one S.I.H. unit of thrombin to absorbance unit is obtained. Using an absorbanve of 1.97 6 x at 1 rng/ml," and a molecular weight of 36,000,*a thrombin concentration of 1 S.I.H. unit per ml corresponds to a concentration of S.64 ill. FibYinogen.-Stock fibrinogens were prepared, each from the plasma of a single animal. Essentially, a number of clotting vomponents are first removed from plasma anticoagulated with sodium oxalate by treatment with BaS04. A cold ethanol precipitate is obtained by a modification of the procedure of Cohn et a1.I2l 3 This is brought into solution and fractionated by ammonium sulfate precipitation according to the procedure of La1~i.I~The details of the preparative procedure, which must be carefully controlled, will be published elsewhere. Stock fibrinogen is at a concentration of about 20 mg/ml in 0.3 M KCl a t pH 7. Preparations contained above 9.5% clottable protein determined by absorbance. A lot of stock fibrinogen is subdivided into -3 ml aliquots in polycarbonate tubes and these are quick frozen in liquid nitrogen and stored a t -90°C. A r-tube will refer to a clear plastic cuvet and fibrinogen content in which a clotting time is to be determined. To establish a set of .r-tubes, aliquots of stock fibrinogen are thawed a t 29"C, air bubbles removed by centrifugation and an aliquot delivered by pipet to a volume of buffer designed t o yield standard environmental conditions and a concentration of 1.26 mg/ml. The resulting solution is incubated for 16 hr a t 29OC, subdivided into 0.5 ml aliquots in r-tubes by an instrument which holds constant all interface which contacts fibrinogen solution. After filling, r-tubes are held a t 15°C. A 7-tube is returned to 29°C 10 min before it is to be used for a clotting time assay. Details of the procedure will be published elsewhere. Calculations.-S.D. is the standard deviation. Where a range is specified, the numbers indicate the 95% confidence limits. Where linear relations are given, they were calculated by linear least squares approximation. S , is the coefficient of variance, the S.D./mean.
Measurement of Thrombin Adsorption by Solution Depletion Clotting times were used to assay for thrombin concentration. A thrombin reservoir provided a known thrombin concentration, which
ADSORPTION O F THROMBIN TO GLASS SURFACES
515
was delivered into a series of small adsorption vessels. Depletion in each of the latter was determined by the difference in concentration between reservoir and vessel. The operation of the procedure is indicated in Figure 1. Reservoir.-In Figure 1, A represents a thrombin reservoir constructed of PMM and having a bottom drain. Initially the drain is empty and closed off a t the inside of the reservoir by a stainless steel stake having an O-ring a t its end. The reservoir, of internal radius 3.75 cm, can accommodate a volume up to 800 ml. The actual volumes used ranged from about 260 to 560 ml, depending on the convenience of establishing accurately the desired thrombin concentration. To establish a thrombin concentration, the closed off reservoir is first filled with standard buffer and a small volume of distilled water so that, after the addition of Gel-T, standard conditions are
- Fig. 1. Measurement of adsorption by solution depletion. (A) represents a large thrombin reservoir, its bottom drain, and a stake with an O-ring a t its end which is used to close off the exit on the inside of the chamber. The bottom drain is used either to fill the side arm thrombin delivery syringe (B) or a set of adsorption vessels (C). After filling, the 10 ml contents of each adsorption vessel is sampled a t a different time by siphoning into a second matched side arm delivery syringe. The thrombin delivery syringe is used to inject a thrombin aliquot into a 7-tube containing fibrinogen for a clotting time determination. Average clotting time is converted into thrombin concentration.
516
WAUGH, ANTHONY, AND NG
met (except for the small amount of phosphate which is introduced with the Gel-T). Gel-T is measured b y calibrated syringe. The contents of the reservoir are mixed slowly for 10 min by a teflon coated magnetic stirring bar and thereafter remain quiescent during a n experiment. After a thrombin concentration is established, the drain (.066 in. i.d. x .095 in. 0.d. polyethylene tubing, Intramedic) is filled with solution by removing the stake and it is then closed a t its open end by a stainless steel plug. After a 1 hr incubation period the drain is briefly purged and is then used to fill either a side arm syringe, or a set of adsorption vessels. The initial thrombin volume is chosen so that, a t termination of a protocol, less than 50y0is removed. Reservoirs are washed after their fabrication with detergent and alkali as described below. Then, after each use, reservoirs are emptied, rinsed with distilled water and allowed to air dry a t room temperature. Side arm syringe.-The syringe is indicated schematically by B in Figure 1. The thrombin delivery syringe, first described as a device to be filled by piercing the surface of the thrombin ~ o l u t i o n , 'was ~ modified t o provide entry into the cavity of the syringe through a hole in the side of the barrel.I6 At this time, the glass plunger was grooved so that its rotation acted as a valve, permitting control of filling and emptying (see Fig. 2 in ref. 12). More recently the barrel has been ground to a more constant internal diameter and glass replaced by 6, closely fitting stainless steel plunger. This plunger has at its side a groove 0.020 in. wide, 0.010 in. deep and just sufficiently long t o accommodate the side arm hole to the total excursion of the plunger. This permits filling without introduction of a n air bubble. The groove leads into a n axial hole which exits a t the end of the plunger. At its other end the plunger threads into a stainless steel disc whose excursion on emptying is limited by the aluminum block to which the barrel is cemented. The barrel of the syringe is cut off t o allow the disc to contact the block. The excursion of the disc on filling is limited by a calibrating stop. A major advantage of the side arm syringe is that it avoids surface piercing, and thus a n attendant alteration in source thrombin concentration. A second advantage is equilibration. When attached to the reservoir, for example, the thrombin in the syringe can be equilibrated with that in the reservoir by a through flow of -0.5 ml. It is noted that when the syringe solenoid is activated, back pressure
ADSORPTION OF THROMBIN TO GLASS SURFACES
517
will tend t o drive thrombin a t equilibrated concentration into connecting spaces, such as that between barrel and plunger. Adsorption vessels.-These are indicated by C in Figure 1. Two types will be examined here: those made from commercial P M M (mainly Lucite) and those of borosilicate glass, G. Vessel dimensions were chosen so that a volume of 10 ml was accommodated. For P M M vessels Lucite tubing, 0.625 in. i.d. X 0.750 in. o.d., was cut into lengths of 2.5 in. One end was machined and t o this was attached a 0.125 in. thick base, using Lucite dissolved in ethylene dichloride as a cement. Tubes were machined t o uniform 0.d. so that their ends fitted into a stirring device. Borosilicate glass test tubes 0.7 in. 0.d. were cut to be 3 in. long. Each vessel was marked a t the position where it contained 10 ml. P M M vessels were washed in Lakeseal detergent in distilled water a t room temperature. They were carefully rinsed and allowed t o soak in distilled water for -3 hr. A set of 36 t o 100 tubes was then rapidly immersed in 6 1. of 1.0 M NaOH brought t o 75°C and the system was gently stirred and allowed to cool for 1 hr (to -55°C). Vessels were then removed and drained, rinsed thoroughly with distilled water, and allowed to air dry. As will be shown elsewhere this treatment so far has been found most effective in reducing day-to-day variance with P M M vessels. Borosilicate glass vessels were washed in the same manner. For a determination of adsorption, a vessel was filled to the 10 ml level by using the reservoir bottom drain held so that a minimum air-solution interface was encountered by the thrombin solution. Thereafter, mixing to facilitate contact between thrombin in solution and the vessel wall was accomplished by vortex free vessel rotation. A single stirring occupied 1 min and consisted of the following. At 10 see intervals a vessel was subjected t o a single rotation of 8 revolutions in 1 sec. The sequence was alternately clockwise and counterclockwise, and vessels were stationary for 9 see between each rotation. Vessels were stirred just after filling, a t 30 and 60 min after filling, and in any case just before assay. Clotting time assays were initially carried out by surface piercing, then by siphoning into a side arm syringe. Usually, a reservoir and 8 adsorption vessels were involved in one experiment. Customarily, a reservoir was examined a t the beginning, the middle and the end of a n experiment. For each examination, 16 clotting times were determined. Each adsorption vessel was
518
WAUGH, ANTHONY, AND NG
examined once by carrying out 8 clotting times. Averages and variances were calculated. Vessel characterjstics were determined from average clotting time, 7,as will be described below. Clotting time assay.-The end point was determined using a BioDynamics Model CA550 coagulation analyzer modified so that the contents of the 7-tube are automatically mixed by magnetic stirrer for 3 sec just after thrombin addition. For assay, 0.1 ml of a solution containing thrombin is added by thrombin delivery syringe to 0.5 ml of fibrinogen solution held a t 29°C. The Bio-Dynamics instrument measures the interval between the addition of thrombin and the moment when the time differential of increasing turbidity reaches a preselected value. This is the clotting time, 7 , which is determined t o the nearest 0.1 see. Before each clotting time determination, a t least 0.5 ml of the thrombin solution to be examined is allowed t o flow through the thrombin delivery syringe and equilibrate i t with the thrombin source. Frequently, the reservoir and adsorption vessels differ significantly in thrombin concentration. To eliminate possible cross contamination, side arm syringes matched by gravimetric and clotting time assay were used: one for the reservoir and one for each set of adsorption vessels. The reference system-The reservoir and set of r-tubes constitute the reference system with which adsorption vessels and the same set of tubes are to be compared. Within the reference system, contributions to clotting time regression can come from botJh the reservoir and the set of r-tubes. Thrombin can disappear in the reservoir, for example, through interaction with the surface or through spontaneous solution denaturation. I n connection with reservoir losses the following are noted. Previous studiess have shown that at 'i = 20 sec the maximum clotting time decay is about 2% of 7 per day in polyethylene containers. The current loss could be smaller, since present reservoirs have a lower surface t o volume ratio (0.65 cm-l) than those used previously and they are sampled through a bottom drain rather than by surface piercing. These previous studies have shown also that concentrated thrombin is stable for long periods of time a t 4°C and that in a large reservoir a thrombin concentration can be reproduced to within 0.5% from day t o day. As a result of thrombin stability compared to fibrinogen, the reservoir concentration of thrombin, based on the concentration of thrombin in Gel-T
ADSORPTION O F THROMBIN TO GLASS SURFACES
519
and a dilution factor, was used as the standard by which the days assay characteristics were established. The instability of dilute fibrinogen is well known.l5,l6 For a large thrombin reservoir and a single set of r-tubes, previous resultss,l0 and present reference system response (an example of which is indicated as R in Fig. 2) show that during an experiment the clotting time-thrombin concentration relationship can be expressed as log ( r
+ at) = y log Th + C
(1)
where r is the clotting time, Th is the thrombin concentration, t is laboratory time, and a, y and C t o a first approximation are constants. Of the constants, y appears to be an intrinsic characteristics of clotting systems in that it has a small variance between sets of r-tubes, and is relatively independent of the source of fibrinogen. For fibrinogens prepared from Armour Fraction-I, and using a visual compaction endpoint for determining clotting time, 7 = -0.679 when T h is in N.I.H. units/ml and r is in sec. The S , of y is 1.9%.lo The turbidimetric endpoint now employed yields 7 = -0.719, and this value is used throughout. With respect to C , when a thrombin concentration giving a n average 7 of 30 sec is reproduced, the variance in F between lots of fibrinogen, each prepared from the plasma of a single animal, corresponds t o a n S , of just less than a%, and a similar value is obtained when sets of r-tubes from the same lot of stock fibrinogen are compared with each other. The procedure for stabilizing r-tubes, described under Fibrinogen, is designed to make a as small as possible, and a n average a could be used. However, the determination of Q for each experiment establishes the satisfactory operation of and permits the individual use of each reference system. When two known thrombin concentrations are compared using one set of r-tubes, elimination of C in eq. (1) gives
Thi Thz
y log-
=
+ +
51 ad log ___ 72 azt
and since the left side is a constant for all t, including t = 0, a1 - -a2 _ -
11
52
(3)
520
WAUGH, ANTHONY, AND NG
14C
cn c
12c
T)
0
0 W
cn
z
.c .-Z I O O -0u
G
c
R
80 200
:
II
t
,t
G
PMM
300
I
I
400
500
t i m e , min Fig. 2. Results of a single solution depletion experiment. Zero time on the abscissa corresponds to 8:40 a.m. laboratory time. The reservoir, R, was established a t 50 min to contain 389 ml of thrombin a t 1 N.I.H. unit/ml. The set of 7-tubes was filled with fibrinogen between 95 and 146 min. The arrows on the abscissa a t G and PMM indicate the times when 10 ml glass or poly(methy1 methacrylate) adsorption vessels, respectively, were filled. The ordinate gives average clotting times and 95% confidence limits of the mean. Regressions are calculated as linear. I n this experiment the thrombin reservoir was examined twice, usually 3 examinations were made.
For any one experiment the regression slope of a reference system is predicted to be proportional to its clotting time. The total number of clotting times required to estimate (Y and 7 for a reservoir, and to determine T for each of a set of adsorption vessels is an important factor in placing an upper limit on the clotting time, thus a lower limit on thrombin concentration, which can be handled during a single experiment. Clotting times between 20 and 300 sec have been used. I n the clotting assay these correspond to 1 and 0.0225 N.I.H. unit thrombin/ml. The sources from which
ADSORPTION O F THROMBIN TO GLASS SURFACES
521
aliquots are taken are 6 times as concentrated. Thus the range of thrombin concentration examined in the reservoirs and adsorption vessels is from 5.18 X lo-* to 1.17 X M.
Examination of Adsorbed Molecules For this purpose we have used Pyrex capillaries of 1 mm. i.d. Lengths of 5 cm were cut and these were cleaned in hot chromic acid cleaning solution, then carefully rinsed in distilled water and air dried. Clean capillaries were used according t o the following procedure. Thrombin solution was introduced into a horizontal capillary a t one end (end A) for a distance of -3.0 cm. After standing for 30 min the capillary was oriented vertically (end A down) and inserted gently into a tube from which buffer was flowing a t a rate of 9 ml/min for washing times less than 30 min and at a rate of -3 ml/mm for longer washing times. After a desired buffer washing the capillary was removed and while vertical was drained of buffer, but not dried. Placed zgain in a horizontal position an aliquot of fibrinogen solution was introduced, through end B, so as just to fill the capillary. Fibrinogen solution thus first passed over a section of the capillary which had not contacted thrombin. A capillary clotting time was taken as the time when the demarkation between untreated and thrombin treated sections of the capillary become visible, with indirect illumination, as a result of the development of turbidity in the treated section. It is well known that, for a solution containing thrombin and fibrinogen, turbidity measured by apparent absorbance remains close to that of the fibrinogen solution until, over a comparatively short time period, it abruptly increases. Turbidity in the capillary has a similar dependence. The capillary clotting time in a suitably washed capillary is taken to reflect the surface concentration of active thrombin, according to a relationship similar t o that of eq. (1).
RESULTS Solution depletion.-Figure 2 gives results obtained in a n experiment where normalized regressions, slope/?, are close t o representing group averages. Average clotting times for reservoir and glass and poly(methy1 methacrylate) adsorption vessels are plotted on the ordinate against laboratory time. I n each case dotting time increases
522
WAUGH, ANTHONY, AND NG
with laboratory time, thus thrombin concentration decreases. The slopes of the clotting time regressions will be indicated as a,SC and SPMM, respectively. The reference system has a = 0.011 sec/min. The time of filling and the succession of T for glass vessels show that there is initially a high rate of increase of clotting time, followed by a slower rate of increase. I n some experiments the rapid initial increase was verified. The slower increase gives SG = 0.21 sec/min. PMM-vessels are similar to glass vessels except that the extent of the rapid increase is less, as is SPMM,.whichis 0.022 (about twice the (Y of the reference system). I n what follows the rapid disappearance of thrombin is associated with approach to adsorption equilibrium, and slower disappearance with progressive surface denaturation. Two aspects of preliminary results became evident. The first is that (as indicated in Fig. 2 ) for any one experiment the adsorption vessels, which come a t the same time from a single batch, appear t o be responding relatively uniformly, as judged by their positions with respect to their lines of regression. For vessels taken a t different times, or from batch-to-batch, adsorption and slow disappearance both have a noticeable variance. This will become evident from what follows. The second aspect is that as the thrombin concentration is increased, the error in clotting time determination makes it increasingly difficult first t o determine SG and SPMM', and a t higher thrombin concentrations the extent of adsorption. Both effects are expected since all values eventually are determined from small differences between large thrombin concentrations. Eighteen final experiments were carried out at initial reservoir thrombin concentrations ranging from 6.0 to 0.7 N.I.H. units/ml (from 5.2 X to 6 X lO-OM). Equilibrium adsorption from solution depletion.-That most of the adsorption t o glass can be described b y a simple average adsorption isotherm is the conclusion of what follows. For a simple isotherm
where 6 is the fraction of maximum (saturation) adsorption, K is a n association constant and The is the equilibrium thrombin concentration. An important problem is introduced by slow thrombin disappearance, which requires a n evaluation of the time when the experi-
ADSORPTION OF THROMBIN TO GLASS SURFACES
,523
mentally determined ambient thrombin concentration, Th, is equal to The. From Th,, 0 is calculated using
e=
(Tho - T h J V A . Us
-
X
(5)
where Tho is the total initial thrombin concentration, V is the adsorption vessel volume (10 ml), A is the solution-wall interfacial area (taken as the apparent area of 27 em2), Us is the surface concentration of thrombin at saturation and X is the amount of thrombin adsorbed a t the solution-air interface. Hexagonal close packing of spherical thrombin molecules a t the solution-solid interface yields a molecular area of -1700 A 2 . Us would then be 5.9 X 10l2mol/cm2 or 1.13 N.I.H. units thrombin/cm2. By eq. ( 5 ) , 0 is sensitive t o the product A . Us. Choice of a n obvious minimum in A and maximum in Us seemed most appropriate. I n adsorption vessels having the geometry used here, the solution-air interface is 2 cm2and adsorption a t this interface according to previous results is about 0.2 N.I.H. units/cm2 (9 X 10" mol/cm2),6'16thus a total of 0.4 N.I.H. units. Adsorption t o the 27 cm2 of solution-glass interface is always in excess of 12 times this value. Since the concentration dependence of adsorption a t the air-solution interface is not known, X = 0.4 N.I.H. units (or its equivalent) was used as a constant. If X = 0.2 is used instead of 0.4, 0 values are increased uniformly by 0.006. Regression in the reference system was examined and taken into consideration as follows. According to eq. ( 3 ) , such regression leads ?~ to slopes at two constant thrombin concentrations related by ( Y ~ / = C Y ~ / T ~If. regression in the reference system were t o account completely for SPMM and S G ,reservoir and vessel regression slopes should be correlated through this relationship. The average ratio for reservoirs is always small, 1.42 f 1.38 X low4min-I (range). Measured S P ~and M SG and the ? a t the midpoints of regressions were used to calculate ratios for a preliminary examination of adsorption vessels. I n all 5 experiments, the ratio for PMM vessels is greater than the a/? of the reference reservoir, and is 3.1 f 3.8 X min-1 (range). I n all 18 experiments, the ratio for glass vessels, much greater than min-' (range). the (Y/T for each partner reservoir, is 1.9 f 1.4 X Evidently, reference system regressions do not account for slow decay in adsorption vessels and for each a progressive thrombin loss
524
WAUGH, ANTHONY, AND NG
(denaturation) is indicated. It is noted that thrombin loss alters the total concentration of active thrombin, Tho, in eq. (5). Progressive thrombin loss in adsorption vessels is related to a residual clotting time regression, X P M M ( , . ) or S G ( ~ =) ( d ~ l d t residual, ) where, for example,
~ )X X 'ivesse~ sec/min. This includes Average S P M ~is( 1.3 effects which have their origins in manipulations to which the reservoir is not subjected, such as filling and stirring, as well as possible denaturation at the PMM surface. The fact that SPMM(~) is small and less than & that for glass suggests that denaturation resulting from the presence of the glass surface is by far the most important effect contributing to progressive thrombin loss. Returning to eqs. (4)and ( 5 ) , it was evident that S G ( ~should ) be back extrapolated to a time such that measured Th corresponds to The. Extrapolation to vessel filling time would neglect the expectation that surface effects occur mainly after thrombin adsorbs. To ~ ) back extrapolated to an arbiapproach this problem, each S G (was trary time of 60 min after vessel filling, and 0 calculated from the observed T h using eq. (5). I n Figure 3, 0 / ( l - 0) is plotted against T h for this time. As seen, there is a small positive intercept a t Th = 0. The slope of the regression of Figure 3 has a value of K = 2 X lo7l./mol, which corresponds to an adsorption free energy close to -10 kcal/mol. Back extrapolation of to less than 60 min produces an increase in Th and this in turn leads to a decrease in both the intercept and slope in Figure 3. At 30 min after vessel filling, the intercept is -0 and the slope is 1.8 x lo7 l./mol, which also is close to -10 kcal/mol. Capillary studies, to be considered below, show that some thrombin molecules are much more strongly bound than average. This observation suggests that the relationship of Figure 3, based on average behavior, should have a small positive intercept. An intermediate value of K = 1.9 X lo7 l./mol has been selected. When concentration is expressed in N.I.H. units/ ml, K = 0.18 ml/unit. The scatter evident in Figure 3 is, on the basis of reference system stability and the small values of S p M M ( , . ) , to be attributed t o variance
ADSORPTION OF THROMBIN TO GLASS SURFACES
,
I
2
I
3
4
525
1 5
Th, 10' M
Fig. 3. Adsorption to borosilicate glass vessels. The abscissa records molar ambient thrombin concentration 60 min after filling adsorption vessels. The ordinate records O / ( l - e), where e is the fraction of the nominal surface area calculated to be covered by adsorbed thrombin. The line, obtained by linear least squares approximation, has the equation e/(l - e) = 2.05 X 10' (Th) 0.04. Scatter is attributed largely to variance between groups of adsorption vessels.
+
between groups of glass adsorption vessels. It is within this variance that adsorption can be described by an average isotherm. Surface inactivation.-Equation (6) yields Sacr) which are considered as follows. For a residual regression, as clotting time increases the thrombin concentration in the assay clotting mixture decreases according to log 7 = y log Th C
+
which on differentiation yields, with y
dTH ~-
at
-
-1.4
=
Th 7
-0.719,
SG(~)
(7)
Multiplying by 60/27 converts this solution rate of decrease to a corresponding rate of decrease a t the vessel surface, R , in N.I.H. units/cm2.min. The specific rate is then
526
WAUGH, ANTHONY, AND NG
Specific rates were calculated for two thrombin-clotting time pairs, the first 60 rnin after vessel filling and the second a t the midpoints of regressions, thus a t about 115 min after vessel filling. Experiments a t high thrombin concentrations, where residual regressions become obscured as a result of short clotting times (
