Adsorption of Plasma Proteins on Hydrophobic Surfaces. 111. Serum, Plasma, and Blood R. D. BAGNALL, Bioengineering and Medical Physics Unit, University of Liverpool, Liverpool, England
Summary Liquid-air and liquid-liquid interfaces were used as models for the liquid-solid system of plasma proteins and hydrophobic surfaces in the study of adsorption of serum, plasma, and blood onto these surfaces. The interfacial tension is determined for three phases: air, methylene iodide, and isooctane. Curves of interfacial tension versus time for the various systems are given from which a triple-intersection point, where the protein solution is in equilibrium with each surface, is found. It is shown that albumin, y-globulin, and a mixed solution of these a t in uiuo concentrations behave in characteristic and constant manners a t the three interfaces of air, methylene iodide, and isooctane. A range of synthetic surfaces which have constant behavior a t equilibrium is deduced and it is concluded that any soft tissue response differences between such surfaces could not be the result of albumin or 7-globulin.
INTRODUCTION Previous papers in this ~ e r i e s l -have ~ described the use of air, methylene iodide, and isooctane as models for studying the adsorption of plasma proteins on hydrophobic surfaces, and in particular for determining the extent to which certain plasma proteins differentiate between surfaces during the adsorption process. From this work, it would appear that in the case of serum albumin and y-globulin the rate of adsorption is surface dependent, but the extent of adsorption a t equilibrium is constant for a wide range of hydrophobic surfaces.2 A means was described for determining the nature of the constant protein surface at equilibrium, and for deducing the range of synthetic surfaces for which constant adsorption a t equilibrium would be expected.2 The relevance of these observations to tissue-material interactions was then discussed. Fibrinogen, on the other hand, was found by this novel technique to develop a coherent film of presumably polymeric material a t the Journal of Biomedical Materials Research, Vol. 12,707-721 (1978) oo21-9304/78/ool2-0707~ol.o~ 01978 John Wiley & Sons, Inc.
BAGNALL
708
adsorbing interface, the extent of which was apparently surface dependent even a t equilibrium.3 The particular relevance of this to nonthrombogenic surfaces was discussed. Other workers have also used albumin, y-globulin, and fibrinogen as models for the interaction of plasma proteins with synthetic surfaces (eg. refs. 4-6), but there have been few reports of the use of genuine tissue fluids, presumably because of the complexity of the solutions involved. Notable exceptions are the work of Vroman et al. on the ellipsometry of proteins adsorbing from plasma and s e ~ u m ,and ~ , ~the recent work of Kim e t al. on radiolabeled protein studies at the blood-polymer i n t e r f a ~ e . ~ In view of the success of the novel technique described in this series of papers, and the simplicity of the procedure, I have therefore extended my current studies to include serum, plasma, and whole blood, in the hope of deducing some general property of one or more of these systems.
BACKGROUND The technique described previously relies on the use of the liquid-air and liquid-liquid interfaces as models for the liquid-solid system, the validity of which has already been discussed.2 The interfacial tension a t equilibrium is given by the Fowkes equation as modified by Owens and Wendt,lo i.e., 712
= 71 + 7 2 - %riirzd)l/2 -Z(Y:Yw
(1)
where y1 is the surface tension of phase 1and 7: and 7: are its dispersive and dipolar components, respectively, so that 71 = 712
= rld
r4' + r:
+ r: + Y2 - 2(rfrzd)1'2- 2(Y,Y2) h h 1/2
(2) (3)
If y12?y$,and are known, eq. (3) can then be solved for all satisfying combinations of yfand y:, these being represented graphically as a curve of yld vs 7:. The novel technique selects three different phases ! y and for each phase a means is found to estimate 7 1 2 of known y$,, with a given protein solution a t equilibrium. Equation (3) is solved in each case, and if all three curves intersect a t one point it is concluded that the protein solution presented the same equilibrium surface to each, the protein surface being described by yf and 7: a t the triple-intersection point. In this paper the three phases are air,
ADSORPTION OF PLASMA PROTEINS. 111
709
methylene iodide, and isooctane, the surface tension components of which have been given previously.2 The method of choice for measuring ylz is the pendant-drop technique for interfacial tension determinations.ll
MATERIALS Fresh frozen pooled human serum; fresh frozen ACD human plasma group A+; and fresh citrated whole blood were provided by the Regional Blood Transfusion Service, Liverpool. Prior to use, samples of serum and plasma were warmed to 37°C for 15 min. 2,2,4-Trimethylpentane (isooctane-spectroscopicgrade) and methylene iodide were obtained from BDH Chemicals Ltd., England.
METHOD Prior to use, all glassware was soaked for 15 min in 50:50 sulphuric acidlnitric acid mixture, washed rigorously with deionized water, rinsed with acetone, and dried with Millipore-filtered compressed air. Interfacial tension determinations were then performed over 1hr a t 37 f 0.1”C in a pendant-drop device described elsewhere12using (1) saturated water vapor atmosphere, (2) isooctane, and (3) methylene iodide as the second phase.
RESULTS Interfacial Tension Measurements Curves of interfacial tension vs time for the various systems studied are shown in Figures 1-3, according to Table I.
“Intersection Point” Analysis of Results The interfacial changes involving serum and plasma were analyzed by eq. ( 3 ) ,the various solution curves being shown in Figures 4 and 5, respectively, according to the following expressions.
Serum isooctane: 20.4 = -yt
+ yt + 19.6 - 2(19.6y$)”2
BAGNALL 50
48
\ AIR
4f
4r
2:
2.
2
1
1
1
1
10
2'0
30
40
g0
6b
-lo
T MlNS Fig. 1. Interfacial tension of human serum with air, isooctane, and methylene iodide a t 37°C as a function of time.
ADSORPTION OF PLASMA PROTEINS. 111
711
47
45
AIR
43
41
39
27
24
‘H2 ‘2
2;
2c
10
20
50
4b
Q
60
fo
TWNS Fig. 2. Interfacial tension of human plasma with air, isooctane, and methylene iodide at 37OC as a function of time.
BAGNALL
10
$0
3b
4b
& I
dl
33
TMINS Fig. 3. Interfacial tension of human blood with isooctane at 37OC as a function of time.
TABLE I Phase 1
Phase 2
Serum Serum Serum Plasma Plasma Plasma Blood
air isooctane methylene iodide air isooctane methylene iodide isooctane
CH212: 14.1 = 7t
+ r,d+ 50.3 - 2(48.47,d)'/' - 2(1.97;)'12
air:
43.6 = yt
+yt
Plasma isooctane:
+ + 19.6 - 2(19.6~:)'/~
23.2 = 7; 7:
Figure 1 1 1 2 2 2 3
ADSORPTION OF PLASMA PROTEINS. I11
713
BAGNALL
714
Fig. 5. Solution curves from eq. (3) for human plasma, demonstrating approximate triple-intersection point.
CH212: 19.2 = $
+ r,"+ 50.3 - 2 ( 4 8 . 4 ~ $ )-~ 2/ ~( 1 . 9 ~ ! ) ~ / ~
38.8 = 7;
+ 7;
air:
The triple-intersection point for each set of curves is shown in Table 11.
DISCUSSION Currently there is much interest in the adsorpti0.n of plasma proteins on biomaterials as one aspect of the relative thrombogenicity of different materials, and there is now evidence to suggest that some surfaces exhibit a degree of selectivity when presented with mixed protein s y ~ t e m s . ~In J ~particular, fibrinogen has been shown to be
ADSORPTION OF PLASMA PROTEINS. 111
715
TABLE I1 Triple-Point Coordinates y d (dynedcm)
System
y (dynedcm)
Serum Plasma
20.2 f 0.3 22.7 f 1.0
23.5 f 0.3 16.9 f 1.0
preferentially adsorbed from its mixed solution with albumin and y-globulin? and a possible connection with platelet adhesion phenomena has been proposed.14 A criticism of these studies is that in many cases they are allowed to proceed to equilibrium (-1 hr), whereas thrombogenesis is a rapid phenomena and presumably occurs under nonequilibrium conditions. On the other hand, I have previously argued that the tissue response to a soft-tissue implant may be sufficiently slow to allow the achievement of true interfacial equilibrium through one or more layers of adsorbed macromolecules,l and I have suggested that this might explain the similar minimal soft-tissue responses observed with pure, stable implants. I have then proposed that differences in softtissue response may only be possible if protein desorption can occur to reveal the underlying implant surface during wound repair, and that this may depend on some critical hydrophiliclhydrophobic balance of the absorbing surface. I have then used the pendant-drop technique to investigate the reversible work of adhesion between proteins and model surfaces: and have shown that albumin, y-globulin, and their mixed solution at in uiuo concentrations behave in characteristic, constant manners at the three interfaces of air, isooctane, and methylene iodide. From these results, I deduced the range of synthetic surfaces for which such constant behavior would be expected, and concluded that any softtissue response differences between such surfaces could not be due to either albumin or y-globulin.
Serum Serum is, of course, not simply a mixture of albumin and y-globulin, but contains many other solutes including other proteins, notably other globulins. It is therefore both pleasing and surprising that Fig. 4 demonstrates a perfect triple point for the systems studied, strongly suggesting that serum presents an essentially similar protein layer
716
BAGNALL
to each of the three interfaces at equilibrium. The properties of this constant surface are then shown in Table 11. Previously, I have shown that the equivalent triple point for albumin/?-globulin mixture can be predicted from the triple points of each protein and their relative concentrations.2 A curious feature of serum, then, is that albumin represents approximately 63%of the total protein, but if all other globulins behave as y-globulin, then according to my previously published figures for the pure proteins serum behaves as though it were only -35% albumin (34%by y d, 36%by y h ) . I have also previously shown that film compression by drop retraction leads to a film-collapse profile which is typical for each protein. Compression of the serum interfacial film at the isooctane interface produced a collapse pattern which was typical of a y-globulin-dominated film, consistent with the serum triple point found in this paper. Quantitative results for these and other compression studies are being prepared for publication in this journal (R. D. Bagnall, P. A. Arundel, and J. A. D. Annis). The present study would then suggest that an in uitro mixture of albumin and y-globulin may not be a good model for serum (and presumably plasma). A conclusion is either that some globulins or other serum components are particularly surface active, or that there may be a factor in serum which interferes with the adsorption of albumin. It is interesting to note that Vroman et al. have separately suggested the latter from ellipsometric studies of protein adsorption onto solid surfaces from both plasma and serum.8
Plasma Previously, I have shown that pure fibrinogen does not lead to a triple-intersection point such as that shown in Figure 4, but that a fibrinogenlalbuminly-globulinmixture at in uiuo concentrations more closely approaches such a triple point at equilibrium than does pure fibrin~gen.~ I have then presented evidence to suggest that this is due to a surface-dependentlateral association of adsorbed fibrinogen molecules, and that such association is impeded by the presence of other adsorbing species. By comparison with my previous data, Fig. 5 approaches even more closely to a triple-intersection point, consistent with my previous hypothesis since there are many more competing species in plasma than in a simple three-protein mixture.
ADSORPTION OF PLASMA PROTEINS. I11
717
Although the three intersection points in Figure 5 are just outside the experimental error required for a true triple point2 (f1.5%),they have been approximated to the triple-intersection point shown in Table I1 for comparison with the triple point for serum. If the systems studied in this paper are good models for the solid-liquid interface, Figure 5 would then suggest that there may be slight interfacial differences a t equilibrium between some biomaterial surfaces when exposed to plasma. One possible difference might then be the degree of lateral association of fibrinogen at the i n t e r f a ~ e . ~ A curious feature of Table I1 is that plasma would appear to present a significantly different surface at each of the three interfaces than does serum, particularly with regard to a much reduced y d. Although there may be several explanations for this, I have previously hypothesized3 that the unfolding of adsorbed protein molecules, particularly fibrinogen, can create a secondary surface pressure which will compress any intervening adsorbed species. Such compression would have much greater effect on y d than y h in a close-packed monolayer, consistent with my previous hypothesis.
Whole Blood Figure 3 shows the change in interfacial tension with time for whole blood against isooctane. Unlike all other systems studied so far,2,3 whole blood showed no tendency to reach interfacial equilibrium, an initially rapid decrease in interfacial tensionbeing followed by a slow linear decrease as shown. Several explanations may be possible, one of which is that the pendant-drop technique assumes constant density whereas erythrocytes will settle during the period of any one experiment. The corresponding change in density gradient will give the erroneous appearance of a slow interfacial tension decrease. A t the same time, hemolysis would release hemoglobin, which Horbett et al. have recently shown to be adsorbed preferentially onto polyethylene from its mixed solution with fibrinogen, albumin, or y-g10bulin.l~ Slow displacement of a primary adsorbed protein layer with the more surface-active hemoglobin would also account for the observed effect. In any event, because of the problem of erythrocyte settling in static systems, the study was not extended to air and methylene iodide. Interestingly, a t the completion of the experiment, the interfacial region was compressed by drop retraction and exhibited a collapse pattern unlike either serum or plasma, quantitative results for which
718
BAGNALL
are being prepared for publication. Whether this is due to formed elements per se at the interface, or to some other macromolecules not present in plasma, is not clear. These results, however, suggest that it might eventually be difficult to model the in uiuo environment by any simple solution of mixed proteins.
Significance of the Results for Biomaterials The triple point of Figure 4 for serum refers to the state of the interface a t equilibrium. If the same analysis is applied to earlier adsorption periods, the separate intersection points of the three solution curves move further apart as the time decreases, suggesting strongly that although the protein surfaces presented to the three chosen interfaces are identical at 1 hr, the rate of achieving this state is surface dependent. This is consistent with my previous findings for albumin and y-globulin at the same interfaces.2 A possible consequence is that protein interactions with hydrophobic surfaces may have much more relevance to blood compatibility than to any soft-tissue response differences between biomaterials. It follows from the demonstration of a triple point for serum and an approximated triple point for plasma that the interfacial free energy between the aqueous medium and the bound protein layer must be zero at equilibrium, according to an argument given previously for albumin and y-globulin.2 If the models used in this paper are typical of the solid-liquid intqface, it then also follows that interfacial free energy considerations may be largely irrelevant in soft-tissue implants after 1 hr of implantation. A means is currently being pursued in this laboratory to demonstrate this directly. From the triple point for serum in Figure 4 and Table 11, it is possible to deduce the range of other surfaces for which such constant behavior at equilibrium would also be expected, according to the following argument. It is assumed that the equilibrium interfacial tension of serum at each interface is a minimum for that interface, so that any protein surface properties leading to a lower interfacial tension by eq. (3) are not possible. If for any second phase of properties y,d and y,h there cannot be found a set of allowed protein surface properties which also lead to a lower interfacial tension [by eq. (3)] than that given by the triple-point coordinates for the protein surface, the triple-point coordinates represent a minimum interfacial tension and must operate
ADSORPTION OF PLASMA PROTEINS. 111
719
35
25
I
z
u.w Ln
z
15
5
C
2b N
40
d
60
4
1( DYNES.CM-’ Fig. 6. Method for determining limits of applicability of the triple-intersection point for serum. Area OKMN represents surfaces for which the triple point applies.
for that surface. Conversely, if there can be found a set of such allowed protein surface properties, the nature of the protein layer presented to that surface becomes uncertain. The solution to this problem for any given triple point has been described in detail elsewhere,Z and is shown for serum in Figure 6, where (1) y,h, r,drefer to any second phase s; (2) dyh/dyd refers to disallowed serum surface properties by eq. (3); for air at the triple point in Figure 4; ( 3 ) -1 is (dyhldyd)serum
720
BAGNALL 26
22.7-----
20.2----.
l!
2 Z
>
a
=A I
, 76.9
--
I
23.5
OKMN represents all surfaces for which adsorption from serum is expected to be constant, and area OK‘MN refers to plasma (see text). Key: a, polyethylene; b, poly(viny1 chloride); c, poly(viny1idene chloride); d, poly(viny1 fluoride); e, poly(vinylidene fluoride); f, poly(trifluoroethy1ene); g, poly(tetrafluoroethy1ene);h, poly(ethylene terephthalate); k, poly(methy1 methacrylate); m, polystyrene; n, n-hexatriacontane; p, paraffin; q, fluorinated poly(methacry1icacid); r, vinylidene chloride: acrylonitrile copolymer (8020); s, poly(dimethy1siloxane); t,u, Nylon 66.
for CH& at the triple point in Figure (4) 0.628 is (dyh/dyd)semm 4; (5) yd’ is a disallowed serum surface property which holds for all r,herurn; and (6) “graphical” refers to a graphical procedure for deducing uncertain combinations of r!, yt, as described previously.2 The area OKMN then represents all surfaces to which it is predicted that serum should present an essentially similar adsorbed protein layer at equilibrium,and this area is reproduced in Fig. 7. The point M then represents methylene iodide; and the point K, a surface having identical properties to the serum triple point. For comparison pur-
ADSORPTION OF PLASMA PROTEINS. 111
721
poses, the equivalent area OK’MN’ for plasma is also shown, according to the following conditions: (1) ( d y h l d y d ) p l a s m a < -1 (2) (dyh/dyd)plasma > 0.974 (3) y$=,, > 123.6 (4) graphical procedure
The significance for biomaterials is shown by the superimposition on Figure 7 of points representing some common hydrophobic polymers.1° If the work reported in this paper is a good model for the early events involving soft-tissue implants, it is then not surprising that many of these materials exhibit similar, minimal soft-tissue responses. If unusual soft-tissue responses’can be elicited by certain as-yetunknown materials, it seems likely that these materials will be found outside the areas OKMN and OK’MN’. This may help to explain, for instance, the unusual carcinogenicity of implanted basic ionogenic polymersl5 and the variable soft-tissue responses of certain hydrogels.16 The author wishes to thank Imperial Chemical Industries Ltd. for the financial support for this work.
References 1. 2. 3. 4. 5. 6. 7. 8.
R. D. Bagnall, J. Biomed. Muter. Res. 11,939 (1977). R. D. Bagnall, J. Biomed. Muter. Res. 11,947 (1977). R. D. Bagnall, J. Biomed. Muter. Res. 12,203 (1978). W. J. Dillman and 1. F. Miller, J. Colloid Interface Sci., 44,221 (1973). J. L. Brash and D. J. Lyman, J. Biomed. Muter. Res., 3,175 (1969). R. G. Lee, C. Adamson, and S. W. Kim, Thromb. Res., 4,485 (1974). L. Vroman and A. L. Adams, J. Biomed. Muter. Res., 3,43 (1969). L. Vroman, A. L. Adams, M. Klings, and G. Fischer, Adu. Chem., Ser. No. 145,255
(1975). 9. S. W. Kim, S. Wisniewski, E. S. Lee, and M. L. Winn, J. Biomed. Muter. Res., Symp. 8,23 (1977). 10. D. K. Owens and R. C. Wendt, J. Appl. Polym. Sci., 13,1741 (1969). 11. J. M. Andreas, E. A. Hauser, and W. B. Tucker, J. Phys. Chem., 42, 1001 (1938). 12. R. D. Bagnall, Euling Reuiew, 2 , l l (1978). 13. T. A. Horbett, P. K. Weathersby, and A. S. Hoffman, J. Bioeng., I, 61 (1977). 14. S. W. Kim, Bioengineering Seminar on Artificial Organs, University of Strathclyde, Glusgow, Scotland, 18-20 August (1976). 15. R. I. Leininger, CRC Crit. Reu. Bioeng., I, 333 (1972). 16. J. J. Rosen, D. F. Gibbons, and L. A. Culp, Polym. P r e p . , 16,555 (1975).
Received July 18,1977
Summary Liquid-air and liquid-liquid interfaces were used as models for the liquid-solid system of plasma proteins and hydrophobic surfaces in the study of adsorption of serum, plasma, and blood onto these surfaces. The interfacial tension is determined for three phases: air, methylene iodide, and isooctane. Curves of interfacial tension versus time for the various systems are given from which a triple-intersection point, where the protein solution is in equilibrium with each surface, is found. It is shown that albumin, y-globulin, and a mixed solution of these a t in uiuo concentrations behave in characteristic and constant manners a t the three interfaces of air, methylene iodide, and isooctane. A range of synthetic surfaces which have constant behavior a t equilibrium is deduced and it is concluded that any soft tissue response differences between such surfaces could not be the result of albumin or 7-globulin.
INTRODUCTION Previous papers in this ~ e r i e s l -have ~ described the use of air, methylene iodide, and isooctane as models for studying the adsorption of plasma proteins on hydrophobic surfaces, and in particular for determining the extent to which certain plasma proteins differentiate between surfaces during the adsorption process. From this work, it would appear that in the case of serum albumin and y-globulin the rate of adsorption is surface dependent, but the extent of adsorption a t equilibrium is constant for a wide range of hydrophobic surfaces.2 A means was described for determining the nature of the constant protein surface at equilibrium, and for deducing the range of synthetic surfaces for which constant adsorption a t equilibrium would be expected.2 The relevance of these observations to tissue-material interactions was then discussed. Fibrinogen, on the other hand, was found by this novel technique to develop a coherent film of presumably polymeric material a t the Journal of Biomedical Materials Research, Vol. 12,707-721 (1978) oo21-9304/78/ool2-0707~ol.o~ 01978 John Wiley & Sons, Inc.
BAGNALL
708
adsorbing interface, the extent of which was apparently surface dependent even a t equilibrium.3 The particular relevance of this to nonthrombogenic surfaces was discussed. Other workers have also used albumin, y-globulin, and fibrinogen as models for the interaction of plasma proteins with synthetic surfaces (eg. refs. 4-6), but there have been few reports of the use of genuine tissue fluids, presumably because of the complexity of the solutions involved. Notable exceptions are the work of Vroman et al. on the ellipsometry of proteins adsorbing from plasma and s e ~ u m ,and ~ , ~the recent work of Kim e t al. on radiolabeled protein studies at the blood-polymer i n t e r f a ~ e . ~ In view of the success of the novel technique described in this series of papers, and the simplicity of the procedure, I have therefore extended my current studies to include serum, plasma, and whole blood, in the hope of deducing some general property of one or more of these systems.
BACKGROUND The technique described previously relies on the use of the liquid-air and liquid-liquid interfaces as models for the liquid-solid system, the validity of which has already been discussed.2 The interfacial tension a t equilibrium is given by the Fowkes equation as modified by Owens and Wendt,lo i.e., 712
= 71 + 7 2 - %riirzd)l/2 -Z(Y:Yw
(1)
where y1 is the surface tension of phase 1and 7: and 7: are its dispersive and dipolar components, respectively, so that 71 = 712
= rld
r4' + r:
+ r: + Y2 - 2(rfrzd)1'2- 2(Y,Y2) h h 1/2
(2) (3)
If y12?y$,and are known, eq. (3) can then be solved for all satisfying combinations of yfand y:, these being represented graphically as a curve of yld vs 7:. The novel technique selects three different phases ! y and for each phase a means is found to estimate 7 1 2 of known y$,, with a given protein solution a t equilibrium. Equation (3) is solved in each case, and if all three curves intersect a t one point it is concluded that the protein solution presented the same equilibrium surface to each, the protein surface being described by yf and 7: a t the triple-intersection point. In this paper the three phases are air,
ADSORPTION OF PLASMA PROTEINS. 111
709
methylene iodide, and isooctane, the surface tension components of which have been given previously.2 The method of choice for measuring ylz is the pendant-drop technique for interfacial tension determinations.ll
MATERIALS Fresh frozen pooled human serum; fresh frozen ACD human plasma group A+; and fresh citrated whole blood were provided by the Regional Blood Transfusion Service, Liverpool. Prior to use, samples of serum and plasma were warmed to 37°C for 15 min. 2,2,4-Trimethylpentane (isooctane-spectroscopicgrade) and methylene iodide were obtained from BDH Chemicals Ltd., England.
METHOD Prior to use, all glassware was soaked for 15 min in 50:50 sulphuric acidlnitric acid mixture, washed rigorously with deionized water, rinsed with acetone, and dried with Millipore-filtered compressed air. Interfacial tension determinations were then performed over 1hr a t 37 f 0.1”C in a pendant-drop device described elsewhere12using (1) saturated water vapor atmosphere, (2) isooctane, and (3) methylene iodide as the second phase.
RESULTS Interfacial Tension Measurements Curves of interfacial tension vs time for the various systems studied are shown in Figures 1-3, according to Table I.
“Intersection Point” Analysis of Results The interfacial changes involving serum and plasma were analyzed by eq. ( 3 ) ,the various solution curves being shown in Figures 4 and 5, respectively, according to the following expressions.
Serum isooctane: 20.4 = -yt
+ yt + 19.6 - 2(19.6y$)”2
BAGNALL 50
48
\ AIR
4f
4r
2:
2.
2
1
1
1
1
10
2'0
30
40
g0
6b
-lo
T MlNS Fig. 1. Interfacial tension of human serum with air, isooctane, and methylene iodide a t 37°C as a function of time.
ADSORPTION OF PLASMA PROTEINS. 111
711
47
45
AIR
43
41
39
27
24
‘H2 ‘2
2;
2c
10
20
50
4b
Q
60
fo
TWNS Fig. 2. Interfacial tension of human plasma with air, isooctane, and methylene iodide at 37OC as a function of time.
BAGNALL
10
$0
3b
4b
& I
dl
33
TMINS Fig. 3. Interfacial tension of human blood with isooctane at 37OC as a function of time.
TABLE I Phase 1
Phase 2
Serum Serum Serum Plasma Plasma Plasma Blood
air isooctane methylene iodide air isooctane methylene iodide isooctane
CH212: 14.1 = 7t
+ r,d+ 50.3 - 2(48.47,d)'/' - 2(1.97;)'12
air:
43.6 = yt
+yt
Plasma isooctane:
+ + 19.6 - 2(19.6~:)'/~
23.2 = 7; 7:
Figure 1 1 1 2 2 2 3
ADSORPTION OF PLASMA PROTEINS. I11
713
BAGNALL
714
Fig. 5. Solution curves from eq. (3) for human plasma, demonstrating approximate triple-intersection point.
CH212: 19.2 = $
+ r,"+ 50.3 - 2 ( 4 8 . 4 ~ $ )-~ 2/ ~( 1 . 9 ~ ! ) ~ / ~
38.8 = 7;
+ 7;
air:
The triple-intersection point for each set of curves is shown in Table 11.
DISCUSSION Currently there is much interest in the adsorpti0.n of plasma proteins on biomaterials as one aspect of the relative thrombogenicity of different materials, and there is now evidence to suggest that some surfaces exhibit a degree of selectivity when presented with mixed protein s y ~ t e m s . ~In J ~particular, fibrinogen has been shown to be
ADSORPTION OF PLASMA PROTEINS. 111
715
TABLE I1 Triple-Point Coordinates y d (dynedcm)
System
y (dynedcm)
Serum Plasma
20.2 f 0.3 22.7 f 1.0
23.5 f 0.3 16.9 f 1.0
preferentially adsorbed from its mixed solution with albumin and y-globulin? and a possible connection with platelet adhesion phenomena has been proposed.14 A criticism of these studies is that in many cases they are allowed to proceed to equilibrium (-1 hr), whereas thrombogenesis is a rapid phenomena and presumably occurs under nonequilibrium conditions. On the other hand, I have previously argued that the tissue response to a soft-tissue implant may be sufficiently slow to allow the achievement of true interfacial equilibrium through one or more layers of adsorbed macromolecules,l and I have suggested that this might explain the similar minimal soft-tissue responses observed with pure, stable implants. I have then proposed that differences in softtissue response may only be possible if protein desorption can occur to reveal the underlying implant surface during wound repair, and that this may depend on some critical hydrophiliclhydrophobic balance of the absorbing surface. I have then used the pendant-drop technique to investigate the reversible work of adhesion between proteins and model surfaces: and have shown that albumin, y-globulin, and their mixed solution at in uiuo concentrations behave in characteristic, constant manners at the three interfaces of air, isooctane, and methylene iodide. From these results, I deduced the range of synthetic surfaces for which such constant behavior would be expected, and concluded that any softtissue response differences between such surfaces could not be due to either albumin or y-globulin.
Serum Serum is, of course, not simply a mixture of albumin and y-globulin, but contains many other solutes including other proteins, notably other globulins. It is therefore both pleasing and surprising that Fig. 4 demonstrates a perfect triple point for the systems studied, strongly suggesting that serum presents an essentially similar protein layer
716
BAGNALL
to each of the three interfaces at equilibrium. The properties of this constant surface are then shown in Table 11. Previously, I have shown that the equivalent triple point for albumin/?-globulin mixture can be predicted from the triple points of each protein and their relative concentrations.2 A curious feature of serum, then, is that albumin represents approximately 63%of the total protein, but if all other globulins behave as y-globulin, then according to my previously published figures for the pure proteins serum behaves as though it were only -35% albumin (34%by y d, 36%by y h ) . I have also previously shown that film compression by drop retraction leads to a film-collapse profile which is typical for each protein. Compression of the serum interfacial film at the isooctane interface produced a collapse pattern which was typical of a y-globulin-dominated film, consistent with the serum triple point found in this paper. Quantitative results for these and other compression studies are being prepared for publication in this journal (R. D. Bagnall, P. A. Arundel, and J. A. D. Annis). The present study would then suggest that an in uitro mixture of albumin and y-globulin may not be a good model for serum (and presumably plasma). A conclusion is either that some globulins or other serum components are particularly surface active, or that there may be a factor in serum which interferes with the adsorption of albumin. It is interesting to note that Vroman et al. have separately suggested the latter from ellipsometric studies of protein adsorption onto solid surfaces from both plasma and serum.8
Plasma Previously, I have shown that pure fibrinogen does not lead to a triple-intersection point such as that shown in Figure 4, but that a fibrinogenlalbuminly-globulinmixture at in uiuo concentrations more closely approaches such a triple point at equilibrium than does pure fibrin~gen.~ I have then presented evidence to suggest that this is due to a surface-dependentlateral association of adsorbed fibrinogen molecules, and that such association is impeded by the presence of other adsorbing species. By comparison with my previous data, Fig. 5 approaches even more closely to a triple-intersection point, consistent with my previous hypothesis since there are many more competing species in plasma than in a simple three-protein mixture.
ADSORPTION OF PLASMA PROTEINS. I11
717
Although the three intersection points in Figure 5 are just outside the experimental error required for a true triple point2 (f1.5%),they have been approximated to the triple-intersection point shown in Table I1 for comparison with the triple point for serum. If the systems studied in this paper are good models for the solid-liquid interface, Figure 5 would then suggest that there may be slight interfacial differences a t equilibrium between some biomaterial surfaces when exposed to plasma. One possible difference might then be the degree of lateral association of fibrinogen at the i n t e r f a ~ e . ~ A curious feature of Table I1 is that plasma would appear to present a significantly different surface at each of the three interfaces than does serum, particularly with regard to a much reduced y d. Although there may be several explanations for this, I have previously hypothesized3 that the unfolding of adsorbed protein molecules, particularly fibrinogen, can create a secondary surface pressure which will compress any intervening adsorbed species. Such compression would have much greater effect on y d than y h in a close-packed monolayer, consistent with my previous hypothesis.
Whole Blood Figure 3 shows the change in interfacial tension with time for whole blood against isooctane. Unlike all other systems studied so far,2,3 whole blood showed no tendency to reach interfacial equilibrium, an initially rapid decrease in interfacial tensionbeing followed by a slow linear decrease as shown. Several explanations may be possible, one of which is that the pendant-drop technique assumes constant density whereas erythrocytes will settle during the period of any one experiment. The corresponding change in density gradient will give the erroneous appearance of a slow interfacial tension decrease. A t the same time, hemolysis would release hemoglobin, which Horbett et al. have recently shown to be adsorbed preferentially onto polyethylene from its mixed solution with fibrinogen, albumin, or y-g10bulin.l~ Slow displacement of a primary adsorbed protein layer with the more surface-active hemoglobin would also account for the observed effect. In any event, because of the problem of erythrocyte settling in static systems, the study was not extended to air and methylene iodide. Interestingly, a t the completion of the experiment, the interfacial region was compressed by drop retraction and exhibited a collapse pattern unlike either serum or plasma, quantitative results for which
718
BAGNALL
are being prepared for publication. Whether this is due to formed elements per se at the interface, or to some other macromolecules not present in plasma, is not clear. These results, however, suggest that it might eventually be difficult to model the in uiuo environment by any simple solution of mixed proteins.
Significance of the Results for Biomaterials The triple point of Figure 4 for serum refers to the state of the interface a t equilibrium. If the same analysis is applied to earlier adsorption periods, the separate intersection points of the three solution curves move further apart as the time decreases, suggesting strongly that although the protein surfaces presented to the three chosen interfaces are identical at 1 hr, the rate of achieving this state is surface dependent. This is consistent with my previous findings for albumin and y-globulin at the same interfaces.2 A possible consequence is that protein interactions with hydrophobic surfaces may have much more relevance to blood compatibility than to any soft-tissue response differences between biomaterials. It follows from the demonstration of a triple point for serum and an approximated triple point for plasma that the interfacial free energy between the aqueous medium and the bound protein layer must be zero at equilibrium, according to an argument given previously for albumin and y-globulin.2 If the models used in this paper are typical of the solid-liquid intqface, it then also follows that interfacial free energy considerations may be largely irrelevant in soft-tissue implants after 1 hr of implantation. A means is currently being pursued in this laboratory to demonstrate this directly. From the triple point for serum in Figure 4 and Table 11, it is possible to deduce the range of other surfaces for which such constant behavior at equilibrium would also be expected, according to the following argument. It is assumed that the equilibrium interfacial tension of serum at each interface is a minimum for that interface, so that any protein surface properties leading to a lower interfacial tension by eq. (3) are not possible. If for any second phase of properties y,d and y,h there cannot be found a set of allowed protein surface properties which also lead to a lower interfacial tension [by eq. (3)] than that given by the triple-point coordinates for the protein surface, the triple-point coordinates represent a minimum interfacial tension and must operate
ADSORPTION OF PLASMA PROTEINS. 111
719
35
25
I
z
u.w Ln
z
15
5
C
2b N
40
d
60
4
1( DYNES.CM-’ Fig. 6. Method for determining limits of applicability of the triple-intersection point for serum. Area OKMN represents surfaces for which the triple point applies.
for that surface. Conversely, if there can be found a set of such allowed protein surface properties, the nature of the protein layer presented to that surface becomes uncertain. The solution to this problem for any given triple point has been described in detail elsewhere,Z and is shown for serum in Figure 6, where (1) y,h, r,drefer to any second phase s; (2) dyh/dyd refers to disallowed serum surface properties by eq. (3); for air at the triple point in Figure 4; ( 3 ) -1 is (dyhldyd)serum
720
BAGNALL 26
22.7-----
20.2----.
l!
2 Z
>
a
=A I
, 76.9
--
I
23.5
OKMN represents all surfaces for which adsorption from serum is expected to be constant, and area OK‘MN refers to plasma (see text). Key: a, polyethylene; b, poly(viny1 chloride); c, poly(viny1idene chloride); d, poly(viny1 fluoride); e, poly(vinylidene fluoride); f, poly(trifluoroethy1ene); g, poly(tetrafluoroethy1ene);h, poly(ethylene terephthalate); k, poly(methy1 methacrylate); m, polystyrene; n, n-hexatriacontane; p, paraffin; q, fluorinated poly(methacry1icacid); r, vinylidene chloride: acrylonitrile copolymer (8020); s, poly(dimethy1siloxane); t,u, Nylon 66.
for CH& at the triple point in Figure (4) 0.628 is (dyh/dyd)semm 4; (5) yd’ is a disallowed serum surface property which holds for all r,herurn; and (6) “graphical” refers to a graphical procedure for deducing uncertain combinations of r!, yt, as described previously.2 The area OKMN then represents all surfaces to which it is predicted that serum should present an essentially similar adsorbed protein layer at equilibrium,and this area is reproduced in Fig. 7. The point M then represents methylene iodide; and the point K, a surface having identical properties to the serum triple point. For comparison pur-
ADSORPTION OF PLASMA PROTEINS. 111
721
poses, the equivalent area OK’MN’ for plasma is also shown, according to the following conditions: (1) ( d y h l d y d ) p l a s m a < -1 (2) (dyh/dyd)plasma > 0.974 (3) y$=,, > 123.6 (4) graphical procedure
The significance for biomaterials is shown by the superimposition on Figure 7 of points representing some common hydrophobic polymers.1° If the work reported in this paper is a good model for the early events involving soft-tissue implants, it is then not surprising that many of these materials exhibit similar, minimal soft-tissue responses. If unusual soft-tissue responses’can be elicited by certain as-yetunknown materials, it seems likely that these materials will be found outside the areas OKMN and OK’MN’. This may help to explain, for instance, the unusual carcinogenicity of implanted basic ionogenic polymersl5 and the variable soft-tissue responses of certain hydrogels.16 The author wishes to thank Imperial Chemical Industries Ltd. for the financial support for this work.
References 1. 2. 3. 4. 5. 6. 7. 8.
R. D. Bagnall, J. Biomed. Muter. Res. 11,939 (1977). R. D. Bagnall, J. Biomed. Muter. Res. 11,947 (1977). R. D. Bagnall, J. Biomed. Muter. Res. 12,203 (1978). W. J. Dillman and 1. F. Miller, J. Colloid Interface Sci., 44,221 (1973). J. L. Brash and D. J. Lyman, J. Biomed. Muter. Res., 3,175 (1969). R. G. Lee, C. Adamson, and S. W. Kim, Thromb. Res., 4,485 (1974). L. Vroman and A. L. Adams, J. Biomed. Muter. Res., 3,43 (1969). L. Vroman, A. L. Adams, M. Klings, and G. Fischer, Adu. Chem., Ser. No. 145,255
(1975). 9. S. W. Kim, S. Wisniewski, E. S. Lee, and M. L. Winn, J. Biomed. Muter. Res., Symp. 8,23 (1977). 10. D. K. Owens and R. C. Wendt, J. Appl. Polym. Sci., 13,1741 (1969). 11. J. M. Andreas, E. A. Hauser, and W. B. Tucker, J. Phys. Chem., 42, 1001 (1938). 12. R. D. Bagnall, Euling Reuiew, 2 , l l (1978). 13. T. A. Horbett, P. K. Weathersby, and A. S. Hoffman, J. Bioeng., I, 61 (1977). 14. S. W. Kim, Bioengineering Seminar on Artificial Organs, University of Strathclyde, Glusgow, Scotland, 18-20 August (1976). 15. R. I. Leininger, CRC Crit. Reu. Bioeng., I, 333 (1972). 16. J. J. Rosen, D. F. Gibbons, and L. A. Culp, Polym. P r e p . , 16,555 (1975).
Received July 18,1977
