J. BIOMED. MATER. RES.
VOL. 10, PP. 455-469 (1976)
The Interaction of Macromolecular Solutions with Macromolecular Monolayers Adsorbed on a Hydrophobic Surface HENRI P. M. FROMAGEOT, JAMES N. GROVES, ALAN R. SEARS, and JOHN F. BROWN, JR., Corporate Research and Development, General Electric, Schenectady, N e w York 12301
Summary I n order to elucidate the general patterns of intermacromolecular surface interactions that may be involved in hemocompatibility phenomena, monolayers of representative macromolecules on an octadecylsilylated glass surface were exposed to solutions of other macromolecules, and the changes in interfacial composition were characterized by zeta potential-pH titration curves, as measured by alternating flow streaming current analysis and, in some cases, by radiotracer labeling. Experiments with poly(vinylpyrro1idone) (PVP) , a blood-compatible linear polymer; bovine serum albumin (BSA), a representative serum protein; whole human serum (HS), a complex mixture of proteins; and erythrocyte surface glycoprotein (GP), an extended-chain macromolecular amphiphile, showed the following: 1) Penetration of the original monolayer occurred within 24 hr in 9 of the 12 possible cases; it did not occur for BSA or HS monolayers exposed to PVP, and probably not for PVP exposed to GP. 2) In all cases, penetration was accompanied by no more than partial displacement of the original monolayer, thereby generating a mixed monolayer. Each of the six possible binary mixed monolayers could be obtained by a t least one of the two possible mixing sequences. 3) I n the three binary systems containing BSA, the formation of the mixed monolayer could be related to increased adsorption in the two-component system. 4) The two components of the mixed monolayers were not equally distributed across their thicknesses : thus, the outer surfaces of the PVP-BSA and (at neutral pH) the PVP-HS mixed monolayers contained only PVP; that of the BSA-HS mixtures only HS. I n the PVP-HS, and probably the GP-BSA and GP-HS mixed monolayers, the composition of the outer surface appeared pH-dependent. The resultant zeta potential versus p H profiles in the latter two cases resembled those of intact blood cells. The results suggest that neither the compact monolayers of globular proteins nor the diffuse monolayers of randomly coiled water-soluble polymers can, by 455 @ 1976 by John Wiley & Sons, Inc.
456
FROMAGEOT ET AL.
their prior adsorption on a synthetic surface, prevent the subsequent adsorption of other globular macromolecules. It is possible that the randomly coiled polymers may impede the adhesion of platelets to the substrate since the results indicate that the adsorption of such polymers causes a displacement of the shear plane.
INTRODUCTION Although the interactions between different macromolecules in solution, and those between individual macromolecules and surfaces, have both been studied extensively, there is virtually no unambiguous information available on the interactions between different types of macromolecules a t surfaces. Evaluation of a t least the general pattern of such interactions is probably essential to any understanding of the physicochemical basis for materials hemocompatibility, since the stability of a blood-materials interface may depend upon several different types of intermacromolecular surface interactions. First, of course, the native blood cell surfaces themselves consist of arrays of membrane proteins, lipids, and glycoproteins, with plasma proteins loosely adsorbed on the outside. Obviously, the structural organization and stability of any such array must depend in considerable part upon attractive interactions among the various molecular constituents of its surface. Second, manifestations of materials incompatibility with blood (e.g., clotting, thrombogenesis, hemolysis) are known to result from either the surface-induced activation of plasma proteins, like Factor XII, which can initiate clotting, or else that of the surfaces of blood cells, such as platelets or erythrocytes, which will result in thrombogenesis or hemolysis, re~pectively.'-~ However, the primary event occurring when a noncompatible material contacts blood is the formation of a fairly stable layer of the major plasma proteins (albumin, globulin, fibrinogens, etc.) which occurs almost instantly.2 Thus, any adsorption of an activatable factor or blood cell surface constituent must involve displacement of or other interaction with the initially formed macromolecular monolayer. Third, one attractive route to the development of blood-compatible surfaces would be to find an inert macromolecule that was so strongly adsorbed as to block access of activatable species to the surface. (Other alternatives are to seek structural materials which either do not adsorb activatable plasma proteins a t all, or else selectively
INTERACTION OF MACROMOLECULAR SOLUTIONS
457
adsorb nonactivatable ones irreversibly.) One might visualize the adsorbable species capable of providing an inert monolayer as either a) a globular protein with even greater affinity for solid surfaces than the ordinary plasma proteins;5 b) a randomly coiled linear polymer, such as heparin, poly(viny1 alcohol) , 5 poly(vinylpyrro1idone) ,6 poly(ethylene oxide),’ or Dextran,8 which are already known to provide partial stabilization of surfaces in contact with blood constituents ; c) a cell surface glycoprotein, such as the readily prepared major glycoprotein of the human erythrocyte, which should endow the material surface with a high degree of surface chemical similarity to the body’s own cells; or d) a combination of the above. It is of interest to note that these alternative approaches would involve completely different types of adsorbates a t the interface: the ordinary globular proteins have a compact structure and can adsorb from solution to form compact monolayers or multilayers depending on the surface p a r a m e t e r ~ ; ~the J ~simple linear polymers exist in the open, randomly coiled configuration in solution and retain this diffuse character on adsorption;11 the red blood cell glycoprotein consists of two dissimilar parts, a highly branched glycosylated polypeptide “head” and a hydrophobic “tail.”12 It forms open monolayers, approximately 120 8 thick,13which are quite similar to those proposed to exist on the surface of the erythrocyte. What is unknown, of course, is how macromolecules even as different as these would behave when allowed to interact a t a surface; e.g., which combinations might give rise to stable interaction products, like those invo1;ed in the formation of cell membranes, and which ones would lead to a displacement of one component by the other. The elucidation of broad, qualitative questions of this type requires no more than a series of experiments in which adsorbed monolayers of each species in question are allowed to interact with solutions of the others, and the change in monolayer composition is then determined. A convenient approach to making such determinations has been provided by the development of the alternating-flow streaming-current analysis technique for characterizing the composition of surfaces or adsorbates thereon via their { potential versus pH titration c u ~ v e s . 1 Accordingly, ~ we decided to undertake an investigation of the interactions of macromolecular solutions with macromolecular monolayers adsorbed on the standard 2 mm octadecylsilylated glass capillaries used in the streaming current apparatus.
458
FROMAGEOT ET AL.
We were particularly interested in the interactions among four macromolecular species : bovine serum albumin (BSA) , a well-characterized serum globular protein;15 whole human serum (HS), a complicated mixture of proteins; poly(vinylpyrro1idone) (PVP), a high molecular weight water-soluble polymer; and human red blood cell glycoprotein (GP). Part of our interest in these substances was due to their relevance to the questions of blood cell surface organization and biocompatible materials design; another reason was that they represent species of grossly different macromolecular forms and hence might show how different types of macromolecules would interact a t surfaces; and finally, because they happened to exhibit widely divergent 1 versus pH profiles, so that even small changes in monolayer composition could be easily recognized. The electrokinetic measurements, of course, can provide information only on the outermost portions of the surface, which are close to the hydrodynamic shear boundary. I n cases where the formation of a multilayered adsorbate with a n overall composition different from that of the outer surface is suspected, radioisotopic labeling or other conventional techniques must be used to define adsorbate composition.
EXPERIMENTAL Hydrophobic surfaces were made by exposing Pyrex glass capillaries to octadecyltrichlorosilane in ~ y r i d i n e . 1 ~The macromolecular adsorbates were prepared by incubating the treated capillaries with a solution of the selected macromolecule in 0.02 M tris(hydroxymethy1)aminomethane hydrochloride (Tris-HC1, pH 7.2) or in 0.01 M phosphate buffer. The incubation procedure consisted of the following steps: 1) A dry octadecylsilylated capillary (radius 2 mm, length 6 cm) was sealed with two silicone stoppers. Each stopper was penetrated by a syringe needle. 2 ) The macromolecular solution was then introduced by expelling the contents of one syringe, thereby partially filling the second. 3) Each capillary was incubated for a t least 6 hr. 4) Three 10-ml syringes containing buffer (TrisHC1) were sequentially attached to one needle and their contents were forced through the capillary, thereby removing the macromolecular solution. The capillary was then placed in the streaming current apparatus. The 1 potential was then determined as a func-
INTERACTION OF MACROMOLECULAR SOLUTIONS
459
tion of pH by titrating with 0.2 M HCl.14 The stability of the initial monolayer to either mixed monolayer formation or displacement by another macromolecule was assessed by incubating the coated and washed capillary with a solution of the second species, usually for 24 hr. The { potential was then determined as a function of pH as described above. For the assays of BSA content in the adsorbates, BSA of specific activity greater than 5 pCi/mg was obtained by reductive methylation in the presence of sodium borotritide.16 The adsorption and
40 -
-w
20 -
E
2
4 Iz
w
0-
G n 4
t; N -20 -
-40-
- 60
3
4
5
6
7
PH
Fig. 1. Titration curves of ( 0 )a BSA monolayer and of BSA monolayers exposed to ( X ) GP or to (A)HS or to ( 0 )PVP. Concentrations: BSA 0.5 mg/ml; GP 0.18 mg/ml; HS 7 mg/ml; PVP 1 mg/ml. All incubations were done at pH 7.2 in 0.02 M Tris buffer.
FROMAGEOT ET AI,.
460
incubation procedures and the solute concentrations employed were otherwise precisely the same as with unlabeled BSA. After incubation, the capillaries were rinsed and inserted in the alternating-flow streaming-current apparatus for 5 min t o simulate any surfaceconditioning effects of the electrokinetic measurement. Each capillary was then transferred to a Pyrex test tube to which 1.5 ml of NCS solubilizer and 0.2 ml of water were added. After 24 hr the capillary was removed, and 15 ml of toluene-based scintillation cocktail was added [24 g PPO (2,5-diphenyloxazole) and 0.4 g POPOP (1 ,4-bis-2(5-phenyloxazolyl)benzene) per gallon of toluene]. The resulting mixture was then analyzed by scintillation counting.
RESULTS The { potential-pH profiles for BSA monolayers before and after exposure to solutions of HS, GP, and PVP are shown in Figure 1. Comparable series for HS, GP, and PVP monolayers are shown in Figures 2, 3, and 4, respectively. The results of experiments using radio-labeled BSA are presented in Table I. Table I1 summarizes the results of both electrokinetic and radio-labeling experiments. TABLE I BSA Concentrations in Surface Films Prepared by Exposing Macromolecular Adsorbates on Octadecylsilylated Glass t o Solutions of Other Macromolecules in Tris Buffers Initial adsorbate BSA BSA BSA PVP GP HS PVP PVP
+ HS + BSA
Macromolecule to which initial adsorbate exposed
Normalized final BSA surface concentration (1.0 = 0.40 pg/cm2)b
PV P HS BSA BSA BSA BSA HS
1.0 1.0 0.6 2.0 1-2 1.8 1.1 0.4
Conditions: Buffer concn, 0.02 M ; pH 7.2; 23°C. Coverage of 0.40 pg/cm2 geometrical. Surface area, uncorrected for any roughness, believed to correspond to approximately 1.0 monolayer of BSA molecules adsorbed s i d e - 0 ~ ~ 3 a
b
INTERACTION OF MACROMOLECULAR SOLUTIONS
461
60 40 -
20 I
>
E
1 4
-
5
0-
I-
L 0
a Iw N
-20 -
-40 -
- 60
3
4
5
6
7
PH
Fig. 2. Titration curves of ( 0 )HS monolayer and of HS monolayers exposed to ( X ) BSA or ( A ) G P or (0)PVP. Concentrations: HS 7 mg/ml; BSA 5 mg/ml; G P 0.18 mg/ml; PVP 1 mg/ml. Incubation conditions are the same as in Fig. 1.
One of several similar curves showing GP-protein mixed monolayer curves under slightly different conditions of measurement is shown in Figure 5 . Here, the buffer was 0.01 M phosphate rather than 0.02 M Tris-HC1. For comparison of curve forms, Figure 5 also shows an electrophoretic mobility versus pH curve for T-type human lymphocytes in 0.29 M sucrose at 0.005 ionic strength that has recently been reported from this laboratory.''
FROMAGEOT ET AL.
462
TABLE I1 Surfacearb and Overall Compositionsc of Macromolecular Adsorbates on Octadecylsilylated Glass after 16-24 H r Exposure to Solutions of Other Macromolecules a t 23°C Adsorbate Solute
PVP
GP
BSA
HS
a Surface composition when different from overall composition is indicated by first entry. b Composition &s indicated by the zeta potential a t shear boundary above p H 5. c As deduced from electrokinetic measurements over wider pH range or from radioisotopic labeling. d Absence of G P inferred from electrokinetic detectability of mixed monolayer formed in reverse sequence. e Possibly few percent BSA also near shear plane. HS also appears near shear plane when measured below pH 5. f
DISCUSSION The results summarized in Table I1 are conveniently discussed by focusing attention first upon the nature of the monolayer exposed to a dissolved macromolecule, and second, upon the nature of the dissolved species. We will therefore examine the behavior of the PVP, GP, and both serum protein monolayers in turn. The monolayers of PVP alone presumably consist of arrays of large, loose, open coils of the intrinsically nonionic polymer." The electrokinetic profiles of such monolayers (Fig. 4) showed { potentials that were very small and virtually independent of pH. Similar profiles have been seen by us for adsorbates of dextran and poly(vinyl alcohol) on hydrophobic surfaces, and a marked decrease in the { potential has been observed following the adsorption of poly(vinyl alcohol) on gibbsite.I8 Presumably, in all these cases, the
INTERACTION OF MACROMOLECULAR SOLUTIONS I
-60L
I
3
1
4
1
I
5
6
463
I
I
7
PH
(m)
Fig. 3. Titration curves of ( 0 )a G P monolayer, an HS monolayer, and of G P monolayers exposed to ( X) BSA or to (A) HS or to (0)PVP. Concentrations: G P 0.2 mg/ml; HS 7 mg/ml; BSA 5 mg/ml; HS 7 mg/ml; BSA 5 mg/ml; PVP 2.5 X 10-3 mg/ml. Incubation conditions are the same as in Fig. 1.
nonionic open-coil structure is not capable of becoming charged by ion adsorption, but is capable of displacing the plane of shear away from the Helmholtz plane associated with the intrinsic or adsorbed charges of the substrate. When the PVP monolayers were exposed to solutions of BSA, penetration occurred to give mixed monolayers which actually contained more BSA than the pure BSA monolayers formed upon the bare substrate (Table I). Exposure of the PVP monolayer to HS
FROMAGEOT E T AL.
464
6o
-20
-60I
I
I
I
I
I
3
4
5
6
7
PH
(m)
Fig. 4. Titration curves for PVP monolayer and for a PVP monolayer exposed to ( X ) BSA or to (0) G P or to ( A ) HS. Concentrations: PVP 1 mg/ml; BSA 5 mg/ml; G P 0.18 mg/ml; HS 7 mg/ml. Incubation conditions are the same as in Fig. 1.
likewise resulted in penetration to give mixed monolayers. This was shown by changes in the adsorptivity for BSA (Table I). Interestingly enough, despite the large content of BSA, the PVP-BSA mixed monolayers showed electrokinetic profiles indistinguishable from those of PVP alone over the entire range of pH, and the PVPHS mixed monolayers showed purely PVP-like profiles above pH 5 . The obvious interpretation of this finding is that in the mixed monolayers the large open coils of the PVP extend out beyond the protein molecules and displace the plane of shear.
INTERACTION OF MACROMOLECULAR SOLUTIONS
-60I
I
I
1
I
I
3
4
5
6
7
465
PH
Fig. 5. Comparison of the titration curves of a ( 0 )BSA monolayer and a GP monolayer exposed to (x)BSA with the titration curve of (0) human type T lymphocytes12 Concentrations: BSA 5 mg/ml; GP 0.15 mg/ml. Both incubations were done at pH 7.4 in 0.01 M potassium phosphate buffer.
When the PVP monolayers were exposed to solutions of GP, no change in electrokinetic profile occurred. This might mean that the GP molecules had likewise penetrated the PVP monolayer and thus became electrokinetically shielded by it. However, the fact that the PVP-GP mixed monolayers prepared by the reverse sequence (GP monolayer exposed to PVP) did show GP near the hydrodynamic shear plane (Fig. 3) argues that penetration of the PVP layer by the GP simply did not occur. The GP molecules are probably considerably more expanded in solution than those of BSA or HS, and hence
FROMAGEOT ET AL.
466
would not be expected to pass through the open coils of the PVP as readily. The monolayers of GP, which presumably consist of open arrays of projecting sialoglycosylated polypeptide chain ends,13 showed titration curves (Fig. 3) that appeared as might have been expected for sialic acid groups of pK, 2.6 a10ne.l~ Upon exposure of such monolayers to BSA, HS, or PVP, there resulted changes to titration curves that were roughly intermediate between those of the two constituents. It was concluded that the open monolayers of G P were readily penetrated by the other macromolecules, yielding mixed monolayers, and that the projecting glycopeptide chains in the mixed monolayers were either too short or too few to provide electrical screening of species like BSA or HS, so that the charges of both components of the mixed monolayer could contribute to that a t the hydrodynamic shear boundary. Globular proteins such as BSA and the various constituents of HS normally show versus pH profiles very similar t o the mobility versus pH profiles observed in solutions. The component(s) of human serum preferentially adsorbed on octadecylsilylated glass showed a n average charge and isoelectric point about like those of a 0-globulin (Figs. 2 and 3). The behavior of such monolayers towards other macromolecules in solution was found to depend markedly upon the nature of the dissolved species. When the attacking macromolecule was PVP, neither the BSA nor HS monolayers underwent any change in electrokinetic profile (Figs. 1 and 2). Due to the large size of the molecule, the possibility of mixed monolayer formation having occurred without a change in the composition of the outer surface is most unlikely. The failure of the flexible PVP chain loops to penetrate the protein layer and attack the substrate surface is surprising; sincc very stable mixed monolayers are formed by the reverse procedure, it may indicate that a sizable number of contiguous attachment points must be available before adsorption of a flexible chain molecule can occur. When the attacking macromolecule was the other serum protein, penetration and mixed monolayer formation occurred. The BSA adsorbate exposed to HS yielded a mixed adsorbate containing the equivalent of 0.6 monolayers of BSA; the HS adsorbate exposed to BSA picked up about 1.8 monolayers of BSA. I n both cases, however, the [ versus pH profile indicated that only the HS was a significant constituent of the region near the
r:
INTERACTION OF MACROMOLECULAR SOLUTIONS
467
hydrodynamic shear boundary. Presumably, this region, furthest from the substrate surface, is dominated by portions of the largest macromolecules of the HS adsorbate, just as it is dominated by the PVP molecules in the PVP-BSA and PVP-HS mixed monolayers. The findings of partial displacement and mixed monolayer formation on octadecylsilylated glass appear in accordance with previous ellipsometric studies on glass.20 Interestingly, other workers have reported that fibrinogen can totally displace a monolayer of BSA adsorbed on silica.21 When the attacking species was the amphiphilic macromolecule GP, ready penetration of both the BSA and HS monolayers occurred (Figs. 1 and 2), yielding mixed monolayers that showed essentially the same { versus pH profiles as those obtained by exposing G P monolayers t o BSA or HS (Fig. 3). This same G P is also known to be able to penetrate the surface of intact human erythrocytes and mycoplasmas.22 The marked tendency towards mixed monolayer formation, regardless of the sequence of exposure to the surface (Table I), is a n intriguing characteristic for a species whose natural habitat is the mixed monolayer of a blood cell surface. The mixed monolayers formed between G P and BSA or HS on octadecylsilylated glass (Figs. 1-3), as well as others we have observed with other proteins and on other substrates, showed { versus p H profiles that were only approximately derivable by interpolation between those of the individual constituents. Particularly in the mixed monolayers with the higher G P contents, there was a tendency towards the appearance of a curve form that first remained quite flat with diminishing pH, and then broke and developed a slope steeper than those of the proteins in approaching the isoelectric point. Similar forms are commonly seen in the electrokinetic titration curves of blood cells, e.g., the lymphocyte curve shown in Figure 5 . We believe that the curve form in both the mixed monolayers and the blood cells results from a pH-dependent conformation change, with the outer diffuse, anionic layer of projecting glycoprotein chains tending to collapse into the underlying cationic layer of adsorbed protein a t low pH. It is hoped that current investigations of GP-protein monolayer composition and variability will shed light on the question of electrokinetic curve forms, and also that of the driving forces for mixed monolayer formation in systems involving the blood cell surface glycoprotein.
468
FROMAGEOT ET AL.
As for the other questions which initially prompted this investigation, it is now evident, first, that the reason why the initially formed monolayer of major plasma proteins on a foreign surface may not offer protection against the activation of minor plasma proteins is that ready displacement and penetration of one adsorbed plasma protein by another can occur. Second, a major limitation on the use of diffuse adsorbed layers of species such as PVP, poly(vinyl alcohol), Dcxtran, or even G P for improving the hemocompatibility of surfaces is apparent: even though such layers may not be displaced by the plasma proteins, they can be penetrated thereby, so that an activatable species may still have access to the substrate surface. What these diffuse layers might be expected to accomplish, however, in view of their ability to displace the hydrodynamic shear boundary well away from the surface or adsorbates thereon, is to keep blood cells away from direct contact with the surface. Available evidence on inhibition of platelet adhesion to glass indicates that this type of protective interaction does in fact occur.’
References 1. R. Biggs, H u m a n Blood Coagulatzon, Hemostasas and Thrombosis, Blackwell Scientific Publications, Oxford, 1972. 2. S. D. Bruck, J . Biomed. Muter. Res., 6, 173 (1972). 3. E. W. Salzman, Blood, 38, 509 (1971). 4. S. D. Brudk, Blood Computable Synthetic Polymers, A n Introduction, Charles C Thomas, Springfield, Illinois, 1974. 5. E. W. Merrill, E. W. Salzman, P. S. L. Wong, T. P. Ashford, A. H. Brown, and W. G. Austen, J . Appl. Physiol., 29, 723 (1970). 6 . J. H. Stimpfling, Trans. Bull., 27, 109 (1961). 7. J. N. George, Blood, 40, 862 (1972). 8. C. R. Ricketts, Brit. J . Anaesth., 45, 958 (1973). 9. 1). J. Lyman, J. I,. Brash, S. W. Chaikin, K. G. Klein, and M. Carini, Trans. SOC.Artif. I n t . Organs, 14, 250 (1968). 10. R. It. Stromberg, B. W. Morrissey, L. E. Smith, W. H. Grant, and C. A.
Fenstermaker, Interaction of Blood Proteans with Solid Surfaces, Annual Report for the Biomaterials Program, National Heart and Lung Institute, Bethesda, Maryland, P B 241-267 (available from the National Technical Information Service), January, 1975. 11. F. Rowland, R. Bulas, E. Rothstein, and I?. R. Eirich, in Chemistry and Physics of Interfaces, American Chemistry Society Publications, Washington, D.C., 1965. 12. J. P. Segrest, J. Kahane, R. L. Jackson, and V. T. Marchesi, Arch. Biochem. Biophys., 155, 167 (1973). 13. E. E. Uzgiris and H. P. M. Fromageot, Biopolymers, 15, 257 (1976).
INTERACTION OF MACROMOLECULAR SOLUTIONS
469
14. J. N. Groves and A. R. Sears, J. Colloid Interface Sci., 53, 83 (1975). 15. F. Soelewey, M. Rosseseu-Motieff, R. Larriote, and H. Peeters, J. Biochem., 71, 705 (1972). 16. G. E. Means and R. E. Feeney, Biochem., 7 , 2192 (1968). 17. E. E. Uzgiris and J. H. Kaplan, Anal. Biochem., 60, 455 (1974). 18. B. V. Kavanagh, A. M. Pasmer, and J. P. Quirk, The Chemical Society, Faraday Division, General Discussion No. 59, 1975. 19. G. M. W. Cook, Biol. Rev., Cambridge Philos. SOC.,43, 363 (1968). 20. L. Vroman, A. L. Adams, and M. Klings, Federation Proc., 30, 1494 (1971). 21. C. A. Fenstermaker, W. H. Grant, B. W. Morrissey, L. E. Smith, and R. R. Stromberg, Interactions of Plasma Proteins with Surfaces, Annual Report for the Biomaterials Program, National Heart and Lung Institute, Bethesda, Md., P B 232-629 (available from the National Technical Information Service), March, 1974. 22. G. Yarrison and G. L. Choules, Biochem. Biophys. Res. Comm., 52,57 (1973).
Received August 24, 1975 Revised October 20, 1975
VOL. 10, PP. 455-469 (1976)
The Interaction of Macromolecular Solutions with Macromolecular Monolayers Adsorbed on a Hydrophobic Surface HENRI P. M. FROMAGEOT, JAMES N. GROVES, ALAN R. SEARS, and JOHN F. BROWN, JR., Corporate Research and Development, General Electric, Schenectady, N e w York 12301
Summary I n order to elucidate the general patterns of intermacromolecular surface interactions that may be involved in hemocompatibility phenomena, monolayers of representative macromolecules on an octadecylsilylated glass surface were exposed to solutions of other macromolecules, and the changes in interfacial composition were characterized by zeta potential-pH titration curves, as measured by alternating flow streaming current analysis and, in some cases, by radiotracer labeling. Experiments with poly(vinylpyrro1idone) (PVP) , a blood-compatible linear polymer; bovine serum albumin (BSA), a representative serum protein; whole human serum (HS), a complex mixture of proteins; and erythrocyte surface glycoprotein (GP), an extended-chain macromolecular amphiphile, showed the following: 1) Penetration of the original monolayer occurred within 24 hr in 9 of the 12 possible cases; it did not occur for BSA or HS monolayers exposed to PVP, and probably not for PVP exposed to GP. 2) In all cases, penetration was accompanied by no more than partial displacement of the original monolayer, thereby generating a mixed monolayer. Each of the six possible binary mixed monolayers could be obtained by a t least one of the two possible mixing sequences. 3) I n the three binary systems containing BSA, the formation of the mixed monolayer could be related to increased adsorption in the two-component system. 4) The two components of the mixed monolayers were not equally distributed across their thicknesses : thus, the outer surfaces of the PVP-BSA and (at neutral pH) the PVP-HS mixed monolayers contained only PVP; that of the BSA-HS mixtures only HS. I n the PVP-HS, and probably the GP-BSA and GP-HS mixed monolayers, the composition of the outer surface appeared pH-dependent. The resultant zeta potential versus p H profiles in the latter two cases resembled those of intact blood cells. The results suggest that neither the compact monolayers of globular proteins nor the diffuse monolayers of randomly coiled water-soluble polymers can, by 455 @ 1976 by John Wiley & Sons, Inc.
456
FROMAGEOT ET AL.
their prior adsorption on a synthetic surface, prevent the subsequent adsorption of other globular macromolecules. It is possible that the randomly coiled polymers may impede the adhesion of platelets to the substrate since the results indicate that the adsorption of such polymers causes a displacement of the shear plane.
INTRODUCTION Although the interactions between different macromolecules in solution, and those between individual macromolecules and surfaces, have both been studied extensively, there is virtually no unambiguous information available on the interactions between different types of macromolecules a t surfaces. Evaluation of a t least the general pattern of such interactions is probably essential to any understanding of the physicochemical basis for materials hemocompatibility, since the stability of a blood-materials interface may depend upon several different types of intermacromolecular surface interactions. First, of course, the native blood cell surfaces themselves consist of arrays of membrane proteins, lipids, and glycoproteins, with plasma proteins loosely adsorbed on the outside. Obviously, the structural organization and stability of any such array must depend in considerable part upon attractive interactions among the various molecular constituents of its surface. Second, manifestations of materials incompatibility with blood (e.g., clotting, thrombogenesis, hemolysis) are known to result from either the surface-induced activation of plasma proteins, like Factor XII, which can initiate clotting, or else that of the surfaces of blood cells, such as platelets or erythrocytes, which will result in thrombogenesis or hemolysis, re~pectively.'-~ However, the primary event occurring when a noncompatible material contacts blood is the formation of a fairly stable layer of the major plasma proteins (albumin, globulin, fibrinogens, etc.) which occurs almost instantly.2 Thus, any adsorption of an activatable factor or blood cell surface constituent must involve displacement of or other interaction with the initially formed macromolecular monolayer. Third, one attractive route to the development of blood-compatible surfaces would be to find an inert macromolecule that was so strongly adsorbed as to block access of activatable species to the surface. (Other alternatives are to seek structural materials which either do not adsorb activatable plasma proteins a t all, or else selectively
INTERACTION OF MACROMOLECULAR SOLUTIONS
457
adsorb nonactivatable ones irreversibly.) One might visualize the adsorbable species capable of providing an inert monolayer as either a) a globular protein with even greater affinity for solid surfaces than the ordinary plasma proteins;5 b) a randomly coiled linear polymer, such as heparin, poly(viny1 alcohol) , 5 poly(vinylpyrro1idone) ,6 poly(ethylene oxide),’ or Dextran,8 which are already known to provide partial stabilization of surfaces in contact with blood constituents ; c) a cell surface glycoprotein, such as the readily prepared major glycoprotein of the human erythrocyte, which should endow the material surface with a high degree of surface chemical similarity to the body’s own cells; or d) a combination of the above. It is of interest to note that these alternative approaches would involve completely different types of adsorbates a t the interface: the ordinary globular proteins have a compact structure and can adsorb from solution to form compact monolayers or multilayers depending on the surface p a r a m e t e r ~ ; ~the J ~simple linear polymers exist in the open, randomly coiled configuration in solution and retain this diffuse character on adsorption;11 the red blood cell glycoprotein consists of two dissimilar parts, a highly branched glycosylated polypeptide “head” and a hydrophobic “tail.”12 It forms open monolayers, approximately 120 8 thick,13which are quite similar to those proposed to exist on the surface of the erythrocyte. What is unknown, of course, is how macromolecules even as different as these would behave when allowed to interact a t a surface; e.g., which combinations might give rise to stable interaction products, like those invo1;ed in the formation of cell membranes, and which ones would lead to a displacement of one component by the other. The elucidation of broad, qualitative questions of this type requires no more than a series of experiments in which adsorbed monolayers of each species in question are allowed to interact with solutions of the others, and the change in monolayer composition is then determined. A convenient approach to making such determinations has been provided by the development of the alternating-flow streaming-current analysis technique for characterizing the composition of surfaces or adsorbates thereon via their { potential versus pH titration c u ~ v e s . 1 Accordingly, ~ we decided to undertake an investigation of the interactions of macromolecular solutions with macromolecular monolayers adsorbed on the standard 2 mm octadecylsilylated glass capillaries used in the streaming current apparatus.
458
FROMAGEOT ET AL.
We were particularly interested in the interactions among four macromolecular species : bovine serum albumin (BSA) , a well-characterized serum globular protein;15 whole human serum (HS), a complicated mixture of proteins; poly(vinylpyrro1idone) (PVP), a high molecular weight water-soluble polymer; and human red blood cell glycoprotein (GP). Part of our interest in these substances was due to their relevance to the questions of blood cell surface organization and biocompatible materials design; another reason was that they represent species of grossly different macromolecular forms and hence might show how different types of macromolecules would interact a t surfaces; and finally, because they happened to exhibit widely divergent 1 versus pH profiles, so that even small changes in monolayer composition could be easily recognized. The electrokinetic measurements, of course, can provide information only on the outermost portions of the surface, which are close to the hydrodynamic shear boundary. I n cases where the formation of a multilayered adsorbate with a n overall composition different from that of the outer surface is suspected, radioisotopic labeling or other conventional techniques must be used to define adsorbate composition.
EXPERIMENTAL Hydrophobic surfaces were made by exposing Pyrex glass capillaries to octadecyltrichlorosilane in ~ y r i d i n e . 1 ~The macromolecular adsorbates were prepared by incubating the treated capillaries with a solution of the selected macromolecule in 0.02 M tris(hydroxymethy1)aminomethane hydrochloride (Tris-HC1, pH 7.2) or in 0.01 M phosphate buffer. The incubation procedure consisted of the following steps: 1) A dry octadecylsilylated capillary (radius 2 mm, length 6 cm) was sealed with two silicone stoppers. Each stopper was penetrated by a syringe needle. 2 ) The macromolecular solution was then introduced by expelling the contents of one syringe, thereby partially filling the second. 3) Each capillary was incubated for a t least 6 hr. 4) Three 10-ml syringes containing buffer (TrisHC1) were sequentially attached to one needle and their contents were forced through the capillary, thereby removing the macromolecular solution. The capillary was then placed in the streaming current apparatus. The 1 potential was then determined as a func-
INTERACTION OF MACROMOLECULAR SOLUTIONS
459
tion of pH by titrating with 0.2 M HCl.14 The stability of the initial monolayer to either mixed monolayer formation or displacement by another macromolecule was assessed by incubating the coated and washed capillary with a solution of the second species, usually for 24 hr. The { potential was then determined as a function of pH as described above. For the assays of BSA content in the adsorbates, BSA of specific activity greater than 5 pCi/mg was obtained by reductive methylation in the presence of sodium borotritide.16 The adsorption and
40 -
-w
20 -
E
2
4 Iz
w
0-
G n 4
t; N -20 -
-40-
- 60
3
4
5
6
7
PH
Fig. 1. Titration curves of ( 0 )a BSA monolayer and of BSA monolayers exposed to ( X ) GP or to (A)HS or to ( 0 )PVP. Concentrations: BSA 0.5 mg/ml; GP 0.18 mg/ml; HS 7 mg/ml; PVP 1 mg/ml. All incubations were done at pH 7.2 in 0.02 M Tris buffer.
FROMAGEOT ET AI,.
460
incubation procedures and the solute concentrations employed were otherwise precisely the same as with unlabeled BSA. After incubation, the capillaries were rinsed and inserted in the alternating-flow streaming-current apparatus for 5 min t o simulate any surfaceconditioning effects of the electrokinetic measurement. Each capillary was then transferred to a Pyrex test tube to which 1.5 ml of NCS solubilizer and 0.2 ml of water were added. After 24 hr the capillary was removed, and 15 ml of toluene-based scintillation cocktail was added [24 g PPO (2,5-diphenyloxazole) and 0.4 g POPOP (1 ,4-bis-2(5-phenyloxazolyl)benzene) per gallon of toluene]. The resulting mixture was then analyzed by scintillation counting.
RESULTS The { potential-pH profiles for BSA monolayers before and after exposure to solutions of HS, GP, and PVP are shown in Figure 1. Comparable series for HS, GP, and PVP monolayers are shown in Figures 2, 3, and 4, respectively. The results of experiments using radio-labeled BSA are presented in Table I. Table I1 summarizes the results of both electrokinetic and radio-labeling experiments. TABLE I BSA Concentrations in Surface Films Prepared by Exposing Macromolecular Adsorbates on Octadecylsilylated Glass t o Solutions of Other Macromolecules in Tris Buffers Initial adsorbate BSA BSA BSA PVP GP HS PVP PVP
+ HS + BSA
Macromolecule to which initial adsorbate exposed
Normalized final BSA surface concentration (1.0 = 0.40 pg/cm2)b
PV P HS BSA BSA BSA BSA HS
1.0 1.0 0.6 2.0 1-2 1.8 1.1 0.4
Conditions: Buffer concn, 0.02 M ; pH 7.2; 23°C. Coverage of 0.40 pg/cm2 geometrical. Surface area, uncorrected for any roughness, believed to correspond to approximately 1.0 monolayer of BSA molecules adsorbed s i d e - 0 ~ ~ 3 a
b
INTERACTION OF MACROMOLECULAR SOLUTIONS
461
60 40 -
20 I
>
E
1 4
-
5
0-
I-
L 0
a Iw N
-20 -
-40 -
- 60
3
4
5
6
7
PH
Fig. 2. Titration curves of ( 0 )HS monolayer and of HS monolayers exposed to ( X ) BSA or ( A ) G P or (0)PVP. Concentrations: HS 7 mg/ml; BSA 5 mg/ml; G P 0.18 mg/ml; PVP 1 mg/ml. Incubation conditions are the same as in Fig. 1.
One of several similar curves showing GP-protein mixed monolayer curves under slightly different conditions of measurement is shown in Figure 5 . Here, the buffer was 0.01 M phosphate rather than 0.02 M Tris-HC1. For comparison of curve forms, Figure 5 also shows an electrophoretic mobility versus pH curve for T-type human lymphocytes in 0.29 M sucrose at 0.005 ionic strength that has recently been reported from this laboratory.''
FROMAGEOT ET AL.
462
TABLE I1 Surfacearb and Overall Compositionsc of Macromolecular Adsorbates on Octadecylsilylated Glass after 16-24 H r Exposure to Solutions of Other Macromolecules a t 23°C Adsorbate Solute
PVP
GP
BSA
HS
a Surface composition when different from overall composition is indicated by first entry. b Composition &s indicated by the zeta potential a t shear boundary above p H 5. c As deduced from electrokinetic measurements over wider pH range or from radioisotopic labeling. d Absence of G P inferred from electrokinetic detectability of mixed monolayer formed in reverse sequence. e Possibly few percent BSA also near shear plane. HS also appears near shear plane when measured below pH 5. f
DISCUSSION The results summarized in Table I1 are conveniently discussed by focusing attention first upon the nature of the monolayer exposed to a dissolved macromolecule, and second, upon the nature of the dissolved species. We will therefore examine the behavior of the PVP, GP, and both serum protein monolayers in turn. The monolayers of PVP alone presumably consist of arrays of large, loose, open coils of the intrinsically nonionic polymer." The electrokinetic profiles of such monolayers (Fig. 4) showed { potentials that were very small and virtually independent of pH. Similar profiles have been seen by us for adsorbates of dextran and poly(vinyl alcohol) on hydrophobic surfaces, and a marked decrease in the { potential has been observed following the adsorption of poly(vinyl alcohol) on gibbsite.I8 Presumably, in all these cases, the
INTERACTION OF MACROMOLECULAR SOLUTIONS I
-60L
I
3
1
4
1
I
5
6
463
I
I
7
PH
(m)
Fig. 3. Titration curves of ( 0 )a G P monolayer, an HS monolayer, and of G P monolayers exposed to ( X) BSA or to (A) HS or to (0)PVP. Concentrations: G P 0.2 mg/ml; HS 7 mg/ml; BSA 5 mg/ml; HS 7 mg/ml; BSA 5 mg/ml; PVP 2.5 X 10-3 mg/ml. Incubation conditions are the same as in Fig. 1.
nonionic open-coil structure is not capable of becoming charged by ion adsorption, but is capable of displacing the plane of shear away from the Helmholtz plane associated with the intrinsic or adsorbed charges of the substrate. When the PVP monolayers were exposed to solutions of BSA, penetration occurred to give mixed monolayers which actually contained more BSA than the pure BSA monolayers formed upon the bare substrate (Table I). Exposure of the PVP monolayer to HS
FROMAGEOT E T AL.
464
6o
-20
-60I
I
I
I
I
I
3
4
5
6
7
PH
(m)
Fig. 4. Titration curves for PVP monolayer and for a PVP monolayer exposed to ( X ) BSA or to (0) G P or to ( A ) HS. Concentrations: PVP 1 mg/ml; BSA 5 mg/ml; G P 0.18 mg/ml; HS 7 mg/ml. Incubation conditions are the same as in Fig. 1.
likewise resulted in penetration to give mixed monolayers. This was shown by changes in the adsorptivity for BSA (Table I). Interestingly enough, despite the large content of BSA, the PVP-BSA mixed monolayers showed electrokinetic profiles indistinguishable from those of PVP alone over the entire range of pH, and the PVPHS mixed monolayers showed purely PVP-like profiles above pH 5 . The obvious interpretation of this finding is that in the mixed monolayers the large open coils of the PVP extend out beyond the protein molecules and displace the plane of shear.
INTERACTION OF MACROMOLECULAR SOLUTIONS
-60I
I
I
1
I
I
3
4
5
6
7
465
PH
Fig. 5. Comparison of the titration curves of a ( 0 )BSA monolayer and a GP monolayer exposed to (x)BSA with the titration curve of (0) human type T lymphocytes12 Concentrations: BSA 5 mg/ml; GP 0.15 mg/ml. Both incubations were done at pH 7.4 in 0.01 M potassium phosphate buffer.
When the PVP monolayers were exposed to solutions of GP, no change in electrokinetic profile occurred. This might mean that the GP molecules had likewise penetrated the PVP monolayer and thus became electrokinetically shielded by it. However, the fact that the PVP-GP mixed monolayers prepared by the reverse sequence (GP monolayer exposed to PVP) did show GP near the hydrodynamic shear plane (Fig. 3) argues that penetration of the PVP layer by the GP simply did not occur. The GP molecules are probably considerably more expanded in solution than those of BSA or HS, and hence
FROMAGEOT ET AL.
466
would not be expected to pass through the open coils of the PVP as readily. The monolayers of GP, which presumably consist of open arrays of projecting sialoglycosylated polypeptide chain ends,13 showed titration curves (Fig. 3) that appeared as might have been expected for sialic acid groups of pK, 2.6 a10ne.l~ Upon exposure of such monolayers to BSA, HS, or PVP, there resulted changes to titration curves that were roughly intermediate between those of the two constituents. It was concluded that the open monolayers of G P were readily penetrated by the other macromolecules, yielding mixed monolayers, and that the projecting glycopeptide chains in the mixed monolayers were either too short or too few to provide electrical screening of species like BSA or HS, so that the charges of both components of the mixed monolayer could contribute to that a t the hydrodynamic shear boundary. Globular proteins such as BSA and the various constituents of HS normally show versus pH profiles very similar t o the mobility versus pH profiles observed in solutions. The component(s) of human serum preferentially adsorbed on octadecylsilylated glass showed a n average charge and isoelectric point about like those of a 0-globulin (Figs. 2 and 3). The behavior of such monolayers towards other macromolecules in solution was found to depend markedly upon the nature of the dissolved species. When the attacking macromolecule was PVP, neither the BSA nor HS monolayers underwent any change in electrokinetic profile (Figs. 1 and 2). Due to the large size of the molecule, the possibility of mixed monolayer formation having occurred without a change in the composition of the outer surface is most unlikely. The failure of the flexible PVP chain loops to penetrate the protein layer and attack the substrate surface is surprising; sincc very stable mixed monolayers are formed by the reverse procedure, it may indicate that a sizable number of contiguous attachment points must be available before adsorption of a flexible chain molecule can occur. When the attacking macromolecule was the other serum protein, penetration and mixed monolayer formation occurred. The BSA adsorbate exposed to HS yielded a mixed adsorbate containing the equivalent of 0.6 monolayers of BSA; the HS adsorbate exposed to BSA picked up about 1.8 monolayers of BSA. I n both cases, however, the [ versus pH profile indicated that only the HS was a significant constituent of the region near the
r:
INTERACTION OF MACROMOLECULAR SOLUTIONS
467
hydrodynamic shear boundary. Presumably, this region, furthest from the substrate surface, is dominated by portions of the largest macromolecules of the HS adsorbate, just as it is dominated by the PVP molecules in the PVP-BSA and PVP-HS mixed monolayers. The findings of partial displacement and mixed monolayer formation on octadecylsilylated glass appear in accordance with previous ellipsometric studies on glass.20 Interestingly, other workers have reported that fibrinogen can totally displace a monolayer of BSA adsorbed on silica.21 When the attacking species was the amphiphilic macromolecule GP, ready penetration of both the BSA and HS monolayers occurred (Figs. 1 and 2), yielding mixed monolayers that showed essentially the same { versus pH profiles as those obtained by exposing G P monolayers t o BSA or HS (Fig. 3). This same G P is also known to be able to penetrate the surface of intact human erythrocytes and mycoplasmas.22 The marked tendency towards mixed monolayer formation, regardless of the sequence of exposure to the surface (Table I), is a n intriguing characteristic for a species whose natural habitat is the mixed monolayer of a blood cell surface. The mixed monolayers formed between G P and BSA or HS on octadecylsilylated glass (Figs. 1-3), as well as others we have observed with other proteins and on other substrates, showed { versus p H profiles that were only approximately derivable by interpolation between those of the individual constituents. Particularly in the mixed monolayers with the higher G P contents, there was a tendency towards the appearance of a curve form that first remained quite flat with diminishing pH, and then broke and developed a slope steeper than those of the proteins in approaching the isoelectric point. Similar forms are commonly seen in the electrokinetic titration curves of blood cells, e.g., the lymphocyte curve shown in Figure 5 . We believe that the curve form in both the mixed monolayers and the blood cells results from a pH-dependent conformation change, with the outer diffuse, anionic layer of projecting glycoprotein chains tending to collapse into the underlying cationic layer of adsorbed protein a t low pH. It is hoped that current investigations of GP-protein monolayer composition and variability will shed light on the question of electrokinetic curve forms, and also that of the driving forces for mixed monolayer formation in systems involving the blood cell surface glycoprotein.
468
FROMAGEOT ET AL.
As for the other questions which initially prompted this investigation, it is now evident, first, that the reason why the initially formed monolayer of major plasma proteins on a foreign surface may not offer protection against the activation of minor plasma proteins is that ready displacement and penetration of one adsorbed plasma protein by another can occur. Second, a major limitation on the use of diffuse adsorbed layers of species such as PVP, poly(vinyl alcohol), Dcxtran, or even G P for improving the hemocompatibility of surfaces is apparent: even though such layers may not be displaced by the plasma proteins, they can be penetrated thereby, so that an activatable species may still have access to the substrate surface. What these diffuse layers might be expected to accomplish, however, in view of their ability to displace the hydrodynamic shear boundary well away from the surface or adsorbates thereon, is to keep blood cells away from direct contact with the surface. Available evidence on inhibition of platelet adhesion to glass indicates that this type of protective interaction does in fact occur.’
References 1. R. Biggs, H u m a n Blood Coagulatzon, Hemostasas and Thrombosis, Blackwell Scientific Publications, Oxford, 1972. 2. S. D. Bruck, J . Biomed. Muter. Res., 6, 173 (1972). 3. E. W. Salzman, Blood, 38, 509 (1971). 4. S. D. Brudk, Blood Computable Synthetic Polymers, A n Introduction, Charles C Thomas, Springfield, Illinois, 1974. 5. E. W. Merrill, E. W. Salzman, P. S. L. Wong, T. P. Ashford, A. H. Brown, and W. G. Austen, J . Appl. Physiol., 29, 723 (1970). 6 . J. H. Stimpfling, Trans. Bull., 27, 109 (1961). 7. J. N. George, Blood, 40, 862 (1972). 8. C. R. Ricketts, Brit. J . Anaesth., 45, 958 (1973). 9. 1). J. Lyman, J. I,. Brash, S. W. Chaikin, K. G. Klein, and M. Carini, Trans. SOC.Artif. I n t . Organs, 14, 250 (1968). 10. R. It. Stromberg, B. W. Morrissey, L. E. Smith, W. H. Grant, and C. A.
Fenstermaker, Interaction of Blood Proteans with Solid Surfaces, Annual Report for the Biomaterials Program, National Heart and Lung Institute, Bethesda, Maryland, P B 241-267 (available from the National Technical Information Service), January, 1975. 11. F. Rowland, R. Bulas, E. Rothstein, and I?. R. Eirich, in Chemistry and Physics of Interfaces, American Chemistry Society Publications, Washington, D.C., 1965. 12. J. P. Segrest, J. Kahane, R. L. Jackson, and V. T. Marchesi, Arch. Biochem. Biophys., 155, 167 (1973). 13. E. E. Uzgiris and H. P. M. Fromageot, Biopolymers, 15, 257 (1976).
INTERACTION OF MACROMOLECULAR SOLUTIONS
469
14. J. N. Groves and A. R. Sears, J. Colloid Interface Sci., 53, 83 (1975). 15. F. Soelewey, M. Rosseseu-Motieff, R. Larriote, and H. Peeters, J. Biochem., 71, 705 (1972). 16. G. E. Means and R. E. Feeney, Biochem., 7 , 2192 (1968). 17. E. E. Uzgiris and J. H. Kaplan, Anal. Biochem., 60, 455 (1974). 18. B. V. Kavanagh, A. M. Pasmer, and J. P. Quirk, The Chemical Society, Faraday Division, General Discussion No. 59, 1975. 19. G. M. W. Cook, Biol. Rev., Cambridge Philos. SOC.,43, 363 (1968). 20. L. Vroman, A. L. Adams, and M. Klings, Federation Proc., 30, 1494 (1971). 21. C. A. Fenstermaker, W. H. Grant, B. W. Morrissey, L. E. Smith, and R. R. Stromberg, Interactions of Plasma Proteins with Surfaces, Annual Report for the Biomaterials Program, National Heart and Lung Institute, Bethesda, Md., P B 232-629 (available from the National Technical Information Service), March, 1974. 22. G. Yarrison and G. L. Choules, Biochem. Biophys. Res. Comm., 52,57 (1973).
Received August 24, 1975 Revised October 20, 1975
