Albumin-binding surfaces for implantable devices James R. Keogh, Fredrik F. Velander, and John W. Eaton* Department of Laboratory Medicine and Pathology and Dight Laborafovies, University of Minnesota, Minneapolis, Minnesota 55455, and Medtronic, Inc., 7000 Central Avenue N.E., Minneapolis, Minnesota 55432 Surfaces of implantable and blood contactdevices accumulate adsorbed and denatured proteins. This anomalous layer of proteins may help trigger unwanted events such as activation of coagulation systems and, perhaps, chronic inflammation. Because, in many experimental systems, the purposeful coating of surfaces with albumin will biologically "passivate" materials, we have attempted to develop polymers which, when exposed to blood or body fluids, will spontaneously, selectively, and reversibly adsorb host albumin. We report here a novel derivatization technique for increasing the albumin affinity of implantable polyetherurethane (PU). The technique is based on the incorporation of high-molecular-weight dextran to which the albumin-binding dye Cibacron Blue is covalently attached. Somewhat surprisingly, the amounts of human albumin adsorbed by Blue Dextran-modified and unmodified PU are quite similar. There
are, however, important differences. First, the binding of albumin to derivatized PU is specific and not readily blocked by proteins in albumin-depleted human serum. Second, the majority of albumin associated with derivatized PU appears to be reversibly bound. Third, the binding of albumin to derivatized PU evidently is mediated primarily through ligandspecific binding of the protein to the albumin-binding dextran-dye conjugate. We conclude that it is possible to produce implantable polymers having surfaces which display albumin-binding dyes that selectively and reversibly bind albumin. Materials with this property, when implanted or exposed to blood, should form an infinitely renewable coating of albumin derived from physiologic fluids. This surface modification strategy may spawn a new generation of implantable materials with improved biologic compatibility.
INTRODUCTION
Blood-contact and implantable artificial medical devices cause a number of iatrogenic effects, including activation of coagulation, chronic inflammation, and colonization by pathogenic bacteria. Upon contact with body fluids, medical polymers almost instantly develop a film of adsorbed and variably denatured proteins.' This protein coat, in turn, may be involved in some or all of the above unwanted effects. For example, blood-contact devices such as vascular catheters, vascular grafts, prosthetic heart valves, and blood oxygenators may serve as nidi for thrombus formation. This prothrombotic effect, in turn, is due to the tendency of the device surface to adsorb procoagulant *To whom correspondence should be addressed at Division of Experimental Pathology, Albany Medical College, 47 New Scotland Avenue, Albany, NY 12208.
Journal of Biomedical Materials Research, Vol. 26, 441-456 (1992) CCC 0021-9304/92/040441-16$4.00 0 1992 John Wiley & Sons, Inc.
442
KEOGH, VELANDER, AND EATON
plasma proteins and, simultaneously, to attract and activate The deposition of platelets may be mediated by interactions of platelet receptors with surface adsorbed protein^.^ One possible solution to the problem of chaotic adsorption of proteins and cells to medical device surfaces is the use of a layer of albumin as "camouf lage." Indeed, in some experimental systems, surfaces coated with albumin show little platelet a d h e ~ e n c e . ~ Unfortunately, -~ selective coating of surfaces with albumin does not occur in vim.Despite the fact that albumin is the predominant plasma protein, many other proteins have at least equal affinity and many of these proteins will later exchange for initially adsorbed albumin some time after first contact with physiologic fluids.8 For these reasons, several groups have attempted to engineer surfaces with enhanced albumin affinity. Perhaps the most successful attempt thus far was based on the well-known ability of albumin to bind free fatty acids with high affinity. Therefore, Munro et al.' earlier made surfaces displaying straightchain 16- or 18-carbon alkyl groups in hopes that this might enhance the albumin affinity. Indeed, the addition of alkyl chains did increase the surface affinity for albumin during brief incubations'"-'2 and improved thromboresistance in an acute canine ex uivo e~periment.'~ Unfortunately, these materials do not have a high degree of selectivity for albumin binding and may, due to the highly hydrophobic nature of the surface, encourage the subsequent denaturation of adsorbed protein. Our work is based on another strategem for the selective binding of albumin. Albumin has the ability to bind and transport a large number of compounds in addition to free fatty acids. One of the physiological functions of serum albumin is the binding of bilirubin. Free bilirubin has toxic effects but binding to albumin renders it nontoxic. Albumin binds unconjugated bilirubin arising from heme breakdown and transports it to the liver for conjugation and secretion into the bile.14 Albumin has one very strong binding site for bilirubin, with an association constant of and two secondary sites having association constants of 1057.14-16 These binding sites are usually free; only one molecule of albumin in 15 contains bound bilirubin." The bilirubin-binding sites on albumin also have high affinity for a number of triazine dyes which have a structural similarity to bilirubin. Both bilirubin and the albumin-binding triazine dyes consist of planar aromatic ring systems and the positions of negatively charged groups can be superimposed in analogous position~.'~ In fact, a commonly used technique for purifying albumin involves the use of an immobilized triazine dye, Cibacron Blue F3GA. Passage of whole human plasma through a column of Cibacron Blue dye-Sepharose will remove approximately 98% of the albumin.IR Cibacron Blue dye has been covalently attached to a variety of support matrices including agarose, Sephacrylo, Sephadex, Sepharose, SpheronB, and Ultrogel@." In the present work, we have elected to use Cibacron Blue dye covalently attached to high-molecular-weight dextran (2 x lo6 daltons) ("Blue Dextran"; see Fig. 1).Dextran was employed as a hydrophilic "spacer" to ensure that albumin, once bound to the dye, would not undergo surface precipitation. Blue Dextran was therefore incorporated into implant-grade
ALBUMIN-BINDING SURFACES FOR IMPLANTABLE DEVICES
0w 2
443
Dextran
-0, H
Figure 1. Structure of blue dextran.
polyetherurethane using a cosolvation technique. The resulting surfacederivatized polymers have many of the desired characteristics, including selective and reversible binding of albumin. It is believed that these materials will also be immune to enzymatic attack; mammals do not produce dextranases and the dye is known to be enzymatically stable. MATERIALS A N D METHODS
Preparation of Blue Dextran-incorporated polyetherurethane
The incorporation of Blue Dextran (Sigma Chemical Co., St. Louis, MO) into Pellethane 2363-55D, a polyetherurethane commercially available from Dow Chemical Co., Midland, MI, was performed by dissolving 0.3 g of Blue Dextran in 1.5 mL of deionized (DI) water. This mixture was then added to 20 g of N,N-dimethylacetamide (DMAC)(Fisher Scientific Co., Pittsburg, PA). To this Blue Dextran/solvent mixture, 1.0 g of Pellethane 2363-55D was added. The mixture was shaken overnight to allow complete solvation of the polyurethane. The solvated Blue Dextran/polyurethane mixture was then cast on a Mylar@release sheet and dried overnight at 50°C to remove the solvent. It is important to note that it was not possible to make control preparations derivatized with dextran alone due to the impossibility of cosolvating dextran of molecular weight 2 x 10' with Pellethane 2363-55D. Adsorption of '251-albumin
Samples (1.0 cm') of derivatized and nonderivatized PU sheet were immersed in '251-humanalbumin (15.7 &mL; specific activity = 8.5 pCi/mg
444
KEOGH, VELANDER, A N D EATON
protein) (Mallinkrodt, St. Louis, MO) dissolved in PBS for 1 h. The samples were then rinsed by repetitive dipping ( ~ 1 0in ) very large volumes of isotonic phosphate buffered saline (PBS), pH 7.4. The samples were then counted in a scintillation counter.
Elution of albumin with NaSCN
The state of surface-adsorbed '251-albumin(which was not removed by PBS washings as described above) was determined by measuring the extent of protein elution by a solution of sodium thiocyanate, a chaotropic agent. Once again, samples (1.0 cm') were exposed to '251-albumin(15.7 pg/mL) for 1 h and repetitively rinsed in PBS as described above. Next, the samples were immersed in large volumes of a solution of 0.5M NaSCN and 50 mM Tris/HCl, pH 8.0 @ 25"C, with constant gentle agitation for 1 h and, finally, rinsed again in PBS. Control samples for determination of the amounts of albumin initially absorbed were exposed to '251-albuminand rinsed only in PBS. All samples were then counted in a scintillation counter. Elution of albumin with 1%SDS
Samples (1.0 cm') were incubated in 'Z51-albumin(15.7 ,ug/mL) for 12 h. All samples were then thoroughly rinsed in PBS. Some samples were counted immediately while others were incubated in 1%SDS for 1 h. These samples were again rinsed in PBS and counted in a scintillation counter.
Adsorption of Iz5I-fibrinogenon materials precoated with albumin
Samples (1.0 cm') were incubated in unlabeled albumin (15.7 pg/mL) for 1 h. Control samples were preincubated in PBS alone. The samples were then rinsed in PBS and incubated in '251-fibrinogen(15.7 pg/mL; specific activity = 115 pCi/mg protein) (Amersham,Arlington Heights, IL) for 1h. The samples were rinsed in PBS and then counted in a scintillation counter. Control samples were preincubated with PBS and subsequently with 1251-fibrinogen. Selective binding of albumin
The selectivity of albumin binding by derivatized and nonderivatized PU in the presence of other human serum proteins was determined using '251-albumin and increasing amounts of albumin-depleted human serum (ADHS). The ADHS was prepared by chromatographing 1.0 mL of whole human serum on an Affi-Gel Blue (Sigma) column (1 cm X 8 cm, bed volume = 6.3 mL, flow rate = 0.5 mL/min) with an elution buffer comprised of PBS. The serum collected from this column was free of detectable albumin as
ALBUMIN-BINDING SURFACES FOR IMPLANTABLE DEVICES
445
judged by SDS-polyacrylamide gel electrophoresis (see below). The resultant ADHS was then used in increasing amounts to competively block nonspecific binding of radiolabeled albumin. Samples (1.0 cm') of derivatized and nonderivatized PU film were immersed in solutions of 1251-humanalbumin (15.7 pg/mL) containing varying amounts of ADHS for 1 h. The samples were then rinsed in PBS three times and counted in a scintillation counter. Reversible binding of albumin
The reversibility of albumin binding to derivatized and nonderivatized PU was determined by measuring the extent of competitive elution of pre-bound '251-albuminby unlabeled human albumin. Samples (1.0 cm') were exposed to '251-albumin(15.7 pg/mL) for 1 h and then rinsed in PBS. Next, the samples were exposed to unlabeled human serum albumin (3 mg/mL) for 1 h and rinsed again with PBS. Controls were exposed to 1251-albuminonly. After the final rinsing the samples were counted in a scintillation counter. Site of albumin binding
The ligand specificity of albumin binding to derivatized and nonderivatized PU was investigated by using solution-phase Blue Dextran to competitively elute albumin bound to PU. Derivatized and nonderivatized PU samples (1.0 cm2) were immersed in '251-albumin (15.7 pg/mL) for 1 h. The samples were then rinsed in PBS and incubated in PBS containing soluble Blue Dextran (5 mg/mL) for 1 h. At the end of the incubation, the samples were rinsed in PBS and counted in a scintillation counter. Electrophoretic analysis of adsorbed proteins
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed on derivatized and nonderivatized PU to determine the types of proteins that adsorbed to the surface from whole, freshly shed unanticoagulated human blood. Samples (1.0 cm') of derivatized and nonderivatized PU film were placed in 12 x 75 mm glass culture tubes. The tubes were then filled with freshly drawn blood (
are, however, important differences. First, the binding of albumin to derivatized PU is specific and not readily blocked by proteins in albumin-depleted human serum. Second, the majority of albumin associated with derivatized PU appears to be reversibly bound. Third, the binding of albumin to derivatized PU evidently is mediated primarily through ligandspecific binding of the protein to the albumin-binding dextran-dye conjugate. We conclude that it is possible to produce implantable polymers having surfaces which display albumin-binding dyes that selectively and reversibly bind albumin. Materials with this property, when implanted or exposed to blood, should form an infinitely renewable coating of albumin derived from physiologic fluids. This surface modification strategy may spawn a new generation of implantable materials with improved biologic compatibility.
INTRODUCTION
Blood-contact and implantable artificial medical devices cause a number of iatrogenic effects, including activation of coagulation, chronic inflammation, and colonization by pathogenic bacteria. Upon contact with body fluids, medical polymers almost instantly develop a film of adsorbed and variably denatured proteins.' This protein coat, in turn, may be involved in some or all of the above unwanted effects. For example, blood-contact devices such as vascular catheters, vascular grafts, prosthetic heart valves, and blood oxygenators may serve as nidi for thrombus formation. This prothrombotic effect, in turn, is due to the tendency of the device surface to adsorb procoagulant *To whom correspondence should be addressed at Division of Experimental Pathology, Albany Medical College, 47 New Scotland Avenue, Albany, NY 12208.
Journal of Biomedical Materials Research, Vol. 26, 441-456 (1992) CCC 0021-9304/92/040441-16$4.00 0 1992 John Wiley & Sons, Inc.
442
KEOGH, VELANDER, AND EATON
plasma proteins and, simultaneously, to attract and activate The deposition of platelets may be mediated by interactions of platelet receptors with surface adsorbed protein^.^ One possible solution to the problem of chaotic adsorption of proteins and cells to medical device surfaces is the use of a layer of albumin as "camouf lage." Indeed, in some experimental systems, surfaces coated with albumin show little platelet a d h e ~ e n c e . ~ Unfortunately, -~ selective coating of surfaces with albumin does not occur in vim.Despite the fact that albumin is the predominant plasma protein, many other proteins have at least equal affinity and many of these proteins will later exchange for initially adsorbed albumin some time after first contact with physiologic fluids.8 For these reasons, several groups have attempted to engineer surfaces with enhanced albumin affinity. Perhaps the most successful attempt thus far was based on the well-known ability of albumin to bind free fatty acids with high affinity. Therefore, Munro et al.' earlier made surfaces displaying straightchain 16- or 18-carbon alkyl groups in hopes that this might enhance the albumin affinity. Indeed, the addition of alkyl chains did increase the surface affinity for albumin during brief incubations'"-'2 and improved thromboresistance in an acute canine ex uivo e~periment.'~ Unfortunately, these materials do not have a high degree of selectivity for albumin binding and may, due to the highly hydrophobic nature of the surface, encourage the subsequent denaturation of adsorbed protein. Our work is based on another strategem for the selective binding of albumin. Albumin has the ability to bind and transport a large number of compounds in addition to free fatty acids. One of the physiological functions of serum albumin is the binding of bilirubin. Free bilirubin has toxic effects but binding to albumin renders it nontoxic. Albumin binds unconjugated bilirubin arising from heme breakdown and transports it to the liver for conjugation and secretion into the bile.14 Albumin has one very strong binding site for bilirubin, with an association constant of and two secondary sites having association constants of 1057.14-16 These binding sites are usually free; only one molecule of albumin in 15 contains bound bilirubin." The bilirubin-binding sites on albumin also have high affinity for a number of triazine dyes which have a structural similarity to bilirubin. Both bilirubin and the albumin-binding triazine dyes consist of planar aromatic ring systems and the positions of negatively charged groups can be superimposed in analogous position~.'~ In fact, a commonly used technique for purifying albumin involves the use of an immobilized triazine dye, Cibacron Blue F3GA. Passage of whole human plasma through a column of Cibacron Blue dye-Sepharose will remove approximately 98% of the albumin.IR Cibacron Blue dye has been covalently attached to a variety of support matrices including agarose, Sephacrylo, Sephadex, Sepharose, SpheronB, and Ultrogel@." In the present work, we have elected to use Cibacron Blue dye covalently attached to high-molecular-weight dextran (2 x lo6 daltons) ("Blue Dextran"; see Fig. 1).Dextran was employed as a hydrophilic "spacer" to ensure that albumin, once bound to the dye, would not undergo surface precipitation. Blue Dextran was therefore incorporated into implant-grade
ALBUMIN-BINDING SURFACES FOR IMPLANTABLE DEVICES
0w 2
443
Dextran
-0, H
Figure 1. Structure of blue dextran.
polyetherurethane using a cosolvation technique. The resulting surfacederivatized polymers have many of the desired characteristics, including selective and reversible binding of albumin. It is believed that these materials will also be immune to enzymatic attack; mammals do not produce dextranases and the dye is known to be enzymatically stable. MATERIALS A N D METHODS
Preparation of Blue Dextran-incorporated polyetherurethane
The incorporation of Blue Dextran (Sigma Chemical Co., St. Louis, MO) into Pellethane 2363-55D, a polyetherurethane commercially available from Dow Chemical Co., Midland, MI, was performed by dissolving 0.3 g of Blue Dextran in 1.5 mL of deionized (DI) water. This mixture was then added to 20 g of N,N-dimethylacetamide (DMAC)(Fisher Scientific Co., Pittsburg, PA). To this Blue Dextran/solvent mixture, 1.0 g of Pellethane 2363-55D was added. The mixture was shaken overnight to allow complete solvation of the polyurethane. The solvated Blue Dextran/polyurethane mixture was then cast on a Mylar@release sheet and dried overnight at 50°C to remove the solvent. It is important to note that it was not possible to make control preparations derivatized with dextran alone due to the impossibility of cosolvating dextran of molecular weight 2 x 10' with Pellethane 2363-55D. Adsorption of '251-albumin
Samples (1.0 cm') of derivatized and nonderivatized PU sheet were immersed in '251-humanalbumin (15.7 &mL; specific activity = 8.5 pCi/mg
444
KEOGH, VELANDER, A N D EATON
protein) (Mallinkrodt, St. Louis, MO) dissolved in PBS for 1 h. The samples were then rinsed by repetitive dipping ( ~ 1 0in ) very large volumes of isotonic phosphate buffered saline (PBS), pH 7.4. The samples were then counted in a scintillation counter.
Elution of albumin with NaSCN
The state of surface-adsorbed '251-albumin(which was not removed by PBS washings as described above) was determined by measuring the extent of protein elution by a solution of sodium thiocyanate, a chaotropic agent. Once again, samples (1.0 cm') were exposed to '251-albumin(15.7 pg/mL) for 1 h and repetitively rinsed in PBS as described above. Next, the samples were immersed in large volumes of a solution of 0.5M NaSCN and 50 mM Tris/HCl, pH 8.0 @ 25"C, with constant gentle agitation for 1 h and, finally, rinsed again in PBS. Control samples for determination of the amounts of albumin initially absorbed were exposed to '251-albuminand rinsed only in PBS. All samples were then counted in a scintillation counter. Elution of albumin with 1%SDS
Samples (1.0 cm') were incubated in 'Z51-albumin(15.7 ,ug/mL) for 12 h. All samples were then thoroughly rinsed in PBS. Some samples were counted immediately while others were incubated in 1%SDS for 1 h. These samples were again rinsed in PBS and counted in a scintillation counter.
Adsorption of Iz5I-fibrinogenon materials precoated with albumin
Samples (1.0 cm') were incubated in unlabeled albumin (15.7 pg/mL) for 1 h. Control samples were preincubated in PBS alone. The samples were then rinsed in PBS and incubated in '251-fibrinogen(15.7 pg/mL; specific activity = 115 pCi/mg protein) (Amersham,Arlington Heights, IL) for 1h. The samples were rinsed in PBS and then counted in a scintillation counter. Control samples were preincubated with PBS and subsequently with 1251-fibrinogen. Selective binding of albumin
The selectivity of albumin binding by derivatized and nonderivatized PU in the presence of other human serum proteins was determined using '251-albumin and increasing amounts of albumin-depleted human serum (ADHS). The ADHS was prepared by chromatographing 1.0 mL of whole human serum on an Affi-Gel Blue (Sigma) column (1 cm X 8 cm, bed volume = 6.3 mL, flow rate = 0.5 mL/min) with an elution buffer comprised of PBS. The serum collected from this column was free of detectable albumin as
ALBUMIN-BINDING SURFACES FOR IMPLANTABLE DEVICES
445
judged by SDS-polyacrylamide gel electrophoresis (see below). The resultant ADHS was then used in increasing amounts to competively block nonspecific binding of radiolabeled albumin. Samples (1.0 cm') of derivatized and nonderivatized PU film were immersed in solutions of 1251-humanalbumin (15.7 pg/mL) containing varying amounts of ADHS for 1 h. The samples were then rinsed in PBS three times and counted in a scintillation counter. Reversible binding of albumin
The reversibility of albumin binding to derivatized and nonderivatized PU was determined by measuring the extent of competitive elution of pre-bound '251-albuminby unlabeled human albumin. Samples (1.0 cm') were exposed to '251-albumin(15.7 pg/mL) for 1 h and then rinsed in PBS. Next, the samples were exposed to unlabeled human serum albumin (3 mg/mL) for 1 h and rinsed again with PBS. Controls were exposed to 1251-albuminonly. After the final rinsing the samples were counted in a scintillation counter. Site of albumin binding
The ligand specificity of albumin binding to derivatized and nonderivatized PU was investigated by using solution-phase Blue Dextran to competitively elute albumin bound to PU. Derivatized and nonderivatized PU samples (1.0 cm2) were immersed in '251-albumin (15.7 pg/mL) for 1 h. The samples were then rinsed in PBS and incubated in PBS containing soluble Blue Dextran (5 mg/mL) for 1 h. At the end of the incubation, the samples were rinsed in PBS and counted in a scintillation counter. Electrophoretic analysis of adsorbed proteins
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed on derivatized and nonderivatized PU to determine the types of proteins that adsorbed to the surface from whole, freshly shed unanticoagulated human blood. Samples (1.0 cm') of derivatized and nonderivatized PU film were placed in 12 x 75 mm glass culture tubes. The tubes were then filled with freshly drawn blood (
