- Email: [email protected]
Albumin adsorption on alksyi chain derivatized polyurettianes
Albuminadsorptionon al@1chain derivatizedpolyurethanes II. The effectof a&y1clminlength W.G. F’itt”, T.G. Graself and S.L. Cooper Department
of Chemical Engineering, University of Wisconsin, Madison, WI 53706,
USA
Presented at Biointeractions ‘87, Cambridge, UK in July 1987
linear alkyl chains containing 2,10 and 18 carbon atoms were grafted to 10% of the urethane nitrogens in a polyether-polyurethane. The polyurethane was synthesized from methylene his@-phenyl isocyanate),l,C butanediol, and polytetramethylene oxide of 1000 molecular weight in a molar ratio of 3/2/l. Fourier transform infrared spectroscopy and attenuated total reflectance optics were used to study the adsorption of 5.0 my/ml human serum albumin (HSA) at 37’C to the derivatized and non-derivatized polymers. Both delipidized HSA and HSA containing 6.5 mol stearic acid per mol of albumin were used to study the effect of chain length upon the initial adsorption rate, the total amount adsorbed in 1 h, and the desorption rate. The initial adsorption rates revealed that non-specific adsorption was similar upon all four polymers. An increase in initial adsorption rate upon the C-l 8 derivatized polymer was attributed to a specific binding interaction between the HSA and the grafted alkyl chains. The conformational stability of the HSA also affected the adsorption rate. The total amount adsorbed after 1 h decreased as the alkyl chain length increased from 2 to 18 methylene groups. The desorption rate decreased in magnitude as the alkyl chain length increased from C-2 to C-l 8. These results support a hypothesis that alkyl chain length influences the interaction between albumin and an alkylated polymer system. Keywords: Polymers, protein adsorption, polyurethane, albumin. alkyl derivatization, FTIR
The alkylation of blood contacting polymers with C-l 8 linear alkyl chains has been shown to reduce thrombus deposition at the blood/polymer interface’, ‘. Munro et al. proposed that the reduction of thrombus formation was a result of albumin adsorbing specifically and strongly to the alkyl chains. This specific adsorption was postulated to increase the amount of albumin on the surface and decrease the number of platelets on the surface available to aggregate into a thrombus. They supported this hypothesis by showing that albumin adsorption from single and binary protein solutions was increased when the polymers were derivatized with C-l 8 and C-l 6 alkyl chainszz3. In order to understand better the molecular mechanisms causing increased human serum albumin (HSA) adsorption to alkylated polymers, we have investigated the adsorption kinetics of HSA on C-2, C-IO, and C-l 8 alkylated polyurethanes using Fourier transform infrared (FTIR) spectroscopy. A previously published portion of this study has *Present address: Department of Chemical Engineering, Brigham Young University, Provo, Utah 84602, USA. tPresent address: W.R. Grace Co., Columbia, MD 21044, USA. Correspondence to Professor S.L. Cooper. 0 1988 36
Butterworth 8 Co (Publishers) Ltd. 0142-9612/88/010036-11$03.00 Biomaterials
1988, Vol9 January
shown that the addition of C-l 8 alkyl chains to a polyurethane increased the initial rate of delipidized HSA adsorption to the polymer, but did not increase the initial adsorption rate when stearic acid was added to the HSA prior to adsorption4. It was established that the increase in initial rate was due to an interaction between the alkyl chains of the polymer and the free fatty acid (FFA) binding site of the HSA. The amount of adsorbed albumin decreased and the adsorption activation energy increased when C-l 8 chains were incorporated into the polymer. This paper reports similar experiments in which the length of the alkyl chain was varied in order to determine the effect of alkyl chain length upon the HSA adsorption. The length of the carbon chain is postulated to influence the adsorption of HSA to the surface because the binding strength of albumin to the alkyl chains of the free fatty acids increases as the number of carbons in the FFA increases from 6 to 18, with singly unsaturated oleic acid being the most tightly bound5. For chains of greater length, the binding strength decreases slightly. The binding of free fatty acids by albumin has been extensively studied and can serve as a model for alkyl chain
Album/n adsorption: W.G. Pitt et al.
binding. Scatchard plots and other models applied to FFA binding data indicate that HSA has six or seven strong binding sites and several weak binding sites5’6. Each binding site pocket has a unique binding constant, and as binding occurs, the conformation of the protein changes to create new binding sites5. As more FFA is bound, the protein expands slightly in size, and becomes less susceptible to denaturation by heat, urea, or guanidine hydrochloride5,7,8. In general HSA binds any organic compound with at least 5 or 6 CH2 groups, whether or not there is an accompanying hydrophilic group such as a carboxylate. sulphonate, sulphate or hydroxylg. The two strongest binding sites appear to be sterically hindered from binding alkyl chains attached to bulky organic molecules, but they can bind alkyl chains attached toa solid polymer substrate”. The binding energy is derived primarily from hydrophobic interactions between the alkyl chain and the hydrophobic side chains of the amino acids forming the binding site ‘pockets’. There are also cationic residues near the mouth of the pockets which can interact with an anionic group such as a carboxylate or sulphate, although the contribution of this electrostatic or hydrogen bonding interaction is small compared to the hydrophobic interactions5. Recent studies in reversed-phase separation of proteins using chromatographic techniques provide insight into the effect of chain length upon the interaction of albumin with alkyl chains covalently attached to a solid support. Some studies have shown that albumin adsorption onto alkyl derivatized agarose increases as the carbon length increases from six to eight methylene groups”. In another study, albumin did not bind to a C-4 chain on a sepharose support, but did bind to longer chains12. However, an opposite observation that chain length has no effect on albumin retention or elution was reported for C-4, C-8 and C-l 8 alkyl chains on an unknown commercial chromatographic s~pp0r-t’~. In addition to the specific interaction between alkyl chains and the alkyl binding sites of albumin, non-specific binding of alkyl chains to hydrophobic patches on albumin or any other protein increases as the chain length increases from 1 to 8 carbons. Beyond this length, the hydrophobic chain appears to fold back on itself, which eliminates any further increase in non-specific hydrophobic interactions14. In an aqueous environment, it is doubtful that an alkyl chain of any length would project out from a polymer substrate since the free energy of mixing (AGmiX) would tend to force the alkyl chains flat against the surface, although transient extensions into the aqueous solution could occur due to thermal vibrations. This phenomenon is commonly observed in reversed-phase chromatography. As one changes the solvent from methanol to water, the pore volume is increased as the alkyl chains are pressed closer to the supporting substrate15, 16. On the other hand, an albumin molecule closely contacting a surface may displace enough water molecules so that the flattened alkyl chains could become free to diffuse into the protein and become bound. In addition to providing alkyl chain interaction with the albumin molecule, alkylation of the polyurethane could influence protein adsorption by disrupting the phase separation between the hard and soft domains of the polymer. Grasel et a/. have shown that C-2 and C-l 8 alkylation of the polyurethane used in this study increases the degree of phase mixing in the polymer by disrupting the hydrogen bonding within the hard segment domains and increasing the hydrogen bonding between the urethane - NH groups and the ether oxygen of the soft segment’. This could change the amount of polar hard segments residing at the
surface, or change the degree of hydrogen bonding in the polymer-water interfacial area. Sanada et al. have shown that changes in the hydrogen bonding properties of a surface can affect the amount and degree of denaturation of adsorbed albumin”. Sakurai studied protein adsorption on two phase block copolymers of polystyrene and poly(hydroxyethyl methacrylate) and found that proteins adsorbed differently on each phase”. However, they could not investigate the influence of the degree of phase mixing upon protein adsorption since their polymers were all very well phase separated. The domain size was also on the order of 100 nm, whereas the co-polyether-polyurethanes used in the present study have a domain size an order of magnitude smaller. One other factor which may influence protein adsorption is the conformational stability of the adsorbing protein. Some investigators have suggested that proteins experience a conformational change upon adsorption, and that the amount of protein adsorbed increases as the degree of conformational change increases4, lgm2’. In the present study, the adsorbing albumin could be made more conformationally stable by incorporation of alkyl chains into the binding sites5,7,8 . Whether the source of alkyl chains is from the polymer surface or from the stearic acid added to the albumin, the change in conformational stability may affect the adsorption process. As in the previous work, this study of the effect of chain length attempts to identify and differentiate between adsorption which occurs due to the binding of the HSA to the alkyl chains at the alkyl chain binding site and adsorption in which the alkyl chain does not bind to the alkyl binding site. The former type of adsorption will be called ‘specific adsorption’, and the latter will be called ‘non-specific adsorption’.
EXPERIMENTAL Polymer
synthesis
and preparation
The details of the synthesis of the base polyurethane and the alkylated derivatives have been described previously’ and are only briefly reviewed here. The base polyurethane was synthesized by reacting 3 mol of methylene bis(p-phenyl isocyanate) (MDI, Polysciences Inc.) with 1 mol of polytetramethylene oxide of 1000 molecular weight (Quaker Oats Co.). The resulting prepolymer was chain extended with 2 mol of 1,4-butanediol (Aldrich Chemical), precipitated in water, and dried under vacuum at 70°C for 1 wk. This base polymer will be referred to as PEU. A portion of this PEU was dissolved in N,N-dimethylacetamide (DMA) and alkylated using a 2-step bimolecular nucleophilic substitution reaction. Enough sodium hydride (NaH) was added to reduce 10% of the urethane hydrogen atoms. An n-alkyl iodide was reacted with the nitrogen anion to produce the derivatized urethane linkage. Three different alkylated polymers were synthesized using n-octadecyl iodide (C’8H371), n-decyl iodide (C’oH2’ I), and ethyl iodide (Cz H5 I). These polymers will be referred to as C-l 8-PEU, C-IO-PEU, and C-2-PEU respectively. The polymers were precipitated in a methanol-water solution and dried under vacuum at 70°C for 1 wk. All four polymers were extracted for 2 d using toluene in a Soxhlet extractor, dried under vacuum to remove residual toluene. For physical property characterization, the polymers were dissolved in DMA, evaporatively cast into sheets, and dried under vacuum at 70°C.
Biomaterials
1988, Vo/ 9 Jarwan/
37
Albumin adsorption: W.G. Pitt et al.
Transmission and attenuated total reflection (ATR) infrared spectra of the polymers were collected at 2 cm-’ resolution using a Nicolet 170SX FTIR as previously described’. Before comparing the transmission and ATR spectra, the ATR spectra were corrected for the depth of penetration of the evanescent wave22. differential scanning calorimetry (d.s.c.) thermograms were recorded from -150” to 230°C at a heating rate of 20”Wmin using a Perkin-Elmer DSC-2 equipped with a data processing unit that allowed subtraction of the background and normalization of the sample thermograms for sample weight. Dynamic mechanical analysis was accomplished with a microprocessor controlled Rheovibron DDV-II. Samples were cooled to - 15O”C, and data were taken at a test frequency of 1 10 Hz and a temperature rise rate of 2”C/min until sample failure. Room temperature uniaxial stress-strain measurements were taken with a lnstron table model tensile testing device at a crosshead speed of 0.5 in/min (33%/min). Samples were prepared using an ASTM D412-D standard die. Germanium internal reflection elements (IRE, 45” aperture, 50 X 20 X 3 mm, Harrick Sci., Ossinning, NY) were cleaned and dip-coated in a solution of 0.1 wt% polymer in DMA4. Each coated IRE was dried in a vacuum oven at 60°C for at least 4 h, and was stored under vacuum until use. All surfaces were prepared within 36 h of use. The thicknesses of the polymer coatings were estimated by infrared transmission spectroscopy and profilomet~ as described previously4. Table I lists the thickness of each film estimated from the two techniques. Underwater contact angles of air and octane in doubledistilled deionized water were determined on the dip-coated surfaces as previously describedz3.
Protein preparation Human serum albumin (HSA) which was free of globulins and fatty acids (delipidized) was purchased from Sigma Chemical (No. A-3782, lot no. 1 14C93 IO). Preparations of 35 mg/ml HSA were made in phosphate buffered saline (PBS, pH 7.35) containing 0.02% sodium azide. Half of this protein was lipidized as previously described4. The lipidized protein is referred to as ‘L-HSA’ while the delipidized protein is refered to as ‘D-HSA’. A titration of the free fatty acid in the D-HSA and L-HSA gave ratios of 0.042 f 0.002 and 6.46 3: 0.44 mol free fatty acid per mol albumin respectively. The proteins were passed through a P-100 (Bio-Rad) gel chromatography column (equilibrated with PBS) in order to remove multimers. The monomer fraction was concentrated to 30 mg/ml by dialysis against 60% polyethylene glycol. The concentrate was then dialysed against PBS for 2 d, snap Table 1 Underwater coated on germanium Polymer
PEU C-2-PEU c-10-PEU c-1 8-PEU
contact angles and thicknesses
Underwater air contact angle (“)”
Undewater octane contact angle (“)
Thickness by fTlR (nm)
32 f 4 39 f 5 33 + 4 50+3
32 f 5 44 + 7 49+3 73 + 5
30 17 33 33
aMean + standard deviation, n = 4. bNot determined.
38
of polymers
6~omate~als
1988, Vol9 January
dip-
Thickness by pfofiiomet~
frozen in a dry ice and ethanol bath, and stored at -70°C. The protein was quickly thawed at 40°C and diluted to 5.0 mg/ml within 48 h of use.
FTIR-ATR
measurement
of adsorption kinetics
Fourier transform infrared spectroscopy (FTI R) coupled with attentuated total reflectance optics (ATR) was used to measure the kinetics of the albumin adsorption to the various polymers at 37°C. The principles of FTIR-ATR and its application to protein adsorption studies are presented elsewhere22.24-26 and are not repeated in detail here. Briefly, infrared radiation reflects internally along the length of a germanium internal reflection element (IRE) in the flow cell. At each reflection, a decaying ‘evanescent’wave is established at the external surface of the IRE. Polymer, buffer and protein within this evanescent wave absorb infrared energy and attenuate the infrared radiation exiting the IRE toward the detector. This attentuation gives rise to an infrared spectrum which the FTIR detects and processes. By subtracting the spectra of polymer and buffer, one obtains the protein spectrum. This spectrum contains contributions from both the adsorbed and the non-adsorbed (or ‘bulk) protein within the evanescent wave, and these contributions can be separated and examined separatelg5. Data collection and processing. The polyurethanes were exposed to the albumin solutions in a flow cell contained inside a constant temperature compartment built into a Nicolet 17OSX FTIR equipped with a narrow band MCT detector (Nicolet, Madison, WI). The flow cell and data collection procedures were identical to those previously reported4. A protein concentration of 5.0 mg/ml and a wall shear rate of 200 s-’ were used in these experiments. Protein was allowed to adsorb for 60 min, at which time any desorption of protein was measured by switching the flow from protein to the buffer. After 30 min of desorption, flow is switched back to the protein for 5 min in order to measure the absorbance due to non-adsorbed protein in the bulk solution. The data processing procedures, including the subtraction of buffer and subtraction of non-adsorbs protein in the bulk solution, were as previously reported4. Quantitation of adsorbed protein. Three independent methods were used and compared to quantitate the amount of adsorbed protein4. These were: (1) using a calibration obtained from 1251-radiolabelled protein adsorbed on PEU and C-l 8-PEU in the flow cell: (2) using a calibration of the amide II area of D-HSA dried on a germanium IRE; and (3) using an internal standard for each experiment based upon the 1550 cm-’ absorbance of the non-adsorbed bulk protein. The three techniques gave very similar results, and the radiolabelled protein calibration was used to quantitate the results presented here.
RESULTS AND DISCUSSION Bulk polymer physical properties
25 NDb
Z
Grasel et al.’ have previously reported that C-2 and C-l 8 alkylation of the polymers decreased the degree of phase separation and changed the mechanical properties from that of the base polyurethane. The physical and thermal properties of the C-lo-PEU also indicate that the degree of phase separation in the bulk polymer is decreased similarly.
Albumin adsorption: W.G. Pitt et al.
Table 2 Polymer
PEU C-2-PEU C-l 0-PEU C-l 8-PEU
Infrared data of alkylated polyurethanes Non-bonded to bonded urethane carbonyl absorbance ratiob
NH to hard segment absorbance ratioa
0.85 0.72 0.73 0.77
Transmission spectra
ATR
0.57 0.63 0.63 0.58
0.52 0.37 0.61 0.62
aA (3340cm-‘)/A (1600cm-‘). bA(1730cm~‘)/A(1703cm~‘).
Table 3 Polymer
PEU C-2-PEU C-l 0-PEU C-l 8-PEU
Glass transition temperatures
of alkylated polyurethanes
Tg by differential scanning calorimetry (“C)
Tg by dynamic mechanical testing (“C)
Midpoint
Onset
Midpoint
Onset
--43 u-37 -m39 -39
-62 -58 -57 -60
-32 -23 -26 -28
-81 -68 -68 -72
Data from the transmission infrared spectra of the four materials show that the hydrogen bonded NH absorption band at 3340 cm-’ was reduced as expected (compared to the 1600 cm-’ benzene ring peak) when 10% of the urethane hydrogens were replaced byalkyl chains (see Tab/e 2). The lack of any detectable non-hydrogen bonded NH absorption at 3460 cm -’ in all spectra indicated that most of the urethane hydrogens were hydrogen bonded. The non-hydrogen bonded and hydrogen bonded urethane carbonyl bands appear at 1730 and 1703 cm-’ respectively. As the base PEU was alkylated, the ratio of nonbonded to bonded carbonyls increased in the transmission spectra, but did not show strong dependency upon chain length. Since NH is the only hydrogen bonding proton donor in the polymer, and since the number of NH groups which are hydrogen bonded appears to remain constant, an increase in the urethane carbonyl which is not hydrogen bonded (1730 cm-’ peak) must be accompanied by an increase in the amount of NH which is hydrogen bonded to the soft segment ether oxygen. Thus the increase in the 1730 cm-‘/l 703 cm-’ ratio of the three alkylated polymers indicates that there is a decrease in the phase separation of the polymer. The ATR spectra of the four polymers show non-bonded to bonded ratios which are similar to the transmission spectra. This is not unexpected since the 400 nm depth of penetration of the infrared radiation in this spectral region is not very surface sensitive. The soft segment glass transition temperature (T,) data obtained from the differential scanning calorimetry data are summarized in Tab/e 3. There was an increase in Tgof the soft segment in all of the polyurethanes compared to the Tgof pure PTMO of molecular weight 1000 (T, = -79”C)27. Although some rise in soft segment Tg may be caused by the loss of chain mobility resulting from the incorporation of the soft segment oligomer into the copolymer, the magnitude of the rise in Tg indicates that there is a significant solubilization of hard segments in the soft segment phase. The additional increase in the Tg of the alkylated polymers above that of the underivatized PEU suggests that the reduction in the ability of the hard segment to hydrogen bond with itself (due to alkylation) has further increased the hard segment solubility
in the PTMO phase. There appears to be no significant effect of alkyl chain length upon the Tg values of the derivatizec! polymers. Endotherms at high temperatures were also observed. Short-range order endotherms occur at approx. 70°C in all samples that were attributable to annealing effects. Endotherms at temperatures over 100°C are associated with the long-range order in the hard segment28. The temperatures of these transitions for the alkylated materials were found to be lower than for PEU, indicating that there was less hard segment ordering in the alkylated materials. The results of the dynamic mechanical testing presented in Tab/e 3 also suggestthat there is more phase mixing in the alkylated polymers. Figure 7 presents the dynamic mechanical data for the C-l 0-PEU polymer which is similar to the data for the other polymers. Using the E” (dynamic loss modulus) maximum in the - 100” to 0°C range as a measure of the soft segment glass transition temperature, the Tgof the three alkylated polymers are 4” to 9°C greater than that of the PEU. The Tg onset values in Tab/e 3 were defined as the minimum in E” in the - 100” to 0°C range. The differences between the Tg data measured by d.s.c. or dynamic mechanical testing is not unexpected since d.s.c. is a static measurement while the other is a dynamic measurement. The tensile properties of the polymers are shown in Figure 2 and summarized in Tab/e 4. Aikylation did not affect the elongation at failure, but it did reduce the ultimate stress of C-IO-PEU and C-l 8-PEU. The Young’s modulus and 100% secant modulus were reduced by the alkylation, although they did not show a strong dependence upon alkyl chain length. These reductions in the physical properties of the alkylated polymers are consistent with decreased phase separation and reduced hard segment ordering upon derivatization. To summarize the results of the mechanical and 10
9
1.0
0.10
T
r-”
D.O1 -100
-50
0
Temperature
50
100
150
(“C)
Figure 1 Dynamic mechanical data on the C- 10-PEU. Storage modulus E’ (--); loss modulus E” I-); and Tan 6 (- - -).
B/omaterials
1988, Vol 9 Januav
39
Albumin adsorption: W.G. Pirt et al.
only half as thick as the other films coated under similar conditions. The contact angles of both polymers dip-coated on germanium are less than those angles observed on thicker polymer coatings, indicating that these thin polymer films have slightly different surface energies’. This difference could be caused by disruption of the bulk polymer morphology in these very thin films. The small differences in surface energies between thick and thin films does not negate the significance of this work, however. In this study of molecular interactions between surface alkyl chains and albumin, the necessary experimental aspect of placing alkyl chains at the surface of the derivatized polymer is supported by the contact angle data. I
I
0
I
I
I
I
400
200
I
I
600
800
% Elongation Figure 2 Tensile properties of the polymers: C, C- 10-PEU; and D; C- 18-PEU.
A
PEU;
B, C-2-PEU;
thermal testing, the degree of phase mixing increases upon derivatization to a similar level in each of the alkylated polymers, although at these levels of substitution it does not appear to be dependent upon the chain length of the alkyl chain. Decreased hard domain cohesion and increased solubility of hard segments in the PTMO soft segment appear to be responsible for the modified physical properties of the polymers. One must note that the degree of phase separation in the bulk polymer may not necessarily correlate with phase separation at the surface of these polymers, although Li et al. have shown similar degrees of bulk and surface phase separation in polybutadiene-polyurethanes”.
Characterization
of dip-coated
polymer films
The underwater contact angles measured through the water phase are shown in Table 1. The octane contact angles increase as the length of the alkyl chain increases, indicating that the surfaces become more hydrophobic with increasing chain length. No significant trends in contact angles using the surface-water-air system were seen, except in the case of C-18-PEU where an increase in the contact angle was observed. Although one could invoke the geometric or harmonic mean approximations to derive a polymer surface energy from the contact angle data3’, the use of such approximations has been suggested to be questionable, and are not performed here31s3’ . This increase in apolar character of the surface as the chain length increases suggests that the alkyl chains may be present in the polymer-water inter-facial region of the coating. The thicknesses of the dip-coated films are slightly less than l/lOth of the 412 nm depth of penetration of the evanescent wave at 1547 wavenumbers”. The measurement of the film thickness is probably only accurate to k 20% in this thickness range. It is not known why the C-2-PEU film is Table 4 Polymer
PEU C-2-PEU C-l 0-PEU C-l 8-PEU
40
Tensile properties of alkylated polyurethanes Ultimate stress
Young’s modulus
100% Secant modulus
(MPa)
Percentage elongation at failure
(MPa)
(MPs)
29 28 19 19
700 750 700 700
30 21 14 15
7.7 5.9 5.0 5.5
Biomaterials
1988, Vol9 January
Initial adsorption rates The initial albumin adsorption rates provide information about the interaction between the protein and the surface at very low surface coverages before protein-protein interactions become large. When protein solution is introduced into the flowcell,thefluxof HSAtothesutfaceisinitiallythesamefor each surface, and the same number of protein molecules strike each surface. However, not all collisions between the surface and protein result in adsorption. Some proteins rebound from the surface while others remain adsorbed. As the fraction of collisions resulting in adsorption increases, the observed initial adsorption rate also increases proportionally. Thus the initial adsorption rate is a measure of the direct adhesive interaction between the protein and surface before the protein-protein interactions start to influence the adsorption process. The initial adsorption rate also shows the effect of blocking the alkyl binding site at short times before the stearic acid has time to diffuse out of the L-HSA. After the protein adsorption spectra were processed to provide kinetic adsorption profiles (amount adsorbed versus time), these data were analysed by fitting (using a non-linear least squares algorithm) the first 0.36 ,ug/cm’ of adsorbed HSA to an adsorption model described previouslyz3. This model allows for one or two layers of protein adsorption with rate constants k, and k2 respectively. The adjustable parameters in this model are k,, k2 and M, the mass of a single layer of protein. The rate equation is: Rate
= dr/dt = dl-,/dt
+ dI-,/dt
(1)
where dr,/dt dlY,/dt
= k,(l = k,(l
-T,/M) - r2/M).
In Equation (1), r, and lY2 are the protein concentrations (mass/area) in the first and second adsorbed layers, and r is the measured adsorption. This model does not require that the first layer of protein becompletelyfilled beforesubsequent layers begin to adsorb. Practical experience with this model has shown that the adjustable parameter k, is fairly independent of the values of k2 and M. The parameter k, is also the initial adsorption rate since it is the slope of the adsorption profile at a time of zero. The initial adsorption data for triplicate experiments using 5.0 mg/ml solutions of D-HSA and L-HSA onto C-2-PEU and C-l 0-PEU are shown in Figure 3. Figure 4 and Tab/e 5 present the initial adsorption rate (k, ) for adsorption on all four surfaces. The initial adsorption rates of L-HSA are lower than those of D-HSA and are similar on the four surfaces. The initial adsorption rates of D-HSA vary on each
Album/n adsorption: W.G. Pitt et al.
Table 5
Adsorption parameter@ PEU
C-2-PEU
C-lo-PEU
c- 18-PEU
37.9 k 6.5 5.7 f 4.2
21.8 i 2.7 5.9 + 3.2
32.2 + 3.8 5.0 t 2.9
0.503 0.437
i- 0.018 rt 0.01 1
0.377 0.403
+ 0.029 + 0.082
0.379 0.340
+ 0.004 + 0.013
0.182 0.189
i- 0.005 i 0.008
0.156 0.175
t 0.002 k 0.008
0.179 0.168
-k 0.003 + 0.006
?I 1.83 rt 1.40
17.32 20.80
+ 2.05 i 1.18
10.91 10.90
k 1.50 t 1.31
Initial adsorphon rates (ng cm-* s“) D-HSA L-HSA
18.6 rt 8.8 6.6 .t 3.4
Mass of HSA at 54 min of adsorption (&cm’) D-HSA L-HSA
0.477 0.389
?I 0.02 1 + 0.057
Mass of a single layer of WA (pg/cm2)b D-HSA L-HSA
0.239 0.181
z!z0.043 f 0.009
HSA desorption rate following 60 min of adsorption (pg cm~2s-1) D-HSA L-HSA
8.19 t 2.06 8.87 z!z2.49
28.57 28.47
‘Mean +_standard deviation, n = 3. ‘Determined from parameter fit of M, monolayer mass, in Equation (1)
Table 6
Statistical examination of the initial adsorption rates Polymer
Delipidizad HSA PEU
C-2-PEU
HSA containi~ C- 10-PEU
stearic acid
C-l 8-PEU
PEU
C-P-PEU
C- 10-PEU
C-l 8-PEU
++
-
-
_
--
-----
P
PEU
*
+++
.Gja
C-2-PEU
- -- -
*
=Q~ 0”
C-lo-PEU C-18-PEU _--~
= - -
+++ =
* ----
++++ *
__--_ -----
___ ----
__-------
F :E ‘im 8 2 I
= PEU ’ C-2-PEU -: C-IO-PEU i C-IS-PEU
+ + + ++
+++++ +++++ +++t+ ++t++
+++++ +++ +++++ +++++
+++++ ++++ +-I-+++ +++++
* = = =
= * = =
= z
=
=
* =
z? = zz *
Notation: The plus sign indicates that the mean of the initial adsorption rateon the surface indicated by the column label is greater than that ofthe row label at the following levels: + a level of srgnificance of 0.1; ++ a level of signrficance of 0.05; +++ a level of significance of 0.025; ++++ a level of significance of 0.0 1; + + + + i- a level of significance of 0.005. The minus sign indicates that the mean of the initial adsorption rate on the surface indicated by the column label is greater than that of the row label at the following levels: _ a level of significance of 0.1; -a level of significance of 0.05: --a level of significance of 0.025; ._--_ a level of sigmficance of 0.0 1; - - - a level of significance of 0.005. = Null hypothesis cannot be rejected at a level of significance of 0.1.
surface. Table 6 presents the results of a statistical test in which each set of experiments is compared against all others to find the significance of any differences in the initial adsorption rates. A one-tailed Student’s t-test assuming equal values for the average initial adsorption rates (the null hypothesis) and assuming unequal and unknown variances was used to determine the level of significance for rejecting the null hypothesis33. This analysis reveals interesting information regarding the effect of alkyl chain length on specific and non-specific adsorption. The lower right quadrant of Table 6 shows the effect of chain length upon non-specific adsorption since the alkyl binding sites of the L-HSA are blocked from interacting specifically with alkyl chains on the polymer. We assume that the presence of stearic acid in the HSA blocks the alkyl chains from binding upon first contact of the protein to the
polymer, but does not preclude the stearic acid from diffusing out or exchanging with alkyl chains after the initial adsorption has occurred. There are no significant differences in initial L-HSA adsorption rates on any of the polymers (see Figure 4) even though there are differences in surface energy and chemistry. Thus during the initial adsorption, there are no differences in non-specific adsorption due to differences in the surface chemistry. There are, however, differences in non-specific adsorption caused by the state of lipidation of the HSA. If adsorption was influenced by the surface only, one would expect that the adsorption rate of L-HSA and D-HSA would be the same on the PEU surface since alkyl chains are absent. One would also expect similar adsorption rates on the C-2-PEU since a C-2 chain has been shown to be too short to specifically bind to albumin. However, the data in Tables 5 and 6 show a
Biomaterials
1988, Vol9 January
4 1
Alburnjn adsorptian: W.G. Pitt et al.
0.4
I
,
I
,
,
,
1
[
I
-
.b
0.0
1
’
0
’
’
’
1
’
2
I
I
I
3
,
4
1 5
Min 0.4,
‘i E 0 9
,
,
,
(
,
Min ,
,
I
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0.0
0.0 0
1
2
3
4
5
Min
0
1
2
3
4
5
Min
Figure 3 Initial adsorption data for triplicate experiments using 5.0 mg/ml HSA at 37°C: a, D-I-ISA adsorption on C-2-PEU; b, L-HSA adsorption on C-2-PEU; c, D-HSA adsorption on C- ?U-PEU: and d, L-HSA adsorption on C- IO-PEU.
significant increase in the initial adsorption rate of D-HSA on PEU and C-2-PEU which can be attributed to a decrease in the conformational stability of the D-HSA molecule. The bottom left quadrant of Tab/e 6 shows that the initial adsorption rate of D-HSA is always significantly greater than L-HSA on any surface. A current hypothesis in protein
L
0
L PEU
D
L
C-2-PEU
D
L
C-lo-PEU
I
dt
C-18.PEU
Figure 4 Mean and standard deviation (n = 3) of the initial adsorption rates, or rate constant k,. D, D-H% L L-HSA.
42
Biomateriafs
1388, Vol9 January
adsorption is that proteins experience a change in conformation when they adsorb to a surface. This ~onformational change allows the protein to maximize its interaction with the surface by allowing the protein surface to flatten and establish mu,ltiple ‘footholds’ on the polymer surface, and by allowing buried non-polar groups to come in contact with hydrophobic portions of the surface. Decreasing the conformational stability of HSA by removing the stearic acid may allow the protein to change its conformation more easily and thus adsorb more avidly. A recent review lists several reports of conformational change upon adsorption34. This observation that the initial albumin adsorption is more dependent upon the state of lipidation than upon the surface chemistry is consistent with the hypotheses that adsorption is strongly influenced by the protein itselflg, “. The effect of alkyl chain length is best observed in the initial adsorption rates of the D-HSA to the various polymers. The initial adsorption rates on the underivatized PEU and the C-l 0-PEU are not significantly different, suggesting that the C-10 chains may not be long enough to participate in the binding of HSA at early adsorption times. It could be possible that the C-IO chains may be buried under the polymer surface in the aqueous environment although the underwater contact angle results suggest otherwise. The initial adsorption rate of D-HSA to C-l 8-PEU is significantly greater than to C- 10-PEU or to the underivatized base PEU. Since the results of the L-HSA adsorption indicated that the four surfaces had similar non-specific
Albumin adsorption: W.G. Pitt et al.
-b
20
0
0
60
40
40
20 Min
Min 0.5
0.5
c
F
F d
r ~
0.4
_
60
--*
c-
MC--
0.4
.
rr-
--
_/--
/-
,/--
-
0.3
” E 0
z
0.2
0.1
.
_..._.......
.
._.:
.
...’
0.3
0.2
0.1
-L
t
0.0
I
0
1
I
20
1
L
40
7
_...LI
0.0
0
60
Min
I
I
I
20
L
40
I 60
Min
Figure 5 Adsorption data for triplicate experiments using 5.0 mg/ml HSA adsorption on C- 1O-PEU: and d. L-HSA adsorption on CI 1O-PEG.
at 37°C:
adsorption properties, the increased initial adsorption rate is attributed to specific binding of the alkyl chain by the albumin molecule. An unexpected finding is that the C-2-PEU surface produces an initial adsorption rate as great as that of the C-l 8-PEU and significantly larger lp < 0.025) than that of the
c (I
a, D-HSA adsorption on C-2-PEU; 6, L-HSA adsorption on C-2-PEU; c, D-HSA
C-l 0-PEU and PEU surfaces. This is unexpected because the C-2 chain is too short to bind in the alkyl binding sites of HSA. Since the initial adsorption with the more conformationally stable L-HSA molecule showed no differences in nonspecific adsorption on the PEU and C-2-PEU, one would not expect large differences between these two surfaces when D-HSA is adsorbed. The influence upon the surface of the increased phase mixing of the bulk C-2-PEU compared to the underivatized PEU cannot be solely responsible for the increased initial adsorption rate because the C-IO-PEU would have shown a similar increase, since the mechanical and thermal testing indicated that the degree of bulk phase separation is similar in all of these alkylated materials.
11
First hour of adsorption
5 shows the adsorption kinetics for the first hour of adsorption onto the C-2-PEU and C-l 0-PEU surfaces. The solid line is the average adsorption, and the dotted lines mark the limits of one standard deviation from the mean. Tab/e 5 and Figure 6 present the average amount of protein adsorbed after 54 min. These values are between two and three times the amount of protein adsorbed in a single monolayer determined from the adsorption model (see Tab/e 5) and from adsorption isotherms35. Although there are no very large differences between the various experiments, the most adsorption was observed for the D-HSA on PEU and
Figure
~
D PEU
L
D
L
C-2-PEU
D
L
C-10.PEU
D
L
C-18.PEU
Figure 6 Mean and standard deviation {n = 3) of the amount of albumin adsorbed in 54 min. D, D-M& L L-H.%
Biomaterials
f988,
!Jof 9 Januatv
43
Albumin adsorption: W.G. Pitt et al.
0.6
I
a
1
I
0.6
I
I
0.5
: ti
x 0.4
Min
0.6
c
c’
’
’
’
0.5
’
?
I
I
I
I
d 0.5
~ L_
:
I
0.6
_L
%w
5n55
4 Ll54annc
3ftxe 444anan4454
0.4
,
I
0.4
1
10
0
I
,
‘
20
I
1
I
I
10
0
30
I
30
20
Min
Min
Figure 7 Albumin desorption following 1 h of adsorption of 5.0 mg/ml HSA at 37°C: a, D-HSA desorption from C-P-PEU; 6, L-HSA desorption from C-2-PEU; c. D-HSA desorption from C- IO-PEU; and d, L-HSA desorption from C- IO-PElJ.
C-2-PEU, while the least adsorption occurred with L-HSA on C-18-PEU. For both L-HSA and D-HSA, the amount adsorbed decreased as the length of the alkyl chain increased from C-2 to C-l 0 to C-l 8. Adsorption of L-HSA is significantly less than that of D-HSA on all surfaces except C-l 0-PEU, in which case there is no significant difference. These observations suggest that less protein adsorbs when the HSA binds alkyl chains either from the stearic acid or the C-l 0 and C-l 8 polymer chains. A possible explanation for this observation is that a less conformationally stable HSA molecule without alkyl chains in its binding sites may beable to squeeze into vacant sites of smaller area on the surface during adsorption, thus allowing more protein to pack onto the surface. It is also possible that the less conformationally stable albumin may be able to expose different amino acid residues at its surface when adsorbed, and thus change the nature of the protein-protein interactions which ultimately determine how closely the molecules adsorb to each other, as well as the extent of multilayer adsorption.
rates cannot be correlated with the initial adsorption rates or the amount adsorbed in 1 h. Contrary to the adsorption data, the desorption rates show very little or no dependence upon the state of lipidation of the HSA. However, there are significant differences between the desorption rates from each polymer. The desorption rates decrease as the length of the alkyl chain increases from C-2 to C-18. A possible explanation for this decrease in desorption rate is that the 30
I
T
I
I
I
Protein desorption Figure 7 shows the average desorption data (n = 3) from the C-2-PEU and C-l 0-PEU surfaces. The scatter at short desorption times is a result of the low signal to noise ratio during the rapid data collection. The desorption rate in each experiment appears to be constant over the 30 min of desorption. Table 5 and Figure 8 present the magnitude of the desorption rates from the four polymers. The desorption
44
Biamaterials
1988, Vol9 January
.l._
0
D
L
C-2-PEU Fi9ura 8 L. L-HSA
c;
D
L
C-IO-PEU
C-18-PEU
L
PEU
Albumin desorption rates following 1 h of adsorption. L?,D-HSR
Albumin adsorption: W.G. Pitt et al.
longer aikyl chains are binding the HSA more strongly to the polymer. However, since the 22 A length of the C-l 8 chain is half as long as the 44 A diam. of the HSA molecule, only the protein layer immediately adjacent to the polymer can bind the alkyl chains and show a desorption rate dependence upon chain length, If the protein-protein adhesion within the multilayer is weaker than the protein-polymer adhesion, then desorption would occur first from the upper layers of protein which are not in direct contact with the alkyl chains. In this case the length of the alkyl chain could still have an indirect effect upon desorption by changing the conformation or the susceptibility to denaturation of the protein which binds the alkyl chain. A change in the conformation of the protein in this layer could then change the protein-protein interactions between this layer and subsequent layers and affect the rate of desorption from the upper layers of protein. Comparison of desorption from the PEU and C-2-PEU polymers indicates that the degree of phase separation may be another factor which influences the desorption rate. Since the C-2 chain is too short to bind albumin, one would expect similar desorption rates on the two surfaces. However, the desorption rate from PEU is much less than C-2-PEU and C-1 0-PEU, and slightly less than the C-l 8-PEU. This result suggests that the decreased phase mixing of the PEU polymer could contribute to its decreased desorption rate.
adsorption rate. There was a large increase in the initial adsorption rate on C-2-PEU which could not be attributed to alkyl chain binding or to increased phase mixing in the polymer. The total amount of HSA adsorbed in the first hour decreased as the chain length increased from C-2 to C-l 8. The L-HSA adsorbed to a lesser extent than the D-HSA. These results suggests that less HSA is adsorbed when it is ~onformationally stabilized by the binding of alkyl chains from either stearic acid or from the alkyl chains grafted to the polymer. The desorption studies showed that as the alkyl chain length increased the desorption rate decreased in magnitude. Comparison of desorption rates from the C-2-PEU and PEU polymers suggested that the desorption rate increases as the phase mixing of the polymer increases. The desorption rate was not affected by the presence or absence of stearic acid in the HSA.
SUMMARY
REFERENCES
The grafting of C-2, C-IO and C-18 alkyl chains to the urethane nitrogen of a polyurethane synthesized from a 3/2/l mol ratio of MDI/BD/PTMO increased the phase mixing in the polymer and affected other physical and thermal properties of the polymer. The degree of phase mixing in the bulk polymer appeared to show little dependence upon the length of the grafted alkyl chain. As the alkyl chain length increased, the polymer surface became less polar as determined from underwater contact angles with octane. The technique of FTIR and ATR optics was employed to study the effect of chain length upon albumin adsorption and desorption kinetics. The binding of albumin to the alkyl chains was studied by comparing the adsorption of HSA without fatty acid in the alkyl chain binding sites, and the adsorption of HSA with 6.5 mol of stearic acid per mol of albumin blocking the alkyl binding sites. The initial adsorption rate on each polymer was obtained to determine the interaction between the HSA and polymer in the absence of the protein-protein interactions which dominate the adsorption at longer times and higher surface coverages. The initial adsorption rate was also useful in revealing the effect of blocking the alkyl binding site at short times before the stearic acid had time to diffuse from the L-HSA. The results showed that alkylation of the surface with the various chain lengths had no effect upon the nonspecific binding of HSA. When the alkyl binding sites were made available by removing the stearic acid, the initial adsorption rate increased on all polymers, including the base PEU which had nografted alkyl chainsThis result suggested that the increased susceptibility of the delipidized HSA toward denaturation increased the initial adsorption rate independent of any effects due to the presence of alkyl chains. There was also an additional increase in initial adsorption rate on the C-l 8-PEU which was attributed to alkyl chain binding. The small and perhaps insignificant increase in initial rate on C-l 0-PEU suggested that the C-f 0 chains may be too short to make a large contribution to the
ACKNOWLEOGEMENTS The authors wish to acknowledge partial support of this work through the National Institutes of Health, Grants HL-21001 and HL-24046, and Fellowship support for WGP from the W.R. Grace Co. and the Standard Oil Co.
1
2
3
4
5 6 7
8
9 10
11 12
13 14 15
16
17
Grasel, T.G.. Pierce, J.A. and Cooper, S.L.. Effects of aikyl grafting on surface properties and blood compatibility of polyurethane block copolymers, J. Biomed. Mater. Res. 1987, 21, 815-842 Munro, M.S., Eberhart, R.C., Make, N.J., Brmk, BE. and Fry. W.J., Thromboresistant alkyl derivatized polyurethanes, Amer. Sot. A& Intern. Organs J. 1983, 6. 65-75 Frautschi, JR., Munro, MS., Lloyd, D.R. and Eberhart, R.C., Alkyl derivatized cellulose acetate membranes wrth enhanced albumin affinity, Trans. Amer. Sot. Artif. Intern. Organs 1983,23,242-245 Pitt, W.G. and Cooper, S.L., Albumin adsorptton on alkyl chain derivatlzed polyurethanes: I. The effect of C-l 8 alkylation. J. Biomed. Mater. Res. 1987, (subm~~ed) Spector, Arthur A., Fatty acid binding to plasma albumin. J. Lipid Res. 1975, 16. 165-179 Goodman, D.S., The interaction of human serum albumln with longcharn fatty acid anions, J. Am. Chem. Sot. 1958,80. 3892-3898 Boyer, P.D., Ballou, G.A. and Luck, J.M.. The combination of fatty acids and related compounds with serum albumin. II. Stabilization against urea and guanidine denaturation, J. ffiol. Chem. 1946,162,199-208 Boyer, P.D.. Lum, F.G.. Ballou, G.A., Luck, J.M. and Rice, R.G.. The combination of fatty acids and related compounds with serum albumin. I. Stabilization against heat denaturatlon, J. Viol. Chem, 1946,162. 181-198 Eendit, E.G., Infrared absorption of tyrosine side chains in proteins, Biopolymers 1967, 5, 525-533 Uzhinova. L.D., Gracheva, N.A. and Plate, N.A., A study of the binding of serum albumin with cetyl methacrylate In solution by the fluorescent probe method, Soviet 1 Bioorganic Chem. 1980, 6, 877-880 Mohr, P. and Pommerening. K.. Affinity Chromaforgraphy Practical and ~eo~tica/~pects, Marcel Dekker, New York, 1985, pp. 225-241 Er-el, Z., Zaidenzaig,Y. and Shaltiel, S., Hydrocarbon-coated sepharoses. Use in the purification of glycogen phosphorylase, Biocbem. Biophys. Res. Comm. 1972,49, 383-390 Steffensen, Fi. and Anderson, Jr, J.J., Reversed-phase separation of proteins, Biochromatog~aphy 1987, 2, 85-89 Lowe, C.R. and Dean, P.D.G.. Affinity Chromafography, Wiley, London, 1974, pp 23-34 McCormrck, R.M. and Karger, 8.L.. Drstributton phenomena of mobilephase components and determmation of dead volume in reversedphase liquid chromatography, Anal. Chem. 1980, 52, 2249-2257 Mockel, H.J. and Freyholdt, T.. Column dead-time for a bonded Cl 8 phase with pure alcohols as eluents, Ch~omatographia 1983, 17. 2 15-220 Sanada, T., Ito. Y., Sisido, M. and Imantshi. Y., Adsorptton of plasma
Biomaterials
7988, Vol9 Januaw
45
Albumin adsorption: W.G. Pitt et al.
proteins to the derivatives of polyaminoetherurethaneurea: The effect of hydr~en-ending property of the material surface. J. Biomed. Mater. Res. 1986,20, 1179-l 195 18
19
20
21 22 23
24
25
46
Sakurai.Y.. Akaike,T., Kataoka, K andOkano,T., interfacial phenomena in biomaterials chemistry in Biomedical Polymers (Ed. E.P. Goldberg and A. Nakajima), Academic Press, New York, 1980 Van Dulm, P. and Norde, W., The adsorption of human plasma albumin on solid surfaces. with special attention to the kinetic aspects,./. Co//. Interface Sci 1983.91, 248-255 Norde. W., Macffitchie, F., Nowicka, G. and Lyklema, J., Protein adsorption at solid-liquid interfaces: Reversibility and conformation aspects, J. Coil. interface Sci. 1986, 112, 447-456 Andrade, J.D. and Hledy, V., Plasma protein adsorption -the big twelve, Anna/s New York Acad. Sci. 1987 (submitted) Harrick, N.J.. fnternal Reflection Spectroscopy, John Wiley. New York, 1967 Pitt, W.G. and Cooper, S.L., FTIR-ATR studies of the effect of shear rate upon albumin adsorption onto polyurethaneurea, Biomateriak 1986,7.340-347 Gendreau, R.M., Leininger. RI.. Winters, S. and Jakobsen, R.J.. Fourier transform infrared spectroscopy for protein-surface studies, in Biomaterials: Interfacial Phenomena and Applications, (Eds S.L. Cooper and N.A. Peppas). ACS, Washington, DC, 1982 Chittur, K.K., Fink, D.J., Leininger, R.I. and Hutson, T.B., FTiR/ATR studies of protein adsorption in flowing systems: Approaches for bulk correction and compositional analysis of adsorbed and bulk proteins in
Biomaterials
1988, Vat 9 January
26 27 28 29
30 31
32 33 34
35
mixtures. J. Coil. interface Sci. 1986, 1 f 1, 419-433 Griffiths, P.R. and deHaseth, J.A., Fourier Transform infrared Spectrometry. Wiley, New York, 1986 Wang, C.B. and Cooper, S.L.. Morphology and properties of segmented polyurethanes, Macromolecules 1983, 16, 775-786 Seymour, R.W. and Cooper, S.L., Thermal analysis of polyurethane block copolymers, Macromolecules 1973, 6. 48-53 Li. C., Goodman, S.L., Albrecht, R.M. and Cooper, S.L., Morphology characterization of segmented ~iyb~~ien~~lyurethane elastomers, ~acromolecoies 1987, (submits) Andrade, J.D., Ma, S.M., King, R.N. and Gregonis, D.E., Contact angles at the solid-water interface, J. Colt. lnferface Sci. 1979,72,488-494 Spelt, J.K.. Absolom, D.R. and Neumann, A.W., Solid surface tension: The interpretation of contact angles by the equation of state approach and the theory of surface tension components, Langmuir 1986, 2. 620-625 Fowkes. F.M., Role of acid-base interfacial bonding in adhesion, J. Adhesion Sci. Tech. 1987, 1, 7-27 Walpole, R.E. and Myers, R.H., Probabiliiyand Statistics for Engineers and Scientists, MacMillan, New York, 1978 Andrade. J.D. and Hlady, V., Protein adsorption and materials bi~ompatibiii~: A tutorial review and suggested hypotheses, A& Polymer Sci 1986, 79, l-63 Young, B.R., Pitt, W.G. and Cooper, S.L., Protein adsorption on polymeric biomaterials: I. Adsorption isotherms, J. Colloid /nterface Sci. 1987, (submitted)
of Chemical Engineering, University of Wisconsin, Madison, WI 53706,
USA
Presented at Biointeractions ‘87, Cambridge, UK in July 1987
linear alkyl chains containing 2,10 and 18 carbon atoms were grafted to 10% of the urethane nitrogens in a polyether-polyurethane. The polyurethane was synthesized from methylene his@-phenyl isocyanate),l,C butanediol, and polytetramethylene oxide of 1000 molecular weight in a molar ratio of 3/2/l. Fourier transform infrared spectroscopy and attenuated total reflectance optics were used to study the adsorption of 5.0 my/ml human serum albumin (HSA) at 37’C to the derivatized and non-derivatized polymers. Both delipidized HSA and HSA containing 6.5 mol stearic acid per mol of albumin were used to study the effect of chain length upon the initial adsorption rate, the total amount adsorbed in 1 h, and the desorption rate. The initial adsorption rates revealed that non-specific adsorption was similar upon all four polymers. An increase in initial adsorption rate upon the C-l 8 derivatized polymer was attributed to a specific binding interaction between the HSA and the grafted alkyl chains. The conformational stability of the HSA also affected the adsorption rate. The total amount adsorbed after 1 h decreased as the alkyl chain length increased from 2 to 18 methylene groups. The desorption rate decreased in magnitude as the alkyl chain length increased from C-2 to C-l 8. These results support a hypothesis that alkyl chain length influences the interaction between albumin and an alkylated polymer system. Keywords: Polymers, protein adsorption, polyurethane, albumin. alkyl derivatization, FTIR
The alkylation of blood contacting polymers with C-l 8 linear alkyl chains has been shown to reduce thrombus deposition at the blood/polymer interface’, ‘. Munro et al. proposed that the reduction of thrombus formation was a result of albumin adsorbing specifically and strongly to the alkyl chains. This specific adsorption was postulated to increase the amount of albumin on the surface and decrease the number of platelets on the surface available to aggregate into a thrombus. They supported this hypothesis by showing that albumin adsorption from single and binary protein solutions was increased when the polymers were derivatized with C-l 8 and C-l 6 alkyl chainszz3. In order to understand better the molecular mechanisms causing increased human serum albumin (HSA) adsorption to alkylated polymers, we have investigated the adsorption kinetics of HSA on C-2, C-IO, and C-l 8 alkylated polyurethanes using Fourier transform infrared (FTIR) spectroscopy. A previously published portion of this study has *Present address: Department of Chemical Engineering, Brigham Young University, Provo, Utah 84602, USA. tPresent address: W.R. Grace Co., Columbia, MD 21044, USA. Correspondence to Professor S.L. Cooper. 0 1988 36
Butterworth 8 Co (Publishers) Ltd. 0142-9612/88/010036-11$03.00 Biomaterials
1988, Vol9 January
shown that the addition of C-l 8 alkyl chains to a polyurethane increased the initial rate of delipidized HSA adsorption to the polymer, but did not increase the initial adsorption rate when stearic acid was added to the HSA prior to adsorption4. It was established that the increase in initial rate was due to an interaction between the alkyl chains of the polymer and the free fatty acid (FFA) binding site of the HSA. The amount of adsorbed albumin decreased and the adsorption activation energy increased when C-l 8 chains were incorporated into the polymer. This paper reports similar experiments in which the length of the alkyl chain was varied in order to determine the effect of alkyl chain length upon the HSA adsorption. The length of the carbon chain is postulated to influence the adsorption of HSA to the surface because the binding strength of albumin to the alkyl chains of the free fatty acids increases as the number of carbons in the FFA increases from 6 to 18, with singly unsaturated oleic acid being the most tightly bound5. For chains of greater length, the binding strength decreases slightly. The binding of free fatty acids by albumin has been extensively studied and can serve as a model for alkyl chain
Album/n adsorption: W.G. Pitt et al.
binding. Scatchard plots and other models applied to FFA binding data indicate that HSA has six or seven strong binding sites and several weak binding sites5’6. Each binding site pocket has a unique binding constant, and as binding occurs, the conformation of the protein changes to create new binding sites5. As more FFA is bound, the protein expands slightly in size, and becomes less susceptible to denaturation by heat, urea, or guanidine hydrochloride5,7,8. In general HSA binds any organic compound with at least 5 or 6 CH2 groups, whether or not there is an accompanying hydrophilic group such as a carboxylate. sulphonate, sulphate or hydroxylg. The two strongest binding sites appear to be sterically hindered from binding alkyl chains attached to bulky organic molecules, but they can bind alkyl chains attached toa solid polymer substrate”. The binding energy is derived primarily from hydrophobic interactions between the alkyl chain and the hydrophobic side chains of the amino acids forming the binding site ‘pockets’. There are also cationic residues near the mouth of the pockets which can interact with an anionic group such as a carboxylate or sulphate, although the contribution of this electrostatic or hydrogen bonding interaction is small compared to the hydrophobic interactions5. Recent studies in reversed-phase separation of proteins using chromatographic techniques provide insight into the effect of chain length upon the interaction of albumin with alkyl chains covalently attached to a solid support. Some studies have shown that albumin adsorption onto alkyl derivatized agarose increases as the carbon length increases from six to eight methylene groups”. In another study, albumin did not bind to a C-4 chain on a sepharose support, but did bind to longer chains12. However, an opposite observation that chain length has no effect on albumin retention or elution was reported for C-4, C-8 and C-l 8 alkyl chains on an unknown commercial chromatographic s~pp0r-t’~. In addition to the specific interaction between alkyl chains and the alkyl binding sites of albumin, non-specific binding of alkyl chains to hydrophobic patches on albumin or any other protein increases as the chain length increases from 1 to 8 carbons. Beyond this length, the hydrophobic chain appears to fold back on itself, which eliminates any further increase in non-specific hydrophobic interactions14. In an aqueous environment, it is doubtful that an alkyl chain of any length would project out from a polymer substrate since the free energy of mixing (AGmiX) would tend to force the alkyl chains flat against the surface, although transient extensions into the aqueous solution could occur due to thermal vibrations. This phenomenon is commonly observed in reversed-phase chromatography. As one changes the solvent from methanol to water, the pore volume is increased as the alkyl chains are pressed closer to the supporting substrate15, 16. On the other hand, an albumin molecule closely contacting a surface may displace enough water molecules so that the flattened alkyl chains could become free to diffuse into the protein and become bound. In addition to providing alkyl chain interaction with the albumin molecule, alkylation of the polyurethane could influence protein adsorption by disrupting the phase separation between the hard and soft domains of the polymer. Grasel et a/. have shown that C-2 and C-l 8 alkylation of the polyurethane used in this study increases the degree of phase mixing in the polymer by disrupting the hydrogen bonding within the hard segment domains and increasing the hydrogen bonding between the urethane - NH groups and the ether oxygen of the soft segment’. This could change the amount of polar hard segments residing at the
surface, or change the degree of hydrogen bonding in the polymer-water interfacial area. Sanada et al. have shown that changes in the hydrogen bonding properties of a surface can affect the amount and degree of denaturation of adsorbed albumin”. Sakurai studied protein adsorption on two phase block copolymers of polystyrene and poly(hydroxyethyl methacrylate) and found that proteins adsorbed differently on each phase”. However, they could not investigate the influence of the degree of phase mixing upon protein adsorption since their polymers were all very well phase separated. The domain size was also on the order of 100 nm, whereas the co-polyether-polyurethanes used in the present study have a domain size an order of magnitude smaller. One other factor which may influence protein adsorption is the conformational stability of the adsorbing protein. Some investigators have suggested that proteins experience a conformational change upon adsorption, and that the amount of protein adsorbed increases as the degree of conformational change increases4, lgm2’. In the present study, the adsorbing albumin could be made more conformationally stable by incorporation of alkyl chains into the binding sites5,7,8 . Whether the source of alkyl chains is from the polymer surface or from the stearic acid added to the albumin, the change in conformational stability may affect the adsorption process. As in the previous work, this study of the effect of chain length attempts to identify and differentiate between adsorption which occurs due to the binding of the HSA to the alkyl chains at the alkyl chain binding site and adsorption in which the alkyl chain does not bind to the alkyl binding site. The former type of adsorption will be called ‘specific adsorption’, and the latter will be called ‘non-specific adsorption’.
EXPERIMENTAL Polymer
synthesis
and preparation
The details of the synthesis of the base polyurethane and the alkylated derivatives have been described previously’ and are only briefly reviewed here. The base polyurethane was synthesized by reacting 3 mol of methylene bis(p-phenyl isocyanate) (MDI, Polysciences Inc.) with 1 mol of polytetramethylene oxide of 1000 molecular weight (Quaker Oats Co.). The resulting prepolymer was chain extended with 2 mol of 1,4-butanediol (Aldrich Chemical), precipitated in water, and dried under vacuum at 70°C for 1 wk. This base polymer will be referred to as PEU. A portion of this PEU was dissolved in N,N-dimethylacetamide (DMA) and alkylated using a 2-step bimolecular nucleophilic substitution reaction. Enough sodium hydride (NaH) was added to reduce 10% of the urethane hydrogen atoms. An n-alkyl iodide was reacted with the nitrogen anion to produce the derivatized urethane linkage. Three different alkylated polymers were synthesized using n-octadecyl iodide (C’8H371), n-decyl iodide (C’oH2’ I), and ethyl iodide (Cz H5 I). These polymers will be referred to as C-l 8-PEU, C-IO-PEU, and C-2-PEU respectively. The polymers were precipitated in a methanol-water solution and dried under vacuum at 70°C for 1 wk. All four polymers were extracted for 2 d using toluene in a Soxhlet extractor, dried under vacuum to remove residual toluene. For physical property characterization, the polymers were dissolved in DMA, evaporatively cast into sheets, and dried under vacuum at 70°C.
Biomaterials
1988, Vo/ 9 Jarwan/
37
Albumin adsorption: W.G. Pitt et al.
Transmission and attenuated total reflection (ATR) infrared spectra of the polymers were collected at 2 cm-’ resolution using a Nicolet 170SX FTIR as previously described’. Before comparing the transmission and ATR spectra, the ATR spectra were corrected for the depth of penetration of the evanescent wave22. differential scanning calorimetry (d.s.c.) thermograms were recorded from -150” to 230°C at a heating rate of 20”Wmin using a Perkin-Elmer DSC-2 equipped with a data processing unit that allowed subtraction of the background and normalization of the sample thermograms for sample weight. Dynamic mechanical analysis was accomplished with a microprocessor controlled Rheovibron DDV-II. Samples were cooled to - 15O”C, and data were taken at a test frequency of 1 10 Hz and a temperature rise rate of 2”C/min until sample failure. Room temperature uniaxial stress-strain measurements were taken with a lnstron table model tensile testing device at a crosshead speed of 0.5 in/min (33%/min). Samples were prepared using an ASTM D412-D standard die. Germanium internal reflection elements (IRE, 45” aperture, 50 X 20 X 3 mm, Harrick Sci., Ossinning, NY) were cleaned and dip-coated in a solution of 0.1 wt% polymer in DMA4. Each coated IRE was dried in a vacuum oven at 60°C for at least 4 h, and was stored under vacuum until use. All surfaces were prepared within 36 h of use. The thicknesses of the polymer coatings were estimated by infrared transmission spectroscopy and profilomet~ as described previously4. Table I lists the thickness of each film estimated from the two techniques. Underwater contact angles of air and octane in doubledistilled deionized water were determined on the dip-coated surfaces as previously describedz3.
Protein preparation Human serum albumin (HSA) which was free of globulins and fatty acids (delipidized) was purchased from Sigma Chemical (No. A-3782, lot no. 1 14C93 IO). Preparations of 35 mg/ml HSA were made in phosphate buffered saline (PBS, pH 7.35) containing 0.02% sodium azide. Half of this protein was lipidized as previously described4. The lipidized protein is referred to as ‘L-HSA’ while the delipidized protein is refered to as ‘D-HSA’. A titration of the free fatty acid in the D-HSA and L-HSA gave ratios of 0.042 f 0.002 and 6.46 3: 0.44 mol free fatty acid per mol albumin respectively. The proteins were passed through a P-100 (Bio-Rad) gel chromatography column (equilibrated with PBS) in order to remove multimers. The monomer fraction was concentrated to 30 mg/ml by dialysis against 60% polyethylene glycol. The concentrate was then dialysed against PBS for 2 d, snap Table 1 Underwater coated on germanium Polymer
PEU C-2-PEU c-10-PEU c-1 8-PEU
contact angles and thicknesses
Underwater air contact angle (“)”
Undewater octane contact angle (“)
Thickness by fTlR (nm)
32 f 4 39 f 5 33 + 4 50+3
32 f 5 44 + 7 49+3 73 + 5
30 17 33 33
aMean + standard deviation, n = 4. bNot determined.
38
of polymers
6~omate~als
1988, Vol9 January
dip-
Thickness by pfofiiomet~
frozen in a dry ice and ethanol bath, and stored at -70°C. The protein was quickly thawed at 40°C and diluted to 5.0 mg/ml within 48 h of use.
FTIR-ATR
measurement
of adsorption kinetics
Fourier transform infrared spectroscopy (FTI R) coupled with attentuated total reflectance optics (ATR) was used to measure the kinetics of the albumin adsorption to the various polymers at 37°C. The principles of FTIR-ATR and its application to protein adsorption studies are presented elsewhere22.24-26 and are not repeated in detail here. Briefly, infrared radiation reflects internally along the length of a germanium internal reflection element (IRE) in the flow cell. At each reflection, a decaying ‘evanescent’wave is established at the external surface of the IRE. Polymer, buffer and protein within this evanescent wave absorb infrared energy and attenuate the infrared radiation exiting the IRE toward the detector. This attentuation gives rise to an infrared spectrum which the FTIR detects and processes. By subtracting the spectra of polymer and buffer, one obtains the protein spectrum. This spectrum contains contributions from both the adsorbed and the non-adsorbed (or ‘bulk) protein within the evanescent wave, and these contributions can be separated and examined separatelg5. Data collection and processing. The polyurethanes were exposed to the albumin solutions in a flow cell contained inside a constant temperature compartment built into a Nicolet 17OSX FTIR equipped with a narrow band MCT detector (Nicolet, Madison, WI). The flow cell and data collection procedures were identical to those previously reported4. A protein concentration of 5.0 mg/ml and a wall shear rate of 200 s-’ were used in these experiments. Protein was allowed to adsorb for 60 min, at which time any desorption of protein was measured by switching the flow from protein to the buffer. After 30 min of desorption, flow is switched back to the protein for 5 min in order to measure the absorbance due to non-adsorbed protein in the bulk solution. The data processing procedures, including the subtraction of buffer and subtraction of non-adsorbs protein in the bulk solution, were as previously reported4. Quantitation of adsorbed protein. Three independent methods were used and compared to quantitate the amount of adsorbed protein4. These were: (1) using a calibration obtained from 1251-radiolabelled protein adsorbed on PEU and C-l 8-PEU in the flow cell: (2) using a calibration of the amide II area of D-HSA dried on a germanium IRE; and (3) using an internal standard for each experiment based upon the 1550 cm-’ absorbance of the non-adsorbed bulk protein. The three techniques gave very similar results, and the radiolabelled protein calibration was used to quantitate the results presented here.
RESULTS AND DISCUSSION Bulk polymer physical properties
25 NDb
Z
Grasel et al.’ have previously reported that C-2 and C-l 8 alkylation of the polymers decreased the degree of phase separation and changed the mechanical properties from that of the base polyurethane. The physical and thermal properties of the C-lo-PEU also indicate that the degree of phase separation in the bulk polymer is decreased similarly.
Albumin adsorption: W.G. Pitt et al.
Table 2 Polymer
PEU C-2-PEU C-l 0-PEU C-l 8-PEU
Infrared data of alkylated polyurethanes Non-bonded to bonded urethane carbonyl absorbance ratiob
NH to hard segment absorbance ratioa
0.85 0.72 0.73 0.77
Transmission spectra
ATR
0.57 0.63 0.63 0.58
0.52 0.37 0.61 0.62
aA (3340cm-‘)/A (1600cm-‘). bA(1730cm~‘)/A(1703cm~‘).
Table 3 Polymer
PEU C-2-PEU C-l 0-PEU C-l 8-PEU
Glass transition temperatures
of alkylated polyurethanes
Tg by differential scanning calorimetry (“C)
Tg by dynamic mechanical testing (“C)
Midpoint
Onset
Midpoint
Onset
--43 u-37 -m39 -39
-62 -58 -57 -60
-32 -23 -26 -28
-81 -68 -68 -72
Data from the transmission infrared spectra of the four materials show that the hydrogen bonded NH absorption band at 3340 cm-’ was reduced as expected (compared to the 1600 cm-’ benzene ring peak) when 10% of the urethane hydrogens were replaced byalkyl chains (see Tab/e 2). The lack of any detectable non-hydrogen bonded NH absorption at 3460 cm -’ in all spectra indicated that most of the urethane hydrogens were hydrogen bonded. The non-hydrogen bonded and hydrogen bonded urethane carbonyl bands appear at 1730 and 1703 cm-’ respectively. As the base PEU was alkylated, the ratio of nonbonded to bonded carbonyls increased in the transmission spectra, but did not show strong dependency upon chain length. Since NH is the only hydrogen bonding proton donor in the polymer, and since the number of NH groups which are hydrogen bonded appears to remain constant, an increase in the urethane carbonyl which is not hydrogen bonded (1730 cm-’ peak) must be accompanied by an increase in the amount of NH which is hydrogen bonded to the soft segment ether oxygen. Thus the increase in the 1730 cm-‘/l 703 cm-’ ratio of the three alkylated polymers indicates that there is a decrease in the phase separation of the polymer. The ATR spectra of the four polymers show non-bonded to bonded ratios which are similar to the transmission spectra. This is not unexpected since the 400 nm depth of penetration of the infrared radiation in this spectral region is not very surface sensitive. The soft segment glass transition temperature (T,) data obtained from the differential scanning calorimetry data are summarized in Tab/e 3. There was an increase in Tgof the soft segment in all of the polyurethanes compared to the Tgof pure PTMO of molecular weight 1000 (T, = -79”C)27. Although some rise in soft segment Tg may be caused by the loss of chain mobility resulting from the incorporation of the soft segment oligomer into the copolymer, the magnitude of the rise in Tg indicates that there is a significant solubilization of hard segments in the soft segment phase. The additional increase in the Tg of the alkylated polymers above that of the underivatized PEU suggests that the reduction in the ability of the hard segment to hydrogen bond with itself (due to alkylation) has further increased the hard segment solubility
in the PTMO phase. There appears to be no significant effect of alkyl chain length upon the Tg values of the derivatizec! polymers. Endotherms at high temperatures were also observed. Short-range order endotherms occur at approx. 70°C in all samples that were attributable to annealing effects. Endotherms at temperatures over 100°C are associated with the long-range order in the hard segment28. The temperatures of these transitions for the alkylated materials were found to be lower than for PEU, indicating that there was less hard segment ordering in the alkylated materials. The results of the dynamic mechanical testing presented in Tab/e 3 also suggestthat there is more phase mixing in the alkylated polymers. Figure 7 presents the dynamic mechanical data for the C-l 0-PEU polymer which is similar to the data for the other polymers. Using the E” (dynamic loss modulus) maximum in the - 100” to 0°C range as a measure of the soft segment glass transition temperature, the Tgof the three alkylated polymers are 4” to 9°C greater than that of the PEU. The Tg onset values in Tab/e 3 were defined as the minimum in E” in the - 100” to 0°C range. The differences between the Tg data measured by d.s.c. or dynamic mechanical testing is not unexpected since d.s.c. is a static measurement while the other is a dynamic measurement. The tensile properties of the polymers are shown in Figure 2 and summarized in Tab/e 4. Aikylation did not affect the elongation at failure, but it did reduce the ultimate stress of C-IO-PEU and C-l 8-PEU. The Young’s modulus and 100% secant modulus were reduced by the alkylation, although they did not show a strong dependence upon alkyl chain length. These reductions in the physical properties of the alkylated polymers are consistent with decreased phase separation and reduced hard segment ordering upon derivatization. To summarize the results of the mechanical and 10
9
1.0
0.10
T
r-”
D.O1 -100
-50
0
Temperature
50
100
150
(“C)
Figure 1 Dynamic mechanical data on the C- 10-PEU. Storage modulus E’ (--); loss modulus E” I-); and Tan 6 (- - -).
B/omaterials
1988, Vol 9 Januav
39
Albumin adsorption: W.G. Pirt et al.
only half as thick as the other films coated under similar conditions. The contact angles of both polymers dip-coated on germanium are less than those angles observed on thicker polymer coatings, indicating that these thin polymer films have slightly different surface energies’. This difference could be caused by disruption of the bulk polymer morphology in these very thin films. The small differences in surface energies between thick and thin films does not negate the significance of this work, however. In this study of molecular interactions between surface alkyl chains and albumin, the necessary experimental aspect of placing alkyl chains at the surface of the derivatized polymer is supported by the contact angle data. I
I
0
I
I
I
I
400
200
I
I
600
800
% Elongation Figure 2 Tensile properties of the polymers: C, C- 10-PEU; and D; C- 18-PEU.
A
PEU;
B, C-2-PEU;
thermal testing, the degree of phase mixing increases upon derivatization to a similar level in each of the alkylated polymers, although at these levels of substitution it does not appear to be dependent upon the chain length of the alkyl chain. Decreased hard domain cohesion and increased solubility of hard segments in the PTMO soft segment appear to be responsible for the modified physical properties of the polymers. One must note that the degree of phase separation in the bulk polymer may not necessarily correlate with phase separation at the surface of these polymers, although Li et al. have shown similar degrees of bulk and surface phase separation in polybutadiene-polyurethanes”.
Characterization
of dip-coated
polymer films
The underwater contact angles measured through the water phase are shown in Table 1. The octane contact angles increase as the length of the alkyl chain increases, indicating that the surfaces become more hydrophobic with increasing chain length. No significant trends in contact angles using the surface-water-air system were seen, except in the case of C-18-PEU where an increase in the contact angle was observed. Although one could invoke the geometric or harmonic mean approximations to derive a polymer surface energy from the contact angle data3’, the use of such approximations has been suggested to be questionable, and are not performed here31s3’ . This increase in apolar character of the surface as the chain length increases suggests that the alkyl chains may be present in the polymer-water inter-facial region of the coating. The thicknesses of the dip-coated films are slightly less than l/lOth of the 412 nm depth of penetration of the evanescent wave at 1547 wavenumbers”. The measurement of the film thickness is probably only accurate to k 20% in this thickness range. It is not known why the C-2-PEU film is Table 4 Polymer
PEU C-2-PEU C-l 0-PEU C-l 8-PEU
40
Tensile properties of alkylated polyurethanes Ultimate stress
Young’s modulus
100% Secant modulus
(MPa)
Percentage elongation at failure
(MPa)
(MPs)
29 28 19 19
700 750 700 700
30 21 14 15
7.7 5.9 5.0 5.5
Biomaterials
1988, Vol9 January
Initial adsorption rates The initial albumin adsorption rates provide information about the interaction between the protein and the surface at very low surface coverages before protein-protein interactions become large. When protein solution is introduced into the flowcell,thefluxof HSAtothesutfaceisinitiallythesamefor each surface, and the same number of protein molecules strike each surface. However, not all collisions between the surface and protein result in adsorption. Some proteins rebound from the surface while others remain adsorbed. As the fraction of collisions resulting in adsorption increases, the observed initial adsorption rate also increases proportionally. Thus the initial adsorption rate is a measure of the direct adhesive interaction between the protein and surface before the protein-protein interactions start to influence the adsorption process. The initial adsorption rate also shows the effect of blocking the alkyl binding site at short times before the stearic acid has time to diffuse out of the L-HSA. After the protein adsorption spectra were processed to provide kinetic adsorption profiles (amount adsorbed versus time), these data were analysed by fitting (using a non-linear least squares algorithm) the first 0.36 ,ug/cm’ of adsorbed HSA to an adsorption model described previouslyz3. This model allows for one or two layers of protein adsorption with rate constants k, and k2 respectively. The adjustable parameters in this model are k,, k2 and M, the mass of a single layer of protein. The rate equation is: Rate
= dr/dt = dl-,/dt
+ dI-,/dt
(1)
where dr,/dt dlY,/dt
= k,(l = k,(l
-T,/M) - r2/M).
In Equation (1), r, and lY2 are the protein concentrations (mass/area) in the first and second adsorbed layers, and r is the measured adsorption. This model does not require that the first layer of protein becompletelyfilled beforesubsequent layers begin to adsorb. Practical experience with this model has shown that the adjustable parameter k, is fairly independent of the values of k2 and M. The parameter k, is also the initial adsorption rate since it is the slope of the adsorption profile at a time of zero. The initial adsorption data for triplicate experiments using 5.0 mg/ml solutions of D-HSA and L-HSA onto C-2-PEU and C-l 0-PEU are shown in Figure 3. Figure 4 and Tab/e 5 present the initial adsorption rate (k, ) for adsorption on all four surfaces. The initial adsorption rates of L-HSA are lower than those of D-HSA and are similar on the four surfaces. The initial adsorption rates of D-HSA vary on each
Album/n adsorption: W.G. Pitt et al.
Table 5
Adsorption parameter@ PEU
C-2-PEU
C-lo-PEU
c- 18-PEU
37.9 k 6.5 5.7 f 4.2
21.8 i 2.7 5.9 + 3.2
32.2 + 3.8 5.0 t 2.9
0.503 0.437
i- 0.018 rt 0.01 1
0.377 0.403
+ 0.029 + 0.082
0.379 0.340
+ 0.004 + 0.013
0.182 0.189
i- 0.005 i 0.008
0.156 0.175
t 0.002 k 0.008
0.179 0.168
-k 0.003 + 0.006
?I 1.83 rt 1.40
17.32 20.80
+ 2.05 i 1.18
10.91 10.90
k 1.50 t 1.31
Initial adsorphon rates (ng cm-* s“) D-HSA L-HSA
18.6 rt 8.8 6.6 .t 3.4
Mass of HSA at 54 min of adsorption (&cm’) D-HSA L-HSA
0.477 0.389
?I 0.02 1 + 0.057
Mass of a single layer of WA (pg/cm2)b D-HSA L-HSA
0.239 0.181
z!z0.043 f 0.009
HSA desorption rate following 60 min of adsorption (pg cm~2s-1) D-HSA L-HSA
8.19 t 2.06 8.87 z!z2.49
28.57 28.47
‘Mean +_standard deviation, n = 3. ‘Determined from parameter fit of M, monolayer mass, in Equation (1)
Table 6
Statistical examination of the initial adsorption rates Polymer
Delipidizad HSA PEU
C-2-PEU
HSA containi~ C- 10-PEU
stearic acid
C-l 8-PEU
PEU
C-P-PEU
C- 10-PEU
C-l 8-PEU
++
-
-
_
--
-----
P
PEU
*
+++
.Gja
C-2-PEU
- -- -
*
=Q~ 0”
C-lo-PEU C-18-PEU _--~
= - -
+++ =
* ----
++++ *
__--_ -----
___ ----
__-------
F :E ‘im 8 2 I
= PEU ’ C-2-PEU -: C-IO-PEU i C-IS-PEU
+ + + ++
+++++ +++++ +++t+ ++t++
+++++ +++ +++++ +++++
+++++ ++++ +-I-+++ +++++
* = = =
= * = =
= z
=
=
* =
z? = zz *
Notation: The plus sign indicates that the mean of the initial adsorption rateon the surface indicated by the column label is greater than that ofthe row label at the following levels: + a level of srgnificance of 0.1; ++ a level of signrficance of 0.05; +++ a level of significance of 0.025; ++++ a level of significance of 0.0 1; + + + + i- a level of significance of 0.005. The minus sign indicates that the mean of the initial adsorption rate on the surface indicated by the column label is greater than that of the row label at the following levels: _ a level of significance of 0.1; -a level of significance of 0.05: --a level of significance of 0.025; ._--_ a level of sigmficance of 0.0 1; - - - a level of significance of 0.005. = Null hypothesis cannot be rejected at a level of significance of 0.1.
surface. Table 6 presents the results of a statistical test in which each set of experiments is compared against all others to find the significance of any differences in the initial adsorption rates. A one-tailed Student’s t-test assuming equal values for the average initial adsorption rates (the null hypothesis) and assuming unequal and unknown variances was used to determine the level of significance for rejecting the null hypothesis33. This analysis reveals interesting information regarding the effect of alkyl chain length on specific and non-specific adsorption. The lower right quadrant of Table 6 shows the effect of chain length upon non-specific adsorption since the alkyl binding sites of the L-HSA are blocked from interacting specifically with alkyl chains on the polymer. We assume that the presence of stearic acid in the HSA blocks the alkyl chains from binding upon first contact of the protein to the
polymer, but does not preclude the stearic acid from diffusing out or exchanging with alkyl chains after the initial adsorption has occurred. There are no significant differences in initial L-HSA adsorption rates on any of the polymers (see Figure 4) even though there are differences in surface energy and chemistry. Thus during the initial adsorption, there are no differences in non-specific adsorption due to differences in the surface chemistry. There are, however, differences in non-specific adsorption caused by the state of lipidation of the HSA. If adsorption was influenced by the surface only, one would expect that the adsorption rate of L-HSA and D-HSA would be the same on the PEU surface since alkyl chains are absent. One would also expect similar adsorption rates on the C-2-PEU since a C-2 chain has been shown to be too short to specifically bind to albumin. However, the data in Tables 5 and 6 show a
Biomaterials
1988, Vol9 January
4 1
Alburnjn adsorptian: W.G. Pitt et al.
0.4
I
,
I
,
,
,
1
[
I
-
.b
0.0
1
’
0
’
’
’
1
’
2
I
I
I
3
,
4
1 5
Min 0.4,
‘i E 0 9
,
,
,
(
,
Min ,
,
I
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0.0
0.0 0
1
2
3
4
5
Min
0
1
2
3
4
5
Min
Figure 3 Initial adsorption data for triplicate experiments using 5.0 mg/ml HSA at 37°C: a, D-I-ISA adsorption on C-2-PEU; b, L-HSA adsorption on C-2-PEU; c, D-HSA adsorption on C- ?U-PEU: and d, L-HSA adsorption on C- IO-PEU.
significant increase in the initial adsorption rate of D-HSA on PEU and C-2-PEU which can be attributed to a decrease in the conformational stability of the D-HSA molecule. The bottom left quadrant of Tab/e 6 shows that the initial adsorption rate of D-HSA is always significantly greater than L-HSA on any surface. A current hypothesis in protein
L
0
L PEU
D
L
C-2-PEU
D
L
C-lo-PEU
I
dt
C-18.PEU
Figure 4 Mean and standard deviation (n = 3) of the initial adsorption rates, or rate constant k,. D, D-H% L L-HSA.
42
Biomateriafs
1388, Vol9 January
adsorption is that proteins experience a change in conformation when they adsorb to a surface. This ~onformational change allows the protein to maximize its interaction with the surface by allowing the protein surface to flatten and establish mu,ltiple ‘footholds’ on the polymer surface, and by allowing buried non-polar groups to come in contact with hydrophobic portions of the surface. Decreasing the conformational stability of HSA by removing the stearic acid may allow the protein to change its conformation more easily and thus adsorb more avidly. A recent review lists several reports of conformational change upon adsorption34. This observation that the initial albumin adsorption is more dependent upon the state of lipidation than upon the surface chemistry is consistent with the hypotheses that adsorption is strongly influenced by the protein itselflg, “. The effect of alkyl chain length is best observed in the initial adsorption rates of the D-HSA to the various polymers. The initial adsorption rates on the underivatized PEU and the C-l 0-PEU are not significantly different, suggesting that the C-10 chains may not be long enough to participate in the binding of HSA at early adsorption times. It could be possible that the C-IO chains may be buried under the polymer surface in the aqueous environment although the underwater contact angle results suggest otherwise. The initial adsorption rate of D-HSA to C-l 8-PEU is significantly greater than to C- 10-PEU or to the underivatized base PEU. Since the results of the L-HSA adsorption indicated that the four surfaces had similar non-specific
Albumin adsorption: W.G. Pitt et al.
-b
20
0
0
60
40
40
20 Min
Min 0.5
0.5
c
F
F d
r ~
0.4
_
60
--*
c-
MC--
0.4
.
rr-
--
_/--
/-
,/--
-
0.3
” E 0
z
0.2
0.1
.
_..._.......
.
._.:
.
...’
0.3
0.2
0.1
-L
t
0.0
I
0
1
I
20
1
L
40
7
_...LI
0.0
0
60
Min
I
I
I
20
L
40
I 60
Min
Figure 5 Adsorption data for triplicate experiments using 5.0 mg/ml HSA adsorption on C- 1O-PEU: and d. L-HSA adsorption on CI 1O-PEG.
at 37°C:
adsorption properties, the increased initial adsorption rate is attributed to specific binding of the alkyl chain by the albumin molecule. An unexpected finding is that the C-2-PEU surface produces an initial adsorption rate as great as that of the C-l 8-PEU and significantly larger lp < 0.025) than that of the
c (I
a, D-HSA adsorption on C-2-PEU; 6, L-HSA adsorption on C-2-PEU; c, D-HSA
C-l 0-PEU and PEU surfaces. This is unexpected because the C-2 chain is too short to bind in the alkyl binding sites of HSA. Since the initial adsorption with the more conformationally stable L-HSA molecule showed no differences in nonspecific adsorption on the PEU and C-2-PEU, one would not expect large differences between these two surfaces when D-HSA is adsorbed. The influence upon the surface of the increased phase mixing of the bulk C-2-PEU compared to the underivatized PEU cannot be solely responsible for the increased initial adsorption rate because the C-IO-PEU would have shown a similar increase, since the mechanical and thermal testing indicated that the degree of bulk phase separation is similar in all of these alkylated materials.
11
First hour of adsorption
5 shows the adsorption kinetics for the first hour of adsorption onto the C-2-PEU and C-l 0-PEU surfaces. The solid line is the average adsorption, and the dotted lines mark the limits of one standard deviation from the mean. Tab/e 5 and Figure 6 present the average amount of protein adsorbed after 54 min. These values are between two and three times the amount of protein adsorbed in a single monolayer determined from the adsorption model (see Tab/e 5) and from adsorption isotherms35. Although there are no very large differences between the various experiments, the most adsorption was observed for the D-HSA on PEU and
Figure
~
D PEU
L
D
L
C-2-PEU
D
L
C-10.PEU
D
L
C-18.PEU
Figure 6 Mean and standard deviation {n = 3) of the amount of albumin adsorbed in 54 min. D, D-M& L L-H.%
Biomaterials
f988,
!Jof 9 Januatv
43
Albumin adsorption: W.G. Pitt et al.
0.6
I
a
1
I
0.6
I
I
0.5
: ti
x 0.4
Min
0.6
c
c’
’
’
’
0.5
’
?
I
I
I
I
d 0.5
~ L_
:
I
0.6
_L
%w
5n55
4 Ll54annc
3ftxe 444anan4454
0.4
,
I
0.4
1
10
0
I
,
‘
20
I
1
I
I
10
0
30
I
30
20
Min
Min
Figure 7 Albumin desorption following 1 h of adsorption of 5.0 mg/ml HSA at 37°C: a, D-HSA desorption from C-P-PEU; 6, L-HSA desorption from C-2-PEU; c. D-HSA desorption from C- IO-PEU; and d, L-HSA desorption from C- IO-PElJ.
C-2-PEU, while the least adsorption occurred with L-HSA on C-18-PEU. For both L-HSA and D-HSA, the amount adsorbed decreased as the length of the alkyl chain increased from C-2 to C-l 0 to C-l 8. Adsorption of L-HSA is significantly less than that of D-HSA on all surfaces except C-l 0-PEU, in which case there is no significant difference. These observations suggest that less protein adsorbs when the HSA binds alkyl chains either from the stearic acid or the C-l 0 and C-l 8 polymer chains. A possible explanation for this observation is that a less conformationally stable HSA molecule without alkyl chains in its binding sites may beable to squeeze into vacant sites of smaller area on the surface during adsorption, thus allowing more protein to pack onto the surface. It is also possible that the less conformationally stable albumin may be able to expose different amino acid residues at its surface when adsorbed, and thus change the nature of the protein-protein interactions which ultimately determine how closely the molecules adsorb to each other, as well as the extent of multilayer adsorption.
rates cannot be correlated with the initial adsorption rates or the amount adsorbed in 1 h. Contrary to the adsorption data, the desorption rates show very little or no dependence upon the state of lipidation of the HSA. However, there are significant differences between the desorption rates from each polymer. The desorption rates decrease as the length of the alkyl chain increases from C-2 to C-18. A possible explanation for this decrease in desorption rate is that the 30
I
T
I
I
I
Protein desorption Figure 7 shows the average desorption data (n = 3) from the C-2-PEU and C-l 0-PEU surfaces. The scatter at short desorption times is a result of the low signal to noise ratio during the rapid data collection. The desorption rate in each experiment appears to be constant over the 30 min of desorption. Table 5 and Figure 8 present the magnitude of the desorption rates from the four polymers. The desorption
44
Biamaterials
1988, Vol9 January
.l._
0
D
L
C-2-PEU Fi9ura 8 L. L-HSA
c;
D
L
C-IO-PEU
C-18-PEU
L
PEU
Albumin desorption rates following 1 h of adsorption. L?,D-HSR
Albumin adsorption: W.G. Pitt et al.
longer aikyl chains are binding the HSA more strongly to the polymer. However, since the 22 A length of the C-l 8 chain is half as long as the 44 A diam. of the HSA molecule, only the protein layer immediately adjacent to the polymer can bind the alkyl chains and show a desorption rate dependence upon chain length, If the protein-protein adhesion within the multilayer is weaker than the protein-polymer adhesion, then desorption would occur first from the upper layers of protein which are not in direct contact with the alkyl chains. In this case the length of the alkyl chain could still have an indirect effect upon desorption by changing the conformation or the susceptibility to denaturation of the protein which binds the alkyl chain. A change in the conformation of the protein in this layer could then change the protein-protein interactions between this layer and subsequent layers and affect the rate of desorption from the upper layers of protein. Comparison of desorption from the PEU and C-2-PEU polymers indicates that the degree of phase separation may be another factor which influences the desorption rate. Since the C-2 chain is too short to bind albumin, one would expect similar desorption rates on the two surfaces. However, the desorption rate from PEU is much less than C-2-PEU and C-1 0-PEU, and slightly less than the C-l 8-PEU. This result suggests that the decreased phase mixing of the PEU polymer could contribute to its decreased desorption rate.
adsorption rate. There was a large increase in the initial adsorption rate on C-2-PEU which could not be attributed to alkyl chain binding or to increased phase mixing in the polymer. The total amount of HSA adsorbed in the first hour decreased as the chain length increased from C-2 to C-l 8. The L-HSA adsorbed to a lesser extent than the D-HSA. These results suggests that less HSA is adsorbed when it is ~onformationally stabilized by the binding of alkyl chains from either stearic acid or from the alkyl chains grafted to the polymer. The desorption studies showed that as the alkyl chain length increased the desorption rate decreased in magnitude. Comparison of desorption rates from the C-2-PEU and PEU polymers suggested that the desorption rate increases as the phase mixing of the polymer increases. The desorption rate was not affected by the presence or absence of stearic acid in the HSA.
SUMMARY
REFERENCES
The grafting of C-2, C-IO and C-18 alkyl chains to the urethane nitrogen of a polyurethane synthesized from a 3/2/l mol ratio of MDI/BD/PTMO increased the phase mixing in the polymer and affected other physical and thermal properties of the polymer. The degree of phase mixing in the bulk polymer appeared to show little dependence upon the length of the grafted alkyl chain. As the alkyl chain length increased, the polymer surface became less polar as determined from underwater contact angles with octane. The technique of FTIR and ATR optics was employed to study the effect of chain length upon albumin adsorption and desorption kinetics. The binding of albumin to the alkyl chains was studied by comparing the adsorption of HSA without fatty acid in the alkyl chain binding sites, and the adsorption of HSA with 6.5 mol of stearic acid per mol of albumin blocking the alkyl binding sites. The initial adsorption rate on each polymer was obtained to determine the interaction between the HSA and polymer in the absence of the protein-protein interactions which dominate the adsorption at longer times and higher surface coverages. The initial adsorption rate was also useful in revealing the effect of blocking the alkyl binding site at short times before the stearic acid had time to diffuse from the L-HSA. The results showed that alkylation of the surface with the various chain lengths had no effect upon the nonspecific binding of HSA. When the alkyl binding sites were made available by removing the stearic acid, the initial adsorption rate increased on all polymers, including the base PEU which had nografted alkyl chainsThis result suggested that the increased susceptibility of the delipidized HSA toward denaturation increased the initial adsorption rate independent of any effects due to the presence of alkyl chains. There was also an additional increase in initial adsorption rate on the C-l 8-PEU which was attributed to alkyl chain binding. The small and perhaps insignificant increase in initial rate on C-l 0-PEU suggested that the C-f 0 chains may be too short to make a large contribution to the
ACKNOWLEOGEMENTS The authors wish to acknowledge partial support of this work through the National Institutes of Health, Grants HL-21001 and HL-24046, and Fellowship support for WGP from the W.R. Grace Co. and the Standard Oil Co.
1
2
3
4
5 6 7
8
9 10
11 12
13 14 15
16
17
Grasel, T.G.. Pierce, J.A. and Cooper, S.L.. Effects of aikyl grafting on surface properties and blood compatibility of polyurethane block copolymers, J. Biomed. Mater. Res. 1987, 21, 815-842 Munro, M.S., Eberhart, R.C., Make, N.J., Brmk, BE. and Fry. W.J., Thromboresistant alkyl derivatized polyurethanes, Amer. Sot. A& Intern. Organs J. 1983, 6. 65-75 Frautschi, JR., Munro, MS., Lloyd, D.R. and Eberhart, R.C., Alkyl derivatized cellulose acetate membranes wrth enhanced albumin affinity, Trans. Amer. Sot. Artif. Intern. Organs 1983,23,242-245 Pitt, W.G. and Cooper, S.L., Albumin adsorptton on alkyl chain derivatlzed polyurethanes: I. The effect of C-l 8 alkylation. J. Biomed. Mater. Res. 1987, (subm~~ed) Spector, Arthur A., Fatty acid binding to plasma albumin. J. Lipid Res. 1975, 16. 165-179 Goodman, D.S., The interaction of human serum albumln with longcharn fatty acid anions, J. Am. Chem. Sot. 1958,80. 3892-3898 Boyer, P.D., Ballou, G.A. and Luck, J.M.. The combination of fatty acids and related compounds with serum albumin. II. Stabilization against urea and guanidine denaturation, J. ffiol. Chem. 1946,162,199-208 Boyer, P.D.. Lum, F.G.. Ballou, G.A., Luck, J.M. and Rice, R.G.. The combination of fatty acids and related compounds with serum albumin. I. Stabilization against heat denaturatlon, J. Viol. Chem, 1946,162. 181-198 Eendit, E.G., Infrared absorption of tyrosine side chains in proteins, Biopolymers 1967, 5, 525-533 Uzhinova. L.D., Gracheva, N.A. and Plate, N.A., A study of the binding of serum albumin with cetyl methacrylate In solution by the fluorescent probe method, Soviet 1 Bioorganic Chem. 1980, 6, 877-880 Mohr, P. and Pommerening. K.. Affinity Chromaforgraphy Practical and ~eo~tica/~pects, Marcel Dekker, New York, 1985, pp. 225-241 Er-el, Z., Zaidenzaig,Y. and Shaltiel, S., Hydrocarbon-coated sepharoses. Use in the purification of glycogen phosphorylase, Biocbem. Biophys. Res. Comm. 1972,49, 383-390 Steffensen, Fi. and Anderson, Jr, J.J., Reversed-phase separation of proteins, Biochromatog~aphy 1987, 2, 85-89 Lowe, C.R. and Dean, P.D.G.. Affinity Chromafography, Wiley, London, 1974, pp 23-34 McCormrck, R.M. and Karger, 8.L.. Drstributton phenomena of mobilephase components and determmation of dead volume in reversedphase liquid chromatography, Anal. Chem. 1980, 52, 2249-2257 Mockel, H.J. and Freyholdt, T.. Column dead-time for a bonded Cl 8 phase with pure alcohols as eluents, Ch~omatographia 1983, 17. 2 15-220 Sanada, T., Ito. Y., Sisido, M. and Imantshi. Y., Adsorptton of plasma
Biomaterials
7988, Vol9 Januaw
45
Albumin adsorption: W.G. Pitt et al.
proteins to the derivatives of polyaminoetherurethaneurea: The effect of hydr~en-ending property of the material surface. J. Biomed. Mater. Res. 1986,20, 1179-l 195 18
19
20
21 22 23
24
25
46
Sakurai.Y.. Akaike,T., Kataoka, K andOkano,T., interfacial phenomena in biomaterials chemistry in Biomedical Polymers (Ed. E.P. Goldberg and A. Nakajima), Academic Press, New York, 1980 Van Dulm, P. and Norde, W., The adsorption of human plasma albumin on solid surfaces. with special attention to the kinetic aspects,./. Co//. Interface Sci 1983.91, 248-255 Norde. W., Macffitchie, F., Nowicka, G. and Lyklema, J., Protein adsorption at solid-liquid interfaces: Reversibility and conformation aspects, J. Coil. interface Sci. 1986, 112, 447-456 Andrade, J.D. and Hledy, V., Plasma protein adsorption -the big twelve, Anna/s New York Acad. Sci. 1987 (submitted) Harrick, N.J.. fnternal Reflection Spectroscopy, John Wiley. New York, 1967 Pitt, W.G. and Cooper, S.L., FTIR-ATR studies of the effect of shear rate upon albumin adsorption onto polyurethaneurea, Biomateriak 1986,7.340-347 Gendreau, R.M., Leininger. RI.. Winters, S. and Jakobsen, R.J.. Fourier transform infrared spectroscopy for protein-surface studies, in Biomaterials: Interfacial Phenomena and Applications, (Eds S.L. Cooper and N.A. Peppas). ACS, Washington, DC, 1982 Chittur, K.K., Fink, D.J., Leininger, R.I. and Hutson, T.B., FTiR/ATR studies of protein adsorption in flowing systems: Approaches for bulk correction and compositional analysis of adsorbed and bulk proteins in
Biomaterials
1988, Vat 9 January
26 27 28 29
30 31
32 33 34
35
mixtures. J. Coil. interface Sci. 1986, 1 f 1, 419-433 Griffiths, P.R. and deHaseth, J.A., Fourier Transform infrared Spectrometry. Wiley, New York, 1986 Wang, C.B. and Cooper, S.L.. Morphology and properties of segmented polyurethanes, Macromolecules 1983, 16, 775-786 Seymour, R.W. and Cooper, S.L., Thermal analysis of polyurethane block copolymers, Macromolecules 1973, 6. 48-53 Li. C., Goodman, S.L., Albrecht, R.M. and Cooper, S.L., Morphology characterization of segmented ~iyb~~ien~~lyurethane elastomers, ~acromolecoies 1987, (submits) Andrade, J.D., Ma, S.M., King, R.N. and Gregonis, D.E., Contact angles at the solid-water interface, J. Colt. lnferface Sci. 1979,72,488-494 Spelt, J.K.. Absolom, D.R. and Neumann, A.W., Solid surface tension: The interpretation of contact angles by the equation of state approach and the theory of surface tension components, Langmuir 1986, 2. 620-625 Fowkes. F.M., Role of acid-base interfacial bonding in adhesion, J. Adhesion Sci. Tech. 1987, 1, 7-27 Walpole, R.E. and Myers, R.H., Probabiliiyand Statistics for Engineers and Scientists, MacMillan, New York, 1978 Andrade. J.D. and Hlady, V., Protein adsorption and materials bi~ompatibiii~: A tutorial review and suggested hypotheses, A& Polymer Sci 1986, 79, l-63 Young, B.R., Pitt, W.G. and Cooper, S.L., Protein adsorption on polymeric biomaterials: I. Adsorption isotherms, J. Colloid /nterface Sci. 1987, (submitted)