- Email: [email protected]
Characterization of plasma proteins adsorbed onto biomaterials by MALDI-TOFMS
Biomaterials 21 (2000) 1701}1710
Characterization of plasma proteins adsorbed onto biomaterials by MALDI-TOFMS R.D. Oleschuk , M.E. McComb , A. Chow, W. Ens, K.G. Standing, H. Perreault *, Y. Marois, M. King Department of Chemistry, University of Alberta, Edmonton, AB, Canada T6G 2G2 Departments of Chemistry and Physics, University of Manitoba, Winnipeg, MB, Canada R3 T 2N2 Departments of Biochemistry and Biophysics, Boston University School of Medicine, Boston, MA 02118-2394, USA Department of Physics, University of Manitoba, Winnipeg, MB, Canada R3 T 2N2 Institut des Biomate& riaux, 10, rue de l+Espinay, Que& bec, QC, Canada G1L 3L5 Received 22 October 1999; accepted 22 February 2000
Abstract The analysis of plasma proteins adsorbed onto a polyurethane (PU) biomaterial was performed using matrix-assisted laser desorption/ionization time-of-#ight mass spectrometry (MALDI-TOFMS). This article marks the "rst study on MALDI-TOFMS analysis of multiple proteins adsorbed from plasma, in vitro, onto the surface of a biomaterial to easily enable their characterization. Plasma standards from three di!erent hosts were placed in contact with non-porous PU, a model biomaterial. Following the use of washing protocols developed in our laboratory, the biomaterial was analyzed, directly, with MALDI-TOFMS. Proteins with molecular weights (M ) ranging from ca. 6.5 to 150 kDa were observed in the mass spectra and characterized upon comparison with proteins of known M . The proteins observed were tentatively identi"ed as those known to adsorb onto PU, both in vitro and in vivo. In an attempt to model in vivo sorption, the PU biomaterial was exposed to freshly collected canine plasma, in vitro, for di!erent lengths of time. Corresponding MALDI-TOFMS spectra displayed increasing protein signal for a number of di!erent proteins with increasing times of exposure to plasma. This method provided qualitative and semi-quantitative analysis of the proteins adsorbed onto the biomaterial surface. 2000 Elsevier Science Ltd. All rights reserved. Keywords: MALDI; TOFMS; Plasma protein; Biomaterial; Biocompatible
1. Introduction Medical devices composed of polymeric materials have become ubiquitous for the treatment of certain disorders, however, implantation of polymeric devices within the human body can lead to a number of problems. In#ammation and thrombosis may cause the body to reject the implant, or the polymer material may be enzymatically degraded, leading to the deterioration or failure of the device. Protein sorption onto the surface of the biomaterial is a major factor contributing to the stability of materials in the body, and ultimately determines the in vivo performance of the polymer [1,2]. The adsorption of certain proteins is bene"cial to inhibit further physio-
* Corresponding author. Tel.:#1-204-474-7418; fax:#1-204-474-7608. E-mail address: [email protected] (H. Perreault).
logical response, while adsorption of others produces a detrimental e!ect, enhancing platelet adhesion and thrombus formation (shown schematically in Fig. 1) and resulting in di$culties with the implanted device. Therefore, the characterization and measurement of surface-adsorbed proteins is important for assessing the biocompatibility of a material. Currently, the characterization of adsorbed proteins is carried out using a number of techniques [3]. Traditional analyses utilizing radio-labeled proteins [4,5] have been performed, but are limited to monitoring the adsorption of one protein at a time. More elaborate protocols involving removal of adsorbed proteins from the biomaterial surface with a surfactant and subsequent analysis by SDS-PAGE [6,7] have also been used, but are extremely time consuming. More recently, techniques such as Fourier transform infrared spectroscopy [8}10] (FTIR), Raman spectroscopy, and surface plasmon resonance [11}13] (SPR) have been utilized to quantify the
0142-9612/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 2 - 9 6 1 2 ( 0 0 ) 0 0 0 5 4 - 5
1702
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
Fig. 1. Schematic representation of protein sorption onto a biomaterial, believed to be responsible for the initial stages of its degradation (not to scale).
amount of adsorbed protein on surfaces of materials, but are also unable to e$ciently analyze several proteins simultaneously. Often, a physiological response to a biomaterial is caused by a complementary interaction [14,15] between two or more proteins. Techniques that can monitor the adsorption of only a single protein would not detect a synergistic interaction. A technique that can monitor and characterize the adsorption of several proteins simultaneously on the biomaterial would be ideal. Matrix-assisted laser desorption/ionization time-of#ight mass spectrometry (MALDI-TOFMS) has become a rapid and convenient method for the characterization of proteins and peptides from biological samples. The technique can simultaneously measure the molecular weights (M ) of several proteins in a sample, allowing them to be identi"ed by comparing experimentally determined M with literature values. MALDI is tolerant of impurities compared with other MS ionization methods. This facilitates the analysis of biological samples which often contain bu!er components in relatively high concentrations. MALDI-TOFMS has been used to analyze proteins prepared on di!erent types of supports [16}21]. Generally, these supports have been polymeric membranes and have been used to remove polar components from protein samples. Sample preparation on a membrane allows the removal of interfering agents such as salts and bu!er components and makes the identi"cation and analysis of proteins in complex mixtures, such as biological #uids, easier. Membranes so far used for the puri"cation of protein samples generally have a high binding a$nity for a wide variety of proteins. In contrast, biomaterials usually adsorb certain proteins preferentially to give them enhanced in vivo stability. It has recently been suggested that MALDI-TOFMS would be a convenient method for the characterization of proteins adsorbed directly onto surfaces, such as biomaterials [22}24]. The biomaterial sample may be analyzed in an analogous manner to membrane MALDI-TOFMS samples and thus provide information about the type, and amount of
protein bound to its surface following exposure to a biological #uid. In this study polyurethane (PU) was chosen as a model biomaterial because of its extensive use in the construction of biomedical implant devices [4]. Protein sorption onto PU in vitro and in vivo is well characterized within the literature, allowing MALDI-TOFMS to be compared with more traditional methods of analysis. In addition, our laboratory has experience with using PU membranes as sample supports for MALDI-TOFMS characterization of peptides and proteins derived from biological samples [19,25,26]. Polyurethane membrane material was exposed to plasma standards from three di!erent hosts. Mass spectra of the plasma standards on the PU biomaterial showed the adsorption of several proteins ranging from 6.5 to 150 kDa. Proteins observed in the mass spectra directly correlate with those shown by other researchers to be adsorbed from plasma onto PU-based materials, thus validating MALDI-TOFMS as a technique for biomaterial analysis. This type of application may also be used for screening among possible future biomaterials.
2. Experimental section 2.1. Reagents and materials All chemicals used in this study were of reagent grade unless otherwise stated. HPLC grade acetonitrile and methanol were from Fisher Scienti"c (Fair Lawn, NJ). Deionized water, prepared with a Milli-Q plus}TOC water puri"cation system (Millipore, Bedford, MA), was employed in all solutions. Plasma protein standards (human, bovine and canine) were obtained from Sigma Chemicals (St. Louis, MO). Sinapinic acid, used as the MALDI matrix, was from Sigma. Protein standards were used as MALDI calibrants. Horse heart myoglobin, 16 951 Da, bovine insulin, 5733 Da, bovine serum albumin, 66 430 Da, were purchased from Sigma, and bovine apotransferrin, 78 030 Da, was obtained from Calbiochem (LaJolla, CA). Solutions of these standards were made up in water, and used without further puri"cation.
2.2. PU membrane model biomaterial The non-porous ether-type PU membrane, 50 lm in thickness (XPR625-FS) was supplied by Stevens Elastomerics (Northampton, MA). The PU membrane used consists of a co-polymer of polytetramethylene ether glycol and diphenylmethane diisocyanate (MDI) [27]. The membrane was washed with water and methanol prior to use in order to remove polar and non-polar surface contaminants.
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
2.3. MALDI-TOFMS
3. Results and discussion
MALDI-TOFMS was performed in the linear mode on a re#ecting time-of-#ight mass spectrometer, an instrument built in-house in the Time-of-Flight Lab, University of Manitoba [28]. Accelerating potentials of 25 and 30 kV were used. Spectra were obtained using a nitrogen laser (337 nm) (VSL 337 ND, Laser Science Inc., Cambridge, MA, USA) with the #uence adjusted slightly above threshold. In order to avoid saturation of the detector by low mass matrix ions, the detector was pulsed on ca. 19 000 ns after each laser shot. External calibrations for measurements using the PU membrane were performed with standards (bovine insulin, equine myoglobin, human transferrin) prepared on similar membrane targets. Spectra presented here are summations of ca. 50}100 consecutive shots.
3.1. Analysis of a human plasma standard
2.4. Preparation of plasma standard samples Samples were prepared on PU based on protocols developed in our laboratory [19]. Samples were prepared on the membrane by "rst depositing 2 lL of the plasma solution onto the surface of the membrane and allowing it to dry. A mark was placed on the underside of the membrane to allow the sample spot to be located after washing. The membrane was then washed with 50 lL of water for 1 min to remove the non-adsorbed plasma components. The washing step was repeated. After drying, immediately before analysis, matrix (2 lL, saturated sinapinic acid in 1 : 1 H O/acetonitrile) was placed on the membrane and allowed to crystallize. The membrane was then trimmed to size, a$xed to the MALDI target using an adhesive spray (Spraymount, 3M) and subsequently analyzed by MALDI-TOFMS.
1703
The nature of protein sorption ultimately determines the in vivo stability of a material. PU possesses a twophase bulk structure consisting of hard (moderately polar) and soft (non-polar) segments as shown in Fig. 2a [29,30]. The chemical structure of each of these segments greatly in#uences the ability of the polymer to adsorb proteins [31]. Protein binding is believed to occur through a combination of hydrogen bonding and hydrophobic interactions between the protein and the surface of the polymer [32]. Several di!erent types of PU have been used and tested for the construction of biomedical devices [33]. For our study the polytetramethylene oxide (PTMO) methylene bis(p-phenyl diisocyanate (MDI)type PU (Fig. 2b) was chosen because it can adsorb larger amounts of protein than PU prepared with more hydrophilic soft segments can [29]. Previously, we have shown that MALDI-TOFMS resolution and mass accuracy for proteins directly deposited onto the PTMO/ MDI-type PU membrane used here were equivalent to those obtained using a metallic target [19,26]. Although MALDI-TOFMS is relatively tolerant of impurities such as salts and bu!er components, the removal of these species yields greater resolution and mass accuracy. In addition, proteins with no a$nity for the membrane material must be removed prior to analysis. To accomplish this a rinsing step may be performed. All of the components with a high a$nity for the biomaterial remain adsorbed while polar components and non-adsorbed proteins are washed o! the material. The washing
2.5. In vitro experiments In vitro experiments were performed at the Institut des BiomateH riaux du QueH bec, UniversiteH Laval. Samples of PU membrane material were cut into 2.0 cm;2.0 cm pieces and exposed to pooled samples of freshly collected canine plasma in neutral glass vials (Anchor Ltd, London, Ontario). Membrane samples were exposed (in triplicate) for periods of time from 0 to 4 h at 373C. Following exposure, the samples were immediately washed with water to remove non-adsorbed plasma components, dried under vacuum and later shipped to our laboratory at the University of Manitoba for MALDI-TOFMS analysis. 2.6. Safety considerations Suitable precautions should be taken when handling biological #uids. Gloves were worn during collection and manipulation of samples.
Fig. 2. (a) Two-phase bulk structure of PU membrane consisting of hard (moderately polar) and soft (non-polar) segments. (b) Chemical structure of PU used in these experiments.
1704
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
Fig. 3. MALDI-TOF mass spectra of human plasma obtained using PU as a sample support. (a) Low M region. (b) High M region. Tentative assignments are found in Table 1.
Table 1 Tentative assignments for human plasma proteins adsorbed on polyurethane and detected by MALDI-TOFMS Protein
Immunoglobulin G Transferrin Comp C3-b Albumin a -HS Glycoprotein Apo A-4 Apo E Apo A-1 Apo D Apo H-III (HDL) Transthyretin Apo C-II Apo C-III Apo C-I
Molecular weight (Da)
Relative peak area
Concentration
Literature
Experimental
RSD
% RSD
(%)
RSD
% RSD
(g/L)
150 000 79 754 75 000 66 472 49 000 43 400 34 236 28 000 22 500 17 000 13 761 8914 8764 6630
148 361 78 513 73 953 66 463 49 575 44 175 33 109 27 502 22 174 16 327 13 555 8946 8783 6631
1900 980 590 * 400 170 200 14 30 460 170 16 10 10
1.3 1.3 0.80 * 0.80 0.40 0.60 0.05 0.15 3 1.3 0.20 0.20 0.20
20 17 31 100 20 16 56 135 17 25 171 * * *
5 3 9 * 3 12 28 107 8 17 159 * * *
23 20 27 * 16 77 49 79 47 68 93 * * *
8}18 2.0}4.0 (1 35}55 0.4}0.85 (1 (1 (1 (1 3 (1 (1 (1 (1
Peak areas of high M proteins normalized to albumin. Concentration in whole blood plasma from Ref. [34]. Literature molecular weights from Refs. [34,35]. RSD and % RSD based on 4 replicate measurements. Albumin was used as the internal calibrant.
process is analogous to membrane puri"cation of proteins prior to MALDI-MS analysis [18,19]. Aliquots of human plasma standard (2 lL) were placed onto the surface of the membrane and allowed to dry. Following rinsing and application of matrix, the samples were placed onto a MALDI probe and inserted in the mass spectrometer for analysis. Fig. 3 shows MALDITOF mass spectra of human plasma proteins adsorbed onto the PU biomaterial (also see Table 1). In order to obtain reasonable signals in the high M range ('20 kDa) it was necessary to suppress the background from the matrix by turning on the microchannel plate detector on, only after the matrix ions had time to reach the detector. Good-quality MALDI mass spectra were thus obtained after removal of excess of salts and bu!er components present in plasma (Table 2). Ions from at least
14 di!erent proteins were visible. The proteins ranged in M from ca. 6.5 to 150 kDa. Mass resolution and accuracy were su$cient for identi"cation upon comparison with M values from the literature (Table 1) [34,35]. MALDI-TOFMS has been used for both the qualitative, semi-quantitative and quantitative analysis of protein and peptide mixtures [36,37] and recently for the quantitative measurement of single protein sorption onto polymer surfaces [24]. This investigation reports on the semi-quantitative and qualitative determination of several plasma proteins simultaneously adsorbed onto a PU membrane using relative MALDI-TOFMS peak intensities. Tentative mass assignments for each of the proteins [38] are shown in Table 1 along with integrated peak areas measured relative to albumin. Reproducibility in M measurements was good, however the % RSD
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710 Table 2 Concentration of bu!er components in plasma Component
Concentration (mM)
Bicarbonate Calcium Chloride Cholesterol Glucose Iron Magnesium Phosphate Potassium Sodium Sulfate Urea Uric acid
42}54 4.2}5.2 95}103 0.038}0.063 0.036}0.055 1.0}2.68;10\ 3.0}5.2 1.8}2.6 8}9.6 136}142 0.4}2.6 0.013}0.033 1.2}3.1;10\
Values adapted from Ref. [38].
measurements of relative peak areas ranged considerably from 16 to 93%. The principle error in peak integration was due to slight di!erences in mass spectra obtained on di!erent samples. Slight variations in the matrix preparation steps led to non-uniform co-crystallization of the plasma proteins with the matrix and the production of irregular crystals on the surface of the PU membrane. This resulted in variable signal intensity in the mass spectra dependent on the location of the target sampled with the laser. Additional care must be taken in the interpretation of the spectra. The MALDI signal intensity observed is not necessarily indicative of the amount of protein on the surface of the membrane, neither is the amount of a particular protein adsorbed onto the membrane indicative of its concentration in solution. The a$nity of the protein for the membrane plays a more signi"cant role than its concentration. This is demonstrated by the absence of ions of several major proteins, and by the presence of several low concentration proteins ions (Fig. 3b, Table 1). Only proteins with an a$nity for the membrane material are adsorbed and later detected, while those with low a$nity are washed o! [21,22]. Additionally, it is known that MALDI signal intensity is dependent on the amount of the protein and the ease with which it will co-crystalize with the matrix and ionize in the MALDI plume [39]. We have shown such discrimination phenomena for mixtures of medium and high M wheat proteins prepared on PU supports vs. steel targets [26]. Plasma proteins which are known to adsorb onto PU materials in vivo include lipoproteins [40,41], albumin [42], immunoglobulin G (IgG) [43], C3, "brinogen as well as others [4]. With the exception of "brinogen, these proteins were observed in the mass spectra obtained on PU (Fig. 3). This indicates a direct correlation between the proteins detected in the mass spectrum of human plasma on PU and proteins known to adsorb onto PU in vivo, determined using other techniques.
1705
Biomaterials development has focused on the adsorption of three proteins in particular, as indicators of biocompatibility [3]. The adsorption levels of albumin, IgG and "brinogen are used to assess the in vivo performance of a material. For biomaterials, albumin adsorption is bene"cial owing to this protein's relative lack of glycosylation, which prevents platelet adhesion. On the other hand, adsorption of IgG and "brinogen may cause a host defense reaction which increases platelet adhesion and promotes a physiological response. Since the adsorption of both albumin and IgG was observed in this study it follows that MALDITOFMS could be used to assess biocompatibility of di!erent materials by analyzing the amount of albumin or IgG adsorbed by a particular material. The third major protein, "brinogen, was not detected because of its high M . In addition, "brinogen is normally observed after longer periods of exposure than were used in this investigation. 3.2. Sample-to-sample reproducibility Initially, the mass spectra of human plasma proteins on PU exhibited relatively poor reproducibility in peak integration. Sample-to-sample reproducibility was explored with care taken to ensure that the preparation/ washing and subsequent MALDI-TOFMS measurements were obtained under nearly identical conditions. The results are shown in Fig. 4, for four replicate samples of bovine plasma standard prepared on PU. Each spectrum shown is the summation of 50 consecutive shots sampled over the entire surface of the target. Good shotto-shot reproducibility was obtained when samples were subjected to the same sample preparation and MALDI conditions. Relative peak areas and heights are given in Table 3. As with the human plasma proteins integration of peak areas resulted in relatively high % RSD values (mean 42%). Improved results were obtained by measuring the peak heights which had a mean % RSD of ca. 28%. Peak height variations overall were less than those obtained with integration and ranged from 9 to 54%. The larger #uctuation in peak areas arose mainly from variation in peak widths possibly due to inconsistent formation of adducts (i.e. with the matrix) from one sample to another. Di!erences in baseline assignment associated with manual integration may have also contributed to the larger error. Comparison of the two semi-quantitative sets of measurements and the errors associated with them are better visualized in Fig. 5. Here it can bee seen that, even though there exists some variance in the measurements, semi-quantitative results of peak area or height may still be obtained and are respective of the amount of protein adsorbed onto the surface of the membrane. Relative MALDI peak areas or heights can thus be used as a semi-quantitative assessment of the concentration
1706
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
Fig. 5. Graphical representation of statistical values from Table 3. Mean and standard deviation in (a) peak area, (b) peak height.
Fig. 4. Reproducibility of MALDI-TOFMS measurements obtained for four replicate samples of bovine plasma standards. Masses correspond to values in Table 3.
of di!erent proteins adsorbed onto the biomaterial. In addition, small mass shifts in the protein adsorption pro"le of a material can be determined and allow better biomaterial characterization.
Table 3 Reproducibility in MALDI-TOFMS measurements of bovine plasma proteins adsorbed on polyurethane, High M Molecular weight
Peak area
Peak height
(Da)
RSD
% RSD
(%)
RSD
% RSD
(%)
RSD
% RSD
12 571 13 657 14 881 15 946 22 034 22 831 27 470 33 153 38 796 48 942 53 007 66 433 73 324 77 606
161 126 147 192 690 510 193 179 292 397 517 * 279 341
1.27 0.92 0.98 1.20 3.10 2.21 0.70 0.54 0.75 0.80 0.97 * 0.38 0.44
75 341 93 28 26 22 157 57 12 9 16 100 31 17
38 104 24 7 21 23 35 5 7 3 8 * 13 8
51 30 26 24 82 103 22 8 61 35 52 * 42 49
288 551 216 96 47 48 190 78 28 27 28 100 40 36
48 87 21 20 12 12 58 10 12 13 15 * 16 19
16 15 9 21 25 24 30 12 44 50 54 * 40 54
Average % RSD
28
Average % RSD Albumin was used as the internal calibrant. Peak areas of high M proteins normalized to albumin. Peak heights of high M proteins normalized to albumin. RSD and % RSD based on four replicate measurements.
42
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
1707
3.3. Plasma standards of diwerent hosts
3.4. Inyuence of sample preparation conditions
Studies suggest that there are slight di!erences in protein adsorption patterns between plasma samples from di!erent hosts and that these are due to variations in hematological pro"les [44]. Plasma samples from three hosts (i.e. human, canine and bovine) were used to determine the ability of MALDI-TOFMS to distinguish between protein adsorption patterns for di!erent plasma standards. PU samples were exposed to each plasma standard and then analyzed to compare adsorption patterns. The spectra of three di!erent plasma standards adsorbed on PU are shown in Fig. 6. These results indicate that proteins of similar high M were adsorbed (Fig. 6b) for each plasma standard. There were di!erences in the low M pro"les (Fig. 6a). The human plasma standard result ed in the highest number of adsorbed proteins in the 5.5}10 kDa mass range, possibly lipoproteins. In comparison, the spectra obtained for bovine and canine plasma showed adsorption of fewer proteins in this region. The adsorption of lipids onto PU surfaces has been associated with the biodegradation of PU implants [45]. It has also been suggested that the interaction between lipoproteins and di!erent domains of the PU facilitates lipid adsorption [4]. The variations between protein adsorption patterns could lead to di!erences in biocompatibility, questioning the validity of testing biomaterials with other hosts [40]. MALDI-TOFMS is capable of distinguishing between di!erent protein adsorption patterns and therefore may be seen as a valid method for biomaterial analysis.
Di!erent sample preparation methods were tested for their e!ect on ion signal intensities and to further determine conditions which may be used for in vitro testing. In Fig. 7a, bovine plasma was applied onto the PU membrane followed by washing and addition of matrix. In Fig. 7b, methanol was applied following the application and drying of the plasma sample. The addition of methanol resulted in an increase in the overall MALDI signal. A similar phenomenon was reported in an earlier publication [19]. We associated this to enhanced protein binding on the membrane, an e!ect commonly exploited in SDS-PAGE when electroblotting onto porous membranes. Methanol enhanced protein adsorption was also described for MALDI-TOFMS on porous membranes [17]. In Fig. 7c, methanol was added to the sample before the plasma sample had a chance to dry on the membrane. This resulted in the precipitation of the proteins from the plasma. Alcohol precipitation is often used to remove proteins from plasma/serum in order to analyze
Fig. 6. MALDI-TOF mass spectra of three di!erent plasma standards: bovine, canine and human on PU membrane. (a) Low M region. (b) High M region.
Fig. 7. MALDI-TOF mass spectra of bovine plasma proteins on PU membrane using di!erent sample preparation methods. (a) Plasma, dried, washed, dried, matrix. (b) Plasma, dried, methanol, dried, washed, dried, matrix. (c) Plasma#methanol, dried, washed, dried, matrix. (d) Plasma#matrix solution, dried, washed, dried, matrix.
1708
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
other components. The result was an increase in signal intensity for some of the proteins, notably, albumin. Fig. 7d shows the e!ect of adding matrix solution to the sample before the plasma had a chance to dry on the membrane. In this case, the spectrum showed substantial di!erences relative to features observed in Fig. 7a}c. Even though the matrix material is dissolved in 50% (v/v) acetonitrile, it is acidic in nature. Protonation in solution will result in a net charge on the plasma proteins and will disfavor their sorption onto the predominantly hydrophobic membrane. The acidity of the matrix facilitates the removal of some bound proteins from the surface so they may co-crystallize with the matrix for analysis by MALDI [19]. The di!erent sample preparation protocols led to changes in the amount and nature of the proteins adsorbed and detected by MALDI. These discrepancies show that a consistent sample preparation method must be employed to directly compare protein adsorption patterns on di!erent biomaterials. With the addition of methanol to the samples (Fig. 7b and c) it was possible to improve spectral quality compared with the neat sample. However, these approaches do not re#ect solution conditions which would be observed in vivo. Therefore, the conditions chosen for the in vitro experiments where based on the "rst protocol used (Fig. 7a). 3.5. In vitro analysis In vitro testing of biomaterials is commonly performed because it is relatively inexpensive and faster than in vivo methods. In vitro experiments involve the isolation of #uid or tissue from a biological source, followed by experimental characterization of this material. In this study, the MALDI analysis of proteins adsorbed in vitro onto PU over di!erent times of exposure to canine plasma was performed. Canine plasma is commonly used by biomaterials researchers and represents the worst case scenario for protein adsorption since it produces extensive platelet adhesion and thrombus formation. Freshly collected canine plasma was placed in contacted with the PU material under speci"ed conditions to facilitate protein sorption. Triplicate samples were exposed to canine plasma for di!erent lengths of time (0, 0.5, 1 and 4 h) at 37C. Representative mass spectra showing the e!ect of the time of exposure of canine plasma on the amount of proteins adsorbed are shown in Fig. 8. In the control sample spectrum, (Fig. 8a) no peaks are observed. Analysis of the sample exposed to plasma for 0.5 h shows several peaks due to adsorption of proteins onto the surface of the PU biomaterial. Continued exposure, up to 4 h, yielded more intense protein signals. This may be interpreted as increasing amounts of protein being adsorbed on the surface of the polymer.
Fig. 8. MALDI-TOF mass spectra obtained on PU membranes exposed to canine plasma for (a) 0 (b) 0.5 (c) 1 and (d) 4 h.
These data supports the notion that protein adsorption occurs quite rapidly upon exposure of the polymeric material [1]. It has been suggested that for relatively short periods of exposure, protein adsorption "ts a Langmuir isotherm [46]. The distribution of proteins adsorbed at the surface of a material is, however, time dependent. Proteins are constantly deposited, and displaced and thus the surface composition changes [47]. The rates of deposition and displacement can be determined by the relative concentrations of proteins within the plasma and their a$nity for the material. Although proteins with a high plasma concentration may be adsorbed "rst, those with low concentrations but with a higher a$nity for a particular surface will eventually displace them. The displacement of proteins occurs after the surface of the polymer has become saturated. This phenomenon was not observed in this study, even over the four hour exposure. Instead, analysis of PU samples exposed to canine plasma in vitro yield spectra suggesting that the longer the exposure time, the larger the amount of protein present on the surface of the material. The mass spectral pro"le observed, although weaker in absolute intensity, was similar to the pro"le shown for the canine standard (Fig. 6b). This resemblance indicates a strong similarity between the proteins observed from the in vitro experiments and those observed for the canine plasma standard. The use of plasma standards for biomaterials testing may be su$cient as an initial
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
screening process for the biomaterial. Here, the use of plasma standards is preferred because it is faster, simpler and less expensive than performing either in vivo or in vitro experiments. 4. Conclusions Among the various kinds of interfacial phenomena occurring in the "eld of biomaterials, the adsorption of proteins at the interface between a liquid and a solid is one of the most signi"cant areas that has been investigated. The analysis of proteins adsorbed onto biomaterials is a new application of MALDI-TOFMS. The technique is able to analyze several proteins simultaneously with su$cient resolution to allow the easy characterization of adsorbed proteins based on literature M . Experiments performed in this study showed direct correlations between the proteins known to adsorb on PU from plasma and those seen in the mass spectra of plasma proteins on PU. The analysis of proteins adsorbed onto PU from plasma standards yielded good-quality mass spectra, displaying a variety of proteins from 6.5 to 150 kDa. The MALDI spectra exhibited good shot-to-shot reproducibility with only minor shifts in peak height. These features enabled the semi-quantitative analysis of adsorbed proteins. Comparison of plasma standards from di!erent hosts showed only small di!erences in the high M pro tein distribution, and more pronounced di!erences for low M proteins. Di!erences in sample manipulation were shown to have a profound e!ect on spectra, therefore care should be taken when developing experimental procedures for biomaterial characterization. This study clearly demonstrates that MALDITOFMS may be used to monitor the adsorption of di!erent proteins onto biomaterials. Therefore, this technique is helpful for determining the biocompatibility of the biomaterial in vivo. The method may be readily expanded to include testing implant devices removed from patients following implantation. Acknowledgements This work was supported by the National Science and Engineering Research Council of Canada (NSERC). We thank the members of the Time-of-Flight Laboratory, University of Manitoba, for helpful discussion. Finally, we wish to thank the University of Manitoba for a Graduate Fellowship awarded to M. E. McComb.
References [1] Vroman L, Adams AL, Klings M, Fischer GC, Munoz PC, Solensky RP. Ann New York Acad Sci 1977;283:65.
1709
[2] Baire RE. Artif Organs 1978;2:422. [3] J.H. Braybrook, Biocompatibility assessment of medical devices and materials, Biomaterials and Engineering Series, Wiley: New York, NY, 1997. [4] Lamba N, Woodhouse K, Cooper S. Polyurethanes in biomedical applications. Boca Raton, FL: CRC Press, 1998. [5] Brundstedt MR, Ziats NP, Robertson SP. J Biomed Mater Res 1993;27:367. [6] Osterberg E, Bergstrom K, Holmberg K, et al. Colloid Surf A 1993;77:159. [7] Lyman DJ, Metcalf LC, Albo D, Richards VK, Lamb. J Trans Am Soc Artif Intern Organs 1974;20:474. [8] Norman ME, Williams P, Illum L. Biomaterials 1992;13:841. [9] Bellissimo JA, Cooper SL. Trans Am Soc Artif Intern Organs 1984;30:359. [10] Holland NB, Qiu Y, Ruegsegger M, Marchant RE. Nature 1998;392:799. [11] Pitt WG, Grasel TG, Cooper SL. Biomaterials 1988;9:36. [12] Mrksich M, Sigal GB, Whitesides GM. Langmuir 1995;11:4383. [13] Green RJ, Davies J, Davies MC, Roberts CJ, Tendler SJB. Biomaterials 1997;18:405. [14] Davies J. Nanobiology 1994;3:5. [15] Janatova J. 23rd Anual Meeting of the Society For Biomaterials. USA: New Orleans, 1977. [16] Strupat K, Karas M, Hillenkamp F. Anal Chem 1994;66:464. [17] Blackledge JA, Alexander AJ. Anal Chem 1995;67:843. [18] Worrall TA, Cotter RJ, Woods AS. Anal Chem 1998;70:750}6. [19] McComb ME, Oleschuk RD, Manley DM, Donald L, Chow A, O'Neil JDJ, Duckworth HW, Ens W, Standing KG, Perreault H. Rapid Commun Mass Spectrom 1997;11:1716. [20] Chaurand M, Stoeckli M, Caprioli RM. Anal Chem 1999; 71:5263. [21] Schleuder D, Hillenkamp F, Strupat K. Anal Chem 1999; 71:3238. [22] Walker AK, Wu Y, Timmons RB, Kinsel GR, Nelson KD. Anal Chem 1999;71:268. [23] McComb ME, Oleschuk RD, Marois Y, King MW, Chow A, Ens W, Standing KG, Perreault H. Proceedings of the 46th ASMS Conference, 1998, p. 175. [24] Walker AK, Land CM, Kinsel GR, Nelson KD. J Am Soc Mass Spectrom 2000;11:62. [25] McComb ME, Oleschuk RD, Chow A, Ens W, Standing KG, Smith M, Perreault H. Anal Chem 1998;70:5142. [26] McComb ME, Oleschuk RD, Chow A. Perreault H, Dworschak RG, Znamirowski M, Ens W, Standing KG, Preston K. J Amer Soc Mass Spectrom, in submission. [27] Oleschuk RD. Separation isolation of metals using polyurethane materials. Winnipeg, MB: University of Manitoba, 1998. [28] Tang X, Beavis RC, Ens W, Lafortune F, Schueler B, Standing KG. Int J Mass Spec Ion Proc 1988;85:43. [29] Silver JH, Meyers CW, Lim F, Cooper SL. Biomaterials 1994;15:695. [30] Petrovic ZS, Fergenson. J Prog Polym Sci 1991;16:695. [31] Ratner BD. In: Polymers in medicine III. Amsterdam: Elsevier, 1988. p. 87. [32] Andrade JD. Surface and interfacial aspects of biomedical polymers. New York, NY: Plenum Press, 1985. [33] Goodman SL, Simmons SR, Cooper SL, Albrecht RM. Trans Soc Biomaterials 1989;15:201. [34] Putnam FW, The plasma proteins, vol IV. New york: Academic Press, 1984 (Chapters I and II). [35] http://expasy.hcuge.ch/sprot/sprot-top.html [36] Jesperson S, Niessen WMA, Tjaden UR, van der Greef J. J Mass Spectrom 1995;30:357. [37] Nelson RW, McLean MA. Anal Chem 1994;66:1408. [38] Anderson SC, Cockayne S. Clinical chemistry: concepts and applications. Toronto, ON: Saunders, 1993.
1710
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
[39] Xiang F, Beavis RC. Rapid Commun Mass Spectrom 1994; 8:199. [40] Dong DE, Andrade JD, Coleman DL. Polymer Prep 1983;24:40. [41] Takahara A, Takahashi K, Kajiyama T. J Biomater Sci Polym Ed 1993;5:183. [42] Kim SW, Lee RG, Oster H, Coleman D, Andrade JD, Lentz DJ, Olsen D. Trans Am Soc Artif Intern Organs 1974;20:449.
[43] Leonard EF, Vroman LJ. Biomater Sci Polym Ed 1991;3:95. [44] Grabowski EF, Didisheim P, Lewis CJ, Franta JT, Stropp JQ. Trans Am Soc Artif Intern Organs 1977;23:141. [45] Sreenivasan K, Jayabalan M, Rao KVC. J Applied Polym Sci 1992;45:2105. [46] Young BR, Pitt WG, Cooper SL. J Coll Interf Sci 1988;28:124. [47] Vroman L, Adams AL. J Biomed Mater Res 1969;3:43.
Characterization of plasma proteins adsorbed onto biomaterials by MALDI-TOFMS R.D. Oleschuk , M.E. McComb , A. Chow, W. Ens, K.G. Standing, H. Perreault *, Y. Marois, M. King Department of Chemistry, University of Alberta, Edmonton, AB, Canada T6G 2G2 Departments of Chemistry and Physics, University of Manitoba, Winnipeg, MB, Canada R3 T 2N2 Departments of Biochemistry and Biophysics, Boston University School of Medicine, Boston, MA 02118-2394, USA Department of Physics, University of Manitoba, Winnipeg, MB, Canada R3 T 2N2 Institut des Biomate& riaux, 10, rue de l+Espinay, Que& bec, QC, Canada G1L 3L5 Received 22 October 1999; accepted 22 February 2000
Abstract The analysis of plasma proteins adsorbed onto a polyurethane (PU) biomaterial was performed using matrix-assisted laser desorption/ionization time-of-#ight mass spectrometry (MALDI-TOFMS). This article marks the "rst study on MALDI-TOFMS analysis of multiple proteins adsorbed from plasma, in vitro, onto the surface of a biomaterial to easily enable their characterization. Plasma standards from three di!erent hosts were placed in contact with non-porous PU, a model biomaterial. Following the use of washing protocols developed in our laboratory, the biomaterial was analyzed, directly, with MALDI-TOFMS. Proteins with molecular weights (M ) ranging from ca. 6.5 to 150 kDa were observed in the mass spectra and characterized upon comparison with proteins of known M . The proteins observed were tentatively identi"ed as those known to adsorb onto PU, both in vitro and in vivo. In an attempt to model in vivo sorption, the PU biomaterial was exposed to freshly collected canine plasma, in vitro, for di!erent lengths of time. Corresponding MALDI-TOFMS spectra displayed increasing protein signal for a number of di!erent proteins with increasing times of exposure to plasma. This method provided qualitative and semi-quantitative analysis of the proteins adsorbed onto the biomaterial surface. 2000 Elsevier Science Ltd. All rights reserved. Keywords: MALDI; TOFMS; Plasma protein; Biomaterial; Biocompatible
1. Introduction Medical devices composed of polymeric materials have become ubiquitous for the treatment of certain disorders, however, implantation of polymeric devices within the human body can lead to a number of problems. In#ammation and thrombosis may cause the body to reject the implant, or the polymer material may be enzymatically degraded, leading to the deterioration or failure of the device. Protein sorption onto the surface of the biomaterial is a major factor contributing to the stability of materials in the body, and ultimately determines the in vivo performance of the polymer [1,2]. The adsorption of certain proteins is bene"cial to inhibit further physio-
* Corresponding author. Tel.:#1-204-474-7418; fax:#1-204-474-7608. E-mail address: [email protected] (H. Perreault).
logical response, while adsorption of others produces a detrimental e!ect, enhancing platelet adhesion and thrombus formation (shown schematically in Fig. 1) and resulting in di$culties with the implanted device. Therefore, the characterization and measurement of surface-adsorbed proteins is important for assessing the biocompatibility of a material. Currently, the characterization of adsorbed proteins is carried out using a number of techniques [3]. Traditional analyses utilizing radio-labeled proteins [4,5] have been performed, but are limited to monitoring the adsorption of one protein at a time. More elaborate protocols involving removal of adsorbed proteins from the biomaterial surface with a surfactant and subsequent analysis by SDS-PAGE [6,7] have also been used, but are extremely time consuming. More recently, techniques such as Fourier transform infrared spectroscopy [8}10] (FTIR), Raman spectroscopy, and surface plasmon resonance [11}13] (SPR) have been utilized to quantify the
0142-9612/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 2 - 9 6 1 2 ( 0 0 ) 0 0 0 5 4 - 5
1702
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
Fig. 1. Schematic representation of protein sorption onto a biomaterial, believed to be responsible for the initial stages of its degradation (not to scale).
amount of adsorbed protein on surfaces of materials, but are also unable to e$ciently analyze several proteins simultaneously. Often, a physiological response to a biomaterial is caused by a complementary interaction [14,15] between two or more proteins. Techniques that can monitor the adsorption of only a single protein would not detect a synergistic interaction. A technique that can monitor and characterize the adsorption of several proteins simultaneously on the biomaterial would be ideal. Matrix-assisted laser desorption/ionization time-of#ight mass spectrometry (MALDI-TOFMS) has become a rapid and convenient method for the characterization of proteins and peptides from biological samples. The technique can simultaneously measure the molecular weights (M ) of several proteins in a sample, allowing them to be identi"ed by comparing experimentally determined M with literature values. MALDI is tolerant of impurities compared with other MS ionization methods. This facilitates the analysis of biological samples which often contain bu!er components in relatively high concentrations. MALDI-TOFMS has been used to analyze proteins prepared on di!erent types of supports [16}21]. Generally, these supports have been polymeric membranes and have been used to remove polar components from protein samples. Sample preparation on a membrane allows the removal of interfering agents such as salts and bu!er components and makes the identi"cation and analysis of proteins in complex mixtures, such as biological #uids, easier. Membranes so far used for the puri"cation of protein samples generally have a high binding a$nity for a wide variety of proteins. In contrast, biomaterials usually adsorb certain proteins preferentially to give them enhanced in vivo stability. It has recently been suggested that MALDI-TOFMS would be a convenient method for the characterization of proteins adsorbed directly onto surfaces, such as biomaterials [22}24]. The biomaterial sample may be analyzed in an analogous manner to membrane MALDI-TOFMS samples and thus provide information about the type, and amount of
protein bound to its surface following exposure to a biological #uid. In this study polyurethane (PU) was chosen as a model biomaterial because of its extensive use in the construction of biomedical implant devices [4]. Protein sorption onto PU in vitro and in vivo is well characterized within the literature, allowing MALDI-TOFMS to be compared with more traditional methods of analysis. In addition, our laboratory has experience with using PU membranes as sample supports for MALDI-TOFMS characterization of peptides and proteins derived from biological samples [19,25,26]. Polyurethane membrane material was exposed to plasma standards from three di!erent hosts. Mass spectra of the plasma standards on the PU biomaterial showed the adsorption of several proteins ranging from 6.5 to 150 kDa. Proteins observed in the mass spectra directly correlate with those shown by other researchers to be adsorbed from plasma onto PU-based materials, thus validating MALDI-TOFMS as a technique for biomaterial analysis. This type of application may also be used for screening among possible future biomaterials.
2. Experimental section 2.1. Reagents and materials All chemicals used in this study were of reagent grade unless otherwise stated. HPLC grade acetonitrile and methanol were from Fisher Scienti"c (Fair Lawn, NJ). Deionized water, prepared with a Milli-Q plus}TOC water puri"cation system (Millipore, Bedford, MA), was employed in all solutions. Plasma protein standards (human, bovine and canine) were obtained from Sigma Chemicals (St. Louis, MO). Sinapinic acid, used as the MALDI matrix, was from Sigma. Protein standards were used as MALDI calibrants. Horse heart myoglobin, 16 951 Da, bovine insulin, 5733 Da, bovine serum albumin, 66 430 Da, were purchased from Sigma, and bovine apotransferrin, 78 030 Da, was obtained from Calbiochem (LaJolla, CA). Solutions of these standards were made up in water, and used without further puri"cation.
2.2. PU membrane model biomaterial The non-porous ether-type PU membrane, 50 lm in thickness (XPR625-FS) was supplied by Stevens Elastomerics (Northampton, MA). The PU membrane used consists of a co-polymer of polytetramethylene ether glycol and diphenylmethane diisocyanate (MDI) [27]. The membrane was washed with water and methanol prior to use in order to remove polar and non-polar surface contaminants.
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
2.3. MALDI-TOFMS
3. Results and discussion
MALDI-TOFMS was performed in the linear mode on a re#ecting time-of-#ight mass spectrometer, an instrument built in-house in the Time-of-Flight Lab, University of Manitoba [28]. Accelerating potentials of 25 and 30 kV were used. Spectra were obtained using a nitrogen laser (337 nm) (VSL 337 ND, Laser Science Inc., Cambridge, MA, USA) with the #uence adjusted slightly above threshold. In order to avoid saturation of the detector by low mass matrix ions, the detector was pulsed on ca. 19 000 ns after each laser shot. External calibrations for measurements using the PU membrane were performed with standards (bovine insulin, equine myoglobin, human transferrin) prepared on similar membrane targets. Spectra presented here are summations of ca. 50}100 consecutive shots.
3.1. Analysis of a human plasma standard
2.4. Preparation of plasma standard samples Samples were prepared on PU based on protocols developed in our laboratory [19]. Samples were prepared on the membrane by "rst depositing 2 lL of the plasma solution onto the surface of the membrane and allowing it to dry. A mark was placed on the underside of the membrane to allow the sample spot to be located after washing. The membrane was then washed with 50 lL of water for 1 min to remove the non-adsorbed plasma components. The washing step was repeated. After drying, immediately before analysis, matrix (2 lL, saturated sinapinic acid in 1 : 1 H O/acetonitrile) was placed on the membrane and allowed to crystallize. The membrane was then trimmed to size, a$xed to the MALDI target using an adhesive spray (Spraymount, 3M) and subsequently analyzed by MALDI-TOFMS.
1703
The nature of protein sorption ultimately determines the in vivo stability of a material. PU possesses a twophase bulk structure consisting of hard (moderately polar) and soft (non-polar) segments as shown in Fig. 2a [29,30]. The chemical structure of each of these segments greatly in#uences the ability of the polymer to adsorb proteins [31]. Protein binding is believed to occur through a combination of hydrogen bonding and hydrophobic interactions between the protein and the surface of the polymer [32]. Several di!erent types of PU have been used and tested for the construction of biomedical devices [33]. For our study the polytetramethylene oxide (PTMO) methylene bis(p-phenyl diisocyanate (MDI)type PU (Fig. 2b) was chosen because it can adsorb larger amounts of protein than PU prepared with more hydrophilic soft segments can [29]. Previously, we have shown that MALDI-TOFMS resolution and mass accuracy for proteins directly deposited onto the PTMO/ MDI-type PU membrane used here were equivalent to those obtained using a metallic target [19,26]. Although MALDI-TOFMS is relatively tolerant of impurities such as salts and bu!er components, the removal of these species yields greater resolution and mass accuracy. In addition, proteins with no a$nity for the membrane material must be removed prior to analysis. To accomplish this a rinsing step may be performed. All of the components with a high a$nity for the biomaterial remain adsorbed while polar components and non-adsorbed proteins are washed o! the material. The washing
2.5. In vitro experiments In vitro experiments were performed at the Institut des BiomateH riaux du QueH bec, UniversiteH Laval. Samples of PU membrane material were cut into 2.0 cm;2.0 cm pieces and exposed to pooled samples of freshly collected canine plasma in neutral glass vials (Anchor Ltd, London, Ontario). Membrane samples were exposed (in triplicate) for periods of time from 0 to 4 h at 373C. Following exposure, the samples were immediately washed with water to remove non-adsorbed plasma components, dried under vacuum and later shipped to our laboratory at the University of Manitoba for MALDI-TOFMS analysis. 2.6. Safety considerations Suitable precautions should be taken when handling biological #uids. Gloves were worn during collection and manipulation of samples.
Fig. 2. (a) Two-phase bulk structure of PU membrane consisting of hard (moderately polar) and soft (non-polar) segments. (b) Chemical structure of PU used in these experiments.
1704
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
Fig. 3. MALDI-TOF mass spectra of human plasma obtained using PU as a sample support. (a) Low M region. (b) High M region. Tentative assignments are found in Table 1.
Table 1 Tentative assignments for human plasma proteins adsorbed on polyurethane and detected by MALDI-TOFMS Protein
Immunoglobulin G Transferrin Comp C3-b Albumin a -HS Glycoprotein Apo A-4 Apo E Apo A-1 Apo D Apo H-III (HDL) Transthyretin Apo C-II Apo C-III Apo C-I
Molecular weight (Da)
Relative peak area
Concentration
Literature
Experimental
RSD
% RSD
(%)
RSD
% RSD
(g/L)
150 000 79 754 75 000 66 472 49 000 43 400 34 236 28 000 22 500 17 000 13 761 8914 8764 6630
148 361 78 513 73 953 66 463 49 575 44 175 33 109 27 502 22 174 16 327 13 555 8946 8783 6631
1900 980 590 * 400 170 200 14 30 460 170 16 10 10
1.3 1.3 0.80 * 0.80 0.40 0.60 0.05 0.15 3 1.3 0.20 0.20 0.20
20 17 31 100 20 16 56 135 17 25 171 * * *
5 3 9 * 3 12 28 107 8 17 159 * * *
23 20 27 * 16 77 49 79 47 68 93 * * *
8}18 2.0}4.0 (1 35}55 0.4}0.85 (1 (1 (1 (1 3 (1 (1 (1 (1
Peak areas of high M proteins normalized to albumin. Concentration in whole blood plasma from Ref. [34]. Literature molecular weights from Refs. [34,35]. RSD and % RSD based on 4 replicate measurements. Albumin was used as the internal calibrant.
process is analogous to membrane puri"cation of proteins prior to MALDI-MS analysis [18,19]. Aliquots of human plasma standard (2 lL) were placed onto the surface of the membrane and allowed to dry. Following rinsing and application of matrix, the samples were placed onto a MALDI probe and inserted in the mass spectrometer for analysis. Fig. 3 shows MALDITOF mass spectra of human plasma proteins adsorbed onto the PU biomaterial (also see Table 1). In order to obtain reasonable signals in the high M range ('20 kDa) it was necessary to suppress the background from the matrix by turning on the microchannel plate detector on, only after the matrix ions had time to reach the detector. Good-quality MALDI mass spectra were thus obtained after removal of excess of salts and bu!er components present in plasma (Table 2). Ions from at least
14 di!erent proteins were visible. The proteins ranged in M from ca. 6.5 to 150 kDa. Mass resolution and accuracy were su$cient for identi"cation upon comparison with M values from the literature (Table 1) [34,35]. MALDI-TOFMS has been used for both the qualitative, semi-quantitative and quantitative analysis of protein and peptide mixtures [36,37] and recently for the quantitative measurement of single protein sorption onto polymer surfaces [24]. This investigation reports on the semi-quantitative and qualitative determination of several plasma proteins simultaneously adsorbed onto a PU membrane using relative MALDI-TOFMS peak intensities. Tentative mass assignments for each of the proteins [38] are shown in Table 1 along with integrated peak areas measured relative to albumin. Reproducibility in M measurements was good, however the % RSD
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710 Table 2 Concentration of bu!er components in plasma Component
Concentration (mM)
Bicarbonate Calcium Chloride Cholesterol Glucose Iron Magnesium Phosphate Potassium Sodium Sulfate Urea Uric acid
42}54 4.2}5.2 95}103 0.038}0.063 0.036}0.055 1.0}2.68;10\ 3.0}5.2 1.8}2.6 8}9.6 136}142 0.4}2.6 0.013}0.033 1.2}3.1;10\
Values adapted from Ref. [38].
measurements of relative peak areas ranged considerably from 16 to 93%. The principle error in peak integration was due to slight di!erences in mass spectra obtained on di!erent samples. Slight variations in the matrix preparation steps led to non-uniform co-crystallization of the plasma proteins with the matrix and the production of irregular crystals on the surface of the PU membrane. This resulted in variable signal intensity in the mass spectra dependent on the location of the target sampled with the laser. Additional care must be taken in the interpretation of the spectra. The MALDI signal intensity observed is not necessarily indicative of the amount of protein on the surface of the membrane, neither is the amount of a particular protein adsorbed onto the membrane indicative of its concentration in solution. The a$nity of the protein for the membrane plays a more signi"cant role than its concentration. This is demonstrated by the absence of ions of several major proteins, and by the presence of several low concentration proteins ions (Fig. 3b, Table 1). Only proteins with an a$nity for the membrane material are adsorbed and later detected, while those with low a$nity are washed o! [21,22]. Additionally, it is known that MALDI signal intensity is dependent on the amount of the protein and the ease with which it will co-crystalize with the matrix and ionize in the MALDI plume [39]. We have shown such discrimination phenomena for mixtures of medium and high M wheat proteins prepared on PU supports vs. steel targets [26]. Plasma proteins which are known to adsorb onto PU materials in vivo include lipoproteins [40,41], albumin [42], immunoglobulin G (IgG) [43], C3, "brinogen as well as others [4]. With the exception of "brinogen, these proteins were observed in the mass spectra obtained on PU (Fig. 3). This indicates a direct correlation between the proteins detected in the mass spectrum of human plasma on PU and proteins known to adsorb onto PU in vivo, determined using other techniques.
1705
Biomaterials development has focused on the adsorption of three proteins in particular, as indicators of biocompatibility [3]. The adsorption levels of albumin, IgG and "brinogen are used to assess the in vivo performance of a material. For biomaterials, albumin adsorption is bene"cial owing to this protein's relative lack of glycosylation, which prevents platelet adhesion. On the other hand, adsorption of IgG and "brinogen may cause a host defense reaction which increases platelet adhesion and promotes a physiological response. Since the adsorption of both albumin and IgG was observed in this study it follows that MALDITOFMS could be used to assess biocompatibility of di!erent materials by analyzing the amount of albumin or IgG adsorbed by a particular material. The third major protein, "brinogen, was not detected because of its high M . In addition, "brinogen is normally observed after longer periods of exposure than were used in this investigation. 3.2. Sample-to-sample reproducibility Initially, the mass spectra of human plasma proteins on PU exhibited relatively poor reproducibility in peak integration. Sample-to-sample reproducibility was explored with care taken to ensure that the preparation/ washing and subsequent MALDI-TOFMS measurements were obtained under nearly identical conditions. The results are shown in Fig. 4, for four replicate samples of bovine plasma standard prepared on PU. Each spectrum shown is the summation of 50 consecutive shots sampled over the entire surface of the target. Good shotto-shot reproducibility was obtained when samples were subjected to the same sample preparation and MALDI conditions. Relative peak areas and heights are given in Table 3. As with the human plasma proteins integration of peak areas resulted in relatively high % RSD values (mean 42%). Improved results were obtained by measuring the peak heights which had a mean % RSD of ca. 28%. Peak height variations overall were less than those obtained with integration and ranged from 9 to 54%. The larger #uctuation in peak areas arose mainly from variation in peak widths possibly due to inconsistent formation of adducts (i.e. with the matrix) from one sample to another. Di!erences in baseline assignment associated with manual integration may have also contributed to the larger error. Comparison of the two semi-quantitative sets of measurements and the errors associated with them are better visualized in Fig. 5. Here it can bee seen that, even though there exists some variance in the measurements, semi-quantitative results of peak area or height may still be obtained and are respective of the amount of protein adsorbed onto the surface of the membrane. Relative MALDI peak areas or heights can thus be used as a semi-quantitative assessment of the concentration
1706
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
Fig. 5. Graphical representation of statistical values from Table 3. Mean and standard deviation in (a) peak area, (b) peak height.
Fig. 4. Reproducibility of MALDI-TOFMS measurements obtained for four replicate samples of bovine plasma standards. Masses correspond to values in Table 3.
of di!erent proteins adsorbed onto the biomaterial. In addition, small mass shifts in the protein adsorption pro"le of a material can be determined and allow better biomaterial characterization.
Table 3 Reproducibility in MALDI-TOFMS measurements of bovine plasma proteins adsorbed on polyurethane, High M Molecular weight
Peak area
Peak height
(Da)
RSD
% RSD
(%)
RSD
% RSD
(%)
RSD
% RSD
12 571 13 657 14 881 15 946 22 034 22 831 27 470 33 153 38 796 48 942 53 007 66 433 73 324 77 606
161 126 147 192 690 510 193 179 292 397 517 * 279 341
1.27 0.92 0.98 1.20 3.10 2.21 0.70 0.54 0.75 0.80 0.97 * 0.38 0.44
75 341 93 28 26 22 157 57 12 9 16 100 31 17
38 104 24 7 21 23 35 5 7 3 8 * 13 8
51 30 26 24 82 103 22 8 61 35 52 * 42 49
288 551 216 96 47 48 190 78 28 27 28 100 40 36
48 87 21 20 12 12 58 10 12 13 15 * 16 19
16 15 9 21 25 24 30 12 44 50 54 * 40 54
Average % RSD
28
Average % RSD Albumin was used as the internal calibrant. Peak areas of high M proteins normalized to albumin. Peak heights of high M proteins normalized to albumin. RSD and % RSD based on four replicate measurements.
42
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
1707
3.3. Plasma standards of diwerent hosts
3.4. Inyuence of sample preparation conditions
Studies suggest that there are slight di!erences in protein adsorption patterns between plasma samples from di!erent hosts and that these are due to variations in hematological pro"les [44]. Plasma samples from three hosts (i.e. human, canine and bovine) were used to determine the ability of MALDI-TOFMS to distinguish between protein adsorption patterns for di!erent plasma standards. PU samples were exposed to each plasma standard and then analyzed to compare adsorption patterns. The spectra of three di!erent plasma standards adsorbed on PU are shown in Fig. 6. These results indicate that proteins of similar high M were adsorbed (Fig. 6b) for each plasma standard. There were di!erences in the low M pro"les (Fig. 6a). The human plasma standard result ed in the highest number of adsorbed proteins in the 5.5}10 kDa mass range, possibly lipoproteins. In comparison, the spectra obtained for bovine and canine plasma showed adsorption of fewer proteins in this region. The adsorption of lipids onto PU surfaces has been associated with the biodegradation of PU implants [45]. It has also been suggested that the interaction between lipoproteins and di!erent domains of the PU facilitates lipid adsorption [4]. The variations between protein adsorption patterns could lead to di!erences in biocompatibility, questioning the validity of testing biomaterials with other hosts [40]. MALDI-TOFMS is capable of distinguishing between di!erent protein adsorption patterns and therefore may be seen as a valid method for biomaterial analysis.
Di!erent sample preparation methods were tested for their e!ect on ion signal intensities and to further determine conditions which may be used for in vitro testing. In Fig. 7a, bovine plasma was applied onto the PU membrane followed by washing and addition of matrix. In Fig. 7b, methanol was applied following the application and drying of the plasma sample. The addition of methanol resulted in an increase in the overall MALDI signal. A similar phenomenon was reported in an earlier publication [19]. We associated this to enhanced protein binding on the membrane, an e!ect commonly exploited in SDS-PAGE when electroblotting onto porous membranes. Methanol enhanced protein adsorption was also described for MALDI-TOFMS on porous membranes [17]. In Fig. 7c, methanol was added to the sample before the plasma sample had a chance to dry on the membrane. This resulted in the precipitation of the proteins from the plasma. Alcohol precipitation is often used to remove proteins from plasma/serum in order to analyze
Fig. 6. MALDI-TOF mass spectra of three di!erent plasma standards: bovine, canine and human on PU membrane. (a) Low M region. (b) High M region.
Fig. 7. MALDI-TOF mass spectra of bovine plasma proteins on PU membrane using di!erent sample preparation methods. (a) Plasma, dried, washed, dried, matrix. (b) Plasma, dried, methanol, dried, washed, dried, matrix. (c) Plasma#methanol, dried, washed, dried, matrix. (d) Plasma#matrix solution, dried, washed, dried, matrix.
1708
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
other components. The result was an increase in signal intensity for some of the proteins, notably, albumin. Fig. 7d shows the e!ect of adding matrix solution to the sample before the plasma had a chance to dry on the membrane. In this case, the spectrum showed substantial di!erences relative to features observed in Fig. 7a}c. Even though the matrix material is dissolved in 50% (v/v) acetonitrile, it is acidic in nature. Protonation in solution will result in a net charge on the plasma proteins and will disfavor their sorption onto the predominantly hydrophobic membrane. The acidity of the matrix facilitates the removal of some bound proteins from the surface so they may co-crystallize with the matrix for analysis by MALDI [19]. The di!erent sample preparation protocols led to changes in the amount and nature of the proteins adsorbed and detected by MALDI. These discrepancies show that a consistent sample preparation method must be employed to directly compare protein adsorption patterns on di!erent biomaterials. With the addition of methanol to the samples (Fig. 7b and c) it was possible to improve spectral quality compared with the neat sample. However, these approaches do not re#ect solution conditions which would be observed in vivo. Therefore, the conditions chosen for the in vitro experiments where based on the "rst protocol used (Fig. 7a). 3.5. In vitro analysis In vitro testing of biomaterials is commonly performed because it is relatively inexpensive and faster than in vivo methods. In vitro experiments involve the isolation of #uid or tissue from a biological source, followed by experimental characterization of this material. In this study, the MALDI analysis of proteins adsorbed in vitro onto PU over di!erent times of exposure to canine plasma was performed. Canine plasma is commonly used by biomaterials researchers and represents the worst case scenario for protein adsorption since it produces extensive platelet adhesion and thrombus formation. Freshly collected canine plasma was placed in contacted with the PU material under speci"ed conditions to facilitate protein sorption. Triplicate samples were exposed to canine plasma for di!erent lengths of time (0, 0.5, 1 and 4 h) at 37C. Representative mass spectra showing the e!ect of the time of exposure of canine plasma on the amount of proteins adsorbed are shown in Fig. 8. In the control sample spectrum, (Fig. 8a) no peaks are observed. Analysis of the sample exposed to plasma for 0.5 h shows several peaks due to adsorption of proteins onto the surface of the PU biomaterial. Continued exposure, up to 4 h, yielded more intense protein signals. This may be interpreted as increasing amounts of protein being adsorbed on the surface of the polymer.
Fig. 8. MALDI-TOF mass spectra obtained on PU membranes exposed to canine plasma for (a) 0 (b) 0.5 (c) 1 and (d) 4 h.
These data supports the notion that protein adsorption occurs quite rapidly upon exposure of the polymeric material [1]. It has been suggested that for relatively short periods of exposure, protein adsorption "ts a Langmuir isotherm [46]. The distribution of proteins adsorbed at the surface of a material is, however, time dependent. Proteins are constantly deposited, and displaced and thus the surface composition changes [47]. The rates of deposition and displacement can be determined by the relative concentrations of proteins within the plasma and their a$nity for the material. Although proteins with a high plasma concentration may be adsorbed "rst, those with low concentrations but with a higher a$nity for a particular surface will eventually displace them. The displacement of proteins occurs after the surface of the polymer has become saturated. This phenomenon was not observed in this study, even over the four hour exposure. Instead, analysis of PU samples exposed to canine plasma in vitro yield spectra suggesting that the longer the exposure time, the larger the amount of protein present on the surface of the material. The mass spectral pro"le observed, although weaker in absolute intensity, was similar to the pro"le shown for the canine standard (Fig. 6b). This resemblance indicates a strong similarity between the proteins observed from the in vitro experiments and those observed for the canine plasma standard. The use of plasma standards for biomaterials testing may be su$cient as an initial
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
screening process for the biomaterial. Here, the use of plasma standards is preferred because it is faster, simpler and less expensive than performing either in vivo or in vitro experiments. 4. Conclusions Among the various kinds of interfacial phenomena occurring in the "eld of biomaterials, the adsorption of proteins at the interface between a liquid and a solid is one of the most signi"cant areas that has been investigated. The analysis of proteins adsorbed onto biomaterials is a new application of MALDI-TOFMS. The technique is able to analyze several proteins simultaneously with su$cient resolution to allow the easy characterization of adsorbed proteins based on literature M . Experiments performed in this study showed direct correlations between the proteins known to adsorb on PU from plasma and those seen in the mass spectra of plasma proteins on PU. The analysis of proteins adsorbed onto PU from plasma standards yielded good-quality mass spectra, displaying a variety of proteins from 6.5 to 150 kDa. The MALDI spectra exhibited good shot-to-shot reproducibility with only minor shifts in peak height. These features enabled the semi-quantitative analysis of adsorbed proteins. Comparison of plasma standards from di!erent hosts showed only small di!erences in the high M pro tein distribution, and more pronounced di!erences for low M proteins. Di!erences in sample manipulation were shown to have a profound e!ect on spectra, therefore care should be taken when developing experimental procedures for biomaterial characterization. This study clearly demonstrates that MALDITOFMS may be used to monitor the adsorption of di!erent proteins onto biomaterials. Therefore, this technique is helpful for determining the biocompatibility of the biomaterial in vivo. The method may be readily expanded to include testing implant devices removed from patients following implantation. Acknowledgements This work was supported by the National Science and Engineering Research Council of Canada (NSERC). We thank the members of the Time-of-Flight Laboratory, University of Manitoba, for helpful discussion. Finally, we wish to thank the University of Manitoba for a Graduate Fellowship awarded to M. E. McComb.
References [1] Vroman L, Adams AL, Klings M, Fischer GC, Munoz PC, Solensky RP. Ann New York Acad Sci 1977;283:65.
1709
[2] Baire RE. Artif Organs 1978;2:422. [3] J.H. Braybrook, Biocompatibility assessment of medical devices and materials, Biomaterials and Engineering Series, Wiley: New York, NY, 1997. [4] Lamba N, Woodhouse K, Cooper S. Polyurethanes in biomedical applications. Boca Raton, FL: CRC Press, 1998. [5] Brundstedt MR, Ziats NP, Robertson SP. J Biomed Mater Res 1993;27:367. [6] Osterberg E, Bergstrom K, Holmberg K, et al. Colloid Surf A 1993;77:159. [7] Lyman DJ, Metcalf LC, Albo D, Richards VK, Lamb. J Trans Am Soc Artif Intern Organs 1974;20:474. [8] Norman ME, Williams P, Illum L. Biomaterials 1992;13:841. [9] Bellissimo JA, Cooper SL. Trans Am Soc Artif Intern Organs 1984;30:359. [10] Holland NB, Qiu Y, Ruegsegger M, Marchant RE. Nature 1998;392:799. [11] Pitt WG, Grasel TG, Cooper SL. Biomaterials 1988;9:36. [12] Mrksich M, Sigal GB, Whitesides GM. Langmuir 1995;11:4383. [13] Green RJ, Davies J, Davies MC, Roberts CJ, Tendler SJB. Biomaterials 1997;18:405. [14] Davies J. Nanobiology 1994;3:5. [15] Janatova J. 23rd Anual Meeting of the Society For Biomaterials. USA: New Orleans, 1977. [16] Strupat K, Karas M, Hillenkamp F. Anal Chem 1994;66:464. [17] Blackledge JA, Alexander AJ. Anal Chem 1995;67:843. [18] Worrall TA, Cotter RJ, Woods AS. Anal Chem 1998;70:750}6. [19] McComb ME, Oleschuk RD, Manley DM, Donald L, Chow A, O'Neil JDJ, Duckworth HW, Ens W, Standing KG, Perreault H. Rapid Commun Mass Spectrom 1997;11:1716. [20] Chaurand M, Stoeckli M, Caprioli RM. Anal Chem 1999; 71:5263. [21] Schleuder D, Hillenkamp F, Strupat K. Anal Chem 1999; 71:3238. [22] Walker AK, Wu Y, Timmons RB, Kinsel GR, Nelson KD. Anal Chem 1999;71:268. [23] McComb ME, Oleschuk RD, Marois Y, King MW, Chow A, Ens W, Standing KG, Perreault H. Proceedings of the 46th ASMS Conference, 1998, p. 175. [24] Walker AK, Land CM, Kinsel GR, Nelson KD. J Am Soc Mass Spectrom 2000;11:62. [25] McComb ME, Oleschuk RD, Chow A, Ens W, Standing KG, Smith M, Perreault H. Anal Chem 1998;70:5142. [26] McComb ME, Oleschuk RD, Chow A. Perreault H, Dworschak RG, Znamirowski M, Ens W, Standing KG, Preston K. J Amer Soc Mass Spectrom, in submission. [27] Oleschuk RD. Separation isolation of metals using polyurethane materials. Winnipeg, MB: University of Manitoba, 1998. [28] Tang X, Beavis RC, Ens W, Lafortune F, Schueler B, Standing KG. Int J Mass Spec Ion Proc 1988;85:43. [29] Silver JH, Meyers CW, Lim F, Cooper SL. Biomaterials 1994;15:695. [30] Petrovic ZS, Fergenson. J Prog Polym Sci 1991;16:695. [31] Ratner BD. In: Polymers in medicine III. Amsterdam: Elsevier, 1988. p. 87. [32] Andrade JD. Surface and interfacial aspects of biomedical polymers. New York, NY: Plenum Press, 1985. [33] Goodman SL, Simmons SR, Cooper SL, Albrecht RM. Trans Soc Biomaterials 1989;15:201. [34] Putnam FW, The plasma proteins, vol IV. New york: Academic Press, 1984 (Chapters I and II). [35] http://expasy.hcuge.ch/sprot/sprot-top.html [36] Jesperson S, Niessen WMA, Tjaden UR, van der Greef J. J Mass Spectrom 1995;30:357. [37] Nelson RW, McLean MA. Anal Chem 1994;66:1408. [38] Anderson SC, Cockayne S. Clinical chemistry: concepts and applications. Toronto, ON: Saunders, 1993.
1710
R.D. Oleschuk et al. / Biomaterials 21 (2000) 1701}1710
[39] Xiang F, Beavis RC. Rapid Commun Mass Spectrom 1994; 8:199. [40] Dong DE, Andrade JD, Coleman DL. Polymer Prep 1983;24:40. [41] Takahara A, Takahashi K, Kajiyama T. J Biomater Sci Polym Ed 1993;5:183. [42] Kim SW, Lee RG, Oster H, Coleman D, Andrade JD, Lentz DJ, Olsen D. Trans Am Soc Artif Intern Organs 1974;20:449.
[43] Leonard EF, Vroman LJ. Biomater Sci Polym Ed 1991;3:95. [44] Grabowski EF, Didisheim P, Lewis CJ, Franta JT, Stropp JQ. Trans Am Soc Artif Intern Organs 1977;23:141. [45] Sreenivasan K, Jayabalan M, Rao KVC. J Applied Polym Sci 1992;45:2105. [46] Young BR, Pitt WG, Cooper SL. J Coll Interf Sci 1988;28:124. [47] Vroman L, Adams AL. J Biomed Mater Res 1969;3:43.