Proteomics Characterization of Protein Adsorption onto Hemodialysis

Dipartimento di Medicina Interna, Universita` di Roma Tor Vergata, Rome, Italy, and IRCCS-Fondazione. Santa Lucia, Rome, Italy. Received April 7, 2006...
0 downloads 0 Views 644KB Size
Proteomics Characterization of Protein Adsorption onto Hemodialysis Membranes Mario Bonomini,*,† Barbara Pavone,‡,§ Vittorio Sirolli,† Francesca Del Buono,‡,§ Moreno Di Cesare,† Piero Del Boccio,‡,§,⊥ Luigi Amoroso,† Carmine Di Ilio,‡,§ Paolo Sacchetta,‡,§ Giorgio Federici,|,⊥ and Andrea Urbani*,‡,§,⊥ Istituto di Clinica Nefrologica, Dipartimento di Medicina, Universita` “G. D’Annunzio” di Chieti-Pescara, Centro Studi sull’Invecchiamento (Ce.S.I.), Chieti, Italy, Dipartimento di Scienze Biomediche, Universita` “G. D’Annunzio” di Chieti-Pescara, Italy, Dipartimento di Medicina di Laboratorio, Policlinico di Tor Vergata and Dipartimento di Medicina Interna, Universita` di Roma Tor Vergata, Rome, Italy, and IRCCS-Fondazione Santa Lucia, Rome, Italy Received April 7, 2006

Protein-adsorptive properties are a key feature of membranes used for hemodialysis treatment. Protein adsorption is vital to the biocompatibility of a membrane material and influences membrane’s performance. The object of the present study is to investigate membrane biocompatibility by correlating the adsorbed proteome repertoire with structural feature of the membrane surfaces. Minidialyzers of identical structural characteristics composed of either cellulose diacetate or ethylenevinyl alcohol materials were employed to develop an ex vivo apparatus to investigate protein adsorption. Adsorbed proteins were eluted by a strong chaotropic buffer condition and investigated by 2-DE coupled to both MALDI-TOF mass spectrometry (MS) mass fingerprinting and fragmentation analysis on a nanoLCMS/MS hybrid instrument. Membrane surface characterization included evaluation of roughness (atomic force microscopy), elemental chemical composition (X-ray-photoelectron-spectroscopy), and hydrophilicity (pulsed nuclear magnetic resonance). The present study identifies a number of different proteins as common or characteristic of filter material interaction, showing that proteomic techniques are a promising approach for the investigation of proteins surface-adsorbed onto hemodialysis membrane. Proteomic analysis enables the characterization of protein layers of unknown composition. Keywords: protein adsorption • hemodialysis • membrane material • minidialyzer • proteomics • mass spectrometry

Introduction When blood is exposed to artificial membrane surfaces, protein deposition and adsorption occurs almost immediately, depending to a large extent on membrane surface characteristics such as hydrophilicity, roughness, charge, and chemistry.1 Adsorptive properties are an important feature of membranes used for chronic hemodialysis therapy in end-stage renal disease patients.2,3 Bio(in)compatibility of membranes, a source of justifiable concern in the hemodialysis field,4-6 is mainly * Authors for correspondence. Prof. Andrea Urbani, Ph.D., Centro Studi sull′invecchiamento (Ce.S.I.), Centre of Investigation on Aging, Universita` degli Studi di Chieti e Pescara, Via Colle dell′ Ara, Chieti Scalo, 66100 Chieti (Italy). E-mail, [email protected]; tel, +39-0871-541580; fax, +39-0871-541598. Prof. Mario Bonomini, MD, Clinica Nefrologica, PO SS. Annunziata, Via dei Vestini, I-66013, Chieti (Italy). E-mail, [email protected]; tel, +390871-540120; fax, +39-0871-574736. † Dipartimento di Medicina, Universita` “G. D’Annunzio” di Chieti-Pescara. ‡ Centro Studi sull’Invecchiamento (Ce.S.I.). § Dipartimento di Scienze Biomediche, Universita` “G. D’Annunzio” di Chieti-Pescara. | Dipartimento di Medicina di Laboratorio, Policlinico di Tor Vergata and Dipartimento di Medicina Interna, Universita` di Roma Tor Vergata. ⊥ IRCCS-Fondazione Santa Lucia.

2666

Journal of Proteome Research 2006, 5, 2666-2674

Published on Web 09/07/2006

governed by protein adsorption. Indeed, the interaction of activated proteins that have been adsorbed at the surface of a membrane may trigger several biological pathways, including the blood coagulation cascade, the complement and the fibrinolytic systems, and cellular mechanisms,7-10 with potential pathophysiologic consequences. On the other hand, adsorption by dialysis membranes of complement fragments C3a and C5a,11 factor D,12 and high molecular weight kininogens13 may increase biocompatibility profile. Adsorption may also contribute to the removal of some noxious compounds such as β2-microglobulin, tumor necrosis factor, and peptides.14,15 The protein layer coating the polymer, however, can reduce the therapeutic usefulness of a membrane by limiting its diffusive and convective capacity. In addition, it has been shown that certain types of synthetic polymers can bind hormones such as erythropoietin16 and medications,17 resulting in unwanted effects. Protein adsorption should thus be carefully controlled during the development of biomaterials for hemodialysis therapy.18 Adequate protein adsorption studies should evaluate the amount, composition, and conformational change of proteins 10.1021/pr060150u CCC: $33.50

 2006 American Chemical Society

research articles

Hemodialysis Membrane Proteomics Table 1. Structural Characteristics of Membranes Equipping Minidialyzers Used in the Experimental Model Materials Effective length Number of fibers Effective surface area Sterilization

Cellulose Diacetate EVAL 80 mm 300 130 cm2 γ-ray

adsorbed onto the surface. However, detailed analysis of protein adsorption onto dialysis membrane materials has been limited by the absence of adequate protein separation and identification techniques. Numerous physicochemical techniques have been adapted to study protein adsorption in situ, but they remain essentially bound to rather simple model systems.1 Immunological methods can only be used to study a small number of proteins and are limited to those proteins for which a sensitive and precise assay is available. Furthermore, standard analytical techniques may not differentiate between the native protein and modified form(s) which may render the protein toxic. For example, glycation of β2-microglobulin has been implicated in dialysis-associated amyloidosis.19 In the post-genomic era, proteomic techniques have been developed to investigate a large number of proteins simultaneously.20 Proteomic methods allow separation and identification of single proteins (including post-translational modifications) from a complex mixture without requiring assumptions as to which protein may be present.21 A proteomic analysis is expected to be capable of providing distribution of proteins on biomaterials with high sensitivity and resolution.22 In the present study, we aimed to ascertain the suitability of proteomic techniques for examining protein adsorption onto hemodialysis membranes and the effect of membrane surface characteristics on protein adsorption. We investigated the protein-binding characteristics of two different materials used for hemodialysis membranes: cellulose diacetate and ethylenevinyl alcohol. In prospective, randomized, crossover studies, we had previously shown a different biocompatibility pattern between the two materials, the cellulosic-derived membrane being much more reactive with blood components.23

Materials and Methods Protein Binding to Minidialyzers. The commercially available dialysis membranes tested included cellulose diacetate (CDA; Hospal, Italy) and ethylenevinyl alcohol (EVAL; Kuraray, Japan). The hollow fiber minidialyzers used in the experimental model were developed as a small scale model of a standard hollow fiber dialyzer. Their structural characteristics are shown in Table 1. Minidialyzers were wet with 0.9% NaCl in water before exposing them to blood from healthy donors (Blood Transfusion Centre, SS. Annunziata Hospital, Chieti, Italy). Whole blood samples were collected in K3EDTA vials (10 mL) and allowed to flow into the minidialyzer by a peristaltic pump in a closed loop system (Figure 1), under meticulous experimental technique.24 The apparatus flow was set to 1 mL/min, and the blood circulated for 30 min before the system was purged with phosphate-buffered saline (PBS). The filters were washed extensively with PBS at a 1 mL/min flow rate. Adsorbed protein was then eluted by a strong chaotropic solubilization buffer containing 6 M urea (Sigma), 2 M thiourea (Sigma), 4% CHAPS (Sigma), and 2% biolytes (Amersham Pharmacia), supplemented with a reducing agent, 40 mM dithiothreitol (Sigma). One milliliter of solubilization buffer was introduced

with a syringe and allowed to stand for 15 min before elution in a polypropylene tube (final volume 800 µL). A successive elution step using either the chaotropic solubilization buffer or an SDS-containing buffer (2% SDS, 40 mM DTT, and 50 mM TRIS-HCl, pH 8) was, as well, evaluated. One milliliter of solubilization buffer was introduced with a syringe and allowed to stand for 1 h before elution in a polypropylene tube (final volume 1000 µL). Total protein quantification was pursued by the bicinchoninic acid method (BCA, Pierce), following the room-temperature protocol supplied by the vendor and using, for the calibration curve, certified BSA (Bio-Rad) standards diluted in the solubilization buffer. 2-DE Protein Separation. Solubilized proteins were employed as directly eluted from the minidialyzer. 2-DE was performed using the immobiline-polyacrylamide system as described.25,26 The IPG (first dimension) was carried out on a nonlinear immobilized pH gradient, pH 3-10; 18 cm long IPG strips (Amersham Biosciences). First dimension IPG strips were loaded with 80 µg of total proteins for analytical and 300 µg of total proteins for preparative gels, respectively. The samples were diluted to a total volume of 340 µL in rehydration solution, which contained 6 M urea, 2 M thiourea, 4.0% CHAPS, 16 mM DTT, and 0.8% (w/v) carrier ampholytes (IPG Buffer 3-10; Amersham Biosciences). The strips were rehydrated and focused on an IPGphor (Amersham Biosciences) at 20 °C. Rehydratation was allowed for 12 h at low voltage (6 h at 30 V and 6 h at 60 V). Protein focusing was obtained by application of 250 V for 1 h, 500 for 1 h, 1000 for 30 min, and 8000 for 4.5 h. The current was limited to 50 µA per strip. After first dimension electrophoresis, IPG strips were equilibrated for 15 min in 6 M urea, 30% (v/v) glycerol, 2% (w/v) SDS, 0.05 M Tris-HCl, pH 6.8, and 2% (w/v) DTT, and subsequently for 15 min in the same urea/SDS/Tris buffer solution but substituting the 2% (w/v) DTT by 2.5% (w/v) iodoacetamide. The second dimension was carried out on 12%T polyacrylamide gels (18 cm × 20 cm × 1.5 mm) at 200 V/gel and 14 °C until the dye front reached the bottom of the gel.27 Monodimensional PAGE gels were obtained according to Laemmli27 after protein precipitation at -20 °C for 48 h with 6 mL of a solution containing 50% ethanol, 25% methanol, and 25% acetone. Analytical gels were stained with silver nitrate fixing the proteins by glutaraldehyde. Preparative gels for mass spectrometry analysis were stained with Colloidal Coomassie.28 2-DE Pattern Analysis. Computer-aided 2-D image analysis was carried out using the PDQuest 7.2 software (Bio-Rad). A synthetic gel was constructed out of four independent analytical gels from each investigated minidialyzer group. The relative abundance of proteins in both samples was evaluated as the integrated density of the protein spot determined by the same imaging software. Comparative analysis was performed on protein spots with a relative intensity above 0.01 and only considering differences above a 3-fold increment or decrease. Protein Identification and Characterization by Mass Spectrometry. Protein identification and characterization were carried out by both MALDI-TOF-MS mass fingerprinting and by fragmentation analysis on a nanoLC-MS/MS hybrid instrument. Excised spots from preparative 2D gels were in-gelreduced, thiol-alkylated, and digested with sequence grade porcine trypsin (Sigma)29 in 50 mM ammonium bicarbonate (Sigma) at 37 °C for 16-18 h. The reaction was stopped by addition of 0.1% TFA (Fluka); samples were frozen at -80 °C. Journal of Proteome Research • Vol. 5, No. 10, 2006 2667

research articles

Bonomini et al.

Figure 1. Diagram of sample preparation system.

MALDI mass spectra were recorded in the positive ion mode with delayed extraction on a Reflex IV time-of-flight instrument equipped with an MTP multiprobe inlet and a 337 nm nitrogen laser. Mass spectra were obtained by averaging 100-300 individual laser shots. Tryptic peptides were extracted by ZipTip C18 (Millipore) reverse phase material and directly eluted and crystallized in a 50% acetonitrile/water (v/v) saturated solution of R-cyano-4-hydroxycinnamic acid. A 50 pmol/µL mix solution of Angiotensin I (1296.68 Da), ACTH 18-39 (2465.19 Da), Bradykinin (fragment 1-9, 1060.57 Da), [Glu1]-Fibrino peptide B (1570.68), and Renin Substrate (1758.93 Da) was used for the external calibration standards. Internal spectrum calibration was performed by a three-point linear fit using the autolysis products of trypsin at m/z 842.50, 1045.56, and 2211.10. A database search with the monoisotopic peptide masses was performed against the NCBI nonredundant database using the peptide search algorithm MASCOT (Matrix Science, http:// www.matrixscience.com). Mass tolerance of 100 ppm, single miss cleavage site per peptide fragment, carboamidomethyl modification of cysteine residues, and the optional presence of methionine oxidation were employed in the database search. Masses corresponding to keratin tryptic fragments or evaluated as environmental contaminants by specific blank controls were excluded. Protein identification by peptide fragmentation analysis was performed on extracted peptide containing injected solutions (6 µL) using a CapLC system (Micromass, Waters) coupled online with a nano-ESI-Q-TOF instrument (Micromass, Waters). The sample was first concentrated into a Waters Symmetry300 C18 5 µm OPTI-PAK Trap Column, 0.35 × 5 mm, by a carrier solvent (water with 0.2% of formic acid) at 20 µL/min for 3 min and subsequently eluted at 0.2 µL/min (using a precolumn splitter) on a C18 column LC-Packings DIONEX PepMap 5 µm, 100 Å ,75 µm inner diameter × 250 mm, with a water/ acetonitrile gradient in the presence of 0.2% formic acid. Peptides were analyzed with a Q-TOF mass spectrometer (Micromass, Waters) equipped with a nano-Lock-Spray source, using a [Glu1]-Fibrino peptide B 500 nM solution in H2O/AcN (1:1) with 0.2% of formic acid, as a reference compound for on-line recalibration data at m/z ) 785.84 [M + 2H]+. A 2.5 kV tension was applied on the PicoTip capillary (NEW OBJECTIVE, PicoTip Emitter; Tip, 10 ( 1 µm). Argon was used as a collision gas. MS/MS spectra were acquired by automatic switching between MS and MS/MS mode. Acquired MS/MS data were converted into a centroid format by ProteinLinx 2.0 software (Micromass) and analyzed using MASCOT MS/MS ion search 2668

Journal of Proteome Research • Vol. 5, No. 10, 2006

Figure 2. Quantitative adsorption of proteins onto cellulose diacetate (CDA) and EVAL membranes contained in minidialyzers. Results are mean ( standard deviations of three experiments. P < 0.01 EVAL vs CDA.

engine (Matrix Science, http://www.matrixscience.com) with NCBInr database. The query was restricted to human proteins, the maximal tolerance for peptide masses was 50 ppm, and the maximal tolerance for MS/MS data was 0.4 Da, searching peptide charge of 2+ and 3+. Peptide modifications were defined as previously reported for mass fingerprint analysis. Analytical Methodologies. The following analytical methodologies were used to characterize the blood-contacting surface of investigated membrane materials: 1. Atomic Force Microscopy. Atomic force microscopy (AFM) has the potential to investigate surface roughness of a material on a nanometer scale. For each of the two test samples examined in this study, images of the inner surface were measured in the air and in water by AFM in a contact mode (DI3100, Veeco Instruments, New York, NY). Both a silicon single-crystal probe (NPS; cantilever length, 200 µm; force constant, 0.12 N/m; resonance frequency, 5-50 kHz) and a tube-type piezo scanner (maximum scan area, 85 µm × 85 µm; maximum scan height, 3.9 µm) were used. The scanning area was 5 µm × 5 µm, scanning speed was less than 2 Hz, and

Hemodialysis Membrane Proteomics

research articles

Figure 3. Proteome maps of proteins eluted from minidialyzers containing cellulose diacetate (A) membranes. Experiments were carried out with blood from healthy donor subjects. The proteins were separated by 2-D PAGE based on their isoelectric points (x-nl gradient pH3-10) and their molecular size. Only protein identifications that achieved a significant MOWSE score are shown in the figure. Spot numbers correspond to the spot number listed in Table 2. Histograms (B) show proteins differentially bound to the materials investigated.

image size was 512 × 512 pixels. Surface roughness was quantified as the mean roughness (Ra) ) the mean value of the surface relative to the center plane.

2. X-ray-Photoelectron-Spectrometry. X-ray-photoelectronspectrometry (XPS) quantitatively determines the elemental composition of solid surfaces to a sensitivity of about 0.1 atom Journal of Proteome Research • Vol. 5, No. 10, 2006 2669

research articles

Bonomini et al.

Figure 4. Proteome maps of proteins eluted from minidialyzers containing EVAL membranes. Experiments were carried out with blood from healthy donor subjects.The proteins were separated by 2-D PAGE based on their isoelectric points (x-nl gradient pH3-10) and their molecular size. Only protein identifications that achieved a significant MOWSE score are shown in the figure. Spot numbers correspond to the spot number listed in Table 2. Histograms (B) show proteins differentially bound to the materials investigated.

percent. The surface elemental composition of the inner surface of the two membranes was determined (AXIS-His, Shimadzu, Kyoto, Japan) after splitting open the hollow fibers longitudinally. 3. Pulse Nuclear Magnetic Resonance (NMR). The membrane sample was set in the NMR sample tube of 7 mm o.d. The sample was maintained at the probe temperature for 30 min before NMR measurement. The mobility of a polymer chain (carbon) of membrane material was evaluated by the transverse relaxation time T2 using the Solid Echo method with a Bruker PC-120 NMR (resonant at 20 MHz). Determination of 2670

Journal of Proteome Research • Vol. 5, No. 10, 2006

the water mobility in each polymer was carried out by a JEOL Mu25 instrument (JEOL Ltd. resonant at 25 MHz). Spin-lattice (T1) and spin-spin (T2) relaxation times were measured by the inversion recovery and the spin-echo [Carr-PurcellMeiboom-Gill] methods, respectively.30

Results 2-DE and Protein Identification. Whole blood proteins were adsorbed onto EVAL and cellulose diacetate membranes through a minidialyzer apparatus. The membranes were then washed with a strong chaotropic buffer in order to release proteins

research articles

Hemodialysis Membrane Proteomics Table 2. Figure of Merit for Mass Spectrometry Identifications of 2-DE Protein Spotsa

spot

score

no. matched peptides

1 2 3

8 9 10 11 12 13 14

110 222 164 443 178 68 84 122 268 73 67 118 150 172 113 155

14 19 18 19 23 13 8 12 5 6 9 10 11 13 10 2

22% 36% 32% 29% 50% 33% 70% 45% 20% 39% 48% 68% 74% 88% 71% 23%

37747855 23307793 23307793 4502027 37499461 45439306 71727271 178775 178775 32189392 999580 71727271 71727271 71727271 71727271 219978

C1 C2 C3 C4 C5 C6 C7 C8 C9 C10

64 150 47 66 104 67 84 167 88 77

2 19 1 11 12 12 11 5 12 9

2% 34% 1% 26% 26% 23% 20% 5% 21% 65%

23307793 23307793 23307793 6013427 23307793 23307793 23307793 23307793 23307793 71727271

90 68 68 76

13 7 8 8

32% 31% 61% 41%

1942629 32189392 32189392 4502419

4 5 6 7

E1 E2 E3 E4

sequence coverage

no. NBCI

name

MW

pI

comments

transferrin serum albumin serum albumin albumin precursor apolipoprotein a-iv aspartyl-tRNA synthetase beta globin proapolipoprotein proapolipoprotein peroxiredoxin 2 isoform a carbonic anhydrase i beta globin beta globin beta globin beta globin transthyretin CDA serum albumin serum albumin serum albumin serum albumin precursor serum albumin serum albumin serum albumin serum albumin serum albumin beta globin EVAL alpha-1-antitrypsin peroxiredoxin 2 isoform a peroxiredoxin 2 isoform a biliverdin reductase b (flavin reductase h)

79310 71344 71344 71317 45371 57499 16101 28944 28944 22049 28792 16101 16101 16101 16101 16023

6.97 6.13 6.13 5.92 5.28 6.11 7.86 5.45 5.45 5.66 6.63 7.86 7.86 7.86 7.86 5.52

MFP MFP MFP nLC-MS/MS MFP MFP MFP MFP nLC-MS/MS MFP MFP MFP MFP MFP MFP nLC-MS/MS

71344 71344 71344 71176 71344 71344 71344 71344 71344 16101

6.13 6.13 6.13 5.91 6.13 6.13 6.13 6.13 6.13 7.86

nLC-MS/MS MFP nLC-MS/MS MFP MFP MFP nLC-MS/MS nLC-MS/MS nLC-MS/MS MFP

44280 22049 22049 22219

5.37 5.66 5.66 7.13

MFP MFP MFP MFP

a Threshold MASCOT scores for significance at p < 0.05 were 63 for MFP experiments and 32 for nLC-MS/MS data. Protein spots named by a single number are common to both filter 2-DE maps. MW and pI values refer to the complete protein sequences.

which had been adsorbed. These procedures were repeated in three independent experiments employing samples from different donors. The total protein adsorption was significantly higher (p < 0.01) for the cellulose-derived membrane than for the synthetic EVAL membrane (Figure 2). In order to possibly evaluate a secondary elution profile from the surface material, we have performed a successive elution step using 2% SDS buffer or the described strong chaotropic buffer. Both solutions extracted a small amount of proteins below our limit of detection, LOQ < 0.05 mg/mL (data not shown). Moreover, monodimensional SDS-PAGE analysis of the concentrated solutions did not show any significant protein bands difference (data not shown). To investigate the protein composition of the adsorbed material in a comparative setup, we performed a coordinated experiment on blood samples collected from a single healthy donor (male, age 35 years) with no reported familial amyloidosis. Proteins obtained from EVAL and cellulose diacetate membrane were separated by linear gradient 2D electrophoresis. To rule out experimental variations not due to the characteristics of the two filters, four gels were analyzed for each sample and two synthetic images were built by computer matching, using the PDQuest software program. Each synthetic image represents all spots constantly present in all gels from the same material. When this procedure was used, the average total number of protein spots was 284 for cellulose diacetate and 280 for EVAL. Quantitative differences were considered as significant when showing at least 3-fold variation in the spot relative to volume.

Comparison of cellulose diacetate synthetic gel (Figure 3A) to EVAL synthetic gel (Figure 4A) showed few common spots in both gels: these were identified by mass spectrometry (Table 2) and preliminary compared by direct inspection with 2D protein maps of plasma and red blood cells available in the database (http://www.expasy.org/cgi-bin/map2/def?PLASMA _HUMAN; http://www.expasy.org/cgi-bin/map2/def?RBC_ HUMAN) (data not shown). Blood-abundant proteins such as albumin, hemoglobin, transferrin, and carbonic anhydrase are visible in both 2D maps of the filters investigated. This evidence would suggest a strong mass action equilibrium in protein binding. Moreover, the presence of a hemoglobin β chain would suggest an erythrocyte shear stress on both membrane materials, resulting in cell rupture. However, this process is limited since we did not observe massive hemolysis in the sample after exposure to the filter material. Some significant spot volume differences were found (Figures 3B and 4B), in particular for 10 spots increased in cellulose diacetate material and 4 spots increased in EVAL filter. These were digested with trypsin and identified by either peptide mass fingerprinting or nLC-MS/MS peptide sequencing analyses followed by a database search. The protein molecular weights of 2D-separated proteins were compared to their theoretical peptide masses as calculated for all proteins in the NCBInr database. Cellulose diacetate material showed a differential distribution of the proteins adsorbed, and a shift to a higher molecular weight than occurred with EVAL. Most of the differentially enriched proteins were, in fact, albumin fragments or isoforms Journal of Proteome Research • Vol. 5, No. 10, 2006 2671

research articles

Bonomini et al.

Figure 5. Atomic force microscopy images of the inner surface of cellulose diacetate (CDA) and EVAL membranes in water. A 5 µm square on one side was examined. The vertical axis is 50 nm.

which had shown a stronger affinity for the cellulose diacetate material. EVAL material specifically retains a lower number of proteins, and among them, peroxiredoxin 2 protein isoforms both shifted in the pI compared to the common isoform present in spot 8. Moreover, alpha-1-antitrypsin and biliveridin reductase were retained significantly more by this material. Characterization of the Membrane Blood-Contacting Surface. To assess the physicochemical determinants of such a differential protein distribution, we investigated the membrane material composition and microscopic surface properties. X-ray analysis was employed to obtain the surface chemical composition of each membrane. This investigation showed that the major constituents of both membrane surfaces under investigation are the elements carbon and oxygen without traces of nitrogen and sulfur. For the EVAL specimen, the carbon percentage was 80 and the oxygen percentage 20; for cellulose diacetate, the percentage of carbon and oxygen was, respectively, 73 and 27. These data indicate a substantially homogeneous atomic composition of the surface regardless of the different raw elemental material composition which would return for EVAL 75% carbon and 25% oxygen; and for cellulose diacetate 56% carbon and 44% oxygen. The surface topography of membranes was examined using AFM analysis in order to assess possible differential surface conformations of the filter materials investigated. Figure 5 shows 3-D atomic force microphotographs of cellulose diacetate (CDA) and EVAL in water at a scan size of 5 µm × 5 µm. It is apparent that the EVAL membrane inner surface is smoother than the cellulose diacetate membrane, that is, there are fewer differences between the surface elevations and depressions. Similar results were observed in AFM images of membrane inner surfaces as examined in air (data not shown). The mean roughness of the membrane inner surface (Ra value) proved to be lower for EVAL than for cellulose diacetate when measured either in air (1.6 nm vs 2.0 nm, respectively) or in water (1.5 nm vs 2.0 nm, respectively). The hydratation behavior of the membrane materials was studied by investigating water molecule mobility on the polymer surface by the pulse NMR method (Table 3). In the table, longer relaxation times (T1, T2) indicate higher mobility. These results highlight the stronger interaction of EVAL with water than with the other material, which is in agreement with a reduced unspecific protein binding effect. 2672

Journal of Proteome Research • Vol. 5, No. 10, 2006

Table 3. Relaxation Time of Water Molecules Inside Membranea T2 (s)

a

specimen

T1 (s)

A

B

EVAL CDA

1.19 2.28

0.21 0.265

0.77 1.04

T2 is resolved into two components, A and B.

Discussion Membranes used for hemodialysis therapy should ideally remove uremic-retained solutes of a defined molecular-weight range and offer minimal activation to blood components following blood-material interaction. Protein adsorptive properties of the membrane material strongly influence both properties, in particular biocompatibility, having either salutary or detrimental effects. Thus, qualitative and quantitative evaluation of a material’s affinity for proteins is essential for assessing the performance of the material and for development of new biomaterials for specific applications with higher biocompatibility. Progress in protein adsorption analysis has long been hampered by the lack of suitable technology. An initial pioneering work to hemodialysis membrane, protein interaction was pursued by the group of J. L. Brash employing monodimensional SDS-PAGE combined with Western blot analysis on tentative candidate proteins.31,32 Such an approach has, in fact, open the route to a systematic evaluation of protein adsorption on different filter material. An effective evaluation of the various components of surface-adsorbed proteins requires a datadriven, high-throughput technology that can monitor and characterize the adsorption of several proteins simultaneously. Proteome analysis has emerged as a new field of protein science offering the possibility to achieve unbiased identification of all proteins and peptides present in biological samples, therefore, overcoming the initial limitations linked to immunological analysis mainly driven by scientist hypothesis.31,32 In the past few years, proteomic approaches have also been increasingly and successfully applied to the study of a range of issues in biomaterial research including surface protein adsorption.22,33-37 Investigations by proteomic techniques of proteins adsorbed onto membrane materials used for hemodialysis treatment, however, have been very scant. Time-of-

Hemodialysis Membrane Proteomics

flight secondary ion mass spectrometry imaging showed distribution of bovine serum albumin adsorbed on the crosssection and the inside and outside surfaces of three commercially available hollow-fiber dialysis membranes.38 In a study published only in abstract form,39 2-DE and peptide mass fingerprinting allowed identification of some proteins which had been adsorbed onto a polyamide membrane dialyzer obtained after clinical treatment. Again, apolipoprotein A-1 (the major component of HDL) was found by SDS-PAGE and immunoblots as a significant component of the protein layer adsorbed from blood by many biomaterial interfaces including a polysulfone hemodialysis membrane.34 In this study, we employed a complete proteomics investigation coupled to a material structural analysis to investigate protein adsorption by hemodialysis membrane materials in a novel experimental model. Direct MALDI-TOF-MS has been suggested as a convenient method for the characterization of proteins adsorbed directly onto biomaterials.33,36,37 However, the poor mass resolution of this method even in the medium molecular weight regimen (above 20 kDa) hampers the discriminatory power of this technology in highlighting subtle protein modifications. We have developed a minidialyzer setup to ex-vivo investigate blood protein adsorption on different surface materials. This setup allows one to recover adsorbed protein by applying a strong chaotropic buffer after removal and washing of cell and protein unbound fraction. To the best of our knowledge, this is the first study on a proteomics analysis of proteins adsorbed from plasma onto the surface of polymeric materials used as the dialyzer membrane in clinical hemodialysis therapy. We examined and compared a cellulosic-derived membrane material (cellulose diacetate) and a synthetic polymer, EVAL. Several proteins were found to be adsorbed onto membranes, though we arbitrarily set the lower threshold for protein identification at a 0.01 relative spot volume of proteins present in proteome maps. It should be stressed, however, that the only aim of the present study was to demonstrate the possibility of identification following the isolation of compounds by a proteomic approach. The two membranes investigated showed similar adsorptive properties for several proteins such as albumin, which was the most abundant protein identified in the eluate fraction obtained from each membrane. However, albumin fragments form one of the main features in the differential adsorption of the two materials. Cellulose diacetate in fact showed a stronger selectivity in binding a number of albumin fragments, which show differences both in the isoelectric point and in the molecular weight associated with the main plasma albumin species. These differences might be associated either with the adsorption of albumin fragments (isoforms) originally present in the blood of the enrolled healthy donor, or with the onset of different molecular species upon binding to the membrane. Such a large distribution of albumin fragments have been observed as well in Western blot studies32 where cellulose acetate membrane was showing at least eight immunoreactive bands. Nevertheless, the clinical and physiological impact of such a behavior remains to be evaluated in experiments employing blood samples from uremic patients. The presence in desorbed fraction from each minidialyzer of erythrocyte proteins, including the hemoglobin β chain, carbonic anhydrase, and peroxiredoxin 2, might represent a sign of shear stress with consequent partial hemolysis. However, the EVAL material showed an increased affinity for two isoforms of peroxiredoxin 2 which

research articles were both more acidic and more basic than the common protein spot. Furthermore, since protein adsorption mainly depends on the surface characteristics of a polymer, it could be possible to approach the polymer’s hemocompatibility and performance through surface modifications. However, some caution is needed at this early stage of development, and a deeper investigation of the low-abundance proteins might reveal other candidate proteins differentially bound to filter materials. In addition, we did not check for possible adsorption onto the membrane surface of post-translationally modified protein forms which are frequently detected in uremia40 and might be of pathophysiologic significance, since for the present experiments we used blood from normal subjects. Differences in the protein makeup of uremic plasma (higher protein concentrations, post-translationally modified proteins) as compared to normal plasma may cause differences in the nature of proteins adsorbed to membranes and to their binding affinity. These issues are actually being addressed in ongoing studies in our laboratory, using blood obtained from patients on chronic hemodialysis. In analyzing the protein adsorptive properties of a membrane material, it pays also to characterize the amount of proteins adsorbed, since the latter is another major factor in evaluating the blood compatibility of materials used for a medical device. We quantified protein concentration in the eluate obtained from each membrane and found a remarkably higher protein amount for CDA than for EVAL. Thus, the two membrane materials differed not only in the type but also in the amount of proteins adsorbed from blood onto their surface. This is in keeping with the concept that each dialysis membrane has multiple and different properties that may contribute, more or less favorably, to interactions with blood components. Protein adsorption onto biomaterials for medical applications depends on the surface characteristics of the polymer and on the actual blood composition. Surface roughness and surface property, hydrophilic or hydrophobic, are important determinants of the biocompatibility and functional characteristics of dialysis membranes.1 Our data show that the surface of EVAL membrane has a more regular lattice than that of CDA, both in air and in water. In addition, EVAL membrane has stronger interactions with water, as evidenced by NMR investigation suggesting higher hydrophilicity than cellulose diacetate. The lower the hydrophilicity of a membrane, the greater the adsorption of proteins during use.41 Thus, differing roughness and hydrophilicity of membrane surfaces might at least partly explain the different protein adsorption pattern of membranes. This , in its turn, might underlie the different blood reactivity displayed by the membranes when used in clinical hemodialysis, EVAL showing less thrombogenicity and cellactivating properties.23,42,43 There are many applications of proteomics foreseen for nephrology, as indicated in a recently edited volume on the topic.44 In the field of uremia, proteome analysis may result in the identification of many more uremic-retained solutes than those presently known.40,45 The current study suggests a potential new use for proteomic approaches to issues related to nephrology, namely, the study of protein adsorption onto membranes used in hemodialysis therapy. Proteomics may be used to monitor the adsorption of different proteins onto biomaterial surfaces and hence for assessing the in vivo biocompatibility of the biomaterial. The present study suggests Journal of Proteome Research • Vol. 5, No. 10, 2006 2673

research articles that these investigations might be eminently suitable to evaluate the protein adsorption processes on biomaterials used for dialysis membranes. Application of proteomic techniques might consequently favor the development of more biocompatible membranes, to the potential benefit of the uremic patient. Whether this concept holds true requires further investigation.

Acknowledgment. We are particularly grateful to Kuraray Analytical Technology Centre for structural characterization of filter material. We thank Mr. Fabrizio Di Giuseppe for technical assistance in running 2-DE. This study has been supported by the Italian Minister of Research PRIN 2004055300_003. References (1) Werner, C.; Jacobasch, H.-J. Int. J. Artif. Organs 1999, 22, 3, 160176. (2) Chanard, J.; Lavaud, S.; Randoux, C.; Rieu, P. Nephrol., Dial., Transplant. 2003, 18, 252-257. (3) Pascual, M.; Tolkoff-Rubin, N.; Schifferli, J. A. Kidney Int. 1996, 49, 309-313. (4) Bonomini, V. Nephrol., Dial., Transplant. 1991, 6 (Suppl.): S1S3. (5) Basile, C.; Drueke, T. Nephron 1989, 52, 113-118. (6) Hakim, R. M. Kidney Int. 1993, 44, 484-494. (7) Lonnemann, G.; Koch, K. M.; Shaldon, S.; Dinarello, C. A. J. Lab. Clin. Med. 1988, 112, 76-86. (8) Anderson, J. M.; Bonfield, T. L.; Ziats, N. P. Int. J. Artif. Organs 1990, 13, 375-382. (9) Johnson, R. Nephrol., Dial., Transplant. 1994, 9 (Suppl. 2), 3645. (10) Franck, R. D.; Weber, J.; Dresbach, H.; Thelen, H.; Weiss, C.; Floege, J. Kidney Int. 2001, 60, 1972-1981. (11) Cheung, A. K.; Parker, C.; Wilcox, L.; Janatova, J. Kidney Int. 1990, 37, 1055-1059. (12) Pascal, M.; Schifferli, J. Kidney Int. 1996, 43, 903-911. (13) Valette, P.; Thomas, M.; Dejardin, P. Biomaterials 1999, 20, 16211634. (14) Goldman, M.; Dhaene, M.; Vanherweghem, J. L. Nephrol., Dial., Transplant. 1987, 2, 576-577. (15) Bouman, C. S.; van Olden, R. W.; Stoutenbeek, C. P. Blood Purif. 1998, 16, 261-268. (16) Cheung, A. K.; Hohnholt, M.; Leypoldt, J. K.; DeSpain, M. Blood Purif. 1991, 9, 153-163. (17) Rumpf, K. W.; Reiger, J.; Anjorg, R.; Doht, B.; Scheler, F. Proc. Eur. Dial. Transplant. Assoc. 1977, 14, 607. (18) Vienken, J. Int. J. Artif. Organs 2002, 25, 5, 470-479. (19) Miyata, T.; Oda, O.; Inagi, R.; Iida, Y.; Araki, N.; Yamada, N.; Horiuchi, S.; Taniguchi, N.; Maeda, K.; Kinoshita, T. J. Clin. Invest. 1993, 92, 1243-1252. (20) Anderson, N. L.; Anderson, N. G. Electrophoresis 1998, 19, 18531861.

2674

Journal of Proteome Research • Vol. 5, No. 10, 2006

Bonomini et al. (21) Banks, R. E.; Dunn, M. J.; Hochstrasser, D. F.; Sanchez, J. C.; Blackstock, W.; Pappin, D. J.; Selby, P. J. Lancet 2000, 356, 17491756. (22) Griesser, H. J.; Kingshott, P.; McArthur, S. L.; McLean, K. M.; Kinsel, G. R.; Timmons, R. B. Biomaterials 2004, 25, 4861-4875. (23) Sirolli, V.; Ballone, E.; Di Stante, S.; Amoroso, L.; Bonomini, M. Int. J. Artif. Organs 2002, 25, 6, 529-537. (24) Haycox, C. L.; Ratner, B. D. J. Biomed. Mater. Res. 1993, 11811193. (25) Go¨rg, A.; Postel, W.; Gu ¨ nther, S. Electrophoresis 1988, 9, 531546. (26) Bjellqvist, B.; Pasquali, C.; Ravier, F.; Sanchez, J.-C.; Hochstrasser, D. F. Electrophoresis 1993, 14, 1357-1365. (27) Laemmli, U. K. Nature 1970, 227, 680-685. (28) Urbani, A.; Poland, J.; Bernardini, S.; Bellicampi, L.; Biroccio, A.; Schnolzer, M.; Sinha, P.; Federici, G. Proteomics 2005, 5, 796804. (29) Mortz, E.; Krogh, T. N.; Vorum, H.; Gorg, A. Proteomics 2001, 11, 1359-1363. (30) Meiboom, S.; Gill, D. Rev. Sci. Instrum. 1958, 29, 688-691. (31) Cornelius, R. M.; Brash, J. L. J. Biomater. Sci., Polym. Ed. 1993, 4, 291-304. (32) Mulzer, S. R.; Brash, J. L. J. Biomed. Mater. Res. 1989, 23, 14831504. (33) Oleschuk, R. D.; McComb, M. E.; Chow, A.; Ens, W.; Standing, K. G.; Perreault, H.; Marois, Y.; King, M. Biomaterials 2000, 21, 17011710. (34) Cornelius, R. M.; Archambault, J.; Brash, J. L. Biomaterials 2002, 23, 3583-3587. (35) Magnani, A.; Barbucci, R.; Lamponi, S.; Chiumento, A.; Paffetti, A.; Trabalzini, L.; Martelli, P.; Santucci, A. Electrophoresis 2004, 25, 2413-2424. (36) Walker, A. K.; Wu, Y.; Timmons, R. B.; Kinsel, G. R.; Nelson, K. D. Anal. Chem. 1999, 71, 268-272. (37) Walker, A. K.; Land, C. M.; Kinsel, G. R.; Nelson, K. D. J. Am. Soc. Mass Spectrom. 2000, 11, 62-68. (38) Aoyagi, S.; Hayama, M.; Hasegawa, U.; Sakai, K.; Tozu, M.; Hoshi, T.; Kudo, M. J. Membr. Sci. 2004, 236, 91-99. (39) Levitski, T. V.; Ward, W.; Black, N. A.; Arthur, J. M.; Hamilton, M. B. J. Am. Soc. Nephrol. 2004, 15, 364A. (40) Ward, R. A.; Brinkley, K. A. Contrib. Nephrol. 2004, 141, 280291. (41) Mujais, S. K.; Ivanovich, P.; Bereza, L. A.; Vidovich, M. Contrib. Nephrol. 1995, 113, 101-109. (42) Mondecchini, M.; Teodori, T.; Collodel, L.; Celeghin, A.; Fregonese, A.; Vinello, A. Blood Purif. 1995, 13, 322-326. (43) Ishii, Y.; Yano, S.; Kanai, H.; Maezawa, A.; Tsuchida, A.; Wakamatsu, R.; Naruse, T. Nephron 1996, 73, 407-412. (44) Proteomics in Nephrology. In Contribution to Nephrology; Thonyboonkend, V. , Klein, J. B., Eds.; Karger: Basel, Switzerland, 2004; Vol 141, pp 1-329. (45) Weissinger, E. M.; Kaiser, T.; Meert, N.; De Smet, R.; Walden, M.; Mischak, H.; Vanholder, R. C. Nephrol., Dial., Transplant. 2004, 19, 3068-3077.

PR060150U