Ultrafiltration for proteomic sample preparation

Electrophoresis 2004, 25, 2461–2468 Elena Chernokalskaya Sara Gutierrez Aldo M. Pitt Jack T. Leonard Millipore Corporation, Life Sciences Division, Danvers, MA, USA 2461 Ultrafiltration for proteomic sample preparation Proteome analysis represents significant challenges to the existing sample preparation techniques. Traditional methods, such as two-dimensional electrophoresis, typically separate high-molecular-weight proteins while discarding low-molecular-weight species. This approach is well justified considering the complexity of any proteome. However, it is desirable to extract the maximum amount of information from each sample to investigate the entire range of biomolecules. We have demonstrated that ultrafiltration not only improves two-dimensional electrophoresis (2-DE) resolution of the protein fraction but also yields the low-molecular-weight fraction amenable for further analysis by high-resolution mass spectrometry. This approach was successfully adapted to the variety of biological samples including cell and tissue lysates and serum. Therefore, ultrafiltration offers an alternative sample preparation technique that enables more thorough analysis of a proteome. Keywords: Biomarkers / Mass spectrometry / Proteomics / Two-dimensional electrophoresis / Ultrafiltration DOI 10.1002/elps.200405998 Dramatically improved analytical techniques have been used to advance the understanding of the human proteome. However, the same advances have further highlighted the tremendous complexity of proteomic analysis and the inadequacy of the existing sample preparation techniques. In most proteomic experiments, researchers investigate only a small section of a proteome to accommodate the capabilities of the available analytical techniques. These sub-proteomes are limited to the high- or medium-abundance proteins, protein complexes isolated together or classes of proteins selected by post-translational modification (phosphorylated or glycosylated) or protein property (for example, kinases or proteases or nuclear proteins, etc.). In most cases, only a particular range of protein molecular weights or pI can be investigated, the choice being limited by the sample preparation and protein separation techniques. For example, in traditional two-dimensional electrophoresis (2-DE), the range of separated proteins is restricted by the resolution of the second-dimensional gel. Proteins larger than 200 kDa or smaller than 10 kDa are usually out of range and thus are not available for investigation. On the other hand, in a typical proteomic pattern discovery study, or “peptidomCorrespondence: Dr. Elena Chernokalskaya, Millipore Corp., 17 Cherry Hill Dr., Danvers, MA 01923, USA E-mail: [email protected] Fax: 1978-762-5386 Abbreviations: BCA, bicinchoninic acid; CHCA, a-cyano-4hydroxycinnamic acid; MWCO, molecular weight cutoff; UF, ultrafiltration  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ics”, the larger-molecular-weight proteins are discarded and all the attention is given to smaller polypeptides [1, 2]. This approach is well justified considering the complexity of even the smallest proteome. However, it may be beneficial to combine both methods and explore the whole range of protein sizes in the same proteomic sample. The success of any protein separation and purification is largely determined by the protein solubilization method. The traditional sample lysis technique for 2-DE involves cell or tissue disruption in the presence of high concentrations of urea, reducing agents, and detergents (typical IPG rehydration buffer). The immobilized pH gradient strip (IPG strip) is then rehydrated with sample and proteins are separated by their pI and molecular weight. Unfortunately, the optimal lysis conditions for 2-DE are not compatible with mass spectrometry (MS), which is commonly used for studying smaller proteins and peptides. Isolation and analysis of naturally occurring peptides requires a totally different sample preparation approach, often involving solid-phase extraction (SPE) and chromatography of the native sample followed by thorough desalting necessary for MS analysis. It would be advantageous to develop a method of proteomic sample preparation compatible both with 2-DE and MS. In this paper, we demonstrate that ultrafiltration (UF) can be used to prepare larger proteins for electrophoresis and simultaneously separate smaller polypeptides for MS analysis, thus permitting more comprehensive proteomic investigation of the same sample. UF has been successfully utilized to concentrate and desalt proteins for decades [3]. UF has previously been reported as a sample Proteomics and 2-DE 1 Introduction 2462 E. Chernokalskaya et al. preparation tool to prepare low-molecular-weight fractions for biomarker analysis [4–7]. We also demonstrate that UF can also be used to prepare larger proteins (. , 10 kDa) for electrophoresis and simultaneously separate low-molecular-weight (, , 10 kDa) polypeptides for MS analysis. We show the successful application of this technique to a variety of commonly studied biological specimens such as tissue, cell lysates, and serum. 2 Materials and methods 2.1 Materials Milli-Q water was used for all solutions. General chemicals were from Sigma Aldrich (St. Louis, MO, USA) and Fisher Scientific (Pittsburgh, PA, USA). RPM1 cell culture media and protease inhibitor cocktail were purchased from Sigma Aldrich. Bicinchoninic acid (BCA) protein assay and serum albumin standard were purchased from Pierce (Rockford, IL, USA). Immobiline DryStrip pH 4–7, 11 cm IPG strips, and Pharmalyte carrier ampholytes were from Amersham Biosciences (Uppsala, Sweden). GelChip 6–15% linear gradient 10615 cm Tris/acetate gels, ProteomlQ colloidal Coomassie blue stain, ProteomIQ Universal Sample buffer, ProteomIQ running buffer, and Tris/Tricine/SDS PreoteomIQ equilibration buffer were from Proteome Systems (Woburn, MA, USA). a-Cyano-4-hydroxycinnamic acid (CHCA) was purchased from Applied Biosystems (Framingham, MA, USA). Amicon Ultra-4 10 000 MWCO centrifugal devices, ZipTipmC18 pipette tips, ZipTipSCX pipette tips, and Montage In-Gel DigestZP Kit were from Millipore (Billerica, MA, USA). Fetal bovine serum was from HyClone Labs (Utah, USA). Human serum was acquired from healthy anonymous donors. MCF-7 (HTB 22) human breast cancer cells were from American Type Culture Collection (ATCC). 2.2 Tissue lysate preparation Forty grams of bovine liver were blended in an industrial blender in 50 mL 50 mM Tris-HCl, 150 mM NaCl, containing protease inhibitor cocktail. The lysate was cleared by centrifugation, aliquoted, and stored at 2207C. Protein concentration of 5 mg/mL was determined by BCA protein assay using bovine serum albumin standard. For 2-DE analysis of the neat lysate, 150 mL of the extract was mixed with 50 mL IPG rehydration buffer (see Section 2.5) and used to rehydrate an IPG strip. For 2-DE analysis of lyophilized lysate, the dried sample was resolubilized in 200 mL of IPG rehydration buffer, cleared by centrifugation, and used to rehydrate an IPG strip. For 2-DE analysis of acetone-precipitated proteins, 150 mL liver lysate was mixed with 4 volumes of cold acetone  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Electrophoresis 2004, 25, 2461–2468 and incubated for 1 h at 2207C, followed by centrifugation at 12 000 rpm for 10 min. The supernatant fluid was decanted and the pellet was resuspended in 200 mL of rehydration buffer. For 2-DE analysis of liver lysate prepared by ultrafiltration, 1 mL lysate was dispensed into an Amicon Ultra-4 10 000 MWCO centrifugal device. The device was spun at 30006g for 40 min until approximately 100 mL of concentrated volume was remaining in the device. Nine-hundred microliters of 8 M urea, 2% CHAPS solution was added to the device, and centrifugation was repeated for 40 min. The concentrated proteins were immediately transferred to a microcentrifuge tube and the volume was adjusted to 1 mL with IPG rehydration buffer (8 M urea, 2% CHAPS, 0.002% bromophenol blue, 40 mM DTT, and 0.5% Pharmalyte). One-hundredfifty microliters of reconstituted ultrafiltrate were used to rehydrate an IPG strip in 200 mL total volume. 2.3 Cell lysate preparation MCF-7 (ATCC # HTB 22) human breast cancer cells were grown in RPM1 cell culture media enriched with 10% fetal bovine serum (FBS). Pellets were washed in 16 PBS (0.01 M phosphate-buffered saline, 0.138 M NaCl, 0.0027 M KCl, pH 7.4) and lysed in 10 mM Tris-HCl, 0.5% CHAPS, 0.5% Triton with protease inhibitor cocktail. Cell pellets were subjected to four freeze-thaw cycle with sonication on ice. One milliliter cell lysate was transferred to Amicon Ultra-4 10K MWCO, and centrifuged for 20 min at 30006g. Filtrates were used for peptide analysis by MS and concentrated proteins used for 2-DE analysis. 2.4 Ultrafiltration of human serum One milliliter of human serum was filtered on Amicon Ultra-4 10 000 MWCO centrifugal device. The ultrafiltration devices were centrifuged in a swinging bucket rotor for 15–30 min at 30006g. Ten microliters of the filtrate was acidified with 5 mL 1% TFA, desalted and cleaned with ZipTipmC18 pipette tips. Co-elution was performed directly onto the MALDI target with 2 mL CHCA matrix (5 mg/mL in 50% acetonitrile, 0.1% TFA). 2.5 Two-dimensional electrophoresis First-dimensional electrophoresis was performed in Immobiline DryStrip pH 4–7, 11 cm gels (Amersham Biosciences). The IPG rehydration buffer contained 8 M urea, 2% CHAPS, 0.002% bromophenol blue, 40 mM DTT, and 0.5% Pharmalyte. The strips were focused on an IPGphor system (Amersham Biosciences) at 207C for Electrophoresis 2004, 25, 2461–2468 Ultrafiltration for proteomic sample preparation 50 000 Vh. Second-dimensional SDS PAGE electrophoresis was performed in an ElectrophoretlQ 2000 system (Proteome Systems) using GelChip 6–15% linear gradient 10615 cm61 mm Tris/acetate chemistry gels. Gels were fixed in 25% methanol 2.5% phosphoric acid, then stained overnight with colloidal Coomassie blue (ProteomlQ blue). Gels were scanned on ScanJet 4c/T scanner (Hewlett-Packard, Palo Alto, CA, USA). teins has failed (Fig. 1A), presumably because of the high salt concentration and low concentration of chaotropic agents. In order to achieve a better separation, it was required to concentrate the sample. Usually concentration is achieved by either protein lyophilization or by protein precipitation. The former method allows one to resolubilize highly concentrated proteins in IPG rehydration buffer, but actually increases the salt concentration. As Fig. 1B shows, the resulting 2-D gel is streaky with no well-defined spots and with protein retained at both ends of the IPG strip. 2.6 Protein identification by in-gel digest and MS Gel spots containing proteins of interest were excised and processed with Montage In-Gel DigestZP kit following the manufacturer’s protocol. Digested proteins were concentrated and desalted in a ZipPlateC18 device included in the kit. Samples were immediately spotted on the MALDITOF target and overlaid with 1 mL 5 mg/mL CHCA matrix in 50% acetonitrile, 0.1% TFA. Samples were analyzed using an Autoflex MALDI TOF mass spectrometer (Bruker Daltonic) in reflector mode. Protein database searching was performed with the Mascot program (Matrix Science) with 100 ppm mass tolerance. The 2-D gel proteins shown in this paper were identified with MASCOT score 80 and above. 2.7 Peptide analysis by MS Peptide containing ultrafiltrates from cell lysates or human serum, were acidified with 1% TFA and concentrated on ZipTipC18 or ZipTipSCX following manufacturer’s directions. All samples were overlaid with 1 mL CHCA matrix (5 mg/mL in 50% acetonitrile, 0.1% TFA) and analyzed on Voyager-DE Workstation (Applied Biosystems) in linear mode, or Autoflex MALDI-TOF mass spectrometer (Bruker Daltonic) in reflector mode. PSD analysis of serum peptides was done on AXIMA-CFR Plus MALDI-TOF mass spectrometer (Shimadzu-Kratos). 3 Results and discussion 3.1 Protein concentration for 2-DE using ultrafiltration Bovine liver lysate was prepared by homogenization in isotonic buffer with protease inhibitors. The protein concentration in the extract was determined to be 5 mg/mL. In order to successfully analyze the sample by 2-DE, it was estimated that approximately 0.75 mg of total protein was required, and thus 150 mL of the lysate was prepared in 200 mL volume by addition of 50 mL IPG rehydration buffer (20% of total volume). The attempt to focus the pro-  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 2463 Typically proteins are concentrated by acetone or TCA precipitation [8, 9]. The disadvantage of protein precipitation is that some of the proteins become insoluble, and can not be resolubilized in IPG buffer. Keeping that in mind, acetone precipitation was done carefully to avoid drying the pellet that can lead to formation of mostly insoluble protein. Following acetone precipitation, the proteins were redissolved in IPG running buffer and analyzed by 2-DE. Figure 1C shows a 2-D gel of acetone precipitated liver lysate. Obviously, the salt content was greatly reduced and a satisfactory separation is achieved. While protein recovery may seem to be complete, it requires an alternative method for protein concentration and desalting in order to evaluate this assumption. UF is a well-known method for protein concentration and desalting. UF, rather than dialysis, was successfully used for protein desalting prior to 2-DE of cerebrospinal fluid from multiple sclerosis patients [8]. Specialized regenerated cellulose UF membranes have extremely low protein binding and dependable molecular weight cutoff (MWCO). For 2-D gel analysis, 10 kDa MWCO seems to be the logical choice to retain the proteins of interest because the range of proteins separated by 2-DE gel usually spans 10–150 kDa. To test the suitability of UF for 2-DE sample preparation, bovine liver proteins were concentrated by centrifugal UF using a 10 000 MWCO membrane. After centrifugation, the retained volume above the UF membrane is known as the retentate, and the volume which passes through the UF membrane is known as the ultrafiltrate. In the process of UF, proteins are retained and thereby concentrated, while the salt concentration does not change and is the same in the retentate and the ultrafiltrate. However, because of the reduced volume of the retentate, the saltto-protein ratio in the retentate is lower, and the concentration is easily rendered suitable for 2-DE after dilution with IPG buffer. Using UF we were able to concentrate the protein and reduce the volume of the lysate 10-fold, from 1 mL to 100 mL. One of the advantages of UF is that it is un- 2464 E. Chernokalskaya et al. Electrophoresis 2004, 25, 2461–2468 Figure 1. 2-DE gel of 0.75 mg bovine liver lysate prepared directly (A) by mixing proteins with IPG rehydration buffer (B); by lyophilization and resuspension in IPG rehydration buffer (C); by acetone precipitation (C), and (D) by UF. Differentially displayed spots: 1, serum albumin; 2, 6, protein disulfide isomerase precursor (PDI); 3, ATP synthase b-chain; 4, ER60; 5, GRP78. affected by the solubility of the concentrated proteins, and it is a distinguishing feature of UF that all proteins are concentrated without dehydration from the filtration device. Concentrated proteins are resolubilized directly in the IPG rehydration buffer. Figure 1D shows a 2-D gel of liver proteins prepared by UF. As expected, the separation was successful, resulting in multiple well-defined spots on 2-D gel. We have analyzed multiple 2-D gels of acetone-precipitated and UF-prepared proteins and observed that while both methods permit good resolution, UF seems to  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim recover more high-molecular-weight proteins without loss of smaller proteins (Figs, 1C, D). It was of interest to identify what proteins were lost in the acetone precipitation method. Multiple differentially presented spots were excised and identified by in-gel digest and MALDI-TOF peptide mass fingerprinting (Figs. 1C, D). Many of them turned out to be most common abundant cellular and serum proteins such as albumin and glucose-regulated protein (GRP78). Interestingly, three of the most disparate spots present in the gel of UF-prepared proteins and absent in the gel of acetone-precipitated proteins were PDI (spots 2 and 6, Fig. 1) and ERP60 (spot 4, Fig. 1) Electrophoresis 2004, 25, 2461–2468 Ultrafiltration for proteomic sample preparation 2465 Figure 2. 2-DE gel of 1.5 mg bovine liver lysate prepared by (A) acetone precipitation and (B) UF. The gels are oriented with acidic proteins on the left and high-molecular-weight proteins at the top. proteins. Both proteins belong to the disulfide isomerase family and are associated with endoplasmic reticulum membrane [10–12]. It is likely that the irreversible precipitation of certain proteins by acetone prevents their migration into the IEF gel. Although we do not have sufficient data to demonstrate the mechanism of protein loss during acetone precipitation, our data suggest that acetone precipitation is impaired in its ability to recover large proteins. One implication of this finding is that the apparent difference in the amount of proteins on 2-D gels in acetoneprecipitated samples could simply be an artifact of incomplete resolubilization. The other significant implication is that informative or clinically relevant proteins could be missed entirely if precipitation methods are employed prior to 2-DE. The UF sample preparation method is even more beneficial for 2-D gels with the higher protein loads that may be necessary in order to access lower-abundance proteins, or to obtain sufficient protein for in-gel digest and MS. Figure 2 shows 2-D gels loaded with 1.5 mg of total liver proteins prepared by acetone precipitation and UF. We have found better resolution and higher protein recovery with UF methods without detectable loss of low-molecular-weight species. Our data agrees with published records that UF improves 2-DE separation of water-soluble proteins [9] and recovery of cerebrospinal fluid proteins [8, 13]. When cell-free supernatant fluid was concentrated using UF, rather than acetone precipitation, a higher percentage and a greater range of proteases were recovered from Streptomyces thermovulgaris [14]. Depletion of the larger proteins by acetone precipitation was also observed in proteomic analysis of Fasciola hepatica [15].  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 3.2 2-DE and peptide analysis of MCF7 breast cancer cells While 2-DE is a powerful method for protein separation and biomarker discovery, it does not resolve smaller proteins and peptides. Peptides and other low-molecularweight molecules have been associated with many pathological states such as cancers [5, 16, 17], cardiovascular disease [18–20], atherosclerosis [21], allergy [22], and Alzheimers disease [23]. The potential research and diagnostic value of these biomarkers have not been more fully exploited because of difficulty in their analysis and detection. Many of these significant and more common molecules (ACTH, etc.) are often measured by radioimmuno assay (RIA) or HPLC techniques after elaborate sample preparation schemes. MS of native peptides recently became a method of choice for peptide pattern elucidation [16, 24, 25]. In addition to providing an efficient sample preparation method for 2-DE, UF offers another valuable piece of information that should not be overlooked. In the UF process, proteins above the MWCO are concentrated in the retentate, while low-molecular-weight polypeptides can be found in the ultrafiltrate and are essentially free of larger species. Analysis of both retentate and filtrate presents a unique opportunity to look for biomarkers within the entire range of molecular weights. To prove this concept, we prepared MCF7 breast cancer cells by hypotonic lysis in the presence of detergents and protease inhibitors with sonication. This method extracts soluble cytosolic proteins and peptides. Proteins, thusly prepared, were treated by UF as described previously, recovered in IPG rehydration buffer, and separated by 2466 E. Chernokalskaya et al. Electrophoresis 2004, 25, 2461–2468 Figure 3. Protein and peptide analysis of MCF7 cytosolic extract. The sample was fractionated by UF through 10 000 MWCO membrane. Larger than 10 kDa species were separated by 2-DE (A) while smaller peptides were analyzed by MALDI-TOF-MS (B). Identified human proteins: 1, elongation factor 2, (EF-2); 2, GMP synthase; 3, glucose-6-phosphate 1-dehydrogenase (G6PD); 4, a-enolase; 5, biliverdin reductase A; 6, heat shock protein 27; 7, phosphoglycerate mutase; 8, triosephosphate isomerase; 9, peroxiredoxin; 10, profilin I; 11, ubiquitin. 2-DE. Figure 3A shows a section of the 2-D gel where specific cytosolic proteins were identified by in-gel digest and peptide mass fingerprinting. Next, the peptides in the ultrafiltrate were analyzed. The recovered peptides were virtually free of larger proteins, but were also highly dilute and contained CHAPS and Triton detergents that are not compatible with MS. These peptides were subsequently concentrated and desalted by cation exchange chromatography on ZipTipSCX pipette tips. Figure 3B shows a mass spectrum of the MCF7 cytosolic peptides. Multiple peaks were observed in 1000–5000 Da range, with low background and high signal-to-noise ratio. 3.3 Preparation of serum peptides for MS Most diagnostic tests are based on blood or urine analysis [24]. Serum is a key source of putative protein biomarkers, and, by its nature, can elucidate organ-confined events. Use of mass spectroscopy coupled with bioinformatics has been demonstrated as being capable of distinguishing serum protein pattern signatures of ovarian cancer in patients with early- and late-stage disease [25]. However, serum and plasma present a major challenge for peptide analysis because of high salt, lipid and protein content. When untreated serum is spotted onto a MALDI plate, it does not produce any useable signal in MS  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim (Fig. 4A). Even after reversed-phase concentration and desalting, only a few peptides are detectable in the mass spectrum (Fig. 4B). This can be explained by the high concentration of proteins and lipids competing with the peptides for the resin binding. Multiple protocols have been developed to extract and enrich peptides from tissues and body fluids, such as extraction with 0.1% TFA to selectively precipitate large proteins while enhancing the solubility of smaller proteins and peptides [26] and batch reversed-phase chromatography over C18 resin [20]. UF of serum was reportedly used to remove the large abundant proteins from serum resulting in the enrichment of the low-molecular-weight proteins or peptides. This ,30 kDa fraction contained several classes of physiologically important proteins such as cytokines, chemokines, peptide hormones, as well as proteolytic fragments of larger proteins [7]. Subsequent proteolysis of those proteins by trypsin followed by MS analysis resulted in identification of more than 340 lower-abundance serum proteins (ibid). In this study, we attempted to isolate naturally occurring serum peptides by UF through a 10 000 MWCO membrane. Peptides were analyzed directly out of the ultrafiltrate, or after being concentrated and desalted by solidphase extraction on C18 resin (Figs. 4C, D). We found that Electrophoresis 2004, 25, 2461–2468 Ultrafiltration for proteomic sample preparation 2467 Figure 4. Comparison of a MALDI-TOF spectra of human serum (A) spotted neat on a MALDI target, (B) after desalting with ZipTipmC18 pipette tips, (C) following UF, and (D) finally after UF, desalting, and concentration with ZipTipmC18 pipette tips. Spectra obtained on Bruker AutoFlex MALDI-TOF mass spectrometer in reflector mode. UF alone slightly improves MALDI-TOF-MS, probably by reducing the complexity of the mixture. Nevertheless, the low concentration of peptides and high salt content require another step of sample preparation prior to MS. The addition of an SPE step after UF improved the mass spectrum remarkably. More than 50 peptides were observed in a single MALDITOF spectrum from human serum. The strong signal and low background allowed the easy identification of peptides by MALDI-TOF-MS with PSD ionization (Fig. 5). Identified peptides included a nontryptic fragment of kininogen L and truncated fibrinopeptide A. Both species are normally present in human serum. The fibrinogen a-peptide results from the activation of fibrinogen by thrombin. The appearance of it in serum indicate the presence of functional proteases in blood [20]. Interestingly, progressive loss of N-terminal amino acids from fibrinopeptide A are caused by aminopeptidase, and can potentially indicate myocardial infarction [20]. In our case, the serum was obtained from a healthy volunteer,  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim and the fibrinopeptide A was truncated on both the Nand C-termini. Kininogen L is a light chain of high-molecular-weight kininogen. HMW kininogen is a thiol protease inhibitor and it plays an important role in blood coagulation [27]. 4 Concluding remarks UF is a simple and effective method for sample preparation in proteomics. The primary advantages over alternative methods are speed and protein recovery. UF offers a unique way to obtain the high-molecular-weight fraction for 2-DE while still preserving the integrity and quality of low-molecular-weight materials for high-resolution MS. We anticipate that this method will be useful for analysis of proteins, peptides, low-molecular-weight metabolites, and cellular constituents from a variety of biological sources including cell lysates, tissue homogenates, body fluids, and environmental samples. 2468 E. Chernokalskaya et al. Electrophoresis 2004, 25, 2461–2468 Figure 5. MALDI-TOF spectrum of human serum peptides prepared by UF and SPE extraction. Some of the identified peptides are indicated. Peptides identified by PSD are labeled: m/z 1465, fibrinopeptide A, with first and last amino acids truncated (DSGEGDFLAEGGGVR); m/z 1943, fragment of kininogen L, high MW (amino acids 63–79 NLGHGHKHERDQGHGHQ). We thank Dr. Gary Smejkal and Dr. Alexander Lazarev from Proteome Systems, Inc., for their assistance with 2-D gels of MCF7 proteins, PSD analysis of serum peptides, and constructive comments. Received January 14, 2004 5 References [1] Raida, M., Schulz-Knappe, P., Heine, G., Forssmann, W. G., J. Am. Soc. Mass. Spectrom. 1999, 10, 45–54. [2] Schulz-Knappe, P., Zucht, H. D., Heine, G., Jurgens, M., Hess, R., Schrader, M., Comb. Chem. High Throughput Screen. 2001, 4, 207–217. [3] Ho, W. S., Sirkar, K. K. (Eds.), Membrane Handbook. Part VII: Ultrafiltration, Van Nostrand Reinhold, New York, NY 1992, pp. 391–454. [4] Schulz-Knappe, P., Schrader, M., Standker, L., Richter, R., Hess, R., Jurgens, M., Forssmann, W.-G., J. Chromatogr. A 1997, 776, 125–132. [5] Basso, D., Valerio, A., Seraglia, R., Mazzza, S., Piva, M. G., Greco, E., Fogar, P., Gall, N., Pedrazzoli, S., Tiengo, A., Plebani, M., Pancreas 2002, 24, 8–14. [6] Prazeres, S., Santos, M. A., Ferreira, H. G., Sobrinho, L. G., Clin. Endocrinol. (Oxf). 2003, 58, 686–690. [7] Tirumalai, R. S., Chan, K. C., Prieto, D. A., Issaq, H. J., Conrads, T. P., Veenstra, T. D., Mol. Cell. Proteomics 2003, 10, 1096–1103. [8] Hammack, B. N., Owens, G. P., Burgoon, M. P., Gilden, D. H., Mult. Scler. 2003, 9, 472–475. [9] Chan, L. L., Lo, S. C., Hodgkiss, I. J., Proteomics 2002, 2, 1169–1186. [10] Yamauchi, K., Yamamoto, T., Hayashi, H., Koya, S., Takikawa, H., Toyoshima, K., Horiuchi, R., Biochem. Biophys. Res. Commun. 1987, 146, 1485–1492.  2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim [11] Hirano, N., Shibasaki, F., Sakai, R., Tanaka, T., Nishida, J., Yazaki, Y., Takenawa, T., Hirai, H., Eur. J. Biochem. 1995, 234, 336–342. [12] Guo, G. G., Patel, K., Kumar, V., Shah, M., Fried, V. A., Etlinger, J. D., Sehgal, P. B., J. Interferon Cytokine Res. 2002, 22, 555–563. [13] Yuan, X., Russell, T., Wood, G., Desiderio, D. M., Electrophoresis 2002, 23, 1185–1196. [14] Yeoman, K. H., Edwards, C., J. Appl. Microbiol. 1997, 82, 149–156. [15] Jefferies, J. R., Brophy, P. M., Barrett, J., Electrophoresis 2000, 21, 3724–3729. [16] Petricoin III, E. F., Ardekani, A. M., Hitt, B. A., Levine, P. J., Fusaro, V. A., Steinberg, S. M., Mills, G. B., Simone, C., Fishman, D. A., Kohn, E. C., Liotta, L. A., The Lancet 2002, 359, 572–577. [17] Wulfkuhle, J. D., Liotta, L. A., Petricoin III, E. F., Nat. Rev. Cancer 2003, 3, 267–275. [18] Jungblut, P. R., Zimny-Arndt, U., Zeindl-Eberhart, E., Stulik, J., Koupilova, K., Pleißner, K.-P., Otto, A., Müller, E.-C., Sokolowska-Köhler, W., Grabher, G., Stöffler, G., Electrophoresis 1999, 20, 2100–2110. [19] Van Eyk, J. E., Curr. Opin. Mol. Ther. 2001, 3, 546–553. [20] Marshall, J., Kupchak, P., Zhu, W., Yantha, J., Vrees, T., Furesz, S., Jacks, K., Smith, C., Kireeva, I., Zhang, R., Takahashi, M., Stanton, E., Jackowski, G., J. Proteome Res. 2003, 2, 361–372. [21] Ballantyne, C. M., Clin. Cardiol. 2001, 24, III13–17. [22] Toda, M., Ono, S. J., Immunology 2002, 106, 1–10. [23] DeMattos, R. B., Bales, K. R., Cummins, D. J., Paul, S. M., Holtzman, D. M., Science 2002, 295, 2264–2267. [24] Bischoff, R., Luider, T. M., J. Chromatogr. B 2004, 803, 27–40. [25] Stevens, E. V., Liotta, L. A., Kohn, E. C., J. Gynecol. Cancer 2003, 13, 133–139. [26] Kline, T. R., Pang, J., Hefta, S. A., Opiteck, G. J., Kiefer, S. E., Scheffler, J. E., Anal. Biochem. 2003, 315, 183–188. [27] Kitamura, N., Ohkubo, H., Nakanishi, S., Cardiovasc. Pharmacol. 1987, 10, S49–S53.