Quantitative Multiplexed C-Reactive Protein Mass Spectrometric Immunoassay Urban A. Kiernan,* Riccardo Addobbati,† Dobrin Nedelkov, and Randall W. Nelson Intrinsic Bioprobes, Inc., 625 S. Smith Rd. Ste. 22, Tempe, Arizona 85281 Received March 23, 2006
Reported in this work is the development and application of a high sensitivity mass spectrometric immunoassay for the quantitative analysis of C-reactive protein from human plasma. Multiplexed affinity retrieval devices and methodology were developed to simultaneously target retinol binding protein, C-reactive protein, serum amyloid P component, as well as an added exogenous internal reference standard (staphylococcal enterotoxin B) for subsequent MALDI-TOF MS analysis. This approach allows for semiquantitative analysis of both retinol binding protein and serum amyloid P component while performing absolute quantitative measurements of C-reactive protein. The ability to qualitatively differentiate between all three human proteins and their associated variants is also maintained. Standard curve, QC, and human plasma samples were analyzed in a high throughput manner, which performed with a CV < 15%. The resultant human plasma sample C-reactive protein quantitative measurements were then compared to those achieved with a high sensitivity latex immunoturbidimetric assay. Keywords: mass spectrometry • Immunoassay • C-reactive protein • quantitation • multiplexing • human plasma
Introduction The field of proteomics has extended its scope beyond basic research and is now attempting to enter clinical application and diagnostics. Historically, clinical biomarker screening has been reserved for classical immunoassay methodologies, but recent findings have shown that pertinent clinical data may lie beyond the fidelity of such approaches 1. The use of protein mass spectrometry in proteomics holds the key to this issue, with the intrinsic ability to qualitatively discriminate between multiple forms of the same target protein. However, a major hurdle for proteomics, in entering the clinical and diagnostic arena, is the inability to readily perform absolute protein quantification. There are many quantitative proteomic approaches currently available, however, methodologies such as Isotope-Coded Affinity Tags (ICAT)2 are only semiquantitative in nature and have restricted data content due to being peptide based. Such peptide based approaches may not allow for accurate qualitative differentiation between the clinically relevant forms of the same protein.3-5 A potential solution to this problem is the mass spectrometric immunoassay (MSIA), a high content proteomics methodology based on immuno-affinity protein isolation combined with mass spectrometric detection. This approach has a longstanding track record of performing both relative6-8 and absolute quantitative analyses,9-12 which is augmented by its unprecedented ability to qualitatively differentiate between multiple forms of the same protein target1,13-19 in a single * To whom correspondence should be addressed. Tel: (480) 804-1778. Fax: (480) 804-0778. Email:
[email protected]. † Currently at IRCCS Burlo Garofolo, Via dell′Istria 65/1, 34127 Trieste, Italy.
1682
Journal of Proteome Research 2006, 5, 1682-1687
Published on Web 05/27/2006
analysis. This ability to discriminate between multiple proteins makes MSIA ideal for the development of multiplexed assays, for combined qualitative and quantitative assessment. Presented in this work is the development and application of a novel quantitative multiplexed C-reactive protein (CRP) MSIA. CRP, a renowned clinical biomarker of inflammation,20 was selected as an ideal protein target because it exhibits a large dynamic concentration range in human plasma,21 has little known phenotypic variation, and possesses strong clinical value.22 This is the first demonstration of a quadraplexed MSIA approach, which is based upon a previously developed semiquantitative methodology.6 However, this approach has been enhanced for the absolute quantitative measurement of CRP through the selective affinity retrieval of an exogenous internal reference standard, while maintaining its semiquantitative capacity for two other targeted human plasma proteins (retinol binding protein and serum amyloid P component) and its qualitative differential capacity for all protein targets. CRP measurements made using this assay were then compared to those generated via a classical immuno-metric method.
Experimental Procedure Affinity Pipets. MSIA protein purification was achieved using antibody derivatized affinity pipets (Intrinsic Bioprobes, Inc., Tempe, AZ). These pipets (commonly referred to as MSIA-Tips) were produced using the same protocol as described previously.23 The affinity ligand used in this study was a mixture of polyclonal antibodies targeting retinol binding protein (RBP: 8.0 mg/mL, Cat No. A0040, DakoCytomation, Carpinteria, CA), C-reactive protein (CRP: 8.3 mg/mL, Cat No. A0073, DakoCytomation, Carpinteria, CA), serum amyloid P component 10.1021/pr0601133 CCC: $33.50
2006 American Chemical Society
research articles
Quantitative C-Reactive Protein Mass Spectrometric Assay
(SAP: 8.1 mg/mL, Cat No. A0302, DakoCytomation, Carpinteria, CA) and staphylococcal enterotoxin B (SEB: 1 mg/mL, Cat No. LBI 202, Toxin Technology, Sarasota, FL). The optimum antibody ratio was empirically determined to be 3:3:1:5 (v/v), respectively. Patient Sample Collection. The samples used in this study were provided by the National Institute of Diabetes & Digestive & Kidney Diseases (NIDDK) repository as a part of their preand type 2 diabetes biomarker discovery program. Citrate plasmas from 35 individuals, which included healthy, prediabetic, and untreated type 2 diabetic patients, were obtained. These classifications were retrospectively determined after the administration of an oral glucose tolerance test. Moreover, samples were provided with pre-screen C-reactive protein (CRP) values that were determined using the Roche Tina-quant high sensitivity latex immunoturbidimetric CRP assay (measurements ranging from 0.1 to 20.0 mg/L) performed on a Hitachi spectrophotometer. Standard Curve and QC Sample Preparation. This study utilized an eight point standard curve (run in duplicate) and low, med and high QC samples (each run in quadruplicate). The standard curves and QC samples were prepared by doping highly purified protein targets into horse serum (Cat No. P5552, Sigma Chemical, St. Louis, MO). Horse serum was selected as the sample matrix of choice due to a lack of cross reactivity between the human protein antibodies used and their equine antigen counterparts, while still being a highly complex biological fluid that is similar in many respects to human. These samples were initially produced in bulk, by spiking neat horse serum (1.38 mL) with purified protein antigens. Spiking included 22.2 µL of 2.9 mg/mL human RBP (Cat No. 30-AR20, Fitzgerald, Concord, MA), 75 µL of 1.0 mg/mL human SAP (Cat No. S-5269, Sigma-Aldrich Co., St. Louis, MO) and finally 22.5 µL of 1.0 mg/mL solution of SEB (Cat No. BT202, Toxin Technology, Sarasota, FL) used as the internal reference standard (IRS). (SEB is a category B toxin, as defined by the U.S. Department of Defense, which requires specific precautions in handling and storage due to its known toxicity to humans and its potential as a biological warfare agent.) The human RBP antigen used in this study was affinity purified from human urine and was truncated (loss of 4 C-terminal amino acids) as compared to the endogenous wild-type plasma protein. Even though the RBP was modified, it was determined to be suitable as a QC standard since its affinity retrieval and MS characteristics did not significantly differ from the endogenous plasma forms. The concentrations of human protein antigens used were selected in order to mimic the physiological concentrations of these targets within human plasma. At this point, the prepared standard samples were split and subsequently spiked with their corresponding amount of 1.0 mg/ mL human CRP antigen (Cat No. CP1000U, Cortex Biochem, San Leandro, CA). The final CRP concentrations of the standard curve samples ranged from 0.025 to 3.0 µg/mL and the QC concentrations used were 0.075, 0.75 and 2.5 µg/mL, corresponding to low, med and high QC, respectively. Fifty microliter aliquots of each were transferred into individual wells of a 96deep well micro-titer plate (Greiner Bio-One, Longwood, FL) and diluted to 1 mL with 0.1 M HEPES buffered saline (pH 7.4) with 0.15 M NaCl, 3 mM EDTA and 0.005% polyoxyethylenesorbitan monolaurate (Tween 20). Patient Sample Preparation. Patient samples were thawed at room temperature and 50 µL aliquots were pipetted into individual wells of a 96-deep well micro-titer plate. Each aliquot
was spiked with 7.5 µL of 0.1 mg/mL SEB internal reference standard and was then diluted to 1 mL with HBS-EP. Sample Analysis. Samples were screened in parallel using a multichannel robotic pipetting workstation (Beckman Coulter, Fullerton, CA) outfitted with the prepared multiplexed affinity pipets. In this study, a total of 63 parallel analyses were performed in a high-throughput manner. The analytical procedure used involved the repetitive flowing of target samples through the affinity pipet tips to simultaneously retrieve and enrich target proteins from the biological fluid. Samples (100 µL) were flown 150 times (repetitive aspirations and dispenses) through the tips, followed by a HBS-EP rinse (100 µL, 10 times) and four serial double distilled water rinses (100 µL, 10 times). At this point, the enriched and purified proteins retained within the affinity pipet tips were eluted directly onto a contrasted 96-spot MALDI-target.11 The elution process involved aspirating 6.0 µL of MALDI matrix solution (saturated sinapinic acid in acetonitrile/water (1:2 v/v) with 0.4% trifluoroacetic acid) into the tips followed by eluant deposition. The samples were allowed to air-dry prior to mass spectrometric analysis. Mass Spectrometric Analyses. Parent protein mass spectrometry was performed on a linear Autoflex MALDI-TOF mass spectrometer (Bruker, Billerica, MA). A linear delayed extraction mode was employed using a 1.45 kV draw out pulse, 670 ns delay and a full accelerating potential of 20.00 kV. Human CRP was used as a calibration standard. Each mass spectrum was of the sum of five 100-laser shot acquisitions. All spectra were then viewed using Proteome Analyzer software (Intrinsic Bioprobes Inc., Tempe, AZ) in which all spectra were aligned and normalized to the integral of the SEB parent signal for intersample comparison.
Results and Discussion Assay Development. This quantitative multiplexed mass spectrometric immunoassay is based upon a previously developed semiquantitative approach that targets the same three human proteins.6 These three protein targets were previously selected for multiplexing based on their size (similar in molecular mass yet different enough to be sufficiently resolved), amicability to MALDI-TOF MS analysis, tolerance to the same MALDI-matrix, and the quality of commercially available antibodies. However, this work is an improvement of the previous approach because of its new assay design, which has the additional ability to target for an exogenous protein for use as an internal reference standard, thus allowing for absolute protein quantification through the generation of a standard curve. SEB was selected as a suitable IRS based on the same criteria listed above as well as it not being an endogenous human blood protein. Other exogenous proteins were tested during the assay development phase (data not shown), however, SEB was found to be the best choice for use with these human protein targets. The appropriate quantity of antigen used as the IRS and the ratio of SEB antibody were determined empirically, with a final working amount of 7.5 µL of 0.1 mg/ mL SEB per 50 µL of human plasma and an antibody ratio of 3 RBP:3 CRP:1 SAP:5 SEB (v/v). Application. The resultant assays were then applied to the standard and human samples. Figure 1A is a representative mass spectrum of the application of the multiplexed MSIA devices in the analysis of human plasma (sample no. 35). MS signals from all four-protein targets, and their associated variants, are clearly observed. A complete list of all identified signals, along with their theoretical and observed m/z values, Journal of Proteome Research • Vol. 5, No. 7, 2006 1683
research articles
Kiernan et al.
Figure 1. Application of the multiplexed MSIA devices and methodology. (A) Representative mass spectrum of the MSIA analysis of human plasma (sample no. 35). M/z signals from RBP, CRP, SAP, and SEB are present. (B) Mass spectra from the MSIA analysis of some standard samples used to generate standard curve 1. Samples displayed were with CRP concentrations of 0.05, 0.25, 1.00, and 3.00 µg/mL. Table 1. Identified Parent Protein Signals from Human Plasma protein identified
m/z theoretical (MH+)
m/z observed (MH+)
RBP RBP -Leu CRP SAP SAP-Val SAP-Sial SEB
21 066.5 20 953.4 23 029.1 25 463.5 25 364.4 25 172.2 28 367.0
21 065.8 ( 3.7 20 952.9 ( 4.3 23 029.2 ( 3.5 25 466.0 ( 6.6 25 360.9 ( 3.0 25 174.5 ( 3.9 28 368.7 ( 4.4
is presented in Table 1. The analysis of the standard curve and QC samples result in very similar protein profiles as those generated from human plasma. The only difference observed was because the purified RBP antigen used in the standard samples was a truncated form compared to the human wildtype found in plasma, with a molecular weight of 20 568.9 Da. Examples of such profiles, normalized to the integral of the internal reference standard, are shown in Figure 1B, which are a portion of the data set used to generate one of the two standard curves. These data clearly illustrate the consistency of the analysis while demonstrating the ability of this MS-based immunoassay to monitor change as the concentration of the CRP antigen is increased. The normalized CRP peak integrals from both the standard curve and QC samples were plotted, shown in Figure 2. The plot in Figure 2A shows how the two standard curves are identical and correlate well with a power series, with R2 values of 0.9893 and 0.9935, respectively. Also included in this plot are the normalized results for the QC samples. For improved clarity at the lower concentration range of the standard curve and QC values, the plots were converted to a Log/Log scale, shown in Figure 2B. The data generated from the two standard curves and the QC samples were also used to determine 1684
Journal of Proteome Research • Vol. 5, No. 7, 2006
Figure 2. Plots of the CRP standard curves and QC samples. (A) Overlay plot of curves 1 and 2, with points that range from 0.025 to 3.000 µg/mL, along with the low (0.075 µg/mL), med (0.750 µg/mL), and high (2.500 µg/mL) QC samples. Plots correlate with a power series, with R2 values of 0.9893 and 0.9935, respectively. (B) Overlay log/log plot of curves 1 and 2 along with the QC samples. Table 2. Coefficients of Variation Determined from the Standard Curve and QC Samples source data
RBP
curve 1 curve 2 low QC (0.075 µg/mL) med QC (0.750 µg/mL) high QC (2.500 µg/mL)
13.634 6.242 4.192 13.752 9.676
CRP
SAP
6.373 7.115 10.134
6.252 11.640 2.354 11.590 11.932
reproducibility, with the coefficient of variation (CV) of the assay. These values are listed in Table 2, demonstrated a mean CV and standard error of the estimate (SEE) of 8.837 and 13.4, respectively. Semiquantitative Determination. The semiquantitative measurements of both RBP and SAP were extracted from the normalized data sets. Plots showing the normalized integral of each protein and their associated variants are presented in Figure 3A and B, respectively. Even though the measurements are only semiquantitative in nature, the intact protein MS analysis allows for differentiation between the multiple variants of each protein that were affinity retrieved. This is a novel feature of this approach since conventional immunoassays are normally blind to such subtle variation. These plots clearly show the varying normalized abundances of three forms of RBP and three forms of SAP consistently detected in each patient
Quantitative C-Reactive Protein Mass Spectrometric Assay
research articles
Figure 3. Semiquantitative protein determination. (A) Normalized abundances of retinol binding protein and detected variants in each of the 35 patient plasma samples. (B) Normalized abundances of serum amyloid P component and detected variants in each of the same patient plasma samples.
sample. Such protein variation may hold significant information regarding the presence of disease. Absolute CRP Quantitation. Regarding the quantitative CRP analysis, the two generated standard curves of the normalized CRP peak integrals verses CRP antigen concentration were averaged, which resulted in the following equation (CRP/SEB ) 1.3092[CRP mg/mL]0.6742). This average equation was then used to translate the observed normalized CRP/SEB peak integral ratios determined from each of the patient samples into CRP concentrations (mg/mL). Only two plasma samples (nos. 17 and 26) of the 35 analyzed using this approach resulted in concen-
tration values outside the range of the standard curve. During the development phase of this assay, we identified that points outside of the established curve range no longer displayed characteristics consistent with the power series (data not shown). Therefore, repeat analyses using sample dilutions would be necessary for accurate measurements of samples with high concentration of CRP. These samples were reanalyzed using a decreased amount of sample plasma (25 µL and 16.5 µL, respectively), so that the results fell within the range of the standard curve, but was then factored for when the final concentration was calculated. Journal of Proteome Research • Vol. 5, No. 7, 2006 1685
research articles
Kiernan et al.
multiplexed proteomic analysis. Moreover, the QC and standard curve analyses demonstrated coefficients of variation that were significantly better than other potential clinical proteomic approaches28 again adding to the potential of the mass spectrometric immunoassay for routine and repetitive target protein(s) analysis, as a needed in a clinical and diagnostic application. Finally, this demonstration of an improved multiplexed MSIA approach sets the foundation for future developmental work, which would include increased multiplexing for more protein targets, as well as the simultaneous generation of multiple standard curves for absolute multi-protein quantitative measurements in a single analysis. Figure 4. Human plasma CRP concentrations. Comparison of the CRP concentrations determined in each of the patient samples (µg/mL) via mass spectrometric immunoassay and immunoturbidimetric assay.
The results of the MSIA CRP-quantitative measurements performed on the patient samples were then plotted with the CRP concentrations determined in the same samples via an immunoturbidimetric assay. This plot is shown in Figure 4, which clearly illustrates that the same trend in CRP concentrations is observed regardless of the immunoassay methodology applied. Discrepancies between the two assay methods used are observed in the two data sets, mostly with the MSIA values being lower than those determined with the immunoturbidimetric approach. The differences observed range from 88 to 627% lower in calculated concentration. However, such differences may be explained by a number of factors, which include the following: (1) the MSIA hs-CRP assay and standard curve were engineered to detect and quantify CRP levels in a concentration range that differed from the turbidity assay (MSIA ranged from 0.025 to 3.000 µg/mL and the turbidity assay ranged from 0.1 to 300.0 µg/mL, (2) the samples were older when the MSIA analysis was performed than when the turbidity assay was performed, and finally (3) the antibodies and standards used to perform the two immunoassays and generate the standard curve samples were different. Such variations in assay development and methodology have shown to differ results between other assays of the same protein target by as much as 10 000%.24-27
Conclusion It has become readily apparent that proteomics has the potential to become integral in clinical application. However, the question still remains as to which form of the multiple proteomics approaches available will translate into the mainstream of clinical and diagnostic use. Presented here was the development and application of a novel multiplexed high sensitivity-CRP mass spectrometric immunoassay. This approach allowed for simultaneous semiquantitative analysis of human retinol binding protein and serum amyloid P component while performing rigorous quantitative measurements of C-reactive protein. Qualitative protein characterization, of all targets, is maintained in this high throughput approach. This methodology described is a novel multiplexed approach that is an improvement of a previously existing assay. The enhancement was achieved through the specific affinity targeting and analysis of an exogenous protein (staphylococcal enterotoxin B) that was added to all samples and co-analyzed for use as an internal reference standard in peak integral normalization, thus making the semi- and absolute quantitative measurements possible. This is the first ever demonstration of such a 1686
Journal of Proteome Research • Vol. 5, No. 7, 2006
Acknowledgment. We would like to thank Dr. Allan L. Bieber for his assistance in preparing and the critical reading of this manuscript. We would also like to thank Dr. Lawrence S. Philips from Emory University for kindly providing the samples and the CRP immunoturbidometric assay data used in this study through the NIDDK repository as a part of Grant No. R18-DK066204 and M01-RR000039. This work was funded by the National Institute of Diabetes & Digestive & Kidney Diseases under Grant No. 1 R42DK071290-01. References (1) Nedelkov, D.; Kiernan, U. A.; Niederkofler, E. E.; Tubbs, K. A.; Nelson, R. W. PNAS 2005, 102, 10852-10857. (2) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R. Nat. Biotechnol. 1999, 17, 994-999. (3) Mahley, R. W.; Huang, Y. Curr. Opin. Lipidol. 1999, 10, 207-217. (4) Yamauchi, K.; Tozuka, M.; Nakabayashi, T.; Sugano, M.; Hidaka, H.; Kondo, Y.; Katsuyama, T. J. Neurosci. Res. 1999, 58, 301-307. (5) Schreiber, G.; Richardson, S. J. Comp. Biochem. Physiol. B Biochem. Mol. Biol. 1997, 116, 137-160. (6) Kiernan, U. A.; Nedelkov, D.; Tubbs, K. A.; Niederkofler, E. E.; Nelson, R. W. Clin. Proteomics J. 2004, 1, 7-16. (7) Niederkofler, E. E.; Nedelkov, D.; Tubbs, K. A.; Kiernan, U. A.; Nelson, R. W. Proceedings of the 51st ASMS Conference on Mass Spectrometry and Allied Topics, 2003. (8) Kiernan, U. A.; Nedelkov, D.; Niederkofler, E. E.; Tubbs, K. A.; Bieber, A. L.; Nelson, R. W. Proceedings of the 53st ASMS Conference on Mass Spectrometry and Allied Topics, 2005. (9) Nelson, R. W.; Krone, J. R.; Bieber, A. L.; Williams, P. Anal. Chem. 1995, 67, 1153-1158. (10) Tubbs, K. A.; Nedelkov, D.; Nelson, R. W. Anal. Biochem. 2001, 289, 26-35. (11) Niederkofler, E. E.; Tubbs, K. A.; Gruber, K.; Nedelkov, D.; Kiernan, U. A.; Williams, P.; Nelson, R. W. Anal. Chem. 2001, 73, 32943299. (12) Nelson, R. W.; Nedelkov, D.; Tubbs, K. A.; Kiernan, U. A. J. Proteome Res. 2004, 3, 851-855. (13) Kiernan, U. A.; Nedelkov, D.; Tubbs, K. A.; Niederkofler, E. E.; Nelson, R. W. Am. Biotech. Lab. 2002, 20, 26-28. (14) Kiernan, U. A.; Nedelkov, D.; Tubbs, K. A.; Niederkofler, E. E.; Nelson, R. W. Proteomics 2004, 4, 1825-1829. (15) Kiernan, U. A.; Tubbs, K. A.; Gruber, K.; Nedelkov, D.; Niederkofler, E. E.; Williams, P.; Nelson, R. W. Anal. Biochem. 2002, 301, 49-56. (16) Kiernan, U. A.; Tubbs, K. A.; Nedelkov, D.; Niederkofler, E. E.; McConnell, E.; Nelson, R. W. J. Proteome Res. 2003, 2, 191-197. (17) Kiernan, U. A.; Tubbs, K. A.; Nedelkov, D.; Niederkofler, E. E.; Nelson, R. W. Biochem. Biophys. Res. Commun. 2002, 297, 401405. (18) Kiernan, U. A.; Tubbs, K. A.; Nedelkov, D.; Niederkofler, E. E.; Nelson, R. W. FEBS Lett. 2003, 537, 166-170. (19) Tubbs, K. A.; Kiernan, U. A.; Niederkofler, E. E.; Nedelkov, D.; Bieber, A. L.; Nelson, R. W. Proteomics 2005, 5, 5002-5007. (20) Zimmerman, M. A.; Selzman, C. H.; Cothren, C.; Sorensen, A. C.; Raeburn, C. D.; Harken, A. H. Arch. Surg. 2003, 138, 220-224. (21) Kushner, I. Ann. N.Y. Acad. Sci. 1982, 389, 39-48. (22) Kushner, I. Science 2002, 297, 520-521. (23) Niederkofler, E. E.; Tubbs, K. A.; Kiernan, U. A.; Nedelkov, D.; Nelson, R. W. J. Lipid Res. 2003, 44, 630-639.
research articles
Quantitative C-Reactive Protein Mass Spectrometric Assay (24) Hafner, G.; Peetz, D.; Erbes, H.; Post, F.; Dahm, M.; Peivandi, A. A.; Sucke, B.; Blankenberg, S.; Rupprecht, H. J.; Prellwitz, W. Scand. J. Clin. Lab. Invest. 2001, 61, 227-235. (25) Maggiore, U.; Cristol, J. P.; Canaud, B.; Dupuy, A. M.; Formica, M.; Pozzato, M.; Panichi, V.; Consani, C.; Metelli, M. R.; Sereni, L.; De Nitti, C.; David, S.; Tetta, C. J. Lab. Clin. Med. 2005, 145, 305-308. (26) Christenson, R. H.; Apple, F. S.; Morgan, D. L.; Alonsozana, G. L.; Mascotti, K.; Olson, M.; McCormack, R. T.; Wians, F. H., Jr.; Keffer, J. H.; Duh, S. H. Clin. Chem. 1998, 44, 52-60.
(27) Apple, F. S.; Maturen, A. J.; Mullins, R. E.; Painter, P. C.; PessinMinsley, M. S.; Webster, R. A.; Spray Flores, J.; DeCresce, R.; Fink, D. J.; Buckley, P. M.; Marsh, J.; Ricchiuti, V.; Christenson, R. H. Clin. Chem. 1999, 45, 206-212. (28) Aivado, M.; Spentzos, D.; Alterovitz, G.; Otu, H. H.; Grall, F.; Giagounidis, A. A.; Wells, M.; Cho, J. Y.; Germing, U.; Czibere, A.; Prall, W. C.; Porter, C.; Ramoni, M. F.; Libermann, T. A. Clin. Chem. Lab. Med. 2005, 43, 133-140.
PR0601133
Journal of Proteome Research • Vol. 5, No. 7, 2006 1687
