Anal. Chem. 2003, 75, 1196-1202
On-Line Immunoaffinity-Liquid Chromatography-Mass Spectrometry for Identification of Amyloid Disease Markers in Biological Fluids Jette W. Sen,† H. Robert Bergen, 3rd,‡ and Niels H. H. Heegaard*,†
Department of Autoimmunology, Statens Serum Institut, Artillerivej 5, DK-2300 Copenhagen S., Denmark, and Mayo Proteomics Research Center, Mayo Foundation, 200 1st Street SW, Rochester, Minnesota 55905
Disease-specific alterations in proteins and peptides such as the appearance of new isoforms, changed relative concentrations of known isoforms, or changed catabolism characterize the group of protein precipitation disorders collectively known as amyloidoses. The goal of this study was to develop an approach for isolating and characterizing the pool of isoforms of a polypeptide of interest from biological fluids for use in development of diagnostic markers and elucidation of pathogenesis. For this purpose, we employed an on-line immunoaffinity-liquid chromatography-mass spectrometry (IA-LC-MS) modular approach using antibodies binding populations of protein isoforms. In this system, crude biological samples, e.g., serum, may be injected and subjected to fast hands-off analysis. The setup consists of an optional preclear column for removal of unspecific binding components, an immunoaffinity column, a short cartridgelike reversedphase column, and an electrospray time-of-flight mass spectrometer. We have tested the system for the automated analysis of three amyloid-related polypeptides, serum amyloid P component, amyloid β-peptide, and β2microglobulin, and we show the feasibility of detection of altered isoforms or determination of relative abundance of isoforms of the proteins from serum or cerebrospinal fluid samples. For each new protein investigated, the only change needed in the system is a new antibody or antibody mixture and the selection of a reversed-phase cartridge of appropriate hydrophobicity. Alterations of polypeptides such as mutations, posttranslational modifications, or changed catabolism are often associated with disease. In the process of discovering and developing new diagnostic tools, the identification of such altered polypeptides is crucial and provides not only diagnostic leads but also clues to understanding the pathogenesis of a given disease. Amyloidosis is a group of chronic diseases associated with a polypeptide aberrantly folding into an insoluble β-sheet structure that forms precipitating fibrils in amyloid deposits surrounded by glycosami* Corresponding author. Phone: +45 32683378. Fax: +45 32683876. E-mail:
[email protected]. † Statens Serum Institut. ‡ Mayo Foundation.
1196 Analytical Chemistry, Vol. 75, No. 5, March 1, 2003
noglycans, serum amyloid P component (SAP), complement factors, and extracellular matrix proteins.1,2 The amyloidogenic core protein is of varying origin with one specific protein or peptide being associated with each type of amyloidosis.3 Mutated and posttranslationally modified proteins as well as proteins with changes in relative abundances of isoforms in, for example, serum or cerebrospinal fluid have been shown to be implicated in several amyloidogenic disorders.4-8 Mass spectrometry is currently the only technique that immediately discriminates between such species. SAP is a serum protein with scavenging and chaperone functions.9;10 For largely unknown reasons, substantial amounts of SAP are bound to all types of amyloid. β2-microglobulin (β2M) is a 99-amino acid-long protein that constitutes the light chain of the major histocompatibility class 1 complex11 but is also found as a free serum protein.12 β2M amyloid causes, for example, carpal tunnel syndrome in renal patients.13 In several chronic diseases such as small cell lung cancer,14,15 rheumatoid arthritis, and systemic lupus erythematosus (SLE)16 and in patients in chronic hemodialysis, an alternatively cleaved and truncated isoform of β2M, des-Lys β2M, may also be present in the blood.16-19 This alternative isoform is generated when the β2M chain is cleaved (1) Cohen, A. S. Curr. Opin. Reumatol. 1994, 6, 68-77. (2) Kisilevsky, R.; Snow, A. Med. Hypotheses 1988, 26, 231-6. (3) Cohen, A. S. Curr. Sci. 1994, 6, 55-68. (4) Date, Y.; Nakazato, M.; Kangawa, K.; Shirieda, K.; Fujimoto, T.; Matsukura, S. J. Neurol. Sci. 1997, 150, 143-8. (5) Golde, T. E.; Eckman, C. B.; Younkin, S. G. Biochim. Biophys. Acta 2000, 1502, 172-87. (6) Moir, R. D.; Lynch, T.; Bush, A. I.; Whyte, S.; Henry, A.; Portbury, S.; Multharp, G.; Small, D. H.; Tanzi, R. E.; Beyreuther, K.; Masters, C. L. J. Biol. Chem. 1998, 273, 5013-9. (7) Motter, R.; Vigo-Pelfrey, C.; Kholodenko, D.; Barbour, R.; Johnson-Wood, K.; Galasko, D.; Chang, L.; Miller, B.; Clark, C.; Green, R. Ann. Neurol. 1995, 38, 643-8. (8) Argiles, A.; Garcia-Garcia, M.; Derancourt, J.; Mourad, G.; Demaille, J. G. Kidney Int. 1995, 48, 1397-405. (9) Coker, A. R.; Purvis, A.; Baker, D.; Pepys, M. B.; Wood, S. P. FEBS Lett. 2000, 473, 199-202. (10) Bharadwaj, D.; Mold, C.; Markham, E.; Du Clos, T. W. J. Immunol. 2001, 166, 6735-41. (11) Solheim, B. G.; Thorsby, E. Nature 1974, 249, 36-8. (12) Plesner, T.; Bjerrum, O. J. Scand. J. Immunol. 1980, 11, 341-51. (13) Drueke, T. B. Nephrol. Dial. Transplant. 2000, 15 Suppl 1, 17-24. (14) Nissen, M. H.; Plesner, T.; Rorth, M. Clin. Chim. Acta 1984, 141, 41-50. (15) Plesner, T. Allergy 1980, 35, 627-37. (16) Plesner, T.; Wiik, A. Scand. J. Immunol. 1979, 9, 247-54. 10.1021/ac026174b CCC: $25.00
© 2003 American Chemical Society Published on Web 01/31/2003
by complement protein C1 at lysine 58, which is then cleaved off by a carboxypeptidase B-like activity in serum. A disulfide bond holds the resulting two chains together into a heterodimer.20 Despite recent evidence for a substantial amount of alternatively folded protein in des-Lys β2M preparations,21 it is not yet clear whether the modified β2M isoforms plays a role in the amyloidosis occurring in renal patients.8,13,22 Another polypeptide prone to forming amyloid is amyloid β-peptide (AβP), which constitutes the core peptide of amyloid plaques in the brain of Alzheimer patients. The peptide is generated from the membrane protein amyloid precursor protein.23 Isoforms of AβP differing in length and side-chain modifications are known to exist.24 In the circulation and in cerebrospinal fluid, the AβP distribution profile of the three most common isoforms measured by ELISA is known to be slightly altered in Alzheimer patients as compared to controls.25,26 However, there is a substantial overlap between Alzheimer patients and other types of dementia and thus this specific isoform distribution does not have a high diagnostic specificity.27,28 On-line coupling of multiple chromatographic columns (LC/ LC) has been used with UV detectors for several years, especially for the analysis of drugs and metabolites.29-31 The introduction of the mass spectrometer as an addition or an alternative to the UV detector in these on-line LC/LC systems was also first implemented for the analysis of drugs, often using tandem mass spectrometers in the single ion monitoring or selected reaction monitoring modes.32,33 The direct analysis of unpurified or crudely separated material by tandem mass spectrometry is not applicable in the analysis of larger biomolecules such as full size proteins. However, with the benefits of purity and fast isolation offered by on-line immunoaffinity chromatography, LC-MS analysis of (17) Vincent, C.; Dennoroy, L.; Revillard, J. P. Biochem. J. 1994, 298 (Pt. 1), 181-7. (18) Nissen, M. H.; Thim, L.; Christensen, M. Eur. J. Biochem. 1987, 163, 218. (19) Nissen, M. H. Dan. Med. Bull. 1993, 40, 56-64. (20) Nissen, M. H. H.; Roepstorff, P.; Thim, L.; Dunbar, B.; Fothergill, J. E. Eur. J. Biochem. 1990, 189, 423-9. (21) Heegaard, N. H.; Roepstorff, P.; Melberg, S. G.; Nissen, M. H. J. Biol. Chem. 2002, 277, 11184-9. (22) Miyata, T.; Inagi, R.; Wada, Y.; Ueda, Y.; Iida, Y.; Takahashi, M.; Taniguchi, N.; Maeda, K. Biochemistry 1994, 33, 12215-21. (23) Yankner, B. A.; Dawes, L. R.; Fisher, S.; Villa-Komaroff, L.; Oster-Granite, M. L.; Neve, R. L. Science 1989, 245, 417-20. (24) Lowenson, J. D.; Clarke, S.; Roher, A. E. Methods Enzymol. 1999, 309, 89105. (25) Andreasen, N.; Minthon, L.; Davidsson, P.; Vanmechelen, E.; Vanderstichele, H.; Winblad, B.; Blennow, K. Arch. Neurol. 2001, 58, 373-9. (26) Kanai, M.; Matsubara, E.; Isoe, K.; Urakami, K.; Nakashima, K.; Arai, H.; Sasaki, H.; Abe, K.; Iwatsubo, T.; Kosaka, T.; Watanabe, M.; Tomidokoro, Y.; Shizuka, M.; Mizushima, K.; Nakamura, T.; Igeta, Y.; Ikeda, Y.; Amari, M.; Kawarabayashi, T.; Ishiguro, K.; Harigaya, Y.; Wakabayashi, K.; Okamoto, K.; Hirai, S.; Shoji, M. Ann Neurol. 1998, 44, 17-26. (27) Hulstaert, F.; Blennow, K.; Ivanoiu, A.; Schoonderwaldt, H. C.; Riemenschneider, M.; De Deyn, P. P.; Bancher, C.; Cras, P.; Wiltfang, J.; Mehta, P. D.; Iqbal, K.; Pottel, H.; Vanmechelen, E.; Vanderstichele, H. Neurology 1999, 52, 1555-62. (28) Shoji, M.; Matsubara, E.; Kanai, M.; Watanabe, M.; Nakamura, T.; Tomidokoro, Y.; Shizuka, M.; Wakabayashi, K.; Igeta, Y.; Ikeda, Y.; Mizushima, K.; Amari, M.; Ishiguro, K.; Kawarabayashi, T.; Harigaya, Y.; Okamoto, K.; Hirai, S. J. Neurol. Sci. 1998, 158, 134-40. (29) Flurer, C. L.; Novotny, M. Anal. Chem. 1993, 65, 817-21. (30) Haasnoot, W.; Schilt, R.; Hamers, A. R.; Huf, F. A.; Farjam, A.; Frei, R. W.; Brinkman, U. A. J. Chromatogr. 1989, 489, 157-71. (31) Farjam, A.; de Jong, G. J.; Frei, R. W.; Brinkman, U. A.; Haasnoot, W.; Hamers, A. R.; Schilt, R.; Huf, F. A. J. Chromatogr. 1988, 452, 419-33. (32) Rule, G. S.; Henion, J. D. J. Chromatogr. 1992, 582, 103-12. (33) Cai, J.; Henion, J. Anal. Chem. 1996, 68, 72-8.
biological macromolecules is now feasible. Immunoaffinity LCMS thus has been used to investigate serum proteins34 and the metabolism of therapeutic peptides,35 as well as nucleic acids.36 We here describe an immunoaffinity-liquid chromatography-mass spectrometry (IA-LC-MS) setup for the analysis of amyloid-related proteins and polypeptides with respect to quantitation and detection of isoforms from a crude sample of biological fluid. We introduce in-line filters and an on-line preclear column consisting of coupled control antibody to minimize unspecific interactions with the immunoaffinity column. Using the appropriate immunoreagents, we show the performance of the approach for three amyloid-related proteins and peptides, SAP, AβP, and β2M in serum or cerebrospinal fluid (CSF). EXPERIMENTAL SECTION Materials. All chemicals were analytical grade or better from Merck (Darmstadt, Germany) or Sigma (St. Louis, MO). Phosphatebuffered saline (PBS) was 43 mM Na2HPO4, 9.8 mM KH2PO4, 82 mM NaCl, pH 7.4. Glycine buffer was 0.1 M glycine-HCl, pH 2.6. In-line filters (0.5-µm stainless steel precolumn microfilter and 0.5µm PEEK in-line microfilter) and columns used for immunoaffinity chromatography (2 cm × 1 mm) were from Upchurch Scientific (Oak Harbor, WA); reversed-phase columns, C4, C8, and C18 (5 mm × 0.3 mm) were from LC Packings/Dionex (Sunnyvale, CA). The antibodies used were polyclonal rabbit anti-SAP and mouse IgG control from Dako (Glostrup, Denmark). Monoclonal antiβ2M (clone 290-03) and anti-AβP were both produced at Statens Serum Institut (sold by AntibodyShop, Copenhagen, Denmark). The anti-AβP used was a mixture of several monoclonals or clone 310-08 alone. Aldehyde-activated resin (Poros AL) used for antibody immobilization was from Applied Biosystems (Foster City, CA) and was coupled to antibodies by following the supplier’s instructions. In short, concentrated antibody in 1 M Na2SO4, NaCNBH3, and resin was kept on a shaker overnight after which the resins were blocked using Tris. SAP was purified from outdated plasma according to ref 37. Synthetic AβP was from Bachem AG (Bubendorf, Switzerland). Native β2M was purified according to ref 18 and des-Lys β2M prepared as described.38 Bovine serum albumin (BSA) was from Sigma. Serum was obtained from healthy donors; the usual safety precautions pertaining to biological samples were taken. Instrumentation. The mass spectrometer was an electrospray time-of-flight mass spectrometer (Mariner Biospectrometry Workstation, Applied Biosystems) operated in the positive ion mode with a sampling time of 5 or 6 s/spectrum. The settings were as follows: nozzle potential 160-200 V, spray tip potential 38004000 V, and nozzle temperature 140 °C. Glycine buffer and PBS were pumped isocratically by an HP1100 HPLC system (Agilent Technologies, Palo Alto, CA) and a 501 HPLC pump (Waters, Milford, MA), respectively. The autosampler (Famos) and the nano LC system (Ultimate) providing the organic gradient were (34) Bergen, H. R.; Lacey, J. M.; O’Brien, J. F.; Naylor, S. Anal. Biochem. 2001, 296, 122-9. (35) Zheng, K.; Lubman, D. M.; Rossi, D. T.; Nordblom, G. D.; Barksdale, C. M. Rapid Commun. Mass Spectrom. 2000, 14, 261-9. (36) Huber, C. G.; Oberacher, H. Mass Spectrom. Rev. 2001, 20, 310-43. (37) Heegaard, N. H. H.; Heegaard, P. M. H.; Roepstorff, P.; Robey, F. A. Eur. J. Biochem. 1996, 239, 850-6. (38) Nissen, M. H.; Johansen, B.; Bjerrum, O. J. J. Immunol. Methods 1997, 205, 29-33.
Analytical Chemistry, Vol. 75, No. 5, March 1, 2003
1197
Figure 1. On-line IA-LC-MS setup. Sample flow path is indicated by dashed line. (A) The sample is injected on the immunoaffinity column and washed in PBS. (B) The immunoaffinity column is then eluted with glycine-HCl, pH 2.6, and the eluate directed to the RP column. (C) The RP column is eluted by an acetonitrile organic gradient into the mass spectrometer (MS). Three LC systems delivering PBS, pH 7.4 (isocratic), 0.1 M glycine-HCl, pH 2.6 (isocratic), and organic gradient were connected by two switching valves with actuators.
from LC Packings/Dionex. The autosampler was equipped with a 100-µL loop. Injection of larger volumes was accomplished by consecutive injections. Switching valves with electric actuators were from Valco Vici (Houston, TX). In-line filters were cleaned for reuse in an ultrasonic bath. IA-LC-MS. Samples were injected in PBS by autosampler into an immunoaffinity column which was subsequently washed in PBS (Figure 1A). The column was then eluted with glycine buffer and the eluate directed to a reversed-phase column (Figure 1B). The column was eluted by a 3-min gradient of 0-100% solution B after 10 min of washing in solution A with a flow of 10-20 µL/min (Solution A: water with 2% acetonitrile and 0.2% formic acid. Solution B: acetonitrile with 2% water, 0.2% formic acid) (Figure 1C). Mass spectra were acquired during the gradient elution. For the analysis of β2M from serum and AβP from CSF, two in-line filters after the injection loop were introduced in addition to a 1198
Analytical Chemistry, Vol. 75, No. 5, March 1, 2003
preclear column between the immunoaffinity column and the filters. The preclear column consisted of the same resin as the immunoaffinity column but coupled to a monoclonal control immunoglobulin (Dako) with specificity for Aspergillus niger glucose oxidase which is not present in mammalian tissues. For some of the samples, injection was followed by a 300-µL salt plug injection (PBS, 1 M NaCl). Both salt plug and preclear column were used to reduce unspecific binding to the immunoaffinity column. For the high-flow (100-250 µL/min) part of the system mainly PEEK tubing (i.d. 0.25 mm) was used whereas in the lowflow (10-20 µL/min) part of the system fused-silica capillary (i.d. 50 µm) was used in order to minimize dead volume and diffusion. One analysis took 33 min including injection and wash of the immunoaffinity column (10 min), elution (5 min) wash of reversedphase column (10 min), elution (3 min), and a 5-min system delay. The duration of a following blank run of the reversed-phase column and regeneration of the preclear column by injection of 0.1 M glycine-HCl pH, 2.6, was 22 min; thus, altogether there was 55 min between injections. Reversed-phase column material (C4, C8, C18) was selected to match the hydrophobicity of the analyte. Even though no actual separation was needed, a fast and complete binding and elution of the analyte is dependent on the correct choice of reversed-phase material. Analysis Procedures. Analysis of Amyloid β-Peptide. For AβP analysis, 10 µL of sample consisting of spiked PBS diluted serum was injected. For AβP(1-28), this was 0.1 µM AβP(1-28) in 10 times-diluted serum. Alternatively, 2 times-diluted serum with 1 µM each of the AβP isoforms (1-28, 10-37, and 1-40) was injected. For CSF AβP analysis, 600 or 900 µL of crude CSF was injected followed by a 300-µL salt plug. Both glycine buffer and PBS were delivered at 100 µL/min, and the reversed-phase column was C18 or C8. For the analysis of low dilutions of sera and CSF, the system contained a preclear column as described under IALC-MS. Analysis of Serum Amyloid P Component. For the SAP analysis, a total volume of 150 µL of serum diluted 1:1 in PBS was injected into the system. PBS was pumped at 250 µL/min and glycine buffer at 75 µL/min, and the reversed-phase column was C8. Analysis of β2M. The β2M setup was used to analyze standard samples of purified des-Lys β2M and normal β2M in PBS containing 77 mg/mL BSA to provide a solution mimicking the conditions of serum samples. Standard samples were injected in volumes of 40 µL. To demonstrate the preclearing strategy, 200 µL of serum was injected followed by injection of 300 µL of PBS with 1 M NaCl or as described in the caption of Figure 2. PBS and glycine were delivered at 100 µL/min, and the reversed-phase column was C4. When a preclear column was included, 300 µL of glycine buffer was injected to regenerate the column after the run. All sera were centrifuged for 20 min at 16000g before injection. Relative Quantitation of β2M Isoforms. For the relative quantitation of isoforms of β2M, the peak area of five charge states from each of the four isoforms was used. As a measure of chemical noise, the summed peak area of the ion chromatogram of five background ions within the chemical background between the β2M ions was subtracted in each run. For the comparison of desLys β2M with native β2M, the areas of oxidized and nonoxidized isoforms were added to give a total measure of the two variants. The native and the des-Lys β2M isoforms analyzed here gave
Figure 2. Demonstration of the effect of preclearing involving preclear column and injection of 300 µL of PBS/1 M NaCl after the injection of a 200 µL of normal serum sample analyzed for β2M. Combined ion chromatogram of two charge state ions (m/z 1110.4 and 1295.1) of the most predominant unspecific binding component (A, C) compared to the combined ion chromatogram of the two most abundant β2M charge states (m/z 1067.3 and 1173.9) (B, D). (A) and (B) were analyzed without preclear, (C) and (D) with preclear.
comparable signals when the system was tested with standard samples, and thus, the relative concentration of the isoforms could be determined without the use of a standard curve. RESULTS AND DISCUSSION With the IA-LC-MS system presented here, the sample to be analyzed was injected into an immunoaffinity column to capture an analyte or a group of analytes, and after washing (Figure 1A), the affinity-purified analytes were eluted onto a reversed-phase column for desalting (Figure 1B). Finally, the isolated and desalted polypeptide was eluted into the mass spectrometer by an organic gradient (Figure 1C). The run time of the entire analysis was ∼55 min, including a final blank run of the reversed-phase column. Because of the automatic nature of the system, samples could be analyzed unattended. The reversed-phase columns used allowed both the higher flow rate (75-100 µL/min) when the immunoaffinity column was being eluted and the lower flow rate (10-20 µL/min) necessary when the reversed-phase column was being eluted into the mass spectrometer. Because only desalting and no actual separation was needed, the short cartridgelike reversedphase columns could be used. The immunoaffinity columns could be used for at least 50-100 analyses before losing their ability to trap the antigen, and repacking with new affinity resin was needed. The need for repacking could be detected by injections of standards which was done regularly. We did not see any run-torun carry-over from the immunoaffinity column. By use of changeable in-line filters in experiments using serum samples, we were able to avoid clogging of the immunoaffinity column, even at higher volumes of sample, e.g., 200 µL, and this also prolonged the lifetime of the reversed-phase column. The filters could easily
be changed and reused after ultrasonication. The use of on-line precolumn preclearing followed by an injection of 300 µL of PBS with 1 M NaCl was very efficient in suppressing nonspecific binding especially when the system was tested using larger amounts of serum (>50 µL) as demonstrated in Figure 2. These amounts of sample caused unacceptable unspecific binding for quantitative analyses in the β2M experiments, but the preclearing procedure solved this problem. As an example, in the analysis for β2M in serum, the presence of a preclear column when 200 µL of serum was injected reduced the main unspecific component by ∼50% (data not shown) and a subsequent plug of 300 µL of PBS with 1 M NaCl further reduced the component to ∼6% of the initial value (Figure 2). This plug, however, also slightly reduced the β2M signals, as seen in Figure 2B compared to Figure 2D. Analysis of AβP. The Alzheimer’s disease (AD)-associated peptides AβP are found in several isoforms up to 43 amino acids in length and with varying degrees of side-chain modifications.24 The distribution of isoforms to some degree appears to be associated with the AD diagnosis though only few of the known isoforms have been evaluated for diagnostic use.5,26,39-41 We have (39) Di Luca, M.; Pastorino, L.; Bianchetti, A.; Perez, J.; Vignolo, L. A.; Lenzi, G. L.; Trabucchi, M.; Cattabeni, F.; Padovani, A. Arch. Neurol. 1998, 55, 1195200. (40) Hulstaert, F.; Blennow, K.; Ivanoiu, A.; Schoonderwaldt, H. C.; Riemenschneider, M.; De Deyn, P. P.; Bancher, C.; Cras, P.; Wiltfang, J.; Mehta, P. D.; Iqbal, K.; Pottel, H.; Vanmechelen, E.; Vanderstichele, H. Neurology 1999, 52, 1555-62. (41) Galasko, D.; Chang, L.; Motter, R.; Clark, C. M.; Kaye, J.; Knopman, D.; Thomas, R.; Kholodenko, D.; Schenk, D.; Lieberburg, I.; Miller, B.; Green, R.; Basherad, R.; Kertiles, L.; Boss, M. A.; Seubert, P. Arch. Neurol. 1998, 55, 937-45.
Analytical Chemistry, Vol. 75, No. 5, March 1, 2003
1199
Figure 4. Raw data of AβP(1-40) ions at charge states of 4 and 5 detected in CSF from two nondemented donors by use of IAP-RPMS with preclear and salt plug. The theoretical average m/z values are 866.98 (z ) 5) and 1083.47 (z ) 4); there is no isotope resolution at this level of material. (A) Raw data from analysis of 600 µL of CSF. (B) Raw data analysis of 900 µL of CSF. Figure 3. Spectra of AβP analyzed by IA-LC-MS. (A) Three synthetic AβP peptides spiked in serum diluted two times in PBS. Upper panel: Raw data from analysis of 10-µL sample containing 1 µM AβP(1-28) (A), AβP(1-40) (B), and AβP(10-37) (C). A carboxypeptidase B-like proteolytic activity in serum converts AβP(1-28) to AβP(1-27) (D), and the AβP(6-40) (E) tryptic fragment of AβP(1-40) appears. Lower panel (B): Deconvoluted spectrum after injection of 10 µL of serum (diluted 10 times with PBS) spiked with 0.1 µM AβP(1-28). The carboxypeptidase B-like cleavage of Cterminal lysine happens practically immediately after spiking the peptide into serum. The theoretical monoisotopic masses and m/z are listed in Table 1. Table 1. Theoretical Monoisotopic Masses at Different Charge (z) States of the AβP Isoforms Detected by IA-LC-MS of Spiked Serum as Shown in Figure 3 AβP isoform
monoisotopic mass m/z, z ) 3 m/z, z ) 4 m/z, z ) 5 m/z, z ) 6
A: 1-28
B: 1-40
C: 10-37
D: 1-27
E: 6-40
3260.5
4327.1
3057.6
3132.4
3708.9
1087.9
1443.4 1082.8 866.3 722.2
1020.2 765.4
1045.2 784.1 627.5
1237.3 928.2
raised monoclonal antibodies against the amino acid 5-16 Nterminal region of AβP as most isoforms span this region. Using these antibodies, we thus anticipated to be able to catch the major fraction of the pool of AβP isoforms, that is, the three major isoforms AβP(1-40/42/43) as well as the several additional C-terminal variants known to exist5 in addition to peptides with N-terminal heterogeneity and posttranslational modifications.24 The immunoaffinity component of the IA-LC-MS on-line system described here for analysis of AβP contained either a single clone (310-08) (Figure 3A and Figure 4) or a mixture of monoclonals (Figure 3B). Analysis of 10 µL of diluted serum spiked to 1 µM AβP(1-28), AβP(1-40), and AβP(10-37) shows in addition to the three spiked isoforms two proteolytically processed isoforms corresponding to AβP(1-28) lacking C-terminal lysine after a carboxypeptidase B-like cleavage giving rise to AβP(1-27) and a tryptic fragment of AβP(1-40), AβP(6-40). The sample was 1200 Analytical Chemistry, Vol. 75, No. 5, March 1, 2003
analyzed a few hours after preparation and yet the AβP(1-28) is almost completely converted into AβP(1-27) (Figure 3A). In Figure 3B, the serum was diluted 10 times and analyzed immediately after spiking to 0.1 µM AβP (1-28) and thus a smaller fraction of peptide is cleaved. Two CSF samples from nondemented donors were subsequently analyzed. By injecting 600-900-µL volumes of undiluted CSF, we were able to detect the most abundant AβP isoform, AβP(1-40) which is present in CSF at ∼1.5 nM, almost 10 times more than the AβP(1-42/43) level28,42 (Figure 4). No other isoforms were readily detected. The limit of detection of the system when AβP is analyzed is the main challenge for work with clinical samples, but the results presented here suggest that suitable sample sizes are beginning to be in the realm of the feasible. The results are very much dependent on the performance of the mass spectrometer and on the propensity of the longer AβP isoforms to aggregate and adhere to the surfaces encountered on the way from sample to mass spectrometer. We obtain acceptable spectra down to ∼1 pmol of injected peptide. The system’s limit of detection is here evaluated with AβP(1-28) and AβP(1-40), and it will most likely be considerably higher for the more amyloidogenic peptide species, i.e., AβP(1-42/43), as they are more prone to precipitation and adsorption during the analysis. When the perfomance specifications of the mass spectrometer are considered, it seems reasonable to expect that the detection limit could be enhanced by a factor of at least 10. Also, an MS/MS option would make it possible to verify the identity of lowabundance peptides and putative new isoforms by sequencing. That might make it feasible to do analysis on biological samples from AD patients and controls and thereby determine whether the mass distribution in the total pool of AβP might be a useful diagnostic marker. The fact that we identify both C-terminally and N-terminally cleaved versions of spiked isoforms of AβP in serum indeed underscores the point that the approach might reveal unexpected modifications in contrast to methods such as those (42) Fukuyama, R.; Mizuno, T.; Mori, S.; Nakajima, K.; Fushiki, S.; Yanagisawa, K. Eur. Neurol. 2000, 43, 155-60.
Figure 5. Spectrum of SAP from normal serum. A total of 150 µL of serum was injected for the IA-LC-MS analysis. (A) Raw data. (B) Deconvoluted spectrum. The measured mass of SAP is m/z 25 467, which corresponds well with the theoretical mass of m/z 25 462.
based on, for example, enzyme-linked immunosorbent assays (ELISA). This suggests the potential of the system in identifying novel disease markers. Analysis of SAP. SAP is known to be a structurally very homogeneous molecule with no known amino acid or glycosylation heterogeneity in normal individuals.43 However, it is unknown whether SAP may be modified posttranslationally in chronic disease states. SAP is tightly linked to amyloidoses by virtue of its presence in all types of amyloid deposits and may play a role in the altered scavenger functions that are possibly involved in systemic lupus erythematosus (SLE).44 Thus, it is of interest to investigate individual patients for SAP modifications and we therefore applied the approach presented here to a study of SLE patients.45 When optimized for the analysis of SAP using commercially available polyclonal antibodies, we were able to use the IA-LC-MS setup to measure the SAP mass directly in a single serum sample of 150 µL (Figure 5). This makes it possible to investigate whether anything has happened to the structure of SAP in patient serum samples. The measured mass of 25 467 observed here is in close agreement with the theoretical mass of SAP at 25 462.43 The glycan of SAP is prone to form sodium adducts, and this most likely gives rise to the shoulder of the peak observed in the spectrum. Analysis of β2M. The reason for the ordered precipitation in amyloid deposits of β2M in renal patients is unknown but might be due to chronically elevated serum levels of β2M in combination with the appearance of new or altered levels of isoforms. The system described here provides a tool for detection of such isoforms and for the relative quantitation of these because they (43) Pepys, M. B.; Rademacher, T. W.; Amatayakul-Chantler, S.; Williams, P.; Noble, G. E.; Hutchinson, W. L.; Hawkins, P. N.; Nelson, S. R.; Gallimore, J. R.; Herbert, J. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 5602-6. (44) Breathnach, S. M.; Kofler, H.; Sepp, N.; Ashworth, J.; Woodrow, D.; Pepys, M. B.; Hintner, H. J. Exp. Med. 1989, 170, 1433-8. (45) Sen, J. W.; Recke, C.; Rahbek, L.; Skogstrand, K.; Heegaard, N. H. H. Scand. J. Immunol. 2002, 56, 645-651.
Figure 6. Quantitation and spectrum of mixed β2M isoforms. Mixtures were prepared in 77 mg/mL BSA to mimic the total protein concentration of serum. (A) Influence of total β2M concentration on the relative quantitation of the des-Lys β2M. The relative concentration of des-Lys β2M is kept at 20% whereas the total β2M concentration is varied from 5 to 90 µg/mL. Each data point represents the mean of three independent measurements; the standard deviations are indicated. (B) Spectrum of the 50 µg/mL data point. (C) Deconvoluted spectrum. See Table 2 for theoretical β2M masses. Table 2. Theoretical Average Masses of β2M Isoforms
native β2M des-Lys β2M
nonoxidized Met99
oxidized Met99
11 729.17 11 619.01
11 745.17 11 635.01
appeared to ionize comparably as seen in Figure 6A. To elucidate the influence of total β2M concentration on the relative signal of the isoforms, the same distribution of isoforms (20% des-Lys β2M) was prepared at total protein concentrations ranging from 5 to 90 µg/mL and the signal distribution between the isoforms was subsequently calculated. The ratio between the signal from desLys β2M and the native β2M was the same throughout the range, though as expected, the variance at a lower total protein concentration was larger (Figure 6A). Raw data and deconvoluted spectra of a standard sample of the β2M isoforms mixed at 20% des-Lys β2M and 80% native β2M (50 µg/mL total β2M) are shown in Figure 6B and C, respectively. Both isoforms also have a fraction oxidized at methionine 99. (Refer to Table 2 for theoretical average masses of the isoforms.) All four isoforms can easily be distinguished in the spectra. In these experiments, albumin was added in large excess (350 times molar excess) to mimic the protein environment of serum. Despite this, the analyses were devoid of albumin signals. Material occasionally precipitated on the interface of the mass spectrometer and lowered the total ion current. To investigate whether this had an impact on the relative Analytical Chemistry, Vol. 75, No. 5, March 1, 2003
1201
signal or the quantitative calculations of the β2M isoforms, standard runs over a couple of months using several different relative concentrations of isoforms were compared. When the relative signal of each isoform was compared as a function of total β2M signal for several relative concentrations of des-Lys β2M, there was no statistically significant correlation; i.e., the relative signal of the isoforms remained the same and thus the quantitation was applicable at all levels of sensitivity of the mass spectrometer (data not shown). CONCLUSION The system described here provides a tool for the identification and quantitation of isoforms and relative abundances of these as well as for measuring masses of selected molecules in individual patient samples as demonstrated by the SAP experiment. Thus, the approach is excellent for discovery and development of new diagnostic leads. Today, ELISA techniques often are the tools of choice for diagnostic tests based on biomarkers in biological fluids, but the use of immunochemsitry alone has limitations when it comes to detection of new markers as it nondiscriminately detects the total pool of antibody-defined species. Due to the extensive similarities between wild-type and pathogenic variants, it can be very difficult if not impossible to raise specific antibodies and indeed to discover the presence of such variants. In the approach described here, known as well as unknown modifications can be detected by use of a monoclonal antibody, combinations of monoclonals, or polyclonal antibody that binds the pool of native protein and variants. Thereby a new test can easily be set up to elucidate whether there are in fact pathological changes associated with a specific biomolecule and what these changes are. For the screening of many patient samples at the same time, the ELISA approach is advantageous by offering parallel sample analysis and a high degree of ruggedness as compared to the IA-LC-MS system. However, upcoming developments in MS systems with multiple channels and parallel columns might, at least in part, live up to the characteristics traditionally in favor of ELISA.46 When one approach is chosen over the other, it has to be considered (46) Deng, Y.; Wu, J. T.; Lloyd, T. L.; Chi, C. L.; Olah, T. V.; Unger, S. E. Rapid Commun. Mass Spectrom. 2002, 16, 1116-23. (47) Atwood, C. S.; Martins, R. N.; Smith, M. A.; Perry, G. Peptides 2002, 23, 1343-50.
1202
Analytical Chemistry, Vol. 75, No. 5, March 1, 2003
whether to test for a known species without regard to the possible presence of related unknown species or whether complete information about modifications is needed. In the case of amyloidosis, much is still unknown with respect to seeding, modifications, levels of different modifications, and effect of modifications on the amyloidogenic potential.47 This makes the MS approach desirable also for understanding disease pathogenesis. Further developments of the system will be improved detection limits and the possibility of doing identification and characterization of modifications as they are detected, e.g., by use of a tandem mass spectrometer. The versatility of the system is apparent from the difference in the analytes involved ranging from peptides and small proteins to larger glycosylated proteins in crude biological fluids. Though the system here is used for analysis of markers associated with amyloidosis, it could in principle be used for the investigation of any set of biomarkers in biological fluids as long as appropriate antibodies are available. The unique exploitation of immunoaffinity specificity for selecting analytes and of the mass resolution and accuracy of MS for detecting them is powerful. We have shown the feasibility of an IA-LC-MS setup for the analysis of three different polypeptides with respect to mass determination, identification of isoforms, and relative quantitation of isoforms. It is a robust, modular, and versatile approach because with few modifications, i.e., a new antibody and possibly a change of reversedphase column, the system can be altered to investigate close to all soluble proteins and peptides from patient samples for the presence of modified species or isoform profiles. ACKNOWLEDGMENT We thank Dr. Mogens H. Nissen for kindly supplying purified β2M and des-Lys β2M. This work was supported by The Danish Medical Research Council, Lundbeckfonden, Fonden til Lægevidenskabens Fremme, M. L. Jørgensen og Gunnar Hansens Fond, and Apotekerfonden af 1991 (NH) and by a grant from Forskningsstyrelsen (J.W.S.).
Received for review September 27, 2002. Accepted December 16, 2002. AC026174B
