Assessment Approach for Evaluating High Abundance Protein Depletion Methods for Cerebrospinal Fluid (CSF) Proteomic Analysis Kevin S. Shores and Daniel R. Knapp* Department of Pharmacology and MUSC Proteomics Center, Medical University of South Carolina, Charleston, South Carolina 29425 Received May 17, 2007
Optimal proteomic analysis of human cerebrospinal fluid (CSF) requires depletion of high-abundance proteins to facilitate observation of low-abundance proteins. The performance of two immunodepletion (MARS, Agilent Technologies and ProteoSeek, Pierce Biotechnology) and one ultrafiltration (50 kDa molecular weight cutoff filter, Millipore Corporation) methods for depletion of abundant CSF proteins were compared using a graphical method to access the depth of analysis using “marker proteins” with known normal concentration ranges. Two-dimensional LC/MS/MS analysis of each depleted sample yielded 171 and 163 unique protein identifications using the MARS and ProteoSeek immunodepletion methods, respectively, while only 46 unique proteins were identified using a 50 kDa molecular weight cutoff filter. The relative abundance of the identified proteins was estimated using total spectrum counting and compared to the concentrations of 45 known proteins in CSF as markers of the analysis depth. Results of this work suggest a clear need for methodology designed specifically for depletion of high-abundance proteins in CSF, as depletion methods designed to deplete high-abundance serum proteins showed little improvement in analysis depth compared to analysis without depletion. The marker protein method should be generally useful for assessing depth of analysis in the comparison of proteomic analysis methods. Keywords: Cerebrospinal fluid • proteomics • LC/MS/MS • abundant protein depletion
1. Introduction Cerebrospinal fluid (CSF) is a translucent fluid that surrounds the ventricles of the brain and spinal cord. The total volume of CSF varies from 150 to 270 mL in adult subjects.1 CSF is produced and absorbed by surrounding tissue at the rate of 0.34 ( 0.13 mL/min resulting in a complete turnover of the entire volume in approximately 7-8 h.1 Because of its continuous contact with the extracellular regions of the brain, CSF contains proteins and other metabolites that reflect the status of the central nervous system (CNS)1 and thus is a potential medium for detection of biomarkers for diseases of the CNS. This has prompted multiple characterizations of the CSF proteome for both normal2-4 and disease states5-8 using twodimensional gel electrophoresis or liquid chromatography separation followed by mass spectrometric analysis. Determination of disease biomarkers in CSF is challenging due to the relatively low total protein concentration of CSF (0.2 - 0.8 mg/mL).1 Furthermore, a large amount of blood-derived proteins permeate the blood-brain barrier and appear in CSF at high concentrations,1 with albumin and immunoglobulins constituting approximately 50% and 15% of the total protein content, respectively.9 In addition, other proteins that are not abundant serum proteins are highly abundant in CSF including prostaglandin D2 synthase and cystatin C. Considering that * Corresponding author. E-mail:
[email protected]. 10.1021/pr070293w CCC: $37.00
2007 American Chemical Society
many potential protein biomarkers exist at far lower concentrations in CSF, and that proteomic analysis methods have a limited dynamic range, high-abundance protein depletion is required to observe lower-abundance proteins. A variety of methods have been used to deplete abundant proteins prior to proteomic analysis. Cibacron Blue chromatography is a commonly used method of removing albumin. However, it is a nonspecific method and has the potential to also bind low-abundance proteins of interest.10 Solvent depletion strategies have been used to selectively fractionate albumin, immunoglobulins, and lower-abundance proteins with several different concentrations of acetonitrile. Subsequent proteomic analyses of the fractions enabled the identification of more than 300 proteins (although as little as one peptide sequence resulting from a single MS/MS spectrum was used for protein identification).3 Using ultrafiltration with 50 kDa molecular weight cutoff filters to remove albumin (∼66 kDa) and other high molecular weight proteins prior to 2D LC/MS/ MS, Noben et al. identified 148 proteins in CSF.5 Immunodepletion products designed for serum analysis have been used to deplete high-abundance proteins from CSF since it contains high levels of serum proteins, including albumin, immunoglobulins, antitrypsin, and haptoglobin. However, CSF also contains several high-abundance proteins that are not targeted by these methods such as prostaglandin D2 synthase, cystatin C, and transthyretin (Prealbumin).1 Maccarrone et al. Journal of Proteome Research 2007, 6, 3739-3751
3739
Published on Web 08/16/2007
research articles processed CSF with the Multiple Affinity Removal System (MARS, Agilent Technologies),2 which includes an HPLC column that binds HSA, transferrin, IgG, IgA, antitrypsin, and haptoglobin. In combination with 2D LC/MS, they identified 100 proteins in CSF. Ogata et al. compared the MARS HPLC column to a spin cartridge containing avian IgY antibodies to HSA, transferrin, IgG, IgA, IgM, and fibrinogen (Beckman Coulter).11 In analysis of pooled normal CSF using 2D gel electrophoresis and MS, they reported superior performance of the MARS HPLC column, presumably due to the removal of antitrypsin and haptoglobin. Righetti et al. have recently demonstrated reduction of the concentration of high-abundance proteins in human serum using a combinatorial solidphase peptide library,12 which theoretically binds every protein in a complex sample up to a saturation limit, thereby lowering the overall dynamic range of protein concentrations in the processed sample. However, there have not been any reports of applying the peptide library method to CSF nor have there been any reports of quantitative applications of this method. The work presented here derives from method development efforts for high-throughput proteomic analysis of CSF. For preparation of CSF samples for LC/MS/MS analysis, we compared three high-abundance protein depletion methods: the six protein MARS spin cartridge (Agilent Technologies); the ProteoSeek spin cartridge (Pierce Biotechnology), an immunodepletion cartridge that removes albumin and IgG; and a 50 kDa molecular weight cutoff filter (Amicon Ultra, Millipore Corporation). We present here a comparison of these depletion methods using a graphical method for visualizing the effect of depletion upon the dynamic range of observable proteins through comparison of the analysis results to proteins of known normal concentration ranges in CSF.1 This marker protein method should also be generally applicable for comparing depletion methods for other kinds of biological samples.
2. Experimental Section 2.1. Materials. MARS spin cartridges were obtained from Agilent Technologies (Wilmington, DE), ProteoSeek spin cartridges were obtained from Pierce Biotechnology (Rockford, IL), and Amicon Ultra 5 kDa and 50 kDa molecular weight cutoff filters were obtained from the Millipore Corporation (Billerca, MA). Urea, dithiothreitol, iodoacetamide, proteomics grade trypsin, and a bicinchoninic acid (BCA) kit for total protein determination were obtained from Sigma-Aldrich corporation (St. Louis, MO). The 2.0 mL C-18 Sep-Pak vacuum manifold solid-phase extraction cartridges (SPE) were obtained from the Waters corporation (Milford, MA). 2.2. Human CSF. A pool of human CSF was generated from multiple deidentified excess clinical specimens obtained under institutional review board approval from the MUSC Hospital clinical chemistry laboratory. Four, 3.2 mL aliquots were centrifuged at 3000g for 30 min, 3.0 mL of supernatant was removed from each aliquot and isolated, and 100 µL of CSF was set aside for total protein determination using a BCA protein assay (Sigma-Aldrich Corporation). The total protein concentration was determined to be 427 µg/mL. The 3.0 mL aliquots were then dried using vacuum centrifugation. 2.3. Abundant Protein Depletion. 2.3.1. Abundant Protein Depletion with the Multiple Affinity Removal System. Immunodepletion using the MARS spin cartridge was performed according to manufacture’s instructions. Proteins from 3.0 mL of CSF were solublized in 1.2 mL of 20.0 mM Tris-buffered saline (TBS) (1.21 g of Tris free base, 8.7 g of NaCl, 500 mL of 3740
Journal of Proteome Research • Vol. 6, No. 9, 2007
Shores and Knapp
water, pH adjusted to 7.4 with 1.0 M HCl) and then split into three, 400 µL aliquots. The spin cartridge was designed to process 8.0 µL of human serum (480 µg protein); therefore, the three 400 µL aliquots of concentrated CSF (427 µg of protein) were processed consecutively. The first aliquot was loaded into the spin cartridge where high-abundance proteins were retained. The spin cartridge was then centrifuged, and the depleted CSF sample was eluted and collected. The spin cartridge was washed with 100 mM Glycine HCl to remove bound proteins and was then reequilibrated with TBS. This process was repeated for the two remaining 400 µL aliquots of CSF, and the three depleted CSF samples were pooled. 2.3.2. Abundant Protein Depletion Using ProteoSeek Immunodepletion Spin Cartridges. ProteoSeek immunodepletion spin cartridges are designed to deplete albumin and IgG from human serum; therefore, the manufacturer’s instructions were scaled proportionally to accommodate for the lower protein concentration of CSF. Proteins from 3.0 mL of CSF were solubulized in 600 µL of 0.2 M NH4HCO3, and 200 µL of solubulized protein (427 µg) was loaded in each of three different spin cartridges. A 400 µL aliquot of a 62% slurry containing immobilized Anti-HSA/Anti-IgG settled gel, capable of processing 450 µg of human serum proteins, was added to each spin cartridge. The cartridges were agitated gently for 30 min at room temperature to maximize binding to the antibodies. Each spin cartridge was then centrifuged, and the albuminand IgG-depleted CSF protein solutions were pooled. 2.3.3. Abundant Protein Depletion Using a 50 kDa Molecular Weight Cutoff Filter. A 3.0 mL aliquot of the CSF pool was transferred to a 15 mL capacity Amicon Ultra 50 kDa molecular weight cutoff filter. The filter was then centrifuged for 45 min at 3000g, and the flow through containing the low molecular weight CSF proteins was collected. The Amicon Ultra molecular weight cutoff filters contain a “V” shaped filter which retains a minimum of 200 µL of unfiltered solution. This feature allows for easy removal of a concentrated sample; however in the present application of the filter device, we sought to maximize flow through of our low molecular weight proteins. This was achieved by adding an additional 3.0 mL of 0.2 M NH4HCO3 to the filter, centrifuging until the retained volume was reduced to 200 µL, and collecting the flow through. The retained high molecular weight proteins (>50 kDa) were discarded, and the flow through fractions were combined. 2.4. Buffer Exchange and Removal of Biological Salts. Depleted CSF proteins from each sample (and supernatant from the undepleted sample) were loaded into individual 15.0 mL capacity 5 kDa Amicon Ultra molecular weight cutoff filters. The volume was then increased to 15.0 mL total by adding 0.2 M NH4HCO3, and the cartridges were centrifuged at 3000g until the volume was reduced to 200 µL. This buffer exchange process was repeated two more times resulting in the substantial reduction of biological salt concentration. The retentates containing CSF proteins were then removed and dried via vacuum centrifugation. 2.5. Reduction, Akylation, and Digestion of CSF Proteins. CSF proteins in each sample were denatured in 450 µL of 8.0 M urea. Reduction of disulfide bonds was performed in 10 mM dithiothreitol for 1 h. at 37 °C in the dark with constant agitation, immediately followed by alkylation in 50 mM iodoacetamide for 1 h. at 37 °C in the dark with constant agitation. The CSF protein mixture was then diluted to 1.85 M urea with addition of 0.2 M NH4HCO3 (pH 8.5). Proteomics grade trypsin was added (20 µg, Sigma-Aldrich), and the
research articles
Evaluating High-Abundance Protein Depletion Methods for CSF Table 1. Proteins Identified in the Nondepleted CSF Samplea protein ID
UniProt ID
Mr
TSC
% abundance
Albumin Hypothetical protein LOC651928 Hypothetical protein LOC649897 Prostaglandin D2-synthase Cystatin C Alpha-1-antitrypsin Transthyretin Transferrin Anti-RhD monoclonal T125 gamma1 heavy chain Apolipoprotein A-I preproprotein Beta-2-microglobulin Apolipoprotein E Vitamin D-binding protein Hemopexin Apolipoprotein A-II preproprotein PREDICTED: similar to Ig gamma-3 chain C region Alpha 2 globin Orosomucoid 2 Serpin peptidase inhibitor, clade A member 3 Orosomucoid 1 Beta globin Haptoglobin-related protein PREDICTED: similar to Ig alpha-1 chain C region isoform 1 Apolipoprotein D Complement component 3 Clusterin isoform 1 Kallikrein 6 isoform B ATP-binding cassette, subfamily D, member 3 Haptoglobin Angiotensinogen preproprotein PREDICTED: similar to Ig gamma-4 chain C region Pancreatic ribonuclease Alpha-2-glycoprotein 1, zinc Epididymal secretory protein E1 Apolipoprotein H Serine (or cysteine) proteinase inhibitor, clade F Complement component 4A preproprotein PREDICTED: similar to Ig gamma-1 chain C region C-type lectin domain family 3, member B Gelsolin isoform B Alpha 1B-glycoprotein Delta globin Ceruloplasmin (ferroxidase) PREDICTED: similar to Ig gamma-2 chain C region Retinol-binding protein 4, plasma Insulin-like growth factor binding protein 6 Dickkopf homologue 3 Insulin-like growth factor binding protein 7 Alpha-2-macroglobulin Complement factor B preproprotein PREDICTED: similar to Ig kappa chain V-III region HAH CD59 antigen p18-20 Kininogen 1 Serine (or cysteine) proteinase inhibitor, clade C Galectin 3 binding protein Secreted phosphoprotein 1 isoform A Complement component 1 inhibitor Family with sequence similarity 3, member C PREDICTED: similar to Ig kappa chain V-III region VG Osteoglycin preproprotein Lumican Superoxide dismutase 3, extracellular CD14 antigen PREDICTED: similar to Ig kappa chain V-I region Walker Fibrinogen, beta chain preproprotein Fibrinogen, gamma chain isoform gamma-B SPARC-like 1 Carnosinase 1 Alpha-2-HS-glycoprotein Complement factor H isoform B UDP-GlcNAc:betaGal beta-1,3-N-acetylglucosaminyltransferase 1 Leucine-rich alpha-2-glycoprotein 1 Apolipoprotein A-IV
P02768 946295 Q49AS2 Q5SQ09 P01034 P01009 P02766 P02787 Q5EFE5 P02647 Q9UM88 P02649 P02774 P02790 P02652 947052 Q86YQ5 Q5T538 P36955 P02763 P68871 P00739 Q8NCL6 P05090 Q6LDJ0 Q96AJ1 Q92876 P33897 P00738 P01019 P01861 P07998 P02765 P61916 P02749 Q4R6H4 Q5JNX2 652050 P05452 137150 P04217 P02042 P00450 P01859 P02753 P24592 Q4R417 Q16270 P01023 P00751 642113 107271 P01042 P32261 P17931 000571 060860 Q5HY75 642113 P20774 P51884 O14618 P08571 642838 P02675 068656 Q14515 Q96KN2 P02765 P08603 O43505 P02750 P06727
69321 26229 22059 21015 15789 46707 15877 77000 52253 30758 13705 36131 52883 51643 11167 18069 15247 23587 47620 23496 15988 39004 28490 21261 187045 57795 15045 75427 45176 53120 47414 17632 34237 16559 38286 46283 192664 64280 22552 80590 54219 16045 122127 39968 22995 25306 38365 29111 163188 85478 16578 14167 47852 52568 65289 35401 55119 24664 16662 33900 38404 25864 40050 13580 55892 51478 75201 56656 39299 50974 47088 38154 45344
2496 162 128 99 70 198 55 247 159 69 24 50 58 56 12 19 15 22 44 21 14 30 21 14 116 35 9 45 25 29 25 9 17 8 18 21 87 29 10 35 23 6 43 14 8 8 12 9 50 26 5 4 13 14 17 9 14 6 4 8 9 6 9 3 12 11 16 12 8 10 8 6 7
32.75 5.62 5.28 4.28 4.03 3.86 3.15 2.92 2.77 2.04 1.59 1.26 1.00 0.99 0.98 0.96 0.89 0.85 0.84 0.81 0.80 0.70 0.67 0.60 0.56 0.55 0.54 0.54 0.50 0.50 0.48 0.46 0.45 0.44 0.43 0.41 0.41 0.41 0.40 0.39 0.39 0.34 0.32 0.32 0.32 0.29 0.28 0.28 0.28 0.28 0.27 0.26 0.25 0.24 0.24 0.23 0.23 0.22 0.22 0.21 0.21 0.21 0.20 0.20 0.20 0.19 0.19 0.19 0.19 0.18 0.15 0.14 0.14
Journal of Proteome Research • Vol. 6, No. 9, 2007 3741
research articles
Shores and Knapp
Table 1 (Continued) protein ID
UniProt ID
Mr
TSC
% abundance
Secretogranin III Complement component 1, q subcomponent, B chain Complement factor D preproprotein Alpha-2-plasmin inhibitor Proprotein convertase subtilisin/kexin type 1 inhibitor Prion protein preproprotein Secreted protein, acidic, cysteine-rich (osteonectin) CutA divalent cation tolerance homologue isoform 1 PREDICTED: similar to Ig heavy chain V-III region VH26 Amyloid like protein 1 isoform 2 Chromogranin A Complement factor H isoform A Histidine-rich glycoprotein Limbic system-associated membrane protein Alpha-1-microglobulin/bikunin PREDICTED: similar to Ceruloplasmin (Ferroxidase) Tissue inhibitor of metalloproteinase 2 Autotaxin isoform 2 preproprotein Afamin Chromogranin B Complement component 1, r subcomponent Plasminogen EGF-containing fibulin-like extracellular matrix protein 1 Fibulin 1 isoform B Cell adhesion molecule with homology to L1CAM Fibrinogen, alpha polypeptide isoform alpha preproprotein Peptidoglycan recognition protein L Insulin-like growth factor binding protein 2, 36kDa Neuronal pentraxin I Fibronectin 1 isoform 3 preproprotein Secretory granule, neuroendocrine protein 1 (7B2 protein) NADH dehydrogenase (ubiquinone) Fe-S protein 8, 23kDa Amyloid beta A4 protein, isoform A Complement component 9 Opioid binding protein/cell adhesion molecule-like isoform A preproprotein I factor (complement) Carboxypeptidase E Neural cell adhesion molecule 1 isoform 1 Complement component 7 VGF nerve growth factor inducible PREDICTED: similar to Probable endonuclease KIAA0830 isoform 3 Chitinase 3-like 1 Inter-alpha (globulin) inhibitor H4 Cathepsin D preproprotein Corticosteroid binding globulin Extracellular matrix protein 1 isoform 1 Contactin 1 isoform 1 ATPase, H+ transporting, lysosomal accessory protein 1 Signal-regulatory protein alpha Inter-alpha globulin inhibitor H2 polypeptide Gelsolin isoform A Neuronal cell adhesion molecule isoform B Neural cell adhesion molecule 2 Secretogranin II Inter-alpha (globulin) inhibitor H1 Complement component 2 Dystroglycan 1 Cartilage acidic protein 1 Calsyntenin 1 isoform 1 Complement component 4B preproprotein Neogenin homolog 1 Fc fragment of IgG binding protein
Q8WXD2 Q5T960 P00746 P08697 P29120 P04156 Q4R5R0 O60888 P01764 P51693 P10645 P08603 P04196 Q13449 Q5TBD7 877964 P16035 1035181 P43652 P05060 Q53HT9 P06733 Q12805 Q59G97 Q59FY0 P02671 Q96PD5 P18065 Q15818 P02751 P05408 A0AV68 P05067 Q9UGI4 Q14982 Q4R955 P16870 P13592 Q8TCS7 Q9UDW8 O94919 P36222 Q59FS1 P07858 P08185 Q59G97 Q12860 Q4R500 P78324 A2RTY6 A2A418 A4D0S3 O15394 P13521 P19827 A2AAQ4 Q14118 Q9NQ79 O94985 Q6U2E9 Q59FP8 P01876
52972 26704 27015 54561 27355 27643 34609 20911 20918 72131 50657 138987 59540 37370 38973 32078 24383 98929 69024 78199 80147 90510 54604 65427 136612 69713 67927 35114 47092 259061 23643 23689 86888 63132 38983 65724 53117 93302 93457 67217 54981 42586 103293 44523 45095 60665 113249 51992 54932 106396 85644 130964 92988 70897 101338 83214 97519 71375 109723 192629 159859 571719
8 4 4 8 4 4 5 3 3 10 7 19 8 5 5 4 3 12 8 9 9 10 6 7 14 7 6 3 4 22 2 2 7 5 3 5 4 7 7 5 4 3 7 3 3 4 7 3 3 5 4 6 4 3 4 3 3 2 2 3 2 4
0.14 0.14 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.12 0.12 0.12 0.12 0.11 0.11 0.11 0.11 0.10 0.10 0.10 0.10 0.10 0.09 0.09 0.08 0.08 0.08 0.08 0.08 0.08 0.07 0.07 0.07 0.07 0.07 0.07 0.07 0.07 0.07 0.06 0.06 0.06 0.06 0.06 0.06 0.05 0.05 0.04 0.04 0.04 0.04 0.04 0.04 0.03 0.03 0.03 0.02 0.01 0.01 0.01
a
Relative abundance was calculated by normalizing the total spectrum count (TSC) of a protein by its molecular weight (Mr).
proteins were digested overnight at 37 °C in the dark with constant agitation. Tryptic peptides were then isolated via solidphase extraction with a C-18 Sep-Pak Vac cartridge (Waters Corporation) according to manufacturer’s instructions. Peptides were eluted with 70% acetonitrile/0.1% formic acid and were dried using vacuum centrifugation. 3742
Journal of Proteome Research • Vol. 6, No. 9, 2007
2.6. Strong Cation Exchange (SCX) Chromatography. SCX chromatography was performed offline using a PolySULFOETHYL A column, 200 mm × 2.1 mm (PolyLC, Inc.). The column was equilibrated with buffer A (10 mM KH2PO4 and 25% acetonitrile, pH < 3.0) for 30 min, after which the CSF peptides were solubulized in 2.0 mL of buffer A and loaded
research articles
Evaluating High-Abundance Protein Depletion Methods for CSF Table 2. Proteins Identified in the MARS Depleted CSF Sample protein ID
UniProt ID
Mr
TSC
% abundance
Cystatin C Prostaglandin D2-synthase Transthyretin Apolipoprotein A-I preproprotein Serpin peptidase inhibitor, clade A, member 3 Hemopexin Apolipoprotein E Orosomucoid 2 Apolipoprotein A-II preproprotein Vitamin D-binding protein Retinol-binding protein 4, plasma Complement component 3 Complement component 4A preproprotein Angiotensinogen preproprotein Serine (or cysteine) proteinase inhibitor, clade C Clusterin isoform 1 Gelsolin isoform b Kallikrein 6 isoform B Alpha 1B-glycoprotein Alpha-2-macroglobulin Pancreatic ribonuclease Complement component 1 inhibitor Albumin Apolipoprotein D Ceruloplasmin (ferroxidase) Superoxide dismutase 3, extracellular Family with sequence similarity 3, member C C-type lectin domain family 3, member B Fibrinogen, beta chain preproprotein Chitinase 3-like 1 Complement factor H isoform B Lumican Epididymal secretory protein E1 Osteoglycin preproprotein Limbic system-associated membrane protein Alpha-1-antitrypsin Secretogranin III Kininogen 1 Prostatic binding protein Apolipoprotein A-IV Fibrinogen, gamma chain isoform gamma-B Galectin 3 binding protein PREDICTED: similar to Ceruloplasmin (Ferroxidase) Insulin-like growth factor binding protein 7 Complement factor B preproprotein Histidine-rich glycoprotein CD14 antigen Complement component 1, s subcomponent CutA divalent cation tolerance homologue isoform 1 Amyloid like protein 1 isoform 2 Complement component 7 Proprotein convertase subtilisin/kexin type 1 inhibitor Complement component 1, r subcomponent Chromogranin B Autotaxin isoform 2 preproprotein Complement component 1, q subcomponent, gamma pol Secreted protein, acidic, cysteine-rich (osteonectin) Alpha-1-microglobulin/bikunin Afamin Prion protein preproprotein CD59 antigen p18-20 [H. sapiens] Alpha-2-HS-glycoprotein SPARC-like 1 Secreted phosphoprotein 1 isoform A Insulin-like growth factor binding protein 6 Inter-alpha (globulin) inhibitor H4 Neural cell adhesion molecule 1 isoform 1 Plasminogen Complement factor H isoform A Neurotrimin Vitronectin Lysozyme Cathepsin D preproprotein
P01034 Q5SQ09 P02766 P02647 P36955 P02652 P02649 Q5T538 P02647 P02774 P02753 Q6LDJ0 Q5JNX2 P01019 P36955 Q96AJ1 137150 Q92876 P04217 P01023 P07998 060860 P02768 P05090 P00450 O14618 Q5HY75 P05452 P02675 P36222 P08603 P51884 P61916 P20774 Q13449 P01009 Q8WXD2 P01042 P30086 P06727 068656 P17931 877964 Q16270 P00751 P04196 P08571 Q53HU9 O60888 P51693 Q8TCS7 P29120 Q53HT9 P05060 1035181 Q5T960 Q4R5R0 Q5TBD7 P43652 P04156 107271 P02765 Q14515 000571 P24592 Q59FS1 P13592 P06733 P08603 Q9P121 P04004 Q8N1E2 P07858
15789 21015 15877 30758 47620 51643 36131 23587 11167 52883 22995 187045 192664 53120 52568 57795 80590 15045 54219 163188 17632 55119 69321 21261 122127 25864 24664 22552 55892 42586 50974 38404 16559 33900 37370 46707 52972 47852 21043 45344 51478 65289 32078 29111 85478 59540 40050 76634 20911 72131 93457 27355 80147 78199 98929 25757 34609 38973 69024 27643 14167 39299 75201 35401 25306 103293 93302 90510 138987 37947 54271 16526 44523
156 165 108 106 143 141 91 57 26 96 34 275 170 46 45 49 58 10 36 105 11 32 40 11 58 12 10 9 22 16 19 14 6 12 13 16 18 16 7 15 17 21 10 9 26 18 12 22 6 20 25 7 20 19 24 6 8 9 15 6 3 8 15 7 5 20 18 17 26 7 10 3 8
13.01 10.34 8.96 4.54 3.95 3.60 3.32 3.18 3.07 2.39 1.95 1.94 1.16 1.14 1.13 1.12 0.95 0.88 0.87 0.85 0.82 0.76 0.76 0.68 0.63 0.61 0.53 0.53 0.52 0.49 0.49 0.48 0.48 0.47 0.46 0.45 0.45 0.44 0.44 0.44 0.43 0.42 0.41 0.41 0.40 0.40 0.39 0.38 0.38 0.37 0.35 0.34 0.33 0.32 0.32 0.31 0.30 0.30 0.29 0.29 0.28 0.27 0.26 0.26 0.26 0.25 0.25 0.25 0.25 0.24 0.24 0.24 0.24
Journal of Proteome Research • Vol. 6, No. 9, 2007 3743
research articles
Shores and Knapp
Table 2 (Continued) protein ID
UniProt ID
Mr
TSC
% abundance
Paraoxonase 1 Cell adhesion molecule with homology to L1CAM Insulin-like growth factor binding protein 2, 36kDa EGF-containing fibulin-like extracellular matrix protein 1 Neuronal cell adhesion molecule isoform B Complement component 9 ATPase, H+ transporting, lysosomal accessory protein 1 Hypothetical protein LOC651928 I factor (complement) Serine (or cysteine) proteinase inhibitor, clade A, member 7 Complement component 1, q subcomponent, B chain Contactin 1 isoform 1 Carnosinase 1 Chromogranin A Inter-alpha (globulin) inhibitor H1 Calsyntenin 1 isoform 1 Complement component 8, beta polypeptide preproprotein Complement component 2 Neuronal growth regulator 1 Fibrinogen, alpha polypeptide isoform alpha preproprotein Decorin isoform C Superoxide dismutase 1, soluble Tissue inhibitor of metalloproteinase 2 Heparin cofactor II Fibulin 1 isoform B Prosaposin VGF nerve growth factor inducible PREDICTED: similar to Triosephosphate isomerase (TIM) PREDICTED: similar to Probable endonuclease KIAA0830 isoform 3 Neural cell adhesion molecule 2 UDP-GlcNAc:betaGal beta-1,3-N-acetylglucosaminyltransferase 1 Contactin 2 Amyloid beta A4 protein, isoform A Neuronal pentraxin receptor isoform 2 WFIKKN2 protein Immunoglobulin superfamily, member 4C Phospholipid transfer protein isoform A Transferrin Aspartate aminotransferase 1 Plasma glutathione peroxidase 3 Secretogranin II Immunoglobulin superfamily, member 4D Extracellular matrix protein 1 isoform 1 Sparc/osteonectin, cwcv and kazal-like domains proteoglycan 1 Calcium channel, voltage-dependent, alpha 2/delta C-reactive protein, pentraxin-related Complement factor H-related 1 Leucine-rich alpha-2-glycoprotein 1 Nel-like 2 Insulin-like growth factor binding protein, acid labile subunit Collagen, type VI, alpha 1 Fibronectin 1 isoform 3 preproprotein Coagulation factor II Cartilage acidic protein 1 Ribonuclease T2 Fibulin 1 isoform C Intercellular adhesion molecule 2 Proenkephalin Hypothetical protein LOC146556 Activated leukocyte cell adhesion molecule Complement component 4 binding protein, alpha chain Coagulation factor XII Chondroitin sulfate proteoglycan 3 (neurocan) Alpha-2-plasmin inhibitor Lactate dehydrogenase B Multiple inositol polyphosphate histidine phosphatase Complement factor H-related 3 PREDICTED: similar to Pyruvate kinase, isozymes M1/M2 Cadherin 2, type 1 preproprotein Inter-alpha globulin inhibitor H2 polypeptide Amyloid beta (A4) like protein 2 Peptidoglycan recognition protein L
P27169 Q59FY0 P18065 Q12805 A4D0S3 Q9UGI4 Q4R500 946295 Q4R955 Q86U78 Q5T960 Q12860 Q96KN2 P10645 Q59FS1 O94985 A1L4K7 A2AAQ4 Q7Z3B1 P02671 P07585 O14618 P16035 P05546 Q59G97 O75905 Q9UDW8 P60174 O94919 O15394 O43505 Q02246 P05067 O95502 Q6UXZ9 Q8NFZ8 Q53H91 P02787 P17174 P22352 P13521 Q8IZP8 Q59G97 Q9BQ16 Q17R45 P02741 Q5TMF4 P02750 Q99435 Q8TAY0 Q8N4Z1 P02751 Q53H04 Q9NQ79 O00584 Q8N9G0 P13598 P01213 Q8N213 Q13740 Q5VVQ8 P00748 Q70JA7 P08697 P07195 Q4R3U9 Q02985 652797 Q9NYQ6 A2RTY6 Q4R413 Q96PD5
39706 136612 35114 54604 130964 63132 51992 26229 65724 46282 26704 113249 56656 50657 101338 109723 22205 83214 38694 69713 23262 15925 24383 57034 65427 58073 67217 27427 54981 92988 47088 113322 86888 52685 63898 42758 54704 77000 46218 35385 70897 48506 60665 49092 123105 25022 37636 38154 91284 65993 108462 259061 69992 71375 29461 74412 30633 30767 48734 65061 66989 67774 142883 54561 36615 55016 37298 39439 99747 106396 86900 67927
7 24 6 9 21 10 8 4 10 7 4 16 8 7 14 15 3 11 5 9 3 2 3 7 8 7 8 3 6 10 5 12 9 5 6 4 5 7 4 3 6 4 5 4 10 2 3 3 7 5 8 19 5 5 2 5 2 2 3 4 4 4 8 3 2 3 2 2 5 5 4 3
0.23 0.23 0.23 0.22 0.21 0.21 0.20 0.20 0.20 0.20 0.20 0.19 0.19 0.18 0.18 0.18 0.18 0.17 0.17 0.17 0.17 0.17 0.16 0.16 0.16 0.16 0.16 0.14 0.14 0.14 0.14 0.14 0.14 0.12 0.12 0.12 0.12 0.12 0.11 0.11 0.11 0.11 0.11 0.11 0.11 0.11 0.10 0.10 0.10 0.10 0.10 0.10 0.09 0.09 0.09 0.09 0.09 0.09 0.08 0.08 0.08 0.08 0.07 0.07 0.07 0.07 0.07 0.07 0.07 0.06 0.06 0.06
3744
Journal of Proteome Research • Vol. 6, No. 9, 2007
research articles
Evaluating High-Abundance Protein Depletion Methods for CSF Table 2 (Continued) protein ID
UniProt ID
Mr
TSC
% abundance
Melanoma cell adhesion molecule Matrix metalloproteinase 2 preproprotein Complement component 6 Nucleobindin 1 Colony stimulating factor 1 receptor CD163 antigen isoform B Biotinidase Fc fragment of IgG binding protein Immunoglobulin superfamily, member 8 Brevican isoform 1 Kelch-like 10 Protein tyrosine phosphatase, receptor type, G EGF-like-domain, multiple 4 Peptidylglycine alpha-amidating monooxygenase isoform B Neurexin 1 isoform alpha Transmembrane protein 132A isoform A Complement component 5 Alpha 1 type I collagen preproprotein Alpha 1 type XVIII collagen isoform 3 Reelin isoform B CD109 Alpha 3 type VI collagen isoform 2 Coagulation factor V Complement component 4B preproprotein Agrin Heparan sulfate proteoglycan 2
P43121 P08253 P13671 Q02818 P07333 Q86VB7 P43251 P01876 Q969P0 Q59F90 Q6JEL2 P23470 Q4R8Q9 620121 P78357 Q24JP5 Q27I61 Q6LAN8 Q2UY07 P78509 Q6YHK3 Q8N4Z1 P12259 Q6U2E9 O00468 Q5SZI5
71562 73834 104775 53846 107915 121516 61093 571719 64993 99056 69474 161956 254402 100754 161779 110128 188185 138826 149840 387950 161616 325090 251543 192629 214704 468528
3 3 4 2 4 4 2 18 2 3 2 4 6 2 3 2 3 2 2 5 2 4 3 2 2 3
0.06 0.05 0.05 0.05 0.05 0.04 0.04 0.04 0.04 0.04 0.04 0.03 0.03 0.03 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.01 0.01 0.01
onto the column. The peptides were eluted with a linear gradient of 0-50% buffer B (10 mM KH2PO4, 1.0 M KCl, and 25% acetonitrile, pH < 3.0) over 60 min. A UV absorption cell (λ ) 214 nm) was used to monitor the elution of CSF peptides. A total of 8 peptide containing fractions were collected in 5 min intervals. Collected fractions were then dried via vacuum centrifugation. 2.7. Reversed-Phase LC/MS/MS. SCX fractions were solubilized in 400 µL of 2% acetonitrile/0.1% formic acid (solvent A), and 20 µL of each fraction was loaded onto a 0.3 mm × 5.0 mm C18 trap column (LC Packings, Dionex) and washed for 30 min. An analytical RP-C18 column, 15 cm × 75 µm (Microtech Scientific) was then incorporated into the flow path, and peptides were separated with a 100 min linear gradient of 0-50% solvent B (95% acetonitrile/0.1% formic acid) followed by a 20 min linear gradient (50-70% solvent B) at a flow rate of 180 nL/min (Dionex ULTIMATE). Eluted peptides were infused directly into a LTQ mass spectrometer (Thermo Electron) with a nanospray source. The mass spectrometer was programmed to perform MS/MS analysis on the 5 most intense ions observed in the MS spectrum collected over m/z 4002000. Nanospray voltage of 2.0 kV, a normalized collision energy of 35%, a default charge state of +5, and an isolation mass window of 2.5 amu were used. Dynamic exclusion was enabled for all experiments with a duration of 3.0 min, a repeat count of 2, a repeat duration of 0.5 min, and a rejection mass window of 2.0 amu. 2.8. Data Analysis. MS/MS spectra were searched against an indexed Homo sapiens database, which was extracted from the NCBI nonredundant H. sapiens database, using the Turbo SEQUEST algorithm,13 a component of the Bioworks 3.2 software suite (Thermo Electron). Peptides with up to two possible missed cleavages were specified as a search parameter. Dynamic chemical modifications of +16 and +57 mass units corresponding to M-oxidized and C-carboxyamidomethyl modifications, respectively, were included as search parameters. A precursor ion accuracy of 2.0 amu was used. Resulting protein
identifications were filtered using two protein and two peptide filters, protein probability P < 0.001, minimum of two unique peptides for a protein identification, peptide probability P < 0.001, and Xcorr (cross correlation) versus charge state of 1.5, 2.0, and 2.5 for +1, +2, and +3 ions, respectively. Using Bioworks 3.2, multi-consensus reports were generated for each set of eight reverse-phase experiments (i.e., LC/MS runs for each of the eight first-dimension separation fractions) for each sample. All protein identifications resulting from only two unique peptides were further examined. Each peptide sequence was searched using the BLAST algorithm, and when both of the two peptide sequences were found in more than two proteins in the NCBI nonredundant H. sapiens database (e.g., multiple protein variants), the protein identification was deemed inconclusive and eliminated from the identification list.
3. Results and Discussion In the development of a method for high-throughput proteomic analysis of human CSF, the performance of three highabundance protein depletion strategies were compared using samples taken from the same pool of CSF, including the Multiple Affinity Removal System spin cartridge (MARS, Agilent Technologies), the ProteoSeek spin cartridge (Pierce Biotechnology), and depletion with an Amicon Ultra 50 kDa molecular weight cutoff filter. A fourth undepleted sample was also analyzed, and the results were used to compare the depletion methods in terms of breadth (number of identified proteins) and depth (concentrations of identified proteins) of analysis of the CSF proteome. 3.1. LC/MS/MS Analysis of Immunodepleted CSF Proteins. Tables 1-4 list the proteins identified in each of the four samples in order of their relative abundance, which was calculated using the total spectrum counting method.14 The total number of peptide MS/MS spectra matched to a particular protein were tabulated for each protein identification (total spectrum count, TSC). Relative protein abundance for each sample was calculated by normalizing the TSC of each protein Journal of Proteome Research • Vol. 6, No. 9, 2007 3745
research articles
Shores and Knapp
Table 3. Proteins Identified in the Proteoseek Depleted CSF Sample protein ID
Uniprot ID
Mr
TSC
% abundance
Cystatin C Alpha-1-antitrypsin Prostaglandin D2-synthase Transferrin Transthyretin Hemopexin Orosomucoid 1 Apolipoprotein A-I preproprotein Vitamin D-binding protein Beta-2-microglobulin Apolipoprotein E Apolipoprotein A-II preproprotein Delta globin Alpha 2 globin Complement component 3 Orosomucoid 2 Albumin Complement component 1 inhibitor Serine (or cysteine) proteinase inhibitor, clade F Beta globin Hypothetical protein LOC651928 Retinol-binding protein 4, plasma Angiotensinogen preproprotein Haptoglobin Complement component 4A preproprotein Kallikrein 6 isoform B Dickkopf homologue 3 Apolipoprotein H Serine (or cysteine) proteinase inhibitor, clade C Ceruloplasmin (ferroxidase) Clusterin isoform 1 Epididymal secretory protein E1 Pancreatic ribonuclease Apolipoprotein A-IV Apolipoprotein D Proprotein convertase subtilisin/kexin type 1 inhibitor Haptoglobin-related protein Kallikrein 6 isoform A preproprotein Gelsolin isoform B Thy-1 cell surface antigen Insulin-like growth factor binding protein 7 CD14 antigen Hypothetical protein LOC649897 Fibrinogen, beta chain preproprotein C-type lectin domain family 3, member B Kininogen 1 PREDICTED: similar to Ceruloplasmin (Ferroxidase) PREDICTED: similar to Ig alpha-1 chain C region isoform 1 Osteoglycin preproprotein Alpha-2-macroglobulin Secretogranin III Limbic system-associated membrane protein Histidine-rich glycoprotein Lysozyme Complement factor B preproprotein Fibrinogen, gamma chain isoform gamma-B Phospholipid transfer protein isoform A Chitinase 3-like 1 Family with sequence similarity 3, member C Superoxide dismutase 3, extracellular Anti-RhD monoclonal T125 gamma1 heavy chain Alpha-2-HS-glycoprotein Complement component 1, q subcomponent, B chain UDP-GlcNAc:betaGal beta-1,3-N-acetylglucosaminyltransferase 1 Prion protein preproprotein Fibrinogen, alpha polypeptide isoform alpha preproprotein Insulin-like growth factor binding protein 2, 36kDa CD59 antigen p18-20 [H. sapiens] Chromogranin A Galectin 3 binding protein Complement component 1, r subcomponent SPARC-like 1 Leucine-rich alpha-2-glycoprotein 1
P01034 P01009 Q5SQ09 P02787 P02766 P02790 P02763 P02647 P02774 Q9UM88 P02649 P02652 P02042 Q86YQ5 Q6LDJ0 Q5T538 P02768 060860 Q4R6H4 P68871 946295 P02753 P01019 P00738 Q5JNX2 Q92876 Q4R417 P02749 P32261 P00450 Q96AJ1 P61916 P07998 P06727 P05090 P29120 P00739 Q6T774 137150 Q59GA0 Q16270 P08571 Q49AS2 P02675 P05452 P01042 877964 Q8NCL6 P20774 P01023 Q8WXD2 Q13449 P04196 Q8N1E2 P00751 068656 Q53H91 P36222 Q5HY75 O14618 Q5EFE5 P02765 Q5T960 O43505 P04156 P02671 P18065 107271 P10645 P17931 Q53HT9 Q14515 P02750
15789 46707 21015 77000 15877 51643 23496 30758 52883 13705 36131 11167 16045 15247 187045 23587 69321 55119 46283 15988 26229 22995 53120 45176 192664 15045 38365 38286 52568 122127 57795 16559 17632 45344 21261 27355 39004 26838 80590 17923 29111 40050 22059 55892 22552 47852 32078 28490 33900 163188 52972 37370 59540 16526 85478 51478 54704 42586 24664 25864 52253 39299 26704 47088 27643 69713 35114 14167 50657 65289 80147 75201 38154
90 213 94 255 52 87 39 51 64 16 33 9 12 11 128 16 47 36 29 10 16 14 32 24 90 7 17 16 21 45 21 6 6 15 7 9 12 8 23 5 8 11 6 15 6 11 7 6 7 33 10 7 11 3 15 9 9 7 4 4 8 6 4 7 4 10 5 2 7 9 11 10 5
10.24 8.19 8.04 5.95 5.88 3.03 2.98 2.98 2.17 2.10 1.64 1.45 1.34 1.30 1.23 1.22 1.22 1.17 1.13 1.12 1.10 1.09 1.08 0.95 0.84 0.84 0.80 0.75 0.72 0.66 0.65 0.65 0.61 0.59 0.59 0.59 0.55 0.54 0.51 0.50 0.49 0.49 0.49 0.48 0.48 0.41 0.39 0.38 0.37 0.36 0.34 0.34 0.33 0.33 0.32 0.31 0.30 0.30 0.29 0.28 0.28 0.27 0.27 0.27 0.26 0.26 0.26 0.25 0.25 0.25 0.25 0.24 0.24
3746
Journal of Proteome Research • Vol. 6, No. 9, 2007
research articles
Evaluating High-Abundance Protein Depletion Methods for CSF Table 3 (Continued) protein ID
Uniprot ID
Mr
TSC
% abundance
Lumican Alpha-2-plasmin inhibitor Amyloid beta A4 protein, isoform A Contactin 1 isoform 1 Complement component 1, s subcomponent Cell adhesion molecule with homology to L1CAM Autotaxin isoform 2 preproprotein Complement component 9 Vitronectin Complement component 7 Neuronal pentraxin I Neurotrimin VGF nerve growth factor inducible Opioid binding protein/ cell adhesion molecule-like isoform A preproprotein Chromogranin B Afamin Proenkephalin Amyloid - like protein 1 GM2 ganglioside activator CutA divalent cation tolerance homologue isoform 1 EGF-containing fibulin-like extracellular matrix protein 1 Cathepsin D preproprotein Inter-alpha (globulin) inhibitor H4 Secreted protein, acidic, cysteine-rich (osteonectin) Secreted phosphoprotein 1 isoform A Extracellular matrix protein 1 isoform 1 Fibronectin 1 isoform 3 preproprotein Transforming growth factor, beta-induced, 68kDa Complement factor H isoform A Phosphatidylethanolamine-binding protein 4 Complement factor H isoform B Complement component 1, q subcomponent, gamma pol Plasminogen Alpha-1-microglobulin/bikunin Complement component 1, q subcomponent, A chain I factor (complement) Carboxypeptidase E Neural cell adhesion molecule 1 isoform 1 Myelin oligodendrocyte glycoprotein isoform alpha Heparin cofactor II Carbonic anhydrase I Ribonuclease T2 Corticosteroid binding globulin Calsyntenin 1 isoform 1 Fibulin 1 isoform B Neuronal cell adhesion molecule isoform B Inter-alpha (globulin) inhibitor H1 Peptidoglycan recognition protein L Neuronal pentraxin receptor isoform 2 Cartilage acidic protein 1 Melanoma cell adhesion molecule Fibulin 1 isoform D Nov precursor Paraoxonase 1 Biotinidase Inter-alpha globulin inhibitor H2 polypeptide WFIKKN2 protein Immunoglobulin superfamily, member 4C Activated leukocyte cell adhesion molecule Collagen, type VI, alpha 1 Coagulation factor XII Aspartate aminotransferase 1 Secretogranin II Complement component 6 Nucleobindin 1 Quiescin Q6 isoform A Contactin 2 Prosaposin Nel-like 2 Calcium channel, voltage-dependent, alpha 2/delta Neural cell adhesion molecule 2 Complement component 5
P51884 P08697 P05067 Q12860 Q53HU9 Q59FY0 1035181 Q9UGI4 P04004 Q8TCS7 Q15818 Q9P121 Q9UDW8 Q14982
38404 54561 86888 113249 76634 136612 98929 63132 54271 93457 47092 37947 67217 38983
5 7 11 14 9 16 11 7 6 10 5 4 7 4
0.23 0.23 0.23 0.22 0.21 0.21 0.20 0.20 0.20 0.19 0.19 0.19 0.19 0.18
P05060 P43652 P01213 P51693 P17900 O60888 Q12805 P07858 Q59FS1 Q4R5R0 000571 Q59G97 P02751 Q15582 P08603 Q8TB40 P08603 Q5T960 P06733 Q5TBD7 Q5T960 Q4R955 P16870 P13592 Q16653 P05546 P00915 O00584 P08185 O94985 Q59G97 A4D0S3 Q59FS1 P19827 O95502 Q9NQ79 P43121 Q96K89 P28686 P27169 P43251 A2RTY6 Q6UXZ9 Q8NFZ8 Q13740 Q8N4Z1 P00748 P17174 P13521 P13671 Q02818 O00391 Q02246 O75905 Q99435 Q17R45 O15394 Q27I61
78199 69024 30767 72131 20824 20911 54604 44523 103293 34609 35401 60665 259061 74634 138987 25414 50974 25757 90510 38973 26000 65724 53117 93302 28174 57034 28852 29461 45095 109723 65427 130964 101338 67927 52685 71375 71562 77190 39135 39706 61093 106396 63898 42758 65061 108462 67774 46218 70897 104775 53846 82525 113322 58073 91284 123105 92988 188185
8 7 3 7 2 2 5 4 9 3 3 5 21 6 11 2 4 2 7 3 2 5 4 7 2 4 2 2 3 7 4 8 6 4 3 4 4 4 2 2 3 5 3 2 3 5 3 2 3 4 2 3 4 2 3 4 3 6
0.18 0.18 0.18 0.17 0.17 0.17 0.16 0.16 0.16 0.16 0.15 0.15 0.15 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.14 0.13 0.13 0.13 0.12 0.12 0.12 0.11 0.11 0.11 0.11 0.11 0.10 0.10 0.10 0.09 0.09 0.09 0.09 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.07 0.07 0.07 0.06 0.06 0.06 0.06 0.06 0.06
Journal of Proteome Research • Vol. 6, No. 9, 2007 3747
research articles
Shores and Knapp
Table 3 (Continued) protein ID
Uniprot ID
Mr
TSC
% abundance
PREDICTED: similar to Ig gamma-1 chain C region Brevican isoform 1 Slit-like 2 Matrix metalloproteinase 2 preproprotein Complement component 2 Gelsolin isoform A Chondroitin sulfate proteoglycan 3 (neurocan) Cadherin 2, type 1 preproprotein Peptidylglycine alpha-amidating monooxygenase isoform B Neogenin homolog 1 Ephrin receptor EphA4 Fc fragment of IgG binding protein Complement component 4B preproprotein Nidogen 2 Coagulation factor V EGF-like-domain, multiple 4 Alpha 3 type VI collagen isoform 2 Heparan sulfate proteoglycan 2
652050 Q59F90 Q6EMK4 P08253 A2AAQ4 A2A418 Q70JA7 Q9NYQ6 620121 Q59FP8 P54764 P01876 Q6U2E9 Q14112 P12259 Q4R8Q9 Q8N4Z1 Q5SZI5
64280 99056 71667 73834 83214 85644 142883 99747 100754 159859 109789 571719 192629 151199 251543 254402 325090 468528
2 3 2 2 2 2 3 2 2 3 2 10 3 2 2 2 2 2
0.06 0.05 0.05 0.05 0.04 0.04 0.04 0.04 0.04 0.03 0.03 0.03 0.03 0.02 0.01 0.01 0.01 0.01
by its molecular weight (Mr). Table 1 lists the proteins identified in the undepleted sample, which provides a benchmark for determining the effectiveness of the protein depletion methods. For example, Table 2 indicates a substantial reduction and, in several cases, complete absence of the proteins targeted by the MARS spin cartridge. The ProteoSeek spin cartridge was also effective in reducing the albumin and immunoglobulins concentrations (Table 3). Figure 1 shows the number of unique proteins identified after using the three depletion strategies. Although effective in removing more high-abundance proteins, the total number of proteins identified using the MARS spin cartridge (171) was only slightly greater than that identified using the ProteoSeek spin cartridge (163). Of these, 129 proteins were identified in each sample. We therefore conclude that neither of the immunodepletion experiment was superior in terms of total number of protein identifications. To maximize the number of observed proteins, we used a 100 min, 0-50% solvent B gradient in the LC/MS/MS analysis. The length of the gradient elution in conjunction with the speed of the Finnigan LTQ mass spectrometer would be expected to enable acquisition of MS/MS spectra of lower-abundance peptides when the number of high-abundance peptides that are present is minimized. This is the case with the MARSdepleted sample, since two of the four most abundant proteins identified in the ProteoSeek experiment include antitrypsin and transferrin and are nearly eliminated in the MARS-depleted sample. The results of the two immunodepletion experiments suggest, however, that the depletion of a few high-abundance
Figure 1. Comparison of proteins identified following depletion. 3748
Journal of Proteome Research • Vol. 6, No. 9, 2007
proteins is not sufficient to shift the overall dynamic range of protein detection and observe lower-abundance species. During the MS analysis, peptide ions from other high-abundance proteins (e.g., cystatin C, prostaglandin D2 synthase, and transthyretin) that are not removed by either immunodepletion method still dominate the ion current. This effectively hinders the instrument’s ability to detect and select low-abundance peptides in a given MS spectrum for analysis. This problem is only partially alleviated by the use of dynamic exclusion, which permits establishing a limit of two successive MS/MS analyses of a given ion for 180 s. We have also performed a separate LC/MS/MS analysis of the ProteoSeek-depleted sample using MS/MS exclusion lists for the initially observed high-abundance peptides. Individual exclusion lists were assembled for the analysis of each of the eight SCX fractions that correspond to high-abundance peptides observed in the initial analysis. The exclusion lists contained m/z values for +1, +2, and +3 charge states for each detected peptide. The combined results of the analysis of all eight fractions yielded significantly less MS/MS spectra of high-abundance proteins (data not shown); however. only 130 unique proteins were identified (compared to 163 from the same sample without exclusion lists). These results further suggest that either high-abundance peptide ions dominate the ion current through a LC/MS/MS run, even when the instrument is programmed to ignore them for MS/MS analysis, or that the sheer number and mass range of exclusions also exclude other peptides of interest. 3.2. LC/MS/MS Analysis of CSF Proteins Depleted with a 50 kDa Molecular Weight Cutoff Filter. A 50 kDa molecular weight cutoff filter was also examined for depletion of highabundance proteins from CSF, since several of the highabundance proteins in CSF have molecular weights that are larger than 50 kDa, (e.g., albumin [66 kDa] and transferrin [77 kDa]). The porous membrane in molecular weight cutoff filters retains proteins that are larger than an average cross-sectional size that corresponds to that of 50 kDa proteins. Therefore, nondenatured CSF was processed which ensured that CSF proteins were filtered in their globular states. Table 4 lists the CSF proteins identified by LC/MS/MS following untrafiltration. Only 46 unique proteins were identified, which is significantly less than were identified using immunodepletion as well as without protein depletion (135 identified proteins). Furthermore, two of the four most abundant proteins seen in the
research articles
Evaluating High-Abundance Protein Depletion Methods for CSF
Table 4. Proteins Identified in CSF Sample Following Depletion Using a 50 kDa Molecular Weight Cutoff Filter protein ID
UniProt ID
Mr
TSC
% abundance
Albumin Prostaglandin D2-synthase Hypothetical protein LOC651928 Transferrin Transthyretin Hypothetical protein LOC649897 Anti-RhD monoclonal T125 gamma1 heavy chain Cystatin C Orosomucoid 1 Alpha-1-antitrypsin Pancreatic ribonuclease Beta-2-microglobulin Secreted phosphoprotein 1 isoform A Kallikrein 6 isoform B Epididymal secretory protein E1 Superoxide dismutase 1, soluble PREDICTED: similar to Ig gamma-3 chain C region PREDICTED: similar to Ig gamma-1 chain C region CD59 antigen p18-20 Hemopexin Serine (or cysteine) proteinase inhibitor, clade F Alpha-2-glycoprotein 1, zinc Chitinase 3-like 1 PREDICTED: similar to Ig gamma-4 chain C region Chromogranin A Vitamin D-binding protein PREDICTED: similar to Ig kappa chain V-III region HAH Apolipoprotein A-IV Family with sequence similarity 3, member C VGF nerve growth factor inducible Serpin peptidase inhibitor, clade A, member 3 Proprotein convertase subtilisin/kexin type 1 inhibitor Insulin-like growth factor binding protein 7 Cathepsin Z preproprotein Chromogranin B Prion protein preproprotein Leucine-rich alpha-2-glycoprotein 1 Alpha-1-microglobulin/bikunin Orosomucoid 2 Dickkopf homologue 3 CD14 antigen Gelsolin isoform B Complement factor B preproprotein Neuronal pentraxin I Alpha 1B-glycoprotein Histidine-rich glycoprotein
P02768 Q5SQ09 946295 P02787 P02766 Q49AS2 Q5EFE5 P01034 P02763 P01009 P07998 Q9UM88 000571 Q92876 P61916 O14618 947052 652050 107271 P02790 Q4R6H4 P02765 P36222 P01861 P10645 P02774 642113 P06727 Q5HY75 Q9UDW8 P36955 P29120 Q16270 Q9UBR2 P05060 P04156 P02750 Q5TBD7 Q5T538 Q4R417 P08571 137150 P00751 Q15818 P04217 P04196
69321 21015 26229 77000 15877 22059 52253 15789 23496 46707 17632 13705 35401 15045 16559 15925 18069 64280 14167 51643 46283 34237 42586 47414 50657 52883 16578 45344 24664 67217 47620 27355 29111 33846 78199 27643 38154 38973 23587 38365 40050 80590 85478 47092 54219 59540
759 134 93 196 39 50 80 23 24 43 16 12 25 9 9 8 9 28 6 19 13 9 10 11 11 11 3 8 4 10 7 4 4 4 9 3 4 4 2 3 3 4 4 2 2 2
25.79 15.02 8.35 6.00 5.79 5.34 3.61 3.43 2.41 2.17 2.14 2.06 1.66 1.41 1.28 1.18 1.17 1.03 1.00 0.87 0.66 0.62 0.55 0.55 0.51 0.49 0.43 0.42 0.38 0.35 0.35 0.34 0.32 0.28 0.27 0.26 0.25 0.24 0.20 0.18 0.18 0.12 0.11 0.10 0.09 0.08
ultrafiltered sample are albumin and transferrin, which were predicted to be removed. It is therefore probable that a portion of the intact proteins that are larger than 50 kDa as well as possibly fragments of these proteins (that would yield a protein identification due to matched peptides) were able to penetrate the filter membrane. Considering that significantly more proteins were identified in the undepleted sample, we also conclude that many low-abundance proteins of interest were apparently bound to larger proteins and were removed by ultrafiltration. In addition, several low-abundance proteins seen in the immunodepleted samples (Tables 2 and 3) are larger than 50 kDa and were likely removed as well. We conclude from this data that use of a 50 kDa molecular weight cutoff filter is not a useful method for the depletion of high-abundance proteins from CSF. 3.3. Dynamic Range Limitations. In an LC/MS/MS analysis of a complex proteomic sample, there is a limited detectable dynamic range of protein concentrations. It is usually desirable to optimize the analysis for observation of the lowest possible concentration range. Absent a way of extending dynamic range, the primary way in which the observed concentration range is
lowered is through the removal of high-abundance proteins. In the present study, we compared the effectiveness of three methods for abundant protein depletion, including two immunodepletion methods and an ultrafiltration method using a 50 kDa molecular weight cutoff filter. We present here a graphical method for visualizing the resulting effect upon the dynamic range of observable proteins by comparing the analysis results for proteins of known normal concentration ranges in CSF. Figure 2 displays the known concentration ranges of 45 proteins in CSF, measured in g/L (adapted from ref 1). The horizontal axis displays a logarithmic distribution of protein concentration spanning 10 orders of magnitude. The columns on the right axis in Figure 2 contains a tabulation of the total spectrum count (TSC) of the identified proteins in the undepleted control sample (C), the MARS-depleted sample (I), the ProteoSeek-depleted sample (II), and the 50 kDa ultrafiltered sample (III). All of the high concentration proteins (10-1 to 10-4 g/L) were observed, whereas proteins with concentrations between 10-4 and 10-5 g/L were only sparsely detected, as seen by relatively low spectrum counts for those proteins that were observed. Only two proteins with normal concentraJournal of Proteome Research • Vol. 6, No. 9, 2007 3749
research articles
Shores and Knapp
Figure 2. Comparison of known concentration ranges of 45 different proteins in CSF. Total spectrum counts associated with proteins identified without depletion (C), depleted with the MARS spin cartridge (I), depleted with the Proteoseek spin cartridge (II), and with a 50 kDa cutoff filter (III) are are listed on the right side of the figure.
tions lower than 10-5 g/L were observed: amyloid beta A4 protein (∼10-6 g/L) and secretogranin-2 (∼10-6 g/L), and it is possible that these were identified by observation of peptides from fragments of the proteins that might be present at higher concentrations than documented of the intact protein’s concentrations resulting from measurements by other methods.1 This is also likely considering that amyloid beta A4 protein and secretogranin-2 peptides were sequenced in both immunodepleted samples and the undepleted sample. Although cumulative repeat MS/MS analyses of each depleted sample would be expected to yield a greater number of unique protein identifications,15 it is unlikely that many additional proteins 3750
Journal of Proteome Research • Vol. 6, No. 9, 2007
with lower concentrations would be observed. We therefore conclude that the dynamic range of detection was not significantly shifted by use of these depletion methods. For detection of potential protein biomarkers, a significantly lower range of detection is desirable, since potential biomarkers may likely exist at lower concentrations in CSF.1 The results shown in Figure 2 indicate that little is gained from employing any of the three abundant protein depletion methods over analysis that does not include abundant protein depletion in terms of shifting the range of observed proteins to lower concentration. These results suggest that there is need for methodology designed for the specific depletion of the high-abundance
research articles
Evaluating High-Abundance Protein Depletion Methods for CSF
proteins in CSF, as depletion methods targeted to the abundant proteins in serum are not sufficient.
4. Conclusion In this work, we have compared the performance of three different high-abundance protein methods in the preparation of CSF for proteomic analysis. Following two-dimensional LC/ MS/MS of aliquots taken from the same CSF pool, 171 unique proteins were identified using the MARS spin cartridge, and 163 unique proteins were identified using the ProteoSeek spin cartridge, with 130 proteins identified in both samples. Ultrafiltration yielded less favorable results, as only 46 unique proteins were identified. Analysis without depletion resulted in the identification of 135 unique proteins. We have assessed the depth of analysis of the CSF proteome using a graphical methods of comparing identified proteins to the concentrations of 45 known proteins in CSF and concluded that use of serum immunodepletion products, while increasing the total number of peptides observed, do not significantly increase the depth of analysis of proteins observed in CSF. For optimal proteomic analysis of CSF, there is need for a depletion method specifically targeted toward the abundant proteins in CSF. Finally, the method of using marker proteins of known normal concentration should be generally useful for assessing the depth of analysis of proteomic analysis methods.
Acknowledgment. This work was supported in part by the NHLBI Proteomics Initiative via N01-HV-28181. Kevin S. Shores is supported by NIH/NHLBI 5 T32 HL007260-30 Training to Improve Cardiovascular Drug Therapy. We thank Dr. J. Baatz for protein assays. References (1) Huhmer, A. F.; Biringer, R. G.; Amato, H.; Fonteh, A. N.; Harrington, M. G. Protein analysis in human cerebrospinal fluid: Physiological aspects, current progress and future challenges. Dis. Markers 2006, 22 (1-2), 3-26. (2) Maccarrone, G.; Milfay, D.; Birg, I.; Rosenhagen, M.; Holsboer, F.; Grimm, R.; Bailey, J.; Zolotarjova, N.; Turck, C. W. Mining the human cerebrospinal fluid proteome by immunodepletion and shotgun mass spectrometry. Electrophoresis 2004, 25 (14), 240212.
(3) Zhang, J.; Goodlett, D. R.; Quinn, J. F.; Peskind, E.; Kaye, J. A.; Zhou, Y.; Pan, C.; Yi, E.; Eng, J.; Wang, Q.; Aebersold, R. H.; Montine, T. J., Quantitative proteomics of cerebrospinal fluid from patients with Alzheimer disease. J. Alzheimers Dis. 2005, 7 (2), 125-33; discussion 173-80. (4) Wenner, B. R.; Lovell, M. A.; Lynn, B. C. Proteomic analysis of human ventricular cerebrospinal fluid from neurologically normal, elderly subjects using two-dimensional LC-MS/MS. J. Proteome Res. 2004, 3 (1), 97-103. (5) Noben, J. P.; Dumont, D.; Kwasnikowska, N.; Verhaert, P.; Somers, V.; Hupperts, R.; Stinissen, P.; Robben, J. Lumbar cerebrospinal fluid proteome in multiple sclerosis: characterization by ultrafiltration, liquid chromatography, and mass spectrometry. J. Proteome Res. 2006, 5 (7), 1647-57. (6) Dumont, D.; Noben, J. P.; Raus, J.; Stinissen, P.; Robben, J. Proteomic analysis of cerebrospinal fluid from multiple sclerosis patients. Proteomics 2004, 4 (7), 2117-24. (7) Hammack, B. N.; Fung, K. Y.; Hunsucker, S. W.; Duncan, M. W.; Burgoon, M. P.; Owens, G. P.; Gilden, D. H. Proteomic analysis of multiple sclerosis cerebrospinal fluid. Mult. Scler. 2004, 10 (3), 245-60. (8) Burkhard, P. R.; Rodrigo, N.; May, D.; Sztajzel, R.; Sanchez, J. C.; Hochstrasser, D. F.; Schiffer, E.; Reverdin, A.; Lacroix, J. S. Assessing cerebrospinal fluid rhinorrhea: a two-dimensional electrophoresis approach. Electrophoresis 2001, 22 (9), 182633. (9) Yuan, X.; Desiderio, D. M. Proteomics analysis of prefractionated human lumbar cerebrospinal fluid. Proteomics 2005, 5 (2), 54150. (10) Gianazza, E.; Arnaud, P. Chromatography, of plasma proteins on immobilized Cibacron, Blue, F3-GA. Mechanism of the molecular interaction. Biochem. J. 1982, 203 (3), 637-41. (11) Ogata, Y. M.; Charlesworth, C.; Muddiman, D. C. Evaluation of protein depletion methods for the analysis of total-, phosphoand glycoproteins in lumbar cerebrospinal fluid. J. Proteome Res. 2005, 4 (3), 837-45. (12) Righetti, P. G.; Boschetti, E.; Lomas, L.; Citterio, A. Protein Equalizer Technology : the quest for a “democratic proteome”. Proteomics 2006, 6 (14), 3980-92. (13) Eng, J. K.; McCormack, A. L.; Yates, J. R., III. An approach to correlate tandem mass spectral data of peptides with amino acid sequences in a protein database. J. Am. Soc. Mass Spectrom. 1994, 5 (11), 976-89. (14) Liu, H.; Sadygov, R. G.; Yates, J. R., 3rd, A model for random sampling and estimation of relative protein abundance in shotgun proteomics. Anal. Chem. 2004, 76 (14), 4193-201. (15) Chong, P. K.; Gan, C. S.; Pham, T. K.; Wright, P. C. Isobaric tags for relative and absolute quantitation (iTRAQ) reproducibility: Implication of multiple injections. J. Proteome Res. 2006, 5 (5), 1232-40.
PR070293W
Journal of Proteome Research • Vol. 6, No. 9, 2007 3751