Reversed-Phase High-Performance Liquid Chromatographic

Reversed-Phase High-Performance Liquid Chromatographic Prefractionation of Immunodepleted Human Serum Proteins to Enhance Mass Spectrometry Identification of Lower-Abundant Proteins James Martosella, Nina Zolotarjova, Hongbin Liu, Gordon Nicol, and Barry E. Boyes* Agilent Technologies, Integrated Biology Solutions, 2850 Centreville Rd., Wilmington, Delaware 19808 Received April 4, 2005

Serum analysis represents an extreme challenge due to the dynamic range of the proteins of interest, and the high structural complexity of the constituent proteins. In serum, the quantities of proteins and peptides of interest range from those considered “high abundance”, present at 2-70% by mass of total protein, to those considered “low abundance”, present at 10-12 M or less. This range of analytical target molecules is outside the realm of available technologies for proteomic analysis. Therefore, in this study, we have developed a workflow toward addressing the complexity of these samples through the application of multidimensional separation techniques. The use of reversed-phase methods for the separation and fractionation of protein samples has been investigated, with the goal of developing an optimized serum separation for application to proteomic analysis. Samples of human serum were depleted of the six most abundant proteins, using an immunoaffinity LC method, then were separated under a variety of reversed-phase (RP) conditions using a macroporous silica C18 surface modified column material. To compare the qualities of the RP separations of this complex protein sample, absorbance chromatograms were compared, and fractions were collected for off-line SDS-PAGE and 2D-LC-MS/MS analysis. The column fractions were further investigated by determination of protein identities using either whole selected fractions, or gel bands excised from SDS-PAGE gels of the fractions. In either case samples underwent tryptic fragmentation and peptide analysis using MALDIMS or LC-MS/MS. The preferred conditions for RP protein separation exhibited reproducibly high resolution and high protein recoveries (>98%, as determined by protein assay). Using the preferred conditions also permitted high column mass load, with up to 500 µg of protein well tolerated using a 4.6 mm ID × 50 mm column, or up to 1.5 mg on a 9.4 mm ID × 50 mm column. Elevated column temperature (80 °C) was observed to be a critical operational parameter, with poorer results observed at lower temperatures. The combination of sample simplification by immunoaffinity depletion combined with a robust and high recovery RP-HPLC fractionation yields samples permitting higher quality protein identifications by coupled LC-MS methods. Keywords: proteomics • pre-fractionation • reversed-phase chromatography • immunodepletion • human serum • mass spectrometry • gel electrophoresis • HPLC

Introduction Interest in the proteomic analysis of human serum has been greatly elevated during the past several years as LC and MS methodologies have evolved sufficiently to investigate this challenging sample. The popularization of multidimensional LC methods, and the ever-improving sensitivity and performance of multi-stage MS instruments, combined with highspeed database searching, is permitting complex protein samples to undergo analysis by identification of constituent tryptic peptide fragments. Proteomic analysis of human serum represents an extreme challenge due to the dynamic range of * To whom correspondence should be addressed. Tel: (302) 993-5883. Fax: (302) 633-8908. E-mail: [email protected].

1522

Journal of Proteome Research 2005, 4, 1522-1537

Published on Web 08/04/2005

the proteins of interest. In serum, the quantities of proteins and peptides range from those considered “high abundance”, present at 2-70% by mass of total protein, to those considered “low abundance”, present at 10-12 or less. This range of analytical target molecules is currently outside the realm of the dynamic range of available technologies for proteomic analyses. A means to address the complexity of these samples is the application of multidimensional separation techniques,1-6 for example, by multidimensional LC fractionation of tryptic peptide fragments, using ion exchange LC, with either continuous or discontinuous sampling prior to reversed-phase LC separation. The well-known MudPIT (multi-dimensional protein identification technique) methodology is an example of this approach, using a continuous sampling of ion exchange 10.1021/pr050088l CCC: $30.25

 2005 American Chemical Society

research articles

Reversed-Phase Chromatography of Human Serum Proteins

Materials and Methods

Figure 1. Experimental Workflow.

resolved peptide fractions, combined with reversed-phase (RP) separation and on-line MS/MS detection.7,17 An alternative approach to complexity reduction is the use of whole protein fractionation, prior to fragmentation into constituent peptide fragments, also known as protein prefractionation methods. At present, two major approaches are being used as protein pre-fractionation techniques: chromatographic and electrophoretic.2,9-16 Chromatographic pre-fractionation methods in broad use have included; ion exchange, size exclusion, partition, hydrophobic interaction, dye-ligand and various affinity methods, as well as reversed-phase LC. The present study combines the use of immunoaffinity depletion with reversed-phase separation modes to reduce the sample complexity of human serum. We selectively immunodepleted six of the most abundant proteins from human serum, then employed gradient elution RP-HPLC to fractionate the remaining serum proteins. Specifically, we have developed a set of preferred RP separation conditions, used with a novel macroporous column material, for enabling higher protein recoveries than can be obtained by the use of conventional LC materials. In addition, sample preparation conditions have been modified to permit high volumetric and mass load tolerance. In the current study, reversed-phase gradient elution conditions and column operational temperature have been studied with specific attention to protein recovery, chromatographic resolution, and protein load tolerance. Protein fractions from a reversed-phase separation were collected and analyzed for determination of protein identity (ID). The workflow used is shown in the flowchart presented as Figure 1. Improved conditions for the RP separation permitted enhanced peak resolution and enabled high protein recoveries (>98%). In particular, use of a macroporous column material operated at elevated temperatures using a multi-segment gradient elution program enhanced selectivity, reduced band broadening, and improved protein recovery. The conditions presented herein permitted robust and reproducible separations and increased the confidence of human serum protein identifications.

Materials. HPLC-grade acetonitrile was purchased from Burdick & Jackson (VWR International). The water used was Milli-Q grade (Millipore, Bedford, MA). Trifluoroacetic acid was purchased from Sigma (St. Louis, MO). Urea, sequanal grade, was obtained from Pierce (Rockford, IL). The 15 mL and 50 mL conical BD polystyrene tubes were obtained from VWR International. Pre-cast gels were obtained from Invitrogen (Carlsbad, CA). The Multiple Affinity Removal System column and buffers were obtained from Agilent Technologies Inc (Wilmington, DE). The reversed-phase columns used in the study were a Zorbax 300 SB-C8 column (300 Å pore size, 5 µm particle size) in a 4.6 mm ID × 50 mm dimension and a prototype macroporous reversed-phase C18 column in 4.6 mm ID × 50 mm and 9.4 mm ID × 50 mm dimensions, all of which were produced by Agilent Technologies (Wilmington, DE). Serum samples were obtained by venupuncture of a healthy male volunteer, with collection in a Becton Dickinson Vacutainer tube (VWR, West Chester, PA) with SST gel and BD clot activator. After clot formation, the sample was centrifuged at 1000 × g for 15 min. The serum was removed, and aliquots stored at -80 °C. Total time for serum processing was less than 60 min. A protease inhibitor cocktail, Complete, (Roche Biochemicals, Indianapolis, IN) was added to serum to reduce proteolytic degradation. Immunoaffinity Depletion of High-Abundant Proteins from Human Serum. High-abundant protein removal from human serum was performed on a 4.6 mm ID × 100 mm immunodepletion column. This column specifically removes albumin, IgG, anti-trypsin, IgA, transferrin and haptoglobin in a single column run. Prior to injection on the column, serum samples were diluted five times with Buffer A - a saltcontaining neutral buffer (pH 7.4) used for loading, washing and reequilibrating. The sample was transferred to a 0.22 µm pore size spin filter for removal of particulates by centrifugation at 16 000 × g for 1 min at room temperature. The prepared samples were maintained at 4 °C in the temperature-controlled autosampler stage of the Agilent 1100 LC system. 180 µL (36 µL sera diluted 5× with Buffer A) of sample was injected onto the 4.6 mm ID × 100 mm column in 100% Buffer A at a flow rate of 0.5 mL/min for 10.0 min. After collection of the flowthrough fraction, the column was washed and the bound proteins eluted with 100% Buffer B (a low pH urea buffer) at a flow rate of 1.0 mL/min for 7.0 min. Afterward, the column was regenerated by equilibrating it with Buffer A (0% B) for 11.0 min for a total run cycle of 28.0 min. The flow-through fraction was collected into 1.5 mL microcentrifuge tubes, and cooled to 4 °C using a thermostat-controlled fraction collector. Processing of Depleted Serum and Bound Fractions for Electrophoretic Analysis. Flow-through fractions or bound fractions from several injections were pooled, buffer-exchanged and concentrated using 4 mL spin concentrators with 5 kDa molecular weight cutoff. The sample was centrifuged at 7500 × g for 20 min at 4 °C and the buffer exchanged into 20 mM Tris-HCl, pH 7.4, by 3 rounds of addition of the buffer, with centrifugation for 20 min each time. The concentrated samples were aliquoted and stored at -80 °C until analysis. Protein concentrations were established using a BCA protein assay kit (Pierce). HPLC Sample Preparation, Separation and Fraction Collection. HPLC separations were performed on an automated Agilent 1100 LC system with operation at specifically defined column temperatures and an autosampler equipped with a 900 Journal of Proteome Research • Vol. 4, No. 5, 2005 1523

research articles µL loop. Approximately 1.0 mL of unconcentrated flow-through samples were collected from a 4.6 mm ID × 50 mm immunodepletion column, denatured and directly loaded onto the reversed-phase column. Specifically, each 1.0 mL of immunodepleted flow-through containing approximately 0.38 mg protein was added to 480 mg of solid urea and acidified by addition of 13 µL of acetic acid (final concentrated ) 6 M urea/ 1.0% AcOH), then permitted to equilibrate at room temperature for at least 30 min before RP separation. The RP separations for each flow-through were performed under a set of preferred conditions using a multi-segment elution gradient, with eluent A (0.1% TFA in water, v/v) and eluent B (0.08% TFA in acetonitrile, v/v). The gradient conditions consisted of three steps with increasing concentrations of the eluent B: 3-30% B 6 min., 30-55% B 33 min, 55-100% B 10 min, hold 100% B 4 min at a flow rate of 0.75 mL/min for a total runtime of 54.0 min. For consecutive runs, a 5.0 min. postrun comprising 3.0% eluent B was added to re-equilibrate the column. Separations conducted on the 4.6 mm ID × 50 mm RP column used a single 900 µL injection of the denatured flowthrough proteins. Higher protein mass loading was accomplished by a series of multiple injections of 900 µL on a 9.4 mm ID × 50 mm column. Multiple injections were performed under isocratic conditions of 3% eluent B for the purpose of concentrating the protein onto the column; the final injection, in the series of such multiple injections, initiated the preferred condition gradient program. The chromatograms were monitored at 280 nm and 50 fractions were collected at 1.0 min. intervals from 3 to 53.0 min. Each fraction was dried in a centrifugal vacuum concentrator (Thermo-Savant, Millford, MA) and stored at -80 °C for subsequent 2D-LC-MS/MS and SDS-PAGE analysis. Electrophoretic Analysis. SDS-PAGE analysis was carried out using Invitrogen Tris-glycine precast gels (4-20% acrylamide, 10 wells, 1 mm) according to the manufacturer’s protocol. Dried protein samples from reversed-phase separations were dissolved in sample preparation buffer, heated, and then loaded onto the gel. Following electrophoresis, proteins were visualized by Coomassie Blue staining using Pierce GelCode Blue. Protein Recovery. To determine protein recovery, 200 µg immunodepleted human serum proteins denatured in 6 M urea/1.0% AcOH were injected onto the RP column operating under the preferred gradient elution conditions as described above. Column effluent was collected into 50 mL polystyrene conical tubes (VWR International). Blank runs were performed in the same manner, however, with the column removed from the flow path. Blanks and column eluates (approximately 38 mL) were dried in a speed vacuum concentrator at medium drying temperatures overnight. Dried samples of column and blank runs were solubilized with 0.5 mL of 3 M Urea, 1% Triton X-100 and 0.25% acetic acid (The final urea concentration, including the urea present from the injected sample, was 5.5 M). Samples were vortexed extensively to solubilize all protein and remove any material adhering to the tube walls. Protein quantitation was performed with EZQ Protein Quantitation kit from Invitrogen/Molecular Probes (similar results have been obtained using the BCA method). An ovalbumin standard was used for the creation of a standard curve, fitted using weighted least-squares. 1 µL of each sample was spotted onto assay paper; the samples were fixed, washed and stained with EZQ stain according to the manufacturer’s protocol. Each sample was then processed in 1524

Journal of Proteome Research • Vol. 4, No. 5, 2005

Martosella et al.

quadruplicate and fluorescence was measured by scanning on a Typhoon 8600 laser scanner (Amersham Biosciences, Piscataway, NJ) using a 532 nm laser for the excitation and the 580 nm emission filter. 2D-LC-MS/MS. The collected protein fractions were lyophilized and redissolved in 100 mM ammonium bicarbonate, 8 M urea, pH 8.5, digested with Lys-C, diluted to 2M in urea, then digested with trypsin. The digestion was quenched with the addition of 3 µL acetic acid.17 For 2D-LC-MS/MS analysis of depleted serum only (∼70 µg total protein), an initial desalting step was required before proceeding with the protocol stated above. Digestion solutions were loaded onto a serially connected RP-SCX column (Agilent Zorbax 300SB-C18, 5 µm, 0.3 mm ID × 35 mm and Agilent Zorbax BIO-SCX II, 3.5 µm, 0.3 mm ID × 35 mm) offline using HPLC (Agilent 1100, Palo Alto, CA). Using methods similar to those previously described,17,18 the two-phase column was connected directly to a capillary RP column (Agilent, Zorbax 300SB-C18, 3.5 µm, 0.1 mm ID × 150 mm) and was analyzed by 2D-LC-MS/MS using an nano-LC-MS system (Agilent 1100 Nano-LC and Agilent XCT ion trap). Peptides were first eluted to the SCX column from the first RP column with neat acetonitrile. A series of 10 steps of increasing ionic strength followed by RP elution were used at the following concentrations of ammonium acetate: 10, 25, 50, 100, 150, 200, 300, 400, 500, 1000 mM. Elution of peptide fragments was accomplished by a gradient elution reversed-phase separation, in which Buffer A is 3.0% ACN and 0.1% formic acid (FA) in water, and Buffer B is 90% ACN and 0.1% FA in water. The elution program was 5% B in 15 min, 5-15% B in 5 min, 15-30% B in 40 min, 30-60% B in 35 min, 60% B in 7 min, re-equilibration of 5% B in 12 min, for a total run time of 114 min. The XCT ion trap mass spectrometer was operated in standard scan mode for MS analysis and in ultra scan mode for MS/MS. The MS/MS data were analyzed with Spectrum Mill (Agilent, Palo Alto, CA) against the International Protein Index (IPI) human database.19 The following filter was used after database searching: peptide score > 8, peptide % SPI > 70 and protein score > 9. Only fully tryptic peptides were considered, with one missed cleavage allowed. All MS/MS derived protein identifications reported in Tables 2 and 3 were also manually examined for fragmentation patterns that support the peptide sequence IDs.

Results and Discussion Immunoaffinity Serum Depletion. Figure 2A is representative of a typical immunoaffinity chromatogram and shows removal of high-abundant proteins from a human serum sample. Ideally, a depletion column should possess little or no binding of nontargeted proteins, and preferentially should bind several high-abundant proteins simultaneously. Antibodybased selection strategies can offer the desired characteristics, since immunoaffinity separations are known to be capable of excellent specificity and acceptable sample capacity.20-25 Electrophoretic analysis of the immunoaffinity fractions indicates high specificity for the removal of the abundant proteins (Figure 2B, Lane 4) and enrichment of low-abundant proteins in the flow-through fraction (Figure 2B, Lane 3). By removing high-abundant proteins, we were able to reduce the total serum protein mass by 85%. From each column pass, we collected 0.38 mg of low-abundant protein in the flow-through fraction, depleting about 2.14 mg of high-abundant proteins from 36 µL of injected human serum.


Figure 2. Chromatogram for the affinity removal of highabundant proteins from human serum and SDS gel electrophoresis of human serum protein fractions from an immunoaffinity column. Panel (A) 36 µL of serum was diluted 5× and injected on a 4.6 mm ID × 100 mm immunoaffinity column (0.50 mL/min) and a flow-through peak (3-5.0 min) was collected for reversed-phase HPLC fractionation. The column was washed with Buffer A and the targeted high-abundant proteins were eluted with Buffer B. Panel (B) An equal amount (10 µg) of crude serum (Lane 2), flow-through (Lane 3) and bound fractions (Lane 4) were separated on 4-20% SDS-PAGE under nonreducing conditions. Lanes 1 and 5 are the molecular weight standards (Mark12) from Invitrogen. On the basis of the protein assay of the flow-through fraction, 85% of total protein was removed from the crude serum.

Flow-through fractions were processed for downstream separation by RP-HPLC (see Material & Methods). Samples were added directly to the acidic urea denaturant and required no other post-depletion processing prior to reversed-phase HPLC. In previous typical practice, immunodepleted serum samples required added steps to desalt and/or concentrate prior to electrophoretic and mass spectral analyses. Some common techniques and procedures rely on the use of membrane filters, dialysis, and spin concentrators. These methods increase the complexity of the workflow and the chance of protein precipitation and losses due to irreversible surface binding. Since HPLC desalting is a well-known and commonly practiced technique, a more practical approach is the introduction of the post-immunodepleted sample directly onto the HPLC column. Immunodepleted flow-through fractions were concentrated on the column using multiple injections with delayed initiation of the gradient elution program. This injection method permits reproducible separations for various column loading mass, with little sensitivity to total sample volumes. We compared a reversed-phase separation of 270 µg crude human serum versus 270 µg immunodepleted human serum. The crude serum separation shown in Figure 3A results in poorly resolved proteins, with a broad and misshapen albumin peak, eluting at about 22 min, and overlapping with many

research articles adjacent components. The chromatogram shown in Figure 3B is a reversed phase separation in which the serum sample has been depleted of six of the most abundant proteins - HSA, IgG, IgA, haptoglobin, transferrin and R-1-antitrypsin. When compared with the crude serum chromatogram in Figure 3A, Figure 3B displays a highly resolved separation of the immunodepleted serum with improved resolution for proteins otherwise obscured by high abundant proteins. Part of the problems associated with crude serum injections may be related to the presence of large quantities of human serum albumin and its fragments, which are believed to have unfavorable reversed-phase adsorption and elution characteristics. Reversed-Phase Column Performance. Conventional reversed-phase column materials and the operational conditions under which they are used to prefractionate, have produced encouraging results to further reduce sample complexity and permit identification of low abundant proteins.5,14,26-28 However, the use of RP chromatography for this application is not without complications. For example, as recently reviewed by Lescuyer et al.,29 RP chromatography can be applied to fractionate complex protein mixtures, but consideration should be given to the protein losses commonly associated with this method. To minimize protein loss, selection of appropriate column materials and operational conditions may be of use. We studied a novel reversed-phase microparticulate macroporous silica column packing material which had been suitably surface modified with an octadecylsilane functionality (mRP-C18). This material was employed to establish preferred conditions for the prefractionation of immunodepleted serum. Our particular interest was in maximizing protein recovery while selecting operational conditions needed to achieve a good quality separation of the serum proteins. Also of importance was the mRP-C18 column performance for delivering reproducible separations and the ability to separate at higher protein loads. To emphasize areas of specific column performance, we compared the results of the mRP-C18 with those obtained from a more traditional porous C8 surface functionalized reversedphase column material (RP-C8). Effect of Temperature on RP Separation. We investigated the effects of column temperature on the separation of an immunodepleted serum sample. Solvent viscosity, diffusivity and mobile phase polarity depend strongly on temperature, and the manipulation of column temperature is a crucial variable in the separation of hydrophobic peptides and proteins.30-32 Elevated temperatures have been previously shown to improve analysis of peptides and protein.33-37 However, elevated temperatures also run the risk of causing stationary phase degradation, requiring the use of hydrolytically stable bonded phase silica column packing materials. The surface of the mRP-C18 material is protected with a stabilized C18 organosilane bonded-phase.35 No change in retention or column efficiency was observed after separation of over 100 samples using the elevated temperature and low pH preferred conditions in the present study. Gradient elution separations of serum proteins from the mRP-C18 column were performed at temperatures of 26 °C, 40 °C, 60 °C, and 80 °C and are shown in Figure 4A. A comparison of the 80 °C and 26 °C chromatograms in Figure 4B,C, details the improvement at elevated temperature for apparent number and spacing of bands when comparing results obtained with the mRP column at 26 °C. Results using intermediate temperatures of 40 °C and 60 °C also showed improvement, but best results were obtained at the highest Journal of Proteome Research • Vol. 4, No. 5, 2005 1525

research articles

Martosella et al.

Figure 3. Comparison of the RP-HPLC elution profiles (absorbance at 280 nm) for crude human serum, shown in Panel (A), and human serum depleted of high-abundant proteins, shown in Panel (B). An aqueous TFA and ACN (TFA) gradient was used at 80 °C at a flowrate of 0.75 mL/min on a 4.6 mm ID × 50 mm mRP-C18 column. Each sample comprised a total of 270 µg protein in 6 M urea/1% AcOH, however, the depleted serum sample also contained salts introduced from the immunodepletion process.

temperature employed. Therefore, separations and fraction collecting were performed at 80 °C. The fractions were dried and further resolved and analyzed by SDS-PAGE. Fractions corresponding to 15, 17, 26, and 28 min on the chromatogram were subsequently analyzed by RP-LC-MS/MS analysis of 1526


tryptic fragments, with identification of protein by database search. R-1-acid-glycoprotein, apolipoprotein A1, hemopexin and complement component C4 were identified as major proteins within these fractions, and correspond to the labeled peaks shown in Figure 4B. R-1-acid-glycoprotein and comple-


research articles

Figure 4. Effect of column operational temperature on immunodepleted serum protein separations using the mRP-C18 column (4.6 mm ID × 50 mm). Multi-segment gradient elution conditions, 4.5% eluent B/min from 3 to 30%, 0.76% B/min. from 30 to 55%, 4.5% B/min. from 55 to 100%, 100% B 4 min at 0.75 mL/min. Panel (A) Four consecutive RP separations at 26 °C, 40 °C, 60 °C, and 80 °C. Blank runs were performed at 80 °C between runs until a stable baseline was reproduced. Panel (B) RP chromatographic separation at 80 °C detailing the resolution of R-1-acid glycoprotein, complement C4, hemopexin, and apolipoprotein A1. Arrows indicate presence of protein identified by SDS-PAGE analysis. Panel (C) Shows a loss in resolution for high-abundant proteins at 26 °C. Note: For temperature comparisons, each flow-through sample after immunodepletion was concentrated and the salts removed by buffer exchange prior to the reversed-phase separation. Thus, a large decrease in UV absorbance at t0-3 min. is observed when compared to all other chromatograms presented. Journal of Proteome Research • Vol. 4, No. 5, 2005 1527

research articles

Martosella et al.

Figure 5. Representative RP-HPLC elution profile of immunodepleted human serum obtained when using linear elution conditions of increasing concentrations of acetonitrile on a 4.6 mm ID × 50 mm mRP-C18 column. Dotted line gradient, 2.3% eluent B/min from 10 to 80%.

ment component C4 have been resolved within 2 fractions for each, #15-16 and #27-28, respectively. Apolipoprotein A1 was identified in fraction #26, and although the chromatogram suggests elution across several fractions, LC-MS/MS and SDSPAGE analysis of the fractions determined the majority of this protein eluted within fraction nos. 26 and 27. At 26 °C the chromatographic profile for many high-abundant proteins, including R-1-acid-glycoprotein, apolipoprotein A1, hemopexin and complement component C4 has changed and shows lower retention times and loss of resolution. SDS-PAGE analysis of this separation confirms band broadening and carryover of proteins across many fractions (between 4 and 8 min apparent peak widths). As an example, apolipoprotein A1, which in Figure 4C does not appear to show a large apparent change on the chromatogram, other than retention shift, exhibits elution across many fractions when visualized by SDS-PAGE. A possible explanation may be the presence of temperature sensitive conformers, which coalesce at elevated column operational temperature (for example, as described in ref 34). The chromatographic separation combined with downstream SDS-PAGE analysis of fractions permits better comparison and definition of protein identity and elution characteristics for this highly complex sample. Preferred RP Gradient Conditions. Our goal was to derive elution conditions for improving the resolution of coeluting serum proteins by obtaining retention across a useful time window, while not inducing excessive band broadening or protein mass losses. Since it is unrealistic to consider the sample as a collection of resolvable components, this goal is equivalent to obtaining a broad distribution of the immunodepleted serum proteins. Thus, preferred conditions should reduce masking lower abundance proteins, when detection occurs through downstream tryptic digestion and 2D-LC-MS/ MS identifications. Initial gradient investigations began with immunodepleted serum proteins denatured at neutral pH in the presence of 6 1528


M guanidine hydrochloride. As gradient work progressed, a switch was made to acidic conditions using 6 M urea as the denaturant. This change permitted comparison of denaturant effects on the resolution and recovery during the RP separation, as well as evaluation of possible effects on the performance of the subsequent SCX fractionation step of the 2D-LC-MS/MS analysis. The two denaturants had little differential effect on the separation with respect to resolution and recovery, but urea was preferred. We observed a negative effect of guanidine hydrochloride on the reproducibility of performing the off-line SCX separation. Gradient profiles were evaluated by systematically changing the water/acetonitrile (TFA) gradient, denaturant, acid modifier concentration (TFA), and flow rate, all at a fixed column temperature of 80 °C. Beginning with linear elution gradients of aqueous TFA and ACN (TFA), the depleted serum separations did not deliver the broad peak distribution preferred for optimal fraction collecting. As shown in Figure 5, the majority of late eluting proteins eluted within a narrow range of increasing organic concentration. Systematically reducing the gradient slope and extending the run times did not produce acceptable separations, exhibiting significant band broadening and run times. Inspection of the linear elution profile in Figure 5 shows that the majority of proteins eluting within the window of 3050% ACN. Consequently, the preferred conditions required a segmented gradient scheme in the 30-50% ACN window, with steeper gradients steps preceding and following this segment, to selectively improve resolution. The gradient profile is then terminated by an isocratic hold of acetonitrile, which we observed to improve repetitive yields of protein recovery and run-to-run reproducibility (see Figure 4B). Following optimization of the gradient program, systematic changes of the TFA modifier concentration were tested. Starting TFA concentrations were varied from 0.1% to 0.5% for water and 0.08% to 0.4% for ACN. We observed limited effect on either selectivity or retention of the immunodepleted serum proteins.


research articles

Figure 6. SDS-PAGE analysis of RP-HPLC fractionated immunodepleted human serum from an mRP-C18 column (9.4 mm ID × 50 mm). 1250 µg of depleted human serum sample was injected onto the column and eluted by a multi-segment gradient (see Materials and Methods, Figure 4A representative of the separation). Fifty fractions were collected (29 shown) for analysis by 4-20% SDS-PAGE. Fraction nos. 7-12 were dissolved in 30 µL of sample buffer and 15 µL loaded onto the gel. Fraction nos. 13-36 were dissolved in 75 µL of sample buffer and 15 µL loaded onto the gel.

TFA concentration was therefore set at 0.1% for water and 0.08% for ACN. SDS-PAGE of Optimized RP Separation. SDS-PAGE patterns were used as another measure of separation efficiency of the fractionated proteins. Examination of the protein patterns on gels determined the effectiveness of the RP separation, based on carryover between fractions. Apparent band intensities also gave an indication of column capacities by the extent of their carryover from one lane to another. As shown in Figure 6, using the preferred conditions, protein carryover between fractions is limited, particularly for the region of the chromatogram where protein mass elution is highest (12-33 min). The gel patterns within this region highlight the separation efficiency of remaining moderately abundant proteins and reveal many distinct and unique bands that would otherwise be obscured. Using the combination of chromatography with electrophoretic analysis enables a more informative approach to define the efficiency of the serum protein separation. We evaluated the separation efficiency using the preferred conditions, comparing the mRP-C18 and 300 Å porous silica C8 columns (Zorbax 300 SB-C8). The separation advantage of using the mRP-C18 is well illustrated in comparing Figures 3B & 7A. There is diminished peak fine structure and significant band overlap and tailing for the SB-C8 absorbance chromatogram, also evidenced by examination of fractions on SDS-PAGE gels. 36 fractions from each separation were visualized on SDSPAGE and the results shown in Figures 6 and 7B. Fractions 1236 represent the regions where the bulk of protein mass is eluted. Within this region, the SB-C8 column fractions show similar high abundant proteins bands in each fraction; the

bands were continuous across many more gel lanes indicative of band broadening, or elution at multiple points. Reproducibility of the RP Separation. mRP-C18 column separation reproducibility was examined using the preferred conditions. Separations of immunodepleted human serum from three injections, and the subsequent blank injections were compared are shown in Figure 8A. The RP separations are reproducible and show no changes from run to run in either the retention times, apparent selectivity, or bandwidths. The blank runs performed after each separation are reproducible and show no indications of peak ghosting or protein carryover. We also collected two identical fractions from each separation and compared them using SDS-PAGE analysis (Figure 8B). The separations revealed no apparent differences in either banding pattern or band intensity. Similar results can be obtained on separations conducted on different days. Following these separations, blank injections were routinely run to establish the absence of ghost peaks. Effect of Packing Material on Column Load. Increasing the load tolerance of the separation enables improved detection and characterization of proteins in downstream MS-based analysis using the workflow described in Figure 1. This is expected via the improvement in the quality of MS data in terms of peptide sequence coverage and quality of the spectra (signal-to-noise ratio). Increased protein loads also aid in visualization of low-abundant serum proteins in downstream electrophoretic analysis. The observed limit for permitted maximum column load for this complex separation was determined by the peak shape and bandwidth of the more abundant proteins in the sample, specifically, apolipoprotein Journal of Proteome Research • Vol. 4, No. 5, 2005 1529

research articles

Martosella et al.

Figure 7. Panel (A) RP-HPLC separation of 270 µg immunodepleted serum performed on a 4.6 mm ID × 50 mm Zorbax 300SB-C8 (300 Å, 5 µm) column under the preferred conditions presented in Materials and Methods. Panel (B) Electrophoretic analysis of the fractionated immunodepleted human serum sample presented in panel A using 4-20% SDS-PAGE.

A1 and R2-macroglobulin. We determined the uppermost column capacity by the extent at which fractionation caused band repetition across gel lanes, in a manner similar to the criteria used for electrophoretic evaluation of column resolution. The 4.6 mm ID × 50 mm mRP-C18 column provided a total protein loading capacity of 400-500 µg and the 9.4 mm ID × 50 mm column gave capacities in the range of 13001500 µg. The capacity evaluation was also performed on the porous SB-C8 column of the same dimensions and showed the mRP-C18 column tolerated a 3-fold higher load capacity before apparent overloading. Determination of Protein Recovery. High protein recovery is a critical attribute for useful protein pre-fractionation strate1530


gies. Any protein losses, whether specific or not, introduce bias in the use of a pre-fractionation strategy for comparative analysis of protein abundance. Similarly, loss of protein during the separation is typically associated with column fouling (usually irreversible binding or precipitate formation within the column), elution of ghost peaks on blank injections, and carry over between samples. Although specific column wash regimes may be employed for column cleanup between injections, for many reasons, it is much more desirable to develop conditions allowing high mass balance and the absence of peak ghosting. A number of specific problematic protein and peptide separations have previously been investigated to define the requirements for full recovery of proteins and hydrophobic pep-


research articles

Figure 8. Panel (A) Overlay of chromatograms from three reversed-phase separations of immunodepleted human serum and the subsequent blank injections after each run, separated on a 9.4 mm ID × 50 mm mRP-C18 column. 1080 µg of depleted serum in 6 M urea/1% AcOH was separated in each run under the preferred conditions described in Materials and Methods. The blocked regions represent areas of fraction collection for SDS-PAGE reproducibility analysis. Three blank runs (no protein injected, but same matrix solution) were conducted between each protein separation, with the results shown on the panel. Panel (B) SDS-PAGE of fractions 19 and 25. The collected fractions were dried and dissolved in SDS sample buffer and loaded onto a 4-20% SDS-PAGE. Lane 1 is the Mark 12 standards (Invitrogen). Lanes 2, 3, and 4 represent RP runs 1, 2, and 3.

tides.38,39 Specific utility has been observed for the use of elevated temperatures, particularly for proteins that have a

tendency to form visible precipitates, due to either high hydrophobicity, propensity to spontaneously form stable exJournal of Proteome Research • Vol. 4, No. 5, 2005 1531

research articles

Martosella et al.

Table 1. Reversed-Phase HPLC Protein Recovery Values for Macroporous and Porous Column Materials at Two Operating Temperatures separation media

recovery, 80 °C n ) 8

recovery, 26 °C n ) 3

mRP-C18 SB300-C8

103.0 ( 5.6 99.1 ( 2.6

70.5 ( 0.4 70.2 ( 6.9

tended aggregates (dimers, tetramers, etc.), or both. In all such previous cases, we noted that the appearance of ghost peaks in the chromatogram was invariably associated with low recovery of protein or peptide mass. For the determination of recoveries of serum proteins using the current RP separation protocol, we performed a series of injections of 200 µg protein from immunodepleted human serum samples onto the mRP-C18 and porous SB-C8 columns, as well as column blanks (no column in the flow path). The solid-phase fluorescence-based protein assay used to quantify protein recovery was chosen due to high tolerance of lipids, detergents and urea, among other interfering agents, to ensure accurate protein quantification.40 The method served to confirm and supplement our earlier observations, using the BCA assay, which indicated high protein recoveries at a column temperature of 80 °C. As shown in Table 1, recovery values for the mRP-C18 and SB-C8 columns were similar when measured at the same column operating temperatures. The RP separations performed at 80 °C, using the preferred gradient conditions, provided high recovery of injected immunodepleted serum. In contrast, separations performed under lower temperature conditions yielded much poorer recoveries, resulting in approximately 30% of total protein losses. Linear gradients are commonly used for protein and peptide separations and typically start at 90/10 water/ACN (modifier) and finish with 80% acetonitrile.41 Postcolumn blank runs under these conditions showed significant ghost peaks at room temperature. Continually adsorbed and bound hydrophobic materials, including proteins, will degrade separation efficiency and expedite column replacement.42 Column regeneration methods are sometimes used, such as trifluoroethanol or guanidine HCl washing, to remove hydrophobic or membrane proteins, but this only extends column life and does not improve recovery.43 For separations of immunodepleted human serum, we found that by completing the gradient with 100% B at 80 °C, and holding for a short period of time (4-5 min), we could regenerate the column surface and maintain very high recoveries. For separations of crude human serum at 80 °C, which contain a significant amount of albumin, the recovery, although not measured quantitatively was lower as evidenced by ghosting peaks in subsequent blank runs. 2D-LC-MS/MS. To define the utility of the new RP method for value in identification of proteins by downstream LC-MS methods, immunodepleted samples were fractionated by RP, the protein ID results generated on selected fractions, and the features were compared to results obtained starting with unfractionated immunodepleted serum samples. Fifty fractions were collected at 1.0 min. time intervals from the separation of 1250 µg of immunodepleted human serum protein using the mRP-C18 column and preferred operating conditions. Four nonadjacent fractions were chosen for 2D-LC/MS/MS analysis. The selected RP fractions, and a serum sample that was immunodepleted only, were analyzed individually by an onlinecoupled nanoscale 2D-LC system and ion-trap mass spectrom1532


eter. For the four collected fractions, (at 16, 20, 29, and 32 min), 144, 114, 96 and 107 proteins were identified (total of 461 IDs), with 360 nonredundant proteins identified from the combined results. The 78% unique identifications (360/461) indicate minimal protein or protein isoform overlap among the fractions. To evaluate the efficacy of pre-fractionation in increasing protein identification, a comprehensive approach would require analysis of every RP fraction, at least in duplicate, then combining the protein identification results, followed by comparison of these results with the combined results from an identical numerical repeat analysis of immunodepleted human serum. This comparative approach is essential since more proteins can be identified when a complex protein sample, such as human serum, is analyzed repeatedly, based on the intrinsic sampling issue associated with shotgun proteomics methodology.8 In theory, as previously described,8 the number of identified proteins would reach a plateau after a certain number of repeat analyses, presumably well below 50 times, which would be the case of depleted human serum without prior RP fractionation. With RP HPLC pre-fractionation one obtains 50 relatively unique protein samples, and each could produce a similar number of identified proteins as the immunodepleted human serum produced. Combining the results from all individually analyzed RP fractions could thus significantly increase the dynamic range of the combined LCMS/MS approach, resulting ultimately in a higher number of identified proteins. The number of samples and analysis effort to undertake this approach is beyond the scope and purpose of the current communication. To simplify the evaluation of our RP fractionation strategy, protein identification results using the two selected RP fractions were compared with the immunodepleted-only human serum. Table 2 presents the results for proteins that were significantly enriched in the fractions, compared to the same proteins in the immunodepleted-only sample. The degree of enrichment in RP fractions is reflected in the increased relative abundance data (number of spectra) and protein sequence coverage. Although these are relative measurements of the quantities of proteins present, we can estimate the effects by examining the proteins involved. The four proteins at lowest apparent abundance in Table 2 show significant improvements in the identification metrics. On the basis of previous measurements, using immunochemical methods,44 these proteins are present at levels of 15-50 µg/mL in normal serum samples. As an example, haptoglobin related protein has been reported in healthy donors to be present in the range of 30-40 µg/mL.45 This protein has been of some potential clinical interest as a biomarker of malignant lymphoma.46 To validate the enrichment of proteins identified in Table 2, we undertook examination of extracted ion chromatograms (EIC) derived from the LC-MS data sets used for the presented identifications. As an example, Figure 9A presents the EICs corresponding to a selected peptide identifying the protein inter-alpha-trypsin inhibitor heavy chain H2 precursor (ITI H2, see Table 2). Figure 9B are exemplary MS/MS spectra for the fraction and the unfractionated sample for a selected peptide that contributed to the positive identification of the protein. The ratio of integrated EIC mass abundance for this peptide was 22-fold higher in fraction 29 compared to unfractionated immunodepleted serum. Similarly, an additional peptide derived from ITI H2 underwent this analysis, generating the ratio of 17. These ratios are comparable to the values of 283 spectra

research articles


Table 2. Reversed-Phase Enriched Proteins Identified in Two Individual RP Fractions Verses Proteins Identified in an Immunodepleted-Only Human Serum Fraction fractionated protein •MWa

not fractionated

protein •pIb

fraction no.

Num Spectrac

Num Pepsd

scoree

% AA coverage

Num Spectrac

Num Pepsd

48721.3 11175.1

5.3 6.3

29 29

785 536

21 8

329.26 123.77

48 69

10 348

5 5

75.61 83.86

18 64

IPI00413704 IPI00017601

106596.2 122205.8

6.5 5.4

29 20

217 187

18 26

277.88 428.14

30 38

3 11

3 9

44.7 129.21

4 15

IPI00465313

166128.3

6.1

29

167

42

603.9

40

109

34

481.79

34

IPI00292530

101389.7

6.3

29

50

8

130.93

14

26

5

78.79

12

IPI00453459

192798.5

6.9

29

45

24

354.24

21

26

17

229.83

18

IPI00019591 IPI00294193

85533.4 103358.9

6.7 6.5

20 20

40 35

15 8

206.49 126.06

29 15

10 18

7 7

98.55 97.08

14 11

IPI00022895

54272.8

5.6

20

24

10

156.59

31

14

6

76.93

26

IPI00021855

9332

8.0

20

14

3

40.13

26

2

2

21.73

24

IPI00296170

43077.8

6.7

20

9

4

55.85

14

2

1

14.86

4

IPI00022426

38999.7

6.0

20

7

6

77.99

23

1

1

16.22

3

IPI00025204

38088.1

5.3

20

5

4

53.98

21

1

1

10.27

6

protein name

protein ID

48 kDa protein Apolipoprotein A-II precursor 106 kDa protein Ceruloplasmin precursor Alpha 2 macroglobulin Inter-alphatrypsin inhibitor heavy chain H1 precursor Complement component 4B proprotein Splice isof orm of complement f actor B precursor Splice isof orm of inter-alphatrypsin Inhibitor heavy chain H4 precursor Alpha-1B-glycoprotein precursor Apolipoprotein C-I precursor Haptoglobinrelated protein AMBP protein precursor CD5 antigenlike precursor

IPI00032215 IPI00021854

scoree

% AA coverage

a Predicted protein molecular weight. b Predicted protein iso-electric point. c Number of matched MS/MS spectra. d Number of unique peptides. e Protein score from Spectrum Mill.

versus 3 spectra (ratio of 94.3) and 30% AA versus 4% AA coverage (ratio of 7.5) obtained using MS/MS-derived relative abundance metrics for protein identification. EICs representing additional peptides from these identified proteins were inspected. No credible signal was observed in the immunodepleted-only datasets at masses or realistic retention times corresponding to such peptides. Examination of several additional protein identifications from Table 2, comparing the ID metrics and EIC analysis of the component peptides yielded a similar pattern of results; in general, we have observed that the use of MS/MS-based measures, such as the number of spectra and % AA coverage, when used as relative abundance measurements, are generally conservative. The difference values obtained for rations from the EIC, number of spectra and sequences coverage are expected to be different. These three measurements have been used as indicators of relative abundance, EIC is indeed the more accurate measurement, but doing this for a large number of peptides is extremely timeconsuming. The use of the number of spectra and the sequence coverage can be used to indicate whether a protein is in greater abundance in the sample. In addition to proteins that were identified as enriched in RP fractions, relative to unfractionated immunodepleted samples, we also identified a large number of proteins that were

identified by LC-MS/MS analysis of the fractions, but not at all in the unfractionated sample. These proteins are shown in Table 3. Note that Table 3 only includes the 34 proteins identified by multiple peptides, although there were also many more single peptide-based protein identifications, an additional 108 so-called one-hits. Several of the identified proteins were further examined by EIC generation for selected peptides. In all cases, peptides were present in the appropriate data for the fraction analysis, and in no cases was evidence obtained to support the detection of the peptides in the unfractionated sample data sets. An example of the identification of a protein of interest in the analysis of a selected fraction is provided by hepatocyte growth factor-like protein precursor, also known as macrophage stimulating 1 (MST), or macrophage stimulatory protein (MSP). This inflammatory cytokine protein is present in normal human serum samples at the level of about 20 ng/ mL, and is responsive to certain pathophysiological events including tissue remodeling/wound healing, response to lipopolysaccharides, as well as hematopoiesis, and bone formation.44,47 An exemplary MS/MS spectra identifying one of the two peptides identifying this protein is shown in Figure 10, demonstrating the definitive fragmentation pattern for this peptide, despite the problematic presence of a pair of adjacent aspartyl-proline residues. Journal of Proteome Research • Vol. 4, No. 5, 2005 1533

research articles

Martosella et al.

Figure 9. Panel (A). Overlay of the EICs (m/z 1121.1-1123.3) corresponding to a selected peptide identifying the protein inter-alphatrypsin inhibitor heavy chain H2 precursor (ITI H2). The elution profile presented is for the nanospray LC-MS for this peptide, which was found in the 200 mM salt fraction. The ratio of the EIC areas was found to be 22. Panel (B). An exemplary MS/MS spectra for the fraction and the unfractionated sample for the selected peptide, which contributed to the positive identification of the protein.

The results in Tables 2 and 3 are based on the single pass analysis of the fractions and of the immunodepleted unfractionated human serum sample. For comparison purposes, replicate analysis was conducted for the unfractionated sample, resulting in a small increment in the MS/MS-based identification metrics (abundance and coverage), relative to those from the summed analysis for the RP fractions. Taken together, the increased unique protein identities, as well as the improved sequence coverage and abundance data, are strong indications that a selected group of proteins are enriched in each RP 1534


fraction, which thereby increases the dynamic range of the overall proteomic analysis.

Conclusions Reversed-phase HPLC separations of human serum, depleted of six high abundant proteins, appears to be a promising approach to fractionate complex serum samples, and thereby to enhance identifications of low abundant proteins. Specifically, the combination of an appropriate RP column material and preferred operational conditions allowed high-resolution

research articles


Table 3. Proteins Identified in Two Reversed-Phase Fractions (at 20 and 29 min) of Immunodepleted Human Serum with at Least Two Unique Peptides Per Protein % AA coverage

protein•MW

protein•pI

fraction no.

585.9

26

256513.2

5.5

20

197.9 104.1 174.2

33 35 29

68611 26234.6 67818.6

6.9 6.3 8.0

20 20 20

20

47651.1

5.3

20

130.4

28

65163.6

6.1

20

9 6 3 13

106.7 99.62 46.89 184.3

23 13 22 37

56111.3 79523.5 26019.3 67047.3

7.8 5.4 6.1 8.5

20 20 20 20

16

8

114.2

11

92375.1

5.9

29

IPI00020091

16

5

67.87

26

23602.8

5.0

20

IPI00400826 IPI00472129 IPI00292946

13 12 12

5 8 6

72.78 112.5 89.57

15 10 26

57832.9 134905.8 46324.8

6.3 6.4 5.9

20 20 29

IPI00446503 IPI00011261

11 11

7 6

103.6 97.15

25 45

53321.6 22219.5

6.1 8.9

29 20

IPI00005439 IPI00339319

10 6

5 5

66.75 60.13

25 3

42094.3 262389.7

6.9 5.4

20 29

IPI00291867

5

4

48.89

7

65720.7

7.7

20

IPI00382938 IPI00026314

5 4

3 4

36.68 53.15

22 9

25977.2 85697.9

6.3 5.9

29 20

IPI00336074

4

3

42.86

14

53391.8

8.1

20

IPI00218192 IPI00021817

4 3

3 3

42 37.85

3 11

101242.5 52071.6

6.2 5.9

29 20

IPI00328609 IPI00022937

3 3

3 3

35.77 29.39

6 2

50612.6 251702.7

8.5 5.7

29 20

IPI00218746

3

2

28.36

11

26703.9

8.8

20

IPI00292218

2

2

33.57

6

81926.5

8.2

20

IPI00009920

2

2

30.92

2

104844.8

6.3

20

IPI00165421

2

2

26.17

8

29263.8

9.0

29

protein name

protein ID

NumSpectra

NumPeps

FIBRONECTIN 1 ISOFORM 4 PREPROPROTEIN IGHM PROTEIN IG KAPPA CHAIN C REGION COAGULATION FACTOR XII PRECURSOR ALPHA-1-ANTICHYMOTRYPSIN PRECURSOR COMPLEMENT COMPONENT C8 ALPHA CHAIN PRECURSOR FLJ00385 PROTEIN 79KDA PROTEIN 26KDA PROTEIN COMPLEMENT COMPONENT C8 BETA CHAIN PRECURSOR PHOSPHATIDYLINOSITOLGLYCAN-SPECIFIC PHOSPHOLIPASE D 1 PRECURSOR ALPHA-1-ACID GLYCOPROTEIN 2 PRECURSOR CLUSTERIN ISOFORM 1 134KDA PROTEIN THYROXINE-BINDING GLOBULIN PRECURSOR MGC27165 PROTEIN COMPLEMENT COMPONENT C8 GAMMA CHAIN PRECURSOR FETUIN-B PRECURSOR SPLICE ISOFORM 11 OF FIBRONECTIN PRECURSOR COMPLEMENT FACTOR I PRECURSOR IGLC2 PROTEIN GELSOLIN PRECURSOR, PLASMA MGC27165 PROTEIN SPLICE ISOFORM 2 OF INTERALPHA-TRYPSIN INHIBITOR HEAVY CHAIN H4 PRECURSOR VITAMIN K-DEPENDENT PROTEIN C PRECURSOR FULL-LENGTH CDNA 5-PRIME END OF CLONE CS0DM009YC13 OF FETAL LIVER OF Homo sapiens COAGULATION FACTOR V PRECURSOR COMPLEMENT COMPONENT 1, Q SUBCOMPONENT, BETA POLYPEPTIDE PRECURSOR HEPATOCYTE GROWTH FACTORLIKE PROTEIN PRECURSOR COMPLEMENT COMPONENT C6 PRECURSOR SERPINC1 PROTEIN

IPI00414283

183

38

IPI00382937 IPI00419424 IPI00019581

108 62 42

13 6 13

IPI00431656

42

5

IPI00011252

23

9

IPI00168728 IPI00412100 IPI00395655 IPI00294395

21 21 21 18

IPI00299503

separations of the serum proteins of this sample. These separations are shown to be very reproducible and exhibit excellent bulk protein recovery. Analysis of protein fractions obtained by RP separation, followed by tryptic cleavage and subsequent 2D-LC-MS/MS of constituent peptides, yielded confident protein identifications of lower abundance proteins. The current study of protein RP separation conditions has demonstrated the strong temperature dependence on protein recovery, as well as a smaller dependency on features of the gradient elution conditions. We observed that recovery of total

score

97.34

protein is improved by roughly 30% on increasing the column operating temperature from 26 °C to 80 °C. The analysis of recovery, by direct protein determination, is expected to only address the question of the total protein mixture, rather than demonstrate the recovery of individual low abundance proteins. Although it is very likely that the preferred conditions will lead to high recovery of many of the individual proteins present in the immunodepleted serum sample, it is also possible that specific lower abundance proteins, or classes of proteins, may show more variable results. At the present time, we have not Journal of Proteome Research • Vol. 4, No. 5, 2005 1535

research articles

Martosella et al.

Figure 10. MS/MS spectra (single spectra, no averaging) for the peptide NPDGDPGGPWCYTTDPAVR used for the identification of hepatocyte growth factor-like protein (MSP). This protein is found in human sera at the level of about 20ng/mL. The higher intensity peaks for Y4 and Y14 are indicative of the facile fragmentation at Asp-Pro bonds. The corresponding B5 and B15 ions are also present.

conducted an analysis of the identities of the proteins that represent the 30% of mass of material lost by operation at the lower temperature, although comparison of absorbance chromatograms suggests that the most problematic proteins elute later in the gradient. The advantage of elevated temperature on recovery has been shown using familiar acetonitrile gradient elution conditions, combined with TFA as the acidic mobile phase modifier. Other choices of gradient elution modifiers may show greater or lesser effects, and there are many examples in the literature of specific protein separations for which these mobile phase modifiers are not optimal, at least for separation selectivity. Although we did not carry out systematic analysis of the effects of additional organic modifiers, we have previously observed that the use of mixtures of organic solvents (for example 1:1 acetonitrile/ 2-propanol, or similar mixtures) has minor effects on recovery, but can have highly specific effects on reversed-phase selectivity for protein separations. The ultimate purpose of the present study was to establish the utility of RP separation as a method for fractionation of a complex protein mixture, toward permitting protein identification by downstream LC-MS/MS methods for proteomic analysis. High protein recovery is judged to be of importance, as this is a requisite to prevent bias and irreproducibility of protein identification when analyzing complex protein samples, such as human serum. The problem with lower recoveries is at least 2-fold; loss of protein can be specific, leading to lost protein identities, and loss of protein during RP will lead to carry-over of protein between runs, possibly leading to miss-identifications in subsequent samples using the same column (or possibly the same chromatograph, if adsorption occurs on system components). Single use devices (disposables) could address the issue of sample carryover, but this approach would likely be expensive, compromise resolution capabilities, and not address the issue of information loss from lost proteins. The demonstration of high recovery using the preferred conditions for RP fractionation cannot ensure that a specific protein present at lower abundance is not lost, but should serve at least to alleviate concerns over sample carry-over and subsequent protein miss-identification. In the current work, we have not conducted a full survey of protein identification in human serum, as this would require reporting the analysis of all of the fractions obtained during 1536


the serum fractionation. Herein, we report on the detailed analysis of only two such fractions, of the 50 obtained. Further work is ongoing. Nevertheless, the utility of RP-LC separation is shown by the enrichment of proteins within selected fractions, leading to the identification of a greater number of peptides by downstream 2D-LC-MS/MS, and subsequent increase in protein sequence coverage (see Table 2). At the limit, we have demonstrated that a much larger number of human serum proteins can be detected using RP prefractionation, than can be determined by analysis of unfractionated serum. In specific cases, such as hepatocyte growth factor related protein precursor, we have shown identification of proteins present at levels as low as 20 ng/mL of human serum. This apparent increase in the dynamic range of protein detection using MS-based proteomic analyses of serum is hoped to be of use for the identification and structure elucidation of serum protein biomarkers of pathological events and of pharmacologic intervention. Further current investigations are examining the extension of this methodological approach for the analysis of proteins present in other biological fluids and tissue extracts.

References (1) Chernokalskaya, E.; Gutierrez, S.; Pitt, A. M.; Leonard, J. T. Electrophoresis 2004, 25 (15), 2461-2468. (2) Duan, X.; Yarmush, D. M.; Berthiaume, F.; Jayaraman, A.; Yarmush, M. L. Electrophoresis 2004, 25 (17), 3055-3065. (3) Fujii, K.; Nakano, T.; Kawamura, T.; Usui, F.; Bando, Y.; Wang, R.; Nishimura, T. J. Proteome Res. 2004, 3 (4), 712-718. (4) Lee, H.; Yi, E. C.; Wen, B.; Reily, T. P.; Pohl, L.; Nelson, S.; Aebersold, R.; Goodlett, D. R. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 2004, 803 (1), 101-110. (5) Tirumalai, R. S.; Chan, K. C.; Prieto, D. A.; Issaq, H. J.; Conrads, T. P.; Veenstra, T. D. Mol. Cell Proteomics 2003, 2 (10), 10961103. (6) Ferguson, P. L.; Smith, R. D. Annu. Rev. Biophys. Biomol. Struct. 2003, 32, 399-424. (7) Link, A. J.; Eng, J.; Schieltz, D. M.; Carmack, E.; Mize, G. J.; Morris, D. R.; Garvik, B. M.; Yates, J. R., 3rd. Nat. Biotechnol. 1999, 17 (7), 676-682. (8) Liu, H.; Sadygov, R. G.; Yates, J. R., 3rd. Anal. Chem. 2004, 76 (14), 4193-4201. (9) Jack, G. W. Mol. Biotech. 1994, 1, 59-86. (10) Janini, G. M.; Chan, K. C.; Conrads, T. P.; Issaq, H. J.; Veenstra, T. D. Electrophoresis 2004, 25 (13), 1973-1980. (11) Lee, W. C.; Lee, K. H. Anal. Biochem. 2004, 324, 1-10.

research articles

Reversed-Phase Chromatography of Human Serum Proteins (12) Marshall, J.; Jankowski, A.; Furesz, S.; Kireeva, I.; Barker, L.; Dombrovsky, M.; Zhu, W.; Jacks, K.; Ingratta, L.; Bruin, J.; Kristensen, E.; Zhang, R.; Stanton, E.; Takahashi, M.; Jackowski, G. J. Proteome Res. 2004, 3 (3), 364-382. (13) Righetti, P. G.; Castagna, A.; Herbert, B.; Reymond, F.; Rossier, J. S. Proteomics 2003, 3 (8), 1397-1407. (14) Rose, K.; Bougueleret, L.; Baussant, T.; Bohm, G.; Botti, P.; Colinge, J.; Cusin, I.; Gaertner, H.; Gleizes, A.; Heller, M.; Jimenez, S.; Johnson, A.; Kussmann, M.; Menin, L.; Menzel, C.; Ranno, F.; Rodriguez-Tome, P.; Rogers, J.; Saudrais, C.; Villain, M.; Wetmore, D.; Bairoch, A.; Hochstrasser, D. Proteomics 2004, 4 (7), 21252150. (15) Zuo, X.; Echan, L.; Hembach, P.; Tang, H. Y.; Speicher, K. D.; Santoli, D.; Speicher, D. W. Electrophoresis 2001, 22 (9), 16031615. (16) Borchers, C.; Peter, J. F.; Hall, M. C.; Kunkel, T. A.; Tomer, K. B. Anal. Chem. 2000, 72, (6), 1163-1168. (17) Washburn, M. P.; Wolters, D.; Yates, J. R., 3rd. Nat. Biotechnol. 2001, 19 (3), 242-247. (18) McDonald, W.; Ohi, R.; Miyamoto, D.; Mitchison, T.; Yates, J. Intl. J. Mass. Spectrosc. 2002, 219 (1), 245-251. (19) Kersey, P. J.; Duarte, J.; Williams, A.; Karavidopoulou, Y.; Birney, E.; Apweiler, R. Proteomics 2004, 4 (7), 1985-1988. (20) Bjorhall, K.; Miliotis, T.; Davidsson, P. Proteomics 2005, 5 (1), 307317. (21) Chromy, B. A.; Gonzales, A. D.; Perkins, J.; Choi, M. W.; Corzett, M. H.; Chang, B. C.; Corzett, C. H.; McCutchen-Maloney, S. L. J. Proteome Res. 2004, 3 (6), 1120-1127. (22) Echan, L.; Tang, H.; Ali-Khan, N.; Lee, K.; Speicher, D. W. Proteomics 2004, submitted. (23) Szafranski, C.; Baily, J.; Turck, C. W.; Nicol, G.; Martosella, J. D.; Zolotarjova, N. Pharmagenomics 2004, 4 (7), 40-46. (24) Zhang, K.; Zolotarjova, N.; Nicol, G.; Martosella, J. D.; Yang, L.; Szafranski, C.; Bailey, J.; Boyes, B. Agilent Technologies Application note 2004. (25) Pieper, R.; Gatlin, C. L.; Makusky, A. J.; Russo, P. S.; Schatz, C. R.; Miller, S. S.; Su, Q.; McGrath, A. M.; Estock, M. A.; Parmar, P. P.; Zhao, M.; Huang, S. T.; Zhou, J.; Wang, F.; Esquer-Blasco, R.; Anderson, N. L.; Taylor, J.; Steiner, S. Proteomics 2003, 3 (7), 1345-1364. (26) Badock, V.; Steinhusen, U.; Bommert, K.; Otto, A. Electrophoresis 2001, 22 (14), 2856-2864.

(27) Neverova, I.; Van Eyk, J. E. Proteomics 2002, 2 (1), 22-31. (28) Van den Bergh, G.; Clerens, S.; Vandesande, F.; Arckens, L. Electrophoresis 2003, 24 (9), 1471-1481. (29) Lescuyer, P.; Hochstrasser, D. F.; Sanchez, J. C. Electrophoresis 2004, 25 (7-8), 1125-1135. (30) Chen, Y.; Mant, C. T.; Hodges, R. S. J. Chromatogr. A 2003, 1010 (1), 45-61. (31) Kalghatgi, K.; Horvath, C. J. Chromatogr. 1987, 398, 335-339. (32) Antia, F. D., Horvath, C. Journal of Chromatography A 1988, 435, 1-15. (33) Guillarme, D.; Heinisch, S.; Rocca, J. L. J. Chromatogr. A 2004, 1052 (1-2), 39-51. (34) Cohen, K. A.; Schellenberg, K.; Benedek, K.; Karger, B. L.; Grego, B.; Hearn, M. T. Anal. Biochem. 1984, 140 (1), 223-235. (35) Boyes, B. E.; Kirkland, J. J.; Peptide Res. 1993, 6, (5), 249-258. (36) Zhu, C.; Goodall, D. M.; Wren, A. C. LC-GC 2005. (37) Chloupek, R. C.; Hancock, W. S.; Marchylo, B. A.; Kirkland, J. J.; Boyes, B. E.; Snyder, L. R. J. Chromatogr. A 1994, 686 (1), 45-59. (38) Boyes, B. E.; Kirkland, J. J., unpublished results. (39) Mant, C. T.; Kondejewski, L. H.; Cachia, P. J.; Monera, O. D.; Hodges, R. S. Methods Enzymol. 1997, 289, 426-469. (40) Agnew, B. J.; Murray, D.; Patton, W. F. Electrophoresis 2004, 25 (15), 2478-2485. (41) Moffatt, F.; Senkans, P.; Ricketts, D. J. Chromatogr. A 2000, 891 (2), 235-242. (42) Bhardwaj, S.; Day, R. A. 1997, 469-480. (43) Bhardwaj, S.; Day, R. A. LC-GC 1999, 17 (4), 354-355. (44) Qian, W. J.; Jacobs, J. M.; Camp, D. G., 2nd; Monroe, M. E.; Moore, R. J.; Gritsenko, M. A.; Calvano, S. E.; Lowry, S. F.; Xiao, W.; Moldawer, L. L.; Davis, R. W.; Tompkins, R. G.; Smith, R. D. Proteomics 2005, 5 (2), 572-584. (45) Tomlinson, S.; Muranjan, M.; Nussenzweig, V.; Raper, J. Mol. Biochem. Parasitol. 1997, 86 (1), 117-120. (46) Epelbaum, R.; Shalitin, C.; Segal, R.; Valansi, C.; Arselan, I.; Faraggi, D.; Leviov, M.; Ben-Shahar, M.; Haim, N. Pathol. Oncol. Res. 1998, 4 (4), 271-276. (47) Bezerra, J. A.; Carrick, T. L.; Degen, J. L.; Witte, D.; Degen, S. J. J. Clin. Invest. 1998, 101 (5), 1175-1183.

PR050088L

Journal of Proteome Research • Vol. 4, No. 5, 2005 1537

Reversed-Phase High-Performance Liquid Chromatographic

Recommend Documents