Anal. Chem. 2006, 78, 5762-5767
Quantification of Intermediate-Abundance Proteins in Serum by Multiple Reaction Monitoring Mass Spectrometry in a Single-Quadrupole Ion Trap Shanhua Lin,† Thomas A. Shaler,† and Christopher H. Becker*
PPD, Inc., 1505 O’Brien Drive, Menlo Park, California 94025
A method is presented to quantify intermediate-abundance proteins in human serum using a single-quadrupole linear ion trap mass spectrometersin contrast, for example, to a triple-quadrupole mass spectrometer. Stableisotope-labeled (tryptic) peptides are spiked into digested protein samples as internal standards, aligned with the traditional isotope dilution approach. As a proof-ofconcept experiment, four proteins of intermediate abundance were selected, coagulation factor V, adiponectin, C-reactive protein (CRP), and thyroxine binding globulin. Stable-isotope-labeled peptides were synthesized with one tryptic sequence from each of these proteins. The normal human serum concentration ranges of these proteins are from 1 to 30 µg/mL (or 20 to 650 pmol/mL). These labeled peptides and their endogenous counterparts were analyzed by LC-MS/MS using multiple reaction monitoring, a multiplexed form of the selected reaction monitoring technique. For these experiments, only one chromatographic dimension (on-line reversed-phase capillary column) was used. Improved limits of detection will result with multidimensional chromatographic methods utilizing more material per sample. Standard curves of the spiked calibrants were generated with concentrations ranging from 3 to 700 pmol/mL using both neat solutions and peptides spiked into the complex matrix of digested serum protein solution where ion suppression effects and interferences are common. Endogenous protein concentrations were determined by comparing MS/MS peak areas of the endogenous peptides to the isotopically labeled internal calibrants. The derived concentrations from a normal human serum pool (neglecting loss of material during sample processing) were 9.2, 110, 120, and 246 pmol/ mL for coagulation factor V, adiponectin, CRP, and thyroxine binding globulin, respectively. These concentrations generally agree with the reported normal ranges for these proteins. As a measure of analytical reproducibility of this single-quadrupole assay, the coefficients of variance based on 12 repeated measurements for each of the endogenous tryptic peptides were 17.0, 25.4, 24.2, and * Corresponding author. Phone: (650) 470-2386. Fax: (650) 470-2400. E-mail:
[email protected]. † These two authors contributed equally to this work.
5762 Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
14.0% for coagulation factor V, adiponectin, CRP, and thyroxine binding globulin, respectively. Various quantitative profiling approaches have been developed as the field of proteomics matures.1-5 Some of these approaches utilize stable isotopes. A label-free quantitative liquid chromatography-mass spectrometry (LC-MS) method has been reported by our group.5 This label-free method enables differential quantitative measurement of large numbers of proteins, peptides, and metabolites in complex matrixes relying on linearity of signal versus concentration and reproducibility of sample processing. As various differential expression approaches are applied to the study of diseases and drug monitoring, potential biomarkers are being discovered in serum and other sample types. As a result, there is a growing need to validate these markers accurately, rapidly, and cost-effectively. Traditionally, most targeted measurements of markers are by antibody-based immunoassays. However, for many newly discovered potential markers, antibodies are not readily available. An alternative, mass spectrometric-based, quantitative approach for validation, and perhaps implementation in the clinic, follows spiking of stable-isotope-labeled compounds as internal standards, known as the isotope dilution method.6,7 For many years isotope dilution mass spectrometry has been used as a quantitative method in the field of analytical chemistry for small compounds such as drug metabolites,8 hormones,9 and peptides.10 Recently isotope dilution MS has also been used to quantify a specific protein by spiking labeled peptides and measuring unique polypeptides formed by enzymatic hydrolysis (1) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R. Nat. Biotechnol. 1999, 17, 994-999. (2) Ji, J.; Chakraborty, A.; Geng, M.; Zhang, X.; Amini, A.; Bina, M.; Regnier, F. J. Chromatogr., B: Biomed. Sci. Appl. 2000, 745, 197-210. (3) Cagney, G.; Emili, A. Nat. Biotechnol. 2002, 20, 163-170. (4) Krieg, R. C.; Paweletz, C. P.; Liotta, L. A.; Petricoin, E. F., 3rd. Technol. Cancer Res. Treat. 2002, 1, 263-272. (5) Wang, W.; Zhou, H.; Lin, H.; Roy, S.; Shaler, T. A.; Hill, L. R.; Norton, S.; Kumar, P.; Anderle, M.; Becker, C. H. Anal. Chem. 2003, 75, 4818-4826. (6) Hamberg, M. Anal. Biochem. 1973, 55, 368-378. (7) Oda, Y.; Huang, K.; Cross, F. R.; Cowburn, D.; Chait, B. T. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 6591-6596. (8) Kostiainen, R.; Kotiaho, T.; Kuuranne, T.; Auriola, S. J. Mass Spectrom. 2003, 38, 357-372. (9) Tai, S. S.; Welch, M. J. Anal. Chem. 2004, 76, 1008-1014. (10) Dass, C.; Fridland, G. H.; Tinsley, P. W.; Killmar, J. T.; Desiderio, D. M. Int. J. Pept. Protein Res. 1989, 34, 81-87. 10.1021/ac060613f CCC: $33.50
© 2006 American Chemical Society Published on Web 07/15/2006
of the protein.11 In that study, tryptic peptides of apolipoprotein A-1 were monitored by LC-MS, a nontandem mass spectrometric approach. The potential advantage of measuring the fragment ions is reduction of background interferences, especially for complex mixtures such as serum or plasma. Fragmentation (by collisions or otherwise) in the majority of cases provides one or more unique fragment ions. The combination of the specific parent mass and the unique fragment ion is generally an unambiguous and sensitive method to selectively monitor and quantify the compound(s) of interest. The mass spectrometric method of measuring specific fragment ions in tandem instruments is known as selective reaction monitoring. When more than one target compound is investigated within one LC-MS/MS run, the method is commonly called multiple reaction monitoring (MRM). In addition to not requiring an effective antibody, the MRM assay has the advantages of specificity, quantification and high throughput. Recently, several groups have reported the results of MRM assays on trypsin-digested plasma or serum using stable-isotopelabeledtrypticpeptidesandatriple-quadrupolemassspectrometer.12-14 Barnidge et al.12 investigated use of two-dimensional LC-MS/ MS assay (SCX and reversed-phase chromatography), stabledisotope-labeled synthetic C-terminal peptide, and MRM to quantify the doped in prostate-specific antigen (PSA) concentration in human serum. Their method involved tryptic digestion of PSA and serum prior to quantification. Kuhn et al.13 used stable-isotopelabeled synthetic peptides and MRM assays to quantify C-reactive protein (CRP) in the serum of patients with rheumatoid arthritis. In that study, abundant proteins were first depleted from serum and the remaining serum was digested by trypsin. Quantification of CRP levels in patient serum was determined by measuring the ratio of the endogenous tryptic peptide of CRP against a 13C-labeled synthetic peptide standard. While the study by Kuhn et al. focused on a single, intermediate-abundance protein concentration in the low-microgram per milliliter range, Anderson and Hunter14 addressed the MRM question of simultaneously quantifying many high- and intermediate-abundance plasma proteins (protein concentration from µg/mL to mg/mL) using labeled tryptic peptides. Their study showed that proteins down to ∼1 µg/mL, such as L-selectin, could be quantified in plasma, yield a dynamic range to 4.5 orders of magnitude in a single experiment. In this publication, we report experimental results on quantification of intermediate-abundance proteins in serum by MRM assay with stable isotopically labeled synthetic tryptic peptides. Unlike previously published work, our study was conducted using a single-quadrupole linear ion trap (Thermo Electron Corp, Waltham, MA; model LTQ). As a proof-of-concept experiment, five intermediate concentration proteins were selected: coagulation factor V, adiponectin, CRP, thyroxine binding globulin, and carboxypeptidase B. These five serum protein concentrations normally range from 1 to 30 µg/mL < or 20 to 650 pmol/mL > in human serum. Five stable-isotope-labeled synthetic peptides were synthesized using one tryptic sequence from each of these (11) Barr, J. R.; Maggio, V. L.; Patterson, D. G., Jr.; Cooper, G. R.; Henderson, L. O.; Turner, W. E.; Smith, S. J.; Hannon, W. H.; Needham, L. L.; Sampson, E. J. Clin. Chem. 1996, 42, 1676-1682. (12) Barnidge, D. R.; Goodmanson, M. K.; Klee, G. G.; Muddiman, D. C. J. Proteome Res. 2004, 3, 644-652. (13) Kuhn, E.; Wu, J.; Karl, J.; Liao, H.; Zolg, W.; Guild, B. Proteomics 2004, 4, 1175-1186. (14) Anderson, L.; Hunter, C. L. Mol. Cell. Proteomics 2006, 5, 573-588.
proteins. The mixture of these synthetic peptides was spiked into the trypsin-digested human serum. We were able to determine the endogenous protein concentration for four out of five proteins with CV of less than or equal to 25% for 12 repeated measurements. The carboxypeptidase B peptide proved insensitive and thus was dropped from the study. Although the results presented here were obtained on a linear single-quadrupole ion trap, they raise the question of using other types of tandem mass spectrometers besides the ion trap, and triple-quadrupole mass spectrometers, such as the LTQ-Orbitrap or Q-TOF models. Whether or not this linear single-quadrupole mass spectrometer obtains results that meet or exceed those of a triple-quadrupole or other type of tandem mass spectrometer, practically successful measurements are possible as shown here. Furthermore, numerous linear ion trap users do not have access to a range of tandem mass spectrometer types or models. EXPERIMENTAL SECTION Materials. Pooled human serum was a mixture from four anonymous healthy donors collected from Stanford Blood Center (Palo Alto, CA). Blood was collected using the 10-mL Vacutainer SST Gel & Clot Activator tube (red/black tiger top tube) following manufacturer recommended protocol (BD, Franklin Lakes, NJ). The handling of these biological materials was performed in accordance with U.S. Department of Health and Human Services guidelines for level 2 laboratory biosafety, as found in Biosafety in Microbiological and Biomedical Laboratories, 4th ed., HHS Publication (CDC) 93-8395. All other general reagents were purchased either from Fisher Scientific International, Inc. (Hampton, NH) or VWR International (West Chester, PA). Stable-Isotope-Labeled Synthetic Peptides. Five stableisotope-labeled synthetic peptides (Table 1) were purchased from Sigma-Aldrich (St. Louis, MO). They were chosen to represent one tryptic sequence from each of the following five human serum proteins: coagulation factor V, adiponectin, CRP, thyroxine binding globulin, and carboxypeptidase B. A 13C-labeled and 15Nlabeled arginine or lysine was incorporated into the C-terminal of each peptide during synthesis, shifting the overall peptide mass by 10 (Arg) or 8 Da (Lys), respectively. Working solutions of these peptides were made by dissolving them in 0.1% formic acid in equal molar ratio to a final concentration for each peptide of 400 fmol/µL. Upon checking these purchased products, the impurity level from the “light” (unlabeled) versions of the synthetic isotopelabeled peptides was determined to be well less than 1% of the labeled product and can be neglected in these experiments. Serum Proteome. The serum (35 microliters) was diluted 1:5 with a proprietary “Buffer A” from Agilent Technologies, Inc. (Palo Alto, CA) as part of their antibody-based depletion system for highly abundant proteins for the purpose of increasing the effective dynamic range of the measurements. The top six abundant proteins (serum albumin, immunoglobulin G, immunoglobulin A, transferrin, haptoglobin, antitrypsin) were depleted. The remaining proteins were denatured using 6 M guanidinium hydrochloride with tris(hydroxymethyl)aminomethane (tris) buffer (pH 8.3). Disulfide bonds were reduced using 10 mM dithiothreitol (4 h, 37 °C), and then sulfhydryl groups were carboxymethylated with 25 mM iodoacetic acid solution neutralized with NaOH (30 min at room temperature). The denaturant and reductionalkylation reagents were removed by buffer exchange with 50 mM Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
5763
Table 1. Summary of the Design and Results of This MRM Experimenta
parent ions m/z, labeled/unlabled (z)
monitored ions m/z, labeled/unlabled
reported serum concn range, pmol/mL (µg/mL)
coagulation factor V (SEAYNTFSER)
607.27/602.27 (2)
997.45/987.45 763.35/753.35
27.8 ( 7.9 (7.0 ( 2)
9.2 ( 1.6
3.3
17.0
adiponectin (IFYNQQNHYDGSTGK)
593.94/591.27 (3)
760.36/756.33 833.91/829.86
200.6 ( 34.0 (5.3 ( 0.9)
110 ( 28.
3.3
25.4
C-reactive protein (ESDTSYVSLK)
568.78/564.77 (2)
764.31/764.31
0.0-319.5 (0.0-8.0)
120 ( 29.
thyroxine binding globulin (TEDSSSFLIDK)
625.30/621.30 (2)
1019.5/1011.5 904.47/896.47
280.6-647.6 (13.0-30.0)
246 ( 34.
carboxypeptidase B (DTGTYGFLLPER)
689.84/684.84 (2)
411.27/401.21 968.47/968.47
103.3-206.6 (5-10)
protein names (peptide sequences)
measured endogenous concn, pmol/mL
LOD, pmol/mL
CV (%)
67 3.3
24.2 14
a The coefficient of variance (CV) is calculated from 12 repeated measurements of the ratio of the peak areas of the endogenous to the spiked (isotope-labeled) peptide for each sequence. The carboxypeptidase B peptide was dropped from the study due to low sensitivity and poor chromatographic behavior.
ammonium bicarbonate buffer (pH 8.3) using a 5-kDa molecular weight cutoff spin filter (Millipore Corp., Bedford, MA). After digestion of the proteins using modified trypsin (Promega Corp., Madison, WI) incubated for 22 h at 37 °C, the mixture was acidified with 1% formic acid, desalted with a C18 solid-phase extraction cartridge (Sep/Pak cartridge by Waters Corp., Milford, MA), dried, and redissolved in 350 µL of 0.1% formic acid for injection into the liquid chromatography-mass spectrometer system. The net dilution factor for the plasma digestion protocol is thus a factor of 10. LC-MS. A binary Agilent capillary-1100 series HPLC system was directly coupled to a Thermo Electron Corp. model LTQ electrospray ionization linear single-quadrupole ion trap mass spectrometer with automatic gain control to avoid space charge limitations. A capillary reversed-phase chromatography column (5-µm C18 material, column dimension 320 µm × 15 cm, MicroTech Scientific, Vista, CA) was used at a flow rate of 8 µL/min. Injection volume was 20 µL for all runs using a Leap Technologies (Carrbora, NC) model HTC PAL autosampler. Gradient elution of the proteome sample was achieved using 100% solvent A (0.1% formic acid in H2O) to 40% solvent B (0.1% formic acid in acetonitrile) over 46 min. A column heater (Hot-Sleeve 25 L, Analytical Sales and Products, Pompton Plains, NJ) was used to maintain the column temperature at 30 °C through the entire runs. Cocktails of labeled peptides at different concentrations were all spiked into the digested serum protein solution at a ratio of 1:3 by volume. Thus, the final dilution factor for the digested serum proteins was 10 × (4/3) ) 13.3. The 20-µL injection volume on column therefore corresponds to 1.5 µL of original volume of serum. Given ∼90% depletion by removing the six abundant proteins, a serum volume of 1.5 µL corresponds to ∼10 µg of digested serum proteins on column for each LC-MS/MS run. MRM Experimental Design. The mass spectrometer was tuned by infusing a solution of angiotensin I peptide, using the instrument autotuning program for optimization of the [M + 3H]3+ signal at m/z 433. The instrument source parameters not adjusted by the autotuning were set as follows: sheath gas flow 80, auxiliary gas flow 0, sweep gas flow 5, spray voltage 5 kV, and capillary temperature 275 °C. A single instrument tuning file was used for 5764
Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
all experiments. The parent ion for each MS2 scan was isolated with a mass window of 2.0 m/z units, fragmented using a collision energy of 35% with an activation time of 30 ms at Q ) 0.25, and the resulting monitored daughter ions scanned in profile mode over a 3.0 m/z unit wide window at a scan rate of 910 µs/u and data point density of 50 points/u. The maximum ion accumulation time was 100 ms, and the number of microscans was set to 1. The chromatographic run was divided up into three segments, with the mass spectrometer set to scan different transitions during each segment according to the peptide(s) that elute at those times. After an initial delay of 19 min, the following transitions were scanned during the first 5-min duration segment: 607.20 f (763.3, 997.4), 602.20 f (753.3, 987.4), 593.90 f (760.4, 833.9), 591.20 f (756.3, 829.8), 568.70 f (764.3), 564.70 f (764.3). The second segment (6 min) monitored the following transitions: 625.30 f (904.4, 1019.5) and 621.30 f (896.4, 1011.5). In the final 6-min segment, the following transitions were monitored: 689.80 f (411.2, 968.4) and 684.80 f (401.2, 968.4). The total cycle time required for each MS/MS scan was 150 ms. In general, the average chromatographic peak width to baseline was ∼30 s, and the fwhm was ∼10-15 s. Therefore, even in the most busy time segment (from 19 to 24 min, where 10 transitions were being monitored over three different peptide sequences and heavy/light versions), the total cycle time was ∼1.5 s. Thus, in this case, 20 points were taken across each entire peptide peak. Future multiplexing considerations are discussed below. Calibration Standard Curves of Stable-Isotope-Labeled Synthetic Peptides. Calibrant standard curves were generated with the stable-isotope-labeled peptide mixture diluted to final (injection) concentrations of 3.3, 6.7, 16.7, 33.3, 66.7, 166.7, 333.3, and 666.7 fmol/µL with 0.1% formic acid (neat) or with the digested human serum solution. Twenty microliters of this mixture was injected for each run. As mentioned above, this volume corresponds to 1.5 µL of serum. The total amounts of isotopelabeled peptide loaded onto the column for each point on the standard curve were 5, 10, 25, 50, 100, 250, 500, and 1000 fmol. For the neat-solution standard curve, one measurement was taken for each concentration. For the serum-spiked standard curve, four
repeated measurements were taken for each concentration. The peak area responses recorded for each peptide analyzed were determined at higher concentrations by using the Qual Browser software (Xcalibur 2.0, Thermo Electron Corp., Waltham, MA). Peak areas constraints were manually adjusted when signals were not intense and the software could not reliably determine peak width or background subtraction. The peak areas of the signal of a selected fragment ion of the labeled peptide (y) were plotted against the concentrations of the spiked, labeled peptides (x) and linearly fit standard curves were derived (y ) mx + b). Endogenous protein concentrations were determined by taking the ratio of the monitored MS/MS peak areas of the endogenous peptides compared with the internal calibrants, averaging 12 repeat measurements per peptide to assess variance. RESULTS Selection of Product Ions for MRM. To determine the chromatographic time segments and which fragment ions to use in the MRM assay, first a neat mixture of the five stable-isotopelabeled peptides was analyzed by on-line LC-MS/MS with the model LTQ ion trap mass spectrometer. After verifying the expected parent ion m/z values, the parent ion m/z of the five peptides was preselected for the continuous MS/MS analysis to determine the strongest fragment ion signal to record for the MRM experiment. The labeled peptides were then spiked into digested serum solution to verify these fragment ions were still the strongest and also to check for any possible interferences. After these investigations, in general two fragment ions per peptide were chosen for MRM monitoring, with the exception of CRP where only one fragment ion was sufficiently strong and not interfered to warrant its use (see Table 1). As noted above, after investigation, the peptide for carboxipeptidase B was dropped from the study due to poor sensitivity and chromatographic behavior. The retention times for the five peptides were as follows: coagulation factor V at 21.7 min, adiponectin at 21.7 min, CRP at 22.5 min, thyroxine binding globulin at 24.5 min, and carboxypeptidase B at 31.0 min. While it was possible to monitor all of these ions quasi-continuously (in rapid repeating succession), it was decided that, given the well-behaved chromatographic behavior of the remaining four peptides, some increase in duty cycle and sensitivity could be gained by using time segments for the peptides. Thus, because of their approximate retention time, the tryptic peptides of coagulation factor V, adiponectin, and thyroxine binding globulin were grouped together in the same time segment, from 19 to 24min, while the thyroxine binding globulin peptide was monitored separately in a time segment from 24 to 28 min. Figure 1 plots exemplary fragment-ion-mass-selected chromatograms from an MRM experiment. In this case, 1000 fmol of the isotope-labeled adiponectin peptide was spiked into digested proteome solution corresponding to an original volume of 1.5 µL of (abundant-protein-depleted) serum. The signals of the doubly charged product ions y13 and y14 from both the labeled and endogenous peptides were recorded as the integrated area under the chromatographic peak. The same experiment was repeated four times, corresponding to the four sets of traces in Figure 1. Some minor retention time shifting (∼0.3 min) was observed. The ratio of the labeled/unlabeled peak areas in this example for the four experimental runs was 6.6, 7.4, 5.4, and 6.9, or an average of
Figure 1. Exemplary ion chromatograms of the MRM experiment. Depicted here are four repeated MRM runs of 1000 fmol of isotopelabeled peptide mixture spiked into the digested human serum. The peak areas of the sum of y13 and y14 fragment ions of the labeled (L) and endogenous (unlabeled, UL) adiponectin peptide IFYNQQNHYDGSTGK are shaded.
6.6 with a CV of 12.8%. This ratio thus indicates (ignoring sample processing losses) an endogenous concentration of 101 pmol/ mL. Assessment of MRM Sensitivity for Proteins in TrypsinDigested Human Serum. To determine the sensitivity for each peptide by this MRM assay, two experiments were conducted. In the first experiment, a serial dilution of neat labeled peptide mixture (in the absence of digested serum) was analyzed by MRM, and the peak areas were plotted against the amount of labeled peptides loaded on column to generate a standard curve (Figure 2). A linear least-squares fit of a line was made for eight different concentrations, with just one recording at each concentration, for each protein’s tryptic peptide. All four peptides were easily detected at levels as low as 5-10 fmol loaded onto the HPLC column. In the second experiment, a similar analysis was performed, but this time the labeled peptide mixture was spiked into the digested human serum to build a more relevant standard curve and address the issue of limit of detection (LOD) within digested serum protein solution. Again, for each amount of labeled peptide, four repeat measurements were recorded. Linear least-squares fitted lines were derived, and the standard deviation for the signal of the labeled peptide fragment ions is plotted as the error bars (( one standard deviation); see Figure 2. The difference in slope of the neat and serum-mixture standard curves demonstrates ion suppression due to the presence of the complex mixture. Clearly, this mixture standard curve is the one that should be considered for the practical situation where labeled peptides are mixed with the digested proteome sample. The LOD for each protein (tryptic peptide) was determined based on where the signal-to-noise ratio Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
5765
Figure 2. Standard curves for both neat solution (labeled peptide mixture alone in 0.1% formic acid) and mixture solution (labeled peptide mixture spiked into digested serum proteins). Each dat a point used in the mixture standard curve was generated from a mean of the peak areas of four repeat measurements. The error bar of each of those four measurements is plotted as plus and minus one standard deviation.
of the labeled peptide in the presence of digested human serum reached 3 (Table 1). The concentration for limit of quantification (LOQ) is generally considered to be at a signal-to-noise ratio of 10, which here is ∼5 times the corresponding LOD values; the noise is a mixture of background and shot noise. Quantification of the Endogenous Protein Concentration and the Reproducibility of the MRM Experiment. The estimated original concentration of each endogenous peptide/protein was calculated by averaging the unlabeled/labeled ratio peak areas of 12 runs using a few different spiked amounts of labeled peptide, choosing spiked amounts so that the labeled peptide signal was comparable to (and somewhat more intense than) the endogenous peptide signal level. For example, the peak area for the endogenous CRP peptide was calculated from runs where 100, 250, and 500 fmol of labeled CRP peptides were injected, with 4 measurements for each or a total of 12 measurements. (In future experiments, a single spiked amount would suffice to determine the endogenous concentration.) Once the endogenous/spiked ratios of peak areas were determined, an estimate of the concentration of the endogenous peptide (and hence protein) was derived for this human serum sample using the known dilution factor in the sample processing steps (a factor of 13.3 here) and neglecting sample processing losses. Results are listed in Table 1 and generally are close to literature concentration ranges.15-19 Since the recovery of these endogenous peptides produced from the tryptic digestion of the endogenous proteins was less than 100%, it is not surprising that the protein concentrations (15) Koistinen, H. A.; Remitz, A.; Koivisto, V. A.; Ebeling, P. Diabetologia 2006, 49, 383-386. (16) Tracy, P. B.; Eide, L. L.; Bowie, E. J.; Mann, K. G. Blood 1982, 60, 59-63. (17) Wilkins, J.; Gallimore, J. R.; Moore, E. G.; Pepys, M. B. Clin. Chem. 1998, 44, 1358-1361. (18) Henry, J., Ed. Clinical diagnosis and management by laboratory methods, 18th ed.; Sauders: Philadelphia, 1991. (19) van Tilburg, N. H.; Rosendaal, F. R.; Bertina, R. M. Blood 2000, 95, 28552859.
5766
Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
determined by our MRM assay were somewhat lower than the expected number for the case of coagulation factor V and thyroxine binding globulin (Table 1). Coefficients of variance for these 12 measurements of endogenous concentrations were also calculated (Table 1). DISCUSSION This proof-of-concept study has demonstrated the utility of the MRM assay with stable-isotope-labeled (tryptic) peptides for quantifying intermediate-abundance proteins in human serum using an “unconventional” type of tandem mass spectrometer for this purpose, namely, a linear single-quadrupole ion trap. Results show sufficient sensitivity with modest CVs to measure endogenous coagulation factor V and thyroxine binding globulin from human serum with a one-dimensional LC-MS/MS assay (one dimension of chromatography). The serum concentrations of these five proteins range from 1 to 30 µg/mL, or 20 to 650 pmol/mL. In the case of coagulation factor V, although its concentration was determined, it is approximately at the same level as its LOD. For the cases of adiponectin and thyroxine binding globulin, the LODs were well below the endogenous concentrations. Another peptide choice will need to be investigated to properly address the case of carboxypeptidase B. The CVs for the measurements of endogenous peptide (protein) concentrations based on the ratios of monitored fragment ions to those of spiked, labeled peptides were 17.0, 25.4, 24.2, and 14.0% for the cases of coagulation factor V, adiponectin, CRP, and thyroxine binding globulin, respectively. Substantial dynamic range was also shown. All four calibration standard curves showed linearity across 3 orders of magnitude. The decision to choose which tryptic peptide for the MRM study was based on the results of previous profiling experiments, where these peptides showed good sensitivity and were identified by MS/MS analysis. In all but one case, these peptides provided good signal responses and low LODs. In the case of carboxypeptidase B, the peptide synthesized was probably chosen on the basis
of a mistaken identification; a future study could investigate other tryptic peptides. For this one-dimensional chromatographic method (with prior depletion of the most abundant serum proteins), this work shows an average LOD of roughly 20 pmol/mL. For a 25 000 Da protein, that would be 0.5 µg/mL, or 1 µg/mL for a 50 000 Da protein. A two-dimensional (2-D) chromatographic approach where more material is used undoubtedly will push the LOD to lower concentrations. It is worth considering the scenario where, within a 2-D method, just one or a subset of the chromatographic fractions is needed to provide the desired analyte(s) concentration thereby reducing the demand on LC-MS/MS machine time while taking advantage of the LOD from a 2-D chromatographic strategy. Also, depletion of a greater number of highly abundant proteins will be beneficial to the LOD. The derived serum concentrations in this work, as mentioned earlier, neglect protein sample processing losses. These results suggest only modest losses by comparison to literature values of normal protein concentration ranges. One approach to determining the amount of sample processing loss is to obtain an independent measure of the protein’s concentration (e.g., by an antibody-based assay) and then compare it with the MRM-derived value. For MRM measurements on small molecules, the labeled standard is generally spiked directly into the raw material, such as serum or urine. While it is desirable in general to introduce a labeled calibrant into the sample as early as possible in the processing steps, the situation with trypsin digestion made that choice impractical here because of the need to perform a buffer exchange after denaturing, reduction, and alkylation. Possibly a different buffer exchange method without use of a molecular weight cutoff spin filter could be employed in future efforts. The fact that there is significant sample processing involved in protein digestion underscores the consideration that consistent and robust experimental practices need to be followed to obtain optimal results in an MRM experiment.
It is envisioned that many more peptides (proteins) can be multiplexed in a single run. This limited set was chosen for this first investigation. There are practical advantages to dividing the data acquisition during an LC-MS/MS analysis into multiple time segments to increase the average duty cycle. The maximum number of peptides that can be multiplexed in a single LC-MS/ MS run will depend on the chromatography elution time distribution of those peptide. Based on the present proof-of-concept experiment, approximately five or six peptides could be readily targeted in one time segment. For the 46-min gradient here, it is reasonable to project the use of 5-10 time segments. Thus it is reasonable to anticipate that multiplexing could reach roughly 50 analytes in one LC-MS/MS run. Future work for this assay with a linear single-quadrupole ion trap may improve upon these CVs, LODs, and LOQs. It may be that triple-quadrupole methods will generally provide lower CVs, LODs, and LOQs although these initially determined values can be considered to represent practical utility and the LODs are comparable to triple-quadrupole results to date for tryptic peptides. Furthermore, many proteomic researchers do not have triplequadrupole instruments but do have recent model ion traps, and it is valuable to understand what analytical alternatives are available. Instruments are constantly being improved. Similarly, these studies could be extended to other types of non-triplequadrupole tandem mass spectrometers. ACKNOWLEDGMENT The authors thank Pamela Owings for preparing the digested human serum and Dr. Hua Lin for fruitful discussions.
Received for review April 3, 2006. Accepted June 19, 2006. AC060613F
Analytical Chemistry, Vol. 78, No. 16, August 15, 2006
5767
