Increased Throughput for Low-Abundance Protein Biomarker

Apr 23, 2009 - X. Tiger Hu and Michaela A. Owens. Journal of Agricultural and Food Chemistry 2011 59 (8), 3551-3558. Abstract | Full Text HTML | PDF |...
0 downloads 0 Views 333KB Size
Download PDF
Anal. Chem. 2009, 81, 3950–3956

Increased Throughput for Low-Abundance Protein Biomarker Verification by Liquid Chromatography/ Tandem Mass Spectrometry Michael Berna* and Bradley Ackermann Eli Lilly and Company, Drug Disposition Biomarker Group, Lilly Corporate Center, Indianapolis, Indiana, 46285 Low-abundance protein quantification has historically been performed using ligand binding techniques. However, due to the time and cost associated with developing enzyme-linked immunosorbent assay (ELISA), mass spectrometric approaches are playing an increasingly important role. Protein quantification at or below the nanogram per milliliter level using liquid chromatography/tandem mass spectrometry (LC/MS/MS) typically utilizes an immunoaffinity (IA) enrichment step such as immunoprecipitation. In order to maximize mass spectrometry (MS) sensitivity, protein enrichment is followed by a proteolytic cleavage step used to generate a surrogate peptide with better mass spectrometric properties. Unlike ELISA, IA-LC/MS/MS is a serial technique that can require up to 3 days for a single batch analysis due to lengthy incubation and digestion steps. This report describes the use of immunoprecipitation in 96-well ELISA format (IPE) and microwave-assisted protein digestion to reduce the time required to perform LC/MS/MS protein analyses to within a single day. The utility of this approach was investigated through its application to previously published LC/MS/MS protein assays from our laboratory for two cardiotoxicity biomarkers, Myl3 and NTproBNP. Using commercially available antibodies, IPE and microwave-assisted digestion were used to repeat intraday validations for these markers, and intraday precision (%CV) and accuracy (%RE) did not exceed 11% or 3% for either assay, respectively. Additionally, lower limits of quantification of 100 pg/mL (NTproBNP) and 0.95 ng/ mL (Myl3) were achieved. The field of proteomics has come under increased scrutiny by those who question its ability to reliably discover novel protein biomarkers. In response to such concerns, Rifai et al. introduced the concept of the protein biomarker pipeline in a landmark paper in 2006 which outlined the steps involved in taking protein candidates discovered by global proteomics through clinical validation.1 The end products of the pipeline are protein biomarkers having sufficient clinical and analytical validation for their intended use. The process described involves four stages: discovery, qualification, verification, and validation. Each successive stage is characterized by a smaller set of candidate proteins along * To whom correspondence should be addressed. Phone: 317-277-6279. E-mail: [email protected]. (1) Rifai, N.; Gillette, M. A.; Carr, S. A. Nat. Biotechnol. 2006, 24, 971–983.

3950

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

with an increase in the number of samples analyzed. The end point of the process is a validated assay in an accessible fluid, typically utilizing an enzyme-linked immunosorbent assay (ELISA). Mass spectrometry (MS) continues to play a central role in the discovery and development of protein biomarkers. During the initial stages of qualification, leads generated by global proteomics are reanalyzed using targeted assays. Western blot analysis has historically been used for corroboration but is limited by antibody availability. Scientists have increasingly turned to targeted analysis using triple quadrupole MS using selected reaction monitoring (SRM) detection. Stable isotope-labeled (SIL) analogues of the tryptic fragments used for protein identification are synthesized for quantitative analysis by liquid chromatography/tandem mass spectrometry (LC/MS/MS). This methodology is often referred to as AQUA (absolute quantification), which was originally proposed by Gygi and co-workers.2,3 Several examples of this strategy have been published4-6 with noteworthy cases involving the analysis of large protein panels7 and targeted assays for highabundance protein biomarkers.8,9 In addition to the AQUA methodology, several other proteomic strategies have been reported which incorporate SIL standards for improved quantification including SILAC (stable isotope labeling with amino acids in cell culture),10 ICAT (isotope-coded affinity tags),11 and iTRAQ (isobaric tag for relative and absolute quantitation).12 A key step in the progression outlined by Rifai et al.1 is the transition from protein biomarker qualification to verification. During this transition, it is necessary to perform targeted analysis in an accessible fluid, which often imposes severe demands on (2) Kirkpatrick, D. S.; Gerber, S. A.; Gygi, S. P. Methods 2005, 35, 265–273. (3) Gerber, S. A.; Rush, J.; Stemman, O.; Kirschner, M. W.; Gygi, S. P. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 6940–6945. (4) Barnidge, D. R.; Dratz, E. A.; Martin, T.; Bonilla, L. E.; Moran, L. B.; Lindall, A. Anal. Chem. 2003, 75, 445–451. (5) Li, N.; Nemirovskiy, O. V.; Zhang, Y.; Yuan, H.; Mo, J.; Ji, C.; Zhang, B.; Brayman, T. G.; Lepsy, C.; Heath, T. G.; Lai, Y. Anal. Biochem. 2008, 380, 211–222. (6) Mirzaei, H.; McBee, J. K.; Watts, J.; Aebersold, R. Mol. Cell. Proteomics 2008, 7, 813–823. (7) Anderson, L.; Hunter, C. L. Mol. Cell. Proteomics 2006, 5, 573–588. (8) Barnidge, D. R.; Goodmanson, M. K.; Klee, G. G.; Muddiman, D. C. J. Proteome Res. 2004, 3, 644–652. (9) Kuhn, E.; Wu, J.; Karl, J.; Liao, H.; Zolg, W.; Guild, B. Proteomics 2004, 4, 1175–1186. (10) Ong, S. E.; Mann, M. Methods Mol. Biol. 2007, 359, 37–52. (11) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R. Nat. Biotechnol. 1999, 17, 994–999. (12) Ross, P. L.; Huang, Y. N.; Marchese, J. N.; Williamson, B.; Parker, K.; Hattan, S.; Khainovski, N.; Pillai, S.; Dey, S.; Daniels, S.; Purkayastha, S.; Juhasz, P.; Martin, S.; Bartlet-Jones, M.; He, F.; Jacobson, A.; Pappin, D. J. Mol. Cell. Proteomics 2004, 3, 1154–1169. 10.1021/ac9002744 CCC: $40.75  2009 American Chemical Society Published on Web 04/23/2009

the sensitivity required. As a general rule, without extensive sample preparation, protein quantification in plasma using LC/ MS/MS cannot access biomarkers below 100 ng/mL. It is for this reason that immunoaffinity (IA) capture methods have become popular for MS quantification of low-abundance proteins. Because antibodies are able to enrich the protein of interest by up to 1000fold,13 limits of quantification below 1 ng/mL can be readily achieved. Moreover, when using antibody capture it is possible to avoid a dedicated step for abundant protein removal. The most widely recognized strategy for IA-LC/MS/MS is a procedure developed by Anderson and co-workers known as stable isotope standards for use with capture by antipeptide antibodies (SISCAPA).7,14 Using SISCAPA, antibodies raised against a target peptide of interest are used to improve analyte recovery after enzymatic digestion. Despite limitations associated with the use of antipeptide antibodies, SISCAPA is well-suited for use in multiprotein assays. Our laboratory has been engaged in a variation of the SISCAPA method that uses antiprotein antibodies as opposed to antibodies targeted against a tryptic fragment.15 Using this approach, a target protein is isolated by immunoprecipitation (IP) prior to enzymatic digestion and LC/MS/MS analysis. Several examples have now been published using antiprotein IA-LC/MS/MS for quantitative analysis of target proteins or peptides below 1 ng/mL.16-20 Because of the selectivity of MS and the need for only a single antibody, quantitative LC/MS/MS methods can often be developed in a fraction of the time and cost needed for a sandwich ELISA. We have previously detailed a strategy for expedited qualification/verification of protein biomarkers using this approach, ultimately leading to faster development of selective ELISA methods for prioritized biomarker candidates.15 Ironically, despite the comparative speed and flexibility of LC/ MS/MS, sample preparation and analysis using standard IP techniques typically occurs over multiple days; far longer than standard sandwich ELISA methods. One limitation is that common IP protocols using either magnetic beads13 or gels16 require the processing of individual tubes as opposed to microtiter plates. Sample preparation time is further prolonged by the need for enzymatic cleavage, which in some cases requires overnight digestion for completion. This article details recent advances toward more streamlined sample preparation by IP for quantitative protein analysis by LC/ MS/MS. IP in a 96-well format using protein A/G to immobilize the capture antibody was used to increase the throughput of sample-processing steps. Although examples of IP in 96-well (13) Whiteaker, J. R.; Zhao, L.; Zhang, H. Y.; Feng, L. C.; Piening, B. D.; Anderson, L.; Paulovich, A. G. Anal. Biochem. 2007, 362, 44–54. (14) Anderson, N. L.; Anderson, N. G.; Haines, L. R.; Hardie, D. B.; Olafson, R. W.; Pearson, T. W. J. Proteome Res. 2004, 3, 235–244. (15) Ackermann, B. L.; Berna, M. J. Expert Rev. Proteomics 2007, 4, 175–186. (16) Berna, M.; Ott, L.; Engle, S.; Watson, D.; Solter, P.; Ackermann, B. Anal. Chem. 2008, 80, 561–566. (17) Berna, M.; Schmalz, C.; Duffin, K.; Mitchell, P.; Chambers, M.; Ackermann, B. Anal. Biochem. 2006, 356, 235–243. (18) Berna, M. J.; Zhen, Y.; Watson, D. E.; Hale, J. E.; Ackermann, B. L. Anal. Chem. 2007, 79, 4199–4205. (19) Zhen, Y.; Berna, M. J.; Jin, Z.; Pritt, M. L.; Watson, D. E.; Ackermann, B. L.; Hale, J. E. Proteomics Clin. Appl. 2007, 1, 661–671. (20) Oe, T.; Ackermann, B. L.; Inoue, K.; Berna, M. J.; Garner, C. O.; Gelfanova, V.; Dean, R. A.; Siemers, E. R.; Holtzman, D. M.; Farlow, M. R.; Blair, I. A. Rapid Commun. Mass Spectrom. 2006, 20, 3723–3735.

format have been reported to improve assay throughput,21 to our knowledge no examples have been reported for LC/MS/MS assays, particularly for the quantification of low-abundance proteins. Accordingly, we refer to this technique as immunoprecipitation in ELISA format or IPE. The procedure is further streamlined through the use of microwave-assisted digestion, an established method for increasing the rate of proteolysis using enzymes such as trypsin.22-26 Optimization of IPE is illustrated by the use of two proteins for which quantitative assays have previously been reported by our laboratory.16,18 We further show that, using IPE sample preparation, analysis by LC/MS/MS can occur in a single day for proteins present in plasma at subnanogram per milliliter concentrations. EXPERIMENTAL SECTION Apparatus. The Quantum Ultra triple quadrupole mass spectrometer was from Thermo Finnigan (San Jose, CA), the LC10-ADvp HPLC system was from Shimadzu (Columbia, MD), and the HTS PAL autosampler was from Leap Technologies (Carrboro, NC). Microwave-assisted proteolytic digestion was accomplished using a CEM Discover microwave system (Matthews, NC). Chemicals and Reagents. Formic acid (88%), dimethyl sulfoxide (DMSO), HPLC grade water, methanol, and acetonitrile were from Thermo Fisher Scientific (Fair Lawn, NJ). Bovine serum albumin (BSA), ethylenediaminetetraacetic acid (EDTA), sodium chloride, ammonium bicarbonate, phosphate-buffered saline (PBS), Triton X-100, Tris-HCl, and protein-G′ were from Sigma-Aldrich (St. Louis, MO). Maxisorp 96-well ELISA plates were from Nunc (Roskilde, Denmark). Disuccinimidyl suberate (DSS) and Seize Protein A/G 96-well plates were from Pierce (Rockford, IL). Trypsin Gold was purchased from Promega (Madison, WI), whereas human and rat sera were from Bioreclamation (Long Island, NY). Rat NTproBNP (H-HPLGSPSQSP EQSTMQKLLE LIREKSEEMA QRQLSKDQGP TKELLKRVLROH) and Myl3 (Swissprot P16409) were expressed at Eli Lilly and Company (Indianapolis, IN) and were characterized using LC/ MS and BCA assay. The stable isotope-labeled peptide (H-ALGQNPTQAEV[2H8]LR-OH) used as the internal standard for Myl3, and stable isotope-labeled NTproBNP containing the tryptic fragment (H-LLELI[13C615N1]R-OH) used as the internal standard for NTproBNP, were synthesized by Midwest BioTech (Fishers, IN); all peptides were characterized by LC/ MS and amino acid analysis. Rabbit polyclonal antibodies (2AG and 3AG) to rat proBNP were from Invitrogen (Carlsbad, CA), and the mouse monoclonal antibodies to rat Myl3 (MLM520) were from Abcam (Cambridge, MA). LC/MS/MS Conditions. The LC/MS/MS conditions used for the analysis of Myl318 and NTproBNP16 were previously published and are described in detail including column switching (21) Asthagiri, A. R.; Horwitz, A. F.; Lauffenburger, D. A. Anal. Biochem. 1999, 269, 342–347. (22) Vesper, H. W.; Mi, L.; Enada, A.; Myers, G. L. Rapid Commun. Mass Spectrom. 2005, 19, 2865–2870. (23) Sun, W.; Gao, S.; Wang, L.; Chen, Y.; Wu, S.; Wang, X.; Zheng, D.; Gao, Y. Mol. Cell. Proteomics 2006, 5, 769–776. (24) Hua, L.; Low, T. Y.; Sze, S. K. Proteomics 2006, 6, 586–591. (25) Chen, W. Y.; Chen, Y. C. Anal. Chem. 2007, 79, 2394–2401. (26) Lill, J. R.; Ingle, E. S.; Liu, P. S.; Pham, V.; Sandoval, W. N. Mass Spectrom. Rev. 2007, 26, 657–671.

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

3951

diagrams. Briefly, for Myl3, online sample cleanup and desalting were achieved with column switching using an Analytical Sales and Services Sprite Armor C18 trap column (40 mm × 2 mm, 5 µm) and a Phenomenex CapCell-Pak C18 MG (2 mm × 35 mm, 5 µm) analytical column. The mobile phases (MP) used for the trap column were HPLC grade water (MP A) and acetonitrile (MP B) delivered at a flow rate of 0.75 mL/min, and those used for the analytical column were 0.1% formic acid (MP A) and 800:200:1 methanol/acetonitrile/formic acid (v/v/v, MP B) delivered at a flow rate of 0.5 mL/min. The trap column gradient profile was (min/% MP B) 0.0/0, 3.0/0, 3.0/100, 3.5/100, 3.5/0, 3.75/0 (end gradient), and the gradient profile for the analytical column (held at 50 °C) was (min/% MP B) 0.0/10, 1.0/10, 3.0/40, 3.0/10, 3.75/ 10 (end gradient). The column switching events were as follows: at 0 min, 50 µL was injected onto the trap column and the effluent was directed to waste, at 1.0 min, the trap and analytical columns were placed in series driven by the analytical column gradient, and at 3.0 min, the trap column effluent was redirected to waste, whereas that of the analytical column was directed to the mass spectrometer. The following positive electrospray ionization (ESI) mass spectrometric conditions were used: scan width 0.1 amu, scan time 0.05 s, Q1/Q3 resolution 0.7/0.7 amu, chrom filter 3 s, Q2 CID (Ar) 1.5 mTorr, spray voltage 4600, sheath gas (N2) 50 AU, aux gas (N2) 5 AU, sweep gas (N2) 5 AU, capillary temperature 350 °C, source CID voltage 5, and MS acquisition time 3.5 min. The SRM transitions (two for each analyte) were m/z 699.0 f 913.5 (CE 26 V, tube lens 130 V), m/z 699.0 f 1027.5 (CE 28 V, tube lens 130 V) for the 13-mer; and m/z 703.1 f 921.6 (CE 26 V, tube lens 124 V), m/z 703.1 f 1035.6 (CE 28 V, tube lens 124 V) for the SIL internal standard. For NTproBNP, online sample cleanup and desalting were achieved with column switching using a Thermo Scientific Sprite Aquasil C18 trap column (20 mm × 2 mm, 5 µm) and a Phenomenex CapCell-Pak C18 MG (1 mm × 35 mm, 3 µm) analytical column. The mobile phases used for the trap column were 0.1% formic acid (MP A) and acetonitrile (MP B) delivered at a flow rate of 0.5 mL/min, and those used for the analytical column were 0.1% formic acid (MP A) and 800:200:1 methanol/ acetonitrile/formic acid (v/v/v, MP B) delivered at a flow rate of 0.18 mL/min. The trap column mobile phases were delivered isocratically (5% MP B), whereas the gradient profile for the analytical column (held at 55 °C) was (min/% MP B) 0.0/10, 1.0/ 10, 4.0/55, 4.0/10, 4.25/10 (end gradient). The column switching events were as follows: at 0 min, 50 µL was injected onto the trap column and the effluent was directed to waste, at 1.0 min, the trap and analytical columns were placed in series driven by the analytical column gradient, and at 2.8 min, the trap column effluent was redirected to waste, whereas that of the analytical column was directed to the mass spectrometer. The following positive ESI mass spectrometric conditions were used: scan width 0.1 amu, scan time 0.1 s, Q1/Q3 resolution 0.7/0.9 amu, chrom filter 5 s, Q2 CID (Ar) 1.5 mTorr, spray voltage 3800, sheath gas (N2) 50 AU, aux gas (N2) 15 AU, capillary temperature 350 °C, source CID voltage 5, and MS acquisition time 4.25 min. The SRM transition was m/z 378.8 f 530.3 (CE 14 V, tube lens 117 V) for LLELIR and m/z 382.2 f 537.3 (CE 14 V, tube lens 117 V) for LLELI7R. The data from both assays were collected and analyzed using Xcalibur v2.0 software (Thermo Scientific). 3952

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

IPE Sample Preparation. For the Myl3 procedure, the wells of Maxisorp ELISA plates were coated with 5 µg of protein-G′ prepared in 200 µL of PBS. The plates were incubated overnight at 4 °C and could be stored for at least 1 month prior to use. Following incubation, the liquid layer was removed and 10 µg of anti-Myl3 (Abcam, Cambridge, MA) was added to each well in 200 µL of blocking buffer (50 mM Tris-HCl, pH 8; 150 mM NaCl; 2 mM EDTA; 0.1% BSA, w/v). The plate was incubated for 1 h at ambient temperature and washed with 200 µL of PBS per well. Anti-Myl3 was cross-linked to protein-G′ using 200 µL of DSS (Pierce, Rockford, IL, 1.3 mg/mL DMSO) per well followed by incubation at ambient temperature for 1 h. The wells were washed four times with 250 µL of blocking buffer and blotted dry. Myl3 serum samples (100 µL aliquots) were diluted with 100 µL of blocking buffer, and the full 200 µL was loaded to the ELISA plate followed by incubation at ambient temperature for 2 h. Next, the wells were washed three times with 250 µL of wash buffer (50 mM Tris-HCl, pH 8; 150 mM NaCl; 2 mM EDTA; 0.01% Triton X-100, v/v), and the plate was blotted dry. The samples were eluted with 250 µL of 5% acetic acid (v/v) for 3 min. The acetic acid was transferred to Eppendorf centrifuge tubes and concentrated to dryness using a Savant. The dried residues were reconstituted with 100 µL of digest buffer (10% methanol in 100 mM ammonium bicarbonate (pH 8.5), v/v) containing 200 ng of trypsin and 712 pM (final concentration) of internal standard, and the samples were digested using a Discover microwave system (CEM, Matthews, NC) for 50 min at 55 °C and 50 W. Following digestion, 50 µL was injected for analysis. For NTproBNP, the wells of a Seize Protein A/G 96-well plate (Pierce) were washed three times with 200 µL of PBS. The wells of the plate were coated with 4 µg of anti-proBNP and were incubated at ambient temperature for 1 h. Next, the wells were washed three times with 200 µL of PBS, and the anti-proBNP was cross-linked to protein A/G using 200 µL of DSS (1.3 mg/mL DMSO) per well followed by incubation at ambient temperature for 1 h. The wells were washed again three times with 200 µL of PBS, and 100 µL blocking buffer, 100 µL of rat serum, and 1 ng of NTproBNP SIL internal standard were added to each well. The samples were incubated at ambient temperature for 2 h and were washed three times with 200 µL of wash buffer. NTproBNP was eluted from the wells with 200 µL of 5% acetic acid (v/v) for 3 min, and the acetic acid was concentrated using a Savant. The dried residues were reconstituted with 100 µL of digest buffer containing 1 µg of trypsin. The samples were digested using the Discover microwave system for 30 min at 55 °C and 50 W. Following digestion, 50 µL was injected for analysis. Protein Digestion. During method development, two parameters were found to be important for efficient microwave-assisted proteolytic digestion: enzyme-to-substrate ratio (E/S, weight trypsin/weight protein) and digestion time. Microwave energy of 50 W and temperature of 55 °C were optimal for trypsin and appear to be independent of substrate based on several different proteins (unpublished data). An experiment was conducted for Myl3 to find the optimal E/S and microwave digestion time. A solution of 26 ng/mL Myl3 was prepared in digest buffer and spiked with SIL internal standard (ALGQNPTQAEV[2H8]LR) at 1.5 ng/mL to correct for analytical variation. Next, trypsin was added at E/S ratios ranging from

1:100 to 1000:1, and the samples were placed in the microwave (50 W, 50 °C) for 60 min. Aliquots (50 µL) were taken for analysis at 10 min intervals ending at 60 min. Initial method development demonstrated that NTproBNP was difficult to digest using trypsin. Therefore, E/S was optimized first followed by digestion time at the optimal E/S ratio. A solution of 2.5 ng/mL NTproBNP was prepared in digest buffer that contained 200 pg/mL of the peptide LLELI[13C615N1]R that was added as an internal standard to correct analytical variability. Next, trypsin was added to aliquots of this solution to span the range of E/S (w/w) from 1:100 to 4000:1. The samples were placed in the microwave and digested for 30 min at 50 W and 55 °C. Following digestion, 50 µL aliquots were injected for analysis. Following the E/S optimization, microwave digestion time was optimized by preparing a second set of samples, but at the optimal E/S ratio of 4000. In this experiment, the samples were digested for 60 min at 50 W and 55 °C, and 50 µL aliquots were removed for analysis at 10 min intervals starting and 10 and ending at 60 min. In order to demonstrate the efficiency of microwave-based digestion, the optimization of E/S was repeated for Myl3 and NTproBNP using an oven-based procedure at 37 °C. Samples were prepared in the same fashion as for the microwave E/S experiment; however, digestion was performed in the oven overnight (∼15 h). In addition to the 24 h experiment, aliquots of the Myl3 digests were removed for analysis after 4 h to demonstrate that digestion could not be performed efficiently within a given day. This experiment was not repeated for NTproBNP, which had only 64% absolute recovery after 15 h of digestion in the oven. Validation Procedures. The accuracy and precision of measuring Myl3 and NTproBNP using IPE and microwave-based digestion were evaluated by performing a fit-for-purpose intraday validation. The results of the validations were compared to previously published results for these proteins using traditional IP and oven-based digestion. For Myl3, standard samples were prepared in duplicate at 0.95, 4.0, 7.0, 16, 26, 36, and 56 ng/mL by supplementing control rat serum (0.95 ng/mL endogenous Myl3) with Myl3. Five replicates of validation samples were prepared in the same fashion at 3.95, 16, and 56 ng/mL. For NTproBNP, standard samples were prepared in duplicate at 0.1, 0.25, 0.50, 0.75, 1.0, and 2.5 ng/mL by spiking NTproBNP into control human serum; the use of human serum as a surrogate matrix for rat serum was previously validated.16 Likewise, validation samples (five replicates) were prepared at 0.10, 0.50, and 2.5 ng/mL. IPE was performed using the procedures outlined in the IPE Sample Preparation section, and microwave-assisted digestion was performed at an E/S of 4000 for NTproBNP and an E/S of 1000 for Myl3. Both NTproBNP and Myl3 required a digestion time of 50 min. Following digestion, duplicate standard curves (front and back) were analyzed along with five replicates of each of the validation concentrations. RESULTS Microwave-Assisted Digestion. The optimization of E/S and microwave digestion time are presented in Figure 1 for Myl3. Using the microwave, optimal digestion efficiency occurred at E/S 1000 for 50 min, which corresponded with an absolute recovery of ∼79%. Absolute recovery takes into account IPE and digestion

Figure 1. Optimization of Myl3 digestion time and enzyme/substrate ratio (weight trypsin/weight Myl3) using the CEM microwave are presented. The E/S ratio and microwave time were varied and plotted against LC/MS percent response of the resulting tryptic fragment (ALGQNPTQAEVLR). Each point represents the mean of triplicate measurements. Key: 10 min (b), 20 min (O), 30 min (2), 40 min (4), 50 min (9), 60 min (0).

Figure 2. Optimization of Myl3 E/S ratio (weight trypsin/weight Myl3) in an oven at 37 °C overnight (∼15 h). E/S ratio was plotted against LC/MS percent response of the resulting tryptic fragment (ALGQNPTQAEVLR). Each point represents the mean of triplicate measurements.

losses and was estimated by comparison to the SIL peptide spiked after sample processing was complete. The optimal E/S for the oven-based procedure (15 h at 37 °C) was 100 (Figure 2) and had a similar absolute recovery of ∼80%. The kinetics with respect to E/S ratio could not be evaluated for NTproBNP as the digestion was incomplete even at the highest practical ratio (4000); this held true for both the microwave and oven-based procedures. As a result, the kinetic profile (E/S vs time) was flat for all but the maximum E/S and time, and the highest practical E/S (4000) and time (50 min) were used for the microwave-based digestion of NTproBNP. Likewise, an E/S of 4000 was used for oven-based digestion overnight at 37 °C. The absolute recovery of the microwave procedure (E/S 4000 for 50 min) was ∼65%, whereas that of the published, oven-based procedure was ∼63%.16 Validation Experiments. To determine the utility of using IPE and microwave-assisted digestion for future IA-LC/MS/MS assays, intraday validations were performed for Myl3 and NTAnalytical Chemistry, Vol. 81, No. 10, May 15, 2009

3953

Table 1. Intraday Validation Statistics for Myl3 Spiked into Rat Serum (Containing 0.95 ng/mL Endogenous Myl3) and Analyzed by LC/MS/MS following IPE and Microwave-Assisted Digestion validation sample concentration (ng/mL) statistic mean (ng/mL) accuracy (%RE) precision (%CV) n

3.95

16.0

56.0

3.48 -11.8 10.4 5

17.4 8.8 10.7 5

52.6 -6.1 1.9 5

Table 2. Intraday Validation Statistics for NTproBNP Spiked into Human Serum and Analyzed by LC/MS/MS following IPE and Microwave-Assisted Digestion validation sample concentration (ng/mL) statistic

0.10

0.50

2.5

mean (ng/mL) accuracy (%RE) precision (%CV) n

0.10 2.5 6.2 5

0.49 -2.1 7.6 5

2.4 -2.2 6.9 5

proBNP. Myl3 validation samples were analyzed at concentrations of 3.95, 16.0, and 56.0 ng/mL (five replicates each) and had precision (%CV) that did not exceed 11% and accuracy (%RE) that was within 12% (Table 1). Similarly, NTproBNP validation samples were prepared and analyzed at 0.1, 0.5, and 2.5 ng/mL and showed good precision (%CV < 8%) and accuracy (%RE within 3%) (Table 2). With the use of this approach, lower limits of quantification (LLOQ) of 100 pg/mL and 0.95 ng/mL were obtained for NTproBNP and Myl3, respectively. Extracted ion chromatograms for NTproBNP from a blank sample with internal standard and from the LLOQ are presented in Figures 3 and 4. DISCUSSION It can be argued that verification has become rate-limiting in protein biomarker development. This statement reflects the time required to develop and validate assays for low-abundance proteins in accessible fluids versus the facility with which hypotheses can be generated and confirmed using standard proteomic approaches. Previously, we proposed a strategy based on IA-LC/MS/MS to address this bottleneck;15 however, two gaps exist. The first is the need for improved IA-LC/MS/MS sensitivity relative to ELISA. As demonstrated by NTproBNP, the limits of quantification for protein biomarkers using IA-LC/MS/MS can reach 100 pg/ mL, whereas those for ELISA are in the 1-10 pg/mL range. We are currently investigating the use of nano-LC/MS/MS to address this shortcoming. The second gap is the relatively low sample throughput of LC/MS/MS compared to ELISA. Reasons for low throughput include the use of individual tubes instead of plates, the need for a lengthy digestion step, and the fact that LC/MS/MS detection is serial (i.e., one sample at a time). In contrast, ELISA is a parallel technique performed in 96-well format, which measures intact proteins. Therefore, the goal of the present investigation was to establish methodology for IP in 96-well format to take advantage of parallel sample processing. In addition, microwave-based protein digestion was added to further reduce the time for sample prepara3954

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

Figure 3. Representative extracted ion chromatogram obtained from an extracted blank serum sample spiked with internal standard. The tryptic fragment LLELI7R resulted from the digestion of the stable isotope-labeled NTproBNP internal standard. The scale of the top pane (analyte) is ∼1% that of the bottom pane (internal standard) to highlight the low level of background noise.

Figure 4. Extracted ion chromatogram obtained from the analysis of a spiked human serum sample at the lower limit of quantification (100 pg/mL). The tryptic fragments from the digestion of NTproBNP (LLELIR) and the internal standard (LLELI7R) are indicated. The top pane presents the analyte, and the bottom pane presents the internal standard.

tion. Although 96-well capable microwave systems are not yet commercially available, the present system allows complete sample preparation to occur within a single day. Myl3. The two most important factors affecting microwaveassisted digestion efficiency are enzyme/substrate ratio and digestion time. Initial method development showed that these parameters were independent, and therefore optimization of E/S was required for all microwave time points. Using the microwave, an E/S of 1000 for 50 min gave approximately the same absolute recovery (80%) of the tryptic peptide ALGQNPTQAEVLR as the

oven-based procedure, which used an E/S of 100 and required approximately 15 h of digestion time at 37 °C. It is possible that a lower E/S and longer microwave digestion time could yield digestion efficiency similar to the above conditions; however, the ultimate goal was to complete sample preparation within a typical work day. To establish that the oven-based procedure was not feasible for within-day sample analysis, triplicate aliquots were removed from the oven after only 4 h at 37 °C, and the absolute recovery of these samples was only 36%. Most importantly, the validation statistics obtained using IPE and the microwave-based procedure (RE within 11% and CV with 12%) were similar to the oven-based procedure (RE within 12.9% and CV within 13.2%).18 NTproBNP. Initial solution digestion experiments uncovered that NTproBNP was difficult to digest and required excess trypsin for both microwave and oven-based digestion. This is most likely explained by the acidic residue (Glu) on the C-terminal side of the NTproBNP tryptic fragment LLELIR(E). The active site of trypsin contains an Asp residue that stabilizes the positive charge of Lys and Arg substrates during catalysis. The charge stabilization necessary for hydrolysis in the trypsin active site is probably disrupted by the negative charge of the proximal Glu residue, which significantly decreases the proteolytic efficiency of trypsin. For this reason, the highest practical E/S ratio (4000) was used to determine the optimal microwave digestion time, which was measured to be 50 min. Likewise, the oven-based procedure also required an E/S ratio of 4000. Even at an E/S of 4000, the absolute recovery of the tryptic fragment LLELIR was only ∼65% for microwave-based and ∼63% for the overnight, oven-based procedure.16 However, it should be noted that the validation statistics obtained using the microwave-based procedure (RE within 3% and CV within 8%) were slightly better than those obtained using the oven-based procedure (RE within 4% and CV within 15%) and that suboptimal digestion efficiency did not prevent a lower limit of quantification of 100 pg/mL from being reached. High Enzyme/Substrate Ratios. Results presented herein reveal the importance of high E/S for low-abundance protein quantification by IA-LC/MS/MS. For Myl3, the optimum E/S using oven-based digestion was 100, representing a 2000-fold excess over what is commonly used for standard tryptic digestion (i.e., E/S 1:20, w/w). This difference can be explained by the relative substrate concentration. The substrate concentration for a standard tryptic digestion is in the range of 10 µg/mL, or roughly 4 orders above the concentration of low-abundance proteins analyzed by IA-LC/MS/MS (i.e., 1 ng/mL). In other words, in order to achieve a comparable probability for enzyme/substrate encounter, excess trypsin on a w/w basis is needed to compensate for the finite level of target protein present in the sample. As illustrated by Myl3, the amount of trypsin required (w/w) scales inversely with the amount of target protein present. The results for Myl3 further indicate that an even higher E/S (i.e., 1000) is needed for optimum microwave-based digestion. Given the fact that microwave-based digestion occurs in about 1/10 the time relative to oven-based digestion, it can be concluded that more trypsin is needed to maintain a comparable number of enzyme/ substrate encounters in 1/10 the duration. In hindsight, the overall process of optimizing E/S can be simplified by changing from units of w/w to enzyme concentration (i.e., millimolar).

Table 3. Timelines for Conventional IP with Oven-Based Digestion and IPE with Microwave-Based Digestion Are Compared activity wash beads or plate wells conjugate antibody (Ab) to protein A/G beads or plate wash to remove unbound Ab cross-link Ab to protein A/G, incubate, and wash incubate samples containing target protein with Ab support wash to remove matrix components elute target protein from Ab, concentrate eluent reconstitute with digest buffer, add protease, digest instrument setup, equilibration LC/MS/MS analysis (60 samples)

conventional IP (h)

IPE (h)

(day 1) 1 2-4

(day 1) 0.25 1

1 2

0.25 1.25

2-overnight (day 2) 1.5

(day 1) 0.25

2-3

2

4-overnighta

2.5b

(day 3) 2

(day 1) c

6

overnight

(day 4)

(day 2)

results available a Myl3 had only 36% recovery after 4 h of digestion. b The optimum digestion time is 50 min; however, the CEM microwave has limited capacity and requires three runs for 60 samples. c Instrument setup and equilibration occurs during incubation steps.

An important logistical point involving the use of elevated protease concentrations is the likelihood of MS ionization suppression resulting from protease autolysis. This issue was mitigated for both assays by using modified trypsin to reduce autolysis and through the use of column switching to perform online cleanup prior to detection. This hypothesis is supported by both the validation statistics (i.e., the desired LLOQs were reached with good accuracy and precision) and from an evaluation of selectivity, which demonstrated the absence of interfering peaks in the blank serum samples. We are currently exploring the utility of microwavebased digestion using unmodified proteases. Comparison of Traditional IP and IPE. One of the main advantages of IPE relative to traditional IP is the use of multichannel pipettes and plate washers to reduce the time for numerous transfer and wash steps. For a typical batch of 60 samples, it takes an hour to perform a single wash for the tubebased procedure using a single-channel pipet and less than 15 min for the plate-based procedure. The time saved becomes significant over the course of the procedure, which requires a sample transfer and/or wash step at every stage. A second advantage is the time saved when using microwave-assisted proteolysis. For the proteins cited here, it was not feasible to achieve efficient digestion within a single day using an oven, while only 50 min (2.5 h for 60 samples) was required for the IPE procedure. The use of IPE and microwave-assisted proteolysis to reduce the time between iterations has substantially reduced method development time. As a result of these changes, the sample throughput for the protein biomarkers Myl3 and NTproBNP has increased by reducing the time required to prepare and analyze a batch of samples from 3 days to 1 (Table 3). Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

3955

CONCLUSIONS In this article, results were presented demonstrating the ability to streamline sample processing associated with low-abundance protein biomarker quantification by IA-LC/MS/MS. IP conducted in 96-well ELISA format, referred to herein as IPE, was shown to significantly reduce the time required for sample preparation compared to conventional bead formats involving single tubes. The process was further accelerated by the incorporation of microwave digestion. Digestion conditions were optimized for two rat cardiotoxicity biomarkers, which revealed the need for elevated enzyme-to-substrate ratios for improved recovery and speed. Quite importantly, single-day validations were conducted for both markers, which showed comparable or improved validation

3956

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

statistics to previously published results obtained using conventional IP sample preparation and digestion. While further investigation and optimization continues, we have adopted IPE and microwave digestion as the default methodology for IA-LC/MS/ MS protein assays; however, it remains to be seen if this approach will be suited to all protein assays. The use of IPE with microwaveassisted digestion is part of a larger effort by our laboratory to reduce the gap between MS and ELISA for the purpose of lowabundance protein biomarker verification. Received for review February 5, 2009. Accepted April 3, 2009. AC9002744