Anal. Chem. 2007, 79, 4199-4205
Strategic Use of Immunoprecipitation and LC/MS/MS for Trace-Level Protein Quantification: Myosin Light Chain 1, a Biomarker of Cardiac Necrosis Michael J. Berna,* Yuejun Zhen, David E. Watson, John E. Hale, and Bradley L. Ackermann
Lilly Research Laboratories, Eli Lilly and Company, Greenfield, Indiana 46140
Myosin light chain 1 (Myl3) is a 23-kDa isoform of one of the subunits of myosin, a protein involved in muscle contraction. Myl3 is presently being studied as a biomarker of cardiac necrosis to predict drug-induced cardiotoxicity, and in the work presented here, an LC/MS/MS assay was developed and validated to measure Myl3 in rat serum. The key steps in this approach involved immunoaffinity purification of Myl3 from serum followed by on-bead digestion with trypsin to release a surrogate peptide. This tryptic peptide was quantified using a synthetic peptide standard and a corresponding stable isotope-labeled internal standard, and the results were stoichiometrically converted to Myl3 serum concentrations. Myl3 concentrations were corrected for peptide recovery following immunoprecipitation and digestion (85%) and showed excellent agreement with synthetic peptide standards. Both the synthetic peptide and HisMyl3 protein were used to evaluate assay accuracy (% RE) and precision (% CV), which were measured on each of 3 days. The synthetic peptide was evaluated over the range of 0.073-7.16 nM, while Myl3 protein QC samples prepared in rat serum were evaluated over the range of 0.13-6.62 nM. To prepare control matrix, endogenous Myl3 was immunodepleted from pooled rat serum. Peptide interday accuracy and precision did not exceed 7.6 and 11.1%, and Myl3 interday accuracy and precision did not exceed 12.9 and 13.2%, respectively. Data are presented from the application of this assay to establish a time course in which rats demonstrated a marked increase in Myl3 serum concentrations following administration of isoproterenol, a β-adrenergic receptor agonist known to induce cardiac injury. This assay is an example of a larger effort in our laboratory to use LC/MS/MS in conjunction with immunoaffinity techniques to evaluate candidate biomarkers of target organ toxicity and to expedite the development of biomarker assays for drug development. The future of the biotechnology and pharmaceutical industries relies in part on their ability to address the rising costs associated with drug discovery and development while continuing to provide * Corresponding author. E-mail:
[email protected]. 10.1021/ac070051f CCC: $37.00 Published on Web 04/21/2007
© 2007 American Chemical Society
valuable medications to the public. It has been reported that a drug entering phase I testing has only an 8% chance of reaching the market,1 while the cost of developing a new drug has been estimated to be as high as $1.7 billion U.S.2 One of the near-term opportunities to address rising drug costs is the use of biomarkers to evaluate drug efficacy, monitor disease progression, and predict drug-related toxicity. In particular, the ability to predict druginduced tissue injury in animals can result in decreased costs by diverting resources from nonviable drug candidates toward candidates that have a greater probability of clinical success. Biochemical markers have been used for decades to provide a noninvasive means of diagnosing cardiac injury.3 Numerous examples exist in the literature describing the use of myocardiumspecific proteins as markers of congestive heart failure, myocardial infarction, cardiac necrosis, and other heart-related ailments.4-6 Examples of cardiac markers include the troponins (cTnT, cTnI), cardiac natriuretic peptides (BNP, ANP), creatinine kinase isoenzyme MB (CK-MB), heart fatty acid binding protein (hFabp3), and myosin light chain (Myl3).7-9 By evaluating the utility of proven clinical cardiac markers in veterinary species, a panel of biomarkers can be identified to diagnose cardiac injury in studies ranging from early discovery through clinical evaluation. Dolci and Panteghini outlined the key characteristics of an ideal cardiac biomarker as high sensitivity (early onset, high concentration, slow clearance), specificity (only present in target tissue, not detected in healthy subjects), good analytical characteristics, and good clinical characteristics (ability to influence patient therapy and outcome).7 Owing to the difficulty in finding a single marker having the cumulative characteristics outlined by Dolci and Panteghini, along with the potential for multiple mechanisms for (1) Koop, R. Drug Discovery Today 2005, 10, 781-788. (2) Gilbert, J.; Henske, P.; Singh, A. In vivo: Business Med. Rep. 2003, 21, 73. (3) Herrmann, J.; Volbracht, L.; Haude, M.; Eggebrecht, H.; Malyar, N.; Mann, K.; Erbel, R. Med. Klin. 2001, 96, 144-156. (4) Sugiura, T.; Takas, H.; Toriyama, T.; Goto, T.; Ueda, R.; Dohi, Y. J. Card. Failure 2005, 11, 504-509. (5) Ravkilde, J. Dan. Med. Bull. 1998, 11, 34-50. (6) Solymoss, B. C.; Bourassa, M. G.; Wesolowska, E.; Dryda, I.; Theroux, P.; Mondor, L.; Perrault, D.; Gilfix, B. M. Clin. Cardiol. 1997, 20, 934-942. (7) Dolci, A.; Panteghini, M. Clinica Chimica Acta 2006, 369, 179-187. (8) Naraoka, H.; Kyoko, I.; Suzuki, M.; Naito, K.; Tojo, H. Clin. Chim. Acta 2005, 361, 159-166. (9) Zhen, Y.; Berna, M. J.; Jin, Z.; Watson, D. E.; Ackermann, B. L.; Hale, J. E. The 54th ASMS Conference on Mass Spectrometry, Seattle, WA, May 29June 1, 2006.
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007 4199
necrosis, the need to establish a panel of biomarkers is clear. The benefits of this strategy are apparent when one considers the advantages and shortcomings of the individual markers. The cardiac troponins are perhaps the gold standard for cardiac injury with respect to specificity, but have a narrow kinetic profile in rat serum due to rapid clearance. In a recent study in our laboratory that used isoproternol to induce cardiac necrosis, cardiac troponin I (cTnI) levels peaked rapidly at 8 h postdose (13.8 ng/mL) and returned to 5 ng/mL after only 24 h (unpublished results). Rat hFabp3 is present in high levels in serum following cardiac damage, but has a kinetic profile similar to the troponins and can be affected by skeletal muscle injury. Myl3 has a superior kinetic profile (see Results and Discussion); however, as with hFabp3, Myl3 is less selective than the cardiac troponins and can be affected by skeletal muscle injury. The advantages of using Myl3 in a cardiac biomarker panel are its broad kinetic window, high sensitivity, excellent analytical characteristics, commercially available antibodies to support ELISA assays, and transferability to the clinic. Historically, ELISA has been the analytical technique of choice for protein quantification due to its high throughput and sensitivity; however, LC/MS/MS is increasingly being used for protein quantification.10-15 Selectivity and method development speed are key advantages of LC/MS/MS, which can be used to quickly evaluate biomarker biology, triage candidate biomarkers for costly ELISA development, and cross-validate ELISA selectivity. Using the LC/MS/MS approach, surrogate peptides of the proteins of interest are produced through proteolysis (in vivo) or enzymatic digestion (ex vivo) and are quantified using stable isotope-labeled peptides as internal standards. An example of this approach was reported by Barnidge et al. and involved measuring prostatespecific antigen.12 While LC/MS/MS with selected reaction monitoring (SRM) detection is common among these methods, the approach taken for sample preparation varies significantly according to the complexity of the sample matrix and the basal level of the target protein. Strategies for using LC/MS/MS to quantitate members of specific protein classes have also been published. This includes methodology by Gerber et al.13 for phosphoproteins and a generalized strategy by Anderson et al.14 that uses polyclonal antibodies to capture and analyze targeted peptides following enzymatic digestion. Using polyclonal antibodies to targeted peptides, Anderson et al. reported that it was possible to enrich a target peptide by 100-1000-fold.14 In our laboratory, we adopted an analogous strategy, which uses antibodies for the target protein to facilitate isolation prior to enzymatic digestion. Among several advantages, this methodology facilitates the rapid development of ELISA assays since it uses similar reagents, and it can provide preliminary evaluation of novel (10) 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, 10, 1676-1682. (11) Kuhn, E.; Wu, J.; Karl, J.; Liao, H.; Zolg, W.; Guild, B. Proteomics 2004, 4, 1175-1186. (12) Barnidge, D. R.; Goodmanson, M. K.; Klee, G. G.; Muddiman, D. C. J. Proteome Res. 2004, 3, 644-652. (13) Gerber, S. A.; Rush, J.; Stemman, O.; Kirschner, M. W.; Gygi, S. P. Proc. Natl. Acad. Sci. U.S.A. 2003, 12, 6940-6945. (14) Anderson, N. L.; Anderson, N. G.; Haines, L. R.; Hardie, D. B.; Olafsom, T. W.; Pearson, T. W. J. Proteome Res. 2004, 3, 235-244. (15) Berna, M.; Schmalz, C.; Duffin, K.; Mitchell, P.; Chambers, M.; Ackermann, B. Anal. Biochem. 2006, 356, 235-243.
4200
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
biomarkers prior to full-scale investment into a time-consuming ELISA assay. In addition, because this strategy involves the close interplay between MS and ELISA methods, the selectivity of the antibodies used in the final ELISA assay can be verified by the enhanced selectivity of MS. In the work presented here, an assay based on liquid chromatography-tandem mass spectrometry was developed and validated to quantify Myl3 in rat serum. Myl3 protein was isolated from 100-µL sample aliquots by immunoprecipitation (IP), and following enzymatic digestion with trypsin, a 13-mer peptide was quantitated by LC/MS/MS in the positive-ion mode using multiple reaction monitoring detection. A practical example of the assay is presented in which elevated levels of Myl3 are detected over a 24-h time course in rat serum following administration of isoproterenol, an agent known to induce cardiac injury. EXPERIMENTAL SECTION Chemicals and Reagents. Formic acid (88%), HPLC acetonitrile, and HPLC methanol were obtained from MallinckrodtBaker (Paris, KY), and HPLC water was acquired from Fisher Scientific (Fair Lawn, NJ). Bovine serum albumin (1% w/v, pH 7), EDTA (0.5 M, pH 8), sodium chloride (5 M), sodium biocarbonate, and Tris-HCl buffer (1 M, pH 8) were obtained from Sigma-Aldrich (St. Louis, MO). Dynal beads (M-280 streptavidin) were purchased from Invitrogen (Carlsbad, CA), the Sulfo-NHSLC-biotin antibody biotinylation kit was from Pierce (Rockford, IL), and the mouse monoclonal antibody to rat Myl3 was obtained from Abcam (product MLM520, Cambridge, MA). Sequence grade trypsin was purchased from Promega (Madison, WI), and control rat serum was from Bioreclamation (Long Island, NY). A 13-mer synthetic peptide to the Myl3 tryptic fragment (ALGQNPTQAEVLR) and associated stable isotope-labeled internal standard (SIL, ALGQNPTQAE(V-d8)LR) were purchased from Midwest BioTech (Fishers, IN). His-Myl3 protein was obtained from Eli Lilly and Co. (Indianapolis, IN). Buffer 1 was prepared by diluting the solutions obtained from Sigma-Aldrich to a final concentration of 50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, and 0.1% bovine serum albumin and adjusting the pH to 7.5. Buffer 2 was prepared in the same fashion but did not contain bovine serum albumin. Due to the presence of endogenous Myl3 in rat serum, immunodepleted (ID) rat serum was used as the control matrix during the validation and sample analysis. Following dilution (1:5) with buffer 1, endogenous Myl3 was removed from control rat serum by IP using M-280 dynal beads and the biotinylated MLM520 monoclonal antibody. Consequently, 500 µL of ID rat serum was transferred for every 100-µL aliquot of naı¨ve serum that was required for analysis. The impact of diluting the control serum was normalized by diluting the study samples in the same fashion prior to IP. Isoproterenol Study Design. Five groups of five SpragueDawley rats were given a single subcutaneous dose of 50 mg/kg isoproterenol; a sixth group of five rats received vehicle consisting of 0.9% sodium chloride. To facilitate multiple biomarker measurements, terminal bleeding of the five groups occurred at 1, 2, 4, 8, and 24 h, and the blood from each of the groups was pooled into a single sample for each time point. Whole blood was collected from the vehicle group at 4 h postdose. The whole blood was allowed to coagulate for 30 min at room temperature, centrifuged,
Figure 1. Product ion spectrum of the 13-mer (ALGQNPTQAEVLR), indicating peptide fragmentation and predominant y ions.
and the serum fraction was collected and stored frozen at -70 °C until the day of analysis. Sample Preparation. To facilitate LC/MS/MS detection, a tryptic fragment (13-mer) of Myl3 was used as a surrogate measure of Myl3 serum concentrations, taking advantage of their stoichiometric relationship. The 13-mer was used to prepare standard samples, which were analyzed in duplicate at 17.9, 7.16, 2.87, 1.15, 0.458, 0.183, and 0.0734 nM. The 17.9 nM standard was prepared by spiking 2.5 µL of a 7.16 µM solution into 1 mL of 10% methanol/50 mM ammonium bicarbonate/712.3 pM SIL (digest buffer). The remaining standards were prepared by serial dilution with digest buffer containing SIL. Peptide (13-mer) QC samples were prepared to evaluate accuracy and precision during the validation at 0.0734, 1.15, and 7.16 nM using the same procedure to prepare the standard samples. Both the synthetic peptide and His-Myl3 protein were corrected for potency using amino acid analysis (Midwest BioTech, Fishers, IN). Double blank (no SIL or 13-mer) and blank (no 13-mer) samples were analyzed with each set of standard samples to verify the absence of peaks that might interfere with the detection of the 13-mer and SIL. Duplicate blank samples were also placed after the highest 13-mer standard sample to evaluate system carryover. Protein QC pools were prepared at 0.132, 1.32, and 6.62 nM and were used to evaluate accuracy and precision during the validation and as a quality control measure during study sample analysis. The 6.62 nM pool was prepared by spiking 28.4 µL of a 186.4 nM solution of Myl3 into 4 mL of ID rat serum. The 1.32 and 0.132 nM pools were prepared by serial dilution of the 6.62 nM pool with ID rat serum. Aliquots of 500 µL of ID rat serum (i.e., 100 µL of serum) were used for sample analysis. Reagent blank, ID rat serum blank, and rat serum blank (no Myl3 depletion) samples were analyzed during each day of the validation to verify the absence of contamination, the absence of endogenous Myl3, and the basal level of Myl3, respectively.
Aliquots (100 µL) of these samples were taken through the IP and digestion steps prior to LC/MS/MS analysis. Immunoprecipitation and Digestion. The M-280 magnetic beads were washed two times with 1 mL of buffer 1 prior to use. On day 1 of the IP procedure, the appropriate mass of biotinylated antibody was added to the M-280 beads at a ratio of 8 µg of antibody/mg of beads. The beads were incubated at room temperature for at least 4 h while rotating. Following incubation, residual unbound antibody was removed by washing two times with buffer 1, and the beads were resuspended with buffer 1 to a final concentration of 10 mg of beads/mL. Next, 0.5 mg (50 µL) of antibody complexed beads were added to a 100-µL aliquot of each rat serum sample, and the total volume was adjusted to 1 mL with buffer 1. The serum samples were incubated/rotated overnight at 4 °C. On day 2, the sample beads were washed 2 times with 1 mL of buffer 2 and were resuspended with 50 µL of 10% methanol in 50 mM ammonium bicarbonate that contained 712 pM SIL internal standard and 2 µg/mL trypsin. The samples were allowed to digest overnight at 37 °C while rotating. Following digestion, the bead supernatants were recovered using a magnet, and 20 µL was injected for analysis. Mass Spectrometric Conditions. Mass spectrometric detection was accomplished using a Finnigan TSQ Quantum Ultra (San Jose, CA) operated in positive ESI mode. The following mass spectrometric settings were used: scan width 0.1 amu, scan time 0.05 s, Q1/Q3 widths 0.7/0.9 amu, chrom filter 3 s, Q2 CID (Ar) 1.5 mTorr, spray voltage 4600, sheath gas (N2) 50 psi, auxiliary gas (N2) 5 psi, sweep gas (N2) 5 psi, capillary temperature 350 °C, quad MS/MS bias -3.5 V, source CID 5 V, and MS acquisition time of 3.5 min. Selected reaction monitoring was used to detect the 13-mer and SIL. In order to improve precision and S/N, two SRM transitions were summed for each analyte as follows: 13-mer, 699.0 f 913.5 (CE 26), 1027.5 (CE 28), tube lens 130; SIL, 703.1 f 921.6 (CE 26), 1035.6 (CE 28), tube lens 124. The product ion spectrum of the 13-mer is presented in Figure 1. Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
4201
Table 1. Intra- and Interday Validation Statistics for the 13-Mer Synthetic Peptide validation sample concentration (nM) day 1 Figure 2. Column-switching diagram presented in the inject position. At 0 min, 20 µL of sample is injected onto the trap column and diverted to waste. At 1.0 min, 13-mer and SIL are eluted from the trap column and through the analytical column driven by a linear gradient. At 3.0 min, the switching valve is placed back in the inject position; the trap column is rinsed for 0.5 min with acetonitrile prior to returning to initial conditions for the next injection.
Chromatographic Conditions. Chromatographic separation was performed using a Shimadzu HPLC system (WoodDale, IL) that consisted of four LC10AD pumps, two SCL-10Avp controllers, and an HTS PAL Leap autosampler (Carrboro, NC). Sample desalting and cleanup steps were performed online using column switching and a Sprite Armor C18 trap column (Analytical Sales & Services). The following mobile phases (MP) were used: MP A was HPLC grade water, MP B was HPLC acetonitrile, MP C was 1000:1 HPLC water/88% formic acid (v/v), and MP D was 800:200:1 methanol/acetonitrile/88% formic acid (v/v/v). A CapCell-Pak C18 MG (5 µm, 2 × 35 mm) analytical column was used following the trap column for chromatographic separation (Phenomenex). The HPLC separation employed column switching (Figure 2), which occurred in a three-step process. In step one, 20 µL of sample was injected onto the trap column (ambient temperature) using MP A delivered at 0.75 mL/min. The eluent was delivered to waste. In step two, at 1.0 min, the trap and analytical columns were placed in series, and a MP C and D linear gradient at 0.5 mL/min was used to elute the 13-mer and SIL into the mass spectrometer. The analytical column was held at 50 °C. In step 3, at 3.0 min, the switching valve was placed back into the inject condition, and MP B was used to briefly wash the trap column prior to the next injection. The trap column gradient profile was as follows (min/% MP B in A): 0.0/0, 3.0/0, 3.0/100, 3.5/100, 3.5/0, and 3.75/0. The analytical column gradient profile was (min/% MP D in C): 0.0/10, 1.0/switch valve, 1.0/10, 3.0/40, 3.0/10, 3.0/switch valve, and 3.75/10. Validation Experiments. The accuracy and precision of the synthetic peptide and His-Myl3 were evaluated on each of 3 days by analyzing five replicates of three concentrations. Validation of the 13-mer was performed at 0.0734, 1.15, and 7.16 nM, while validation of Myl3 was at 0.132, 1.32, and 6.62 nM; these concentrations represent the range of in vivo Myl3 concentrations. The validation samples were bracketed by duplicate standard curves that spanned the dynamic range of 0.0734-7.16 nM. The assay intra- and interday accuracy (percent relative error, % RE) and precision (percent coefficient of variation, % CV) were calculated and are presented in Table 1 for the 13-mer and in Table 2 for Myl3. Due to the presence of endogenous Myl3 in rat serum and the high degree of Myl3 homology across species, ID rat serum was used as the control matrix during the validation and study sample analysis. Assay selectivity with respect to the ID rat serum 4202 Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
2
3
all
statistic
0.0734
1.15
7.16
mean (nM) accuracy (% RE) precision (% CV n mean (nM) accuracy (% RE) precision (% CV) n mean (nM) accuracy (% RE) precision (% CV) n mean (nM) accuracy (% RE) precision (% CV) n
0.0810 10.4 10.1 5 0.0776 5.7 12.0 5 0.0784 6.8 11.2 5 0.0790 7.6 11.1 15
1.14 -0.9 8.5 5 1.14 -0.9 2.5 5 1.12 -2.6 8.4 5 1.13 -1.7 6.5 15
7.60 6.1 6.0 5 7.84 9.5 4.8 5 7.60 6.1 3.8 5 7.68 7.3 4.9 15
Table 2. Intra- and Interday Validation Statistics for Myl3 in Rat Serum validation sample concentration (nM) day 1
2
3
all
statistic
0.132
1.32
6.62
mean (nM) accuracy (% RE) precision (% CV) N mean (nM) accuracy (% RE) precision (% CV) N mean (nM) accuracy (% RE) precision (% CV) N mean (nM) accuracy (% RE) precision (% CV) n
0.138 4.5 8.5 5 0.142 7.6 5.7 5 0.168 27.3 25.4 5 0.149 12.9 13.2 15
1.09 -17.4 7.9 5 1.13 -14.4 5.0 5 1.30 -1.5 9.3 5 1.17 -11.4 7.4 15
5.96 -10.0 4.8 5 5.75 -13.1 3.2 5 6.80 2.7 6.4 5 6.17 -6.8 4.8 15
was evaluated during the validation through the analysis of ID rat serum blanks (no 13-mer, spiked SIL). In addition, the basal level of Myl3 in the control rat serum, prior to ID, was measured. Analyte carryover was evaluated during the validation by analyzing duplicate carryover blanks (no analyte, plus SIL) following the highest standard concentration. The percent absolute carryover and percent relative carryover (to the LLOQ) were calculated and are presented in the Results and Discussion. The analyte and SIL internal standard stock solutions were prepared in 50:50:0.1 water/acetonitrile/88% formic acid (v/v/v). These solutions were stored at ∼ -70 °C when not in use. The stability of the stock solutions was evaluated by comparing the results obtained from replicate injections made over time to an initial mean value obtained following the preparation of the solutions. Room-temperature matrix stability was evaluated in rat serum over a 24-h period. Rat serum was supplemented with Myl3 (212 pM final concentration), and five 100-µL aliquots were incubated for 24 h at room temperature. Following incubation, the stability samples were immunoprecipitated, digested, and analyzed by
LC/MS/MS. Room-temperature stability was evaluated by comparing the incubated samples to freshly prepared (T ) 0) samples. Long-term freezer stability was not evaluated during the validation, but will be evaluated during study sample analysis if the need for freezer storage arises. Extract stability was evaluated during each batch analysis by bracketing the validation and study samples with duplicate standard curves. Any degradation of the analyte over the course of the analysis would cause the slopes of the two standard curves to diverge. During the validation and sample analysis, calibration curves were obtained by plotting the peak area ratio of the 13-mer to its internal standard versus concentration. A weighted (1/concentration2) least-squares regression was used to obtain a linear equation over the range of the calibration graph. The origin was not used in the standard curve calculations. RESULTS AND DISCUSSION Method Development. An LC/MS/MS assay was developed for Myl3 after initial attempts to develop and validate a rat serum ELISA failed; ELISA was successfully employed to measure dog and primate concentrations. Immunoprecipitation was selected as the most suitable procedure to isolate Myl3 from rat serum for two reasons: (i) the use of a monoclonal antibody eliminated the need to remove abundant proteins prior to digestion and provided clean sample extracts which enhanced detection limits; (ii) by using the ELISA capture antibody to perform IP, the LC/MS/MS procedure could be used to cross-validate ELISA selectivity. In addition, the LC/MS/MS assay provided the means to investigate Myl3 biomarker biology in rat serum concomitant with working out the ELISA assay issues. ELISA is a parallel technique that is much more cost-effective for routine sample analysis than is LC/MS/MS, so our strategy relies on ELISA assays for routine study sample analysis whenever possible. A synthetic peptide to the Myl3 tryptic fragment was used to prepare calibration samples and ultimately measure Myl3 serum concentrations. Alternatively, Myl3 protein could have been used to prepare calibration samples; however, this approach was abandoned due to a variety of technical challenges related to optimizing recovery, digestion efficiency, and accuracy. The use of the synthetic peptide to measure Myl3 protein concentrations provided the means to optimize assay conditions and resulted in greater IP and digestion recovery, improved accuracy, and ultimately a lower limit of quantification. Because the internal standard is not added to the Myl3 samples until after the IP (i.e., the peptide internal standard does not contain the antibody epitope and would be washed away), great care had to be taken during the IP to prevent poor recovery and precision. In addition to validating the assay to measure both the synthetic peptide and Myl3 protein, Myl3 QC samples were included in each run to further validate assay performance. Due to the presence of endogenous Myl3 in rat serum, control rat serum was prepared using immunodepletion. Using this approach, a 2-mL serum pool was diluted to 9 mL with buffer 1 and spiked with 1 mL (10 mg) of antibody-complexed beads (8 µg of Ab/mg of beads). The serum was incubated overnight at 4 °C, and the beads were removed using a magnetic stand. Myl3supplemented ID rat serum concentrations could be accurately
and precisely predicted using independently prepared synthetic peptide standards. Assay Validation. The assay as described was linear over the range of 0.0734-7.16 nM. The results of the intraday and interday accuracy and precision for the 13-mer and for Myl3 are presented in Tables 1 and 2, respectively. The interday accuracy (% RE) and precision (% CV) of the assay with respect to the 13-mer was 7.6 and 11.1%, respectively, over the range of validation concentrations. Likewise, the interday accuracy and precision of Myl3 was 12.9 and 13.2%, respectively. Representative extracted ion chromatograms obtained from the analysis of a standard at the lower limit of quantitation (0.07 nM) and from a blank sample are presented in Figures 3and 4. Additionally, percent relative error statistics are presented in Table 3 for the standard samples on each of the 3 days of the validation. Key measures of assay selectivity were evaluated during the validation. With respect to matrix interferences, the background in the 13-mer blank (plus SIL) at the same retention time as the 13-mer was determined to be ∼1.4% of the lower limit of quantification (0.0734 nM). This represents the background present in the digestion matrix (10% methanol in 50 mM ammonium bicarbonate with 712 pM SIL). Similarly, the background present in an ID rat serum blank (plus SIL) was found to be ∼4.5% of the Myl3 lower limit of quantification and was probably due to endogenous Myl3 that remained following immunodepletion. Neither of the interferences had a significant impact on assay performance. Endogenous Myl3 in serum obtained from healthy rats varies. The endogenous level of Myl3 in the rat serum pool used to prepare the control matrix was estimated to be 1.32 pM. Analyte carryover was also evaluated and was not a significant issue with this assay. Absolute carryover was found to be ∼0.04% of the preceding sample (worst case). The relative carryover, carryover from the highest standard as a percent of the lower limit of quantification, was determined to be 5.6%. One of the key advantages to using a synthetic peptide to measure the tryptic fragment obtained from the digestion of Myl3 is the ability to provide accurate measurements, not just precise ones. The 13-mer standard curves were prepared in relatively neat digest buffer (10% methanol, ammonium bicarbonate, SIL, no trypsin), while the Myl3 QCs and study samples were immunoprecipitated from serum and contained trypsin used for the digestion. Throughout the validation and sample analysis, the difference between the mean protein QC concentrations and the peptide standards was -15% (i.e., the protein concentrations had -15% residual error when extrapolated from the peptide standard curve). This negative bias is related to total recovery, which in turn is related to the sum of the losses due to IP recovery, digestion efficiency, and ionization suppression. In order to provide accurate Myl3 concentrations during sample analysis, these losses were adjusted by applying a correction factor (1.18, i.e., 85% total recovery) to the protein QC and study samples. Divergence from this factor would be illustrated by poor Myl3 QC accuracy during sample analysis and the validation. Analyte stability in solution and in the matrixes (e.g., ID rat serum, digestion matrix) was investigated. The synthetic peptide (13-mer), 13-mer SIL, and Myl3 were found to be stable for at least 8 months in 50:50:0.1 water/acetonitrile/formic acid (v/v/v) when stored at -70 °C. Room-temperature stability was Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
4203
Figure 3. Extracted ion chromatogram of the 13-mer at the LLOQ (0.0734 nM) and SIL at 0.712 nM. The SRM channels (699.0 f 913.5, 1027.6) were monitored for the 13-mer, and (703.0 f 921.5, 1035.7) were monitored for SIL.
Figure 4. Extracted ion chromatogram of a blank sample (no standard) showing the SRM channels for the 13-mer (699.0 f 913.5, 1027.6) and SIL 703.0 f 921.5, 1035.7) is presented.
evaluated in rat serum, and Myl3 was found to be stable for at least 24 h. Extract stability, the maximum time the 13-mer can spend in digest matrix while waiting to be injected, was found to be at least 9 h. In addition, extract stability is evaluated during each batch analysis by injecting duplicate (front/back) standard curves and evaluating the residual errors of the coregressed 4204
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
standard curve samples (Table 3). Long-term freezer stability will be evaluated, as necessary, during the course of sample analysis. Isoproterenol Study. Two hours following subcutaneous administration of 50 mg/kg isoproterenol, the Myl3 serum concentration increased ∼10-fold from below the lower limit of quantification (0.07 nM) in the vehicle group to 0.64 nM. Myl3
Table 3. Percent Relative Error (% RE) Values for the Standard Samples on Each Day of the Validationa % RE day
standard (nM)
front curve
back curve
1
0.0734 0.183 0.458 1.15 2.87 7.16 0.0734 0.183 0.458 1.15 2.87 7.16 0.0734 0.183 0.458 1.15 2.87 7.16
4.0 -12.6 -6.1 -8.3 -2.5 16.3 9.2 -1.4 -4.9 -5.5 -9.0 7.2 6.4 3.4 -8.8 -2.8 -5.8 9.6
4.6 -4.7 -3.0 2.0 -0.3 10.7 -8.5 0.8 6.2 -8.0 10.9 2.9 -1.7 -7.7 -13.4 9.2 4.9 6.8
2
3
a The standard curves were run in duplicate (front/back) for all analyses to verify the absence of extract instability.
Figure 5. Time course of Myl3 response in rat serum following a 50 mg/kg subcutaneous dose of isoproterenol, a β-adrenergic receptor agonist known to induce cardiac injury. (The vehicle dose group had Myl3 concentrations below the lower limit of quantification.)
peaked at 1.7 nM after 8 h and remained at 1.3 nM following the last collection point, 24 h (see Figure 5). A similar pharmacokinetic profile was observed following a 10 mg/kg subcutaneous dose with a peak Myl3 concentration of 1.9 nM occurring at 8 h postdose (unpublished data). (These nM levels represent study sample concentrations ranging from below 1 to 45 ng/mL.) The lack of dose linearity between the 10 and 50 mg/kg isoproterenol studies suggests dose saturation. Myl3 demonstrates several of the ideal cardiac biomarker characteristics. Myl3 serum concentrations in healthy rats were low and ranged from below the lower limit of quantification (0.07 nM) to ∼0.15 nM. A greater than 25-fold increase in Myl3 serum
concentration occurred following cardiac injury with isoproterenol and provided a good kinetic window out to 24 h due to slow elimination. The advantageous kinetic profile of Myl3 in rat serum addresses the limited diagnostic window demonstrated by hFabp3 and the cardiac troponins and reduces the chance of a false negative result that may be indicated by markers with relatively short half-lives. Myl3 can be potentially affected my skeletal muscle injury, but when used in conjuction with a panel of markers chosen for their complementary characteristics, a false positive result due to skeletal muscle injury would be apparent (e.g., by low cTnI levels). CONCLUSION A strategy was presented based on the use of antibodies to isolate a protein of interest from biological matrixes, such as plasma, followed by proteolytic digestion and LC/MS/MS detection. The key advantages of LC/MS/MS are speed of method development and selectivity, which enable the triage of biomarker characteristics prior to costly and time-consuming ELISA development. Using this strategy, a sensitive and selective LC/MS/MS assay was developed to measure Myl3 serum concentrations in rat. The assay utilizes immunoprecipitation to selectively isolate Myl3 from the matrix followed by digestion with trypsin to release a surrogate peptide. Column switching was used to automate online SPE to further clean the samples following digestion. A synthetic peptide to the tryptic fragment and an associated stable isotope-labeled internal standard were used to quantify the tryptic fragment, which was stoichiometrically related to the intact protein concentration. The assay was validated using both the intact protein and the synthetic peptide to demonstrate the validity of this approach. Following validation, the assay was used to evaluate Myl3 as a biomarker for drug-induced cardiac injury in rat. A subcutaneous administration of 50 mg/kg isoproterenol, a known cardiotoxic agent, caused Myl3 serum concentrations to increase from nearly undetectable to 1.7 nM, a greater than 25-fold increase. Serum concentrations remained high after 24 h, and Myl3 confirmed its utility as a cardiac marker. Most importantly, Myl3 demonstrated a broad diagnostic window following cardiac injury, which addressed a distinct disadvantage of the other members of our biomarker panel. Because of the difficulty in finding a single ideal biomarker for evaluating cardiac injury, a strategy was proposed to develop a panel of markers with a proven history in the clinic that could be translated into veterinary species to support research from early discovery through clinical evaluation. ACKNOWLEDGMENT The authors thank Mike Pritt and Steve Engle for their hard work in conducting the isoproterenol study, and Yue-Wei Qian for his efforts in synthesizing the His-Myl3. Received for review January 10, 2007. Accepted March 27, 2007. AC070051F
Analytical Chemistry, Vol. 79, No. 11, June 1, 2007
4205
