ARTICLE pubs.acs.org/ac
Comparison of MALDI-TOF-MS and HPLC-ESI-MS/MS for Endopeptidase Activity-Based Quantification of Anthrax Lethal Factor in Serum Zsuzsanna Kuklenyik,† Anne E. Boyer,† Renato Lins,†,§ Conrad P. Quinn,‡ Maribel Gallegos-Candela,† Adrian Woolfitt,† James L. Pirkle,† and John R. Barr*,† †
National Center for Environmental Health, Centers for Disease Control and Prevention, 4770 Buford Hwy, NE Atlanta, Georgia 30341, United States ‡ National Center for Immunization and Respiratory Diseases, Centers for Disease Control and Prevention, 1600 Clifton Road Atlanta, Georgia 30333, United States § Battelle Atlanta Analytical Services, 2987 Clairmont Road NE, Suite 450 Atlanta, Georgia 30329, United States ABSTRACT: Diagnosing and treating anthrax at the earliest stage of disease is critical. We developed a method to diagnose anthrax at early stages of infection by detecting anthrax lethal factor (LF) at the attomol/mL level in plasma or serum. This method uses antibody capture and quantification of LF endoproteinase activity by isotope dilution matrix-assisted laser-desorption ionization (MALDI) time-of-flight (TOF) mass spectrometry (MS). Many public health laboratories do not use MALDI-TOF-MS; thus, we have adapted the LF method for detection by electrospray ionization (ESI) tandem MS (MS/MS), which allowed comparison of both MS platforms for LF quantification. Calibration curves were linear from 0.05-2.5 ng/mL when measured after 2 h and from 0.005-1.0 ng/mL after 18 h incubation time. The limit of detection was 0.005 ng/mL using a 200 μL sample. The coefficient of variation for quality control samples was 6-12% for both MS platforms. Samples used to perform cross-validation included 158 serum samples from a study in rabbits exposed to anthrax spores by inhalation. Some were treated with anthrax immune globulin before exposure. Concentrations measured by ESI-MS/MS matched those by MALDITOF-MS with p = 0.99 (r2 = 0.997) and -0.25% mean relative difference ((9% standard deviation). This study shows that isotope dilution MALDI-TOF-MS is a robust and precise quantitative MS platform.
T
he development of technologies with high diagnostic specificity and sensitivity for bioterrorism select agents and toxins is critical to public health preparedness. Biotoxins and biotoxin producing bacteria have the highest fatality rates among the Category A bioterrorism agents1 and require rapid, sensitive, and specific diagnosis to ensure timely treatment options and effective prophylaxis. Of the available diagnostic approaches for early detection of biological toxins, mass spectrometry (MS)-based proteomics offers superior speed, specificity, and sensitivity without extensive sample purification and preconcentration. Results are available in a few hours, compared to days for classical microbiology or toxicology approaches.2,3 Matrix-assisted laser desorption/ionization (MALDI) timeof-flight (TOF) MS and electrospray ionization (ESI) with tandem quadrupole (MS/MS) mass filtering provide flexible platforms that are ideal for targeted analysis of biological toxins: proteins and small molecules. ESI-MS/MS is typically coupled online with high performance liquid chromatography (HPLC), allowing for both preconcentration and analytical separation from matrix interferences, providing enhanced analytic sensitivity and selectivity. With the use of isotope dilution internal r 2011 American Chemical Society
standardization, HPLC-ESI-MSMS is a highly rugged and reproducible detection technique and is considered by many to be the gold standard for quantification. Compared to ESI-MS/MS, MALDI-TOF has the advantage of being able to accommodate more complex samples.2 For MALDI-TOF-MS, samples mixed with desorption/ionization assisting matrix (MALDI matrix) are spotted and dried on a metal target plate without preconcentration. Furthermore, MALDI-TOF-MS has the advantage of high instrument throughput, each spot requiring as short as 10-15 s acquisition time, versus HPLC-ESI-MS/MS with several minutes per sample. However, currently, MALDI-TOF-MS has rarely been used for accurate and precise quantification and there are very few validated clinical methods that use MALDI-TOF-MS.4 To enhance emergency response and public health laboratory preparedness, we developed a MS-based method for quantification of anthrax lethal factor (LF) which is a specific biomarker of active infection by Bacillus anthracis.5 Our measurement Received: November 16, 2010 Accepted: January 13, 2011 Published: February 08, 2011 1760
dx.doi.org/10.1021/ac1030144 | Anal. Chem. 2011, 83, 1760–1765
Analytical Chemistry approach relies on antibody enrichment/separation of LF, coupled with LF specific hydrolytic activity on an optimally designed synthetic peptide substrate, followed by MS detection of unique hydrolysis products.6,7 We call this approach “EndopepMS” because it measures the LF-specific endoproteolytic activity by MS detection. Our previously published method uses MALDITOF-MS detection with high throughput (10 times signal-to-noise (S/N) ratios and at least five times more intense signals at the 0.005 ng/mL limit of detection (LOD) level (Figure 2, left) than for the blank serum (Figure 2, right). The ratio of the cleavage product to [MALDI matrix þ internal standard] and spotting one time 0.5 μL of the mix on the MALDI plate was determined for optimal crystallization of the peptides. The volume injected into the HPLC-ESI-MS/MS
system was also limited to 20 μL. These volumes represented only 1/600 versus 1/20, respectively, of the total cleavage product or LF amount per sample. The capacity of the HPLC-ESIMS/MS allows the use of a much greater portion of the cleavage products providing a better S/N ratio. We found, however, that using greater portion of the cleavage products in the HPLC injection volume increased the cleavage product background peak signal for the blank serum samples. This was likely caused by zinc mediated chemical degradation of the substrate. Although the cleavage products in the blank serum represented only low fmol/mL contribution from the 50 nmol/mL substrate solution, it was enough to cause a background peak signal. Thus, injecting more of the cleavage products improved the S/N ratio but the detectable level versus blank sample signal ratio remained similar, only moderately improving the HPLC-ESI-MS/MS LOD. Therefore, the analytic sensitivity of the method is limited by the amount of substrate in the incubation mix rather than the capability of the MS/MS instrument. Reproducibility. Variations in the LF measurement can originate not only from the MS detection but also from the antibody capture and the enzymatic reaction conditions. We developed and optimized the sample preparation protocol such that variations due to sample preparation and cleavage reaction were minimal. We used magnetic beads and an automated sample handling system where the beads were moved from plate-to-plate with magnetic sleeves instead of adding new wash reagents to the same plate. This provided a much cleaner end product with minimal loss of LF-mAb beads. We also prepared mAb beads in a minimum of 1 mL batches. Thus, we used the same batch of mAb beads for capturing LF from calibration standards and unknown samples, automatically correcting for potential changes in mAb bead quality from sample-run-tosample-run, caused by a slight decrease in binding capacity during storage (up to 1 month). 1763
dx.doi.org/10.1021/ac1030144 |Anal. Chem. 2011, 83, 1760–1765
Analytical Chemistry
ARTICLE
Figure 2. Comparison of the C-terminal cleavage product signals from 200 μL of serum after 18 h substrate cleavage time; spiked with 0.005 ng/mL LF (left) and blank serum (right): (A) MALDI-TOF-MS spectra, spot contained the equivalent of 1/600 volume of the cleavage reaction mix or LF amount per sample. (B) HPLC-ESI-MS/MS chromatogram, HPLC injection volume, contained 1/20 of the cleavage reaction mix of LF amount per sample.
Because of the enzymatic approach, the cleavage product response ratios were not only LF-concentration dependent but also reaction-time dependent. We ran a standard series along with every unknown batch which caused the incubation and sampling of the reaction mix to occur at the same time for calibration standards and unknowns. This eliminated the effect of small reaction time variations on reproducibility of the LF concentration measurements. We assessed overall method precision with 20 independent measurements of four levels of quality controls (0.014 and 0.15, ng/mL and 10 times diluted 1.5 and 15 ng/mL) in spiked serum. The accuracy and precision of the QCs ranged from 86-108% to 6-12%, respectively, regardless of the use of MALDI-TOF-MS or HPLC-ESI-MS/MS detection (Table 2). The ESI analyses tended to have somewhat lower CVs ranging from 6 to 9% while the MALDI TOF-MS analysis yielded CVs of 10-12%. The close match of accuracy and precision for the two techniques demonstrates that isotope dilution MALDI-TOF-MS is a valid quantification technique. The use of an isotopic-labeled internal standard and high speed laser (200 shots/s) and averaging 2000 random laser shots successfully compensates for the heterogeneous solid nature of the MALDI spots. Matrix and Sample Dilution Effects. To ensure the highest diagnostic sensitivity and specificity for human cases, the method was designed to achieve the best possible analytic sensitivity (0.005 ng/mL LOD) using 200 μL of serum or plasma. To minimize the effect of the sample matrix on recovery, all samples were diluted at least 5 times (up to 200 μL of sample into 1 mL). If the LF concentration in the test sample was greater than the LOL (2.5 ng/mL after 2 h cleavage reaction time), the analysis was repeated using a smaller sample volume, and the measured value was corrected with a dilution factor. To cover a wider concentration range in a single run, unknown samples were
Table 2. Mean, Percent Spiked Concentration Recovery (% Accuracy), and Coefficient of Variation (CV%) for Quality Control (QC) Samplesa spiked concentration 15 ng/mL
mean
accuracy %
CV%
MALDI (N = 20)
16.19
108
11
ESI (N = 20)
16.07
98
6
107
10
1.5 ng/mL
MALDI (N = 20)
1.60
ESI (N = 20)
1.62
95
6
0.14 ng/mL
MALDI (N = 20)
0.15
108
10
0.014 ng/mL
ESI (N = 20) MALDI (N = 20)
0.12 0.014
86 100
9 12
ESI (N = 20)
0.012
86
9
The 15 and 1.5 ng/mL QCs were analyzed using 20 μL sample aliquots, and the 0.14 and 0.14 ng/mL QCs used 200 μL.
a
prepared using 20 μL (or less) and 200 μL aliquots in the same batch. To provide quality control for diluted samples, we used 20 μL of the 15 and 1.5 ng/mL QCs, representing 10 times sample dilution. Accuracy and reproducibility for these diluted QCs were 107-108% (Table 2), which indicated that dilution did not significantly affect method accuracy. Comparison of Measurements in Rabbit Samples. To assess the effect of the matrix and the relative reproducibility of MALDI-TOF-MS versus HPLC-ESI-MS/MS detection modes, we used both methods to analyze serum samples from rabbits prior to and after exposure to anthrax spores by inhalation and in some rabbits that were protected from infection with antitoxin AIG of varying doses. This sample set provided roughly equal numbers of expected negative results and a wide range of positive results from the lowest LF levels in early infection to high LF 1764
dx.doi.org/10.1021/ac1030144 |Anal. Chem. 2011, 83, 1760–1765
Analytical Chemistry
ARTICLE
comparable quantitative accuracy of MALDI-TOF-MS to the well-defined isotope-dilution MS/MS quantitative platform. In addition, the LC-MS/MS method for LF anthrax diagnosis provides a platform that has potential for transfer to multiple state laboratories within the Laboratory Response Network with existing ESI-MS/MS systems, expanding the potential for response during an anthrax public health emergency.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. Phone: 770-488-7848.
’ REFERENCES
Figure 3. Comparison of HPLC-ESI-MS/MS versus MALDI-TOF data for rabbit serum samples.
levels in late infection. The LF concentrations ranged from LOD and 82 samples had
