A NOVEL METHOD OF M/Z DRIFT CORRECTION FOR OA-TOF MASS SPECTROMETERS BASED ON CONSTRUCTION OF LIBRARIES OF MATRIX COMPONENTS. Martin R Green*, Keith Richardson, John Chipperfield, Nick Tomczyk, Martin Palmer. Waters Corporation, Wilmslow, UK, SK9 4AX OVERVIEW PURPOSE- Investigation of the use of common matrix ions for m/z drift correction in complex mixtures. METHODS- Construction of m/z, RT library from repetitive injections of human urine matrix. Accurate mass measurement of eight forensic toxicology standards spiked into urine using matrix library. Comparison of accurate mass measurement using external lock mass. RESULTS- 1ppm RMS mass accuracy and improved precision compared to external lock mass method. INTRODUCTION Orthogonal acceleration Time-of-Flight (oa ToF), accurate mass measurement is a well-established analytical technique. Compensation for temperature-dependent m/z drift is commonly performed by periodic measurement of internal or external reference peaks introduced using a purpose built interface. In many analyses, analytes of interest are minor components within a complex chemical matrix, the composition of which is largely invariant regardless of sample origin. Many common biological, environmental and food matrices are known. In this paper, a library of m/z and retention time (RT) is constructed for common components of a matrix and used to correct for m/z drift for subsequent analyses. This approach is compared with results from periodic external lock mass correction. TO DOWNLOAD A COPY OF THIS POSTER NOTE, VISIT WWW.WATERS.COM/POSTERS 1
METHODS All experiments were performed in positive ion electrospray ionization on a Waters Xevo G2-S QToF mass spectrometer using an ACQUITY UPLC system. Urine samples from three different male subjects (A, B, C) were collected and diluted by 10x before analysis. An aliquot of urine sample (A) was spiked with an eight component mixture of standard forensic toxicology compounds at a concentration of 25pg/μL. Leucine Enkephalin m/z 556.2771 was used to provide a single point calibration correction at one minute intervals during the chromatographic separation introduced via a separate electropspray nebulizer. ToF data was acquired in MS E mode (alternating low and high collision energy) at 10 spectra / second. ACQUITY UPLC Conditions: Column: Mobile Phase: Gradient: ACQUITY HSS T3 2.1mm x 100mm (1.8μm). A. 100% Water + 5mM Ammonium Formate B. 100% Acetonitrile + 0.1% Formic Acid 87%A with a linear ramp to 50%B at 10 minutes then a linear ramp to 95%B at 11 minutes. Column temperature: 40 C Flow rate: 0.4mL/min Injection volume 10μL RESULTS Chromatographic peaks corresponding to matrix components are often intense and reproducible for a given matrix. Matrix peaks often appear in the same RT and m/z range as targeted analytes and are therefore ideal to use for internal m/z reference. Figure 1 shows the base peak chromatogram (BPI) of matrix components from three distinct types of matrix. Urine was used as the matrix in all further work presented. Figure 1. BPI chromatograms from three common matrices. 2
Figure 2 shows BPI chromatograms of the urine samples from the three subjects used in this study. Although differences may be seen there is clearly a great deal of commonality between the main components of this matrix. These peaks are from matrix compounds such as urea, creatinine, uric acid, citrate, host/pathogen DNA, host/ pathogen RNA and amino acids. To validate the method in Figure 3, results from ten repeat injections of Urine A were used to generate a m/z, RT library. Figure 4 shows a plot of m/z vs RT for entries in this library. Approximately 10 of the largest peaks in each 1min RT window were selected. Figure 4. Library containing 147 peaks from sample A Figures 5 and 6 show peaks in Urine samples B and C respectively, which can be matched to the library entries created from Urine A. The plots of m/z error (ppm) vs RT (min) show the high degree of commonality between samples from the three subjects. A more exhaustive library could have been constructed by combining peaks from the three samples and including solvent ions at each retention time. Figure 3 shows a simplified flow diagram of the method of m/z drift correction using matrix ions. Figure 5. m/z error (ppm) vs RT (min) for peaks in Urine B matched to matrix li-brary constructed from analysis of Urine A. 127 peaks matched. Figure 6. m/z error (ppm) vs RT (min) for peaks in Urine C matched to matrix li-brary constructed from analysis of Urine A. 131 peaks matched. Urine A was spiked with an eight component forensic toxicology mixture at 25pg/μL. 10 consecutive analyses of this mixture were performed. An artificially large m/z drift was forced by directing a cooling fan at the system. Analysis included monitoring an internal lock mass at one minute intervals for comparison. Peaks corresponding to those in the library (Figure 4) were identified. Figure 7 shows m/z error (ppm) vs. RT for the identified library peaks in each injection. Figure 8 shows the m/z error (ppm) for the external lock mass during each run. Figure 3. Outline of method used to correct for m/z drift using matrix ions. 3
Figure 10 shows the mass error (ppm) vs RT for the eight spiked compounds in each of the ten injections after correction using the library matched matrix peaks. For comparison Figure 11 shows the same data m/z corrected using the periodic external reference peak. Figure 7. Library matched data for 10 repeat injections Urine A with spiked standards Figure 8. External lock mass data for 10 repeat injections Urine A with spiked standards A trend line was calculated for each of the ten sets of library matched peaks. An example trend line is shown in Figure 9. The trend line in Figure 9 was generated by determining the median ppm error value of all data points in a 2 minute RT window. The centre of the window was shifted by 1 minute and the median recalculated to determine consecutive points. A linear extrapolation was used between each point. From this trend line a m/z correction factor was calculated at each RT and used to correct the measured values for the eight spiked compounds. For comparison the same data was corrected using the periodic external lock mass. Figure 10. Plot of mass measurement error (ppm) for eight spiked compounds corrected using matrix ions. Figure 11. Plot of mass measurement error (ppm) for eight spiked compounds corrected using external lock mass. Figure 12 shows a table of RMS mass measurement error in ppm for all eight spiked compounds over the ten injections. Results from both data corrected using identified matrix ions and correction using the external lock mass are presented. Figure 12. RMS mass measurement error (ppm) Figure 9. Example of calculated trend line for library matched data allowing calculation of m/z correction factor at each RT. 4
DISCUSSION The results in Figure 12 show an increased mass precision when using matrix ions for mass correction, compared to when using an external reference. This is in part due to the large number and intensity of the matrix ions. Introduction of a more intense external lock mass signal would also have resulted in improved precision. All experiments were preformed in MS E mode resulting in a second, high collision energy (CE) function associated with each data file. No attempt was made to use these data, however adding high CE data to the library would increase the number of library entries significantly. In addition, the m/z range of the matrix peaks at each RT would be increased compared to Figure 4. This would allow more detailed correction of mass calibration, if required. Over the course of the experiments presented no significant shift in RT was observed. However, the method discussed may also be used to compensate for RT time shifts as columns age or due to differences between systems. Although not shown in this study, the matrix library may also include ion mobility (IMS) data. The collision cross sections measured in LC-IMS-MS data have been shown to be highly reproducible. Using m/z, drift time and CCS and sophisticated pattern matching software allows matrix ions to be identified regardless of RT. For example, when measuring analytes from the same matrix but with different chromatographic conditions. CONCLUSION m/z drift correction using matrix ions has been shown to give comparable results to using a periodic external reference. The large number of matrix ions at each RT improves measurement precision allowing sub ppm accuracy. Matrix libraries may be built for many matrix types and transferred between different instruments. The high degree of reproducibility of CCS measurements using ion mobility may be used to add specificity to library identification and compensate for retention time changes. The method described removes the need for external or internal references, simplifying the process of mass measurement. 2015 Waters Corporation 5