2014 Volume 3 Issue Special_Issue Pages S0030
Parallel Fragmentation Monitoring (PFM), which is an analogue of selected reaction monitoring (SRM), is a recently developed method for quantification of small molecules by MALDI-TOF/TOF mass spectrometry (MS). It is well known that isobaric interference substances can be occasionally present in complex biological samples, and affect the accuracy of measurement by SRM. Unfortunately, by design it is not possible to assess whether isobaric interference happens in a SRM analysis. In contrast, the unique design of PFM should allow quick inspection for isobaric interference and subsequent correction. In this study, using arginine as an example, interference effect of isobaric structural analogs on the quantification of citrulline by PFM was evaluated. Our results showed that the presence of arginine affected the measured concentrations of citrulline standard solutions in a concentration dependent manner. Such interference could be observed readily in the MS/MS spectra, and contributed by [arginine+H–NH3]+ fragment ion. Because of having highly similar mass, 13C-isotope of [arginine+H–NH3]+ fragment ion overlapped with monoisotope of [citrulline+H–NH3]+ fragment ion, which was used as the report ion for quantification. However, such interference could be mathematically eliminated or minimized through estimation of the signal intensity of the 13C-isotopic peak of [arginine+H–NH3]+ from the intensity of the corresponding monoisotopic peak according to isotope distribution of elements. Furthermore, the presence of interfering fragment ions could be avoided by optimizing MALDI ionization condition. In conclusion, isobaric interference can happen in PFM, but can be easily identified in the mass spectra and eliminated (minimized) with simple methods.
Selected reaction monitoring (SRM) is the most common mass spectrometry (MS) method for quantification of small molecules such as amino acids,1,2) hormones,3) pesticides,4) and drug metabolites.5,6) SRM is mainly carried out with liquid chromatography electrospray ionization triple-quadrupole (LC-ESI-QqQ) MS platform. Unambiguous identification of small molecules is one of the current challenges for the analytical laboratories. Accuracy of mass measurement and chromatographic retention time, and reproducibility of ultraviolet spectra and fragment ion pattern are important factors in confirming identities of target compounds in SRM.7) Occasionally, endogenous isobaric compounds that have transitions (precursor/fragment ion pairs) identical to those of target compounds can be present in complex biological specimens (i.e., plasma, tissue lysates or cell lines), and co-eluted with the target compounds, thus making quantification inaccurate. Unfortunately, by design, SRM does not provide sufficient product ion information for structural elucidation and subsequent assessment the isobaric interferences, resulting in false positive signals.8,9) In addition, substantially shorten LC run can increase risks of co-eluting isobaric interfering compounds from complex samples.10)
The need for high-throughput MS-based methods for quantification of small-molecule metabolites has been increasing in clinical laboratories. Matrix-assisted laser desorption/ionization time-of-flight/time-of-flight mass spectrometry (MALDI-TOF/TOF MS) has potentials to be applied to the qualification and quantification of small molecules.11–14) It can avoid the need for LC separation prior to MS analysis, resulting in reduction of turn-around time. Moreover, it can allow quick assessment of result quality by identifying interfering signal peaks near the target analyte in the entire MS and/or MS/MS spectra.15)
The overall performances of LC-ESI-QqQ MS and MALDI-QqQ MS on quantification of small molecules analysis were shown to be comparable.11,16) However, it is logical to expect that exclusion of the LC separation step in MALDI-MS analysis will result in higher chance of the presence of interfering isobaric compounds, especially when analyzing highly complex biological samples. Sources of isobaric interference are not only originated from compounds having molecular weights (MWs) that are too close to be resolved by an MS instrument (e.g. mass difference <0.05 Da in MALDI-TOF MS), but also those with a mass difference of about 1 Da. In the latter case, MS or MS/MS signal peaks generated from the 13C-isotope of a compound can overlap with those generated from the monoisotope of a target analyte, and vice versa.17) For example, among the common amino acids in human specimens, there are 3 pairs of structural analogs with a mass difference about 1 Da, and they are asparagine (MW 132.12) and aspartic acid (MW 133.10), glutamine (MW 146.15) and glutamic acid (MW 147.13), and arginine (MW 174.20) and citrulline (MW 175.19). Recently, using citrulline for proof-of-concept, a novel method named parallel fragmentation monitoring (PFM), which is an analogue of SRM, has been developed for small molecules quantification by using MALDI-TOF/TOF MS.14) In contrast to ESI-QqQ-MS in SRM mode, MALDI-TOF/TOF MS provides mass spectra covering the entire acquisition range during PFM analysis. This should allow quick inspection of isobaric interference, and subsequent correction for unwanted signals by using a simple mathematical method. Using arginine as an example, interference effect of an isobaric structural analog on the quantification of citrulline by the PFM assays was evaluated in this study. Practical solutions for this problem were also explored.
Mesoporous graphitized carbon nanoparticles (<500 nm particle sizes), L-citrulline (≥98% purity), L-arginine (≥98% purity) and L-fructose (≥97% purity), were from Sigma-Aldrich (St. Louis, MO, USA). Graphene nano-flake (8 nm thick) was from Graphene Laboratories Inc. (Reading, MA, USA). [Ureido-13C] citrulline (99% purity) was from Cambridge Isotope Laboratories (Andover, MA, USA). Acetonitrile (ACN), ethanol and trifluoroacetic acid (TFA) (spectroscopy grade) were from Merck (Darmstadt, Germany). Purified water was obtained from Milli-Q water purification system (Millipore, Milford, MA, USA).
Standard solutions, graphitized carbon nanoparticles and graphene flakesArginine, citrulline and [ureido-13C] citrulline (internal standard) were dissolved in 50% ethanol in water (v/v) and prepared at concentrations of 100–200 μM, 10–250 μM, and 250 μM, respectively. Graphitized carbon nanoparticles were washed and resuspended in 20% ethanol in water (v/v) at a concentration of 10 mg/mL as a stock solution. Working suspension of the graphitized carbon nanoparticles was freshly prepared by mixing the stock solution with equal volume of fructose solution (50 mg/mL fructose in water), as previously described.14) Graphene flakes were initially washed with ACN and water, and then resuspended in 50% ethanol in water (v/v) at a concentration of 2 mg/mL. Graphitized carbon nanoparticles and graphene flake solution were dispersed by sonication for 10 min before use.
MALDI-TOF/TOF mass spectrometry and PFMMass spectra were acquired using Ultraflex III MALDI-TOF/TOF mass spectrometer (Bruker Daltonics) equipped with a smartbeam Nd-YAG laser operating at a wavelength of 355 nm. Unless otherwise stated, the instrument conditions for reflectron and LIFT tandem MS as well as peak detection algorithm of Flex Analysis software (version 2.0) were identical to those previously reported.14) PFM was also performed as previously reported.14) Briefly, the setting of the timed ion selector gate was optimized to obtain the maximum amount of precursor ions for both citrulline and [ureido-13C] citrulline simultaneously (i.e., ions with m/z values between 176.10 and 177.10 for protonated ions, and between 198.10 and 199.10 for sodiated ions). Resolution and signal-to-noise ratios of all selected precursor ions were >1,400 and >500, respectively. The selected precursor ions were subjected to LIFT for parallel fragmentation. The target fragment ions from citrulline and [ureido-13C] citrulline for quantification (i.e., ions at m/z 159.1 and 160.1 for protonated ions or at m/z 153.1 and 154.1 for sodiated ions) were subsequently detected in the same tandem MS spectrum. The peak area ratio of the selected fragment ion from citrulline to the one from the internal standard (i.e., [ureido-13C] citrulline) was used for quantification. A calibration curve was generated by plotting peak area ratio against citrulline concentration.
Estimation of elemental compositions and isotope distributionPossible elemental compositions of an ion were obtained by searching the METLIN database18) for protonated metabolites having a theoretical m/z unit matched to an observed m/z value within an error tolerance of 0.05 m/z unit. Relative abundances of various isotopes of a molecule were calculated from its elemental composition by using an isotope distribution calculator IDCalc (Department of Genome Sciences, University of Washington, USA).
Arginine has a molecular weight (i.e., MW 174.19) 1 Da less than that of citrulline (i.e., MW 175.19) and it can be converted to citrulline by deimination. Arginine shares proximity in MW and similar structure with citrulline, but differs by only 1 functional group. Theoretical masses of the natural 13C-isotope of arginine and the monoisotope of citrulline are only different by 0.019 Da. Using graphitized carbon nanoparticles as matrix, the naturally occurring 13C-isotope of arginine overlapped with the monoisotope of citrulline in the MS spectrum (Figs. 1A and B). Various amounts of arginine (i.e., 0 μM, 100 μM, and 200 μM) were spiked to citrulline standard solutions containing various amounts of citrulline (i.e., 20 μM, 40 μM, and 125 μM) and a fixed amount of internal standard (i.e., 250 μM [ureido-13C] citrulline). When more arginine was added, the measured citrulline concentration deviated more from the expected value (Table 1). For a citrulline standard solution of 20 μM, the presence of 200 μM of arginine resulted in an over-estimation of 41%.
Arginine loading (μM) | 0 | 100 | 200 | 0 | 100 | 200 | 0 | 100 | 200 |
---|---|---|---|---|---|---|---|---|---|
Expected citrulline concentration (μM) | Measured citrulline concentration (μM) [% error] | ||||||||
Without correction | With correction for a known compound (i.e., arginine) | With correction for an unknown compound | |||||||
20 | 19.8 | 23.4 | 28.2 | 19.4 | 20.7 | 22.8 | 19.4 | 20.0 | 21.5 |
[1.0%] | [17.0%] | [41.0%] | [3.0%] | [3.5%] | [14.0%] | [3.0%] | [0%] | [7.5%] | |
40 | 43.0 | 43.0 | 48.4 | 42.6 | 40.3 | 43.3 | 42.5 | 39.7 | 42.0 |
[7.5%] | [7.5%] | [21.0%] | [6.5%] | [0.7%] | [8.2%] | [6.3%] | [0.7%] | [5.0%] | |
125 | 133.5 | 131.1 | 141.6 | 133.1 | 122.6 | 136.5 | 133.0 | 121.8 | 135.3 |
[6.8%] | [4.9%] | [13.3%] | [6.5%] | [1.9%] | [9.2%] | [6.4%] | [2.6%] | [8.2%] |
In the PFM assay, an area ratio of the citrulline fragment ion at m/z 159.1 to the corresponding fragment ion of the internal standard (i.e., [ureido-13C] citrulline) at m/z 160.1 was used for quantification. Reviews of the MS/MS spectra showed that [M+H–NH3]+ fragment ion of arginine 13C-isotope overlapped with that of the citrulline monoisotope at m/z 159.1 (Figs. 1C, D), and contributed to the positive bias on the measurement of citrulline concentration (Fig. 2).
As a TOF/TOF analyzer provides entire mass spectra of precursor and fragment ions, we speculated that the isobaric interference effect could be eliminated or minimized by estimating the 13C-isotopic peak area of [arginine+H–NH3]+ fragment ion (m/z 159.1) from the area of the corresponding monoisotopic peak (m/z 158.1) on the basis of the isotope distribution. According to the theoretical isotopic distribution of C6H12N3O2, the interference signal contributed by the 13C-isotopic peak area of [arginine+H–NH3]+ is equivalent to 7.9% of the monoisotopic peak area. In order to obtain accurate measurement of the monoisotopic peak area of [arginine+H–NH3]+ fragment ion, the setting of the timed ion selector gate was optimized to obtain the maximum amount of precursor ions for arginine, citrulline and [ureido-13C] citrulline simultaneously (i.e., ions with m/z values between 175.10 and 177.10). Then the interfering signal was estimated from the monoisotopic peak area of [arginine+H–NH3]+. After subtracting the estimated interfering signal from the peak area of citrulline fragment ions at m/z 159.1, the degree of over-estimation of concentrations of citrulline standard solutions was greatly reduced to <15% (Table 1).
In real practice, it may not be able to elucidate the structure of an interfering isobaric substance from the MS and MS/MS data. Without the structural information, it is impossible to calculate the relative abundances of its isotopic species. However, we could estimate the relative abundances of the isotopes from the possible elemental compositions derived from its monoisotopic m/z value. Seven possible elemental compositions were found by searching the METLIN database for protonated metabolites having an m/z value of 158.1±0.05. Relative abundance of the 13C-isotope for each possible elemental composition was calculated. For these 7 elemental compositions, the average value of 13C-isotopic peak area at m/z 159.1 is equivalent to 9.8% of the monoisotopic peak area (Table 2). After using this average value to correct for the isobaric interference, the accuracies of the observed concentrations of citrulline standard solutions were found to be acceptable. The relative errors were <15% (Table 1).
Elemental compositions | Exact m/z | Natural abundance (%) of m/z 158.1±0.05 | Abundance (%) of the first isotope | Relative abundance (%) |
---|---|---|---|---|
C8H15NS | 158.0998 | 86.5816 | 8.7774 | 10.1 |
C11H11N | 158.0964 | 88.2383 | 11.0997 | 12.6 |
C6H11N3O2 | 158.0924 | 92.0497 | 7.2834 | 7.9 |
C8H15NO2 | 158.1176 | 90.6781 | 8.5458 | 9.4 |
C7H11NO3 | 158.0812 | 91.5016 | 7.6035 | 8.3 |
C10H7NO | 158.0601 | 89.0396 | 10.208 | 11.5 |
C6H15N5 | 158.1400 | 91.7678 | 7.9196 | 8.6 |
For situations similar to this arginine case, the isobaric interference effect can be minimized by using the monoisotopic peak area of an interfering ion. However, this approach is not applicable for the presence of an isobaric compound having a mass value identical to that of a target molecule. For this reason, we further investigated whether a particular MALDI ionization condition could generate any quantifiable fragment ions that were free from isobaric interference. Using graphene flake as the inorganic matrix, we found that sodiated fragment ions (i.e., [M+Na–COOH]+) at m/z 153.1 and m/z 154.1 were detected for citrulline and [ureido-13C] citrulline, respectively, while arginine did not generate any fragment ions at m/z 152.1, m/z 153.1 or m/z 154.1 (Figs. 3A, B, C, D). By using graphene flake as matrix and the specific ion transition of citrulline (i.e., m/z 153.1), the presence of arginine did not falsely increase the peak intensity of the report fragment ion of citrulline (Fig. 3E). The isobaric interference from arginine completely disappeared. When using the areas of the peaks at m/z 153.1 and m/z 154.1 instead of at m/z 159.1 and m/z 160.1 for citrulline quantification, a similar linear calibration curve was observed. Because fragmentation pathways are different for various types of ion species (e.g., protonated, lithiated, sodiated and potassiated), the disappearance of the interfering isobaric fragment ion of arginine could have been contributed by the formation of sodiated precursor ion, but not the use of graphene flake. To test this, we added 10 mM NaCl into the matrix solution. In the presence of 10 mM NaCl, the use of graphitized carbon nanoparticles promoted the formation of sodiated citrulline ion, and resulted in a fragmentation pattern similar to that obtained with graphene flake. However, only a portion of citrulline molecules formed the sodiated precursor ions, resulting in decreased detection sensitivity. In contrast, the use of graphene flake allowed the formation of sodiated precursor ions from all citrulline molecules.
A number of approaches and algorithms were previously reported for identification of isobaric interference substances and correction of target signals in SRM assays.19,20) All of them were developed for ESI-MS, but not for MALDI-MS. When using ESI-QqQ MS with flow injection, Piraud et al. observed interfering signals during quantification of amino acids, and they were resulted from isobaric structural analogs, in-source collision-induced dissociation, or natural isotopic distribution of the elements.21) Optimized LC separation before SRM was needed to minimize the isobaric interference.21) By design, SRM only monitors ion signals of a selected transition pair at specific m/z values. Signal data at other m/z values are not available. Hence, it is difficult to discover the presence of isobaric interference if it happens. In contrast, MALDI-TOF/TOF mass spectrometer provides all peak signals in the entire acquisition range. As shown in this study, the presence of isobaric interference substance in PFM can be easily identified from the MALDI-TOF/TOF mass spectra by simple inspection. Furthermore, in the case of an isotope of a compound having a mass value identical to a target analyte, availability of signal data at other m/z values allows one to estimate and correct for the degree of isobaric interference according to distribution of the elements by using a simple mathematical method. For an isobaric compound having a mass value identical to that of a target molecule, an interfering fragment ion can be removed by optimizing the MALDI ionization condition. Theoretically, the isobaric interference from arginine can be eliminated by using the high resolution MS with the resolution larger than 10,000 because the mass difference between citrulline and 13C-isotope of arginine is 0.0193. However, the costs of high resolution MS platforms (e.g., high resolution q-TOF MS, orbitrap MS) are relatively high. Although current MALDI-TOF/TOF MS instruments do not provide sufficient resolving power to differentiate most isobaric precursor/fragment ions, our result shows that isobaric interference effect can be substantially reduced or eliminated with simple methods. In the future, MALDI-based bench-top/hand-held tandem MS devices with limited resolution may be developed for bed-side applications. Then our methods for elimination of isobaric interference will become useful. In conclusion, isobaric interference can happen in PFM, but can be easily identified and eliminated (minimized) with simple methods.
This project was supported by RGC General Research Fund (project No. 403413) from the University Grants Committee, Hong Kong and by the Li Ka Shing Foundation.