RAPID COMMUNICATIONS IN MASS SPECTROMETRY Rapid Commun. Mass Spectrom. 2008; 22: 1359–1366 Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rcm.3511
Application of ion trap technology to liquid chromatography/mass spectrometry quantitation of large peptides Petia Shipkova*, Dieter M. Drexler, Robert Langish, James Smalley, Mary Ellen Salyan and Mark Sanders Bristol-Myers Squibb Company, PO Box 5400, Princeton, NJ 08543, USA Received 9 October 2007; Revised 11 January 2008; Accepted 23 February 2008
Triple quadrupole mass spectrometers are generally considered the instrument of choice for quantitative analysis. However, for the analysis of large peptides we have encountered some cases where, as the data presented here would indicate, ion trap mass spectrometers may be a good alternative. In general, specificity and sensitivity in bioanalytical liquid chromatography/mass spectrometry (LC/MS) assays are achieved via tandem MS (MS/MS) utilizing collision-induced dissociation (CID) while monitoring unique precursor to product ion transitions (i.e. selected reaction monitoring, SRM). Due to the difference in CID processes, triple quadrupoles and ion traps often generate significantly different fragmentation spectra of product ion species and intensities. The large peptidic analytes investigated here generated fewer fragments with higher relative abundance on the ion trap as compared to those generated on the triple quadrupole, resulting in lower limits of detection on the ion trap. Copyright # 2008 John Wiley & Sons, Ltd. The increasing interest in peptides as drugs for therapeutic treatment,1,2 or as biomarkers in clinical diagnosis,1,3 necessitates the development of sensitive and selective analytical methodologies for their quantitative assessment. In general, pharmacokinetic and pharmacodynamic studies on peptidic drug candidates are performed utilizing liquid chromatography/mass spectrometry (LC/MS) employing electrospray ionization (ESI) and using stable isotope labeled (i.e. 13C, 15N, 18O, 2H) peptides4–8 or other related peptides5,9,10 as internal standards. Targeted bioanalytical assays typically involve the detection and quantitation of an analyte in complex matrices, e.g. plasma, urine, bile, tissue homogenate. The peptidic analytes can be isolated by a variety of traditional off-line sample preparation and clean-up procedures, e.g. two-dimensional gel electrophoresis, liquid-liquid extraction, solid-phase extraction, immunoprecipitation, etc.11 Additionally, selective mass spectrometry methods are used to reduce the background and to provide optimum detection limits. In most cases, selected reaction monitoring (SRM) techniques are used, where a target-specific fragmentation transition(s) from precursor to product ions are monitored. If a suitable fragmentation transition is not observed, as in the case of some sterols or fatty acids, single (selected) ion monitoring (SIM) can be used, where only the precursor ion is monitored. Unless high resolution, accurate mass instruments (e.g. sector, FTMS or ToF) are used, SIM methods often
*Correspondence to: P. Shipkova, Bristol-Myers Squibb Company, PO Box 5400, Princeton, NJ 08543, USA. E-mail:
[email protected]
lack selectivity and could lead to inadequate detection limits if there are significant matrix interferences or other chemical noise. In this article we will limit the discussion to MS/MS quantitative methods and a comparison between ion traps and triple quadrupole instruments. In general, for small molecule quantitation, the instrument of choice is a triple quadrupole mass spectrometer, due to a high duty cycle and a large dynamic range in tandem mass spectrometry (MS/MS) mode, i.e. SRM, affording superior analytical figures of merit such as sensitivity, precision and accuracy, especially for simultaneous analysis of multiple analytes. Following this trend, most peptide quantitative studies reported in the literature utilize triple quadrupoles.12–16 However, as the practical examples in this publication demonstrate, the advantages provided by triple quadrupoles for quantitation of small molecules are not always applicable for large peptides, where quantitation by ion trap mass spectrometers could be an excellent alternative. It should be mentioned that all data shown here was obtained on a linear ion trap, that has been shown to have a superior sensitivity and scanning speed compared to spherical ion traps.17,18 The process of the MS/MS fragmentation in a triple quadrupole is very different from that in an ion trap. In a triple quadrupole fragments are produced from collisions with a gas, typically argon or nitrogen, as the precursor ion is accelerated through a pressurized collision cell (Q2). This leads to a collision cascade or a multiple-step fragmentation19 where fragment ions undergo collisions and can further fragment. In the case of peptides, generally, more energy is required to dissociate bonds as compared to small molecules and the resultant fragmentation tends to generate Copyright # 2008 John Wiley & Sons, Ltd.
1360 P. Shipkova et al.
a large number of product ions and thus relatively low abundance of the analyte ion(s). Also, the associated collisional scattering owing to high translational energy disperses ions in the collision cell and the subsequent mass analyzer, therefore reducing the number of ions reaching the detector. This decrease in intensity for the targeted product ion can lead to a reduction in the sensitivity of a quantitative SRM measurement method. In contrast, product ions generated in an ion trap are formed via applying a mass-to-charge (m/z) correlated resonant excitation frequency to the precursor ion which is then fragmented by collisions with the helium gas in the ion trap. This typically affords a single-step fragmentation, since the product ions have different masses to that of the precursor ion. They are no longer in resonance with the excitation frequency and, therefore, undergo fewer additional collisions and, generally, less fragmentation.19 The result is a product ion spectrum with fewer ions at higher relative intensities. In fact, if there are facile water or ammonia losses, the product ion spectrum can consist solely of the water/ammonia loss product ions, which do not give suitable SRM transitions due to lack of specificity. For molecules where water or ammonia loss is the predominant pathway, as observed in many peptides, wideband excitation19,20 is used, where resonance excitation frequency is applied over a mass range of 20 Da lower than the selected precursor ion, thus causing fragmentation of the precursor and the ions from a water or ammonia loss (pseudo MS/MS/ MS).19 The formation of fewer fragment ions of higher relative intensity has a direct correlation and positive impact on the overall sensitivity of a quantitation assay. This phenomenon has been previously described in detail21 for small molecules, where CID processes were directly compared for quadrupole ion traps and triple quadrupoles and it was shown that ion traps have greater efficiencies of fragmentation, collection, mass selection and transmission of the product ions to the detector. However, MS/MS on an ion trap is performed ‘in time’, meaning a serial process occurs for collection, isolation of precursor ions, fragmentation and detection of product ions. This result is a low duty cycle. In the case of a small molecule present at low levels in a LC peak, there may not be sufficient time to fill the ion trap with enough ions to make an accurate quantitative measurement. On the other hand, if the trap is allowed to fill, there may be an insufficient number of data points across the LC peak resulting in poor reproducibility and therefore a higher limit of quantitation (LOQ). However, with the utilization of the automatic gain control (AGC) available on modern ion traps, the ion population in the storage quadrupole is well controlled and, in general, prevents space charging and yields acceptable analytical figures of merit such as precision and accuracy in quantitation experiments. On a triple quadrupole the double-stage SRM analysis is carried out ‘in space’ with a continuous ion beam affording a nearly 100% duty cycle which provides better precision and thus better LOQs. Combining the advantages and disadvantages provided by duty cycle and fragmentation efficiency for the two types of mass spectrometers, in most cases triple quadrupoles give superior performance for quantitation of small molecules, Copyright # 2008 John Wiley & Sons, Ltd.
while ion traps hold an advantage for analysis of large peptides where collisional scattering and fragmentation efficiency become more significant factors. Collision and fragmentation efficiency has been defined by three parameters: fragmentation efficiency (EF), collection efficiency (EC) and collisional efficiency (ECID).22,23 They all contribute to the LOQ of an LC/MS/MS assay. The fragmentation efficiency describes the degree to which the precursor ion is fragmented, the collection efficiency characterizes the transmission efficiency through the collision cell, and collisional efficiency represents the ratio of the initial precursor ion to the collectable fragment ions. Another way of characterizing the fragmentation process is by using MS/MS efficiency (EM), which is described to provide a more realistic measure of the efficiency of conversion of the precursor ions entering the collision cell into detected ions in the MS/MS mode.22 In this study we have simply compared the ‘loss of signal’ or ratio of optimized signal of the precursor ion in full scan MS mode to the optimized signal of the measured product ion in MS/MS mode. There are a small number of literature examples utilizing ion trap technology for quantitation of small peptides.24–29 In this study we compare the detection limits of commercially available peptide analytes, utilizing both triple quadrupole and linear ion trap mass spectrometers, that were achieved during method development for LC/MS/MS assays employing SRM methods.
EXPERIMENTAL Chemicals and standards HPLC-grade solvents, water, acetonitrile (ACN) and formic acid (FA) were purchased from VWR International (West Chester, PA, USA) and were used without further purification. Glucogen-like peptide (GLP-1, 7-36), reported amino acid sequence: EGTFTSDVSSYLEGQAAKEFIAWLVKGR, MW 3089 Da, was purchased from Sigma (St. Louis, MO, USA). Exendin-4, reported amino acid sequence: HGEGTFTSDLSKQMEEEAVRLFIEWLKNGGPSSGAPPPS, MW 4185 Da, was purchased from SynPep Corporation (Dublin, CA, USA). Arginine Vasopressin (AVP), reported amino acid sequence: CYFQNCPRG, MW 1084 Da, was obtained from American Custom Chemicals Corporation (San Diego, CA, USA) and was used without further purification. Oxytocin, reported amino acid sequence: CYIQNCPLG, MW 1007 Da, was obtained from Fluka (Buchs, Switzerland). All solutions were made in the corresponding mobile phase A as indicated in the next section at the specified concentrations.
LC/MS conditions GLP-1 study The ion trap experiments were carried out using an LTQ mass spectrometer (ThermoElectron, San Jose, CA, USA) in positive electrospray ionization (ESI) mode. The SRM transition used was m/z 773.4 (4þ charge state) ! 765.4 with normalized collision energy (CE) ¼ 36%, 1 microscan, 50 ms ion time. The samples were introduced into the mass Rapid Commun. Mass Spectrom. 2008; 22: 1359–1366 DOI: 10.1002/rcm
Application of ion trap technology to LC/MS of large peptides
spectrometer via high-performance liquid chromatography (HPLC) using a Finnigan Surveyor HPLC pump (ThermoElectron) and a CTC PAL autosampler (Leap Technologies, Carrboro, NC, USA). An Atlantis C18, 2 20 mm, 5 m HPLC column (Waters, Milford, MA, USA) with a linear gradient from 0%B to 100%B in 2.5 min and a flow rate of 1 mL/min was used. Mobile phase A was 95:5 water/acetonitrile (ACN), 0.1% formic acid (FA) and mobile phase B was ACN, 0.1% FA. The triple quadrupole MS/MS data were generated using a TSQ Quantum Ultra mass spectrometer (ThermoElectron) by infusion with a T-connection to a 0.2 mL/min flow rate of 50% solvent A/50% solvent B, where solvents A and B were as described above for the ion trap experiments. The collision energy was varied between 20 and 40 V.
Exendin study The ion trap experiments were carried out using an LTQ mass spectrometer in positive ESI mode. The SRM transition used was m/z 1047.6 (4þ charge state) ! 1264.5 with CE ¼ 20%, 1 microscan, 100 ms ion time. The samples were introduced into the mass spectrometer via HPLC using a Finnigan Surveyor HPLC pump and a CTC PAL autosampler. A Zorbax SB 5 m C18 1 50 mm HPLC column (Agilent Technologies, Santa Clara, CA, USA) with a linear gradient from 10%B to 100%B in 3 min and a flow rate of 150 mL/min was used. Mobile phase A was 98:2 water/ACN, 0.1% FA and mobile phase B was 90:10 ACN/water, 0.1% FA. The triple quadrupole data were generated using a Turbo-Ionspray API 4000 mass spectrometer (Applied Biosystems, Foster City, CA, USA). The experimental conditions were optimized to give a maximum response for the detected fragment ion. The collision energy was set at 25 eV with 50 ms dwell time. The transition used was m/z 838.1 ! 949.1.
Vasopressin study The ion trap experiments were carried out using an LTQ mass spectrometer in positive ESI mode. The SRM transition used was m/z 543.2 (2þ charge state) ! 328.2 with CE ¼ 30%, 1 microscan, 50 ms ion time. The samples were introduced into the mass spectrometer via HPLC using a Finnigan Surveyor HPLC pump and a CTC PAL autosampler. An Atlantis C18, 2 20 mm, 5 m column with a linear gradient from 0%B to 100%B in 2.5 min and a flow rate of 1 mL/min was used. Mobile phase A was 95:5 water/ACN, 0.1% FA and mobile phase B was ACN, 0.1% FA. Oxitocin, a structurally related peptide, was used an internal standard.
RESULTS AND DISCUSSION The differences observed in the fragmentation profiles of large peptides are described below with a few practical examples of method development. Different mass spectrometers available in our lab were employed to find the most suitable analyzer and each instrument was optimized to give the best signal. The results may be somewhat confounded by the use of instruments with different ion source designs and so in some cases different charge states were used on the different platforms, but the goal of this Copyright # 2008 John Wiley & Sons, Ltd.
1361
study was not to further investigate the theoretical differences between the MS/MS process in triple quadrupoles and ion traps, but to determine the most suitable mass spectrometer for a particular analysis during method development. The first example involves the development of a quantitative assay to measure GLP-1 (7-36), an insulinotropic hormone, that is of interest in treatment of type 2 diabetes. GLP-1 (7-36) is readily cleaved in the human body via loss of a His-Ala dipeptide to its inactive form GLP-1 (9-36) by the ubiquitous enzyme, dipeptidyl peptidase IV (DPP4). Compounds that inhibit the deactivation of GLP-1 are being investigated as potential drugs that could exploit the therapeutic effects of elevated levels of GLP-1, thus necessitating the development of a sensitive and specific LC/MS assay to measure GLP-1 (7-36) levels. Figure 1 shows the product ion spectra for GLP-1 obtained on a Thermo Quantum triple quadrupole with increase of the collision energy (CE). The top panel (panel A) shows the full scan spectrum for GLP-1 with the two most abundant charged states at m/z 103.5 (þ3 charge state) and 773.0 (þ4 charge state). Panel B, with CE ¼ 20 eV, shows no fragmentation of the isolated precursor ion, þ3 charge state, the most abundant charged state generated by ESI. Panel C shows an increase of CE to 30 eV, leading to 5-fold loss of intensity of the intact precursor ion (from 1.9e8 in panel B to 3.7e7 in panel C) and formation of numerous low intensity product ions. Panel D corresponds to CE of 33 eV, where there is a 10-fold loss of precursor ion intensity (from 1.9e8 in panel B to 1.5e7 in panel D) without generation of a significant fragment ion that can be used for an SRM transition. Panel E corresponds to a CE of 35 eV and a 50-fold loss of precursor intensity (from 1.9e8 in panel B to 3.6e6 in panel E) and formation of numerous fragment ions. In this case, a selected ion monitoring (SIM) detection would be the logical choice; however, due to its low specificity it did not give acceptable sensitivity. Figure 2 shows the fragmentation pattern for GLP-1 (7-36) obtained on a linear ion trap. Panel A shows the full scan MS spectrum, where the most abundant peak corresponds to the 4þ charged state at m/z 773.4. In addition, the 2þ and 3þ charge states are also detected at m/z 1545.1 and 1030.4, respectively. The signal was optimized on the most abundant ion at m/z 773.4. Panel B shows the product ion spectrum at a normalized collision energy of 35%. Contrary to what was observed with the triple quadrupole, the precursor ion is completely dissociated and a simple product ion spectrum is observed with one high intensity fragment peak, an obvious choice for an SRM transition. The ion trap approach was further developed to support an in vitro quantitative assay for detection of GLP-1. With the optimized transition of m/z 773.4 ! m/z 765.4, excellent linearity, reproducibility and low LOQ can be achieved, as shown in Fig. 3. An LOQ of 100 pg/mL (or 30 pM) corresponds to 2 pg on-column. Linearity is observed over 3 orders of magnitude (from 30 pM to 60 nM), which falls in the desired detection range for this assay. The% Difference and% relative standard deviation (RSD) obtained on the linear ion trap are all
