High Mass Analysis with a Fourier Transform Ion Cyclotron Resonance Mass Spectrometer: From Inorganic Salt Clusters to Antibody Conjugates and Beyond

Abstract

Analysis of proteins and complexes under native mass spectrometric (MS) and solution conditions was typically performed using the time-of-flight analyzers, due to their routine high m/z transmission and detection capabilities. However, over recent years, the ability of Orbitrap-based mass spectrometers to transmit and detect a range of high molecular weight species is well documented. Herein, we describe how a 15 Tesla Fourier transform ion cyclotron resonance (15 Tesla FT-ICR) instrument is more than capable of analyzing a wide range of ions in the high m/z scale (>5000), in both positive and negative instrument polarities, ranging from the inorganic ceasium iodide salt clusters; a humanized IgG1k monoclonal antibody (mAb; 148.2 kDa); a IgG1-mertansine mAb drug conjugate (148.5 kDa, drug-to-antibody ratio, DAR 2.26); a siRNA-IgG1 mAb conjugate (159.1 kDa; ribonucleic acid to antibody ratio, RAR 1); the membrane protein aquaporin-Z (97.2 kDa) liberated from a C8E4 detergent micelle; the empty MSP1D1-nanodisc (142.5 kDa) and the tetradecameric chaperone protein complex GroEL (806.2 kDa; GroEL dimer at 1.6 MDa). We also investigate different regions of the FT-ICR which impact ion transmission and desolvation. Finally, we demonstrate how the transmission of these species and resultant spectra are highly consistent with those previously generated on both quadrupole time-of-flight (Q-ToF) and Orbitrap instrumentation. This report serves as an impactful example of how FT-ICR mass analyzers are competitive to Q-ToFs and Orbitraps for high mass detection at high m/z.

TOC

The TOC briefly summarizes the diverse samples that can be analyzed on a commercial 15 Tesla FT-ICR MS instrument

Introduction

Since the initial nano-electrospray ionization (nESI) mass spectrometric (MS) experiments demonstrating the preservation and transmission of biomolecular complexes from solution into the gas-phase, under native conditions over two decades ago [1–4], native-MS has become an increasingly important research tool in both industry [5–8] and academia [9–11]. As an analytical technique, native-MS lends its self particularly well to accurate protein complex molecular weight (MW) [10, 12] and subunit stoichiometry determination [13, 14]; protein-ligand stoichiometry and associated binding constant determinations [15]; gas-phase protein conformation measurements [14, 16] and gas phase stability assessment [17, 18]. Most recently, significant progress has been made characterizing membrane proteins, nanodiscs and associated complexes [19–22].

Incremental instrument improvements to MS performance, in the form of restricting ion source pumping [3, 23, 24]; source ion guide sleeves [25]; gas-bleed valves in to the source hexapole ion optics [24]; the use of more massive, mono and polyatomic collision gases [26–29]; low frequency quadrupoles [24] and lower frequency time-of-flight (ToF) pusher optics [24]. All of in the above modifications led to improved high m/z transmission, detection (> m/z 5000) and spectral quality, in terms of ion transmission and protein complex desolvation [30, 31] as well as preservation of multimeric protein stoichiometries [29, 30], some of which are in the MDa range [32–34].

Since the early 2000s, all of the aforementioned instrumental improvements were implemented exclusively on ToF [23] or quadrupole-ToF (Q-ToF [4, 24–27, 35]) instrumentation. During the subsequent decade of native-MS applications research and development, very little improvement to the ion source and ion optics were realized, specifically addressing charge state peak width, as a function of efficient ion desolvation; the charge states remained broadly adducted (much wider than the theoretical peak width; [30]). Previous native-MS experiments performed on the Q-ToF instrument, improved ion desolvation and cooling occurred in the ion source [3, 24, 27] and in the hexapole [25] or travelling wave [29] collision cell, using large monoatomic (xenon, Xe) or polyatomic (sulpurhexafluoride, SF₆) collision gases. The ions were still far from fully desolvated, adding approximately 0.5% to 1.0% to the MW [30].

Only recently has the study of proteins and complexes under native-MS and solution conditions been transferred to trapping-based instruments such as the Orbitrap [8, 21, 31, 36–38] and the benefits been realized. In 2012 and 2013, Rose [36] and Kelleher [37] respectively, demonstrated a paradigm shift in native-MS protein spectral quality. Close to theoretical peak width distributions for GroEL [36] and pyruvate kinase [37] charge sates and base-line separation of sequential additions of nucleotide and sugar ligands, respectively, were achieved on the Orbitrap-EMR instrument. This level of spectral quality implied a superior level of ion desolvation, assumed to be occurring in the instrument source and HCD cell. Similar spectral quality was also achieved for monoclonal antibodies (mAbs) [36]. It must be noted however that Heck, in 2004, documented native-MS data acquisitions of vanillyl-alcohol oxidase on a modified linear ion trap [39] (LCQ-Deca) reporting well resolved charge states at m/z values 10,000 to 12,000. It is not absolutely clear which regions of the Orbitrap instrument are critical for the observed superior ion desolvation currently being described within the literature, as Rose [36] and Gault [31] have demonstrated that the HCD collision gas and HCD cell voltage, respectively, are important factors that should be considered for improved native protein desolvation and transmission. Further insights into the superior spectral quality can also be ascertained from Lange [40] where m/z data is displayed as combined absorption and magnitude mode spectrum from an apodised transient, therefore an improvement in spectral baseline noise levels and resolution are achieved by signal processing, however this is beyond the scope of this brief discussion and will be described in detail in a subsequent publication. A less frequently used instrument for native-MS analysis is the Fourier transform ion cyclotron (FT-ICR) and multiple groups have demonstrated its utility for mAbs, nanodiscs, membrane proteins and large multimeric complexes [7, 21, 41–44].

Herein, we present the use of a 15 Tesla FT-ICR instrument to acquire extended m/z range data of the humanized NIST IgG1k mAb; caesium iodide (CsI) clusters; the detergent encapsulated membrane protein aquaporin (AqpZ); an IgG1-siRNA conjugate, an IgG1-mertansine drug (DM1) conjugate; an empty MSP1D1-nanodisc (MSP1D1-Nd); and the tetradecameric chaperone protein complex GroEL, all under native-MS and solution conditions on an unmodified instrument. We have also investigated the regions within the FT-ICR instrument which may afford improved levels of mAb desolvation such as the source ion transfer capillary temperature, source skimmer voltage, collision cell voltage, collision cell and ICR-cell trapping times and monoatomic versus polyatomic collision gases. Finally, we highlight the spectral quality is highly comparable to that obtained on the Q-ToF and Orbitrap instruments, for similar protein complexes.

Material and Methods

Humanized IgG1k mAb was acquired from NIST (RM-8671[45]) at a stock concentration of 10 μg/μL. CsI was acquired from Sigma-Aldrich and a MS-working solution of 50 μg/μL in 50% (v/v) acetonitrile was used (no formic acid). AqpZ was purified and prepared using a previously described method [22]. The antibody drug conjugate (ADC) is an IgG1 covalently modified (via native lysine residues) with DM1 [46]. The siRNA-IgG1 SEFL2 [47] conjugate was prepared and purified as previously described [48]. Briefly, the IgG1 (SELF2 format) was conjugated with duplex siRNA (7.5 kDa) via an engineered cysteine. The empty MSP1D1-Nd was assembled and purified as described previously [21]. GroEL was purchased from Sigma-Aldrich and prepared as previously described [29]. All protein samples were chromatographed through an 5/150 GL Superdex column (GE Healthcare; 25 mM Tris, 100 mM NaCl) as a final purification step and re-concentrated to a low volume, high concentration stock. Immediately prior to native-MS, samples were buffer exchanged into 200mM ammonium acetate (0.5 % v/v C8E4, 2 x CMC, for AqpZ) using a BioRad P6 spin column. All working native-MS solutions were approximately 5 μM.

Mass Spectrometry

All data were acquired on a Bruker SolariX 15T FT-ICR MS system (Bruker Daltonics, Germany) operated in positive nESI mode over the m/z range 153 to 20,000 at 512 k (and negative nESI mode for the siRNA-IgG1 conjugate) equating to a resolution value of 4100 FWHM at m/z 6108.3 ([Cs₂₃I₂₃]Cs⁺). Argon (Ar) or Sulphurhexafluoride (SF₆) were used as the collision gas. All samples were infused into the MS system using gold-coated glass capillaries (Waters narrow thin-walled M956232AD1-S) using the Bruker nESI source. Typical instrument voltage and pressures are noted in either the discussion or the figure legends. All data were acquired in magnitude mode and the free induction decay (FID) transients were processed using a symmetrical Hann (F = 0.5) function. Additionally, the effect of FID transient apodization is demonstrated in Figure S1 (Supporting Information). Detailed instrument parameters are documented in the Supporting Information.

Results and Discussion

High m/z and Mass Optimization on the 15 Tesla FT-ICR Instrument

There are a number of different instrument parameters which can influence the level of ion desolvation and transmission, such as the ion transfer glass capillary temperature, applied source skimmer and collision cell voltage, collision cell storage and ICR residence times (Figure 1a), all of which were systematically investigated using the NIST IgG1k mAb standard.

Figure 1.

The 15 Tesla SolariX instrument schematics and measured effect of highlighted instrument regions on NIST mAb glycoform desolvation and transmission: a) The specific regions tested and optimized for high MW and high m/z transmission are highlighted in red. Heated ion transfer capillary (30 to 100 °C); source Skimmer 1 activation (25 to 100 V); collision cell activation (0 to 30 V); collision gas (Ar or SF₆ indicated pressures 1.5 to 5.2e⁻⁶ mbar); collision cell accumulation time (0 to 5 s) and ICR detection delay time (0 to 5 s). Image courtesy of Bruker Daltonics, Billerica, MA, USA; b) native-MS spectrum of the NIST IgGk mAb; c-e) the effect of the ion transfer capillary temperature (CID and Skimmer1 held constant at 30 V and 50 V, respectively); f-h) the effect of source Skimmer 1 voltage (CID and capillary temperature held constant at 0 V and 100 °C, respectively); i-k) and the effect of collision cell voltage (Skimmer 1 and capillary temperature held constant at 50 V and 100 °C, respectively) on desolvation, separation and transmission of the NIST IgG1k ion series m/z 5470–5520 z = 27+, glycoforms G0F/G0F, G0F/G1F, G1F/G1F, G1F/G2F and G2F/G2F. All data normalized to an ion count of 8.0e⁹. Zoom-ins are included for low intensity spectra. All data displayed in magnitude mode (512 kW) using the Hann windowing function of F = 0.5.

The effect of the Source Temperature, Source Skimmer and Collision Cell Voltage:

The NIST IgIk mAb [49] (Figure 1b) is represented by five well resolved glycoforms, G0F/G0F, G0F/G1F, G1F/G1F, G1F/G2F and G2F/G2F for each charge state (Figure 1e). Figures 1c to 1e display the effect of the ion source transfer capillary gas temperature (30 °C, 60 °C and 100 °C) on glycoform desolvation and transmission for charge state z = 27+. Increased ion transfer capillary temperature results in reduced levels of adducting. This reduction in charge state adducting and therefore desolvation can also be achieved by increasing Skimmer 1 (Figures 1f to 1h) and the collision cell voltage (Figure 1i to 1k). Improved ion transmission is also observed in all cases; the ion current is spread over less channels/adducts. Skimmer 1 voltage is less efficient at removing adducts. For example, 100V applied to the skimmer (CID set to 0 V) is required for efficient adduct removal and transmission (Figure 1h) as opposed to 30 V applied to the collision cell (Figure 1k; skimmer set to 50 V) where increased transmission is observed. These changes in levels of charge state adducting as a function of ion activation are consistent with those previously reported on different MS instruments [30, 31, 36] where higher activation voltages were required for improved transmission, desolvation and MW determination of large protein complex ions. Additionally, Figure S2 (Supporting Information) graphically demonstrates the improvement in ion desolvation (and peak width FWHM).

General Source Considerations:

Increasing the source backing pressure and subsequent improved high m/z ion transmission by increasing both radial and axial collisional cooling was previously documented [23–25]. Typically the source backing pressure of the Z-Spray ion source (Waters Corporation) is increased from the standard operating pressure of 1.5–2.0 mbar to 6.0 mbar, with dramatic effects [23, 30]. For efficient high m/z transmission on the FT-ICR described herein (Figures 1b–k and 2a–f) the source pressure was not adjusted (held constant at an indicated pressure of 2 mbar). This indicates that a significant amount of ion cooling occurs in the heated ion transfer capillary and the region of the source immediately prior to entry into the first ion funnel. It is also conceivable that the ion funnels are very efficient at focusing and transmitting ions of high m/z value. The pressure in the ion transfer capillary is not measured on the commercial 15 T SolariX instrument. However, it is assumed that the pressure is significantly higher than the ion funnel region which is held at an indicated 2.0 mbar.

Figure 2.

A selection of proteins and complexes displaying charge states over a wide m/z range, acquired under standard instrument voltages and pressures. All data displayed in magnitude mode using the Hann windowing function of F = 0.5. a) CsI, 50 μg/uL; b) the tetrameric AqpZ complex liberated from a C8E4-micelle, 10 μM (1RC2, downloaded from www.rcsb.org); c) a siRNA-IgG1 conjugate, 5 μM (displaying a single cysteamine modification); d) an IgG1-DM1 conjugate, 5 μM; e) the empty MSP1D1 nanodisc, 10 μM and e) the tetradecameric chaperone complex GroEL, 5 μM (4V43, downloaded from www.rcsb.org). All data acquired over the m/z range 153–20,000 (m/z 1000 to 25,000 for CsI). Transient times ranged from 0.04 s (128 kW) to 0.17 s (512 kW). All spectral zoom-ins and FIDs are displayed in Figures S10 and S11 (Supporting Information).

Monoatomic versus a polyatomic collision gases:

Historically, larger more massive, monoatomic collision gases (Ar, Kr, Xe) were used for small molecule ion trap CID experiments [50]. More recently, monoatomic and polyatomic collision gases (Xe, N₂, SF₆) were reported to improve collisional focusing, cooling and transmission of large monomeric and multimeric protein complexes on Q-ToF instruments, in both MS and tandem-MS modes of operation [25] [26, 28, 29, 51]. In 2012, Rose demonstrated that Xe could be used in the HCD cell of the Orbitrap EMR instrument [36]. The larger, polyatomic collision gas SF₆ as opposed to Ar, was selected to test transmission gains on the FT-ICR described herein. The results demonstrated a modest improvement in ion transmission (~10%; Figures S3a–c, Supporting Information) for the NIST IgG1k mAb. However, the signal improvement for CsI cluster transmission was dramatic (Figures S3d–f; Supporting Information). One important point to note is that the SolariX quadrupole has an effective upper selection limit of m/z 6000, making it impossible to test the effects of lager polyatomic collision gases in protein complex tandem-MS experiments [28].

Collision Cell and ICR trapping times:

There were a number of gas-phase trapping experiments performed on proteins and their structural evolution was measured. The early work of Clemmer [52] demonstrates the gas-phase structural transition of ubiquitin was measured as a function of trapping time (10 ms to 30 s) prior to ion mobility. In addition the work of McLafferty [53] demonstrates trapping cytochrome-c in a 7 Tesla ICR from picoseconds to minutes and probing the structure by electron capture dissociation. There are two specific regions within the 15 Tesla SolariX where ions are trapped and accumulated prior to release or detection, the collision cell and the ICR cell. It was hypothesized that trapping times within different regions of the 15 Tesla FT-ICR may have an impact on the levels of ion desolvation. The increased residence time within the ICR cell was achieved by setting a delay prior to ion detection at 512 kW. Additionally, a low level of declustering potential was induced by raising the source skimmer to 75 V (CID was set to 0 V) only to allow for observable glycoform separation. Figures S4 (Supporting Information) demonstrate the effects of increasing trapping times and residence times within the trap hexapole and the ICR cell respectively. In all cases, increasing the times from 0.0 sec to 5.0 sec had no measurable effect on the level of adducting/declustering of the detected charge states. Due to lack of improvement over 0.0 to 5.0 sec, longer trapping times were not investigated.

High Mass Analysis

Figures 2a–f demonstrate the high mass and high m/z transmission capabilities of an unmodified commercial 15 Tesla SolariX FT-ICR MS instrument. CsI represents a sample which displays multiple ion clusters and charge states (z = 1+ to 5+ [29]) over a wide m/z range 500 to 20,000. The peak series labelled [Cs₂₉I₂₉]Cs⁺, [Cs₃₃I₃₃]Cs⁺, [Cs49I49]Cs⁺, [Cs₆₂I₆₂]Cs⁺ and [Cs₇₄I₇₄]Cs⁺ represent the stable cubic clusters (magic-numbers [54, 55]). They also display signal enhancement due to constructive overlap [29, 55], with corresponding m/z values 7667.5, 8706.8, 12,863.9, 16,241.8 and 19,359.8. This spectrum is highly consistent with those previously reported on Q-ToF [24, 29, 55] and Orbitrap [36] systems and historically, on a 3 Tesla ICR instrument [56]. However, ions above m/z 20,000, as previously reported by Lebrilla [57] are not observed. Since the instrument described herein (15 T SolariX FT-ICR) is commercial, it is possible enhancements may be required to improve ultra-high m/z transmission, such as increased source pressure for improved collisional cooling and ion transmission of the larger CsI clusters. It is also interesting to note that the level of CsI cluster detection between m/z values 13,000 to 20,000 is sparse compared to data previously described [24, 29, 55]. Only the larger stable clusters are detected (m/z 16,241.8 and 19.359.8, as less stable clusters may not survive the transmission and subsequent detection within this instrument.

Over the past fifteen years the Q-ToF instrument was the instrument of choice for native membrane protein analysis [58, 59]. Only since 2015, was the Orbitrap used for membrane protein and nanodisc analysis [31, 60]. However, the FT-ICR is more than capable of analyzing membrane proteins liberated from a detergent micelle [21, 61]. Figure 2b shows the tetrameric water channel AqpZ liberated from the detergent C8E4, as a function of skimmer and collision cell voltage (50 V and 30 V respectively). Between m/z 5500 to 7000, a series of well resolved charge states (z = 12+ to 18+) correspond to a tetrameric MW of 97.2 kDa, consistent with previously reported data in C8E4 [62]. Each charge state consists of an additional superposition of peaks corresponding to multiple N-terminal formylation modifications [61] (Figure S5, Supporting Information).

Two therapeutically relevant mAb conjugates were chosen: a siRNA conjugate [48] and a DM1 [46] conjugate, analysed in negative and positive nESI modes, respectively. The position of the charge states in the m/z scale differ from negative to positive modes of operation: m/z 7,000–9,000 (charge states z = 24- to 19-) for the siRNA-IgG1 conjugate and m/z range 5000–7000 (charge states z = 28+ to 22+) for the mAb-DM1 conjugate. The IgG1-siRNA conjugate is in the ribonucleic acid-to-antibody ratio of 1 (RAR1). Since the siRNA duplex is highly negatively charged, it was decided to measure and display this molecule in negative nESI mode. This siRNA-IgG1 conjugate was demonstrated to ionize equally well in positive nESI mode [63]. The IgG1-DM1 conjugate displays up to 8 DM1 covalent additions (Figure 2d), all of which are baseline separated. The calculated DAR for this native nESI FT-ICR data is 2.26, compared to a previously calculated DAR value 3.1 to 3.3 [46]. The differences between native-MS and denaturing-MS derived DAR values have previously been addressed [7, 64]. In both cases, mild voltages are used to induce efficient ion desolvation (Skimmer 1 50V; collision cell 30V).

Figure 1e represents the empty MSP1D1 nanodisc (MW 142.5 kDa) and was described elsewhere, using different data processing parameters [21, 42]. Briefly, a highly polydisperse spectrum is displayed over the m/z range 6000–8000, representing charge states in the range of z = 21+ to 24+ (Figure S6, Supporting Information). The use of a symmetric Hann apodization function (F = 0.5) improves the spectral quality over that previously described [21, 42].

The chaperone protein GroEL (Figure 2f) represents the largest molecule analyzed within this study (806.2 kDa) with charge states ranging from z = 79+ to 70+ detected between m/z 10,000 and 12,000. The GroEL dimer is also observed in the m/z range 13,500 to 15,000 (Figure S7, Supporting Information). The individual tetradecamer charge states are fully baseline resolved. However, the resolution, or more precisely, the peak width, previously reported on the Orbitrap-EMR [36, 37], is not observed here. There may be several reasons for this. First, the GroEL sample analysed here displayed multiple N-terminal methionine oxidations (Figures S8a–d, Supporting Information), which leads to sample micro-heterogeneity, therefore broader individual charge state peak width. It is also possible that the Orbitrap source and HCD cell are more efficient at ion desolvation. Furthermore, the GroEL construct and purification procedure were most likely different. Figure S9 (Supporting Information) displays GroEL analysed at increasing levels of instrument resolution (32 kW to 4 MW). In the context of the aforementioned results, it is important to note that the Orbitrap data are displayed as a function of mixed absorption and magnitude modes [40] resulting in increased spectral resolution (approximately 1.7 times [65, 66]). Figures S10 and S11 (Supporting Information) display zoom-in m/z regions and FID transients for all spectra discussed above.

Conclusions

The 15 Tesla SolariX FT-ICR MS is more than capable of transmitting and detecting high m/z ions and high MW species efficiently, and importantly, on an unmodified commercial instrument. The data presented herein are highly consistent with historical data acquired on Q-ToF and more recently Orbitrap-EMR and UHMR instruments. Additioally, there is an increased level of therapeutic biomolecule MS characterization required from regulatory agencies [67] [68] [69] [70] therefore, the data presented herein further demonstrates that the 15 Tesla SolariX is a bona fide platform for membrane protein, protein complex and pharmaceutically relevant protein (under native-MS and solution conditions) and high MW analysis.

The highest detected and assigned ion was CsI, m/z 19,359.8, [Cs₇₄I₇₄]Cs⁺. The highest MW detected was the GroEL dimer measured at 1.61 MDa, consistent with calculated trapping limit m_critical = 1.20607×10⁷zB²a²/V_trapα [44, 71]. Additionally, Li recently demonstrated electron transfer dissociation charge reduced generated ions of m/z values 40,000 detected on a 15 Tesla magnet [44]. However, one must note that GroEL can also be detected on a smaller 7 Tesla magnet (Figure S12). The NIST IgG1k mAb and the IgG1-DM1 conjugate demonstrated baseline resolved glycoforms and DM1 covalent modifications, respectively, at m/z values of 5000–6500. The empty MSP1D1-Nd spectrum is highly polydisperse and consistent with previously published data. The membrane protein AqpZ liberated from the C8E4-micelle can be efficiently transmitted and detected as an intact tetrameric complex. All the aforementioned species were measured under nESI positive instrument polarity. The siRNA-mAb conjugate was measured effectively under negative instrument polarity. Data quality, in terms of charge state peak widths are significantly improved when the FID transients are apodized using a symmetric Hann function (F = 0.5).

To achieve efficient protein complex and high charge state transmission, specifically desolvation for reduced levels of charge state adducting, instrument parameter optimizations were minimal. The most effective optimizations were increasing the source Skimmer 1 voltage (up to 100 V) in combination with mild collisional activation (collision cell; up to 30 V). Operating the source ion transfer capillary to 100 °C resulted in maximum charge state desolvation. Using SF₆, as opposed to Ar, led to (species dependent) improvements in high m/z and high MW transmission. Increasing ion residence times within the hexapole trap and the ICR cell, made no measurable difference to levels of native charge state desolvation.

Finally, we wish to clarify that care should be taken when referring to native-MS data as high resolution. High resolution implies obtaining isotopic resolution and observing isotopic fine structure. To date, most, if not all Orbitrap-EMR (and UHMR) native protein data lack evidence of isotopic fine structure. These are typically acquired at a spectral resolution ranging from 3200 FWHM (12 Hz) to 25,000 FWHM (1.5 Hz) at m/z 6000 [36], where many of the charge states are detected; this is far from high resolution data. We therefore suggest using the terms “highly desolvated” or “superior spectral quality” to describe the native-MS protein data currently published and acquired on both Orbitrap and FT-ICR instruments.