Subscriber access provided by YORK UNIV
Article
Differential Ion Mobility Separations/Mass Spectrometry with High Resolution in Both Dimensions Matthew A. Baird, Gordon A. Anderson, Pavel Vyacheslavovich Shliaha, Ole Nørregaard Jensen, and Alexandre A. Shvartsburg Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b04518 • Publication Date (Web): 13 Dec 2018 Downloaded from http://pubs.acs.org on December 13, 2018
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
Differential Ion Mobility Separations/Mass Spectrometry with High Resolution in Both Dimensions Matthew A. Baird,† Gordon A. Anderson,¶ Pavel V. Shliaha,‡ Ole N. Jensen,‡ Alexandre A. Shvartsburg†,* † ¶
Department of Chemistry, Wichita State University, 1845 Fairmount, Wichita, Kansas 67260, United States GAACE, 101904 Wiser Parkway Ste 105, Kennewick, Washington 99338, United States
‡
Department of Biochemistry and Molecular Biology, VILLUM Center for Bioanalytical Sciences, University of Southern Denmark, DK-5230 Odense M, Denmark ABSTRACT: Strong orthogonality to mass spectrometry makes differential ion mobility spectrometry (FAIMS) a powerful tool for isomer separations. However, high FAIMS resolution and 2-D FAIMS/MS peak capacity were overall achieved only with buffers rich in He or H2. That obstructed coupling to Fourier-transform mass spectrometers operating in ultrahigh vacuum, but exceptional m/z resolution and accuracy of FTMS are indispensable for frontline biological and environmental applications. By raising the waveform amplitude to 6 kV, we enabled high FAIMS resolution using solely N2 and thus straightforward integration with any MS platform - here Orbitrap XL with electron transfer dissociation (ETD) option. The initial evaluation for complete histone tails (50 residues) with diverse post-translational modifications on alternative sites demonstrates broad capability to separate and confidently identify the PTM localization variants in middle-down range. Mass-spectrometric (MS) analyses of complex samples are usually preceded by separations.1,2 These are increasingly effected in the gas phase via ion mobility spectrometry (IMS), comprising linear IMS based on the absolute ion mobility K (normally at moderate electric field intensity, E)3-6 and differential or field asymmetric waveform IMS (FAIMS) relying on the increment of K between two E levels (∆K).7-10 To elicit that, a periodic asymmetric waveform of some amplitude (dispersion voltage, DV) is loaded across a gap between two electrodes through which ions are carried by gas flow. All ions are deflected toward either electrode and would be lost upon neutralization, if not for a DC compensation voltage (CV) superimposed on the waveform.9 Each species is equilibrated and can pass the gap at certain CV determined by the K(E) function, and scanning CV reveals the spectrum of ions entering the gap. For consistency across devices with unequal gap widths, CV is converted to the compensation field EC. The ion size and thus mass (m) for similar species is tightly correlated to the ion-molecule collision cross section Ω and thus (within a charge state) K by the Mason-Schamp equation:11 𝛺=
3
[
2𝜋
16 (𝑘𝐵 𝑇)
1
1
𝑧𝑒
𝑚
𝑀
𝑁𝐾
( + )]1/2
(1)
(where ze is the ion charge, kB is the Boltzmann constant, and T, M and N are the gas temperature, molecular mass, and number density), as the ubiquitous trend lines in IMS/MS plots manifest.12,13 The correlation between m and ∆K is much weaker.14 Thus FAIMS has far greater orthogonality to the MS dimension than linear IMS, for typical peptides and lipids by 3 - 4 fold: ~50% and ~10 - 15%, respectively.14-18 That enables FAIMS to broadly separate structural isomers that are challenging or impossible to resolve by other means including liquid chromatography (LC), such as positional isomers of derivatized aromatics,19 lipids with transposed acyl chains or double bonds,14,20-22
peptides with variant localization of post-translational modifications (PTMs),23,24 single protein conformers,25 and isotopomers.26 Most of these separations need resolving power (RFAIMS) over 100 and often ~200 - 300, which was attained using planar-gap FAIMS devices with homogenous field.27,28 To date, high-resolution FAIMS stages were fitted only to MS instruments of modest resolving power (RMS < 104), primarily quadrupole ion traps (Thermo LTQ series).19,22-26,28 That served to advance the FAIMS technology and exemplary applications, but not address complex natural samples that require higher RMS and mass measurement accuracy (mma) to disentangle and identify all constituents. Modern biomedical and environmental analyses increasingly rely on Fourier-Transform (FT) MS, which delivers the ultimate RMS (105 - 106) and mma (~1 ppm) that permit distinguishing nominal isobars and assigning stoichiometries based on the “exact” m/z up to ~1 kDa.29-31 Both FT Ion Cyclotron Resonance (FTICR) and Orbitrap MS platforms were coupled to FAIMS devices with curved gaps and thus inhomogeneous field.32-35 Those low-resolution systems (RFAIMS ~ 10) can filter compound classes or charge states (e.g., to suppress the chemical noise in bottom-up proteomics by pulling the sought multiply-charged peptides away from 1+ interferences),17,34,35 but generally not provide fine isomer separations like listed above.36,37 Hence, the 2-D peak capacity of FAIMS/MS was capped at 5×106 - much below ~108 created in principle from highest-resolution FAIMS and FTMS platforms. The lack of that combination has been for a reason. High FAIMS resolution generally hinged on buffers with major fraction (40 - 75%) of He or H2 gas.19,22-25,28,38,39 Those are aspirated into the downstream MS stage, but turbo pumps are inevitably less efficient for light molecules with faster Brownian motion. Commercial MS instruments are not intended for that scenario,
ACS Paragon Plus Environment
Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
and consequent rise of pressure throughout the vacuum manifold causes losses of MS resolution, mma, and sensitivity (unless the interlock shuts down the instrument first). Also, a damaging arc discharge in MS hardware can occur as He and H2 have lower electrical breakdown thresholds than N2 or air, and (here on the left branch of Paschen curve) elevated pressure reduces them further.40 These issues apply to all MS systems, but are worse for FTMS that necessitates ~10−10 Torr pressure versus ~10−5 Torr for quadrupoles or Paul traps and ~10−7 Torr for time-of-flight (ToF) instruments.41 That has hindered our early attempts to configure high-resolution FAIMS with FTMS. Here we traverse that conundrum by reaching high resolution with N2 buffer in a new FAIMS system employing the greatestever DV up to 6 kV. That unit can be mounted on any MS instrument without affecting its operation. The tool of choice to detect PTMs and their attachment sites is electron transfer dissociation (ETD) that severs the peptide backbone at every residue without abstracting the weakly bound PTMs.42,43 Inter alia, ETD was deployed to identify the variant peptides separated by FAIMS.24,36,37,44 Present integration with Orbitrap XL (with ETD module) is the gateway to bring highresolution FAIMS to real proteomic analyses. This work focuses on histones - the exceedingly important and intriguing proteins that spool DNA in cell nuclei and govern their transcription and replication via a combinatoric PTM language (histone code).45-49 While histones contain >100 residues, essentially all PTMs reside on the enzymatically cleavable Nterminal domains of ~50 residues (“tails”) protruding from the nucleosome.47,49 Characterization of these tails approaches that of whole histones. Uniquely among separation methods, IMS (including FAIMS) has resolved the isoforms with alternative localization of topical PTMs - methylation (me), trimethylation (me3), acetylation (ac), and phosphorylation (p).24,50 We are revisiting that benchmarked variant set here. Experimental Methods Gas buffers rich in He or H2 enhance the resolution in FAIMS because (a) with planar gaps, RFAIMS is approximately proportional51 to K1/2, and K increases in lighter gases as the reduced mass and Ω (always lower with smaller and less polarizable molecules) in eq (1) decrease; (b) deviations from Blanc’s law in gas mixtures amplify the nonlinearity of K(E) and thus the FAIMS effect.52,53 On the other hand, RFAIMS with planar gap scales9 (to the 1st order) as E3, and adding He or (to a lesser extent) H2 reduces the possible E as stated.38,39 The two factors largely cancel, about conserving RFAIMS along the breakdown E(He%) curve.38,54 That allows replacing He by higher E in systems with both curved and planar gaps.33-35,55 However, the 1.88 mm gap in our high-definition analyzer can sustain40 up to 7 kV in N2, whereas FAIMS drivers were limited to DV = 5 kV.34,35,54,55 As anticipated, that regime (field well under the breakdown limit) has severely constrained resolution. The optimum bisinusoidal waveform is generated using two resonating circuits, tuned to the fundamental frequency (here 1 MHz) and its 2nd harmonic. As keeping a stable waveform caps the consumed power at ~100 W, the DV is limited by the circuit quality factor. Here we increased it by upgrading the electronics (with same mechanical embodiment) to combat losses. For example, all coils are wound from Litz wire to reduce the skin effect losses by maximizing the surface area of conductors. The inductive coupling loop taking energy to the coils is fashioned
Page 2 of 8
from thicker wire and uses specially designed insulators to minimize dielectric losses and couple the primary winding to coil more closely. The coils are cooled by air circulated by multiple fans. These and other modifications allow reaching DV = 6.0 kV (dispersion field ED = 32 kV/cm with normalized field of 130 Td) with peak-to-peak amplitude of 9.2 kV (Figure S1). As previously,19,22,24-26 the FAIMS unit is connected to the LTQ XL ion trap (now joined to the Thermo Orbitrap XL stage) via a custom slit aperture/electrodynamic ion funnel interface. Ions were generated by electrospray ionization (ESI) using a chemically etched glass emitter (20 μm i.d.). Modified H3.1 tails were produced by Fmoc solid-state synthesis and native chemical ligation, and purified by LC.24 Here we explore the set of 15 variants (two me, three me3, five ac, and five p), Table 1. (One me and two me3 species examined 24 by FAIMS in He/N2 were omitted because of sample shortage.) Table 1. Sequence of tail and PTM localizations Sequence
PTM
Positions
ART3K4QT6ARK9S10TG
me
K4, K23
GK14APRK18QLATK23A ARK27S28APATGGVK36 KPHRY41RPGTVALRE (monoisotopic 5337 Da)
me3
K9, K23, K36
ac
K9, K14, K18, K27, K36
p
T3, T6, S10, S28, Y41
The individual standards or mixtures were dissolved to 5 μM in 99.9:0.1 water/formic acid. Samples were infused to the ESI source at 0.5 μL/min. The carrier gas (UHP N2) was formulated by digital flow meter26 at the rate of Q = 0.9 - 3 L/min and passed through a universal filter. The EC scan speed was 1.3 1.8 V/(cm × s). The EC scale was anchored utilizing the 4+ ion of C-terminal tail segment CRKSAPATGGVKKPHRYRPGT VALRE (monoisotopic 2834 Da) with confirmation by variant mixtures. (We avoided the unmodified tail calibrant24 to preclude artifacts due to the hypothetical PTM detachment upon field heating of ions in FAIMS). The ETD data were recorded by selecting the CVs of major peak apexes and fragmenting the transmitted ions.24,44 In MS and MS/MS modes, we set RMS = 30,000 and achieved that at the top of charge state envelope (z ~ 10), Figure 1a. Results and Discussion Demonstration of capability The most impressive isoform separations24 were for me variants, where the PTM made just 0.25% of the 5.5 kDa tail mass. The K4me and K23me forms were resolved near-baseline for z = 10 by both linear IMS50 and FAIMS (using 65% He).24 The FAIMS spectra for these species in N2 (Figure 1b) at DV = 5.9 kV and Q = 1.4 L/min resemble those in He/N2 medium,24 with current EC values higher by ~10% (and EC for K23me above that for K4me by same 6 - 7%). The increase of EC and associated peak spacing upon substitution of augmented field for He is expected.38,55 That often does not improve RFAIMS as the peaks also widen (because of lower K values in N2),38,51,55 here from 1.3 to 2.1 V/cm for the full width at half maximum (w). Still, we achieved baseline resolution as the minor feature of K23me (that coincided with the major K4me peak)24 has van
ACS Paragon Plus Environment
Page 3 of 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
(a)
R 30 K
K4me 10+ Q = 1.4 L/min
536.5
m/z
(b)
B 5.4 kV
4.8 kV B
B
K23me 10+
5.9 kV
C C C
A
537.0
A A
Q = 3 L/min Q = 2 L/min Q = 0.9 L/min
B
C
(c)
60
80
100
120
140
160
EC, V/cm
A D
B1 C1 130
140
150
Figure 2. Spectra for K4me/K23me mixture (z = 10) at three DV values as labeled (the trace at 5.9 kV is from Figure 1c).
160
170
EC, V/cm
Figure 1. Data for K4me and K23me tails: mass spectrum (a) and normalized FAIMS spectra (z = 10) for individual standards (b) and their 1:1 mixture at three Q values (c). ished. Such spectral evolution depending on the DV or gas composition is common for large peptides, reflecting their conformational annealing induced by stronger field heating at higher DV or He content.23,25 This phenomenon can be exploited to disentangle merged variants by adjusting the trade-off between DV and He % while retaining high resolving power. The resolution in FAIMS depends on the filtering time t that scales as 1/Q.51,56 Extending t from ~200 to ~300 ms by throttling56 Q to 0.9 L/min narrows the peaks to w = 1.4 V/cm and splits the tail of K4me and front of K23me from major peaks B and C into distinct features B1 and C1 (Figure 1c). Conversely, reducing t to ~140 and ~100 ms by increasing Q to 2.0 and 3.0 L/min broadens the peaks and worsens resolution. The feature D comes from the K9me impurity (in mixture only) that may also contribute signal between B and C,24 hence K4me and K23me may be resolved better than appears. However, free diffusion causes the ion flux through the gap to decay exponentially51 with t. The actual drop-off is yet faster (here by 100-fold from 3.0 to 0.9 L/min), probably as “self-cleaning” eliminates flexible species that change conformation and thus EC in the gap.25,57,58 Indeed, compact polypeptides generally have greater EC than elongated conformers59,60 and K23me is more compact than K4me by linear IMS data.50 Then the diminishing C/B intensity ratio at longer t agrees with self-cleaning upon progressive peptide unfolding during the FAIMS process. This K4me/K23me mixture is separated only partly at lower DVs of 4.8 and 5.4 kV (Figure 2), showing the benefit of present record DV. The EC values are expected9 to scale as ~ED3. In fact, the measured EC for peaks A and B increase by 1.87× between DV of 4.8 to 5.9 kV - versus 1.86× in theory. This further validates proper FAIMS operation up to DV = 6 kV. Performing ETD pre-supposes plenty of precursor because of limited total efficiency and partition of signal into many dissociation channels (especially for larger peptides). This moves the
optimum resolution/sensitivity balance of FAIMS in conjunction with ETD toward the latter, to the degree controlled by overall signal attenuation during ETD. Here, we got quality results at Q = 2 L/min. The MS resolution constraints of ion trap had compelled long ETD reaction times tETD (up to 150 ms) to alleviate the spectral congestion by taking all fragments to low z through consecutive ETD steps.24 This is moot with present MS resolution, where fragments of all z from single step can be disentangled. The fewer steps and shorter tETD (15 ms) have significantly raised the product counts, yielding informative fragments with excellent isotopic envelopes in 0.98, Figure 5), though admittedly the insight from r2 with just three points is limited. Of more relevance, the regressions are parallel to those for ac and p variants. The separations for latter with z = 8, 9, 11, 12 broadly follow those in He/N2 buffer24 by visual inspection, but multiple peaks of close height with the top flipped by minute experimental changes or between technical replicates render that hard to quantify. However, there are divergences in detail. For instance, S28p 9+ had24 one major peak h co-eluting with S10p (Figure 4). Beyond that, here we see a large peak i at higher EC. Likewise, T3p 11+ now features, in addition to the peak at lower EC k, another main peak l co-eluting with S28p. Some present spectra are, to the contrary, simpler, e.g., that for T3p 12+ comprises one peak rather than two previously.24 These variations must originate from distinct annealing endpoints ensuing from unequal magnitude of field heating in the N2 and He/N2 buffers. Again, this may facilitate resolving specific variants in spite of lower formal RFAIMS, e.g., S28p is now easily separated from other isoforms at its peak apex in z = 9, which was unfeasible24 for any z using He/N2. The new z = 6 and 7 provide extra separation options. For example, S28p and S28p/Y41p mix is cleanly filtered near peak apexes using 6+ and 7+, respectively. The results over wider z = 6 - 12 range show the variant resolution to be worst at z = 8 with maximum conformational heterogeneity, improving at lower and higher z. This mirrors the picture in linear IMS,50 where unfolding (driven by Coulomb repulsion) was observed at z = 8. Such transitions involve57,61,62 numerous intermediates trapped along the pathways from more delineated compact to elongated conformers at respectively lower and higher z. For analytical purposes, this implies two charge state regions conducive to proteoform separations in both IMS methods. Conclusions and Future Directions We established high-resolution FAIMS in N2 gas using novel electronics to output stable waveform with amplitude up to 6 kV. Eliminating He or H2 from the buffer enables immediate coupling to high-resolution Fourier-Transform mass spectrometry on the Orbitrap instrument. (Some FTICR platforms with multiple differential pumping stages can deal with limited He loads with little consequence, but production Orbitraps cannot.)
ACS Paragon Plus Environment
Page 4 of 8
Page 5 of 8
Trimethylated
Acetylated
Phosphorylated
K9 K23 K36
K9 K14 K18 K27 K36
T3 T6 S10 S28 Y41
The MS resolving power of 30,000 here is enough for many proteomic assays even when higher RMS is available.63 This RMS is also provided by ToF systems 6+ that operate at lower vacuum and thus can typically accept greater He or H2 loads. However, same FAIMS unit is compatible with all Orbitrap and FTICR instruments, 29-31,35,64 with RMS 7+ including 6those up to ~10 . The combination of high resolution in each dimension and their mutual orthogonality endows such platforms with unprecedented 2-D IMS/MS peak capac8+ ity. That is particularly pertinent as integrating linear IMS to FTMS has been impeded by mismatch between long MS timescale (mani h datory to precisely determine the ion cyclotron frequencies) with 9+ fast transients arriving from dispersive (drift tube, traveling wave, or trapped) IMS devices.6470 a b c e f g d The brute-force solution of turning IMS into a filtering technique via the tandem-gate arrange10+ ment66,70 means (at high resolution) a duty cycle of ~0.1%. That drastically impairs sensitivity: l k e.g., a drift-tube IMS/Orbitrap system had70 LoD ~20 pmol (4 µM concentration) on the precur11+ sor level even at the minimum RMS = 17,500. While that problem can be mitigated by sophisticated frequency-modulation,68 selective accumulation,67 oversampling,64 69 12+ and nonlinear scan schemes, the duty cycle for targeted high-resolution linear IMS stays69 below 1% (versus 100% for FAIMS). In 120 130 140 150 160 170 180 120 130 140 150 160 170 180 120 130 140 150 160 170 180 any event, the fundamental correEC, V/cm lation between the linear IMS and Figure 4. The FAIMS spectra for me3, ac, and p variants with z = 6 - 12 (as labeled) measured MS dimensions persists. using DV = 5.9 kV and Q = 1.4 L/min. The new platform was tested in MS and MS/MS analyses of the set of histone tail variants representative of proteoforms in the middle-down range (~5.5 kDa). All variant separations aligned with those in He/N2 buffer,24 but average resolving power decreased by ~30%. Still, some variants got resolved better upon spectral changes due to conformational annealing (induced by field heating), and all were fractionated at least into binary mixtures amenable to complete characterization by ETD. As high MS resolution lifted the peak congestion, we could robustly assign ETD fragments without charge reduction. That permitted cutting the reaction time tenfold, raising the signal to reach s/n ~ 100 in same 10-min acquisition or compressing that to
