Differential Ion Mobility Separations in the Low-Pressure Regime

Presently near the lower end of ν range, we expect only a minor effect. The system ... widths are conserved (~0.2 - 0.3 Td) at all ED/N (in line with...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/ac

Cite This: Anal. Chem. XXXX, XXX, XXX−XXX

Differential Ion Mobility Separations in the Low-Pressure Regime Alexandre A. Shvartsburg,*,† Anisha Haris,‡ Roch Andrzejewski,‡ Andrew Entwistle,‡ and Roger Giles‡ †

Department of Chemistry, Wichita State University, 1845 Fairmount, Wichita, Kansas 67260, United States Shimadzu Research Laboratory, Wharfside, Trafford Wharf Road, Manchester M17 1GP, United Kingdom



ABSTRACT: Ion mobility spectrometry (IMS) in conjunction with mass spectrometry (MS) has emerged as a powerful platform for biological and environmental analyses. An inherent advantage of differential or field asymmetric waveform IMS (FAIMS) based on the derivative of mobility vs electric field over linear IMS based on absolute mobility is much greater orthogonality to MS. Effective coupling of linear IMS to MS and diverse IMS/MS arrangements and modalities impossible at ambient buffer gas pressure were enabled at much reduced pressures. In contrast, FAIMS devices operate at or near atmospheric pressure, which complicated integration with MS. Here, we show FAIMS at ∼15−30 Torr using a planar-gap stage within the MS instrument envelope. Fields up to ∼300 Td permitted by the Paschen law at these pressures greatly raise the separation speed, providing fair resolution in ∼10 ms and FAIMS scans in under 5 s. Rapid separation and efficient ion collection at low pressure minimize losses in the FAIMS step. Separations for key analyte classes and their dependences on electric field mirror those at ambient pressure. The potential for proteomics is demonstrated by separations of isomeric peptides with variant localization of post-translational modifications.

D

translational modifications, PTM),13−16 metabolomics (to distinguish isomeric lipids or carbohydrates),17,18 imaging (to discriminate against matrix interference),19 food quality monitoring,20 pharmaceutical studies,21 and isotopic analyses.22 As the K(E/N) derivative is correlated to ion mass (m) weaker than K itself,14,17,23 FAIMS is much more orthogonal to MS rather than linear IMS and makes an exceptional tool for fine isomer separations. The IMS invented in the early 1900s was a standalone technique with ions separated at ambient pressure and detected by a charge collector.24 The emergence of IMS at low pressure (P < 10 Torr) coupled to MS in the 1960s transformed the field9,25,26 through hugely advancing the grasp of IMS principles by allowing extraordinary E/N up to ∼700 Td (where highfield effects dominate), ensuring unequivocal ion assignments that revolutionized research into the nanocluster27,28 and biomolecular29−31 structure, permitting ion annealing by injection into gas at a controlled energy32 prior to IMS, and addressing ions from sources optimally run in vacuum or at low pressure (such as matrix-assisted laser desorption ionization).33 Of practical importance, the rf ion focusing in Dehmelt pseudopotential has so far succeeded (at pertinent m/z) below ∼30 Torr only.34 So concurrent35 or subsequent36 radial focusing can compress ion packets in low-pressure IMS to the tube axis (countering the diffusion and Coulomb repulsion) and thus dramatically improve their transmission to the following MS stage.36 Further, the rf ion confinement in Paul

espite tremendous progress of mass spectrometry, the complexity of most biological and environmental samples mandates prior separations.1−3 As the improved MS peak capacity has ameliorated spectral congestion and thus the need to separate isobars of different stoichiometry, isomer resolution has come to the forefront of omic sciences such as proteomics,4 metabolomics,5 and petroleomics.6 Traditional condensedphase approaches are increasingly replaced or complemented by rapid ion mobility separations (IMS) with unique specificity.7,8 Ion transport in gases depends on the reduced electric field E/N, where E is the field strength and N is the gas number density.9,10 Hence IMS techniques belong to two groups: linear7−9 based on the absolute mobility (K) and differential or field asymmetric waveform IMS (FAIMS) based on the derivative of K(E/N) function10 elicited in a periodic asymmetric field10,11 created by waveform of some amplitude (dispersion voltage, DV) applied across two electrodes. Ions pulled through the gap between those electrodes by gas flow are pushed toward either with velocity set by the difference between K at two E/N values and neutralized on contact.10,11 However, some ions are equilibrated by compensation field (EC/N) due to compensation voltage (CV) superposed on the waveform and pass to the detector: scanning CV reveals the spectrum of present species. The resolving power (R) depends on the gap shape, maximizing for planar gaps where homogeneous field can exactly balance only one species at any CV.10,12 The FAIMS/MS strategies are now deployed in proteomics (to improve the sequence coverage and detection limits by suppressing chemical noise and to resolve peptides with inverted sequences or alternative locations of post© XXXX American Chemical Society

Received: September 25, 2017 Accepted: November 27, 2017 Published: November 27, 2017 A

DOI: 10.1021/acs.analchem.7b03925 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

lower pressure (with w scaling as P−1/2). As for any separation in media, the diffusion-limited resolving power scales50 as t1/2. The decrease of R due to faster separation may outweigh its rise thanks to stronger field, resulting in worse resolution and peak capacity. That said, the peak capacity production rate can grow drastically:44 merging the proportionalities of R to (ED/N)3 and t1/2 yields the scaling of t as (ED/N)−6 at constant resolution (i.e., doubling the field nominally accelerates separation 64fold). This physics underlies the FAIMS microchips (particularly, in Owlstone Lonestar and μ-FAIMS devices). There, channels with g = 35 and 100 μm permitted ED/N up to ∼700 and ∼300 Td respectively, although (i) and (iii) imposed the cap of ∼200−300 Td.44,51,52 Most analytes vanish in that range, probably via “self-cleaning” where isomerization or fragmentation changes CVs beyond the peak width and products are destroyed by FAIMS action.47,48 The waveform remained roughly businusoidal, with extreme ν = 28 MHz required to keep the ion oscillation amplitude (Δd) under the gap width.10 One can calculate Δd as

or electrodynamic funnel traps at the IMS entrance during the shut gate periods greatly raises the ion utilization.36,37 These capabilities are crucial to the Waters Synapt traveling-wave (TW) IMS and Agilent 6560 drift tube (DT) IMS/time-offlight MS instruments,38,39 which over the past decade had turned IMS/MS from a niche pursuit into a widely accepted broad-purpose analytical tool. Combining IMS separations with orthogonal rf focusing at low P has also enabled (i) stacking IMS stages with isomer selection and activation in between for (IMS)n analyses40 to elucidate sequential isomerization steps in parallel to MSn, (ii) trap IMS platforms by Bruker, which deliver excellent R up to ∼300 in a compact package,5,41 (iii) cyclotron IMS that attained record R up to ∼1000 by lengthening the ion path to ∼200 m through many turns in a circular enclosure,42 and (iv) high-resolution TW IMS in long serpentine channels of “SLIM” modules.43 The diversity and stature of IMS/MS today would not be reached with just ambient-pressure IMS. Those low-pressure systems were for linear IMS, while FAIMS was only characterized near 1 atm.10,11,44 The sole work at P < 1 atm stopped at 0.4 atm,45 which was far too high for rf ion focusing and therefore did not open the novel analytical avenues noted above for linear IMS. Here we report the development of low-pressure FAIMS compatible with such processes.

Δd = K (0)E DΔF /ν

where ΔF is a dimensionless factor governed by the waveform profile.10 The diffusional losses in such tiny gaps limit sensible t to ∼0.02−0.2 ms, which is 103−104 times smaller than t ∼ 200 ms in high-resolution planar FAIMS devices. 44,51 The consequent resolving power loss of ∼30−100× is only partly mitigated by ∼10× gain due to higher ED/N, resulting in poor resolution44,51,52 (“regular” devices would be useless on the same time scale). Less challenging separations remain feasible and can now be nested in a standard LC elution peak (∼30 s), fitting FAIMS into the mainstream liquid chromatography (LC)/MS and capillary electrophoresis/MS pipelines.53 Instead of narrowing the gap by some factor (in microchips, 19−54× less than g = 1.9 mm in “full-size” devices), one can establish identical ED/N by reducing the pressure by an equal factor: i.e., from 1 atm to 14−40 Torr. The dependence of FAIMS performance on pressure was not well examined in theory or experiment. The monography on FAIMS anticipated superior resolution at lower pressures.10 Though obviously unphysical over the whole range (as R would have become infinite in the zero-pressure limit), that ostensibly agreed with then available measurements down to P = 0.4 atm.45 We now understand that the argument was not framed in the right units, and the resolving power actually scales as P1/2: in E/N terms, the peak position EC/N is pressure-independent whereas the width increases as stated above. (The resolution seemed to improve at lower pressure45 because of fixed DV and thus higher ED/N, which is not extendable to yet lower P in view of breakdown.) Hence extreme-field FAIMS at reduced pressure is basically not equivalent to that in narrower gaps: the resolving power is lower with all other parameters equal (by 4−7 times for above pressures). However, drop of DV at equal E/N in proportion to P permits widening the gap to lengthen t at the same diffusional loss and still decrease the DV. Extending t from ∼0.02−0.2 ms in microchips to present 7−8 ms increases R by ∼6−20 ×, at least canceling the effect of lower pressure. Further, the decreases of DV, load capacitance, and waveform frequency possible with wider gaps cumulatively relax the electrical engineering challenge a lot. That permitted us to upgrade the waveform from bisinusoidal to near the ideal rectangular, doubling the key ⟨F3⟩ and ⟨F5⟩ moments and thus the resolving



FOUNDATIONS As any function, K(E/N) is expandable in an infinite power series, here comprising even powers only:9,10 K (E /N )/K (0) = 1 + a(E /N )2 + b(E /N )4 + c(E /N )6 + d(E /N )8 ...

(2)

(1)

In the simplest model, truncating eq 1 at the quadratic term and treating ion diffusion as independent of E/N, one finds R to scale10 as (ED/N)3 where ED is the peak field at DV. This approximation is apt at ED/N ∼ 60−120 Td in “regular” FAIMS devices that feature P ∼ 1 atm in macroscopic gaps (width g ∼ 0.5−2 mm).12,14,15,17 At much higher ED/N employed here, the b(E/N)4 and c(E/N)6 terms become crucial, and yet higher orders may matter. In sum, those terms partially offset a(E/N)2, flattening the K(E/N) and thus CV(DV) dependences.10 Also, intense field heating of ions at extreme E/N renders their diffusion anisotropic and significantly accelerates it (especially in the field direction).9 This broadens the peaks and thus diminishes the R value, moderating its scaling toward (ED/N)2 in the infinite-field limit.10,44,46 The dependences of R on ED/N remain steep nonetheless, motivating one to maximize ED/N for best resolution. Elevating ED/N is checked by (i) output constraints of the driving circuitry, (ii) electrical (arc) breakdown across the gap, and (iii) ion fragmentation upon field heating.10,47,48 As the engineering efforts relax (i), the fundamental considerations (ii) and (iii) come to the forefront. With respect to (ii), the maximum reduced field in any gas is defined according to Paschen law by the product of pressure and gap width, increasing as P × g decreases.49 For N2 at ambient conditions and g = 2 mm, the threshold is ∼130 Td, exceeding which requires a narrower gap and/or lower pressure. Either necessitates shortening the filtering time (t) because diffusion eliminates all ions faster from narrower gaps at equal pressure (with the packet width w scaling as t1/2) and from same gaps at B

DOI: 10.1021/acs.analchem.7b03925 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 1. Schemes of experimental platforms involving (a) quadrupole MS and (b) ToF MS.

power roughly proportional to them.10,54 In the end, low pressure allows higher resolution than narrower gaps at equal P × g and thus ED/N, while providing the sensitivity and operational flexibility advantages. This concept is realized here.



METHODS The FAIMS cell consists of planar steel electrodes held in PEEK and ceramic insulators. The gap has a 100 mm length, 20 mm span, and 5 mm width to roughly match the diffusional broadening for common ion species after t = 10 ms (below). The device was installed into two Shimadzu MS platforms (Figure 1): (i) LCMS 2020 (single quadrupole) and (ii) modified LCMS-IT-TOF (quadrupole ion trap/time-of-flight). The MS inlet capillary of 0.5 mm diameter and 80 mm length was heated to 250 °C. The cell was housed in a new chamber placed into the first vacuum region after the capillary and connected to downstream Q-array ion optics via a skimmer with 2 mm aperture. The pressure inside was adjusted by admitting N2 with a leak valve and measured by a manometer. Ions are carried through the cell by gas flow. The residence time (t) was extracted by replacing the FAIMS waveform by a symmetric rectangular one and tuning its frequency to halve the signal. That time dropped from 11 to 6 ms as the pressure rose from 15 to 30 Torr (Figure 2a). Hence the proportionalities of R to P1/2 and t1/2 about cancel, especially over the upper half of the range (Figure 2a), and the highest pressure does not always maximize the resolution because of decreasing E/N (Figure 2b) and likely deviations from laminar gas flow. The nominally rectangular waveform is generated by digital key switching between two opposite-polarity dc voltages with absolute ratio f that determines the profile.54 We could ramp ν up to 800 kHz with DV up to the breakdown at any pressure (Figure 2b). The corresponding E/N ∼ 300 Td is close to the breakdown level51 in microchips with g = 100 μm and exceeds the point of signal disappearance for nearly all species. The quantity ΔF for rectangular waveforms is10 ΔF = 1/(f + 1)

Figure 2. Dependences of the measured FAIMS filtering time and (time × pressure)1/2 quantity (a) and breakdown voltage and field (b) on the cell pressure. The range selected herein is shaded.

satisfying the Δd ≪ g criterion is ∼100 kHz, wherein Δd for lighter ions with K0 = 2 cm2/(V s) at ED/N = 250 Td is 2.4−3.0 mm. A perfect rectangular waveform with any f implies an impossible infinite electric current at vertical potential walls, and real profiles have finite rise/fall times (tris) that reduce the true ⟨F3⟩ and other ⟨F2n+1⟩ moments below nominal metrics with proportional impact on CVs and FAIMS resolution. The tris depends on the electronics but ordinarily not the frequency, thus increasing in importance for higher ν. The tris = 50 ns here amounts to a critical 40% of the high-voltage excursion for f = 4 and ν = 800 kHz (with ⟨F3⟩ and ⟨F5⟩ at 82% and 76% of the nominal values) but marginal 5% for 100 kHz (⟨F3⟩ and ⟨F5⟩ at 96% and 94% of the same values). The fidelity for f = 3 at the same frequencies is slightly better yet, because of longer highvoltage excursion. We chose ν = 200 kHz (⟨F3⟩ and ⟨F5⟩ at 95% and 92%, which in practice is as good as 100%) to balance maximizing10 ⟨F2n+1⟩ with minimizing Δd (Figure 3). The CV scan speed is freely variable; we used ∼0.5−1 Td/s.

(3)

Although ⟨F3⟩ reaches maximum of 0.25 at f = 2, extremefield FAIMS may benefit from greater f to reduce Δd per eqs 2, 3 and the high-field diffusion component.10 Here we adopted f of 3 or 4 with respective theoretical ⟨F3⟩ of 2/9 and 3/16 (i.e., 89−75% of the maximum) and ΔF = 1/4 and 1/5 (i.e., 75−60% of ΔF = 1/3 at f = 2). Then the minimum reasonable ν C

DOI: 10.1021/acs.analchem.7b03925 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

generated by ambient-pressure electrospray ionization (ESI) sources original to the MS platforms. These compounds and their mixtures were dissolved to 0.3−300 μM in a usual solvent of 50:49:1 water/methanol/acetic acid and infused at 1−9 μL/ min flow rate.



RESULTS Characterization of Low-Pressure FAIMS across Experimental Regimes and Analyte Classes. First, we probed the influence of gas pressure in FAIMS across ion masses and charge states. As exemplified for 1+ protonated papaverine (Pv, m/z = 340) and 5+ insulin (Is, m/z = 1163), the curves of EC/N versus ED/N virtually coincide at all pressures (Figure 4). The peak widths are conserved (∼0.2−0.3 Td) at all ED/N (in line with theory10,46 and experiments at ambient pressure) and pressures (reflecting faster separations at higher P, as discussed above). As the result, the resolving power appears not systematically pressure-dependent, with different trends for different species. Both Pv and Is belong (in N2) to type C where K(E/N) function steadily decreases (a < 0),10,55 but the dependence is steeper for Is. That pattern is usual for 1+ ions in this mass range and multiply charged peptides.23,56−59 The maximum resolving power is ∼25 for Pv and ∼80 for Is. A much greater R for multiply charged ions, reflecting higher EC/N (presently >10 Td for Is vs 5 Td for Pv) is a hallmark of planar FAIMS.56−59 This ensues here from Is having both steeper EC(ED) dependences and surviving to higher ED/N (∼300 Td vs ∼250 Td for Pv), perhaps due to said self-cleaning (fragmentation is favored for smaller more mobile ions that experience stronger field heating9,10 and have fewer vibrations to dissipate it during the peak field). The faster anisotropic diffusion and greater Δd for those ions also serve50 to remove them from the gap at lower ED/N, because of ion losses being both greater at any E/N and increasing more rapidly at higher E/N. Given the minute and case-specific dependence of separations on the pressure, we have settled on intermediate P = 24−26 Torr (Figure 2). The signal for most species also about maximized in that range, typically at ∼10−

Figure 3. Readback of the waveform with f = 4 and ν = 200 kHz.

As all IMS methods, FAIMS relies on ions essentially maintaining the terminal velocity controlled by immediate field with no memory of past dynamics. That is, the relaxation time (trel) ought to be much shorter than any material element of the waveform. One can estimate:10 trel = 2mK /(ze)

(4)

where z is the ion charge state and e is the elementary charge. The mobility and thus trel scale as 1/P. Hence, the relaxation negligible in regular FAIMS grows in significance at low pressure, especially at higher waveform frequencies. We see indications of that, to be detailed in future work. Presently near the lower end of the ν range, we expect only a minor effect. The system was evaluated in positive mode using standards with masses and charge states representative of small molecule, metabolomic, and proteomic (bottom-up and middle-down) analyses: 1+ amino acids (glycine, isoleucine, isobaric glutamine and lysine, and tryptophan), 1+ papaverine, and peptides with z = 2−5 (syntide 2, peptides from the τ protein phosphorylated on differing sites, and insulin). The phosphopeptides were synthesized by standard Fmoc protocol;15 other chemicals were purchased from Sigma-Aldrich (Dorset, U.K.). Ions were

Figure 4. Measured separation parameters (top), peak widths (middle), and resolving powers (bottom) for papaverine and insulin ions as a function of gas pressure (single quad platform). The error margin of peak widths is within ±10%. D

DOI: 10.1021/acs.analchem.7b03925 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 5. Measured CV/DV palettes for Gly, Ile, Trp (ToF platform), and Gln/Lys mixture (single quad platform). The spectrum for Gln/Lys at marked ED/N was acquired in 4 s.

unfolding) at any DV. That pattern persists here (Figure 6), and syntide 2 remains a good calibrant for bottom-up proteomics.

30% compared to the no-FAIMS benchmark. Some of the losses occur at FAIMS stage interfaces (rather than within the cell) and would likely be reduced by ongoing design optimization. Amino acids (aa) make a convenient finite set to systematically test separations of small molecules and were so employed in IMS and FAIMS development.60−62 All aa readily form 1+ protonated ions in ESI. In regular FAIMS62 with N2, they were observed at EC < 0, behaving as type A ions [i.e., K(E/N) continually increasing]10,55 up to the sampled E/N ∼ 65 Td. That is inevitably an artifact of ending the measurement or signal cutoff at ED/N below the turn.10 Indeed, the K(E/N) for most aa were fit62 using a > 0 and b < 0, and the negative b(E/ N)4 term must prevail at some E/N (type B behavior). That happens by ED/N ∼ 200 Td in microchips,44,52 with the peak for Leu (and presumably Ile and heavier aa) shifting to EC > 0. Present patterns are similar (Figure 5). The lightest H+Gly (m/z = 76) continues as type A up to ED/N ∼ 120 Td where the signal disappears. The medium-size H+Ile (m/z = 132), H+Gln (m/z = 147), H+Lys (m/z = 147), and heaviest H+Trp (m/z = 205) are type B, with K decreasing from ED/N ∼ 120 Td until the signal vanishes at ∼160 Td. While the increment of K from low to high field stays positive and all peaks come at EC < 0, modest K(E/N) extrapolations employing fit {a, b} values yield EC > 0 by ED/N ∼ 200 Td, in line with the spectra obtained using microchips.52 Lysine is heavier than glutamine by 36 mDa, allowing baseline mass resolution above R ∼ 10 000 in modern MS systems including the present ToF. The isotopic envelopes and spectral congestion still make preseparation helpful in such instances. As in regular FAIMS,62 the coefficient a is much lower for H+Lys than H+Gln, and they are trivially resolved at moderate ED/N = 120 Td in a scan of few seconds (Figure 5).62 Like MS and other IMS methods, FAIMS benefits from the calibration of separation axis. Best is furnished by standards close to unknowns in m and z (without m/z overlap) and EC to minimize the time lag between them in CV scans and ensure their parallel evolution depending on experimental variables. That is, 1+ lipids are optimally served by others of close chemistry and mass, and peptides by similar peptides. Our analyses of smaller modified peptides typically employed56,57,59 syntide 2 (PLARTLSVAGLPGKK, 1507 Da) that made prominent well-shaped peaks with z = 2−4. Those were type C in regular FAIMS using N2, with CVs increasing from 2+ to 3+ and notably decreasing to 4+ (perhaps because of

Figure 6. Separation parameters for syntide 2 ions depending on DV (ToF platform).

All above observations confirm the integrity of data, verify at most marginal relaxation under present conditions (perhaps manifested in slightly higher EC/N at higher pressures in Figure 4), and overall prove that FAIMS at mere 1/50 atm copies the mechanism and principles known at ambient pressure. The maximum R values for Pv and Is are ∼1/4 of ∼100 and ∼300 achieved for analogous ions in ambient-pressure FAIMS,17,56−59 but present separations are much faster and involve no He or H2. Mixtures of those light gases with N2 dramatically enhance resolution for ions of all sizes and charge states through nonBlanc phenomena in regular FAIMS devices17,56−59,63,64 and, less but still substantially, microchips at present51 E/N. Those buffers may also lift the R metrics here. The resolution of ambient-pressure FAIMS for some species may be augmented by (polar) vapor additives that preferentially adsorb on ions with certain chemical functionalities or pronouncedly nonuniform charge distributions that yield strong charge−dipole interactions.65,66 This approach remains to be investigated at higher E/N here, although other vapors may work better. Anyhow, vapors have not been useful in proteome analyses (below), partly because of intense proton transfer from multiply charged peptides causing dramatic overall signal losses and phantom features for reduced charge states that grossly interfere with separations.66−68 Separations of Isomeric Peptides. With the performance of low-pressure FAIMS validated, we started exploring its E

DOI: 10.1021/acs.analchem.7b03925 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 7. (a) Measured FAIMS spectra for the isomolar phosphopeptide mixture (ED/N of 220 Td for z = 2 and 190 Td for z = 3 and 4) with individual variants overlaid; (b) CVs for all variants relative to 1, depending on DV (ToF platform). The CV scales were calibrated by linear dilation using the dominant Syntide 2 peak with same z as the target ion(s).

z = 4, the 2 coincident with 1 at 190 Td progressively pulls to higher EC at lower ED with a wide margin by ∼100 Td (Figure 7b)in line with regular FAIMS.72 So present separations generally match previous findings in the same field range. Separated isomeric peptides need to be identified, preferably without the requirement for standards that severely constricts the utility of approach. A priori localization of PTMs is best achieved via ETD,15,73−75 but long reaction time (∼10−100 ms) prohibits it after dispersive IMS separations with typical temporal peak widths of ∼0.1−1 ms such as DTIMS and TWIMS (including SLIM).43 Filtering methods like FAIMS have no intrinsic time scale and are readily coupled to ETD15,73,74 in a mode resembling single or multiple reaction monitoring (SRM/MRM) where CV is fixed or stepped over few defined values.15,76 This approach successful with ambientpressure FAIMS would equally work at low pressures. While we employed quadrupole and time-of-flight MS here, low-pressure FAIMS can likewise be introduced into other mass spectrometer types, including ion traps and Fourier-transform ion cyclotron resonance and Orbitrap MS that were used with ambient-pressure FAIMS stages.15,16,72,73,77,78 Present rapid scans also fit into normal LC peaks, allowing full-range FAIMS within LC/MS pipelines that was previously feasible only with low-resolution microchips.53

topical applications. One is characterizing the mixtures of peptides with variant PTM localization coming from the digestion of proteoforms that are ubiquitous in vivo and have distinct biomedical roles or activities. A headline example is histones, where attachment of various PTMs to designated sites codes an epigenetic language that controls DNA transcription.15,69,70 Another occurrence is the human τ (tau), with variant phosphorylation relevant to Alzheimer’s disease.71−73 Despite vital importance to proteomics, localization of PTMs remains difficult. A major reason is that tandem MS techniques fail to disentangle more than two variants for lack of informative fragments,15,73,74 while LC separations are often ineffective or unsuitably lengthy.71 Recently, FAIMS has broadly resolved variants involving various PTMs for peptides up to ∼5 kDa.15,59,70,72,73 Arguably the most consequential PTM is phosphorylation, and four monophosphorylated peptides from the τ226−240 segment {pT231 (1), pS235 (2), pS237 (3), pS238 (4)} make a quality benchmark for localization procedures.71−73 High-resolution FAIMS at ambient pressure has separated all those baselines for 3+ and partly for 2+ ions.72,73 We nearly reproduced that outcome here, with 1 and 4 resolved worse for 3+ but baseline for 4+ ions (not encountered for 4 previously72) and partial separation for 2+ ions. As usual, the resolution for each z is maximized at highest ED/N with decent signal, here 190−220 Td (Figure 7). Separations are clearly uncorrelated across charge states (as at ambient pressure),15,70 while the peak orders differ. The EC in z = 3 increases in the order (1, 4, 2, 3), while the best resolution in regular FAIMS72 was achieved with (1, 4, 3, 2) using ED/N = 87 Td and He/N2 mix with 60−70% He. The features for sizable peptides in CV spectra often swap upon conformational transitions driven by field heating that is controlled by ED/N and gas composition (increasing at higher He fractions in He/ N2).70,72 The peaks for 2 and 3 at z = 3 in regular FAIMS moved closer with raising ED/N and/or He percentage to about cross under the hottest conditions tried (117 Td, 50% He).72 Those two switch places at ED/N ∼ 130 Td here (Figure 7b). For z = 2, the present order {3; 1 + 4; 2} at ED/N ∼ 180−220 Td switches to {1 + 4; 3; 2} at ∼120−160 Td that is close to {1 + 4; 2; 3} at the maximum heating in regular FAIMS.72 For



CONCLUSIONS We have implemented FAIMS at gas pressures down to 15 Torr, allowing insertion into the vacuum manifold and effective integration with MS stages using established means for ion conveyance by rf and dc fields. The inherently lower resolution at reduced pressure is offset by extreme fields (up to ∼300 Td) and near-ideal rectangular waveform profiles permitted by Paschen law and modest drive requirements. While inferior to that achievable at 1 atm, the resulting peak capacity suffices for challenging applications such as resolving the PTM localization variants for modified peptides. The isolation of FAIMS stage from ion source allows choosing the carrier gas with no effect on ESI (a problem with buffers containing64,79−81 SF6 or He) and gases precluded in ambient-pressure FAIMS (with high gas consumption and substantial exhaust to the lab space) by cost or toxicity. Broad trends for benchmark species across masses F

DOI: 10.1021/acs.analchem.7b03925 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

(18) Gabryelski, W.; Froese, K. L. J. Am. Soc. Mass Spectrom. 2003, 14, 265−277. (19) Griffiths, R. L.; Creese, A. J.; Race, A. M.; Bunch, J.; Cooper, H. J. Anal. Chem. 2016, 88, 6758−6766. (20) Sander, L. C.; Sharpless, K. E.; Satterfield, M. B.; Ihara, T.; Phinney, K. W.; Yen, J. H.; Wise, S. A.; Gay, M. L.; Lam, J. W.; McCooeye, M.; Gardner, G.; Fraser, C.; Sturgeon, R.; Roman, M. Anal. Chem. 2005, 77, 3101−3112. (21) Hatsis, P.; Brockman, A. H.; Wu, J. T. Rapid Commun. Mass Spectrom. 2007, 21, 2295−2300. (22) Kaszycki, J.; Bowman, A. P.; Shvartsburg, A. A. J. Am. Soc. Mass Spectrom. 2016, 27, 795−799. (23) Shvartsburg, A. A.; Mashkevich, S. V.; Smith, R. D. J. Phys. Chem. A 2006, 110, 2663−2673. (24) Tyndall, A. M. The Mobility of Positive Ions in Gases; Cambridge University Press: New York, 1938. (25) Albritton, D. L.; Miller, T. M.; Martin, D. W.; McDaniel, E. W. Phys. Rev. 1968, 171, 94−102. (26) James, D. R.; Graham, E.; Akridge, G. R.; McDaniel, E. W. J. Chem. Phys. 1975, 62, 740−741. (27) Jarrold, M. F.; Constant, V. A. Phys. Rev. Lett. 1991, 67, 2994− 2997. (28) von Helden, G.; Hsu, M. T.; Kemper, P. R.; Bowers, M. T. J. Chem. Phys. 1991, 95, 3835−3837. (29) Jarrold, M. F. Acc. Chem. Res. 1999, 32, 360−367. (30) Ruotolo, B. T.; Giles, K.; Campuzano, I.; Sandercock, A. M.; Bateman, R. H.; Robinson, C. V. Science 2005, 310, 1658−1661. (31) Cole, H. L.; Kalapothakis, J. M. D.; Bennett, G.; Barran, P. E.; MacPhee, C. E. Angew. Chem., Int. Ed. 2010, 49, 9448−9451. (32) Hunter, J.; Fye, J. L.; Jarrold, M. F. Science 1993, 260, 784−786. (33) von Helden, G.; Wyttenbach, T.; Bowers, M. T. Int. J. Mass Spectrom. Ion Processes 1995, 146, 349−364. (34) Ibrahim, Y.; Tang, K.; Tolmachev, A. V.; Shvartsburg, A. A.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2006, 17, 1299−1305. (35) Thalassinos, K.; Slade, S. E.; Jennings, K. R.; Scrivens, J. H.; Giles, K.; Wildgoose, J.; Hoyes, J.; Bateman, R. H.; Bowers, M. T. Int. J. Mass Spectrom. 2004, 236, 55−63. (36) Tang, K.; Shvartsburg, A. A.; Lee, H. N.; Prior, D. C.; Buschbach, M. A.; Li, F.; Tolmachev, A. V.; Anderson, G. A.; Smith, R. D. Anal. Chem. 2005, 77, 3330−3339. (37) Hoaglund, C. S.; Valentine, S. J.; Clemmer, D. E. Anal. Chem. 1997, 69, 4156−4161. (38) Ruotolo, B. T.; Benesch, J. L. P.; Sandercock, A. M.; Hyung, S. J.; Robinson, C. V. Nat. Protoc. 2008, 3, 1139−1152. (39) May, J. C.; Goodwin, C. R.; Lareau, N. M.; Leaptrot, K. L.; Morris, C. B.; Kurulugama, R. T.; Mordehai, A.; Klein, C.; Barry, W.; Darland, E.; Overney, G.; Imatani, K.; Stafford, G. C.; Fjeldsted, J. C.; McLean, J. A. Anal. Chem. 2014, 86, 2107−2116. (40) Merenbloom, S. I.; Koeniger, S. L.; Valentine, S. J.; Plasencia, M. D.; Clemmer, D. E. Anal. Chem. 2006, 78, 2802−2809. (41) Silveira, J. A.; Ridgeway, M. E.; Park, M. A. Anal. Chem. 2014, 86, 5624−5627. (42) Glaskin, R. S.; Ewing, M. A.; Clemmer, D. E. Anal. Chem. 2013, 85, 7003−7008. (43) Deng, L.; Webb, I. K.; Garimella, S. V. B.; Hamid, A. M.; Zheng, X.; Norheim, R. V.; Prost, S. A.; Anderson, G. A.; Sandoval, J. A.; Baker, E. S.; Ibrahim, Y. M.; Smith, R. D. Anal. Chem. 2017, 89, 4628− 4634. (44) Shvartsburg, A. A.; Smith, R. D.; Wilks, A.; Koehl, A.; RuizAlonso, D.; Boyle, B. Anal. Chem. 2009, 81, 6489−6495. (45) Nazarov, E. G.; Coy, S. L.; Krylov, E. V.; Miller, R. A.; Eiceman, G. A. Anal. Chem. 2006, 78, 7697−7706. (46) Shvartsburg, A. A.; Tang, K.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2004, 15, 1487−1498. (47) Shvartsburg, A. A.; Li, F.; Tang, K.; Smith, R. D. Anal. Chem. 2007, 79, 1523−1528. (48) Menlyadiev, M. R.; Tarassov, A.; Kielnecker, A. M.; Eiceman, G. A. Analyst 2015, 140, 2995−3002.

and charge states track those at atmospheric pressure. Some differences in peak order for fragile macromolecular ions appear largely due to conformational annealing upon presently stronger field heating and mostly vanish at lower field. Full CV scans measurable in seconds are readily compatible with LC/MS and CE/MS, enabling synergetic use of condensedphase and FAIMS separations.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Alexandre A. Shvartsburg: 0000-0003-4004-481X Notes

The authors declare the following competing financial interest(s): The corresponding author is a consultant to Shimadzu Research Laboratory.



ACKNOWLEDGMENTS We thank Matthew Gill, Jeff Chadbourne, Stuart Harley, Cong Gao, Richard Witter, and Ioannis Nikolos (SRL) for experimental help, Matthew Baird and Andrew Bowman (WSU) for preparing the samples, and Dr. Pavel V. Shliaha (U. of Southern Denmark) for synthesizing the phosphopeptides. The work at WSU was funded by NSF CAREER Award (CHE-1552640). A.S. also holds a faculty appointment at the Moscow Engineering Physics Institute (MEPhI), Russia.



REFERENCES

(1) Niessen, W. M. A. Liquid Chromatography−Mass Spectrometry, 3rd ed.; CRC Press: Boca Raton, FL, 2006. (2) Dettmer, K.; Aronov, P. A.; Hammock, B. D. Mass Spectrom. Rev. 2007, 26, 51−78. (3) Wolters, D. A.; Washburn, M. P.; Yates, J. R. Anal. Chem. 2001, 73, 5683−5690. (4) Abshiru, N.; Caron-Lizotte, O.; Rajan, R. E.; Jamai, A.; Pomies, C.; Verreault, A.; Thibault, P. Nat. Commun. 2015, 6, 8648. (5) Pu, Y.; Ridgeway, M. E.; Glaskin, R. S.; Park, M. A.; Costello, C. E.; Lin, C. Anal. Chem. 2016, 88, 3440−3443. (6) Lalli, P. M.; Corilo, Y. E.; Rowland, S. M.; Marshall, A. G.; Rodgers, R. P. Energy Fuels 2015, 29, 3626−3633. (7) Eiceman, G. A.; Karpas, Z.; Hill, H. H. Ion Mobility Spectrometry; CRC Press: Boca Raton, FL, 2013. (8) Wilkins, C. L.; Trimpin, S. Ion Mobility Spectrometry−Mass Spectrometry: Theory and Applications; CRC Press: Boca Raton, FL, 2010. (9) Mason, E. A.; McDaniel, E. W. Transport Properties of Ions in Gases; Wiley: New York, 1988. (10) Shvartsburg, A. A. Differential Ion Mobility Spectrometry; CRC Press: Boca Raton, FL, 2008. (11) Guevremont, R. J. Chromatogr. A 2004, 1058, 3−19. (12) Shvartsburg, A. A.; Li, F.; Tang, K.; Smith, R. D. Anal. Chem. 2006, 78, 3706−3714. (13) Canterbury, J. D.; Yi, X.; Hoopmann, M. R.; MacCoss, M. J. Anal. Chem. 2008, 80, 6888−6897. (14) Shvartsburg, A. A.; Creese, A. J.; Smith, R. D.; Cooper, H. J. Anal. Chem. 2011, 83, 6918−6923. (15) Shliaha, P. V.; Baird, M. A.; Nielsen, M. M.; Gorshkov, V.; Bowman, A. P.; Kaszycki, J. L.; Jensen, O. N.; Shvartsburg, A. A. Anal. Chem. 2017, 89, 5461−5466. (16) Bridon, G.; Bonneil, E.; Muratore-Schroeder, T.; Caron-Lizotte, O.; Thibault, P. J. Proteome Res. 2012, 11, 927−940. (17) Bowman, A. P.; Abzalimov, R. R.; Shvartsburg, A. A. J. Am. Soc. Mass Spectrom. 2017, 28, 1552−1561. G

DOI: 10.1021/acs.analchem.7b03925 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry (49) Meek, J. M.; Craggs, J. D. Electrical Breakdown of Gases; Wiley: New York, 1978. (50) Shvartsburg, A. A.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2007, 18, 1672−1681. (51) Shvartsburg, A. A.; Ibrahim, Y.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2014, 25, 480−489. (52) Shvartsburg, A. A.; Tang, K.; Smith, R. D.; Holden, M.; Rush, M.; Thompson, A.; Toutoungi, D. Anal. Chem. 2009, 81, 8048−8053. (53) Arthur, K. L.; Turner, M. A.; Reynolds, J. C.; Creaser, C. S. Anal. Chem. 2017, 89, 3452−3459. (54) Shvartsburg, A. A.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2008, 19, 1286−1295. (55) Purves, R. W.; Guevremont, R.; Day, S.; Pipich, C. W.; Matyjaszczyk, M. S. Rev. Sci. Instrum. 1998, 69, 4094−4105. (56) Shvartsburg, A. A.; Danielson, W. F.; Smith, R. D. Anal. Chem. 2010, 82, 2456−2462. (57) Shvartsburg, A. A.; Smith, R. D. Anal. Chem. 2011, 83, 23−29. (58) Shvartsburg, A. A.; Smith, R. D. Anal. Chem. 2011, 83, 9159− 9166. (59) Shvartsburg, A. A.; Seim, T. A.; Danielson, W. F.; Norheim, R.; Moore, R. J.; Anderson, G. A.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2013, 24, 109−114. (60) Asbury, G. R.; Hill, H. H. J. Chromatogr. A 2000, 902, 433−437. (61) Beegle, L. W.; Kanik, I.; Matz, L.; Hill, H. H. Anal. Chem. 2001, 73, 3028−3034. (62) Guevremont, R.; Barnett, D. A.; Purves, R. W.; Viehland, L. A. J. Chem. Phys. 2001, 114, 10270−10277. (63) Barnett, D. A.; Purves, R. W.; Ells, B.; Guevremont, R. J. Mass Spectrom. 2000, 35, 976−980. (64) Shvartsburg, A. A.; Tang, K.; Smith, R. D. Anal. Chem. 2004, 76, 7366−7374. (65) Schneider, B. B.; Covey, T. R.; Coy, S. L.; Krylov, E. V.; Nazarov, E. G. Anal. Chem. 2010, 82, 1867−1880. (66) Schneider, B. B.; Nazarov, E. G.; Londry, F.; Vouros, P.; Covey, T. R. Mass Spectrom. Rev. 2016, 35, 687−737. (67) Purves, R. W.; Barnett, D. A.; Ells, B.; Guevremont, R. J. Am. Soc. Mass Spectrom. 2001, 12, 894−901. (68) Blagojevic, V.; Koyanagi, G. K.; Bohme, D. K. J. Am. Soc. Mass Spectrom. 2014, 25, 490−497. (69) Jenuwein, T.; Allis, C. D. Science 2001, 293, 1074−1080. (70) Shvartsburg, A. A.; Zheng, Y.; Smith, R. D.; Kelleher, N. L. Anal. Chem. 2012, 84, 4271−4276. (71) Singer, D.; Kuhlmann, J.; Muschket, M.; Hoffmann, R. Anal. Chem. 2010, 82, 6409−6414. (72) Shvartsburg, A. A.; Singer, D.; Smith, R. D.; Hoffmann, R. Anal. Chem. 2011, 83, 5078−5085. (73) Baird, M. A.; Shvartsburg, A. A. J. Am. Soc. Mass Spectrom. 2016, 27, 2064−2070. (74) Xuan, Y.; Creese, A. J.; Horner, J. A.; Cooper, H. J. Rapid Commun. Mass Spectrom. 2009, 23, 1963−1969. (75) Wiesner, J.; Premsler, T.; Sickmann, A. Proteomics 2008, 8, 4466−4483. (76) Canterbury, J. D.; Yi, X.; Hoopmann, M. R.; MacCoss, M. J. Anal. Chem. 2008, 80, 6888−6897. (77) Isenberg, S. L.; Armistead, P. M.; Glish, G. L. J. Am. Soc. Mass Spectrom. 2014, 25, 1592−1599. (78) Robinson, E. W.; Williams, E. R. J. Am. Soc. Mass Spectrom. 2005, 16, 1427−1437. (79) McCooeye, M. A.; Ells, B.; Barnett, D. A.; Purves, R. W.; Guevremont, R. J. Anal. Toxicol. 2001, 25, 81−87. (80) Xia, Y. Q.; Wu, S. T.; Jemal, M. Anal. Chem. 2008, 80, 7137− 7143. (81) Saba, J.; Bonneil, E.; Pomies, C.; Eng, K.; Thibault, P. J. Proteome Res. 2009, 8, 3355−3366.

H

DOI: 10.1021/acs.analchem.7b03925 Anal. Chem. XXXX, XXX, XXX−XXX