Determining Collisional Cross Sections from Ion Decay with Individual Ion Mass Spectrometry

Abstract

Collision cross section (CCS) measurements determined by ion mobility spectrometry (IMS) provide useful information about gas-phase protein structure that is complementary to mass analysis. Methods for determining CCS without a dedicated IMS system have been developed for Fourier transform mass spectrometry (FT-MS) platforms by measuring the signal decay during detection. Individual ion mass spectrometry (I²MS) provides charge detection and measures ion lifetimes across the length of an FT-MS detection event. By tracking lifetimes for entire ion populations, we demonstrate simultaneous determination of charge, mass, and CCS for proteins and complexes ranging from ~8 to ~232 kDa.

Graphical Abstract

INTRODUCTION

Ion mobility spectrometry (IMS) measures collision cross sections (CCS) of gas-phase analytes to provide conformational information.^1–4 Typical IMS experiments utilize drift tube, traveling wave, or trapped ion mobility in conjunction with time-of-flight mass spectrometers to measure analyte CCS and mass-to-charge ratio.^5,6 Determination of biopolymer CCS has allowed interrogation of oligonucleotide structural dynamics and protein ion conformations.^7–10 These methods have since been applied to intact protein complexes to understand protein–protein interactions and to provide orthogonal information to traditional structural biology techniques.^11–15 As the applications of CCS measurements have developed, the need for improved m/z resolving power has motivated the hyphenation of IMS to Fourier transform mass spectrometers (FT-MS).^4,16

Recently, methods have been developed for determining protein CCS directly from FT-MS data, including FT-ICR and Orbitrap systems, without additional IMS instrumentation.^17–21 These methods monitor dephasing events (i.e., in the form of signal decay) that remove ions from stable orbitals during mass analysis. Determining CCS from FT-MS signal provides a unique alternative to IMS systems and opens the path for integrating CCS methods with other modes of FT-MS analysis. Though this method has been operated using ion ensembles, integration with individual ion mass spectrometry (I²MS) and thus charge detection capabilities is possible.

Individual ion characterization provides a database from which a large amount of information can be determined from each ion signal. First, the precise centroiding of individual ion isotope signals yield high-resolution m/z values.^22,23 Second, the analysis of the induced current of each ion on the Orbitrap analyzer’s outer electrode yields ion charge through STORI plot analysis.^24,25 Now, by utilizing the lifetime information on individual ions (i.e., when the ion can no longer be detected), decay curves may be generated from thousands of individual ions to produce decay constants for CCS calculation. Herein, we demonstrate an I²MS processing technique that maximizes the information gleaned from a single ion data set for both denatured and native analytes; we employ I²MS to determine ion charge, mass, and CCS values.

METHODS

Sample Preparation and MS Analysis.

For native analysis, carbonic anhydrase (29 kDa, UniProt: P18915, Sigma-Aldrich: C7749), enolase (93 kDa, UniProt: P00925, Sigma-Aldrich: E6126), alcohol dehydrogenase (143 kDa, UniProt: P00330, Sigma-Aldrich: A3263), pyruvate kinase (232 kDa, UniProt: P11974, Sigma-Aldrich: PKRO), and beta-galactosidase (466 kDa, UniProt: P00722, Sigma-Aldrich: G3153), all obtained from Sigma-Aldrich, were dissolved in H₂O to final concentrations of 1 mg/mL and spin filtered with 100 mM ammonium acetate 10 times and H₂O twice at 11 000g in 3, 10, 30, or 50 kDa Amicon Ultra centrifugal filters (Merck Millipore) based on the molecular weight of the species. After desalting, the species were diluted to 500 nM in 100 mM ammonium acetate. For denatured analysis, ubiquitin (8.6 kDa, UniProt: A0A3Q1M4K3, Sigma-Aldrich: U6253), cytochrome C (12.2 kDa, UniProt: P00004, Sigma-Aldrich: C7752), myoglobin (16.9 kDa, UniProt:P68082, Sigma-Aldrich: M5696), and carbonic anhydrase (29 kDa) were obtained from Sigma-Aldrich, dissolved in H₂O at 1 mg/mL, and diluted to a final concentration of 1 μM in 50/50 H₂O/MeOH (Fisher Scientific) with 0.1% (v/v) formic acid (Fisher Scientific).

Denatured samples were analyzed by an Orbitrap Exploris 480 (Thermo Fisher Scientific) coupled to an Ion Max source (Thermo Fisher Scientific) with a flow rate of 1.5–5 μL/min, a spray voltage between 2.5 and 3 kV, an in-source collisioninduced-dissociation (sCID) value of 0–5 V, and a source temperature of 320 °C. Trapping gas pressure was reduced between 0.02 and 0.05 arbitrary units (UHV = 2.60e–11 – 2.85e–11 mbar), and AGC was disabled to allow for fixed injection times. The instrument resolution was set to 480 000 (at m/z 200) with eFT switched off, corresponding to a 2 s transient length to monitor individual ion decay times. Native samples were analyzed by a Q Exactive Ultra High Mass Range (UHMR) mass spectrometer (Thermo Fisher Scientific) with a Nanoflex ion source (Thermo Fisher Scientific) and borosilicate capillary emitters (Thermo Fisher Scientific) and the commercially available Direct Mass Technology mode (DMTmode) activated. Spray voltage was maintained between 1.3 and 1.8 kV, sCID was kept between 20 and 60 V, AGC was disabled, eFT was turned off, and the source temperature was kept at 300 °C. Ion decay was induced by adjusting the trapping pressure to 1–3 arbitrary units (UHV = 3.85e–11 – 1.25e–10 mbar) and a resolution setting of 800 000 (at m/z 200) corresponding to a ~5.5 s acquisition to monitor the decay of native ions. Maximum injection times for all samples varied between 0.5 and 20 ms, and 1000 spectra were acquired in each experiment.

Data Analysis and Collision Cross Section Determination.

A depiction of how STORI information can be utilized to determine the protein charge, mass, and CCS is presented in Figure 1. Individual ion spectra were processed using the commercially available processing software STORI-board (Proteinaceous) with signal-to-noise set to 0, duration threshold set to 0, minimum lifetime set to 0, and R² = 0.99 to maximize available ion decay information from each processed ion signal for the entire length of the detection event. Mass domain spectra were generated similarly with signal-to-noise set to 3, duration threshold set to 0.42, and R² = 0.99. In-house software developed in MATLAB 2021b and Python 3.11 were used to track ion decay across the duration of the measurement by binning ions by lifetime in 10.5 and 4.02 ms bins for native and denatured analyses, respectively. Binned ions were selected by filtering ions by m/z to track specific charge states and by slope to remove multi-ion events.

Figure 1.

Analysis of individual ion signals by STORI plot analysis to determine charge assignment, mass distributions (orange trace and inset), and precise ion signal decay information (blue trace and inset) for the CCS calculation.

Decay profiles for each ion signal were subject to truncation on both the onset and termination of the STORI collection. I²MS analysis relies on FT-MS signal induced by singular orbiting ions at unique frequency (or m/z) values. As a result, ion signals must be observed for long enough to become differentiable from instrument background noise.²⁴ The amount of time it takes for a real ion signal to be differentiated from background noise is related to its charge value and the length of survival within the acquisition event; keeping the survival time constant, an ion with higher charge induces a larger signal and therefore is differentiated from noise more quickly (higher signal-to-noise ratio). The “noise band” prevents the collection of ion information before the signal is confidently detected, so ion decay that occurs before the ion is differentiated cannot be utilized for CCS calculation. Additionally, close to the termination of the acquisition event, any remaining ions must be ejected from the Orbitrap analyzer to prepare for the next acquisition event. Consequently, ion deaths that are not related to neutral gas collisions occur near the end of the acquisition period. Decay profiles for native and denatured protein ions were ultimately sampled from 1 to 4.5 and 0.5–1.75 s, respectively. Sampled decay profiles were then log-transformed, and decay constants were determined from the slope of the linear fit. Supplemental Figure S1 illustrates this thresholding process and its effect on the linear regressions used to extract decay constants for CCS determination. Supplemental Figure S2 shows log-transformed charge state decay plots for two proteins of interest. Log-transformed decay plots typically yield R² > 0.99 for abundant charge states.

$$\text{CCS}=\frac{c}{L{f}_z{n}_g}$$ (1)

Denatured CCS were determined using eq 1 (described by Sanders et al.¹⁸), where c is the decay constant (1/s), L is the average path length of the ion during one oscillation (m), f_z is the frequency of the ion (Hz), and n_g, the density of gas molecules, was determined for denatured analytes by calibrating with the ubiquitin 9+ charge state as previously described and by the ideal gas law for native analytes.^18–20 The equations for converting from m/z to frequency for both instruments used are available in Supplemental Scheme 1.

$$\text{corrected}\text{CCS}=\frac{\text{uncorrected}\text{CCS}}{1-\left(\frac{\text{Ecm}*-9.6+223.24}{100}\right)}$$ (2)

Native CCS were determined using the collision energy correction in eq 2 as previously described by James et al.²⁰ The center of mass collision energy (Ecm) for a protein charge state is used to estimate the percent difference between the Orbitrap CCS measurements and IMS CCS measurements to correct values from eq 1. Native CCS were ultimately calculated by weighted average using the charge state CCS and charge state ion count as the literature shows that native protein CCS does not change significantly between charge states.^5,12–14

RESULTS AND DISCUSSION

After sampling, decay constants are extracted from the population decay plots. Figure 2a shows population decay curves for three (+22, +25, and +28) exemplary denatured carbonic anhydrase charge states. This trend is consistent with prior IMS and Orbitrap CCS findings, which have shown that denatured protein ion CCS increase with charge due to additional Coulombic repulsion between the charged sites.^{18,19,26–28} Supplemental Figure S3 validates that this trend holds across all denatured protein charge states analyzed in this study. Figure 2b demonstrates a linear decay rate increase across charge state distributions for ubiquitin, cytochrome C, myoglobin, and carbonic anhydrase. In addition, the overall decay values between these denatured proteins of increasing mass follow a linear trend, allowing CCS determination for protein analytes of varied mass.

Figure 2.

Decay plots for denatured carbonic anhydrase and extracted decay constants for all of the denatured species studied. I²MS-based Orbitrap CCS determination enables the simultaneous tracking of multiple carbonic anhydrase charge states, whose populations decay at different rates (a). Extracted decay constants for denatured ubiquitin (UB), cytochrome C (CY), myoglobin (MY), and carbonic anhydrase (CA) were plotted against charge (b).

Although the CCS determined for denatured proteins increase linearly across charge states and mass ranges, proteins that retain their native conformations in the gas phase have been observed to deviate from this trend. Dissociation or dephasing events for denatured proteins typically occur after an ion collides with a single gas molecule, but native proteins may endure many more collisions. This results in underestimated CCS by eq 1, and the extracted CCS values for native proteins require additional correction by eq 2.²⁰ After calculating raw CCS values for native proteins, the percent deviation between the Orbitrap CCS and IMS CCS values is plotted in Supplemental Figure S4 to determine the Ecm correction factor in eq 2. Charge state CCS values were then weighted by the number of ions collected for each charge state and averaged together as prior literature suggests native protein CCS does not change significantly across the native charge state distribution.^5,12–14 Figure 3 demonstrates that CCS values determined from I²MS data show strong agreement with IMS values for both denatured and postcorrection native proteins. Tabulated values for denatured proteins may be found in Supplemental Table S1, and values for native proteins may be found in Supplemental Table S2 containing exact values for I²MS determined CCS and percent deviation from known IMS CCS values.

Figure 3.

I²MS analysis provides the determination of analyte charge, mass, and CCS for both denatured and native protein analytes. Orbitrap CCS for native carbonic anhydrase (nCA), enolase dimer (nEN), alcohol dehydrogenase (nAD), pyruvate kinase (nPK), and beta-galactosidase (nBG) are represented as weighted averages. Error bars represent one standard deviation for the CCS determined from replicate measurements (n = 3). The nBG CCS value (n = 1) in the bottom right of the plot is provided to illustrate the limitations for accurately determining increasingly large protein CCS using the correction scheme in eq 2.

The reduced decay observed for native protein ions makes CCS determination by decay analysis challenging, requiring the initial determination of the correction factor, as shown in eq 2. The retention of globular structure for native proteins reduces CCS in comparison to denatured counterparts, resulting in a higher mean free path and less frequent neutral gas collisions,²⁹ which may contribute to this observed difference. However, for native and denatured proteins of comparable CCS, such as native alcohol dehydrogenase and denatured carbonic anhydrase, there remains a discrepancy in decay. This discrepancy may be exacerbated by the distribution of collisional energy across the additional degrees of freedom afforded by globular protein structures, thereby reducing the rate of dissociation compared to a denatured protein with an extended structure.³⁰ Megadalton-sized particles have been shown to primarily undergo desolvation rather than ion dissociation when mass analyzed for extended periods of time, suggesting that these large complexes are generally unaffected by collisions with nitrogen gas.³¹ While the proteins analyzed in this study did not exhibit frequency shifting, their general resistance to decay is similar to that exhibited by these other large globular analytes. Frequency shifts have mostly been observed with large, native complexes over 1 MDa.³¹ These considerations suggest that CCS determination by signal decay analysis may not be practical for large globular analytes. Alternative methods that measure the change in ion velocity have been shown to provide measurements of ion mobility, but extracting CCS is challenging due to the varied ion velocities within the trap.³²

As shown in Figure 3, calculating the CCS for increasingly large analytes like beta-galactosidase becomes impossible with this correction scheme. Supplemental Figure S5 further illustrates the loss of linearity in the correction model for beta-galactosidase, where each charge state returns a CCS percent difference of ~98%. The reduced decay for large analytes yields extremely low CCS values in eq 1, so the absolute difference between the known CCS determined by IMS and the CCS determined by signal decay analysis becomes too large to accurately correct on a relative basis, resulting in nonsensical values returned by eq 2. The core assumption that Orbitrap CCS values linearly deviate from known IMS CCS breaks down for complexes larger than ~232 kDa, ultimately plateauing near 100% without additional space to differentiate complexes with increasingly large CCS. Though the prospect of determining CCS for any native-like analyte by Orbitrap signal decay analysis is desirable, a more inclusive model that can effectively determine CCS across any mass range without additional correction is required for a practical approach to this type of analysis. Although outside the scope of this manuscript, future work will focus on the development of a model that accounts for collisions that do not directly lead to decay.

Besides providing CCS values that agree with values determined by IMS, I²MS also provides a detailed analysis of the ion charge and subsequent determination of mass. Isotopic resolution is readily achieved for all proteins besides alcohol dehydrogenase and pyruvate kinase in Figure 3 and could be achieved for larger protein complexes by extending the acquisition time to increase the number of ions collected.²³ The Ecm correction factor determined herein showed stability over the course of the study (~8 months), which is consistent with the correction factor developed by James et al.²⁰ This approach may also be applied to previously collected data using DMTmode if sufficient ion decay was observed. Determining CCS by I²MS provides an orthogonal measurement that may be extracted from I²MS data sets without increasing the experimental time required, allowing for more detailed characterization of proteins and protein complexes with Orbitrap mass spectrometers.

CONCLUSIONS

The use of individual ion time domain data to calculate CCS values maximizes the information that can be extracted from a single I²MS data set. I²MS-driven CCS of denatured proteins shows strong agreement with CCS determined previously by IMS, while native protein CCS determined using the Ecm correction strategy and weighted charge state averages provide CCS values at high m/z regimes. Integrating I²MS with established Orbitrap CCS determination strategies provides charge, mass, and CCS values for proteins and complexes weighing ~8 kDa to 232 kDa within a single experiment for targeted studies analyzing one proteoform or global analysis of complex proteoform mixtures.