Cyclic Ion Mobility Separations of Free Oxylipins

Abstract

Oxylipins are oxygenated lipid metabolites derived from polyunsaturated fatty acids (PUFAs). They play interconnected roles in physiological processes such as inflammation and vascular function, and their dysregulation has been linked to diseases, including cardiovascular, metabolic, and immunological disorders. Liquid chromatography (LC) has a limited ability to resolve the large number of isomers and isobars within this diverse lipid class. Ion mobility spectrometry (IMS) offers an orthogonal gas-phase separation technique and has been shown to improve peak capacity for isomers that differ by more than 1–2% in collision cross section (CCS). High-resolution IMS platforms have progressively enabled the separation of isomers with increasingly similar gas-phase structures and offer the potential to separate complex mixtures in experiments such as mass spectrometry imaging where LC is not available. Here, we demonstrate the application of high-resolution cyclic ion mobility spectrometry–mass spectrometry (cIMS–MS) to separate and characterize six complex mixtures of authentic oxylipin standards. Results show that deprotonated oxylipin species often exhibit multiple mobility features that emerge only after extended separations, whereas metal-adducted species, especially sodium adducts, yield clearer single peaks. Optimized traveling wave parameters enabled the separation of mixtures containing three to five oxylipins within 3–10 m of effective path length before ion wrap-around occurred, resolving features with Δ^DTCCS% of approximately 0.5% without front-end chromatographic separation. Multipass ^cIMSCCS_N2 measurements under optimized conditions were highly reproducible and closely matched the ^DTCCS reference values from Moran-Garrido et al. and da Silva et al., with average deviations of 0.33–0.76%. These findings support the use of multipass cIMS experiments and ^cIMSCCS measurements as valuable tools for oxylipin separation and annotation, particularly in the absence of chromatographic retention time data.

Introduction

Oxylipins are a diverse group of bioactive lipids, characterized by numerous isomers, produced via enzymatic and radical-driven pathways from the oxidative metabolism of polyunsaturated fatty acids (PUFAs). Usually stored in esterified forms within larger lipid classes, oxylipins are released as free fatty acids to influence inflammation and vascular function; their dysregulation has been implicated in respiratory, cardiovascular, and neurodegenerative diseases. − Free oxylipin isomeric complexity arises from enzymatic oxidation by cyclooxygenase (COX), lipoxygenase (LOX), and cytochrome P450 (CYP) enzymes, as well as autoxidation. This structural diversity manifests as linear and cyclized isomers bearing a variety of hydroxyl, epoxide, and ketone modifications, many of which share identical elemental composition and produce highly similar or indistinguishable tandem mass spectra, posing significant analytical challenges for confident identification. ,

Reversed-phase liquid chromatography–mass spectrometry (LC–MS) is the primary method for oxylipin analysis because of its high sensitivity and selectivity, enabling detection of low-abundance species. Targeted LC–MS techniques match retention times and tandem mass spectrometry (MS/MS) fragmentation patterns to authentic standards, facilitating differentiation of many isomers and isobars and identification of coeluting analytes when they have distinct product ion spectra; however, this approach is often challenged by isomers that coelute and produce highly similar or indistinguishable MS/MS spectra. , The sensitivity of multiple reaction monitoring (MRM) methods has recently enabled the direct detection of free oxylipins from tissue surfaces via mass spectrometry imaging (MSI) under ambient conditions. However, this method requires prior LC–MS/MS experiments, for example, on homogenized tissue, to establish which oxylipins are present and to inform MRM MSI transitions. When standards are unavailable, nontargeted LC–MS provides a comprehensive oxylipin profile, often using data-dependent acquisition to determine oxidation sites and types (such as epoxy, hydroxy, and keto groups), making it ideal for discovery studies. , However, nontargeted approaches can be less sensitive, and limited precursor-ion selection resolution before MS/MS can make it difficult to distinguish structurally similar compounds.

Ion mobility spectrometry (IMS) quickly separates ions in the gas phase by size, shape, and charge, often within milliseconds. When paired with LC–MS, it boosts peak capacity. Drift tube IMS (DTIMS), the traditional variant, employs a uniform electric field to steer ions through a drift chamber containing neutral buffer gas, with no net flow. The mobility data obtained help identify compounds and are usually expressed as collision cross section (CCS) values, indicating an ion’s effective size and shape in the gas phase. − Under DTIMS conditions, the time for an ion to reach its natural drift velocity is minimal, permitting CCS calculation via the Mason–Schamp equation (^DTCCS). Variations in functional groups, their positions, adduct profiles, and double bonds influence the gas-phase size and shape of oxylipins, thereby impacting their mobility and enabling many isomers to be distinguished. , Derivatization strategies such as the Paternò–Büchi reaction, which promotes cycloaddition across a carbon–carbon double bond, can alter lipid CCS so they fall outside the densely populated lipid region. Metal adducts generally increase ΔCCS% between oxylipin isomers compared to protonated or deprotonated forms, thus enhancing resolution. , Studies integrating LC with DTIMS–MS show that very few free oxylipin isomers are completely unresolved across all adducts, underscoring the importance of comprehensive ^DTCCS libraries for compound annotation. Nevertheless, many metal-adducted oxylipins still have ΔCCS% values below 2%, and methods such as MSI, which lack front-end separation, require high-resolution IM–MS techniques to distinguish isomers.

The popularization of IMS has led to the development of many high-resolution platforms with resolving powers reaching hundreds. Among these, traveling-wave (TW) IMS techniques, like structures for lossless ion manipulation (SLIM) and cyclic TWIMS (cIMS), achieve higher resolving power by extending the ion path. , In TWIMS, a traveling electric field propels ions, and separation occurs because of differences in rollover frequency. cIMS allows ions to pass multiple times through a 1-meter separation cell because there is no net electric potential drop from start to end, and it has been used to separate and identify fatty acyl and other lipid isomers. , Unlike DTIMS, CCS values in cIMS (^cIMSCCS) are not directly calculable; instead, they are calibrated against ^DTCCS, often using power-law relationships. , CCS is crucial for oxylipin annotation when retention time data are missing, and MS/MS spectra are very similar or identical. Although most ^cIMSCCS calibration techniques are designed for single-pass experiments, accurate multipass ^cIMSCCS measurements are vital for analyzing complex mixtures of closely related oxylipins that cannot be separated in a single pass. Our tool, Multipass CCS Refiner, applies perturbation-corrected periodic drift times via the Lin and Costello method, streamlining CCS calibration in multipass cIMS by correcting for charge, reduced mass, and mass-dependent flight-time effects, as suggested by Ruotolo et al. −

This study examines the ion mobility properties of 23 free oxylipins (10 cyclized and 13 linear) grouped into 6 isomeric mixtures, using multipass cIMS in both positive and negative ion modes. By optimizing TW conditions and employing tandem cIMS (cIMS/cIMS), we resolve mixtures containing 3–5 oxylipin isomers. TW conditions of 375 m/s velocity and 18 V wave height, with 60-70 ms separation times, served as a practical starting point for oxylipin mixtures to separate isomers with similar gas-phase structures while limiting ion wrap-around. We measured multipass ^cIMSCCS_N2 values across four TW conditions to assess reproducibility and validated them against a DTIMS library and the literature. , The Multipass CCS Refiner reduced the time required for post-processing CCS calibration curve creation and for studying unknown samples, enabling a workflow that completes within minutes.

Materials and Methods

Chemicals

Oxylipin standards (Table and Figure S1) were obtained from Cayman Chemical (Ann Arbor, MI, USA). Sodium, lithium, and potassium acetate salts were purchased from Sigma-Aldrich (St. Louis, MO, USA). Optima-grade ACN and H₂O were obtained from Fisher Scientific (Pittsburgh, PA, USA). Sodium formate and Major Mix calibration solutions were purchased from Waters Corporation (Wilmslow, UK). The SpheriCal IM–MS Low Range calibration mixture was obtained from Polymer Factory (Stockholm, Sweden).

1. Elemental Composition, Neutral Mass, Compound Abbreviations, FA Shorthand Notation, Structural Category, and Isomerism Summary for the 23 Oxylipins Investigated, Grouped by Isomeric Mixture.

mixture # (composition)	neutral mass (Da)	compound abbreviation	structural category	isomerism summary
1	354.2406	PGD1	cyclized	constitutional isomers
FA 20:3;O3(C20H3 4O5)		PGE1		PGF2 diastereomers
		PGF2α
		PGF2β
2	320.2351	14R,15S-EpETrE	linear	constitutional isomers
FA 20:4;O(C20H32O3)		15(R)-HETE		15-HETE enantiomers
		15(S)-HETE
		5(S)-HETE
		(±)-8,9-EpETrE
3	336.2301	5(S),12(S)-DiHETE	linear and cyclized	constitutional isomers
FA 20:4;O2(C20H32O4)		6-trans-LTB4		6-trans-LTB4 and 5(S),12(S)-DiHETE diastereomers
		8(S),15(S)-DiHETE
		PGA1
4	352.2250	LXA4	linear and cyclized	constitutional isomers
FA 20:4;O3(C20H32O5)		PGD2
		PGE2
5	318.2195	14(15)-EpETE	linear	constitutional isomers
FA 20:5;O(C20H30O3)		15(S)-HEPE
		15-oxo-ETE
		5-oxo-ETE
6	334.2144	PGA2	cyclized	constitutional isomers
FA 20:5;O2(C20H30O4)		PGB2
		PGJ2

^aOxylipin abbreviations are detailed in the Supporting Information (Table S1).

Sample Preparation

Oxylipin standards were grouped into 6 isomeric mixtures (Table S1), with full structural descriptions provided in the Supporting Information. Standards were prepared as individual compounds and as isomeric mixtures at 500 nM. Standards were initially dissolved in 100% H₂O and then serially diluted in H₂O/ACN (50:50, v/v). Acetate salt solutions (10 mM) of lithium, potassium, and sodium were prepared in H₂O/ACN (50:50, v/v) and further diluted to a final sample concentration of 1 mM. Each standard was analyzed both individually and as part of its isomeric mixture.

cIMS–MS Analysis

IMS separations were performed on a cIMS–mass spectrometry (cIMS–MS) platform (Waters Corporation, Wilmslow, U.K.), a commercially available instrument based on the SYNAPT G2-Si platform, in which the conventional 25 cm colinear separation cell is replaced by a 98 cm cyclic mobility cell positioned orthogonally to the primary ion path and operated with pulsed DC traveling waves. Electrospray ionization (ESI) was conducted in both positive and negative ion modes at applied voltages of 3 kV and 2 kV, respectively, and at a flow rate of 2 μL/min. Instrument parameters were tuned to preserve labile species, including a cone voltage of 20 V, a desolvation gas temperature of 200 °C and a flow of 600 L/h, a source offset of 10 V, a step-wave gradient of 10 V, trap and transfer cell collision energies of 5 and 3 V, respectively, and a transfer radio frequency of 150 V. Precursor ions were isolated in the quadrupole prior to separation, and N₂, maintained at a pressure of 1.79 mbar, was used as the buffer gas throughout all experiments.

TW heights and velocities were optimized independently for each isomeric mixture and adduct type (Table S2). TW heights were first scanned to identify conditions that increased arrival-time differences between isomeric species. TW velocities were then adjusted to shift relative ion positions within the wave and further improve separation. Optimization was performed over TW heights of 15–30 V and velocities of 350–475 m/s, as higher wave energies often produced fragmentation within the separation cell, predominantly from water and CO₂ losses. Ions underwent multiple passes through the cyclic mobility array, with pass counts determined manually for each separation. Separations were extended to the onset of ion wrap-around, where the fastest-moving ions complete an additional lap of the mobility cell and arrive at drift times comparable to the slowest-moving ion populations.

cIMS/cIMS–MS Analysis

cIMS/cIMS separations were performed by isolating specific mobility regions in the separation cell, storing them in the pre-store array, and reinjecting them for additional passes, while remaining mobility populations were transmitted directly to the time-of-flight detector without acquisition. , Transmitting ions to the pre-store array remixes the population and separation is lost. Therefore, sliced ion populations must first undergo the same number of passes as were conducted in the cIMS experiment to return to the same level of resolution prior to additional separation. The arrival time distribution (ATD) obtained from each reinjected mobility slice was adjusted during data analysis to account for remixing within the pre-store array. Following mobility separation, ion populations were ejected to the time-of-flight detector, operated in V-mode, and data were collected over 2 min using 3 pushes per bin over 200 bins.

Data Processing

Raw data were processed using MassLynx v4.2 and Quartz software (Waters Inc., Wilmslow, U.K.). Extracted arrival times were exported as. xlsx files. ATDs of individual species (dashed traces) and mixtures (solid black traces) were overlaid, plotted, and normalized to unity using custom R scripts (v4.5.2). Mobility feature centroids were determined by Gaussian fitting with nonlinear least-squares regression implemented using the nls function. Two-peak resolution (R_pp) for peaks within isomeric mixtures was calculated as the difference in drift times between peak centroids divided by the sum of the peak widths, multiplied by a factor of 1.18 derived from the half-maximum definition (2.355σ/2). , Half-height separation was defined at R_pp = 0.83, while baseline separation was defined as R_pp > 1.5. R_pp values were calculated from mixtures using the major mobility peak of each isomer under TW conditions optimized for group separation (Table S2).

Multipass CCS Calculations

The SpheriCal® IM–MS Low Range calibration mixture was analyzed over a mass range of 300–1200 m/z under four TW conditions spanning 15–20 V and 350–450 m/s. Multipass calibration curves were generated from perturbation-corrected periodic drift times (t_pp) obtained from [M + Na]⁺ adducts, using separation times (t_s) of 0, 2, 10, 20, 30, 32, 50, 75, 100, and 110 ms. Automated t_pp values were calculated from Multipass CCS Refiner using linear regressions of (t_n – t_s – t₀)/n versus t_s/n, where t_n represents the measured arrival time, t₀ the bypass time, and n the number of passes through the mobility cell. ATDs exhibiting ion wrap-around were excluded from the calibration datasets.

Calibrant ^DTCCS_N2 values were converted to reduced CCS (Ω′) to account for ion charge state and reduced mass effects. Calculated t_pp values were further adjusted for mass-dependent flight time contributions using the instrument-specific enhanced duty cycle delay coefficient, yielding t_D′. , Calibration curves were generated by linear regressions of ln­(Ω′) versus ln­(t_D′), consistent with the power-law relationship between drift time and CCS, and the resulting exponential factors were used to calculate calibrated ^cIMSCCS_N2 values.

All 23 oxylipin standards were subsequently analyzed as [M + Na]⁺ ions under the same four sets of TW conditions used for calibration and on the same day. The t_pp values were determined in the same manner as for the calibrants and were used to calculate ^cIMSCCS_N2 values using the corresponding calibration curves. Multipass ^cIMSCCS_N2 values were evaluated for variability across TW conditions and compared with recently published ^DTCCS_N2 databases and, when absent, with literature values. , ATD data for calibrants and oxylipin standards were downloaded in. xlsx format from the introduction tab of Multipass CCS Refiner under the Using the Application section at: https://ericgier.shinyapps.io/Multipass-CCS-Refiner/. Instructions for processing the data within the application can be found in the first tab of the SpheriCal calibration example file (“Start Here”), and a tutorial is available through YouTube: www.youtube.com/watch?v=AXrbocbQGY8&t=0s. Additional updates to the application since its release include import/export functionality for calibration files and built-in or user-supplied calibrant tables, reducing manual data entry. Current and planned updates are summarized in the Announcements section, and complete documentation is available at: https://github.com/facundof2016/Multipass-CCS-Refiner.

Results and Discussion

Oxylipin standards were selected to represent various oxygen-containing functional groups, including diols, epoxides, hydroxyls, and ketones within linear or cyclized structures. This structural diversity encompasses distinct types of constitutional isomerism: positional isomers that differ in the site of a shared functional group (e.g., 5­(S)-HETE versus 15­(S)-HETE, which differ only in hydroxyl position, or 14R,15S-EpETrE versus (±)-8,9-EpETrE, which differ only in epoxide position), and functional group isomers that differ in the identity of the oxygen-containing modification itself at a given site (e.g., 15­(S)-HETE versus 14R,15S-EpETrE) (Table ). Diastereomeric relationships are also represented within this panel: PGF_2α and PGF_2β share identical connectivity but differ in configuration at C9, while 6-trans-LTB₄ and 5­(S),12­(S)-DiHETE differ in both a stereocenter and E/Z double-bond configuration. In biological samples, variations in the position and type of functional groups, along with double-bond geometry, create numerous regio- and stereoisomers that often require high resolution to differentiate. , These structural differences affect gas-phase ion conformations and their CCS, allowing separation by IMS. Notably, metal adducts frequently exhibit ΔCCS% values exceeding 2%, thereby aiding isomer resolution. , cIMS was used to analyze 6 mixtures containing three to five oxylipin isomers without chromatographic separation, demonstrating its potential for MSI workflows. Due to the low endogenous oxylipin levels, especially in tissue sections, efficient separations with short path lengths and minimal reliance on cIMS/cIMS are necessary to minimize losses. Separations of complex oxylipin mixtures were optimized for a panel of 23 eicosanoids, including 18 from arachidonic acid (AA), 3 from dihomo-γ-linolenic acid (DGLA), and 2 from eicosapentaenoic (EPA) (Table S1).

Ionization Behavior

Oxylipins easily form deprotonated ions in ESI(−) because they contain acidic groups like carboxylic acids. These [M – H]⁻ ions are common in LC–MS workflows due to their high ionization efficiency. , In IMS, oxylipins appear in a narrow CCS range, with many isomers showing ΔCCS% < 0.5%. , Their fluid gas-phase structures cause broad mobility peaks, often limiting or preventing separation. Moran-Garrido et al. reported a 100% detection rate for 74 oxylipins as [M – H]⁻ ions, with only 10% of isomer pairs showing ΔCCS% > 4%. Significant ΔCCS% differences were mainly seen in FA 20:5;O species, such as those in mixture 1, where 5-oxo-ETE showed three different deprotonated forms, two of which were separated from other isomers and accounted for most of the resolved species. To test how [M – H]⁻ oxylipins behave under multipass cIMS, separations with drift paths ranging from 1 to 7 meters were tested.

All 23 oxylipins were detected as [M – H]⁻ ions. Single-pass separations yielded ATDs with minimal resolution, aligning with DTIMS results. , Extending the separation path to 7 m uncovered two or more gas-phase conformations for 17 of the 23 species (Figure S2). TW conditions were fine-tuned between 350–400 m/s and 15–18 V to enhance separation while limiting fragmentation within the mobility cell. Despite multiple deprotonated forms, some species showed partial or baseline resolution before ion wrap-around occurred. For instance, the main mobility feature of PGD₁ clearly separated from the dominant features of other isomers in the FA 20:3;O3 mixture (Figure S2A), though minor features overlapped. The dominant mobility features of 5­(S),12­(S)-DiHETE and 8­(S),15­(S)-DiHETE were also well resolved (Figure S2C), with minor overlaps. However, all species in the FA 20:4;O3 mixture showed a single dominant mobility feature, and LXA₄ was clearly separated from PGD₂ and PGE₂ (Figure S2D). The separation of the FA 20:5;O mixture was limited to a 3 m path length because 5-oxo-ETE rapidly occupied the entire separation space, displaying at least three deprotonated forms. Two of these forms were resolved from other isomers, consistent with DTIMS observations (Figure S2E).

In ESI­(+) experiments, only the FA 20:5;O mixture consistently formed [M + H]⁺ adducts across more than half of its isomers. Similar to deprotonated species, many cases exhibited minor mobility features before ion wrap-around. Water-loss in-source fragmentation products were common and, in some instances, highly abundant, often showing broad mobility distributions. Longer pathlength separations heightened distribution complexity but also uncovered additional mobility features that could shed light on structural differences among isomers.

Oxylipins readily formed metal adducts in ESI­(+), with sodiated species observed for most standards. When sodium acetate was added, all 23 oxylipins appeared as [M + Na]⁺ adducts, and many also as sodium acetate clusters in ESI(−), though with lower abundances. [M + Li]⁺ and [M + K]⁺ adducts were observed for over 85% of standards when their respective salts were present. Metal adducts of free oxylipins formed narrow mobility peaks and typically appeared as a single dominant feature, even after extended (>10 m) separations. These properties prompted a focus on metal adducts, particularly [M + Na]⁺, to optimize separation. Their stable conformations, strong ionization, and large ΔCCS% allowed for more effective resolution.

cIMS Behavior of Prostaglandins

Prostaglandins feature a characteristic cyclopentane ring, a C13–C14 double bond, and a hydroxyl group at C15. Metal adducts with sodium, potassium, and lithium enable the differentiation of both constitutional isomers and stereoisomers. For instance, the larger gas-phase structure of [M + Na]⁺ of PGF_2α differs markedly from its isomers PGD₁, PGE₁, and the epimer PGF_2β. Previous DTIMS studies showed Δ^DTCCS% of about 3% between PGF_2α and PGE₁, and about 4% between PGF_2α and PGD₁. , Meanwhile, PGF_2β, being more compact, is only partially separated from PGE₁, even with high-resolution demultiplexing (HRDm). Likewise, PGD₁ and PGE₁, which differ only by the position of the keto group at C11 versus C9, show limited separation as [M + Na]⁺ adducts in DTIMS.

Previous studies examined these isomers in only three species, but this study investigated all four. , Despite the added complexity, the FA 20:3;O3 mixture was clearly separated using cIMS after 8 meters of separation (Figure A). Before ion wrap-around, PGD₁ and PGE₁ nearly reached baseline resolution (R_pp = 1.48), and PGF_2β and PGE₁ were fully separated (R_pp = 2.68). The most challenging pair was PGD₁ and PGF_2β (R_pp = 1.07), but additional passes caused ion wrap-around, requiring cIMS/cIMS to further enhance resolution.

1.

cIMS–MS separations of FA 20:3;O3 (mixture 1) as their [M + Na]⁺ (A) and [M + K]⁺ (B) adducts, FA 20:4;O3 (mixture 4) as their [M + K]⁺ (C) and [M + Li]⁺ (D) adducts, and FA 20:5;O2 (mixture 6) as their [M + Na]⁺ (E) and [M + K]⁺ (F) adducts, under individually optimized TW conditions (Table S2). Separations were extended to the onset of ion wrap-around. Solid black traces represent isomeric mixtures, and dashed traces correspond to individual isomer standards analyzed separately.

The separation of the FA 20:3;O3 isomeric mixture serves as a clear example of how adduct profile affects ion mobility and can enhance the resolution of isomeric pairs. Potassium adducts changed the relative mobilities of prostaglandin isomers, and after a 10-meter separation, they improved resolution between PGD₁ and PGE₁ (R_pp = 2.38), as well as between PGD₁ and PGF_2β (R_pp = 2.12) (Figure B). Importantly, the migration order depended on the type of adduct: PGF_2α, as lithium and sodium adducts, arrived last and was well separated from the other isomers. Conversely, with potassium adducts, PGF_2α arrived before PGE₁, reducing the resolution between these species compared to sodium and lithium adducts (Figure S3A).

Adduct-dependent mobility was also observed for FA 20:4;O3 isomers. PGD₂ and PGE₂, which demonstrate nearly identical MS/MS fragmentation patterns and are usually separated by LC, were easily distinguished as sodium adducts (Figure S3B). However, adding the linear, LOX-derived LXA₄ revealed a limitation of sodium adduct separations. LXA₄ and PGE₂ remained unresolved before ion wrap-around and showed minimal improvement even after an additional 40 m of separation, consistent with their small reported Δ^DTCCS% (∼0.2%). Conversely, [M + K]⁺ (Figure C), [M + Li]⁺ (Figure D), and [M – H + CH₃COONa]⁻ (Figure S3C) all improved resolution between LXA₄ and PGE₂, although these enhancements involved shifts in migration order and trade-offs in resolution with PGD₂.

Cyclized dehydration products of PGE₂ and PGD₂, including PGA₂, PGB₂, and PGJ₂, form a set of isomers that exhibit ion wrap-around in as little as 3 meters, with Δ^DTCCS% values reaching up to 6.5%. While PGB₂ and PGJ₂ can be difficult to distinguish chromatographically, DTIMS has achieved their separation across various adduct profiles, and PGJ₂ and PGA₂ have also been nearly fully resolved by reversed-phase LC. In this study, all three isomers were examined together to evaluate mixture separation under conditions where ion wrap-around occurs rapidly. Lithium adducts showed significant separation between PGJ₂ and PGB₂ (R_pp = 6.02), but PGA₂ and PGB₂ were barely resolved (R_pp = 0.80) (Figure S3D). Although large ΔCCS% improves the separation of single species, it was, in this case, disadvantageous for the mixture, as it limited the separation window before wrap-around. More balanced CCS differences, such as those seen with sodium and potassium adducts, allowed baseline resolution of all three isomers within 3–4 meters (Figure E,F). While multiple mobility features were less common in metal adducts, PGA₂ displayed a minor additional mobility feature in sodium, lithium, and potassium adducts, which in some cases was fully resolved in a few passes. In contrast, sodium acetate clusters needed more than 10 meters for similar resolution, and multiple mobility features were not detected for PGA₂.

cIMS Behavior of Linear Oxylipins

A similar correlation between structural lability and adduct profile was observed in linear oxylipins, including those in the FA 20:4;O and FA 20:5;O mixtures. Metal adduct formation again resulted in narrow mobility peaks, usually with a single dominant feature. DTIMS analyses of FA 20:5;O isomers revealed that [M + Na]⁺ of 14(15)-EpETE and 15-oxo-ETE are well separated; however, adding more isomers reduced the separation and complicated compound identification. Although higher mixture complexity also limited resolution in cIMS, a separation path length of 7 m was enough to distinguish 14(15)-EpETE from 15-oxo-ETE and also 15­(S)-HEPE from 5-oxo-ETE, which showed intermediate arrival times (Figure A). All four FA 20:5;O isomers were additionally separated as potassium and lithium adducts, with a minimum resolution of R_pp ≈ 1.1 between neighboring peaks before ion wrap-around (Figure B and Figure S3E).

2.

cIMS–MS separations of FA 20:5;O (mixture 5) as their [M + Na]⁺ (A) and [M + K]⁺ (B) adducts, FA 20:4;O (mixture 2) as their [M + Na]⁺ (C) and [M + Li]⁺ (D) adducts, and FA 20:4;O2 (mixture 3) as their [M + Na]⁺ adducts (E) and [M – H + CH₃COONa]⁻ clusters (F), under individually optimized TW conditions (Table S2). Separations were extended to the onset of ion wrap-around. Solid black traces represent isomeric mixtures, and dashed traces correspond to individual isomer standards analyzed separately.

The FA 20:4;O mixture, which includes five linear oxylipins, proved more difficult to distinguish. da Silva et al. noted that a resolving power of approximately 180 is necessary to separate the constitutional isomers of this mixture as sodium acetate clusters. In cIMS, both 15-HETE enantiomers were partially resolved from epoxide-containing species beginning after 3–4 m of separation and optimized between 9–10 m as metal adducts in ESI­(+) (Figure C,D). Yet, 5-HETE showed intermediate arrival times, sometimes overlapping with nearby isomers depending on the adduct. Investigating sodium acetate clusters with cIMS revealed that the order of 5-HETE and the 15-HETE enantiomers shifted (Figure S3F) relative to sodium and lithium adducts, although both remained well separated from each other and from (±)-8,9-EpETrE. As expected, the 15­(R/S)-HETE enantiomers were not resolved under optimized TW conditions, even after extended runs of up to 40 m. Within this mixture, positional isomers such as 5-HETE and 15-HETE are also readily distinguished by reversed-phase LC, similar to the EpETrE regioisomer pair, and only the 15-HETE enantiomers remain challenging to resolve by either technique. Among these linear oxylipins, epoxide-containing species consistently had the shortest arrival times, consistent with previous findings by da Silva et al. However, the relative arrangement and separation of isomers depended heavily on factors such as adduct profile, functional groups, and number of double bonds.

FA 20:4;O2 proved comparatively resistant to full resolution, likely reflecting the close structural similarity among 5­(S),12­(S)-DiHETE, 6-trans-LTB₄, and 8­(S),15­(S)-DiHETE, which share the same diol substitution pattern and differ only in hydroxyl position and double-bond geometry. Linear FA 20:4;O2 oxylipin isomers were clearly separated after 7 m as sodium adducts and sodium acetate clusters. Adding PGA₁ reduced group separation in sodium adducts and positioned itself between 6-trans-LTB₄ and 5­(S),12­(S)-DiHETE. In ESI(−), PGA₁ sodium acetate clusters formed a more compact, early-moving peak, decreasing the number of passes before ion wrap-around (Figure E,F). These species also showed poor resolution as potassium adducts, and 6-trans-LTB₄ exhibited low ionization with lithium.

No single adduct consistently maximized resolution; sodium and potassium adducts generally offered the most balanced compromise between separation and ion wrap-around, while lithium adducts, despite often producing the largest ΔCCS%, more frequently induced premature wrap-around in complex mixtures. Across all tested adduct types and TW conditions, 65.3% of isomer pair measurements achieved baseline resolution, 22.1% were partially resolved, and 12.6% remained unresolved (Table S2). Within standard (single-pass) cIMS, only the enantiomeric 15­(R)/15­(S)-HETE pair and the positional epoxide isomers 14R,15S-EpETrE and (±)-8,9-EpETrE remained unresolved across all tested adducts. Isomers that are unresolved or only partially resolved by cIMS alone may benefit from LC–cIMS workflows; the EpETrE pair, for example, is routinely resolved by conventional reversed-phase LC via retention time, reducing reliance on extended path lengths and extensively optimized TW conditions. For LC-free workflows, particularly in complex mixtures of four or more oxylipin isomers, cIMS/cIMS experiments can instead be used to improve resolution.

cIMS/cIMS

Sodium adducts were chosen for cIMS/cIMS experiments because they generally produced single, distinct mobility features, showed efficient ionization, and exhibited little signal loss over long path lengths. Partially separated regions from multipass runs were isolated within the separation cell, transferred to the pre-store array, and reinjected for more passes. The regions, marked by vertical dashed boundaries in the ATDs, represent the mobility windows subjected to extended path lengths. Final reconstructed ATDs for each population were obtained by combining corrected, isolated portions of the ATD collected from individual reinjection events, after correcting for remixing within the pre-store array (see Figure S4). Regions outside the dashed boundaries were ejected directly to the detector without collection. While this method might be less effective for low-abundance species, cIMS/cIMS enhances the resolution of complex mixtures that would otherwise be limited by ion wrap-around in continuous cIMS experiments.

cIMS/cIMS was applied to mixtures of FA 20:3;O3, FA 20:4;O, and FA 20:4;O2 isomers (Figure and Figure S5). For FA 20:3;O3, a total pathlength of 18 m was required to resolve PGD₁ and PGF_{2 β}, which exhibit a small Δ^DTCCS of ∼0.57% (Figure A,B). Similarly, for FA 20:4;O2, a total pathlength of 17 m enabled resolution of the closely spaced three-component region comprising PGA₁, 6-trans-LTB₄, and 8­(S),15­(S)-DiHETE (Figure S5), where neighboring species differed in ^DTCCS by ∼0.9–1.2%.

3.

Multipass cIMS–MS and cIMS/cIMS–MS separations of FA 20:3;O3 (mixture 1) and FA 20:4;O (mixture 2) isomers as their [M + Na]⁺ adducts extended to the onset of ion wrap-around. The 8 m separation of FA 20:3;O3 was divided into three mobility slices defined by vertical boundaries (A), each extended by an additional 10 m, yielding a reconstructed ATD with a total pathlength of 18 m (B). The 9 m separation of FA 20:4;O was divided into two slices (C), each extended by 15 m to a total cIMS pathlength of 24 m (D). Solid traces represent mixtures; dashed traces represent individual standards.

The FA 20:4;O mixture, containing five isomers, was subjected to a combined 24 m of separation until the second onset of ion wrap-around in the three-component region highlighted in purple (Figure C,D). Although an additional 15 m of pathlength improved separation relative to cIMS experiments alone, complete resolution was not achieved. These observations are consistent with the small Δ^DTCCS% between adjacent 14R,15S-EpETrE and 5-HETE (∼0.49%) and between (±)-8,9-EpETrE and 14R,15S-EpETrE (∼0.56%), as well as the near co-mobility of 15­(R)- and 15­(S)-HETE. Together, these results demonstrated that cIMS/cIMS experiments can substantially improve resolution in complex oxylipin mixtures while highlighting limitations for isomers with minor differences in gas-phase structure.

Multipass CCS Calculations

Mobility values determined from first-principles in DTIMS are considered a primary standard and serve as the basis for calculating ^DTCCS values using the Mason–Schamp equation. These measurements are validated through rigorous quality control procedures and generally show excellent reproducibility, with intralaboratory RSD typically ≤0.1%. Moran-Garrido et al. conducted the largest interlaboratory comparison of oxylipin ^DTCCS_N2 values to date, evaluating oxylipins across five laboratories in both ionization modes. Average interlaboratory deviations were generally <1.5%, with maximum deviations up to 3%.

In TWIMS, CCS calibration is typically performed by fitting power functions or polynomials to reference ^DTCCS values for calibrant ions. Existing calibration strategies for ^cIMSCCS have largely been developed for single-pass separations; however, complex oxylipin mixtures often remain unresolved after only 1 m of separation. As a result, accurate multipass ^cIMSCCS values are highly desirable for future oxylipin studies in biological samples to improve annotation confidence. Extending ^cIMSCCS calibration to multipass experiments can be both time- and labor-intensive, requiring the generation of multiple calibration curves in which analytes and calibrants undergo identical separation path lengths.

Modern calibration strategies leveraging average ion velocities and periodic drift times, the duration to complete a cycle through the separation region, have been shown to produce accurate ^cIMSCCS values, usually within about 1% of ^DTCCS reference data. , In this study, a periodic drift time-based method, developed by Lin and Costello, was used to address perturbations in electric potential caused by field switching during ion ejection. It also provides a simple, practical framework for calculating multipass ^cIMSCCS. Calibration was carried out using t_pp values, derived over various separation times (Figure S6A). Further corrections included adjusting CCS values for ion charge state and reduced mass, as well as refining t_pp values to account for mass-dependent flight-time variations (Figure S6B). Multipass ^cIMSCCS_N2 values for oxylipin standards were then derived from their respective t_D′ measurements under the same TW conditions and compared to literature ^DTCCS_N2 values (Figure A–C and S6C). Selecting calibrant species carefully was critical to avoid multiple mobility features at longer path lengths, which can complicate Gaussian fitting.

4.

Comparison of calculated multipass ^cIMSCCS_N2 values with ^DTCCS_N2 reference values for [M + Na]⁺ adducts of oxylipins. Scatter plots are shown for three TW conditions optimized for oxylipins: 15 V/350 m/s, 18 V/350 m/s, and 20 V/450 m/s (A–C). The dashed line represents unity, and data points are colored by isomeric mixture. Variability in multipass ^cIMSCCS_N2 within each TW condition is shown, where horizontal bars indicate the average and dots indicate the maximum percent |ΔCCS| (D). The average and maximum percent |ΔCCS| values between multipass ^cIMSCCS_N2 and ^DTCCS_N2 are summarized in (E). The complete list of multipass ^cIMSCCS_N2 and ^DTCCS_N2 values are provided in the Supporting Information (Table S3).

Calibration curves were created across four TW conditions using Multipass CCS Refiner to assess variability in multipass ^cIMSCCS_N2 under conditions mimicking optimized separations for complex oxylipin mixtures (Figure S7). [M + Na]⁺ adducts were chosen for these experiments because they were common to all species, well-documented in existing libraries, and produced single mobility features. Pairwise comparison of [M + Na]⁺ oxylipin isomers across TW conditions showed average ^cIMSCCS_N2 differences ranging from 0.20% to 0.77%, with a maximum single measurement deviation of 1.44% for (±)-8,9-EpETrE (Figure D). Out of 92 measurements (23 analytes tested under four TW conditions), only 12 values differed by more than 1.0% from the ^DTCCS_N2 reference data, with overall average deviations between 0.33 and 0.76% (Figure E). This consistency aligns with known interlaboratory variability and supports the reliability of multipass ^cIMSCCS for oxylipin analysis. Implementing calibration workflows applicable across various cIMS TW conditions and adaptable to cIMS/cIMS should further improve the analysis of complex oxylipin mixtures. Recent studies indicate that multipass ^cIMS/cIMSCCS can be calculated by modifying t₀ to account for all injection, separation, and storage steps outside the final separation, with platforms such as SLIM further validating TWIMS CCS values over longer path lengths. ,

Conclusions

This work demonstrates that adduct selection and TW parameter optimization are key variables for resolving complex oxylipin mixtures with cIMS, and that meaningful separations of isomeric species are achievable, even at short effective path lengths. Although annotation of oxylipins is challenging in the absence of retention time information, our findings support multipass ^cIMSCCS as a reliable annotation metric to facilitate this task, which we believe will be particularly important for species such as PGE₂ and PGD₂, whose MS/MS fragmentation spectra are nearly identical. While minimizing path length mitigates ion losses inherent to extended cIMS separations, the sensitivity of cIMS–MS platforms may still be the ultimate limiting factor for trace-level analysis of oxylipins; enhanced ionization strategies, such as the addition of metal salts to the sprayed solvent, will likely be necessary for improving oxylipin detection and increasing coverage.

LC–cIMS-MS approaches offer complementary advantages in oxylipin studies, including reduced sample complexity, decreased ion suppression, and orthogonal separation, which will likely reduce the need for extensive TW optimization and extended cIMS/cIMS separations. Nevertheless, for complex samples analyzed without LC, longer separation path lengths, such as those achievable with cIMS/cIMS, will likely be necessary to fully resolve the oxylipin isomeric landscape. We anticipate that high-quality cIMS/cIMS CCS measurements will play an increasingly important role in oxylipin identification, enabling more confident annotation in both LC-coupled and LC-free workflows, ideally through methods that remain time-efficient and broadly accessible to the community.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.analchem.6c03450.

• Detailed chemical standard information, free oxylipin structures, cIMS–MS instrumental parameters, and separation path lengths for the workflows are provided. Additional ATD data for oxylipin separations are included for deprotonated and metal-adduct species, as individual and isomeric mixtures. Additional cIMS/cIMS experiments, including reconstruction procedures for extended separations, are presented. Calibration and multipass CCS determination workflow plots are included for regression analyses across multiple TW conditions. Calculated CCS values, R_pp, and comparisons between multipass ^cIMSCCS_N2 and literature ^DTCCS_N2 values are also provided (PDF)

E.C.G.: Conceptualization, methodology, investigation, data curation, formal analysis, writing (original draft), and writing (review and editing). F.M.F.: Conceptualization, methodology, resources, supervision, and writing (review and editing). All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.