Ion Mobility-Enhanced LA-REIMS Improves Molecular Resolution in Ambient Biofluid Metabolomics

13 14 Ambient metabolomics techniques such as laser -assisted rapid evaporative ionization mass 15 spectrometry (LA -REIMS) enable fast, preparation-free fingerprinting of biological samples 16 but are inherently limited by spectral congestion in the absence of chromatographic 17 separation. While ion mobility spectrometry provides additional gas-phase separation, 18 maintaining ion transmission under the transient signals' characteristic of laser desorption , 19 remains analytically challenging. Here, we define operating conditions for cyclic traveling -20 wave ion mobility spectrometry (cIMS) that preserve transmission under LA-REIMS duty-cycle 21 constraints and systematically evaluate how cIMS integration reshapes biofluid fingerprints 22 and enhances chemical specificity in chromatography-free metabolomics analysis. 23 24 Under optimized single-pass conditions, cIMS separation reorganized LA-REIMS spectra into 25 structured mass /mobility feature domains, enabling selective mobility -based filtering of 26 matrix-derived salt cluster ions. This reduced non-biological background contributions by up 27 to 35% of total spectral intensity while preserving over 90% of detected untargeted features. 28 Although cIMS operation introduced a sensitivity penalty relative to time -of-flight-only 29 acquisition, approximately 80% of the total ion current was recovered under optimized 30 conditions. Mobility-resolved data revealed coherent homologous series and class -specific 31 structural trends, particularly for lipids, supporting class -level annotation. Analysis of 101 32 metabolite and lipid standards covering a broad physicochemical range (logP -5.30 to 19.40) 33 demonstrated comprehensive molecular coverage, high mass accuracy (mean 2.4 ppm), and 34 good agreement with reference CCS values (mean deviation 4 .0%), with isomer separation 35 observed for biologically important secondary bile acids in extended separation cycles . 36 Collectively, these results establish LA -REIMS-cIMS as a practical analytical strategy for 37 enhancing chemical specificity and spectral interpretability in support of high -throughput 38 large-scale metabolic fingerprinting. 39 40 41

ambient ionization mass spectrometry ; cyclic ion mobility spectrometry; high-42 throughput metabolomics; biofluid analysis; matrix interference; isomer discrimination ; 43 collision cross section 44 45 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 2 46 Graphical abstract: Ion mobility spectrometry adds an orthogonal gas -phase separation to 47 LA-REIMS, reorganizing complex biofluid spectra into distinct mass -mobility feature bands 48 and improving molecular resolution in rapid ambient ionization metabolomics. 49 50

51 52 Ambient ionization mass spectrometry (AIMS) enables rapid molecular fingerprinting of 53 biological samples with minimal or no sample preparation and without chromatographic 54 separation, in contrast to conventional metabolomics workflows 1,2. These characteristics 55 support high analytical throughput and operational simplicity, making AIMS particularly 56 attractive for in -field applications, large -sample-number metabolomics studies, and 57 workflows requiring fast, on -site analytical feedback, such as epidemiological screening, 58 environmental monitoring, and manufacturing process control , among others 1,3,4. In 59 biomedical contexts specifically, AIMS has attracted growing interest for the sensitive 60 molecular characterization of clinical specimens, where rapid and robust analysis can support 61 point-of-care decision making and population -level surveillance initiatives aligned with 62 emerging 5P medicine frameworks5,6. 63 64 Among AIMS techniques, laser-assisted rapid evaporative ionization mass spectrometry (LA -65 REIMS) occupies a distinct position due to its applicability to a wide range of biological 66 samples, including non-invasively collected urine, saliva, and feces 7. LA-REIMS generates ions 67 via rapid laser desorption and post -ionization, producing short -lived ion populations that 68 closely reflect the bulk biochemical composition of the sample 8. While this mechanism 69 enables exceptional analytical speed and robustness to sample morphology, it also leads to 70 highly congested mass spectra characterized by isobaric overlap, multiple adduct formation, 71 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 3 and matrix-derived background signals, particularly in complex biofluids 9. In the absence of 72 chromatographic separation and prolonged ion accumulation , opportunities to disentangle 73 endogenous metabolites from co -ionized matrix components are inherently limited. 74 Improving chemical specificity without compromising throughput therefore remains a central 75 challenge for biological fingerprint interpretation and confident compound assignment in 76 AIMS-based biofluid metabolomics10. 77 78 In this context, ion mobility spectrometry (IMS) offers a promising, non -chromatographic 79 means of increasing chemical specificity while preserving the rapid analysis characteristic of 80 ambient ionization workflows 11. By introducing a gas -phase separation based on ion size, 81 shape, and charge, IMS adds an orthogonal conformational dimension to datasets acquired 82 without prior chromatographic separation, that redistributes ions across arrival time, 83 reducing spectral congestion and enabling more reliable association of signals with individual 84 compounds12. In addition, IMS allows for the determination of collision cross section (CCS) 85 values, which are more transferable across platforms than chromatographic retention times 86 and can support compound annotation through standardized reference databases 13. These 87 attributes have established IMS as a valuable complement to mass spectrometry and 88 motivate its systematic evaluation in AIMS workflows under the stringent duty -cycle 89 constraints imposed by ambient ionization. 90 91 Among available IMS methods, cyclic traveling -wave ion mobility spectrometry (cTWIMS , 92 cIMS) offers a combination of features that are particularly relevant for addressing these 93 constraints. In uniform-field drift tube IMS, resolving power is fixed and linked to drift time 94 (and thus drift length and operating conditions), creating an inherent limitation to resolution 95 and separation time tuning14. By contrast, cTWIMS enables users to select the effective 96 separation length by varying the number of passes around the cyclic device, increasing 97 resolving power with longer cycle times15. Unlike differential mobility methods such as high-98 field asymmetric waveform IMS , which function as differential ion filters with transmission 99 efficiencies that can vary strongly between ion classes, cTWIMS operates as a time-dispersive 100 separation that can be acquired in a broadband manner 16,17. This enables short, single -pass 101 operation for high -throughput measurements while retaining the option to extend 102 separations for targeted analyses when additional mobility resolution is required. 103 104 Building on these considerations, this work presents the integration of cIMS with LA -REIMS 105 for high -throughput biofluid analysis. While cIMS has previously been applied in 106 chromatography-free settings12, its implementation has not previously been adapted for ion 107 sources generating transient ion packets , such as LA-REIMS. Here, we explore the potential 108 of cIMS to operate in a manner compatible with transient ion generation and rapid 109 acquisition, while providing mobility -resolved information that enhances spectral 110 interpretability. By evaluating mobility performance, ion transmission, and duty -cycle trade-111 offs, and by assessing the extent to which mobility separation mitigates matrix-driven spectral 112 complexity while enabling metabolite class annotation and isomer discrimination in the 113 absence of chromatography, this study defines both the potential and practical limits of cIMS 114 for fit-for-purpose integration into ambient biofluid metabolomics. 115 116 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 4

117 118 1. Chemical standards and solvents 119 120 Isopropyl alcohol (IPA, LC -MS grade) was purchased from Fisher Scientific (UK). Ultrapure 121 water (UPW) was produced using a Sartorius Arium 661 UV water purification system 122 (Sartorius, Belgium). Analytical standards were purchased from Sigma -Aldrich (USA), ICN 123 Biomedicals Inc. (USA), TLC Pharmchem (Canada), or Waters Corporation (USA) as listed 124 previously18. Sodium chloride (NaCl, food -grade purity) was dissolved in ultrapure water to 125 yield a 0.9% (w/v) solution corresponding to isotonic saline conditions. 126 127 2. Samples and study design 128 129 Biological samples were obtained from existing, ethically approved clinical cohorts of which 130 re-use was allowed via biobank BC-184. Saliva samples originated from the FAME study19 and 131 urine and feces from the MetaBEAse cohort5. Study specimens were composed as pooled 132 samples comprising equal parts from three normal -weight and three overweight children, 133 classified according to International Obesity Task Force (IOTF) criteria 20, extending chemical 134 diversity of the biological pool by including individuals with different phenotypical traits. 135 Pooling was applied to minimize inter-individual biological variability and to enable controlled 136 assessment of analytical performance. 137 138 Following at-home collection, samples were immediately frozen at -18 °C. Upon receipt at 139 LIMET (within 48 h), fecal samples were freeze -dried, homogenized by grinding and sieving, 140 and subsequently stored at -80 °C to rapidly quench ongoing microbial, enzymatic, and 141 chemical degradation21. Saliva and urine samples were stored at -80 °C without additional 142 processing. Prior to analysis, samples were thawed at room temperature (22 ± 2 °C). Urine 143 and saliva samples were vortex -mixed and aliquoted (100 µL) into 96 -well plates for direct 144 analysis without further purification 7. Freeze-dried fecal samples were resuspended with 145 UPW (1:4, w/v) to generate fecal water, facilitating homogenization and subsequent infrared 146 laser desorption 5. An aqueous sodium chloride solution (0.9% (w/v), physiological saline 147 concentration) prepared in ultrapure water was analyzed in triplicate under identical LA -148 REIMS-cIMS conditions and used as a reference for the identification of salt -derived cluster 149 ions in the ion mobility space. 150 151 3. Instrumentation and acquisition modes 152 153 LA-REIMS experiments were performed using a mid -infrared laser ablation system 154 (Opolette™ 2940, OPOTEK, USA) coupled to a REIMS ionization source (Waters Corporation, 155 Manchester, UK) as described previously7. A continuous flow (0.2 mL/min) of IPA was supplied 156 to the source throughout all experiments to support post -ionization and ion transfer. The 157 REIMS ionization source was coupled to a SELECT SERIES™ Cyclic™ traveling-wave ion mobility 158 (cIMS) time-of-flight (ToF) MS (Waters Corporation, Manchester, UK)15. Nitrogen was used as 159 the ion mobility drift gas. Data were acquired in both positive and negative ionization modes 160 for targeted experiments, with polarity -specific lock -mass correction applied as described 161 below. Untargeted biofluid fingerprinting experiments were performed in negative ionization 162 mode only, because positive mode produced high levels of background -derived peaks . 163 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 5 General ion optics parameters, solvent delivery conditions, and ToF scan rates were based on 164 previously optimized biofluid -specific LA -REIMS protocols 7 and adapted to the instrument 165 configuration used in this study. Detailed parameter settings are provided in SI Note 1. 166 167 4. cIMS operation and optimization 168 169 cIMS method optimization was performed using pooled urine as a representative biological 170 matrix, selected due to its high molecular fingerprint reproducibility, as previously 171 demonstrated for ToF-based LA-REIMS analyses7. Urine was used consistently throughout 172 cIMS optimization to minimize matrix -related variability and enable robust comparison of 173 experimental conditions. The cIMS mode was initially operated using default (manufacturer-174 set) parameters (total cycle time 51.6 ms, traveling wave (TW) velocity 375 m/s, TW height 175 26 V). Optimized cIMS experiments were performed using a TW voltage ramp (total cycle time 176 37.8 ms, TW velocity 375 m /s, TW starting height 10 V, ending height 26 V, ramping rate 1 177 V/ms) in combination with optimized ToF parameters ( SI Note 1). The cIMS cycle time , and 178 TW ramp starting and ending voltages were optimized iteratively during acquisition by 179 monitoring the total ion current (TIC). TW ramping rate and wave velocity were further fine-180 tuned using Design of Experiments (DoE) modelling to promote a broad ion arrival -time 181 distribution, thereby enhancing ion mobility separation and resolving power under short 182 cycle-time conditions as detailed in SI Note 2. Benchmarking experiments were conducted 183 using serial measurements (n = 10) of pooled urine sample aliquots under each acquisition 184 mode to allow direct comparison of signal intensity, feature detection, and reproducibility. 185 186 5. Untargeted LA-REIMS(-cIMS) 187 188 Untargeted metabolomics analyses were performed on pooled urine, saliva, and fecal water 189 samples (n = 10 per matrix). Data were acquired using both ToF -only acquisition ( cIMS 190 bypassed), default (manufacturer-set) single-pass cIMS mode and optimized single-pass cIMS 191 conditions. Physiological saline samples were analyzed to support identification of salt -192 derived cluster ion bands in mobility-resolved spectra. Mobility-domain filtering was applied 193 by excluding arrival-time regions corresponding to these clusters. Average feature counts and 194 summed feature intensities were compared before and after filtering to assess the 195 contribution of salt-derived clusters to the overall signal structure. 196 197 6. Targeted LA-REIMS-(cIMS) 198 A panel of 101 in-house metabolite and lipid standards spanning a wide range of molecular 199 weights (88.02 to 1152.71 Da ) and lipophilicity (logP -5.30 to 19.40), covering the major 200 metabolome and lipidome (9 HMDB superclasses and 39 HMDB classes22) was analyzed to 201 assess compound detectability, ion transmission, and annotation performance (SI Table 4). 202 To isolate ToF and cIMS transmission effects from laser -sample interactions, standards were 203 introduced by direct infusion into the REIMS source. 204 Standards were grouped into two polarity -specific mixtures (polar metabolites and lipids 205 prepared separately based on stock solution compatibility), with each compound present at 206 a final concentration of 10 ng/µL. When required, mixtures were reanalyzed at 100 ng/µL per 207 compound ( SI Table 4). IPA blanks were measured before analytical standard mixtures in 208 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 6 triplicate for downstream background subtraction. Each mixture was analyzed in triplicate in 209 both positive and negative ionization modes and detectability metrics were based on 210 averaged measurements. Lock-mass correction was applied using cholic acid (negative mode) 211 and anandamide C20:4 (positive mode), selected as polarity -specific references due to their 212 consistent high-intensity signals under REIMS conditions. For each measurement, 50 µL of the 213 analytical standard mixture was infused and the source was flushed with 500 µL IPA after each 214 acquisition. 215 In ToF-only mode, compound matching was based on averaged m/z values using a tolerance 216 of 10 ppm. In cIMS -enabled mode, compound matching was based on combined averaged 217 m/z and collision cross section (CCS) values. Averaged m/z and CCS values were calculated 218 from replicate measurements (n = 3), except for compounds used for CCS calibration, for 219 which two replicates were averaged and one reserved exclusively for calibration. CCS 220 calibration was performed post-acquisition with subset of standards with available literature 221 CCS values spanning a broad m/z and mobility range (SI Table 3). For cIMS-based evaluation, 222 a CCS deviation cut-off of 10% was applied to compounds with available empirical reference 223 CCS values from open databases, as detailed in SI Table 5. 224 Analytical standards of secondary bile acids (deoxycholic acid, ursodeoxycholic acid, 225 chenodeoxycholic acid and their glycine - and taurine -conjugated forms; Table 1) and 226 positional isomers of phosphocholine lipids (PC 16:0/18:0 and PC 18:0/16:0) . were analyzed 227 at 10 ng/µL concentration in triplicate, accompanied by solvent blanks, as described above. 228 Compounds were first analyzed individually (10 ng/µL) to record arrival times and 229 subsequently as defined mixtures under single -pass and extended cIMS conditions (2 - 200 230 ms effective separation time). All other source, ion optics, and ToF parameters were held 231 constant. Arrival -time distributions of deprotonated monomers and selected adduct or 232 cluster ions were compared across individual and mixed analyses. 233 234 7. Data processing and statistical analysis 235 236 Raw mass spectra were acquired using MassLynx (v. 4.2, Waters Corporation). Spectra 237 containing mobility information were processed in DriftScope (v. 3.0, Waters Corporation) for 238 individual compound detection and with MS -DIAL23 (v. 5.0, RIKEN) for batch processing . 239 Progenesis Bridge and QI (v. 2.3, Waters Corporation) w ere used for signal alignment, 240

subtraction, peak picking and TIC normalization. DoE modelling24 for ToF and IMS 241 parameters optimization was performed in JMP® (v. 18, SAS Institute Inc., Cary, NC, USA) , 242 fitting standard least -squares regression to assess response effects . Model effects were 243 evaluated by analysis of variance (ANOVA), and factor significance was determined using F -244 tests with a significance threshold of α = 0.05 and using FDR correction . The JMP Prediction 245 Profiler was used to visualize factor effects and to identify parameter settings that maximized 246 feature count and repeatability based on model response predictions. 247 248 For selected untargeted analyses, mobility-domain filtering was applied by excluding arrival-249 time regions corresponding to salt-derived cluster ion bands identified in IMS m/z vs mobility 250 arrival time maps. Feature counts and summed feature intensities were calculated before and 251 after filtering to assess the contribution of these clusters to overall signal structure. 252 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 7 253 For targeted data analysis, .raw directories were converted to .mzML format using 254 msConvert25. The resulting files were then visually curated in a custom R Shiny application to 255 confirm the presence of the expected compound peaks (SI Note 3). Target lists were 256 generated based on the known compound identity and expected adducts per polarity. For 257 each replicate, 10 scans around the TIC apex were selected and summarized. Detected peaks 258 were manually curated based on signal presence in samples relative to blanks, and detection 259 quality metrics (mass deviation before and after lock mass correction , signal-to-noise (S/N) 260 ratio were recorded. Curated results were exported for downstream analysis. 261 262 CCS calibration was performed using a subset of experimentally detected standards with 263 visually confirmed peaks and reference CCS values obtained from the Ion Mobility Collision 264 Cross Section Compendium 26. Calibrants were selected to span a broad m/z and CCS range 265 and to avoid isomeric or closely related compounds. A peak intensity threshold of 80 counts 266 (negative mode) or 100 counts (positive mode), based on polarity -specific noise levels, and 267 an m/z tolerance of 0.1 Da were applied during calibration. 268 269 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 8

and discussion 270 1. Duty-cycle-aware optimization enables compatible integration of cIMS with LA-REIMS 271 Initial experiments were designed to evaluate the impact of cIMS incorporation on the quality 272 of untargeted LA -REIMS biofluid fingerprints, using pooled urine (10 replicates) as a 273 representative matrix. When cIMS was enabled using manufacturer -set default parameters, 274 a pronounced reduction in overall sample signal was observed compared to ToF -only 275 acquisition. The average total ion current (TIC) decreased by approximately one order of 276 magnitude, from 2.87 × 10⁷ ± 5.24 × 10⁵ in ToF mode to 1.54 × 10⁶ ± 2.59 × 10⁵ in default cIMS 277 mode (Fig. 1A). In addition to reduced TIC, relative peak abundances were markedly altered 278 across the full detected mass range (50 -1200 Da), with both decreases and increases in 279 individual feature intensities observed upon cIMS activation compared to ToF -only 280 acquisition (Fig. 1, A-C, right panel ). These observations are consistent with duty -cycle and 281 ion transmission losses associated with ion mobility operation under the transient ion 282 generation conditions characteristic of LA -REIMS, indicating that default cIMS settings 283 optimized for more continuous ion sources are not directly compatible with ambient 284 ionization workflows27. This prompted systematic optimization toward duty -cycle-aware 285 cIMS operation. 286 To mitigate signal loss, optimization efforts focused on reducing the total cIMS duty cycle 287 while maintaining sufficient ion mobility separation and avoiding wrap -around artifacts28,29. 288 Shortening the total cycle time from default 51.6 to 37.8 ms and introducing a traveling-wave 289 (TW) voltage ramp during both the mobility separation and ion ejection phases , described 290 previously as signal temporal compression effect 30, resulted in substantial signal recovery. 291 Under these conditions, the average TIC increased to 1.23 × 10⁷ ± 1.90 × 10⁵ ( Fig. 1B), 292 representing a near order -of-magnitude improvement relative to default cIMS operation . 293 Moreover, the relative abundance of individual peaks in the averaged untargeted urine 294 fingerprint was comparable to that observed in ToF-only acquisition (Fig. 1C), indicating that 295 the optimized cIMS settings restored signal transmission while largely preserving spectral 296 profiles. 297 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 9 298 Fig. 1. Comparison of the spectral signal of pooled urine following LA-REIMS analysis under 299 different acquisition modes. The total ion current (TIC) of the pool is presented left, specific 300 molecular features in the spectra are presented right: (A) ToF-only acquisition (ion mobility 301 bypassed), (B) optimized IMS transmission conditions and (C) default cIMS settings. 302 Following establishment of a shortened, wrap -around-free cycle, a design -of-experiments 303 (DoE) approach was applied to evaluate the influence of TWIMS parameters on analytical 304 performance ( SI Note 2). Effect tests showed that the TW ramping rate was the primary 305 determinant of total feature count (F = 4.59, p = 0.043), whereas TW velocity exhibited a 306 weaker, non -significant effect (F = 3.18, p = 0.088). The lowest TW height ramping rate 307 achievable by the instrument (1 V/ms) produced the highest number of detected untargeted 308 features within the tested voltage bounds. This observation is consistent with TW height 309 governing ion transmission efficiency across broad ion mobility distributions 31. Repeatability 310 analyses reinforced this trend, with TW ramping rate showing the strongest overall model 311 impact (F = 35.61, p < 0.0001). Although increasing TW velocity ( from 375 to 425 m/s) 312 produced modest gains in total feature counts, these improvements did not translate into 313 higher numbers of reproducibly detected features. Collectively, these findings establish TW 314 ramping rate as the dominant control parameter for robust feature detection under duty -315 cycle restricted conditions, whereas velocity adjustments provide only marginal, non -316 reproducible gains (detailed parameter settings in SI Table 2). 317 318 2. LA-REIMS-cIMS supports selective filtering of matrix -derived background signals from 319 raw biofluid spectra 320 321 Having established stable single-pass cIMS operating conditions, we next examined how the 322 addition of mobility information reshapes the structure and chemical composition of 323 untargeted LA-REIMS biofluid fingerprints. For this purpose, pooled biofluid samples (urine, 324 saliva, and feces; 10 replicates each ) were analyzed under these conditions (SI Table 2) to 325 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 10 evaluate systematic signal organization across the m/z and ion mobility arrival -time 326 dimensions. Ion mobility separation revealed distinct, recurring signal populations within the 327 overall feature space (Fig. 2). 328 329 In untargeted LA -REIMS analyses of urine and saliva, but not feces, ion mobility separation 330 consistently revealed a secondary signal band, defined as a dense, contiguous population of 331 features in the arrival-time space32, in addition to the primary metabolite-rich band (Fig. 2A). 332 This secondary band was characterized by highly regular m/z spacing and arrival time patterns 333 distinct from the endogenous metabolite signal space. Based on the observed spacing and 334 mobility behaviour, these features were tentatively attributed to salt -derived cluster ions, 335 which have been reported previously (in e.g. electrospray and matrix -assisted desorption 336 ionization33,34) in relation to biofluids and other salt -rich matrices. The assignment of this 337 secondary mobility band to salt-derived clusters was experimentally confirmed by analysis of 338 a physiological sodium chloride solution under identical acquisition conditions, which 339 reproduced the characteristic mobility pattern observed in urine and saliva samples (SI Fig. 340 1). Accordingly, the secondary mobility band predominantly reflects a systematic inorganic 341

that does not contribute to the intrinsic biochemical fingerprint of the sample 342 and primarily reduces analytical clarity by obscuring biologically informative metabolite 343 features. 344 345 Fig 2. Ion mobility -resolved identification and filtering of salt -derived cluster ions in 346 untargeted LA -REIMS-cIMS biofluid spectra. (A) Representative m/z versus ion mobility 347 arrival-time heatmap of single pooled urine spectrum showing a primary metabolite -rich 348 mobility band and a secondary high -density band attributed to salt -derived cluster ions 349 (highlighted in green). (B) Mirrored raw-intensity spectra (550-800 m/z) of averaged pooled 350 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 11 urine processed without (blue) and with mobility filtering (orange). The broken y-axis (0-80 351 and 150-200 intensity counts) highlights dominant salt clusters in unfiltered data and their 352 elimination after filtering. (C) Feature count and summed intensity before (blue) and after 353 mobility filtering (orange), calculated from average values and expressed relative to unfiltered 354 spectra (100%). Feature number is largely retained (94.3%), whereas total intensity decreases 355 (66.4%), indicating selective removal of high-intensity matrix signals. 356 357 Because the observed salt-cluster band overlaps with the mass range in which many 358 endogenous metabolites are detected (approximately 300 to 800 Da), it can artificially inflate 359 the number of detected spectral features and interfere with fragmentation spectra. In 360 addition, these cluster ions are typically among the most intense species in the spectrum, 361 thereby dominating the overall signal intensity and increasing ion competition during 362 acquisition. Notably, filtering out th is salt-cluster mobility band from urine spectra reduced 363 the number of detected peaks by only 5.7% (from 1693.5 ± 55 to 1596.5 ± 51), whereas the 364 summed intensity of all detected peaks decreased by 3 3.6% (from 2.14 × 10⁴ to 1.42 × 10⁴). 365 Importantly, filtering out this dominant background also reduced spectral overlap and 366 improved detectability of low -abundance organic features from the same region, without 367 substantially altering overall feature count compared to non-filtered spectral data (Fig 2B and 368 C). It is imperative to acknowledge though that ion mobility filtering is incapable of eradicating 369 undesirable signals in untargeted spectra that do not demonstrate distinguishable mobility 370 behaviour (e.g., detector-associated artifacts, SI Fig. 2). In such instances, its implementation 371 should be accompanied by additional computational background signal removal 372 methodologies, such as peak intensity and S/N ratio thresholds and blank subtraction34. 373 374 3. Targeted profiling of biofluid metabolome and lipidome class representatives 375 demonstrates broad molecular coverage of LA-REIMS-cIMS 376 377 To evaluate the intrinsic molecular coverage of LA -REIMS and to assess how ion mobility 378 integration alters compound detectability and annotation potential, a targeted panel of 379 analytical standards representative of major metabolome and lipidome classes (n = 101, logP 380 -5.30 to 19.40, SI Table 4) was analyzed under both ToF-only and cIMS acquisition modes. 381 382 Under ToF -only acquisition, all standards (100% of the panel) were consistently detected 383 across replicate analyses (n = 3), demonstrating broad intrinsic molecular coverage of REIMS 384 for chemically diverse small molecules and lipids (Fig. 3). A total of 63 compounds (62% of the 385 panel) were observed in both ionization modes, while a subset showed polarity -specific 386 detection (24 negative-only, 14 positive-only; SI Table 4), reflecting expected class-dependent 387 ionization behaviour (e.g., preferential protonation of acylcarnitines35). Multiple adduct types 388 were frequently observed, including chloride adducts in negative mode and alkali metal 389 adducts positive mode, consistent with the presence of inorganic salts in untreated biological 390 matrices36,37. 391 392 By comparison, UHPLC-HRMS analysis of the same analytical standard panel, performed using 393 a Vanquish™ Duo UHPLC system coupled to an Orbitrap Exploris™ 120 HRMS (HESI-II source, 394 polarity switching) and based on our previously validated fecal metabolomics and lipidomics 395 methods18, yielded spectra dominated by protonated and deprotonated molecular ions. This 396 difference in adduct distribution reflects effective salt removal during chromatography and 397 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 12 differences in ionization physics rather than intrinsic differences in compound detectability18. 398 Despite differences in ionization mechanisms , compound detectability showed substantial 399 overlap between REIMS and HESI -based ionization, with no systematic compound -class-400 specific ionization bias observed within the tested panel and concentration range. 401 402 403 Fig 3. Chemical space and detection characteristics of the targeted metabolite panel 404 following LA-REIMS-cIMS analysis. (A) Nested donut chart summarizing the taxonomic 405 distribution and polarity coverage of the 101 -compound targeted panel. The outer ring 406 represents HMDB superclasses 22, with segment size proportional to the number of 407 compounds in each superclass (n = 101). The inner ring indicates polarity coverage. ( B) 408 Distribution of targeted compounds in physicochemical space plotted as m/z versus logP. 409 410

of cIMS under optimized conditions resulted in a reduction in compound 411 detectability, with 85 of the 101 standards remaining detectable (Fig 3B; SI Table 5). To assess 412 whether compound loss followed systematic trends, detectability was examined as a function 413 of singly charged m/z and polarity (LogP). No clear monotonic dependence was observed 414 across either parameter, indicating that transmission through the ion mobility device is not 415 governed by simple univariate physicochemical cut -offs. Nonetheless, compounds at the 416 lower extreme of the investigated m/z range (≤ 100 Da; e.g., valeric and isovaleric acid) and 417 those exhibiting low signal intensities under ToF-only conditions were more frequently absent 418 under cIMS conditions. These boundary effects suggest that transmission losses arise from 419 multivariate ion transport and duty -cycle constraints rather than predictable linear trends . 420 This observation is c onsistent with the results of previous TWIMS/cIMS studies, which 421 indicated that ions with low m/z values are more prone to probabilistic transmission loss38,39. 422 Importantly, however, the majority of structurally and chemically diverse standards remained 423 detectable in cIMS, indicating that mobility integration largely preserves broad metabolome 424 coverage at the compound class level. 425 426 Beyond detectability, the targeted standards dataset was used to evaluate the accuracy of 427 feature descriptors underpinning untargeted analysis, namely mass accuracy and collision 428 cross section (CCS) performance under LA-REIMS-cIMS conditions (SI Table 5). Following CCS 429 calibration, the relative error in CCS assignment across detected standards was moderate 430 (mean deviation 4.0%), while mass accuracy after drift-time correction remained high (mean 431 2.4 ppm). The observed CCS deviations are consistent with previously reported inter -432 laboratory and inter -platform variability and likely reflect systematic differences between 433 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 13 literature CCS values , frequently determined using drift tube or other IMS platforms, and 434 those measured on the cyclic TWIMS used here13. Although CCS accuracy alone is insufficient 435 for low-tier compound identification in untargeted scenarios 40, the combination of high -436 accuracy m/z measurements with reproducible CCS values provides orthogonal information 437 that improves downstream feature matching and annotation beyond what accurate mass 438 alone can achieve 41. In the context of ambient LA -REIMS analysis, this combined descriptor 439 supports class -level metabolite annotation, facilitates exclusion of false positives, and 440 provides structural context that is otherwise inaccessible in the absence of chromatographic 441 separation. 442 443 4. LA-REIMS cIMS reveals lipid organization and enhances isomer discrimination via 444 multiple-cycle separation 445 446 Next, we examined how ion mobility separation extends the structural interpretability of 447 untargeted LA -REIMS datasets beyond targeted compound detection and enables 448 discrimination of closely related isomeric species within an ambient workflow. Across pooled 449 urine samples, mobility-resolved LA-REIMS spectra revealed compact ion clusters occupying 450 shared arrival-time domains and clearly separated from the salt-cluster band (Fig. 4A). These 451 features, predominantly within the m/z range of 500-800 Da, formed coherent mobility bands 452 consistent with lipid homologous series. Within these domains, ions sharing common 453 backbone structures and similar total carbon numbers grouped together, while variations in 454 acyl-chain composition produced systematic yet subtle mobility offsets 42. Comparable lipid 455 organization was also observed in saliva and fecal pools, indicating that mobility -resolved 456 clustering of lipid homologues is a consistent feature across biofluid matrices. 457 458 To illustrate this behaviour in a chemically simplified system, fecal free fatty acids were 459 investigated more in depth (Fig. 4B). Unlike complex glycerophospholipids, free fatty acids 460 lack bulky headgroups and extensive adduct diversity, providing a clearer view of mobility 461 shifts associated solely with saturation state. Adjacent species differing by one double bond 462 were separated by 2.016 Da, corresponding to the mass difference of two hy drogen 463 losses17,42. Under single -pass cIMS conditions, each additional double bond induced a 464 reproducible drift -time shift of approximately 0.13 ms in this fatty acid series . In this 465 framework, the core molecular structure defines the absolute drift -time range, while 466 incremental changes in unsaturation induce reproducible relative shifts in drift time within a 467 given dataset 11,43. Together, these relationships enable biologically meaningful grouping of 468 related lipid features directly from intact LA -REIMS-cIMS fingerprints, providing structurally 469 information. 470 471 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 14 472 Fig 4. Mobility-resolved organization of lipid homologues and fatty acid saturation forms in 473 LA-REIMS-cIMS biofluid spectra. (A) Pooled urine spectrum showing lipid homologues which 474 form a distinct mobility band separated from salt -derived cluster ions. (B) Pooled f ecal 475 spectrum showing an analogous mobility -resolved pattern for free fatty acids, including 476 linoleic acid (C18:2; m/z 279.23), oleic acid (C18:1; m/z 281.25), and stearic acid (C18:0; m/z 477 283.26), along with their 13C isotopic peaks and corresponding mass spectrum . (C) Arrival 478 times of the bile acid isomers deoxycholic acid (DCA), ursodeoxycholic acid (UDCA), and 479 chenodeoxycholic acid (CDCA) plotted as a function of effective cyclic IMS separation time (2-480 32 ms). 481 482 Having demonstrated that ion mobility separation adds structurally meaningful organization 483 to untargeted LA-REIMS fingerprints, we next evaluated the extent to which such separation 484 can be leveraged for explicit isomer discrimination under practical acquisition constraints . In 485 the subsequent targeted experiments, analytical standard solutions of secondary bile acids, 486 their glycine- and taurine-conjugated forms, and phosphocholine (PC) lipids were examined 487 as model compounds to assess the performance of the established LA-REIMS-cIMS workflow 488 to generate isomer-resolved chemical signatures under both short and extended separation 489 conditions. The selected analytes comprise structurally similar isomers, are abundant in fecal 490 samples, and are biologically relevant 44,45, making them suitable benchmarks for assessing 491 cIMS performance in the context of biofluid analysis. 492 493 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 15 For bile acids, multiple ion forms were detected, including deprotonated monomers, 494 hydrogen-bonded dimers, sodium-bound dimers, and, when their m/z values fell within the 495 acquisition mass range , hydrogen-bonded trimers (Table 1), consistent with the structural 496 complexity and adduct diversity observed in bile acid research46. While individual standards 497 exhibited distinct arrival times, this separation did not translate to isomer resolution in 498 mixtures under single -cycle IMS conditions. Arrival -time differences between deprotonated 499 monomers were small (<5 ms) and insufficient to resolve isomeric bile acids based on their 500 [M-H]⁻ ions, irrespective of conjugation state. In contrast, partial isomer discrimination was 501 observed for hydrogen -bonded dimers. Under single -cycle IMS conditions, the 502 chenodeoxycholic acid (CDCA)-dominated hydrogen-bonded dimer exhibited higher mobility 503 than other bile acid dimers, with an arrival time difference of 5.08 ms relative to nearest co-504 migrating dimer , enabling its resolution in mixtures. This effect diminished upon glycine 505 conjugation and was absent for taurine -conjugated bile acids, which collapsed into a single 506 unresolved mobility feature , reflecting the dependence of IMS separation on the ion 507 conformation and adduct chemistry47. Extending the cIMS separation time enabled improved 508 bile acid isomer resolution. In mixtures of deoxycholic acid (DCA), ursodeoxycholic acid 509 (UDCA), and CDCA, deprotonated CDCA was resolved from the other isomers at a 12 ms 510 separation cycle, while all three bile acids produced distinct mobility -resolved signals at 32 511 ms (as their [M -H]⁻ forms ; Fig 4C ). Although cIMS provides sufficient tunability to resolve 512 closely related bile acid isomers, optimal separation performance is compound - and class-513 dependent and requires targeted optimization48,49. 514 515 Table 1. Bile acid analytical standards and their detected forms with arrival times following 516 LA-REIMS-cIMS analysis. 517 518 Compound Chemical Formula Arrival time (ms) [M-H]⁻ [2M-H]⁻ [3M-H]⁻ [2M+Na-2H]⁻ Deoxycholic acid C24H40O4 31.31 70.47 22.01 ND Ursodeoxycholic acid C24H40O4 32.66 71.24 22.56 ND Chenodeoxycholic acid C24H40O4 32.92 65.39 21.65 ND Glycoursodeoxycholic acid C26H43NO5 31.18 82.21 OOR 83.52 Glycodeoxycholic acid C26H42NO5 31.26 83.52 OOR 84.11 Glycochenodeoxycholic acid C26H42NO5 31.07 79.86 OOR 83.52 Taurochenodeoxycholic acid C26H44NO6S 33.88 89.69 OOR 91.40 Taurodeoxycholic acid C26H44NO6S 34.19 88.34 OOR 94.53 Tauroursodeoxycholic C26H45NO6S 34.14 88.30 OOR 94.55 ND: not detected; OOR: outside of tested m/z range (50 – 1200 Da) 519 520 In contrast, fully saturated positional phosphocholine isomers with mirrored acyl -chain 521 distributions (PC 16:0/18:0 and PC 18:0/16:0) remained unresolved under all tested 522 conditions. No separation was observed even at extended cIMS separation cycles of 100 and 523 200 ms (arrival times of 89.49 and 89.53 ms for the respective chlorine adducts). The minimal 524 0.04 ms difference is consistent with the negligible collision cross section variation expected 525 for exchange of two linear, fully saturated acyl chains, which introduces only marginal 526 perturbation of the overall ion conformation50. Notably, these species are also challenging to 527 resolve chromatographically in routine workflows, as reversed-phase retention is dominated 528 by bulk hydrophobicity (total carbon number and unsaturation) and HILIC retention by 529 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 16 headgroup polarity rather than acyl -chain position51,52. Consequently, robust differentiation 530 of structurally subtle positional isomers often requires complementary structural information 531 from MS/MS -based strategies (e.g., ozone dissociation or related reaction/fragmentation 532 methods), which can be incorporated alongside both LC-MS and IMS-MS measurements53,54. 533 534 5. Analytical scope, applications, and limitations of LA -REIMS-cIMS in ambient biofluid 535 metabolomics 536 537 cIMS introduces multidimensionality into the chromatography-free AIMS data and can 538 substantially improve the specificity of ambient biofluid fingerprints by providing an 539 orthogonal separation dimension. This is particularly valuable under conditions of high 540 chemical noise or extensive matrix interferences, and when structural resolution , such as 541 isomer discrimination or CCS -based annotation , is central to the analytical objective. In 542 addition, CCS -based annotation is generally more reproducible across platforms than 543 chromatographic retention time matching, making it more amenable to 544 standardization13,26,41. 545 546 However, for high -throughput metabolic fingerprinting workflows where analytical 547 performance is driven primarily by global spectral patterns rather than feature -level 548 annotation, MS-only acquisition is often sufficient and can offer practical advantages. In such 549 scenarios, the added dimensionality provided by ion mobility may not always translate into 550 proportional gains in downstream interpretability, while increased acquisition speed and 551 sensitivity may be achieved by maximizing duty cycle. When matrix -derived artefacts are 552 limited, or when biological variability dominates spectral differences, omitting IMS can 553 therefore represent a pragmatic and effective strategy. In this context, ambient ionization 554 platforms capable of operating in either IMS-enabled or MS-only mode provide flexibility to 555 address complementary analytical needs, with the optimal acquisition strategy determined 556 by matrix complexity and study objectives. 557 558 6. Conclusion 559 560 This study establishes practical conditions under which cyclic ion mobility spectrometry can 561 be integrated with LA -REIMS for rapid, chromatography -free biofluid metabolomics despite 562 the transient ion generation inherent to ambient ionization workflows. By implementing 563 duty-cycle-aware optimization, single-pass cIMS operation was achieved that preserved a 564 substantial fraction of the total ion current while remaining compatible with the temporal 565 constraints of LA-REIMS acquisition. While the magnitude of signal recovery and separation 566 performance depends on instrument architecture, ion source design and sample acquisition 567 method, the operational principles described here are broadly applicable to transient 568 ambient ionization workflows. 569 570 Furthermore, ion mobility separation reorganizes complex LA-REIMS spectra into structured 571 mass-mobility feature domains, enabling selective suppression of matrix -derived salt cluster 572 ions and revealing chemically coherent signal populations, particularly within lipid classes. 573 The combination of accurate mass and reproducible collision cross -section measurements 574 supported class -level metabolite annotation and enabled conditional discrimination of 575 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 17 selected isomeric species, illustrating both the analytical capabilities and intrinsic limitations 576 of cIMS under short separation times. 577 578 Collectively, these results define the analytical operating space of LA -REIMS-cIMS and clarify 579 the trade-offs between sensitivity, throughput, and chemical specificity associated with ion 580 mobility integration. The framework presented here provides guidance for informed selection 581 of IMS-enabled versus MS-only acquisition strategies in ambient metabolomics applications 582 where chromatographic separation is impractical and enhanced chemical resolution is 583 required. 584 585 Acknowledgments 586 587 This work was funded in part by the European Union (ERC project MeMoSA, 2023 -CoG, 588 101124151). Views and opinions expressed are however those of the author(s) only and do 589 not necessarily reflect those of the European Union or the European Research Council. 590 Neither the European Union nor the granting authority can be held responsible for them. The 591 authors acknowledge Waters Corporation for assistance with the LA-REIMS-cIMS integration 592 and technical support. 593 594 Conflict of interest 595 596 SELECT SERIES™ Cyclic™ traveling-wave ion mobility time -of-flight mass spectrometer was 597 purchased under scientific collaboration agreement with the Waters Corporation ; the 598 company had no involvement in the conduct or reporting of the study. 599 600 Author contributions 601 602 V.P.: Conceptualization, methodology, investigation, data curation, formal analysis, 603 visualization, and writin g - original draft ; N.V.d.V.: Investigation, data curation, formal 604 analysis, visualization, interpretation , reviewing and writing ; A.V.B.: Investigation, data 605 curation, formal analysis, and interpretation ; J.D.D.M.: Conceptualization, reviewing and 606 writing. L.V.: Conceptualization, funding acquisition, supervision, project administration , 607 reviewing and writing. All authors contributed to manuscript revision and approved the final 608 version. 609 610 Data and code availability 611 612 Processed data, representative spectra, and additional figures supporting the findings of this 613 study are provided in the Supporting Information. Raw mass spectrometry data will be made 614 publicly available via MetaboLights repository MTBLS2605 upon publication . The R Shiny 615 application code for spectra curation is available via https://github.com/UGent-616 LIMET/MS_spectra_curation. 617 618 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 18

619 620 1. Feider, C. L., Krieger, A., Dehoog, R. J. & Eberlin, L. S. Ambient Ionization Mass 621 Spectrometry: Recent Developments and Applications. Anal. Chem.; 622 https://doi.org/10.1021/acs.analchem.9b00807 (2019). 623 2. Plekhova, V., De Windt, K., De Spiegeleer, M., De Graeve, M. & Vanhaecke, L. Recent 624 advances in high-throughput biofluid metabotyping by direct infusion and ambient 625 ionization mass spectrometry. TrAC Trends in Analytical Chemistry 168, 117287 626 (2023). 627 3. Zhou, X., Zhang, W. & Ouyang, Z. Recent advances in on-site mass spectrometry 628 analysis for clinical applications. TrAC - Trends in Analytical Chemistry vol. 149 629 https://doi.org/10.1016/j.trac.2022.116548 (2022). 630 4. Yang, S.-S., Shi, M.-Y., Tao, Z.-R., Wang, C. & Gu, Z.-Y. Recent applications of metal–631 organic frameworks in matrix-assisted laser desorption/ionization mass 632 spectrometry. Anal. Bioanal. Chem. 411, 4509–4522 (2019). 633 5. De Spiegeleer, M. et al. Point-of-care applicable metabotyping using biofluid-specific 634 electrospun MetaSAMPs directly amenable to ambient LA-REIMS. Sci. Adv. 9, 635 eade9933 (2023). 636 6. Palmnäs, M. et al. Perspective: Metabotyping—A Potential Personalized Nutrition 637 Strategy for Precision Prevention of Cardiometabolic Disease. Advances in Nutrition 638 11, 524–532 (2020). 639 7. Plekhova, V. et al. Rapid ex vivo molecular fingerprinting of biofluids using laser-640 assisted rapid evaporative ionization mass spectrometry. Nat. Protoc. 641 https://doi.org/10.1038/s41596-021-00580-8 (2021) 642 8. Jones, E. A. et al. Matrix assisted rapid evaporative ionization mass spectrometry. 643 Anal. Chem. https://doi.org/10.1021/acs.analchem.9b01441 (2019). 644 9. Schrimpe-Rutledge, A. C., Codreanu, S. G., Sherrod, S. D. & McLean, J. A. Untargeted 645 Metabolomics Strategies—Challenges and Emerging Directions. J. Am. Soc. Mass 646 Spectrom. 27, 1897–1905 (2016). 647 10. Kuo, T. H., Dutkiewicz, E. P., Pei, J. & Hsu, C. C. Ambient Ionization Mass Spectrometry 648 Today and Tomorrow: Embracing Challenges and Opportunities. Anal. Chem. 92, 649 2353–2363 (2020). 650 11. Burnum-Johnson, K. E. et al. Ion mobility spectrometry and the omics: Distinguishing 651 isomers, molecular classes and contaminant ions in complex samples. TrAC Trends in 652 Analytical Chemistry 116, 292–299 (2019). 653 12. Wen, S.-S., Zhou, H.-S., Zhu, C.-S., Li, P. & Gao, W. Direct infusion electrospray 654 ionization-ion mobility-mass spectrometry for rapid metabolite marker discovery of 655 medicinal Phellodendron Bark. J. Pharm. Biomed. Anal. 219, 114939 (2022). 656 13. Paglia, G. et al. Ion Mobility Derived Collision Cross Sections to Support Metabolomics 657 Applications. Anal. Chem. 86, 3985–3993 (2014). 658 14. Spangler, G. E. Expanded theory for the resolving power of a linear ion mobility 659 spectrometer. Int. J. Mass Spectrom. 220, 399–418 (2002). 660 15. Giles, K. et al. A Cyclic Ion Mobility-Mass Spectrometry System. Anal. Chem. 91, 8564–661 8573 (2019). 662 16. Richardson, K. & Giles, K. Travelling Wave Ion Mobility Separation: Basics and 663 Calibration. in Ion Mobility – Mass Spectrometry: Fundamentals and Applications 664 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 19 (eds. Ashcroft, A. E. & Sobott, F.) 0 (The Royal Society of Chemistry, 2021). 665 doi:10.1039/9781839162886-00083. 666 17. Paglia, G., Smith, A. J. & Astarita, G. Ion mobility mass spectrometry in the omics era: 667 Challenges and opportunities for metabolomics and lipidomics. Mass Spectrom. Rev. 668 41, 722–765 (2022). 669 18. Vangeenderhuysen, P. et al. Dual UHPLC-HRMS Metabolomics and Lipidomics and 670 Automated Data Processing Workflow for Comprehensive High-Throughput Gut 671 Phenotyping. Anal. Chem. 95, 8461–8468 (2023). 672 19. De Paepe, E. et al. A repository of the salivary metabolome and its key drivers in 1436 673 European children. EBioMedicine 122, 106019 (2025). 674 20. Cole, T. J. & Lobstein, T. Extended international (IOTF) body mass index cut-offs for 675 thinness, overweight and obesity. Pediatr. Obes. https://doi.org/10.1111/j.2047-676 6310.2012.00064.x (2012) 677 21. De Spiegeleer, M. et al. Impact of storage conditions on the human stool metabolome 678 and lipidome: Preserving the most accurate fingerprint. Anal. Chim. Acta 679 https://doi.org/10.1016/j.aca.2020.02.046 (2020) 680 22. Wishart, D. S. et al. HMDB 5.0: the Human Metabolome Database for 2022. Nucleic 681 Acids Res. 50, D622–D631 (2022). 682 23. Takeda, H. et al. MS-DIAL 5 multimodal mass spectrometry data mining unveils 683 lipidome complexities. Nat. Commun. 15, 9903 (2024). 684 24. Lee, R. Statistical Design of Experiments for Screening and Optimization. Chemie 685 Ingenieur Technik 91, 191–200 (2019). 686 25. Adusumilli, R. & Mallick, P. Data Conversion with ProteoWizard msConvert. in 687 Proteomics: Methods and Protocols (eds. Comai, L., Katz, J. E. & Mallick, P.) 339–368 688 (Springer New York, New York, NY, 2017). 689 26. May, J. C., Morris, C. B. & McLean, J. A. Ion Mobility Collision Cross Section 690 Compendium. Anal. Chem. 89, 1032–1044 (2017). 691 27. Dodds, J. N. & Baker, E. S. Ion Mobility Spectrometry: Fundamental Concepts, 692 Instrumentation, Applications, and the Road Ahead. J. Am. Soc. Mass Spectrom. 30, 693 2185–2195 (2019). 694 28. Naylor, C. N. & Nagy, G. Recent advances in high-resolution traveling wave-based ion 695 mobility separations coupled to mass spectrometry. Mass Spectrom. Rev. 44, 581–696 598 (2025). 697 29. Breen, J., Hashemihedeshi, M., Amiri, R., Dorman, F. L. & Jobst, K. J. Unwrapping 698 Wrap-around in Gas (or Liquid) Chromatographic Cyclic Ion Mobility–Mass 699 Spectrometry. Anal. Chem. 94, 11113–11117 (2022). 700 30. Williamson, D. L. & Nagy, G. Evaluating the Utility of Temporal Compression in High-701 Resolution Traveling Wave-Based Cyclic Ion Mobility Separations. ACS Measurement 702 Science Au 2, 361–369 (2022). 703 31. Shvartsburg, A. A. & Smith, R. D. Fundamentals of Traveling Wave Ion Mobility 704 Spectrometry. Anal. Chem. 80, 9689–9699 (2008). 705 32. Luo, M.-D., Zhou, Z.-W. & Zhu, Z.-J. The Application of Ion Mobility-Mass 706 Spectrometry in Untargeted Metabolomics: from Separation to Identification. J. Anal. 707 Test. 4, 163–174 (2020). 708 33. May, J. C. et al. Conformational Ordering of Biomolecules in the Gas Phase: Nitrogen 709 Collision Cross Sections Measured on a Prototype High Resolution Drift Tube Ion 710 Mobility-Mass Spectrometer. Anal. Chem. 86, 2107–2116 (2014). 711 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 20 34. Konermann, L., McAllister, R. G. & Metwally, H. Molecular Dynamics Simulations of 712 the Electrospray Process: Formation of NaCl Clusters via the Charged Residue 713 Mechanism. J. Phys. Chem. B 118, 12025–12033 (2014). 714 35. Zhang, J. et al. Simultaneously quantifying hundreds of acylcarnitines in multiple 715 biological matrices within ten minutes using ultrahigh-performance liquid-716 chromatography and tandem mass spectrometry. J. Pharm. Anal. 14, 140–148 (2024). 717 36. Aron, A. T. et al. Native mass spectrometry-based metabolomics identifies metal-718 binding compounds. Nat. Chem. 14, 100–109 (2022). 719 37. Zhu, J. & Cole, R. B. Formation and decompositions of chloride adduct ions, [M + Cl]−, 720 in negative ion electrospray ionization mass spectrometry. J. Am. Soc. Mass 721 Spectrom. 11, 932–941 (2000). 722 38. Hamid, A. M. et al. Characterization of Traveling Wave Ion Mobility Separations in 723 Structures for Lossless Ion Manipulations. Anal. Chem. 87, 11301–11308 (2015). 724 39. May, J. C. & McLean, J. A. Integrating ion mobility into comprehensive 725 multidimensional metabolomics workflows: critical considerations. Metabolomics 18, 726 104 (2022). 727 40. Schymanski, E. L. et al. Identifying Small Molecules via High Resolution Mass 728 Spectrometry: Communicating Confidence. Environ. Sci. Technol. 48, 2097–2098 729 (2014). 730 41. Zhou, Z. et al. Ion mobility collision cross-section atlas for known and unknown 731 metabolite annotation in untargeted metabolomics. Nat. Commun. 11, 4334 (2020). 732 42. Leaptrot, K. L., May, J. C., Dodds, J. N. & McLean, J. A. Ion mobility conformational 733 lipid atlas for high confidence lipidomics. Nat. Commun. 10, 985 (2019). 734 43. Kyle, J. E. et al. Uncovering biologically significant lipid isomers with liquid 735 chromatography, ion mobility spectrometry and mass spectrometry. Analyst 141, 736 1649–1659 (2016). 737 44. Brown, E. M., Clardy, J. & Xavier, R. J. Gut microbiome lipid metabolism and its impact 738 on host physiology. Cell Host Microbe 31, 173–186 (2023). 739 45. Peng, Y.-L. et al. Effects of bile acids on the growth, composition and metabolism of 740 gut bacteria. NPJ Biofilms Microbiomes 10, 112 (2024). 741 46. Hadavi, D., Borzova, M., Porta Siegel, T. & Honing, M. Uncovering the behaviour of 742 ions in the gas-phase to predict the ion mobility separation of isomeric steroid 743 compounds. Anal. Chim. Acta 1200, 339617 (2022). 744 47. Feng, H. & Yamaguchi, T. Effects of Different Adduct Ions, Ionization Temperatures, 745 and Solvents on the Ion Mobility of Glycans. Molecules 30, (2025). 746 48. de Bruin, C. R., Hennebelle, M., Vincken, J.-P. & de Bruijn, W. J. C. Separation of 747 flavonoid isomers by cyclic ion mobility mass spectrometry. Anal. Chim. Acta 1244, 748 340774 (2023). 749 49. Nytka, M., Wan, J., Tureček, F. & Lemr, K. Cyclic Ion Mobility of Isomeric New 750 Psychoactive Substances Employing Characteristic Arrival Time Distribution Profiles 751 and Adduct Separation. J. Am. Soc. Mass Spectrom. 35, 1733–1742 (2024). 752 50. Fabritius, M. & Yang, B. Analysis of triacylglycerol and phospholipid sn-positional 753 isomers by liquid chromatographic and mass spectrometric methodologies. Mass 754 Spectrom. Rev. 45, 4–36 (2026). 755 51. Zhang, Z. et al. Development of a targeted hydrophilic interaction liquid 756 chromatography-tandem mass spectrometry based lipidomics platform applied to a 757 coronavirus disease severity study. J. Chromatogr. A 1708, 464342 (2023). 758 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint 21 52. Cajka, T. & Fiehn, O. Comprehensive analysis of lipids in biological systems by liquid 759 chromatography-mass spectrometry. TrAC Trends in Analytical Chemistry 61, 192–760 206 (2014). 761 53. Xia, T. et al. Deep-profiling of phospholipidome via rapid orthogonal separations and 762 isomer-resolved mass spectrometry. Nat. Commun. 14, 4263 (2023). 763 54. Michael, J. A. et al. Deep Characterisation of the sn-Isomer Lipidome Using High-764 Throughput Data-Independent Acquisition and Ozone-Induced Dissociation. 765 Angewandte Chemie International Edition 63, e202316793 (2024). 766 767 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 12, 2026. ; https://doi.org/10.64898/2026.03.10.709786doi: bioRxiv preprint