Hydrophilic Interaction Liquid Chromatography–Hydrogen/Deuterium Exchange–Mass Spectrometry (HILIC-HDX-MS) for Untargeted Metabolomics

Liquid chromatography with mass spectrometry (LC-MS)-based metabolomics detects thousands of molecular features (retention time–m/z pairs) in biological samples per analysis, yet the metabolite annotation rate remains low, with 90% of signals classified as unknowns. To enhance the metabolite annotation rates, researchers employ tandem mass spectral libraries and challenging in silico fragmentation software. Hydrogen/deuterium exchange mass spectrometry (HDX-MS) may offer an additional layer of structural information in untargeted metabolomics, especially for identifying specific unidentified metabolites that are revealed to be statistically significant. Here, we investigate the potential of hydrophilic interaction liquid chromatography (HILIC)-HDX-MS in untargeted metabolomics. Specifically, we evaluate the effectiveness of two approaches using hypothetical targets: the post-column addition of deuterium oxide (D2O) and the on-column HILIC-HDX-MS method. To illustrate the practical application of HILIC-HDX-MS, we apply this methodology using the in silico fragmentation software MS-FINDER to an unknown compound detected in various biological samples, including plasma, serum, tissues, and feces during HILIC-MS profiling, subsequently identified as N1-acetylspermidine.

LC-MS-based metabolomics can generate thousands of molecular features/ion peaks (i.e., retention time-m/z pairs) in biological samples per single analysis.However, it is estimated that only 10% of these features can be annotated according to mass spectra library matches [10,11], leaving the remaining 90% as unknowns.Thus, metabolomics faces a low identification rate [12] despite the overcounting of features that either do not trigger the accompanying MS/MS spectra or need to be accounted for differently due to adducts [13,14].Untargeted metabolomics studies often provide only sets of annotated metabolites, unknowns are not further investigated because additional resources are needed, and the structure elucidation process is time-consuming.However, studies focused, for instance, on circadian rhythms [12,15], cardiovascular diseases [16], type 2 diabetes [17], lung cancer [18], or nutrition [19] have shown that these unknowns can be statistically significant, and focus on their structure elucidation may bring valuable insights into metabolic pathways, potentially uncovering novel biomarkers or therapeutic targets associated with these specific physiological and pathological conditions [20].
The metabolite annotation rate can be increased by collecting information from the MS analysis of certified standards and making these data publicly accessible in mass spectral repositories [21,22].However, acquiring the MS/MS spectra of all authentic metabolites in every laboratory is unrealistic.As a more viable solution, tandem mass spectral libraries were developed to annotate the MS/MS spectra obtained during metabolomics experiments.Despite the availability of various commercial and public MS/MS libraries such as NIST 23, METLIN Gen2, MassBank, and MoNA [21,23,24], many false positive annotations may be reported [9,25,26].
Employing computational simulations to predict the mass spectra from input structures can further increase the annotation rate [27].Various existing in silico fragmentation software programs, including Mass Frontier, CSI:FingerID, CFM-ID, MS-FINDER, MIDAS-G, and MetFrag [27][28][29][30][31][32], are utilized for identifying unknown metabolites and determining their chemical structures.Typically, these tools convert mass data into molecular fragments using combinatorial structure generation techniques with limited success.Hence, it is crucial to integrate additional orthogonal information to determine the correct structure.Such methods may be used to annotate unknown compounds while also reducing false positive annotations originating from MS/MS search and in silico fragmentation software.Hydrogen/deuterium exchange mass spectrometry (HDX-MS), a well-established technique in proteomics and for the structural elucidation of pharmaceutical impurities and drug metabolites, though not fully explored in metabolomics, provides a promising approach [33][34][35][36][37][38][39][40][41][42][43].
Exchangeable hydrogen atoms (also called active, acidic, or labile), bound to heteroatoms such as oxygen, nitrogen, and sulfur, readily exchange with deuterium, while those bound to carbon remain unaltered [44].When metabolites are exposed to deuterium oxide (D 2 O), labile hydrogens within various functional groups such as −NH−, −NH 2 , −OH, −COOH, and −SH are substituted with deuterium.Using a mass spectrometer, the number of replaced hydrogens in the molecule can be determined by measuring the molecular mass before and after HDX.Such information can be a valuable filter, narrowing down potential candidates for structure elucidation [45,46].
Here, we evaluate the potential of LC-HDX-MS in untargeted metabolomics.Specifically, we compare the performance of two methods: the post-column addition of D 2 O and the on-column HILIC-HDX-MS method.In the latter case, we also evaluate variations when only water is replaced with D 2 O and non-deuterated mobile-phase modifiers are used (partial HILIC-HDX-MS) or both D 2 O and deuterated mobile-phase modifiers are used (full HILIC-HDX-MS).We exemplify this approach by identifying an unknown metabolite detected in various biological samples (biofluids, tissues, feces) of rats, mice, and humans.
Untargeted metabolomics and lipidomics studies typically report hundreds of annotated metabolites, usually after combining data from various LC-MS platforms.However, many more metabolites remain unidentified.Elemental formulas can be correctly assigned in 95% of all unknowns [9], but annotating these compounds is challenging.Chemical transformations such as HDX may reveal substructure information and be used to discard false positive isomer structures from chemical database queries to differentiate lists of possible isomers.
Considering the simplicity of operation, we initially focused on evaluating the performance of the standard HILIC-MS profiling method, modified only with the post-column addition of D 2 O.In this case, the previously injected extracts in an acetonitrile/water (4:1, v/v) mixture were subsequently analyzed using the same HILIC-MS method but with a post-column addition of D 2 O introduced using an infusion pump (Figure 1, Table 1).The advantage of this approach is that no further modification of the mobile phases and resuspended extracts was needed.
Considering the simplicity of operation, we initially focused on evaluating the performance of the standard HILIC-MS profiling method, modified only with the postcolumn addition of D2O.In this case, the previously injected extracts in an acetonitrile/water (4:1, v/v) mixture were subsequently analyzed using the same HILIC-MS method but with a post-column addition of D2O introduced using an infusion pump (Figure 1, Table 1).The advantage of this approach is that no further modification of the mobile phases and resuspended extracts was needed.
Next, we turned our attention to on-column HILIC-HDX-MS methods with modified mobile phases and the resuspension of dry extracts.For partial HILIC-HDX-MS, only water was replaced by D2O, and non-deuterated mobile-phase modifiers (ammonium formate and formic acid) were used.The dry extracts were resuspended in an acetonitrile/D2O (4:1, v/v) mixture.The advantage of this approach is that only the key H/D exchange solvent (D2O) was needed.On the other hand, for full HILIC-HDX-MS, both D2O and deuterated mobile-phase modifiers (D5-ammonium formate and D2-formic acid) were employed, and the dry extracts were again resuspended in an acetonitrile/D2O (4:1, v/v) mixture (Figure 1, Table 1).All the H/D exchange components were replaced in this last tested method, thus requiring more individual D-labeled chemicals during the run.1).All the H/D exchange components were replaced in this last tested method, thus requiring more individual D-labeled chemicals during the run.

HILIC-MS with the Post-Column Addition of D 2 O
For the first setup, we used the post-column addition of D 2 O introduced via a Tee connector into the LC effluent using an infusion pump.We first evaluated this approach for a range of polar metabolites with varying numbers of exchangeable hydrogen atoms.For these hypothetical targets, we assessed the extent of H/D exchange.As Figure 2a shows, the post-column co-infusion of D 2 O at a flow rate of 50 µL/min against a HILIC column flow of 400 µL/min led to the detection of the molecules as [M+D] + ions as base peaks (100%), and only limited H/D exchange occurred.A slight improvement was observed with the increased addition of D 2 O (100 µL/min), leading to the detection of base peaks for [M(D 1 )+D] + ions with one H/D exchange (Figure 2b).We also included in Figure 2 an internal standard, 12-[[(cyclohexylamino)carbonyl]amino]-dodecanoic acid (CUDA), eluting at 0.45 min, in which case the co-infusion of D 2 O led to complete H/D exchange ([M(D 3 )+D] + ).This can be explained by the low content of H 2 O (Figure 3a) in the mobile phase with co-infused D 2 O (~83% D 2 O, Figure 3b).
However, most polar metabolites were eluted during HILIC-MS at the composition of the mobile phases, favoring a high content of H 2 O (3-5 min); thus, the content of coinfused D 2 O (100 µL/min) is reduced to 25-40% compared to H 2 O from the mobile phases (Figure 3b).This shortcoming led to the incomplete H/D exchange mass spectra of the evaluated polar metabolites, which would be difficult to interpret if these were unknowns.Shah et al. [37] showed with the example of drug metabolites that the post-column addition of D 2 O was efficient when the LC output was 50 µL/min and a D 2 O infusion rate of 126 µL/min (~1:2.5 ratio) was used.However, splitting the LC flow was required, possibly leading to a signal reduction and requiring further modification of the LC-MS setup.Of note, resuspending the samples in an acetonitrile/D 2 O (4:1, v/v) mixture with follow-up analysis did not improve the H/D exchange.Thus, the post-column addition of D 2 O appeared to provide limited potential for H/D exchange, as also highlighted by Liu et al. [33] during pharmaceutical compound analysis.

Partial and Full HILIC-HDX-MS
In the subsequent two setups of on-column HILIC-HDX-MS, we replaced H 2 O with D 2 O as the component of both mobile phases but kept unlabeled mobile-phase modifiers (ammonium formate, formic acid), i.e., partial HILIC-HDX-MS, or also replaced the mobilephase modifiers with deuterated counterparts (D 5 -ammonium formate, D 2 -formic acid), i.e., full HILIC-HDX-MS.
However, for mobile phase B containing acetonitrile and D 2 O (95:5, v/v), we observed solubility issues with both 10 mM ammonium formate and 10 mM D 5 -ammonium formate compared to the acetonitrile/H 2 O (95:5, v/v) mixture with 10 mM ammonium formate.Based on our experience, the light turbidity disappears after sonication when preparing fully non-labeled mobile phase B. However, this was not the case when D 2 O was used instead of H 2 O.For complete dissolving of the salts, sonication at 25 • C for 10 mM ammonium formate or even 30 • C for the D 5 -ammonium formate solutions was needed.
Unfortunately, after cooling both solutions to lab temperature (23 • C), we observed crystals of the salts, although the solutions were not turbid.We found that 7.5 mM of ammonium formate or D 5 -ammonium formate was a limiting concentration when D 2 O was used as a part of mobile phase B. However, this modification of mobile phase B did not impact the metabolites' retention time shift or peak shape.
However, for mobile phase B containing acetonitrile and D2O (95:5, v/v), we observed solubility issues with both 10 mM ammonium formate and 10 mM D5-ammonium formate compared to the acetonitrile/H2O (95:5, v/v) mixture with 10 mM ammonium formate.Based on our experience, the light turbidity disappears after sonication when preparing fully non-labeled mobile phase B. However, this was not the case when D2O was used instead of H2O.For complete dissolving of the salts, sonication at 25 °C for 10 mM ammonium formate or even 30 °C for the D5-ammonium formate solutions was needed.
Unfortunately, after cooling both solutions to lab temperature (23 °C), we observed crystals of the salts, although the solutions were not turbid.We found that 7.5 mM of ammonium formate or D5-ammonium formate was a limiting concentration when D2O was used as a part of mobile phase B. However, this modification of mobile phase B did not impact the metabolites' retention time shift or peak shape.We should also note that the H/D exchange efficiency has been reported to be pHdependent [53].For example, amides demonstrate a minimum exchange rate around a pH of 2.5 [54], while carbohydrates, primarily composed of hydroxyls, exhibit a minimum exchange rate around a pH of 6.5, with the exchange rate increasing as the solution becomes more acidic or basic [55].The pH of the mobile phases used in the HILIC method was approximately 3, aligning with the observed pH-dependent trends in amides and carbohydrates.However, as metabolomic profiling encompasses diverse groups of metabolites, the H/D exchange efficiency can vary based on the pH of the mobile phase.Figure 4 shows an example of the amino acid lysine with five labile hydrogen atoms acquired using conventional HILIC-MS with detected [M+H] + ions (Figure 4a), followed by incomplete H/D exchange when using the post-column addition of D 2 O, providing ions from [M+H] + to [M(D 5 )+D] + (Figure 4b), and complete H/D exchange when the full HILIC-HDX-MS method was used, with dominating [M(D 5 )+D] + ions (Figure 4c).All these data show that on-column H/D exchange is advantageous since it generally produces a high yield of fully deuterated compounds owing to the adequate mixing time of the metabolites with the deuterated mobile phases.
utilizing the post-column addition of D2O (100 μL/min), whereas approximately 600 of mobile phase A with D2O and 2.8 mL of mobile phase B (including ~140 μL D2O) w required for the partial or full HILIC-HDX-MS setups.Thus, both approaches involv comparable volumes of D2O.However, the partial or full HILIC-HDX-MS setu provided more efficient deuterium incorporation into the analytes, making them preferred choice for achieving complete H/D exchange in untargeted metabolom studies.

Structure Elucidation of N 1 -Acetylspermidine Using HILIC-HDX-MS
When processing raw HILIC-MS files from multiple studies focused on nutritional intervention [56], circadian rhythms [57], drug treatment [58], or heart failure [59], we have repeatedly observed an unknown metabolite eluted at a retention time of 4.3 min and m/z 188.1757 in various biological matrices (biofluids, tissues, feces) of rats, mice, and humans (Figure S1).
Since we did not obtain a positive spectral match when using the combined NIST 23 and MoNA MS/MS libraries, we submitted MS1 isotopic ions and the MS/MS spectrum from MS-DIAL to the MS-FINDER software [27] for structure elucidation.Only one formula (C 9 H 21 N 3 O), within a mass error of <0.005 Da (set up as a criterion for mass tolerance), was reported.
This formula provided the source to 23 local databases in MS-FINDER with 101 possible unique structures.Focusing on such a high number of potential candidates would be impractical; thus, we applied an additional filter from the HILIC-HDX-MS experiment.Specifically, we obtained information based on the isotopic ions from the conventional HILIC-MS used for polar metabolite profiling and full HILIC-HDX-MS that four labile hydrogens were present in the molecule.Using this information, we reduced the number of potential candidates to 22; thus, 78% of false positive structures were filtered out with this additional filter.
Based on this information and using the two most scored candidates, we analyzed the standards of N 1 -acetylspermidine and N 8 -acetylspermidine using HILIC-MS and HILIC-HDX-MS.Figure 5 shows that the N 1 -acetylspermidine detected in mouse feces matched the standard, including the retention time and MS1 and MS/MS spectra from HILIC-MS and HILIC-HDX-MS after H/D exchange.A similar observation was also confirmed for rat feces (Figure S2) and human plasma NIST SRM 1950 (Figure S3). Figure 6 also shows the MS/MS fragmentation of N 1 -acetylspermidine under HILIC-MS and full HILIC-HDX-MS conditions, further confirming the identity of this unknown metabolite based on H/D exchange in a series of MS/MS fragments.Of note, the standard of N 8 -acetylspermidine provided a longer retention time (0.05 min shift) than the peak of N 1 -acetylspermidine detected in mouse feces, and the MS/MS spectrum did not match (Figure S4).The MS/MS spectra of N 1 -acetylspermidine and N 8 -acetylspermidine can be downloaded in a mass searchable format (MSP) for storing MS/MS spectra (m/z and intensity of mass peaks) from the Supplementary Materials.In addition, N 1 -acetylspermidine appeared statistically significant during our experiments investigating the impact of diet and antibiotic treatment on mice.As Figure 7 shows, there was no statistical difference between the groups of mouse feces under normal (chow) and high-fat diets.On the other hand, after antibiotic treatment, over one order of magnitude difference in signal intensity was observed between these two groups.Interestingly, in mice on a chow diet, the level of N 1 -acetylspermidine decreased 3.8-fold after antibiotics treatment, while on a high-fat diet, the opposite trend was observed (4.1-fold increase), indicating differences in the polyamine pathway of the microbiota.
The mouse and rat feces samples were from the Institute of Physiology of the Czech Academy of Sciences, Prague, Czech Republic.The human serum (catalog no.S7023-100ML), NIST SRM 1950 plasma (catalog no.NIST1950), and a mixture of 17 amino acids (catalog no.79248) were from Merck.

Experiments with Animals
The age-matched 6-week-old male C57BL/6J mice were from Charles River Laboratories (Sulzfeld, Germany).After their arrival, the mice were individually housed in cages and maintained at 22 • C and according to a 12 h light/dark cycle (light from 6:00 a.m.).The mice were maintained on a chow or high-fat diet (ssniff Spezialdiäten, Soest, Germany) ad libitum.The chow diet contained 16% calories from fat, 27% from protein, and 57% from carbohydrates.The high-fat diet contained 60% calories from fat, 20% from protein, and 20% from carbohydrates.One group of mice was on a chow diet (n = 16) for three weeks prior to the subsequent experiments, while the other group of mice (n = 16) was first on a chow diet for one week (acclimatization), followed by two weeks on a high-fat diet.After these initial periods, each group was split into (i) a control group (no antibiotics) (n = 8) and (ii) a treatment group (with antibiotics) (n = 8) for two weeks.Ampicillin, streptomycin, and clindamycin at a ratio of 1:1:1 were provided in sterile drinking water at a final concentration of 1 g/L.These antibiotics were chosen due to their broad spectrum capacity and well-documented impacts on the intestinal microbiota [60].The animals were allowed to drink ad libitum during the experiment, with water replacement at 3-day intervals.Feces samples were collected and stored at −80 • C until further analysis.

Sample Preparation
The metabolites were extracted using a biphasic solvent system of cold methanol, methyl tert-butyl ether, and water [52,61].
For the extraction of tissues and feces, 20 mg of the sample in a 2 mL tube was homogenized (1.5 min) with 275 µL of methanol using a grinder.Subsequently, 1 mL of MTBE was added, and this mixture was shaken (30 s).Finally, 275 µL of 10% methanol was added, and after vortexing (10 s), the tubes were centrifuged (24,328× g, 5 min, 4 • C) [61].
A 70 µL aliquot of the bottom phase was collected and evaporated.The dry extracts were further processed as described for biofluids.
The ESI source and MS settings were as follows: sheath gas pressure, 50 arbitrary units; aux gas flow, 13 arbitrary units; sweep gas flow, 3 arbitrary units; capillary temperature, 260 • C; aux gas heater temperature, 425 • C; spray voltage, 3.6 kV; polarity, positive; MS1 mass range, m/z 60-900; MS1 resolving power, 35,000 FWHM; the number of datadependent scans per cycle, 3; MS/MS resolving power, 17,500 FWHM; stepped normalized collision energies, 20, 30, and 40% [52].For HILIC-MS with the post-column addition of D 2 O, the tubing from the HILIC column and the infusion pump (Fusion 100, Chemyx, Stafford, TX, USA) introducing the D 2 O were coupled with a Tee unit before directing the flow into an ion source.

Statistical Analysis
The peak height intensities of N 1 -acetylspermidine from the HILIC-MS profiling were normalized to the amount (mg) and log 10 -transformed before processing using GraphPad Prism software v. 10.1, Boston, MA, USA to compare groups (two-way ANOVA, Tukey's); p < 0.05 was considered significant.

Figure 1 .
Figure 1.Experimental design of LC-HDX-MS experiments: setup (1) of conventional HILIC-MS method with a post-column addition of D 2 O, setup (2) of partial on-column LC-HDX-MS method, and setup (3) of full on-column LC-HDX-MS method.ACN, acetonitrile; AmF, ammonium formate; FA, formic acid.Next, we turned our attention to on-column HILIC-HDX-MS methods with modified mobile phases and the resuspension of dry extracts.For partial HILIC-HDX-MS, only water was replaced by D 2 O, and non-deuterated mobile-phase modifiers (ammonium formate and formic acid) were used.The dry extracts were resuspended in an acetonitrile/D 2 O (4:1, v/v) mixture.The advantage of this approach is that only the key H/D exchange solvent (D 2 O) was needed.On the other hand, for full HILIC-HDX-MS, both D 2 O and deuterated mobile-phase modifiers (D 5 -ammonium formate and D 2 -formic acid) were employed, and the dry extracts were again resuspended in an acetonitrile/D 2 O (4:1, v/v) mixture (Figure 1, Table1).All the H/D exchange components were replaced in this last tested method, thus requiring more individual D-labeled chemicals during the run.

Figure 2 .
Figure 2. Evaluation of selected target compounds containing 2-7 labile hydrogens (#H Bond Donors) under (a) conventional HILIC-MS with post-column addition of D2O (50 μL/min); (b) conventional HILIC-MS with post-column addition of D2O (100 μL/min); (c) partial HILIC-HDX-MS with D2O used in the mobile phases; (d) full HILIC-HDX-MS with D2O, D5-ammonium formate, D2-formic acid used in the mobile phases.Full H/D exchange is indicated with a solid box (□).The most abundant ion (base peak) in the MS1 spectrum is labeled blue (), and the color of lower intensities is proportionally scaled.For details of the composition of mobile phases and used modifiers, see Table 1.

Figure 2 .
Figure 2. Evaluation of selected target compounds containing 2-7 labile hydrogens (#H Bond Donors) under (a) conventional HILIC-MS with post-column addition of D 2 O (50 µL/min); (b) conventional HILIC-MS with post-column addition of D 2 O (100 µL/min); (c) partial HILIC-HDX-MS with D 2 O used in the mobile phases; (d) full HILIC-HDX-MS with D 2 O, D 5 -ammonium formate, D 2 -formic acid used in the mobile phases.Full H/D exchange is indicated with a solid box (□).The most abundant ion (base peak) in the MS1 spectrum is labeled blue (■), and the color of lower intensities is proportionally scaled.For details of the composition of mobile phases and used modifiers, see Table 1.
Figure 2c,d show that the partial and full HILIC-HDX-MS setups provided complete H/D exchange for the evaluated metabolites compared to the setup with the post-column addition of D2O.The full HILIC-HDX-MS method demonstrated slightly better performance, as the intensities of the [M(Dx-1)+D] + ions were lower (with an average absolute difference of ~10%) compared to the [M(Dx)+D] + ions, which represent the full H/D exchange species with x deuterium atoms.The increased ratio of [M(Dx-

Figure
Figure 2c,d show that the partial and full HILIC-HDX-MS setups provided complete H/D exchange for the evaluated metabolites compared to the setup with the post-column addition of D 2 O.The full HILIC-HDX-MS method demonstrated slightly better performance, as the intensities of the [M(D x-1 )+D] + ions were lower (with an average absolute difference of ~10%) compared to the [M(D x )+D] + ions, which represent the full H/D exchange species with x deuterium atoms.The increased ratio of [M(D x-1 )+D] + /[M(D x )+D] + ions in the partial HILIC-HDX-MS method can be attributed to the presence of non-deuterated mobile-phase modifiers providing hydrogen atoms, thereby reducing the yield of fully deuterated ions.We should also note that the H/D exchange efficiency has been reported to be pHdependent[53].For example, amides demonstrate a minimum exchange rate around a pH of 2.5[54], while carbohydrates, primarily composed of hydroxyls, exhibit a minimum exchange rate around a pH of 6.5, with the exchange rate increasing as the solution becomes more acidic or basic[55].The pH of the mobile phases used in the HILIC method was approximately 3, aligning with the observed pH-dependent trends in amides and carbohydrates.However, as metabolomic profiling encompasses diverse groups of metabolites, the H/D exchange efficiency can vary based on the pH of the mobile phase.Figure4shows an example of the amino acid lysine with five labile hydrogen atoms acquired using conventional HILIC-MS with detected [M+H] + ions (Figure4a), followed by incomplete H/D exchange when using the post-column addition of D 2 O, providing ions from [M+H] + to [M(D 5 )+D] + (Figure4b), and complete H/D exchange when the full HILIC-HDX-MS method was used, with dominating [M(D 5 )+D] + ions (Figure4c).All these data show that on-column H/D exchange is advantageous since it generally produces

Figure 4 .
Figure 4. MS1 spectra of amino acid lysine obtained using (a) conventional HILIC-MS (m/z 147.1128, [M+H] + ), (b) conventional HILIC-MS with post-column addition of D 2 O (100 µL/min), and (c) full HILIC-HDX-MS (m/z 153.15046, [M(D 5 )+D] + ).The lysine structure indicates the molecule's possible exchangeable hydrogens (in red).Regarding D 2 O consumption, 850 µL of D 2 O per injection was necessary when utilizing the post-column addition of D 2 O (100 µL/min), whereas approximately 600 µL of mobile phase A with D 2 O and 2.8 mL of mobile phase B (including ~140 µL D 2 O) was required for the partial or full HILIC-HDX-MS setups.Thus, both approaches involved comparable volumes of D 2 O.However, the partial or full HILIC-HDX-MS setups provided more efficient deuterium incorporation into the analytes, making them the preferred choice for achieving complete H/D exchange in untargeted metabolomics studies.

Figure 5 . 30 Figure 6 .
Figure 5. Extracted ion chromatograms (EICs) and MS1 and MS/MS spectra of N 1 -acetyl in mouse feces and analytical standard analysis under conventional HILIC-MS and HDX-MS conditions.In HILIC-MS, the EIC at m/z 188.1757, corresponding to [M( displayed.Conversely, in HILIC-HDX-MS, the EIC at m/z 193.2071, corresponding to [M shown.MS/MS spectra were acquired at stepped normalized collision energies of 20, 30

Figure 7 .
Figure 7. Box plots of N 1 -acetylspermidine during experiments investigating the im (chow vs. high-fat diet) and antibiotic (ABX) treatment on mice.For statistical analy ANOVA was used.

Figure 7 .
Figure 7. Box plots of N 1 -acetylspermidine during experiments investigating the impact of diet (chow vs. high-fat diet) and antibiotic (ABX) treatment on mice.For statistical analysis, two-way ANOVA was used.

Table 1 .
Overview of the key characteristics of evaluated HILIC-MS methods.