Journal of Analysis and Testing

https://doi.org/10.1007/s41664-020-00133-0

REVIEW

The Application of Ion Mobility‑Mass Spectrometry in Untargeted

Metabolomics: from Separation to Identification
Ming‑Du Luo1,2 · Zhi‑Wei Zhou1,2 · Zheng‑Jiang Zhu1

Received: 26 March 2020 / Accepted: 14 May 2020

© The Nonferrous Metals Society of China 2020

Abstract
Untargeted metabolomics aims to comprehensively profile metabolites as many as possible in biological samples. Recently,
ion mobility-mass spectrometry (IM-MS) has emerged as a powerful technology for untargeted metabolomics. The emerging
role of IM-MS in untargeted metabolomics enables the separation of metabolite isomers and generation of multidimension
data to support the identification of metabolites. In this review, we first introduced the basic principles of IM-MS instruments
commonly used for untargeted metabolomics. Then, we demonstrated the application of IM-MS for metabolite separation and
identification of both known and unknown metabolites. Finally, we discussed the future developments of IM-MS technology
to improve untargeted metabolomics.

Keywords Ion mobility-mass spectrometry · Untargeted metabolomics · Collison cross section (CCS) · Metabolite
identification · Isomer separation

1 Introduction [6]. Specifically, liquid chromatography–mass spectrom-

etry (LC–MS) is one of the most popular techniques for
Untargeted metabolomics aims to comprehensively pro- untargeted metabolomics due to its high selectivity, high
file small molecule metabolites in cells, tissues, or whole sensitivity, and broad dynamics range [2, 5]. High-res-
organisms, and to identify changes of metabolites directly olution LC–MS can acquire both accurate mass (MS1)
reflecting the physiological and pathological status [1, 2]. As and tandem MS spectra (MS/MS or MS2) to support the
the downstream omics technology of genomics, transcrip- accurate identification of metabolites [7]. However, several
tomics and proteomics, metabolomics links between geno- challenges are still existing for LC–MS-based untargeted
type and phenotype [3], and is regarded as an essential part metabolomics. First, metabolites have a high diversity of
of systems biology [4]. Metabolites are endogenous small chemical structures and polarities, and require a technol-
molecules produced or consumed during metabolism with ogy with high separation and selectivity for analysis [2].
molecular weight less than 2000 Da. Metabolites in biologi- Second, metabolite identification is still the key challenge
cal systems have a huge number, a vast diversity of chemical in LC–MS-based untargeted metabolomics. The identifica-
structures, numerous isomers, and wide concentration ranges tion is mainly achieved by matching MS1 and MS/MS to the
[2, 5]. These features have presented significant analytical standard metabolite databases, such as METLIN, MassBank,
challenges for untargeted analysis of metabolites. Nuclear MoNA, NIST. However, the coverage of MS/MS spectra in
magnetic resonance (NMR) and mass spectrometry (MS) spectral databases is very limited [8]. Third, there are enor-
are the popular technologies for untargeted metabolomics mous isomeric metabolites existing in the biological samples
[9]. LC–MS has a limited capability to separate and identify
all isomers. Therefore, the development of new technologies
* Zheng‑Jiang Zhu
[email protected] with improved selectivity and accuracy plays an important
role in untargeted metabolomics.
1
Interdisciplinary Research Center on Biology and Chemistry, Recently, ion mobility-mass spectrometry (IM-MS) has
Shanghai Institute of Organic Chemistry, Chinese Academy emerged as a potential technology for untargeted metabo-
of Sciences, Shanghai 200032, China
lomics [10–12]. Ion mobility technology separates ions
2
University of Chinese Academy of Sciences, Beijing 100049, according to the differential movements of ions in neutral
China

13
Vol.:(0123456789)
 Journal of Analysis and Testing

buffer gas under the influence of an electric field, and enables instruments and labs [17]. The coupling of IM with MS pro-
to separate the ions with different charges, structures, and vides a multidimensional separation and analysis technology
conformations [10, 13–15]. The ion mobility (IM) derived with improved selectivity and sensitivity and reduced chemi-
collision cross-section (CCS) represents the effective area cal noise in MS detection [18]. In addition, IM-MS could
for the interaction between an individual ion and the neutral be further hyphenated with the front-end LC separation to
gas through which it travels [16]. CCS has been demon- improve the peak capacity and separation capability (Fig. 1a)
strated as a valuable physiochemical property for metabo- [10]. In untargeted metabolomics, LC–IM-MS technology
lite identification with high reproducibility across different demonstrates several distinct advantages over conventional

Fig. 1  The ion mobility-mass spectrometry (IM-MS) technologies: in commercial available IM-MS instruments, including drift tube ion
(a) The multidimensional separation and characterization in LC– mobility spectrometry (b), travelling wave ion mobility spectrometry
IM-MS/MS; (b–d) The schematic illustration of the IM technologies (c), and trapped ion mobility spectrometry (d)

13
Journal of Analysis and Testing

technologies: (1) improved peak capacity [10], (2) separa- schematic illustration of the IM separation principles of
tion of metabolite isomers [13], and (3) generation of multi- commonly used IM-MS instruments is shown in Fig. 1.
dimensional data to facilitate metabolite identification [19]. More detailed information about the physical principles
In this review, we first introduced the basic principles of of different IM-MS instruments can be found in these spe-
IM-MS instruments commonly used for untargeted metabo- cialized reviews [14, 24]. Here, we focus on their imple-
lomics. Then, the recent applications of IM-MS in metabo- mentation for untargeted metabolomics, and their major
lite analysis and untargeted metabolomics were summarized, characteristics are summarized in Table 1.
including (1) effective separation of metabolites; (2) accu- Drift tube ion mobility-mass spectrometry (DTIM-MS)
rate metabolite identification using IM derived CCS values; is widely used for untargeted metabolomics and enables to
(3) identification of unknown metabolites using IM-MS. generate gold standard CCS values using the stepped-field
Finally, we discussed the future developments of IM-MS approach. DTIM-MS is currently commercialized by Agi-
technology for untargeted metabolomics. lent Technologies, named as Agilent 6560 DTIM-QTOF
[22]. In DTIM-MS, ring electrodes are stacked to form the
drift cell where a weak uniform electric field is applied to
2 IM‑MS Instruments for Untargeted drive the movement of ions through it (Fig. 1b). Buffer
Metabolomics gas, such as N ­ 2 or He, fills the drift cell as a stationary
phase. Ions with different structures have different drift
The commercially available IM-MS instruments can be times, enabling the separation of isomeric ions with differ-
classified into three types, including spatially-dispersive, ent structures. The drift time can be affected by the buffer
time-dispersive, and confinement and selective release. gas and its velocity. Currently, most commercial IM-MS
Spatially-dispersive IM instruments enable to separate instruments use nitrogen as the buffer gas. Compared
ions along different paths in the drift cell, such as field with drift time, CCS is reproducible and independent
asymmetric waveform ion mobility spectrometry (FAIMS) from instrument settings, therefore, it can be used to char-
[20] and differential mobility spectrometry (DMS) [21]. acterize the metabolites [25]. Some protocols have been
They are typically implemented for targeted metabolite reported to optimize the conditions and maximize the IM
analysis due to the high selectivity. However, spatially- separation [26, 27]. The coupling DTIM to a QTOF mass
dispersive IM is not common in untargeted metabolomics spectrometer enables the simultaneous detection of mass-
because no CCS values could be generated for metabolite to-charge (m/z) and drift time of an ion. In addition, col-
identification. In untargeted metabolomics, time-dispersive lision induced dissociation (CID) can be performed post
IM instruments, including drift tube ion mobility spec- mobility to generate mobility-correlated MS/MS spectra
trometry (DTIMS) [22] and travelling wave ion mobility via the all-ion fragmentation (AIF) [14]. The integration of
spectrometry (TWIMS) [12], and confinement and selec- m/z, drift time (or CCS) and MS/MS facilitates the metab-
tive release IM instruments, such as trapped ion mobility olite identification in untargeted metabolomics. Currently,
spectrometry (TIMS) [23], are commonly utilized. The the IM resolution of Agilent DTIM-MS is about 60 [22].

Table 1  The comparisons of four common commercial IM-MS instruments for untargeted metabolomics
DTIM-MS TWIM-MS TIMS-MS Cyclic IM-MS

Separation type Time-dispersive Time-dispersive Confinement and selective Time-dispersive

release
Commercial instrument Agilent 6560 DTIM-QTOF Waters Synapt; Waters Bruker timsTOF; Bruker Waters Select Series Cyclic
Vion IMS-QTOF timsTOF Pro IMS
MS type Time-of-flight Time-of-flight Time-of-flight Time-of-flight
Resolving power ~ 60 ~ 40 ~ 200 ~ 350 (16 passes)
CCS measurement Calibrant-dependent or Calibrant-dependent Calibrant-dependent Calibrant-dependent
-independent
IM separation of fragment No Yes No Yes
ions
Advantages Gold standard CCS values High versatility on ion High resolution; High duty High resolution; ­IMSn
fragmentation cycle
Limitations Low IM resolution; Lim- Low IM resolution; Lim- Space charge effect; Mass Resolution depends on the
ited orthogonality to MS ited orthogonality to MS range-dependent IM number of passes
resolution

13
 Journal of Analysis and Testing

13
 Journal of Analysis and Testing

◂Fig. 2  The applications of IM-MS for metabolite separation: (a) the of m/z and CCS ranges, and highly structural similarity to
correlation CCS values and m/z (left panel), and the trend lines fit- metabolites [32]. The proper applications of calibrants for
ted with measured CCS values (right panel); (b) drift times and CCS
values of monosaccharide isomers, including d-Mannose, d-Allose,
specific metabolites could provide more accurate calibration
d-Glucose, and myo-Inositol (data generated from Zhu group with and CCS measurements, such as lipids [33]. These calibrants
chemical standards on a Agilent 6560 DTIM-QTOF instrument); (c) were summarized in a recent review [25]. The IM resolu-
the chelate complex ions [(DPro)2 + D/LAA + CuII-H]+ for the differ- tion of TWIM-MS Synapt is about 40 [14]. Recently, cyclic
entiation of enantiomers of phenylalanine, tryptophan and tyrosine.
(a was reproduced from Ref. 22 with permission from the American
IM-MS [34] and structures for lossless ion manipulations
Chemical Society; c was reproduced from Ref. 47 with permission (SLIM) [35] technologies were developed based on TWIM-
from the Royal Society of Chemistry) MS with IM resolution as high as 400.
Trapped ion mobility spectrometry–mass spectrometry
(TIMS-MS) is one of the newest commercial IM-MS instru-
In Agilent DTIM-MS, there are two methods to measure ments produced by Bruker with a great potential in untar-
CCS values including the stepped-field and the single-field geted metabolomics [23]. In an ion funnel, ions are dragged
methods [19]. The stepped-field method utilizes at least 5 by a gas flow toward the exit funnel while the applied
different drift voltages to build a linear regression curve opposed electric field pushes ions back to the entrance fun-
between the drift times and the inversed drift voltages. nel (Fig. 1d). Ions with equal charge number experience the
Then, the CCS can be derived from the slope of the linear equal electric force but unequal dragging force from the gas
regression curve according to the Mason–Schamp equation flow due to different CCS. The duty cycle is low because
[22, 27]. The stepped-field method in DTIM-MS has high of the accumulation procedure. The second generation
measurement reproducibility of CCS values with a rela- of TIMS-MS (i.e., timsTOF Pro) employed two stages of
tive standard deviation (RSD) of 0.29% [27]. The method TIMS. The first ion trap is used for accumulation of ions,
is calibrant-independent, and regarded as the gold stand- whereas the second one is used for IM separation and meas-
ard for CCS measurements. However, the method requires urement, achieving nearly 100% duty cycle. TIMS has an IM
a long cycle time for each measurement, and is not suitable resolution as high as about 200. In TIMS-MS, TOF is also
for LC–IM-MS-based untargeted metabolomics. Instead, implemented as the mass analyzer. Most importantly, MS/
the single-field method is developed for CCS measurement MS spectra can be acquired with both data-dependent and
which requires a fixed drift voltage [26]. The method is data-independent acquisition modes [14, 23]. Similarly, the
calibrant-dependent. A mixture of chemical standards with measurement of CCS values in TIMS-MS is also calibrant-
known CCS values, such as Agilent ESI-L Low Concentra- dependent, and Agilent ESI-L Low Concentration Tune Mix
tion Tune Mix, was first analyzed to establish a correlation is often used for calibration [23, 36, 37]. However, when
between the drift time and CCS and to calculate the related compared with DTIM-MS and TWIM-MS, available experi-
coefficients, including the mobility-independent flight time mental CCS values from TIMS-MS is quite limited.
(tfix) and instrument-dependent proportionality coefficient
(β). Then, CCS values of metabolites can be derived from
the measured drift times and the coefficients. The Agilent 3 IM‑MS for Metabolite Separation
ESI-L Low Concentration Tune Mix mainly consists of
betaine, trifluoracetic acid ammonium salt, and a class of In IM-MS-based untargeted metabolomics, IM enables to
hexakisphosphate compounds, which was reported by Stow separate metabolites according to their structures and con-
et al. [27]. Currently, DTIM-MS has been commonly used formations, while MS enables to separate ions according to
in both targeted and untargeted metabolomics analyses [9, their m/z. The rapid 2D separation provides the correlation
28–31]. between the molecular structure and m/z, which could be
Travelling wave ion mobility-mass spectrometry (TWIM- visualized in the CCS-m/z 2D plot (Fig. 2a) [22]. The certain
MS) is the most popular IM instrument for untargeted classes of metabolites locate in distinct spaces in the plot,
metabolomics, and is commercialized by Waters [17]. In and the trend lines can be fitted through power–law func-
TWIM, dynamic AC voltage is applied to the piled elec- tion to define and describe the specific molecular classes
trodes to generate voltage waves and drive the ion through (Fig. 2a). The trend line technology is powerful to deline-
the drift cell (Fig. 1c). Different from DTIM-QTOF, the ate a molecule into a specific metabolite class. May et al.
TWIM drift cell is placed between quadrupole and TOF, measured CCS values of 63 ammonium salts, 314 lipids, 92
and CID can take place before or after the IM separation peptides, and 125 carbohydrates with DTIM-MS (Fig. 2a)
with the data-independent acquisition (i.e., ­MSE mode) [14]. [22]. Trend lines of different classes are distinct from oth-
In TWIM, CCS values are measured via calibrant-depend- ers. Similarly, Hernandez-Mesa et al. proposed a CCS data-
ent manner. Polyalanine oligomer is one of the most widely base of 300 steroids [38]. They found the good correlations
used calibrants due to the long-term stability, wide coverage between CCS values and m/z can be observed when steroids

13
 Journal of Analysis and Testing

were divided into different subclasses according to the ster- such as D-Mannose, D-Allose and D-glucose, have distinct
oid skeleton. They also observed that better correlations differences in CCS values ranging from 1.4 to 3.9%, and
were observed for steroid monomers than dimers. can be separated by DTIM-MS (Fig. 2b). Hofmann et al.
IM-MS has also demonstrated a great potential in sepa- also reported the separation of trisaccharide diastereomers
rating and differentiating the metabolites isomers [39]. difference in α and β configuration with TWIM-MS [45].
Metabolite isomers are molecules with the same formulas, An enantiomer is one of two stereoisomers that are mirror
but different chemical structures. There are enormous iso- images of each other, therefore, enantiomers have the same
meric metabolites existing in the biological samples. For CCS values. A chiral selector must be introduced to enan-
example, in KEGG database, 10,137 out of 17,935 metabo- tiomers and allows to form the chelate diastereomers for
lites (> 58%) have an isomeric metabolite. Metabolite iso- separation on IM-MS [46]. This strategy has been widely
mers tend to share similar fragments in MS/MS spectra, reported. For example, Domalain et al. utilized TWIMS to
which presents a challenge for metabolite identification. differentiate the enantiomers of aromatic amino acids (AA),
IM-MS also enables the effective separation of metabolite including phenylalanine, tryptophan and tyrosine, through
isomers, including both constitutional and stereo isomers. the introduction of cooper(II) and the chiral selector D-pro-
Dobbs et al. investigated a total of 11 leucine/isoleucine line [47]. The d- or l-amino acids formed as the complex
isomers, and evaluated the potential to utilize IM-MS for ion [(DPro)2 + D/LAA + CuII-H]+ with a 50% of baseline
separating the isomeric mixtures [39]. In this study, con- separation (Fig. 2c). Similarly, Yu et al. reported the use of
stitutional isomers had 0.3–6.9% difference in CCS values TWIMS to separate enantiomers of amino acids by form-
and most of them can be separated by DTIM-MS. Similarly, ing the copper-bound complex ion ­[L/DM + 3LY + 2Cu2+-
our group also utilized DTIM-MS to measure the CCS val- 4H + Na]+, in which Y is the chiral selectors like trypto-
ues of 4 monosaccharide isomers. We found the CCS value phan and histidine, and M is the chiral amino acids [48].
of myo-Inositol is significantly different with its constitu- Significant enantiomer recognition was observed for amino
tional isomers (d-Mannose, d-Allose, and d-glucose, shown acids with aromatic rings like tyrosine and tryptophan or
in Fig. 2b). Several other studies have also demonstrated long side chains, such as glutamic acid. More isomer separa-
the use of IM-MS to separate constitutional isomers, such tion examples in real samples have been reported for steviol
as lipids [40], peptides [34], drug-like compounds [41] and glycosides in food commodities [49] and lipids in human
environmental toxins [42]. plasma samples [37, 44].
Stereo isomers, such as cis–trans isomers and opti-
cal isomers, share highly similar structures, and usu-
ally require high resolution of IM for separation [39]. 4 IM‑MS for Metabolite Identification
For cis–trans isomers, Jeanne et al. utilized TIMS-MS
to separate phosphatidylcholines (PC) and diacylglyc- In untargeted metabolomics, metabolite identification is still
erols (DG) cis–trans isomers with IM resolution of the key challenge. IM-MS derived CCS value is high repro-
more than 300, such as PC (16:1(9Z)/16:1(9Z)) and PC ducible property of metabolite ion, and enabled metabolite
(16:1(9E)/16:1(9E)), DG (22:1(13Z)/22:1(13Z)/0:0) and DG identification. Therefore, the curation of CCS database is
(22:1(13E)/22:1(13E)/0:0), DG (22:1(13Z)/0:0/22:1(13Z)), one of most important parts for metabolite identification in
and DG (22:1(13E)/0:0/22:1(13E)) [37]. Other high- IM-MS. In the past several years, several CCS databases
resolution IM technologies, such as SLIM can also be have been reported for small molecules, covering metab-
used to directly separate cis–trans isomers. Wojcik olites [30], lipids [31], drugs [28], and other xenobiotics
reported the separation of PC (16:1(9Z)/16:1(9Z)) and PC [29]. The details for the CCS database curation and related
(16:1(9E)/16:1(9E)) on SLIM [43]. In addition, the ozo- resources were summarized in our recent review article [19].
nolysis reaction coupled to IM-MS also provided the pos- Here, we summarized several representative CCS databases
sibility to separate the cis/trans isomers of phosphatidyle- for metabolomics (Table 2). Generally, these databases are
thanolamines (PE), such as PE (18:1(9Z)/ 18:1(9Z)) and PE classified into experimental and in silico types. One of the
(18:1(9E)/18:1(9E)) [44]. most representative experimental CCS database is the CCS
For stereo isomers, the systematic evaluation of the rela- compendium developed by McLean group [50], which col-
tionship between CCS difference and the resolution has been lected a total of 3833 experimental CCS values in DTIM-
investigated by Dobbs et al. [39]. They demonstrated that MS. Although the experimental CCS values increased
IM resolution > 300 could to separate stereo isomers with rapidly, the coverage still is very limited as compared to
CCS difference less than 0.5%. Diastereomers have one the number of known metabolites. In contrast, the in silico
or more opposite stereochemistry for given stereo centers generated CCS database has an indispensable advantage for
and relatively large structural differences with each other. metabolite coverage. Different approaches to generate in
For example, our group demonstrated that diastereomers silico CCS values, such machine-learning-based prediction

13
Journal of Analysis and Testing

Table 2  The major available CCS databases for metabolomics

References Database type Instrument No. of CCS values No. of compounds Web resource

Zheng et al. (2017) [29] Experimental CCS DTIM-MS 826a 459a Yesc
Hines et al. (2017) [28] Experimental CCS TWIM-MS 1,440a 1,425a No
Hernandez- Mesa et al. (2018) Experimental CCS TWIM-MS 1,080a 300a No
[38]
Nichols et al. (2018) [9] Experimental CCS DTIM-MS 1,246a 554a No
Picache et al. (2019) [50] Experimental CCS DTIM-MS 3, ­271a 1,142a Yesd
Schroeder et al. (2019) [23] Experimental CCS TIMS-MS 343a 146a No
Zhou et al. (2016) [30] Predicted CCS DTIM-MS 176,015b 35,203b Yese
Colby et al. (2019) [54] Experimental CCS Predicted DTIM-MS 1,454a 2,111,575b 879a 718,595b Yesf
CCS
Zhou et al. (2020) [56] Experimental CCS Predicted DTIM-MS TWIM-MS 5,119a 11,697,711b 2,193a 1,670,596b Yesg
CCS

Superscripts c-g are links for web resource

a, b
Represent experimental CCS and predicted CCS, respectively
c
https​://panom​ics.pnnl.gov/metab​olite​s/
d
https​://lab.vande​rbilt​.edu/mclea​n-group​/colli​sion-cross​-secti​on-datab​ase/
e
https​://www.zhula​b.cn/MetCC​S/
f
https​://metab​olomi​cs.pnnl.gov/
g
https​://allcc​s.zhula​b.cn/

[19, 31, 51–53] and quantum chemistry-based theoretical CCS library [49]. More importantly, IM-MS generated
calculation [54, 55] have also been reported. To ensure the multidimensional properties (i.e. MS1, RT, CCS, and MS/
prediction reliability, it generally requires the following con- MS) could be integrated to achieve accurate identification
ditions: (1) a large and representative training dataset; (2) of metabolites in untargeted metabolomics (Fig. 3a). For
a well-trained prediction model without over-fitting; (3) the example, Schroeder et al. identified 40 metabolites in plant
high structural similarity between the input structure and with an experimental database containing RT, MS1, CCS,
the training dataset. More details have been discussed in and MS/MS spectra [23]. This strategy was also used for
our recent review [19]. Several in silico CCS databases have the pesticide screening in fish feed, in which tebuconazole
been curated, such as MetCCS [30], LipidCCS [31] and ISi- and piperonil butoxide were identified for the first time in
CLE [54], providing over one million CCS values. However, these samples [58]. However, metabolite identification is
currently, it lacks unified platforms to host and share both significantly limited by the coverage of the experimental
experimental and predicted CCS values. Very recently, our CCS database. Instead, the in silico CCS database could
group have developed a new CCS atlas, namely, AllCCS, significantly expand the annotation coverage. For example,
to embrace both experimental and predicted CCS values Zhou et al. putatively identified 1284 metabolites in human
[56]. AllCCS database included more than 5000 experi- plasma using the MetCCS database [30]. Bijlsma et al. also
mental CCS records, and ~ 12 million predicted CCS values screened more than 600 pesticides in spinach samples, and
for > 1.6 million chemical compounds. This database is the putatively annotated 10 pesticides using the predicted CCS
first unified CCS database with the widest coverage (https​ values [52]. Finally, IM-MS also enables to identify more
://allcc​s.zhula​b.cn/). metabolites with improved separation. Zhang et al. recently
For metabolite identification, the most common strategy compared the performance of LC–IM-MS to traditional
is matching MS1 and CCS values with experimental CCS LC–MS, and IM-MS increased additional 22% metabolic
database. This strategy can reduce false identifications espe- peaks that overlapping in mass and retention time [18].
cially when the standard MS/MS spectra are not available. The recent advancements in bioinformatics enabled the
For example, Zheng et al. identified metabolites in urine integration of multidimensional information to provide a
using measured MS1 and CCS values with their experimen- holistic and large-scale identification of small molecules
tal CCS database [29]. Similarly, Stephen et al. also utilized in complex biological samples. For example, Zhou et al.
this strategy for contaminant screening in wastewater analy- developed a software tool, namely, LipidIMMS Analyzer
sis [57]. McCullagh et al. screened the steviol glycosides in (https:​ //imms.zhulab​ .cn/LipidI​ MMS/), to integrate 4-dimen-
the food commodities using TWIM-MS with an established sional information (i.e., MS1, RT, CCS and MS/MS) for

13
 Journal of Analysis and Testing

Fig. 3  The applications of IM-MS for metabolite identification: (a) related metabolites through the in silico predicted CCS values (b was
multidimensional data generated in IM-MS for annotation of both reproduced from Ref. 50 with permission from the Royal Society of
known and unknown metabolites; (b) the class-specific trend lines for Chemistry; c was reproduced from Ref. 62 with permission from the
the classification of an unknown feature, and the reduction of candi- American Chemical Society)
dates by the CCS database; (c) the annotation of unknown pesticide

lipid identification in LC–IM-MS/MS-based lipidomics provided a comprehensive lipidome atlas with retention
[59]. More recently, Tsugawa et al. released MS-DIAL4 time, CCS, and MS/MS information. However, these strate-
to provide an all-in-one solution from raw data processing gies were mostly applied in untargeted lipidomics, because
to multidimensional lipid identification [60]. It formulated lipids have well-predicted and accurate CCS values, reten-
mass spectral fragmentations for 117 lipid subclasses, and tion time, and MS/MS spectra. In untargeted metabolomics,

13
Journal of Analysis and Testing

the multidimensional metabolite identification is still a chal- addition, there are increasingly demands to couple IM-MS
lenge since the accuracy of predicting retention time and instruments with different ionization sources, and enable
MS/MS spectra is still low. imaging analysis of metabolites. Some successful examples
In addition to known metabolites, the identification of have been demonstrated to implement ionization sources,
unknowns is another grand challenge for untargeted metabo- such as matrix-assisted laser desorption ionization (MALDI)
lomics. The certain classes of metabolites locate in distinct [63], desorption electrospray ionization (DESI) [64], and
spaces in the CCS-m/z plot, and the trend lines can be fitted laser ablation electrospray ionization (LAESI) [65] with
to define and describe the specific molecular classes. There- IM-MS for metabolite imaging analysis. Coupling IM with
fore, the trend line technology in IM-MS could facilitate imaging MS technology facilitate to achieve the spatial
unknown metabolite annotation. For example, Goodwin information of metabolites in biological sample, such as a
et al. used the trend lines to distinguish secondary cyclic tissue [63]. Finally, the quantitation performance of IM-MS
peptides from crude microbial extract [61]. Picache et al. for untargeted metabolomics is less studied. Zhang et al.
also employed the class-specific trend lines to putatively systematically evaluated the quantitation performance of
identify an unknown feature as PC and PE [50]. They also Agilent DTIM-QTOF 6560 with the comparison to regular
demonstrated that the CCS data can also be used to filter LC-QTOF [18]. They found that LC–IM-MS had a limited
most of the false candidates (Fig. 3b). In addition, other linear dynamic range and limit of detection (LOD) due to the
studies combined the in silico metabolic reaction with the limited trapping capacity of ion funnel trap prior to the IM
predicted CCS for unknown metabolite annotation. For drift tube. In IM-MS, the sensitivity drop in IM-MS has been
example, Bijlsma et al. utilized the in silico reactions to observed with different instrument types and mass ranges.
generate possible the metabolites of insecticide pirimiphos- Specifically, the sensitivity has been evaluated with a signifi-
methyl (PM) in fish (Fig. 3c) [62]. They used TWIM-MS to cant decrease about 1–4 folds in the drift tube IM for low m/z
acquire the untargeted metabolomics data and match the data ions (m/z ≤ 250), and a slightly decrease for high m/z ions
to the properties of possible metabolites, including MS1, [66]. In a recent publication [66], the multiplexing approach
predicted CCS and retention time. They identified 3 new PM was developed to effectively improve the sensitivity without
metabolites in different tissues of fish, such as liver, kidney, increasing the sample injection. The improvement of quanti-
bile, muscle and fat. The study demonstrated the effective- fication in IM-MS is also required in the future.
ness of this strategy to identify the unknown endogenous Second, the development of software tools and databases
metabolites derived from exogenous molecules, like drugs for IM-MS data processing is urgently needed. Currently,
and pesticides. We foresee that the combination of large- available software for IM-MS data processing is mainly pro-
scale CCS database and other in silico MS/MS spectra will vided by instrument vendors, like Agilent IM-MS browser
facilitate to identify more unknown metabolites. and Mass Profiler, Waters Progenesis QI and Bruker Meta-
boScape. These tools are specifically designed for their own
instruments. Recently, the open source and freely available
5 Outlook software MS-DIAL has updated to version 4, and supports
the processing IM-MS data from different vendors’ instru-
IM-MS has demonstrated its unique advantages for untar- ments [60]. After processing of raw data with the software
geted metabolomics, including the separation of metabolite mentioned above, the exported feature table can be imported
isomers and metabolite identification. Regarding the whole to traditional metabolomics platforms (e.g., SIMCA, and
workflow of IM-MS-based metabolomics, we think the MetaboAnalyst) to conduct pairwise or multiple compari-
low IM resolution, limited quantitation, and complicated son. Meanwhile, it is also very important to further develop
data processing are possible pitfalls in the current IM-MS user-friendly and freely accessible CCS databases in the
application. In the future, a series of improvements are IM-MS field. The AllCCS atlas developed by our group is
further required for instrumentation, data processing, and an example, and enables to support different applications
applications. First, the continuous improvement of IM-MS of IM-MS-based untargeted metabolomics [56]. Finally, we
instrument performances, such as IM resolution, imaging look forward in seeing more software tools and databases
analysis, and quantitation, is necessary. Currently, most of developed for untargeted metabolomics.
commercial IM-MS instruments have low IM resolutions With the development of IM-MS instruments and soft-
of 40–60, which severely limits the application of isomer ware, it opens a new stage for the application of IM-MS in
separation. The good news is that several high-resolution untargeted metabolomics. The ultra-high separation speed
IM-MS instruments are commercialized, such as Bruker tim- of IM has a potential to replace the liquid chromatography to
sTOF Pro [23] and Waters Cyclic IM-MS [34]. The avail- achieve high-throughput metabolomics analyses. Interesting
ability of high-resolution IM-MS will open a new stage for examples have been demonstrated by the Fernańdez group
the application of IM-MS in untargeted metabolomics. In [67] and Baker group [68]. They utilized the flow injection

13
 Journal of Analysis and Testing

and solid-phase extraction (SPE) for IM-MS-based high- 9. Nichols CM, Dodds JN, Rose BS, Picache JA, Morris CB,
throughput metabolomics, respectively. We believe that the Codreanu SG, May JC, Sherrod SD, McLean JA. Untargeted
molecular discovery in primary metabolism: collision cross
combination of IM-MS with automatic sample preparing section as a molecular descriptor in ion mobility-mass spec-
and injection technologies (e.g., Agilent RapidFire system) trometry. Anal Chem. 2018;90(24):14484–92. https​: //doi.
enable to high-efficient and large-scale metabolomics. Then, org/10.1021/acs.analc​hem.8b043​22.
the population-based cohort study such as precision medi- 10. Lapthorn C, Pullen F, Chowdhry BZ. Ion mobility spectrom-
etry-mass spectrometry (IMS-MS) of small molecules: sepa-
cine will get benefits. IM-MS has a promising potential for rating and assigning structures to ions. Mass Spectrom Rev.
single cell metabolomics when it is equipped with in-situ 2013;32(1):43–71.
ionization source. For example, Vertes group combined cap- 11. May JC, Gant-Branum RL, McLean JA. Targeting the untargeted
illary microsampling with ESI-IM-MS for in-situ metabolic in molecular phenomics with structurally-selective ion mobility-
mass spectrometry. Curr Opin Biotechnol. 2016;39:192–7.
analysis of the single plant cell [69]. They demonstrated 12. Paglia G, Astarita G. Metabolomics and lipidomics using
the metabolic differences existing in different types of plant traveling-wave ion mobility mass spectrometry. Nat Protoc.
cells. Overall, IM-MS is increasingly becoming a powerful 2017;12(4):797–813. https​://doi.org/10.1038/nprot​.2017.013.
technology for untargeted metabolomics with the unique 13. Lanucara F, Holman SW, Gray CJ, Eyers CE. The power of
ion mobility-mass spectrometry for structural characteriza-
advantages in metabolite separation and identification. tion and the study of conformational dynamics. Nat Chem.
2014;6(4):281.
Acknowledgements The work was supported by National Natural 14. May JC, McLean JA. Ion mobility-mass spectrometry: time-
Science Foundation of China (Grant No. 31971356), and Shang- dispersive instrumentation. Anal Chem. 2015;87(3):1422–36.
hai Municipal Science and Technology Major Project (Grant No. 15. Zheng X, Wojcik R, Zhang X, Ibrahim YM, Burnum-Johnson
2019SHZDZX02). KE, Orton DJ, Monroe ME, Moore RJ, Smith RD, Baker ES.
Coupling front-end separations, ion mobility spectrometry,
Funding The work was supported by National Natural Science Foun- and mass spectrometry for enhanced multidimensional bio-
dation of China (Grant No. 31971356), and Shanghai Municipal Sci- logical and environmental analyses. Ann Rev Anal Chem.
ence and Technology Major Project (Grant No. 2019SHZDZX02). 2017;10:71–92.
16. May JC, Morris CB, McLean JA. Ion mobility collision cross
Data Availability Not applicable. section compendium. Anal Chem. 2016;89(2):1032–44. https​://
doi.org/10.1021/acs.analc​hem.6b049​05.
17. Paglia G, Williams JP, Menikarachchi L, Thompson JW, Tyldes-
Compliance with ethical standards ley-Worster R, Halldórsson S, Rolfsson O, Moseley A, Grant D,
Langridge J, Palsson BQ, Astarita G. Ion mobility derived col-
Conflict of interest The authors declare that they have no conflict of lision cross sections to support metabolomics applications. Anal
interest. Chem. 2014;86(8):3985–93. https​://doi.org/10.1021/ac500​405x.
18. Zhang X, Kew K, Reisdorph R, Sartain M, Powell R, Armstrong
M, Quinn K, Cruickshank-Quinn C, Walmsley S, Bokatzian
S, Darland E, Rain M, Imatani K, Reisdorph N. Performance
of a high-pressure liquid chromatography–ion mobility-mass
References spectrometry system for metabolic profiling. Anal Chem.
2017;89(12):6384–91. https ​ : //doi.org/10.1021/acs.analc​
1. Nicholson JK, Lindon JC. Metabonomics. Nature. hem.6b046​28.
2008;455(7216):1054–6. 19. Zhou Z, Tu J, Zhu ZJ. Advancing the large-scale CCS database for
2. Patti GJ, Yanes O, Siuzdak G. Metabolomics: the apogee of the metabolomics and lipidomics at the machine-learning era. Curr
omics trilogy. Nat Rev Mol Cell Biol. 2012;13(4):263–9. https​:// Opin Chem Biol. 2018;42:34–41.
doi.org/10.1038/nrm33​14. 20. Kolakowski BM, Mester Z. Review of applications of high-
3. Fiehn O. Metabolomics—the link between genotypes and pheno- field asymmetric waveform ion mobility spectrometry (FAIMS)
types. In: Functional genomics. Springer, 2002; pp 155–171. and differential mobility spectrometry (DMS). Analyst.
4. Kitano H. Systems biology: a brief overview. Science. 2007;132(9):842–64.
2002;295(5560):1662–4. https​://doi.org/10.1126/scien​ce.10694​ 21. Zhang JD, Mohibul Kabir KM, Lee HE, Donald WA. Chiral
92. recognition of amino acid enantiomers using high-definition dif-
5. Cajka T, Fiehn O. Toward merging untargeted and targeted meth- ferential ion mobility mass spectrometry. Int J Mass Spectrom.
ods in mass spectrometry-based metabolomics and lipidomics. 2018;428:1–7. https​://doi.org/10.1016/j.ijms.2018.02.003.
Anal Chem. 2016;88(1):524–45. 22. May JC, Goodwin CR, Lareau NM, Leaptrot KL, Morris CB,
6. Fernie AR, Trethewey RN, Krotzky AJ, Willmitzer L. Metabolite Kurulugama RT, Mordehai A, Klein C, Barry W, Darland E,
profiling: from diagnostics to systems biology. Nat Rev Mol Cell Overney G, Imatani K, Stafford GC, Fjeldsted JC, McLean
Biol. 2004;5(9):763–9. JA. Conformational ordering of biomolecules in the gas phase:
7. Zhu ZJ, Schultz AW, Wang J, Johnson CH, Yannone SM, Patti nitrogen collision cross sections measured on a prototype high
GJ, Siuzdak G. Liquid chromatography quadrupole time-of-flight resolution drift tube ion mobility-mass spectrometer. Anal Chem.
mass spectrometry characterization of metabolites guided by the 2014;86(4):2107–16. https​://doi.org/10.1021/ac403​8448.
METLIN database. Nat Protoc. 2013;8(3):451–60. 23. Schroeder M, Meyer SW, Heyman HM, Barsch A, Sumner LW.
8. Vinaixa M, Schymanski EL, Neumann S, Navarro M, Salek RM, Generation of a collision cross section library for multi-dimen-
Yanes O. Mass spectral databases for LC/MS- and GC/MS-based sional plant metabolomics using UHPLC-trapped ion mobility-
metabolomics: state of the field and future prospects. TrAC, MS/MS. Metabolites. 2019;10(1):13. https​://doi.org/10.3390/
Trends Anal Chem. 2016;78:23–35. metab​o1001​0013.

13
Journal of Analysis and Testing

24. Dodds JN, Baker ES. Ion mobility spectrometry: fundamental con- 38. Hernandez-Mesa M, Le Bizec B, Monteau F, Garcia-Campana
cepts, instrumentation, applications, and the road ahead. J Am Soc AM, Dervilly-Pinel G. Collision cross section (CCS) data-
Mass Spectrom. 2019;30(11):2185–95. base: an additional measure to characterize steroids. Anal
25. Gabelica V, Shvartsburg AA, Afonso C, Barran P, Benesch JLP, Chem. 2018;90(7):4616–25. https​://doi.org/10.1021/acs.analc​
Bleiholder C, Bowers MT, Bilbao A, Bush MF, Campbell JL, hem.7b051​17.
Campuzano IDG, Causon T, Clowers BH, Creaser CS, Pauw 39. Dodds JN, May JC, McLean JA. Investigation of the com-
ED, Far J, Fernandez-Lima F, Fjeldsted JC, Giles K, Groessl M, plete suite of the leucine and isoleucine isomers: toward pre-
Hogan CJ Jr, Hann S, Kim HI, Kurulugama RT, May JC, McLean diction of ion mobility separation capabilities. Anal Chem.
JA, Pagel K, Richardson K, Ridgeway ME, Rosu F, Sobott F, 2016;89(1):952–9. https​://doi.org/10.1021/acs.analc​hem.6b041​
Thalassinos K, Valentine SJ, Wyttenbach T. Recommendations 71.
for reporting ion mobility mass spectrometry measurements. Mass 40. Leaptrot KL, May JC, Dodds JN, McLean JA. Ion mobility confor-
Spectrom Rev. 2019;38(3):291–32020. https​://doi.org/10.1002/ mational lipid atlas for high confidence lipidomics. Nat Commun.
mas.21585​. 2019;10(1):985. https​://doi.org/10.1038/s4146​7-019-08897​-5.
26. Kurulugama RT, Darland E, Kuhlmann F, Stafford G, Fjeldsted J. 41. Ross DH, Seguin RP, Xu L. Characterization of the impact of drug
Evaluation of drift gas selection in complex sample analyses using metabolism on the gas-phase structures of drugs using ion mobil-
a high performance drift tube ion mobility-QTOF mass spectrom- ity-mass spectrometry. Anal Chem. 2019;91(22):14498–50707.
eter. Analyst. 2015;140(20):6834–44. https​://doi.org/10.1021/acs.analc​hem.9b032​92.
27. Stow SM, Causon TJ, Zheng X, Kurulugama RT, Mairinger T, 42. Zheng X, Dupuis KT, Aly NA, Zhou Y, Smith FB, Tang K, Smith
May JC, Rennie EE, Baker ES, Smith RD, McLean JA, Hann RD, Baker ES. Utilizing ion mobility spectrometry and mass spec-
S, Fjeldsted JC. An interlaboratory evaluation of drift tube ion trometry for the analysis of polycyclic aromatic hydrocarbons,
mobility–mass spectrometry collision cross section measure- polychlorinated biphenyls, polybrominated diphenyl ethers and
ments. Anal Chem. 2017;89(17):9048–55. their metabolites. Anal Chim Acta. 2018;1037:265–73.
28. Hines KM, Ross DH, Davidson KL, Bush MF, Xu L. Large-scale 43. Wojcik R, Webb I, Deng L, Garimella S, Prost S, Ibrahim Y, Baker
structural characterization of drug and drug-like compounds ES, Smith RD. Lipid and glycolipid isomer analyses using ultra-
by high-throughput ion mobility-mass spectrometry. Anal high resolution ion mobility spectrometry separations. Int J Mol
Chem. 2017;89(17):9023–30. https​://doi.org/10.1021/acs.analc​ Sci. 2017;18(1):183. https​://doi.org/10.3390/ijms1​80101​83.
hem.7b017​09. 44. Poad BLJ, Zheng X, Mitchell TW, Smith RD, Baker ES, Blanksby
29. Zheng X, Aly NA, Zhou Y, Dupuis KT, Bilbao A, Paurus VL, SJ. Online ozonolysis combined with ion mobility-mass spec-
Orton DJ, Wilson R, Payne SH, Smith RD, Baker ES. A struc- trometry provides a new platform for lipid isomer analyses. Anal
tural examination and collision cross section database for over 500 Chem. 2018;90(2):1292–300. https​://doi.org/10.1021/acs.analc​
metabolites and xenobiotics using drift tube ion mobility spec- hem.7b040​91.
trometry. Chem Sci. 2017;8(11):7724–36. https:​ //doi.org/10.1039/ 45. Hofmann J, Hahm HS, Seeberger PH, Pagel K. Identification of
c7sc0​3464d​. carbohydrate anomers using ion mobility–mass spectrometry.
30. Zhou Z, Shen X, Tu J, Zhu ZJ. Large-scale prediction of col- Nature. 2015;526(7572):241–4. https​://doi.org/10.1038/natur​
lision cross-section values for metabolites in ion mobility-mass e1538​8.
spectrometry. Anal Chem. 2016;88(22):11084–91. https​://doi. 46. Wu Q, Wang JY, Han DQ, Yao ZP. Recent advances in differentia-
org/10.1021/acs.analc​hem.6b030​91. tion of isomers by ion mobility mass spectrometry. TrAC Trends
31. Zhou Z, Tu J, Xiong X, Shen X, Zhu ZJ. LipidCCS: prediction in Anal Chem. 2020;124:115801. https​: //doi.org/10.1016/j.
of collision cross-section values for lipids with high precision to trac.2019.11580​1.
support ion mobility–mass spectrometry-based lipidomics. Anal 47. Domalain V, Hubert-Roux M, Tognetti V, Joubert L, Lange CM,
Chem. 2017;89(17):9559–666. https​://doi.org/10.1021/acs.analc​ Rouden J, Afonso C. Enantiomeric differentiation of aromatic
hem.7b026​25. amino acids using traveling wave ion mobility-mass spectrometry.
32. Bush MF, Campuzano ID, Robinson CV. Ion mobility mass spec- Chem Sci. 2014;5(8):3234–9. https:​ //doi.org/10.1039/c4sc00​ 443d​
trometry of peptide ions: effects of drift gas and calibration strate- .
gies. Anal Chem. 2012;84(16):7124–30. 48. Yu X, Yao ZP. Chiral differentiation of amino acids through binu-
33. Hines KM, May JC, McLean JA, Xu L. Evaluation of colli- clear copper bound tetramers by ion mobility mass spectrometry.
sion cross section calibrants for structural analysis of lipids by Anal Chim Acta. 2017;981:62–70. https​://doi.org/10.1016/j.
traveling wave ion mobility-mass spectrometry. Anal Chem. aca.2017.05.026.
2016;88(14):7329–36. 49. McCullagh M, Douce D, Van Hoeck E, Goscinny S. Exploring
34. Giles K, Ujma J, Wildgoose J, Pringle S, Richardson K, Langridge the complexity of steviol glycosides analysis using ion mobility
D, Green M. A cyclic ion mobility-mass spectrometry system. mass spectrometry. Anal Chem. 2018;90(7):4585–95. https​://doi.
Anal Chem. 2019;91(13):8564–73. https​://doi.org/10.1021/acs. org/10.1021/acs.analc​hem.7b050​02.
analc​hem.9b018​38. 50. Picache JA, Rose BS, Balinski A, Leaptrot KL, Sherrod SD, May
35. Tolmachev AV, Webb IK, Ibrahim YM, Garimella SV, Zhang JC, McLean JA. Collision cross section compendium to anno-
X, Anderson GA, Smith RD. Characterization of ion dynam- tate and predict multi-omic compound identities. Chem Sci.
ics in structures for lossless ion manipulations. Anal Chem. 2019;10(4):983–93. https​://doi.org/10.1039/c8sc0​4396e​.
2014;86(18):9162–8. 51. Plante PL, Francovic-Fontaine E, May JC, McLean JA, Baker
36. Vasilopoulou CG, Sulek K, Brunner AD, Meitei NS, Schweiger- ES, Laviolette F, et al. Predicting ion mobility collision cross-
Hufnagel U, Meyer SW, Barsch A, Mann M, Meier F. Trapped sections using a deep neural network: DeepCCS. Anal Chem.
ion mobility spectrometry and PASEF enable in-depth lipidomics 2019;91(8):5191–9. https​://doi.org/10.1021/acs.analc​hem.8b058​
from minimal sample amounts. Nat Commun. 2020;11(1):1–11. 21.
37. Jeanne Dit Fouque K, Ramirez CE, Lewis RL, Koelmel JP, Garrett 52. Bijlsma L, Bade R, Celma A, Mullin L, Cleland G, Stead S, Her-
TJ, Yost RA, Fernandez-Lima F. Effective liquid chromatography- nandez F, Sancho JV. Prediction of collision cross-section values
trapped ion mobility spectrometry-mass spectrometry separation for small molecules: application to pesticide residue analysis.
of isomeric lipid species. Anal Chem. 2019;91(8):5021–7. https​ Anal Chem. 2017;89(12):6583–9. https​://doi.org/10.1021/acs.
://doi.org/10.1021/acs.analc​hem.8b049​79. analc​hem.7b007​41.

13
 Journal of Analysis and Testing

53. Colby SM, Nunez JR, Hodas NO, Corley CD, Renslow RR. Deep by in silico prediction tools. Anal Chem. 2019;91(9):6321–8. https​
learning to generate in silico chemical property libraries and can- ://doi.org/10.1021/acs.analc​hem.9b012​18.
didate molecules for small molecule identification in complex 63. Trim PJ, Henson CM, Avery JL, McEwen A, Snel MF, Claude
samples. Anal Chem. 2020;92(2):1720–9. https:​ //doi.org/10.1021/ E, Marshell PS, West A, Princivalle AP, Clench MR. Matrix-
acs.analc​hem.9b023​48. assisted laser desorption/ionization-ion mobility separation-mass
54. Colby SM, Thomas DG, Nuñez JR, Baxter DJ, Glaesemann KR, spectrometry imaging of vinblastine in whole body tissue sec-
Brown JM, Pirrung MA, Govind N, Teeguarden JG, Metz TO, tions. Anal Chem. 2008;80(22):8628–34.
Renslow RS. ISiCLE: a quantum chemistry pipeline for estab- 64. Škrášková K, Claude E, Jones EA, Towers M, Ellis SR, Heeren
lishing in silico collision cross section libraries. Anal Chem. RM. Enhanced capabilities for imaging gangliosides in murine
2019;91(7):4346–56. https:​ //doi.org/10.1021/acs.analch​ em.8b045​ brain with matrix-assisted laser desorption/ionization and des-
67. orption electrospray ionization mass spectrometry coupled to ion
55. Mesleh M, Hunter J, Shvartsburg A, Schatz GC, Jarrold M. Struc- mobility separation. Methods. 2016;104:69–78.
tural information from ion mobility measurements: effects of the 65. Li H, Smith BK, Márk L, Nemes P, Nazarian J, Vertes A. Ambient
long-range potential. J Phys Chem. 1996;100(40):16082–6. molecular imaging by laser ablation electrospray ionization mass
56. Zhou Z, Luo M, Chen X, Yin Y, Xiong X, Zhu ZJ. AllCCS Web spectrometry with ion mobility separation. Int J Mass Spectrom.
server. https​://allcc​s.zhula​b.cn/. Accessed 25 April 2020. 2015;377:681–9.
57. Stephan S, Hippler J, Köhler T, Deeb AA, Schmidt TC, Schmitz 66. Causon TJ, Si-Hung L, Newton K, Kurulugama RT, Fjeldsted J,
OJ. Contaminant screening of wastewater with HPLC-IM-qTOF- Hann S. Fundamental study of ion trapping and multiplexing using
MS and LC+LC-IM-qTOF-MS using a CCS database. Anal Bio- drift tube-ion mobility time-of-flight mass spectrometry for non-
anal Chem. 2016;408(24):6545–55. https:​ //doi.org/10.1007/s0021​ targeted metabolomics. Anal Bioanal Chem. 2019;411(24):6265–
6-016-9820-5. 74. https​://doi.org/10.1007/s0021​6-019-02021​-8.
58. Regueiro J, Negreira N, Berntssen MH. Ion-mobility-derived 67. Zang X, Monge ME, Gaul DA, Fernandez FM. Flow injection-
collision cross section as an additional identification point for traveling-wave ion mobility-mass spectrometry for prostate-cancer
multiresidue screening of pesticides in fish feed. Anal Chem. metabolomics. Anal Chem. 2018;90(22):13767–74. https​://doi.
2016;88(22):11169–77. https ​ : //doi.org/10.1021/acs.analc​ org/10.1021/acs.analc​hem.8b042​59.
hem.6b033​81. 68. Zhang X, Romm M, Zheng X, Zink EM, Kim YM, Burnum-John-
59. Zhou Z, Shen X, Chen X, Tu J, Xiong X, Zhu ZJ. LipidIMMS son KE, Orton DJ, Apffel A, Ibrahim YM, Monroe ME, Moore
analyzer: integrating multi-dimensional information to support RJ, Smith JN, Ma J, Renslow RS, Thomas DG, Blackwell AE,
lipid identification in ion mobility—mass spectrometry based Swinford G, Sausen J, Kurulugama RT, Eno N, Darland E, Staf-
lipidomics. Bioinformatics. 2019;35(4):698–700. ford G, Fjeldsted J, Metz TO, Teeguarden JG, Smith RD, Baker
60. Tsugawa H, Ikeda K, Takahashi M, Satoh A, Mori Y, Uchino H, ES. SPE-IMS-MS: an automated platform for sub-sixty second
Okahashi N, Yamada Y, Tada I, Bonini P, Higashi Y, Okazaki Y, surveillance of endogenous metabolites and xenobiotics in bioflu-
Zhou ZW, Zhu ZJ, Koelmel J, Cajka T, Fiehn O, Saito K, Arita M, ids. Clin Mass Spectrom. 2016;2:1–10. https​://doi.org/10.1016/j.
Arita M. MS-DIAL 4: accelerating lipidomics using an MS/MS, clinm​s.2016.11.002.
CCS, and retention time atlas. bioRxiv. 2020:2020.02.11.944900. 69. Zhang L, Foreman DP, Grant PA, Shrestha B, Moody SA, Vil-
doi:10.1101/2020.02.11.944900. liers F, Kwak JM, Vertes A. In situ metabolic analysis of single
61. Goodwin CR, Fenn LS, Derewacz DK, Bachmann BO, McLean plant cells by capillary microsampling and electrospray ioniza-
JA. Structural mass spectrometry: rapid methods for sepa- tion mass spectrometry with ion mobility separation. Analyst.
ration and analysis of peptide natural products. J Nat Prod. 2014;139(20):5079–85. https​://doi.org/10.1039/c4an0​1018c​.
2012;75(1):48–53. https​://doi.org/10.1021/np200​457r.
62. Bijlsma L, Berntssen MHG, Merel S. A refined nontarget work-
flow for the investigation of metabolites through the prioritization

13