Most untargeted IMS techniques exist in a nested timescale between LC and MS such that a comprehensive annotation of retention time, drift/arrival time, and m/z can often be collected simultaneously [15]. The caveat, however, of the LC- IMS- MS workflows is the time expense of LC as a majority of lipid gradients require 20–40 minutes per injection. This is confounded in lipidomics as lipids are com- monly analyzed in both the positive and negative modes, thereby requiring approxi- mately an hour of instrument time for each sample. Shotgun lipidomics has become a prominent, high- throughput lipidomic analysis technique where extracted lipi- dome aliquots are directly injected into the mass spectrometer [106]. Conversely, MS imaging (MSi) prioritizes the annotation of an analyte’s spatial distribution in heterogeneous tissue and/or cell types, which is also incompatible with chromato- graphic separations [107]. As both these techniques solely collect MS information, the greatest weakness of both shotgun and MSi is the depth of lipid speciation achievable. For shotgun lipidomics, assigning MS/MS spectra can be impeded by
the prevalence of shared structural motifs, the diverse biological abundances of lipids, and the propensity of isomeric and isobaric species. MSi lipid identifications are even more challenged by the limited sample preparation capabilities needed to isolate lipids from other biomolecular classes. Additionally, MS/MS data collection in MSi applications is non- trivial, and challenges occur when spatially aligning the alternating MS and MS/MS scans causing a majority of imaging experiments to be collected in the MS- only mode [107]. However, the timescale of IMS readily allows it to be integrated into both shotgun and imaging workflows. Given that these two lipidomic techniques suffer from similar obstacles, we have combined the discus- sion of how IMS can be beneficially integrated into both these applications.
IMS can facilitate the reduction of spectral complexity through the partitioning of interfering chemical signals. For MSi, this capability is especially beneficial given the diversity of biomolecular classes represented in tissue before any extraction.
Figure 6.8a,b depicts an example spectrum of differentiating isobaric peptide and protein species from lipid signals based on unique mass versus mobility trends [108].
Additionally, DMS integration into the shotgun lipidomic analyses has facilitated the distinction of isobars arising from ether versus diacyl linkages in addition to PC and sphingomyelin (SM) lipids [109]. The mass versus mobility trendlines can also be used to identify charge states and biomolecule classes, as evidenced by Figure 6.8c, and the shotgun annotation of wax and sterol esters that were used to develop a meibum lipidome library [108, 110]. Reduced spectral complexity with IMS also enhances S/N ratios that can facilitate the annotation of low- level lipids. A desorp- tion electrospray ionization (DESI) imaging experiment comparing lipidome cover- age with and without FAIMS shows an increase in the S/N ratio of ~50% and the annotation of numerous cardiolipin species that were otherwise indifferentiable from background noise (Figure 6.8d) [108]. Additionally, MALDI- TIMS MS work has demonstrated improvements of peak capacity by 250% for imaging experiments by integrating IMS into the analyses [94]. The annotation of CCS also takes drift time separations a step further as this descriptor normalizes the separation and with m/z information can be used to filter data and increase confidence for analyte iden- tification. Integration of CCS with m/z has been shown to facilitate lipid identifica- tions by making mass accuracy criteria less stringent and providing another dimension to assess data. A shotgun lipidomics study by Paglia et al. demonstrated how the addition of a CCS molecular descriptor can limit both false- negative and false- positive lipid identifications [100]. SLIM has also aided in ganglioside isomer distinction given its heightened resolution relative to commercially available DTIMS and TWIMS systems [111]. Taken together, the integration of IMS into imaging and shotgun lipidomic workflows can facilitate accurate lipid identifications by reduc- ing spectral complexity and enhancing the number of descriptors being investigated while still maintaining the benefits of high- throughput shotgun lipidomics and the spatial information of imaging MS workflows.
6.1.5.2 IMS- MS/MS and Novel Speciation Approaches
As mentioned previously, the orthogonal collection of LC, IMS, MS, and MS/MS information provides a multitude of descriptors to increase the analyte
6 Ion Mobility Spectrometry 170
identification confidence. A sterol- omics study, for example, demonstrated that
~80% of derivatized isomers could be separated with LC and IMS together, while IMS or LC separations could only resolve 60% and 40% of isomers alone [112]. For other lipids such as phospholipids, sphingolipids, and glycerolipids, shared fatty acyl and head group moieties and overlapping mass regions preclude the confident assignment of fragments with either data independent acquisition (DIA) or data dependent acquisition (DDA) workflows, even with chromatography. In traditional LC- MS/MS workflows, DDA has become a common fragmentation method where ions are selected based on specific targets or abundance and then isolated and
Separation of PC lipid from isobaric peptide ion.
Separation of lipid species from isobaric protein ions. Separation of cardiolipins from isobaric lipid species.
Improved S/N by ~50% and detection of new species.
Improved identification and spatial determination.
4
3
2
1
00 200 400 600
Drift time (ms)
m/z800 m/z 788.5
MS at DT 1.89 ms
protein ions MS at DT 2.79 ms lipid ions
Relative abundance100%
(d) 0%
(c)
(b) (a)
m/z 1224.5
1000 1200 1400
Without FAIMS
With FAIMS
m/z 761.998
CL 78 : 12 m/z 749.998
CL 76 : 10 747.521 PG 34 : 1
713.494 CL 70 : 4
724.484 CL 72 : 7
737.494 CL 74 : 8
749.495 CL 76 : 10
761.495 CL 78 : 12 773.494
CL 80 : 14 784.488 CL 82 : 17
788.547 PS 34 : 1
788.547 PS 40 : 6 774.546 PEP-40 : 6
700 710 400730
420 440 460 480 500 520 540
740 750 760
Arrival time distribution (us)
m/z
770 780 790
720 730
Relative intensity (%)
740 750 760
m/z 770 780 790 800 774.546 PEP-40 : 6
Lipid imaging MALDI + DTIMS
LAESI + TWIMS DESI + FAIMS
DESI + TWIMS
Lipids separation based on head group, size, and charge.
Enhanced ganglioside imaging and data simplification.
100
0
Relative abundance (%)
Improved imaging accuracy
[Phosphatidylcholine(PC) 34 : 2+H]+
[Phosphatidylcholine 34 : 2+H]+
[RPPGFSP+H]+
Drift time (Bins) 1
2 4 3
m/z 917.5 [GD1 (d18 : 1/18 : 0)–2H]2–
m/z 1063.5 [GT1 (d18 : 1/18 : 0)–2H]2–
m/z 708.6 [GT1 (d18 : 1/18 : 0)–3H]3–
m/z
PC 34 : 2
RPPGFSP
RPPGFSP PC 34 : 2
Figure 6.8 Imaging IMS lipidomic applications. (a) MALDI- DTIMS- MS demonstrates the utility of IMS in imaging for the deconvolution of isomeric lipid signals. (b) DESI- TWIMS- MS and (c) Laser ablation electrospray ionization (LAESI)- TWIMS- MS illustrate drift time versus m/z relationships that help eliminate spectral complexity in biological samples, facilitating deeper lipidome coverage. (d) DESI- FAIMS- MS showcases cardiolipin identification by improving the S/N ratio with mass versus mobility selection capabilities. Source: Adapted with permission of Sans et al. [108], Elsevier.
fragmented. Conversely, in DIA, all ions or specific m/z windows of ions are frag- mented concurrently to produce extremely complex MS/MS spectra. With IMS inte- gration prior to fragmentation, the drift time separation allows for the deconvolution of MS and MS/MS information for enhancing the confidence in identifications as the precursor and product ions are drift time aligned (Figure 6.9) [113]. To imple- ment IMS- DIA workflows, a series of alternating scans of MS and MS/MS are typi- cally collected as ions elute from the IMS cell. The resulting spectral clean- up leveraging IMS has made IMS- DIA a common practice for data collection. In the HDMS Waters system, a TWIMS can be operated in the MSE mode such that co- eluting molecules are separated in the TWIMS cell and then fragmented using time- aligned parallel (TAP) fragmentation for cleaner MS/MS spectra [114, 115].
Additionally, the trapping capabilities of the tims TIMS cell in a Bruker tims TOF allow for a number of isolation, accumulation, and/or fragmentation options using their PASEF TIMS operation [53]. To optimize fragmentation efficiency on an Agilent DTIMS system, All Ions DIA can be employed similar to the TWIMS HDMS on a Waters platform [116]. Agilent DTIMS systems have also demonstrated the util- ity of ramping collision energy for collision induced dissociation (CID) such that ions with later drift times and therefore greater size experience collisions with more energy to facilitate fragmentation [117, 118]. Instrument modifications have also been used to allow for quadrupole ion selection on the Agilent system, further
DDADIA
1. Precursor survey scan (MS)
After 5 MS/MS experiments
Find fragments with same drift time
High CE frame 4D feature finding
Mass/charge ratio Chromatogram
m/z Time
Intensity
Intensity Drift time
m/z
Drift time
m/z m/z
Drift time
Mass/charge ratio
Intensity
2
3 4 5
1
Frame 1 No CE
Frame 2 High CE
2. Isolation and fragmentation (MS/MS) Consecutive isolation of 5 precursors
Figure 6.9 Example of MS/MS collection strategies. On top, DDA is used to selectively fragment ions based on their intensity for compound annotation. On bottom, DIA is used to fragment all ions regardless of intensity. By having the drift region prior to fragmentation, IMS- DIA allows for the separation of precursor signals by observed drift time to clean up the MS/MS spectra. Product ions can then be drift time aligned with precursor signals to map fragments to the correct precursor molecule. Source: Figure reproduced with permission from Nys et al. [113].
6 Ion Mobility Spectrometry 172
enhancing the sensitivity and S/N ratio in a fragmentation strategy similar to DDA [119].
IMS platforms have also been integrated with several fragmentation and/or deri- vatization strategies to enhance lipid speciation capabilities. Ozonolysis, for exam- ple, is a reaction- based annotation of double- bond location that has been implemented with IMS to facilitate the alignment of double- bond fragments for enhancing speciation [120]. Similarly, a Paterno–Büchi reaction that also cleaves double bonds was integrated with a supercritical fluid chromatography (SFC)- IMS separation on a Waters HDMS system where TAP fragmentation facilitated the assignment of fragments for over 500 glycerolipids and 30 sterols [121]. Electron impact excitation of ions from organics (EIEIO) is a novel fragmentation strategy that leverages an electron beam to fragment positive mode ions to identify sn- position and cis/trans isomers for specific lipid classes [122]. Recently, EIEIO frag- mentation has been integrated with DMS and an isopropanol modifier to facilitate isomer separations [123]. Together, these workflows are significant steps toward annotating full lipid speciation. However, a major caveat of these techniques is that the resulting data are extensive, and software to facilitate data processing is limited.
Therefore, these platforms have largely focused on proof- of- concept work with standards or example case studies that would be challenging to replicate for larger numbers of complex samples. As this data analysis becomes less manual, the inte- gration of IMS with the novel lipid speciation strategies is expected to greatly expand annotation capabilities.