Targeting Modified Lipids during Routine Lipidomics Analysis using HILIC and C30 Reverse Phase Liquid Chromatography coupled to Mass Spectrometry

Lipids are important biomolecules in all biological systems and serve numerous essential cellular functions. The global analysis of complex lipids is very challenging due to the extreme diversity in lipid structures. Variation in linkages and positions of fatty acyl chain(s) on the lipid backbone, functional group modification, occurrence of the molecular species as isomers or isobars are among some of the greatest challenges to resolve in lipidomics. In this work, we describe a routine analytical approach combining two liquid chromatography platforms: hydrophilic interaction (HILIC) and C30 reversed-phase chromatography (C30RP) coupled to high resolution mass spectrometry (HRMS) as complementary high throughput platforms to analyze complex lipid mixtures. Vascular plants (kale leaves and corn roots), rat brain and soil microbes were used as proxies to evaluate the efficiency of the enhanced approach to resolve traditional, as well as, modified lipids during routine lipidomics analysis. We report for the first time, the observation of a modified class of acylphosphatidylglycerol (acylPG) in corn roots by HILIC, and further resolution of the isomers using C30RP chromatography. We also used this approach to demonstrate the presence of high levels of N-monomethyl phosphatidylethanolamine (MMPE) in soil microbes, as well as to determine the regioisomers of lysophospholipids in kale leaves. Additionally, neutral lipids were demonstrated using C30RP chromatography in positive ion mode to resolve triacylglycerol isomers in rat brain. The work presented here demonstrates how the enhanced approach can more routinely permit novel biomarker discovery, or lipid metabolism in a wide range of biological samples.

Lipidomics is the omics science concerned with the comprehensive identification and quantification of cellular lipid molecular species and their function in biological systems. Improvement in chromatography and mass spectrometry have contributed significantly to the development and rapid expansion of lipidomics as a sub discipline in metabolomics over recent years 1,2 . The high mass resolution and mass accuracy integrated in advanced mass spectrometers provides accurate measurement of the mass of precursor ions, which enables isobaric separation, as well as fragment ions to eliminate possible false positive identification 3 . Lipids comprise an essential part of all biological systems, and serve numerous structural and functional roles in these systems inclusive of providing structural molecules for forming cellular membrane bilayers 4 , act as signaling molecules in cell communications (e.g., diacylglycerols and ceramides) 5,6 , energy storage (e.g., triacylglycerols in adipose tissues) and transport 7 . Lipids can be classified mainly into eight categories, namely, fatty acids (FA), glycerolipids (GL), added. The sample mixture was vortexed, then added 0.8 mL of water and centrifuged at 5,000 × g for 15 min. Following centrifugation, the organic (bottom) layer containing the lipids was transferred to a pre-weighed 2 mL sample vial having a PTFE lined cap (VWR). The sample was dried under a stream of nitrogen, and the vial re-weighed to determine the amount of lipid recovered 36 . The recovered sample was then re-suspended in 1 mL chloroform:methanol (1:1 v/v) and stored at −20 °C for analysis by liquid chromatography coupled with high resolution tandem mass spectrometry.
Analysis of the lipid mixture using HILIC or C30-RPLC. The HILIC column (Luna 3 µm, 100 × 2 mm I.D., particle size: 3 µm, pore diameter: 200 Å) was purchased from Phenomenex (Torrance, CA, USA). The mobile phase system used was as follows: acetonitrile:water (97:3 v/v) containing 10 mM ammonium acetate (solvent A), and solvent B containing 10 mM ammonium acetate in pure water. HILIC separation was carried out at 25 °C (column oven temperature) with a flow rate of 0.2 mL/min, and 10 µL of the lipid extract suspended in chloroform:methanol (1:1 v/v) injected in the machine. The system gradient was maintained at 100% solvent A for 2 min; solvent B was then increased to 10% over 23 min, solvent B was increased from 10-15% over 10 min, and then kept at 15% B for 5 min. The column was re-equilibrated to starting conditions (100% solvent A) for 10 min prior to each new injection.
The Accucore C30 column (150 × 2 mm I.D., particle size: 2.6 µm, pore diameter: 150 Å) was obtained from ThermoFisher Scientific (ON, Canada). The mobile phase system consisted of solvent A (acetonitrile: H 2 O 60:40 v/v) and solvent B (isopropanol:acetonitrile:water 90:10:1 v/v/v) both containing 10 mM ammonium formate and 0.1% formic acid. C30-RPLC separation was carried out at 30 °C (column oven temperature) with a flow rate of 0.2 mL/min, and 10 µL of the lipid extraction suspended in chloroform:methanol (1:1 v/v) was injected onto the column. The following system gradient was used for separating the lipid classes and molecular species: 30% solvent B for 3 min; then solvent B increased to 43% over 5 min, then to 50% B in 1 min, then to 90% B over 9 min, then to 99% B over 8 min and finally kept at 99% B for 4 min. The column was re-equilibrated to starting conditions (70% solvent A) for 5 min prior to each new injection.
The method was applied in parallel as complementary system to take advantages of both (i) HILIC column to screen any novel lipid class present in the samples and (ii) C30-RPLC to confirm the HILIC results and provide further information on the modified lipids.
High resolution tandem mass spectrometry analysis. Lipid analyses were carried out using a Q-Exactive Orbitrap mass spectrometer controlled by X-Calibur software 4.0 (ThermoScientific, MO, USA) with an automated Dionex UltiMate 3000 UHPLC system controlled by Chromeleon software. The following parameters were used for the Q-Exactive mass spectrometer -sheath gas: 40, auxiliary gas: 2, ion spray voltage: 3.2 kV, capillary temperature: 300 °C; S-lens RF: 30 V; mass range: 200-2000 m/z; full scan mode at a resolution of 70,000 m/z; top-20 data dependent MS/MS at a resolution of 35,000 m/z and collision energy of 35 (arbitrary unit); isolation window: 1 m/z; automatic gain control target: 1e5. The instrument was externally calibrated to 1 ppm using ESI negative and positive calibration solutions (ThermoScientific, MO, USA). Tune parameters were optimized using a mixture of lipid standards (Avanti Polar Lipids, Alabama, USA) in both negative and positive ion modes.

Results and Discussion
Analysis of the complex lipid standard mixture using HILIC or C30-RPLC-HRMS/MS. We utilized a solvent system consisting of acetonitrile, water and ammonium acetate buffer for HILIC chromatography, which efficiently separated 15 classes of the lipids present in the complex lipid standard mix in negative ion mode (Fig. 1a). The resolution of the lipid classes (inter-class differentiation) based on their headgroup composition occurred in the following order: SQDG, PG, DMPE, P-PC, PC, MMPE, O-PC, P-PE, SM, PI, P-LPE, LPE, PA, DLCL, and finally LPA (Fig. 1a). In comparison to HILIC, a C30-RPLC coupled with HESI-MS in negative ion mode is shown in Fig. 1b for the same complex lipid standards mixture. It is noted here that the buffer conditions were different between the two platforms. Formate buffer was used in the C30RP method whereas acetate buffer was utilized in the HILIC method for column optimization. Therefore, some lipid classes form acetate adducts [M + CH 3 COO] − under HILIC-MS conditions, and formate adducts [M + HCOO] − when C30-RPLC-MS was performed. The observed mass (m/z) and ion types of each lipid standard from both chromatograms are included in Table S-1. Examples of the HILIC-HESI-MS and C30-RPLC-MS spectra are shown in Fig. 1c Fig. 1e,f with the characteristic ions for diagnosis of the inositol headgroup at m/z 241.01 and the two fatty acid composition m/z 283.26 (C18:0) and m/z 303.23 (C20:4). The fragmentation of PI was known to be more complicated and different from other phospholipid classes in which the intensity of sn-2 FA generated from PI is either relatively lower than or nearly equal to that of the sn-1 fatty acid ion 32,37 . The m/z 419.26 product ion arising from further inositol loss after initial C20:4 fatty acid loss (i.e. m/z 581.31) was occurring preferentially at the sn-2 position 37,38 . This preference was used to assign the positions of FA chains on the glycerol backbone of PI. Thus the regiochemical assignment was consistent with PI 18:0/20:4 from bovine liver included in the lipid standard mix. The HRMS/MS spectrum from HILIC ( Fig. 1e) was compared with that from C30RP in (Fig. 1f), and all the major fragment ions were present in the two MS 2 spectra as expected. HRMS/MS permitted the accurate assignment of the PI molecular species following both C30RP and HILIC chromatography.
We assessed the efficiency of HILIC chromatography in resolving representative modified polar lipids in a complex standard mixture. Optimization of the mobile phase gradient using an isocratic elution period www.nature.com/scientificreports www.nature.com/scientificreports/ consisting of 5% solvent B from 5 min to 15 min improved the modified lipid subclasses separation in the first half of HILIC chromatogram (Fig. 2a) Fig. 2d) and 28 Da for N,N-dimethyl phosphoethanolamine ion (m/z 168 for DMPE headgroup in Fig. 2e). The product ion at m/z 168 is also present in PC and SM mass spectra associated with de-methylation of the phosphocholine headgroup and can be differentiated from PE by a companion of CH 3 COOCH 3 loss (i.e., -74 Da observed in Fig. 2f  Application of HILIC chromatography to analyze modified (methylated) PE lipids in the membrane of soil microbes. The complex lipid profile of soil microbes was examined using the HILIC method in negative ion mode (Fig. 3a). We observed the presence of monomethylated PE (MMPE) occurred at significant levels along with isomeric forms of diacyl PE within the cell membrane of soil microbes living under cool climatic www.nature.com/scientificreports www.nature.com/scientificreports/ conditions in podzolic soils (Fig. 3). The HILIC chromatography is efficient in separating the two PE subclasses, which allowed for facile identification and relative quantitation. LC-MS/MS spectra were used to assign the molecular species composition of both the monomethylated and diacyl PE subclasses (Table 1).
For example, two isomers at m/z 700.49 were clearly resolve following HILIC chromatography (Fig. 3b). The assignments were made as follows: m/z 140.01 and 154.03 fragment ions were used to distinguish between PE and MMPE head groups. It is known for PE lipids that the pathways leading to the formation of the carboxylate anion and of the ions corresponding to a ketene loss are sterically more favorable at the sn-2 position under low-energy CID 37,40 . The modified monomethyl PE lipid in soil microbes was identified as MMPE 16:1/16:1 which produced only one carboxylate anion m/z 253.22 from C16:1 fatty acid (Fig. 3c). The sn-2 ketene loss of C16:1 forming m/z 464.28 ion (Fig. 3c) same as the minor ion observed in Fig. 3d.
Assignments for the major diacyl and methylated PE isomers found in soil microbes are listed in Table 1 along with their retention times. Extracted area counts were used for relative quantitation ( Table 1). As such, MMPE accounted for 18.73% of the total PE lipid class composition observed in soil microbes living under cool climatic conditions in podzolic soils. These results demonstrate potential applications of the enhanced HILIC method for the inclusion of methylated PE (modified PE) during routine lipidomics analysis of complex biological samples.  41 . HILIC chromatography was applied to assess silage corn root membrane lipid class composition when cultivated under cool climatic conditions (Fig. 4a). Three categories of lipids including phytosphingolipids (phytoSP), glycolipids and phospholipids (8 classes) were detected in the complex lipid mixture obtained from silage corn roots. The corn root membrane polar lipid classes were observed to consist of PG, PC, PE, LPC, PI, LPE, PA and LPA in which intra-class separation of individual molecular species can be partly resolved following HILIC-HRMS/MS analysis (see Fig. S1).
Using this approach, we observe the presence of modified PG in corn roots. PG lipids are known to have important correlation with the plant sensitivity toward chilling temperature 13 . The modified PG was identified as acylphosphatidylglycerol (acylPG) following HRMS/MS analysis (Fig. 4). Both the modified (acylPG) and the un-modified PG were observed in the silage corn roots as sodiated [M + Na-2H] − and deprotonated ([M-H] − ) adducts ( Table 2, Figs 4 and 5). It is important to note that no sodium buffer was added to the solvent system. Roots are known to have high levels of sodium from soil uptake, and appears to be the source of endogenous sodium that formed the adducts with PG 42 . The acylated modification occurred at the glycerol headgroup of PG, and were detected in pairs with the un-acylated form of PG. The modified headgroup resulted in acylPG being less polar than PG, and as such eluted before PG during HILIC chromatography. Thus, acylPG was observed to elute at 8.53 min immediately before the PG peak eluted at 10.67 min during HILIC chromatography (Fig. 4a). Confirmation of the identity of acylPG was done based on HRMS/MS fragmentation patterns (Fig. 4e,f) www.nature.com/scientificreports www.nature.com/scientificreports/ PG head group was identified by its deprotonated [M-H] − precursor ions (Fig. 4c,d) or sodiated adduct ions (Fig. 4e,f). The determination of fatty acyl chain positions on the glycerol backbone was based on relative abundance of the carboxylate anions and ketene losses arising from sn-1, sn-2 and sn-3′ positions (sn-3′ represents the acylated position at the phosphoglycerol headgroup). The mass spectral interpretation is in agreement with fragmentation of acylPG reported previously 12 . Of the total silage corn root PG and acylPG mass spectral intensity; 25.22% was from acylPG ( Table 2).     12 which reported that the relative abundance of fatty acid loss at sn-1 or sn-3′ is higher than that for the ketene loss (i.e. m/z 727.49 in Fig. 4d corresponding to C16:0 acid loss). Intriguingly, the same compound as sodiated adduct [M + Na-2H] − produced a more noticeable fragment ion (m/z 749.52 in Fig. 4f) for the loss of C16:0 fatty acid than for example the ketene loss and was assigned as the sn-3′ fatty acid. Thus, acylPG 50:2 in corn root membrane was assigned as acylPG 16 Fig. 4e,f). As discussed earlier, the position of the fatty acyl chains in PGs from corn roots may have significance in terms of indicating their biosynthesis pathway and thus identifying their positional isomers are targeted using chromatography.
Analysis of acylPG using both HILIC and C30RP approaches in Fig. 5 demonstrated the usefulness of using C30RP complementary to HILIC to resolve the acylPG isomers. HILIC resulted in all the acylPG molecular species eluting as one peak and were resolved into individual molecular species following separation using high resolution tandem mass spectrometry (Fig. 5a,c). On the other hand, the molecular species were separated based on chain length and unsaturation using C30-RPLC (Fig. 5b). For example, acylPG 52:4 in corn root was observed from HILIC-HRMS/MS in Fig. 5d to exist as two isomers of acylPG 16:0/18:2-(18:2) and 18:2/18:2-(16:0). The www.nature.com/scientificreports www.nature.com/scientificreports/ presence of both isomers was assigned from sn-3′ C16:0 and C18:2 fatty acid losses, represented by almost the same abundance of m/z 773.52 and 749.52, respectively (Fig. 5d). Both acylPG 52:4 isomers were also observed following C30-RPLC to be present at the beginning and end of the peak eluting at 24.51 min (Fig. 5b) and their corresponding C30RPLC-MS/MS spectra further confirming the results obtained using HILIC. Major differences between fragmentation of the two positional acylPG isomers are seen in Fig. 5(e,f) This result demonstrates the application of the combined LC platforms to permit the detection of a new lipid in silage corn root during routine lipidomics. We also observed that using the HILIC complementary to C30 reverse phase chromatography aided in the identification of this headgroup modified PG in corn roots, while the C30RP was able to confirm the identity of the modified lipids with further intra-class separation of acylPG molecular species. Acylated phosphatidylglycerol (acylPG) are rare occurrence in plants, with a few exceptions reported in oats (Avena sativa) 43 and Arabidopsis leaves 44 as potential response to climatic conditions. The occurrence of acylPG observed in corn roots may well be associated with cultivation condition under cool climatic conditions in Newfoundland.
Application of the HILIC chromatography to analyze regioisomers of lysophospholipids in kale leaf samples. We also investigated HILIC chromatography to resolve regioisomers of lysophospholipids during routine analysis of kale lipidome following cultivation in different natural media amendments 35 . Following HILIC chromatography, not only the kale LPA, LPE and LPC lipids were clearly separated from each other, but also the regioisomers (differ based on the position of fatty acid at the sn-1 or sn-2 positions on the glycerol backbone) within each class of lysophospholipids were resolved. The sn-1 regioisomers were observed to have longer retention times, and higher relative abundance in comparison to sn-2 isomers consistent with previous findings 45 . HILIC-HRMS/MS spectra of the isomeric ion pairs are shown in Fig. 6, indicating a higher relative abundance of fatty acid ions formed by the cleavage of fatty acyl chain at the sn-2 position compared to the ions formed at the sn-1 position allows the differentiation of (lyso)phospholipid regioisomers.
The separation of the lysophospholipid classes and their regioisomers using HILIC chromatography permitted unambiguous identification and relative quantitation of each individual species present in the sample. Relative quantification of each regioisomer was based on the peak area obtained from the extracted ion chromatograms www.nature.com/scientificreports www.nature.com/scientificreports/ (XIC) along with the assigned retention times. The sn-1 regioisomer of lysophospholipids (1-LPE and 1-LPC) in kale leaves were observed at significantly higher level as compared to their sn-2 counterparts (Fig. 6 and Table S -2). Although the application of HILIC chromatography for separating regioisomers of lysophospholipids have been reported previously, this is the first time this information is reported for kale lysophospholipid regioisomers. The main purpose for this section is to demonstrate the applicability of the enhanced HILIC approach to facilitate the analysis of regioisomers present in lysophospholipids, as well as the analysis of modified phospholipids during routine lipidomics.

C30-RPLC-HRMS/MS for the separation and identification of neutral lipids as a complementary platform to HILIC-HRMS/MS. One limitation of HILIC chromatography is the poor retention of
neutral (non-polar) lipids such as mono/di/triacylglycerols, cholesterol, free fatty acids, etc. These neutral lipids tend to co-elute very early close to the solvent front (i.e., void volume peak) during HILIC. Thus C30-RPLC was chosen for analyzing neutral lipids as a complementary approach to HILIC. C30-RPLC has been compared with other reverse phase chromatographic methods (C8 and C18) and has been demonstrated to be very efficient in separating neutral lipids, especially triacylglycerols with high selectivity and geometric peak shapes 24,25,28 . In addition, the combination of C30RP with high resolution accurate mass spectrometry offer superior capabilities in separating neutral lipid isobars and isomers present in biological samples. We applied the C30-RPLC approach to separate the neutral lipids present in our standard mix using positive ion mode as shown in Fig. 7.
Triglyceride (TG) molecular species in the standard mixed were clearly resolved form each other based on the fatty acid composition and chain length ranging from TG 24:0 to TG 48:0 (Fig. 7a). The TG molecular species  (Fig. 7i,j) (Fig. 7j). The glycolipids were also observed to be clearly resolved from the TG, DG, and ceramide species following C30-RPLC-MS/MS analysis (Fig. 7a-j). MGDG molecular species were observed to elute between 16.95 and 18.01 minutes followed by Cer d36:1 at 21.18 minutes, DG 38:4 at 21.76 minutes, Cer d42:2 at 21.61 minutes and O-DG at 24.70 minutes (Fig. 7b). The TGs were interspersed between these molecular species and were clearly resolved before or after these species and classes based on the chain length (Fig. 7a,b).
Applying this approach to rat brain samples (Fig. 8), we demonstrated that C30-RPLC-MS/MS was efficient in resolving brain neutral lipids consistent with previous reports in the literature using this (C30) and other reverse phase (C8 or C18) columns 24,48 . Rat brain triglycerides are presented as an example of the C30-RPLC technique www.nature.com/scientificreports www.nature.com/scientificreports/ to analyze neutral lipids in biological samples. High resolution of rat brain TG isomers was accomplished using the C30-RP separation (Fig. 8b-e). For example, the extracted ion chromatogram shown in Fig. 8b of m/z 896.77 observed in rat brain TG gave three clearly resolved peaks eluted at 25.10, 25.60 and 26.21 mins (Fig. 8b) corresponding to three isomers (Fig. 8c- 46 . Using this as a guide, the three [TG 54:6 + NH 4 ] + isomers were assigned as TG 18:2/18:2/18:2, TG 16:0/20:4/18:2 and TG 16:0/22:6/16:0 molecular species (Fig. 8c-e). The purpose of this section is to demonstrate that using C30-RPLC-HRMS/MS as a complementary technique to HILIC-HRMS/MS is a suitable approach to analyze neutral lipids, modified lipids, as well as, regioisomers (i.e., sn-positional isomers) of lipids during routine lipidomics of a wide range of biological samples (microbes, plants and animals).

Conclusion
In this work, we present the utilization of two enhanced high resolution liquid chromatography platforms coupled to high resolution accurate mass tandem mass spectrometry, for the analysis of modified lipids and regioisomers during routine lipidomics analysis. We also demonstrate that this platform is suitable for investigating the lipidome across different biological samples (animals, plants and microbes). HILIC separates lipids into classes including modified headgroups according to their polarity and electrostatic interactions. Regioisomers of sn-LPE, sn-LPC and sn-PI in kale leaves were separated and quantified using HILIC in the negative ion mode, while isomeric structures of neutral lipids, i.e., triacylglycerols in rat brain were demonstrated with C30-RPLC operated in the positive ion mode. In the current study, an enhanced HILIC method was demonstrated to be superior in resolving modified lipids present in a complex standard mixture. When we applied the enhanced approached during routine lipidomics to a range of biological samples we observed the following: (1) Discovery of modified PG (acylPG) in silage corn roots as potential response to the cultivation condition under cool climate. On the other hand, the optimized C30RP chromatography permitted excellent intra-class resolution (separation) of lipid isomers with different fatty acid composition or head group modification. The combination of both chromatography as complementary platform with high resolution tandem mass spectrometry allowed excellent resolution of lipid isomers or isobars, and therefore accurately distinguish di/triglycerides, plasmalogens and ether iso-forms of lipids in a diverse mix of biological samples. The work presented in this study demonstrated that one can use HILIC complementary with C30 reverse phase ultra-high performance liquid chromatography and high resolution accurate mass spectrometry as a high throughput lipidomic platform to analyze complex lipids (isobars, isomers, linkage or modified headgroup) in diverse biological systems. We believe this method could facilitate the inclusion and analysis of some complex lipid classes that are subject of more targeted lipid analysis into a more routine lipidomic analyses; and maybe of value to the general scientific community.