Polar lipidomic profile shows Chlorococcum amblystomatis as a promising source of value-added lipids

There is a growing trend to explore microalgae as an alternative resource for the food, feed, pharmaceutical, cosmetic and fuel industry. Moreover, the polar lipidome of microalgae is interesting because of the reports of bioactive polar lipids which could foster new applications for microalgae. In this work, we identified for the first time the Chlorococcum amblystomatis lipidome using hydrophilic interaction liquid chromatography-high resolution electrospray ionization- tandem mass spectrometry (HILIC–HR–ESI–MS/MS). The Chlorococcum amblystomatis strain had a lipid content of 20.77% and the fatty acid profile, determined by gas chromatography-mass spectrometry, has shown that this microalga contains high amounts of omega-3 polyunsaturated fatty acids (PUFAs). The lipidome identified included 245 molecular ions and 350 lipid species comprising 15 different classes of glycolipids (6), phospholipids (7) and betaine lipids (2). Of these, 157 lipid species and the main lipid species of each class were esterified with omega-3 PUFAs. The lipid extract has shown antioxidant activity and anti-inflammatory potential. Lipid extracts also had low values of atherogenic (0.54) and thrombogenic index (0.27). In conclusion, the lipid extracts of Chlorococcum amblystomatis have been found to be a source of lipids rich in omega-3 PUFAs for of great value for the food, feed, cosmetic, nutraceutical and pharmaceutical industries.

To assess the potential health benefits of Chlorococcum amblystomatis, the atherogenic (AI), thrombogenic (TI), hypocholesterolemic/hypercholesterolemic (h/H) and the PUFA n-6/ PUFA n-3 indices were calculated. The AI and TI of the lipid extracts were 0.5 and 0.3, respectively, while the h/H ratio was 1.4 and the n-6/n-3 ratio was 0.1.

Polar lipid composition of Chlorococcum amblystomatis. The polar lipid profile was revealed by
hydrophilic interaction liquid chromatography-high resolution electrospray mass spectrometry (HILIC-HR-ESI-MS and HILIC-HR-ESI-MS/MS) (Supplementary file S1). Using this approach, a total of 245 molecular ions, with a minimum of 350 molecular species, belonging to 15 classes of lipids were identified. Of these, 101 molecular ions were glycolipids (GL), 54 molecular ions were betaine lipids (BL), 89 molecular ions were phospholipids (PL), and 1 molecular ion was an inositol phosphoceramide (PI-Cer) (Tables 2 and 3). Figures explaining how the MS/MS data were interpreted can be found in the Supplementary Figures S1-S13. The m/z of the fragment ions used to identify each molecular ion and each molecular species can be found in Supplementary Table S1 and S2.
The PL classes identified included phosphatidylcholine (PC), lysophosphatidylcholine (LPC), phosphatidylethanolamine (PE) and lysophosphatidylethanolamine (LPE), identified in the HILIC-HR-ESI-MS analysis as the molecular ion [M+H] + (Fig. 1); and phosphatidylglycerol (PG) and phosphatidylinositol (PI) identified as the negative molecular ion, [M−H] − (Fig. 2). The only sphingolipid species found in Chlorococcum amblystomatis was a PI-Cer, which has been identified in the analysis on the negative ion mode. A total of 32 molecular ions of PC, 11 of LPC, 13 of PE, 9 of LPE, 22 of PG, 2 of PI and 1 of PI-Cer were recognized. Analysis of the MS/MS data allowed to recognize that each ion can correspond to several molecular species, with different combinations of fatty acyl chains, but with the same (C:N) (Supplementary Table S2). The molecular species identified per class were of 39 PC, 11 LPC, 14 PE, 9 LPE, 32 PG, 2 PI, and 1 PI-Cer.
As previously indicated, the identification of the lipid species was based on the mass accuracy, retention time and interpretation of MS/MS data. The interpretation of the MS/MS data of the molecular ions of the PC, LPC, PE and LPE species allowed to confirm their identity (Supplementary Table S2). This was done by confirming for each species the presence of the polar head groups, in particular by the presence of the product ion at m/z 184 (for PC and LPC) and fragment ions arising from the neutral loss of 141 Da (for PE and LPE) 16 The HILIC-HR-ESI-MS/MS spectra of the negative molecular ions belonging to the PG, PI and PI-Cer classes allowed to confirm the presence of the corresponding polar heads by observing for each species the product ions at m/z 171 (for PG) and m/z 241 (for PI and PI-Cer), and identify the fatty acyl chains by the presence of RCOO − carboxylate anions (for PG and PI) 16,[44][45][46][47][48] (Supplementary Table S2). The only PI-Cer detected was identified by the accurate mass measurement, the retention time, and the presence of the polar head at Among the PG molecular ions identified, a few oxidized PG species were found (Table 3), the most abundant being PG(34:2-OH), identified as PG(18:0-OH-16:2). All the oxidized PG species were identified based on exact mass measurements, retention time and MS/MS spectra analysis (Supplementary Table S3). In the MS/MS spectra, the presence of mass shifts of + 16 Da relative to the [M−H] − ions and of RCOO − carboxylate hydroxyl anions was indicative of the formation of hydroxy derivatives 49,50 . No product ions with oxidized polar head groups were observed 45 .
The betaine lipids identified in the HILIC-HR-ESI-MS data included the classes of monoacylglycerol-trimethylhomoserine (MGTS) and diacylglycerol-trimethylhomoserine (DGTS), and were identified in positive ion mode as [M+H] + ions (Table 4, Fig. 3). A total of 38 molecular ions of DGTS and 16 molecular ions of MGTS were identified, corresponding to 68 molecular species of DGTS and 16 molecular species of MGTS. In Table 1. Fatty acid profile identified in the total lipid extract of Chlorococcum amblystomatis by GC-MS. Values are expressed in relative abundance (%) and represent the mean of five analytical samples ± standard deviation (SD). SFA saturated fatty acids, MUFAs monounsaturated fatty acids, PUFAs polyunsaturated fatty acids, AI atherosclerotic index, TI thrombogenic index, (h/H) (hypocholesterolemic/hypercholesterolemic) ratio.  44,47 , and the fatty acyl composition was determined by observation of the fragment ions resulting from the neutral losses of the fatty acyl chains as an acid (-RCOOH) and ketene (-R=C=O) derivatives 44,47 (Supplementary Table S2). The most abundant ions in each class of betaine lipids were MGTS (18:4) and DGTS (34:4) assigned as DGTS (18:4-16:0) and DGTS (18:3-16:1) (Table 4). Interestingly, among the identified DGTS molecular ions, oxidized DGTS molecular ions were also found, the most abundant being DGTS (34:7-OH), identified as a combination of two molecular species of DGTS (16:4-OH-18:3) and DGTS (16:3-OH-18:4). These oxidized ions were identified based on exact mass measurements, retention times, and by analysis of MS/MS spectra (Supplementary Table S3). The MS/MS of the oxidized DGTS molecular ions showed a neutral loss of H 2 O and the product ions formed due to the loss of oxidized fatty acyl chains (acid and keto derivatives). Also, the product ion characteristic of betaine lipids at m/z 236 was present in all MS/MS spectra, therefore no product ions with oxidized polar head groups were observed.
The typical fragmentation observed in the HILIC-HR-ESI-MS/MS spectra of neutral glycolipids, as [M+NH 4 ] + ions, allowed to confirm the polar head group by the neutral loss of 197 Da (for MGDG) or 359 Da (for DGDG), which corresponds to the loss of the carbohydrate moiety combined with the loss of NH 3 44,47 . The assignment of the fatty acyl composition was corroborated by the presence of product ions corresponding to each fatty acyl group as acylium plus 74 [RCO+74] + ion 44,47 (Supplementary Table S2). On the other hand, the HILIC-HR-ESI-MS/MS spectra of the [M−H] − ions of the acidic glycolipids (SQDG and SQMG), when available, showed the product ion of the sulfoquinovosyl polar head group at m/z 225 and the fatty acyl composition was confirmed by the neutral loss of fatty acyl chains as carboxylic acid (-RCOOH) and by the presence of carboxylate RCOO − anions (Supplementary Table S2).   Table S3).
The relative quantification of the identified species was carried out as described in the methods section. Figure 6 shows the most abundant lipid species, identified as molecular ions, in the polar lipidome of Chlorococcum   LC-MS data was acquired using an internal standard for each class of phospholipids. LC-MS data were normalized against their assigned internal standard, to semi-quantify each molecular ion. Due to the lack of commercially available GL and BL standards, internal standards from classes with a retention time closer to those identified for GL and BL were used to normalize their respective data, as performed in our laboratory 45,[52][53][54] . The use of one internal standard is acceptable for semi-quantification because the ionization efficiency of polar lipids depends mainly on the polar head, while the length of the chain and the degree of unsaturation contribute little 55,56 . However, we would like to point out the limitations of our inter-class quantitation approach, because despite the similar retention time, the response factors for each class, even between phospholipids, tend to differ in ESI-MS, which prevents a robust and accurate quantification [51][52][53] . Normalized data were then used to semi-quantify the abundances of the identified molecular ions. For a clearer overview of the composition of the polar lipidome of Chlorococcum amblystomatis, we have summed up the relative percentage of all the detected lipid species belonging to glycolipids, betaines and phospholipids (Fig. 7A). According to the largest number of molecular ions, the highest cumulative levels were observed for glycolipids (51.9%), followed by betaine lipids (30.1%) and phospholipids (17.9%). The sum of the relative percentage of all lipid species belonging to each class of lipids was also calculated. The highest cumulative levels were observed for DGTS (27.6%) followed by SQDG (23.7%) (Fig. 7B).

Evaluation of in vitro anti-inflammatory and antioxidant properties of Chlorococcum amblystomatis lipid extracts.
The anti-inflammatory potential of lipid extracts from Chlorococcum amblystomatis were evaluated using a test kit for screening for inhibition of human COX-2. With 10 µl of an extract of 50 µg mL −1 (5 µg), we have observed an inhibition of 87.5 ± 0.1 of COX-2 activity (Fig. 8). These results suggest that the lipid extracts of Chlorococcum amblystomatis have anti-inflammatory potential.
The antioxidant potential of lipid extracts from Chlorococcum amblystomatis were evaluated using free radical DPPH • and ABTS •+ scavenging assays (Fig. 9), as described in the methods section. The results obtained for the DPPH assay revealed that the concentration of the extract resulting in a 40% inhibition (IC40) of DPPH • was at an estimated concentration of 226.81 ± 2.99 µg mL −1 . The average antioxidant activity (Trolox equivalent, TE) was 76.25 ± 1.02 Trolox µmol g −1 of lipid extract. In contrast, the lipid extract which resulted in a 50% inhibition

Discussion
Chlorococcum amblystomatis is a green microalga with great potential to be a sustainable resource of commercially important essential lipids 1,2,29 , with promising applications in several industrial sectors 32,33 . In our work, the lipid extracts represented 20.77 ± 0.57%, of the dry weight of the biomass, which is consistent with the lipid content reported for Chlorococcum (8.71-32.3%) 10,11,[18][19][20] . This wide range of lipid content determined by different studies is consistent with the variation in lipid content due to growing conditions 32,36,[38][39][40] and the use of different extraction methods 54 . The FA profile identified by GC-MS included as the most abundant FA: C16:0, C18:3 (n-3), C18:0, C16:4 (n-3), C20:5 (n-3) and C16:1-9, representing respectively 23%, 19%, 14%, 11%, 9% and 7% of the total FA pool. It is important to note that this is the first time that FA C16:4 (n-3) and C20:5 (n-3) have been reported in Chlorococcum amblystomatis. The values of relative abundance collected in the present work are different from those reported in the literature to Chlorococcum sp 32,33,38,39 . Such differences could result from differences in growth conditions 55 , different strains, different species, different derivatization methods or even GC detectors (FID or MS) 56 . Chlorococcum amblystomatis extracts were rich in omega-3 PUFAs, representing 43.2%, with a high contribution of the most abundant FA, C18:3 (n-3), C16:4 (n-3) and C20:5 (n-3). Omega-3 PUFAs are known to have a variety of health benefits, such as preventing chronic diseases, as they are associated with anti-inflammatory and antioxidant protection, benefits for the cardiovascular system, prevention of breast cancer, improvement of neurological capacities and visual development [57][58][59]      www.nature.com/scientificreports/ and n-6/n-3 (0.1). The AI and TI indexes are commonly used to assess the potential of the matrix to stimulate platelet aggregation, and the lower the indexes the more beneficial it is in reducing the prevalence of heart disease 60 , and have already been used to assess these benefits in microalgae, seaweeds and fish [61][62][63][64][65] . Our results show that these indexes of the lipid extract of Chlorococcum amblystomatis were similar to those observed in fish oils 66,67 . Compared to Spirulina platensis, Nannochloropsis gaditana, Nannochloropsis oculata and Porphyridium tricornutum, Chlorococcum amblystomatis had lower AI values. Their AI values ranged between 0.6 and 1.7 61 .
As for the TI index, Chlorococcum amblystomatis had lower values than Spirulina platensis, Nannochloropsis gaditana and Porphyridium Cruentum 61 , which were 1.       18,19 , this will favour its valorization for food and nutraceutical formulations, where they can be used as value-added ingredients, and also in the cosmetic industry, as they can be used to create moisturizing emulsions 69 . Interestingly, oxidized phosphatidylglycerols (PG) with bonded oxylipins have been identified in this study. Some studies have suggested the anti-inflammatory potential of oxylipins 70 , as reported for Nannochloropsis gaditana and Chlamydomonas debaryana 71 . Indeed, we observed a COX-2 inhibitory activity from the lipid extracts of Chlorcococcum amblystomatis, to which these oxylipins could have contributed. Consequently, extracts rich in polar lipids from Chlorococcum amblystomatis could constitute an interesting opportunity for the nutraceutical or pharmaceutical industries.
The  72 . However, BL remains poorly studied to date, and little is known about its bioactive potential.
The polar lipidome of Chlorococcum amblystomatis was particularly rich in glycolipids (MGDG, DGDG and SQDG) which represented approximately half of the polar lipids content (Fig. 7). There is an interest in characterizing GL because of their reported bioactive properties 16 . For example, SQDG of the microalga Porphyridium purpureum have been associated with antioxidant activity, due to their inhibitory effect on the generation of superoxide in peritoneal mononuclear cells 73 . GL with PUFAs have shown anti-inflammatory activity by inhibiting the release of nitric oxide by macrophages 20,21 . There is also a patent for MGDG with EPA to be used as an anti-inflammatory compound 74 . Another interesting application for glycolipids comes from the use of DGDG and SQDG from seaweeds as chemotherapy agents 22 . SQDG esterified in EPA appears to have an antiproliferative effect by inhibiting the key enzymes telomerase 23  Regarding sulfur-containing lipids, SQDG (34:3), assigned as SQDG (16:0-18:3), and SQDG (32:0), were among the most abundant SQDG identified in Chlorococcum amblystomatis, and were linked to anti-inflammatory 79 and antiviral activities 80 , respectively. Also, SQDG (16:0-20:4) and SQDG (16:0-18:4), identified in extracts of Chlorococcum amblystomatis, have been reported for their antiproliferative properties 81,82 . Finally, SQMG (16:0), although detected in low abundance, has been reported to have antimicrobial and antitumor activities 83 .
The polar lipids of Chlorococcum amblystomatis may contribute to the antioxidant activity observed in the DPPH and ABTS radical scavenging assays, as some of the reported polar lipids were previously associated with antioxidant activity 73 . Natural antioxidants are highly sought after for their biological effects 84 and their wide applications in the food and pharmaceutical sectors 85 . As such, microalgal biomass extracts have been widely explored regarding their potential antioxidant activity 86,87 . Ethanolic and aqueous extracts are mainly composed of phenolic compounds, while methanolic extracts are rich in polar lipids 88 . In a recent study, the DPPH assay was employed to evaluate antioxidant activity on 9 microalgae. The methanolic extracts of the different microalgae strains showed antioxidant activity, with percentage inhibition of DPPH ranging from 15 to 45% (IC15 to IC45; at 200 μg mL −1 extracts concentration) 86 . In this work, the dichloromethane:methanol extracts of Chlorococcum amblystomatis resulted in a percentage of inhibition of 36% at 200 µg mL −1 , displaying a value better compared to 7 strains of microalgae from the reported work.
COX-2 is an important component of inflammation, associated with pro-inflammatory activity, responsible for the production of prostaglandin E 2 89 . In the present work, the lipid extracts of Chlorococcum amblystomatis exhibit a COX-2 inhibiting activity. It is the first time that such a response has been described in Chlorococcum sp. However, recent work, also measuring COX-2 inhibition, compared the anti-inflammatory potential of aqueous and ethanolic extracts of two Tetraselmis sp. strains and a Skeletonema sp. strain 81 . The highest value of antiinflammatory activity (82 ± 2%) was measured for the ethanolic extract of Skeletonema sp. at a concentration of 1 mg mL -1 . Our results showed that the COX inhibition of dichloromethane: methanol extracts of Chlorococcum www.nature.com/scientificreports/ amblystomatis was 87.5 ± 0.1%, at a concentration of 50 μg mL −1 . These results are consistent with the hypothesis that Chlorococcum amblystomatis has a higher COX-2 inhibitory power than Tetraselmis sp. and Skeletonema sp. The high COX-2 inhibitory activity of Chlorococcum amblystomatis extracts results in promising antiinflammatory potential. Future studies should explore the use of lipid extracts of Chlorococcum amblystomatis in inflammatory cells to explore this anti-inflammatory potential.

Conclusions
The polar lipidome of Chlorococcum amblystomatis was characterized for the first time in this study. The Chlorococcum amblystomatis strain used revealed a high content of omega-3 PUFAs. PUFAs are associated with several health benefits, such as the prevention of cardiovascular disease. The HILIC-MS/MS lipidomic approach identified 245 molecular ions of polar lipids, in Chlorococcum amblystomatis, revealing to be a microalga rich in glyco-and betaine lipids. Some of the identified polar lipids have already been reported with biological activity, for example, DGTS (20:5-20:5), SQDG (16:0-18:3), MGDG (20:5-18:2) and DGDG (20:5-18:2) which were associated with anti-inflammatory activity. In addition, extracts rich in polar lipids had COX-2 inhibiting activity and antioxidant activity. In conclusion, due to its chemical, biochemical, bioactive, and health-promoting properties, the lipid extracts of Chlorococcum amblystomatis have been found to be of high value for application in food, feed, cosmetic, nutraceutical, and pharmaceutical applications.

Materials and methods
Reagents. HPLC  Portugal Chlorococcum amblystomatis 0066 CA was cultivated autotrophically in Guillard's F2 culture medium, whose composition was adapted to local water, using nitrates as the source of nitrogen 91 . Briefly, 5 L flask reactors were cultivated from 7 to 15 days, under continuous exposure to light. Five 5 L flask reactors were used to inoculate one 0.1 m 3 L outdoor Flat Panel (FP) reactor, which was later sequentially scaled as follows: 0.25 m 3 L to a 0.5 m 3 L to a 1 m 3 Flat Panels. Four of the later reactors were used as inoculum of a 10 m 3 photobioreactor (PBR). The reactor was operated for 21 days, exposed to the environmental light and temperature conditions, at an average temperature of 15.5 °C and light irradiance of 20.10 MJ m −2 day −1 . pH was maintained constant, 7.0-8.0, by pulse injections of CO 2 and the temperature was kept under 28 °C by a sprinkler-like irrigation system. After growing period, the biomass was industrially collected by centrifugation and further spray-drying.
Lipid extraction procedure. Lipid extraction was carried out using a mixture of dichloromethane: methanol solvents (DM) (2:1, v/v). The lipids were extracted from 25 mg of lyophilized Chlorococcum amblystomatis biomass using the solvent mixture. The suspension was centrifuged (Selecta JP Mixtasel, Abrera, Barcelona, Spain) at 2000 rpm for 10 min and the supernatant was collected in a new pre-weighed glass tube. This process was repeated four times until the extraction solvent lost the green colour. The combined supernatants were dried under a stream of nitrogen.
The Folch extraction method was used to the obtained dried supernatants 54,92 . The extracts were redissolved in 2 mL of dichloromethane, and 1 mL of methanol and 0.75 mL of Milli-Q water were added. The mixture was vortexed for 2 min followed by phase separation by centrifugation at 2000 rpm for 10 min. The organic phase was collected in a new pre-weighed tube, and the aqueous phase was reextracted with 2 mL of dichloromethane, two more times. The combined organic phases were dried under a stream of nitrogen and weighted.
Each series of extracts was repeated five times and the total lipid content was determined by gravimetry. The yield of lipids extracted from dry biomass extracts (DW) was calculated as follows (Eq. 1): COX-2 inhibition assay. A commercial cyclooxygenase (COX-2) inhibitory screening assay kit-Cayman test kit-701080 (Cayman Chemical Company, Ann Arbor, MI, USA)-was used to assess their anti-inflammatory potential 93  www.nature.com/scientificreports/ arachidonic acid (AA, C20:4 [n-6]) in the cyclooxygenase reaction. This assay was carried out according to the instructions described by the supplier of the assay kit, using an aliquot of test extract or DMSO. For this assay, lipid extracts of Chlorococcum amblystomatis (500 μg, 250 μg, 125 μg and 50 μg) were dissolved in DMSO and the final volume of reaction was 1000 μl. Positive and negative controls were provided by the assay kit protocol. The positive control used inactivated COX-2 enzyme, and negative control used the enzyme with 100% initial activity without any inhibitor. The assay was performed in three replicates. Interferences were considered by subtracting COX-2 inhibition from the blank assays. The results were expressed as a percentage of inhibited COX-2.

Polar lipidome analysis by hydrophilic interaction liquid chromatography coupled to high-resolution tandem mass spectrometry (HILIC-HR-MS/MS). The polar lipidome was determined
according to the methodology previously described 34,35 . The dried samples were dissolved in CH 2 Cl 2 to a final concentration of 1 ug uL -1 . From each sample, a volume of 10 µL (10 µg of lipid extract) was taken and transferred to an appropriate vial, followed by the addition of 86 µL of a solvent system composed of two mobile phases in a proportion of 90% v/v eluent B and 10% v/v eluent A and 4 µL of a mixture of internal standards (dMPC-0.02 μg, dMPE-0.02 μg, SM (17:0/d18:1)-0.02 μg, LPC-0.02 μg, dPPI-0.08 μg, dMPG-0.012 μg, dMPS-0.04 μg). The composition of eluent B was 60% v/v of acetronitrile, 40% v/v of methanol, and 5 mM of ammonium acetate and the composition of eluent A was 50% v/v of acetonitrile, 25% v/v of methanol, 25% v/v of water, and 5 mM ammonium acetate. The lipids were separated by hydrophilic interaction liquid chromatography (HILIC) using a microbore Ascentis Si column (10 cm × 1.0 mm, 3 µm; Sigma-Aldrich) and a high performance-liquid chromatography (HPLC) system (Ultimate 3000 Dionex, Thermo Fisher Scientific, Bremen, Germany) with an autosampler coupled to the Q-Exactive hybrid quadrupole Orbitrap mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). A 5 µL aliquot of each sample mixture was injected into the HPLC column, at a flow rate of 50 µL min −1 and a temperature of 35 °C. The following gradient was applied: 10% A (0-2 min), 10-90% A (2-15 min), 90% A (15-17 min). The mass spectrometer with Orbitrap technology was operated simultaneously in positive (electrospray voltage 3.0 kV) and negative (electrospray voltage − 2.7 kV) modes, with the following configuration: high resolution with 70,000, AGC target of 1 × 10 6 , capillary temperature 250 °C, and sheath gas flow of 15 U. The tandem mass spectrometry experiments were performed according to the following configuration: resolution of 17,500, AGC target of 1 × 10 5 , with one full scan mass spectrum and 10 data-dependent MS/MS scans. The cycles were repeated continuously throughout the experiments with the dynamic exclusion of 60 s and an intensity threshold of 1 × 10 4 . Normalized collision energy (CE) ranged between 25, 30, and 35 eV. Data acquisition was performed using the Xcalibur data system (V3.3, Thermo Fisher Scientific, USA). Five independent biological replicas were carried out.

Data analysis.
To identify the classes of the polar lipids in the lipid extracts acquired spectra were analysed using Xcalibur v3.3 (Thermo Fisher Scientific, USA). The identification was carried out according to a standard approach in our laboratory 98 : this approach consists of the localization of the species according to the retention time of internal standards (information related to internal standards can be found in Supplementary Table S1), accurate mass measurements (5 ppm) and identification of recurring fragmentation patterns (Supplementary Figures S1-S13) and their comparison with those of internal standards and published information on fragmentation patterns [98][99][100][101] . After identification, the quantification of molecular species was carried out by integrating the chromatographic peaks using the MZmine v2.42 software. The software allows filtering and smoothing, peak detection, peak processing and assignment against an internal database 102 . All peaks of raw intensity bellow 1 × 10 4 were excluded.
Relative quantification was performed by exporting the values of the peak areas to a computer spreadsheet (Excel, Microsoft, Redmond, WA). To normalize the data, the peak areas of the extracted ion chromatograms (EIC) of each lipid molecular species were divided by the EIC peak areas of the selected internal standards. Relative abundances were calculated using dMPC, dMPE, SM(17:0/d18:1), LPC, dPPI, dMPG and dMPS as internal standards. To normalize DGTS and MGTS we used PE as internal standard, for SQDG we used PG as internal standard, and for MGDG, DGDG, DGMG and MGMG, we used Ceramide as internal standard. For clear visualization, normalized data was transformed in percentage. The normalized data was calculated as follows (Eq. 7): The relative percentage of betaines, phospholipids and glycolipids was calculated as follows (Eq. 8): The relative percentage of the distribution of omega-3 and omega-6 fatty acids across the different main classes of polar lipids (betaines, phospholipids and glycolipids) was calculated as follows (Eq. 9): Raw abundances, normalized data and relative quantification can be found on the additional spreadsheet (Supplementary file S1).

Data availability
Raw datasets generated during this study are available from the corresponding authors upon reasonable request. (6)