Lipidomic analysis of moss species Bryum pseudotriquetrum and Physcomitrium patens under cold stress

Abstract Bryophytes, which lack lignin for protection, support themselves in harsh environments by producing various chemicals. In response to cold stress, lipids play a crucial role in cell adaptation and energy storage. Specifically, bryophytes survive at low temperatures by producing very long‐chain polyunsaturated fatty acids (vl‐PUFAs). The in‐depth understanding of the lipid response to cold stress of bryophytes was studied by performing lipid profiling using ultra‐high‐performance liquid chromatography‐quadrupole time of flight mass spectrometry (UHPLC‐QTOF‐MS). Two moss species (Bryum pseudotriquetrum and Physcomitrium patens) cultivated at 23°C and at 10°C were included in this study. Relative quantitative lipid concentrations were compared and the potential lipid biomarkers were identified by multivariate statistical analysis in each species. In B. pseudotriquetrum, it was observed that the phospholipids and glycolipids increased under cold stress, while storage lipids decreased. The accumulation of the lipids with high unsaturation degrees mostly appears in phospholipids and glycolipids for both mosses. The results also indicate that two unusual lipid classes in plants, sulfonolipids and phosphatidylmethanol are biosynthesized by the bryophytes. This has not been seen previously and show that bryophytes have a very diverse chemistry and substantially different from other plant groups.

point of the cells and keep the plant membrane fluidity together with other lipids (Tarazona et al., 2015). Although the lipid compositions of different moss species were reported before to contain high contents of vl-PUFAs, most of the studies only examined their free fatty acid profiles using conventional lipid analysis method such as thin-layer chromatography (TLC) or gas chromatography (GC; Dembitsky et al., 1993;Gellerman et al., 1975;Hartmann et al., 1986;Pejin et al., 2012). Recent development of lipidomics approaches allow a fast, high sensitivity and detailed molecular structural analysis of lipids. Herein, we performed plant lipidomics by ultra-high-performance liquid chromatography-quadrupole time of flight mass spectrometry (UHPLC-QTOF-MS) of two moss species to achieve a broader coverage of the moss lipidomes.
In this study, we present the lipid composition in two moss species, and report the changes in lipid composition under cold stress when cultivated at 23 and 10°C by using a lipidomics approach. Bryum pseudotriquetrum was selected for its relative fast growth rate at both 23 and 10°C in liquid culture. Second, Physcomitrium patens was examined as a model organism for studying non-seed plants (Rensing et al., 2020) and have been studied previously for it lipid composition (Girke et al., 1998).
In this study, the lipid composition in different species, B. pseudotriquetrum and P. patens, were investigated and the lipid molecular changes during cold stress were examined. Additionally, potential lipid biomarkers that are upregulated and downregulated under cold stress were identified, and the relative concentrations of identified lipid molecular species and the total amount of each lipid class were quantified. The changes and the impact of these are discussed. HPLC-grade chloroform and methanol, LC-MS-grade acetonitrile, isopropanol, ammonium acetate, tributylamine, and potassium chloride (KCl, purity≥99.0%) were all purchased from Sigma-Aldrich.

| Chemicals
Hexakis(2,2,3,3-tetrafluoropropoxy)phosphazene was purchased from Apollo Scientific Ltd, UK. Glass tubes with PTFE coated caps was used for the analysis (DWK Life Sciences, UK). Ultra-pure water was obtained from a Milli-Q system. Moss liquid culture was grown in Knop media (Reski & Abel, 1985).

| In vitro cultivation
Prior to the experiment, moss culture was blended in sterile water for 30 s and inoculation started with a biomass concentration of 100-300 mg/L. Two hundred milliliters of blended moss liquid culture was inoculated in 600-ml cell culture flasks (Corning) and kept on a cell culture rocker, and two biological replicates were grown.
In the standard condition, the moss liquid cultures were maintained in a growth chamber with a temperature of 23 ± 1°C, and the light intensity was kept at 15-20 Wm −2 as described by Pan et al. (2015).
Cold-stressed moss was kept in a growth chamber at 10 ± 1°C with the same light intensity as the standard condition. Day/night cycle was 16 h/8 h in both conditions. One hundred milliliters of biomass was harvested at day 21 from each biological replicate for extraction of moss at room temperature and the rest of the biomass was left at 10°C, and then harvested after another 24 h. The harvested biomass was filtered using Nylon cell strainers (pore size 70 μm) and kept at −20°C until extraction. As isotope-labeled lipid internal standards are not all commercially available, Welti et al. (2002) used hydrogenated MGDG and DGDG as internal standards for plant lipid profiling since those lipids do not exist in Arabidopsis thaliana. However, hydrogenated glycolipids are not suitable for using as internal standards for bryophytes because we have detected DGDG 32:0 (16:0_16:0) and SQDG 32:0 (16:0_16:0) produced by P. patens in our test run. Therefore, a deuterated glycolipid DGTS (d9) was used for glycolipids (MGDG, DGDG, and SQDG) quantification.

| Lipid extraction
Due to the low growth rate of mosses, the harvested moss material in biological replicates were combined in each growth condition, and the combined material was divided into three aliquots for lipid extraction. The fresh moss material was first ground in liquid nitrogen.
In each aliquot, 200 mg fresh frozen moss material was weighed into a glass tube; 3 ml chloroform/methanol (2:1, v/v) was added into the tube together with IS. The mixture was ultrasonicated for 20 min in dark, and then vortexed on a vortex mixer before adding 0.75 ml 1 M KCl. The mixture was vortexed again and centrifuged at 2000g for 5 min at 4°C. The organic phase was collected into a new glass tube by using a Pasteur pipette. The remaining mixture was washed twice with 1 ml chloroform, vortexed, and centrifuged; the organic phases were combined and evaporated under nitrogen stream and stored at −20°C until analysis. The dried lipid residue was re-suspended in 150 μl of reconstitution solvent (one portion of chloroform/methanol (1:1, v/v) and nine portions of isopropanol/acetonitrile/water (2:1:1, v/v/v)). The solution was transferred to an Eppendorf tube and centrifuged at 13,000g for 5 min at room temperature, 100 μl of the supernatant was transferred to an HPLC vial with a glass insert for analysis. Quality control (QC) was prepared by pooling 10 μl aliquots of all samples.

| Lipid quantification and recovery
To perform relative quantification of detected lipids, calibration curves of each IS were made with at least six concentration points (Table 1). Eight ISs were used for quantification in ESI+ mode and five in ESI-mode. The lowest quantitative concentration was 0.125 μg/ ml and good linearity was observed up to 40 μg/ml. All ISs had a R 2 higher than .99.
The recovery of the ISs was tested by comparing the peak areas of their corresponding m/z before (pre-spike) and after (post-spike) extraction procedure. For lipids that generate more than one form of ions, the most abundant ion form was chosen for relative quantification Most of the internal standards showed more than 85% recovery ( Figure S1). PI 15:0-18:1(d7) had with the lowest recovery in both ESI+ and ESI-mode (73% and 78%, respectively), as expected (Aldana et al., 2020).

| Data analysis
Agilent MassHunter Qualitative Analysis (B.07.00) was used for firsthand chromatogram visualization and for integration of the internal standards. The raw data files were imported directly to MS-DIAL (version 4.60) for further data analysis (peak peaking, deconvolution, compound identification, and alignment). Data normalization was achieved by calculating relative concentrations by using the internal standard representing the same lipid class. Parameters used for importing data is available at Data S1. The batch results used a set of criteria such as m/z error <0.01 Da compared to the theoretical mass, RSD < 30% of the replicates, and retention time deviation <0.05 min to ensure data quality (Broadhurst et al., 2018). The alignment result was imported to SIMCA 17 (Umetrics AB) for Principal Components Analysis (PCA) and Projection to Latent Structures with Discriminant Analysis (PLS-DA; Bruce et al., 2008;Eriksson et al., 2006Eriksson et al., , 2008. Data were pareto-scaled (to keep the original impact of the raw data) and log2-transformed (to correct the skewed distributions; van den Berg et al., 2006). T-tests were performed for calculating statistical significance of total lipid concentration changes between the moss species under different temperatures.

| Lipid compositions in B. pseudotriquetrum and P. patens
To review the lipid composition in B. pseudotriquetrum and P. patens, the identified lipids were classified accordingly to their lipid classes in ESI+ and ESI-mode, the lipid classes and the corresponding numbers of lipid metabolites are shown in Figure 1.

| Identification of biomarkers under cold stress
To identify potential lipid biomarkers from B. pseudotriquetrum and P. patens, chemometric approaches were applied by using unsupervised PCA and supervised PLS-DA. First, PCA plots were generated to visualize the group information and to monitor the quality of the data ( Figure S2). The two-moss species show clear separation and the quality control samples are clustered tightly together, indicating that the batch is of good quality. B. pseudotriquetrum shows higher inter-species variations between room temperature and cold temperature in ESI+ mode, this may indicate that B. pseudotriquetrum has a stronger response to cold stress than P. patens.
To discriminate the samples that belong to 23 and 10°C in each species, individual PLS-DA plots were built for B. pseudotriquetrum and P. patens for ESI+ and ESI-mode dataset, respectively ( Figure 2).
During this process it was also clear that an OPLS-DA did not fit the data, possibly due to the small numbers of samples, thus PLS-DA was chosen. All PLS-DA plots showed clear separation of two groups of samples with high cumulative X and Y matrix variations (R 2 X and R 2 Y, respectively) and high predictability (The second quartile, Q2).
As seen in Figure 2, the PLS-DA plots resulted in one predictive and two orthogonal components. A total X variance (R 2 X) of 0.449 can be explained, and the predicted variance R 2 Y was 0.961. The predictive ability Q2Y = 0.676, which indicates good predictability (Eriksson et al., 2006). To test the validity of the PLS-DA plots, permutation tests with 999 iterations were performed for all four PLS-DA plots.
The permutation tests are shown in Figure S2.  et al., 2007). Here all of the Q2 intercepts were below 0 but the R 2 are greater than 0.4, and the slopes of both R 2 and Q2 were larger than 0, thus, the PLS-DA plots can be considered to provide a good indication of the change in the lipid composition as a result of the cold stress.
The potential biomarkers for cold stress were screened based on three criteria (fold change of larger than 2 or smaller than 0.5, p < .05, and VIP >1). The lipid molecular species that fulfilled all three criteria were designated as biomarkers for cold stress (Figure 3).

Detailed lists for biomarker screening can be found in Data S2
and Figure S3a- showed significant increase in both moss species ( Figure S4). The presence of PMeOH is assumed that the choline head group from PC reacts with methanol when phospholipase D is active and generates PMeOH and free choline. However, we speculate the PMeOH is also produced endogenously in P. patens since the total amount of PC decreased while total amount of PMeOH increased (Figure 4, Data S3).

| DISCUSS ION
The major phospholipids, signaling lipids and storage lipids found in both moss species are common lipid classes also found in algae, higher plants, and other bryophytes (Chen et al., 2013;Conde et al., 2021;Okazaki & Saito, 2018;Vu et al., 2014), and the change observed here (Figure 4) is not unusual, though some for plants unusual lipids was found to be biosynthesized in bryophytes.

The unusual lipid class PMeOH detected in both moss species
has been reported to be an artifact from lipid extraction when using methanol (Roughan et al., 1978). The PMeOH lipid was identified by its characteristic m/z 110.981 (CH4OP-) in ESI-mode, which corresponds to the head group of a phosphatidic acid with a methyl group added to the phosphate, an example of the MS/MS spectrum of PMeOH 16:0_18:2 is shown in Figure S5. Tsugawa et al. (2019) analyzed the lipid compositions of nine algal species, but only detected several PMeOH lipids in one of them, Euglena gracilis, even if all algal species were extracted with the same method. In our study, the moss materials were kept at −20°C right after harvest, and were ground in liquid nitrogen to quench the lipid metabolism, before lipid extraction to limit methylation. A difference in the cold response is also seen for PC and PMeOH, for example, PC 16:0_18:2 decreased in cold stress in P. patens, whereas PMeOH 16:0_18:2 increased (Data S3). The different cold response of PC and PMeOH suggests that PMeOH is likely arise from a biosynthetic pathway in both mosses.
Another unusual lipid class, sulfonolipids (SLs, or N-acyl-capnine) again detected in both moss species, are structurally related to ceramides, but has a sulfonic acid group in the sphingoid base (Walker et al., 2017). is shown in Figure S5. SLs were previously only described in diatoms (Anderson et al., 1978) and some bacterial species (Walker et al., 2017). SL is likely produced by N-acylation of its precursor F I G U R E 2 PLS-DA score plot of B. pseudotriquetrum at room temperature and under cold stress in ESI+ mode (R 2 X = 0.449, R 2 Y = 0.997, Q2 = 0.676) and in ESI-mode (R 2 X = 0.728, R 2 Y = 0.961, Q2 = 0.804), P. patens at room temperature and under cold stress in ESI+ mode (R 2 X = 0.768, R 2 Y = 0.986, Q2 = 0.693) and in ESI-mode (R 2 X = 0.893, R 2 Y = 0.969, Q2 = 0.740) capnine with fatty acids (Godchaux & Leadbetter, 1980). The presence of SLs may again suggest that bryophytes have different biosynthetic pathways for lipids than vascular plants. Alternatively, the SL could arise from a bacterial contamination in the liquid culture though we did not observe this in our previous study by using nonaxenic moss materials (Lu et al., 2021).
Among the more common lipids, considerable amounts of vl-PUFAs were detected in both moss species, including FA's longer than 20 carbons as seen in previous studies (Beike et al., 2014;Girke et al., 1998), that also described 24:0, 25:0, and 26:0 FAs in several moss species. It is known that vl-PUFAs provide freezing tolerance for the mosses (Glime, 2017;Hansen & Rossi, 1991;Hartmann et al., 1986;Lu et al., 2019), one may expect to find higher PUFAs . The x-axis shows the log 2 FC (fold change) and the y-axis represent the -log 10 of the p-values. The red color represents the down-regulated significant changes in cold stress (VIP > 1, p < .05, FC < 0.5), while the green color represents the up-regulated significant changes in cold stress (VIP > 1, p < .05, FC > 1). The blue color shows the variables of VIP > 1, but p > .05. For full annotation please see Figure S3a-d patens (Figure 4), which indicates the TG breaks down in the moss cells when exposed to cold stress (Chen et al., 2013), and use the FAs released from TG for synthesis of phospholipids and glycolipids.
Resemann (2018) reported a slight increase of TG in wild type P. patens under cold stress at 4°C. However, they have conducted a much longer stress period (7 days), we therefore assume the TG synthesis undergoes a decrease when exposed to acute cold stress (24 h

| CON CLUS ION
Using mass spectrometry-based lipidomic approach, we identified bryophytes to react to cold stress.

ACK N OWLED G M ENT
This study is funded by Marie Sklodowska-Curie Actions, Innovative Training Networks under European Union Horizon 2020 programme under grant agreement No. 765115-MossTech.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data that support the findings of this study are available at DTU data repository, Lu et al (2022)