The seed of Begonia (B. grandis Dry subsp. Sinensis) was obtained from Dr. Yin Yi’s laboratory at school of life sciences, Guizhou Normal University, China. The seed of Begonia was stratified at 4°C for 7 d and then sown into peat soils (The Pindstrup Group, Denmark). The seeded pots were then transferred and grown in a phytotron under long-day conditions, 16 h of white light (80 μmol m⁻² s⁻¹) and 8 h of dark, with 60% relative air humidity at 22°C ± 1°C. The plants of Begonia were watered three times each week and fertilized weekly.

Fully extended leaves were detached from 173-d-old plants of Begonia. The detached leaves were placed on plastic square Petri dishes with three-layer Whatman filter paper at the bottom immersed in 10 ml of deionized water. The plastic square Petri dishes containing the detached leaves were subsequently incubated at control (22°C for 2 h) or heat stress (45°C for 2 h) conditions in dark.

Following control or heat treatment, five biological samples of detached leaves for control or heat treatment were taken, respectively. Each sample of detached leaves (1 g FW) was quickly frozen in liquid nitrogen and stored at -80°C. Total lipids were extracted according to the protocol (Bligh and Dyer, 1959) with modifications. The frozen leaves were ground in liquid nitrogen, and the resulting tissue pellets were re-suspended in 2 ml de-mineralized water followed by the sequential addition of 2.4 ml precooled methanol and 8 ml of methyl tert-butyl ether with vigorous mixing after adding each solvent. Ultrasound was applied for 20 min at low temperature water bath, and the resulted sample was kept at room temperature for 30 min. Organic and water phases were separated by centrifugation at 14000 g for 15 min at 10°C. The upper organic phase was transferred into a glass vial, dried under a stream of nitrogen gas at 37°C, and stored at −20°C.

Dried lipid extracts were reconstituted in 2 ml 90% isopropanol/acetonitrile with vigorous mixing and centrifuged at 14000 g for 15 min, and finally, 3 μl of the resulting sample was injected for analysis. Reverse phase chromatography was selected for LC separation using CSH C18 column (1.7 μm, 2.1 mm × 100 mm, Waters). Mobile phases were consisted of (A): acetonitrile–water (6:4, v/v) with 0.1% formic acid and 0.1 mM ammonium formate; and (B): acetonitrile–isopropanol (1:9, v/v) with 0.1% formic acid and 0.1 mM ammonium formate. The initial mobile phase was 30% solvent B at a flow rate of 300 μl/min. It was held for 2 min and then linearly increased to 100% solvent B in 23 min, followed by equilibrating at 5% solvent B for 10 min.

Mass spectra were acquired by Q Exactive Plus mass spectrometer (Thermo Fisher Scientific) in positive and negative mode, respectively. ESI parameters were optimized and preset for all measurements as follows: Source temperature, 300°C; Capillary Temp, 350°C; the ion spray voltage was set at 3000 V; S-Lens RF Level was set at 50%, and the scan range of the instruments was set at m/z 200–1800. “Lipid Search” is a search engine for the identification of lipid species based on MS/MS data. Lipid Search contains more than 30 lipid classes and more than 1,500,000 fragment ions in the database. Both mass tolerance for precursor and fragment were set to 5 ppm.

Lipid search software (Thermo Fisher, Waltham, United States) was used for peak identification, retention time correction, and automatic integration pretreatment. After editing, the data matrices were imported into SIMCA-P 13.0 (Umetrics, Umea, Sweden), centered on the mean, and proportionally adjusted to Pareto variance. The multifactor analysis was then carried out. Partial least square discriminant analysis and orthogonal projection latent structure discriminant analysis (OPLS-DA) were used to analyze the data. The variable importance in the projection (VIP) value of the OPLS-DA model (VIP ≥ 1) and independent sample t-test (p < 0.05) were used to screen differential metabolites. Hierarchical cluster analysis was performed using pheatmap package of R language.

Cellular membrane systems including plasma and chloroplast membranes are very sensitive and easy to be altered when subjected to adverse environmental conditions, such as high-temperature stress (Hou et al., 2016; Higashi and Saito, 2019). At high temperature, glycerolipids as the major constituents of membranes can be changed into a very dynamic remodeling process by adjusting the glycerolipid composition of the related membranes with keeping the integrity and optimal fluidity of the membrane systems (Higashi and Saito, 2019). The contents of diacylglycerols (DG) and triacylglycerols (TG) significantly increased in leaves of B. grandis in response to heat stress (45°C, 2 h) whereas the reduced content was measured for monoacylglycerols (MG; Figures 1A–C) with considerable variations among different glycerol lipid species (Figures 2A–C). Accumulated data from the published reports suggest that a cumulative effect of lipid remodeling has been marked by the decreases in unsaturation levels of membrane lipids occurring in response to high-temperature stress (Higashi et al., 2015; Higashi and Saito, 2019; Shiva et al., 2020). With respect to the levels of triacylglycerols containing less unsaturated fatty acids, such as oleic (18:1) and linoleic (18:2) acids, and/or saturated fatty acids, such as palmitic (16:0) acid, the heat-induced accumulations of TG (18:0/16:0/16:0), TG (16:0/16:0/18:1) were detected (Figure 3), suggesting that the significant increase in saturation index was driven by accumulation of these two TG species under heat stress in leaves of B. grandis. In addition to the increased levels of triacylglycerols containing less unsaturated and saturated fatty acids as mentioned above, a dramatic (26.7-fold) increase was detected in the level of TG containing highly unsaturated fatty acids, such as linolenic acid (18:3), in leaves of B. grandis when challenged with heat stress (Figure 3). Similar to our findings, this lipidomic remodeling patterns were also reported in Arabidopsis (Higashi et al., 2015; Mueller et al., 2015, 2017). Triacylglycerols are thought to act as buffers for the homeostasis of cellular acyl lipids with occurrence of a very dynamic recycling of 18:3 fatty acids in TGs (Hernandez et al., 2012). This suggests a possible sequestration of 18:3 acyl chains from plastidic and extraplastidic lipid species into TGs at high temperature. In combination with previous studies in other plant species, our findings suggest that fine-tuning the metabolic processes of glycerolipids is important for high-temperature stress acclimation in Begonia plants.

Phosphatidic acid (PA) has been identified as a new class of lipid mediators involved in regulating plant growth and developments, such as membrane tethering during cell development and structural effects on cell membranes (Zhang et al., 2004; Wang et al., 2006; Higashi and Saito, 2019). In addition to its structural roles, extensive studies have been focusing on PA serving as a lipid signaling molecule regulating the responses of plants to a variety of biotic and abiotic stresses, such as heat, chilling, drought, pathogen attacks, and hormonal signaling of abscisic acid (Zhang et al., 2004, 2019; Wang et al., 2006; Hou et al., 2016; Higashi and Saito, 2019; Li and Wang, 2019; Li et al., 2019b). In this study, heat stress resulted in a dramatic increase in the total amount of PA species in leaves of B. grandis in response to heat stress (45°C, 2 h; Figure 4A). This heat-induced accumulations were mainly attributed from the molecular species of PA, such as PA (18:3/18:2) and PA (25:0e) whereas the reduction in the levels of PA (21:4p) and PA (24:4e) was detected in response to heat stress (Figure 5A). Our results are in agreement with previous reports that heat stress induced a rapid accumulation of PA in leaves of Arabidopsis (Mishkind et al., 2009; Horvath et al., 2012; Higashi and Saito, 2019).

In heat-challenged leaves of B. grandis, 15 species of phosphatidylethanolamine (PE) were measured (Figure 5B). Heat stress also increased the total amount of PE (Figure 4B), and this increase was mainly driven by accumulation of three PE species including (31:2p), (20:0/18:3), and (40:5; Figure 5B). Inconsistent with our results, the total amount of PE was reduced in leaves of Arabidopsis (Higashi et al., 2015) and wheat (Narayanan et al., 2016b). In agreement with our results, heat stress led to a significant increase in the content of PE whereas heat treatment decreased the levels of 18:3-containing PE species (34:3-PE, 36:5-PE, and 36:6-PE) in turfgrasses and tall fescue (Su et al., 2009; Zhang et al., 2019). We also found that the levels of 18:2- and 18:3-containing PE species (36:4-PE and 36:10-PE) were reduced in response to heat stress (Figure 5B), which is consistent with the results reported in turfgrasses and tall fescue as mentioned above. With respect to the changes of phosphatidylserine (PS), heat stress-induced accumulations of PS were reported in a variety of plant species, such as wheat (Narayanan et al., 2016a), turf grasses (Zhang et al., 2019), and desert shrub (A. lentiformis of the Chenopodiaceae family; Li et al., 2015). In this study, the accumulation of PS was also measured in leaves of B. grandis in response to heat stress (45°C, 2 h; Figure 4C).

Phosphatidylcholine (PC) is one of the major structural components of cellular membranes and plays other critical roles in plant vegetative and reproductive developmental stages(Wang et al., 2006; Higashi and Saito, 2019;Nakamura, 2021; Yu et al., 2021). PC is required for cell proliferation and differentiation (Nakamura, 2021; Yu et al., 2021). Based on the previous studies, the content of the major phospholipids PC was changed in different patterns in a variety of plant species, which were not very consistent, probably due to dynamic turnover of the PC lipids and differences in plant sources and growth conditions (Nakamura, 2021; Yu et al., 2021). In this study, total amount of the PC species was found to decrease significantly in response to heat stress (45°C, 2 h) in leaves of B. grandis (Figure 4D). With respect to the detected 6 species of PC, 19:0/13:0-PC, 8:0e/8:0-PC, 12:0/22:3-PC, and 34:3-PC were found to decrease whereas the levels of 18:3-containing PC increased in response to heat stress (Figure 5C). In agreement with our findings, heat stress treatment was reported to increase the accumulations of 18:3-containing PC in leaves of Arabidopsis (Higashi et al., 2015; Mueller et al., 2017). Phosphatidylglycerol (PG) is mainly chloroplast-localized as one of the phosphorous-containing membrane lipids. As shown in Figure 4E, heat stress caused a significant reduction in total amount of the PG species in leaves of B. grandis. Ten of the PG species were measured, and the heat-induced reduction was mainly resulted from the molecular species of PG, such as 17:0/20:1-PG, 29:4-PG, 40:5-PG, 16:0/16:0-PG, 16:0/18:1-PG, 40:7-PG, 44:9-PG, 43:6-PG, and 9:0/10:4-PG whereas the level of 16:1/18:3-PG increased under heat stress (Figure 5D). Accumulated data suggest that the largest proportion of lipid species contain 36 and 34 acyl carbons, indicating two 18-C fatty acid chains or a combination of 16-C and 18-C chains when lipidomic analysis was performed in a variety of plant species (Devaiah et al., 2006; Okazaki and Saito, 2014; Higashi et al., 2015; Mueller et al., 2015, 2017; Yu et al., 2021). In contrast, the lipid species with 38 acyl carbons or more, containing one chain of 20–26 carbons (very long-chain fatty acids, VLCFAs) plus one 18-C chain are generally less abundant (Devaiah et al., 2006; Okazaki and Saito, 2014; Higashi et al., 2015; Mueller et al., 2015, 2017; Yu et al., 2021). As we found, the four decreasing species with VLCFAs are highly desaturated, such as 40:5-PG, 40:7-PG, 43:6-PG, and 44:9-PG, which may contain one chain of 20–26 carbons (Figure 5D). In general, a significant decrease was measured in the highly unsaturated PG species with VLCFAs, which could be involved in stabilizing membrane systems given the importance of desaturation levels of lipids in adjusting the membrane fluidity. It is worth noting that 36-C PG was not detected in this study, which is consistent with results reported previously in Arabidopsis (Welti et al., 2002; Devaiah et al., 2006). However, another report suggests that 36-C PG is present in Arabidopsis, albeit at low signal intensity(Higashi et al., 2015).

Phosphatidylinositols (PIs) are a class of cellular signaling lipids acting in cellular processes, including membrane trafficking, cytoskeleton organization, polar tip growth, and abiotic and biotic stress responses (Thole and Nielsen, 2008; Ischebeck et al., 2010; Hou et al., 2016; Meijer et al., 2017). As measured in leaves of B. grandis in response to heat stress (45°C, 2 h), the total amount of the PI species decreased dramatically in comparison with control condition (Figure 4F). There were 18 PI species detected with the heat-induced reduction in the levels of the PI species including 21:0/12:3-PI, 32:1e-PI, 20:1p/14:1-PI, 32:0p-PI, 31:1-PI, 39:5-PI, 36:2-PI, 38:5-PI, 36:1p-PI, 38:4p-PI, 32:3-PI, 34:3p-PI, 16:0/16:0-PI, and 32:4-PI and the levels of 41:2-PI, 42:7-PI, 30:1-PI, and 12:0e/20:3-PG increased under heat stress (Figure 5E). Based on the reported results, heat stress increased the total amount of the PI species in Arabidopsis (Higashi et al., 2015) and turf grasses(Zhang et al., 2019) whereas no significant difference was detected in the total amount of the PI species in wheat(Narayanan et al., 2016b). In comparison with the heat-induced accumulation patterns of the PI species, the plants of B. grandis showed specific patterns different from previous reports.

As bioactive molecules, lysophospholipids (LysoPLs) are usually the hydrolysis products of glycerophospholipids catalyzed by phospholipases at the sn-1 or sn-2 position of the glycerol backbone. The most common LysoPLs usually include lysophosphatidic acid (lysoPA), lysophosphatidylcholine (lysoPC), lysophosphatidylethanolamine (lysoPE), lysophosphatidylglycerol (lysoPG), and lysophosphatidylinositol (lysoPI; Higashi et al., 2015; Hou et al., 2016). Accumulation of LysoPLs was reported in a variety of plant species when challenged with abiotic and biotic stresses, such as freezing, wounding, pathogen infection, or the application of elicitors (Lee et al., 1997; Narvaez-Vasquez et al., 1999; Scherer, 2002; Viehweger et al., 2002; Welti et al., 2002; Wi et al., 2014). Under heat stress, no significant changes were detected in the contents of three lysophospholipid classes (lysoPC, lysoPE, and lysoPG) in two representative species of alpine scree plants (Saussurea medusa and Solms-Laubachia linearifolia; Zheng et al., 2011). In this study, total amounts of the most common LysoPLs species were detected to decrease significantly in leaves of B. grandis in response to heat stress (45°C, 2 h; Figures 6A–E). These LysoPLs species include two species of lysoPA (16:0 and 18:2), five species of lysoPC (16:0, 18:1, 18:2, 18:3, and 20:5), two species of lysoPE (16:0 and 18:2), three species of lysoPG (16:0, 16:1, and 18:2), and two species of lysoPI (14:0e and 16:0; Figure 7). LysoPLs are thought to be marker molecules representing instant responses to a variety of stress conditions given that they are usually the hydrolysis products of glycerophospholipids catalyzed by phospholipases in a variety of plant species when challenged with abiotic and biotic stresses(Wi et al., 2014; Higashi et al., 2015; Higashi and Saito, 2019).

Sphingolipids constitute up to 40% of those lipids making the plasma membrane (PM), forming a significant portion of the lipids present in higher plants (Cacas et al., 2016). In plants, sphingolipids are thought to be major components of plasma membrane, tonoplast, and endomembranes (Markham et al., 2013). It is generally accepted that sphingolipids accumulate in the outer leaflet of the PM and are basically characterized by a ceramide formed by a sphingoid base amide linked to an acyl chain (Markham et al., 2013; Carmona-Salazar et al., 2021). In this study, three molecular species of sphingolipid-related metabolites (So, sphingosine; Sm, sphingomyelin; Cer, ceramide) were detected in leaves of B. grandis and the total amounts of three molecular species were significantly reduced in response to heat stress (45°C, 2 h) in comparison with control condition (Figures 8A–C). Extensive studies suggest sphingolipid-related metabolites as potential signaling molecules in mammalian cells as well as in some studies with yeast. Previous work showed that sphingolipid-deficient strains of Saccharomyces cerevisiae are heat-sensitive, suggesting that sphingolipids are necessary for surviving heat stress(Wells et al., 1998). It was reported that sphingolipid long-chain bases act as signaling molecules that regulate growth and responses to heat stress in S. cerevisiae (Dickson et al., 2006). In yeast, phytosphingosine, thought to be a putative sphingolipid second messenger, acts as a mediator in regulation of heat stress signaling and activation of ubiquitin-dependent proteolysis via the endocytosis vacuolar degradation and 26 S proteasome pathways (Chung et al., 2000). Based on the accumulated findings, sphingolipids function as bioactive signals and play critical roles in a variety of physiological processes and environmental responses, including programmed cell death (Liang et al., 2003; Shi et al., 2007; Simanshu et al., 2014), pathogen-induced hypersensitive response (HR; Liang et al., 2003; Wang et al., 2008; Gan et al., 2009), ABA-dependent guard cell closure (Ng et al., 2001; Coursol et al., 2003; Worrall et al., 2008), host–pathogen interactions (Takahashi et al., 2009; Peer et al., 2010; Rivas-San Vicente et al., 2010; Sanchez-Rangel et al., 2015), and low-temperature stress responses (Cantrel et al., 2011; Dutilleul et al., 2012). According to the studies in yeast, sphingolipids could act as signaling molecules functioning in the physiological adaptations to heat stress in plants. In higher plants, how sphingolipids affect heat stress signaling and thermotolerance has remained unexplored (Markham et al., 2013; Batsale et al., 2021).