1 INTRODUCTION

The importance of virus infection in controlling phytoplankton population and bloom dynamics is increasingly being elucidated (Brussaard, 2004; Vardi et al., 2009; Rosenwasser et al., 2019). In marine environment, the annual Emiliania huxleyi spring blooms are routinely terminated by the infection of a specific giant double-stranded DNA virus, the E. huxleyi virus (EhV), which belongs to coccolithoviruses (Bratbak et al., 1993; Schroeder et al., 2002; Rosenwasser et al., 2016), and the fate of E. huxleyi blooms may have a critical impact on carbon and sulfur biogeochemical cycles (Sheyn et al., 2018; Vincent et al., 2021). In the course of infection, virus may control the metabolic process of host to facilitate the production of virus progenies, resulting the disruption and changes in cytological, physiological and biochemical aspects. For example, lytic EhVs infection can restructure the metabolism of fatty acids and the synthesis of highly saturated triglycerides (TAG), coupled with a remarkable accumulation of neutral lipids within lipid droplets in E. huxleyi cells (Malitsky et al., 2016; Zeng et al., 2019). Interestingly, EhVs can not only be regarded as passive consumers of host metabolic products, also introduce new virus-encoded auxiliary metabolic genes (vAMGs) to manipulate even rewire the host sphingolipid metabolic networks, generating a unique metabolic state that fulfills their high metabolic requirements during infection (Bidle et al., 2011; Breitbart et al., 2012; Rosenwasser et al., 2014; Ziv et al., 2016). Recent studies highlighted that sphingolipid metabolism had been considered to be greatly important to viral infectivity, production of the virion membranes and assembly (Evans et al., 2009; Rosenwasser et al., 2014, 2016, 2019). The virus-encoded metabolic pathway gives rise to the production of unique virusderived glycosphingolipids (vGSLs), which are the major component of the EhV membrane structure (Rose et al., 2014). Besides its structural role, vGSL acts as a signaling molecule, which can induce a host programmed cell-death (PCD) during the lytic phase (Vardi et al., 2009; Liu et al., 2018). Also, vGSL can induce the biosynthesis of highly saturated triacylglycerols (TAG) (Bidle et al., 2007; Malitsky et al., 2016). Although a wealth of genomic resources of E. huxleyi and its specific EhVs is now available, a deep understanding of virus-encoded metabolic genes and their functional role in rewiring host sphingolipid metabolism is still limited.

The ability to define specific metabolic states by metabolomic signatures greatly facilitates the use of metabolic fingerprints as biomarkers to assess active viral infection in the ocean. Sphingolipidomics is a valuable emerging tool for understanding the structure-specific of all the individual molecular species of sphingolipids, which can provide clues for promoting a perfect, delicate and worldwide map of sphingolipids metabolism (Sakamoto et al., 2005; Merrill et al., 2008; Haynes et al., 2009). Furthermore, detection of unique virocell-derived molecules can provide novel insights into the employed vAMG-resulting metabolic strategy. It is important, therefore, to understand the dynamic changes and regulation of host sphingolipids metabolism during virus infection (Liang et al., 2003).

In this study, we performed lipidomics analysis on cultures of uninfected and EhV99B1 infected E. huxleyi BOF92 strain at different time points, by using ultra-performance liquid chromatography coupled to mass spectrometry (UPLC-MS) to explore the composition and content of sphingolipids during viral infection, providing a comprehensive insight into the host-virus biotic interaction mechanism and enabling a deeper understanding of global host-virus biotic interactions mechanism.

2 MATERIAL AND METHOD 2.1 Algal culture and viral infection

Emiliania huxleyi BOF92, isolated from the West Coast of Norway was cultured in f/2-Si medium at 18±1 ℃ and a 14-h꞉10-h light꞉dark cycle. Light was supplied by cool white LEDs and the light intensity was proximately 60 μmol photons/(m²·s). Viral infection was performed with exponential-phase cultures of E. huxleyi (~6×10⁵ cells/mL), and a 1꞉50 volumetric ratio of an EhV 99B1 lysate was added to the cultures (multiplicity of infection of ~1꞉1 viral particle per cell).

During viral infection (0–45 hpi), samples (2 mL) for algal and viral counts were fixed with 1.0% and 0.5% glutaraldehyde (final concentration) respectively, frozen in liquid N₂, and stored at -80 ℃ for flow cytometry analysis. After viral infection 6 and 45 h, 300-mL cell culture was collected by centrifugation at 2 800×g for 5 min at 4 ℃, while the control cultures (virus-free) took the same treatment. Each condition was set up five parallels and the algal cell pellets were lyophilized and stored at -80 ℃ for subsequent lipid study. Another 300 mL of culture at each time point was harvested and the algal cell pellets were stored at -80 ℃ for total RNA extraction. Each sample was performed as three parallels.

2.2 Material and chemical

Chloroform, acetonitrile, isopropanol, and methanol used in this study were of high-performance liquid chromatography grade, while all of the other reagents were of analytical grade. Chloroform was purchased from Xilong Scientific Co. (Guangdong, China). Isopropanol and methanol were purchased from Merck & Co. (Darmstadt, Germany). Acetonitrile was purchased from Sigma-Aldrich (St. Louis, Missouri, USA). Ultrapure water was prepared by a Millipore Milli-Q system (Billerica, MA, USA).

Sphingolipid standards, Cer (d18꞉1/16꞉0), Cer (d18꞉1/18꞉0), GluCer (d18꞉1/18꞉0), and GalCer (d18꞉ 1/12꞉0) were diluted with 39꞉60꞉1 chloroform/methanol/water; Cer (t18꞉0/16꞉0) were diluted with methanol, formulated into 10-mg/mL sphingolipid standard liquor, and then diluted to the working solution of 100 pg/mL with methanol. They were stored at -20 ℃ for further use.

2.3 Flow cytometry counting of algae and viruses

The sample analysis was performed with an Epics Attra Ⅱ flow cytometer (Beckman-Dickinson) equipped with an external quantitative sample injector (Harvard Apparatus PHD 2000) and a water-cooled laser providing 0.100±0.01 W at 488 nm. For enumeration of algal cells, frozen samples were quickly thawed and run with an injection flow rate of 50–100 events/s and with the discriminator set on red fluorescence. Virus enumerations were performed on frozen samples that were quickly thawed, diluted from 1꞉10 to 1꞉ 1 000 in TE buffer (10-mmol/L Tris, 1-mmol/L EDTA, pH 8.0), and stained for 10 min at 80 ℃ with SYBR Green-1 (Molecular Probes, Eugene, OR, USA) at a final concentration of 10^-4 of the commercial solution. The samples were analyzed at a flow rate of 50–100 events/s with the discriminator set on green fluorescence.

2.4 Sample preparation for lipid analysis

The dry powder of the algal cell pellets in each replicate was accurately weighed and transferred to an Eppendorf tube.

The total lipid extract from each biological replicate was analyzed to conduct a non-targeted lipidomics study on an ACQUITY UPLC system (Waters, USA) coupled with mass spectrometry on Q-Exactive HF (Thermo Fisher Scientific, USA). Details of the lipidomics analysis including sample preparation, equipment, and methods are based on a previously method (Hu et al., 2008). Firstly, 300 μL of methanol containing internal standards was added to each sample (2.8 μg/mL of Cer d18꞉1/16꞉0-d3, 0.67 μg/mL of sphingomyelin (SM) d18꞉1/12꞉0). Five beads were added to each sample then homogenized three times using a homogenizer (Tissuelyser-24, Shanghai Jingxin Industrial Development Co., Ltd., China) for 2.5 min at 65 Hz. Adding 600 μL of chloroform and 250 μL of Milli-Q water to the homogenate successively. And the phases were thoroughly mixed by vortexing. Next, centrifuged at 13 523×g for 10 min at 8 ℃ to form a two-phase system. The down-layer was extracted as a sample, and the quality control (QC) sample is prepared by mixing an appropriate amount of the lipid extract in each sample into aliquots. The QC sample was then pretreated as a real sample. Finally, the sample was vacuum dried in a Speedvac concentrator (Thermo Scientific, USA), and stored at -80 ℃ for subsequent analysis.

2.5 Lipidomics analysis by means of UPLC-Q-Exactive-MS

Lipidomics analysis was performed with an ACQUITY ultra performance liquid chromatography (UPLC, Waters, USA) coupled with a Q-Exactive HF mass spectrometry (Thermo Fisher Scientific, USA). A reversed phase BEH C8 column (2.1 mm× 100 mm, 1.7 μm) (Waters, USA) was used for UPLC separation. The column temperature was set at 55 ℃. The mobile phase consisted of acetonitrile/water (6꞉4, v/v) with 10-mmol/L ammonium (mobile phase A), and isopropanol/acetonitrile (9꞉1, v/v) with 10-mmol/L ammonium (mobile phase B). The elution gradient was maintained for 1.5 min from the initial proportion of 32% of mobile phase B, then increased to 85% B at 15.5 min, and then to 97% B at 15.6 min, and kept until 18 min. Then, the gradient was quickly changed back to 32% B at 18.1 min, and equilibrated for another 1.9 min until the next injection. The flow rate was maintained at 0.26 mL/min for 20 min.

In MS analysis, full-scan MS and data-dependent MS/MS (ddMS2) acquisition of sphingolipids were performed. When applying a heated electrospray (HESI) in positive ion mode, the parameters are set as follows: sheath gas flow rate, 50; aux gas flow rate, 13; spray voltage, 3.5 kV; capillary temperature, 263 ℃; scan range, 100–1 500 m/z. The sample injection method is an automatic injection of the needle pump, each injection volume was 5 μL, and the injection temperature was set at 10 ℃. When HESI was performed in negative ion mode, most of the parameters are the same as those used in positive ion mode except that the spray voltage was reset to 3.0 kV. To monitor the robustness of lipidomics analysis, QC samples were regularly inserted in the analytical sequence.

2.6 Lipidomics data processing and statistical

Identification of the lipid species was based on accurate m/z, MS/MS fragmentation pattern and retention behavior, by using of the database of LIPID MAPS, Xcalibur, and LipidSearch softwares. Trace Finder software (Thermo, USA) was subsequently used to quantity lipids with m/z tolerance of ±0.000 5%, retention time extraction window: +/-15 s. Peak check and noise cancellation are performed to reduce errors.

Peak areas of all samples were then normalized by dry weight of algal cells, and the internal standards area, SIMCA-P software (Umetrics, Sweden) were used for multivariate analysis with unit variance (UV) scaling. A supervised partial least squares discriminant analysis (PLS-DA) was used to analyze the group-predictive spectral features. The S-plot was generated to screen potential biomarkers. Then orthogonal partial least squares discriminant analysis (OPLS-DA) was used for identification of discrimination. The OPLS-DA model was estimated with the relevant R² and Q². The variable importance in projection (VIP) was employed to select feature lipids. Library Multi-Experiment Viewer (MeV) was used for evaluating the univariate statistical significance based on the Mann-Whitney-Wilcoxon test and Benjamini-Hochberg FDR correction. Heat map was generated by using MeV, to visualize the specific classes of sphingolipid and levels of sphingolipid in all control and experimental groups. Metabolic pathway analysis was carried out with the reference of LIPID MAPS and KEGG database.

2.7 RNA isolation and qRT-PCR analysis

Total RNA was isolated by following the protocol of Tiangen RNAsimple Total RNA Kit (DP419, TIANGEN China) and the RNA quality was evaluated by electrophoresis on a 1.0% agarose gel. The amount of the RNA was measured using a NanoDrop spectrophotometer. Equal amounts of RNA were used for cDNA synthesis with HiScript Ⅱ Q RT SuperMix (R223, vazyme). Seven genes encoding enzymes related to sphingolipid metabolism, such as viral serine palmitoyltransferase (vSPT), host serine palmitoyltransferase (hSPT), ceramide synthetase (CERS), viral fatty acid hydroxylase/sterol desaturase (vFAH), dihydroceramidase (YDC1), dihydroceramide desaturase (DCD), and host glucosyltransferase (hUGCG) were selected for qRT-PCR analysis. Primers were designed using Primer Premier v5.0 software and given in Supplementary Table S1.

qRT-PCR was performed on a Roche LightCycler^® 480Ⅱ/96 Real-time PCR System (Roche, Switzerland) using Universal SYBR Green Supermix (Vazyme, China) in 96-well plates according to manufacturer's recommendations. Reactions were performed in a volume of 10 μL containing 1.6 μL of cDNA, 5 μL of 1×SYBR Green Premix, 0.2 μL of each reverse and forward primer, 3.0 μL of ddH₂O, and in a 96-well optical plate. The conditions were set up as follows: an initial denaturation for 30 s at 95 ℃, 40 cycles for 10 s at 95 ℃, 30 s at 58 ℃ for annealing, 30 s at 72 ℃ with fluorescent signal recording. Experiments were performed in biological duplicates and experimental triplicates. The results of qRT-PCR experiments were calculated using the 2^-ΔΔCt method. The transcript abundance was calculated by normalizing the results to the expression of cyclin-dependent kinase A (CDKA) in each sample (Zhang et al., 2021).

2.8 Statistical analysis

Statistical analysis was performed using SPSS 17.0 software (SPSS Inc., USA). Inter-group variation was measured by one-way ANOVA and subsequent unpaired two-tailed Student's t-test. All data were presented as means±standard deviations (S.D.). The significance level was denoted using *0.01 < P < 0.05, **0.001 < P < 0.01, and ***P < 0.001.

3 RESUTL 3.1 Growth inhibition in E. huxleyi BOF92 during EhV 99B1 infection

In host-virus culture system, viral infection caused a significant decline in algal density and accompanied by the release of large amounts of virions around 20–45 hpi (Fig. 1), indicating that viral infection was successful in this study.

3.2 Method validation of sphingolipids analysis

To analyze the lipidomics associated with distinct phases of infection, the sphingolipid profiling of E. huxleyi was acquired based on the ultra performance liquid chromatography (UPLC)-Q-Exactive approach (Fig. 2a–b). The quality control (QC) samples were used to evaluate the quality of acquired sphingolipids data. The distribution of relative standard deviation (%RSD) for QC samples indicated that 92.31% of the sum of sphingolipids had %RSD of less than 10% (Supplementary Fig.S1). The results ensure the acceptable stability and reproducibility of lipidomics data.

3.3 Sphingolipid alterations of E. huxleyi after viral infection

Principal component analysis (PCA) of the identified sphingolipids showed a clear separation between the differently infected samples. Control and viral infection groups displayed a clear distinction trend at both 6 and 45 hpi, indicating a substantial change of the sphingolipid metabolites during viral infection (Fig. 3). Particularly, a more remarkable difference of sphingolipids could be observed at 45 hpi compared with that at 6 hpi (Fig. 4).

Analysis of the difference in sphingolipids showed a significant change in host sphingolipid metabolism during viral infection (Fig. 5). The sum of responses for each sphingolipid category was UV scaled and clustered hierarchically and displayed by a heat map (Fig. 5a).

Sixteen sphingolipids content significantly changed at 6 and 45 hpi after viral infection (Table 1; Fig. 5b). Compared with the contents in control group at 6 hpi, the contents of C38꞉1-Cer, C40꞉1-Cer, C40꞉2-Cer (isomer1), C40꞉2-Cer (isomer2), C40꞉3-Cer, C41꞉1-Cer, C42꞉1-Cer, C42꞉2-Cer, and C40꞉1-CerG1 in infection group were significantly reduced, and the contents of C39꞉0-Cer, C39꞉0-CerG1 in infection group were significantly increased (P < 0.05). Compared with the contents in control group at 45 hpi, the contents of SPH 18꞉1, C42꞉2-Cer, and C40꞉2-GlcCer were significantly reduced, and the contents of C39꞉0-Cer, C39꞉0-CerG1 were significantly increased in infection group at 45 hpi (P < 0.05). At the early stage of virus infection (6 hpi), the contents of C38꞉ 1-Cer, C40꞉1-Cer, C40꞉2-Cer (isomer1), C40꞉2-Cer (isomer2), C40꞉3-Cer, C41꞉1-Cer, C42꞉1-Cer, C42꞉2- Cer, and C40꞉1-GlcCer in the host cells was significantly down regulated than that in the late stage (45 hpi). Compared with the early stage (6 hpi), the down regulation of C18꞉1-SPH was growing in change magnitude at the later stage (45 hpi), while the content of C39꞉0-Cer, C39꞉0-CerG1 were significantly up regulated at the late stage. These results indicate that viral infection played a significant role in the regulation of sphingolipid levels of host cells.

The differential sphingolipids based on corrected absolute contents were used to further extract the potential markers for distinguishing viral infection stages at 6 and 45 hpi and indicate stable disturbances upon viral infection. Biomarker candidates including Cer 40꞉1;2 [i.e., Cer(d18꞉1/22꞉0)], Cer 40꞉2;2_isomer [i.e., Cer(d18꞉1/22꞉1)], CerG1 39꞉0; 2 [i.e., CerG1 (t17꞉0/22꞉0+O)], and Cer 39꞉0;2 [i.e., Cer(t17꞉0/22꞉ 0+O)] might indicate the general dysfunctions in E. huxleyi in response to viral infection at early and late stages, respectively (Fig. 6).

3.4 Quantitative analysis of genes related to sphingolipid metabolism

Genes participating in de novo sphingolipid biosynthesis based on quantitative real-time PCR (qRT-PCR) data revealed a significant up-regulation of host genes, including hSPT, CerS, DCD, and UGCG at 6 hpi in infected cells (Fig. 7; Supplementary Fig.S2; Supplementary Table S2). However, the levels of host C18-LCB (long-chain base) based SLs such as C18DHCer 40꞉0;2, C18 Cer 40꞉1;2, and C18 CerG1 40꞉1;2 decreased significantly during viral infection (Fig. 5b). Attractively, at 6 hpi, transcription level of host YDC1 was markedly increased and accelerated the conversion of C18 DHCer 40꞉0;2 to C18 DH SPH, resulting in the decrease of host C18-LCBs. At 45 hpi, almost all of the host genes above were significantly down-regulated (Fig. 7). Correspondingly, we detected a significant reduction of host ceramide (Cer) during viral infection (Fig. 5b). In addition, viral infection induced host UGCG gene expression and promoted to the accumulation of virus specific C17 sphingoid bases (C17Cer G139꞉0;2) other than host C18 CerG1 40꞉1; 2 (Figs. 5b & 7), suggesting glycosylation of viral ceramide intermediates via UGCG, which was absent in virus genome.

The mRNA levels of vSPT and vFAH genes encoded by virus increased at 6 hpi and was over forty times higher than that of the control group at 45 hpi. This rapid induction was followed by a profound accumulation of hydroxylated odd chain C17 LCB (t17꞉0) in infected cells (Fig. 8). In addition, an increase in vFAH gene expression induced a subsequently hydroxylated C18 SPH 18꞉1 at 6 hpi (Fig. 8). Interestingly, we detected a significant induction of phytoceramides (C17 phytoCer 39꞉0;2) and glycosyl phytoceramides (C17 CerG 139꞉0;2) (Figs. 5 & 8) during viral infection, in agreement with the up-regulated expression of vSPT, vFAH, and UGCG. These results indicated that virus induced metabolic shift of the sphingolipid synthesis pathway was due to down-regulation of host gene and up-regulation of viral homologous gene.

4 DISCUSSION

We applied lipidomic approach to characterize the metabolic rewiring of sphingolipid (SL) biosynthesis during the interaction between cosmopolitan bloom-forming alga E. huxleyi and its specific virus EhV. A total of sixteen significantly changed sphingolipids (SLs) (including Cer, CerG1, and SPH) were identified and the data profiling revealed a nearly complete decrease of host SLs (hSLs) and increase of viral SLs (vSLs) with the development of viral infection (Fig. 5). The outcome of this metabolic shift from host to virus-specific sphingolipid production may suggest the competitive inhibition of host de novo sphingolipid synthesis to support virion assembly (Rosenwasser et al., 2014; Ziv et al., 2016). hSLs (Cer and CerG1) decreased at both early and late stages as lytic infection proceeded and such downregulation upon viral infection was non-progressive between different stages (Fig. 5). Accordingly, hostencoded genes involved in sphingolipid metabolism exhibited upregulated at 6 hpi, while downregulated at 45 hpi during lytic infection (Fig. 7). This decoupling may imply a possible host defense mechanism to suppress virus-induced SLs synthesis in the early stages of lytic infection. Interestingly, we detected an induction of host C18 SPH 18꞉1 during lytic infection, which was accumulated at 6 hpi and decreased at 45 hpi (Fig. 5). We supposed that C18 DH SPH might be used as a potential substrate for vFAH during viral infection, thus an increase in vFAH gene expression induced a subsequently hydroxylated C18 SPH 18꞉1 during the first 6 h of infection (Fig. 5). Multiple hydroxylation of the ceramide backbones of SLs increased the stability of membrane and decreased its permeability, as well as better membrane resilience to environmental stress (Lynch and Dunn, 2004).

Our data showed that viral infection induced a shift in LCB profile of the host cells, suggesting that vSPT was responsible for this remodeling of host sphingolipid metabolism to produce virus-specific vSLs composed of t17꞉0 sphingolipid bases (C17 phytoCer 39꞉0;2) during lytic viral infection (Fig. 8). t17꞉0 was previously putatively identified as C17-phytosphingosine (C17-phytoSO) in EhV201 infected E. huxleyi strain CCMP374 (Ziv et al., 2016). Hydroxylated LCBs as phytoSOs are highly abundant in plants and fungal membranes (Markham et al., 2013), and have a significant impact on the biophysical characteristics of the membrane in which they are embedded. The hydrophobic backbones of SLs are able to increase the stability and reduce permeability of membranes, and improve membrane elasticity to environmental stress conditions (Hannun et al., 1995, 2002, 2008).

These vSLs are essential for viral assembly and infectivity. A previous study showed that the vSPT preferentially used myristoyl-CoA as a substrate, rather than the classical host hSPT that used palmitoyl-CoA (Han et al., 2006). Indeed, the production of vSLs was recently demonstrated to have an important role in viral assembly, since they are the major constituents of viral membranes (Vardi et al., 2009, 2012; Fulton et al., 2014). vSLs are facilitated by viral homologous genes that encode de novo sphingolipid enzymes. Our data showed that vSPT and vFAH genes were up-regulated, whereas the host homologous genes were almost undetected during viral infection (Fig. 7), suggesting that the vSPT and vFAH were responsible for the accumulation of C17 LCB sphingolipid (C17 PhytoCer 39꞉0;2 and C17 CerG1 39꞉0; 2). Serine palmitoyltransferase (SPT) activity data showed high utilization of C15-CoA as a preferential substrate during viral infection (Ziv et al., 2016). Since sphingolipid bases with shorter carbon chains are less hydrophobic, they are likely to exhibit some special biophysical properties, that might influence their subcellular localization and distribution in membranes. Taken together, these data indicated that viral infection was dependent on SPT leading to a shift in the substrate specificity to form unique sphingolipid base products composed of C16 and C17 chains. In addition, biomarker candidates Cer40꞉2;2, Cer C39꞉0;2, and Cer G139꞉0;2 might thus indicate the general dysfunctions in E. huxleyi in response to viral infection. Our results may provide novel insights into the chemical arms race between E. huxleyi-EhV interaction, and the potential influence on large-scale biogeochemical processes.

5 CONCLUSION

This study provided an insight into the potential mechanism of sphingolipid metabolism in the EhV infection process. The sphingolipid changes clearly showed a special viral strategy to involve in host cells sphingolipid metabolism along the infection process. Ultimately, EhVs competitively inhibited host de novo sphingolipid biosynthesis and promoted the catabolism of toxic ceramides, supporting their unique requirements in life cycle. Sensitive biomarkers of unique bioactive sphingolipids derived from EhVinfected cells could assess the possible impact of viral infections in marine food webs. Therefore, further investigations into the underlying mechanism of host-virus interactions are essential and may also provide insights into sphingolipid-based chemical arms race during host-virus dynamics in the ocean.

6 DATA AVAILABILITY STATEMENT

The datasets generated during this study are available from the corresponding author upon reasonable request.

7 ACKNOWLEDGMENT

We cordially thank Prof. Gunnar Bratbak (University of Bergen) for friendly providing E. huxleyi BOF92 and the E. huxleyi virus 99B1 strains.

Electronic supplementary material

Supplementary material (Supplementary Tables S1–S2 and Figs.S1–S2) is available in the online version of this article at https://doi.org/10.1007/s00343-022-1442-5.