MicroRNA-mediated regulation of lipid metabolism in virus-infected Emiliania huxleyi

The interactions between Emiliania huxleyi and E. huxleyi virus (EhV) regulate marine carbon and sulfur biogeochemical cycles and play a prominent role in global climate change. As a large DNA virus, EhV has developed a novel “virocell metabolism” model to meet its high metabolic needs. Although it has been widely demonstrated that EhV infection can profoundly rewire lipid metabolism, the epigenetic regulatory mechanisms of lipid metabolism are still obscure. MicroRNAs (miRNAs) can regulate biological pathways by targeting hub genes in the metabolic processes. In this study, the transcriptome, lipidome, and miRNAome were applied to investigate the epigenetic regulation of lipid metabolism in E. huxleyi cells during a detailed time course of viral infection. Combined transcriptomic, lipidomic, and physiological experiments revealed reprogrammed lipid metabolism, along with mitochondrial dysfunction and calcium influx through the cell membrane. A total of 69 host miRNAs (including 1 known miRNA) and 7 viral miRNAs were identified, 27 of which were differentially expressed. Bioinformatic prediction revealed that miRNAs involved in the regulation of lipid metabolism and a dual-luciferase reporter assay suggested that phosphatidylinositol 3-kinase (PI3K) gene might be a target of ehx-miR5. Further qPCR and western blot analysis showed a significant negative correlation between the expression of ehx-miR5 and its target gene PI3K, along with the lower activity of its downstream components (p-Akt, p-TOR, SREBP), indicating that lipid metabolism might be regulated by ehx-miR5 through the PI3K-Akt-TOR signaling pathway. Our findings reveal several novel mechanisms of viral strategies to manipulate host lipid metabolism and provide evidence that ehx-miR5 negatively modulates the expression of PI3K and disturbs lipid metabolism in the interactions between E. huxleyi and EhV.


Enumeration
. Oligonucleotide primers used for mRNA qPCR. Table S2. Oligonucleotide primers used for miRNA qPCR. Table S3. List of antibodies used in this study. Table S4. Illumina sequencing statistics. Table S5. Assembly quality statistics. Table S6. Overview of all unigene annotation. Table S7. Assignment of the identified metabolites to lipid classes and amount of lipid species identified in this study. Table S8. Annotation and accession numbers of unigenes verified by qPCR. Table S9. Differentially expressed miRNAs in terms of log2 (fold change) from the transcriptome and -ΔΔCt from qPCR. Table S10. Plant-type targets of host miRNAs which were related to lipid metabolism. Figure S1. Infection dynamics, ultrastructure, and neutral lipid accumulation during EhV infection.

Enumeration of cell and virus abundance
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 watercooled laser providing 0.100 ± 0.01W 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 -1 and with the discriminator set on red fluorescence. Virus enumerations were performed on frozen samples that were quickly thawed, diluted from 1:10 to 1:1000 in TE buffer (10mM Tris, 1mM EDTA, pH 8.0), and stained for 10 min at 80 o C 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 -1 with the discriminator set on green fluorescence.

Transmission electron microscopy (TEM) analysis
Fixed cells were washed three times with 0.1 M PBS (pH 7.4) and post-fixed for 1 h in 1.0 % osmium tetraoxide at 4℃. Samples were washed in buffer, dehydrated in graded series of ethanol and embedded in Epon. Sections were cut using an LKB 2088 ultramicrotone, collected on 200-mesh copper grids, and stained with uranyl acetate and lead citrate. The stained sections were photographed with a JEM-100CXII electron microscope.

Intracellular ATP detection
The ATP levels in cells were detected using an ATP assay kit (Beyotime, Shanghai, China) according to the protocol provided by the manufacturer. In brief, cells were gathered with centrifuging for 5 min at 4°C, 3000 rpm and lysed in the lysis solution, collecting supernatant after centrifuging for 5 min at 4°C, 12,000 rpm. ATP detection reagent was diluted to one tenth with ATP detection reagent dilution, incubated for 35 min at room temperature. Then 100 μL ATP detection working fluid were mixed with 20 μL prepared standards and samples separately in 96-well plate. The results were detected by a microplate reader (Beckman Coulter, USA).

Nucleoprotein extraction
The extraction and isolation of nuclear and cytoplasmic protein were performed according to the Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime, Shanghai, China). First, 200 mL of algal cultures were centrifuged for 5 min at 8000 rpm at 4°C and the pellet was dissolved with 400 μL of cytoplasmic protein extraction agent A supplemented with PMSF. After vortex for 30 s, the tubes were incubated for 15 min on ice to promote lysis. Next, add 20 μL of the cytoplasmic protein extraction agent B, vortex for 5 s and incubated on ice for 1 min. Then the samples were centrifuged for 5 min at 13,500 rpm at 4°C and the supernatant, consisting of the cytosolic fraction, was immediately frozen for further analysis. The pellet was resuspended in 50 μL of nuclear protein extraction agent supplemented with PMSF. After vortexing the tubes 15-20 times for 30 min and centrifuging for 10 min at 13,500 rpm at 4°C, the supernatants containing the nuclear extracts were obtained.

Quality evaluation and global changes of lipid profiling
Nontargeted lipidomic analysis was performed to explore host-virus interactions. The large-scale lipid profiling of E. huxleyi revealed approximately 500 lipid features (Fig. S5). A total of 246 lipid species were identified (Supplemental Dataset 1), including 53 phospholipids (PLs), 26 sphingolipids (SLs), 144 glycerolipids (GLs), 20 FAs, 2 wax esters (WEs) and 1 of CmE (Table S7). The quality of the resulting lipid profiling was examined by evaluating quality control (QC) samples and was confirmed to be satisfactory for complex biological samples according to our published methods [1] (Fig. S6). Prior to subsequent analysis, lipids with a percentage relative standard deviation (%RSD) higher than 30% in all QCs were removed from the dataset. An overview of E. huxleyi lipidome deregulation upon viral infection was visualized using a multivariate partial least squares discriminant analysis (PLS-DA) model, and six types of metabolic disturbance induced by viral infection were clearly visible on the score plot along the first principal components (Fig. S7A). The PLS-DA model was validated without overfitting based on response permutation test with 200 iterations (Fig. S7B).
The relative levels of all lipid metabolites were scaled by z score and subjected to hierarchical clustering heatmap to visualize the variation tendencies during viral infection. These lipid classes were clustered into four major change models (clusters 1-4) (Fig. S8A), and the lipids from different clusters were subsequently subject to KEGG enrichment analysis (Fig. S8B-E). Moveover, the sum of lipid levels was also visualized by a plotted heatmap (Fig. S9). The level trends of lipids from cluster 1 gradually decreased with the duration of infection, while the lipid levels in cluster 2 were upregulated during 6-24 hpi. The lipids in cluster 1 and cluster 2 were both significantly enriched in glycerophospholipid, sphingolipid and glycerolipid metabolism (Fig. S8B, C), specifically including most PLs (PSs, PEs, PIs, PAs, PGs, and CLs), host SLs (hCers, hGSLs, and SMs) and GLs (MGDGs, DGs, and TGs) (Fig. S9).
The diversity of metabolic disruptions during virus infection implied complicated dynamics of the hostvirus interaction.

Reprogrammed lipid metabolism during viral infection
FA metabolism-The expression patterns of genes and metabolites related to FA biosynthesis were strongly affected by EhV infection (Fig. S12). Although the expression trends of acetyl-CoA carboxylase (ACC), the rate-limiting enzyme of FA biosynthesis, was fluctuant, it reached the highest expression level at 12 hpi (1.72-fold compared with that at 0 h). Simultaneously, most of the FA synthase except FabD also showed a trend towards higher expression in early infection (0-12 hpi). However, two types of free saturated FAs (FA 16:0 and 18:0) detected in our lipidome data both gradually decreased during infection (Fig. S12). This might be explained by the following reasons. First, we found that nearly all genes participating in host β-oxidation (except ACADL) were highly expressed in early stage (0-12 hpi), which might result in high decomposition efficiency to FAs (Fig. S12). Second, long chain acyl-CoA synthase (ACSL), which catalyzes the acylation of long chain FAs, was gradually upregulated and reached its maximum at 24 hpi (3.02-fold), further consuming FAs to synthesize other types of lipids (Fig. S12). In addition, we speculated that some continuously increased unsaturated FAs (FA 20:4, 16:1 and 20:5) were transformed from saturated FAs (Fig. S12). These observations indicated that there was a high demand for FAs during the early phase of viral infection.
Glycerolipid metabolism-The metabolic map of glycerolipid metabolism (including glycerophospholipid metabolism) was shown in Fig. S13. Almost all genes involved in glycerolipid biosynthesis were induced and reached their maximal levels during early infection, such as glpK (3.05fold at 12 hpi), GPAT (2.01-fold at 12 hpi), hLPP (2.10-fold at 12 hpi) and DGAT (2.99-fold at 6 hpi) (Fig. S13). Simultaneously, the high expression of vLPP (851.76-fold higher at 48 hpi than at 24 hpi) made it possible to promote glycerolipid biosynthesis at late infection stage (Fig. S13). In agreement, the metabolic levels of sum of triacylglycerols (TGs) were elevated during the whole infection process, reaching the top at 60 hpi (Fig. S13). This increased abundance of TGs coincided with a decrease in the relative abundance of monogalactosyldiacylglycerols (MGDGs) and diacylglycerols (DGs) during infection (Fig. S13). Despite the increased profiling of sum of TGs, the abundance of partial TGs showed a gradual decrease during the infection process (Fig. S8A) concomitantly with a significant upregulation of host TGL (hTGL, 2.70-and 2.24-fold at 12 and 24 hpi, respectively) and viral TGL (vTGL, 701.91and 1887.55-fold higher at 48 hpi and 60 hpi than at 24 hpi, respectively) (Fig. S13). We speculated that this part of TGs was hydrolyzed by TGL to release FAs that could be oxidated to generate energy for viral production, which might account for the increase of some unsaturated FAs (Fig. S13). The abundance of PIs was decreased compared with the control group (1.24-1.69-fold decrease) (Fig. S13).
It was assumed that most PIs were phosphorylated to generate PIP, PIP2 or PIP3, which served as signal molecules playing a prominent role in virus entry [2]. Other PLs were gradually increased during the infection, reaching the maximum levels at 12 hpi (PG and PE), 24 hpi (CL) or 48 hpi (PC).
Sphingolipid Metabolism-The expression pattern of genes (including viral genes) related to sphingolipid metabolism was strongly consistent with the abundance of sphingolipid metabolites (Fig.   S14). During early infection stage (0-12 hpi), host genes participating in de nove sphingolipid biosynthesis except 3-ketodihydrosphingosine reductase (KSR) showed a high expression, corresponding to the abundance of the detected SLs, including dihydroceramide and ceramide, which were significantly accumulated in early infection stage (Fig. S14). Reversely, during the late infection stage (24-60 hpi), the host expression in both transcriptional level and metabolite level were sharply decreased (Fig. S14). Other deviants of sphingolipids like glucosylceramide and sphingomyelin exhibited a similar expression trend (Fig. S14). In contrast to host sphingolipid expression patten, all viral genes participating in de novo sphingolipid biosynthesis, such as serine palmitoyltransferase (SPT), dihydroceramide desaturase (DCD), and ceramide synthase (CerS) showed a profound overexpression (210-1668-fold increase compared with 24 hpi) at the late stages (48 hpi and 60 hpi) of infection (Fig.   S14). In addition, virus-encoded FA hydroxylase (vFAH) was highly expressed at the late stage of infection (Fig. S14), which hydroxylases the FA chain bases of dihydrosphingosine or ceramide to generate phytosphingosine or phytoceramide, respectively. Conformably, virus-specific ceramides and glucosylceramides also showed extremely high abundance at 48 and 60 hpi and contained hydroxylated FA chain bases (Fig. S13). Besides, according to the previous study [3], vSPT could make use of C14-CoA or C15-CoA as substrates to produce ceramides, which is consistent with the C16 or C17 FA chains contained in viral phytoceramide from our data (Fig. S14). Taken together, these observations reveal the inhibited host sphingolipid metabolism and a metabolic shift towards viral sphingolipid metabolism during EhV infection in E huxleyi-EhV system. Tables   Table S1. Oligonucleotide primers used for mRNA qPCR.