Viral Glycosphingolipids Induce Lytic Infection and Cell Death in Marine Phytoplankton

See allHide authors and affiliations

Science  06 Nov 2009:
Vol. 326, Issue 5954, pp. 861-865
DOI: 10.1126/science.1177322


Marine viruses that infect phytoplankton are recognized as a major ecological and evolutionary driving force, shaping community structure and nutrient cycling in the marine environment. Little is known about the signal transduction pathways mediating viral infection. We show that viral glycosphingolipids regulate infection of Emiliania huxleyi, a cosmopolitan coccolithophore that plays a major role in the global carbon cycle. These sphingolipids derive from an unprecedented cluster of biosynthetic genes in Coccolithovirus genomes, are synthesized de novo during lytic infection, and are enriched in virion membranes. Purified glycosphingolipids induced biochemical hallmarks of programmed cell death in an uninfected host. These lipids were detected in coccolithophore populations in the North Atlantic, which highlights their potential as biomarkers for viral infection in the oceans.

Marine phytoplankton are the basis of marine food webs and are responsible for nearly half the global carbon-based net primary production (1). The coccolithophorid Emiliania huxleyi (Prymnesiophyceae, Haptophyte) is a cosmopolitan unicellular photoautotroph whose intricate calcite skeletons account for about a third of the total marine CaCO3 production. E. huxleyi forms massive annual blooms in the North Atlantic that have been shown to be infected and terminated by lytic, giant double-stranded DNA containing coccolithoviruses (2, 3), a subset of the larger Phycodnaviridae group that infects microalgae (4). As the most abundant biological entities in aquatic environments, viruses turn over more than a quarter of the photosynthetically fixed carbon, thereby fueling microbial food webs and short-circuiting carbon export to higher trophic levels and the deep sea (5, 6). In addition, marine viruses stimulate the lateral transfer of genes from one host cell to another, which contributes to the diversification and adaptation of plankton in the oceans (7).

Very little is known about the molecular mechanisms and signal transduction pathways mediating phytoplankton cell death by marine viruses. However, it was recently shown that lytic viral infection of E. huxleyi induces hallmarks of autocatalytic, programmed cell death (PCD), including metacaspase expression and caspase activity, which were required for successful viral replication (8). The recent availability of genomic resources for an E. huxleyi host (9) and E. huxleyi lytic virus strain 86 (EhV86) (10) provides an unprecedented opportunity to explore cellular pathways triggered during execution of viral infection and to gain insights into the origin of PCD in these unicellular photoautotrophs. The genome sequence of EhV86, the type strain for the Coccolithoviruses, revealed an unexpected cluster of putative sphingolipid biosynthetic genes (10), a pathway that has not been described in a viral genome. Recent phylogenetic evidence for the transfer of seven genes in the sphingolipid biosynthesis pathway between E. huxleyi and EhV86 suggests a critical role in host-virus interactions (11). De novo sphingolipid biosynthesis is initiated by serine palmitoyltransferase (SPT) (12) and leads to ceramide production, a potent inducer of PCD in animals and plants (13, 14). The EhV86-encoded SPT gene is expressed during infection (10, 15) and encodes an active SPT with a unique biochemical preference for myristoyl–coenzyme A (myristoyl-CoA) as a substrate, when expressed heterologously in yeast (16). Nonetheless, it is not known whether these lipids play any functional role during infection of E. huxleyi.

We examined the polar membrane composition of uninfected and EhV86-infected sensitive (Ehux374) and resistant (Ehux373) E. huxleyi strains during the course of lytic infection. Using high-performance liquid chromatography with electrospray ionization mass spectrometry (HPLC/ESI-MS) (17, 18), we compared the lipid composition of uninfected and infected host cells. We detected glycosphingolipids (GSLs) in uninfected host cells that appeared to be composed of predominantly hydroxyl-sphingoid bases derived from palmitoyl-CoA (Fig. 1A). These host sphingoid bases are consistent with the expected products of the host SPT, which utilizes palmitoyl-CoA, and are common in plants (19). However, the lipids from EhV86-infected Ehux374 had unique GSLs yielding fragmentation ions that were indicative of multiply hydroxylated sphingoid bases derived from myristoyl-CoA (Fig. 1B and fig. S1). These sphingoid bases are the expected products of the viral SPT, based on its preference for myristoyl-CoA (16). The virus-induced myristoyl-base GSLs were absent in uninfected cells and were unique to lytic viral infection (Fig. 1B). Both resistant and susceptible hosts produced significant concentrations of host palmitoyl GSLs, which were structurally distinct from the viral myristoyl GSLs (Fig. 1, A and D).

Fig. 1

Production of viral GSLs in EhV86-infected E. huxleyi cells and in purified EhV86 virions. Summed ion HPLC/MS chromatograms showing relative abundances (normalized to an internal standard) of GSLs extracted from (A) susceptible Ehux374, (B) Ehux374 infected with EhV86 52 hours post infection, (C) purified EhV86 on a CsCl2 gradient, and (D) resistant Ehux373 infected with EhV86 52 hours post infection. (Insets) TEM micrographs of respective treatments. Arrows in (B) depict intracytoplasmic vacuoles (C) that contain viral particles (V).

To ascertain the origin of the myristoyl GSLs, viruses were purified using a cesium chloride (CsCl2) density gradient and ultracentrifugation. Identical myristoyl GSLs were dominant components of the lipids extracted from purified EhV86 viruses (Fig. 1C); host palmitoyl GSLs were absent. These observations indicate that the viral GSLs are a component of the membranes packaged with this virus. Although GSLs are distributed in some prokaryotes, they have not been found in phytoplankton or microalgal viral membranes (20). Proteomic analysis of the EhV86 virion determined that 23 out of 28 proteins are predicted to be membrane proteins (21), which corroborates our observations of GSLs and membrane structures. Sphingolipids, such as GSLs, are common constituents of membrane lipids, which may form lipid rafts in eukaryotes. For example, lipid rafts may be involved in the entry and budding of HIV and hepatitis C, yet their role is not well understood (20, 22). Transmission electron microscopy (TEM) micrographs of infected Ehux374 cells revealed viral particles within intracytoplasmic vacuoles, consistent with this type of strategy (Fig. 1B, inset).

The accumulation of viral GSLs in Ehux374 during EhV86 infection was accompanied by a reduction in cell abundance, a severely compromised photochemical quantum yield of photosystem II (PSII) (declining to 0.22 after 48 hours), and an ~30-fold induction in caspase specific activity (Fig. 2). Induction of this biochemical PCD marker occurred concomitantly with de novo synthesis of viral GSLs and viral replication (reaching 3 × 108 virus/ml), leading to the demise of Ehux374 at the onset of the lytic phase 25 hours post infection. Although elevated viral GSLs were detected at 3.5 hours in EhV86-infected cultures, presumably because of the presence of free viruses, de novo production of viral GSLs began after the first 12 hours (Fig. 2D, inset), trailing gene expression dynamics of the virally encoded SPT at 2 hours post infection (15). In late infection (>50 hours), caspase specific activity and GSL production exceeded levels seen in uninfected cells or in resistant Ehux373 cells by a factor of >100 (Fig. 2, C and D). Linear regression demonstrated a high correlation (R2 = 0.815) between viral GSL production and caspase specific activity over the course of lytic viral infection (fig. S2). In contrast, the EhV86-resistant strain Ehux373 exhibited slightly better growth than control, uninfected Ehux374 cells (Fig. 2A). Only trace levels of viral GSLs were detected in infected Ehux373, probably originating from the viral inoculum (Figs. 1D and 2D).

Fig. 2

Onset of the lytic phase during EhV86 infection is mediated by induction of caspase activity and viral GSL production. Viral infection dynamics of susceptible Ehux374 or resistant Ehux373 strains as monitored by the following parameters: (A) host abundance; (B) photochemical quantum yield (Fv/Fm) of PS II , a proxy for photosynthetic health; (C) caspase specific activity via normalized cleavage of IETD-AFC in cell extracts; and (D) de novo synthesis of viral, myristoyl GSLs. Data in (A) to (C) is derived from four to six biological replicates for infected cultures. Analytical error for GSL measurement is from two independent experiments, each analyzed in triplicate.

We monitored the ability of viral GSLs to modulate host physiology after purification from EhV86-infected Ehux374 cells by preparative HPLC and addition to uninfected Ehxu374 at various concentrations (Fig. 3). The viral GSLs suppressed cell growth compared with control cells treated with dimethyl sulfoxide (DMSO, a solvent) and control cells treated with phosphatidylglycerol (PG), which had a similar HPLC retention time (Fig. 3A). The cells treated with viral GSLs exhibited a dose-dependent induction of cell death above a threshold concentration (>0.06 µg/ml). Induction of cell death compromised photosynthetic efficiency of PSII (Fig. 3A) and resulted in elevated in vivo caspase activity for about 20 to 25% of Ehux374 cells treated with 0.3 and 1.5 µg/ml of viral GSLs after 48 hours (Fig. 3B). The induction of in vivo caspase activity was assessed by cell staining with the fluorescently labeled caspase probe, VAD-FMK conjugated with fluorescein isothiocyanate (FITC), and flow cytometric analysis (Fig. 3B) (18). Likewise, 18 and 56.2% of the cells at these two concentrations were positively stained with SYTOX, a DNA binding fluorescent indicator of compromised cell membrane integrity (Fig. 3C); this observation is consistent with previous findings of late stages of PCD in phytoplankton (23). Only 5.6 to 10.3% of cells were positively stained with VAD-FMK-FITC in the control treatments and in cells that were treated with a sublethal GSL concentration (e.g., 0.06 µg/ml), which indicated minimal induction of PCD. Likewise, a visual comparison of control (DMSO or PG) and GSL treatments revealed massive cell lysis only at GSL concentrations >0.06 µg/ml (Fig. 3C).

Fig. 3

Application of purified viral GSL to uninfected E. huxleyi cells mimics infection by inducing PCD. Dose-dependent induction of cell death in uninfected Ehux374 cells over 72 hours by application of purified GSLs. (A) Cell abundance (bars) and photochemical quantum yield of PS II (circles). Error bars, standard deviation of four biological replicates. (B) In vivo caspase activity (measured by flow cytometry). Cytograph plots represent the fluorescence distribution of 5000 cells 48 hours post treatment and after staining with VAD-FMK-FITC (CaspACE). The percentage of positively stained cells is given. The dashed line represents the threshold fluorescence above which cells are positively stained (as determined by unstained cells for each treatment). (C) Viral GSL-treated cultures exhibited massive cell lysis after 72 hours. The percentage of SYTOX-positive cells, which serves as a proxy for dying cells, is given. Control treatments consisted of DMSO (solvent) or phosphatidylglycerol (PG).

Given their potent ability to trigger E. huxleyi’s PCD response in a dose-dependent manner and their presence in purified EhV virions, we propose that viral GSLs may be part of a timing mechanism for viral release. In such a mechanism, host lysis is dependent on the accumulation of viral myristoyl GSLs to a critical effective concentration, above which host PCD is induced. According to our measurements, an EhV86 virion contains ~0.1 to 0.3 fg of myristoyl GSLs. At a typical burst size of ~800 to 1000 viruses per cell (8), an effective intracellular concentration of about 80 to 300 fg per cell is reached at the peak of the lytic phase; this concentration is consistent with the 100 to 200 fg per cell observed at the onset of lytic infection and PCD activation (>50 hours post infection) (Fig. 2, C and D, and fig. S3). Given that ceramide enrichment in cell membranes can serve to modulate entry and release of viruses (20, 22), GSL enrichment in intact EhV86 virions and the profound accumulation of viral GSL during lytic phase may suggest a similar mechanism for the timed release of EhV86 virions.

We hypothesize that such bioactive molecules have the potential also to elicit cell death in surrounding, uninfected cells under natural bloom densities and, hence, may serve as a bloom termination signal. It has been suggested that induction of PCD in E. huxleyi can act as a “viral exclusion” strategy (8, 24) to limit production of viruses during infection and, ultimately, their propagation through clonal populations. We calculated the “sphere of influence” distance around a given lysed cell that would contain an adequate viral GSL concentration to induce PCD in surrounding cells. Using our measurements of viral GSL concentrations (~200 fg per cell) and a threshold concentration of 0.2 µg/ml to induce PCD and cell lysis (fig. S4) and assuming the GSL is isotropically dispersed, we estimated that a sphere with a diameter of a few hundred micrometers would be sufficient to elicit cell death. Given that E. huxleyi blooms can reach about 100,000 cells/ml (25) and that cells clearly respond to external GSL application (Fig. 3), it is plausible that GSLs act as intercellular signals. Similar findings were recently reported for diatom-derived oxylipins found to act as infochemicals either to potentiate PCD or to induce resistance in sublethal doses (26, 27).

Because of their potent bioactivity and unique chemical signature, we also propose virus-derived GSLs can act as a novel biomarker for viral infection of natural phytoplankton assemblages. We collected natural plankton for GSL analyses during a cruise in the North Atlantic along 65°W between 43°N and 33°N during April of 2008. We used concentrations of 19′-hexanoyloxyfucoxanthin (19′-hex), a pigment marker for prymnesiophytes, including coccolithophores, as a proxy for host abundances (Fig. 4). Palmitoyl GSLs, similar to those seen in uninfected Ehux374 cultures, were observed throughout the North Atlantic transect at stations characterized by high 19′-hex pigment concentrations, which we postulate are indicative of healthy coccolithophore populations. In contrast, at a local minimum in 19′-hex pigment concentrations, we observed a GSL molecule with an HPLC retention time and mass spectrum that was consistent with the myristoyl GSL observed in EhV86-infected cultures of Ehux374 (compare Fig. 1, Fig. 4, and fig. S3). Together with our laboratory-based findings showing viral GSL production during viral lysis and its incorporation into virions (Fig. 1, 2), the observation of a putative viral GSL at the 19′-hex minimum is suggestive of virus-induced demise of a natural E. huxleyi population at this location. Detected expression of some host and viral sphingolipid biosynthetic genes in natural E. huxleyi populations corroborate our findings (28). On the basis of the diagnostic preference of the viral SPT for myristoyl-CoA (16), these viral GSLs may serve as a proxy for viral infection of coccolithophore populations in the sea. Currently, biomarkers to quantify active viral infection in the oceans are lacking, which hinders our understanding of the role and activity of viruses and virus-mediated processes in the oceans. We posit that future studies will elucidate the relative importance of biotic (viral infection as in this study) and abiotic [phosphorus limitation (17)] stress conditions, in shaping community structure of marine microbes, through the detection of different classes of stress-specific lipids.

Fig. 4

Viral GSLs represent a potential biomarker for viral infection of natural coccolithophore populations. Vertical distribution of 19′-hex along a meridianal transect at 65°W in the North Atlantic (43°N to 33°N). A multiply hydroxylated myristoyl GSL, which we posit is of viral origin, was only detected in a distinct zone of minimum 19′-hex in pigment values (35°N 65°W, 5-m depth; designated by arrows). This observation is consistent with population demise by viral infection. (Inset) Tentative chemical structure of the observed myristoyl GSL.

Although the origin of PCD in unicellular organisms is still unclear, its functional conservation among phylogenetically diverse phytoplankton lineages suggests key evolutionary and ecological drivers in aquatic environments (29). The retention and expression of a nearly complete, virus-based sphingolipid biosynthetic pathway, along with its requirement for viral replication and regulation of the host cell fate, now underscores the pivotal role of a chemical-based, coevolutionary “arms race” in mediating host-virus interactions.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S4


  • * Present address: Department of Plant Sciences, Weizmann Institute of Science, Rehovot 76100, Israel.

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. The authors acknowledge M. Soule and E. Kujawinski and the funding sources of the Woods Hole Oceanographic Institution Fourier-Transform Mass Spectrometry Facility (NSF’s Major Research Instrumentation Program OCE-0619608 and the Gordon and Betty Moore Foundation) for assistance with MS data acquisition. We thank D. Repeta for assistance and advice on pigment analysis, V. Starovoytov for TEM analysis, and C. Brown and M. Frada for helpful discussions and valuable feedback. This work was supported by NSF grant IOS-0717494 to K.D.B. and A.V., as well as NSF grant OCE-0646944 to B.V.M.
View Abstract

Navigate This Article