Targeted Sorting of Single Virus-Infected Cells of the Coccolithophore Emiliania huxleyi

Discriminating infected from healthy cells is the first step to understanding the mechanisms and ecological implications of viral infection. We have developed a method for detecting, sorting, and performing molecular analysis of individual, infected cells of the important microalga Emiliania huxleyi, based on known physiological responses to viral infection. Of three fluorescent dyes tested, FM 1-43 (for detecting membrane blebbing) gave the most unequivocal and earliest separation of cells. Furthermore, we were able to amplify the genomes of single infected cells using Multiple Displacement Amplification. This novel method to reliably discriminate infected from healthy cells in cultures will allow researchers to answer numerous questions regarding the mechanisms and implications of viral infection of E. huxleyi. The method may be transferable to other virus-host systems.


Introduction
Viral infection greatly influences the biogeochemistry and genetic variability that sustain marine phytoplankton communities by accelerating the lysis of bacteria and phytoplankton and serving as vectors for horizontal gene transfer [1]. However, the inability to separately investigate infected and healthy phytoplankton cells in the environment limits our understanding of the ecological and biogeochemical implications of viral infection. Subtle changes in infected cells of a particular phytoplankton species are almost impossible to detect from bulk measurements using standard mass filtering [2] and subsequent biogeochemical or molecular analysis. This is particularly true when the species of interest is not dominant and/or when infected cells only represent a small fraction of the community. These limitations evidence the need for a method that allows discriminating and isolating infected phytoplankton cells from environmental samples.
Previous studies have shown that viral infection of the marine microalga Emiliania huxleyi leads to intracellular accumulation of reactive oxygen species (ROS) [3] and plasma membrane patchiness (blebbing) due to increased production of a dense lipid excreted to the cell surface [4]. Here, we took advantage of these physiological responses to develop a method for distinguishing single infected cells within a phytoplankton culture using flow cytometry prior to high throughput physical separation, whole genome amplification and molecular analysis. We chose the E. huxleyi host-virus system because of its wide distribution, high abundance and importance in the ocean's biogeochemistry [5,6], as well as the ease with which the host can be grown and the virus propagated in the laboratory. Additionally, recurring vast E. huxleyi blooms have been reported to be terminated by viral infection [7,8,9] and extensive sequence information is available for E. huxleyi strain CCMP 1516 (http://genome.jgi-psf.org/Emihu1/ Emihu1.home.html) and E. huxleyi-specific virus strain EhV-86 [10], facilitating molecular analysis.
We tested and compared three flow cytometric assays for the discrimination of healthy and infected cells: 1) lipid dye FM 1-43 for detection of membrane blebbing [4]; 2) CM-H 2 DCFDA for detection of accumulated intracellular ROS [3]; 3) DNA dye SYBR Green I for detection of increased total DNA in infected cells as virus progeny accumulates intracellularly prior to lysis (as reported for Chlorella NC64A [11]). We then judged the suitability of the sorted cells for whole genome multiple displacement amplification (MDA) to generate sufficient good quality genomic DNA for downstream molecular analysis. MDA amplicons were screened by PCR using generic and specific primers for E. huxleyi and E. huxleyi-specific viruses (EhVs) respectively.

Results
In an initial experiment we compared the efficacy of three fluorescence probes for discriminating infected from healthy E. huxleyi strain CCMP 1516 cells at different times during the infection process (2 h, 6 h, 20 h, 24 h, 42 h and 48 h postinoculation, PI). At each PI time point three aliquots from both a virus-free and a virus-added culture were each labeled with either fluorescent dye FM 1-43, CM-H 2 DCFDA or SYBR Green I. A fourth aliquot received no stain. In the virus-free aliquots all the cells were uniformly labeled and appeared as a single group (Fig. 1). FM 1-43 was found to be the most suitable dye for our host-virus system, allowing the most distinctive discrimination between cell subpopulations in the virus-added culture (one subpopulation with high-orange fluorescence as that in the virusfree culture and a second with low-orange fluorescence) compared to CM-H 2 DCFDA and SYBR Green I (Figs. 1B-D). Furthermore, distinctive subpopulations were resolved with FM 1-43 from 6 hours PI while the other two fluorescent dyes distinguished cell subpopulations (based on acquired green fluorescence) only from 20 hours PI (Figs. S1, S2, S3). In the absence of fluorescence dyes the cells' red autofluorescence and low green fluorescence levels did not change throughout the 48 h sampling period both in the virus-free and the virus-added cultures (Fig. S4).
In an independent experiment analogous to the previous one, we mixed FM 1-43 dye-labeled aliquots from virus-free and virusadded E. huxleyi cultures (1:1 ratio) at 20 h PI and sorted 84 single cells with high-and 84 single cells with low-orange fluorescence (red and green clusters, respectively, in Fig. 1B) into a 384-well plate. Additionally, we included 22 blank wells (no-drop deposition) and 1 positive sorting control well (50 cells in a single well) from each cell subpopulation. After sorting, MDA and real-time PCR screening with both MCP [12] and GPA [13] primers (to detect virus and host respectively) were performed on each well.
Close to 100% of the MDA reactions in wells containing single cells were operationally-defined successful, i.e. critical amplification point (Cp), the time necessary to reach half of the maximum accumulated fluorescence for each sample, was less than 12 hours. In general, low-orange fluorescence cells were MDA-amplified faster than high-orange fluorescence cells ( Table 1 and Fig. S5). Cp mean for multiple-cell wells was approximately 4:30 hours and Cp mean for blank wells was 14:35 hours, indicating the overall effectiveness of the sorting process, the suitability of the MDA reaction conditions and the lack of DNA contamination in the blanks and the MDA reagents (Fig. S5).
MCP and GPA PCR reactions yielded amplicons for the multiple-cell wells and did not yield products for any of the blank wells or the few MDA reactions with Cp$12 hours (data not shown), indicating the suitability of the reaction and the lack of contamination across wells. Ninety seven percent of the MDA products from single sorts with Cp,8 hours from the low-orange fluorescence cells yielded an EhV-MCP amplicon indicating that those single sorted cells were indeed infected with EhV-86. In contrast, we could only amplify the MCP fragment from approximately 17% of MDA amplicons with Cp,8 hours from highorange fluorescence cells ( Table 1). E. huxleyi-specific GPA reactions were overall more successful on high-orange fluorescence cells compared to infected cells and on MDA products with Cp.8 hours ( Table 1).
Although FM 1-43 dye was chosen in preference to CM-H 2 DCFDA and SYBR Green I dyes for our host-virus system, we also proved in a separate test that MDA amplification and PCR efficiencies did not depend on whether the cells had been dyelabeled prior to sorting or on the type of dye utilized (File S1).

Discussion
The method we describe here allows the unequivocal identification, isolation, and whole genome amplification of single E. huxleyi infected cells from a culture. A sensible starting point when developing a new technique is to investigate manageable components of the oceanic microbial ecosystem, ideally those that are quantitatively significant; hence, our choice of E. huxleyi. Yet, it is possible that this method can be used on a range of microalgae or other host species infected by viruses.
Certain changes in the host physiology can be used as indicators of viral infection. For instance, decreased photosynthetic efficiency of infected hosts has been measured as changes in fluorescence quenching [14,15,16]. The decline of autofluorescence can sometimes be detected and quantified by flow cytometry but it is not always evident, especially at early stages of infection. Our results show that in the absence of fluorescent dye, the red autofluorescence of both healthy and infected E. huxleyi cells remained undistinguishable for at least the first 24 h of study ( Fig.  S4) evidencing the need for a different approach to discriminate healthy from infected cells at earlier stages of infection when the host's genetic material is still intact [17]. Accumulation of intracellular ROS and membrane and cytoplasmic blebbing are indicators of programmed cell death induced by environmental factors such as excessive ultraviolet radiation, nutrient limitation, oxidative stress or viral infection [18]. Viruses have been suggested to be the ultimate cause of phytoplankton loss for which the death apparatus might be seen as a product of host and virus coevolution, with each trying to control it [19]. In this study we took advantage of those known physiological changes observed in E. huxleyi in response to viral infection, and found that membrane blebbing as detected with the fluorescent dye FM 1-43 [4] was most suitable for early discrimination of infected cells.
FM 1-43 revealed two cell populations distinguished by the level of orange fluorescence. Low-orange fluorescence cells, infected with EhV-86, had on average lower MDA Cp values than highorange fluorescence cells, of which only 20% yielded EhV-MCP amplicons. These amplicons could be due to the presence of noninfectious EhV particles nonspecifically attached to some of the high-orange fluorescence cells or it could be that those highorange fluorescence cells were indeed infected, but at such early stage of infection that membrane blebbing was not yet evident. Infected E. huxleyi cells exhibit plasma membrane patchiness only after approximately 2 h PI [4]. The EhV genome is an easier template than the E. huxleyi genome, which with its many repeats, GC-rich regions and extensive secondary structure, is a difficult template for amplification (http://genome.jgi-psf.org/Emihu1/ Emihu1.home.html). This likely explains the lower MDA Cp values for low-orange fluorescence (infected) cells compared to high-orange fluorescence cells (mostly non-infected). On the contrary, the higher success of E. huxleyi-specific GPA reactions on high-orange, compared to low-orange fluorescence cells (Table 1) could be explained by the fact that in infected cells (low orange fluorescence signal): 1) the EhV genomes are preferentially amplified by MDA; 2) most of the host genome has already been degraded for the production of new virions hindering amplification of the GPA fragment [17]. However, this may not be the case for other host-virus systems with different viral infection strategies and different genome composition and structures. This novel method for targeted sorting of infected E. huxleyi cells is a powerful tool that opens a range of new opportunities for the investigation of viral infection of this microalgae and possibly of other host-virus systems. The availability of genomic data from single microalgae infected cells can, for instance, help discern the factors that determine virus specificity and host susceptibility as well as gene transfer and co-evolution by facilitating the search for common shared patterns among the genomes of individual host cells and their viruses. Probably the most important improvement that this method offers is the ability to detect and genetically analyze individual infected cells without the need to maintain the host-virus system in the laboratory. Naturally occurring blooms of E. huxleyi are frequently terminated by viral infection [7,8,9] and therefore the majority of sorted cells that exhibit membrane blebbing or elevated intracellular ROS or total DNA content are likely to be infected. Moreover, results obtained this way from environmental samples will be more ecologically relevant than those from manipulated laboratory conditions which often do not reflect true ecological conditions.

Host culture
Non-axenic clonal Emiliania huxleyi strain CCMP 1516 (3-5 mm) was obtained from the Provasoli-Guillard Center for the Cultivation of Marine Phytoplankton (CCMP, Maine, USA; http:// ccmp.bigelow.org/). The cultures were maintained at 15uC and kept at mid-exponential growth phase (approx. 1-2610 6 cells ml 21 ) by periodically transferring 5-10% (v/v) culture in fresh f/2-Si seawater medium [20]. Virus pathogen E. huxleyi-virus (EhV) strain EhV-86 [12] was obtained from the Plymouth Virus Collection (UK). Fresh working solutions of EhV-86 lysate were produced prior to performing an experiment. Briefly, 1 ml lysate was added to 50 ml of an exponentially growing culture of E. huxleyi strain CCMP 1516. Once clearing of the host culture was observed, the lysate was passed through a 0.2 mm syringe filter (Sartorius AG, Germany) and the filtrate containing virus was stored at 4uC. Virus concentration was calculated by flow cytometry using SYBR Green I as described by Brussaard [22].  [3] or SYBR Green I (5610 5 dilution of commercial stock) for 30 min, 60 min or 15 min, respectively, in the dark at 15uC. In an initial experiment, an E. huxleyi culture aliquot was incubated for 10 minutes with 100 mM H 2 O 2 to artificially elevate the intracellular ROS concentration prior to addition of CM-H 2 DCFDA dye, as a control to verify the effectiveness of the CM-H 2 DCFDA labeling. H 2 O 2 is the most stable of the ROS and is capable of rapid diffusion across cell membranes [23]. Labeled E. huxleyi cells were discriminated by flow cytometry on the basis of their red autofluorescence at 610 nm versus side scatter, the green fluorescence of the CM-H2DCFDA and SYBR Green I dyes at 522 nm or the orange fluorescence of the FM 1-43 dye at 488 nm, accordingly.

Fluorescence-activated cell sorting
Prior to cell sorting, samples were diluted 10-fold with sterilefiltered seawater and pre-screened through a 70 um mesh-size cell strainer (BD). Sorting was done with a MoFlo TM (Beckman Coulter) flow cytometer using a 488 nm argon laser for excitation, a 70 mm nozzle orifice and a CyClone TM robotic arm for droplet deposition into microplates. The ''single 1 drop'' mode was used for maximal sort purity, which ensures the absence of non-target particles within the target cell drop and the drops immediately surrounding the cell.Extreme care was taken to prevent sample contamination by any non-target DNA. Instruments and reagents were decontaminated as previously described [24]. Cell sorting was performed in a HEPA-filtered environment. The cytometer was triggered on side scatter, the sort gates were based on red autofluorescence and side scatter for not-labelled cultures, on red autofluorescence versus orange fluorescence for FM 1-43-labelled cells, on red autofluorescence or green fluorescence versus side scatter for CM-H 2 DCFDA-labelled and on red autofluorescence versus side scatter or green fluorescence for SYBR Green Ilabelled cells. Cells from the virus-free aliquots and from the notlabelled virus-added aliquot were sorted by setting a gate that included the entire population. For virus-added cultures labeled with FM 1-43, the sorting gates included cells with either reduced or normal orange fluorescence, compared to the virus-free culture. In the case of virus-added cultures labeled with CM-H 2 -DCFDA or SYBR Green I, we set a double-gate criterion for cells with increased green fluorescence and relatively low red fluorescence and side scatter signal. Cells were deposited into 384-well plates containing 0.6 mL per well of 16 TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) and stored at 280uC until further processing. Sorted plates included single cells, blanks (no doplet deposition), and positive sorting controls (multiple cells into one well).

MDA reaction
Sorted cells were lysed and their DNA was denatured using cold KOH [25]. The genomic DNA was amplified using real-time multiple displacement amplification (MDA) [25,26] in 10 mL final volume reactions. The MDA reactions contained 2 U/uL Repliphi polymerase (Epicentre Biotechnologies, Madison, WI, USA), 1x reaction buffer (Epicentre Biotechnologies, Madison, WI, USA), 0.4 mM each dNTP (Epicentre Biotechnologies, Madison, WI, USA), 2 mM DTT (Epicentre Biotechnologies, Madison, WI, USA), 50 mM phosphorylated random hexamers (IDT) and 1 mM SYTO-9 (Invitrogen Co., Carlsbad, CA, USA). The MDA reactions were conducted at 30uC for 16 h followed by a polymerase denaturation step at 65uC for 15 min. Successful reactions were determined based on the real-time kinetics (increase in SYTO-9 fluorescence signal) and the melting curves measured with a FLUOstar Omega (BMG) plate reader. Time for amplification was determined using an in-house algorithm developed to calculate the critical amplification point (Cp), described as the time necessary to reach half of the maximum accumulated fluorescence for each sample. The Cp is inversely correlated to the amount of DNA template [27]. The amplified genomic DNA was stored at 280uC until further use for PCR screening.

PCR screening of MDA products
The MDA products were diluted 50-fold in sterile TE buffer and 1 ml or 0.5 mL aliquots of the dilute products served as template in 50 ml standard or 5 mL real-time PCR reactions, respectively. Previously described primers (Table S1) were used to amplify genes encoding the Major Capsid Protein (MCP) in EhVs [12], a calcium binding protein (GPA) in E. huxleyi [13], universal eukaryotic 18S rRNA (primers Euk1A [28] and Euk516R [29]) and prokaryotic and plastid 16S rRNA (primers 27F and 1492R) [30]. The PCR reactions contained 1U Taq   increasing cell subpopulation with high-green fluorescence and relatively higher red fluorescence and side scatter signals (marked by squares) throughout the 48 h period of study, probably as a result of the accumulation of intracellular ROS in some cells due to normal cellular metabolism. From the 20 h post-inoculation sampling point onwards the virus-added culture also showed a cell subpopulation with high-green fluorescence (lower than that in the virus-free culture), however, these cells had relatively lower red fluorescence and side scatter signals (marked by circles).