Marine viruses are the most abundant biological entity in the ocean and are considered as major evolutionary drivers of microbial life (Suttle, 2007). Yet, we lack quantitative approaches to assess their impact on the marine ecosystem. Here, we provide quantification of active viral infection in the bloom forming single celled phytoplankton Emiliania huxleyi infected by the large virus EhV, using high-throughput single molecule mRNA in situ hybridization of both virus and host transcripts. In natural samples, viral infection reached only 25% of the population despite synchronized bloom demise exposing co-existence of infected and non-infected subpopulations. We prove that photosynthetically active cells chronically release viral particles through non-lytic infection, and that viral-induced cell lysis can occur without viral release, thus challenging major assumptions regarding the life cycle of giant viruses. We could also assess active infection in cell aggregates, linking viral infection and carbon export to the deep ocean (Laber et al., 2018) and suggest a potential host defense strategy by enrichment of infected cells in sinking aggregates. Our approach can be applied to diverse marine microbial systems, opening a mechanistic dimension to the study of biotic interactions in the ocean.

Culture growth and viral infection

The non-calcifying E. huxleyi strain CCMP 2090 was used for this study. Cells were cultured in K/2 medium with antibiotics (Ampicillin and Kanamycin) and incubated at 18°C with a 16:8h light-dark illumination cycle. A light intensity of 100 μmol photons m^-2 s^-1 was provided by cool white light-emitting diode lights. All experiments started with exponential phase cultures (5*10^5 cells mL^-1). The virus used for this study is EhV201 propagated on CCMP2090 with antibiotics. Five day before infection, Most Probable Number assays were performed to assess the fraction of infectious viruses in the stock (48). E. huxleyi was infected with 5:1 multiplicity of infection (MOI) ratio of infectious virus per cell, hence guaranteeing that all cells encountered an infectious particle 30 minutes post infection. Time course of infected and non-infected cultures were done simultaneously, in triplicates. A second MPN experiment was performed on the day of the experiment with the same virus and same algae to assess the exact MOI at the beginning of the experiment (Table S1).

mcp, and psbA probe design and conjugation for smFISH

The smFISH technique identifies a single mRNA based on the binding of multiple small probes targeted to different locations to the mRNA of interest. Fasta sequences of the target genes mcp and psbA were first submitted to Stellaris Probe Designer to obtain potential probes. We designed 48 probes per gene with a probe length of 20 nucleotides. Probes with more than 70% GC content were discarded. To discard off-targets and decrease nonspecific signal, each probe sequence was blasted against the E. huxleyi transcriptome and EhV201 genome. Probes with an off-target gene matching over 17 nucleotides were discarded. A minimum of 20 probes per gene are necessary to detect a single molecule. Validated probes were ordered with 3’ amine groups through the Custom Oligo Service of BioSearch Technologies (see Probe_Sequences.xlsx). Fluorophores with succinimidyl ester group were ordered from Click Chemistry Tools: Tetramethylrhodamine (TMR) and sulfo-Cyanine5 (Cy5). Probes were coupled to fluorophores and purified according to (23). All the conjugated probes of a given gene are used as a mixture each time we target the expression of that gene. Therefore, the mcp mRNA was detected using 47 probes of 20 nucleotides each, and psbA mRNA was detected using 48 probes of 20 nucleotides each. The position of the probes on the reference gene are shown in Figure S9A and S9B, and their respective positions and sequences indicated in the file Probe_Sequences.xlsx

Sample fixation and hybridization for smFISH in laboratory samples

At each time point, 50 mL of each flask were fixed in cold 1% paraformaldehyde final concentration, and incubated for 1h at 4°C with gentle agitation. Each sample was then centrifuged for 2 min at 4°C and 3000g. The supernatant was discarded, the pellet resuspended in 1 mL of cryopreservant solution (prepared in 1X PBS containing 4% paraformaldehyde and 30% sucrose), transferred to a 1.7 mL Eppendorf and incubated 1h at 4°C with agitation. Tubes were then centrifuged for 2 min at 4°C and 3000g, the supernatant removed, and stored at 80°C until hybridization.

A day before ImageStreamX acquisition, selected samples were thawed and chlorophyll extracted by a first wash using 900 μL of 70% ethanol applied for 3 min, followed by centrifugation 3 min at 3000g and removal of supernatant. A second wash was performed with 900 μL of 100% ethanol applied for 3 min, followed by centrifugation of 3 min at 3000g and removal of supernatant. Samples were treated with 500 μL ProteinaseK at 10 μg/mL final concentration (Ambion #AM2546) for 10 min at room temperature and washed with 3 minutes at 3000g centrifugation. Samples were then resuspended in 50 mL of hybridization buffer (17.5% formamide concentration) containing equal concentrations of the different target genes at 0.1ng/mL final concentration. Mcp probes were conjugated to TMR, and psbA probes were coupled to Cy5. Hybridization was performed overnight in 30°C shakers.

The day of ImageStreamX acquisition, hybridization buffer was washed away by 3 min centrifugation at 3000g. Samples were stained for dsDNA with 500 μL of GLOX buffer (prepared in nuclease free water with 0.4% final concentration of glucose, 2X final concentration SSC Ambion #AM9765, and 10 mM final concentration of Tris pH 8.0) with DAPI in 10μg/mL final concentration (except the single stains). DAPI staining was done for 30 min in 30°C, followed by centrifugation 3 min at 3000g, removal of supernatant and resuspension in 40μL of GLOX buffer before being acquired in the ImageStreamX. DAPI was used for several reasons. The first one is technical: DAPI is required for the fluorescent microscope to localize cells when Brightfield channel is not available. We used DAPI to draw the contour of the nucleus using Cell Profiler in Figure 1 and Figure S12. DAPI is also important for gating of the cells in the imaging flow-cytometer. Gating on brightfield signal and on the DAPI achieves better detection of the cells and thus better downstream statistics. The second major reason to use DAPI is that it provides important information regarding the general DNA content of the cells, its cell cycle phase, and is a nucleus marker that serves as a reference point in our cells.

The following day, samples were resuspended in 300 μL of GLOX with the 3,3'-Dihexyloxacarbocyanine Iodide membrane stain at final concentration of 1μM (ThermoFisher Catalog number D273). Excessive dye was removed by 3 min centrifugation at 3000g and removal of supernatant, followed by resuspension in 4 μL GLOX. 2 μL were deposited on a microscope slide with 2 μL of antibleach solution, and the sample was imaged with the epifluorescent microscope.

Imaging Flow Cytometry acquisition, compensation and analysis for laboratory samples

Samples were acquired using the ImageStreamX MarkII machine (ISX, Amnis, Luminex). Three excitation wavelengths were used: 405 nm (DAPI - Channel 7- 50 mW), 561 nm (TMR - Channel 3 – 200 mW) and 642 nm (Cy5 - Channel 11 – 120 mW). For each sample, at least 50,000 cells were acquired and imaged with 60X magnification. Single stained samples were acquired using the compensation wizard of ISX. On average, less than 10 μL of 2x10^{^6} cell stock concentration were necessary to collect the optimal amount of cells, except towards the end of the experiment where cells were scarce.

Data was analyzed using IDEAS6.2 (Amnis, Luminex). The compensation matrix was built using the IDEAS wizard and manually checked, before being applied to all the acquired files. Based on the Area (the number of microns squared in a mask) and Circularity (the degree of the mask’s deviation from a circle) of DAPI three populations are identified as single cells (mainly DAPI area < 60 a.u), doublets and aggregates (mainly DAPI area > 60 a.u). Single cells were additionally selected in the same focal plane using the BF gradient and contrast (both gradient and contrast measure the sharpness quality of an image by detecting large changes of pixel values in the image). All gates were defined on a single file before being applied to the total data set (See Fig. S10). Each file was then manually inspected to check the accuracy of single cell and aggregates gating. All the data (fluorescent intensities, morphological features, populations) was then exported for each cell of each file for analysis in R. 

Similarity calculation. As defined in the IDEAs User Manual (Imagestream software) "The Similarity feature is the log transformed Pearson’s Correlation Coefficient and is a measure of the degree to which two images are linearly correlated within a masked region.". In Fig S4D, we correlate DAPI and mcp masks for each cell of two subpopulations (mcp+/psbA+, mcp+/psbA-) and plot its distribution. The threshold used to declare co-localisation was Similarity > 1.2

Software

The files can be open with the software IDEAS 6.5.

Files types:

Files with .cif extension contain the compensation matrix used to analyze the data based on single stain acquisition.

Files with .daf extension contain the actual data for each single cell of that time point, as well as the gates used to define the different populations.

IDEAS requires both .cif and .daf files for a given sample.

Naming scheme:

Files are named with the following scheme: Treatment_Replicate_TimePoint

If Treament = Ctrl: the samples are from non-infected E. huxleyi cultures

If Treament = EhV201: the samples are from E. huxleyi cultures infected with EhV201

Replicate = R1-R2-R3

The Control R1 was kept for microscopy controls.

TimePoint = 0, 8, 24, 48, 72 hours post infection (hpi).