Synergistic Viral-Bacterial Co-Infection Accelerates Algal Bloom Collapse

This study examined interactions among the coccolithophore Gephyrocapsa huxleyi, its lytic virus EhV-M1 and associated bacteria during laboratory experiments and induced blooms. The researchers isolated the alga, virus and bacterial assemblage (BAM) from a 2018 AQUACOSM VIMS-Ehux mesocosm experiment near Bergen, Norway, then combined culture experiments, flow cytometry, single-molecule fluorescence in situ hybridization (smFISH), 16S rRNA sequencing, bacterial genome analysis, chemotaxis assays and field samples from North Atlantic blooms. They quantified how bacteria altered viral algal mortality and how viral infection affected bacterial growth, community composition and attraction to algal cell extracts.

Isolation of the alga, virus and BAM

G. huxleyi strain RCC6946, the virus EhV-M1 and the bacterial assemblage BAM were isolated from an induced G. huxleyi bloom during the AQUACOSM VIMS-Ehux mesocosm experiment near Bergen, Norway, in 2018²⁴. To isolate the alga, researchers collected water from mesocosm bag 4 during the bloom phase, on day 13 (6 June 2018), and isolated individual cells at the Roscoff Culture Collection (RCC) within two weeks.

BAM was isolated from mesocosm bag 7, which showed the greatest increase in viral abundance as the bloom declined. Water collected during the demise phase, on day 23 (16 June 2018), was filtered through GF/F glass microfibre filters with a nominal pore size of approximately 0.7 µm. The filtrate was cryopreserved with glycerol at a final concentration of 15%. EhV-M1 was isolated from the same mesocosm experiment using G. huxleyi CCMP 374 as a susceptible host, and its genome was sequenced⁶².

G. huxleyi cultures and viral infection experiments

RCC6946 was maintained in batch culture at 18 °C under a 16 h light:8 h dark cycle. Cool-white LEDs supplied 100 µmol photons m⁻² s⁻¹. The culture medium consisted of autoclaved, filtered seawater (FSW) adjusted to 35‰ and supplemented with modified K/2 medium⁶³, as described in Supplementary Data 3. An axenic culture was generated from the xenic culture by adding ampicillin and kanamycin at final concentrations of 100 µg ml⁻¹ and 50 µg ml⁻¹, respectively.

The absence of bacteria was checked periodically by plating 100 µl on Difco Marine Agar 2216 (MA; BD Diagnostics) and by counting bacteria with flow cytometry (Extended Data Fig. 12e). Fresh medium was added by dilution to keep the algal cultures in exponential growth.

For infection experiments, 20 ml of exponentially growing axenic cultures, containing approximately 8 × 10⁵ cells ml⁻¹, were infected with freshly propagated EhV-M1. The virus was grown on RCC6946 and concentrated using 100 kDa Amicon filters. The multiplicity of particles (MOP) was 1:2, expressed as alga:virus.

To test whether antibiotics affected algal mortality, ampicillin and kanamycin were added to xenic RCC6946 cultures at final concentrations of 100 µg ml⁻¹ and 50 µg ml⁻¹, respectively, at the same time as viral co-infection at an alga:virus MOP of 1:5. The experiment used four biological replicates and was independently repeated twice with similar results. Log₁₀-transformed data were analysed with a two-sided linear mixed-effects model, followed by Holm-adjusted pairwise comparisons between treatments at each time point using lmerTest (v3.2-1) and emmeans (v2.0.3) in R (v4.5.1).

Infections of additional bloom-forming microalgae

Xenic cultures of the haptophyte Phaeocystis globosa G(A), the green alga Micromonas commoda LAC38 and Ostreococcus tauri RCC4221 were used to propagate their respective viruses: P. globosa virus PgV-07T, M. commoda virus McV-BIO217, and the O. tauri viruses OtV5, a specialist virus, and OtV06-12, a generalist virus. The procedures followed those described above, with modifications intended to slow infection and permit bacterial synergy.

P. globosa was grown in f/2 medium⁶⁴ and infected at 5 × 10⁵ cells ml⁻¹ with PgV-07T filtered at 0.22 µm, using an MOP of 1:0.1. M. commoda was grown in FSW adjusted to 27‰ and supplemented with f/2 medium, then infected at 10⁶ cells ml⁻¹ with McV-BIO217 filtered at 0.45 µm, using an MOP of 1:1. O. tauri was grown in K/2 medium at 50 µmol photons m⁻² s⁻¹. It was infected either with OtV5 at 10⁷ cells ml⁻¹, using virus filtered at 0.45 µm and concentrated on 100 kDa filters at an MOP of 1:0.001, or with OtV06-12 at 10⁶ cells ml⁻¹, using virus filtered at 0.22 µm and concentrated on 100 kDa filters at an MOP of 1:0.1.

Ampicillin and kanamycin were added at 100 µg ml⁻¹ and 50 µg ml⁻¹, respectively, on day −1 for P. globosa, day 0 for O. tauri infected with OtV5, or days 1 and 2 for M. commoda and O. tauri infected with OtV06-12, depending on the microbiome response. All experiments used three biological replicates and were independently repeated twice with similar results. Log₁₀-transformed data were analysed with two-sided linear mixed-effects models and Holm-adjusted pairwise comparisons at each time point.

Enrichment of BAM on G. huxleyi

The cryopreserved BAM was enriched on G. huxleyi using media containing organic material produced by algal exudation, mechanical lysis by sonication or viral lysis. To prepare exponential conditioned medium (Exp. CM) and sonication lysate (SL), algal cultures were grown to approximately 8 × 10⁵ cells ml⁻¹ and centrifuged at 600g for 5 min. The supernatant was passed through 0.22 µm PES filters and used as Exp. CM.

The cell pellet was resuspended in 0.3 ml K/2 medium and disrupted by probe sonication for five cycles of 30 s, each followed by 30 s of rest. The lysate was resuspended in 500 ml K/2 medium, filtered through 0.22 µm PES filters and used as SL. To prepare viral-free lysate (VFL), algal cultures were infected with EhV-M1 as described above. After four days, lysed cultures were passed sequentially through 2 µm cellulose membrane filters, 0.22 µm PES filters and 100 kDa centrifugal filters.

The three filtrates were stored at 4 °C and used within 24 h. BAM glycerol stock was inoculated directly into 30 ml of each filtrate in triplicate and incubated for three days in an orbital shaker at 18 °C. Triplicates were combined, and 9 ml aliquots of each medium were mixed with glycerol to a final concentration of 15% before cryopreservation.

Tripartite alga–virus–bacteria experiments

Algal cultures were infected with EhV-M1 and co-inoculated with an overnight BAM culture. BAM was grown overnight at 28 °C with agitation at 160 rpm in Difco Marine Broth 2216 (MB; BD Diagnostics). The culture was washed three times with 0.2 µm-filtered FSW by centrifugation at 13,800g for 1 min. Bacteria were counted by flow cytometry and added to virus-infected algal cultures at 10⁴ cells ml⁻¹.

The experiment used five biological replicates and was independently repeated at least twice with similar results. Statistical significance was assessed using log₁₀-transformed data, a two-sided linear mixed-effects model and Holm-adjusted pairwise comparisons at each time point.

Flow-cytometric quantification of algae, bacteria and viruses

Flow-cytometry measurements were performed with a CytoFLEX S instrument (Beckman Coulter). Algae were identified by high forward scatter (FSC) and high chlorophyll autofluorescence. FSC was measured with 488 nm excitation and a height threshold of 4 × 10⁴ arbitrary units (a.u.), while chlorophyll was measured with 561 nm excitation and emission at 665–715 nm. Only cells with chlorophyll autofluorescence above 4 × 10⁴ a.u. were counted (Extended Data Fig. 12a,b).

For bacterial and virus counts, samples were fixed with glutaraldehyde at a final concentration of 0.5% for at least 30 min at 4 °C, snap-frozen in liquid nitrogen and stored at −80 °C. After thawing, samples were stained with SYBR Gold (Invitrogen), diluted 1:10⁴ in Tris-EDTA buffer, for 20 min at 80 °C. Samples were analysed by plotting violet side scatter, measured with 405 nm excitation and an area threshold of 1 × 10³ a.u., against SYBR fluorescence, measured with 488 nm excitation and 500–550 nm emission with an area threshold of 1 × 10³ a.u. (Extended Data Fig. 12c–f).

Data were inspected and processed in CytExpert v2.4 (Beckman Coulter). Technical outliers caused by unstable timelines or shifts in signal intensity were removed. Statistical analyses were performed in GraphPad Prism v11.0.2 and R v4.5.1.

Detection of infected algal cells by smFISH

Alga–virus–bacteria experiments were conducted in 500 ml cultures. Each culture was sampled daily for flow-cytometric quantification of algae, bacteria and viruses and for smFISH analysis, following a previously described method³.

At each time point, 25–30 ml of culture was fixed with paraformaldehyde at a final concentration of 1% for 1 h and centrifuged for 3 min at 3,000g. The cell pellet was incubated with cryopreservation solution for 1 h, centrifuged again for 3 min at 3,000g and stored at −80 °C. Chlorophyll was removed with sequential 70% and 100% ethanol washes. Proteinase K, at a final concentration of 10 µg ml⁻¹ in 2× sodium chloride-sodium citrate (SSC), was used to degrade enzymes.

Cells were incubated overnight in hybridization buffer containing 17.5% formamide in 2× SSC and probes targeting the viral mcp transcript. The probes were conjugated to tetramethylrhodamine (TMR) and used at a final concentration of 0.1 ng ml⁻¹. Cells were washed with 17.5% formamide in 2× SSC, resuspended in 10 mM Tris at pH 8.0 in 2× SSC and analysed by flow cytometry using the probe fluorescence at 561 nm excitation and 564–606 nm emission.

Measurements used an FSC area threshold of 4 × 10³ a.u., more than 1,000 total events in the target gate at 30 µl min⁻¹ and an abort rate below 7%. Cells with TMR signal area above 1 × 10³ a.u. were classified as mcp-positive (mcp⁺) virus-infected cells (Extended Data Fig. 12g). Unstained samples provided the background-fluorescence reference. The experiment used four biological replicates and was independently repeated twice with similar results. Significance was assessed with a two-sided linear mixed-effects model and Holm-adjusted pairwise comparisons for each time point.

16S rRNA analysis of bacterial communities

Alga–virus–bacteria experiments were established in five 400 ml replicates. Forty-millilitre aliquots were gravity-filtered through autoclaved 47 mm GF/C filters (Whatman), and the filtrate was collected on autoclaved 47 mm, 0.22 µm PVDF filters (Durapore, Millipore). Filters were transferred directly into the lysis buffer of the DNeasy PowerWater Kit (Qiagen), frozen in liquid nitrogen and stored at −20 °C. DNA was extracted according to the manufacturer’s instructions.

The V4 region of the 16S rRNA gene was amplified with primers 515F (GTGYCAGCMGCCGCGGTAA)⁶⁵ and 806R (GGACTACNVGGGTWTCTAAT)⁶⁶. PCR consisted of an initial 3 min denaturation at 94 °C, followed by 35 cycles of 94 °C for 45 s, 50 °C for 1 min and 72 °C for 1 min 30 s. A final 10 min extension was performed at 72 °C. Sequencing used paired-end 2 × 250 bp Illumina MiSeq at the RUSH Genomics and Microbiome Core Facility.

Reads were processed with DADA2 v1.28.0⁶⁷ in R v4.3.2 for quality filtering, error correction and amplicon sequence variant (ASV) inference. Quality was assessed with plotQualityProfile; low-quality bases were removed with filterAndTrim, using truncation values selected from quality-score distributions. Error rates were learned with learnErrors, reads were dereplicated with derepFastq, ASVs were inferred with dada and chimeras were removed with removeBimeraDenovo.

Taxonomy was assigned using the SILVA_nr99 reference database⁶⁸ v138.1. Mean sequencing depth was 67,413 ± 26,969 reads per sample, ranging from 24,150 to 132,743. Chloroplast reads represented 28 ± 23%, with a maximum of 73%, and were removed. Processed data were analysed with Phyloseq v1.44.0⁶⁹ in R. The nine most abundant genera were shown individually; genera accounting for less than 5% were combined as “Others”.

Calculation of SI and BI

The synergy index (SI) compared the area under the curve (AUC) for virus-infected G. huxleyi growth in the absence and presence of bacteria. The complete growth curves were used because AUC represents cumulative biomass change, captures the full temporal dynamics, incorporates three measurements from each experimental day and is less sensitive to experimental variability than a single time point. Normalizing the AUC difference by another AUC produces a dimensionless measure that can be compared across bacterial strains and experimental conditions.

SI was expressed as:

Here, AUC_virus is the algal population dynamics during viral infection alone, and AUC_tripartite is the algal population dynamics during co-infection with virus and bacteria.

AUC was calculated for every replicate and treatment with GraphPad Prism v11.0.2. The mean AUC for each treatment (n = 3) was used to calculate SI. Its standard error was estimated by analytical error propagation using the delta method:

The variance was calculated as:

A repeated-measures two-way ANOVA, using a two-tailed significance level of α = 0.05, tested treatment differences between tripartite and virus-only demise dynamics and evaluated the effects of time and bacterial strain on algal abundance. Because calcification can cause cells to sediment, raw data were detrended with a linear mixed-effects model using a fixed slope and a random intercept for each day in the robustlmm package v3.4-3 in R v4.5.1. Results are reported as SI ± SE.

The bacterial index (BI) quantified differences in bacterial growth over time by comparing AUC under tripartite conditions with AUC for bacterial growth alone:

In this equation, AUC_tripartite represents bacterial growth when co-cultured with virus-infected G. huxleyi, while AUC_bacteria represents bacterial growth with G. huxleyi but without viral infection.

Bacterial AUC was calculated for each replicate and treatment between days 1 and 6. Day 0 was excluded because the bacterial inoculum was below the flow-cytometry detection limit. Mean treatment AUCs (n = 3) were used to calculate BI. The standard error was estimated as:

Using the delta method, the variance was calculated as:

A repeated-measures two-way ANOVA with a two-tailed α of 0.05 evaluated differences between tripartite and bacterial growth dynamics and tested the effects of time and bacterial strain on bacterial abundance. A mixed-effects model was used when values were missing. Results are reported as BI ± SE.

Isolation of bacteria from BAM and induced blooms

Glycerol stocks were sampled on different days of the induced G. huxleyi blooms and after various enrichment procedures (Supplementary Data 1). Seawater was collected during the bloom phase on day 13 and the demise phase on days 21 and 23. Samples were filtered through 10 µm and GF/F filters, cryopreserved with 15% glycerol, snap-frozen in liquid nitrogen and stored at −80 °C. BAM glycerol stock was prepared as described above.

To enrich taxa associated with algae and viral infection, mesocosm and BAM stocks were inoculated into uninfected and EhV-M1-infected G. huxleyi cultures before bacterial isolation. To enrich G. huxleyi-associated taxa, xenic RCC6946 cultures containing bacteria co-isolated with algal cells from the mesocosm bloom were infected with EhV-M1 before isolation.

Overnight cultures from glycerol stocks and enrichment cultures were plated at 100×, 1,000× and 10,000× dilution, using 100 µl per dilution on MA plates. Plates were incubated at 28 °C until individual colonies appeared. Selected colonies were streaked onto fresh MA three times or until uniform colony morphology indicated purity. Colonies were grown overnight in 3 ml MB at 28 °C with agitation at 160 rpm and then cryopreserved in glycerol. Tripartite experiments used biological triplicates and were independently repeated twice with similar results.

For complete 16S rRNA sequencing, cells from pure colonies were lysed in TE buffer containing 0.1% Triton-X100 by heating at 95 °C for 5 min. Lysates were centrifuged at 13,300 rpm for 10 min, and 2 µl of supernatant was used for PCR. Universal primers 16S-27F (AGAGTTTGATCMTGGCTCAG)⁷⁰ and 16S-1492R (CGGTTACCTTGTTACGACTT)⁷¹ were used in 20 µl reactions with 2× HY-Taq (HyLabs) or REDTaq (Sigma) Ready Mix.

PCR involved 3 min at 94 °C, followed by 25–35 cycles of 94 °C for 30 s, 57 °C for 30 s and 72 °C. A final 10 min extension at 72 °C was followed by a 12 °C hold. Products were visualized on 1% w/v agarose gels containing ethidium bromide and purified with the illustra ExoProStar 1-Step enzymatic cleanup kit (GE Healthcare Life Sciences). A 5 µl aliquot was incubated at 37 °C for 15 min and then at 80 °C for 15 min to inactivate the enzymes. Purified amplicons were Sanger-sequenced by the Life Sciences Core Facilities at the Weizmann Institute of Science. Sequences were classified to genus level with the SINA classifier tool⁷² v1.2.12 and the SILVA rRNA database.

Quantification of algal cell death with SYTOX

Alga–virus–bacteria cultures were sampled daily for flow-cytometric enumeration of algae, bacteria and viruses and for measurement of dying, membrane-permeable cells using SYTOX Green (Invitrogen). Samples were stained with 1 µM SYTOX Green for at least 20 min in the dark at room temperature.

Flow cytometry used an FSC height threshold of 2.5 × 10⁴ a.u. and a violet side-scatter area threshold of 1 × 10³ a.u. SYTOX fluorescence was measured with 488 nm excitation and 505–545 nm emission. In samples with at least 400 cells in the parent gate, cells with SYTOX signal above 1 × 10⁵ a.u. were counted as SYTOX⁺, membrane-permeable cells (Extended Data Fig. 12h). Cells with chlorophyll autofluorescence above 2 × 10⁵ a.u. were counted as high-chlorophyll cells. Unstained samples provided the background reference. The experiment used four biological replicates and was independently repeated twice with similar results. Significance was assessed with a two-sided linear mixed-effects model and Holm-adjusted pairwise comparisons for each time point.

Genome sequencing and metabolic pathway analysis

Bacterial isolates listed in Supplementary Data 1 were grown overnight at 28 °C with agitation at 160 rpm in MB. Cultures were centrifuged at 12,000 rpm for 1 min at room temperature, pellets were washed three times with FSW and frozen in liquid nitrogen. Genomic DNA was extracted using a modified version⁷³ of a published protocol⁷⁴.

DNA integrity was checked by gel electrophoresis or with Genomic DNA ScreenTape on a TapeStation 4150 (Agilent Technologies). Concentration and purity were measured with a Qubit fluorometer using the dsDNA HS Assay Kit (Thermo Fisher Scientific) and a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific). Samples were concentrated when necessary with the NEBNext Ultra II FS DNA Library Prep Kit (New England Biolabs) or AMPure XP beads (Beckman Coulter).

For Illumina sequencing, shotgun metagenomic libraries were prepared with the NEBNext Ultra II FS DNA Library Prep Kit using enzymatic shearing. Libraries were sequenced on an Illumina NovaSeq 6000 S2 flow cell with 2 × 150-base reads. Reads were quality-trimmed with Trimmomatic v0.39 using: ILLUMINACLIP:TruSeq3-PE-2.fa:2:30:10, LEADING:10, TRAILING:10, SLIDINGWINDOW:4:15 and MINLEN:50.

Quality-filtered reads were assembled with SPAdes v3.13.0 using default k-mer sizes of 21, 33, 55 and 77 bp. Contigs were retained at mean coverage of at least 3× and length of at least 300 bp, then binned with MaxBin2 v2.2.7. Assembly quality was assessed using CheckM2 v1.0.2. Bins from the same isolate were concatenated into single genomes when their completeness and taxonomic annotations indicated that they complemented one another.

Genes were predicted with Prodigal v2.6.3 in metagenomic mode. Genes were retained if they were at least 300 bp long, began with a start codon at the first trimer and lacked a premature stop codon. Retained genes were clustered with MMseqs2 v13.45111 using coverage mode 0, minimum coverage 0.8 and minimum sequence identity 0.9. Representative genes were annotated with eggNOG-mapper v2.1.3.

For PacBio sequencing, multiplexed microbial libraries were prepared with the SMRTBell Express Prep Kit 2.0 (Pacific Biosciences). A single library was used for SMRTbell template preparation according to the manufacturer’s protocol. Templates were annealed to sequencing polymerase with a Sequel binding kit 2.0 and sequenced at a 12 pM on-plate concentration on a Sequel I System in CLR mode, using one 1M SMRT Cell (PacBio) at the WIS Genomics Unit. Polymerase reads were processed with SMRT Link v8.0. Genomes were assembled with Trycycler v0.5.5 and Flye v2.9.6 using default parameters. Assembly assessment and gene prediction, filtering, clustering and annotation followed the Illumina workflow.

Nanopore sequencing was performed by Plasmidsaurus using its standard bacterial genome sequencing service, Oxford Nanopore Technology and custom analysis and annotation. Genome sequences were classified at species level with GTDB-Tk v2.4.1 and GTDB database v220⁷⁵ (Supplementary Data 1). The genome of Sulfitobacter sp. D7 was sequenced as previously described⁷⁶.

To examine links between bacterial metabolic pathways and synergy, KEGG pathways and their component KEGG orthologues were retrieved through the KEGG API. For each taxon, pathway coverage was calculated from eggNOG-based annotations as the proportion of pathway orthologues present in the strain relative to the total orthologues in that pathway. FAD-pathway coverage was correlated with SI using Spearman’s rank correlation, and the correlation coefficient ρ and corresponding P value were determined.

Algae and viruses in induced blooms

The abundance of G. huxleyi and intracellular EhV during the induced bloom in mesocosm bag 7 was measured as previously described²⁴. Water samples were pre-filtered through 40 µm cell strainers and immediately analysed with an Eclipse iCyt flow cytometer using ec800 software v1.3.7 (Sony Biotechnology). Calcified G. huxleyi cells were identified by high side scatter and high chlorophyll, measured with 488 nm excitation and 663–737 nm emission.

For intracellular EhV quantification, samples were pre-filtered through 20 µm filters, and DNA was extracted from 2 µm filters. EhV abundance was measured by quantitative PCR targeting the major capsid protein (mcp) gene. Absolute abundance was calculated from a calibration curve prepared with DNA at known concentrations.

16S rRNA sequencing during induced blooms

Planktonic bacteria larger than 0.2 µm and smaller than 2 µm, and particle-attached bacteria larger than 20 µm and smaller than 200 µm, were assessed throughout the induced bloom in mesocosm bag 7 using 16S rRNA sequencing as previously described²⁴.

Water was pre-filtered through 200 µm and then passed through 20 µm, 2 µm and 0.22 µm polycarbonate filters (Isopore, 47 mm; Merck Millipore). Filters were immediately frozen in liquid nitrogen. DNA was extracted from the 20 µm and 0.22 µm filters with the DNeasy PowerWater kit (Qiagen). Sequencing followed the EMP 16S amplicon protocol targeting the V4 region with 515F-806R primers⁷⁷, including degeneracy to reduce bias against Thaumarchaeota and SAR11^65,66.

Particle-attached samples were sequenced on an Illumina MiniSeq paired-end 2 × 153 bp platform at the RUSH Genomics and Microbiome Core Facility. Planktonic data were analysed with DADA2 and ASVs were annotated with the RDP classifier. For particle-attached samples, DADA2 was run with “justConcatenate = TRUE”, merging forward and reverse reads with 10× “N” inserted between them, and ASVs were annotated with SILVA_nr99. Mean sequencing depth for the particle-attached fraction in bag 7 was 20,734 ± 6,517 reads per sample, ranging from 8,057 to 29,629. Chloroplast reads represented 7 ± 8%, ranging from 1% to 36%, and were removed.

Kruskal–Wallis tests compared planktonic and particle-attached communities for each family and bloom phase: pre-bloom, days 1–6; bloom, days 7–14; and viral infection, days 15–23. Pairwise Welch’s t-tests with Holm–Bonferroni correction compared log₂-transformed fold changes between phases. Relative abundances were interpolated for the missing day 2 planktonic sample and day 1 particle-attached sample. Pseudoalteromonadaceae was absent from some planktonic samples, preventing calculation of its log₂ fold change.

Chemotaxis assays during induced blooms

To prepare chemoattractants, G. huxleyi CCMP 2090 was grown in two 19 l carboys containing modified K/2 medium with stirring and aeration. One exponentially growing culture, at 1.2 × 10⁶ cells ml⁻¹, was infected with EhV 201 at an MOP of 1:1; the other was an uninfected control. Viruses were propagated on CCMP 2090, washed, concentrated on 100 kDa Amicon-ultra filters (EMD Millipore) and stored at 4 °C for use within two months. Cells were counted daily with a Multisizer 4 Coulter Counter (Beckman Coulter).

After 48 h, uninfected and infected cultures contained 4.4 × 10⁶ and 0.9 × 10⁶ cells ml⁻¹, respectively. Cultures were filtered onto pre-combusted 47 mm GF/C filters (Whatman) using a gentle vacuum of 600 mbar under-pressure. Filters were transferred immediately to 500 ml pre-cooled methanol at −20 °C, extracted overnight at 4 °C while shaking, sonicated for 30 min and filtered through pre-combusted GF/C filters. Extracts were evaporated to 20 ml with a Rotovap at 30 °C and 70–100 rpm, centrifuged at 20,800g for 10 min at 4 °C, and dried completely in a SpeedVac at 28 °C and 1,980 rpm.

Dried extracts were redissolved in methanol to reduce salt, divided among eight centrifuge tubes, dried again and stored at −80 °C. Dry weights were 57.5 ± 0.3 mg for the uninfected extract and 38.3 ± 0.5 mg for the infected extract.

In situ chemotaxis assays (ISCAs) containing these extracts were deployed on days 3, 9, 12, 18 and 20 of the mesocosm succession. For each deployment, 12.5 ml of bloom water was collected from each of mesocosm bags 1–4, combined to 50 ml and subjected to quadruple filtration: one pass through a 0.2 µm Millex FG filter, two passes through a 0.2 µm Sterivex filter and a final pass through a 0.02 µm Anatop filter (all Millipore or Whatman). The filtration was intended to remove microorganisms and was validated by flow cytometry and DNA sequencing⁷⁸.

Filtered bloom water served as the ISCA control and was also used to resuspend dried extracts at 1 mg ml⁻¹. Resuspended extracts were passed through a 0.2 µm Millex FG filter to remove undissolved material. This concentration represents environmental hotspots: chemical concentrations decline exponentially with distance from the ISCA opening, so concentrations sensed by bacteria in surrounding water are substantially lower than those inside the wells and are within the same order of magnitude as concentrations produced by lysis of a phytoplankton cell⁷⁹.

Filtered bloom water and the two cell extracts were randomly assigned to ISCA rows, each containing five 110 µl wells. Each treatment was replicated on three ISCA devices, which served as biological replicates (n = 3). Devices were placed in flow-damping acrylic enclosures using an established protocol⁸⁰. Each enclosure was filled with equal-volume mixtures of water from mesocosm bags 1–4, sealed and incubated in situ at 1 m depth for 2 h⁷⁸. After retrieval, wells were sampled with sterile 1 ml syringes and 27G needles.

ISCA samples for DNA extraction were snap-frozen. DNA was extracted using a microvolume physical-lysis method⁸¹. Libraries were prepared with the Nextera XT DNA Sample Preparation Kit (Illumina) following a low-input protocol⁸². Primers targeting the V4–V5 region were 515F and 926R (CCGYCAATTYMTTTRAGTTT)⁶⁵. Libraries were sequenced on an Illumina MiSeq platform with paired-end 2 × 300 bp reads, yielding approximately 300 bp per sample. DADA2 and the SILVA_nr99 database were used for community analysis. On day 18, a Welch’s ANOVA followed by two-sided pairwise t-tests with Benjamini–Hochberg correction compared Alteromonadaceae fractions among chemoattractants. The R “stats” package v4.5.1 was used, with Shapiro–Wilk and Levene tests assessing normality and variance homogeneity.

Isolation and chemotactic specificity of bloom-associated bacteria

To isolate chemotactic bacteria, agar plates were prepared with 0.22 µm-filtered bloom water and supplemented with 1 mg ml⁻¹ of the relevant G. huxleyi extract. Plates were inoculated with bacterial communities from ISCA wells containing uninfected or virus-infected extracts and incubated at 10 °C, the fjord-water temperature, until colonies appeared.

Individual colonies were grown overnight in 10% MB prepared with autoclaved seawater, at room temperature and at 10 °C. Each isolate underwent two further purification rounds on plates and in liquid culture. Isolates were identified by 16S rRNA Sanger sequencing⁷⁸. Selected strains listed in Supplementary Data 1 were used for genome sequencing and chemotaxis characterization.

For genome sequencing, overnight cultures were inoculated into MB at 0.1 OD₆₀₀. Exponentially growing cultures at 1.0 OD₆₀₀ were centrifuged, and 15–30 mg cell pellets were resuspended in 0.5 ml Zymo DNA/RNA Shield. Plasmidsaurus performed DNA extraction and Nanopore sequencing using its standard bacterial genome sequencing service. Assemblies were classified at species level with GTDB-Tk.

For laboratory chemotaxis assays, strains were plated from cryostocks on MA. Several colonies from each strain were inoculated into 1% MB in artificial seawater containing 33 g l⁻¹ Instant Ocean (Aquarium Systems) and incubated at 180 rpm and 27 °C for 4 h to promote motility. An aliquot was fixed with 1% glutaraldehyde, stained with SYBR Green I for 15 min and counted with a CytoFLEX S flow cytometer.

Each ISCA was loaded with four filtered treatments and incubated at room temperature for 1 h in artificial seawater containing 10⁶ bacteria ml⁻¹. Treatments were uninfected CCMP 2090 extract, virus-infected CCMP 2090 extract, 10% MB as a positive control and artificial seawater as a negative control for random entry. Extracts were used at 1 mg ml⁻¹. Treatments were introduced with sterile 1 ml syringes and 27G needles. Each row of five wells was a technical replicate, and each treatment was deployed simultaneously in four ISCA devices.

After incubation, well contents were recovered with sterile syringes and needles, technical replicates were pooled⁸⁰, and 200 µl per sample were fixed, stained and counted by flow cytometry. Chemotactic index (I_c) was the cell concentration in treatment wells divided by the mean artificial-seawater control concentration. I_c values were log1p-transformed, as ln(1 + x). Levene and Shapiro–Wilk tests assessed variance homogeneity and normality.

All bacteria differed significantly between the artificial-seawater control and the treatments, including the MB control, according to Dunnett’s test with P < 0.001. Differences between the two cell extracts were tested with a two-sided Welch’s t-test and Holm–Bonferroni correction. For fold-change analysis, transformed I_c values were log₂-transformed, normality was checked with a Shapiro–Wilk test, and strain-specific treatment differences were assessed with two-sided Welch’s t-tests and Benjamini–Hochberg correction. Log₂-transformed fold change (LFC) between infected and control extracts was calculated as LFC = mean(log(Inf)) − mean(log(Ctl)), with s.e.m. = √(Var_Inf/n_Inf + Var_Ctl/n_Ctl).

North Atlantic coccolithophore blooms

Coccolithophore blooms were sampled during the NA-VICE cruise in 2012, which provided broad spatial coverage¹⁸, and during the Tara Breizh cruise in 2019, which provided high temporal resolution⁴³.

During NA-VICE, satellite remote sensing identified blooms along a 2,000-mile transect between the Azores and Iceland. Samples were collected at 30 stations aboard the RV Knorr from 13 June to 16 July 2012, as previously described¹⁸. Water was collected between the surface and 150 m with Niskin bottles, pre-filtered through 200 µm and collected on 0.8 µm polycarbonate filters.

DNA was extracted and amplified in a dual-PCR protocol targeting the V1–V3 region of the 16S rRNA gene. The primers were CS1_27F_YM (ACACTGACGACATGGTTCTACA-AGAGTTTGATYMTGGCTCAG) and CS2_534R_Y (TACGGTAGCAGAGACTTGGTCT-TYACCGCGGCTGCTGG)⁸³. The first PCR used AccuPrime SuperMix II (Life Technologies): 5 min at 95 °C, 28 cycles of 95 °C for 30 s, 56 °C for 30 s and 68 °C for 5 min, followed by 7 min at 68 °C. A second PCR added sample barcodes using 5 min at 95 °C, eight cycles of 95 °C for 30 s, 60 °C for 30 s and 68 °C for 30 s, followed by 7 min at 68 °C. Products were pooled, purified with AMPure XP beads and quantified with Qubit. Sequencing used paired-end 2 × 300 bp Illumina MiSeq at the DNA Services Facility, University of Illinois at Chicago.

During the Tara Breizh cruise, satellite observations identified a G. huxleyi bloom patch in the Celtic Sea. The SV Tara sampled the patch from 27 May to 2 June 2019, as previously described^21,43. Samples were collected at maximum turbidity depth by pumping water through a 20 µm prefilter and sequentially filtering it onto 3 µm and 0.2 µm polycarbonate filters. V4–V5 16S rRNA amplicons were sequenced twice with paired-end 2 × 250 bp Illumina MiSeq. FASTQ files were obtained from the European Nucleotide Archive (ENA: PRJEB50692).

Intracellular virus abundance was measured from DNA extracted from the 3 µm filters using droplet digital PCR with primers targeting the EhV mcp gene. Flow-cytometry samples were collected at bloom depth with a Niskin bottle and immediately analysed using a BD Accuri C6 Plus Personal Flow Cytometer (BD Biosciences). Calcified G. huxleyi cells were identified by high chlorophyll, measured with 488 nm excitation and 663–737 nm emission, and high side scatter.

All paired-end amplicons were processed with QIIME 2⁸⁴ and ASVs were inferred with the DADA2 paired-end workflow. For NA-VICE V1–V3 reads, 20 bp were trimmed from the 5′ ends of both reads, leaving 280 bp of forward sequence and 230 bp of reverse sequence. Samples retaining fewer than 10,000 non-chimeric reads were excluded. Taxonomy was assigned with the QIIME 2 classify-sklearn classifier and SILVA v138, and ASV tables were aggregated to genus-level relative abundance.

Tara Breizh V4–V5 data were processed with QIIME 2 as previously described⁴³, with additional processing using SeqKit2 v2.3.1⁸⁵ and CutAdapt v4.2. Empty reads were removed, read pairs were repaired and primer orientation was handled in two branches corresponding to the original and swapped R1/R2 orientations. Reads lacking primers, shorter than 75 bp or containing insertions or deletions were removed. DADA2 denoising used pseudo-pooling. Reads were truncated to 215 bp forward and 190 bp reverse in one branch, and 190 bp forward and 215 bp reverse in the other. The second branch was reverse-complemented before merging.

ASV identifiers were normalized by sequence content and feature tables were merged by summing identical sequences across runs and branches. Chimeras were removed with de novo UCHIME in vsearch v2.22.1⁸⁶, and non-chimeric counts from duplicate sequencing runs were summed. Taxonomy was assigned with the SILVA v138 99% Naive Bayes classifier. For Fig. 5, samples with exactly zero relative abundance received a pseudocount of 1 × 10⁻⁵. Sample-location maps were produced with cartopy v0.25.0 in Python v3.10 using Natural Earth map data.

Comparison with other oceanic bacterial communities

To compare bacterial relative abundances from North Atlantic coccolithophore blooms with other marine environments, taxonomic profiles from Tara Oceans⁴⁴ were obtained from https://www.ocean-microbiome.org/. Relative abundance at genus and family levels was calculated by dividing each taxon’s read count by the total bacterial and archaeal reads in the sample.

Only planktonic bacterial fractions were used: 0.2–3 µm and 0.2–1.6 µm size fractions. NA-VICE data were restricted to cast samples, excluding trap samples. For Tara Breizh, samples labelled as outside the bloom were excluded. Pairwise comparisons were made at matched taxonomic levels and depth ranges.

Alteromonas relative abundance from the surface to 150 m was compared between Tara Oceans and NA-VICE. Because limited sequencing depth reduced reliable detection of Alteromonas, the Tara Oceans–Tara Breizh comparison used the Alteromonadaceae family from the surface to 25 m. Differences were tested with two-sided Mann–Whitney U tests. For Extended Data Fig. 11, samples with exactly zero relative abundance received a pseudocount of 1 × 10⁻⁵. Maps were made with cartopy v0.25.0 in Python v3.10 using Natural Earth map data.

Reporting summary

Additional information about the research design is available in the Nature Portfolio Reporting Summary linked to the article.

Source: www.nature.com