Strains, media and cultivation

The diatom C. radiatus was obtained from the Roscoff culture collection RCC 7277 (https://roscoff-culture-collection.org/rcc-strain-details/7277). The bacteria Mameliella sp. CS4 and Marinobacter sp. CS1 were isolated in our lab previously22. C. radiatus was cultivated in sterile Guillard’s (f/2) enrichment medium (G9903, Sigma-Aldrich) prepared with natural seawater (ATI). To optimize the growth of diatoms, silicate was added to reach the final concentration of 400 μM and the medium was noted as f/2+Si. Depending on the volume, cultures were maintained in tissue culture flasks (T175 for 150 ml cultures, T25 for 20 ml cultures, Sarstedt). Cultures (300 μl) were grown in 48-well plates, and 100 μl cultures were grown in 96-well plates (Sarstedt). Cultures were maintained at 13 °C with a light:dark cycle of 15 h:9 h and light intensity of 25 μmol m−2 s−1 for running experiments and 20 μmol m−2 s−1 for storage.

For the preparation of purified cultures, we applied a previously used method to minimize the bacterial community from C. radiatus22. Briefly, the cells were washed with 20 μg ml−1 Triton X-100 (X100, Sigma-Aldrich) on a filtering device (43-50040-51, 40 μm pore size, pluriSelect), followed by treatment with the antibiotics kanamycin (50 μg ml−1) and ciprofloxacin (20 μg ml−1) (K1377 and 23265, Sigma-Aldrich) overnight. After another filtering process, the antibiotics were removed by repeated centrifugation and washing with excess fresh f/2+Si medium. Cells were resuspended in a sterile Falcon tube. Young cells were stored under low light intensity in an incubator and used within 1 week. Old cells were used for the experiment on the same day of the filtration process. No bacteria were detectable under the microscope and after plating out on marine broth (MB, Sigma-Aldrich) agar plates (1.5% agar). Of note, since the diatoms are closely associated with bacteria, the cultures were strongly reduced in bacteria after the treatment but not necessarily entirely axenic.

Bacterial cultures were prepared from single colonies picked from marine broth (MB 76448, Sigma-Aldrich) agar plates (1.5% agar). Picked colonies were transferred into 3 ml liquid MB medium in 10 ml sterile tubes and grown at 28 °C with 150 r.p.m. shaking. When the cultures reached the late exponential phase at an optical density at 600 nm (OD600) of 0.4–0.7, 30 μl were transferred into 3 ml of fresh MB. When the cultures reached an OD600 = 0.4–0.7, they were collected by centrifugation at room temperature, washed two times with f/2+Si medium to remove the MB medium and resuspended in f/2+Si medium to reach an OD600 = 0.4 (~5 × 106 cells per ml).

Inorganic nutrients

For the determination of the inorganic nutrient content, C. radiatus RCC 7277 cultures were grown in 150 ml of f/2+Si medium as described above. Every other day, 2 ml were sampled and 100–300 μl of this sample were transferred to 96-well plates for counting of live and EV-producing cells. Nitrate, phosphate and silicate were determined by spectrophotometric methods following previously reported procedures41.

C. radiatus aging

Young and old cultures of C. radiatus were selected on the basis of two criteria: (1) The cultivation time after initial inoculation was less than 14 days for young cultures and longer than 14 days but less than 20 days for old cultures grown in 150 ml medium. (2) When most of the cells in the culture showed light yellow pigmentation under the microscope and the chloroplast spots were visible with a clear edge, the culture was regarded as young. When most of the cells in the culture showed dark brown pigmentation, the culture was classified as old (Fig. 1b).

Cell counting, time-lapse video capture and vesicle observation

Algal cell counting was done under a Leica DM2500 microscope with a scanning objective lens (×4) combined with a ×10 eyepiece lens. The microscope was equipped with a CCD system and the software NIC-elements D 4.30.00 was used for the photographic documentation. Live and dead cells were distinguished according to previous criteria22 and live cells were counted to monitor algal growth. When the individual live cell is attached to or surrounded by vesicle(s), this cell was recorded as an EV-producing cell. The number of EV-producing cells was normalized to the number of total live cells to give the EV-producing cell percentage. For 150 ml cultures, samples of 300 μl were pipetted to 96-well plates. After inoculation of the well plate at 13 °C for 6 h to allow time for EV production, the live cells and EV-producing cells were recorded as described above. For the alga cultivated in well plates, we directly counted the total cells from each well (100 μl for 96-well plate and 300 μl for 48-well plate). The counting was done within 1 h in the microscope room where the temperature was maintained at 18 °C. No obvious negative effects on C. radiatus caused by the short period outside the culture room could be observed. In a single-well counting pretest, the recognition of live cells and EV-producing cells was done with five independent observers (Yun 31 (live)/10 (EV producers), Vera 29/11, Fatemeh 30/9, Mona 31/9 and Mimi 30/10). Time-lapse video capture was carried out using the same microscope system in the auto-exposure mode. The videos were generated from the time-lapse pictures and edited using the software ImageJ v.1.53c. The size distribution of EVs was determined by microscopy from an old culture. The diameters of all vesicles in the field of vision were determined using the ImageJ software. The fate of the vesicles was determined by evaluation of nine video recordings (among them Supplementary Video 1). Thirty-one EVs that were in contact with a neighbouring algal cell were observed over up to 18 h.

Single-cell tracking

Six independent cultures were grown to prepare old cells. The old cells were then purified to minimize bacteria following the process described above. The cultures were diluted to a density of 8 cells per ml and then 100 μl was added to each well of a 96-well plate. Individual cells were tracked under the microscope every day for 1 week except day 4. The final fate (rejuvenation or non-rejuvenation) of the cells producing EV was recorded.

cLSM observation of C. radiatus

The fluorescence probes, 5(6)-carboxy-2’-7’dichlorofluorescin diacetate (CDCFDA) (Sigma-Aldrich) and Nile Red (Cayman Chemical) were used for detection of ROS, autophagy and lipids in C. radiatus cells and extracellular vesicles. The stock solution of CDCFDA was prepared using dimethylsulfoxide at concentrations of 24.7 mM and 100 mM. Nile Red was dissolved in acetone at a concentration of 0.3 mM. The final concentration was 5 μM for CDCFDA and Nile Red. Cells and vesicles were stained by each dye for at least 15 min in the dark. C. radiatus micrographs were acquired using a cLSM 880 microscope (Zeiss) equipped with a ×20/0.8 Plan Apochromat objective (Zeiss) at a resolution of 1,024 × 1,024 pixels with a pixel scaling of 210 × 210 nm, zoom factor 2 and 12 bit depth. CDCFDA, Nile Red, as well as chlorophyll autofluorescence were excited with 25% transmission of a 405 nm laser diode and emissions were recorded via an main beam splitter (MBS) 405, with a pinhole of 1.04 AU, 1.03 µs pixel dwell time and 4 times averaging of unidirectional scans. Spectral windows for the detection channels were defined as 480–540 nm for CDCFDA, 554–638 nm for polar lipids of Nile Red staining, 515–585 nm for neutral lipids of Nile Red staining, as well as 670–740 nm for the chlorophyll autofluorescence running at a detector gain of 500–525 nm. The bright-field signal was acquired as the transmitted light of the 405 nm laser diode recorded by a transmitted light photo multiplier tube (T-PMT) at a gain of 200.

Quantification of ROS production induced by the bacterial strain CS1

CS1 in MB culture was collected by centrifugation at 13,200 g for 5 min and adjusted to OD600 = 0.2 with f/2+Si medium. Old C. radiatus cells were treated with dense (OD600 = 0.2) CS1 in f/2+Si medium at a ratio of 1:1 for 24 h. Control cultures were maintained in parallel. CDCFDA dye at 5 µM final concentration was then added to the culture. After washing with dye-free f/2+Si medium three times, the fluorescence of cells and EVs was recorded using a Varioskan Flash multimode reader (Thermo Fisher) with the following parameters: the black solid 96-well plate was shaken at a speed of 300 r.p.m. and diameter of 5 mm for 20 s and then stopped for 50 s before measurement. For CDCFDA fluorescence detection, the excitation wavelength was 488 nm with a bandwidth of 12 nm, and the emission wavelength was set to 520 nm. The acquisition mode was set as ‘read from plate top’. The measurement time was 100 ms. Multipoint mode was used with a safety zone of 1.4 mm.

To prepare the bacterial culture filtrates for metabolomics experiments, bacteria were transferred in f/2+Si medium supplemented with the amino acids glycine, glutamic acid, threonine, tyrosine, serine, leucine, isoleucine and valine (each at 0.3 mM final concentration) and grown to an OD600 of 0.1 (after ~2 days at 28 °C with 150 r.p.m. shaking). The bacteria were removed by centrifugation at 16,100 g for 1 min and subsequent filtration through 0.2 µm filters. The young and old C. radiatus cells were inoculated into these bacterial filtrates. The f/2+Si+amino acids medium was used as a control. For metabolomics experiments, incubations were run for 3 days. In a pre-experiment, it was verified that the amino acids did not affect algal growth and EV production.

Metabolomics analysis

For untargeted metabolomics profiling of algal intra- and extracellular metabolites, 20 ml of cultures were transferred to Falcon tubes (50 ml) and centrifuged at 1,200 g for 15 min. Approximately 12 ml of the supernatant containing dissolved metabolites as well as the vesicles were carefully transferred to new Falcon tubes. The remaining cultures were centrifuged again at 1,200 g for 5 min and the remaining supernatant was pipetted off and combined with the first fraction. The combined supernatant was loaded on a solid-phase extraction cartridge (SPE, Oasis PRiME HLB 3 cc column, Waters) and eluted according to manufacturer protocol.

The cell pellet was resuspended with 3 ml methanol (Sigma-Aldrich) and the suspension was sonicated using a Bandelin Sonopuls HD 2070 ultrasonic homogenizer for 20 s. The resulting lysate was centrifuged for 20 min at 4,000 g to remove cell debris. The resulting methanolic extracts were transferred into new glass vials. After dilution with water, the supernatant was loaded on an SPE cartridge (Oasis PRiME HLB 3 cc column, Waters) for extraction following manufacturer recommendations. All samples were dried in a Vacufuge plus vacuum concentrator (Eppendorf) overnight. The dried samples were dissolved in 300 µl methanol and transferred into 1.5 ml Eppendorf tubes for centrifugation at 16,100 g for 10 min. The supernatant was transferred into the insert of a glass vial and submitted for MS analysis. Volumes of 5 µl of each sample were combined to obtain the quality control (QC) pools for endometabolites and exometabolites. A 1 µl sample was injected into the UHPLC–HR–MS (UltiMate 3000 UHPLC Dionex) equipped with an Accucore C18 column (100 × 2.1 mm, 2.6 µm, Thermo Fisher) coupled to a Q-Exactive Plus Orbitrap mass spectrometer (Thermo Fisher). The LC separation was performed starting with 100% of A: aqueous phase (2% acetonitrile, 0.1% formic acid in water) and increasing with B: acetonitrile phase (100% acetonitrile) from 0.2 to 8 min until reaching 100% B. This was held for 3 min before switching back to 100% of the water phase and equilibration for 1 min. The flow rate was 0.4 ml min−1. The column oven was controlled at 25.0 °C. Mass spectrometry was conducted in positive and negative modes with a scan range of m/z 80–1,200 at a peak resolution of 70,000, and heated electrospray ionization was used for ionization. AGC target was set to 3 × 106 and maximum ion time was set to 200 ms. The MS/MS spectra of precursor ions, selected with an inclusion list, were obtained from cell extracts with the above-mentioned parameters and within an isolation window of m/z 0.4 at a peak resolution of 280,000 (NCE 15, 35, 45). The raw dataset was uploaded and is available at https://www.ebi.ac.uk/metabolights/editor/MTBLS5368/descriptors.

Quantification of methionine

Samples were prepared as described for the exometabolome samples, but before SPE extraction, isotope labelled 13C-methionine (490083, Sigma-Aldrich) dissolved in water was added to reach a final concentration of 20 μM for old cells and and 2 μM for young cells as internal standards. Quantification was performed by integrating peak areas of the labelled and unlabelled methionine pseudomolecular ion.

Statistical analysis and pathway assignment

Peak picking, deconvolution and tentative identification of the metabolites from raw data were performed using the Compound Discoverer software v.3.3.0.550 (Thermo Fisher). Mass tolerance for MS identification was 5 ppm, retention time tolerance was 0.2 min, minimum MS peak intensity was 1 × 105, and intensity tolerance for isotope search was 50%. Relative standard deviation value was set to 50%. The compound list was exported as a .csv file. The injection sequence of the samples and normalization factors according to cell density are listed in Supplementary Table 2. Statistical analysis and enrichment analysis were performed using MetaboAnalyst 5.0. Principal component analysis was performed to compare overall metabolite pattern similarities among intra- and extracellular extracts. Pairwise comparisons for endometabolomics and exometabolomics profiles were made between (1) old control/young control, (2) old CS1/old control, (3) young CS1/young control, (4) old CS4/old control and (5) young CS4/young control. The differential expression metabolites (P < 0.05, FC > 1.5) lists from each pair comparison were exported and used for drawing Venn diagrams using the tool https://bioinformatics.psb.ugent.be/webtools/Venn/. The annotated features were uploaded for enrichment analysis. The identity of selected ions was confirmed by comparison of retention time and MS2 spectra between the QC pool and analytical standards.

Bioassays of microalgal growth and EV production

Bioassays were carried out in 96-well plates. After treatment with antibiotics, young and old algal cells were inoculated with an initial cell density of ~20 cells per well (in 100 µl of f/2+Si medium). The respective compound was added to the culture medium as aqueous solution, and cell counts and EV counts were recorded every other day. The EV-producing cell percentage was calculated as EV-producing cells/total live cells × 100%. IC50 values were determined on day 7 or day 8, and EC50 values were determined on day 4 by fitting the resulting plots.

EVs extracts and effects on bacteria

Old cells were used for induction of EV production 2 days after transfer into fresh medium. Diatom cells were removed by filtration on 40 μm filters. The filtrates containing EVs were left to sediment for 30 min for EV enrichment. The supernatant was removed by pipetting to reach an EV density of 2,500–3,000 ml−1. EVs were lysed by freezing at −80 °C and thawing. The lysate was filtered through 0.2 µm filters to remove bacteria. The supernatant collected after EV sedimentation from the same preparation was treated identically and served as control. A total of 15 bacterial strains were cultivated as previously described22. Bacterial cultures with OD600 = 0.6 were collected by centrifugation at room temperature and resuspended in fresh MB medium to reach an OD600 = 0.4. For each well, 100 μl of test culture was prepared by combining 50 μl of EV extract or control extract, 1 μl of bacterial inoculum and 49 μl of MB medium in a 96-well plate. Bacterial growth was recorded by measuring OD600 using a Varioskan Flash multimode reader (Thermo Fisher) at room temperature for 40 cycles at 30 min per cycle.

Metabolomics profiling of FACS-purified EVs

The experimental workflow is depicted in Extended Data Fig. 10a. The EV enrichment was performed as described above. The enriched EV samples were stained with the ROS probe CDCFDA for 15 min in the dark before FACS analysis. Vesicles were analysed and sorted with a fluorescence activated cell sorter (BD FACS Aria Fusion (BD Biosciences)). A solution containing 3.5% sea salt (Instant Ocean, Aquarium Systems) was used as FACS running buffer. Sorting temperature was kept at room temperature. Vesicles were sorted into 5 ml FACS tubes (Falcon). Chlorophyll a fluorescence was detected in the BV650 channel. The ROS dye CDCFDA was detected in the FITC channel. As single staining for ROS dye CDCFDA could not be created, compensation was not performed. To control for media contamination in the sorted vesicle-containing droplets, Accu-drop beads (BD) were diluted into the corresponding supernatant and sorted as above. Data analysis was performed using BD FACSDiva Software v.8 (BD Biosciences). The total sorted events for each sample are listed in Supplementary Table 3. FACS-purified EV and control samples were dried in a Vacufuge Plus vacuum concentrator (Eppendorf) for 2 h. The dried samples were dissolved in 300 μl of methanol and centrifuged at 16,100 g for 10 min. The supernatant of 10 μl was loaded on the LC–MS for untargeted metabolomics profiling as described above. The raw dataset was uploaded to www.ebi.ac.uk/metabolights/MTBLS5401.

Analysis of EVs for fatty acids and oxylipins using UHPLC–MS/MS

The enriched EV preparation was performed as described above, except that EVs were collected from 2-day-old cells after treatment with 100 μM homocysteine or 10 μM methyl methionine to enhance EV production (Fig. 5b). Cell samples were prepared as described for the metabolomics analysis. All samples were dried in a Vacufuge Plus vacuum concentrator (Eppendorf) overnight. The measurement was based on a previously established protocol for lipidomics analysis42. Dried samples were resuspended in 100 µl of methanol and subjected to UHPLC–MS/MS. Fatty acids and oxylipins were analysed with a Nexera X2 UHPLC system (Shimadzu) and a QTRAP 5500 mass spectrometer (AB Sciex) equipped with a Turbo V Source and electrospray ionization. Lipids were separated using an ACQUITY UPLC BEH C18 column (1.7 μm, 2.1 × 100 mm; Waters) at 50 °C with methanol:water:acetic acid ratio of 42.0:58.0:0.01 (v/v/v) at a flow rate of 0.3 ml min−1 that was ramped to 80.8:19.2:0.01 (v/v/v) over 11 min and then to 98:2:0.01 (v/v/v) for 5 min. The QTRAP 5500 was operated in negative-ionization mode using scheduled multiple reaction monitoring coupled with information-dependent acquisition. Optimized parameters (collision energy, entrance potential, declustering potential, collision cell exit potential) for fatty acid and oxylipin analysis were adopted, and the curtain gas pressure was set to 35 psi. The retention time and at least 6 diagnostic ions for each compound were confirmed using an external commercially available standard (Cayman Chemical). Quantification was achieved using linear calibration curves for each metabolite.

Statistics and reproducibility

Sample size was determined on the basis of a previous study that reached a significant result (P < 0.05). For example, up to 6 replicates were set for the algal cultivation experiment for nutrients measurement, but 3–4 replicates were set for bacteria–diatom co-cultivation and bioassay experiments following previous experimental results for this diatom strain22. For metabolomics experiments, 5 independent biological replicates were used for extraction. For all assays showing error bars, the mean values and standard deviations or standard errors across multiple biological replicates were set as the measures of centre and spread. The number of replicates for each experiment and the type of statistical test used to determine significance are included in respective figure legends. Only statistical comparisons for which P < 0.05 are shown in graphs. The graphs, data statistical analysis and fitting for EC50 and IC50 values were applied using Microsoft Office Excel 2016 and GraphPad Prism v.8.00.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.