Precultivation of N. oceanica in f/2 media
Nannochloropsis oceanica IMET1 was inoculated on f/2 agar plates and grown for 6 days in darkness at 25 °C and then grown for 2 weeks at 50-µE white light until green colonies of N. oceanica were visible. Usually, one f/2 agar plate has 50–100 colonies.
Single clones from these colonies were inoculated in 100-ml modified liquid f/2 medium made with aquarium sea salt (Real Ocean, USA) supplemented with NaNO3 at 1 g/l as only nitrogen source. Cells were grown under continuous white light (roughly 80 ± 5 μmol photons/m2 s = µEinstein) at room temperature (RT) and aerated by bubbling with compressed air in the 250-ml flask. Cultures were grown until they reached the mid-exponential phase (optical density at 750 nm (OD750) of 5.0 ± 0.5). OD750 was measured using appropriate dilution in isometric germ-free sodium chloride solution. To check for bacterial contamination, 90 µl was taken from each culture and mixed with 10 µl 0.01 µg/µl DAPI (Solarbio C0060) . Mixtures were incubated in dark for 20 min, then analyzed with a fluorescence microscope and laser Ex 300–385 nm, Em 420 nm and dichroism at 400 nm. Bacteria stained by DAPI appear blue in field of vision, whereas algae due to chloroplast autofluorescence turn red.
Main cultivation in Erlenmeyer flasks
Appropriate cultures were taken from the preculture medium and diluted to OD750 of 1 in 100 ml; for +N, modified liquid f/2 medium with NaNO3 at 1 g/l, and for –N. only modified liquid f/2 medium was used. Each culture was transferred to a 250-ml flask and grown for 2 weeks at 50 µE, 25 °C  until OD750 of about 5 was reached and shaken once per day, before sorting of the two subpopulations in the sample. Each week, the medium was replaced with fresh one.
Recultivation in Erlenmeyer flasks
For the recultivation experiment to investigate stability of heterogeneity in the N. oceanica IMET1 cultures, FACS-sorted +N subpopulations were grown in modified liquid f/2 medium with 1 g/L NaNO3. To inoculate medium, 4 million cells of each +N P3 and +N P4 population (defined in “FACS analysis and cell sorting of N. oceanicacells”) were sorted onto a sterile 0.22-µm filter cartridge. Filter material containing the cells was extracted from the filter casing in the clean bench; the extracted filters  were then transferred to 50 ml +N f/2 medium in a 250-ml flask and grown for 40 days in a light cupboard with 50-µE light at 25 °C. Flasks were daily checked for growth and shaken. After cultures had reached an OD750 value of 1, filter membrane fragments were separated by gravitational precipitation overnight. The supernatant was centrifuged at 6800g for 10 min at room temperature (RT) and the cell pellet was resuspended in 200 ml +N f/2 and cultivated in 500-ml flasks as described in main cultivation in Erlenmeyer flasks until OD750 of 5 was reached in 2 weeks.
Main cultivation in small-scale photo bioreactor
Cells from preculture were injected to OD750 value of 0.1, and cultures grown in two 1-L glass columns (inner diameter: 5 cm) containing 600 ml +N f/2 medium until mid-exponential phase with OD750 = 5 ± 0.5, room temperature was 25 °C with 80 ± 5 μmol/m2 s continuous white light irradiation. Cultures were bubbled with compressed air. After 2-week culture, Nannochloropsis cultivations were individually transferred into centrifuge tubes and centrifuged at 6800g for 10 min at RT (HITACHI, centrifuge R9A rotor). Algae pellets were washed twice with 100-ml nitrogen-deficiency f/2 medium (f/2 medium without sodium nitrate). After centrifugation (Beckman Coulter Allegra X-12R, SX4750 Rotor, 2000g, 10 min) pellets from three cultures were resuspended with 600-ml modified f/2 medium containing 1 g/l NaNO3and pellets from the other three replicates were resuspended with 600-ml nitrogen-deficient f/2 medium. After 10 days of cultivation, cells were harvested from the cultures under a clean bench. Each cultivation was divided into three sections corresponding to the top, middle and bottom parts of the reactor.
Light microscope-based cell counting
Microscopic cell counting of N. oceanica cells was performed using a Thoma cell counting chamber. Cells were counted in the c-fields of one b-field of the Thoma cell counting chamber with a 40× objective. All c-fields in a b-field were counted. All c-fields were averaged over the four b-fields. The average number of cells for all c-fields was multiplied by 4 million to obtain the number of cells per milliliter.
Lipid staining of N. oceanica cells
Nile Red staining of neutral lipids was used to sort individual N. oceanica cells and to observe heterogeneity with FACS according to their lipid content. For staining with Nile Red, N. oceanica cells were aliquoted in the required number as established by cell counting. The aliquoted cells were washed three times using 35 g/l sea salt water (2380g, 5 min, Eppendorf 5804, at RT) and then diluted to a cell concentration of 20 million cells per milliliter. First, the Nile Red stain was optimized with fluorescence microscopy to achieve highest Nile Red intensity for lipid-containing N. oceanica cells using fluorescence spectroscopy (Additional file 1: Figure S1). Accordingly, cells were diluted in a mixture of 0.9% sodium chloride with 15% DMSO. Nile Red was added to this solution in a volume ratio of 1:100 from a 219 µmol/l Nile Red stock in MeOH solution according to . The samples were quickly vortexed and then stored in the dark for at least 10 min.
FACS analysis and cell sorting of N. oceanica cells
To prove occurrence of population heterogeneity in N. oceanica, single-cell analysis with FACS was chosen. Sorting of N. oceanica cells was performed using a MoFlo™ XDP High-speed cell sorter. Before cell sorting, beads (Flow-Check™ Fluorospheres 6605359, Beckman Coulter, Inc. USA) were used for fine alignment of the instrument. FL2-Log-Height fluorescence channel was applied for Nile Red fluorescent intensity detection. FSC-Height, FSC-Width and SSC-Height, presenting the morphological features of the input particles here, were used to measure the N. oceanica cells. For acquisition of results from cell-sorting experiment, the sorting protocol of the MoFlo™ XDP High-speed cell sorter was set to consist of three histograms: in the first histogram holding a gate R1, signals lower than 128 in FSC-Height and lower than 224 in SSC-Height were filtered out; in the second histogram, a spindly rectangle gate R2 was applied in the FSC-Height to FSC-Width histogram to remove cells which were not single; then in the last histogram, the target single cells were sorted, so that single cells for one subpopulation would be sorted to gate R3 (referred to as subpopulation P3), and another subpopulation would be sorted to gate R4 (referred to as subpopulation P4). For –N N. oceanica cells, the two sorted subpopulations are referred to as −N P3 and −N P4; For +N N. oceanica cells, the two sorted subpopulations are referred to as +N P3 and +N P4.
After Nile Red addition, cells in the staining solution were diluted to a cell concentration of 20 million cells per ml using 0.9% sodium chloride which was filtered with a 0.22-µm filter. Subpopulations were identified by the SSC channel filtering at 488 nm with a 10-nm bandpass and the Nile Red fluorescence channel filtering at 580 nm with a bandpass filter, filtering out the signal with wavelength below 500 nm and above 580 nm. Droplet formation was optimized and drop delay was determined using the Drop Delay Wizard. The sorting process was controlled by the Summit software 126.96.36.19980 (Beckman Coulter, Inc.). 10,000,000 cells of each subpopulation were sorted into 30 ml of Puraflow 8× Sheath Fluid, and cells were stored at − 80 °C and then lyophilized using an ultradry lyophilizator to obtain dried samples that were shipped from China to Germany for proteome analysis.
Digestion of proteins on filter well plates
The lyophilized cells were concentrated on 96 filter well plates by resuspension in 15-ml MilliQ water per falcon tube and the solution was homogenized by short vortexing. 300-µl portions of this solution were applied to each well of the 96-well filter plate (Millipore MSGVS2210) and the liquid was aspirated using a 96-well plate table top aspirator (KNF Neuberger Laboport). To increase the speed of the aspiration process, each subpopulation was aspirated in three separate wells; after aspiration, membranes with cells were washed with MilliQ grade water for three times.
The digestion protocol used for N. oceanica was based on . Filter membranes containing cells from the same sample were first collected in one tube and then divided into smaller pieces by a scalpel and dissolved in 32-μl dissolution buffer (25-mM ammonium bicarbonate, pH 7.8 containing 2-µl acetonitrile). Subsequently, cells were proteolytically digested with 8-μl porcine trypsin (Promega, Mannheim, Germany) resulting in a working concentration of 0.25 μg/μl at 37 °C with continuous shaking at 400 rpm for 2 h. Afterwards, cell debris and filter membranes were removed by centrifugation at 13,000g for 10 min at RT. Supernatants were collected in a glass vial and vacuum concentrated until the glass vial was dry. Dried peptides were resuspended in buffer A (2% acetonitrile 0.1% formic acid) by sonification for 5 min in a water bath. After centrifugation (2 min at 380g RT), samples were measured with mass spectrometry (MS).
Protein identification via 1D-nLC-ESI–MS
Protein identification was conducted in the same way as described in . Measurements were performed using a nanoAcquity UltraPerformance LC system connected to an auto-sampler equipped with a HSS T3 analytical column (1.8-µm particle, 75 µm × 150 mm) kept at 45 °C, and a Symmetry C18 trap column (5-µm particle, 180 µm × 20 mm), all Waters, USA. This setup was connected to an LTQ Orbitrap Elite. A 180-min gradient was used: (0–5 min: 99% buffer A and 1% buffer B, 5–10 min 99–94% A, 10–161 min: 94–60% A, 161–161.5 min: 60–14% A, 161.5–166.5 min: 14–4% A, 166.5–167.1 min: 99% A, 167.1–180 min: 99% A). To optimize the method for small cell numbers, the MS/MS max ion inject time was increased to 400 ms.
MS data processing
The identified proteins from the N. oceanica samples were identified by the Andromeda algorithm embedded in the MaxQuant software version 188.8.131.52 . All datasets were searched against an N. oceanica IMET1 database containing 9915 sequences . Only tryptic peptides with minimum of seven amino acids length were considered for the identification of proteins. No fixed modifications were used in the searches, while oxidation of methionine and N-terminal acetylation were set as variable modifications. The mass tolerance for CID fragment ion matches was set to 0.5 Da. Proteins were quantified by label-free quantification (LFQ) integrated into the MaxQuant software, and used for quantification only, when they had two or more unique peptides . Presented area values for N. oceanica are based on these quantification results. Significantly regulated proteins were established by a two-sample t test based on median normalized log2 abundance values of three replicates, and FDR (q value of ≤ 0.05) calculation by randomizing samples. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE  partner repository with the dataset identifier PXD008721.
Recultivation and reanalysis of subpopulations
For the purpose of recultivation and reanalysis, the subpopulations +N P3 and +N P4 were obtained by sorting exactly as described in previous sections. After 40 days, the recultivation reached a cell density sufficient for FACS sorting. Triplicate samples for each recultivated subpopulation (+N P3, +N P4) were used. Each sample for resorting was stained with the same staining method referred to in the previous sections. The protocol as referred to in section “FACS analysis and cell sorting of N. oceanica” was applied to detect the appearances of subpopulations in the sorted cells upon recultivation. After Nile Red staining, cells in the staining solution were diluted into the concentration of 20 million cells per milliliter using sodium chloride which was filtered with a 0.22-µm filter for sorting with a MoFlo™ XDP High-speed cell sorter in the purify sorting mode. 10,000 cells of each subpopulation were sorted into 30 ml of Puraflow 8× Sheath Fluid, for collecting the subpopulations’ features for further analysis.
Statistical assessment of differences between cultivation conditions
To analyze the sample differences among samples based on all MS data acquired in this study, the relative abundances of co-occurring proteins were analyzed by principal component analysis (PCA) implemented in R (https://www.r-project.org/) with ade4 package .