ABSTRACT
How oligotrophic marine cyanobacteria position themselves in the water column is currently unknown. The current paradigm is that these organisms avoid sinking due to their reduced size and passive drift within currents. Here, we show that one in four picocyanobacteria encode a type IV pilus which allows these organisms to increase drag and remain suspended at optimal positions in the water column, as well as evade predation by grazers. The evolution of this sophisticated floatation mechanism in these purely planktonic streamlined microorganisms has profound implications for our current understanding of microbial distribution in the oceans, predator-prey interactions and, ultimately, will influence future models of carbon flux dynamics in the oceans.
A quarter of all primary production on Earth occurs in large nutrient deplete oceanic gyres1. Primary production in these large biomes is mainly driven by the dominant marine cyanobacteria i.e. Prochlorococcus and Synechococcus2. Gyres are permanently thermally stratified, where a lack of upward physical mixing poses a challenge for the microbial communities that inhabit them. How then do purely planktonic cyanobacterial cells in suspension combat the downward pull of gravity through the biological pump –i.e. drawing fixed carbon towards the ocean interior? Moreover, how do these highly-specialised planktonic microbes place themselves in their ‘preferred spot’, such as the well-established vertical distribution of high and low light-adapted Prochlorococcus ecotypes3–5? Marine picocyanobacteria lack flagellar structures for swimming or gas vacuoles for flotation6, and only a limited number of Synechococcus strains possess non-conventional mechanisms for motility7,8. Therefore, it has been assumed that these free-living microbes avoid sinking due to their lower density and reduced size9, and bloom when they encounter their optimal environmental conditions while drifting randomly within marine currents.
To date, type IV pili are known to provide functions such as twitching motility, surface attachment, biofilm formation, pathogenicity, as well as conjugation, exogenous DNA acquisition and competence10–12. These extracellular appendages can be rapidly extended and retracted by polymerising and depolymerising cycles of the major pilin subunit e.g. PilA, requiring a defined transmembrane apparatus and the consumption of energy in the form of ATP11,13. Functional analysis of most type IV pili has focused on pathogenic microbes and their use of surfaces or substrates for pilus anchoring. However, analysis of pili from freshwater cyanobacteria has revealed these appendages can be used for twitching motility during phototaxis in Synechocystis14 or exogenous DNA acquisition in both Synechocystis and Synechococcus elongatus15,16. This latter function requires an additional set of proteins for competence such as ComEA and ComEC. While acquisition of DNA by S. elongatus is performed by the third of three PilA-like proteins encoded by this strain (PilA3)16, no known function has been attributed to PilA1 and PilA2 other than being dispensable for attachment and biofilm formation17. A mutant in S. elongatus that no longer produced PilB –the protein responsible for pilus elongation– abolished the production of pili appendages, made up of PilA1, and was reported to suppress planktonic growth of this strain17. The presence of pilus genes in purely planktonic marine microbes has previously been reported, but their role remains enigmatic18.
Here, we show that almost a quarter of all marine picocyanobacteria encode a PilA1-like pilus. We show that this extracellular appendage produced in these purely planktonic organisms –which rarely encounter any kind of surface in their natural habitat– allows cells to increase drag and remain in an optimal position in the water column as well as avoid being preyed upon. This provides yet another biological function to these filamentous appendages and sheds light on the ecological role of type IV pili in marine ecosystems.
RESULTS AND DISCUSSION
Abundant production of a type IV pilus in Synechococcus sp. WH7803
We first detected an abundant PilA protein (i.e. SynWH7803_1795) in the extracellular proteomes of the model marine cyanobacterium Synechococcus sp. WH7803, accounting for up to 25% of the exoproteome19,20. Transmission electron microscopy confirmed the existence of the macromolecular pili structures (Fig 1A and Fig S1). Unlike Synechocystis sp. PCC6803 that simultaneously produces thick and thin pili14, this marine picocyanobacterium presented multiple pili of similar thickness, each ~10 μm in length. The amino acid sequence of PilA revealed a typical type II secretion signal and a conserved GFTLxE motif at the N-terminus of the protein (Fig 1B) that is known to be cleaved in the cytoplasm before the protein is translocated to the base of the pili for assembly10. After cleavage, the N-terminal of PilA can be post-translationally modified, e.g. methylated, to increase the hydrophobicity and stability of the pilin10,21, although we were unable to detect this modified N-terminus tryptic peptide during proteomic analyses. In close proximity to pilA in the Synechococcus sp. WH7803 genome we found five other type IV-like pilin genes (Fig 1C), all with the conserved GFTLxE motif (Fig 1D).
Using the pilus apparatus from the freshwater cyanobacterium S. elongatus PCC 7942 as a reference16 and the established architecture for type IV pilus machinery11, we were able to find all components necessary for pilus assembly in Synechococcus sp. WH7803 (Fig 1E). The genetic cluster encoding the six pilin-like proteins (Fig 1C and 1D) likely produces three types of pili that use the same transmembrane pilus structure. Based on homology with the annotated genes from S. elongatus16 and conserved domains found using the CD-search tool in NCBI, we suggest the three pili types: PilA1-PilE, PilA2-PilV and PilA3-PilW (Fig 1C). Of these types, shotgun proteomic analyses have only ever detected PilA1-PilE19,20, although PilA2 was also detected in low abundance in cellular –but not extracellular– proteomic datasets22. Unlike in S. elongatus, where PilA1 and the contiguously-encoded pilin-like protein are almost identical, the amino acid sequence of PilA1 and PilE in Synechococcus sp. WH7803 are clearly distinguishable. Although in much lower abundance, PilE seems to be correlated with PilA1 in the exoproteomes of this cyanobacterium19,20 and, therefore, it is possible that PilE and PilA1 form subunits of the same pilus apparatus.
Pilus distribution amongst picocyanobacterial isolates and Single-cell Assembled Genomes (SAGs)
Genomic analysis of sequenced marine picocyanobacterial isolates downloaded from the Cyanorak database revealed that 74% of sequenced Synechococcus (n=46) and 33% of Prochlorococcus (n=43) encoded pilA1 (Fig 2 and Table S1). In Synechococcus, the pilus was prevalent in all clades (93%; n=28) except for clades II and III where it was less abundant (44%; n=18). Interestingly, all low light Prochlorococcus isolates from clades III and IV encoded pilA1 (n=7; Fig 2). Most of these pilA1-containing strains also encoded a pilE homologue in close proximity. Genes pilA2 and pilA3 were also abundantly found in Synechococcus (59 and 74%, respectively), although were much less prevalent in Prochlorococcus (12 and 9%, respectively). As expected, all strains that encode at least one of the pilA types also possessed the transmembrane pilus apparatus, whereas this apparatus was completely absent or partially lost in strains lacking pilA (Fig 2). PilA3 is known to be involved in DNA uptake and competence in S. elongatus, requiring additional competence proteins to do so16. Marine picyanobacteria are not known for being naturally competent but, interestingly, all strains encoding PilA3 also contained the competence genes encoding ComEA and ComEC (Fig 2). Further work is needed to investigate the conditions under which the PilA3-type pilus becomes active in these organisms and, therefore, when exogenous DNA might be taken up.
The GFTLxE motif is conserved in 87.5% of PilA1 sequences encoded by cultured marine picocyanobacteria (i.e. 42 of the 48 sequences; Table S1). The remaining six PilA1 sequences possess a GFSLxE motif, five of which were in Prochlorococcus strains. Across the full length of the mature PilA1 protein, which on average is 140 amino acids long, only the first ~50 N-terminal amino acids starting from the conserved GFTLxE motif are well conserved amongst all sequences, a commonly observed feature in PilA-like proteins11. The C-terminus of the protein showed a remarkably high variability even between closely related strains. Despite this high variability, their predicted structures were still similar to those of known pili subunits (Fig S2)23,24. During pili assembly, the helix encoded by the conserved N-terminus of PilA remains in the pilus core and only the variable C-terminus –that producing anti-parallel β-sheets– is exposed to the milieu11. We hypothesise that the hyper-variability of the exposed C-terminus is a strategy to escape phage attachment, it being a known pathway used by phage for host encounter and infection25,26. Similarly, flagella have a hyper-variable region, which has also been attributed to phage and immune system evasion27.
The screening of 190 picocyanobacterial Single-cell Assembled Genomes (SAGs) obtained from surface seawater across the globe28 –those with over 75% completeness– revealed the presence of pilA1 and genes encoding components of the pilus apparatus in almost one in four marine picocyanobacteria (24% encoded pilA1 and 21-25% the pilus apparatus; Table 1 and Table S1). As expected by its abundance in the oceans, Prochlorococcus comprised almost 96% of all 190 SAGs (Table 1). The prevalence of the pilus was much higher amongst the low light Prochlorococcus SAGs from clades II/III (67-80%) than in those belonging to other ecotypes, and Synechococcus showed the abundance observed in cultures isolates (~76%; Table 1).
Global distribution and expression of picocyanobacterial pilA1 in marine pelagic ecosystems
The distribution and expression of pilA1 in the surface ocean was determined by analysing its presence in the global marine TARA metagenome and metatranscriptome datasets. An HMM profile generated from the PilA1 sequences (cultured isolates and SAGs; Table S1) was used to search the TARA datasets in the Ocean Gene Atlas portal29 retrieving 903 and 837 individual hits from the metagenomes and metatranscriptomes, respectively (using a cut-off E-value < 10−10). Sequences assigned to Prochlorococcus represented 85% in both datasets, whereas those assigned to Synechococcus represented 12% and 11% of the metagenomes and metatranscriptomes, respectively. BLAST analysis of these hits against PilA1, PilA2 and PilA3 sequences was used to confirm the specificity of our HMM profile, proving effective in discriminating against PilA2 and PilA3 sequences (each representing less than 1% of the hits).
The abundance and transcription of genes encoding PilA1 from Prochlorococcus and Synechococcus across all oceanic regions, marine biomes and water depths (Fig 3) revealed a similar abundance of pilA1 from Prochlorococcus in both metagenomic and metatranscriptomic datasets. In contrast, pilA1 from Synechococcus was enriched in the metatranscriptomes, mainly driven by the increased expression in the North Atlantic Ocean and Mediterranean Sea as well as in coastal and westerlies, biomes where Synechococcus are known to thrive. As expected, a large reduction in the presence and expression of the picocyanobacterial pilA1 was noted in the polar oceans, where both genus are not abundant. Furthermore, the presence and transcription of cyanobacterial pilA1 decreased drastically in the aphotic mesopelagic layer, in accordance with picocyanobacteria naturally populating only euphotic layers of the ocean.
PilA1-type pili increase drag and allow cells to remain planktonic
The extensive distribution and expression of such an extracellular appendage begs the question why such a complex structure is so prevalent in such streamlined planktonic cyanobacteria. To assess this, and to assign a biological function to this extracellular structure in the context of the ecology of marine planktonic bacteria more generally, we abolished the production of PilA1 and PilE in Synechococcus sp. WH7803. As expected, the fully segregated pili mutant strain no longer produced the extracellular structure (Fig 1A and Fig S1). Most remarkably, we observed a clear loss in the strain’s ability to remain suspended in its typical planktonic form (Fig 4A). By tracking wild type and pili mutant cells of Synechococcus sp. WH7803, we determined that the lack of pili produced an average cell sinking rate of 8.4 ± 0.4 mm/day, while the wild type had an average uplifting drift of 0.8 ± 0.3 mm/day (Fig 4B, Suppl. Movies 1-4 and full data in Table S2). While avoiding sedimentation, the pili did not appear to confer motility, with both mutant and wild type strains producing ‘pin-prick’ colonies in sloppy agar plates as opposed to fuzzy colonies characteristic of motile strains8. Furthermore, neither the mutant nor the wild type strain aggregated when grown in shaken liquid cultures.
We performed a comparative proteomic analysis between wild type Synechococcus sp. WH7803 and the pili mutant to assess any additional effects of disrupting the PilA1-PilE pilus (Table S3). Apart from the complete absence of the PilA1 and PilE proteins, only three other proteins were significantly down-regulated in the cellular proteome of the pili mutant strain (Fig S3): SynWH7803_0049 (12.9 fold) and SynWH7803_1797 (3.2 fold), both of unknown function and, most interestingly the pilus retraction protein PilT (SynWH7803_0020; 2.9 fold reduction). SynWH7803_1797 is located just downstream of PilA1-PilE and is predicted to encode a secreted protein that is usually found in low abundance in the exoproteome of Synechococcus sp. WH780320. Indeed, it was also found down-regulated in the exoproteome of the pili mutant (2.5x; Fig S3). The exoproteomes also revealed a generalised shift of proteins more abundantly detected in the pili mutant. These were mainly low abundance cytoplasmic proteins that were barely detected in the wild type strain (Fig S3 and Table S3). Most likely, the absence of the abundant PilA1 protein from the exoproteome of the mutant strain caused an artifactual increase in the detection of lower abundance proteins by mass spectrometry and, despite efforts to normalise the data, there was an apparent upregulation of most proteins.
We had previously observed that Prochlorococcus sp. MIT9313 (a strain that encodes PilA1; Fig 2) routinely switches between planktonic and sedimenting lifestyles when grown in different media (Pro99 and PCR-S11, respectively; Fig 4C). Whilst Prochlorococcus remains genetically intractable30, we compared the exoproteomes of this strain in both medias. Commensurate with our findings in Synechococcus sp. WH7803, PilA1 was not detected in the exoproteomes of sedimenting Prochlorococcus sp. MIT9313 cultures, whereas it was present in all of the planktonic ones (i.e. PMT_0263; Table S4), suggesting Prochlorococcus can coordinate production of the pili in response to the distinct nutrient environments present in both medias.
PilA1 allows cells to retain an optimal position within the water column
We further explored the effect of nutrient stress on pili production in Synechococcus sp. WH7803, with nitrogen and metal depletion showing the strongest decline (i.e. 11.5 and 3.4-fold decrease in PilA1 production, respectively), whilst phosphorus depletion had no effect (Table S5). Although in low abundance, PilA1 of Synechococcus sp. WH7803 was detected even under natural oligotrophic conditions (i.e. when incubated in natural seawater; representing 0.3% of the exoproteome) and showed a slight increase after adding environmentally-relevant concentrations of nutrients (0.5% of the exoproteome in the presence of 8.8 μM N and 0.18 μM P). Pili detection became most obvious under higher nutrient conditions (60-fold increase in pili at 88 μM N and 1.8 μM P when compared to nutrient deplete seawater; Table S6). Considering the biological significance of this phenotype in the context of marine planktonic organisms –where the production of pili reduces sedimentation by increasing the viscous drag of the cell– the extension/retraction of pili would allow a cell to position itself at an optimal position in the water column, e.g. in patches of high nutrient availability. To further investigate this, we set up a nutrient step gradient. While the pili mutant, as expected, sank through the gradient independently of nutrient availability, the wild type strain was able to position itself in the nutrient-replete layer where it remained via the production of pili (Fig 4D and Fig S4).
Synechococcus sp. WH7803 also retained pili over a 24 h period of darkness. Nevertheless, after three days under dark conditions –during which cultures are known to remain viable31, the detection of pili dropped drastically (i.e. >20-fold drop in pili abundance in the exoproteome; Table S7). Marine cyanobacteria occupying photic layers of the ocean will therefore keep their pili structures over normal diurnal light-dark cycles to maintain their position, but cells will cease to retain their position once they sink out of the euphotic zone, retracting their pili possibly as a strategy to recover energy while awaiting an uplift back to photic layers by upwelling currents.
Pili prevent grazing
Given the cell surface nature of this pilus, we also assessed whether it could mediate ecological interactions with other organisms. We found that as well as preventing sedimentation, pili allow bacteria to evade predation by protist grazers. Thus, Synechococcus sp. WH7803 and the pili mutant grown in the presence of two bacterivorous protozoa i.e. the preying ciliate Uronema and suspension feeding flagellate Cafeteria, strikingly showed that whilst the wild type strain was able to completely evade grazing by Cafeteria and largely delay culture depletion by Uronema, the pili mutant was efficiently grazed by both protists (Fig 5). Presumably, the long pili appendages interfere with the way bacterivorous protozoa access their prey and, hence, this reduces their susceptibility to grazing.
Concluding remarks
Together, our data suggests a novel and sophisticated mechanism that enables oligotrophic marine picocyanobacteria to stay buoyant and reduce cell death due to grazing through the use of pili, adding a new biological function to these extracellular appendages. This PilA1-type pilus, widely distributed and expressed amongst dominant marine oligotrophic cyanobacteria, allow these purely planktonic organisms to increase cell drag and, consequently, maintain an optimal position in the water column. This mechanism is under tight regulation in response to discrete stimuli e.g. nutrients, which may vary depending on the ecological adaptation and preferred niche of each individual organism. Therefore, as opposed to flagellated bacteria that show positive chemotaxis towards nutrient hotspots in the oceans32, non-motile picocyanobacteria may apply a more passive strategy which consists of elongating their pili when they encounter preferable conditions to remain in an optimal position while drifting in a water body. These long appendages also interfere with the access of bacterivorous protozoa to their prey allowing pili-producing cells to evade grazing. Further research should define: i) the biophysical differences between pili that allow attachment and floatation, ii) the resources required for this floatation system and advantages over other mechanisms such as flagella, and iii) the ecological trade-offs of having such extracellular appendages. Thus, besides being beneficial, pili could also be a handicap to those cells that produce them as, being phage binding sites25,26, they will increase the chance of interacting with phage and hence their susceptibility to phage infection.
The biological carbon pump and microbial loop pose important challenges for streamlined marine cyanobacteria, which have moved away from canonical flagellar motility and have evolved this more passive mechanism for flotation that may require less resources and no additional convoluted tactic systems. This discovery changes our ecological perception of this dominant marine bacterial group, and will have important consequences for our future understanding of predator-prey and carbon flux dynamics in the oceans.
Methods
Culture conditions
Synechococcus sp. WH7803 was grown in ASW medium and oligotrophic seawater using conditions previously described22. Experiments were performed using 20 ml cultures contained in 25 cm² rectangular cell culture flasks (Falcon) with vented caps. Cultures were incubated under optimal growth conditions i.e. at 22°C at a light intensity of 10 μmol photons m−2 s−1 with shaking (140 r.p.m.), unless otherwise stated in the text. To study the influence of nutrients on pili production, media was prepared by i) not adding different nutrient sources into ASW media, i.e. nitrogen, phosphorus and trace metals, or ii) diluting ASW in oligotrophic seawater (i.e. 1:1000, 1:100 and 1:10). The low light-adapted ecotype Prochlorococcus sp. MIT9313 was grown in 40 ml Pro99 medium and PCR-S11 medium with no additional vitamins33. Different light intensities (i.e. 4 and 15 μmol photons m−2 s−1) and temperatures (i.e. 14 and 22°C) were tested. Cyanobacterial culture growth was routinely monitored by flow cytometry (BD Fortessa).
Grazing experiments were performed using ASW-washed Uronema marinum (isolated from Qingdao Bay) and Cafeteria roenbergensis CCAP 1900/1 cells. Briefly, 10 ml culture was subjected to centrifugation at 4000g for 15 min and the pellet resuspended in 10 ml ASW. One ml of washed grazers was used to inoculate 20 ml cultures of wild type and pili mutant of Synechococcus sp. WH7803 at an initial concentration of 3-4 x 106 cells ml−1. Triplicates cultures were incubated under optimal conditions, including shaking to avoid pili mutant sedimentation (see above).
Pilus knockout mutant in Synechococcus sp. WH7803
Genes SynWH7803_1795 and SynWH7803_1796 (pilA1 and pilE, respectively) were replaced by a gentamicin cassette via a double recombination event to generate the Synechococcus sp. WH7803 pili mutant. Two flanking regions of 700 bp from the genome of Synechococcus sp. WH7803 and the gentamicin cassette from pBBR-MCS34 were amplified by PCR using primers indicated in Table M1, and inserted into vector pK18mobsacB35 using the Gibson assembly method following manufacturer’s instructions (New England Biolabs). The detailed protocol to generate mutants in Synechococcus sp. WH7803 is given in the Supplementary Information. All three transconjugant colonies that were picked had the same sinking phenotype, had doubly recombined (as checked by sequencing the overlapping regions) and were fully segregated. One mutant was subsequently selected to make axenic by eliminating the ‘helper’ strain and used for further experimentation. The pilus mutation is stable and has retained its sinking phenotype over time.
Transmission Electron Microscopy
Optimally-grown Synechococcus sp. WH7803 and pilus mutant cultures were fixed using 3% (v/v) final concentration glutaraldehyde after which 5 μl were delicately transferred onto a glow-discharge formvar/carbon coated grid and left 2-3 min for cells to attach. After blotting the excess media, negative staining was achieved by applying a drop of 2% uranyl acetate to the grid for 1 min. The excess stain was blotted off and left to air dry before imaging using a JEOL 2011 TEM with Gatan Ultrascan.
Tracking and imaging of sinking cells
The movement of wild type Synechococcus sp. WH7803 and pilus mutant through a nutrient step gradient was performed by placing washed cells in oligotrophic SW on top of a column where nutrient layers (nutrient deplete and amended layers as indicated, using ASW media) were achieved by increasing sucrose concentration (i.e. 2.5% w/v per layer).
Sedimentation tracking and velocity measurements were conducted using a setup as previously described36. Briefly, a sample chamber was prepared by gluing a square glass capillary (inner dimensions 1.00 x 1.00 mm, length 50 mm; CM Scientific, UK) onto a glass slide using an optical glue (#81; Norland, USA). Tubings (Masterflex Transfer Tubing, Tygon® ND-100-80 Microbore, 0.020" ID x 0.060" OD; inner dimension 0.51 mm; Cole-Parmer, USA) were attached to both ends of the sample chamber using blunt dispensing nozzle tips (Polypro Hubs; Adhesive Dispensing, UK), one-way stopcock valves (WZ-30600-00; Cole-Parmer, USA), and Luer connectors. Synechococcus was pulled into the capillary through the inlet by manual suction using a 2 ml syringe connected to the outlet of the capillary system. After introducing the sample, the microfluidic system was isolated by closing the stopcock valves. Samples were left to settle for 30-60 minutes before recording. The slide-capillary system was held vertically by an adjustable translation stage (PT1B/M; Thorlabs, USA), and placed between a white LED ring light source and a continuously focusable objective (InfiniProbe TS-160; Infinity Photo-Optical Company, USA) for dark-field imaging. Images were acquired with magnification set at 16×, using a CMOS FLIR Grasshopper3 (GS3-U3-23S6M-C; Point Grey Research Inc., Canada) operated with FlyCapture2 (FLIR Systems UK). Recordings were done at 1 fps for either 1, 2, or 5 minutes. The system was calibrated with a resolution target (R2L2S1P Positive NBS 1963A; Thorlabs, USA). Sedimentation velocities were calculated with custom-made codes in MATLAB 2019a, based on particle tracking code from Crocker and Grier37. Briefly, features corresponding to individual cells were selected based on shape and image intensity, and the distribution of frame-to-frame displacements was calculated from all the tracks lasting at least 20s. All the experiments were checked for outliers.
Proteome preparation and shotgun analysis
20 ml cell cultures were subjected to centrifugation at 4000g for 15 min at 4°C. Cell pellets were used for cell proteome analyses whereas the supernatants, which were further filtered to remove any remaining cells (0.22 μm), were used for exoproteomic analyses. Exoproteomes were concentrated using a trichloroacetic acid precipitation protocol as previously described38. Cell pellets and exoproteome precipitates were resuspended using 1x LDS buffer (ThermoFisher) containing 1% beta-mercaptoethanol and prepared for LC-MS/MS via an in-gel trypsin digestion as described previously38. Tryptic peptides were analysed with an Orbitrap Fusion mass spectrometer (Thermo Scientific) coupled to an Ultimate 3000 RSLCnano system (Dionex), using conditions described in Christie-Oleza et al39. Mass spectra were identified using Synechococcus sp. WH7803 and Prochlorococcus sp. MIT9313 UniProt databases (downloaded on 09/11/2017) and quantified using default parameters in MaxQuant v1.6.10.4340. Comparative proteomic analyses were performed using MS intensity values in Perseus v1.6.2.241.
In silico analysis of pilus proteins
Major pilin proteins and pilus machinery were searched in Synechococcus sp. WH7803 using reference proteins from S. elongatus PCC 794216 and proposed modelled structure11. Genomes from cultured picocyanobacteria (Synechococcus, n=46, and Prochlorococcus, n=43) were downloaded from Cyanorak and re-annotated using PROKKA vs. 1.742. Genomes were then screened for the presence of PilA-like proteins, structural proteins PilQ, PilB and PilC, and competence proteins ComEA and ComEC, via a local BLAST server using the amino acid sequences from Synechococcus sp. WH7803.
The 190 picocyanobacterial SAGs from Berube et al28, all with over 75% estimated genome completeness, were downloaded and annotated in-house using PROKKA vs. 1.7. Annotated SAGs were screened for pilus associated proteins using those from Synechococcus sp. WH7803 (Table S1), with further verification using the Conserved Domain search tool from NCBI and manual curation in order to eliminate redundant matches within each SAG and select PilA1-like proteins which had a PilE-like protein encoded in close proximity in the genome.
The PilA1 sequences obtained from the cultured isolates and SAGs were used to generate an HMM profile in Unipro UGENE vs. 3343 implemented with the hmmbuild programme from HMM344 using the default parameters. The resulting HMM profile was used to search the TARA oceans metagenomes and metatranscriptomes via the functions offered in the Ocean Gene Atlas portal29.
The modelled 3D protein structure of PilA-like proteins was performed with mature sequences using the I-TASSER server’s default settings45.
Supporting information
Author contribution
JC-O conceived the study. MA-F generated the knockout mutant. JC-O and MA-F performed the sinking experiments. VZ, AC and DS performed the proteomic analyses. JC-O, RB and RJP performed the bioinformatics analyses. ML and MP carried out the tracking and imaging of sinking cells. JC-O and MA-F wrote the manuscript with large input from DJS and RJP.
Competing interests
The authors declare no competing interests.
Data and materials availability
All detailed methods and data is available as supplementary information and Tables. The mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifiers: PXD018394, PXD018395, PXD018396, PXD018524 and PXD019315.
Supplementary Information
Protocol for generating mutants in marine Synechococcus
Extended data Table M1 and Figures S1 to S4
Extended data Tables S1 to S7
Extended data movies 1 to 4
Acknowledgments
We thank Dr Bakker and the Advanced Bioimaging Platform as well as Dr Hernandez-Fernaud and the WPH Proteomic Facility, at the University of Warwick, for respective support in imaging and proteomics. We thank Dr Guillonneau for providing the protist cultures and WISB centre (grant ref. BB/M017982/1) for access to equipment. This work was supported by NERC grant NE/K009044/1, Ramón y Cajal contract RYC-2017-22452 and the MINECO project CTM2015-70180-R.