Ecogenomics and cultivation reveal distinctive viral-bacterial communities in the surface microlayer of a Baltic Sea slick

Abstract

Visible surface films, termed slicks, can extensively cover freshwater and marine ecosystems, with coastal regions being particularly susceptible to their presence. The sea-surface microlayer (SML), the upper 1-mm at the air-water interface in slicks (herein slick SML) harbors a distinctive bacterial community, but generally little is known about SML viruses. Using flow cytometry, metagenomics, and cultivation, we characterized viruses and bacteria in a brackish slick SML in comparison to non-slick SML as well as seawater below slick and non-slick areas (subsurface water = SSW). Size-fractionated filtration of all samples distinguished viral attachment to hosts and particles. The slick SML contained higher abundances of virus-like particles, prokaryotic cells, and dissolved organic carbon compared to non-slick SML and SSW. The community of 428 viral operational taxonomic units (vOTUs), 426 predicted as lytic, distinctly differed across all size fractions in the slick SML compared to non-slick SML and SSW. Specific metabolic profiles of bacterial metagenome-assembled genomes and isolates in the slick SML included a prevalence of genes encoding motility and carbohydrate-active enzymes (CAZymes). Several vOTUs were enriched in slick SML, and many virus variants were associated with particles. Nine vOTUs were only found in slick SML, six of them being targeted by slick SML-specific clustered-regularly interspaced short palindromic repeats (CRISPR) spacers likely originating from Gammaproteobacteria. Moreover, isolation of three previously unknown lytic phages for Alishewanella sp. and Pseudoalteromonas tunicata, abundant and actively replicating slick SML bacteria, suggests that viral activity in slicks contributes to biogeochemical cycling in coastal ecosystems.

Fig. 1. Slicks in the marine environment and sampling sites for this study.

Representative example of surface slicks (none of the ones sampled), observed in the Kalmar Sound with Öland in the background (a). Map illustrating slick sampling sites 1, 2, and 3 close to Ljungnäs (Rockneby, Sweden) in the Baltic Sea. Map was generated using Ocean Data View v.5.6.2 [128] (b).

Fig. 2. Accumulation of organic matter, surfactants, prokaryotic cells, and virus-like particles (VLPs) in water samples.

Dissolved organic carbon (DOC) measured in technical duplicates (a), mean +/- standard deviation for concentration of surfactants (n = number of technical replicates) (b), and counts of VLPs and prokaryotic cells in slick SML, non-slick SML, slick SSW and non-slick SSW (c). DOC and surfactants were measured from sampling site #1 only. SML sea-surface microlayer, SSW subsurface water (~70 cm depth).

Fig. 3. Diversity and indices of replication (iRep) for slick-associated bacteria.

Shannon-Wiener Index for PA and FL bacterial fractions from slick versus non-slick samples (a), relative abundance of bacterial classes among PA and FL fractions (b), and families of Gammaproteobacteria in greater detail (c). In situ replication rates (based on iRep) for bacterial metagenome-assembled genomes (d). FL free-living fraction (5–0.2 µm), PA particle-associated fraction (>5 µm), SML sea-surface microlayer, SSW subsurface water (~70 cm depth).

Fig. 4. Relative fraction of genes from specific functional pathways with differential abundance between MAG groups, identified using ALDEx2 and displayed as CLR-transformed relative gene abundances.

Several genes from functionally related categories (e.g. che chemotaxis, fli/flh flagellum genes) were combined, showing the average CLR value. lacI/galR LacI family transcriptional regulator, xthA exodeoxyribonuclease, motY sodium−type flagellar protein, pobR AraC family transcriptional regulator, sufS cysteine desulfurase/selenocysteine lyase, sufC Fe − S cluster assembly ATP-binding protein, metF methylenetetrahydrofolate reductase, gapA glyceraldehyde 3−phosphate dehydrogenase, pduO cob(I)alamin adenosyltransferase, crtB 15−cis−phytoene synthase, ccmBCF cytochrome/heme biogenesis/transport (Table S4c) (a). Diversity of CAZyme families in different MAG groups (see Table S4d for details) (b). CAZyme profiles of slick SML isolates, showing presence/absence of carbohydrate-binding module (CBM), glycoside hydrolase (GH), and polysaccharide lyase (PL) gene families (Table S2e). The four R. baltica isolates featured identical CAZyme diversity; therefore, only SMS3 is shown as representative. Due to lower completeness of SMS9, only SMS8 is shown for Alishewanella sp. (c).

Fig. 5. Viral diversity, relative abundance, and clustering with known phages.

Shannon-Wiener Index for vOTUs for the four different sample types (a). Relative abundance of viral clusters (VC) including the top 100 abundant vOTUs show an increase in relative abundance of certain VC in slick SML (VC_1425_0 & VC_988_0), while other VC decreased in relative abundance (VC1424_0 or VC_572_0) compared to reference samples. Further information about VCs and closest associated viruses is given in Table S6. Outliers and singletons refer to unclustered, presumably unknown viruses (b). Many vOTUs (nodes) from this study clustered with known phages of Flavobacterium, Pelagibacter, or Synechococcus based on shared protein clusters (interactions with known phages indicated by eleven purple frames) with a virus reference database from July 2022. Several vOTUs clustered only with other vOTUs from this study indicating unknown viruses (c). FL free-living fraction (5–0.2 µm), PA particle-associated fraction (>5 µm), SML sea-surface microlayer, SSW subsurface water (~70 cm depth), Vir viral fraction (<0.2 µm).

Fig. 6. Viral enrichment in slick versus non-slick SML, and auxiliary metabolic genes.

Enrichment of viruses in slick SML and non-slick SML across different filtered fractions. Shown are coverage ratios >1 indicating enrichment of a virus in the SML compared to the corresponding reference subsurface water sample. Black areas indicate depletion of a virus (ratio <1), while white areas show absence of the virus in the nominator or denominator of the ratio. Values and VCs of the heatmap’s y-axis are given in Table S8a (a). Coverage of vOTUs carrying an auxiliary metabolic gene (AMG), sorted by KEGG category (b). Only vOTUs being present in a sample based on read mapping were considered for this analysis. Full information on involved AMGs is given in Table S8b. FL free-living fraction (5–0.2 µm), PA particle-associated fraction (>5 µm), SML sea-surface microlayer, SSW subsurface water, Vir viral fraction (<0.2 µm).

Fig. 7. Phage-host interactions and viral micro-diversity.

Based on k-mer frequency patterns, vOTUs were predicted to match host MAGs and isolate genomes (middle). Further host evidence (left) was derived from vConTACT2 viral clustering (VC) with known phages from reference database and CRISPR spacer matches from MAGs. The heatmap (right) depicts the coverage of vOTUs in the three size fractions. D2* is a dissimilarity measure (the lower, the higher the similarity) (a) CRISPR-spacer to vOTU protospacer matches at 100% similarity reveal ten clusters with slick SML derived spacers, with C1-C5 including a slick SML specific vOTU from (a), framed in purple (b). Viral micro-diversity for different water sample types and filtered size fractions. Open circles represent strain variants of the viral OTUs (closed circles) and lines indicate the sample in which the respective variant has been detected. This figure corresponds to the results shown in Table S12b. FL free-living fraction (5–0.2 µm), PA particle-associated fraction (>5 µm), SML sea-surface microlayer, SSW subsurface water, Vir viral fraction (<0.2 µm) (c).

Fig. 8. Transmission electron microscopy images, abundance, CRISPR spacer matches and synteny of lytic phage isolates extracted from slick SML.

Negatively stained electron micrographs reveal myovirus morphology of Slickus and Slicko and podovirus morphology of Slicky, scale bar 100 nm (a). Coverage of reads based on mapping to the host MAGs Alishewanella sp. (MAG_01) and Pseudoalteromonas tunicata (MAG_66) as well as phages vB_AspM_Slickus01, vB_AspM_Slicko01, and vB_PtuP_Slicky01 (b). CRISPR spacer extracted from reads matching at 100% similarity to phage genomes from isolates and the SMS8 prophage of 50 kb. No matches from SSW spacers were detected. c Proteomic tree with placement of Slickus, Slicko, and Slicky (indicated by red stars) and closest related phages from the Refseq database (d). Based on insights from the tree, genomic alignments with related phages with indicated identity of proteins based on tBLASTX results (e). Annotations for the phage isolates are given in Table S14. SML sea-surface microlayer, SSW subsurface water.