Alteromonas sp. 76-1 was isolated in April 2012 from a microcosm with surface seawater collected at the Patagonian continental shelf (47° 56′41″S 61° 55′23″W) amended with 0.001% (w/v) sodium alginate (Wietz et al., 2015). Purity was confirmed by 16S rRNA gene PCR after several rounds of subculturing. Polysaccharide utilization was analyzed in seawater minimal medium (SWM) (Zech et al., 2009) supplemented with 0.1% sodium alginate (cat. no. A2158; Sigma-Aldrich, St. Louis, MO, United States) or 0.2% ulvan (cat. no. YU11689; Carbosynth, United Kingdom) as sole carbon sources in comparison to SWM + 0.1% glucose. All cultures were incubated in triplicates at 20°C and 100 rpm with regular photometric determination of growth (diluted if OD600 >0.4) followed by calculation of maximal growth rate (μmax) and doubling time (ln2/μmax). In addition, hydrolysis of 18 AZO-CL labeled polymers was tested according to Panschin et al. (2016). Briefly, AZO-CL substrates (Megazyme, Ireland) were distributed to individual microtiter wells in triplicate. To each well, 100 μL HaHa medium (Hahnke and Harder, 2013) were added, plus each 100 μL of a starved culture or 100 μL medium as control. Cultures were incubated at 25°C for up to 14 days and evaluated for hydrolytic activity on the basis of color change.

Genomic DNA was extracted with the Genomic-tip 100/G kit (Qiagen, Hilden, Germany). After shearing using g-tubes (Covaris, Woburn, MA, United States) and monitoring the size range by pulse field gel electrophoresis, DNA fragments were end-repaired and ligated to hairpin adapters using P6 chemistry (Pacific Biosciences, Menlo Park, CA, United States). SMRT sequencing was carried out on a PacBio RSII instrument (Pacific Biosciences). PacBio reads were assembled de novo using the RS_HGAP_Assembly.3 protocol in the SMRT Portal v2.3. Indel errors were corrected using 5,759,952 paired-end reads of 112 bp from prior Illumina GAIIx sequencing on Nextera XT libraries (Illumina, San Diego, CA, United States) performed at Göttingen Genomics Laboratory (Germany). Illumina reads were mapped using the Burrows-Wheeler Aligner (Li and Durbin, 2009) followed by variant detection using VarScan v2.3.6 (Koboldt et al., 2012), consensus calling using GATK 3.1-1 (Mckenna et al., 2010), and trimming of the final assembly. Chromosome and plasmid were circularized (in total 4,817,656 bp) and uploaded to both ENA¹ and IMG² under accessions numbers PRJEB28726 and 2784132050, respectively. Phylogenetic analysis was carried out with 92 core genes identified using the UBCG pipeline (Na et al., 2018) including Pseudoalteromonas atlantica T6c as outgroup (RefSeq NC_008228.1). The concatenated nucleotide alignment was manually curated and the best substitution model (GTR+G) computed using jModelTest2 (Darriba et al., 2012). A maximum-likelihood phylogeny with 1000 bootstrap replicates was calculated using RaxML (Stamatakis, 2014) implemented on CIPRES (Miller et al., 2010).

Genomes of strain 76-1 and publicly available Alteromonas spp. (Supplementary Table S1) were compared using a variety of bioinformatic tools. Average amino acid identities and genome-to-genome distances were calculated using the Enveomics (Rodriguez-R and Konstantinidis, 2016) and GGDC (Meier-Kolthoff et al., 2013) web applications, respectively. Unique genes were identified using BPGA (Chaudhari et al., 2016) with a threshold of 50% amino acid identity, only considering proteins >50 amino acids (Supplementary Table S2). CAZymes were identified using dbCAN2 (Zhang et al., 2018), only considering hits with e < 10⁻²³ and >80% query coverage. Annotations of selected genes were checked using UniprotKB-Swissprot, KAAS and MEROPS databases (Moriya et al., 2007; The UniProt Consortium, 2017; Rawlings et al., 2018). Genomic regions were visualized using genoPlotR (Guy et al., 2010) and Circos (Krzywinski et al., 2009) followed by manual inspection. Homologous PUL in related strains were identified using MultiGeneBlast (Medema et al., 2013), only considering proteins with >50% amino acid identity and >50% sequence coverage. Completeness of draft genomes used for comparative analyses was estimated using CheckM (Parks et al., 2015).

Polysaccharide lyases of strain 76-1 were compared to enzymes deposited in the curated databases CAZY (Lombard et al., 2014) and PDB (Berman et al., 2000). In addition, related sequences deposited at NCBI were identified by BLASTp, only considering hits with >90% query coverage and >60% amino acid identity. PL sequences were aligned using ProbCons (Do et al., 2005) followed by manual inspection. Protein substitution models (WAG+G+F best for all alignments) were calculated using Modeltest-NG³, an improved successor of ProtTest3 (Darriba et al., 2011). Phylogenies were computed using RaxML with 1000 bootstrap replicates (Stamatakis, 2014). To determine the occurrence of PL transcripts in marine habitats, one PL from each alginolytic system (alt76_01684, alt76_03417) was searched against 272 marine metatranscriptomes publicly available at IMG (Supplementary Table S3). Only a co-detection of PL homologs with >70% amino acid identity was considered positive.

Two grams of XAD-16 resin (Dow Chemical Company, Midland, MI, United States) were washed several times with dH₂O and methanol, concentrated on a metal filter, resuspended in 100 mL dH₂O, and autoclaved. Strain 76-1 was precultured in sea salt medium (4% Sigma sea salts, 0.3% casamino acids, and 0.4% glucose) for 16 h at 20°C. Preculture was inoculated at 2% (v/v) into 100 mL sea salt medium mixed with 100 mL XAD-16 solution and cultivated at 20°C and 100 rpm for 24 h. Resin was collected by filtration and extracted with methanol:dH₂O (80:20). Extracts were concentrated in a rotary evaporator using a two-step vacuum (337 mbar, 100 mbar) and heating to 40°C. Concentrated extract was tested for antimicrobial activity in a well-diffusion assay (Hjelm et al., 2004) with Alteromonas macleodii D7 as target strain and methanol:dH₂O (80:20) as negative control. Siderophore production was tested with sterile-filtered supernatant from overnight cultures in both iron-deplete and iron-replete minimal medium using a modified CAS assay (Schwyn and Neilands, 1987; Alexander and Zuberer, 1991). Positive and negative controls were deferoxamine mesylate and sterile medium, respectively.

Strain 76-1 was isolated from alginate-supplemented seawater collected at the Patagonian continental shelf (Wietz et al., 2015) on solid media containing alginate as sole carbon source. Colonies appear round and cream-colored, assuming a tough surface after several days. Cells are motile as confirmed by light microscopy. Whole-genome sequencing revealed a chromosome of 4.7 Mb and a 126 Kb plasmid, encoding a total of 4100 proteins (Table 1). Core genome phylogeny demonstrated close relationship with Alteromonas naphthalenivorans, with 95.5% average amino acid identity (AAI) to A. naphthalenivorans SN2^T but only 63.5% DNA-DNA relatedness. Hence, strain 76-1 can only be assigned to genus level and might represent a novel Alteromonas species. Strains 76-1 and SN2 form a phylogenetic clade with A. stellipolaris and A. addita (89% AAI), separated from the more distant relatives A. australica, A. macleodii, and A. mediterranea (75–77% AAI) (Figure 1).

The unclear taxonomic affiliation motivated an initial comparison of strain 76-1 with the closest relative, A. naphthalenivorans SN2^T. In addition to a shared core of 3401 genes, strains 76-1 and SN2 encode 621 and 793 unique genes relating to their contrasting habitats (Table 1 and Supplementary Table S2). Notably, the origin of strain 76-1 from pelagic waters with high primary production and carbohydrate availability (Acha et al., 2004; Garcia et al., 2008) is reflected by a third of unique genes belonging to KEGG class “Carbohydrate Metabolism” (Figure 2A) and a diverse CAZyme repertoire (Figure 2B). Strain 76-1 encodes a total of 96 CAZymes, including thirteen polysaccharide lyases from six families and 41 glycoside hydrolases from 20 families (Table 2 and Figure 2B). This CAZyme content approximates that of algae-associated Flavobacteriia (Mann et al., 2013) whereas most other Alteromonas encode lower numbers (Supplementary Table S1).

Based on these observations, this study focuses on the genomics and ecophysiology of polysaccharide degradation, as detailed below, but 76-1 features further unique adaptations. Specifically, unique gene clusters for siderophore synthesis and methylamine metabolism (Supplementary Figure S1) could sustain iron and nitrogen supply under limiting conditions in pelagic habitats (Boiteau et al., 2016; Taubert et al., 2017). The siderophore cluster is co-localized with polyketide-nonribosomal peptide (PKS-NRPS)-encoding genes that may enhance competitive abilities (Patin et al., 2017). We experimentally confirmed functionality of these clusters by showing iron scavenging and antibacterial activity using CAS and well-diffusion assays (Supplementary Figure S1). Homology of the PKS-NRPS to tyrocidine/gramicidin-related gene clusters in Bacillus (Mootz and Marahiel, 1997) and differing G+C content compared to the chromosome (38 vs. 44%) suggest horizontal acquisition from low-GC Gram-positive bacteria. In contrast, proteins for naphthalene and urea degradation (Math et al., 2012; Jin et al., 2016) as well as conjugal transfer (Supplementary Table S2) are unique to A. naphthalenivorans SN2^T and probably facilitate adaptation to higher anthropogenic input (Cozzi et al., 2014) and genetic exchange (Fuchsman et al., 2017) in coastal sediments. The lower numbers and diversity of CAZymes (Figure 2) suggests a minor role of polysaccharide degradation in the habitat of SN2 (Math et al., 2012).

A comprehensive CAZyme inventory of Alteromonas sp. 76-1 revealed genetic machineries for degradation of alginate, ulvan and xylan, common polysaccharides from marine algae (Michel and Czjzek, 2013) and potentially contained in algal-derived microgels (Aluwihare and Repeta, 1999). Furthermore, we identified functional genes for the degradation of different glucans with simpler molecular structure.

Alginate degradation of strain 76-1 relates to two separate alginolytic systems (designated AS1 and AS2) encoding PL6 and PL7 alginate lyases, TonB-dependent receptors, MFS transporters and enzymes for monomer processing (Figure 3A). Additional alginate lyases from families PL7, PL17, and PL18 are located outside of AS1 and AS2 (Figure 3B and Supplementary Table S2). The role of these enzymes in alginate degradation is supported by distinct homologies to biochemically characterized alginate lyases (Table 3 and Supplementary Figure S2) and co-localization of genes for downstream processing. Alginate lyases encoded in AS1 are probably exolytic, including alt76_01689 with homology to a PL6 from Paraglaciecola chathamensis (Xu et al., 2017) and alt76_01684 with homology to a PL7 from Zobellia galactanivorans that is specifically upregulated in the presence of alginate (Thomas et al., 2017). AS2 encodes another exolytic candidate PL6 and two homologs of endolytic PL7 from Vibrio splendidus releasing trisaccharides (Lyu et al., 2018). In line with other studies, the separate PL17 (alt76_2604) probably degrades released oligomers, exemplified by distinct similarity to PL17_4NEI from Saccharophagus degradans (Park et al., 2014). The separate PL18 (alt76_00345) shares >50% amino acid similarity to PL18_4Q8L from Pseudoalteromonas sp. 0524 with complete conservation of the seven central residues, indicating endolytic release of di- and trisaccharides (Dong et al., 2014). The PL18 furthermore harbors two predicted CBMs from families 16 and 32, which may enable binding different alginate motifs (Sim et al., 2017) or contribute to protein maturation (Dong et al., 2014).

Accordingly, strain 76-1 grew proficiently with alginate as sole carbon source (Figure 4). Faster growth than with glucose was notable, as degradation of the complex alginate polymer (comprising heterogeneous blocks of mannuronate and guluronate) involves more enzymatic steps and induces a prolonged lag phase in other Alteromonas strains (Neumann et al., 2015). Utilization may be boosted by FabG-like enzymes encoded in both AS (Supplementary Figure S3) with homology to alginate-specific reductases in Flavobacteriia (Inoue et al., 2015). These enzymes possibly accelerate the central downstream conversion of 4-deoxy-L-erythro-5-hexoseulose uronate to 2-keto-3-deoxy-D-gluconate and hence prevent bacteriostasis from accumulation of metabolic intermediates (Fuhrman et al., 1998).

The alginolytic machinery of 76-1 is found in several Alteromonas strains, indicating a broader distribution of alginate degradation among this taxon than previously assumed. AS1 and the separate PL17/18 are conserved in the phylogenetic clade comprising strain 76-1, A. naphthalenivorans, A. stellipolaris, and A. addita (Figure 1 and Supplementary Figure S3). AS1 also occurs in the next related species (A. australica) but including the PL17 that is encoded separately in 76-1 (Figure 3B). AS2 shows greater structural variability between strains; with three genes including a putative TBDR plug domain (The UniProt Consortium, 2017) missing in 76-1 and a further reduced version in A. australica and some A. mediterranea (Figure 3B). Different PUL architectures and PL rearrangements are consistent with Alteromonas genome plasticity (López-Pérez et al., 2017) and may confer ecological differentiation, as lyase copy number correlates with enzymatic activity due to gene dosage effect (Hehemann et al., 2016). In the broader eco-evolutionary context, the alginolytic machinery of 76-1 might play an important role, allowing degradation of different alginate structures (Peteiro, 2018) and natural polysaccharide pulses that frequently occur at the productive Patagonian shelf (Acha et al., 2004; Romero et al., 2006). Accordingly, an OTU with 99% 16S rRNA gene similarity was abundant in the alginate-supplemented microcosm from which 76-1 was isolated (Wietz et al., 2015). The ecological relevance was underlined by co-detecting transcripts of alginate lyases from both AS near the original isolation site (Supplementary Table S3), suggesting active roles in natural habitats.

The plasmid of Alteromonas sp. 76-1 is markedly devoted to polysaccharide degradation (Figure 5A and Table 2). Twenty percent of plasmid genes encode CAZymes (plus additional sulfatase and monomer-processing genes) in comparison to 2% CAZymes on the chromosome. This CAZyme condensation essentially denotes the plasmid as an extrachromosomal PUL, matching the size of the largest PUL hitherto described in marine bacteria (Schultz-Johansen et al., 2018). Comparison to characterized CAZymes suggest that the plasmid is linked to degradation of ulvan, a common polysaccharide of green algae (Michel and Czjzek, 2013) mainly consisting of 3-sulfated rhamnose, iduronic acid, glucuronic acid and xylose. Accordingly, strain 76-1 showed distinct growth on ulvan as sole carbon source, with doubling times comparable to glucose (Figure 4). This activity relates to three PL24 and two PL25 ulvan lyases, multiple GHs including predicted rhamnosidases, as well as transporters and monomer-processing genes in the plasmid-encoded ulvanolytic system (Figure 5A). Highly homologous proteins in similar organization also occur in Alteromonas sp. LOR and Pseudoalteromonas sp. PLSV (Kopel et al., 2016; Foran et al., 2017), and several of these homologs have been characterized biochemically (Table 3). All encoded PL24 of strain 76-1 are homologous to the PL24_6BYP ulvan lyase from Alteromonas sp. LOR, although phylogenetic clustering of alt76_04143 with flavobacterial PLs suggests another evolutionary origin (Figure 5B). The proposed functionality of the ulvanolytic system is supported by structural alignments (Figure 5C), showing conservation in residues essential for endolytic ulvan cleavage (Ulaganathan et al., 2017, 2018). Activity assays in strains LOR and PLSV have illustrated potential routes of downstream degradation, showing release of two uronic tetrasaccharides by PL24_6BYP (Ulaganathan et al., 2018) which were in turn degraded by PL25_5UAM from strain PLSV (Qin et al., 2018). Considering sizeable genomic similarities and physiological evidence, we assume similar action by the homologous enzymes in 76-1.

Although regulatory mechanisms remain to be determined, the rapid initiation of growth indicates that the plasmid localization accelerates ulvan degradation, as plasmids are less affected by DNA supercoiling (Dorman and Dorman, 2016) and early-replicating regions (due to the proximity of lyases and origin of replication) are characterized by higher expression (Oliveira et al., 2017). Moreover, the common induction of PUL by substrate availability (Thomas et al., 2017; Koch et al., 2019) suggests that enzymes are only produced in presence of ulvan, reducing fitness costs and the risk of plasmid loss (MacLean and San Millan, 2015). Intriguingly, the encoding contigs in Alteromonas sp. LOR and Pseudoalteromonas sp. PLSV (estimated completeness of draft genomes >99%) harbor partitioning proteins with 83% identity to those on the 76-1 plasmid (alt76_04111–12), suggesting equivalent localization on plasmids acquired in separate horizontal transfer events. Strains LOR and PLSV originate from sea slugs feeding on ulvan-rich macroalgae, indicating ecological relevance of ulvan degradation in host-microbe interactions (Gobet et al., 2018a; Konasani et al., 2018). Isolation of LOR/PLSV from the Brittany coast and hence the opposite side of the Atlantic suggest that ulvanolytic plasmids are distributed over wide geographic scales and maintained in strains from certain niches. Although PUL on plasmids are known in marine bacteria (Zhong et al., 2001) this is only the second report of a plasmid almost fully dedicated to polysaccharide metabolism (Gobet et al., 2018b). Mobile CAZyme elements are likely an important eco-evolutionary factor, representing efficient vehicles of PUL that provide recipient strains with access to specific “polysaccharide niches” on short time scales (Hülter et al., 2017).

The predisposition of Alteromonas sp. 76-1 toward polysaccharides was underlined by testing hydrolysis of 18 AZO-CL labeled substrates (Panschin et al., 2016). Strain 76-1 hydrolyzed diverse glucans with α-(1–4), β-(1–3), and β-(1–4) bonds, including amylose, pullulan, pachyman and barley β-glucan (Supplementary Table S4). Amylase and pullulanase activity likely relates to a gene cluster with four GH13 enzymes plus an additional GH13 in a distant genomic location. Pachyman and barley β-glucan are probably hydrolyzed by the same endo-β-(1–3,4)-glucanase from family GH16. Commonly described xylanase families (GH10, GH30, GH67, and GH115) were not detected in the genome, but xylanolytic activity likely corresponds to a unique chromosomal cluster harboring a GH3 β-(1–4)-xylosidase, a GH9 and several xylose-processing genes (Supplementary Table S4). Proteolytic activity exemplified by casein hydrolysis demonstrated that 76-1 also utilizes other polymer types, attributed to 97 encoded peptidases and proteases predicted by MEROPS (Supplementary Table S2). However, this number is only about half compared to related Alteromonas such as A. macleodii, which in turn encode less CAZymes (Supplementary Table S1). This observation indicates differing preferences between Alteromonas lineages toward polysaccharides and proteins.