Sweet spheres: succession and CAZyme expression of marine bacterial communities colonizing a mix of alginate and pectin particles

particles, whereas elevated methylcitrate and glyoxylate cycles suggested nutrient limitation in surrounding waters. The bacterial preference for alginate, whereas pectin primarily served as colonization scaffold, illuminates substrate-driven dynamics within mixed polysaccharide pools. These insights expand our understanding of bacterial niche specialization and the biological carbon pump in macroalgae-rich habitats.


Introduction
Polysaccharides produced by marine macroalgae and phytoplankton are important ecological and biogeochemical agents, serving as structural and storage components for the algae as well as nutrient source for heterotrophic bacteria (Arnosti et al., 2021).A considerable fraction of algal polysaccharides is bound in particles, hotspots of microbial activity with central implications for the biological carbon pump (Stocker, 2012).Hydrogels and transparent exopolymer particles (TEP), a subset of polysaccharide particles forming by self-assembly of anionic polysaccharides in seawater, constitute a global amount of $70 gigatons and are indispensable for the study of particle-microbe interactions (Verdugo et al., 2004;Verdugo, 2012;Cordero and Datta, 2016).The building blocks of marine hydrogels largely originate from macroalgae, in which anionic gelling polysaccharides such as alginate can constitute >50% of the biomass (Mabeau and Kloareg, 1987).In addition, phytoplankton produce anionic polymers and contribute to the marine hydrogel pool (Mühlenbruch et al., 2018).Natural processes of decay or exudation, such as the release of alginate and rhamnogalacturonan from widespread macroalgae (Koch et al., 2019a), presumably result in the formation of hydrogel scaffolds that represent hotspots for microbial life.These events potentially play an ecological role at rocky coasts of temperate seas, which harbor dense forests of macroalgae.
The chemical and structural complexity of marine hydrogels challenges the identification of specific particle-microbe relationships.To reduce this complexity, exposing synthetic model particles to natural bacterioplankton helps understanding the dynamics and drivers of particle colonization.Such approaches have identified hydrogels and other polysaccharide particles as active microbial microhabitats, which harbor distinct communities compared to the surrounding water (Mitulla et al., 2016;Sperling et al., 2017;Zäncker et al., 2019).Furthermore, attached microbes can undergo a temporal succession of primary degraders and opportunistic taxa (Datta et al., 2016;Enke et al., 2018Enke et al., , 2019)).The main indicator of hydrolytic capacities is the presence and diversity of carbohydrate-active enzymes (CAZymes), foremost polysaccharide lyases (PLs) and glycoside hydrolases (GHs), in bacterial genomes (Hehemann et al., 2014).CAZymes are commonly clustered in polysaccharide utilization loci (PUL), operon-like regions facilitating efficient hydrolysis (Grondin et al., 2017).CAZyme numbers, diversity and genomic organization can distinguish bacteria in primary degraders for initial polymer breakdown, and secondary consumers utilizing oligosaccharides, monosaccharides or other compounds released by primary degraders.These types occur across taxonomic boundaries and also within single species (Hehemann et al., 2016;Koch et al., 2020).
Nonetheless, it remains enigmatic how bacterial particle utilization proceeds within the natural 'particlescape' presumably containing a mixture of particle types with different polysaccharide compositionand how these processes are shaped by the taxonomic and functional diversity of the ambient microbiota.The co-availability of hydrogels with different polysaccharide composition might initiate a segregation of bacterial populations by substrate preferences, comparable to hydrolyzing model isolates (Zhu et al., 2016;Koch et al., 2019a).In this context, the CAZyme repertoire is considered to be a stronger driver of niche specialization than phylogenetic relationships (Hehemann et al., 2016;Wolter et al., 2021a).The colonization and utilization of particle resources might also include interactions with free-living microbes, which might benefit from oligosaccharides and other compounds released into the surrounding water.In addition, microbial competition and cooperation can coincide with successional patterns and specific interactions (Ebrahimi et al., 2019;Gralka et al., 2020).
To evaluate particle-specific bacterial dynamics in a mixture of hydrogels, the present study co-exposed alginate and pectin particles to bacterioplankton communities from Helgoland, an island in the southern North Sea surrounded by dense macroalgal forests (Bartsch and Kuhlenkamp, 2000;Uhl et al., 2016).Due to the gelling capacities of alginate and pectin and their demonstrated release from Helgoland macroalgae (Koch et al., 2019a), we assume that related particles occur in this habitat and constitute microhabitats for specialized microbiota.The co-incubation followed by magnetic separation allowed deciphering community composition, functional potential and gene expression depending on particle type and in relation to the free-living fraction.Opposed to our original hypothesis that alginate and pectin particles are utilized by different members of the ambient community, we observed similar compositional and functional patterns with predominant expression of alginate lyases.The identification of alginate as preferred substrate, whereas pectin primarily served as colonization scaffold, illuminates bacterial microhabitat ecology and substrate cycling in macroalgae-rich habitats with diverse polysaccharide budgets.

Results and discussion
We studied taxonomic diversity, functional capacities and gene expression of particle-attached (PA) marine bacterial communities on alginate (AlgP) and pectin (PecP) particles in comparison to their free-living (FL) counterparts.For this purpose, synthetic AlgP and PecP were co-exposed to bacterioplankton collected near Helgoland Island, surrounded by dense macroalgal forests and hence considerable polysaccharide budgets (Supplementary Fig. 1A).The ambient water was sequentially filtered through 100 and 20 μm before AlgP/PecP addition to exclude naturally occurring particles and larger organisms.For the targeted separation of communities, we then carried out triplicate co-incubations in different combinations of magnetic and non-magnetic particles: (i) magnetic AlgP and non-magnetic PecP, (ii) non-magnetic AlgP and magnetic PecP, and (iii) controls without particles.Applying magnetic force allowed the specific recovery of each particle type (Supplementary Video 1).The FL fraction was obtained by 5 μm filtration to remove non-magnetic particles and collecting the flow-through on 0.2 μm filters (Supplementary Fig. 1B).

Do AlgP, PecP and FL harbour specific communities with temporal variability?
Amplicon sequencing of bacterial 16S rRNA genes revealed significant differences between PA and FL communities (PERMANOVA, p < 0.001) but substantial overlap between AlgP and PecP (Fig. 1A, Supplementary Fig. 2A).FL communities from both particle combinations were congruent as expected (Fig. 1A), and FL data were thus combined in subsequent analyses.Furthermore, significant compositional differences of PA and FL communities to those in the ambient seawater and controls without added particles (PERMANOVA, p < 0.001) confirmed the observations as true biological dynamics.Amplicon sequence variants (ASVs) affiliated with Tenacibaculum (Bacteroidetes: Flavobacteriales), Colwellia, Psychromonas and Psychrobium (Gammaproteobacteria: Alteromonadales) constituted up to 60% of both AlgP and PecP communities (Fig. 1B), with significant enrichment compared to the FL fraction (Kruskal-Wallis test, p < 0.001).Hence, particle colonization largely related to few dominant taxa, comparable to other marine polysaccharide particles (Datta et al., 2016;Enke et al., 2019).The finding of related strains with considerable CAZyme repertoires on marine macroalgae (Dong et al., 2012;Martin et al., 2015;Gobet et al., 2018;Christiansen et al., 2020) supports the ecological relevance of our observations.Notably, both Colwellia and Tenacibaculum can be enriched on decaying algae (Fernandes et al., 2012;Zhu et al., 2017) and hence under circumstances when algal polysaccharides might be released and self-assemble into particles.Furthermore, Tenacibaculum and Psychromonas frequently occur during phytoplankton blooms near Helgoland, when bacterial dynamics are largely driven by algal carbohydrates (Teeling et al., 2012;Kappelmann et al., 2019;Krüger et al., 2019).High adaptability and metabolic rates, illustrated by the rapid stimulation of multiple Colwellia ASVs from nearly undetectable levels in the ambient community (Supplementary Fig. 3), could be a competitive advantage during such events.
Glaciecola (Alteromonadales) dominated the FL community (Kruskal-Wallis test; p < 0.0003), with an average abundance of >30% during the first 48 h with low alphadiversity (Fig. 1B, Supplementary Fig. 2B).Notably, the Glaciecola population was dominated by a single ASV (Supplementary Fig. 3), suggesting that nutrient scarcity in FL favoured highly competitive genotypes.This finding underlined that specific biogeochemical conditions can stimulate the predominance of single community members (Pedler et al., 2014).We hypothesize that Glaciecola largely persisted as secondary consumer of particle-derived substrates, supported by genomic evidence from the major corresponding MAG (see below).
The substantial overlap between AlgP and PecP microbiomes contradicted our original hypothesis that the ambient community segregates by particle type.Furthermore, there was little temporal variability in community composition, although we possibly missed rapid successional dynamics as observed in related studies (Datta et al., 2016;Enke et al., 2019).One exception was Pseudoalteromonas, whose sole occurrence at 24 h on both particle types (Fig. 1B; Kruskal-Wallis test, p = 0.002) signifies a polysaccharide pioneer (Hehemann et al., 2016).This notion is supported by alginolytic and pectinolytic capacities of various Pseudoalteromonas species, which generally respond quickly to nutrient input (Ivanova et al., 2014;Hehemann et al., 2017).The AlgP microbiota established within the first 24 h and then remained unchanged, whereas Tenacibaculum established with temporal delay on PecP with peak abundances at 60 h (Fig. 1B; Kruskal-Wallis test, p = 0.04).Faster stabilization of the AlgP community indicates that alginate was the major nutrient source, as discussed below in context of metagenomic and metatranscriptomic evidence.One notable exception was Catenovulum (Alteromonadales), which solely established on PecP after 60 h (Fig. 1B; Kruskal-Wallis test, p = 0.01) and was the only taxon linked to pectin degradation (see below).

Do AlgP, PecP and FL communities differ in functional diversity and gene expression?
As taxonomic and metagenomic richness are overall connected (Salazar et al., 2019), we expected contrasting functional potentials in PA and FL communities, whereas metabolic capacities of AlgP and PecP communities should be largely congruent.However, AlgP and PecP microbiomes might differ in gene expression patterns, as these can be independent from taxonomic composition (Salazar et al., 2019).For instance, certain taxa encode both alginate and pectate lyases (Koch et al., 2019a) and might express the corresponding genes differentially depending on particle type.To evaluate these aspects, we analyzed the metagenome (24 and 60 h) and metatranscriptome (60 h) of AlgP and PecP communities in relation to the FL fraction (Supplementary Table 1).This approach included both community-wide and genome-centric perspectives through MAGs.
The metagenomic library of 21 gigabases comprised $192 000 genes predicted by Prokka, 47% of which were functionally annotated using UniProtKB, KEGG and/or COG databases.Two percent of all genes were predicted to encode CAZymes according to dbCAN2 (Supplementary Table 2).We first assessed overarching differences between PA (i.e.occurring on both AlgP and PecP) and FL communities to identify general signatures of planktonic and attached niches.Transcripts from the citric acid cycle, glycolysis/gluconeogenesis and amino acid metabolism were abundant in both PA and FL metatranscriptomes but numerous pathways differed (Supplementary Table 3).Overall, $60% of all transcripts were differentially abundant between PA and FL communities (Fig. 2A, Supplementary Table 3), matching metatranscriptomic evidence in other marine ecosystems (Satinsky et al., 2014).Transcript abundances of glutamine synthetase, one key enzyme of bacterial nitrogen assimilation converting ammonium into glutamine, peaked in PA communities (Fig. 2B).This observation suggests considerable ammonium uptake to meet the nitrogen demand for fuelling polysaccharide-derived carbon into protein biosynthesis, supported by abundant transcripts of related transporter and regulator genes (Fig. 2B; Wilcoxon rank-sum test, p < 0.05).Notably, the biosynthesis of valine, leucine and isoleucine peaked in PA, but their degradation in FL communities (Supplementary Table 3; Wilcoxon rank-sum test, p < 0.01).We interpret this observation as provision of amino acids from actively growing PA to substrate-limited FL bacteria.In this context, leucine exchange between bacteria on polysaccharide particles and the surrounding water (Enke et al., 2019) might be a stabilizing component of their interactions (Johnson et al., 2020).Glyoxylate, dicarboxylate and pyruvate metabolism peaked in FL communities (Supplementary Table 3).Furthermore, induction of the methylcitrate cycle and the glyoxylate shunt (Fig. 2B; Wilcoxon ranksum test, p < 0.01) supports the notion of substrate limitation in the FL niche, matching transcriptomic responses of starved bacterioplankton (Kaberdin et al., 2015).These pathways likely promoted persistence by generating energy from short-chain fatty acids but might also alleviate iron limitation or oxidative stress (Palovaara et al., 2014;Ahn et al., 2016;Dolan et al., 2018;Koedooder et al., 2018;Serafini et al., 2019).In Alteromonas macleodii, similar expression patterns were interpreted as maintenance metabolism (van Bodegom, 2007;Beste and McFadden, 2010;Koch et al., 2019b).
Next, we specifically compared AlgP and PecP to identify polysaccharide-specific patterns.Communities on AlgP and PecP only slightly differed in functional potential and gene expression (Fig. 2A), compliant with their compositional overlap (Fig. 1).Only 2% of transcripts were differentially abundant, without community-wide patterns in specific functional categories (Supplementary Table 3).On AlgP, higher transcript abundances of alkaline phosphatase genes possibly counteracted beginning phosphate limitation, comparable to late stages of natural TEP colonization (Berman-Frank et al., 2016).Furthermore, higher transcript abundances of predicted prophages (Fig. 2C, Supplementary Table 3) indicates the induction of lytic cycles and corresponding release of organic matter (Breitbart et al., 2018).These events potentially stimulated secondary consumers such as Aureispira, which only appeared after 60 h (Kruskal-Wallis test, p = 0.01).This predatory taxon can feed on metabolic products or cell debris from other bacteria, fuelled by its capacity to adhere to anionic polysaccharides (Furusawa et al., 2015).On PecP, a single MAG related to Catenovulum accounted for the vast majority of differentially abundant transcripts, supporting the predisposition of this taxon towards pectin (see below).The PecP-specific upregulation of lipopolysaccharide-related mla, lpt and kds genes presumably stimulated biofilm formation, an important advantage for colonization and assimilation of particulate substrates (Sivadon et al., 2019).
Saccharina and Fucus, which are abundant in our sampling area and release alginate into the water column (Koch et al., 2019a), offers an explanation why alginatedegrading genes and organisms predominated.In contrast, we only detected three PL1 pectate lyases and few other pectin-related genes (CE8, GH28, GH105).These results indicate that pectin is not a prime bacterial substrate in kelp forests, although pectinolytic bacteria occur in diverse marine habitats (Van Truong et al., 2001;Hehemann et al., 2017;Hobbs et al., 2019) and pectinous substrates are exuded by Helgoland macroalgae (Koch et al., 2019a).Instead, we hypothesise that PecP primarily served as colonization scaffolds for alginolytic bacteria.We propose that the predominant taxa are generally adapted to life on (polysaccharide) particles, favoring cross-particle colonization especially as AlgP were available nearby.The fast sinking of the relatively large particles (diameter $200 μm) resulted in a loose bottom layer, with close spatial contact of both particle types.This 'particlescape' potentially allowed cross-particle interactions and utilization of alginate, even if attached to PecP.Nonetheless, significantly higher abundances of alginate lyase transcripts on AlgP (Wilcoxon rank-sum test, p = 0.0002 to 10 À16 ) indicates that PecP associates were less alginolytic, possibly attributed to diffusion losses.
Approximately 3% of genes in the major PA-MAGs were annotated as CAZymes (Table 1), mostly PL6, PL7 and PL18 alginate lyases with CBM32, CBM16 or CBM6 domains (Fig. 5).Considerable transcript abundances of alginate lyase and monomer-processing genes kdgA, kdgF, kdgK and dehR illustrate the complete metabolization of alginate (Fig. 5, Supplementary Table 4, Supplementary Fig. 5).Approximately half of CAZymes from Colwellia, Psychrobium and Psychromonas MAGs harbour predicted signal peptides (Supplementary Table 4) and were hence likely secreted, although we cannot discern whether these were indeed free enzymes or anchored to the cell membrane.Presumably, CAZyme secretion into the polysaccharide matrix facilitated particle utilization, enhancing polymer hydrolysis and subsequent oligomer uptake (Vetter et al., 1998).MAG Gla-32 affiliated with the dominant FL taxon Glaciecola encoded only three PLs but 18 GHs, with the highest transcript abundances of families GH3, GH13 and GH23 (Fig. 5).The lower fraction of signal peptides in its CAZymes (30%) indicates that secreted enzymes are less relevant when free-living, pointing towards opportunistic interactions with primary hydrolyzers.
Overall, only some CAZymes of each MAG's repertoire showed elevated transcript abundances (Fig. 5).We assume that the 'silent' CAZymes enable the degradation of other carbohydrates.For instance, the Colwellia-MAG Col-24 encodes a homolog of the rarely described PL29 family (locus tag 50313), potentially activated in presence of chondroitin sulfate, dermatan sulfate or hyaluronic acid (Ndeh et al., 2018).Although Col-24 clusters with the hydrolytic model isolate Colwellia echini A3 (Fig. 4A) at 80% average nucleotide identity (ANI), a BLASTp survey revealed that CAZymes targeting agar, carrageenan and furcellaran are not shared with strain A3 (Supplementary Table 4).Divergent CAZyme repertoires in related Colwellia spp.presumably reflect their different habitats (Christiansen et al., 2020).
Diversity of PL7 homologues in Psym-73.The presence of 14 PL7 genes in the Psychromonas-MAG Psym-73 signifies a marked specialization towards alginate, as A. Maximum-likelihood phylogeny based on 92 single-copy core genes in the context of related genomes.Dots designate nodes with >90% bootstrap support.Supplementary Fig. 4 shows an extended tree including medium-quality MAGs and additional related genomes.B. Normalized coverage in metagenomes at 24 and 60 h.The scales of y-axes differ for better visualization.hydrolytic activity scales with CAZyme number (Hehemann et al., 2016).Two of these homologs (locus tags 20477 and 38641) exhibited the highest transcript abundances of all PL7 genes in our dataset (Supplementary Table 4).Both are related to biochemically characterized lyases from Vibrio strains (Supplementary Fig. 8) isolated from macroalgae or seawater (Roux et al., 2009;Badur et al., 2015;Sun et al., 2019).For instance, PL7_38641 has 72% amino acid identity to AlyD of Vibrio splendidus, an endolytic lyase releasing three oligomer fractions from guluronate-rich sections (Badur et al., 2015).Prediction of a Lipo signal peptide while lacking a CBM indicates that PL7_38641 is anchored as outer membrane lipoprotein (Supplementary Table 4), comparable to AlyA5 from Zobellia galactanivorans (Thomas et al., 2013).
The two adjacent PL7 genes from different subfamilies (Fig. 6B) possess predicted Sec and Lipo signal peptides respectively (Supplementary Table 4), indicating complementary membrane-bound versus secreted localization to maximize alginate utilization.Homologues of PL7_75913 with >50% amino acid identity also occur in Simiduia (Cellvibrionales), Reichenbachiella and Marinoscillum (Cytophagales), indicating wide ecological relevance (Spring et al., 2015).PL7_75915 from the poorly described subfamily 3 has 55% amino acid identity to a structurally resolved lyase from Persicobacter (Sphingobacteriales) specialized towards alginate of high molecular weight (Sim et al., 2017), suggesting a role in initial depolymerization.We hypothesize that the two PL variants originate from separate horizontal acquisition events with subsequent insertion into the same genomic locus, considering their low similarity and different branching in the phylogenetic tree (Supplementary Fig. 8).Overall, highly variable transcript abundances of PL7 genes (Supplementary Fig. 8) suggest that different variants are activated by specific biochemical conditions, for instance, different alginate characteristics (e.g.polymer length; dissolved or particulate form; or the ratio of mannuronate to guluronate monomers).
represent related species with presumably wide ecological relevance on polysaccharide particles.Psym-73 and strain B3M02 share nine homologous PLs (Supplementary Table 5), however, encoded in different genomic contexts.This variable organization, together with the higher PL count in Psym-73, indicates considerable CAZyme diversity and genomic rearrangements among hydrolytic Psychromonas.Ten-26 and strain E3R01 share 29 homologous CAZymes.However, no PLs were detected in E3R01 while Ten-26 encodes 17 (Supplementary Table 5).These observations indicate CAZyme-related niche specialization among Tenacibaculum species, consistent with CAZyme variability in Tenacibaculum type strains ranging from eight PLs in T. jejuense to none in T. mesophilum (Lombard et al., 2014).
A single, rare pectin degrader.Despite the compelling evidence that alginate was the preferred bacterial substrate, MAG21 is a candidate for pectin utilization.MAG21 accounted for $95% of differentially abundant transcripts on PecP (Fig. 7A), contributing to significantly elevated GH abundances and normalized coverage compared to AlgP (Fig. 7B, Kruskal-Wallis test, p < 10 À16 ).MAG21 encodes several genes for galacturonate degradation and processing of pectin monomers.A GH53 endo-galactanase gene with the fourth-highest transcript abundance of all CAZymes on PecP (locus tag 118003) might cleave galacturonate-rich side chains from pectinous substrates by endolytic activity (Benoit et al., 2012).Moreover, colocalized GH53 and GH2 plus two carbohydrate transporter genes (Fig. 7B) resemble a galacturonan-related PUL in Bacteroides thetaiotaomicron (Luis et al., 2018).Thus, MAG21 might utilize oligomeric side chains of pectin, processing the resulting galacturonate via tagaturonate and altronate to 2-keto-3-deoxy-D-gluconate through UxuABC (Supplementary Table 4, Supplementary Fig. 5).Genome-based taxonomy assigned MAG21 to the uncultured taxon GCA-2401725, whereas the majority of CAZymes possess homologs in Catenovulum spp.(Supplementary Table 4), recently described for pectinolytic capacities (Furusawa et al., 2021).Growth of the Catenovulum isolate CCB-QB4 on unsaturated galacturonate has been linked to GH28 and GH105 enzymes, which are also encoded by MAG21 (Supplementary Table 4).Together with the restriction of Catenovulum ASVs to PecP (Fig. 1C) we propose that MAG21 is taxonomically and functionally related to Catenovulum and degrades galacturonate.The medium quality of MAG21 (76% estimated completeness) may explain why pectate lyases are missing compared to CCB-QB4.Alternatively, MAG21 indeed only encodes an incomplete degradation cascade, only accessing oligomeric side chains or oligomers released by the activity of primary degraders that encode pectate lyases.
An additional PecP-specific pattern occurred in Psychromonas-MAG Psym-73, with significantly higher transcript abundances of a hybrid gene cluster for the biosynthesis of a siderophore as well as spermidine (Supplementary Table 3, Supplementary Fig. 7B).Homologues of the siderophore-encoding section have been identified in diverse marine bacteria with shown ironchelating activity (Koch et al., 2019b), indicating a similar functionality in Psym-73.The upstream spermidinerelated section has $40% amino acid similarity to the polyamine synthesis pathway of Vibrio, indicating a PecP-specific role in biofilm formation (Lee et al., 2009).

Ecological conclusions
The predominance of alginolytic pathways demonstrates alginate particles as preferred microbial substrate, whereas pectin was primarily a colonization scaffold.The establishment of similar communities contradicted our original hypothesis of community segregation by substrate preferences.On the contrary, the expression of alginate lyases when attached to pectin signify the concept of a 'particlescape' encompassing cross-particle interactions.Such a scenario might resemble natural processes when algal polysaccharide exudates enter the water column, self-assemble into particles and sink to the seafloor.Under such circumstances, bacteria might utilize alginate even if attached to neighboring microhabitats, e.g. by secreting extracellular CAZymes or exploiting hydrolytic activity of co-occurring microbes.The predominance of few taxa indicates that polysaccharide availability stimulates only certain community members, outcompeting most other strains by their extensive CAZyme repertoire.Nonetheless, identification of a single MAG with pectin-specific dynamics suggests that numerically rare but competitive bacteria can establish in specific niches.Single-particle incubations and the application of 13 C-labelled substrates followed by NanoSIMS or stable isotope probing might answer these open questions in future studies.Altogether, our study illuminates central elements of the biological carbon pump in macroalgae-rich habitats, with implications for microscale ecology, niche specialization and bacteriaalgae interactions.

Characteristics of polysaccharide particles
Custom polysaccharide particles consisting of alginate (CAS #9005-38-3) or pectin (CAS #9000-69-5), approximately 200 μm in diameter, were fabricated by geniaLab (Braunschweig, Germany) by immersing polysaccharide in calcium chloride solution containing metallic beads (Supplementary Methods).Four different versions were produced: magnetic alginate and pectin particles (polysaccharide coated on magnetite core) as well as nonmagnetic alginate and pectin particles (polysaccharide coated on ferrous iron core).Particles contained on average 4% polysaccharide (Supplementary Table 6), approximating the 1:100 solid:solvent ratio in natural hydrogels (Verdugo et al., 2004).Magnetic and nonmagnetic particles of each polysaccharide allowed coincubation of both polysaccharides and hence similar selection pressures per treatment (i.e.always two available particle types, only varying in magnetism).Applying external magnetic force (Supplementary File 1) subsequently allowed the targeted sampling of particle types.

Seawater sampling and experimental set-up
Seawater was sampled from approx. 1 m depth above macroalgal forests at Helgoland Island (54.190556 N, 7.866667 E) in June 2017.Seawater was filtered through a 100 μm mesh, brought to the lab within 2 h, and filtered again through a 20 μm mesh to remove larger particles and organisms.Each 12 L of filtered seawater were distributed into 20 L Clearboy bottles (Nalgene, Rochester, NY) previously rinsed with the same seawater.Per bottle, 10 μM NaNO 3 and 1 μM NaH 2 PO 4 (wt./vol.)were added as additional nitrogen and phosphorous source to avoid limitation.Three experiments were set up in triplicate: (i) magnetic alginate particles and non-magnetic pectin particles, (ii) non-magnetic alginate particles and magnetic pectin particles, and (iii) control without particles (Supplementary Fig. 1).Each particle type was added at 3500 L À1 , resulting in $42 000 particles per bottle.Bottles were incubated statically at 15 C (approx. in situ temperature) in the dark.
Sampling and nucleic acid extraction of particleassociated and free-living cells 250 ml of the original seawater were filtered onto 0.2 μm polycarbonate filters for determination of the ambient in situ community (start).Filters were flash-frozen in liquid nitrogen and stored at À80 C. Incubations were sampled after 24, 48 and 60 h.At each sampling point, bottles were mixed by inversion and ca.550 ml withdrawn into rinsed measuring cylinders.Each sample was distributed (2 Â 250 ml) into sterile RNase-free Nunc tubes (cat. no. 376814;Thermo Fisher Scientific,Waltham,MA). Particle-associated communities on magnetic AlgP or PecP were sampled by holding a neodymium magnet (cat.no.Q-40-10-10-N; Supermagnete, Gottmadingen, Germany) next to the tube.AlgP and PecP were washed with sterile seawater (filtered through 100 and 20 μm; mixed 3:1 with ddH 2 O to prevent salt precipitation during autoclavation for 20 min at 121 C) and transferred to 2 ml RNase-free microcentrifuge tubes.The supernatant was transferred to a separate tube and non-magnetic particles were removed by filtration through 5 μm polycarbonate filters.The flow-through was captured on 0.2 μm polycarbonate filters to obtain the FL community.All samples were directly flash-frozen in liquid nitrogen and stored at À80 C. Simultaneous extraction of DNA and RNA was done using a modification of Schneider et al. (2017).Purified DNA and RNA were sent on dry ice to DNASense (Aalborg, Denmark) for quality control and sequencing.For particles, several subsamples per replicate were pooled to obtain sufficient DNA and RNA (Supplementary Table 1).

Metatranscriptomics
RNA was quantified in duplicate per sample using the Qubit BR RNA assay (Thermo Fisher Scientific).RNA quality and integrity were confirmed using TapeStation with RNA ScreenTape (Agilent, Santa Clara, CA).rRNA was depleted using the Ribo-Zero Magnetic kit (Illumina) and residual DNA removed using the DNase MAX kit (Qiagen, Hilden, Germany).Following sample cleaning and concentrating using the RNeasy MinElute Cleanup kit (Qiagen), rRNA removal was confirmed using Tap-eStation HS RNA ScreenTapes (Agilent).Sequencing libraries were prepared using the TruSeq Stranded Total RNA kit (Illumina), quantified using the Qubit HS DNA assay (Thermo Fisher Scientific) and size-estimated using TapeStation D1000 ScreenTapes (Agilent).For RNA from particle samples, four to five subsamples per replicate were pooled in equimolar concentrations and sequenced on a HiSeq2500 in a 1 Â 50 bp Rapid Run (Illumina).As the first sequencing run did not deliver sufficient data for seven metatranscriptomes, a second run was performed on the same library.PCA confirmed consistent sequencing runs (data not shown), and read counts were subsequently aggregated.Raw fastq sequence reads were trimmed using USEARCH v10.0.2132 (Edgar, 2010) using -fastq_filter and settings -fastq_minlen 45 -fastq_truncqual 20.rRNA reads were removed using BBDuk (http://jgi.doe.gov/data-and-tools/bb-tools) using the SILVA database as reference (Quast et al., 2013).Reads were mapped to the predicted genes using Minimap2, discarding reads with sequence identities <0.98.Relative transcript abundances were obtained by dividing raw counts by the length of each gene (RPK) and normalized by per-million scaling factors.Resulting transcripts per million (TPM) were summed per gene annotation (Supplementary Table 2).Differential transcript abundances were calculated on raw read counts using the default DESeq2 workflow in R v3.6 (Love et al., 2014;R Core Team, 2018) in RStudio (https:// rstudio.com),only considering log2-fold changes >2 with p adj < 0.001 (Supplementary Table 3).

Fig 1 .
Fig 1. Bacterial community composition based on amplicon sequence variants.A. Non-metric multidimensional scaling reveals different communities on alginate (AlgP) and pectin particles (PecP) compared to the free-living fraction (FL), control without particles (CTR), and the ambient seawater (start).B. Relative abundances of dominant genera on AlgP and PecP in comparison to FL, CTR and start communities.

Fig 3 .
Fig 3. Community-wide diversity and expression of CAZymes. A. Average transcript abundances of polysaccharide lyases (PL), carbohydrate-binding modules (CBM) and glycoside hydrolases (GH) with mean TPM >50.B. Top100 CAZymes with the highest expression in the different bacterial fractions.

Fig 4 .
Fig 4. Phylogeny and abundance of metagenome-assembled genomes (MAGs).A. Maximum-likelihood phylogeny based on 92 single-copy core genes in the context of related genomes.Dots designate nodes with >90% bootstrap support.Supplementary Fig.4shows an extended tree including medium-quality MAGs and additional related genomes.B. Normalized coverage in metagenomes at 24 and 60 h.The scales of y-axes differ for better visualization.

Fig 5 .
Fig 5. Diversity and expression of CAZymes in metagenome-assembled genomes.Mean transcript abundance (heatmaps) and number (bars) of CAZyme-encoding genes in MAGs affiliated with the dominant PA and FL community members.

Fig 7 .
Fig 7. MAG21 as candidate for pectin degradation.A. Numbers of differentially abundant transcripts compared to the major PA-MAGs on PecP versus AlgP.B. Elevated transcript abundances of glycoside hydrolase genes on PecP.C, left panel: PUL with similarities to the pectinolytic operon BT4667-4673 in Bacteroides thetaiotaomicron.C, right panel: PUL encoding GH105 and GH73 genes plus a hypothetical protein with 60% amino acid identity to an alpha-amylase from Paraglaciecola arctica (UniProtKB accession K6ZD77; indicated by asterisk).