Cultivation of Glossomastix

The microalgal strain, Glossomastix sp. PLY432, was obtained from the Roscoff Culture Collection (RCC3688) and cultivated in NEPCC medium (MediaDive: 1724)⁵⁶ at 15 °C, with 140 µmol m⁻² s⁻¹ photosynthetic photon flux density provided by cool-white fluorescent lamps under a 12 h/12 h light/dark cycle. Pre-cultures grown for 2 weeks were used to inoculate fresh NEPCC medium. For large-scale cultures, 1 l culture was grown in 2 l Fernbach flasks with 20 ml pre-culture. For small-scale cultures, 150 ml NEPCC medium in 250 ml cell culture flasks was inoculated with pre-culture to an initial concentration of 3.05 × 10⁴ cells ml⁻¹. Growth and carbohydrate production of Glossomastix was monitored over 60 days by sampling 2 ml of the mixed culture every second day, or every fourth day when Glossomastix was grown in phosphate-limited medium. Phosphate-limited NEPCC medium was obtained by adjusting the final concentration of β-glycerol phosphate. The initial concentration of algal cells in phosphate-limited cultures was 2.25 × 10⁴ cells ml⁻¹. Cell counting was performed with a Neubauer haemocytometer (Marienfeld-superior, 0640111) using 10 µl of culture. Total carbohydrate was determined using 200 µl of culture as described below. The remainder of the 2-ml samples was stored at −20 °C for total monosaccharide determination. After 3 months of growth, Glossomastix cultures were sampled from the top and bottom part of the culture flask to assess low and high viscosity fractions, respectively. Cell morphology was examined with an EVOS FL Auto Microscope (Thermo Fisher) and the size of mucus layers was evaluated manually using ImageJ⁵⁷.

For the semi-continuous growth experiments with inorganic phosphate, KH₂PO₄ was used instead of β-glycerol phosphate as the phosphate source. Pre-cultures were grown for 1 week and inoculated at 5% (v/v) into fresh NEPCC medium supplemented with 14 μM KH₂PO₄ (100 ml + 5 ml) as Batch 1. Each treatment was set up in triplicate. Batch 1 was cultured for 6 days until exponential growth. At this stage, 50 ml of culture was transferred to Batch 2 (50 ml fresh medium without phosphate) while retaining the remainder for continued cultivation. For sampling, 1 ml of cultures was collected for cell counting (10 μl), phosphorus analysis (0.5 ml filtered, −20 °C) and fucose quantification (remainder, −20 °C). Subsequent batches (2–5) followed an identical 3-day transfer protocol: each time, 50 ml from the previous batch was transferred to fresh medium (maintaining a 1:1 dilution) while continuing cultivation of the remaining culture, with consistent sampling at each transfer point.

Phylogenetic analysis of Glossomastix

Glossomastix sp. PLY432 genomic DNA was extracted using the DNeasy Blood and Tissue kit (Qiagen). The V4–V5 region of the 18S rRNA gene was amplified from purified genomic DNA using the 18S universal primers (574F-CGGTAAYTCCAGCTCYAV and 1192R-CAGGTGAGTTTTCCCGTGTT) in Q5 High-Fidelity 2× Master Mix according to manufacturer instructions^58,59. PCR products were then purified with the QIAquick PCR Purification kit (Qiagen) and sequenced at Eurofins Genomics. The 18S rRNA genes of 34 species from 5 classes of the Ochrophyta (Pinguiophyceae, Eustigmatophyceae, Bacillariophyceae (Diatom), Phaeophyceae (Brown algae) and Xanthophyceae) were obtained from NCBI and used to construct a phylogenetic tree along with the 18S rRNA gene of PLY432. The 18S rRNA gene of six species from Saccharomyces was likewise retrieved and used as an outgroup. Details of the genes are listed in Supplementary Table 9. Sequences were aligned using MUSCLE (v.3.8.31) in MPI Bioinformatics Toolkit, with non-aligned regions removed^60,61,62. The phylogenetic tree was constructed via IQ-TREE 2 (ref. ⁶³) under the automatic optimal model selection (Tne+I+G4), calculated with 1,000 bootstrap replications and visualized with TVBOT⁶⁴.

Total carbohydrate quantification

The total carbohydrate content in Glossomastix cultures was measured continuously during growth using the phenol–sulfuric acid method⁶⁵. Briefly, 0.2 ml sample, 5% phenol and concentrated sulfuric acid were combined in a 2-ml tube (1:1:5) and gently mixed. After 10 min at room temperature, the samples were placed in a water bath at 30 °C for 20 min. The content of each tube was cooled to r.t. and transferred to cuvettes, and the absorbance of each sample was then detected at optical density (OD)₄₉₀ using a BioSpectrometer (Eppendorf AG), or 100 μl of sample in a 96-well plate was read using the SpectraMax iD3 plate reader. A calibration curve was constructed by analysing the linear relationship between the concentration and the OD_490 nm of the 99.99% standard glucose stock in the range of 0.02–0.5 mg ml⁻¹.

Phosphate quantification assay

Phosphate concentration was quantified using a colorimetric assay modified from ref. ⁶⁶. Briefly, a standard curve was generated using KH₂PO₄ solutions ranging from 0 to 40 μM. In a 48-well plate, 200 μl of freshly prepared reagent (a mixture of 10% ascorbic acid, 2.5% ammonium molybdate, 6 N sulfuric acid and deionized water at the ratio of 1:1:1:2) was added to 200 μl of each phosphate standard or sample. The plate was sealed with parafilm and incubated at 37 °C with shaking at 500 r.p.m. for 1.5 h. Absorbance was recorded at 820 nm via a SpectraMax iD3 plate reader and phosphate concentrations of samples were calculated on the basis of the standard curve. This assay has a detection limit of ~1 μM, below which phosphate concentrations cannot be accurately quantified.

Deoxy-sugar quantification assay

The quantification of deoxy-sugars was performed using an L-cysteine assay modified from refs. ^67,68. For the assay, a standard curve was prepared using L-fucose solutions ranging from 0 to 10 μg ml⁻¹ in ultra-pure water. Samples (100 μl) and standards were aliquoted into 1.5 ml microcentrifuge tubes, followed by the addition of 234 μl 97% H₂SO₄. The tubes were allowed to cool to r.t. before being heated at 99 °C for 10 min with shaking (800 r.p.m.). After rapid cooling on ice, 10 μl 3% (v/v) L-cysteine HCl was added to each tube. The samples were vortexed and incubated in the dark at r.t. for 90 min. Absorbance measurements were taken at 396 nm and 427 nm using a microplate reader in a 96-well plate. The difference in absorbances (Abs 396–427) was calculated, and deoxy-sugar concentrations were determined on the basis of the standard curve. To ensure accuracy, all measurements were performed in triplicate and a calibration curve with an R² > 0.99 was used for quantification.

Extraction of total polysaccharides from Glossomastix

After 2 months of cultivation, EDTA was added to 1 l Glossomastix culture (50 mM final concentration) and autoclaved (121 °C, 15 min). Whatman glass microfibre filters (Grade GF/F, 0.70 μm) and Millipore Express PLUS membranes (0.22 μm) were used to separate polysaccharide fractions from the supernatant sequentially. The filtered culture supernatant was concentrated on an Amicon stirred cell (Millipore) equipped with a 30-kDa ultrafiltration membrane to collect the high molecular weight (HMW) polysaccharide fraction. The concentrate was continuously dialysed with ultra-pure water (UPW) until the conductivity of the filtrate no longer changed. Conductivity was measured using the SevenCompact Duo S213 meter. The final concentrate was collected and made up to 100 ml with UPW, followed by stirring overnight to detach polymers from the membrane. The desalted samples were lyophilised and stored at r.t. until further analysis. Further separation of high-purity polysaccharides was carried out using AEX and SEC (Extended Data Fig. 4a).

AEX

Crude polysaccharide extracts were further concentrated and purified by anion-exchange chromatography on an XK 26/40 column packed with 90 ml ANX FF resin. The packed column attached to an ÄKTA pure system was first equilibrated with Tris-HCl buffer (50 mM, pH 7.5, degassed) at 5 ml min⁻¹, followed by sample application. Crude sample in 100 ml Tris-HCl buffer (1 g l⁻¹) was filtered (0.22 μm) to remove insoluble materials and then loaded onto the equilibrated resin. Following sample injection, the column was washed with two column volumes of Tris-HCl buffer to remove unconsolidated fractions, followed by two column volumes of Tris-HCl buffer containing 0.5 M NaCl. Finally, fraction collection started immediately using two column volumes of Tris-HCl buffer containing 5 M NaCl as elution buffer to wash the column. The eluates were concentrated and desalted via Amicon stirred cells with 30-kDa ultrafiltration membrane as described above, and lyophilised.

SEC

Final purification of polysaccharides was performed using size exclusion chromatography on two HiPrep 16/60 Sephacryl S-400 HR (120 ml per column) connected in series to a Knauer FPLC system (Azura Bio Purification System) equipped with a refractive index detector. Lyophilised sample (100 mg after AEX) was dissolved in 2 ml Tris-HCl buffer, filtered (0.22 μm) and loaded onto the columns that were equilibrated with 300 ml Tris-HCl buffer (50 mM, pH 7.5, degassed) at a rate of 1 ml min⁻¹ before sample injection. Columns were eluted with 300 ml Tris-HCl buffer and the polysaccharide-containing fractions were pooled, desalted and concentrated via Amicon stirred cells as above.

Quantification of monosaccharides with HPAEC-PAD

For the quantification of monosaccharides, samples were analysed on an ICS-5000+ system (Dionex) with pulsed amperometric detection (PAD) equipped with a CarboPac PA10 analytical column (2 × 250 mm) and a CarboPac PA10 guard column (2 × 50 mm)⁶⁹. In brief, 200 µl lyophilised pure polysaccharide samples (1 mg ml⁻¹) or 200 μl microalgae culture (with cells) were hydrolysed with 200 µl 2 M HCl at 100 °C for 24 h in a pre-combusted (450 °C, 4 h) vial. Supernatants from 100 µl bacterial cultures were hydrolysed with 100 µl 2 M HCl. After complete acid hydrolysis, 100 µl Glossomastix culture samples were dried by speed vacuum to remove HCl and then resuspended in 100 µl UPW, followed by a 1:100 (v/v) dilution. The other samples were diluted with UPW at a ratio of 1:200 (v/v) and then centrifuged at 14,800 r.p.m. (~21,000 × g) (Thermo Scientific Fresco 21 microcentrifuge) for 10 min. Note that after acid hydrolysis, Glossomastix culture samples were dried by speed vacuum and then resuspended in 100 µl UPW, followed by a 100-fold dilution. Subsequently, 100 µl supernatant was analysed by direct injection onto the HPAEC-PAD system. Monosaccharide standard (Supplementary Table 10) mix ranging from 1–10 to 1,000 μg l⁻¹ was used to identify peaks by retention time and to construct standard curves for quantifying the amount of monosaccharide products in the reaction mixture.

Desulfation of fucoidan

A complete desulfation of fucoidan was conducted using a modified version of the solvolytic desulfation protocol outlined in ref. ⁷⁰. First, sodium cations were exchanged with pyridinium ions by dissolving 20 mg of fucoidan in water and passing the solution through an AG 50 W cation exchange resin (Bio-Rad) pre-equilibrated with pyridine (Sigma-Aldrich). The eluate was neutralized with 0.3 ml pyridine and subjected to lyophilisation. Subsequently, the fucoidan–pyridinium salt was dissolved in 15 ml of DMSO (Sigma-Aldrich), and 75 µl of UPW was introduced. The mixture was incubated at 80 °C for 30 min and then subjected to dialysis (8,000 molecular weight cut-off) against 1 M NaCl and UPW before lyophilisation.

NMR characterization of Glossomastix fucoidan

SEC-purified fucoidan is used only for structural and elemental analysis. Purified fucoidan and desulfated fucoidan (10 mg) were first dissolved in 1 ml 99.9% D₂O (Sigma-Aldrich) and lyophilised to reduce the residual water signal. Subsequently, the samples were dissolved in 200 μl D₂O (D-99.96%; Sigma-Aldrich) and transferred to a 3 mm LabScape Stream NMR tube (Bruker LabScape). For NMR analyses, all homo- and heteronuclear experiments were acquired on a Bruker AV-IIIHD 800 MHz spectrometer (Bruker BioSpin) equipped with a 5 mm cryogenic CP-TCI z-gradient probe. Chemical shifts were calibrated using the residual water signal for ¹H (4.75 ppm at 25 °C). The ¹³C chemical shift was internally referenced to DSS (4,4-dimethyl-4-silapentane-1-sulfonic acid) using the absolute frequency ratio⁷¹ (¹³C/¹H = 0.251449530). For chemical shift assignment, the following one- and two-dimensional NMR experiments were recorded at a temperature of 25 °C: 1D proton with water suppression (zgesgp), ¹H-¹³C HSQC with multiplicity editing (hsqcedetgpsisp2.3), ¹H-¹³C heteronuclear two-bond correlation (H2BC) spectroscopy (h2bcetgpl3pr), ¹H-¹³C heteronuclear multiple bond coherence (HMBC) with suppression of one-bond correlations (hmbcetgpl3nd), ¹H-¹H in-phase correlation spectroscopy (IP-COSY) with water suppression with excitation sculpting (ipcosyesgp-tr)⁷², ¹H-¹H total correlation spectroscopy (TOCSY) with 70 ms mixing time and water suppression (clmlevphpr), ¹H-¹³C HSQC-TOCSY with 70 ms mixing time protons (hsqcdietgpsisp.2), ¹H-¹H nuclear Overhauser effect spectroscopy (NOESY) with 80 ms mixing time and water suppression with excitation sculpting and gradients (noesyesgpph). The spectra were recorded, processed and analysed using the TopSpin 3.5 or 4.0.1 software (Bruker BioSpin).

Sulfate quantification

The released sulfate from acid-hydrolysed polysaccharide was measured on a Metrohm 761 compact ion chromatograph equipped with a Metrosep A Supp 5 column and suppressed conductivity detection with 0.5 M H₂SO₄. Ions were separated by an isocratic flow of carbonate buffer (3.2 mM Na₂CO₃, 1 mM NaHCO₃) and the duration of each run was 20 min, with sulfate eluting at 16 min. Chromatograms were analysed with the instrument’s software MagIC Net v.3.2.

Elemental analysis

Elemental analysis was performed using an Elementar modern elemental analyser. Lyophilised sample (0.1–1.0 mg) was transferred to a dry and pre-weighed tin boat, and a small amount of tungsten oxide was added. For the calibration curve, sulfanilamide (0.1–1 mg) was prepared in the same way. Before each test, the samples were degassed by compressing the tin boats.

ELISA

Enzyme-linked immunosorbent assay was used to assess the binding of four fucoidan-specific rat mAbs, namely BAM1, BAM2, BAM3 and BAM4 (ref. ²³) from SeaProbes (https://www.sb-roscoff.fr/en/seaprobes), to Glossomastix fucoidan purified by AEX. The purified fucoidan was dissolved in water and underwent dilution in PBS to 200 μg ml⁻¹, followed by five 2-fold dilutions in PBS, resulting in a total of 6 concentrations. Each antibody–fucoidan combination was tested in triplicate. For the ELISA, 100 μl of the fucoidan solution were added to wells of a 96-well plate (NUNC MaxiSorp, Thermo Fisher). After overnight incubation at 4 °C, wells were washed six times with tap water, and unbound sites were blocked with 200 μl PBS buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.7 mM KH₂PO₄, pH 7.4) containing 5% (w/v) low-fat milk powder (5% MPBS) for 2 h at r.t. After washing nine times with tap water, 100 µl of the mAbs diluted 1:10 in 5% MPBS were added to each well and incubated for 1.5 h at r.t. After washing nine times with tap water, 100 µl of anti-rat secondary antibody conjugated to horseradish peroxidase (A9037, Sigma-Aldrich) diluted 1:1,000 in 5% MPBS were added to each well and incubated for 1.5 h at r.t. Wells were washed with tap water nine times. The plate was developed by adding 100 µl ELISA tetramethylbenzidine (TMB) substrate per well. The enzyme reaction was stopped after 5–10 min by addition of 100 µl 1 M HCl to each well. Absorbance at 450 nm (mAb binding intensities) was measured with a SPECTROstar Nano absorbance plate reader using the MARS software (BMG Labtech). Fucoidan from Laminaria (Glycomix, PSa13) was used as a positive control. Appropriate negative controls were run on every plate.

Media and monitoring of growth

Unless otherwise stated, bacteria were cultivated in marine minimal Tris-HCl medium⁷³ supplemented with vitamins⁷⁴ and iron. The base minimal medium (MMT) contained 2.3% (w/v) sea salts (S9883, Sigma-Aldrich), 9 mM NH₄Cl, 26 mg l⁻¹ ammonium ferric citrate (F5879, Sigma-Aldrich), 50 mM Tris-HCl (pH 7.8) and 1× vitamin mix (1 l 1,000× vitamin mix: 10 mg biotin, 10 mg lipoate, 50 mg Ca-D-pantothenate, 50 mg vitamin B12, 100 mg nicotinate, 100 mg pyridoxamine dihydrochloride, 100 mg thiamine hydrochloride, 40 mg aminobenzoate and 30 mg folate). MMT was further supplemented with 0.03% (w/v) yeast extract (MMT-YE), 0.03% (w/v) casamino acids (MMT-CA) or 50–100 µM KH₂PO₄ (MMT-KDP). At a concentration of 0.03% (w/v) Bacto casamino acids, the medium already contains ~48.9 μM Pi⁷⁵. All media were sterilized by 0.22 µm filtration. Unless otherwise stated, bacteria were grown at r.t. (400 r.p.m.) in 24-well plates (Sarstedt, 83.3922.500), and growth was monitored as OD₆₀₀ using a SpectraMax iD3 plate reader (Molecular Devices).

Bacteria and growth conditions

‘Lentimonas’ sp. CC4 was obtained previously⁹ and Wenyingzhuangia fucoidanilytica CZ1127^T (DSM 100787) was purchased from DSMZ (German Collection of Microorganisms and Cell Cultures). Bacteria were isolated from non-axenic Glossomastix by plating 1:10⁴ diluted microalgal culture onto Difco Marine Agar 2216. Single colonies were obtained from plates after incubation at 15 °C for 12 days. All bacteria were cultured in MMT-CA medium containing 0.05% (w/v) glucose for 5 days, and then inoculated 1:1,000 (v/v) into MMT-CA medium containing 0.05% (w/v) Glossomastix fucoidan. OD₆₀₀ values of the cultures were measured on day 6 and ability to degrade fucoidan was assessed by carbohydrate polyacrylamide gel electrophoresis (C-PAGE) as described below.

Isolation of fucoidan-degrading bacterium V_227

Sediment-associated bacteria were sampled on 27 February 2022 at low tide from the North Beach of Helgoland, Germany, and used to inoculate 13 ml polypropylene tubes (Sarstedt, 62.515.006) containing 3 ml MMT-YE medium supplemented with 0.2% (w/v) fucoidan (glycans eluted in AEX using 0.5 M NaCl). Enrichment cultures were incubated at 17 °C at 115 r.p.m. for 2 weeks, and then diluted 1:100 (v/v) into fresh MMT-YE medium with fucoidan (AEX 5 M NaCl fraction) and cultivated for an additional 7 days. Cultures showing an increase in optical density were diluted 1:10⁵ with fresh MMT medium and plated on solid MMT-YE with 0.05% (w/v) fucoidan and 1% (w/v) agarose. After incubating at r.t. for 5–6 days, the putative fucoidan-degrading isolates appearing as colonies on the plates were re-inoculated into fresh MMT-YE medium to confirm growth and degradation of fucoidan.

Detection of fucoidan degradation

Bacterial degradation of fucoidan in the culture medium was analysed qualitatively by the BSA–acetate method^76,77 or C-PAGE^78,79, and quantitatively by a toluidine blue (TB) assay⁸⁰.

BSA–acetate method

For the BSA–acetate assay, 20 µl culture supernatant was mixed in a 96-well plate with 180 µl acid albumin solution (per litre: 3.26 g of sodium acetate, 4.56 ml of glacial acetic acid and 1 g of bovine serum albumin, pH adjusted to 3.72 to 3.78). Degradation of fucoidan was assessed by observing the formation of cloudy white precipitates against a black background; the degree of turbidity correlates positively with the concentration of acidic polysaccharides and thereby, a clear transparent solution indicated fucoidan degradation.

C-PAGE

Bacterial cultures with fucoidan were centrifuged at 12,000 r.p.m. (13,850 × g) (Thermo Scientific Fresco 21 microcentrifuge) for 10 min, and 24 µl of supernatant was mixed with 6 µl 5× phenol red loading dye before loading onto an acrylamide gel (25% resolving/5% stacking). Electrophoresis was performed for 30 min at 100 V, followed by 1 h at 200 V in native running buffer (1 l: 3 g Tris, 15 g glycine). The gel was stained with 0.005% (w/v) Stains-all in 30% ethanol overnight, and de-stainied with UPW until the background of the gel was clear.

TB method

For the TB assay, 10 µl culture supernatant was mixed with 990 µl TB solution (0.03 mM toluidine blue in 20 mM citrate buffer, pH 3.0, 0.22-µm filtered). Then, 100 μl solution was transferred to a 96-well plate and absorbance measured at 632 nm in a SpectraMax iD3 plate reader (Molecular Devices). Absorbance values were converted to concentration via standard curves constructed on Glossomastix fucoidan (0–1 mg ml⁻¹), where sulfated fucoidan concentration is inversely proportional to OD_632 nm (Supplementary Table 11).

Growth physiology of V_227

Carbohydrate assimilation experiments

Growth assays with mono- and polysaccharides were performed in multiwell culture plates (0.5–1 ml medium) or 13 ml culture tubes (4 ml medium) depending on the availability of carbohydrate substrate and the volume needed for glycan detection. For growth assays with Glossomastix fucoidan, a 4-day-old pre-culture of V_227 grown in MMT-YE was inoculated 1:100 (v/v) into 4 ml fresh MMT-YE medium (20 °C, 110 r.p.m.). Negative control means that no carbohydrates were added. This growth experiment was performed in 13 ml polypropylene tubes (Sarstedt, 62.515.006). Sampling was performed at intervals of 6–14 h and the OD₆₀₀ was measured in 10 mm cuvettes using a BioSpectrometer (Eppendorf AG). Supernatants were centrifuged at 12,000 r.p.m. (~13,850 × g) (Thermo Scientific Fresco 21 microcentrifuge) for 10 min and collected for fucoidan detection (all time points) and monosaccharides detection (some time points) after acid hydrolysis via HPAEC-PAD. This growth experiment was performed in test tubes. In subsequent growth experiments, pre-cultures were inoculated into fresh medium at a ratio of 1:1,000.

Monosaccharides (Supplementary Table 10) were used in the growth assay at a final concentration of 0.05% (w/v). For the fucose, rhamnose, galacturonate and glucuronate, V_227 was incubated for 3 weeks (MMT-CA, 24-well plate or test tubes, 20 °C, 110 r.p.m.). For glucose and galactose, V_227 was incubated for 1 week (MMT-KDP, 24-well plate, r.t., 400 r.p.m.).

Growth experiments with Fucus vesiculosus fucoidan and Glossomastix fucoidan as carbohydrates were carried out in MMT-CA medium at a final concentration of 0.05% (w/v), and OD₆₀₀ was measured at 6–24 h intervals (24-well plate, r.t., 400 r.p.m.). Negative control means that no carbohydrates were added. We simultaneously tested the growth of V_227 with the addition of different polysaccharides in MMT-KDP medium (100 µM KH₂PO₄) (Supplementary Table 10), and the cultivation was performed in 48-well plates (Sarstedt, 83.3923.500) (r.t., 400 r.p.m.). OD₆₀₀ was measured via BioSpectrometer (Eppendorf) after 1 week incubation.

Phosphate-limitation experiments

V_227 was cultivated in MMT-KDP medium (1–50 µM KH₂PO₄ final concentration; for pectin, 1–30 μM) with 0.05% (w/v) polysaccharide as the sole carbon source: Glossomastix fucoidan (5,592.98 μM glycan carbon, 24-well plate, r.t., 400 r.p.m.), Eisenia bicyclis laminarin (14,911.39 μM glycan carbon, 24-well plate, r.t., 400 r.p.m.), corncob xylan (15,381.25 μM glycan carbon, 24-well plate, r.t., 450 r.p.m.) and sugar beet pectin (12,294.61 μM glycan carbon, 24-well plate, r.t., 450 r.p.m.) (Supplementary Table 10). Growth was monitored during 144 h of incubation, after which the cultures were centrifuged at 12,000 r.p.m. (~13,850 × g) (Thermo Scientific Fresco 21 microcentrifuge) for 10 min. Supernatants were subjected to monosaccharide composition analysis via HPAEC-PAD and total carbohydrate quantification as described above. The converted glycan carbon is the sum of the initial molar concentrations of all monosaccharides minus the sum of the molar concentrations of all monosaccharides remaining in the supernatant at 144 h, and then the value obtained was multiplied by 6 for these hexoses having 6 carbon atoms.

Calibration of OD against cell counts

V_227 was grown in MMT-CA medium with 0.05% (w/v) fucoidan. When V_227 was in the logarithmic growth phase (OD₆₀₀ = 0.262), cultures were diluted into groups with different cell densities. The OD₆₀₀ of the diluted cultures was measured in 24-well plates using a SpectraMax iD3 plate reader. Each serial dilution was then further diluted 10,000 times, and 100 μl was spread on MMT-CA plates (0.05% fucoidan and 1% agar) and incubated for 9 days. Linear correlation analysis was performed between the number of colonies counted and the OD₆₀₀ value (Supplementary Table 12).

Super-resolution microscopy of selfish fucoidan uptake

Fucoidan was fluorescently labelled with fluoresceinamine isomer II (Sigma, 51649-83-3) as previously described⁸¹. Pre-cultures of V_227 were diluted 1:1,000 in 1 ml MMT-KDP medium (50 μM KH₂PO4) containing 0.4% (w/v) fluorescently labelled fucoidan, and triplicate cultures were grown in a 24-well plate (r.t., 400 r.p.m.). At several time points over 7 days, 100-μl cultures were fixed with 2% (v/v) formaldehyde and diluted 1:10 in 1× PBS buffer. The fixed cells were collected on polycarbonate filters (0.2 μm, Ø47 mm) before staining with 4′,6-diamidino-2-phenylindole (DAPI), and then mounted onto glass slides using a Citifluor/VectaShield (4:1) solution. Likewise, 100 μl V_227 pre-culture with 100 μl MMT-KDP medium was fixed, DAPI-stained and applied to glass slides as controls. Stained V_227 cells (day 3) were visualized by epifluorescence microscopy on an AxioImager.Z2 microscopic stand (Carl Zeiss) equipped with automated imaging and light-emitting diodes^82,83. Images were acquired at ×63 magnification. For super-resolution microscopy, the cells were visualized on a Zeiss ELYRA PS.1 (Carl Zeiss) using 488 and 405-nm lasers, and BP 502–538 and BP 420–480 + LP 750 optical filters. Z-stack images were taken with a Plan-Apochromat ×63/1.4 oil objective and processed with the ZEN2011 software (Carl Zeiss).

Genome sequencing and assembly

Genomic DNA was extracted with the Gentra Puregene Yeast/Bact kit (Qiagen) using a 5-ml culture of V_227 that was grown in MMT-YE medium (20 °C, 110 r.p.m.) for 4 days, and the purified genomic DNA was then sequenced on a PacBio Sequel II platform at the Max Planck Genome Centre Cologne. Assembly of the V_227 genome was carried out using HiCanu (v.2.2)⁸⁴ based on PacBio hifi reads, and assembly quality was evaluated using checkM (v.1.2.2)⁸⁵.

Relative abundance in TARA and Helgoland spring bloom water

We performed read mapping against the TARA read dataset using Bowtie2 (v.2.3.5.1)⁸⁶. Samtools (v.1.7)⁸⁷ was used to convert the SAM files to BAM, which were then sorted. The trimmed_mean values across all reads were obtained from the sorted BAM files via CoverM v.0.6.1 (https://github.com/wwood/CoverM). The trimmed mean was normalized using the genome equivalence of the RpoB gene as calculated from the TARA dataset. This was then repeated using the Helgoland spring bloom read dataset (mid-March ~ mid-May of 2010, 2011, 2012, 2016, 2018 and 2020)^88,89, except that the trimmed mean was then normalized using the genome equivalence of the RpoB gene as calculated from the Helgoland dataset. The genome equivalence of the RpoB gene was obtained by mapping all reads in the TARA dataset against a collection of reference RpoB sequences^29,90 using the same tools detailed above. The genome equivalence for each sample was calculated as follows: (total number of hits to the RpoB references × average read length) ÷ the average length of the RpoB references. This analysis was then repeated using the Helgoland dataset.

Phylogenetic tree of fucoidan-degrading Verrucomicrobiaceae sp. isolate

Genomes and MAGs belonging to the Akkermansiaceae family were obtained from GTDB (v.214.1)⁹¹. Genomes used were selected on the basis of their quality (completion 5×, contamination ≥50) as determined using checkM (v.1.2.2)⁸⁵ (lineage_wf) and then dereplicated using a 99% ANI threshold for the secondary clustering step in dRep (v.3.4.3)⁹². A total of 246 selected genomes (Supplementary Table 7) were further used for phylogenetic reconstruction based on 120 conserved genes determined using GTDB-tk (v.2.3.2)⁹³ and the v.214.1 GTDB release. A maximum likelihood tree was determined using IQ-TREE v.2.2.2.6 (-m MFP)⁶³ and visualized using the interactive Tree of Life (iTol)⁹⁴.

Genome annotation

Preliminary annotation of the V_227 genome was performed with Prokka (v.1.14.5)⁹⁵. The protein domain families were annotated using Pfam-A HMMs⁹⁶ via HMMER (v.3.3.2, http://hmmer.org) with a ‘cut_tc’ thresholding, and the hit with the highest score value for each sequence was extracted. CAZymes were predicted using a combination of HMMER (E-value < 1 × 10⁻¹⁵, query coverage >35%) against the dbCAN v.11 (https://bcb.unl.edu/dbCAN2/download/Databases/V11/) HMM database and Diamond blastp⁹⁷ (v.2.0.14.152, E-value < 1 × 10⁻²⁰, identity >40% and query coverage >50%) against the 2022 CAZy database⁹⁸. Only results with consistent dbCAN and CAZy annotations were considered reliable. Sulfatases were annotated on the basis of Diamond blastp (v.2.0.14.152, E-value < 1 × 10⁻²⁰, identity >40% and query coverage >50%) searches against the SulfAtlas database (v.1.3)⁹⁹. Transporters were annotated using the Transporter Automatic Annotation Pipeline (TransAAP)¹⁰⁰. Alpha-L-fucosidases were annotated using reference HMM models (PF01120) via HMMER. The MetaCyc database^101,102,103 and BlastKOALA¹⁰⁴ were used for metabolic pathway reconstruction and confirmed by InterPro¹⁰⁵ or CDD^106,107. Prodigal (v.2.6.3)¹⁰⁸ was used to obtain amino acid sequences of proteins encoded in 245 bacterial genomes selected from the Akkermansiaceae family. Proteins in the phosphate transport high-affinity system (PstA, PstB, PstC and PstS) were identified in the 245 genomes via HMMER using the individual HMM modules: PstA (TIGR00974.1, E-value < 1 × 10⁻¹⁵, query coverage >50%), PstB (TIGR00972.1, E-value < 1 × 10⁻⁵⁰, query coverage >50%), PstC (TIGR02138.1, E-value < 1 × 10⁻¹⁵, query coverage >50%) and PstS (TIGR00975.1, E-value < 1 × 10⁻⁰⁵, query coverage >35%). A bacterium was considered to encode Pst when at least three homologues were identified in its genome. The AAI matrix was determined using the aai.rb script from the enveomics collection¹⁰⁹. CAZymes, sulfatases and transporters encoded by the other SW10 genomes were annotated using the same approach detailed above, but a length filter was added to select matches with at least 50% of alignment coverage to the subject.

Data visualization and statistics

Most data visualizations and statistical analyses were generated using Python with Matplotlib¹¹⁰, Pandas¹¹¹, NumPy¹¹², SciPy¹¹³ and Statsmodels¹¹⁴ packages. Growth rate was obtained with R using the Growthcurver¹¹⁵ package. The synteny of the regions encoding the GH141 genes was plotted in R using gggenes¹¹⁶. Figure 3f and Extended Data Figs. 3a and 4d were created with BioRender.com. Extended Data Fig. 4a was created with Google Maps.

Sustainability and inclusion statement

The here-conducted experiments, methods, instruments and organisms are broadly accessible, hence this work is inclusive for many scientists around the globe. The most advanced instrument was a nuclear magnetic resonance machine. The next advanced instrument was an HPAEC-PAD machine for detection of glycans with pulsed amperometric detection. HPLC machines with other modes of detection can also work for the analysis of glycans. Other commonly accessible instruments include a spectrophotometer. Only one bacterial genome was sequenced for this study, limiting the amount of data that require long term storage and sustainability.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.