Metagenome-derived SusD-homologs affiliated with Bacteroidota bind to synthetic polymers

ABSTRACT Starch utilization system (Sus)D-homologs are well known for their carbohydrate-binding capabilities and are part of the sus operon in microorganisms affiliated with the phylum Bacteroidota. Until now, SusD-like proteins have been characterized regarding their affinity toward natural polymers. In this study, three metagenomic SusD homologs (designated SusD1, SusD38489, and SusD70111) were identified and tested with respect to binding to natural and non-natural polymers. SusD1 and SusD38489 are cellulose-binding modules, while SusD70111 preferentially binds chitin. Employing translational fusion proteins with superfolder GFP (sfGFP), pull-down assays, and surface plasmon resonance (SPR) has provided evidence for binding to polyethylene terephthalate (PET) and other synthetic polymers. Structural analysis suggested that a Trp triad might be involved in protein adsorption. Mutation of these residues to Ala resulted in an impaired adsorption to microcrystalline cellulose (MC), but not so to PET and other synthetic polymers. We believe that the characterized SusDs, alongside the methods and considerations presented in this work, will aid further research regarding bioremediation of plastics. IMPORTANCE SusD1 and SusD38489 can be considered for further applications regarding their putative adsorption toward fossil-fuel based polymers. This is the first time that SusD homologs from the polysaccharide utilization loci (PUL), largely described for the phylum Bacteroidota, are characterized as synthetic polymer-binding proteins.

chitin (13), and many others.Distinct Bacteroidota can present a range of SusD-like proteins with varied substrate specialization (14,15).To our knowledge, SusD-like proteins with putative binding affinity toward synthetic polymers were not characterized before.The importance of finding suitable proteins with high affinity to plastics relies on the problem that they are ubiquitous pollutants, and recycling is not effective due to the alarming pollution rate (16)(17)(18)(19)(20)(21).Plastics are modified by weathering, which results in the production of micro-and nanoparticles (20,22,23).These particles not only affect the fauna and flora but also human health, once they accumulate into the food chain and might affect homeostasis (24).
Previously, it was shown that bacteroidotal enzymes can hydrolyze polyethylene terephthalate (PET) (25).Therefore, in this work, we studied if and to which extent Bacteroidota might harbor PET binding surface proteins.In this study, we identified three SusD-homologs ( SusD1, SusD38489, and SusD70111) that primarily bind cellulose or chitin, but also PET, bis-(2-hydroxyethyl)terephthalate (BHET) and nylon (PA6) with varying strength.

Origin of the proteins
SusD38489 and SusD70111 were identified from the elephant feces metagenome (26).The 6-year old Asian elephant Elephas maximus was living in the zoo Hagenbecks Tierpark (Hamburg, Germany) and was mainly fed with grass, hay, leaves, and twigs, in addition to fruits and vegetables.Besides, the elephant was breast-feeding.The work did not include endangered or protected species, and hence specific permissions were not required.The samples were collected aseptically by the zoo staff (26).On the other hand, SusD1 was identified by the hidden Markov model (HMM), and it was obtained from the cow rumen metagenome (fosmid Sc00044) (27) (Table 1).The rumen samples were collected from Bos indicus consuming Rhodes grass Chloris gayana in Rockhampton (QLD, Australia) (27).The sampling was carried out in accordance with the protocols approved by the Rendel Laboratory Animal Experimentation and Ethics Committee (27).The genes were previously amplified from the metagenomes, and they are available in our collection.

Phylogenetic analysis
Protein BLAST was carried out on using the NCBI "Non-redundant (NR) protein sequen ces" Database (available at BLAST: Basic Local Alignment Search Tool (nih.gov);accessed on 16/01/2024).Only the sequences with 90% of coverage and at least 70% of identity were selected.Besides, entries lacking species identification, belonging to multispecies or without the description of the isolated environment, were not included.The proteins were aligned using T-coffee structural alignment v.11.00 in expresso mode (28)(29)(30).A neighbor-joining tree with bootstrap adjusted to 1000 was built with MEGAX v.10.2.4, and the tree was colored using iTOL v.6 (31).
SusD-homologs from hot springs were found in the NCBI protein database (available at National Center for Biotechnology Information (nih.gov);accessed on 16/01/2024).We selected these proteins because they represent a distinct niche (32).The terms "SusD, " "hot spring, " "NOT hypothetical, " and "NOT putative" were applied to the search.Receptor antigen B (RagB) proteins were also searched using the NCBI protein database.In this case, the search for "RagB, " "Bacteroidota, " "NOT partial, " "NOT hypothetical, " and "NOT putative" retrieved thousands of hits, in which most of them consisted of the oral pathogen Porphyromonas gingivalis.For this reason, the protein sequences were selected based on their availability in the literature (33)(34)(35)(36)(37)(38).RagB was considered an external group, and the tree was rooted in this node.Additionally, forty-seven crystal structures of distinct SusDs from the PDB database (39,40) were compared in an "all against all" analysis with the Dali server (41).The co-crystallized SusC/SusD structures were not considered.
A search for the evolutionary conserved motif named tetratricopeptide repeat (TPR), present in sugar-binding proteins such as SusD and RagB (42), was carried out with default parameters, using the online server TPRpred (43) (available at Bioinformatics Toolkit (mpg.de);accessed on 27/09/2023).The tool calculates the probability that a protein present TPR by using P-value-dependent scores (43).

Gene cloning, fusion, and site-directed mutagenesis
Each susD gene was flanked by the restriction enzymes NdeI and SalI (New England Biolabs, USA).Heat shock transformation occurred in E. coli DH5α using standard protocols (44), and the recombinant proteins were expressed in E. coli BL21(DE3).The respective cultures were grown under aerobic conditions in lysogeny broth (LB) medium plus 100 µg/µL of ampicillin, at 37°C, 150 rpm, overnight.Cryotubes with 20% glycerol (vol/vol) and bacterial culture were kept long term at −70°C.N-terminal signal peptides were predicted using the SignalP DTU v.6.0 server (45).Tables S1 to S4 provide detailed information regarding the primer pairs, PCR cycles, and final product size (bp).The cloning was performed in accordance with Sambrook (44), and the DNA was submit ted to Sanger sequencing (Microsynth Seqlab, Germany).The sequence was aligned, trimmed, and analyzed using ChromasPro v.2.1.10(Technelysium Pty Ltd, Australia) and NCBI BLAST (available at BLAST: Basic Local Alignment Search Tool (nih.gov);National Library of Medicine, National Center for Biotechnology Information).Further information regarding the colony PCR primer pair and cycle can be found in Table S4.Table 2 presents detailed information regarding the strains used in this work.
The C-terminal of each protein was fused with the GGGGS linker to sfGFP (46) via blunt-end cloning.The first PCR consisted of the amplification of SusD with the linker, alongside the overhangs to anneal the backbone (sfGFP in pET21a+).In the second PCR, SusD was applied to the master mix as a mega primer.Before heat shock transformation, the PCR product was treated with the restriction enzyme DpnI (NEB, Germany).a The empty vector pET21a(+) was purchased from Novagen/Merck (Darmstadt, Germany).The vector characteristics include the following: lacI, Amp R , T7-lac-promoter, and C-terminal His-6-tag coding sequence.

Protein expression and purification
The proteins were expressed with autoinduction medium (47) plus 100 µg/µL of ampicillin.The cultures were incubated at 37°C with constant shaking until the OD 600 reached 0.7 and then placed at 28°C overnight.The centrifugation occurred at 13,000 rpm, 20 minutes at 4°C (Beckman Coulter Avanti JXN-30; Rotor JA-10).
The cells were resuspended in NPI-10 lysis buffer (50 mM NaH 2 PO 4 (Merck, Germany), 300 mM of NaCl, and 10 mM of imidazole (Carl Roth, Germany)) and disrupted with French Pressure Cell Press (American Instrument Company) at 1250 psi.The crude cell extract was incubated with one bed volume of Ni-NTA agarose (Macherey-Nagel, Germany), both placed in a polypropylene column (Qiagen, Germany), and washed with NPI buffer containing 20 mM imidazole.Elution of the purified protein was performed with NPI buffer containing 250 mM imidazole.
The buffer was changed to 0.1 M potassium phosphate buffer with pH 7 (prepared from 1 M of K 2 HPO 4 and 1 M of KH 2 PO 4 stock solutions (Carl Roth, Germany)), with a Vivaspin 20 column (Sartorius, Germany).The quality was checked with SDS PAGE and Western blot, and the concentration was accessed through NanoPhotometer NP80 (Implem, Munich, Germany) with the molecular weight and extinction coefficient estimated by the Expasy ProtParam tool (48).Protein purity was estimated with ImageJ v.1.53k(Wayne Rasband and contributors, National Institutes of Health, United States).

Standard SusD pull-down assay
SusD proteins were tested with natural (MC; Sigma-Aldrich, Germany) and synthetic polymers (PET with and without 30 days of UV-C treatment; polyamide/nylon 6 (PA6) and low-density polyethylene (LDPE) powder (GoodFellow, England)).PET and LDPE powder comprise approximately 50% of crystallinity.Final concentrations of 0.4 mg/mL (6.18 µM) and 1 mg/mL (14.86 µM) of SusD1 and SusD38489, respectively, were diluted in potassium phosphate buffer 0.1 M pH 7 to a final volume of 10 mL.Preliminary tests with 6.18 µM of SusD38489 were performed, and no putative adsorption was detected.Furthermore, SusD70111 had a low protein concentration, and only the crude cell extract could be successfully tested.
Each protein was incubated with 0.1 g of the polymer of interest for 1 hour at 22°C, with slow shaking.Similar to the protein purification, a Qiagen polypropylene column was used, in which the substrate could precipitate and form a bed volume.The column was washed twice with phosphate-buffered saline (PBS) 1 x (137 mM NaCl, 10 mM Na 2 HPO 4 (Carl Roth, Germany), 2.7 mM KCl, and 2 mM K 2 HPO 4 (ChemSolute, Germany), and the bound protein was eluted with Triton X-100 2% (vol/vol) (Gerbu, Germany).The quantifications with Bradford assay (Roti Quant, Karl Roth, Germany) or simply by measuring the fractions with the NanoPhotometer NP80 (Implem, Munich, Germany) did not work since the detergent alone interacted with the Bradford reagent and with the NanoDrop device.For this reason, the fractions were analyzed via Western blot.

SusD binding assay with fluorescence readout
MC or chitin powder (0.02 g) was immersed in 20 µM of SusD::(GGGGS)::sfGFP diluted with potassium phosphate buffer 0.1 M pH 6 to a final volume of 200 µL in 1.5-mL microtubes.The mixture was incubated at 22°C for 1 hour at 300 rpm.Afterward, the substrate was vigorously washed twice with potassium phosphate buffer 0.1 M pH 6.The powder was resuspended in 200 µL of buffer and added to a 96-well polystyrene flat-bottomed plate, black-walled and clear bottom (Invitrogen, Thermo Fisher Scientific, United States).The negative controls included sfGFP and potassium phosphate buffer 0.1 M pH 6.The screening was performed on a PlateReader (Synergy H1 microplate reader, BioTek, Agilent Technologies, Santa Clara, United States) at the excitation wavelength 485 nm/emission wavelength 510 nm.
Similarly, amorphous foils of PET, PA6, and LDPE were cut with a hole punched to a diameter of 6 mm.The incubation was carried out under the same conditions, with the foil being further transferred to the plate.The measurements from top and bottom were collected, and they were selected in accordance with the substrate background (bottom for the natural polymers and top for the synthetic polymers).
Since we encountered the following limitations: 1) sfGFP presented putative adsorption to PET and 2) the control PET amorphous foil plus buffer presented a high background, the adsorption of the WT proteins toward this substrate could not be measured.However, the fluorescence intensity could be normalized by subtraction of SusD to each of the negative controls described previously.The values used for normalization represented the average of three independent fluorescence intensity measurements.

Binding kinetics and affinity of SusD to distinct substrates using surface plasmon resonance (SPR) spectroscopy
SPR assays were performed in a Biacore T200 using CM5 carboxymethyl dextran sensor chips (Cytiva, USA).The chips were previously coated with anti-His antibodies (Biacore His-capture kit, Cytiva, USA) so that the chip surface allows for complete regeneration of His 6 -tagged molecules from a sensor chip.First, the chips were equilibrated with HBS-EP buffer (10 mM HEPES [pH 7.4], 150 mM NaCl, 3 mM EDTA, 0.005% (vol/vol) detergent P20) until the dextran matrix had swollen.Afterward, the flow cells of each sensor chip were activated by injecting a 1:1 mixture of N-ethyl-N-(3-dimethylaminopropyl) carbodiimide hydrochloride and N-hydroxysuccinimide, using the standard amine-coupling protocol.All flow cells were loaded with a final concentration of 50 µg/mL of the anti-His 6 antibody in 10 mM acetate (pH 4.5) using a contact time of 420 seconds so that the surfaces contained antibody densities equivalent to ~8.500 response units (RU).Free binding sites on the flow cells were saturated by injection of 1 M ethanolamine/HCl (pH 8.0).During the course of each experiment, a reference curve was generated by the injection of the same buffer used to dilute the proteins (potassium phosphate buffer 0.1 M pH 7).To avoid bulk analysis, the experimental curves were subtracted by the reference curve.
The His 6 -tagged pure proteins SusD1, SusD38489, and SusD70111 (15 µg/mL each) were captured onto the chip in HBS-EP buffer for 600 seconds at a flow rate of 10 µL/min so that a final response of 600-1,800 RU was reached.The natural and synthetic polymers were injected over the chip using the single cycle kinetics protocol at a flow rate of 30 µL/min.Increasing concentrations (1 nM, 10 nM, 100 nM, 1.000 nM, and 10.000 nM) of CMC or BHET were sequentially injected on the flow cells without interim regeneration using a contact time of 180 seconds each and a final dissociation of 1,200 seconds.The chip was regenerated by injection of 10 mM glycine at pH 1.5 for 60 seconds at a flow rate of 30 µL/min all flow cells, which completely removed the SusD proteins from the chip.Furthermore, blank single-ycle kinetics were recorded by sequentially injecting the buffer instead of increasing concentrations of the polymers after capturing the SusD proteins.Each single-cycle kinetic was performed four times in a row at 25°C.Sensorgrams were recorded using the Biacore T200 Control software 3.2 and analyzed with the Biacore T200 Evaluation software 3.2.The surface of flow cell 1 was used to obtain blank sensorgrams for subtraction of bulk refractive index background.Buffer controls on the second, third, and fourth surface were subtracted from the sensorgrams obtained with the respective polymer to normalize drifts on the surface.The reference sensorgrams were then normalized to a baseline of 0. Peaks in the sensorgrams at the beginning and the end of the injections emerged from the runtime difference between the flow cells of each chip.R max was calculated measuring the maximal binding response for a 1:1 interaction and using the formula: Binding stoichiometry (n) was then calculated using the formula:

Bioinformatics
In order to identify SusD conserved domains, the amino acid sequences of SusD70111, SusD38489, and SusD1 were evaluated against the database CDD v.3.21 using default parameters (available at NCBI Conserved Domain Search (nih.gov),accessed on 24/04/2024).

Main characteristics of SusD38489, SusD70111, and SusD1
The complete protein sizes are in accordance with previous reports (61).SusD70111 has 588 amino acids (64.5 kDa), SusD38489 603 amino acids (67.3 kDa), and SusD1 580 amino acids (64.8 kDa).The N-terminal lipoprotein signal peptide Sec/SPII, which mediates the transport by the Sec translocon and cleaved by signal peptidase II (Lsp), was removed from each protein.The signal peptide removal did not affect the results of the assays.As expected, neighbor analysis revealed that each susD gene was positioned besides susC.Apart from SusD70111, glycoside hydrolases (GHs) of distinct families were identified downstream (Fig. 1).

Phylogenetic analysis of SusD38489, SusD70111, and SusD1
In a previous work, SusD from B. thetaiotaomicron and RagB from P. gingivalis were described to present structural similarity, which implies on a similar sugar-binding function (42).For this reason, six RagB protein sequences were included in the neighborjoining tree, which are displayed in the same cluster.
As expected, SusD1 and SusD38489 clustered with the protein sequences identified by BLAST, in which most of them belonged to herbivore gut metagenomes.Only three of the sequences identified by BLAST belonged to distinct niches (wastewater/food fermentation metagenomes).Intriguingly, SusD70111 clustered with the SusDs from hot springs (Fig. 2).
The "all against all" analysis performed with the Dali server retrieved a heat map and a dendrogram, displayed in Fig. S1 and S2.The proteins SusD1 and SusD38489 have a shared identity of 58.7% and are placed in a small subcluster, alongside the PDB structure 5E75 (64).SusD70111 displayed no close similarity to other structures and a shared identity of 28.7% and 29.1% to SusD1 and SusD38489, respectively.
Finally, a search using TPRpred (43) only predicted a TPR region for SusD70111, between the residues 70 and 104 in the wild-type.Altogether, the structural and phylogenetic data suggest that SusD70111 is the most conserved protein, while SusD1 and SusD38489 share structural and sequence similarity.

Cellulose-and chitin-binding studies using recombinant and tagged SusD proteins
Protein adsorption toward MC was addressed by quantitative screening based on fluorescence detection and qualitative pull-down assays.Heterologous expression of the SusD proteins, lacking the N-terminal signal peptide, was carried out in E. coli.Besides, each SusD had the C-terminus fused to sfGFP, generating a translational fusion protein.A His 6 -tag was positioned at the C-terminus of sfGFP.SusD1 and SusD38489 expression yielded relatively high concentrations of the protein (>100 mg per liter culture), while SusD70111 expression resulted in low protein concentrations (<10 mg per liter culture).
Pure proteins with final concentration of 20 µM were incubated with 0.01 g of the polymer, at the final volume of 200 µL.The binding tests implied that SusD1 is poten tially the protein with best activity because approximately 88.3% (in comparison to the amount of proteins lost in the flow through) could bind to MC (Fig. 3A; Fig. S3).While SusD38489 presented a similar behavior (86.9% of the protein could still bind to MC), SusD70111 bound poorly to MC.Because most of the protein was already lost in the flow through, the PlateReader detected an overflow for this fraction (Fig. S3).For this reason, an estimate percentage of SusD70111 protein retention could not be calculated.In the same incubation procedure, we observed that SusD70111 prefers chitin and, therefore, it might be characterized as a chitin-binding module (Fig. 3B).
Since the structural analysis showed multiple tryptophan residues involved in each potential binding site, we set out to mutate three tryptophan residues to verify this hypothesis.In the case of SusD38489, the tryptophan residues at the amino acid positions 258, 280, and 283 have been selected.The mutations were constructed in the backbone of the sfGFP translational fusion by exchanging each tryptophan to an alanine residue.The obtained mutants, with one (SusD38489Δ1-25 W258A ) or all three tryptophan residues exchanged (SusD38489Δ1-25 W258A,W280A,W283A ), were expressed, and the pure proteins were used for binding studies, as described above.SusD38489Δ1-25 W258A and SusD38489Δ1-25 W258A,W280A,W283A displayed 45.6% and 19.9% of binding activity toward MC in comparison to the wild-type (WT), respectively (Fig. 3A).For SusD38489Δ1-25 W258A,W280A,W283A , most of the protein was lost in the flow through (Fig. S3).
For the pull-down assays, there were no differences between each WT and their respective mutant lacking the N-terminal signal peptide.Since each protein had a His-6-tag positioned to the C-terminal, the binding was further assayed using Western blot and detection of the bound protein by monitoring the His-6-tag with a monoclo nal antibody.SusD1 bound MC with the concentration of 0.4 mg/mL (6.18 µM), while SusD38489 only bound with 1 mg/mL (14.86 µM) (Fig. 4A).Due to the reasons men tioned previously, SusD70111 tests only succeeded with the crude cell extract, where we estimated an amount of 3 mg/mL (46.5 µM) in 5 g of cells (obtained from each 1 L culture).
Other natural polymers, for instance, starch and xylan, could not be successfully tested due to the high background measurements obtained for sfGFP alone.Besides, FIG 2 Neighbor-joining tree built with the bootstrap method adjusted to 1000.Six RagB sequences (in pink) belonging to the oral pathogen P. gingivalis were selected based on their availability in the literature (33)(34)(35)(36)(37)(38).Six SusD-homologs from hot springs (in purple) were also included in the tree (32) and clustered with SusD70111.SusDs from gut, wastewater, or food fermentation metagenomes (29,62,63) were identified by BLAST using the NCBI non-redundant database.
These proteins are clustered alongside SusD1 and SusD38489.lichenan (Iceland moss) and lyophilized algae could not be tested in this method, as the powder absorbed most of the liquid content.

SusD proteins bind the synthetic polymers PET and PA6
SusD::His 6 -tag adsorption to PET was first verified with pull-down assays.Each protein was incubated with 0.02 g of PET powder, and the fractions representing the flow through, washing, and elution steps were collected and analyzed with Western blot (Fig. 4A).Putative adsorption was considered when a protein band could be detected after elution with the detergent, and the elution band was thicker than the second wash, meaning that the protein was not just being washed out from the column.Based on the Western blot analysis, it is suggested that SusD1 presented the best putative adsorption to PET, as the protein bands were still visible until the third elution.Besides, SusD1 binding to PET powder occurred at the lowest protein concentration of 6.18 µM.However, the weakest protein seems to be SusD38489 because even with the increased concentration of 14.86 µM, the protein band was fading and only detectable on the first elution.SusD1 and SusD38489 were also evaluated with other synthetic polymers, showing a putative binding to PA6 and, to a lower extent, to LDPE (Fig. 4B).When PET was incubated under UV-C light for 30 days, protein adsorption was avoided (Fig. S4).
The SusD70111 protein band was only visible when the crude cell extract was tested.Because SusD70111 was the only protein that presented two predicted binding sites with PET trimer, we believe that the interaction with PET is not stable (Fig. S5).A future approach would include the expression of SusD70111 in a Bacteroidotal system, to reject the hypothesis that the protein might be toxic to E. coli, yielding the observed low concentrations.
With the fluorescence analysis, SusD1 was normalized to 391.7 and 2064 fluorescence intensity (AU), when incubated with PA6 foil.SusD38489 presented normal ized measurements of 47 and 1719.2 fluorescence intensity (AU) to PA6.In general, PA6 had a lower sfGFP adsorption and lower background than PET.The LDPE foil had the lowest background, but the WT proteins were not measurable due to the high sfGFP background.Intriguingly, when compared to SusD38489 WT, the mutants SusD38489Δ1-25 W258A and SusD38489Δ1-25 W258A,W280A,W283A presented an increase of 16.3% and 212.7%, respectively, in the adsorption to LDPE.Furthermore, the mutants presented increased adsorption of 10.3% and 71% to PA6, as well as 4.8% and 55.8% to PET, respectively.

Binding kinetics of SusD1, SusD38489, and SusD70111 toward cellulose and BHET
To analyze the kinetics between each SusD and the polymers, SPR spectroscopy was performed.SusD70111 could not bind to the analytical chip and, therefore, it was not included.In this case, we used CMC, which is a variant of cellulose, due to its solubility in water.PET is soluble in DMSO, but it is not stable for long-term usage and incompatible with the microfluids in the SPR device.We selected BHET, which is a degradation product from PET (Fig. 5), soluble and stable in DMSO.Besides, it was previously reported that PET hydrolases commonly present BHET hydrolase activity as well, due to the structural similarity between the substrates (65).
The first step was to capture His 6 -tagged SusD1 and SusD38489 onto a CM5 sensor chip previously immobilized with anti-His antibodies.Increasing concentrations of CMC and BHET were injected over the chip in a single-cycle approach without the regen eration steps between the injections of the respective analyte concentrations.The sensorgrams were evaluated using the 1:1 binding algorithm.All SusD homologs bound cellulose as well as BHET with high affinity.SusD1 interacted with CMC with an associa tion rate of 5.2 × 10 3 /M*s (k a ) and a dissociation rate of 7.0 × 10 −5 /s (k d ), resulting in an overall affinity of K D = 14 nM (Fig. 6A).The association of BHET to SusD1 was higher (k a = 5.0×10 5 /M*s), while the dissociation rate was comparable (k d = 1.2×10 −4 /s), resulting in an overall affinity of 0.3 nM (Fig. 6B).However, the shapes of the sensorgrams reveal that the interaction of SusD1 to CMC is not truly 1:1.From the maximal response units (R max = 15), a stoichiometry of n = 6-8 can be concluded (calculated R max = 2.5), revealing more than one binding site for CMC in SusD1.Therefore, it might be possible that the different binding sites differ in affinity to each other so that the calculated binding kinetics and the affinity only reflect an average value from the sum of all binding events of CMC within the protein.Since the response units (RUs) increased after injecting the higher concentrations of the analyte, it can be concluded that SusD1 harbors a mixture of high and low affinity binding sites for CMC.The measured R max of BHET to SusD1 was comparable to the calculated R max , revealing a binding stoichiometry of 1:1.
The association rate of CMC to SusD38489 was comparable to that of SusD1 (k a = 5.0×10 3 /M*s); however, the dissociation rate was tenfold lower (k d = 6.3×10 −4 /s), resulting in a tenfold lower affinity (K D = 124 nM) due to the less stable interaction (Fig. 6C).The same sensorgram shape was also observed for the interaction of SusD38489 with BHET (k a = 1.2×10 5 /M*s; k d = 1.0×10 −3 /s), resulting in an affinity of 10 nM (Fig. 6D).The measured R max of CMC (7 RU) as well as that of BHET (1.5 RU) was lower compared to the SusD1, also revealing more than one binding sites (putatively 4-6) for CMC within SusD38489 and only one for BHET.Since the calculated R max of 6 could not be measured for the interaction with BHET, it is likely that not all protein molecules captured onto the chip bound the ligand, possibly due to inaccessibility or inactive or protein molecules.In summary, our data imply that both SusD1 and SusD38489 interacted with high affinity toward CMC and BHET.Due to the highest dissociation rates observed for SusD38489 and each ligand, the SusD1 interaction was more stable.Furthermore, multiple binding sites for CMC might be present in the SusD homologs and only one for BHET.We speculate that the binding kinetics as well as affinities could be like those of BHET for PET, which could not be measured in this approach due to technical prerequi sites of the SPR technique.In terms of protein concentration and binding stability, the SPR spectroscopy showed that SusD1 is the protein with best adsorption toward each substrate tested, which is also in accordance with the Western blot and fluorescence readings.

Modeling of the SusD 3D structures and structural analysis
To gain insight into the putative binding sites, all three SusD proteins were modeled using AlphaFold.We selected a total of six SusD-homologs from the PDB Database, which were co-crystallized with substrates and had their binding residues identified.These six proteins were SusD from B. thethaiotaomicron (pdb 3CKC; with maltotriose 3CKB) (66) (70).For structures resolved as a homo dimer, only the respective chain A was used for the comparative analysis to fit the modeled structures monomeric character.The binding areas of each protein model were compared in the structures with and without the bound substrate, and the binding residues and positions were taken from the respective publications.
When the structures were overlapped, it was possible to identify homologous residues in the loop-rich region of the protein.The putative binding residues were chosen based on the structural comparison to natural polymers.We highlighted the amino acids methionine, isoleucine, and valine due to their involvement in PET binding of previously described PETases (21,25).Eight putative binding residues were identified in each model and the structure model of SusD38489 is displayed in Fig. 7A.The amino acids residues predicted to be involved in substrate binding for SusD1 and SusD38489 were nearly identical, apart from R388 and R405 for SusD1 and SusD38489, respectively.Even though the residues are identified distinctly, they are structurally placed in the same position.The other putative binding residues were R43, W58, S61, N77, W258, W280 and W283.For SusD70111, this comparison resulted in the following residues identified as potential binding ones: H46, N66, N67, T295, W293, R310, W327, and Q419.Interestingly, at least two (SusD70111: W293 and W327) and up to four (SusD1 and SusD38489: W58, W258, W280, and W283) tryptophan residues located in the predicted binding areas were identified (Fig. 7B).
Surface color-coded hydrophobicity revealed that SusD1 and SusD38489's putative binding sites are relatively hydrophilic, with few hydrophobic residues in SusD38489 (Fig. 7C).However, the surface of the potential binding site of SusD70111 contains hydropho bic residues on its centre, while the hydrophilic and neutral ones are distributed on the upper surface.The degree of conservation implied that the inner regions of SusD1 and SusD38489 are scored average to conserved, while the outer portion is variable.Interestingly, the predicted binding site is conserved in both proteins, with the residues R43 and W258 being highly conserved (score of 9).SusD70111, on the other hand, presented slightly more variable residues in the helix and the only highly conserved putative binding residue (score of 9) is R310 (Fig. 7D).
According to the analysis performed with GeoMine (60), each protein presented two pockets (Table 3).Within the studies performed with AutoDock Vina (53), one pocket from SusD1 was structurally in the same position than one pocket from SusD38489.Furthermore, the following amino acids were identified: Y80, W83, I86, H88, Y146, and N147 (Fig. S6).Interestingly, the residue I86 was also found in SusD70111.
Recently, Sahihi and collaborators (71) provided relevant data on the molecular dynamics between a PETase and PET.It was reported that hydrophobic interactions were important for enzymatic adsorption toward PET (71).In our work, we observed a similar trend with SusD, revealed by the predicted presence of hydrophobic pockets within the proteins' surface.Because of the hydrophobicity, volume, and surface area, these pockets could be involved in the adsorption to small synthetic polymers (such as PET trimer; Table 3; Fig. S6).

DISCUSSION
Hitherto, there were no published reports with respect to plastic polymer binding by SusD proteins from the phylum Bacteroidota.Thus, we asked whether SusD-like proteins, which are known to bind a plethora of polymers in nature, were able to adhere to synthetic substrates.SusD-like proteins were previously described to bind a variety of substrates, which were in accordance with the niche occupied by the host.For instance, GMSusD from a marine Bacteroidota could recognize the β-glucan algae compounds laminarin and pustulan (12).
The SusDs characterized in this work were originally from herbivores' microbiota and, therefore, cellulose variants were expected to behave as the natural polymer.Furthermore, the fosmid Sc00044 (containing SusD1) displayed putative binding to CMC by activity screening on LB agar with 0.05% (wt/vol) of the substrate (27).The glyco side hydrolases from the cellulase family 5 and mannanase family 26 were identified adjacent to SusD1 and SusD38489, respectively (26,27).SusD70111, on the other hand, is predicted to be a chitin-binding protein, as it presented poor adsorption to MC but increased adsorption to chitin (Fig. 3A and B; Fig. 4A).
Bacteroidota and Firmicutes were abundant in the elephant feces metagenome samples, but Bacteroidota was responsible for cellulose degradation due to the large number of PULs found in the data set (26).Cellulose and chitin are the most abundant polysaccharides on earth and their main difference is that cellulose has β(1,4)-linked D-glucose chain, while chitin has β(1,4)-linked N-acetyl-D-glucosamine (13).Overall, considering the phylogenetic and structural differences elucidated here, SusD70111 seems to have evolved in a distinct manner than SusD1 and SusD38489.The predicted absence of TPR in SusD1 and SusD38489, which is a highly conserved amino acid signature believed to be responsible for the protein-protein interaction with SusC (2, 14, 15), makes for a strong case.
In a recent publication, CBM2 fused with sfGFP presented higher affinity to crystal line PET than amorphous foil (72).PET and LDPE powder used in this work presented 50% of crystallinity (data provided by the manufacturer GoodFellow, UK).The proteins bound following similar trends: while SusD38489 is probably the protein with weakest affinity toward PET (Fig. 4A), SusD1 and SusD38489 also displayed a putative weak binding toward LDPE (Fig. 4B).Besides, PA6 was also screened, and both proteins have a hypothetically good binding strength toward this polymer (Fig. 4D).When it comes to crystallinity changes, (73) described a shortening of PET chains after UV-C pretreatment, due to intrachain scissions.It is also believed that UV-C light acts on polymers with aromatic compounds (21).An increased surface crystallinity is proposed to affect the enzymatic efficiency toward this substrate (73), and physical properties such as color change from transparent to slightly yellow, as well as decreased hydrophobicity, have been reported (21,74).These changes might as well affect SusD adsorption, as the prior incubation of PET under UV-C light for 30 days resulted in the lack of initial binding between SusD and PET (Fig. S4).
In general, SusD-like proteins were previously described to present a flexible structure that accommodates the substrate after the preliminary binding (67).BtSusD has a flexible binding site that recognizes preferably cyclic maltooligosaccharide molecules due to 3D structure recognition, rather than the monosaccharide components (6,66).SPR analysis data show that SusD1 and SusD38489 bound steady to CMC and BHET, constituents of PET (Fig. 5 and 6).With one predicted binding site for BHET, it is suggested that this might be positioned within the pocket area (Table 3) comprising high hydrophobic ity.Regarding CMC, more than one binding site was predicted because the binding stoichiometry was not truly 1:1.This is in accordance with our in silico analysis using AutoDock Vina (53), where two to three binding sites were predicted for CMC and MC.The Gibbs free energy (ΔG) values were not significant and, for this reason, the in silico data are provided upon request.With the SPR analysis, SusD1 presented better adsorption to BHET and CMC than SusD38489 (Fig. 6).This is interesting because SusD1 was the protein with best performance toward distinct substrates, screened by fluorescence and pull-down assays.
In summary, other options should be considered for future research on this topic.For example, it is possible to speculate if SusD-like proteins would present an enhanced adsorption toward PET when incubated in complex with SusC.SusE/SusF are also binding proteins that form a complex (5) and could be considered for further experi ments and compared with SusD's strength.Furthermore, the α-amylase SusG carries a CBM58 inserted in domain B, which does not interfere with the catalytic domain (75,76).A knockout mutant lacking CBM58 was constructed, and the enzyme presented threefold increased activity toward the soluble substrate, but a decreased activity toward insoluble substrates (75).Therefore, further incubations of SusG and synthetic polymers could be performed.
The qualitative pull-down assays followed by Western blot can be considered for the preliminary discovery and characterization of new binding modules, as similar approaches were also performed with SusDs in previous literature reports, but not with synthetic polymers (10,68,77,78).In this work, we also showed that the adsorption to synthetic polymers was proportional to the amount of tryptophan residues replaced by alanine, in the putative binding site.For this reason, it was observed that the mutant SusD38489Δ1-25 W258A,W280A,W283A presented increased adsorption toward PET, PA6, and LDPE, which was a surprising finding.A previous work described a CBM with high affinity to the PET film due to the interaction with an exposed tryptophan triad and confirmed by tryptophan quenching (79).However, the trends seem to be distinct for PETases.Joo et al. performed docking studies of IsPETase, showing that the residue R280 was positioned in the binding pocket and responsible for stable adsorption to PET (80).However, when the residue was replaced by alanine (R280A), the enzyme presented enhanced activity (80).

Conclusion
The predicted structure and function of two cellulose-and one chitin-binding SusDhomologs from the phylum Bacteroidota were characterized.SusD70111 is substan tially distinct from SusD1 and SusD38489, as observed by structural and phylogenetic analyses.Binding of SusD toward natural and synthetic polymers was confirmed through pull-down assay, fluorescence measurements, and SPR.We ranked the proteins having the best to the worst putative adsorption toward distinct substrates: SusD1, SusD38489, and SusD70111.Surprisingly, SusD38489Δ1-25 W258A and specially SusD38489Δ1-25 W258A,W280A,W283A presented enhanced adsorption toward PET, PA6, and LDPE.To our knowledge, this is the first time SusD-like proteins are described to bind PET and other synthetic polymers, presenting a promising alternative for micro-and nanoplastic detection.

FIG 1
FIG 1 Operon map of SusD38489, SusD70111, and SusD1.SusD38489 and SusD70111 were derived from the elephant feces metagenome (26) and SusD1 from the cow rumen metagenome (27).The arrows indicate the open-reading frames (ORFs), and the distinct colors indicate the predicted function.Because of the shorter contig with a final size of 11 kb, glycoside hydrolase (GH) sequences were missing for SusD70111.The SusD38489 contig had 16 kb, while SusD1 was not defined.

FIG 3
FIG 3 Adsorption of SusD1, SusD38489, and SusD70111 measured by fluorescence analysis.(A) After discarding the flow through and washing twice, MC was resuspended in 200 µL of fresh potassium phosphate buffer 0.1 M pH 6 and transferred to a 96-well microtiter plate, with transparent flat bottom and black walls.After gently shaking the plates for 3 seconds, the fluorescence measurements were collected from the bottom at excitation wavelength of 485 nm and emission wavelength at 510 nm.The bars represent the mean value of the measurements collected in triplicates.Error bars represent the standard deviation.The incubations were carried out with 20 µM of protein.sfGFP or the buffer plus the substrate were the negative controls.SusD38489Δ1-25 W258A and SusD38489Δ1-25 W258A,W280A,W283A are identified in the graph as W258A and 3xWA, respectively.(B) Following the same procedure described prevviously, 20 µM of protein was incubated with 0.01 g of chitin.Chitin powder was resuspended in 200 µL of fresh buffer and transferred to a microtiter plate.The negative controls were sfGFP or buffer plus chitin.

TABLE 1
SusD proteins and their origin