Cadherin Domains in the Polysaccharide-Degrading Marine Bacterium Saccharophagus degradans 2-40 Are Carbohydrate-Binding Modules

Abstract

The complex polysaccharide-degrading marine bacterium Saccharophagus degradans strain 2-40 produces putative proteins that contain numerous cadherin and cadherin-like domains involved in intercellular contact interactions. The current study reveals that both domain types exhibit reversible calcium-dependent binding to different complex polysaccharides which serve as growth substrates for the bacterium.

Saccharophagus degradans strain 2-40 is a Gram-negative, aerobic bacterium isolated from decaying salt marsh cord grass in the Chesapeake Bay watershed (1, 7). This bacterium can degrade at least 10 different complex polysaccharides, including agar, chitin, alginic acid, cellulose, β-glucan, laminarin, pectin, pullulan, starch, and xylan, and utilize them as sole carbon and energy sources (4, 5, 15). It has an extraordinary range of catabolic capabilities, and many of the enzymes exhibit unusual architectures, including novel combinations of catalytic and substrate-binding modules (17). The S. degradans genome (GenBank accession number NC_007912) encodes a set of 127 identifiable carbohydrate-binding modules (CBMs) and contains 52 cadherin (CA) and cadherin-like (CADG) domains in five large, secreted proteins (CabD/Sde_0798, Sde_1294, Sde_2834, Sde_3233, and CabC/Sde_3323) (6).

The cadherins constitute a large family of calcium-dependent cell adhesion proteins which, in higher organisms, play a major role in development and tissue morphogenesis (10, 16). At the molecular level, homotypic adhesion between cells arises from homophilic interactions between extracellular tandemly repeated cadherin domains (9, 10). Each of these domains, consisting of approximately 110 amino acids, forms a β-sandwich with Greek key folding topology (9). The cadherins have been broadly studied in the metazoan lineage (12), but in the prokaryotic world (8), their biological role has been poorly explored. In most cases, annotations of the deduced bacterial proteins that contain CA and CADG domains are not particularly informative, with most annotated as hypothetical proteins or proteins of unknown function.

Previously, we showed that CA and CADG domains are involved in protein-protein interactions and bind directly to bacterial cell surfaces, thus indicating a possible role in mediating cell-cell contact (6). Here, we report that CA and CADG doublet domains display carbohydrate-binding features in a calcium-dependent manner, thus suggesting an additional biological role of bacterial cadherins. This carbohydrate-binding function may be extremely important in aquatic environments where resources are dispersible and limited.

The CA and CADG doublet domains from CabC (Sde_3323) (amino acids 2288 to 2496) and CabD (Sde_0798) (amino acids 2261 to 2500), respectively, were cloned and expressed using standard techniques (13) as described earlier (6). Both proteins were examined for direct binding to the insoluble complex carbohydrates (ICPs) pectin, starch, agar, agarose, lichenan, cellulose, xylan from birch wood, xylan from oat spelt, and chitin (Fig. 1) according to previously described methodology (18). Briefly, the recombinant proteins, at an optimal concentration of 0.5 mg/ml, were mixed with ICPs (15 mg) suspended in a final volume of 0.2 ml of TBS-Ca buffer (50 mM Tris-HCl, 300 mM NaCl, 7 mM CaCl₂·2H₂O, pH 7.6). After a 20-min incubation period, the suspension was centrifuged and washed, and the contents of the pellet (bound) and supernatant (unbound) fractions were examined by SDS-PAGE. As shown by the results in Fig. 1, both the CA and CADG doublet domains bound to most of the ICPs listed above in a calcium-dependent manner. The proteins failed to bind to the ICPs in the presence of 10 mM EDTA. Intriguingly, the binding reaction was found to be reversible. All bound CA and CADG could be removed from the ICPs by EDTA (data not shown).

FIG. 1.

Interactions of CA and CADG doublet domains with different complex carbohydrates. The CA and CADG doublet domains were mixed with the indicated ICPs in the presence of 7 mM CaCl₂ or 10 mM EDTA as indicated. Samples were separated into supernatant (−) and pellet (+) fractions, which were subjected to SDS-PAGE.

The binding capacities of the CA and CADG doublet domains for the different ICPs were quantified by Bradford assay using bovine serum albumin as a standard (Fig. 2). The CA and CADG doublet domains bound at similar levels to chitin and agar (4.5 to 3.6 mg/g carbohydrate), as well as cellulose and agarose (2.7 to 2.2 mg/g carbohydrate). However, their binding capacities to other complex carbohydrates were different. The binding capacities of the CA doublet domain were highest for lichenan and the xylans (4.6 to 3.7 mg/g carbohydrate), less for pectin (2.0 mg/g carbohydrate), and lowest for starch (0.8 mg/g carbohydrate). In contrast, the binding capacity of the CADG doublet module was highest for pectin (4.2 mg/g carbohydrate) and lower for starch, lichenan, and the xylans (2.7 to 1.3 mg/g carbohydrate). According to the results of kinetic experiments, all binding reactions reached saturation by 40 to 50 min, indicating a half-life (t_1/2) of 20 to 25 min (data not shown).

FIG. 2.

Binding capacities of CA and CADG doublet domains for different complex carbohydrates. The concentration of each protein in the unbound fraction was determined, and binding capacities for the different ICPs were calculated for CA and CADG separately. Error bars show standard deviations.

The kinetics of the interactions between the CA and CADG doublet domains and the ICPs exhibiting the highest binding capacities were further investigated by determining the distribution of bound and free fractions and analyzing the data by Scatchard plots (Fig. 3). Incremental concentrations of the selected proteins were incubated with 15 mg of the appropriate polysaccharide. The suspensions were centrifuged, and the pellets washed. The bound fractions were subjected to SDS-PAGE, and the protein concentrations in the unbound fractions at the end of the incubation period were estimated using Bradford reagent. The results were linearized using Scatchard plots (14), whereby the X intercept indicated the maximum specific binding (B_max) of CA/CADG to each of the complex carbohydrates, and the slope was used to determine the apparent dissociation constants (K_d). The respective binding parameters for the given proteins are shown in Table 1. The K_d values ranged from 0.57 to 1.74 μM, and the B_max values ranged from 12.7 to 17.6 μM. These values would indicate that the CA and CADG doublet domains exhibit only moderate binding affinities, according to an early classification scheme (11). These relatively weak interactions may be compensated for by avidity resulting from multivalent interactions, when clustered carbohydrate-binding sites interact simultaneously with the ligand. We suggest that tandem repeats of CA and CADG domains in the five S. degradans proteins may thus participate in multivalent interactions with different ICPs.

FIG. 3.

Scatchard plot analysis of CA and CADG binding to selected complex carbohydrates. Incremental concentrations of CA and CADG doublet domains were mixed with the indicated ICPs. Samples were separated into supernatant (free) and pellet (bound) fractions, and the amount of protein was determined using the Bradford method.

TABLE 1.

Binding parameters for the interaction of CA and CADG with selected ICPs

Protein and substrate	Value (μM [mean± SD]) for parameter
	Kd	Bmax
CA
Chitin	1.74 ± 0.1	16.7 ± 2.3
Xylan (from birch wood)	1.37 ± 0.15	14.6 ± 1.8
Lichenan	1.45 ± 0.12	17.6 ± 2.5
CADG
Chitin	1.08 ± 0.1	15 ± 1.5
Pectin	0.57 ± 0.11	12.7 ± 1.5

CBMs are defined as contiguous amino acid sequences with discrete folds within the modular structures of carbohydrate active enzymes and cellulosomal scafoldins (2, 11). The explosion in genome sequence data has resulted in the detection by bioinformatics means of CBM family members that are not directly appended to enzymatically active modules (2). The wealth of CBM-encoding genes in S. degradans dramatically expands the number of modules classified in family 6 (by 39 members) and includes many novel combinations of CBMs, including those attached to domains of unknown function (17). Here, we propose that the CA and CADG domains may contribute to an additional expansion in the number of CBMs in S. degradans, due to their direct carbohydrate-binding features as demonstrated in this work. The CA and CADG doublet domains in S. degradans bind to different complex carbohydrates and, therefore, by definition act as distinct CBMs. Our experiments strongly support the previous suggestion that the bacterial cadherin domains may have carbohydrate-binding ability (3).

In conclusion, S. degradans cadherin and cadherin-like domains are located on five exported large proteins (133.2 to 828.6 kDa) that bind to several different complex carbohydrates which are abundant in aquatic environments. We propose that these large proteins, with more than 52 cadherin and cadherin-like domains, are involved in cell-cell contact by mediating cell aggregation and/or chain formation (6). Moreover, these domains may also be involved in the critical function of binding (and consequent targeting of the bacterium) to the different insoluble carbohydrates which serve as substrates for S. degradans.