Alpha-Agarases Define a New Family of Glycoside Hydrolases, Distinct from Beta-Agarase Families

Abstract

The gene encoding the α-agarase from “Alteromonas agarilytica” (proposed name) has been cloned and sequenced. The gene product (154 kDa) is unrelated to β-agarases and instead belongs to a new family of glycoside hydrolases (GH96). The α-agarase also displays a complex modularity, with the presence of five thrombospondin type 3 repeats and three carbohydrate-binding modules.

Agars are the main cell wall components of numerous red macroalgae. These polymers consist of 3,6-anhydro-l-galactoses and d-galactoses alternately linked by α-(1,3) and β-(1,4) linkages (10). Agars constitute a crucial carbon source for a number of marine bacteria which secrete agarolytic enzymes, mainly β-agarases (EC 3.2.1.81), which hydrolyze the β-(1,4) linkages (for a review, see reference 12). They are found in three distinct families of glycoside hydrolases (GH), families GH16, GH50, and GH86 (CAZY database [6]). Structural data are available only for the GH16 β-agarases from Zobellia galactanivorans, AgaA_Zg and AgaB_Zg (1, 2, 9). In contrast, “Alteromonas agarilytica” (proposed name) secretes an α- agarase (EC 3.2.1.158) which randomly hydrolyzes the α-(1,3) linkages in agars, releasing agarotetraose as its main end product (13, 15, 18). In this context, we sought to determine whether the agarases of the α type share any structural relationships with β-agarases. We report here the cloning of the α-agarase gene agaA from “A. agarilytica,” and we demonstrate that α-agarases define a new GH family.

Cloning of the agaA gene.

The α-agarase from “A. agarilytica” (AgaA_Aa) was purified as previously described (13), and N-terminal and internal peptide sequences were determined (Pasteur Institute, France) as follows: (i) ETLELQAESFANSGGFF and (ii) QPRVYNPNEHIVAEIQGPAT, respectively. With degenerated oligonucleotides derived from these microsequences (CARGCIGARTCITTYGCIAA and TAYAAYCCNAAYGARCAYAT, respectively), a 2.5-kb DNA fragment was amplified by PCR using “A. agarilytica” genomic DNA and labeled with [α-³²P]dCTP. This radiolabeled probe was used to screen an “A. agarilytica” genomic library, prepared as previously described (5). Among the ∼5,000 recombinant clones, two positive clones (pAA1 and pAA2) were identified, with inserts of 7.4 kb and 17.9 kb in length, respectively. Southern blotting analysis and plasmid mapping indicated that both inserts encompassed the same gene, which is present in only one copy in the genome. Plasmid pAA1 was sequenced on both strands over 4,651 bp, and a single open reading frame was identified and referred to as agaA (4,287 bp). Potential −35 and −10 promoter regions (TTGATC and TACACA, respectively) and a Shine-Dalgarno sequence (GGAG) were identified upstream of the start codon. A possible transcription termination codon was found downstream of the TAA stop codon. The deduced gene product is a preprotein of 1,429 residues (154 kDa) which includes the peptides A and B. A signal peptide cleaved between A26 and E27 is predicted by SIGNALP (4), consistent with the N-terminal sequencing of the purified extracellular α-agarase.

Several attempts were made to overexpress AgaA_Aa and its isolated modules with Escherichia coli, with pET or pGEX vectors, under various conditions. Unfortunately, the constructs always yielded inclusion bodies. However, within 1 week of culture at 22°C on Zd agar broth (3), the E. coli clones harboring the plasmid pAA1 dug a hole in the substratum, indicating agar degradation. Therefore, under the control of its own promoter, agaA was successfully translated into an active, recombinant enzyme, confirming that this gene indeed encodes the α-agarase.

AgaA_Aa is a complex, modular protein with a catalytic domain that defines a new GH family.

Only the N-terminal region of AgaA_Aa displays significant sequence similarity with proteins in the UniProt database. Based on InterProScan (14) analysis, eight distinct modules were identified in this region (Fig. 1A), including five thrombospondin type 3 (TSP3) repeats (TSP3-1 [amino acids D171 to G203], TSP3-2 [D360 to L392], TSP3-3 [D393 to A425], TSP3-4 [D426 to L458], and TSP3-5 [D459 to G491]) and three carbohydrate-binding modules from family 6 (CBM6-1 [amino acids E27 to R159], CBM6-2 [E209 to T343], and CBM6-3 [S659 to L792]).

FIG. 1.

(A) Modular architecture of the α-agarase AgaA_Aa. CBM6 and GH96 refer to carbohydrate-binding modules of family 6 and the glycoside hydrolase module of family 96, respectively. Gray boxes correspond to the TSP3 repeats. (B) Structure-based alignment of the TSP3 repeats of the human thrombospondin (PDB code 1UX6), of the α-agarase AgaA_Aa, and of the cellulase CelG from Pseudoalteromonas haloplanktis (trEMBL code O86099). (C) Structure-based alignment of the three CBM6s from the α-agarase AgaA_Aa and of the agar-specific CBM6s tethered to the β-agarases Aga16B_Sd (GenPept no., ABD80437) and Aga86E_Sd (GenPept no., ABD81915) from Saccharophagus degradans. These modules are compared to the secondary structures of the CBM6 appended to Aga16B_Sd (PDB code 2CDO). Alpha helices and beta strands are represented as helices and arrows, respectively, and beta turns are marked with TT. The black triangles mark the residues involved in the recognition of the nonreducing end of agarose chains. Figure 1B and C were prepared using ESPrit software (7) and use the same shading codes.

The characterized protein that most closely matches the TSP3 repeats of AgaA_Aa is the cellulase CelG from Pseudoalteromonas haloplanktis. In CelG, the TSP3 repeats constitute an extended linker connecting the GH5 catalytic module and a C-terminal CBM5 (17). The TSP3 repeats of AgaA_Aa present about 50% sequence identity with their counterparts in CelG. They also display ∼30% sequence identity with the “true” type 3 repeats found in human thrombospondin, whose crystal structure has been solved (11). These modules lack secondary structures and are organized around a core of calcium ions coordinated by conserved aspartates [DXDXDGXX(D/N)XXDXC motif]. The conserved cysteine is involved in a disulfide bridge linking adjacent TSP3 repeats, strengthening their stability (11). This motif is strictly conserved in each of the TSP3 repeats of AgaA_Aa (Fig. 1B), indicating that these modules adopt a similar structure and likely bind calcium ions. This is consistent with the observation that α-agarase activity is stabilized by the presence of calcium ions (13, 18).

In BLASTp searches with the three CBM6 modules from AgaA_Aa, the highest E values are always obtained for the CBM6 sequences tethered to β-agarases, while they decrease significantly with CBM6 linked to nonagarolytic enzymes. Only the CBM6s attached to the β-agarases from Saccharophagus degradans, Aga16B_Sd and Aga86E_Sd, were shown to actually bind agarose (8). A pair-wise comparison indicates a strong sequence identity (51%) between the CBM6-1 and the CBM6 from Aga16B_Sd (Fig. 1C), while those of CBM6-2 and CBM6-3 are more divergent (28% and 26%, respectively). The crystal structure of the CBM6 from Aga16B_Sd revealed that five residues are critical for the recognition of the nonreducing end of the agarose chain, Asn39, Tyr40, Trp97, Trp127, and Asn130 (8). Four of these residues are strictly conserved in CBM6-1, while Tyr40 is replaced by a similar aromatic amino acid, Phe64 (Fig. 1C). Together, these results strongly suggest that CBM6-1 is an agar-binding module and likely displays selectivity toward the nonreducing termini of agarose chains. In contrast, the five critical residues are only partially conserved in CBM6-2 and CBM6-3. Therefore, the specificities of these CBM6 sequences are less certain.

Finally, a BLASTp search of the patent data bank at NCBI identified two proteins from a marine bacterium with a strong sequence identity with the C-terminal region of AgaA_Aa (49% and 77%; Fig. 2). These proteins, also described as α-agarases (16), encompass three modules, two N-terminal CBM6s and the C-terminal module conserved with AgaA_Aa. Since the N-terminal region of AgaA_Aa encompasses only additional, noncatalytic modules, its conserved C-terminal region (Asn809 to His1429) likely contains its active site. Together, these three catalytic modules (∼620 residues) constitute a new family of glycoside hydrolases, referred to as family GH96 (CAZY database). Therefore, α-agarases are structurally unrelated to the β-agarases from the families GH16, GH50, and GH86.

FIG. 2.

Sequence alignment of the catalytic module of AgaA from A. agarilytica and of Aga14 and Aga15 (16). Shaded boxes enclose conserved positions. Figure 2 was prepared using ESPrit software and the same shading codes as seen in Fig. 1B and C.

Accession numbers.

The nucleotide sequence of the α-agarase gene from “Alteromonas agarilytica” has been deposited in GenBank with the accession number AF121273. Its amino acid sequence is available in Swiss-Prot under the accession number Q9LAP7.