01% w/v arabinose for E. coli clones, both solidified with 12% gelatin (Oxoid, Adelaide, Australia). Colonies were grown at 25 °C for 5 days and then cooled at 4 °C for 3 h before checking for liquefaction by adding 3 μL of the 6 × gel loading dye (Fermentas Inc., Glen Burnie, MD) to each well. Evidence of liquefaction was established if the dye diffused rapidly (within 5 s) through the well and sank to the bottom. The Pseudoalteromonas tunicata D2 wild-type strain

(Holmström et al., 1998) and a genomic library of P. tunicata DNA, which was constructed by Burke et al. (2007) and which used the same fosmid vector and host strain as the metagenomic library described above, were used as positive controls. Cultures exhibiting activities on the solidified gelatin were subjected to a further assay EMD 1214063 in vivo using Azocoll, an insoluble, ground collagen, to which an azo-dye is attached. The assay was conducted in triplicates. Strains were grown for approximately 48 h at room temperature in MB and bacterial cells were harvested by centrifugation GPCR Compound Library at 8000 g for 10 min. Cell pellets were resuspended in the Azocoll substrate at a concentration of 5 × 108 CFU mL−1, supplemented

with a final concentration of 1 mM CaCl2. To prepare the substrate, 2.0 mg mL−1 of Azocoll (Sigma, St. Louis, MO) was washed twice using 0.01 M phosphate-buffered saline (pH 7.4) as described in Jiang et al. (2007). The tubes were incubated at room temperature with shaking at 90 r.p.m. for 24 h before centrifugation for 5 min to remove the undegraded Azocoll. Supernatants were taken for the measurement of OD520 nm. Escherichia coli Epi 300 pCC1FOS and Pseudomonas aeruginosa PAO1 strain were used as a negative and a positive control, respectively. The shotgun metagenome-sequencing data (92.6 Mbp of unique sequence) of the bacterial community associated with two C. concentrica specimen (BBAY04 and BBAY15) described in Thomas

et al. (2010) were searched for genes that were annotated as collagenase/matrix proteinase-related genes. Searches were performed on KEGG (Kanehisa & Goto, Cyclin-dependent kinase 3 2000), COG (Tatusov et al., 2003), Swiss-Prot (Boeckmann et al., 2003) and TIGRFAM (Haft et al., 2003) annotations using the keywords: ‘collagenase’, ‘Zn-dependent aminopeptidase’, ‘metalloproteinase’, ‘matrixin’ and ‘matrix proteinase’. The results were checked manually and matches that had an e-value lower than 1 × 10−20 in at least one annotation were regarded as putative collagenase protein sequences. In addition, collagenase-related proteins were retrieved from NCBI’s protein sequence database and the curated Swiss-Prot database (Boeckmann et al., 2003) using the keywords: ‘gelatinase’, ‘microbial collagenase’ and ‘matrix proteinase’ as well as proteins with the M9 peptidase and peptidase U32 conserved domains (which are domains in collagenases). Those database sequences were searched against the C. concentrica protein dataset with blastp (Altschul et al., 1990).