In silico identification of bacterial seaweed-degrading bioplastic producers

Abstract

There is an urgent need to replace petroleum-based plastic with bio-based and biodegradable alternatives. Polyhydroxyalkanoates (PHAs) are attractive prospective replacements that exhibit desirable mechanical properties and are recyclable and biodegradable in terrestrial and marine environments. However, the production costs today still limit the economic sustainability of the PHA industry. Seaweed cultivation represents an opportunity for carbon capture, while also supplying a sustainable photosynthetic feedstock for PHA production. We mined existing gene and protein databases to identify bacteria able to grow and produce PHAs using seaweed-derived carbohydrates as substrates. There were no significant relationships between the genes involved in the deconstruction of algae polysaccharides and PHA production, with poor to negative correlations and diffused clustering suggesting evolutionary compartmentalism. We identified 2 987 bacterial candidates spanning 40 taxonomic families predominantly within Alphaproteobacteria, Gammaproteobacteria and Burkholderiales with enriched seaweed-degrading capacity that also harbour PHA synthesis potential. These included highly promising candidates with specialist and generalist specificities, including Alteromonas, Aquisphaera, Azotobacter, Bacillus, Caulobacter, Cellvibrionaceae, Duganella, Janthinobacterium, Massilia, Oxalobacteraceae, Parvularcula, Pirellulaceae, Pseudomonas, Rhizobacter, Rhodanobacter, Simiduia, Sphingobium, Sphingomonadaceae, Sphingomonas, Stieleria, Vibrio and Xanthomonas. In this enriched subset, the family-level densities of genes targeting green macroalgae polysaccharides were considerably higher (n=231.6±68.5) than enzymes targeting brown (n=65.34±13.12) and red (n=30.5±10.72) polysaccharides. Within these organisms, an abundance of FabG genes was observed, suggesting that the fatty acid de novo synthesis pathway supplies (R)-3-hydroxyacyl-CoA or 3-hydroxybutyryl-CoA from core metabolic processes and is the predominant mechanism of PHA production in these organisms. Our results facilitate extending seaweed biomass valorization in the context of consolidated biorefining for the production of bioplastics.

Keywords: PHA; algae; bioplastic; genome; macroalgae; marine; polyhydroxyalkanoates; seaweed.

Fig. 1.

Phylogenetic distribution and density of CAZymes involved in the degradation of seaweed polysaccharides together with the genes involved in PHA synthesis, regulation and degradation. The tree and density values were constructed from 68 256 SPD enzymes identified within the CAZyme database and 231 611 PHA related genes identified within the NCBI protein database using ete3 [27]. Taxa identified from PHA genes not concomitantly identified within the CAZy database have been filtered. Seaweed-degrading density values are displayed as number of unique CAZy families. PHA-related gene density values within the heatmap bars are based on the number of occurrences of genes within each taxonomic family. The figure was prepared using the iTOL server [28].

Fig. 2.

Taxonomic distribution of PHA-related genes. Genomes with greater than or equal to two unique SPD enzymes per algal type are displayed (n=2987). Unknown taxa have been filtered for clarity. Data displayed as log₂ density values per taxonomic family for each gene encoded for within respective genomes.

Fig. 3.

Hierarchal edge bundling co-occurrence network. Densities of SPD- and PHA-related genes for 41 745 genomes containing PhaC, PhaC2 or PhaE. Vertex size and edge weight are proportional to log₂ normalized density values. Stochastic nested block model bundling is visualized with graph-tools [31]; blue squares indicate first-level hierarchical organization.

Fig. 4.

Pearson correlation for genomes with marine isolation sources (n=801). Circle size denotes sample size. Row and column colours delineate function; Green, green-targeting enzymes; red, red-targeting enzymes; brown, brown-targeting enzymes; grey, PHA-related genes.

Fig. 5.

Euclidean clustering of the PHA gene profile within genomes. Hierarchal clustering was not performed for SPD enzymes. Density values are constructed using the number of occurrences of a gene per genome. Genomes have been filtered to retain genomes containing PhaC, PhaC2 or PhaE and greater than or equal to three unique SPD genes per algal type (n=05). Singleton and unknown taxa have been filtered. Circular heatmap (R, B, G) denote the number of unique SPD genes per algal type for each genome.

Fig. 6.

Phylogenetic tree and genome architecture of genomes greater than or equal to the 70th percentile for genomes containing greater than or equal to three unique algae-degrading CAZy per algae type, thus representing potential candidates of biotechnological utility. Species or strains with more than one available genome have been filtered for clarity. Gene positions have been interspersed by 0.1% of the genome length for observability.