Novel Toxin Biosynthetic Gene Cluster in Harmful Algal Bloom-Causing Heteroscytonema crispum: Insights into the Origins of Paralytic Shellfish Toxins

Abstract

Caused by both eukaryotic dinoflagellates and prokaryotic cyanobacteria, harmful algal blooms are events of severe ecological, economic, and public health consequence, and their incidence has become more common of late. Despite coordinated research efforts to identify and characterize the genomes of harmful algal bloom-causing organisms, the genomic basis and evolutionary origins of paralytic shellfish toxins produced by harmful algal blooms remain at best incomplete. The paralytic shellfish toxin saxitoxin has an especially complex genomic architecture and enigmatic phylogenetic distribution, spanning dinoflagellates and multiple cyanobacterial genera. Using filtration and extraction techniques to target the desired cyanobacteria from nonaxenic culture, coupled with a combination of short- and long-read sequencing, we generated a reference-quality hybrid genome assembly for Heteroscytonema crispum UTEX LB 1556, a freshwater, paralytic shellfish toxin-producing cyanobacterium thought to have the largest known genome in its phylum. We report a complete, novel biosynthetic gene cluster for the paralytic shellfish toxin saxitoxin. Leveraging this biosynthetic gene cluster, we find support for the hypothesis that paralytic shellfish toxin production has appeared in divergent Cyanobacteria lineages through widespread and repeated horizontal gene transfer. This work demonstrates the utility of long-read sequencing and metagenomic assembly toward advancing our understanding of paralytic shellfish toxin biosynthetic gene cluster diversity and suggests a mechanism for the origin of paralytic shellfish toxin biosynthetic genes.

Keywords: biosynthetic gene cluster; cyanobacteria; horizontal gene transfer; saxitoxin; selection.

Fig. 1.

Identification of the cyanobacterial contigs in the H. crispum UTEX LB1556 assembly. Two blob plots that display the GC content of contigs (x axis) against their abundance (y axis) are shown. Left panel: the associated phylum-level annotations are labeled, indicated by the colored dots. Cyanobacterial contigs (lower middle of the plot, yellow) clearly resolve into a separate cluster. Right panel: the GC content of the same contigs is plotted, this time with the corresponding bin assignments labeled in color. Bin 11 (lower middle of the plot, brown) is observed in the same location as the cyanobacterial cluster from the left panel.

Fig. 2.

An illustration of the H. crispum saxitoxin biosynthetic gene cluster identified in this work. Coding genes are represented by arrows, and the direction indicates the gene orientation (e.g. a gene on the forward strand will point from left to right). Gene IDs are written directly on the gene where space permits; otherwise, the Gene ID is indicated by the color in the legend. The precise coordinates of the genes with respect to the H. crispum assembly are provided in supplementary table S2, Supplementary Material online.

Fig. 3.

The structure of the sxt gene cluster and patterns of HGT among cyanobacterial strains. a) Gene clusters with at least 20 unique BLAST hits to the H. crispum sxt genes are displayed. The phylogeny on the left displays evolutionary relationships of genomes based on the 16S rRNA gene, estimated with IQ-TREE. Vertical arrows between branches represent the top five most common transfer events between branches and are colored according to the number of inferred gene transfers. sxt genes are represented by horizontal arrows on the right and are colored according to their IDs as determined by protein similarity. Lines between gene clusters in closely related strains connect genes with at least 80% amino-acid identity. Only one representative for each of A. gracile and C. raciborskii is displayed. b) The estimated rate of HGT across the sxt cluster (bold vertical line) is significantly higher (P < 0.01) than the mean rate across 100 randomly sampled gene sets (distribution shown to the left of the sxt cluster line) of the same size as the sxt cluster.