A targeted approach to enrich host-associated bacteria for metagenomic sequencing

Abstract

Multicellular eukaryotic organisms are hosts to communities of bacteria that reside on or inside their tissues. Often the eukaryotic members of the system contribute to high proportions of metagenomic sequencing reads, making it challenging to achieve sufficient sequencing depth to evaluate bacterial ecology. Stony corals are one such complex community; however, separation of bacterial from eukaryotic (primarily coral and algal symbiont) cells has so far not been successful. Using a combination of hybridization chain reaction fluorescence in situ hybridization and fluorescence activated cell sorting (HCR-FISH + FACS), we sorted two populations of bacteria from five genotypes of the coral Acropora loripes, targeting (i) Endozoicomonas spp, and (ii) all other bacteria. NovaSeq sequencing resulted in 67-91 M reads per sample, 55%-90% of which were identified as bacterial. Most reads were taxonomically assigned to the key coral-associated family, Endozoicomonadaceae, with Vibrionaceae also abundant. Endozoicomonadaceae were 5x more abundant in the 'Endozoicomonas' population, highlighting the success of the dual-labelling approach. This method effectively enriched coral samples for bacteria with <1% contamination from host and algal symbionts. The application of this method will allow researchers to decipher the functional potential of coral-associated bacteria. This method can also be adapted to accommodate other host-associated communities.

Keywords: FACS; FISH; coral; host-associated; hybridization chain reaction; metagenomics; symbiosis.

Figure 1.

Setup for A. loripes (left) and MC (right) samples where A was labelled with EUBMix338 and Endo probes using HCR-FISH, and B was an unlabelled negative control for HCR-FISH. Replicate samples for each coral genotype and MC were processed on three separate days. The MC comprised Endozoicomonas and one other cnidarian-associated bacterium (Ruegeria, Roseovarius, or Vibrio).

Figure 2.

FACS sorting profiles for unlabelled (A-E) and labelled (F-J) A. loripes tissue homogenate (genotype Al08). First, background noise (P1) from the FACS was excluded (A, F). Large coral and Symbiodiniaceae cells were then identified by forward (B, G) and side (C, H) scatter and excluded through sequential gating. Fluorescent channels were used to identify autofluorescence of the remaining events in the unlabelled fraction in the 582 nm (D-E x-axis) and 450 nm (D-E y-axis) emissions. Events in the labelled fraction that surpassed this baseline autofluorescence in both axes were collected as the dual labelled ‘Endozoicomonas’ population (P5) and those that had greater signal on the x-axis only were collected as the single labelled ‘all-bacteria’ population (P3) (I-J).

Figure 3.

A labelled A. loripes sample (colony Al13) prior to sorting with FACS (A–C) or sorted cell populations that were single (‘all-bacteria’, D–F), or dual (‘Endozoicomonas’, G–I) labelled. Each sample was visualized on a Nikon A1R confocal laser scanning microscope with channels for brightfield (A, D, G), 561 nm excitation (B, E, H), and 405 nm excitation (C, F, I). S=Symbiodiniaceae, Cn=host cnidocyst, H=uncharacterised host cell, B=bacteria. All scale bars are 25 µm.