Host DNA depletion on frozen human respiratory samples to enable shotgun metagenomic sequencing for microbiome studies

Titile: Host DNA depletion on frozen human respiratory samples to enable shotgun metagenomic sequencing for microbiome studies Abstract: Background: Most respiratory microbiome studies have focused on amplicon rather than metagenomics sequencing due high host DNA content. Therefore species-level taxonomic composition and predicted function of the respiratory microbiome is not well understood. Methods: We evaluated the efficacy of five host DNA depletion methods on previously frozen human nasal swab, sputum, and bronchoalveolar lavage (BAL) fluid. Methods included two previously published approaches (osmotic lysis followed by propidium monoazide treatment (lyPMA) or benzonase) and three commercial kits (Zymo HostZero, Molzym Molysis, Qiagen QIAamp). Untreated samples, host-depleted samples, mock community, and negative reagent only samples underwent ultra-low input library prep and metagenomics sequencing. Host depletion efficiency was evaluated based on sequencing failure rate, change in % host DNA proportion, increase in final non-human reads, change in species richness, presence of potential bias compared to the untreated community, and presence of contamination. Results: The median sequencing depth was 76.4 million reads per sample. Untreated nasal, sputum and BAL samples had a median 94.1%, 99.2% and 99.7% host reads. We observed significant differences between treatment methods based on sample type. Most of the methods depleted host DNA for nasal swabs and sputum. Only Host zero and Molysis significantly lowered the % host DNA of BAL although these two methods also had the highest sequencing failure rates overall. All treatment methods increased final microbial reads as well as species richness; the increase in species richness was due to a higher effective sequencing depth based on mediation analysis. Most successful methods also had significant changes in Morisita-Horn distances when comparing treated to untreated samples indicating a shift in overall microbial community composition; some gram negative bacteria were less abundant after host depletion. Weighing all criteria, Molysis for BAL, QIAamp for nasal swabs, and Host zero for sputum were selected as the optimal host depletion methods. Conclusions: Deeper sequencing alone cannot overcome the challenges of high host DNA content in respiratory microbiome studies. Host DNA depletion increases the effective sequencing depth thus increasing detected species richness in respiratory samples. The best approach differs based on sample type likely due to differences in the underlying sample matrix. Contact: Peggy S. Lai. This submission was powered by METAGENOTE (https://metagenote.niaid.nih.gov).