(663f) Metabolic Modeling of Cystic Fibrosis Airway Communities Predicts Mechanisms of Pathogen Dominance

Cystic fibrosis (CF) is a genetic disease which results in excessive mucus production that reduces lung function. Approximately 80-95% of CF deaths are attributable to respiratory failure due to chronic airway infections and associated inflammation. The Cystic Fibrosis Foundation estimates that approximately 70,000 CF patients are living worldwide and about 1,000 new CF cases are diagnosed in the United States each year. With advent of culture independent techniques such as 16S rRNA gene amplicon library sequencing, sputum and bronchoscopy samples from CF patients can be analyzed systematically with respect to the diversity and abundance of bacterial taxa present. Numerous studies have shown that CF airway infections are rarely monomicrobial, but rather the CF lung harbors a complex community of bacteria that originate from the mouth, skin, intestine and the environment. While the identities and relative abundances of the genera present can be determined by 16S rRNA gene sequencing, different analysis techniques are required to understand the interactions between the multiple bacterial taxa and the CF lung environment, the role of the individual microbes in shaping community composition and behavior, and the impact of community composition on the efficacy of antibiotic treatment regimens.

In this paper, we utilized 16S rRNA gene amplicon library sequencing data from three published studies to develop a 17-species bacterial community model for predicting species abundances in CF airway communities. The 16S rRNA gene sequence data covers 75 distinct sputum samples from 46 adult CF patients, and captures the heterogeneity of CF polymicrobial infections with respect to taxonomic diversity and the prevalence of pathogens including Pseudomonas, Streptococcus, Burkholderia, Achromobacter and Enterobacteriaceae. The in silico community model was used to predict when each pathogen may dominate the polymicrobial infection by using the 16S rRNA gene sequence data to restrict which pathogens were present in the simulated community. By randomly varying the availability of host-derived nutrients, the model was used to simulate sample-by-sample heterogeneity of community compositions across patients and to understand how metabolite cross-feeding enhanced pathogen abundances. To our knowledge, this study represents the first attempt to metabolically model the CF airway bacterial community rather than model the individual metabolism of common CF pathogens.