Dayama et al., 2019. Interactions between the gut microbiome and host gene regulation in cystic fibrosis. BioRxiv.

This is a summary of our DalMUG journal club discussion of this paper written by Casey Jones.

Summary

Cystic fibrosis (CF) is a genetically inherited disease that causes dysfunction to gastrointestinal, respiratory, and reproductive systems. CF may be most known for its effects on respiration, but less recognized is that people with CF have an earlier onset of colorectal cancer (CRC) than those without CF - up to 30 years earlier compared to the relative population.

What’s more is that cancerous polyps develop at a younger age in those with CF & CRC, and progress faster too. Some evidence points to the causative gene of CF, the cystic fibrosis transmembrane conductance regulator (CFTR), playing a role in CFC pathogenesis. In the colon, CFTR is highly expressed in the stem cell compartment of the intestinal crypt, which is where CFC originates.

We read with interest a recent pre-print from Ran Blekman’s lab, first-authored by Gargi Dayama. Through analysis of mucosal biopsies, they were able to assess the mucosal microbiome and host gene expression of CF patients and non-CF patients undergoing routine colonoscopy CRC screening.

They found decreased microbial diversity, differential abundance of select bacterial taxa, and differential expression of 1543 genes in CF patients compared to controls. They then modeled associations between select taxa and host gene expression, hinting at associations with CRC.

See the main points below:

Points of Interest

• The authors found several differentially expressed human genes between CF patients and healthy individuals through analysis of RNA transcripts using DESeq2. One of the top genes was LCN2 that encodes for lipocalin 2, a protein that limits bacterial growth via siderophore sequestration. This provides one host-microbial link for lower bacterial abundance in CF patients.

• At different taxonomic ranks, the authors uncovered 51 OTUs that were differentially abundant between CF and healthy individuals.

• Using Spearman correlation, 50 host gene - microbe correlations were found. They highlight the strongest associations, including the negative correlation between Ruminococcaceae abundance and LCN2 transcript abundance in Figure 3C. There seemed to be a good separation of CF and healthy patient samples in these correlation plots.

Points of Confusion

• The authors removed two CF patient samples from downstream analysis since they had an alternate expression pattern of host genes compared to the rest.

• Why were Q-values thresholds not kept consistent between tests? (i.e. q-value < 0.05 for differentially expressed host genes, q-value < 0.1 for DE bacterial taxa (Figure 2B/C)

• We were confused by the taxa selection in Figure 2B. The supplementary table shows 6 OTUs at Faecalibacterium prausnitzii, but were not shown in the figure.