Figure 3. Identification of genetic determinants of fitness in vivo.

(A) The transposon mutant population is largely stable in vivo. The relative abundance of mutations in each gene (points) was compared between input and output (median from wild-type monoassociated mice, n = 15) populations. Genes that show a statistically significant change (q<0.05) in representation in all three cohorts of mice are shown in red, others in gray. The relative abundances of 80 gene-sized ‘neutral loci’ are shown in black (no significant change) and green (three ‘neutral loci’ that pass the significance criteria). (B) Individual strains retrieved from the archived collection demonstrate that genes critical for fitness in vivo are dispensable in vitro. Each strain was cultured individually in TYG medium and doubling time was calculated from OD₆₀₀ measurements. Mutants predicted by INSeq to have in vitro growth defects are marked with arrows. N/D, no growth detected. rnf, Na⁺-transporting NADH:ubiquinone oxidoreductase (BT0616-22); S-layer, putative S-layer locus (BT1953-7); nqr, Na⁺-translocating NADH-quinone reductase (BT1155-60); hly, hemolysin A (BT3459). (C) Insertions in the CPS4 locus (BT1338-58) show a consistent in vivo fitness defect. Individual insertion locations (open arrowheads) in a representative gene (BT1346; green arrow) are shown at top. Read counts for input (black) and output (orange) samples at each insertion location are indicated (output counts represent the median from the ceca of wild-type monoassociated mice, n = 15). Median output:input ratios for each gene (black/green arrows) across the CPS4 locus are shown below. Asterisks indicate the average FDR-corrected p-value (q) across 3 experimental cohorts (n = 5 mice/cohort): *q< 0.05; **q< 0.01.