Genomic sequencing-based mutational enrichment analysis identifies motility genes in a genetically intractable gut microbe

Abstract

A major roadblock to understanding how microbes in the gastrointestinal tract colonize and influence the physiology of their hosts is our inability to genetically manipulate new bacterial species and experimentally assess the function of their genes. We describe the application of population-based genomic sequencing after chemical mutagenesis to map bacterial genes responsible for motility in Exiguobacterium acetylicum, a representative intestinal Firmicutes bacterium that is intractable to molecular genetic manipulation. We derived strong associations between mutations in 57 E. acetylicum genes and impaired motility. Surprisingly, less than half of these genes were annotated as motility-related based on sequence homologies. We confirmed the genetic link between individual mutations and loss of motility for several of these genes by performing a large-scale analysis of spontaneous suppressor mutations. In the process, we reannotated genes belonging to a broad family of diguanylate cyclases and phosphodiesterases to highlight their specific role in motility and assigned functions to uncharacterized genes. Furthermore, we generated isogenic strains that allowed us to establish that Exiguobacterium motility is important for the colonization of its vertebrate host. These results indicate that genetic dissection of a complex trait, functional annotation of new genes, and the generation of mutant strains to define the role of genes in complex environments can be accomplished in bacteria without the development of species-specific molecular genetic tools.

Keywords: comparative genomics; gene annotation; motility; mutagenesis.

Fig. 1.

Identification of genes required for motility in E. acetylicum. (A) Schematic of the approach coupling chemical mutagenesis and DNA sequencing to perform a MEAPS and identify genes required for swarming motility. (B) Transmission electron micrograph of E. acetylicum flagella. (Scale bar: 0.5 μm.) (C) MEAPS of E. acetylicum strains defective for motility. E. acetylicum mutants were selected based on the loss of swarming (n = 440) or not selected (n = 700), and their genomes were sequenced. The normalized frequency of nonsynonymous mutations reveals two chromosomal regions (RI and RII) that preferentially accumulate mutations in nonmotile E. acetylicum strains. Red bars highlight predicted motility genes based on their similarities to motility genes in other bacteria. (D) Genetic map of ORFs in motility region I. Genes with homology to predicted motility genes (Dataset S1) are shown as blue arrows, and the number of nonsense and nonsynonymous mutations identified, after correction, are represented by gray and white squares, respectively. (E) Cartoon schematic of the Gram-positive flagellar apparatus displaying all components conserved in E. acetylicum. Components identified by MEAPS are shown in different font colors to reflect confidence of their relative association with motility (Dataset S2).

Fig. S1.

Simulations of the rate of identification of motility genes based on the sequencing of nonmotile mutants. Mutations in the E. acetylicum genome were generated in silico at random GC base pairs. The mutagenesis rates were altered to generate 5, 10, or 20 mutations per genome, with the assumption that only one mutation was responsible for the loss of motility. The number of motility genes was fixed at 100. The simulations indicate that, at an average of 10 mutations per genome, sequencing of 400–500 nonmotile mutants will lead to the identification of 70–75% of motility genes, based on their high frequency of accumulation of nonsynonymous mutations.

Fig. S2.

Isolation of nonmotile E. acetylicum mutants. (A) Chemically mutagenized bacteria were inoculated on 0.3% agar plates to enrich for nonmotile strains based on their inability to migrate away from the inoculation site. After three to four passages, individual clones were derived from bacteria that remained near the inoculation site and were individually tested in a 96-well plate format. Each clone was inoculated at the edge of the well and incubated for 24 h. Motile strains covered the entire well and appeared turbid. Strains impaired for motility remained near the edge. (B) Representative image of a 96-well plate used to screen for motility defects. Candidate nonmotile mutants are indicated with red arrows. (C) Secondary confirmation tests for the loss of motility. Saturated bacterial cultures were inoculated in the middle of 0.3% agar in 24-well plates and incubated for 24 h. The relative diameter of the bacterial colony was used as criteria to identify mutants with strong defects in swarming motility.

Fig. 2.

Functional characterization of motility genes in E. acetylicum. (A) Overlapping set of putative E. acetylicum motility genes identified by reciprocal BLAST homology searches, Pfam associated with motility, or MEAPS. (B) Suppressors of nonsense mutations in putative structural flagellar genes confirm their role in motility. Strains with nonsense mutations in flagellar components were passaged in soft agar to enrich for spontaneous motile variants. (B, Inset) Hag^Q222* and its Hag^Q222W suppressor. Sequence analysis indicated the presence of reversions and intragenic suppressor mutations that restored the reading frame. The relevant sequence of mutated and suppressed codons is shown. (C) Intragenic suppressor of a hag1 nonsense allele restores the formation of wild-type flagellar structures in E. acetylicum. (D) Motility enhances E. acetylicum colonization of germ-free zebrafish. Rifampin (Rif)-resistant and -sensitive versions of an E. acetylicum strain with a nonsense mutation in Hag1 (Hag1^Q222*) and its motile suppressor isogenic derivative (Hag1^Q222W) were placed in direct competition for colonization of 6 d postfertilization germ-free zebrafish embryos. Inoculum medium consisted of Rif-resistant strains mixed with sensitive strains at a 1:500 ratio. The relative frequency of each strain in the inoculum media at 6 d postfertilization, in association with animals immediately after colonization at 0 d postinoculation (0 dpi) and after 3 d of association (3 dpi) were determined by assessing the percentage of Rif-resistant bacteria. Error bars represent SD. Letters indicate P < 0.05 compared with respective inoculum medium (a) or 0-dpi larvae (b) using the Kruskal–Wallis test.

Fig. S3.

Comparative analysis of the distribution of synonymous mutations among E. acetylicum mutants. The location and abundance of synonymous mutations in E. acetylicum mutants that were selected or not selected based on motility was determined by sequencing the genomes of pools of mutants. The corrected frequency of synonymous mutation in nonmotile E. acetylicum mutants revealed no clear bias in mutations in any particular locus. Number of GC bases per gene (Bottom) indicates the likelihood of mutational biases based solely on targets for EMS mutagenesis. Note: The regions between 500 and 1,600 were omitted from analysis because this region contains rDNA and other repetitive regions that confound alignment of sequencing reads to a unique locus.

Fig. S4.

Comparative analysis of mutations in noncoding regions. Distribution of SNVs in noncoding regions of E. acetylicum mutants that had been selected based on their motility is shown. The frequency of mutation was normalized to the number of nucleotides in the noncoding region after correcting for mutation biases. In the regions exhibiting abnormally high frequencies of mutations, the nucleotide variants were mapped to predicted ribosome sites of cheY and flhA, respectively. The consequence of mutations in the intergenic region between rutD3 and ea3382 on motility is currently unknown because these genes do not appear to be directly involved in motility.

Fig. S5.

Motility contributes to the efficient colonization of zebrafish by E. acetylicum. Rifampin (Rif)-resistant and -sensitive versions of an E. acetylicum strain with a nonsense mutation in flagellar biosynthesis gene, flhA, (Flh^AQ393*), and wild-type strain were placed in direct competition for colonization of 6-dpf germ-free zebrafish embryos. Inoculum medium consisted of Rif-resistant strains mixed with sensitive strains at a 1:500 ratio. The relative frequency of each strain in the inoculum medium at 6 dpf, in association with animals immediately after colonization at 0 dpi and after 3 d of association (3 dpi) were determined by assessing the percentage of Rif-resistant bacteria. Error bars represent SD. Letters indicate P < 0.05 compared with respective inoculum medium (a) or 0-dpi larvae (b) using the Kruskal–Wallis test.

Fig. 3.

Motility region II of E. acetylicum encodes for new motility genes. (A) Nonsynonymous mutations in uncharacterized genes and close homologs of predicted motility genes (blue arrows) in the RII motility region are overrepresented among nonmotile E. acetylicum mutants. The number of independent nonsense and nonsynonymous mutations identified are represented by gray and white squares, respectively. (B) GGDEF/EAL domain proteins in E. acetylicum. Only genes with more than three total nonsynonymous mutations are shown. Pie charts indicate the proportion of total mutations identified among nonmotile (black) and unselected (gray) mutant pools. Squares on top and bottom of each gene represent the location of mutations identified in the selected (nonmotile) and unselected group, respectively. EAL, phosphodiesterase domain; GGDEF, diaguanylate cyclase domain; MYHT, bacterial signaling (Pfam03707) ); PalC, proteolytic processing (pfam08733); PAS, sensor for signal transduction; Serinc, serine incorporator (pfam03348. (C) Motility behavior and flagellar assembly of a FtsX^Q82* nonsense E. acetylicum mutant and a spontaneous variant that regained motility (FtsX^Q82W). TEM analysis of mutants and their suppressor indicated that FtsX is not required for flagellar assembly.

Fig. S6.

TEM images of FtsX^Q82* mutants.

Fig. 4.

Extragenic suppression analysis of nonmotile E. acetylicum mutants identifies a role for cell-wall modifications and c-di-GMP sensing in commensal Firmicutes motility. (A) Genetic suppressors of loss-of-function alleles in two novel motility genes reveal that loss of motility in mutants defective for ea2619 and ea2862 can be bypassed by changes in flagellar rotor switch control and chemotaxis. Motile variants of mutants bearing independent nonsense alleles of ea2619 or ea2862 were isolated. Common suppressor mutations (open circles) were mapped to chemotaxis genes (brown), rotor control genes (green), regulators of flagellar gene transcription (blue), and cell membrane homeostasis (orange). Thickness of lines connecting nodes is proportional to the number of independent suppressor mutations identified (Dataset S3). (B) Schematic of suppressor mutations linking Ea2619 and Ea2862 to the regulation of swimming speed and direction. Ea2157 indicate a direct link for the levels of c-di-GMP regulated by Ea2862 as central to the control of flagellar motility in E. acetylicum (red arrows). ea2157 was independently identified by MEAPS as a putative motility gene.