The impact of motility on the localization of Lactobacillus agilis in the murine gastrointestinal tract

Background While the overall composition of the mammalian gut microbiota has been intensively studied, the characteristics and ecologies of individual gut species are incompletely understood. Lactobacilli are considered beneficial commensals in the gastrointestinal mucosa and are relatively well-studied except for the uncommon species which exhibit motility. In this study, we evaluate the importance of motility on gut colonization by comparing motile and non-motile strains of Lactobacillus agilis in mice models. Results A flagellated but non-motile L. agilis strain was constructed by mutation of the motB gene. Colonization of the wild type and the mutant strain was assessed in both antibiotic-treated female Balb/c mice and gnotobiotic mice. The results suggest that the motile strain is better able to persist and/or localize in the gut mucosa. Chemotaxis assays indicated that the motile L. agilis strain is attracted by mucin, which is a major component of the intestinal mucus layer in animal guts. Conclusions Motility and chemotactic ability likely confer advantages in gut colonization to L. agilis. These findings suggest that the motile lactobacilli have unique ecologies compared to non-motile commensals of the lactic acid bacteria. Electronic supplementary material The online version of this article (10.1186/s12866-018-1219-3) contains supplementary material, which is available to authorized users.


Background
Next generation sequencing technology has unveiled the diverse nature of the gut microbiota [1][2][3]. Albeit recent intensive studies reported key functions of these complex microbial communities [4][5][6], the ecology and roles of individual microbial species in the gastrointestinal (GI) tract have not been elucidated in detail. Lactic acid bacteria are culturable and beneficial microorganisms residing in animal guts, and thus their ecologies are relatively well-studied [7,8]. Most of those lactic acid bacteria are non-motile, but a few members of the lactobacilli possess flagella and exhibit motility [9][10][11][12]. It is obvious that motility is not essential for gut colonization, which raises the question of why the energy-consuming machinery is maintained, while in most other members it has been lost during evolution. The most likely explanation is that motility provides certain advantages on survivability and persistence for these organisms in the gut mucosa. Hence, in this study we hypothesize that the motility of these lactobacilli strains contributes to colonization in the gastrointestinal tract.
Lactobacillus ruminis and Lactobacillus agilis are motile lactobacilli isolated from the GI tract of mammals [13][14][15]. Since established genetic tools are available [16], the latter one seems to be less difficult to use as a model microbe for analysis. In the current study, we have been able to construct a non-motile derivative strain from L. agilis BKN88, a highly motile strain [17]. This mutant is flagellated but lacks motility due to malfunction of a motor-switch protein. In two different murine models and in vitro assays, the colonization, localization, and chemotactic abilities of the motile and non-motile L. agilis strains were compared.

Results
Construction and validation of a motB (D23A) mutant of L. agilis A consistently motile isolate of L. agilis JCM1048, originally isolated from a chicken, was obtained previously and designated BKN88 (Additional file 1). A non-motile derivative of L. agilis BKN88, was constructed through replacement of the wild type motB gene with a mutant motB (Fig. 1a). The resulting mutation, obtained by conversion of a single amino acid residue (23rd Asp to Ala), resulted in malfunction of MotB, the motor switch protein of the flagellar machinery. After sequence analysis confirmed the mutation of motB (D23A), the motility of the mutant strain, BKN134, was assessed in soft-agar culture. As shown in Fig. 1b, the mutant L. agilis strain exhibited no motility. This flagellar malfunction in BKN134 was also observed using optical microscopy (Additional file 2). The flagella were fully equipped in the mutant strain, and no structural difference was recognized between the wild type and the mutant strain (Fig. 1c).

Antibiotic-assisted colonization of the L. agilis stains in mice
In a preliminary experiment, mice without antibiotic-treatment received L. agilis via the intragastric route; however, all L. agilis cells passed through the gastrointestinal tract within 2 days. For further experimentation, streptomycin resistant derivatives of BKN88 (BKN136) and BKN134 (BKN141) were isolated after growth on agar plates with 100 μg/ml of streptomycin. The streptomycin-resistant L. agilis strains were able to colonize for a reasonable duration in antibiotic-treated mice. Both strains were predominant in the first few days and then gradually decreased in number in mouse feces. As shown in Fig. 2, significantly higher numbers of motile L. agilis colonies were detected in comparison with the mutant for several of the data points. After removing antibiotics, the mice started shedding both L. agilis strains and eliminated all of the streptomycin resistant bacteria in a month. The motile or non-motile phenotype of the cells in the recovered colonies did not change throughout the experiment.

Colonization of the L. agilis strains in Gnotobiotic mice
Germ-free mice were administered the L. agilis strains and housed in isolators for a month. Fecal samples collected weekly stably maintained 10 10 cfu/g of the lactobacilli throughout the experiment (Fig. 3a). No difference in numbers between motile and non-motile strains was found. After euthanasia, samples of the small intestines and ceca were collected. Total RNA was isolated from cecal contents, and the expression of the genes in the motility operon of L. agilis was confirmed by RT-PCR. As shown in Fig. 3b, both motA and fliC2 were expressed in the murine gut. Numbers of lactobacilli in the local mucosa of small intestine samples were counted using roughly fractionated lavage fluids. Overall, the bacterial cells were more predominant in the upper gastrointestinal tract than the lower, and more predominant in the luminal fractions than mucus fractions (Fig. 3c). Significant differences were found only in the mucus fractions of the ilea. In this case, the occurrence of the motile strain was approximately 1-log higher than in the non-motile strain.

Chemotaxis and penetration of the L. agilis strains in simulated mucus
Mucin is a major component of the gastrointestinal mucus layer. A capillary assay was used to determine whether the motile L. agilis would exhibit chemotactic ability toward mucin. As shown in Fig. 4a, L. agilis BKN88 was clearly attracted by mucin. Penetration through the simulated mucus layer by the Lactobacillus strains was also assessed. While small numbers of the non-motile strain passively passed through the layer, the motile strain could penetrate the layer to a much greater extent (Fig. 4b).

Discussion
The motility of flagellated enteropathogenic bacteria has been intensively investigated. These pathogens use the function to penetrate through mucus layers and invade host cells [18][19][20]. In return, the immune cells of the host recognize flagellar proteins via specific receptors such as TLR5 and NLRC4/IPAF to elicit innate immune responses [21][22][23][24][25]. Our previous work indicates that flagellins of L. agilis exhibit much lower immunological activity than those of major pathogenic bacteria [17]. This result implies that a host allows such bacteria to colonize as commensals. In the present study, we suggest that L. agilis takes advantage of its lower immunological activity in colonizing and/or localizing in the gut mucosa.
As described above, bacterial flagella possess at least two different functions, motility and immune-stimulating activity, Hence, a flagellated but non-motile strain was required to evaluate the exclusive impact of motility on colonization of the gastrointestinal mucosa of the host. A single amino acid mutation of the MotB protein confers a non-motile phenotype in L. agilis without loss of the flagella, as in a previous report in L. monocytogenes [26]. The currently constructed mutant seems to be an optimal strain to test our hypothesis.
Under antibiotic pressure, the streptomycin-resistant L. agilis strains could colonize the murine gut most likely because the bacteriocidal reagent substantially eliminated other competitive microbes. After discontinuing feeding of the antibiotic, L. agilis numbers reduced over time and eventually disappeared. In general, the motile L. agilis strain exhibited higher persistence in the gut than the non-motile mutant overtime. Meanwhile, similar amounts of bacterial cells were recovered regardless of motility in gnotobiotic mice, despite the fact that the flagella-associated genes were expressed in vivo. These dissimilar results among the two murine models seem conflicting, but might suggest that motility confers advantages on colonization only in case where the lactobacilli were surrounded by competitors. Albeit the total amount of lactobacilli in feces was similar in the gnotobiotic mice, the motile strain was detected in the mucosal/epithelial layer more frequently than the non-motile strain. Other experiments in vitro showed that only the motile L. agilis strain was attracted to mucin and had the ability to penetrate the mucus layer. Taken together, these results could support a hypothesis that the motile L. agilis cells actively localize in the middle of the mucus layer of the gut for robust colonization. Recent studies found that some gut microbes utilize mucin as a scaffold for cell-adhesion and/or as a carbon source [27][28][29][30][31]. Most lactobacilli in animal guts are also understood to utilize adhesins to attach to the local mucosa [32][33][34][35]. In contrast, this study suggests that L. agilis likely utilizes motility instead of or in addition to adhesion factors for its colonization.
We are aware of criticisms that the motile L. agilis strain recruited in this study is not a natural member of the gut microbiota of mice. Unfortunately, no motile lactobacilli have been isolated from mice or other rodents to the best of our knowledge. Thus, further studies need to be done in more appropriate animal models which include natural host-microbe combinations. Nevertheless, this study provides new and noteworthy

Conclusions
We assessed the impact of motility on the colonization of L. agilis in the gastrointestinal mucosa in murine models. The results suggest that the bacteria could take advantage of motility to establish a niche which is likely distinct from other non-motile lactic acid bacteria. This study reveals an unexplored ecological feature of certain motile lactobacilli residing in animal guts.

Bacterial strains and growth conditions
Lactobacillus agilis BKN88 [17] and derivatives were grown statically (liquid culture) or anaerobically (plate culture) using AnaeroPouch-Anaero Anerobic Gas Generators (Mitsubishi Gas Chemical) in MRS broth/agar (Difco/BD) with or without 5 μg/ml of erythromycin or 100 μg/ml streptomycin at 37°C. E. coli mc1061 was propagated aerobically in LB broth/agar with or without 200 μg/ml of erythromycin. All strains used in this study are described in Table 1. Motilities of Lactobacillus strains were determined by visual examination after inoculation into semi-solid MRS medium with 0.2% agar. Bacterial motility was also observed using an optical microscope (Keyence, Osaka, Japan).

Transmission electron microscopy (TEM)
Bacterial cells at exponential phase (OD 600 = 0.8) were collected from liquid culture in MRS-broth. The bacterial cells and the flagellar filaments were negatively stained and visualized using a transmission electron microscope (JEM1200EX, JEOL Ltd., Tokyo, Japan) at 80 kV. This experiment was done by Hanaichi Ultra-Structure Research Institute (Aichi, Japan).

Construction of a motB mutant of L. agilis
In Listeria monocytogenes, a single amino acid (23rd aspartate) replacement in MotB protein resulted in a non-motile phenotype without loss of the flagella structure [26]. L. agilis possesses an orthologous protein, and the specific amino acid residue is conserved. A DNA fragment containing mutant motB (D23A) and flanking region was generated by overlap PCR. Two separately amplified DNA fragments, an upstream fragment (Primer pair: DOKJ4, ATA TGG ATC CAG GAT TAT TAG CGC TAG AGG, and DOKJ7, AGG TCA TCA TAG CGG AGT AAG GTA GTA ACC) and a downstream fragment (Primer pair: DOKJ6, TAC TCC GCT ATG ATG ACC TTA CTA TTA TCC, and DOKJ5, ATA TGA ATT CAG CGG TAT CGT TAC TTG C), were assembled by subsequent PCR using DOKJ4 and DOKJ5. This PCR product was then digested with BamHI and EcoRI followed by insertion into pG + host5 [36] using E. coli mc1061 as a cloning host. The constructed plasmid, pG + host5::motB (D23A), was introduced into L. agilis BKN88 by electroporation in accordance with a protocol reported by Stephenson et al. [16]. A L. agilis isolate with the integrated plasmid at the target locus, BKN126, was selected. The integrated pG + host5 with wild type motB gene was then excised, and a non-motile isolate, BKN134, was selected. The replacement of motB sequence was confirmed by sequencing.

Colonization of L. agilis in antibiotic-treated mice
Mice were housed and cared for in accordance with the committee for the assessment of laboratory animal care standards and the guidelines of Tokyo University of Agriculture. To discriminate L. agilis from other gut microbes, naturally occurred streptomycin-resistant strains were isolated by plating L. agilis cultures onto MRS-agar containing 100 μg/ml of streptomycin. The antibiotic-resistant L. agilis were derived from either motile (BKN136) or non-motile (BKN141) strains. Female Balb/c mice were obtained from Crea Japan, Inc. Mice (four mice per group) were gavaged with 1 × 10 9 cfu of either motile or non-motile L. agilis strains. After the gavage, the mice received drinking water supplemented with 100 μg/ml of streptomycin. Fecal samples were collected twice a week and streptomycin resistant colonies were enumerated. Motility of randomly selected colonies (10 colonies/mouse) recovered from fecal samples was tested periodically. After a month, streptomycin was removed from the drinking water. Before starting this experiment, no streptomycin-resistant bacteria was detected from the feces.

Colonization of L. agilis in gnotobiotic mice
Care of gnotobiotic mice and collection of samples were operated by Sankyo Labo Service Co. (Tokyo, Japan). This experimental design was approved by the ethical committee of the company. Six weeks old female germ-free Balb/ c mice were administered with either L. agilis BKN88 or BKN134. Feces were collected once a week and the cfu of each strain was determined. Motility of isolated colonies was tested as described above. After a month, mice were euthanized to collect specimens: stomachs, jejuna, ilea, and ceca. Luminal contents of these tissues except for ceca were then washed with PBS followed by treatment with dithiothreitol (DTT, 1 mM)/ EDTA (5 mM)-supplemented PBS to collect epithelial/mucosal lavage fluids. After serial dilution, those were spread onto MRS-agar and incubated anaerobically until colonies appeared.  (TTT GCT TGA TGG  TCT TAG G) for motA, DOKJ51 (TTT CGG TAC AGG TGC A) and DOKJ52 (CTT TCT TGA TAG CAG C) for fliC2, were used respectively. Reverse transcription followed by PCR was performed with PrimeScript One

RNA-isolation from cecal contents and RT-PCR
Step RT-PCR Kit (Takara). In order to check DNA-contamination, PCR was also carried out with Takara Ex-taq. PCR-products were analyzed by 2% agarose-gel electrophoresis.

Chemotaxis assay
Chemotaxis of L. agilis BKN88 to mucin was tested as described by Worku et al. with minor modifications [37]. Glass microcapillary tubes of 10 μl capacity were filled with 1% mucin from porcine stomach (SIGMA-Aldrich) in chemotaxis buffer (0.1 M potassium phosphate, 0.1 M glucose, 0.5 M EDTA, in pure water) and then sealed at the upper end with plastic film. The capillaries were inserted into to 1.5 ml microtubes with bacterial cells at mid-log phase suspended in the chemotaxis buffer at a concentration of 10 6 cells/ml. After 1 h incubations, the outside of the capillary tubes was washed intensively with PBS followed by collection of the inner liquid. After serial dilution, the bacterial suspensions were spread onto MRS-agar plate for enumeration.

Penetration of simulated mucus layers
Simulated mucus layers were prepared as reported previously [38]. Briefly, 0.1 ml of simulated mucus, 0.5% melting agarose with 12.5% mucin from porcine stomach (SIGMA-Aldrich), was transferred into cell culture inserts (8.0 μm pore) in 24-well plates (Corning). After gelling, 0.2 ml of MRS-broth was overlaid on the mucus layer and 1.0 ml of bacterial suspension at mid-log phase (1.0 × 10 8 cells/ml) in MRS-broth was added to the well-plate before incubation at 37°C. At designated time points, 20 μl of liquid-phase was removed from the insert followed by dilution and plating on MRS-plates for enumeration.

Availability of data and materials
The datasets during and/or analyzed during the current study available from the corresponding author on reasonable request.
Authors' contributions AK was involved in all experiments and prepared the manuscript. SS and SI contributed to prepare and review the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate Animals were housed and cared for in accordance with the committee for the assessment of laboratory animal care standards and the guidelines of Tokyo University of Agriculture. This animal study was approved by the Ethics Committee of the Tokyo University of Agriculture. Care of gnotobiotic mice and collection of samples were operated by Sankyo Labo Service Co. (Tokyo, Japan). This experimental design was approved by the ethical committee of the company.