FliL Functions in Diverse Microbes to Negatively Modulate Motor Output via Its N-Terminal Region

ABSTRACT The flagellar motor protein FliL is conserved across many microbes, but its exact role has been obscured by varying fliL mutant phenotypes. We reanalyzed results from fliL studies and found they utilized alleles that differed in the amount of N- and C-terminal regions that were retained. Alleles that retain the N-terminal cytoplasmic and transmembrane helix (TM) regions in the absence of the C-terminal periplasmic domain result in loss of motility, while alleles that completely lack the N-terminal region, independent of the periplasmic domain, retain motility. We then tested this prediction in Helicobacter pylori fliL and found support for the idea. This analysis suggests that FliL function may be more conserved across bacteria than previously thought, that it is not essential for motility, and that the N-terminal region has the negative ability to regulate motor function.

The FliL extracytoplasmic domain forms a circle of rings, each coaxially sandwiched between MotA and the peptidoglycan-binding domain of MotB of a respective stator unit (9,11). Overall, FliL appears to be important for flagellar motility, but it has been difficult to assess its exact role and whether it plays similar roles in diverse microbes. Here, we provide insight into this issue by identifying that previous work compared different types of fliL mutants. Our analysis suggests that a key variation is inclusion or exclusion of the N-terminal cytoplasmic and transmembrane regions. We support our ideas with a direct test in Helicobacter pylori. Our findings suggest that FliL is not essential for flagellar motility, but instead that the FliL N-terminal region acts as a motility inhibitor when retained without the extracytoplasmic C-terminal region.
fliL mutant phenotypes are reported to vary between and even within the same bacterial species (4,12,13). To begin to understand these divergent phenotypes, we analyzed all published fliL mutants (Table S1). Our goal was to evaluate whether there were patterns to the type of fliL alleles and their phenotypes. We were not able to use all the reports, however, because in some cases the bacterial species had more than one flagellar system with each system encoding a FliL with unclear relations between them (10,11,14). Some reports suggested that the mutants were polar or otherwise non-complementable, or the reports lacked full motility data (Table S1). Our data set contained fliL mutants from eight species belonging to Alphaproteobacteria (Rhodobacter sphaeroides, Caulobacter crescentus [15]), Betaproteobacteria (Herminiimonas arsenicoxydans [16]), Gammaproteobacteria (Escherichia coli [4,7,12], Salmonella enterica serovar Typhimurium [7,12], Proteus mirabilis [4]), Campylobacterota (Helicobacter pylori [9]), and Firmicutes (Bacillus subtilis [15,17]) (Table S1). FliL proteins from these bacteria share a conserved secondary structure: a short, 2-28-residue N-terminal cytoplasmic region; an ;23-residue TM; a variable-length linker; and an ;200-residue C-terminal extracytoplasmic domain (Fig. 1). The conserved structure suggests that these FliL proteins perform similar functions.
We then analyzed the details of the different DfliL mutants. We found that there was significant variation in the length of the N-terminal and C-terminal regions retained in the DfliL mutants. At the N-terminus, there were three types of variations: full loss of the TM (DfliL1); retention of the cytoplasmic region plus part of the TM (DfliL2); or retention of the cytoplasmic region plus the whole TM (DfliL3) (Fig. 2). At the C-terminus, the variation could be classified as complete loss, retention of about half of the After classifying these types of fliL alleles, we then analyzed whether there were any patterns to fliL alleles that retained or lost motility on ;0.3% soft agar plates. One pattern immediately jumped out: alleles that deleted the entire TM (DfliL1) caused minimal motility defects, and in some cases even resulted in enhanced motility compared to wild type ( Fig. 2A). In contrast, alleles that retained all of the TM (DfliL3) showed severe motility defects ( Fig. 2A). Alleles with a partial TM (DfliL2) showed intermediate and variable motility phenotypes. In contrast to the N-terminal region, there was no obvious correlation with types of C-terminal mutations ( Fig. 2A). These results suggest that flagellar motors can function without FliL if it is fully removed, but that retention of partial N-terminal FliL sequences results in loss of soft-agar migration.
Loss of movement on 0.3% soft agar plates could be due to defects in motility, chemotaxis, and/or growth. We thus examined the studies of DfliL3 alleles to further explore the nature of the defects. Although the data are limited, motility was lost in the two analyzed for this ability, C. crescentus and H. pylori (8,9). One S. enterica species was found to retain motility in liquid, which was slowed but had normal switching (7) (Table S1). Overall, these results suggest that retention of the FliL N-terminal region results in loss or slowed motility but does not affect switching.
The DfliL mutant previously constructed in H. pylori was a DfliL3 allele and nonmotile (9) (Fig. 2A). We experimentally tested our hypothesis that the role of FliL in motility is associated with its TM by constructing an H. pylori DfliL1 mutant lacking the entire TM ( Fig. 2A). This mutant retained migration on soft agar (Fig. 2B). Indeed, the DfliL1 mutant in H. pylori showed even greater soft agar migration than wild type (Fig. 2B). Overall, these results support the idea that fliL is not required for motility.
We were curious about the observation that the DfliL1 allele showed elevated soft agar migration. Because FliL has been suggested to play a role in surface-associated responses, we examined fliL mutant phenotypes on soft-agar plates with high agar concentrations, between 0.5% and 1%, concentrations that support the surface-associated behavior called swarming (18). DfliL1 mutants in E. coli (4) and P. mirabilis (4) migrated to a greater extent than their wild types (WTs) on ;0.5% soft agar ( Fig. 2A). This response was similar to that of H. pylori (Fig. 2), suggesting the removal of fliL can result in motility that is more effective under elevated agar conditions. However, two other E. coli DfliL1 alleles (3,11), with deletions of the same regions as the allele above, were found to have high agar migration defects ( Fig. 2A), indicating other unknown factors are associated with the function of FliL under high agar conditions.
Our analysis suggests that loss of fliL has a more consistent phenotype on ;0.3% soft agar across microbes than previously expected. FliL is not needed for flagellar motor function under this condition, and indeed, it appears to negatively regulate motility via its N-terminal region, including the TM helix. Our work suggests that future fliL alleles should be made with care to exclude the TM region if seeking a null allele. The idea that FliL N-terminal region exerts negative motor control is new, and it is not yet clear how this might occur. One idea comes from the observation that motility defects were restored in DfliL3 mutants by extragenic suppressor mutations in the region of the motB gene corresponding to the plug (3,5). Given that FliL and MotB are close to each other in the motor and interact in vitro (3,9,19), we propose that the FliL N-terminal region interacts with MotB so as to prevent plug opening, block ion flow, and inhibit motility. In WT FliL, because we found that alleles that retained the N-terminal region without the C-terminal one resulted in loss of motility, we propose that the C-terminal domain may act to regulate the N-terminal region.
Although our analysis did not contain bacteria with dual flagellar systems, the inhibiting function of the FliL N-terminus on motor output has also been reported in V. alginolyticus (10,20), B. diazoefficiens (11), and V. fischeri (14). These studies show that DfliL1 alleles retain motility in both polar or lateral systems, while DfliL3 mutants show severe defects (Table S1). These combined results strongly support FliL is not an essential component of the flagellar motor, explaining why the fliL gene avoided detection by classical loss-offunction genetic analysis for a long time.
In addition to the differences in terms of what parts of fliL were deleted, we also noticed that fliL mutant phenotypes may be influenced by experimental conditions. For example, decreases in temperature alter the FliL soft-agar phenotype (4). In addition, the motility of Borrelia burgdorferi DfliL was tested by adding agarose instead of agar (6). These variations have not been systematically addressed, making it hard to compare results. Our improved understanding of fliL mutant construction will allow these variables to be studied.
FliL has been proposed to function in the surface-associated response in P. mirabilis (21), V. alginolyticus (10,20), and B. diazoefficiens (11). The only direct evidence examining the function of FliL on mechanosensing comes from studies in E. coli (22,23), which found there was no difference in force generation between WT and DfliL strains under external load. The two DfliL mutants (PL111 and PL62) used in these studies are fliL2 types of alleles, a type with variable phenotypes (Fig. 2; Table S1), suggesting this result needs to be reexamined. Instead, the greater motile phenotypes of DfliL1 on 0.5% agar plates suggest that FliL does have a function on motor output under viscosity. The stop or slow down of flagellar rotation in DfliL3 suggests that the role of FliL on high viscosity might be achieved by affecting torque production, with the activity of the MotB ion channel regulated by the C-terminal FliL domain, which is similar to the function of the domain found in FliL, SPFH, in eukaryotic stomatin (10,24,25).
The information about regions deleted in the fliL mutants and cognate phenotypes on ;0.3% and ;0.5% agar plates was collected from published articles by May 2022 as cited in Table S1. The sequence alignment was performed using PRALINE (https://www .ibi.vu.nl/programs/pralinewww/) and edited in Jalview. The secondary structure and transmembrane regions were predicted using YASPIN and TMHMM (http://www.cbs.dtu .dk/services/TMHMM/), respectively. The DfliL1 in H. pylori SS1 was constructed using natural transformation with a DfliL::aphA3 that retained first 40 bp and the last 32 bp of fliL (primers available upon request). H. pylori motility on soft agar plates was tested using plates composed of Brucella broth, 2.5% (vol/vol) heat-inactivated fetal bovine serum (FBS), and either 0.35% or 0.55% Bacto agar. These plates were inoculated with H. pylori from overnight Brucella broth/10% FBS cultures adjusted to an OD 600 of 0.15. Plates were incubated at 37°C, 10% CO 2 , 5% O 2 , 85% N 2 .

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TABLE S1, DOCX file, 0.1 MB.

ACKNOWLEDGMENTS
We thank Pushkar Lele (Texas A&M) for his thoughtful comments on this article. The described project was supported by the National Institute of Allergy and Infectious Diseases (NIAID) grant 1R01AI164682-01 to K.M.O., the Australian Research Council grant DP210103056 to A.R., and a student fellowship from the China Scholarship Council 201904910692 to X.L. The funders had no role in study design, data collection, and interpretation, or decision to submit the work for publication.