Anti-σ²⁸ Factor FlgM Regulates Flagellin Gene Expression and Flagellar Polarity of Treponema denticola

ABSTRACT

FlgM, an antagonist of FliA (also known as σ²⁸), inhibits transcription of bacterial class 3 flagellar genes. It does so primarily through binding to free σ²⁸ to prevent it from forming a complex with core RNA polymerase. We recently identified an FliA homolog (FliA_Td) in the oral spirochete Treponema denticola; however, its antagonist FlgM remained uncharacterized. Herein, we provide several lines of evidence that TDE0201 functions as an antagonist of FliA_Td. TDE0201 is structurally similar to FlgM proteins, although its sequence is not conserved. Heterologous expression of TDE0201 in Escherichia coli inhibits its flagellin gene expression and motility. Biochemical and mutational analyses demonstrate that TDE0201 binds to FliA_Td and prevents it from binding to the σ²⁸-dependent promoter. Deletions of flgM genes typically enhance bacterial class 3 flagellar gene expression; however, deletion of TDE0201 has an opposite effect (e.g., the mutant has a reduced level of flagellins). Follow-up studies revealed that deletion of TDE0201 leads to FliA_Td turnover, which in turn impairs the expression of flagellin genes. Swimming plate, cell tracking, and cryo-electron tomography analyses further disclosed that deletion of TDE0201 impairs spirochete motility and alters flagellar number and polarity: i.e., instead of having bipolar flagella, the mutant has flagella only at one end of cells. Collectively, these results indicate that TDE0201 is a FlgM homolog but acts differently from its counterparts in other bacteria.

IMPORTANCE Spirochetes are a group of bacteria that cause several human diseases. A unique aspect of spirochetes is that they have bipolar periplasmic flagella (PFs), which bestow on the spirochetes a unique spiral shape and distinct swimming behaviors. While the structure and function of PFs have been extensively studied in spirochetes, the molecular mechanism that regulates the PFs’ morphogenesis and assembly is poorly understood. In this report, FlgM, an anti-σ²⁸ factor, is identified and functionally characterized in the oral spirochete Treponema denticola. Our results show that FlgM regulates the number and polarity of PFs via a unique mechanism. Identification of FliA and FlgM in T. denticola sets a benchmark to investigate their roles in other spirochetes.

INTRODUCTION

Bacterial flagella are complex molecular machines that are important for motility, biofilm formation, and pathogenesis. More than 30 different proteins are coordinately assembled to constitute three distinct parts of the flagella: the basal body (motor), hook (joint), and filament (propeller) (1–3). The basal body is embedded within the cell envelope and works as a reversible rotary motor (4–6). The hook acts as a flexible joint that connects the basal body and filament, a long helical structure composed of tens of thousands of subunits of a single protein called flagellin (7). Flagellin proteins are synthesized in the cytoplasm and exported through the hook-basal body (HBB) (8). To cope with sophisticated flagellar structure, most bacteria have evolved a complex transcriptional hierarchy to control their flagellar gene expression both temporally and spatially (1, 2). Within this regulatory hierarchy, the operons that encode flagellar, motility, and chemotaxis proteins can be divided into three transcriptional classes. Class 1 genes encode FlhC and FlhD, two master regulators that initiate the expression of class 2 genes, which mainly encode HBB structural proteins and two transcriptional regulators, FliA (also called σ²⁸) and FlgM (anti-σ²⁸). As a flagellum-specific alternative σ factor, FliA initiates the expression of class 3 genes, such as those encoding flagellin and chemotaxis proteins. As an antagonist of FliA, FlgM negatively regulates the expression of class 3 genes through direct interaction with FliA, preventing it from forming a complex with core RNA polymerase (RNAP) (9, 10). FlgM can also attack and destabilize the FliA-RNAP holoenzyme and inhibit transcription initiation (11). FlgM exerts its inhibitory role on flagellar gene expression through a partner-switching mechanism in response to HBB assembly and completion (12, 13). Specifically, FlgM and FliA remain bound as a complex in the cytoplasm until HBB assembly is completed, whereupon FlgM is exported out of cells, releasing FliA to initiate transcription of class 3 genes. This regulatory paradigm was first established in the two model organisms Escherichia coli and Salmonella enterica serovar Typhimurium (1, 2, 13). Later, a similar regulatory scheme was found in numerous motile bacteria, such as Campylobacter jejuni (14), Helicobacter pylori (15, 16), Pseudomonas aeruginosa (17), Vibrio cholerae (18), and Bacillus subtilis (13, 19). Deletions of fliA led to mutants deficient in flagellar filaments and nonmotile, highlighting the gene’s essential role in bacterial flagellar synthesis and motility (16, 18, 20–22). In contrast, deletions of flgM typically increase class 3 flagellar gene expression: e.g., deletions of the flgM gene in H. pylori and S. Typhimurium enhance flagellin gene expression by several fold (16, 23).

Spirochetes are a group of bacteria that cause several human diseases: e.g., Lyme disease (Borrelia burgdorferi), syphilis (Treponema pallidum), leptospirosis (Leptospira interrogans), and periodontal disease (Treponema denticola) (24–27). A unique aspect of spirochetes is that they have bipolar flagella encased within the periplasmic space. Therefore, spirochetal flagella are named endoflagella or periplasmic flagella (PFs) (28, 29). Spirochetes are diverse in terms of flagellar number and length and whether or not PFs overlap at the middle of cells (29). For example, L. interrogans has a single short flagellum inserted at each pole that does not overlap the other (30). In contrast, B. burgdorferi has 7 to ~11 long PFs subterminally inserted at the cell poles. These PFs form two bundles of ribbon-like structures that overlap at the middle regions of the cells (31, 32). PFs confer upon spirochetes a distinctive flat-wave or spiral morphology and unique form of screw-like motility that empower them to translocate through viscous substrates and tissues to access otherwise inaccessible host niches (29, 33). Therefore, in pathogenic spirochetes, motility facilitates invasion and dissemination and is essential for infection (26, 34). Like external flagellates, PFs are also composed of a motor, hook, and filament; however, they are more complex in terms of motor structures and flagellar filament compositions (35). For example, PFs have a unique structure called the “collar,” a large multiprotein complex consisting of an inner core and an outer turbine-like structure. The collar has a remarkable structural plasticity essential not only for assembly of flagellar motors in the highly curved membrane of spirochetes but also for generation of the high torque necessary for spirochete motility (36–38). The flagellar hook protein FlgE in spirochetes self-catalyzes the formation of an unusual intersubunit lysinoalanine cross-link critical for cell motility (39, 40). Spirochetes are complex in terms of flagellar filament compositions. Most spirochetes have multiple flagellin proteins (28, 29). For instance, the flagellar filaments of T. denticola and T. pallidum comprise at least one sheath protein, FlaA, and three core FlaB proteins (41–45). The flagellar filament of Leptospira is even more complex as it is composed of two FlaA proteins, at least one FlaB protein, and two novel Leptospira-specific flagellar filament proteins, FcpA and FcpB (30, 46, 47). It is noteworthy that FlaB proteins are homologs of FliC. FlaA proteins are unique to spirochetes and do not share sequence similarity with bacterial flagellin proteins (28).

While the structure and function of PFs have been extensively studied in spirochetes (29, 30), the molecular mechanism that regulates PFs’ morphogenesis is poorly understood. Thus far, it remains unclear if spirochetes have evolved the aforementioned transcriptional hierarchy to regulate flagellar gene expression and assembly. In our recent studies, we have been seeking to fill this gap by using T. denticola as a model organism. Similar to other spirochetes, the flagellar filaments of T. denticola are composed of at least one FlaA protein and three FlaB proteins (FlaB1, FlaB2, and FlaB3). The genes encoding these four proteins are regulated by different promoters (41, 44). For instance, a conserved σ²⁸-dependent promoter is mapped upstream of flaB2, and a σ⁷⁰-dependent promoter is mapped upstream of flaA and flaB1 of T. denticola (41). Additionally, we recently identified and functionally characterized an FliA homolog (FliA_Td) in T. denticola and found that it functions as a flagellum-specific σ²⁸ factor that positively regulates the flagellin genes of T. denticola (48). However, FlgM, an antagonist of FliA, has not been identified in T. denticola or other spirochetes. In this report, by using the approaches of bioinformatics, genetics, biochemistry, and cryo-electron tomography (cryo-ET), we provide several lines of evidence that TDE0201 is an FlgM homolog that regulates the flagellin gene expression and flagellar number and polarity of T. denticola.

RESULTS AND DISCUSSION

TDE0201 is an FlgM homolog.

Our recent study shows that TDE2683 is a FliA (σ²⁸) homolog that regulates the later class 3 flagellar genes of T. denticola (48), yet its antagonist FlgM has not been identified. FlgM proteins (PF04316 and PF05998) are not conserved, and their sizes are diverse, ranging from 65 to 131 amino acids (aa) (49). The function of FlgM has been well characterized in S. Typhimurium, and its three-dimensional (3D) structure was determined in Aquifex aeolicus as a complex with σ²⁸ by X-ray crystallography (11, 50, 51). Using these two bacterial FlgM proteins as queries, we searched the genome of T. denticola for their homologs and found that TDE0201 has 15.73% shared sequence identity with S. Typhimurium FlgM (FlgM_St) and 20.48% shared identity with A. aeolicus FlgM (FlgM_Aa). We also performed large-scale sequence alignment analyses and found that TDE0201 shares a very limited sequence identity to its counterparts from other bacteria out of the Treponema genus, indicating that TDE0201 is not conserved. This is also true for other bacterial FlgM homologs: e.g., H. pylori FlgM (HP1122) is ~11% identical to the E. coli homolog (15). TDE0201 consists of 93 aa, and its predicted molecular weight (MW) is 10.54 kDa. Previous structural studies indicate that both FlgM_Aa and FlgM_St are composed of four α-helices (H1′, H2′, H3′, and H4′) (50, 51). Sequence alignment analysis revealed that, similar to FlgM_Aa, TDE0201 also comprises four α-helices, among which H2′ and H3′ align well and contain several conserved residues (e.g., Val33, Ser36, Leu42, and Ile63), although their overall conservation is relatively low (Fig. 1A). To further determine if TDE0201 is an FlgM homolog, we generated its structural model using chain D from the cocrystal structure of σ²⁸/FlgM_Aa complex (PDB code 1rp3) (50) as a template and found that the two proteins share a similar structural architecture and their H3′ helices are well aligned (Fig. 1B). The alignment also uncovered significant differences in the positioning of the N-terminal helices, likely due to their disordered nature in the absence of binding to the σ²⁸ factors (51). Collectively, sequence alignment and homology modeling analyses suggest that TDE0201 is an FlgM homolog; thus, here, it is denoted FlgM_Td.

FIG 1.

TDE0201 is a homolog of FlgM. (A) Sequence alignment of FlgM homologs. The numbers show the positions of amino acids of T. denticola TDE0201 (WP_002666518), S. Typhimurium FlgM (St-FlgM) (NP_460143), and A. aeolicus FlgM (Aa-FlgM) (WP_010880181.1). The rods represent four helices of FlgM, including H1′, H2′, H3′, and H4′. The alignments were conducted using the program Clustal Omega. (B) Homology modeling of TDE0201. The structural model was generated using chain D from the cocrystal structure of A. aeolicus σ²⁸/FlgM complex (PDB code 1rp3) (50) as a template with MODELLER version 10.0 (83) and the MPI Bioinformatics Toolkit (84). The H3′ helices of TDE0201 and FlgM_Aa were superimposed.

The flgM_Td gene is regulated by a σ⁷⁰-dependent promoter.

The flgM_Td gene (TDE0201) and its upstream TDE0200 gene are in the same transcriptional orientation (Fig. 2A). TDE0200 encodes a rRNA small subunit methyltransferase I (RsmI) responsible for the 2′-O-methylation of cytidine 1402 in bacterial 16S rRNA (52). Co-reverse transcription-PCR (co-RT-PCR) analysis showed that only the pair of primers bridging TDE0200 and TDE0201 yielded a positive band, whereas the other two pairs of primers across the intergenic regions of TDE0199-TDE0200 and TDE0201-TDE0202 were both negative (Fig. 2B), suggesting that TDE0201 is cotranscribed with its upstream TDE0200 gene but not with its downstream TDE0199 gene, a gene in an opposite transcriptional direction. By using 5′ rapid amplification of cDNA ends (5′-RACE), we identified a transcriptional start site (TSS) at the upstream region of TDE0200, 21 bp from its start codon (Fig. 2C). The −10 and −35 regions from the TSS contain a consensus sequence of σ⁷⁰-dependent promoter (TTGATT-N17-TATAAT) (Fig. 2D), suggesting that the expression of flgM_Td is most likely controlled by σ⁷⁰.

FIG 2.

TDE0201 resides in a gene cluster that is regulated by a σ⁷⁰ promoter. (A) Diagram showing the genes adjacent to TDE0201. Arrows represent the relative positions and orientations of RT-PCR primers that span the intergenic regions between individual genes as labeled. (B) RT-PCR analysis. For each pair of primers, chromosomal DNA was used as a positive control. The numbers below the primers are predicted sizes of RT-PCR and PCR products. (C) 5′-RACE analysis. The arrow shows the sequencing direction, and an asterisk indicates the transcriptional start site (TSS). The red underlined sequence (GTG) is the start codon of TDE0200. (D) Sequence comparison between the canonical E. coli σ⁷⁰ promoter sequence and that identified upstream of TDE0200.

FlgM_Td inhibits E. coli motility and fliC expression.

FlgM_Td shares about 16% sequence identity and a similar structural topology with E. coli FlgM (FlgM_Ec). We speculated that if FlgM_Td functions as an anti-σ²⁸ factor, overexpressing this gene in E. coli may impair its late flagellar gene expression (e.g., the flagellin gene fliC) and motility. To determine if this is the case, we ectopically expressed flgM_Td in an E. coli DH5α strain and then assessed its impact on motility using swimming plate assays and the level of FliC using immunoblotting analysis. Supporting our speculation, these two experiments showed that induction of flgM_Td with IPTG (isopropyl-β-d-1-thiogalactopyranoside) substantially inhibited E. coli motility (Fig. 3A) and expression of FliC (Fig. 3B). FlgM proteins are exported when HBB assembly is completed. To determine if FlgM_Td can be exported in E. coli, after addition of IPTG, we monitored the presence of FlgM_Td in the growth medium of DH5α by immunoblotting with the FLAG antibody (anti-Flag). We detected FlgM_Td in the whole-cell lysates of DH5α but not in the supernatants concentrated from the growth medium (data not shown), suggesting that no FlgM_Td was secreted into the growth medium via the flagellar export apparatus of E. coli or that it was secreted but its level in the growth medium was too low to be detected. This result is understandable because the domain essential for FlgM_Ec export lies in its N-terminal region (53), which is not conserved in FlgM_Td (Fig. 1A). Altogether, the heterologous expression of FlgM_Td in E. coli indicates that it shares some common characteristics with FlgM_Ec (i.e., binding to FliA and repressing flagellin gene expression and motility).

FIG 3.

Heterologous expression of flgM_Td inhibits E. coli motility and fliC expression. (A) Swimming plate assays with/without IPTG induction. This assay was carried out on agar plates containing 1% tryptone, 0.5% NaCl, and 0.3% Bacto agar as previously documented (86). The plates were incubated at 30°C overnight. Numbers on the plates represent the following conditions: 1, DH5a alone; 2, DH5a transformed with the empty vector pFLAG-CTS; and 3, DH5a transformed with the vector pFLAG-CTS containing flgM_Td. (B) Detection of FliC by immunoblotting analysis. For this experiment, equivalent amounts (~10 μg) of whole-cell lysates of three E. coli strains were analyzed by SDS-PAGE and then probed with specific antibodies against E. coli GroEL (αGroEL) and FliC (αFliC). GroEL was used as a loading control.

FlgM_Td interacts with FliA_Td.

As an anti-σ²⁸ factor, FlgM exerts its inhibitory role on class 3 flagellar gene expression via binding to FliA (11, 13). To determine if FlgM_Td interacts with FliA_Td, we first conducted protein-protein docking analysis by comparing their homology models, which were generated from the cocrystal structure of the σ²⁸/FlgM complex in A. aeolicus (PDB code 1rp3) (50). The docking analysis suggested that FlgM_Td binds to FliA_Td primarily through its C-terminal H3′-H4′ α-helices, whereby H4′ is buried in a deep hydrophobic pocket of FliA_Td (Fig. 4A). To experimentally test the interaction between FliA_Td and FlgM_Td in vivo, we carried out coimmunoprecipiation (co-IP) assays using T. denticola whole-cell lysates and a specific antibody against FliA_Td (anti-FliA_Td). The resulting samples were then subjected to immunoblotting probed against anti-FlgM_Td to determine if FlgM_Td was precipitated along with FliA_Td. FlgM_Td was detected in the eluates from the wild type (WT) but not in those from the ΔflgM mutant (Fig. 4B), suggesting that FlgM_Td and FliA_Td form a complex; otherwise, it would not have been pulled down by anti-FliA_Td. To corroborate this result, we coexpressed His-FliA_Td and FlgM_Td-FLAG in E. coli BL21 cells as previously described (54). After induction with IPTG, two recombinant proteins were copurified under native conditions using FLAG beads. The resulting coeluates were then subjected to SDS-PAGE, followed by immunoblotting probed against either anti-His or anti-Flag antibodies. Our results show that His-FliA_Td was copurified along with FlgM_Td-FLAG (Fig. 4C and D). Previous structural and mutagenesis studies indicate that FlgM interacts with FliA mainly via its C-terminal H3′-H4′ helices, which bind to the σ4 domain of FliA (11, 50, 53). To determine if the interaction between FlgM_Td and FliA_Td is specific, we mutated E221 and V231, two conserved residues in the σ4 domain of FliA_Td (48), and truncated the H4′ helix (79 to 93 aa) of FlgM_Td. We first repeated the copurification experiment using three mutated FliA_Td proteins (E221D, V231E, and E221D V231E) and found that the single mutation in V231 and double mutations in both E221 and V231 nearly abolished the interaction of two proteins, whereas the mutation in E221 had no impact (Fig. 4C and D). We then repeated this experiment using the truncated FlgM_Td (ΔH4′) and found that the truncation completely abolished the interaction between FlgM_Td and FliA_Td (Fig. 4E and F). Based on these results, we infer that FlgM_Td binds to FliA_Td mainly through its C-terminal H4′ helix.

FIG 4.

FliA_Td interacts with FlgM_Td. (A) Protein-protein docking analysis of FliA_Td and FlgM_Td. The homology models of FliA_Td and FlgM_Td were first generated using the cocrystal structure of A. aeolicus s²⁸/FlgM complex (PDB code 1rp3), and then protein-protein docking analysis was conducted on the ClusPro server, with FliA_Td shown as a molecular surface and FlgM_Td shown as a backbone ribbon. (B) Coimmunoprecipitation (co-IP). This assay was performed using T. denticola whole-cell lysates and a polyclonal antibody against FliA_Td (αFliA_Td). The final co-IP eluates were probed using anti-FlgM_Td. A deletion mutant of flgM_Td (ΔflgM) was used as a negative control. (C to F) Copurification of N-terminal His-tagged FliA_Td (His-FliA_Td) and C-terminal FLAG-tagged FlgM_Td (FlgM_Td-FLAG). The two tagged proteins and their corresponding site-directed mutated or truncated proteins were coexpressed in the E. coli BL21 Star(DE3) strain and then purified using FLAG beads. The resulting samples were subjected to SDS-PAGE (C and E), followed by immunoblotting analysis (D and F) using either anti-FLAG (αFlag) or anti-His (αHis) antibodies. In panels C and D, in addition to the wild-type FliA_Td protein (lane 1), three site-directed mutated recombinant proteins were included: E221D (lane 2), V231E (lane 3), and a double mutation of E221D andV231E (lane 4). In panels E and F, lane 1 shows the wild-type FlgM_Td protein and lane 2 shows a truncated FlgM_Td protein without the H4′ helix (79 to 93 aa).

FlgM_Td prevents FliA_Td from binding to the σ²⁸-dependent promoter.

Our recent study shows that FliA_Td binds to σ²⁸-dependent promoters, such as the flaB2 promoter (P_flaB2) (48). We reasoned that if FlgM_Td acts as an anti-σ²⁸ factor, it should sequester FliA_Td and prevent it from binding to σ²⁸-dependent promoters. To test this hypothesis, we carried out an electrophoretic mobility shift assay (EMSA) using biotin-labeled P_flaB2 as a DNA probe. Our results showed that the recombinant FliA_Td (rFliA_Td) bound to P_flaB2, which was titrated by the recombinant FlgM_Td (rFlgM_Td) in a dose-dependent manner (Fig. 5). Notably, rFlgM_Td alone did not bind to P_flaB2. Along with the above result, we conclude that FlgM_Td functions as an anti-σ²⁸ factor that interacts with FliA_Td and prevents it from binding to the σ²⁸-dependent promoter.

FIG 5.

FlgM_Td sequesters FliA_Td from binding to the flaB2 promoter. EMSA was performed using biotin-labeled flaB2 promoter (Bio-P_flaB2) as a DNA probe. For this assay, Bio-P_flaB2 was incubated with a fixed amount of recombinant FliA_Td protein (rFliA_Td) and then titrated with increasing concentrations of recombinant FlgM_Td protein (rFlgM_Td) as labeled.

FlgM_Td controls T. denticola flagellar filament gene expression.

To investigate the role of FlgM_Td in T. denticola, we constructed an flgM_Td deletion mutant (ΔflgM) and its isogenic complemented strain (cflgM) using a strategy illustrated in Fig. S1 in the supplemental material. Immunoblotting assays with anti-FlgM_Td showed that the production of FlgM_Td was abolished in the ΔflgM mutant and then restored in the cflgM strain (Fig. 6A). Deletions of flgM genes typically increased class 3 flagellar gene expression (16, 23). Unexpectedly, our immunoblotting results uncovered that deletion of flgM_Td significantly reduced the level of four flagellar filament proteins in T. denticola, including FlaA, FlaB1, FlaB2, and FlaB3, whereas the level of the hook protein FlgE remained unchanged (Fig. 6A). The result from FlgE is expected because it belongs to the class 2 flagellar genes, which are regulated by σ⁷⁰ in the flagellar transcriptional cascade of E. coli and S. Typhimurium (2, 55). To elucidate the mechanism underpinning the reduction of four flagellar filament proteins in the ΔflgM mutant, we monitored their transcript (mRNA) levels during growth by using quantitative RT-PCR (qRT-PCR) and found that the expression of flaA, flaB1, and flaB3 genes was constantly impaired in the ΔflgM mutant at the mid-log, early stationary, and stationary phases (Fig. 6B). Compared to the wild type, expression of the flaB2 gene in the ΔflgM mutant was more diverse (Fig. 6B), ranging from a slight reduction at log phase, transient increase at early stationary phase, and ~50% reduction at stationary phase. We repeated this experiment three times with different biological samples and found a similar pattern. This result indicates that the reduction of four flagellar filament proteins in the ΔflgM mutant mainly occurs at the transcriptional level.

FIG 6.

Deletion of flgM_Td impairs T. denticola flagellar filament gene expression. (A) Immunoblotting analysis. Equivalent amounts of whole-cell lysates of the wild-type (WT), ΔflgM, and cflgM strains were analyzed by SDS-PAGE and then probed with specific antibodies against the following proteins: DnaK (loading control), FlaA, FlaBs, FlgE, FlgM_Td, and FliA_Td. (B) qRT-PCR analysis. For this experiment, the levels of four flagellar filament genes (flaA, flaB1, flaB2, and flaB3) were measured by qRT-PCR as previously described (41, 48). The dnaK gene transcript was used as an internal control to normalize the readouts of qRT-PCR. The results are expressed as the level of individual gene transcripts in the ΔflgM mutant relative to that of the wild type at three different growth phases, including the mid-logarithmic (Mid-log), early stationary (Early-stat), and stationary (Stat) phases. Asterisks indicate that the difference between the wild-type and ΔflgM strains was statistically significant at a P value of <0.01 (one-way ANOVA followed by Tukey’s multiple-comparison test).

Our recent reports show that the flaB2 gene is positively regulated by FliA_Td (41, 48). Thus, we expected that deletion of FlgM_Td, an anti-σ²⁸ factor, should increase its expression. Along with this expectation, we found that the level of flaB2 mRNA was transiently elevated at the early stationary phase of the ΔflgM mutant. However, its expression level was attenuated at the stationary phase, which is contradictory to the role of FlgM_Td as an anti-σ²⁸ factor. To explore its underlying mechanism, we first measured the level of FliA_Td by immunoblotting with anti-FliA_Td and found that its level significantly decreased in the mutant (Fig. 6A). FlgM interacts with FliA, forming a complex in the cytoplasm (50). We reasoned that deletion of FlgM_Td may affect the stability of FliA_Td, which in turn impairs flagellin gene expression in the mutant. To test this possibility, we carried out protein turnover assays to monitor the stability of FliA_Td. Our results uncovered that while FliA_Td was stable in wild type during the course of 24 h, it was rapidly degraded in the ΔflgM mutant: e.g., 6 h after arresting protein synthesis with spectinomycin, FliA_Td was almost undetectable (Fig. 7A). This result can explain the reduction of flaB2 mRNA, a gene regulated by FliA_Td (σ²⁸). It is possible that without FlgM_Td, free FliA_Td in the cytoplasm is degraded by certain proteases that are activated when T. denticola enters the stationary phase, which in turn impairs the expression of the flaB2 gene. Supporting this possibility, a similar scenario is reported in E. coli, whereby FlgM protects FliA from proteolysis, which mainly depends on the Lon protease (56).

FIG 7.

Deletion of flgM_Td leads to FliA_Td degradation and increases expression of the csrA gene. (A) Protein turnover assays of the wild type (WT) and the ΔflgM mutant. Protein translation was arrested by adding spectinomycin to the mid-log cultures. Samples were collected at the indicated time points and analyzed by immunoblotting. Equivalent amounts of whole-cell lysates were analyzed by SDS-PAGE and then probed with specific antibodies to DnaK (αDnaK) and FliA_Td (αFliA_Td), and DnaK was used as a loading control. (B) Detection of csrA (TDE2355) gene transcripts using qRT-PCR. Similar to the above qRT-PCR analysis, dnaK was used as an internal control to normalize the readouts of qRT-PCR, and the results are expressed as the level of individual gene transcripts in the ΔflgM mutant relative to those of the wild type at the mid-log, early stationary, and stationary phases. Asterisks indicate that the difference between the wild-type and ΔflgM strains was statistically significant at a P value of <0.01.

It was unexpected that deletion of flgM_Td led to the reduction of flaA, flaB1, and flaB3 transcripts (Fig. 6) because they are regulated by σ⁷⁰ rather than σ²⁸ (41, 57). CsrA (carbon storage regulator A) is a small RNA binding protein that represses flagellin gene expression in several bacteria both transcriptionally and translationally (32, 58, 59). CsrA is also a global regulator that orchestrates vast gene expression, including σ factors such as FliA (60). The genome of T. denticola encodes a CsrA homolog (TDE2355) (61), and its binding site is found at the upstream untranslated regions (UTRs) of the flaA, flaB1, and flaB3 genes (data not shown); thus, we speculated that deletion of flgM_Td may increase expression of TDE2355 which, in turn, represses expression of the flaA, flaB1, and flaB3 genes. To determine if this possibility stands, we measured the TDE2355 mRNA using qRT-PCR and found that its level was significantly increased in the mutant at the log, early stationary, and stationary phases (Fig. 7B), which coincides with the reduction of flaA, flaB1, and flaB3 mRNA. A similar phenotype is also reported in Yersinia pseudotuberculosis (62). Taken together, these results indicate that while FlgM_Td is an anti-σ²⁸ factor, its role and mechanism of action are more complex than those of its counterparts in other bacteria. This unique regulatory mechanism may allow the spirochete to finely govern the level of multiple flagellin proteins to assemble flagellar filaments, which are structurally more complex than most external flagellates.

FlgM_Td is essential for the motility of T. denticola.

Deletion of flgM_Td has no impact on T. denticola growth because the growth rate of the ΔflgM mutant was similar to those of the wild-type and cflgM strains (Fig. 8A). Under dark-field microscopy, we found that the ΔflgM mutant swam poorly and was unable to displace in 1% methylcellulose (Videos S1 to S3), suggesting that its motility was impaired. To confirm this observation, we carried out swimming plate assays (Fig. 8B) and found that the swimming rings of the ΔflgM mutant (5.25 ± 0.17 mm; n = 16 plates) were significantly smaller than those of the wild type (18.75 ± 0.24 mm; n = 16 plates) and the complemented strain (11.62 ± 0.38 mm; n = 16 plates) (Fig. 8C). This result was further confirmed by bacterial motion tracking analysis, wherein individual bacterial cells were tracked for at least 30 s in 1% methylcellulose, and their average velocities (μm/s, n = 26 cells) were measured as follows: wild type, 10.31 ± 0.6 μm/s; ΔflgM strain, 1.39 ± 0.23 μm/s; and cflgM strain, 6.82 ± 0.39 μm/s. These results indicate that FlgM_Td is indispensable for T. denticola motility, in contrast to its counterparts in other bacteria, where deletions of flgM genes either have no significant impact on motility or lead to abnormal swimming behaviors (16, 23, 63, 64). We noticed that the motility of cflgM was not fully restored. This may be because the expressional level or timing of flgM_Td in the complemented strain does not fully coincide with that of wild type. This strain was created by cis-complementation. We also tried to complement the ΔflgM mutant using pBFC, a shuttle vector of T. denticola (65, 66); however, this vector is poorly transformable in the parental wild-type ATCC 35405 strain (65).

FIG 8.

Deletion of flgM_Td has no impact on T. denticola growth but inhibits its motility. (A) Growth curves of the wild-type (WT), ΔflgM, and cflgM strains. For this experiment, 10⁵ T. denticola cells were inoculated into the TYGVS medium and enumerated every 24 h for up to 6 days using a Petroff-Hausser counting chamber. (B) Swimming plate assays. This assay was carried out on 0.35% agarose containing the TYGVS medium diluted 1:1 with PBS. The plates were incubated anaerobically at 37°C for 3 days. The Δtap1 strain, a nonmotile mutant, was used as a control to determine initial inoculum sizes. (C) Cell tracking analysis of the wild-type, ΔflgM, and cflgM strains. T. denticola cells were tracked in the presence of 1% methylcellulose as previously described (57). The results are expressed as average cell velocities (micrometers per second) of at least 20 cells. The data were analyzed by one-way ANOVA, followed by Tukey’s multiple-comparison test at P < 0.01.

Deletion of flgM_Td alters the flagellar polarity of T. denticola.

To determine if deletion of flgM_Td impairs T. denticola flagellar assembly, we applied cryo-ET to directly visualize PFs inside the cells. In total, we generated 20 whole-cell reconstructions, including wild-type (3 cells), ΔflgM (10 cells), and cflgM (7 cells) strains. We also enumerated the number of flagellar motors and measured the length of PFs (Table 1). As in our previous reports (41, 42, 48, 57), T. denticola wild-type cells uniformly have bipolar flagella (two PFs at both cell poles), which form two helical bundles wrapping around the cell cylinders (Fig. 9A). Interestingly, the two bundles of PFs are asymmetrical in terms of length (i.e., the PFs from one end of the cells are usually longer than those from the opposite end). Jutras et al. uncovered that T. denticola forms a single division zone and divides near the middle of the cells (67); thus, a nascent daughter cell might inherit one bundle of PFs from the old cell pole and newly assemble the other bundle of PFs at the new cell pole. Therefore, we referred to the pole with long PFs as the “old pole” and the one with short PFs as the “new pole.” We measured the lengths of PFs and found that those from “old pole” (5.0 ± 0.6 μm; n = 6 PFs) are nearly two times longer than those from “new pole” (2.9 ± 0.7 μm; n = 6 PFs). A similar phenotype was also observed in the complemented strain (Table 1). The phenotype of the ΔflgM mutant is quite striking. Instead of having bipolar PFs, as observed in both the wild-type and cflgM strains, the majority of mutant cells (8/10) had two PFs at only one end of cells (Fig. 9B and Table 1). Of the remaining two mutant cells, one had a single flagellum at each cell end, and the other had no PFs. We found that the number of flagellar motors (3.3 motors per cell) in the mutant is slightly lower than that of wild type (4 motors per cell) but not statistically significant (Table 1). We also measured the length of PFs in situ and found that the remaining two PFs in the mutant (5.2 ± 1.1 μm; n = 18 PFs) are longer (P < 0.05) than those in the wild type (3.9 ± 1.3 μm; n = 12 PFs). Occasionally, those PFs were too long and even protruded from the cells (Fig. 9B). These results indicate that deletion of flgM_Td has limited impact on the number of flagellar motors but does alter the flagellar filament number, length, and polarity in T. denticola. To our knowledge, a similar phenotype has not been reported in other bacteria.

TABLE 1.

Cryo-ET analysis of the wild type, ΔflgM, and cflgM strains

Strains	No. of motors	PF length(s) (μm) ina:
		New pole	Old pole
Wild type
Cell 1	4	2.1 and 2.1	4.8 and 6.1
Cell 2	4	3.5 and 2.5	5.0 and 5.5
Cell 3	4	3.4 and 3.5	4.6 and 4.6
Avg	4	2.9 ± 0.7 (n = 6)	5.1 ± 0.6 (n = 6)
ΔflgM mutant
Cell 1	3	2.5	5.0
Cell 2	4	ND	ND
Cell 3	3	ND	5.5 and 5.5
Cell 4	4	ND	6.0 and 6.0
Cell 5	3	ND	6.5 and 6.5
Cell 6	3	ND	4.0 and 4.5
Cell 7	3	ND	4.0 and 6.0
Cell 8	3	ND	5.8 and 6.0
Cell 9	3	ND	4.0 and 6.0
Cell 10	4	ND	5.0 and 6.0
Avg	3.3	ND	5.6 ± 0.7 (n = 17)
cflgM complemented clone
Cell 1	4	4.0 and 4.0	5.0 and 5.0
Cell 2	4	3.0 and 4.5	5.0 and 5.0
Cell 3	4	4.5 and 5.0	5.0 and 5.0
Cell 4	4	1.5 and 1.5	6.0 and 6.0
Cell 5	4	0.5 and 1.5	5.5 and 5.5
Cell 6	4	3.0 and 6.5	6.5 and 6.5
Cell 7	4	6.5 and 5.0	6.5 and 6.0
Avg	4	3.6 ± 1.9 (n = 14)	5.6 ± 0.7 (n = 14)

^aND, no PFs detected; n, number of PFs.

FIG 9.

Whole-cell cryo-ET analysis of the wild-type, ΔflgM, and cflgM cells. Shown are projection images of a wild-type cell (A), ΔflgM cell (B), and cflgM cell (C). PFs originating from one end of cells are colored in yellow and those from the opposite end in orange.

Summary.

The above studies provide several lines of evidence that TDE0201 is a FlgM homolog: (i) it has a structural topology similar to that of its counterparts from other bacteria (Fig. 1), (ii) it represses E. coli motility and fliC expression (Fig. 2), and (iii) it interacts with the T. denticola σ²⁸ (FliA_Td) and prevents it from binding to the σ²⁸-dependent promoters (Fig. 4 and 5). Identification of FlgM_Td and FliA_Td indicates that T. denticola has also evolved to regulate its later flagellar gene expression through the partner-switching mechanism between σ²⁸ and its antagonist FlgM. We also searched the genomes of other spirochetes using FliA_Td and FlgM_Td as queries and identified their homologs in T. pallidum, Brachyspira, and Leptospira species: e.g., FliA_Td and FlgM_Td share 69% and 39% sequence identities with TP_0709 and TP_0974 of T. pallidum (68), respectively, suggesting that a similar regulatory mechanism also exists in other spirochetes. Therefore, the results presented in this report also shed light on the role of FliA and FlgM homologs in other spirochetes. Our results also demonstrate that FlgM_Td acts differently from its counterparts in other bacteria in several ways: (i) FlgM_Td protects FliA_Td from being degraded (e.g., deletion of flgM_Td leads to FliA_Td rapid turnover) (Fig. 7), (ii) deletion of flgM_Td represses flagellin gene expression in T. denticola (Fig. 6), rather than increasing flagellin gene expression as reported in other bacteria, (iii) FlgM_Td interplays with other regulators, such as CsrA (Fig. 7), and (iv) deletion of flgM_Td alters the flagellar polarity from bipolar to monopolar (Fig. 9). We found that the reduction of flagellin gene expression in the ΔflgM mutant is mainly due to FliA_Td turnover and CsrA upregulation. However, the mechanism by which FlgM_Td regulates flagellar polarity remains unknown. Previous studies have shown that small GTPase FlhF and FlhG act cooperatively to harness flagellar number and placement in several polar flagellates, such as B. burgdorferi (31), C. jejuni (69), and Vibrio alginolyticus (70). Additionally, FapA (flagellar assembly protein A) controls the flagellar polarity of Vibrio vulnificus (71, 72). We recently found FlhF (TDE2686), FlhG (TDE2685), and FapA (TDE2682) homologs in a large motility gene operon of T. denticola (48). Therefore, it is possible that FlgM_Td regulates flagellar polarity by directly or indirectly modulating these proteins. We are planning to perform chromatin immunoprecipitation sequencing (ChIP-seq) to map the promoters recognized by FliA_Td and transcriptome sequencing (RNA-seq) to determine genes coordinately regulated by FliA_Td and FlgM_Td. Results from these studies will provide insights into the role that FlgM_Td plays in control of T. denticola flagellar polarity.

MATERIALS AND METHODS

Bacterial strains, culture conditions, and oligonucleotide primers.

Treponema denticola ATCC 35405 (wild type [WT]) was used in this study (61). Cells were grown in tryptone-yeast extract-gelatin-volatile fatty acids-serum (TYGVS) medium at 37°C in an anaerobic chamber in the presence of 85% nitrogen, 5% carbon dioxide, and 5% hydrogen (73). T. denticola isogenic mutants were grown with appropriate antibiotic for selective pressure as needed: erythromycin (50 μg/mL) and gentamicin (20 μg/mL). The Escherichia coli DH5α strain (New England Biolabs, Ipswich, MA) was used for DNA cloning. BL21 Star(DE3) (Invitrogen, Waltham, MA) was used for preparing recombinant proteins. The E. coli strains were cultivated in lysogeny broth (LB) supplemented with appropriate concentrations of antibiotics for selective pressure as needed: erythromycin, 400 μg/mL; gentamicin, 20 μg/mL; kanamycin 50 μg/mL; and ampicillin, 100 μg/mL. The oligonucleotide primers for PCR and RT-PCR used in this study are listed in Table S1 in the supplemental material. These primers were synthesized in Integrated DNA Technologies (IDT, Coralville, IA).

Electrophoresis and immunoblotting analysis.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting analyses were carried out as previously described (48). For immunoblotting, T. denticola cells were harvested in the mid-logarithmic (mid-log) phase (~5 × 10⁸ cells/mL). Equal amounts of whole-cell lysates (1 to 5 μg) were separated by SDS-PAGE and then transferred to polyvinylidene difluoride (PVDF) membranes. The immunoblots were probed with specific antibodies against T. pallidum FlaB and T. denticola FlaA, DnaK, FlgE, and FliA as described in previous publications (48, 74, 75). The antibody against T. denticola TDE0201 was generated in rats in this study (see below for details). His tag and FLAG tag antibodies were purchased from Thermo Fisher Scientific (Waltham, MA) and Sigma-Aldrich (St. Louis, MO). E. coli flagellin (FliC) and GroEL antibodies were purchased from Abcam (Cambridge, United Kingdom). Immunoblots were developed using a horseradish peroxidase-conjugated secondary antibody with an enhanced chemiluminescence (ECL) assay, and signals were quantified using the Molecular Imager ChemiDoc system with the Image Lab software (Bio-Rad Laboratories, Hercules, CA) as previously described (57).

Preparation of TDE0201 recombinant protein and its antiserum.

The gene encoding the full-length TDE201 protein was PCR amplified with primers P₁/P₂ using Pfx DNA polymerase (Life Technologies, Grand Island, NY) and then cloned into the pET200/d-TOPO expression vector (Life Technologies), which encodes a 6-histidine-tag (6×His) at the N terminus. The resulting plasmids were then transformed into BL21 Star(DE3) (Invitrogen). The expression of TDE0201 was induced using 1 mM isopropyl-β-d-1-thiogalactopyranodside (IPTG). For antiserum preparation, the recombinant TDE0201 protein was purified using Ni-nitrilotriacetic acid (NTA) agarose (Qiagen, Valencia, CA) under denaturing conditions according to the manufacturer’s protocol (Qiagen). For the electrophoretic mobility shift assay (EMSA), the recombinant proteins were purified using Ni-NTA agarose (Qiagen) under native conditions in the presence of 1% Sarkosyl according to the manufacturer’s protocol (Qiagen). The purified proteins were then dialyzed in 20 mM Tris-HCl buffer (pH 8.0) at 4°C overnight using 3.0-kDa molecular-weight-cutoff Spectra/Por dialysis bags (Spectrum Laboratories, Rancho Dominguez, CA). The concentrations of purified proteins were determined using a Bio-Rad Protein assay kit (Bio-Rad). The antibody against TDE0201 was produced by General Bioscience (Brisbane, CA) as previously described (48). In brief, 5 mg of purified recombinant TDE0201 was used to immunize two rats following a standard immunization procedure. The primers for preparation of the recombinant proteins are listed in Table S1.

Heterologous expression of TDE0201 in E. coli.

The gene encoding the full-length TDE201 protein was PCR amplified with primers P₄₀/P₄₁, and the resulting amplicon was first cloned into pGEM-T Easy vector (Promega, Madison, WI) and then subcloned into pFLAG-CTS expression vector (Sigma-Aldrich). The resulting plasmid (pFLAG-CTS-TDE0201) and pFLAG-CTS (negative control) were transformed into the E. coli DH5α strain, respectively. The resulting DH5α strains were subjected to immunoblotting analysis with FLAG antibody (anti-Flag) for detection of TDE0201 and anti-FliC for detection of E. coli FliC expression, as well as to swimming plate assays for assessing the impact of TDE0201 on E. coli motility. Swimming plate assays of E. coli were performed on soft agar plates containing 1% tryptone, 0.5% NaCl, and 0.3% agar as previously described (48). In brief, overnight E. coli cultures were diluted to 1:100. After 8 h of growth at 37°C, 1 μL of the culture was inoculated onto the soft agar plates with or without IPTG (1 mM), and then the plates were incubated at 30°C for 16 h. The average diameters of swimming rings were calculated from four independent plates; the results are represented as the mean of diameters ± standard error of the mean (SEM). The data were statistically analyzed by one-way analysis of variance (ANOVA), followed by Tukey’s multiple-comparison test at P < 0.01.

Construction of a TDE0201 deletion mutant and its complemented strain.

The TDE0201::erm (Fig. S1A) plasmid was constructed to replace the entire open reading frame of TDE0201 with a previously documented erythromycin resistance cassette (ERM-F/AM) (76). The vector TDE0201^com (Fig. S1B) was constructed to cis-complement the TDE0201 deletion mutant. These two plasmids were constructed by two-step PCR and DNA cloning. To construct TDE0201::erm, the TDE0201 upstream region, ERM-F/AM, and the TDE0201 downstream region were PCR amplified with primers P₃/P₄, P₅/P₆, and P₇/P₈, respectively. The resulting PCR fragments were cloned into the pGEM-T Easy vector (Promega, Madison, WI). The upstream and downstream regions were ligated. The resulted fragment was then inserted by ERM-F/AM at an engineered XbaI cut site, generating TDE0201::erm, as illustrated in Fig. S1A. To delete TDE0201, TDE0201::erm was linearized and transformed into T. denticola wild-type competent cells via heat shock as previously described (77). The deletion was confirmed by PCR and immunoblotting analyses. The resulting mutant was designated the ΔflgM strain. To construct TDE0201^com, the full-length TDE0201 gene along with its upstream region and the aacC1 cassette were PCR amplified with primers P₃/P₉ and P₁₀/P₁₁, respectively, and then fused together with primers P₃/P₁₁, generating TDE0201-aacC1. The downstream region of TDE0201 was PCR amplified with primers P₁₂/P₈ and then fused to TDE0201-aacC1 by PCR using primers P₃/P₈ (Fig. S1B). The obtained DNA fragment was cloned into the pGEM-T Easy vector, generating the vector TDE0201^com. To restore the expression of TDE0201, TDE0201^com was linearized and transformed into the ΔflgM mutant via heat shock (77). The resulting complemented clones were confirmed by PCR and immunoblotting analyses. One complemented clone (cflgM) was selected for further characterizations. The primers for constructing these two plasmids are listed in Table S1.

Coimmunoprecipitation.

Coimmunoprecipitation (co-IP) was carried out as previously described (57). Briefly, 500 mL of the mid-log-phase T. denticola wild-type and ΔflgM mutant (negative control) cultures (~5 × 10⁸ cells/mL) was centrifuged at 5,000 × g for 20 min at 4°C. The obtained cell pellets were washed three times with phosphate-buffered saline (PBS [pH 7.4]) and then resuspended in 2 mL of TSEA buffer (50 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, 0.05% sodium azide [pH 7.5]). The resulting cell suspensions were sonicated on ice, followed by centrifugation (14,800 × g for 30 min at 4°C) to collect the cell supernatants. For co-IP, ~500 μL supernatant was incubated with 100 μL of FliA_Td antibody (anti-FliA_Td) for 16 h at 4°C in the presence of 1% bovine serum albumin (BSA). After the incubation, ~50 μL of protein A agarose (Sigma-Aldrich) was added, and the mixture was further incubated for 16 h at 4°C. The resulting samples were centrifuged (1,600 × g at 4°C) and then washed three times with 1 mL of PBS buffer containing 0.05% Tween 20. The final pellets were suspended in 50 μL of electrophoresis sample buffer, boiled for 5 min, and then briefly centrifuged as before to collect the supernatants, which were then subjected to SDS-PAGE and immunoblotting analysis.

Coexpression of recombinant TDE2683 (FliA_Td) and TDE0201 (FlgM_Td) proteins.

To determine if FlgM_Td binds to FliA_Td, these two proteins were coexpressed in E. coli with either a His tag (His-FliA_Td) or a FLAG tag (FlgM_Td-FLAG) as previously described (58, 78). Briefly, the genes encoding the full-length FliA_Td and FlgM_Td were amplified with primers P₁₃/P₁₄ and P₁₅/P₁₆, respectively, and then fused together with primers P₁₃/P₁₆. Of note, the sequence encoding FLAG was added to the 3′ end of TDE0201 and a stop codon was added between these two genes. The resulting amplicons were cloned into pGEM-T Easy vector (Promega) and then subcloned into pQE80L expression vector (Qiagen), which encodes the 6×His tag at the N terminus. The resulting plasmids were transformed into BL21 Star(DE3) (Invitrogen). Recombinant His-FliA_Td and FlgM_Td-FLAG proteins were induced using 1 mM isopropyl β-d-1-thiogalactosidase (IPTG) and then purified using either Ni-NTA agarose (Qiagen) or anti-Flag M2 affinity gel (Sigma-Aldrich) under native conditions according to the manufacturers’ protocols.

Bacterial swimming plate assay and motion tracking analysis.

Swimming plate assays were performed as previously described (41). Briefly, 4 μL of T. denticola cultures (10⁹ cells/mL) was inoculated onto 0.35% agarose containing TYGVS medium diluted 1:1 with PBS. The plates were incubated anaerobically at 37°C for 3 days to allow the cells to swim out. The diameters of swimming rings were measured in millimeters. As a negative control, a previously constructed nonmotile mutant of T. denticola (Δtap1) was included to determine the initial inoculum size (79). The average diameters of each strain were calculated from at least four independent plates; the results are represented as the mean of diameters ± standard error of the mean (SEM). The velocity of bacterial cells was measured using a computer-based bacterial tracking system as previously described (57). Briefly, 100 μL of mid-log-phase T. denticola cultures was first diluted (1:50) in oral bacterial growth medium (OBGM), and then 10 μL of diluted cultures was mixed with an equal volume of 2% methylcellulose with a viscosity of 4,000 cp (80). T. denticola cells were videotaped and tracked. For each bacterial strain, 26 cells were recorded for up to 30 s. The average cell swimming velocities (micrometers per second) of tracked cells were calculated.

Measuring T. denticola growth rates.

For this study, a total of 10⁵ cells/mL of late-log-phase T. denticola cultures was inoculated into 5 mL TYGVS medium and then were enumerated every 24 h using a Petroff-Hausser counting chamber (Hausser Scientific, Horsham, PA). Each growth curve is representative of at least three independent cultures, and the results are represented as the mean of cell numbers ± standard error of mean (SEM).

Measuring the level of FlgM_Td throughout the growth phase of T. denticola.

For this study, the wild-type strain and a previously constructed flgE-deficient mutant (ΔflgE) of T. denticola were used (61, 76). A total of 2 × 10⁶ cells/mL of late-log-phase T. denticola cultures was inoculated into the TYGVS medium. T. denticola cells in the cultures were enumerated every 24 h using a Petroff-Hausser counting chamber (Hausser Scientific, Horsham, PA). Samples were harvested at different time points and subjected to immunoblotting analysis.

RNA preparations, qRT-PCR, and 5′-RACE.

RNA isolations were performed as previously described (42). Briefly, a total of 10⁵ cells/mL of late-log-phase T. denticola cultures was inoculated into 100 mL TYGVS medium. T. denticola cells in the cultures were enumerated every 24 h using a Petroff-Hausser counting chamber (Hausser Scientific). Samples were harvested at different time points. Total RNA was extracted using TRI reagent (Sigma-Aldrich), following the manufacturer's instructions. The resulting samples were treated with Turbo DNase I (Thermo Fisher Scientific) at 37°C for 2 h to eliminate genomic DNA contamination. The resulting RNA samples were reextracted using acid phenol-chloroform (Ambion), precipitated in isopropanol, and washed once with 70% ethanol. The RNA pellets were resuspended in RNase-free water. cDNA was generated from the purified RNA (1 μg) using a SuperScript IV VILO cDNA synthesis kit (Thermo Fisher Scientific). qRT-PCR was performed using iQ SYBR green supermix and an MyiQ thermal cycler (Bio-Rad). The gene encoding DnaK (TDE0628) was used as an internal control to normalize the qRT-PCR data as previously described (42). The results were expressed as the normalized difference of threshold cycles (ΔΔC_T) between the wild type and ΔflgM mutant. 5′-RACE analysis was performed using the SMART RACE 5′/3′ kit (TaKaRa Bio USA) according to the manufacturer’s protocol. The primers for RT-PCR, qRT-PCR, and 5′-RACE are listed in Table S1.

Protein turnover assay.

The protein turnover assay was carried out as previously described (42, 57). Briefly, T. denticola strains were first grown to the mid-log phase (~5 × 10⁸ cells/mL). Subsequently, spectinomycin (100 μg/mL) was added into the cultures, followed by incubation at 37°C. Samples (10 mL) were harvested at the indicated time points and subjected to immunoblotting analysis.

EMSA.

The electrophoretic mobility shift assay (EMSA) was performed using recombinant proteins FliA_Td (rFliA_Td) and FlgM_Td (rFlgM_Td), as well as the flaB2 promoter (P_flaB2) as a DNA probe, as previously described (48). The DNA probe was synthesized by PCR using primers P₁₇/P₁₈ (Table S1) and labeled with biotin using the Biotin DecaLabel DNA labeling kit (Thermo Fisher Scientific). EMSA was performed by adding increasing amounts of rFlgM_Td (25.8 μM, 65 μM, and 130 μM) to biotin-labeled P_flaB2 (2 mM) in binding buffer (10 mM Tris-HCl [pH 7.5]). The mixtures were incubated at room temperature for 30 min. After the incubation, 25.8 μM rFliA_Td was added to the reaction mixtures and the mixtures were further incubated for another 1 h. DNA loading dye was then added to the reaction mixtures immediately before loading. The reaction mixtures were separated on 4 to 20% Mini-Protean TGX gels using 0.5× Tris Borate EDTA (TBE) buffer and then transferred to a nylon membrane and cross-linked using UV light. The detection was performed using a Chemiluminescent nucleic acid detection module kit (Thermo Fisher Scientific).

Cryo-ET sample preparation, data collection, and image processing.

The freeze-hydrated specimens of T. denticola were prepared as previously described (42). Briefly, T. denticola cultures were mixed with 10-nm colloidal gold solutions and then deposited on a freshly glow-discharged, holey carbon grid for about 1 min. The grids were blotted with a small piece of filter paper for ~4 s and then rapidly plunged into liquid ethane using a gravity-driven plunger apparatus. The grids were then transferred to a 300-kV Krios electron microscope (Thermo Fisher Scientific) equipped with a field emission gun, Volta phase plate (VPP), and a direct detection detector (Gatan). To generate three-dimensional (3D) reconstructions of whole bacterial cells, SerialEM (81) was used to acquire multiple tilt series along the cell at a ×19,000 magnification. The pixel size at the specimen level is 5.4 Å. Tilt series were collected in the low-dose mode with the VPP at an ~0.5-μm defocus. A total dose of 60 e⁻/Ų was distributed among 33 tilt images covering angles from −48° to +48° at tilt steps of 3°. The tilt series were aligned and reconstructed by IMOD (82). Multiple reconstructions from different segments of the same cell were integrated into one large reconstruction. In total, we generated over 90 tomograms to build 20 reconstructions of the entire cells from the wild-type (3 cells), ΔflgM (10 cells), and cflgM (7 cells) strains.

Bioinformatics and statistical analyses.

Sequence alignment analyses were performed using Clustal Omega. Homology modeling was generated using MODELLER version 10.0 (83), as incorporated in the MPI Bioinformatics Toolkit (84), and protein-protein docking analysis was conducted using the ClusPro server (85). Statistical significance was analyzed by one way ANOVA followed by Tukey’s multiple-comparison test at P < 0.01.