Structure of Salmonella FlhE, conserved member of a flagellar Type III secretion operon

Abstract

The bacterial flagellum is assembled by a multicomponent transport apparatus categorized as a Type III secretion (T3S) system. The secretion of proteins that assemble into the flagellum is driven by the proton motive force. The periplasmic protein FlhE is a member of the flhBAE operon in the majority of bacteria where FlhE is found. FlhA and FlhB are established components of the flagellar T3S system. The absence of FlhE results in a proton leak through the flagellar system, inappropriate secretion patterns, and cell death, indicating that FlhE regulates an important aspect of proper flagellar biosynthesis. We isolated FlhE from the periplasm of Salmonella and solved its structure to 1.5 Å resolution. Possible roles of FlhE, including that of a chaperone, are discussed.

Introduction

The importance of protein secretion to bacterial pathogens is exemplified by the array of mechanisms that have evolved for this purpose [1–6]. Most pathogens use a combination of several protein secretion systems for successful infection. The secretion system categorized as Type III plays an essential role in building the organelles that help pathogens arrive at the site of infection (flagella), and once there, deliver effector proteins into the host cells (needles/injectisomes). Bacteria that use Type III secretion (T3S) employ these systems to infect humans, plants and insects. The overall architecture of the basal structure of flagella and injectisomes is similar, as are their substrate recognition and secretion mechanisms [7–12]. A core of eight proteins is highly conserved among the T3S systems found in bacterial flagella and in all known injectisomes. These proteins are part of a multicomponent T3S apparatus. Six of these conserved proteins are integral membrane (IM) proteins that together transport substrates across the membrane and into the periplasm or extracellular space using proton motive force (PMF) as the energy source [13–16]; the other two proteins are part of a cytoplasmic ATPase complex that delivers substrates to the T3S gate. While knowledge of the structure and function of T3S substrates is advanced, that of the highly conserved secretion apparatus itself is not.

In Salmonella enterica and Escherichia coli flagella, FlhA and FlhB are two of the six conserved, essential IM proteins, the other four being FliF (the MS ring protein), FliP, FliQ and FliR; FliO is an essential flagellar protein not found in injectisomes (Fig. 1). The large cytoplasmic domains of FlhA and FlhB serve as a substrate delivery platform that interacts with the cytoplasmic ATPase complex [17]. While FlhA is thought to harbor a proton channel [18], and FlhB is known to regulate secretion specificity [19], not much is known about the function of the other IM proteins [20]. The secretion pore had been hypothesized to pass through the transmembrane portions of FlhA and FlhB; however, this view needs to be reconsidered in light of the nonameric ring structure observed for the cytoplasmic domain of the FlhA homolog MxiA (MxiA_c) from Shigella flexineri injectisomes [21]. Homology predictions suggest that the ring/torus feature of this protein is conserved in all T3S systems, and the MxiA_c torus can be superimposed on the density attributed to FlhA_c in in situ cryotomograms of S. enterica flagellar T3S. The transmembrane portion of FlhA is likely also toroidal and could house FlhB, and FliOPQR as depicted in Fig. 1. Of the latter five proteins, FlhB regulates the switch in specificity of secreted substrates in response to a secreted ‘ruler’ protein FliK: when the rod-hook structure reaches a specific length, the T3S system stops secreting rod-hook substrates and starts secreting filament substrates [22, 23]. The secretion pore itself may be comprised by a subset of the FliOPQR proteins (Fig. 1).

Figure 1. Schematic of the T3S architecture of a Salmonella flagellum.

Flagella biogenesis is thought to begin with self-assembly of the MS ring in the inner membrane (IM), followed by assembly of the attached cytoplasmic C-ring [8, 19]. The T3S apparatus is inserted in the IM patch enclosed by the MS ring, and secretes flagellar substrates that will form the periplasmic rod, short external hook and long flagellar filament (not shown). FlhA (likely a nonameric ring) is thought to conduct protons that drive secretion. FlhA and FlhB have large cytoplasmic domains that interact with an ATPase complex, which delivers substrates to the secretion gate or pore. The identity of the secretion pore is not known, but is likely constituted by a subset of the FliOPQR proteins. FlhE is a soluble, periplasmic protein. How it participates in flagellar assembly is unknown. See text.

FlhE is part of the flagellar regulon in all bacteria where it is found, but closely linked to genes encoding FlhA and FlhB in the majority of these bacteria [24]. While T3S systems of injectisomes contain homologs of FlhA and FlhB, they do not have FlhE. The following observations suggest that FlhE is an important component of the flagellar T3S. First, flhE is co-transcribed with flhB and flhA in S. enterica 24]. Second, the absence of FlhE results in a proton leak through the flagellar system in both E. coli and S. enterica [25]. Third, FlhE is secreted into the periplasm and found included in the basal body lumen, and is thus ideally located to regulate T3S secretion from the periplasmic side [25] (Fig. 1). Here we report the 1.5 Å-resolution structure of FlhE and discuss its potential role.

Results and Discussion

Purification of FlhE-His₆ from S. enterica periplasm

The flhE gene product of S. enterica consists of 130 amino acids. The first 16 residues encoding the signal sequence are removed as the protein is secreted into the periplasm [26] (Fig. S1). The functional 114 residue FlhE with a C-terminal His₆ tag was overexpressed from a plasmid, isolated from S. enterica periplasm by an osmotic shock procedure, and purified over a nickel column as described in Methods (Fig. S2). The protein and its selenomethionine derivative readily crystallized (Figs. S3-4).

Overall structure

FlhE-His₆ crystallized in space group P2₁2₁2₁, with cell dimensions a, b, c = 25.4, 36.8, 111.5 Å and 1 molecule per asymmetric unit (V_M = 3.1 Å/Da). Initial phases for the structure were determined by single-wavelength anomalous diffraction (SAD) from selenomethionine-labeled FlhE; anomalous signal was present to 2.8-Å resolution. Two selenium sites were located. With the initial phases and 1.5 Å-resolution native data, a 113-residue model was built by Phenix using the AutoBuild protocol. The final 1.5-Å structure consists of 115 residues and includes the first histidine residue of the C-terminal His₆ tag (Figures 2A and S1). The crystal packing of FlhE suggests that it is monomeric. Crystallographic data are shown in Table 1, and a typical region of the 2F_o-F_c electron density map is shown in Figure 2B.

Figure 2. FlhE structure.

A) Stereo diagrams of two sides of the β-sandwich. The invariant disulfide (between C59 and C64) lies within a shallow, hydrophobic groove. B) Representative electron density is shown (simulated annealing omit, contoured at 1.0 rmsd) around β8. Residues L86 and V92 are labeled.

Table 1.

Crystallographic Data and Refinement Statistics

	native
Space group	P212121
Cell constants	a=25.4, b=36.8, c=111.5 Å
Resolution (Å) (outer shell)	50.-1.51 (1.54-1.51)
Rmerge (%) (outer shell)	0.063 (0.263)
<I/σI> (outer shell)	17.3 (6.2)
Rpim[1] (outer shell)	0.034 (0.107)
CC1/2[2] (outer shell)	0.997 (0.966)
Completeness (%) (outer shell)	92.3 (80.2)
Unique reflections	15,949
Redundancy	9.6 (7.7)
# of residues	115
# of protein atoms	855
# of solvent atoms	86
Rworking	0.200
Rfree	0.226
Average B factor for protein atoms (Å2)	13.5
Average B factor for solvent atoms (Å2)	21.4
rms deviation from ideality   bonds (Å)   angles (°)	0.007 1.217
Ramachandran plot
% of residues in favored region	98.2
% of residues in additional allowed region	1.8

Values in parentheses correspond to highest resolution shell

^[1]Weiss MS. Global indicators of X-ray data quality. J Appl Crystallography 2001;34:130-5.

^[2]Karplus PA, Diederichs K. Linking crystallographic model and data quality. Science. 2012;336:1030-3.

FlhE is ~20 × 20 × 45 Å in size and possesses a distinct β-sandwich fold; there are no known sequence homologous structures. The β-strand connectivity of FlhE is most similar to that of a jellyroll. Each of its 2 β-sheets contains anti-parallel β-strands, and one β-sheet is larger than the other. There are several conserved hydrophobic residues which form the hydrophobic core of the molecule between the two β-sheets: W6, M37, V40, W42, F77, L102, V109, and V111 from the large sheet and L54, L58, L67, L86, and F88 from the small sheet. These conserved residues are shown in Fig. 3.

Figure 3. Conserved features of FlhE.

The FlhE sequence from Salmonella enterica is compared with the FlhE’s of 33 other bacteria through a sequence alignment [47]. Each contains a periplasmic signal sequence that is cleaved when FlhE enters the periplasm. Residue numbering and secondary structure are based on the S. enterica FlhE structure (fifth in the alignment). Residues that are at least 80% conserved are indicated by orange arrows and displayed orange in the ribbon diagram. The disulfide-containing sequence (CX_4-7C) is indicated both in the alignment and in the ribbon diagram. The "variable colors" in the alignment are the default colors for the program ClustalX, used to make the figure. The colors codes, followed by the residue type and conservatism, are as follows: Blue, hydrophobic (ACFHILMVWY), >60%; Magenta, negatively charged (DE), >50%; Red, positively charged (KR), >60%; Green, polar (STQN), >50%; Cyan, aromatic (FYW), >50%; Pink, cysteines, >85%: Orange, glycines, >85%; Yellow, prolines, >85%.

Ancient scaffold

Results from a search for similar structures in the Protein Data Bank (PDB) using the Dali-server [27] revealed that FlhE is distantly related to other known structures. The most similar structural homologs are the non-catalytic P-domain of the kexin family of serine proteases (PDB Code 3HJR [28], 2ID4 [29], 1P8J [30]) with Dali Z-scores of 12.6, 11.5, and 10.9. The rms distance of corresponding α-carbons of FlhE compared with the P-domain of the kexin-like serine protease from Aeromonas sobria (PDB code 3HJR) is 2.1 Å. A superposition of the two proteins is shown in Fig. 4A. The β-sandwich folds of the two proteins are quite similar; but the A. sobria protein (ASP) P-domain does contain several extended loops between β–strands, some of which contain additional secondary structure (Fig. 4). The sequence identity between FlhE and the ASP P-domain is only 14%, but hydrophobic core residues are conserved indicating that FlhE is a distant evolutionary relative of the kexin P-domain. These residues are highlighted in the structure-based sequence alignment in Fig. 4B. The role of the P-domain is not well understood, but it does have a surface hydrophobic patch which is involved in contact with the subtilisin-like catalytic domain of these proteases. This hydrophobic patch is not conserved in FlhE, indicating that FlhE does not have a corresponding catalytic subunit. Unlike these structural homologs, FlhE is a single domain protein.

Figure 4. Comparison of FlhE with structural homologs.

A) The α-carbon trace of FlhE (colored green) is superimposed with that of the P-domain of kexin-like ASP (blue) and the brome mosaic virus coat protein (red) in this stereo picture. The N and C-termini of FlhE are labeled. B) The structure-based sequence alignment of FlhE with the P-domain of kexin-like ASP and the brome mosaic virus coat protein. The sequence numbering and secondary structural elements of FlhE are shown on top. The secondary structural elements of the ASP P-domain and the virus coat protein are shown at the bottom. Identical hydrophobic core residues are highlighted in black. Additional conserved hydrophobic core residues are highlighted in gray.

The next most similar group of structural homologs is a domain of virus coat proteins, e.g. PDB code 1YC6[31] with Z-score 8.1, rmsd of 3.2 Å, and 6% sequence identity. The superposition and the structure-based sequence alignment of FlhE with the coat protein of the brome mosaic virus (PDB code 1YC6) are shown in Fig. 4. Most of the FlhE hydrophobic core residues are conserved, indicating that FlhE and these virus coat protein structural homologs have a common evolutionary ancestor.

Clearly, the β-sandwich fold of FlhE is an ancient scaffold that has been adapted by nature to various purposes. The distant structural homologs do not reveal the exact biological role of FlhE but do provide some clues. Its role is likely structural in nature rather than catalytic, and it may involve interactions with other protein molecules.

Surface of FlhE

The curvature of both β-sheets faces the same direction, and a groove is formed on the surface of the smaller β-sheet. A disulfide bond is located in this groove, formed across the only β-hairpin of the structure (β5/β6) by an invariant CX_4-7C motif. The disulfide and neighboring moieties (V89 side chain and nonpolar portions of R57 and R87) create a hydrophobic patch in the center of the groove. Residues that are identical in at least 80% of sequences in an alignment of 35 FlhE sequences are indicated with orange arrows in Fig. 3, including surface residues that could interact with other proteins (e.g., R43, G53, C59, R63, C64, G72, G96). The lengths of the β7/β8 and β8/β9 loops are highly conserved. The β3/β4 and β8/β9 loops may be rigidified through a close association of the β8/β9 loop with the R18 side chain (Figure 2A).

Role of FlhE

FlhE is one of the last components of the flagellar assembly apparatus for which a role has not been elucidated. Its placement in the T3S operon flhBAE, and the phenotypes associated with its absence, suggest that FlhE regulates T3S secretion. One phenotype is acidification of the cytoplasm, suggesting that FlhE might be an architectural element of a channel or a pore; absence of FlhE would then cause a proton leak [25]. We had originally speculated that FlhE might plug the proton channel in FlhA [18], but given that there are nine FlhA subunits [21], this is not a likely scenario. However, it might function as a plug for the secretion pore, likely comprised of a subset of FliOPQR proteins (Fig. 1). An alternative explanation of the proton leakage phenotype is that FlhE serves as a chaperone for all or a subset of the T3S IM proteins, or for the secreted flagellar components, resulting in improper assembly of either the T3S apparatus or the flagellar structure in its absence.

A second flhE phenotype is a premature switch in secretion specificity from early (rod-hook) to late substrates (filament) in a Salmonella mutant background where the rod does not exit the outer membrane [32]. Since this switch normally requires a timely interaction between the C-terminus of FliK and that of FlhB, one interpretation is that this interaction is hastened by acidification of the cytoplasm [25]. Alternatively, FlhE could regulate FliK/FlhB interactions through docking to the periplasmic portion of FlhB transmembrane loops, or by directly binding FliK.

A third phenotype associated with absence of FlhE is a change in the outer membrane properties of S. enterica 24], and an increased expression of genes activated by the Rcs signaling pathway [25, 33]. These two observations could be related, and likely a secondary consequence of acidification of the cytoplasm, which might trigger a cellular stress response.

The structure of FlhE itself is suggestive of a substrate chaperone. For example, it shares characteristics with cytoplasmic effector chaperones from T3S systems (e.g., SycE, SycH, SicP), which are relatively small (<15 kDa) and bind unfolded substrates through a hydrophobic groove formed by an antiparallel β-sheet [34–36], reminiscent of the hydrophobic patch surrounding the disulfide in FlhE, which lies within a groove on the surface of the smaller β-sheet (Fig. S5). The cytosolic flagellar chaperones (FlgN for FlgK and FlgL, FliT for FliD, and FliS for FliC) have a different architecture; however, FliS is known to bind an unfolded region of its substrate FliC [37]. While these parallels hint at a role for FlhE as a chaperone, they do not exclude a structural role such as a plug for the secretion pore. With the FlhE structure now at hand, we can begin to investigate its specific function.

Materials and Methods

Purification of FlhE-His₆ from S. enterica periplasm

FlhE-His₆, expressed from pTrc99a (pMJ68; [38]) was purified for crystallization by an osmotic shock procedure that releases periplasmic proteins into the medium [13], with the following modifications. An overnight culture of S. enterica flhE mutant cells (ST004) containing pTrc99a-flhE-His was diluted 1:100 in a 2 L flask with a working volume of 1 L LB supplemented with ampicillin and grown at 37 °C for 2 h. To induce FlhE-His₆, IPTG, at the final concentration of 50 μM, was added. After 3.5 h, cells were centrifuged at 4000 × g and 4 °C for 20 min. All procedures for FlhE-His₆ purification and concentration were performed at 4 °C. Cells (3.5 g of wet weight) were resuspended in 100 mL solution containing 30% sucrose and 30 mM Tris-HCl (pH 8.0), and 500 mM EDTA was added dropwise to 1 mM final concentration. The cells were incubated for 15 min in a rotation carousel and centrifuged at 8000 × g for 25 min. The majority of FlhE-His₆ was recovered in this supernatant (Fig. S2). The supernatant was saved, and the pellet was resuspended in 100 mL of cold 5 mM MgSO₄ solution. The suspension was incubated and centrifuged as described above. Supernatants (total 200 mL) from the sucrose and the MgSO₄ solutions were transferred to a Spectra/Por®3 Dialysis Membrane (SPECTRUM®LABORATORIES) and dialyzed twice against 2 L of buffer A (50 mM NaH₂PO₄300 mM NaCl, 10 mM imidazole, pH 8.0) for 2 h before continuing with the purification. Five mL of 50% Ni-NTA matrix (Ni-NTA Fastflow, Qiagen) equilibrated in buffer A were added to a column (Flex-Column, Kontes).

The supernatant was passed through the column 3 times, and the matrix was extensively washed 4 times with 15 mL buffer A containing 50 mM imidazole. FlhE-His₆ was eluted 6 times with 5 mL of buffer A containing 250 mM imidazole. Eluted fractions were dialyzed against 1 L of buffer B (20 mM Tris-HCl (pH 7.0), 10% (v/v) glycerol) twice and concentrated to 7 mg mL⁻¹ for crystallization by a centrifugal filter unit (Amicon® Ultra-15, MILLPORE) and a centrifugal filter device (Microcon, MILLPORE).

Size-exclusion chromatography (SEC)

To determine the molecular weight of FlhE, and confirm whether it was monomeric in solution as suggested by the crystal packing data, the protein was fractionated on Superdex and Biogel SEC columns in two buffers used during purification (Tris-glycerol and NaH₂PO₄-NaCl). FlhE did not elute from either column. Next we tried Multi-Angle Light Scattering (MALS) as described [39]. However, this method involved an SEC step using TSK-gel column; FlhE did not elute from this column either. These results indicate that FlhE is a ‘sticky’ protein.

Purification of FlhE containing selenomethionine

To produce selenomethionine-labeled FlhE-His₆cells were prepared as described [40] with the following modifications. ST1004 containing pJM68 was grown overnight in LB, and diluted 1:25 in 100 mL of M9 minimal medium. For the M9-medium preparation, 5× M9 minimal salts was prepared using 33.9 g Na₂HPO₄, 15 g KH₂PO₄, 5 g NH₄Cl, and 2.5 g NaCl in 1 L dH₂O. After being autoclaved, 200 mL of 5× M9 salts was diluted to 800 mL of autoclaved dH₂O and supplemented with MgSO₄ (2 mg mL⁻¹), CaCl₂ ( 0.1 mg mL⁻¹), 0.4% glucose, and ampicillin (100 μg/ml). The overnight culture from M9 minimal medium was diluted 1:100 in a 2 L flask with a working volume of 1 L M9 medium supplemented with ampicillin and grown at 37 °C for 6 h until OD₆₀₀ reached 0.6. Before FlhE-His₆ induction, amino acid mix (100 mg of lysine, phenylalanine, and threonine, 50 mg of isoleucine, leucine, and valine, and 60 mg of seleno-methionine) was added to the culture and incubated for an additional 15 min. The addition of the amino acid mix inhibits the methionine synthesis pathway by targeting an aspartokinase, thus enabling selenomethionine to be incorporated. FlhE-His₆ induction and purification were performed as described above.

Protein crystallization and data collection

FlhE-His₆ was initially screened for crystallization through the sitting-drop vapor diffusion method using a commercially available 96-screen kit (Crystal Screen HT, Hampton Research). Crystallization drops were set up by a Phoenix crystallization robot (Art Robbins Instruments) with two 96-well protein crystallization plates (Intelli-Plate™ 96-3 LVR, Hampton Research). Each protein sample drop was prepared by mixing 0.1 μL reservoir solution with either 0.1 μL or 0.2 μL protein solution. Drops were equilibrated against 50 μL reservoir solutions. The plates were stored at both 4 °C and room temperature for two weeks. Microcrystals were then grown using the hanging drop method from one of the identified crystallization conditions. Each drop was set up on a cover-slide (Siliconized Glass Circle Cover Slide, Hampton Research) with 1 μL reservoir solution and 1 μL protein solution, equilibrating against a reservoir of 500 μL 1.5 M NaCl, 10% ethanol. The plates were sealed with the cover slides and subsequently stored at 4 °C. To increase the size of crystals, a streak-seeding method was performed. Protein and reservoir solution were mixed 1:1 to a final drop volume of 2 μl on a slide. By using a microneedle (CrystalProbe, Hampton Research), several microcrystals were streaked on the drop, and the plate filled with 500 μL of the reservoir solution was sealed with the slide. After 24 h, crystals large enough (~ 0.1 mm in one dimension) to be used for data collection were observed (Fig. S3). Prior to data collection, a FlhE-His₆ crystal was transferred briefly to a drop of the 1.5 M NaCl, 10% ethanol solution containing 20% glycerol for cryoprotection. The crystals mounted in a cryoloop (Hampton Research, Laguna Niguel, CA) were flash frozen in liquid nitrogen.

X-ray diffraction data from a native crystal were collected at 100 K at the Advanced Light Source (ALS) beamline 5.0.2 at the Lawrence Berkeley National Laboratory. Data from a selenomethonine crystal were collected at 100 K at ALS beamline 5.0.3. Diffraction images were processed and data were reduced using HKL2000 [41].

Structure determination and analysis

Initial phases were determined by single-wavelength anomalous diffraction (SAD) using the selenomethionine data with Phenix [42]. An electron density map calculated with native data and the initial phases was used to build the first model of 53 residues with the automated building capability of Phenix. Additional rounds of automatic building and refinement produced a model containing 113 residues with the correct sequence and had R_work = 0.21. Additional model building was carried out using Coot [43]. Additional refinement of models was done with Phenix. To facilitate manual building, an F_o-F_c difference map and a 2F_o-F_c map (σA-weighted to eliminate bias from the model [44]) were prepared. 5% of the diffraction data were set aside throughout refinement for cross-validation [45]. MolProbity was used to determine areas of poor geometry and to make Ramachandran plots [46]. A search for structural homologs was done with DALI [27]. Coordinates of the refined model of FlhE have been deposited in the PDB with accession code 4QXL.

Highlights.

• FlhE plays an important role in flagella biosynthesis

• A 1.5 Å-resolution structure for Salmonella FlhE is reported

• Potential roles of FlhE as a chaperone or plug are discussed