Crystal structure of the flagellar accessory protein FlaH of Methanocaldococcus jannaschii suggests a regulatory role in archaeal flagellum assembly

Abstract

Archaeal flagella are unique structures that share functional similarity with bacterial flagella, but are structurally related to bacterial type IV pili. The flagellar accessory protein FlaH is one of the conserved components of the archaeal motility system. However, its function is not clearly understood. Here, we present the 2.2 Å resolution crystal structure of FlaH from the hyperthermophilic archaeon, Methanocaldococcus jannaschii. The protein has a characteristic RecA‐like fold, which has been found previously both in archaea and bacteria. We show that FlaH binds to immobilized ATP—however, it lacks ATPase activity. Surface plasmon resonance analysis demonstrates that ATP affects the interaction between FlaH and the archaeal motor protein FlaI. In the presence of ATP, the FlaH‐FlaI interaction becomes significantly weaker. A database search revealed similarity between FlaH and several DNA‐binding proteins of the RecA superfamily. The closest structural homologs of FlaH are KaiC‐like proteins, which are archaeal homologs of the circadian clock protein KaiC from cyanobacteria. We propose that one of the functions of FlaH may be the regulation of archaeal motor complex assembly.

Short abstract

Interactive Figure 1

Introduction

Flagellation is widespread in the prokaryotic world. Both bacteria and archaea use rotating flagella for swimming in liquid environments. However, in spite of functional similarities, bacterial, and archaeal motility systems are not homologous. Instead, archaeal flagella are more closely related to other bacterial surface structures called type IV pili.1, 2, 3, 4

In Archaea, all flagella‐related proteins are encoded in one fla operon. Typically it starts with one or more (up to 5) flagellin genes (flaA and/or flaB) followed by several flagella‐associated genes (flaC to flaJ). Genes flaF, G, H, I, J are conserved throughout the kingdom of Archaea. Each of these genes is required for flagellar assembly and motility.5, 6, 7, 8, 9 Three proteins, FlaH, FlaI, and FlaJ, are thought to form a core structure involved in flagellar assembly and rotation.1 Two of them, FlaI and FlaJ, are homologs to the components of bacterial type IV pili: PilT/PilB ATPases and the inner membrane protein PilC, respectively.3, 10, 11

FlaH contains a highly conserved Walker A motif and a less conserved Walker B motif.12 Both motifs are required for ATP binding and hydrolysis, implying ATPase activity of FlaH. Although FlaH has no significant sequence similarity to any bacterial proteins, it was predicted to belong to the RecA family, which includes both archaeal and bacterial ATPases that use the energy of nucleotide hydrolysis to perform mechanical work.3, 13, 14 The exact role of FlaH in the archaeal motility system is poorly understood. Deletion of the flaH gene results in non‐motile cells lacking any flagella structures.6, 8, 9 However, such ΔflaH mutants express flagellin in its mature form without leader peptide at the same level as wild‐type cells, implying that flaH deletion does not affect the processing of flagellin. These observations suggest involvement of FlaH either in flagellin secretion and/or flagellum assembly.

Cell fractionation analysis shows that both FlaI and FlaH are localized in the cell membrane fraction, implying that these two proteins might form a complex.12 Recently, interaction between FlaH and FlaI of Sulfolobus acidocaldarius in vitro has been confirmed by various biochemical methods.15

Determination of protein structure is a first step to understanding its function. So far, our knowledge about the structures of the components of archaeal motility system was limited only to one organism, namely crenarchaeon S. acidocaldarius.16, 17, 18 No structural information on motility systems from other archaeal phyla is available. Here, we report the crystal structure of the FlaH protein of Methanocaldococcus jannaschii, a member of one of the major archaeal phylum ‐ Euryarchaeota. Our results provide new insights into the role of FlaH in archaeal motility.

Results and Discussion

We solved the crystal structure of the FlaH protein by multi‐wavelength anomalous diffraction using selenomethionine‐labeled protein as described previously.19 Crystals belong to the P3₁21 space group with three protein molecules in an asymmetric unit. The final model consists of residues 5–233 for chains A and B, and residues 5‐230 for chain C. Refinement statistics are summarized in Table 1.

Table 1.

X‐ray refinement statistics

Resolution (Å)	50.0–2.2
R work	20.6
R free	23.3
No. of atoms
Protein	5378
Ligand	71
Water	70
Average B‐factor
Protein	68.9
Ligand	61.9
Water	53.6
R.m.s deviations
Bond lengths (Å)	0.015
Bond angles (°)	1.854
Ramachandran plot (%)
Favoured	97.3
Additional allowed	2.7
Outliers	0

The protein FlaH has a well‐known RecA‐like fold that was previously found in many other proteins, both archaeal and bacterial. FlaH consists of a central, mostly parallel, twisted β‐sheet surrounded by several α‐helices (Fig. 1). A Walker A motif, or phosphate binding loop (P‐loop), is located between β3 and α2, and a Walker B motif lies on β6. The highly conserved Asp127 of Walker B motif forms hydrogen bond with Ser41 of Walker A motif. In the RecA protein this interaction coordinates position of Mg²⁺ ion which is important for ATP hydrolysis.20 The Asp127‐Ser128 peptide bond of the Walker B motif is in the cis‐conformation that has also been observed in other RecA superfamily members and seems to be a common feature of all RecA‐like fold proteins.20, 21, 22

Figure 1.

Ribbon representation of M. jannaschii FlaH. The protein has typical RecA‐like fold with Walker A and Walker B motifs located between β3 and α2, and on β6, respectively. Sulfate ions bound to P‐loop (Walker A) are shown as ball‐and‐stick models. An interactive view is available in the electronic version of the article.

FlaH crystals have been obtained in the absence of ATP or its analogues. Co‐crystallization or post‐soaking attempts were unsuccessful. No electron density corresponding to the ligand molecule was found in the final electron density map. Two sulfate ions, likely from crystallization solution that contained 0.65M Rb₂SO₄, were located in the P‐loop of the protein (Fig. 1). At high sulfate concentration, it is feasible that sulfate anions may compete with ATP phosphates for binding to the protein. To confirm association of FlaH with ATP we checked its binding to ATP‐immobilized agarose. FlaH binds to immobilized ATP and can be specifically eluted by addition of free ATP (Fig. 2). The ATP‐binding site is the most conserved region on the FlaH surface (Fig. 3), suggesting that ATP binding is an essential property of FlaH proteins.

Figure 2.

Binding of M. jannaschii FlaH to immobilized ATP. 10 μg of FlaH were incubated with 20 μL of ATP‐agarose in 20 mM HEPES, pH 8.0, 100 mM NaCl, 5 mM MgCl₂ for 2 h at 30°C. Samples were analyzed by SDS‐PAGE. 1, loaded protein; 2, supernatant after incubation; 3, ATP elution; 4, elution with SDS‐PAAG loading buffer. Agarose was used as a blank control.

Figure 3.

Conserved properties of FlaH. A: Multiple sequence alignment of FlaH from M. jannaschii and FlaH proteins from different archaea species. Identical residues are highlighted in red; similar residues are in red letters. The secondary structure elements of M. jannaschii FlaH are shown above the alignment. B: Evolutionarily conserved residues of FlaH. Residues are colored according to conservation in amino acid sequences of 150 different FlaH homologs.

We tested FlaH for the ability to hydrolyze ATP. ATPase activity of FlaI S. acidocaldarius shows temperature dependence;23 therefore, the FlaH activity experiment has been performed at several different temperatures. However, under all temperatures tested no ATPase activity of the protein was detected (data not shown). We suggest that FlaH may need additional factor(s) to enable ATPase activity, or that perhaps it lacks ATPase activity but ATP acts as a co‐factor, affecting FlaH function.

Many RecA‐like proteins can assemble into higher oligomeric structures.13 We examined the oligomeric state of FlaH by cross‐linking and by gel‐filtration chromatography. Both methods gave similar results: in solution FlaH exists mostly as monomer; Mg and ATP do not affect its oligomeric state (Fig. 4). This finding may explain why FlaH does not show ATPase activity. It is known that oligomerization is essential for the activity of proteins with a RecA‐like fold, since ATP hydrolysis requires residues from adjacent RecA‐like domains.

Figure 4.

Analysis of oligomerization states of M. jannaschii FlaH. A: Cross‐linking product of FlaH analyzed by SDS‐PAGE. 10 μg of FlaH with protein concentration of 1 mg/mL were cross‐linked with 0.2% glutaraldehyde for 1 h at 30°C. B: Elution profiles of FlaH in the presence or absence of Mg‐ATP on a Superdex 200 10/300 gel‐filtration column. The high peak at ∼22 mL corresponds to ATP. The data indicate that FlaH in solution exists mostly as a monomer, and neither MgCl₂ nor Mg‐ATP affect the oligomeric state of the protein.

It has been shown previously that FlaH interacts with the ATPase FlaI, another archaeal motility system protein.15 To assess the effect of ATP on the interaction of these two proteins, we analyzed the FlaI‐FlaH interaction in the presence or absence of ATP using surface plasmon resonance. Monomeric FlaI of M. jannaschii was immobilized on the Biacore CM5 sensorchip surface and allowed to bind FlaH from the mobile phase. The obtained sensorgrams were fitted with a two‐state reaction model (Fig. 5) and an overall equilibrium dissociation constant (K_D) was calculated. In the presence of 2 mM ATP, affinity between FlaH and FlaI was significantly reduced: K_D is 1.14 ± 0.08 × 10⁻⁷ M in the absence of ATP, and it is 4.41 ± 0.48 × 10⁻⁷ M in the presence of ATP. FlaI is an ATP‐binding protein.23 This experiment does not discriminate which protein, FlaH or FlaI (or both), is affected by ATP binding. Nevertheless, our analysis clearly demonstrates that in the presence of ATP, the interaction between FlaH and FlaI becomes weaker. This decreased affinity between FlaH and FlaI in the presence of ATP may be a requirement for functioning flagella. Until the structure of the archaeal basal body is determined, an association of FlaH and FlaI as part of the motor complex remains speculative. However, a current model suggests that FlaI, which has a dynamic conformation and is responsible for ATP hydrolysis, is a key protein involved in flagellar rotation.16

Figure 5.

Kinetic analysis of FlaH binding to immobilized FlaI in absence (A) or presence of 2 mM ATP (B). Concentrations of each FlaH injection are noted. Sensorgrams were fitted with a two‐state reaction model and an overall equilibrium dissociation constant (K _D) was calculated using Biacore T200 Evaluation Software. Analysis shows that in the presence of ATP, affinity between FlaH and FlaI is decreased.

A database search using the DALI24 and SSM servers25 revealed a number of DNA‐binding proteins belonging to the RecA superfamily, which are structurally similar to FlaH. None of these proteins was previously known to participate either in the archaeal motility system or in the bacterial type IV pili system. Besides FlaH from S. acidocaldarius, the closest structural homologs of FlaH with DALI Z‐scores above 20 were two archaeal KaiC‐like proteins: PH0186 from Pyrococcus horikoshii 26 and SSO2452 from Sulfolobus solfataricus,22 and N‐terminal domain (KaCI) of KaiC protein from cyanobacterium Synechococcus elongates.27 FlaH superimposes onto PH0186 with a root mean square deviation (r.m.s.d.) of 1.96 Å for 185 aligned Cα atoms, onto SSO2452 with r.m.s.d. of 1.93 Å for 180 aligned Cα atoms, and onto KaiCI with r.m.s.d. of 2.15 Å for 188 aligned Cα atoms (Fig. 6). Central core elements that form ATP binding sites are very similar in all these proteins, while peripheral α‐helices show larger deviations.

Figure 6.

Superposition of M. jannaschii FlaH (orange); P. horikoshii KaiC‐like protein PH0186 (blue), S. solfataricus SSO2452 (green), and KaiCI S. elongatus (gray).

KaiC is a crucial regulator of circadian rhythm in cyanobacteria.28 Cyanobacterial KaiC consists of two RecA domains, CI and CII, joined together head‐to‐tail.29 Homologs of KaiC were found in almost all major archaeal taxa.30, 31 But in contrast to cyanobacteria, most archaeal KaiC homologs are single‐domain proteins. Shorter KaiC homologs consistently have higher similarity to KaiCI.31 The evolutionary history of KaiC is uncertain. Nevertheless, it is commonly accepted that the ancestral KaiC was a single‐domain protein that may have differentiated into groups of proteins with diverse functions. The function of KaiC‐like proteins in archaea remains obscure: no circadian rhythm was found in archaea. However, their distribution and abundance (many archaea have more than one kaiC gene) suggests an important role of these proteins. Similarity of archaeal KaiC‐like proteins to cyanobacterial KaiC and other DNA‐binding proteins of RecA superfamily implies that archaeal KaiC‐like proteins may also act on DNA. Indeed, in vitro experiments demonstrated that SSO2452 could bind tightly to single‐stranded DNA.22 Archaeal flagella‐related genes are clustered into a single fla operon. For methanogenic archaea multiple polycistronic mRNA transcripts were detected.1, 5, 12 All of them originate from one promoter located upstream of the flagellin gene—flaB1. It is interesting that mRNAs encoding the major structural proteins—flagellins—are much more abundant than transcripts encoding other flagellum‐associated proteins. However, it is not clear how the process of transcription in archeal flagellar genes is regulated. So far no transcriptional regulators have been found in Euryarchaeota. Possible DNA‐binding properties of FlaH have not been investigated, which would be an interesting topic for future research.

Materials and Methods

Structure determination

Details of the purification of M. jannaschii FlaH, crystallization and data collection have been described.19 Briefly, the FlaH structure was solved by multiwavelength anomalous diffraction (MAD) using selenomethionine‐labeled protein. Diffraction data were indexed, integrated and scaled with iMOSFLM.32 Heavy‐atom search, phasing and density modification were performed with AutoSol33 in the PHENIX program suite.34 The initial protein model was built automatically with Buccaneer35 from the CCP4 package.36 The model was refined through an iterative combination of refinement with Refmac537 and manual model building in COOT.38 In the final round, TLS refinement39 was carried out treating each FlaH monomer in an asymmetric unit as an individual TLS group. The quality of the final model was validated with MolProbity.40 All protein structure figures were prepared with PyMOL (http://www.pymol.org). Most of software was supported by SBGrid.41

Bioinformatics

The alignment figure was prepared with Crystal Omega 42 and ESPript3.43 The estimation of evolutionary conservation of amino acids in the FlaH protein was done using ConSurf.44

ATP binding assay

20 µL of ATP‐agarose were equilibrated with binding buffer (20 mM HEPES, pH 7.5, 100 mM NaCl, 5 mM MgCl₂). 10 µg of the protein in binding buffer were added to the ATP‐agarose, and incubated at 30°C for 2 h. The agarose was washed five times with 500 µL of washing buffer (20 mM HEPES, pH 7.5, 500 mM NaCl, 5 mM MgCl₂). Bound protein was eluted with 20 mM HEPES, pH 7.5, 100 mM NaCl, 5 mM ATP. Finally, ATP‐agarose was washed with SDS‐loading buffer. Samples were analyzed on SDS‐PAGE.

ATPase activity

FlaH ATPase activity was measured using an ATPase/GTPase Activity Assay Kit (Sigma‐Aldrich, USA) accordingly to the manufacturer's protocol. Measurements were performed at 40, 50, 60, and 70°C.

Chemical cross‐linking

Totally, 10 μg of FlaH (protein concentration–1 mg/mL) were cross‐linked with 0.2% glutaraldehyde in 20 mM HEPES, pH 8.0, 100 mM NaCl supplemented with either 1 mM ATP or 5 mM MgCl₂/1 mM ATP. The reaction was carried out for 1 h at 30°C. The treatment was quenched with 200 mM Tris‐HCl, pH 8.0. Cross‐linking products were analyzed by SDS‐PAGE.

Analytical gel filtration

Purified FlaH was incubated with 1 mM ATP and 5 mM MgCl₂ in a buffer containing 20 mM HEPES, pH 8.0, 100 mM NaCl at 30°C for 1 h. The sample was then applied to a Superdex 200 10/300 gel filtration column (GE Healthcare) pre‐equilibrated with the same buffer. Ferritin (440 kDa), alcohol dehydrogenase (150 kDa), BSA (67 kDa), carbonic anhydrase (29 kDa), and ribonuclease A (13.7 kDa) were used as size markers.

Purification of FlaI

The M. jannaschii FlaI gene (GeneID: 1453798) was cloned into the pTXB1 expression vector (New England BioLabs), generating an in‐frame fusion with intein and with the chitin‐binding domain (CBD).

The recombinant vector containing the FlaI‐intein‐CBD fusion protein gene was transformed into Escherichia coli strain Rosetta (DE3) (Novagen). Transformed cells were cultured at 37°C in 5 l Luria‐Bertani medium supplemented with 50 μg mL⁻¹ ampicillin and 34 μg mL⁻¹ chloramphenicol. Expression was induced with 0.2 mM isopropyl β‐D‐1‐thiogalactopyranoside at an OD₆₀₀ of 0.5–0.6. Induction was performed at 16°C overnight. Cells were harvested by centrifugation at 8,000 g for 15 min and suspended in 200 mL buffer A (40 mM HEPES, pH 8.0, 500 mM NaCl, 15% glycerol). Suspended cells were sonicated on ice. Cell debris was removed by centrifugation at 30,000 g for 30 min at 5°C. The clarified supernatant was applied to a 20 mL Chitin Beads column (New England BioLabs) equilibrated with buffer A. The column was washed with 200 ml buffer A and then with 100 mL of buffer A supplemented with 100 mM DTT to induce on‐column cleavage. FlaI protein was eluted with buffer A after 24 h incubation at 4°C. Eluted protein was dialyzed against 40 mM HEPES, pH 8.0, 50 mM NaCl, and applied to a Mono Q column (GE Healthcare) equilibrated with the same buffer. The protein was eluted from the column with a linear gradient of NaCl from 0.05 to 1M in 20 mM HEPES, pH 8.0. Pure protein was applied to a Superdex 200 10/300 gel filtration column (GE Healthcare) pre‐equilibrated with 10 mM HEPES, pH 8.0, 50 mM NaCl. Fractions corresponding to monomeric FlaI were collected and used for kinetic analysis. To confirm proper folding of monomeric FlaI, hexamerization of the protein in the presence of non‐hydrolysable ATP analogue, AMPPNP, has been verified by negative‐stain electron microscopy (Fig. 7).

Figure 7.

Negative‐stain electron micrograph of M. jannaschii FlaI in the presence of AMPPNP. The expressed protein folds properly into particles of uniform size and shape.

Biacore Analysis

All analyses were carried out on a Biacore T100 (T200 Sensitivity Enhanced) (GE Healthcare). About 1000 response units of FlaI were immobilized on a CM5 chip with amine cross‐linking. FlaH of various concentrations in binding buffer (10 mM HEPES pH 7.4, 150 mM NaCl, 10 mM MgCl₂, 0.05% (v/v) Tween 20) was passed over the sensor surface at a flow rate of 40 µL min⁻¹ for 6 min. Dissociation of the sample was monitored in binding buffer for 12 min. The surface was regenerated using 10 mM Glycine, pH 2.0, 500 mM NaCl. All measurements were performed in triplicate at 25°C. Data were analyzed with Biacore T200 Evaluation software.

Electron Microscopy

FlaI (200 µg mL⁻¹) in 10 mM HEPES, pH 8.0, 50 mM NaCl, 5 mM MgCl₂, 1 mM AMPPNP was incubated for 30 min at 40°C and 4 μL sample was applied onto carbon grids, which had been glow‐discharged immediately before use. The sample was stained by 2% uranyl acetate, pH 5.0. Grids were examined using a JEM‐1230R transmission electron microscope at 100 keV equipped with a Gatan Ultrascan digital camera.

Accession code

Atomic coordinates and structural factors have been deposited in the Protein Data Bank with accession code 4WIA.