Structure and function of a spectrin-like regulator of bacterial cytokinesis

Abstract

Bacterial cell division is facilitated by a molecular machine--the divisome--that assembles at mid-cell in dividing cells. The formation of the cytokinetic Z-ring by the tubulin homologue FtsZ is regulated by several factors, including the divisome component EzrA. Here we describe the structure of the 60-kDa cytoplasmic domain of EzrA, which comprises five linear repeats of an unusual triple helical bundle. The EzrA structure is bent into a semicircle, providing the protein with the potential to interact at both N- and C-termini with adjacent membrane-bound divisome components. We also identify at least two binding sites for FtsZ on EzrA and map regions of EzrA that are responsible for regulating FtsZ assembly. The individual repeats, and their linear organization, are homologous to the spectrin proteins that connect actin filaments to the membrane in eukaryotes, and we thus propose that EzrA is the founding member of the bacterial spectrin family.

ARTICLE

Received 25 Feb 2014 | Accepted 30 Sep 2014 | Published 18 Nov 2014

Robert M. Cleverley1, Jeffrey R. Barrett2, Arnaud Basl1, Nhat Khai Bui1,w, Lorraine Hewitt1, Alexandra Solovyova3, Zhi-Qiang Xu2, Richard A. Daniel1, Nicholas E. Dixon2, Elizabeth J. Harry4, Aaron J. Oakley2, Waldemar Vollmer1 & Richard J. Lewis1

Bacterial cell division is facilitated by a molecular machinethe divisomethat assembles at mid-cell in dividing cells. The formation of the cytokinetic Z-ring by the tubulin homologue FtsZ is regulated by several factors, including the divisome component EzrA. Here we describe the structure of the 60-kDa cytoplasmic domain of EzrA, which comprises ve linear repeats of an unusual triple helical bundle. The EzrA structure is bent into a semicircle, providing the protein with the potential to interact at both N- and C-termini with adjacent membrane-bound divisome components. We also identify at least two binding sites for FtsZ on EzrA and map regions of EzrA that are responsible for regulating FtsZ assembly. The individual repeats, and their linear organization, are homologous to the spectrin proteins that connect actin laments to the membrane in eukaryotes, and we thus propose that EzrA is the founding member of the bacterial spectrin family.

DOI: 10.1038/ncomms6421 OPEN

Structure and function of a spectrin-like regulator of bacterial cytokinesis

1 Institute for Cell and Molecular Biosciences, Newcastle University, Newcastle upon Tyne NE2 4HH, UK. 2 Centre for Medical and Molecular Bioscience, University of Wollongong, Wollongong, New South Wales 2522, Australia. 3 NUPPA, Devonshire Building, Newcastle University, Newcastle upon Tyne NE1 7RU, UK. 4 The ithree Institute, University of Technology, Sydney, New South Wales 2007, Australia. w Present address: Helmholtz Centre for Infection

Research, Helmholtz Institute for Pharmaceutical Research, Saarland University, 66123 Saarbrcken, Germany. Correspondence and requests for materials should be addressed to R.J.L. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 5:5421 | DOI: 10.1038/ncomms6421 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6421

Bacterial cell division is directed by cytoskeletal proteins. The bacterial homologue of eukaryotic tubulin, FtsZ1, polymerises in the presence of GTP2 and is attached to the

membrane at the mid-cell position to form the Z-ring3,4. The hydrolysis of GTP to GDP results in the dissociation of FtsZ polymers into monomers that disengage from the Z-ring5. Hence, the Z-ring is highly dynamic and constantly exchanges FtsZ molecules with the pool of its freely diffusible monomers6. The Z-ring is stabilized and anchored to the cytoplasmic membrane by interacting with FtsA, a bacterial homologue of actin, which is itself capable of polymerizing at the membrane7. FtsZ and FtsA recruit other cell division proteins, including cell wall synthesis enzymes and many proteins of uncharacterized function, to the mid-cell to form the mature divisome8. This large macromolecular assembly synthesizes the cell wall that will form the new poles of the cell after division9.

The polymerization of FtsZ into the Z-ring is a necessary and tightly regulated step in divisome assembly. The short C-terminal tail of FtsZ is essential for self-assembly and for interactions with its regulators3. Z-ring assembly is regulated by multiple, widely conserved cell division proteins, including FtsA, ZapA, the protease ClpX, SepF and EzrA. FtsA both recruits FtsZ laments to the cell membrane and counteracts the clustering of FtsZ laments on it10. ZapA and SepF promote the bundling of FtsZ laments into higher-order structures in vitro1113. SepF additionally stabilizes the Z-ring at mid-cell in vivo14 by binding to both FtsZ and the membrane13.

By contrast, the integral membrane protein EzrA was identied as an inhibitor of FtsZ polymerization and its absence results in the inappropriate formation of multiple FtsZ rings, not only at mid-cell but also at the cell poles15,16. However, EzrA is also one of the rst proteins to be recruited to the mid-cell by FtsZ8, suggesting it has an additional regulatory or scaffolding role in the formation of the divisome. As an inhibitor of FtsZ polymerization, EzrA is assumed to be required to maintain the dynamics of the Z-ring as it constricts during cell division16. Other evidence supports a positive role for EzrA in the assembly of the divisome. In Bacillus subtilis, EzrA recruits the major bi-functional, glycosyltransferase-transpeptidase peptidoglycan synthase, penicillin-binding protein 1 (PBP1, aka PonA)17 to the septum. Bacteria require the synthesis of new peptidoglycan to grow, and peptidoglycan must be deposited as cells divide to ensure a continuum of cell wall around both halves of the dividing cell. Consequently, PBPs that synthesize new cell wall material must localize to areas of cell growth or division. Deletion of ezrA changes the cellular localization patterns of both FtsZ and PBP1, causing marked growth defects and alterations in cell morphology, division and elongation17,18. Generally, ezrA knockouts display a delay in cell division, resulting in cells about twice as long as wild type in B. subtilis13,15,19 and larger, spherical cells in Staphylococcus aureus18. The ezrA knockout has been described as lethal to S. aureus20, Streptococcus pneumoniae21,22 and Listeria monocytogenes23, and thus constitutes a novel target for antibacterial agents. In B. subtilis, the concurrent knockout of ezrA with other divisome components including zapA11, sepF14,17 or gpsB17 causes severe growth defects. Therefore, EzrA plays key roles both as a regulator of cell division and in the maintenance of cell size and shape.

EzrA comprises a single, N-terminal trans-membrane (TM) spanning helix followed by an B60-kDa cytoplasmic domain15, which sequence analysis suggests forms an a-helical coiled-coil. To understand further the role of EzrA during cell division, we report the crystal structure of the intact, cytoplasmic domain ofB. subtilis (Bs) EzrA at 3.95 resolution. Its three-dimensional structure comprises ve, B100 amino-acid long repeats, each of

which forms an anti-parallel triple helical bundle; these are arranged linearly to form a complete semi-circle with a diameter of 120 . The individual repeats and their linear head-to-tail arrangement are reminiscent of the spectrin family of eukaryotic cytoskeletal proteins that associate with actin laments and membrane-bound proteins. Two copies of the spectrin repeats are also observed in a 2.6- resolution structure of a 20-kDa cytoplasmic fragment of EzrA from S. aureus (Sa), which has enabled direct molecular comparisons to be made between eukaryotic spectrin repeats and those found in EzrA. FtsZ-binding sites on the BsEzrA arch are mapped and a binding site that inhibits FtsZ oligomerization is identied. We propose a role for EzrA that is analogous to that of spectrins in eukaryotes, in linking the dynamics of the bacterial cytoskeleton to the membrane-anchored components of the divisome, the proper localization of which is essential for the bacterial cell cycle.

ResultsEzrA has a semi-circular structure. The crystal structure ofB. subtilis EzrA lacking the N-terminal 21 residue TM helix (that is, BsEzrA22562) was solved by selenomethionine single wave

length anomalous dispersion (SAD) to a resolution of 3.95 . Despite extensive crystallization screening, the reliable diffraction limit could not be extended. However, through the use of highly redundant data, the experimentally phased electron density maps of BsEzrA22562 (Supplementary Fig. 1a) enabled the building of

Table 1 | Crystallographic statistics.

BsEzrA22562 SaEzrA24214

Data collectionSpace group R32 P43212

Cell dimensionsa, b, c (), 222.1, 222.1, 183.5 96.1, 96.1, 97.1 a, b, g () 90, 90, 120 90, 90, 90

Resolution () 29.23.95(4.163.95)*

48.562.6(2.722.6)*

Rsym 0.104 (0.351) 0.081 (0.972) Rpim 0.03 (0.266) 0.031 (0.368) I/sI 21.3 (3.6) 15.3 (2.8)

Completeness (%) 99.6 (99.2) 100 (100) Redundancy 22.4 (4.0) 15.3 (2.8) Anomalouscompleteness (%)

99.2 (95.7) NA

Anomalous redundancy (%)

11.4 (2.5) NA

Solvent content (%) 82.2 74.7

RenementResolution () 29.23.95 48.562.6 No. of reections 15,307 13,770 Rwork/Rfree 0.320/0.348 0.209/0.249

No. of atoms

Protein 4,264 1,586 Ligand/ion 0 0 Water 0 0 B-factors

Protein (2) 155.6 88.4 RMSD

Bond lengths () 0.003 0.0102 Bond angles () 0.785 1.34

Molprobity geometry score

2.6 (98th percentile)

1.6 (100th percentile)

Molprobity clash score 9.4 (97th percentile)

1.15 (100th percentile)

NA, not applicable; RMSD, root mean square deviation.*Where values in parentheses refer to the highest resolution shell.

2 NATURE COMMUNICATIONS | 5:5421 | DOI: 10.1038/ncomms6421 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6421 ARTICLE

a

96% of the 540 amino acids in the structure. The only regions that could not be built were six residues at the N-terminus and a loop between Ser177 and Asn190, which the sequence predicted to be disordered. Although only a few side chain features are visible at this resolution, the nal rened model, which includes sidechains, has geometric and crystallographic renement parameters that compare very favourably with other structures of similar resolution (Table 1). The nal model correctly aligns methionine residues with the experimentally determined positions of the selenium atoms from the SAD analysis (Supplementary Fig. 1b).

Overall, the BsEzrA22562 monomer is a curved rod, forming a complete semicircle with a diameter of 120 . Groups of three a-helices pack together along its length (Fig. 1a), except at the

C-terminus where there is a four-helix bundle. The formation of a semicircle confers the potential of the C-terminal four-helix bundle to interact with the membrane at the same time that the N-terminus of EzrA is embedded in it through its N-terminal TM helix; this can be modelled trivially by manually extending the N-terminal a-helix of repeat 1 (Supplementary Fig. 1c). An anti-parallel dimer is observed in the BsEzrA22562 crystal lattice as a

result of crystallographic symmetry (Supplementary Fig. 2a). The dimerization of BsEzrA22562 in the crystal buries some 3,300 2

of surface area (B9% of the total) in an extensive, molecule-long interface. The anti-parallel dimer could not be detected in solution by sedimentation velocity, which indicates that the monomer and an oligomer of unknown arrangement are in equilibrium (Supplementary Fig. 2b). However, bacterial two-hybrid analysis in vivo shows an interaction between the N- and C-terminal regions of BsEzrA (Supplementary Fig. 2c), which is consistent with the anti-parallel dimer seen in the crystals and which is the only oligomeric arrangement in the crystal that would enable the N-terminal TM helices of both subunits to insert into the same face of a membrane in vivo.

To obtain higher resolution structural information on EzrA, several fragments from different EzrA homologues were subjected to crystallization. A 190 residue N-terminal fragment of S. aureus EzrA, SaEzrA24214 (that also lacks its N-terminal TM helix),

yielded crystals that diffracted to 2.6 resolution. Its structure (Fig. 1b, Supplementary Fig. 1d) was solved by molecular replacement using model phases from the equivalent portion of BsEzrA22562, and was rened to a nal crystallographic Rfree of

0.25 (Table 1). The structures of SaEzrA24214 and BsEzrA22562,

which share only 20% sequence identity, superpose with root mean square deviations (RMSDs) of 2.4 on 154 paired Ca atoms (Supplementary Fig. 3a). The BsEzrA22562 and the shorter

SaEzrA24214 rods comprise multiple copies of the same B100 amino acid, triple helical bundle; BsEzrA22562 contains ve such

repeats and SaEzrA24214 contains just two. The individual repeating units can all be superimposed with RMSDs of between1.5 and 2.3 on paired Ca atoms (Supplementary Fig. 3b).

Structural homologues of EzrA occur only in eukaryotes. A search for structural homologues of BsEzrA22562 using

PDBefold24, DALI25 and BLAST-3D26 using default matching criteria did not identify any structures that superimposed convincingly along the entire length of the structure; only short segments could be aligned onto structural homologues. Representative matches from a BLAST-3D search included the eukaryotic cytoskeletal proteins vinculin27 and a-actinin28, the nuclear importers karyopherin29 and importin30 and four-helical coiled coils such as the bacterial chemotaxis receptor, CheA31 (Supplementary Fig. 4a). The same searches were repeated with the structure of SaEzrA24214 as the query. PDBefold identied

ve matches, all of which resembled the hits from the BLAST-3D search with BsEzrA22562; one match was to vinculin and the

other four were to members of the spectrin superfamily to which a-actinin belongs.

Although EzrA structures could be aligned partially onto other proteins that also have an up-down-up arrangement of helices, the match with the spectrin superfamily of structures stood out as most signicant. Despite having an up-down-up helix conguration, vinculin, karyopherin, importin and CheA all lack discrete triple-helical bundles; in all of these protein families, each helix packs intimately against three, rather than two, other helices. In addition, although DALI searches matched both EzrA structures with other rod-shaped triple helical bundles, such as BAR domains32 (Supplementary Fig. 4a), in these proteins there is only a single, triple helical bundle that is longer (4150 amino acids)

than the repeat unit in EzrA. The linear head-to-tail arrangement of interconnected B100 amino-acid triple helical bundles appears to be unique to spectrins and to EzrA. Finally, no matter which search tool was used, the C-terminal four-helix bundle of EzrA superimposed only weakly on matching structures in the PDB, but it is notable that it has a supercial resemblance to the helical bundle fold of the actin-binding calponin homology domains33, which can be found at the termini of spectrin-type proteins.

EzrA is the founding member of the bacterial spectrin family. To our knowledge, the structural similarity of EzrA or any other bacterial protein to spectrins had not been predicted. From the logic below, we conclude that EzrA is the founding member of a

C-terminus

N-terminus

b

N-terminus

C-terminus

c

EzrA22562

22 132 215 302 410 497

R1

R2

R3

R4 R5

R5 CT

CT

EzrA22217

EzrA216562

R1 R2

R3 R4

Figure 1 | Structure of the cytoplasmic domain of EzrA. (a) Ribbon diagram of the BsEzrA22562 monomer coloured by triple helical bundle

repeats from N- to C-terminus, red, yellow, green, blue and magenta. The C-terminal, four-helix bundle is coloured silver. This and all other structure representations were made in PyMOL (http://www.pymol.org). (b) Ribbon diagram of the SaEzrA24214 monomer, coloured as in a. (c) Schematic illustration of the BsEzrA fragments used in this study, with blocks (numbered R1R5) representing the spectrin repeat units and the C-terminal four-helix bundle (labelled CT) and coloured as in a. The N- and C-termini and the junctions between spectrin repeats are numbered above the schematic.

NATURE COMMUNICATIONS | 5:5421 | DOI: 10.1038/ncomms6421 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6421

new family of bacterial spectrin-like proteins: (i) the eukaryotic structural homologues of EzrA share the same extended rod-like character, comprising repeating units of B100 amino acids arranged as triple helical anti-parallel bundles in a head-to-tail linear array (Fig. 2a, Supplementary Fig. 4b). Eukaryotic spectrins contain 420 of these units, called spectrin repeats, whereas EzrA contains 5 of them. (ii) Four of the ve BsEzrA22562 spectrin

repeats and the two in SaEzrA24214 can be superimposed on

representative eukaryotic spectrin repeats with RMSDs of between 1.9 and 2.6 on matched Ca atoms (Fig. 2b); this is a high degree of similarity that is comparable to RMSDs among the EzrA spectrin repeats. Moreover, each superimposition accounts for greater than 70% of the EzrA spectrin repeat with paired amino acids coming from all three of the helices. (iii) The structural similarity of EzrA to the spectrin superfamily extends to the junctions and interfaces among adjacent repeats. As with all other crystallized spectrins, the repeat junctions in EzrA are formed from long continuous a-helices that form the third helix of one repeat and the rst helix of the next. As also observed in several spectrins3436, the SaEzrA24214 structure reveals intimate interactions at the interface between repeats involving the linker helix and loop residues. For example, the linker residues Tyr135 and Ile131 interact with loop residues Thr62, Leu56 and Tyr193 (Supplementary Fig. 5a,b). (iv) Although there are multiple examples of spectrin structures containing dual repeats, arranged

like EzrA in a head-to-tail linear arrangement (Figs 1a and 2a, Supplementary Fig. 4b), the semi-circular curvature (Figs 1a and 2a) of BsEzrA22562 is unique. The principal axes of adjacent

repeats in BsEzrA22562 intersect with angles of 37, 38, 33 and 53, resulting in the pronounced overall bend. The angle between the two repeats in SaEzrA24214 is 34, suggesting that full-length

EzrA from S. aureus will also have a pronounced bend. The magnitude of these angles in EzrA is similar to that between erythrocytic b-spectrin repeats 14 and 15 (ref. 35), which is 37

(Fig. 2c). Furthermore, a study of the six-repeat plakin domain of desmoplakin in solution by small-angle X-ray scattering was consistent with the plakin adopting an overall L-shape37, but BsEzrA22562 is the rst structure with a sufcient bend between

enough repeats to generate a full semi-circle. The absence of longer chains of spectrin repeats from the PDB may simply reect a crystallization bias towards the shorter, relatively rigid, dual spectrin repeats that predominate in the database.

Nevertheless, there are facets of the EzrA structure that differentiate it from spectrins. For instance, there is no sequence homology between EzrA and any spectrin, although short segments of both proteins share a heptad repeat periodicity of hydrophobic amino acids (Supplementary Fig. 6). Although the orientations of the three a-helices in the EzrA repeats are similar to those of spectrin repeats (Fig. 2a,b), the relative positions and therefore the connectivities of the helices are different (Fig. 2d).

c

36.8

36.9

34

a

b

3EDU

BsEzrA24217

SaEzrA24214

BsEzrA24129

SaEzrA24133 SaEzrA134214

d

BsEzrA130217

BsEzrA220303

BsEzrA304409

BsEzrA24129

1HCI630746

C 1HCI630746

C N

BsEzrA24129

B A A

C B

N

C

EzrA Spectrin

Figure 2 | Comparison of EzrA with spectrin family proteins. (a) Structure of the spectrin protein a-actinin (PDB 1HCI), with alternating triple helical bundle spectrin repeats coloured red and blue. (b) Superimposition of spectrin units from BsEzrA22562 or SaEzrA24214 (coloured red, yellow, green and

blue, as in Fig. 1a) onto representative spectrin units coloured cyan from erythroid spectrin (top; PDB 3KBT), a-actinin (middle; PDB 1HCI) and brain spectrin (bottom; PDB 1U5P). (c) The angle between spectrin repeats 1 and 2 in BsEzrA22562 is 37 and in SaEzrA24214 it is 34; the angle is also

37 between erythrocytic b-spectrin repeats 14 and 15 (PDB 3EDU). (d) TopOrthogonal views of the relative positioning and different connectivity of the three a-helices in EzrA spectrin repeat 1 (residues 24129) and in the fourth repeat (residues 630745) of the rod domain of a-actinin (PDB 1HCI).

The N-terminal helix in the bundle (a) is coloured blue, the middle helix (b) yellow and the C-terminal helix (c) red. BottomSchematic illustration of the different connectivity in the EzrA and spectrin three-helix bundles, viewed along the helix axes. An up arrow (helix a, c) represents an up helix viewed along the helix axis in an NC direction (N-terminus of helix axis above plane of page). A down arrow represents a down helix viewed in the opposite direction.

4 NATURE COMMUNICATIONS | 5:5421 | DOI: 10.1038/ncomms6421 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6421 ARTICLE

For example, the rst up helix in the EzrA repeat, helix A (blue, Fig. 2d), is superimposed on the third up helix in the spectrin repeat from a-actinin, helix C (red, Fig. 2d). The second down

helix in the EzrA repeat is superimposed well on the a-actinin equivalent (helix B, coloured yellow in Fig. 2d), whereas the third up helix in the EzrA repeat, helix C, is superimposed on the rst up helix in the a-actinin spectrin repeat, helix A. In other words, the left-handedness of the connectivity between helices in eukaryotic spectrin repeats (helices AC are arranged in a counter-clockwise manner when viewed end-on) is reversed to be right-handed in EzrA (helices AC are arranged clockwise).

Direct interaction of EzrA with FtsZ. To map which regions of BsEzrA interact with B. subtilis FtsZ (BsFtsZ), a qualitative crosslinking approach using 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC), which conjugates amines directly to carboxyl groups, was combined with quantitative surface plasmon resonance (SPR) measurements. EDC-mediated zero-length crosslinks were found between BsFtsZ and the BsEzrA22562

fragment, but not to unrelated control proteins (Fig. 3a). To determine more precisely the crosslinking site within EzrA, the BsEzrA22562BsFtsZ adduct was cleaved site-specically

into dened fragments using the cysteine-specic cleavage reagent 2-nitro-5-thiocyanatobenzoic acid (NTCB). For this, a BsEzrA22562 variant (K135C/C222A) with a single cysteine between spectrin repeats 1 and 2 was prepared (Fig. 3b). Cleavage of the BsEzrA22562(K135C/C222A)BsFtsZ adduct with NTCB yielded a fragment consistent with the attachment of BsFtsZ to the N-terminal region of BsEzrA (residues 22134) comprising the rst spectrin repeat (Fig. 3c).

Guided by limited proteolysis studies and the crystal structure, two truncated BsEzrA fragments were produced. BsEzrA22217

encompassed repeats one and two and BsEzrA216562 included the

other three and the C-terminal domain (Fig. 1c). BsEzrA22217

bound to an SPR chip surface coated with immobilized BsFtsZ with fast kinetics, enabling equilibrium binding levels to be measured (Fig. 4a). Fitting the binding isotherm to a 1:1 binding model yielded a dissociation constant (Kd) of B13 mM for the

BsFtsZ:BsEzrA22217 complex (Fig. 4b). The BsEzrA216562 fragment also bound to the BsFtsZ-coated chip, but with considerably slower kinetics (Supplementary Fig. 7a), and tting the sensorgrams to a 1:1 binding model yielded a Kd of 2 mM. The specicity of the observed interactions is supported by the lack of any interaction of either BsEzrA131217 or bovine serum albumin

with the FtsZ chip surface under equivalent buffer conditions (dashed cyan sensorgrams, Fig. 4a and Supplementary Fig. 7c).

The binding of both BsEzrA22217 and BsEzrA216562 to BsFtsZ

implies that there are either two distinct FtsZ-binding sites on an EzrA monomer or a single site that extends across the interface between repeats 2 and 3. However, to be consistent with the crosslinking of BsFtsZ to repeat 1, a single binding site at the interface between repeats 2 and 3 would need to extend also to spectrin repeat 1, implying an intimate association of FtsZ with repeat 2. However, a fragment encompassing only repeat 2, BsEzrA131217, showed no detectable interaction with immobilized

BsFtsZ (red dashed traces, Fig. 4a) in SPR, despite it having appreciable secondary and tertiary structure content (Supplementary Fig. 8a,b). Therefore, it seems most likely that there are at least two distinct FtsZ interaction sites on an EzrA monomer, one of which is in spectrin repeat 1.

Effect of EzrA on FtsZ oligomerization. The BsEzrABsFtsZ interaction was analysed to map regions of EzrA responsible for the inhibition of FtsZ oligomerization. GTP-induced increase in

c

NTCB +

a

FtsZ

+ + + + +

+ +

EzrA BSA

Carbonic anhydrase RNAase

b

+

+ +

92 66

170

130

100

70

55

40

45

30

20

14

NTCB

+

170

130

100

70

55

40

35

Figure 3 | Qualitative binding of EzrA spectrin repeat 1 to FtsZ. (a) EDC crosslinking of BsFtsZ to BsEzrA22562. BsFtsZ with a 15-residue recombinant

N-terminal T7 epitope tag was incubated with BsEzrA22562 or various control proteins with the crosslinking agent EDC. The crosslinking reactions

were analysed by SDSPAGE and by western blotting to detect the T7 epitope tag. Only the reaction containing BsEzrA22562 and BsFtsZ yields a unique

product (highlighted with an arrow) with a mobility consistent with a 1:1 EzrA22562:FtsZ adduct of mass B100 kDa. The units of the masses listed

at the side of this and the other two panels in this gure are in kDa. (b) Cysteine-specic fragmentation of the BsEzrABsFtsZ adduct between spectrin repeats 1 and 2. The BsEzrA22562 mutant K135C/C222A has only one cysteine, at the interface between spectrin repeats 1 and 2. NTCB cleavage of isolated, puried K135C/C222A generates fragments of the expected size (16 kDa for residues 22134 plus the N-terminal hexahistidine expression tag; 49 kDa for residues 135562) on SDSPAGE. (c) Cleavage of the BsEzrA22562K135C/C22ABsFtsZ crosslinked adduct with NTCB. The cleavage product was analysed by SDSPAGE and by western blotting to detect the T7 epitope tag on FtsZ. After NTCB cleavage, a band (marked with an arrow) appears close to the 55-kDa marker, which is consistent with the 16-kDa N-terminal BsEzrA fragment (residues 22134) crosslinked to the 40-kDa BsFtsZ.

NATURE COMMUNICATIONS | 5:5421 | DOI: 10.1038/ncomms6421 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6421

130

110

90

70

50

30

10

30

50

10 70 120 170

a

a

FtsZ alone

0.98

FtsZ + EzrA22562

50, 100, 200 M

Response difference (RU)

FtsZ + EzrA22217

FtsZ + EzrA216562

0.78

0.58

0.38

0.18

0.02

1.2

1

0.8

0.6

0.4

0.2

0

0 5 10 15 20 25 30

FtsZ +

EzrA 22562

22.5 M

7.5 M

light scattering (AU)

FtsZ + EzrA22217 + EzrA216562

+GTP

Time (s)

Time (min)

b

100

Increase in response units (RU)

b

90 80 70 60 50 40 30 20 10

0 0 50 100 150 200

light scattering (AU)

[EzrA22217], M

FtsZ +

EzrA 22217

FtsZ +

EzrA 216562

FtsZ +

EzrA 22217+

EzrA 216562

Figure 4 | Quantication of binding of EzrA spectrin repeats 12 to FtsZ. (a) Binding of BsEzrA22217 to immobilized BsFtsZ was monitored by SPR.

Solid black sensorgrams represent the titration of increasing amounts of BsEzrA22217 over a chip on which BsFtsZ was immobilized by amine

coupling. The dashed cyan sensorgram represents an injection of 50 mM bovine serum albumin (BSA) over the same surface under the same conditions; the dashed red sensorgrams represent injections of 25, 50 and 100 mM of the BsEzrA131217 fragment. (b) The response units of

BsEzrA22217 binding the BsFtsZ surface as a function of injected protein were t with a curve corresponding to a 1:1 binding interaction, yielding a Kd of 13.71.5 mM. The actual binding process may be more complex than a simple 1:1 interaction because of the self-association of the BsEzrA22217

fragment, apparent from size exclusion chromatography-multi angle laser light scattering analysis (SEC-MALLS) (see Supplementary Fig. 7b).

FtsZ alone

Figure 5 | Inhibition of FtsZ oligomerization by EzrA fragments.(a) The GTP-induced oligomerization of FtsZ was monitored by light scattering. All samples contained 12.5 mM BsFtsZ and 25 mM of the indicated BsEzrA fragment. The arrow corresponds to the time when GTP was added to a nal concentration of 1 mM. (b) The average GTP-induced increase in light scattering in the presence of the BsEzrA fragments. The results shown are the means.d. from three independent measurements. The BsEzrA22217 fragment, encompassing spectrin repeats 12, reduces

FtsZ polymerization to undetectable levels.

light scattering provides a measure of the polymerization of FtsZ monomers into laments and the bundling of laments into higher order structures; we refer to these processes collectively as oligomerization. Previous studies have shown that a twofold excess of BsEzrA over BsFtsZ reduced FtsZ oligomerization between two- and tenfold16,38. The pH was increased from 6.5, which was used in previous studies with this assay, to 7.3, to reduce the aggregation of BsEzrA22562. Under these conditions, a

twofold excess of BsEzrA22562 inhibited BsFtsZ oligomerization about twofold (Fig. 5a,b), consistent with previous studies16,38.

Remarkably, identical concentrations of BsEzrA22217 in this assay

inhibited BsFtsZ oligomerization almost completely (Fig. 5a,b), whereas a twofold excess of BsEzrA216562 reduced BsFtsZ

oligomerization by about one-third. An equimolar mixture of BsEzrA22217 and BsEzrA216562 also completely inhibited

BsFtsZ assembly (Fig. 5a,b). The stronger inhibition of BsFtsZ assembly by the smaller BsEzrA22217 fragment in comparison

to BsEzrA22562 may reect the masking of FtsZ-interacting

sites by the oligomerization of BsEzrA22562; size exclusion

chromatography of BsEzrA22562 revealed that it formed an oligomer in the low salt buffer conditions used in the light scattering assay (Supplementary Fig. 8c). The relative levels of inhibition of BsFtsZ assembly by different BsEzrA fragments must be interpreted cautiously, but the marked inhibition by BsEzrA22217 points once more to a functionally important

interaction of the N-terminal region of EzrA with FtsZ.

To ascertain whether the marked inhibition of BsFtsZ oligomerization by BsEzrA22217 reects direct inhibition of FtsZ

polymerization into laments, or the blocking of lament bundling into higher order structures, its effect on the GTPase activity of BsFtsZ was examined. Head-to-tail polymerization of FtsZ into laments is dependent upon GTP binding39, and dissociation of FtsZ polymers into monomers is dependent upon GTP hydrolysis5. Therefore, either the inhibition of GTP binding or the stimulation of the GTPase activity could result in the inhibition of FtsZ head-to-tail polymerization by EzrA. The presence of BsEzrA22217 did not markedly affect the binding of

the uorescent BODIPY-GTP-g-S to BsFtsZ (Supplementary Fig. 9a), a result that is consistent with the very slight reduction

6 NATURE COMMUNICATIONS | 5:5421 | DOI: 10.1038/ncomms6421 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6421 ARTICLE

(1.6-fold) in GTP afnity for BsFtsZ in the presence of BsEzrA measured previously40. The GTPase activity of BsFtsZ was measured in the presence and absence of BsEzrA22217, to

determine whether EzrA affected GTP hydrolysis and thus FtsZ oligomerization. Both BsEzrA22562 and BsEzrA22217 had

marginal effects on GTP turnover in a coupled GTPase assay (Supplementary Fig. 9b) and in a malachite green assay that monitored release of free phosphate (Supplementary Fig. 9c). It would thus seem that EzrA has little impact on the binding of GTP by FtsZ or on its hydrolysis.

Finally, we examined the impact of BsEzrA22217 on the

oligomerization of BsFtsZ1366, which lacks the C-terminal 17

amino acids required for bundling of FtsZ laments41,42. Consistent with previous observations41, the change in light scattering on adding GTP to BsFtsZ1366 was reduced markedly,

by B15-fold relative to wild-type BsFtsZ (Supplementary Fig. 10a). As has been observed previously with BsEzrA22562

(ref. 43), EzrA22217 had no impact on the residual GTP-induced

oligomerization activity of BsFtsZ1366 (Supplementary Fig. 10). These results suggest that EzrA does not affect the head-to-tail polymerization of FtsZ into laments but instead largely acts by preventing the bundling of FtsZ laments into higher-order assemblies.

DiscussionSpectrins are ubiquitous proteins involved in the organization of the cytoskeleton and membrane processes in animals. A typical spectrin has an elongated molecular shape due to multiple, connected repeats of three anti-parallel a-helices, which are slightly curved and coil around each other. Individual spectrin repeats are approximately 50 long and 20 wide and are similar in length (99122 amino acids). Their primary amino-acid sequences differ signicantly, but the bundling of the helices is stabilized by regularly spaced hydrophobic side chains. EzrA shares all of these features with the spectrin superfamily of proteins, and the structural similarity of the EzrA repeats with those of other spectrin-like proteins is striking (Fig. 2). Although many proteins are known to comprise bundles of antiparallel a-helices, EzrA is, to the best of our knowledge, the only bacterial protein that is formed from serial spectrin-like repeats.

Structure-based sequence alignments do not show any meaningful sequence homology either among the ve EzrA repeats, or between EzrA and spectrins. Such a lack of sequence conservation is not without precedence within the spectrin superfamily44. The spectrin repeat is a degenerate sequence established predominantly on the heptad repeat pattern of hydrophobic amino acids in the primary sequence, a characteristic of left-handed helical supercoiled proteins45. The degeneracy is exemplied by the observation that there are only four invariant residues in an alignment of the four spectrin repeats in human a-actinin 2, one invariant residue in an alignment of the ten spectrin repeats in the human Duchenne muscular dystrophy protein and none in the seventeen spectrin repeats of human erythrocyte b-spectrin. Therefore, even the tryptophan in the rst helix in the repeat, originally described as critical for thermodynamic stability of the spectrin repeat46,47, is not strictly maintained. The rst and third repeats in EzrA each contain a highly conserved proline residue in helix A; Pro45 of SaEzrA in repeat 1 (Supplementary Fig. 11a) is conserved in eight out of nine phylogenetically diverse paralogues, whereas Pro228 (Pro227 in BsEzrA) in repeat 3 (Supplementary Fig. 11b) is conserved in all of them. Although sequence analyses of spectrins have identied that a proline is found frequently in helix B48, neither helix A nor C has been reported to be interrupted by proline residues.

There is signicant variation in the relative orientation of adjacent repeat units in spectrin structures35,49. However, EzrA is the rst spectrin-like repeat-containing structure in which there is a pronounced enough bend between a sufcient number of repeats to form a complete semicircle. Molecular dynamics studies suggest the exibility of the junction regions in spectrins is constrained appreciably, with the force constants for changing the angle between adjacent repeats determined36 to be between 20 and 60 kcal mol 1 rad 2. A considerable input of energy is thus likely to be necessary to convert the semicircular EzrA into a straight rod. The solvent contents of the EzrA crystals are unusually high (BsEzrA22562 80%; SaEzrA24214 74%); 480% of

all entries in the PDB have a solvent content o60% and only1.1% have a solvent content 475%. The high solvent content of these crystals does not concur with the release of sufcient energy during crystallization to drive thermodynamically unfavourable events. Therefore, the semicircular shape is highly unlikely to represent an artefact induced by crystal packing. The extended curved structure is a property EzrA shares with SepF, an FtsZ bundling protein required for normal septum formation. B. subtilis SepF assembles into polymeric arcs13 and rings12 with a diameter (55 nm) larger than that of the EzrA semicircle (12 nm). FtsZ laments are proposed to bind on the outside of the SepF curve, whereas the inner face interacts with the surface of the invaginating septum membrane13. An analogous intimate association of the much smaller inner face of the EzrA semicircle with the membrane in vivo would require extreme curvature of the lipid bilayer, which has not been reported to date.

A question raised by the BsEzrA22562 structure is the orientation of full-length EzrA relative to the lipid bilayer. A tilted orientation of the EzrA arch with respect to the membrane could result from the curvature of the membrane, the bending of the TM helix or exibility in the junction between the TM helix and the cytoplasmic domain. The conserved proline in helix 1 of repeat 1 (Pro45 in S. aureus EzrA) may confer some additional exibility in the junction between the body of the cytoplasmic domain and the membrane. However, an intimate association of the EzrA cytoplasmic domain with the surface of the membrane, as seen in the curve-shaped helical BAR domain proteins32, is not supported by electrostatic potential calculations on either of the EzrA structures presented here. A tilted orientation of the arch would position the C-terminal four-helix bundle away from the surface of the membrane. This domain contains a conserved motif between residues 505 and 511, termed the QNR patch (Supplementary Fig. 1c), that is important for mid-cell localization38. The QNR patch has been proposed recently to bind directly to FtsZ, based upon the inhibition of FtsZ oligomerization by different EzrA fragments in a light scattering assay in vitro50.

In the simplest model for the orientation of EzrA, the TM helix would form a straight extension of the rst few turns of helix A of the rst spectrin repeat; secondary structure predictions suggest a continuous uninterrupted helix extends into the rst helix in the cytoplasmic domain. This orientation would result in the cytoplasmic domain of EzrA forming an arch with the C-terminal helical bundle positioned close to the membrane surface (Supplementary Fig. 1c). Although the sequence register in this region of the structure must be regarded cautiously, because of the paucity of methionine residues in this region of the sequence, this model would position key solvent-exposed residues in the QNR patch (R510, R512 and F511) to interact with the membrane, and preclude their direct interaction with Z-ring FtsZ polymers (Supplementary Fig. 1c). This suggests rather that the QNR patch controls EzrA localization through direct interaction with the membrane or specic membrane-associated protein(s).

NATURE COMMUNICATIONS | 5:5421 | DOI: 10.1038/ncomms6421 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6421

An arch over the membrane could allow EzrA to regulate FtsZ oligomerization by trapping FtsZ laments in the gap between the membrane and the inside of the EzrA curve. Molecular modelling with the Methanococcus janaschii FtsZ dimer51 and the Thermotoga maritima FtsA:FtsZ complex7 reveals there is sufcient space beneath the arch, in both monomeric and dimeric forms of EzrA, to accommodate an FtsZ lament or even to simultaneously accommodate FtsA-mediated membrane-anchored FtsZ and FtsA protolaments (Fig. 6). Trapping FtsZ protolaments inside the arch is intuitively an effective means to disrupt lateral interactions between protolaments and could additionally anchor contractile protolaments to the membrane. An individual lament inside the arch cannot engage in lateral interactions with laments outside of it (Fig. 6), providing an additional level of Z-ring regulation. The ability of the arch to accommodate FtsZ protolaments could be modulated depending on the cellular location of EzrA by the curvature of the cell membrane and/or by competing interactions of EzrA with other divisome components such as GpsB. In addition, or alternatively, FtsA and FtsZ may interact with the outside face of the EzrA arch only, leaving the inner surface for mediating alternative proteinprotein interactions.

A bacterial origin for the eukaryotic cytoskeleton has been suggested previously, in part, because of the similarity in structure and function of FtsZ to tubulin and FtsA to actin, although the primary amino-acid sequences are poorly related1,52,53 and their structures differ in some respects. For example, two C-terminal a-helices (residues 381437), which cover the surface of tubulin and are intimately associated with the globular body of the protein, are not visible in many FtsZ structures from different organisms5357 or are replaced by b-strands51. In addition, the relative positioning of the four subdomains differs in FtsA and actin58, and these two proteins are further distinguished by the unique presence in FtsA of subdomain 1C58. Our study extends the similarities between the eukaryotic and prokaryotic cytoskeletons to proteins that contain spectrin-like repeats. An established function of spectrins in eukaryotes is to connect actin laments to other membrane-associated proteins and, at least in the case of plectin, to also regulate actin dynamics59. An analogous role can thus be envisaged for EzrA during cell division, co-ordinating FtsA/FtsZ dynamics inside the cell to peptidoglycan biosynthesis on the outside, an activity that is essential for cytokinesis and which is effected by the recruitment of PBP1 to the site of division by its interaction with EzrA17.

Methods

Full details of the Methods are supplied in the Supplementary Information.

Expression plasmid construction. The genes encoding fragments of Bs and SaEzrA and T7 epitope-tagged FtsZ were cloned into pET-based vectors. The plasmids pBS58 and pCXZ for overexpressing FtsZ were a gift from David Adams (Newcastle University).

Expression and purication of EzrA and EzrA fragments. Recombinant EzrA proteins were produced using E. coli strain BL21(DE3) carrying the appropriate plasmid. Expression was induced in Luria Bertani (LB) medium with isopropyl b-

D-1-thiogalactopyranoside or by auto-induction. For the isopropyl b-D-1-thiogalactopyranoside -induced expression of selenomethionine-labelled proteins, the methionine-auxotrophic host strain B834 (DE3) was grown in a dened, minimal medium based on M9 supplemented with appropriate ions, vitamins and 19 amino acids; methionine was replaced by selenomethionine. For the BsEzrA proteins, the cells were harvested by centrifugation, before the cell pellet was resuspended in lysis buffer. Cells were lysed by sonication and claried by centrifugation. The super-natant was ltered and then puried by Ni-nitrilotriacetic acid (Ni-NTA) afnity chromatography with elution effected by imidazole. The hexahistidine tag was removed by overnight incubation with thrombin, before further purication by size exclusion chromatography. The puried proteins were snap-frozen in liquid nitrogen for storage. SaEzrA24214 was produced without a tag, and was puried by ammonium sulphate fractionation and anion exchange chromatographic steps.

Expression and purication of BsFtsZ. Untagged BsFtsZ was co-expressed withE. coli FtsQ, FtsA and FtsW using plasmids pCXZ and pBS58, as described previously60. FtsZ was puried by ion exchange chromatography and serial ammonium sulphate precipitation at 20, 30 and 40% (w/v). The proteins precipitated with 40% ammonium sulphate were resuspended and puried further by size exclusion chromatography, before snap-freezing in liquid nitrogen for storage. T7-tagged BsFtsZ was puried by Ni-NTA afnity chromatography following the same procedures as used for His-tagged BsEzrA fragments, and the His-tag was removed by thrombin.

Crystallization and structure determination. BsEzrA22562 was crystallized using

a mother liquor of 50 mM MES/NaOH, pH 6.7, 1 M tri-sodium citrate, 0.1 M ammonium sulphate, 0.2 M non-detergent sulphobetaine-201. The structure of EzrA was solved by single wavelength anomalous dispersion from a single selenomethionine-labelled crystal. The heavy atoms substructure and initial phasing were determined in SHELX61. SHELXE61 and PHASER62 were used for subsequent density modication. The atomic model was built by manually docking a-helices into the electron density in COOT63. The amino-acid sequence was assigned by matching the positions of selenium atoms to methionine positions in the protein sequence. SLOOP64 was used to nd loops that best t the electron density map. The geometry of the loophelix junctions was corrected using RAPPER65 and PHENIX. REFINE66 was used for renement of the model.

SaEzrA24214 was crystallized in a buffer of 34% 2-methyl-2, 4-pentanediol,0.1 M Tris-HCl, pH 7.5, and its structure was solved by molecular replacement using equivalent atoms from the BsEzrA22562 structure as a search model in

FtsZ filaments

Front view

Top view

Lipid bilayer

FtsA filaments

FtsA/ FtsZ filament axis

Figure 6 | Model of the interaction of EzrA with the cytoskeleton at the membrane surface. Model of FtsZ and FtsA protolaments situated both alongside and beneath the inner face of the arch-shaped EzrA molecule (here represented as the crystallographic anti-parallel dimer, component subunits coloured light and dark blue). The modelled TM helix of EzrA is coloured grey. The coordinates of T. maritima FtsA bound to a C-terminal fragment of FtsZ (PDB 4A2A) and the M. jannaschii FtsZ dimer (PDB 1W59) and B. subtilis FtsZ monomer (PDB 2VAM) were used to model FtsA (light and dark red) and FtsZ (light and dark green) protolaments, respectively. FtsA has been oriented to place the membrane-anchored C-terminus close to the lipid bilayer and its binding site for FtsZ near the FtsZ lament. The FtsZ lament has been positioned to place its C-terminal FtsA-binding region (disordered in FtsZ crystal structures) close to the FtsA lament. An FtsZ lament trapped beneath the EzrA arch cannot bundle together with laments outside the arch. Although the crystallographic dimer is represented, the same effect on bundling could be made by EzrA monomers.

8 NATURE COMMUNICATIONS | 5:5421 | DOI: 10.1038/ncomms6421 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6421 ARTICLE

PHASER62 and rened to convergence using COOT63 and REFMAC5 (ref. 67). Summaries of the data collection and model renement statistics are provided in Table 1, and Molprobity68 was used for structure validation reports.

EDC crosslinking. EzrA fragments were mixed with T7-tagged FtsZ and cross-linked with EDC before analysis by SDSpolyacrylamide gel electrophoresis (SDSPAGE) and western blotting. To map the crosslink site on EzrA, crosslinked proteins were puried from non-crosslinked proteins by Ni-NTA before NTCB cleavage, SDSPAGE and immunoblotting with an anti-T7 antibody.

Light scattering. Light scattering measurements used a Varian Cary Eclipse uorimeter at 30 C, with all samples in a buffer of 50 mM TrisHCl, pH 7.3, 20 mM KCl, 3 mM magnesium acetate. Samples were pre-incubated at 30 C, centrifuged in a microfuge and the supernatant transferred to a quartz micro-cuvette. GTP was added to a nal concentration of 1 mM and light scattering recorded for at least 30 min. The GTP-induced change in light scattering in the presence of EzrA and FtsZ was normalized relative to the average GTP-induced scattering change for FtsZ alone.

Surface plasmon resonance. All SPR experiments used a Biacore X100 instrument with the chip surface at 25 C. Proteins were immobilized on Biacore CM5 chip surfaces by standard protocols. The binding data were all t using the Biacore X-100 evaluation software; for afnity ts, the response on the surface 4 s before the end of the injection were t as a function of the concentration of the injected EzrA fragment. Each SPR titration shown was performed independently at least twice.

GTPase assay. The GTPase activity of FtsZ was measured under the same buffer conditions as used for the light scattering assays. In an NADH oxidation-coupled continuous regenerative assay69, 150 ml samples were incubated at 30 C in a microplate reader after the addition of GTP to 1 mM. NADH consumption was measured, which is coupled to GDP production, from the decrease in absorption at 340 nm. For malachite green assays, the release of inorganic phosphate was measured 20 min after adding GTP to 2 mM. A standard curve was generated using phosphate ions in the concentration range 50450 mM.

GTP binding. Fluorescence was measured at 22 C in a Varian Cary Eclipse uorimeter. 150 ml of 250 nM BODIPY GTP-g-S in 50 mM HEPES, pH 7, 100 mM KCl, 5 mM MgCl2 was preincubated for 45 min before adding 410 ml of a stock solution of either 8 mM FtsZ or 8 mM FtsZ with 16 mM BsEzrA22217 in the same

buffer.

References

1. Lwe, J. & Amos, L. Crystal structure of the bacterial cell-division protein FtsZ. Nature 391, 203206 (1998).

2. Bramhill, D. & Thompson, C. GTP-dependent polymerization of Escherichia coli FtsZ protein to form tubules. Proc. Natl Acad. Sci. USA 91, 58135817 (1994).

3. Adams, D. & Errington, J. Bacterial cell division: assembly, maintenance and disassembly of the Z ring. Nat. Rev. Microbiol. 7, 642653 (2009).4. Bi, E. & Lutkenhaus, J. FtsZ ring structure associated with division in Escherichia coli. Nature 354, 161164 (1991).

5. Mukherjee, A. & Lutkenhaus, J. Dynamic assembly of FtsZ regulated by GTP hydrolysis. EMBO J. 17, 462469 (1998).

6. Anderson, D., Gueiros-Filho, F. & Erickson, H. Assembly dynamics of FtsZ rings in Bacillus subtilis and Escherichia coli and effects of FtsZ-regulating proteins. J. Bacteriol. 186, 57755781 (2004).

7. Szwedziak, P., Wang, Q., Freund, S. M. & Lwe, J. FtsA forms actin-like protolaments. EMBO J. 31, 112 (2012).

8. Gamba, P., Veening, J., Saunders, N., Hamoen, L. & Daniel, R. Two-step assembly dynamics of the Bacillus subtilis divisome. J. Bacteriol. 191, 41864194 (2009).

9. Osawa, M., Anderson, D. E. & Erickson, H. P. Reconstitution of contractile FtsZ rings in liposomes. Science 320, 792794 (2008).

10. Loose, M. & Mitchison, T. J. The bacterial cell division proteins FtsA and FtsZ self-organize into dynamic cytoskeletal patterns. Nature Cell Biol. 16, 3846 (2014).

11. Gueriros-Filho, F. & Losick, R. A widely conserved bacterial cell division protein that promotes assembly of the tubulin-like protein FtsZ. Genes Dev. 16, 25442556 (2002).

12. Gndodu, M. et al. Large ring polymers align FtsZ polymers for normal septum formation. EMBO J. 30, 617626 (2011).

13. Duman, R. et al. Structural and genetic analyses reveal SepF as a new membrane anchor for the Z-ring. Proc. Natl Acad. Sci. USA 110, E4601E4610 (2013).

14. Hamoen, L. W., Meile, J. C., de Jong, W., Noirot, P. & Errington, J. SepF,a novel FtsZ-interacting protein required for a late step in cell division. Mol. Microbiol. 59, 989999 (2006).

15. Levin, P. A., Kurtser, I. G. & Grossman, A. D. Identication and characterization of a negative regulator of FtsZ ring formation in Bacillus subtilis. Proc. Natl Acad. Sci. USA 96, 96429647 (1999).

16. Haeusser, D., Schwartz, R., Smith, A., Oates, M. & Levin, P. EzrA prevents aberrant cell division by modulating assembly of the cytoskeletal protein FtsZ. Mol. Microbiol. 52, 801814 (2004).

17. Claessen, D. et al. Control of the cell elongation-division cycle by shuttling of PBP1 protein in Bacillus subtilis. Mol. Microbiol. 68, 10291046 (2008).

18. Jorge, A. M., Hoiczyk, E., Gomes, J. P. & Pinho, M. G. EzrA contributes to the regulation of cell size in Staphylococcus aureus. PLoS ONE 6, e27542 (2011).

19. Chung, K. M., Hsu, H. H., Govindan, S. & Chang, B. Y. Transcription regulation of ezrA and its effect on cell division of Bacillus subtilis. J. Bacteriol. 186, 59265932 (2004).

20. Steele, V. R., Bottomley, A. L., Garcia-Lara, J., Kasturiarachchi, J. & Foster, S. J. Multiple essential roles for EzrA in cell division of Staphylococcus aureus. Mol. Microbiol. 80, 542555 (2011).

21. Van Opijnen, T., Bodi, K. & Camill, A. Tn-seq: high-throughput parallel sequencing for tness and genetic interaction studies in microorganisms. Nat. Methods 6, 767772 (2009).

22. Thanassi, J. A., Hartman-Neumann, S. L., Dougherty, T. J., Dougherty, B. A. & Pucci, M. J. Identication of 113 conserved essential genes using a high-throughput gene disruption system in Streptococcus pneumoniae. Nucleic Acids Res. 30, 31523162 (2002).

23. Considine, K. M., Sleator, R. D., Kelly, A. L., Fitzgerald, G. F. & Hill, C. Identication and characterization of an essential gene in Listeria monocytogenes using an inducible gene expression system. Bioeng. Bugs 2, 150159 (2011).

24. Krissinel, E. & Henrick, K. Secondary-structure matching (SSM), a new tool for fast protein structure alignment in three dimensions. Acta Crystallogr. D60, 22562268 (2004).

25. Holm, L. & Rosenstrm, P. Dali server: conservation mapping in 3D. Nucleic Acids Res. 38, W545W549 (2010).

26. Tung, C. H., Huang, J. W. & Yang, J. M. Kappa-alpha plot derived structural alphabet and BLOSUM-like substitution matrix for rapid search of protein structure database. Genome Biol. 8, R31 (2007).

27. Borgon, R. A., Vonrhein, C., Bricogne, G., Bois, P. R. & Izard, T. Crystal structure of human vinculin. Structure 12, 11891197 (2004).

28. Liu, J., Taylor, D. W. & Taylor, K. A. A 3-D reconstruction of smooth muscle alpha-actinin by CryoEM reveals two different conformations at the actin-binding region. J. Mol. Biol. 338, 115125 (2004).

29. Cansizoglu, A. E., Lee, B. J., Zhang, Z. C., Fontoura, B. M. & Chook, Y. M. Structure-based design of a pathway-specic nuclear import inhibitor. Nat. Struct. Mol. Biol. 14, 452454 (2004).

30. Grnwald, M. & Bono, F. Structure of Importin13-Ubc9 complex: nuclear import and release of a key regulator of sumoylation. EMBO J. 30, 427438 (2011).

31. Park, S. Y. et al. Reconstruction of the chemotaxis receptor-kinase assembly. Nat. Struct. Mol. Biol. 13, 400407 (2006).

32. Peter, B. J. et al. BAR domains as sensors of membrane curvature: the amphiphysin BAR structure. Science 495, 495499 (2004).

33. Banuelos, S., Saraste, M. & Djinovi Carugo, K. Structural comparisons of calponin homology domains: implications for actin binding. Structure 6, 14191431 (1998).

34. Kusunoki, H., MacDonald, R. I. & Mondragn, A. Structural insights intothe stability and exibility of unusual erythroid spectrin repeats. Structure 12, 645656 (2004).

35. Stabach, P. R. et al. The structure of the ankyrin-binding site of b-spectrin reveals how tandem spectrin-repeats generate unique ligand-binding properties. Blood 113, 53775384 (2009).

36. Mirijanian, D., Chu, J., Ayton, G. & Voth, G. Atomistic and coarse-grained analysis of double spectrin repeat units: the molecular origins of exibility.J. Mol. Biol. 365, 523534 (2007).37. Al-Jassar, C. et al. The nonlinear structure of the desmoplakin plakin domain and the effects of cardiomyopathy-linked mutations. J. Mol. Biol. 411, 1049 1061 (2011).

38. Haeusser, D. P., Garza, A. C., Buscher, A. Z. & Levin, P. A. The division inhibitor EzrA contains a seven-residue patch required for maintaining the dynamic nature of the medial FtsZ ring. J. Bacteriol. 189, 90019010 (2007).

39. Mukherjee, A. & Lutkenhaus, J. Guanine nucleotide-dependent assembly of FtsZ into laments. J. Bacteriol. 176, 27542758 (1994).

40. Chung, K. M., Hsu, H. H., Yeh, H. Y. & Chang, B. Y. Mechanism of regulation of prokaryotic tubulin-like GTPase FtsZ by membrane protein EzrA. J. Biol. Chem. 282, 1489114897 (2007).

41. Buske, P. & Levin, P. Extreme C-terminus of bacterial cytoskeletal protein FtsZ plays fundamental role in assembly independent of modulatory proteins. J. Biol. Chem. 287, 1094510957 (2012).

42. Buske, P. & Levin, P. A exible C-terminal linker is required for proper FtsZ assembly in vitro and cytokinetic ring formation in vivo. Mol. Microbiol. 89, 249263 (2013).

NATURE COMMUNICATIONS | 5:5421 | DOI: 10.1038/ncomms6421 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6421

43. Singh, J. K., Makde, R. D., Kumar, V. & Panda, D. A membrane protein, EzrA, regulates assembly dynamics of FtsZ by interacting with the C-terminal tail of FtsZ. Biochemistry 46, 1101311022 (2007).

44. Nicolas, A., Delalande, O., Hubert, J. F. & Le Rumeur, E. The spectrin family of proteins: a unique coiled-coil fold for various molecular surface properties.J. Struct. Biol. 186, 392401 (2014).45. Lupas, A. & Gruber, M. Historical review: another 50th anniversary--new periodicities in coiled coils. Trends Biochem. Sci. 28, 679685 (2003).

46. Pantazatos, D. P. & MacDonald, R. I. Site-directed mutagenesis of either the highly conserved Trp-22 or the moderately conserved Trp-95 to a large, hydrophobic residue reduces the thermodynamic stability of a spectrin repeating unit. J. Biol. Chem. 272, 2105221059 (1997).

47. MacDonald, R. I., Musacchio, A., Holmgren, R. A. & Saraste, M. Invariant tryptophan at a shielded site promotes folding of the conformational unit of spectrin. Proc. Natl Acad. Sci. USA 91, 12991303 (1994).

48. Pascual, J., Castresana, J. & Saraste, M. Evolution of the spectrin repeat. Bioessays 19, 811817 (1997).

49. Grum, V. L., Li, D., MacDonald, R. I. & Mondragn, A. Structures of two repeats of spectrin suggest models of exibility. Cell 98, 523535 (1999).

50. Land, A. D., Luo, Q. & Levin, P. A. Functional domain analysis of the cell division inhibitor EzrA. PLoS ONE 9, e102616 (2014).

51. Oliva, M., Cordell, S. & Lwe, J. Structural insights into FtsZ protolament formation. Nat. Struct. Mol. Biol. 11, 12431250 (2004).

52. van den Ent, F., Amos, L. A. & Lwe, J. Prokaryotic origin of the actin cytoskeleton. Nature 413, 3944 (2001).

53. van den Ent, F., Amos, L. & Lwe, J. Bacterial ancestry of actin and tubulin. Curr. Opin. Microbiol. 4, 634638 (2001).

54. Li, Y. et al. FtsZ protolaments use a hinge-opening mechanism for constrictive force generation. Science 341, 392395 (2013).

55. Haydon, D. J. et al. An inhibitor of FtsZ with potent and selective anti-staphylococcal activity. Science 321, 16731675 (2008).

56. Matsui, T. et al. Structural reorganization of the bacterial cell-division protein FtsZ from Staphylococcus aureus. Acta Crystallogr. D68, 11751188 (2012).

57. Oliva, M. A., Trambaiolo, D. & Lwe, J. Structural insights into the conformational variability of FtsZ. J. Mol. Biol. 373, 12291242 (2007).

58. van den Ent, F. & Lwe, J. Crystal structure of the cell division protein FtsA from Thermotoga maritima. EMBO J. 19, 53005307 (2000).

59. Andra, K., Nikolic, B., Stcher, M., Drenckhahn, D. & Wiche, G. Not just scaffolding: plectin regulates actin dynamics in cultured cells. Genes Dev. 12, 34423451 (1998).

60. Wang, X. & Lutkenhaus, J. The FtsZ protein of Bacillus subtilis is localized at the division site and has GTPase activity that is dependent upon FtsZ concentration. Mol. Microbiol. 9, 435442 (1993).

61. Sheldrick, G. M. A short history of SHELX. Acta Crystallogr. D64, 112122 (2008).

62. McCoy, A. et al. PHASER crystallographic software. J. Appl. Cryst. 40, 658674 (2007).

63. Emsley, P., Lohkamp, B., Scott, W. & Cowtan, K. Features and development of Coot. Acta. Crystallogr. D66, 486501 (2010).

64. Cowtan, K. Completion of autobuilt protein models using a database of protein fragments. Acta. Crystallogr. D68, 328335 (2012).

65. Furnham, N. et al. Knowledge-based real-space explorations for low-resolution structure determination. Structure 14, 13131320 (2006).

66. Adams, P. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D66, 213221 (2010).

67. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Renement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D53, 240255 (1997).

68. Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D66, 1221 (2010).

69. Ingerman, E. & Nunnari, J. A continuous, regenerative coupled GTPase assay for dynamin-related proteins. Methods Enzymol. 404, 611619 (2005).

Acknowledgements

We thank Bert van den Berg, Martin Embley, Jeff Errington, Robert Lightowlers and Tom Williams for useful discussions and comments on the manuscript; David Adams for advice on working with FtsZ; John Berrisford for assistance with low-resolution model building; and Helen Waller, Joseph Newman and Jeremy Lakey for assistance with SPR. We also acknowledge the Diamond Light Source and the Australian Synchrotron for access to synchrotron radiation beamlines. SEC-MALLS analyses were undertaken at the University of Manchester Biomolecular Analysis Facility. This work has been supported by the UK BBSRC (BB/G015902/1).

Author contributions

R.M.C., W.V. and R.J.L. wrote the manuscript; R.M.C. crystallized and solved the BsEzrA structure; R.M.C. and R.J.L. analysed the BsEzrA structure; R.M.C., N.K.B. and L.H. performed biochemical experiments with BsEzrA; A.S. performed sedimentation analyses of BsEzrA; R.A.D. performed two-hybrid analyses; R.M.C., A.S., R.A.D., W.V. and R.J.L. analysed the biochemical and genetic data; J.R.B., Z.-Q.X., N.E.D., E.J.H. and A.J.O. generated the SaEzrA fragments; R.M.C., Z.-Q.X., A.J.O., A.B. and R.J.L. crystallized and solved the SaEzrA structure; R.M.C., A.J.O., A.B., Z.-Q.X. and R.J.L. analysed the SaEzrA structure and R.M.C., W.V., N.E.D., A.J.O. and R.J.L. revised the manuscript.

Additional information

Accessions codes. The atomic co-ordinates and the associated structure factors have been deposited in the Protein Data Bank (http://http://www.pdbe.org

Web End =www.pdbe.org ) with accession code 4UXV for BsEzrA22562 and 4UY3 for SaEzrA24214.

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Cleverley, R. M. et al. Structure and function of a spectrin-like regulator of bacterial cytokinesis. Nat. Commun. 5:5421 doi: 10.1038/ncomms6421 (2014).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

10 NATURE COMMUNICATIONS | 5:5421 | DOI: 10.1038/ncomms6421 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.