Structure of the Tubulin/FtsZ-Like Protein TubZ from ... · Structure of the Tubulin/FtsZ-Like...

of 10/10
Structure of the Tubulin/FtsZ-Like Protein TubZ from Pseudomonas Bacteriophage ΦKZ Christopher H. S. Aylett, Thierry Izoré, Linda A. Amos and Jan Löwe MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge CB2 0QH, UK Correspondence to Jan Löwe: [email protected]; [email protected] Edited by E. Nogales Abstract Pseudomonas ΦKZ-like bacteriophages encode a group of related tubulin/FtsZ-like proteins believed to be essential for the correct centring of replicated bacteriophage virions within the bacterial host. In this study, we present crystal structures of the tubulin/FtsZ-like protein TubZ from Pseudomonas bacteriophage ΦKZ in both the monomeric and protofilament states, revealing that ΦKZ TubZ undergoes structural changes required to polymerise, forming a canonical tubulin/FtsZ-like protofilament. Combining our structures with previous work, we propose a polymerisationdepolymerisation cycle for the Pseudomonas bacteriophage subgroup of tubulin/FtsZ-like proteins. Electron cryo-microscopy of ΦKZ TubZ filaments polymerised in vitro implies a long- pitch helical arrangement for the constituent protofilaments. Intriguingly, this feature is shared by the other known subgroup of bacteriophage tubulin/FtsZ-like proteins from Clostridium species, which are thought to be involved in partitioning the genomes of bacteriophages adopting a pseudo-lysogenic life cycle. © 2013 Elsevier Ltd. All rights reserved. Introduction Bacteriophages are believed to be the most abundant organisms in the world. 1 Despite their number and diversity, the small size and simplicity of the majority of bacteriophages combined with their use of co-opted host proteins appears to have limited the requirement for their own cytomotive (formerly cytoskeletal) filament systems. 2,3 Bacteriophage- encoded cytomotive filament proteins remain infre- quent despite the number of sequenced genomes now available. However, although actin-like proteins have not yet been identified within a bacteriophage, both deviant Walker A cytoskeletal ATPases 4 and tubulin/FtsZ family cytoskeletal proteins 5,6 have been found. Bacteriophage-encoded cytomotive filament pro- teins were first identified in the partitioning systems of pseudolysogenic bacteriophages. Such temper- atebacteriophages do not cause immediate lysis of the host cell but are capable of maintaining themselves within the bacterial cytoplasm as a separate plasmid. This necessitates the presence of a bacteriophage-genome-encoded cytomotive filament to organise the accurate partitioning of the prophage into both daughter cells at cell division. The prototypical deviant Walker A, parABS plasmid partitioning system is encoded by Escherichia coli prophage P1 7 and is responsible for the consistent segregation of the prophage as a plasmid. These systems were not immediately identified as parts of the bacterial cytoskeleton; however, later studies have shown that they may form filaments within the bacterial cell. 810 Only two bacteriophage tubulin/FtsZ-like proteins have so far been reported. Each has been proposed to represent an exemplar of its own independent subgroup of bacteriophage tubulin/FtsZ-like pro- teins; however, the evolutionary relationships within bacteriophage tubulins remain ill-defined at this early juncture and will remain so until many further structures and functional data are available. In order to avoid prejudging the relationships between these proteins, we refer to both subgroups (and the newly discovered subject of this study) using the general name TubZto denote tubulin/FtsZ-like proteins that are different from, but related to both tubulin and FtsZ. 5,6,11,12 The first bacteriophage-encoded tubulin/FtsZ-family cytomotive filament protein to be reported was also a 0022-2836/$ - see front matter © 2013 Elsevier Ltd. All rights reserved. J. Mol. Biol. (2013) 425, 21642173
  • date post

  • Category


  • view

  • download


Embed Size (px)

Transcript of Structure of the Tubulin/FtsZ-Like Protein TubZ from ... · Structure of the Tubulin/FtsZ-Like...

  • Structure of the Tubulin/FtsZ-Like Protein TubZ fromPseudomonas Bacteriophage KZ

    Christopher H. S. Aylett, Thierry Izor, Linda A. Amos and Jan Lwe

    MRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge CB2 0QH, UK

    Correspondence to Jan Lwe: [email protected]; [email protected]:// by E. Nogales


    Pseudomonas KZ-like bacteriophages encode a group of related tubulin/FtsZ-like proteins believed to beessential for the correct centring of replicated bacteriophage virions within the bacterial host. In this study, wepresent crystal structures of the tubulin/FtsZ-like protein TubZ from Pseudomonas bacteriophageKZ in boththe monomeric and protofilament states, revealing that KZ TubZ undergoes structural changes required topolymerise, forming a canonical tubulin/FtsZ-like protofilament. Combining our structures with previous work,we propose a polymerisationdepolymerisation cycle for the Pseudomonas bacteriophage subgroup oftubulin/FtsZ-like proteins. Electron cryo-microscopy ofKZ TubZ filaments polymerised in vitro implies a long-pitch helical arrangement for the constituent protofilaments. Intriguingly, this feature is shared by the otherknown subgroup of bacteriophage tubulin/FtsZ-like proteins from Clostridium species, which are thought to beinvolved in partitioning the genomes of bacteriophages adopting a pseudo-lysogenic life cycle.

    2013 Elsevier Ltd. All rights reserved.


    Bacteriophages are believed to be the mostabundant organisms in the world.1 Despite theirnumber and diversity, the small size and simplicity ofthe majority of bacteriophages combined with theiruse of co-opted host proteins appears to have limitedthe requirement for their own cytomotive (formerlycytoskeletal) filament systems.2,3 Bacteriophage-encoded cytomotive filament proteins remain infre-quent despite the number of sequenced genomesnow available. However, although actin-like proteinshave not yet been identified within a bacteriophage,both deviant Walker A cytoskeletal ATPases4 andtubulin/FtsZ family cytoskeletal proteins5,6 havebeen found.Bacteriophage-encoded cytomotive filament pro-

    teins were first identified in the partitioning systemsof pseudolysogenic bacteriophages. Such temper-ate bacteriophages do not cause immediate lysis ofthe host cell but are capable of maintainingthemselves within the bacterial cytoplasm as aseparate plasmid. This necessitates the presenceof a bacteriophage-genome-encoded cytomotivefilament to organise the accurate partitioning of the

    0022-2836/$ - see front matter 2013 Elsevier Ltd. All rights reserve

    prophage into both daughter cells at cell division.The prototypical deviant Walker A, parABS plasmidpartitioning system is encoded by Escherichia coliprophage P17 and is responsible for the consistentsegregation of the prophage as a plasmid. Thesesystems were not immediately identified as parts ofthe bacterial cytoskeleton; however, later studieshave shown that they may form filaments within thebacterial cell.810

    Only two bacteriophage tubulin/FtsZ-like proteinshave so far been reported. Each has been proposedto represent an exemplar of its own independentsubgroup of bacteriophage tubulin/FtsZ-like pro-teins; however, the evolutionary relationships withinbacteriophage tubulins remain ill-defined at this earlyjuncture and will remain so until many furtherstructures and functional data are available. Inorder to avoid prejudging the relationships betweenthese proteins, we refer to both subgroups (and thenewly discovered subject of this study) using thegeneral name TubZ to denote tubulin/FtsZ-likeproteins that are different from, but related to bothtubulin and FtsZ.5,6,11,12

    The first bacteriophage-encoded tubulin/FtsZ-familycytomotive filament protein to be reported was also a

    d. J. Mol. Biol. (2013) 425, 21642173

    mailto:[email protected]:[email protected]://

  • 2165Structure of KZ TubZ

    plasmid partitioning protein encoded by a pseudo-lysogenic bacteriophage (Clostridium botulinum pseu-dolysogenic prophage C-ST).5 Based on sequenceand structure, it has been designated as belonging tothe subgroup of TubZs believed to be responsible foraccurate plasmid partitioning in a number of membersof the Bacillus species.11 Congruently, the plasmidsand bacteriophages involved are typically those uponwhich toxins are found, supporting a common historyfor these systems.5 Notably, both the C-ST andBacillus TubZ proteins share the distinctive feature ofpolymerising to form double-helical twisted filaments,which have not been found in other members of thetubulin/FtsZ family of proteins.5,1214 The exactbiological role of TubZ from prophage C-ST is notknown and hence its classification as a partitioningprotein remains preliminary.In contrast to the bacteriophage TubZ identified in

    Clostridium, the TubZ found within a Pseudomonasbacteriophage has not been proposed to be involvedin partitioning of a pseudolysogenic prophage.Pseudomonas TubZ (which is also referred to asPhuZ) was first identified in bacteriophage 2012-1as a homologue of tubulin/FtsZ.6 While various closehomologues have been found in other bacterio-phages, it constitutes the only bacteriophage cyto-motive filament protein for which no close plasmid orchromosomal homologues are known. Bacterio-phage 2012-1 is a member of the KZ-like groupof viruses,15 notable due to the size of their genomeand coat.16,17 Although KZ-like bacteriophagesare known pseudolysogens,18 there is no evidencefor a role in partitioning. 2012-1 TubZ (PhuZ) isinstead thought to localise bacteriophages within thebacterium, forming cytomotive filaments within thehost cell that position virions at the cell centre forefficient release upon lysis.6

    A previous crystal structure of 2012-1 TubZ(PhuZ) revealed a protofilament-like arrangementwith a weak longitudinal contact between adjacentsubunits and exhibited an extended C-terminusrunning along the side of the protofilament bearingan acidic knuckle, which mutation revealed wascritical for polymerisation.6 The difference betweenthe active site from the 2012-1 TubZ (PhuZ)structure and that found in canonical tubulin/FtsZprotofilaments posed the question of whether or notPseudomonas TubZs function in the same way.Electron micrographs of polymerised 2012-1 TubZ(PhuZ) suggested filaments consisting of flat ribbonsof paired protofilaments, similar to those in thecrystal structure.6

    In this study, we looked at the protofilamentcontacts and filament structure formed by TubZfrom Pseudomonas bacteriophage KZ. We havesolved the crystal structure of KZ TubZ in themonomeric and polymeric forms, revealing structuralchanges in KZ TubZ during polymerisation andshowing that KZ TubZ forms canonical tubulin/

    FtsZ subunitsubunit contacts and a functionaltubulin/FtsZ active site. We propose that the threeavailable structures of Pseudomonas TubZs (onefrom 2012-1 and two from KZ) describe apolymerisationdepolymerisation cycle. Electroncryo-microscopy of KZ TubZ filaments poly-merised in vitro implies a long-pitch twisted helicalarrangement in which the constituent protofilamentsare entwined around one another; different filamentarchitectures could be observed, all showing twistedfilaments. This is notably also a feature of filamentsof the other known subgroup of bacteriophage-borneTubZs identified in Clostridium.5

    Results and Discussion

    Bacteriophage tubulin/FtsZ (TubZ) fromPseudomonas bacteriophage KZ

    In order to investigate the Pseudomonas group ofbacteriophage tubulin/FtsZs, we synthesised the geneencoding TubZ from Pseudomonas bacteriophageKZ (Taxon 169683; UniProt ID Q8SDC3), over-produced protein in E. coli and purified it to homoge-neity (Fig. 1b). To gain a structural understanding ofthe protein, we began by crystallising and solving thestructure of KZ TubZ through molecular replace-ment.We solved the structure of crystals ofKZ TubZto 2.0 (R/Rfree = 0.17/0.21); the asymmetric unitcontained a single subunit of the protein in amonomeric state (Fig. 1c; Table 1). The GTPaseactive site within the subunit was occupied by GDP,which was retained by KZ TubZ through thepurification process, and only the extreme N- and C-termini and a single region of surface loop T3 (residues5860) were not resolved within the electron density.

    KZ TubZ adopts a tubulin/FtsZ fold

    The crystal structures ofKZ TubZ revealed that itexhibits a tubulin/FtsZ protein family fold consistingof the canonical N-terminal GTP-binding domain,connected to the C-terminal GTPase activationdomain through a conserved helix (H7) (Fig. 1c).Tubulin/FtsZ family proteins frequently possess N-and C-terminal extensions to the conserved core,and althoughKZTubZ has no significant N-terminalextensions to the canonical fold, it possesses aflexible C-terminal helix (H11) similar to that found inBacillus TubZs, C-ST TubZ,5 and 2012-1 TubZ(PhuZ),6 which tails into a long extended coil ataround residue 300 (of 327) (Fig. 1a).Comparison ofKZ TubZ to other members of the

    tubulin/FtsZ family of proteins reveals significantsimilarity to 2012-1 TubZ (PhuZ), the proteinssuperimposing with a C RMSD of 2.4 (Fig. 1e).Given that both tubulin/FtsZs are found within the

  • Fig. 1. Monomeric and protofilament crystal structures ofKZ TubZ. (a) Cartoon representation of three subunits of theKZ TubZ protofilament crystal structure; bound GDP shown as spheres. (b) Coomassie-stained SDS-PAGE of KZTubZ protein; molecular weight standards are expressed in kilodaltons. (c) Cartoon representation of the crystal structureof a monomer of KZ TubZ annotated with the named tubulin/FtsZ secondary structural elements;19 bound GDP shownas spheres. (d) Cartoon representation of three subunits of the KZ TubZ protofilament crystal structure; bound GDPshown as spheres, rotated 180 and with the central subunit coloured grey in order to highlight the C-terminal knucklebinding site. (e) Structural alignment of the three resolved bacteriophage tubulin/FtsZs. Colour scheme for all plates:green, GTPase domain; yellow, helix 7; magenta, activation domain; cyan, C-terminal helix; nucleotide in cyan and CPKcolours.

    2166 Structure of KZ TubZ

    KZ-like family of bacteriophages and that theyshare significant sequence identity (31%), this is notunexpected. Significantly, other tubulin/FtsZ familystructures are more distantly related, including theother known phage tubulin, C-ST TubZ, the RMSDfrom these structures lying between 3 and 4 (C-ST

    TubZ, 3.8 , 16% sequence identity; in the phylo-genetic tree presented by Kraemer et al.,6 C-STcorresponds to cst, KZ to GP39, and 2012-1 toGP59). Such segregation is compatible with theinterpretation that the Pseudomonas TubZs mayexist as an evolutionarily separate subgroup,

  • Table 1. Crystallographic statistics for KZ TubZ

    Pseudomonas phageKZTubZ (monomer)

    Pseudomonas phageKZTubZ (protofilament)

    ProteinUniProt ID Q8SDC3 Q8SDC3Taxon ID 169683 169683

    CollectionBeamline DiamondI24 ESRFID14-1Wavelength () 0.9787 0.9334

    CrystalSpace group P21 P21Cell dimensiona, b, c () 41.5, 76.3, 54.1 43.5, 66.2, 52.6 () 89.9 112.7

    ScalingResolution 2.0 1.7Completeness

    (%)a99.8 (99.5) 99.2 (95.2)

    Multiplicitya 3.4 (3.3) 3.7 (2.9)Half shell

    correlationa,b0.991 (0.712) 0.998 (0.817)

    (I)/(I)a 8.9 (2.6) 14.2 (1.7)Rmerge

    a 0.124 (0.540) 0.075 (0.645)Rpim

    a 0.079 (0.348) 0.046 (0.453)Wilson B-factor 17.9 17.4


    c 0.1723/0.2103 0.1680/0.2032Bond length

    rmsd ()0.007 0.007

    Bond ang l ermsd ()

    1.126 1.010

    Most favoured(%)d

    98.3 98.7

    Disallowed (%)d 0 0

    DepositionPDB ID 3ZBP 3ZBQ

    a Values in parentheses refer to the highest recorded resolutionshell.

    b Correlation coefficient between half sets, calculated from theintensities.

    c Five percent of reflections were randomly selected beforerefinement in order to calculate Rfree.

    d Percentage of residues lying in the chosen areas of theRamachandran plot (PROCHECK).

    2167Structure of KZ TubZ

    perhaps one exclusive to the KZ-like bacterio-phages alone, although the sample number remainstoo small to be sure at this point.

    The surface of monomeric KZ TubZ isincompatible with polymerisation

    The monomeric structure of KZ TubZ exhibitsseveral clear and interesting differences from thoseso far obtained for other tubulin/FtsZ family proteins.In terms of primary structure, there is a significantinsertion within tubulin homology loop T3, whichforms an -helix instead of the conserved extendedloop found in the majority of tubulin structures (Figs.1c and 2a and c). Given the role of T3 in forming the

    subunitsubunit interface within a canonical proto-filament, this would preclude formation of such aprotofilament without significant rearrangement ofthe secondary structure. This is not the only suchsurface change, as the H10S9 loop within theactivation domain also lies in a conformation thatwould clash with an adjacent subunit (Fig. 2d) andthe partially disordered C-terminal tail prominent inprevious TubZ structures lies across the subunitinterface, blocking polymerisation. This implied to usthat either several conformational changes would berequired for polymerisation into a canonical tubulin/FtsZ protofilament or that there might be differencesin the protofilament.

    KZ TubZ undergoes structural changes inorder to form a protofilament

    In order to discover how KZ TubZ formedprotofilaments, we crystallised and solved thestructure of KZ TubZ in the presence of the weaklyhydrolysable GTP analogue GTPS. The structureof a crystal form containing a crystallographicprotofilament was obtained to 1.7 (R/Rfree =0.17/0.20) (Table 1). On examination, surprisingly,the GTPase active site was once again found to beoccupied by GDP, not GTPS, suggesting thatnucleotide hydrolysis had occurred within the drop,possibly after forming GTPase-enabled protofila-ments within the crystals. We discovered that theprotein had undergone significant and intriguingchanges in structure between the monomeric andfilamentous crystal forms (Fig. 1a and d).The structure of a protofilament of KZ TubZ

    revealed that the monomer-to-protofilament transi-tion substantially changes the conformation of theprotein, returning all three aforementioned elementsof the structure, which are inimical to the formation ofa canonical tubulin/FtsZ-like protofilament to a similarstate to that found in the majority of other tubulin andFtsZ homologues. This change entails the reorgani-sation of the C-terminal helix H11, and of both the T3and H10S9 surface loops, while both T3 and H10S9 also lose their -helical character, and would beessential for polymerisation (Fig. 2ad).A curvedstraight transition reorienting the

    GTPase and activation domains through movementof H7 has been proposed during protofilamentformation by tubulin and FtsZ; differences in domainorientation between the structures of related proteinsin different states have provided some evidence forsuch movement in the case of tubulin while thestructure of the Staphylococcus aureus FtsZ proto-filament exhibits a different domain orientation to thatobserved in most FtsZ monomers.2022 Significantly,in contrast to the surface changes required forprotofilament formation, there was no evidence ofany change in domain orientation for KZ TubZ, thedomains superimposing perfectly in both monomeric

  • 2168 Structure of KZ TubZ

    and protofilament crystal forms. This report repre-sents only the second pair of such states for a singletubulin/FtsZ-like protein, Bacillus thuringiensis TubZrepresenting the first such pair.14,23 In both cases,no evidence for domain movement during polymer-isation has been observed. The absence of such atransition for these two TubZs implies that suchmovements cannot be positively required for thepolymerisation of tubulin/FtsZ-like proteins but can-not rule them out for other cases.

    KZ TubZ forms canonical protofilaments with acompetent GTPase site

    KZ TubZ formed crystallographic protofilamentswith a subunitsubunit interface essentially identicalwith that resolved for tubulin [Protein Data Bank(PDB) ID: 1JFF] and FtsZ (PDB ID: 4DXD).22,24

    Superimposition of these three structures revealsthat loop T7, the base of helix H8 and strand S9occupy the same space in all three protofilaments(Fig. S1). The formation of a canonical tubulin/FtsZprotofilament explains the requirement for thestructural changes observed between monomericand polymeric states; the correct formation of thesubunitsubunit interface is paramount for polymer-isation, and it cannot otherwise take place.The canonical tubulin/FtsZ subunitsubunit inter-

    face also resolves the question of how catalyticactivity must occur in Pseudomonas TubZ protofila-ments and implies that hydrolysis will weaken thisinterface in the samemanner as in other tubulin/FtsZ-like proteins. Whereas the catalytic aspartate withinloop T7 of 2012-1 TubZ (PhuZ) is found 12.2 fromthe -phosphate of GDP, too long a distance forefficient nucleotide hydrolysis, in protofilaments oftubulin (PDB ID: 1JFF) and FtsZ (PDB ID: 4DXD), thisdistance is 5.7 and 7.1 , respectively.22,24 TheKZ TubZ protofilament distance of 6.9 liesbetween the tubulin and FtsZ figures (Fig. 2e). Theformation of a canonical protofilament substantiallyincreases the surface area buried at the subunitinterface. Whereas 2012-1 TubZ (PhuZ) has anextremely small longitudinal surface contact (188 2

    buried by both contributing subunits excluding theC-terminal tail), the KZ interface encompasses986 2 (buried by both contributing subunits exclud-ing the C-terminal tail), a comparable figure to that oftubulin and FtsZ protofilaments (PDB IDs 1JFF and4DXD bury 1666 and 1151 2, respectively).22,24 Theacidic knuckle region at the C-terminus of KZ TubZbecomes ordered in the protofilament, occupying theside of the adjacent subunit within the protofilament inthe same location and orientation as found in 2012-1TubZ (PhuZ) and forming a similarly sized interface(1035 2 and 1027 2 are buried by both contributingsubunits in the respective structures). This supportsthe proposed wider significance of this contact inthese proteins and implies that this contact forms

    independently from the canonical subunitsubunitinterface (Figs. 1d and 2f).6

    It seems to us that the most likely interpretation isthat the three Pseudomonas TubZ structures nowavailable fortuitously represent different snapshots ofthe polymerisationdepolymerisation cycle of theseproteins, theKZprotofilament structure representinga tight polymerised protofilament complex, and the2012-1 structure representing a weaker protofila-ment encounter/departure complex. Combining struc-tural information from both homologues, we proposethat Pseudomonas TubZs may occupy a differentconformation as a free monomer, surface loops beingincorrectly ordered and the C-terminal helix and tailextending flexibly in solution as found in the mono-meric KZ structure. GTP exchange will favourformation of the active site, ordering surface loopsand facilitating the rearrangement of T3 and H10S9so that a canonical tubulin/FtsZ subunitsubunitinterface can form as the protein polymerises. Thepresence of adjacent subunits in the protofilament willthen allow the C-terminal helix and tail to fold into theacidic knuckle, forming an extensive peptide interac-tion along the side of the protofilament, and substan-tially expanding the subunitsubunit interface.Stabilisation of the protofilament by the C-terminusmay be important during hydrolysis. Once nucleotidehydrolysis has occurred, the canonical subunitsubunit interface dissolves, destabilising this interac-tion, but leaving the C-terminal tail still ordered on thesurface of the filament, and allowing the protofilamentto move between the canonical tubulin situationrepresented by the KZ protofilament structure anda departure state similar to that of 2012-1. Thisweakened protofilament state would collapse oncethe C-terminal tail becomes disordered, releasing afree monomer.It is also possible that the weaker protofilament

    structure might represent an intermediate state inassembly. If GTP only favours organisation of theactive-site loops, rather than fixing their conforma-tion, a loose encounter complex with a longer inter-subunit spacing would result during the early stagesof assembly. GTP would then be hydrolysed slowlywithin a protofilament due to subsequent rearrange-ment of T3 and H10S9 to their active conforma-tions. In the future, it will be important to investigatethe exact role of the acidic knuckle in the polymer-isation/depolymerisation process to discover wheth-er it simply acts as a long, initial tether enhancingassembly and opposing disassembly, or whether itmight also be directly involved in the regulation ofGTPase activity.

    KZ TubZ filaments are helical and composedof intertwined protofilaments

    We went on to examine the polymerisation ofKZTubZ in bulk through light scattering. KZ TubZ

  • 2169Structure of KZ TubZ

    filaments proved extremely dynamic in the presenceof GTP; a 10-fold excess of GTP was insufficient forsignal to reach a plateau; however, a hundred-fold

    Fig. 2 (legend o

    excess was saturating, producing full polymerisation(Fig. 3a). Samples saturated with GTP were frozenin vitreous ice and viewed by electron cryo-

    n next page)

    image of Fig.2

  • Fig. 3. KZ TubZ forms dynamic polymers from intertwined protofilaments. (a) Light scattering ofKZ TubZ on additionof GTP. Cyan trace indicates dynamic polymerisation and depolymerisation on addition of 20 M GTP; magenta traceindicates polymerisation to a plateau on the addition of a saturating concentration of GTP (200 M). (b) Electron cryo-microscopy of KZ TubZ filaments with saturating concentrations of GTP, showing both polymerised bundles andseparated filaments. (c) Fourier transform of a single thick KZ TubZ filament; gyre and pitch layer lines are indicatedalongside. (d) Electron micrograph of four single thick KZ TubZ filaments, aligned to highlight the repeat and twist,adjacent to a filtered thick filament produced from the marked layer lines on its Fourier transform.

    2170 Structure of KZ TubZ

    microscopy, revealing thatKZ TubZ protofilamentscombine to form a variety of filamentous structures invitro (Fig. 3b). Thin filaments that appear to consist ofseveral intertwined protofilaments predominate;however, larger structures are also present; someof these larger bundles appear to consist of multiplethin filaments that have coalesced, while others

    Fig. 2. Conformational changes in KZ TubZ during polymand protofilament crystal structures ofKZ TubZ. (b) Expansio11 and the C-terminus on filament formation. (c) Expansion shofilament formation. (d) Expansion showing the conformationformation. (e) Structural superimposition of the protofilaments osubunit interface. (f) Structural superimposition of the C-termColour scheme: cyan, monomeric KZ TubZ; magenta/purpleprotofilament; coloured arrows denote the same region in differexpressed in angstrms.

    might possibly be fatter filaments consisting of largernumbers of protofilaments.Separated filaments suitable for analysis were

    visible in electron micrographs (Fig. 3d); Fouriertransformation of these filaments of KZ TubZindicates a subunit repeat of roughly 4.3 nm (byFourier transform). This figure is in agreement with

    erisation. (a) Structural superimposition of the monomericn showing the conformational changes undergone by helixwing the conformational change undergone by loop T3 onal changes undergone by the H10S9 loop on filamentf KZ and 2012-1 TubZ (PhuZ) comparing the subunitinal knuckle regions of KZ and 2012-1 TubZ (PhuZ)., KZ TubZ protofilament; yellow, 2012-1 TubZ (PhuZ)ent structures. All structures are C ribbons; distances are

    image of Fig.3

  • 2171Structure of KZ TubZ

    the figure for the protofilament crystal structure ofKZ TubZ (43.5 ). We interpret this repeat lengthas implying that the filamentous state ofKZ TubZ islikely to be similar in nature to the canonical tubulin/FtsZ-like protofilament state found in ourKZ crystalstructure. One small distortion relative to the crystalstructure that must be present is clear, however.Whereas filaments of 2012-1 TubZ (PhuZ) arecurrently believed to be formed by flat pairs ofprotofilaments adjacent to one another in a ribbon-like arrangement,6 a particular feature of KZ TubZfilaments visible in electron micrographs was quiteunexpected: there is a slight but visible twist to theprotofilaments within the filament, which implies thatthe filaments form a helix with a long protofilamentgyre length of ~51 nm (by Fourier transform). Theconstituent protofilaments are wrapped around oneanother, and it appears that there are at least threeprotofilaments in each bundle, as three separatetraces can be clearly followed at each filamentcrossover. The width of the filaments (~15 nm)could, however, accommodate up to four or fiveprotofilaments in theory. This is significant as thelarge distance required for each helical repeat (atleast 153 nm for a three-filament bundle) explainsclearly why a straight filament is possible in thecrystals with very little distortion; the twist over eachsubunit interface will remain low.The presence of twisted helical filaments of KZ

    TubZ is interesting as it is only the third tubulin/FtsZ-like protein to form twisted filaments that has beenidentified; the others being the Bacillus TubZs14 andthe other known bacteriophage tubulin/FtsZ, C-STTubZ.5 Notably, this encompasses both TubZ sub-groups identified within bacteriophages. One possi-ble reason for these proteins to share such a featureis convergent evolution of cytomotive filamentarchitectures; twist may be a favourable feature fortasks involving the movement of DNA or largecellular components, possibly acting as rifling toaid a linear direction of progress during polymerisa-tion. A further possibility is shared ancestry; bothproteins may have originated from a commonancestor in bacteriophages, which has adapted tocarry out different tasks in pseudolysogenic pro-phages, some of which have become plasmids5 andlarge bacteriophages requiring virion centring.6

    Could features of the Pseudomonas and theBacillus and Clostridium TubZs extend toone another?

    The similarities we have identified between the twosubgroups of TubZ proteins pose significant ques-tions and hint at interesting possibilities. The Bacillusand Clostridium TubZs are believed to transportDNA within the cell; could this perhaps prove to bethe case for Pseudomonas TubZs, the phagegenome being localised prior to encapsulation?

    The C-terminal tail of Bacillus/Clostridium TubZs isalso known to be involved in cofactor recruitment;23

    however, that of Pseudomonas TubZs is clearlyinvolved in polymerisation.6 Is it possible that both ofthese events occur in both proteins? This wouldprovide an elegant manner in which a cofactor couldaffect the polymerisation of these proteins, throughbinding to the same site. Further work will be neededto unravel the mysteries these proteins present.

    Materials and Methods


    Unless stated, chromatography equipment was provid-ed by GE Healthcare, chemicals were provided by SigmaAldrich, crystallography consumables were provided byHampton Research, and molecular graphics were gener-ated using PyMOL (Schrdinger).

    DNA, genes, and vectors

    The gene encoding Pseudomonas bacteriophage KZ(Taxon ID 169683) TubZ (UniProt ID Q8SDC3) wassynthesised codon optimised (GenScript, Hong Kong) invector pET28a. Construct pET28a-KZ-tubZ encoded thecomplete published sequence of KZ TubZ withoutmodifications and was used for light scattering andelectron microscopy, whereas construct pET28a-KZ-tubZ-His6 encoded an additional six histidine residues atthe C-terminus and was used for crystallography.


    C41 E. coli (Invitrogen) carrying either of the twoexpression vectors were grown in 12 L of 2xYT broth.Cultures were grown at 37 C and supplemented with50 g/L kanamycin, until reaching an optical density at600 nm of 0.6. Expression was induced by the addition of afinal concentration of 1 mM isopropyl--D-1-thiogalacto-pyranoside, and after expression overnight at 20 C, thecells were harvested by centrifugation at 4g.

    Protein purification

    The cell pellet from 12 L of culture was resuspended in200 mL of 100 mM TrisCl and 500 mMNaCl, pH 8.0, andbroken at 40 kPSI, 4 C, using a cell disruption system(Constant Systems). Debris was removed by centrifuga-tion at 45,000g. KZ TubZ-His6 was retrieved by nickelaffinity chromatography (5 mL HisTrap HP, 100 mM TrisCl, 500 mM NaCl, and 01 M imidazole gradient, pH 8.0),and crystallographic purity was achieved by ion exchange(1 mL HiTrap Q HP, 25 mM TrisCl, and 0500 mM NaClgradient, pH 8.0) followed by size-exclusion chromatogra-phy [HiLoad Sephacryl S200 16/60, 25 mM TrisCl,200 mM NaCl, 1 mM ethylenediaminetetraacetic acid(EDTA), and 1 mM NaN3, pH 8.0].Wild-type KZ TubZ was retrieved by stepwise (10%

    sat. step) precipitation with saturated ammonium sulfate,

  • 2172 Structure of KZ TubZ

    pH 8.0. KZ TubZ-containing pellets were pooled in25 mM TrisCl, pH 8.0, and 1 mM EDTA, followed bypurification by ion exchange (5 mL HiTrap Q HP, 25 mMTrisCl, and 0500 mM NaCl, pH 8.0) and size-exclusionchromatography (HiLoad Sephacryl S200 16/60, 25 mMTrisCl, 200 mM NaCl, 1 mM EDTA, and 1 mM NaN3,pH 8.0).

    90 Light scattering

    Light-scattering experiments were performed using aPerkin Elmer LS55 Luminescence spectrometer in 25 mMTrisCl, 200 mM KCl, 5 mM MgCl2, and 0.5 mM EDTA,pH 8.0, at 25 C with constant stirring of the 1-mL quartzcuvette. Excitation and emission wavelengths were bothheld at 400 nm, while the photon multiplier was set to650 V. KZ TubZ was added to a final concentration of2 M while either 20 or 200 M GTP was added asindicated.

    Electron microscopy

    Polymerised samples ofKZ TubZ produced as in light-scattering experiments (3 L) were applied to glow-discharged holey carbon grids (Quantifoil R2/2 Cu/Rh200 mesh; Agar Scientific) for 15 s, blotted and plunge-frozen in liquid ethane using a FEI Vitrobot. Grids weretransferred to a FEI Polara G2 microscope operated at300 kV. Images were acquired with defocus ranging from1 to 3 m on a back-thinned FEI Falcon 4k detector at76,700 nominal magnification, leading to a dose of 34 e

    2, and processed using the MRC suite for electronmicroscopy.25 The magnification and pixel resolution ofthe microscope were calibrated using the molecular latticeof graphite before we undertook our experiment.


    Initial conditions were identified at the MRC-LMB crystal-lisation facility.26KZTubZ-His6 crystals were produced in500 nL to 500 nL protein to precipitant drops: the mono-meric crystal form in 150 mM TrisCl, pH 8.0, 8.0% (v/v)ethylene glycol, and 20% (w/v) polyethylene glycol 5000monomethyl ether, and the filamentous crystal form in100 mM Nacitrate pH 5.5 and 20% (w/v) polyethyleneglycol 3000. Artificial mother liquor supplemented to 25%(v/v) glycerol was used as a cryo-protectant. Diffractionfrom KZ TubZ-His6 crystals was collected at EuropeanSynchrotron Research Facility beamline ID14eh1 andDiamond beamline I24. Data were processed withXDS,27 POINTLESS,28 and SCALA.29 Initial phaseswere determined by molecular replacement from PDB ID3R4V using Phaser,30 and themodel was built with MAIN31

    and refined with REFMAC532 and PHENIX.33

    Structural calculations and accession numbers

    Structural superimpositions and alignments were carriedout using the DALI-lite webserver.34 Surface area calcu-lat ions were performed using the PDBe-PISAwebserver.35 Coordinates and structure factors havebeen deposited in the PDB with accession numbers

    3ZBP and 3ZBQ for the monomeric and protofilamentforms of KZ TubZ, respectively.Supplementary data to this article can be found online at


    We would like to thank Fabrice Gorrec and SonjaKuhlman for their help at the MRC-LMB crystal-lisation facility, Chen Shaoxia and Gregory McMul-lan for their aid with electron microscopes at theMRC-LMB, and Yu Minmin and Raphael Gasper-Schnenbrcher for their assistance with X-raycrystallography at the MRC-LMB. We acknowledgeboth the European Synchrotron Radiation Facilityand Diamond Light Source for their excellent serviceand support. This work was supported by theMedical Research Council (grant U105184326).

    Author Contributions. C.H.S.A. and T.I.carried out all experiments. Experimental design,analysis, and manuscript preparation were per-formed by C.H.S.A., T.I., L.A.A., and J.L.

    Conflict of Interest Statement. The authorsdeclare that they have no conflict of interest.

    Received 31 January 2013;Received in revised form 7 March 2013;

    Accepted 8 March 2013Available online 22 March 2013


    filamentous protein;X-ray crystallography;

    electron cryo-microscopy

    Abbreviations used:PDB, Protein Data Bank;

    EDTA, ethylenediaminetetraacetic acid.


    1. Bergh, O., Brsheim, K. Y., Bratbak, G. & Heldal, M.(1989). High abundance of viruses found in aquaticenvironments. Nature, 340, 467468.

    2. Lwe, J. & Amos, L. A. (2009). Evolution of cytomotivefilaments: the cytoskeleton from prokaryotes to eu-karyotes. Int. J. Biochem. Cell Biol. 41, 323329.

    3. Aylett, C. H. S., Lwe, J. & Amos, L. A. (2011). Newinsights into the mechanisms of cytomotive actin andtubulin filaments. Int. Rev. Cell Mol. Biol. 292, 171.

    4. Austin, S. & Abeles, A. (1983). Partition of unit-copyminiplasmids to daughter cells. I. P1 and F

  • 2173Structure of KZ TubZ

    miniplasmids contain discrete, interchangeable se-quences sufficient to promote equipartition. J. Mol.Biol. 169, 353372.

    5. Oliva, M. A., Martin-Galiano, A. J., Sakaguchi, Y. &Andreu, J. M. (2012). Tubulin homolog TubZ in aphage-encoded partition system. Proc. Natl Acad. Sci.USA, 109, 77117716.

    6. Kraemer, J. A., Erb, M. L., Waddling, C. A.,Montabana, E. A., Zehr, E. A., Wang, H. et al.(2012). A phage tubulin assembles dynamic filamentsby an atypical mechanism to centre viral DNA withinthe host cell. Cell, 149, 14881499.

    7. Austin, S., Hart, F., Abeles, A. & Sternberg, N. (1982).Genetic and physical map of a P1 miniplasmid.J. Bacteriol. 152, 6371.

    8. Michie, K. A. & Lwe, J. (2006). Dynamic filaments ofthe bacterial cytoskeleton. Annu. Rev. Biochem. 75,467492.

    9. Ringgaard, S., van Zon, J., Howard, M. & Gerdes, K.(2009). Movement and equipositioning of plasmids byParA filament disassembly. Proc. Natl Acad. Sci. USA,106, 1936919374.

    10. Ptacin, J. L., Lee, S. F., Garner, E. C., Toro, E., Eckart,M., Comolli, L. R. et al. (2010). A spindle-likeapparatus guides bacterial chromosome segregation.Nat. Cell Biol. 12, 791798.

    11. Larsen, R. A., Cusumano, C., Fujioka, A., Lim-Fong,G., Patterson, P. & Pogliano, J. (2007). Treadmilling ofa prokaryotic tubulin-like protein, TubZ, required forplasmid stability in Bacillus thuringiensis. J. GenesDev. 21, 13401352.

    12. Chen, Y. & Erickson, H. P. (2008). In vitro assemblystudies of FtsZ/tubulin-like proteins (TubZ) fromBacillus plasmids: evidence for a capping mechanism.J. Biol. Chem. 283, 81028109.

    13. Hoshino, S. & Hayashi, I. (2012). Filament formationof the FtsZ/tubulin-like protein TubZ from the Bacilluscereus pXO1 plasmid. J. Biol. Chem. 287,3210332112.

    14. Aylett, C. H. S., Wang, Q., Michie, K. A., Amos, L. A. &Lwe, J. (2010). Filament structure of bacterial tubulinhomologue TubZ. Proc. Natl Acad. Sci. USA, 107,1976619771.

    15. Krylov, V. N., Dela Cruz, D. M., Hertveldt, K. &Ackermann, H. W. (2007). KZ-like viruses, aproposed new genus of myovirus bacteriophages.Arch. Virol. 152, 19551959.

    16. Mesyanzhinov, V. V., Robben, J., Grymonprez, B.,Kostyuchenko, V. A., Bourkaltseva, M. V., Sykilinda,N. N. et al. (2002). The genome of bacteriophageKZof Pseudomonas aeruginosa. J. Mol. Biol. 317, 119.

    17. Fokine, A., Kostyuchenko, V. A., Efimov, A. V.,Kurochkina, L. P., Sykilinda, N. N., Robben, J. et al.(2005). A three-dimensional cryo-electron microscopystructure of the bacteriophage KZ head. J. Mol. Biol.352, 117124.

    18. Pletnev, E. A., Krylov, S. V., Shaburova, O. V.,Burkal'tseva, M. V., Miroshnikov, K. A. & Krylov,V. N. (2010). Pseudolysogeny of Pseudomonasaeruginosa bacteria infected with KZ-like bacterio-phages. Russ. J. Genet. 46, 2025.

    19. Nogales, E., Downing, K. H., Amos, L. A. & Lwe, J.(1998). Tubulin and FtsZ form a distinct family ofGTPases. Nat. Struct. Biol. 5, 451458.

    20. Elsen, N. L., Lu, J., Parthasarathy, G., Reid, J. C.,Sharma, S., Soisson, S. M. & Lumb, K. J. (2012).Mechanism of action of the cell-division inhibitorPC190723: modulation of FtsZ assembly cooperativ-ity. J. Am. Chem. Soc. 134, 1234212345.

    21. Matsui, T., Yamane, J., Mogi, N., Yamaguchi, H.,Takemoto, H., Yao, M. & Tanaka, I. (2012). Structuralreorganization of the bacterial cell-division proteinFtsZ from Staphylococcus aureus. Acta Crystallogr.,Sect. D, 68, 11751188.

    22. Lwe, J., Li, H., Downing, K. H. & Nogales, E. (2001).Refined structure of alpha beta-tubulin at 3.5 resolution. J. Mol. Biol. 313, 10451057.

    23. Ni, L., Xu, W., Kumaraswami, M. & Schumacher, M. A.(2010). Plasmid protein TubR uses a distinct mode ofHTH-DNA binding and recruits the prokaryotic tubulinhomolog TubZ to effect DNA partition. Proc. NatlAcad. Sci. USA, 107, 1176311768.

    24. Tan, C. M., Therien, A. G., Lu, J., Lee, S. H., Caron, A.,Gill, C. J. et al. (2012). Restoring methicillin-resistantStaphylococcus aureus susceptibility to -lactamantibiotics. Sci. Transl. Med. 4, 126ra35.

    25. Crowther, R. A., Henderson, R. & Smith, J. M. (1996).MRC image processing programs. J. Struct. Biol. 116,916.

    26. Stock, D., Perisic, O. &Lwe, J. (2004). Robotic nanolitreprotein crystallisation at the MRC Laboratory of Molec-ular Biology. Prog. Biophys. Mol. Biol. 88, 311327.

    27. Kabsch, W. (2010). XDS. Acta Crystallogr., Sect. D,66, 125132.

    28. Evans, P. (2006). Scaling and assessment of dataquality. Acta Crystallogr., Sect. D, 62, 7282.

    29. Collaborative Computational Project Number 4 (1994).The CCP4 suite: programs for protein crystallography.Acta Crystallogr., Sect. D, 50, 760763.

    30. McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D.,Winn, M. D., Storoni, L. C. & Read, R. J. (2007).Phaser crystallographic software. J. Appl. Crystallogr.40, 658674.

    31. Turk, D. (1992). Weiterentwicklung eines Programmsfr Moleklgraphik und Elektronendichte-Manipula-tion und seine Anwendung auf verschiedene ProteinStrukturaufklrungen (PhD thesis, Technical Univer-sity of Munich, Germany).

    32. Murshudov, G. N., Vagin, A. A. & Dodson, E. J.(1997). Refinement of macromolecular structures bythe maximum-likelihood method. Acta Crystallogr.,Sect. D, 53, 240255.

    33. Adams, P. D., Afonine, P. V., Bunkczi, G., Chen, V.B., Davis, I. W., Echols, N. et al. (2010). PHENIX: acomprehensive Python-based system for macromo-lecular structure solution. Acta Crystallogr., Sect. D,66, 213221.

    34. Holm, L.&Park, J. (2000).DaliLiteworkbench for proteinstructure comparison. Bioinformatics, 16, 566567.

    35. Krissinel, E. & Henrick, K. (2007). Inference ofmacromolecular assemblies from crystalline state.J. Mol. Biol. 372, 774797.

    Structure of the Tubulin/FtsZ-Like Protein TubZ from Pseudomonas Bacteriophage KZIntroductionResults and DiscussionBacteriophage tubulin/FtsZ (TubZ) from Pseudomonas bacteriophage KZKZ TubZ adopts a tubulin/FtsZ foldThe surface of monomeric KZ TubZ is incompatible with polymerisationKZ TubZ undergoes structural changes in order to form a protofilamentKZ TubZ forms canonical protofilaments with a competent GTPase siteKZ TubZ filaments are helical and composed of intertwined protofilamentsCould features of the Pseudomonas and the Bacillus and Clostridium TubZs extend to one another?

    Materials and MethodsSourcesDNA, genes, and vectorsExpressionProtein purification90 Light scatteringElectron microscopyCrystallographyStructural calculations and accession numbers