COMMENTARY

When cytoskeletal worlds collide

Eva Nogales The Howard Hughes Medical Institute, Molecular and Biology Department, University of California, Berkeley, CA 94720- 3220; and The Lawrence Berkeley National Laboratory, Berkeley, CA 94720

ytoskeletal aficionados and mo- domain of the next along the filament, lecular evolutionists are in for burying the nucleotide between them (the C a surprising treat in PNAS. C-terminal domain contributing essential Löwe and colleagues, who have residues for nucleotide hydrolysis, and for some years now brought to our atten- thus coupling hydrolysis with polymeriza- tion the conservation of the and tion) (10, 11). Surprisingly, TubZ main- across kingdoms tains the same orientation of the N- through their structural studies (1, 2), are terminal and C-terminal domains across now giving us new striking images to think an interface as that used by tubulin along about (3). Just test your knowledge of protofilaments, but these two domains, structure by looking at their within one subunit, have dramatically ro- figure 4. Do not think twice and say aloud tated with respect to each other compared what you think those filaments are. Now with the tubulin and FtsZ cases. The result read the title of their article. Surprised? Fig. 1. Distinct filament structure with the same is that whereas the latter form linear ar- Read more. assembly interfaces. Tubulin (brighter colors) and rays where one subunit is simply translated TubZ was recently identified as a tubu- TubZ (lighter colors) share a conserved interface along the filament axis, in TubZ there is along the filament that sandwiches the nucleotide lin/FtsZ-like involved in plasmid (shown in green). The twist in the TubZ filament a twist reminiscent of that in a single segregation in (4). This process (filament on the right) originates from a rotation F-actin strand. TubZ has produced a radi- appears to require a minimal set of two of the two domains within one subunit (compare cally different polymer structure while plasmid-encoded genes, one correspond- relative orientation of the magenta and blue do- maintaining the main longitudinal in- ing to a centromere-binding protein, the mains for bright and light subunits). Figure cour- terface, which is under very strong evolu- other a self-assembling, filament-forming tesy of Gabriel C. Lander. tionary pressure to preserve nucleotide element that in many cases is either an binding. Instead, TubZ has significantly actin-like (ParM) or a tubulin-like pro- significant when it concerns the residues modified its overall tertiary structure via tein (TubZ) (4, 5). This minimalist seg- involved in nucleotide binding. Impor- domain rearrangements (Fig. 1). The regation system parallels the complex tantly, and true for all members of the change is driven by a new structural ele- kinetochore– spindle appara- tubulin family thus far identified, the nu- ment at the N-terminal end of TubZ (helix tus of eukaryotes, in a fitofefficient sim- cleotide site happens to sit at the mono- H0) that wedges between the N-terminal plicity. Although some obvious parallels mer–monomer longitudinal interface. This and C-terminal domains. This same helix do exist, they break down once we start arrangement is in strong contrast to what appears to play a second role in the in- thinking about the molecular details. The occurs for actin and its bacterial homologs, teraction between strands in the double dynamic character of the cytoskeletal where the ATP binding site is not directly filament (3). polymer is in both cases essential for the involved in the contact between mono- It is interesting to compare the actin and movement of plasmids/chromosomes (6). mers. Interestingly, ParM lacks the poly- tubulin assemblies with those of the third But whereas move chromo- merization interfaces present in actin and cytoskeletal system involving a nucleotide- somes to the poles by depolymerization uses an alternative assembly arrangement binding protein: septin filaments (thus far during anaphase (7), the ParM and TubZ (filaments are left-handed rather than identified only in eukaryotes—but see ref. 6 filaments seem to act mostly by pushing right-handed) (9), although ultimately ac- for an interesting hypothesis). Interestingly, during polymer growth or relying on tin and ParM filaments are clearly struc- these three cytoskeletal systems play in- treadmilling (4, 8). The work of Aylett turally related. So, how can it be that termingled roles in cell division across et al. now indicates that the distinct roles tubulin and TubZ, which are highly con- kingdoms: in eukaryotes, tubulin segregates that TubZ and tubulin play in plasmid and served at assembly interfaces, give rise to chromosomes in mitosis, whereas septins chromosomal segregation, respectively, such distinct polymers? It is not the in- assemble abut to the membrane in cytoki- determine the very different architectures terface, but the protein structure, that has nesis to define the site of septation and re- of the TubZ filament and the microtubule. significantly changed! cruit an actomyosin ring for constriction Whereas tubulin forms hollow cylinders Following the nomenclature used by (12). In bacterial cell division, FtsZ defines of laterally interacting, (almost) straight Aylett et al., the structure of the tubulin the site of septation and carries out the protofilaments, TubZ forms a two- family members consists of an N-terminal, constriction itself, whereas plasmids are stranded filament that resembles actin and nucleotide-binding domain and a C- segregated by ParM or TubZ filaments. ParM! Löwe and colleagues put forward terminal domain (for tubulin, the term “C- Whereas F-actin is polymerized from several arguments about why these two- terminal domain” has traditionally been actin monomers, and microtubules are stranded TubZ and ParM filaments would reserved for the tubulin-unique C-terminal made of αβ-tubulin heterodimers, the as- be optimal for their role in plasmid seg- helices that define the crest of protofila- sembly units for septin filaments are non- regation. Although this is a hard question ments on the microtubule—and which are polar multimers, of different order from to answer, it is interesting to think of how lacking in FtsZ and TubZ—whereas the the distinctiveness of the TubZ and tubu- rest of the protein up to the nucleotide- lin protofilaments could arise. binding domain has been referred to as the Author contributions: E.N. wrote the paper. As when comparing tubulin with FtsZ, “intermediate domain”). For all family The author declares no conflict of interest. the sequence conservation between tubu- members, the N-terminal domain of one See companion article on page 19766. lin and TubZ is very low overall but very subunit interacts with the C-terminal E-mail: [email protected].

www.pnas.org/cgi/doi/10.1073/pnas.1014665107 PNAS | November 16, 2010 | vol. 107 | no. 46 | 19609–19610 Downloaded by guest on September 26, 2021 organism to organism, that elongate into actin-like and tubulin-like systems. On the Each tubulin-like protein characterized a nonpolar filament by end-on interaction other hand, they do have lateral polarity thus far appears to form a distinct type of (13–15). The septin filament has no twist, (left–right polarity) that is present in the polymer that has been optimized for its just like the tubulin protofilaments and FtsZ and tubulin protofilaments (but not function: microtubules for eukaryotic tu- unlike the TubZ or actin filaments. Inter- in microtubules!) and is lacking in fila- bulin, straight filaments for FtsZ, and estingly, each septin protein along this ments that twist, such as actin, ParM, and twisted, double-stranded filaments for arrangement makes two different types of TubZ. It is interesting that eukaryotic tu- interactions. One involves septin-unique bulin itself is capable of assembly into regions that extend N- and C-terminal of Each tubulin-like protein different polymer forms, involving differ- the Ras-like GTP-binding domain. The characterized thus far ent curvatures and lateral arrangements, other involves the nucleotide-binding do- although always via protofilaments with mains of two consecutive septins, where appears to form a distinct little or no twist. There is an innate ca- the nucleotides themselves constitute a pacity of the tubulin subunits to use al- significant part of the interface (13). Thus, type of polymer. ternative lateral interfaces, which may be along the septin filaments, interfaces that unique to eukaryotic , as well as involve a highly conserved nucleotide- to change conformation at the monomer binding domain (like tubulin and TubZ) TubZ. This septin and FtsZ property level (16). The latter occurs via rotations alternate with interfaces away from the seems to make sense functionally, as fila- of the N-terminal versus C-terminal do- nucleotide (like in actin and ParM). Al- ments must interact with the plasma mem- main that are slightly reminiscent of, but though both interfaces are relatively well brane (or a membrane-bound protein) qualitatively distinct from, those seen conserved across septins, that involving along their length using one side, while the when comparing tubulin with TubZ. Im- the nucleotide is markedly more so, in other, facing the cytosol, is available for portantly, some of these conformational agreement with the concept that retain- binding to other cellular factors. Finally, changes give rise to alternative polymer ing nucleotide binding is a stronger evo- septin filaments appear to lack the “cyto- forms that correspond to intermediates lutionary constraint than maintaining self- motive” property of actin and tubulin in the microtubule assembly and disas- assembly contacts. families (6), or at least lack the nucleotide- sembly processes (17, 18). Such inter- Septins lack the polarity along the fila- dependent dynamic behavior of actin mediates are likely to play important ment (up–down polarity) present in all and tubulin. cellular roles (19).

1. Löwe J, Amos LA (1998) Crystal structure of the bacte- 8. Møller-Jensen J, et al. (2003) Bacterial mitosis: ParM of 14. John CM, et al. (2007) The Caenorhabditis elegans sep- rial cell-division protein FtsZ. Nature 391:203–206. plasmid R1 moves plasmid DNA by an actin-like inser- tin complex is nonpolar. EMBO J 26:3296–3307. 2. van den Ent F, Amos LA, Löwe J (2001) Prokaryotic tional polymerization mechanism. Mol Cell 12:1477– 15. Bertin A, et al. (2008) Saccharomyces cerevisiae septins: – origin of the actin cytoskeleton. Nature 413:39 44. 1487. Supramolecular organization of heterooligomers and fi 3. Aylett CHS, Wang Q, Michie KA, Amos LA, Löwe J 9. Orlova A, et al. (2007) The structure of bacterial ParM the mechanism of lament assembly. Proc Natl Acad – (2010) Filament structure of bacterial tubulin homo- filaments. Nat Struct Mol Biol 14:921–926. Sci USA 105:8274 8279. – logue TubZ. Proc Natl Acad Sci USA 107:19766 19771. 10. Nogales E, Wolf SG, Downing KH (1998) Structure of 16. Gigant B, et al. (2000) The 4 Å X-ray structure of a tu- 4. Larsen RA, et al. (2007) Treadmilling of a prokaryotic bulin:-like domain complex. Cell 102:809–816. the αβ tubulin dimer by electron crystallography. Na- tubulin-like protein, TubZ, required for plasmid stabil- 17. Chrétien D, Fuller SD, Karsenti E (1995) Structure of ture 391:199–203. ity in Bacillus thuringiensis. Genes Dev 21:1340–1352. growing microtubule ends: Two-dimensional sheets 11. Oliva MA, Cordell SC, Löwe J (2004) Structural insights 5. Møller-Jensen J, Jensen RB, Gerdes K (2000) Plasmid close into tubes at variable rates. J Cell Biol 129: into FtsZ protofilament formation. Nat Struct Mol Biol and chromosome segregation in . Trends 1311–1328. – Microbiol 8:313–320. 11:1243 1250. 18. Mandelkow E-M, Mandelkow E, Milligan RA (1991) Mi- 6. Löwe J, Amos LA (2009) Evolution of cytomotive fila- 12. Tolliday N, Bouquin N, Li R (2001) Assembly and regu- crotubule dynamics and microtubule caps: A time-re- ments: The cytoskeleton from prokaryotes to eukar- lation of the cytokinetic apparatus in budding yeast. solved cryo-electron microscopy study. J Cell Biol 114: yotes. Int J Biochem Cell Biol 41:323–329. Curr Opin Microbiol 4:690–695. 977–991. 7. Koshland DE, Mitchison TJ, Kirschner MW (1988) Pole- 13. Sirajuddin M, et al. (2007) Structural insight into fila- 19. Nogales E, Wang HW (2006) Structural intermediates wards chromosome movement driven by microtubule ment formation by mammalian septins. Nature 449: in microtubule assembly and disassembly: How and depolymerization in vitro. Nature 331:499–504. 311–315. why? Curr Opin Cell Biol 18:179–184.

19610 | www.pnas.org/cgi/doi/10.1073/pnas.1014665107 Nogales Downloaded by guest on September 26, 2021