The molecular mechanisms of prokaryotic cell division and eukaryotic mitosis seem to have little in common, yet similarities between the structures of two GTPases published in this week's issue1,2 (pages 199 and 203) will reinforce the need for microbiologists and cell biologists to compare notes. Nogales et al.1 report the 3.8-Å image reconstruction of the tubulin α/β heterodimer, which assembles into the microtubules of, for example, eukaryotic mitotic spindles. Löwe and Amos2, on the other hand, describe the 2.8-Å crystallographic structure of Methanococcus jannaschii FtsZ, which forms a filamentous ring-shaped septum at the division site of prokaryotes.
It has been said that many of the best intelligence-analysts are trained as biologists, because they can solve complex problems from fragmentary, and sometimes conflicting, evidence. The 3.8-Å structure1 of the α/β tubulin dimer is a testament to this skill. It confirms many aspects of the working model built up over the past 30 years from indirect biochemical analysis, such as the effects due to the reactivity of specific cysteine residues on ligand binding, sequence analysis and site-directed mutagenesis. The structure is, of course, more tangible than any previous model, even though its resolution is insufficient to assign functions to specific residues. But it will greatly influence studies into the mechanisms of microtubule assembly, microtubule-based motility and the effects of the commercially important anti-mitotic ligands.
Nogales et al.1 have determined the structure of Zn2+-induced sheets of anti-parallel α/β tubulin protofilaments, with GTP bound to the α-subunit, and taxolFootnote 1and GDP bound to the β-subunit. The apparent simplicity of the microtubule ultrastructure belies the fact that the assembled dimer can adopt a bewildering variety of conformational states. These lead to treadmilling3 (whereby steady-state microtubules preferentially assemble at one end and dissociate at the other) and dynamic instability4 (whereby steady-state microtubules show stochastic length changes). Direct measurements of the subunit-dissociation rate constants and the protofilament subunit-subunit periodicity5 have established that these states result from assembly-dependent hydrolysis of the β-tubulin GTP, and from differences in the kinetic properties of the two microtubule ends.
Analysis of severed microtubules indicates that there are three conformational states6 at each microtubule end. These may correspond to subunits at the two ends with bound GTP, GDP + inorganic phosphate, and GDP. In fact, this may underestimate the complexity, because different states may also be induced by binding the microtubule-associated proteins and other modulators of the assembly kinetics. Indeed, the requirement for this conformational plasticity may partly account for the extraordinarily high conservation of the tubulins — 11 per cent of the β-tubulin residues are completely conserved in a sample of 208 sequences representing almost all phylogenetic orders.
Structural information is going to be needed for each of the conformational states before the dynamic properties of microtubules are fully understood. We also need to know the relevant ligand association and conformational rate constants (many of which are likely to depend on the isotype). The 3.8-Å structure of the tubulin dimer is a veritable tour de force, but only one step towards the full story.
Determination of the structure of FtsZ- GDP by Löwe and Amos2 may represent another step. FtsZ is thought to be the bacterial homologue of the eukaryotic tubulins7, although the sequences are less than 20 per cent identical. But similarities include hydrolysis of the bound GTP during the in vitro assembly of FtsZ into filaments; the longitudinal periodicities of these filaments and the microtubule protofilaments; and the assembly of both proteins into sheets and rings or spirals8,9.
The crystallographic structure shows that FtsZ has two main domains. One of these binds a guanine nucleotide at a site that is atypical compared with the more conventional GTPases such as p21ras. The modified Rossman fold of this domain is, however, strikingly similar to that of the tubulin subunits (Fig. 1). Moreover, FtsZ and β-tubulins share many of the residues that are implicated in coordination of the bound nucleotide.
But this structural similarity between the nucleotide-binding domains of FtsZ and the tubulins presents a paradox. Is it a quaint coincidence, with atypical GTPases having unrelated functions in prokaryotic and eukaryotic cell division? Or have the two protein families evolved from a common ancestor, such that microtubules and FtsZ filaments share some functional properties? If this is the case, FtsZ may provide a way to dissect the conformational complexity of the tubulin subunit — and microbiologists and cell biologists will have a lot to talk to one another about.
*Bristol-Myers Squibb has registered Taxol as a trademark and wishes the scientific community to use the name paclitaxel.
Nogales, E., Wolf, S. G. & Downing, K. H. Nature 391, 199–203 (1998).
Löwe, J. & Amos, L. A. Nature 391, 203–206 (1998).
Margolis, R. L. & Wilson, L. Cell 13, 1–8 (1978).
Mitchison, T. J. & Kirschner, M. Nature 312, 237–242 (1984).
Hyman, A. A., Chretian, D., Arnal, I. & Wade, R. H. J. Cell Biol. 128, 117–125 (1995).
Tran, P. T., Walker, R. A. & Salmon, E. D. J. Cell Biol. 138, 105–117 (1997).
Erickson, H. P. Cell 80, 367–370 (1995).
Mukherjee, A. & Lutkenhaus, J. J. Bacteriol. 176, 2754–2758 (1994).
Erickson, H. P., Taylor, D. W., Taylor, K. A. & Bramhill, D. Proc. Natl Acad. Sci. USA 93, 519–523 (1996).
About this article
Paclitaxel-resistant Human Ovarian Cancer Cells Undergo c-Jun NH2-terminal Kinase-mediated Apoptosis in Response to Noscapine
Journal of Biological Chemistry (2002)
An Overview of Endosymbiotic Models for the Origins of Eukaryotes, Their ATP-Producing Organelles (Mitochondria and Hydrogenosomes), and Their Heterotrophic Lifestyle
Biological Chemistry (2001)