The surface lattice of microtubules in vivo consists of 13 protofilaments with a 12 nm pitch on a three start helix (Figure 6). There is an axial repeat of 4 nm which corresponds to half the diameter of the heterodimer, i.e., the diameter of a single alpha- or beta-tubulin. There is an approximately 0.9 nm stagger between adjacent protofilaments thus totalling 12 nm for 13 protofilaments. This is three times the axial monomer repeat, therefore three helices or a three-start helix is required to fill the lattice, each with an inclination of approximately ten degrees. Along the helix are either heterologous (-alpha-beta- and -beta-alpha-) or homologous (-alpha-alpha- and -beta-beta-) lateral interactions. These lateral interactions between filaments can be organized in two possible ways, the A-lattice and the B-lattice (Mandelkow et al., 1995; Amos, 1995) (See Figures 6 and 7). To complete the microtubule structure, a variety of proteins may be attached to the surface of the lattice; for example, microtubule based motor proteins such as dynein or kinesin, and microtubule-associated proteins (MAPs). These proteins may bind along the cylindrical structure at specific points (Díaz-Nido et al., 1990; also see text edited by Kreis and Vale, 1993).
Amos and colleagues propose the A-lattice structure (Amos, 1995), with staggered contacts between adjacent protofilaments, for native single microtubules. They believe this to be the true structure for the complete A-subfibre of flagellar-doublet microtubules (refer back to Figure 1), with the presence of B-lattice structures only existing as the result of faulty reassembly or in the B-subfiber, which is the incomplete cylinder appended to the A-subfibre in flagellar-doublet microtubules (refer back to Figure 1). The terminology A-lattice and B-lattice actually arises from this proposed relationship to the structure of A- and B-subfibres. The A-lattice has a perfectly helical symmetry, containing the usual 13 protofilaments (Tilney et al., 1973). The tubulin heterodimers form polar protofilaments with beta-tubulin located on the growing plus-end of the microtubule (Mitchison, 1993) (Figure 6A). The alpha-tubulin is thus found at the opposite pole (the minus-end) bound to gamma-tubulin and associated with the MTOC. Amos postulated that in vivo, 13 gamma-tubulin subunits group together with other associated proteins to initiate the 13 protofilament microtubule structure. The associated proteins would require three gamma-tubulin binding sites and one alpha/beta-dimer binding site due to the three start helical configuration as indicated in Figure 6A. On the plus-end there are multiple places for addition of tubulin for a growing microtubule, conversely there are as many sites to lose tubulin for a shrinking microtubule.
Research by Mandelkow's group on the diffraction patterns of in vitro reassembled microtubules decorated with the head group of the motor protein kinesin indicated that all microtubules exist as a B-lattice, with aligned contacts between adjacent protofilaments (Mandelkow et al., 1995). Mandelkow's findings are supported by other researchers using X-ray diffraction and electron microscopy of axonemal microtubules and repolymerized brain microtubules (Linck and Langevin, 1981; Wais-Steider et al., 1987; Lanzavecchia et al., 1994). On a complete cylindrical microtubule, the B-lattice shows a seam of discontinuity, thus having no symmetry in vitro (Figure 6B). This seam is postulated to contain A-lattice type interactions between protofilaments. Mandelkow also believes that the alpha-tubulin is on the plus-end while the beta-tubulin subunit is found at the distal end associated with gamma-tubulin and the MTOC. The lattice would require an MTOC with 13 gamma-tubulin binding sites and a single alpha/beta-dimer binding site as shown in Figure 6B. In summary, both Amos and Mandelkow have shown good evidence for a particular lattice, and this debate may only be settled by determining a high resolution X-ray diffraction structure of tubulin.
A more recent study by Nogales et al. (1995) determined the structure of tubulin to 6.5 Å resolution. Their three-dimensional reconstruction of tubulin resulted from data obtained by electron crystallography of zinc-induced two-dimensional crystals of tubulin with bound taxol. Zinc-sheets are planar polymeric structures formed by tubulin in the presence of zinc (Larsson et al., 1976; Tamm et al., 1979). These highly ordered structures are composed of a lattice of protofilaments in antiparallel order, unlike the cylindrical microtubule where protofilaments are parallel. Interactions occur which prevent the formation of cylindrical tubules, but instead allow the formation of flat sheets. In both the tubular and sheet formation, the alternating alpha- and beta-tubulin arrangement is maintained, with only the interactions between protofilaments being altered. The data obtained by Nogales and colleagues showed that the external surface of tubulin was fairly smooth in comparison to the internal surface, which contained ragged projections due to the presence of beta-sheets. The presence of an alpha-helix was observed, presumably at the carboxyl-terminal, as previously predicted from the sequence studies of Little et al. (1981). Furthermore, beta-sheets appear to comprise a substantial part of the axial contacts between alpha- and beta-tubulin molecules within a single protofilament. They were able to determine the taxol binding site on the outside surface of the beta-tubulin subunit, but were not able to trace the polypeptide chain or to clearly define a domain organization that would indicate the possible nucleotide binding site. Furthermore, though the observable resolution within the plane was quite good, the resolution on the third axis perpendicular to the plane of the two-dimensional crystal was much poorer. Furthermore, the study of zinc-sheets cannot resolve the issue of the A- or B-lattice of natural microtubules, since the protofilament arrangement is antiparallel.
Many microtubule functions are based on tubulin's ability to polymerize and depolymerize readily (Kirschner, 1978). For polymerization, it has been shown that tubulin can bind two molecules of GTP at two different sites, referred to as the E-site and the N-site. The E-site binds one molecule of GTP and allows free exchange with the GTP in solution while the N-site binds a single GTP molecule tightly and non-exchangeably. The E-site is known to be situated on the -tubulin subunit (Linse and Mandelkow, 1988; Hesse et al., 1987; Geahlen and Haley, 1977). There is still some question as to the location of the N-site, though the high homology of the alpha- and beta-tubulin sequences suggests that this site is within the alpha-tubulin subunit, and is occluded by formation of the alpha/beta-dimer. The association constants for the binding of GTP and GDP to beta-tubulin are approximately 5 × 107 M-1 and 2.5-2.7 × 107 M-1 (respectively) in the presence of Mg2+ and 1.6 × 107 M-1 and 1.4 × 104 M-1 (respectively) in the absence of Mg2+ (Correia et al., 1987; Mejillano and Himes, 1991). Many GTP analogues, including non-hydrolysable forms such as GMPPNP, can also bind at the E-site.
GTP hydrolysis at the E-site is thought to play a key role in generating or controlling microtubule dynamic properties. Microtubules continuously exchange tubulin subunits with those in the unassembled tubulin pool at a higher rate than expected for simple equilibrium by way of two identified dynamic behaviours: (1) treadmilling (Cleveland, 1982) and (2) dynamic instability (Mitchison and Kirschner, 1984). The first process discovered was treadmilling (Figure 8A). In treadmilling there is an inherent asymmetry in the microtubule resulting from the differential rates of subunit gain and loss at the two microtubule ends. Though treadmilling is too slow to contribute to motile processes in vivo, it was used to define the orientation of a microtubule with regard to the plus and minus ends. Current research places more emphasis on dynamic instability (Figure 8B), for which observed rates are consistent with in vivo processes.
Both dynamic instability and treadmilling arise from inherent asymmetry of the two microtubule ends, reinforced by the driving force of GTP hydrolysis (Mejillano et al., 1990) catalyzed by a GTPase activity within tubulin, which is stimulated when a tubulin-GTP dimer adds to the microtubule end (Andreu and Timasheff, 1981). Treadmilling results from differential stability of the two microtubule ends (Figure 8A), the energy discrepancy for subunit addition being made-up by GTP hydrolysis.
In cell-free polymerization reactions, addition of tubulin occurs primarily at the ends of pre-existing microtubules (Lodish et al., 1995). If a preparation of pure tubulin heterodimers is used in an assembly reaction (at elevated temperatures of 30-37oC and at typical tubulin concentrations greater than 10 µM) tubulin-GTP will spontaneously polymerize into microtubules, though the initial rate of formation of microtubules is very slow (for a review, see Bayley et al., 1994). Addition of fragments of microtubules increases this rate of formation (Lodish et al., 1995), indicating that microtubule nucleation is slower than elongation of an existing polymer lattice.
In the process of dynamic instability (Figure 8B), microtubules are anchored at the minus-end (as identified from kinetic properties), while elongation occurs at the plus-end. Anchorage at the minus-end occurs naturally in vivo by growth of microtubules out of the centrosome. The binding of tubulin-GTP to the microtubule plus-end stabilizes the growing polymer while hydrolysis of in situ GTP destabilizes it. A stabilized plus-end can now elongate, since addition of tubulin-GTP exceeds the rate of GTP-hydrolysis. In contrast, an unstable plus-end shrinks when hydrolysis of GTP is faster than subunit addition, leading to the loss of the cap and destabilisation of the end. Thus, growth and shrinkage of microtubule ends is a dynamic process, with individual members of the microtubule population growing or shrinking under identical global conditions. Under defined conditions of ion composition, there is a free tubulin subunit concentration at which steady-state turnover is reached; that is, the addition of dimers to some microtubules is balanced by the loss of dimers from others. This is referred to as the critical concentration, Cc, of tubulin. Above this concentration, microtubules grow, below it, they tend to shrink. This ability to assemble and disassemble at will allows for quick local responses in cell shape to environmental stimuli (Kirschner and Mitchison, 1986). This behaviour has been observed in different cell types (Hayden et al., 1990; Mitchison, 1989) and is thought to be important in mitotic spindle morphogenesis, chromosome separation and cell shape changes (Kirschner and Mitchison, 1986).
During the purification of tubulin, certain proteins were found to co-purify with the microtubule fraction (Keates, 1984; Weingarten, et al., 1975; Murphy and Borisy, 1975; Murphy et al., 1977). These proteins can be separated from tubulin by ion-exchange chromatography (Keates, 1984). These co-purifying proteins (or MAPs) have been shown to promote microtubule self-assembly (Weingarten et al., 1975). The term MAP may be used as a generic abbreviation for proteins co-purifying with microtubules, but among the MAPs are a number of specific and well characterized proteins identified by a numeric suffix, e.g. MAP1A, MAP2A, etc.
MAPs can be loosely divided into groups depending on their molecular weight. Two groups contain high molecular weight proteins and are termed MAP1 (Wiche et al., 1991; Müller et al., 1994) and MAP2 (Matus, 1994). MAP1 can be resolved into three unrelated polypeptides: MAP1A, MAP1B and a cytoplasmic dynein termed MAP1C (Paschal and Vallee, 1987). MAP1A has been shown to be distributed along the lengths of microtubules and is thought to contribute to their stability (Schoenfeld et al., 1989), while the function of MAP1B (which has been called MAP5 by some authors) is yet unknown, although it is thought to have a role in neuronal morphogenesis (Aletta et al., 1989). MAP2 is a group of heat-stable high molecular weight proteins which exist in the closely related forms MAP2A and MAP2B and a lower molecular weight form, MAP2C, a product of alternate splicing of the MAP2B gene (Matus, 1988). Another group are small molecular weight (50-70 kDa) proteins known as tau factors (Goedert et al., 1994; Cleveland et al., 1977a, b; Lee et al., 1988; Weingarten et al., 1975). Other characterized MAPs also include the 220 kDa MAP4 (Parysek et al., 1984; Olmsted et al., 1986; Bulinski, 1994), the 180 kDa MAP3 (Huber et al., 1985) and the 250K MAP in Drosophila (Irminger-Finger et al., 1990). Further classes of MAPs can also be found in other cell types (for reviews, see Chapin and Bulinski, 1992; Bloom, 1992). Table 1 shows a summary of different MAP proteins, their respective molecular weights in kDa as indicated by SDS-PAGE gels and from known sequences, as well as their general distribution in cells.
It should be noted that the nucleotide sequence of many of the genes encoding these proteins project smaller masses for the polypeptides (Lee et al., 1988; Noble et al., 1989; Langkopf et al., 1992; Müller et al., 1994) than the original gel electrophoresis determination, which suggested higher molecular masses. This indicates proteins with surprising qualities. The abnormally high values of apparent molecular weight are consistent with the anomalous mobilities in SDS-PAGE of filamentous proteins (Goedert et al., 1991).
MAPs generally act as regulators of microtubule dynamics and have been shown to play a role in the nucleation and elongation of microtubules by increasing the yield of polymerization in vitro and by conferring stability in vivo. MAPs can bind tubulin and stabilize its polymerization into microtubules in vivo (Cleveland et al., 1977a, b).
In addition to MAPs, motor proteins also associate themselves with tubulin and microtubules (Bloom, 1992). Motor proteins such as dynein and kinesin convert chemical energy from ATP hydrolysis into force and motion along the microtubule. Motor proteins move along the microtubule proximal surface either towards the distal end, which in vivo is the plus-end (generally kinesins) or towards the minus-end anchored at the centrosome in vivo (generally dyneins) (Porter et al., 1987). For example, kinesin hydrolyzes ATP to ADP and uses the released chemical energy for anterograde (towards the microtubule plus-end) transport of organelles along the microtubule in vivo (Vale et al., 1986). Conversely, non-claret disjunctional (ncd), a microtubule motor found in Drosophila, is identified as a kinesin by its sequence, but is instead a minus-end directed motor protein (Scholey et al., 1985). MAP1C or cytoplasmic dynein is thought to be responsible for retrograde (towards the microtubule minus-end) organelle movement and possibly for some aspects of mitotic movement (Paschal and Vallee, 1987). Figure 9 shows a schematic representation of two motor proteins associated with microtubules. Vesicles move along the surface of microtubules through the interaction between the microtubules and organelles, driven by kinesin, an ATPase motor protein which exerts a translocating and outward directing force along the microtubule (Vale et al., 1986); similarly, dynein acts to direct transport to the proximal end of a microtubule (Vale and Toyoshima, 1989). The consistency of direction arises from the intrinsic polarity of microtubules, which have a uniform radial orientation in the cell (Heideman, 1986).
Colchicine has long been acknowledged as being cytotoxic, due to its interference with the mitotic spindle, even before microtubules had been recognized ultrastructurally (Dustin, 1984). The tubulin-colchicine complex inhibits addition to growing microtubules (Keates and Mason, 1981). Colchicine acts to disrupt the mitotic spindle of a cell by preventing polymerization of tubulin to microtubules. The action of this toxin is not limited to mitosis, and it is quite detrimental to non-cancerous growing cells, as well as being neurotoxic due to its disruption of neuronal microtubules (Hamel, 1990; Wilson and Meza, 1973). Another medical use for colchicine is its proven effectiveness against gout (Gutman, 1965) through an as yet uncharacterized mechanism.
The anti-tumour agents vinblastine, vincristine and vindesine are a class of molecules called Vinca alkaloids. These compounds halt cell division by disrupting tubulin polymerization, with formation of abnormal helical structures leading to the formation of polytene cells (Raxworthy, 1988). Vinca alkaloids show slight specificity for blood cells, and show enhanced cytotoxicity mostly towards cells having high rates of proliferation, such as bone marrow cells and cancer cells. These drugs are unable to penetrate the blood-brain barrier and are therefore not as neurotoxic as colchicine (Hamel, 1990).
The compound taxol is a recent and highly promising chemotherapeutic agent. During cell division microtubules are the main functional components of the mitotic spindle and the targets of mitosis-arresting drugs (see Hamel, 1990, for a review). Isolated from the Pacific yew tree (Taxus brevifolia), taxol is unusual in that it has the ability to induce microtubule assembly (Schiff et al., 1979). From photoaffinity labelling experiments, taxol was found to bind within the first 31 residues of the amino terminus of beta-tubulin (Rao et al., 1994). Taxol has been found to stabilize normal microtubules, promoting tubulin polymerization and rendering them resistant to depolymerization. This action prevents the changeover from interphase microtubules to mitotic spindle microtubules, blocking further proliferation of cancer cells. The unique activity arrests cell division but will not kill normal non-dividing cells (Hamel, 1990; Schiff and Horowitz, 1980).
Drugs acting on tubulin function in mitosis may serve as possible cures for cancers. The action of colchicine is too detrimental for therapeutic use due to its toxicity, but others such as vinblastine and taxol have been found to be useful drugs. Vinca alkaloids have proven to be 90% effective against Hodgkin's disease, and taxol has been effective in clinical trials against human breast and ovarian carcinomas (McGuire et al., 1989). Taxol or taxol-like compounds are very promising, but use is restricted by limited natural availability and difficulty of synthesis of these compounds. Taxol is found in small quantities in the rare Pacific yew trees and organic synthesis of taxol-like molecules yields small amounts through expensive and lengthy multi-step processes (Nicolaou et al., 1994). If the three-dimensional structure of tubulin was obtained, it could aid in efforts to design taxol analogues which might be less difficult to synthesize yet still bind in the same site. Details of the taxol-tubulin interaction could also be used as a guide to design taxol analogues that would bind more tightly to tubulin and have improved pharmacological properties. From a full structure of tubulin, a better understanding of the polymerization process and dynamic instability may result. Since tubulin is important in many aspects of eukaryote cell morphology and dynamics, knowledge of its structure would help to reveal the ways in which a variety of cellular processes are controlled.