The XMAP215 (Xenopus microtubule-associated protein 215) and CLASP [CLIP-170 (cytoskeletal linker protein 170) associated protein] microtubule plus end tracking families play central roles in the regulation of interphase microtubule dynamics and the proper formation of mitotic spindle architecture and flux. XMAP215 members comprise N-terminally-arrayed hexa-HEAT (huntingtin, elongation factor 3, the PR65/A subunit of protein phosphatase 2A and the lipid kinase Tor) repeats known as TOG (tumour overexpressed gene) domains. Higher eukaryotic XMAP215 members are monomeric and have five TOG domains. Yeast counterparts are dimeric and have two TOG domains. Structure determination of the TOG domain reveals that the six HEAT repeats are aligned to form an oblong scaffold. The TOG domain face composed of intra-HEAT loops forms a contiguous, conserved tubulin-binding surface. Nested within the conserved intra-HEAT loop 1 is an invariant, signature, surface-exposed tryptophan residue that is a prime determinant in the TOG domain–tubulin interaction. The arrayed organization of TOG domains is critical for the processive mechanism of XMAP215, indicative that multiple tubulin/microtubule-binding sites are required for plus end tracking activity. The CLASP family has been annotated as containing a single N-terminal TOG domain. Using XMAP215 TOG domain structure determinants as a metric to analyse CLASP sequence, it is anticipated that CLASP contains two additional cryptic TOGL (TOG-like) domains. The presence of additional TOGL domains implicates CLASP as an ancient XMAP215 relative that uses a similar, multi-TOG-based mechanism to processively track microtubule ends.
A web of microtubule plus end tracking proteins
Microtubules are highly dynamic polymers of the αβ-tubulin GTPase that oscillate between phases of growth, pause, depolymerization (catastrophe) and rescue . While dynamics can be reconstituted in vitro with purified tubulin and GTP, the observed dynamics do not correlate with those observed in vivo. The microtubule plus end is the prime, polarized site for microtubule dynamics. It is here that a number of microtubule associated proteins termed +TIPs have been observed to dynamically localize, modulate microtubule dynamics, recruit cellular factors and translate cellular signalling cues into cytoskeletal outcomes. The four dominant +TIPs families are EB1 (end-binding 1; because it was serendipitously found to bind to the C-terminus of the adenomatous polyposis coli tumour suppressor), CLIP-170 (cytoskeletal linker protein 170), CLASP (CLIP-170- associated protein) and the family defined by the XMAP215 (Xenopus microtubule-associated protein 215) [2–6]. The domains in these proteins that mediate microtubule plus end localization are: EB1, a calponin homology domain; CLIP-170, CAP-Gly (cytoskeletal associated protein glycine-rich) domains; and XMAP215 and CLASP, TOG (tumour overexpressed gene) domains, named after the human XMAP215 homologue ch-TOG . While these three +TIP domains are structurally diverse, one central, common theme is that multiple domains are either required for or greatly enhance microtubule plus end localization . Whether multiple +TIP domains enable an avidity effect for microtubule plus end localization or are part of a formin-like hand-over-hand processivity mechanism remains to be determined [8–10]. In addition to core +TIP domains, the above four +TIP families all contain binding sites for one or more of the other +TIP proteins. This web of +TIP interactions has complicated the in vivo analysis of single +TIP members and, as a result, mechanistic analysis has shifted to in vitro microtubule plus end tracking reconstitution assays. To date, the solo plus end tracking activity of EB1 and XMAP215 members has been reconstituted, while CLIP-170 plus end tracking activity has been reconstituted only in the presence of EB1 . Studies in the +TIP field now focus on defining the mechanism of these proteins – whether they co-polymerize with tubulin, track processively or some combination thereof. Recent in vitro work shows that EB1 can processively track plus ends and promote formation of microtubules with A-lattice geometry instead of the standard in vitro B-lattice, implicating EB1 as a processive microtubule polymerization chaperone . In contrast with EB1 and CLIP-170, where microtubule plus end tracking activity is conferred by small N-terminal domains, robust plus end tracking activity of XMAP215 and CLASP requires full-length protein, making structure–function studies of these families an in vitro challenge.
The TOG domain plus end tracking protein families XMAP215 and CLASP
XMAP215 and CLASP localize to microtubule plus ends during interphase, promoting growth and modulating dynamics in a cell cycle-dependent manner [13,14]. Depletion of the Drosophila counterparts have dramatic mitotic phenotypes: depletion of Drosophila CLASP [MAST (multiple asters)] yields multiple asters, while depletion of Drosophila Msps (Minispindles) yields a spindle in accordance with its name as well [15,16]. XMAP215 family members come in two general domain organization schemes characterized by arrayed TOG domains (Figure 1A). In all cases, TOG domain arrays oscillate between domains with a net negative charge followed by a domain with a net positive charge. The yeast homologues Stu2p (Saccharomyces cerevisiae) and Dis1p (Saccharomyces pombe) have two N-terminal TOG domains followed by a coiled coil domain that mediates homodimerization. C-terminal to the dimerization domain is a basic region implicated in microtubule binding . The higher eukaryotic XMAP215 members are monomeric and have five arrayed TOG domains at their N-terminus followed by a conserved, unique CTD (C-terminal domain) with predicted HEAT (huntingtin, elongation factor 3, the PR65/A subunit of protein phosphatase 2A and the lipid kinase Tor) repeats and microtubule binding capability [18,19]. The CTD also confers binding to the TACC (transforming acidic coiled coil) protein which recruits XMAP215 to the centrosome during mitosis . In contrast, CLASP family members contain a single canonical N-terminal TOG domain (Figure 1A). The C-terminus of CLASP contains a conserved domain that binds CLIP-170 . During mitosis, CLASP is recruited to the kinetochore where it acts to incorporate microtubule subunits into kinetochore fibres [21,22].
Organization of TOG domains in the XMAP215 and CLASP families of microtubule plus end tracking proteins and conservation within the first intra-HEAT loop
Structure of the TOG domain
To date, the structures of three TOG domains have been determined by X-ray crystallography [7,23]. All belong to the XMAP215 family and include the S. cerevisiae member, Stu2p, TOG domain 2; the Drosophila melanogaster Msps TOG domain 2; and the Caenorhabditis elegans Zyg-9, TOG domain 3. All three structures represent TOG domains that can be subclassified as having a net positive charge. The structure of the TOG domain is an oblong scaffold formed by six HEAT repeats A–F, arrayed in a linear fashion with overall dimensions of 20×30×60Å (1Å=0.1 nm) (Figure 2A). Canonical HEAT repeats are composed of two antiparallel helices in which the N-terminal helix undergoes a near-90° bend (analogous to an armadillo repeat), moving from an orthogonal to an antiparallel orientation relative to the HEAT repeat's C-terminal helix. Most of the TOG domain's paired helices are not canonical HEAT repeats and lack the orthogonal kink in the N-terminal helix. While most HEAT repeat structures exhibit a global twist, the TOG domain is relatively straight due to a shift in the organization between HEAT repeats C and D (Figures 2A and 2C). Least-squares fitting of the TOG structures shows that the highest degree of structural conservation maps to the face delineated by intra-HEAT loops (Face A, Figures 2A and 2C). Examination of TOG domain homology across species reveals that Face A is the most conserved surface-exposed determinant. The conservation maps primarily to the intra-HEAT loops of repeats A–C. A signature tryptophan residue (Figures 1B and 2B) is surface exposed on repeat A, its ability to torsionally engage the hydrophobic core is restricted by a conserved asparagine residue and a buried salt bridge between an arganine residue and an aspartic acid residue that bridge repeats A and B.
TOG domain structure, conserved elements and potential model for tubulin interaction
Tandem TOG domains from XMAP215 members show robust tubulin binding in vitro [7,24]. Mutating the conserved tryptophan residue in intra-HEAT loop 1 to a glutamic acid residue decreases the affinity for tubulin when made in either of the tandem TOG domains. Mutating the tryptophan residue in both TOG domains to a glutamic acid residue completely abrogates the interaction of the tandem TOG domains with tubulin. It is predicted that of the two tandem TOG domains, one is specific for α-tubulin and the other for β-tubulin. This would explain the oscillatory nature of the arrayed TOG domains, with TOG domains 1 and 3 exhibiting the greatest homology with each other, and a similar degree of homology delineating TOG domains 2 and 4 as a second subgroup. Which TOG domain prefers α-tubulin and β-tubulin is not known, nor is the polarity for engagement. The length of the TOG domain's Face A (52 Å) and the length of a tubulin subunit (40 Å) would require that tandem TOG domains adopt an offset orientation on the tubulin heterodimer (tentative model presented in Figure 2D). A disconnection between TOG domains would explain why they can be proteolytically cleaved from each other and subsequently fail to bind tubulin in trans.
CLASP contains cryptic TOGL (TOG-like) domains
Given that XMAP215 members require tandem TOG domains to bind αβ-tubulin, it was mysterious that CLASP contained a single N-terminal TOG domain. However, structure-based analysis of CLASP revealed two central regions with predicted dodeca-helical content, reminiscent of TOG domain architecture. While the predicted helices showed no conservation with TOG domains, the alternate loop regions spanning the helices showed significant homology to the intra-HEAT repeat loops that mediate tubulin binding. It has thus been predicted that CLASP contains two central TOGL domains that diverged from canonical TOG domains but have functionally retained the conserved platform that facilitates tubulin binding . It is likely that XMAP215 and CLASP are ancient plus end tracking relatives that have diverged to perform independent, modulatory functions on the microtubule cytoskeleton.
Microtubule polymerization mechanisms
CLASP and XMAP215 contain multiple tubulin/microtubule-binding sites, but how these are co-ordinated to afford plus end tracking activity and modulate microtubule dynamics is a point of contention. Recent work on Stu2p and XMAP215 indicate that these proteins bind a single tubulin heterodimer and deliver the heterodimer to the microtubule tip by processively surfing the plus end . This model requires that the five TOG domains of higher eukaryotic XMAP215 members be non-equivalent in their mode of tubulin/microtubule association. This inequality is especially required in the case of the homodimeric yeast XMAP215 members Stu2p and Dis1p, where two functionally equivalent tandem TOG arrays would need to break symmetry to bind a single tubulin heterodimer . However, this model is powerful in that it enables parallel TOG domains to interact with the microtubule and facilitates co-ordinated binding activity between the tandem TOG arrays. A processive mechanism, akin to the action of formins in actin polymerization, requires that one domain bind the polymer, while the other incorporates the polymer's building block into the lattice . These processive domains must work in a co-ordinated, out of phase, bind-and-release fashion, processively driven by polymerization. Other polymerization models postulate the TOG-dependent templating of multiple tubulin oligomers on to the microtubule plus end . In support of this model, in vitro optical trap experiments have shown the XMAP215-dependent incorporation of multiple tubulin subunits en block to the microtubule plus end .
Both CLASP and XMAP215 contain multiple TOG domains. While their plus end tracking activities are similar, there are also fundamental differences in their localization, effect on mitotic spindle dynamics and effect on interphase microtubule dynamics. The TOG domain is apt to confer a uniform mode of tubulin binding. This mode of interaction is likely to be manipulated and utilized to modulate microtubule dynamics differentially, depending on the number of TOG domains arrayed, their quaternary organization and the presence or absence of other microtubule-binding or -regulatory domains present in the protein. For example, five TOG domains arrayed on a single polypeptide may promote a protofilament-like organization of tubulin heterodimers, while a dimer of two TOG domains promotes lateral interactions. To address the interaction between TOG domain proteins and tubulin, it will be important to determine the role multiple TOG domains play in tubulin/XMAP215 and tubulin/CLASP stoichiometries. With the structures of multiple TOG domains determined, the next structural goal will be to elucidate the molecular nature of the TOG–tubulin complex. While trapping tubulin in a polymerization-incompetent state for crystallographic structure determination has been feasible, trapping tubulin in complex with domains that promote polymerization will present unique challenges, and will necessitate the use of X-ray crystallography, small-angle X-ray scattering, cryoelectron microscopy and cross-linking studies. CLASP and XMAP215 members also contain unique CTDs that contribute to their respective plus end tracking activities. Elucidating the role these unique domains play and determining how they synergize with TOG domains will aid our molecular understanding of how XMAP215 and CLASP function. Correlating structural studies with single-molecule in vitro reconstitution assays will enable the field to develop dynamic molecular models of the TOG domain plus end tracking mechanism.
The Dynamic Cell: Joint Biochemical Society and British Society for Cell Biology Focused Meeting held at Appleton Tower, University of Edinburgh, U.K., 1–4 April 2009. Organized and Edited by Ian Dransfield (Edinburgh, U.K.), Margarete Heck (Edinburgh, U.K.), Kairbaan Hodivala-Dilke (Cancer Research UK, London, U.K.), Robert Insall (Beatson Institute for Cancer Research, Glasgow, U.K.), Andrew McAinsh (Marie Curie Research Institute, Oxted, U.K.) and Barbara Reaves (Bath, U.K.).
cytoskeletal associated protein glycine-rich
cytoskeletal linker protein 170
huntingtin, elongation factor 3, the PR65/A subunit of protein phosphatase 2A and the lipid kinase Tor
tumour overexpressed gene
Xenopus microtubule-associated protein 215
K.C.S. is supported by a Klingenstein Fellowship in the Neurosciences and a Basil O'Connor Research Starter Scholar Award from the March of Dimes.