Multiple tubulins: evolutionary aspects and biological implications


For correspondence (e-mail


Plant tubulin is a dimeric protein that contributes to formation of microtubules, major intracellular structures that are involved in the control of fundamental processes such as cell division, polarity of growth, cell-wall deposition, intracellular trafficking and communications. Because it is a structural protein whose function is confined to the role of microtubule formation, tubulin may be perceived as an uninteresting gene product, but such a perception is incorrect. In fact, tubulin represents a key molecule for studying fundamental biological issues such as (i) microtubule evolution (also with reference to prokaryotic precursors and the formation of cytomotive filaments), (ii) protein structure with reference to the various biochemical features of members of the FstZ/tubulin superfamily, (iii) isoform variations contributed by the existence of multi-gene families and various kinds of post-translational modifications, (iv) anti-mitotic drug interactions and mode of action, (v) plant and cell symmetry, as determined using a series of tubulin mutants, (vi) multiple and sophisticated mechanisms of gene regulation, and (vii) intron molecular evolution. In this review, we present and discuss many of these issues, and offer an updated interpretation of the multi-tubulin hypothesis.


Tubulin may be considered one of the most conserved proteins in eukaryotes, in which it was firstly recognized as the target of colchicine, an alkaloid that is capable of arresting mitosis in mammals (Borisy and Taylor, 1967; Weisenberg et al., 1968). Tubulin, a dimeric protein comprised of an α and a β monomer, was thus identified as the primary component of microtubules (MTs). Since then, the extraordinary importance of this protein in the course of evolution, which has been systematically confirmed for eukaryotes, has gained further credit with the identification of FtsZ (filamentous temperature-sensitive protein Z), a prokaryotic homolog that is structurally equivalent to tubulin (Lowe and Amos, 1998). As a consequence, eukaryotic tubulins are now classified as members of the large tubulin/FtsZ superfamily. In all these systems, tubulin acts as the key molecule for cell division, a fundamental role onto which other additional functions have been progressively added, in what appears to be a remarkable process of evolutionary optimization. The original function of providing a molecular tool for bacterial cell constriction and cilia/flagella motility in unicellular eukaryotes, has been progressively added to, with additional diverse intracellular commitments such as chromosome separation, control of cell symmetry and polarity of growth, cellulose deposition, intracellular trafficking and connections, and even storage of long-term memory in humans, as recently proposed (Craddock et al., 2012). What is fascinating about the development of this multi-tasking role is that such functional optimization has occurred without denying the original mandate that is control of cell division. The multicellular commitments, added to the key role in cell division, have probably been achieved through the increase in size of both the α- and β-tubulin gene families. In 1976, this led to the proposition, by Fulton and Simpson, of the multi-tubulin hypothesis, which interpreted tubulin diversity as the result of the requirement for various isoforms for formation of different MT structures (Fulton and Simpson, 1976). Tubulin diversity, based upon the presence of genetically determined multiple α- and β-tubulin isotypes, may be further enhanced by a whole set of post-translational modifications (Luduena, 1998). So far, there is very little convincing experimental evidence of functional commitment for specific tubulin isotypes in animals and none in plants. Thus tubulin gene number redundancy appears much more related to the need to secure expression of this fundamental protein under variations of the intracellular and the extracellular environmental conditions.

At the same time, tubulin gene redundancy has resulted in development of specific sub-classes of the α- and β-tubulin gene family, characterized by differences in the nucleotide sequence, none of which compromise those regions that encode amino acid domains fundamental to the key role played in cell division. In contrast, major nucleotide differences are observed either in those regions that control gene expression (promoters and 5′ and 3′ UTR sequences) or those that encode the C–terminal tail (CTT), which is known to be a major site for the interaction with intracellular components. Tubulin introns have also been identified as an important additional source of polymorphism that may influence the site and level of expression of the various tubulin isotypes. Because of their conserved positioning in plants and vertebrates, introns may be used as molecular markers for evolutionary and phylogenetic studies (Breviario et al., 2008). This review addresses these issues, with a special emphasis on aspects of the plant tubulin gene family.

The tubulin/FtsZ superfamily

Proteins of the actin and tubulin superfamilies are major components of the cytoskeleton in all three domains of life: archaea, eubacteria and eukaryotes (Shih and Rothfield, 2006; Pogliano, 2008; Lowe and Amos, 2009). The tubulin/FtsZ superfamily is ubiquitous in living cells, where the protein performs essential functions such as DNA segregation and cell division.

Proteins of the superfamily share the same structure, consisting of an N-terminal GTP-binding domain, a GTPase-activating domain and highly divergent C-termini. They assemble polar cytomotive protofilaments in which the GTP-binding site is at the interface between successive subunits (Lowe and Amos, 2009). Contact between the GTP-binding domain and the GTPase-activating domain of the next subunit along the protofilament activates GTP hydrolysis, which eventually triggers disassembly and polymer turnover. Most bacteria and archea contain at least one form of FtsZ, the prokaryotic homolog of tubulin (Lowe and Amos, 1998), which probably evolved from a common ancestor. Despite only limited primary sequence homology centered on the GTP-binding motif, which is termed the ‘tubulin signature sequence’, FtsZ and eukaryotic tubulin show extensive structural similarity (Nogales et al., 1998), characterized by the presence of a common fold of two domains connected by a central helix (Figure 1).

Figure 1.

Structural alignments of three members of the tubulin/FtsZ superfamily.

(a) α-tubulin with bound GTP from Ovis aries (Nawrotek et al., 2011).

(b) β-tubulin with bound GDP from Ovis aries (Nawrotek et al., 2011).

(c) FtsZ with bound GDP from Aquifex aeolicus (Oliva et al., 2007).

Color code: helices, blue; β strands, yellow; GDP/GTP, red. Protein structures were aligned using MultiProt software (Shatsky et al., 2004).

Recently, knowledge of the tubulin/FtsZ superfamily has grown further, to incorporate new members such as the tubulin-like protein TubZ, which plays a role in segregation of some bacterial plasmids (Aylett et al., 2010), and the BtubA/B genes, two tubulin-like genes found in the genus Prosthecobacter of Verrucomicrobia (Jenkins et al., 2002), the function of which remains unknown as it has been demonstrated that FtsZ remains in control of cell division (Pilhofer et al., 2007). The sequence and structure of BtubA/B proteins are highly similar to those of eukaryotic α/β-tubulin, and this fact, combined with the finding that no other gene outside this genus is so closely related to any eukaryotic counterparts, sustains the hypothesis of horizontal gene transfer from eukaryotes (Schlieper et al., 2005; Martin-Galiano et al., 2011). However, this hypothesis has recently been challenged by the observation of ‘bacterial microtubules’ composed of five protofilaments, as revealed by electron cryomicroscopy (Pilhofer et al., 2011). The authors of this study suggested that the BtubA/B MTs represent primordial structures that preceded the modern eukaryotic 13 protofilament-based MT structure.

In the plant nuclear genome, tubulin genes co-exist with ftsZ genes, which have moved from their original plastidial location. Phylogenetic analyses revealed that ftsZ genes may be traced back to their cyanobacterial ancestor, and that the split into various families occurred in a progenitor of the Viridiplantae (Osteryoung and Pyke, 1998; Kiessling et al., 2004; Rensing et al., 2004). While plant α-, β- and γ-tubulin define MT functions, FtsZ proteins acts in chloroplasts, where they remain in charge of organelle division and shaping (Reski, 2002; Martin et al., 2009).

Eukaryotic tubulin

α-, β- and γ-tubulin appear to be present in every eukaryote, and in many cases are the only tubulins present. Other tubulins (δ, ε, ζ and η), which are not able to form polymeric structures, have been found in some lineages, particularly protists, but are absent in land plants and fungi. This evidence, together with immunolocalization assays, suggests that they act in association with the centrioles and basal bodies of protist, algae and animal cells (Oakley, 2000; McKean et al., 2001; Fromherz et al., 2004).

α-, β- and γ-tubulin thus represent the minimal set of tubulins required to define MT function in eukaryotic cells. Comparison of members of the tubulin family across eukaryotes reveals that the α-tubulins share 89–95% amino acid homology, β-tubulins are 88–94% similar, and γ-tubulins have similarities ranging from 72 to 94% (Fygenson et al., 2004). At the structural level, the α-, β- and γ-tubulin groups each exhibit major differences in net electric charge, dipole moment and dipole vector orientation (Tuszynski et al., 2006). These properties may influence functional characteristics such as MT stability and assembly kinetics, due to their effects on the strength of protein–protein interactions.

Eukaryotic α- and β-tubulin monomers differ from the single-subunit bacterial protein FtsZ in that they form stable heterodimers with non-exchangeable GTP. α/β-tubulin dimers contain a variable C-terminal domain, which is absent in FtsZ, that extends outside the MTs (Nogales et al., 1999). Although the structure of the tubulin dimer within the MT lattice has been determined (Gigant et al., 2000; Lowe et al., 2001; Li et al., 2002a; Sept et al., 2003), the biochemical roles for many amino acids present in the α- and β-tubulin chains remain obscure. In multicellular organisms, further levels of diversification among the multiple tubulin isotypes present in the MT lattice may be associated with the presence of species-specific sequence variations and patterns of differential expression. However, the structures of the α- and β-tubulin moieties are known to be quite similar, being almost indistinguishable at approximately 6 Å (Li et al., 2002a). Tubulin has three main domains, the N-terminal nucleotide-binding domain, the intermediate domain involved in taxol-binding and the C-terminal ligand-binding domain, which shows relevant structural fluctuations (Keskin et al., 2002; Freedman et al., 2011).

Once synthesized, both α- and β-tubulin proteins undergo a complex process of folding that, assisted by prefoldins, chaperonins and co-factors, the folding process is active at both the monomer and dimer level (Lewis and Cowan, 2002; Steinborn et al., 2002; Dhonukshe et al., 2005).

Each individual tubulin heterodimer is 8 nm long, and they interact laterally and longitudinally to form protofilaments and then MTs (Downing and Nogales, 1998). The assembled tubulin dimer hydrolyzes a bound GTP molecule to GDP at its exchangeable nucleotide-binding site. The kinetics of this process in α/β-tubulin is critical in regulating dynamic instability as it affects the status of the ‘lateral cap’, i.e. the top tubulin layer in the growing MT, which is known to stabilize the entire structure. The number of eukaryotic tubulin protofilaments per MT is usually 13, but may vary from 8–15 depending on special conditions (Pierson et al., 1978; Eichenlaub-Ritter and Tucker, 1984; Raff et al., 1997; Topalidou et al., 2012). The exclusion zone between MTs (Dustin, 1978) may be influenced by electrostatic repulsion among C-terminal ends, which have been proposed to either project to a considerable distance from the MT or to be collapsed onto the surface, depending on physiological conditions, biochemical interactions and tubulin isotype composition (Tuszynski et al., 2006). According to this model, CTTs for both α- and β-tubulins are also involved in MT assembly and function, as they may influence the conformation of the whole tubulin molecule. Reciprocal interactions between α and β-tubulin CTTs and the respective H11 domains have been proposed (Freedman et al., 2011). These alter the conformation of the tubulin dimer to such an extent as to influence MT formation and interaction with ligands. Tail–body interactions within tubulin and tubulin polymers may also influence the binding or efficacy of the tubulin-targeted cancer chemotherapy drugs colchicine and vinblastine (Rai and Wolff, 1998; Chakraborty et al., 2004).


γ-tubulin, first identified in the filamentous fungus Aspergillus nidulans (Oakley and Oakley, 1989), is an essential component of the MT nucleation complex in all eukaryotic cells (Aldaz et al., 2005; Wiese and Zheng, 2006). The nucleation of new MTs from the surface of pre-existing MTs was first documented in the green alga Nitella tasmanica (Wasteneys and Williamson, 1989) at approximately the same time as the discovery of γ-tubulin. This key step may have resulted from changes in γ-tubulin or γ-tubulin complexes (Murata and Hasebe, 2007). In animal and most algal cells, γ-tubulin is tightly associated with centrioles and the centrosome, whereas, in plants, it is associated with microtubule-organizing centers, diffuse and mobile entities that change in form and location during the cell and life cycles (Brown and Lemmon, 2007). Depletion of plant γ-tubulin by RNAi results in randomization of cortical MTs in A. thaliana (Binarova et al., 2006). This is consistent with the finding that cortical MTs of interphase plant cells are predominantly nucleated from γ-tubulin-containing sites on lattices of previously established MTs (Murata et al., 2005). γ-tubulin is found in complexes with additional proteins, termed γ-tubulin complex proteins (GCPs). The γ-tubulin small complex is a core nucleation unit comprising two molecules of γ-tubulin associated with one molecule each of GCP2 and GCP3 (Seltzer et al., 2007). A single point mutation in GCP2 (spr3) of A. thaliana has been shown to significantly affect the angle of MT nucleation with a minor influence on MT nucleation frequency or dynamics (Nakamura and Hashimoto, 2009). Animal cells also contain a larger complex called the γ-tubulin ring complex, which includes the additional proteins GCP4, GCP5 and GCP6 and shows high nucleation activity in vitro. A different set of proteins is associated with the larger γ-tubulin complex in yeast (S. cerevisiae) (Wiese and Zheng, 2006). Sequenced genomes of higher plants revealed the presence of other accessory GCP proteins homologous to those present in animal γ-tubulin ring complexes (Pastuglia and Bouchez, 2007), implying that the fundamental organization of plant nucleation complexes is similar to that of animal cells. Spc98p and Spc97p are two such important components of the nucleation complex formed in higher plants (Erhardt et al., 2002; Schmit, 2002). More recently, two additional GCP3-interacting proteins (GIP1 and GIP2) have been characterized as part of the γ-tubulin complex in A. thaliana (Janski et al., 2012). GIP1 and GIP2 are required for γ-tubulin complex protein localization, spindle integrity, and chromosomal stability and segregation.

During plant cell mitosis, the critical role of γ-tubulin in MT nucleation is exerted in concert with the augmin complex, which is made up of eight subunits in A. thaliana (Hotta et al., 2012). Augmin plays an important role in γ-tubulin localization and MT generation from prometaphase to cytokinesis. MT formation in phragmoplasts was severely compromised after RNAi knockdown of an augmin subunit, leading to incomplete expansion of phragmoplasts and cytokinesis failure. Knockdown of the γ-tubulin complex affected MT formation throughout mitosis (Nakaoka et al., 2012).

Tubulin post-translational modifications

Once synthesized, α- and β-tubulin may undergo several post-translational modifications, the functional role of which remains largely obscure. The lack of clear functional indications led to reconsideration of the idea that these modifications not only impart distinct biochemical properties but may mask isotype-specific differences, reducing tubulin heterogeneity (McRae, 1997). Tubulin dimers may be altered by diverse post-translational modifications, such as acetylation of lysine 40 in α-tubulin (Sasse and Gull, 1988; Gardiner et al., 2007), or glycylation and glutamylation of the CTTs of both α- and β-tubulins (Verhey and Gaertig, 2007). In plants, although polyglycinated tubulin has not been detected, and reports of polyglutamylated tubulin are limited (Smertenko et al., 1997; Wang et al., 2004), acetylation of MTs has been fairly well documented. According to more recent studies, MT acetylation acts as a signal for ER aggregation, a pre-requisite for ER organization and association (Perdiz et al., 2011). Consistently, it has been shown that acetylation of MTs favors ER-ring formation in the preprophase band of angiosperms (Zachariadis et al., 2003; Quader et al., 2006; Giannoutsou et al., 2012). Use of acetylation-specific anti-tubulin monoclonal antibodies (Piperno and Fuller, 1985; Gilmer et al., 1999; Perdiz et al., 2011) revealed the presence of acetylated tubulin in MT systems other than the preprophase band, such as the mitotic spindle and the interphase MT band of guard cells. Acetylation of tubulin may be considered a consequence rather than a cause of an arrest of MT dynamicity. In addition to acetylation, recent mutational experiments have shown the structural importance of K40 of α-tubulins in plant growth (Xiong et al., 2013). Replacement of lysine with alanine or glutamine, but not arginine, severely affects cell division and expansion.

The CTT of α-tubulin is also the target of a tyrosination/detyrosination modification cycle that appears to be evolutionarily conserved across eukaryotes (Preston et al., 1979). In fact, although tyrosine is the C-terminal amino acid encoded by most plant and animal α-tubulin genes, de-tyrosinated tubulin has been ubiquitously found in many eukaryotes, including plants (Luduena, 1998). Tyrosination of tubulin, which is commonly interpreted as a signature for rapid growth, is probably involved in controlling the binding of plus-end tracking proteins and motor proteins (Peris et al., 2009). The occurrence of tubulin phosphorylation and nitrotyrosination in plants has also been reported recently (Blume et al., 2008a; Jovanovic et al., 2011; Yemets et al., 2011). These two post-translational modifications may actually target the same tyrosine residues with different effects. While phosphotyrosine residues may be detected within the normal cortical MT array of interphase cells (Blume et al., 2010), tubulin nitrotyrosination is reported to affect microtubular functions relevant to cell division and cell-wall deposition (Jovanovic et al., 2011). The importance of plant tubulin phosphorylation for cell-cycle progression and MT array transition has been well documented with reference to cyclin-dependent serine/threonine kinase/phosaphatase (Blume et al., 2008b).

More generally, it is assumed that tubulin post-translational modifications differentially mark distinct MT sub-populations, affecting MT stability and influencing binding to microtubule-associated proteins, motor proteins and drugs. It is also assumed, and it can be experimentally verified, that the status of the tubulin dimer is characterized, at any given time, by the presence of multiple and heterogeneous post-translational modifications. Tubulin post-translational modifications are evolutionarily conserved, suggesting that they play important roles in vivo, such as binding of molecular motors to the external surface of MTs (Rosenbaum, 2000).

Tubulin and drugs

Various pharmacological properties have been attributed to tubulins of different kingdoms (Downing, 2000). They are determined by the presence of distinct amino acids that differ between protozoa, fungi, vertebrates and plants but are conserved within each of these groups. Based on this, a large number of small molecules targeting the eukaryotic α/β-tubulin dimer have been produced, with the principal aim of inhibiting cell division when undesirable. Such a need may easily be foreseen for control of the growth of animal and plant pathogens (protozoa and fungi), infestants (helminths), weeds and cancerous cells. It is generally accepted that these drugs act by shifting the normal equilibrium between free dimers and polymers to alter the stability of MTs.

Among the many compounds that are capable of interacting with almost all tubulin isotypes, a few are selectively active against phylogenetically restricted subsets of tubulins. Dinitroanilines represent one important group of such selective drugs. They are synthetic compounds that disassemble MTs in plants and protozoa but are inactive against the MTs of vertebrates and fungi (Morrissette and Sept, 2008). This has been largely established by binding studies performed with tubulin extracted from plants, protozoans and vertebrates (Morejohn et al., 1987; Hugdahl and Morejohn, 1993). Modeling studies, supported by experimental findings from resistant biotypes, have identified the target site for dinitroanilines at the dimer interface (Blume et al., 2003; Delye et al., 2004). Distinct amino acid residues of α-tubulin contribute to the binding site (Delye et al., 2004; Hashim et al., 2011). Dinitroanilines, particularly trifluralin, have been used in commercial herbicide formulations for decades. As a result, a few α-tubulin mutations conferring resistance to dinitroanilines have emerged in weeds such as the goosegrass Eleusine indica (Anthony et al., 1998; Yamamoto et al., 1998).

In plants, a common mechanism of action for almost all of the anti-MT herbicides whereby the tubulin heterodimer–herbicide complex sequesters the MT subunits from integration into the growing ends of MTs has been proposed (Vaughn and Nick, 2000; Giles et al., 2009). The consequence of tubulin sequestering is that the equilibrium of dynamic MT growth shifts towards disassembly, eventually resulting in MT catastrophe. This effect is much more enhanced on a highly dynamic MT population. Such a mechanism has been evoked for anti-microtubular drugs of the dinitroaniline and cyanoacrylate categories, which are capable of producing, in a dose-dependent manner, complete loss of all MT structures, from cortical arrays to the preprophase band, spindle and phragmoplasts. This wide range of effectiveness may not hold true for other anti-microtubular herbicides, which may not share the same mechanism of action. It is not clear, for instance, if the mechanism of tubulin sequestering is attributable to the herbicides of the carbamate class. In fact, Isopropyl-N-phenylcarbamate has been reported to primarily affect microtubule-organizing centers rather than MTs (Yemets et al., 2008), while ethyl-N-phenylcarbamate does not produce the root swelling and loss of cell anisotropy in rice seedlings (Oryza sativa) or in vitro cultivated cells that is observed with dinitroanilines (Morettini et al., 2013). This suggests that ethyl-N-phenylcarbamate, although efficient in inhibiting cell growth, is less effective on cortical MT arrays. Moreover, a mutant rice cell line resistant to ethyl-N-phenylcarbamate remains susceptible to the dinitroaniline oryzalin, indicating that the two herbicides may act differently or have different priority targets (Morettini et al., 2013). Citral, a monoterpene, has been reported to primarily affect mitotic MTs rather than cortical MTs (Chaimovitsh et al., 2012). In contrast, flamprop-M-methyl is an anti-microtubular drug that, although ineffective on MT polymerization, severely affects the orientation of the MTs that are part of the spindle and the phragmoplast arrays, thereby causing minus-end disassembly (Tresch et al., 2008).

While the hypersensitivity of helical growth mutants to oryzalin may be easily explained by a further reduction of GTP capping, which is already compromised by the increase in GTPase activity caused by the majority of tubulin mutations, it is difficult to understand how the dimer-binding and inhibition model may explain the organ twisting triggered by low doses of oryzalin. The fact that organ twisting, such as left-hand torsion of A. thaliana roots, may also be induced by low doses of taxol, a drug that stabilizes MTs, and prozyamide, suggests that oryzalin may also directly interact with MTs.

Plant tubulin mutants

Mutations in tubulin genes tend to be either lethal, in those organisms in which few genes are present, or subtle, if masked by gene redundancy. Also, they may or may not be apparent under the particular screening conditions employed. This is understandable given the key role exerted by this protein in the cell division process of all eukaryotes. In addition, it should also be remembered that, at any given time, plant cells and MTs contain a heterogeneous population of tubulin, such that tubulin properties reflect those of the isotype mixture.

Thus, it is not surprising that the vast majority of the tubulin and tubulin-associated mutations characterized so far primarily affect the cortical MT structure and the related processes of growth expansion and cell chirality. Although important for cell expansion and cellulose deposition (Paredez et al., 2006; Wasteneys and Fujita, 2006; Lloyd and Chan, 2008), the cortical MT array may tolerate a certain degree of abnormality compared to more fundamental structures such as the spindle.

With regard to the chirality of cells in multicellular organisms, it has recently been shown that a fundamental contribution is made by the cytoskeleton (see below). The chirality of cytoskeletal structures due to the presence of mutations in tubulin or microtubule-associated proteins may also be amplified by drugs or physiological mechanisms. Left/right (L/R) asymmetry relates to how plant cells utilize the MT cytoskeleton to initiate and rigorously maintain straight growth, and the cortical MT array plays a fundamental role on this. Models that indicate how this role may be exerted have been proposed (Furutani et al., 2000; Wasteneys and Collings, 2004; Buschmann et al., 2009).

Consistent with all this, it has been demonstrated that the majority of helical growth mutants with fixed handedness are the result of mutations in regulators of MT function, and are defective in cortical MT organization. Usually, cortical MT handedness shows the opposite orientation to the helical growth of the mutants, although exceptions in which organ twisting occurs in the absence of a preferential MT orientation have been reported (Whittington et al., 2001). Helical growth organization is presumed to result from a compromised MT state, and further detrimental effects on MTs (e.g. a low dose of disrupting agents) lead to more disorganized or fragmented arrays.

Mutations affecting plant symmetry may be found in the α- and β-tubulin genes as well as in genes encoding γ-tubulin-associated proteins or microtubule-associated proteins. Arabidopsis lefty mutants (Thitamadee et al., 2002) were the first to be reported among mutations that exclusively affect tubulin-encoding genes. These mutants are characterized by an altered L/R symmetry of their organs that causes left-handed helical growth. The altered phenotype is due to a single dominant-negative mutation occurring in either of the α-tubulin genes TUA4 or TUA6. The resultant serine to phenylalanine substitution at position 180 is located at the intra-dimer interface of tubulin. This reflects in a reduced stability of MTs that results in left-handed helical growth. Suppressor mutations of the lefty phenotype map to the same α-tubulin gene, but to a different location. These mutations are capable of restoring the original growth properties to the extent that they are indistinguishable from the parent ecotype (Thitamadee et al., 2002). lefty mutations predominantly affect the function and organization of cortical MTs rather than mitotic or cytokinetic MT arrays. This does not reflect differences in tubulin gene expression, which is substantially cell cycle-independent.

After the original identification of the lefty mutants, numerous additional twisted growth mutants associated with dominant-negative mutations occurring in either the α- or the β-tubulin genes of A. thaliana were characterized (Ishida et al., 2007). Dominant-negative mutations in either TUA or TUB genes caused the formation of either right or left-handed cortical MT arrays in root epidermal cells, which twisted in a direction approximately perpendicular to the orientation of the array. This shows that the cortical array dictates the direction of cell growth. In addition to tubulin mutations proximal to the GTPase-activating region, other mutations have been mapped to the longitudinal interface between tubulin subunits and the lateral contact regions among adjacent protofilaments. All these dominant-negative tubulin mutants were incorporated into the MT polymer and formed shallow helical arrays of distinct handedness along the long axis of the root epidermal cells. Several semi-dominant mutants exhibited severe phenotypes, such as isotropic cell expansion, imperfect cytokinesis, disconnected vasculature and deformed root hairs and trichomes, and did not grow to the reproductive phase or set few seeds (Ishida et al., 2007).

Recently, a twisted dwarf 1-1 (Tid1-1) mutation occurring in a rice α-tubulin ortholog of the A. thaliana genes affected in the lefty mutants (TUA4 and TUA6) has also been characterized (Sunohara et al., 2009). The Tid1-1 mutation not only affects anisotropic growth but heavily influences meristematic activity and gross plant morphology. With respect to the A. thaliana lefty mutants, a threonine to isoleucine substitution at residue 56 causes right-handed growth with left-handed MTs and dwarfism. These results demonstrate that the direction of growth may depend on the position of the mutation.

A point mutation (C213Y) in the TUA6 gene of A. thaliana that determines root skewing only in the presence of low doses of propyzamide has also been reported (Ishida and Hashimoto, 2007). This conditional twisting mutant exhibits normal growth and development in the absence of the drug. Structural tubulin studies suggest that this point mutation affects the GTP/GDP nucleotide-binding domain (Ishida and Hashimoto, 2007). Organ specificity is explained on the basis of the higher expression in roots of the TUA6 isotype compared to the level of expression of other TUA genes (TUA2 and TUA4) that belong to the same sub-family (Ishida and Hashimoto, 2007). The fact that skewing is not revealed except in the presence of low doses of herbicide suggests that the alteration of MT organization and dynamics caused by the C213Y mutation is too small to show a perceivable effect. Thus, MT twisting, which is often associated with altered anisotropic growth, is a pre-condition for MT disorganization and eventual depolymerization. This predisposition is made apparent by the effect of drugs known to interfere with MT dynamics.

tor2 (tortifolia) is a conservative replacement mutation (arginine to lysine at position 2) in the TUA4 gene of A. thaliana (Buschmann et al., 2009). It has been used experimentally to propose a new model for helical growth that is active at the single-cell level and is not influenced by cell–cell interactions. According to structural models, the tor2 mutation affects inter-dimer contacts, weakening the hydrogen bonds that are normally established between the arginine residue at position 2 of α-tubulin and the GTPase domain of β-tubulin. Thus, the α-tubulin arginine to lysine mutation alters the GTPase-activating function of β-tubulin, thereby affecting MT assembly. Experiments performed on tor2 suspension cells revealed that torsion is inherent within single cells. This has been confirmed in trichomes of the same mutant that show a right-handed twisting phenotype in freely growing cells with no contribution of multicellular, tissue-based interactions (Buschmann et al., 2009). These experiments demonstrate that helical twisting occurs at the level of single cells. Thus, the organ twisting observed in the A. thaliana tor2 mutant may be simply explained by the additive effect of helical expansion of individual cells. Through use and manipulation of hypocotyls, organs that do not show cell division, these experiments definitively show that helical organ growth is exclusively based on cell elongation, as previously proposed (Wasteneys and Collings, 2004). Although apparently contradictory to the previous theories of helical growth based either on staggered helical division patterns of cell files (Wasteneys and Collings, 2004) or differential cell elongation occurring between the epidermal and cortical cells of seedling roots (Furutani et al., 2000; Hashimoto, 2002), the single-cell determination model may still be compatible with those theories. Single-cell determination of L/R patterning is also consistent with recent data on chirality obtained by microinjecting mutated plant homologous α-tubulin mRNAs into animal cells (Lobikin et al., 2012).

As anticipated, spiral growth may also be caused by mutations occurring in proteins of the γ-tubulin-containing complex, as is the case for the spiral3 mutant of A. thaliana (Nakamura and Hashimoto, 2009), where a missense mutation in the GCP2 gene compromises the interaction between GCP2 and GCP3, another subunit of the γ-tubulin-containing complex, resulting in very severe right-handed helical growth.

To determine whether the same tubulin proteins implicated in Arabidopsis asymmetry also control the large-scale asymmetry of both vertebrate and plant systems, molecular mutations responsible for twisting phenotypes were introduced into the corresponding positions of two Xenopus laevis genes encoding for α-tubulin and the y tubulin-associated protein Tubgcp2 (Lobikin et al., 2012). Corresponding synthetic mRNAs were transfected into Xenopus laevis, Caenorhabditis elegans and human HL-60 cells to verify how much their influence on cell asymmetry extended throughout eukaryotes. Transfection of mutated tua4 and tubgcp2 mRNA randomized asymmetry in all these organisms, inducing a significant level of heterotaxia (altered expression of mRNA and protein). These experiments show that the same tubulin mutations that randomize asymmetry in plants also do so in nematode and vertebrate systems. Data on Xenopus embryos demonstrated that the mutated tubulin isoform is capable of influencing the entire pattern of development at its very origin, when the left/right organ asymmetry is normally established. Activation of symmetry thus appears to be influenced by the orientation of the cytoskeletal tracks, as may be the case for handedness of the plant cortical MTs.

Tubulin gene expansion in plants

Convincing evidence has been obtained for separation of α-, β- and even γ-tubulin genes from a common ancestor before the divergence of extant eukaryotes (Keeling and Doolittle, 1996). As in other eukaryotic lineages, gene duplication events led to an increase in tubulin gene family size throughout evolution. While unicellular green algae possess one α-tubulin gene, two almost identical β-tubulin genes (Silflow and Rosenbaum, 1981; Harper and Mages, 1988) and one δ-tubulin gene (Dutcher and Trabuco, 1998), flowering plants may have up to 20 different β-tubulin genes as found in Populus tremuloides (aspen) and 13 α-tubulin genes, as in Gossypium hirsutum (cotton) (Oakley et al., 2007; He et al., 2008). Conversely, γ-tubulin is encoded by either one or two genes.

Gene redundancy is more common in plants than in animals due to their higher genome plasticity, which reflects the occurrence of polyploidization events introduced either by evolution or artificially during selection of domesticated crops. Tubulin gene expansion may be traced back to these genome-wide duplication events, which were then followed by chromosome rearrangements and single-gene losses and/or duplications. The whole process eventually led to a distribution of the tubulin loci across the whole genome, a feature that makes it possible to use β-tubulin introns as effective molecular markers (Breviario et al., 2007). Tubulin family expansion is already found in lower land plants. The moss Physcomitrella patens contains six genes encoding almost identical β-tubulins (96–99% amino acid identity) that show a moderate degree of differentiation in their expression patterns and minimal variations in the hypervariable CTT (Jost et al., 2004). This group forms a separate clade in a similarity tree of plant β-tubulins, suggesting that sub-families observed in vascular plants originated after their separation from bryophythes.

In the fern Ceratopteris richardii, at least four α-tubulin loci are present, encoding proteins with greatest diversity at their CTT. Phylogenetic analysis of nucleotide sequences placed three of the four Ceratopteris α-tubulin genes in a clade with homologs of Pseudotsuga menziesii (a gymnosperm tree) and the fern Anemia phylliditis, while the fourth gene was clearly separated in the clade of the algal α-tubulin genes. This is consistent with a history of two gene duplication events, one preceding the divergence of algae and land plants, and the other following the divergence of ferns and seed plants (Scott et al., 2007).

In contrast to vertebrates, in which the α- and β-tubulin gene families are similar in size, with six or seven members in four α- and six β-tubulin classes (Khodiyar et al., 2007), the number of higher-plant β-tubulin isotypes typically exceeds that of α-tubulins, and the size of the whole tubulin gene family varies greatly. Minimum-evolution trees, based on alignment of amino acids sequences of all known α- and β-tubulins of six species, three monocots and three dicots, are shown in Figures 2 and 3. It has long been established that α-tubulins in flowering plants are clearly split into two sub-families (I and II), according to specific DNA sequence features (Villemur et al., 1992). In contrast, the definition of distinct classes of β-tubulin is relatively recent and clades are less clearly resolved. Although some difference in phylogenetic analysis may arise depending on the pool of sequences aligned and the algorithm employed, β-tubulins may be roughly grouped into four main clades (Oakley et al., 2007; Radchuk, 2008), with minor but clearly separated branches (Figure 3) that are not representative of all species. Class I β-tubulins are highly represented in cereals (three members in Zea mays, Hordeum vulgare and Oryza sativa), and also in aspen (six genes, mostly expressed in xylem), while only one gene is present in the A. thaliana genome. The genes possess a long third intron in the 5′ UTR with regulatory functions. Class II is more expanded in dicots, with 10 members in cotton, but is represented by a single gene in each cereal. Class III proteins contain additional amino acid residues (1–4) at position 39 within the N-terminal hypervariable region. The maize (Zea mays) genes ZmTub3-4 and their putative rice ortholog OsTub8, the only members of class III in these species, are expressed almost exclusively in anthers (Rogers et al., 1993; Yang et al., 2007). Interestingly, GhTub6, 8, 11 and 13, which are all members of class III, are the only cotton β-tubulin genes that fail growth complementation in S. cerevisiae (He et al., 2008). This evidence suggests that the additional amino acid residues may confer different properties to these tubulins. In contrast to class II, class IV is more expanded in monocots. Some pollen-specific genes from eudicots, e.g. Arabidopsis AtTub9 (Cheng et al., 2001) and aspen PtTuB7 and PtTUB8 (Oakley et al., 2007), belong to this class.

Figure 2.

Unrooted-minimum evolution tree for α-tubulin amino acid sequences of Hordeum vulgare (Hvu), Zea mays (Zma), Oryza sativa (Osa), Arabidopsis thaliana (Ath), Gossypium hirsutum (Ghy) and Populus tremuloides (Ptr). Red circles and blue triangles indicate tubulin sequences of the two reference genomes for dicot (A. thaliana) and monocot (O. sativa) species, respectively. Phylogenetic analysis was performed using MEGA 5.1 (Tamura et al., 2011). The numbers on major branches indicate the percentage of bootstrap support evaluated using the interior-branch test with 1000 replicates. Accession numbers of protein sequences are listed in Table S1.

Figure 3.

Unrooted minimum-evolution tree for β-tubulin amino acid sequences of Hordeum vulgare (Hvu), Zea mays (Zma), Oryza sativa (Osa), Arabidopsis thaliana (Ath), Gossypium hirsutum (Ghy) and Populus tremuloides (Ptr). Red circles and blue triangles indicate tubulin sequences of the two reference genomes for dicot (A. thaliana) and monocot (O. sativa) species, respectively. Phylogenetic analysis was performed as for Figure 2. Accession numbers of protein sequences are listed in Table S1.

Generally, it is clear that the number of tubulin isotypes within each cluster is not conserved across species: new genes may have arisen following specific duplication and other genes have been lost or mutated to pseudogenes, so that identification of orthologs is not straightforward among distantly related species. Examples of recent gene duplications are represented by tandem genes that encode highly similar or identical proteins (e.g. AtTUA3 and 5, AtTUA2 and 4, AtTUB2 and 3, ZmTtub3 and 4, and ZmTub5 and 8) and reported cases of pseudogenization (Oakley et al., 2007). However, the asymmetrical expansion of particular classes in some species requires repeated single-gene duplications or differential selective losses following genome-wide duplication events. It thus appears that selective forces shape tubulin classes to respond to specific needs. To obtain further insights into the evolution of β-tubulin sub-families, it is important to add new members to the α-and β-tubulin phylogenetic trees, particularly from less studied and more ancient seed plant groups, such as cycas, gymnosperms or early monocots (e.g. palms).

The multi-tubulin hypothesis

The key question concerning tubulin gene expansion is why all the members are maintained. Is it a matter of different function or different regulation? The multi-tubulin hypothesis, firstly proposed by Fulton and Simpson in 1976, was an attempt to offer a rationale for this observation. They suggest that the various forms of the tubulin proteins contribute to distinct MT structures, possibly with equally distinct functions. Originally, this hypothesis was supported by biochemical studies performed with isotypically pure α- and β-tubulins extracted from bovine brain. By producing various α/β heterodimers in which one α-tubulin subunit bound to one of three β-tubulin polypeptides, it was found that the tubulin isotype composition influences in vitro MT dynamics (Panda et al., 1994). The multi-tubulin hypothesis is apparently supported by the finding that different expression patterns are observed in almost all eukaryotes when more tubulin genes are present, but direct evidence for a clear functional specificity of individual tubulin isotypes remains limited to few examples. The best known examples of functional specificity come from two invertebrate species, C. elegans (Savage et al., 1994; Fukushige et al., 1999) and Drosophila melanogaster (Hoyle and Raff, 1990; Raff et al., 2000), but no such functional specificity has been reported in plants despite the large expansion of their tubulin gene families. On the other hand, some examples of sub-functionalization, i.e. tubulin isotypes with a specific but not strictly exclusive role, have been found. Sub-functionalization is often the consequence of a differential pattern of expression. In the last 20 years, many plant tubulin genes have been cloned and many tubulin gene sequences have been retrieved from genome sequencing projects, thus allowing collection of expression data through in situ hybridization, Northern blotting, (Breviario and Nick, 2000; Breviario, 2008) RT-PCR (Radchuk, 2008), reporter gene expression in transgenic plants, isotype-specific antibodies and microarray data analysis (Carpenter et al., 1992, 1993; Tonoike et al., 1994; Uribe et al., 1998; Stotz and Long, 1999; Breviario and Nick, 2000; Jeon et al., 2000; Cheng et al., 2001; Li et al., 2002b, 2007; Breviario, 2008; Griffin and Wick, 2008; Gianì et al., 2009). Specificity of expression of various tubulin genes was thus observed in distinct organs, tissues or even cell types, and potential roles for any given tubulin isotype have been proposed.

The overall picture emerging from these studies is that of preferential co-existence of expression between more general and more specific members of the tubulin gene family. Expression patterns of the latter may be finely tuned, not only for tissue specificity but also in response to many external or hormonal stimuli. However, expression patterns are often partially overlapping, such that, in every cell type, more than one tubulin isotype is expressed at the same time (Radchuk, 2008). Pollen germination and plant responses to low temperatures represent conditions for which there is more convincing evidence for sub-functionalization of specific tubulin isotypes.

In fact, both conditions are characterized by extensive modifications of the MT cortical array. During pollen tube elongation, an explosive assembly of MTs, formerly absent in the vegetative cytoplasm, is observed (Huang et al., 1993; Taylor and Hepler, 1997). Similarly, during cold acclimation, plant MTs undergo transient disorganization, followed by formation of cold-stable MTs (Nick and Nick, 2000; Abdrakhamanova et al., 2003). Such rearrangements are preceded in various plant species by transcription of subsets of α- and β-tubulin genes, specific for pollen (Carpenter et al., 1992; Rogers et al., 1993; Villemur et al., 1994; Cheng et al.,2001; Fiume et al., 2004; Yang et al., 2007; Oakley et al., 2007) or triggered by cold (Kerr and Carter, 1990; Chu et al., 1993; Ridha Farajalla and Gulick, 2007). Such observations are in accordance with the multi-tubulin hypothesis, suggesting that MT arrays endowed with specific features may be required under particular circumstances, although a functional need for specific tubulin isotypes has not been demonstrated in these cases.

In accordance, tubulin gene disruption or silencing, and even gene swap experiments in multicellular organisms, has not led to any macroscopic phenotypic alteration, and hence any clear functional assignment, with the exception of the two cases of functional specificity reported for C. elegans (Savage et al., 1994; Fukushige et al., 1999) and Drosophila melanogaster (Hoyle and Raff, 1990; Raff et al., 2000). This is probably due to the following three factors, which may concurrently contribute.

  1. Gene redundancy. In every cell, more than one type of tubulin is present at any time, probably with a similar structure and overlapping functions.
  2. ‘Blind’ self-assembly. Whatever the degree of sub-functionalization may be, any α- and β-tubulin must be able to fulfil the minimal requirement of being capable of self-assembling into MTs. The results of gene replacement experiments in various systems agree on one point: any tubulin isotype that is introduced or expressed in a cell is readily incorporated into the growing MTs, but with different kinetics.
  3. Co-regulation. Knockout or ectopic expression of a single tubulin gene may affect the synthesis of endogenous tubulins, resulting in enhanced expression or down-regulation of other genes as a consequence of complex regulatory mechanisms (Bao et al., 2001).

The difficulty in identification of macroscopic functional difference among isotypes in plants does not mean that differences among tubulin isotypes do not play any role at a lower level of resolution. In fact, particular isotype combinations may better support more specific functions, thus explaining their adaptive role as a whole. This is the concept of ‘isovariant dynamics’ (Meagher et al., 1999) proposed to explain the co-expression of closely related actin proteins in the same cell. Such highly networked biochemical systems may provide more robust and highly buffered responses, increasing individual fitness, and thus being favored by selection.

Thus, determination of the functional specificity of any given tubulin isotype may be hindered by the fact that individual MTs are composed of combinations of various isotypes. Hence, functionality may not depend on the presence of a specific isotype but is rather influenced by specific ratios between tubulin isotypes that may vary during the MT lifetime, depending on internal and external stimuli.

At the same time, protein diversification is not mandatory, and very similar proteins with different expression patterns may occur together without further important modifications, as in the case of P. patens β-tubulin genes (Jost et al., 2004). In P. patens, it is likely that tubulin gene diversification ceased at an early stage in the absence of evolutionary forces that may act when the number of different cell types and specialized cellular functions increase, such an halt is likely to cause little diversification of the expression patterns and no further differentiation at the protein level.

In conclusion, the multi-tubulin hypothesis does not need to be verified for each single isotype in order to be accepted. Its acceptance is not in contradiction with the existence of genes that encode almost identical proteins and may differ only in their regulatory sequences. Both are products of the same evolutionary process, that may promote isotype functional specificity by a two-step process. Gene duplication is the first prerequisite for evolution of functional diversity in genes that play fundamental roles in cell life. These genes tolerate minor amino acid variations and tend to lose function entirely rather than providing an altered function in response to changes in their amino acid sequence. Conversely, mutations in regulatory regions are well tolerated, and their evolution leads to changes in the pattern of expression over time (cell cycle) for single-cell organisms, and/or space (tissue or cell type) for multicellular species (sub-functionalization). This is probably sufficient to stabilize newly duplicated genes in complex organisms, in association with the co-evolution of subsets of cell-specific transcription factors. Temporal and spatial expression differentiation of such fundamental genes provides the organism with the opportunity to submit new amino acid combinations to the pressure of selection. Whatever changes occur in the protein structure that influence the dynamic properties of the MT may have a detrimental effect if the mutated protein is expressed in a constitutive way, but may provide specialized functions in special cell types.

Tubulin gene introns

Tubulin genes are interrupted by introns in most eukaryotes. The number, length and position of tubulin introns varies widely between lineages (Ayliffe et al., 2001; Perumal et al., 2005; Nielsen et al., 2010). High numbers of small introns in variable positions are found in fungi and many invertebrates (Einax and Voigt, 2003; Edvardsen et al., 2004; Nielsen et al., 2010), while introns of both α-and β-tubulin genes are few and are located at highly conserved positions in vertebrates and plants (Perumal et al., 2005).

In the plant kingdom, differences in intron distribution are observed between green algae and land plants (Figure 4). In unicellular green algae, β-tubulin genes have three introns (amino acids positions 8, 55/56 and 131), while α-tubulin genes have two or three (positions 15, 89 and 210) (Silflow and Rosenbaum, 1981; Harper and Mages, 1988). Starting from mosses (P. patens), the two most upstream introns of algae β-tubulin genes were lost, while the third (position 131) is conserved and another was added at position 221 in subsequent lineages. This additional intron is not found in some class II β-tubulin genes of grasses (rice Ostub2 and maize Zmtub1). A third additional intron within the 5′ UTR is a typical feature of class I β-tubulin genes of seed plants. Interestingly, one of the six P. patens β-tubulin paralogs bears a unique intron in the 3′ UTR, just after the stop codon.

Figure 4.

Comparison of intron positions among α- and β-tubulin genes of green algae and flowering plants. Numbers indicate the position in the corresponding amino acids sequence (not to scale).

Flowering plant α-tubulin gene structure differs among the two classes: members of class II have four introns (positions 38, 109, 176/177 and 346), while class I genes have one to three introns (31/32, 110 and 233/234), most frequently three (Kopczak et al., 1992; Li et al., 2007; Oakley et al., 2007).

Although very conserved in terms of position, plant tubulin introns vary in length from <100 bp to more than 1000. Regulatory sequences acting at the transcription level have been found to reside within tubulin introns of insects, vertebrates and plants (Gasch et al., 1989; Hoyle and Raff, 1990).

Rice tubulin introns, particularly the most upstream ones, residing either in the coding region for alpha- or the 5′ UTR for β-tubulin genes, have been shown to enhance flanking promoter activities in transient transformation assays (Morello et al., 2002; Fiume et al., 2004) as well as in transgenic plants (Jeon et al., 2000; Gianì et al., 2009) through a mechanism called intron-mediated enhancement of gene expression. Intron presence also affects the site of gene expression, as chimeric tub::GUS genes show different profiles of expression in different tissues and organs depending on the presence of the first regulatory intron (Gianì et al., 2009).

The effect of the intron on gene expression appears to be exerted at the post-transcriptional level, as it is impaired by mutations affecting the splice sites (Morello et al., 2011). These mechanisms are not specific for tubulin, as they have been documented for other plant genes, but it is indisputable that they play a role in modulating the amount of tubulin produced within the cell.

The difference in length among introns from paralogous genes has been exploited as a source of DNA polymorphism useful for plant genotyping: PCR amplification with intron-crossing primers, designed based on conserved exon sequences, allows attribution of characteristic genomic profiles to plant species or varieties (Bardini et al., 2004; Breviario et al., 2007). The use of plant tubulin universal primers was also exploited for rapid characterization of the β-tubulin gene family of the oil plant Camelina sativa (Galasso et al., 2011).

Expression control of tubulin genes

The first step of tubulin gene expression regulation is exerted, as for any gene, at the transcriptional level, as demonstrated in studies that used gene fusion constructs between tubulin promoters and reporter genes (Carpenter et al., 1992; Cheng et al., 2001; Li et al., 2002b). Despite this, critical information about tubulin-specific cis-acting regulatory sequences and transcription factors has remained largely elusive, mostly restricted to studies describing specific regulatory sequences identified in β-tubulin promoters of Chlamydomonas reinhardtii (Davies and Grossman, 1994) and in the 5′ and 3′ flanking sequences of a soybean (Glycine max) gene (Doyle and Han, 2001). In addition, tubulin transcriptional regulation also involves other steps that require further investigation, such as synthesis of multiple transcripts of the same gene that have different 5′ UTR sequences through use of alternative promoters (Morello et al., 2002). The regulatory role of tubulin introns has been discussed above, but it would also be interesting to investigate the relationship among intron sequence variability and specificity in the regulation of tubulin gene expression. The availability of families of highly conserved genes, which differ in their expression pattern, offers a rare opportunity to study their regulatory elements.

In addition to nuclear transcription, the availability of α- and β-tubulin monomers may also depend on control mechanisms that operate at the level of mRNA stability and protein synthesis. In animal cells, tubulin mRNA stability has been shown to be based on a mechanism that operates on the nascent chain of newly synthesized tubulin polypeptides that contain the MREI motif at their N-terminal end (Gay et al., 1989). This is an amino acid signature that typically characterizes (with sporadic exceptions) all animal and plant β-tubulins and class I α-tubulins in plants. Although well described in animals, such a mechanism of regulation has been inferred in plants but never demonstrated.

Another important issue concerning regulation of the synthesis of the heterodimeric tubulin protein relates to the requirement for balanced expression of the α- and β-tubulin monomers. Such a requirement, originally postulated in yeast to explain the lethality observed in strains over-expressing the β-tubulin gene TUB2 (Burke et al., 1989), was further supported by transformation experiments performed on maize calli (Anthony and Hussey, 1998) and tobacco leaf discs (Anthony et al., 1999). Transfection of this material with exogenous, mutated forms of α-tubulin genes allowed recovery of primary transformants only when co-transformed with a wild-type copy of a β-tubulin gene. Transformants were characterized by progressive replacement of the endogenous tubulin mRNA and protein with that produced by the exogenously added transgenes (Anthony and Hussey, 1998; Anthony et al., 1999). Conversely, many tubulin mutants affecting helical growth have been obtained by use of constructs containing single tubulin genes (Abe and Hashimoto, 2005; Ishida et al., 2007). The resulting transgenic plants were reported to be viable, and the transgenes were maintained and expressed for several viable generations of plants. Moreover, MTs that incorporated the GFP-TUB fusion proteins were reported to be functional (Ueda et al., 1999). Evaluation of transcript and protein expression patterning in relation to MT dynamics and plant phenotype (Abe and Hashimoto, 2005) did not reveal any effect of unbalanced tubulin expression due to over-expression of a monomeric tubulin. Over-expression of GFP::TUA6 in A. thaliana actually mimicked the MT structural rearrangements and dynamics that were originally observed in the lefty mutants. This similarity is commonly explained by preferential incorporation of the modified α-tubulins into cortical MTs (although it occurs efficiently in other MT arrays as well), resulting in a right-handed helical arrangement that represents a transitional state toward a more fragmented and destabilized organization of MTs. In fact, the cortical MTs in the GFP::TUA6 lefty double mutant are highly fragmented and randomly oriented, similar to those observed in the temperature-sensitive mor1-1 mutant (Whittington et al., 2001; Abe et al., 2004).

The issue of balanced tubulin expression, although conflicting, must be taken as an opportunity to study the regulatory circuit that controls and modulates the amount of intracellular tubulin.

Tubulin mutations capable of conferring helical growth are also interesting with respect to gene expression. Effective mutations occur in highly expressed tubulin isotypes, as is the case for TUA4 and TUA6 or TUB4 in Arabidopsis (Ishida et al., 2007). This makes it difficult to attribute any possible isotype-dependent functional specificity. Most likely, it is their high level of expression that accounts for the severity of the mutant phenotype.

Translational control may also contribute to the regulation of intracellular tubulin. Shown to occur for α-tubulin in mammalian cells (Gonzalez-Garay and Cabral, 1996), it operates at the level of 5′ UTR sequences and may explain the data reported by Anthony et al., (1999) on transformed tobacco cells and those obtained by treating rice cells with oryzalin (Gianì et al., 2002). In this latter context, it was shown that oryzalin causes a rapid decrease in de novo synthesis of rice α- and β-tubulin, but does not change the level of tubulin mRNA. An additional effect of oryzalin on tubulin protein degradation was also reported (Gianì et al., 2002).

Tubulin folding is another key step that regulates the amount of tubulin subunits available for MT polymerization, in concert with translational and transcriptional controls. Tubulin folding is performed by a cytosolic chaperone complex consisting of several proteins. Some are specific for monomeric α-tubulin (tubulin-folding co-factors B and E) or β-tubulin (tubulin-folding co-factors A and D), whereas others (tubulin-folding co-factor C) are specific for heterodimeric tubulin (Steinborn et al., 2002). The cytosolic chaperone complex may actually be involved in maintaining the correct balance between intracellular α- and β-tubulin monomer pools as shown in Arabidopsis, where a mutation in Kis, the Arabidopsis ortholog of tubulin-folding co-factor A, was rescued by over-expression of α-tubulin (Kirik et al., 2002). This is consistent with the idea that over-expression of α-tubulin in kis mutants counterbalances the excess of free monomeric β-tubulin resulting from reduced activity of the chaperone protein tubulin-folding co-factor A. On the other hand, mutations occurring in prefoldin, a protein involved in tubulin biogenesis (Gu et al., 2008), led to a reduced concentration of the intracellular tubulin pool that, in turn, increased MT stability.

Concluding remarks

The fact that tubulin is a protein that is largely defined at both the structural and functional levels should be regarded as an experimental advantage and a special opportunity, rather than a limitation. While studying the many aspects that affect tubulin (genome evolution, gene regulation, post-translational modifications, protein–protein interactions), the researcher remains well aware that the ultimate purpose is the production of a globular protein that is capable of self-assembly in cytomotive filaments that are essential for cell division, cell growth, cell symmetry, intracellular connections and trafficking. There is no room for speculation about the role of tubulin, and this is a reassuring and firm end-point to which any related issue, any experimental approach, must eventually relate. This is the very same philosophy behind the multi-tubulin hypothesis originally formulated to explain the presence of multiple, apparently unnecessary, isoforms of a protein that was only required to make MTs as its final purpose. However, this functional limit has been largely overcome by the work of the last 36 years, and the chances are that research into tubulin will continue to provide us with exciting and truly innovative contributions in the fields of evolutionary and cell biology, protein biochemistry and gene expression.


We are indebted to Philip Winter and Jack Tuszynski for their kind contribution of Figure 1. We also thank Luca Braglia for his help with the phylogenetic analysis.