The evolution and diversification of plant microtubule-associated proteins


For correspondence (e-mail


Plant evolution is marked by major advances in structural characteristics that facilitated the highly successful colonization of dry land. Underlying these advances is the evolution of genes encoding specialized proteins that form novel microtubular arrays of the cytoskeleton. This review investigates the evolution of plant families of microtubule-associated proteins (MAPs) through the recently sequenced genomes of Arabidopsis thaliana, Oryza sativa, Selaginella moellendorffii, Physcomitrella patens, Volvox carteri and Chlamydomonas reinhardtii. The families of MAPs examined are AIR9, CLASP, CRIPT, MAP18, MOR1, TON, EB1, AtMAP70, SPR2, SPR1, WVD2 and MAP65 families (abbreviations are defined in the footnote to Table 1). Conjectures are made regarding the evolution of MAPs in plants in relation to the evolution of multicellularity, oriented cell division and vasculature. Angiosperms in particular have high numbers of proteins that are involved in promotion of helical growth or its suppression, and novel plant microtubular structures may have acted as a catalyst for the development of novel plant MAPs. Comparisons of plant MAP gene families with those of animals show that animals may have more flexibility in the structure of their microtubule cytoskeletons than plants, but with both plants and animals possessing many MAP splice variants.


Microtubules are a component of the plant cytoskeleton. Composed of heterodimers of αβ–tubulin, they perform many cellular roles, perhaps most notably coordinating the deposition of cellulose microfibrils in the cell wall. A number of microtubule-associated proteins (MAPs) interact with microtubules and regulate their dynamics, including the rates at which microtubules grow or shrink (Sedbrook, 2004; Sedbrook and Kaloriti, 2008; Nick, 2012).

The sequencing of the Arabidopsis thaliana, Oryza sativa, Selaginella moellendorffii, Physcomitrella patens, Volvox carteri and Chlamydomonas reinhardtii genomes (Flavell, 2009) means that it is now possible to examine the evolution of plant MAPs in a constructive manner. This extends the work of a previous study that used data mining to look at conserved MAPs in Oryza sativa (Guo et al., 2009). Chlamydomonas reinhardtii is a single-celled alga with a cell wall as well as a microtubular flagellum. Unlike many other algae (including green, red and other forms), there is no cellulosic cell wall, and Chlamydomonas is quite unusual in that it has a largely proteinaceous cell wall throughout most of its development. Volvox carteri is closely related to C. reinhardtii but is multicellular. It would be useful to include some algae that are more closely related to land plants, such as the Charales, but this will have to wait for further genome sequencing. Physcomitrella patens is a moss, with hydroids and motile sperm. Selaginella moellendorffii is a lycophyte, lacking true leaves and roots but with vasculature, stomata and motile sperm. Oryza sativa and A. thaliana are both angiosperms, with O. sativa being a monocot and A. thaliana a dicot. Both lack motile sperm. Studies of ancestral plants have revealed that the microtubule cytoskeleton is likely to be regulated differently from more modern lineages including O. sativa and A. thaliana (e.g. Cleary et al., 1993).

The development of plant structures that enabled the colonization of dry land relies upon specialized microtubule arrays. Land plants have a life cycle with two distinct multicellular generations: a haploid gametophyte and diploid sporophyte (Pires and Dolan, 2012). These specialized generations require differing suites of MAPs in order for correct cell division in meiosis and mitosis, although some proteins are involved in both these processes. For example, MOR1 is required for pollen mitosis 1 (Park et al., 1998), and its homologues in yeast and animals are required for both mitosis and meiosis (Moon and Hazelrigg, 2004), suggesting a probable role for MOR1 in plant meiosis also. TETRASPORE is a kinesin that is required for male meiotic cytokinesis in A. thaliana (Yang et al., 2003), and a separase is required for the development of radial microtubule arrays during male meiosis but not female meiosis (Yang et al., 2011). Thus, there is a distinction between the proteins involved in microtubule function even between male and female meiosis, suggesting that there will also be distinctions between proteins involved in meiosis and mitosis. No role for either TETRASPORE or the separase was observed in mitosis 1. γ–tubulin complex protein 3-interacting proteins are required for γ–tubulin complex protein localization, spindle integrity, and chromosomal stability during mitosis (Janski et al., 2012). Studies from animals also show distinct suites of MAPs involved in mitosis and meiosis (Gache et al., 2010). Likewise, leaves possess unique MAPs, including ANGUSTIFOLIA (Kim et al., 2002) and RIC1 (Fu et al., 2005), and other organs developed by plants that enabled the colonization of land, including flowers (Zhang et al., 2012), have specialized microtubule arrays.

The evolution and diversification of plant MAPs

Plants share some MAPs with animals and possess others that are unique to plants

Plants have several MAPs that are shared with metazoan, protist and fungal lineages. These include microtubule plus-end binding proteins. CLASP (Ambrose et al., 2007; Kirik et al., 2007) is a microtubule binding protein that is involved in the regulation of cell division plane rotation through a PLETHORA transcription factor and auxin-dependent pathway (Dhonukshe et al., 2012). It has functions in trichome morphogenesis (Kirik et al., 2007) and microtubule–cortex interactions (Ambrose and Wasteneys, 2008). However, it has not been shown to bind microtubule plus-ends in plants. EB1 (Chan et al., 2003) is another microtubule plus end-binding protein with three isoforms in A. thaliana. EB1a has been shown to localize to plant microtubule plus-ends, with GFP fusion proteins forming ‘comets’ (Chan et al., 2003). Reduced expression of EB1c compromises the alignment of spindle and phragmoplast microtubules (Komaki et al., 2010), and the EB1 family of proteins is involved in root responses to gravity and touch (Bisgrove et al., 2008). The roles of plus-end binding proteins in plants have been reviewed previously (Young and Bisgrove, 2010).

WVD2 is a trapoxin homologue and is required for anisotropic growth. Presumably it has other functions as well, but these are yet to be characterized (Perrin et al., 2007). MOR1 is a homologue of the XMAP215 family of tubulin polymerases that control microtubule assembly in all eukaryotes. In plants, this function is conserved, and several mutant alleles have been identified that affect the formation of interphase cortical arrays, preprophase bands, phragmoplasts and spindles (Park et al., 1998; Whittington et al., 2001; Twell et al., 2002). Plants possess a homologue of CRIPT, which is a neuronal MAP in animals that links microtubules to the post-synaptic density (Passafaro et al., 1999). The MAP65 family is another conserved family of MAPs (Chan et al., 1999). This is perhaps the best characterized family of plant MAPs, with exciting recent work showing that MAP65–3 cross-links antiparallel microtubules in the phragmoplast (Ho et al., 2012). Indeed, this formation of antiparallel microtubule arrays is a function that is conserved with the metazoan homologue Ase1 (Braun et al., 2011). So far, all members of this family investigated in plants interact with microtubules (Van Damme et al., 2004). Similarities in MAPs between plants and animals have been reviewed previously (Gardiner and Marc, 2003, 2011). Plants may also possess extremely distant homologues of Tau/MAP2/MAP4 (Gardiner et al., 2012a) and lamin (the AtMAP70 family of proteins; Gardiner et al., 2011).

Plants have evolved a number of unique MAPs. SPR1 is a unique plant MAP that is required for anisotropic plant growth (Nakajima et al., 2006). Its function is compromised under salt stress in A. thaliana (Shoji et al., 2006). SPR2 is another protein that is required to prevent helical twisting of A. thaliana (Shoji et al., 2004), but it is not known whether this family of proteins has a similar role in plants with more complex tissues. Determination of this role may be difficult due to likely redundancy of function among MAPs. Other unique plant MAP families include MAP70, a member of which (AtMAP70–5) is required for proper development of the cell wall in xylem vessels (Pesquet et al., 2010). MAP18 is a novel plant MAP from A. thaliana that destabilizes cortical microtubules and regulates directional cell growth (Wang et al., 2007).

WVD2 homologues (proteins with animal trapoxin homologues) are likely to affect microtubule dynamics, with knockout plants exhibiting reduced anisotropic growth (Perrin et al., 2007). This suggests that uncharacterized members of other plant MAP families are also likely to interact with microtubules. One reason why these homologues have not yet been assigned a function may be protein redundancy, which means that mutant plants may not show an obvious phenotype. Indeed, MAP redundancy may be of great importance to organismal homeostasis (Gardiner et al., 2012b).

Bioinformatic identification of MAP homologues

MAPs were selected if their function appeared to be predominantly microtubule-related. Some MAPs were not included due to the difficulty in determining whether related proteins are likely to interact with microtubules. For example, katanins were not surveyed as they are extremely closely related to various non-MAP ATPases. For each protein family with more than one member, the first member of that family was used in the BLASTP search (Altschul et al., 1990), i.e. TON1a, EB1a, AtMAP70–1, MAP65–1 etc. Homologous proteins for other species were found using the A. thaliana proteins and performing BLASTP on the National Center for Biotechnology Information website ( A cut-off E value of less than or equal to 1e−05 was used to predict homology. The E value is the number of proteins expected to be found at a given level of similarity, and thus this represents a 1 in 100 000 chance of this level of similarity. However, this value is arbitrary, and so it is likely that some distant homologues will be excluded from the analysis, for example a V. carteri flagellar protein with a high E value may be related to TON1.

Unfortunately, this is unavoidable, because dropping the E threshold further risks introduction of proteins that are not true homologues. This level of similarity also means that proteins are highly likely to have a conserved microtubule-based function. Annotations by genome-sequencing consortia were used to ensure the MAPs in Table 1 were not duplicated. Some proteins were excluded as the homologies detected were for domains that are not involved with microtubules, for example a number of putative AIR9 homologues were not included as the homologies detected were for a short leucine-rich domain, and two S  moellendorffii homologues of EB1a were not included as the homologies were to a calponin domain.

Table 1. Number of plant MAPs of each type, full list of MAPs is given in supplementary Data S1
 Arabidopsis thaliana (147 Mbp)Oryza sativa (430 Mbp)Selaginella moellendorffii (212 Mbp)aPhyscomitrella patens (480 Mbp)Volvox carteri (140 Mbp)Chlamydomonas reinhardtii (104 Mbp)
  1. a

    Protein names (abbreviated) are shown in the left column. The top row gives the names of the model organisms with the size of their genomes in mega base pairs (Mbp). The numbers in the remainder of the table show the total number of genes for these MAPs. S. moellendorffii has an estimated genome size of 110 Mbp; however, the sequences provided by the current genome consortium (Banks et al., 2011) are from 212 Mbp sequence contigs, representing a mixture of two diversified haplotypes. Even if this were not the case, the R2 value for the number of MAPs in each family compared to the numbers in A. thaliana would still be high, and indeed the pooling of data gives a more accurate assessment of relative MAP numbers.

  2. AIR9, auxin-induced in root cultures 9; CLASP, cytoplasmic linker protein- associated protein; CRIPT, cysteine-rich interactor of PDZ3; MAP, microtubule-associated protein; MOR1, microtubule organization 1; TON, tonneau; EB1, end-binding 1; SPR, spiral 1; WVD2, wave-dampened 2.


Organisms possess a wide range of non-coding RNAs, including small nucleolar RNAs, telomerase RNAs, tRNAs and rRNAs that function in translation, and it is now possible to predict these non-coding RNAs using computer technology, although this is still in its infancy (Song et al., 2009). However, there is nothing to suggest that any of the putative MAP genes detected in this study transcribe to non-coding RNAs. Indeed gene prediction software, on which the results of this study are predicated, works best for long, protein-encoding genes (Eddy, 2001). Although some of the genes identified as MAP homologues have yet to be shown to actually encode proteins, leaving them out of this study potentially excludes genuine protein-encoding genes, especially as there are only a limited number of expressed sequence tags for some of the species in this study. Thus the list of MAP homologues is as complete and accurate as is possible at the current time.

A previous study showed that MAPs conserved between plants and metazoans share similar secondary structure, suggesting a conservation of function (Gardiner and Marc, 2003). Indeed, several of these MAPs have subsequently been shown to be plant MAPs, including CLASP (Ambrose et al., 2007; Kirik et al., 2007), MAP1A light chain (Ketelaar et al., 2004) and prefoldin (Gu et al., 2008). Thus, using data mining to discover MAPs within biological kingdoms is even more likely to be an accurate method than discovering MAPs between the kingdoms, an approach which has already proven so successful. Many proteins behave as MAPs; for example, a proteomic screen in A. thaliana identified 122 tubulin-binding proteins (Chuong et al., 2004). Experience from animal systems suggests that most tubulin-binding proteins have microtubular functions. For this study, those proteins whose primary function is microtubule-based have been included; including all tubulin-binding proteins from plants is beyond the scope of this review, but the conclusions drawn may be limited by this.

Most plant microtubule-associated proteins are present in algae

Of the 12 plant MAP families surveyed in this study, eight are present in the algae. Indeed, a previous review suggests that most of the proteins responsible for the multicellular structures of extant land plants were present in the earliest land plants, and thus presumably in the algae that gave rise to land plants (Pires and Dolan, 2012). This may include a WVD2 homologue in C. reinhardtii, which fell just outside the cut-off E value of 5e−05 but appears to be a genuine homologue of the WVD2 family proteins. There are no TON or AIR9 homologues in the algae, lending support to the theory that these proteins are primarily important for formation of the preprophase band during cell division (Buschmann et al., 2006; Spinner et al., 2010), which algae lack. It is interesting that the preprophase band and xylem-forming microtubule arrays may have evolved at around the same point in plant evolution, perhaps reflecting similar processes for formation of both microtubular structures (Ambrose and Wasteneys, 2012).

The single-celled protist Trypanosoma brucei has an AIR9 homologue that is involved in cleavage furrow placement (May et al., 2012). Cleavage furrows are structures that are conserved between protists and animals. The presence of this homologue at the cleavage furrow shows that algae have probably lost this protein during evolution. This also suggests a homology between the animal cleavage furrow and plant microtubule arrays, including the preprophase band and phragmoplast, as discussed previously (Otegui et al., 2005), particularly as microtubules are required in both cleavage furrows (Albertson et al., 2008) and phragmoplast (Verma, 2001) to coordinate vesicle transport to the plane of cell division. This in turn may allow some of the knowledge about the cleavage furrow to be transferred to plant biology. For example, contact of astral microtubules with the cortex is required for cleavage furrow initiation, but the microtubule dynamic state is irrelevant here (Strickland et al., 2005). Is the same true for the preprophase band and phragmoplast? Another possible insight from animal cell biology is the opposite effects of stable and dynamic microtubules in focusing myosin activation to the cell equator during cytokinesis (Foe and von Dassow, 2008).

The lack of TON and AIR9 homologues in algae shows that development of the preprophase band acted as a catalyst for the development of diversity in plant MAPs. Indeed, the A. thaliana MAP TANGLED also marks the division plane during mitosis and cytokinesis (Walker et al., 2007). It may be distantly related to the animal MAP adenomatous polyposis coli. There is conservation of function between TON1 isoforms in P. patens and A. thaliana, with successful reciprocal cross-complementation achieved (Spinner et al., 2010). The development of the phragmoplast may have acted as a catalyst for the diversification of plant MAPs as well, with MAP65–3 fulfilling a specific role here as mentioned above. Proteins present in algae have been co-opted in higher plants to play developmental roles that are not required in algae. For example, SPR1 (Nakajima et al., 2004) and WVD2 (Perrin et al., 2007) are required to prevent aberrant helical torsion in developing A. thaliana plants.

Vascular plants have many more MAPs than algae

From an examination of the results of this study, it is clear that vascular plants have many more isoforms of various MAPs than algae. This is particularly true for EB1, AtMAP70, WVD2, SPR1, SPR2 and MAP65. It appears likely that the development of vasculature, leaves and roots (or at least leaf-like and root-like structures) requires an array of MAPs. For example, it has been shown that the xylem in A. thaliana requires an AtMAP70 homologue for correct development (Pesquet et al., 2010).

P. patens may have more MAPs than A. thaliana

For several families of MAPs, P. patens has more MAPs than the angiosperms. The reasons for this are unclear, but it is possible that it may be related to the mode of reproduction. Both S. moellendorffii and P. patens have motile sperm with a flagellum. Flagella are composed of microtubules and associated proteins, so it is to be expected that there will be a range of proteins that have a specialized function here. Interestingly, in angiosperms, ciliary (or flagellar) proteins appear to have taken on new functions, but many are still involved in reproduction, with high expression levels of former ciliary proteins in pollen (Hodges et al., 2011).

Multicellularity does not require an increase in MAPs

Both C. reinhardtii and V. carteri have very similar numbers of MAPs. This suggests that multicellularity, in itself, does not require an increase in the complexity of the microtubule cytoskeleton. This is in agreement with a recent study that found that multicellular complexity may readily evolve from unicellular eukaryotes, using Saccharomyces cerevisiae as a model organism (Ratcliff et al., 2012). S.cerevisiae is descended from a multicellular organism (Scannell et al., 2007), and it is probably more difficult for organisms that have only ever been single-celled to develop multicellularity. The study by Ratcliff et al. (2012) shows that, under certain conditions, there is strong selective pressure for organisms to develop multicellularity.

The number of MAPs does not appear to be linked to genome size

The number of MAPs found in a given plant does not appear to be linked to genome size. Both O. sativa and P. patens have genomes that are much larger than those of the other model plants in this study, but do not have relatively more MAP isoforms than the other two vascular plants in the study (A. thaliana and S. moellendorffii). This shows that the number of MAPs for a vascular plant appears to be regulated independently of genome size. This is perhaps not surprising considering that, while the size of nuclear genomes varies some 300 000-fold, transcriptome size varies only 17-fold (Cavalier-Smith, 2005).

Each model plant has a specific suite of MAPs but there is some correlation

With the exception of algae, where a bare minimum of MAPs are found, each lineage of plants examined in this study has its own unique suite of MAPs. For example, six CLASP isoforms are found in S. moellendorffii, many more than any other plant in this study. Fourteen WVD2 isoforms are found in P. patens, again many more than in any other plant in this study. Therefore, the complexity of microtubule function may arise in many ways, with each lineage of comparable complexity showing differing suites of MAPs. This also suggests that there is certain randomness in the numbers of plant MAPs in a given family of proteins. It appears that unique MAPs may arise in a given lineage, e.g. MAP18 is only found in A. thaliana and not any of the other model plants (although there may be a distant homologue in O. sativa), but this is rare. Interestingly, A. thaliana only has a single CLASP homologue, in contrast to all the other plant species surveyed here, including the algae.

A regression analysis performed on a plot of the number of MAPs in A. thaliana versus the number in other vascular plants (Figure 1) showed that, in general, there is good correlation between A. thaliana MAP gene numbers and those of other plants in this study: O. sativa (y = 0.7049 + 0.2052, R2 = 0.7121), S. moellendorffii (= 0.5411 x + 1.6239, R= 0.3237) and P. patens (= 0.8752 x + 8428, R2 = 0.6226). As expected, O. sativa has the closest correlation of MAP numbers, followed by Ppatens and then S. moellendorffii.

Figure 1.

Plot of the number of plant MAPs of each type. The number of MAP genes in each Arabidopsis thaliana MAP gene family is plotted against numbers from other plants in this study. Regression analysis and graph production were performed using Excel (Microsoft, The top trend line represents A. thaliana MAPs versus Selaginella moellendorffii, the middle line represents A. thaliana versus Physcomitrella patens, and the bottom line represents A. thaliana versus Oryza. sativa.

Angiosperms and P. patens show diversification in protein families required to prevent or promote helical growth

As mentioned above, several A. thaliana proteins, namely SPR1, SPR2 and WVD2, prevent aberrant helical growth of root and shoot (probably among other functions), whereas another protein, AtMAP70–5 (Pesquet et al., 2010), causes helical growth if over-expressed. Mutation of MOR1 also causes helical growth (Sugimoto et al., 2003), but this is probably not its ‘principal function’, as knockouts of this protein are fatal. Interestingly, in P. patens and the two angiosperms A. thaliana and O. sativa, these four protein families are amongst the families with the greatest number of genes.

This suggests that, in angiosperms, there is close regulation of anisotropic growth through regulation of the microtubule cytoskeleton. Conversely, angiosperm MAPs that are not involved in helical growth, or its prevention, have not diversified to the same extent. This should perhaps be seen as a general trend, rather than a hard and fast rule. The major exception to this is the MAP65 family, which has high numbers of genes in both A. thaliana and O. sativa. This family may be involved in the diversification of cell types in various tissues. For example, the ATTED–II gene co-expression tool ( shows that AtMAP65–8 is highly co-expressed with the xylem-specific cellulose synthase IRX3 (irregular xylem 3). This suggests a specific role for AtMAP65–8 in xylem development. The MAP65 family may be among the most versatile of the plant MAP families, with its microtubule cross-linking ability being required for many subcellular processes. MOR1, with two homologues in each of the four land plant model organisms surveyed here, is also involved in suppression of helical growth.

Interestingly, most of the MAPs involved in helical growth (or its absence) have yet to be shown to play roles in cell division, while most other MAP families surveyed here, including MAP65 (Smertenko et al., 2000; Dhonukshe et al., 2012), EB1 (Komaki et al., 2010), AIR9 (Buschmann et al., 2006, 2007), CLASP (Ambrose et al., 2007) and TON1 (Azimzadeh et al., 2008), are important at various stages during cell division as well as interacting with interphase microtubules. MOR1 (Park et al., 1998; Whittington et al., 2001) is again the exception.

A previous paper (Gardiner et al., 2012b) suggested that plant microtubules have a built-in propensity to form helical fractals, and plant tubulins may have developed structures that actively counteract this. Fractals are structures that are self-similar at various scales. Here, it is hypothesized that helical microtubules give rise to helical cellulose, which then causes helical growth. Point mutations in tubulins causing helical growth of A. thaliana (Ishida et al., 2007) reveal the underlying helical fractal propensity of the microtubules or may interfere with the interaction of MAPs with the microtubules. This may be the pattern among all angiosperms, as there is evidence from Osativa that microtubules are important in underlying helical growth, with a mutation in a rice α–tubulin gene causing dwarfism and right-handed helical growth (Sunohara et al., 2009).

Diversification of plant microtubular structures may have driven MAP diversification

As mentioned above, the preprophase band, phragmoplast, cortical and xylem-forming microtubule arrays all have their own specific MAPs. One xylem MAP not yet discussed in this review is MIDD1, a plasma membrane-associated MAP that suppresses microtubule rescue events, thus locally depleting microtubules and allowing the formation of cell wall pits (Oda et al., 2010). Leaves and cotyledons have unique microtubule arrays that give rise to inter-digitating epidermal cells. RIC1, another microtubule-interacting protein, promotes well-ordered microtubules in leaves and antagonizes the outgrowth-promoting ROP2 (Rho of plants 2) pathway (Fu et al., 2005). As well as controlling helical growth of plants, cortical MAPs are required to prevent isotropic growth. Indeed, the plant MAP MOR1 prevents radial swelling of roots (among other crucial functions as outlined above) (Whittington et al., 2001). Trichomes have distinctive net-like microtubule arrays. The MAP CLASP is important for correct trichome development (Kirik et al., 2007), as well as other proteins, including STICHEL (Ilgenfritz et al., 2003). The mitotic spindle also has its own array of MAPs, including MAP65–4, which promotes microtubule bundle elongation (Fache et al., 2010). The diversity of microtubule structures in plants thus accompanies a great number of different MAPs, each with a unique and critical function.

Plant MAPs are required for diverse cellular and organismal functions

The plant microtubule cytoskeleton integrates many cellular processes. MAPs are crucial here. Plants deficient in EB1 MAPs show alterations in response to touch and gravity, although microtubule dynamics do not appear important here (Bisgrove et al., 2008). EB1a also interacts with tobacco mosaic virus movement protein, and thus may play a role in the trafficking of proteins to plasmodesmata (Brandner et al., 2008).

Animal microtubule cytoskeleton compared to that of plants

Analysis of animal MAP families

Animal MAP gene family number was examined similarly to the analysis of plant MAP gene families in order to compare and contrast MAP genes between the two kingdoms (Table 2). Danio rerio, or zebrafish, is a fellow vertebrate to Homo sapiens. Drosophila melanogaster and Caenorhabditis elegans are more distantly related multicellular animals. Monosiga brevicollis is a choanoflagellate, and is considered the closest unicellular relative to animals.

Table 2. Number of animal MAPs of each type, full list of MAPs is given in supplementary Data S2
 Homo sapiens (3000 Mbp)Danio rerio (1700 Mbp)Drosophila melanogaster (150 Mbp)Caenorhabditis elegans (97 Mbp)Monosiga brevicollis (42 Mbp)
  1. Protein names (abbreviated) are shown in the left column. The top row gives the names of the model organisms with the size of their genomes in mega base pairs (Mbp). The numbers in the remainder of the table show the total number of genes for these MAPs present in each genome.

  2. CRIPT, cysteine-rich interactor of PDZ3; APC, adenomatous polyposis coli; STIM1, stromal interaction molecule 1; EB1, end-binding 1; MAP, microtubule-associated protein; CLASP, cytoplasmic linker protein-associated protein; CLIP cytoplasmic linker protein; SCG10, superior cervical ganglion 10.


As well as MAP families shared with plants, a number of unique animal MAP families are included in this survey. SCG10 (superior cervical ganglion 10) is a protein that destabilizes microtubules by sequestering tubulin subunits (Curmi et al., 1999). The adenomatous polyposis coli protein has multiple roles, and is both a centrosomal protein (Bahmanyar et al., 2009) and a protein that is involved in neuronal polarization (Barth et al., 2008). The Tau/MAP2/MAP4 and MAP1B families of animal MAPs are involved in the stabilization of microtubules in neurons (Halpain and Dehmelt, 2006; Ke et al., 2012), and CLIP (cytoplasmic linker protein) is another plus-end binding protein (Slep, 2010) as is STIM (stromal interaction molecule 1) (Smyth et al., 2012). Evidently, animal MAPs have complex functions in a similar fashion to plant MAPs.

Animal MAP gene numbers are more flexible than those of plant species

Plotting H. sapiens MAP gene numbers against those of D. rerio, D. melanogaster and C. elegans and performing regression analysis revealed a high degree of correlation between H. sapiens and D. rerio, as may be expected (y = 1.8382 x−0.6618, R2 = 0.8448) (Figure 2). However, for the other two animals, D. melanogaster and C. elegans, there is very little correlation with the number of H. sapiens MAPs. This may be partly due to the large difference in genome size between the vertebrates and other animals. However, this shows that there is more variation in the numbers of animal MAP genes than in plants. This suggests that plants are more constrained in the relative numbers of MAPs of each family, and that they have become more evolutionarily canalized with respect to the microtubule cytoskeleton than animal lineages. These data again suggest that, in animals, as in plants, randomness plays a certain role in determining MAP numbers. MAP families in animals also appear not to have diversified to the same extent as in plants. Of all the MAPs surveyed in the animal lineages, the highest number of MAPs for a given family is 6, whereas many plant MAP lineages exceed this number.

Figure 2.

Plot of the number of animal MAPs of each type. The number of MAP genes in each Homo sapiens MAP gene family is plotted against numbers from other animals in this study. Regression analysis and graph production were performed using Excel. The top two trend lines represent H. sapiens versus Drosophila melanogaster and H. sapiens versus Caenorhabditis elegans, the bottom line represents is H. sapiens versus Danio rerio.

The MAP gene number data for M. brevicollis, a choanoflagellate considered the closest single-celled relative of metazoans, show that it lacks many of the MAP families present in multicellular animals (Table 2). Some of the animal MAPs that are absent in M. brevicollis are shared with plant lineages, while others are specific to animals. This again suggests that plants, even single-celled plants such as C. reinhartii, are more constrained in their use of MAP families, and perhaps more evolutionarily canalized with respect to the proteins regulating the microtubule cytoskeleton.

Common and distinct properties of animal and plant MAPs

Post-translational modification of tubulin

Although both plants and animals have post-translationally modified tubulin, it appears that animals may have a more complex range of modifications than plants. In H. sapiens, there are no fewer than 11 proteins of the tubulin-tyrosine ligase-like (TTLL) family. These proteins not only catalyse the C–terminal tyrosination of α–tubulin (Szyk et al., 2011), but also tubulin-glycine ligation (Wloga et al., 2009) and tubulin polyglutamylation (Janke et al., 2005). Plants have TTLLs, but A. thaliana, for example, only appears to possess one isoform (Gardiner and Marc, 2003) and its role remains to be determined. Another α-tubulin post-translational modification is lysine acetylation. In animals, this is crucial to the function of kinesin motor proteins, with kinesin being targeted to acetylated microtubules (Reed et al., 2006). Two proteins, elongator protein 3 (ELP3) (Creppe et al., 2009) and α–tubulin acetyltransferase (ATAT) (Akella et al., 2010), acetylate α–tubulin in animals. Although plants possess ELP3, they lack ATAT, again suggesting that the mechanisms of post-translational tubulin modification may be simpler in plants than in animals.

Splice variants of MAPs in animals and plants

A proteomic analysis of Bos taurus brain MAPs found a large number of splice variants, particularly of Tau protein (Kozielski et al., 2011). This was confirmed when performing BLAST searches against animal proteomes, with large numbers of MAP splice variants annotated in the National Center for Biotechnology Information database. These splice variants are of functional importance, with mis-regulation of the 5′ splice site of Tau exon 10, for example, leading to abnormal ratios of Tau isoforms and causing frontotemporal dementia (Wang et al., 2011). Few plant MAP splice variants have been described in the literature, although there are two splice variants of A. thaliana MAP65–1 in the National Center for Biotechnology Information database: one of 587 amino acids and the other of 616 amino acids. A recent study surveyed genome-wide A. thaliana splice variants, with implications for MAPs (Marquez et al., 2012). For example, this study revealed four splice variants of the plant MAP MOR1. Ab initio prediction of plant MAP splice variants is possible with the AUGUSTUS program (Stanke et al., 2006). Alternative splicing in plants has influenced the evolution of complex networks of regulation of gene expression, and variation in alternative splicing helps plants adapt to environmental conditions (Syed et al., 2012).

MAPs that are not included in the analysis

As mentioned above, some MAP families were not be included in the bioinformatics analysis. Perhaps the most important of these families is the motor proteins, kinesins. Plants possess many more kinesins than animals, with A. thaliana having 61 of them (Lee and Liu, 2004). This suggests that plants may have evolved alternative ways to introduce variability between taxa and complexity of the microtubule cytoskeleton within taxa, apart from the number of members of MAP families (not including kinesins). Another major MAP family that has not been discussed is the microtubule-severing katanin family, for which a number of A. thaliana mutants with interesting phenotypes are available (e.g. Burk and Ye, 2002; Uyttewaal et al., 2012). Thus the data from katanin mutants suggests that plants may use other mechanisms than animals to increase complexity, including multi-tasking by MAPs such as katanin.

A diversity of microtubule-binding sites supports a diversity of MAPs

Both plant and animal MAPs use a range of microtubule-binding motifs and microtubule-binding sites. In animals, a number of plus end-binding MAPs possess an SxIP microtubule-binding motif (Honnappa et al., 2009). It is not yet known whether plant MAPs utilize this motif, but there are over 3000 proteins in the A. thaliana proteome with the motif (Young and Bisgrove, 2010). TTLL uses an ancient, recently defined, binding site (Szyk et al., 2011) that is probably conserved between plants and animals. Two other tubulin post-translational modification proteins, ELP3 and ATAT, are predicted to bind in the microtubule lumen (Creppe et al., 2009; Akella et al., 2010). Plant katanin may have a unique microtubule-binding domain (Stoppin-Mellet et al., 2007), and A. thaliana MAP18 and animal MAP1B may share a commons microtubule-binding motif (Wang et al., 2007). It is possible that the Tau/MAP2/MAP4 microtubule-binding site is conserved between animals and plants, as a MAP4 microtubule-binding domain–GFP fusion protein localizes to plant microtubules (Marc et al., 1998), but the identities of any plant MAPs with this microtubule-binding motif remain unclear. A diversity of functional binding sites on tubulins has probably acted as a catalyst to the development of multiple MAP lineages in both plants and animals.


The recent sequencing of several plant genomes has paved the way for studies comparing protein genes across evolutionary history, as I have attempted to do in this study. The sequencing of genomes has opened the door to more detailed studies of MAPs in various ancestral species, which should cast light on their function in economically important species as well.

By comparison of plant MAP gene families with animal MAP gene families, it is apparent that animals may have more variation in their microtubule cytoskeleton structure in terms of the MAP families present and number of genes within MAP families, although plants may have other ways of introducing some flexibility to their microtubule cytoskeletons. The assertion that animals have more variation in the microtubule cytoskeletons is supported by data from the single-celled choanoflagellate M. brevicollis, which lacks many MAP gene families seen in multicellular animals, whereas the single-celled plant C. reinhardtii and the related alga V. carteri possess most of the MAP gene families present in multicellular plants.


Jan Marc (The School of Biological Sciences, The University of Sydney) assisted with the editing.