WVD2 is a novel microtubule-associated protein in Arabidopsis thaliana


For correspondence (fax +1 608 262 2976; e-mail phmasson@wisc.edu).


Arabidopsis WAVE-DAMPENED 2 (WVD2) was identified by forward genetics as an activation-tagged allele that causes plant and organ stockiness and inversion of helical root growth handedness on agar surfaces. Plants with high constitutive expression of WVD2 or other members of the WVD2-LIKE (WDL) gene family have stems and roots that are short and thick, have reduced anisotropic cell elongation, are suppressed in a root-waving phenotype, and have inverted handedness of twisting in hypocotyls and roots compared with wild-type. The wvd2-1 mutant shows aberrantly organized cortical microtubules in peripheral root cap cells as well as reduced branching of trichomes, unicellular leaf structures whose development is regulated by microtubule stability. Orthologs of the WVD2/WDL family are found widely throughout the plant kingdom, but are not similar to non-plant proteins with the exception of a C-terminal domain distantly related to the vertebrate microtubule-associated protein TPX2. in vivo, WVD2 and its closest paralog WDL1 are localized to interphase cortical microtubules in leaves, hypocotyls and roots. Recombinant glutathione-S-transferase:WVD2 or maltose binding protein:WVD2 protein bind to and bundle microtubules in vitro. We speculate that a C-terminal domain of TPX2 has been utilised by the WVD2 family for functions critical to the organization of plant microtubules.


The cytoskeleton is a major regulator of plant growth and form, and microtubules are particularly important determinants of phenotype. Growth and morphology may be perturbed by mutations in genes encoding tubulin isoforms (Abe et al., 2004; Thitamadee et al., 2002), application of microtubule-stabilizing or -destabilizing drugs (Nakamura et al., 2004), overexpression of tubulin fused to epitope tags or reporter proteins (Abe and Hashimoto, 2005), or mutations in microtubule-associated proteins (MAPs; Buschmann et al., 2004; Nakajima et al., 2004; Sedbrook, 2004; Sedbrook et al., 2004; Shoji et al., 2004; Smith et al., 2001; Whittington et al., 2001). Many of the treatments or mutant alleles that have been described cause organs such as roots or hypocotyls to twist. This twisting occurs with a characteristic left- or right-handedness that is directionally distinct from the wild-type state. Such helical growth phenotypes may arise from an interplay between cortical microtubules underlying the plasma membrane and cellulose microfibrils synthesized at the plasma membrane and extruded into the cell wall, as orientation of these two components are often (but not always) correlated (Baskin, 2001; Paredez et al., 2006). However, recent studies suggest that the relationship between cortical microtubules and the load-bearing cellulose microfibrils may be indirect (Baskin et al., 2004; Himmelspach et al., 2003; Sugimoto et al., 2003; Wasteneys, 2004), or at least complex (Paredez et al., 2006).

Several plant MAPs have been identified in recent years (Sedbrook, 2004). Some, such as MOR1, have homologs in non-plant species but may have specialized functions in plants (Gard et al., 2004; Hussey et al., 2002; Kawamura et al., 2006; Whittington et al., 2001). Others, such as TOR1 and SPR1, appear to be specific to plants (Buschmann et al., 2004; Korolev et al., 2005; Nakajima et al., 2004; Sedbrook et al., 2004; Shoji et al., 2004; Smertenko et al., 2000). The biological functions of some of these proteins are unknown, while others have been proposed to have roles in microtubule nucleation, cross-linking, severing or stabilization (Hashimoto and Kato, 2005; Sedbrook, 2004).

The WAVE-DAMPENED 2 (WVD2) locus was identified by a forward genetic screen in Arabidopsis (Yuen et al., 2003). Overexpression of WVD2 by an activation-tagged allele results in inhibition of cell elongation, right-handed root twisting (when the seedlings grown on nutrient plates are viewed from above), left-handed leaf petiole twisting, and suppression of a root-waving growth phenotype on tilted plates containing hard agar medium (Yuen et al., 2003). The WVD2 protein is not highly similar to any characterized protein yet described, but putative orthologs occur widely throughout the plant kingdom and are present in monocots, dicots, gymnosperms and the non-vascular moss Physcomitrella patens. Additionally, the C-termini of these proteins share distant similarity to the C-terminal domain of an unconventional vertebrate MAP named Targeting Protein for Xklp2 (TPX2) that is critical for vertebrate mitotic spindle assembly (Gruss and Vernos, 2004; Wittmann et al., 2000). The wvd2-1 mutant is more sensitive to root growth inhibition by the microtubule-destabilizing drug oryzalin than wild-type, and cortical microtubule orientation is altered in epidermal cells of the root elongation zone (Yuen et al., 2003). In this study, we find that WVD2 and its closest paralog WVD2-LIKE-1 (WDL1) are localized to microtubules in vivo. Overexpression of WVD2 results in alteration of cortical microtubule orientation in peripheral root cap cells. Increased expression of WVD2 also alters the number of trichome branches, unicellular leaf surface structures whose form is influenced by microtubule stability (Abe and Hashimoto, 2005; Mathur and Chua, 2000). In addition, recombinant WVD2 protein binds to and bundles microtubules in vitro. We discuss possible functions of WVD2/WDL proteins based on the protein localization patterns, in vitro biochemical activity, and in vivo effects on microtubule organization and dynamics.


Anti-GST–WVD2 signal overlaps with anti-tubulin signal

To examine the intracellular localization of the WVD2 protein, we raised polyclonal antibodies against purified recombinant GST–WVD2 protein. Immunoblotting was performed on protein extracts from plants that overexpress WVD2 (wvd2-1 in Ler ecotype background or 35S::WVD2 transgenic plants in Col-0 ecotype background), wild-type plants, or plants homozygous for an intragenic suppressor of wvd2-1, called wvd2-2. wvd2-2 is derived from wvd2-1 by a mutation that converts Arg152 to a stop codon; it restores wild-type root waving and skewing on hard agar surfaces (Yuen et al., 2003). The premature stop codon is 44 amino acids downstream of the conserved KLEEK motif (Figure 1a and Yuen et al., 2003). Thus, the wvd2-2 allele results in constitutive expression of a truncated form of the protein lacking the C-terminal 66 amino acid residues. The anti-GST–WVD2 antibodies recognized a protein with apparent molecular weight of 32 kDa in wild-type and WVD2-overexpressing seedlings. Signal was strong in the overexpressing lines and hardly detectable in wild-type seedlings, in agreement with the low steady-state gene expression levels for this locus (Yuen et al., 2003). A truncated signal of apparent molecular weight 25 kDa was observed in wvd2-2 seedlings (Figure 1b). Signal corresponding to full-length WVD2 was not observed in wvd2-2 seedlings. Immunoblotting experiments consistently showed both native and recombinant WVD2 protein migrating in denaturing Tris–glycine SDS–PAGE gels at a location higher than expected for its calculated molecular weight of 23.3 kDa (Figure 1b, and data not shown).

Figure 1.

 Analysis of WVD2 by immunoblotting and immuno-confocal microscopy.
(a) Gene structure of WVD2. In wvd2-1, a 35S element within a Ds insertion upstream of the open reading frame (Yuen et al., 2003) directs constitutive overexpression of the locus. A suppressor screen of wvd2-1 identified wvd2-2, which bears a C to T transition resulting in a premature stop codon. Thick arrows indicate translational start and stop sites. The black bar indicates the coding sequence region conserved with vertebrate TPX2. The figure is drawn to scale.
(b) Immunoblot using anti-GST–WVD2 antibodies against total protein extracted from 7-day-old seedlings. The upper marker shown by a horizontal black bar on the left is 33.8 kDa; the lower marker is 26.9 kDa.
(c) Immuno-confocal image of freeze-fractured epidermal cells from rosette leaves of a 4-week-old 35S::WVD2 plant. Left, signal from anti-GST–WVD2 antibodies detected with rhodamine-labeled secondary antibodies; middle, signal from anti-tubulin antibodies detected with FITC-labeled secondary antibodies; right, merged image. Scale bar = 20 μm. The smaller images on the right are representative immuno-confocal images of mitotic structures in a root tip of a 7-day-old Col-0 wild-type seedling (one of three biological repeats, each with four tissue samples). Red signal, anti-GST–WVD2; green signal, anti-tubulin. Shown clockwise: mitotic spindle; preprophase band; phragmoplast. Scale bar = 10 μm for all root tip micrographs.

We performed immunolocalization experiments using antibodies to GST–WVD2 and α-tubulin on freeze-fractured leaf tissue of transgenic plants overexpressing WVD2. The anti-WVD2 antibody signal co-localizes with the anti-tubulin antibody signal in leaf epidermal pavement cells (Figure 1c). In dividing cells of wild-type root tips, anti-WVD2 signal did not localize to the microtubular preprophase band or the phragmoplast (Figure 1c). Instead, signal appeared to be localized to the nuclear periphery at these points in the cell cycle. WVD2 signal was also proximal to the spindle poles, but it was not possible to discern whether this represented localization to chromosomes or to the spindle itself.

WVD2 and WDL1 localize to cortical microtubules in vivo

To study in further detail the subcellular localization of this protein and its closest homolog, WVD2-LIKE 1 (WDL1), we generated stably transformed transgenic Arabidopsis lines expressing WVD2–GFP, WDL1–GFP or GFP–WDL1 under the control of the CaMV 35S constitutive promoter. The GFP signal in all of these lines indicated that the fusion proteins localized to cortical microtubules in leaf epidermal pavement cells, guard cells, epidermal cells of hypocotyls, trichomes and roots (Figure 2b,c, and data not shown). The linear structures are labeled relatively uniformly throughout their length. GFP signal was examined in lines with a range of expression levels, and the localization patterns were not appreciably different in eight lines that showed strong expression of the transgene and a wvd2-1-like helical growth phenotype (Figures 2c and 3) compared with 16 lines that showed weaker transgene expression and a wild-type helical growth phenotype (Figure 2b). Overexpression of either GFP–WDL1 (N-terminal fusion) or WDL1–GFP (C-terminal fusion) both caused a wvd2-1-like growth phenotype upon overexpression, indicating that the location of the GFP fusion did not affect function in vivo (Y. Wang and P.H. Masson, data not shown.) Untransformed controls (Figure 2c, and data not shown) indicated that autofluorescence was limited to plastids and stomatal lips under the conditions used.

Figure 2.

 WVD2 and WDL1 microtubule localization.
GFP signal observed in 7-day-old transgenic 35S::WVD2–GFP plants (a, b), untransformed plants (c) or 35S::GFP–WDL1 plants (d–f). (a, c, d) Leaf epidermis; (b) hypocotyl epidermal cells; (e, f) leaf trichome. Panel (c) shows the autoflorescence from plastids and stomatal lips observed under these conditions. Scale bar = 10 μm for all micrographs. The results shown are representative for 48–54 seedlings per genotype.

Figure 3.

 GFP signal in 35S::WDL1–GFP seedlings treated with the microtubule-disrupting drug oryzalin.
Seven-day-old seedlings were incubated with DMSO (solvent control) or 25 μm oryzalin, and GFP signal was assessed in hypocotyl epidermal cells. (a) DMSO treatment for 30 min; (b) oryzalin treatment for 30 min; (c) oryzalin treatment for 26 min; (d) oryzalin treatment for 30 min. Scale bar = 10 μm (applies to all micrographs). Representative results are shown from two experimental repeats.

To confirm that the structures observed were indeed cortical microtubules, 7-day-old seedlings expressing WDL1–GFP were incubated on microscope slides in the presence or absence of the microtubule-destabilizing drug oryzalin. GFP signal in hypocotyl epidermal cells was monitored over time (Figure 3). Over the course of 30 min, the GFP signal in seedlings incubated with oryzalin became diffuse, while the GFP signal in seedlings incubated in buffer was unaffected. In summary, these results indicate that WVD2 and WDL1 can act as MAPs in vivo.

Recombinant WVD2 protein binds to and bundles microtubules in vitro

Microtubule binding studies were conducted to determine whether recombinant WVD2 protein could associate with microtubules in vitro. Recombinant WVD2 tagged with c-myc and His6 epitopes was expressed in Escherichia coli, purified, and used for a microtubule spin-down assay (Cytoskeleton Inc. http://www.cytoskeleton.com). Bovine tubulin was polymerized in vitro, taxol-stabilized, and exposed to a known microtubule-associated protein (bovine MAP2), a negative control protein that does not associate with microtubules (bovine serum albumin), buffer or WVD2-c-myc-His6. The samples were then centrifuged at 100 000 g through a glycerol cushion, and proteins present in the supernatant and pellet fractions were analyzed by SDS–PAGE and immunoblotting (Figure 4a,b). MAP2 was found to co-sediment with polymerized tubulin in the pellet fraction, while BSA did not. Densitometry quantification using NIH ImageJ software (National Institutes of Health, Bethesda, MD, USA) showed that 4.2-fold more WVD2:c-myc:His6 was present in the pellet fraction in the presence of microtubules than in their absence, indicating that recombinant WVD2 protein had co-sedimented with polymerized microtubules. A similar result was observed using GST–WVD2 recombinant protein in place of WVD2:c-myc:His6 (data not shown). To determine whether WVD2 had any effect on microtubule bundling in vitro, rhodamine-labeled tubulin was polymerized and then the resulting microtubules were incubated with recombinant WVD2 [either GST- or maltose binding protein (MBP)-tagged purified protein], control fusion tag protein [purified GST (Figure 4c) or MBP (data not shown)], or buffer. Microtubules incubated with recombinant WVD2 protein formed bundles (Figure 4c). In some cases, WVD2-induced microtubule bundling resulted in extensive, knot-like structures of multiple co-aligned microtubules. Visualization of WVD2-bundled microtubules by transmission electron microscopy showed that microtubules were tightly aligned in close proximity (Figure 4c).

Figure 4.

in vitro association of recombinant WVD2 with microtubules.
(a) Microtubule co-sedimentation assay. Tubulin was polymerized in vitro and incubated in presence or absence of mammalian microtubule-associated protein (MAP2), the negative control protein bovine serum albumin (BSA), or buffer (- cont). Following incubation, the reactions were centrifuged at 100 000 g through a glycerol cushion, and the pellet (P) and supernatant (S) were analyzed by SDS–PAGE and stained with Coomassie brilliant blue (left panel) to detect control proteins (MAP2, 300 kDa; BSA, 67 kDa) or tubulin (migrating as a doublet around 50 kDa).
(b) WVD2:c-myc:His6 protein, detected by immunoblot using anti-c-myc antibodies, in the absence or presence of microtubules. Representative results are shown from three experimental replicates.
(c) Rhodamine-labeled tubulin was polymerized in vitro and incubated with fusion tag protein (GST), no protein (buffer) or recombinant WVD2 protein (GST–WVD2 or MBP–WVD2) as indicated, followed by direct visualization using confocal microscopy. Scale bar = 10 μm. Representative results are shown from two (MBP–WVD2) or four (GST–WVD2) experimental replicates. Far right, a GST–WVD2-induced microtubule bundle was visualized directly using transmission electron microscopy. Scale bar = 0.25 μm.

Cortical microtubule orientation is altered in wvd2-1 root cap cells

Peripheral root cap cells are critical for root responses to tactile stimuli (Fasano et al., 2002; Massa and Gilroy, 2003), and overexpression of WVD2 alters root growth on hard agar media, a bioassay that requires coordination of gravity signal transduction and mechano-transduction (Buer et al., 2003; Okada and Shimura, 1990; Thompson and Holbrook, 2004). We examined microtubule orientation in peripheral root cap cells by immuno-confocal microscopy (Figure 5). Microtubule orientation was quantified in root cap cells of each genotype using ImageJ software. Microtubules were largely transversely oriented in wild-type and wvd2-2 peripheral root cap cells, while organization was greatly perturbed in wvd2-1 root caps (Figure 5, lower panels). A variety of microtubule orientations were observed in wvd2-1root cap cells, including longitudinal and oblique positions less frequently observed in wild-type or wvd2-2 cells. The overall trend in wvd2-1 root cap cells was towards either right-handed or left-handed oblique orientation, accompanied by a great reduction in transversely oriented microtubules.

Figure 5.

 Microtubule orientation in roots of WVD2 overexpressing or wild-type seedlings.
Microtubules in root caps of 7-day-old seedlings were visualized using anti-tubulin antibodies. (a) Wild-type (Ler); (b) wvd2-2; (c) wvd2-1. Scale bar = 10 μm (applies to all micrographs). Lower panels, corresponding histograms of microtubule orientation in root tip cells. Microtubule angles were defined as: transverse, 0° (mid-point of histogram x-axis); left-handed oblique, from −90 to 0° (left half of x-axis); right-handed oblique, from 0 to 90° (right half of x-axis). The results are based on n = 429 microtubules from 24 cells of six seedlings (Ler), 318 microtubules from 23 cells of four seedlings (wvd2-2), or 333 microtubules from 27 cells of four seedlings (wvd2-1).

Increased expression of WVD2 affects trichome branching

Trichomes are elaborate unicellular structures that emerge from the leaf epidermis and branch into a characteristic number of spike-like projections. As microtubule stability is reported to play a role during the well-characterized process of trichome branching (Abe and Hashimoto, 2005; Mathur and Chua, 2000), we evaluated the branching phenotype of populations of trichomes from wvd2-1, wvd2-2 and wild-type rosette leaves (Table 1). Trichome branching is greatly reduced in wvd2-1 compared to its wild-type (Ler) ecotype control, with wvd2-1 showing a 21-fold increase in trichomes bearing two tips, and half as many three-tipped trichomes compared to Ler (Table 1). No four-branched trichomes were observed in wvd2-1 rosette leaves, although they were occasionally seen in Ler leaves. By comparison, wvd2-2 showed only a slight decrease in trichome branching.

Table 1.   Trichome branching in rosette leaves of WVD2 overexpression lines and controls
LineNo. trichome branch tips (% of total)Total
  1. Branching was quantified for the indicated number of trichomes in 5–12 rosette leaves of 21-day-old plants grown on hard agar medium. ND, none detected. Values with identical superscripts are not significantly different (one-way anova, 95% confidence intervals for the mean based on pooled SD). Data shown are from one of two experimental repetitions.

wvd2-1 (Ler)ND46.06c53.94cNDND330
wvd2-2 (Ler)ND10.70a88.48d0.82aND243

Microtubule interactions are altered in the wvd2-1 background

To examine the effect of WVD2 overexpression on cortical microtubule dynamics, we crossed wvd2-1 to a line expressing enhanced GFP (eGFP) fused to AtTUB1A (Shaw et al., 2003) and identified lines homozygous for both loci. Because the eGFP–AtTUB1A reporter is not well-expressed in root tissue, we analyzed time-lapse images of cortical microtubule signal in hypocotyl epidermal cells for all lines. We noticed that the microtubule populations observed in wild-type consisted mainly of transversely oriented microtubules and a highly dynamic population of longitudinally and obliquely oriented microtubules (Videos S1 and S2.) However, this sub-population of cortical microtubules appeared less dynamic in a wvd2-1 background compared to a Col-0 background (Video S3). When microtubule interactions observed in real time were quantified and categorized as ‘crossover’, ‘catastrophic’ or ‘zippering’ events (Dixit and Cyr, 2004), the wvd2-1 allele was associated with a 1.9-fold increase in zippering, defined as an interaction between the plus end of a growing microtubule with an existing microtubule, followed by a change in direction of growth to result in co-alignment (Figure 6). There was a corresponding decrease in crossover events, in which a microtubule intersecting with another microtubule continues on its original growth vector.

Figure 6.

 Microtubule interactions observed in eGFP:TUB1-7 lines expressed in wild-type (Col-0) or wvd2-1 allele backgrounds.
Microtubule interactions were quantified in time-lapse series of eGFP signal in hypocotyl epidermal cells of 3-day-old transgenic eGFP:TUB1-7 seedlings in wild-type (Col-0) ecotype background (n = 49) or in homozygous wvd2-1 mutant background (n = 32). Microtubule interactions were categorized as co-alignment (‘zippering’), crossover or catastrophic according to the definitions given by Dixit and Cyr (2004), and angles of incidence were measured using ImageJ.


Orthologs of WVD2 and WDL proteins are found widely throughout the plant kingdom, but are not well-conserved with non-plant proteins. Within the eight-member WVD2/WDL family in Arabidopsis, a region of 57 residues rich in highly charged residues at the C-terminus is most well conserved and includes an invariant KLEEK motif, whereas the N-termini vary greatly in size and sequence. Amino acid identity ranges from 19.2% to 55.3% when compared throughout the full length of the WVD2/WDL proteins and from 50% to 88% when compared within the KLEEK region alone. Residues 92–136 of WVD2, which are within the KLEEK domain, are predicted by PAIRCOILS (Berger et al., 1995) to form a coiled-coil structure (probability = 0.896). Such structures are frequent mediators of protein–protein interactions (Burkhard et al., 2001), including MAP–microtubule interactions (Korolev et al., 2005) or dimerization between MAPs (Wade and Kozielski, 2000). Additionally, although the level of amino acid identity is low, the amino acid composition of the KLEEK region is strikingly similar to a corresponding domain at the C-terminus of TPX2, including six absolutely conserved residues (Figure 7). This region is also predicted to form a coiled-coil structure in TPX2 (Eyers and Maller, 2004). It should be noted that the WVD2 and WDL family members are distinct from the likely full-length Arabidopsis ortholog of TPX2 (At1g03780). In addition to At1g03780, three additional loci predicted to encode truncated TPX2 homologs are present in the Arabidopsis genome.

Figure 7.

 A C-terminal domain is conserved between the WVD2/WDL gene family and the vertebrate MAP TPX2.
Top panel, domain conserved between WVD2/WDL family members of Arabidopsis and TPX2 proteins. Amino acid residues are colored using the clustalx scheme: orange, glycine (G); yellow, proline (P); blue, small and hydrophobic amino acids (A, V, L, I, M, F, W); green, hydroxyl and amine amino acids (S, T, N, Q); red, charged amino acids (D, E, R, K); cyan, histidine (H) and tyrosine (Y). Conserved identical residues are indicated by a black circle and similar residues are indicated by an asterisk. Lower panel, phylogenetic tree based on clustalw alignment of the conserved domain for the Arabidopsis WVD2 and WDL protein family, two Oryza sativa (rice) WDL proteins (OsWDLA and OsWDLB), a WDL homolog of Physcomytrella patens (PpWDL) and TPX2 homologs from Mus musculans (mouse) (MmTPX2), Xenopus laevis (XlTPX2) and Arabidopsis thaliana (AtTPX2).

TPX2 was first described as a protein necessary for mitotic spindle formation in Xenopus laevis oocyte extracts (Wittmann et al., 1998). Originally identified for its role in localization of a kinesin-like protein (Xklp2) to spindle poles (Wittmann et al., 1998, 2000), it has since been found to participate in spindle assembly in a Ran-GTP-mediated pathway (Gruss and Vernos, 2004; Gruss et al., 2001, 2002; Trieselmann et al., 2003; Tsai et al., 2003). Several recent biochemical studies have analyzed the functional domains of TPX2 (Brunet et al., 2004; Eyers and Maller, 2004; Schatz et al., 2003; Stewart and Fang, 2005; Trieselmann et al., 2003). In truncation experiments, the C-terminal domain overlapping with the WVD2/WDL family region (Figure 6) did not bind to or nucleate microtubules in vitro (Brunet et al., 2004). However, it was essential but not sufficient for microtubule nucleation in a Xenopus egg cell extract assay (Brunet et al., 2004). This difference between in vitro and cell-based assay function was interpreted as involving a complex mechanism requiring unidentified importin α-interacting factors that depend upon Ran-GTP. Another possible explanation is that this domain of TPX2, which was found to self-aggregate (Brunet et al., 2004), has a higher affinity for self-interaction than its affinity for microtubules in vitro.

Regardless, it is interesting to note that recombinant fusion tag versions of full-length WVD2 containing a region distantly similar to TPX2, but lacking all other domains found in TPX2, could bind to and bundle microtubules in vitro. It is highly likely that the WVD2 family member functions are at least partially distinct from those of the full-length Arabidopsis homolog of TPX2. We speculate that a C-terminal domain of TPX2 has been utilised by the WVD2/WDL family for functions critical to the cortical microtubules. The function of WVD2 and potentially WDL1 as interphase MAPs is even more interesting in light of the fact that vertebrate TPX2 does not seem to play a major role in interphase cells (Harel and Forbes, 2004).

WVD2 and WDL1 were observed to co-localize with cortical microtubules in all tissues examined. The WVD2 signals observed did not differ between wild-type and overexpression alleles (Figures 1 and 2, and data not shown), suggesting that WVD2 does not act in a neomorphic way upon overexpression. The fact that overexpression of these proteins alters anisotropic cell expansion suggests that the WVD2 family plays a key regulatory role during growth. Considering the microtubule bundling activity observed using recombinant WVD2, one possibility is that WVD2 may affect the association or orientation of microtubules in elongating cells. It has been proposed that left-handed (as viewed from above the seedling) helical growth phenotypes may arise as an intermediate state of slightly destabilized microtubules, with further destabilization resulting in isotropically swollen cells (Abe and Hashimoto, 2005; Abe et al., 2004). Proteins capable of bundling microtubules in vitro are often assumed to have a stabilizing function in vivo, and yet roots of wvd2-1 plants show increased sensitivity to the microtubule-destabilizing drug oryzalin (Yuen et al., 2003), and the reduced trichome-branching phenotype of wvd2-1 is similar to that of mutants with destabilized microtubules (Folkers et al., 2002; Mathur and Chua, 2000; Oppenheimer et al., 1997; Qiu et al., 2002).

These results could be reconciled if the ability of the microtubule cytoskeleton to respond to changing conditions within the cell brought about by increased turgor during growth or perturbation by drugs requires a period of transient destabilization before adaptation to newly established conditions. In this scenario, constitutive stabilization could result in a less responsive cortical array. Indeed, although short-term exposure to the microtubule stabilizing drug taxol could elicit branch-point formation in the trichome mutant zwichel, continuous exposure to the drug only resulted in isotropic swelling (Mathur and Chua, 2000). It should also be cautioned that not all microtubule stabilization functions are necessarily equal. Even when considering only in vitro activity, it is noteworthy that MAPs may not have the same stabilization role in every condition. For example, the tobacco MAP-65 isoform NtMap65-1b protected bundled microtubules from cold-induced polymerization in vitro, but not from katanin-induced depolymerization (Wicker-Planquart et al., 2004). Finally, is it possible that WVD2/WDL proteins have additional biochemical roles in vivo, or interact with other factors that modify their function, perhaps in a regulated manner. Notably, the partial trichome-branching defect observed in wvd2-2 suggests that this truncated allele may retain some function in the regulation of trichome development.

Further analysis on the roles of WVD2 and WDL members during mitosis and meiosis is necessary, but the results presented here suggest that WVD2 may co-localize with the mitotic spindle poles or condensed chromatin and is absent from other cell division structures. This is in contrast to several isoforms of the plant MAP-65 family, which localize to the spindle mid-zone and phragmoplast (Mao et al., 2005; Muller et al., 2004), SPR1/SKU6, which localizes diffusely to the spindle as well as all other division structures (Furutani et al., 2000; Sedbrook et al., 2004), and SPR2, which is localized to all division structures and is diffuse throughout the spindle but absent from the mid-zone (Buschmann et al., 2004; Shoji et al., 2004). No obvious mitotic or meiotic defects are observed in WVD2 and GFP-tagged WDL1-overexpressing plants, but a comprehensive study of spindle formation during the cell cycle is needed in mutant and overexpression lines. Of course, care must be taken when interpreting data from activation-tagged alleles with regard to the role of such genes when expressed at wild-type levels and spatio-temporal patterns; however, analysis of loss-of-function alleles of some members of the WDL family also supports roles in growth regulation (Y. Wang, J. Will, B. Harrison and P.H. Masson, unpublished data).

In addition to its roles in growth regulation, WVD2 can also modulate environmental responses to mechano-stimulation. In Arabidopsis roots, touch stimulation negatively and transiently regulates responses to gravity, which may reflect the need for roots to integrate both mechanical and gravitational signals while navigating through soil (Evans, 2003; Massa and Gilroy, 2003; Massa et al., 2003). Laser ablation of peripheral root cap cells abolished the ability of roots to respond to touch stimulation (Massa and Gilroy, 2003), but the exact mechanism of the touch-response pathway remains unclear. The tilted hard agar plate bioassay presents the growing root tip with a combination of gravity and mechanical stimuli and may be influenced by other factors as well, and the exact mechanisms by which these signals are integrated is a matter of some debate (Buer et al., 2000, 2003; Okada and Shimura, 1990; Thompson and Holbrook, 2004). Under properly controlled conditions, the Arabidopsis root-waving growth response in this bioassay is very robust. Several mutants with aberrant root-waving growth phenotypes are affected in MAPs or tubulin subunits (Abe et al., 2004; Buschmann et al., 2004; Furutani et al., 2000; Sedbrook et al., 2004; Shoji et al., 2004; Thitamadee et al., 2002). We observed that cortical microtubule organization is markedly affected in wvd2-1 root cap cells. As wvd2-1 shows no major defects in gravitropism (Yuen et al., 2003, and data not shown), altered microtubule organization in peripheral cap cells may modify the root's ability to respond properly to mechanical signals experienced in the hard agar bioassay, explaining the dampened-wave phenotype. Other tissues in plants are capable of responding to touch stimuli, for example during thigmomorphogenesis, a developmental re-programming in which plant shoots adapt a short and thick stature in response to repeated touch stimuli (Braam, 2005). To date, few candidates for global regulators of mechano-transduction in plants have been reported. However, WVD2 may be involved in these responses, as we have also observed alterations in the thigmomorphogenesis response in WVD2-overexpressing plants (Figure S1).

In conclusion, WVD2 and WDL1 are members of a novel plant-specific family of MAPs that are localized to cortical microtubules and whose expression level dictates morphology and helical handedness at both the organ and cellular level. Overexpression of WVD2 affects microtubule organization in the peripheral root cap cells that are critical for responses to mechano-stimulation. WVD2 is capable of binding to and bundling microtubules in vitro, suggesting a role in regulating the stability or organization of the cytoskeleton. Further study of this protein family is needed, particularly with regard to the roles of family members in the context of environmental stimuli and within regulatory pathways.

Experimental procedures

Identification of wvd2-2

A wvd2-1 line, back-crossed five times to Ler, was mutagenized using ethyl methane sulfonate (EMS) as described by Chourey and Schwartz (1971), except the EMS concentration was decreased to 0.4%. Mutagenized seeds were sown directly onto sterilized soil using 0.15% agar solution and grown at 22°C, 70% relative humidity, 16 h light (60 μmol ml2 s-1)/8 h darkness. M1 plants that showed restoration of wild-type height were used for preliminary screening. The WVD2 locus was sequenced using primers AC-CY1 (5′-CTAGCTCTACCGTTTCCGTTTCCGTTTAC), wvd2(EII; 5′-GGAAGCTTTTTGGGTGACCTCATTCTACCACAC), seqWVD.R21 (5′-CTCATTCCAACTTACTGTCATATTTCACCA), seqWVD2.F21 (5′-TTGGTGAAATATGACAGTAAGTTGGAATGA), seqWVD2.R22 (5′-GAATTCTCCTCAGTGCATTCTTTCTCCTCA) and seqWVD2.R23 (5′-ACTAATACCACAAAAAGACTTCAAAATCAG). Line WE9-11.1 bearing a missense mutation (C to T) that changes R152 to a stop codon was named wvd2-2. This line was back-crossed to Ler three times before phenotypic analysis was performed. Root wave assays were performed as described previously (Yuen et al., 2003).

Generation of polyclonal antibodies to WVD2

cDNA encoding the open reading frame of WVD2 (Yuen et al., 2003) was cloned into pGEX-4T-1, resulting in a construct encoding a GST–WVD2 recombinant protein. The construct was verified by sequencing. GST–WVD2 was expressed in the Origami(DE3)pLysSE. coli strain (Novagen, http://www.novagen.com), isolated using BugBuster reagent (Novagen), and purified using glutathione–Sepharose (Amersham Biosciences, http://www.amershambiosciences.com), eluting in buffer containing 50 mm, Tris pH 8.0, 10 mm reduced glutathione and protease inhibitor cocktail (Calbiochem, http://www.calbiochem.com). Purified protein was desalted with Sephadex G-25 in PBS and concentrated using Centricon YM-3 filter units (Millipore, http://www.millipore.com). Purified protein was used for polyclonal antibody production in New Zealand White rabbits (Cocalico Biologicals, Inc., http://www.cocalicobiologicals.com).

Whole-mount immunoconfocal assays

Rosette leaves of 10-day-old plants were subjected to a freeze-shattering step as described by Wasteneys et al. (1997). Following freeze-shattering, samples were fixed under vacuum for 1 h in PEM buffer (50 mm PIPES-KOH, pH 6.9, 2 mm EGTA, 2 mm MgSO4) containing 4% para-formaldehyde. To facilitate transfer of solutions during sample processing, O-shaped pieces of Parafilm were melted onto Probe-On slides (Fisher Scientific, http://www.fishersci.com) to create a sample well in the middle of each slide. Samples were washed three times in PEM buffer with 0.1% Triton X-100 (PEMT), digested for 20 min in PEMT containing 0.5% macerozyme and 0.1% pectolyase (Karlan Research Products, http://www.karlan.com), washed three times in PEM, permeabilized for 45 min in PEM containing 0.5% Nonidet P-40 (NP-40) and 10% dimethylsulfoxide (DMSO), and blocked for at least 30 min in PEM plus 3% v/v BSA and 1% v/v goat serum. Samples were then incubated in a wet chamber for durations ranging from 3 h to overnight in primary antibody diluted in blocking buffer (anti-GST–WVD2 used at 1:1000 to 1:5000, anti-GST (Santa Cruz Biotechnology, http://www.scbt.com) used at 1:3000, rat anti-tubulin YOL1/34 (Serotec, http://www.serotec.com) used at 1:100, or pre-immune sera at a dilution matching anti-GST–WVD2 immune sera). Samples were washed three times in PEM buffer and incubated in secondary antibody [goat anti-rabbit conjugated to rhodamine (Jackson ImmunoResearch, http://www.jacksonimmuno.com) at 1:200, or goat anti-rat IgG (heavy + light chain), FITC-conjugated (Jackson ImmunoResearch) at 1:100] for 2 h in blocking buffer. Samples were washed five times with PEM buffer, mounted in Vectashield (Vector Laboratories, http://www.vectorlabs.com), and analyzed using an MRC-1024 confocal laser scanning microscope (Bio-Rad, http://www.bio-rad.com) at the W.M. Keck Center for Biological Imaging, University of Wisconsin-Madison.

Generation and analysis of GFP transgenic lines

To generate transgenic Arabidopsis lines expressing WVD2 fused to GFP, the stop codon of a genomic WVD2 cassette was engineered to encode Thr and cloned into pCAMBIA 1302 to result in a 35S::WVD2–GFP construct. Wild-type ecotypes (Col-0, Ler and No-O) were transformed with this construct using the floral dip method (Clough and Bent, 1998) and selected on hygromycin-containing media, and transformation was verified by phenotype and by PCR using primers specific for GFP (genoGFPF1, 5′-CAGTGGAGAGGGTGAAGGTGATGC-3′; genoGFPR1, 5′-TTTCGAAAGGGCAGATTGTGTGGAC-3′). Constructs for transgenic plants expressing WDL1–GFP or GFP–WDL were generated by PCR amplification of the cDNA sequence of WDL1 from the CD4-34 library (ABRC Stock Center, Ohio State University, OH, USA). The WDL1 open reading frame was then amplified using KOD Hot Start DNA polymerase (Novagen) with primers WDL1+Start (5′-CACCATGGGAAGAGAAGTTGTTGAG-3′) and WDL1+Stop (5′-TCAAGCTTCTTCTGAGGACTC-3′) for inclusion of the stop codon, or WDL1+Start and WDL1–Stop (5′-AGCTTCTTCTGAGGACTCGT-3′) for product without the stop codon. The PCR fragments were cloned into the Gateway entry vector pENTR/D-TOPO using the pENTR directional TOPO cloning kit (Invitrogen, http://www.invitrogen.com/). The WDL1 coding region with or without the stop codon was then recombined using the Gateway LR Clonase Enzyme Mix (Invitrogen) to the Gateway destination vectors pK7WGF2 or pK7FWG2 (Karimi et al., 2002) to create an N-terminal or C-terminal GFP-tagged WDL1 overexpression construct, respectively (pK7WGF2/WDL1 or pK7FWGF2/WDL1). The constructs were used to transform Col-0 plants as described above. Positive transformants were selected by plating out and germinating the T1 seeds on the media described above supplemented with 100 μg ml−1 carbenicillin, 50 μg ml−1 vancomycin and 50 μg ml−1 kanamycin. Transgenic lines were evaluated for the level and segregation of the wvd2-1-like phenotype and GFP signal. To generate tubulin reporter lines in a WVD2 overexpression background, a 35S::WVD2 transgenic line in the Col-0 ecotype showing strong wvd2-1-like phenotype was crossed to eGFP:TUB1-7 in the Col-0 ecotype background provided by Dr David Ehrhardt (Stanford University, CA, USA). Segregation analysis was performed, and F2 lines homozygous for both loci were used for further characterization. To generate transgenic plants expressing WVD2 under the control of the CaMV 35S promoter, Col-0 plants were transformed with a construct bearing genomic WVD2 in vector pCAMBIA1302. Transgenic lines were screened on the basis of a wvd2-1-like phenotype, and lines showing particularly compact rosette and inflorescence stem phenotypes (Figure S1 and data not shown) were assessed at the molecular level and used for further experiments.

In vitro microtubule co-sedimentation and bundling assays

Either recombinant WVD2:c-myc:His6 protein expressed in E. coli using the pET28 vector and purified using His-Bind affinity resin (Novagen) or recombinant GST–WVD2 was used for microtubule co-sedimentation assays (Cytoskeleton Inc., Denver, CO, USA) Pellet and supernatant samples were subjected to electrophoresis on SDS–PAGE and stained with Coomassie brilliant blue. Recombinant WVD2 protein was detected by immunoblotting, and signal was quantified by densitometry using NIH ImageJ software (NIH, http://rsb.info.nih.gov/nih-image). For microtubule bundling assays, microtubules were assembled by incubating tubulin and rhodamine-labelled tubulin (2.6 μg μgl-1 or 0.65 μg μl-1, respectively, Cytoskeleton Inc., http://ww.cytoskeleton.com) in PEM-G buffer (80 mM PIPES, pH 6.8, 1mM, MgCl2, 1mM, EGTA, 1 mM, GTP) with 10% glycerol in a total reaction volume of 5 μl at 37°C for 10 mins. Following this, 5 μl of GST–WVD2 (5 μm final concentration), GST (5 μm final concentration) or PEM-G was added, and incubation was continued at 37°C for an additional 10 min. Microtubules were fixed using 1% glutaraldehyde in PEM buffer, diluted with 80% glycerol, and 4–5 μl were analyzed by confocal microscopy. Assays were also performed as above with purified recombinant maltose binding protein–WVD2 protein expressed in E. coli Bl21 (DE3) pLysS stain using the pMAL-c2X vector (New England Biolabs, http://www.neb.com). Similar assays conducted using bovine tubulin without rhodamine labeling were analyzed by negative staining and imaged using transmission electron microscopy as described previously (Chan et al., 1999).

Thigmomorphogenesis experiments

Seeds of wild-type (Col-0) or a WVD2-overexpressing transgenic line (35S::WVD2) were suspended in 0.1% agar and sown directly on sterilized potting soil. Pot positions were randomized with regard to genotype, and plants were grown under 8 h darkness/16 h light (60 μmol ml2 s-1), 22°C, 70% relative humidity. Starting at the age of 2 weeks, rosette leaves of touch-treated plants were gently patted by hand every 5 sec for a total of 30 sec. The force applied to the leaves was approximately 0.025 N per touch. This treatment was applied twice a day until maturity (8–10 weeks). Inflorescence stems were harvested and measured.


The authors wish to thank Lance Rodenkirch and the W.M. Keck Laboratory for Biological Imaging at the University of Wisconsin-Madison for technical discussion. Funding was provided by University of Wisconsin College of Agricultural and Life sciences United States Department of Agriculture Hatch grant (grant number WIS04784 to P.H.M.), National Institutes of Health Research Service Award (NIH-NRSA) (grant number F32 GM069184 to R.M.P.) and National Aeronautics and space Administration (NASA) (grant number NNA04CC71G to P.H.M.). This is manuscript number 3632 from the University of Wisconsin Laboratory of Genetics.