The Microtubule-Associated Protein IQ67 DOMAIN5 Modulates Microtubule Dynamics and Pavement Cell Shape

The microtubule-associated protein IQD5 binds directly to microtubules to promote their stability, which affects subsequent cell shape formation in Arabidopsis.

Abstract

The dynamic arrangement of cortical microtubules (MTs) plays a pivotal role in controlling cell growth and shape formation in plants, but the mechanisms by which cortical MTs are organized to regulate these processes are not well characterized. In particular, the dynamic behavior of cortical MTs is critical for their spatial organization, yet the molecular mechanisms controlling MT dynamics remain poorly understood. In this study, we used the puzzle piece-shaped pavement cells of Arabidopsis (Arabidopsis thaliana) leaves as a model system in which to study cortical MT organization. We isolated an ethyl methanesulfonate mutant with reduced interdigitation of pavement cells in cotyledons. This line carried a mutation in IQ67 DOMAIN5 (IQD5), which encodes a member of the plant-specific IQ motif protein family. Live-cell imaging and biochemical analyses demonstrated that IQD5 binds to MTs and promotes MT assembly. MT-depolymerizing drug treatment and in vivo MT dynamics assays suggested that IQD5 functions to stabilize MTs. Hence, our findings provide genetic, cell biological, and biochemical evidence that IQD5 regulates MT dynamics that affect MT organization and subsequent cell shape formation.

Cell shape formation is critical for plant development and morphogenesis. The particular cell shape is not only important for its function but also offers advantages during environmental adaptation (Malinowski, 2013). Thus, there has been interest and progress in investigating the mechanisms underlying the formation of plant cell shapes (Yang, 2008; Qian et al., 2009; Mirabet et al., 2011; Ivakov and Persson, 2013; Qin and Dong, 2015). Cell walls play important roles in controlling cell morphogenesis. Due to large intracellular turgor pressure, the cell shape in plants is defined by the cell wall via differential cell wall expansion, which is determined by both cell wall loosening and its anisotropic growth. Both of these processes are linked to the spatial organization and dynamics of cortical microtubule (MT) arrays (Lloyd and Chan, 2004; Fu et al., 2005; Yang, 2008; Xiao and Anderson, 2016). Cortical MTs may influence cell wall loosening via regulation of the distribution of cell wall-loosening factors (Fu et al., 2005; Mathur, 2006; Guimil and Dunand, 2007). Cortical MTs also play important roles in controlling anisotropic growth by spatially modulating the deposition of cellulose microfibrils (Wasteneys and Galway, 2003; Lloyd and Chan, 2004; Ivakov and Persson, 2013; Hamada, 2014; Armour et al., 2015; Hashimoto, 2015; Kong et al., 2015). Since MTs guide microfibril deposition, plant cells expand preferentially in the direction perpendicular to the well-ordered parallel cortical MT arrays, leading to anisotropic expansion (Fu et al., 2005, 2009; Hamada, 2014; Hashimoto, 2015). In addition, the mechanical heterogeneities of the anticlinal cell walls between adjacent cells also contribute to plant cell shape formation (Majda et al., 2017; von Wangenheim et al., 2017; Belteton et al., 2018).

How cortical MTs form parallel arrays perpendicular to the preferred growth axis is an important question that has been pursued but not yet fully understood. The cortical MTs in plants exhibit a unique dynamic behavior known as sustained treadmilling (Shaw et al., 2003; Dixit and Cyr, 2004a). In addition, intermicrotubule encounters have variable outcomes (Shaw et al., 2003; Dixit and Cyr, 2004b). The shallow-angle encounters are associated with MT zippering, which promotes MT coalignment and bundling, whereas the steep-angle encounters are associated with catastrophic collisions or MT crossover, which promote MT destabilization (Shaw et al., 2003; Dixit and Cyr, 2004b). Modeling shows that these two behaviors work together, allowing the self-organizing formation of parallel MTs (Eren et al., 2012; Deinum and Mulder, 2013). However, nascent cortical MTs are formed predominantly by branch nucleation from an existing cortical MT at an angle of roughly 40° (Murata et al., 2005; Chan et al., 2009; Nakamura et al., 2010). How can branched cortical MTs become ordered in parallel? The plant-specific RIC1 MT-associated protein was implicated in recruiting the MT-severing protein katanin to the branch point to release branched MTs (Lin et al., 2013). Katanin severs MTs at crossovers, thus explaining how katanin promotes well-ordered cortical MT arrays (Lin et al., 2013; Lindeboom et al., 2013; Zhang et al., 2013). The new plus ends created by katanin severing provides a mechanism for MT amplification, driving array reorientation in response to signals such as blue light (Lindeboom et al., 2013).

The self-organizing ordering, katanin-mediated branch and crossover severing, and MT amplification can explain the formation of well-ordered cortical MTs, but the mechanisms for this organization are more complicated, as several other microtubule-associated proteins (MAPs) are known to be required for MT ordering. For example, MOR1, a member of the MAP215/DIS1 family that has a role in controlling MT length by activating the dynamic instability, is critical in the organization of cortical MTs (Whittington et al., 2001; Twell et al., 2002). Cytoplasmic linker-associated protein1 (CLASP), an MT plus end-binding protein, was shown to counter cell edge-induced MT depolymerization and is involved in the organization of cortical MTs in cells with shape edges (Ambrose et al., 2007, 2011; Kirik et al., 2007; Ambrose and Wasteneys, 2008). In addition, the models described above are based on intra- and inter-MT dynamic behaviors, but the mechanisms underlying the regulation of these dynamic behaviors are poorly understood.

A plant-specific IQ67-domain (IQD) protein family was implicated recently as a potential family of MAPs important in plant development and responses to the environment. These proteins are recognized by the presence of a central region of 67 conserved amino acid residues composed of IQ motifs (Abel et al., 2005). The IQ motif is responsible for recruiting calmodulin, a Ca²⁺ sensor that regulates the activities of numerous downstream proteins (Abel et al., 2013). IQDs have been implicated in the regulation of the basal defense response in plants (Levy et al., 2005), developmental processes (Zentella et al., 2007), and cell shape formation (Xiao et al., 2008; Wu et al., 2011; Bürstenbinder et al., 2017). When overexpressed in Arabidopsis (Arabidopsis thaliana) transgenic plants, some GFP-IQDs colocalized with MTs and induced changes in pavement cell shape consistent with the altered organization of cortical MTs (Bürstenbinder et al., 2017). These observations support the association of IQDs with MTs and their roles in regulating MT organization. However, the biological functions of these IQDs are still largely unknown, and their roles in MT organization remain uncharacterized.

Arabidopsis leaf pavement cells, with their interlocking, puzzle-piece patterns, are a useful model system in which to investigate complex cell shape formation (Fu et al., 2005; Jacques et al., 2014; Sampathkumar et al., 2014; Majda et al., 2017). Understanding the mechanisms regulating leaf pavement cell morphogenesis helps to decipher general rules governing cell shape development (Majda et al., 2017; Belteton et al., 2018; Sapala et al., 2018). Pavement cell shape is affected by disrupting MT polymerization and depolymerization (Akita et al., 2015) or MAP functions such as CLASP (Ambrose et al., 2007, 2011; Kirik et al., 2007; Ambrose and Wasteneys, 2008), Katanin (Burk et al., 2001; Lin et al., 2013; Zhang et al., 2013), or MAP18 (Wang et al., 2007). In addition, the coordination between ordered cortical MT arrays associated with the indentations and cortical actin microfilaments localized to the lobes is important for pavement cell morphogenesis (Fu et al., 2005; Panteris and Galatis, 2005).

To further elucidate the mechanisms for the formation of pavement cell shape, we isolated Arabidopsis mutants with altered interdigitation patterns in cotyledon pavement cells and identified IQD5. The cortical MT arrays in iqd5 mutants were organized aberrantly and were more sensitive to the MT-depolymerizing drug oryzalin. IQD5 bound to MTs in vivo and in vitro and promoted MT assembly by enhancing its stability. Our findings demonstrate that IQD5 is a MAP and it regulates MT dynamics that affect MT organization and subsequent cell shape formation.

RESULTS

iqd5 Mutants Exhibit Abnormal Pavement Cell Shape

To explore additional components or mechanisms that control pavement cell morphogenesis, a genetic screen was conducted for mutants with abnormal pavement cell shapes from an ethyl methanesulfonate (EMS)-induced mutagenesis. The mutant line bQ18E, with abnormal pavement cell shape that lacked interdigitation of lobes, was isolated from M2 seedlings (Fig. 1A). We further found that 100 of 438 F2 seedlings (approximately 22.8%) of bQ18E that had been backcrossed with wild-type Col-0 plants showed the pavement cell phenotype. Genetic analysis revealed a single recessive mutation responsible for the observed phenotype. We identified the gene by next-generation sequencing of bulked segregants (Zhu et al., 2012) and found that the bQ18E mutation was a G-to-A change leading to a premature stop in the coding region of IQD5 (At3g22190). Two T-DNA insertional alleles that disrupt this locus, iqd5-1 (Salk_015580) and iqd5-2 (Salk_098610), also had abnormal pavement cell morphology (Fig. 1A), and iqd5-1 eliminated the IQD5 transcript (Supplemental Fig. S1C). To confirm that bQ18E is allelic to iqd5-1, we crossed iqd5-1 with bQ18E. bQ18E failed to complement iqd5-1 because the F1 progeny exhibited the pavement cell phenotype.

Figure 1.

Pavement cell shape phenotype in the iqd5 mutants. A, Comparison of cotyledon pavement cell shape using confocal images of wild-type ecotype Columbia (Col-0; wt), EMS mutant bQ18E, iqd5-1, and iqd5-2. Bars = 20 µm. B and C, Quantitative analysis of pavement cell shape changes for wild-type Col-0, EMS mutant bQ18E, iqd5-1, and iqd5-2. Lobe length and neck width were measured as described previously (Lin et al., 2013) with some modifications. For lobe length, only the true lobes were assessed and the tricellular corners were avoided. For neck width, only the distance between two indents located at opposing sides of the cell was measured. The differences in lobe lengths (B) and neck widths (C) between the wild type and iqd5 mutants (bQ18E, iqd5-1, and iqd5-2) were significant (Mann-Whitney test; **, P < 0.01). No significant differences were detected among iqd5 mutants (bQ18E, iqd5-1, and iqd5-2). Data are means ± SD; n = 3 replicates (100–200 cells from five to six seedlings were checked in each replicate).

We quantitatively analyzed lobe lengths and indentation widths using cotyledons from 3-d-old seedlings. iqd5 mutants showed reduced lobe lengths (Fig. 1B) and increased neck widths (Fig. 1C), which were significantly different from wild-type Col-0, whereas no significant differences among bQ18E, iqd5-1, and iqd5-2 were detected. To further compare the cell shape differences, we measured the cell area, perimeter, and circularity. Circularity is a dimensionless shape factor; the circularity value decreases when the complexity of the shape increases (Zhang et al., 2011). The differences in cell perimeter, cell area, and circularity between the wild type and iqd5 mutants (bQ18E, iqd5-1, and iqd5-2) were significant (P < 0.0001), but no significant differences were detected among iqd5 mutants (bQ18E, iqd5-1, and iqd5-2; Table 1). Taken together, these results indicate that IQD5 influences pavement cell morphogenesis in Arabidopsis.

Table 1. Pavement cell size of wild-type Col-0 and iqd5 mutants.

The parameters cell area, perimeter, and circularity were measured from 3-d-old seedlings by ImageJ software. Data are means ± SD. Asterisks indicate statistically significant differences (Mann-Whitney test; **, P < 0.0001); n = 3 replicates (100–200 cells from five to six seedlings were measured in each replicate).

Plant	Area	Perimeter	Circularity
	µm2	µm
Col-0	2,997 ± 1,019	349 ± 98	0.34 ± 0.07
bQ18E	2,717 ± 1,110**	230 ± 48**	0.60 ± 0.09**
iqd5-1	2,716 ± 957**	232 ± 48**	0.60 ± 0.09**
iqd5-2	2,688 ± 966**	237 ± 52**	0.58 ± 0.09**

IQD5 Localized to Cortical MTs in Dynamic Pattern and Labeled All MT Arrays

To determine the subcellular distribution of IQD5 and to monitor its dynamics in vivo, we generated functional GFP and mCherry fluorescent protein fusion constructs using the entire IQD5 genomic sequence. GFP or mCherry was fused to the N terminus of IQD5 under the control of its endogenous regulatory elements. These constructs fully rescued the pavement cell defects in the iqd5-1 mutant, suggesting that the fusion proteins were functional (Fig. 2A). GFP-IQD5 colocalized with the MT marker mCherry-TUB6 (mCherry fused to β-tubulin6) in both leaf pavement cells (Fig. 2, B and C) and hypocotyl epidermal cells (Supplemental Fig. S2A). GFP-IQD5 also localized to preprophase bands, mitotic spindles, and phragmoplasts of mitotic cells in root tips (Fig. 2D). Together, these data show that IQD5 colocalizes with MTs in vivo.

Figure 2.

Genetic complementation of iqd5-1 and IQD5 localization. A, The genomic GFP-IQD5 and mCherry-IQD5 fusion complemented the pavement cell defects of the iqd5-1 mutant. Bar = 75 µm. B and C, GFP-labeled IQD5 colocalized with cortical MTs (mCherry-TUB6) in Arabidopsis pavement cells. GFP-IQD5, mCherry-TUB6, and merged images are shown. Bars = 5 µm (B) and 10 µm (C). D, GFP-IQD5 is associated with preprophase band, spindle, and phragmoplast MT arrays in mitotic cells of Arabidopsis roots. Bars = 5 µm.

mCherry/GFP-IQD5 labeled the outline of pavement cells in a discontinuous and punctate manner, accumulating in the indentation regions (Supplemental Fig. S2, B and C; Supplemental Movie S1). To determine whether IQD5 localization depended on intact MT arrays, mCherry-IQD5 seedlings were treated with the MT-depolymerizing drug oryzalin. We found that mCherry-IQD5 in both MTs and the cell cortex was abolished . Indeed, mCherry-IQD5 and mCherry-TUB6 cell cortex fluorescence was equally strongly affected after 2 h of oryzalin treatment (Supplemental Fig. S3), suggesting that IQD5 localization on the cell cortex (Supplemental Movie S2) was dependent on the presence of cortical MTs. These localization and pharmacological data indicated that IQD5 localizes to cortical MTs and acts alongside MTs to label the cell cortex.

The localization pattern of GFP-IQD5 suggests that it labels the entire length of cortical MTs evenly. To further test this, we investigated the IQD5 dynamic behavior by time-lapse imaging. Dynamic analysis revealed that IQD5 also was highly dynamic (Fig. 3A; Supplemental Movie S3); the growth and shrinkage rates of IQD5 were 5.48 ± 1.12 and 12.76 ± 2.63 µm min⁻¹, respectively (Fig. 3B), showing no significant difference from the behavior of cortical MTs themselves (Table 2). Kymograph analysis confirmed that IQD5 bound to both the growing and shrinking MT ends (Fig. 3C). Therefore, IQD5 binds to the MT lattice and decorates all MT arrays.

Figure 3.

Dynamicity of IQD5. A, Individual frames from time-lapse confocal images showing a single GFP-IQD5-labeled structure exhibiting dynamic growth and shrinkage behavior, similar to MT catastrophe and rescue. Numbers indicate time in seconds. Bar = 2 µm. B, IQD5 growth and shrinkage rates. The histogram depicts the average growth and shrinkage velocities, respectively. Data are means ± sd (n represents the number of observed growth or shrinkage events of GFP-IQD5; n ≥ 50 from a total of 20 to 30 cells from five seedlings). C, Kymograph of a 295-s recording of a GFP-IQD5 fluorescent track showing consecutive polymerization and depolymerization events within the IQD5 trajectory. The solid lines indicate polymerization, and the dashed lines represent depolymerization. The red arrow indicates a pause. Bar = 2 µm.

Table 2. Cortical MT dynamics.

MT parameters were measured from time-lapse images of 3-d-old mCherry-TUB6, Col-0 and mCherry-TUB6, iqd5-1 cotyledons. Values for growth rates, shrinkage rates, and dynamicity are means ± sd. For catastrophe and rescue frequencies, the values with SD were calculated by averaging the inverse time needed for each individual switch and the values without SD were obtained from the inverse of the mean time spent in growth and shrinkage, respectively, as described previously (Dhonukshe and Gadella, 2003; Abe and Hashimoto, 2005). Asterisks indicate statistically significant differences (for each measurement listed, we used 10–20 seedlings per genotype to observe 20–50 cells from which we assessed n ≥ 50 MT dynamic properties: Student’s t test, two tailed, type 3; *, P < 0.05 and **, P < 0.01).

Dynamic Parameters	mCherry-TUB6, Col-0	mCherry-TUB6, iqd5-1
Growth rate (µm min−1)	5.39 ± 1.13	4.34 ± 0.85**
Shrinkage rate (µm min−1)	12.87 ± 2.69	14.40 ± 2.92*
Dynamicity (µm min−1)	6.12 ± 1.25	6.92 ± 1.71*
Catastrophe frequency (events s−1)	0.0192	0.0198
	0.020 ± 0.005	0.022 ± 0.008
Rescue frequency (events s−1)	0.065	0.067
	0.071 ± 0.021	0.073 ± 0.020

iqd5-1 Is Hypersensitive to the MT-Disrupting Drug Oryzalin

The iqd5-1 mutants exhibited a pavement cell phenotype consistent with impaired MT dynamics; therefore, we treated iqd5-1 with the MT-depolymerizing drug oryzalin. Wild-type Col-0, iqd5-1, and iqd5-1 rescued lines were grown on medium containing oryzalin (300 nM). After 7 d, roots of the light-grown iqd5-1 seedlings were significantly shorter than those of the wild type, but there were no significant differences between the wild type and iqd5-1 GFP-IQD5 rescued lines (Supplemental Fig. S4A). The average length of dark-grown iqd5-1 hypocotyls also was significantly shorter than that of the wild type upon oryzalin treatment (Supplemental Fig. S4B). These results indicated that the iqd5-1 mutants were hypersensitive to oryzalin treatment.

Next, we observed the MT array response to oryzalin treatment in iqd5-1 mutants. We treated the 3-d-old light-grown seedlings of mCherry-TUB6, Col-0 and mCherry-TUB6, iqd5-1 with 10 µM oryzalin for 30 and 60 min. As shown in Figure 4, the cortical MT arrays in the cotyledon were disrupted faster in iqd5-1 pavement cells than in wild-type cells. The 60-min oryzalin treatment removed almost all MTs in iqd5-1, whereas in mCherry-TUB6, Col-0 plants, incomplete depolymerization of the labeled MTs was observed (Fig. 4, A and B). We counted the number of cortical MTs or MT bundles in pavement cells before and after oryzalin treatment. More cortical MTs were disrupted in the iqd5-1 mutant than in the wild type after oryzalin treatment (Fig. 4C). Together, these results suggested that MTs were more prone to depolymerization in iqd5-1 seedlings treated with oryzalin.

Figure 4.

Cortical MTs are more sensitive to oryzalin treatment in iqd5-1 than in the wild type. A and B, Cortical MTs in pavement cells of mCherry-TUB6, Col-0 (A) and mCherry-TUB6, iqd5-1 (B) treated with oryzalin (10 µm) for 0, 30, and 60 min. Maximum projections of z series are presented. Bars = 10 µm. C, Number of MTs and MT bundles in the pavement cells of mCherry-TUB6, Col-0 and mCherry-TUB6, iqd5-1 treated with oryzalin (10 µm) for 0, 30, and 60 min. A line of fixed length (10 µm) was drawn, and the number of cortical MTs across the line was counted using ImageJ software. Asterisks indicate statistically significant differences from the wild type (Student’s t test, two tailed, type 3; *, P < 0.05 and **, P < 0.01). Error bars are SD; n = 3 seedlings (at least 10 cells were used in each replicate, and five fixed-length lines were drawn from each cell).

Disruption of IQD5 Influences MT Organization and Dynamics

The hypersensitivity of iqd5-1 to oryzalin treatment led us to explore whether MT organization or stability also was affected in iqd5-1. First, we compared the MTs of Col-0 (wild type) and iqd5-1 plants. Live-cell imaging showed that cortical MTs in pavement cells of iqd5-1 were randomly oriented and fragmented compared with the ordered distribution in wild-type cells (Fig. 5, A and B). In the indentation regions of wild-type pavement cells, cortical MTs are typically aligned perpendicular to the indentation. To compare MT organization in the indented region, we defined the axis as the direction in which most of the MTs aligned parallel to it in this area, and then the angles of MTs were calculated to this axis (Fig. 5C). Quantification of MT orientation relative to the defined axis showed that iqd5-1 had fewer well-aligned MTs compared with the wild-type MT arrays. We also measured the anisotropy score in the indented region of wild-type and iqd5-1 mutants by FibrilTool (Boudaoud et al., 2014; Fig. 5D). Statistical analysis showed that the anisotropy score in the wild type was significantly higher than that in iqd5-1, indicating that the MT arrays in iqd5-1 were less organized. The MT arrays in hypocotyl epidermal cells also were less ordered in iqd5-1 compared with the wild type, leading to a decrease in the anisotropy score of iqd5-1 (Supplemental Fig. S5, A and B). Together, these data suggest that IQD5 promotes proper MT array organization in Arabidopsis.

Figure 5.

Comparison of MT arrays between wild-type Col-0 (wt) and iqd5-1 mutants. A, Cortical MTs in pavement cells of mCherry-TUB6, Col-0 and mCherry-TUB6, iqd5-1 cotyledons. Maximum projections of two images are presented. Bars = 10 µm. B, Cortical MTs in pavement cells of mCherry-TUB6, Col-0 and mCherry-TUB6, iqd5-1 cotyledons. Bars = 5 µm. C, Quantitative analysis of MT orientation. Data are displayed as frequency distribution histograms (Mann-Whitney test; P < 0.0001). Error bars are sd; n = 5 seedlings (at least 80 MTs from two to three cells were analyzed for each replicate). D, Quantitative analysis of MT array anisotropy by the ImageJ FibrilTool plugin. Data are means ± sd; n = 5 seedlings. At least 10 regions of interest (surface area,∼50 µm²; perimeter, ∼28 µm) from three to four cells were analyzed for each replicate. Asterisks indicate statistically significant differences from the wild-type distribution (Student’s t test, two tailed, type 3; **, P < 0.01).

MT dynamic behavior usually contributes to changes in MT array organization. The altered MT dynamics might influence the overall array organization and could lead to changes in anisotropic growth mediated by a change in the deposition of cellulose microfibrils (Dixit and Cyr, 2004a; Elliott and Shaw, 2018). To determine whether the MT dynamic behavior changed in iqd5-1 mutants, the individual MT dynamics in vivo were monitored by time-lapse imaging, and the MT dynamic parameters were compared between wild-type Col-0 and iqd5-1 mutant pavement cells in cotyledons (Table 2). The MT growth rate decreased while both the shrinkage rate and the dynamicity increased in iqd5-1, indicating that MTs were destabilized in iqd5-1 pavement cells. The catastrophe and rescue frequency were not significantly different between wild-type Col-0 and the iqd5-1 mutant (Table 2). Thus, our data indicate that IQD5 affects the stability and organization of cortical MT arrays in cotyledon pavement cells. We propose that diminished MT stability influenced the organization of cortical MT arrays; the less organized cortical MT arrays caused the abnormal cell shape in iqd5-1 mutants.

IQD5 Binds to MTs and Promotes MT Assembly in Vitro

To test whether IQD5 binds directly to MTs, we conducted MT cosedimentation assays using the full-length IQD5 fused with the 6×His tag (Supplemental Fig. S6C). When MTs were present, His-IQD5 was observed in the pellet fraction. In the absence of MTs, His-IQD5 remained in the supernatant after ultracentrifugation (Fig. 6A). These results indicated that His-IQD5 bound MTs directly. The molecular mass of the recombinant His-IQD5 (47.8 kD) is close to that of tubulin (55 kD); therefore, the His-IQD5 band was masked by the tubulin band on Coomassie Blue-stained SDS-PAGE gels. Subsequently, western blotting with anti-His antibodies was used to detect the His-IQD5 protein in supernatant and pellet fractions. To determine the stoichiometry and affinity of His-IQD5 and MTs, various amounts of MTs were ultracentrifuged with a constant amount of His-IQD5 protein (5 µg; Fig. 6B), which showed that His-IQD5 bound MTs in a concentration-dependent manner. When 10 µL of MTs (8 µm) was added, no His-IQD5 was detected from the supernatant fraction (Fig. 6B). The binding was saturated at a stoichiometry of approximately 1.23 m His-IQD5 per mol of tubulin (Supplemental Fig. S6A).

Figure 6.

Recombinant IQD5 binds MTs in vitro. A, His-IQD5 fusion protein (5 µg) was cosedimented with paclitaxel-stabilized MTs (8 µm). His-IQD5 was detected only in the pellet when MTs were present. Without MTs, His-IQD5 was detected only in the supernatant fraction. MAPF (MAP fraction; 280 kD) was used as a positive control; Tub, tubulin (55 kD); BSA (68 kD) was used as a negative control. P, Pellet fraction; S, supernatant fraction. Staining, Samples were analyzed by SDS-PAGE and Coomassie Brilliant Blue staining; Western blot, samples were detected by western blot using α-His antibodies. His-IQD5 protein (47.8 kD) is indicated by the arrowheads. The experiments were repeated three times with similar results. B, IQD5 bound MTs in a concentration-dependent manner. Five micrograms of His-IQD5 was incubated with different volumes of MTs (8 µm). After ultracentrifugation, the amount of His-IQD5 in the pellet fraction increased with the increased amount of added MTs before reaching saturation. The experiments were repeated three times with similar results. C, Charge plots of IQD5. The basic amino acid (AA)-enriched region (amino acids 86–152) is indicated by the bracket.

IQD5 encodes a polypeptide with 422 amino acids containing the typical IQ motif (amino acids 87–138) and a domain of unknown function (DUF) in the C-terminal region (DUF4005). To investigate whether the DUF had MT-binding activity, we expressed a truncated version of IQD5, namely His-IQD5-F3 (amino acids 300–422; Supplemental Fig. S6D), and tested the MT-binding activity by MT cosedimentation assay. His-IQD5-F3 cosedimented with MTs, although the proportion of His-IQD5-F3 in the pellet fraction was lower than that of the full-length IQD5 (Supplemental Fig. S6B). His-IQD5 was detected only in the supernatant fraction in the absence of MTs; however, when MTs were present, some His-IQD5-F3 still was detected in the supernatant fraction (Supplemental Fig. S6B). The binding capacity of the truncated protein is much lower compared with the full-length IQD5 protein, indicating that His-IQD5-F3, to some extent, can bind to MTs in vitro. We further analyzed the amino acid distribution of IQD5 and found that, besides the C-terminal F3 region, the region between amino acids 86 and 152 covering the IQ motifs also was rich in basic amino acid residues, typical of MT-associated proteins (Fig. 6C).

Since IQD5 binds MTs directly and iqd5-1 exhibited decreased MT stability, we hypothesized that IQD5 binding stabilizes MTs. To test this hypothesis, we determined how IQD5 influenced MT polymerization in vitro. The MT in vitro polymerization assay was performed by incubating IQD5-His protein, rhodamine-labeled tubulin, and GTP at 37°C for 1 h. We anticipated that if IQD5 stabilized MTs, we would observe an IQD5-mediated increase in MT length. In agreement, MTs were longer in the samples coincubated with His-IQD5 (Fig. 7A). In the presence of the His-IQD5 protein, the assembled MT lengths were increased significantly compared with the His-GFP negative control (Fig. 7B). Together with the in vivo experiments, these data indicate that IQD5 stabilizes MTs.

Figure 7.

IQD5 promotes MT elongation in vitro. A, Rhodamine-labeled tubulins (10 µm) were incubated with His-IQD5 (1 µm) and His-GFP (1 µm) at 37°C for 30 min. MTs were visualized by confocal microscopy. Bars = 25 µm. B, Quantitative analysis of MT length. Data are displayed as frequency distribution histograms (Mann-Whitney test; P < 0.0001). Error bars are sd; n = 3 replicates (400–500 MTs were measured in each replicate).

DISCUSSION AND CONCLUSION

In this study, we demonstrate that IQD5 is a MAP required for pavement cell morphogenesis in Arabidopsis. IQD5 directly binds MTs and promotes MT stability both in vivo and in vitro. Consequently, IQD5 regulates pavement cell morphogenesis, probably via the stabilization of MT arrays. Mutations in tubulin subunits are thought to impact the organization of cortical MTs through their effects on MT stability (Abe and Hashimoto, 2005), and MOR1 also regulates MT stability (Whittington et al., 2001; Kawamura and Wasteneys, 2008). Here, our findings provide strong in vivo and in vitro evidence that a member of the large IQD gene family encodes a functional MAP that modulates MT dynamics and organization during pavement cell morphogenesis.

IQD5 Encodes a MAP Belonging to the Plant-Specific IQD Family

IQD5 belongs to the plant-specific IQD gene family, which was described initially in Arabidopsis and rice (Oryza sativa; Abel et al., 2005) and later was annotated systematically in six other plant species (Huang et al., 2013; Feng et al., 2014; Ma et al., 2014; Cai et al., 2016; Wu et al., 2016). A genome-wide assessment of the subcellular distribution patterns for all 33 Arabidopsis IQDs suggests that different clades of IQDs may have distinct subcellular localization patterns and cellular functions (Bürstenbinder et al., 2017). Indeed, IQD1 was shown to localize to the nucleus and regulate glucosinolate accumulation (Levy et al., 2005). A widespread role for IQDs in the regulation of MT organization or functions has been suggested by their association with kinesins or MTs, but their in vivo role has not been determined to date (Bürstenbinder et al., 2013, 2017). Our findings here, including the altered MT dynamics and organization in iqd5-1, IQD5 association with MTs in vivo, and IQD5 direct binding with MTs and impact on MT assembly in vitro, demonstrate that IQD5 is a MAP that regulates MT dynamics.

The high pI values resulting from basic amino acid residues were proposed to be responsible for the MT-binding activity in many MAPs (Hashimoto, 2015). Positively charged domains can interact with the acidic tails of α- and β-tubulins, which are rich in Glu and Asp residues, through electrostatic interaction (Mishima et al., 2007; Roll-Mecak, 2015). The pI value of IQD5 is 11.05. However, as IQD5 or other IQDs lack known MT-binding motifs, we reasoned that IQD5 binds MTs via a basic region-mediated MT-binding mode, as there is differential binding affinity between the full-length IQD5 and IQD5-F3 (Fig. 6; Supplemental Fig. S6B) and more than one basic amino acid-rich regions exists in IQD5 protein, as revealed by the amino acid distribution (Fig. 6C). Additionally, although IQD proteins in Arabidopsis are highly diverse in terms of size (amino acid residues 103–794) and predicted molecular mass (11.8–86.8 kD), they all share the common features of high pI (∼10.3) and high fractions of Ser and Arg/Lys residues (Abel et al., 2013). This hallmark feature of IQD proteins likely contributes to their colocalization with MTs through direct binding. Moreover, the MAPs GROWING PLUS-END TRACKING1 (GPT1) and GPT2 in Arabidopsis also showed very high pI values, and the binding affinity (around 2.5 mol per mol of tubulin dimer; Wong and Hashimoto, 2017) was quite similar to that of IQD5 (1.23 m His-IQD5 per mol of tubulin), suggesting that they might share a similar MT-binding mode. Based on our findings on IQD5 here, the molecular characteristics of all Arabidopsis IQDs, and their subcellular localization, we propose that IQDs define a large family of MAPs.

IQD5 Regulates MT Dynamicity and Organization

Our results indicate that IQD5 regulates MT dynamicity and promotes proper MT organization. We found that iqd5-1 mutants exhibited less ordered organization of cortical MTs in cotyledon pavement cells and hypocotyl epidermal cells. In vivo imaging and in vitro studies suggest that these changes in MT organization are the result of reduced MT stability. First, both the cell shape phenotype and cortical MTs in iqd5-1 were hypersensitive to the MT-disrupting drug oryzalin compared with those in wild-type plants. Second, in vivo imaging showed that iqd5-1 displayed increased MT dynamicity, indicating that IQD5 may stabilize MTs by reducing their dynamicity in vivo. Third, our in vitro MT polymerization assays suggested that IQD5 promotes MT elongation, consistent with its role in stabilizing MTs. Our live-cell imaging data showed that IQD5 labeled cortical MTs uniformly. The even distribution pattern is similar to that of MAP65-1/2, which cross-link the antiparallel MTs to facilitate MT bundle formation (Lucas et al., 2011). However, in vitro assays showed that IQD5 does not affect MT bundling but promoted MT elongation. Taken together, these findings indicate that IQD5 directly stabilizes MTs by reducing dynamicity, thereby regulating the ordering of cortical MTs and pavement cell morphogenesis in Arabidopsis. Nonetheless, we cannot rule out that IQD5 might additionally influence MT stability by regulating MT bundle formation through interacting with other proteins, such as MAP65 proteins, which also uniformly decorate MTs. The MT bundling is proposed to increase the cortical MT stability and allows the short cortical MTs to associate with one another to form a higher order structure (Dixit and Cyr, 2004a). The role for IQD5 in the regulation of MT dynamicity is likely to be general for all MT arrays in plant cells, as it is associated with all MT arrays (Fig. 2).

Arabidopsis MOR1, which belong to the MAP215 family, can enhance MT dynamics by promoting MT growth and shrinkage rates as well as catastrophe frequency to affect general MT organization (Kawamura and Wasteneys, 2008). Similarly, SPR2 localizes to all MT arrays and binds MTs directly in vitro; it can enhance MT dynamics by promoting both MT growth rate and catastrophe frequency (Yao et al., 2008). In contrast, IQD5 only inhibits MT dynamicity. As a result, the IQD5-regulated MT dynamicity is important for MT ordering. By binding uniformly to cortical MTs and stabilizing them, it is possible that IQD5 may be important for sustained treadmilling. Further studies should determine whether IQD5 indeed is important for sustained treadmilling, a hallmark of plant cortical MTs critical for MT ordering, whereas the mechanism underlying this unique dynamic behavior of plant cortical MTs remains unknown.

IQD5 Regulates Anisotropic Cell Growth and Shape Formation

Well-organized cortical MT arrays are essential for anisotropic growth and, consequently, shape formation for cells undergoing diffuse growth in plants. The disordering of cortical MTs resulted in isotropic growth and loss of indentation in puzzle piece-shaped pavement cells (Lloyd and Chan, 2004; Horio and Murata, 2014). Ectopic expression of several IQDs, including IQD11, IQD14, IQD16, and IQD25, altered pavement cell shape by either increasing or decreasing the ordering of cortical MTs in pavement cells, suggesting that these IQDs might have a role in the regulation of anisotropic growth and shape formation by modulating the ordering of cortical MTs (Bürstenbinder et al., 2017). However, the biological functions of IQDs in anisotropic growth and cell shape regulation had been enigmatic due to a lack of phenotypes in the iqd mutants studied.

Here, we used the mutant iqd5-1 to demonstrate conclusively that IQD5 is important for pavement cell morphogenesis. Although iqd5-1 exhibits a strong pavement cell phenotype, the whole adult plant grows normally and is almost indistinguishable from wild-type plants. It is possible that IQD5 might function redundantly with other IQD members or other proteins to influence cell shape formation in other developmental stages. The relatively large number of IQD genes also raises the question of whether individual IQD family members regulate distinct aspects of development. The phylogenic tree indicates that IQD5 is most closely related to IQD6, IQD7, IQD8, and IQD21 in Arabidopsis; they cluster in the same clade (Supplemental Fig. S7). We compared the expression patterns for all 33 IQD members in Arabidopsis through GENEVESTIGATOR (https://genevestigator.com/gv/; Supplemental Fig. S8) and found that the IQD5 expression level was relatively high in early developmental stages compared with other IQD members, especially the other four members (IQD6, IQD7, IQD8, and IQD21) in the same clade.

According to the phylogenetic tree (Supplemental Fig. S7), IQD5, IQD11, IQD14, IQD16, and IQD25 clustered in different clades. Among them, IQD25 not only labeled MTs but also labeled distinct and immobile plasma membrane subdomains (Bürstenbinder et al., 2017). Overexpressing IQD25 induced rounding and more compact leaves and cells, whereas overexpressing IQD11 or IQD16 induced significantly elongated epidermal pavement cells and altered the orientation of cortical MTs (Bürstenbinder et al., 2017). Overexpressing IQD14 induced organ twisting and no changes in leaf elongation (Bürstenbinder et al., 2017). These results indicated that IQD family members may interact with different proteins within individual plasma membrane subdomains and suggested specialized functions of individual IQDs in regulating the direction of cell elongation and, possibly, in cell polarity establishment (Bürstenbinder et al., 2017). Future studies on the roles of IQDs will prove illuminating. While this article was under review, three preprint articles on IQD proteins were posted (Mitra et al., 2018; Wendrich et al., 2018; Yang et al., 2018). In one, the same iqd5 T-DNA insertion line altered pavement cell phenotype, IQD5 localization, and IQD5 accumulation in the indentation region were reported, consistent with our findings (Mitra et al., 2018). The other two articles reported AtIQD15 to AtIQD18 (Wendrich et al., 2018) and OsIQD14 (Yang et al., 2018) involvement in integrating auxin and calcium signaling pathways to control cytoskeleton dynamics and cell shape in Arabidopsis and rice separately.

MATERIALS AND METHODS

Plant Materials and Growth Conditions

Arabidopsis (Arabidopsis thaliana) Col-0 was used as the genetic background in this study. The EMS mutant was a generous gift from Juan Dong (Waksman Institute of Microbiology). The iqd5-1 (SALK_015580) and iqd5-2 (SALK_098610C) mutants were obtained from the Arabidopsis Biological Resource Center. The T-DNA insertions were verified by PCR (Supplemental Fig. S1B), and sequencing was performed using the primers described in Supplemental Table S1. Arabidopsis seeds were surface sterilized and dispersed on one-half-strength Murashige and Skoog (1/2 MS) medium containing 1% (w/v) Suc and 0.8% (w/v) agar. Plated seeds were incubated for 3 d at 4°C in darkness to ensure synchronous germination, and then plants were grown under 16 h of light/8 h of dark in a 22°C growth room. For observation of the seedlings, the plates were positioned vertically.

Construction and Plant Transformation

The MultiSite Gateway system (Invitrogen) was used to simultaneously clone multiple DNA fragments to generate the pIQD5::GFP-IQD5 constructs. Phusion DNA polymerase (New England Biolabs) was used in PCR. The IQD5 promoter including the 1,963-bp upstream region from the start codon ATG was amplified from the genomic DNA with primers attB1_IQD_F and attB4_IQD_R. The GFP-encoding sequence was amplified with primers attB4r_GFP_F and attB3r_GFP_R. The IQD5 genomic fragment, including its coding region and 3′ untranslated region, was amplified from genomic DNA using the primers attB3_IQD_F and attB2_IQD_R. Then, the amplified fragments were cloned into separate entry vectors pDONR P1P4, pDONR P4rP3r, and pDONR P3P2 by BP recombination reaction. Finally, three entry vectors, pDONR P1P4_pIQD5, pDONR P4rP3r_GFP, and pDONR P3P2_IQD5, were delivered into destination vector pGWB501 by LR recombination reaction. The pIQD5::mCherry-IQD5 construct was generated by the same method, only replacing the entry vector pDONR P4rP3r_GFP by pDONR P4rP3r_mCherry. The mCherry-encoding sequence was amplified with primers attB4r_mCherry_F and attB3r_mCherry_R.

For the pTUB6::mCherry-TUB6 construct, the TUB6 (At5g12250) promoter and DNA fragment were amplified from the genomic DNA with primer pairs attB1_TUB6_F/attB4_TUB6_R and attB3_TUB6_F/attB2_TUB6_R separately and cloned into entry vector to generate pDONR P1P4_pTUB6 and pDONR P3P2_TUB6. Then, three entry vectors, pDONR P1P4_pTUB6, pDONR P4rP3r_mCherry, and pDONR P3P2_TUB6, were delivered into destination vector pGWB601 by LR recombination reaction.

The plasmids pGWB501_ pIQD5::GFP-IQD5 and pGWB501_pIQD5::mCherry-IQD5 were transferred into iqd5-1 and wild-type Col-0 plants, and hygromycin (30 µg mL⁻¹) was used to select transgenic plants. The pGWB601_pTUB6::mCherry-TUB6 construct was introduced into wild-type Col-0 plants. Transgenic plants were selected by resistance to glufosinate (Basta; Bayer), and only the transgenic plants without any noticeable phenotypes were selected as MT marker lines. The standard floral dipping method was used for Arabidopsis transformation as described before (Clough and Bent, 1998). Agrobacterium tumefaciens strain GV3101 was used for transformation. Homozygous T3 seedlings were used for phenotype analysis, confocal microscopy observations, and immunoblotting tests.

Drug Treatments

Three-day-old seedlings from the growth room were submerged in liquid 1/2 MS medium containing 10 µm oryzalin (Sigma) diluted from a 100 mm stock solution dissolved in DMSO (Sigma) and incubated for 30 to 120 min. To analyze root elongation, 300 nm oryzalin was added to 1/2 MS medium solidified with 0.8% (w/v) agar. Seeds were sown on the plates, and root lengths in 7-d-old seedlings were measured in ImageJ.

Confocal Microscopy Observation

Fluorescence images were acquired by confocal microscope (Leica SP5) using either a 20× (numerical aperture, 0.75) or a 40× (numerical aperture, 1.1) water-immersion lens. GFP and mCherry were excited at 488 and 561 nm and collected at 500 to 550 nm and 580 to 630 nm, respectively. For propidium iodide and FM4-64 dye, a 561-nm excitation laser was used and 580- to 630-nm emission was collected. Time-lapse images were acquired every 5 s during the course of 4 min 55 s.

RNA Isolation and Reverse Transcriptase-Mediated PCR Analysis

Total RNA was extracted by the RNeasy Plant Mini Kit (Qiagen) from 100 mg of cotyledons. On-column DNase digestion was performed by using the RNase-Free DNase Set (Qiagen), and 1 µg of RNA were reverse transcribed in a 20-µL reaction using SuperScript III reverse transcriptase (Invitrogen) with oligo(dT)₁₈ primer according to the manufacturer’s instructions. The synthesized first-strand cDNA was used as a template for PCR with primer pair IQD5_RT_F/R for IQD5 (Supplemental Table S1), and UBQ was used as an internal control.

Protein Expression

Full-length IQD5 cDNA plus 6×His Tag at the N-terminal end was synthesized and cloned into expression vector pET30a, expressed, and purified by Genscript (https://www.genscript.com/). Recombinant His-IQD5 proteins were in 50 mm MES buffer, pH 6. Protein samples of 2 μg were analyzed using 10% SDS-PAGE gels. The coding sequence of IQD5-F3 was cloned into vector pET28a using NheI and XhoI restriction enzyme cleavage sites to generate the His-IQD5_F3 construct. The specific primers were pETIQD5_F3_F and pETIQD5_F3_R (Supplemental Table S1). The expression construct was transformed and expressed in Escherichia coli Rosetta (DE3) strain (Novagen). The recombinant His-IQD5_F3 proteins were purified using Ni-NTA resin columns according to the manufacturer’s instructions (Thermo Scientific). Purified proteins were dialyzed using the Slide-A-Lyzer G2 Dialysis Cassette (Thermo Scientific), and the buffer was changed to 50 mm MES, pH 6. Protein concentration was measured by Bio-Rad protein assay kit. Protein samples were analyzed using 12% SDS-PAGE gels.

Image Analysis

All images were processed and analyzed in ImageJ (https://imagej.nih.gov/ij). The background was subtracted using the subtract background plugin in ImageJ. ImageJ software was used to quantify the numbers of cortical MTs in pavement cells. The fixed-length (10 µm) line perpendicular to the orientation of the most cortical MTs was drawn, and the number of cortical MTs across the line was counted. Five fixed-length lines were drawn from each cell, and at least 10 cells were used in each replicate. The average MT number was calculated before and after oryzalin treatments. Fluorescence signals were counted as MTs or MT bundles because it is impossible to distinguish between individual MTs and bundles using light microscopy. Student’s t test was used to analyze the significance of differences. MT angle measurement was performed using the angle measure tool of ImageJ. The defined axis was used as reference angle 0°, and all measured angles were normalized to it. MT array anisotropy was measured by the ImageJ FibrilTool plugin (Boudaoud et al., 2014). At least 10 regions of interest (surface area, ∼50 µm²; perimeter, ∼28 µm) from three to four cells were analyzed for each replicate. The anisotropy score was 0 for no order and 1 for perfectly ordered (Boudaoud et al., 2014). The MT growth and shrinkage rates were calculated by measuring the distance grown or shortened during the time spent in that event (Abe and Hashimoto, 2005). Dynamicity was calculated by dividing the sum of both the grown and shortened lengths by the total time (Abe and Hashimoto, 2005). Catastrophe and rescue frequency were calculated from the inverse of the time spent in growth and shrinkage, respectively (Abe and Hashimoto, 2005).

MT Spin-Down Assays

MT spin-down assays were performed using the MT Binding Protein Spin-Down Assay Kit (Cytoskeleton) according to the manufacturer’s instructions. The supernatant and pellet fractions were mixed with 5× or 1× SDS-PAGE sample buffer separately and analyzed using 10% or 12% SDS-PAGE gels followed by staining the gel with Coomassie Brilliant Blue R 250 or immunoblot analysis with anti-His (EMD Millipore) antibodies. The band intensities were quantified by ImageLab software 5.2.1 (Bio-Rad).

Tubulin Polymerization Assay

The MT in vitro polymerization assay was performed by incubating 1 µm IQD5-His protein, 10 µm rhodamine-labeled tubulin (Cytoskeleton), and 1 mm GTP (Cytoskeleton) in general tubulin buffer (without glycerol; Cytoskeleton) at 37°C for 1 h. A concentration of 1 µm GFP-His protein (Thermo Scientific) was used as a control. The tubulin polymerization was stopped by adding 1% (v/v) glutaraldehyde. Confocal microscopy images were acquired using a filter set of 530- to 550-nm excitation and 580- to 600-nm emission. The lengths of MTs were measured in ImageJ.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Identification of the IQD5 gene.

• Supplemental Figure S2. Subcellular localization of IQD5.

• Supplemental Figure S3. The localization of IQD5 to the cell cortex is abolished by oryzalin treatment.

• Supplemental Figure S4. The iqd5-1 mutant is sensitive to oryzalin treatment at the whole-plant level.

• Supplemental Figure S5. Comparison of MT arrays in hypocotyl epidermal cells.

• Supplemental Figure S6. His-IQD5 binds to MTs in vitro.

• Supplemental Figure S7. Phylogenetic tree of IQD proteins in Arabidopsis.

• Supplemental Figure S8. Expression patterns of all 33 IQD genes in Arabidopsis.

• Supplemental Table S1. List of primers used in this study.

• Supplemental Movie S1. IQD5 accumulates preferentially in the indentation region.

• Supplemental Movie S2. 3D reconstruction of the colocalization of GFP-IQD5 and the cell cortex.

• Supplemental Movie S3. Dynamic behavior of GFP-IQD5 in Arabidopsis pavement cells.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• GPT2 Gramene: AT1G61800

• GPT2 Araport: AT1G61800

• IQD1 Gramene: AT3G09710

• IQD1 Araport: AT3G09710

• IQD11 Gramene: AT5G13460

• IQD11 Araport: AT5G13460

• IQD14 Gramene: AT2G43680

• IQD14 Araport: AT2G43680

• IQD16 Gramene: AT4G10640

• IQD16 Araport: AT4G10640

• IQD21 Gramene: AT3G49260

• IQD21 Araport: AT3G49260

• IQD25 Gramene: AT4G29150

• IQD25 Araport: AT4G29150

• IQD6 Gramene: AT2G26180

• IQD6 Araport: AT2G26180

• IQD7 Gramene: AT1G17480

• IQD7 Araport: AT1G17480

• IQD8 Gramene: AT1G72670

• IQD8 Araport: AT1G72670

• MAP18 Gramene: AT5G44610

• MAP18 Araport: AT5G44610

• MES CHEBI: CHEBI:39010

• MOR1 Gramene: AT2G35630

• MOR1 Araport: AT2G35630

• RIC1 Gramene: AT2G33460

• RIC1 Araport: AT2G33460

• drug CHEBI: CHEBI:23888

• iqd5 Gramene: AT3G22190

• iqd5 Araport: AT3G22190