Identification of legume RopGEF gene families and characterization of a Medicago truncatula RopGEF mediating polar growth of root hairs


(fax +530 754 6617; e-mail


Root hairs play important roles in the interaction of plants with their environment. Root hairs anchor the plant in the soil, facilitate nutrient uptake from the rhizosphere, and participate in symbiotic plant–microbe interactions. These specialized cells grow in a polar fashion which gives rise to their elongated shape, a process mediated in part by a family of small GTPases known as Rops. RopGEFs (GEF, guanine nucleotide exchange factor) activate Rops to effect tip growth in Arabidopsis pollen and root hairs, but the genes mediating tip growth in legumes have not yet been characterized. In this report we describe the Rop and RopGEF gene families from the model legume Medicago truncatula and from the crop legume soybean. We find that one member of the M. truncatula gene family, MtRopGEF2, is required for root hair development because silencing this gene by RNA interference affects the cytosolic Ca2+ gradient and subcellular structure of root hairs, and reduces root hair growth. Consistent with its role in polar growth, we find that a GFP::MtRopGEF2 fusion protein localizes in the apex of emerging and actively growing root hairs. The amino terminus of MtRopGEF2 regulates its ability to interact with MtRops in yeast, and regulates its biological activity in vivo.


Root hairs are specialized cells that play important roles in plant growth by increasing the surface area of the root, thereby both anchoring the plant and facilitating the uptake of water and nutrients from the soil. Root hairs develop in a polarized fashion from specialized cells known as trichoblasts. Turgor pressure, in conjunction with a localized exocytosis of growth machinery and materials, culminates in polar tip growth and an elongated cell shape.

Recent studies have begun to elucidate the molecular mechanism underlying root hair development and polar tip growth. The rate of root hair growth oscillates with peaks of maximum elongation preceding periodic oscillations in calcium and reactive oxygen species (ROS) which may serve to either negatively regulate tip growth or to reinforce the new cell wall (Monshausen et al., 2007). One or more members of a family of small GTPases known as Rops function very early in root hair development to effect these processes (Molendijk et al., 2001; Jones et al., 2002). Arabidopsis Rop GTPases regulate the production of ROS by the NADPH oxidase rhd2 (Jones et al., 2007). Reactive oxygen species activate Ca2+ channels leading to a tip-focused calcium gradient that is critical in organizing the F-actin network (Jones et al., 2002; Foreman et al., 2003). The actin cytoskeleton mediates vesicle targeting and membrane cycling, leading to tip growth (Bloch et al., 2005). Calcium also binds to and activates RHD2, leading to positive feedback and polar growth of the root hair (Takeda et al., 2008).

Rop GTPases behave as molecular switches that cycle between GTP-bound active states and GDP-bound inactive states. Polar cell growth is a function of both Rop localization and the nature of the bound nucleotide. Accessory proteins regulate the cycling of Rop between the GTP- and GDP-bound forms in a spatio-temporal manner (Kost, 2008). The GTPase activating proteins (RopGAPs) accelerate GTP hydrolysis, thereby attenuating Rop signaling and restricting it to specific growth sites. The Rop GDP dissociation inhibitor proteins (RopGDIs) are required for proper Rop localization and for recycling inactive Rop from the plasma membrane for retargeting to growing cell tips. Here, guanine nucleotide exchange factors (GEFs) reactivate Rops by catalyzing the exchange of GTP for GDP. Plants contain two distinct classes of GEFs. The first class is similar to the DOCK family of RhoGEFs from animals. Arabidopsis contains a single DOCK-type GEF called SPIKE1 (SPK1). Spk1 mutants exhibit aberrant cell morphogenesis and SPK1 interacts with the SCAR/WAVE and ARP2 complexes to regulate actin-dependent growth in trichomes and pavement cells (Basu et al., 2008). A second, plant-specific class of RopGEFs was identified in Arabidopsis (Berken et al., 2005; Gu et al., 2006). This class contains a highly conserved PRONE (plant Rop nucleotide exchanger) domain flanked by variable N- and C-termini. The PRONE RopGEFs catalyze nucleotide exchange by Rops in vitro and modulate the polar growth of pollen tubes and root hairs in vivo. The PRONE domain is necessary and sufficient to effect nucleotide exchange while the N- and C-termini regulate this GEF activity (Berken et al., 2005; Gu et al., 2006; Won et al., 2009).

Legume species such as Medicago truncatula and Glycine max (soybean) possess the unique ability to participate in a symbiotic relationship with nitrogen-fixing bacteria known as rhizobia. This interaction is of great agronomic and ecological importance as it represents one of the principal conduits for inert, atmospheric dinitrogen to become biologically available. For symbiosis to occur, the bacteria must gain entry into the host’s root system, a process that is intimately linked to the polar growth of root hairs. Rhizobia secrete molecules known as Nod factors which reorient the direction of root hair growth, culminating in the entrapment of bacteria within a tight root hair curl from which the bacteria enter the plant (Esseling et al., 2003). It is therefore likely that the signaling pathways mediating Nod factor perception and root hair growth converge. Elucidating the mechanism underlying root hair development in M. truncatula may therefore provide insights as to how Nod factors modulate root hair growth to promote infection.

Although a connection between symbiotic nitrogen fixation and root hair growth is well established, the genes regulating polar growth have not been well described in legumes. Here we describe both the Rop GTPase and PRONE RopGEF gene families from two legume species, M. truncatula and G. max. Importantly, we report that a single M. truncatula RopGEF, MtRopGEF2, is required for proper root hair development. MtRopGEF2 lacks the C-terminal domain that has been shown to autoregulate other RopGEFs. Instead, we identify the variable N-terminus as the critical domain that regulates RopGEF2 activity in vivo and the interaction with Rop substrates in yeast.


Characterization of legume Rop GTPases

The M. truncatula genome encodes seven highly conserved Rop GTPase paralogs (Szucs et al., 2006; Liu et al., 2010). A high level of functional redundancy within this group of proteins in diverse angiosperm species has typically constrained their genetic analysis to methods involving protein misexpression or the introduction of protein isoforms with altered function. With the objective of studying the role of legume Rop proteins in root hair polarity, we used the Arabidopsis proteins to query the soybean genome sequences and identified sets of 20 soybean Rop homologs (GmRops). Maximum likelihood analysis involving the coding sequences of Medicago, soybean, Arabidopsis, and rice revealed that the phylogeny of this gene family is dominated by relatively simple orthologous relationships (Figure S1). For the legumes, pairs of GmRop homeologs are typically sister to a single MtRop homolog; the two exceptional cases of GmRops without a corresponding M. truncatula Rop are most simply explained by either deletions or incomplete sequence information from the M. truncatula genome. Of particular interest are the Type IV Rops, which regulate the polar growth of root hairs and pollen tubes in Arabidopsis (Zheng and Yang, 2000; Christensen et al., 2003; Szucs et al., 2006). Within the Type IV Rops there is evidence of lineage-specific duplication and deletion, as shown by the example of the AtRop1, -3, 5 triplication and corresponding duplications in the MtRop3, -5 and MtRop6 lineages. This same set of Medicago paralogs (i.e. MtRop3, -5, -6) are typically the most abundant transcripts and they have expression patterns that are highly symmetrical between tissues, with the exception of elevated ratios of MtRop5 in late seed development (Liu et al., 2010; Figure S2). These patterns of overall high sequence conservation and a history of plant family-specific gene duplication are consistent with the reported functional redundancy among these Arabidopsis paralogs (Jones et al., 2007).

To assess the possible role of legume Rop proteins in polar growth of root hairs, each MtRop was tested individually by over-expression of constitutively active (CA) and dominant negative (DN) isoforms. Transformation with CA or DN isoforms of the Type IV MtRops caused short, thick root hairs. Although similar to the DN Rop phenotype, CA Rops more severely depolarized growth and caused ballooning of root hair tips (Figure S3e–h). The most severe root hair phenotype was caused by over-expression of wild-type or CA MtRop10 (Figure S3i–l) which resembles the phenotype caused by CA Rop11 in Arabidopsis (Bloch et al., 2005). These data indicate that members of the M. truncatula Rop gene family possess conserved biochemical properties; however, they do not permit the assignment of individual paralogs to specific biological or developmental pathways.

Identification of RopGEFs from M. truncatula

Originally identified in tomato and Arabidopsis (Berken et al., 2005; Kaothien et al., 2005; Gu et al., 2006), the RopGEF gene family has not been well described in any legume species. We used the Arabidopsis RopGEF sequences to search the M. truncatula genome and EST databases for RopGEF homologs. Altogether we identified 10 M.  truncatulaRopGEF homologs which we cloned using 5′ and 3′ rapid amplification of cDNA ends (RACE), and RT-PCR (Table 1). The M. truncatula RopGEFs were numbered according to the closest Arabidopsis RopGEF based on amino acid identity and phylogenetic analyses (Figures 1 and 2, see below). All 10 Medicago proteins contained the highly conserved PRONE domain, while eight of the proteins contained both N- and C-terminal variable domains. MtRopGEF2 and MtRopGEF3 were unique because they lacked a significant C-terminal regulatory domain (Figure 1).

Table 1. Medicago truncatula Rop guanine nucleotide exchange factors (MtRopGEF) homologs identified in MtGI and NCBI databases
  1. EST, expressed sequence tag; TC, tentative consensus.

  2. aDenotes bacterial artificial chromosome (BAC) end survey sequence.

MtRopGEF1EX528980, BQ140633, TC138820mth-225K18aUnknown
MtRopGEF2TC123974, TC131647mth4-23B24FM1aUnknown
MtRopGEF3TC127088, AW691989mth2-189e27
MtRopGEF6TC128949, CX550738mth2-27n195
MtRopGEF14TC119326, TC115035, BI308670, CX521491mth2-81g165
Figure 1.

 Alignment of Medicago truncatula Rop guanine nucleotide exchange factors (MtRopGEFs).
clustalw alignment of the amino acid sequences of the 10 M. truncatula RopGEFs with the Arabidopsis AtRopGEF1 shown for comparison. The region of MtRopGEF2 used to create the RNA interference (RNAi) fragment is underlined and the site where MtRopGEF2 was truncated is indicated with an arrow. The three subdomains of the PRONE (plant Rop nucleotide exchanger) domain are highlighted with roman numerals.

Phylogenetic analyses of the legume RopGEF gene family

To gain insight into the evolution of this gene family in legumes, we used the M. truncatula RopGEF sequences to query the soybean genome sequence and identified 24 soybean RopGEF genes. The maximum likelihood tree generated using the deduced full-length peptide sequences formed two primary clades. Additionally, there were five well-supported subgroups which contained representatives of all five angiosperm species examined. In all cases, the phylogenetic relationships of the RopGEFs within these clades recapitulated the taxonomic relationships of these species, suggesting that these clades represent a radiation of five ancestral RopGEF genes that existed prior to the monocot–dicot divergence. There is evidence for multiple duplication events in groups II–IV, giving rise to the majority of the extant members of the RopGEF gene family. Group II contains two MtRopGEFs and includes AtRopGEF4 which is required for root hair elongation in Arabidopsis (Won et al., 2009). There are two monophyletic subgroups within group III, each of which contains closely related Arabidopsis, soybean and M. truncatula RopGEFs. Group III may have arisen from two ancient RopGEFs that pre-dated the split of monocots from dicots; however, low bootstrap values preclude us from inferring their precise relationships between the rice group III genes and the corresponding dicot subgroups (Figure 2). Arabidopsis AtRopGEF12 regulates pollen tube growth and occurs within group IV along with MtRopGEF8 and -12 (Zhang and McCormick, 2007). AtRopGEF1 can affect polar growth when ectopically expressed in pollen (Gu et al., 2006); this gene resides within group I and displays simple orthologous relationships to genes in Medicago, soybean and rice.

Figure 2.

 Phylogenetic analysis of the RopGEF family.
Maximum likelihood tree constructed using the amino acid sequences from Medicago truncatula, Glycine max, Arabidopsis thaliana, Oryza sativa, and Physcomitrella patens. Medicago Rop guanine nucleotide exchange factors (RopGEFs) were numbered based on their phylogenetic relationship to Arabidopsis RopGEFs as described in the text. Arabidopsis and rice sequences were previously described (Berken et al., 2005; Gu et al., 2006; Zhang and McCormick, 2007), G. max RopGEFs were named based on their genome position according to genome release version 1.0. Physcomitrella and rice sequences are listed according to their GenBank IDs and contain the prefixes XM_ or NM_ respectively. The tree was rooted using four related Physcomitrella sequences and bootstrap values <50 were omitted from the tree.

Expression analyses of Medicago RopGEFs

We performed semi-quantitative RT-PCR analyses of aerial (leaves, flowers, pods) and below ground (roots, root tips, nodules) M. truncatula tissues to examine patterns of RopGEF expression. All 10 MtRopGEFs could be amplified from at least one type of tissue. Most RopGEFs were expressed in multiple tissues with the exception of MtRopGEF8, which was specifically expressed in flowers. Notably, MtRopGEF8 is phylogenetically related to AtRopGEF12 which mediates pollen tube growth (Figure 2). The Affymetrix data at the Medicago truncatula Gene Expression Atlas (MtGEA) provide a more quantitative assessment of transcript abundance. Most of the RopGEFs were weakly expressed relative to the MtRops (compare Figure 3b with Figure S1) with the notable exception of MtRopGEF14, which was two- to 10-fold more highly expressed in aerial tissues than the other RopGEFs (Figure 3b, inset). MtRopGEF2, -6, -12, and -14 were the most abundant RopGEFs in root tissues and are good candidates to positively regulate the Rops that affect polar root hair growth. The strong sequence conservation and high level of overlap in expression between many MtRopGEFs suggests the possibility of functional redundancy among certain paralogs, as has been reported for the Rop GTPases (Jones et al., 2007) and RopGEFs in Arabidopsis pollen tubes (Gu et al., 2006).

Figure 3.

 Rop guanine nucleotide exchange factors (RopGEFs) are expressed ubiquitously in Medicago truncatula.
(a) Semi-quantitative RT-PCR using gene-specific primers in roots (R), toot tips (RT), 10-day-old root nodules (N), leaves (L), pods (P), or open flowers (F).
(b) Quantitative analysis of the MtRopGEFs derived from data obtained from the Medicago Gene Expression Atlas (Benedito et al., 2008). Note the differences in the values on the Y-axes indicating the high MtRopGEF14 expression relative to the other M. truncatula RopGEFs. Error bars represent standard error of the mean of the individual Affymetrix values for each gene.

MtRopGEF2 regulates root hair growth

To determine if any of the MtRopGEFs are required for root hair growth, we individually silenced each of the 10 MtRopGEF genes using RNA interference (RNAi). The RNAi constructs targeted the 5′ ends of the coding sequences, which were poorly conserved even among closely related RopGEFs (see Figure 1). Only the MtRopGEF2 RNAi (MtRopGEF2i) construct resulted in root hairs that differed from control root hairs (Figure 4a–f). Mature MtRopGEF2i root hairs were approximately one-third the length of full-grown root hairs from vector control plants or from root hairs expressing each of the other nine RNAi constructs (Figure 4i) and resembled root hairs expressing dominant negative forms of the Rop GTPases (Figure 4h). Frequently MtRopGEF2i root hairs manifested one or more regions of constriction (Figure 4f), suggesting a variable reduction in growth rate caused by RopGEF2 RNAi.

Figure 4.

Medicago truncatula Rop guanine nucleotide exchange factor 2 (MtRopGEF2) is required for root hair elongation.
Darkfield (a, b) or fluorescent (c, d) stereoscopic micrographs of roots transformed either with an empty pK7WIWG2(II)::DsRED vector (a, c), or MtRopGEF2 RNAi construct (b, c). Note that each micrograph contains one transformed and one untransformed root as an internal control as indicated by the presence or absence of DsRED fluorescence.
(e–h) Differential interference contrast (DIC) images of full-grown root hairs from transgenic roots expressing either (e) empty pK7WIWG2(II)::DsRED, (f) MtRopGEF2i, (g) empty pKGW-RR, or (h) pKGWRR::35S::Rop3(D121A).
(i) Quantitative analysis of root hair length in plants expressing RNAi constructs against each of the 10 MtRopGEFs. At least 50 root hairs from five different plants were measured for each genotype. Error bars represent standard error. *Indicates significant difference ( 0.001 general linear model anova, sas) (SAS Institute,
(j) Semi-quantitative RT-PCR of MtRopGEF2 and MtRopGEF3 performed on cDNA isolated from either two independent roots expressing a MtRopGEF2i construct or two independent roots containing an empty pK7WIWG2(II)::DsRED vector.
Scale bars represent 300 μm in (a–d) and 50 μm in (e–h).

To test the specificity our RNAi constructs, we generated cDNA from RNA isolated from RopGEF2i and vector control roots and conducted semi-quantitative (sq)RT-PCR analyses using gene-specific primers. Expression of MtRopGEF2 in MtRopGEF2i-expressing roots was consistently lower than in vector controls. In contrast, the expression of MtRopGEF3, the closest MtRopGEF2 homolog (see Figure 2), did not differ between MtRopGEF2i roots and vector controls, indicating that our hairpin constructs were specific for MtRopGEF2 (Figure 4j). Two additional RNAi constructs targeting different regions of MtRopGEF2 exhibited an identical, aberrant root hair phenotype (Figure S4). These results indicate that MtRopGEF2 is specifically required for root hair elongation.

MtRopGEF2 is required to maintain cytoarchitecture and Ca2+ gradients in growing root hairs

Root hair cells possess a tip-focused gradient of cytosolic free Ca2+([Ca2+]cyt) that is required for polar growth. We investigated this [Ca2+]cyt gradient in actively growing wild-type and MtRopGEF2i root hairs using the Yellow Chameleon 3.6 (YC3.6) calcium sensor. Elongating control root hairs expressing YC3.6 were phenotypically similar to untransformed root hairs. They were 237.78 ± 107.52 μm (= 75) in length and contained a region of dense subapical cytoplasm (Figure 5a). The nuclei in these root hairs were positioned at the base of the subapical cytoplasm, while the subtending root hair shaft consisted primarily of a central vacuole surrounded by a thin layer of cytoplasm. In contrast, elongating MtRopGEF2i root hairs were short and swollen (47.91 ± 18.63 μm, = 75) (Figure 5d) and were devoid of the dense subapical cytoplasm observed in the control root hairs. The nuclei of MtRopGEF2i root hairs were often positioned at the base of the root hair shaft (Figure 5d) or completely absent from the shaft, an architecture typically associated with full-grown root hairs (Esseling et al., 2004).

Figure 5.

 Silencing Medicago truncatula Rop guanine nucleotide exchange factor 2 (MtRopGEF2) disrupts the subcellular architecture and calcium gradients of elongating root hairs.
Analysis of subcellular structure and cytosolic free Ca2+([Ca2+]cyt) levels in control (a–c) or MtRopGEF2i (d–f) root hairs expressing the Chameleon YC3.6 calcium reporter. Differential interference contrast (DIC) microscopy shows that vector control root hairs (a) contain dense subapical cytoplasm and nuclei (arrowhead) whereas the apices of MtRopGEF2i root hairs (d) consist primarily of vacuole with nuclei localized to the base of the root hairs. A fluorescent micrograph in the fluorescence resonance energy transfer (FRET) channel shows an elevated [Ca2+]cyt gradient (shown as red fluorescence) in the control root hair apex (b) which is absent from the MtRopGEF2i root hair (e). Quantitative analysis of FRET intensity along a longitudinal transect from tip to base of root hairs ios shown in (c) and (f). The negative slope of the best-fit curve of the control root hair (c) shows a high level of free [Ca2+]cyt at the root hair apex and a low level of free [Ca2+]cyt at the base. The positive slope of the curve in (f) indicates a high level of free [Ca2+]cyt at the base of MtRopGEF2i root hairs near the nuclei, not at the apex. The fragment used for MtRopGEF2-RNAi is identical to that used in Figure 4. Graphs and images depicted are representative of vector control (= 6, slope = −0.00695 ± 0.0021) and MtRopGEF2i (= 6, slope = +0.00483 ± 0.0028) root hairs. Error bars represent 10 μm.

In control M. truncatula root hairs the slope of the fluorescence resonance energy transfer (FRET)/cyan fluorescent protein (CFP) ratio, from the root hair tip to the root hair base, was negative (−7.4 nm−1), indicating a higher [Ca2+]cyt at the tip than the base (Figure 5c). This [Ca2+]cyt gradient was severely affected in MtRopGEF2i root hairs, with an inverted, base-focused calcium gradient evidenced by a positive (8.4 nm−1) FRET/CFP slope from the root hair apex towards base (Figure 5f). Thus, aberrant root hair growth in MtRopGEF2i root hairs is correlated with altered cytoarchitecture and a disrupted [Ca2+]cyt gradient.

Over-expression and localization of MtRopGEF2

Constitutively expressing the Arabidopsis genes RopGEF1 or RopGEF12 in tobacco pollen causes tip swelling and tube thickening, respectively. Both of these proteins have the canonical RopGEF domain structure, with a central conserved PRONE domain and variable N- and C-termini. Independent studies document the important role of the variable C-terminus in regulating the function of RopGEF. Gu et al. (2006) demonstrated that C-terminal deletion constructs of RopGEF1 were enhanced in their ability to stimulate in vitro GTPase activity of the Rop1 substrate, while Zhang and McCormick (2007) observed that deletion of the C-terminal regulatory domain of AtRopGEF12 was sufficient to obtain wavy and branched pollen tubes in over-expression assays. In M. truncatula, MtRopGEF2 does not contain a variable domain C-terminal (Figure 1) and thus is potentially analogous to the hyper-stimulatory C-terminal deletions of RopGEF1 and RopGEF12 (Gu et al., 2006; Zhang and McCormick, 2007). Using the strategy of Zhang and McCormick (2007), we translationally fused GFP to the N-terminus of MtRopGEF2 and expressed this construct in M. truncatula roots under the control of the constitutive 35S promoter. Over-expression of the GFP::MtRopGEF2 construct did not affect root hair growth, as root hairs were of similar length and width to root hairs expressing GFP controls (Figure 6a,b,g,h). Furthermore, we did not observe either wavy or branched cells as had been observed in pollen tubes over-expressing RopGEFs and root hairs over-expressing Rops or RopGEFs (Jones et al., 2002; Gu et al., 2006; Zhang and McCormick, 2007; Won et al., 2009).

Figure 6.

 The N-terminus of Medicago truncatula Rop guanine nucleotide exchange factor 2 (MtRopGEF2) regulates its activity.
Differential interference contrast (DIC) images of fully grown root hairs expressing (a) 35S::GFP, (b) 35S::GFP::MtRopGEF2, or (c) 35S::GFP::ΔNMtRopGEF2. Confocal micrographs of (d) emerging root hair expressing 35S::GFP, (e) emerging root hair expressing 35S::GFP::MtRopGEF2, (f) elongating root hair expressing 35S::GFP::ΔNMtRopGEF2. Roots were stained with propidium iodide prior to imaging. (g, h), quantitative analysis of the length (g) and width (h) of root hairs expressing 35S::GFP, 35S::GFP::MtRopGEF2, or 35S::GFP::ΔNMtRopGEF2 as indicated. At least 50 root hairs from five different plants were measured for each genotype. Error bars represent standard error. *Indicates significant difference, (P ≤ 0.001 general linear model anova, sas) (SAS Institute) Scale bars represent 50 μm in (a–c), 10 μm in (d–f).

In their study of RopGEF1, Gu et al. (2006) observed partial stimulation of GEF activity on deletion of the N-terminal regulatory region, suggesting that the N-terminus negatively regulates GEF activity. In contrast, Zhang and McCormick (2007) observed that co-expression of YFP-RopGEF12 with CFP-PRK2 severely depolarized pollen tubes whereas co-expressing YFP-RopGEF12ΔN with CFP-PRK2 did not, suggesting a positive role for the N-terminus of RopGEF12. To test the possibility that the variable N-terminus of MtRopGEF2 might also have regulatory properties, we deleted the N-terminus of MtRopGEF2 and cloned GFP onto the N-terminus of the deletion construct (GFP:ΔNMtRopGEF2). In contrast to the full-length GFP::MtRopGEF2, over-expression of GFP:ΔNMtRopGEF2 resulted in root hairs that were shorter and more swollen relative to either GFP- or GFP::MtRopGEF2-expressing root hairs (Figure 6c), being approximately two-thirds the length and twice the width of root hairs expressing either the GFP control or GFP::MtRopGEF2 (Figure 6g,h).

The subcellular location of GFP::MtRopGEF2 and GFP:ΔNMtRopGEF2 was assessed by means of confocal microscopy. Fluorescence from the GFP control was observed throughout the nucleus and cytoplasm (Figure 6d), while GFP::MtRopGEF2 localized in the growing tips of emerging and elongating root hairs (Figure 6e), consistent with its role in mediating root hair growth. Analyses of root hairs expressing the chimeric N-terminal deletion (GFP:ΔNMtRopGEF2) revealed the protein predominantly at the tip of growing root hairs, similar to full-length GFP:MtRopGEF2, indicating that the N-terminus is not required for localization at the root hair tip (Figure 6f).

MtRopGEF2 interacts with multiple MtRop GTPases

RopGEFs re-activate RopGTPases by catalyzing their exchange of GDP for GTP. Nucleotide exchange is caused by a conformational change induced by the physical interaction between the two proteins (Berken et al., 2005; Gu et al., 2006; Thomas et al., 2006). We asked if MtRopGEF2 preferentially interacted with any of the Rop GTPases in a yeast two-hybrid assay, using MtRopGEF2 and the seven MtRops as ‘bait’ and ‘prey’, respectively. We found that yeast strains co-expressing each of the seven MtRops with MtRopGEF2 were able to grow in the absence of histidine, indicating that MtRopGEF2 could interact with all seven Rop proteins (Figure 7). However, there was significant variability in the strength of the interactions. MtRopGEF2 interacted most strongly with MtRop9 and only weakly with MtRop7 and MtRop8. MtRopGEF2 did not activate expression of the HIS reporter gene when expressed in the presence of either an empty prey vector or the arbitrary prey protein AvrB, indicating that the observed growth was not due to autoactivation of the reporter gene by MtRopGEF2. Likewise, yeasts expressing either the arbitrary bait protein Rin4 or an empty bait vector were unable to grow in the absence of histidine, indicating that the Rop GTPases did not autoactivate the HIS reporter.

Figure 7.

 The amino terminus of Medicago truncatula Rop guanine nucleotide exchange factor 2 (MtRopGEF2) regulates the interaction with multiple Rop GTPases.
Yeast strains expressing the bait constructs listed on the left, were mated with yeast strains of the opposite mating type expressing the prey constructs listed at the top. Numbers refer to specific M. truncatula Rop GTPases as described in the text using the nomenclature proposed by Szucs et al., (2006). ‘V’ denotes the empty prey vector pLAW11 whereas +H and −H respectively denote the presence or absence of histidine in the medium. AvrB and Rin4 are arbitrary bait and prey proteins which interact with each other (Mackey et al., 2002) but are not known to interact with any Rop or RopGEF. All photographs were taken from the same +H or −H plates at the same time.

Removing the N-terminal variable domain from Arabidopsis AtRopGEF1 increases its GEF activity in vitro (Gu et al., 2006) whereas a similar MtRopGEF2 deletion construct gains the ability to affect root hair growth in vivo (this study). We therefore asked if removing the N-terminus would alter the ability of MtRopGEF2 to interact with the Rop GTPases. We expressed ΔNMtRopGEF2 in yeast with each of the seven MtRops and tested the ability to grow in the absence of histidine. ΔNMtRopGEF2 interacted more robustly with all seven MtRops, including MtRop7 and -8, than did full-length MtRopGEF2, suggesting that the N-terminus inhibits or otherwise regulates the interaction. Similar to the full-length MtRopGEF2, yeast expressing ΔNMtRopGEF2 did not grow in the presence of either an empty prey vector or AvrB, indicating that ΔNMtRopGEF2 did not autoactivate the His reporter gene. These results suggest that the N-terminus may inhibit the ability of MtRopGEF2 to interact with its Rop substrates.


MtRopGEF2 regulates root hair elongation and structure in M. truncatula

The molecular mechanism underlying polar root hair growth is a topic that is particularly relevant in legume biology as root hairs constitute the entry point through which symbiotic bacteria gain access to their hosts. Here we described the Rop and RopGEF gene families from two legume species, M. truncatula and G. max, and begin functional analysis of the gene families in M. truncatula. Of the 10 M. truncatula MtRopGEFs, RopGEF2, -6, and -14 were the most highly expressed in roots but only MtRopGEF2 is required for root hair elongation. Silencing MtRopGEF2, but not the other RopGEFs causes short root hairs with aberrant cytoarchitecture, altered calcium gradients and depolarized growth. The closest RopGEF2 homolog, MtRopGEF3, shares 72% amino acid identity; importantly, MtRopGEF3 expression was not altered in MtRopGEF2i roots. Moreover, MtRopGEF3 does not appear to function in root hair elongation, as it is weakly expressed in roots and MtRopGEF3i root hairs were of similar length to control root hairs. This indicates that M. truncatula requires MtRopGEF2 for proper root hair elongation and that the other endogenous MtRopGEFs cannot fully compensate for MtRopGEF2 function.

Polar tip growth in plant cells such as root hairs and pollen tubes is controlled by a high [Ca2+]cyt gradient at the growing apex (Rosen et al., 1964; Reiss and Herth, 1979; Pierson et al., 1996; Bibikova et al., 1997; Felle and Hepler, 1997; Wymer et al., 1997; Li et al., 1999; Monshausen et al., 2007, 2008). This gradient is critical for maintaining proper cytoskeletal and cytological structure and can be regulated by Rop GTPases (Molendijk et al., 2001; Gu et al., 2005; Hwang et al., 2005). Over-expressing constitutively active Rop GTPases in Arabidopsis creates multiple [Ca2+]cyt foci in depolarized root hairs (Molendijk et al., 2001). In pollen tubes, only GTP-bound (active) Rop1 interacts with downstream effectors to organize [Ca2+]cyt gradients and F-actin to regulate polar growth processes. Perturbing these processes by mis-expressing Rop1 effectors or by pharmacologically altering [Ca2+]cyt or F-actin levels disrupts polar growth (Gu et al., 2005).

In M. truncatula, silencing MtRopGEF2 disrupted both the [Ca2+]cyt gradient and the cytoarchitecture of the root hair tip, suggesting that MtRopGEF2 activates one or more Rop GTPases that function to maintain the [Ca2+]cyt gradient and cytoarchitecture that are characteristic of growing root hairs. Which Rops are regulated by MtRopGEF2 in root hairs remains an open question. Ectopic expression studies implicate Arabidopsis Rop2, -4, -6, and -11 in various aspects of root hair development. Over-expressing AtRop2 affects both root hair initiation and elongation, whereas over-expression of constitutively active and dominant negative AtRop2, -4, -6, and -11 mutants perturbs root hair tip polarity (Molendijk et al., 2001; Jones et al., 2002; Bloch et al., 2005). MtRopGEF2 interacted most strongly with MtRop9 in a yeast two-hybrid assay, and to a lesser degree with MtRop3, -5, -6, and -10. Based on expression, MtRop3, -5, -6, and -10 would seem to be better candidates for regulating root hair development because they are more strongly expressed in roots than MtRop9 for which expression is most correlated with seed development.

In addition to root hair elongation, Rop GTPases also regulate the initiation of root hair formation. Rops localize to the site of root hair growth and over-expression of either AtRop2 or AtRopGEF4 causes root hair branching and ectopic root hair development (Molendijk et al., 2001; Jones et al., 2002; Won et al., 2009). We did not observe gross differences in the numbers of root hairs in MtRopGEF2i roots (Figure 4), nor did we observe root hair branching in roots over-expressing any of the M. truncatula Rop GTPases or GFP::MtRopGEF2 (Figures 6 and S3). This situation may reflect a fundamental difference in the pattern of root hair development between the two species. In contrast to Arabidopsis, which forms discrete cell files of hair and non-hair cells, every epidermal cell in M. truncatula forms a root hair (Sieberer and Emons, 2000). Nevertheless, we cannot exclude the possibility that MtRopGEF2 is involved in root hair initiation because we were unable to completely eliminate MtRopGEF2 expression using RNAi. A more conclusive analysis of RopGEF2 function in root hair initiation and branching would involve the identification of a null allele of MtRopGEF2. We are currently searching for such mutants in M. truncatula by Tnt1 insertion and TILLING populations.

Won et al. recently demonstrated that AtRopGEF4 is required for root hair elongation in Arabidopsis (Won et al., 2009). Maximum likelihood analysis indicates that AtRopGEF4 clusters with MtRopGEF2 in group II. Although similar in structure and biological function, there are several notable differences between MtRopGEF2 and AtRopGEF4. While MtRopGEF2i root hairs were approximately one-third the length of control root hairs, eliminating AtRopGEF4 only reduced root hair length by approximately 20%. In contrast to M. truncatula, which contains two group II RopGEFs, Arabidopsis contains three homologous genes in group II. One of these genes, AtRopGEF3, is more similar to MtRopGEF2, and is more highly expressed in roots than is AtRopGEF4 (Zhang and McCormick, 2007). Furthermore, while AtRopGEF4 exhibits root hair-specific expression, AtRopGEF3, like MtRopGEF2, is expressed in diverse aerial tissues (Zhang and McCormick, 2007). Although an AtRopGEF3 mutant has not yet been described, based on its higher similarity to MtRopGEF2 we hypothesize that AtRopGEF3 may function cooperatively with AtRopGEF4 to mediate root hair elongation in Arabidopsis.

The N-terminus regulates MtRopGEF2 activity

Most RopGEFs contain a well-conserved catalytic PRONE domain that is flanked by highly variable regulatory domains on both the N- and C-termini. These variable regions have been proposed to allow the RopGEFs to activate Rop signaling in response to diverse environmental and developmental signals (Shichrur and Yalovsky, 2006). The C-terminus in particular plays a crucial role in autoinhibiting RopGEF activity, because removing or mutating this domain constitutively activates RopGEFs both in vitro and in vivo (Gu et al., 2006; Zhang and McCormick, 2007). The C-terminus of Arabidopsis RopGEF12 is necessary and sufficient to interact with the pollen receptor PRK2, which may activate RopGEF12 to promote polar pollen tube growth. MtRopGEF2 differs from the previously characterized Arabidopsis pollen tube RopGEFs in that it contains a variable N-terminus but not a C-terminus. This suggests that the regulatory mechanism differs from that of the RopGEFs that regulate pollen tube polarity. We found that removing the N-terminus significantly affects its activity in vivo, because constitutively expressing GFP:ΔNMtRopGEF2 causes a significant shortening and thickening of root hairs, indicative of a wider zone of active Rop and therefore a wider zone of growth at the tip of the root hair cell. Notably, the expression of CA MtRops also produced shorter and thicker root hairs under our growth conditions in which the root hairs are grown in contact with the air (Figure S3). The increased activity of N-terminal deletions in planta is also consistent with our yeast two-hybrid results, which demonstrate increased affinity of ΔNMtRopGEF2 for all Rop substrates. The MtRopGEF2 N-terminus may sterically hinder the Rop–RopGEF interaction, for example by occluding the binding pockets for certain Rop substrates. Increased interaction with Rop substrates could also be achieved through an indirect mechanism. For example, structural studies demonstrate that RopGEFs function as homodimers and that dimerization is required for the Rop–RopGEF interaction (Thomas et al., 2007; Berken and Wittinghofer, 2008). Removing the N-terminus may promote dimerization and therefore the interaction with Rop. The mechanism by which the variable domains regulate RopGEF function remains an interesting and open question and may provide important insights into how specificity in Rop signaling is achieved.

How is MtRopGEF2 regulated?

The plant hormones auxin and ethylene function in a coordinated fashion with Arf-mediated secretion both to determine the position of root hair initiation and to regulate root hair elongation (Pitts et al., 1998; Fischer et al., 2006). Accumulation of Rop2 at the site of future root hair formation precedes bulge formation in Arabidopsis trichoblasts. Site-specific localization is a function of auxin and ethylene perception and requires a functional Arf GTPase pathway (Xu and Scheres, 2005; Fischer et al., 2006). Likewise, localized auxin perception triggers rapid, site-specific Rop2 and Rop6 activation, leading to the interdigitation of Arabidopsis pavement cells (Xu et al., 2010). If RopGEF2, like Rop2, is also required for root hair bulge formation, it is likely that localization of RopGEF2 may also be a function of localized hormone perception. How hormone perception influences the localized accumulation of tip growth machinery remains to be determined.

The rate of pollen tube and root hair elongation fluctuates, such that growth oscillates with the same periodicity as Rop activation, calcium and ROS increases, and F-actin accumulation (Hwang et al., 2005; Monshausen et al., 2007, 2008). These oscillations are likely to reflect both the positive and negative feedback that takes place during polar tip growth (Hwang et al., 2005; Takeda et al., 2008). In Arabidopsis root hairs, ROS production by RHD2 triggers release of Ca2+, which feeds back to stimulate further ROS production. In Arabidopsis pollen, the Rop-dependent secretion of the GTPase activation protein REN1 inactivates Rop at the pollen tube tip to prevent ectopic growth (Hwang et al., 2008). Is the concentration of RopGEF2 at the tip regulated in an oscillatory manner (similar to REN1) or is its biochemical activity regulated by some unidentified factor (such as Ca2+ in the case of RHD2)? Alternatively both RopGEF2 concentration and activity may remain constant with the availability of GDP-Rop substrate at the tip of the root hair providing the limiting factor for GEF activity. Future work must integrate RopGEF2 into the model describing oscillating tip growth of root hairs.

Zhang et al. suggested that leucine-rich repeat receptor kinases co-localize with RopGEFs and promote Rop activation at the plasma membrane of pollen tubes (Zhang and McCormick, 2007). Membrane-spanning receptors may allow root hair growth to respond to changes in environmental conditions such as physical contact, or the presence of symbiotic microbes for example. Could MtRopGEF2 also be regulated by receptor kinases as has been observed for the RopGEFs mediating pollen tube growth? There are multiple receptor kinases that affect different aspects of root hair development and polar growth which could potentially regulate MtRopGEF2 (Esseling et al., 2004; Jones et al., 2006; Kwak and Schiefelbein, 2007). However the lack of a variable C-terminal domain on MtRopGEF2 suggests that the intermolecular mechanism differs from that proposed for the pollen RopGEFs. The Rop4–PRONE8–GDP structure suggests that the region between the second and third subdomain of the PRONE may face the plasma membrane and could serve as an alternative point of contact with membrane-bound receptor kinases (Berken and Wittinghofer, 2008). Identification of novel MtRopGEF2-interacting proteins may help to elucidate the upstream signaling components that translate environmental and developmental cues into polar root hair growth.

We have only begun to elucidate the mechanism that underlies polar root hair growth. The identification of a RopGEF that positively affects root hair elongation has provided an important link between the Rop-regulated tip-growth machinery and the signals that turn this pathway on and off. Future work will address the mechanism by which MtRopGEF2 is regulated, how its activity is integrated into current models describing polar tip growth, and if this pathway facilitates infection by symbiotic microorganisms.

Experimental procedures

Identification and cloning of legume Rops and RopGEFs

We identified Medicago RopGEFs by querying both the M. truncatula expressed sequence tag (EST) database (MtGI), and the M. truncatula genome sequence with the coding sequences of the Arabidopsis RopGEFs. Bacterial artificial chromosomes identified as containing possible RopGEFs were analyzed using softberry fgenesh gene prediction tools to identify possible open reading frames in MtRopGEFs ( Tentative consensus (TC) sequences identified in MtGI were also used to design primers for 5′ and 3′ RACE RT-PCR using the SMART RACE cDNA amplification kit (Clontech, and TA cloned into pENTR-D TOPO for sequencing. Full-length cDNAs for all 10 Medicago RopGEFs were subsequently amplified using primers downstream of the first, in-frame predicted stop codon. The resulting clones were sequenced to confirm each of the coding sequences.

The coding sequences for the soybean Rops and RopGEFs were identified by querying the soybean genome with the MtRop and MtRopGEF sequences using the blast algorithm. The annotated genes were aligned with M. truncatula and Arabidopsis sequences using clustalw ( Genes appearing less then full-length were further analyzed using fgensh to predict the coding sequence from the surrounding genomic sequence. The predicted sequences were subsequently realigned with the Arabidopsis and M. truncatula genes and indels were manually curated using the genome sequence as template.

Phylogenetic analyses

Nucleotide (Rop) and amino-acid (RopGEF) alignments were performed using clustalw and edited using jalview. Maximum likelihood analyses were performed through the cipres web portal ( utilizing RAxML with the BLOSUM62 substitution matrix. The number of bootstraps was automatically determined by the software (Stamatakis et al., 2008). Trees were visualized and edited using mega 4.0 (

Semi-quantitative RT-PCR

RNA was isolated from the indicated tissues using Qiagen RNeasy kits ( Single-stranded (ss)DNA template was generated from equivalent quantities of RNA using a Quantitect Reverse transcription kit (Qiagen) and equal volumes were used for PCR using gene-specific primers. Actin primers were used as a loading control.

Yeast two-hybrid assays

MtRopGEF2, ΔNMtRopGEF2, and MtRop coding sequences were amplified using gene-specific primers, and cloned into pDONR207 (Invitrogen, To prevent membrane association of the GAL4 fusion proteins, we mutated the putative isoprenylation site (Cys to Ser) in MtRop C-termini (Berken and Wittinghofer, 2008). pDONR207:MtRopGEF2 and ΔNMtRopGEF2 were recombined into pLAW10 which contains the GAL4 DNA-binding domain (bait), and MtRops were recombined into pLAW11 which contained the Gal4 activation domain (prey) using LR reaction (Invitrogen). The Gateway-compatible yeast two-hybrid vectors pLAW10 and pLAW11 were a generous gift of Dr Richard Michelmore (University of California, Davis). Yeast transformation and mating were performed according to the Matchmaker GAL4 Two-Hybrid System 3 User Manual and Yeast Protocols Handbook (Clontech). The interactions were analyzed on SC-Trp-Leu-His medium supplemented with 2.0 mm 3-aminotriazole (Sigma,

FRET assays to measure [Ca2+]cyt gradient in root hairs

The MtRopGEF2-specific RNAi fragment was cloned into the pH7GWIWG2(II)-YC3.6 Gateway® based binary vector by LR recombination (Invitrogen) (den Os et al., University of Wisconsin Madison, Madison, WI, unpublished data). The RNAi vector construct was subsequently sequenced and transformed into Agrobacterium rhizogenes strain MSU440. The construct was then transformed into M. truncatula Jemalong A17 by hairy root transformation (Boisson-Dernier et al., 2001). Plants transformed with the empty pH7GWIWG2(II)-YC3.6 vector were used as a control. Transgenic hairy roots were selected under an epifluorescence binocular microscope utilizing yellow fluorescent protein (YFP) fluorescence. The transgenic plants (RNAi and control) were mounted on glass slides covered with Fahraeus medium (0.9% agar) and incubated in a growth chamber at an angle of 45° for at least 16 h before observation. Root hair phenotype and Ca2+ measurements were performed under Zeiss LSM 510 Meta confocal microscope (Carl Zeiss Inc., Young and emerging root hairs from the apical region of the root elongation zone were chosen for Ca2+ measurements. The Ca2+ sensor was excited at 458 nm using an argon laser. The CFP (473–505 nm) and FRET (536–546 nm) emissions were collected using a 458-nm primary dichroic mirror and the meta-detector of the microscope. Bright field images were acquired simultaneously by utilizing the transmission detector of the confocal microscope. For time-lapse analyses, images were collected every 2.5 sec. To measure the [Ca2+]cyt gradient, CFP and FRET intensities were collected from the tip to the base of root hairs. Background CFP and FRET signal intensities were subtracted from the signals from the root hairs. The means of 25 frames were calculated for each pixel along the root hair length and the FRET/CFP ratio was calculated from the root hair tip to the base along the cytoplasm. The central vacuole region of the root hairs and the nuclei were avoided for [Ca2+]cyt gradient measurement. The FRET/CFP ratio was plotted against the distance from the root hair tip (μm). Regression analyses were performed and the slope of the linear regression curve was estimated for each data set. A representative sample of six measurements is shown for control and MtRopGEF2-silenced root hairs.


We would like to thank Dr Richard Michelmore and Luis Williams for assistance with the yeast two-hybrid assay, Ben Rosen for assistance with the phylogenetic analysis and Stephanie Porter for her help with the statistical analyses. This research was supported by USDA Hatch no. WIS01163 and National Science Foundation no. 0701846 to JMA. HH was supported by Fundamental Research Funds for the Central Universities of China #200-124030. Support to DRC and BKR was provided by DOE Energy Biosciences award DE-FG02-01ER15200.