IMPa-4, an Arabidopsis Importin α Isoform, Is Preferentially Involved in Agrobacterium-Mediated Plant Transformation

Abstract

Successful transformation of plants by Agrobacterium tumefaciens requires that the bacterial T-complex actively escorts T-DNA into the host's nucleus. VirD2 and VirE2 are virulence proteins on the T-complex that have plant-functional nuclear localization signal sequences that may recruit importin α proteins of the plant for nuclear import. In this study, we evaluated the involvement of seven of the nine members of the Arabidopsis thaliana importin α family in Agrobacterium transformation. Yeast two-hybrid, plant bimolecular fluorescence complementation, and in vitro protein–protein interaction assays demonstrated that all tested Arabidopsis importin α members can interact with VirD2 and VirE2. However, only disruption of the importin IMPa-4 inhibited transformation and produced the rat (resistant to Agrobacterium transformation) phenotype. Overexpression of six importin α members, including IMPa-4, rescued the rat phenotype in the impa-4 mutant background. Roots of wild-type and impa-4 Arabidopsis plants expressing yellow fluorescent protein–VirD2 displayed nuclear localization of the fusion protein, indicating that nuclear import of VirD2 is not affected in the impa-4 mutant. Somewhat surprisingly, VirE2–yellow fluorescent protein mainly localized to the cytoplasm of both wild-type and impa-4 Arabidopsis cells and to the cytoplasm of wild-type tobacco (Nicotiana tabacum) cells. However, bimolecular fluorescence complementation assays indicated that VirE2 could localize to the nucleus when IMPa-4, but not when IMPa-1, was overexpressed.

INTRODUCTION

The Gram-negative soil bacterium Agrobacterium tumefaciens is the causative agent of crown gall tumors in plants. Plant transformation results from interkingdom transfer of genetic material from bacteria to plant cells and is the end product of multiple events initiating in the bacterium and terminating in the infected plant (reviewed in Gelvin, 2000, 2003; Zupan et al., 2000; Tzfira and Citovsky, 2001; Lacroix et al., 2006; McCullen and Binns, 2006). During transformation, T-DNA, a part of the bacterial Ti-plasmid, is transferred from the bacterium to the plant cell with the aid of a number of Virulence (Vir) proteins encoded by the Ti-plasmid. Processing of T-DNA is initiated by the border-specific VirD1/VirD2 endonuclease (Yanofsky et al., 1986; Jayaswal et al., 1987; Filichkin and Gelvin, 1993), which nicks the bottom strand of T-DNA at T-DNA borders (Stachel et al., 1986; Wang et al., 1987). Nicking at border sequences is coupled to covalent linkage of VirD2 via its Tyr-29 residue to the 5′ end of the resulting single-stranded (ss) T-DNA molecule (Vogel and Das, 1992), the T-strand (Herrera-Estrella et al., 1988; Ward and Barnes, 1988; Durrenberger et al., 1989). VirE2, a nonspecific ssDNA binding (SSB) protein, cooperatively binds to the T-strand and protects it from endonucleases and exonucleases (Citovsky et al., 1989; Sen et al., 1989). The T-strand, with a single affixed VirD2 and coated with VirE2, is called the T-complex. Formation of the T-complex most likely occurs within the plant because T-strands and VirE2 proteins are transferred independently to plant cells (Otten et al., 1984; Citovsky et al., 1992; Binns et al., 1995; Stahl et al., 1998, Lee et al., 1999). A currently popular model posits that the assembled T-complex translocates to the nucleus, where T-DNA subsequently integrates into the plant genome.

In eukaryotes, the nuclear membrane separates the major genome from the rest of the cell. The double membrane nuclear envelope maintains structural integrity of the nucleus and allows molecular communication through specific channels, the nuclear pores, which contain proteins that form the nuclear pore complex. These channels serve as conduits for the passive diffusion of small macromolecules (<40 to 60 kD), metabolites, and ions. By contrast, nuclear translocation of most macromolecules requires an energy-dependent active import/export process (reviewed in Stewart et al., 2007), which is mediated by a variety of soluble nuclear transport factors. Many karyophilic proteins contain cis-acting nuclear localization signal (NLS) sequences that are recognized by these nuclear transport factors (Goldfarb et al., 2004).

Import of a NLS-containing protein is initiated by the energy-independent formation of a cargo-receptor complex, the nuclear pore-targeting complex (PTAC) (Quimby and Corbett, 2001; reviewed in Gasiorowski and Dean, 2003). PTAC contains two factors, importin α and importin β. Importin α acts as an adaptor by recognizing the NLS cargo and then binds to importin β. Importin β is the carrier component of PTAC and targets the complex to the nuclear pore by binding to nuclear pore proteins collectively termed nucleoporins (Görlich et al., 1995; Bednenko et al., 2003). Once translocated to the nucleus, the complex is disassembled and importin α and importin β are recycled back to the cytoplasm (for a recent review of nucleocytoplasmic transport, see Terry et al., 2007).

Because any piece of DNA placed between T-DNA borders can be translocated from Agrobacterium into plant nuclei, information directing nuclear import of the T-complex must reside in the accompanying protein constituents, including VirD2 and VirE2. Both proteins can be imported into plant nuclei (Herrera-Estrella et al., 1990; Howard et al., 1992; Tinland et al., 1992; Citovsky et al., 1994; Mysore et al., 1998; Ziemienowicz et al., 2001). In VirD2, a highly conserved single cluster of four amino acids at the N terminus (R-K-G-K) and two short stretches of four to five residues at the C terminus resemble prototypic monopartite and bipartite NLSs, respectively. The close proximity of the N-terminal NLS to Tyr-29, to which T-DNA links (Vogel and Das, 1992), is likely to mask its in vivo nuclear import function. In isolation from the rest of the VirD2 protein, the C-terminal bipartite NLS can direct reporter protein fusions to plant, yeast, and animal nuclei (Howard et al., 1992; Shurvinton et al., 1992; Tinland et al., 1992; Rossi et al., 1993; Citovsky et al., 1994; Guralnick et al., 1996; Mysore et al., 1998).

VirE2 contains two central bipartite NLSs. Each of these can direct reporter protein fusions into plant nuclei individually, but efficient nuclear import requires both (Citovsky et al., 1992, 1994). Zupan et al. (1996) reported the nuclear localization of VirE2 complexed with ssDNA in microinjected plant cells. However, Ziemienowicz et al. (2001) claimed that nuclear import of large oligonucleotides in permeabilized plant cells required both VirE2 and VirD2 activity. Because the ssDNA binding and NLS domains of VirE2 overlap, VirE2 complexed with T-DNA may be unable to function in nuclear import of the T-complex (Citovsky et al., 1992; Ziemienowicz et al., 2001). At present, whether both VirD2 and VirE2 are required for nuclear import of the T-complex remains controversial.

Several studies have suggested the involvement of importin α in Agrobacterium-mediated transformation. Nuclear import of VirD2 and VirE2 proteins, and of synthetically generated T-complexes, can be inhibited by nonhydrolyzable GTP analogs (Zupan et al., 1996; Ziemienowicz et al., 2001), suggesting use of the importin pathway. Ballas and Citovsky (1997) and Bako et al. (2003) identified three Arabidopsis thaliana importin α members that interact with VirD2 in yeast. Because importin α is encoded by a multigene family in Arabidopsis, it is not clear whether multiple importin α isoforms or a specific importin α isoform is required for Agrobacterium-mediated transformation. Ballas and Citovsky (1997), however, were unable to demonstrate the interaction of VirE2 with Arabidopsis KAPα (also known as At KAPα; Ballas and Citovsky, 1997) in yeast. Rather, Tzfira et al. (2001) suggested that nuclear import of VirE2 is mediated through VIP1 (for VirE2-Interacting Protein1), an Arabidopsis protein belonging to a basic Leu zipper family. These authors showed that transgenic tobacco (Nicotiana tabacum) plants that express a VIP1 antisense gene are recalcitrant to transformation, possibly because of a deficiency in nuclear targeting of VirE2 but not of VirD2. Li et al. (2005) further showed the importance of VIP1 in T-DNA integration as well as nuclear import.

In this study, we investigated the role of seven Arabidopsis importin α isoforms in Agrobacterium-mediated transformation. Using yeast two-hybrid analysis, plant bimolecular fluorescence complementation (BiFC), and in vitro interaction assays, we show that multiple importin α isoforms can interact not only with VirD2 but also with VirE2. However, only mutation of IMPa4, but not of the other importin α genes investigated, resulted in a transformation deficiency of the mutant plants.

RESULTS

Multiple Arabidopsis Importin α Isoforms Interact with VirD2 and VirE2

The Arabidopsis genome contains nine genes that encode importin α proteins (Figure 1; see Supplemental Data Set 1 online). We initially utilized a yeast two-hybrid approach to investigate the interaction of four of these importin α isoforms (KAPα [IMPa-1], IMPa-2, IMPa-3, and IMPa-4) (Figure 1; see Supplemental Figure 1 online) with VirD2 and VirE2. In these assays, the prey plasmid pGAD424 expressed either VirD2 or VirE2 as a translational fusion with the Gal4 activation domain. Correspondingly, the bait plasmid pSTT91 expressed importin α isoforms 1 to 4 as translational fusions with the LexA DNA binding domain. β-Galactosidase activity in these yeast strains was quantified using o-nitrophenyl β-d-galactoside (ONPG) assays. All four importin α isoforms tested interacted with both VirD2 and VirE2 (Figures 2A and 2C).

Figure 1.

Cladogram of the Nine Importin α Proteins Investigated in This Study.

The genes encoding the corresponding importin α proteins are indicated in parentheses. Alignments were performed using ClustalW. The alignment information was used to generate a cladogram using MegAlign 7.1 (DNASTAR) and a bootstrap value of 1000, with seed of 111. Asterisks denote the importin α isoforms investigated in this study. Sequences used to generate the cladogram are presented in Supplemental Data Set 1 online.

Figure 2.

KAPα, IMPa-2, IMPa-3, and IMPa-4 Interact with VirE2 and VirD2.

(A) and (C) In yeast two-hybrid assays, yeast strains containing various bait/prey combinations were tested for LacZ activity using liquid ONPG as a β-galactosidase substrate. The black bars indicate LacZ activity in yeast strains containing importin α bait/VirD2 prey (A) or importin α bait/VirE2 prey (C) combinations. The gray bars show the LacZ activity in the yeast strains containing importin α bait/nonspecific GTPase prey. Error bars represent se obtained with three biological replicates.

(B) and (D) KAPα, IMPa-2, IMPa-3, and IMPa-4 bind to VirD2 and VirE2 in vitro. GST-tagged importin α proteins and GST alone were linked to separate glutathione–Sepharose columns and treated with lysates from E. coli expressing VirD2 (B) or VirE2 (D). Bound proteins were eluted, and equimolar fraction volumes were analyzed by immunoblot using either rabbit anti-octopine VirD2 (B) or anti-nopaline VirE2 (D) polyclonal antiserum as the primary antibody (1:1000 dilution) and goat anti-rabbit IgG conjugated to horseradish peroxidase as the secondary antibody (1:1000 dilution). Lane 1, VirD2 (B) or VirE2 (D) crude lysate. Lanes 2 to 6, eluted fractions from different columns: lane 2, GST-KAPα; lane 3, GST-IMPa-2; lane 4, GST-IMPa-3; lane 5, GST-IMPa-4; lane 6, GST. The asterisk represents a lower molecular weight, partially degraded VirE2 protein previously reported in the lysates of VirE2-expressing E. coli cells (Dombek and Ream, 1997).

In vitro glutathione-S-transferase (GST) pull-down assays further demonstrated direct protein–protein interactions of KAPα, IMPa-2, IMPa-3, and IMPa-4 with both VirD2 and VirE2 (Figures 2B and 2D). Our VirD2 interaction results confirm and extend earlier reports of Ballas and Citovsky (1997) and Bako et al. (2003) with regard to KAPα and IMPa-3/IMPa-4 interaction with VirD2, respectively. Our results further indicate that multiple Arabidopsis importin α isoforms can recognize VirE2, an interaction not detected by Ballas and Citovsky (1997).

Mapping VirD2 and VirE2 Domains Important for Interaction with Importin α

VirD2 has monopartite- and bipartite-type NLSs located in the N-terminal and C-terminal regions of the protein, respectively (Figure 3A) (Herrera-Estrella et al., 1990; Howard et al., 1992; Shurvinton et al., 1992; Tinland et al., 1992; Rossi et al., 1993; Mysore et al., 1998). VirE2 has two central bipartite NLSs (NLS1 and NLS2) (Citovsky et al., 1992; Dombek and Ream, 1997) (Figure 3C). We generated a series of VirD2 and VirE2 deletion mutants (Figures 3A and 3C) and tested their interaction with IMPa-4 in yeast (Figures 3B and 3D). We found that a 46–amino acid domain located between the two C-terminal NruI sites (R-372 to R-417) of VirD2 is important for interaction with IMPa-4 (Figure 3B). This domain of VirD2 includes the C-terminal bipartite-type NLS (KRPR-[X]₉-RKRAR). Interestingly, the mutant VirD2 fusion protein containing a precise deletion of the bipartite NLS residues (VirD2ΔNLS) also interacted with IMPa-4, albeit at a slightly lower strength (as indicated by units of β-galactosidase activity) than did the full-length VirD2 prey fusion (Figure 3B).

Figure 3.

IMPa-4–Interacting Domains of VirD2 and VirE2.

(A) Schematic representation of the various VirD2 deletion mutants used. VirD2 protein or its truncated derivatives are shown as gray boxes. VirD2 contains two NLS, a N-terminal monopartite NLS (stippled box) and a C-terminal bipartite NLS (black boxes). Thin lines indicate deleted regions. The N- and C-terminal NLS residues are underlined.

(B) Results of interaction assays in yeast with IMPa-4 bait and the various VirD2 deletion mutant preys. VirD2 interacts with IMPa-4 via residues Arg-372 to Arg-417. The LacZ activity was measured by liquid ONPG assays. Error bars represent se of three biological replicates.

(C) Schematic representation of the various VirE2 deletion mutants used. VirE2 protein or its truncated derivatives are shown as gray boxes. VirE2 contains two central bipartite NLSs (black boxes). The NLS motifs are underlined.

(D) Results of interaction assays in yeast with IMPa-4 bait and the various VirE2 deletion mutant preys. VirE2 interacts with IMPa-4 via its two central NLS domains and also via the last 158 amino acids in its C terminus. Full-length and various deletion constructions of VirD2 preys, VirE2 preys, or the nonspecific GTPase prey were tested for interaction in yeast with IMPa-4 as a bait. The LacZ activity was measured by liquid ONPG assays. The x axis indicates the various bait plasmids used. Error bars represent se of three biological replicates.

Interaction assays of IMPa-4 with the various VirE2 deletion proteins indicated that VirE2 interacts in yeast with IMPa-4 in a NLS dose-dependent manner. Yeast strains expressing the IMPa-4 bait fusion protein showed higher β-galactosidase activity with a truncated VirE2 protein containing both NLSs (N-ter 2-349) than with a mutant VirE2 protein containing only NLS1 (N-ter 2-242; Figure 3D). Surprisingly, a C-terminal 158–amino acid domain of VirE2 also interacted with IMPa-4 (Figure 3D). This domain of VirE2 does not contain any signature NLS sequence and overlaps with the ssDNA binding domain of VirE2.

We also performed in vitro pull-down assays using full-length and various deletion derivatives of VirD2 and VirE2 expressed as GST fusions and T7 epitope-tagged IMPa-4 (see Supplemental Figure 2 online). These assays further supported the observations obtained using yeast two-hybrid assays. Taken together, these data indicate that for both VirD2 and VirE2, the IMPa-4–interacting domains include but are not restricted to previously identified NLS residues (Citovsky et al., 1992; Howard et al., 1992).

SSB and Importin α Binding of VirE2 Are Not Mutually Exclusive

Based on the partial overlap observed between the SSB domain and the NLSs of VirE2, several groups have argued that, once bound to T-strands, the NLSs of VirE2 may be inaccessible to nuclear import receptors and hence may not have a direct role in the nuclear import of the T-complex (Citovsky et al., 1992; Ziemienowicz et al., 2001). To investigate this, we performed gel mobility shift and supershift assays. We found that VirE2-bound ssDNA migrates at lower mobility and in a VirE2 dose-dependent manner through polyacrylamide gels in comparison with free ssDNA. Approximately 30 ng of His-tagged VirE2 was required to saturate 5 ng of labeled ssDNA (see Supplemental Figure 3 online). Furthermore, T7-tagged IMPa-4 decreased the mobility of a preformed ssDNA-VirE2 complex in a dose-dependent manner (Figure 4A, compare lane 2 with lanes 3 to 8). T7-tagged IMPa-4 protein did not bind to free ssDNA at the highest concentration tested (Figure 4A, lane 9). These results indicate that ssDNA-bound VirE2 protein maintains a conformation necessary for binding importin α.

Figure 4.

SSB and Importin α Binding Activities of VirE2 Are Not Mutually Exclusive.

(A) ssDNA-bound VirE2 protein binds importin α in vitro. ³²P end–labeled ssDNA complexed with 30 ng of His-VirE2 was incubated with increasing amounts (20 to 400 ng; amounts used are indicated below each lane) of T7–IMPa-4 protein. The samples were subjected to nondenaturing PAGE and visualized by autoradiography. Lane 1, labeled ssDNA; lanes 2 to 8, labeled ssDNA (5 ng) + VirE2 (30 ng) with increasing amounts of T7–IMPa-4 (20 to 400 ng); lane 9, labeled ssDNA (5 ng) + 400 ng of T7–IMPa-4.

(B) An importin α-VirE2 complex binds ssDNA in vitro. A total of 400 ng of IMPa-4 was preincubated with 30 ng of VirE2, and 5 ng of ³²P end–labeled labeled ssDNA was added before electrophoresis. Lane 1, labeled ssDNA; lane 2, labeled ssDNA (5 ng) + VirE2 (30 ng); lane 3, preincubated T7–IMPa-4 (400 ng) + VirE2 (30 ng) followed by 5 ng of labeled ssDNA; lane 4, labeled ssDNA (5 ng) + 400 ng T7–IMPa-4.

We performed an additional in vitro gel mobility supershift assay in which His-tagged VirE2 was first incubated with an excess of T7-tagged IMPa-4. Following complex formation, we added labeled ssDNA. Figure 4B shows that VirE2 in a preformed IMPa-4-VirE2 complex can bind ssDNA (cf. lanes 2 and 3). These results support the previous observation that importin α–bound VirE2 maintains a conformation necessary to bind ssDNA. The SSB and importin α binding activities of VirE2, therefore, are not mutually exclusive.

Several Arabidopsis Importin α Mutants Are Not Impaired in Agrobacterium-Mediated Transformation

Previous studies suggested that several importin α proteins may interact with VirD2 (Ballas and Citovsky, 1997; Bako et al., 2003). In order to determine the functional relevance of these interactions and whether specific importin α members are essential for Agrobacterium-mediated transformation, we investigated the transformation proficiency of Arabidopsis lines with T-DNA disruptions in some of the individual importin α genes. We identified plants homozygous for T-DNA insertions in KAPα (Salk_001092), IMPa-2 (Salk_017914), and IMPa-3 (Salk_025919). RT-PCR analysis indicated that full-length importin α transcripts were not detectable in the corresponding mutants (see Supplemental Figures 4A to 4C online). We tested the susceptibility of these mutant plants to transient (transfer of T-DNA not requiring integration) and stable Agrobacterium-mediated root transformation. T-DNA insertion lines individually harboring disruptions in each of these three genes were as susceptible to transformation by Agrobacterium as wild-type plants (Figure 5A). Figure 5B shows representative tumors observed on the roots of wild-type and importin α mutant plants. These data indicate that KAPα, IMPa-2, and IMPa-3 are not individually essential for Agrobacterium-mediated root transformation.

Figure 5.

Arabidopsis Importin α Mutants kapα, impa-2, and impa-3 Show Wild-Type Agrobacterium-Mediated Transformation Efficiencies.

(A) Root segments from Arabidopsis mutants homozygous for the disruption of KAPα, IMPa-2, or IMPa-3 and wild-type control plants were infected with the tumorigenic strain Agrobacterium A208 (for stable transformation assays; multipatterned bars) or Agrobacterium strain At849 (containing a plant-active gusA gene for transient transformation assays; stippled black bars). Stable transformation efficiency (indicated by the presence of tumors) was scored and categorized after 1 month. The sizes and colors of the tumors are indicated by bar patterns. The larger and more green the tumors, the more highly susceptible the plant tissue is for transformation. For transient transformation, root segments were stained with 5-bromo-4-chloro-3-indolyl-β-glucuronic acid (X-gluc) at 6 d after infection, and the percentage of roots showing GUS activity was calculated. For each assay, 10 to 15 plants and >80 segments per plant were used. Error bars represent se for the 10 to 15 plants assayed for each treatment.

(B) Representative plates showing tumorigenesis efficiencies (stable transformation) and tumor morphology on wild-type and importin α mutant root segments. Bar = 2 cm.

An Arabidopsis impa-4 Mutant Is a rat Mutant

Because the Salk collection did not contain an impa-4 mutant, we screened the Feldmann T-DNA insertional mutant collection (ecotype Wassilewskija [Ws-2]) using a reverse genetics–based approach (Zhu et al., 2003). We identified a homozygous mutant containing a T-DNA disruption in the seventh intron of the IMPa-4 gene (Figure 6A). RT-PCR analysis revealed only extremely low levels of IMPa-4 transcript in this mutant (see Supplemental Figure 3D online). To quantify the amount of IMPa-4 protein synthesized in this mutant line, we generated a peptide antibody against amino acids 6 to 19. This region of the protein shares no significant homology with the other eight Arabidopsis importin α proteins, nor to other Arabidopsis proteins. Supplemental Figure 5 online shows that this peptide antibody reacts exclusively with IMPa-4. Using this specific antibody, we probed protein blots of extracts from leaf or root tissue from 4-week-old wild-type (ecotypes Columbia and Ws-2) and impa-4 mutant plants. Figure 6B shows that IMPa-4 protein was detectable only in wild-type plants but not in impa-4 plants. These results indicate that the T-DNA insertion in the seventh intron of IMPa-4 does not permit the accumulation of importin α-4 protein.

Figure 6.

An Arabidopsis impa-4 Mutant Is a rat Mutant.

(A) Schematic representation of the IMPa-4 gene. The Arabidopsis impa-4 mutant contains a T-DNA insertion in the seventh intron. A part of the T-DNA/plant DNA junction sequence is shown. The sequence corresponding to the T-DNA right border primer TR1 is underlined. The sequence for intron 7 (shaded box) is followed by a portion of exon 8 sequence. The position of the T-DNA right border TR1 primer is indicated with an arrow.

(B) Protein blot analysis of extracts isolated from leaf and root tissue of Arabidopsis ecotypes Ws-2 (lanes 1 and 4), Columbia (lanes 2 and 5), and impa-4 mutant plants (lanes 3 and 6) with an antibody raised against an IMPa-4 peptide. Recombinant GST-tagged IMPa-4 protein was used as a control (lane 7). Top, immunostain; bottom, portion of a Coomassie blue–stained gel showing equal protein loading within each tissue sample.

(C) The Arabidopsis impa-4 mutant shows reduced levels of transient and stable Agrobacterium-mediated root transformation. Root segments from wild-type and impa-4 mutant plants were infected either with the tumorigenic strain Agrobacterium A208 (for stable transformation assays; multipatterned bars) or Agrobacterium At849 (for transient transformation assays; gray bars). For stable transformation, the tumors were scored after 1 month. For transient transformation, root segments were stained with X-gluc at 6 d after infection, and the percentage of roots showing GUS activity as a result of infection by Agrobacterium At849 was calculated. For each assay, 10 to 15 plants and >80 segments per plant were used. Error bars represent se. Representative plates of tumorigenesis assays are shown. White bars = 1 cm.

Transient and stable transformation assays showed that impa-4 is a rat mutant. impa-4 mutant plants formed tumors at an efficiency sevenfold to eightfold lower than that of wild-type plants (Figure 6C). Whereas root segments from infected wild-type plants formed mostly large green teratomas, root segments from impa-4 mutant plants formed mainly small yellow tumors. impa-4 mutant plants have a greater than threefold reduction in transient transformation efficiency compared with wild-type plants (Figure 6C). Because transient expression requires nuclear import and conversion of the incoming ssT-DNA to a double-stranded transcription-competent form, the lower transient transformation efficiency of impa-4 mutant plants indicates a hindrance in the transformation process at a step most likely prior to nuclear import of the T-complex. Thus, IMPa-4 is important for an early step in Agrobacterium-mediated root transformation.

impa-4 RNA Interference Plants Show a rat Phenotype

To confirm that IMPa-4 plays a role in Agrobacterium-mediated transformation, we generated impa-4 RNA interference (RNAi) plants. The RNAi vector used to make these plants contained a hairpin construction of a region of the IMPa-4 gene, mostly within the 3′ untranslated region (UTR), that does not show homology to other Arabidopsis importin α genes. We tested root segments from numerous T1 generation transgenic plants for transformation competence using the tumorigenic strain Agrobacterium A208. Figure 7 shows that silencing of IMPa-4 by RNAi decreases susceptibility to Agrobacterium-mediated transformation for many of the RNAi lines. This result supports our previous findings that IMPa-4 is involved in Agrobacterium-mediated transformation.

Figure 7.

Inhibition of Agrobacterium-Mediated Root Transformation by Expression of an RNAi Construct Targeted Specifically against IMPa-4.

Root segments from wild-type (ecotype Ws-2), impa-4 mutant, and RNAi plants were inoculated with Agrobacterium A208 at 10⁷ colony-forming units/mL, and tumorigenesis assays were conducted. Tumors were scored after 6 weeks. Each bar indicates data from an individual plant, and average values for each set are indicated. Numbers above the error bars indicate the number of plants analyzed. Error bars indicate se.

Arabidopsis Importin α Transcripts Are Differentially Expressed in Roots

In higher eukaryotes, importin α gene members demonstrate tissue-specific expression patterns and levels. This specificity likely reflects dynamic and regulated nucleocytoplasmic transport demands of individual tissues (Köhler et al., 1997). We performed real-time RT-PCR using primers specific to various importin α isoforms to investigate their expression levels in Arabidopsis. In roots, the investigated importin α isoforms display modest but significant differences in expression levels (Figure 8). IMPa-4 transcripts accumulate in roots to levels 1.5- to 2-fold lower than those of KAPα and IMPa-2, respectively. Similarly, IMPa-3 transcripts accumulate to levels threefold to fourfold lower than those of KAPα and IMPa-2, respectively. We also quantified the transcript levels of various importin α isoforms in flowers and leaves (see Supplemental Figure 6 online). The relative levels of expression in these tissues paralleled those seen in roots.

Figure 8.

Differences in Transcript Levels of Various Importin α Isoforms in Wild-Type and impa-4 Mutant Roots as Determined by RT-PCR.

Total RNA from wild-type and impa-4 mutant roots was reverse-transcribed using oligo(dT) primers and then subjected to real-time PCR analysis using various importin α gene-specific primers. RT-PCR with ubiquitin primers was performed as a control to correct for sample concentrations. Gray and black bars represent the transcript levels of the indicated importin α isoforms in wild-type and impa-4 mutant roots, respectively. Error bars represent se of four technical replicates.

We performed a quantitative RT-PCR analysis to investigate whether the loss of IMPa-4 transcripts in the impa-4 mutant affects transcript levels of the other importin α isoforms. Figure 8 shows that in this mutant, levels of IMPa-4 transcripts are greatly reduced (∼250-fold) compared with those in wild-type plants. While the transcript levels of KAPα and IMPa-2 were comparable in the impa-4 mutant and wild-type roots, we detected a slight increase in IMPa-3 transcript levels in the impa-4 mutant relative to the wild type. Thus, the reduction of IMPa-4 transcripts in the impa-4 mutant is not compensated for by an increase in accumulation of the other importin α transcripts investigated.

Overexpression of Heterologous Importin α Isoforms Rescues the rat Phenotype of impa-4 Mutant Plants

Although multiple isoforms of importin α interact with VirD2 and VirE2 in yeast and in vitro, only the impa-4 mutant is a rat mutant. Thus, the other importin α isoforms are not redundant with IMPa-4 with respect to Agrobacterium-mediated transformation. This lack of redundancy could derive from several causes. The various importin α proteins could have different properties in planta (i.e., they may have different cargo preferences for transporting the T-complex to the nucleus). Alternatively, the various importin α proteins may be expressed at different levels, or with different patterns, in root cells. To address the apparent discrepancy between the interaction and the transformation assays, we hypothesized that if the level or expression pattern of IMPa-4 was important for transformation, then constitutive overexpression of not only IMPa-4 but also of heterologous importin α isoforms should restore transformation competence to impa-4 mutant plants. If, however, IMPa-4 plays a unique role in Agrobacterium-mediated transformation, then overexpression of only IMPa-4 but not of other importin α isoforms would phenotypically complement the impa-4 mutant.

Therefore, we generated individual transgenic lines in the impa-4 mutant background overexpressing either an IMPa-4 cDNA or cDNAs of other importin α isoforms. Transient and stable Agrobacterium-mediated transformation assays on independent lines of plants indicated that overexpression of not only IMPa-4 but also of KAPα, IMPa-2, IMPa-3, IMPa-6, IMPa-7, and IMPa-9 reversed the rat phenotype of the impa-4 mutant (Figure 9; see Supplemental Tables 1A and 1B online). These results indicate that the rat phenotype of the impa-4 mutant can be rescued by overexpression of other importin α isoforms, although IMPa-6, -7, and -9 do so less efficiently than do IMPa-1, -2, -3, and -4.

Figure 9.

Overexpression of Importin α Isoforms Rescues the Rat Phenotype of impa-4 Mutant Roots.

(A) Transgenic plants individually overexpressing At KAPα, IMPa-2, IMPa-3, IMPa-4, IMPa-6, IMPa-7, or IMPa-9 cDNAs in the impa-4 mutant background show increased susceptibility to Agrobacterium-mediated root transformation compared with parent impa-4 mutant plants. Root segments from several T2 generation transgenic plants (2 to 3 independent lines each for KAPα, IMPa-2, IMPa-3, or IMPa-4 overexpression) or T1 generation transgenic plants (14 to 15 independent plants each for IMPa-6, IMPa-7, or IMPa-9 overexpression) were infected with either the tumorigenic strain Agrobacterium A208 (for stable transformation assays; multipatterned bars) or Agrobacterium At849 (for transient transformation assays; gray bars). Infected root segments from wild-type and impa-4 mutant plants were used as positive and negative controls, respectively. For stable transformation, the tumors were scored after approximately 1 month. For transient transformation, root segments were stained with X-gluc at 6 d after infection, and the percentage of roots showing GUS activity resulting from infection by Agrobacterium At849 was calculated. N.D., not determined. Error bars indicate se.

(B) Representative plates of tumorigenesis assays. Bar = 2 cm.

Taken together, these data indicate that when the native expression levels of the tested importin α isoforms are overridden, they are functionally redundant with IMPa-4 with respect to Agrobacterium-mediated transformation.

Localization of VirD2 and VirE2 in Wild-Type and impa-4 Mutant Plants

Previous studies have shown that when transiently expressed, both VirD2 and VirE2 proteins localize to the nuclei of plant cells (Howard et al., 1992; Tinland et al., 1992; Rossi et al., 1993; Citovsky et al., 1994; Guralnick et al., 1996; Mysore et al., 1998). In order to explore the cause of the rat phenotype of the impa-4 mutant, we investigated the localization VirD2 and VirE2 in roots of impa-4 mutant and wild-type plants. We generated transgenic wild-type and impa-4 mutant plants overexpressing yellow fluorescent protein (YFP), YFP-VirE2, or VirE2-YFP fusion proteins. We were unable to generate transgenic plants expressing YFP fusions with full-length VirD2. Previous work in our laboratory indicated that transgenic plants overexpressing full-length VirD2 are nonviable (Hwang et al., 2006). Therefore, we generated transgenic plants expressing YFP fusions to the C-terminal NruI (Arg-372 to Arg-417) fragment of VirD2 shown above as the domain necessary and sufficient for interaction with IMPa-4. We denote this fusion as YFP-VirD2^*. T1 generation plants from several independent lines were analyzed for protein localization by confocal microscopy.

As expected, YFP alone localized throughout the cells of both wild-type and impa-4 mutant roots (Figures 10A to 10D). In roots of both wild-type and impa-4 mutant plants, YFP-VirD2^* localized exclusively to the nucleus, as verified by the coincidence of YFP fluorescence and nuclear staining with Hoechst 33242 (Figures 10E to 10H). These results indicate that the NruI (Arg-372 to Arg-417) fragment of VirD2 transfers foreign protein (in this case, YFP) to the nucleus. In addition, the data indicate that nuclear import of VirD2 is not compromised in roots of impa-4 mutant plants.

Figure 10.

Localization of YFP-VirD2, YFP-VirE2, and VirE2-YFP in Wild-Type and impa-4 Mutant Roots.

Single-plane confocal optical sections show YFP localization (green fluorescence) throughout cells (both in the cytoplasm and in the nucleus) in the root tip of wild-type ([A] and [B]) and impa-4 mutant ([C] and [D]) plants. YFP-VirD2 preferentially localizes to the nuclei of root tip cells of both wild-type ([E] and [F]) and impa-4 mutant ([G] and [H]) plants. YFP-VirD2 colocalizes with nuclear Hoechst 33242 stain (blue fluorescence). In both wild-type ([I] and [J]) and impa-4 mutant ([K] and [L]) root tip cells, YFP-VirE2 localizes at cell poles (arrowheads). VirE2-YFP also aggregates in the cytoplasm (arrowheads) in both wild-type ([M] and [N]) and impa-4 mutant ([O] and [P]) root tip cells. In (A) to (L), blue indicates Hoechst 33242 staining of nuclei. Bars = 15 μm in (A), (C), (E), (G), (I), (K), (M), and (O) and 50 μm in (B), (D), (F), (H), (J), (L), (N), and (P).

YFP fluorescence in roots of transgenic plants expressing either YFP-VirE2 or VirE2-YFP never localized to the nucleus. In plants expressing YFP-VirE2, the fusion protein localized preferentially to the cell poles (Figures 10I and 10J, arrowheads). Tobacco BY-2 cells transiently expressing YFP-VirE2 also showed exclusive cytoplasmic localization of YFP fluorescence; this fluorescence was frequently in strands (see Supplemental Figure 7 online). Transgenic Arabidopsis roots expressing VirE2-YFP showed somewhat punctate aggregation in the cytoplasm (Figures 10M and 10N, arrowheads). Localization of YFP-VirE2 or VirE2-YFP fusion protein in roots of impa-4 mutant plants was similar to that observed in roots of wild-type plants (Figures 10K, 10L, 10O, and 10P). These results indicate that the localization of VirE2, although nonnuclear, is unaffected by mutation of impa-4.

VirE2 can form multimers utilizing specific peptide domains (Citovsky et al., 1997; Dombek and Ream, 1997; Zhou and Christie, 1999). We used BiFC (Citovsky et al., 2006) to investigate the subcellular site of VirE2–VirE2 interaction in electroporated tobacco BY-2 cells. In BiFC, a YFP molecule is split into two sections, neither of which fluoresces on its own. Refolding of the two partial YFP moieties restores fluorescence (fluorescence complementation). Refolding can occur if each YFP moiety is affixed to proteins that themselves interact, thus bringing together the YFP peptides (Hu et al., 2002). Figure 11A and Supplemental Table 2 online show that VirE2 proteins interact with each other; interaction localizes exclusively to the cytoplasm, confirming the cytoplasmic localization of VirE2 shown in Figures 10 and 11.

Figure 11.

Interaction and Localization of VirE2 with Itself and of VirD2 and VirE2 with Importin α Proteins in Plant Cells.

BiFC constructs were electroporated into tobacco BY-2 cells, and the cells were visualized by fluorescence microscopy 48 h later. Constructs are as follows: VirE2-nYFP plus VirE2-cYFP (A); nYFP-VirD2 plus KAPα-cYFP (B); nYFP-VirD2 plus IMPa-4–cYFP (C); nYFP-VirD2 plus IMPa-7–cYFP (D); nYFP-VirD2 plus IMPa-9–cYFP (E); nYFP-VirD2 plus IMPa-4–cYFP genomic clone (F); VirE2-nYFP plus KAPα-cYFP (G); VirE2-nYFP plus IMPa-4–cYFP ([H] and [I]); VirE2-nYFP plus IMPa-7–cYFP (J); VirE2-nYFP plus IMPa-9–cYFP (K); VirE2-nYFP plus IMPa-4–cYFP genomic clone (L). In (A) and (E), blue indicates Hoechst 33242 staining of nuclei. In (B), (C), (D), (G), (J), and (K), blue indicates false-color image of the entire cell. In (F), (H), (I), and (L), red fluoresecent protein images (marking the entire cell) are merged with the YFP images. Bars = 20 μm.

Our observation that the VirE2 fusion protein localized to the cytoplasm differs from the findings of others, in which a β-glucuronidase (GUS) protein fused to the N terminus of VirE2 localized to the nucleus (Citovsky et al., 1992, 1994; Guralnick et al., 1996; Li et al., 2005). Because it is possible that protein tags may alter protein function, we tested the functionality of our VirE2 fusion proteins using root-based extracellular complementation (Otten et al., 1984) transient transformation assays on transgenic Arabidopsis plants expressing either YFP-VirE2 or VirE2-YFP. We inoculated root segments of these plants with Agrobacterium At1565, a virE2⁻ mutant strain containing the T-DNA binary vector pBISN1 (Narasimhulu et al., 1996). As a positive control for extracellular complementation, we used a transgenic Arabidopsis line expressing untagged VirE2. The results of these assays showed that roots of transgenic plants expressing VirE2-YFP, but not YFP-VirE2, were transformable by At1565 (Figure 12). The transient transformation efficiency was comparable to that of the positive control plants. As expected, roots of wild-type Arabidopsis plants were not transformed by the virE2⁻ mutant Agrobacterium strain. These results indicate that, with respect to Agrobacterium-mediated transformation, a VirE2-YFP fusion protein retains functionality whereas a fusion to the N terminus of VirE2 (YFP-VirE2) does not.

Figure 12.

Wild-Type and impa-4 Mutant Roots Overexpressing VirE2-YFP, but Not YFP-VirE2, Complement a virE2 Mutant Agrobacterium in Transient Transformation Assays.

Several independent T1 transgenic plants overexpressing YFP-VirE2 or VirE2-YFP in the wild type or in the impa-4 mutant background were infected with Agrobacterium At1565. Root segments were stained with X-gluc at 6 d after infection, and the percentage of roots showing GUS activity resulting from infection by Agrobacterium At89 was calculated. Root segments from wild-type plants overexpressing untagged VirE2 were used as positive controls (left bar). Nontransgenic wild-type plants and impa-4 mutant plants were used as negative controls. For each assay, >80 segments per plant were used. Error bars represent se.

Several Importin α Isoforms Can Interact with VirD2 and VirE2 in Planta

We used BiFC to examine whether interactions between virulence proteins D2 or E2 and various importin α proteins could also occur in plants. Supplemental Table 2 online and Figures 11B to 11E show that nYFP-VirD2 interacted with KAPα–cYFP, IMPa-4–cYFP, IMPa-7–cYFP, and IMPa-9–cYFP. In these experiments, the proteins were expressed from cDNA clones under the regulation of a cauliflower mosaic virus (CaMV) 35S promoter. The interacting protein partners localized to the nucleus, confirming the localization results shown in Figures 10E and 10F. By contrast, VirE2-nYFP interacted with these importin α proteins mostly in the cytoplasm (Figures 12G to 12K; see Supplemental Table 3 online), confirming the results shown in Figures 10M and 10N. However, YFP fluorescence sometimes localized to the nucleus, especially with IMPa-4 (Figure 11I). These results suggest that IMPa-4 may be more efficient at delivering VirE2 to the nucleus than KAPα is, again indicating the importance of IMPa-4 in Agrobacterium-mediated transformation.

It is possible that overexpression of IMPa-4 from the CaMV 35S promoter may result in mislocalization of the interacting protein partners. Therefore, we constructed a plasmid that would express IMPa-4–cYFP at native levels. This construction contained the IMPa-4 promoter, introns, exons, and 3′ UTR. Figures 11F and 11L show that when expressed from this plasmid, IMPa-4 again interacted with VirD2 in the nucleus and with VirE2 predominantly in the cytoplasm.

In the BiFC experiments described above, we could visualize the interaction of VirD2 or VirE2 with only one importin α isoform at a time. It is possible, however, that these Vir proteins interact simultaneously with several importin α isoforms. Therefore, we conducted multicolor BiFC experiments, as described by Hu and Kerppola (2003). In these experiments, the Vir protein (either VirD2 or VirE2) was tagged with cCFP (for cyan fluorescent protein), and the importin α partners were tagged with either nVenus or nCerulean (Shyu et al., 2006). Expression cassettes encoding the tagged Vir protein and both of the tagged importin α proteins were transiently transfected into tobacco BY-2 cells. Interaction of the Vir protein with importin α tagged with nVenus would generate yellow/green fluorescence, whereas interaction of the Vir protein with importin α tagged with nCerulean would generate blue fluorescence. To indicate whether cells were competent to receive DNA, we cotransfected the cells with an expression cassette that expressed mCherry; these cells would fluoresce red. As a control, we swapped nGFP (for green fluorescent protein) derivative fluorescent tags on the importin α proteins.

Figures 13A to 13H show that VirD2 could simultaneously interact with both IMPa-1 and IMPa-4. In each instance, interaction occurred in the nucleus, validating the results shown in Figures 11B and 11C. The interaction of VirE2 with IMPa-1 and IMPa-4 was more complex. Interaction of VirE2 with IMPa-1 always occurred in the cytoplasm (Figures 13J and 13N), confirming the results shown in Figure 11G. However, interaction of VirE2 with IMPa-4 occurred both in the nucleus and in the cytoplasm (Figures 13K and 13O), confirming the results shown in Figures 11H and 11I. Swapping the fluorescent tags on the two importin α proteins did not alter these results. Thus, VirE2 localized to the nucleus only when IMPa-4, but not IMPa-1, was overexpressed. These results again suggest the importance of IMPa-4 in the transformation process.

Figure 13.

Multicolor BiFC Indicates the Site of Interaction of VirD2 and VirE2 with IMPa-1 (KAPα) and IMPa-4.

VirD2-cCFP ([A] to [H]) and VirE2-cCFP ([I] to [P]) were used as baits to interact simultaneously with IMPa-1 or IMPa-4. mCherry indicates transfected BY-2 protoplasts; red fluorescence localizes both to the cytoplasm and to the nucleus. The various combinations of labeled proteins coexpressed in the protoplasts are indicated above each series of four panels. Cells were visualized using an epifluorescence microscope: (A), (E), (I), and (M), red (mCherry) channel; (B), (G), (J), and (O), yellow/green (Venus) channel; (C), (F), (K), and (N), blue (Cerulean) channel; (D), (H), (L), and (P), overlay of yellow and blue channels. Bars = 20 μm.

DISCUSSION

Several Importin α Isoforms Interact with both VirD2 and VirE2

Importin α proteins in Arabidopsis are encoded by a multigene family consisting of nine members (Figure 1; see Supplemental Figure 1 online). In this study, VirD2 and VirE2 interacted in yeast and in vitro with each of four importin α isoforms tested (KAPα and IMPa-2, -3, and -4; Figure 2). VirD2 and VirE2 additionally interacted in plants with IMPa-7 and IMPa-9 (Figure 11). The interaction of VirD2 with KAPα, IMPa-3, and IMPa-4 in yeast is consistent with the reports of Ballas and Citovsky (1997) and Bako et al. (2003). Our finding that VirE2 also interacts with these importin α isoforms contrasts with the observations of Ballas and Citovsky (1997), who failed to detect an interaction between VirE2 and KAPα. Ballas and Citovsky (1997) expressed VirE2 as a bait protein and KAPα as a prey protein, whereas we expressed VirE2 as a prey and the various importin α isoforms as baits. It is possible that when VirE2 is used as a prey but not as a bait the fusion may maintain a conformation necessary to interact with importin α proteins. Interaction of VirD2 and VirE2 with IMPa-4 in plants cannot be attributed to overexpression of the importin α protein, because expression of tagged IMPa-4 at normal levels from its native gene context also indicated interaction with these virulence proteins (Figure 11F).

To identify domains of VirD2 and VirE2 that are important for interaction with importin α, we tested the interaction of a series of VirD2 and VirE2 deletion mutants with IMPa-4 as the prototype importin α member (Figure 3). We chose this isoform for these experiments because mutational analysis of the various importin α genes indicated that in Arabidopsis, IMPa-4 is the only isoform tested that is required for Agrobacterium-mediated root transformation (Figures 5 and 6). The Agrobacterium VirD2 protein contains N-terminal and C-terminal regions of amino acids resembling those found in classical monopartite and bipartite NLSs, respectively (Robbins et al., 1991). Based on various mutations, the N-terminal NLS was proposed not to be essential for the nuclear import of T-strands (Shurvinton et al., 1992; Rossi et al., 1993). This NLS does not interact with IMPa-4 (Figure 3B). Sequences resembling classical NLSs are not always recognized by importin α (Neumann et al., 1997; Wang et al., 1997).

The VirD2 C-terminal bipartite NLS, which has been suggested to play a role in the nuclear import of T-strands (Rossi et al., 1993), was not essential for interaction with importin α. Truncated VirD2 with a precise deletion of the C-terminal NLS residues (deletions of KRPR and KRAR) still interacted in yeast with IMPa-4, although at a slightly lower level than did wild-type VirD2 (Figures 3A and 3B). This observation is in contrast with that of Ballas and Citovsky (1997), who reported a complete abolition of KAPα interaction with a VirD2 protein containing a precise deletion of the C-terminal bipartite NLS residues (deletions of KRPR and RKRER). Although the reason for these differences is not clear, the use of slightly different deletion mutants may have resulted in these contradictory observations. Our assays showed that an internal deletion of 45 amino acid residues (Arg-372 to Arg-417) that include the bipartite NLS (VirD2ΔNruI) abolished interaction of the mutant VirD2 protein with IMPa-4. This result suggests that, whereas the C-terminal NLS of VirD2 may modulate the strength of interaction with importin α, additional adjoining residues are also important for this interaction. Our interaction results are consistent with previous reports showing that virulence of a mutant Agrobacterium strain with precisely the same deletion of the VirD2 C-terminal NLS residues (VirD2ΔNLS) shows little reduction in transformation (Shurvinton et al., 1992; Mysore et al., 1998). By contrast, a strain expressing the same VirD2ΔNruI protein that we examined in this report was avirulent (Shurvinton et al., 1992) or severely compromised in virulence (Rossi et al., 1993). Taken together, our results suggest that a 46–amino acid region (Arg-372 to Arg-417) of octopine-type VirD2 that includes the C-terminal NLS and adjoining amino acids is important for nuclear import of T-strands, presumably through interactions with importin α.

Interaction assays of various VirE2 deletion mutants with IMPa-4 in yeast further support the specificity of the VirE2 interaction with importin α (Figure 3). Interaction of VirE2 proteins containing both bipartite NLSs results in a higher level of β-galactosidase activity than does VirE2 containing only one NLS. This observation is in agreement with that of Citovsky et al. (1992), who reported that efficient nuclear import of VirE2 requires both NLSs.

Importin α May Mediate T-Strand Nuclear Import via Interactions with VirE2

The T-complex is proposed to consist of multiple molecules of VirE2 coating the T-strand, which is covalently linked at its 5′ terminus to VirD2. VirE2 presumably protects T-strands from nucleolytic attack in the plant (Yusibov et al., 1994; Rossi et al., 1996). The partial overlap of the ssDNA binding domain and the second NLS of VirE2 raises the question of whether VirE2 NLSs are accessible to plant nuclear import components once VirE2 protein is bound to the T-strand. Our in vitro gel mobility supershift results (Figure 4) suggest that the ssDNA and importin α binding activities of VirE2 are not mutually exclusive and that VirE2 can mediate the nuclear import of T-strands in an importin α–dependent manner.

Tzfira et al. (2001) identified VIP1, an Arabidopsis basic Leu zipper protein, which interacts with VirE2. VIP1 also interacts with KAPα (Tzfira et al., 2002). Because they were unable to demonstrate interaction of VirE2 with KAPα, these authors suggested that VIP1 may serve as an adaptor molecule to facilitate the import of VirE2-bound T-strands into the nucleus. Our data demonstrate, however, that VirE2 can directly bind all tested importin α isoforms (KAPα and IMPa-2, -3, -4, -7, and -9; Figures 2 and 11). It is possible that VirE2 may utilize several cellular mechanisms for nuclear import, thereby creating additional opportunities for T-complex entry into the nucleus.

With Regard to T-Complex Nuclear Import, to What Extent Are Importin α Isoforms Redundant?

The interaction of VirD2 and VirE2 with several Arabidopsis importin α isoforms suggests redundancy. However, disruption of only IMPa-4, but not KAPα, IMPa-2, or IMPa-3, reduced Agrobacterium-mediated root transformation of Arabidopsis. Relative to wild-type plants, the impa-4 mutant expresses greatly reduced amounts of IMPa-4 mRNA and undetectable levels of IMPa-4 protein (Figures 6 and 8). Disruption of IMPa-4 expression was not compensated for by increased expression of the other importin α genes examined (Figure 8). Therefore, we investigated whether manipulation of native expression patterns and levels of heterologous importin α genes, by overexpression from a constitutive CaMV 35S promoter, could phenotypically complement the transformation deficiency of the impa-4 mutant. Overexpression not only of IMPa-4 but also of KAPα, IMPa-2, IMPa-3, IMPa-6, IMPa-7, and IMPa-9 restored transformation competence to the impa-4 mutant, showing functional redundancy among these importin α proteins (Figure 9). The rat phenotype of the impa-4 mutant, in which the other importin α genes are expressed, likely results from differing patterns of importin α gene expression, especially in those root cells most susceptible to transformation (Yi et al., 2002, 2006). Our observations are similar to those of Mason et al. (2002), who reported complementation of male sterility defects in Drosophila resulting from loss of importin α-2 by overexpression of either α-2 or of the heterologous α-1 and α-3 transgenes.

Subcellular Localization of VirD2 and VirE2 in Wild-Type and impa-4 Mutant Plants

The lack of transformation proficiency of the impa-4 mutant could result from the inability of this plant to support nuclear localization of VirD2 and/or VirE2 and, thus, the T-complex. To test this, we examined the localization of YFP-VirD2^*, YFP-VirE2, and VirE2-YFP fusions in wild-type and impa-4 mutant plants. YFP-VirD2^* localized to the nucleus in both wild-type and impa-4 mutant plants, suggesting that the rat phenotype of these plants did not result from lack of VirD2 nuclear import (Figures 10A to 10H).

We did not detect nuclear localization of either YFP-VirE2 or VirE2-YFP fusion protein in roots of wild-type or impa-4 plants (Figures 10I to 10P). Whereas YFP-VirE2 localized to the cell poles, VirE2-YFP showed cytoplasmic aggregation that sometimes resulted in punctate patterns. These results contrast with previous reports indicating that transiently expressed VirE2 (with either a fluorescent tag or as a fusion with reporter GUS or GFP proteins) localized to nuclei (Citovsky et al., 1992, 1994; Zupan et al., 1996; Tzfira et al., 2001; Ziemienowicz et al., 2001). Our results, however, are consistent with the observation that VirE2 can associate with membranes (Dumas et al., 2001; Duckely et al., 2005) and remains in the cytoplasm (Duckely and Hohn, 2003). Indeed, Grange et al. (2008) recently showed that when tobacco BY-2 cells were bombarded with a VirE2-hemagglutinin expression construct, VirE2 localized exclusively to the cytoplasm. Our results further demonstrate that VirE2 localization in impa-4 mutant roots does not differ from that observed in wild-type plants.

Several reports delineating the functional domains of VirE2 suggested that some mutations or tag insertions in VirE2 may render the resulting protein either nonfunctional or may interfere with the ssDNA binding or cooperativity of VirE2 (Dombek and Ream, 1997; Simone et al., 2001). Therefore, we investigated whether the observed lack of nuclear localization of YFP-VirE2 or VirE2-YFP resulted from altered properties conferred by the YFP tag. To test for functionality of the two VirE2 fusions, we infected roots of transgenic plants expressing either YFP-VirE2 or VirE2-YFP with the virE2⁻ mutant strain Agrobacterium At1565. Our assays showed that wild-type plants expressing VirE2-YFP, but not YFP-VirE2, were susceptible to transformation by Agrobacterium At1565 at levels comparable to those of transgenic plants expressing untagged VirE2 (Figure 12). These results demonstrate that VirE2-YFP, but not YFP-VirE2, is fully functional with respect to Agrobacterium-mediated transformation, despite its inability to localize to the nucleus.

T-strands affixed to VirD2 and VirE2 are transferred independently from Agrobacterium to the plant cell (Binns et al., 1995; Lee et al., 1999; Vergunst et al., 2000), where they are likely assembled into a mature T-complex. Our observation that VirE2 localizes predominantly to the cytoplasm (Figures 11 and 13; see Supplemental Figure 6 online) is consistent with this model. BiFC indicated that cytoplasmic VirE2 associates with importin α. However, when coexpressed with IMPa-4, some of the pool of VirE2 may move to the nucleus (Figures 11 and 13). These observations may explain the role of VirE2 in protecting T-strands against nucleolytic degradation in the plant cytoplasm (Yusibov et al., 1994) and its role in the nuclear import of the T-complex (Gelvin, 1998; Ziemienowicz et al., 2001). Whether some of the cytosolic VirE2 localizes to the nucleus after forming a complex with T-strands requires further investigation.

Our findings that mutation of IMPa4 reduces Agrobacterium-mediated transformation but does not apparently affect the localization of T-complex components VirD2 and VirE2 may at first appear contradictory. We speculate that only a small portion of the cytoplasmic VirE2 population may participate in T-strand nuclear transport and that this subpopulation is below our level of detection using our imaging assays. Alternatively, VirD2 and VirE2 may behave differently when complexed with T-strands compared with our investigations that imaged free protein molecules. Finally, IMPa-4 may be responsible for nuclear targeting of other bacterial or plant proteins, which we have not investigated, that are important for transformation. Future investigations into the mechanism of T-complex assembly, cytoplasmic trafficking, and nuclear import will contribute not only to our knowledge of Agrobacterium-mediated transformation but also to a better understanding of the nuclear import of protein–nucleic acid complexes.

METHODS

Phylogenetic Analysis of the Nine Arabidopsis Importin α Proteins

Sequence searches of the Arabidopsis thaliana genome identified nine importin α isoforms. These are KAPα/IMPa-1 (At3g06720), IMPa-2 (At4g16143), IMPa-3 (At4g02150), IMPa-4 (At1g09270), IMPa-5 (At5g49310), IMPa-6 (At1g02690), IMPa-7 (At3g05720), IMPa-8 (At5g52000), and IMPa-9 (At5g03070). The proteins encoded by these genes were aligned using the ClustalW module of the MegAlign 7.1 program (DNASTAR). The sequence alignment is shown in Supplemental Figure 1 online and was used for the construction of a cladogram. In the phylogenetic tree, construction bootstrapping was performed for 1000 replicates with a seed of 111.

PCR Amplification of Arabidopsis Importin α cDNAs

Total RNA was extracted from 2- to 3-week-old plants using TRIZOL reagent (GIBCO) following the instructions provided by the manufacturer. Two micrograms of total RNA was reverse-transcribed using poly(dT) primers and AMV reverse transcriptase (Promega) to obtain single-stranded cDNA. Double-stranded cDNA synthesis was performed using the gene-specific primer sequences presented in Supplemental Table 3 online. Except for primers used to amplify IMPa-6, all forward primers contained a 5′-EcoRI restriction endonuclease site followed by sequences from the second codon. Reverse primers contained a 5′-BamHI restriction endonuclease site followed by sequences including the stop codon. IMPa-6 forward primers contained SalI and BglII sites, and the reverse primer contained a SacI site. PCR was performed in a 50-μL reaction volume with 2 units of Pwo polymerase (Roche). PCR amplification conditions were as follows: 95°C for 5 min (1 cycle); 95°C for 40 s, 58°C for 1 min, and 72°C for 5 min (36 cycles); and 72°C for 1 min (1 cycle). The PCR products were cloned into pBluescript KS+ (Stratagene) and verified by sequencing (Purdue Sequencing Facility).

Yeast Two-Hybrid Plasmid Construction and Assays

From the plasmids mentioned above, the corresponding importin α cDNAs were individually subcloned as EcoRI-BamHI fragments into bait and prey vectors pSTT91 and pGAD424, respectively (Clontech). The constructed plasmids and the bait and prey clones expressing full-length and truncated forms of octopine VirD2 (Hwang et al., 2006) and VirE2 are shown in Supplemental Table 4 online. As negative controls, empty baits, empty preys, or an Aspergillus nidulans GTPase protein were used as prey plasmids (see Supplemental Table 4 online). Yeast transformations were performed using the lithium acetate method (Becker and Lundblad, 1994). The cells were plated onto synthetic dropout–agar plates lacking the appropriate amino acids and incubated for 2 d at 30°C. Interactions were tested qualitatively by transferring the yeast transformants to synthetic dropout–agar plates containing the chromogenic substrate 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside. The plates were incubated for 2 d at 30°C. Positive interactions were detected by the expression of the reporter LacZ gene that converts the substrate 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside to an insoluble blue dye, 5-bromo 4-choloro indole.

Quantitative β-galactosidase assays, using ONPG (Sigma-Aldrich) as substrate, were performed on three independent colonies of each bait–prey combination. The assay was performed according to the protocol in the Matchmaker two-hybrid system user manual (Clontech). The absorbance of the reaction fluids was measured using a spectrophotometer (Spectra Max Plus 384; Molecular Devices). Protein quantification of the cell suspensions was performed using a Bradford assay. β-Galactosidase activity was reported as micromoles of ONPG hydrolyzed per minute per milligram of total protein.

Recombinant Protein Expression, Purification, and in Vitro Protein Binding Assays

Plasmids expressing recombinant importin α proteins were tagged on their N termini by GST or T7 epitope and are listed in Supplemental Table 4 online. The expression vectors for various tagged and untagged versions of either the full-length or truncated forms of VirD2 and VirE2 proteins are also described in Supplemental Table 4 online.

The expression and purification of GST fusion proteins were performed according to Ausubel et al. (2003) with some minor modifications. Isopropyl β-d-thiogalactopyranoside–induced Escherichia coli BL21 (DE3) cells containing the corresponding expression plasmids were collected by centrifugation and resuspended in chilled PBS containing 1 mM DTT and 1 mM phenylmethylsulfonyl fluoride (PMSF). Cell lysis was performed with 0.5 mg/mL (w/v) chicken egg white lysozyme (Sigma-Aldrich) for 2 h on ice. Triton X-100 was added to the cell suspension to a final concentration of 1% (v/v) and gently mixed by inversion. The lysate was centrifuged at 12,000 rpm for 30 min at 4°C to remove cell debris. The supernatant solution was added to PBS-preswelled glutathione–Sepharose beads (Sigma-Aldrich) to enable the GST fusion proteins to bind overnight at 4°C. The next day, the slurry of beads and supernatant solution was transferred to a column and the supernatant solution was allowed to flow through the column. The column containing the glutathione–Sepharose beads was washed with 25 bed volumes of chilled PBS containing DTT and PMSF. These glutathione–Sepharose beads containing the bound GST fusion proteins either were used directly for the in vitro GST pull-down assay (described below) or the bound proteins were eluted from the beads. Bound proteins were eluted in 1-mL fractions with chilled 50 mM Tris-HCl (pH 8) containing 5 mM reduced glutathione (Sigma-Aldrich). The fractions were analyzed by SDS-PAGE, and proteins were visualized by staining with Coomassie Brilliant Blue R 250.

Purification of T7-tagged importin α proteins was performed according to the instructions provided in the T7 tag affinity purification kit (Novagen). The fractions were analyzed by SDS-PAGE, and proteins were visualized by staining with Coomassie Brilliant Blue. Fractions containing the T7-tagged fusion protein were pooled.

For purification of His-tagged octopine VirE2, E. coli BL21 (DE3) cells containing the expression plasmid (pE2477; see Supplemental Table 4 online) were induced with isopropyl β-d-thiogalactopyranoside and the bacterial pellet was resuspended in resuspension solution (4 M urea, 50 mM Na-phosphate buffer [pH 8], 300 mM NaCl, and 20 mM imidazole [Sigma-Aldrich]). The suspension was sonicated and then centrifuged to remove cell debris as described earlier. The supernatant solution was added to a column containing nickel-nitrilotriacetic acid agarose resin (Qiagen) previously washed with 10 bed volumes of chilled resuspension buffer. The supernatant solution was allowed to flow through, and the resin was washed with 20 bed volumes of chilled resuspension solution. Bound proteins were eluted in 1-mL fractions with chilled elution buffer (4 M urea, 50 mM Na-phosphate buffer [pH 8], 300 mM NaCl, and 250 mM imidazole). The fractions were dialyzed overnight at 4°C against a buffer containing 50 mM Na-phosphate buffer (pH 8), 300 mM NaCl, and 20% glycerol. The fractions were analyzed by SDS-PAGE, and proteins were visualized by staining with Coomassie Brilliant Blue. Fractions containing His-tagged VirE2 fusion proteins were pooled. For all protein quantification, a Bradford assay was used with BSA as standard.

The GST fusion protein was bound to glutathione–Sepharose beads as described above. As a negative control, GST alone was linked to glutathione–Sepharose beads. The protein-bound beads were washed with 20 bed volumes of chilled binding buffer (20 mM Tris-HCl [pH 7.5], 75 mM KCl, 50 mM NaCl, 1 mM EDTA [pH 8], 0.05% Nonidet P-40, 10% glycerol, 1 mM DTT, and 1 mM PMSF). E. coli cell lysates from cultures expressing the T7-tagged (or untagged) proteins were passed through the glutathione–Sepharose column containing the corresponding GST-tagged protein (or GST alone). The column was washed with 20 bed volumes of chilled binding buffer, and the bound proteins were eluted with 50 mM Tris-HCl [pH 8] containing 5 mM reduced glutathione. The eluted fractions were analyzed by SDS-PAGE. Eluted fraction volumes corresponding to equimolar amounts of the first protein were subjected to SDS-PAGE, and after transfer to a nitrocellulose membrane, the proteins were subjected to protein gel blot analysis using the appropriate antibodies.

In Vitro Gel Mobility Shift Assay

A 237-bp Arabidopsis glyceraldehyde 3-phosphate dehydrogenase cDNA fragment obtained from the plasmid pE2565 by EcoRI digestion was used for the assay. The double-stranded DNA fragment was quantified, and fragment amounts equivalent to 10 pmol of 5′ ends were 5′ end labeled using [γ-³²P]dATP (Amersham) following the instructions in the DNA 5′ end labeling kit (Promega). The two denatured ssDNA strands had significant differences in mobility and were separated by electrophoresis at 4°C through a 4% polyacrylamide gel cast in low-ionic-strength buffer (6.7 mM Tris-HCl [pH 7.5], 1 mM EDTA, and 3.3 mM sodium acetate). The gel was prerun at a potential of 65 V for 20 min prior to loading of sample. Electrophoresis was performed at 4°C at a similar voltage for 24 h. The gel was gently transferred to a plastic wrap and exposed to x-ray film (Kodak). The x-ray film was developed and aligned to the gel. A gel slice corresponding to the more rapidly migrating ssDNA was cut, and the DNA was eluted by electrophoresis, precipitated, and resuspended in binding buffer (20 mM Tris-HCl [pH 7.5], 75 mM KCl, 50 mM NaCl, 1 mM EDTA [pH 8], 0.05% Nonidet P-40, 10% glycerol, 1 mM DTT, and 1 mM PMSF). The labeled ssDNA was quantified by measuring the A₂₆₀. His-tagged octopine VirE2 and T7-tagged IMPa-4 proteins were dialyzed overnight at 4°C against binding buffer.

To determine the minimal concentration of VirE2 protein required to saturate fully the ssDNA binding sites, labeled ssDNA (5 ng for each reaction) was incubated on ice for 15 min with increasing amounts of purified His-tagged octopine VirE2 protein (7 to 50 ng VirE2, 20 μL of each reaction volume). Loading solution (6 μL of 10% glycerol and 0.1% bromphenol blue) was added to the reaction mixture, and electrophoresis was performed as described above. The gel was dried and subjected to autoradiography.

To determine whether ssDNA-bound VirE2 protein interacted with importin α, labeled ssDNA was preincubated with VirE2 (5 ng of ssDNA per 30 ng of VirE2) on ice for 15 min followed by addition of increasing amounts of T7-tagged IMPa-4 protein (20 to 400 ng, 20 μL of each reaction volume). The mixture was incubated on ice for 20 min followed by the addition of loading solution. As a control, 400 ng of IMPa-4 was incubated with 5 ng of labeled ssDNA in the absence of VirE2. Electrophoresis was performed as described above. The gel was dried and subjected to autoradiography.

To investigate whether the preformed IMPa-4–VirE2 protein complex binds ssDNA, 400 ng of IMPa-4 was preincubated on ice for 15 min with 30 ng of VirE2. To this mixture, 5 ng of labeled ssDNA was added (reaction volume of 20 μL) followed by additional incubation on ice for 20 min. Loading solution was added, and electrophoresis was performed as discussed above. To identify the ssDNA–VirE2–IMPa-4 complex, a parallel lane containing ssDNA-bound VirE2 complex was used as a control (30 ng of VirE2 protein bound to 5 ng of labeled ssDNA). The negative control lane contained 400 ng of IMPa-4 preincubated with 5 ng of labeled ssDNA. The gel was dried and subjected to autoradiography.

Identification of Arabidopsis Lines from the Feldmann and Salk T-DNA Insertion Collection with T-DNA Insertions in the Individual Importin α Genes

A PCR-based reverse genetics approach similar to that described by Zhu et al. (2003) was used to identify an Arabidopsis line (ecotype Ws-2) with a T-DNA insertion in the IMPa-4 gene. Briefly, PCR primers specific to each of the four members of the Arabidopsis importin α gene family were designed and used in combination with the mutagenizing T-DNA right or left border primer. T-DNA– and IMPa-4–specific primer sequences are presented in Supplemental Table 3 online. PCR was successively performed on genomic DNA isolated from pools of 1000, 100, and 20 T-DNA insertion lines. PCR amplification conditions were as follows: 95°C for 5 min; 36 cycles at 94°C for 40 s, 56°C for 1 min, 72°C for 3 min, 72°C for 10 min, and 4°C hold. PCR products of the expected sizes were sequenced to identify the T-DNA–plant DNA junctions. Using these approaches, the PCR product was narrowed to a single Arabidopsis plant. Seeds harvested from this plant were used for genotyping and RT-PCR analyses.

Seeds of Arabidopsis (ecotype Columbia; CS60,000) with T-DNA insertions in the importin α genes (KAPα, IMPa-2, and IMPa-3) were identified using the SIGnAL T-DNA Express Arabidopsis Gene Mapping Tool (http://signal.salk.edu/; Alonso et al., 2003) and were obtained from the ABRC at Ohio State University and subsequently used for genotyping and RT-PCR analyses.

Genotyping and RT-PCR Analyses of Arabidopsis Lines with T-DNA Insertions in the Individual Importin α Genes

Genomic DNA isolated from 3-week-old plants was used for PCR-based genotypic analyses. For genotyping, the impa-4 mutant obtained from the Feldmann collection, the right border, the left border, and the IMPa-4–specific primers were as mentioned in Supplemental Table 3 online. For genotyping the importin α mutants obtained from the Salk collection, the primer sequences are presented in Supplemental Table 3 online. PCR products of appropriate sizes, when obtained, were sequenced to identify the T-DNA–plant DNA junctions.

For RT-PCR analysis, 2 μg of total RNA from 2- to 3-week-old plants was first reverse-transcribed using oligo(dT) primers and then PCR-amplified with importin α–specific primers using the amplification conditions described above. For KAPα, IMPa-2, and IMPa-4, the primers were the same as those used for the genotyping. For IMPa-3, the forward primer used for genotyping was used in combination with a reverse primer as presented in Supplemental Table 3 online. Arabidopsis ACTIN2 (ACT2)-specific primers were used as a positive control to demonstrate equivalent efficiencies of double-stranded cDNA amplification in the samples investigated. The sequences of the actin2 primers are presented in Supplemental Table 3 online.

Measurement of Transcript Levels by Real-Time PCR

Total RNA was isolated as described earlier (Hwang and Gelvin, 2004). A 5-μg aliquot of purified RNA was reverse-transcribed in a 20-μL reaction volume using the SuperScript II first-strand cDNA synthesis system and oligo(dT)_12-18 according to the manufacturer's instructions (Invitrogen). The reaction was terminated by heat inactivation at 70°C for 15 min. Subsequently, the reaction mix was diluted 1:25 with water, and 20 μL of the diluted cDNA (corresponding to 100 ng of total RNA) was used as a template for real-time PCR analysis with the Rotor-Gene 2000 (Corbett Research). Amplification mixtures (20 μL per reaction) consisted of 1× Ex Taq buffer (Takara Bio), 0.2 mM deoxynucleotide triphosphate, 0.2 M each primer, 1:20,000 times diluted SYBR Green I (Molecular Probes), 0.5 units of TaKaRa Ex Taq (Takara Bio), and cDNA corresponding to 100 ng of total RNA. Cycling conditions were as follows: 5 min at 94°C; 35 cycles of 20 s at 94°C, 20 s at 54°C, and 30 s at 72°C; product extension at 72°C for 5 min. This was followed by a melting-curve program (72 to 99°C, with a 5-s hold at each temperature). Fluorescence data were acquired at the 72°C step and during the melting-curve program. For the real-time PCR analysis, gene-specific primers were prepared as presented in Supplemental Table 3 online. The identities of the amplicons and the specificity of the reaction were verified by agarose gel electrophoresis, melting-curve analysis, and sequencing. Absolute and relative copy numbers of individual mRNA species were calculated with standard curves of known copy numbers for each of the importin α cDNAs. Four replicates of each sample and standard concentrations were included with every experiment, along with no-template controls. Rotor-Gene 5.0.37 software (Corbett Research) was used to determine the optimal cycle threshold from the dilution series of standards and to calculate the mean expression level for each set of four replicates for each cDNA.

Peptide Antibody Synthesis and Protein Blot Analysis

Synthesis of an IMPa-4–specific peptide, generation of polyclonal antibodies against this peptide in rabbits, and affinity purification of the raised antibodies were performed at GenScript. In order to link the IMPa-4–specific peptide STRAELRKKIYKTG to the carrier protein KLH (for Keyhole Limpet Hemocyanin), a Cys residue was added to the C terminus of this peptide.

For protein blot analysis, plant tissue was ground to a fine powder and resuspended in 1 volume of 2× Laemmli buffer in an Eppendorf tube. Samples were incubated in boiling water for 15 min and centrifuged at 14,000 rpm at room temperature, and the supernatant solution was transferred to a new Eppendorf tube. For protein blot analysis of E. coli extracts, cells were pelleted and resuspended in 2× Laemmli buffer and boiled for 5 min. Proteins were separated through 10% SDS polyacrylamide gels. Proteins were either stained with Coomassie Brilliant Blue or transferred to a nitrocellulose membrane (BioTraceNT; Pall Corporation). For immunostaining, membranes were blocked in 1× TBS-T (0.05 M Tris-HCl, 150 mM NaCl, and 0.1% [v/v] Tween, pH 7.6) with 5% (w/v) skim milk for at least 1 h. After washing the membranes three times for 5 min in 1× TBS-T, the membranes were incubated for 60 min in 1× TBS-T with 0.5% skim milk and a 1:1000 dilution of the affinity-purified antibody raised against IMPa-4. Membranes were washed four times for 5 min in 1× TBS-T and incubated for 60 min in 1× TBS-T with 0.5% skim milk and the secondary antibody, a 1:10,000 dilution of goat anti-rabbit IgG horseradish peroxidase conjugate (12-348; Upstate). Membranes were washed four times for 10 min in 1× TBS-T. To visualize IMPa-4 protein, we used a SuperSignal West Femto trial kit (Pierce). Membranes were exposed to CL-Xposure film (Pierce).

Construction of IMPa-4 RNAi Lines

We generated an RNAi construct specifically targeting IMPa-4 as follows. We PCR-amplified a 789-bp fragment containing 79 bp of the coding sequence and 710 bp of the 3′ UTR. This region shows no significant homology to other Arabidopsis thaliana IMPa genes or to other Arabidopsis genes. We designed primers such that two restriction sites were added to each end of the PCR product: XbaI and AscI restriction sites at the 5′ end and BamHI and SwaI sites at the 3′ end. The primer sequences are presented in Supplemental Table 3 online. PCR was performed using DNA from the Arabidopsis BAC clone T12M4 (obtained from the ABRC). The PCR product was cloned into pUC19 digested with HincII, the sequence was verified, and the construction was designated pE3201. The RNAi construct was made according to the instructions at www.chromdb.org. First, a SwaI/AscI fragment from pE3201 was cloned into the RNAi vector pFGC5941 (Kerschen et al., 2004) digested with the same enzymes, then a XbaI/BamHI fragment from pE3201 was cloned into the previous construct digested with XbaI/BamHI. This final construct was transformed into Agrobacterium tumefaciens GV3101 (Koncz and Schell, 1986). Arabidopsis plants (ecotypes Ws-2 and Columbia) were transformed by a floral dip protocol (Clough and Bent, 1998). Transgenic plants were selected on B5 medium supplemented with 100 μg/mL timentin and 10 μg/mL phosphinothricin.

Construction of Binary Vectors and Generation of Importin α cDNA Overexpression Lines in the impa-4 Mutant Background

The KAPα, IMPa-2, IMPa-3, or IMPa-4 cDNAs were cloned as XbaI/HincII fragments into the XbaI-T4 polymerase–filled SacI sites of pE1546, a T-DNA binary vector with a CaMV 35S promoter, to express the cDNAs and a hptII gene as the plant selectable marker. The cDNA of IMPa-6 was cloned as a SalI/SacI fragment into pE1546 digested with the same enzymes. The cDNA of IMPa-7 was cloned as a SalI/EcoRI Klenow filled-in fragment into pE1546 digested with SalI and filled in using Klenow. The cDNA of IMPa-9 was cloned as a BamHI Klenow-filled fragment into pE1546 digested with SalI and filled in using Klenow. These various binary vectors were introduced into the disarmed Agrobacterium strain GV3101 (Koncz and Schell, 1986), and the resulting Agrobacterium strains were used to generate transgenic Arabidopsis plants in the impa-4 mutant background using a floral dip transformation protocol (Clough and Bent, 1998).

Construction of Binary Vectors and Generation of Transgenic Plants Expressing YFP-Tagged VirD2 and VirE2

For the binary vector expressing a YFP-VirD2 fusion protein, the C-terminal NruI fragment (∼135 bp) of octopine VirD2 was fused in the correct reading frame downstream of the EYFP gene at the SmaI site of pCAM-35S-EYFP-C1 (a kind gift from Erik Nielsen at the Donald Danforth Plant Science Center), resulting in the plasmid pE2837. For the binary vectors expressing YFP-VirE2 or VirE2-YFP, the corresponding YFP-VirE2 or VirE2-YFP fusion sequences, obtained as a NcoI/XbaI fragment from pE2774 or pE2775, respectively, were individually subcloned into the NcoI–XbaI sites of pCAM-35S-EYFP-C1 to generate pE2838 or pE2836, respectively. A polylinker sequence (5′-GAGAGGAAAGAGAGG-3′) separates the two fusion proteins in YFP-VirE2 or VirE2-YFP. The above binary vectors along with a control vector, pCAM-35S-EYFP-C1 (pE2829), were used to generate transgenic plants in the wild type and in the impa-4 mutant background utilizing the Agrobacterium-mediated floral dip transformation protocol as described above.

BiFC

Genes encoding VirD2 and VirE2 from pTiA6 were cloned into the various BiFC vectors pSAT6-nEYFP-C1, pSAT6-nEYFP-N1, and pSAT6-cEYFP-N1 (Citovsky et al., 2006). virD2 was cloned as an EcoRI/SacII fragment, and virE2 was cloned as a NcoI/BamHI fragment. cDNAs (lacking translation stop codons) of KAPα, IMPa-4, IMPa-7, and IMPa-9 were cloned into the BiFC vectors pSAT6-nEYFP-N1 and pSAT6-cEYFP-N1 as NcoI/XmaI fragments. For multicolor BiFC, proteins were tagged with cCFP (amino acids 155 to 239), nVenus (amino acids 1 to 174), or nCerulean (amino acids 1 to 174) as described (Shyu et al., 2006). In some instances, mRFP (for red fluoresecent protein; Campbell et al., 2002) or mCherry (Shaner et al., 2004) was coexpressed from a CaMV double 35S promoter (Restrepo et al., 1990). Tobacco (Nicotiana tabacum) BY-2 protoplasts were prepared and coelectroporated with 20 μg of each BiFC construction as described previously (Mysore et al., 1998). In some instances, the cells were stained with 2 μg/mL Hoechst 33242 (Molecular Probes) in PBS–0.4 M mannitol buffer, and fluorescence was visualized using a Nikon Eclipse E600 fluorescence microscope.

We generated IMPa-4–nYFP and –cYFP constructions within the native context of the gene as follows. We cloned a 6351-bp HincII fragment, harboring the IMPa-4 gene, from BAC clone T12M4 into pUC19 digested with HincII and designated this clone pE3203. This HincII fragment contains 2257 bp upstream of the putative translational start codon, the entire coding region of the IMPa-4 gene (exons and introns), and 766 bp downstream of the stop codon. This construction is flanked by two SalI sites. To enable generation of in-frame fusions, we digested pSat6-n(1-174)EYFP-N1 and pSat6-c(175-end)EYPF-N1 (Tzfira et al., 2005) with XmaI and filled in the overhanging ends using Klenow fragment of DNA polymerase, then digested the vectors with XhoI. We subcloned a XhoI/MspA1I fragment, containing the 3′ end of the IMPa-4 coding sequence, into the digested pSat vectors and partially sequenced them to confirm that the IMPa-4 open reading frames were in-frame with the YFP fragments. These clones were designated pE3204 (nYPF-tagged IMPa-4 fragment) and pE3205 (cYPF-tagged IMPa-4 fragment). We next used PCR to amplify a 612-bp fragment harboring the 3′ UTR and the putative terminator region of IMPa-4, using the primers presented in Supplemental Table 3 online. This amplification added a XbaI site to the 5′ end and a NotI site to the 3′ end of the PCR product. The PCR product was first cloned into pBluescript KS+ digested with SmaI, then the XbaI/NotI fragment of this clone was ligated into both pE3204 and pE3205 digested with the same enzymes. Finally, we individually cloned XhoI/ClaI fragments from pE3204 and pE3205 into pE3203 that was digested with the same enzymes. The resulting constructs were designated pE3292 (nYPF-tagged IMPa-4) and pE3293 (cYPF-tagged IMPa-4). SalI fragments from pE3203, pE3292, and pE3293 were ligated into the T-DNA binary vector pCAMBIA1300 cut with the same enzyme, and the constructs were transformed into Agrobacterium GV3101 and EHA105 by electroporation (Nagel et al., 1990).

Agrobacterium-Mediated Transient and Stable Transformation Assays

All Agrobacterium strains were cultured in 5 mL of yeast extract peptone (Lichenstein and Draper, 1986) supplemented with the appropriate antibiotics (rifampicin, 10 μg/mL; kanamycin, 25 μg/mL) at 30°C. Two milliliters of an overnight bacterial culture was inoculated into 25 mL of YEP medium with appropriate antibiotics and grown at 30°C to an A₆₀₀ of 0.85 (1 to 2 × 10⁹ cells/mL). The cells were washed with sterile 0.9% NaCl and resuspended in 0.9% NaCl at 2 × 10⁸ cells/mL for stable and transient root transformation assays.

Seeds from the wild type and from the various transgenic plants generated were surface-sterilized with a solution containing 50% commercial bleach and 0.1% SDS for 10 min and rinsed five times with sterile distilled water. Seeds were resuspended in sterile distilled water and incubated in the dark at 4°C for 2 d. Seeds were then placed on Gamborg's B5 medium (GIBCO) containing 0.75% Bactoagar (BD Biosciences) and appropriate antibiotics (timentin, 100 μg/mL; kanamycin, 50 μg/mL for the T-DNA mutants; hygromycin, 20 μg/mL for the various importin α overexpression lines or the YFP, YFP-VirD2, YFP-VirE2, and VirE2-YFP overexpression lines). Wild-type seeds were placed on medium lacking the antibiotics. The plates were incubated for 10 d under a 16-h-light/8-h-dark photoperiod at 25°C. Seedlings were transferred individually into baby food jars containing solidified B5 medium without any antibiotics, grown for 3 weeks, and then used for transformation assays as described by Nam et al. (1997, 1998, 1999) and Zhu et al. (2003).

For transient transformation assays, infection was performed with Agrobacterium GV3101 (Koncz and Schell, 1986) containing the binary vector pBISN1 (Narasimhulu et al., 1996) as described (Nam et al., 1997, 1998, 1999; Zhu et al., 2003). pBISN1 contains within the T-DNA a gusA intron gene under the control of a superpromoter (Narasimhulu et al., 1996). Transient transformation efficiency was reported as the percentage of root segments showing GUS staining.

For stable transformation assays, the tumorigenic strain Agrobacterium A208 was used. After a 2-d cocultivation period, root bundles were separated into single root segments onto Murashige and Skoog basal medium plates containing 100 μg/mL timentin. The plates were incubated at 23°C for 4 weeks, after which they were scored for tumor development and the tumors were categorized according to morphology (large green, small green, large yellow, or small yellow). Stable transformation efficiency was reported as the percentage of total root segments showing tumors.

Confocal Microscopy Analyses of the YFP-Expressing Arabidopsis Lines

Several independent 1-week-old transgenic T1 plants expressing either YFP or YFP-Vir protein fusions were placed on their sides on B5 agar plates containing appropriate antibiotics (timentin, 100 μg/mL; hygromycin, 20 μg/mL) and grown in a vertical position for an additional 2 weeks. Root segments (2 to 3 mm) containing the tip meristematic regions were excised and immediately fixed for 1 h at room temperature with fixative solution (4% [v/v] paraformaldehyde [Sigma-Aldrich], 0.2% Nonidet P-40 in PEM 50:5:5 [pH 5.7]). PEM 50:5:5 contains 50 mM PIPES, 5 mM EGTA, and 5 mM MgSO₄. The root segments were washed twice for 5 min each with PEM 50:5:5 and then counterstained for 15 min with Hoechst 33242 (2 μg/mL). After an additional wash for 5 min with PEM 50:5:5, the root segments were mounted on slides containing a drop of fluorescent mounting medium (0.1 M Tris-HCl [pH 9.0], 50% glycerol, and 1 mg/mL p-phenylenediamine hydrochloride [Sigma-Aldrich]). Confocal microscopy analysis was performed using the Bio-Rad Radiance 2100 MP rainbow system microscope equipped with 20× multi-immersion and 60× water objectives.

Accession Numbers

Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: KAPα, At3g06720; IMPa-2, At4g16143; IMPa-3, At4g02150; IMPa-4, At1g09270; IMPa-5, At5g49310; IMPa-6, At1g02690; IMPa-7, At3g05720; IMPa-8, At5g52000; IMPa-9, At5g03070; ACT2, At3g18780; UBQ5, At3g62250.

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure 1. Amino Acid Alignment of the Seven Importin α Proteins Investigated.

• Supplemental Figure 2. Importin α–Interacting Domains of VirD2 and VirE2 Are Not Restricted to NLS Residues.

• Supplemental Figure 3. Saturation Kinetics of ssDNA Binding Activity by VirE2.

• Supplemental Figure 4. Homozygous Arabidopsis Importin α Mutants Show Reduction in the Corresponding Importin α Transcripts.

• Supplemental Figure 5. Peptide Antibody against Arabidopsis IMPa-4 Detects Recombinant IMPa-4 but Not IMPa-1, -2, or -3.

• Supplemental Figure 6. Expression of Various Importin α Transcripts in Tissues of Wild-Type Plants.

• Supplemental Figure 7. VirE2-YFP Localizes to the Cytoplasm of Electroporated Tobacco BY-2 Cells.

• Supplemental Table 1A. Stable and Transient Transformation Assays on T2 Lines Overexpressing Various Importin α Isoforms in the impa-4 Mutant Background.

• Supplemental Table 1B. Stable Transformation Assays on T1 Lines Overexpressing Various Importin α Isoforms in the impa-4 Mutant Background.

• Supplemental Table 2. Localization of BiFC Signal in Electroporated Tobacco BY-2 Cells.

• Supplemental Table 3. Primers Used in PCR.

• Supplemental Table 4. Plasmids Used in Yeast Two-Hybrid, in Vitro, and in Planta Interaction Assays.

• Supplemental Data Set 1. Sequences Used to Generate the Cladogram in Figure 1.