Plant Transformation by Agrobacterium tumefaciens

Agrobacterium tumefaciens infects plant cells by the transfer of DNA. A key factor in this process is the bacterial virulence protein VirE2, which associates stoichiometrically with the transported single-stranded (ss) DNA molecule (T-strand). As observed in vitro by transmission electron microscopy, VirE2-ssDNA readily forms an extended helical complex with a structure well suited to the tasks of DNA protection and nuclear import. Here we have elucidated the role of the specific molecular chaperone VirE1 in regulating VireE2-VirE2 and VirE2-ssDNA interactions. VirE2 alone formed functional filamentous aggregates capable of ssDNA binding. In contrast, co-expression with VirE1 yielded monodisperse VirE1–VirE2 complexes. Cooperative binding of VirE2 to ssDNA released VirE1, resulting in a controlled formation mechanism for the helical complex that is further promoted by macromolecular crowding. Based on this in vitro evidence, we suggest that the constrained volume of the VirB channel provides a natural site for the exchange of VirE2 binding from VirE1 to the T-strand.

The transfer of DNA from the soil bacterium Agrobacterium tumefaciens into plant cells is an efficient process utilizing both bacterial and host machineries (1)(2)(3). Detection of phenolic compounds released from wounded plant tissues leads to adhesion and induction of the bacterial virulence (vir) machinery. Vir proteins are responsible for the excision of the single-stranded transfer DNA (tDNA), a specific segment of the bacterial tumor-inducing plasmid. Its delivery to the host cell cytoplasm occurs in complex with a single VirD2 molecule attached at its 5Ј end (the T-strand).
The T-strand is wrapped with numerous VirE2 proteins to produce a mature transfer complex (T-complex) (3), which is then imported into the host cell nucleus. Additional bacterial proteins and host factors are involved in genomic integration and expression of the encoded genes (1)(2)(3)(4)(5)(6).
It is well established that the T-strand and several other virulence proteins (i.e. VirE2, VirF, VirE3, and VirD5) are transported independently into the host plant cytoplasm via a type IV secretion system (7,8). However, the site of T-complex formation remains unclear. Co-inoculation of plants with a combination of individually avirulent Agrobacterium strains (⌬virE2 and tDNA) resulted in tumorigenesis (9), suggesting that the mature T-complex forms in the plant cell rather than in the bacterium. This observation was further corroborated by transformation of VirE2-expressing tobacco plants with a ⌬virE2 Agrobacterium mutant (10). The mechanism by which VirE2 interaction with tDNA is inhibited in the bacterium, before translocation of the molecular substrates to the plant host, is still unresolved.
VirE2 is a large (63.5 kDa) single-stranded DNA-binding protein. It binds without sequence specificity (10 -12) in a highly cooperative manner (13,14) that is readily detected in vitro. The resulting complex forms hollow cylindrical filaments with a solenoidal structure (e.g. a coiled telephone cord) (15,16) that protects the tDNA from cytoplasmic nucleases in the plant (17). VirE2 also interacts with the plant host factor VIP1 (18) and the bacterial protein VirE3 (19), both of which can mediate interaction with the karyopherin-␣ nuclear import receptor (20). In vitro, VirE2 forms channels in lipid bilayers (21), suggesting a role in mediating DNA import across the plant plasma membrane (22). The binding of VirE2 by its molecular chaperone VirE1, a 7.1-kDa protein located on the same operon as VirE2 (23,24), is suggested to prevent premature association of VirE2 with the T-strand in bacterial cells (25). VirE1-VirE2 association preserves the latter in a soluble, nonaggregated form essential for virulence (24 -26). Agrobacterium strains lacking the virE1 gene or not expressing it in cis to virE2 are avirulent (7). It was therefore proposed that VirE1 preserves VirE2 in a state suitable for translocation through the VirB/D4 channel (27), into which VirE2 and T-strand are targeted independently (28).
Here we have addressed the role of VirE1 in VirE2-ssDNA 2 complex formation from a structural point of view. We found that in the absence of VirE1, VirE2 formed soluble filamentous aggregates that bind ssDNA. In contrast, when the two genes were co-expressed, a monodisperse species of VirE2 in complex with VirE1 was formed. VirE1 was dissociated from VirE2 upon binding of the latter to ssDNA. The dissociation and protein-DNA complex assembly processes then proceeded in a controlled manner that was promoted by macromolecular crowding effects.

Protein Expression and Purification
VirE2 Purification-Nopaline virE2 (29) was cloned into the pET-28b vector (Novagen) and expressed in the Escherichia coli BL21(DE3) strain by induction with 1 mM isopropyl ␤-D-thiogalactopyranoside at 37°C for 3 h. Following sonication in buffer A (10 mM Hepes, pH 7.4, 100 mM NaCl, 1 mM EDTA, 1 mM DTT, and 10% glycerol) plus 1 mM phenylmethylsulfonyl fluoride (Sigma) and 170 units of protease inhibitor mixture (Calbiochem), inclusion bodies were isolated and purified by three washes in buffer A plus 2% Triton X-100. The inclusion bodies were then washed in buffer B (20 mM Hepes, pH 7.4, 1 M NaCl, 1 mM EDTA) and denatured in buffer A plus 5 M urea. Following refolding by dialysis against buffer C (10 mM KH 2 PO 4 , pH 7.4, 0.1 mM CaCl 2 , 1 mM DTT) VirE2 was purified on a ceramic hydroxyapatite column using a linear gradient of phosphate buffer ranging from 10 to 500 mM. VirE2 eluted at 150 mM KH 2 PO 4 and was further purified by gel filtration (Pharmacia HiLoad 16/60 Superdex 200). Purified fractions of the protein were pooled, concentrated separately to 0.5 mg/ml, and stored at Ϫ80°C.
VirE1-VirE2 Co-expression and Purification-The virE1 and virE2 genes were cloned into pACYCDuet-1 (Novagen). VirE1 expressed from this vector bears an N-terminal His 6 tag followed by a tobacco etch virus protease cleavage site (ENLY-FQG), whereas VirE2 is untagged. BL21(DE3) cells harboring the VirE1-VirE2 duet plasmid were induced by 0.05 mM isopropyl ␤-D-thiogalactopyranoside at A 600 0.6. The culture was harvested after additional growth for 18 h at 15°C. Bacterial pellets were sonicated in buffer D (50 mM Tris, pH 8.0, 500 mM NaCl, 1 mM DTT). The VirE1-VirE2 complex was captured on a nickel-chelating column (Pharmacia HiTrap-chelating) and eluted at 200 mM imidazole. Following buffer exchange by dialysis to buffer C, the complex was purified on a heparin column (Pharmacia HiTrap heparin) with a linear salt gradient to 1 M NaCl. The VirE1-VirE2 complex eluted at 200 mM NaCl and was treated with tobacco etch virus protease to remove the His 6 tag from VirE1. Final purification of the complex was done by gel filtration in buffer T (50 mM Tris, pH 8.0, 150 mM NaCl, 1 mM DTT, 1 mM EDTA).
Expression and Purification of N-and C-terminal His 6 -tagged VirE2-Nopaline virE2 (29) was His 6 -tagged at its N or C terminus by cloning into pET-28b (Novagen). For the N-terminal fusion, virE2 was cloned between the HindIII and XhoI sites, whereas for the C-terminal fusion, it was cloned between the NcoI and XbaI sites. Both constructs were expressed and purified as described above. BL21(DE3) pLysS cells containing the expression plasmids were induced with 0.1 mM isopropyl ␤-Dthiogalactopyranoside. Clean inclusion bodies containing tagged VirE2 were denatured in buffer A (10 mM Hepes, pH 7.4, 100 mM NaCl, 1 mM EDTA, 1 mM DTT, and 10% glycerol) plus 6 M guanidine hydrochloride. Soluble protein was captured on a nickel-chelating column (Pharmacia HiTrap-Chelating). Buffer was then changed to buffer E (10 mM Hepes, pH 7.4, 500 mM NaCl, 1 mM DTT, 6 M urea) and protein refolded on the nickel column by gradual exchange to buffer A without urea. Refolded protein was eluted from the column at 200 mM imidazole. Fractions containing the protein were pooled, dialyzed against buffer T, and stored in aliquots at 0.5 mg/ml at Ϫ80°C.

Sample Preparation and Transmission Electron Microscopy
Single-stranded DNA (M13mp18 ssDNA, New England Biolabs) or 20-mer oligo-DNA with an arbitrary sequence (AGC-TACCATGGCTTCATCAG; Metabion FPLC pure) were denatured by heating to 70°C for 5 min and then cooled on ice for 5 min. VirE2 or VirE1-VirE2 was added and allowed to interact for ϳ3 h at room temperature in buffer H (10 mM Hepes, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM DTT). Mixing ratios are denoted as w/w; complete coverage of VirE2:ssDNA corresponds to a 10:1 VirE2:ssDNA ratio. Samples of 5-10 l at a final protein concentration of 50 -100 g/ml were deposited on glow-discharged carbon-coated 400 copper grids and stained with 1% uranyl acetate for 30 s. The interaction of DNA with uranyl acetate rendered some of the specimens positively stained, allowing for the detection of both protein and ssDNA. Images were recorded on a charge-coupled device camera (TVIPS) using either an FEI Tecnai T12 or F20 transmission electron microscopy (TEM).

Complex Composition and Gel Electrophoresis
To check for complex formation between VirE2 or VirE1-VirE2 and ssDNA, samples were analyzed relative to naked DNA on 0.8% agarose gel and stained with GelStar (Cambrex). Alternatively, 5Ј 32 P radioactively labeled 20-mer ssDNA substrate (sequence above) was used, and the samples were analyzed on 12% acrylamide nondenaturing gel. The gel was then dried on filter paper and analyzed by phosphoimager (Fuji, Inc.). The composition of the complex formed between VirE1-VirE2 and ssDNA was determined by filtration and gel electrophoresis. VirE1-VirE2 complex was mixed with M13 ssDNA at a 25:1 w/w ratio for ϳ3 h and then filtered through a NanoSep device (Pall, Inc.) with 300-kDa cutoff. Filtration was repeated twice, increasing the volume of the concentrate to 200 l before each filtration step. The fractions were probed on a 15% Tricine gel (30) followed by silver staining to reveal the 7.1 kDa-VirE1 protein.

Analytical Ultracentrifugation
Equilibrium analytical ultracentrifugation was performed in UV absorption mode at 4°C in a Beckman XL-A instrument using sample cells fitted with 6-channel centerpieces. The VirE1-VirE2 protein stock solution was diluted successively in buffer T to obtain solutions of 3.0, 1.0, and 0.5 mg/ml protein.
To extend the measurement range to higher concentrations, absorbance data were collected at 250 nm as well as at 280 nm. Measurements were repeated over 48 h at 7000 rpm following preliminary trials at 6000, 9000, and 10000 rpm. The data were FEBRUARY 9, 2007 • VOLUME 282 • NUMBER 6 analyzed by plotting the natural logarithm of the optical absorption (proportional to concentration) versus the square of the radius from the center of rotation. For a homogeneous sample the slope of this plot is proportional to the molecular mass (MM) according to Equation 1,

VirE1 Modulates T-complex Formation
where R is the gas constant, is the angular velocity, is the solution density (assumed equal to unity), and is the partial specific volume of the protein (calculated to be 0.72 from amino acid content by the UltraScan software suite).

Mass Spectrometry
To determine the stoichiometry of the VirE1-VirE2 complex, electrospray ionization (ESI) mass spectrometry (MS) and tandem MS (MS/MS) experiments were preformed on a QSTAR XL instrument. Prior to MS analysis, 30 l of a 1.8 mg/ml solution of His 6 -VirE1-VirE2 in 10 mM KH 2 PO 4 and 0.2 M NaCl was buffer-exchanged twice into 0.5 M ammonium acetate solution using Bio-Rad Biospin columns and stored on ice. Typically, a 2-l aliquot was loaded for sampling via NanoESI capillaries prepared in-house from borosilicate glass tubes as described previously (31). The following experimental parameters were used: capillary voltage up to 1.2 kV, declustering potential 150 V, focusing potential 250 V, second declustering potential 30 V focusing rod offset 40 V, and microchannel plate detector 2350 V. In MS/MS the relevant m/z value was selected in the quadrupole, and collision energies up to 120 V were employed. Argon was used as a collision gas at maximum pressure. All spectra were calibrated externally by using a solution of cesium iodide (100 mg/ml). Spectra in Fig. 4 are shown with minimal smoothing and without background subtraction.

Small-angle Solution X-ray Scattering (SAXS) Experiments
SAXS data were collected on the X33 camera of the European Molecular Biology Laboratory at the storage ring DORIS III (Deutsches Elektronen-Synchrotron (DESY), Hamburg) (32). Purified solutions of the co-expressed VirE1-VirE2 complex were measured in buffer T at 7.8 and 3.9 mg/ml concentrations (determined by UV absorption at 280 nm, with ⑀ ϭ 49,760 cm Ϫ1 M Ϫ1 ) using a mar345 Image Plate Detector in the range of momentum transfer 0.13 Ͻ s Ͻ 3.0 nm Ϫ1 (s ϭ 4 sin()/), where 2 is the scattering angle and ϭ 0.15 nm is the x-ray wavelength). Data were processed following standard procedures by the programs PRIMUS (33) and GNOM (34) to yield the final scattering pattern, overall parameters, and the distance distribution function p(r) of the particle. The molecular mass of the solute was evaluated by comparison with the scattering from a reference solution of bovine serum albumin. The shape of VirE1-VirE2 was restored ab initio using the simulated annealing program DAMMIN (35). Multiple shape reconstructions were analyzed by the program DAMAVER (36) yielding the most representative models of VirE1-VirE2. For display, these models were converted to a volume of uniform density with a Gaussian low-pass filter of 3.0 nm using pdb2mrc from the EMAN image processing suite (37).

RESULTS
VirE2 Alone Forms Functional Aggregates-VirE2 forms inclusion bodies when overexpressed in E. coli (12). After refolding, size exclusion gel filtration analysis revealed that the major fraction of protein eluted from the column in the void volume, indicating soluble aggregates of molecular weight greater than the cutoff of 600,000 Da (Fig. 1A). A 20-mer DNA oligomer readily associates with the VirE2 aggregates as seen by gel shift analysis (Fig. 1B). Twenty bases correspond to the binding footprint of VirE2 (63.5 kDa) to ssDNA (average molecular weight 315/base) at a 10:1 w/w ratio corresponding to the observed saturation of protein binding, i.e. maximal gel shift or, equivalently, "full coverage" of the DNA by protein (15,16). Thus the aggregated protein retains DNA binding activity. The maximal gel shift of the complex formed by this protein with 20-mer ssDNA required a protein:DNA ratio of 15:1 (w/w), 50% higher than the saturating ratio observed with long ssDNA substrates. This indicates a coexistence of DNA-bound and unbound protein.
Purified soluble aggregates of VirE2 alone, and VirE2 in complex with oligo-DNA, were analyzed by TEM. In the absence of DNA, the VirE2 protein stock appeared as filamentous aggregates of indistinct shape with a typical diameter of ϳ14 nm ( Fig.  2A). After incubation for several hours at room temperature, a few percent of these aggregates adopted a helical structure ( Fig.  2A, inset) similar to that of complexes formed between VirE2 and long ssDNA (Fig. 2B). Compact aggregates are identified with tetrameric rings described previously (16). Well ordered helical complexes were also formed following incubation of VirE2 with 20-mer oligo-DNA (Fig. 2C). These complexes reached several micrometers in length despite the fact that this DNA substrate matches the binding footprint of a single VirE2 protein. Thus, the transition of VirE2 aggregates into helical conformation is promoted by the binding to ssDNA. The effect on complex formation of peptide tags at either end of VirE2 was examined by fusion of a His 6 tag to the N or C terminus of VirE2. Tagged VirE2 retained ssDNA binding activity (gel shift, not shown), although the complexes that were formed lacked an extended helical morphology and appeared instead as "beads on a string" (Fig. 2D). Complexes formed with N-and C-terminally-tagged VirE2 were indistinguishable in the electron microscope. Fusions at either terminus therefore appear to
VirE2 Forms Soluble Complexes When Co-expressed with VirE1-To imitate the natural state of VirE2 in Agrobacterium, we co-expressed it together with VirE1. A soluble VirE1-VirE2 complex was formed (Fig. 3A), which could be concentrated to more than 15 mg/ml. Gel filtration indicated that the complex is monodisperse (Fig. 3B), whereas comparison with calibration standards suggested a molecular mass of 135 kDa (not shown). Because protein migration in size exclusion chromatography depends on molecular shape as well as mass, we performed equilibrium analytical ultracentrifugation to assess the oligomerization state of the VirE1-VirE2 complex. Analysis of a VirE1-VirE2 sample at 1 mg/ml concentration yielded a molecular mass of 67.9 kDa. Other protein concentrations gave similar results, with sample-to-sample variations on the order of 10%, and no systematic trend was found with the sample concentration. The mass calculated from analytical ultracentrifugation experiments is clearly inconsistent with more than one VirE2 protein (63.5 kDa) in the assembly. It was not possible, however, to determine whether one or more VirE1 molecules bound per VirE2.
To determine the VirE1-VirE2 stoichiometry, mass spectrometry measurements were conducted (Fig. 4). The mass spectrum of the complex gave rise to a charge series centered around 3,750 m/z. The measured mass of 72,956 Ϯ 21 corresponds to a 1:1 stoichiometry between VirE1 and VirE2. To confirm the composition of the complex, the 19ϩ charge state was selected for tandem MS experiments. The complex dissociates into a relatively highly charged VirE1 monomer in the low m/z region and a VirE2 protein in the 5,000 -7,000 m/z region. The measured mass for VirE2 is 63,451 Da, which is in close agreement with the mass calculated from the protein sequence.
Structural Characterization of VirE1-VirE2 Complex by SAXS-To gain further insight into VirE1-VirE2 complex structure, SAXS analysis was performed. Fig. 5A presents the background-corrected and processed scattering curve. Because primarily ordered regions contribute to the scattering signal, it can be inferred that VirE2 is maintained in an ordered (i.e. folded) conformation within the complex. The radius of gyration and maximum size (4.2 Ϯ 0.2 and 14 Ϯ 1 nm, respectively) point to a rather anisometric structure. This interpretation is confirmed by the profile of the distance distribution function typical for elongated particles shown in Fig. 5B. Several independent simulated annealing runs of the ab initio program DAMMIN (35) without symmetry imposition (P1) yielded reproducible shapes all neatly fitting the experimental data with discrepancy ϭ 0.90 (solid curve in Fig. 5A). A representative solution is presented in Fig. 5C, showing a suggested molecular outline, or surface, of the VirE2 protein generated from the scattering data.
VirE1 Competes with ssDNA for Binding to VirE2-The interaction of VirE1-VirE2 with ssDNA resulted in the formation of a nucleoprotein complex (Fig. 6A). VirE2-ssDNA complex formation was substantially improved when a crowding agent, polyethylene glycol (PEG 8000), was added to the reaction mixture. The excluded volume environment introduced by the crowding agent (38) thus promoted subunit association and cooperative elongation of the VirE2-ssDNA filament (13,14). In contrast to VirE2 alone, however, 20-mer DNA did not form complexes when mixed with VirE1-VirE2 (Fig. 6B), indicating that complex formation in the presence of VirE1 requires long ssDNA substrates. To determine whether the complex formed by the mixing of VirE1-VirE2 with long ssDNA sub-

VirE1
Modulates T-complex Formation FEBRUARY 9, 2007 • VOLUME 282 • NUMBER 6 strates still contains VirE1, its composition was analyzed using a filtration device with a cutoff of 300 kDa. In such a membrane, the 7.2 kb-M13mp18 ssDNA substrate is retained, whereas the VirE1-VirE2 complex can pass. The retained fraction as well as the two eluted filtrates was probed on a 15% Tricine gel (Fig.  6C). VirE1 was present only in the filtrates, whereas VirE2 appeared in all fractions, demonstrating that VirE1 is displaced by the ssDNA rather than retained in the nucleoprotein complex.
Association of VirE2 to ssDNA in the Presence of VirE1-The interaction of VirE1-VirE2 complex and M13 ssDNA at various mixing ratios was characterized by TEM. Positively stained specimens prepared at sub-saturating VirE2:ssDNA ratios, in which the DNA was not fully coated with VirE2, allowed the detection of both protein and DNA along the same complex (Fig. 7A). In addition to bare and VirE2-coated DNA, we observed compact rings (Fig. 7A, black arrowhead), which can be compared with similar structures seen previously with VirE1-free VirE2 (16). Saturating protein ratios seen by gel shift analysis (Fig. 6A) resulted in continuous coating of the ssDNA by VirE2 (Fig. 7B), with a structure identical to that of complexes formed between VirE2 aggregates and ssDNA (Fig. 2, B and C).

DISCUSSION
Data presented here suggest that VirE1 inhibits the formation of VirE2 aggregates and regulates the assembly of VirE2-ssDNA complexes. In the absence of VirE1, VirE2 tends to selfassociate both in vivo and in vitro (24,25). Indeed, we observed that VirE2 adopted an irregular filamentous form in the absence of ssDNA. This aggregated protein retained its biochemical function, however, as helically ordered filaments formed upon addition of ssDNA of various lengths. In addition, we demonstrated that VirE2 filamentous aggregates adopted the characteristic helical structure at low frequency even in the absence of ssDNA. Thus it appears that ssDNA binding to VirE2 induces a conformational change that affects proteinprotein interactions and that the aggregates represent an active form of this unusual protein.
In previous work we noted the presence of small rings under certain conditions when VirE2 was mixed with ssDNA (16). These rings tended to lie nearby the helical complexes. Image averaging revealed a tetrameric form. As the solenoidal complex contains just slightly more than 4 protomers per helical turn, we speculated that these rings represent a nucleation intermediate and that solenoidal growth extends continuously from an out-of-plane slip in the ring. Here we have detected similar rings when starting with VirE1-VirE2 (e.g. Fig. 7A), clearly in association with ssDNA. Along the same strand we see the assembled helical complex. These observations together with the previous structural characterization of the tetrameric rings support the notion that in response to dissociation from VirE1, VirE2 assembles by nucleation and extension along the DNA template.
VirE2 variants with His 6 fusions at the N or C terminus fail to form extended helical structures but instead take a beads-on-a-string morphology (Fig. 2D) as if the tetrameric rings were stabilized. Therefore both termini of VirE2 appear to be involved in proteinprotein contacts along the helical filament. Indeed, previous studies indicated that VirE2-VirE2 interac- A, the intact complex is observed between 3,500 and 4,250 m/z. Its measured mass (72,956 Da) confirms a 1:1 stoichiometry between VirE1 and VirE2. B, the 19ϩ charge state was selected for tandem MS (shadowed) to confirm the stoichiometry. The complex dissociates into a relatively highly charged VirE1 monomer in the low m/z and a VirE2 protein in the 5,000 -7,000 m/z region. The calculated mass for VirE2 is 63,451 Da. Two species of VirE1 can be detected, with masses of 9,569 and 9,378 Da, indicating coexistence of wild type and a 3-residue C-terminal truncated (⌬LAG) form of VirE1 with TEV cleavage site, His 6 tag, and spacers at its N terminus.

VirE1 Modulates T-complex Formation
tions require both N-and C-terminal sequences (25). Fusions to either terminus of VirE2 result in an avirulent strain of Agrobacterium (26), and the integrity of the C-terminal region in particular is essential for the VirB-mediated transport of VirE2 into plant cells (39). In the three-dimensional structure of the VirE2-ssDNA complex, the VirE2 protomers are oriented with the long axis of the protein wrapping tangentially around the hollow core (16). Combined with the evidence above, this ordering suggests that VirE2 self-association proceeds in a head-to-tail manner involving both termini.
The formation of VirE2 aggregates in vitro demonstrates the important role of VirE1 in regulating VirE2 self-association in Agrobacterium. Such aggregates, although functional in terms of T-strand binding and T-complex formation, are not likely to translocate from the bacterium to the plant through the VirB/D4 channel. Several lines of evidence indicate that VirE1 plays a central role in preserving VirE2 in a competent state for translocation and binding to the T-strand (7, 24 -27, 40, 41). In the absence of VirE1, VirE2 accumulates in the bacterium but is unable to move into the plant cell (27). Furthermore, overexpression of VirE1 in planta results in reduced susceptibility to Agrobacterium infection 3 (42). Here we have demonstrated that VirE1 preserved VirE2 in a nonaggregated form. In agreement with another recent study (43), we found that VirE1 was dissociated from VirE2 upon competitive binding of VirE2 to ssDNA.
The formation of VirE2-ssDNA complexes was substantially promoted in the presence of crowding agents. Because VirE1 competes with ssDNA for the binding to VirE2, removal of VirE1 from VirE2 can be achieved by an effectively higher concentration of ssDNA, mimicked in vitro by the addition of PEG 8000 (38,44). Prior studies have shown that VirE2 interacts more strongly with VirE1 than with itself, using both in vitro binding and LacZ reporter gene assays (24), and that the C-terminal ssDNA binding region of VirE2 also interacts with VirE1 (25). In addition, we have shown here that VirE2 was not transferred from VirE1-VirE2 complexes to 20-mer ssDNA. Altogether, the evidence indicates that dissociation of VirE1 from VirE2 in response to ssDNA binding may be efficient only as cooperative self-association of VirE2 on the single-stranded DNA template is induced.
Macromolecular crowding stabilizes protein-protein interactions and enhances the formation of rod-like aggregates of certain fiber-forming proteins such as amyloids (45) and the bacterial FtsZ (46). VirE2-ssDNA is clearly stabilized in vitro. However, the VirE1-VirE2 complex may also be stabilized in a crowded environment. Efficient dissociation of VirE1 from VirE2 would be preferred at high relative concentrations of ssDNA. In the bacterium, the tumor-inducing plasmid exists as a single copy (47,48) so that only one molecule of T-strand is likely to be present at any given time. VirE2, on the other hand, may become the most abundant protein in the bacterium following activation (29,49). This scenario results in an extremely low concentration of ssDNA relative to the VirE1-VirE2 complex, such that the binding of VirE2 to tDNA is unlikely to occur in the Agrobacterium cytoplasm. Where, then, does VirE2 associate with the T-strand in vivo?
Combining our observation of the VirE1-regulated polymerization of VirE2 on ssDNA with the effect of crowding on this exchange process, we suggest that the association between VirE2 and tDNA may occur in the VirB/D4 channel or near its exit. Although the type IV secretion channel has not yet been resolved structurally, it certainly represents a tightly constrained volume. As the ssDNA threads through, its local concentration will be relatively much higher than in the larger volume of either adjoining cell. Dissociation of VirE1 would be promoted under such conditions. VirE1 should remain in the bacterium, in agreement with previous observations (27). Because VirE2 was not observed to interact with tDNA within the channel of induced but nonadherent agrobacteria (8), secretion is an apparent requisite for binding between the substrates. The trans side of the channel may therefore be the site 3 W. Ream, personal communication.  1B). Lane 1, 20-mer without VirE2; lanes 2-5, VirE2:20-mer at w/w ratios 1:1, 5:1, 10:1, and 15:1, respectively. C, interaction of VirE1-VirE2 complexes with M13 ssDNA. Complexes were prepared at a 20:1 VirE1-VirE2:ssDNA w/w ratio. Composition was assayed by filtering through a 300-kDa NanoSep separation membrane. The complex was filtered twice, and the concentrate as well as the filtrates was analyzed on a 15% PAGE-Tricine-SDS gel and detected by silver staining. lane 1, concentrated sample; lanes 2 and 3, filtrates.  FEBRUARY 9, 2007 • VOLUME 282 • NUMBER 6 of interaction. VirE2 association with the T-strand could then provide a ratchet mechanism to prevent its return to the bacterium, akin to retention models proposed for translocation of DNA through the nuclear pore (50,51).

VirE1 Modulates T-complex Formation
The ssDNA binding activity of the aggregated form of VirE2 demonstrated here neatly explains the artificial transformation assays where tDNA and VirE2 arrive from different sources (9,10), as aggregates in the plant cytosol should retain normal function in DNA binding. However, T-complex formation in the cytoplasm leaves the tDNA susceptible to degradation by cytoplasmic nucleases prior to VirE2 binding, which may be detrimental under natural conditions. By the proposed mechanism of regulation by VirE1, VirE2 self-association and premature binding to the T-strand is prevented inside the bacterium, whereas the T-complex arrives to the plant cytosol fully assembled, ready to interact with host co-factors that guide it into the nucleus.