The Agrobacterium tumefaciens Chaperone-Like Protein, VirE1, Interacts with VirE2 at Domains Required for Single-Stranded DNA Binding and Cooperative Interaction

Abstract

Agrobacterium tumefaciens transfers single-stranded DNA (ssDNA) into plants. Efficient tumorigenesis requires VirE1-dependent export of ssDNA-binding (SSB) protein VirE2. VirE1 binds VirE2 domains involved in SSB and self-association, and VirE1 may facilitate VirE2 export by preventing VirE2 aggregation and the premature binding of VirE2 to ssDNA.

Agrobacterium tumefaciens transfers part of its tumor-inducing (Ti) plasmid, the T-DNA, into the plant nuclear genome (10). Border sequences delimit the T-DNA (41, 42, 49, 60). T-DNA transfer requires vir genes located elsewhere on the Ti plasmid (23, 52). The VirD1-VirD2 endonuclease nicks border sequences, and VirD2 attaches to the 5′ end of the nicked strand (50). Subsequent events produce single-stranded DNA (ssDNA) called T strands (2, 31, 53).

A transmembrane channel, comprising 11 VirB proteins and VirD4, exports VirD2-T strand complexes and VirE2 into plants (11, 39, 50, 57, 62, 63). VirB proteins resemble Legionella pneumophila Icm (also called Dot), Helicobacter pylori Cag, and Bordetella pertussis Ptl toxin export proteins (1, 46, 47, 59, 62). Rickettsia prowazekii contains virD4 and virB homologs (3). Proteins that mediate plasmid conjugation resemble VirB and VirD4 (5, 32, 34, 35, 62). The VirB-VirD4 channel mediates plasmid conjugation from A. tumefaciens into other bacteria (6, 21, 22, 25, 56), plants (8), or fungi (9, 17). L. pneumophila Icm and Dot proteins mobilize plasmids into other bacteria (46, 59). This family of transmembrane proteins exports proteins and protein-DNA complexes into a broad range of recipients.

VirE2 binds ssDNA cooperatively (14, 48), contains two bipartite nuclear localization signals (NLSs) (15), and plays an important role in tumorigenesis (24, 45). The C-terminal half of VirE2 is essential for ssDNA binding (SSB) (15, 19, 48). Some alterations in the N-terminal half of VirE2 retain their function, while others affect cooperativity (15, 19). VirE2 protects T strands from nuclease attack (14, 45, 48, 63), but the absence of VirE2 does not diminish T-strand accumulation in A. tumefaciens (45, 54, 58). The NLSs, which overlap the SSB domain, function when VirE2 binds ssDNA (65). VirD2 NLS mutants transform plants, implicating VirE2 in the nuclear uptake of T strands (38, 51). Transgenic VirE2-producing tobacco plants are transformed by a virE mutant of A. tumefaciens (15), showing that VirE2 is necessary only inside plant cells.

VirE2 export requires VirE1 (57). Although VirE2 is stable in A. tumefaciens without VirE1 (57), VirE1 stabilizes VirE2 in Escherichia coli (37), suggesting that these proteins interact. We used protein interaction cloning in yeast (Saccharomyces cerevisiae) to study this interaction. Recently, Deng et al. reported similar data (18).

Protein interaction assays.

Table 1 describes the yeast strains used. Bait fusions contained virE2 sequences fused to lexA in pEG202 (26). Prey fusions contained virE1 or virE2 fused to a transcriptional activator (TA) domain in pJG4-5 (26); pSH18-34 contains lacZ downstream of the GAL1 promoter and four lexA operators (26). Recruitment of prey protein to the operators via interaction with the bait protein induces lacZ (26). β-Galactosidase activity indicated protein interaction (26, 43).

TABLE 1.

Yeast strains^a

CSY strain		Bait fusion (virE2 codons)	Prey fusion (codons)	Lac phenotype (test strain/control strain)
Test	Control
1	41	2–533b	virE1	+/−
5	45	2–242b	virE1	+/+
6	46	2–349b	virE1	+/+
20	62	326–533b	virE1	−/−
21	63	288–533b	virE1	+/−
22	67	288–495c	virE1	+/−
23	70	288–495b	virE1	+/−
24	71	242–495b	virE1	+/−
28	41	2–533b	virE2 (2–533)	+/−
29	43	190–533b	virE2 (2–533)	−/−
37	65	2–495c	virE2 (2–533)	−/−
39	68	2–495b	virE2 (2–533)	−/−
76	84	190–436b	virE1	−/−
78	86	242–436b	virE1	−/−
104	126	288–495::Xho 378c	virE1	−/−
109	131	288–495::Xho 378b	virE1	−/−
110	132	288–495::Xho 472b	virE1	−/−
155		46–349b	virE1	−/ND
157		46–533b	virE2 (2–533)	−/ND
165	173	2–533::Xho 378b	virE1	+/−
166	174	2–533::Xho 472b	virE1	+/−
168	176	2–386c	virE1	+/−
169	177	190–495b	virE1	+/−
170	178	190–495c	virE1	+/−
181	41	2–533b	virE1 (25–65)	+/−
182	67	288–495c	virE1 (25–65)	+/−
184	173	2–533::Xho 378b	virE1 (25–65)	+/−
185	174	2–533::Xho 472b	virE1 (25–65)	+/−
186	126	288–495::Xho 378c	virE1 (25–65)	−/−
187	127	288–495::Xho 472c	virE1 (25–65)	−/−

^aPrey plasmids in test strains contained either full-length virE1, the 41 C-terminal codons (25 to 65) of virE1, or virE2 (codons 2 to 533) fused in frame to the vector’s TA coding sequence. Control strains contained the prey vector plasmid, pJG4-5, with no DNA inserted. Lac phenotype is shown as present (+) or absent (−). ND, not determined. 

^bBait fusions were made in pEG202. 

^cBait fusions were made in pCD2M1. 

Plasmid construction.

Portions of virE2 were amplified by PCR; primers contained NcoI (downstream) and BamHI (upstream) sites. virE2 contains BglII and BclI sites phased to allow their use instead of the BamHI site in the upstream primer. Some primers contained a BsaI site that generated NcoI- or BamHI-compatible ends. We inserted virE2 sequences into pEG202 at the BamHI and NcoI sites. Duplicate bait constructs were prepared in pCD2M1, a pEG202 derivative containing a simian virus 40 NLS (PKKKRKV [44]) fused to lexA at the EcoRI and BamHI sites. Bait fusions in both plasmids gave equivalent results. The TA sequence was fused to full-length virE1 or virE2 or truncated virE1 by inserting PCR amplicons with EcoRI- and XhoI-compatible ends into pJG4-5 at these sites.

PCR was performed with Stratagene Pfu DNA polymerase with a template (pGR1) containing the virE operon (57). Thermocycle ligations (36) were performed with T4 DNA ligase; DNA was transformed into E. coli as previously described (4). All gene fusions were sequenced by the dye terminator method (4). virE2 differed from the published sequence (61), with G-C and C-G transversions at positions 1073 and 1074. These changes were present in pGR1. This double mutation changed codon 358 from cysteine to serine, the residue present in nopaline-type VirE2 (27).

VirE1 interacts with VirE2.

A bait fusion containing full-length VirE2 interacted with VirE1 prey (CSY1) (Fig. 1 to 3). This LexA-VirE2 protein did not activate lacZ in yeasts that harbor pJG4-5 without virE1 (Fig. 2), nor did VirE2 interact with the TA domain encoded by pJG4-5. Several bait fusions with only a portion of VirE2 did not interact with the TA-VirE1 prey (Fig. 1 and 2), proving that VirE1 did not bind the LexA moiety of the bait. The LexA-VirE1 fusion alone, in cells devoid of a prey plasmid, activated lacZ (data not shown). This often occurs when LexA is joined to an acidic protein (26) such as VirE1 (pI 4.7 [61]). VirE1-VirE2 interactions were characterized with virE2 bait fusions and virE1 prey.

FIG. 1.

Interaction of VirE2 bait proteins and VirE1 prey fusion proteins. The large box represents the genetic and physical map of VirE2. Domains involved in cooperativity (Coop.) and DNA binding (ssDNA binding) are overlined. The C-terminal VirE1-binding domain is underlined. Numbers above the map indicate codons. Triangles mark XhoI linker insertions after codons 378 and 472. Hatched boxes on the map represent the NLSs. Bars show the region of VirE2 present in each bait fusion protein. Solid bars indicate VirE2 bait-VirE1 prey interaction (shown under the column at the left as a +), open bars indicate no interaction (shown by a −), and the hatched bar indicates VirE2 bait self-activation (shown by s.a.), in which the bait plasmid activated transcription of lacZ in the presence of pJG4-5 (without virE1 sequences).

FIG. 3.

Interaction of VirE2 bait proteins with truncated or full-length VirE1 prey proteins. On the left, open bars show regions included in VirE2 bait and VirE1 prey proteins; numbers (Ser288, etc.) indicate VirE1 or VirE2 codons included in each fusion. Triangles mark XhoI linker insertions after codons 378 and 472. β-gal, β-galactosidase activity, indicated as positive (+) or negative (−). Panels A to H (right) indicate the corresponding β-galactosidase assays.

FIG. 2.

Interaction of VirE2 bait proteins with VirE1 or VirE2 prey proteins; β-galactosidase assays. The top bar on the left represents the map of VirE2; hatched boxes are NLSs, triangles indicate XhoI linker insertions after codons 378 and 472, and the C-terminal VirE1-binding domain is underlined. Bars show the regions of VirE2 present in each bait fusion protein. Solid bars indicate VirE2 bait-VirE1 prey interaction, open bars indicate no interaction, and the hatched bar indicates VirE2 bait self-activation. On the right, the Bait/Prey column shows β-galactosidase assays of yeast strains that harbored both bait and prey fusion plasmids, whereas the Bait column shows results for strains that harbored a bait fusion plasmid together with pJG4-5. Dark patches indicate β-galactosidase activity corresponding to VirE2 bait-VirE1 prey interaction. Light patches indicate no activity. VirE2 bait self-activation is shown by dark patches in both the Bait/Prey and Bait columns. Vertical bars indicate whether VirE2 bait fusions were tested with VirE2 (top panel) or VirE1 (lower panels) prey fusions. Numbers on the left (Ile 190, etc.) indicate the VirE2 codons included in the bait fusion.

N-terminal VirE1-binding domain.

VirE2 residues 2 to 386 interacted with VirE1 but not with the TA domain of pJG4-5 (CSY168) (Fig. 1). Baits containing shorter segments of VirE2 (residues 2 to 349 or 2 to 242) activated lacZ in the presence of pJG4-5 (CSY46 and CSY45) (Table 1; Fig. 2), which prevented further mapping of this domain’s distal end. The first 73 residues of VirE2, although poorly conserved (27, 61), were required for the interaction of this domain with VirE1; residues 46 to 349 did not interact with VirE1 (CSY155) (Fig. 1). Because VirE2 residues 2 to 386 contain cooperativity and VirE1 interaction domains, VirE1 binding at the VirE2 N terminus may prevent VirE2 self-association by occupying domains crucial for self-interaction. Indeed, VirE1 prevents VirE2 aggregation in vitro (18).

C-terminal SSB and VirE1-binding domains overlap.

VirE2 residues 288 to 533 interacted with VirE1, but residues 326 to 533 did not (CSY21 and CSY20) (Fig. 1), indicating that residues between 288 and 326 were necessary for binding VirE1. The C-terminal boundary of this VirE1-binding domain was established with VirE2 fusions that extended from residue 190 toward the C terminus. One fusion extended through residue 495 and interacted with VirE1 (CSY169) (Fig. 1). VirE2 residues 190 to 436 failed to interact with VirE1 (CSY76) (Fig. 1 and 2), indicating that residues between 436 and 495 were required for binding VirE1. Thus, residues 288 to 495 encompass this VirE1-binding domain. CSY23 contained this VirE2 bait and interacted with VirE1 (Fig. 1 to 3), confirming that a VirE2 domain within residues 288 to 495 bound VirE1. Mutations that abolish SSB activity define a domain that includes residues 287 to 497 (13, 19, 48). The smallest VirE2 bait that bound VirE1, residues 288 to 495, corresponds to the SSB domain (Fig. 1). XhoI linker insertions in the SSB domain destroyed SSB activity (19) and abolished its interaction with VirE1 (CSY109 and CSY110) (Fig. 1 to 3). VirE1 and ssDNA interact with VirE2 features disrupted by the same mutations, so VirE1 may compete with ssDNA to bind VirE2. Full-length VirE2 baits containing these insertions interacted with VirE1, indicating that these mutations did not affect the N-terminal VirE1-binding domain (CSY165 and CSY166) (Fig. 1 and 3).

Bait proteins are stable.

Yeast lysates contained VirE2 bait proteins detectable in immunoblots probed with antibodies to LexA (data not shown).

The VirE1 C terminus interacts with VirE2.

VirE1 contains two conserved domains, residues 1 to 28 and 35 to 62 (27, 61). We fused virE1 codons 25 to 65 to the TA domain in pJG4-5. This VirE1 prey interacted with bait proteins containing either full-length VirE2 or the SSB–VirE1-binding domain (Fig. 3D and H). Bait fusions containing the SSB–VirE1-binding domain disrupted by an XhoI linker insertion did not interact with the truncated VirE1 prey (Fig. 3B and C). In full-length VirE2, these insertions did not affect the binding of truncated VirE1 to the VirE2 N terminus (Fig. 3F and G). Thus, truncated and full-length VirE1 preys behaved the same. Because the same region of VirE1 binds both protein interaction domains of VirE2, separate VirE1 molecules probably bind each domain.

VirE2-VirE2 interaction requires N- and C-terminal sequences.

VirE2 contains two regions important for cooperativity (15, 19). Mutations in these regions affect cooperativity without abolishing SSB activity, implicating them in self-interaction. We fused codons 2 to 533 to the TA domain in pJG4-5, creating a full-length VirE2 prey which interacted with full-length VirE2 bait (CSY28) (Table 1; Fig. 2). Removing the first 189 residues of the VirE2 bait eliminated one cooperativity domain and abolished VirE2-VirE2 interaction (CSY29) (Table 1). VirE2 bait that included both cooperativity domains (residues 46 to 533) failed to bind full-length VirE2 prey (CSY157) (Table 1). VirE2 bait (residues 2 to 495) missing the last 38 residues did not interact with VirE2 prey (CSY37 and CSY39) (Table 1). VirE2-VirE2 interaction required both N- and C-terminal VirE2 sequences outside known cooperativity domains.

VirE2 and T-strand transfer may occur independently.

VirE2 export requires VirE1, but T-strand transfer does not (57). VirB-VirD4-dependent mobilization of RSF1010 from A. tumefaciens into plants abolishes tumorigenesis by preventing VirE2 export; RSF1010 reduces but does not eliminate T-strand transfer (7, 55). The oncogenesis-suppressing Osa protein inhibits VirE2 export but not T-strand transfer (33). Thus, T-strand transfer occurs even though VirE2 export is blocked. A. tumefaciens lacking VirE1 and VirE2 transfers T strands into plants (24, 57, 63), and VirE2 export does not require T-strand production (12, 40, 57). Tumors form readily when a single plant wound is inoculated with two nonpathogenic strains of A. tumefaciens: one lacking T-DNA and another lacking virE2 (12, 40, 57). Because both strains must bind plant cells (12), VirE2 and T strands are probably exported directly, and independently, into plants.

VirE1 is a secretory chaperone for VirE2.

Others have suggested that VirE1 may promote the proper folding of VirE2 (64). This seems unlikely because transgenic plants express functional VirE2 in the absence of VirE1 (15). Instead, VirE1 may act as a molecular escort of VirE2. Because VirE1 binds VirE2 at sites critical for SSB and self-interaction, VirE1 may hinder these interactions. VirE1 may promote the export of VirE2 by limiting its association with ssDNA and by preventing VirE2 aggregation. Similar molecular escorts participate in protein secretion in many pathogenic bacteria, where they prevent the premature association of secreted proteins with each other or with the translocation machinery (16, 20, 28). Yersinia outer membrane protein (Yop) export requires a specific Yop chaperone (Syc), which binds its target Yop (29, 30). The physical properties of Syc proteins resemble those of VirE1: all are small (123 to 168 amino acids), acidic (pI 4.5 to 4.88) proteins with a hydrophobic C terminus (16). Syc proteins release their Yop as it exits the cell (30). The export of monomer subunits that form a complex following secretion is a common strategy for assembling large extracellular complexes. If formed inside the cell, complexes might be too large to pass through the transmembrane channel. Secretory chaperones promote protein export by preventing the assembly of large complexes in the cytoplasm. VirE1 appears to act as a secretory chaperone for VirE2.