![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,
Julieta Aguilar,
Doyle Ward,
and
Patricia Zambryski*
Department of Plant and Microbial Biology, Koshland Hall, University of California, Berkeley, Berkeley, California 94720-3102
Received 29 March 2007/ Accepted 29 June 2007
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
An understanding of the structure, as well as the operation, of the vir-T4SS is emerging (14, 15). This apparatus is assembled from 11 proteins encoded by the virB operon and an additional coupling protein, VirD4, that brings the T complex to the T4SS for export. Genetic and biochemical studies suggest that the core of the apparatus is a transmembrane complex composed of VirB6, VirB7, VirB8, VirB9, and VirB10 (10, 36, 42, 44). Energy for assembly and function of the vir-T4SS is provided by three ATPases—VirB4, VirB11, and VirD4 (1, 9, 17, 18, 43, 65).
The other structural feature of the vir-T4SS is the T pilus (31). The major component is VirB2, an F-pilin homolog (50), and VirB5 is an additional minor component (62). VirB7 also has been detected in high-molecular-weight, extracellular structures purified by centrifugation (60). Early steps in T-pilus assembly involve the intramolecular cyclization of processed VirB2 monomers (28), the formation of VirB2-VirB5 complexes in the membranes, and the formation of VirB7 homodimers (47). VirB4 interactions withVirB3 and VirB8 also facilitate T-pilus formation (74). However, exactly how VirB2 is mobilized from the inner membrane to the site of T-pilus biogenesis is not known. It has been speculated (49, 51) that chaperones and ushers, such as those that regulate the assembly of P pili (68), may deliver T-pilin subunits through the outer membrane.
VirB1, the first product of the virB operon, is an incompletely characterized vir-T4SS component. This lack of interest may stem from reports that VirB1 is not absolutely required for DNA transfer: whereas deletion of other VirB genes completely abolished DNA transfer, deletion of virB1 reduced virulence to 
3% (6, 8, 29). Nevertheless, these latter data suggest that VirB1 plays a significant role in Agrobacterium-mediated DNA transfer.
Sequence similarity between the N terminus of VirB1 and chicken egg white lysozyme and bacterial lytic transglycosylases suggests this protein provides local lysis of the peptidoglycan cell wall to create the space required for assembly of a complex as large as the T4SS (21, 45, 54). Recent data confirm this prediction by demonstrating VirB1 transglycosylase activity in vitro (75). Interaction between VirB1 and VirB8, as well as VirB9 and VirB10, may target the lytic transglycosylase activity of VirB1 to the site of T4SS assembly (72) and suggests that VirB1 has a periplasmic function. The C-terminal third of the protein, VirB1*, is cleaved from VirB1 and secreted by an unknown mechanism (3). In Brucella suis, the C terminus of VirB1, although not processed, interacts with VirB9 (a periplasmic protein), which again suggests a periplasmic function for this VirB1 homolog (37). Complementation of an in-frame deletion of virB1 with constructs expressing either the N-terminal lytic transglycosylase domain or C-terminal VirB1* resulted in tumors intermediate in size and frequency compared to the wild type (53). Thus, each domain contributes an independent function to virulence.
Here we provide genetic evidence that VirB1* is an essential factor for T-pilus assembly. A nopaline virB1 deletion strain of Agrobacterium fails to mobilize VirB2 and VirB5 to the cell exterior for T-pilus biogenesis. VirB2 and VirB5 are restored to isolated T pili when the deletion is complemented with full-length VirB1 or VirB1* but not with the N-terminal lytic transglycosylase domain. Protein-protein interactions between VirB1* and both VirB2 and VirB5 were detected in the yeast two-hybrid assay. At least 50% of VirB1* is exported to the cell exterior (3), and we show that a portion of extracellular VirB1* copurifies with the T pilus. Thus, VirB1* may promote T-pilus assembly by first binding VirB2 and VirB5 and then acting as a chaperone for T-pilus subunits as they are mobilized to sites of T-pilus assembly and subsequently to the cell exterior.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
Construction of vectors for expression of wild-type VirB1 and VirB1 mutants.
All procedures for plasmid DNA isolation and manipulations, such as digestion with restriction endonucleases or ligation, were performed as described previously (61). For complementation studies, pDW029 was constructed to enhance vir-regulated protein expression. The ribosome-binding site (RBS) sequence and the XhoI and ClaI sites of pBP21 (53) were replaced with the RBS corresponding to A. tumefaciens 16S ribosomal DNA. In the introduced sequence, 5'-CTCGAGGAGGAGGTTTGTCATGATCGAT-3', the RBS is in italics, and a BspHI site spanning the translation initiation codon (in boldface) is underlined. This plasmid carries the pVS1 origin of replication (40) for replication in A. tumefaciens. Protein expression from this promoter is 
3-fold higher relative to pBP21 as estimated by green fluorescent protein (GFP) expression (J. Zupan, D. Ward, and P. Zambryski, data not shown).
pDW101 is a pDW029 derivative that encodes full-length VirB1 from pTiC58. The virB1 coding sequence was PCR amplified from pGK217 (48) such that a BspHI site was introduced at the initiation codon and a PstI site introduced 3' of the termination codon. Oligonucleotides DWB1Bs165 and DWB1P_140 were used as primers for this PCR. The PCR fragment was first cloned into pCR-BluntII-Topo (Invitrogen) and then cloned as a BspHI-PstI fragment into pDW029 to create pDW101. The protein encoded by this plasmid is referred to as VirB1 throughout the text.
Amino acid substitutions and deletions in VirB1 were produced by inverse PCR (56) using pDW101 as the template. PCR products were generated with oligonucleotides described in Table 1 and then digested with appropriate restriction endonucleases, followed by ligation. Constructs were propagated in Escherichia coli and confirmed by sequencing.
pJZ034 encodes the VirB1 N-terminal 179 amino acids (aa), which correspond to the signal peptidase I (SPI) domain that targets the protein to the general secretory pathway (GSP) and the lytic transglycosylase domain. A stop codon was introduced after the first alanine of the VirB1* cleavage site. Oligonucleotides LysF029 and LysR029 were used as primers for inverse PCR to produce this plasmid. The protein encoded by this plasmid is referred to as VirB1-LT throughout the text.
pJZ036 encodes a protein with a large internal deletion in the VirB1 lytic transglycosylase domain. The first 10 aa after the SPI domain and the 10 aa N-terminal to the VirB1* cleavage site were retained to provide sequence context at these processing sites. Oligonucleotides B1*F029 and B1*R029 were used as primers for inverse PCR to produce this plasmid. The protein encoded by this plasmid is referred to as VirB1* throughout the text.
pJZ038 encodes VirB1 without the SPI domain. Oligonucleotides SPF029 and SPR029 were used as primers for inverse PCR to produce this plasmid. The protein encoded by this plasmid is referred to as VirB1
SP.
pJZ040 encodes VirB1 with two amino acid substitutions at the VirB1* cleavage site. Alanine 179 and alanine 180 were changed to proline and arginine, respectively, by the use of the nucleotide sequence for the restriction endonuclease AvrII. Oligonucleotides B1no*F029 and B1no*R029 were used as primers for inverse PCR to produce this plasmid. The protein encoded by this plasmid is referred to as VirB1(AA/PR) throughout the text.
pJZ042 encodes VirB1 with an amino acid substitution in the lytic transglycosylase active site. Glutamic acid 60 was changed to glutamine. This mutation is in a domain conserved among transglycosylases and thought to form the catalytic site (54). When this mutation has been introduced into other transglycosylases, enzymatic activity was abolished. Oligonucleotides LysE60QF029 and LysE60QR029 were used as primers for inverse PCR to produce this plasmid. The protein encoded by this plasmid is referred to as VirB1(E/Q) throughout the text.
pJZ076 is a pDW029 derivative that encodes nopaline VirB1 with a single amino acid change in the VirB1* processing site. This change, glutamine 187 to serine, produces a contiguous stretch of 19 aa that are identical to the sequence of octopine VirB1. Oligonucleotides B1QtoS.F and B1QtoS.R were used as primers for inverse PCR to produce this plasmid. The protein encoded by this plasmid is referred to as VirB1(Q/S) throughout the text.
pJZ079 is a pDW029 derivative that encodes octopine VirB1 with a hemagglutinin epitope tag at the carboxy terminus. Oligonucleotides MTX19 and OctB1HA.R2 were used as primers for PCR to introduce the coding sequence for the hemagglutinin epitope tag at the 3' of the octopine virB1 gene. The template for this PCR was pMTX124 (53). The PCR product was subcloned by using a topoisomerase vector (pCR-BluntII-Topo; Invitrogen). The intermediate plasmid was digested with HindIII and PstI, and the gel-purified insert was ligated into similarly digested pDW029.
pJZ093 is a pDW029 derivative that encodes octopine VirB1 with an internal hemagglutinin epitope tag between aa 188 and 189. Oligonucleotides FOctB1NcoI, 5OctB1IntHA.R, 3OctB1IntHA.F, and 3OctB1IntHA.R were used as primers for overlap extension PCR. The template was pMTX124. The PCR product was subcloned into pCR-BluntII-Topo (Invitrogen). The intermediate plasmid was digested with NcoI and PstI, and the gel-purified insert was ligated into pDW029 digested with BspHI and PstI.
Protein analysis. Preparation of the cell lysates and supernatant fractions by precipitation with acetone, as well as analysis of VirB1 products by gel electrophoresis and Western blotting, was performed as previously described (3). T-pilus isolation was performed as described earlier (38). Polyclonal anti-VirB2 antibody was generated against the peptide QSAGGGTDPATMVNN (aa 48 to 62, the N-terminal 15 aa of the mature VirB2 T pilin).
The well-documented and established procedure for isolating T pili is to remove surface structures from vir-induced cells by application of shearing forces such as are generated by passing a concentrated cell suspension through a small-diameter needle (2, 38, 41, 49, 50, 63). Low-speed centrifugation then separates bacterial cells from any material liberated from the cell surface by the shear forces generated in the syringe. Purification of high-molecular-weight complexes or structures is achieved by high-speed centrifugation of the supernatant from the first centrifugation. This procedure results in the enrichment of the major and minor T-pilus subunits, VirB2 and VirB5, in the high-speed pellet from a vir-induced culture of wild-type Agrobacterium. Electron microscopy images have confirmed the purification of filamentous structures specifically from vir-induced cells with dimensions that correspond to T pili (62).
Isolation of periplasmic proteins was based on the procedure of De Maagd and Lugtenberg (19). Briefly, cells were harvested and resuspended in 50 mM potassium phosphate buffer (pH 5.5) and then pelleted by centrifugation at 7,000 x g. An aliquot of total cells was frozen in denaturing sample buffer. The remaining cells were resuspended in an osmotic shock solution of 50 mM Tris (pH 8), 20% sucrose, 2 mM EDTA, 87 µg of phenylmethylsulfonyl fluoride/ml, and 200 µg of lysozyme/ml, followed by incubation at room temperature for 30 min. The mixture was centrifuged at 7,000 x g, and the supernatant (periplasmic fraction) was concentrated by acetone precipitation.
Sequence comparisons. To determine whether other bacterial species showed conservation of the VirB1* domain, we first performed a PSI-BLAST search with A. tumefaciens (strain C58) VirB1 (NP_536285.1) residues 168 to 252 against the nonredundant database. In the second iteration of the search, 12 sequences were identified with an E-value of 0.005 or better, including a probable VirB1 protein from M. loti (CAD31462). No new sequences were found in subsequent iterations.
To search more broadly for homologs, we aligned A. tumefaciens (C58) VirB1, the M. loti probable VirB1, and the B. abortus VirB1 (AAF73894) by using MUSCLE (27). The alignment was cropped to exclude all but the C-terminal domain of VirB1, from aa 154 to the end, using the alignment editor Jalview (16). A hidden Markov model (HMM) was constructed by using the program Hmmbuild on the cropped alignment (26). Sequences matching the HMM were selected from the Uniprot database by using the Hmmsearch program and aligned to the model by using the Hmmalign program. The new alignment was cropped to the relevant region, edited to remove redundancy, and then used as the basis for building another HMM. The process was repeated, resulting in a final multiple sequence alignment of 27 homologs.
Yeast two-hybrid analysis.
Protein-protein interactions were characterized by the yeast two-hybrid system (25). Coding sequences for full-length fragments of VirB1 were cloned in fusion to the Gal-4 DNA-binding domain in either the CEN vector, pCD.2 (72), or the 2µm vector, pAS2 (24). Coding sequences for full-length protein or fragments of VirB2 were cloned in fusion to the Gal-4 activation domain in the CEN vector, pC-ACT.2 (25, 72) or the 2µm vector, pACTII (24). To construct pCD.B1*+44, the nucleotide sequence encoding VirB1* and the 44 aa N-terminal to the VirB1* cleavage site was PCR amplified from pAS2.B1SP with oligonucleotides CH010 and CH013 (Table 1). The restriction site for AvrII was introduced at the 5' end of the coding sequence and PstI at the 3' end of the PCR product. The PCR product was purified, digested with AvrII and PstI, and ligated into similarly restricted pCD.1 to produce pCH001. In addition, laboratory stock plasmids pAS2.B1SP and pACTII.B5SP were constructed as described previously (4). Briefly, the coding sequence for VirB1SP and VirB5SP were PCR amplified, digested with appropriate restriction endonucleases, and ligated to pACTII (24). Yeast two-hybrid plasmids were transformed into yeast strain YD116 (25), and interactions were detected by growing transformants on appropriate selective media (25). Interactions were quantified by ß-galactosidase activity using chorophenol-red-ß-D-galactopyranoside substrate (CPRG; Roche Chemical) as described previously (24).
Transmission electron microscopy. Bacterial cells were adjusted to an A600 of 2.0 in 50 mM potassium phosphate buffer (pH 7) and deposited on carbon-Formvar films on 200 mesh, 3-mm copper grids. Then, 10 µl of bacterial suspension was placed on each grid for 2 to 3 min to allow bacteria to settle on the grid. Excess cell suspension was then wicked away, and each grid was washed three times by floating it on a drop of water for a few seconds. Grids were then stained with 1% phosphotungstic acid-0.01% glucose (pH 7) for 1 min. The specimens were examined in a FEI Tecnai 12 transmission electron microscope.
A. tumefaciens C58 cells were grown, and vir gene expression was induced with 200 µM AS. After 24 h of induction the bacterial cells were transferred to copper disks (50 µm thick, 1.44 mm in diameter) and frozen by using a Bal-Tec HPM 010 high-pressure freezer. Frozen samples were maintained under liquid nitrogen and transferred into a 2-ml cryovial containing 0.2% glutaraldehyde and 0.1% uranyl acetate in anhydrous acetone. The samples were placed into the Bal-Tec HPM 010 unit, freeze substituted under controlled temperatures, and then brought up to 0°C over 72 h. Samples were covered in LR white resin and polymerized in a microwave at 650 W for 45 min. Embedded samples were sectioned with the Ultramicrotome-MT 6000. Ultrathin sections (60 nm) were collected on Formvar 100 mesh nickel grids coated and treated by ion discharge in a vacuum evaporator.
For immunogold labeling, grids containing bacterial sections were sequentially placed on a 20-µl drop of the following solutions: (i) 0.1% glycine for 15 min to quench unreacted aldehyde groups; (ii) 5% normal goat serum (blocking agent) for 15 min; (iii) primary antibody (
-VirB1) diluted 1:500 in blocking agent at 4°C overnight; (iv) three washes in 0.5 M NaCl in phosphate-buffered saline (PBS) of 5 min each; (v) five washes in PBS-0.2%Tween for 3 min each; (vi) five washes in PBS for 3 min each; (vii) secondary antibody buffer (1% bovine serum albumin, 0.1% cold water fish gelatin, 0.25% Tween 20) for 15 min; (viii) secondary antibody (10-nm gold conjugated goat anti-rabbit) diluted 1:200 in secondary antibody buffer for 60 min at room temperature; (ix) five washes in PBS-0.2% Tween for 5 min each; (x) five washes in PBS for 3 min each; (xi) 0.5% glutaraldehyde in PBS for 10 min; (xii) one wash in PBS for 1 min; and (xiii) five washes in water for 1 min each. The grids were then stained with 2% uranyl acetate and lead citrate. Samples were observed and photographed by using an FEI Tecnai 12 100 kV transmission electron microscope.
Tumor studies. Tumor assays on Kalachoe diagremontiana were performed as previously described (53). Briefly, 1-cm wound sites were created by carefully scratching the surface of a leaf with a razor blade. Wounds were immediately inoculated with 109 CFU (resuspended in 10 µl of double-distilled H2O) of the strains described. We determined that this concentration of bacteria is optimal for reproducibly inciting tumors by testing different dilutions of bacteria. Each strain was inoculated multiple times on several independent plants. Inoculated plants were incubated at 19°C for 48 h and then transferred to room temperature. The fraction of inoculations that incited the tumors was calculated. The sizes of the tumors relative to that of tumors incited by A348 (wild type) were estimated by visual inspection on a scale of 0 to 4. The mean relative tumor size incited by each strain was then calculated. Relative tumorigenesis was quantified as follows: (tumor frequency x tumor size x 0.25) x 100; a strain that produces tumors of maximal size in every inoculation would then have a relative index of 100%.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Expression of VirB1 and derivatives.
Figure 1A shows the amino acid sequence of wild-type nopaline VirB1 (NP_536285.1) (34). Residues flanking the SPI cleavage site, the glutamic acid presumed to be required for catalytic activity of the N terminus (54), and residues flanking the C-terminal processing site for VirB1* (3) are highlighted. Figure 1B diagrams the domains of wild-type VirB1 and six deletion or mutant derivatives used in our complementation studies. VirB1-LT comprises the N terminus, including the SPI domain. VirB1* includes the N-terminal SPI domain and the C-terminal VirB1* (10 wild-type amino acids adjacent to each of these domains are also encoded for sequence context at the two cleavage sites). VirB1SP contains all of VirB1 except the N-terminal 35 aa. VirB1(AA/PR) is full-length VirB1 in which the alanine-alanine VirB1* processing site is changed to proline-arginine. VirB1(E/Q) is full-length VirB1 in which the glutamic acid (aa 60) required in the active site is changed to glutamine. VirB1(Q/S) is full-length VirB1 in which a glutamine (aa 182) near the VirB1* processing site is changed to serine. This change makes the processing site region identical to that found in octopine VirB1 (see Fig. 6, below).
|
SP, employ coding sequences identical to those used in our previous study (53). Two constructs, VirB1-LT and VirB1*, differ slightly due to different cloning strategies. Three of the complementing constructs are new, VirB1(AA/PR), VirB1(E/Q), and VirB1(Q/S). To assess expression and C-terminal processing of VirB1 and VirB1 mutants, whole cells from the wild-type strain C58, the virB1 deletion strain CB1001 (62), and CB1001 complemented with the constructs described in Fig. 1B were subjected to Western blot analysis (Fig. 2A). To assess secreted forms of VirB1, cell surface proteins were released from the cell surface by vortexing and then precipitated and subjected to Western blot analysis (Fig. 2B).
|
SP was observed only as a single protein (Fig. 2A, lane 7), corresponding to the lower protein band seen in lanes 2 and 4 (and lanes 8 and 9, see below). VirB1-LT was not detected (Fig. 2A, lane 5) because our anti-VirB1 antibody does not recognize the N terminus. However, an increase in tumor formation by a virB1 deletion strain containing VirB1-LT suggests that this deletion product is expressed and functional (see Table 2) (53). The VirB1* construct also yielded two novel proteins that are approximately the correct size for the full-length VirB1* protein that would be produced before and after cleavage of the SPI domain (Fig. 2A, lane 6). The mutations with amino acid changes at the VirB1* processing site and the catalytic site in the lytic transglycosylase domain produce abundant amounts of the full-length proteins, again as doublets (Fig. 2A, lanes 8 and 9).
|
Unexpectedly, three C-terminal products, slightly larger than bona fide VirB1*, were produced by the strain that expressed VirB1(AA/PR), mutated at the VirB1* processing site (Fig. 2A, lane 8). The same proteolytic mechanism that produces wild-type VirB1* may also cleave at secondary sites that are not utilized when the wild-type sequence is available. Alternatively, substitution of PR for AA may render the mutant protein susceptible to another periplasmic protease(s) that cleaves at a different site(s). Mutation of glutamic acid to glutamine in the active site of the lytic transglycosylase domain did not affect processing of VirB1* from VirB1 (Fig. 2A, lane 9).
We next assayed for VirB1* secretion to the exterior of the bacterial cells versus that present in whole cells described above. In assays for secreted VirB1* and isolated T pili (below), control Western blots were probed with anti-VirB8 antiserum to determine whether these fractions contained whole cells. VirB8 was not detected even after prolonged exposure of films, suggesting that these fractions were not contaminated with whole cells (data not shown).
Slightly more VirB1* is secreted by CB1001 expressing wild-type VirB1 in trans compared to C58 (Fig. 2B, lanes 2 and 4) (likely due to enhanced expression mentioned above), although the amounts secreted are approximately proportional to the amounts in whole-cell pellets. CB1001 expressing VirB1* also efficiently secreted this protein (Fig. 2B, lane 6). Since CB1001 expressing VirB1SP did not produce VirB1* (Fig. 2A, lane 7), no VirB1* was secreted (Fig. 2B, lane 7). Of the three products derived from the C terminus of VirB1(AA/PR) in CB1001 expressing VirB1(AA/PR), the two lower-molecular-weight forms are secreted to a higher level than the largest-molecular-weight form (Fig. 2B, lane 8), reflecting their higher production in whole cells. CB1001 expressing VirB1(E/Q) secreted VirB1* (Fig. 2B, lane 9). Thus, the lytic transglycosylase function of the N terminus of VirB1 is not required for secretion of VirB1*.
As in our previous study (53), deletion of virB1 reduced tumorigenesis by 90 to 99% and expression of full-length VirB1 restored wild-type virulence (Table 2). The two strains that express VirB1 with an intact lytic transglycosylase domain but that carry a deletion or mutation of the VirB1* domain [VirB1-LT and VirB1(AA/PR)] were restored to 28% of wild-type virulence. Similarly, the two strains that express VirB1 with a large deletion or mutation in the lytic transglycosylase domain [VirB1* and VirB1(E/Q)] were restored to 28 and 22% of wild-type virulence, respectively. Expression of VirB1SP in the virB1 deletion restored virulence to ca. 6% of the wild type. This latter result is an increase in virulence over the previous study (53) and may be related to the enhanced expression from the plasmid used for complementation experiments in the present study (see also the discussion of T-pilus formation in this strain below).
Analyses of VirB2 and VirB5 in isolated T pili. We next used established procedures for purification of T pili (2, 38, 41, 49, 50, 63) whereby T pili are first sheared from the surface of vir-induced cells and then concentrated by high-speed centrifugation. Although VirB2 is abundant in T-pilus preparations from the wild-type strain, VirB2 is not detected in the same fraction from the virB1 deletion strain CB1001 (Fig. 3A, lanes 2 and 3). trans complementation of CB1001 with a plasmid that expresses full-length VirB1 restores VirB2 to the high-speed pellet but not to wild-type levels (Fig. 3A, lanes 2 and 4). VirB2 is also recovered in the T-pilus preparation specifically from CB1001 expressing VirB1* or VirB1(E/Q), the lytic transglycosylase active-site mutation (Fig. 3A, lanes 6 and 9). CB1001 complemented with VirB1* or full-length VirB1 produce comparable amounts of extracellular VirB2 (Fig. 3A, lanes 4 and 6).
|
SP resulted in the accumulation of a small but detectable and reproducible amount of VirB2 in the T-pilus preparations (Fig. 3A, lane 7). Since this VirB1 variant more efficiently binds the mature form of VirB2 than VirB1* (shown below in Table 3), we suggest that its binding to VirB2 mediates a small amount of VirB1SP transport to the periplasm. Such periplasmic VirB1SP then may be processed to produce a minute amount of VirB1* (not detected by Western blotting, Fig. 2A, lane 7, and Fig. 2B, lane 7) to facilitate VirB2 transit to the cell exterior for T-pilus generation. The partial function of VirB1SP is supported by tumor assays (Table 2).
|
VirB5 accumulation in T-pilus preparations mirrors that found for VirB2. Complementation of CB1001 with wild-type VirB1 resulted in the reappearance of VirB5 in T-pilus preparations (Fig. 3B, lanes 3 and 4). Expression of VirB1* was as effective as full-length VirB1 in restoring accumulation of VirB5 in isolated T pili (Fig. 3B, lanes 4 and 6). As for VirB2, accumulation of VirB5 was slightly higher in T pili from CB1001 expressing VirB1(E/Q) compared to full-length VirB1 (Fig. 3B, lanes 4 and 9).
These data highlight that the stoichiometry between VirB2 and VirB5 is consistent in all T-pilus preparations (Fig. 3A and B, lanes 2, 4, 6, 7, and 9). As shown above with VirB2, accumulation of VirB5 was partially supported by trans complementation with VirB1SP (Fig. 3B, lane 7). trans complementation with VirB1-LT or VirB1(AA/PR) (Fig. 3B, lanes 5 and 8), however, did not increase the accumulation of VirB5 in T pili relative to CB1001. Therefore, both the major and minor T-pilus components, VirB2 and VirB5, are significantly reduced in the high-speed pellet from a virB1 deletion strain, and their presence is restored by trans complementing plasmids that express and secrete VirB1*. In contrast, the lytic transglycosylase domain of VirB1 does not contribute to the accumulation of VirB2 and VirB5 in isolated T pili. Reduced levels of VirB2 and VirB5 in CB1001 extracts are not due to decreased transcription, since both proteins are found in cell pellets in amounts comparable to wild type (data not shown). Potentially, translational coupling plays a role in VirB1 mediated processing of VirB2 for pilus biogenesis.
Notably, VirB1* copurifies with T pili in wild-type cells and not in CB1001 (Fig. 3C, lanes 2 and 3). This association is found in all strains that produce VirB1* (Fig. 3C, lanes 2, 4, and 6) but not in strains only producing the N terminus of VirB1 (Fig. 3C, lane 5). We do not intend to imply that VirB1* is a structural component of T pili; indeed, VirB1* can be removed from vir-induced T-pilus-containing cells by vortexing. Nevertheless, that VirB1* is found in isolated T pili provides supportive evidence for a role for VirB1* in T-pilus formation.
Visual detection of T pili. In C58 grown under virulence gene-inducing conditions, the T pilus was identified as a filamentous structure that is shorter than a flagellum with a much smaller diameter (Fig. 4A) (30, 49). This structure was present in ca. 50% of the cells observed (n > 100). This number is lower than 100% since T pili are notoriously sensitive to loss during sample preparation for transmission electron microscopy. T pili were often inserted close to the cell pole. T pili were never observed in uninduced cells (data not shown). Vegetative flagella were less frequent under virulence gene-inducing conditions as previously reported (49). T pili were never observed on the virB1 deletion strain CB1001 (data not shown). When CB1001 was complemented with virB1, T-pilus biogenesis was restored (Fig. 4B). In some cases, the tip of the T pilus contained a knob-like structure (insets in Fig. 4A and 4B). Similar structures were reported in isolated T pili (2), although whether these structures were at the base or distal tip of the isolated T pili could not be determined. The present data suggest the knob occurs at the distal tip of the T pili.
|
|
B1). VirB1* was processed (Fig. 6B) and secreted (Fig. 6C) from VirB1 (Q/S) in both the nopaline (lane 3) and the octopine (lane 4) strains, These data suggest that both strains are able to process and secrete VirB1* from a VirB1 variant whose amino acid sequence around the processing site is identical to octopine VirB1 (pTiA6).
Although A348 appears to have the capacity to process and secrete a nopaline/octopine hybrid VirB1*, octopine VirB1* has not been reported. Figure 7B shows that a protein of approximately the expected 13 kDa was detected with an anti-hemagglutinin antibody in vir-induced cell lysates of A348B1 that expressed wild-type octopine VirB1 with hemagglutinin epitope tags (Fig. 7B, lanes 2 and 4). Epitope tags (Fig. 7A) were placed at the C terminus of octopine VirB1 (pJZ079) or between aa 188 and 189 (pJZ093), 12 aa downstream from the conserved region close to the processing site, see below). Octopine VirB1* with the internal hemagglutinin tag was efficiently processed from octopine VirB1 (Fig. 7, lane 4). VirB1* with the C-terminal hemagglutinin tag was also processed from VirB1 but at much lower efficiency (Fig. 7, lane 2). Thus, octopine VirB1 is also processed to release octopine VirB1*.
|
|
SP and virB1* was 2 orders of magnitude less efficient than transformations with full-length virB1 or virB1*+44 (Fig. 9A). Our interpretation is that VirB1SP and VirB1* adopt conformations that expose a domain in the C terminus that is toxic to yeast. The conformations of full-length VirB1 and VirB1*+44 presumably mask the toxic domain. Interestingly, VirB1* toxicity is also alleviated by coexpression of the mature but uncyclized form of VirB2 (Fig. 9B); these data suggest that protein-protein interaction may mask the toxic domain.
|
SP or VirB1*+44 and VirB2m, which corresponds to the mature T-pilus subunit prior to cyclization. A moderately strong interaction was obtained with VirB1*+44 and full-lengthVirB2. VirB1* interacted moderately with VirB2m. Strong interactions were also detected between VirB1*+44 or VirB1* and VirB5SP. In summary, interaction between VirB1* and VirB2 or VirB5 is consistently detected. The VirB2 interaction domain appears to reside in the N terminus of the mature T-pilin subunit (aa 48 to 94), since the C-terminal VirB2 (aa 95 to 121) interacts extremely poorly with VirB1 derivatives. The VirB5 interaction domain was not more narrowly defined. These data provide independent data to suggest VirB1* interacts with VirB2 and VirB5. Protein-protein interactions are consistent with a role of VirB1* in facilitating T-pilus formation. However, given the toxicity of VirB1* to yeast, we do not intend that the data in Table 3 represent an exact quantitative assessment of these interactions. Nevertheless, the data are qualitatively sound and highly reproducible. Interestingly, we have shown previously that reduced expression of toxic proteins leads to higher levels of yeast two-hybrid interactions (as assessed by measuring the ß-galactosidase activity) (25). Thus, protein levels per se cannot be used to monitor the strength of different protein-protein interactions.
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
No role for the C terminus of VirB1, VirB1*, has been proposed previously. A database search with the amino acid sequence of VirB1* did not identify any related proteins other than VirB1 homologs primarily from strains of A. tumefaciens or closely related species (nine from A. tumefaciens, one from A. rhizogenes, one from Rhizobium etli, and one from Mesorhizobium loti; J. Zupan, O. Draper, R. Middleton, and P. Zambryski, data not shown). Therefore, no role for VirB1* can be inferred from proteins that function in systems or processes other than T4SS.
Evidence for processing a C-terminal product from VirB1 homologs is emerging. The machinery to process and secrete VirB1* from VirB1 (pTiC58, nopaline) is conserved in an octopine strain (A348) of A. tumefaciens (53). In addition, we now detect a protein of approximately 13 kDa with an anti-hemagglutinin antibody in cell lysates of vir-induced A348B1 that expressed octopine VirB1 with hemagglutinin tags. VirB1* processing, or stability, in the octopine system was affected by the position of the epitope tag. Octopine VirB1* with an internal hemagglutinin tag was produced at an efficiency comparable to the nopaline system. In contrast, the hemagglutinin tag at the carboxy terminus of VirB1* severely reduced the amount of VirB1* produced. Both nopaline and octopine strains process and secrete a nopaline VirB1 variant [VirB(Q/S)], where the VirB1* processing site was altered to conform to the amino acid sequence of octopine VirB1 in this region. Recently, a VirB1*candidate has been identified in the cell lysate of Brucella abortus (20). Finally, homology searches identified a set of VirB1 homologs that exhibit sequence similarity in their C termini in addition to the region of lytic transglycosylase similarity. The carboxy-terminal homology among these latter proteins is confined to a region of about 10 aa at the VirB1* processing site; sequences distal to this domain diverge (see Fig. S1 in the supplemental material). Together, these data suggest that C-terminal processing of VirB1-like proteins may not be restricted to A. tumefaciens.
In the present study, the abundance of VirB2 and VirB5 in isolated T pili was positively correlated with the presence of VirB1* but not the lytic transglycosylase N-terminal domain of VirB1. VirB1* also is found in isolated T pili. In addition, VirB1* bound significantly to mature T-pilus subunits VirB2 (aa 48 to 121) and VirB5SP in the yeast two-hybrid assay. Thus, genetic, cell fractionation, microscopy data from A. tumefaciens and physical interaction data from the yeast two-hybrid system together suggest that VirB1* facilitates T-pilus biogenesis by binding to VirB2 as well as to VirB5.
The interaction between VirB1* and the structural T-pilus components, VirB2 and VirB5, may contribute to the regulation of T-pilus biogenesis. Assembly of a polymeric, extracellular structure such as the T-pilus likely requires mechanisms that ensure temporal and spatial control of subunit interaction. Presumably, T-pilus assembly is at least partly driven by the interaction between regions on the surface of the VirB2 and VirB5 proteins. VirB1* may bind VirB2 and VirB5 monomers at the inner membrane or in the periplasm, thereby masking VirB2-VirB2 or VirB2-VirB5 interaction domains until they are utilized at the site of T-pilus assembly. Since significant amounts of VirB1* are found on the cell exterior and associated with the T pilus, some VirB1* may remain associated with T-pilus subunits during their transit across the outer membrane.
Many type 2, type 3, and type 4 secretion systems mediate the biogenesis of cell surface-associated filamentous or fibrous structures that mediate adhesion to the host cell surface (67). In each case, there is a requirement for the spatiotemporal regulation of assembly of these structures (68). In some cases, regulation of assembly is mediated by periplasmic chaperones; the best known are members of the PapD-like superfamily (39). PapD-like proteins not only bind P-pilus subunits in the periplasm of pathogenic E. coli but also facilitate proper folding after the subunit emerges from the GSP into the periplasm in an unfolded state. Although VirB1* is proposed to bind T-pilus subunits and promote their accumulation in T pili, it is not known whether VirB1* performs all of the functions provided by PapD-like chaperones. Additional chaperone-like characteristics of VirB1* are its low molecular weight and acidic pI, which are common characteristics of type III secretion system chaperones (57). Alternatively, VirB1* binding may mediate some other step in T-pilus biogenesis, such as recruitment of VirB2 and VirB5 to the core transporter via interactions between VirB1 and VirB8, VirB9, or VirB10 (72).
We propose that VirB1 is a bifunctional protein for the biogenesis of the vir-T4SS (Fig. 10). The N-terminal lytic transglycosylase domain would provide localized modification of the peptidoglycan to create pores large enough for the assembly of the vir-T4SS trans-envelope core. The C-terminal VirB1* would regulate T-pilus assembly by binding T-pilus subunits, possibly to prevent their premature association. VirB1* on the cell exterior and in the T-pilus fraction may be cotransported with T-pilus subunits or may reflect residual binding of VirB1* to VirB2 or VirB5 after T-pilus assembly. Future studies will address the impact of virB1 deletion on the assembly of the T4SS core, as well as the identity of the VirB1* protease. These studies likely have broad significance since many T4SS, as well as type III secretion systems, have VirB1 homologs (46), and their assembly requires both penetration of the peptidoglycan layer and mobilization of structural components to the exterior of the cell.
|
|
ACKNOWLEDGMENTS |
|---|
B1. We thank Kent McDonald, director of the University of California, Berkeley, Electron Microscope Laboratory, for assistance with transmission electron microscopy. We thank Rebecca Middleton for help with sequence analysis. C.A.H. was supported by an NSF predoctoral fellowship. J.A. was supported by a UC Berkeley Chancellors Fellowship. This research was supported by a grant from the NSF (MCB-0343566) awarded to P.Z.
|
FOOTNOTES |
|---|
Published ahead of print on 13 July 2007.
Supplemental material for this article may be found at http://jb.asm.org/.
Present address: Department of Biology, West Valley Community College, 14000 Fruitvale Ave., Saratoga, CA 95070.
Present address: Program in Molecular Medicine, University of Massachusetts Medical School, 373 Plantation St., Worcester, MA 01605.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
