A Putative Transmembrane Leucine Zipper of Agrobacterium VirB10 Is Essential for T-Pilus Biogenesis but Not Type IV Secretion

Abstract

The Agrobacterium tumefaciens VirB/VirD4 type IV secretion system is composed of a translocation channel and an extracellular T pilus. Bitopic VirB10, the VirB7 lipoprotein, and VirB9 interact to form a cell envelope-spanning structural scaffold termed the “core complex” that is required for the assembly of both structures. The related pKM101-encoded core complex is composed of 14 copies each of these VirB homologs, and the transmembrane (TM) α helices of VirB10-like TraF form a 55-Å-diameter ring at the inner membrane. Here, we report that the VirB10 TM helix possesses two types of putative dimerization motifs, a GxxxA (GA₄) motif and two leucine (Leu1, Leu2) zippers. Mutations in the Leu1 motif disrupted T-pilus biogenesis, but these or other mutations in the GA₄ or Leu2 motif did not abolish substrate transfer. Replacement of the VirB10 TM domain with a nondimerizing poly-Leu/Ala TM domain sequence also blocked pilus production but not substrate transfer or formation of immunoprecipitable complexes with the core subunits VirB7 and VirB9 and the substrate receptor VirD4. The VirB10 TM helix formed weak homodimers in Escherichia coli, as determined with the TOXCAT assay, whereas replacement of the VirB10 TM helix with the strongly dimerizing TM helix from glycophorin A blocked T-pilus biogenesis in A. tumefaciens. Our findings support a model in which VirB10's TM helix contributes to the assembly or activity of the translocation channel as a weakly self-interacting membrane anchor but establishes a heteromeric TM-TM helix interaction via its Leu1 motif that is critical for T-pilus biogenesis.

INTRODUCTION

The type IV secretion systems (T4SSs) of Gram-negative bacteria deliver DNA or protein substrates to other bacterial or eukaryotic cell targets (1). In the pKM101 conjugation system, an early step in the T4SS assembly pathway involves the formation of a stabilizing machine subassembly consisting of 14 copies each of the TraN, TraO, and TraF subunits. The structure of this complex, termed the “core complex,” was solved by cryoelectron microscopy (CryoEM), and a portion of the complex was also solved by X-ray crystallography (2–4). The entire core complex is a large 1.5-MDa barrel that spans the entire Gram-negative cell envelope. This complex is postulated to serve as a structural scaffold for the assembly of the translocation channel, as well as an extracellular filament termed the conjugative or T pilus (5, 6). Homologs of TraN, TraO, and TraF are phylogenetically widely distributed, and thus, the pKM101 core complex is considered a structural archetype for equivalent machine subassemblies of Gram-negative bacterial T4SSs (2, 7).

A particularly intriguing feature of the core complex is that TraF spans the entire Gram-negative cell envelope (3), a topology that is unique among bacterial membrane proteins characterized to date. Our laboratory is characterizing the VirB/VirD4 T4SS of Agrobacterium tumefaciens, and on the basis of the pKM101 structural data and our structural modeling and biochemical findings, we have proposed that the TraF homolog VirB10 similarly adopts an envelope-spanning architecture such that (i) residues 1 to 30 and 31 to 50, respectively, make up the cytoplasmic and inner membrane (IM) transmembrane (TM) domains, (ii) residues 61 to 114 and 115 to 377, respectively, make up the proline-rich and β-barrel domains that span the periplasm, and (iii) residues 285 to 335 make up an unusual helix-loop-helix domain termed the antenna projection (AP) that in the assembled core complex forms an outer membrane (OM) pore through which substrates pass and the T pilus extends (3, 7).

In the pKM101 core complex, the N-proximal TM domain helices of the 14 monomers of TraF form a 55-Å-diameter ring that is envisioned to encircle the IM translocase (2, 5, 6). The architecture of the IM translocase is unknown, but in the A. tumefaciens VirB/VirD4 system, several VirB subunits, including the VirB11 ATPase, polytopic VirB6, and bitopic VirB8, are likely constituents, as deduced from results of formaldehyde cross-linking studies demonstrating close associations of DNA substrates with these subunits during translocation (8). The N-terminal region of VirB10 might form contacts with one or more of these translocase subunits to promote channel assembly. Additionally, there is evidence that VirB10-like subunits interact with VirD4-like substrate receptors (9–11). The VirD4-like proteins belong to a family of FtsK/SpoIIIE-like ATPases termed type IV coupling proteins (T4CPs) that are widely distributed among T4SSs (11, 12). The T4CP nomenclature arose from findings that VirD4-like subunits serve to “couple” early conjugative DNA processing reactions with transfer through the T4SS channel (13–15). The interaction of VirD4 with VirB10 thus might form the basis of this “coupling” activity in the A. tumefaciens VirB/VirD4 T4SS.

Besides VirB10's involvement in channel assembly through N-proximal contacts with translocase subunits and the VirD4 receptor, VirB10 undergoes a conformational transition that is required for DNA substrate passage through the channel. This structural change occurs in response to ATP utilization by the IM ATPases VirD4 and VirB11 (16). Recently, we showed that docking of the DNA substrate with VirD4 and its transfer to VirB11 are also required for the VirB10 structural transition (17). Moreover, a VirB10 mutation (G272R) located near the AP domain appears to “lock” VirB10 in an ATP-insensitive, activated conformation that results in substrate leakage to the cell surface (18). Together, these findings support a model in which VirB10 senses ligand binding and ATP utilization by the VirD4 and VirB11 ATPases and transduces these intracellular signals via a conformational change to gate the distal portion of the translocation channel, allowing substrate passage across the OM (17).

In view of VirB10's overall transenvelope architecture, we predicted that the N-proximal TM domain is directly involved in the ligand/energy-coupled translocation reaction through homo- or heterospecific contacts. Intriguingly, however, insertions of Ala-Cys dipeptides (designated i2 insertions) into the TM domain did not disrupt substrate transfer or VirB10 complex formation with VirD4 and the VirB9 and VirB7 core components but instead abolished or diminished the production of T pili. These findings pointed to the importance of the VirB10 TM domain for the biogenesis of the T pilus but not the translocation channel (19). For the present study, it is also important to note that VirD4-like T4CPs are required for the assembly of translocation channels but not T pili (20, 21).

On further examination, we discovered that the VirB10 TM domain has two types of potential dimerization motifs. The first, a GxxxA (GA₄) motif, closely resembles the strongly dimerizing GG₄ motif of glycophorin A (GpA) and other membrane proteins (22). The second is made up of two Leu(Ile) zippers that in a heptad wheel align on opposite faces of the α helix. These motifs resemble Leu zippers shown to mediate dimer formation among soluble proteins and eukaryotic and viral membrane proteins (23). To our knowledge, Leu zippers have not been reported to mediate TM-TM helix dimerization among bacterial membrane proteins, with the exception that an artificial Leu zipper functionally substituted for a TM domain of the A. tumefaciens VirA sensor kinase (24). The i2 mutations that conferred the transfer-positive, pilus-negative (Tra⁺, Pil⁻) phenotype are located in these dimerization motifs and might exert their effects through disruption of the helix register (19).

Here, we further defined the contribution of the VirB10 TM helix to VirB/VirD4 machine assembly through directed mutagenesis of the putative dimerization motifs and TM domain swaps. We report that one of the Leu zipper motifs is essential for T-pilus biogenesis, but neither this nor another dimerization motif is critical for VirB10 interactions with other channel components or for substrate transfer. In fact, the entire TM domain can be swapped with a nondimerizing TM domain sequence without loss of channel activity. We discuss our findings in the context of VirB10's overall domain architecture, ligand- and ATP energy-coupling activity, and contributions to the biogenesis of the transfer channel and T pilus.

MATERIALS AND METHODS

Bacterial strains and growth/induction conditions.

The A. tumefaciens and Escherichia coli strains used in this study are listed in Table 1. Conditions for growth of A. tumefaciens cells and induction of vir genes with 100 μm acetosyringone were previously described (25). Plasmids were maintained in A. tumefaciens or E. coli by the addition of appropriate antibiotics to the medium at the following concentrations (in μg/ml): carbenicillin, 100; tetracycline, 5; kanamycin, 100; gentamicin, 100; spectinomycin, 500 (19).

Table 1.

Bacterial strains and plasmids

Strain or plasmida	Relevant characteristic(s)	Source
E. coli strains
DH5α	ϕ80dlacZΔM15 Δ(lacZYA-argF)U169 recA1 endA1 hsdR17 (rK− mK+) phoA supE44 λ− thi-1 gyrA96 relA1	Gibco-BRL/Invitrogen
CJ236(pCJ105)	Camr dut ung thi relA	26
S17-1	Carries tra genes from broad-host-range plasmid RP4 in the chromosome	34, 37
MM39	malE	22
A. tumefaciens strains
A348	A. tumefaciens containing octopine-type Ti plasmid pTiA6NC	69
A348Spcr	Spcr derivative of octopine strain A348	34
PC1010	A348 pTiA6NC with a nonpolar ΔvirB10 mutation	30
Plasmids
pBSIISK+NdeI	Crbr; cloning vector containing NdeI restriction site at the translational start site of lacZ	30
pSW172	Tetr; broad-host-range IncP plasmid	27
pXZ151	Kanr; derivative of pSW172 encoding a Kanr cassette replacing Tetr	28
pML122ΔKm	Genr; mobilizable IncQ derivative (RSF1010)	36
pKVD10	Crbr; pBSIISK+NdeI coding for native VirB10	19
pIG024	Crbr; pKVD10 coding for VirB10 G37I	This study
pIG025	Crbr; pKVD10 coding for VirB10 A41I	This study
pIG026-33 seriesb	Crbr; pKVD10 coding for VirB10 variants with single and triple Ala substitutions of Leu1 and Leu2 motif residues	This study
pIG034	Crbr; pKVD10 coding for VirB10-FtsNTM	This study
pIG023	Crbr; pKVD10 coding for VirB10-FtsNCyto/TM	This study
pIG037-41 seriesb	Crbr; pKVD10 coding for VirB10 variants with pLA TM domain swaps	This study
pIG042	Crbr; pKVD10 coding for VirB10 W48A	This study
pccKan	Crbr, Kanr; TOXCAT cloning vector	22
pMal-c2	Crbr; pBR322 producing MBP lacking its signal sequence	New England Biolabs
pMal-p2	Crbr; pBR322 producing MBP with a signal sequence	New England Biolabs
pccGpA	Crbr; pccKan with the GpA TM domain	22
pccGpA.G83I	Crbr; pccKan with the GpA(G83I) TM domain	22
pIG070-82 seriesb	Crbr; pccKan with native and mutant VirB10 TM domain sequences	This study
pLS23	Spcr, IncP vector coding for FLAG-VirE2	70
pKVD117	Crbr; pKVD10 coding for VirB10 ΔN30	This study
pKVD136	Crbr; pKVD10 coding for VirB10 ΔN46ss	19
pIG5-50	Crbr; pKVD10 coding for VirB10 derivatives with Ala-Cys insertions immediately following the residue number indicated	19
pIG013	Crbr; pKVD10 coding for VirB10 S61C	19
pIG014	Crbr; pKVD10 coding for VirB10 N69C	19

^aColE1 plasmids of interest were ligated to broad-host-range IncP plasmid pSW172 or pXZ151 for introduction into A. tumefaciens.

^bNames of plasmids for each VirB10 mutation are listed in Table 2.

Construction of virB10 insertions, substitutions, and TM domain swaps.

Codon insertions and substitutions in virB10 were generated by oligonucleotide-directed mutagenesis (26). Plasmid pKVD10, containing P_lac-virB10, served as the template for codon substitutions generated with oligonucleotides containing 12- to 15-bp sequence complementarity to regions both upstream and downstream of the mutation. Mutations were verified by DNA sequence analysis. ColE1 plasmids expressing the mutant alleles were ligated to broad-host-range plasmid pSW172 (27) or pXZ151 (28), and the resulting cointegrated plasmids were introduced into A. tumefaciens cells by electroporation (29).

virB10 alleles encoding VirB10 derivatives with TM domain swaps were created by PCR amplification of the 5′ end of virB10 together with the respective TM domain sequence change. PCRs utilized pKVD10 (P_lac-virB10) as the template and primers listed in Table 2. PCR products incorporated NdeI (CAT ATG) and SphI (GCA TGC) sites at the 5′ and 3′ ends, respectively. The digested PCR fragments were introduced into pIG50, which contains a phenotypically silent SphI (GCA TGC) site at codon 50 of virB10 (19). Resulting plasmids encoded full-length VirB10 with the variant TM domain sequences.

Table 2.

Oligonucleotides used for construction of VirB10 mutations

Substitution or protein (plasmid namea)	Oligonucleotideb
TM domain substitutions
G37I (pIG024)	5′-CGCGAGAACGACAATTCCGACGATAAGCTT1CTGAGACCC-3′
A41I (pIG025)	5′-TAGCCATATG2AGGCTTAACGATAAAATGAGAACGACACC-3′
L33A (pIG026)	5′-ACCTCCGACGATCGCTTTCTGGGATCC3CGAAAGACGCCG-3′
V35A (pIG027)	5′-GAACGACACCTCCGGCGATAAGCTT1CTGAGACCC-3′
L40A (pIG028)	5′-CTAGCCATATG2AGGCTTAACGATAACGCGGCAACGACACC-3′
L42A (pIG029)	5′-CCAAATGAGGCTTAAGCTAGC4CGCGAGAACGACACC-3′
I47A (pIG030)	5′-CACCTTCTTTTGGCGGCCGC5CTAGCCAAGCGAGGCTTAAC-3′
L49A (pIG031)	5′-CACCTTCTTTTGGCGGCCGC5CTGCCCAAATGAGGCT-3′
L33A, L40A, I47A (pIG032)	5′-CCCACCTAGCCAAGCGAGGCTTAACGATAACGCGGCAACGACACCTCCGACGATCGCTTTCTGGGATCC3CGAAAGACG-3′
V35A, L42A, L49A (pIG033)	5′-GACGCCCACCTGCCCAAATGAGGCTTAAGCTAGC4CGCGAGAACGACACCTCCGGCGATCAATTTCTG-3′
Fusion proteins for TOXCAT
FP-TM33	5′-GGGTCTCAGAAATCTAGA6TTGATCGTCGGAGGT-3′
RP-TM50	5′-CTTTTGACGCCCGGATCC3CTAGCCAAATGAG-3′
FP-TM33-(L33A)	5′-GGGTCTCAGAAATCTAGA6GCGATCGTCGGAGGT-3′
FP-TM33-(V35A)	5′-GGGTCTCAGAAATCTAGA6TTGATCGCCGGAGGT-3′
RP-TM50-(I47A)	5′-CTTTTGACGCCCGGATCC3CTAGCCAAGCGAG-3′
RP-TM50-(L49A)	5′-CTTTTGACGCCCGGATCC3CTGCCCAAATGAG-3′
FP-TM33-(35AC)	5′-CGTCTTTCGGGGTCTCAGAAATCTAGA6TTGATCGTC-3′
TM domain swaps
B10-FP	5′-CATATG2AATAACGATAGTCAGCAAGCGGC-3′
FtsN (35-50) TM domain swap (pIG034)	5′-GCATGC7AATGAAGTACAGACCACCGATAAAGGTCACAAGAACGGCGGCAGCGACGATAAGCTTCTGAGACCCCGAAAG-3′
FtsN (1-50) TM domain swap (pIG035)	5′-CACCTTCTTTTGACGCCCGCATGCAATGAAGTATGCACCACCGATAAAGGTCACAAGAAC-3′
pLA (33-50, W48) (pIG037)	5′-GCATGC7CGCCAGCCACAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGTTTCTGAGACCCCGAAAGACGCCG-3′
pLA (33-50) (pIG038)	5′-GCATGC7CGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAATTTCTGAGACCCCGAAAGACGCCG-3′
pLA (27-50) (pIG039)	5′-GCATGC7CGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCCAGCGCAAGACGCCGGCGATGTTTGTCGGAGACCAGGGATCC-3′
W48A (pIG042)	5′-CTTCTTTTGTCGAC8CACCTAGCGCAATGAGGCTTAAC-3′
pLA (33-50, W48, VL/SL) (pIG040)	5′-CTTCTTTTGACGCCCGCATGC7CGCCAGCCACAGCGCCAGTAACGACGCCAGGAGAACCGCCAGCGCCAGCGCCAATTTCTGAGACCCCGAAAGACGCCG-3′
pLA (33-50, W48, L40) (pIG041)	5′-CTTCTTTTGACGCCCGCATGC7CGCCAGCCACAGCGCCAGCGCCAGCGCCAGCAGCAGCGCCAGCGCCAGCGCCAATTTCTGAGACCCCGAAAGACGCCG-3′
GpA (33-50) (pIG043)	5′-CTTCTTTTGACGCCCGCATGC7ACCTAGCCAAATGAGGGTCCCAATAACGCCAGCCATGACTCCGAAAATAATCAATTTCTGAGACCCCGAAG-3′

^aPlasmids expressing the ToxR-TM-MBP fusion proteins: pIG070 (native VirB10 TM domain), pIG071 (35AC), pIG72 (40AC), pIG73 (45AC), pIG074 (G37I), pIG075 (A41I), pIG076 (L33A L40 I47A), pIG077 (V35A L42A L49A), pIG078 (L33A), pIG079 (L40A), pIG080 (pLA W48), pIG081 (pLA VL/SL), and pIG082 (pLA L40).

^bBases in bold are silent restriction sites identified by superscript numbers as follows: 1, HindIII; 2, NdeI; 3, BamHI; 4, NheI; 5, NotI; 6, XbaI; 7, SphI; 8, SalI.

Protein analysis by Western blotting.

vir-induced cells were harvested, cell numbers were normalized by adjustment to equivalent optical densities at 600 nm (OD₆₀₀s), and cells were resuspended in Laemmli buffer (50 mM Tris-HCl, 4% sodium dodecyl sulfate [SDS], 20% glycerol, 20 mM dithiothreitol, 1% β-mercaptoethanol, 0.1% bromophenol blue, pH 6.8) and boiled for 10 min. The boiled cell extracts were resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) or Tricine-SDS-PAGE as previously described (30). Separated proteins were then transferred to nitrocellulose membranes and detected immunologically with anti-VirB or anti-VirD4 antibodies. For TOXCAT assay, the steady-state levels of the ToxR-TM-maltose binding protein (MBP) fusion were assessed by immunostaining with an anti-MBP monoclonal antibody conjugated to horseradish peroxidase, followed by enhanced chemiluminescence assay (Thermo Scientific).

Cysteine cross-linking.

Oxidative cross-linking was carried out at 30°C for 5 min by adding copper phenanthroline (180 mM phenanthroline, 60 mM CuSO₄, and 50 mM NaH₂PO₄ in H₂O) to a final concentration of 300 mM to a cell suspension (OD₆₀₀ = 1.0). Reactions were terminated by the addition of N-ethylmaleimide (1 mg/ml final concentration) and EDTA (5 mM final concentration) (31, 32). Cross-linked products were detected by analysis of protein samples by SDS-PAGE under nonreducing conditions, followed by immunostaining of blots with anti-VirB10 antibodies.

Membrane solubilization, protein cross-linking, and immunoprecipitation.

A. tumefaciens cultures (500 ml) were induced for vir gene expression in AB minimal medium containing acetosyringone (ABIM) for 18 h at 22°C (33). Cells were harvested and lysed by French press treatment, and total membranes were recovered as previously described (33). Membranes were resuspended in 50 mM sodium phosphate buffer (pH 7.4) containing 20% sucrose, 0.1 mg/ml DNase I, 0.1 mg/ml RNase A, and EDTA-free protease inhibitor cocktail (Roche Molecular Biochemicals, Indianapolis, IN). Approximately 2 mg of total membrane in 200 μl of 10 mM sodium phosphate buffer (pH 7) was cross-linked with dithiobis(succinimidyl propionate) (DSP; 20-mg/ml [final concentration] stock, 0.5 μg/μl) for 30 min on ice, and the reaction was quenched with 0.5 M l-lysine at pH 8.5 (final concentration, 62.5 mM) for 15 min on ice. The cross-linked sample was washed once with 50 mM Tris-HCl, pH 7.4, and the total membranes were pelleted at 4°C by centrifugation for 1 h at 125,000 × g in a Beckman tabletop ultracentrifuge. The pellet was solubilized with 1.5% N-lauroyl sarcosine in 50 mM Tris-HCl (pH 7.4)–1 mM EDTA at 4°C for 4 h with gentle rocking. The solubilized material was centrifuged at 15,000 × g for 15 min, and the supernatant was used for immunoprecipitation.

The solubilized material was rocked gently at 4°C for 12 h with 10 mg protein A-Sepharose CL-4B beads (Amersham Biosciences; Piscataway, NJ) to remove nonspecific binding proteins. The samples were centrifuged for 5 min at 2,000 × g at 4°C, and the supernatant was divided into two aliquots for incubation with preimmune or immune serum at room temperature for 6 h. Samples were incubated with 10 mg of protein A-Sepharose CL-4B beads for 12 h at 4°C with slow rocking. The samples were centrifuged at 2,000 × g, washed three times with 50 mM Tris-HCl (pH 7.4)–0.15% N-lauroyl sarcosine–1 mM EDTA, and resuspended in 5× Laemmli buffer. The samples were boiled for 5 min and analyzed by SDS-PAGE and immunostaining.

T-pilus isolation.

T pili were isolated by mechanical shearing and ultracentrifugation as previously described (19, 34). Briefly, a 500-μl aliquot of vir-induced A. tumefaciens cells was spread on inducing plates containing 100 μM acetosyringone and then incubated for 3 to 5 days at 18°C. Cells were gently scraped from the inducing plates and collected in 1 ml KPO₄ buffer (50 mM, pH 5.5). The cell suspension was repeatedly passed through a 25-gauge needle, sheared cells were separated from T pili by centrifugation at 14,000 × g for 30 min at 4°C, and the supernatant was filtered through a cellulose acetate membrane (0.22-μm pore size; VWR sterile syringe filter). T pili were then pelleted by ultracentrifugation at 100,000 × g for 1 h at 4°C. The pellet was resuspended in Laemmli buffer and electrophoresed through a Tricine-SDS-polyacrylamide gel. The presence of pilin, monitored by Western transfer of proteins to nitrocellulose and immunostaining with anti-VirB2 antibodies, served as a diagnostic of T-pilus production (34, 35).

Detection of surface VirB2 pilin by colony immunoblotting.

A colony blot assay was also used to detect VirB2 pilin on the cell surface as an indicator of T-pilus production (19). One-milliliter cultures of vir-induced cells with equivalent OD₆₀₀s were pelleted by centrifugation, and cells were resuspended in 100 μl of liquid ABIM. This suspension was then spotted in 25-μl aliquots onto a vir induction plate and incubated at 18°C for 3 days. Cells were then transferred to a nitrocellulose membrane, and the resulting membrane was subjected to immunostaining with anti-VirB2 antibodies for the detection of surface-exposed pilin. Essentially, the same colony immunoblotting assay was used to detect the release of the VirE2 substrate to the cell surface through a defective VirB/VirD4 T4SS, as shown previously (18).

Virulence assays.

Translocation of DNA and protein substrates through the VirB/VirD4 T4SS was assessed with a plant tumor formation assay as previously described (30). Each leaf was coinoculated with wild-type (WT) strain A348 (positive control) and an avirulent ΔvirB10 mutant strain (negative control), and the virulence of a given strain was affirmed by the inoculation of at least three different leaves. Inoculated leaves were observed for tumor formation over a 4- to 6-week period. Tumor production was scored as follows: −, no tumors; +, small tumors arising 3 to 4 weeks after those incited by the WT A348 strain; ++, tumors that were reproducibly smaller and delayed in appearance by a few days to a week compared with those incited by A348; +++, WT tumors as judged by size, morphology, and time course of appearance.

Conjugation assays.

DNA transfer through the VirB/VirD4 system was also monitored by interbacterial conjugation (19, 36). The mobilizable IncQ plasmid pML122ΔKm (an RSF1010 derivative) was introduced into A. tumefaciens strains of interest by diparental mating with E. coil S17-1(pML122ΔKm) (34, 37). A. tumefaciens strains carrying pML122ΔKm then served as donors in mating experiments with a Spc^r A348 derivative (19, 36). Transfer frequencies were reported as the number of transconjugants per donor cell. For each strain, mating experiments were carried out in triplicate and results of a representative experiment with standard deviations are presented.

TOXCAT assay.

ToxR-TM-MBP expression plasmids were constructed as follows. E. coli malE mutant strain MM39 and vectors pccKan, pccGpA-WT, and pccGpA-G83I were generously supplied by D. Engelman (Yale University, New Haven, CT) (22). Plasmids containing the various TM domain sequences served as the templates for PCRs to amplify the TM domain sequence of interest with the addition of 5′- and 3′-flanking XbaI and BamHI sites, respectively. Primers used for amplification of TM domains are listed in Table 2. Primers FP-TM33 and RP-TM50 were used for the amplification of the native and most mutant TM domain sequences; additional primers listed were used for the amplification of TM domain sequences with mutations located near the 5′ or 3′ end of the TM domain sequence. PCR products and pccKan were similarly digested and ligated, generating the pIG071-to-pIG082 series of plasmids encoding ToxR-TM-MBP fusion proteins with variant TM domain sequences. TM domain sequences in the TOXCAT vector were confirmed by DNA sequencing.

The steady-state abundance of the ToxR-TM-MBP fusion proteins produced by MM39 strains was monitored by SDS-PAGE and immunostaining with anti-MBP (MalE) antibodies. Cells were also assayed for growth on maltose to confirm the correct orientation of the bitopic protein at the IM. Cells were incubated overnight with shaking in 0.4% glucose M9 minimal medium, streaked onto a 0.4% maltose M9 minimal medium plate, and then incubated for 2 days at 37°C to assay for growth. Cells expressing MBP in the periplasm (pMal-p2) and cytoplasm (pMal-c2) (New England BioLabs) served as positive and negative controls, respectively (22).

To quantify TM domain-mediated dimerization of ToxR, cell lysates were assayed for chloramphenicol acetyltransferase (CAT) activity (22, 38). Briefly, overnight cultures of cells expressing ToxR-TM-MBP fusion proteins were inoculated into 5 ml fresh LB and the resulting cultures were incubated at 37°C to an OD₄₂₀ of ∼1.0. Cells were normalized to the same OD₄₂₀ value, harvested, washed, and resuspended in 2 ml sonication buffer (25 mM Tris, 2 mM EDTA, pH 8.0). Cells were sonicated on ice for 4 min (Branson 250 Analog sonicator; output, 2; duty cycle, 50%). A sample of the whole-cell lysate was subjected to Western blot analysis to confirm the production of the fusion proteins. The remainder of the lysate was centrifuged at 13,000 × g for 20 min, and the supernatant was assayed for CAT activity by tracing absorbance at 412 nm for 5 min with a Beckman DU 530 UV/VIS spectrophotometer in a reaction mixture consisting of 10 μl cell lysate, 10 μl of 2.5 mM chloramphenicol, and 250 μl of reaction buffer [100 μM acetyl coenzyme A, 0.4 mg/ml of 5,5′-dithiobis(2-nitrobenzoic acid), 100 μM Tris-HCl, pH 7.8]. A blank reaction mixture lacking chloramphenicol was traced to determine the background absorbance. Reported CAT activities are mean values and standard deviations from three separate experiments.

RESULTS

The GA₄ dimer motif is dispensable for VirB10 function.

Large-scale surveys have shown that the GG₄ motif and the closely related GA₄ motif are significantly overrepresented pairs of amino acids in single-pass TM domain helices (39, 40). These motifs are the principal components of dimerization interfaces, but such helix-helix contacts also include additional flanking hydrophobic residues constituting a more extended LIxxGVxxGVxxT motif (where the underlined residues correspond to the GG₄ motif) (41–44). These extended motifs conform to an (abcd)_n repeat where the a and b residues adopt a right-handed helix-helix interface (40, 43). Here, we sought to determine whether a highly similar GA₄ motif (LIxxGVxxALxxS) in the A. tumefaciens VirB10 TM domain might promote homo- or heterotypic interactions required for the assembly or function of the VirB/VirD4 T4SS (Fig. 1A).

Fig 1.

Effects of GA₄ and Leu motif mutations on VirB10 function. (A) Schematics showing VirB10 domains, the TM domain sequence, and putative GA₄ and Leu dimerization motifs. Domains: Cyto, cytoplasmic; PRR, proline rich region; Linker, region between the PRR and the β-barrel domain; Barrel/AP, C-terminal β-barrel domain positioned near the OM/AP and extending across the OM. Numbers in the schematic and above the TM domain sequence correspond to residues relative to VirB10's N terminus. The extended GA₄ motif is depicted with the GA₄ residues in bold and underlined. The Leu1 motif is in bold and underlined, and the Leu2 motif is in plain letters. The helical-wheel diagram at the right depicts the VirB10 TM domain sequence and the putative Leu1 and Leu2 zipper motifs. (B) T-pilus production by nonpolar ΔvirB10 (ΔB10) mutant strain PC1010 lacking or producing native (B10) or the GA₄ or Leu motif mutant proteins shown. Top to bottom: B2 surface, colony immunoblot assays developed with anti-VirB2 antibodies; B2 shear, VirB2 comprising the T pilus detected in extracellular shear fractions; B2 cellular, VirB2 pilin detected in total cell lysates; B10, VirB10 detected in total cell lysates by immunostaining with anti-VirB10 antibodies. (C) Effects of mutations on T-DNA transfer as monitored by virulence on wounded Kalanchoe leaves (black bars; −, avirulent; +++, WT virulence) and transfer of the mobilizable IncQ plasmid pML122 to A. tumefaciens recipients (gray bars; Tc's/D, number of transconjugants per donor cell). For IncQ plasmid transfer, results of a single experiment with standard deviations from triplicate matings are presented.

In the well-characterized, strongly dimerizing TM domain of GpA, replacement of either glycine residue in the GG₄ motif with a bulky Ile disrupts the formation of TM domain homodimers (22). We similarly analyzed the effects of Ile replacement of Gly37 or Ala41 within VirB10's TM domain (Fig. 1A). The G37I and A41I mutant proteins comigrated with native VirB10 (48 kDa) and accumulated at abundant levels when produced in the ΔvirB10 mutant PC1010, suggesting that the GA₄ mutations were not destabilizing (Fig. 1B). Interestingly, A. tumefaciens strains producing these mutant proteins were proficient for the delivery of oncogenic T-DNA and protein effectors to susceptible plant hosts, as shown by results of plant virulence assays. The mutant strains incited the formation of plant tumors similar in size, morphology, and time course of appearance to those incited by the VirB10-producing strain (Fig. 1C). Both mutant strains also mobilized the transfer of the IncQ plasmid pML122, an RSF1010 derivative, to agrobacterial recipient cells. The G37I and A41I mutations conferred a slight attenuation of substrate transfer as monitored by both transfer assays (Fig. 1C) but clearly did not abolish protein function, in sharp contrast to the corresponding GG₄ mutations in GpA (45, 46).

The VirB/VirD4 T4SS also elaborates a T pilus, an extracellular filament important for the establishment of contacts between A. tumefaciens donors and bacterial or eukaryotic target cells. The G37I and A41I mutations did not disrupt T-pilus production, as shown by the presence of VirB2 pilin on the cell surface with a colony immunoblot assay and in high-molecular-weight material derived from the shearing of cells (Fig. 1B). Levels of pilin were comparable for strains producing native VirB10 and the GA₄ mutant proteins, whereas the ΔvirB10 mutant lacked detectable amounts of surface-exposed pilin. The VirB2 pilin undergoes maturation and assembly as a pool in the IM for subsequent recruitment to build the T pilus. As expected, ΔvirB10 mutant strains lacking or producing native or mutant forms of VirB10 accumulated similar levels of this cellular form of pilin (Fig. 1B).

The putative Leu1 zipper motif is important for T-pilus biogenesis.

In contrast to GG₄/GA₄-mediated dimerization, Leu zippers have a characteristic heptad motif (abcdefg)_n in which the Leu or Ile residue at the “a” position contributes to interhelical packing (44). Leu zipper motifs within TM domain helices of viral and eukaryotic membrane proteins mediate homo- or heterotypic TM domain dimers (44, 47, 48). The two possible Leu zipper motifs in the VirB10 TM domain, designated Leu1 (L33, L40, I47) and Leu2 (V35, L42, L49), are located on opposite faces of a heptad wheel (Fig. 1A). This arrangement was of interest in view of the pKM101 core complex structure wherein the TM domain helices of the 14 TraF monomers form a ring in the IM. We thus envisioned that the putative Leu zippers might mediate side-by-side contacts between the adjacent TM domain helices necessary for the assembly or stability of the core complex.

To assay for the functional importance of the Leu motifs of VirB10, we replaced residues in each motif with Ala because such substitutions have been shown to disrupt Leu zipper-mediated dimerization of other TM domain helices (44). The mutant proteins accumulated at abundant levels, suggesting that the mutations did not significantly disrupt protein stability (Fig. 1B). The mutations did not abolish substrate transfer through the VirB/VirD4 T4SS, as evidenced by infection assays monitoring T-DNA transfer to plants and IncQ plasmid transfer to agrobacterial recipients (Fig. 1C). Strains producing mutant proteins bearing single Ala substitutions (L33A, L40A, I47A, V35A, L42A, L49A) incited tumor production on plants at or near WT levels and also efficiently mobilized IncQ plasmid transfer to agrobacterial recipients. Strains producing the Leu1 and Leu2 triple mutant proteins (L33A L40A I47A and V35A L42A L49A) also translocated substrates, although at reduced levels compared with those of the WT strain. The putative Leu motifs thus contribute to but are not absolutely required for the assembly or activity of the translocation channel.

Surprisingly, however, the Leu1 and Leu2 zipper mutations had strikingly different effects on the assembly of T pili (Fig. 1B). Strains producing the Leu1 I47A and L33A L40A I47A mutant proteins lacked detectable surface-exposed pilin, and strains producing the L33A and L40A mutant proteins accumulated very low levels of this form of pilin, as monitored by colony immunoblot and shear assays. In contrast, strains producing the Leu2 mutant proteins accumulated abundant levels of surface-exposed pilin suggestive of WT T-pilus production. Although the blot presented in Fig. 1B shows some variation in the cellular form of VirB2 among the Leu1 mutant strains, in several repetitions of this assay, all of the Leu1 and Leu2 mutant strains accumulated comparable and abundant levels of the cellular form of pilin. The Leu1 mutations therefore appear to specifically block a step in the T-pilus biogenesis pathway subsequent to the formation of the IM pilin pool.

TM domain swaps have differential effects on substrate transfer and pilus biogenesis.

In spite of our findings that the GA₄ and Leu motif mutations did not abolish the elaboration of functional translocation channels, it remained possible that VirB10's TM domain mediates interactions with VirD4 or another channel subunit via an unidentified dimerization motif. To address this question, we replaced VirB10's TM domain sequence with two heterologous TM domain sequences, one from the cell division protein FtsN and a second composed of a nondimerizing poly-Leu/Ala (pLA) sequence (Fig. 2A).

Fig 2.

Effects of TM domain swaps on VirB10 function. (A) Alignment of TM domain sequences of native VirB10 (B10) and mutant TM domains bearing replacements of FtsN′s TM domain sequence and synthetic pLA sequences. Substituted residues are in bold, residues at positions corresponding to the Leu1 motif are gray shaded, and the position of the Leu1 motif is underlined at the bottom. (B) T-pilus production by the ΔvirB10 (ΔB10) mutant producing native (B10) or mutant proteins with the TM domain residue substitutions indicated. Two chimeric proteins with substitutions of the FtsN TM domain or cytoplasmic and TM domains (Cyto/TM) were analyzed. Top to bottom: B2 surface, surface-exposed VirB2 pilin protein; B2 shear, pilin in shear fraction; B2 cellular, total cellular levels of pilin; B10, total cellular levels of VirB10. (C) Effects of mutations on T-DNA transfer as monitored by plant tumor production (black bars; −, avirulent; +++, WT virulence) and mobilization of IncQ plasmid pML122 to A. tumefaciens recipients (gray bars; Tc's/D, number of transconjugants per donor cell).

FtsN is a bitopic protein that functions in the stabilization of FtsZ and other subunits of the divisome at the site of cell division (49). We replaced the VirB10 TM domain sequence (residues 35 to 50) with a corresponding TM domain sequence from FtsN (residues 38 to 54) (Fig. 2A). The VirB10-FtsN_TM chimeric protein accumulated at abundant levels, although a proteolytic breakdown product of ∼40 kDa was also abundant (Fig. 2B). Surprisingly, VirB10-FtsN_TM supported substrate transfer, as monitored by virulence on plants and IncQ plasmid transfer to agrobacterial recipients, although there was some attenuation of substrate transfer (Fig. 2C). Of further interest, the chimeric protein supported T-pilus production, as evidenced by the accumulation of abundant levels of VirB2 pilin on the cell surface and in the shear fraction (Fig. 2B). We also replaced both the cytoplasmic and TM domains of VirB10 with the corresponding regions of FtsN (residues 1 to 54). The resulting strain accumulated abundant levels of the chimeric protein but failed to translocate substrates or elaborate detectable levels of T pili (Fig. 2B and C). The FtsN TM domain substitution therefore does not block the formation of a functional translocation channel or the T pilus, whereas a swap of both the cytoplasmic and TM domains completely abolishes these activities.

Interestingly, VirB10-FtsN_TM has an intact Leu1 motif composed of Leu33 from the VirB10 TM domain sequence and Leu40 and Leu47 from the FtsN TM domain (Fig. 2A). Because this or another unspecified dimerization motif might account for the functionality of this chimeric protein, we next replaced the TM domain with a pLA sequence to more precisely evaluate the importance of TM domain dimerization for VirB10 function. Such sequences adopt α-helical secondary structures in membranes, but they do not form TM-TM helix interactions and thus have been exploited to assess the functional importance of TM domain dimerization of various proteins of interest (see below and references 44, 50–52, and 53). We replaced VirB10 TM domain residues 27 to 50 or 33 to 50 with pLA sequences of equivalent lengths. Note that the mutant pLA TM domain sequences retain Leu1 motif residues at positions 33 and 47 but have an Ala residue at position 40. The pLA TM domain swaps also eliminated an aromatic residue (Trp48) located near the membrane-periplasm interface that could serve as a topogenic signal (54). Consequently, we also constructed a pLA 33-to-50 swap that retained Trp48, as well as a W48A substitution, in the native VirB10 TM domain sequence (Fig. 2A).

The VirB10-pLA_TM chimeric proteins accumulated at abundant levels, as shown by SDS-PAGE, but also exhibited altered mobility in protein gels, as well some apparent degradation. VirB10-pLA27-50_TM and VirB10-pLA33-50/W48_TM comigrated with native VirB10 (48 kDa) but also formed lower-molecular-mass species of ∼40 and 42 kDa. VirB10-pLA33-50_TM migrated with an apparent molecular mass of ∼42 kDa, although an ∼40-kDa species also was detected (Fig. 2B). The W48A substitution in the native TM domain did not detectably alter protein stability (Fig. 2B).

Despite effects of the pLA TM domain swaps on protein mobility or stability, strains expressing the chimeric proteins elaborated functional translocation channels, as shown by their capacity to incite tumors on plants and mobilize IncQ plasmid transfer (Fig. 2C). The TM domain swaps did confer diminished IncQ plasmid transfer by 1 to 2 orders of magnitude, but regardless, the findings firmly show that VirB10 TM-mediated interactions are not essential for channel assembly or activity. In striking contrast, the pLA substitutions completely abolished T-pilus production, as shown by a lack of surface pilin in the colony immunoblot and shear assays (Fig. 2B). The W48A substitution in otherwise native VirB10 was phenotypically silent, establishing that Trp48 is not a critical topogenic signal (Fig. 2B and C).

We next asked if restoration of the Leu1 motif in the nondimerizing pLA motif would restore pilus biogenesis, and to this end, we constructed two types of amino acid changes (Fig. 3A). First, we introduced the A40L mutation to restore only the Leu1 motif in the pLA33-50/W48_TM chimeric protein. Second, we introduced Val, Leu, Ser, and Leu at positions corresponding, respectively, to residues 39, 40, 43, and 44 of native VirB10. These substitutions were made because studies elsewhere have shown that noninterfacial residues can contribute to Leu zipper-mediated TM domain dimerization (44), and by sequence alignment, we found that the possible noninterfacial residues Val39, Ser43, and Leu44 are invariant among close VirB10 homologs (see http://mmg.uth.tmc.edu/Assets/pdf/VirB10%20PD%20Allignment.pdf). As shown for VirB10-pLA33-50/W48_TM, the isogenic Leu40 and Val39 Leu40 Ser43 Leu44 (designated VL/SL) substitution mutant proteins migrated as ∼48-, 42-, and 40-kDa species in gels, although the VL/SL mutant protein migrated predominantly as the full-length (48-kDa) protein (Fig. 3B).

Fig 3.

Effects of pLA TM domain sequences bearing substitutions of residues conserved among close VirB10 homologs on VirB10 function. (A) Alignment of the VirB10 TM domain and pLA (33-50, W48) sequences. Substituted residues are in bold, residues at positions corresponding to the Leu1 motif are gray shaded, and the position of the Leu1 motif is highlighted at the top. The two variant pLA sequences bearing the conserved VL/SL and Leu40 residues are indicated below. (B) T-pilus production by the ΔvirB10 (ΔB10) mutant producing native (B10) or pLA mutant proteins with the VL/SL or Leu40 (L40) substitutions indicated. Top to bottom: B2 surface, surface-exposed VirB2 pilin protein; B2 shear, pilin in shear fraction; B2 cellular, total cellular levels of pilin; B10, total cellular levels of VirB10. (C) Effects of mutations on T-DNA transfer as monitored by plant tumor production (black bars; −, avirulent; +++, WT virulence) and mobilization of IncQ plasmid pML122 to A. tumefaciens recipients (gray bars; Tc's/D, number of transconjugants per donor cell).

Both of the pLA TM domain variants supported elevated levels of substrate transfer compared with those of the original pLA mutant (Fig. 3B and C). More strikingly, however, strains producing the pLA variants possessed abundant levels of surface-accessible VirB2 pilin (Fig. 3B). Both the Leu40 and VL/SL substitutions restored T-pilus production to similar extents, arguing against strong contributions by the Val, Ser, and Leu residues to protein function. The observed correlation between the introduction of a Leu1 motif into a nondimerizing pLA sequence and restoration of T-pilus production further supports our proposal that this motif is critical for the biogenesis of this extracellular organelle.

VirB10 TM helix mutations do not disrupt the VirD4 coupling protein interaction.

Because VirD4 is required for elaboration of the translocation channel but not the T pilus (20, 21), we predicted that the TM domain mutations conferring a transfer-positive (even if attenuated) and pilus-negative phenotype would not block requisite contacts between VirD4 and VirB10 or the core complex. As shown in Fig. 4 and previously (19), a complex composed of the core components VirB10, VirB9, and VirB7 and of VirD4 is precipitable from extracts of WT cells with anti-VirB10 antibodies. Similarly, we found that anti-VirB10 antibodies coprecipitated these four proteins from extracts of strains producing the GA₄ and Leu substitution mutants, as well as the FtsN and pLA TM domain-swapped mutants. In contrast, the core complex components, but not VirD4, were precipitated from extracts of strains producing the ΔN30 and ΔN46ss mutant forms, which lack the cytoplasmic domain and the cytoplasmic and TM domains, respectively. The core components, but not VirD4, were also precipitated from extracts of a VirB10-FtsN_Cyto/TM-producing strain. Together, these findings indicate that a stable VirD4 interaction with the core complex requires the cytoplasmic domain of VirB10 but not a specific sequence composition of the TM domain.

Fig 4.

Effects of TM domain mutations on VirB10 partner interactions. Extracts from DSP-treated and detergent-solubilized cells were immunoprecipitated with anti-VirB10 antibodies, and precipitates were analyzed by immunoblot development with antibodies specific to VirB10, VirD4, VirB9, and VirB7. Strains: PC1010 lacking (ΔB10) or producing native VirB10 (B10) or the TM domain mutants listed; Leu1, L33A L40A I47A variant (see Fig. 1); FtsN and pLA 33-50, W48 TM domain swaps (see Fig. 3); ΔN30 and ΔN46, VirB10 N-terminal truncations; FtsN Cyto/TM, VirB10 chimera with cytoplasmic and TM domains from FtsN.

VirB10 TM domain mutations do not disrupt OM channel gating.

Previously, we showed that the G272R mutation located near the OM-spanning AP domain of VirB10 promotes leakage of the VirE2 effector to the cell surface and also appears to “lock” VirB10 in an energy-activated state (Fig. 5; see reference 18). To determine if the VirB10 TM domain mutations might similarly disrupt energy sensing and channel gating, we assayed the mutant strains for the leaky-channel phenotype. However, none of the VirB10 TM domain mutations (GA₄ or Leu motif substitutions, i2 insertions, FtsN or pLA swaps) phenocopied the G272R mutation by mediating the release of VirE2 to the cell surface (Fig. 5). The VirB10 TM domain thus does not contribute directly to core-VirD4 complex formation (Fig. 4) or sensing of intracellular signals for channel gating (Fig. 5).

Fig 5.

Effects of TM domain mutations on release of VirE2 to the cell surface. Surface: extracellular FLAG-VirE2 (FL-E2) was assessed by colony immunoblotting with anti-FLAG antibodies. Cellular: total cellular levels of FLAG-VirE2 and native and mutant forms of VirB10 (B10) were monitored by immunoblot development with anti-FLAG or anti-VirB10 antibodies. Strains: PC1010 (ΔB10) producing native VirB10 (B10), the substrate-leaky mutant (G272R), or the VirB10 TM domain mutant proteins listed.

The VirB10 TM domain weakly self-associates.

Cys residues introduced throughout the periplasmic region of VirB10 form putative disulfide-cross-linked homodimers, but Cys residues located in the N-terminal cytoplasmic or TM domain do not spontaneously cross-link (19). To further assess whether VirB10's N-terminal region dimerizes, we pretreated cells producing Cys-containing VirB10 derivatives with dichloro(1,10-phenanthroline)copper (Cu-OP), an oxidative catalyst that can induce the formation of cross-linking of Cys residues within membranes (31). As shown in Fig. 6, Cu-OP pretreatment yielded cross-linked species of ∼100 kDa among all of the protein derivatives bearing N-terminal Cys residues, consistent with the formation of VirB10 dimers. Cys residues in the first 15 residues of the protein and in the TM domain formed intermolecular cross-links only in the presence of Cu-OP, whereas those located proximal to the TM domain formed cross-links even in the absence of the oxidative catalyst, suggestive of a close association of adjacent VirB10 monomers in the vicinity of the membrane. With the exception of the 45.i2 mutant bearing a Cys residue near the membrane-periplasm interface, which migrated exclusively as putative cross-linked dimers, mutant proteins with Cys residues within the TM domain migrated in approximately equal amounts as cross-linked dimeric and uncrosslinked monomeric species. This was of interest in view of the fact that the Cys residues were located at five-residue intervals and therefore along different faces of the TM helix, suggesting that the TM helix does self-associate but only weakly or in a conformationally flexible or switchable state.

Fig 6.

Oxidative cross-linking of Cys-containing VirB10 derivatives. PC1010 lacking (ΔB10) or producing native VirB10 (B10) or derivatives with the mutations indicated were left untreated (−) or treated with Cu-OP (+). Total cell extracts were prepared and electrophoresed through gels in the absence of reductant, and VirB10 species were detected by immunostaining with anti-VirB10 antibodies. The i2 mutants carry Ala-Cys insertions immediately after the residue number indicated. The positions of VirB10 monomer (B10) and putative dimer (B10-B10) species are indicated at the right. MW, molecular mass markers with sizes (in kilodaltons) listed at the left.

To quantitate the strength of the predicted TM-TM interaction, we made use of the TOXCAT assay (22). This assay measures the interaction affinities of TM domains in the E. coli membrane through the fusion of a TM domain sequence of interest at its N terminus to the ToxR DNA-binding domain and at its C terminus to periplasmic MalE (here termed MBP). Insertion of the helix into the membrane and self-association result in ToxR dimerization and activation of the ToxR-dependent ctx promoter, which drives the expression of the CAT-encoding gene (cat) (22). The amount of CAT activity in cell lysates serves as a measure of the strengths of helix-helix interactions. As controls and for comparisons of dimerization affinities, we quantitated the CAT activities of cells producing the strongly dimerizing ToxR-GpA_TM-MBP and nondimerizing ToxR-GpA(G83I)_TM-MBP fusion proteins (22).

As shown in Fig. 7A, the reporter strain for the native VirB10 TM domain exhibited CAT activity at a level only ∼25% of that of the GpA_TM reporter strain but still appreciably higher than that of the nondimerizing GpA(G83I)_TM reporter. All three fusion proteins accumulated at comparable levels in immunoblot analyses with anti-MBP antibodies and were properly oriented in the membrane, as monitored by the growth of E. coli MM39 (malE) strains engineered to produce the fusion proteins on maltose-containing plates (Fig. 7A and B) (22, 55).

Fig 7.

Self-association of native and mutant VirB10 TM domains as quantitated by TOXCAT assay. (A) Schematic of the MBP-TM-ToxR fusion protein. TM-mediated dimerization activates the ToxR-dependent ctx promoter controlling cat gene expression; total cellular CAT activity reflects the strength of TM domain self-association. Histogram depicts CAT activities normalized to that of the reporter strain for the strongly dimerizing GpA TM domain sequence. (B) Growth of MM39 cells expressing ToxR-TM-MBP proteins on maltose minimal medium. pMal-c2 producing a cytoplasmic form of MBP and pMal-p2 producing a periplasmic form of MBP served as negative and positive controls, respectively. (C) Histogram depicting CAT activities of reporter strains for the mutant TM domain sequences normalized to the CAT activity of the reporter strain for the native VirB10 TM domain. For panels A and C, the steady-state abundance of the fusion proteins was assessed by immunoblot development with anti-MBP antibodies.

Next, we tested the effects of various VirB10 TM domain substitution mutations on dimerization. CAT activities are reported as percentages of the CAT activity of the ToxR-B10_TM-MBP reporter strain (Fig. 7C). CAT activities of reporter strains for the GA₄ and Leu2 mutant TM domains were similar to the levels of the reporter for the native TM domain, consistent with the idea that these domains do not contribute to dimerization. In contrast, the CAT activities of reporter strains for TM domains bearing the i2 and Leu1 mutations were diminished by ∼40 to 50% compared to that of the reporter for the native TM domain. Additionally, the reporter strain for the pLA sequence exhibited CAT activity comparable to that of the reporter for the nondimerizing GpA(G83I) motif, and reporter strains for the variant pLA TM domain sequences (VL/SL, Leu40 substitutions) exhibited higher levels of CAT activities approaching that of the reporter for the native TM domain. All ToxR-TM-MBP proteins accumulated at comparable levels, and the E. coli MM39 strain producing these fusion proteins grew on maltose, which is suggestive of proper orientation across the membrane (Fig. 7B and C). Taken together, the results of our TOXCAT assays confirm that the VirB10 TM domain weakly self-associates and further suggest that the Leu1 motif contributes to some extent to dimerization.

Weak self-association of the VirB10 TM domain is biologically important.

Finally, we were intrigued by the findings that the Leu1 motif contributes to weak self-association of the TM domain and is also important for the biogenesis of the T pilus. We postulated that weak TM domain self-association might be essential for T-pilus production, and to test this model, we examined the effect of substituting the strongly dimerizing TM helix of GpA for that of VirB10. A. tumefaciens cells accumulated abundant levels of the VirB10-GpA_TM chimeric protein but showed diminished substrate transfer to plants and the IncQ plasmid to bacterial recipients (Fig. 8A and C). These cells also possessed undetectable levels of surface-exposed VirB2 pilin (Fig. 8B). Therefore, TM domain replacements with a nondimerizing (pLA) or a strongly dimerizing (GpA) TM domain sequence exert similar modest effects on channel assembly or activity while abolishing T-pilus biogenesis. These findings further reinforce the idea that weak self-association is an important biological feature of the TM domain.

Fig 8.

Effect of a GpA TM domain swap on VirB10 function (see Fig. 3). (A) Alignment of TM domain sequences of native VirB10 (B10) and the GpA TM domain variant. Substituted residues are in bold, residues at positions corresponding to the Leu1 motif are gray shaded, and the position of the extended GA₄ motif is underlined at the bottom. (B) T-pilus production by the ΔvirB10 (ΔB10) mutant producing the native (B10) or VirB10-GpA_TM mutant protein. Top to bottom: B2 surface, surface-exposed VirB2 pilin protein; B2 shear, pilin in shear fraction; B2 cellular, total cellular levels of pilin; B10, total cellular levels of VirB10. (C) Effects of the GpA_TM swap on substrate transfer as monitored by plant tumor production (black bars; −, avirulent; +++, WT virulence) and mobilization of IncQ plasmid pML122 to A. tumefaciens recipients (gray bars; Tc's/D, number of transconjugants per donor cell).

DISCUSSION

This study addressed the contributions of VirB10's TM helix to the assembly of the A. tumefaciens VirB/VirD4 translocation channel and T pilus. Our interest in this domain derives from evidence that VirB10 is a structural scaffold for the T4SS (2, 6) and also dynamically couples DNA ligand binding and ATP energy signals at the IM to channel gating at or near the OM (16, 17). In view of these activities, we sought to discriminate between two possible modes of action for the TM helix. The helix could form tight homo- or heteromeric interactions, for example, with polytopic VirB6 or bitopic VirB8 to coordinate channel assembly or with the VirD4 T4CP for signal sensing and channel activation. Alternatively, the helix could adopt a flexible conformation through the establishment of weak TM-TM helix contacts. Such weak interactions would be more appropriate for a dynamic, switchable system, since contacts that are too tight could impede conformational transitions between the quiescent and activated states.

For elaboration of the translocation channel, our data support the latter model, whereby the TM domain serves predominantly as a weakly self-interacting membrane anchor. Initially, we were surprised by the findings that mutations of the GA₄ motif did not alter VirB10 function or confer strong TM domain homodimer formation in the E. coli membrane, because such motifs commonly mediate dimerization among membrane proteins (44). However, besides the presence of small residues at the GxxxG/A positions, other features of the TM domain, including the position of the GG/A₄ motif relative to the ends of the TM helix and the composition of noninterfacial residues, contribute to TM-TM helix association affinities (56–58). In GpA (59) but not other membrane proteins (60), for example, an Ala substitution for the second Gly residue of a GG₄ motif strongly disrupts dimerization. Thus, despite the fact that VirB10's GA₄ motif (LIxxGVxxALxxS) closely resembles that of the extended GG₄ consensus motif (LIxxGVxxGVxxT), the C-proximal portion of VirB10's motif or other contextual features must disrupt dimer packing. Contextual features similarly might account for the failure of the Leu1 and Leu2 motifs to mediate strong helix-helix self-association, but as discussed further below, the Leu1 motif is absolutely required for T-pilus biogenesis, suggesting that it does mediate a TM-TM helix association that is critical for the elaboration of this organelle.

TM helix dimer interfaces and the functional importance of dimerization have been extensively explored for single-pass membrane proteins (41, 44, 56), but comparatively few studies have addressed the contribution of TM helix dimerization to the assembly of large, macromolecular complexes such as the envelope-spanning T4SSs. Our findings unequivocally establish that the sequence composition of the VirB10 TM domain, and hence TM helix-mediated dimerization, is not essential for elaboration of a functional VirB/VirD4 translocation channel. VirB10 forms a number of intermolecular contacts with other channel subunits, most notably, VirB7 and VirB9, to build the core complex. The assembly of this core complex is both intrinsically stabilizing for the core subunits and stabilizing for most of the other VirB channel subunits (61). Yet, nucleation of the core complex proceeds independently of TM helix-mediated interactions, a conclusion supported by results of our coimmunoprecipitation studies (Fig. 4) (19) and recent structural findings. By CryoEM, it was shown that a pKM101 core complex designated the D1 complex and made up of N-terminally truncated TraF (VirB10-like), TraN, and TraO structurally closely resembles the core complex formed by the native forms of these subunits (2).

VirB10-like subunits also interact with VirD4-like T4CPs via their N-terminal regions (9–11), but here we showed that a wide array of TM domain mutants does not abolish channel activity (Fig. 1 to 3), the formation of a precipitable core-VirD4 complex (Fig. 4), or signal-coupled channel gating (Fig. 5). In early genetic studies, evidence was presented that some chimeric conjugation machines are functional when assembled from a T4CP from one conjugation system and the translocation channel from another (9, 13–15). For example, the TraG T4CP from plasmid RP4 functionally replaces the TrwB T4CP to mobilize the transfer of an IncQ plasmid through the R388-encoded Trw transfer channel (13). Interestingly, however, we have not been able to identify any common dimerization motifs or even much sequence relatedness in the TM domains of exchangeable T4CPs, which argues against the assembly of high-affinity TM helix dimers as a basis for the functionality of these chimeric systems. Indeed, one mobilizable plasmid, pCloDF13, encodes its own T4CP that is capable of interfacing with different conjugation machines to mediate pCloDF13 transfer (62, 63). pCloDF13-encoded MobB is an unusual T4CP in that it completely lacks an N-terminal TM domain, as predicted by TMHMM, TopPred, TmPred, or other topology prediction algorithms (our unpublished data). Consequently, MobB most likely interacts via cytoplasmic domain contacts with other components of the translocation channel.

Although VirB10's TM domain does not contribute directly to the assembly of the core complex or coupling of VirD4 with the core complex, this domain also is not simply a passive membrane anchor. Mutations in the TM domain, particularly those in the Leu motifs and the pLA swaps (Fig. 2 and 3), attenuated substrate transfer, as did a TM helix swap with the strongly dimerizing GpA domain (Fig. 8). These findings suggest that weak TM domain self-association is important for optimal channel assembly or function. Weak self-association might contribute to the assembly of productive heteromeric interactions among the adjacent cytoplasmic or periplasmic domains, as suggested by our finding that VirB10's cytoplasmic domain is important for the precipitation of a core-VirD4 complex (Fig. 4). In addition, we recently showed that various mutations in the N-terminal periplasmic loop of VirD4 abolish the DNA ligand/ATP energy-induced conformational change of VirB10, suggestive of requisite contacts between the periplasmic regions of VirB10 and VirD4 for signal sensing and channel activation (17). Finally, if, as proposed elsewhere (2, 3, 61), the core complex is assembled at an early stage of T4SS machine biogenesis, conformational flexibility of VirB10's TM domain might be necessary for other channel subunits to gain entry to or be correctly positioned within the core's interior.

One of the most intriguing discoveries from this work was that VirB10's TM domain contributes in distinct ways to the assembly of the translocation channel versus the T pilus, as evidenced by the isolation of “uncoupling” mutations that are permissive for the assembly of functional translocation channels, albeit sometimes with diminished efficiency, but completely abolish the biogenesis of T pili. In fact, such mutations have been isolated in a number of the VirB subunits, including VirB2, VirB6, VirB9, VirB10, and VirB11, prompting a proposal that an early morphogenetic checkpoint exists to drive the assembly of either the translocation channel or the T pilus (19, 34, 61, 64–66). However, until now, our studies of these mutations have not offered any mechanistic insights into the nature of the signals or protein-protein interactions associated with this morphogenetic switch (61). Our present findings strongly imply the involvement of VirB10's TM domain, specifically, the Leu1 motif, in the promotion of T-pilus production.

How might the Leu1 motif specifically mediate T-pilus biogenesis? Although this motif contributes to the formation of weak TM-TM domain dimers (Fig. 7), this likely is not its only contribution to T-pilus assembly, given that weak TM domain self-association also seems to be important for the elaboration of fully functional translocation channels (Fig. 1 and 8). We propose instead that the Leu1 motif establishes another, heterotypic interaction of critical importance for T-pilus assembly. Further studies are needed to identify the putative binding partner(s), but two recent findings point to a possible interaction with the membrane-spanning (67) VirB4 ATPase or the small membrane protein VirB3 that acts in partnership with VirB4 (68). First, in a thiol accessibility study of Cys-substituted VirB2 variants, we gained evidence that VirB4 catalyzes a change in the topological state of membrane-integrated VirB2 consistent with its function as a pilin dislocase (65). Second, the Waksman group determined that the pKM101-encoded VirB4 homolog TraB associates with the side of the TraN/TraO/TraF core complex (4). Incorporating these findings into a general model of T-pilus biogenesis, VirB4 might catalyze the extraction of mature pilin monomers from the IM pool and, through its interactions with the core complex, promote the delivery of the pilins into the central chamber of the core complex for pilus polymerization (5, 6). Although more extensive interactions were predicted between VirB4-like TraB and the N-terminal region of VirB9-like TraN, the CryoEM structure clearly allows for TraB contacts with TraF's TM domain (4). Thus, a VirB10-VirB4 complex, mediated by the Leu1 motif, might serve as a point of entry for membrane-extracted pilin monomers across the core's wall into the central chamber. Such a translocation pathway would necessitate conformational flexibility in the core's wall, as would be afforded by weak self-association of VirB10's TM helices.

In sum, VirB10 has critical and complex functions both as a component of a structural scaffold for the T4SS and as a sensor/transducer of intracellular ligand- and ATP-energy signals for channel activation. Here, we report evidence that VirB10's TM domain contributes in mechanistically distinct ways to the assembly of the translocation channel and the T pilus. Membrane anchoring and establishment of weak TM domain homomeric interactions are important for the elaboration of both organelles. For the assembly of the translocation channel, the TM domain likely acts in partnership with the adjacent cytoplasmic or periplasmic domain to coordinate interactions with VirD4 and other channel subunits. For the assembly of T pili, the TM domain's Leu1 motif is essential, possibly forming a specific interaction with the VirB4 ATPase for pilin entry into the core's chamber. While the Leu1 motif might contribute in ways other than that of a Leu zipper or by mediating a VirB4 interaction, our findings nevertheless set the stage for further mechanistic studies aimed at deciphering how VirB10 coordinates early stages of T4SS machine morphogenesis.