Decoupling Filamentous Phage Uptake and Energy of the TolQRA Motor in Escherichia coli

Filamentous phages are widely distributed symbionts of Gram-negative bacteria, with some of them being linked to genome evolution and virulence of their host. However, the precise mechanism that permits their uptake across the cell envelope is poorly understood. The canonical phage model Fd requires the TolQRA protein complex in the host envelope, which is suspected to translocate protons across the inner membrane. In this study, we show that phage uptake proceeds in the presence of the assembled but nonfunctional TolQRA complex. Moreover, our results unravel an alternative route for phage import that relies on the ExbB-ExbD proteins. This work provides new insights into the fundamental mechanisms of phage infection and might be generalized to other filamentous phages responsible for pathogen emergence.

ABSTRACT

Filamentous phages are nonlytic viruses that specifically infect bacteria, establishing a persistent association with their host. The phage particle has no machinery for generating energy and parasitizes its host’s existing structures in order to cross the bacterial envelope and deliver its genetic material. The import of filamentous phages across the bacterial periplasmic space requires some of the components of a macrocomplex of the envelope known as the Tol system. This complex uses the energy provided by the proton motive force (pmf) of the inner membrane to perform essential and highly energy-consuming functions of the cell, such as envelope integrity maintenance and cell division. It has been suggested that phages take advantage of pmf-driven conformational changes in the Tol system to transit across the periplasm. However, this hypothesis has not been formally tested. In order to decouple the role of the Tol system in cell physiology and during phage parasitism, we used mutations on conserved essential residues known for inactivating pmf-dependent functions of the Tol system. We identified impaired Tol complexes that remain fully efficient for filamentous phage uptake. We further demonstrate that the TolQ-TolR homologous motor ExbB-ExbD, normally operating with the TonB protein, is able to promote phage infection along with full-length TolA.

IMPORTANCE Filamentous phages are widely distributed symbionts of Gram-negative bacteria, with some of them being linked to genome evolution and virulence of their host. However, the precise mechanism that permits their uptake across the cell envelope is poorly understood. The canonical phage model Fd requires the TolQRA protein complex in the host envelope, which is suspected to translocate protons across the inner membrane. In this study, we show that phage uptake proceeds in the presence of the assembled but nonfunctional TolQRA complex. Moreover, our results unravel an alternative route for phage import that relies on the ExbB-ExbD proteins. This work provides new insights into the fundamental mechanisms of phage infection and might be generalized to other filamentous phages responsible for pathogen emergence.

INTRODUCTION

Living cells produce various sources of energy in order to sustain essential reactions and physiological functions. In Gram-negative bacteria, the envelope is the location of heavy energetic processes. It is composed of two hydrophobic bilayers of lipids, namely, the inner membrane (IM) and outer membrane (OM), imbedded with or associated with numerous proteins. The two membranes delimit the periplasmic space, an aqueous compartment containing a thin layer of peptidoglycan (PG), and devoid of ATP. The ionic gradients across the IM constitute the proton motive force (pmf) and power a variety of dynamic processes that are essential for cell survival and multiplication, including ATP synthesis, flagellar rotation, energization of OM transporters, and trafficking of molecules against a concentration gradient (1–3). The protein complexes that convert the electrochemical energy into mechanical movements are referred to as molecular motors and are thought to cycle like gears between energized and nonenergized states. One of the best-characterized pmf-energized motor is the MotA/MotB IM complex that powers the rotation of the bacterial flagellum (2).

Filamentous phages are obligate parasites of Gram-negative bacteria that deliver their genetic material from one susceptible bacterial cell to another in order to replicate. However, phages have no protein machinery for generating energy (4, 5). While more than 70 different filamentous phages have been reported in the literature, the understanding of the host infection process is largely based on two types of viruses. Ff coliphages include fd, M13, and f1 and specifically infect Escherichia coli cells. They have served the development of extensive applications in genetic engineering and phage display technology (4, 5). The CTX vibriophage carries the genes encoding the cholera toxin in its genome and converts Vibrio cholerae to a deadly pathogen upon infection (6). In both cases, the general mechanism of filamentous phage infection involves the phage minor coat protein pIII located at the tip of the particle and two sequential receptors of the host, namely, a type IV pilus, which is somehow dispensable but increases the phage infection efficiency, and the TolQRA proteins, which are absolutely required for phage uptake (7–10). First, the phage specifically binds to the tip of the pilus protruding from the host cell surface (reception step) thanks to the central domain of pIII (pIII-N2) (11, 12). Pili are dynamic structures that normally undergo cycles of extension and retraction driven by ATPase activity at the cytoplasmic side of the IM (13). It is thought that coliphages like Fd and CTX vibriophage are brought close to the OM following pilus retraction of their target host (F pilus and toxin-coregulated pilus TCP, respectively) in a process that might not require ATP hydrolysis (14–16).

Once in the periplasmic space, filamentous phages require TolA, TolQ, and TolR for efficient infection (translocation step) (7, 8, 12, 17, 18). A direct interaction between the TolA C-terminal domain (TolAIII) and the phage pIII N-terminal domain (pIII-N1) has been documented (19–23), while the role of TolQ and TolR proteins remains unclear. TolA, TolQ, and TolR proteins are part of the Tol-Pal system, a pmf-dependent molecular motor conserved in Gram-negative bacteria. It is involved in maintaining OM integrity, in OM lipid homeostasis, and in the late stages of cell division (Fig. 1) (24–32). TolA is the central hub of the system. It is anchored to the IM thanks to a transmembrane (TM) domain and protrudes into the cell periplasmic space with a predicted long helical domain (TolAII) and a globular C-terminal domain (TolAIII). Besides TolA, the complex is composed of TolQ and TolR, which are embedded in the IM thanks to three TM domains and one TM domain, respectively. TolQ and TolR both interact with the TM domain of TolA, forming an IM subcomplex with a stoichiometry of four to six TolQs, two TolRs, and one TolA (33–37). The OM-associated subcomplex is composed of the peptidoglycan-associated lipoprotein (Pal) and the periplasmic protein TolB (38–40). TolQ and TolR are thought to form an ion channel at the protein TM helix interfaces, which allows the flow of protons from the periplasm to the cytoplasm (Fig. 1). As TolQ, TolR, and TolA interact in the IM, it is believed that the use of the pmf by the TolQ-TolR motor results in a conversion of the electrochemical potential into mechanical movements that will eventually trigger the stretch of TolAII across the periplasm and conformational changes in TolAIII. This leads to the formation of a transient TolAIII-Pal complex that has been observed by coimmunoprecipitation experiments in vivo (40). The system may alternate cycles of TolA-Pal binding and release, coordinated with the pmf-induced mechanical movements occurring in the IM subcomplex TolQ-TolR-TolA (41, 42). Deletion of one of the Tol proteins or dissipation of the pmf with the protonophore carbonyl cyanide m-chlorophenylhydrazone (CCCP) abolishes the link between the IM and OM Tol subcomplexes (40), resulting in pleiotropic Tol⁻ phenotypes, such as high sensitivity to detergents and chaining morphology under low osmolarity growth conditions (26).

FIG 1.

Schematic representation of the two homologous proton-motive force coupled systems. (A) The different components of Tol-Pal and the Ton-ExbB-ExbD are presented according to previously published literature. (B) Periplasmic view of TolA, TolR, and TolQ TM segment organization. While the system has been described to show a stochiometry of 1 TolA, 2 TolRs, and 4 to 6 TolQs, here only one protein of each is shown for clarity. The interfacial area delimitating the hypothetic aqueous channel is light gray. Residues predicted to be important for proton conductivity are indicated on TolQ TM2, TolQ TM3, and TolR TM segments. Residues S18 and H22, described as involved in energy transduction from the TolQ-TolR motor to TolA, are part of an SHLS motif and indicated on TolA TM domain. TBDT, TonB-dependent transporter.

The TolQR motor is evolutionary related to the pmf-dependent MotA-MotB motor generating the rotation of the flagellum and the AglR-AglS-AglQ system energizing a gliding motility machinery (36, 43). It also shares homologies with the ExbB-ExbD motor that energizes the TonB protein and is responsible for the energy-dependent uptake of iron-charged siderophores and vitamin B₁₂ (Fig. 1A) (36, 44, 45). The Ton system has also recently been shown to be involved in the secretion of a protease (46). Similarly to TolA being energized by the TolQR motor, ExbBD transduces the energy derived from the pmf of the IM to generate conformational changes into TonB that are required for active transport through the TonB-dependent transporters (47, 48).

Tol and Ton systems are both parasitized by bacterial toxins called colicins. These small proteins, produced by and active against Escherichia coli strains, are classified as group A and group B. Group A colicins, such as ColA and ColE1, use a subset of the TolQRAB proteins, whereas group B colicins, such as ColB, use the ExbBD-TonB proteins to penetrate and kill their target bacteria. It is thought that some colicins depend on energy-induced conformational changes in the Tol or Ton system to perform translocation, while others are still active in energy-deprived systems (26, 49, 50).

Filamentous phage translocation across the envelope has been suggested to require active pulling driven by pmf-induced conformation cycling of Tol system. Indeed, phage infection requires the host proteins TolA, TolQ, and TolR, which are part of a macrocomplex that uses the pmf for its normal functions. Moreover, cell exposure to the protonophore CCCP, which is able to dissipate the pmf, has been reported to strongly affect CTX phage transduction in V. cholerae (51). However, this chemical treatment is rather nonspecific, as it acts at the whole-cell scale and cannot differentiate between the energy requirement for pilus assembly and maintenance at the cell surface, pilus retraction and disassembly, and periplasmic transit of the phage particle.

In this study, we aimed to investigate the requirements for phage translocation across the host periplasm using a genetic targeted approach. We and others have previously identified polar residues on specific faces of the TolQ and TolR TM helices (Fig. 1B) that are essential for pmf-dependent functions of the Tol system (36, 42, 52). These residues are conserved within the TolQ or TolR protein families but also in the homologous bacterial motors ExbB-ExbD, AlgR-AglS, and MotA-MotB proteins. We specifically targeted residues in the TM helices of TolQ, TolR, or TolA in order to affect the assembly of the TolQR-TolA IM complex, to abolish pmf-dependent functions of the TolQR motor, or to impair energy transduction from TolQR to the phage receptor protein TolA. We also questioned if the ExbBD motor could be exchangeable with the TolQR motor for phage uptake.

RESULTS AND DISCUSSION

E. coli GM1 tol mutants display typical phenotypes.

Filamentous phage uptake in E. coli requires the F-conjugative pilus at the surface of the bacterium as primary receptors (reception) and the TolQRA proteins in the envelope as secondary receptors (translocation). tol mutants have been reported to be unaffected for both the synthesis of F pili, as well as the ability to undergo conjugation. Finally, the Leviviridae icosahedral phage f2, of which import is dependent on the F pilus but independent of the Tol system, is still able to infect E. coli Tol-defective mutants (53). These features allow the study of filamentous phage translocation in the periplasm of wild-type and tol-mutant strains independently of the initial F pilus-dependent reception step. We first generated various E. coli GM1 F-piliated strains lacking tolA, tolQ-tolR (referred hereafter as the ΔtolQR mutant), and exbB-exbD (ΔexbBD mutant) genes by P1 transduction and verified that they displayed typical phenotypes of tol and exb mutants. Compared with the GM1 wild-type strain, tol mutants showed an increased sensitivity to SDS (Fig. 2A), consistent with defects in the maintenance of OM integrity and resistance to two Tol-dependent colicins, namely, ColA (dependent on TolA, TolB, TolQ, and TolR) and ColE1 (dependent on TolA and TolQ) (Fig. 2B). As expected, the ΔtolQR ΔexbBD quadruple mutant was resistant to ColA and ColE1, as well as to the TonB-ExbBD-dependent colicin ColB (Fig. 2B). The mutant phenotypes were rescued by ectopic expression from compatible vectors of the wild-type copy of tolA, tolQR, or exbBD genes, as indicated (Fig. 2; see Fig. S1 in the supplemental material).

FIG 2.

Phenotypic analysis of E. coli GM1 wild-type strain and tol mutants used in this study. (A) Cell sensitivity to SDS is reported as the percentage of growth for each strain cultivated with 0.2% SDS compared with the same strain grown in LB medium. Experiments were conducted in triplicate. (B) Colicin sensitivity was estimated by 10-fold serial spot dilutions. One microliter of colicin (A, E1, or B) was spotted onto a growing lawn of cells. Clear zones indicate cell death. ColA and ColE1 are dependent on the Tol system, while ColB is dependent on the TonB-ExbBD system for uptake.

Phage infection occurs at low frequency in E. coli tolQ and tolR mutants but not in tolA mutant.

Filamentous phages have been proposed to hijack TolA, TolQ, and TolR proteins in order to transit across the periplasm of their target host and to position themselves for DNA ejection into the cytoplasm. So far, TolA is the only protein that has shown direct interaction with the phage in the periplasm (9). In E. coli, the deletion of tolQ or tolR was reported to totally abolish Ff phage infection, similarly to a tolA mutant (7, 8, 10, 53, 54). However, in V. cholerae, tolA deletion results in a CTX phage-resistant strain, while tolQ or tolR can still be infected at low frequency (17). These discrepancies might reflect differences in previous experimental setups used to conduct phage infection assays. In this study, we measured phage infection frequency in the different E. coli GM1 strain backgrounds using Fd-Tc coliphages carrying a tetracycline-resistant marker. In order to increase the sensitivity of our assay, we used a high multiplicity of infection with the aim to saturate the phage entrance pathway in the host cell (infection frequency, >10⁻¹ for the wild-type strain) (Fig. 3). With this experimental setup, we observed that the GM1 ΔtolA mutant was fully resistant to phage infection (infection frequency below the detection threshold of 1 × 10⁻⁶), while a ΔtolQR mutant remained susceptible to the Fd-Tc phage, albeit with a 5-log-unit decrease in infection efficiency compared with the wild-type strain (Fig. 3A). As a control, we verified the abundance of TolA in a ΔtolQR mutant compared with the wild-type (WT) strain (Fig. S2). The tolQR deletion resulted in a slight decrease in the total amount of TolA in all the tested strains, even when fully complemented with the pTolQR plasmid for the tested phenotypes (Fig. 2). This suggests that the decrease in phage susceptibility in ΔtolQR mutant was not the result of the decrease in the abundance of its receptor TolA. We concluded that the Fd phage strictly requires the host receptor TolA, while TolQ and TolR are important but not essential for efficient infection.

FIG 3.

Sensitivity to Fd-Tc phage and infection frequency for the WT, △tolA, △tolQR (A), and △tolQR△exbBD strains (B). Cells were incubated with the phage during 30 min and 10-fold serially diluted and spotted onto LB plates (left) and on LB plates supplemented with tetracycline (LB+Tc, right) in order to numerate the total number of CFU and Fd phage-infected CFU, respectively. Experiments were conducted in triplicate. The frequency of infection indicated on the right was calculated as the mean of the 3 infections with standard deviations.

The ExbBD motor can partially replace the TolQR motor for phage uptake but not for maintaining OM integrity.

Because a low level of infection is still observed in a ΔtolQR mutant (Fig. 3A), we hypothesized that the phage receptor TolA might be functioning with another IM-related system in the absence of the TolQ-TolR complex. ExbB-ExbD and TolQ-TolR motors share structural and functional homologies and have been reported to perform cross talk. Indeed, the TolQR complex can partially replace the ExbBD motor to energize TonB and to couple the energy of the pmf with OM transport cycle of vitamin B₁₂ (55) and with Ton-dependent colicin uptake. However, the reciprocal complementation of a TolQR mutant by the endogenous ExbBD motor does not support Tol-dependent OM integrity (for review, see reference 26 and Fig. 2B). In the case of a cross talk between the two systems for Fd phage uptake, we would expect to totally abolish phage infection in a quadruple mutant deleted of tolQ, tolR, exbB, and exbD. Indeed, the ΔtolQR-exbBD mutant was fully resistant to phage infection, as shown in Fig. 3B. Moreover, we observed that overexpressing the ExbB-ExbD-encoding genes from a plasmid in a ΔtolQR-exbBD mutant background promoted phage uptake to the wild-type level (Fig. 3B), as well as colicin E1 killing (Fig. 2B). As ColE1 import relies on TolA and TolQ but not TolR and, consequently, is thought to be energy independent, these data also suggest that ExbB can partially replace TolQ in the ColE1 translocation pathway. Of note, ExbB-ExbD protein did not cross-complement the mutant strain for ColA uptake, which is consistent with a previous study reporting that ColA translocation in the host requires a direct interaction with TolR (56). On the contrary, cells producing TolA with the ExbBD complex were sensitive to SDS, suggesting that pmf-dependent OM integrity is still impaired in this background (Fig. 2A). Taken together, these phenotypes suggest that in the absence of TolQR, the phage uses the ExbBD complex along with TolA as an alternative route to perform its translocation in the host envelope, in a process that is not coupled to the normal functioning of the Tol system.

The phage receptor domain TolAIII is strictly required but not sufficient for phage uptake.

We then questioned if the isolated phage receptor domain TolAIII and the TolQR complex would be sufficient for processing the uptake of the phage particle across the periplasmic space. Indeed, TolAIII is sufficient for binding the Fd phage protein pIII-N1 in vitro (19, 23) and in a bacterial 2-hybrid assay (21). It was also shown that overexpressing the isolated TolAIII domain in the host periplasm disturbs the normal functioning of the Tol system and decreases cell susceptibility to phage infection, possibly by sequestering the phage in the periplasm and preventing the infection to proceed to the next step (7). Using a similar approach in Fig. 4, we expressed the isolated TolAIII domain in the WT and in the ΔtolA cell periplasm using a pIN-III-ompA2 vector carrying a fusion between the sec-dependent ompA signal sequence and the tolAIII sequence (57). As a control, we used the pIN-pIII^ΔYGT vector that has been previously reported to produce an inactive pIII variant, which is unable to bind TolA in the periplasm (58). We first verified that periplasmic production of TolAIII in the WT strain impaired the ability of the cell to resist to SDS (Fig. 4A) and decreased the frequency of infection by Fd-Tc (Fig. 4B), while periplasmic production of the control pIII^ΔYGT had no effect. Then, we used the pIN-TolAIII vector to produce TolAIII in the periplasm of GM1ΔtolA cells and measured the frequency of Fd phage infection in this background. The TolAIII domain did not allow phage uptake despite the presence of the endogenous TolQR motor and contrary to the full-length TolA protein. We concluded that TolAIII deprived of its TM domain TolAI and periplasmic helical stretch TolAII is not sufficient for infection. This suggests that TolAI and/or TolAII also participate in the uptake process. Consistent with this hypothesis, an interaction assay conducted on isolated domains by surface plasmon resonance (SPR) analysis reported that TolAII binds the phage pIII-N2 domain with an apparent affinity of 1.3 to 2.4 μM (59). However, the significance of this interaction is unclear, as the deletion of the full helical domain TolAII reduced the efficiency of infection by a 5.5 factor but does not abolish it (7). It has been shown that TolAIII exists as different conformations depending on the interaction of TolAI with the TolQR complex in the IM and the energetic status of the cell (36, 40, 52). Thus, TolAII might serve as a dynamic spring, transducing mechanical movements from its IM anchor TolAI that will eventually facilitate TolAIII positioning in the periplasm for efficient phage translocation.

FIG 4.

Phenotypic characterization of various strains producing the isolated TolA receptor domain for the phage (TolAIII) in the periplasm compared with a control protein PIII^ΔYGT unable to bind TolA. (A) Cell sensitivity to SDS is reported as the percentage of growth for each strain cultivated with 0.2% SDS compared with the same strain grown in LB medium. Experiments were conducted in triplicate. (B) Sensitivity to Fd-Tc phage and infection frequency were determined as indicated for Fig. 3.

Phage uptake requires interaction between the TolA TM domain and TolQR.

We then questioned if TolA needed to be in interaction with TolQ and TolR in the IM to promote phage uptake. It has been previously reported that the TolQR complex assembles with TolA thanks to a conserved SHLS motif in the TolAI domain (52, 60, 61). Mutations of Ser18 or His22 residues in the TolA SHLS motif strongly impair the interaction of TolA with the TolQ-TolR motor, as seen in an in vivo cross-linking experiment (60) (Fig. S3 in the supplemental material) and prevent energy-dependent conformational changes of TolA and interaction with Pal in the OM (52). We first constructed plasmids producing TolA carrying a S18W or a H22W mutation and tested the ability of these variants to complement a ΔtolA mutant phenotype. The cells producing either TolA_S18W or TolA_H22W variants showed sensitivity to SDS, similarly to the ΔtolA mutant carrying a control plasmid (Fig. 5A), demonstrating that the mutations impair the ability of TolQ, TolR, and TolA to form an assembled and functional IM complex. However, the TolA_S18W and TolA_H22W proteins were able to support colicin E1 entry (Fig. 5B), which is known to depend only on the presence of TolA and TolQ, independently of the energy state of the Tol motor. Finally, the strains expressing the TolA_S18W or TolA_H22W proteins were more sensitive to TonB-dependent colicin B than the strain expressing wild-type TolA. These data are consistent with the observation that mutations in TolA TM domain impair interaction with the TolQR motor (Fig. S3), which may then become available to cross-complement the absence of ExbBD and to energize TonB, allowing ColB entry. Together, these data show that, unlike the wild-type TolA, the TolA_S18W and TolA_H22W variants are not properly assembled with the TolQR complex. Finally, we measured the frequency of phage infection in the GM1 cells expressing TolQR along with TolA_S18W or TolA_H22W proteins and found that the cells were less susceptible to phage infection (about a 3-log-unit decrease) than the cells expressing wild-type TolA (Fig. 5C). This suggests that optimum phage uptake requires interaction of TolA with the TolQR complex in the IM. However, vanishing interactions between the TolA mutants and TolQR might be sufficient for low-frequency phage transduction because uptake of a single phage results in the acquisition of the Tet resistance cassette by the host. These phenotypes are reminiscent of those obtained when the endogenous ExbBD complex partially cross complements the absence of TolQR for TolA-dependent functions (Fig. 2 and 3).

FIG 5.

Phenotypic consequences of the production of TolQ, TolR, and TolA or their variants in the △tolQR △tolA △exbBD mutant background. (A) Sensitivity to colicins and percentage of growth in the presence of 0.2% SDS compared with standard LB conditions. (B) Colicin sensitivity was estimated by 10-fold serial spot dilutions, as indicated for Fig. 2. (C) Sensitivity to Fd-Tc phage and infection frequency. Experiments were conducted as indicated for Fig. 3.

Phage uptake proceeds in the presence of an assembled but nonfunctional IM TolQRA complex.

We decided to further investigate whether Fd phages require that the Tol system be in a pmf-energized operating state to transit across the cell envelope, similarly to what has been described for group B colicins that parasitize the homologous TonB-ExbBD system (49). It is thought that energization of the Tol-Pal system relies on an ion channel that forms at the TM interface between TolR and TolQ. The conversion of the electrochemical energy into mechanical movements in the Tol system involves the TolR_D23, TolQ_T145, and TolQ_T178 polar residues, located in TolR TM, in TolQ TM2, and in TolQ TM3 segments, respectively (41, 42) (Fig. 1B).

Mutation of TolR D23 residue.

The Asp23 in TolR has been proposed to be the key ionizable residue of the channel and is also present and essential to the function of the homologous ExbD (Asp25), AglQ (Asp28), AglS (Asp41), and MotB (Asp32) proteins (2, 43). In cells producing a TolR_D23C variant, the IM TolA-TolQ-TolR_D23C complex is assembled (41) (Fig. S3) but totally defective for both energy-dependent and energy-independent functions. Indeed, the TolQ-TolR_D23C motor is unable to promote the interaction between TolA and Pal anchored in the OM, resulting in tol⁻ phenotypes (41, 42, 62). In our experiments, rescue of the ΔtolQR ΔtolAΔexbBD mutant was not observed with ectopic expression of the tolR_D23C variant along with WT copies of TolA and TolQ (Fig. 5), despite correct expression of the proteins (Fig. S1 and S3). We verified that this result was not caused by a pmf defect at the whole-cell level, as inactivating both the TolQRA and the ExbBD systems does not significantly disturb the membrane potential compared with the WT cells (see Fig. S4 in the supplemental material).

Finally, we found that filamentous phages were unable to penetrate the cells producing the TolQ-TolR_D23C-TolA complex. We hypothesize that in this energy-less state of the motor, TolA conformation prevents its binding to the phage or, alternatively, that the phage is stuck in the periplasm and cannot progress in the translocation pathway.

Mutation of TolQ threonine 145 and 178 residues.

Finally, we took advantage of mutants TolQ_T145A, and TolQ_T178A that have been reported to present discriminative phenotypes (62). In these backgrounds, the state of the motor differs from the TolQ-TolR_D23C-TolA, as the system is defective (TolA does not bind to Pal in vivo) but can still support pmf-independent processes like colicin uptake (62). As expected, the cells expressing either the TolQ_T145A or TolQ_T178A variants were sensitive to SDS but were still able to be killed by ColA and ColE1 (Fig. 5A and B). It is interesting to note that, unlike wild-type TolQ, these two mutants are unable to perform cross talk for ColB uptake (Fig. 5B), which requires a pmf-energized state of TonB, consistently with the hypothesis that the motors composed of TolR and TolQ_T145A or TolQ_T178A variants are unable to transduce energy to a partner. Finally, we observed that filamentous phages were able to infect cells expressing the TolQ_T145A or TolQ_T178A variants at a frequency similar to that for the WT strain. We concluded that Fd phage translocation in the host periplasm was not dependent on a functional TolQRA system. Thus, our data challenge the model that filamentous phages use the energy-dependent pulling action and cycling of TolA to cross the periplasm and reach the IM.

The data presented here have to be interpreted in the context of normal functioning of the Tol-Pal system for OM maintenance and during cell division. Indeed, it has been speculated that TolA cycles between different conformations, depending on its energized or nonenergized states regulated by the TolQR motor, similarly to TonB energized by the ExbBD motor (63). In the current working model, energy-loaded TolA interacts with the OM lipoprotein Pal, bridging the IM and the OM layers. During cell division, conformational changes in the Tol system provide the energy needed for the OM to be pulled toward the pinch point (29, 32). The mechanisms allowing progression of OM invagination remain to be fully elucidated but might require cycles of TolA-Pal interaction and release. Taken together, our data on TolR_D23, TolQ_T145, and TolQ_T178 highlight the existence of distinct nonfunctional states of the TolQRA complex that might reflect sequential steps in motor cycling, coordinated with the ion conduction pathway inside the complex. They also suggest that phage import requires assembly and the initial activation state of the TolQRA complex, which cannot be reached when the TolR Asp23 residue is mutated. At this point, it is unclear if this initial activation step of the TolQRA complex is dependent on proton progression in the motor channel or not. Previous data on V. cholerae showed that dissipation of the pmf by the protonophore CCCP abolishes CTX phage infection (51). It is possible that CCCP treatment results in a nonfunctional TolQR complex that blocks TolA in a conformation that is not able to support phage translocation in the periplasm. This frozen state might be similar to the one obtained when TolA interacts with the TolR_D23C variant protein (Fig. 5). An alternative explanation could be that CCCP treatment impacts other mechanisms, upstream or downstream of the Tol reception step. Indeed, both pmf and ATP depletion can be observed following CCCP treatment (64). Moreover, exposure to CCCP has been reported to decrease the total number of F pili extruding from the cell surface (63), and eclipse of the F pili is an ATP-requiring process that could be impacted by a general decrease of the pmf. Finally, opening of the phage head particle at the OM and final DNA injection in the cytoplasm might require the presence of a pmf across the cytoplasmic membrane of the host cell, as reported for T4 phage (65). While pmf does not seem to power phage translocation, this energy could serve other steps of the process, such as pilus retraction or opening of the phage particle head in the IM. Alternatively, the data might reflect the high energy requirement for cell OM integrity maintenance, which cannot be reached when TolA operates with the TolQ_T145A or TolQ_T178A variants, or with the homologous ExbBD motor. In this case, the defective Tol IM complexes could either operate with a very slow cycling or poor torque. Measuring the effect of TolQ or TolR point mutations on channel conductance would help the understanding of the different stages in the Tol motor cycling but requires reconstitution of the system in vitro (for example, in planar lipid bilayers). While this approach has been set up for the homologous motor ExbB-ExbD (44, 66), it has not been set up for the TolQ-TolR system so far.

In conclusion, our data demonstrate that filamentous Fd phage transit in the periplasm requires the presence of an assembled TolQ-TolR-TolA complex anchored to the IM. The system needs to reach a first activation state (depending on TolR Asp23 residue) but does not require it to be functional to promote phage uptake (as attested by TolQ T145 and T178 variants). We also show that the homologous ExbBD complex can support phage uptake along with TolA in the absence of the TolQR complex. Although a detailed mechanical analysis of the Tol proteins is beyond the scope of this study, it would be of great interest to understand the structural and dynamic basis for the coordination between the phage particle and the Tol components during the cell cycle. As some the filamentous phages are linked to the pathogenicity potential of their host, deciphering the energy requirements for these potentially important bacterial parasites might lead to novel intervention strategies.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacterial strains and plasmids used in this study are listed in Table S1 in the supplemental material. Bacteria were cultivated in Luria-Bertani (LB) broth at 37°C. When indicated, antibiotics were added to the medium at the following concentrations: 100 μg/ml, streptomycin; 50 or 100 μg/ml, ampicillin; 50 μg/ml, kanamycin; and 15 μg/ml, tetracycline. For E. coli GM1 deletion strains, the recombinant genes were transferred from W3110 strain (32) to the desired strain background by P1 transduction (67).

Plasmid construction.

PCRs were performed using the Q5 high-fidelity DNA polymerase (New England BioLabs). Primer sets required to generate genetic constructs were synthesized by Sigma-Aldrich. Enzymes (New England BioLabs) were used according to the manufacturer’s instructions. Plasmids were constructed using standard cloning technics, as previously described (21, 41). Mutations on pBAD-tolQR and pOKtolA plasmids were performed by QuikChange site-directed mutagenesis using complementary pairs of oligonucleotides (listed in Table S2 in the supplemental material) and Pfu Turbo polymerase. All constructs were confirmed by DNA sequencing (Eurofins, MWG).

SDS-PAGE and immunoblotting.

Protein samples resuspended in 2× loading buffer (100 mM Tris-HCl [pH 6.8], 2% SDS, 10% glycerol, 0.01% bromophenol blue, and 5% 2-β-mercaptoethanol) were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). For detection by immunostaining, proteins were transferred onto nitrocellulose membranes, immunoblots were probed with primary antibodies, and goat secondary antibodies were coupled to alkaline phosphatase and developed in alkaline buffer in the presence of 5-bromo-4-chloro-3-indolylphosphate (BCIP) and nitroblue tetrazolium (NBT). The anti-TolAIII and anti-TolR polyclonal antibodies are from our laboratory collection, while the antihemaglutinin (anti-HA) monoclonal antibody (Invitogen) and alkaline phosphatase-conjugated goat anti-rabbit and anti-mouse antibodies (Millipore) were purchased as indicated.

Phenotypic analysis.

Cells harboring the empty plasmid as a control or the plasmid encoding the constructs of interest were grown in LB medium until stationary phase and then back diluted 100× in LB supplemented with 0.02% l-arabinose or 100 μM isopropyl-β-d-thiogalactopyranoside (IPTG) if necessary. After 1-h incubation at 37°C, cultures were adjusted to an optical density at 600 nm (OD₆₀₀) of 0.2 in LB supplemented or not with 0.2% SDS, transferred to a 96-well plate (final volume of 300 μl), and static grown for 5 h at 37°C. The OD₆₀₀ was measured using a Tecan instrument. The percentage of surviving cells was estimated from the turbidity ratio of the SDS-treated cells and the control samples. Experiments were performed in triplicates.

Colicin susceptibility.

Colicin activities were observed by the presence of clear halos on a lawn of the strain to be tested, as described previously (56). Briefly, overnight cultures of the strains were spread onto LB agar petri dishes supplemented with antibiotics, and plasmid expression was induced as specified, using l-arabinose (0.02%) and IPTG (100 μM). After drying, 1 μl of serial dilutions (10-fold) of colicins was spotted on the bacterial lawn. Plates were incubated at 37°C for 16 h. The data are reported as the maximal dilution of the colicin stock sufficient for inhibiting cell growth. Experiments were conducted in duplicates.

Fd-Tc phage preparation and titration.

A 5-ml preculture of the GM1 strain was grown at 37°C. At an OD₆₀₀ of 0.5 to 0.6, it was inoculated with a colony of E. coli JM101 Fd-Tet (ATCC 37000) and incubated for 2 hours at 37°C. The preculture was then used to inoculate a 500-ml culture of LB and incubated for 1 hour at 37°C with shaking. At this point, tetracycline (15 μg/ml) was added to select infected cells only. After overnight incubation at 30°C with shaking, the culture was centrifuged at 10,800 × g for 15 min at 4°C. The supernatant containing the phages was precipitated with one-fourth volume (V) of a polyethylene glycol (PEG)-NaCl solution (20% polyethylene glycol 8000; 2.5 M NaCl) for 2 hours on ice, and phages were collected by centrifugation at 10,800 × g for 30 min at 4°C. The pellet was resuspended in 8 ml PBS, and a second round of precipitation with one-fourth V PEG/NaCl solution was performed 20 min on ice. The phage solution was centrifuged at 3,300× g for 30 min, and the pellet was resuspended in 5 ml PBS. Remaining bacterial debris were removed by centrifugation at 11,600 × g for 10 min and filtration with a 0.2-μm syringe. Phage preparations were checked for sterility by plating onto an LB plate.

Titration of the phage suspension was performed as follows: the phage suspension was serially diluted in sterile PBS (10-fold), and 10 μl of each phage dilution was incubated with 200 μl of GM1 receptor cells grown at OD₆₀₀ of 0.5 to 0.6. After 30 min of incubation at room temperature without shaking, the cells were vortexed for 5 s, and 5 μl was dropped onto an LB plate supplemented with 15 ng/μl tetracycline (Tc). After overnight incubation at 37°C, CFU were counted in the highest dilution test, leading to titer determination of 10¹³ to 10¹⁶ phages/ml.

Susceptibility to Fd-Tc phage infection assays.

Strains of interest were cultivated to reach an OD₆₀₀ of 0.7 to 0.8 and normalized to the same initial OD₆₀₀. A total of 10 μl of phage suspension was added to 200 μl cells (multiplicity of infection of 10,000 phages per bacterium). Infection assays were performed in triplicate in 96-well plates, during 30 min of incubation at room temperature without shaking. The cells were vigorously homogenized by pipetting and immediately serially diluted 10-fold in sterile PBS. A total of 5 μl was dropped onto an LB plate (total recipient CFU) or LB agar supplemented with Tc (15 ng/μl) (phage-infected CFU). After overnight incubation at 37°C, isolated CFU were counted in the appropriate dilution test. The frequency of infection was determined by dividing the number of infected cells by the number of total recipient cells. Experiments were conducted in triplicate.