The Twin-arginine Leader-binding Protein, DmsD, Interacts with the TatB and TatC Subunits of the Escherichia coli Twin-arginine Translocase*

The twin-arginine translocase (Tat) pathway is involved in the targeting and translocation of fully folded proteins to the inner membrane and periplasm of bacteria. Proteins that use this pathway contain a characteristic twin-arginine signal sequence, which interacts with the receptor complex formed by the TatBC subunits. Recently, the DmsD protein was discovered, which binds to the twin-arginine signal sequences of the anaerobic respiratory enzymes dimethylsulfoxide reductase (DmsABC) and trimethylamine N-oxide (TMAO) reductase. In this work, the targeting of DmsD within Escherichia coli was investigated. Using cell fractionation and Western blot analysis, DmsD is found to be associated with the inner membrane of wild-type E. coli and a dmsABC mutant E. coli under anaerobic conditions. In contrast, DmsD is predominantly found in the cytoplasmic fraction of a ΔtatABCDE strain, which suggests that DmsD interacts with the membrane-associated Tat complex. Under aerobic conditions DmsD was also found primarily in the cytoplasmic fraction of wild-type E. coli, suggesting that physiological conditions have a significant effect upon the targeting of DmsD to the inner membrane. Size exclusion chromatography data and membrane washing studies indicate that DmsD is interacting tightly with an integral membrane protein and not with the lipid component of the E. coli inner membrane. Additional investigation into the nature of this interaction revealed that the TatB and TatC subunits of the translocase are important for the interaction of DmsD with the E. coli inner membrane.

In prokaryotes, proteins can be transported across the cytoplasmic membrane in various manners. A common method for protein translocation is the general secretory (Sec) 1 pathway, which threads proteins across the membrane in an unfolded state (1,2). In contrast, the discovery of a Sec-independent pathway reveals that fully folded proteins can also be translocated across the membrane (3)(4)(5). This pathway is now referred to as the twin-arginine translocase (Tat) protein-export pathway. Proteins that use the Tat system, including many respiratory enzymes that bind redox-active cofactors, contain a characteristic twin-arginine motif SRRXFLK in their signal peptides (6). This conserved twin-arginine signal sequence targets precursor proteins to the membrane-bound Tat complex (7).
TatA and TatE are homologous proteins with overlapping functions in the Tat pathway, and recent studies have suggested that tatE may be a cryptic gene duplication of tatA (4,5,17). It has also been suggested that TatA might form the transport channel (7,18,19). Though TatB is sequence related to TatA and TatE (4,5), it has a distinct role in the translocation of proteins (12). The tatC gene product is predicted to be a polytopic membrane protein with four (20) or six (19) transmembrane helices and is considered capable of forming a specific binding site for the twin-arginine signal sequence (7,18). Studies involving the thylakoid Tat system have shown that TatBC homologues form a complex that recognizes precursor proteins (21) while in the bacterial system TatB and TatC have also been shown to form a functional and structural unit, which acts as a receptor complex for the twin-arginine leader sequence of Tat-bound proteins (10,14,19).
Recently, a protein that specifically binds the twin-arginine signal sequence of the E. coli dimethylsulfoxide (Me 2 SO) reductase (DmsA subunit) was discovered (22). This 204-residue protein, DmsD, has homology to the TorD family of molecular chaperones (19). Sequence analysis predicts that members of this family comprise at least two distinct structural domains (22,23). DmsD has also been shown to interact with the precursor form of trimethylamine N-oxide (TMAO) reductase (TorA) (22), a DmsA homologue that also binds a molybdopterin cofactor. However, DmsD is unable to interact with the fully folded mature forms of DmsA and TorA, suggesting that it interacts with the twin-arginine signal sequence (22). Recent speculation proposes that DmsD might also have the ability to interact with the cofactorless (unfolded) mature portions of the two redox enzymes (18,19).
An understanding of the role of DmsD in the Tat system is still in a preliminary stage. Previously, it was proposed that specific leader binding proteins might exist to escort twinarginine containing proteins to the Tat translocase (22). In this study, the targeting of DmsD to the E. coli inner membrane was investigated. It was shown that under anaerobic conditions, DmsD is associated with the inner membrane in wild type (WT) and dmsABC mutant strains of E. coli. However, in the absence of the Tat translocase complex, DmsD is found primarily in the cytoplasmic fraction. DmsD was also predominantly located in the cytosolic fraction of WT E. coli under aerobic conditions, indicating that the cellular physiology strongly influences the targeting of DmsD. Upon investigation of the nature of the interaction of DmsD with the membrane under anaerobic conditions, it was shown that DmsD interacts tightly with an integral membrane protein. Further investigation suggests that the TatB and TatC subunits are important for the interaction of DmsD with the membrane-associated Tat translocase. Thus, this work shows that the twin-arginine leader binding protein, DmsD, interacts with the TatBC twinarginine signal sequence receptor complex on the E. coli inner membrane.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-E. coli strains (Table I) DSS301, DSS640, DSS642, DSS643, and DSS644 were kindly provided by Dr. D. Sambasivarao and Dr. J. Weiner (University of Alberta, Edmonton, Alberta, Canada). Strains ELV16, J1M1, JARV16 and DADE were kindly provided by Dr. F. Sargent (University of East Anglia, Norwich, UK). Each strain was transformed with pTDMS28, a vector expressing DmsD with an N-terminal T 7 epitope tag (22). Transformants were verified by performing a Qiagen plasmid preparation.
Growth Conditions of Anaerobic Cultures and Cell Fractionation-1 liter anaerobic cultures were grown for 58 h at 37°C in a modified peptone fumarate media (24), supplemented with 2% glucose and 0.1 mM ampicillin. Upon harvesting, the cell paste was resuspended in a 10-fold (w/v) excess of 20 mM Tris, pH 7.9 and 1 mM dithiothreitol (buffer A). The cell suspension was passed twice at 16,000 p.s.i through a French Pressure cell then centrifuged at 7800 ϫ g for 25 min to remove unlysed cells, particulate, and protein aggregates (particulate fraction). Next, the clarified cell free extract was centrifuged at 165,000 ϫ g for 90 min to separate the cytoplasmic and membrane fractions. The membrane fractions were resuspended and thoroughly washed in a 20-fold (w/v) excess of buffer A by 5 min of homogenization using a hand-held Wheaton 5-ml homogenizer. The membrane suspensions were centrifuged at 424,000 ϫ g for 35 min to obtain the soluble and washed membrane fractions. Washed membranes were resuspended by homogenation in a 10-fold (w/v) excess of buffer A. All centrifugations were performed at 4°C, and an aliquot of each fraction was kept. Pure DmsD was obtained following the methods of Oresnik et al. (22) and was used within a week of purification. A modified Folin-Lowry method (25) was used to determine the protein concentration of each fraction and bovine serum albumin Bio-Rad Protein Assay Standard II was used to obtain the standard curve.
Protein expression levels were checked by Western blot using an antibody against the T 7 epitope tag to ensure equivalent DmsD expression in all strains. DmsD was expressed through the leakiness of the T 7 promotor, not through induction. Thus, DmsD was not overexpressed in the various strains in the studies reported here. Analysis of DmsD Localization-SDS-PAGE (12%) was performed using 50 g of the 5 fractions of interest: the particulate, cytoplasmic, soluble, membrane, and washed membrane fractions. Additionally, 3.75 g of purified DmsD was loaded onto each gel. The presence of DmsD in the various fractions was determined through Western blot analysis against the T 7 epitope tag, using a 1:5000 dilution of T 7 -HRP conjugate antibody (Novagen) in a 0.05% Tween-20 Tris buffered saline solution, pH 7.5. The bands were visualized using an HRP development buffer (Bio-Rad), and quantitated with NIH Imaging software. The percentage of DmsD in the cytoplasmic and washed membrane fractions was determined as follows: First, the micrograms of DmsD present in each band from each cell fraction was calculated by comparing the density of each band to the purified DmsD band, following Equation 1. [DmsD] fraction of interest ϭ (g DmsD) band of interest /(l fraction) loaded onto gel (Eq. 2) The percentage of DmsD localized to the membrane fractions relative to the cytoplasmic fractions was calculated by totaling the micrograms of DmsD in both fractions, and setting that value to 100% as in Equation 4.
Growth Conditions of Aerobic Cultures-Aerobic cultures (1 liter) of TG1 and tatABC mutant E. coli strains, transformed with pTDMS28 (Table I) were grown in 4-liter Fernbach flasks at 37°C in LB media in a rotary shaker (250 rpm). Two of the cultures were induced with a final concentration of 0.1 mM IPTG at OD 600 ϭ 0.5, while the two other cultures were not induced. All four cultures were grown for an additional 2.5 h prior to harvesting. Cell fractions were purified and the presence of DmsD determined following the procedure described for the anaerobic cultures.
Preparation of E. coli Lipid Vesicles (Liposomes)-Small unilamellar E. coli lipid vesicles (liposomes) were made using 125 l of Avanti E. coli polar lipid extract (25 mg/ml). The lipid was dried with N 2 (g) then resuspended in 1 ml of buffer A. The solution was frozen at Ϫ70°C then thawed with warm water and vortexed. The freeze-thaw cycle was repeated five times to create multilamellar vesicles. Small unilamellar vesicles were created by repetitive sonication of the multilamellar vesicles using a Microson Ultrasonic Cell Disruptor at level 10 for 10 s, until the cloudy solution became clear (ϳ6ϫ). 3 ml of buffer A was added to the liposomes, which were stored at Ϫ20°C and used within 5 days of preparation. Liposome Interaction Studies Using Size Exclusion Chromatography-These studies were based on the methods of Millman et al. (26). Purified DmsD (0.5 mg/ml) was incubated with various ratios (wt/wt) of liposomes [2:1, 1:1, 1:2, and 1:10] in 100 l for 1 h at 23°C. The solutions were loaded onto an XK-16 Amersham Biosciences column containing 12 ml of Sepharose CL-4B resin (Amersham Biosciences), with inclusion limits of 60 kDa to 2000 kDa. The column was equilibrated with a minimum of two column volumes of buffer A, filtered with a 0.20 m White Nylon Millipore filter. Samples were also eluted with buffer A, and 1-ml fractions were collected. The column was calibrated using blue dextran as a marker of void volume (V o ), and the elution volumes of DmsD and the liposomes were also determined and used for reference. High molecular weight DmsD aggregates were removed prior to experimentation by centrifugation at 10,000 ϫ g for 10 min then the supernatant was filtered using a 0.20 m sterile syringe. Chromatograms were analyzed using Unicorn 3.21 software (Amersham Biosciences).
Proteins present in each 1-ml fraction were precipitated by the addition of 100 l of 100% w/v trichloroacetic acid. The fractions were vortexed then incubated at Ϫ20°C for 20 min. Next, samples were centrifuged at 10,000 ϫ g for 5 min, and decanted. The precipitate was resuspended in 10 l of 0.1 N NaOH and 10 l of Laemmli Solubilization Buffer (LSB: 50 mM Tris, pH 6.8, 0.1 M dithiothreitol, 2% SDS, 2% glycerol, and 0.1% bromphenol blue) then analyzed by 12% SDS-PAGE. The presence of the T 7 epitope-tagged DmsD was visualized by Western blot, as described in previous sections.
Washing Studies of E. coli Membranes-Washed membranes derived from WT (MC4100 and TG1) E. coli transformed with pTDMS28 (described above) were resuspended with 5 min of homogenization and washed with a 35-fold (v/v) excess of various buffers: double distilled H 2 O, buffer A, 10 mM NH 4 HCO 3 , 1% Triton X-100, 20% isopropyl alcohol, 2% w/v n-dodecyl-␤-D-maltoside, 10 mM EDTA, and 4 M urea. 1 ml of the respective wash buffer was added to the WT/pTDMS28 membranes at 4°C, containing 200 g of total protein. The samples were vortexed for 10 s, and then incubated on ice for 15 min. This was repeated four times, then samples were centrifuged for 35 min at 424,000 ϫ g at 4°C. The supernatants were removed and proteins were precipitated as described in the previous section. The membrane pellets were resuspended in 20 l of LSB, and in the case of the membrane samples washed with acetic acid, an additional 1 l of 1 M NaOH was added. Both membrane and supernatant samples were analyzed by 12% SDS-PAGE, and the presence of DmsD in the various fractions was determined through Western blot analysis against the T 7 epitope tag.

RESULTS
DmsD Interacts with the E. coli Inner Membrane under Anaerobic Conditions-In order to determine the location of DmsD in WT MC4100 E. coli transformed with pTDMS28, anaerobically grown cultures were harvested and the cells were fractionated. Western blot analysis of whole cells revealed equivalent expression of tagged DmsD in all strains and expression levels were further verified by Western blotting of the particulate fractions and cell free lysates (data not shown). As previously mentioned, DmsD was not overexpressed in these cells, thus targeting within the cell would not be due to expression artifact. Western blot results against the T 7 epitope tag clearly show that under anaerobic conditions, DmsD is associated with the E. coli inner membrane fraction and is not found in the cytoplasmic fraction (Fig. 1, lane 1). Further rigorous washing of the membranes with buffer did not result in dissociation of DmsD from the membrane since DmsD was not present in the soluble fraction following additional buffer washes (results not shown). This result was obtained in 5 independent experiments, using both MC4100 and TG1 E. coli strains (Fig. 1, lanes 1 and 6).
Location of DmsD in Various tat Mutants-In contrast to the WT results, DmsD was found in both the cytoplasmic and membrane fractions of a ⌬tatABCDE strain transformed with pTDMS28 (Fig. 1, lane 2). The number of micrograms of protein loaded into each well of the SDS-PAGE gel was normalized, thus the Western blot band intensities are not indicative of the relative percentage of DmsD located in a given fraction of E. coli. In order to determine the percent DmsD in the cytoplasmic fractions relative to the membrane fractions of E. coli cells arising from a 1-liter culture, the Western blot data was quantitated. Quantitation of DmsD present in the ⌬tatABCDE 1-liter culture revealed that 95% of DmsD was located in the cytoplasmic fraction compared with only 5% that remained associated with the membrane fraction (Fig. 2). As in the WT cells, a rigorous buffer wash did not result in dissociation of the remaining DmsD from the membranes of the tat mutants (Fig. 1).
To determine which subunits were important for the interaction of DmsD with the Tat translocase, the location of DmsD in 6 tat mutants was assessed. As in WT E. coli, DmsD was located exclusively in the membrane fractions of strains with tatA, tatD, tatE, or tatAE gene deletions (Fig. 1). In contrast, the ⌬tatB and ⌬tatC strain results (Fig. 1, lanes 7 and 8) were comparable to the results obtained for the total tat deletion strain (Fig. 1, lane 2). For both ⌬tatB and ⌬tatC strains, the majority of DmsD (88%) was located in the cytoplasm, while only 12% was associated with the inner membrane (Fig. 2). In all strains, a buffer wash did not dissociate the remaining DmsD from the membrane (Fig. 1, washed membranes).
Location of DmsD in a dmsABC Mutant-To assess whether or not the targeting of DmsD to the inner membrane was dependent upon the presence of DmsABC, a ⌬dmsABC mutant was studied. In this mutant, DmsD was located exclusively in  Table  I for details). Lanes 1-5 are MC4100 based while lanes 6 -9 are TG1-based E. coli strains. 50 g of total protein was loaded into each lane. Consequently, band intensities are not indicative of the relative percentage of DmsD in each fraction (see Fig. 3). DmsD was analyzed by 12% SDS-PAGE and Western blotted using a T 7 -HRP conjugate antibody against the DmsD N-terminal T 7 epitope tag. the membrane fraction and remained on the membrane even after an extensive buffer wash (Fig. 3, lane 2).
Effect of Aerobic Conditions on the Location of DmsD-To investigate the effect of an aerobic environment on the targeting of DmsD, WT and ⌬tatABC E. coli strains transformed with pTDMS28 were grown under aerobic conditions. In contrast to the results obtained for WT E. coli in anaerobic environments, DmsD was primarily located in the cytoplasm (Fig. 4). Likewise, DmsD was also found in the cytoplasmic fraction of a tatABC-deficient strain. The induction of pTDMS28 with IPTG had no apparent effect on the location of DmsD within the cell, with only trace amounts found associated with the membrane.
DmsD Does Not Interact with E. coli Lipid Vesicles-To determine whether the interaction of DmsD with the membrane was protein or lipid mediated, purified DmsD was incubated with E. coli liposomes, then studied using SEC. The void volume (V o ) of the column was 3.4 ml, determined using blue dextran as a marker (data not shown). Liposomes eluted at V o (Fig. 5A, solid line), while a purified and filtered sample of DmsD eluted at 6.6 ml (Fig. 5A, dashed line). There was no evidence of interaction between DmsD and E. coli lipid upon the incubation of DmsD with a 10-fold excess of liposomes. This was indicated by two separate elution peaks at 3.4 ml and 6.6 ml (Fig. 5B) corresponding to E. coli lipid vesicles and purified DmsD. Western blot analysis of the trichloroacetic acid-precip-itated fractions confirmed that DmsD was not present in the liposome elution peak of 3.4 ml (data not shown). The results were similar for all ratios of DmsD to liposomes (data not shown).
Nature of the Interaction of DmsD with the E. coli Inner Membrane-To elucidate what type of interaction is occurring between DmsD and the E. coli inner membrane, buffer-washed membranes derived from the WT (TG1)/pTDMS28 E. coli strain were resuspended and washed vigorously with various solutions. DmsD remained on the membrane in the presence of all eight solutions used (Fig. 6, membranes). Only in the presence of 4 M urea was any significant amount of DmsD located in   2, 4, 6, and  8). Others were not induced; DmsD was expressed through the leakiness of the T 7 promoter (lanes 1, 3, 5, and 7). DmsD was analyzed by 12% SDS-PAGE and Western blotted using a T 7 -HRP conjugate antibody against the DmsD N-terminal T 7 epitope tag. the soluble fraction corresponding to the washed membranes (Fig. 6, soluble fraction, lane 5). The results were similar using an MC4100/pTDMS28 strain (data not shown).
To investigate the strength with which DmsD was associated with the membrane, WT (TG1)/pTDMS28 E. coli membranes were solubilized with increasing concentrations of urea. DmsD remained tightly associated with the membrane at low concentrations of urea (0.5-3.0 M) (Fig. 7, membranes, lanes 1-4) with only trace amounts present in the corresponding soluble fractions (Fig. 7, soluble fraction, lanes 1-4). Above 3 M urea, increasing the concentration of urea resulted in an increasing amount of DmsD located in the soluble fractions (Fig. 7, lanes  5-8). However, only in the presence of 8 M urea was the vast majority of DmsD dissociated from the membrane (Fig. 7, lane  8). An MC4100/pTDMS28 strain gave similar results (data not shown).

DISCUSSION
In this study various E. coli fractions were analyzed for the presence of DmsD, with the goal of determining the subcellular location of DmsD within E. coli. This was achieved by Western blot analysis using an antibody against the T 7 epitope tag. Under anaerobic conditions DmsD is found on the E. coli inner membrane of both WT and dmsABC mutant strains. In contrast, the majority of DmsD was found in the cytoplasmic fraction of a total tat deletion mutant under the same growth conditions. This interaction is strongly influenced by the presence or absence of oxygen in the growth environment. Our next goal was to elucidate the nature of the interaction of DmsD with the E. coli inner membrane under anaerobic conditions. SEC data and vigorous membrane washing studies suggest that DmsD is strongly interacting with an integral membrane protein. Further investigation revealed that the interaction of DmsD with the E. coli cytoplasmic membrane is dependent upon the TatB and TatC subunits.
The results presented in this study illustrate that DmsD is associated with the E. coli inner membrane under anaerobic conditions. Strains lacking the TatA, TatD, or TatE subunits give rise to the same result as WT strains, where DmsD is associated with the inner membrane. TatD is not required for protein translocation using the Tat pathway (9), thus it is not surprising that DmsD behaves similarly in both ⌬tatD and WT strains. Likewise, the targeting of DmsD is not affected by the absence of the TatA or TatE subunits. The identical results obtained with the ⌬tatA, ⌬tatE and ⌬tatAE strains are in accordance with genetic experiments, which show that the TatA and TatE homologues have overlapping functions in the Tat pathway (4,5). DmsD was also associated with the E. coli inner membrane in a ⌬dmsABC mutant strain. This suggests that the targeting of DmsD to the E. coli inner membrane is not driven by its association with the DmsA preprotein.
The absence of the Tat complex has a dramatic effect on the localization of DmsD within E. coli. Although a small fraction of DmsD is still targeted to the inner membrane, the absence of the translocase appears to greatly hinder the interaction of DmsD with the membrane. This was evident in the ⌬tatAB-CDE, ⌬tatB, and ⌬tatC strains, in which DmsD was found primarily in the cytoplasmic fractions. These results suggest that DmsD interacts with the TatB and TatC subunits of the translocase complex. A mutation in either subunit appears to prevent the interaction of DmsD with the Tat complex. However, the absence of both subunits does not seem to result in a statistically greater percentage of DmsD located in the cytoplasmic fraction. This suggests that both subunits are equally important for the interaction of DmsD with the Tat complex.
These results are in accordance with previous work, which has demonstrated that the TatB and TatC subunits act in concert and form a functional and structural unit (10,14,17). Recent studies have also shown that a TatBC complex containing some TatA protein is capable of binding a Tat signal peptide, while a complex composed predominantly of TatA with small amounts of TatB was incapable of this interaction (15). Thus, the interaction of DmsD, a twin-arginine leader-binding protein, with the TatB and TatC subunits is consistent with current literature which suggests that the TatBC proteins form a receptor complex capable of recognizing the twin-arginine signal sequence of substrate proteins (18,21).
The liposome SEC and membrane washing results provide additional support for the interaction of DmsD with a protein component of the E. coli inner membrane. In a related study, Millman et al. (26) generated phopholipid liposomes and used SEC to investigate how FtsY, a membrane-associated signal recognition particle (SRP) receptor, binds to the inner membrane of E. coli. SRP is involved in the co-translational targeting of nascent polypeptide chains to the E. coli inner membrane through an interaction with FtsY (27,28). To investigate whether or not DmsD interacts with the E. coli inner membrane via lipid interactions, Millman's methods were followed. The SEC data, verified by Western blot analysis against the T 7 -epitope tag of DmsD, demonstrate that DmsD is not interacting with the lipid component of the membrane, suggesting that the interaction is protein mediated.
Vigorous washing of the anaerobic membranes with various solutions reveal that a chaotropic agent, such as urea, is needed to disrupt the interaction of DmsD with the membrane. This suggests that DmsD is interacting tightly with an integral  1-8, respectively). DmsD remained associated with the membrane following all 8 buffer washes. Only in the presence of 4 M urea was a significant amount of DmsD dissociated from the E. coli membrane (lane 5). DmsD was analyzed by 12% SDS-PAGE and detected by Western blot using a T 7 -HRP conjugate antibody against the DmsD N-terminal T 7 epitope tag.  1-8, respectively). Above 3.0 M urea, DmsD was increasingly dissociated from the E. coli inner membrane, as shown by the soluble fraction bands. Proteins were resolved by 12% SDS-PAGE and DmsD was detected by Western blot using a T 7 -HRP conjugate antibody against the DmsD N-terminal T 7 epitope tag. membrane protein or proteins. Folding studies 2 have shown that DmsD starts to denature in the presence of ϳ4 M urea, which is consistent with the concentration above which a significant amount of DmsD was dissociated from the E. coli inner membrane. Thus, it appears that DmsD denaturation is required for the disruption of its interaction with the membrane.
Another important observation from this study is that the physiological conditions have a significant effect on the localization of DmsD in WT E. coli, as evidenced by the distinct results obtained for WT E. coli grown aerobically versus anaerobically. DmsD binds the twin-arginine signal sequences of dimethylsulfoxide and TMAO reductase (22), two enzymes that are physiologically active under anaerobic conditions (29 -31). Therefore, it follows that the interaction of DmsD with the membrane would be dependent upon an anaerobic environment. Although DmsD is associated with the membrane in anaerobic environments, hydropathy analysis of the sequence suggests that DmsD is not a membrane protein. 3 The evidence presented in this study demonstrates that under anaerobic conditions, DmsD is targeted to the membranebound Tat complex in E. coli and is interacting tightly with the TatBC signal sequence receptor complex. It was also shown that this interaction is independent of the DmsA preprotein and is dependent upon the physiological conditions of the cell. Thus, this work further supports the prediction that DmsD might function as a targeting chaperone (22), assisting in the interaction of DmsA with the Tat translocase. An interesting challenge for the future will be to determine the mechanism by which this occurs.