JBC Biosymposia, Inc.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Originally published In Press as doi:10.1074/jbc.M409192200 on October 12, 2004

J. Biol. Chem., Vol. 279, Issue 51, 53747-53754, December 17, 2004
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental Data
Right arrow All Versions of this Article:
279/51/53747    most recent
M409192200v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Henderson, N. S.
Right arrow Articles by Thanassi, D. G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Henderson, N. S.
Right arrow Articles by Thanassi, D. G.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

Topology of the Outer Membrane Usher PapC Determined by Site-directed Fluorescence Labeling*{boxs}

Nadine S. Henderson, Stephane Shu Kin So, Cheryl Martin{ddagger}, Ritwij Kulkarni, and David G. Thanassi§

From the Department of Molecular Genetics and Microbiology, Center for Infectious Diseases, Stony Brook University, Stony Brook, New York 11794-5120

Received for publication, August 11, 2004 , and in revised form, September 27, 2004.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
In contrast to typical membrane proteins that span the lipid bilayer via transmembrane {alpha}-helices, bacterial outer membrane proteins adopt a {beta}-barrel architecture composed of antiparallel transmembrane {beta}-strands. The topology of outer membrane proteins is difficult to predict accurately using computer algorithms, and topology mapping protocols commonly used for {alpha}-helical membrane proteins do not work for {beta}-barrel proteins. We present here the topology of the PapC usher, an outer membrane protein required for assembly and secretion of P pili by the chaperone/usher pathway in uropathogenic Escherichia coli. An initial attempt to map PapC topology by insertion of protease cleavage sites was largely unsuccessful due to lack of cleavage at most sites and the requirement to disrupt the outer membrane to identify periplasmic sites. We therefore adapted a site-directed fluorescence labeling technique to permit topology mapping of outer membrane proteins using small molecule probes in intact bacteria. Using this method, we demonstrated that PapC has the potential to encode up to 32 transmembrane {beta}-strands. Based on experimental evidence, we propose that the usher consists of an N-terminal {beta}-barrel domain comprised of 26 {beta}-strands and that a distinct C-terminal domain is not inserted into the membrane but is located instead within the lumen of the N-terminal {beta}-barrel similar to the plug domains encoded by the outer membrane iron-siderophore uptake proteins.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
All bacterial outer membrane (OM)1 proteins with known structures span the membrane via a series of transmembrane (TM) antiparallel {beta}-strands arranged to form a {beta}-barrel (1–4). Proteins found in the OM of mitochondria and chloroplasts also adopt similar {beta}-barrel folds (5). Only six to nine residues are required for a {beta}-strand to cross the central non-polar region of the lipid bilayer (4), and only half of these need to be hydrophobic since only alternating residues face into the bilayer. This makes prediction of TM {beta}-strands by sequence analysis difficult. Furthermore protocols typically used to map the topology of {alpha}-helical membrane proteins are not applicable to OM proteins. For example, alkaline phosphatase (PhoA) fusions are commonly used to identify periplasmic regions of inner membrane proteins (6). However, PhoA cannot be used to map OM proteins reliably as the large PhoA insertion folds in the periplasm and disrupts proper assembly of proteins in the OM (7). Smaller insertions are generally tolerated in surface and periplasmic loops of {beta}-barrel proteins, and introduction of epitopes and protease cleavage sites has been used to map OM proteins (7–11). However, the mapping of such insertions in whole bacteria may be critically affected by the membrane or protein structure, and mapping of periplasmic regions requires disruption of the OM with the potential for misleading results (7, 9–11). To overcome these difficulties, we adapted a site-directed fluorescence labeling protocol for topology mapping of OM proteins. Our method only requires insertion or substitution of single cysteine residues and permits analysis of OM proteins in their native state in intact bacteria.

We used this protocol to map the topology of the PapC usher protein, which is required for assembly and secretion of P pili in uropathogenic Escherichia coli (12, 13). P pili are composite structures built from multiple pilus subunits (14, 15). The PapG adhesin, located at the pilus tip, binds to Gal{alpha}(1–4)Gal moieties present in kidney glycolipids, and P pili are key virulence factors for the development of pyelonephritis (16, 17). P pili are assembled by the chaperone/usher secretion pathway, which is responsible for biogenesis of a superfamily of surface structures associated with virulence (18). Pilus subunits cross the inner membrane in an unfolded form via the Sec general secretory pathway (19, 20) and then must interact with the PapD chaperone in the periplasm. The chaperone acts by a mechanism termed donor strand complementation that couples subunit folding with the simultaneous capping of subunit interactive surfaces (21–23). Chaperone-subunit complexes must next target to the PapC usher in the OM for pilus biogenesis. The usher serves as a pilus assembly platform as well as a secretion channel (13, 24–26). Pilus assembly is thought to occur at the periplasmic face of the usher concomitant with secretion of the pilus fiber through the usher to the cell surface (18). Chaperone-subunit complexes initially target to an N-terminal region of the usher consisting of the first 124–139 residues of the mature N terminus (27, 28). The complexes subsequently form stable interactions with the usher C terminus, which is required for subunit assembly into pili and secretion to the cell surface (29, 30). Pilus assembly occurs by a mechanism termed donor strand exchange in which chaperone-subunit interactions are replaced by subunit-subunit interactions (21, 31, 32). Donor strand exchange allows subunits to undergo a topological transition to a lower energy state, presumably providing the driving force for pilus biogenesis at the usher.

Mature PapC contains 809 amino acids and has a molecular mass of 88.2 kDa. Computer modeling of the PapC topology predicted 24 TM {beta}-strands, and analysis of PapC by circular dichroism confirmed a largely {beta}-sheet secondary structure (30). The usher C terminus forms a distinct domain that is not required for proper folding of the {beta}-barrel in the OM (30). PapC assembles into an oligomeric complex and forms channels 2–3 nm in diameter (24), large enough to allow secretion of folded pilus subunits. Detailed knowledge of the arrangement of the usher in the lipid bilayer is essential for understanding the molecular mechanisms of pilus assembly and secretion across the bacterial OM. Based on our topology mapping results and experiments showing that the usher C terminus is tightly associated with the OM, we propose that the usher folds with an N-terminal {beta}-barrel domain comprised of 26 TM {beta}-strands and that the C terminus is located within the barrel channel in a fashion analogous to the plug domains found in OM iron-siderophore uptake proteins (33–36).


   EXPERIMENTAL PROCEDURES
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
RESULTS
 DISCUSSION
 REFERENCES
 
Strains and Plasmids—E. coli strain DH5{alpha} (37) was used for plasmid manipulations; strain AAEC185 (38), which lacks the chromosomal fim genes coding for type 1 pili, was used for hemagglutination assays (HAs); and strain SF100 (39), which lacks the OmpT protease, was used for analysis of the PapC constructs and for the topology mapping experiments. Strain XL-3 was used for analysis of the OxlT constructs (40). Strains were grown in LB broth containing appropriate antibiotics at 37 °C with aeration. Plasmids used in this study are listed in Supplemental Table I. PapC expression was induced at an A600 of 0.6 for 1 h with 0.1% L-arabinose, and the {Delta}papC pap operon from plasmid pMJ2 was induced with 0.05 mM isopropyl {beta}-D-thiogalactoside. Plasmids expressing the single cysteine OxlT variants G49C or D104C were induced at an A600 of 0.6 for 1 h with 1 mM isopropyl {beta}-D-thiogalactoside.

Plasmids designated pNH (Supplemental Table I), encoding single cysteine substitution or insertion mutants of PapC, were derived from plasmid pMJ3, which encodes WT PapC with a C-terminal hexahistidine tag (His tag), using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). Primers used for mutagenesis are listed in Supplemental Table I. All mutations were verified by sequencing. Random insertion of tobacco etch virus (TEV) protease sites into PapC was performed using the TnTAP transposon as described previously (41). Briefly pMM1 was transformed into DH5{alpha}/pMJ3, and a single Ampr, Tetr colony was selected for overnight growth and plasmid isolation. Plasmids were digested with NheI and transformed into DH5{alpha}, selecting for Ampr, Kanr, and Tets. The colonies were then screened for PhoA activity using 5-bromo-4-chloro-3-indolyl phosphate indicator plates. Blue colonies were selected and screened by PCR for insertion of TnTAP into papC using the primer UPSTRMPAPC (5'-GAAGCGCTGGATTACACCCTCAG-3'), which binds upstream of the papC open reading frame, and primer TNTAPPHOA (5'-GCAGTAATATCGCCCTGAGCAGC-3'), which binds to phoA within the TnTAP transposon. Plasmids containing TnTAP insertions in papC were isolated, and the insertion junctions were determined by sequencing using the TNTAPPHOA primer. Finally the phoA and neo reporter genes of TnTAP were removed by digesting the plasmids with NotI and religating to produce the plasmids designated pHG (Supplemental Table I), which code for PapC proteins containing in-frame insertions of the 24-residue sequence LTLIHKFENLYFQSAAAILVYKSQ (the TEV protease recognition site is underlined).

HA—Analysis of pilus biogenesis by HA was performed as described previously (30) using AAEC185/pMJ2 ({Delta}papC pap operon) complemented with pMON6235{Delta}cat (vector), pMJ3 (WT PapC), or one of the PapC mutants.

Heat-modifiable Mobility Shift Assay—Expression and folding of the various PapC mutants in the OM was analyzed using a heat-modifiable mobility shift assay as described previously (27).

Digestion of TnTAP Insertion Mutants of PapC with TEV Protease— PapC mutants containing TnTAP insertions were tested for susceptibility to cleavage by TEV protease as described previously (41) using whole bacteria or isolated OM. OM was isolated as described previously (30). Samples were subjected to SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Osmonic Inc., Gloucester, MA). Cleavage of PapC was monitored by blotting with anti-PapC or anti-His tag (Covance, Richmond, CA) antibodies followed by an alkaline phosphatase-conjugated secondary antibody (Sigma). The blots were developed with 5-bromo-4-chloro-3-indolyl phosphate-nitro blue tetrazolium substrate (KPL, Gaithersburg, MD).

Site-directed Fluorescence Labeling—This protocol was adapted from the method used by Ye and coworkers (40) to map the inner membrane protein OxlT. A 20-ml culture of SF100 expressing WT PapC or one of the PapC mutants was harvested, resuspended in 2.6 ml of buffer A (100 mM potassium sulfate, 50 mM potassium phosphate, pH 8.0), and divided into two equal aliquots. Freshly prepared 4-acetamido-4'-male-imidylstilbene-2–2'-disulfonic acid (AMS, Molecular Probes, Eugene, OR) was added to one aliquot to 100 µM final concentration, and the aliquots were incubated at room temperature (RT, 25 °C) for 7 min. Both aliquots were washed three times with buffer A, and then Oregon green 488 maleimide (OGM, Molecular Probes) was added to 40 µM final concentration to both aliquots. The samples were incubated in the dark at RT for 15 min, the reaction was quenched by addition of 6 mM {beta}-mercaptoethanol, and the aliquots were washed three times with buffer A. The bacteria were then resuspended in 1 ml of 20 mM Tris-HCl (pH 8.0) containing protease inhibitors for isolation of OM as described previously (27). The OM pellet was resuspended in 20 mM Tris (pH 8.0), 0.3 M NaCl, 0.5% dodecylmaltoside (DDM, Anatrace, Maumee, OH) and solubilized overnight by rocking at 4 °C. After spinning out insoluble material (16,100 x g, 30 min, 4 °C), imidazole was added to 20 mM final concentration, and the mixture was rocked for 30 min at RT with 50 µl of 50% nickel-nitrilotriacetic acid resin (Qiagen, Chatsworth, CA) to bind the His-tagged PapC. The nickel-nitrilotriacetic acid resin was then washed three times with 20 mM Tris (pH 8.0), 0.3 M NaCl, 20 mM imidazole, 0.1% DDM. The supernatant was removed, 25 µl of 2x SDS-PAGE sample buffer was added to the resin, the sample was incubated at 95 °C for 10 min, and 18 µl was loaded on a 10% gel. After electrophoresis, the fluorescence profile was obtained by scanning with a STORM fluorescence imaging system (Amersham Biosciences) using the blue fluorescent chemifluorescence mode (excitation wavelength of 450 ± 30 nm). The same gel was then stained with Coomassie Brilliant Blue R-250, scanned using a GS-710 densitometer (Bio-Rad), and analyzed with Quantity One software (Bio-Rad). The fluorescence value for each PapC band was normalized to its protein level determined from Coomassie staining. Then the ratio of the normalized OGM signal without AMS pretreatment to the normalized OGM signal with AMS pretreatment was calculated for each PapC to give the -fold reduction in OGM labeling due to pretreatment with AMS. Control PapC constructs containing cysteines at known surface and periplasmic locations were included in all assays; this was important to monitor the quality of the AMS reagent. All PapC mutants were analyzed at least twice.

As an alternative to AMS, Alexa Fluor 594 C5 maleimide (Molecular Probes) was used in the above protocol at a final concentration of 100 µM and an incubation time of 15 min. The specificity of AMS as a blocking agent for surface-exposed residues was tested using the OxlT mutants G49C and D104C, which contain periplasmic cysteines (40). The OxlT mutants were assayed using AMS as a blocking agent according to our protocol described above or using methanethiosulfonate ethyltrimethylammonium (Biotium Inc., Hayward, CA) as a blocking agent (2 mM final concentration for 10 min) as described previously (40).

PapC mutants that did not label with OGM in intact bacteria were subsequently analyzed by labeling with OGM after solubilization with DDM or denaturation with SDS (42). For labeling of solubilized PapC, OM was extracted with DDM as described above, but before the addition of imidazole samples were incubated for 15 min at RT with 40 µM OGM. The reaction was quenched with 6 mM {beta}-mercaptoethanol, and protein was purified and analyzed as described above. For labeling of denatured PapC, purified protein was isolated as described above but boiled in sample buffer without {beta}-mercaptoethanol. OGM was added to 1 mM, and the samples were incubated for 15 min at RT. The samples were then subjected to SDS-PAGE and analyzed as described above.

Association of the PapC C Terminus with the OM—OM was isolated from SF100/pMJ3 as described previously (30). A final concentration of 50 µg/ml L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Sigma) was added to 0.2 mg OM (as determined by the bicinchoninic acid protein assay, Pierce). After incubation on ice for 2 min, digestion was terminated by addition of 1 ml of 20 mM HEPES (pH 7.5), 0.3 M NaCl, 1 mM phenylmethylsulfonyl fluoride. The OM was pelleted by centrifugation (80,000 x g, 1 h, 4 °C) and resuspended in 1 ml of 20 mM HEPES (pH 7.5), 0.3 M NaCl, 1 mM phenylmethylsulfonyl fluoride. Samples were treated on ice for 1 h with NaOH (0.01, 0.05, or 0.1 M), urea (4 or 6 M), 1 M NaCl, or 0.2 M Na2CO3 or were sonicated (4 min, 30 s on, and 30 s off) in a water bath sonicator (Misonix). The membrane and soluble fractions were then separated by centrifugation (80,000 x g,1h, 4 °C). The supernatant fractions were precipitated by addition of trichloroacetic acid. Both the membrane and trichloroacetic acid pellets were treated with SDS-PAGE sample buffer at 95 °C for 10 min, resolved by SDS-PAGE, and blotted with an anti-His tag antibody as described above.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
DISCUSSION
 REFERENCES
 
Identification of Two Surface Loops of PapC by TEV Protease Cleavage—As a first approach to map the topology of the usher, we generated random insertions of the TEV protease recognition sequence in PapC using the TnTAP transposon (41). TnTAP was designed specifically for analysis of OM proteins and results in a final in-frame insertion of a 24-residue sequence carrying the TEV protease cleavage site. We isolated 29 unique TnTAP insertions in PapC that were stably expressed in the OM (Fig. 1). Approximately one-third of the insertions were isolated more than once with a pronounced hotspot for insertion following residue Val-76 (V76TEVins) in a large periplasmic loop of PapC (Fig. 1). This loop also contained the highest concentration of TnTAP insertions. In other regions of PapC, notably the C terminus, we were unable to isolate insertions. The insertion sites in PapC define regions flexible enough to accommodate the 24-residue TnTAP sequence. The non-random distribution of insertions was thus as least partly attributable to the screening procedure, which selected for stable, in-frame insertions in PapC (41).



View larger version (44K):
[in this window]
[in a new window]
  		FIG. 1.
PapC topology model. Site-directed fluorescence labeling results are color coded green for cysteines that labeled as surface, red for cysteines that labeled as periplasm, yellow for cysteines that required solubilization with DDM to label, and blue for cysteines that required denaturation with SDS to label. Triangles indicate cysteine insertions, and squares denote cysteine substitutions. Symbols with black borders indicate an HA defect (HA titer of 0–8 compared with a WT PapC titer of 128), and hollow symbols indicate a defect in the heat-modifiable mobility assay. Cysteines that may have mispaired with the native C-terminal cysteines are marked with an asterisk (*). TnTAP insertions are marked by black or white triangles. White triangles indicate an HA defect (HA titer of 0–8 compared with a WT PapC titer of 64). Triangles marked with an S denote TnTAP insertions that cleaved with TEV protease in whole bacteria. The gray bar indicates the predicted location of the OM bilayer. The dark gray region marks the 26 TM {beta}-strands proposed to form the PapC {beta}-barrel, and the light gray region marks the PapC C-terminal domain proposed to insert within the {beta}-barrel. Alternating hydrophobic residues within the TM regions are presented in outline.

 
To confirm that the PapC mutants containing TnTAP insertions were folded properly in the OM, we took advantage of the fact that PapC, in common with other OM {beta}-barrel proteins, is resistant to denaturation by SDS unless incubated at high temperatures and thus exhibits a characteristic heat-modifiable mobility on SDS-PAGE (24). Each of the TnTAP insertion mutants exhibited heat-modifiable mobility, indicating proper folding of the PapC {beta}-barrel (data not shown). We next assessed functionality of the TnTAP insertion mutants by testing their ability to complement a {Delta}papC pap operon for assembly of adhesive pili by HA. The majority of the insertion mutants (22 of 29, Fig. 1) were defective for pilus biogenesis, having HA titers of 0–8 compared with a WT HA titer of 64 (Supplemental Table II). The functional defects found with the TnTAP insertion mutants are not surprising given the large size of the inserted sequence.

Although most of the TnTAP insertion mutants were not functional for pilus biogenesis, we proceeded with topology mapping studies as all were able to adopt a correct {beta}-barrel fold in the OM. TEV protease was added to whole bacteria or to isolated OM, and cleavage of the usher was monitored by immunoblot analysis using anti-His tag or anti-PapC antibody. Three of the TnTAP insertion mutants (A124TEVins, R138TEVins, and S194TEVins) were clearly cleaved in whole bacteria, indicating a surface location (Fig. 2). These insertions occurred in two N-terminal loops predicted to be surface-exposed by a prior modeling study (30). These two loops are also the largest surface loops present in the {beta}-barrel region of PapC (Fig. 1, see below). Three additional TnTAP insertion mutants (G30TEVins, A69TEVins, and D74TEVins) were cleaved by TEV protease in isolated OM but not in whole bacteria, suggesting a periplasmic location (data not shown). However, we could not discriminate against the possibility that these insertions instead were located at the surface but only became exposed upon isolation of the OM. This is a major limitation of topology mapping protocols that require membrane disruption to reach internal sites. None of the remaining TnTAP insertions in PapC was cleaved by TEV protease in either whole bacteria or OM (data not shown). This was surprising given that the insertion mutants were stably expressed, and the large size of the inserted sequence should favor exposure of the cleavage site. The TEV protease may not have been able to recognize the cleavage sites due to local conformational effects, or the sites may have been sterically blocked by protein or membrane structure (7, 9–11).



View larger version (18K):
[in this window]
[in a new window]
  		FIG. 2.
Cleavage of PapC TnTAP insertion mutants in whole bacteria. Whole bacteria expressing WT PapC or the TnTAP insertion mutant A124TEVins, R138TEVins, or S194TEVins were incubated in the absence (-) or presence (+) of TEV protease. Bacteria were subjected to SDS-PAGE and blotted with an anti-His tag antibody.

 
Development of a Site-directed Fluorescence Labeling Protocol for OM Proteins—Given the problems encountered with topology mapping using protease cleavage, we sought to develop a protocol that would allow accurate determination of periplasmic regions and would be less susceptible to steric interference. Small molecule probes that differ in their ability to cross the OM would meet these criteria. Sulfhydryl-reactive molecules have been used extensively in combination with cysteine mutagenesis to study the topology of {alpha}-helical membrane proteins (43). In particular, Ye and coworkers (40) used site-directed fluorescence labeling to map the topology of the inner membrane oxalate/formate transporter OxlT in intact bacteria. They created single cysteine mutants of OxlT and measured reactivity with the thiol-directed fluorescent probe OGM. OGM labeled periplasmic cysteines of OxlT in intact bacteria, indicating that OGM freely crossed the OM presumably through the porins. They also demonstrated that pretreatment of intact bacteria with a non-fluorescent, thiol-specific probe blocked OGM labeling of periplasmic cysteines (40). Based on these results, we reasoned that OGM should label both periplasmic and surface-exposed cysteines of an OM protein, providing a control for exposure of the residue to the medium and ability of the residue to be labeled. To differentiate surface-exposed from periplasmic cysteines, we needed a blocking agent that would not cross the OM. The OmpF porin allows diffusion of small molecules up to ~500 Da in size with a preference for cationic molecules (44). The thiol-reactive molecule AMS, which has been used as a blocking agent for topology mapping studies (43, 45), has a molecular mass of 536 Da and a dianionic charge; therefore, it should only slowly cross the OM through the porins. AMS does not fluoresce at the excitation wavelength used to detect OGM and could therefore function as a blocking agent.

Studies using sulfhydryl-reactive probes typically are performed on proteins naturally lacking or engineered to lack cysteines. Alternatively cysteines may be present in the native protein if they are not available to interact with the probe and if they do not interfere with analysis of the introduced cysteines (43). WT PapC has conserved cysteine pairs at its N and C termini (Fig. 1). These cysteine pairs form intramolecular disulfide bonds that stabilize their respective Nand C-terminal periplasmic regions but are not required for proper folding of the PapC {beta}-barrel (27, 30). WT PapC did not react with OGM, confirming that the cysteines were paired. A cysteineless PapC was constructed in which all four cysteines were substituted with alanine, but this mutant was not functional for pilus biogenesis and was too unstable to be used for topology mapping.2 We also tested PapC mutants in which only one of the cysteine pairs was replaced with alanines. Although the N-terminal cysteine pair was required for PapC function (27), the C-terminal cysteine pair was not3; however, both constructs were subject to increased degradation that limited their usefulness for the topology mapping studies. Therefore, we performed our mapping experiments with WT PapC containing the four native cysteines, reasoning that mispairing of the native cysteines with an introduced cysteine would cause obvious structural and/or functional defects in PapC. Importantly the majority of the PapC cysteine mutants we constructed were functional for pilus biogenesis and able to fold correctly in the OM (see below).

We tested our experimental design using the PapC cysteine substitution mutant S137C and the alanine substitution mutant C805A, which contain surface and periplasmic cysteines, respectively. S137C is located in the surface loop identified by cleavage of R138TEVins (Fig. 1). Cys-805 and Cys-787 form the conserved C-terminal cysteine pair of PapC. The C805A mutation leaves Cys-787 unpaired and available for labeling. Previous studies have located the conserved cysteine pairs of the ushers to the periplasm (30, 46). Both the S137C and C805A PapC mutants folded correctly in the OM as determined by the heat-modifiable mobility assay and were functional for pilus biogenesis as measured by HA (Supplemental Table III). Incubating intact bacteria expressing PapC S137C or C805A with 40 µM OGM for 15 min labeled both cysteines equally well, confirming that OGM readily crosses the OM. By varying the conditions, we found that pretreatment of bacteria with 100 µM AMS for 7 min resulted in a 12-fold reduction in subsequent OGM labeling of the surface-exposed S137C. In contrast, OGM labeling of the periplasmic Cys-787 was reduced by only 2.4-fold, indicating slow permeation of AMS across the OM as predicted. We also tested the thiol-reactive probe Alexa Fluor 594 C5 maleimide as an alternative blocking agent to AMS. Alexa Fluor 594 has a molecular mass of 909 Da, is negatively charged, and does not fluoresce at the excitation wavelength used to detect OGM. Results with the Alexa Fluor were equivalent to those generated with AMS (7.3-fold reduction in OGM labeling of S137C and a 2.2-fold reduction for C805A). The fact that we did not get better discrimination with the larger Alexa Fluor may have been due to the longer incubation time we used with this reagent (see "Experimental Procedures"). Also the penetration of larger molecules across the OM is not strictly correlated with size but depends on their structural flexibility as well as their charge state (44). We chose to use AMS for the subsequent mapping experiments due to its lower cost. To confirm the specificity of AMS as a blocking agent for surface residues, we tested our protocol on the OxlT mutants G49C and D104C that contain single cysteines located in the periplasm (40). Pretreatment with AMS reduced OGM labeling of the OxlT constructs by only 2.5- and 1.9-fold, respectively. In contrast, pretreatment with methanethiosulfonate ethyltrimethylammonium, a blocking agent that readily crosses the OM (40), reduced OGM labeling by 24- and 27-fold, respectively. This confirms that AMS preferentially blocks surface residues and can be used in combination with OGM to map the topology of OM proteins.

Site-directed Fluorescence Labeling of PapC—Having established and verified our protocol, we constructed an extensive set of single cysteine insertions and substitutions throughout PapC. We initially chose sites predicted to be in surface or periplasmic loops based on previous modeling (30). We then constructed additional mutations as needed to clarify and revise the PapC topology. The native N- and C-terminal cysteine pairs of PapC were analyzed in the fluorescence assay using mutants in which one of the cysteines was substituted with alanine as described above. In total, we analyzed 115 PapC cysteine mutants as summarized in Fig. 1 and Supplemental Table III. Each of the PapC single cysteines mutants was also analyzed for function by HA and for effects on protein folding by the heat-modifiable mobility assay. Cysteines are generally well tolerated in proteins, and most of the mutants were expected to be functional. Indeed only 18 of the 115 cysteine mutants were defective for pilus biogenesis with HA titers from 0–8 (Fig. 1, symbols with black borders, and Supplemental Table III). Furthermore only four cysteine mutants caused defects in heat-modifiable mobility, indicating an effect on stability of the {beta}-barrel (Fig. 1, open symbols, and Supplemental Table III). These PapC mutants most likely could adopt a correct overall fold as they were functional for pilus biogenesis by HA.

Cysteines were judged to be located in the periplasm if pretreatment with AMS reduced OGM labeling by 1–4-fold (Fig. 1, marked red). Cysteines were judged to be located at the cell surface if they showed a 6-fold or higher reduction in OGM labeling (Fig. 1, marked green). Note that the distinction between surface and periplasmic cysteines was generally very clear with the majority (23 of 35) of surface-located cysteines fully blocked (no detectable OGM signal) by pretreatment with AMS; only seven surface-located cysteines showed less than a 10-fold reduction in labeling (Supplemental Table III). Representative labeling results are presented in Fig. 3. A subset of the PapC cysteine mutants did not label with OGM in intact bacteria. One class of these mutants could be labeled following extraction of PapC from the OM with DDM (Figs. 1, marked yellow, and 3B). These cysteines mapped near the membrane interface or in periplasmic loops and were likely located in regions that were protected from labeling by the membrane or were in structured regions of PapC that blocked access of OGM. Solubilization of PapC from the OM either exposed residues that were blocked by the membrane or caused an opening of the usher structure to allow OGM access. Interestingly no cysteines in this class were found in surface loops. The larger periplasmic loops of PapC must be structured in the native state of the protein so as to block access of OGM to the cysteines. A second class of PapC cysteine mutants that did not label in intact bacteria could only be labeled following denaturation of the protein with SDS (Figs. 1, marked blue, and 3C). These cysteines presumably were located in TM regions of PapC that remained protected by detergent even after solubilization from the OM.



View larger version (52K):
[in this window]
[in a new window]
  		FIG. 3.
Site-directed fluorescence labeling of PapC. A, whole bacteria expressing the indicated PapC cysteine mutant were pretreated without (-) or with (+) AMS and then labeled with OGM. PapC was purified and subjected to SDS-PAGE. The upper panel shows the OGM fluorescence profile for the purified PapC, and the lower panel shows the corresponding Coomassie Blue profile. AMS pretreatment of M252C reduced OGM labeling by 21-fold, indicating a surface location. AMS pretreatment of G347Cins reduced OGM labeling by 2.0-fold, indicating a periplasmic location. V356Cins and V520Cins did not label in whole bacteria. B and C, V356Cins and V520Cins, which did not label in whole bacteria, were labeled with OGM after solubilization (s) from the OM with DDM or denaturation (d) with SDS. The upper and lower panels show the fluorescence and Coomassie profiles as in A. V356Cins labeled after solubilization or denaturation, whereas V520Cins labeled only after denaturation.

 
Anomalous labeling results were obtained for a small number of cysteine mutants. The cysteine insertion following residue Gly-397 (G397Cins) and six cysteine substitution mutants (S248C, T484C, S544C, S553C, T570C, and S581C) labeled in the fluorescence assay as though they were periplasmic (Supplemental Table III). However, these residues were assigned to the surface (Fig. 1) based on consideration of mapping data from cysteines in the surrounding region as well as additional criteria as described below to generate the best possible overall fit with the topology model. Each of these cysteines that mis-labeled as periplasmic were located at or close to the OM surface (Fig. 1). AMS may have difficulty accessing these residues possibly due to electrostatic or steric effects at the membrane surface that do not affect OGM to the same extent. Thus, AMS would not block efficiently, thereby erroneously reporting a periplasmic location. Anomalous results were also obtained for four cysteine mutants (S345C, Y531Cins, G533Cins, and N560Cins), which labeled in the fluorescence assay as though they were surface-exposed but were assigned to the periplasm based on data from surrounding mutants (Fig. 1). PapC is capable of forming channels 2–3 nm in diameter (24). OGM and AMS would be expected to easily pass through a channel of this size, which, if this occurred, would have prevented discrimination of periplasmic from surface residues. Previous analysis demonstrated that the PapC channels are likely in a predominantly closed conformation (24). Three of the four residues that labeled as surface (Y531Cins, G533Cins, and N560Cins) are located near each other in the PapC topology map and also close to residue S507C, which labels as surface but maps to the center of a TM strand (Fig. 1). These residues may define a probe-accessible channel present in the closed state of PapC.

Two additional cysteines (R688Cins and K727Cins) labeled in the fluorescence assay as though they were surface-exposed but were assigned to the periplasm (Fig. 1). These two cysteine mutants, as well as G758Cins, were not functional for pilus biogenesis with HA titers of zero. Note that these defective cysteine mutants alternated in the C terminus with functional mutants containing surface-exposed cysteines (D666Cins, S707C, and S741C). These results suggest that the R688Cins, K727Cins, and G758Cins cysteines (Fig. 1, marked with an asterisk) were located in the periplasm and mis-disulfide bonded with the native C-terminal cysteine pair, causing the HA defects and giving a false localization signal. In contrast, the D666Cins, S707C, and S741C cysteines were located at surface sites where they could not interact with the native cysteines and thus were fully functional and properly labeled in the fluorescence assay. Note that the mispaired cysteines did not interfere with proper folding of the PapC {beta}-barrel as determined by the heat-modifiable mobility assay. This is not unexpected as it was previously shown that the C terminus does not form part of the core {beta}-barrel structure of PapC (30).

Topology Model for PapC—Taken together, the mapping results described above are consistent with PapC having 32 TM {beta}-strands as presented in Fig. 1. Several criteria in addition to the topology mapping results were used to position the TM strands with respect to the OM bilayer. An alignment of 98 PapC homologs, generated by the Jnet program (47), was used to identify conserved stretches of alternating hydrophobic residues. Such conserved regions are characteristic of TM {beta}-strands (4). A minimum of six residue or three consecutive alternating hydrophobic residues are required for a {beta}-strand to cross the central non-polar region of the bilayer, although typically seven to nine residues are used (4). The TM {beta}-strands of PapC were positioned to meet this minimum requirement with most strands having four or more consecutive alternating hydrophobic residues (Fig. 1). OM {beta}-barrel proteins commonly have belts of aromatic residues located at the positions of the phospholipid head groups (1, 3). This characteristic was also used as a guide for positioning the TM {beta}-strands of PapC (Fig. 1). In support of our model, TnTAP insertions were not isolated in central regions of the TM strands where they would be expected to disrupt the {beta}-barrel structure. Insertion of single cysteines at the beginning or middle of a TM {beta}-strand also might disrupt protein structure or function as this would change the alternating hydrophobic register of the strand. Three cysteine insertions (G397Cins, Q518Cins, and V520Cins) were created at such locations (Fig. 1). The Q518Cins and V520Cins mutations had no effect on PapC structure or function, but the altered register created by these insertions also encodes hydrophobic amino acids (Fig. 1). In the case of G397Cins where the altered register is not hydrophobic, this mutant had a HA defect.

The last 169 residues of PapC (following Ala-640) are not required for correct folding of the usher into an SDS-resistant structure in the OM (30). Therefore, this region of the C terminus cannot form part of the core {beta}-barrel structure of PapC. Since all OM proteins with known structures have an even number of TM {beta}-strands (3), we propose that the PapC {beta}-barrel consists of 26 TM strands and is comprised of residues from the N terminus through Asn-587 (Fig. 1). The region of PapC from Ser-588 to the end would then form a separate domain. Interestingly a new computer algorithm for structural analysis of OM {beta}-barrel proteins (48) identified TM {beta}-strands in PapC from the N terminus through residue Leu-586 and predicted that the remainder of the usher resides in the periplasm. In contrast, a previous computer-generated topology model of PapC identified TM strands throughout the C terminus (30), and our current mapping data identified both surface and periplasmic residues in this region, consistent with the presence of six TM {beta}-strands (Fig. 1). Three possible models could account for these results. (i) The C terminus does contain TM {beta}-strands, and these form a separate integral membrane region; (ii) the C terminus does not contain transmembrane regions but instead is located entirely in the periplasm, and the regions that mapped as surface are located near or below the PapC channel; or (iii) the C terminus does not contain TM {beta}-strands but instead inserts within the N-terminal 26-stranded {beta}-barrel, similar to the plug domain found in FhuA and related iron-siderophore uptake proteins (33–36).

The PapC C Terminus Is Tightly Associated with the OM—To begin to differentiate among the models proposed above, we analyzed the association of the PapC C terminus with the OM. Limited trypsin digestion of isolated OM cleaves PapC between Arg-652 and Leu-653, producing two fragments: an ~70-kDa N-terminal fragment and an ~20-kDa C-terminal fragment (30). We reasoned that if the C terminus resided strictly in the periplasm, the 20-kDa fragment should move to the supernatant fraction following cleavage. If it was peripherally associated with the OM or the PapC {beta}-barrel, we should be able to extract it from the membrane under relatively mild conditions. In FhuA, the plug domain makes extensive contacts with the interior of the {beta}-barrel, including nine salt bridges and more than 60 hydrogen bonds, and would not be easily removed (34, 35). Likewise if the PapC C terminus was located within the {beta}-barrel, it should be extracted only under harsh conditions. Harsh conditions would also be required to extract the C terminus from an integral membrane location.

Without treatment, full-length PapC and the 20-kDa C-terminal fragment were found exclusively in the membrane fraction (Fig. 4). Thus, the C terminus does not exist in the periplasm free of the rest of PapC or the OM. We next subjected protease-treated OM to extraction with various agents. Under relatively mild conditions (bath sonication, 1 M NaCl, 0.2 M Na2CO3, or 0.01 M NaOH), full-length PapC and the 20-kDa fragment were located solely in the membrane fraction (Fig. 4 and data not shown), suggesting tight association of the C terminus with the OM. Under harsher conditions (4 or 6 M urea or 0.05 or 0.1 M NaOH), the C-terminal fragment partly or completely moved to the soluble fraction (Fig. 4 and data not shown). Note that full-length PapC was partly extracted under these conditions but to a much lesser extent than the C-terminal fragment. These data are consistent with insertion of the C terminus within the usher {beta}-barrel. However, an integral membrane location cannot be ruled out as a C-terminal domain containing six TM {beta}-strands would be extracted from the OM more readily than full-length PapC containing up to 32 TM strands.



View larger version (31K):
[in this window]
[in a new window]
  		FIG. 4.
The PapC C terminus is tightly associated with the OM. OM containing full-length PapC was subjected to limited trypsin proteolysis, treated with the indicated agents, and centrifuged to separate membrane (P) from soluble (S) fractions. The fractions were subjected to SDS-PAGE and blotted with an anti-His tag antibody. A, full-length PapC; B, the C-terminal fragment.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
REFERENCES
 
We report here the development of a site-directed fluorescence labeling technique that allows the topology of bacterial OM proteins to be mapped in intact bacteria using small molecule probes. Most importantly, this method permits accurate differentiation of periplasmic from surface residues and is less susceptible to steric interference than methods relying on proteases or antibodies (7, 9–11). We used site-directed fluorescence labeling to map the topology of the PapC usher. Based on our mapping results and analysis of the usher C terminus, we propose that PapC, and presumably other members of the usher superfamily, consist of an N-terminal 26-stranded {beta}-barrel and a separate C-terminal domain. The C-terminal domain contains regions exposed to the surface and periplasm and is tightly associated with the OM. To explain these results, we propose that the C terminus inserts within the lumen of the N-terminal {beta}-barrel. This topology model for PapC is similar to the topology revealed by crystallography for FhuA and related OM proteins that function as siderophore-mediated iron importers (33–36). The FhuA C terminus folds as a 22-stranded {beta}-barrel, and the N terminus forms a separate domain that inserts within the {beta}-barrel, forming a plug. The plug domain of FhuA is composed of 159 residues, comparable to the 222 residues we propose for the C-terminal domain of PapC.

Previous analysis predicted that PapC contained 24 TM {beta}-strands spread throughout the PapC sequence (30). The previous model matches well with our current model for the N-terminal half of the usher but differs considerably for the C-terminal half. Topology modeling studies also misidentified the N-terminal plug domain of the iron-siderophore import proteins as containing TM strands with alternating hydrophobic residues that formed part of the {beta}-barrel (8, 11). This highlights the difficulty of predicting the topology of {beta}-barrel OM proteins. Interestingly a new hidden Markov model-based prediction program (48) agreed with our identification of the N-terminal {beta}-barrel region of PapC and predicted that the C terminus lacks TM strands. However, this program predicted an altered arrangement for the TM strands in the N-terminal {beta}-barrel presumably because the program was trained using typical OM proteins that have large surface loops and only short periplasmic regions. The ushers, as shown in this and previous studies (30, 46, 49), contain short extracellular loops with larger regions exposed to the periplasm. The FasD usher required for biogenesis of 987P pili in enterotoxigenic E. coli was predicted to have 28 TM {beta}-strands (49), and the FaeD usher required for biogenesis of K88 pili in enterotoxigenic E. coli was predicted to have 22 TM {beta}-strands (46). Similar to our model, experimental analysis of FaeD localized a large C-terminal domain to the periplasm despite the identification of TM strands by computer analysis (46). In contrast to our model, the first ~130 residues of the N termini of both FaeD and FasD were proposed to lack TM strands and reside in the periplasm (25, 46, 49). However, this analysis was based on localization of PhoA fusions and epitope insertions, and, as noted here, these methods cannot be used to map periplasmic regions reliably.

The short surface loops of PapC may account for some of the difficulties we encountered with the TEV protease mapping experiments. The outer leaflet of the OM is composed of lipopolysaccharide (44), and TEV cleavage sites located in short surface loops may have been obscured by the surface polysaccharides of lipopolysaccharide. Such shielding effects by lipopolysaccharide have been noted previously (10, 11). In agreement with this, we were unable to detect c-Myc epitopes that were added to various TnTAP insertion sites by either whole bacteria enzyme-linked immunosorbent assay or immunofluorescence microscopy.4 Isolated OM reacted well with the anti-c-Myc antibody, suggesting that surface-located epitopes were obscured in intact bacteria. This again highlights the difficulty of differentiating periplasmic from surface regions using epitope or protease mapping techniques.

Residues 1–124 of PapC, comprising the mature N terminus and the large loop containing the N-terminal cysteine pair, form the targeting site for chaperone-subunit complexes and are also required for subsequent pilus assembly events (27). Residues 2–11 are an essential part of the chaperone-subunit targeting site (27). In agreement with this, the L8TEVins mutant was completely defective for pilus biogenesis (HA titer of zero), whereas a TnTAP insertion before residue Val-1 (this insertion remained part of the mature N terminus) had no effect on usher function. Six TnTAP insertions were isolated in the periplasmic loop containing the Cys-70 and Cys-97 N-terminal cysteine pair (Fig. 1). Interestingly the two insertions prior to Cys-70 (S58TEVins and A69TEVins) did not disrupt PapC function, whereas ushers containing insertions between the cysteines were not functional (all had HA titers of zero). These results demonstrate that the region between the N-terminal cysteines is critical for usher function with the cysteine pair presumably acting to properly structure this loop region. We were unable to isolate TnTAP insertions in the PapC C terminus between residue Thr-681 and the final residue Lys-809 (Fig. 1), implying that the structure of the C terminus is sensitive to perturbation. This seems contrary to the finding that the C terminus is not necessary for formation of a stable {beta}-barrel by PapC. However, we have previously noted that small deletions or alterations in the C terminus can destabilize the usher, leading to protein degradation (30).5 A similar phenomenon was noted for the C terminus of the FaeD usher (46) and also was described for the N-terminal plug domain of FhuA (8).

As expected, the majority of the single cysteine insertion or substitution mutations in PapC did not affect usher function. Most of the cysteine mutations that did cause HA defects were located in surface loops (Fig. 1). This was contrary to expectations as interactions of PapC with chaperone-subunit complexes are thought to occur mainly on the periplasmic face of the usher. In particular, six of seven mutations in the surface loop from Leu-246 to Gly-259 completely disrupted usher function, giving HA titers of zero (Fig. 1 and Supplemental Table III). This loop may be important for interactions with the pilus fiber during secretion. Alternatively the loop may fold back into the usher channel to contact pilus subunits, or it could be important for conformational changes that occur in the usher during pilus biogenesis.

A recent high resolution electron microscopy study of PapC demonstrated that the usher assembles as a dimeric, twin pore complex.6 Each PapC monomer has dimensions of ~5 x 7 nm in diameter and contains an apparent central channel of ~2nm in diameter. By comparison, FhuA measures ~4 x 5 nm in diameter, and the channel of the 22-stranded FhuA {beta}-barrel, which is occluded by the N-terminal plug domain, measures roughly 2–3 nm in diameter (34, 35). Therefore, a 26-stranded PapC {beta}-barrel could easily form a 2-nm channel. Electron microscopy analysis of a PapC truncation mutant lacking the C-terminal 169 residues revealed that it has the same overall shape, size, and apparent central channel as full-length PapC but with an altered dimerization interface.6 Together with the topology model presented here, this strongly suggests that the PapC channel is formed by the N-terminal {beta}-barrel domain and that the C terminus forms a structurally distinct domain. The location of the C-terminal domain could not be discerned from the electron microscopy structures,6 but the tight association of the C terminus with the OM argues for either an integral membrane location or insertion within the N-terminal {beta}-barrel. Based on the evidence described above and by structural homology with FhuA, we favor the latter interpretation. The PapC C terminus is required for stages of pilus biogenesis subsequent to the initial targeting of chaperone-subunit complexes (30). A pluglike C-terminal domain might act to control access of chaperone-subunit complexes to the usher channel. Following the targeting of chaperone-subunit complexes to the usher N terminus, complexes form stable interactions with the usher C terminus (29, 30). Interaction of chaperone-subunit complexes with the usher C terminus might induce rearrangement or gating of the C terminus, opening the usher channel for fiber assembly and secretion to the cell surface.


   FOOTNOTES
 
* This work was supported by Public Health Service Grant GM62987 from the National Institute of General Medical Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Back

{boxs} The on-line version of this article (available at http://www.jbc.org) contains Supplemental Tables I–III. Back

{ddagger} Present address: Dept. of Medicine, Stony Brook University, Stony Brook, NY 11794-8151. Back

§ To whom correspondence should be addressed: 242 Center for Infectious Diseases, Stony Brook University, Stony Brook, NY 11794-5120. Tel.: 631-632-4549; Fax: 631-632-4294; E-mail: David.Thanassi{at}stonybrook.edu.

1 The abbreviations used are: OM, outer membrane; TM, transmembrane; PhoA, alkaline phosphatase; HA, hemagglutination assay; TEV, tobacco etch virus; AMS, 4-acetamido-4'-maleimidylstilbene-2–2'-disulfonic acid; RT, room temperature; OGM, Oregon green 488 maleimide; DDM, dodecylmaltoside; WT, wild type. Back

2 T. Ng and D. G. Thanassi, unpublished data. Back

3 S. Shu Kin So and D. G. Thanassi, unpublished data. Back

4 N. S. Henderson, C. Martin, and D. G. Thanassi, unpublished data. Back

5 D. G. Thanassi, unpublished data. Back

6 Li, H., Qian, L., Chen, Z., Thahbot, D., Liu, G., Liu, T., and Thanassi, D. G. (2004) J. Mol. Biol., in press. Back


   ACKNOWLEDGMENTS
 
We thank Michael Ehrmann for providing materials for the TnTAP experiments and for helpful discussions. We thank Ashwin Basavaraj, Jigna Mehta, and Eirini Tsangalidis for assistance with the TnTAP experiments. We thank Jason Hall and Peter Maloney for providing the OxlT plasmids and for helpful discussions. We thank Jorge Benach for critical reading of the manuscript.



   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Wimley, W. C. (2003) Curr. Opin. Struct. Biol. 13, 404-411[CrossRef][Medline] [Order article via Infotrieve]
  2. Tamm, L. K., Arora, A., and Kleinschmidt, J. H. (2001) J. Biol. Chem. 276, 32399-32402[Free Full Text]
  3. Schulz, G. E. (2000) Curr. Opin. Struct. Biol. 10, 443-447[CrossRef][Medline] [Order article via Infotrieve]
  4. Koebnik, R., Locher, K. P., and Van Gelder, P. (2000) Mol. Microbiol. 37, 239-253[CrossRef][Medline] [Order article via Infotrieve]
  5. Paschen, S. A., Waizenegger, T., Stan, T., Preuss, M., Cyrklaff, M., Hell, K., Rapaport, D., and Neupert, W. (2003) Nature 426, 862-866[CrossRef][Medline] [Order article via Infotrieve]
  6. Manoil, C., Mekalanos, J. J., and Beckwith, J. (1990) J. Bacteriol. 172, 515-518[Abstract/Free Full Text]
  7. Newton, S. M., Klebba, P. E., Michel, V., Hofnung, M., and Charbit, A. (1996) J. Bacteriol. 178, 3447-3456[Abstract/Free Full Text]
  8. Koebnik, R., and Braun, V. (1993) J. Bacteriol. 175, 826-839[Abstract/Free Full Text]
  9. Guédin, S., Willery, E., Tommassen, J., Fort, E., Drobecq, H., Locht, C., and Jacob-Dubuisson, F. (2000) J. Biol. Chem. 275, 30202-30210[Abstract/Free Full Text]
  10. Merck, K. B., de Cock, H., Verheij, H. M., and Tommassen, J. (1997) J. Bacteriol. 179, 3443-3450[Abstract/Free Full Text]
  11. Murphy, C. K., Kalve, V. I., and Klebba, P. E. (1990) J. Bacteriol. 172, 2736-2746[Abstract/Free Full Text]
  12. Dodson, K. W., Jacob-Dubuisson, F., Striker, R. T., and Hultgren, S. J. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 3670-3674[Abstract/Free Full Text]
  13. Norgren, M., Baga, M., Tennent, J. M., and Normark, S. (1987) Mol. Microbiol. 1, 169-178[CrossRef][Medline] [Order article via Infotrieve]
  14. Bullitt, E., and Makowski, L. (1995) Nature 373, 164-167[CrossRef][Medline] [Order article via Infotrieve]
  15. Kuehn, M. J., Heuser, J., Normark, S., and Hultgren, S. J. (1992) Nature 356, 252-255[CrossRef][Medline] [Order article via Infotrieve]
  16. Soto, G. E., and Hultgren, S. J. (1999) J. Bacteriol. 181, 1059-1071[Free Full Text]
  17. Bock, K., Breimer, M. E., Brignole, A., Hansson, G. C., Karlsson, K.-A., Larson, G., Leffler, H., Samuelsson, B. E., Strömberg, N., Svanborg-Edén, C., and Thurin, J. (1985) J. Biol. Chem. 260, 8545-8551[Abstract/Free Full Text]
  18. Thanassi, D. G., Saulino, E. T., and Hultgren, S. J. (1998) Curr. Opin. Microbiol. 1, 223-231[CrossRef][Medline] [Order article via Infotrieve]
  19. Pugsley, A. (1993) Microbiol. Rev. 57, 50-108[Abstract/Free Full Text]
  20. Van den Berg, B., Clemons, W. M., Jr., Collinson, I., Modis, Y., Hartmann, E., Harrison, S. C., and Rapoport, T. A. (2004) Nature 427, 36-44[CrossRef][Medline] [Order article via Infotrieve]
  21. Barnhart, M. M., Pinkner, J. S., Soto, G. E., Sauer, F. G., Langermann, S., Waksman, G., Frieden, C., and Hultgren, S. J. (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 7709-7714[Abstract/Free Full Text]
  22. Choudhury, D., Thompson, A., Stojanoff, V., Langermann, S., Pinkner, J., Hultgren, S. J., and Knight, S. D. (1999) Science 285, 1061-1066[Abstract/Free Full Text]
  23. Sauer, F. G., Fütterer, K., Pinkner, J. S., Dodson, K. W., Hultgren, S. J., and Waksman, G. (1999) Science 285, 1058-1061[Abstract/Free Full Text]
  24. Thanassi, D. G., Saulino, E. T., Lombardo, M.-J., Roth, R., Heuser, J., and Hultgren, S. J. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 3146-3151[Abstract/Free Full Text]
  25. Valent, Q. A., Zaal, J., de Graaf, F. K., and Oudega, B. (1995) Mol. Microbiol. 16, 1243-1257[CrossRef][Medline] [Order article via Infotrieve]
  26. Klemm, P., and Christiansen, G. (1990) Mol. Gen. Genet. 220, 334-338[Medline] [Order article via Infotrieve]
  27. Ng, T. W., Akman, L., Osisami, M., and Thanassi, D. G. (2004) J. Bacteriol. 186, 5321-5331[Abstract/Free Full Text]
  28. Nishiyama, M., Vetsch, M., Puorger, C., Jelesarov, I., and Glockshuber, R. (2003) J. Mol. Biol. 330, 513-525[CrossRef][Medline] [Order article via Infotrieve]
  29. Saulino, E. T., Thanassi, D. G., Pinkner, J. S., and Hultgren, S. J. (1998) EMBO J. 17, 2177-2185[CrossRef][Medline] [Order article via Infotrieve]
  30. Thanassi, D. G., Stathopoulos, C., Dodson, K. W., Geiger, D., and Hultgren, S. J. (2002) J. Bacteriol. 184, 6260-6269[Abstract/Free Full Text]
  31. Sauer, F. G., Pinkner, J. S., Waksman, G., and Hultgren, S. J. (2002) Cell 111, 543-551[CrossRef][Medline] [Order article via Infotrieve]
  32. Zavialov, A. V., Berglund, J., Pudney, A. F., Fooks, L. J., Ibrahim, T. M., MacIntyre, S., and Knight, S. D. (2003) Cell 113, 587-596[CrossRef][Medline] [Order article via Infotrieve]
  33. Ferguson, A. D., Chakraborty, R., Smith, B. S., Esser, L., van der Helm, D., and Deisenhofer, J. (2002) Science 295, 1715-1719[Abstract/Free Full Text]
  34. Ferguson, A. D., Hofmann, E., Coulton, J. W., Diederichs, K., and Welte, W. (1998) Science 282, 2215-2220[Abstract/Free Full Text]
  35. Locher, K. P., Rees, B., Koebnik, R., Mitschler, A., Moulinier, L., Rosenbusch, J. P., and Moras, D. (1998) Cell 95, 771-778[CrossRef][Medline] [Order article via Infotrieve]
  36. Buchanan, S. K., Smith, B. S., Venkatramani, L., Xia, D., Esser, L., Palnitkar, M., Chakraborty, R., van der Helm, D., and Deisenhofer, J. (1999) Nat. Struct. Biol. 6, 56-63[CrossRef][Medline] [Order article via Infotrieve]
  37. Grant, S. G., Jessee, J., Bloom, F. R., and Hanahan, D. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 4645-4649[Abstract/Free Full Text]
  38. Blomfield, I. C., McClain, M. S., and Eisenstein, B. I. (1991) Mol. Microbiol. 5, 1439-1445[Medline] [Order article via Infotrieve]
  39. Baneyx, F., and Georgiou, G. (1990) J. Bacteriol. 172, 491-494[Abstract/Free Full Text]
  40. Ye, L., Jia, Z., Jung, T., and Maloney, P. C. (2001) J. Bacteriol. 183, 2490-2496[Abstract/Free Full Text]
  41. Ehrmann, M., Bolek, P., Mondigler, M., Boyd, D., and Lange, R. (1997) Proc. Natl. Acad. Sci. U. S. A. 64, 13111-13115
  42. Fu, D., and Maloney, P. C. (1998) J. Biol. Chem. 273, 17962-17967[Abstract/Free Full Text]
  43. Seal, R. P., Leighton, B. H., and Amara, S. G. (1998) Methods Enzymol. 296, 318-331[CrossRef][Medline] [Order article via Infotrieve]
  44. Nikaido, H., and Vaara, M. (1985) Microbiol. Rev. 49, 1-32[Free Full Text]
  45. Loo, T. W., and Clarke, D. M. (1995) J. Biol. Chem. 270, 843-848[Abstract/Free Full Text]
  46. Harms, N., Oudhuis, W. C., Eppens, E. A., Valent, Q. A., Koster, M., Luirink, J., and Oudega, B. (1999) J. Mol. Microbiol. Biotechnol. 1, 319-325[Medline] [Order article via Infotrieve]
  47. Cuff, J. A., and Barton, G. J. (2000) Proteins 40, 502-511