Disparate Subcellular Localization Patterns of Pseudomonas aeruginosa Type IV Pilus ATPases Involved in Twitching Motility

Bacterial strains and growth media.

The bacterial strains and plasmids used in this study are described in Table 1. Loss of twitching motility, swimming motility, or T2S in transposon mutants from the University of Washington collection was confirmed by functional assays (swimming on 0.3% agar, twitching motility in 1% agar, and T2S on skim milk agar plates). PCR analysis was performed to verify the site of transposon insertion in each mutant (data not shown). Bacteria were grown on Luria-Bertani (LB) agar. Antibiotics were used at the following concentrations: 15 μg of gentamicin ml⁻¹ for E. coli and 200 μg of gentamicin ml⁻¹ and 50 μg of tetracycline ml⁻¹ for P. aeruginosa. Plasmids constructs were made in E. coli and transformed by heat shock. All final constructs were introduced into P. aeruginosa strains by electroporation.

Plasmid construction.

All plasmids were purified by using Qiaprep spin miniprep columns (QIAGEN). Standard recombinant DNA manipulation techniques were used (29). Enzymes were purchased from Invitrogen and used as recommended by the manufacturer. Oligonucleotides synthesized by ACGT Co. (Toronto, Ontario, Canada) were used for sequencing and PCR. The oligonucleotide primer sequences used for PCRs are shown in Table 2. All PCRs were carried out with an annealing temperature of 57°C for 30 to 45 s, and the extension time varied depending on the expected size of the amplicon.

Sequential ligation of the pilB, pilT, and pilU genes and yfp was performed to make the fluorescent fusion protein constructs. Briefly, pilB, pilT, and pilU were amplified by PCR from the start codon of each gene and then ligated into pUCP20Gm by using restriction sites within the PCR primers. The yellow fluorescent protein (YFP) gene, yfp, was amplified from pEYFP-N1 in order to encompass a Shine-Delgarno sequence at the 5′ end (designed into the primer) and a linker region at the 3′ end (from the vector) and then ligated upstream of the pilB, pilT, or pilU gene that was cloned into pUCP20Gm by using restriction sequences that were also designed into the PCR primers. All constructs were sequenced to verify the integrity of the fusion sequence and the reading frame.

To make the deleted fusion proteins used to test the length required for polar localization, downstream PCR primers, each containing a stop codon, were designed at the desired point of deletion. Full-length yfp-pilT and yfp-pilU constructs were used as the templates for amplification of partially deleted fusion products, which were subsequently cloned into pUCP20Gm by using restriction sites introduced by PCR primers.

pilT and pilU chimeras were engineered by using a Quik Change XL site-directed mutagenesis kit (Stratagene, La Jolla, Calif.) as described by Geiser et al. (16). For example, a yfp-N′pilT-C′pilU chimera was produced by exchanging the N-terminal portion of pilU by using a PCR fragment of yfp linked to the N terminus of pilT with a region of homology (∼22 bp) to the 5′ end of the C-terminal portion of pilU. The PCR fragment was utilized as an overlapping mutagenic primer, while the full-length yfp-pilU construct was used as the template for the synthesis of the chimera. The template DNA was removed by digestion with DpnI for 3 h, and this was followed by transformation into E. coli XL-10 competent cells as described by the manufacturer. The entire transformation mixture was plated on LB agar supplemented with 15 μg of gentamicin per ml and incubated at 37°C overnight. Random clones were selected, the plasmids were isolated, and the correct construction of the chimera was confirmed by PCR and DNA sequencing.

Sub-agar-surface twitching motility assay.

Thin (3-mm) 1% LB agar plates were prepared for the twitching assay. Sterile toothpicks were used to inoculate a single colony by stabbing into the bottom of the agar plate (47). The plates were incubated at 37°C for 24 h.

Fluorescent microscopy.

Single colonies were streaked on agar plates with the appropriate antibiotics and then incubated for 18 h at 37°C. Each sample was prepared by mixing a single colony with 10 μl of phosphate-buffered saline (pH 7.1) on a glass slide. Polylysine-coated coverslips were used to mount the cells for photography. The polylysine-coated coverslips were prepared in advance by dipping coverslips in a polylysine solution (catalog no. P8920; Sigma Diagnostic Inc.) and then allowing them to dry at room temperature.

By using fluorescence illumination, slides were viewed at a magnification of ×100 with oil immersion by using a Leica DMR upright microscope fitted with a Hamamatsu ORCA camera. Open Lab V 3.1.3 (Improvision) software was used to collect fluorescent images. Images were processed with PhotoShop 7.0.1 (Adobe).

Characterization of ATPase fusions to spectral variants of GFP.

To study the intracellular localization of PilB, PilT, and PilU in P. aeruginosa, we took a fluorescent protein fusion approach. Initially, we used green fluorescent protein (GFP), but low levels of native autofluorescence detected in the green channel prompted a switch to two spectral variants of GFP, cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP). cfp and yfp genes optimized for eukaryotic expression were chosen instead of prokaryotic versions because we determined that their high G+C codon usage closely resembled that preferred by P. aeruginosa (55).

First, CFP was translationally fused to the C terminus of PilT, and the fusion was expressed from the lac promoter of pUCP20Gm in P. aeruginosa. The PilT-CFP fusion was not functional, in that it was unable to complement the twitching defect in a pilT mutant (Fig. 1). Next, we fused CFP or YFP to the N terminus of PilT to generate CFP-PilT and YFP-PilT. Both fusions complemented the twitching defect of a pilT mutant (Fig. 1), suggesting that the C terminus, but not the N terminus, may be involved in oligomerization or the activity of PilT. Since YFP fusions were brighter than CFP fusions in P. aeruginosa and since there was no difference in the localization of proteins fused to the two spectral variants, YFP was used to generate subsequent fusions. An N-terminal fusion of YFP to PilU also complemented the pilU mutant (Fig. 1). Although our construct containing pilB alone complemented the pilB mutant, the N-terminal YFP fusions to pilB could not complement twitching, while a C-terminal YFP fusion to pilB produced an unstable fusion protein. All other fluorescent fusions, including those unable to effect complementation, were expressed at the expected molecular masses on Western blots of whole-cell lysates from mutants harboring the relevant constructs, probed with a monoclonal antibody to YFP (data not shown).

Localization of YFP-PilB, YFP-PilT, and YFP-PilU.

Figures 2A and B show the pilT mutant expressing CFP alone and the nonfunctional, C-terminal PilT-CFP fusion. Both transformants exhibited diffuse cytoplasmic fluorescence. However, cells expressing the N-terminal CFP-PilT fusion (Fig. 2C) showed punctate bipolar fluorescence. Similarly, pilT mutants expressing YFP-PilT (Fig. 2E) showed punctate bipolar fluorescence, while YFP alone showed diffuse fluorescence in the pilT mutant background (Fig. 2D). Representative cells are shown, and more than 90% of the cells exhibited bipolar localization of the fusions. Approximately 5% of the cells were nonfluorescent, while the remaining cells were fluorescent at both poles and at the division septum (Fig. 2F), suggesting that there was recruitment of PilT to the division septum (and the future new pole) prior to cell division. The localization of the PilT fusions in the wild-type strain was identical (data not shown).

pilB mutants expressing YFP alone exhibited diffuse cytoplasmic fluorescence (Fig. 2G), while mutants expressing YFP-PilB showed punctate bipolar fluorescence (Fig. 2H). In contrast, while pilU mutants expressing YFP exhibited diffuse fluorescence (Fig. 2I), pilU mutants expressing YFP-PilU showed unipolar fluorescence (Fig. 2J). Approximately 90% of the pilU mutants expressing YFP-PilU exhibited unipolar fluorescence and 5% were nonfluorescent, while the remaining cells had bipolar fluorescence but lacked localization to the septum, indicating that the two resulting daughter cells would retain unipolar localization of YFP-PilU.

Orientation of YFP-PilU relative to the piliated or flagellated cell pole.

To orientate YFP-PilU relative to the TFP and flagellum (15), we used time-lapse images captured by fluorescent microscopy. When free-swimming pilU mutant cells expressing YFP-PilU were photographed at 1-s intervals by fluorescent microscopy, unipolar YFP-PilU could be observed at the trailing pole, which was the origin of the flagellum-powered propulsion. Therefore, PilU localized to the piliated or flagellated pole of the cell (Fig. 3).

Localization of YFP-PilB, YFP-PilT, and YFP-PilU in P. aeruginosa TFP mutants.

To determine whether other proteins may play a role in the polar localization of PilB, PilT, and PilU, the fluorescent fusions were expressed in P. aeruginosa strains carrying mutations in genes related to TFP assembly and function, in genes related to the function of other polar structures, or in genes encoding proteins previously shown to be localized to the pole. We focused on TFP gene products that have been shown or predicted to be membrane associated. Strains with mutations in the following genes were tested (Table 1): pilA (TFP subunit); pilB, pilT, and pilU (TFP ATPases); pilC, pilD, pilJ, pilM, pilN, pilO, pilP, pilQ, pilY1, and fimV (TFP assembly and regulation); fliC (flagellum synthesis); xcpY and xcpZ (T2S); and pilS and fimX (TFP proteins previously demonstrated to be polar) (3, 19). YFP-PilT and YFP-PilU retained the characteristic polar localization pattern in all strains tested, suggesting that they localized independent of their substrate PilA, other TFP ATPases, TFP membrane proteins, the flagellum, and the other polar TFP-associated gene products PilS and FimX. YFP-PilB remained bipolar in all of the mutant backgrounds mentioned above except for the pilC mutant background, in which it exhibited diffuse cytoplasmic fluorescence (Fig. 4).

Localization of YFP-PilT and YFP-PilU deletion mutants.

pilT and pilU are paralogues (67% amino acid similarity) that appear to have arisen through gene duplication based on their tandem arrangement in the P. aeruginosa chromosome. Loss of either gene product results in a nontwitching, hyperpiliated phenotype. However, the dissimilar localization patterns of these proteins suggest that each gene contains unique information. Therefore, we focused on the two proteins to determine the location of the signal required for bipolar and unipolar localization.

As TFP ATPases are members of the AAA+ protein family, they possess characteristic His and Asp boxes with unknown functions in the vicinity of the nucleotide-binding site in the C terminus (42). We first created arbitrary truncations of YFP-PilU containing approximately the first one-third, first two-thirds, and last two-thirds of PilU (Fig. 5) to determine whether the N- or C-terminal fragments alone were sufficient for polar localization. Fusions PilU_1-117 (first one-third) and PilU_118-382(last two-thirds) had diffuse cytoplasmic distributions (Fig. 5), suggesting that neither the N-terminal fragment alone nor the C-terminal fragment alone was sufficient for polar localization. A YFP-PilU deletion lacking the C terminus beyond Phe₂₄₄ (first two-thirds intact) retained polar localization but could not restore twitching motility (Fig. 5), suggesting that it was possible to uncouple the polar localization and motor function of these proteins.

Next, deletions of YFP-PilT and YFP-PilU downstream of conserved AAA+ protein motifs, such as the Asp box, Walker B box, and His box, were generated (Fig. 5). Both YFP-PilT and YFP-PilU deletions lacking the region beyond the His or Walker B box retained the characteristic patterns of polar localization, while deletions lacking the region beyond the Asp or Walker A box resulted in diffuse fluorescence. None of the deletions, including those localizing to the pole, could restore the twitching motility of the corresponding mutant strains (data not shown). These results suggest that an intact ATP-binding site is required for twitching motility but is dispensable for polar localization.

Localization of PilT-PilU chimeras.

Since partial loss of the C terminus did not affect the characteristic localization of PilU and PilT, it is plausible that the information specifying unipolar or bipolar localization of PilU and PilT resides in the N-terminal portions of these proteins. By using the recently solved crystal structure of EpsE, a T2S ATPase, as a model (42), the structures of PilT and PilU were predicted by using the 3D-PSSM fold recognition server (24). A significant match in the structure of these proteins to EpsE was predicted (PSSM e-value, 1.59e-05; 95% confidence level). Like EpsE, PilT and PilU are predicted to have globular N-terminal and C-terminal domains joined by a short central linker region (Fig. 6).

Using this information, we created a fluorescent chimera, YFP-N′PilT-C′PilU, by fusing the N terminus of PilT (up to and including Ser₁₀₁ in the linker region) to the C terminus of PilU (after Thr₁₀₁ in the linker region) to test the hypothesis that the information required for the characteristic polar localization of PilT or PilU was contained in the N-terminal domain. If the chimeric protein demonstrated bipolar localization like YFP-PilT, the results would suggest that the region critical for bipolar localization resides in the N terminus. If, however, the chimera was unipolar like YFP-PilU, the results would suggest that the C terminus was responsible for specifying localization.

The YFP-N′PilT-C′PilU chimera exhibited bipolar localization in both pilT and pilU mutants of P. aeruginosa (Fig. 7) but could not complement twitching motility in either background (data not shown). The inverse chimera, YFP-N′PilU-C′PilT, also could not complement twitching motility in pilT and pilU mutants but exhibited unipolar localization in approximately 5% of the cells; the remaining cells exhibited diffuse cytoplasmic fluorescence (Fig. 7).

Localization of TFP ATPases in P. aeruginosa.

Previously, the localization of only sensory-related TFP proteins, such as PilS and FimX, has been examined in P. aeruginosa. In this work, we focused on the mechanical components of the TFP system, using fluorescent protein fusions to the TFP ATPases to examine their spatial relationship to each other and to TFP.

Based on twitching assay results, fluorescent proteins fused to the C terminus of PilT resulted in a nonfunctional fusion product. It is conceivable that fusions to this region of PilT, a protein that forms hexamers (18), resulted in steric hindrance of subunit interactions, localization to a hypothetical basal structure, or interaction with other accessory proteins. The ATP hydrolase activity of the hexameric traffic ATPase HP0525 from the T4S system of H. pylori was proposed to modulate a conformational shift in the C-terminal domain of the subunits, resulting in opening of one end of the hexameric ring to allow translocation of substrates (59). It is possible that the C-terminal fusion interferes with ATP-hydrolysis-induced conformational changes in PilT (18). EpsE, a T2S ATPase, localizes to the flagellated old pole in V. cholerae (46). Its N terminus interacts with EpsL and EpsM, which is required for membrane and unipolar localization (44). The crystal structure of EpsE revealed that the N terminus of one subunit also interacts with the C terminus of an adjacent subunit to mediate oligomerization (42). Based on the evidence that the N termini of AAA+ ATPases are involved in interactions both with adjacent subunits and with at least one cytoplasmic membrane anchor protein, it is interesting that fluorescent protein fusions to the N termini of P. aeruginosa PilT and PilU did not interfere with twitching function. In contrast, the N-terminal fusions of YFP to PilB resulted in a loss of twitching motility. PilB (∼62 kDa) has a C-terminal domain whose size is similar to the sizes of the C-terminal domains of PilT (∼38 kDa) and PilU (∼42 kDa) but a larger N-terminal domain; therefore, it is conceivable that the as-yet-unidentified function of this region of PilB was affected by the YFP fusion.

We originally hypothesized that all three ATPases would localize to the piliated cell pole. This hypothesis was supported by the colocalization of PilU with TFP, but unexpectedly, PilB and PilT also localized to the nonpiliated cell pole. The bundle-forming pilus system of EPEC has homologues of PilB (BfpD) and PilT (BfpF) (49). BfpF was recently shown by immunofluorescence to localize to one pole in EPEC (20). Collectively, these results suggest that TFP ATPases generally localize to the poles, but the exact pattern may be species and protein dependent.

The bipolar localization of PilB and PilT raises some interesting questions. What feature(s) of the ATPases determines uni- versus bipolar localization? What function(s) do the ATPases PilB and PilT have at the nonpilated pole, if any? In addition to pilus extension and retraction, could PilB and PilT have other functions? PilB is not required for T2S (50), and we ruled out the possibility that PilT and PilU contribute to Xcp-based T2S using a simple skim milk agar-based elastase assay (data not shown). In Myxococcus xanthus, which exhibits TFP-based social motility, most cells are piliated at only one pole, although a small proportion can be pilated at both poles (23). Recent findings of Sun et al. (51) suggest that pilus assembly and retraction switches from one pole to another in an frz chemosensory-dependent manner, but only one pole is active at any given time. Similar observations have been reported by Semmler et al. (47), who observed that individual twitching P. aeruginosa cells rapidly reversed the direction of motility. Taken together, these observations suggest that P. aeruginosa might also exhibit alternating twitching motility at the poles, which may explain the bipolar localization of PilB and PilT. The significance of the unipolar localization of PilU in the context of alternating polar twitching is not clear at this time; however, we might need to revisit the notion that P. aeruginosa TFP are unipolar, a finding that was reported 35 years ago by Fuerst and Hayward (15).

Homologues of PilB and PilT are found in other TFP-possessing species, while PilU is present only in a subset of these species (17, 18, 32, 39, 57). It is reasonable that PilB and PilT, which are more widely conserved, have similar localization patterns, while PilU exhibits a dissimilar pattern. pilU may have evolved by gene duplication in P. aeruginosa or its ancestor, or it may have been common in TFP-possessing species at one time but lost during evolution in all but a few species in which there are new functions for PilU. The fact that other phenotypic differences have been reported for P. aeruginosa pilT and pilU mutants (8, 57) is in agreement with the dissimilar localization patterns of these proteins. We and other workers have observed compromised pilus integrity in PilU mutants (7, 57), suggesting that PilU could be involved in reinforcing the pilus at its proximal end, possibly by mediating interaction with an as-yet-unidentified anchor. Figure 8 shows another previously unreported difference in the appearance of pilT and pilU mutant colonies on LB agar. pilT mutants in both PAK and PAO1 backgrounds are glossy, like the corresponding wild types, but pilU mutants of both strains have a matte appearance that can be complemented to the glossy phenotype when the gene is supplied in trans (data not shown). The basis for this unusual colony phenotype of the pilU mutants is currently under investigation, but this characteristic further emphasizes the distinct functional roles of the PilT and PilU proteins.

Localization of TFP ATPases in P. aeruginosa mutants.

Previous fractionation experiments showed that PilT from N. gonorrhoeae is found both in the cytoplasm and associated with the cytoplasmic membrane (6). Similar observations were made with TadA, the T4S ATPase of Actinobacillus actinomycetemcomitans (2). In the cyanobacterium Synechocystis sp. strain PCC 6803, PilT was found in the membrane fraction but not in the cytosolic fraction (36). PilB, PilT, and PilU of P. aeruginosa lack predicted transmembrane domains required for direct integration within the membrane, as well as secretion signals indicative of transport into the periplasm. These observations, together with the requirement for a cytoplasmic ATP substrate, strongly suggest that the TFP ATPases of P. aeruginosa are likely to be peripherally associated with the inner membrane, like other PulE-VirB11-type ATPases.

The T2S ATPase EpsE requires EpsL, an inner membrane protein, as a tether for membrane association in V. cholerae (43). Therefore, we focused on monitoring the localization of PilB, PilT, and PilU in P. aeruginosa strains with mutations in TFP-associated membrane proteins that could conceivably act as anchors for retention at the poles. We were especially interested in the roles that the PilN and PilO proteins may have in the localization of TFP ATPases. PilMNOPQ is thought to form a membrane-spanning apparatus for TFP transport or assembly. PilP and PilQ are outer membrane proteins, while PilN and PilO are inner membrane proteins (12, 30, 31). In particular, PilN has limited homology with the V. cholerae anchor protein, EpsL. However, the lack of dependence on PilMNOPQ for the localization of TFP ATPases suggests that this apparatus does not serve as a platform for anchoring these ATPases to the cytoplasmic membrane. In P. aeruginosa, the T2S ATPase XcpR requires XcpY, also an EpsL homologue, to associate with the cytoplasmic membrane (1). PilB, PilT, and PilU fusions continued to display polar localization in an XcpY mutant, suggesting that T2S and TFP, although evolutionarily related, do not share an ATPase-anchoring platform protein.

Neither the flagellin (FliC) nor pilin (PilA) was required for the polar localization of TFP ATPases. In N. gonorrhoeae, it was shown that the rate of pilus retraction was independent of the PilT concentration, suggesting that a single PilT functional unit binds to a single pilus (28). If PilB and PilT were constitutively associated with the pilus, then a loss of localization at the piliated pole could be expected in a pilA mutant strain. Therefore, we prefer a model in which the interaction between PilB or PilT and the pilus is transient, perhaps responding selectively to a message delivered by chemotactic response regulators, such as PilG and PilH (11). Characteristic localization of each TFP ATPase was also independent of the two other ATPases, suggesting that mixed oligomers are not formed and that a tertiary complex containing two or more ATPases is not required for localization. PilT and PilU localization was also independent of other TFP membrane proteins that we tested, including FimX and PilS, which were shown previously to have unipolar and bipolar localizations, respectively (3, 19).

Interestingly, YFP-PilB, but not YFP-PilT or YFP-PilU, lost polar localization in the pilC mutant background. The pilC gene is located immediately downstream of pilB and is predicted to encode a ∼37-kDa integral membrane protein (35). PilC has 21% identity and 67% similarity to BfpE from EPEC, which has been shown by a yeast two-hybrid approach to interact with BfpD (PilB) and BfpF (PilT) (9). PilC might act as an anchor for retaining PilB at the cell pole; however, neither the localization of PilC nor its exact function is known at this time. The unchanged polar localizations of YFP-PilT and YFP-PilU in a pilC mutant background suggest that PilT and PilU do not need PilC for localization; however, further experiments are needed to address whether PilC interacts with PilT-PilU to mediate twitching motility.

Uncoupling of localization and twitching.

Analysis of the localization patterns of truncated versions of YFP-PilT and YFP-PilU showed that the twitching function of these ATPases could be uncoupled from their polar localization. Given that oligomerization of AAA+ ATPases is required for their function, might oligomerization be required for polar localization? Sequence alignment of PilT and PilU from P. aeruginosa with EpsE of V. cholerae showed conservation of residues implicated in intersubunit contacts (42). From the alignments reported by Robien et al., we oriented our truncations with respect to these conserved residues.

Our results indicate that neither the N terminus alone nor the C terminus alone could localize, possibly due to the inability to form the requisite N₁:C₂ contact for oligomerization. However, the deletions YFP-PilU_1-244 (retaining all four AAA+ motifs) and YFP-PilT-PilU_1-221 (lacking the His box), with intact putative intersubunit contact regions in the C terminus, retained polar localization. Two deletions (YFP-PilT-PilU_1-156 and YFP-PilT-PilU_1-168) were located in the vicinity of the Asp box, correlating to approximately one-half the length of the region in EpsE proposed to make intersubunit contact. Interestingly, although longer by only 12 amino acids, the larger fragment exhibited polar localization, while the shorter fragment resulted in diffuse fluorescence. A single residue (D₂₉₈ in EpsE) in that 12-residue stretch, located between the eighth and ninth β-strands in EpsE, was shown to be one of the important contact points for oligomerization. Conservative substitutions in this residue are found in PilB (E₃₆₂), PilT (E₁₆₇), and PilU (Q₁₆₇). Taken together, the results suggest that neither an intact AAA+ motif region nor the ability to hydrolyze ATP is required for polar localization of PilT and PilU; hence, localization can be uncoupled from function. Our results do not directly show whether ATP binding or hydrolysis is involved in localization; site-directed mutagenesis studies targeting conserved catalytic residues within the Walker boxes are under way to answer this question.

Localization and function of chimeric PilT-PilU.

The preservation of bipolar localization in a chimera containing the N terminus of PilT and the C terminus of PilU supports our hypothesis that the information directing bipolar localization resides in the N terminus of PilT. However, as shown from our truncation studies, the N terminus alone is not sufficient for polar localization. The inverse chimera with the N terminus of PilU and the C terminus of PilT was unipolar 5% of the time, while the rest of the cells had diffuse cytoplasmic fluorescence. This suggests that the localization of YFP-N′PilU-C′PilT is unstable, but the instability was not due to degradation, as Western blot analysis showed that both chimeras were expressed at the predicted size and at the same levels as other ATPase fusions (data not shown). In the cells for which localization of the chimera was evident, it was always unipolar and never bipolar, supporting the hypothesis that the N termini of the ATPases likely determine localization. It is possible that specific sequence motifs within the N termini of PilT and PilU moderate interactions with unidentified unipolar and bipolar membrane anchor proteins, resulting in different patterns of localization. Although the C-terminal nucleotide-binding domains are highly conserved in the two ATPases, they do not appear to be interchangeable, suggesting that both N and C domains play specific functional roles.

In conclusion, we found that the TFP ATPases are located at one (PilU) or both (PilB and PilT) poles of P. aeruginosa, that the information for localization likely resides in the N termini of the proteins, and that localization and function can be uncoupled. Localization may depend on oligomerization, but it appears to be independent of the ability to hydrolyze ATP, since loss of the Walker B motif does not prevent localization. The integral membrane protein, PilC, is required for the polar localization of PilB. Future studies will be directed at identification of the factors that tether PilT and PilU at their characteristic locations in the cell.