TEXT

Top Abstract Text References

Type IV pili (34), long, thin appendages present on the surfaces of many gram-negative bacteria, are involved in bacterial adherence to various surfaces (9, 10, 31), pathogenicity (5, 10, 21), and related phenomena (38). Type IV pilus biogenesis involves a machinery composed of up to 16 proteins that are located in the different bacterial compartments (19). The precise role of each piliation protein and the way their genes are regulated are unclear. However, it is known that type IV pili contain a major pilin subunit that is proteolytically processed and N methylated by a prepilin peptidase (13, 18, 33, 35). This enzyme, together with the major and minor pilin subunits (2), an outer membrane protein (secretin) (6, 11, 22) with its chaperone-like protein (12), and an ATP binding protein (37) are the best-characterized components of the type IV piliation machinery.

Escherichia coli K-12 is not known to possess type IV pili, but surprisingly, 16 genes at seven different loci (Table 1) that could encode components of a piliation machinery have been identified in its chromosome (7). Recently, Francetic et al. showed that one of these genes, the pppA gene coding for a prepilin peptidase, is poorly expressed and that the levels of PppA protein and activity are four times higher when E. coli K-12 is grown at 37°C than at 30°C (16). In this report, we examine the expression of other putative piliation genes in E. coli K-12 grown under laboratory conditions and examine whether the putative major pilin subunit, PpdD, can be assembled into pili.

First, we examined the transcription of hypothetical type IV piliation genes in E. coli K-12. A reverse transcription-PCR (RT-PCR) assay was used with gene-specific oligonucleotide primers and total bacterial RNA to analyze the expression of ppdD (region A) (Table 1), hofQ (region C; putative outer membrane secretin), and yggR (region D; putative ATP binding protein). As a positive control, expression of the noninduced chromosomal malE gene (8) was also tested by RT-PCR. Total RNA was isolated from bacteria of E. coli strain MG1655 or MC4100 grown to exponential or stationary phase in Luria-Bertani broth (LB) using an RNeasy total RNA kit (Qiagen). Contaminating DNA was removed by digestion with DNase I (Boehringer Mannheim). RNA (5 µg) was reverse transcribed and amplified using Ready To Go RT-PCR (Pharmacia) and 200 pmol of specific oligonucleotide primers. For each pair of RT reactions, one tube was preheated 10 min at 95°C to inactivate reverse transcriptase and to test for DNA contamination. Tubes were then incubated in a thermocycler as specified by Pharmacia. The RT-PCR products were analyzed on a 1% agarose gel. ppdD, hofQ, and yggR were not reverse transcribed and amplified, whereas an RT-PCR product was obtained with chromosomal malE and with ppdD cloned on a plasmid under lac promoter control (see below; also data not shown).

To validate the results of the RT-PCR analysis and to compare the expression of the genes tested with that of the previously studied pppA gene (16), we analyzed the expression of putative type IV piliation genes by constructing two transcriptional fusions with lacZ. The hypothetical promoter region of the hofMNOPQ operon upstream of hofM does not contain a consensus promoter sequence. Therefore, we used PCR to amplify a 314-bp fragment whose 5' end is 275 bp upstream from the translation start of hofM within the upstream gene, mrcA. This PCR fragment was then cloned into pRS551 (15 to 30 copies/cell) between a transcription terminator sequence and promoter-less lacZYA to give pCHAP3103 [(hofM'-lacZ)] (see Tables 2 and 3 for detailed descriptions of the plasmids and strains used in this study). This operon fusion was also introduced by homologous recombination into the chromosome of strain MC4100 to give PAP3003. The region upstream of ppdA in the ppdABC operon also lacks a consensus promoter sequence, so we amplified a 578-bp fragment whose 5' end is 150 bp upstream from the stop codon of thyA, the gene upstream of ppdA. This PCR product was then cloned into pRS551 to give pCHAP3122 [(ppdA'-lacZ)].

-Galactosidase was assayed in E. coli K-12 derivatives grown in LB at 37°C. Plasmid-borne (ppdA'-lacZ) and (hofM'-lacZ) were poorly expressed (117 to 330 and 165 to 320 Miller units (29), respectively, compared to 15 to 40 Miller units for the empty vector), irrespective of whether the bacteria were harvested in exponential or stationary phase. The -galactosidase activity in E. coli carrying chromosomal (hofM'-lacZ) was 17 Miller units, comparable to the previously observed low-level expression of pppA (16). Thus, both RT-PCR and -galactosidase assays show that the hofMNOPQ operon is poorly transcribed. We conclude that the putative type IV piliation genes in regions A, B, C, and D (Table 1) are all poorly expressed.

To characterize the product of the putative major pilin gene, ppdD, we raised rabbit antibodies against a periplasmic MalE-PpdD hybrid protein. To provide a source of PpdD protein, a PCR fragment comprising ppdD with its Shine-Dalgarno sequence was cloned under the control of the lac promoter in different vectors. Immunoblotting of extracts of these strains revealed the presence of an isopropyl--D-thiogalactoside (IPTG)-inducible PpdD (19-kDa) band that was absent from control strains carrying empty vector. We then tested whether PpdD, like all other type IV pilins (34), has an intramolecular disulfide bond and is processed by prepilin peptidases. To do this, plasmids carrying the cloned ppdD PCR fragment were introduced into E. coli K-12 strains with or without PulO (the prepilin peptidase of Pul secreton [14, 29]) or into a Pseudomonas aeruginosa PAK mutant with the major pilin gene pilA deleted but with a functional prepilin peptidase gene pilD (PAK A) (26). When cellular proteins were examined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting, PpdD was found to migrate more slowly when reduced with dithiothreitol (20 kDa; not shown). Thus, PpdD, which has four cysteines, has at least one intramolecular disulfide bond. In the absence of exogenous prepilin peptidase, PpdD was poorly cleaved at 30°C and 50% processed at 37°C in E. coli K-12 (Fig. 1, lanes 1 and 2) but was almost fully processed by PulO and PilD prepilin peptidases in E. coli and P. aeruginosa, respectively (Fig. 1, lanes 3 and 4). Therefore, PpdD has the typical characteristics of a type IV pilin. The predicted PpdD precursor has a short N-terminal leader peptide, similar to that of pilins in Neisseria and Pseudomonas (type IVa) and different from the type IVb pilins typified by Vibrio cholerae TcpA, which have longer (>20-amino-acid) leader peptides (34).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Processing of PpdD by different prepilin peptidases, as determined by SDS-PAGE (12% acrylamide) and immunoblotting with antiserum against MalE-PpdD. Secondary antibodies are horseradish peroxidase-labeled anti-rabbit immunoglobulin G (IgG) and the immunoblots were developed by enhanced chemiluminescence (ECL kit; Amersham). Lane 1, E. coli MC4100(pCHAP3098) grown at 30°C in LB containing ampicillin (100 µg/ml); lane 2, same as lane 1 but grown at 37°C; lane 3, E. coli MC4100(pCHAP3098 pCHAP155 [pulO+]) grown at 37°C in LB containing ampicillin and chloramphenicol (34 µg/ml); lane 4, P. aeruginosa PAK A(pCHAP3117) grown at 37°C in LB containing carbenicillin (150 µg/ml). IPTG (1 mM) was used to induce expression of cloned genes. The upper band corresponds to unprocessed pre-PpdD. The MalE-PpdD antibody was diluted 1/2,000, and proteins were dissolved in sample buffer (5% SDS, 12.5% glycerol, 0.1 M Tris HCl [pH 8.0]) with or without 15 mM dithiothreitol as appropriate.

The MalE-PpdD antibodies did not detect PpdD in E. coli K-12 or in 18 other E. coli isolates that were shown to carry the ppdD gene (16) (not shown). In addition, E. coli K-12 strains lacking the global regulator proteins integration host factor, Lrp, StpA, Fis, and H-NS were indistinguishable from isogenic parent strains with respect to ppdD and hofM expression (data not shown). Therefore, the inability of E. coli to express endogenous type IV piliation genes is not restricted to strain K-12.

The results of the studies presented so far suggested that E. coli K-12 does not produce type IV pili because ppdD and possibly one or more of the piliation genes are poorly expressed. Electron microscopy and immunogold labeling with the MalE-PpdD antibodies were used to determine whether increasing the level of expression of ppdD or of ppdD together with the adjacent piliation genes hofB and hofC (Table 1) could promote piliation in E. coli K-12. Cells from colonies of E. coli PAP5023 (carrying a malE deletion to avoid detection of the MalE protein by the MalE-PpdD antibodies), PAP5023(pCHAP3098) (plasmid carrying only ppdD), and PAP5023(pCHAP3101) (plasmid carrying ppdD, hofB, and hofC) grown on LB plates were examined (see Tables 2 and 3 for descriptions of the plasmids and strains). None of these three strains produced detectable pili that reacted with anti MalE-PpdD (Fig. 2). However, the surface of bacteria producing PpdD was covered by gold particles, indicating that PpdD might be exposed on the cell surface (Fig. 2B and C). We do not have a plausible explanation for this observation.

View larger version (46K): [in this window] [in a new window]   FIG. 2.   E. coli K-12 overexpressing the ppdD, hofB, and hofC genes do not produce type IV pili, as demonstrated by electron microscopic examination of immunogold-labeled bacteria grown on agar plates. Bacteria scraped from the agar surface with a cotton swab were suspended in phosphate-buffered saline (PBS). A drop (20 to 50 µl) of this suspension was placed on a sheet of Parafilm. A Formvar-coated nickel grid (made hydrophilic by glow discharge) was placed on the drop, with the Formvar facing the drop, for 2 min and then sequentially onto drops on the following reagents (at room temperature): PBS-0.1% glutaraldehyde (5 min), PBS-50 mM NH4Cl (5 min), PBS-1% bovine serum albumin (BSA)-1% normal goat serum (PBS-1% BSA-1% NGS) (5 min), and then rabbit anti-MalE-PpdD antiserum diluted 1/100 in PBS-1% BSA-1% NGS (for 30 min). After three washes in PBS-0.1% BSA (for 2 min each), the grid was placed on a drop of immunoglobulin-gold-conjugated anti-rabbit immunoglobulin G (IgG) (heavy and light chains) (10-nm-diameter gold particles) diluted in PBS-0.01% fish skin gelatin (30 min). The grids were then subjected to three washes in PBS (3 min each), fixed in 1% glutaraldehyde in PBS (5 min), and washed twice in distilled water (for 5 min each). The grids were then treated with 4 or 5 drops of 2% sodium phosphotungstate for contrast and with 10 µg of bacitracin per ml. Grids were rapidly air dried and then coated with a 3-nm-thick layer of carbon in a vacuum evaporator. Specimens were examined using a Philips CM12 transmission electron microscope operated under standard conditions in the 80- to 120-kV accelerating voltage range. Strain PAP5023 (A), IPTG-induced strain PAP5023(pCHAP3100) (overexpressing ppdD) (B), and IPTG-induced PAP5023(pCHAP3101) (overexpressing ppdD, hofB, and hofC) (C) are shown. Bars, 100 nm.

Since the expression of genes involved in bacterial pathogenicity, and in particular those involved in pilus biogenesis, is usually regulated and can be induced by specific growth conditions (28), we anticipated that the piliation genes of E. coli K-12 might be subject to similar environmental control. However, the expression of ppdD, as determined by immunoblotting, and hofM, as determined by assaying the -galactosidase activity of E. coli MC4100 (hofM'-lacZ), was unaffected by temperature (20, 30, 37, and 42°C) in LB or minimal medium containing glycerol (0.5%) (25), by known inducers of some pathogenicity genes (LB plus Congo red [27], minimal essential medium and Dulbecco modified Eagle medium [28]; both from Gibco-BRL) or anaerobic conditions in LB or Dulbecco modified Eagle medium (not shown).

In view of our failure to find conditions that increased hofM or ppdD expression, we looked directly for genes that might control their transcription. First, we attempted to inactivate a putative repressor in E. coli MC4100 (hofM'-lacZ) by transposition mutagenesis with ::Tn1098 (41) with selection for tetracycline resistance (Tc^r; 16 µg/ml) and Lac⁺ phenotypes on M63 minimal medium containing lactose (0.2%) (25). None of the approximately 15,000 potential Tn1098 insertion mutants (Tc^r) were Lac⁺. We also performed Tn1098 mutagenesis of MC4100 carrying a low-copy-number plasmid in which ppdD under its own promoter was fused to a promoter-less gene coding kanamycin resistance [(ppdD-aphA3)] (pCHAP3121) (Table 2), with selection for resistance to chloramphenicol (34 µg/ml), tetracycline (16 µg/ml), and kanamycin (Km^r; 30 µg/ml) on Luria-Bertani agar (L agar). None of the approximately 20,000 potential Tc^r mutants obtained was Km^r.

We also tried to overexpress a putative activator gene present in E. coli K-12 by electroporating five independent chromosomal banks (HindIII, XbaI, Sau3A1, and in two separate experiments, BamHI-cleaved MG1655 chromosomal DNA cloned into pUC18) into MC4100 (pCHAP3121) and selecting for Cm^r Ap^r Km^r colonies. From around 20,000 Ap^r colonies that carried a plasmid bearing a fragment of the E. coli K-12 chromosome, 40 exhibited increased Km^r but none of them produced detectable levels of PpdD (not shown).

Due to our failure to find conditions allowing type IV pilus formation in E. coli K-12, we could not determine whether PpdD can, in fact, form filaments resembling pili. Since P. aeruginosa can form type IV pili from pilin subunits encoded by genes cloned from other species of bacteria (4, 15, 20, 23, 40), we tested whether PpdD could form pili in P. aeruginosa PAK A lacking endogenous type IV pilin. Typical type IV pili were observed by electron microscopy in the control strains PAK and PAK AA⁺ (Fig. 3A and C), whereas none was observed with PAK A (Fig. 3B). PAK A PpdD⁺ produced pili, but they appeared to be shorter than those formed by PAK (Fig. 2D).

View larger version (111K): [in this window] [in a new window]   FIG. 3.   Visualization by electron microscopy of pili produced by P. aeruginosa PAK derivatives. Bacteria bearing plasmids carrying cloned genes were grown on medium containing 150 µg of carbenicillin per ml. Basic electron microscopy procedures were as described in the legend to Fig. 2. P. aeruginosa PAK (A), PAK A (B), PAK A A+(pAWJ103) (C), and PAK A PpdD+(pCHAP3117) (D) are shown. The arrows in panel D indicate the short filaments produced by PAK A PpdD+. The insert in panel D shows a higher-magnification view of a short filament. Bars, 100 nm.

To test whether the pili produced by PAK A PpdD⁺ contained PpdD and whether their biogenesis was dependent on PilB, an essential component of the type IV pilus biogenesis machinery (26), we examined PAK and PAK carrying a pilB deletion (PAK B) by immunogold labeling and electron microscopy with PilA antibodies, and PAK A PpdD⁺ and PAK B PpdD⁺ with PpdD antibodies (Fig. 4). As expected, gold particle-coated pili were observed with PAK (Fig. 4A) but not with PAK B (Fig. 4B). Likewise, gold particle-coated pilus-like structures were seen with PAK A PpdD⁺ and not with PAK B PpdD⁺ (Fig. 4C and D). Thus, we conclude that the pili observed by electron microscopy (Fig. 3D) are composed of PpdD and that their biogenesis is PilB dependent.

View larger version (107K): [in this window] [in a new window]   FIG. 4.   Visualization by electron microscopy of immunogold-labeled pili produced by P. aeruginosa PAK derivatives. P. aeruginosa PAK (A) and PAK B (B) with antiserum against PilA and with immunoglobulin-gold-conjugated anti-mouse immunoglobulin G (IgG) diluted to 1/100 and P. aeruginosa PAK A PpdD+ (C) and PAK B PpdD+ (D) with antiserum against PpdD and with immunoglobulin-gold-conjugated anti-rabbit IgG diluted to 1/100. The gold particles were 5 nm in diameter. Bars, 100 nm.

As an alternative approach to detect surface-anchored pilin subunits that could be derived from assembled pili, we extracted cell surface proteins from P. aeruginosa PAK, PAK B, PAK with pilC deleted (another gene essential for type IV pilus biogenesis in P. aeruginosa [26]; PAK C), PAK A PpdD⁺, PAK B PpdD⁺, and PAK C PpdD⁺ by shearing. Proteins that were released were examined by SDS-PAGE and immunoblotting (Fig. 5). PilA was extracted only from strain PAK, and PpdD was extracted only from PAK A PpdD⁺ cells (Fig. 5A and B). As an additional control, we extracted surface proteins from PAK expressing both pilA and ppdD. In this case, both proteins were extracted in amounts equivalent to those obtained with bacteria expressing only one of the pilin genes (Fig. 5C). Thus, production of pilA does not interfere with the assembly of PpdD into surface structures in P. aeruginosa PAK. These two different techniques show that PpdD can replace PilA to form type IV pili in Pseudomonas. Thus, PpdD can act as a major pilin subunit.

View larger version (45K): [in this window] [in a new window]   FIG. 5.   Extraction of pili by shearing of P. aeruginosa PAK derivatives and examination of released and retained proteins by SDS-PAGE (12% acrylamide) and immunoblotting with antiserum against PilA (A) or PpdD (B) or both (C; strain PAK PpdD+ only). P. aeruginosa derivatives were grown overnight on LB agar containing appropriate antibiotics. Colonies were then resuspended in LB containing 10 mM MgCl2 and passed three times through a 25-gauge hypodermic needle on a syringe. The suspensions were then centrifuged at 18,300 × g in a microcentrifuge for 5 min to separate the bacteria from the pilus-enriched supernatant, which was precipitated with trichloroacetic acid. Both fractions were resuspended in sample buffer (5% SDS, 12.5% glycerol, 0.1 M Tris HCl [pH 8.0]) and analyzed by immunoblotting. Gels were loaded with material derived from the same starting volume of bacterial suspension. The MalE-PpdD antibody was diluted 1/2,000, and PilA antibody was diluted 1/1,000. Secondary antibodies are horseradish peroxidase-labeled anti-rabbit immunoglobulin G (IgG) and anti-mouse IgG, respectively. S, supernatant; P, pellet.

It is remarkable that E. coli K-12 contains two sets of related, cryptic genes, one (the gsp gene cluster) presumed to encode a secreton that is part of the general secretory pathway (17) and one (the ppd and hof genes) presumed to be involved in type IV piliation (this study). Together, these genes account for almost 1% of the E. coli K-12 genome (7). Why does E. coli fail to produce type IV pili? We show here that the failure to produce PpdD pilin is not the only explanation. Another cause could be the low level of expression of other putative pilus assembly factors. We did not identify any conditions that increased the expression of these genes, although the mutagenesis and cloning experiments might not have been saturating and many environmental conditions were not tested. However, we would like to raise the possibility that at least some piliation genes are constitutively expressed at very low levels, that the regulatory factors that would normally control their expression are absent from E. coli K-12, that more than one regulatory element is involved, or that the genes are controlled by an essential repressor protein. It is worth noting, however, that other E. coli strains also failed to produce PpdD under laboratory conditions. A further explanation for the absence of type IV pili in E. coli K-12 could be that one or more of the pilus assembly genes is defective.