An inhibitor of DNA binding and uptake events dictates the proficiency of genetic transformation in Neisseria gonorrhoeae: mechanism of action and links to Type IV pilus expression

Authors

  • Finn Erik Aas,

    1. Biotechnology Centre of Oslo, 0317 Oslo, Norway.
    2. Department of Microbiology, Institute of Pharmacy
    3. Centre for Molecular Biology and Neuroscience, and
    Search for more papers by this author
  • Cecilia Løvold,

    1. Biotechnology Centre of Oslo, 0317 Oslo, Norway.
    2. Department of Biology, University of Oslo, 0316 Oslo, Norway.
    Search for more papers by this author
  • Michael Koomey

    Corresponding author
    1. Biotechnology Centre of Oslo, 0317 Oslo, Norway.
    2. Department of Microbiology, Institute of Pharmacy
    3. Centre for Molecular Biology and Neuroscience, and
    Search for more papers by this author

Summary

Although natural genetic transformation is a widely disseminated form of genetic exchange in prokaryotic species, the proficiencies with which DNA recognition, uptake and processing occur in nature vary greatly. However, the molecular factors and interactions underlying intra- and interspecies diversity in levels of competence for natural genetic transformation are poorly understood. In Neisseria gonorrhoeae, the Gram-negative aetiologic agent of gonorrhoea, DNA binding and uptake involve components required for Type IV pilus (Tfp) biogenesis as well as those which are structurally related to Tfp biogenesis components but dispensable for organelle expression. We demonstrate here that the gonococcal PilV protein, structurally related to Tfp pilin subunits, is an intrinsic inhibitor of natural genetic transformation which acts ultimately by reducing the levels of sequence-specific DNA uptake into the cell. Specifically, we show that DNA uptake is enhanced in strains bearing pilV mutations and reduced in strains overexpressing PilV. Furthermore, we show that PilV exerts its effect by acting as an antagonist of ComP, a positive effector of sequence-specific DNA binding. As it prevents the accumulation of ComP at a site where it can be purified by shear extraction of intact cells, the data are most consistent with PilV either obstructing ComP trafficking or altering ComP stability. In addition, we report that ComP and PilV play overlapping and partially redundant roles in Tfp biogenesis and document other genetic interactions between comP and pilV together with the pilE and pilT genes required for the expression of retractile Tfp. Together, the results reveal a novel mechanism by which the levels of competence are governed in prokaryotic species and suggest unique ways by which competence might be modulated.

Introduction

Competence for natural genetic transformation is a programmed physiological state which enables bacteria to take up and process exogenous DNA (Dubnau, 1999). Because the frequencies of productive recombination events associated with transformation are limited by the levels of DNA entering the cell, the competent state represents a crucial step in generating and controlling genetic exchange. It is not surprising then that competence in many species is highly regulated. However, the degrees of competence exhibited in vitro vary greatly from species to species and even within a species (Sikorski et al., 2002) and the selective forces which have shaped the varying propensities to bind, take up and process DNA are poorly understood. Furthermore, the molecular factors and interactions which account for intra-species and intra-strain variability in competence proficiency remain unknown. Genetic characterization of randomly mutagenized cells and directed mutagenesis of candidate targets have led to the identification of genes whose products are required for expression of the competent phenotype. In some instances, mutants isolated in these screens displayed a hypercompetent phenotype and in nearly all of these cases, the mutations resulted in increased or deregulated expression of a competence regulon (Kong et al., 1993; Turgay et al., 1998; Martin et al., 2000; Ma and Redfield, 2000; Lacks and Greenberg, 2001; Ween et al., 2002). A unique exception to this situation may be found in hypertransformable pilAII null mutants of Pseudomonas stutzeri (Graupner and Wackernagel, 2001). PilAII contains the N-terminal domain conserved among type IV pilus (Tfp) subunit proteins and their related family of pilin-like proteins and is most closely similar to PilAI, the likely Tfp subunit of P. stutzeri (Graupner and Wackernagel, 2001). PilAII has been proposed to act by antagonizing an unidentified step in transformation which occurs subsequent to DNA uptake into the cell.

A common theme in most naturally competent species is the requirement for components structurally related to those required for Tfp biogenesis and Type II secretion (Dubnau, 1999). In virtually all cases where it has been directly examined, transformation mutants lacking such components have been shown to be defective in DNA uptake. Transformation in the human pathogenic neisseria species is particularly important to genetic exchange and diversity as it accounts for the extensive polymorphism underlying their panmictic population structure (Smith et al., 1993). Like Haemophilus influenzae, pathogenic neisseria take up double-stranded DNA based on the presence of sequence elements which are distributed in their genomes and those of related species (Danner et al., 1980; Elkins et al., 1991). As defined in N. gonorrhoeae, sequence-specific DNA uptake is clearly correlated with Tfp expression because loss-of-function mutations in any one of a large number of genes encoding biogenesis components lead to a block at this step (Wolfgang et al., 1998a). Three proteins dispensable to Tfp biogenesis have also been defined as being essential to gonococcal DNA uptake: PilT, a TrbB/GspE family member involved in pilus retraction (Wolfgang et al., 1998a), ComP, a Tfp prepilin-like protein (Wolfgang et al., 1999), and ComE, a periplasmic protein with intrinsic DNA binding activity which is closely related to the ComEA competence factor of B. subtilis (Chen and Gotschlich, 2001).

Using the gonococcal system, we recently demonstrated that sequence-specific DNA binding and uptake are dissociable events and we were able to show that ComP contributes in a dose dependent fashion to sequence-specific DNA binding events whereas PilT appears to function primarily at the uptake step (Aas et al., 2002). Despite these findings, it remains unclear how ComP might function in the DNA binding step and how its function might relate to those served by Tfp biogenesis factors. Attempts to address these questions have been complicated by the fact that in wild-type cells, it has not been possible to detect ComP by standard, sensitive immunodetection techniques.

Neisseria gonorrhoeae expresses another Tfp pilin-like protein dispensable for Tfp biogenesis termed PilV which co-purifies with Tfp and is required for efficient Tfp mediated adherence of gonococci to human epithelial cells (Winther-Larsen et al., 2001). Most evidence supports the idea that the adherence defect of null mutants reflected a diminished level or activity of the pilus associated PilC adhesin. In the study cited, it was reported that pilV null mutants were not reduced in transformability relative to the wild-type strain. Here we show using pilV mutants and overexpression that PilV antagonizes transformability by ultimately reducing DNA uptake. The results indicate that PilV acts in this capacity by disrupting the accumulation of ComP and thus most likely exerts its effect at the level of DNA binding. The results also demonstrate that the range of the comP and pilV interactions extends to include pilE and pilT thus providing a direct, albeit genetic, link to components of the Tfp biogenesis machinery.

Results

pilV mutants display enhanced transformability in association with increased DNA uptake

Gonococcal pilV mutants are defective in Tfp-associated adherence to human epithelial cells although they show no quantitative reductions in levels of Tfp expression (Winther-Larsen et al., 2001). In the study cited, pilV mutants were noted to be uncompromised in their genetic transformability. Further investigation here revealed in fact that the transformation efficiencies of mutants bearing either a pilV frameshift mutation (the pilVfs allele) or a kanamycin gene cassette insertion in pilV (the pilV::kan allele) were significantly elevated (approximately 10-fold) over the value seen for the isogenic wild-type parent (Table 1). This hypertransformable phenotype was paralleled by a nearly 40-fold increase in the amounts of DNA taken up into a DNase I resistant state by the cells. The hypertransformable and enhanced DNA uptake phenotypes were directly attributable to the absence of PilV because reintroduction of a wild-type pilV gene copy at another locus in these backgrounds restored both parameters to the levels seen in the wild-type parent. We next examined what effect overexpression of mature PilV would have using a strain carrying a translational fusion of the PilV ORF to the highly expressed pilE gene (expressed from the pilVoe allele). Introduction of this gene construct resulted in a 10-fold decrease in transformability and 60-fold decrease in DNA uptake relative to the wild-type strain. This strain retained normal levels of piliation such that the transformation defects could not be explained by a quantitative reduction in Tfp expression (not shown).

Table 1. . Effects of pilV on transformability and DNA uptake.
Relevant genotypeTransformation
frequency (%)a
Total cell associatedb
(% of DNA added)
DNA uptakeb
(% of DNA added)
  • a

    . Transformation frequency as percentage of recipient cells. Carried out with mutants made in a wild-type recombination background (strains: VD300, GV16, GV17, GV22, GV23, GV29 and GV30, see Table 3).

  • b

    . Total amount of cell associated DNA and uptake was carried out with mutants made in a recA6 background (strains: N400, GV1, GV5, GV2, GV3, GV27 and GV28, see Table 3).

  • Values represent the mean ± standard deviation of at least three individual experiments.

  • ND, not determined.

wt  5.4 ± 3.0  1.0 ± 0.2  0.6 ± 0.2
pilV fs 46.8 ± 26.339.4 ± 8.431.6 ± 9.8
pilV fs , iga::pilV  7.8 ± 2.2  1.3 ± 0.5  0.9 ± 0.2
pilV::kan58.4 ± 22.425.3 ± 8.818.0 ± 6.7
pilV::kan, iga::pilV11.6 ± 8.3NDND
pilV oe   0.4 ± 0.4  0.2 ± 0.10.01 ± 0.002
pilV oe , comP oe 38.1 ± 10.3  9.3 ± 2.5  6.7 ± 1.2

Mutations in a number of gonococcal genes elimi-nate sequence specific DNA uptake. If pilV mutations bypassed the normal DNA binding and uptake pathway, transformation in the absence of PilV might be independent of these genes. To examine this possibility, we tested double-mutant strains carrying a pilV frameshift mutation and concurrently failing to express either the pilin subunit PilE (GV12), ComP (GV10), PilT (GV6) or the Tfp biogenesis components PilF and PilQ (unpublished). In all instances, double mutants lacking PilV were non-transformable (transformation frequencies of less than 10−6) and therefore indistinguishable from the corresponding single mutants (Wolfgang et al., 1998a; 1999; Aas et al., 2002). We also found that DNA uptake in pilV mutants remained dependent on the presence of the 10 bp neisserial DNA uptake sequence (DUS, not shown). Therefore pilV mutations enhance transformability and DNA uptake through the normal transformation pathway. Overexpression of ComP increases transformability by enhancing the levels of DNA binding and uptake (Aas et al., 2002). To further investigate the mechanism of PilV inhibition, a strain simultaneously overexpressing both PilV and ComP (using the comPoe allele) at the same relative levels was tested and found to be hypertransformable and to have 600-fold enhanced DNA uptake relative to the strain only overexpressing PilV (Table 1). These phenotypes were very similar to those seen for the ComP overexpressing strain although the levels of transformability and DNA uptake were somewhat reduced (Aas et al., 2002). We conclude that ComP overexpression is therefore dominant over PilV overexpression with regard to these properties.

PilV and ComP play overlapping and redundant roles in Tfp biogenesis

In the course of characterizing mutants lacking both ComP and PilV, we noted a defect in Tfp-associated auto-agglutination and its associated unique colonial mor-phology. This phenotypic alteration very often reflects diminished Tfp expression and when examined by electron microscopy, these strains had reduced levels of piliation (not shown). When assessed by a quantitative purification scheme (Winther-Larsen et al., 2001), the yields of Tfp from these strains were reduced approximately 10-fold relative to the wild-type or single mutant backgrounds (Fig. 1, top panel). Therefore, ComP and PilV play overlapping and redundant roles in Tfp biogenesis. As previously shown for other biogenesis mutants, reductions in Tfp levels for these strains were correlated with the degradation of PilE into a faster migrating form lacking the first 39 mature residues termed S-pilin (Fig. 1, bottom panel) (Wolfgang et al., 2000). Surprisingly, the defect in biogenesis as a result of their simultaneous absence was corrected by a mutation in the pilT gene whose product is required for pilus retraction (Fig. 1, top panel) to levels exceeding those seen in either a wild-type or pilT background. Suppression by the loss of PilT function localized the biogenesis defect in the comP, pilV mutant to a step at which net Tfp expression is controlled by the equilibrium dynamics of pilus fibre growth and retraction (Wolfgang et al., 2000).

Figure 1.

PilV and ComP play overlapping and redundant roles in Tfp biogenesis. Upper panel, Coomassie-stained SDS-PAGE gel showing the relative amounts of Tfp based on the relative levels of PilE in purified pilus preparations obtained by shear extraction. The amounts of sample loaded were standardized based on the amounts of whole cells from which pili were isolated. Lower panel, immunoblotting of whole cell lysates using rabbit antibodies specific for PilE. S-pilin is a proteolytic degradation product of PilE seen in association with defects in Tfp biogenesis. With regard to PilV, ComP and PilT expression ‘+’ denotes the wild-type allele, whereas ‘−’ denotes a null allele or in the case of PilT expression the non-induced pilTind allele. Strains: 1, N400 (wild-type); 2, GV10 (pilVfs, comP::mTnerm23); 3, GV13 (pilVfs, pilTind, comP::mTnerm23); 4, GV24 (pilV::kan, comP::mTnerm23); 5, GV26 (pilV::kan, pilTind, comP::mTnerm23).

Effects of pilV in ComP mutants retaining partial function

Given the evidence for interactions between pilV and comP, we investigated what effects pilV might have in comP mutants possessing residual function in transformability and DNA uptake. These mutants included strains expressing ComP with single missense substitutions at residues G−1 and E+5 (Aas et al., 2002) and two having transposon insertion mutations in the comP promoter region (Wolfgang et al., 1999). In each instance, introduction of a pilV knockout mutation into these backgrounds led to a dramatic increase in transformability (Fig. 2A). These findings reinforce the previous data that pilV mutations act through the normal ComP-dependent pathway and demonstrate that they can suppress both defects in ComP structure and quantitative defects in comP expression. The enhanced transformability in the double-mutants relative to those carrying the comP mutations alone was particularly noteworthy because it was associated with a reduction in levels of purifiable Tfp. For the comPG−1S allele, levels of purifiable Tfp were reduced fivefold by the absence of PilV whereas for the comPE+5V allele, levels were reduced threefold (Fig. 2B). These findings indicate that the altered proteins encoded by these comP alleles are also partially defective vis a vis biogenesis. Reductions in Tfp expression were also seen for the pilV strains with the comP promoter region transposon insertions (not shown).

Figure 2.

A pilV null mutation suppresses transformation defects in mutants retaining partial ComP function.

A. Graph shows transformability as percentage of recipient cells. Grey bars represent the presence of PilV, whereas white bars represent the absence of PilV. Strains: 1, N400/GV1 (wild-type/pilVfs); 2, MW117/GV20 (comPG-1S /comPG-1S, pilVfs); 3, GP104/GV21 (comPE+5V /comPE+5V, pilVfs); 4, MW102/GV9 (ΔORF1::mTnerm22/ΔORF1::mTnerm22, pilVfs); 5, MW103/GV11 (comP ::mTncm64/comP ::mTncm64, pilVfs). Values used are mean ± standard error of the mean; n= 3 except for the strains MW102 and MW103 that were only assayed once, but are previously reported to be 15% and 0.02% relative to wild-type respectively. By way of reference, the transformation frequency observed for the isogenic strain GV10 lacking both PilV and ComP (pilVfs, comP ::mTnerm23) was less than 10−6.

B. Coomassie-stained SDS-PAGE gel showing the relative amounts of PilE in purified pili obtained by shear extraction. The strains used are from the left to the right the same ones as listed from 2 to 3 above. With regard to PilV expression ‘+’ denotes the wild-type allele, whereas ‘−’ denotes the pilVfs null allele.

PilV prevents accumulation of ComP in the shear extractable cell fraction

As inactivating mutations in pilV generated a phenocopy of strains overexpressing ComP with regard to transformability and DNA uptake, the status of ComP expression in the pilV backgrounds was assessed. As noted in previous studies, ComP is normally expressed at levels too low to be detected by immunoblotting of whole cell lysates or purified pilus samples (Wolfgang et al., 1999; Aas et al., 2002). Similarly, it could not be detected here in whole cell lysates from pilV strains (not shown). Remarkably however, ComP was detected in cell fractions of the pilV mutants derived using a method identical to that used to purify gonococcal Tfp (Fig. 3, lanes 3 and 5). These results were specifically a result of the absence of PilV as reintroduction of a wild-type pilV gene copy into these backgrounds resulted in the inability to detect ComP in these fractions (Fig. 3, lanes 4 and 6). PilV therefore prevents accumulation of ComP at a site in the cell where it can be readily recovered by shear force extraction. Data from this work and a previous study (Aas et al., 2002) taken together with their structural similarities suggested that PilE and ComP might interact. Accordingly, we asked if ComP localization in the shear extractable fraction was influenced by PilE and found that ComP was not recoverable in the simultaneous absence of PilE and PilV (Fig. 3, lanes 7 and 8).

Figure 3.

PilV prevents ComP accumulation in the shear extractable cell fraction enriched in Tfp. +, denotes the wild-type allele; −, denotes a null allele or in the case of PilE expression the non-induced pilEind allele; −/+, denotes complementation of a pilV null mutation. Strains: 1, N400 (wild-type); 2, MW104 (comP::Tnerm23); 3, GV1 (pilVfs); 4, GV5 (pilVfs, iga::pilV ); 5, GV2 (pilV::kan); 6, GV3 (pilV::kan, iga::pilV ); 7, GV1 (pilVfs); 8, GV12 (pilVfs, pilEind).

Upper panel. Coomassie-stained SDS-PAGE gel showing the relative amounts of PilE in purified pilus samples obtained by shear extraction.

Lower panel. Immunoblotting of purified pilus samples using rabbit antibodies specific for ComP.

Genetic interactions of pilV with ComP overexpression: the influences of PilE and PilT

Given the common phenotypes associated with pilV mutations and ComP overexpression along with the evidence for interactions between pilV and comP, it was of interest to see what effects the combination of the absence of PilV and ComP overexpression might have on the system. Attempts at constructing these strains in an otherwise wild-type background were unsuccessful as no transformants could be recovered by selection for antibiotic resistance marker linked forms of the comPoe allele (which overexpresses ComP) in a pilV null mutant or of a pilV null allele into a comPoe background. We also attempted to cross a pilV frameshift mutation into a comPoe background by virtue of its linkage to an antibiotic resistance gene tagged transposon insertion less than 700 bp away. Although ComP overexpressing transformants with the transposon insertion were recovered, none were found which had crossed in the linked pilV point mutation. It was formally possible that the inability to recover mutants with both alleles was caused by a defect in the capacity of such double-mutants to grow in the presence of antibiotics even though they had in fact picked up the altered alleles. To assess this possibility, we tried transforming in the unmarked pilV frameshift mutation into the strains under conditions in which very high levels of transformation are achievable (Gunn and Stein, 1996). Whereas greater than 50% of the non-selected colonies arising from wild-type cells exposed to the DNA had acquired the defective pilV allele, we were unable to find any in the ComP overexpressing background which had picked up the mutation (not shown).

To examine the parameters contributing to these negative results, the influences of pilE and pilT to the process were assessed. Remarkably, strains simultaneously overexpressing ComP but lacking PilV could be recovered in backgrounds conditionally failing to express either the PilE subunit or the PilT retraction protein. Together, these findings suggest that the combination of ComP overexpression together with the absence of PilV creates a severe growth defect which involves the simultaneous expression of both PilE and PilT.

Genetic interactions of pilV with ComP overexpression: effects on transformability and DNA binding

The strains used in constructing the ComP overexpressing, pilV backgrounds utilized transcriptional fusions of lac promoter and operator sequences to pilE and pilT such that both tight negative regulation and derepression were feasible. The derepressible nature of the pilEind and pilTind alleles was exploited to examine the interactions of comPoe and pilV on transformation related processes. In the absence of derepression, both strains were non-transformable showing that the comPoe/pilV combination does not bypass the requirement for either PilE or PilT (not shown). When maximally derepressed, both the pilEind and pilTind strains were viable although they grew slightly slower than seen under non-inducing conditions. The differences between these findings and the observations made for the wild-type background likely reflect the fact that the levels of expression achievable from the derepressible constructs are significantly lower than that of the native promoters (Rudel et al., 1995; Wolfgang et al., 1998a). Derepression in the comPoe/pilEind/pilV background led to transformation frequencies approaching 100% but as the values seen for the pilV and comPoe alterations alone in the pilEind background were close to 60%, it was difficult to demonstrate the significance of the effect in the comPoe/pilV strain. De-repression in the pilTind background together with the comPoe/pilV combination led to transformation frequencies significantly higher than that seen for the pilT background in conjunction with either comPoe or pilV alone (Table 2, pre-induced values). These data clearly demonstrate synthetic enhancement between comPoe and pilV. The same basic effect was seen when pilT derepression was carried out only after cells had been exposed to transforming DNA and then washed extensively to remove any DNA not bound to the cell surface (Table 2, post-induced values) (Aas et al., 2002). This result strongly suggested that increased transformability is a consequence of increased DNA binding to the cell surface. We therefore tested the effects of a pilV null mutation on DNA association with cells under conditions where DNA uptake was blocked (Table 2). Unlike the case for comPoe, the absence of PilV did not lead to statistically significant enhanced DNA binding in the absence of uptake. In combination with ComP overexpression however, the absence of PilV led to an approximately 35-fold increase in sequence-specific DNA binding. Because this value significantly exceeded the sum of the effects of pilV and comPoe individually, the co-operative interactions were clearly synergistic. The comPoe/pilV combination led to synergistic enhancement of non-specific DNA binding as well which was significantly lower than that seen for the DUS containing substrate. To address if the comPoe/pilV combination had rendered the uptake system less stringent or enhanced a non-specific DNA binding activity unrelated to transformation, we examined whether the DUS lacking substrate was taken up into the cell when pilT was derepressed. Although the DUS containing substrate was readily taken up into a DNase I resistant state, the DUS lacking substrate was not (not shown), arguing that the altered DUS independent DNA binding activity did not lead to entry into the transformation pathway.

Table 2. . Interactions of pilV with ComP overexpression.
Relevant genotypeTransformation freq. (%) aDNA bindingb
prepostDUS+DUS−
  • a

    . Transformation frequency as percentage of recipient cells. Carried out with mutants made in a wild-type recombination background carrying an inducible pilT allele (strains: GT104, GV18, GP120 and GV19, see Table 3). pre, denotes induction of pilT before incubation with DNA; post, denotes induction after washing to remove unassociated DNA.

  • b

    . DNA binding as percentage of DNA added. Carried out with mutants made in a recA6 background carrying a pilT null-mutation (strains: GT107, GV8, GP118 and GV15, see Table 3).

  • Values represent the mean ± standard deviation of at least three individual experiments.

  • ND, not determined.

pilT 0.41 ± 0.170.01 ± 0.0010.08 ± 0.040.05 ± 0.04
pilT, pilV fs  7.0 ± 1.2 2.8 ± 1.50.11 ± 0.04ND
pilT, comP oe  9.1 ± 1.8 4.3 ± 2.20.19 ± 0.020.01 ± 0.01
pilT, pilV fs , comP oe 46.8 ± 30.819.7 ± 10.82.80 ± 0.200.45 ± 0.05

To investigate the mechanisms behind the co-operative effects of pilV and comPoe, the levels of ComP expression were assessed in these backgrounds (Fig. 4). As seen in the wild-type background, the absence of PilV in the pilT background led to ComP in the surface shear preparation fraction. Likewise, the absence of PilV in the comPoe background enhanced ComP levels in the surface preparation fraction and the co-operative interactions of these alleles appeared to be synergistic in this property.

Figure 4.

PilV inhibits ComP accumulation in the shear extractable cell fraction when ComP is overexpressed. +, denotes the wild-type allele; ++, denotes the ComP overexpressing allele (comPoe); −, denotes the pilVfs null allele. Strains: 1, MW4 (pilTind); 2, GV6 (pilTind, pilVfs); 3, GP119 (pilTind, iga::comPoe); 4, GV14 (pilTind, iga::comPoe, pilVfs).

Upper panel. Coomassie-stained SDS-PAGE gel showing the relative amounts of PilE in purified pili obtained by shear extraction.

Lower panel. Immunoblotting of purified pilus samples using rabbit antibodies specific for ComP.

Discussion

Graupner and Wackernagel (2001) were the first to show in their studies of P. stutzeri that transformability could be governed by the antagonistic action of a Tfp pilin subunit-like protein. Here, we have likewise shown that PilV acts as an intrinsic inhibitor of genetic transformation in N. gonorrhoeae. In contrast to the P. stutzeri PilAII situation however, PilV acts ultimately by quantitatively reducing DNA uptake. Furthermore, genetic and biochemical evidence strongly suggests that PilV exerts its effect by disrupting ComP function and thus antagonizing ComP-dependent processes. As recent studies have shown that the role of ComP in DNA uptake can most directly be accounted for by its ability to promote sequence-specific DNA binding (Aas et al., 2002), PilV presumably impacts indirectly on the system at this same step. However, the absence of PilV did not enhance sequence-specific DNA binding in the absence of uptake as was shown for strains overexpressing ComP (Table 2). This may be because the level of increase in this background is too low to be detected by the assay used. This explanation is supported by the finding that when a pilV mutation was placed into a ComP overexpressing strain, a dramatic increase in sequence-specific DNA binding in the absence of uptake was observed (Table 2).

Cumulatively, the evidence favours a model in which PilV impinges on the system by reducing the levels of functional ComP. As levels of comP mRNA are not elevated by pilV mutations in any of the backgrounds described here (not shown), PilV acts in a post-transcriptional manner to inhibit ComP accumulation. These effects could be exerted by either directly targeting ComP to a degradative pathway or preventing ComP trafficking to its active site in the cell such that it undergoes default degradation. Although these models are not mutually exclusive, the data that ComP in pilV mutants can be detected specifically in the surface shear fraction favours the latter notion. How might PilV then alter ComP trafficking? Given their shared, highly conserved N-terminal domains and the abundant genetic evidence that they together interact with the Tfp biogenesis machinery and its constituents, one plausible scenario would be that PilV competes with ComP for a common interactive component or translocative site with limiting abundance. Alternatively, they may form mixed, non-functional multimers which titrate out ComP from the system. It is also possible that PilV may indirectly affect ComP turnover or trafficking by altering some other component of Tfp or the Tfp biogenesis machinery. The ideas that PilV and ComP interact with one another or a common second partner are strengthened by their structural relatedness to one another. They share 37% identity encompassing mature residues 37–80 (of PilV) in addition to their conserved, N-terminal domains (Fig. 5). Database searching using gonococcal PilV and ComP or the orthologues found in the two N. meningitidis genomes available reveals that these two molecules are more closely related to one another than they are to any other proteins (not shown). These data appear to indicate that PilV and ComP have a common genetic origin and that they have divergently evolved to carry out their unique roles in Tfp biology.

Figure 5.

Extended structural identity and similiarity between PilV and ComP. A region extending more than 50 residues downstream of the N-terminal part conserved among type IV prepilins and other GspG family members is shown. +, denotes conserved residues. The N-terminal consensus sequence shared by type IV prepilins is shown beneath the ComP sequence. An arrow indicates the PilD prepilin processing site. Alignment was performed using the clustalw alignment tool (Thompson et al., 1994).

We were unable to recover strains overexpressing ComP and lacking PilV in a wild-type background. Although it was not possible to demonstrate conclusively that this gene combination results in synthetic lethality, the overall data favour this interpretation. PilV can then be interpreted as allowing the system to tolerate elevated levels of ComP in the context of wild-type expression levels of PilE and PilT and this ameliorating effect can most simply be accounted for by the ability of PilV to prevent ComP accumulation. Similarly, the contribution of PilE to the defect is most readily attributable to its ability to promote ComP accumulation. The mechanism by which PilT might be involved in these processes is unclear. As toxicity appears to reflect the high level accumulation of ComP in a specific cell compartment, we speculate nonetheless that PilT also influences ComP trafficking or stability. This effect might relate directly to Tfp retraction as fibre disassembly would presumably lead to increased steady state levels of PilE in the membrane and it is clear that PilE contributes quantitatively to the process.

Pilin related molecules (also termed pseudopilins or minor pilins) are known to affect Tfp biogenesis in a number of systems (Mattick, 2002). However, the observation here that pilin-related molecules play overlapping and partially redundant roles in organelle biogenesis is novel. Moreover, the findings here are particularly unique in that the biogenesis defect in the double-mutants can be defined to the specific step in the pathway which involves the dynamics of fibre growth and retraction. The finding that PilV and ComP contribute to biogenesis also raises some uncertainty as to how they might contribute to Tfp-associated phenotypes. For example, single mutants lacking either one may in fact have subtle biogenesis defects undetectable by current methodologies which might account for the altered phenotypes. This could be particularly relevant to the case of pilV null mutants as normal Tfp expression in this background is solely reliant on ComP which is expressed at low levels. However, pilV strains bearing pilT null mutations (which override the need for ComP in biogenesis in a pilV background) remain defective in epithelial cell adherence (Winther-Larsen et al., 2001). It is therefore not possible to attribute defects in adherence associated with pilV mutations to reduced levels of Tfp expression.

Whereas the results here underscore the importance of ComP to gonococcal DNA binding and uptake, they do not demonstrate how ComP functions in the process. The data does, nonetheless, strongly suggest that ComP acts by virtue of its trafficking to a site in the cell where it can be released by physical shear force, because its accumulation in this fraction correlates extremely well with enhanced transformability and sequence-specific DNA binding. In addition, ComP recovery in this fraction requires the PilE pilus subunit. To our knowledge, this is the first time in any system that the presence of a Tfp pilin subunit or prepilin-like protein has been demonstrated to influence the expression pattern of a transformation component. Whether the co-localization of ComP to a fraction enriched in Tfp and the role of PilE in the process reflects the formation of a ComP/PilE heteropolymer or ComP indirectly parasitizes the Tfp biogenesis machinery remains to be seen. Efforts to address the former scenario by electron microscopic immunodetection are currently hampered by the low levels of ComP in the purified pilus samples (unpublished data). In summary, we characterized a unique set of interactions between comP and pilV whose products act, respectively, as positive and negative effectors of DNA binding and uptake. Whereas the gonococcal transformation system defined in vitro appears to be constitutively expressed, these findings raise the possibility that alterations in the relative levels of ComP and PilV in vivo might provide a novel mechanism through which competence could be modulated.

Experimental procedures

Strains, plasmids and mutants

The bacterial strains used in this study are described in Table 3. Escherichia coli and gonococcal strains were grown as described (Freitag et al., 1995). Escherichia coli strain HB101 was used for plasmid propagation and cloning experiments. For DNA uptake/binding experiments, the radiolabelled substrates used were pHSS6 (Seifert et al., 1990) and pUP6, a derivative of pHSS6 that carries two gonococcal DNA uptake sequences (Wolfgang et al., 1999).

Table 3. .N. gonorrhoeae strains used in this study.
StrainParent strainRelevant genotypeReference
  • a

    . VD300 is an Opa derivative of MS11.

  • b

    . recA6 is an IPTG-inducible allele of recA.

  • c

    . Transposon insertion mutation in the comP promoter region (Wolfgang et al., 1999).

  • d

    . comPoe (comPoverexpressed) is the new designation of the former pilE::comP allele (Wolfgang et al., 1999).

  • e

    . pilV fs is a null-allele of pilV with a frame-shift mutation at the G−1 codon (Winther-Larsen et al., 2001).

VD300aMS11  Koomey and Falkow (1987)
N400VD300 recA6(tetM)b Tønjum et al. (1995)
N401N400 recA6(kan) Wolfgang et al. (1998b)
GT104VD300 pilT ind Wolfgang et al. (1998a)
MW4N401 pilT ind Wolfgang et al. (1998b)
MW24N401 pilE ind Wolfgang et al. (2000)
GT17N400 pilT::mTncm17 Park et al. (2002)
MW102N400ΔORF1::mTnerm22c Wolfgang et al. (1999)
MW103N400 comP::mTncm64c Wolfgang et al. (1999)
MW104N400 comP::mTnerm23 Wolfgang et al. (1999)
MW117N400 comP G-1S Wolfgang et al. (1999)
GP104N400 comP E+5V Aas et al. (2002)
MW120N400 comP oe d Wolfgang et al. (1999)
GP117N400 iga::comPoe Aas et al. (2002)
GP118GP117 iga::comPoe, pilT::mTncm17This study
GP119GP117 iga::comPoe, pilTindThis study
GP120GT104 iga::comPoe, pilTindThis study
GV1N400 pilV fs e Winther-Larsen et al. (2001)
GV5GV1 pilV fs , iga::pilV Winther-Larsen et al. (2001)
GV6MW4 pilV fs , pilTind Winther-Larsen et al. (2001)
GV8GV1 pilV fs , pilT::mTncm17This study
GV9GV1 pilV fs , ΔORF1::mTnerm22This study
GV10GV1 pilV fs , comP::mTnerm23This study
GV11GV1 pilV fs , comP::mTncm64This study
GV12MW24 pilV fs , pilEindThis study
GV13GV6 pilV fs , pilTind, comP::mTnerm23This study
GV14GV6 pilV fs , pilTind, iga::comPoeThis study
GV15GV14 pilV fs , pilT::cm, iga::comPoeThis study
GV16VD300 pilV fs This study
GV17GV16 pilV fs , iga::pilVThis study
GV18GV16 pilV fs , pilTindThis study
GV19GV18 pilV fs , pilTind, iga::comPoeThis study
GV20MW117 pilV fs , comPG−1SThis study
GV21GP104 pilV fs , comPE+5VThis study
GV2N400 pilV::kan Winther-Larsen et al. (2001)
GV3GV2 pilV::kan, iga::pilV Winther-Larsen et al. (2001)
GV22VD300 pilV::kanThis study
GV23GV22 pilV::kan, iga::pilVThis study
GV24GV22 pilV::kan, comP::mTnerm23This study
GV25GV22 pilV::kan, pilTindThis study
GV26GV25 pilV::kan, pilTind, comP::mTnerm23This study
GV27N400 iga::pilVoeThis study
GV28GV27 iga::pilVoe, comPoeThis study
GV29VD300 iga::pilVoeThis study
GV30GV29 iga::pilVoe, comPoeThis study

Construction of PilV overexpressing strains

A plasmid construct overexpressing PilV was generated by fusing the transcriptional and translational signals present in the highly expressed pilE gene to the PilV ORF employing a method previously used to overexpress ComP (Wolfgang et al., 1999). The 5′-end of pilE was amplified by PCR using the primers pilE5′ (Bam) (5′-CTAGGATCCGACCCAATCAA CACAC-3′) and pilE-pilV3′ (5′-CAGCTCGAGAAGGGTAAA GCCTTTTTGAAGGGTAT-3′). The product was then cleaved with BamHI and XhoI (underlined sequences) and was used to replace the BamHI-XhoI fragment containing the pilV 5′ region of pPilV1 (Winther-Larsen et al., 2001) to create pPilV4. The resulting construct expresses a translational fusion of PilE to PilV at residue L+3 such that the mature (PilD processed) protein was identical to that derived from the wild-type pilV allele, and this allele was designated pilVoe(oe for overexpression). Gonococcal strains with pilVoe were generated by first cloning the gene fusion into pPilE2 (Aas et al., 2002). This was achieved by digesting pPilV4 DNA with BspE1, a fill in reaction with Klenow fragment to create blunt ends, and digestion with Bsu36I. The Bsu36I-blunted fragment was then used to replace the corresponding Bsu36I-StuI fragment of pPilE2. The DNA from the resulting plasmid pPilEV was used to introduce pilVoe into the iga locus of strains N400 and VD300 by transformation and selection for the linked ermC marker. Direct DNA sequencing of PCR products derived from the gonococcal transformants was done using custom primers at GATC Biotech AG (Konstanz, Germany) to insure the correct introduction of the alleles and the absence of any other alterations.

Transformation assays

The transformation assays were carried using 1 µg ml−1 of plasmid pSY6 DNA (Stein et al., 1991) mixed with 0.5 ml of cells (5 × 107 ml−1), supplemented with 7 mM MgCl2, and incubated for 30 min (37°C, 5% CO2). After the incubation the samples were diluted 10 × in Gc broth and grown for 3 h before appropriate dilutions were plated onto agar medium with and without 1 µg ml−1 of nalidixic acid. The conditional transformation assays used to assess functional DNA binding utilized strains in which pilE or pilT could be rapidly derepressed, and it was performed as the transformation assay with modifications. Cells were incubated with 1 µg ml−1 pSY6 DNA for 15 min, centrifuged and washed five times in Gc broth and then grown in the presence of 250 µM IPTG for 4 h to release repression of pilE or pilT (Aas et al., 2002). These results were compared to those in which cells were grown under derepressing conditions overnight before DNA exposure and washing.

DNA binding/uptake assay

DNA binding and uptake was measured using 32P-labelled plasmid DNAs from pHSS6 (DUS–) and pUP6 (DUS+). The plasmid DNAs were linearized by NheI, treated with ExoIII and a fill in reaction with Klenow Exo was performed in the presence of 25 µM dGTP/dTTP and 1 µM [α-32P]-dATP/dCTP (3000 Ci mmol−1, Amersham Pharmacia Biotech). Unincorporated nucleotides were removed using QIAquick mini columns (QIAGEN), and the resulting product had a specific activity of 4.0 × 107 cpm µg−1. Colonies of bacteria were resuspended from Gc agar plates (20 h, 37°C, 5% CO2) to 1 ml suspensions containing 5 × 108 colony forming units (cfu), supplemented with 7 mM MgCl2, mixed with 500 ng of 32P-labelled DNA, and tumbled for 30 min at 37°C. The samples were then split into two 0.5 ml fractions. One of these 0.5 ml fractions received 100 µg ml−1 DNase I (Roche) for 10 min at RT to measure DNase I resistant uptake, while the other was placed on ice. Subsequently, both samples where washed by centrifugation, three times in 1 ml of ice-cold Gc broth. The samples were counted in 4 ml of liquid scintillation cocktail in a liquid scintillation analyzer (Packard 1600 TR).

Tfp purification and quantification, SDS-PAGE and immunoblotting

Pilus purification was carried out as described (Wolfgang et al., 1998a). Relative quantification was achieved by comparison of the PilE band intensity of the mutants to that seen for serial dilutions of Tfp preparations from the wild-type strain after samples were first standardized based on amounts of whole cells from which the preparations were made (Winther-Larsen et al., 2001). Procedures for SDS-PAGE, Coomassie staining and immunoblotting have been described previously (Freitag et al., 1995). PilV, ComP and PilE were detected by immunoblotting of purified pili and whole cell lysates using rabbit polyclonal antibodies and alkaline phosphatase coupled goat anti-rabbit antibodies (Tago). PilE-, ComP- and PilV-specific sera have been described previously (Drake and Koomey, 1995; Wolfgang et al., 1999; Winther-Larsen et al., 2001).

Acknowledgements

This work was supported by funds from the Norwegian Research Council (to F.E.A. and M.K.) and in part by a subcontract of PHS Grant AI27837, National Institutes of Health (to M.K.).

Ancillary