H. influenzae is a small gram-negative bacterium frequently commensal in the human upper respiratory tract, and a common cause of serious infections. Serotype b strains are the major cause of meningitis in infants and small children, with a 6% mortality rate and residual damage to hearing or intellect in about 50% of cases (40). A conjugated anti-type b vaccine effective in infants has recently become available, but there is no vaccine for other serotypes, which are a major cause of childhood ear infections and of respiratory disease in infants and elderly or immunocompromised individuals, especially people with AIDS (36). Competence and transformation have medical relevance at several levels. Multi-drug-resistant H. influenzae are increasing in Canada and elsewhere (18,48), and antibiotic resistance genes, virulence determinants and genes specifying the different capsular serotypes can be spread by transformation (24,25). Our working hypothesis is that competence is part of a suite of adaptations to a mucosal environment, allowing cells to use the abundant host DNA in respiratory tract mucus as a source of nucleotides (28). Competence is regulated by growth conditions, so dissecting the regulatory signals will clarify the control of fundamental metabolic processes in this pathogen.
Overview of competence and transformation in H. influenzae: (Figure 1 summarizes the steps involved and our knowledge of the genes responsible): Competent cells bind up to 300kb of double-stranded DNA, preferentially binding fragments carrying the uptake signal sequence AAGTGCGGT (the 'core USS'), which is highly over-represented in the H. influenzae genome (46). DNA uptake is a two-stage process; fragments or plasmids are first transferred intact into the periplasmic space (or possibly into a membrane-bound vesicle called the transformasome) where the DNA is inaccessible to both extracellular DNases and cytoplasmic restriction endonucleases (8). During the subsequent translocation of DNA into the cytoplasm one strand is degraded, and the resulting nucleotides are rapidly reused for new DNA synthesis (circular molecules are not translocated) (38). The remaining strand may recombine into any homologous segment of the chromosome (the resident strand is displaced and degraded); otherwise it too is degraded and reused (38). A cell whose genotype is changed by this recombination is said to be transformed.
1. Genetic recombination: Most biologists have assumed that competence evolved to allow transformation to create recombinant genomes. However, research into the evolution of sex has found the benefits of recombination to be elusive (34), and our theoretical work extending this analysis to transformation reveals significant costs (41). Most hypotheses about the benefits of recombination can only be tested experimentally in bulk cultures and in chemostats, conditions highly prone to cryptic selection and other experimental artefacts.
2. Recombinational DNA Repair: Competent cells may use some of the DNA they take up as templates for recombinational repair of otherwise-irreparable DNA damage (35). Such physical lesions are very frequent, and lethal if unrepaired, so use of incoming DNA for repair is potentially more beneficial than recombination. However, uptake of homologous DNA does not detectably increase H. influenzae's ability to survive DNA damage (4), and competent cells are actually less able to repair UV-induced DNA lesions than are normal cells (10,44) .
3. Nucleotide Salvage from DNA: One reliable benefit of DNA uptake is the nucleotides released when the DNA or the strand it has displaced is degraded inside the cell. H. influenzae has the genes to synthesize its own nucleotides (pyrimidine synthesis requires citrulline, explained by the absence of a carAB homolog) (17), but calculations done for E. coli indicate that this will consume about 16% of the cell's total energy budget (47). It is thus not surprising that H. influenzae makes efficient use of DNA and RNA precursors provided in the culture medium. However, although its natural environment is very rich in DNA (>=300ug/ml) (28), it does not secrete nucleases that would hydrolyze the DNA for uptake by standard salvage pathways (23). We speculate that this is because the competence pathway allows external DNA to be hydrolyzed to dNMPs inside the cell and immediately reused, bypassing the degradation to bases plus deoxyribose required when dNMPs are generated externally and then imported. Uptake of DNA also avoids extracellular losses by diffusion of nucleases or nucleotides. Use of DNA as a source of carbon, nitrogen, or energy makes economic sense only if it is taken up in excess of the cell's need for nucleotides.
Costs of competence: Competent cells must undergo the expense of synthesizing the DNA uptake machinery, and suffer increased sensitivity to DNA damage (10,44). In the natural environment most recombination events are likely to be either selectively neutral or harmful, and most available DNA is of poor genetic quality. In particular, uptake of phage or prophage DNA can kill cells (12), and the H. influenzae phage HP1 has a high density of the preferred uptake signal sequences (16). Furthermore, DNA uptake (especially of heterologous DNA) often induces the SOS response, which causes cell death by inducing a prophage that is resident in the H. influenzae genome (45).
Competence is tightly regulated (43); the regulatory effects likely evolved to reflect the shifting balance of benefits and costs associated with DNA uptake. Each of the above roles for competence makes predictions about regulation. The simplest, made by the DNA repair hypothesis, is that competence should be inducible by DNA damage. However our experiments in H. influenzae and B. subtilis have shown that this is not the case, although both species have damage-inducible SOS responses, and in both competence is inducible by nutritional limitation (3). Thus any benefits from DNA repair are probably minor. If the primary benefit is acquisition of nucleotides by DNA salvage, regulation would be expected to tie competence to both energy resources and the availability of free nucleotides or bases in the growth environment. Under this scheme, nucleotides or bases would be taken up when available, and, if this uptake were insufficient, synthesized de novo when energy resources permitted and salvaged from extracellular DNA when energy resources were limiting. Because the benefits of recombination are still controversial, regulatory predictions are highly speculative. If competence's main role is recombination, it might be subject to multiple regulatory systems, causing it to be induced only when other regulatory mechanisms are unable to maintain growth or survival.
In the laboratory, H. influenzae cells are usually grown in brain-heart infusion broth supplemented with hemin and NAD (sBHI) (9), and competence is induced by transfer of exponentially growing cells to a starvation medium (MIV) which contains amino acids but no purines, pyrimidines or cofactors (21). The competence of a culture can be directly assayed by measuring the uptake of radiolabelled DNA, but a more sensitive indicator is the transformation frequency (TF), the fraction of the cells that become transformed by DNA carrying a chromosomal antibiotic resistance allele or other marker. When a fully competent culture is mixed with a saturating amount (about 200ng) of a cloned marked DNA fragment, most cells will take up several fragments and about one in five will become transformed (TF~0.2). It is usually more convenient to measure transformation with total genomic DNA, which gives TF of about 10E-2. This should be compared with the TF of <=10E-8 exhibited by non-competent cultures (43);.
Regulation by cAMP, CRP and the PTS: Induction of competence absolutely requires the transcription factor cyclic AMP regulatory protein (CRP) and its cofactor cyclic AMP (cAMP) (1, 15); these are well-characterized signals of catabolite limitation in many bacteria (13). In E. coli, high intracellular cAMP concentration allows CRP to bind to and activate over 100 promoters controlling chemotaxis to and use of sugars, amino acids, nucleotides and bases. Our experiments support similar roles for cAMP and CRP in H. influenzae, as mutations eliminating either prevent fermentation of most sugars (1, 2) (assayed using phenol red indicator broth; see Table 1 in (2)). The role of CRP in H. influenzae catabolite regulation is supported by the newly-available complete sequence of the H. influenzae genome (17). As in E. coli, a consensus CRP binding site precedes the putative promoters of genes controlling galactose, ribose and xylose utilization (galT (HI#0820), rbsD (HI#0501) and xylH (HI#1109) respectively). [Each H. influenzae gene has been assigned a 'HI' number.] Addition of cAMP also induces competence in exponentially growing cultures, although only about 1% of the cells become competent (1, 52). Control of competence by cAMP and CRP therefore implies an energy-sparing role for the DNA cells take up, and directs attention to the processes that regulate cAMP and CRP.
In E. coli the activity of CRP is regulated primarily through changes in intracellular cAMP concentration. This in turn depends mainly on the activity of adenylate cyclase, itself regulated by the availability of sugars to the phosphoenoylpyruvate:glycose phosphotransferase system (the PTS) (39). Other potential controls on cAMP have not to date been implicated in catabolite regulation and may serve only to maintain basal levels of cAMP; these include breakdown and excretion of cAMP and repression of the cya (HI#0604) and crp (HI#0957) genes by CRP (13).
The PTS has been best characterized in E. coli, where it has two roles relevant here (39). First, it uses group translocation to drive uptake of PTS-transported sugars such as glucose and fructose, channeling high-energy phosphate to the sugar via the general PTS proteins EI and HPr and one or more sugar-specific PTS proteins. Second, the PTS activates adenylate cyclase when no PTS sugar is available, raising intracellular cAMP and thus permitting the CRP-dependent expression of genes for use of less-favoured sugars such as lactose and galactose (transported by non-PTS permeases). This activation is thought to be mediated by the glucose-specific PTS protein EIIA-Glc in its phosphorylated state. PTS proteins specific for the other PTS sugars also activate adenylate cyclase when their substrate is absent, by shifting the balance between phosphorylated and nonphosphorylated EIIA-Glc) .
The H. influenzae PTS appears to be much simpler than its E. coli counterpart; only five PTS components have been identified in the genome sequence (17) (Figure 2 ). Fructose is the only sugar for which all PTS-transport components are present; this agrees with our finding that only fructose fermentation is prevented by a ptsI null mutation (EI, HI#1712) when cAMP is provided (2). A glucose permease gene (EIIBC-Glc) has not been found; it is unlikely to have been overlooked because the other PTS proteins showed 68-88% similarity to enteric homologs. EIIA-Glc) (product of crr, HI#1711) is present and, despite the absence of its permease, probably regulates the activity of adenylate cyclase in H. influenzae as it does in E. coli. Our ptsI null mutant is unable to carry out processes dependent on cAMP (both sugar fermentation and competence) (2), and these defects are corrected by exogenously-supplied cAMP. Furthermore, the E. coli cya gene can restore normal competence regulation to a H. influenzae cya mutant, suggesting that adenylate cyclase is regulated similarly in the two organisms. (animated model of the H. influenzae PTS (note: big file, ~200K).
One major role of cAMP in competence, possibly its only role, is to permit transcription of the competence activator gene sxy (HI#0601) (5).(also called tfoX (56)). We first identified this gene as the site of the sxy-1 mutation described below (43). It encodes a 19kD basic protein with no evident processing signals or transmembrane segments and no homology to known proteins. A null mutation (sxy::Tn) eliminates all competence (5) and reduces competence-induced transcription of the DNA-uptake genes com101A and dpr A ((56) and J.-F. Tomb, personal communication). The sxy gene is preceded by a consensus CRP-binding sequence, and its transcription is reduced in a cya null mutant and stimulated in wildtype cells by cAMP and by transfer to competence-inducing conditions, regulatory effects consistent with an essential role as an activator of competence-specific transcription.
A second mode of regulation acts at sxy: Analysis of sxy-1 and other sxy mutations causing hypercompetence supports the long-standing speculation that competence induction requires a second, cAMP-independent, signal (53) and provides clues to the nature of this signal and its mode of action. The need for a second signal is typical for processes regulated by cAMP and CRP, and is consistent with the failure of exogenous cAMP to induce full competence.
We have now isolated 5 distinct point mutations in sxy, by selecting for hypercompetent cells. All cause similar hypercompetence phenotypes, and all are single-base substitutions that potentially disrupt secondary structure in sxy mRNA. The best characterized of these mutations is sxy- 1, which causes moderate competence (TF~10E-5) to develop under conditions that are noninducing for wildtype cells, and full competence under conditions that normally cause only partial competence (TF~10E-2) (43). Although elevated, competence remains absolutely dependent on cAMP and CRP. The sxy-1 mutation increases DNA uptake and transformation, but does not affect growth or DNA repair, consistent with the hypothesis that sxy is a competence-specific regulatory gene (43). It also increases transcription of com101A [J.-F. Tomb, personal communication].
The sxy-1 mutation is a G-A transition causing a Val-Ile substitution in codon 19 of the Sxy ORF (5). This very minor alteration must be responsible for the hypercompetence because an identical mutation created by site-directed mutagenesis gives the same phenotype (5). Three other hypercompetence mutations (sxy-2, -3, and -4) do not alter the predicted Sxy protein at all: sxy-2 is a silent substitution in the 3rd position of codon 17, and sxy-3 and sxy-4 are point mutations at adjacent sites in the 5'-untranslated region, 37 and 38 bases upstream of the AUG. The identity of each has been confirmed by site-directed mutagenesis. These mutations could not exert their effects by changing the protein, but analysis of possible secondary structure of the sxy mRNA showed that each mutation weakens potential base pairing between two short segments of the mRNA, one in the 5'-untranslated region (sites of mutations 3, 4) and one in the coding region (sites of mutations 1 and 2) (illustrated in Figure 3 ). We have now isolated four additional hypercompetent strains carrying sxy mutations that would disrupt the same stem; two are identical to sxy-1 and sxy-2, the other two define a new allele, sxy-5.
We speculate that the RNA secondary structure suggested by these mutations also regulates competence in wild type cells, acting independently of CRP to further limit sxy expression except when nucleotide pools are depleted. Precedents in other bacteria establish how mRNA secondary structure may link changes in nucleotide pools to appropriate changes in gene expression. Availability of CTP influences the precise transcriptional start site of the E. coli pyrC gene, in turn permitting or preventing formation of a short stem that will block the ribosome binding site (30). The E. coli pyrBI operon encodes aspartate transcarbamylase (the first step in pyrimidine biosynthesis); its transcription is regulated 50-fold by a UTP-sensitive attenuation mechanism. Low UTP causes pyrBI transcription to pause, allowing transcription and translation to become tightly coupled and thus preventing formation of the attenuation stem (29). In B. subtilis, pyrimidine biosynthesis is controlled by three transcription attenuator sites active only when pyrimidines are abundant (49). In H. influenzae we have found that the sxy-1 mutation increases the amount of sxy transcript several-fold, suggesting that the ability to form the postulated secondary structure may normally limit sxy transcription.
These results suggest the following simple model for the regulation of competence. The first step is a rise in intracellular cAMP, caused at least in part by PTS-dependent activation of adenylate cyclase when fructose is absent. In the presence of cAMP CRP allows initiation of sxy transcription, but Sxy protein will not be expressed unless the second signal is also present. This specific signal is depletion of nucleotide pools, which, by limiting or preventing the formation of secondary structure in sxy mRNA, allows transcription to produce functional mRNA. The Sxy protein, in turn, activates transcription of DNA uptake genes, and the ensuing DNA uptake relieves pressure on nucleotide pools.
New gene(s) affecting competence:
We have isolated hypercompetent mutant strains whose mutations are not in or near the sxy gene and whose phenotypes are inconsistent with mutations in known competence genes. One of these, RR753, confers the same phenotype as sxy-1; the others (RR749-752) cause an even more extreme hypercompetence, with TF ~10E-3 even during exponential growth. The mutant phenotypes were efficiently backcrossed into wildtype cells by transformation, suggesting that each strain's phenotype is probably caused by a single mutation. The mutations are not linked to Str-resistance (or sxy) in transformation assays, and preliminary mapping by transposon tagging places one mutation well away from known competence genes. The other mutations were isolated independently but may affect the same gene.
Two challenges to the DNA-salvage model for competence: What we know about the regulation of competence supports the hypothesis that its primary role is to provide the cells with nucleotides. The following two sections address aspects of H. influenzae competence and transformation that are not predicted by this hypothesis, for which they therefore provide critical tests.
Other interpretations have not been investigated. One possibility is that each recombination event occurs between a replicating phage genome and a single-stranded fragment of phage DNA brought into the cell by the competence pathway. Phage HP1 resembles the P2-like phage family of E. coli (J. Scocca, personal communication), a group that recombines very poorly (22), and phage stocks typically contain substantial free phage DNA that would be taken up by competent cells. Phage recombination occurs with highly purified phage stocks, but the presence of free phage DNA was not ruled out (12). This hypothesis would be consistent with the otherwise-puzzling finding that mutations blocking DNA translocation by competent cells (rec-2, HI#0061; com101A) also prevent phage recombination (26).
Summary: Neither the uptake specificity nor the apparent induction of recombination is incompatible with the nucleotide-scavenging role we propose for competence, although neither is predicted by it. Furthermore, competence is clearly regulated by the PTS and by cAMP and CRP, a signal transduction system that in other bacteria serves to optimize the use of energy resources. The possible involvement of mRNA secondary structure suggests parallels with the regulation by nucleotide limitation seen in other bacteria, and expected from the hypothesized role. The experiments proposed below examine this role both by dissecting the regulation of competence and by critically examining evidence for the induction of recombination and the evolution of the uptake specificity.
Cassette-mutagenesis without cloning: The method is described here for fruB, and illustrated in Figure 4 . A 4-5kb fragment containing the fruB gene will be amplified, and the Cm-resistance cassette CAT19 will be ligated into it, replacing an internal fruB subfragment. The ligation mix will then be added to competent H. influenzae cells (prepared in advance and stored frozen), and the cells plated to select Cm-resistant transformants, which will contain the disrupted version of the gene. The mutant will be checked by PCR and by Southern blotting to ensure that a single copy of the disrupted gene has replaced the original version. This method uses the natural transformation system to circumvent the need to clone genes, and in principle can generate chromosomal null mutants in a single day. It is limited only by the locations of convenient restriction sites and by the length of the initially-amplified fragment, which should be at least several kb to compensate for the usual exonucleolytic degradation of the ends of transforming DNA fragments (38). However we have successfully carried out a worst-case test using a short fragment of the cya gene, with the cassette flanked by only 600bp of cya sequence on one side and 1kb on the other. Two colonies were isolated; both carried the desired gene disruption.
Two collaborations will be very helpful with the PTS analysis. We are exchanging strains and information with Milton Saier and Jonathan Reizer of UCSD, who have extensive experience in identifying genes of the PTS and assaying their function (43a). They have carried out in vivo and in vitro sugar-uptake assays on our H. influenzae Rd strain; these results will be incorporated into (2). We are also collaborating with T. Gaasterland and R. Overbeek (Argonne) and V. Selkov (Puschino, Russia), who are integrating information from the H. influenzae genome sequence into the Enzyme and Metabolic Pathway database, with the goal of reconstructing a complete picture of H. influenzae's metabolic pathways. These collaborations will be especially useful if our results do not match the prediction of our model, as they may suggest other modes of regulation by the PTS (e.g. inducer-exclusion effects, detectable as a difference in phenotype between crr and ptsHI mutants, and as effects that are independent of cAMP (33)), or involvement of other pathways.
If time permits we will also examine the effects of PTS mutations on the partial competence that develops at the end of exponential growth; this requires adenylate cyclase but is not stimulated by added cAMP, suggesting that cAMP levels do not limit competence at this time. The ptsI mutation delays this competence slightly but does not reduce it (2), suggesting that cAMP levels may also be controlled by a PTS-independent mechanism, such as the 3'-5' cAMP phosphodiesterase mentioned below.
The main difficulties are developing a suitable promoter and selecting cells that have integrated it into the chromosome. An inducible promoter would be best, but none have been characterized in H. influenzae. One option is the well-characterized CRP-independent lacUV5 promoter from E. coli (11); E. coli promoters usually work well in H. influenzae, and this one is inducible by IPTG in the presence of the lac repressor. Other promoter options are described below. Because plasmids carrying sxy are toxic and unstable in both H. influenzae and E. coli, the desired modifications will be created using a truncated version of sxy in E. coli, and recombined into the H. influenzae chromosome by transformation (Figure 5). A fragment containing lacI and the lacUV5 promoter will be amplified from E. coli chromosomal DNA and ligated into pDJM90 (5) placing the normal sxy leader sequences under PlacUV5 control (The normal leader is essential because of its postulated role in regulation.) The hybrid insert can then be transformed into cya (-) H. influenzae made competent in the presence of cAMP. Figure 5 shows a Tet (R) gene inserted between rec-1 and sxy to allow selection for the construct; we will test that an insertion here does not interfere with competence development, transformation, or DNA repair. If it does, we will try placing the sacB gene (counterselectable with sucrose in other bacteria) in the recipient chromosome and selecting for its loss. Direct selection and counterselection at sxy will be useful in other experiments, so this will be an important initial project .
To confirm that the construct is CRP-independent, RNase protection assays will be used to measure sxy mRNA in exponentially growing and competence-induced cells in the presence and absence of cAMP. Because the amount of Sxy produced will be a critical factor, sxy mRNA levels in cells with a native sxy promoter will be measured under competence-inducing conditions, and the induction of the construct by IPTG adjusted to give comparable levels. (We cannot directly assay for Sxy protein, as it has not been purified and no antibodies are available.) Competence will then be assayed by DNA uptake and by transformation. If CRP is needed only at sxy, the construct strain should become as competent as wild type cells, independent of added cAMP. Lower competence in the presence of cAMP will imply that the promoter is working poorly. Low competence only in the absence of cAMP will mean that CRP is required at another step.
An alternate approach would be to use the H. influenzae crp promoter, which is predicted to be repressed by CRP and thus induced by the absence of cAMP (15). We can also relieve the CRP-dependence of the sxy promoter itself, by changing its -10 sequence from TATAAA to TATAAT and improving the -35 sequence (S. Busby, R. Ebright, personal communication). This promoter would likely be constitutive rather than inducible (checked by RNase protection assays), but would have the advantage of retaining any other important promoter or leader sequences. CRP-independent expression of sxy is unlikely to be toxic, as fully-inducing levels of cAMP do not interfere with cell growth.
If competence remains cAMP-dependent when sxy expression is not, the second step must be sought. There is no evidence that any of the other known competence genes are regulated by CRP, but we can examine their mRNA levels in the construct strain, in the presence and absence of cAMP. The non-sxy hypercompetence mutations do not circumvent the requirement for cAMP and CRP (double-mutants are not competent), but this may be because of the known CRP dependence of sxy. Once we have identified the gene (or genes) involved, we will introduce the CRP-independent sxy construct into the cya-minus derivatives of these mutants and again test competence. If none of the known genes are found to be regulated by CRP, we can use the construct strain as a genetic background in which to select mutations relieving the CRP-dependence of the postulated step, using the same strategy used to isolate hypercompetence mutations (43).
Direct analysis of sxy mRNA secondary structure in vitro (done with G. Mackie, UBC Biochem.): This will test whether the positions identified by the mutations do pair with each other (forming the stem shown in Figure 2), and whether the sxy-1-sxy-5 mutations decrease this pairing. The 5' end of sxy will be cloned under a T7 promoter, and 5'-labeled RNA will be transcribed in vitro. After a folding step the RNA will be incubated with each of the following ribonucleases: RNase T1 (cleaves at unpaired guanosines), RNase V1 (paired or helically stacked residues), RNase T2 (unpaired residues), RNase CL3 (unpaired cytidines), aiming for an average of one cleavage per molecule. We will analyze transcripts of two cloned fragments; one just large enough to resolve the region thought to be involved in the pairing, and a longer fragment used to confirm the first analysis. We will also check the results using chemical cleavage followed by primer extension.
Do mutations that restore or increase base pairing in the proposed stem reduce competence? If pairing is reasonably strong, the in vitro analysis should reveal which regions are involved. A negative result will not be conclusive, as pairing may be real but too weak or transient to detect in vitro. The critical test for base pairing is the effect of compensatory mutations in vivo. For example, sxy-1 and sxy-3 both cause hypercompetence, and both affect the same base pair (Figure 3). We will use a variant of the cassette-mutagenesis procedure described above to introduce both mutations into the sxy leader, creating a potential A:U pair. If disruption of pairing is responsible for the single-mutant phenotypes, then the presence of both mutations in the same gene should restore pairing and reduce competence. If pairing is not involved, then the double mutant should be at least as competent as each single mutant. More generally, any mutation that increases the number of well-paired bases in the postulated stem should decrease or prevent competence development. The choice of bases that can be mutated is limited by the need to maintain the wildtype amino acid sequence; we will test the combined effects of a T-to-C transition at a site four bases to the left of the sxy-3 mutation and a G-to-A transition at the site of sxy-5, changing postulated G:U and U:G pairs to G:C and U:A pairs. The in vitro analysis may direct our attention to other regions of pairing; these can be similarly evaluated by site-directed mutagenesis.
Do fusions to sxy show position-dependent effects? If 5' secondary structure in sxy mRNA is indeed regulating competence, then lacZ fusion cassettes inserted within and outside of this region should show position-dependent expression. We will test both operon fusions and gene fusions. The operon fusions could be most easily generated in E. coli by transposition of miniTn10lac-kan into pDJM90 (a pGEM derivative carrying the 5' half of sxy and all of rec-1 (5)) and would be tested by beta-galactosidase assays in both H. influenzae (naturally Lac (-)) and E. coli. If the model is correct all fusions should be regulated by CRP and cAMP. In the presence of 1 mM cAMP gene fusions whose sxy secondary-structure region is intact (i.e. the insertion is 3' of the base-paired region) should show low expression during exponential growth and higher expression when competence is induced by transfer to MIV medium. Gene fusions whose secondary structure is disrupted (fusions within the pairing region) should give high expression independent of growth conditions. Operon fusions upstream of, or disrupting, the pairing region should be unaffected by transfer to MIV.
If the above experiments do not support involvement of the proposed secondary structure in regulation of sxy, the same approaches can be used to investigate alternative effects of the sxy-1-5 mutations. The in vitro RNA analysis may suggest other secondary structures. Involvement of antisense RNA has not been ruled out. It is also possible that alternative polypeptides are translated from the known or other transcripts; there is a 12aa open reading frame just 5' of sxy, and initiation of translation at an internal AUG has been observed in E. coli (56). The non-sxy hypercompetence mutations may identify a protein (or RNA) which acts in the sxy leader sequence.
Is in vitro production of Sxy protein by an coupled transcription-translation system higher when nucleotides are limiting? Use of an in vitro system allows the concentration of ribonucleotides to be directly controlled. The model predicts that decreasing the supply of one or more nucleotide will increase the amount of sxy or sxy-lac-fusion protein produced. The best control will be a mutant known to have reduced secondary structure. A detailed protocol for preparation and use of S30 extracts, with a discussion of potential problems, is provided by Mackie et al. (32), and we will have the assistance of the author himself.
In vivo experiments using a lacZ protein fusion: These will extend the analysis of sxy-lacZ fusions described above. We will examine expression of a fusion retaining the complete 5' region implicated in secondary structure, in both E. coli and H. influenzae. Pyrimidine effects will be examined first as these are regulatory in the known examples (29,30,49). Because H. influenzae is a pyrimidine auxotroph, pools can be manipulated by controlling the provision of precursors (citrulline or uracil) in the medium. The effects can be examined at several levels in both wildtype and sxy- 1 cells: i) transformation frequencies; ii) direct analysis of sxy transcripts; iii) expression of sxy::lacZ fusions. We can dissect it by mutating specific steps. For example, a codA mutation would let us discriminate between effects of cytosine and uracil. In order to manipulate intracellular purine levels, we need to first mutate genes involved in purine biosynthesis. Because H. influenzae is a purine prototroph, before manipulating purine pools we will mutate one or more genes in the purF...J pathway. The transcriptional regulator ppGpp slows mRNA transcriptional elongation in E. coli. (14, 50). We will use inhibitors to manipulate concentrations of ppGpp; 1-10-phenanthroline inhibits the ppGpp-ase activity of the spoT gene product, causing ppGpp levels to rise, and decoyinine depletes ppGpp by blocking GMP synthetase (guaA). Both genes are present in H. influenzae (17).
Mapping: Our plan is to first map one mutation, and to then check whether the others are closely linked. If not, we will know we have at least two genes, and will proceed to map another one. Mapping is slow because large numbers of potential recombinants must be individually tested for hypercompetence, and the genome sequence is not useful until after the mutation has been mapped. We have completed the first steps, by isolating a Tn916 insertion linked to the hypercompetence mutation in strain RR749(6% cotransformation with hypercompetence), and mapping the insertion by CHEF gel analysis to a 170 kb-segment defined by ApaI D and SmaI E restriction fragments (27). This insertion is now being used to select a more tightly linked miniTn10kan marker by reapplying the same strategy.
Several candidate genes will also be checked. The mapped mutation is linked to a homolog of the B. subtilis competence gene comE1, which has not to date been implicated in H. influenzae competence. Another possibility is that the mutations increase cAMP levels independent of the PTS, for example by inactivating a gene for a 3'-5' cAMP phosphodiesterase. The only known bacterial sequence has no obvious homolog in the H. influenzae genome We are checking the existence of such a gene by constructing a purine auxotroph, which will be tested for ability to use cAMP as a purine source. Expression of sxy in these mutants will also be examined.
We will begin with key genes representing different recombination pathways. One obvious candidate is the ATP-dependent exonuclease RecBCD, which in E. coli plays a major role in DNA processing and recombinational repair of DNA damage, especially double strand breaks. The enzyme was well characterized in H. influenzae 20 years ago. Although mutations that eliminated the nuclease activity did not dramatically reduce transformation or phage recombination (51), these mutations are now thought to be all in a single complementation group that may encode the RecD component, and thus would not eliminate most of the 'recombinase' activity (37). The recBCD analysis will be complemented by inactivation of RecO or RecR, which are required for plasmid recombination and repair of single strand gaps. Eventually Dr. Myers plans to mutate all of the genes in Table 1, as part of a larger project investigating recombination pathways in H. influenzae.
To determine whether free phage DNA can participate in phage recombination, competent cells will be infected with HP1ts2 in the presence of DNA purified from HP1ts3. To test whether the competence DNA uptake pathway is involved in this recombination, an excess of host DNA will be added to a standard mixed-infection recombination assay-this should reduce phage recombination if one of the participants uses the same uptake pathway as the host DNA. To interpret these experiments we will need to know how much phage and host DNA is normally present in lysates; this can be determined by incubating lysates with restriction enzymes and examining the amounts of digested and undigested phage DNA and the background of host DNA fragments. The complete sequence of HP1 has been determined by J. Scocca's group (Johns Hopkins) and he has provided us with a detailed restriction map. The restriction assay can then be used to monitor preparation of purified phage stocks (by centrifugation, nuclease digestion, or passage over a column of DNA-binding resin) which can be retested for recombination. These lysates can also be used in experiments asking whether host recombination enzymes become induced only as a consequence of DNA fragments entering the cytoplasm of competent cells. This will be important in assessing the induction of recombination genes described above.
A clean 'yes' would explain why we see competence-induced phage recombination, and why we only see it in transformation-proficient cells. There would then be no reason to assume that phage recombination proficiency is co-induced with competence, or that the benefits of competence require recombination.
Several approaches can be used to model the coevolution of an abundant USS and the gene for its receptor. In collaboration with H. O. Smith (Johns Hopkins), I have developed an analytical model of USS evolution, in which the abundance of the USS is set by the balance between the frequency of transformation and the bias of the receptor, both factors can be estimated by experimentation. I am also developing a simple computer simulation model in which the USS becomes abundant because it has a direct funtion in non-competent cells. Each of these models must eventually be coupled with a model addressing the evolution of the receptor gene. This latter is potentially a complex problem, and will likely be deferred until more information is available about the genes encoding the biased DNA uptake system.
1. Dorocicz, I., P. Williams and R. J. Redfield. 1993. The Haemophilus influenzae adenylate cyclase gene: cloning, sequence and essential role in competence. J. Bacteriol. 175:7142-7149.
2. Macfadyen, L. P., I. Dorocicz and R. J. Redfield. 1995. Regulation of competence development in Haemophilus influenzae Rd by a phosphoenolpyruvate:fructose phosphotransferase system.
3. Redfield, R. J. 1993. Evolution of natural transformation: testing the DNA repair hypothesis in Bacillus subtilis and Haemophilus influenzae. Genetics 133:755-761.
4. Redfield, R. J. 1993. Genes for breakfast: the have your cake and eat it too of transformation. J. Hered. 84.:400-404.
5. Williams, P. M., L. A. Bannister and R. J. Redfield. 1994. The Haemophilus influenzae sxy-1 mutation is in a newly identified gene essential for competence. J. Bacteriol. 176:6789-6794.
6. Albritton, W. L., J. K. Setlow, M. Thomas, F. Sottnek and A. G. Steigerwalt. 1984. Heterospecific transformation in the genus Haemophilus. Molecular & General Genetics 193:358-63.
7. American Genetics Association, 1992. Symposium on the Evolution of Sex. J. Hered. 84:
8. Barany, F., M. E. Kahn and H. O. Smith. 1983. Directional transport and integration of donor DNA in Haemophilus influenzae transformation. Proc.Nat. Acad. Sci. USA 80:7274-8.
9. Barcak, G. J., M. S. Chandler, R. J. Redfield and J.-F. Tomb. 1991. Genetic systems in Haemophilus influenzae. Methods in Enzymol. 204:321-342.
10. Beattie, K. L. and J. K. Setlow. 1969. Killing of Haemophilus influenzae cells by integrated ultraviolet-induced lesions from transforming deoxyribonucleic acid. Journal of Bacteriology 100:1284-8.
11. Beckwith, J. 1987. The lactose operon. 1444-1452 In F. C. Neidhardt (ed.), Escherichia coli and Salmonella typhimurium. ASM Press, Washington.
12. Boling, M. E., J. K. Setlow and D. P. Allison. 1972. Bacteriophage of Haemophilus influenzae. I. Differences between infection by whole phage, extracted phage DNA and prophage DNA extracted from lysogenic cells. Journal of Molecular Biology 63:335-48.
13. Botsford, J. L. and J. G. Harman. 1992. Cyclic AMP in prokaryotes. Microbiol. Rev. 56:100-122.
14. Bremer, H. and E. M. 1995. Guanosine tetraphosphate as a global regulator of bacterial RNA synthesis: a model involving RNA polymerase pausing and queuing. Biochim. Biophys. Acta 1262:15-36.
15. Chandler, M. S. 1992. The gene encoding cyclic AMP receptor protein is required for competence development in Haemophilus influenzae Rd. Proc. Natl. Acad. Sci. USA 89:1626-1630.
16. Fitzmaurice, W. P., R. C. Benjamin, P. C. Huang and J. J. Scocca. 1984. Characterization of recognition sites on bacteriophage HP1c1 DNA which interact with the DNA uptake system of Haemophilus influenzae Rd. Gene 31:187-96.
17. Fleischmann, R. D. and 30 others. 1995. Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science 269:496-512.
18. Goldstein, F. W. and J. F. Acar. 1995. Epidemiology of antibiotic resistance in Haemophilus influenzae. Microb. Drug Res. in press:
19. Goodgal, S. H. and M. A. Mitchell. 1990. Sequence and uptake specificity of cloned sonicated fragments of Haemophilus influenzae DNA. Journal of Bacteriology 172:5924-8.
20. Goodman, S. D. and J. J. Scocca. 1988. Identification and arrangement of the DNA sequence recognized in specific transformation of Neisseria gonorrhoeae. Proc. Natl. Acad. Sci. USA 85:6982-6.
21. Herriott, R. M., E. M. Meyer and M. Vogt. 1970. Defined nongrowth media for stage II development of competence in Haemophilus influenzae. J. Bacteriol. 101:517-24.
22. Hudnik-Plevnik, T. and G. Bertani. 1980. Recombination in bacteriophage P2: recA-dependent enhancement by ultraviolet irradiation and by transfection with mixed DNA dimers. Mol. Gen. Genet. 178:131.
23. Kilian, M. and W. Frederiksen. 1981. Identification tables for the Haemophilus-Pasteurella-Actinobacillus group. 281-290 In M. Kilian, W. Frederiksen and E. L. Biberstein (ed.), Haemophilus, Pasteurella, and Actinobacillus. Academic Press, London.
24. Kroll, J. S. and E. R. Moxon. 1990. Capsulation in distantly related strains of Haemophilus influenzae type b: genetic drift and gene transfer at the capsulation locus. Journal of Bacteriology 172:1374-9.
25. Kroll, J. S., E. R. Moxon and B. M. Loynds. 1994. Natural genetic transfer of a putative virulence-enhancing mutation to Haemophilus influenzae type a. Journal of Infectious Diseases 169:676-9.
26. Kupfer, D. M. and D. McCarthy. 1992. rec-2-dependent phage recombination in Haemophilus influenzae. J. Bacteriol. 174:4960-4966.
27. Lee, J. J., H. O. Smith and R. J. Redfield. 1989. Organization of the Haemophilus influenzae Rd genome. Journal of Bacteriology 171:3016-24.
28. Lethem, M. I., S. L. James, C. Marriott and J. F. Burke. 1990. The origin of DNA associated with mucus glycoproteins in cystic fibrosis sputum. European Respiratory Journal 3:19-23.
29. Liu, C., L. S. Heath and C. L. Turnbough. 1994. Regulation of pyrBI operon expression in Escherichia coli by UTP-sensitive reiterative RNA synthesis during transcriptional initiation. Genes Dev. 8:2904-2912.
30. Liu, J. and C. L. Turnbough. 1994. Effects of transcriptional start site sequence and position on nucleotide-sensitive selection of alternative start sites at the pyrC promoter in Escherichia coli. J. Bacteriol. 176:2938-2945.
31. Lorenz, M. G. and W. Wackernagel. 1994. Bacterial gene transfer by natural genetic transformation in the environment. Microb. Rev. 58:563-602.
32. Mackie, G. A., B. C. Donly and P. C. Wong. 1990. Coupled transcription-translation of ribosomal proteins. 191-211 In G. Spedding (ed.), Ribosomes and Protein Synthesis. Oxford University Press, Oxford.
33. Meadow, N., D. Fox and S. Roseman. 1990. The bacterial phosphoenolpyruvate: glycose phosphotransferase system. Ann. Rev. Biochem. 59:497-542.
34. Michod, R. E. and B. R. Levin. 1988. The Evolution of Sex.
35. Michod, R. E. and M. F. Wojciechowski. 1994. DNA repair and the evolution of transformation IV. DNA damage increases transformation. J. evol. Biol. 7:147-175.
36. Moxon, E. R. and R. Wilson. 1991. The role of Haemophilus influenzae in the pathogenesis of pneumonia. Reviews of Infectious Diseases
37. Myers, R. S. and F. W. Stahl. 1994. Chi and the RecBCD enzyme of Escherichia coli. Ann. Rev. Genet. 28:49-70.
38. Pifer, M. L. and H. O. Smith. 1985. Processing of donor DNA during Haemophilus influenzae transformation: analysis using a model plasmid system. Proceedings of the National Academy of Sciences of the United States of America 82:3731-5.
39. Postma, P., J. Lengeler and G. Jacobson. 1993. Phosphoenolpyruvate:carbohydrate phosphotransferase systems of bacteria. Microb. Rev. 57:543-594.
40. Quagliarello, V. and W. Scheld. 1992. Bacterial meningitis: pathogenesis, pathophysiology, and progress. New. Engl. J. Med. 327:864-872.
41. Redfield, R. J. 1988. Evolution of bacterial transformation: is sex with dead cells ever better than no sex at all? Genetics 119:213-21.
42. Redfield, R. J. 1991. Bacterial Mating Preferences. Nature 352:25-26.
43. Redfield, R. J. 1991. sxy-1, a Haemophilus influenzae mutation causing greatly enhanced competence. J. Bacteriol. 173:5612-5618.
43a. Reizer, J., V. Michotey, A. Reizer and M. H. Saier. 1994. Novel phosphotransferase system genes revealed by bacterial genome analysis: Unique, putative fructose- and glucoside-specific systems. Protein Science 3:44-=450.
44. Setlow, J. K. and M. E. Boling. 1972. Bacteriophage of Haemophilus influenzae. II. Repair of ultraviolet-irradiated phage DNA and the capacity of irradiated cells to make phage. Journal of Molecular Biology 63:349-62.
45. Setlow, J. K., M. E. Boling, D. P. Allison and K. L. Beattie. 1973. Relationship between prophage induction and transformation in Haemophilus influenzae. Journal of Bacteriology 115:153-61.
46. Smith, H. O., J.-F. Tomb, B. A. Dougherty, R. D. Fleischmann and J. C. Venter. 1995. Frequency and distribution of DNA uptake signal sequences in the Haemophilus influenzae Rd genome. Science 269:538-540.
47. Stouthamer, A. H. 1979. The search for correlation between theoretical and experimental growth yields. Int. Rev. Biochem. 21:1-47.
48. Tremblay, L., J. L'Ecuyer, P. Provencher and M. Bergeron. 1990. Susceptibility of Haemophilus influenzae to antimicrobial agents used in Canada. Can. Med. Assoc. J. 143:895-901.
49. Turner, R. J., Y. Lu and R. L. Switzer. 1994. Regulation of the Bacillus subtilis pyrimidine biosynthetic (pyr) gene cluster by an autogenous transcriptional attenuation mechanism. Journal of Bacteriology 176:3708-22.
50. Vogel, U. and K. Jensen. 1994. Effects of guanosine 3',5'-bisdiphosphate (ppGpp) on rate of transcription elongation in isoleucine-starved Escherichia coli. J. Biol. Chem. 269:16236-16241.
51. Wilcox, K. W. and H. O. Smith. 1975. Isolation and characterization of mutants of Haemophilus influenzae deficient in an adenosine 5'-triphosphate-dependent deoxyribonuclease activity. Journal of Bacteriology 122:443-53.
52. Wise, E. M., S. Alexander and M. Powers. 1973. Adenosine 3':5'-cyclic monophosphate as a regulator of bacterial transformation. Proc. Natl. Acad. Sci. USA. 70:471-474.
53. Zoon, K. C., M. Habersat and J. J. Scocca. 1975. Multiple regulatory events in the development of competence for genetic transformation in Haemophilus influenzae. Journal of Bacteriology 124:1607-9.
54. Zuker, M. and P. Stiegler. 1981. Optimal computer folding of large RNA sequences using thermodynamics and auxiliary information. Nucleic Acids Res. 9:133-148.
55. Zulty, J. and G. Barcak. 1993. Structural organization, nucleotide sequence, and regulation of the Haemophilus influenzae rec-1+ gene. J. Bacteriol. 175:7269-7281.
56. Zulty, J. and G. J. Barcak. 1995. Identification of a DNA transformation gene required for com101A-plus expression and supertransformer phenotype in Haemophilus influenzae. Proc. Natl. Acad. Sci. USA 92: 3616-3620.
Back to Redfield Lab Research Page
