Rate of Inversion of the Salmonella enterica Shufflon Regulates Expression of Invertible DNA
http://www.100md.com
感染与免疫杂志 2005年第9期
Department of Biochemistry, Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong, China
ABSTRACT
Salmonella enterica serovar Typhi and some strains (Vi+) of serovar Dublin use type IVB pili to facilitate bacterial self-association, but only when the PilV proteins (potential minor pilus proteins) are not synthesized. Pilus-mediated self-association may be important in the pathogenesis of enteric fever. We have suggested that the rate of Rci-catalyzed inversion of DNA encoding the C-terminal portions of the PilV proteins controls PilV protein synthesis. This potentially represents a novel means of transcriptional control. Here, it is initially shown that DNA inversion per se is required for inhibition of gene expression from invertible DNA. Binding, without DNA scission, of Rci to its substrate sequences on DNA cannot explain the data obtained. Next, it is shown that inversion frequencies of xylE-encoding DNA, bracketed by Rci substrate sequences, may be modulated by changes in the 19-bp consensus sequences which are essential components of Rci substrate DNA. The affinity of Rci for these sequences affects inversion frequencies, so that a greater affinity is predictive of faster inversion, and therefore less synthesis of product encoded by invertible DNA. Inversion events may inhibit transcription of DNA from external promoters. In vivo, the frequency of Rci-mediated inversion is influenced by the extent of DNA supercoiling, with increasing levels of expression of invertible genes as novobiocin inhibits DNA supercoiling and thus Rci action. This inhibition of DNA supercoiling results in increased synthesis of PilV proteins as Rci activity decreases, and, in turn, bacterial self-association (particularly in serovar Dublin) decreases.
INTRODUCTION
The Salmonella enterica pil operon is located in Salmonella pathogenicity island 7 (SPI7), which also carries the viaB gene cluster required for the synthesis of the Vi antigen (15). SPI7 is unstable in serovar Typhi, with excision possibly facilitated by long-term strain storage or growth under laboratory conditions (2, 14). The pil operons of S. enterica serovar Typhi, and some strains (Vi+) of serovar Dublin, encode type IVB pili. These pili mediate bacterial self-association, but only when the presumptive minor pilus proteins PilV1 and PilV2 are not expressed (12, 13). The N-terminal regions of both PilV proteins are identical, but the C termini differ as they are encoded by invertible DNA at the end of the pil operon. The adjacent rci gene encodes a recombinase active to effect the DNA inversion. The C-terminal part of the pilV gene and the adjacent rci gene, encoding a tyrosine site-specific recombinase, together comprise the S. enterica shufflon. The Rci protein acts on DNA sequences, including conserved 19-bp inverted repeats to invert DNA in the C-terminal portion of the pilV gene, and is active only on supercoiled DNA (4, 5, 12).
Apart from playing a role in bacterial self-association, the pil operon may also be important in attachment of invading bacteria to eukaryotic cells during infection. The type IVB structural pilin, PilS, binds to the cystic fibrosis transmembrane conductance regulator, the recognized eukaryotic cell receptor for serovar Typhi (9, 16, 19).
Since these type IVB pili-mediated events cannot be effected by serovars lacking pil, such as Salmonella enterica serovar Typhimurium, we have suggested that the expression of type IVB pili by certain serovars (particularly Typhi) might be important in explaining why only certain S. enterica serovars cause enteric fever in humans (13, 21). While serovar Paratyphi C (Vi+) carries a pil operon very similar to that of serovar Typhi, the shufflon is inactive because the 19-bp Rci consensus inverted repeat sequences have been modified to be 20 bp in length, and DNA between such sequences is not invertible by Rci. Since serovar Paratyphi is less virulent than is serovar Typhi, it is possible that the observed differences in pathogenicity are partly due to the demonstrated changes in pil (18). In the absence of tests involving the administration of serovar Typhi pil mutants to humans, the observation that inactivation of a similar pil operon in Yersinia pseudotuberculosis decreased mouse virulence, with an increase in the 50% lethal dose value of 0.7 log, is the only reported test of the effect of a pil mutation on virulence (3).
We have suggested that control of the synthesis of the PilV proteins is effected by rapid DNA inversion activity of the shufflon (7, 12, 20). Through-transcription of the pilV gene, from a promoter located outside the invertible DNA, may not be possible with rapidly inverting DNA. The idea that control of protein synthesis is effected by the rate of DNA inversion of a portion of the gene encoding the protein is novel, and worthy of further exploration. In the present study, we examine the model in some detail and offer evidence that it is an accurate representation of events in cells of pil+ strains of S. enterica.
MATERIALS AND METHODS
Materials and DNA manipulation. Reagents were of molecular biology grade. Enzymes active on DNA were obtained from either Invitrogen or Roche and were used as directed by the suppliers. X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) and IPTG (isopropyl--D-thiogalactopyranoside) were purchased from Amersham. Bio-Rad was the supplier of polyvinylidene difluoride membrane. Plasmid preparation, restriction enzyme digestion, and other DNA manipulations were done as described by Sambrook et al. (17).
Media. Luria-Bertani broth (LB) was prepared as described by Miller (11). Solid medium contained 1.5% (wt/vol) agar. Antibiotics were added, when appropriate, to 5 to 15 μg/ml (Tc), 50 μg/ml (Str), or 100 μg/ml (Ap).
Bacterial strains. Serovar Typhi J341 (Ty2 Vi–) (20) is the wild-type strain used here. The construction of site-directed pilS and pilV mutants of this strain has been described (20, 21). The wild-type serovar Dublin strain 124 (Vi+), and the pilS and pilV mutants derived therefrom, were used in earlier work (13). Serovar Typhi J341 pilS/pRU670 (Rts1::Tn1731) (Tcr) was the donor, in liquid matings, of the conjugative plasmid pRU670 (12). For use as recipients in liquid matings, spontaneous rpsL (Strr) mutants of Typhi and Dublin, obtained earlier (12, 13), were used. Escherichia coli K-12 DH5 [supE44 lacU169 (80 lacZM15) hsdR17 recA1 endA1 gyrA96 (Nalr) thi-1 relA1] was the usual host for recombinant plasmids. E. coli JM109 [F' traD36 lacIq (lacZ)M15 proA+B+/e14–(McrA–) (lac-proAB) thi gyrA96 (Nalr) endA1 hsdR17 (rK– mK–) relA1 supE44 recA1] was used as the host for plasmids carrying the xylE gene between DNA potentially invertible by Rci. The xylE gene was obtained by PCR from pCM20 (10). Plasmids carrying genes for proteins tagged with glutathione S-transferase (GST), His6, or Trx were expressed in E. coli strain BL21(DE3) [F– ompT hsdSB (rB– mB–) gal dcm (DE3); Novagen], which carries a chromosomal copy of the T7 RNA polymerase gene under control of the lacUV5 promoter.
Purification of a fusion protein containing Rci. The rci gene was subcloned between the BamHI and XhoI sites of plasmid pET-trx, which is a modification of plasmid pET-32a (Novagen), in which the S-tag and the enterokinase cleavage site have been removed. In this construct, termed pET-trx-rci, the N-terminal DNA of rci was joined in-frame to, sequentially, DNA specifying the thioredoxin (Trx) and hexahistidine (His6) fusion tags to encode a protein, Trx-His6-Rci, of 57,565 Da (514 amino acids [aa]), which includes the entire Rci sequence. The 155-aa control protein (Trx-His6) expressed from plasmid pET-trx was 16,689 Da in molecular size. In both cases, the T7 promoter drives expression of the genes encoding the fusion proteins. To prepare the proteins, 5-ml amounts of stationary-phase cultures of E. coli BL21(DE3) hosting pET-trx-rci or pET-trx were inoculated into 1-liter amounts of fresh LB medium with Ap and shaken at 37°C to optical cell densities at 600 nm of ca. 0.4. IPTG was added to 0.4 mM, and the cultures were shaken at 30°C for a further 3 h. After centrifugation (5,887 x g for 30 min at 4°C), bacterial pellets were frozen at –80°C to facilitate membrane rupture. Pelleted material from 500 ml of culture was thawed and resuspended in 30 ml of binding buffer (20 mM Tris-HCl, 500 mM NaCl, 5 mM imidazole [pH 8]), and protease inhibitors (pepstatin to 1 μg/ml, leupeptin to 1 μg/ml, and phenylmethylsulfonyl fluoride to 1 mM) were added. The resuspended bacterial pellet was placed on ice and subjected to sonication by using a B. Braun Labsonic U machine with a power level of ca. 100 W and a duty cycle of 0.3 s, for 10 cycles each of 30 s with 30 s of a cooling pause between each cycle. The sonicate was centrifuged, and the supernatant (30 ml) was collected for Ni-iminodiacetic acid protein affinity purification under native conditions. Agarose beads with immobilized iminodiacetic acid were purchased from Novagen and charged with Ni2+ to permit binding of fusion proteins with the His6 tag. The beads, prepacked in a small column, were extensively washed with binding buffer (above), and the 30 ml of supernatant was loaded to 1.5 ml of packed beads and held at 4°C for 1 h. The column was washed with 400 ml of binding buffer, and fusion proteins eluted by gradually increasing the concentration of imidazole in elution buffers (all 20 mM Tris-HCl, 500 mM NaCl, with 40, 60, 100, 250, or 500 mM imidazole [pH 8]). For each imidazole concentration, 5 to 7.5 ml of elution buffer was used. Fusion proteins were detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) of aliquots of the eluates. SDS-PAGE was performed on polyacrylamide at 5% (wt/vol) stacking gel to 15% (wt/vol) separating gel. Immunoblotting with mouse monoclonal immunoglobulin G (IgG) anti-His6 antibody (Santa Cruz Biotechnology) was performed to confirm the identities of purified proteins. The primary antibody was used at a dilution of 1:500 in TBST (20 mM Tris-HCl, 140 mM NaCl, 0.1% [wt/vol] Tween 20). The secondary antibody (Amersham), used at a dilution of 1:2,000 in TBST, was a sheep anti-mouse IgG peroxidase-linked antibody. Development of immunoblots employed the enhanced chemiluminescence system (Amersham), with the substrate H2O2. Fractions of interest were dialyzed against buffer D (20 mM Tris-HCl, 50 mM NaCl [pH 8]) and stored at –80°C in 10% (vol/vol) glycerol. The fusion protein Trx-His6-Rci was eluted in fractions containing imidazole at 250 and 500 mM. Purified control Trx-His6 protein was eluted in fractions containing 500 mM imidazole (Fig. 1).
Detection of Rci self-association. A pull-down assay was used to detect self-association of Rci protein. The fusion proteins GST-His6-Rci, GST-His6, and His6-Rci were used. To prepare GST-His6-Rci, the rci gene was first cloned between the BamHI and XhoI sites of pET-42a (Novagen) to give pET-42a-rci. The GST-His6 protein was encoded by pET-42a. To construct a plasmid encoding His6-Rci, the rci gene was cloned between the BamHI and XhoI sites of plasmid pET-6H modified from pET-32a (Novagen) with the Trx-tag, S-tag, and enterokinase cleavage site removed to give plasmid pET-6H-rci. The His6-Rci protein was prepared in a manner similar to that described above for the Trx-His6-Rci protein, except that His6-Rci was eluted with a buffer containing only 100 mM imidazole. To prepare the GST-His6-Rci and GST-His6 proteins, E. coli BL21(DE3) strains hosting pET-42a-rci or pET-42a were grown to an optical density at 600 nm of ca. 0.7 under shaking conditions at 37°C and then induced with IPTG at 0.5 mM, followed by further shaking growth at 25°C for 3 to 4 h prior to harvest. Processing of the bacterial pellet to obtain purified fusion proteins was similar to that described above for the Trx-His6-Rci and Trx-His6 proteins except (i) phosphate-buffered saline (pH 7.4) was substituted for binding buffer, (ii) the affinity beads used were Glutathione-Sepharose beads from Amersham, and (iii) the elution buffer was 10 mM reduced glutathione in 50 mM Tris-HCl (pH 8.0). All three protein samples were dialyzed against buffer D (above), and the GST-His6-Rci and His6-Rci fusion proteins were concentrated by ultrafiltration.
Two tubes received, in a final volume of 400 μl of buffer A (20 mM Tris-HCl, 50 mM NaCl, 0.05% [vol/vol] Triton X-100, 1% [wt/vol] bovine serum albumin [pH 8.0]), either (i) GST-His6-Rci (16 μg) and His6-Rci (10 μg) (test) or (ii) GST-His6 (7 μg) and His6-Rci (10 μg) (control). The mixtures were gently incubated at 4°C for 4 h on a rotator, and then each tube received 50 μl of a suspension of GST-Sepharose beads (50% [vol/vol], preincubated in buffer A for 4 h). The mixtures were gently agitated at 4°C for a further 2 h. After brief centrifugation the supernatant was discarded, and the proteins in the bead-containing pellets were eluted with 40 μl of SDS-PAGE loading buffer. After gel electrophoresis of 10-μl samples, an anti-His6 antibody (above) was used as the primary antibody in immunoblotting.
Nature of DNA sequences in the pilV region and the terminology used to describe them. In wild-type serovars Typhi and Dublin, DNA invertible by Rci lies between two 19-bp inverted repeat sequences, differing by a single bp, and termed the V1 or V2 sequences (12, 13). Use of the term "V1 orientation" indicates that a promoter external to potentially invertible DNA reads first through the V1 sequence, across invertible DNA, and out through the V2 sequence, while use of the term "V2 orientation" indicates that the locations of the repeat sequences are reversed. In all plasmids in the present study containing 19-bp sequences recognized by Rci, the 19-bp sequence located just before the xylE gene and forming part of the leftmost inversion site in plasmids with two 19-bp sequences is preceded by the 12-bp sequence TGCCACACTTTC, which is the sequence found naturally in serovars Typhi and Dublin (see GenBank AF000001). The 31-bp sequence composed of the 12-bp sequence and the 19-bp sequence together form the Rci substrate site on the DNA (5). Likewise, the 19-bp sequence located just before the rci gene and forming part of the rightmost inversion site in plasmids with two 19-bp sequences is followed by the natural S. enterica 12-bp sequence GTATGTCCTTAC to form a 31-bp Rci substrate site. The use of the natural S. enterica 12-bp sequences as components of the 31-bp Rci recognition sites is important, since changes in these 12-bp sequences can dramatically affect the steady-state levels of the two possible orientations of invertible DNA (5). Thus, pUST164 (18) differs from pUST170 (the present study) only in the replacement (in pUST164) of the leftmost natural S. enterica 12-bp sequence with unrelated vector DNA. In pUST170, >90% of the DNA is in the orientation permissive for transcription (this work), whereas ca. 100% of the DNA is in the other orientation in pUST164 (18).
Measurement of XylE activity. Expression of the reporter xylE gene, behind the lac promoter, was used as a measure of DNA transcription. The xylE gene of pCM20 has been widely used in such an experimental context (10). This assay, using sonicates of stationary-growth-phase cultures of E. coli JM109-based strains, has been described (12). Levels of catechol 2,3-dioxygenase were assayed. Catechol was from Aldrich, Inc. Enzyme units are expressed as nanomoles of 2-hydroxymuconic semialdehyde produced per minute per microgram of protein. A Bio-Rad protein assay kit was used for estimation of protein concentration.
Electrophoretic mobility shift assay. Four picomoles of double-stranded DNA molecules (termed V1, V2, V3, or V4 19-bp sequences [described below]) were radiolabeled with 40 μCi of Redivue adenosine 5'-[32P-PO4]triphosphate (Amersham) by using T4 polynucleotide kinase (Amersham). Purification of the radiolabeled DNA molecules was effected with the QiaQuick nucleotide removal kit (QIAGEN). For each DNA-binding reaction, 50 ng of Trx-His6-Rci protein (test) or Trx-His6 protein (control) was added to a 20-μl final reaction mixture containing 0.02 pmol (5,000 cpm) of labeled probe, 1 μg of bovine serum albumin/μl, and 4 mM spermidine in binding buffer (30 mM Tris-HCl, 80 mM KCl, 15 mM NaCl, 10% [vol/vol] glycerol [pH 8.0]), with or without unlabeled probe DNA in molar excesses of 5-, 10-, 20-, or 30-fold. After incubation for 30 min at room temperature, the samples were loaded onto a 6% (wt/vol) native polyacrylamide gel, and electrophoresis was performed at 300 V for 2.5 h at 4°C. The gel was dried and exposed by autoradiography at –80°C to Fuji Super RX X-ray film with intensifying screens. To quantify labeled probe-protein complexes formed, the films were scanned and analyzed with a Bio-Rad imaging densitometer, which measured units of band intensity on a grayscale as optical densities at 400 to 750 nm per square centimeter.
Effect of novobiocin on XylE expression from invertible DNA. E. coli JM109-based strains hosting either pUST170 (test plasmid) or pUST168 (positive control plasmid) were grown in shaking (220 rpm) culture at 37°C for 24 h, to stationary phase, with Ap, IPTG (0.2 mM), and various concentrations of novobiocin (0 to 150 μg/ml). The cells were sonicated and XylE assays performed on the cell lysates.
Liquid mating tests. These tests were conducted as described previously (12). Briefly, a donor strain (a pilS mutant of serovar Typhi carrying a conjugative Tcr plasmid, and Strs) in the logarithmic growth phase, and recipients (various Strr strains of serovars Typhi or Dublin) in the stationary growth phase were mixed in a 10:1 ratio of recipients to donors, and liquid matings proceeded for 2 h (serovar Typhi recipients) or 4 h (serovar Dublin recipients) at 30°C. Dilutions of the mating mixtures were plated with Tc (selecting for the R-factor) and Str (counterselecting the donor) for colony enumeration. In some tests, novobiocin (15 or 20 μg/ml) was present during growth of the recipients and during the matings.
RESULTS AND DISCUSSION
Inversion rate of xylE-containing DNA regulates XylE expression. E. coli JM109-based strains hosting various pUC19-derived plasmids with a promoterless xylE gene behind the lac promoter and between DNA sequences potentially invertible by Rci were constructed (Fig. 2A). Plasmids pUST168 and pUST169 are positive and negative controls, respectively. In either plasmid, the xylE gene is noninvertible due to a deletion in rci. In pUST168, the xylE gene is fixed in the orientation permissive for transcription; the reverse orientation is found in pUST169. In plasmids pUST170 and pUST171 (in which the xylE gene is in the V1 or V2 orientation, respectively), the xylE gene is potentially invertible. The favored orientation in either plasmid is the orientation in which xylE may be transcribed by the lac promoter (Fig. 2B). This indicates that the 19-bp Rci substrate sequences do not determine the preferred orientation of DNA inserted between such sequences. Instead, the sequence of the contained DNA appears to dictate the steady-state levels of the two possible orientations. Expression of XylE by a strain containing pUST170 was very low (Fig. 2C), despite the fact that >90% of the invertible DNA, in the steady-state, was in the orientation permissive for xylE transcription (Fig. 2B). The same was true of a strain containing pUST171 (Fig. 2B and C). Poor expression of the xylE gene indicated a defect in transcription, since a viable lac promoter preceded the reporter gene. The low expression of the xylE gene was accompanied by active inversion. Thus, transformation of preparations of either pUST170 or pUST171 into E. coli JM109, followed by examination of plasmid DNA isolated from 50 separate colonies or transformation, showed that all plasmids had achieved the equilibrium orientations of Fig. 2B, even though the transforming DNA was a mixture (in the ratio of ca. 9:1) of two distinct plasmids.
It is important to note that in pUST170 and pUST171, transcription of the rci gene will occur from the rci promoter only. Read-through transcription of rci from the lac promoter will not occur in these plasmids because the failure, in strains hosting these plasmids, to transcribe the xylE gene from the lac promoter precludes transcription of a gene (rci) located further downstream. Therefore, the bias in orientation of the insert DNA of pUST170 and pUST171 should not be due to an effect of the lac promoter on Rci expression level. To confirm this, insert DNA of pUST168, pUST170, and pUST171 was cloned into pUC18 (which placed the lac promoter to the right of the rci gene), and the lac promoter excised from these plasmids by removal of a 364-bp PciI-Asp718 fragment. It was confirmed (i) that none of the resulting plasmids expressed any XylE activity and (ii) that the equilibrium levels of the two possible insert orientations in the plasmids obtained from pUST170 and pUST171 were identical to those seen in pUST170 and pUST171 (data not shown).
The pUST170/pUST171 data suggested two possibilities. First, it might be that an Rci-mediated rapid rate of DNA inversion inhibited transcription of the xylE gene, since RNA polymerase might not effectively transcribe DNA in the process of undergoing inversion. Second, it could be that Rci binding to a substrate DNA region including the leftmost 19-bp fragment (V1 in pUST170, for example) inhibited through-transcription of xylE from the lac promoter. The rci genes of plasmids pUST170 and pUST171 were then subjected to site-directed mutagenesis to replace Tyr272 by Phe (rci encodes a protein of 383 aa). This change in an amino acid residue essential for Rci activity (1) should eliminate DNA inversion activity by Rci but should not destroy the ability of Rci to bind to substrate DNA (4). Plasmids with previously invertible DNA fixed in either orientation were obtained, showing that the inversion activity of Rci was indeed eliminated by this mutation. Plasmids pUST172 and pUST173, in which the xylE gene is fixed in the productive orientation, were derived from pUST170 and pUST171, respectively (Fig. 2B), and strains hosting these plasmids produced high levels of XylE (Fig. 2C). This observation argued against the possibility that binding per se of Rci to DNA between the lac promoter and xylE, without DNA scission was sufficient to reduce xylE transcription. Instead, it seemed that Rci-mediated inversion activity might be essential to the reduction of XylE levels (from that of E. coli JM109/pUST168) noted in strains hosting pUST170 or pUST171.
It remained possible, however, that the Tyr272 to Phe mutation in rci reduced (compared to wild-type Rci) the binding of Rci to DNA containing a 19-bp substrate sequence. Mutations in the C-terminal region of the prototype tyrosine recombinase, integrase, may reduce the affinity of the enzyme for its substrate sites in DNA (1). The task thus was to construct a plasmid with a wild-type rci gene and a wild-type xylE gene, bracketed by wild-type 19-bp Rci substrate sequences, but to somehow render the DNA between the 19-bp sequences noninvertible. We have established (unpublished data) that the introduction of DNA with the potential to form secondary structure into hitherto-invertible DNA eliminates Rci-mediated inversion of that DNA. Therefore, the trpA terminator was inserted, after the xylE gene, into DNA which was previously invertible (Fig. 3A). Plasmids pUST174 and pUST175 both carry this modification and have the hitherto-invertible DNA in the V1 or V2 orientations, respectively. The DNA between the 19-bp sequences was noninvertible (Fig. 3B), and E. coli JM109-based strains hosting either plasmid showed high levels of XylE activity (Fig. 3C). These data strongly suggested that inversion per se was required for the observed inhibition of XylE expression from plasmids with invertible xylE genes compared to the XylE levels synthesized by the positive control E. coli JM109/pUST168. It was clear that binding of Rci to the leftmost 19-bp sequence could not inhibit xylE transcription from the lac promoter. However, as a final control, it was necessary to show that the activity of the rci promoter was not affected by the upstream insertion of DNA containing the trpA terminator (Fig. 4). Plasmid pUST176 was constructed to have the rci promoter, with 65 bp of natural S. enterica DNA sequence upstream of the –35 sequence, driving the xylE gene, whereas plasmid pUST177 contained the same insertion of DNA with the trpA terminator used in the construction of plasmid pUST174. Assay of XylE levels (performed three times, each time in duplicate) in E. coli JM109 strains hosting pUST176 or pUST177 yielded values of 11.9 (standard deviation of 1.5) or 12.7 (standard deviation of 0.7) enzyme units, respectively (ca. 20% of the value shown by an induced lac promoter in other experiments). These values did not differ significantly. It may therefore be concluded that insertion of the trpA terminator 211 bp upstream of the rci –35 sequence did not affect the activity of the rci promoter. To confirm that Rci expression was identical from pUST170 and pUST174, the His6 epitope tag was added to the C termini of the rci genes of these plasmids, and Rci expression levels in lysates of cells hosting the plasmids compared by immunoblotting, after SDS-PAGE, of lysate dilutions, using the mouse monoclonal IgG anti-His6 as primary antibody. The Rci expression levels in strains hosting either plasmid were very low (ca. 850 molecules/cell, which was at the limit of detection of the assay), due to (negative) autoregulation of rci transcription by binding of Rci to the –35 sequence of its own promoter, which is located in the rightmost 19-bp substrate sequence (Fig. 4 and unpublished data). There was no difference in Rci expression levels between strains hosting the rci-tagged derivatives of pUST170 or pUST174.
Our four principal findings can be summarized as follows. (i) In pUST170, pUST171, pUST174, and pUST175, transcription of the rci gene occurs from the rci promoter only. As mentioned above, read-through transcription of rci from the lac promoter will not occur in pUST170 or pUST171 because the failure, in strains hosting these plasmids, to transcribe the xylE gene from the lac promoter precludes transcription of the rci gene located further downstream. The lac promoter cannot transcribe rci in pUST174 or pUST175 because a transcriptional terminator is located between xylE and rci. (ii) Insertion of a transcriptional terminator 211 bp upstream of the rci –35 sequence does not affect the activity of the rci promoter. Therefore, Rci protein expression from plasmids pUST174 and pUST175 may be expected to be at levels comparable to those from similar plasmids (pUST170 and pUST171) lacking the DNA insertion with the trpA terminator. The elimination of DNA inversion caused by the trpA insertion is presumably due to a change in DNA topology. (iii) The fact that the active site Tyr272-to-Phe mutation of plasmids pUST172 or pUST173 eliminated Rci activity but still permitted high XylE expression in strains hosting these plasmids suggested that DNA inversion per se was required to inhibit XylE expression from invertible DNA compared to XylE levels synthesized by the positive control strain E. coli JM109/pUST168. However, the Tyr272-to-Phe mutation might also conceivably lessen the binding of Rci to its substrate sequences (1). Hence, inhibition of XylE synthesis by binding, without DNA scission, of Rci to its leftmost substrate sequence (between the lac promoter and xylE) could not be ruled out. (iv) The Rci expressed from plasmids pUST174 and pUST175 is, however, wild type. Therefore, the fact that E. coli JM109 strains hosting plasmids pUST174 or pUST175 express high XylE levels compared to XylE levels synthesized by the positive control strain E. coli JM109/pUST168 indicates that DNA inversion per se is required to inhibit XylE expression from invertible DNA.
The affinity of Rci for DNA containing 19-bp substrate sequences affects DNA inversion rate. If the rate of Rci inversion activity controls the expression of invertible DNA carrying the xylE gene, it follows that modulation of this rate should be reflected in variation of XylE expression levels. An obvious way in which to attempt to modulate Rci activity is to vary the 19-bp Rci substrate sequences. Plasmid R64 contains five such sequences (there are seven 19-bp sequences in the R64 shufflon; three are identical) (5), whereas the plasmid R721 shufflon (which contains six 19-bp sequences in all) (6) contains one 19-bp sequence which differs from both the R64 sequences and those of S. enterica. A total of at least eight distinct 19-bp sequences which are actual or potential substrates for S. enterica Rci is thus known. All eight sequences were examined for relative Rci affinities by the techniques to be described below. Together with the natural S. enterica V1 and V2 sequences, two R64-derived sequences (Fig. 5), termed V3 and V4, were chosen for detailed examination, since they represented the extremes (high and low, respectively) of affinity in Rci affinity assays. It is important to note that although the –35 sequence of the rci promoter (8) is contained in these 19-bp sequences (Fig. 5), this sequence is not affected by the nucleotide differences between the four 19-bp sequences. Possible effects of differences in these 19-bp sequences on Rci activity should therefore not be due to differences in rci transcription.
Plasmids pUST178-pUST181 (Fig. 6A) contain the xylE gene bracketed by 19-bp V1, V2, V3, or V4 pairs, respectively. In all four cases, DNA inversion activity may be seen (Fig. 6B), with the steady-state levels of DNA in the orientation permissive for xylE transcription being high and rather uniform in the four E. coli JM109 strains hosting the various plasmids. The XylE expression levels differed between strains, however (Fig. 6C), and were influenced by the nature of the 19-bp inverted repeats bracketing the xylE gene, decreasing in the order V4 > V2 > V1 > V3.
Next, experiments involving the affinity of a purified fusion protein containing full-length Rci for 19-bp sequences forming portions of Rci substrate sites on DNA were designed. Since Rci cleaves DNA as a dimer (5), it was useful to confirm that Rci-containing fusion protein monomers retained the ability to interact in solution. This was determined to be the case (Fig. 7). Pulldown tests, with beads binding GST-containing proteins, showed that GST-His6-Rci protein bound to His6-Rci in solution, thus indicating that the Rci component of Trx-His6-Rci (Fig. 1) should function like untagged Rci in affinity tests with Rci substrate sequences of DNA.
The affinity of the purified Rci-containing fusion protein to the different 19-bp sequences was ranked in an assay in which an excess of unlabeled DNA containing various 19-bp sequences was used to compete with labeled DNA associating with Rci. Complexes of labeled DNA and Rci were detected by autoradiography after gel electrophoresis (Fig. 8 is the example with labeled V1-containing DNA). The abilities of unlabeled DNA fragments containing 19-bp sequences to compete with a labeled V1-containing sequence for purified Rci dimer could be ranked in the order: V3 > V1 > V2 > V4 (Fig. 8). This is particularly clear when the intensities of the labeled bands in the presence of a 30-fold excess of unlabeled challenge DNA are examined (Fig. 8). The challenge effectiveness ranking of V3 > V1 > V2 > V4 was seen not only when the labeled sequence contained the V1 19-bp sequence but also when the labeled sequence was any of the other three 19-bp sequences (V2, V3, or V4) (Fig. 9). The relative affinities of different 19-bp sequences for Rci explains the differences in XylE expression levels in the four E. coli JM109 strains hosting pUST178-pUST181. The stronger the affinity of a 19-bp substrate sequence for Rci, the less XylE is produced in an E. coli JM109 strain hosting plasmids such as pUST178-pUST181. Binding, without DNA scission, of Rci to its substrate sequences cannot explain these data, as discussed above. Instead, the data strongly suggest that the relative affinity of Rci for DNA containing a 19-bp substrate sequence determines, in turn, the frequency of inversion, so that greater affinity results in higher inversion frequency and thus less through-transcription of xylE from the lac promoter, resulting in lower XylE levels.
Inhibition of DNA supercoiling affects Rci activities. In S. enterica serovars Typhi and Dublin, portions of the genes encoding the PilV1 and PilV2 proteins are located in DNA invertible by Rci (13, 20) and will therefore not be synthesized when Rci-mediated inversion activity is high. We have suggested that DNA supercoiling, required for Rci activity (4, 5, 12), may ultimately regulate shufflon activity, resulting in PilV-free pili when inversion activity is high, and PilV-tipped pili when such activity is low. In turn, bacterial self-association, mediated by PilV-free pili, may therefore occur preferentially under circumstances permissive for DNA supercoiling, such as in the anaerobic environment of the human gut, as a prelude to bacterial invasion.
Since novobiocin inhibits DNA supercoiling, it was of interest to examine possible effects of the drug on quantifiable Rci functions. First, E. coli JM109 strains hosting either pUST170 (Fig. 2A; the plasmid contains a xylE gene, in the V1 orientation, invertible by Rci) or pUST168 (Fig. 2A; positive control plasmid) were grown in the presence of various amounts of novobiocin and XylE production levels assayed (Fig. 10). Increasing amounts of novobiocin decreased XylE production gradually in E. coli JM109/pUST168, presumably reflecting an inhibition by the drug of gyrase activities useful to facilitate efficient DNA transcription. With E. coli JM109/pUST170, however, the low XylE levels synthesized in the absence of novobiocin increased, by just over a factor of 2, as novobiocin concentrations increased, to peak when the novobiocin level was 75 μg/ml (Fig. 10). This suggested that a gradual reduction in DNA supercoiling inhibited Rci activity and thus allowed more transcription of the xylE gene. At novobiocin levels greater than 75 μg/ml, XylE production decreased, again presumably because of adverse effects of novobiocin on general transcription.
As previously described (12), a liquid mating assay, in which potentially self-associating bacterial recipients are in 10-fold excess to donors, provides a measure of bacterial self-association, since the recipients tend to enmesh donor bacteria in a developing pellet. This entrapment of donors is reflected in higher transfer of a conjugative plasmid, from donor to recipient, than is the case when the recipient strains do not self-associate. A pilS donor of serovar Typhi was used in these tests, since this strain does not make pili and therefore cannot contribute to any pilus-mediated bacterial association. The pilS-to-pilS background transfer frequencies in individual tests ranged from 3 x 10–3 to 11 x 10–3/donor bacterium, as before (12), and mating frequencies with non-pilS recipients are shown as multiples of these figures (Fig. 11). In liquid mating assays, in the absence of novobiocin, the pili mediated bacterial self-association in both the wild-type and pilV strains of serovars Typhi and Dublin, as shown by the fact that the transfer frequencies to such strains were higher than those to pilS mutants (Fig. 11) (12, 13). The pilV mutants were more effective recipients than were the wild-type strains, since the pilV mutants never make PilV proteins and therefore self-associate to levels greater than those of the wild-type strains. In wild-type bacteria, PilV protein synthesis may occur when the shufflon inversion frequency is sufficiently low to permit transcription of the shufflon-contained portions of the pilV genes. When novobiocin was added to the mating mixtures, the plasmid transfer frequencies to pilV mutants were not significantly altered (Fig. 11). With the wild-type strains as recipients, however, novobiocin reduced recipient ability. This reduction was barely significant (the standard deviation error bars almost overlap) when wild-type serovar Typhi was the recipient but was much clearer when wild-type serovar Dublin was used in this capacity (Fig. 11). This is due to the fact that the recipient ability of wild-type serovar Typhi, in the absence of novobiocin, is already low (Fig. 11) (13), being reproducibly about half that of serovar Dublin. This may be due to a lower level of shufflon inversion in laboratory-grown serovar Typhi compared to serovar Dublin. Thus, it may be suggested that in serovar Dublin an effect of novobiocin (at 15 or 20 μg/ml) on wild-type recipient ability is more easily noted due to lower expression (compared to serovar Typhi) of PilV proteins by serovar Dublin grown in the absence of novobiocin (Fig. 11). The effect of novobiocin in reducing the recipient ability of wild-type serovar Dublin and the absence of such an effect when the pilV mutant of serovar Dublin was the recipient may be explained by decreases in DNA supercoiling in the presence of novobiocin (compared to the extent of supercoiling in novobiocin-free medium). This results in decreases in Rci-mediated shufflon inversion frequency and consequent increases in the synthesis of PilV proteins which, finally, cap the type IVB pili and render the pili ineffective in bacterial self-association. This process clearly cannot occur in a pilV mutant.
The S. enterica shufflon is a bacterial self-association control mechanism regulated, in turn, by the extent of DNA supercoiling. The S. enterica type IVB pili mediate bacterial self-association when the presumptive minor pilus proteins PilV1 and PilV2 are not expressed (12, 13). We have suggested that control of PilV protein synthesis is effected by rapid DNA inversion activity of the shufflon, composed of the C-terminal part of the pilV gene and the adjacent rci gene. In the present study, the model is examined in some detail. Plasmids containing the xylE gene, in invertible DNA, and potentially transcribed from the lac promoter located in stable DNA outside the invertible DNA, were initially used in an examination of Rci function. It was shown that Rci-mediated DNA inversion per se is necessary for inhibition of expression of xylE located in the invertible DNA. Binding, without DNA scission, of Rci to a substrate site lying between the lac promoter and xylE, which might tend to inhibit the movement of RNA polymerase from the promoter into the invertible DNA, is not an adequate explanation of the data. The expression levels of xylE in invertible DNA were modulated by changes in the 19-bp Rci substrate sequences, and the changes correlated with the affinity of a purified Rci-containing fusion protein to the 19-bp sequences. Thus, higher affinity led to a faster DNA inversion rate and a lower expression of the invertible gene. An inversion event may be divided into (i) a phase in which Rci binds to the 19-bp substrate sequences (two Rci molecules/sequence); (ii) a period of formation of a complex in which the four Rci molecules, with their substrate DNA sequences and possibly accessory proteins, are apposed; and (iii) a resolution of this complex to give an inverted DNA fragment. During steps ii and iii, the DNA under inversion may be inaccessible to RNA polymerase. In intervals between the end of one inversion event and the commencement of the next such event, RNA polymerase may be able to traverse the potentially invertible DNA. When one inversion event is followed by another such event after only a short interval (fast inversion), the frequency of complete transcription of potentially invertible DNA may be low. If a pause, sufficiently long to permit RNA polymerase transcription of the potentially invertible DNA, ensues between one inversion event and the next (slow inversion), then expression of a potentially invertible gene may be measured.
As reviewed earlier, it is known that Rci is active only on supercoiled DNA. Novobiocin, an inhibitor of DNA supercoiling, was used in tests exploring possible effects of DNA supercoiling on genes fully or partly invertible by Rci. The expression of an invertible xylE gene initially increased with increasing levels of novobiocin in the growth medium. This is interpreted to mean that gradual inhibition of DNA supercoiling rendered Rci action less effective so that a lower frequency of inversion of xylE-containing DNA allowed an increased level of completion of gene transcription, leading to higher levels of expressed XylE enzyme. Novobiocin also caused decreased self-association of, particularly, wild-type serovar Dublin. This again suggested that a lower frequency of inversion of pilV-containing DNA allowed an increased level of completion of gene transcription, leading to higher levels of PilV proteins which may cap the pili, rendering them ineffective in bacterial self-association.
Use of the natural S. enterica 19-bp sequences, V1 and V2, to bracket DNA, affords Rci inversion frequencies intermediate between those seen when the DNA is bracketed by V3 or V4 19-bp sequences. It may be that the use of the V1 and V2 sequences in the S. enterica shufflon represents an evolutionary fine-tuning of shufflon response to environmental conditions.
In summary, the data presented here are in agreement with the suggestion that the rate of inversion of Rci-invertible DNA determines the expression level of genes contained within that DNA. Lower inversion frequencies afford greater opportunities for through-transcription of the invertible DNA from external promoters, such as the lac promoter in many plasmids used above, or the pil promoter in S. enterica. Supercoiling of DNA is required for Rci activity. Ultimately, therefore, environmental conditions, such as anoxia in the human gut, determine the level of self-association by pil+ strains of S. enterica.
ACKNOWLEDGMENTS
This study was supported by the Hong Kong Government Research Grants Council.
REFERENCES
1. Azaro, M. A., and A. Landy. 2002. Integrase and the Int family, p. 118-148. In N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz (ed.), Mobile DNA II. American Society for Microbiology Press, Washington, D.C.
2. Bueno, S. M., C. A. Santiviago, A. A. Murillo, J. A. Fuentes, A. N. Trombert, P. L. Rodas, P. Youderian, and G. C. Mora. 2004. Precise excision of the large pathogenicity island, SPI7, in Salmonella enterica serovar Typhi. J. Bacteriol. 186:3202-3213.
3. Collyn, F., M.-A. Lety, S. Nair, V. Escuyer, A. Ben Younes, M. Simonet, and M. Marceau. 2002. Yersinia pseudotuberculosis harbors a type IV pilus gene cluster that contributes to pathogenicity. Infect. Immun. 70:6196-6205.
4. Gyohda, A., and T. Komano. 2000. Purification and characterization of the R64 shufflon-specific recombinase. J. Bacteriol. 182:2787-2792.
5. Gyohda, A., N. Furuya, N. Kogure, and T. Komano. 2002. Sequence-specific and nonspecific binding of the Rci protein to the asymmetric recombination sites of the R64 shufflon. J. Mol. Biol. 318:975-983.
6. Kim, S.-R., and T. Komano. 1992. Nucleotide sequence of the R721 shufflon. J. Bacteriol. 174:7053-7058.
7. Komano, T., S.-R. Kim, T. Yoshida, and T. Nisioka. 1994. DNA rearrangement of the shufflon determines recipient specificity in liquid mating of IncI1 plasmid R64. J. Mol. Biol. 243:6-9.
8. Kubo, A., A. Kusukawa, and T. Komano. 1988. Nucleotide sequence of the rci gene encoding shufflon-specific DNA recombinase in the IncI1 plasmid R64: homology to the site-specific recombinases of integrase family. Mol. Gen. Genet. 213:30-35.
9. Lyczak, J. B., and G. B. Pier. 2002. Salmonella enterica serovar Typhi modulates cell surface expression of its receptor, the cystic fibrosis transmembrane conductance regulator, on the intestinal epithelium. Infect. Immun. 70:6416-6423.
10. Marx, C. J., and M. E. Lidstrom. 2001. Development of improved versatile broad-host-range vectors for use in methylotrophs and other gram-negative bacteria. Microbiology 147:2065-2075.
11. Miller, J. H. 1972. Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.
12. Morris, C., I. S. M. Tsui, C. M. C. Yip, D. K.-H. Wong, and J. Hackett. 2003. The shufflon of Salmonella enterica serovar Typhi regulates type IVB pilus-mediated bacterial self-association. Infect. Immun. 71:1141-1146.
13. Morris, C., C. K. P. Tam, T. S. Wallis, P. W. Jones, and J. Hackett. 2003. Salmonella enterica serovar Dublin strains which are Vi antigen-positive use type IVB pili for bacterial self-association and intestinal cell entry. Microb. Pathog. 35:279-284.
14. Nair, S., S. Alokam, S. Kothapalli, S. Porwollik, E. Proctor, C. Choy, M. McClelland, S.-L. Liu, and K. E. Sanderson. 2004. Salmonella enterica serovar Typhi strains from which SPI7, a 134-kilobase island with genes for Vi exopolysaccharide and other functions, has been deleted. J. Bacteriol. 186:3214-3223.
15. Parkhill, J., G. Dougan, K. D. James, N. R. Thomson, D. Pickard, J. Wain, C. Churcher, K. L. Mungall, S. D. Bentley, M. T. Holden, M. Sebaihia, S. Baker, D. Basham, K. Brooks, T. Chillingworth, P. Connerton, A. Cronin, P. Davis, R. M. Davies, L. Dowd, N. White, J. Farrar, T. Feltwell, N. Hamlin, A. Haque, T. T. Hien, S. Holroyd, K. Jagels, A. Krogh, T. S. Larsen, S. Leather, S. Moule, P. O'Gaora, C. Parry, M. Quail, K. Rutherford, M. Simmonds, J. Skelton, K. Stevens, S. Whitehead, and B. G. Barrell. 2001. Complete genome sequence of a multiple drug resistant Salmonella enterica serovar Typhi CT18. Nature 413:848-852.
16. Pier, G. B., M. Grout, T. Zaidi, G. Meluleni, S. S. Mueschenborn, G. Banting, R. Ratcliff, M. J. Evans, and W. H. Colledge. 1998. Salmonella typhi uses CFTR to enter intestinal epithelial cells. Nature 393:79-82.
17. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
18. Tam, C. K. P., J. Hackett, and C. Morris. 2004. Salmonella enterica serovar Paratyphi C carries an inactive shufflon. Infect. Immun. 72:22-28.
19. Tsui, I. S. M., C. M. C. Yip, J. Hackett, and C. Morris. 2003. The type IVB pili of Salmonella enterica serovar Typhi bind to the cystic fibrosis transmembrane conductance regulator. Infect. Immun. 71:6049-6050.
20. Zhang, X.-L., C. Morris, and J. Hackett. 1997. Molecular cloning, nucleotide sequence, and function of a site-specific recombinase encoded in the major "pathogenicity island" of Salmonella typhi. Gene 202:139-146.
21. Zhang, X.-L., I. S. M. Tsui, C. M. C. Yip, A. W. Y. Fung, D. K.-H. Wong, X. Dai, Y. Yang, J. Hackett, and C. Morris. 2000. Salmonella enterica serovar Typhi uses type IVB pili to enter human intestinal epithelial cells. Infect. Immun. 68:3067-3073.(Connie K. P. Tam, Jim Hac)
ABSTRACT
Salmonella enterica serovar Typhi and some strains (Vi+) of serovar Dublin use type IVB pili to facilitate bacterial self-association, but only when the PilV proteins (potential minor pilus proteins) are not synthesized. Pilus-mediated self-association may be important in the pathogenesis of enteric fever. We have suggested that the rate of Rci-catalyzed inversion of DNA encoding the C-terminal portions of the PilV proteins controls PilV protein synthesis. This potentially represents a novel means of transcriptional control. Here, it is initially shown that DNA inversion per se is required for inhibition of gene expression from invertible DNA. Binding, without DNA scission, of Rci to its substrate sequences on DNA cannot explain the data obtained. Next, it is shown that inversion frequencies of xylE-encoding DNA, bracketed by Rci substrate sequences, may be modulated by changes in the 19-bp consensus sequences which are essential components of Rci substrate DNA. The affinity of Rci for these sequences affects inversion frequencies, so that a greater affinity is predictive of faster inversion, and therefore less synthesis of product encoded by invertible DNA. Inversion events may inhibit transcription of DNA from external promoters. In vivo, the frequency of Rci-mediated inversion is influenced by the extent of DNA supercoiling, with increasing levels of expression of invertible genes as novobiocin inhibits DNA supercoiling and thus Rci action. This inhibition of DNA supercoiling results in increased synthesis of PilV proteins as Rci activity decreases, and, in turn, bacterial self-association (particularly in serovar Dublin) decreases.
INTRODUCTION
The Salmonella enterica pil operon is located in Salmonella pathogenicity island 7 (SPI7), which also carries the viaB gene cluster required for the synthesis of the Vi antigen (15). SPI7 is unstable in serovar Typhi, with excision possibly facilitated by long-term strain storage or growth under laboratory conditions (2, 14). The pil operons of S. enterica serovar Typhi, and some strains (Vi+) of serovar Dublin, encode type IVB pili. These pili mediate bacterial self-association, but only when the presumptive minor pilus proteins PilV1 and PilV2 are not expressed (12, 13). The N-terminal regions of both PilV proteins are identical, but the C termini differ as they are encoded by invertible DNA at the end of the pil operon. The adjacent rci gene encodes a recombinase active to effect the DNA inversion. The C-terminal part of the pilV gene and the adjacent rci gene, encoding a tyrosine site-specific recombinase, together comprise the S. enterica shufflon. The Rci protein acts on DNA sequences, including conserved 19-bp inverted repeats to invert DNA in the C-terminal portion of the pilV gene, and is active only on supercoiled DNA (4, 5, 12).
Apart from playing a role in bacterial self-association, the pil operon may also be important in attachment of invading bacteria to eukaryotic cells during infection. The type IVB structural pilin, PilS, binds to the cystic fibrosis transmembrane conductance regulator, the recognized eukaryotic cell receptor for serovar Typhi (9, 16, 19).
Since these type IVB pili-mediated events cannot be effected by serovars lacking pil, such as Salmonella enterica serovar Typhimurium, we have suggested that the expression of type IVB pili by certain serovars (particularly Typhi) might be important in explaining why only certain S. enterica serovars cause enteric fever in humans (13, 21). While serovar Paratyphi C (Vi+) carries a pil operon very similar to that of serovar Typhi, the shufflon is inactive because the 19-bp Rci consensus inverted repeat sequences have been modified to be 20 bp in length, and DNA between such sequences is not invertible by Rci. Since serovar Paratyphi is less virulent than is serovar Typhi, it is possible that the observed differences in pathogenicity are partly due to the demonstrated changes in pil (18). In the absence of tests involving the administration of serovar Typhi pil mutants to humans, the observation that inactivation of a similar pil operon in Yersinia pseudotuberculosis decreased mouse virulence, with an increase in the 50% lethal dose value of 0.7 log, is the only reported test of the effect of a pil mutation on virulence (3).
We have suggested that control of the synthesis of the PilV proteins is effected by rapid DNA inversion activity of the shufflon (7, 12, 20). Through-transcription of the pilV gene, from a promoter located outside the invertible DNA, may not be possible with rapidly inverting DNA. The idea that control of protein synthesis is effected by the rate of DNA inversion of a portion of the gene encoding the protein is novel, and worthy of further exploration. In the present study, we examine the model in some detail and offer evidence that it is an accurate representation of events in cells of pil+ strains of S. enterica.
MATERIALS AND METHODS
Materials and DNA manipulation. Reagents were of molecular biology grade. Enzymes active on DNA were obtained from either Invitrogen or Roche and were used as directed by the suppliers. X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) and IPTG (isopropyl--D-thiogalactopyranoside) were purchased from Amersham. Bio-Rad was the supplier of polyvinylidene difluoride membrane. Plasmid preparation, restriction enzyme digestion, and other DNA manipulations were done as described by Sambrook et al. (17).
Media. Luria-Bertani broth (LB) was prepared as described by Miller (11). Solid medium contained 1.5% (wt/vol) agar. Antibiotics were added, when appropriate, to 5 to 15 μg/ml (Tc), 50 μg/ml (Str), or 100 μg/ml (Ap).
Bacterial strains. Serovar Typhi J341 (Ty2 Vi–) (20) is the wild-type strain used here. The construction of site-directed pilS and pilV mutants of this strain has been described (20, 21). The wild-type serovar Dublin strain 124 (Vi+), and the pilS and pilV mutants derived therefrom, were used in earlier work (13). Serovar Typhi J341 pilS/pRU670 (Rts1::Tn1731) (Tcr) was the donor, in liquid matings, of the conjugative plasmid pRU670 (12). For use as recipients in liquid matings, spontaneous rpsL (Strr) mutants of Typhi and Dublin, obtained earlier (12, 13), were used. Escherichia coli K-12 DH5 [supE44 lacU169 (80 lacZM15) hsdR17 recA1 endA1 gyrA96 (Nalr) thi-1 relA1] was the usual host for recombinant plasmids. E. coli JM109 [F' traD36 lacIq (lacZ)M15 proA+B+/e14–(McrA–) (lac-proAB) thi gyrA96 (Nalr) endA1 hsdR17 (rK– mK–) relA1 supE44 recA1] was used as the host for plasmids carrying the xylE gene between DNA potentially invertible by Rci. The xylE gene was obtained by PCR from pCM20 (10). Plasmids carrying genes for proteins tagged with glutathione S-transferase (GST), His6, or Trx were expressed in E. coli strain BL21(DE3) [F– ompT hsdSB (rB– mB–) gal dcm (DE3); Novagen], which carries a chromosomal copy of the T7 RNA polymerase gene under control of the lacUV5 promoter.
Purification of a fusion protein containing Rci. The rci gene was subcloned between the BamHI and XhoI sites of plasmid pET-trx, which is a modification of plasmid pET-32a (Novagen), in which the S-tag and the enterokinase cleavage site have been removed. In this construct, termed pET-trx-rci, the N-terminal DNA of rci was joined in-frame to, sequentially, DNA specifying the thioredoxin (Trx) and hexahistidine (His6) fusion tags to encode a protein, Trx-His6-Rci, of 57,565 Da (514 amino acids [aa]), which includes the entire Rci sequence. The 155-aa control protein (Trx-His6) expressed from plasmid pET-trx was 16,689 Da in molecular size. In both cases, the T7 promoter drives expression of the genes encoding the fusion proteins. To prepare the proteins, 5-ml amounts of stationary-phase cultures of E. coli BL21(DE3) hosting pET-trx-rci or pET-trx were inoculated into 1-liter amounts of fresh LB medium with Ap and shaken at 37°C to optical cell densities at 600 nm of ca. 0.4. IPTG was added to 0.4 mM, and the cultures were shaken at 30°C for a further 3 h. After centrifugation (5,887 x g for 30 min at 4°C), bacterial pellets were frozen at –80°C to facilitate membrane rupture. Pelleted material from 500 ml of culture was thawed and resuspended in 30 ml of binding buffer (20 mM Tris-HCl, 500 mM NaCl, 5 mM imidazole [pH 8]), and protease inhibitors (pepstatin to 1 μg/ml, leupeptin to 1 μg/ml, and phenylmethylsulfonyl fluoride to 1 mM) were added. The resuspended bacterial pellet was placed on ice and subjected to sonication by using a B. Braun Labsonic U machine with a power level of ca. 100 W and a duty cycle of 0.3 s, for 10 cycles each of 30 s with 30 s of a cooling pause between each cycle. The sonicate was centrifuged, and the supernatant (30 ml) was collected for Ni-iminodiacetic acid protein affinity purification under native conditions. Agarose beads with immobilized iminodiacetic acid were purchased from Novagen and charged with Ni2+ to permit binding of fusion proteins with the His6 tag. The beads, prepacked in a small column, were extensively washed with binding buffer (above), and the 30 ml of supernatant was loaded to 1.5 ml of packed beads and held at 4°C for 1 h. The column was washed with 400 ml of binding buffer, and fusion proteins eluted by gradually increasing the concentration of imidazole in elution buffers (all 20 mM Tris-HCl, 500 mM NaCl, with 40, 60, 100, 250, or 500 mM imidazole [pH 8]). For each imidazole concentration, 5 to 7.5 ml of elution buffer was used. Fusion proteins were detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) of aliquots of the eluates. SDS-PAGE was performed on polyacrylamide at 5% (wt/vol) stacking gel to 15% (wt/vol) separating gel. Immunoblotting with mouse monoclonal immunoglobulin G (IgG) anti-His6 antibody (Santa Cruz Biotechnology) was performed to confirm the identities of purified proteins. The primary antibody was used at a dilution of 1:500 in TBST (20 mM Tris-HCl, 140 mM NaCl, 0.1% [wt/vol] Tween 20). The secondary antibody (Amersham), used at a dilution of 1:2,000 in TBST, was a sheep anti-mouse IgG peroxidase-linked antibody. Development of immunoblots employed the enhanced chemiluminescence system (Amersham), with the substrate H2O2. Fractions of interest were dialyzed against buffer D (20 mM Tris-HCl, 50 mM NaCl [pH 8]) and stored at –80°C in 10% (vol/vol) glycerol. The fusion protein Trx-His6-Rci was eluted in fractions containing imidazole at 250 and 500 mM. Purified control Trx-His6 protein was eluted in fractions containing 500 mM imidazole (Fig. 1).
Detection of Rci self-association. A pull-down assay was used to detect self-association of Rci protein. The fusion proteins GST-His6-Rci, GST-His6, and His6-Rci were used. To prepare GST-His6-Rci, the rci gene was first cloned between the BamHI and XhoI sites of pET-42a (Novagen) to give pET-42a-rci. The GST-His6 protein was encoded by pET-42a. To construct a plasmid encoding His6-Rci, the rci gene was cloned between the BamHI and XhoI sites of plasmid pET-6H modified from pET-32a (Novagen) with the Trx-tag, S-tag, and enterokinase cleavage site removed to give plasmid pET-6H-rci. The His6-Rci protein was prepared in a manner similar to that described above for the Trx-His6-Rci protein, except that His6-Rci was eluted with a buffer containing only 100 mM imidazole. To prepare the GST-His6-Rci and GST-His6 proteins, E. coli BL21(DE3) strains hosting pET-42a-rci or pET-42a were grown to an optical density at 600 nm of ca. 0.7 under shaking conditions at 37°C and then induced with IPTG at 0.5 mM, followed by further shaking growth at 25°C for 3 to 4 h prior to harvest. Processing of the bacterial pellet to obtain purified fusion proteins was similar to that described above for the Trx-His6-Rci and Trx-His6 proteins except (i) phosphate-buffered saline (pH 7.4) was substituted for binding buffer, (ii) the affinity beads used were Glutathione-Sepharose beads from Amersham, and (iii) the elution buffer was 10 mM reduced glutathione in 50 mM Tris-HCl (pH 8.0). All three protein samples were dialyzed against buffer D (above), and the GST-His6-Rci and His6-Rci fusion proteins were concentrated by ultrafiltration.
Two tubes received, in a final volume of 400 μl of buffer A (20 mM Tris-HCl, 50 mM NaCl, 0.05% [vol/vol] Triton X-100, 1% [wt/vol] bovine serum albumin [pH 8.0]), either (i) GST-His6-Rci (16 μg) and His6-Rci (10 μg) (test) or (ii) GST-His6 (7 μg) and His6-Rci (10 μg) (control). The mixtures were gently incubated at 4°C for 4 h on a rotator, and then each tube received 50 μl of a suspension of GST-Sepharose beads (50% [vol/vol], preincubated in buffer A for 4 h). The mixtures were gently agitated at 4°C for a further 2 h. After brief centrifugation the supernatant was discarded, and the proteins in the bead-containing pellets were eluted with 40 μl of SDS-PAGE loading buffer. After gel electrophoresis of 10-μl samples, an anti-His6 antibody (above) was used as the primary antibody in immunoblotting.
Nature of DNA sequences in the pilV region and the terminology used to describe them. In wild-type serovars Typhi and Dublin, DNA invertible by Rci lies between two 19-bp inverted repeat sequences, differing by a single bp, and termed the V1 or V2 sequences (12, 13). Use of the term "V1 orientation" indicates that a promoter external to potentially invertible DNA reads first through the V1 sequence, across invertible DNA, and out through the V2 sequence, while use of the term "V2 orientation" indicates that the locations of the repeat sequences are reversed. In all plasmids in the present study containing 19-bp sequences recognized by Rci, the 19-bp sequence located just before the xylE gene and forming part of the leftmost inversion site in plasmids with two 19-bp sequences is preceded by the 12-bp sequence TGCCACACTTTC, which is the sequence found naturally in serovars Typhi and Dublin (see GenBank AF000001). The 31-bp sequence composed of the 12-bp sequence and the 19-bp sequence together form the Rci substrate site on the DNA (5). Likewise, the 19-bp sequence located just before the rci gene and forming part of the rightmost inversion site in plasmids with two 19-bp sequences is followed by the natural S. enterica 12-bp sequence GTATGTCCTTAC to form a 31-bp Rci substrate site. The use of the natural S. enterica 12-bp sequences as components of the 31-bp Rci recognition sites is important, since changes in these 12-bp sequences can dramatically affect the steady-state levels of the two possible orientations of invertible DNA (5). Thus, pUST164 (18) differs from pUST170 (the present study) only in the replacement (in pUST164) of the leftmost natural S. enterica 12-bp sequence with unrelated vector DNA. In pUST170, >90% of the DNA is in the orientation permissive for transcription (this work), whereas ca. 100% of the DNA is in the other orientation in pUST164 (18).
Measurement of XylE activity. Expression of the reporter xylE gene, behind the lac promoter, was used as a measure of DNA transcription. The xylE gene of pCM20 has been widely used in such an experimental context (10). This assay, using sonicates of stationary-growth-phase cultures of E. coli JM109-based strains, has been described (12). Levels of catechol 2,3-dioxygenase were assayed. Catechol was from Aldrich, Inc. Enzyme units are expressed as nanomoles of 2-hydroxymuconic semialdehyde produced per minute per microgram of protein. A Bio-Rad protein assay kit was used for estimation of protein concentration.
Electrophoretic mobility shift assay. Four picomoles of double-stranded DNA molecules (termed V1, V2, V3, or V4 19-bp sequences [described below]) were radiolabeled with 40 μCi of Redivue adenosine 5'-[32P-PO4]triphosphate (Amersham) by using T4 polynucleotide kinase (Amersham). Purification of the radiolabeled DNA molecules was effected with the QiaQuick nucleotide removal kit (QIAGEN). For each DNA-binding reaction, 50 ng of Trx-His6-Rci protein (test) or Trx-His6 protein (control) was added to a 20-μl final reaction mixture containing 0.02 pmol (5,000 cpm) of labeled probe, 1 μg of bovine serum albumin/μl, and 4 mM spermidine in binding buffer (30 mM Tris-HCl, 80 mM KCl, 15 mM NaCl, 10% [vol/vol] glycerol [pH 8.0]), with or without unlabeled probe DNA in molar excesses of 5-, 10-, 20-, or 30-fold. After incubation for 30 min at room temperature, the samples were loaded onto a 6% (wt/vol) native polyacrylamide gel, and electrophoresis was performed at 300 V for 2.5 h at 4°C. The gel was dried and exposed by autoradiography at –80°C to Fuji Super RX X-ray film with intensifying screens. To quantify labeled probe-protein complexes formed, the films were scanned and analyzed with a Bio-Rad imaging densitometer, which measured units of band intensity on a grayscale as optical densities at 400 to 750 nm per square centimeter.
Effect of novobiocin on XylE expression from invertible DNA. E. coli JM109-based strains hosting either pUST170 (test plasmid) or pUST168 (positive control plasmid) were grown in shaking (220 rpm) culture at 37°C for 24 h, to stationary phase, with Ap, IPTG (0.2 mM), and various concentrations of novobiocin (0 to 150 μg/ml). The cells were sonicated and XylE assays performed on the cell lysates.
Liquid mating tests. These tests were conducted as described previously (12). Briefly, a donor strain (a pilS mutant of serovar Typhi carrying a conjugative Tcr plasmid, and Strs) in the logarithmic growth phase, and recipients (various Strr strains of serovars Typhi or Dublin) in the stationary growth phase were mixed in a 10:1 ratio of recipients to donors, and liquid matings proceeded for 2 h (serovar Typhi recipients) or 4 h (serovar Dublin recipients) at 30°C. Dilutions of the mating mixtures were plated with Tc (selecting for the R-factor) and Str (counterselecting the donor) for colony enumeration. In some tests, novobiocin (15 or 20 μg/ml) was present during growth of the recipients and during the matings.
RESULTS AND DISCUSSION
Inversion rate of xylE-containing DNA regulates XylE expression. E. coli JM109-based strains hosting various pUC19-derived plasmids with a promoterless xylE gene behind the lac promoter and between DNA sequences potentially invertible by Rci were constructed (Fig. 2A). Plasmids pUST168 and pUST169 are positive and negative controls, respectively. In either plasmid, the xylE gene is noninvertible due to a deletion in rci. In pUST168, the xylE gene is fixed in the orientation permissive for transcription; the reverse orientation is found in pUST169. In plasmids pUST170 and pUST171 (in which the xylE gene is in the V1 or V2 orientation, respectively), the xylE gene is potentially invertible. The favored orientation in either plasmid is the orientation in which xylE may be transcribed by the lac promoter (Fig. 2B). This indicates that the 19-bp Rci substrate sequences do not determine the preferred orientation of DNA inserted between such sequences. Instead, the sequence of the contained DNA appears to dictate the steady-state levels of the two possible orientations. Expression of XylE by a strain containing pUST170 was very low (Fig. 2C), despite the fact that >90% of the invertible DNA, in the steady-state, was in the orientation permissive for xylE transcription (Fig. 2B). The same was true of a strain containing pUST171 (Fig. 2B and C). Poor expression of the xylE gene indicated a defect in transcription, since a viable lac promoter preceded the reporter gene. The low expression of the xylE gene was accompanied by active inversion. Thus, transformation of preparations of either pUST170 or pUST171 into E. coli JM109, followed by examination of plasmid DNA isolated from 50 separate colonies or transformation, showed that all plasmids had achieved the equilibrium orientations of Fig. 2B, even though the transforming DNA was a mixture (in the ratio of ca. 9:1) of two distinct plasmids.
It is important to note that in pUST170 and pUST171, transcription of the rci gene will occur from the rci promoter only. Read-through transcription of rci from the lac promoter will not occur in these plasmids because the failure, in strains hosting these plasmids, to transcribe the xylE gene from the lac promoter precludes transcription of a gene (rci) located further downstream. Therefore, the bias in orientation of the insert DNA of pUST170 and pUST171 should not be due to an effect of the lac promoter on Rci expression level. To confirm this, insert DNA of pUST168, pUST170, and pUST171 was cloned into pUC18 (which placed the lac promoter to the right of the rci gene), and the lac promoter excised from these plasmids by removal of a 364-bp PciI-Asp718 fragment. It was confirmed (i) that none of the resulting plasmids expressed any XylE activity and (ii) that the equilibrium levels of the two possible insert orientations in the plasmids obtained from pUST170 and pUST171 were identical to those seen in pUST170 and pUST171 (data not shown).
The pUST170/pUST171 data suggested two possibilities. First, it might be that an Rci-mediated rapid rate of DNA inversion inhibited transcription of the xylE gene, since RNA polymerase might not effectively transcribe DNA in the process of undergoing inversion. Second, it could be that Rci binding to a substrate DNA region including the leftmost 19-bp fragment (V1 in pUST170, for example) inhibited through-transcription of xylE from the lac promoter. The rci genes of plasmids pUST170 and pUST171 were then subjected to site-directed mutagenesis to replace Tyr272 by Phe (rci encodes a protein of 383 aa). This change in an amino acid residue essential for Rci activity (1) should eliminate DNA inversion activity by Rci but should not destroy the ability of Rci to bind to substrate DNA (4). Plasmids with previously invertible DNA fixed in either orientation were obtained, showing that the inversion activity of Rci was indeed eliminated by this mutation. Plasmids pUST172 and pUST173, in which the xylE gene is fixed in the productive orientation, were derived from pUST170 and pUST171, respectively (Fig. 2B), and strains hosting these plasmids produced high levels of XylE (Fig. 2C). This observation argued against the possibility that binding per se of Rci to DNA between the lac promoter and xylE, without DNA scission was sufficient to reduce xylE transcription. Instead, it seemed that Rci-mediated inversion activity might be essential to the reduction of XylE levels (from that of E. coli JM109/pUST168) noted in strains hosting pUST170 or pUST171.
It remained possible, however, that the Tyr272 to Phe mutation in rci reduced (compared to wild-type Rci) the binding of Rci to DNA containing a 19-bp substrate sequence. Mutations in the C-terminal region of the prototype tyrosine recombinase, integrase, may reduce the affinity of the enzyme for its substrate sites in DNA (1). The task thus was to construct a plasmid with a wild-type rci gene and a wild-type xylE gene, bracketed by wild-type 19-bp Rci substrate sequences, but to somehow render the DNA between the 19-bp sequences noninvertible. We have established (unpublished data) that the introduction of DNA with the potential to form secondary structure into hitherto-invertible DNA eliminates Rci-mediated inversion of that DNA. Therefore, the trpA terminator was inserted, after the xylE gene, into DNA which was previously invertible (Fig. 3A). Plasmids pUST174 and pUST175 both carry this modification and have the hitherto-invertible DNA in the V1 or V2 orientations, respectively. The DNA between the 19-bp sequences was noninvertible (Fig. 3B), and E. coli JM109-based strains hosting either plasmid showed high levels of XylE activity (Fig. 3C). These data strongly suggested that inversion per se was required for the observed inhibition of XylE expression from plasmids with invertible xylE genes compared to the XylE levels synthesized by the positive control E. coli JM109/pUST168. It was clear that binding of Rci to the leftmost 19-bp sequence could not inhibit xylE transcription from the lac promoter. However, as a final control, it was necessary to show that the activity of the rci promoter was not affected by the upstream insertion of DNA containing the trpA terminator (Fig. 4). Plasmid pUST176 was constructed to have the rci promoter, with 65 bp of natural S. enterica DNA sequence upstream of the –35 sequence, driving the xylE gene, whereas plasmid pUST177 contained the same insertion of DNA with the trpA terminator used in the construction of plasmid pUST174. Assay of XylE levels (performed three times, each time in duplicate) in E. coli JM109 strains hosting pUST176 or pUST177 yielded values of 11.9 (standard deviation of 1.5) or 12.7 (standard deviation of 0.7) enzyme units, respectively (ca. 20% of the value shown by an induced lac promoter in other experiments). These values did not differ significantly. It may therefore be concluded that insertion of the trpA terminator 211 bp upstream of the rci –35 sequence did not affect the activity of the rci promoter. To confirm that Rci expression was identical from pUST170 and pUST174, the His6 epitope tag was added to the C termini of the rci genes of these plasmids, and Rci expression levels in lysates of cells hosting the plasmids compared by immunoblotting, after SDS-PAGE, of lysate dilutions, using the mouse monoclonal IgG anti-His6 as primary antibody. The Rci expression levels in strains hosting either plasmid were very low (ca. 850 molecules/cell, which was at the limit of detection of the assay), due to (negative) autoregulation of rci transcription by binding of Rci to the –35 sequence of its own promoter, which is located in the rightmost 19-bp substrate sequence (Fig. 4 and unpublished data). There was no difference in Rci expression levels between strains hosting the rci-tagged derivatives of pUST170 or pUST174.
Our four principal findings can be summarized as follows. (i) In pUST170, pUST171, pUST174, and pUST175, transcription of the rci gene occurs from the rci promoter only. As mentioned above, read-through transcription of rci from the lac promoter will not occur in pUST170 or pUST171 because the failure, in strains hosting these plasmids, to transcribe the xylE gene from the lac promoter precludes transcription of the rci gene located further downstream. The lac promoter cannot transcribe rci in pUST174 or pUST175 because a transcriptional terminator is located between xylE and rci. (ii) Insertion of a transcriptional terminator 211 bp upstream of the rci –35 sequence does not affect the activity of the rci promoter. Therefore, Rci protein expression from plasmids pUST174 and pUST175 may be expected to be at levels comparable to those from similar plasmids (pUST170 and pUST171) lacking the DNA insertion with the trpA terminator. The elimination of DNA inversion caused by the trpA insertion is presumably due to a change in DNA topology. (iii) The fact that the active site Tyr272-to-Phe mutation of plasmids pUST172 or pUST173 eliminated Rci activity but still permitted high XylE expression in strains hosting these plasmids suggested that DNA inversion per se was required to inhibit XylE expression from invertible DNA compared to XylE levels synthesized by the positive control strain E. coli JM109/pUST168. However, the Tyr272-to-Phe mutation might also conceivably lessen the binding of Rci to its substrate sequences (1). Hence, inhibition of XylE synthesis by binding, without DNA scission, of Rci to its leftmost substrate sequence (between the lac promoter and xylE) could not be ruled out. (iv) The Rci expressed from plasmids pUST174 and pUST175 is, however, wild type. Therefore, the fact that E. coli JM109 strains hosting plasmids pUST174 or pUST175 express high XylE levels compared to XylE levels synthesized by the positive control strain E. coli JM109/pUST168 indicates that DNA inversion per se is required to inhibit XylE expression from invertible DNA.
The affinity of Rci for DNA containing 19-bp substrate sequences affects DNA inversion rate. If the rate of Rci inversion activity controls the expression of invertible DNA carrying the xylE gene, it follows that modulation of this rate should be reflected in variation of XylE expression levels. An obvious way in which to attempt to modulate Rci activity is to vary the 19-bp Rci substrate sequences. Plasmid R64 contains five such sequences (there are seven 19-bp sequences in the R64 shufflon; three are identical) (5), whereas the plasmid R721 shufflon (which contains six 19-bp sequences in all) (6) contains one 19-bp sequence which differs from both the R64 sequences and those of S. enterica. A total of at least eight distinct 19-bp sequences which are actual or potential substrates for S. enterica Rci is thus known. All eight sequences were examined for relative Rci affinities by the techniques to be described below. Together with the natural S. enterica V1 and V2 sequences, two R64-derived sequences (Fig. 5), termed V3 and V4, were chosen for detailed examination, since they represented the extremes (high and low, respectively) of affinity in Rci affinity assays. It is important to note that although the –35 sequence of the rci promoter (8) is contained in these 19-bp sequences (Fig. 5), this sequence is not affected by the nucleotide differences between the four 19-bp sequences. Possible effects of differences in these 19-bp sequences on Rci activity should therefore not be due to differences in rci transcription.
Plasmids pUST178-pUST181 (Fig. 6A) contain the xylE gene bracketed by 19-bp V1, V2, V3, or V4 pairs, respectively. In all four cases, DNA inversion activity may be seen (Fig. 6B), with the steady-state levels of DNA in the orientation permissive for xylE transcription being high and rather uniform in the four E. coli JM109 strains hosting the various plasmids. The XylE expression levels differed between strains, however (Fig. 6C), and were influenced by the nature of the 19-bp inverted repeats bracketing the xylE gene, decreasing in the order V4 > V2 > V1 > V3.
Next, experiments involving the affinity of a purified fusion protein containing full-length Rci for 19-bp sequences forming portions of Rci substrate sites on DNA were designed. Since Rci cleaves DNA as a dimer (5), it was useful to confirm that Rci-containing fusion protein monomers retained the ability to interact in solution. This was determined to be the case (Fig. 7). Pulldown tests, with beads binding GST-containing proteins, showed that GST-His6-Rci protein bound to His6-Rci in solution, thus indicating that the Rci component of Trx-His6-Rci (Fig. 1) should function like untagged Rci in affinity tests with Rci substrate sequences of DNA.
The affinity of the purified Rci-containing fusion protein to the different 19-bp sequences was ranked in an assay in which an excess of unlabeled DNA containing various 19-bp sequences was used to compete with labeled DNA associating with Rci. Complexes of labeled DNA and Rci were detected by autoradiography after gel electrophoresis (Fig. 8 is the example with labeled V1-containing DNA). The abilities of unlabeled DNA fragments containing 19-bp sequences to compete with a labeled V1-containing sequence for purified Rci dimer could be ranked in the order: V3 > V1 > V2 > V4 (Fig. 8). This is particularly clear when the intensities of the labeled bands in the presence of a 30-fold excess of unlabeled challenge DNA are examined (Fig. 8). The challenge effectiveness ranking of V3 > V1 > V2 > V4 was seen not only when the labeled sequence contained the V1 19-bp sequence but also when the labeled sequence was any of the other three 19-bp sequences (V2, V3, or V4) (Fig. 9). The relative affinities of different 19-bp sequences for Rci explains the differences in XylE expression levels in the four E. coli JM109 strains hosting pUST178-pUST181. The stronger the affinity of a 19-bp substrate sequence for Rci, the less XylE is produced in an E. coli JM109 strain hosting plasmids such as pUST178-pUST181. Binding, without DNA scission, of Rci to its substrate sequences cannot explain these data, as discussed above. Instead, the data strongly suggest that the relative affinity of Rci for DNA containing a 19-bp substrate sequence determines, in turn, the frequency of inversion, so that greater affinity results in higher inversion frequency and thus less through-transcription of xylE from the lac promoter, resulting in lower XylE levels.
Inhibition of DNA supercoiling affects Rci activities. In S. enterica serovars Typhi and Dublin, portions of the genes encoding the PilV1 and PilV2 proteins are located in DNA invertible by Rci (13, 20) and will therefore not be synthesized when Rci-mediated inversion activity is high. We have suggested that DNA supercoiling, required for Rci activity (4, 5, 12), may ultimately regulate shufflon activity, resulting in PilV-free pili when inversion activity is high, and PilV-tipped pili when such activity is low. In turn, bacterial self-association, mediated by PilV-free pili, may therefore occur preferentially under circumstances permissive for DNA supercoiling, such as in the anaerobic environment of the human gut, as a prelude to bacterial invasion.
Since novobiocin inhibits DNA supercoiling, it was of interest to examine possible effects of the drug on quantifiable Rci functions. First, E. coli JM109 strains hosting either pUST170 (Fig. 2A; the plasmid contains a xylE gene, in the V1 orientation, invertible by Rci) or pUST168 (Fig. 2A; positive control plasmid) were grown in the presence of various amounts of novobiocin and XylE production levels assayed (Fig. 10). Increasing amounts of novobiocin decreased XylE production gradually in E. coli JM109/pUST168, presumably reflecting an inhibition by the drug of gyrase activities useful to facilitate efficient DNA transcription. With E. coli JM109/pUST170, however, the low XylE levels synthesized in the absence of novobiocin increased, by just over a factor of 2, as novobiocin concentrations increased, to peak when the novobiocin level was 75 μg/ml (Fig. 10). This suggested that a gradual reduction in DNA supercoiling inhibited Rci activity and thus allowed more transcription of the xylE gene. At novobiocin levels greater than 75 μg/ml, XylE production decreased, again presumably because of adverse effects of novobiocin on general transcription.
As previously described (12), a liquid mating assay, in which potentially self-associating bacterial recipients are in 10-fold excess to donors, provides a measure of bacterial self-association, since the recipients tend to enmesh donor bacteria in a developing pellet. This entrapment of donors is reflected in higher transfer of a conjugative plasmid, from donor to recipient, than is the case when the recipient strains do not self-associate. A pilS donor of serovar Typhi was used in these tests, since this strain does not make pili and therefore cannot contribute to any pilus-mediated bacterial association. The pilS-to-pilS background transfer frequencies in individual tests ranged from 3 x 10–3 to 11 x 10–3/donor bacterium, as before (12), and mating frequencies with non-pilS recipients are shown as multiples of these figures (Fig. 11). In liquid mating assays, in the absence of novobiocin, the pili mediated bacterial self-association in both the wild-type and pilV strains of serovars Typhi and Dublin, as shown by the fact that the transfer frequencies to such strains were higher than those to pilS mutants (Fig. 11) (12, 13). The pilV mutants were more effective recipients than were the wild-type strains, since the pilV mutants never make PilV proteins and therefore self-associate to levels greater than those of the wild-type strains. In wild-type bacteria, PilV protein synthesis may occur when the shufflon inversion frequency is sufficiently low to permit transcription of the shufflon-contained portions of the pilV genes. When novobiocin was added to the mating mixtures, the plasmid transfer frequencies to pilV mutants were not significantly altered (Fig. 11). With the wild-type strains as recipients, however, novobiocin reduced recipient ability. This reduction was barely significant (the standard deviation error bars almost overlap) when wild-type serovar Typhi was the recipient but was much clearer when wild-type serovar Dublin was used in this capacity (Fig. 11). This is due to the fact that the recipient ability of wild-type serovar Typhi, in the absence of novobiocin, is already low (Fig. 11) (13), being reproducibly about half that of serovar Dublin. This may be due to a lower level of shufflon inversion in laboratory-grown serovar Typhi compared to serovar Dublin. Thus, it may be suggested that in serovar Dublin an effect of novobiocin (at 15 or 20 μg/ml) on wild-type recipient ability is more easily noted due to lower expression (compared to serovar Typhi) of PilV proteins by serovar Dublin grown in the absence of novobiocin (Fig. 11). The effect of novobiocin in reducing the recipient ability of wild-type serovar Dublin and the absence of such an effect when the pilV mutant of serovar Dublin was the recipient may be explained by decreases in DNA supercoiling in the presence of novobiocin (compared to the extent of supercoiling in novobiocin-free medium). This results in decreases in Rci-mediated shufflon inversion frequency and consequent increases in the synthesis of PilV proteins which, finally, cap the type IVB pili and render the pili ineffective in bacterial self-association. This process clearly cannot occur in a pilV mutant.
The S. enterica shufflon is a bacterial self-association control mechanism regulated, in turn, by the extent of DNA supercoiling. The S. enterica type IVB pili mediate bacterial self-association when the presumptive minor pilus proteins PilV1 and PilV2 are not expressed (12, 13). We have suggested that control of PilV protein synthesis is effected by rapid DNA inversion activity of the shufflon, composed of the C-terminal part of the pilV gene and the adjacent rci gene. In the present study, the model is examined in some detail. Plasmids containing the xylE gene, in invertible DNA, and potentially transcribed from the lac promoter located in stable DNA outside the invertible DNA, were initially used in an examination of Rci function. It was shown that Rci-mediated DNA inversion per se is necessary for inhibition of expression of xylE located in the invertible DNA. Binding, without DNA scission, of Rci to a substrate site lying between the lac promoter and xylE, which might tend to inhibit the movement of RNA polymerase from the promoter into the invertible DNA, is not an adequate explanation of the data. The expression levels of xylE in invertible DNA were modulated by changes in the 19-bp Rci substrate sequences, and the changes correlated with the affinity of a purified Rci-containing fusion protein to the 19-bp sequences. Thus, higher affinity led to a faster DNA inversion rate and a lower expression of the invertible gene. An inversion event may be divided into (i) a phase in which Rci binds to the 19-bp substrate sequences (two Rci molecules/sequence); (ii) a period of formation of a complex in which the four Rci molecules, with their substrate DNA sequences and possibly accessory proteins, are apposed; and (iii) a resolution of this complex to give an inverted DNA fragment. During steps ii and iii, the DNA under inversion may be inaccessible to RNA polymerase. In intervals between the end of one inversion event and the commencement of the next such event, RNA polymerase may be able to traverse the potentially invertible DNA. When one inversion event is followed by another such event after only a short interval (fast inversion), the frequency of complete transcription of potentially invertible DNA may be low. If a pause, sufficiently long to permit RNA polymerase transcription of the potentially invertible DNA, ensues between one inversion event and the next (slow inversion), then expression of a potentially invertible gene may be measured.
As reviewed earlier, it is known that Rci is active only on supercoiled DNA. Novobiocin, an inhibitor of DNA supercoiling, was used in tests exploring possible effects of DNA supercoiling on genes fully or partly invertible by Rci. The expression of an invertible xylE gene initially increased with increasing levels of novobiocin in the growth medium. This is interpreted to mean that gradual inhibition of DNA supercoiling rendered Rci action less effective so that a lower frequency of inversion of xylE-containing DNA allowed an increased level of completion of gene transcription, leading to higher levels of expressed XylE enzyme. Novobiocin also caused decreased self-association of, particularly, wild-type serovar Dublin. This again suggested that a lower frequency of inversion of pilV-containing DNA allowed an increased level of completion of gene transcription, leading to higher levels of PilV proteins which may cap the pili, rendering them ineffective in bacterial self-association.
Use of the natural S. enterica 19-bp sequences, V1 and V2, to bracket DNA, affords Rci inversion frequencies intermediate between those seen when the DNA is bracketed by V3 or V4 19-bp sequences. It may be that the use of the V1 and V2 sequences in the S. enterica shufflon represents an evolutionary fine-tuning of shufflon response to environmental conditions.
In summary, the data presented here are in agreement with the suggestion that the rate of inversion of Rci-invertible DNA determines the expression level of genes contained within that DNA. Lower inversion frequencies afford greater opportunities for through-transcription of the invertible DNA from external promoters, such as the lac promoter in many plasmids used above, or the pil promoter in S. enterica. Supercoiling of DNA is required for Rci activity. Ultimately, therefore, environmental conditions, such as anoxia in the human gut, determine the level of self-association by pil+ strains of S. enterica.
ACKNOWLEDGMENTS
This study was supported by the Hong Kong Government Research Grants Council.
REFERENCES
1. Azaro, M. A., and A. Landy. 2002. Integrase and the Int family, p. 118-148. In N. L. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz (ed.), Mobile DNA II. American Society for Microbiology Press, Washington, D.C.
2. Bueno, S. M., C. A. Santiviago, A. A. Murillo, J. A. Fuentes, A. N. Trombert, P. L. Rodas, P. Youderian, and G. C. Mora. 2004. Precise excision of the large pathogenicity island, SPI7, in Salmonella enterica serovar Typhi. J. Bacteriol. 186:3202-3213.
3. Collyn, F., M.-A. Lety, S. Nair, V. Escuyer, A. Ben Younes, M. Simonet, and M. Marceau. 2002. Yersinia pseudotuberculosis harbors a type IV pilus gene cluster that contributes to pathogenicity. Infect. Immun. 70:6196-6205.
4. Gyohda, A., and T. Komano. 2000. Purification and characterization of the R64 shufflon-specific recombinase. J. Bacteriol. 182:2787-2792.
5. Gyohda, A., N. Furuya, N. Kogure, and T. Komano. 2002. Sequence-specific and nonspecific binding of the Rci protein to the asymmetric recombination sites of the R64 shufflon. J. Mol. Biol. 318:975-983.
6. Kim, S.-R., and T. Komano. 1992. Nucleotide sequence of the R721 shufflon. J. Bacteriol. 174:7053-7058.
7. Komano, T., S.-R. Kim, T. Yoshida, and T. Nisioka. 1994. DNA rearrangement of the shufflon determines recipient specificity in liquid mating of IncI1 plasmid R64. J. Mol. Biol. 243:6-9.
8. Kubo, A., A. Kusukawa, and T. Komano. 1988. Nucleotide sequence of the rci gene encoding shufflon-specific DNA recombinase in the IncI1 plasmid R64: homology to the site-specific recombinases of integrase family. Mol. Gen. Genet. 213:30-35.
9. Lyczak, J. B., and G. B. Pier. 2002. Salmonella enterica serovar Typhi modulates cell surface expression of its receptor, the cystic fibrosis transmembrane conductance regulator, on the intestinal epithelium. Infect. Immun. 70:6416-6423.
10. Marx, C. J., and M. E. Lidstrom. 2001. Development of improved versatile broad-host-range vectors for use in methylotrophs and other gram-negative bacteria. Microbiology 147:2065-2075.
11. Miller, J. H. 1972. Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.
12. Morris, C., I. S. M. Tsui, C. M. C. Yip, D. K.-H. Wong, and J. Hackett. 2003. The shufflon of Salmonella enterica serovar Typhi regulates type IVB pilus-mediated bacterial self-association. Infect. Immun. 71:1141-1146.
13. Morris, C., C. K. P. Tam, T. S. Wallis, P. W. Jones, and J. Hackett. 2003. Salmonella enterica serovar Dublin strains which are Vi antigen-positive use type IVB pili for bacterial self-association and intestinal cell entry. Microb. Pathog. 35:279-284.
14. Nair, S., S. Alokam, S. Kothapalli, S. Porwollik, E. Proctor, C. Choy, M. McClelland, S.-L. Liu, and K. E. Sanderson. 2004. Salmonella enterica serovar Typhi strains from which SPI7, a 134-kilobase island with genes for Vi exopolysaccharide and other functions, has been deleted. J. Bacteriol. 186:3214-3223.
15. Parkhill, J., G. Dougan, K. D. James, N. R. Thomson, D. Pickard, J. Wain, C. Churcher, K. L. Mungall, S. D. Bentley, M. T. Holden, M. Sebaihia, S. Baker, D. Basham, K. Brooks, T. Chillingworth, P. Connerton, A. Cronin, P. Davis, R. M. Davies, L. Dowd, N. White, J. Farrar, T. Feltwell, N. Hamlin, A. Haque, T. T. Hien, S. Holroyd, K. Jagels, A. Krogh, T. S. Larsen, S. Leather, S. Moule, P. O'Gaora, C. Parry, M. Quail, K. Rutherford, M. Simmonds, J. Skelton, K. Stevens, S. Whitehead, and B. G. Barrell. 2001. Complete genome sequence of a multiple drug resistant Salmonella enterica serovar Typhi CT18. Nature 413:848-852.
16. Pier, G. B., M. Grout, T. Zaidi, G. Meluleni, S. S. Mueschenborn, G. Banting, R. Ratcliff, M. J. Evans, and W. H. Colledge. 1998. Salmonella typhi uses CFTR to enter intestinal epithelial cells. Nature 393:79-82.
17. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
18. Tam, C. K. P., J. Hackett, and C. Morris. 2004. Salmonella enterica serovar Paratyphi C carries an inactive shufflon. Infect. Immun. 72:22-28.
19. Tsui, I. S. M., C. M. C. Yip, J. Hackett, and C. Morris. 2003. The type IVB pili of Salmonella enterica serovar Typhi bind to the cystic fibrosis transmembrane conductance regulator. Infect. Immun. 71:6049-6050.
20. Zhang, X.-L., C. Morris, and J. Hackett. 1997. Molecular cloning, nucleotide sequence, and function of a site-specific recombinase encoded in the major "pathogenicity island" of Salmonella typhi. Gene 202:139-146.
21. Zhang, X.-L., I. S. M. Tsui, C. M. C. Yip, A. W. Y. Fung, D. K.-H. Wong, X. Dai, Y. Yang, J. Hackett, and C. Morris. 2000. Salmonella enterica serovar Typhi uses type IVB pili to enter human intestinal epithelial cells. Infect. Immun. 68:3067-3073.(Connie K. P. Tam, Jim Hac)