ABSTRACT
The Pseudomonas aeruginosa aacC3 gene was expressed inEscherichia coli after cloning of the single gene behind the strong tac promoter. In the originalPseudomonas strain, aacC3 is preceded bycysC; together they form a single transcription unit. The ribosome-binding site and start codon of aacC3 are involved in a putative intercistronic hairpin, the stability of which interfered with the aminoglycoside resistance level. In Northern blots, full-length transcripts comprising both cysC andaacC3 could not be detected either in the originalPseudomonas strain or in E. coli harboring a plasmid with the cloned operon. In contrast, cysCtranscripts were abundant. Cloning of the operon between thetac promoter and a transcription termination signal resulted in higher mRNA levels and phenotypic expression in E. coli. The absence of a transcription termination signal in the wild-type cysC-aacC3 sequence is associated with transcripts of heterogeneous size that were undetected in Northern blots. Our results shed more light on the expression of this gentamicin resistance determinant, although the discrepancies between its expression in E. coli and Pseudomonas are not fully solved.
The presence of aminoglycoside-modifying enzymes is the most frequent cause of bacterial resistance to aminoglycosides (8). Although many genes that encode aminoglycoside-modifying enzymes are plasmid borne or are associated with transposons, some, notably, a number of aminoglycoside acetyltransferase-encoding genes, are located on the chromosome (7, 17, 18, 26). The aacC3 gene, encoding aminoglycoside-(3)-N-acetyltransferase III [AAC(3)-III], has been found thus far only in Pseudomonas aeruginosa (5, 20). This gene was cloned and could be expressed in Pseudomonas putida KT2440 but not inEscherichia coli (31).
The aacC3 gene is the second of a polycistronic operon, since deletion of the first open reading frame or its upstream regulatory region led to the loss of gentamicin resistance (31). The nucleotide sequence of this open reading frame is homologous to the E. coli cysC gene, encoding adenosine 5′-phosphosulfate kinase, an enzyme involved in cysteine biosynthesis (24). In an E. coli-based in vitro transcription-translation system, the cysC gene product was present, whereas the aacC3 gene product could not be detected (31). In an attempt to explain the discrepant results observed with Pseudomonas and E. coli, the expression of both the aacC3 gene and thecysC gene was studied in both species.
MATERIALS AND METHODS
Bacterial strains, plasmids, and culture conditions.The strains and plasmids used in the study are listed in Table1. Bacteria were grown either in Luria-Bertani medium or in M9 minimal medium (23) with the following supplements: thiamine, l-leucine, andl-proline (all at 40 μg/ml) and either 0.4% glucose forE. coli HB101 or 0.3% citrate for P. aeruginosaand P. putida. The antibiotic concentrations used for selection were as follows: ampicillin, 100 μg/ml; gentamicin, 5 μg/ml; tetracycline, 12.5 μg/ml; streptomycin, 100 μg/ml; and trimethoprim, 200 μg/ml.
Bacterial strains and plasmids used in this study
Aminoglycoside susceptibility testing.Aminoglycoside susceptibilities were tested by the disc diffusion and agar dilution methods, both of which were described previously (29).
Preparation, analysis, and manipulation of DNA.Plasmid DNA was isolated by alkaline lysis (23) and, when necessary, was purified in a cesium chloride density gradient. All enzymes exceptAsp700 (Boehringer Mannheim), NaeI (New England Biolabs), and SuperTth DNA polymerase (SphaeroQ) were purchased from Gibco BRL and all were used as recommended by the manufacturer. Heat-shock-competent E. coli cells were transformed with plasmid DNA as described by Inoue et al. (15). DNA fragments were isolated from agarose gels with the GeneClean kit (Bio 101). Small fragments (<500 bp) were isolated from low-melting-point agarose gels (Bio-Rad) by repeated phenol extractions and subsequent precipitation.
PCR.The DNA fragments required for cloning or hybridization were obtained by PCR as described previously (30), except that 0.2 U of SuperTth DNA polymerase was used and the annealing temperature was set at 60°C. Approximately 2.5 ng of plasmid DNA was used as a template. The PCR products used for cloning were treated with proteinase K to remove DNA polymerase. After phenol-chloroform extraction, the fragments were phosphorylated by T4 polynucleotide kinase and subsequently filled in by the Klenow enzyme to enable blunt-end cloning. The primers and the sequences of the primers used in the PCRs were as follows: P1 and P2, as described by Vliegenthart et al. (31); P3A, 5′-gtcaaaagcttctgcagCATAGGGGTACACCCATGACCG-3′; P3B, 5′-gtcaaaagcttctgcagCATAGGGGTACACATATGACCG-3′; P4, 5′-gtcaaaagcttTGATGGCGCTTCCGCGGTGCG-3′; P5, 5′-TTGTCGTCGTTCATCTCGCC-3′; P19, 5′-ACCATGATTACGAATTCGAGC-3′; and P20, 5′-GTACATATTGTCGTTAGAACGC-3′ (noncomplementary nucleotides are given as lowercase letters, the introduced HindII and PstI restriction sites are underlined, and the nucleotides that introduced mutations are in boldface type).
DIG labeling and detection.Appropriate PCR fragments were randomly labeled with digoxigenin (DIG)-11-dUTP and were detected with a chemiluminescent substrate according to the manufacturer’s instructions (Boehringer Mannheim).
DNA sequencing.Double-stranded DNA sequencing was performed with plasmid DNA by using the T7 sequencing kit (Pharmacia) and [α-33P]dATP or α-35S-dATP (Amersham). Sequence-specific synthetic oligonucleotide primers (n= 17 to 33 nucleotides [nt]) were synthesized with a 391 DNA synthesizer (Applied Biosystems).
Total RNA isolation.Total RNA was isolated as described by Aiba et al. (1), with minor modifications. After phenol extractions, total RNA was selectively precipitated by adding an equal volume of 4 M LiCl and overnight incubation at 4°C (2). The RNA was pelleted by centrifugation, rinsed with 70% ethanol, and dissolved in 100 μl of diethyl pyrocarbonate-treated water. RNA yield was determined by measuring the optical densities at 260 and 280 nm.
Northern (RNA) blot analysis.Samples containing 5 μg of total RNA were analyzed by electrophoresis in 1% MOPS (morpholinepropanesulfonic acid)–formaldehyde agarose gels by standard methods (23). An RNA size marker (Gibco BRL) was always included. Ethidium bromide staining of the gels allowed visual inspection of the relative intensities and integrity of the rRNA bands. The RNA was then transferred to Hybond N+ (Amersham) filters by overnight dry capillary blotting or electroblotting in 1× MOPS at 15 mA for 1.5 h. Hybridization was performed as described by Ghosn et al. (12). Washing conditions and detection of DIG-labeled probes were as recommended by the manufacturer.
RT-PCR.Four micrograms of each RNA sample was treated with 10 U of DNase I in a volume of 20 μl in the presence of RNAguard (12.3 U; Gibco BRL) at 37°C for 20 min. Control RNA samples were processed similarly but without DNase I. After phenol extraction and ethanol precipitation, the RNA pellet was dissolved in diethyl pyrocarbonate-treated water and 100 ng of primer P2 was added. After denaturation (5 min, 94°C) the primer was annealed to the RNA (10 min, 60°C) and the reverse transcriptase (RT) reaction with SuperScript RT was then carried out in a final volume of 20 μl. A 5-μl aliquot of the mixture served as a template in a PCR as described above.
Asymmetric PCR generating single-stranded DIG-labeled DNA probes.Unidirectional PCR was performed as described by Ghosn et al. (12) with minor modifications. The first PCR was performed as described above. PCR products were isolated from an agarose gel and were used as templates in the second, asymmetric PCR. The resulting single-stranded PCR products were used as probes in hybridizations without further processing.
RESULTS
Effect of hairpin stability on aminoglycoside susceptibility.The intercistronic region between cysC and aacC3contains a putative stem-loop structure (ΔG 37 0 = −2.7 kcal/mol) involving both the ribosome-binding site (RBS) and the AUG start codon of theaacC3 gene (Fig. 1). To test the effect of hairpin stability on expression of aacC3, the gene was cloned with the stem-loop region but without the precedingcysC gene. In two separate PCRs the aacC3 gene, including either the wild-type stem-loop region (primers P3A and P4; A1 in Fig. 1) or a destabilized one (primers P3B and P4; B4 in Fig. 1), was amplified. The RBS and start codon sequences themselves were unaffected by this procedure. The PCR products were ligated into theEcoRV site of pBluescript KS(−). Sequencing of the inserts to ascertain the accuracy of the stem-loop regions revealed a third PCR product, in which the constructed hairpin was more stable than that in the wild type (B1 in Fig. 1). Subsequently, the 900-bpPstI-HindIII fragments containing theaacC3 gene and the different hairpins, hairpins A1, B1, and B4, were subcloned into vector pSPT19 and expression vector pKK223-3, resulting in the constructs pSPT-A1, pSPT-B1, pSPT-B4, pKK-A1, pKK-B1, and pKK-B4, respectively. Disc diffusion tests with E. coliHB101 strains with these plasmids showed that the AAC(3)-III phenotype was found in E. coli strains when aacC3 was positioned after the strong tac promoter (plasmids pKK-A1, pKK-B1, and pKK-B4) (9). Furthermore, the diameters of the inhibition zones decreased with decreasing hairpin stability, as shown for gentamicin in Table 2. The zone diameters of strains harboring plasmids lacking the tac promoter were considerably larger. Nevertheless, in these constructs a direct correlation between the sizes of the inhibition zones and hairpin stability was shown. The MICs of gentamicin corroborated the results of the disc diffusion tests.
(A) Genetic organization of the aac operon; nucleotides are numbered as described by Vliegenthart et al. (31). Bold arrows indicate the two open reading frames containing cysC and aacC3. The wavy arrow indicates the approximate position of the promoter. Asterisks represent the positions of the cysC and aacC3 probes used for hybridization. The position of primers used for PCR cloning are shown by arrows. Both primers have nonhybridizing tails withPstI and HindIII sites for cloning. (B) Intercistronic region containing the putative hairpin structure shown in detail. The different hairpins with various stabilities are shown: A1, wild type; B1, stabilized; B4, destabilized. The nucleotides of interest are indicated by black boxes; mutations have an asterisk. ThecysC stop codon, RBS, and aacC3 start codon are boxed. The calculated free energies, ΔG 37 0, were determined by using the fold program of the GCG Sequence Analysis Software Package (Genetics Computer Group, Madison, Wis.).
Gentamicin susceptibilities ofaacC3-containing plasmids in E. coli HB101
mRNA levels in Northern blots.The observed susceptibilities of the strains with the different plasmids could not fully explain the lack of resistance previously seen in E. coli harboring plasmid pJV305 (31). Therefore, we examined in Northern blots the levels of aacC3 mRNA in P. aeruginosaPST-I, in P. putida with pJV305, and in E. coliHB101 with several aacC3-containing plasmids (Fig.2A). In strains with plasmids harboring the single aacC3 gene (pKK-A1, pKK-B1, and pKK-B4),aacC3 transcripts were detected at 1,200 nt, the approximate length of transcripts starting at the tac promoter. When the plasmid also contained the cysC gene, noaacC3-specific transcripts were detected.
Northern blot hybridized with the aacC3 probe (A) and the cysC probe (B). Lane 1, P. aeruginosaPST-I; lanes 2 to 4, P. putida KT2440 with pJV305 (lane 2) and pLAFR3 (lane 3) and wild type (lane 4); lanes 5 to 12, E. coli HB101 with pKK-A1 (lane 5), pKK-B1 (lane 6), pKK-B4 (lane 7), pKK223-3 (lane 8), pSPT-A1 (lane 9), pSPT::KpnI (lane 10), pSPT19 (lane 11), and wild type (lane 12).
As a control, a duplicate blot was hybridized with a cysCprobe. In all strains harboring the cysC gene as part of the polycistronic operon, cysC mRNA was detected at 600 nt, the approximate length of a single-gene transcript (Fig. 2B). No signal was visible at 1,600 nt, the expected length of the polycistronic transcript. The 1,200-nt band in lane 2 of Fig. 2B corresponds to thecysC transcript starting at the lacZ promoter of the vector (27).
Detection of polycistronic mRNA using dot blots and RT-PCR.To demonstrate the presence of aacC3 mRNA in strains containing the polycistronic operon, DNase I-treated samples of total RNA were dot blotted directly onto nylon membranes. The blots were hybridized with single-stranded antisense probes recognizing the coding regions of aacC3 and cysC. Both probes yielded positive results, although the signal generated by the aacC3probe was considerably less intense than that caused by thecysC probe (data not shown).
The polycistronic character of the transcripts was further analyzed by RT-PCR with discriminating primer sets (Fig.3). Copy DNA generated from primer P2 served as the template for all subsequent, separate PCRs. In this strategy, the formation of a PCR product with primer set P1-P2 would be indicative of the presence of aacC3 mRNA, whereas a PCR product with primer set P5-P2 would prove that the transcript consisted of both cysC and aacC3. To test the completeness of DNase treatment, PCRs with primers located in vector sequences upstream from aacC3 inserts and thus in a nontranscribed region (P19 and P20) were also performed. The results (Fig.4) demonstrated that in all strains that harbor the entire aac operon either on a plasmid or on the chromosome the polycistronic transcript was present.
Schematic overview of primers used for RT-PCR. P19 and P20 are located in vector sequences upstream from aacC3inserts in plasmids pJV305 and pSPT19::KpnI, respectively. The sizes of the generated PCR products with the different primer sets were as follows: P1-P2, 185 bp; P5-P2, 1,168 bp; P19-P2, 1,886 bp; P20-P2, 1,863 bp. The wavy arrow indicates the approximate position of the promoter.
Agarose gel electrophoresis of RT-PCR products with the three primer sets outlined in Fig. 3. Lane numbers refer to the strains mentioned on the left-hand side. Lanes C1 and C2, plasmid DNA preparations of pJV305 and pSPT19::KpnI, respectively. Plus and minus signs indicate DNase I treatment or no treatment, respectively, of the sample prior to the reaction with RT.
Introduction of a transcription termination signal.The absence of aacC3 transcripts in Northern blots could be caused by size heterogeneity because of the lack of a transcription termination signal. To investigate this possibility, a 2.5-kbKpnI fragment (blunted) containing the aac operon was cloned into the PstI site (also blunted) of pKK223-3, resulting in plasmid pKKaac. In this way the transcription terminator derived from the rRNA gene rrnB (6) was placed 0.7 kb downstream from the aacC3 stop codon. To reduce the distance between the stop codon and the terminator, another plasmid, pKK52T, was constructed. In pKK52T the transcription terminator sequence is located at 52 nt beyond the aacC3 stop codon. Two other plasmids were generated during the construction of pKK52T, namely, pUCaac and pUC52T (Table 1). These also contained theaac operon with and without the transcription terminator but without the tac promoter and were therefore included in Northern blot analysis. Total RNA from E. coli HB101 harboring each of the four terminator constructs was blotted and hybridized with antisense aacC3 and cysC probes (Fig. 5). A signal at 0.6 kb generated by thecysC probe is present in all four lanes, presumably representing cysC mRNA expressed by the wild-type promoter. The 1.1-kb band seen in pKK52T and pKKaac (Fig. 5, lanes 1 and 2) corresponds to the cysC transcript starting from thetac promoter. The estimated distance between thetac promoter and the presumed wild-type promoter is 0.4 kb, which agrees well with our results. The 2.2-kb band in Fig. 5, lane 1 (pKK52T), and the 2.8-kb band in Fig. 5, lane 2 (pKKaac), found with both the cysC and the aacC3 probes, are in accord with the expected sizes of the polycistronic cysC-aacC3 mRNA transcribed from the tac promoter, which are 2.25 and 2.8 kb for these two plasmids, respectively. The difference in size corresponds to different positions of the transcription terminator.
Northern blot analysis of terminator constructs inE. coli HB101. (A) aacC3 probe; (B)cysC probe. Lane 1, pKK52T; lane 2, pKKaac; lane 3, pUC52T; lane 4, pUCaac.
Susceptibility data for these strains showed that the presence of thetac promoter and the transcription terminator resulted in the highest expression level (MIC of gentamicin for pKK52T and pKKaac, 8 μg/ml; Table 2). When only the transcription terminator was present, the MIC decreased to 4 μg/ml for pUC52T. In the absence of both transcriptional features (pUCaac and also pSPT19::KpnI), the observed expression was very low (1 μg/ml; see also inhibition zone diameters in Table 2).
DISCUSSION
Some P. aeruginosa gentamicin resistance determinants have been reported to be poorly expressed in E. coli(16, 19). We have shown that the expression ofaacC3 is not restricted to P. aeruginosa but, in contrast to earlier findings (31), can also be achieved inE. coli when aacC3 is cloned behind the strongtac promoter. High-level aminoglycoside resistance was obtained in the absence of the upstream cysC gene; when both genes were cloned behind the tac promoter, resistance was of an intermediate level. The wild-type promoter preceding thecysC gene was shown to be active in E. coli, which eliminates the possibility that expression is hampered by host heterology.
Resistance levels of E. coli varied with differences in stability of the stem-loop region preceding the aacC3 gene. Since the RBS, the start codon, and the spacing between them were unaltered by the mutations, we attribute the differences in aminoglycoside susceptibility to changes in hairpin stability. This type of secondary structure involving the RBS and sometimes also the start codon is known to interfere with translation efficiency, even when the stem-loop stability is low (−1 to −4 kcal/mol) (11, 13,28). However, our results indicate that this does not fully account for the previously observed lack of expression ofaacC3 in E. coli.
The results of the Northern blots and RT-PCR reveal that discreteaacC3 transcripts were found only in the presence of a strong transcription termination signal. Effective transcription termination leads to size homogeneity of the mRNA population (22). An intrinsic terminator structure is lacking in the wild-type sequence downstream from aacC3. Therefore, transcripts of this gene may have different sizes and are not detected in Northern blots but can be identified by RT-PCR.
Beside polycistronic transcripts, single-gene transcripts ofcysC are present in all E. coli andPseudomonas strains that harbor the gene, but it is still unclear how this transcript is generated. Possibly, the intercistronic hairpin functions as a transcription terminator (10, 25) or serves as a signal for nucleolytic processing (4). In both cases, the stem-loop structure, albeit of limited size, may protect the upstream sequence against 3′-5′ exonuclease degradation (4), whereas the downstream fragment generated by processing of the transcript is readily degraded (21).
In the search for an explanation of the discrepant resistance phenotypes of P. aeruginosa, P. putida, andE. coli, all of which harbored the same aacC3gene, the problem appeared to be more complex than anticipated. Rather than a single cause, all factors described may be involved and thus may contribute to the observed variations in aacC3 expression in different hosts.
ACKNOWLEDGMENTS
We thank Anniek van den Broek for technical assistance. The continuing support of C. P. A. van Boven is greatly appreciated.
This work was supported financially by the Stichting ter Bevordering van Medisch Microbiologisch Onderzoek, Leiden, The Netherlands.
FOOTNOTES
- Received 11 March 1998.
- Returned for modification 15 May 1998.
- Accepted 17 September 1998.
- Copyright © 1998 American Society for Microbiology
