Genetic regulation of the ramA locus and its expression in clinical isolates of Klebsiella pneumoniae

Abstract

Tigecycline resistance has been attributed to ramA overexpression and subsequent acrA upregulation. The ramA locus, originally identified in Klebsiella pneumoniae, has homologues in Enterobacter and Salmonella spp. In this study, we identify in silico that the ramR binding site is also present in Citrobacter spp. and that Enterobacter, Citrobacter and Klebsiella spp. share key regulatory elements in the control of the romA–ramA locus. RACE (rapid amplification of cDNA ends) mapping indicated that there are two promoters from which romA–ramA expression can be regulated in K. pneumoniae. Correspondingly, electrophoretic binding studies clearly showed that purified RamA and RamR proteins bind to both of these promoters. Hence, there appear to be two RamR binding sites within the Klebsiella romA–ramA locus. Like MarA, RamA binds the promoter region, implying that it might be subject to autoregulation. We have identified changes within ramR in geographically distinct clinical isolates of K. pneumoniae. Intriguingly, levels of romA and ramA expression were not uniformly affected by changes within the ramR gene, thereby supporting the dual promoter finding. Furthermore, a subset of strains sustained no changes within the ramR gene but which still overexpressed the romA–ramA genes, strongly suggesting that a secondary regulator may control ramA expression.

1. Introduction

Klebsiella pneumoniae is a major nosocomial pathogen that causes both community- and hospital-acquired infections. Strains with chromosomal and/or plasmid-mediated resistance mechanisms coupled with efflux/influx-related mutations are being increasingly identified [1–3]. The continued antimicrobial challenge of K. pneumoniae has precipitated the emergence of clones harbouring a plethora of resistance mechanisms to clinically relevant antibiotics (e.g. fluoroquinolones, third-generation cephalosporins and carbapenems) and emerging pandrug-resistant clones have left few therapeutic strategies available to combat this pathogen [4].

Tigecycline is a new glycylcycline with substantial anti-Gram-negative activity that has been introduced for the treatment of community-acquired Gram-negative infections caused by extended-spectrum β-lactamase-producing K. pneumoniae and Escherichia coli [5]. A recent study has shown that tigecycline is effective against pandrug-resistant K. pneumoniae and Enterobacter spp. Tigecycline is able to evade the classical mechanisms that confer resistance to tetracycline and minocycline, such as Tet(A–E)-mediated efflux and ribosomal protection conferred by Tet(M) [5]. However, tigecycline appears to be vulnerable to efflux by chromosomally encoded efflux pumps [6–10].

Studies into tigecycline resistance in members of the Enterobacteriaceae have shown that it is mediated via upregulation of efflux pumps that are controlled by certain regulatory loci [6–8,10]. For instance, several studies have demonstrated that tigecycline resistance results from upregulation of AraC family transcriptional regulators such as MarA or RamA, which in turn are linked to increased expression of the AcrAB efflux pump [6,7,10,11]. This pump not only functions as a clinically relevant drug exporter but has also been demonstrated to affect significantly the virulence potential of the bacterium [12,13]. The AcrAB efflux pump is locally controlled by the transcriptional repressor AcrR [14]. However, transcription factors such as RamA and MarA are able to override AcrR-mediated repression and upregulate expression of the AcrAB efflux pump [6,7,10,11]. Accordingly, strains with wild-type AcrR and increased RamA or MarA expression are linked to increased AcrAB levels [11]. Whether levels of AcrA upregulation in the absence of acrR mutations are as significant as those seen when acrR is mutated is not clear.

RamA is a member of the AraC/XylS family and is closely related to the MarA and SoxS proteins [15]. The ramA locus is only found in Klebsiella, Salmonella, Enterobacter and Citrobacter spp.; the genetic organisation in Salmonella differs from the others owing to lack of the romA gene. Like MarA, increased ramA expression is linked to upregulation of the AcrAB efflux pump, which confers a multidrug-resistant phenotype to a variety of different antibiotic classes [7,10,11,16]. This phenotype is mediated exclusively through the acrAB gene, as ramA overexpression in an acrAB-deleted strain does not produce a similar phenotype [11].

Expression of the ramA gene is controlled at a transcriptional level. Studies both in Salmonella enterica serovar Typhimurium [17] and K. pneumoniae [16] have identified a tetR-like gene, called ramR, that lies upstream of the ramA gene and acts as its repressor (Fig. 1). Of note, the stop codon of the ramR gene overlaps the start codon of the ybdF gene, implying that regulation of these genes is very likely linked. Studies both into S. Typhimurium [17] and K. pneumoniae [16] have reported that ramR mutations are directly linked to ramA overexpression. Both in Salmonella and Klebsiella, bioinformatic analyses suggest that the RamR protein binds a palindromic sequence that is either overlapping or downstream of the –10 sequence of the ramR gene [16,17]. In K. pneumoniae and Enterobacter spp., the genomic organisation of ramR, its corresponding palindromic binding sites and ramA is conserved compared with Salmonella, although these bacteria also harbour the romA gene (Fig. 1). Mutations within the ramR gene have been shown to result in ramA upregulation [16,17], however the effect of this derepression on the surrounding genes within the locus, i.e. romA, is not known.

Fig. 1.

(A) Alignment of the RamR palindromic binding sequence at the pI and pII promoter in Salmonella, Klebsiella, Enterobacter and Citrobacter spp. (B) Regulatory elements within the romA–ramA locus. The inverted palindromic sequences are in bold and indicated by arrows. The two putative promoters are depicted as pI and pII. The transcription start of ramA as mapped by 5′ RACE (rapid amplification of cDNA ends) is indicated by +1. The translation start site of both romA and ramA is indicated in bold and superscript. All promoter changes are in grey, with those hypothesised to be associated with ramA overexpression in grey and with an asterisk. IR, intergenic region.

In studies involving clinical strains, ramA upregulation is not normally the sole mechanism of resistance but appears to act in combination with other mutations, e.g. cefuroxime-resistant isolates also showed a decrease in levels of the outer membrane protein OmpK35 [18], and in fluoroquinolone-resistant isolates upregulation of the ramA gene was in association with target-specific mutations in the topoisomerase genes (e.g. gyrA and parC) and mutations within acrR, the repressor of the acrAB efflux pump [11]. Hence, the aims of this study were to elucidate the genetic regulation of the ramA locus and whether its overexpression is always linked to changes within the ramR gene.

2. Materials and methods

2.1. Clinical isolates

All isolates were identified as K. pneumoniae and were isolated from intensive care wards in hospitals within Chile, Turkey and Pakistan from 2006–2008. Three strains from Singapore characterised in a previous study [11] as well as the original strains (Ecl8 and Ecl8Mdr1) described by George et al. [15] were also included. Other strains used as controls in this study were: for validation of the AcrA Western blot analyses, AG100A (deleted for acrA) and AG100B (deleted for acrR, AcrA-overexpresser); and for validation of the ramA reverse transcription polymerase chain reaction (RT-PCR), TS67 (deleted for ramA) and TS68 (deleted for ramR) [19]. The mutants TS67 and TS68 were generated from K. pneumoniae Ecl8 using a modified protocol as described by Merlin et al. [20].

2.2. Minimum inhibitory concentration (MIC) testing

MICs of tigecycline (gift from Wyeth Pharmaceuticals) for the clinical isolates were determined and interpreted according to British Society for Antimicrobial Chemotherapy (BSAC) protocols [21]. Strains exhibiting tigecycline MICs ≤1 μg/mL and >2 μg/mL were classed as sensitive and resistant, respectively.

2.3. Mapping the transcriptional start site of the ramA locus

The transcriptional start site of the ramA locus was mapped according to the manufacturer's instructions using the 5′ RACE (rapid amplification of cDNA ends) system (Invitrogen, Carlsbad, CA). Briefly, 0.5 μg of RNA extracted at mid log phase was reverse transcribed into first-strand cDNA, which was then terminal deoxynucleotidyl transferase (TdT)-tailed. Subsequently, the TdT-tailed cDNA was amplified by PCR and was assessed for positive PCR products. A second round of amplification using nested gene-specific primers (Table 1) and Abridged Universal Amplification Primer (AUAP) was performed prior to the resulting products being subcloned into pGEM^®-T Easy vector (Promega, Southampton, UK) for sequencing.

Table 1.

Primer sequences used in the study.

Primer	Sequence
ramAF	5′-AGCCTGGGGCGCTATATT-3′
ramAR	5′-GTGGTTCTCTTTGCGGTAGG-3′
romAF	5′-GAAGCGTAACCAGACGCTGT-3′
romAR	5′-CTGGTCATACTGCCCGTTCT-3′
ramRF	5′-AACTGCAGTCGTCAAGACGATTTTCAATTTT-3′
ramRR	5′-AAAAGTACTAGTGTTTCCGGCGTCATTAG-3′
acrRF	5′-TTAAGCTGACAAGCTCTCCG-3′
acrRR	5′-ACGTAACCTCTGTAAAGTCAT-3′
pIF	5′-GGGCCAGTTTTCTGTT-3′
pIR	5′-ATAGTATCAATCACCTGAGC-3′
pIIF	5′-CTACTTTTTTCCTCACGCAG-3′
pIIR	5′-CCCTGCGGCGCCTTACCA-3′
16SF	5′-GTTACCCGCAGAAGAAGCAC-3′
16SR	5′-CTACGCATTTCACCGCTACA-3′
ramAR1	5′-TTGCAGATGCCATTTCGA-3′
ramAR2	5′-TATCATCAATACGCAGCG-3′
ramAR3	5′-GGGGTACCATAGTATCAATCACCTGAGC-3′

2.4. Gene expression analyses using RT-PCR

RT-PCR was performed to assess the expression levels of the ramA and romA genes. RNA samples were extracted from the different clinical isolates at mid log phase [optical density at 600 nm (OD₆₀₀) 0.5–0.8] using TRIzol^®. Total RNA was digested with DNase I to ensure the removal of contaminating genomic DNA prior to cDNA synthesis. For cDNA synthesis, a SuperScript^® VILO™ cDNA Synthesis Kit (Invitrogen) was used. Briefly, 300 ng of DNase I-treated total RNA was converted to cDNA and was used in PCRs with gene-specific primers shown in Table 1. A 1 in 10 cDNA dilution was used for amplification of the 16S gene. The resulting PCR products were subjected to electrophoresis on a 1.5% agarose gel. Densitometric analyses using Multi Gauge FujiFilm software of the gel bands were performed and normalised to 16S expression. The fold increase in ramA expression relative to the sensitive strain Ecl8 (ramA non-expresser) is shown in Fig. 2A.

Fig. 2.

(A) Bar chart showing levels of romA and ramA expression in Klebsiella pneumoniae clinical strains. Levels of romA and ramA were normalised to K. pneumoniae Ecl8 (ramA non-expresser). (B) Bar chart showing levels of AcrA expression. Relative fold increases in the AcrA levels were quantified after comparisons with a wild-type sensitive K. pneumoniae isolate Ecl8.

2.5. Electrophoretic mobility shift assay (EMSA)

The romA–ramA promoter regions were amplified and subjected to EMSA with both purified RamA and RamR proteins. Purified RamA and RamR proteins were extracted from recombinant pET constructs containing the ramA and ramR genes using metal chelation chromatography on nickel/nitrilotriacetate superflow agarose (QIAGEN, Crawley, UK). Briefly, end-labelled (using [γ-³²P] ATP; Perkin Elmer, Boston, MA) PCR products were incubated with increasing concentrations (200 nM and 400 nM) of RamA or RamR in binding buffer (125 mM Tris–Cl, 250 mM KCl, 5 mM dithiothreitol [DTT], 160 ng of salmon sperm DNA and 25% glycerol). The complexes were run on 5% native polyacrylamide gel electrophoresis (PAGE) gels for 2.5 h. The gel was then dried and exposed to the phosphor screen for image analysis. To confirm that the interactions between RamA or RamR and the promoter regions were specific, competition experiments with bovine serum albumin (BSA) as a negative control and with cold promoter were also performed.

2.6. Mutations within the promoter regions and the ramR and acrR genes

For those strains where ramA expression was elevated, the ramR and acrR genes and the pI/pII promoter regions were amplified using the primers shown in Table 1. The BigDye™ reaction (Applied Biosystems, Warrington, UK) was set up prior to the products being sequenced at the Genomics Core Facility in Belfast City Hospital (Belfast, UK).

2.7. Western blot analyses

Levels of AcrA were determined by Western blot analyses as described previously by Schneiders et al. [11]. Briefly, cultures were grown to mid log phase (OD₆₀₀ ca. 0.6–0.7) prior to protein extraction using sonication. Then, 15 μg of total protein was loaded onto a 10% NuPAGE^® gel (Invitrogen) prior to transfer to a nitrocellulose membrane as per the manufacturer's instructions. The blots were hybridised with the primary antibody anti-AcrA (kind gift from H. Zgurskaya, University of Oklahoma, Norman, OK) (1:40 000) overnight at 4 °C. The secondary antibody IRDye-800 goat anti-rabbit IgG (Li-Cor Biosciences, Lincoln, NE) was incubated with the blots for 1 h at room temperature and the membranes were scanned using the Odyssey Infrared Imaging system, which is based on near infrared fluorescence detection. The membrane was scanned in two channels at 700 nM and 800 nM, which detects the green and red channels, respectively. All images were quantified using the reading from the 800 nM channel. Analysis of band intensities to assess AcrA levels was performed using the Odyssey software v3 (Li-Cor Biosciences). Levels of AcrA protein of all test strains were compared against the sensitive K. pneumoniae strain Ecl8 to report a relative fold increase in AcrA levels. Both AG100A and AG100B were also used as controls in the AcrA Western blots.

3. Results

3.1. Minimum inhibitory concentrations

The MIC₅₀ and MIC₉₀ values (MICs for 50% and 90% of the organisms, respectively) of tigecycline, minocycline, tetracycline, ciprofloxacin and ceftazidime against the strains are shown in Table 2. Of the 157 strains tested, 24 were resistant to tigecycline based on the BSAC breakpoint criteria [21]. MICs to tigecycline generally ranged from 0.25 μg/mL to 4 μg/mL. Minocycline and tetracycline MIC₉₀ values were 128 μg/mL and >128 μg/mL, respectively; however, cross-resistance to tigecycline was not observed, consistent with previous survey data [22]. Tigecycline-resistant isolates exhibited cross-resistance to ceftazidime and ciprofloxacin.

Table 2.

Minimum inhibitory concentrations (MICs) for 157 clinical isolates.

Antimicrobial agent	MIC (μg/mL)
	MIC50	MIC90
Ciprofloxacin	128	>128
Ceftazidime	>128	>128
Tetracycline	>128	>128
Minocycline	32	128
Tigecycline	1	2

MIC_50/90, MIC for 50% and 90% of the organisms, respectively.

3.2. Bioinformatic analyses for RamR binding sites in Klebsiella pneumoniae (accession no. NC_009648), Citrobacter koseri (accession no. NC_009648) and Enterobacter sp. 638 (accession no. NC_009648)

In S. Typhimurium, ramR has been shown to function as its cognate negative regulator [17]. Given the similarities in the genomic organisation of the ramR and romA–ramA loci between Klebsiella, Enterobacter, Citrobacter and Salmonella, we sought to confirm whether ramR controlled the expression of the romA–ramA locus directly in Klebsiella and whether other accessory binding sites existed for RamR within the romA–ramA locus. Bioinformatic analyses using ClustalW (http://www.ebi.ac.uk/Tools/clustalw2/index.html) showed that the palindromic binding sites identified for RamR in Salmonella spp. are conserved within the intergenic region between the ramR and romA coding sequences of K. pneumoniae, Enterobacter and Citrobacter spp. (Fig. 1A). There appeared to be no perfect palindromic site for RamR upstream of the ramA gene. Thus, we surmised that regulation of the romA–ramA locus mediated by RamR must occur via the palindromic sequence that lies upstream of the romA gene.

3.3. RACE mapping

The transcriptional start site of ramA as determined by RACE mapping is shown in Fig. 1B. The transcriptional start site (hereby called pII) (Fig. 1B) for ramA was found upstream of the romA stop codon, which implies that romA and ramA are part of an operon and are likely co-transcribed in K. pneumoniae. The transcriptional start site mapped corresponds with our previous observation that two transcripts sized ca. 0.6 kb and ca. 0.9 kb were obtained with Northern blotting. The larger transcript must correspond to the promoter that lies upstream of the romA gene. The presence of the pII promoter also suggests that ramA expression can be modulated independently of romA. Correspondingly, this differential regulation is evident in the differing levels of transcription of the romA and ramA genes as shown in Fig. 2A.

3.4. Electrophoretic mobility shift assay

Bioinformatic analyses indicate the presence of a putative but conserved palindromic binding site for RamR within the pI region. However, no such site could be observed for the pII promoter. We also wondered whether RamA, like other similar regulators (i.e. MarA), autoregulates itself by binding to its own promoter region. The intergenic regions, designated pI and pII, bound both purified RamR and RamA proteins (Fig. 3A). The interactions of the pI and pII promoter regions with the RamA and RamR proteins were found to be specific, as no shifts were observed with BSA alone (Fig. 3A) and addition of cold promoter was able to reduce binding of the proteins to the labelled probe (data not shown). Of note, the fragment of DNA (pII*) that contained no transcriptional signals was found not to bind either RamA or RamR (Fig. 3B). As indicated previously, the pI promoter has a conserved RamR binding site that is required for RamR-mediated control. However, the lack of a similar or identical RamR binding site (by bioinformatic analyses) indicates that RamR-mediated regulation of the pII promoter may be mediated via a more degenerate palindromic binding site. Accordingly, bioinformatic analyses using the EMBOSS Pairwise Alignment Tool (http://www.ebi.ac.uk/Tools/emboss/align/) of the pII promoter region implies that there is a putative but degenerate palindromic site upstream of the ramA gene (Fig. 1A).

Fig. 3.

Electrophoretic mobility shift assay (EMSA) of the promoter regions with purified RamA or RamR proteins. (A) EMSA showing the binding of purified RamR and RamA to the pI and pII promoter regions. (B) pII* represents the region underlined (dashed lines) in Fig. 1B.

3.5. DNA mutations in the promoter regions pI and pII and the ramR and acrR genes

3.5.1. Mutations within the ramR gene and levels of romA and ramA expression

Of the 24 strains that were resistant or intermediately susceptible to tigecycline, only 10 harboured mutations within the ramR gene. Changes within the ramR gene were found in both the DNA- and ligand-binding domains of the protein (prediction based on PSIPREDv3 [23]), suggesting that there are no mutational hotspots within ramR, although the majority of changes were clustered within the ligand-binding domain. As previously suspected, not all of the changes observed within RamR resulted in ramA expression, e.g. changes H186N and A187E observed in TS202 and change E41K in TS215 did not result in ramA overexpression. The change I141T was observed in four of the isolates tested (Ecl8, Ecl8Mdr1, TS262 and TS293) (Table 3A). Since this change is also found in the sensitive Klebsiella strain Ecl8 it is likely that it is not associated with ramA overexpression. In addition, strains TS173, TS221, TS257 and TS259 did not harbour changes within the ramR gene (Table 3A) and/or the intergenic regions but still overexpressed ramA (Fig. 2A). Several strains exhibiting MICs ≥ 4 μg/mL were found to overexpress ramA but not to harbour any changes within the ramR gene. Interestingly, levels of ramA expression between those strains that harboured changes within ramR were not higher than those strains that sustained no changes within the gene. In addition, transcriptional levels of the romA gene did not appear to be linked to ramA expression (Fig. 2A), although increased expression of romA was generally associated with strains that harboured changes within the ramR gene (Fig. 2A). Ideally, complementation studies with wild-type ramR would have been performed to confirm whether the ramR changes were directly associated with ramA overexpression. However, this was not possible owing to the multidrug-resistant profiles of the different strains.

Table 3A.

Association between RamR and AcrR changes and tigecycline minimum inhibitory concentration (MICs).

Strain	RamR	AcrR	Tigecycline MIC (μg/mL)	Origin
S7	No change	[11]	8	Singapore
S28	15 nt Δ (558–573 bp)	[11]	2	Singapore
S29	7 nt insertion at nt position 561	[11]	16	Singapore
TS152	T119P	No change	4	Turkey
TS165	T119P	GAG440GAA (E147E)	4	Turkey
TS170	No change	No change	2	Turkey
TS173	No change	No change	2	Turkey
TS184	No change	No change	4	Turkey
TS202	H186N, A187E	No change	2	Chile
TS215	E41K	CCC465CCT (P155P)	4	Chile
TS221	No change	GAG440GAA (E147E)	2	Chile
TS238	A19V	GAG440GAA (E147E)	4	Pakistan
TS240	No change	CCC465CCT (P155P)	4	Pakistan
TS245	No change	TTG641TTT (L214F)	4	Pakistan
TS248	No change	ACC14AAC (T5N)	4	Pakistan
TS250	No change	GAG440GAA (E147E)	4	Pakistan
TS251	No change	No change	8	Pakistan
TS257	No change	No change	16	Pakistan
TS259	No change	No change	2	Pakistan
TS261	No change	GAG440GAA (E147E)	4	Pakistan
TS262	I141T	No change	4	Pakistan
TS267	No change	No change	4	Pakistan
TS293	I141T	CCC465CCU (P155P)	4	Pakistan
TS308	W94Stop	No change	4	Pakistan

nt, nucleotide.

3.5.2. Promoter (pI and pII) mutations

Unlike previous reports, the majority of the clinical isolates did not harbour any changes within the pI promoter. However, the lack of promoter-associated changes linked to increased ramA expression is not unique to this cluster of isolates, as the original ramA-overexpressing strain Ecl8MdrI (as reported by George et al. [15]) also does not harbour any changes within the ramR gene, the palindrome or the −10 and −35 hexamers at the pI promoter. Only one strain (TS170), which sustained a change within the palindromic sequence (C → T, recognised by RamR) but harboured no changes within the ramR gene, was found. Correspondingly, TS170 exhibited a small increase in ramA expression (Fig. 2A). Several changes associated with the pII promoter region were located downstream of the second putative palindrome but upstream of the RACE-mapped transcriptional start site of ramA (Fig. 1B). Three changes (T206C, G229A and A235T) were consistently identified in four strains (S7, TS152, TS165 and TS261) (Table 3B), all of which overexpressed ramA. Of these strains, only S7 and TS261 harboured no changes within the ramR gene (Table 3A).

Table 3B.

Changes observed within the pI and pII promoter regions.

Strain	pI	pII
S7	T141C	T206Ca
S28	–	–
S29	–	–
TS 152	–	T206C, G229A, A235T
TS 165	–	T206C, G229A, A235T
TS 170	T141C, C56Tb	–
TS 243	T141C	–
TS 245	T141C	–
TS 248	T141C	–
TS 250	T141C	–
TS 251	T141C	–
TS 257	T141C	–
TS 261	T141C	T206C, G229A, A235T

^aBold indicates a change within the putative – 35 hexamer sequence upstream of the romA gene.

^bIndicates a recurring change that occurs in clinical isolates from geographically distinct locations.

3.5.3. Mutations within the acrR gene and levels of AcrA protein

Of the 27 strains tested, 9 sustained changes within the acrR gene. Of these changes, only two (T5N and L214F) have been previously documented [11]. The other nucleotide substitution, but not amino acid change (E147E), was found in a region previously shown to be implicated in AcrA overexpression. One new silent change (P155P, CCC → CCT) was observed. Given that most of the changes observed were silent, it is unlikely that they would have contributed significantly to AcrA levels. Strains TS165, TS173, TS257 and TS293, which significantly overexpressed ramA, appear not to produce greater levels of the AcrA protein (Fig. 2B).

4. Discussion

Recent studies have linked antibiotic resistance, particularly tigecycline resistance, to increased expression of the ramA gene. In this study, we identified in silico that key regulatory features of the ramA locus are conserved amongst Klebsiella, Enterobacter, Citrobacter and Salmonella spp.

This work also shows that both the RamR and RamA proteins bind the pI and pII promoters specifically, implying the presence of recognition sites for both proteins within these promoter regions. Like other orthologous systems, such as MarRAB and SoxRS [24], binding of purified RamA to the pI and pII promoters indicates that ramA autoregulates its own expression. In addition, RamR appears to be able to bind regions that either contain a perfect palindrome or more degenerate palindromic sites (Figs. 1A and 3A). Other TetR-like regulators have been shown to regulate via palindromic binding sites that are identified as lower-affinity binding sites owing to key differences in the palindromic binding sequence.

The present results show that regulation of the romA–ramA locus is not entirely identical to that observed in Salmonella as there are two promoters within the romA–ramA locus. The presence of these two promoters supports the possibility that the romA–ramA genes can be regulated independently of each other or as part of an operon (Fig. 2A). Consistent with this genetic arrangement, changes were identified within the pI and putative pII promoter regions in the ramA-overexpressing strains (Table 3B). Interestingly, levels of romA and ramA expression are not linked, supporting the finding that two different promoters control regulation of the romA–ramA locus (Fig. 2A).

We chose to extend the molecular findings by determining the regulation of ramA expression in clinical isolates that were resistant or intermediately susceptible to tigecycline. The results demonstrate that in geographically diverse isolates of K. pneumoniae, 18 isolates exhibited tigecycline MICs ≥ 4 μg/mL (Table 3A). A notable example is strain TS257, which exhibits a tigecycline MIC of 16 μg/mL, and where no mutations were observed within the ramR–romAramA locus but it still overexpressed the ramA gene. Two key observations arise from these data: first, ramA is not always associated with ramR-mediated derepression; and second, decreased tigecycline susceptibility is not always associated with ramA expression. We also did not observe any significant mutations within the promoter regions, as has been observed previously [16,17]. An interesting point is that the isolates in this study pre-date the use of tigecycline in the relevant hospitals, which underscores the likelihood that most antibiotics have the propensity to select for changes that result in ramA overexpression, a similar situation that is noted for the marRAB and soxRS systems [24]. The lack of a direct association between ramA and acrA expression has been noted previously [25] and supports the current data.

Overexpression of transcriptional regulators such as RamA triggers the expression of a multitude of genes that confer pleiotropic phenotypes [26] (Schneiders, unpublished data). Given the varied bacterial response that is mounted with ramA overexpression, we should consider the broader implications of these intrinsic resistance mechanisms in the development and persistence of antimicrobial resistance.