Introduction

Quinolones constitute an important class of broad-spectrum antibacterial drugs and are used for the treatment of various infections. Since the introduction of the first molecule, nalidixic acid, into clinical therapy [1], other molecules have been developed, extending their spectrum of activity [2]. However, increased use of quinolones has led to increasing resistance to these antimicrobials [3]. Indeed, apart from chromosomally encoded quinolone resistance, in the 2000s new mechanisms of quinolone resistance emerged, plasmid-mediated with three mechanisms known so far: Qnr peptides, the aminoglycoside acetyltransferase variant AAC(6′)-Ib-cr, and plasmidic efflux pumps QepA and OqxAB [4]. During the last decade plasmid-mediated quinolone resistance (PMQR) has been increasingly reported in Enterobacteriaceae worldwide [5], including Mediterranean countries that constitute a region with an heterogeneous population from the north of Africa and the south of Europe and the Middle East. This literature review presents a description of the chromosomal and plasmidic fluoroquinolone resistance mechanisms known to date and the epidemiology of occurrence of PMQR determinants in Enterobacteriaceae in Mediterranean countries, with aim of gaining a better understanding of the current situation.

Quinolone structure and clinical use

Quinolones are a family of synthetic broad-spectrum antimicrobial drugs developed by modification of 1-alkyl-1,8-naphthyridin-4-one-3-carboxylic acid [6]. Nalidixic acid was the first molecule introduced into clinical therapy in the early 1960s that was obtained during the synthesis of chloroquine. It had limited clinical use because it was only sufficient for treatment of urinary tract infections and because of the early emergence of resistance [1]. Pharmacological improvements on the basic quinolone ring with the addition of fluorine at position C6 and piperazinyl or related rings at position C7 yielded the fluoroquinolones [7]. Thereby, newer class quinolones that expand the spectrum of activity to include Gram-positive bacteria and even anaerobics were developed in the 1980s (Fig. 1).

Fig. 1
figure 1

Chemical structure of some of the quinolones used in clinical practice (http://www.chemspider.com/)

High bioavailability, relatively low toxicity, and favorable pharmacokinetics have resulted in the clinical success of fluoroquinolones. Thus, these molecules have been widely prescribed to treat respiratory tract infections, including tuberculosis, urinary tract infections (UTIs), intra-abdominal infections, skin and skin structure infections, sexually transmitted diseases, and bone and joint infections [2]. The molecules most frequently used in clinical practice are ciprofloxacin, levofloxacin, and moxifloxacin. They achieve higher serum levels leading to better tissue penetration and greater potency against Enterobacteriaceae and other species [8]. The quinolones that are clinically available have been classified into several generations based on their spectrum of activity (Table 1) [9].

Table 1 Generations and spectrum of activity of quinolones [9]

Some quinolone derivatives demonstrated inhibitory potential against eukaryote topoisomerase II and substantial dose-dependent cytotoxic potential against some cancerous cells [10]. Moreover, some quinolones, such as 6-chloro-7-methoxy4(1H)-quinolones, have good antimalarial activity and good activity against multiple stages of Plasmodium infection [11]. Fluoroquinolones, such as enrofloxacin, have also been extensively used in veterinary medicine to treat and prevent bacterial infections in food-producing animals, aquaculture, and pets, along with growth promoters [12].

Mechanism of action

Quinolones inhibit bacterial DNA synthesis by interfering with the action of two essential bacterial enzymes, DNA gyrase and topoisomerase IV. Both are heterotetrameric type II topoisomerase enzymes breaking transiently both strands of a duplex and pass another double-helical segment through the break by ATP hydrolysis [13]. The DNA gyrase introduces negative supercoils (or relaxes positive supercoils) into DNA, whereas topoisomerase IV exhibits a potent decatenation activity. Those enzymes are essential for bacterial growth as they control the topological status of the chromosomal DNA to facilitate replication, transcription, recombination, and DNA repair [13].

These enzymes comprise two copies of each of either a GyrA (97 kDa) and GyrB (90 kDa) subunit or a ParC (75 kDa) and ParE (70 kDa; GrlA and GrlB in the Staphylococcus aureus) subunit respectively. A few bacteria are able to function with only DNA gyrase, such as Mycobacterium tuberculosis, Helicobacter pylori, and Treponema pallidum, but most bacteria have both enzymes [8]. The targeting of either DNA gyrase or topoisomerase IV as the primary target by fluoroquinolones varies with bacterial species and the specific fluoroquinolone; however, as a broad generalization, the key target in Gram-negative bacteria is DNA gyrase, whereas in Gram-positive microorganisms topoisomerase IV is preferentially targeted [14]. Nevertheless, some recently developed quinolones, such as clinafloxacin and moxifloxacin, have a similar affinity for both targets [15].

Quinolones bind the DNA–enzyme complex of topoisomerase, forming a quinolone–enzyme–DNA complex (known as a ternary complex) in which the type II topoisomerase is trapped with the bound DNA [14]. The drug–enzyme–DNA complexes block cell growth. The bactericidal activity is due to the releasing of double-stranded DNA breaks from those complexes (chromosomal fragmentation process). Furthermore, an indirect route could be responsible for the lethal effect of quinolones, related to reactive oxygen species and/or toxin–antitoxin systems [16, 17]. A bacteriostatic effect is observed at low quinine concentrations (around the minimum inhibitory concentrations [MICs]), with cleaved complexes blocking DNA replication reversibly and inducing the SOS stress response, which in Escherichia coli results in upregulation of various stress response genes that enhance DNA repair capability, leading to the formation of filamentous cells because of the inhibition of cell division [18].

Chromosome-encoded resistance mechanisms

Alterations in target enzymes

The most common mechanism of high-level fluoroquinolone resistance is due to mutations in one or more of the genes that encode the primary and secondary targets of these drugs, the type II topoisomerases (gyrA, gyrB, parC, and parE) [19, 20]. These alterations have been localized mainly in the amino terminal domains of GyrA (residues 67 to 106 for E. coli numbering) or ParC (residues 63–102) and are in proximity to the active site of tyrosines (Tyr122 for GyrA, Tyr120 for ParC), which are covalently linked to DNA in an enzyme intermediate in both enzymes. This domain has been termed the “quinolone resistance determining region” (QRDR) [19, 21, 22]. However, some mutations localized outside the QRDR region, such as position 51, can also lead to decreased susceptibility to the quinolones [15].

Mutations in the QRDR of the topoisomerase genes, resulting in amino acid substitutions, alter the target protein structure and subsequently the fluoroquinolone binding affinity of the enzyme, leading to quinolone resistance [8]. In E. coli and in some other Gram-negative bacteria, such as Citrobacter freundii, Shigella spp., Neisseria gonorrhoeae, and Acinetobacter baumannii, mutations in gyrA appear most frequently at codon Ser83 and mutations at codon Asp87 are the second most commonly observed [2325]. These two amino acids (Ser83 and Asp87) likely play a crucial role in mediating quinolone–enzyme interactions. Indeed, a recent study explained that the interaction between quinolones and topoisomerases was due to the formation of a water–metal ion bridge between the oxygen molecules in the amine group of the drug and the hydroxyl residues in conserved serine or acidic residues in the enzyme, mediated by Mg2+ ion. Thus, a mutation in the place of serine or nearby another amino residue decreases the affinity of the enzyme for quinolones at the noncatalytic Mg2+ site [26].

Different amino acid substitutions at the same position result in different quinolone susceptibility levels. Furthermore, the presence of a single mutation in the QRDR of gyrA usually results in high-level resistance to nalidixic acid only (the MIC of nalidixic acid is greater than 64 mg/L), but to obtain high levels of resistance to fluoroquinolones (ciprofloxacin MICs ranging from 16 to >256 mg/L), the presence of additional mutation(s) in gyrA and/or in another target such as parC is required (Table 2) [15, 27]. Mutations in specific domains of GyrB and ParE have also been shown to cause quinolone resistance, although they are substantially less common in resistant clinical bacterial isolates than mutations in GyrA or ParC [34, 35].

Table 2 Fluoroquinolone resistance mechanisms associated with changes in MIC of ciprofloxacin in E. coli

Reduction in cytoplasmic drug concentrations

As the target of fluoroquinolones is intracellular, these molecules must have the ability to enter the bacterial cell. Quinolones may cross the outer membrane in two different ways: through specific porins (e.g., ciprofloxacin) or by diffusion through the phospholipid bilayer (e.g., nalidixic acid). The rate of diffusion of a quinolone is largely associated with its level of hydrophobicity [36].

Gram-negative bacteria can regulate membrane permeability by altering expression of outer membrane porin proteins that form channels for passive diffusion, such as outer membrane proteins OmpF and OmpC in E. coli. Downregulation of these channels or mutations in their structural genes can lead to reductions in cytoplasmic drug concentrations that may contribute as a resistance mechanism [36].

Another chromosome-encoded resistance mechanism leading to decreased intracellular drug accumulation of quinolones is the upregulation of native efflux pumps. These efflux systems are largely responsible for the intrinsic susceptibility of many species to fluoroquinolones and other drugs, but are also responsible for increased MICs resulting from derepression of the transporter [37]. Acquired quinolone resistance can occur by mutations in genes encoding regulatory proteins that control the transcription of efflux pump or porin genes. Less often, mutations in efflux pump structural genes have been associated with changes in pump substrate profiles that include quinolones [37, 38].

Several Enterobacteriaceae species possess a chromosomal native AcrAB-TolC efflux pump belonging to the resistance–nodulation–division (RND) family. AcrA is a membrane fusion protein, AcrB is an inner-membrane pump, and TolC is an outer-membrane protein and they build up an efflux pump [39]. Mutations in acrR (a repressor of acrAB) increase pump activity. On the other hand, mutations that inactivate marR (a repressor of marA) allow MarA to activate acrAB, tolC, and a gene that decreases the translation of ompF, thus collectively decreasing the influx and increasing the efflux of quinolones [40]. A recent study suggested that marA expression is a sensitive marker for early detection of development of levofloxacin resistance [41]. In addition to the Mar regulon, the operon SoxRS regulates the levels of expression of OmpF and some efflux pumps in E. coli [42].

A decrease in the level of expression of OmpF is related to an increase in the resistance to some quinolones, such as norfloxacin and ciprofloxacin, and a variety of other antibacterial agents [43, 44], but does not affect the MICs of other quinolones, such as tosufloxacin or sparfloxacin [45]. Despite the fact that both reduced quinolone uptake and the overexpression of efflux pumps lead to low-level quinolone resistance (change in ciprofloxacin MIC of 4- to 8-fold; Table 2), they can have an additive effect, with QRDR mutations contributing to higher-level quinolone resistance. Moreover, they provide a favorable environment for other types of resistance to antibacterial agents [29, 30].

Plasmid-mediated resistance mechanisms

Target protection: qnr proteins

Plasmid-mediated resistance to quinolones was first reported in 1998 by Martínez-Martínez et al. in a K. pneumoniae clinical strain that was isolated in July 1994 from the urine of a patient at the University of Alabama at Birmingham Medical Center (UAB) [46]. The gene responsible was termed qnr (later named qnrA1). This gene was mediated by plasmid pMG252, which could transfer low-level resistance to nalidixic acid, but susceptibility to fluoroquinolones to a variety of Gram-negative recipients (Citrobacter, Salmonella, and even P. aeruginosa) [46].

The PMQR protein QnrA binding to gyrase or topoisomerase IV inhibits the gyrase–DNA interaction, and could account at least in part for the protection against quinolones by minimizing opportunities for these agents to stabilize the lethal gyrase–DNA–quinolone complex [47]. This QnrA1 is a 218-amino-acid protein that belongs to the pentapeptide repeat family, of which more than 500 members are known, distributed in prokaryotic and eukaryotic cells [48]. These proteins are made of tandemly repeated amino acid sequences with a consensus sequence [Ser, Thr, Arg, or Val] [Asp or Asn] [Leu or Phe] [Ser, Thr, or Arg] [Gly] [48].

The pentapeptide repeat proteins also include immunity proteins such as McbG, which, together with McbE and McbF, protects a microcin B17-producing strain from self-inhibition [49]. Microcins are a class of small inhibitory proteins (less than 10 kDa) that are encoded by bacterial plasmids and differ in their mechanisms of action. Microcin B17, like quinolones, targets DNA gyrase, and McbEFG has been reported to produce low-level resistance to quinolones such as sparfloxacin [50]. The second Qnr relative is MfpA, a protein cloned from the genome of Mycobacterium smegmatis, that has a 4-fold effect on susceptibility to ciprofloxacin. Qnr, McbG, and MfpA have < 20% amino acid identity and, thus, are not closely related, but their existence suggests that Qnr could have evolved from an immunity protein designed to protect DNA gyrase and DNA topoisomerase IV from some naturally occurring inhibitor [3].

The description of the qnrA gene was subsequently followed by the discovery of plasmid-mediated qnrS in Shigella flexneri 2b [51], qnrB in K. pneumoniae [52], qnrC in Proteus mirabilis [53], and qnrD in Salmonella enterica [54]. The qnrVC (from Vibrio cholerae) has been found on plasmids in Aeromonas punctata [55] and Vibrio fluvialis [56]. It is an atypical member of the Qnr family [57] that differs from the plasmid-mediated qnr genes because of the presence of the attC site, which is characteristic of integron cassettes [58]. These qnr genes generally differ in sequence by 35% or more from qnrA and from each other. Furthermore, most of them contain allelic variants differing by 10% or less (qnrA: 8, qnrS: 9, qnrB: 88, qnrC: 1, qnrD: 2, and qnrVC: 7) in which qnrB constitutes the most heterogeneous cluster of the qnr gene family (http://www.lahey.org/qnrstudies/).

The origin of the plasmid-mediated qnr genes has been found on the chromosome of many species, especially in aquatic bacteria, including species of Aeromonas, Photobacterium, Shewanella, and Vibrio [4]. The progenitor of three QnrA-like determinants (termed QnrA3, QnrA4, and QnrA5) was identified as being the chromosome of the environmental bacterium Shewanella algae [59]. Vibrio splendidus could be the source of QnrS1-like determinants as chromosomal-encoded Qnr-like proteins shared about 84 and 88% amino acid identity with the plasmid-mediated determinants QnrS1 and QnrS2 respectively [60]. On the other hand, the source of qnrB determinants was supposed to be Citrobacter spp. including Citrobacter braakii, Citrobacter werkmanii, and Citrobacter youngae [61]. QnrC is 72% identical to chromosomal Qnr in V. orientalis or V. cholerae [53]. However, qnrD can be found in other Enterobacteriaceae, but are especially likely to be found in Proteeae, such as Proteus, Providencia, and Morganella and may have originated there [62]. Vibrionaceae are the source of qnrVC genes, the qnrVC1 attC site has 89% identity with V. parahaemolyticus repeats, and the qnrVC2 putative attC site has 96% identity with a V. cholerae repeat sequence [57].

Some qnr variants are regulated by the SOS system. The SOS response is triggered by DNA damage, such as that generated by quinolones [63]. The RecA protein is activated by single-stranded DNA and acts as a coprotease to cleave the LexA protein, which otherwise binds as a dimer to LexA boxes, repressing the expression of adjacent genes. Most of these genes are involved in DNA repair or the regulation of cell division [64]. The LexA binding site has been identified in the promoter region of all qnrB alleles [65]. This LexA box is also present in qnrD and Smaqnr (pentapeptide repeat protein reported from the chromosome of Serratia marcescens), but not in the other qnr variants [66]. SOS regulation could serve to protect the host cell from the potentially toxic effects of QnrB, while allowing augmented production upon exposure to quinolone antimicrobial agents [65]. In the uninduced state, the LexA protein is bound to its site at the promoter region of qnrB. Thus, the qnrB gene is expressed at a basal level [67]. On induction of the SOS response, by ciprofloxacin for example, single-stranded DNA (ssDNA) is produced and the co-protease activity of the RecA protein is activated by binding to ssDNA [68]. The interaction between LexA and the nucleoprotein filament RecA/ssDNA results in autoproteolytic cleavage of LexA and subsequently in qnrB derepression [67]. Induced expression of qnrB leads to an increase in the MIC of ciprofloxacin (up to 9-fold). However, a qnrB-containing strain does not express quinolone resistance in non-inducing conditions (silent resistance gene); this resistance will be activated only under selective antibiotic pressure [65, 67]. Moreover, it was shown that the direct SOS-dependent regulation of a low-level fluoroquinolone resistance mechanism could be in response to other antimicrobials such as β-lactams and trimethoprim [66].

Enzymatic inactivation: aminoglycoside acetyl transferase AAC(6 ′)-Ib-cr

The aminoglycoside acetyltransferase AAC(6′)-Ib is usually responsible for resistance to tobramycin, amikacin, and kanamycin [69]. Two codon changes, namely Trp102Arg and Asp179Tyr, are found in the variant AAC(6′)-Ib-cr (ciprofloxacin resistance phenotype), which are required to confer reduced susceptibility to certain fluoroquinolones compared with the wild-type AAC(6′)-Ib [31]. The function of the enzyme is to N-acetylate ciprofloxacin at the amino nitrogen on its piperazinyl substituent [70]. Norfloxacin also has free amino nitrogen acting as a substrate of the enzyme, whereas other fluoroquinolones lacking an unsubstituted piperazinyl nitrogen, such as levofloxacin and moxifloxacin, are unaffected [70]. The effectiveness of this enzyme is modest, leading to an increase in the MICs of ciprofloxacin and norfloxacin 3-fold to 4-fold within the range of low-level resistance (Table 2) [31].

Plasmid-mediated efflux pumps

QepA determinant

The qepA gene was first discovered in two E. coli clinical isolates from Japan and Belgium in 2007 [32, 71]. This gene encodes a 511-amino-acid deduced protein (53 kDa) that shares significant identity with various 14-transmembrane-segment (14-TMS) putative efflux pump belonging to the major facilitator superfamily (MFS) of proton-dependent transporters [32, 71]. The natural reservoir of qepA remains unknown. However, it may consist of Actinomycetales species as QepA had significant amino acid identity with likely membrane transporters of the members of the order of Actinomycetales (such as Streptomyces globisporus, Streptomyces coelicolor, Nocardia farcinica, or Polaromonas spp.), and its high GC% content (72%) is compatible with this origin [5].

This protein confers significantly decreased susceptibility to the hydrophilic quinolones (e.g., norfloxacin, ciprofloxacin, and enrofloxacin) with an 8- to 32-fold increase in MICs compared with a wild-type susceptibility profile (Table 2) [32]. Since its initial discovery, a variant of qepA possessing two amino acid substitutions has been identified and named QepA2 [72]. Wang et al. recently identified a new qepA allele, qepA3, in Enterobacteriaceae isolates from China [73].

OqxAB determinant

The OqxAB belongs to the RND family, being one of the first plasmid-borne efflux pumps described in 2004 in E. coli porcine isolates in Denmark and Sweden, on a conjugative plasmid designated pOLA52 [74, 75]. To our knowledge, this is the first identified genetic resistance mechanism towards olaquindox, an agent used as a growth enhancer in pigs. Moreover, it was shown to mediate resistance to other molecules, such as chloramphenicol, trimethoprim, nalidixic acid, norfloxacin, and ciprofloxacin [33].

The oqxA gene codes for the OqxA membrane fusion protein whereas the oqxB codes for the OqxB protein, containing 12-transmembrane α-helices for the inner-membrane pump. This system requires the TolC outer-membrane protein to function fully [74].

The OqxAB pump is common (usually 75% or more) on the chromosome of K. pneumoniae isolates [76] and it is also prevalent among other species such as Enterobacter spp. [77]. It has been reported that the oqxAB operon, together with a transcriptional regulator (orf68) in the chromosome of K. pneumoniae, was captured by IS26 transposase and transferred to foreign plasmids, which were subsequently disseminated to other bacterial species that do not harbor oqxAB-like elements in the chromosome [77].

As for other PMQR, OqxAB alone confers low-level resistance to fluoroquinolones (Table 2). In K. pneumoniae, overexpression of the nearby rarA gene is associated with increased oqxAB expression, whereas increased expression of the adjacent oqxR gene downregulates OqxAB production [78]. The overexpression of this efflux pump, resulting from a point mutation in the repressor oqxR gene, is responsible for the multidrug resistance phenotype and also for the increased virulence potential in some cases [79].

Impact of PMQR on the clinical resistance level and mutant development

The PMQR genes cause only a modest decrease in ciprofloxacin susceptibility (Table 2). However, it has been shown that PMQR plays an important role in the acquisition of clinical resistance leading to the selection of high-level quinolone resistance [27]. These data indicate that these low-level quinolone-resistant phenotypes may be critical stages (depending on the genotype) in resistance development, including chromosome- and plasmid-mediated mechanisms, at which some low-fitness mutants below the resistance breakpoint are able to evolve clinical resistance with just one or two mutations, and show increased fitness [27].

The clinical importance and the selection of qnr-positive cells to high-level fluoroquinolone resistance can be demonstrated by the mutant preventive concentration (MPC), which is the lowest concentration at which no mutants are observed. It represents the MIC of the least-susceptible single-step mutant [80]. The MPC is measured after plating a large inoculum (1010 bacteria) on concentrations of a quinolone that are above its MIC. Resistant mutants are selectively enriched only in the concentration range between the MIC and the MPC, which is designated as the mutant selection window. Below the MIC, no mutant will be enriched because selective pressure is absent; above MPC, no mutant will be selected because a double mutation is required for growth [81]. Thus, the best schedule for quinolone administration is one that maintains a quinolone concentration that exceeds the MPC for as much of the dosage interval as possible, thereby minimizing the opportunity for mutant selection [3].

Rodríguez-Martínez et al. showed that the expression of qnrA considerably increased the MPC compared with strains without this gene. In the presence of qnrA, mutations in gyrA and parC genes were easily selected to produce high levels of quinolone resistance [82]. Another study reported that the qepA2 gene also increases the MPCs for ciprofloxacin 4- to 16-fold [83].

Experimental animal models have been used to evaluate the in vivo importance of PMQR genes. Indeed, it has been shown that low-level fluoroquinolone resistance conferred by aac(6′)Ib-cr is associated with the reduced bactericidal activity of ciprofloxacin in vivo and to lead to ciprofloxacin therapeutic failure in pyelonephritis. The reduction in bactericidal activity could be explained by in situ N-acetylation of ciprofloxacin [84]. A further work described the significant role of the aac(6′)-Ib-cr gene in the acquisition of a clinical level of ciprofloxacin and norfloxacin resistance, when combined with three or four chromosomal mutations, both in vitro and in vivo [85].

In addition, several studies used the mouse model of pyelonephritis for in vivo experiments to study the interplay in fluoroquinolone resistance mutations and bacterial fitness. Michon et al. [86] investigated host fitness of E. coli isogenic strains after acquisition of the qnrA3 gene, inserted either alone onto a small plasmid or harbored on a large conjugative native plasmid found in a clinical isolate. Indeed, it was shown that qnrA3 acquisition enhanced bacterial fitness, which may explain qnr emergence and suggests the regulatory role of qnr. However, fitness was reduced when qnrA3 was inserted onto multidrug-resistant plasmids and this can slow down its dissemination without antibiotic exposure. Another study based on in vitro and in vivo models reported that most of the combination of target mutations and qnrA1 had a significant fitness cost, whereas most of the combinations harboring qnrS1 had significantly enhanced bacterial fitness. These positive relationships could represent an additional driving force in the development of increased resistance to quinolones [27].

On the other hand, a recent study reported that in E. coli J53 carrying qnrA and aac(6′)Ib-cr or qepA, gyrA mutations are rare, whereas alterations in pumps, porins, and lipopolysaccharides are more common. Mutations upregulating drug efflux (mutations in regulatory genes marR or soxR) may have a fitness cost, but are relevant to resistance evolution because of their pleiotropic effects on susceptibility to other agents and potential for occurrence at a rate higher than that required for specific changes in the QRDR of gyrA [87].

Antibiotic markers, plasmids, and mobilizing elements associated with PMQR genes

Determinants of PMQR have been strongly associated with extended spectrum beta-lactamases (ESBLs) and AmpC ß-lactamases, as shown in Table 3. They have been described in multiresistant clones with worldwide distribution, such as E. coli ST131 and K. pneumoniae ST11 [119, 120]. Moreover, the PMQR genes have often been transferred on multiresistant conjugative plasmids varying in size, which facilitated their global emergence involving several plasmids with different incompatibility groups [4].

Table 3 Prevalence of plasmid-mediated quinolone-resistant (PMQR) genes associated with bla genes in Mediterranean countries

The PMQR qnrA1, qnrA3, qnrA6, qnrB2, qnrB4, qnrB6, and qnrB10 have usually been associated with the mobilizing element insertion sequence common region (ISCR), whereas qnrB1 and qnrB20 have been linked to IS26 and Orf1005 [4]. The qnrB19 allele has been found in a transposon comprising ISEcp1 [121]. The qnrS1 has been associated with an upstream Tn3-like transposon or upstream of the insertion sequence ISEcl2 [122], in several plasmids containing an active TEM-1 gene [4]. Nevertheless, most plasmids carrying qnrS genes are relatively smaller than plasmids carrying qnrA and qnrB and are rarely associated with other antibiotic markers [58]. The qnrC gene has been found to be linked to ISPmi1 [4], whereas the qnrD gene has been found in small nontransmissible plasmids (2.6 to 4 kb) in Proteeae and located inside mobile insertion cassette (mic) elements [62].

The aac(6′)-Ib-cr gene is usually found in a cassette as part of an integron with IS26 in a multiresistance plasmid, which may contain other PMQR genes such as qnrB and also bla CTX-M-15 gene [58]. Moreover, the aac(6′)-Ib-cr gene can also be inserted in the chromosome [123].

The qepA1 determinant has also often been found on plasmids encoding aminoglycoside ribosomal methylase rmtB. IS26 elements and ISCR3C have been implicated in mobilizing the qepA1 gene to plasmids [124], whereas the qepA2 gene was not associated with the rmtB gene and IS26 elements, but was identified to be flanked by two copies of an ISCR-like element [72]. A novel genetic structure surrounding the qepA3 gene has been described. In some strains, qepA3 was linked to the non-rmtB- or non-ISCR3C-producing and was associated with bla CTX-M-14 or bla TEM-12 [73].

Besides being commonly intrinsic in Klebsiella spp. and Enterobacter spp., the OqxAB-encoding genes are often located on conjugative plasmids, within a composite transposon Tn6010 flanked by IS26 [75].

Epidemiology of PMQR determinants: Mediterranean countries

The PMQR determinants emerged in various Enterobacteriaceae species, especially E. coli, Enterobacter spp., Klebsiella spp., and Salmonella spp. [4] and are less common in other Gram-negative bacteria such as Aeromonas spp. [125] A. baumannii [126] and P. aeruginosa [127].

Currently, these determinants have been increasingly reported in Enterobacteriaceae worldwide, including several Mediterranean countries. Although many reports about PMQR determinants have been published in Spain, Tunisia, Italy, France, Algeria, Egypt, and Turkey, there are no published data yet from Libya, Syria, Lebanon, Albania, Bosnia and Herzegovina, Montenegro, and Malta. The geographic distribution of PMQR determinants in the Mediterranean basin is presented in Fig. 2.

Fig. 2
figure 2

Geographic distribution of plasmid-mediated quinolone resistance (PMQR) determinants in Mediterranean countries. Yellow no case reported, orange sporadic or some outbreaks of PMQR-producing isolates

The PMQR genes have been found in different sources, including humans (clinical [96, 111, 117, 128, 129] and community isolates [91, 95, 105, 130]), animals (companion animals [103, 131133] and food-producing animals [90, 94, 134]), and the environment (waste water treatment plant [135] and hospital effluents [91, 136]).

The prevalence of PMQR genes varies by both organism and geographic region and depending on the selection criteria of the strains studied (ESBL-producing, nalidixic acid, fluoroquinolones, etc.) [5]. However, the most commonly reported PMQRs remain the aac(6′)Ib-cr gene followed by the qnr gene. Thus, the aac(6′)-Ib-cr gene had the highest prevalence at 22%, whereas qnr genes were present in 1.3% among a collection of uropathogenic E. coli isolated in a Tunisian university hospital [129]. In contrast, qnr-related genes (qnrA, qnrB, qnrS) and the aac(6′)-Ib-cr gene were identified in 14.3 and 3.6% respectively in Enterobacter cloacae and Klebsiella spp. isolates recovered from cases of clinical and sub-clinical mastitis in Egypt [137]. In many cases, qnrB genes seem to be more prevalent overall than other qnr genes. In Italy, the prevalence rates of qnrA, qnrB, and qnrS were 1.4, 4.3, and 1.4% respectively in Klebsiella spp. isolates from companion animals [103].

An outbreak of K. pneumoniae carrying qnrB1, aac(6′)Ib-cr, and bla CTX-M-15 was reported in a French intensive care unit between 2008 and 2009 [128].

In Spain, the first report of clustered infections caused by a non-ESBL-producing K. pneumoniae strain harboring bla OXA-1 and aac(6’)Ib-cr and belonging to ST14 was identified. The clonal isolates were detected from May 2007 until December 2009 from clinical samples of 38 patients admitted to the University Hospital of Bellvitge, Barcelona, Spain [138]. Dissemination of qnr and aac(6’)Ib-cr genes in E. coli is closely associated with the worldwide spread of a single clone, ST131, which is the main cause of clinic- and community-acquired urinary tract infections [139], as described in community E. coli strains harboring aac(6′)Ib-cr and bla CTX-M-15 isolated in Algeria in 2011 and 2014 [91]. Recent reports have highlighted the Mediterranean Sea as being a potential reservoir of various antibiotic markers, notably PMQR determinants. Indeed, qnr (qnrA, qnrB, and qnrS) and aac(6′)-Ib-cr genes, were identified in 16 (5.8%) and 3 (1.1%) isolates respectively in Gram-negative bacteria isolated from water samples taken from fish farms in the northern part of Egypt at the Mediterranean Sea [94]. Moreover, both aac(6′)Ib-cr and qnrB genes have been identified in CTX-M-15-producing E. coli strains belonging to the pandemic clone ST 131 recovered from wild fish in the Mediterranean Sea far from Bejaia (Algeria) [140].

QepA efflux pump is usually identified with a low prevalence. France was the first Mediterranean country to report the QepA determinant (QepA2 variant) in 2008 in one (0.8%) E. coli ESBL-producing isolate [72]. In Spain, only one isolate (2.4%) showed the presence of qepA among a collection of Salmonella enterica clinical isolates [141]. Very recently, QepA1 was identified in two E. coli clinical strains isolated in 2011 in Algeria (Yanat et al., unpublished data). This determinant was also described in nalidixic acid-resistant Salmonella isolates recovered from various food-animal products in Tunisia, with a rate of 1.16% associated with the qnr gene [142].

The identification of OqxAB plasmidic efflux pump is rare in the Mediterranean basin. This determinant is more frequent in China, mainly from animal isolates [143]. Nevertheless, oqxAB genes were detected in E. coli isolates from animals in Italy with a rate of 15% associated with ST 238 [131] and in K. pneumoniae and E. coli clinical isolates from Spain and Turkey respectively [76, 144].

Conclusion

In parallel to a high consumption of quinolones, different mechanisms of quinolone resistance were developed in Gram-negative bacteria. PMQR emerged in Enterobacteriaceae isolates with a high prevalence in Mediterranean countries, probably because of the great diversity of the population and the exchange relationship between the regions. As the presence of these determinants is often difficult to detect based on the phenotype observed on a routine basis in the laboratory, the real prevalence of PMQR-positive bacteria could be higher than recorded in the published reports. Therefore, long-term surveillance is needed for the future monitoring of trends in the occurrence of PMQR genes in different regions and in various settings. Also, future studies are required focusing on plasmid analysis to understand more about the successful dissemination of PMQR genes.